Archive for the ‘Lipids’ Category

Excess Eating, Overweight, and Diabetic

Larry H Bernstein, MD, FCAP, Curator



You Did NOT Eat Your Way to Diabetes!


The myth that diabetes is caused by overeating also hurts the one out of five people who are not overweight when they contract Type 2 Diabetes. Because doctors only think “Diabetes” when they see a patient who fits the stereotype–the grossly obese inactive patient–they often neglect to check people of normal weight for blood sugar disorders even when they show up with classic symptoms of high blood sugar such as recurrent urinary tract infections or neuropathy.

Where Did This Toxic Myth Come From?

The way this myth originated is this: Because people with Type 2 Diabetes are often overweight and because many people who are overweight have a syndrome called “insulin resistance” in which their cells do not respond properly to insulin so that they require larger than normal amounts of insulin to lower their blood sugar, the conclusion was drawn years ago that insulin resistance was the cause of Type 2 Diabetes.

It made sense. Something was burning out the beta cells in these people, and it seemed logical that the something must be the stress of pumping out huge amounts of insulin, day after day. This idea was so compelling that it was widely believed by medical professionals, though few realized it had never been subjected to careful investigation by large-scale research.

That is why any time there is an article in the news about Type 2 Diabetes you are likely to read something that says, “While Type 1 diabetes (sometimes called Juvenile Diabetes) is a condition where the body does not produce insulin, Type 2 Diabetes is the opposite: a condition where the body produces far too much insulin because of insulin resistance caused by obesity.”

When your doctor tells you the same thing, the conclusion is inescapable: your overeating caused you to put on excess fat and that your excess fat is what made you diabetic.

Blaming the Victim

This line of reasoning leads to subtle, often unexpressed, judgmental decisions on the part of your doctor, who is likely to believe that had you not been such a pig, you would not have given yourself this unnecessary disease.

And because of this unspoken bias, unless you are able to “please” your doctor by losing a great deal of weight after your diagnosis you may find yourself treated with a subtle but callous disregard because of the doctor’s feeling that you brought this condition down on yourself. This bias is similar to that held by doctors who face patients who smoke a pack a day and get lung cancer and still refuse to stop smoking.

You also see this bias frequently expressed in the media. Articles on the “obesity epidemic” blame overeating for a huge increase in the number of people with diabetes, including children and teenagers who are pictured greedily gorging on supersized fast foods while doing no exercise more strenuous than channel surfing. In a society where the concepts “thin” and “healthy” have taken on the overtones of moral virtue and where the only one of the seven deadly sins that still inspires horror and condemnation is gluttony, being fat is considered by many as sure proof of moral weakness. So it is not surprising that the subtext of media coverage of obesity and diabetes is that diabetes is nothing less than the just punishment you deserve for being such a glutton.

Except that it’s not true.

Obesity Has Risen Dramatically While Diabetes Rates Have Not

The rate of obesity has grown alarmingly over the past decades, especially in certain regions of the U.S. The NIH reports that “From 1960-2 to 2005-6, the prevalence of obesity increased from 13.4 to 35.1 percent in U.S. adults age 20 to 74.7.”

If obesity was causing diabetes, you’d exect to see a similar rise in the diabetes rate. But this has not happened. The CDC reports that “From 1980 through 2010, the crude prevalence of diagnosed diabetes increased …from 2.5% to 6.9%.” However, if you look at the graph that accompanies this statement, you see that the rate of diabetes diagnoses rose only gradually through this period–to about 3.5% until it suddenly sped upward in the late 1990s. This sudden increase largely due to the fact that in 1998 the American Diabetes Association changed the criteria by which diabetes was to be diagnosed, lowering the fasting blood sugar level used to diagnose diabetes from 141 mg/dl to 126 mg/dl. (Details HERE)

Analyzing these statistics, it becomes clear that though roughtly 65 million more Americans became fat over this period, only 13 million more Americans became diabetic.

And to further confuse the matter, several factors other than the rise in obesity and the ADA’s lowering of the diagnostic cutoff also came into play during this period which also raised the rate of diabetes diagnoses:

Diabetes becomes more common as people age as the pancreas like other organs, becames less efficient. In 1950 only 12% of the U.S. population was over 65. By 2010 40% was, and of those 40%, 19% were over 75.(Details HERE.)

At the same time, the period during which the rate of diabetes rose was also the period in which doctors began to heavily prescribe statins, a class of drugs we now know raises the risk of developing diabetes. (Details HERE.)

Why Obesity Doesn’t Cause Diabetes: The Genetic Basis of Diabetes

While people who have diabetes are often heavy, one out of five people diagnosed with diabetes are thin or normal weight. And though heavy people with diabetes are, indeed, likely to be insulin resistant, the majority of people who are overweight will never develop diabetes. In fact, they will not develop diabetes though they are likely to be just as insulin resistant as those who do–or even more so.

The message that diabetes researchers in academic laboratories are coming up with about what really causes diabetes is quite different from what you read in the media. What they are finding is that to get Type 2 Diabetes you need to have some combination of a variety of already-identified genetic flaws which produce the syndrome that we call Type 2 Diabetes. This means that unless you have inherited abnormal genes or had your genes damaged by exposure to pesticides, plastics and other environmental toxins known to cause genetic damage, you can eat until you drop and never develop diabetes.

Now let’s look in more depth at what peer reviewed research has found about the true causes of diabetes

Twin Studies Back up a Genetic Cause for Diabetes

Studies of identical twins showed that twins have an 80% concordance for Type 2 Diabetes. In other words, if one twin has Type 2 Diabetes, the chance that the other will have it two are 4 out of 5. While you might assume that this might simply point to the fact that twins are raised in the same home by mothers who feed them the same unhealthy diets, studies of non-identical twins found NO such correlation. The chances that one non-identical twin might have Type 2 Diabetes if the other had it were much lower, though these non-identical twins, born at the same time and raised by the same caregivers were presumably also exposed to the same unhealthy diets.

This kind of finding begins to hint that there is more than just bad habits to blame for diabetes. A high concordance between identical twins which is not shared by non-identical twins is usually advanced as an argument for a genetic cause, though because one in five identical twins did not become diabetic, it is assumed that some additional factors beyond the inherited genome must come into play to cause the disease to appear. Often this factor is an exposure to an environmental toxin which knocks out some other, protective genetic factor.

The Genetic Basis of Type 2 Diabetes Mellitus: Impaired Insulin Secretion versus Impaired Insulin Sensitivity. John E. Gerich. Endocrine Reviews 19(4) 491-503, 1998.

The List of Genes Associated with Type 2 Keeps Growing

Here is a brief list of some of the abnormal genes that have been found to be associated with Type 2 Diabetes in people of European extraction: TCF7L2, HNF4-a, PTPN, SHIP2, ENPP1, PPARG, FTO, KCNJ11, NOTCh3, WFS1, CDKAL1, IGF2BP2, SLC30A8, JAZF1, and HHEX.

People from non-European ethnic groups have been found to have entirely different sets of diabetic genes than do Western Europeans, like the UCP2 polymorphism found in Pima Indians and the three Calpain-10 gene polymorphisms that have been found to be associated with diabetes in Mexicans. The presence of a variation in yet another gene, SLC16A11, was recently found to be associated with a 25% higher risk of a Mexican developing Type 2 diabetes.

The More Diabetes Genes You Have The Worse Your Beta Cells Perform

A study published in the Journal Diabetologia in November 2008 studied how well the beta cells secreted insulin in 1,211 non-diabetic individuals. They then screened these people for abnormalities in seven genes that have been found associated with Type 2 Diabetes.

They found that with each abnormal gene found in a person’s genome, there was an additive effect on that person’s beta cell dysfunction with each additional gene causing poorer beta cell function.

The impact of these genetic flaws becomes clear when we learn that in these people who were believed to be normal, beta cell glucose sensitivity and insulin production at meal times was decreased by 39% in people who had abnormalities in five genes. That’s almost half. And if your beta cells are only putting out half as much insulin as a normal person’s it takes a lot less stress on those cells to push you into becoming diabetic.

Beta cell glucose sensitivity is decreased by 39% in non-diabetic individuals carrying multiple diabetes-risk alleles compared with those with no risk alleles L. Pascoe et al. Diabetologia, Volume 51, Number 11 / November, 2008.

Gene Tests Predict Diabetes Independent of Conventional “Risk Factors”

A study of 16,061 Swedish and 2770 Finnish subjects found that

Variants in 11 genes (TCF7L2, PPARG, FTO, KCNJ11, NOTCh3, WFS1, CDKAL1, IGF2BP2, SLC30A8, JAZF1, and HHEX) were significantly associated with the risk of Type 2 Diabetes independently of clinical risk factors [i.e. family history, obesity etc.]; variants in 8 of these genes were associated with impaired beta-cell function.

Note that though the subjects here were being screened for Type 2 Diabetes, the defect found here was NOT insulin resistance, but rather deficient insulin secretion. This study also found that:

The discriminative power of genetic risk factors improved with an increasing duration of follow-up, whereas that of clinical risk factors decreased.

In short, the longer these people were studied, the more likely the people with these gene defects were to develop diabetes.

Clinical Risk Factors, DNA Variants, and the Development of Type 2 Diabetes Valeriya Lyssenko, M.D. et. al. New England Journal of Medicine, Volume 359:2220-2232, November 20, 2008,Number 21.

What A Common Diabetes Gene Does

A study published in July of 2009 sheds light on what exactly it is that an allele (gene variant) often found associated with diabetes does. The allele in question is one of TCF7L2 transcription factor gene. The study involved 81 normal healthy young Danish men whose genes were tested. They were then given a battery of tests to examine their glucose metabolisms. The researchers found that:

Carriers of the T allele were characterised by reduced 24 h insulin concentrations … and reduced insulin secretion relative to glucose during a mixed meal test … but not during an IVGTT [intravenous glucose tolerance test].

This is an interesting finding, because what damages our bodies is the blood sugar we experience after eating “a mixed meal” but so much research uses the artificial glucose tolerance (GTT) test to assess blood sugar health. This result suggests that the GTT may be missing important signs of early blood sugar dysfunction and that the mixed meal test may be a better diagnostic test than the GTT. I have long believed this to be true, since so many people experience reactive lows when they take the GTT which produces a seemingly “normal reading” though they routinely experience highs after eating meals. These highs are what damage our organs.

Young men with the TCF7L2 allele also responded with weak insulin secretion in response to the incretin hormone GLP-1 and “Despite elevated hepatic [liver] glucose production, carriers of the T allele had significantly reduced 24 h glucagon concentrations … suggesting altered alpha cell function.”

Here again we see evidence that long before obesity develops, people with this common diabetes gene variant show highly abnormal blood sugar behavior. Abnormal production of glucose by the liver may also contribute to obesity as metformin, a drug that that blocks the liver’s production of glucose blocks weight gain and often causes weight loss.

The T allele of rs7903146 TCF7L2 is associated with impaired insulinotropic action of incretin hormones, reduced 24 h profiles of plasma insulin and glucagon, and increased hepatic glucose production in young healthy men. K. Pilgaard et al. Diabetologia, Issue Volume 52, Number 7 / July, 2009. DOI 10.1007/s00125-009-1307-x

Genes Linked to African Heritage Linked to Poor Carbohydrate Metabolism

It has long been known that African-Americans have a much higher rate of diabetes and metabolic syndrome than the American population as a whole. This has been blamed on lifestyle, but a 2009 genetic study finds strong evidence that the problem is genetic.

The study reports,

Using genetic samples obtained from a cohort of subjects undergoing cardiac-related evaluation, a strict algorithm that filtered for genomic features at multiple levels identified 151 differentially-expressed genes between Americans of African ancestry and those of European ancestry. Many of the genes identified were associated with glucose and simple sugar metabolism, suggestive of a model whereby selective adaptation to the nutritional environment differs between populations of humans separated geographically over time.

In the full text discussion the authors state,

These results suggest that differences in glucose metabolism between Americans of African and European may reside at the transcriptional level. The down-regulation of these genes in the AA cohorts argues against these changes being a compensatory response to hyperglycemia and suggests instead a genetic adaptation to changes in the availability of dietary sugars that may no longer be appropriate to a Western Diet.

In conclusion the authors note that the vegetarian diet of the Seventh Day Adventists, often touted as proof of the usefulness of the “Diet Pyramid” doesn’t provide the touted health benefits to people of African American Heritage. Obviously, when hundreds of carbohydrate metabolizing genes aren’t working properly the diet needed is a low carbohydrate diet.

The study is available in full text here:

Stable Patterns of Gene Expression Regulating Carbohydrate Metabolism Determined by Geographic AncestryJonathan C. Schisler et. al. PLoS One 4(12): e8183. doi:10.1371/journal.pone.0008183

Gene that Disrupts Circadian Clock Associated with Type 2 Diabetes

It has been known for a while that people who suffer from sleep disturbances often suffer raised insulin resistance. In December of 2008, researchers identified a gene, “rs1387153, near MTNR1B (which encodes the melatonin receptor 2 (MT2)), as a modulator of fasting plasma glucose.” They conclude,

Our data suggest a possible link between circadian rhythm regulation and glucose homeostasis through the melatonin signaling pathway.

Melatonin levels appear to control the body clock which, in turn, regulates the secretion of substances that modify blood pressure, hormone levels, insulin secretion and many other processes throughout the body.

A variant near MTNR1B is associated with increased fasting plasma glucose levels and type 2 diabetes risk. Nabila Bouatia-Naji et al. Nature Genetics Published online: 7 December 2008, doi:10.1038/ng.277

There’s an excellent translation of what this study means, translated into layman’s terms at Science Daily:

Body Clock Linked to Diabetes And High Blood Sugar In New Genome-wide Study


The Environmental Factors That Push Borderline Genes into Full-fledged Diabetes

We’ve seen so far that to get Type 2 Diabetes you seem to need to have some diabetes gene or genes, but that not everyone with these genes develops diabetes. There are what scientists call environmental factors that can push a borderline genetic case into full fledged diabetes. Let’s look now at what the research has found about what some of these environmental factors might be.


Your Mother’s Diet During Pregnancy May Have Caused Your Diabetes

Many “environmental factors” that scientists explore occur in the environment of the womb. Diabetes is no different, and the conditions you experienced when you were a fetus can have life-long impact on your blood sugar control.

Researchers following the children of mothers who had experienced a Dutch famine during World War II found that children of mothers who had experienced famine were far more likely to develop diabetes in later life than a control group from the same population whose mothers had been adequately fed.

Glucose tolerance in adults after prenatal exposure to famine. Ravelli AC et al.Lancet. 1998 Jan 17;351(9097):173-7.,

A study of a Chinese population found a link between low birth weight and the development of both diabetes and impaired glucose regulation (i.e. prediabetes) that was independent of “sex, age, central obesity, smoking status, alcohol consumption, dyslipidemia, family history of diabetes, and occupational status.” Low birth weight in this population may well be due to less than optimal maternal nutrition during pregnancy.

Evidence of a Relationship Between Infant Birth Weight and Later Diabetes and Impaired Glucose Regulation in a Chinese Population Xinhua Xiao et. al. Diabetes Care31:483-487, 2008.

This may not seem all that relevant to Americans whose mothers have not been exposed to famine conditions. But to conclude this is to forget how many American teens and young women suffer from eating disorders and how prevalent crash dieting is in the group of women most likely to get pregnant.

It is also true that until the 1980s obstetricians routinely warned pregnant women against gaining what is now understood to be a healthy amount of weight. When pregnant women started to gain weight, doctors often put them on highly restrictive diets which resulted in many case in the birth of underweight babies.

Your Mother’s Gestational Diabetes May Have Caused Your Diabetes

Maternal starvation is not the only pre-birth factor associated with an increased risk of diabetes. Having a well-fed mother who suffered gestational diabetes also increases a child’s risk both of obesity and of developing diabetes.

High Prevalence of Type 2 Diabetes and Pre-Diabetes in Adult Offspring of Women With Gestational Diabetes Mellitus or Type 1 Diabetes The role of intrauterine hyperglycemia Tine D. Clausen, MD et al. Diabetes Care 31:340-346, 2008

Pesticides and PCBs in Blood Stream Correlate with Incidence of Diabetes

A study conducted among members of New York State’s Mohawk tribe found that the odds of being diagnosed with diabetes in this population was almost 4 times higher in members who had high concentrations of PCBs in their blood serum. It was even higher for those with high concentrations of pesticides in their blood.

Diabetes in Relation to Serum Levels of Polychlorinated Biphenyls and Chlorinated Pesticides in Adult Native Americans Neculai Codru, Maria J. Schymura,Serban Negoita,Robert Rej,and David O. Carpenter.Environ Health Perspect. 2007 October; 115(10): 1442-1447.Published online 2007 July 17. doi: 10.1289/ehp.10315.

It is very important to note that there is no reason to believe this phenomenon is limited to people of Native American heritage. Upstate NY has a well-known and very serious PCB problem–remember Love Canal? And the entire population of the U.S. has been overexposed to powerful pesticides for a generation.

More evidence that obesity may be caused by exposure to toxic pollutants which damage genes comes in a study published January of 2009. This study tracked the exposure of a group of pregnant Belgian woman to several common pollutants: hexachlorobenzene, dichlorodiphenyldichloroethylene (DDE) , dioxin-like compounds, and polychlorinated biphenyls (PCBs). It found a correlation between exposure to PCBs and DDE and obesity by age 3, especially in children of mothers who smoked.

Intrauterine Exposure to Environmental Pollutants and Body Mass Index during the First 3 Years of Life Stijn L. Verhulst et al., Environmental Health Perspectives. Volume 117, Number 1, January 2009

These studies, which garnered no press attention at all, probably have more to tell us about the reason for the so-called “diabetes epidemic” than any other published over the last decade.

BPA and Plasticizers from Packaging Are Strongly Linked to Obesity and Insulin Resistance

BPA, the plastic used to line most metal cans has long been suspected of causing obesity. Now we know why. A study published in 2008 reported that BPA suppresses a key hormone, adiponectin, which is responsible for regulating insulin sensitivity in the body and puts people at a substantially higher risk for metabolic syndrome.

Science Daily: Toxic Plastics: Bisphenol A Linked To Metabolic Syndrome In Human Tissue

The impact of BPA on children is dramatic. Analysis of 7 years of NHANES epidemiological data found that having a high urine level of BPA doubles a child’s risk of being obese.

Bisphenol A and Chronic Disease Risk Factors in US Children. Eng, Donna et al.Pediatrics Published online August 19, 2013. doi: 10.1542/peds.2013-0106

You, and your children are getting far more BPA from canned foods than what health authorities assumed they were getting. A research report published in 2011 reported that the level of BPA actually measured in people’s bodies after they consumed canned soup turned out to be extremely high. People who ate a serving of canned soup every day for five days had BPA levels of 20.8 micrograms per liter of urine, whereas people who instead ate fresh soup had levels of 1.1 micrograms per liter.

Canned Soup Consumption and Urinary Bisphenol A: A Randomized Crossover Trial Carwile, JL et al. JAMA. November 23/30, 2011, Vol 306, No. 20

Nevertheless, the FDA caved in to industry pressure in 2012 and refused to regulate BPA claiming that, as usual, more study was needed. (FDA: BPA)

BPA is not the only toxic chemical associated with plastics that may be promoting insulin resistance. . Phthalates are compounds added to plastic to make it flexible. They rub off on our food and are found in our blood and urine. A study of 387 Hispanic and Black, New York City children who were between six and eight years old measured the phthalates in their urine and found that the more phthalates in their urine, the fatter the child was a year later.

Associations between phthalate metabolite urinary concentrations and body size measures in New York City children.
Susan L. Teitelbaum et al.Environ Res. 2012 Jan;112:186-93.

This finding was echosed by another study:

Urinary phthalates and increased insulin resistance in adolescents Trasande L, et al. Pediatrics 2013; DOI: 10.1542/peds.2012-4022.

And phthalates are everywhere. A study of 1,016 Swedes aged 70 years and older found that four phthalate metabolites were detected in the blood serum of almost all the participants. High levels of three of these were associated with the prevalence of diabetes. The researchers explain that one metabolite was mainly related to poor insulin secretion, whereas two others were related to insulin resistance. The researchers didn’t check to see whether this relationship held for prediabetes.

Circulating Levels of Phthalate Metabolites Are Associated With Prevalent Diabetes in the Elderly.Lind, MP et al. Diabetes. Published online before print April 12, 2012, doi: 10.2337/dc11-2396

Chances are very good that these same omnipresent phthalates are also causing insulin resistance and damaging insulin secretion in people whose ages fall between those of the two groups studied here.

Use of Herbicide Atrazine Maps to Obesity, Causes Insulin Resistance

A study published in April of 2009 mentions that “There is an apparent overlap between areas in the USA where the herbicide, atrazine (ATZ), is heavily used and obesity-prevalence maps of people with a BMI over 30.”

It found that when rats were given low doses of this pesticide in thier water, “Chronic administration of ATZ decreased basal metabolic rate, and increased body weight, intra-abdominal fat and insulin resistance without changing food intake or physical activity level.” In short the animals got fat even without changing their food intake. When the animals were fed a high fat,high carb diet, the weight gain was even greater.

Insulin resistance was increased too, which if it happens in people, means that people who have genetically-caused borderline capacity to secrete insulin are more likely to become diabetic when they are exposed to this chemical via food or their drinking water.

Chronic Exposure to the Herbicide, Atrazine, Causes Mitochondrial Dysfunction and Insulin Resistance PLoS ONE Published 13 Apr 2009

2,4-D A Common Herbicide Blocks Secretion of GLP-1–A Blood Sugar Lowering Gastric Peptide

In 2007 scientists at New York’s Mount Sinai Hospital discovered that the intestine has receptors for sugar identical to those found on the tongue and that these receptors regulate secretion of glucagon-like peptide-1 (GLP-1). GLP-1 is the peptide that is mimicked by the diabetes drug Byetta and which is kept elevated by Januvia and Onglyza. You can read about that finding in this Science Daily report:

Science Daily: Your Gut Has Taste Receptors

In November 2009, these same scientists reported that a very common herbicide 2,4 D blocked this taste receptor, effectively turning off its ability to stimulate the production GLP-1. The fibrate drugs used to lower cholesterol were also found to block the receptor.

Science Daily: Common Herbicides and Fibrates Block Nutrient-Sensing Receptor Found in Gut and Pancreas

What was even more of concern was the discovery that the ability of these compounds to block this gut receptor “did not generalize across species to the rodent form of the receptor.” The lead researcher was quoted as saying,

…most safety tests were done using animals, which have T1R3 receptors that are insensitive to these compounds,

This takes on additional meaning when you realize that most compounds released into the environment are tested only on animals, not humans. It may help explain why so many supposedly “safe” chemicals are damaging human glucose metabolisms.

Trace Amounts of Arsenic in Urine Correlate with Dramatic Rise in Diabetes

A study published in JAMA in August of 2008 found of 788 adults who had participated in the 2003-2004 National Health and Nutrition Examination Survey (NHANES) found those who had the most arsenic in their urine, were nearly four times more likely to have diabetes than those who had the least amount.

The study is reported here:

Arsenic Exposure and Prevalence of Type 2 Diabetes in US Adults. Ana Navas-Acien et al. JAMA. 2008;300(7):814-822.

The New York Times report about this study (no longer online) added this illuminating bit of information to the story:

Arsenic can get into drinking water naturally when minerals dissolve. It is also an industrial pollutant from coal burning and copper smelting. Utilities use filtration systems to get it out of drinking water.

Seafood also contains nontoxic organic arsenic. The researchers adjusted their analysis for signs of seafood intake and found that people with Type 2 Diabetes had 26 percent higher inorganic arsenic levels than people without Type 2 Diabetes.

How arsenic could contribute to diabetes is unknown, but prior studies have found impaired insulin secretion in pancreas cells treated with an arsenic compound.

Prescription Drugs, Especially SSRI Antidepressants Cause Obesity and Possibly Diabetes

Another important environmental factor is this: Type 2 Diabetes can be caused by some commonly prescribed drugs. Beta blockers and atypical antipsychotics like Zyprexa have been shown to cause diabetes in people who would not otherwise get it. This is discussed here.

There is some research that suggests that SSRI antidepressants may also promote diabetes. It is well known that antidepressants cause weight gain.

Spin doctors in the employ of the drug companies who sell these high-profit antidepressants have long tried to attribute the relationship between depression and obesity to depression, rather than the drugs used to treat the condition.

However, a new study published in June 2009 used data from the Canadian National Population Health Survey (NPHS), a longitudinal study of a representative cohort of household residents in Canada and tracked the incidence of obesity over ten years.

The study found that, “MDE [Major Depressive Episode] does not appear to increase the risk of obesity. …Pharmacologic treatment with antidepressants may be associated with an increased risk of obesity. [emphasis mine]. The study concluded,

Unexpectedly, significant effects were seen for serotonin-reuptake-inhibiting antidepressants [Prozac,Celexa, Lovox, Paxil, Zoloft] and venlafaxine [Effexor], but neither for tricyclic antidepressants nor antipsychotic medications.

Scott B. Patten et al. Psychother Psychosom 2009;78:182-186 (DOI: 10.1159/000209349)

Here is an article posted by the Mayo Clinic that includes the statement “weight gain is a reported side effect of nearly all antidepressant medications currently available.

Antidepressants and weight gain –

Here is a report about a paper presented at the 2006 ADA Conference that analyzed the Antidepressant-Diabetes connection in a major Diabetes prevention study:

Medscape: Antidepressant use associated with increased type 2 diabetes risk.

Treatment for Cancer, Especially Radiation, Greatly Increases Diabetes Risk Independent of Obesity or Exercise Level

A study published in August 2009 analyzed data for 8599 survivors in the Childhood Cancer Survivor Study. It found that after adjusting for body mass and exercise levels, survivors of childhood cancer were 1.8 times more likely than the siblings to report that they had diabetes.

Even more significantly, those who had had full body radiation were 7.2 times more likely to have diabetes.

This raises the question of whether exposure to radiation in other contexts also causes Type 2 diabetes.

Diabetes Mellitus in Long-term Survivors of Childhood Cancer: Increased Risk Associated With Radiation Therapy: A Report for the Childhood Cancer Survivor Study.Lillian R. Meacham et al. Arch. Int. Med.Vol. 169 No. 15, Aug 10/24, 2009.

More Insight into the Effect of Genetic Flaws

Now that we have a better idea of some of the underlying physiological causes of diabetes, lets look more closely at the physiological processes that takes place as these genetic flaws push the body towards diabetes.

Insulin Resistance Develops in Thin Children of People with Type 2 Diabetes

Lab research has come up with some other intriguing findings that challenge the idea that obesity causes insulin resistance which causes diabetes. Instead, it looks like the opposite happens: Insulin resistance precedes the development of obesity.

One of these studies took two groups of thin subjects with normal blood sugar who were evenly matched for height and weight. The two groups differed only in that one group had close relatives who had developed Type 2 Diabetes, and hence, if there were a genetic component to the disorder, they were more likely to have it. The other group had no relatives with Type 2 Diabetes. The researchers then and examined the subjects’ glucose and insulin levels during a glucose tolerance test and calculated their insulin resistance. They found that the thin relatives of the people with Type 2 Diabetes already had much more insulin resistance than did the thin people with no relatives with diabetes.

Insulin resistance in the first-degree relatives of persons with Type 2 Diabetes. Straczkowski M et al. Med Sci Monit. 2003 May;9(5):CR186-90.

This result was echoed by a second study published in November of 2009.

That study compared detailed measurements of insulin secretion and resistance in 187 offspring of people diagnosed with Type 2 diabetes against 509 controls. Subjects were matched with controls for age, gender and BMI. It concluded:

The first-degree offspring of type 2 diabetic patients show insulin resistance and beta cell dysfunction in response to oral glucose challenge. Beta cell impairment exists in insulin-sensitive offspring of patients with type 2 diabetes, suggesting beta cell dysfunction to be a major defect determining diabetes development in diabetic offspring.

Beta cell (dys)function in non-diabetic offspring of diabetic patients M. Stadler et al. Diabetologia Volume 52, Number 11 / November, 2009, pp 2435-2444. doi 10.1007/s00125-009-1520-7

Mitochondrial Dysfunction is Found in Lean Relatives of People with Type 2 Diabetes

One reason insulin resistance might precede obesity was explained by a landmark 2004 study which looked at the cells of the “healthy, young, lean” but insulin-resistant relatives of people with Type 2 Diabetes and found that their mitochondria, the “power plant of the cells” that is the part of the cell that burns glucose, appeared to have a defect. While the mitochondria of people with no relatives with diabetes burned glucose well, the mitochondria of the people with an inherited genetic predisposition to diabetes were not able to burn off glucose as efficiently, but instead caused the glucose they could not burn and to be stored in the cells as fat.

Impaired mitochondrial activity in the insulin-resistant offspring of patients with type 2 diabetes. Petersen KF et al. New England J Med 2004 Feb 12; 350(7);639-41

More Evidence that Abnormal Insulin Resistance Precedes Weight Gain and Probably Causes It

A study done by the same researchers at Yale University School of Medicine who discovered the mitochondrial problem we just discussed was published in Proceedings of the National Academy of Science (PNAS) in July 2007. It reports on a study that compared energy usage by lean people who were insulin resistant and lean people who were insulin sensitive.

The role of skeletal muscle insulin resistance in the pathogenesis of the metabolic syndrome Petersen,KF et al. PNAS July 31, 2007 vol. 104 no. 31 12587-12594.

Using new imaging technologies, the researchers found that lean but insulin resistant subjects converted glucose from high carbohydrate meals into triglycerides–i.e. fat. Lean insulin-sensitive subjects, in contrast, stored the same glucose in the form of muscle and liver glycogen.

The researchers conclude that:

the insulin resistance, in these young, lean, insulin resistant individuals, was independent of abdominal obesity and circulating plasma adipocytokines, suggesting that these abnormalities develop later in the development of the metabolic syndrome.”

In short, obesity looked to be a result, not a cause of the metabolic flaw that led these people to store carbohydrate they ate in the form of fat rather than burn it for energy.

The researchers suggested controlling insulin resistance with exercise. It would also be a good idea for people who are insulin resistant, or have a family history of Type 2 Diabetes to cut back on their carb intake, knowing that the glucose from the carbs they eat is more likely to turn into fat.

Beta Cells Fail to Reproduce in People with Diabetes

A study of pancreas autopsies that compared the pancreases of thin and fat people with diabetes with those of thin and fat normal people found that fat, insulin-resistant people who did not develop diabetes apparently were able to grow new beta-cells to produce the extra insulin they needed. In contrast, the beta cells of people who developed diabetes were unable to reproduce. This failure was independent of their weight.

Beta-Cell Deficit and Increased Beta-Cell Apoptosis in Humans With Type 2 Diabetes. Alexandra E. Butler, et al. Diabetes 52:102-110, 2003

Once Blood Sugars Rise They Impair a Muscle Gene that Regulates Insulin Sensitivity

Another piece of the puzzle falls into place thanks to a research study published on Feb 8, 2008.

Downregulation of Diacylglycerol Kinase Delta Contributes to Hyperglycemia-Induced Insulin Resistance. Alexander V. Chibalin et. al. Cell, Volume 132, Issue 3, 375-386, 8 February 2008.

As reported in Diabetes in Control (which had access to the full text of the study)

The research team identified a “fat-burning” gene, the products of which are required to maintain the cells insulin sensitivity. They also discovered that this gene is reduced in muscle tissue from people with high blood sugar and type 2-diabetes. In the absence of the enzyme that is made by this gene, muscles have reduced insulin sensitivity, impaired fat burning ability, which leads to an increased risk of developing obesity.

“The expression of this gene is reduced when blood sugar rises, but activity can be restored if blood sugar is controlled by pharmacological treatment or exercise”, says Professor Juleen Zierath. “Our results underscore the importance of tight regulation of blood sugar for people with diabetes.”

In short, once your blood sugar rises past a certain point, you become much more insulin resistant. This, in turn, pushes up your blood sugar more.

A New Model For How Diabetes Develops

These research findings open up a new way of understanding the relationship between obesity and diabetes.

Perhaps people with the genetic condition underlying Type 2 Diabetes inherit a defect in the beta cells that make those cells unable to reproduce normally to replace cells damaged by the normal wear and tear of life.Or perhaps exposure to an environmental toxin damages the related genes.

Perhaps, too, a defect in the way that their cells burn glucose inclines them to turn excess blood sugar into fat rather than burning it off as a person with normal mitochondria might do.

Put these facts together and you suddenly get a fatal combination that is almost guaranteed to make a person fat.

Studies have shown that blood sugars only slightly over 100 mg/dl are high enough to render beta cells dysfunctional.

Beta-cell dysfunction and glucose intolerance: results from the San Antonio metabolism (SAM) study. Gastaldelli A, et al. Diabetologia. 2004 Jan;47(1):31-9. Epub 2003 Dec 10.

In a normal person who had the ability to grow new beta cells, any damaged beta cells would be replaced by new ones, which would keep the blood sugar at levels low enough to avoid further damage. But the beta cells of a person with a genetic heritage of diabetes are unable to reproduce So once blood sugars started to rise, more beta cells would succumb to the resulting glucose toxicity, and that would, in turn raise blood sugar higher.

As the concentration of glucose in their blood rose, these people would not be able to do what a normal person does with excess blood sugar–which is to burn it for energy. Instead their defective mitochondria will cause the excess glucose to be stored as fat. As this fat gets stored in the muscles it causes the insulin resistance so often observed in people with diabetes–long before the individual begins to gain visible weight. This insulin resistance puts a further strain on the remaining beta cells by making the person’s cells less sensitive to insulin. Since the person with an inherited tendency to diabetes’ pancreas can’t grow the extra beta cells that a normal person could grow when their cells become insulin resistant this leads to ever escalating blood sugars which further damage the insulin-producing cells, and end up in the inevitable decline into diabetes.

Low Fat Diets Promote the Deterioration that Leads to Diabetes in People with the Genetic Predisposition

In the past two decades, when people who were headed towards diabetes begin to gain weight, they were advised to eat a low fat diet. Unfortunately, this low fat diet is also a high carbohydrate diet–one that exacerbates blood sugar problems by raising blood sugars dangerously high, destroying more insulin-producing beta-cells, and catalyzing the storage of more fat in the muscles of people with dysfunctional mitochondria. Though they may have stuck to diets to low fat for weeks or even months these people were tormented by relentless hunger and when they finally went off their ineffective diets, they got fatter. Unfortunately, when they reported these experiences to their doctors, they were almost universally accused of lying about their eating habits.

It has only been documented in medical research during the past two years that that many patients who have found it impossible to lose weight on the low fat high carbohydrate can lose weight–often dramatically–on a low carbohydrate diet while improving rather than harming their blood lipids.

Very low-carbohydrate and low-fat diets affect fasting lipids and postprandial lipemia differently in overweight men. Sharman MJ, et al. J Nutr. 2004 Apr;134(4):880-5.

An isoenergetic very low carbohydrate diet improves serum HDL cholesterol and triacylglycerol concentrations, the total cholesterol to HDL cholesterol ratio and postprandial lipemic responses compared with a low fat diet in normal weight, normolipidemic women. Volek JS, et al. J Nutr. 2003 Sep;133(9):2756-61.

The low carb diet does two things. By limiting carbohydrate, it limits the concentration of blood glucose which often is enough to bring moderately elevated blood sugars down to normal or near normal levels. This means that there will be little excess glucose left to be converted to fat and stored.

It also gets around the mitochondrial defect in processing glucose by keeping blood sugars low so that the body switches into a mode where it burns ketones rather than glucose for muscle fuel.

Relentless Hunger Results from Roller Coaster Blood Sugars

There is one last reason why you may believe that obesity caused your diabetes, when, in fact, it was undiagnosed diabetes that caused your obesity.

Long before a person develops diabetes, they go through a phase where they have what doctors called “impaired glucose tolerance.” This means that after they eat a meal containing carbohydrates, their blood sugar rockets up and may stay high for an hour or two before dropping back to a normal level.

What most people don’t know is that when blood sugar moves swiftly up or down most people will experience intense hunger. The reasons for this are not completely clear. But what is certain is that this intense hunger caused by blood sugar swings can develop years before a person’s blood sugar reaches the level where they’ll be diagnosed as diabetic.

This relentless hunger, in fact, is often the very first diabetic symptom a person will experience, though most doctors do not recognize this hunger as a symptom. Instead, if you complain of experiencing intense hunger doctors may suggest you need an antidepressant or blame your weight gain, if you are female, on menopausal changes.

This relentless hunger caused by impaired glucose tolerance almost always leads to significant weight gain and an increase in insulin resistance. However, because it can take ten years between the time your blood sugar begins to rise steeply after meals and the time when your fasting blood sugar is abnormal enough for you to be diagnosed with diabetes, most people are, indeed, very fat at the time of diagnosis.

With better diagnosis of diabetes (discussed here) we would be able to catch early diabetes before people gained the enormous amounts of weight now believed to cause the syndrome. But at least now people with diabetic relatives who are at risk for developing diabetes can go a long way towards preventing the development of obesity by controlling their carbohydrate intake long before they begin to put on weight.

You CAN Undo the Damage

No matter what your genetic heritage or the environmental insults your genes have survived, you can take steps right now to lower your blood sugar, eliminate the secondary insulin resistance caused by high blood sugars, and start the process that leads back to health. The pages linked here will show you how.

How To Get Your Blood Sugar Under Control

What Can You Eat When You Are Cutting The Carbs?

What is a Normal Blood Sugar

Research Connecting Blood Sugar Level with Organ Damage

The 5% Club: They Normalized Their Blood Sugar and So Can You


Read Full Post »

Obesity Issues

Larry H. Bernstein, MD, FCAP, Curator



The Changing Face of Obesity

Science tells us obesity is a chronic disease. Why does the outmoded and injurious notion that it is a problem of willpower persist?

By Joseph Proietto | November 1, 2015

In Dante Alighieri’s Divine Comedy the narrator meets a man named Ciacco who had been sent to Hell for the “Damning sin of Gluttony.” According to Catholic theology, in order to end up in Hell one must willfully commit a serious sin. So Dante believed that fat people chose to be fat. This antiquated view of the cause of obesity is still widespread, even among medical professionals. The consequences of this misconception are significant, because it forms the basis for the discrimination suffered by the obese; for the wasting of scarce resources in attempts to change lifestyle habits by public education; and for the limited availability of subsidized obesity treatments.

While obesity is often labeled a lifestyle disease, poor lifestyle choices alone account for only a 6 to 8 kg weight gain. The body has a powerful negative feedback system to prevent excessive weight gain. The strongest inhibitor of hunger, the hormone leptin, is made by fat cells. A period of increased energy intake will result in fat deposition, which will increase leptin production. Leptin suppresses hunger and increases energy expenditure. This slows down weight gain. To become obese, it may be necessary to harbor a genetic difference that makes the individual resistant to the action of leptin.

Evidence from twin and adoption studies suggests that obesity has a genetic basis, and over the past two decades a number of genes associated with obesity have been described. The most common genetic defect in European populations leading to severe obesity is due to mutations in the gene coding for the melanocortin 4 receptor (MCR4). Still, this defect can explain severe obesity in only approximately 6 percent to 7 percent of cases (J Clin Invest, 106:271-79, 2000). Other genes have been discovered that can cause milder increases in weight; for example, variants of just one gene (FTO) can explain up to 3 kg of weight variation between individuals (Science, 316:889-94, 2007).

Genes do not directly cause weight gain. Rather, genes influence the desire for food and the feeling of satiety. In an environment with either poor access to food or access to only low-calorie food, obesity may not develop even in persons with a genetic predisposition. When there is an abundance of food and a sedentary lifestyle, however, an obesity-prone person will experience greater hunger and reduced satiety, increasing caloric intake and weight gain.

Since the 1980s, there has been a rapid rise in the prevalence of obesity worldwide, a trend that likely results from a variety of complex causes. There is increasing evidence, for example, that the development of obesity on individual or familial levels may be influenced by environmental experiences that occur in early life. For example, if a mother is malnourished during early pregnancy, this results in epigenetic changes to genes involved in the set points for hunger and satiety in the developing child. These changes may then become fixed, resulting in a tendency towards obesity in the offspring.

The biological basis of obesity is further highlighted by the vigorous defense of weight following weight loss. There are at least 10 circulating hormones that modulate hunger. Of these, only one has been confirmed as a hunger-inducing hormone (ghrelin), and it is made and released by the stomach. In contrast, nine hormones suppress hunger, including CCK, PYY, GLP-1, oxyntomodulin, and uroguanylin from the small bowel; leptin from fat cells; and insulin, amylin, and pancreatic polypeptide from the pancreas.


After weight loss, regardless of the diet employed, there are changes in circulating hormones involved in the regulation of body weight. Ghrelin levels tend to increase and levels of multiple appetite-suppressing hormones decrease. There is also a subjective increase in appetite. Researchers have shown that even after three years, these hormonal changes persist (NEJM, 365:1597-604, 2011; Lancet Diabetes and Endocrinology, 2:954-62, 2014). This explains why there is a high rate of weight regain after diet-induced weight loss.

Given that the physiological responses to weight loss predispose people to regain that weight, obesity must be considered a chronic disease. Data show that those who successfully maintain their weight after weight loss do so by remaining vigilant and constantly applying techniques to oppose weight regain. These techniques may involve strict diet and exercise practices and/or pharmacotherapy.

It is imperative for society to move away from a view that obesity is simply a lifestyle issue and to accept that it is a chronic disease. Such a change would not only relieve the stigma of obesity but would also empower politicians, scientists and clinicians to tackle the problem more effectively.

Joseph Proietto was the inaugural Sir Edward Dunlop Medical Research Foundation Professor of Medicine in the Department of Medicine, Austin Health at the University of Melbourne in Australia. He is a researcher and clinician investigating and treating obesity and type 2 diabetes.



A Weighty Anomaly

Why do some obese people actually experience health benefits?

By Jyoti Madhusoodanan | November 1, 2015

THE ENDOCRINE THEORY: Some researchers have posited that fat cells may secrete molecules that affect glucose homeostasis in muscle or liver tissue.COURTESY OF MITCHELL LAZAR

In the early 19th century, Belgian mathematician Adolphe Quetelet was obsessed with a shape: the bell curve. While helping with a population census, Quetelet proposed that the spread of human traits such as height and weight followed this trend, also known as a Gaussian or normal distribution. On a quest to define a “normal man,” he showed that human height and weight data fell along his beloved bell curves, and in 1823 devised the “Quetelet Index”—more familiar to us today as the BMI, or body mass index, a ratio of weight to height.

Nearly two centuries later, clinicians, researchers, and fitness instructors continue to rely on this metric to pigeonhole people into categories: underweight, healthy, overweight, or obese. But Quetelet never intended the metric to serve as a way to define obesity. And now, a growing body of evidence suggests these categories fail to accurately reflect the health risks—or benefits—of being overweight.

Although there is considerable debate surrounding the prevalence of metabolically healthy obesity, when obesity is defined in terms of BMI (a BMI of 30 or higher), estimates suggest that about 10 percent of adults in the U.S. are obese yet metabolically healthy, while as many as 80 percent of those with a normal BMI may be metabolically unhealthy, with signs of insulin resistance and poor circulating lipid levels, even if they suffer no obvious ill effects. “If all we know about a person is that they have a certain body weight at a certain height, that’s not enough information to know their health risks from obesity,” says health-science researcher Paul McAuley of Winston-Salem State University. “We need better indicators of metabolic health.”

The dangers of being overweight, such as a higher risk of heart disease, type 2 diabetes, and other complications, are well known. But some obese individuals—dubbed the “fat fit”—appear to fare better on many measures of health when they’re heavier. Studies have found lower mortality rates, better response to hemodialysis in chronic kidney disease, and lower incidence of dementia in such people. Mortality, it’s been found, correlates with obesity in a U-shaped curve (J Sports Sci, 29:773-82, 2011). So does extra heft help or hurt?

To answer that question, researchers are trying to elucidate the metabolic reasons for this obesity paradox.

In a recent study, Harvard University epidemiologist Goodarz Danaei and his colleagues analyzed data from nine studies involving a total of more than 58,000 participants to tease apart how obesity and other well-known metabolic risk factors influence the risk of coronary heart disease. Controlling these other risk factors, such as hypertension or high cholesterol, with medication is simpler than curbing obesity itself, Danaei explains. “If you control a person’s obesity you get rid of some health risks, but if you control hypertension or diabetes, that also reduces health risks, and you can do the latter much more easily right now.”

Danaei’s team assessed BMI and metabolic markers such as systolic blood pressure, total serum cholesterol, and fasting blood glucose. The three metabolic markers only explained half of the increased risk of heart disease across all study participants. In obese individuals, the other half appeared to be mediated by fat itself, perhaps via inflammatory markers or other indirect mechanisms (Epidemiology, 26:153-62, 2015). While Danaei’s study was aimed at understanding how obesity hurts health, the results also uncovered unknown mechanisms by which excess adipose tissue might exert its effects. This particular study revealed obesity’s negative effects, but might these unknown mechanisms hold clues that explain the obesity paradox?

Other researchers have suggested additional possibilities—for example, that inflammatory markers such as TNF-α help combat conditions such as chronic kidney disease, or that obesity makes a body more capable of making changes to, and tolerating changes in, blood flow depending on systemic needs (Am J Clin Nutr, 81:543-54, 2005).

According to endocrinologist Mitchell Lazar at the University of Pennsylvania, the key to explaining the obesity paradox may be two nonexclusive ways fat tissue is hypothesized to function. One mechanism, termed the endocrine theory, suggests that fat cells secrete, or don’t secrete enough of, certain molecules that influence glucose homeostasis in other tissues, such as muscle or liver. The first such hormone to be discovered was leptin; later studies reported several other adipocyte-secreted factors, including adiponectin, resistin, and various cytokines.

The other hypothesis, dubbed the spillover theory, suggests that storing lipids in fat cells has some pluses. Adipose tissue might sequester fat-soluble endotoxins, and produce lipoproteins that can bind to and clear harmful lipids from circulation. When fat cells fill up, however, these endotoxins are stashed in the liver, pancreas, or other organs—and that’s when trouble begins. In “fat fit” people, problems typically linked to obesity such as high cholesterol or diabetes may be avoided simply because their adipocytes mop up more endotoxins.

“In this model, one could imagine that if you could store even more fat in fat cells, you could be even more obese, but you might be protected from problems [associated with] obesity because you’re protecting the other tissues from filling up with lipids that cause problems,” says Lazar. “This may be the most popular current model to explain the fat fit.”

Although obesity greatly increases the risk of type 2 diabetes—up to 93-fold in postmenopausal women, for example—not all obese people suffer from the condition. Similarly, a certain subtype of individuals with “normal” BMIs are at greater risk of developing insulin resistance and type 2 diabetes than others with BMIs in the same range. Precisely what distinguishes these two cohorts is still unclear. “Just as important as explaining why some obese people don’t get diabetes is to explain why other subgroups—normal-weight people or those with lipodystrophy—sometimes get it,” Lazar says. “If there are multiple subtypes of obesity and diabetes, can we figure out genetic aspects or biomarkers that cause one of these phenotypes and not the other?”

To Lazar, McAuley, and other researchers, it’s increasingly evident that BMI may not be that metric. Finding better ways to assess a healthy weight, however, has proven challenging. Researchers have tested measures, such as the body shape index (ABSI) or the waist-hip ratio, which attempt to gauge visceral fat—considered to be more metabolically harmful than fat in other body locations. However, these metrics have yet to be implemented widely in clinics, and few are as simple to understand as the BMI (Science, 341:856-58, 2013).

Independent of metrics, however, the health message regarding weight is still unanimous: exercise and healthy dietary choices benefit everyone. “At a certain point, despite all the so-called fit-fat people, the demographics say that there’s a huge risk of diabetes and heart disease at very high BMI,” notes Lazar. “We can’t assume we’ll be one of the lucky ones who will have a BMI in the obese category but will still be protected from heart disease.”

Correction (November 2): The original version of this article misattributed the pull quote above. The attribution for this quote has been corrected, and The Scientist regrets the error.




 Science 23 Aug 2013;  341(6148): 856858     DOI:
Obesity paradoxes.
In this review, we examine the original obesity paradox phenomenon (i.e. in cardiovascular disease populations, obese patients survive better), as well as three other related paradoxes (pre-obesity, “fat but fit” theory, and “healthy” obesity). An obesity paradox has been reported in a range of cardiovascular and non-cardiovascular conditions. Pre-obesity (defined as a body mass index of 25.0-29.9 kg · m⁻²) presents another paradox. Whereas “overweight” implies increased risk, it is in fact associated with decreased mortality risk compared with normal weight. Another paradox concerns the observation than when fitness is taken into account, the mortality risk associated with obesity is offset. The final paradox under consideration is the presence of a sizeable subset of obese individuals who are otherwise healthy. Consequently, a large segment of the overweight and obese population is not at increased risk for premature death. It appears therefore that low cardiorespiratory fitness and inactivity are a greater health threat than obesity, suggesting that more emphasis should be placed on increasing leisure time physical activity and cardiorespiratory fitness as the main strategy for reducing mortality risk in the broad population of overweight and obese adults.
Obesity, insulin resistance, and cardiovascular disease.
Recent Prog Horm Res. 2004;59:207-23.
The ability of insulin to stimulate glucose disposal varies more than six-fold in apparently healthy individuals. The one third of the population that is most insulin resistant is at greatly increased risk to develop cardiovascular disease (CVD), type 2 diabetes, hypertension, stroke, nonalcoholic fatty liver disease, polycystic ovary disease, and certain forms of cancer. Between 25-35% of the variability in insulin action is related to being overweight. The importance of the adverse effects of excess adiposity is apparent in light of the evidence that more than half of the adult population in the United States is classified as being overweight/obese, as defined by a body mass index greater than 25.0 kg/m(2). The current epidemic of overweight/obesity is most-likely related to a combination of increased caloric intake and decreased energy expenditure. In either instance, the fact that CVD risk is increased as individuals gain weight emphasizes the gravity of the health care dilemma posed by the explosive increase in the prevalence of overweight/obesity in the population at large. Given the enormity of the problem, it is necessary to differentiate between the CVD risk related to obesity per se, as distinct from the fact that the prevalence of insulin resistance and compensatory hyperinsulinemia are increased in overweight/obese individuals. Although the majority of individuals in the general population that can be considered insulin resistant are also overweight/obese, not all overweight/obese persons are insulin resistant. Furthermore, the cluster of abnormalities associated with insulin resistance – namely, glucose intolerance, hyperinsulinemia, dyslipidemia, and elevated plasma C-reactive protein concentrations — is limited to the subset of overweight/obese individuals that are also insulin resistant. Of greater clinical relevance is the fact that significant improvement in these metabolic abnormalities following weight loss is seen only in the subset of overweight/obese individuals that are also insulin resistant. In view of the large number of overweight/obese subjects at potential risk to be insulin resistant/hyperinsulinemic (and at increased CVD risk), and the difficulty in achieving weight loss, it seems essential to identify those overweight/obese individuals who are also insulin resistant and will benefit the most from weight loss, then target this population for the most-intensive efforts to bring about weight loss.
Long-Term Persistence of Hormonal Adaptations to Weight Loss

Priya Sumithran, Luke A. Prendergast, Elizabeth Delbridge, Katrina Purcell, Arthur Shulkes, Adamandia Kriketos, and Joseph Proietto

N Engl J Med 2011; 365:1597-1604   October 27, 2011

After weight loss, changes in the circulating levels of several peripheral hormones involved in the homeostatic regulation of body weight occur. Whether these changes are transient or persist over time may be important for an understanding of the reasons behind the high rate of weight regain after diet-induced weight loss.

Weight loss (mean [±SE], 13.5±0.5 kg) led to significant reductions in levels of leptin, peptide YY, cholecystokinin, insulin (P<0.001 for all comparisons), and amylin (P=0.002) and to increases in levels of ghrelin (P<0.001), gastric inhibitory polypeptide (P=0.004), and pancreatic polypeptide (P=0.008). There was also a significant increase in subjective appetite (P<0.001). One year after the initial weight loss, there were still significant differences from baseline in the mean levels of leptin (P<0.001), peptide YY (P<0.001), cholecystokinin (P=0.04), insulin (P=0.01), ghrelin (P<0.001), gastric inhibitory polypeptide (P<0.001), and pancreatic polypeptide (P=0.002), as well as hunger (P<0.001).

What’s new in endocrinology and diabetes mellitus

Large genome wide association studies have demonstrated that variants in the FTO gene have the strongest association with obesity risk in the general population, but the mechanism of the association has been unclear. However, a nonocoding causal variant in FTO has now been identified that changes the function of adipocytes from energy utilization (beige fat) to energy storage (white fat) with a fivefold decrease in mitochondrial thermogenesis [17]. When the effect of the variant was blocked in genetically engineered mice, thermogenesis increased and weight gain did not occur, despite eating a high-fat diet. Blocking the gene’s effect in human adipocytes also increased energy utilization. This observation has important implications for potential new anti-obesity drugs. (See “Pathogenesis of obesity”, section on ‘FTO variants’.)

Liraglutide for the treatment of obesity (July 2015)

Along with diet, exercise, and behavior modification, drug therapy may be a helpful component of treatment for select patients who are overweight or obese. Liraglutide is a glucagon-like peptide-1 (GLP-1) receptor agonist, used for the treatment of type 2 diabetes, and can promote weight loss in patients with diabetes, as well as those without diabetes.

In a randomized trial in nondiabetic patients who had a body mass index (BMI) of ≥30 kg/m2 or ≥27 kg/m2 with dyslipidemia and/or hypertension, liraglutide 3 mg once daily, compared with placebo, resulted in greater mean weight loss (-8.0 versus -2.6 kg with placebo) [18]. In addition, cardiometabolic risk factors, glycated hemoglobin (A1C), and quality of life improved modestly. Gastrointestinal side effects transiently affected at least 40 percent of the liraglutide group and were the most common reason for withdrawal (6.4 percent). Liraglutide is an option for select overweight or obese patients, although gastrointestinal side effects (nausea, vomiting) and the need for a daily injection may limit the use of this drug. (See “Obesity in adults: Drug therapy”, section on ‘Liraglutide’.)

In a trial designed specifically to evaluate the effect of liraglutide on weight loss in overweight or obese patients with type 2 diabetes (mean weight 106 kg), liraglutide, compared with placebo, resulted in greater mean weight loss (-6.4 kg and -5.0 kg for liraglutide 3 mg and 1.8 mg, respectively, versus -2.2 kg for placebo) [19]. Treatment with liraglutide was associated with better glycemic control, a reduction in the use of oral hypoglycemic agents, and a reduction in systolic blood pressure. Although liraglutide is not considered as initial therapy for the majority of patients with type 2 diabetes, it is an option for select overweight or obese patients with type 2 diabetes who fail initial therapy with lifestyle intervention and metformin.  (See “Glucagon-like peptide-1 receptor agonists for the treatment of type 2 diabetes mellitus”, section on ‘Weight loss’.)

The Skinny on Fat Cells

Bruce Spiegelman has spent his career at the forefront of adipocyte differentiation and metabolism.

By Anna Azvolinsky | November 1, 2015

Bruce Spiegelman
Stanley J. Korsmeyer Professor of Cell Biology
and Medicine
Harvard Medical School
Director, Center for Energy Metabolism
and Chronic
Disease, Dana-Farber Cancer Institute, Boston

It’s hard to know whether you have the right stuff to be a scientist, but I had a passion for the research,” says Bruce Spiegelman, professor of cell biology at Harvard Medical School and the Dana-Farber Cancer Institute. After receiving his PhD in biochemistry from Princeton University in 1978, Spiegelman sent an application to do postdoctoral research to just one lab. “I wasn’t thinking I should apply to five different labs. I just marched forward more or less in a straight line,” he says. Spiegelman did know that he had no financial backup and depended on research fellowships throughout the early phase of his science career. “I thought it was fantastic, and still think so, that a PhD in science is supported by the government. I certainly appreciated that, because many of my friends in the humanities had to support themselves by cobbling together fellowships and teaching every semester, whereas we didn’t face similar challenges in the sciences.”

Since his graduate student days, Spiegelman has realized his potential, pioneering the study of adipose tissue biology and metabolism. He was introduced to the field in Howard Green’s laboratory, then at MIT, where Spiegelman began his one and only postdoc in 1978. Green had recently developed a system for culturing adipose cells and asked Spiegelman if he wanted to study fat cell differentiation. “I knew nothing about adipose tissue, but I was really interested in any model of how one cell switches to another. Whether skin or fat didn’t matter too much to me, because I was not coming at this from the perspective of physiology but from the perspective of how do these switches work at a molecular level?”

Spiegelman has stuck with studying the biology and differentiation of fat cells for more than 30 years. While looking for the master transcriptional regulator of fat development—which his laboratory found in 1994—Spiegelman’s group also discovered one of the first examples of a nuclear oncogene that functions as a transcription factor, and, more recently, the team found that brown fat and white fat come from completely different origins and that brown and beige fat are distinct cell types. Spiegelman was also the first to provide evidence for the connection between inflammation, insulin resistance, and fat tissue.

Here, Spiegelman talks about his strong affinity for the East Coast, his laboratory’s search for molecules that can crank up brown fat production and activity, and the culture of his laboratory’s weekly meeting.

Spiegelman Sets Out

First publication. Spiegelman grew up in Massapequa, New York, a town on Long Island. “Birds, insects, fish, and animals were fascinating to me. As a kid, I imagined I would be a wildlife ranger,” he says. Spiegelman and his brother were the first in their family to attend college; Spiegelman entered the College of William and Mary in 1970 thinking he would major in psychology. But before taking his first psychology course, he had to take a biology course, really loved it, and switched his major. For his senior thesis, he chose one of the few labs that did biochemistry-related research. He studied cultures of the filamentous fungus Aspergillus ornatus in which he induced the upregulation of a metabolic enzyme. Spiegelman applied a calculus transformation that related the age of the culture to the age of individual cells, something that had not been previously done. The work earned him his first first-author publication in 1975. “It was not a great breakthrough, but I think it showed that I was maybe applying myself more than the typical undergraduate.”

Full steam ahead. “My interest in laboratory research was intense. Even though it was not particularly inspired work, the first-author publication in a college where not many of the professors published a lot gave me a lot of confidence. It was probably out of proportion to the quality of the actual work.” That confidence and Spiegelman’s interest in the chemistry of living things led him to pursue a PhD in biochemistry at Princeton University. “Very early on, I felt that I couldn’t understand biology if it didn’t go to the molecular level. To me, just describing how an animal lived without understanding how it worked was very unsatisfying. I think it was one of the best decisions that I made in my life, to do a PhD in biochemistry,” he says, “because if you really want to understand living systems, you are very limited in how you can understand them without having a strong background in biochemistry because these are, essentially, chemical systems.”

Embracing molecular biology. Spiegelman initially joined Arthur Pardee’s laboratory, but switched when Pardee left Princeton for Harvard University in 1975. Because he was already collaborating with Marc Kirschner, a cell biologist and biochemist who studies the regulation of the cell cycle and how the cytoskeleton works, it was an easy transition to transfer to the new laboratory. In Kirschner’s group, Spiegelman became the cell biologist among many protein biochemists working on microtubule assembly in vitro. Rather than understanding how the proteins fit together to form the filamentous structures, Spiegelman wanted to understand what controlled their assembly inside cells. Working in mammalian cells, Spiegelman published three consecutive Cell papers on how microtubule assembly occurs in vivo. The firstpaper, from 1977, demonstrated that a nucleotide functions to stabilize the tubulin molecule rather than to regulate tubulin assembly in vivo.

Spiegelman Simmers

A new tool. For his next move, Spiegelman wanted to marry his background in biochemistry and molecular biology with a good cellular model system. He became interested in differentiation at the end of his PhD, while studying how the cytoskeleton is reorganized during neural differentiation, and settled on Green’s MIT laboratory for his postdoc. Green had developed a way to study both skin and fat cell differentiation. Again, Spiegelman was the odd man out, working on the molecular biology of fat cell differentiation while most of the graduate students and postdocs focused on the cellular biology of skin cell differentiation. While there, Spiegelman learned how to clone cDNA—a new method that some researchers thought was just another new fad, he says. “I thought it was pretty obvious that this was a tool that would be a game changer. I could see how I could clone some of the cDNAs and genes that were regulated in the fat cell lineage and then try to understand the regulation of these genes.”

Setting the stage. Spiegelman demonstrated that cAMP regulates the synthesis of certain enzymes in fat cells during differentiation. But while this was the most influential paper from his postdoc, says Spiegelman, it was his demonstration of cloning mRNAs from adipocytes, published in 1983, that set the stage for cloning fat-selective genes. The work, mostly done when Spiegelman was already a new faculty member at the Dana-Farber Cancer Institute, stemmed from his learning molecular cloning in Phillip Sharp’s lab at MIT and Bryan Roberts’s lab at Harvard. “This was the raw material from which we eventually cloned PPARγ and showed it to be the master regulator of fat [cell] development.”

Roots. Spiegelman became an assistant professor at the Harvard Medical School in 1982, when he was not yet 30. Although he had entertained the idea of moving to the West Coast with his wife, whom he had met at Princeton where she obtained a PhD in French literature, Spiegelman says he is really an East Coaster at heart. “My wife and I came to love Boston and were very comfortable there. Our families were both in New York, which was close, but not too close, and we really enjoyed the culture and pace of Boston; it was more ‘us.’ We really liked to visit California but didn’t particularly want to move there. We’re both real Northeastern people.”

Relating to Sisyphus. The transition from doing a postdoc to setting up his own laboratory was “very exciting and terribly stressful,” says Spiegelman. “When I think back, I always tried to be professional with my laboratory, but I was so stressed at suddenly being on my own with no management training.” The people resources he had encountered in his graduate and postdoctoral training labs were also not there yet, and he says his first publication as a principal investigator was like pushing a rock up a hill. But eventually, Spiegelman’s lab built a reputation and reached a critical mass of talented people who advanced the science. Again in 1983, Spiegelman produced a publication showing that morphological manipulation can affect gene expression and adipose differentiation.

End goal. Spiegelman’s goal was to find a master molecule that  orchestrates the conversion of adipocyte precursor cells into bona fide fat cells. Piece by piece, his lab identified the enhancers, promoters, and other regulatory elements involved in adipocyte differentiation. In 1994, graduate student Peter Tontonoz finallyfound that the PPARγ gene, inserted via a retroviral vector into fibroblasts, could induce the cells to become adipose cells. “It took 10 years,” Spiegelman says. Along the way, the laboratory found that c-fos, the product of a famous nuclear oncogene, bound to the promoters of fat-specific genes and worked as a transcription factor. “It was not really known how nuclear oncogenes worked. This was one of the first papers showing that these oncogenes bound to gene promoters and were transcription factors.”

A wider scope. In 1993, graduate student Gökhan Hotamisligil found that tumor necrosis factor-alpha(TNF-α), is induced in the fat tissue of rodent models of obesity and diabetes. The paper sparked the formation of the field of immunometabolism and resulted in the expansion of Spiegelman’s lab into the physiology arena, partly thanks to the guidance of C. Ronald Kahn and Jeff Flier, who both study metabolism and diabetes. But the work initially encountered pushback, says Spiegelman, partly because it was the merging of two fields.

Spiegelman Scales Up

Fat color palette. Brown fat tissue, abundant in infants but scarce in adults, is a metabolically active form of fat that is chock full of mitochondria and is found in pockets in the body distinct from white fat tissue.Pere Puigserver, then a postdoc in Spiegelman’s lab, found that the coactivator PCG-1, binding to PPARγ and other nuclear receptors, could stimulate mitochondrial biogenesis. The PCG-1 gene is turned on by stimuli such as exercise or a cold environment. Later, postdoc Patrick Seale, Spiegelman, and their colleagues showed brown fat cells derive from the same lineage that gives rise to skeletal muscle. “This was a big surprise, maybe the biggest surprise we ever uncovered in the lab,” says Spiegelman.

A paler shade of brown. More recently, in 2012, Spiegelman’s laboratory showed that within adult white adipose tissue, there are pockets of a yet another type of fat tissue that he called beige fat. “I think the evidence is very good from rodents that if you activate brown and beige fat, you get metabolic benefit both in obesity and diabetes. So the question now is: Can that be done in humans in a way that’s beneficial and not toxic?”  The lab is now looking to identify molecules that can either ramp up the activity of brown and beige fat or increase the production of both cell types as possible therapeutics for metabolic disorders or even cancer-associated cachexia. “Anyone who says that either approach will work better is being foolish. We just don’t know enough to go after just one or the other.”

On the irisin controversy. After reporting in 2012 that a muscle-related hormone called irisin could switch white fat to metabolically active brown fat, Spiegelman became embroiled in a media-covered debate about whether the molecule really exists; he was also the victim of a potential fraud plot. Most recently, Spiegelman provided thorough evidence that irisin does in fact exist. On the controversy, he says it’s a fine line between defending his scientific integrity and not adding more fuel to the fire or engaging with his harassers. “We have a long track record of doing credible and reproducible science and it was not that complicated to address the paper that claimed irisin was ‘a myth.’ That study used very outmoded scientific approaches.”

Raw talent. Many of Spiegelman’s trainees have gone on to become very successful scientists, including Tontonoz, Hotamisligil, Evan Rosen, and Randy Johnson. “It’s a quantum change in the experience of doing science when you get people who have their own visions. I would have thought that interacting with smart people would mainly help me get my scientific vision accomplished. And that was partly true, but also it changed my vision. When you have people challenging you on a day-to-day basis, you learn from them through the questions they ask and the way they challenge you in a constructive way. They made me a much better scientist.”

Rigorous mentorship.  “I feel very passionately that a major part of my job is to prepare the next generation of scientists. Everyone who comes through my lab will tell you that I take that very seriously. We make sure my students give a lot of talks and get critical assessments of their presentations to our lab group. I am very hands-on both scientifically and in developing the way students project their vision. I had a very good mentor, Marc Kirschner, and I’d like to think that I learned how to be a mentor from him. I want to make sure that when people walk out of my lab they are prepared to run independent research programs.”

Greatest Hits

  • Identified the master regulator of adipogenesis, the nuclear receptor PPARγ
  • Was the first to show that a nuclear oncogene, c-fos, codes for a transcription factor that binds to the promoters of genes
  • Demonstrated that adipose tissue synthesizes tumor necrosis factor-alpha (TNF-α), providing the first direct link between obesity, inflammation, insulin resistance, and fat tissue.
  • Showed that brown fat cells are not developmentally related to white fat
  • Identified beige fat as a distinct cell type, different from either white or brown fat


Fanning the Flames

Obesity triggers a fatty acid synthesis pathway, which in turn helps drive T cell differentiation and inflammation.

By Kate Yandell | November 1, 2015


The paper
Y. Endo et al., “Obesity drives Th17 cell differentiation by inducing the lipid metabolic kinase, ACC1,” Cell Reports, 12:1042-55, 2015.

Cell Rep. 2015 Aug 11;12(6):1042-55. Epub 2015 Jul 30.
Obesity Drives Th17 Cell Differentiation by Inducing the Lipid Metabolic Kinase, ACC1.
  • A high-fat diet augments Th17 cell development and the expression of Acaca
  • ACC1 controls Th17 cell development in vitro and Th17 cell pathogenicity in vivo
  • ACC1 modulates RORγt function in developing Th17 cells
  • Obesity in humans induces ACACA and IL-17A expression in CD4 T cells

Chronic inflammation due to obesity contributes to the development of metabolic diseases, autoimmune diseases, and cancer. Reciprocal interactions between metabolic systems and immune cells have pivotal roles in the pathogenesis of obesity-associated diseases, although the mechanisms regulating obesity-associated inflammatory diseases are still unclear. In the present study, we performed transcriptional profiling of memory phenotype CD4 T cells in high-fat-fed mice and identified acetyl-CoA carboxylase 1 (ACC1, the gene product of Acaca) as an essential regulator of Th17 cell differentiation in vitro and of the pathogenicity of Th17 cells in vivo. ACC1 modulates the DNA binding of RORγt to target genes in differentiating Th17 cells. In addition, we found a strong correlation between IL-17A-producing CD45RO(+)CD4 T cells and the expression of ACACA in obese subjects. Thus, ACC1 confers the appropriate function of RORγt through fatty acid synthesis and regulates the obesity-related pathology of Th17 cells.

Figure thumbnail fx1

FEEDING INFLAMMATION: When mice eat a diet high in fat, their CD4 T cells show increased expression of the fatty acid biosynthesis gene Acaca, which encodes the enzyme ACC1 (1). Products of the ACC1 fatty acid synthesis pathway encourage the transcription factor RORγt to bind near the gene encoding the cytokine IL-17A (2). There, RORγt recruits an enzyme called p300 to modify the genome epigenetically and turn on IL-17A. The memory T cells then differentiate into inflammatory T helper 17 cells.
See full infographic: PDF

Obesity often comes with a side of chronic inflammation, causing inflammatory chemicals and immune cells to flood adipose tissue, the hypothalamus, the liver, and other areas of the body. Inflammation is a big part of what makes obesity such an unhealthy condition, contributing to Type 2 diabetes, heart disease, cancers, autoimmune disorders, and possibly even neurodegenerative diseases.

To better understand the relationship between obesity and inflammation, Toshinori Nakayama, Yusuke Endo, and their colleagues at Chiba University in Japan started with what often leads to obesity: a high-fat diet. They fed mice rich meals for a couple of months and looked at how gene expression in the animals’ T cells compared to gene expression in the T cells of mice fed a normal diet. Most notably, they found increased expression ofAcaca, a gene that codes for a fatty acid synthesis enzyme called acetyl coA carboxylase 1 (ACC1). They went on to show that the resulting increase in fatty acid levels pushed CD4 T cells to differentiate into inflammatory T helper 17 (Th17) cells.

Th17 cells help fight off invading fungi and some bacteria. But these immune cells can also spin out of control in autoimmune diseases such as multiple sclerosis. Nakayama’s team showed that either blocking ACC1 activity with a drug called TOFA or deleting a key portion of Acaca in mouse CD4 T cells reduced the generation of pathologic Th17 cells. Overexpressing Acaca increased Th17-cell generation.

The researchers also demonstrated that mice fed a high-fat diet had elevated susceptibility to a multiple sclerosis–like disease, and that TOFA reduced the symptoms.

“This is a very intriguing finding, suggesting not only that obesity can directly induce Th17 differentiation but also indicating that pharmacologic targeting of fatty acid synthesis may help to interfere with obesity-associated inflammation,” Tim Sparwasser of the Twincore Center for Experimental and Clinical Infection Research in Hannover, Germany, says in an email. Sparwasser and his colleagues had previously shown that ACC1 is required for the differentiation of Th17 cells in mice and humans.

Nakayama explains that CD4 T cells must undergo profound metabolic changes as they mature and differentiate. “The intracellular metabolites, including fatty acids, are essential for cell proliferation and cell growth,” he says in an email. When fatty acid levels in T cells increase, the cells are activated and begin to proliferate.

“It’s a nice illustration of how, really, immune response is so highly connected to the metabolic state of the cell,” says Gökhan S. Hotamisligil of Harvard University’s T.H. Chan School of Public Health who was not involved in the study. “The immune system launches its responses commensurate with the sources of nutrients and energy from the environment,” he adds in an email.

There are still missing pieces in the path from high-fat diet to increased Acaca expression to ACC1’s influence on T-cell differentiation. It also remains to be seen how this plays out in obese humans, although Nakayama and colleagues did show that inhibiting ACC1 reduced pathologic Th17 generation in human immune cell cultures, and that the T cells of obese humans contain elevated levels of ACC1 and show signs of increased differentiation into Th17 cells.


The prevalence of obesity has been increasing worldwide, and obesity is now a major public health problem in most developed countries (Gregor and Hotamisligil, 2011, Ng et al., 2014). Obesity-induced inflammation contributes to the development of various chronic diseases, such as autoimmune diseases, metabolic diseases, and cancer (Kanneganti and Dixit, 2012, Kim et al., 2014,Osborn and Olefsky, 2012, Winer et al., 2009a). A number of studies have pointed out the importance of reciprocal interactions between metabolic systems and immune cells in the pathogenesis of obesity-associated diseases (Kaminski and Randall, 2010, Kanneganti and Dixit, 2012, Kim et al., 2014, Mauer et al., 2014, Stienstra et al., 2012, Winer et al., 2011).

Elucidating the molecular mechanisms by which naive CD4 T cells differentiate into effector T cells is crucial for understanding helper T (Th) cell-mediated immune pathogenicity. After antigen stimulation, naive CD4 T cells differentiate into at least four distinct Th cell subsets: Th1, Th2, Th17, and inducible regulatory T (iTreg) cells (O’Shea and Paul, 2010, Reiner, 2007). Several specific master transcription factors that regulate Th1/Th2/Th17/iTreg cell differentiation have been identified, including T-bet for Th1 (Szabo et al., 2000), GATA3 (Yamashita et al., 2004, Zheng and Flavell, 1997) for Th2, retinoic-acid-receptor-related orphan receptor γt (RORγt) for Th17 (Ivanov et al., 2006), and forkhead box protein 3 (Foxp3) for iTreg (Sakaguchi et al., 2008). The appropriate expression and function of these transcription factors is essential for proper immune regulation by each Th cell subset.

Among these Th cell subsets, Th17 cells contribute to the host defense against fungi and extracellular bacteria (Milner et al., 2008). However, the pathogenicity of IL-17-producing T cells has been recognized in various autoimmune diseases, including multiple sclerosis, psoriasis, inflammatory bowel diseases, and steroid-resistant asthma (Bettelli et al., 2006, Coccia et al., 2012, Ivanov et al., 2006,Leonardi et al., 2012, McGeachy and Cua, 2008, Nylander and Hafler, 2012,Stockinger et al., 2007, Sundrud et al., 2009).

An HFD Promotes Th17 Cell Differentiation and Affects the Expression of Fatty Acid Enzymes in Memory CD4 T Cells In Vivo

Inhibition of ACC1 Function Results in Decreased Th17 Cell Differentiation and Ameliorates the Development of Autoimmune Disease

ACC1 Controls the Differentiation of Th17 Cells Both In Vitro and In Vivo

ACC1 Controls the Function, but Not Expression, of RORγt in Differentiating Th17 Cells

Extrinsic Fatty Acid Supplementation Restored Acaca−/− Th17 Cell Differentiation through the Functional Improvement of RORγt

Obese Subjects Show Upregulation of ACACA and Increased Th17 Cells in CD45RO+ Memory CD4 T Cells

We herein identified a critical role that ACC1 plays in Th17 cell differentiation and the pathogenicity of Th17 cells through the control of the RORγt function under obese circumstances. High-fat-induced obesity augments Th17 cell differentiation and the expression of enzymes involved in fatty acid metabolism, including ACC1. Pharmacological inhibition or genetic deletion of ACC1 resulted in impaired Th17 cell differentiation in both mice and humans. In contrast, overexpression of Acaca induced Th17 cells in vivo, leaving the expression ofIfng and Il4 largely unchanged. ACC1 modulated the binding of RORγt to theIl17a gene and the subsequent p300 recruitment in differentiating Th17 cells. Memory CD4 T cells from peripheral blood mononuclear cells (PBMCs) of obese subjects showed increased IL-17A production and ACACA expression. Furthermore, a strong correlation was detected between the proportion of IL-17A-producing cells and the expression level of ACACA in memory CD4 T cells in obese subjects. Thus, our findings provide evidence of a mechanism wherein obesity can exacerbate IL-17-mediated pathology via the induction of ACC1.

Read Full Post »

Peptides and anti-Cancer activity

Larry H. Bernstein, MD, FCAP, Curator



PE and PS Lipids Synergistically Enhance Membrane Poration by a Peptide with Anticancer Properties

Natália Bueno Leite, Anders Aufderhorst-Roberts, Mario Sergio Palma, Simon D. Connell, João Ruggiero Neto, Paul A. Beales
Biophys J 1 Sept 2015; 109(5):936–947.   DOI:
Polybia-MP1 (MP1) is a bioactive host-defense peptide with known anticancer properties. Its activity is attributed to excess serine (phosphatidylserine (PS)) on the outer leaflet of cancer cells. Recently, higher quantities of phosphatidylethanolamine (PE) were also found at these cells’ surface. We investigate the interaction of MP1 with model membranes in the presence and absence of POPS (PS) and DOPE (PE) to understand the role of lipid composition in MP1’s anticancer characteristics. Indeed we find that PS lipids significantly enhance the bound concentration of peptide on the membrane by a factor of 7–8. However, through a combination of membrane permeability assays and imaging techniques we find that PE significantly increases the susceptibility of the membrane to disruption by these peptides and causes an order-of-magnitude increase in membrane permeability by facilitating the formation of larger transmembrane pores. Significantly, atomic-force microscopy imaging reveals differences in the pore formation mechanism with and without the presence of PE. Therefore, PS and PE lipids synergistically combine to enhance membrane poration by MP1, implying that the combined enrichment of both these lipids in the outer leaflet of cancer cells is highly significant for MP1’s anticancer action. These mechanistic insights could aid development of novel chemotherapeutics that target pathological changes in the lipid composition of cancerous cells.

The antimicrobial peptide Polybia-MP1 (IDWKKLLDAAKQIL-NH2), or simply MP1, has unexpectedly been shown to exhibit selective inhibition against several types of cancerous cells and therefore could prove advantageous in the development of novel chemotherapies. Extracted from the Brazilian waspPolybia paulista, MP1 has a broad spectrum of bactericidal activities against Gram-negative and Gram-positive bacteria without being hemolytic and cytotoxic (1). Surprisingly, MP1 also selectively inhibits proliferating bladder and prostate cancer cells (2), and multidrug-resistant leukemic cells (3). Recently, it has been observed that this peptide is cytotoxic against leukemic T lymphocytes and very selective in recognizing these cells compared to healthy lymphocytes (4).

Cancer cell membranes are now known to lose the asymmetric transmembrane distribution of phospholipids that is observed in healthy cells (5, 6). In healthy mammalian cells, the anionic aminophospholipid PS (phosphatidylserine) is predominant in the inner membrane leaflet and zwitterionic phospholipids are predominant in outer membrane leaflet. In such cells, the phospholipid asymmetry is maintained by a family of aminophospholipid translocases that catalyze the transport of PS from the outer to the inner membrane leaflets (7). However, in apoptotic and cancer cells, PS is found to also be located in the outer monolayer of the plasma membrane in significant proportions (5, 6).

The molecular-scale mechanistic basis for MP1’s anticancer properties is yet to be established. Changes in the distribution and/or composition of lipids (e.g., PS) within the plasma membrane of malignant cells could be the origin of MP1’s cancer selectivity. This is a reasonable hypothesis, based upon the well-established selectivity of antimicrobial peptides for bacterial membranes over eukaryotic membranes due to their higher anionic lipid content (8, 9, 10, 11). Recently, the effect of PS on the pore-forming activity of MP1 was investigated by multiple techniques, namely, conductance measurements in planar bilayer lipid membranes, binding assays, and lytic activity on large unilamellar vesicles (4). Although an increase in affinity and lytic activity of MP1 for lipid vesicles containing PS was observed, MP1’s pore-formation activity in BLM showed no difference between PC (phosphatidylcholine) and mixed PC/PS bilayers. Significantly, it was recently reported that PE (phosphatidylethanolamine) lipids, naturally found on the inner plasma membrane of normal cells, are also externalized to the outer monolayer of the plasma membrane of apoptotic and tumor endothelial cells due to both PS and PE lipids being coregulated by the same transporters (7). These authors observed that the exposure to the outer monolayer of one of these phospholipids leads to the exposure of the other. Therefore, it is important for future work to establish the role of increased concentrations of both PE and PS lipids in the interaction of MP1 with membranes.

In this work, we address this challenge by establishing the roles of PE and PS lipids in the effects of MP1 on the structure and permeability of model membranes. Primarily, we study the permeability of giant unilamellar vesicles (GUVs) at the single vesicle level. Fluorescence confocal microscopy was used to determine the size-dependent macromolecular permeability of lipid membranes in GUV model systems by analyzing the influx of three fluorescent dyes with molecular masses of 0.37, 3.0, and 10.0 kDa into these vesicles (Fig. 1). We deconvolve the effects of PS and PE lipids by exploring their effects within DOPC (PC) membranes both separately and in combination: DOPC/POPS 80:20 (PC/PS), DOPC/DOPE 90:10 (PC/PE), and DOPC/DOPE/POPS 70:10:20 (PC/PE/PS). These experiments are corroborated by circular dichroism (CD) spectroscopy to quantify peptide binding to the membrane, fluorescence spectroscopy experiments to establish the leakage mechanism in an ensemble system of nanoscale large unilamellar vesicles (LUVs), and atomic-force microscopy (AFM) imaging of supported lipid bilayers to reveal the nanoscale perturbations of membrane structure induced by the peptide. By combining these approaches, we show that, while PS lipids significantly enhance MP1’s binding onto the membrane, PE lipids impart the most significant contribution to the rate and extent of membrane permeabilization by MP1, facilitating the opening of larger membrane defects than in bilayers lacking in PE.

Thumbnail image of Figure 1. Opens large image 1

Schematic representation of membrane disruption by peptides and the experimental system. The helical peptide Polybia-MP1 is shown according to the helical wheel projections. Amino acids: (blue) polar with positive net charge; (purple) polar with negative net charge; (red) polar noncharged; and (green) nonpolar. Confocal microscopy was performed to investigate the influx of three dyes with distinct sizes in GUVs in the presence and absence of PE lipids: 0.37 kDa CF (green), 3k-CB (blue), 10k-AF647 (magenta), and the scale bars correspond to 10 μm. Lipid membranes are labeled with Rh-DOPE (red). The peptide interacts with the GUVs, disturbs their structure, and then enables the passage of fluorescent dyes by formation of pore-like structures. To see this figure in color, go online.

Jump to Section
Materials and Methods
  Peptide synthesis and purification
  Mass spectrometry analysis
  GUV formation
  LUV preparation
  CD spectroscopy for binding isotherms
  Confocal microscopy
  Analysis of confocal images and movies
  ANTS/DPX requenching measurements
  Phase contrast microscopy
  PS lipids significantly enhance peptide binding to the membrane
  MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations
  Confirmation of the pore-formation hypothesis in lipid vesicles
  Synergistic enhancement of GUV leakage kinetics by PE and PS lipids
  PE lipids significantly enhance pore size and membrane permeability
  Direct imaging of peptide-induced pores by AFM
  Biophysical implications for MP1-lipid membrane interactions
  Implications for the chemotherapeutic potential of MP1 peptides
Author Contributions
Supporting Material


Jump to Section
Materials and Methods
  Peptide synthesis and purification
  Mass spectrometry analysis
  GUV formation
  LUV preparation
  CD spectroscopy for binding isotherms
  Confocal microscopy
  Analysis of confocal images and movies
  ANTS/DPX requenching measurements
  Phase contrast microscopy
  PS lipids significantly enhance peptide binding to the membrane
  MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations
  Confirmation of the pore-formation hypothesis in lipid vesicles
  Synergistic enhancement of GUV leakage kinetics by PE and PS lipids
  PE lipids significantly enhance pore size and membrane permeability
  Direct imaging of peptide-induced pores by AFM
  Biophysical implications for MP1-lipid membrane interactions
  Implications for the chemotherapeutic potential of MP1 peptides
Author Contributions
Supporting Material

PS lipids significantly enhance peptide binding to the membrane

The overall efficacy of a peptide at disrupting a target membrane can be broken down into the combination of two sequential steps: 1) binding of the peptide to the membrane surface, and 2) the efficiency of membrane disruption by the bound peptide resulting in membrane poration or leakage. First, we investigate the membrane binding isotherms of the MP1 peptide to our four lipid compositions of interest by CD spectroscopy by titrating a 10 μM MP1 solution with increasing lipid (LUV) concentrations (Fig. 2). Fitting these binding isotherms revealed that the partition coefficient (Kp) of the peptide was 7–8 times higher for membrane compositions containing PS (Kp values were PC 4600 M−1, PC/PE 4000 M−1, PC/PS 33,000 M−1, and PC/PE/PS 30,000 M−1). It is also interesting to note from this data that PE lipids slightly suppress peptide binding by a factor of ∼10%. Due to the cationic nature of MP1 (net charge of +2e), it is highly likely that the enhanced peptide binding to anionic-PS-containing membranes is primarily driven by electrostatic interactions.

large Image

Figure 2

Binding isoterms show that MP1 has a higher affinity for PS-containing membranes. The binding isoterms and the partition coefficients (Kp) obtained using CD by lipid titration at 10 μM MP1 solution. LUVs are composed of (a) PC, (b) PC/PE, (c) PC/PS, and (d) PC/PE/PS.

Jump to Section
Materials and Methods
  Peptide synthesis and purification
  Mass spectrometry analysis
  GUV formation
  LUV preparation
  CD spectroscopy for binding isotherms
  Confocal microscopy
  Analysis of confocal images and movies
  ANTS/DPX requenching measurements
  Phase contrast microscopy
  PS lipids significantly enhance peptide binding to the membrane
  MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations
  Confirmation of the pore-formation hypothesis in lipid vesicles
  Synergistic enhancement of GUV leakage kinetics by PE and PS lipids
  PE lipids significantly enhance pore size and membrane permeability
  Direct imaging of peptide-induced pores by AFM
  Biophysical implications for MP1-lipid membrane interactions
  Implications for the chemotherapeutic potential of MP1 peptides
Author Contributions
Supporting Material

MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations

To investigate the efficiency of membrane disruption, we measured the leakage of macromolecules across GUV model membranes by confocal fluorescence microscopy. Fluorescent passive leakage markers of different sizes were simultaneously employed: 0.37 kDa carboxyfluorescein (CF), 3 kDa dextran labeled with Cascade Blue (3k-CB), and 10 kDa dextran labeled with Alexa Fluor 647 (10k-AF647). GUVs were composed of PC, PC/PS, PC/PE, or PC/PE/PS. The dose-response of the membranes to the addition of MP1 was characterized for each membrane composition and passive leakage marker by evaluating the normalized fluorescence intensities of the probes in the intravesicular lumen of the GUVs after 30 min incubation time (Fig. 3 and Fig. S1 in the Supporting Material). Each data point in Figs. 3 and S1 shows the mean leakage of 50 individual GUVs from a minimum of two independent experiments. For determining the percentage of leaked vesicles (Fig. 3), a threshold of 20% leakage (normalized to the background probe concentration) was used to define a filled vesicle. Alternatively, this data can be analyzed in terms of the average leakage into GUVs as a percentage of the probe concentration in the external medium (Fig. S1).

large Image
Figure 3
Dose-response curves show increased leakage of PE containing GUVs at lower peptide concentrations. (a and c) Percentage of GUVs filled by CF (0.37 kDa) after 30 min incubation time with MP1. (b and d) Percentage of GUVs filled by 10k-AF647 after 30 min incubation time with MP1. All vesicles presenting >20% of dye entry were accounted as filled. The data is plotted as a function of (a and b) total peptide concentration and (c and d) the concentration of peptide bound to the membranes. Fifty GUVs were used to construct each data point. Vesicles are composed of PC, PC/PS, PC/PE, and PC/PE/PS. To see this figure in color, go online.

The integrity of membranes containing both PE and PS lipids is perturbed by lower concentrations of MP1 peptide than the other membrane compositions we investigated. PC/PE/PS GUVs show significant (40–65%) leakage to the CF probe at 0.4 and 1.2 μM MP1 concentrations, whereas other membrane compositions studied leaked <30% within this concentration range (Fig. S1a).

Larger pore defects, evidenced by leakage of the larger 10k-AF647 probe, are shown to be significantly enhanced in membranes containing 10% PE. Almost all GUVs (98%) containing PE lipids are observed to leak the 10k-AF647 probe when in the presence of 4.0 μM MP1, compared to <60% of GUVs for other membrane compositions at the same peptide concentration (Fig. 3b). This is the most significant enhancement in selective perturbation for specific lipid membrane compositions observed within the dose-response data in Figs. 3 andS1. At this MP1 concentration, membranes under native conditions would be susceptible to the leakage of biological macromolecules such as small proteins and RNAs.

Interestingly, we also plot the GUV leakage data as a function of the concentration of bound peptide on the membrane using the specific partition coefficients of the peptide for different lipid compositions that were calculated inFig. 2 (see also Figs. 3, c and d, and S1, c and d). This representation of the data clearly shows that PE lipids increase the susceptibility of PC membranes to disruption by the MP1 peptide, with PC/PE lipids leaking at significantly lower bound peptide concentrations. Due to the higher bound concentration of peptide to membranes containing PS lipids, this lipid decreases the apparent susceptibility of the membrane to leakage as observed by the onset of leakage shifting to higher bound peptide concentrations. For PC/PE/PS GUVs, the apparent competing effects of PE and PS lipids on the membrane’s leakage susceptibility roughly cancel each other out, leading to intermediate membrane disruption susceptibility for a given bound peptide concentration. However, the effect of increased bound peptide concentrations due to PS far outweighs its apparent inhibition of membrane leakage, making PC/PE/PS GUVs the most susceptible to leakage for a given total peptide concentration. Therefore, the combined roles of PS in increasing membrane binding and PE in increasing the susceptibility of the membrane are both important in increasing the membrane disruptive efficacy of MP1.

Jump to Section
Materials and Methods
  Peptide synthesis and purification
  Mass spectrometry analysis
  GUV formation
  LUV preparation
  CD spectroscopy for binding isotherms
  Confocal microscopy
  Analysis of confocal images and movies
  ANTS/DPX requenching measurements
  Phase contrast microscopy
  PS lipids significantly enhance peptide binding to the membrane
  MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations
  Confirmation of the pore-formation hypothesis in lipid vesicles
  Synergistic enhancement of GUV leakage kinetics by PE and PS lipids
  PE lipids significantly enhance pore size and membrane permeability
  Direct imaging of peptide-induced pores by AFM
  Biophysical implications for MP1-lipid membrane interactions
  Implications for the chemotherapeutic potential of MP1 peptides
Author Contributions
Supporting Material

Confirmation of the pore-formation hypothesis in lipid vesicles

Fluorescence spectroscopy experiments using LUVs give ensemble-averaged measurements with high statistics on a large population of vesicles, complementing single-vesicle GUV imaging experiments that inherently have lower statistics but yield information on the distribution of behaviors and rare events within a sample. The fluorescence requenching method (18) enables us to distinguish the type of leakage mechanism induced by MP1 for the lipid compositions under investigation. One possibility is the all-or-none mechanism where some vesicles release all of their internal contents while the others remain intact. This is attributed to pore-formation mechanisms of membrane perturbation, or complete vesicle lysis. Another possibility is the gradual leakage mechanism where vesicles only release a fraction of their encapsulated contents during a leakage event. This is associated with transient perturbations of the membrane. A fluorophore (ANTS) and a quencher (DPX) are encapsulated within lipid vesicles at high concentrations such that the fluorescence is initially quenched; vesicle leakage results in the release of both ANTS and DPX, but quenching is decreased due to dilution of these probes. The externalized ANTS fluorescence can be suppressed by additional titration of DPX such that the remaining fluorescence signal is only due to the ANTS inside intact vesicles. The data can be represented by a plot of the degree of quenching (Qin) against the released ANTS fraction (fout). In the case of an all-or-none leakage mechanism, the plot of Qin versus fout will show no dependence of Qin on fout. In contrast, the gradual leakage mechanism causes release of only a fraction of the encapsulated contents within individual vesicles and so Qin increases with increasing fout (18).

large Image

Figure 4

Fluorescence requenching assays for MP1 reveal all-or-none leakage in the four lipid compositions studied. Qin is constant as a function of fout for MP1, which is in agreement with the all-or-none mechanism of dye release.) (Lines) Theoretical curves for ideal graded and all-or-none dye release (18). To see this figure in color, go online.

Fig. 4 shows that the values of Qin remain constant with the increase of fout and the consequent increase of peptide/lipid molar ratios. This clearly shows that MP1 exhibits the all-or-none leakage mechanism for all lipid compositions studied, which is in contrast to what has been observed for antimicrobial peptides mastoparan X and mastoparan MP (19). We propose that this all-or-none leakage is related to peptide-induced pore formation (20, 21, 22, 23), where the vesicles are able to release all their internal contents through pore-like structures that are sufficiently long lived (23, 24, 25, 26, 27, 28). We do not solely attribute the all-or-none leakage to lysis of the vesicles because nonlysed, leaky vesicles are observed in our GUV experiments (Figs. 3 and S1). However, we do not discount the possibility that lysis might play a role in the LUV leakage at the highest peptide concentrations used in this assay. Furthermore, pore-like activity of MP1 has previously been identified from electrophysiology measurements in planar lipid bilayers composed of phytanoyl-PC and phytanoyl-PC/PS (70:30) (4).

Our fluorescence requenching results show that stable pores form with a lifetime that persists long enough for the dye efflux to reach equilibrium in LUV systems. However, this does not discount the possibility that pores might be transient over longer timescales, for example during the leakage of much larger vesicles such as GUVs where the encapsulated volume of dye that needs to be released during a leakage event is ∼106 times greater than for the LUV model system. Indeed, we will see some evidence for transient pore events and dynamic changes in membrane permeability in the single GUV leakage kinetics data that follows. Nevertheless, all-or-none leakage is clearly evident in GUVs after 30 min incubation with 1.2 and 4.0 μM MP1. Leakage histograms of the individual GUVs (an alternative representation of data shown in Figs. 3 and S1) predominantly show either Math Eq20% (unleaked) or Math Eq80% (fully leaked) leakage (Fig. S2).
Jump to Section
Materials and Methods
  Peptide synthesis and purification
  Mass spectrometry analysis
  GUV formation
  LUV preparation
  CD spectroscopy for binding isotherms
  Confocal microscopy
  Analysis of confocal images and movies
  ANTS/DPX requenching measurements
  Phase contrast microscopy
  PS lipids significantly enhance peptide binding to the membrane
  MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations
  Confirmation of the pore-formation hypothesis in lipid vesicles
  Synergistic enhancement of GUV leakage kinetics by PE and PS lipids
  PE lipids significantly enhance pore size and membrane permeability
  Direct imaging of peptide-induced pores by AFM
  Biophysical implications for MP1-lipid membrane interactions
  Implications for the chemotherapeutic potential of MP1 peptides
Author Contributions
Supporting Material

Synergistic enhancement of GUV leakage kinetics by PE and PS lipids

Analysis of the time delay from the start of our GUV experiments (addition of the peptide) to observations of the onset of GUV leakage reveals a synergistic reduction in this lag time for PC/PE/PS membranes (Table 1 and Fig. 5). In these GUV experiments, we add 4.0 μM MP1 to our samples and monitor the time taken for initial leakage events of GUVs to the 0.37 kDa CF probe to occur (t0tCF). This MP1 concentration is chosen as it causes significant leakage of GUVs within 30 min of peptide addition across all four lipid compositions of interest. The onset of leakage occurs approximately twice as quickly for PC/PE/PS GUVs than for other membrane compositions, with only a very slight reduction of the lag time for PC/PS membranes compared with PC/PE and PC GUVs. Therefore, this is not a purely electrostatic effect from the increased rate and extent of peptide binding to anionic PS-containing membranes; it also requires the presence of PE to significantly increase the susceptibility of the membrane to permeabilization.

Table 1Lag times between the onset of leakage of each dye and the time interval before the initial leakage takes place after the addition of peptide
Time Delays (s) PC/PE PC/PE/PS PC/PS PC
t0tCF 1600 ± 110 760 ± 120 1400 ± 60 1600
tCFt3k-CB 1.8 ± 0.6 1.5 ± 0.3 41 ± 5a 160 ± 110
tCFt10k-AF647 4.2 ± 1.5 2.0 ± 0.4 52 ± 6 220 ± 66
t3k-CBt10k-AF647 2.7 ± 0.9 1.4 ± 0.4 9.6 ± 1.2 60 ± 42

The errors represent the standard deviation of the observed GUV data set.

aFor PC/PS GUVs, the tCFt3k-CB data only includes GUVs that leaked to all three dyes; these sample conditions contained two distinct populations of GUV leakage behaviors where a second population of GUVs only leaked to the CF and 3k-CB dyes with a time delay of tCFt3k-CB = 4.8 ± 0.6 s.
large Image
Figure 5
GUV permeabilization kinetics are synergistically enhanced by PE and PS lipids. (a) Comparison between dye influx kinetics of three distinct dyes (CF-0.37kDa, 3k-CB, and 10k-AF647) for PC/PE/PS and PC/PS GUVs. The time axis represents the time after peptide addition. These individual GUV leakage profiles were chosen as they represent the average behavior of the GUVs observed under these conditions. (b) Schematic representation of the dye influx kinetics for PC/PS and PC/PE/PS GUVs in the presence of CF and 10k-AF647 passive leakage markers. This shows the average lag times for GUV leakage after the addition of 4.0 μM MP1 and the typical average leakage extent of the GUVs that resulted in these experiments. To see this figure in color, go online.
We also quantify the average delay times between leakage of the different-sized fluorescent probes between CF and 3k-CB (tCFt3k-CB), CF and 10k-AF647 (tCFt10k-AF647), and 3k-CB and 10k-AF647 (t3k-CBt10k-AF647). Once the initial leakage event occurs, PE-containing GUVs rapidly become leaky to fluorescent probes of larger sizes (3 and 10 kDa). For PC/PE/PS and PC/PE membranes, GUVs become leaky to larger 3k-CB, then 10k-AF647 passive leakage markers within seconds of permeabilization to the smallest CF (0.37 kDa) probe (Table 1). The consecutive delay times between CF and 3k-CB probes and 3k-CB and 10k-AF647 probes were approximately an order-of-magnitude longer for PC/PS membranes, and almost two orders-of-magnitude longer for purely PC membranes. This strongly implies that the presence of PE significantly enhances the favorability and rate of formation of larger membrane defects or pores.
Jump to Section
Materials and Methods
  Peptide synthesis and purification
  Mass spectrometry analysis
  GUV formation
  LUV preparation
  CD spectroscopy for binding isotherms
  Confocal microscopy
  Analysis of confocal images and movies
  ANTS/DPX requenching measurements
  Phase contrast microscopy
  PS lipids significantly enhance peptide binding to the membrane
  MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations
  Confirmation of the pore-formation hypothesis in lipid vesicles
  Synergistic enhancement of GUV leakage kinetics by PE and PS lipids
  PE lipids significantly enhance pore size and membrane permeability
  Direct imaging of peptide-induced pores by AFM
  Biophysical implications for MP1-lipid membrane interactions
  Implications for the chemotherapeutic potential of MP1 peptides
Author Contributions
Supporting Material

PE lipids significantly enhance pore size and membrane permeability

We use time-series confocal microscopy imaging to quantify the membrane permeability of GUVs during initial leakage. Quantification of the fluorescence intensity of the leakage markers in the intravesicular and extravesicular medium allows us to calculate the fractional leakage of individual GUVs as a function of time. The leakage kinetics of individual GUVs are monitored for up to 30 min after the addition of 4.0 μM MP1. This concentration is chosen as all lipid compositions show significant leakage within 30 min; a higher MP1 concentration of 10 μM is observed to induce significant lysis of GUV samples (Fig. S3). These experiments are conducted on GUVs of all four membrane compositions under investigation, using the CF, 3k-CB, and 10k-AF647 leakage markers simultaneously. This allows the time evolution of membrane permeability to different molecular sizes to be simultaneously measured for individual GUVs (Figs. 5a and S4). To the best of our knowledge, this is the first example of simultaneous size-dependent permeability measurements in GUVs for three different-sized leakage markers.

Typical leakage kinetic profiles for different membrane compositions and probe sizes are shown in Figs. S4 and 5a. It can be qualitatively seen from these example profiles that membrane compositions containing 10% PE exhibit full and rapid membrane leakage for all three sizes of leakage marker, consistent with the leakage kinetics data in Table 1, which is also outlined in Fig. 5b. For membrane compositions lacking PE, the leakage rates can sometimes be seen to increase and decrease intermittently, sometime plateauing before full leakage is achieved; this is particularly evident in the leakage profile of a single PC/PS GUV shown in Fig. S5. We attribute these observations to membrane self-healing events, where the pores/defects reseal and the membrane regains its permeability barrier, followed by later phases of increased leakage. This is particularly observed for the larger 3k-CB and 10k-AF647 leakage probes. Therefore, the membrane permeability for PC and PC/PS GUVs, in particular, can change dynamically during the observed leakage events; this is a result of the competition between the lipid bilayer and peptides in maintaining their barrier properties and inducing membrane pores, respectively (Fig. S5).

large Image

Figure 6

PE lipids facilitate much greater membrane permeability in GUV membranes. (a) Typical log-linear plot of time-dependent dye influx: −R/3ln(1−c) (105) versus time, for the three dyes in a single GUV of PC/PS. (b) Distributions of the obtained permeabilities in single GUVs composed of PC/PS. (c) Typical log-linear plot of time-dependent dye influx: −R/3ln(1−c) (10−5) versus time, for the three dyes in a single GUV of PC/PE/PS. (d) Distributions of the obtained permeabilities in single GUVs composed of PC/PE/PS. The permeabilities are obtained from the slopes of the log-linear plots of the time-dependent influx of dyes into single GUVs. To see this figure in color, go online.

Our leakage kinetic profiles were used to calculate the membrane permeability to the different-sized probes using a diffusional model for membrane translocation (29); the membrane permeability is the gradient of the log-linear plot as seen in the example data in Fig. 6, a and c. Average permeability values for each membrane composition to each probe size during the initial leakage events are shown in Table 2. It can be seen that, for all membrane compositions tested, average permeability decreases with increasing probe size. However, the most significant finding from this data is the large, one-order-of-magnitude increase, in membrane permeability for membrane compositions containing 10% PE. This can be observed for all three leakage markers studied. It can also be seen that the presence of PS in the membrane imparts a modest, but significant, increase in permeability on the membranes upon perturbation by MP1. This effect can be seen further in Fig. 6, b and d, which show the distributions of permeability measurements for PC/PS and PC/PE/PS GUVs to the CF and 10k-AF647 leakage markers, respectively. For both probe sizes, the majority of permeability measurements for PC/PS membranes were in the 0–25 nm/s range, whereas when PE was included in the membrane formulations, a large proportion of permeability measurements were >500 nm/s.

Table 2Average permeability values (〈Pm〉) and the average fractional permeated area per vesicle (〈Ap〉/〈Av〉) obtained from the average permeability values for each probe size and membrane composition
PmCF(nm/s) Ap〉/〈AvCF(106) Pm3k-CB(nm/s) Ap〉/〈Av3k-CB(106) Pm10k-AF647(nm/s) Ap〉/〈Av10k-AF647 (106)
PC 46 ± 14 0.45 ± 0.14 17 ± 6 0.50 ± 0.16 8 ± 2 0.44 ± 0.08
PC/PS 59 ± 12 0.58 ± 0.12 29 ± 6 0.90 ± 0.20 23 ± 4 1.30 ± 0.20
PC/PE 466 ± 143 4.60 ± 1.40 207 ± 102 6.50 ± 3.40 158 ± 53 8.80 ± 2.90
PC/PE/PS 589 ± 142 5.80 ± 1.40 333 ± 73 10.40 ± 2.80 169 ± 52 9.40 ± 2.90

The errors represent the standard deviation of the observed GUV data set.

It should be noted that the observed permeability distributions (Fig. 6, b and d) are broad due to the fact that peptide-induced pores do not have well-defined structures, pore formation events are stochastic, and the membrane interfaces are fluid, giving rise to this wide distribution of individual permeability events when measured at the single vesicle level. Indeed, it has previously been reported that the initial pores that form during peptide-induced pore formation might be far from equilibrium and can, for example, relax to a smaller size over longer timescales as has been observed for the peptides Bax-α5 (23) and magainin 2 (30).

The permeability data was used to calculate the effective fractional permeable area of the membrane for each probe size using the expression (29)

Math Eq

where Ap is the permeable area of membrane on a GUV, Av is the total area of the vesicle, Pm is the permeability, and δ is the thickness of the membrane. The Stokes-Einstein diffusion constant of the leakage marker is D0 = kT/6πηR0, where kT is the thermal energy, η is the solvent viscosity, and R0 is the hydrodynamic radius of the fluorescent dye that was estimated with the relationR0 = 0.0332(Mw)0.463 in nanometers (31); Mw is the molecular weight of the dye. A brief derivation of this equation is presented in the Supporting Material. It should be noted that this equation is most accurate for the formation of large membrane pores as it assumes that the diffusion constant of the dye within the pore is the same as its diffusion constant in bulk solution. However, we believe this to be a reasonable assumption because these passive leakage markers will have a very short residence time within the pore itself due to the bilayer only being ∼5-nm thick; these solutes are not expected to interact strongly with the membrane itself.

Values of the fractional permeable areas are shown in Table 2. The fractional permeable areas were also found to be an order-of-magnitude greater for membrane compositions containing PE than for those that did not. Note that slightly larger permeable areas were measured for the larger leakage markers; these represent a later time point in the membrane disruption of GUVs by MP1 as the smaller leakage markers translocate the membrane at earlier times (Table 1). This extended delay time therefore allows for a greater area of membrane disruption to occur before the initiation of leakage to the larger Mw dyes.

Besides the order-of-magnitude increase in membrane permeabilization in the presence of PE lipids, we found an interesting correlation between PE content and membrane morphological response to MP1. Without PE, PC and PC/PS GUVs exhibited bright spots of fluorescent lipids at specific locations on the membrane surface in the presence of 4.0 μM MP1 (Fig. 7). We attribute these observations to local aggregation of peptides and lipids at the GUV surface. These peptide-induced lipid aggregates may be in competition with the pore-/defect-forming activity of the peptides. Such dense lipid structures were not seen on GUVs containing PE (PC/PE and PC/PE/PS) upon introduction of the peptide. Therefore, we speculate that the PE suppresses the intramembrane lipid aggregation by more easily facilitating the poration of the membrane.

large Image
Figure 7

Lipid aggregation is observed within the membranes of GUVs lacking in PE. Images of local lipid aggregation at the GUV surface (bright localized spots of fluorescence) seen after peptide addition (Cp = 4.0 μM). This effect is frequently observed in PC and PC/PS GUVs, but not for the lipid mixtures containing 10 mol % PE. To see this figure in color, go online.

While localized lipid aggregation was not observed on the surface of the PE-containing GUVs, these GUVs were observed to decrease in diameter by ∼10–15% over a period of ∼1 h after peptide addition (Fig. S6). Contrary to this, PC and PC/PS GUVs remained at a constant size for up to 2 h after addition of MP1. Therefore, MP1 results in the significant loss of lipid from only those GUVs that contain PE lipids.

Jump to Section
Materials and Methods
  Peptide synthesis and purification
  Mass spectrometry analysis
  GUV formation
  LUV preparation
  CD spectroscopy for binding isotherms
  Confocal microscopy
  Analysis of confocal images and movies
  ANTS/DPX requenching measurements
  Phase contrast microscopy
  PS lipids significantly enhance peptide binding to the membrane
  MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations
  Confirmation of the pore-formation hypothesis in lipid vesicles
  Synergistic enhancement of GUV leakage kinetics by PE and PS lipids
  PE lipids significantly enhance pore size and membrane permeability
  Direct imaging of peptide-induced pores by AFM
  Biophysical implications for MP1-lipid membrane interactions
  Implications for the chemotherapeutic potential of MP1 peptides
Author Contributions
Supporting Material

Direct imaging of peptide-induced pores by AFM

AFM imaging of supported lipid bilayers confirms the role of PE in potentiating the formation of larger transmembrane pores. MP1 was added at 10 μM concentration to induce significant pore formation on the relatively small patches of membrane imaged by the AFM within a reasonable experimental timescale (<2 h); resultant pores/defects were observed to be much larger in PC/PE/PS membranes (250 ± 110 nm in diameter) compared to PC/PS (54 ± 30 nm in diameter) (Fig. 8). Similar-sized transmembrane pores were observed in PC/PE/PS and PC/PE membranes (290 ± 200 nm in diameter), but significantly fewer defects formed in PC/PE membranes. Note that the large standard deviations in these average pore diameters represent a significant size polydispersity in the defects formed. No pores were evident in PC membranes 2 h after peptide addition (Fig. S7); however, pores would need to be several nanometers in diameter to be observable by AFM, considerably larger than those that can be detected by passive dye influx into the GUVs we used to investigate the early stages of GUV poration.

Our AFM studies also clearly show a difference in pore formation and growth mechanisms dependent on the presence of PE. The large transmembrane pores in PC/PE/PS and PC/PE membranes are seen to grow by the stepwise loss of lipid aggregates from the edge of the pore, implying that vesicle micellization is important for pore growth in the these membranes (Fig. S8). This is consistent with the small decrease (within experimental error) in GUV size observed for PE-containing GUVs by phase contrast microscopy (Fig. S6). Conversely, in PC/PS membranes, raised areas of lipid are first seen to form on the membrane (Fig. S9), which may correlate to the dense lipid structures observed in Fig. 7. These raised areas of membrane later evolve into comparatively small pores; many of the defects seen in Fig. 8 (bottom left) only span half the bilayer, with only the center of a few of these defects showing full bilayer pores (Fig. S10). This indicates that pores in these membranes may form via a half-membrane intermediate state. Finally the timescale for observation of membrane defects by AFM was much faster for PC/PE/PS membranes than for other lipid mixtures, with defects observed almost immediately after peptide addition (Fig S7), compared with a few tens of minutes for other mixtures.


Jump to Section
Materials and Methods
  Peptide synthesis and purification
  Mass spectrometry analysis
  GUV formation
  LUV preparation
  CD spectroscopy for binding isotherms
  Confocal microscopy
  Analysis of confocal images and movies
  ANTS/DPX requenching measurements
  Phase contrast microscopy
  PS lipids significantly enhance peptide binding to the membrane
  MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations
  Confirmation of the pore-formation hypothesis in lipid vesicles
  Synergistic enhancement of GUV leakage kinetics by PE and PS lipids
  PE lipids significantly enhance pore size and membrane permeability
  Direct imaging of peptide-induced pores by AFM
  Biophysical implications for MP1-lipid membrane interactions
  Implications for the chemotherapeutic potential of MP1 peptides
Author Contributions
Supporting Material

Biophysical implications for MP1-lipid membrane interactions

We have shown a synergistic enhancement of the rate and extent of membrane permeabilization by MP1 peptides when PE and PS lipids are present in the lipid membrane. This picture is confirmed and corroborated by complementary experiments using three different model membrane systems: LUVs, GUVs, and planar-supported bilayers. We consider the perturbation of the membrane by MP1 peptides in two steps: 1) binding of the peptides to the membrane, and 2) perturbation of the bilayer structure by bound peptides to induce leakage.

Binding isotherms (Fig. 2) reveal that PS lipids cause a 7–8-fold increase in peptide bound to the membrane. This strongly outweighs the small ∼10% reduction in bound peptide concentration caused by the PE lipids. Therefore, we find that the dominant role of PS lipids’ contribution to the membrane disruption by MP1 is a large increase in peptide binding to the membrane.

The role of PE lipids in MP1-induced membrane disruption is twofold: 1) PE increases the susceptibility of the membrane to permeabilization by bound peptides, and 2) PE facilitates the formation of larger transmembrane pores. First, when the extent of GUV leakage is normalized to bound peptide concentration in the dose-response curves in Fig. 3, c and d, it can be seen that 4–5 times lower bound peptide concentration is required to induce a similar leakage response compared to comparable GUVs without PE lipids. Second, GUV and AFM experiments corroborate the effect of larger pores forming in the presence of PE. Quantitative analysis of GUV leakage profiles in Fig. 6 andTable 2 reveal that the presence of PE increases the permeability of membranes by an order of magnitude compared to membranes lacking in PE. Furthermore, once pores formed in GUVs, they quickly (within seconds) grew large enough in size to allow larger macromolecules (3 and 10 kDa) to permeate the membrane (Table 1); this compared to several tens of seconds for larger pores to form in GUVs lacking PE. Crucially, the formation of larger pores for PE-containing membranes is directly visualized by AFM (Fig. 8), where the observed pore diameters are ∼5 times larger in the presence of PE (and hence ∼20–30 times larger in average pore area, consistent with the order-of-magnitude increase in permeability reported for the GUVs).

The formation of transmembrane pores was confirmed by complementary experimental systems and techniques. Rapid translocation of membrane-impermeable leakage markers across GUV membranes, an all-or-none LUV fluorescence leakage assay, and direct visualization of transmembrane defects by AFM imaging of planar bilayers, all confirm this to be true. While these pores are fairly long lived, the membranes were sometimes observed to temporarily reseal, regaining their barrier properties. This can clearly be seen in the leakage profiles of individual GUVs in Figs. 5a, S4, and S5. GUV and planar bilayer imaging experiments also strongly suggest differences in the mechanism of pore formation depending on whether PE lipids are present. Images of GUVs that did not contain PE lipids often exhibited bright spots of increased local lipid concentrations on the membrane, which we interpret to be local aggregation of peptides and lipid (Fig. 7). Similarly, AFM images showed locally raised regions of lipid scattered across the membrane for these lipid compositions (Fig. S9) before the formation of pores (Fig. S10). This contrasted to the pore-formation mechanism observed in the presence of PE, where local aggregates were not directly observed on the GUV surface and time-resolved AFM imaging showed pore growth to occur by the stepwise micellization and loss of lipid from the edge of the pores (Fig. S8).

Besides the increased binding due to PS and the increased membrane susceptibility and pore size due to PE, the synergistic enhancement of membrane disruption facilitated by these lipids can be observed in the kinetics of initial permeabilization events. GUV experiments showed that PC/PE/PS GUVs leaked a factor-of-two quicker than other membrane compositions (Table 1). This is again corroborated by the AFM studies where defects were observed in PC/PE/PS membranes almost immediately after peptide addition, whereas perturbations of other membrane compositions took a few 10 s of minutes to evolve. The complementary pore-promoting effects of PS on bound peptide concentrations and PE on membrane susceptibility far outweigh their slight inhibitory effects on each other’s roles (PE causes a slight reduction in binding affinity (Fig. 2) and PS causes a decrease in the membrane susceptibility to bound peptide (Fig. 3, c and d)). This is apparent from the effects of MP1 on GUVs, where PC/PE/PS membranes experience the greatest membrane perturbation for any given total peptide concentration (Fig. 3, a and b) and the larger number of pores observed on the membrane surface by AFM (Fig. 8). Therefore, our combined results provide a detailed mechanistic picture of the importance of increased PS and PE lipid concentrations in synergistically enhancing the membrane’s propensity for significant disruption of its barrier properties by MP1 peptides.

Variations in lipid composition are responsible for differences in membrane properties such as charge, fluidity, lateral pressure profiles, and mechanical moduli. Changes in these fundamental membrane properties directly affect their interactions with extraneous compounds, such as antimicrobial peptides. The cationic nature of the MP1 peptide is likely an important component in the initial step of peptide action, in which the peptide recognizes the target membrane due to electrostatic interactions and binds to it in a structured form, most of the time as a helix. Therefore, the inclusion of anionic PS lipids in these membranes increases these electrostatic interactions with the MP1 peptide (net charge = +2e). However, MP1-membrane interactions cannot be solely driven by electrostatics as these peptides were also found to disrupt zwitterionic PC and PC/PE membranes, likely through secondary hydrophobic initial binding interactions that lead to a significantly lower bound concentration of peptide compared to the anionic membranes.

Next, insertion of the peptide into the bilayer likely takes place due to the hydrophobic effect, where nonpolar MP1 residues insert into the bilayer core, and defects may then be opened within the membrane structure, leading to its disruption. Furthermore, taking account of the fact that MP1 is a short peptide (14 amino acids) and hence not long enough to form a bilayer-spanning barrel stave pore (9, 32), we anticipate that these pores will be disorganized toroidal structures formed by lipids and peptides, as described by many molecular-dynamics studies (33, 34). PE is known to significantly modulate the lateral pressure profile through membranes and thereby induce negative curvature stress in the bilayer. Negative curvature stress has been shown to enhance the formation of toroidal lipid pores within a membrane by stabilizing the curvature of these structures (35). Therefore, PE would be expected to favor the formation of pore-like defects in the membrane, consistent with the increase susceptibility of these membranes to MP1-induced poration and the order-of-magnitude increase in membrane permeability that we find for PE-containing membranes upon interaction with MP1 peptides.

Jump to Section
Materials and Methods
  Peptide synthesis and purification
  Mass spectrometry analysis
  GUV formation
  LUV preparation
  CD spectroscopy for binding isotherms
  Confocal microscopy
  Analysis of confocal images and movies
  ANTS/DPX requenching measurements
  Phase contrast microscopy
  PS lipids significantly enhance peptide binding to the membrane
  MP1 dose-response studies reveal that PE and PS lipids enhance membrane permeability at lower peptide concentrations
  Confirmation of the pore-formation hypothesis in lipid vesicles
  Synergistic enhancement of GUV leakage kinetics by PE and PS lipids
  PE lipids significantly enhance pore size and membrane permeability
  Direct imaging of peptide-induced pores by AFM
  Biophysical implications for MP1-lipid membrane interactions
  Implications for the chemotherapeutic potential of MP1 peptides
Author Contributions
Supporting Material

Implications for the chemotherapeutic potential of MP1 peptides

The MP1 peptide has been shown to have selective inhibition against numerous cancer lines compared to healthy cells (2, 3). Such malignant cells are also known to have increased expression of PS and PE lipids on their outer plasma membrane (5, 6, 7). This study strongly correlates the enhanced tumor inhibitory effects of these peptides with this pathological change in plasma membrane lipid composition, where the upregulation of PS and PE lipids can synergistically enhance the membrane-permeabilizing activity of MP1. This membrane permeabilization is likely to be the primary mechanism of cancer cell death induced by these peptides.

This suggests that MP1 might be a candidate therapeutic for development of novel cancer therapies, or at least guide the development of novel lead compounds for treatment of these diseases. One challenge for the application of antimicrobial peptides in medicine is that they often do not show high enough selectivity to their target cells to result in a favorable therapeutic index for these compounds (36). However, MP1 does not exhibit hemolytic activity to rat erythrocytes but presents chemotaxis for polymorphonucleated leukocytes and potent antimicrobial action against Gram-positive and Gram-negative bacteria (12), suggesting it could have favorable selectivity. It may also be of interest to test MP1 in a combination therapy with other chemotherapeutics that have intracellular targets. The selectivity of the MP1 peptide to disrupt the membranes of cancer cells may act synergistically with these other drugs to significantly enhance the therapeutic potency. Therefore, the therapeutic potential of this and other membrane-active peptides within the field of oncology is worthy of further investigation.

Read Full Post »

Obesity Variant Circuitry

Larry H. Bernstein, MD, FCAP, Curator



FTO Obesity Variant Circuitry and Adipocyte Browning in Humans

Melina Claussnitzer,  Simon N. Dankel, Kyoung-Han Kim,  Gerald Quon,  Wouter Meuleman,  Christine Haugen,  Viktoria Glunk,  Isabel S. Sousa, et al.

N Engl J Med 2015; 373:895-907  Sept 3, 2015    DOI: http://dx.10.org1/056/NEJMoa1502214


Genomewide association studies can be used to identify disease-relevant genomic regions, but interpretation of the data is challenging. The FTO region harbors the strongest genetic association with obesity, yet the mechanistic basis of this association remains elusive.

Full Text of Background…



We examined epigenomic data, allelic activity, motif conservation, regulator expression, and gene coexpression patterns, with the aim of dissecting the regulatory circuitry and mechanistic basis of the association between the FTO region and obesity. We validated our predictions with the use of directed perturbations in samples from patients and from mice and with endogenous CRISPR–Cas9 genome editing in samples from patients.

Full Text of Methods…



Our data indicate that the FTO allele associated with obesity represses mitochondrial thermogenesis in adipocyte precursor cells in a tissue-autonomous manner. The rs1421085 T-to-C single-nucleotide variant disrupts a conserved motif for the ARID5B repressor, which leads to derepression of a potent preadipocyte enhancer and a doubling of IRX3 and IRX5 expression during early adipocyte differentiation. This results in a cell-autonomous developmental shift from energy-dissipating beige (brite) adipocytes to energy-storing white adipocytes, with a reduction in mitochondrial thermogenesis by a factor of 5, as well as an increase in lipid storage. Inhibition of Irx3 in adipose tissue in mice reduced body weight and increased energy dissipation without a change in physical activity or appetite. Knockdown of IRX3 or IRX5 in primary adipocytes from participants with the risk allele restored thermogenesis, increasing it by a factor of 7, and overexpression of these genes had the opposite effect in adipocytes from nonrisk-allele carriers. Repair of the ARID5B motif by CRISPR–Cas9 editing of rs1421085 in primary adipocytes from a patient with the risk allele restored IRX3 and IRX5 repression, activated browning expression programs, and restored thermogenesis, increasing it by a factor of 7.


Effect of the FTO Locus on IRX3 and IRX5 in Human Adipocyte Progenitor Cells

To identify the cell types in which the causal variant may act, we examined chromatin state maps15,16 of the FTO obesity region across 127 cell types. An unusually long enhancer (12.8 kb) in mesenchymal adipocyte progenitors indicated a major regulatory locus (Figure 1B; and Fig. S1A, S1B, and S1C in the Supplementary Appendix). Haplotype-specific enhancer assays showed activity in association with the risk haplotype that was 2.4 times as high as that associated with the nonrisk haplotype in human SGBS adipocytes (i.e., adipocytes derived from a patient with the Simpson–Golabi–Behmel syndrome), which indicated genetic control of enhancer activity (Figure 1C). Enhancers in brain cells and other cell types were considerably shorter than those in mesenchymal adipocyte progenitors and lacked allelic activity (Fig. S1C and S1D in the Supplementary Appendix).


Figure 1. Activation of a Superenhancer in Human Adipocyte Progenitors by the FTO Obesity Risk Haplotype.

Panel A shows the genetic association with body-mass index (BMI) for all common FTO locus variants,14 including the reported single-nucleotide variant (SNV) rs1558902 (red diamond) and the predicted causal SNV rs1421085 (red square). Gray shading delineates consecutive 10-kb segments. CEU denotes a population of Utah residents with northern and western European ancestry, and LD linkage disequilibrium. Panel B shows chromatin state annotations for the locus across 127 reference epigenomes (rows) for cell and tissue types profiled by the Roadmap Epigenomics Project.15,16 For information on the colors used to denote chromatin states, see Figure S1A in the Supplementary Appendix. Vertical lines delineate the consecutive 10-kb segments shown in Panel A. ESC denotes embryonic stem cell, HSC hematopoietic stem cell, and iPSC induced pluripotent stem cell. Panel C shows human SGBS adipocyte enhancer activity, for 10-kb tiles, of the risk and nonrisk haplotypes with the use of relative luciferase expression. The boxes indicate means from seven triplicate experiments, and T bars indicate standard deviations.

To predict putative target genes, we examined large domains that had long-range three-dimensional chromatin interactions surrounding FTO and identified eight candidate genes (Figure 2A and 2B)



Figure 2. Activation of IRX3 and IRX5 Expression in Human Adipocyte Progenitors by the FTO Obesity Risk Genotype.

Panel A shows gene annotations and LD with array tag variant rs9930506 in a 2.5-Mb window; LD is expressed as r2 values in the CEU population. Arrows indicate the direction of transcription of annotated genes in the locus. Panel B shows chromosome conformation capture (Hi-C) interactions contact probabilities in human IMR90 myofibroblasts,22 revealing a 2-Mb topologically associating domain, and LD mean r2 statistics for all SNV pairs at 40-kb resolution. Panel C shows box plots for expression levels, after 2 days of differentiation, in human adipose progenitors isolated from 20 risk-allele carriers and 18 nonrisk-allele carriers, evaluated by means of a quantitative polymerase-chain-reaction analysis for all genes in the 2.5-Mb locus. The horizontal line within each box represents the median, the top and bottom of each box indicate the 75th and 25th percentile, and I bars indicate the range.

Among them, the developmental regulators IRX3 and IRX5 had genotype-associated expression, which indicated long-range (1.2-Mb) genetic control in primary preadipocytes (Figure 2C). Genotype-associated expression was not observed in whole-adipose tissue, a finding consistent with previous reports23,24; this indicated that the effect was cell type–specific and restricted to preadipocytes, which represent a minority of cells in adipose tissue (Fig. S2A in the Supplementary Appendix).


Effect of the FTO Locus on Mitochondrial Thermogenesis and Lipid Storage

To identify the biologic processes affected by altered IRX3 and IRX5expression in adipocytes, we used genomewide expression patterns in brown adipocyte–containing perirenal adipose tissue from a separate cohort of 10 nongenotyped, healthy kidney donors to identify genes with expression that was positively or negatively correlated with IRX3 and IRX5 expression. Genes that are associated with mitochondrial functions were found to have a negative correlation with IRX3 and IRX5, and genes with FXR and RXR lipid-metabolism functions were found to have a positive correlation, which suggests thatIRX3 and IRX5 may play roles in energy dissipation and storage


Figure 3A


Regulation of Obesity-Associated Cellular Phenotypes in Human Adipocytes by IRX3and IRX5., and Table S1 in the Supplementary Appendix). IRX3 and IRX5 had consistently higher mean expression in white adipose tissue from nine participants, as well as negative correlation with PGC1A and UCP1expression, as assessed with the use of interindividual expression patterns in perithyroid brown adipose tissue (Fig. S2B and S2C in the Supplementary Appendix); these findings indicated potential roles for IRX3 and IRX5 in the repression of thermogenesis.

To examine the trans-eQTL genetic control of energy balance by the FTOobesity locus, we used primary preadipocytes from risk-allele carriers and nonrisk-allele carriers to evaluate the genes with mitochondrial and FXR and RXR functions that had expression patterns most closely correlated with those of IRX3 and IRX5, as well as several known markers of energy-balance regulation (Fig. S2D and S2E in the Supplementary Appendix). As compared with nonrisk-allele carriers, risk-allele carriers had lower expression of mitochondrial, browning, and respiration genes and higher expression of lipid-storage markers, which indicated a shift from energy dissipation to energy storage.

These differences in expression were also reflected in the cellular signatures of obesity. Risk-allele carriers had increased adipocyte size, reduced mitochondrial DNA content, and a loss of UCP1 response to β-adrenergic stimulus or cold exposure (Figure 3B and 3C, and Fig. S2F in theSupplementary Appendix), as well as resistance to isoproterenol-mediated uncoupling, a decreased basal oxygen consumption rate, and a reduction in mitochondrial thermogenesis by a factor of 5 (Fig. S2G in the Supplementary Appendix); this indicated excessive accumulation of triglycerides, reduced mitochondrial oxidative capacity, reduced white adipocyte browning, and reduced thermogenesis.

Adipocyte-Autonomous Effects of IRX3 and IRX5 on Energy Balance

We next quantified the effect that manipulation of IRX3 and IRX5 expression had on thermogenesis in primary preadipocytes that were isolated from both risk-allele carriers and nonrisk-allele carriers. In preadipocytes from risk-allele carriers, IRX3 and IRX5 knockdown restored oxygen consumption and thermogenesis response to nonrisk levels, increased thermogenesis by a factor of 7 (Figure 3D), and restored UCP1 expression levels (Fig. S3A in the Supplementary Appendix). In preadipocytes from nonrisk-allele carriers, IRX3 and IRX5 overexpression reduced basal respiration and thermogenesis to risk-allele levels (with thermogenesis reduced by a factor of 8) (Figure 3D) and decreased the expression of UCP1, other regulators of mitochondrial function and thermogenesis (PGC1A, PGC1B, and PRDM16), and the β-adrenergic receptor (ADRB3), which also regulates UCP1-independent thermogenesis programs (Fig. S3B and S3C in the Supplementary Appendix). These manipulations had no significant effect on preadipocytes from participants with the reciprocal genotypes, which indicated that IRX3 and IRX5 levels recapitulate the effect that the FTO genetic variant has on thermogenesis.

To examine the organism-level effects of the repression of Irx3 in adipose tissue, we used adipose Irx3 dominant-negative (aP2-Irx3DN) mice. These mice had pronounced antiobesity characteristics, including reduced body size, body weight, fat mass, white and brown fat depots, and adipocyte size (Fig. S4A through S4G in the Supplementary Appendix). These aP2-Irx3DN mice also had resistance to weight gain on a high-fat diet, increased energy expenditure both at night and during the day, and increased oxygen consumption both at room temperature (22°C) and in thermoneutral conditions (30°C), but they did not have significant differences from control mice in food intake or locomotor activity (Fig. S4A and S4H through S4L in the Supplementary Appendix). At the molecular and cellular levels, these mice had increased mitochondrial activity and thermogenesis marker expression, reduced lipid-storage marker expression in both white and brown fat compartments, and markedly smaller adipocytes than did control mice (Fig. S4M, S4N, and S4O in the Supplementary Appendix).

Figure 4. Disruption of a Conserved ARID5B Repressor Motif by Causal SNV rs1421085 in Humans.

Panel A shows disruption of an ARID5B repressor motif in the evolutionarily conserved motif module surrounding rs1421085. The sequences shown at the top of the panel indicate the frequencies of each nucleotide, with the size scaled to indicate the information content (measured as entropy) at each position. Panel B shows adapted phylogenetic module complexity analysis (PMCA)25 scores in the FTO region for all 82 noncoding SNPs in LD (r2≥0.8) with tag SNV rs1558902, which was identified in a genomewide association study26; rs1421085 had the maximal score. Chromatin state annotation is shown for Roadmap Epigenomics reference genome E025, which corresponds to adipose-derived mesenchymal stem cells; for information on the colors used to denote chromatin states, see Figure S1A in the Supplementary Appendix. Panel C shows increased endogenous expression of IRX3 and IRX5 on single-nucleotide T-to-C editing of rs1421085 in the nonrisk haplotype of a nonrisk-allele carrier, using CRISPR–Cas9 (five clonal expansions). CRISPR–Cas9 re-editing from the engineered C risk allele back to a T nonrisk allele with the use of an alternative single guide RNA restores low endogenous IRX3 and IRX5 gene expression. Panel D shows reduced expression of IRX3 and IRX5 on C-to-T editing of the risk allele in adipocyte progenitors from a risk-allele carrier. Knockdown of ARID5B increases IRX3 and IRX5 levels, as compared….

We next evaluated the tissue-autonomous versus brain-mediated roles of Irx3 by comparing the aP2-Irx3DN mice with hypothalamus dominant-negative Ins2-Irx3DN mice.19 The aP2-Irx3DN mice had a reduction in fat-mass ratio that was 3 times as great as that in Ins2-Irx3DN mice (a reduction of 57% vs. 19%), despite the fact that transgene expression in the hypothalamus was 3 times lower than that in Ins2-Irx3DN mice (Fig. S4P and S4Q in the Supplementary Appendix), which indicated that Irx3 has a hypothalamus-independent regulatory role in whole-body energy regulation. The phenotypic effects of Irx3 repression in aP2-Irx3DN mice were also stronger than those in whole-body Irx3 knockout mice, which suggested potential dominant repressor effects in adipocytes or other tissues, and were independent of Fto gene expression, which did not change (Fig. S4P and S4R in the Supplementary Appendix).

Our findings indicate that both Irx3 and Irx5 have cell-autonomous roles: manipulation of Irx3 andIrx5 led to energy-balance differences in three mouse cellular models, including mouse embryonic fibroblast–derived adipocytes, white 3T3-L1 preadipocytes, and β-adrenergic–stimulated beige ME3 preadipocytes (Fig. S5 in the Supplementary Appendix). In each case, our results indicated that Irx3 and Irx5 induced adipocyte lipid accumulation and repressed thermogenesis in a cell-autonomous way.


Determination of the Causal Variant and Disruption of Repression by ARID5B

To predict the causal variant, the disruption of which is necessary and sufficient to cause IRX3 andIRX5 dysregulation in human preadipocytes, we used phylogenetic module complexity analysis (PMCA)25

(Figure 4A 


Disruption of a Conserved ARID5B Repressor Motif by Causal SNV rs1421085 in Humans., and Fig. S6A and S6B in the Supplementary Appendix). The highest PMCA score was found for the rs1421085 T-to-C SNV, which is in perfect linkage disequilibrium with the most significant reported SNV, rs1558902, across multiple populations (1000 Genomes Phase 1 data), a finding that is consistent with a potentially causal role.

To evaluate whether rs1421085 plays a causal role in enhancer activity, we introduced the C allele into the nonrisk haplotype in our luciferase reporter assay. The T-to-C single-nucleotide alteration increased enhancer activity levels for 10-kb and 1-kb segments centered on the variant, in both orientations and both upstream and downstream of the transcription start, which indicated a gain of enhancer activity in association with the rs1421085 risk allele (Fig. S6C and S6D in the Supplementary Appendix).

To evaluate the effect of the variant on regulator binding, we used electrophoretic mobility-shift assays (EMSAs) of adipocyte nuclear extract with probes for the risk allele and the nonrisk allele of rs1421085. We found binding for the nonrisk allele, T, which lacked enhancer activity, but no binding for the risk allele, C; this indicated that the increased enhancer activity associated with the risk allele is probably due to a loss of repressor binding rather than to a gain of activator binding (Fig. S6E in the Supplementary Appendix).

We examined disrupted motifs and regulator expression to identify potential upstream regulators. The T-to-C substitution disrupted conserved motifs for NKX6-3, LHX6, and the ARID family of regulators (Figure 4A). Among them, ARID5B had the highest expression in adipose tissue and adipocytes and was bound specifically to the nonrisk allele in EMSA competition experiments (Fig. S6E and S6F in the Supplementary Appendix). ARID5B is known to play both repressive and activating roles and was previously implicated in adipogenesis and lipid metabolism in mice.27,28. Among nonrisk-allele carriers, expression of ARID5B was negatively correlated with expression ofIRX3 and IRX5, a finding consistent with ARID5B having a repressive role. No correlation was found in risk-allele carriers, which indicates a loss of ARID5B regulation (Fig. S6G in the Supplementary Appendix).

To evaluate the causal role of ARID5B, we next examined the effects of its knockdown and overexpression on IRX3 and IRX5. ARID5B knockdown increased IRX3 and IRX5 expression in primary preadipocytes from nonrisk-allele carriers to risk-allele levels, which indicates a loss of repression, but it had no effect on preadipocytes from risk-allele carriers, which indicates epistasis with the obesity-risk haplotype (Fig. S6H in the Supplementary Appendix). Consistent with this finding, in SGBS enhancer assays, ARID5B knockdown increased the activity of preadipocytes with the nonrisk allele to risk-allele levels, which indicates a loss of repression, but had no effect on risk-allele constructs, indicating epistasis with the rs1421085 risk allele (Fig. S6I in the Supplementary Appendix). ARID5B overexpression further reduced IRX3 and IRX5 levels in nonrisk-allele carriers, which indicated that repression was strengthened, but had no significant effect on risk-allele carriers, a finding consistent with impaired ARID5B repression in association with the risk haplotype (Fig. S6J in the Supplementary Appendix).

We also evaluated the cellular effects of ARID5B-directed perturbations in primary preadipocytes from risk-allele carriers and nonrisk-allele carriers. In preadipocytes from nonrisk-allele carriers,ARID5B knockdown reduced basal oxygen consumption and lipolysis (Fig. S6K and S6L in theSupplementary Appendix) and shifted expression patterns from mitochondrial to lipid markers (Fig. S2E in the Supplementary Appendix), which indicated that ARID5B plays causal roles in energy-balance regulation. In contrast, ARID5B knockdown had no effect on preadipocytes from risk-allele carriers, a finding consistent with a loss of ARID5B control.

These results suggest that the FTO obesity variant acts through disruption of ARID5B binding in the risk haplotype, leading to a loss of repression, a gain of enhancer activity, and increases inIRX3 and IRX5 expression (Fig. S6M in the Supplementary Appendix).


C-to-T Editing of the rs1421085 Risk Variant and the Effect on Thermogenesis

Targeted genome editing technology involving CRISPR–Cas929 makes it possible to test the phenotypic effect of altering the predicted causal nucleotide rs1421085 in its endogenous genomic context, in isolation from the other obesity-associated genetic variants in the same haplotype. We used CRISPR–Cas9 in primary preadipocytes with two separate guide RNAs, one for rs1421085 C-to-T rescue of the ARID5B motif disruption in risk-allele carriers and one for rs1421085 T-to-C disruption of the ARID5B motif in nonrisk-allele carriers.

We first evaluated the effect of rs1421085 editing on IRX3 and IRX5 expression levels. Starting from preadipocytes of a nonrisk-allele carrier, T-to-C editing doubled endogenous IRX3 and IRX5expression, to levels seen in risk-allele carriers; starting from the edited preadipocytes, C-to-T re-editing back to the nonrisk allele restored low expression levels (Figure 4C). Starting from the risk haplotype, C-to-T editing reduced IRX3 and IRX5 to nonrisk-allele levels, but only in the presence of ARID5B (Figure 4D); this established that disruption of ARID5B repression by rs1421085 is the mechanistic basis of the IRX3 and IRX5 dysregulatory event that mediates the effects of the FTOlocus on obesity.

Next, we evaluated the role of rs1421085 editing during differentiation of white and beige adipocytes, by studying differences in expression between edited and unedited preadipocytes during differentiation. Unedited adipocytes from a risk-allele carrier had a peak in IRX3 and IRX5expression during days 0 and 2 of preadipocyte differentiation into adipocytes; expression during early differentiation was reduced to nonrisk-allele levels by rs1421085 editing, which indicated a causal role of rs1421085 in developmental gene expression programs.

(Figure 5A


Rescue of Metabolic Effects on Adipocyte Thermogenesis through Editing of SNV rs1421085 in a Risk-Allele Carrier. The causal role of rs1421085 was further reflected in a significant increase in the expression of thermogenesis regulators (ADRB3, DIO2, PGC1A, and UCP1) and mitochondrial markers (NDUFA10, COX7A, and CPT1) in differentiating preadipocytes (Figure 5B), which indicated that C-to-T editing of the risk allele rescued thermogenesis regulatory programs.

Last, we evaluated the role of rs1421085 editing in cellular signatures of obesity by quantifying phenotypic differences between edited and unedited adipocytes. A causal role in the regulation of energy balance was indicated by the fact that C-to-T rescue of rs1421085 in edited adipocytes resulted in a reduction in gene expression for lipid storage and lipolytic markers (Fig. S2E and S8A in the Supplementary Appendix), an increase by a factor of 4 in basal metabolic rate and β-adrenergic oxygen consumption, and an increase by a factor of 7 in thermogenesis (Figure 5C, and Fig. S7B in the Supplementary Appendix). In particular, rescue of the ARID5B motif in C-to-T edited preadipocytes restored the strong dependence of mitochondrial respiration on ARID5B that is seen in nonrisk-allele carriers (Fig. S7C in the Supplementary Appendix).

These results indicate that the rs1421085 T-to-C single-nucleotide alteration underlies the association between FTO and obesity by disrupting ARID5B-mediated repression of IRX3 andIRX5. This disruption leads to a developmental shift from browning to whitening programs and loss of mitochondrial thermogenesis (Figure 5D).


Our work elucidates a potential mechanistic basis for the genetic association between FTO and obesity and indicates that the causal variant rs1421085 can disrupt ARID5B repressor binding; this disruption results in derepression of IRX3 and IRX5 during early adipocyte differentiation. This process could lead to a cell-autonomous shift from white adipocyte browning and thermogenesis to lipid storage, increased fat stores, and body-weight gain.

To translate the results of genomewide association studies into mechanistic insights, we combined public resources (epigenomic annotations, chromosome conformation, and regulatory motif conservation), targeted experiments for risk and nonrisk haplotypes (enhancer tiling, gene expression, and cellular profiling), and directed perturbations in human primary cells and mouse models (regulator–target knockdown and overexpression and CRISPR–Cas9 genome editing). These methods are specific to the elucidation of noncoding variants, which constitute the majority of signals in genomewide association studies; 80% of the trait-associated loci identified in such studies lack protein-altering variants, and 93% of the top hits are noncoding.30

The FTO association with obesity is unusual in many ways. First, rs1421085 has both a high frequency and a strong effect size,31 which suggests positive selection or bottlenecks (e.g., 44% frequency in European populations vs. 5% in African populations). Second, rs1421085 has switchlike behavior in enhancer activity, target-gene expression, and cellular phenotypes, possibly because of selective pressures on energy-balance control for rapid adaptation. Third, rs1421085 acts specifically in the early differentiation of preadipocytes, which emphasizes the importance of profiling diverse tissues, cell types, and developmental stages. Fourth, enhancer activity is found only for the risk allele, which emphasizes the importance of profiling both alleles. Finally, rs1421085 leads to a gain of function (increased enhancer, IRX3, and IRX5 activity); this is a rare property in protein-coding variants but may be common in noncoding variants.

The apparent genetic link between obesity and cell-autonomous adipocyte browning suggests a central role of beige adipocyte thermogenesis in whole-body energy metabolism in humans, a role that is consistent with that suggested in recent reports on PRDM16 in mice.9 IRX3 and IRX5 have evolutionarily conserved roles, and the ARID5B motif lies in a module that is functionally conserved across multiple mammalian species; this indicates that adaptive thermogenesis circuits are conserved, and IRX3 and IRX5 probably play both UCP1-dependent and UCP1-independent roles. Even though IRX3 and IRX5 dysregulation by rs1421085 was restricted to early differentiation, their effects persisted in mature adipocytes, and the targeting of these genes can have broader effects.

Last, we found that direct manipulation of the ARID5B–rs1421085–IRX3/IRX5 regulatory axis in primary cell cultures of adipocytes from patients reversed the signatures of obesity. This indicates that in addition to changes in physical activity and nutrition, manipulation of mitochondrial thermogenesis26 offers a potential third pathway for shifting between energy storage and expenditure in a brain-independent and tissue-autonomous way in humans.

In summary, our work elucidates a mechanistic basis for the strongest genetic association with obesity. Our results indicate that the SNV rs1421085 underlies the genetic association between theFTO locus and obesity. The SNV disrupts an evolutionarily conserved motif for the ARID5B repressor, which leads to loss of binding, derepression of a potent preadipocyte superenhancer, and activation of downstream targets IRX3 and IRX5 during early differentiation of mesenchymal progenitors into adipocyte subtypes. This results in a cell-autonomous shift from white adipocyte browning to lipid-storage gene expression programs and to repression of basal mitochondrial respiration, a decrease in thermogenesis in response to stimulus, and an increase in adipocyte size. Manipulation of the uncovered pathway, including knockdown or overexpression of the upstream regulator ARID5B, genome editing of the predicted causal variant rs1421085, and knockdown or overexpression of target genes IRX3 and IRX5, had a significant effect on obesity phenotypes.

Read Full Post »

Metabolic Genomics and Pharmaceutics, Vol. 1 of BioMed Series D available on Amazon Kindle

Metabolic Genomics and Pharmaceutics, Vol. 1 of BioMed Series D available on Amazon Kindle

Reporter: Stephen S Williams, PhD


Leaders in Pharmaceutical Business Intelligence would like to announce the First volume of their BioMedical E-Book Series D:

Metabolic Genomics & Pharmaceutics, Vol. I

SACHS FLYER 2014 Metabolomics SeriesDindividualred-page2

which is now available on Amazon Kindle at

This e-Book is a comprehensive review of recent Original Research on  METABOLOMICS and related opportunities for Targeted Therapy written by Experts, Authors, Writers. This is the first volume of the Series D: e-Books on BioMedicine – Metabolomics, Immunology, Infectious Diseases.  It is written for comprehension at the third year medical student level, or as a reference for licensing board exams, but it is also written for the education of a first time baccalaureate degree reader in the biological sciences.  Hopefully, it can be read with great interest by the undergraduate student who is undecided in the choice of a career. The results of Original Research are gaining value added for the e-Reader by the Methodology of Curation. The e-Book’s articles have been published on the Open Access Online Scientific Journal, since April 2012.  All new articles on this subject, will continue to be incorporated, as published with periodical updates.

We invite e-Readers to write an Article Reviews on Amazon for this e-Book on Amazon.

All forthcoming BioMed e-Book Titles can be viewed at:

Leaders in Pharmaceutical Business Intelligence, launched in April 2012 an Open Access Online Scientific Journal is a scientific, medical and business multi expert authoring environment in several domains of  life sciences, pharmaceutical, healthcare & medicine industries. The venture operates as an online scientific intellectual exchange at their website and for curation and reporting on frontiers in biomedical, biological sciences, healthcare economics, pharmacology, pharmaceuticals & medicine. In addition the venture publishes a Medical E-book Series available on Amazon’s Kindle platform.

Analyzing and sharing the vast and rapidly expanding volume of scientific knowledge has never been so crucial to innovation in the medical field. WE are addressing need of overcoming this scientific information overload by:

  • delivering curation and summary interpretations of latest findings and innovations on an open-access, Web 2.0 platform with future goals of providing primarily concept-driven search in the near future
  • providing a social platform for scientists and clinicians to enter into discussion using social media
  • compiling recent discoveries and issues in yearly-updated Medical E-book Series on Amazon’s mobile Kindle platform

This curation offers better organization and visibility to the critical information useful for the next innovations in academic, clinical, and industrial research by providing these hybrid networks.

Table of Contents for Metabolic Genomics & Pharmaceutics, Vol. I

Chapter 1: Metabolic Pathways

Chapter 2: Lipid Metabolism

Chapter 3: Cell Signaling

Chapter 4: Protein Synthesis and Degradation

Chapter 5: Sub-cellular Structure

Chapter 6: Proteomics

Chapter 7: Metabolomics

Chapter 8:  Impairments in Pathological States: Endocrine Disorders; Stress

                   Hypermetabolism and Cancer

Chapter 9: Genomic Expression in Health and Disease 






Read Full Post »

Hypoxia Inducible Factor 1 (HIF-1)

Writer and Curator: Larry H Bernstein, MD, FCAP

7.9  Hypoxia Inducible Factor 1 (HIF-1)

7.9.1 Hypoxia and mitochondrial oxidative metabolism

7.9.2 Hypoxia promotes isocitrate dehydrogenase-dependent carboxylation of α-ketoglutarate to citrate to support cell growth and viability

7.9.3 Hypoxia-Inducible Factors in Physiology and Medicine

7.9.4 Hypoxia-inducible factor 1. Regulator of mitochondrial metabolism and mediator of ischemic preconditioning

7.9.5 Regulation of cancer cell metabolism by hypoxia-inducible factor 1

7.9.6 Coming up for air. HIF-1 and mitochondrial oxygen consumption

7.9.7 HIF-1 mediates adaptation to hypoxia by actively downregulating mitochondrial oxygen consumption

7.9.8 HIF-1. upstream and downstream of cancer metabolism

7.9.9 In Vivo HIF-Mediated Reductive Carboxylation

7.9.10 Evaluation of HIF-1 inhibitors as anticancer agents



7.9.1 Hypoxia and mitochondrial oxidative metabolism

Solaini G1Baracca ALenaz GSgarbi G.
Biochim Biophys Acta. 2010 Jun-Jul; 1797(6-7):1171-7

It is now clear that mitochondrial defects are associated with a large variety of clinical phenotypes. This is the result of the mitochondria’s central role in energy production, reactive oxygen species homeostasis, and cell death. These processes are interdependent and may occur under various stressing conditions, among which low oxygen levels (hypoxia) are certainly prominent. Cells exposed to hypoxia respond acutely with endogenous metabolites and proteins promptly regulating metabolic pathways, but if low oxygen levels are prolonged, cells activate adapting mechanisms, the master switch being the hypoxia-inducible factor 1 (HIF-1). Activation of this factor is strictly bound to the mitochondrial function, which in turn is related with the oxygen level. Therefore in hypoxia, mitochondria act as [O2] sensors, convey signals to HIF-1directly or indirectly, and contribute to the cell redox potential, ion homeostasis, and energy production. Although over the last two decades cellular responses to low oxygen tension have been studied extensively, mechanisms underlying these functions are still indefinite. Here we review current knowledge of the mitochondrial role in hypoxia, focusing mainly on their role in cellular energy and reactive oxygen species homeostasis in relation with HIF-1 stabilization. In addition, we address the involvement of HIF-1 and the inhibitor protein of F1F0 ATPase in the hypoxia-induced mitochondrial autophagy.

Over the last two decades a defective mitochondrial function associated with hypoxia has been invoked in many diverse complex disorders, such as type 2 diabetes [1] and [2], Alzheimer’s disease [3] and [4], cardiac ischemia/reperfusion injury [5] and [6], tissue inflammation [7], and cancer [8][9][10],[11] and [12].

The [O2] in air-saturated aqueous buffer at 37 °C is approx. 200 μM [13]; however, mitochondria in vivo are exposed to a considerably lower [O2] that varies with tissue and physiological state. Under physiological conditions, most human resting cells experience some 5% oxygen tension, however the [O2] gradient occurring between the extracellular environment and mitochondria, where oxygen is consumed by cytochrome c oxidase, results in a significantly lower [O2] exposition of mitochondria. Below this oxygen level, most mammalian tissues are exposed to hypoxic conditions  [14]. These may arise in normal development, or as a consequence of pathophysiological conditions where there is a reduced oxygen supply due to a respiratory insufficiency or to a defective vasculature. Such conditions include inflammatory diseases, diabetes, ischemic disorders (cerebral or cardiovascular), and solid tumors. Mitochondria consume the greatest amount (some 85–90%) of oxygen in cells to allow oxidative phosphorylation (OXPHOS), which is the primary metabolic pathway for ATP production. Therefore hypoxia will hamper this metabolic pathway, and if the oxygen level is very low, insufficient ATP availability might result in cell death [15].

When cells are exposed to an atmosphere with reduced oxygen concentration, cells readily “respond” by inducing adaptive reactions for their survival through the AMP-activated protein kinase (AMPK) pathway (see for a recent review [16]) which inter alia increases glycolysis driven by enhanced catalytic efficiency of some enzymes, including phosphofructokinase-1 and pyruvate kinase (of note, this oxidative flux is thermodynamically allowed due to both reduced phosphorylation potential [ATP]/([ADP][Pi]) and the physiological redox state of the cell). However, this is particularly efficient only in the short term, therefore cells respond to prolonged hypoxia also by stimulation of hypoxia-inducible factors (HIFs: HIF-1 being the mostly studied), which are heterodimeric transcription factors composed of α and β subunits, first described by Semenza and Wang [17]. These HIFs in the presence of hypoxic oxygen levels are activated through a complex mechanism in which the oxygen tension is critical (see below). Afterwards HIFs bind to hypoxia-responsive elements, activating the transcription of more than two hundred genes that allow cells to adapt to the hypoxic environment [18] and [19].

Several excellent reviews appeared in the last few years describing the array of changes induced by oxygen deficiency in both isolated cells and animal tissues. In in vivo models, a coordinated regulation of tissue perfusion through vasoactive molecules such as nitric oxide and the action of carotid bodies rapidly respond to changes in oxygen demand [20][21][22][23] and [24]. Within isolated cells, hypoxia induces significant metabolic changes due to both variation of metabolites level and activation/inhibition of enzymes and transporters; the most important intracellular effects induced by different pathways are expertly described elsewhere (for recent reviews, see [25][26] and [27]). It is reasonable to suppose that the type of cells and both the severity and duration of hypoxia may determine which pathways are activated/depressed and their timing of onset [3][6][10][12][23] and [28]. These pathways will eventually lead to preferential translation of key proteins required for adaptation and survival to hypoxic stress. Although in the past two decades, the discovery of HIF-1 by Gregg Semenza et al. provided a molecular platform to investigate the mechanism underlying responses to oxygen deprivation, the molecular and cellular biology of hypoxia has still to be completely elucidated. This review summarizes recent experimental data concerned with mitochondrial structure and function adaptation to hypoxia and evaluates it in light of the main structural and functional parameters defining the mitochondrial bioenergetics. Since mitochondria contain an inhibitor protein, IF1, whose action on the F1F0 ATPase has been considered for decades of critical importance in hypoxia/ischemia, particular notice will be dedicated to analyze molecular aspects of IF1 regulation of the enzyme and its possible role in the metabolic changes induced by low oxygen levels in cells.

Mechanism(s) of HIF-1 activation

HIF-1 consists of an oxygen-sensitive HIF-1α subunit that heterodimerizes with the HIF-1β subunit to bind DNA. In high O2 tension, HIF-1α is oxidized (hydroxylated) by prolyl hydroxylases (PHDs) using α-ketoglutarate derived from the tricarboxylic acid (TCA) cycle. The hydroxylated HIF-1α subunit interacts with the von Hippel–Lindau protein, a critical member of an E3 ubiquitin ligase complex that polyubiquitylates HIF. This is then catabolized by proteasomes, such that HIF-1α is continuously synthesized and degraded under normoxic conditions [18]. Under hypoxia, HIF-1α hydroxylation does not occur, thereby stabilizing HIF-1 (Fig. 1). The active HIF-1 complex in turn binds to a core hypoxia response element in a wide array of genes involved in a diversity of biological processes, and directly transactivates glycolytic enzyme genes [29]. Notably, O2 concentration, multiple mitochondrial products, including the TCA cycle intermediates and reactive oxygen species, can coordinate PHD activity, HIF stabilization, hence the cellular responses to O2 depletion [30] and [31]. Incidentally, impaired TCA cycle flux, particularly if it is caused by succinate dehydrogenase dysfunction, results in decreased or loss of energy production from both the electron-transport chain and the Krebs cycle, and also in overproduction of free radicals [32]. This leads to severe early-onset neurodegeneration or, as it occurs in individuals carrying mutations in the non-catalytic subunits of the same enzyme, to tumors such as phaeochromocytoma and paraganglioma. However, impairment of the TCA cycle may be relevant also for the metabolic changes occurring in mitochondria exposed to hypoxia, since accumulation of succinate has been reported to inhibit PHDs [33]. It has to be noticed that some authors believe reactive oxygen species (ROS) to be essential to activate HIF-1 [34], but others challenge this idea [35], therefore the role of mitochondrial ROS in the regulation of HIF-1 under hypoxia is still controversial [36]. Moreover, the contribution of functional mitochondria to HIF-1 regulation has also been questioned by others [37][38] and [39].

Major mitochondrial changes in hypoxia

Major mitochondrial changes in hypoxia

Fig. 1. Major mitochondrial changes in hypoxia. Hypoxia could decrease electron-transport rate determining Δψm reduction, increased ROS generation, and enhanced NO synthase. One (or more) of these factors likely contributes to HIF stabilization, that in turn induces metabolic adaptation of both hypoxic cells and mitophagy. The decreased Δψm could also induce an active binding of IF1, which might change mitochondrial morphology and/or dynamics, and inhibit mitophagy. Solid lines indicate well established hypoxic changes in cells, whilst dotted lines indicate changes not yet stated. Inset, relationships between extracellular O2concentration and oxygen tension.

Oxygen is a major determinant of cell metabolism and gene expression, and as cellular O2 levels decrease, either during isolated hypoxia or ischemia-associated hypoxia, metabolism and gene expression profiles in the cells are significantly altered. Low oxygen reduces OXPHOS and Krebs cycle rates, and participates in the generation of nitric oxide (NO), which also contributes to decrease respiration rate [23] and [40]. However, oxygen is also central in the generation of reactive oxygen species, which can participate in cell signaling processes or can induce irreversible cellular damage and death [41].

As specified above, cells adapt to oxygen reduction by inducing active HIF, whose major effect on cells energy homeostasis is the inactivation of anabolism, activation of anaerobic glycolysis, and inhibition of the mitochondrial aerobic metabolism: the TCA cycle, and OXPHOS. Since OXPHOS supplies the majority of ATP required for cellular processes, low oxygen tension will severely reduce cell energy availability. This occurs through several mechanisms: first, reduced oxygen tension decreases the respiration rate, due first to nonsaturating substrate for cytochrome c oxidase (COX), secondarily, to allosteric modulation of COX[42]. As a consequence, the phosphorylation potential decreases, with enhancement of the glycolysis rate primarily due to allosteric increase of phosphofructokinase activity; glycolysis however is poorly efficient and produces lactate in proportion of 0.5 mol/mol ATP, which eventually drops cellular pH if cells are not well perfused, as it occurs under defective vasculature or ischemic conditions  [6]. Besides this “spontaneous” (thermodynamically-driven) shift from aerobic to anaerobic metabolism which is mediated by the kinetic changes of most enzymes, the HIF-1 factor activates transcription of genes encoding glucose transporters and glycolytic enzymes to further increase flux of reducing equivalents from glucose to lactate[43] and [44]. Second, HIF-1 coordinates two different actions on the mitochondrial phase of glucose oxidation: it activates transcription of the PDK1 gene encoding a kinase that phosphorylates and inactivates pyruvate dehydrogenase, thereby shunting away pyruvate from the mitochondria by preventing its oxidative decarboxylation to acetyl-CoA [45] and [46]. Moreover, HIF-1 induces a switch in the composition of cytochrome c oxidase from COX4-1 to COX4-2 isoform, which enhances the specific activity of the enzyme. As a result, both respiration rate and ATP level of hypoxic cells carrying the COX4-2 isoform of cytochrome c oxidase were found significantly increased with respect to the same cells carrying the COX4-1 isoform [47]. Incidentally, HIF-1 can also increase the expression of carbonic anhydrase 9, which catalyses the reversible hydration of CO2 to HCO3 and H+, therefore contributing to pH regulation.

Effects of hypoxia on mitochondrial structure and dynamics

Mitochondria form a highly dynamic tubular network, the morphology of which is regulated by frequent fission and fusion events. The fusion/fission machineries are modulated in response to changes in the metabolic conditions of the cell, therefore one should expect that hypoxia affect mitochondrial dynamics. Oxygen availability to cells decreases glucose oxidation, whereas oxygen shortage consumes glucose faster in an attempt to produce ATP via the less efficient anaerobic glycolysis to lactate (Pasteur effect). Under these conditions, mitochondria are not fueled with substrates (acetyl-CoA and O2), inducing major changes of structure, function, and dynamics (for a recent review see [48]). Concerning structure and dynamics, one of the first correlates that emerge is that impairment of mitochondrial fusion leads to mitochondrial depolarization, loss of mtDNA that may be accompanied by altered respiration rate, and impaired distribution of the mitochondria within cells [49][50] and [51]. Indeed, exposure of cortical neurons to moderate hypoxic conditions for several hours, significantly altered mitochondrial morphology, decreased mitochondrial size and reduced mitochondrial mean velocity. Since these effects were either prevented by exposing the neurons to inhibitors of nitric oxide synthase or mimicked by NO donors in normoxia, the involvement of an NO-mediated pathway was suggested [52]. Mitochondrial motility was also found inhibited and controlled locally by the [ADP]/[ATP] ratio [53]. Interestingly, the author used an original approach in which mitochondria were visualized using tetramethylrhodamineethylester and their movements were followed by applying single-particle tracking.

Of notice in this chapter is that enzymes controlling mitochondrial morphology regulators provide a platform through which cellular signals are transduced within the cell in order to affect mitochondrial function [54]. Accordingly, one might expect that besides other mitochondrial factors [30] and [55] playing roles in HIF stabilization, also mitochondrial morphology might reasonably be associated with HIF stabilization. In order to better define the mechanisms involved in the morphology changes of mitochondria and in their dynamics when cells experience hypoxic conditions, these pioneering studies should be corroborated by and extended to observations on other types of cells focusing also on single proteins involved in both mitochondrial fusion/fission and motion.

Effects of hypoxia on the respiratory chain complexes

O2 is the terminal acceptor of electrons from cytochrome c oxidase (Complex IV), which has a very high affinity for it, being the oxygen concentration for half-maximal respiratory rate at pH 7.4 approximately 0.7 µM [56]. Measurements of mitochondrial oxidative phosphorylation indicated that it is not dependent on oxygen concentration up to at least 20 µM at pH 7.0 and the oxygen dependence becomes markedly greater as the pH is more alkaline [56]. Similarly, Moncada et al. [57] found that the rate of O2 consumption remained constant until [O2] fell below 15 µM. Accordingly, most reports in the literature consider hypoxic conditions occurring in cells at 5–0.5% O2, a range corresponding to 46–4.6 µM O2 in the cells culture medium (see Fig. 1 inset). Since between the extracellular environment and mitochondria an oxygen pressure gradient is established [58], the O2 concentration experienced by Complex IV falls in the range affecting its kinetics, as reported above.

Under these conditions, a number of changes on the OXPHOS machinery components, mostly mediated by HIF-1 have been found. Thus, Semenza et al. [59] and others thereafter [46] reported that activation of HIF-1α induces pyruvate dehydrogenase kinase, which inhibits pyruvate dehydrogenase, suggesting that respiration is decreased by substrate limitation. Besides, other HIF-1 dependent mechanisms capable to affect respiration rate have been reported. First, the subunit composition of COX is altered in hypoxic cells by increased degradation of the COX4-1 subunit, which optimizes COX activity under aerobic conditions, and increased expression of the COX4-2 subunit, which optimizes COX activity under hypoxic conditions [29]. On the other hand, direct assay of respiration rate in cells exposed to hypoxia resulted in a significant reduction of respiration [60]. According with the evidence of Zhang et al., the respiration rate decrease has to be ascribed to mitochondrial autophagy, due to HIF-1-mediated expression of BNIP3. This interpretation is in line with preliminary results obtained in our laboratory where the assay of the citrate synthase activity of cells exposed to different oxygen tensions was performed. Fig. 2 shows the citrate synthase activity, which is taken as an index of the mitochondrial mass [11], with respect to oxygen tension: [O2] and mitochondrial mass are directly linked.

Citrate synthase activity

Citrate synthase activity

Fig. 2. Citrate synthase activity. Human primary fibroblasts, obtained from skin biopsies of 5 healthy donors, were seeded at a density of 8,000 cells/cm2 in high glucose Dulbecco’s Modified Eagle Medium, DMEM (25 mM glucose, 110 mg/l pyruvate, and 4 mM glutamine) supplemented with 15% Foetal Bovine Serum (FBS). 18 h later, cell culture dishes were washed once with Hank’s Balanced Salt Solution (HBSS) and the medium was replaced with DMEM containing 5 mM glucose, 110 mg/l pyruvate, and 4 mM glutamine supplemented with 15% FBS. Cell culture dishes were then placed into an INVIVO2 humidified hypoxia workstation (Ruskinn Technologies, Bridgend, UK) for 72 h changing the medium at 48 h, and oxygen partial pressure (tension) conditions were: 20%, 4%, 2%, 1% and 0.5%. Cells were subsequently collected within the workstation with trypsin-EDTA (0.25%), washed with PBS and resuspended in a buffer containing 10 mM Tris/HCl, 0.1 M KCl, 5 mM KH2PO4, 1 mM EGTA, 3 mM EDTA, and 2 mM MgCl2 pH 7.4 (all the solutions were preconditioned to the appropriate oxygen tension condition). The citrate synthase activity was assayed essentially by incubating 40 µg of cells with 0.02% Triton X-100, and monitoring the reaction by measuring spectrophotometrically the rate of free coenzyme A released, as described in [90]. Enzymatic activity was expressed as nmol/min/mg of protein. Three independent experiments were carried out and assays were performed in either duplicate or triplicate.

However, the observations of Semenza et al. must be seen in relation with data reported by Moncada et al.[57] and confirmed by others [61] in which it is clearly shown that when cells (various cell lines) experience hypoxic conditions, nitric oxide synthases (NOSs) are activated, therefore NO is released. As already mentioned above, NO is a strong competitor of O2 for cytochrome c oxidase, whose apparent Km results increased, hence reduction of mitochondrial cytochromes and all the other redox centres of the respiratory chain occurs. In addition, very recent data indicate a potential de-activation of Complex I when oxygen is lacking, as it occurs in prolonged hypoxia [62]. According to Hagen et al. [63] the NO-dependent inhibition of cytochrome c oxidase should allow “saved” O2 to redistribute within the cell to be used by other enzymes, including PHDs which inactivate HIF. Therefore, unless NO inhibition of cytochrome c oxidase occurs only when [O2] is very low, inhibition of mitochondrial oxygen consumption creates the paradox of a situation in which the cell may fail to register hypoxia. It has been tempted to solve this paradox, but to date only hypotheses have been proposed [23] and [26]. Interestingly, recent observations on yeast cells exposed to hypoxia revealed abnormal protein carbonylation and protein tyrosine nitration that were ascribed to increased mitochondrially generated superoxide radicals and NO, two species typically produced at low oxygen levels, that combine to form ONOO [64]. Based on these studies a possible explanation has been proposed for the above paradox.

Finally, it has to be noticed that the mitochondrial respiratory deficiency observed in cardiomyocytes of dogs in which experimental heart failure had been induced lies in the supermolecular assembly rather than in the individual components of the electron-transport chain [65]. This observation is particularly intriguing since loss of respirasomes is thought to facilitate ROS generation in mitochondria [66], therefore supercomplexes disassembly might explain the paradox of reduced [O2] and the enhanced ROS found in hypoxic cells. Specifically, hypoxia could reduce mitochondrial fusion by impairing mitochondrial membrane potential, which in turn could induce supercomplexes disassembly, increasing ROS production[11].

Complex III and ROS production

It has been estimated that, under normoxic physiological conditions, 1–2% of electron flow through the mitochondrial respiratory chain gives rise to ROS [67] and [68]. It is now recognized that the major sites of ROS production are within Complexes I and III, being prevalent the contribution of Complex I [69] (Fig. 3). It might be expected that hypoxia would decrease ROS production, due to the low level of O2 and to the diminished mitochondrial respiration [6] and [46], but ROS level is paradoxically increased. Indeed, about a decade ago, Chandel et al. [70] provided good evidence that mitochondrial reactive oxygen species trigger hypoxia-induced transcription, and a few years later the same group [71] showed that ROS generated at Complex III of the mitochondrial respiratory chain stabilize HIF-1α during hypoxia (Fig. 1 and Fig. 3). Although others have proposed mechanisms indicating a key role of mitochondria in HIF-1α regulation during hypoxia (for reviews see [64] and [72]), the contribution of mitochondria to HIF-1 regulation has been questioned by others [35][36] and [37]. Results of Gong and Agani [35] for instance show that inhibition of electron-transport Complexes I, III, and IV, as well as inhibition of mitochondrial F0F1 ATPase, prevents HIF-1α expression and that mitochondrial reactive oxygen species are not involved in HIF-1α regulation during hypoxia. Concurrently, Tuttle et al. [73], by means of a non invasive, spectroscopic approach, could find no evidence to suggest that ROS, produced by mitochondria, are needed to stabilize HIF-1α under moderate hypoxia. The same authors found the levels of HIF-1α comparable in both normal and ρ0 cells (i.e. cells lacking mitochondrial DNA). On the contrary, experiments carried out on genetic models consisting of either cells lacking cytochrome c or ρ0 cells both could evidence the essential role of mitochondrial respiration to stabilize HIF-1α [74]. Thus, cytochrome c null cells, being incapable to respire, exposed to moderate hypoxia (1.5% O2) prevented oxidation of ubiquinol and generation of the ubisemiquinone radical, thus eliminating superoxide formation at Complex III [71]. Concurrently, ρ0 cells lacking electron transport, exposed 4 h to moderate hypoxia failed to stabilize HIF-1α, suggesting the essential role of the respiratory chain for the cellular sensing of low O2 levels. In addition, recent evidence obtained on genetic manipulated cells (i.e. cytochrome b deficient cybrids) showed increased ROS levels and stabilized HIF-1α protein during hypoxia [75]. Moreover, RNA interference of the Complex III subunit Rieske iron sulfur protein in the cytochrome b deficient cells, abolished ROS generation at the Qo site of Complex III, preventing HIF-1α stabilization. These observations, substantiated by experiments with MitoQ, an efficient mitochondria-targeted antioxidant, strongly support the involvement of mitochondrial ROS in regulating HIF-1α. Nonetheless, collectively, the available data do not allow to definitely state the precise role of mitochondrial ROS in regulating HIF-1α, but the pathway stabilizing HIF-1α appears undoubtedly mitochondria-dependent [30].

Overview of mitochondrial electron and proton flux in hypoxia

Overview of mitochondrial electron and proton flux in hypoxia

Overview of mitochondrial electron and proton flux in hypoxia

Fig. 3. Overview of mitochondrial electron and proton flux in hypoxia. Electrons released from reduced cofactors (NADH and FADH2) under normoxia flow through the redox centres of the respiratory chain (r.c.) to molecular oxygen (blue dotted line), to which a proton flux from the mitochondrial matrix to the intermembrane space is coupled (blue arrows). Protons then flow back to the matrix through the F0 sector of the ATP synthase complex, driving ATP synthesis. ATP is carried to the cell cytosol by the adenine nucleotide translocator (blue arrows). Under moderate to severe hypoxia, electrons escape the r.c. redox centres and reduce molecular oxygen to the superoxide anion radical before reaching the cytochrome c (red arrow). Under these conditions, to maintain an appropriate Δψm, ATP produced by cytosolic glycolysis enters the mitochondria where it is hydrolyzed by the F1F0ATPase with extrusion of protons from the mitochondrial matrix (red arrows).

Hypoxia and ATP synthase

The F1F0 ATPase (ATP synthase) is the enzyme responsible of catalysing ADP phosphorylation as the last step of OXPHOS. It is a rotary motor using the proton motive force across the mitochondrial inner membrane to drive the synthesis of ATP [76]. It is a reversible enzyme with ATP synthesis or hydrolysis taking place in the F1 sector at the matrix side of the membrane, chemical catalysis being coupled to H+transport through the transmembrane F0 sector.

Under normoxia the enzyme synthesizes ATP, but when mitochondria experience hypoxic conditions the mitochondrial membrane potential (Δψm) decreases below its endogenous steady-state level (some 140 mV, negative inside the matrix [77]) and the F1F0 ATPase may work in the reversal mode: it hydrolyses ATP (produced by anaerobic glycolysis) and uses the energy released to pump protons from the mitochondrial matrix to the intermembrane space, concurring with the adenine nucleotide translocator (i.e. in hypoxia it exchanges cytosolic ATP4− for matrix ADP3−) to maintain the physiological Δψm ( Fig. 3). Since under conditions of limited oxygen availability the decline in cytoplasmic high energy phosphates is mainly due to hydrolysis by the ATP synthase working in reverse [6] and [78], the enzyme must be strictly regulated in order to avoid ATP dissipation. This is achieved by a natural protein, the H+ψm-dependent IF1, that binds to the catalytic F1 sector at low pH and low Δψm (such as it occurs in hypoxia/ischemia) [79]. IF1 binding to the ATP synthase results in a rapid and reversible inhibition of the enzyme [80], which could reach about 50% of maximal activity (for recent reviews see [6] and [81]).

Besides this widely studied effect, IF1 appears to be associated with ROS production and mitochondrial autophagy (mitophagy). This is a mechanism involving the catabolic degradation of macromolecules and organelles via the lysosomal pathway that contributes to housekeeping and regenerate metabolites. Autophagic degradation is involved in the regulation of the ageing process and in several human diseases, such as myocardial ischemia/reperfusion [82], Alzheimer’s Disease, Huntington diseases, and inflammatory diseases (for recent reviews see [83] and [84], and, as mentioned above, it promotes cell survival by reducing ROS and mtDNA damage under hypoxic conditions.

Campanella et al. [81] reported that, in HeLa cells under normoxic conditions, basal autophagic activity varies in relation to the expression levels of IF1. Accordingly, cells overexpressing IF1 result in ROS production similar to controls, conversely cells in which IF1 expression is suppressed show an enhanced ROS production. In parallel, the latter cells show activation of the mitophagy pathway (Fig. 1), therefore suggesting that variations in IF1 expression level may play a significant role in defining two particularly important parameters in the context of the current review: rates of ROS generation and mitophagy. Thus, the hypoxia-induced enhanced expression level of IF1[81] should be associated with a decrease of both ROS production and autophagy, which is in apparent conflict with the hypoxia-induced ROS increase and with the HIF-1-dependent mitochondrial autophagy shown by Zhang et al. [60] as an adaptive metabolic response to hypoxia. However, in the experiments of Zhang et al. the cells were exposed to hypoxia for 48 h, whereas the F1F0-ATPase inhibitor exerts a prompt action on the enzyme and to our knowledge, it has never been reported whether its action persists during prolonged hypoxic expositions. Pertinent with this problem is the very recent observation that IEX-1 (immediate early response gene X-1), a stress-inducible gene that suppresses production of ROS and protects cells from apoptosis [85], targets the mitochondrial F1F0-ATPase inhibitor for degradation, reducing ROS by decreasing Δψm. It has to be noticed that the experiments described were carried out under normal oxygen availability, but it does not seem reasonable to rule out IEX-1 from playing a role under stress conditions as those induced by hypoxia in cells, therefore this issue might deserve an investigation also at low oxygen levels.

In conclusion, data are still emerging regarding the regulation of mitochondrial function by the F1F0 ATPase within hypoxic responses in different cellular and physiological contexts. Given the broad pathophysiological role of hypoxic cellular modulation, an understanding of the subtle tuning among different effectors of the ATP synthase is desirable to eventually target future therapeutics most effectively. Our laboratory is actually involved in carrying out investigations to clarify this context.

Conclusions and perspectives

The mitochondria are important cellular platforms that both propagate and initiate intracellular signals that lead to overall cellular and metabolic responses. During the last decades, a significant amount of relevant data has been obtained on the identification of mechanisms of cellular adaptation to hypoxia. In hypoxic cells there is an enhanced transcription and synthesis of several glycolytic pathway enzymes/transporters and reduction of synthesis of proteins involved in mitochondrial catabolism. Although well defined kinetic parameters of reactions in hypoxia are lacking, it is usually assumed that these transcriptional changes lead to metabolic flux modification. The required biochemical experimentation has been scarcely addressed until now and only in few of the molecular and cellular biology studies the transporter and enzyme kinetic parameters and flux rate have been determined, leaving some uncertainties.

Central to mitochondrial function and ROS generation is an electrochemical proton gradient across the mitochondrial inner membrane that is established by the proton pumping activity of the respiratory chain, and that is strictly linked to the F1F0-ATPase function. Evaluation of the mitochondrial membrane potential in hypoxia has only been studied using semiquantitative methods based on measurements of the fluorescence intensity of probes taken up by cells experiencing normal or hypoxic conditions. However, this approach is intrinsically incorrect due to the different capability that molecular oxygen has to quench fluorescence [86] and [87] and to the uncertain concentration the probe attains within mitochondria, whose mass may be reduced by a half in hypoxia [60]. In addition, the uncertainty about measurement of mitochondrial superoxide radical and H2O2 formation in vivo [88] hampers studies on the role of mitochondrial ROS in hypoxic oxidative damage, redox signaling, and HIF-1 stabilization.

The duration and severity of hypoxic stress differentially activate the responses discussed throughout and lead to substantial phenotypic variations amongst tissues and cell models, which are not consistently and definitely known. Certainly, understanding whether a hierarchy among hypoxia response mechanisms exists and which are the precise timing and conditions of each mechanism to activate, will improve our knowledge of the biochemical mechanisms underlying hypoxia in cells, which eventually may contribute to define therapeutic targets in hypoxia-associated diseases. To this aim it might be worth investigating the hypoxia-induced structural organization of both the respiratory chain enzymes in supramolecular complexes and the assembly of the ATP synthase to form oligomers affecting ROS production [65] and inner mitochondrial membrane structure [89], respectively.

7.9.2 Hypoxia promotes isocitrate dehydrogenase-dependent carboxylation of α-ketoglutarate to citrate to support cell growth and viability

DR WisePS WardJES ShayJR CrossJJ Gruber, UM Sachdeva, et al.
Proc Nat Acad Sci Oct 27, 2011; 108(49):19611–19616

Citrate is a critical metabolite required to support both mitochondrial bioenergetics and cytosolic macromolecular synthesis. When cells proliferate under normoxic conditions, glucose provides the acetyl-CoA that condenses with oxaloacetate to support citrate production. Tricarboxylic acid (TCA) cycle anaplerosis is maintained primarily by glutamine. Here we report that some hypoxic cells are able to maintain cell proliferation despite a profound reduction in glucose-dependent citrate production. In these hypoxic cells, glutamine becomes a major source of citrate. Glutamine-derived α-ketoglutarate is reductively carboxylated by the NADPH-linked mitochondrial isocitrate dehydrogenase (IDH2) to form isocitrate, which can then be isomerized to citrate. The increased IDH2-dependent carboxylation of glutamine-derived α-ketoglutarate in hypoxia is associated with a concomitant increased synthesis of 2-hydroxyglutarate (2HG) in cells with wild-type IDH1 and IDH2. When either starved of glutamine or rendered IDH2-deficient by RNAi, hypoxic cells are unable to proliferate. The reductive carboxylation of glutamine is part of the metabolic reprogramming associated with hypoxia-inducible factor 1 (HIF1), as constitutive activation of HIF1 recapitulates the preferential reductive metabolism of glutamine-derived α-ketoglutarate even in normoxic conditions. These data support a role for glutamine carboxylation in maintaining citrate synthesis and cell growth under hypoxic conditions.

Citrate plays a critical role at the center of cancer cell metabolism. It provides the cell with a source of carbon for fatty acid and cholesterol synthesis (1). The breakdown of citrate by ATP-citrate lyase is a primary source of acetyl-CoA for protein acetylation (2). Metabolism of cytosolic citrate by aconitase and IDH1 can also provide the cell with a source of NADPH for redox regulation and anabolic synthesis. Mammalian cells depend on the catabolism of glucose and glutamine to fuel proliferation (3). In cancer cells cultured at atmospheric oxygen tension (21% O2), glucose and glutamine have both been shown to contribute to the cellular citrate pool, with glutamine providing the major source of the four-carbon molecule oxaloacetate and glucose providing the major source of the two-carbon molecule acetyl-CoA (45). The condensation of oxaloacetate and acetyl-CoA via citrate synthase generates the 6 carbon citrate molecule. However, both the conversion of glucose-derived pyruvate to acetyl-CoA by pyruvate dehydrogenase (PDH) and the conversion of glutamine to oxaloacetate through the TCA cycle depend on NAD+, which can be compromised under hypoxic conditions. This raises the question of how cells that can proliferate in hypoxia continue to synthesize the citrate required for macromolecular synthesis.

This question is particularly important given that many cancers and stem/progenitor cells can continue proliferating in the setting of limited oxygen availability (67). Louis Pasteur first highlighted the impact of hypoxia on nutrient metabolism based on his observation that hypoxic yeast cells preferred to convert glucose into lactic acid rather than burning it in an oxidative fashion. The molecular basis for this shift in mammalian cells has been linked to the activity of the transcription factor HIF1 (810). Stabilization of the labile HIF1α subunit occurs in hypoxia. It can also occur in normoxia through several mechanisms including loss of the von Hippel-Lindau tumor suppressor (VHL), a common occurrence in renal carcinoma (11). Although hypoxia and/or HIF1α stabilization is a common feature of multiple cancers, to date the source of citrate in the setting of hypoxia or HIF activation has not been determined.

Here, we study the sources of hypoxic citrate synthesis in a glioblastoma cell line that proliferates in profound hypoxia (0.5% O2). Glucose uptake and conversion to lactic acid increased in hypoxia. However, glucose conversion into citrate dramatically declined. Glutamine consumption remained constant in hypoxia, and hypoxic cells were addicted to the use of glutamine in hypoxia as a source of α-ketoglutarate. Glutamine provided the major carbon source for citrate synthesis during hypoxia. However, the TCA cycle-dependent conversion of glutamine into citric acid was significantly suppressed. In contrast, there was a relative increase in glutamine-dependent citrate production in hypoxia that resulted from carboxylation of α-ketoglutarate. This reductive synthesis required the presence of mitochondrial isocitrate dehydrogenase 2 (IDH2). In confirmation of the reverse flux through IDH2, the increased reductive metabolism of glutamine-derived α-ketoglutarate in hypoxia was associated with increased synthesis of 2HG. Finally, constitutive HIF1α-expressing cells also demonstrated significant reductive-carboxylation-dependent synthesis of citrate in normoxia and a relative defect in the oxidative conversion of glutamine into citrate. Collectively, the data demonstrate that mitochondrial glutamine metabolism can be rerouted through IDH2-dependent citrate synthesis in support of hypoxic cell growth.

Some Cancer Cells Can Proliferate at 0.5% O2 Despite a Sharp Decline in Glucose-Dependent Citrate Synthesis.

At 21% O2, cancer cells have been shown to synthesize citrate by condensing glucose-derived acetyl-CoA with glutamine-derived oxaloacetate through the activity of the canonical TCA cycle enzyme citrate synthase (4). In contrast, less is known regarding the synthesis of citrate by cells that can continue proliferating in hypoxia. The glioblastoma cell line SF188 is able to proliferate at 0.5% O2 (Fig. 1A), a level of hypoxia that is sufficient to stabilize HIF1α (Fig. 1B) and predicted to limit respiration (1213). Consistent with previous observations in hypoxic cells, we found that SF188 cells demonstrated increased lactate production when incubated in hypoxia (Fig. 1C), and the ratio of lactate produced to glucose consumed increased demonstrating an increase in the rate of anaerobic glycolysis. When glucose-derived carbon in the form of pyruvate is converted to lactate, it is diverted away from subsequent metabolism that can contribute to citrate production. However, we observed that SF188 cells incubated in hypoxia maintain their intracellular citrate to ∼75% of the level maintained under normoxia (Fig. 1D). This prompted an investigation of how proliferating cells maintain citrate production under hypoxia.

SF188 glioblastoma cells proliferate at 0.5% O2 despite a profound reduction in glucose-dependent citrate synthesis.

SF188 glioblastoma cells proliferate at 0.5% O2 despite a profound reduction in glucose-dependent citrate synthesis.

Fig. 1. SF188 glioblastoma cells proliferate at 0.5% O2 despite a profound reduction in glucose-dependent citrate synthesis. (A) SF188 cells were plated in complete medium equilibrated with 21% O2 (Normoxia) or 0.5% O2 (Hypoxia), total viable cells were counted 24 h and 48 h later (Day 1 and Day 2), and population doublings were calculated. Data are the mean ± SEM of four independent experiments. (B) Western blot demonstrates stabilized HIF1α protein in cells cultured in hypoxia compared with normoxia. (C) Cells were grown in normoxia or hypoxia for 24 h, after which culture medium was collected. Medium glucose and lactate levels were measured and compared with the levels in fresh medium. (D) Cells were cultured for 24 h as in C. Intracellular metabolism was then quenched with 80% MeOH prechilled to −80 °C that was spiked with a 13C-labeled citrate as an internal standard. Metabolites were then extracted, and intracellular citrate levels were analyzed with GC-MS and normalized to cell number. Data for C and D are the mean ± SEM of three independent experiments. (E) Model depicting the pathway for cit+2 production from [U-13C]glucose. Glucose uniformly 13C-labeled will generate pyruvate+3. Pyruvate+3 can be oxidatively decarboxylated by PDH to produce acetyl-CoA+2, which can condense with unlabeled oxaloacetate to produce cit+2. (F) Cells were cultured for 24 h as in C and D, followed by an additional 4 h of culture in glucose-deficient medium supplemented with 10 mM [U-13C]glucose. Intracellular metabolites were then extracted, and 13C-enrichment in cellular citrate was analyzed by GC-MS and normalized to the total citrate pool size. Data are the mean ± SD of three independent cultures from a representative of two independent experiments. *P < 0.05, ***P < 0.001.

Increased glucose uptake and glycolytic metabolism are critical elements of the metabolic response to hypoxia. To evaluate the contributions made by glucose to the citrate pool under normoxia or hypoxia, SF188 cells incubated in normoxia or hypoxia were cultured in medium containing 10 mM [U-13C]glucose. Following a 4-h labeling period, cellular metabolites were extracted and analyzed for isotopic enrichment by gas chromatography-mass spectrometry (GC-MS). In normoxia, the major 13C-enriched citrate species found was citrate enriched with two 13C atoms (cit+2), which can arise from the NAD+-dependent decarboxylation of pyruvate+3 to acetyl-CoA+2 by PDH, followed by the condensation of acetyl-CoA+2 with unenriched oxaloacetate (Fig. 1 E and F). Compared with the accumulation of cit+2, we observed minimal accumulation of cit+3 and cit+5 under normoxia. Cit+3 arises from pyruvate carboxylase (PC)-dependent conversion of pyruvate+3 to oxaloacetate+3, followed by the condensation of oxaloacetate+3 with unenriched acetyl-CoA. Cit+5 arises when PC-generated oxaloacetate+3 condenses with PDH-generated acetyl-CoA+2. The lack of cit+3 and cit+5 accumulation is consistent with PC activity not playing a major role in citrate production in normoxic SF188 cells, as reported (4).

In hypoxic cells, the major citrate species observed was unenriched. Cit+2, cit+3, and cit+5 all constituted minor fractions of the total citrate pool, consistent with glucose carbon not being incorporated into citrate through either PDH or PC-mediated metabolism under hypoxic conditions (Fig. 1F). These data demonstrate that in contrast to normoxic cells, where a large percentage of citrate production depends on glucose-derived carbon, hypoxic cells significantly reduce their rate of citrate production from glucose.

Glutamine Carbon Metabolism Is Required for Viability in Hypoxia.

In addition to glucose, we have previously reported that glutamine can contribute to citrate production during cell growth under normoxic conditions (4). Surprisingly, under hypoxic conditions, we observed that SF188 cells retained their high rate of glutamine consumption (Fig. 2A). Moreover, hypoxic cells cultured in glutamine-deficient medium displayed a significant loss of viability (Fig. 2B). In normoxia, the requirement for glutamine to maintain viability of SF188 cells can be satisfied by α-ketoglutarate, the downstream metabolite of glutamine that is devoid of nitrogenous groups (14). α-ketoglutarate cannot fulfill glutamine’s roles as a nitrogen source for nonessential amino acid synthesis or as an amide donor for nucleotide or hexosamine synthesis, but can be metabolized through the oxidative TCA cycle to regenerate oxaloacetate, and subsequently condense with glucose-derived acetyl-CoA to produce citrate. To test whether the restoration of carbon from glutamine metabolism in the form of α-ketoglutarate could rescue the viability defect of glutamine-starved SF188 cells even under hypoxia, SF188 cells incubated in hypoxia were cultured in glutamine-deficient medium supplemented with a cell-penetrant form of α-ketoglutarate (dimethyl α-ketoglutarate). The addition of dimethyl α-ketoglutarate rescued the defect in cell viability observed upon glutamine withdrawal (Fig. 2B). These data demonstrate that, even under hypoxic conditions, when the ability of glutamine to replenish oxaloacetate through oxidative TCA cycle metabolism is diminished, SF188 cells retain their requirement for glutamine as the carbon backbone for α-ketoglutarate. This result raised the possibility that glutamine could be the carbon source for citrate production through an alternative, nonoxidative, pathway in hypoxia.

Glutamine carbon is required for hypoxic cell viability

Glutamine carbon is required for hypoxic cell viability

Glutamine carbon is required for hypoxic cell viability

Fig. 2. Glutamine carbon is required for hypoxic cell viability and contributes to increased citrate production through reductive carboxylation relative to oxidative metabolism in hypoxia. (A) SF188 cells were cultured for 24 h in complete medium equilibrated with either 21% O2 (Normoxia) or 0.5% O2(Hypoxia). Culture medium was then removed from cells and analyzed for glutamine levels which were compared with the glutamine levels in fresh medium. Data are the mean ± SEM of three independent experiments. (B) The requirement for glutamine to maintain hypoxic cell viability can be satisfied by α-ketoglutarate. Cells were cultured in complete medium equilibrated with 0.5% O2 for 24 h, followed by an additional 48 h at 0.5% O2 in either complete medium (+Gln), glutamine-deficient medium (−Gln), or glutamine-deficient medium supplemented with 7 mM dimethyl α-ketoglutarate (−Gln +αKG). All medium was preconditioned in 0.5% O2. Cell viability was determined by trypan blue dye exclusion. Data are the mean and range from two independent experiments. (C) Model depicting the pathways for cit+4 and cit+5 production from [U-13C]glutamine (glutamine+5). Glutamine+5 is catabolized to α-ketoglutarate+5, which can then contribute to citrate production by two divergent pathways. Oxidative metabolism produces oxaloacetate+4, which can condense with unlabeled acetyl-CoA to produce cit+4. Alternatively, reductive carboxylation produces isocitrate+5, which can isomerize to cit+5. (D) Glutamine contributes to citrate production through increased reductive carboxylation relative to oxidative metabolism in hypoxic proliferating cancer cells. Cells were cultured for 24 h as in A, followed by 4 h of culture in glutamine-deficient medium supplemented with 4 mM [U-13C]glutamine. 13C enrichment in cellular citrate was quantitated with GC-MS. Data are the mean ± SD of three independent cultures from a representative of three independent experiments. **P < 0.01.

Cells Proliferating in Hypoxia Maintain Levels of Additional Metabolites Through Reductive Carboxylation.

Previous work has documented that, in normoxic conditions, SF188 cells use glutamine as the primary anaplerotic substrate, maintaining the pool sizes of TCA cycle intermediates through oxidative metabolism (4). Surprisingly, we found that, when incubated in hypoxia, SF188 cells largely maintained their levels of aspartate (in equilibrium with oxaloacetate), malate, and fumarate (Fig. 3A). To distinguish how glutamine carbon contributes to these metabolites in normoxia and hypoxia, SF188 cells incubated in normoxia or hypoxia were cultured in medium containing 4 mM [U-13C]glutamine. After a 4-h labeling period, metabolites were extracted and the intracellular pools of aspartate, malate, and fumarate were analyzed by GC-MS.

In normoxia, the majority of the enriched intracellular asparatate, malate, and fumarate were the +4 species, which arise through oxidative metabolism of glutamine-derived α-ketoglutarate (Fig. 3 B and C). The +3 species, which can be derived from the citrate generated by the reductive carboxylation of glutamine-derived α-ketoglutarate, constituted a significantly lower percentage of the total aspartate, malate, and fumarate pools. By contrast, in hypoxia, the +3 species constituted a larger percentage of the total aspartate, malate, and fumarate pools than they did in normoxia. These data demonstrate that, in addition to citrate, hypoxic cells preferentially synthesize oxaloacetate, malate, and fumarate through the pathway of reductive carboxylation rather than the oxidative TCA cycle.

IDH2 Is Critical in Hypoxia for Reductive Metabolism of Glutamine and for Cell Proliferation.

We hypothesized that the relative increase in reductive carboxylation we observed in hypoxia could arise from the suppression of α-ketoglutarate oxidation through the TCA cycle. Consistent with this, we found that α-ketoglutarate levels increased in SF188 cells following 24 h in hypoxia (Fig. 4A). Surprisingly, we also found that levels of the closely related metabolite 2-hydroxyglutarate (2HG) increased in hypoxia, concomitant with the increase in α-ketoglutarate under these conditions. 2HG can arise from the noncarboxylating reduction of α-ketoglutarate (Fig. 4B). Recent work has found that specific cancer-associated mutations in the active sites of either IDH1 or IDH2 lead to a 10- to 100-fold enhancement in this activity facilitating 2HG production (1517), but SF188 cells lack IDH1/2 mutations. However, 2HG levels are also substantially elevated in the inborn error of metabolism 2HG aciduria, and the majority of patients with this disease lack IDH1/2 mutations. As 2HG has been demonstrated to arise in these patients from mitochondrial α-ketoglutarate (18), we hypothesized that both the increased reductive carboxylation of glutamine-derived α-ketoglutarate to citrate and the increased 2HG accumulation we observed in hypoxia could arise from increased reductive metabolism by wild-type IDH2 in the mitochondria.

Reductive carboxylation of glutamine-derived α-ketoglutarate to citrate in hypoxic cancer cells is dependent on mitochondrial IDH2

Reductive carboxylation of glutamine-derived α-ketoglutarate to citrate in hypoxic cancer cells is dependent on mitochondrial IDH2

Reductive carboxylation of glutamine-derived α-ketoglutarate to citrate in hypoxic cancer cells is dependent on mitochondrial IDH2

Fig. 4. Reductive carboxylation of glutamine-derived α-ketoglutarate to citrate in hypoxic cancer cells is dependent on mitochondrial IDH2. (A) α-ketoglutarate and 2HG increase in hypoxia. SF188 cells were cultured in complete medium equilibrated with either 21% O2 (Normoxia) or 0.5% O2 (Hypoxia) for 24 h. Intracellular metabolites were then extracted, cell extracts spiked with a 13C-labeled citrate as an internal standard, and intracellular α-ketoglutarate and 2HG levels were analyzed with GC-MS. Data shown are the mean ± SEM of three independent experiments. (B) Model for reductive metabolism from glutamine-derived α-ketoglutarate. Glutamine+5 is catabolized to α-ketoglutarate+5. Carboxylation of α-ketoglutarate+5 followed by reduction of the carboxylated intermediate (reductive carboxylation) will produce isocitrate+5, which can then isomerize to cit+5. In contrast, reductive activity on α-ketoglutarate+5 that is uncoupled from carboxylation will produce 2HG+5. (C) IDH2 is required for reductive metabolism of glutamine-derived α-ketoglutarate in hypoxia. SF188 cells transfected with a siRNA against IDH2 (siIDH2) or nontargeting negative control (siCTRL) were cultured for 2 d in complete medium equilibrated with 0.5% O2. (Upper) Cells were then cultured at 0.5% O2 for an additional 4 h in glutamine-deficient medium supplemented with 4 mM [U-13C]glutamine. 13C enrichment in intracellular citrate and 2HG was determined and normalized to the relevant metabolite total pool size. (Lower) Cells transfected and cultured in parallel at 0.5% O2 were counted by hemacytometer (excluding nonviable cells with trypan blue staining) or harvested for protein to assess IDH2 expression by Western blot. Data shown for GC-MS and cell counts are the mean ± SD of three independent cultures from a representative experiment. **P < 0.01, ***P < 0.001.

In an experiment to test this hypothesis, SF188 cells were transfected with either siRNA directed against mitochondrial IDH2 (siIDH2) or nontargeting control, incubated in hypoxia for 2 d, and then cultured for another 4 h in hypoxia in media containing 4 mM [U-13C]glutamine. After the labeling period, metabolites were extracted and analyzed by GC-MS (Fig. 4C). Hypoxic SF188 cells transfected with siIDH2 displayed a decreased contribution of cit+5 to the total citrate pool, supporting an important role for IDH2 in the reductive carboxylation of glutamine-derived α-ketoglutarate in hypoxic conditions. The contribution of cit+4 to the total citrate pool did not decrease with siIDH2 treatment, consistent with IDH2 knockdown specifically affecting the pathway of reductive carboxylation and not other fundamental TCA cycle-regulating processes. In confirmation of reverse flux occurring through IDH2, the contribution of 2HG+5 to the total 2HG pool decreased in siIDH2-treated cells. Supporting the importance of citrate production by IDH2-mediated reductive carboxylation for hypoxic cell proliferation, siIDH2-transfected SF188 cells displayed a defect in cellular accumulation in hypoxia. Decreased expression of IDH2 protein following siIDH2 transfection was confirmed by Western blot. Collectively, these data point to the importance of mitochondrial IDH2 for the increase in reductive carboxylation flux of glutamine-derived α-ketoglutarate to maintain citrate levels in hypoxia, and to the importance of this reductive pathway for hypoxic cell proliferation.

Reprogramming of Metabolism by HIF1 in the Absence of Hypoxia Is Sufficient to Induce Increased Citrate Synthesis by Reductive Carboxylation Relative to Oxidative Metabolism.

The relative increase in the reductive metabolism of glutamine-derived α-ketoglutarate at 0.5% O2 may be explained by the decreased ability to carry out oxidative NAD+-dependent reactions as respiration is inhibited (1213). However, a shift to preferential reductive glutamine metabolism could also result from the active reprogramming of cellular metabolism by HIF1 (810), which inhibits the generation of mitochondrial acetyl-CoA necessary for the synthesis of citrate by oxidative glucose and glutamine metabolism (Fig. 5A). To better understand the role of HIF1 in reductive glutamine metabolism, we used VHL-deficient RCC4 cells, which display constitutive expression of HIF1α under normoxia (Fig. 5B). RCC4 cells expressing either a nontargeting control shRNA (shCTRL) or an shRNA directed at HIF1α (shHIF1α) were incubated in normoxia and cultured in medium with 4 mM [U-13C]glutamine. Following a 4-h labeling period, metabolites were extracted and the cellular citrate pool was analyzed by GC-MS. In shCTRL cells, which have constitutive HIF1α expression despite incubation in normoxia, the majority of the total citrate pool was constituted by the cit+5 species, with low levels of all other species including cit+4 (Fig. 5C). By contrast, in HIF1α-deficient cells the contribution of cit+5 to the total citrate pool was greatly decreased, whereas the contribution of cit+4 to the total citrate pool increased and was the most abundant citrate species. These data demonstrate that the relative enhancement of the reductive carboxylation pathway for citrate synthesis can be recapitulated by constitutive HIF1 activation in normoxia.

Reprogramming of metabolism by HIF1 in the absence of hypoxia

Reprogramming of metabolism by HIF1 in the absence of hypoxia

Reprogramming of metabolism by HIF1 in the absence of hypoxia is sufficient to induce reductive carboxylation of glutamine-derived α-ketoglutarate.

Fig. 5. Reprogramming of metabolism by HIF1 in the absence of hypoxia is sufficient to induce reductive carboxylation of glutamine-derived α-ketoglutarate. (A) Model depicting how HIF1 signaling’s inhibition of pyruvate dehydrogenase (PDH) activity and promotion of lactate dehydrogenase-A (LDH-A) activity can block the generation of mitochondrial acetyl-CoA from glucose-derived pyruvate, thereby favoring citrate synthesis from reductive carboxylation of glutamine-derived α-ketoglutarate. (B) Western blot demonstrating HIF1α protein in RCC4 VHL−/− cells in normoxia with a nontargeting shRNA (shCTRL), and the decrease in HIF1α protein in RCC4 VHL−/− cells stably expressing HIF1α shRNA (shHIF1α). (C) HIF1-induced reprogramming of glutamine metabolism. Cells from B at 21% O2 were cultured for 4 h in glutamine-deficient medium supplemented with 4 mM [U-13C]glutamine. Intracellular metabolites were then extracted, and 13C enrichment in cellular citrate was determined by GC-MS. Data shown are the mean ± SD of three independent cultures from a representative of three independent experiments. ***P < 0.001.

Compared with glucose metabolism, much less is known regarding how glutamine metabolism is altered under hypoxia. It has also remained unclear how hypoxic cells can maintain the citrate production necessary for macromolecular biosynthesis. In this report, we demonstrate that in contrast to cells at 21% O2, where citrate is predominantly synthesized through oxidative metabolism of both glucose and glutamine, reductive carboxylation of glutamine carbon becomes the major pathway of citrate synthesis in cells that can effectively proliferate at 0.5% O2. Moreover, we show that in these hypoxic cells, reductive carboxylation of glutamine-derived α-ketoglutarate is dependent on mitochondrial IDH2. Although others have previously suggested the existence of reductive carboxylation in cancer cells (1920), these studies failed to demonstrate the intracellular localization or specific IDH isoform responsible for the reductive carboxylation flux. Recently, we identified IDH2 as an isoform that contributes to reductive carboxylation in cancer cells incubated at 21% O2 (16), but remaining unclear were the physiological importance and regulation of this pathway relative to oxidative metabolism, as well as the conditions where this reductive pathway might be advantageous for proliferating cells.

Here we report that IDH2-mediated reductive carboxylation of glutamine-derived α-ketoglutarate to citrate is an important feature of cells proliferating in hypoxia. Moreover, the reliance on reductive glutamine metabolism can be recapitulated in normoxia by constitutive HIF1 activation in cells with loss of VHL. The mitochondrial NADPH/NADP+ ratio required to fuel the reductive reaction through IDH2 can arise from the increased NADH/NAD+ ratio existing in the mitochondria under hypoxic conditions (2122), with the transfer of electrons from NADH to NADP+ to generate NADPH occurring through the activity of the mitochondrial transhydrogenase (23). Our data do not exclude a complementary role for cytosolic IDH1 in impacting reductive glutamine metabolism, potentially through its oxidative function in an IDH2/IDH1 shuttle that transfers high energy electrons in the form of NADPH from mitochondria to cytosol (1624).

In further support of the increased mitochondrial reductive glutamine metabolism that we observe in hypoxia, we report here that incubation in hypoxia can lead to elevated 2HG levels in cells lacking IDH1/2 mutations. 2HG production from glutamine-derived α-ketoglutarate significantly decreased with knockdown of IDH2, supporting the conclusion that 2HG is produced in hypoxia by enhanced reverse flux of α-ketoglutarate through IDH2 in a truncated, noncarboxylating reductive reaction. However, other mechanisms may also contribute to 2HG elevation in hypoxia. These include diminished oxidative activity and/or enhanced reductive activity of the 2HG dehydrogenase, a mitochondrial enzyme that normally functions to oxidize 2HG back to α-ketoglutarate (25). The level of 2HG elevation we observe in hypoxic cells is associated with a concomitant increase in α-ketoglutarate, and is modest relative to that observed in cancers with IDH1/2 gain-of-function mutations. Nonetheless, 2HG elevation resulting from hypoxia in cells with wild-type IDH1/2 may hold promise as a cellular or serum biomarker for tissues undergoing chronic hypoxia and/or excessive glutamine metabolism.

The IDH2-dependent reductive carboxylation pathway that we propose in this report allows for continued citrate production from glutamine carbon when hypoxia and/or HIF1 activation prevents glucose carbon from contributing to citrate synthesis. Moreover, as opposed to continued oxidative TCA cycle functioning in hypoxia which can increase reactive oxygen species (ROS), reductive carboxylation of α-ketoglutarate in the mitochondria may serve as an electron sink that decreases the generation of ROS. HIF1 activity is not limited to the setting of hypoxia, as a common feature of several cancers is the normoxic stabilization of HIF1α through loss of the VHL tumor suppressor or other mechanisms. We demonstrate here that altered glutamine metabolism through a mitochondrial reductive pathway is a central aspect of hypoxic proliferating cell metabolism and HIF1-induced metabolic reprogramming. These findings are relevant for the understanding of numerous constitutive HIF1-expressing malignancies, as well as for populations, such as stem progenitor cells, which frequently proliferate in hypoxic conditions.

7.9.3 Hypoxia-Inducible Factors in Physiology and Medicine

Gregg L. Semenza
Cell. 2012 Feb 3; 148(3): 399–408.

Oxygen homeostasis represents an organizing principle for understanding metazoan evolution, development, physiology, and pathobiology. The hypoxia-inducible factors (HIFs) are transcriptional activators that function as master regulators of oxygen homeostasis in all metazoan species. Rapid progress is being made in elucidating homeostatic roles of HIFs in many physiological systems, determining pathological consequences of HIF dysregulation in chronic diseases, and investigating potential targeting of HIFs for therapeutic purposes. Oxygen homeostasis represents an organizing principle for understanding metazoan evolution, development, physiology, and pathobiology. The hypoxia-inducible factors (HIFs) are transcriptional activators that function as master regulators of oxygen homeostasis in all metazoan species. Rapid progress is being made in elucidating homeostatic roles of HIFs in many physiological systems, determining pathological consequences of HIF dysregulation in chronic diseases, and investigating potential targeting of HIFs for therapeutic purposes.


Oxygen is central to biology because of its utilization in the process of respiration. O2 serves as the final electron acceptor in oxidative phosphorylation, which carries with it the risk of generating reactive oxygen species (ROS) that react with cellular macromolecules and alter their biochemical or physical properties, resulting in cell dysfunction or death. As a consequence, metazoan organisms have evolved elaborate cellular metabolic and systemic physiological systems that are designed to maintain oxygen homeostasis. This review will focus on the role of hypoxia-inducible factors (HIFs) as master regulators of oxygen homeostasis and, in particular, on recent advances in understanding their roles in physiology and medicine. Due to space limitations and the remarkably pleiotropic effects of HIFs, the description of such roles will be illustrative rather than comprehensive.

O2 and Evolution, Part 1

Accumulation of O2 in Earth’s atmosphere starting ~2.5 billion years ago led to evolution of the extraordinarily efficient system of oxidative phosphorylation that transfers chemical energy stored in carbon bonds of organic molecules to the high-energy phosphate bond in ATP, which is used to power physicochemical reactions in living cells. Energy produced by mitochondrial respiration is sufficient to power the development and maintenance of multicellular organisms, which could not be sustained by energy produced by glycolysis alone (Lane and Martin, 2010). The modest dimensions of primitive metazoan species were such that O2 could diffuse from the atmosphere to all of the organism’s thousand cells, as is the case for the worm Caenorhabditis elegans. To escape the constraints placed on organismal growth by diffusion, systems designed to conduct air to cells deep within the body evolved and were sufficient for O2delivery to organisms with hundreds of thousands of cells, such as the fly Drosophila melanogaster. The final leap in body scale occurred in vertebrates and was associated with the evolution of complex respiratory, circulatory, and nervous systems designed to efficiently capture and distribute O2 to hundreds of millions of millions of cells in the case of the adult Homo sapiens.

Hypoxia-Inducible Factors

Hypoxia-inducible factor 1 (HIF-1) is expressed by all extant metazoan species analyzed (Loenarz et al., 2011). HIF-1 consists of HIF-1α and HIF-1β subunits, which each contain basic helix-loop-helix-PAS (bHLH-PAS) domains (Wang et al., 1995) that mediate heterodimerization and DNA binding (Jiang et al., 1996a). HIF-1β heterodimerizes with other bHLH-PAS proteins and is present in excess, such that HIF-1α protein levels determine HIF-1 transcriptional activity (Semenza et al., 1996).

Under well-oxygenated conditions, HIF-1α is bound by the von Hippel-Lindau (VHL) protein, which recruits an ubiquitin ligase that targets HIF-1α for proteasomal degradation (Kaelin and Ratcliffe, 2008). VHL binding is dependent upon hydroxylation of a specific proline residue in HIF-1α by the prolyl hydroxylase PHD2, which uses O2 as a substrate such that its activity is inhibited under hypoxic conditions (Epstein et al., 2001). In the reaction, one oxygen atom is inserted into the prolyl residue and the other atom is inserted into the co-substrate α-ketoglutarate, splitting it into CO2 and succinate (Kaelin and Ratcliffe, 2008). Factor inhibiting HIF-1 (FIH-1) represses HIF-1α transactivation function (Mahon et al., 2001) by hydroxylating an asparaginyl residue, using O2 and α-ketoglutarate as substrates, thereby blocking the association of HIF-1α with the p300 coactivator protein (Lando et al., 2002). Dimethyloxalylglycine (DMOG), a competitive antagonist of α-ketoglutarate, inhibits the hydroxylases and induces HIF-1-dependent transcription (Epstein et al., 2001). HIF-1 activity is also induced by iron chelators (such as desferrioxamine) and cobalt chloride, which inhibit hydroxylases by displacing Fe(II) from the catalytic center (Epstein et al., 2001).

Studies in cultured cells (Jiang et al., 1996b) and isolated, perfused, and ventilated lung preparations (Yu et al., 1998) revealed an exponential increase in HIF-1α levels at O2 concentrations less than 6% (~40 mm Hg), which is not explained by known biochemical properties of the hydroxylases. In most adult tissues, O2concentrations are in the range of 3-5% and any decrease occurs along the steep portion of the dose-response curve, allowing a graded response to hypoxia. Analyses of cultured human cells have revealed that expression of hundreds of genes was increased in response to hypoxia in a HIF-1-dependent manner (as determined by RNA interference) with direct binding of HIF-1 to the gene (as determined by chromatin immunoprecipitation [ChIP] assays); in addition, the expression of hundreds of genes was decreased in response to hypoxia in a HIF-1-dependent manner but binding of HIF-1 to these genes was not detected (Mole et al., 2009), indicating that HIF-dependent repression occurs via indirect mechanisms, which include HIF-1-dependent expression of transcriptional repressors (Yun et al., 2002) and microRNAs (Kulshreshtha et al., 2007). ChIP-seq studies have revealed that only 40% of HIF-1 binding sites are located within 2.5 kb of the transcription start site (Schödel et al., 2011).

In vertebrates, HIF-2α is a HIF-1α paralog that is also regulated by prolyl and asparaginyl hydroxylation and dimerizes with HIF-1β, but is expressed in a cell-restricted manner and plays important roles in erythropoiesis, vascularization, and pulmonary development, as described below. In D. melanogaster, the gene encoding the HIF-1α ortholog is designated similar and its paralog is designated trachealess because inactivating mutations result in defective development of the tracheal tubes (Wilk et al., 1996). In contrast, C. elegans has only a single HIF-1α homolog (Epstein et al., 2001). Thus, in both invertebrates and vertebrates, evolution of specialized systems for O2 delivery was associated with the appearance of a HIF-1α paralog.

O2 and Metabolism

The regulation of metabolism is a principal and primordial function of HIF-1. Under hypoxic conditions, HIF-1 mediates a transition from oxidative to glycolytic metabolism through its regulation of: PDK1, encoding pyruvate dehydrogenase (PDH) kinase 1, which phosphorylates and inactivates PDH, thereby inhibiting the conversion of pyruvate to acetyl coenzyme A for entry into the tricarboxylic acid cycle (Kim et al., 2006Papandreou et al., 2006); LDHA, encoding lactate dehydrogenase A, which converts pyruvate to lactate (Semenza et al. 1996); and BNIP3 (Zhang et al. 2008) and BNIP3L (Bellot et al., 2009), which mediate selective mitochondrial autophagy (Figure 1). HIF-1 also mediates a subunit switch in cytochrome coxidase that improves the efficiency of electron transfer under hypoxic conditions (Fukuda et al., 2007). An analogous subunit switch is also observed in Saccharomyces cerevisiae, although it is mediated by a completely different mechanism (yeast lack HIF-1), suggesting that it may represent a fundamental response of eukaryotic cells to hypoxia.

Regulation of Glucose Metabolism nihms-350382-f0001

Regulation of Glucose Metabolism nihms-350382-f0001

Regulation of Glucose Metabolism
Figure 1
Regulation of Glucose Metabolism

It is conventional wisdom that cells switch to glycolysis when O2 becomes limiting for mitochondrial ATP production. Yet, HIF-1α-null mouse embryo fibroblasts, which do not down-regulate respiration under hypoxic conditions, have higher ATP levels at 1% O2 than wild-type cells at 20% O2, demonstrating that under these conditions O2 is not limiting for ATP production (Zhang et al., 2008). However, the HIF-1α-null cells die under prolonged hypoxic conditions due to ROS toxicity (Kim et al. 2006Zhang et al., 2008). These studies have led to a paradigm shift with regard to our understanding of the regulation of cellular metabolism (Semenza, 2011): the purpose of this switch is to prevent excess mitochondrial generation of ROS that would otherwise occur due to the reduced efficiency of electron transfer under hypoxic conditions (Chandel et al., 1998). This may be particularly important in stem cells, in which avoidance of DNA damage is critical (Suda et al., 2011).

Role of HIFs in Development

Much of mammalian embryogenesis occurs at O2 concentrations of 1-5% and O2 functions as a morphogen (through HIFs) in many developmental systems (Dunwoodie, 2009). Mice that are homozygous for a null allele at the locus encoding HIF-1α die by embryonic day 10.5 with cardiac malformations, vascular defects, and impaired erythropoiesis, indicating that all three components of the circulatory system are dependent upon HIF-1 for normal development (Iyer et al., 1998Yoon et al., 2011). Depending on the genetic background, mice lacking HIF-2α: die by embryonic day 12.5 with vascular defects (Peng et al., 2000) or bradycardia due to deficient catecholamine production (Tian et al., 1998); die as neonates due to impaired lung maturation (Compernolle et al., 2002); or die several months after birth due to ROS-mediated multi-organ failure (Scortegagna et al., 2003). Thus, while vertebrate evolution was associated with concomitant appearance of the circulatory system and HIF-2α, both HIF-1 and HIF-2 have important roles in circulatory system development. Conditional knockout of HIF-1α in specific cell types has demonstrated important roles in chondrogenesis (Schipani et al., 2001), adipogenesis (Yun et al., 2002), B-lymphocyte development (Kojima et al., 2002), osteogenesis (Wang et al., 2007), hematopoiesis (Takubo et al., 2010), T-lymphocyte differentiation (Dang et al., 2011), and innate immunity (Zinkernagel et al., 2007). While knockout mouse experiments point to the adverse effects of HIF-1 loss-of-function on development, it is also possible that increased HIF-1 activity, induced by hypoxia in embryonic tissues as a result of abnormalities in placental blood flow, may also dysregulate development and result in congenital malformations. For example, HIF-1α has been shown to interact with, and stimulate the transcriptional activity of, Notch, which plays a key role in many developmental pathways (Gustafsson et al., 2005).

Translational Prospects

Drug discovery programs have been initiated at many pharmaceutical and biotech companies to develop prolyl hydroxylase inhibitors (PHIs) that, as described above for DMOG, induce HIF activity for treatment of disorders in which HIF mediates protective physiological responses. Local and/or short term induction of HIF activity by PHIs, gene therapy, or other means are likely to be useful novel therapies for many of the diseases described above. In the case of ischemic cardiovascular disease, local therapy is needed to provide homing signals for the recruitment of BMDACs. Chronic systemic use of PHIs must be approached with great caution: individuals with genetic mutations that constitutively activate the HIF pathway (described below) have increased incidence of cardiovascular disease and mortality (Yoon et al., 2011). On the other hand, the profound inhibition of HIF activity and vascular responses to ischemia that are associated with aging suggest that systemic replacement therapy might be contemplated as a preventive measure for subjects in whom impaired HIF responses to hypoxia can be documented. In C. elegans, VHL loss-of-function increases lifespan in a HIF-1-dependent manner (Mehta et al., 2009), providing further evidence for a mutually antagonistic relationship between HIF-1 and aging.


Cancers contain hypoxic regions as a result of high rates of cell proliferation coupled with the formation of vasculature that is structurally and functionally abnormal. Increased HIF-1α and/or HIF-2α levels in diagnostic tumor biopsies are associated with increased risk of mortality in cancers of the bladder, brain, breast, colon, cervix, endometrium, head/neck, lung, ovary, pancreas, prostate, rectum, and stomach; these results are complemented by experimental studies, which demonstrate that genetic manipulations that increase HIF-1α expression result in increased tumor growth, whereas loss of HIF activity results in decreased tumor growth (Semenza, 2010). HIFs are also activated by genetic alterations, most notably, VHL loss of function in clear cell renal carcinoma (Majmunder et al., 2010). HIFs activate transcription of genes that play key roles in critical aspects of cancer biology, including stem cell maintenance (Wang et al., 2011), cell immortalization, epithelial-mesenchymal transition (Mak et al., 2010), genetic instability (Huang et al., 2007), vascularization (Liao and Johnson, 2007), glucose metabolism (Luo et al., 2011), pH regulation (Swietach et al., 2007), immune evasion (Lukashev et al., 2007), invasion and metastasis (Chan and Giaccia, 2007), and radiation resistance (Moeller et al., 2007). Given the extensive validation of HIF-1 as a potential therapeutic target, drugs that inhibit HIF-1 have been identified and shown to have anti-cancer effects in xenograft models (Table 1Semenza, 2010).

Table 1  Drugs that Inhibit HIF-1

Process Inhibited Drug Class Prototype
HIF-1 α synthesis Cardiac glycosidemTOR inhibitorMicrotubule targeting agent

Topoisomerase I inhibitor



HIF-1 α protein stability HDAC inhibitorHSP90 inhibitorCalcineurin inhibitor

Guanylate cyclase activator



Heterodimerization Antimicrobial agent Acriflavine
DNA binding AnthracyclineQuinoxaline antibiotic DoxorubicinEchinomycin
Transactivation Proteasome inhibitorAntifungal agent BortezomibAmphotericin B
Signal transduction BCR-ABL inhibitorCyclooxygenase inhibitorEGFR inhibitor

HER2 inhibitor

ImatinibIbuprofenErlotinib, Gefitinib


Over 100 women die every day of breast cancer in the U.S. The mean PO2 is 10 mm Hg in breast cancer as compared to > 60 mm Hg in normal breast tissue and cancers with PO2 < 10 mm Hg are associated with increased risk of metastasis and patient mortality (Vaupel et al., 2004). Increased HIF-1α protein levels, as identified by immunohistochemical analysis of tumor biopsies, are associated with increased risk of metastasis and/or patient mortality in unselected breast cancer patients and in lymph node-positive, lymph node-negative, HER2+, or estrogen receptor+ subpopulations (Semenza, 2011). Metastasis is responsible for > 90% of breast cancer mortality. The requirement for HIF-1 in breast cancer metastasis has been demonstrated for both autochthonous tumors in transgenic mice (Liao et al., 2007) and orthotopic transplants in immunodeficient mice (Zhang et al., 2011Wong et al., 2011). Primary tumors direct the recruitment of bone marrow-derived cells to the lungs and other sites of metastasis (Kaplan et al., 2005). In breast cancer, hypoxia induces the expression of lysyl oxidase (LOX), a secreted protein that remodels collagen at sites of metastatic niche formation (Erler et al., 2009). In addition to LOX, breast cancers also express LOX-like proteins 2 and 4. LOX, LOXL2, and LOXL4 are all HIF-1-regulated genes and HIF-1 inhibition blocks metastatic niche formation regardless of which LOX/LOXL protein is expressed, whereas available LOX inhibitors are not effective against all LOXL proteins (Wong et al., 2011), again illustrating the role of HIF-1 as a master regulator that controls the expression of multiple genes involved in a single (patho)physiological process.

Translational Prospects

Small molecule inhibitors of HIF activity that have anti-cancer effects in mouse models have been identified (Table 1). Inhibition of HIF impairs both vascular and metabolic adaptations to hypoxia, which may decrease O2 delivery and increase O2 utilization. These drugs are likely to be useful (as components of multidrug regimens) in the treatment of a subset of cancer patients in whom high HIF activity is driving progression. As with all novel cancer therapeutics, successful translation will require the development of methods for identifying the appropriate patient cohort. Effects of combination drug therapy also need to be considered. VEGF receptor tyrosine kinase inhibitors, which induce tumor hypoxia by blocking vascularization, have been reported to increase metastasis in mouse models (Ebos et al., 2009), which may be mediated by HIF-1; if so, combined use of HIF-1 inhibitors with these drugs may prevent unintended counter-therapeutic effects.

HIF inhibitors may also be useful in the treatment of other diseases in which dysregulated HIF activity is pathogenic. Proof of principle has been established in mouse models of ocular neovascularization, a major cause of blindness in the developed world, in which systemic or intraocular injection of the HIF-1 inhibitor digoxin is therapeutic (Yoshida et al., 2010). Systemic administration of HIF inhibitors for cancer therapy would be contraindicated in patients who also have ischemic cardiovascular disease, in which HIF activity is protective. The analysis of SNPs at the HIF1A locus described above suggests that the population may include HIF hypo-responders, who are at increased risk of severe ischemic cardiovascular disease. It is also possible that HIF hyper-responders, such as individuals with hereditary erythrocytosis, are at increased risk of particularly aggressive cancer.

O2 and Evolution, Part 2

When lowlanders sojourn to high altitude, hypobaric hypoxia induces erythropoiesis, which is a relatively ineffective response because the problem is not insufficient red cells, but rather insufficient ambient O2. Chronic erythrocytosis increases the risk of heart attack, stroke, and fetal loss during pregnancy. Many high-altitude Tibetans maintain the same hemoglobin concentration as lowlanders and yet, despite severe hypoxemia, they also maintain aerobic metabolism. The basis for this remarkable evolutionary adaptation appears to have involved the selection of genetic variants at multiple loci encoding components of the oxygen sensing system, particularly HIF-2α (Beall et al., 2010Simonson et al., 2010Yi et al., 2010). Given that hereditary erythrocytosis is associated with modest HIF-2α gain-of-function, the Tibetan genotype associated with absence of an erythrocytotic response to hypoxia may encode reduced HIF-2α activity along with other alterations that increase metabolic efficiency. Delineating the molecular mechanisms underlying these metabolic adaptations may lead to novel therapies for ischemic disorders, illustrating the importance of oxygen homeostasis as a nexus where evolution, biology, and medicine converge.

7.9.4 Hypoxia-inducible factor 1. Regulator of mitochondrial metabolism and mediator of ischemic preconditioning

Semenza GL1.
Biochim Biophys Acta. 2011 Jul; 1813(7):1263-8.

Hypoxia-inducible factor 1 (HIF-1) mediates adaptive responses to reduced oxygen availability by regulating gene expression. A critical cell-autonomous adaptive response to chronic hypoxia controlled by HIF-1 is reduced mitochondrial mass and/or metabolism. Exposure of HIF-1-deficient fibroblasts to chronic hypoxia results in cell death due to excessive levels of reactive oxygen species (ROS). HIF-1 reduces ROS production under hypoxic conditions by multiple mechanisms including: a subunit switch in cytochrome c oxidase from the COX4-1 to COX4-2 regulatory subunit that increases the efficiency of complex IV; induction of pyruvate dehydrogenase kinase 1, which shunts pyruvate away from the mitochondria; induction of BNIP3, which triggers mitochondrial selective autophagy; and induction of microRNA-210, which blocks assembly of Fe/S clusters that are required for oxidative phosphorylation. HIF-1 is also required for ischemic preconditioning and this effect may be due in part to its induction of CD73, the enzyme that produces adenosine. HIF-1-dependent regulation of mitochondrial metabolism may also contribute to the protective effects of ischemic preconditioning.

The story of life on Earth is a tale of oxygen production and utilization. Approximately 3 billion years ago, primitive single-celled organisms evolved the capacity for photosynthesis, a biochemical process in which photons of solar energy are captured by chlorophyll and used to power the reaction of CO2 and H2O to form glucose and O2. The subsequent rise in the atmospheric O2 concentration over the next billion years set the stage for the ascendance of organisms with the capacity for respiration, a process that consumes glucose and O2 and generates CO2, H2O, and energy in the form of ATP. Some of these single-celled organisms eventually took up residence within the cytoplasm of other cells and devoted all of their effort to energy production as mitochondria. Compared to the conversion of glucose to lactate by glycolysis, the complete oxidation of glucose by respiration provided such a large increase in energy production that it made possible the evolution of multicellular organisms. Among metazoan organisms, the progressive increase in body size during evolution was accompanied by progressively more complex anatomic structures that function to ensure the adequate delivery of O2 to all cells, ultimately resulting in the sophisticated circulatory and respiratory systems of vertebrates.

All metazoan cells can sense and respond to reduced O2 availability (hypoxia). Adaptive responses to hypoxia can be cell autonomous, such as the alterations in mitochondrial metabolism that are described below, or non-cell-autonomous, such as changes in tissue vascularization (reviewed in ref. 1). Primary responses to hypoxia need to be distinguished from secondary responses to sequelae of hypoxia, such as the adaptive responses to ATP depletion that are mediated by AMP kinase (reviewed in ref 2). In contrast, recent data suggest that O2 and redox homeostasis are inextricably linked and that changes in oxygenation are inevitably associated with changes in the levels of reactive oxygen species (ROS), as will be discussed below.

HIF-1 Regulates Oxygen Homeostasis in All Metazoan Species

A key regulator of the developmental and physiological networks required for the maintenance of O2homeostasis is hypoxia-inducible factor 1 (HIF-1). HIF-1 is a heterodimeric transcription factor that is composed of an O2-regulated HIF-1α subunit and a constitutively expressed HIF-1β subunit [3,4]. HIF-1 regulates the expression of hundreds of genes through several major mechanisms. First, HIF-1 binds directly to hypoxia response elements, which are cis-acting DNA sequences located within target genes [5]. The binding of HIF-1 results in the recruitment of co-activator proteins that activate gene transcription (Fig. 1A). Only rarely does HIF-1 binding result in transcriptional repression [6]. Instead, HIF-1 represses gene expression by indirect mechanisms, which are described below. Second, among the genes activated by HIF-1 are many that encode transcription factors [7], which when synthesized can bind to and regulate (either positively or negatively) secondary batteries of target genes (Fig. 1B). Third, another group of HIF-1 target genes encode members of the Jumonji domain family of histone demethylases [8,9], which regulate gene expression by modifying chromatin structure (Fig. 1C). Fourth, HIF-1 can activate the transcription of genes encoding microRNAs [10], which bind to specific mRNA molecules and either block their translation or mediate their degradation (Fig. 1D). Fifth, the isolated HIF-1α subunit can bind to other transcription factors [11,12] and inhibit (Fig. 1E) or potentiate (Fig. 1F) their activity.

Mechanisms by which HIF-1 regulates gene expression. nihms232046f1

Mechanisms by which HIF-1 regulates gene expression. nihms232046f1

Mechanisms by which HIF-1 regulates gene expression.

Fig. 1 Mechanisms by which HIF-1 regulates gene expression. (A) Top: HIF-1 binds directly to target genes at a cis-acting hypoxia response element (HRE) and recruits coactivator proteins such as p300 to increase gene transcription.

HIF-1α and HIF-1β are present in all metazoan species, including the simple roundworm Caenorhabitis elegans [13], which consists of ~103 cells and has no specialized systems for O2 delivery. The fruit flyDrosophila melanogaster evolved tracheal tubes, which conduct air into the interior of the body from which it diffuses to surrounding cells. In vertebrates, the development of the circulatory and respiratory systems was accompanied by the appearance of HIF-2α, which is also O2-regulated and heterodimerizes with HIF-1β [14] but is only expressed in a restricted number of cell types [15], whereas HIF-1α and HIF-1β are expressed in all human and mouse tissues [16]. In Drosophila, the ubiquitiously expressed HIF-1α ortholog is designatedSimilar [17] and the paralogous gene that is expressed specifically in tracheal tubes is designated Trachealess[18].

HIF-1 Activity is Regulated by Oxygen

In the presence of O2, HIF-1α and HIF-2α are subjected to hydroxylation by prolyl-4-hydroxylase domain proteins (PHDs) that use O2 and α-ketoglutarate as substrates and generate CO2 and succinate as by-products [19]. Prolyl hydroxylation is required for binding of the von Hipple-Lindau protein, which recruits a ubiquitin-protein ligase that targets HIF-1α and HIF-2α for proteasomal degradation (Fig. 2). Under hypoxic conditions, the rate of hydroxylation declines and the non-hydroxylated proteins accumulate. HIF-1α transactivation domain function is also O2-regulated [20,21]. Factor inhibiting HIF-1 (FIH-1) represses transactivation domain function [22] by hydroxylating asparagine residue 803 in HIF-1α, thereby blocking the binding of the co-activators p300 and CBP [23].

Negative regulation of HIF-1 activity by oxygen nihms232046f2

Negative regulation of HIF-1 activity by oxygen nihms232046f2

Negative regulation of HIF-1 activity by oxygen

Fig. 2 Negative regulation of HIF-1 activity by oxygen. Top: In the presence of O2: prolyl hydroxylation of HIF-1a leads to binding of the von Hippel-Lindau protein (VHL), which recruits a ubiquitin protein-ligase that targets HIF-1a for proteasomal degradation;

When cells are acutely exposed to hypoxic conditions, the generation of ROS at complex III of the mitochondrial electron transport chain (ETC) increases and is required for the induction of HIF-1α protein levels [24]. More than a decade after these observations were first made, the precise mechanism by which hypoxia increases ROS generation and by which ROS induces HIF-1α accumulation remain unknown. However, the prolyl and asparaginyl hydroxylases contain Fe2+ in their active site and oxidation to Fe3+would block their catalytic activity. Since O2 is a substrate for the hydroxylation reaction, anoxia also results in a loss of enzyme activity. However, the concentration at which O2 becomes limiting for prolyl or asparaginyl hydroxylase activity in vivo is not known.

HIF-1 Regulates the Balance Between Oxidative and Glycolytic Metabolism

All metazoan organisms depend on mitochondrial respiration as the primary mechanism for generating sufficient amounts of ATP to maintain cellular and systemic homeostasis. Respiration, in turn, is dependent on an adequate supply of O2 to serve as the final electron acceptor in the ETC. In this process, electrons are transferred from complex I (or complex II) to complex III, then to complex IV, and finally to O2, which is reduced to water. This orderly transfer of electrons generates a proton gradient across the inner mitochondrial membrane that is used to drive the synthesis of ATP. At each step of this process, some electrons combine with O2 prematurely, resulting in the production of superoxide anion, which is reduced to hydrogen peroxide through the activity of mitochondrial superoxide dismutase. The efficiency of electron transport appears to be optimized to the physiological range of O2 concentrations, such that ATP is produced without the production of excess superoxide, hydrogen peroxide, and other ROS at levels that would result in the increased oxidation of cellular macromolecules and subsequent cellular dysfunction or death. In contrast, when O2levels are acutely increased or decreased, an imbalance between O2 and electron flow occurs, which results in increased ROS production.

MEFs require HIF-1 activity to make two critical metabolic adaptations to chronic hypoxia. First, HIF-1 activates the gene encoding pyruvate dehydrogenase (PDH) kinase 1 (PDK1), which phosphorylates and inactivates the catalytic subunit of PDH, the enzyme that converts pyruvate to acetyl coenzyme A (AcCoA) for entry into the mitochondrial tricarboxylic acid (TCA) cycle [25]. Second, HIF-1 activates the gene encoding BNIP3, a member of the Bcl-2 family of mitochondrial proteins, which triggers selective mitochondrial autophagy [26]. Interference with the induction of either of these proteins in hypoxic cells results in increased ROS production and increased cell death. Overexpression of either PDK1 or BNIP3 rescues HIF-1α-null MEFs. By shunting pyruvate away from the mitochondria, PDK1 decreases flux through the ETC and thereby counteracts the reduced efficiency of electron transport under hypoxic conditions, which would otherwise increase ROS production. PDK1 functions cooperatively with the product of another HIF-1 target gene, LDHA [27], which converts pyruvate to lactate, thereby further reducing available substrate for the PDH reaction.

PDK1 effectively reduces flux through the TCA cycle and thereby reduces flux through the ETC in cells that primarily utilize glucose as a substrate for oxidative phosphorylation. However, PDK1 is predicted to have little effect on ROS generation in cells that utilize fatty acid oxidation as their source of AcCoA. Hence another strategy to reduce ROS generation under hypoxic conditions is selective mitochondrial autophagy [26]. MEFs reduce their mitochondrial mass and O2 consumption by >50% after only two days at 1% O2. BNIP3 competes with Beclin-1 for binding to Bcl-2, thereby freeing Beclin-1 to activate autophagy. Using short hairpin RNAs to knockdown expression of BNIP3, Beclin-1, or Atg5 (another component of the autophagy machinery) phenocopied HIF-1α-null cells by preventing hypoxia-induced reductions in mitochondrial mass and O2 consumption as a result of failure to induce autophagy [26]. HIF-1-regulated expression of BNIP3L also contributes to hypoxia-induced autophagy [28]. Remarkably, mice heterozygous for the HIF-1α KO allele have a significantly increased ratio of mitochondrial:nuclear DNA in their lungs (even though this is the organ that is exposed to the highest O2 concentrations), indicating that HIF-1 regulates mitochondrial mass under physiological conditions in vivo [26]. In contrast to the selective mitochondrial autophagy that is induced in response to hypoxia as described above, autophagy (of unspecified cellular components) induced by anoxia does not require HIF-1, BNIP3, or BNIP3L, but is instead regulated by AMP kinase [29].

The multiplicity of HIF-1-mediated mechanisms identified so far by which cells regulate mitochondrial metabolism in response to changes in cellular O2 concentration (Fig. 3) suggests that this is a critical adaptive response to hypoxia. The fundamental nature of this physiological response is underscored by the fact that yeast also switch COX4 subunits in an O2-dependent manner but do so by an entirely different molecular mechanism [33], since yeast do not have a HIF-1α homologue. Thus, it appears that by convergent evolution both unicellular and multicellular eukaryotes possess mechanisms by which they modulate mitochondrial metabolism to maintain redox homeostasis despite changes in O2 availability. Indeed, it is the balance between energy, oxygen, and redox homeostasis that represents the key to life with oxygen.

Regulation of mitochondrial metabolism by HIF-1  nihms232046f3

Regulation of mitochondrial metabolism by HIF-1 nihms232046f3

Regulation of mitochondrial metabolism by HIF-1α

Fig. 3 Regulation of mitochondrial metabolism by HIF-1α. Acute hypoxia leads to increased mitochondrial generation of reactive oxygen species (ROS). Decreased O2 and increased ROS levels lead to decreased HIF-1α hydroxylation (see Fig. 2) and increased HIF-1-dependent 


7.9.5 Regulation of cancer cell metabolism by hypoxia-inducible factor 1

Semenza GL1.
Semin Cancer Biol. 2009 Feb; 19(1):12-6.

The Warburg Effect: The Re-discovery of the Importance of Aerobic Glycolysis in Tumor Cells

The induction of hypoxia-inducible factor 1 (HIF-1) activity, either as a result of intratumoral hypoxia or loss-of-function mutations in the VHL gene, leads to a dramatic reprogramming of cancer cell metabolism involving increased glucose transport into the cell, increased conversion of glucose to pyruvate, and a concomitant decrease in mitochondrial metabolism and mitochondrial mass. Blocking these adaptive metabolic responses to hypoxia leads to cell death due to toxic levels of reactive oxygen species. Targeting HIF-1 or metabolic enzymes encoded by HIF-1 target genes may represent a novel therapeutic approach to cancer.

7.9.6 Coming up for air. HIF-1 and mitochondrial oxygen consumption

Simon MC1.
Cell Metab. 2006 Mar;3(3):150-1.

Hypoxic cells induce glycolytic enzymes; this HIF-1-mediated metabolic adaptation increases glucose flux to pyruvate and produces glycolytic ATP. Two papers in this issue of Cell Metabolism (Kim et al., 2006; Papandreou et al., 2006) demonstrate that HIF-1 also influences mitochondrial function, suppressing both the TCA cycle and respiration by inducing pyruvate dehydrogenase kinase 1 (PDK1). PDK1 regulation in hypoxic cells promotes cell survival.

Comment on

Oxygen deprivation (hypoxia) occurs in tissues when O2 supply via the cardiovascular system fails to meet the demand of O2-consuming cells. Hypoxia occurs naturally in physiological settings (e.g., embryonic development and exercising muscle), as well as in pathophysiological conditions (e.g., myocardial infarction, inflammation, and solid tumor formation). For over a century, it has been appreciated that O2-deprived cells exhibit increased conversion of glucose to lactate (the “Pasteur effect”). Activation of the Pasteur effect during hypoxia in mammalian cells is facilitated by HIF-1, which mediates the upregulation of glycolytic enzymes that support an increase in glycolytic ATP production as mitochondria become starved for O2, the substrate for oxidative phosphorylation (Seagroves et al., 2001). Thus, mitochondrial respiration passively decreases due to O2 depletion in hypoxic tissues. However, reports by Kim et al. (2006) and Papandreou et al. (2006) in this issue of Cell Metabolism demonstrate that this critical metabolic adaptation is more complex and includes an active suppression of mitochondrial pyruvate catabolism and O2consumption by HIF-1.

Mitochondrial oxidative phosphorylation is regulated by multiple mechanisms, including substrate availability. Major substrates include O2 (the terminal electron acceptor) and pyruvate (the primary carbon source). Pyruvate, as the end product of glycolysis, is converted to acetyl-CoA by the pyruvate dehydrogenase enzymatic complex and enters the tricarboxylic acid (TCA) cycle. Pyruvate conversion into acetyl-CoA is irreversible; this therefore represents an important regulatory point in cellular energy metabolism. Pyruvate dehydrogenase kinase (PDK) inhibits pyruvate dehydrogenase activity by phosphorylating its E1 subunit (Sugden and Holness, 2003). In the manuscripts by Kim et al. (2006) and Papandreou et al. (2006), the authors find that PDK1 is a HIF-1 target gene that actively regulates mitochondrial respiration by limiting pyruvate entry into the TCA cycle. By excluding pyruvate from mitochondrial metabolism, hypoxic cells accumulate pyruvate, which is then converted into lactate via lactate dehydrogenase (LDH), another HIF-1-regulated enzyme. Lactate in turn is released into the extracellular space, regenerating NAD+ for continued glycolysis by O2-starved cells (see Figure 1). This HIF-1-dependent block to mitochondrial O2 consumption promotes cell survival, especially when O2 deprivation is severe and prolonged.



Figure 1. Multiple hypoxia-induced cellular metabolic changes are regulated by HIF-1

By stimulating the expression of glucose transporters and glycolytic enzymes, HIF-1 promotes glycolysis to generate increased levels of pyruvate. In addition, HIF-1 promotes pyruvate reduction to lactate by activating lactate dehydrogenase (LDH). Pyruvate reduction to lactate regenerates NAD+, which permits continued glycolysis and ATP production by hypoxic cells. Furthermore, HIF-1 induces pyruvate dehydrogenase kinase 1 (PDK1), which inhibits pyruvate dehydrogenase and blocks conversion of pyruvate to acetyl CoA, resulting in decreased flux through the tricarboxylic acid (TCA) cycle. Decreased TCA cycle activity results in attenuation of oxidative phosphorylation and excessive mitochondrial reactive oxygen species (ROS) production. Because hypoxic cells already exhibit increased ROS, which have been shown to promote HIF-1 accumulation, the induction of PDK1 prevents the persistence of potentially harmful ROS levels.

Papandreou et al. demonstrate that hypoxic regulation of PDK has important implications for antitumor therapies. Recent interest has focused on cytotoxins that target hypoxic cells in tumor microenvironments, such as the drug tirapazamine (TPZ). Because intracellular O2 concentrations are decreased by mitochondrial O2 consumption, HIF-1 could protect tumor cells from TPZ-mediated cell death by maintaining intracellular O2 levels. Indeed, Papandreou et al. show that HIF-1-deficient cells grown at 2% O2 exhibit increased sensitivity to TPZ relative to wild-type cells, presumably due to higher rates of mitochondrial O2 consumption. HIF-1 inhibition in hypoxic tumor cells should have multiple therapeutic benefits, but the use of HIF-1 inhibitors in conjunction with other treatments has to be carefully evaluated for the most effective combination and sequence of drug delivery. One result of HIF-1 inhibition would be a relative decrease in intracellular O2 levels, making hypoxic cytotoxins such as TPZ more potent antitumor agents. Because PDK expression has been detected in multiple human tumor samples and appears to be induced by hypoxia (Koukourakis et al., 2005), small molecule inhibitors of HIF-1 combined with TPZ represent an attractive therapeutic approach for future clinical studies.

Hypoxic regulation of PDK1 has other important implications for cell survival during O2 depletion. Because the TCA cycle is coupled to electron transport, Kim et al. suggest that induction of the pyruvate dehydrogenase complex by PDK1 attenuates not only mitochondrial respiration but also the production of mitochondrial reactive oxygen species (ROS) in hypoxic cells. ROS are a byproduct of electron transfer to O2, and cells cultured at 1 to 5% O2 generate increased mitochondrial ROS relative to those cultured at 21% O2 (Chandel et al., 1998 and Guzy et al., 2005). In fact, hypoxia-induced mitochondrial ROS have also been shown to be necessary for the stabilization of HIF-1 in hypoxic cells (Brunelle et al., 2005Guzy et al., 2005 and Mansfield et al., 2005). However, the persistence of ROS could ultimately be lethal to tissues during chronic O2 deprivation, and PDK1 induction by HIF-1 should promote cell viability during long-term hypoxia. Kim et al. present evidence that HIF-1-deficient cells exhibit increased apoptosis after 72 hr of culture at 0.5% O2 compared to wild-type cells and that cell survival is rescued by enforced expression of exogenous PDK1. Furthermore, PDK1 reduces ROS production by the HIF-1 null cells. These findings support a novel prosurvival dimension of cellular hypoxic adaptation where PDK1 inhibits the TCA cycle, mitochondrial respiration, and chronic ROS production.

The HIF-1-mediated block to mitochondrial O2 consumption via PDK1 regulation also has implications for O2-sensing pathways by hypoxic cells. One school of thought suggests that perturbing mitochondrial O2consumption increases intracellular O2 concentrations and suppresses HIF-1 induction by promoting the activity of HIF prolyl hydroxylases, the O2-dependent enzymes that regulate HIF-1 stability (Hagen et al., 2003 and Doege et al., 2005). This model suggests that mitochondria function as “O2 sinks.” Although Papandreou et al. demonstrate that increased mitochondrial respiration due to PDK1 depletion results in decreased intracellular O2 levels (based on pimonidazole staining), these changes failed to reduce HIF-1 levels in hypoxic cells. Another model for hypoxic activation of HIF-1 describes a critical role for mitochondrial ROS in prolyl hydroxylase inhibition and HIF-1 stabilization in O2-starved cells (Brunelle et al., 2005Guzy et al., 2005 and Mansfield et al., 2005) (see Figure 1). The mitochondrial “O2 sink” hypothesis can account for some observations in the literature but fails to explain the inhibition of HIF-1 stabilization by ROS scavengers (Chandel et al., 1998Brunelle et al., 2005Guzy et al., 2005 and Sanjuán-Pla et al., 2005). While the relationship between HIF-1 stability, mitochondrial metabolism, ROS, and intracellular O2 redistribution will continue to be debated for some time, these most recent findings shed new light on findings by Louis Pasteur over a century ago.

Selected reading

Brunelle et al., 2005

J.K. Brunelle, E.L. Bell, N.M. Quesada, K. Vercauteren, V. Tiranti, M. Zeviani, R.C. Scarpulla, N.S. Chandel

Cell Metab., 1 (2005), pp. 409–414

Article  PDF (324 K) View Record in Scopus Citing articles (357)

Chandel et al., 1998

N.S. Chandel, E. Maltepe, E. Goldwasser, C.E. Mathieu, M.C. Simon, P.T. Schumacker

Proc. Natl. Acad. Sci. USA, 95 (1998), pp. 11715–11720

View Record in Scopus Full Text via CrossRef Citing articles (973)

Doege et al., 2005Doege, S. Heine, I. Jensen, W. Jelkmann, E. Metzen

Blood, 106 (2005), pp. 2311–2317

View Record in Scopus Full Text via CrossRef Citing articles (84)

Guzy et al., 2005

R.D. Guzy, B. Hoyos, E. Robin, H. Chen, L. Liu, K.D. Mansfield, M.C. Simon, U. Hammerling, P.T. Schumacker

Cell Metab., 1 (2005), pp. 401–408

Article  PDF (510 K) View Record in Scopus Citing articles (593)

Hagen et al., 2003

Hagen, C.T. Taylor, F. Lam, S. Moncada

Science, 302 (2003), pp. 1975–1978

View Record in Scopus Full Text via CrossRef Citing articles (450)

7.9.7 HIF-1 mediates adaptation to hypoxia by actively downregulating mitochondrial oxygen consumption

Papandreou I1Cairns RAFontana LLim ALDenko NC.
Cell Metab. 2006 Mar; 3(3):187-97.

The HIF-1 transcription factor drives hypoxic gene expression changes that are thought to be adaptive for cells exposed to a reduced-oxygen environment. For example, HIF-1 induces the expression of glycolytic genes. It is presumed that increased glycolysis is necessary to produce energy when low oxygen will not support oxidative phosphorylation at the mitochondria. However, we find that while HIF-1 stimulates glycolysis, it also actively represses mitochondrial function and oxygen consumption by inducing pyruvate dehydrogenase kinase 1 (PDK1). PDK1 phosphorylates and inhibits pyruvate dehydrogenase from using pyruvate to fuel the mitochondrial TCA cycle. This causes a drop in mitochondrial oxygen consumption and results in a relative increase in intracellular oxygen tension. We show by genetic means that HIF-1-dependent block to oxygen utilization results in increased oxygen availability, decreased cell death when total oxygen is limiting, and reduced cell death in response to the hypoxic cytotoxin tirapazamine.

Comment in

Tissue hypoxia results when supply of oxygen from the bloodstream does not meet demand from the cells in the tissue. Such a supply-demand mismatch can occur in physiologic conditions such as the exercising muscle, in the pathologic condition such as the ischemic heart, or in the tumor microenvironment (Hockel and Vaupel, 2001 and Semenza, 2004). In either the physiologic circumstance or pathologic conditions, there is a molecular response from the cell in which a program of gene expression changes is initiated by the hypoxia-inducible factor-1 (HIF-1) transcription factor. This program of gene expression changes is thought to help the cells adapt to the stressful environment. For example, HIF-1-dependent expression of erythropoietin and angiogenic compounds results in increased blood vessel formation for delivery of a richer supply of oxygenated blood to the hypoxic tissue. Additionally, HIF-1 induction of glycolytic enzymes allows for production of energy when the mitochondria are starved of oxygen as a substrate for oxidative phosphorylation. We now find that this metabolic adaptation is more complex, with HIF-1 not only regulating the supply of oxygen from the bloodstream, but also actively regulating the oxygen demand of the tissue by reducing the activity of the major cellular consumer of oxygen, the mitochondria.

Perhaps the best-studied example of chronic hypoxia is the hypoxia associated with the tumor microenvironment (Brown and Giaccia, 1998). The tumor suffers from poor oxygen supply through a chaotic jumble of blood vessels that are unable to adequately perfuse the tumor cells. The oxygen tension within the tumor is also a function of the demand within the tissue, with oxygen consumption influencing the extent of tumor hypoxia (Gulledge and Dewhirst, 1996 and Papandreou et al., 2005b). The net result is that a large fraction of the tumor cells are hypoxic. Oxygen tensions within the tumor range from near normal at the capillary wall, to near zero in the perinecrotic regions. This perfusion-limited hypoxia is a potent microenvironmental stress during tumor evolution (Graeber et al., 1996 and Hockel and Vaupel, 2001) and an important variable capable of predicting for poor patient outcome. (Brizel et al., 1996Cairns and Hill, 2004Hockel et al., 1996 and Nordsmark and Overgaard, 2004).

The HIF-1 transcription factor was first identified based on its ability to activate the erythropoetin gene in response to hypoxia (Wang and Semenza, 1993). Since then, it is has been shown to be activated by hypoxia in many cells and tissues, where it can induce hypoxia-responsive target genes such as VEGF and Glut1 (Airley et al., 2001 and Kimura et al., 2004). The connection between HIF-regulation and human cancer was directly linked when it was discovered that the VHL tumor suppressor gene was part of the molecular complex responsible for the oxic degradation of HIF-1α (Maxwell et al., 1999). In normoxia, a family of prolyl hydroxylase enzymes uses molecular oxygen as a substrate and modifies HIF-1α and HIF2α by hydroxylation of prolines 564 and 402 (Bruick and McKnight, 2001 and Epstein et al., 2001). VHL then recognizes the modified HIF-α proteins, acts as an E3-type of ubiquitin ligase, and along with elongins B and C is responsible for the polyubiquitination of HIF-αs and their proteosomal degradation (Bruick and McKnight, 2001Chan et al., 2002Ivan et al., 2001 and Jaakkola et al., 2001). Mutations in VHL lead to constitutive HIF-1 gene expression, and predispose humans to cancer. The ability to recognize modified HIF-αs is at least partly responsible for VHL activity as a tumor suppressor, as introduction of nondegradable HIF-2α is capable of overcoming the growth–inhibitory activity of wild-type (wt) VHL in renal cancer cells (Kondo et al., 2003).

Mitochondrial function can be regulated by PDK1 expression. Mitochondrial oxidative phosphorylation (OXPHOS) is regulated by several mechanisms, including substrate availability (Brown, 1992). The major substrates for OXPHOS are oxygen, which is the terminal electron acceptor, and pyruvate, which is the primary carbon source. Pyruvate is the end product of glycolysis and is converted to acetyl-CoA through the activity of the pyruvate dehydrogenase complex of enzymes. The acetyl-CoA then directly enters the TCA cycle at citrate synthase where it is combined with oxaloacetate to generate citrate. In metazoans, the conversion of pyruvate to acetyl-CoA is irreversible and therefore represents a critical regulatory point in cellular energy metabolism. Pyruvate dehydrogenase is regulated by three known mechanisms: it is inhibited by acetyl-CoA and NADH, it is stimulated by reduced energy in the cell, and it is inhibited by regulatory phosphorylation of its E1 subunit by pyruvate dehydrogenase kinase (PDK) (Holness and Sugden, 2003 and Sugden and Holness, 2003). There are four members of the PDK family in vertebrates, each with specific tissue distributions (Roche et al., 2001). PDK expression has been observed in human tumor biopsies (Koukourakis et al., 2005), and we have reported that PDK3 is hypoxia-inducible in some cell types (Denko et al., 2003). In this manuscript, we find that PDK1 is also a hypoxia-responsive protein that actively regulates the function of the mitochondria under hypoxic conditions by reducing pyruvate entry into the TCA cycle. By excluding pyruvate from mitochondrial consumption, PDK1 induction may increase the conversion of pyruvate to lactate, which is in turn shunted to the extracellular space, regenerating NAD for continued glycolysis.

Identification of HIF-dependent mitochondrial proteins through genomic and bioinformatics approaches

In order to help elucidate the role of HIF-1α in regulating metabolism, we undertook a genomic search for genes that were regulated by HIF-1 in tumor cells exposed to hypoxia in vitro. We used genetically matched human RCC4 cells that had lost VHL during tumorigenesis and displayed constitutive HIF-1 activity, and a cell line engineered to re-express VHL to establish hypoxia-dependent HIF activation. These cells were treated with 18 hr of stringent hypoxia (<0.01% oxygen), and microarray analysis performed. Using a strict 2.5-fold elevation as our cutoff, we identified 173 genes that were regulated by hypoxia and/or VHL status (Table S1 in the Supplemental Data available with this article online). We used the pattern of expression in these experiments to identify putative HIF-regulated genes—ones that were constitutively elevated in the parent RCC4s independent of hypoxia, downregulated in the RCC4VHL cells under normoxia, and elevated in response to hypoxia. Of the 173 hypoxia and VHL-regulated genes, 74 fit the putative HIF-1 target pattern. The open reading frames of these genes were run through a pair of bioinformatics engines in order to predict subcellular localization, and 10 proteins scored as mitochondrial on at least one engine. The genes, fold induction, and mitochondrial scores are listed in Table 1.

HIF-1 downregulates mitochondrial oxygen consumption

Having identified several putative HIF-1 responsive gene products that had the potential to regulate mitochondrial function, we then directly measured mitochondrial oxygen consumption in cells exposed to long-term hypoxia. While other groups have studied mitochondrial function under acute hypoxia (Chandel et al., 1997), this is one of the first descriptions of mitochondrial function after long-term hypoxia where there have been extensive hypoxia-induced gene expression changes. Figure 1A is an example of the primary oxygen trace from a Clark electrode showing a drop in oxygen concentration in cell suspensions of primary fibroblasts taken from normoxic and hypoxic cultures. The slope of the curve is a direct measure of the total cellular oxygen consumption rate. Exposure of either primary human or immortalized mouse fibroblasts to 24 hr of hypoxia resulted in a reduction of this rate by approximately 50% (Figures 1A and 1B). In these experiments, the oxygen consumption can be stimulated with the mitochondrial uncoupling agent CCCP (carbonyl cyanide 3-chloro phenylhydrazone) and was completely inhibited by 2 mM potassium cyanide. We determined that the change in total cellular oxygen consumption was due to changes in mitochondrial activity by the use of the cell-permeable poison of mitochondrial complex 3, Antimycin A. Figure 1C shows that the difference in the normoxic and hypoxic oxygen consumption in murine fibroblasts is entirely due to the Antimycin-sensitive mitochondrial consumption. The kinetics with which mitochondrial function slows in hypoxic tumor cells also suggests that it is due to gene expression changes because it takes over 6 hr to achieve maximal reduction, and the reversal of this repression requires at least another 6 hr of reoxygenation (Figure 1D). These effects are not likely due to proliferation or toxicity of the treatments as these conditions are not growth inhibitory or toxic to the cells (Papandreou et al., 2005a).

Since we had predicted from the gene expression data that the mitochondrial oxygen consumption changes were due to HIF-1-mediated expression changes, we tested several genetically matched systems to determine what role HIF-1 played in the process (Figure 2). We first tested the cell lines that had been used for microarray analysis and found that the parental RCC4 cells had reduced mitochondrial oxygen consumption when compared to the VHL-reintroduced cells. Oxygen consumption in the parental cells was insensitive to hypoxia, while it was reduced by hypoxia in the wild-type VHL-transfected cell lines. Interestingly, stable introduction of a tumor-derived mutant VHL (Y98H) that cannot degrade HIF was also unable to restore oxygen consumption. These results indicate that increased expression of HIF-1 is sufficient to reduce oxygen consumption (Figure 2A). We also investigated whether HIF-1 induction was required for the observed reduction in oxygen consumption in hypoxia using two genetically matched systems. We measured normoxic and hypoxic oxygen consumption in murine fibroblasts derived from wild-type or HIF-1α null embryos (Figure 2B) and from human RKO tumor cells and RKO cells constitutively expressing ShRNAs directed against the HIF-1α gene (Figures 2C and 4C). Neither of the HIF-deficient cell systems was able to reduce oxygen consumption in response to hypoxia. These data from the HIF-overexpressing RCC cells and the HIF-deficient cells indicate that HIF-1 is both necessary and sufficient for reducing mitochondrial oxygen consumption in hypoxia.

HIF-dependent mitochondrial changes are functional, not structural

Because addition of CCCP could increase oxygen consumption even in the hypoxia-treated cells, we hypothesized that the hypoxic inhibition was a regulated activity, not a structural change in the mitochondria in response to hypoxic stress. We confirmed this interpretation by examining several additional mitochondrial characteristics in hypoxic cells such as mitochondrial morphology, quantity, and membrane potential. We examined morphology by visual inspection of both the transiently transfected mitochondrially localized DsRed protein and the endogenous mitochondrial protein cytochrome C. Both markers were indistinguishable in the parental RCC4 and the RCC4VHL cells (Figure 3A). Likewise, we measured the mitochondrial membrane potential with the functional dye rhodamine 123 and found that it was identical in the matched RCC4 cells and the matched HIF wt and knockout (KO) cells when cultured in normoxia or hypoxia (Figure 3B). Finally, we determined that the quantity of mitochondria per cell was not altered in response to HIF or hypoxia by showing that the amount of the mitochondrial marker protein HSP60 was identical in the RCC4 and HIF cell lines (Figure 3C)

PDK1 is a HIF-1 inducible target protein

After examination of the list of putative HIF-regulated mitochondrial target genes, we hypothesized that PDK1 could mediate the functional changes that we observed in hypoxia. We therefore investigated PDK1 protein expression in response to HIF and hypoxia in the genetically matched cell systems. Figure 4A shows that in the RCC4 cells PDK1 and the HIF-target gene BNip3 (Greijer et al., 2005 and Papandreou et al., 2005a) were both induced by hypoxia in a VHL-dependent manner, with the expression of PDK1 inversely matching the oxygen consumption measured in Figure 1 above. Likewise, the HIF wt MEFs show oxygen-dependent induction of PDK1 and BNip3, while the HIF KO MEFs did not show any expression of either of these proteins under any oxygen conditions (Figure 4B). Finally, the parental RKO cells were able to induce PDK1 and the HIF target gene BNip3L in response to hypoxia, while the HIF-depleted ShRNA RKO cells could not induce either protein (Figure 4C). Therefore, in all three cell types, the HIF-1-dependent regulation of oxygen consumption seen in Figure 2, corresponds to the HIF-1-dependent induction of PDK1 seen in Figure 4.

In order to determine if PDK1 was a direct HIF-1 target gene, we analyzed the genomic sequence flanking the 5′ end of the gene for possible HIF-1 binding sites based on the consensus core HRE element (A/G)CGTG (Caro, 2001). Several such sites exist within the first 400 bases upstream, so we generated reporter constructs by fusing the genomic sequence from −400 to +30 of the start site of transcription to the firefly luciferase gene. In transfection experiments, the chimeric construct showed significant induction by either cotransfection with a constitutively active HIF proline mutant (P402A/P564G) (Chan et al., 2002) or exposure of the transfected cells to 0.5% oxygen (Figure 4D). Most noteworthy, when the reporter gene was transfected into the HIF-1α null cells, it did not show induction when the cells were cultured in hypoxia, but it did show induction when cotransfected with expression HIF-1α plasmid. We then generated deletions down to the first 36 bases upstream of transcription and found that even this short sequence was responsive to HIF-1 (Figure 4D). Analysis of this small fragment showed only one consensus HRE site located in an inverted orientation in the 5′ untranslated region. We synthesized and cloned a mutant promoter fragment in which the core element ACGTG was replaced with AAAAG, and this construct lost over 90% of its hypoxic induction. These experiments suggest that it is this HRE within the proximal 5′ UTR that HIF-1 uses to transactivate the endogenous PDK1 gene in response to hypoxia.

PDK1 is responsible for the HIF-dependent mitochondrial oxygen consumption changes

In order to directly test if PDK1 was the HIF-1 target gene responsible for the hypoxic reduction in mitochondrial oxygen consumption, we generated RKO cell lines with either knockdown or overexpression of PDK1 and measured the oxygen consumption in these derivatives. The PDK1 ShRNA stable knockdown line was generated as a pool of clones cotransfected with pSUPER ShPDK1 and pTK-hygro resistance gene. After selection for growth in hygromycin, the cells were tested by Western blot for the level of PDK1 protein expression. We found that normoxic PDK1 is reduced by 75%, however, there was measurable expression of PDK1 in these cells in response to hypoxia (Figure 5A). When we measured the corresponding oxygen consumption in these cells, we found a change commensurate with the level of PDK1. The knockdown cells show elevated baseline oxygen consumption, and partial reduction in this activity in response to hypoxia. Therefore, reduction of PDK1 expression by genetic means increased mitochondrial oxygen consumption in both normoxic and hypoxic conditions. Interestingly, these cells still induced HIF-1α (Figure 5A) and HIF-1 target genes such as BNip3L in response to hypoxia (data not shown), suggesting that altered PDK1 levels do not alter HIF-1α function.



PDK1 expression directly regulates cellular oxygen consumption rate

Figure 5. PDK1 expression directly regulates cellular oxygen consumption rate

  1. A)Western blot of RKO cell and ShRNAPDK1RKO cell lysates after exposure to 24 hr of normoxia or 0.5% O2. Blots were probed for HIF 1α, PDK1, and tubulin as a loading control.
  2. B)Oxygen consumption rate in RKO and ShRNAPDK1RKO cells after exposure to 24 hr of normoxia or 0.5% O2.
  3. C)Western blot of RKOiresGUS cell and RKOiresPDK1 cell lysates after exposure to 24 hr of normoxia or 0.5% O2. Blots were probed for HIF 1α, PDK1, and tubulin as a loading control.
  4. D)Oxygen consumption rate in RKOiresGUS and RKOiresPDK1 cells after exposure to 24 hr of normoxia or 0.5% O2.
  5. E)Model describing the interconnected effects of HIF-1 target gene activation on hypoxic cell metabolism. Reduced oxygen conditions causes HIF-1 to coordinately induce the enzymes shown in boxes. HIF-1 activation results in increased glucose transporter expression to increase intracellular glucose flux, induction of glycolytic enzymes increases the conversion of glucose to pyruvate generating energy and NADH, induction of PDK1 decreases mitochondrial utilization of pyruvate and oxygen, and induction of LDH increases the removal of excess pyruvate as lactate and also regenerates NAD+ for increased glycolysis.

For all graphs, the error bars represent the standard error of the mean.

We also determined if overexpression of PDK1 could lead to reduced mitochondrial oxygen consumption. A separate culture of RKO cells was transfected with a PDK1-IRES-puro expression plasmid and selected for resistance to puromycin. The pool of puromycin resistant cells was tested for PDK1 expression by Western blot. These cells showed a modest increase in PDK1 expression under control conditions when compared to the cells transfected with GUS-IRES-puro, with an additional increase in PDK1 protein in response to hypoxia (Figure 5C). The corresponding oxygen consumption measurements showed that the mitochondria is very sensitive to changes in the levels of PDK1, as even this slight increase was able to significantly reduce oxygen consumption in the normoxic PDK1-puro cultures. Further increase in PDK1 levels with hypoxia further reduced oxygen consumption in both cultures (Figure 5D). The model describing the relationship between hypoxia, HIF-1, PDK1, and intermediate metabolism is described inFigure 5E.

Altering oxygen consumption alters intracellular oxygen tension and sensitivity to hypoxia-dependent cell killing

The intracellular concentration of oxygen is a net result of the rate at which oxygen diffuses into the cell and the rate at which it is consumed. We hypothesized that the rate at which oxygen was consumed within the cell would significantly affect its steady-state intracellular concentrations. We tested this hypothesis in vitro using the hypoxic marker drug pimonidazole (Bennewith and Durand, 2004). We plated high density cultures of HIF wild-type and HIF knockout cells and placed these cultures in normoxic, 2% oxygen, and anoxic incubators for overnight treatment. The overnight treatment gives the cells time to adapt to the hypoxic conditions and establish altered oxygen consumption profiles. Pimonidozole was then added for the last 4 hr of the growth of the culture. Pimonidazole binding was detected after fixation of the cells using an FITC labeled anti-pimonidazole antibody and it was quantitated by flow cytometry. The quantity of the bound drug is a direct indication of the oxygen concentration within the cell (Bennewith and Durand, 2004). The histograms in Figure 6A show that the HIF-1 knockout and wild-type cells show similar staining in the cells grown in 0% oxygen. However, the cells treated with 2% oxygen show the consequence of the genetic removal of HIF-1. The HIF-proficient cells showed relatively less pimonidazole binding at 2% when compared to the 0% culture, while the HIF-deficient cells showed identical binding between the cells at 2% and those at 0%. We interpret these results to mean that the HIF-deficient cells have greater oxygen consumption, and this has lowered the intracellular oxygenation from the ambient 2% to close to zero intracellularly. The HIF-proficient cells reduced their oxygen consumption rate so that the rate of diffusion into the cell is greater than the rate of consumption.

Figure 6. HIF-dependent decrease in oxygen consumption raises intracellular oxygen concentration, protects when oxygen is limiting, and decreases sensitivity to tirapazamine in vitro

  1. A)Pimonidazole was used to determine the intracellular oxygen concentration of cells in culture. HIF wt and HIF KO MEFs were grown at high density and exposed to 2% O2or anoxia for 24 hr in glass dishes. For the last 4 hr of treatment, cells were exposed to 60 μg/ml pimonidazole. Pimonidazole binding was quantitated by flow cytometry after binding of an FITC conjugated anti-pimo mAb. Results are representative of two independent experiments.
  2. B)HIF1α reduces oxygen consumption and protects cells when total oxygen is limited. HIF wt and HIF KO cells were plated at high density and sealed in aluminum jigs at <0.02% oxygen. At the indicated times, cells were harvested, and dead cells were quantitated by trypan blue exclusion. Note both cell lines are equally sensitive to anoxia-induced apoptosis, so the death of the HIF null cells indicates that the increased oxygen consumption removed any residual oxygen in the jig and resulted in anoxia-induced death.
  3. C)PDK1 is responsible for HIF-1’s adaptive response when oxygen is limiting. A similar jig experiment was performed to measure survival in the parental RKO, the RKO ShRNAHIF1α, and the RKOShPDK1 cells. Cell death by trypan blue uptake was measured 48 hr after the jigs were sealed.
  4. D)HIF status alters sensitivity to TPZ in vitro. HIF wt and HIF KO MEFs were grown at high density in glass dishes and exposed to 21%, 2%, and <0.01% O2conditions for 18 hr in the presence of varying concentrations of Tirapazamine. After exposure, cells were harvested and replated under normoxia to determine clonogenic viability. Survival is calculated relative to the plating efficiency of cells exposed to 0 μM TPZ for each oxygen concentration.
  5. E)Cell density alters sensitivity to TPZ. HIF wt and HIF KO MEFs were grown at varying cell densities in glass dishes and exposed to 2% O2in the presence of 10 μM TPZ for 18 hr. After the exposure, survival was determined as described in (C).

For all graphs, the error bars represent the standard error of the mean.

HIF-induced PDK1 can reduce the total amount of oxygen consumed per cell. The reduction in the amount of oxygen consumed could be significant if there is a finite amount of oxygen available, as would be the case in the hours following a blood vessel occlusion. The tissue that is fed by the vessel would benefit from being economical with the oxygen that is present. We experimentally modeled such an event using aluminum jigs that could be sealed with defined amounts of cells and oxygen present (Siim et al., 1996). We placed 10 × 106 wild-type or HIF null cells in the sealed jig at 0.02% oxygen, waited for the cells to consume the remaining oxygen, and measured cell viability. We have previously shown that these two cell types are resistant to mild hypoxia and equally sensitive to anoxia-induced apoptosis (Papandreou et al., 2005a). Therefore, any death in this experiment would be the result of the cells consuming the small amount of remaining oxygen and dying in response to anoxia. We found that in sealed jigs, the wild-type cells are more able to adapt to the limited oxygen supply by reducing consumption. The HIF null cells continued to consume oxygen, reached anoxic levels, and started to lose viability within 36 hr (Figure 6B). This is a secondary adaptive effect of HIF1. We confirmed that PDK1 was responsible for this difference by performing a similar experiment using the parental RKO cells, the RKOShRNAHIF1α and the RKOShRNAPDK1 cells. We found similar results in which both the cells with HIF1α knockdown and PDK1 knockdown were sensitive to the long-term effects of being sealed in a jig with a defined amount of oxygen (Figure 6c). Note that the RKOShPDK1 cells are even more sensitive than the RKOShHIF1α cells, presumably because they have higher basal oxygen consumption rates (Figure 5B).

Because HIF-1 can help cells adapt to hypoxia and maintain some intracellular oxygen level, it may also protect tumor cells from killing by the hypoxic cytotoxin tirapazamine (TPZ). TPZ toxicity is very oxygen dependent, especially at oxygen levels between 1%–4% (Koch, 1993). We therefore tested the relative sensitivity of the HIF wt and HIF KO cells to TPZ killing in high density cultures (Figure 6D). We exposed the cells to the indicated concentrations of drug and oxygen concentrations overnight. The cells were then harvested and replated to determine reproductive viability by colony formation. Both cell types were equally resistant to TPZ at 21% oxygen, while both cell types are equally sensitive to TPZ in anoxic conditions where intracellular oxygen levels are equivalent (Figure 6A). The identical sensitivity of both cell types in anoxia indicates that both cell types are equally competent in repairing the TPZ-induced DNA damage that is presumed to be responsible for its toxicity. However, in 2% oxygen cultures, the HIF null cells displayed a significantly greater sensitivity to the drug than the wild-type cells. This suggests that the increased oxygen consumption rate in the HIF-deficient cells is sufficient to lower the intracellular oxygen concentration relative to that in the HIF-proficient cells. The lower oxygen level is significant enough to dramatically sensitize these cells to killing by TPZ.

If the increased sensitivity to TPZ in the HIF ko cells is determined by intracellular oxygen consumption differences, then this effect should also be cell-density dependent. We showed that this is indeed the case in Figure 6E where oxygen and TPZ concentrations were held constant, and increased cell density lead to increased TPZ toxicity. The effect was much more pronounced in the HIF KO cells, although the HIF wt cells showed some increased toxicity in the highest density cultures, consistent with the fact they were still consuming some oxygen, even with HIF present (Figure 1). The in vitro TPZ survival data is therefore consistent with our hypothesis that control of oxygen consumption can regulate intracellular oxygen concentration, and suggests that increased oxygen consumption could sensitize cells to hypoxia-dependent therapy.


The findings presented here show that HIF-1 is actively responsible for regulating energy production in hypoxic cells by an additional, previously unrecognized mechanism. It has been shown that HIF-1 induces the enzymes responsible for glycolysis when it was presumed that low oxygen did not support efficient oxidative phosphorylation (Iyer et al., 1998 and Seagroves et al., 2001). The use of glucose to generate ATP is capable of satisfying the energy requirements of a cell if glucose is in excess (Papandreou et al., 2005a). We now find that at the same time that glycolysis is increasing, mitochondrial respiration is decreasing. However, the decreased respiration is not because there is not enough oxygen present to act as a substrate for oxidative phosphorylation, but because the flow of pyruvate into the TCA cycle has been reduced by the activity of pyruvate dehydrogenase kinase. Other reports have suggested that oxygen utilization is shifted in cells exposed to hypoxia, but these reports have focused on other regulators such as nitric oxide synthase (Hagen et al., 2003). NO can reduce oxygen consumption through direct inhibition of cytochrome oxidase, but this effect seems to be more significant at physiologic oxygen concentrations, not at severe levels seen in the tumor (Palacios-Callender et al., 2004).

7.9.8 HIF-1. upstream and downstream of cancer metabolism

Semenza GL1.
Curr Opin Genet Dev. 2010 Feb; 20(1):51-6

Hypoxia-inducible factor 1 (HIF-1) plays a key role in the reprogramming of cancer metabolism by activating transcription of genes encoding glucose transporters and glycolytic enzymes, which take up glucose and convert it to lactate; pyruvate dehydrogenase kinase 1, which shunts pyruvate away from the mitochondria; and BNIP3, which triggers selective mitochondrial autophagy. The shift from oxidative to glycolytic metabolism allows maintenance of redox homeostasis and cell survival under conditions of prolonged hypoxia. Many metabolic abnormalities in cancer cells increase HIF-1 activity. As a result, a feed-forward mechanism can be activated that drives HIF-1 activation and may promote tumor progression. Hypoxia-inducible factor 1 (HIF-1) plays a key role in the reprogramming of cancer metabolism by activating transcription of genes encoding glucose transporters and glycolytic enzymes, which take up glucose and convert it to lactate; pyruvate dehydrogenase kinase 1, which shunts pyruvate away from the mitochondria; and BNIP3, which triggers selective mitochondrial autophagy. The shift from oxidative to glycolytic metabolism allows maintenance of redox homeostasis and cell survival under conditions of prolonged hypoxia. Many metabolic abnormalities in cancer cells increase HIF-1 activity. As a result, a feed-forward mechanism can be activated that drives HIF-1 activation and may promote tumor progression.

Metastatic cancer is characterized by reprogramming of cellular metabolism leading to increased uptake of glucose for use as both an anabolic and catabolic substrate. Increased glucose uptake is such a reliable feature that it is utilized clinically to detect metastases by positron emission tomography using 18F-fluorodeoxyglucose (FDG-PET) with a sensitivity of ~90% [1]. As with all aspects of cancer biology, the details of metabolic reprogramming differ widely among individual tumors. However, the role of specific signaling pathways and transcription factors in this process is now understood in considerable detail. This review will focus on the involvement of hypoxia-inducible factor 1 (HIF-1) in both mediating metabolic reprogramming and responding to metabolic alterations. The placement of HIF-1 both upstream and downstream of cancer metabolism results in a feed-forward mechanism that may play a major role in the development of the invasive, metastatic, and lethal cancer phenotype.

O2 concentrations are significantly reduced in many human cancers compared to the surrounding normal tissue. The median PO2 in breast cancers is ~10 mm Hg, as compared to ~65 mm Hg in normal breast tissue [2]. Reduced O2 availability induces HIF-1, which regulates the transcription of hundreds of genes [3*,4*] that encode proteins involved in every aspect of cancer biology, including: cell immortalization and stem cell maintenance; genetic instability; glucose and energy metabolism; vascularization; autocrine growth factor signaling; invasion and metastasis; immune evasion; and resistance to chemotherapy and radiation therapy [5].

HIF-1 is a transcription factor that consists of an O2-regulated HIF-1α and a constitutively expressed HIF-1β subunit [6]. In well-oxygenated cells, HIF-1α is hydroxylated on proline residue 402 (Pro-402) and/or Pro-564 by prolyl hydroxylase domain protein 2 (PHD2), which uses O2 and α-ketoglutarate as substrates in a reaction that generates CO2 and succinate as byproducts [7]. Prolyl-hydroxylated HIF-1α is bound by the von Hippel-Lindau tumor suppressor protein (VHL), which recruits an E3-ubiquitin ligase that targets HIF-1α for proteasomal degradation (Figure 1A). Asparagine 803 in the transactivation domain is hydroxylated in well-oxygenated cells by factor inhibiting HIF-1 (FIH-1), which blocks the binding of the coactivators p300 and CBP [7]. Under hypoxic conditions, the prolyl and asparaginyl hydroxylation reactions are inhibited by substrate (O2) deprivation and/or the mitochondrial generation of reactive oxygen species (ROS), which may oxidize Fe(II) present in the catalytic center of the hydroxylases [8].

HIF-1 and metabolism  nihms156580f1

HIF-1 and metabolism nihms156580f1

HIF-1 and metabolism

Figure 1 HIF-1 and metabolism. (A) Regulation of HIF-1α protein synthesis and stability and HIF-1-dependent metabolic reprogramming. The rate of translation of HIF-1α mRNA into protein in cancer cells is dependent upon the activity of the mammalian 

The finding that acute changes in PO2 increase mitochondrial ROS production suggests that cellular respiration is optimized at physiological PO2 to limit ROS generation and that any deviation in PO2 — up or down — results in increased ROS generation. If hypoxia persists, induction of HIF-1 leads to adaptive mechanisms to reduce ROS and re-establish homeostasis, as described below. Prolyl and asparaginyl hydroxylation provide a molecular mechanism by which changes in cellular oxygenation can be transduced to the nucleus as changes in HIF-1 activity. This review will focus on recent advances in our understanding of the role of HIF-1 in controlling glucose and energy metabolism, but it should be appreciated that any increase in HIF-1 activity that leads to changes in cell metabolism will also affect many other critical aspects of cancer biology [5] that will not be addressed here.

HIF-1 target genes involved in glucose and energy metabolism

HIF-1 activates the transcription of SLC2A1 and SLC2A3, which encode the glucose transporters GLUT1 and GLUT3, respectively, as well as HK1 and HK2, which encode hexokinase, the first enzyme of the Embden-Meyerhoff (glycolytic) pathway [9]. Once taken up by GLUT and phosphorylated by HK, FDG cannot be metabolized further; thus, FDG-PET signal is determined by FDG delivery to tissue (i.e. perfusion) and GLUT/HK expression/activity. Unlike FDG, glucose is further metabolized to pyruvate by the action of the glycolytic enzymes, which are all encoded by HIF-1 target genes (Figure 1A). Glycolytic intermediates are also utilized for nucleotide and lipid synthesis [10]. Lactate dehydrogenase A (LDHA), which converts pyruvate to lactate, and monocarboxylate transporter 4 (MCT4), which transports lactate out of the cell (Figure 1B), are also regulated by HIF-1 [9,11]. Remarkably, lactate produced by hypoxic cancer cells can be taken up by non-hypoxic cells and used as a respiratory substrate [12**].

Pyruvate represents a critical metabolic control point, as it can be converted to acetyl coenzyme A (AcCoA) by pyruvate dehydrogenase (PDH) for entry into the tricarboxylic acid (TCA) cycle or it can be converted to lactate by LDHA (Figure 1B). Pyruvate dehydrogenase kinase (PDK), which phosphorylates and inactivates the catalytic domain of PDH, is encoded by four genes and PDK1 is activated by HIF-1 [13,14]. (Further studies are required to determine whether PDK2PDK3, or PDK4 is regulated by HIF-1.) As a result of PDK1 activation, pyruvate is actively shunted away from the mitochondria, which reduces flux through the TCA cycle, thereby reducing delivery of NADH and FADH2 to the electron transport chain. This is a critical adaptive response to hypoxia, because in HIF-1α–null mouse embryo fibroblasts (MEFs), PDK1 expression is not induced by hypoxia and the cells die due to excess ROS production, which can be ameliorated by forced expression of PDK1 [13]. MYC, which is activated in ~40% of human cancers, cooperates with HIF-1 to activate transcription of PDK1, thereby amplifying the hypoxic response [15]. Pharmacological inhibition of HIF-1 or PDK1 activity increases O2 consumption by cancer cells and increases the efficacy of a hypoxia-specific cytotoxin [16].

Hypoxia also induces mitochondrial autophagy in many human cancer cell lines through HIF-1-dependent expression of BNIP3 and a related BH3 domain protein, BNIP3L [19**]. Autocrine signaling through the platelet-derived growth factor receptor in cancer cells increases HIF-1 activity and thereby increases autophagy and cell survival under hypoxic conditions [21]. Autophagy may also occur in a HIF-1-independent manner in response to other physiological stimuli that are associated with hypoxic conditions, such as a decrease in the cellular ATP:AMP ratio, which activates AMP kinase signaling [22].

In clear cell renal carcinoma, VHL loss of function (LoF) results in constitutive HIF-1 activation, which is associated with impaired mitochondrial biogenesis that results from HIF-1-dependent expression of MXI1, which blocks MYC-dependent expression of PGC-1β, a coactivator that is required for mitochondrial biogenesis [23]. Inhibition of wild type MYC activity in renal cell carcinoma contrasts with the synergistic effect of HIF-1 and oncogenic MYC in activating PDK1 transcription [24].

Genetic and metabolic activators of HIF-1

Hypoxia plays a critical role in cancer progression [2,5] but not all cancer cells are hypoxic and a growing number of O2-independent mechanisms have been identified by which HIF-1 is induced [5]. Several mechanisms that are particularly relevant to cancer metabolism are described below.

Activation of mTOR

Alterations in mitochondrial metabolism

NAD+ levels

It is of interest that the NAD+-dependent deacetylase sirtuin 1 (SIRT1) was found to bind to, deacetylate, and increase transcriptional activation by HIF-2α but not HIF-1α [42**]. Another NAD+-dependent enzyme is poly(ADP-ribose) polymerase 1 (PARP1), which was recently shown to bind to HIF-1α and promote transactivation through a mechanism that required the enzymatic activity of PARP1 [43]. Thus, transactivation mediated by both HIF-1α and HIF-2α can be modulated according to NAD+ levels.

Nitric oxide

Increased expression of nitric oxide (NO) synthase isoforms and increased levels of NO have been shown to increase HIF-1α protein stability in human oral squamous cell carcinoma [44]. In prostate cancer, nuclear co-localization of endothelial NO synthase, estrogen receptor β, HIF-1α, and HIF-2α was associated with aggressive disease and the proteins were found to form chromatin complexes on the promoter of TERT gene encoding telomerase [45**]. The NOS2 gene encoding inducible NO synthase is HIF-1 regulated [5], suggesting another possible feed-forward mechanism.

7.9.9 In Vivo HIF-Mediated Reductive Carboxylation

Gameiro PA1Yang JMetelo AMPérez-Carro R, et al.
Cell Metab. 2013 Mar 5; 17(3):372-85.

Hypoxic and VHL-deficient cells use glutamine to generate citrate and lipids through reductive carboxylation (RC) of α-ketoglutarate. To gain insights into the role of HIF and the molecular mechanisms underlying RC, we took advantage of a panel of disease-associated VHL mutants and showed that HIF expression is necessary and sufficient for the induction of RC in human renal cell carcinoma (RCC) cells. HIF expression drastically reduced intracellular citrate levels. Feeding VHL-deficient RCC cells with acetate or citrate or knocking down PDK-1 and ACLY restored citrate levels and suppressed RC. These data suggest that HIF-induced low intracellular citrate levels promote the reductive flux by mass action to maintain lipogenesis. Using [1–13C] glutamine, we demonstrated in vivo RC activity in VHL-deficient tumors growing as xenografts in mice. Lastly, HIF rendered VHL-deficient cells sensitive to glutamine deprivation in vitro, and systemic administration of glutaminase inhibitors suppressed the growth of RCC cells as mice xenografts.

Cancer cells undergo fundamental changes in their metabolism to support rapid growth, adapt to limited nutrient resources, and compete for these supplies with surrounding normal cells. One of the metabolic hallmarks of cancer is the activation of glycolysis and lactate production even in the presence of adequate oxygen. This is termed the Warburg effect, and efforts in cancer biology have revealed some of the molecular mechanisms responsible for this phenotype (Cairns et al., 2011). More recently, 13C isotopic studies have elucidated the complementary switch of glutamine metabolism that supports efficient carbon utilization for anabolism and growth (DeBerardinis and Cheng, 2010). Acetyl-CoA is a central biosynthetic precursor for lipid synthesis, being generated from glucose-derived citrate in well-oxygenated cells (Hatzivassiliou et al., 2005). Warburg-like cells, and those exposed to hypoxia, divert glucose to lactate, raising the question of how the tricarboxylic acid (TCA) cycle is supplied with acetyl-CoA to support lipogenesis. We and others demonstrated, using 13C isotopic tracers, that cells under hypoxic conditions or defective mitochondria primarily utilize glutamine to generate citrate and lipids through reductive carboxylation (RC) of α-ketoglutarate by isocitrate dehydrogenase 1 (IDH1) or 2 (IDH2) (Filipp et al., 2012Metallo et al., 2012;Mullen et al., 2012Wise et al., 2011).

The transcription factors hypoxia inducible factors 1α and 2α (HIF-1α, HIF-2α) have been established as master regulators of the hypoxic program and tumor phenotype (Gordan and Simon, 2007Semenza, 2010). In addition to tumor-associated hypoxia, HIF can be directly activated by cancer-associated mutations. The von Hippel-Lindau (VHL) tumor suppressor is inactivated in the majority of sporadic clear-cell renal carcinomas (RCC), with VHL-deficient RCC cells exhibiting constitutive HIF-1α and/or HIF-2α activity irrespective of oxygen availability (Kim and Kaelin, 2003). Previously, we showed that VHL-deficient cells also relied on RC for lipid synthesis even under normoxia. Moreover, metabolic profiling of two isogenic clones that differ in pVHL expression (WT8 and PRC3) suggested that reintroduction of wild-type VHL can restore glucose utilization for lipogenesis (Metallo et al., 2012). The VHL tumor suppressor protein (pVHL) has been reported to have several functions other than the well-studied targeting of HIF. Specifically, it has been reported that pVHL regulates the large subunit of RNA polymerase (Pol) II (Mikhaylova et al., 2008), p53 (Roe et al., 2006), and the Wnt signaling regulator Jade-1. VHL has also been implicated in regulation of NF-κB signaling, tubulin polymerization, cilia biogenesis, and proper assembly of extracellular fibronectin (Chitalia et al., 2008Kim and Kaelin, 2003Ohh et al., 1998Thoma et al., 2007Yang et al., 2007). Hypoxia inactivates the α-ketoglutarate-dependent HIF prolyl hydroxylases, leading to stabilization of HIF. In addition to this well-established function, oxygen tension regulates a larger family of α-ketoglutarate-dependent cellular oxygenases, leading to posttranslational modification of several substrates, among which are chromatin modifiers (Melvin and Rocha, 2012). It is therefore conceivable that the effect of hypoxia on RC that was reported previously may be mediated by signaling mechanisms independent of the disruption of the pVHL-HIF interaction. Here we (1) demonstrate that HIF is necessary and sufficient for RC, (2) provide insights into the molecular mechanisms that link HIF to RC, (3) detected RC activity in vivo in human VHL-deficient RCC cells growing as tumors in nude mice, (4) provide evidence that the reductive phenotype ofVHL-deficient cells renders them sensitive to glutamine restriction in vitro, and (5) show that inhibition of glutaminase suppresses growth of VHL-deficient cells in nude mice. These observations lay the ground for metabolism-based therapeutic strategies for targeting HIF-driven tumors (such as RCC) and possibly the hypoxic compartment of solid tumors in general.

Functional Interaction between pVHL and HIF Is Necessary to Inhibit RC

Figure 1  HIF Inactivation Is Necessary for Downregulation of Reductive Carboxylation by pVHL

We observed a concurrent regulation in glucose metabolism in the different VHL mutants. Reintroduction of wild-type or type 2C pVHL mutant, which can meditate HIF-α destruction, stimulated glucose oxidation via pyruvate dehydrogenase (PDH), as determined by the degree of 13C-labeled TCA cycle metabolites (M2 enrichment) (Figures 1D and 1E). In contrast, reintroduction of an HIF nonbinding Type 2B pVHL mutant failed to stimulate glucose oxidation, resembling the phenotype observed in VHL-deficient cells (Figures 1D and 1E). Additional evidence for the overall glucose utilization was obtained from the enrichment of M3 isotopomers using [U13-C6]glucose (Figure S1A), which shows a lower contribution of glucose-derived carbons to the TCA cycle in VHL-deficient RCC cells (via pyruvate carboxylase and/or continued TCA cycling).

To test the effect of HIF activation on the overall glutamine incorporation in the TCA cycle, we labeled an isogenic pair of VHL-deficient and VHL-reconstituted UMRC2 cells with [U-13C5]glutamine, which generates M4 fumarate, M4 malate, M4 aspartate, and M4 citrate isotopomers through glutamine oxidation. As seen in Figure S1BVHL-deficient/VHL-positive UMRC2 cells exhibit similar enrichment of M4 fumarate, M4 malate, and M4 asparate (but not citrate) showing that VHL-deficient cells upregulate reductive carboxylation without compromising oxidative metabolism from glutamine. …  Labeled carbon derived from [5-13C1]glutamine can be incorporated into fatty acids exclusively through RC, and the labeled carbon cannot be transferred to palmitate through the oxidative TCA cycle (Figure 1B, red carbons). Tracer incorporation from [5-13C1]glutamine occurs in the one carbon (C1) of acetyl-CoA, which results in labeling of palmitate at M1, M2, M3, M4, M5, M6, M7, and M8 mass isotopomers. In contrast, lipogenic acetyl-CoA molecules originating from [U-13C6]glucose are fully labeled, and the labeled palmitate is represented by M2, M4, M6, M8, M10, M12, M14, and M16 mass isotopomers.

Figure 2 HIF Inactivation Is Necessary for Downregulation of Reductive Lipogenesis by pVHL

To determine the specific contribution from glucose oxidation or glutamine reduction to lipogenic acetyl-CoA, we performed isotopomer spectral analysis (ISA) of palmitate labeling patterns. ISA indicates that wild-type pVHL or pVHL L188V mutant-reconstituted UMRC2 cells relied mainly on glucose oxidation to produce lipogenic acetyl-CoA, while UMRC2 cells reconstituted with a pVHL mutant defective in HIF inactivation (Y112N or Y98N) primarily employed RC. Upon disruption of the pVHL-HIF interaction, glutamine becomes the preferred substrate for lipogenesis, supplying 70%–80% of the lipogenic acetyl-CoA (Figure 2C). This is not a cell-line-specific phenomenon, but it applies to VHL-deficient human RCC cells in general; the same changes are observed in 786-O cells reconstituted with wild-type pVHL or mutant pVHL or infected with vector only as control (Figure S2).

HIF Is Sufficient to Induce RC (reductive carboxylation) from Glutamine in RCC Cells

As shown in Figure 3C, reintroduction of wild-type VHLinto 786-O cells suppressed RC, whereas the expression of the constitutively active HIF-2α mutant was sufficient to stimulate this reaction, restoring the M1 enrichment of TCA cycle metabolites observed in VHL-deficient 786-O cells. Expression of HIF-2α P-A also led to a concomitant decrease in glucose oxidation, corroborating the metabolic alterations observed in glutamine metabolism (Figures 3D and 3E).

Figure 3 Expression of HIF-2α Is Sufficient to Induce Reductive Carboxylation and Lipogenesis from Glutamine in RCC Cells

Expression of HIF-2α P-A in 786-O cells phenocopied the loss-of-VHL with regards to glutamine reduction for lipogenesis (Figure 3G), suggesting that HIF-2α can induce the glutamine-to-lipid pathway in RCC cells per se. Although reintroduction of wild-type VHL restored glucose oxidation in UMRC2 and UMRC3 cells (Figures S3B–S3I), HIF-2α P-A expression did not measurably affect the contribution of each substrate to the TCA cycle or lipid synthesis in these RCC cells (data not shown). UMRC2 and UMRC3 cells endogenously express both HIF-1α and HIF-2α, whereas 786-O cells exclusively express HIF-2α. There is compelling evidence suggesting, at least in RCC cells, that HIF-α isoforms have overlapping—but also distinct—functions and their roles in regulating bioenergetic processes remain an area of active investigation. Overall, HIF-1α has an antiproliferative effect, and its expression in vitro leads to rapid death of RCC cells while HIF-2α promotes tumor growth (Keith et al., 2011Raval et al., 2005).

Metabolic Flux Analysis Shows Net Reversion of the IDH Flux upon HIF Activation

To determine absolute fluxes in RCC cells, we employed 13C metabolic flux analysis (MFA) as previously described (Metallo et al., 2012). Herein, we performed MFA using a combined model of [U-13C6]glucose and [1-13C1]glutamine tracer data sets from the 786-O derived isogenic clones PRC3 (VHL−/ −)/WT8 (VHL+) cells, which show a robust metabolic regulation by reintroduction of pVHL. To this end, we first determined specific glucose/glutamine consumption and lactate/glutamate secretion rates. As expected, PRC3 exhibited increased glucose consumption and lactate production when compared to WT8 counterparts (Figure 4A). While PRC3 exhibited both higher glutamine consumption and glutamate production rates than WT8 (Figure 4A), the net carbon influx was higher in PRC3 cells (Figure 4B). Importantly, the fitted data show that the flux of citrate to α-ketoglutarate was negative in PRC3 cells (Figure 4C). This indicates that the net (forward plus reverse) flux of isocitrate dehydrogenase and aconitase (IDH + ACO) is toward citrate production. The exchange flux was also higher in PRC3 than WT8 cells, whereas the PDH flux was lower in PRC3 cells. In agreement with the tracer data, these MFA results strongly suggest that the reverse IDH + ACO fluxes surpass the forward flux in VHL-deficient cells. The estimated ATP citrate lyase (ACLY) flux was also lower in PRC3 than in WT8 cells. Furthermore, the malate dehydrogenase (MDH) flux was negative, reflecting a net conversion of oxaloacetate into malate in VHL-deficient cells (Figure 4C). This indicates an increased flux through the reductive pathway downstream of IDH, ACO, and ACLY. Additionally, some TCA cycle flux estimates downstream of α-ketoglutarate were not significantly different between PRC and WT8 (Table S1). This shows that VHL-deficient cells maintain glutamine oxidation while upregulating reductive carboxylation (Figure S1B). This finding is in agreement with the higher glutamine uptake observed in VHL-deficient cells. Table S1 shows the metabolic network and complete MFA results. …

Addition of citrate in the medium, in contrast to acetate, led to an increase in the citrate-to-α-ketoglutarate ratio (Figure 5L) and absolute citrate levels (Figure S4H) not only in VHL-deficient but alsoVHL-reconstituted cells. The ability of exogenous citrate, but not acetate, to also affect RC in VHL-reconstituted cells may be explained by compartmentalization differences or by allosteric inhibition of citrate synthase (Lehninger, 2005); that is, the ability of acetate to raise the intracellular levels of citrate may be limited in (VHL-reconstituted) cells that exhibit high endogenous levels of citrate. Whatever the mechanism, the results imply that increasing the pools of intracellular citrate has a direct biochemical effect in cells with regards to their reliance on RC. Finally, we assayed the transcript and protein levels of enzymes involved in the reductive utilization of glutamine and did not observe significant differences between VHL-deficient andVHL-reconstituted UMRC2 cells (Figures S4I and S4J), suggesting that HIF does not promote RC by direct transactivation of these enzymes. The IDH1/IDH2 equilibrium is defined as follows:


Figure 5 Regulation of HIF-Mediated Reductive Carboxylation by Citrate Levels

We sought to investigate whether HIF could affect the driving force of the IDH reaction by also enhancing NADPH production. We did not observe a significant alteration of the NADP+/NADPH ratio between VHL-deficient and VHL-positive cells in the cell lysate (Figure S4I). Yet, we determined the ratio of the free dinucleotides using the measured ratios of suitable oxidized (α-ketoglutarate) and reduced (isocitrate/citrate) metabolites that are linked to the NADP-dependent IDH enzymes. The determined ratios (Figure S4J) are in close agreement with the values initially reported by the Krebs lab (Veech et al., 1969) and showed that HIF-expressing UMRC2 cells exhibit a higher NADP+/NADPH ratio. Collectively, these data strongly suggest that HIF-regulated citrate levels modulate the reductive flux to maintain adequate lipogenesis.

Reductive Carboxylation from Glutamine Is Detectable In Vivo

Figure 6 Evidence for Reductive Carboxylation Activity In Vivo

Loss of VHL Renders RCC Cells Sensitive to Glutamine Deprivation

We hypothesized that VHL deficiency results in cell addiction to glutamine for proliferation. We treated the isogenic clones PRC3 (VHL-deficient cells) and WT8 (VHL-reconstituted cells) with the glutaminase inhibitor 968 (Wang et al., 2010a). VHL-deficient PRC3 cells were more sensitive to treatment with 968, compared to the VHL-reconstituted WT8 cells (Figure 7A). To confirm that this is not only a cell-line-specific phenomenon, we also cultured UMRC2 cells in the presence of 968 or diluent control and showed selective sensitivity of VHL-deficient cells (Figure 7B).

Figure 7 VHL-Deficient Cells and Tumors Are Sensitive to Glutamine Deprivation

(A–E) Cell proliferation is normalized to the corresponding cell type grown in 1 mM glutamine-containing medium. Effect of treatment with glutaminase (GLS) inhibitor 968 in PRC3/WT8 (A) and UMRC2 cells (B). Rescue of GLS inhibition with dimethyl alpha-ketoglutarate (DM-Akg; 4 mM) or acetate (4 mM) in PRC3/WT8 clonal cells (C) and polyclonal 786-O cells (D). Effect of GLS inhibitor BPTES in UMRC2 cells (E). Student’s t test compares VHL-reconstituted cells to control cells in (A), (B), and (E) and DM-Akg or acetate-rescued cells to correspondent control cells treated with 968 only in (C) and (D) (asterisk in parenthesis indicates comparison between VHL-reconstituted to control cells). Error bars represent SEM.

(F) GLS inhibitor BPTES suppresses growth of human UMRC3 RCC cells as xenografts in nu/nu mice. When the tumors reached 100mm3, injections with BPTES or vehicle control were carried out daily for 14 days (n = 12). BPTES treatment decreases tumor size and mass (see insert). Student’s t test compares control to BPTES-treated mice (F). Error bars represent SEM.

(G) Diagram showing the regulation of reductive carboxylation by HIF.

In summary, our findings show that HIF is necessary and sufficient to promote RC from glutamine. By inhibiting glucose oxidation in the TCA cycle and reducing citrate levels, HIF shifts the IDH reaction toward RC to support citrate production and lipogenesis (Figure 7G). The reductive flux is active in vivo, fuels tumor growth, and can potentially be targeted pharmacologically. Understanding the significance of reductive glutamine metabolism in tumors may lead to metabolism-based therapeutic strategies.

Along with others, we reported that hypoxia and loss of VHL engage cells in reductive carboxylation (RC) from glutamine to support citrate and lipid synthesis (Filipp et al., 2012Metallo et al., 2012Wise et al., 2011). Wise et al. (2011) suggested that inactivation of HIF in VHL-deficient cells leads to reduction of RC. These observations raise the hypothesis that HIF, which is induced by hypoxia and is constitutively active inVHL-deficient cells, mediates RC. In our current work, we provide mechanistic insights that link HIF to RC. First, we demonstrate that polyclonal reconstitution of VHL in several human VHL-deficient RCC cell lines inhibits RC and restores glucose oxidation. Second, the VHL mutational analysis demonstrates that the ability of pVHL to mitigate reductive lipogenesis is mediated by HIF and is not the outcome of previously reported, HIF-independent pVHL function(s). Third, to prove our hypothesis we showed that constitutive expression of a VHL-independent HIF mutant is sufficient to phenocopy the reductive phenotype observed in VHL-deficient cells. In addition, we showed that RC is not a mere in vitro phenomenon, but it can be detected in vivo in human tumors growing as mouse xenografts. Lastly, treatment of VHL-deficient human xenografts with glutaminase inhibitors led to suppression of their growth as tumors.

7.9.10 Evaluation of HIF-1 inhibitors as anticancer agents

Semenza GL1.
Drug Discov Today. 2007 Oct; 12(19-20):853-9

Hypoxia-inducible factor 1 (HIF-1) regulates the transcription of many genes involved in key aspects of cancer biology, including immortalization, maintenance of stem cell pools, cellular dedifferentiation, genetic instability, vascularization, metabolic reprogramming, autocrine growth factor signaling, invasion/metastasis, and treatment failure. In animal models, HIF-1 overexpression is associated with increased tumor growth, vascularization, and metastasis, whereas HIF-1 loss-of-function has the opposite effect, thus validating HIF-1 as a target. In further support of this conclusion, immunohistochemical detection of HIF-1α overexpression in biopsy sections is a prognostic factor in many cancers. A growing number of novel anticancer agents have been shown to inhibit HIF-1 through a variety of molecular mechanisms. Determining which combination of drugs to administer to any given patient remains a major obstacle to improving cancer treatment outcomes.

Aurelian Udristioiu


Aurelian Udristioiu

Lab Director at Emergency County Hospital Targu Jiu

Mechanisms that control T cell metabolic reprogramming are now coming to light, and many of the same oncogenes importance in cancer metabolism are also crucial to drive T cell metabolic transformations, most notably Myc, hypoxia inducible factor (HIF)1a, estrogen-related receptor (ERR) a, and the mTOR pathway.
The proto-oncogenic transcription factor, Myc, is known to promote transcription of genes for the cell cycle, as well as aerobic glycolysis and glutamine metabolism. Recently, Myc has been shown to play an essential role in inducing the expression of glycolytic and glutamine metabolism genes in the initial hours of T cell activation. In a similar fashion, the transcription factor (HIF)1a can up-regulate glycolytic genes to allow cancer cells to survive under hypoxic conditions

Read Full Post »


Writer and Curator: Larry H. Bernstein, MD, FCAP 

7.8  Sirtuins

7.8.1 Function and regulation of the mitochondrial Sirtuin isoform Sirt5 in Mammalia

7.8.2 Substrates and Regulation Mechanisms for the Human Mitochondrial Sirtuins- Sirt3 and Sirt5

7.8.3 The mTORC1 Pathway Stimulates Glutamine Metabolism and Cell Proliferation by Repressing SIRT4

7.8.4  Rab1A and small GTPases Activate mTORC1

7.8.5 PI3K.Akt signaling in osteosarcoma

7.8.6 The mTORC1-S6K1 Pathway Regulates Glutamine Metabolism through the eIF4B-Dependent Control of c-Myc Translation

7.8.7 Localization of mouse mitochondrial SIRT proteins

7.8.8 SIRT4 Has Tumor-Suppressive Activity and Regulates the Cellular Metabolic Response to DNA Damage by Inhibiting Mitochondrial Glutamine Metabolism

7.8.9 Mitochondrial sirtuins and metabolic homeostasis

7.8.10 Mitochondrial sirtuins

7.8.11 Sirtuin regulation of mitochondria: energy production, apoptosis, and signaling


7.8.1 Function and regulation of the mitochondrial Sirtuin isoform Sirt5 in Mammalia

Gertz M1Steegborn C.
Biochim Biophys Acta. 2010 Aug; 1804(8):1658-65

Sirtuins are a family of protein deacetylases that catalyze the nicotinamide adenine dinucleotide (NAD(+))-dependent removal of acetyl groups from modified lysine side chains in various proteins. Sirtuins act as metabolic sensors and influence metabolic adaptation but also many other processes such as stress response mechanisms, gene expression, and organismal aging. Mammals have seven Sirtuin isoforms, three of them – Sirt3, Sirt4, and Sirt5 – located to mitochondria, our centers of energy metabolism and apoptosis initiation. In this review, we shortly introduce the mammalian Sirtuin family, with a focus on the mitochondrial isoforms. We then discuss in detail the current knowledge on the mitochondrial isoform Sirt5. Its physiological role in metabolic regulation has recently been confirmed, whereas an additional function in apoptosis regulation remains speculative. We will discuss the biochemical properties of Sirt5 and how they might contribute to its physiological function. Furthermore, we discuss the potential use of Sirt5 as a drug target, structural features of Sirt5 and of an Sirt5/inhibitor complex as well as their differences to other Sirtuins and the current status of modulating Sirt5 activity with pharmacological compounds.

removal of acetyl groups from modified lysine side chain

removal of acetyl groups from modified lysine side chain
removal of acetyl groups from modified lysine side chain

sirtuin structure

sirtuin structure
sirtuin structure

7.8.2 Substrates and Regulation Mechanisms for the Human Mitochondrial Sirtuins- Sirt3 and Sirt5

Schlicker C1Gertz MPapatheodorou PKachholz BBecker CFSteegborn C
J Mol Biol. 2008 Oct 10; 382(3):790-801

The enzymes of the Sirtuin family of nicotinamide-adenine-dinucleotide-dependent protein deacetylases are emerging key players in nuclear and cytosolic signaling, but also in mitochondrial regulation and aging. Mammalian mitochondria contain three Sirtuins, Sirt3, Sirt4, and Sirt5. Only one substrate is known for Sirt3 as well as for Sirt4, and up to now, no target for Sirt5 has been reported. Here, we describe the identification of novel substrates for the human mitochondrial Sirtuin isoforms Sirt3 and Sirt5. We show that Sirt3 can deacetylate and thereby activate a central metabolic regulator in the mitochondrial matrix, glutamate dehydrogenase. Furthermore, Sirt3 deacetylates and activates isocitrate dehydrogenase 2, an enzyme that promotes regeneration of antioxidants and catalyzes a key regulation point of the citric acid cycle. Sirt3 thus can regulate flux and anapleurosis of this central metabolic cycle. We further find that the N- and C-terminal regions of Sirt3 regulate its activity against glutamate dehydrogenase and a peptide substrate, indicating roles for these regions in substrate recognition and Sirtuin regulation. Sirt5, in contrast to Sirt3, deacetylates none of the mitochondrial matrix proteins tested. Instead, it can deacetylate cytochrome c, a protein of the mitochondrial intermembrane space with a central function in oxidative metabolism, as well as apoptosis initiation. Using a mitochondrial import assay, we find that Sirt5 can indeed be translocated into the mitochondrial intermembrane space, but also into the matrix, indicating that localization might contribute to Sirt5 regulation and substrate selection.

Mitochondria are central organelles in cellular energy metabolism, but also in processes such as apoptosis, cellular senescence, and lifespan regulation.1 and 2 Failures in mitochondrial function and regulation contribute to aging-related diseases, such as atherosclerosis3 and Parkinson’s disease,4 likely by increasing cellular levels of reactive oxygen species and the damage they cause.1 Emerging players in metabolic regulation and cellular signaling are members of the Sirtuin family of homologs of “silent information regulator 2” (Sir2), a yeast protein deacetylase.5 and 6 Sir2 was found to be involved in aging processes and lifespan determination in yeast,7 and 8 and its homologs were subsequently identified as lifespan regulators in various higher organisms.89 and 10 Sirtuins form class III of the protein deacetylase superfamily and hydrolyze one nicotinamide adenine dinucleotide (NAD +) as cosubstrate for each lysine residue they deacetylate.11 and 12 The coupling of deacetylation to NAD + was proposed to link changes in cellular energy levels to deacetylation activity,13 and 14 which would indicate Sirtuins as metabolic sensors. Other known regulation mechanisms for Sirtuin activity are the modulation of the expression levels of their genes6 and the autoinhibitory effect of an N-terminal region on the yeast Sirtuin “homologous to SIR2 protein 2” (Hst2).15

The seven mammalian Sirtuin proteins (Sirt1–Sirt7) have various substrate proteins that mediate functions in genetic, cellular, and mitochondrial regulation.5 and 6 The best-studied mammalian Sir2 homolog, Sirt1, was shown to regulate, among others, transcription factor p53, nuclear factor-kappa B, and peroxisome proliferator-activated receptor gamma coactivator-1-alpha.6 Three human Sirtuin proteins are known to be located in the mitochondria, Sirt3, Sirt4, and Sirt5,161718 and 19 although Sirt3 was reported to change its localization to nuclear when coexpressed with Sirt5.20 The recent identification of the first substrates for mitochondrial Sirtuins—acetyl coenzyme A synthetase 221 and 22 and glutamate dehydrogenase (GDH)16—as targets of Sirtuins 3 and 4, respectively, revealed that these Sirtuins control a regulatory network that has implications for energy metabolism and the mechanisms of caloric restriction (CR) and lifespan determination.23 Sirt3 regulates adaptive thermogenesis and decreases mitochondrial membrane potential and reactive oxygen species production, while increasing cellular respiration.24 Furthermore, Sirt3 is down-regulated in several genetically obese mice,24 and variability in the human SIRT3 gene has been linked to survivorship in the elderly. 25 In contrast to the deacetylases Sirt3 and Sirt5, Sirt4 appears to be an ADP ribosyltransferase. 16 Through this activity, Sirt4 inhibits GDH and thereby down-regulates insulin secretion in response to amino acids. 16 For Sirt5, however, there is no report yet on its physiological function or any physiological substrate. It is dominantly expressed in lymphoblasts and heart muscle cells,17 and 26 and its gene contains multiple repetitive elements that might make it a hotspot for chromosomal breaks. 26 Interestingly, the Sirt5 gene has been located to a chromosomal region known for abnormalities associated with malignant diseases. 26

A proteomics study found 277 acetylation sites in 133 mitochondrial proteins;27 many of them should be substrates for the mitochondrial Sirtuins mediating their various functions, but up to now, only one physiological substrate could be identified for Sirt3,21 and 22 and none could be identified for Sirt5. Our understanding of substrate selection by Sirtuins is incomplete, and knowledge of specific Sirtuin targets would be essential for a better understanding of Sirtuin-mediated processes and Sirtuin-targeted therapy. A first study on several Sirtuins showed varying preferences among acetylated peptides.28 Structural and thermodynamic analysis of peptides bound to the Sirtuin Sir2Tm from Thermatoga maritima indicated that positions − 1 and + 2 relative to the acetylation site play a significant role in substrate binding. 29 However, these studies were conducted with nonphysiological Sirtuin/substrate pairs, and other studies indicated little sequence specificity; instead, the yeast Sirtuin Hst2 was described to display contextual and conformational specificity: Hst2 deacetylated acetyl lysine only in the context of a protein, and it preferentially deacetylated within flexible protein regions. 30 Finally, statistical analysis of a proteomics study on acetylated proteins identified preferences at various positions such as + 1, − 2, and − 3, and deacetylation sites appeared to occur preferentially in helical regions. 27 Thus, our present knowledge of Sirtuin substrates and of factors determining Sirtuin specificity is incomplete and insufficient for sequence-based identification of physiological substrates.

Here, we describe the identification of novel targets for the mitochondrial deacetylases Sirt3 and Sirt5. We show that Sirt3 can deacetylate and thereby activate the enzymes GDH and isocitrate dehydrogenase (ICDH) 2—two key metabolic regulators in the mitochondrial matrix. We find that the N- and C-terminal regions of Sirt3 influence its activity against GDH and a peptide substrate, indicating roles in regulation and substrate recognition for these regions. Furthermore, we find that Sirt5 can deacetylate cytochrome c, a protein of the mitochondrial intermembrane space (IMS) with a central function in oxidative metabolism and apoptosis.

The upstream sequence contributes to the target specificity of Sirt3 and Sirt5

Sirtuins have been reported to have little sequence specificity,30 but other studies indicated a sequence preference dominated by positions − 1 and + 2.29 We tested the importance of the amino acid pattern preceding the acetylation site for recognition by the mitochondrial Sirtuins Sirt3 and Sirt5 through a fluorescence assay. First, the fluorogenic and commercially available modified p53-derived tetrapeptide QPK-acetylK, originally developed for Sirt2 assays but also efficiently used by Sirt3, was tested. Even 60 μg of Sirt5 did not lead to any deacetylation signal, whereas 0.35 μg of Sirt3 efficiently deacetylated the peptide (Fig. 1a). We then tested Sirt3 and Sirt5 on a second modified p53-derived tetrapeptide, RHK-acetylK. Sirt3 (0.5 μg) showed a slightly increased activity against this substrate as compared to QPK-acetylK (Fig. 1b); more importantly, 0.5 μg of Sirt5 showed significant activity against this peptide. These results show that the mitochondrial Sirtuins Sirt3 and, especially, Sirt5 indeed recognize the local target sequence, and target positions further upstream of − 1 seem to be involved in substrate recognition. For identification of novel substrates for the mitochondrial Sirtuins and further characterization of their target recognition mechanisms, we then turned to testing full-length proteins, as the downstream sequence and the larger protein context of the deacetylation site might also contribute to substrate selection.

Sirtuin substrate specificity

Sirtuin substrate specificity

Fig. 1. Testing the substrate specificity of Sirt3 and Sirt5 with peptides. (a) Sirt3, but not Sirt5, deacetylates the fluorogenic peptide QPK-acetylK. (b) Sirt3 efficiently deacetylates the fluorogenic peptide RHK-acetylK, and Sirt5 also significantly deacetylates this substrate.

Sirt3 deacetylates and activates GDH

In order to identify novel physiological substrates of the mitochondrial Sirtuins, we used proteins isolated in their partly acetylated form from natural sources (i.e., from mammalian mitochondria). These proteins, carrying physiological acetylations, were tested as Sirt3 and Sirt5 substrates in vitro in an ELISA system using an antibody specific for acetylated lysine. In a recent proteomics study, 27 GDH, a central regulator of mitochondrial metabolism, was identified to be acetylated in a feeding-dependent manner. With our ELISA, we found that Sirt3 and Sirt5 can both deacetylate pure GDH isolated from mitochondria, but with very different efficiencies ( Fig. 2a). Sirt3 significantly deacetylated GDH, but even large amounts of Sirt5 decreased the acetylation level of this substrate only slightly. We next tested the effect of GDH deacetylation on its activity. Deacetylation of GDH through incubation with Sirt3 and NAD + before its examination in a GDH activity assay increased its activity by 10%, and a stronger stimulation of GDH activity was seen when larger amounts of Sirt3 were used for deacetylation ( Fig. 2b). GDH is colocalized with Sirt3 in the mitochondrial matrix 1618 and 19 and, thus, likely could be a physiological substrate of this Sirtuin. Indeed, GDH from a Sirt3 knockout mouse was recently shown to be hyperacetylated compared to protein from wild-type mice. 31 Thus, Sirt3 deacetylates GDH in vivo, and our results show that this direct deacetylation of GDH by Sirt3 leads to GDH activation.

sirtuin structure

sirtuin structure

Fig. 2. Sirt3 can deacetylate and thereby activate GDH. (a) Deacetylation of GDH tested in ELISA. Sirt3 efficiently deacetylates GDH, whereas Sirt5 has only a small effect on the acetylation state. (b) GDH activity is increased after deacetylation of the enzyme by Sirt3. The increase in GDH activity depends on the amount of Sirt3 activity used for deacetylation.

Sirt3 can deacetylate and thereby activate ICDH2

In the proteomics study by Kim et al., the mitochondrial citric acid cycle enzymes fumarase and ICDH2 (a key regulator of this metabolic cycle) were found to be acetylated in a feeding-dependent manner. 27 In our ELISA system, we found that Sirt3 efficiently deacetylated the ICDH2 substrate isolated from mitochondria ( Fig. 3a). Western blot analysis (data not shown) and mass spectrometry confirmed that, indeed, the ICDH2 fraction of the partially purified protein was deacetylated by Sirt3. In contrast, even large amounts of Sirt5 did not significantly decrease the acetylation level of this substrate ( Fig. 3a). As expected, deacetylation of ICDH2 by Sirt3 was dependent on NAD +. Fumarase, in contrast, could not be deacetylated as efficiently as ICDH2 through treatment with either Sirt3 or Sirt5 ( Fig. 3b). The low absolute values over background for the ELISA with fumarase, however, might indicate low acetylation levels of the natively purified protein, and a stronger effect might be attainable when testing fumarase with a higher acetylation level.

Fig. 3. Sirt3 deacetylates ICDH2, but not fumarase. (a) Deacetylation of ICDH2 by Sirt3 and Sirt5 tested in ELISA. Sirt3, but not Sirt5, deacetylates ICDH2 in a NAD +-dependent manner. (b) Fumarase acetylation determined through ELISA cannot be significantly decreased by incubation with recombinant Sirt3 or Sirt5. (c) ICDH2 activity measured in a spectrophotometric assay based on the formation of NADPH. ICDH2 activity (continuous line) is increased after deacetylation of the enzyme by Sirt3 (dashed line). (d) The stimulatory effect of deacetylation on ICDH2 activity depends on the amount of deacetylase activity added during pretreatment. (e) ICDH2 with and without Sirt3 treatment analyzed by mass spectrometry after proteolytic digest. The decrease in the signal at 962.3 Da and the increase in signal at 903.5 Da indicate deacetylation at either K211 or K212.

In order to analyze the potential physiological function of ICDH2 deacetylation, we tested the effect of Sirt3-mediated ICDH2 deacetylation on its activity. Incubation of ICDH2 with Sirt3 and NAD + prior to its analysis in an ICDH activity assay increased its activity (Fig. 3c). The stimulation of ICDH2 activity was further increased when larger amounts of Sirt3 were used for deacetylation (Fig. 3d), and no significant increase in ICDH2 activity was observed when the Sirtuin inhibitor dihydrocoumarin was present during incubation with Sirt3 (data not shown). Sirt3 and ICDH2 are colocalized in the mitochondrial matrix,1619 and 32 and we therefore assume that ICDH2 is likely a physiological substrate for Sirt3, which activates ICDH2 by deacetylation.

Sirt3 can deacetylate KK motifs in substrate proteins

In order to identify the site of ICDH2 deacetylation upon treatment with Sirt3, we analyzed ICDH2 by mass spectrometry. For analyzing pure ICDH2, we excised its band from an SDS gel before mass spectrometry analysis. In the proteomics study by Kim et al., two acetylation sites were reported for ICDH2: K75 and K241 (numbering of the partial sequence of the unprocessed precursor; SwissProt entry P33198). 27 After digest of ICDH2, we could not detect peptides comprising K75 and, therefore, could not determine its acetylation status, and we only observed the deacetylated form of K241. We identified an additional acetylation site, however, by detecting signals at m/z = 903.5 and m/z = 962.3 for the peptide QYAIQKK (residues 206–212) carrying one and two acetyl groups, respectively ( Fig. 3e; calculated m/z = 903.5 and 962.5). Sirt3 treatment decreased the signal for the double-acetylated form and increased the signal for the single-acetylated form as compared to internal peptides [e.g., m/z = 890.5 (calculated m/z = 890.5) andm/z = 1041.4 (calculated m/z = 1041.5)]. These data indicate that Sirt3 deacetylates either position K211 or K212 of this KK motif located at a surface-exposed end of a helix that flanks the active site of ICDH2. 33Deacetylation of a KK motif by Sirt3 is consistent with the efficient use of the tested peptide substrates (see above) that both carry KK motifs.

Fig. 4. Increased activity of N- and C-terminally truncated Sirt3. (a) Specific activity against a peptide substrate of the longest Sirt3 form after proteolytic processing that covers residues 102–399. N-terminal truncation increases the specific activity dramatically, and an additional C-terminal truncation activates the catalytic core further. (b) Homology model of Sirt3 based on the crystal structure of Sirt2. The part comprising the catalytic core is shown in red. The NAD + and peptide ligands were manually placed into their binding sides based on the crystal structure of their complex with a bacterial Sir2 homolog from T. maritima. Parts removed in N- and C-terminal truncation constructs are shown in cyan and blue, respectively. (c) Level of acetylation of GDH tested in ELISA. The shortest Sirt3 form Sirt3(114–380) deacetylates more efficiently than Sirt3(114–399) and Sirt3(102–399), which show activities comparable to each other.

Sirt5 can deacetylate cytochrome c

Sirt5 can deacetylate cytochrome c

Sirt5 can deacetylate cytochrome c

The Sirt5 protein that we used for our study comprises residues 34–302, corresponding to the fully active catalytic core determined for Sirt3 (see above). This protein is indeed active against a peptide substrate, but it showed no significant activity against the acetylated mitochondrial matrix proteins tested so far: GDH, ICDH2, and fumarase. We thus picked cytochrome c, a central protein in energy metabolism and apoptosis localized in the mitochondrial IMS, from the list of acetylated mitochondrial proteins 27 for testing as deacetylation substrate. Sirt5 showed deacetylation activity against pure cytochrome c in our ELISA system, whereas Sirt3 had almost no activity against this substrate ( Fig. 5a). Even the more active shortened form of Sirt3(114–380) showed no considerable activity against this substrate.

Fig. 5.  Sirt5 can deacetylate cytochrome c. (a) Deacetylation of cytochrome c tested in ELISA. Sirt5 uses cytochrome c as substrate for deacetylation, whereas Sirt3 treatment leaves the acetylation level of cytochrome c unchanged. (b) Model of the action of the mammalian Sirtuins Sirt3, Sirt4, and Sirt5 in mitochondria. CAC: citric acid cycle. (c) Digest of Sirt5 synthesized in vitro with PK. The protein is fully degraded at proteinase concentrations of 25 μg/ml and above. (d) Import of Sirt5 into isolated yeast mitochondria. Sirt5 reaches an inner mitochondrial compartment in the presence and in the absence of the mitochondrial membrane potential (ΔΨ), whereas Sirt3, as a control for a matrix-targeted protein, is not imported into uncoupled mitochondria. (e) Intramitochondrial localization of Sirt5. Part of the imported Sirt5 is sensitive to PK after swelling (SW) and thus localized in the IMS, but another part of the protein remains protease-resistant and therefore appears to be localized to the matrix. Atp3, a protein localized at the matrix site of the mitochondrial inner membrane, and an IMS-located domain of translocase of inner membrane 23 detected by Western blot analysis served as controls for matrix transport and swelling, respectively. aTim23: anti-Tim23. (f) Scheme of the domain organizations of Sirt3 and Sirt5. Numbers in brackets are residue numbers for boundaries of protein parts. NLS: nuclear localization sequence; MLS: mitochondrial localization sequence; R1, regulatory region 1; R2: regulatory region 2.

Cytochrome c might be a physiological substrate of Sirt5 if this Sirtuin is localized to the mitochondrial IMS (Fig. 5b). A recent study on overexpressed tagged mouse Sirt5 in COS7 cells 20 indeed indicated that Sirt5, at least from mouse, is localized in the IMS. In order to test whether human Sirt5 can be localized to the IMS, we performed import experiments with human Sirt3 and Sirt5 using isolated yeast mitochondria as a model system. 3 Sirt3 and Sirt5 proteins were incubated with mitochondria, followed by PK treatment for degradation of nonimported protein ( Fig. 5d). In a parallel reaction, mitochondria were uncoupled prior to the import reaction by addition of valinomycin (− ΔΨ). Sirt3, a protein known to be located in the mitochondrial matrix, 19 was only efficiently imported in the presence of a membrane potential. Dependence on the mitochondrial potential is a hallmark of matrix import, 38 and the results thus show that Sirt3 is imported into the correct compartment in our experimental system. Sirt5, in contrast, reaches an inner-mitochondrial compartment both in the presence and in the absence of the membrane potential, suggesting that Sirt5 may accumulate in the IMS.

In order to further test the localization of Sirt5, we removed the outer mitochondrial membrane after the import reaction by osmotic swelling, followed by PK digest of then accessible proteins (Fig. 5e). Rupture of the outer membrane was confirmed by monitoring the accessibility of an IMS-exposed domain of endogenous translocase of inner membrane 23 (detected by Western blot analysis). Part of the imported Sirt5 was degraded by PK, indicating its localization in the IMS.

Sirtuins are involved in central physiological regulation mechanisms, many of them with relevance to metabolic regulation and aging processes.5 and 6 Therefore, the seven mammalian Sirtuin isoforms are emerging targets for the treatment of metabolic disorders and aging-related diseases.39 For most Sirtuin effects, however, the specific signaling mechanisms and molecular targets are not yet known. We have identified novel potential targets for Sirtuins in mitochondria, the major metabolic centers in cells. We found that Sirt3 can deacetylate and thereby activate ICDH2, a key regulation point for flux throughout the citric acid cycle. Interestingly, the ICDH isoform regulated by Sirt3 forms NADPH instead of the NADH used for ATP synthesis. This activity is assumed to be important for the NADPH-dependent regeneration of antioxidants,40 and its stimulation by Sirt3 should thus help to slow oxidative damage and cellular aging processes. Furthermore, Sirt3 deacetylates GDH in vitro (this study) and in vivo31 and we find that this modification also stimulates GDH activity that promotes glucose and ATP synthesis by enabling amino acids to be used as fuels for citric acid cycle and gluconeogenesis. 41 Consistently, Sirt3 was reported to increase respiration, 24 which is needed for ATP synthesis but also for conversion of amino acids into glucose and urea. 41 The enzyme previously identified to be activated by Sirt3, acetyl coenzyme A synthetase 2, 21 and 22 also fuels the citric acid cycle independently of glycolysis by activating free acetate (Fig. 5b). Interestingly, a shift away from liver glycolysis is one of the metabolic changes observed under CR, a feeding regimen with 20–40% fewer calories than consumed ad libitum that is found to extend the lifespan of a variety of organisms. 6 CR was previously reported to increase GDH activity in the liver, 42where Sirt3 is highly expressed, 17 and Sirt3 activity is known to be increased by CR. 6 and 24 It thus appears that Sirt3 mediates some of the effects of CR and lifespan regulation, consistent with its implication in survivorship in the elderly 25 and 43 and the prominent role of Sirtuins in CR found for various organisms,6 and 44 and it also appears that GDH activation likely contributes to the Sirt3-dependent effects.

Little is known about additional factors regulating the activity and specificity of Sirtuin enzymes. Their requirement for NAD + indicates that the NAD +/NADH ratio should regulate Sirtuins,13 and 14 but even changes to ratios observed under extreme conditions such as CR appear to influence Sirtuin activity only slightly.35 Furthermore, NAD + levels would influence all Sirtuins similarly, but a more specific tuning of individual Sirtuin activities appears necessary in order to orchestrate the many effects mediated by Sirtuins (see, e.g., discussion above).6 and 45 A deeper insight into the regulation of Sirtuin enzymes would also be required for the development of more specific Sirtuin inhibitors—a prerequisite for Sirtuin-targeted therapy.39 The regulatory parts flanking the catalytic cores might be interesting target sites (Fig. 5f). N-terminal extensions between ∼ 30 and 120 residues are present in all human Sirtuins but show little conservation, indicating that they might respond to various regulators. Our results indicate that the corresponding N-terminal region in Sirt3 also blocks productive binding for small peptides (Fig. 4a), but enables access for entire protein substrates (Fig. 4c). The C-terminal truncated part in our experiments (Sirt3 residues 380–399) is formed by α14 (secondary structure numbering for Sirt236) whose end corresponds to the N-terminus of Hst2 α13 that partly occupies the NAD +binding site.15 In Sirt3, however, the C-terminal truncation alone lowers activity only slightly, and we assume that it has no regulatory function on its own but might instead assist the N-terminal autoinhibitory region. This module of the N-terminus and the C-terminus (Figs. 4b and 5f) appears to contribute to the substrate specificity of the enzyme, and ligands binding to it might enable or block rearrangements opening up the active site and thereby regulate the enzyme’s activity. Alternatively, the flanking parts might be removed by proteolytic processing or alternative splicing, thereby changing Sirtuin activity and specificity.

7.8.3 The mTORC1 Pathway Stimulates Glutamine Metabolism and Cell Proliferation by Repressing SIRT4

Csibi A1Fendt SMLi CPoulogiannis GChoo AYChapski DJ, et al.
Cell. 2013 May 9; 153(4):840-54.

Proliferating mammalian cells use glutamine as a source of nitrogen and as a key anaplerotic source to provide metabolites to the tricarboxylic acid cycle (TCA) for biosynthesis. Recently, mTORC1 activation has been correlated with increased nutrient uptake and metabolism, but no molecular connection to glutaminolysis has been reported. Here, we show that mTORC1 promotes glutamine anaplerosis by activating glutamate dehydrogenase (GDH). This regulation requires transcriptional repression of SIRT4, the mitochondrial-localized sirtuin that inhibits GDH. Mechanistically, mTORC1 represses SIRT4 by promoting the proteasome-mediated destabilization of cAMP response element binding-2 (CREB2). Thus, a relationship between mTORC1, SIRT4 and cancer is suggested by our findings. Indeed, SIRT4 expression is reduced in human cancer, and its overexpression reduces cell proliferation, transformation and tumor development. Finally, our data indicate that targeting nutrient metabolism in energy-addicted cancers with high mTORC1 signaling may be an effective therapeutic approach.

Proliferating mammalian cells use glutamine as a source of nitrogen and as a key anaplerotic source to provide metabolites to the tricarboxylic acid cycle (TCA) for biosynthesis. Recently, mTORC1 activation has been correlated with increased nutrient uptake and metabolism, but no molecular connection to glutaminolysis has been reported. Here, we show that mTORC1 promotes glutamine anaplerosis by activating glutamate dehydrogenase (GDH). This regulation requires transcriptional repression of SIRT4, the mitochondrial-localized sirtuin that inhibits GDH. Mechanistically, mTORC1 represses SIRT4 by promoting the proteasome-mediated destabilization of cAMP response element binding-2 (CREB2). Thus, a relationship between mTORC1, SIRT4 and cancer is suggested by our findings. Indeed, SIRT4 expression is reduced in human cancer, and its overexpression reduces cell proliferation, transformation and tumor development. Finally, our data indicate that targeting nutrient metabolism in energy-addicted cancers with high mTORC1 signaling may be an effective therapeutic approach.

Nutrient availability plays a pivotal role in the decision of a cell to commit to cell proliferation. In conditions of sufficient nutrient sources and growth factors (GFs), the cell generates enough energy and acquires or synthesizes essential building blocks at a sufficient rate to meet the demands of proliferation. Conversely, when nutrients are scarce, the cell responds by halting the biosynthetic machinery and by stimulating catabolic processes such as fatty acid oxidation and autophagy to provide energy maintenance (Vander Heiden et al., 2009). Essential to the decision process between anabolism and catabolism is the highly conserved, atypical Serine/Threonine kinase mammalian Target of Rapamycin Complex 1 (mTORC1), whose activity is deregulated in many cancers (Menon and Manning, 2008). This complex, which consists of mTOR, Raptor, and mLST8, is activated by amino acids (aa), GFs (insulin/IGF-1) and cellular energy to drive nutrient uptake and subsequently proliferation (Yecies and Manning, 2011). The molecular details of these nutrient-sensing processes are not yet fully elucidated, but it has been shown that aa activate the Rag GTPases to regulate mTORC1 localization to the lysosomes (Kim et al., 2008Sancak et al., 2008); and GFs signal through the PI3K-Akt or the extracellular signal-regulated kinase (ERK)-ribosomal protein S6 kinase (RSK) pathways to activate mTORC1 by releasing the Ras homolog enriched in brain (RHEB) GTPase from repression by the tumor suppressors, tuberous sclerosis 1 (TSC1)– TSC2 (Inoki et al., 2002Manning et al., 2002Roux et al., 2004). Finally, low energy conditions inhibit mTORC1 by activating AMPK and by repressing the assembly of the TTT-RUVBL1/2 complex. (Inoki et al., 2003Gwinn et al., 2008Kim et al., 2013).

Glutamine, the most abundant amino acid in the body plays an important role in cellular proliferation. It is catabolized to α-ketoglutarate (αKG), an intermediate of the tricarboxylic acid (TCA) cycle through two deamination reactions in a process termed glutamine anaplerosis (DeBerardinis et al., 2007). The first reaction requires glutaminase (GLS) to generate glutamate, and the second occurs by the action of either glutamate dehydrogenase (GDH) or transaminases. Incorporation of αKG into the TCA cycle is the major anaplerotic step critical for the production of biomass building blocks including nucleotides, lipids and aa (Wise and Thompson, 2010). Recent studies have demonstrated that glutamine is also an important signaling molecule. Accordingly, it positively regulates the mTORC1 pathway by facilitating the uptake of leucine (Nicklin et al., 2009) and by promoting mTORC1 assembly and lysosomal localization (Duran et al., 2012;Kim et al., 2013).

Commonly occurring oncogenic signals directly stimulate nutrient metabolism, resulting in nutrient addiction. Oncogenic levels of Myc have been linked to increased glutamine uptake and metabolism through a coordinated transcriptional program (Wise et al., 2008Gao et al., 2009). Hence, it is not surprising that cancer cells are addicted to glutamine (Wise and Thompson, 2010). Thus, considering the prevalence of mTORC1 activation in cancer and the requirement of nutrients for cell proliferation, understanding how mTORC1 activation regulates nutrient levels and metabolism is critical. Activation of the mTORC1 pathway promotes the utilization of glucose, another nutrient absolutely required for cell growth. However, no study has yet investigated if and how the mTORC1 pathway regulates glutamine uptake and metabolism. Here, we discover a novel role of the mTORC1 pathway in the stimulation of glutamine anaplerosis by promoting the activity of GDH. Mechanistically, mTORC1 represses the transcription of SIRT4, an inhibitor of GDH. SIRT4 is a mitochondrial-localized member of the sirtuin family of NAD-dependent enzymes known to play key roles in metabolism, stress response and longevity (Haigis and Guarente, 2006). We demonstrate that the mTORC1 pathway negatively controls SIRT4 by promoting the proteasome-mediated degradation of cAMP-responsive element-binding (CREB) 2. We reveal that SIRT4 levels are decreased in a variety of cancers, and when expressed, SIRT4 delays tumor development in a Tsc2−/− mouse embryonic fibroblasts (MEFs) xenograft model. Thus, our findings provide new insights into how mTORC1 regulates glutamine anaplerosis, contributing therefore to the metabolic reprogramming of cancer cells, an essential hallmark to support their excessive needs for proliferation.

The mTORC1 pathway regulates glutamine metabolism via GDH

The activation of the mTORC1 pathway has recently been linked to glutamine addiction of cancer cells (Choo et al., 2010), yet it remains to be resolved if mTORC1 serves as a regulator of glutamine anaplerosis. To investigate this possibility, we first determined the effect of mTORC1 activity on glutamine uptake. We measured glutamine uptake rates in Tsc2 wild-type (WT) and Tsc2−/− MEFs. We found that Tsc2−/− MEFs consumed significantly more glutamine (Figure 1A), showing that mTORC1 activation stimulates the uptake of this nutrient. In addition, re-expression of Tsc2 in Tsc2−/− cells reduced glutamine uptake (Figure S1A). Similarly, mTORC1 inhibition with rapamycin resulted in decreased glutamine uptake in MEFs (Figure 1A). The decreased on glutamine uptake was significantly reduced after 6h of rapamycin treatment when compared to control (data not shown). To further confirm the role of mTORC1 on glutamine uptake, we used human embryonic kidney (HEK) 293T cells stably expressing either WT-RHEB or a constitutively active mutant (S16H) of RHEB. Increased mTORC1 signaling, as evidenced by sustained phosphorylation of S6K1 and its target rpS6, was observed in RHEB-expressing cells (Figure S1B). The activation of the mTORC1 pathway nicely correlated with an increase in glutamine consumption, therefore confirming that changes in mTORC1 signaling are reflected in cellular glutamine uptake (Figure S1B). To determine whether the modulation of glutamine uptake by the mTORC1 pathway occurs in cancer cells, we examined glutamine uptake rates in conditions of mTORC1 inhibition in human epithelial tumor cell lines, including the colon carcinoma DLD1, and the prostate cancer DU145. Rapamycin treatment resulted in decreased proliferation (data not shown) and yielded a decreased glutamine uptake in both cell lines (Figure 1B & data not shown). Glutamine is the major nitrogen donor for the majority of ammonia production in cells (Figure 1C) (Shanware et al., 2011). Consistent with decreased glutamine uptake, we found that ammonia levels were also diminished after rapamycin treatment (Figure S1C).

Figure 1  The mTORC1 pathway regulates glutamine metabolism via glutamate dehydrogenase

We next examined the fate of glutamine in conditions of mTORC1 inhibition, using gas chromatography/mass spectrometry (GC/MS) analysis to monitor the incorporation of uniformly labeled [U-13C5]-Glutamine into TCA cycle intermediates. Direct glutamine contribution to I̧KG (m+5), succinate (m+4), malate (m+4) and citrate (m+4) was decreased in rapamycin treated cells (Figure S1D) indicating that rapamycin impaired glutamine oxidation and subsequent carbon contribution into the TCA cycle.

To test whether glutamine uptake or glutamine conversion is limiting, we measured the intracellular levels of glutamine and glutamate in DLD1 cells. Increased levels of glutamine and/or glutamate will show that the catalyzing enzyme activity is limiting and not glutamine transport itself (Fendt et al., 2010). Rapamycin treatment resulted in increased intracellular levels of both glutamine and glutamate, showing that glutamate to αKG conversion is the critical limiting reaction (Figures 1D & 1E). To further confirm the implication of the glutamate catalyzing reaction we also measured αKG levels. If glutamate conversion is indeed critical we expect no alteration in αKG levels. This is expected because αKG is downstream of the potentially limiting glutamate conversion step, and it has been shown that product metabolite concentrations of limiting metabolic enzymes stay unaltered, while the substrate metabolite concentrations change to keep metabolic homeostasis (Fendt et al., 2010). We found that αKG levels were unaltered after rapamycin treatment, corroborating that the limiting enzymatic step is glutamate conversion (Figure 1F). To further confirm the limitation in glutamate-to-αKG conversion, we measured flux through this reaction. Strikingly, this flux was significantly reduced during rapamycin treatment (Figure 1G). Additionally, the inhibition of mTORC1 resulted in increased glutamate secretion (Figure 1H), thus confirming that the glutamate-to-αKG conversion step is a major bottleneck in the glutamine pathway during rapamycin treatment.

Glutamate conversion can be conducted by GDH (Figure 1C), suggesting that the mTORC1 pathway potentially regulates this enzyme. In agreement, rapamycin treatment resulted in decreased GDH activity in DLD1 cells (Figure 1I). To exclude that transaminases play a role in the mTORC1-induced regulation of glutamine metabolism, we used amin ooxyacetate (AOA) at a concentration shown to effectively inhibit the two predominant transaminases, alanine aminotransferase (ALT) and aspartate aminotransferase (AST) (Figure 1C) (Wise et al., 2008), or rapamycin in the presence of α-15N-labeled glutamine. Subsequently, we measured 15N-labeling patterns and metabolite levels of alanine, an amino acid that is predominately produced by a transaminase-catalyzed reaction (Possemato et al., 2011). We found that AOA dramatically decreased 15N contribution and metabolite levels of alanine, while rapamycin only mildly affected the 15N contribution to this amino acid and showed no effect on alanine levels compared to the control condition (Figures 1J & S1E). In conclusion, these data demonstrate that GDH, not transaminases, plays a major role in the regulation of glutamine metabolism downstream of mTORC1.

mTORC1 controls GDH activity by repressing SIRT4

As our results show that mTORC1 regulates glutamate dehydrogenase, we sought to identify the molecular mechanism. SIRT4 is a negative regulator of GDH activity through ADP-ribosylation (Haigis et al., 2006), thus suggesting that mTORC1 potentially controls this step of glutamine metabolism via SIRT4. To test this possibility, we first assessed the ADP-ribosylation status of GDH by introducing biotin-labeled NAD followed by immunoprecipitation using avidin-coated beads. Rapamycin treatment led to an increase in the mono-ADP-ribosylation status of GDH, similar to that observed in cells stably expressing SIRT4 (Figure 2A). Importantly, we found that the knockdown of SIRT4 abrogated the rapamycin-induced decrease in the activity of GDH (Figures 2B & S2A). Strikingly, SIRT4 protein levels were increased upon mTORC1 inhibition in MEFs (Figures 2C). This regulation was confirmed in both DLD1 and DU145 cells (Figures 2D). Remarkably, rapamycin potently increased SIRT4 levels after 6h of treatment (Figure S2B), correlating with reduced glutamine consumption at the same time point (data not shown). In contrast, SIRT4 levels were not influenced by the treatment of MEFs with U0216, an inhibitor of MEK1/2 in the MAPK pathway (Figure S2C). All other mTOR catalytic inhibitors tested in Tsc2−/− MEFs also resulted in increased SIRT4 protein levels (Figure S2D). To evaluate a potential regulation of SIRT4 by mTORC2, we performed RNA interference (RNAi) experiments of either raptor or the mTORC2 component, rictor, in Tsc2−/− MEFs. The knockdown of raptor, but not rictor, was sufficient to increase SIRT4 protein levels, confirming the role of the mTORC1 pathway in the regulation of SIRT4 (Figure 2E). To investigate whether mTORC1 regulation of SIRT4 occurs in tumor samples, a TSC-xenograft model was used. We injected a TSC2−/− rat leiomyoma cell line; ELT3 cells, expressing either an empty vector (V3) or TSC2 (T3), in the flank of nude mice. SIRT4 levels were dramatically increased in TSC2-expressing tumors compared to empty vector samples (Figure S2E). In addition, we assessed the levels of SIRT4 in both ELT3 xenograft tumors and in mouse Tsc2+/− liver tumors after rapamycin treatment. As expected, these tumor samples exhibited robust elevation of SIRT4 after rapamycin treatment (Figures 2F & S2F). Thus, these data demonstrate that the mTORC1 pathway represses SIRT4 in several tumor systems.

Figure 2  mTORC1 controls glutamate dehydrogenase activity by repressing SIRT4

CREB2 regulates the transcription of SIRT4 in an mTORC1-dependent fashion

We next asked whether the mTORC1-dependent regulation of SIRT4 occurred at the mRNA level. Quantitative RT-PCR results show that rapamycin treatment significantly increased the expression of SIRT4mRNA in Tsc2−/− MEFs (Figure 3A). SIRT4 mRNA levels were dramatically reduced in Tsc2−/− MEFs compared to their WT counterpart (Figure 3B). Similar results were obtained from transcriptional profiling analysis of the SIRT4 gene from a previously published dataset (GSE21755) (Figure 3C) (Duvel et al. 2010). Altogether, our data demonstrate that mTORC1 negatively regulates the transcription of SIRT4. To determine whether CREB2 is involved in the mTORC1-dependent regulation of SIRT4, we performed RNAi experiments. The silencing of CREB2 abolished the rapamycin-induced expression of SIRT4 (Figures 3E & S3A). The knockdown of CREB1 did not affect the upregulation of SIRT4 upon mTORC1 inhibition, thus demonstrating the specificity of CREB2 to induce SIRT4 (Figure S3B), and the knockdown of CREB2 significantly abrogated the rapamycin-induced increase in the activity of the SIRT4 promoter.

Figure 3  SIRT4 is regulated at the mRNA level in an mTORC1-dependent fashion

mTORC1 regulates the stability of CREB2

We next investigated whether the mTORC1 pathway regulates CREB2. Although we did not observe major changes in Creb2 mRNA in normal growth conditions (Figure S4A), mTORC1 inhibition resulted in accumulation of CREB2 protein levels by 2h of rapamycin treatment (Figure 4A). U0126 failed to cause the accumulation of CREB2 (Figure S4B). In contrast, CREB1 protein levels were not affected after 24h rapamycin treatment (Figure S4C). As observed for SIRT4, mTOR catalytic inhibitors, and the specific knockdown of mTOR, resulted in upregulation of CREB2 protein levels (Figures S4D & S4E). CREB2 is upregulated in diverse cell types as a response to a variety of stresses, including hypoxia, DNA damage, and withdrawal of GFs, glucose, and aa (Cherasse et al., 2007Rouschop et al., 2010Yamaguchi et al., 2008;Whitney et al., 2009). Interestingly, mTORC1 is negatively regulated by all of these environmental inputs (Zoncu et al., 2011). Since mTORC1 signaling in Tsc2−/− MEFs is insensitive to serum deprivation, we assessed the role of aa withdrawal and re-stimulation on CREB2 levels. As shown in Fig. 4B, CREB2 accumulated upon aa deprivation, and was decreased following aa re-addition. This phenomenon required the action of the proteasome as MG132 efficiently blocked CREB2 degradation following aa re-addition. Importantly, we found that mTORC1 inhibition abrogated the aa-induced decrease of CREB2 (Figure 4B).

Figure 4  mTORC1 regulates the stability of CREB2

mTORC1 activation promotes the binding of CREB2 to βTrCP and modulates CREB2 ubiquitination

Next, we attempted to identify the E3 ubiquitin ligase that might be responsible for CREB2 turnover. Consistent with a recent study, we found CREB2 to bind the E3 ligase, βTrCP (Frank et al., 2010). However, other related E3 ligases including Fbxw2, Fbxw7a, and Fbxw9 did not bind to CREB2 (data not shown). The interaction of CREB2 with Flag-βTrCP1 was enhanced in the presence of insulin, and was abolished by rapamycin pretreatment (Figure 4D). Importantly, insulin treatment promoted the ubiquitination of CREB2 in an mTORC1-dependent fashion (Figure 4E). Altogether, our results support the notion that the mTORC1 pathway regulates the targeting of CREB2 for proteasome-mediated degradation. βTrCP binds substrates via phosphorylated residues in conserved degradation motifs (degrons), typically including the consensus sequence DpSGX(n)pS or similar variants. We found an evolutionary conserved putative βTrCP binding site (DSGXXXS) in CREB2 (Figure 4F). Interestingly, we noted a downward mobility shift in CREB2 protein with mTORC1 inhibition, consistent with a possible decrease in the phosphorylation of CREB2. (Figure 4A). Frank et al. (2010) showed that phosphorylation of the first serine in the degron motif corresponding to Ser218 is required for the CREB2/βTrCP interaction, and this modification acts as a priming site for a gradient of phosphorylation events on five proline-directed residues codons (T212, S223, S230, S234, and S247) that is required for CREB2 degradation during the cell cycle progression (Frank et al., 2010). Consistent with these observations, we found that the mutation of the five residues to alanine (5A mutant) resulted in strong stabilization of CREB2, comparable to the serine-to-alanine mutation on the priming Ser218 phosphorylation site (Figure S4G).

SIRT4 represses bioenergetics and cell proliferation

We observed that glutamine utilization is repressed by rapamycin treatment (Figure 1) and SIRT4 is induced by mTORC1 inhibition (Figure 2). Thus, we tested whether SIRT4 itself directly regulates cellular glutamine uptake. The stable expression of SIRT4 resulted in the repression of glutamine uptake in Tsc2−/− MEFs and DLD1 cells (Figures 5A & 5B). Glucose uptake was not affected by SIRT4 expression (data not shown). Because glutamine can be an important nutrient for energy production, we examined ATP levels in SIRT4 expressing cells. Consistent with reduced glutamine consumption, the expression of SIRT4 in Tsc2−/− cells resulted in decreased ATP/ADP ratio compared to control cells (Figure 5C). Cells produce ATP via glycolysis and oxidative phosphorylation (OXPHOS). To test the contribution of mitochondrial metabolism versus glycolysis to ATP, we measured the ATP/ADP ratio after the treatment with oligomycin, an inhibitor of ATP synthesis from OXPHOS. Importantly, the difference of the ATP/ADP ratio between control and SIRT4 expressing cells was abrogated by oligomycin (Figure 5C), further demonstrating that SIRT4 may repress the ability of cells to generate energy from mitochondrial glutamine catabolism. Mitochondrial glutamine catabolism is essential for energy production and viability in the absence of glucose (Yang et al., 2009Choo et al., 2010). Thus, we examined the effect of SIRT4 on the survival of Tsc2−/− MEFs during glucose deprivation. Control cells remained viable following 48h of glucose deprivation. Conversely, SIRT4 expressing cells showed a dramatic increase in cell death under glucose-free conditions, which was rescued by the addition of the cell permeable dimethyl-I̧KG (DM-I̧KG) (Figure 5D). Conversely, the expression of SIRT4 did not affect the viability of glucose-deprived Tsc2 WT MEFs (Figure S5A). Glucose deprivation also induced death of the human DU145 cancer cell line stably expressing SIRT4 (data not shown).

Figure 5  SIRT4 represses bioenergetics and proliferation

Glutamine is an essential metabolite for proliferating cells, and many cancer cells exhibit a high rate of glutamine consumption (DeBerardinis et al., 2007). Thus, decreased glutamine uptake in DLD1 and DU145 cancer cells expressing SIRT4 might result in decreased proliferation. Indeed, these cells grew significantly slower than did control cells. Remarkably, DM-I̧KG completely abrogated the decreased proliferation of SIRT4 expressing cells (Figure 5E & 5F), suggesting that repressed glutamine metabolism drove the reduced proliferation of cells expressing SIRT4. The expression of SIRT4 also slowed the proliferation of Tsc2−/− MEFs but did not affect Tsc2 WT MEFs (Figures S5B & S5C). Finally, to rule out that the effect on proliferation was due to aberrant localization and to off-target effects of the overexpressed protein, we examined the localization of HA-SIRT4. We found that SIRT4 is co-localized with the MitoTracker, a mitochondrial-selective marker (Figure S5D). Taken together, these data demonstrate that SIRT4 is a critical negative regulator of mitochondrial glutamine metabolism and cell proliferation.

SIRT4 represses TSC-tumor development

Recent studies have demonstrated a major role of glutamine metabolism in driving oncogenic transformation of many cell lines (Gao et al., 2009Wang et al., 2011). Since SIRT4 expression represses glutamine uptake and cell proliferation (Figure 5), we hypothesized that it could affect tumorigenesis. To test this idea, we assessed the role of SIRT4 in cell transformation by using an anchorage-independent growth assay. SIRT4 expression reduced the ability of Tsc2−/−p53−/− MEFs to grow in soft agar. However, the expression of SIRT4 in Tsc2+/+p53−/− did not impair their colony formation properties (Figure 6A). Tumor incidence in mice injected with Tsc2+/+p53−/− MEFs was not affected by SIRT4 (data not shown). Conversely, in the Tsc2−/−p53−/− cohort, SIRT4 reduced tumor incidence by 20 days at median (Figure 6B). SIRT4 expression inTsc2−/−p53−/− MEFs resulted in reduction of Ki-67 positivity by 60% (Figure 6E), consistent with the finding that SIRT4 inhibits the proliferation of these cells in vitro (Figure S5B). Finally, we performed a comprehensive meta-analysis of SIRT4 expression in human tumors and found significantly lower expression levels of SIRT4, relative to normal tissue, in bladder, breast, colon, gastric, ovarian and thyroid carcinomas (Figure 6F). Interestingly, loss of SIRT4 expression showed a strong association with shorter time to metastasis in patients with breast cancer (Figures 6G & 6H). Altogether, these data strongly suggest that SIRT4 delays tumorigenesis regulated by the mTORC1 pathway.

Figure 6
SIRT4 suppresses TSC-tumor development

The pharmacologic inhibition of glutamine anaplerosis synergizes with glycolytic inhibition to induce the specific death of mTORC1 hyperactive cells

The activation of mTORC1 leads to glucose and glutamine addiction as a result of increased uptake and metabolism of these nutrients (Choo et al., 2010Duvel et al., 2010 & Figure 1). These observations suggest that targeting this addiction offers an interesting therapeutic approach for mTORC1-driven tumors. The alkylating agent, mechlorethamine (Mechlo), incites cell toxicity in part by the inhibition of the GAPDH step of glycolysis via poly-ADP ribose polymerase (PARP)-dependent cellular consumption of cytoplasmic NAD+. The ultimate consequence is glycolytic inhibition, thus mimicking glucose deprivation (Zong et al., 2004). Treatment of Tsc2−/− MEFs with Mechlo decreased both NAD levels and lactate production (Figure 7A and data not shown). The decrease in NAD+ levels was rescued by addition of DPQ (Figure 7A), a PARP inhibitor (Zong et al., 2004). We next tested the ability of glutamine inhibition to determine the sensitivity of Tsc2−/− MEFs to Mechlo. As shown in Figure 7B, the treatment with EGCG, a GDH inhibitor (Figure 1G), potently synergized with Mechlo to kill Tsc2−/− MEFs with the greatest effect observed at 30μM (Figure 7B). As a result, this combination dramatically increased the cleavage of PARP, an apoptotic marker (Figure 7E). Similarly, glutamine deprivation sensitized Tsc2−/− MEFs to Mechlo (data not shown). The RNAi-mediated knockdown of GDH also synergized with Mechlo to induce death of Tsc2−/− MEFs (Figure 7D). Importantly, at these concentrations the combination did not induce death of a Tsc2-rescued cell line (Figure 7C).

Figure 7 The combination of glutamine metabolism inhibitors with glycolytic inhibition is an effective therapy to kill Tsc2−/− and PTEN−/− cells

Because the metabolic properties of cells with activated mTORC1 by Tsc2– deficiency can be efficiently targeted, we also examined other cell types in which mTORC1 is hyperactive by the loss of PTEN. We found that the combination of Mechlo and EGCG was also effective to induce specific toxicity of PTEN−/− MEFs, while PTEN+/+ MEFs were not affected (Figures S7A & S7B). In addition, the PTEN-deficient human prostate adenocarcinoma cell line, LNCaP, was also sensitive to treatment with Mechlo and EGCG (Figure 7F). This effect was specifically due to lack of TCA cycle replenishment as pyruvate supplementation completely reversed the synergistic effect (Figure 7F). The combination of Mechlo with the GLS1 inhibitor, BPTES (Figure 1G), also resulted in decreased viability of Tsc2−/− cells but not of Tsc2-reexpressing cells (Figures S7C & S7D). Again, death in Tsc2−/− cells was rescued with pyruvate or OAA (Figure S7E). To further investigate if the potent cell death in Tsc2−/− was restricted to Mechlo, we used 2-DG, a glycolytic inhibitor. The combination of 2-DG with either EGCG or BPTES resulted in enhanced cell death of Tsc2−/− MEFs compared to single agent treatments (Figure S7F). This effect was also specific to Tsc2−/− cells, since this combination was less toxic in Tsc2-reexpressing MEFs (Figure S7G). Taken together, our results demonstrate that the combination treatments aimed at inhibiting glycolysis and glutaminolysis potently synergize to kill cells with hyperactive mTORC1 signaling.

Here, we define a novel mTORC1-regulated pathway that controls glutamine-dependent anaplerosis and energy metabolism (Figure 7G). We discovered that the mTORC1 pathway regulates glutamine metabolism by promoting the activity of GDH (Figures 1​-3).3). We show that this regulation occurs by repressing the expression of SIRT4, an inhibitor of GDH (Figures 2 & 3). Molecularly, this is the result of mTORC1-dependent proteasome-mediated degradation of the SIRT4 transcriptional regulator, CREB2 (Figure 4). Interestingly, the modulation of CREB2 levels correlates with increased sensitivity to glutamine deprivation (Ye et al., 2010Qing et al., 2012), fitting with our model of glutamine addiction as a result of mTORC1 activation (Choo et al., 2010). Our data suggest that mTORC1 promotes the binding of the E3 ligase, βTrCP, to CREB2 (Figure 4D), promoting CREB2 degradation by the proteasome (Figure 4E). A previous study has demonstrated that five residues in CREB2 located next to the βTrCP degron are required for its stability (Frank et al., 2010). Accordingly, the mutation of these residues to alanine resulted in stabilization of CREB2 and SIRT4 following insulin and aa-dependent mTORC1 activation (Figure 4G). Future work is aimed at determining if mTORC1 and/or downstream kinases are directly responsible for the multisite phosphorylation of CREB2.

The identification of CREB2 as an mTORC1-regulated transcription factor increases the repertoire of transcriptional regulators modulated by this pathway including HIF1α (glycolysis), Myc (glycolysis) and SREBP1 (lipid biosynthesis) (Duvel et al., 2010Yecies and Manning, 2011). The oncogene Myc has also been linked to the regulation of glutamine metabolism by increasing the expression of the surface transporters ASCT2 and SN2, and the enzyme GLS. Thus, enhanced activity of Myc correlates with increased glutamine uptake and glutamate production (Wise et al., 2008Gao et al., 2009). Our findings describe a new level of control to this metabolic node as shown by the modulation of the glutamate-to-αKG flux (Figure 2). This regulation is particularly relevant as some cancer cells produce more than 50% of their ATP by oxidizing glutamine-derived αKG in the mitochondria (Reitzer et al JBC, 1979). Therefore, these studies support the notion that Myc and CREB2/SIRT4 cooperate to regulate the metabolism of glutamine to αKG.

7.8.4  Rab1A and small GTPases Activate mTORC1 Rab1A Is an mTORC1 Activator and a Colorectal Oncogene

Thomas JD1Zhang YJ2Wei YH3Cho JH3Morris LE3Wang HY4Zheng XF5.
Cancer Cell. 2014 Nov 10; 26(5):754-69.


  • Rab1A mediates amino acid signaling to activate mTORC1 independently of Rag
  • Rab1A regulates mTORC1-Rheb interaction on the Golgi apparatus
  • Rab1A is an oncogene that is frequently overexpressed in human cancer
  • Hyperactive amino acid signaling is a common driver for cancer

Amino acid (AA) is a potent mitogen that controls growth and metabolism. Here we describe the identification of Rab1 as a conserved regulator of AA signaling to mTORC1. AA stimulates Rab1A GTP binding and interaction with mTORC1 and Rheb-mTORC1 interaction in the Golgi. Rab1A overexpression promotes mTORC1 signaling and oncogenic growth in an AA- and mTORC1-dependent manner. Conversely, Rab1A knockdown selectively attenuates oncogenic growth of Rab1-overexpressing cancer cells. Moreover, Rab1A is overexpressed in colorectal cancer (CRC), which is correlated with elevated mTORC1 signaling, tumor invasion, progression, and poor prognosis. Our results demonstrate that Rab1 is an mTORC1 activator and an oncogene and that hyperactive AA signaling through Rab1A overexpression drives oncogenesis and renders cancer cells prone to mTORC1-targeted therapy. Regulation of TOR by small GTPases

Raúl V Durán1 and Michael N Halla,1
EMBO Rep. 2012 Feb; 13(2): 121–128.

TOR is a conserved serine/threonine kinase that responds to nutrients, growth factors, the bioenergetic status of the cell and cellular stress to control growth, metabolism and ageing. A diverse group of small GTPases including Rheb, Rag, Rac1, RalA and Ryh1 play a variety of roles in the regulation of TOR. For example, while Rheb binds to and activates TOR directly, Rag and Rac1 regulate its localization and RalA activates it indirectly through the production of phosphatidic acid. Here, we review recent findings on the regulation of TOR by small GTPases.

The growth-controlling TOR signalling pathway is structurally and functionally conserved from unicellular eukaryotes to humans. TOR, an atypical serine/threonine kinase, was originally discovered inSaccharomyces cerevisiae as the target of rapamycin (Heitman et al, 1991). It was later described in many other organisms including the protozoan Trypanosoma brucei, the yeast Schizosaccharomyces pombe, photosynthetic organisms such as Arabidopsis thaliana and Chlamydomonas reinhardtii, and in metazoans such as Caenorhabditis elegansDrosophila melanogaster and mammals. TOR integrates various stimuli to control growth, metabolism and ageing (Avruch et al, 2009Kim & Guan, 2011Soulard et al, 2009;Wullschleger et al, 2006Zoncu et al, 2011a). In mammals, mTOR is activated by nutrients, growth factors and cellular energy, and is inhibited by stress. Thus, the molecular regulation of TOR is complex and diverse. Among the increasing number of TOR regulators, small GTPases are currently garnering much attention. Small GTPases (20–25 kDa) are either in an inactive GDP-bound form or an active GTP-bound form (Bos et al, 2007). GDP–GTP exchange is regulated by GEFs, which mediate the replacement of GDP by GTP, and by GAPs, which stimulate the intrinsic GTPase activity of a cognate GTPase to convert GTP into GDP (Fig 1). Upon activation, small GTPases interact with effector proteins, thereby stimulating downstream signalling pathways. Small GTPases constitute a superfamily that comprises several subfamilies, such as the Rho, Ras, Rab, Ran and Arf families. Rheb, Rag, RalA, Rac1 and Ryh1, all members of the small GTPase superfamily, play a role in the concerted regulation of TOR by different stimuli. This review summarizes recent advances in the understanding of TOR regulation by these small GTPases.

Regulation of small GTPases by GEFs and GAPs

Regulation of small GTPases by GEFs and GAPs

Figure 1 Regulation of small GTPases by GEFs and GAPs. A guanine nucleotide exchange factor (GEF) replaces GDP with GTP to activate the signalling function of the GTPase. Conversely, a GTPase-activating protein (GAP) stimulates hydrolysis of GTP into GDP

The TOR complexes

TOR is found in two functionally and structurally distinct multiprotein complexes, named TORC1 and TORC2 (Avruch et al, 2009Kim & Guan, 2011Soulard et al, 2009Wullschleger et al, 2006Zoncu et al, 2011a). TORC1 regulates several cellular processes including protein synthesis, ribosome biogenesis, nutrient uptake and autophagy. TORC2, in turn, regulates actin cytoskeleton organization, cell survival, lipid synthesis and probably other processes. TORC1 and TORC2 are rapamycin-sensitive and rapamycin-insensitive, respectively, although in some organisms, for example A. thaliana and T. brucei, this rule does not apply (Barquilla et al, 2008Mahfouz et al, 2006). Nevertheless, long-term treatment with rapamycin can also indirectly inhibit TORC2 in mammalian cell lines (Sarbassov et al, 2006). Furthermore, there is accumulating evidence that not all TORC1 readouts are rapamycin-sensitive (Choo & Blenis, 2009Dowling et al, 2010Peterson et al, 2011).

Upstream of TOR

Four main inputs regulate mTORC1: nutrients, growth factors, the bioenergetic status of the cell and oxygen availability. It is well established that growth factors activate mTORC1 through the PI3K–AKT pathway. Once activated, AKT phosphorylates and inhibits the heterodimeric complex TSC1–TSC2, a GAP for Rheb and thus an inhibitor of mTORC1 (Avruch et al, 2009). The TSC1–TSC2 heterodimer is a ‘reception centre’ for various stimuli that are then transduced to mTORC1, including growth factor signals transduced through the AKT and ERK pathways, hypoxia through HIF1 and REDD1, and energy status through AMPK (Wullschleger et al, 2006). In addition to the small GTPases Rheb and Rag (see below), PA also binds to and activates mTORC1 (Fang et al, 2001). Pharmacological or genetic inhibition of PA production, through the inhibition of PLD, impairs activation of mTORC1 by nutrients and growth factors (Fang et al, 2001). Moreover, elevated PLD activity leads to rapamycin resistance in human breast cancer cells (Chen et al, 2003), further supporting a role for PA as an mTORC1 regulator. As discussed below, the small GTPase RalA participates in the mechanism by which PA activates mTORC1 (Maehama et al, 2008Xu et al, 2011).

In the case of nutrients, amino acids in particular, several elements mediate the activation of TORC1. As discussed below, the Rag GTPases are necessary to activate TORC1 in response to amino acids (Binda et al, 2009Kim et al, 2008Sancak et al, 2008). In mammals, it has also been proposed that amino acids stimulate an increase in intracellular calcium concentration, which in turn activates mTORC1 through the class III PI3K Vps34 (Gulati et al, 2008).

Downstream of TOR

TORC1 regulates growth-related processes such as transcription, ribosome biogenesis, protein synthesis, nutrient transport and autophagy (Wullschleger et al, 2006). In mammals, the best-characterized substrates of mTORC1 are S6K and 4E-BP1, through which mTORC1 stimulates protein synthesis. mTORC1 activates S6K, which is a positive regulator of protein synthesis, and inhibits 4E-BP1, which is a negative regulator of protein synthesis. Upon phosphorylation by mTORC1, 4E-BP1 releases eIF4E. Once released from 4E-BP1, eIF4E interacts with the eIF4G subunit of the eIF4F complex, allowing initiation of translation. In mammals, 4E-BP1 participates mainly in the regulation of cell proliferation and metabolism (Dowling et al, 2010). In S. cerevisiae, the main substrate of TORC1 is the S6K orthologue Sch9 (Urban et al, 2007). Sch9 is required for the activation of ribosome biogenesis and translation initiation stimulated by TORC1. Furthermore, it participates in TORC1-dependent inhibition of G0 phase entry.

Regulation of TOR by Rheb

The small GTPase Rheb was first identified in 1994 in a screen for genes induced in neurons in response to synaptic activity (Yamagata et al, 1994), and was first described to interact with the Raf1 kinase (Yee & Worley, 1997). A later report showed that loss of Rhb1, the Rheb orthologue in S. pombe, causes a starvation-like growth arrest (Mach et al, 2000). In 2003, several independent groups working with mammalian cells in vitro and Drosophila in vivo demonstrated that Rheb is the target of the TSC1–TSC2 GAP and a TORC1 activator (Avruch et al, 2009).

Interestingly, the Rheb–mTOR interaction both in vivo and in vitro does not depend on GTP loading of Rheb. This is unusual for GTPases as GTP loading usually regulates effector binding. However, GTP loading of Rheb is crucial for the activation of mTOR kinase activity (Sancak et al, 2007). Conversely, mTOR becomes inactive after association with a nucleotide-deficient Rheb (Long et al, 2005a; Fig 2). Similar results were obtained in S. pombe, making use of mutations that hyperactivate Rheb by increasing its overall GTP : GDP binding ratio (Urano et al, 2005). In contrast to the situation in mammals, interaction of Rheb with SpTOR2 in fission yeast is detected only with a hyperactive Rheb mutant. This suggests that, in S. pombe, Rheb binds to SpTOR2 in a GTP-dependent manner.

Rheb activates TORC1

Rheb activates TORC1

Figure 2 Rheb activates TORC1 both directly and indirectly. GTP-bound Rheb interacts directly with TORC1 to activate TORC1 kinase. GTP-bound Rheb also activates RalA, which activates PLD to increase production of PA. PA in turn interacts with TORC1

In addition to the direct interaction between mTOR and Rheb, activation of PA production by Rheb is an additional mechanism by which Rheb might regulate mTORC1. Rheb binds to and activates PLD in a GTP-dependent manner (Sun et al, 2008). PLD produces PA, which binds directly to and upregulates mTORC1. This finding reveals cross-talk between the TSC–Rheb and the PA pathways in the regulation of mTORC1 signalling. A recent study by Yoon and colleagues further demonstrated the role of PLD in mTORC1 regulation (Yoon et al, 2011). They showed that amino acids activate PLD through translocation of PLD to the lysosomal compartment. This translocation is positively regulated by human Vps34 and is necessary for the activation of mTORC1 by amino acids. These authors propose the existence of a Vps34–PLD1 pathway that activates mTORC1 in parallel to the Rag pathway (Yoon et al, 2011).

Although Rheb is required for the activation of mTORC1 by amino acids, Rheb itself does not participate in amino acid sensing, and GTP-loading of Rheb is not affected by amino acid depletion (Long et al, 2005b). Furthermore, amino acid depletion inhibits mTORC1 even in TSC2−/− fibroblasts (Roccio et al, 2006). Nevertheless, interaction of mTORC1 with Rheb depends on amino acid availability (Long et al, 2005b). As discussed below, the current model proposes that amino acids mediate translocation of mTORC1 to the lysosomal surface where mTORC1 interacts with and is activated by GTP-loaded Rheb (Sancak et al, 2008).

Regulation of TOR by Rag

Rag GTPases have unique features among the Ras GTPase subfamily members: they form heterodimers and lack a membrane-targeting sequence (Nakashima et al, 1999Sekiguchi et al, 2001). Gtr1 in S. cerevisiaewas the first member of this GTPase subfamily to be identified (Bun-Ya et al, 1992). The mammalian RagA and RagB GTPases were later described as Gtr1 orthologues (Hirose et al, 1998). Gtr2 in yeast (Nakashima et al, 1999) and its mammalian orthologues RagC and RagD (Sekiguchi et al, 2001) were subsequently discovered due to their ability to form heterodimers with Gtr1 in yeast and RagA and RagB in mammals, respectively. The crystal structure of the Gtr1–Gtr2 complex has been determined recently (Gong et al, 2011). Gtr1 and Gtr2 have similar structures, organized in two domains: an amino-terminal GTPase domain (designated as the G domain) and a carboxy-terminal domain. The Gtr1–Gtr2 heterodimer presents a pseudo-twofold symmetry resembling a horseshoe. The crystal structure reveals that Gtr1–Gtr2 dimerization results from extensive contacts between the C-terminal domains of both proteins, while the G domains do not contact each other (Gong et al, 2011).

Rag proteins mediate the activation of TORC1 in response to amino acids.

Rag proteins mediate the activation of TORC1 in response to amino acids.

Figure 3 Rag proteins mediate the activation of TORC1 in response to amino acids. The RagA/B–RagC/D heterodimer is anchored to the MP1–p14–p18 complex on the surface of the lysosome.

Overexpressed Rheb is mislocalized throughout the cell, and therefore interaction of mTORC1 with Rheb does not require amino-acid-induced translocation of mTORC1 to the lysosome. The model is further supported by observations in Drosophila showing that expression of a constitutively active mutant of RagA significantly increases the size of individual cells, whereas expression of a dominant negative mutant of RagA reduces cell size (Kim et al, 2008). Moreover, Rag plays a role in TORC1-mediated inhibition of autophagy both in Drosophila (Kim et al, 2008) and in human cells (Narita et al, 2011).

mTOR and small GTPases are therapeutic targets in the treatment of cancer (Berndt et al, 2011Dazert & Hall, 2011). Aberrant activation of GTPases, including Ras, Rho, Rab or Ran GTPases, promotes cell transformation and cancer (Agola et al, 2011Ly et al, 2010Pylayeva-Gupta et al, 2011), in some cases by acting in the mTOR pathway. Targeting GTPases by using farnesyltransferase inhibitors or geranylgeranyltransferase inhibitors affects signal transduction pathways, cell cycle progression, proliferation and cell survival. Both types of inhibitor are currently under investigation for cancer therapy, although only a small subset of patients responds to these inhibitors (Berndt et al, 2011). A better understanding of the relationship between GTPases and mTOR is essential for the design of combined therapies.

From a mechanistic point of view, research on TOR in different systems is continually adding new insight on the role of TOR in cell biology. However, what is lacking is an integration of the various proposed regulators of TOR, in particular small GTPases (see Sidebar A).

Sidebar A | In need of answers

  1. How are amino acids sensed by the cell?
  2. What is the mechanism by which amino acids regulate the GTP-loading of Rag proteins? What are the GEF and GAP for the Rag proteins?
  3. Is there a GEF that regulates the GTP-loading of Rheb?
  4. What is the molecular mechanism by which Rheb activates TORC1?
  5. How is the dual effect of Rac1 being both upstream and downstream from TOR regulated?
  6. How are the diverse GTPases that impinge on TOR integrated?

7.8.5 PI3K.Akt signaling in osteosarcoma

Zhang J1Yu XH2Yan YG1Wang C1Wang WJ3.
Clin Chim Acta. 2015 Apr 15; 444:182-192.


  • Activation of the PI3K/Akt signaling regulates various cellular functions.
  • The PI3K/Akt signaling may play a key role in the progression of osteosarcoma.
  • Targeting the PI3K/Akt signaling has therapeutic potential for osteosarcoma.

Osteosarcoma (OS) is the most common nonhematologic bone malignancy in children and adolescents. Despite the advances of adjuvant chemotherapy and significant improvement of survival, the prognosis remains generally poor. As such, the search for more effective anti-OS agents is urgent. The phosphatidylinositol 3-kinase (PI3K)/Akt pathway is thought to be one of the most important oncogenic pathways in human cancer. An increasing body of evidence has shown that this pathway is frequently hyperactivated in OS and contributes to disease initiation and development, including tumorigenesis, proliferation, invasion, cell cycle progression, inhibition of apoptosis, angiogenesis, metastasis and chemoresistance. Inhibition of this pathway through small molecule compounds represents an attractive potential therapeutic approach for OS. The aim of this review is to summarize the roles of the PI3K/Akt pathway in the development and progression of OS, and to highlight the therapeutic potential of targeting this signaling pathway. Knowledge obtained from the application of these compounds will help in further understanding the pathogenesis of OS and designing subsequent treatment strategies.

PK.Akt signaling

PK.Akt signaling

PI3K/Akt signaling

PI3K.Akt signaling pathway

PI3K.Akt signaling pathway

PI3K/Akt signaling pathway

PK.Akt therapeutic target

PK.Akt therapeutic target

PK/Akt therapeutic target

7.8.6 The mTORC1-S6K1 Pathway Regulates Glutamine Metabolism through the eIF4B-Dependent Control of c-Myc Translation

Csibi A1Lee G1Yoon SO1Tong H2,…, Fendt SM4Roberts TM2Blenis J5.
Curr Biol. 2014 Oct 6; 24(19):2274-80.

Growth-promoting signaling molecules, including the mammalian target of rapamycin complex 1 (mTORC1), drive the metabolic reprogramming of cancer cells required to support their biosynthetic needs for rapid growth and proliferation. Glutamine is catabolyzed to α-ketoglutarate (αKG), a tricarboxylic acid (TCA) cycle intermediate, through two deamination reactions, the first requiring glutaminase (GLS) to generate glutamate and the second occurring via glutamate dehydrogenase (GDH) or transaminases. Activation of the mTORC1 pathway has been shown previously to promote the anaplerotic entry of glutamine to the TCA cycle via GDH. Moreover, mTORC1 activation also stimulates the uptake of glutamine, but the mechanism is unknown. It is generally thought that rates of glutamine utilization are limited by mitochondrial uptake via GLS, suggesting that, in addition to GDH, mTORC1 could regulate GLS. Here we demonstrate that mTORC1 positively regulates GLS and glutamine flux through this enzyme. We show that mTORC1 controls GLS levels through the S6K1-dependent regulation of c-Myc (Myc). Molecularly, S6K1 enhances Myc translation efficiency by modulating the phosphorylation of eukaryotic initiation factor eIF4B, which is critical to unwind its structured 5′ untranslated region (5’UTR). Finally, our data show that the pharmacological inhibition of GLS is a promising target in pancreatic cancers expressing low levels of PTEN.


  • The mTORC1 pathway positively regulates GLS and glutamine flux
  • mTORC1 controls the translation efficiency of Myc mRNA
  • S6K1 regulates Myc translation through eIF4B phosphorylation
  • Inhibition of GLS decreases the growth of pancreatic cancer cells

Figure 1. The mTORC1 Pathway Regulates GLS1 (A–C and E) GLS protein levels in whole cell lysates from Tsc2 WT and Tsc22/2 MEFs treated with rapamycin (Rapa) for 8 hr (A); HEK293T cells stably expressing Rheb WT, the mutant S16H Rheb, or EV and treated with rapamycin for 24 hr (B); Tsc22/2 MEFs treated with rapamycin at the indicated time points (C); and Tsc2 WT and Tsc22/2 MEFs treated with the indicated compounds for 8 hr (E). The concentrations of the compounds were as follows: rapamycin, 20 ng/ml; LY294002 (LY), 20 mM; and BEZ235, 10 mM. (D) Time course of glutamine consumption in Tsc22/2 MEFs incubated with or without 20ng/ml rapamycin for 24 hr. Each time data point is an average of triplicate experiments. (F) Intracellular glutamine levels in Tsc22/2 MEFs treated with rapamycin for 24 hr. (G) Glutamineflux inTsc22/2 MEFs expressing an EV or re-expressingTSC2 treated with theindicated compounds for 24hr.The concentrations of the compounds were as follows: rapamycin 20 ng/ml; LY294002, 20 mM; BEZ235, 10 mM; BPTES, 10 mM; and 6-diazo-5-oxo-l-norleucine, 1mM. The mean is shown. Error bars represent the SEM from at least three biological replicates. Numbers below the immunoblot image represent quantification normalized to the loading control. See also Figure S1.

Figure2. The mTORC1 Pathway Regulates GLS1 via Myc GLS and Myc protein levels in whole cell lysates from BxPC3 cells transfected with a nontargeting control (NTC) siRNA or four independent siRNAs against Myc for 72 hr (A), Tsc2 WT and Tsc22/2 MEFs treated with rapamycin (20 ng/ml) for 8 hr (B), and Tsc22/2 MEFs stably expressing Myc or EV and treated with rapamycin (20 ng/ml) for 24 hr (C).

Figure 3. The mTORC1 Substrate S6K1 Controls GLS through Myc mRNA Translation (A) Normalized luciferase light units of Tsc22/2 MEFs stably expressing a Myc-responsive firefly luciferase construct (Myc-Luc) or vector control (pCignal Lenti-TRE Reporter). Myc transcriptional activity was measured after treatment with rapamycin (20 ng/ml) or PF4708671 (10 mM) for 8 hr. (B) GLS and Myc protein levels in whole cell lysates from HEK293T cells expressing HA-S6K1-CA (F5A-R3A-T389E) or EV treated with rapamycin (20 ng/ml) for 24 hr. HA, hemagglutinin. (CandD) Intracellular glutamine levels of Tsc22/2 MEFs stably expressing S6K-CA(F5A/R5A/T389E, mutating either the three arginines or all residues within the RSPRR motif to alanines shows the same effect; [10]) or empty vector and treated with rapamycin (20 ng/ml) or DMSO for 48 hr (C) or transfected with NTC siRNA or siRNA against both S6K1/2 (D). 24 hr posttransfection, cells transfected with NTC siRNA were treated with PF4708671 (10 mM) or DMSO for 48 hr. (E) Glutamine consumption of Tsc22/2 MEFs transfected with NTC siRNA or siRNA against both S6K1/2. 72 hr posttransfection, media were collected, and levels of glutamine in the media were determined. (F) Normalized luciferase light units of Tsc2WTMEFs transfected with thepDL-N reporter construct containing the 50 UTR of Myc under the control of Renilla luciferase. Firefly luciferase was used as an internal control. 48hr posttransfection, cells were treated with rapamycin (20ng/ml) or PF4708671 (10mM) for 8h. (G) Relative levels of Myc, Gls, and Actin mRNA in each polysomal gradient fraction. mRNA levels were measured by quantitative PCR and normalized to the 5S rRNA level. HEK293T cells were treated with rapamycin (20 ng/ml) for 24 hr, and polysomes were fractionated on sucrose density gradients. The values are averaged from two independent experiments performed in duplicate, and the error bars denote SEM (n = 4). (Hand I) GLS and Myc protein levels in whole cell lysates from Tsc22/2 MEFs transfected with NTC siRNA or two independent siRNAs against eIF4B for 72hr (H) and Tsc22/2 MEFs stably expressing eIF4B WT, mutant S422D, or EV) and treated with rapamycin for 24 hr (I). The mean is shown. Error bars represent the SEM from at least three biological replicates. The asterisk denotes a nonspecific band. The numbers below the immunoblot image represent quantification normalized to the loading control. See also Figures S2 and S3.

Figure 4. Inhibition of GLS Reduces the Growth of Pancreatic Cancer Cells (A) GLS and Myc protein levels in whole cell lysates from BxPC3, MIAPaCa-2, or AsPC-1 cells treated with rapamycin (20 ng/ml) or BEZ235 (1 mM) for 24 hr. (B) Glutamine consumption of BxPC3 or AsPC-1 cells 48 hr after plating. (Cand D) Soft agar assays with BxPC3 or AsPC-1 cells treated with BPTES (10 mM), the combination of BPTES (10 mM) + OAA (2 mM) (C) and BxPC3 or AsPC-1 cells treated with BPTES, and the combination of BPTES (10 mM) + NAC (10 mM) (D). NS, not significant. The mean is shown. Error bars represent the SEM from at least three biological replicates.

7.8.7 Localization of mouse mitochondrial SIRT proteins

Nakamura Y1Ogura MTanaka DInagaki N.
Biochem Biophys Res Commun. 2008 Feb 1; 366(1):174-9

Yeast silent information regulator 2 (SIR2) is involved in extension of yeast longevity by calorie restriction, and SIRT3, SIRT4, and SIRT5 are mammalian homologs of SIR2 localized in mitochondria. We have investigated the localization of these three SIRT proteins of mouse. SIRT3, SIRT4, and SIRT5 proteins were localized in different compartments of the mitochondria. When SIRT3 and SIRT5 were co-expressed in the cell, localization of SIRT3 protein changed from mitochondria to nucleus. These results suggest that the SIRT3, SIRT4, and SIRT5 proteins exert distinct functions in mitochondria. In addition, the SIRT3 protein might function in nucleus

Fig. 1. Localization of SIRT3, SIRT4, and SIRT5 in mitochondria. (A) Confocal microscopy. SIRT3-myc (upper panels), SIRT4-myc (middle panels), and SIRT5-FLAG (lower panels) were expressed in COS7 cells and immunostained with anti-myc antibody or anti-FLAG antibody. Mitochondria and nuclei were stained by MitoTracker Red and DAPI, respectively, and fluorescent images were obtained using a confocal microscope. (B) Fractionation of post-nuclear supernatant. SIRT3-myc, SIRT4-myc, and SIRT5-FLAG proteins each was expressed in COS7 cells, and the obtained PNS was fractionated into mitochondria-enriched precipitate (P1), microsome-enriched precipitate (P2), and supernatant (S) fractions. The three fractions were separated by SDS–PAGE and then analyzed by Western blotting using anti-myc antibody for SIRT3-myc and SIRT4-myc or anti-FLAG antibody for SIRT5-FLAG. Hsp60, calnexin, and GAPDH were used as endogenous markers for mitochondria, microsome, and cytosol, respectively. (C) Alkaline treatment of mitochondria. Mitochondria prepared from the COS7 cells expressing each of the SIRT3-myc, SIRT4-myc, and SIRT5-FLAG proteins were treated with Na2CO3. The reaction mixture was centrifuged to separate the precipitate and supernatant fractions, containing membrane-integrated proteins and soluble proteins, respectively. The two fractions were analyzed by Western blotting. Cytochrome c (cytc) and hsp60 were used as endogenous protein markers for mitochondrial soluble protein. (D) Submitochondrial fractionation. The mitochondria from COS7 cells expressing one of three SIRT proteins were treated with either H2O (hypotonic) or TX-100, and then treated with trypsin. The reaction mixtures were analyzed by Western blotting. Cytochrome c and hsp60 were used as endogenous markers for mitochondrial intermembrane space protein and matrix protein, respectively.

Fig. 2. Localization of SIRT3 when co-expressed with SIRT5. (A) Confocal microscopic analysis of COS7 cells expressing two of the three mitochondrial SIRT proteins. SIRT3-myc and SIRT5-FLAG (upper panels), SIRT3-myc and SIRT4-FLAG (middle panels), and SIRT4-myc and SIRT5-FLAG (lower panels) were co-expressed in COS7 cells, and immunostained using antibodies against myc tag and FLAG tag. Nuclei were stained by DAPI. (B) Subcellular fractionation of PNS. PNS of COS7 cells co-expressing SIRT3-myc and SIRT5-FLAG was fractionated into mitochondria-enriched precipitate (P1), microsome-enriched precipitate (P2), and supernatant (S) fractions, and these fractions along with whole cell lysate were analyzed by Western blotting. (C) Subcellular fractionation using digitonin. COS7 cells expressing either SIRT3-myc (left) or SIRT5-FLAG (middle) or both (right) were solubilized by digitonin, and the obtained lysate was centrifuged and fractionated into nuclear-enriched insoluble (INS), and soluble (SOL) fractions. Hsp60 and laminA/C were used as endogenous markers for mitochondria protein and nucleus protein, respectively.

Because the segment containing amino acid residues 66– 88 potentially forms a basic amphiphilic a-helical structure, it could serve as a MTS. To examine the role of this segment, SIRT3 mutant SIRT3mt, in which the four amino acid residues 72–75 were replaced by four alanine residues, was constructed (Fig. 3A). When SIRT3mt alone was expressed in COS7 cells, SIRT3mt protein was not detected in mitochondria but was widely distributed in the cell in confocal microscopic analysis (Fig. 3B, upper panels). In addition, when SIRT3mt and SIRT5 were co-expressed, the distribution of SIRT3mt protein was not changed compared to that expressed alone (Fig. 3B, lower panels). In fractionation of PNS, SIRT3mt protein was fractionated into S fraction both when SIRT3mt was expressed alone and when SIRT3mt and SIRT5 were co-expressed. SIRT5 protein was localized in mitochondria when SIRT3mt and SIRT5 were co-expressed (Fig. 3C). These results indicate that the MTS is necessary not only for targeting SIRT3 to mitochondria in the absence of SIRT5 but also for targeting SIRT3 to nucleus in the presence of SIRT5.

Fig. 3. Effect of disruption of putative mitochondrial targeting signal of SIRT3. (A) Alanine replacement of putative MTS of SIRT3. Four residues of the putative MTS of SIRT3 (amino acid residues 72–75) were replaced with four alanine residues. In the SIRT3mt sequence, amino acid residues identical with wild-type SIRT3 protein are indicated with dots. (B) Confocal microscopy. Immunofluorescent images of COS7 cells expressing SIRT3mt-myc alone (upper panels) or both SIRT3mt-myc and SIRT5-FLAG (lower panels) are shown. Mitochondria and nuclei were stained by MitoTracker Red and DAPI, respectively. (C) Subcellular fractionation of PNS. PNSs of COS7 cells expressing SIRT3mt-myc alone (an upper panel) or co-expressing SIRT3mt-myc and SIRT5-FLAG (middle and lower panels) were centrifuged and fractionated into mitochondria-enriched precipitate (P1), microsome-enriched precipitate (P2), and supernatant (S) fractions. The fractions were analyzed by Western blotting.

Fig. 4. Effect of disruption of putative nuclear localization signal of SIRT3. (A) Comparison of the amino acid sequences of putative NLS of SIRT3, SIRT3nu, and SV40 large T antigen. Three basic amino acid residues of the putative NLS of SIRT3 (amino acid residues 214–216) were replaced with three alanine residues. In the SIRT3nu sequence, amino acid residues identical with wild-type SIRT3 protein are indicated with dots. The classical NLS of SV40 large T antigen also is shown (SV40). (B) Confocal microscopy. Immunofluorescent images of COS7 cells expressing SIRT3nu-myc alone (upper panels) or both SIRT3nu-myc and SIRT5-FLAG (lower panels) are shown. Mitochondria and nuclei were stained by MitoTracker Red and DAPI, respectively. (C) Subcellular fractionation of PNS. PNSs of the COS7 cells expressing SIRT3nu-myc alone (an upper panel) or co-expressing SIRT3numyc and SIRT5-FLAG (middle and lower panels) were fractionated into mitochondria-enriched precipitate (P1), microsome-enriched precipitate (P2), and supernatant (S) fractions. The fractions were analyzed by Western blotting.

The sequence containing amino acid sequence 213-219 of the SIRT3 closely resembles the putative protein classical NLS of the SV40 T antigen (Fig. 4A). To examine whether this sequence functions as a NLS, the mutant SIRT3 protein SIRT3nu, in which the three basic amino acid residues (214–216) in the putative NLS of SIRT3 were replaced by three alanine residues (Fig. 4A), was constructed. When SIRT3nu alone was expressed in COS7 cells, it was localized in mitochondria (Fig. 4B, upper panels). In the cells co-expressing SIRT3nu and SIRT5, a shift of SIRT3nu protein to the nucleus was not observed, and SIRT3nu protein and a part of SIRT5 protein were scattered widely in the cell in confocal microscopic analysis (Fig. 4B, lower panels). In fractionation of PNS, all of the SIRT3nu protein and nearly half of the SIRT5 protein were shifted from P1 fraction to S fraction by co-expression (Figs. 1B and 4C). These results suggest that the segment containing amino acid residues 213–219 of SIRT3 plays an important role in the localization shift of SIRT3 protein to nucleus when co-expressed with SIRT5. Furthermore, SIRT5 may well hamper SIRT3nu localization in mitochondria through interaction with SIRT3nu. However, further study is required to elucidate the mechanism of the localization shift of SIRT3 protein. Interestingly, recent study has reported that human prohibitin 2 (PHB2), known as a repressor of estrogen receptor (ER) activity, is localized in the mitochondrial inner membrane, and translocates to the nucleus in the presence of ER and estradiol [18]. Although the mechanism of regulation of the expression level of SIRT5 remains unknown, SIRT3 might play a role in communication between nucleus and mitochondria in a SIRT5-dependent manner. The function of mitochondrial SIRT proteins is still not well known. In the present study, we determined the exact localization of mouse SIRT3, SIRT4, and SIRT5 proteins in mitochondria. In addition, we demonstrated that SIRT3 can be present in nucleus in the presence of SIRT5. It has been reported that SIRT3 deacetylates proteins that are not localized in mitochondria in vitro such as histone-4 peptide and tubulin [14]. Thus, if SIRT3 is present in nucleus in vivo, SIRT3 protein might well deacetylate nuclear proteins. These results provide useful information for the investigation of the function of these proteins.


[1] J.C. Tanny, G.J. Dowd, J. Huang, H. Hilz, D. Moazed, An enzymatic activity in the yeast Sir2 protein that is essential for gene silencing, Cell 99 (1999) 735–745.
[2] S. Imai, C.M. Armstrong, M. Kaeberlein, L. Guarente, Transcriptional silencing and longevity protein Sir2 is an NAD-dependent histone deacetylase, Nature 403 (2000) 795–800.
[3] M. Gotta, S. Strahl-Bolsinger, H. Renauld, T. Laroche, B.K. Kennedy, M. Grunstein, S.M. Gasser, Localization of Sir2p: the nucleolus as a compartment for silent information regulators, EMBO J. 16 (1997) 3243–3255.
[4] I. Muller, M. Zimmermann, D. Becker, M. Flomer, Calendar life span versus budding life span of Saccharomyces cerevisiae, Mech. Aging Dev. 12 (1980) 47–52.
[5] S.J. Lin, M. Kaeberlein, A.A. Andalis, L.A. Sturtz, P.A. Defossez, V.C. Culotta, G.R. Fink, L. Guarente, Calorie restriction extends Saccharomyces cerevisiae lifespan by increasing respiration, Nature 418 (2002) 344–348.
[6] S.J. Lin, P.A. Defossez, L. Guarente, Requirement of NAD and SIR2 for life-span extension by calorie restriction in Saccharomyces cerevisiae, Science 289 (2000) 2126–2128.

7.8.8 SIRT4 Has Tumor-Suppressive Activity and Regulates the Cellular Metabolic Response to DNA Damage by Inhibiting Mitochondrial Glutamine Metabolism

Jeong SM1Xiao CFinley LWLahusen TSouza ALPierce KLi YH, et al.
Cancer Cell. 2013 Apr 15; 23(4):450-63.

DNA damage elicits a cellular signaling response that initiates cell cycle arrest and DNA repair. Here we find that DNA damage triggers a critical block in glutamine metabolism, which is required for proper DNA damage responses. This block requires the mitochondrial SIRT4, which is induced by numerous genotoxic agents and represses the metabolism of glutamine into TCA cycle. SIRT4 loss leads to both increased glutamine-dependent proliferation and stress-induced genomic instability, resulting in tumorigenic phenotypes. Moreover, SIRT4 knockout mice spontaneously develop lung tumors. Our data uncover SIRT4 as an important component of the DNA damage response pathway that orchestrates a metabolic block in glutamine metabolism, cell cycle arrest and tumor suppression.

DNA damage initiates a tightly coordinated signaling response to maintain genomic integrity by promoting cell cycle arrest and DNA repair. Upon DNA damage, ataxia telangiectasia mutated (ATM) and ataxia telangiectasia and RAD3-related protein (ATR) are activated and induce phosphorylation of Chk1, Chk2 and γ-H2AX to trigger cell cycle arrest and to initiate assembly of DNA damage repair machinery (Abraham, 2001Ciccia and Elledge, 2010Su, 2006). Cell cycle arrest is a critical outcome of the DNA damage response (DDR) and defects in the DDR often lead to increased incorporation of mutations into newly synthesized DNA, the accumulation of chromosomal instability and tumor development (Abbas and Dutta, 2009Deng, 2006Negrini et al., 2010).

The cellular metabolic response to DNA damage is not well elucidated. Recently, it has been shown that DNA damage causes cells to upregulate the pentose phosphate pathway (PPP) to generate nucleotide precursors needed for DNA repair (Cosentino et al., 2011). Intriguingly, a related metabolic switch to increase anabolic glucose metabolism has been observed for tumor cells and is an important component of rapid generation of biomass for cell growth and proliferation (Jones and Thompson, 2009Koppenol et al., 2011). Hence, cells exposed to genotoxic stress face a metabolic challenge; they must be able to upregulate nucleotide biosynthesis to facilitate DNA repair, while at the same time limiting proliferation and inducing cell cycle arrest to limit the accumulation of damaged DNA. The molecular events that regulate this specific metabolic program in response to DNA damage are still unclear.

Sirtuins are a highly conserved family of NAD+-dependent deacetylases, deacylases, and ADP-ribosyltransferases that play various roles in metabolism, stress response and longevity (Finkel et al., 2009;Haigis and Guarente, 2006). In this study, we studied the role of SIRT4, a mitochondria-localized sirtuin, in cellular metabolic response to DNA damage and tumorigenesis.

DNA damage represses glutamine metabolism

To investigate how cells might balance needs for continued nucleotide synthesis, while also preparing for cell cycle arrest, we assessed the metabolic response to DNA damage by monitoring changes in the cellular consumption of two important fuels, glucose and glutamine, after DNA-damage. Strikingly, treatment of primary mouse embryonic fibroblasts (MEFs) with camptothecin (CPT), a topoisomerase 1 inhibitor that causes double-stranded DNA breaks (DSBs), resulted in a pronounced reduction in glutamine consumption (Figure 1A). Glutamine metabolism in mammalian cells is complex and contributes to a number of metabolic pathways. Glutamine is the primary nitrogen donor for protein and nucleotide synthesis, which are essential for cell proliferation (Wise and Thompson, 2010). Additionally, glutamine provides mitochondrial anaplerosis. Glutamine can be metabolized via glutaminase (GLS) to glutamate and NH4+, and further converted to the tricarboxylic acid (TCA) cycle intermediate α-ketoglutarate via glutamate dehydrogenase (GDH) or aminotransferases. This metabolism of glutamine provides an important entry point of carbon to fuel the TCA cycle (Jones and Thompson, 2009), and accounts for the majority of ammonia production in cells (Yang et al., 2009). CPT-induced reduction of glutamine consumption was accompanied by a reduction in ammonia secretion from cells (Figure 1B). Notably, under these conditions, we observed no obvious decrease in glucose uptake and lactate production (Figures 1C and 1D), consistent with previous studies showing that intact glucose utilization through the PPP is important for a normal DNA damage response (Cosentino et al., 2011). Preservation of glucose uptake also suggests that repression of glutamine consumption may be a specific metabolic response to genotoxic stress and not reflective of a non-specific metabolic crisis.

Figure 1 Glutamine metabolism is repressed by genotoxic stress

To examine the metabolic response to other forms of genotoxic stress, we monitored the metabolic response to ultra-violet (UV) exposure in primary MEFs. Similar to CPT treatment, UV exposure reduced glutamine uptake, without significant changes in glucose consumption (Figures 1E and 1F). Similarly two human cell lines, HepG2 and HEK293T, also demonstrated marked reductions in glutamine uptake in response to DNA damaging agents without comparable changes in glucose uptake (Figures 1G and 1HFigures S1A and S1B). Taken together, these results suggest that a variety of primary and tumor cell lines (from mouse or human) respond to genotoxic stress by down-regulating glutamine metabolism.

To examine in more detail the changes in cellular glutamine metabolism after genotoxic stress, we performed a global metabolomic analysis with transformed MEFs before and after DNA damage. As previously reported, we observed that PPP intermediates were increased in response to DNA damage (Figures 1I and 1J). Remarkably, we observed a decrease in measured TCA cycle intermediates after UV exposure (Figures 1I and 1K). Moreover, we found that HepG2 cells showed a similar metabolomic shift in response to DNA damage (Figure S1D). We did not observe a clear, coordinated repression of nucleotides or glutamine-derived amino acids after exposure to DNA damage (Figure S1C).

To determine whether reduction in TCA cycle metabolites was the consequence of reduced glutamine metabolism, we performed a time-course tracer study to monitor the incorporation of [U-13C5]glutamine into TCA cycle intermediates at 0, 2 and 4 hr after UV treatment. We observed that after UV exposure, cells reduced contribution of glutamine to TCA cycle intermediates in a time-dependent manner (Figure 1L). Moreover, the vast majority of the labeled fumarate and malate contained four carbon atoms derived from [U-13 C5]glutamine (Figure S1F, M+3 versus M+4), indicating that most glutamine was used in the non-reductive direction towards succinate, fumarate and malate production. We were able to observe little contribution of glutamine flux into nucleotides or glutathione in control or UV-treated cells at these time points (data not shown), suggesting that the mitochondrial metabolism of glutamine accounts for the majority of glutamine consumption in these cells. Taken together, the metabolic flux analysis demonstrates that DNA damage results in a reduction of mitochondrial glutamine anaplerosis, thus limiting the critical refueling of carbons into the TCA cycle.

To assess the functional relevance of decreased glutamine metabolism after DNA damage, we deprived cells of glucose, thereby shifting cellular dependence to glutamine to maintain viability (Choo et al., 2011Dang, 2010). If DNA damage represses glutamine usage, we reasoned that cells would be more sensitive to glucose deprivation. Indeed, following 72 hr of glucose deprivation, cell death in primary MEFs was significantly elevated at 10 hr after UV exposure (Figure S1E). However, cells cultured with glucose remained viable in these conditions. Thus, these data demonstrate that genotoxic stress limits glutamine entry into the central mitochondrial metabolism of the TCA cycle.

SIRT4 is induced in response to genotoxic stress

Because sirtuins regulate both cellular metabolism and stress responses (Finkel et al., 2009Schwer and Verdin, 2008), we examined whether sirtuins were involved in the metabolic adaptation to DNA damage. We first examined the expression of sirtuins in the response to DNA damage. Specifically, we probed SIRT1, which is involved in stress responses (Haigis and Guarente, 2006), as well as mitochondrial sirtuins (SIRT3–5), which have been shown to regulate amino acid metabolism (Haigis et al., 2006Hallows et al., 2011Nakagawa et al., 2009). Remarkably, SIRT4 mRNA levels were induced by nearly 15-fold at 15 hr after CPT treatment and 5-fold after etoposide (ETS), a topoisomerase 2 inhibitor, in HEK293T cells (Figure 2A). Interestingly, the induction of SIRT4 was significantly higher than the induction of SIRT1 and mitochondrial SIRT3 (~2-fold), sirtuins known to be induced by DNA damage and regulate cellular responses to DNA damage (Sundaresan et al., 2008Vaziri et al., 2001Wang et al., 2006). Moreover, overall mitochondrial mass was increased by only 10% in comparison with control cells (Figure S2A), indicating that the induction of SIRT4 is not an indirect consequence of mitochondrial biogenesis. These data hint that SIRT4 may have an important, previously undetermined role in the DDR.

Figure 2 SIRT4 is induced by DNA damage stimuli

To test the induction of SIRT4 in the general genotoxic stress response, we treated cells with other types of DNA damage, including UV and gamma-irradiation (IR). SIRT4 mRNA levels were also increased by these genotoxic agents (Figures S2B and S2C) and low doses of CPT and UV treatment also induced SIRT4expression (Figures S2D and S2E). We observed similar results with MEFs (Figures 2B and 2DFigure S2F) and HepG2 cells (Figure S2G). DNA damaging agents elevated SIRT4 in p53-inactive HEK293T cells (Figures 2A and 2C) and in p53-null PC3 human prostate cancer cells (Figure S2H), suggesting that SIRT4can be induced in a p53-independent manner.

To examine whether the induction of SIRT4 occurred as a result of cell cycle arrest, we measured SIRT4levels after the treatment of nocodazole, which inhibits microtubule polymerization to block mitosis. While treatment with nocodazole completely inhibited cell proliferation (data not shown), SIRT4 expression was not elevated (Figure S2I). In addition, we analyzed SIRT4 expression in distinct stages of the cell cycle in HepG2 cells synchronized with thymidine block (Figure S2J, Left). SIRT4 mRNA levels were measured at different times after release and were not elevated during G1 or G2/M phases (Figure S2J, Right), suggesting thatSIRT4 is not induced as a general consequence of cell cycle arrest. Next, we re-examined the localization of SIRT4 after DNA damage. SIRT4 localizes to the mitochondria of human and mouse cells under basal, unstressed conditions (Ahuja et al., 2007Haigis et al., 2006). Following CPT treatment, SIRT4 colocalized with MitoTracker, a mitochondrial-selective marker, indicating that SIRT4 retains its mitochondrial localization after exposure to DNA damage (Figure S2K). Taken together, our findings demonstrate that SIRT4 is induced by multiple forms of DNA damage in numerous cell types, perhaps to coordinate the mitochondrial response to genotoxic stress.

SIRT4 represses glutamine anaplerosis

We observed that glutamine anaplerosis is repressed by genotoxic stress (Figure 1) and SIRT4 is induced by DNA damage (Figure 2). Additionally, previous studies reported that SIRT4 represses glutamine anaplerosis (Haigis et al., 2006). We next tested whether SIRT4 directly regulates cellular glutamine metabolism and contribution of glutamine to the TCA cycle. Like DNA damage, SIRT4 overexpression (SIRT4-OE) in HepG2, HeLa or HEK293T cells resulted in the repression of glutamine consumption (Figure 3AFigures S3A–C). Conversely, SIRT4 knockout (KO) MEFs consumed more glutamine than did wild-type (WT) cells (Figure 3B).

Figure 3 SIRT4 represses mitochondrial glutamine metabolism in response to DNA damage

Mitochondrial glutamine catabolism refuels the TCA cycle and is essential for viability in the absence of glucose (Choo et al., 2011Yang et al., 2009). Thus, we examined the effect of SIRT4 on cell survival during glucose deprivation. Overexpression of SIRT4 in HEK293T or HeLa cells increased cell death in glucose-free media compared to control cells (Figure 3CFigure S3D). Importantly, this cell death was completely rescued by the addition of pyruvate or cell permeable dimethyl α-ketoglutarate (DM-KG), demonstrating that SIRT4 overexpression reduced the ability of cells to utilize glutamine for mitochondrial energy production. Moreover, cell death was equally maximized in the absence of glucose and presence of the mitochondrial ATPase inhibitor oligomycin (Figure 3C). These findings are in line with the model that SIRT4 induction with DNA damage limits glutamine metabolism and utilization by the TCA cycle

We next utilized a metabolomic approach to interrogate glutamine usage in the absence of SIRT4. SIRT4 KO MEFs demonstrated elevated levels of TCA cycle intermediates (Figure 3J, WT versus KO), whereas intermediates of glycolysis were comparable with WT cells (data not shown). Nucleotides and other metabolites downstream of glutamine metabolism were not coordinately regulated by SIRT4 loss (Figure S3E and data not shown). Next, we analyzed glutamine flux in WT and SIRT4 KO MEFs in medium containing [U-13C5]glutamine for 2 or 4 hours and measured isotopic enrichment of TCA cycle intermediates. Loss of SIRT4 promoted a higher rate of incorporation of 13C-labeled metabolites derived from [U-13C5]glutamine in all TCA cycle intermediates measured (Figure 3D). These data provide direct evidence that SIRT4 loss drives increased entry of glutamine-derived carbon into the TCA cycle.

Next, we examined the mechanisms involved in this repression of glutamine anaplerosis. GLS is the first required enzyme for mitochondrial glutamine metabolism (Curthoys and Watford, 1995) and its inhibition limits glutamine flux into the TCA cycle (Wang et al., 2010; Le et al., 2012; Yuneva et al., 2012). Treatment with bis-2-(5-phenylacetoamido-1,2,4-thiadiazol-2-yl)ethyl sulfide (BPTES) (Robinson et al., 2007), an inhibitor of GLS1, repressed glutamine uptake and completely rescued the increased glutamine consumption of SIRT4 KO cells (Figure 3E). Moreover, SIRT4 overexpression no longer inhibited glutamine uptake when GLS1 was reduced by using short hairpin RNAs (shRNAs) (Figures 3F and 3G), demonstrating that SIRT4 regulates mitochondrial glutamine metabolism. SIRT4 is a negative regulator of GDH activity (Haigis et al., 2006) and SIRT4 KO MEFs exhibited increased GDH activity in comparison with WT MEFs (Figure S3F). To test whether SIRT4 regulates mitochondrial glutamine metabolism via inhibiting GDH activity, we measured glutamine uptake in WT and SIRT4 KO cells in the presence of EGCG, a GDH inhibitor (Choo et al., 2011Li et al., 2006). The treatment of EGCG partially rescued the increased glutamine uptake of KO cells (Figure S3G), suggesting that GDH contributes to the role of SIRT4 in glutamine metabolism.

SIRT4 represses mitochondrial glutamine metabolism after DNA damage

SIRT4 regulates cell cycle progression and genomic fidelity in response to DNA damage

Figure 4 SIRT4 is involved in cellular DNA damage responses

SIRT4 represses tumor proliferation

Figure 5 SIRT4 has tumor suppressive function

(A and B) Growth curves of WT and SIRT4 KO MEFs (n = 3) cultured in standard media (A) or media supplemented with BPTES (10 μM) (B). Data are means ±SD.

(C and D) Growth curves of Vector and SIRT4-OE HeLa cells (n = 3) cultured in standard media (C) or media supplemented with BPTES (10 μM) (D). Data are means ±SD.

(E) Focus formation assays with transformed WT and SIRT4 KO MEFs (left). Cells were cultured with normal medium or medium without glucose or glutamine for 10 days and stained with crystal violet. The number of colonies was counted (right) (n =3 samples of each condition). n.d., not determined.

(F) Focus formation assays with transformed KO MEFs reconstituted with SIRT4 or a catalytic mutant of SIRT4 (n = 3). Cells were cultured for 8 days and stained with crystal violet.

(G) Contact inhibited cell growth of transformed WT and SIRT4 KO MEFs cultured in the presence of DMSO or BPTES (10 μM) for 14 days (left). The number of colonies was counted (right). Data are means ±SEM. n.s., not significant. *p < 0.05, **p < 0.005. See also Figure S5.

SIRT4 represses tumor formation in vivo

To investigate SIRT4 function in human cancers, we examined changes in SIRT4 expression. SIRT4 mRNA level was reduced in several human cancers, such as small cell lung carcinoma (Garber et al., 2001), gastric cancer (Wang et al., 2012), bladder carcinoma (Blaveri et al., 2005), breast cancer (TCGA) and leukemia (Choi et al., 2007) (Figure 6A). Of note, lower SIRT4 expression associated with shorter time to death in lung tumor patients (Shedden et al., 2008) (Figure 6B). Overall the expression data is consistent with the model that SIRT4 may play a tumor suppressive role in human cancers.

Figure 6 SIRT4 is a mitochondrial tumor suppressor

SIRT4 regulates glutamine metabolism in lung tissue

To test further the biological relevance of this pathway in lung, we examined whether SIRT4 is induced in vivo after exposure to DNA damaging IR treatment. Remarkably, Sirt4 was significantly induced in lung tissue after IR exposure (Figure 7A). We next examined whether IR repressed glutamine metabolism in vivo, as observed in cell culture by examining GDH activity in lung tissue from WT and SIRT4 KO mice with or without IR exposure. GDH activity was elevated in lung tissue extracts from SIRT4 KO mice compared with WT lung tissue (Figure 7B). Importantly, GDH activity was significantly decreased in lung tissue from WT mice after IR exposure, whereas not in lung tissue from KO mice (Figure 7C). Thus, these findings recapitulate our cellular studies and are in line with the model that SIRT4 induction with DNA damage limits mitochondrial glutamine metabolism and utilization.

SIRT4 inhibits mitochondria glutamine metabolism in vivo

SIRT4 inhibits mitochondria glutamine metabolism in vivo

Figure 7 SIRT4 inhibits mitochondria glutamine metabolism in vivo

To assess whether the functions of SIRT4 can be reproduced in these lung tumors, cells derived from SIRT4 KO lung tumors were reconstituted with wild type SIRT4 (Figure S7A). As expected, SIRT4 reconstitution reduced glutamine uptake, but not glucose uptake (Figures 7D and 7E) and repressed proliferation (Figure S7B) of lung tumor cells.

Here, we report that SIRT4 has an important role in cellular metabolic response to DNA damage by regulating mitochondrial glutamine metabolism with important implication for the DDR and tumorigenesis. First, we discovered that DNA damage represses cellular glutamine metabolism (Figure 1). Next, we found that SIRT4 is induced by genotoxic stress (Figure 2) and is required for the repression of mitochondrial glutamine metabolism (Figure 3). This metabolic response contributes to the control of cell cycle progression and the maintenance of genomic integrity in response to DNA damage (Figure 4). Loss of SIRT4 increased glutamine-dependent tumor cell proliferation and tumorigenesis (Figure 5). In mice, SIRT4 loss resulted in spontaneous tumor development (Figure 6). We demonstrate that SIRT4 is induced in normal lung tissue in response to DNA damage where it represses GDH activity. Finally, the glutamine metabolism-genomic fidelity axis is recapitulated in lung tumor cells derived from SIRT4 KO mice via SIRT4 reconstitution (Figure 7). Our studies therefore uncover SIRT4 as a important regulator of cellular metabolic response to DNA damage that coordinates repression of glutamine metabolism, genomic stability and tumor suppression.

The DDR is a highly orchestrated and well-studied signaling response that detects and repairs DNA damage. Upon sensing DNA damage, the ATM/ATR protein kinases are activated to phosphorylate target proteins, leading to cell cycle arrest, DNA repair, transcriptional regulation and initiation of apoptosis (Ciccia and Elledge, 2010Su, 2006). Dysregulation of this pathway is frequently observed in many tumors. Emerging evidence has suggested that cell metabolism also plays key roles downstream of the DDR-induced pathways.


7.8.9 Mitochondrial sirtuins and metabolic homeostasis

Pirinen E1Lo Sasso GAuwerx J.
Best Pract Res Clin Endocrinol Metab. 2012 Dec; 26(6):759-70.

The maintenance of metabolic homeostasis requires the well-orchestrated network of several pathways of glucose, lipid and amino acid metabolism. Mitochondria integrate these pathways and serve not only as the prime site of cellular energy harvesting but also as the producer of many key metabolic intermediates. The sirtuins are a family of NAD+-dependent enzymes, which have a crucial role in the cellular adaptation to metabolic stress. The mitochondrial sirtuins SIRT3, SIRT4 and SIRT5 together with the nuclear SIRT1 regulate several aspects of mitochondrial physiology by controlling posttranslational modifications of mitochondrial protein and transcription of mitochondrial genes. Here we discuss current knowledge how mitochondrial sirtuins and SIRT1 govern mitochondrial processes involved in different metabolic pathways.

Mitochondria are organelles composed of a matrix enclosed by a double (inner and outer) membrane (1). Major cellular functions, such as nutrient oxidation, nitrogen metabolism, and especially ATP production, take place in the mitochondria. ATP production occurs in a process referred to as oxidative phosphorylation (OXPHOS), which involves electron transport through a chain of protein complexes (I-IV), located in the inner mitochondrial membrane. These complexes carry electrons from electron donors (e.g. NADH) to electron acceptors (e.g. oxygen), generating a chemiosmotic gradient between the mitochondrial intermembrane space and matrix. The energy stored in this gradient is then used by ATP synthase to produce ATP (1). One well-known side effect of the OXPHOS process is the production of reactive oxygen species (ROS) that can generate oxidative damage in biological macromolecules (1). However, to neutralize the harmful effects of ROS, cells have several antioxidant enzymes, including superoxide dismutase, catalase, and peroxidases (1). The sirtuin silent information regulator 2 (Sir2), the founding member of the sirtuin protein family, was identified in 1984 (2). Sir2 was subsequently characterized as important in yeast replicative aging (3) and shown to posses NAD+-dependent histone deacetylase activity (4), suggesting it could play a role as an energy sensor. A family of conserved Sir2-related proteins was subsequently identified. Given their involvement in basic cellular processes and their potential contribution to the pathogenesis of several diseases (5), the sirtuins became a widely studied protein family.

In mammals the sirtuin family consists of seven proteins (SIRT1-SIRT7), which show different functions, structure, and localization. SIRT1 is mostly localized in the nucleus but, under specific physiological conditions, it shuttles to the cytosol (6). Similar to SIRT1, also SIRT6 (7) and SIRT7 (8) are localized in the nucleus. On the contrary, SIRT2 is mainly present in the cytosol and shuttles into the nucleus during G2/M cell cycle transition (9). Finally, SIRT3, SIRT4, and SIRT5, are mitochondrial proteins (10).

The main enzymatic activity catalyzed by the sirtuins is NAD+-dependent deacetylation, as known for the progenitor Sir2 (4,11). Along with histones also many transcription factors and enzymes were identified as targets for deacetylation by the sirtuins. Remarkably, mammalian sirtuins show additional interesting enzymatic activities. SIRT4 has an important ADP-ribosyltransferase activity (12), while SIRT6 can both deacetylate and ADP-ribosylate proteins (13,14). Moreover, SIRT5 was recently shown to demalonylate and desuccinylate proteins (15,16), in particular the urea cycle enzyme carbamoyl phosphate synthetase 1 (CPS1) (16). The (patho-)physiological context in which the seven mammalian sirtuins exert their functions, as well as their biochemical characteristics, are extensively discussed in the literature (17,18) and will not be addressed in this review; here we will focus on the emerging roles of the mitochondrial sirtuins, and their involvement in metabolism. Moreover, SIRT1 will be discussed as an important enzyme that indirectly affects mitochondrial physiology.

Sirtuins are regulated at different levels. Their subcellular localization, but also transcriptional regulation, post-translational modifications, and substrate availability, all impact on sirtuin activity. Moreover, nutrients and other molecules could affect directly or indirectly sirtuin activity. As sirtuins are NAD+-dependent enzymes, the availability of NAD+ is perhaps one of the most important mechanisms to regulate their activity. Changes in NAD+ levels occur as the result of modification in both its synthesis or consumption (19). Increase in NAD+ amounts during metabolic stress, as prolonged fasting or caloric restriction (CR) (2022), is well documented and tightly connected with sirtuin activation (4,19). Furthermore, the depletion and or inhibition of poly-ADP-ribose polymerase (PARP) 1 (23) or cADP-ribose synthase 38 (24), two NAD+consuming enzymes, increase SIRT1 action.

Analysis of the SIRT1 promoter region identified several transcription factors involved in up- or down-regulation of SIRT1 expression. FOXO1 (25), peroxisome proliferator-activated receptors (PPAR) α/β (26,27), and cAMP response element-binding (28) induce SIRT1 transcription, while PPARγ (29), hypermethylated in cancer 1 (30), PARP2 (31), and carbohydrate response element-binding protein (28) repress SIRT1 transcription. Of note, SIRT1 is also under the negative control of miRNAs, like miR34a (32) and miR199a (33). Furthermore, the SIRT1 protein contains several phosphorylation sites that are targeted by several kinases (34,35), which may tag the SIRT1 protein so that it only exerts activity towards specific targets (36,37). The beneficial effects driven by the SIRT1 activation – discussed below- led the development of small molecules modulators of SIRT1. Of note, resveratrol, a natural plant polyphenol, was shown to increase SIRT1 activity (38), most likely indirectly (22,39,40), inducing lifespan in a range of species ranging from yeast (38) to high-fat diet fed mice (41). The beneficial effect of SIRT1 activation by resveratrol on lifespan, may involve enhanced mitochondrial function and metabolic control documented both in mice (42) and humans (43). Subsequently, several powerful synthetic SIRT1 agonists have been identified (e.g. SRT1720 (44)), which, analogously to resveratrol, improve mitochondrial function and metabolic diseases (45). The precise mechanism of action of these compounds is still under debate; in fact, it may well be that part of their action is mediated by AMP-activated protein kinase (AMPK) activation (21,22,46), as resveratrol was shown to inhibit ATP synthesis by directly inhibiting ATP synthase in the mitochondrial respiratory chain (47), leading to an energy stress with subsequent activation of AMPK. However, at least in β-cells, resveratrol-mediated SIRT1 activation and AMPK activation seem to regulate glucose response in the opposite direction, pointing to the existence of alternative molecular targets (48).

Another hypothesis to explain the pleitropic effects of resveratrol suggests it inhibits cAMP-degrading phosphodiesterase 4 (PDE4), resulting in the cAMP-dependent activation of exchange proteins activated by cyclic AMP (Epac1) (40). The consequent Epac1-mediated increase of intracellular Ca2+ levels may then activate of CamKKβ-AMPK pathway (40), which ultimately will result in an increase in NAD+ levels and SIRT1 activation (21). Interestingly, also PDE4 inhibitors reproduce some of the metabolic benefits of resveratrol representing yet another putative way to activate SIRT1.

The regulation of the activity of the mitochondrial sirtuins is at present poorly understood. SIRT3 expression is induced in white adipose (WAT) and brown adipose tissues upon CR (49), while it is down-regulated in the liver of high-fat fed mice (50). SIRT3 activity changes also in the muscle after fasting (51) and chronic contraction (52). All these processes are associated with increase (20,53) or decrease (50) in NAD+ levels. From a transcriptional point of view, SIRT3 gene expression in brown adipocytes seems under the control of peroxisome proliferator-activated receptor gamma coactivator-1α (PGC-1α) -estrogen-related receptor α (ERRα) axis, and this effect is crucial for full brown adipocyte differentiation (54,55). SIRT4 expression is reported to be reduced during CR (12), while the impact of resveratrol on SIRT4 is still under debate (56). Finally, upon ethanol exposure, SIRT5 gene expression was shown to be decreased together with the NAD+levels (57), probably explaining the protein hyperacetylation caused by alcohol exposure (58).

Metabolic homeostasis

The maintenance of metabolic homeostasis is critical for the survival of all species to sustain body structure and function. Metabolic homeostasis is achieved through complicated interactions between metabolic pathways that govern glucose, lipid and amino acid metabolism. Mitochondria are organelles, which integrate these metabolic pathways by serving a physical site for the production and recycling of metabolic intermediates.

Glucose metabolism


Glucose homeostasis is regulated through various complex processes including hepatic glucose output, glucose uptake, glucose utilization and storage. The main hormones regulating glucose homeostasis are insulin and glucagon, and the balance between these hormones determines glucose homeostasis. Insulin promotes glucose uptake in peripheral tissues (muscle and WAT), glycolysis and storage of glucose as glycogen in the fed state, while glucagon stimulates hepatic glucose production during fasting. Sirtuins influence many aspects of glucose homeostasis in several tissues such as muscle, WAT, liver and pancreas.


The body’s ability to synthesise glucose is vital in order to provide an uninterrupted supply of glucose to the brain and survive during starvation. Gluconeogenesis is a cytosolic process, in which glucose is formed from non-carbohydrate sources, such as amino acids, lactate, the glycerol portion of fats and tricarboxylic acid (59) cycle intermediates, during energy demand. This process, which occurs mainly in liver and kidney, shares some enzymes with glycolysis but it employs phosphoenolpyruvate carboxykinase, fructose-1,6-bisphosphatase and glucose-6-phosphatase to control the flow of metabolites towards glucose production. These three enzymes are stimulated by glucagon, epinephrine and glucocorticoids, whereas their activity is suppressed by insulin.

The role of mitochondrial sirtuins in the control of gluconeogenesis is not well established. SIRT3 is suggested to induce fasting-dependent hepatic glucose production from amino acids by deacetylating and activating the mitochondrial conversion of glutamate into the TCA cycle intermediate α-ketoglutarate, via the enzyme glutamate dehydrogenase (GDH) (Fig. 1A) (60,61). As SIRT3−/− mice do not display changes in GDH activity (62), the mechanism requires further clarification. In contrast to SIRT3, SIRT4 inhibits GDH via ADP-ribosylation under basal dietary conditions (Fig. 1A-B) (12). Conversely, SIRT4 activity is suppressed during CR resulting in activation of GDH, which fuels the TCA cycle and possibly also gluconeogenesis (12). Therefore, mitochondrial sirtuins may function to support gluconeogenesis during energy limitation, but further research is required to understand the exact roles of mitochondrial sirtuins in gluconeogenesis.

Summary of mitochondrial sirtuins’ role in mitochondrial pathways

Summary of mitochondrial sirtuins’ role in mitochondrial pathways

Figure 1 Summary of mitochondrial sirtuins’ role in mitochondrial pathways

Glucose utilization

 Lipid metabolism

Urea metabolism

The recent discoveries in the biology of mitochondria have shed light on the metabolic regulatory roles of the sirtuin family. To maintain proper metabolic homeostasis, sirtuins sense cellular NAD+ levels, which reflect the nutritional status of the cells, and translate this information to adapt the activity of mitochondrial processes via posttranslational modifications and transcriptional regulation. SIRT1 and SIRT3 function to stimulate proper energy production via FAO and SIRT3 also protects from oxidative stress and ammonia accumulation during nutrient deprivation. SIRT4 seems to play role in the regulation of gluconeogenesis, insulin secretion and fatty acid utilization during times of energy limitation, while SIRT5 detoxifies excess ammonia that can accumulate during fasting. However, we are only at the beginning of our understanding of the roles of the mitochondrial sirtuins, SIRT3, SIRT4 and SIRT5 in complex metabolic processes. In the coming years, further research should identify and verify novel sirtuin targets in vivo and in vitro. We need also to elucidate the regulation and tissue-specific functions of these mitochondrial sirtuins, as well as to understand the potential crosstalk and synchrony between the different sirtuins in different subcellular compartments. Ultimately, the understanding of mitochondrial sirtuin functions may open new possibilities, not only for treatment of cancer and metabolic diseases characterized by mitochondrial dysfunction, but also for disease prevention and health maintenance.

7.8.10 Mitochondrial sirtuins

Huang JY1Hirschey MDShimazu THo LVerdin E.
Biochim Biophys Acta. 2010 Aug; 1804(8):1645-51.

Sirtuins have emerged as important proteins in aging, stress resistance and metabolic regulation. Three sirtuins, SIRT3, 4 and 5, are located within the mitochondrial matrix. SIRT3 and SIRT5 are NAD(+)-dependent deacetylases that remove acetyl groups from acetyllysine-modified proteins and yield 2′-O-acetyl-ADP-ribose and nicotinamide. SIRT4 can transfer the ADP-ribose group from NAD(+) onto acceptor proteins. Recent findings reveal that a large fraction of mitochondrial proteins are acetylated and that mitochondrial protein acetylation is modulated by nutritional status. This and the identification of targets for SIRT3, 4 and 5 support the model that mitochondrial sirtuins are metabolic sensors that modulate the activity of metabolic enzymes via protein deacetylation or mono-ADP-ribosylation. Here, we review and discuss recent progress in the study of mitochondrial sirtuins and their targets.

mitochondrial sirtuins

mitochondrial sirtuins

mitochondrial sirtuins
Fig.1 .NAD+ -dependent deacetylation of sirtuins. The two step catalytic reaction mechanism. In this diagram ADPR = acetyl-ADP-ribose, NAM = nicotinamide, 1-O-AADPR = 1-O-acetyl ADP-ribose and βNAD = beta nicotinamide adenine dinucleotide.

Table 1 Shows subcellular localization, substrates and functions of different types of sirtuins.

Fig.2. Sirt3 regulated pathways in mitochondrial metabolism. Schematic diagram demonstrating the different roles of Sirt3 in the regulation of the main metabolic pathways of mitochondria.In this diagram LCAD = long-chain acyl-CoA dehydrogenase, ACeS2 = acetyl coenzyme synthetase 2, Mn SOD = manganese superoxide dismutase, CypD = cyclophilin D, ICDH2 = isocitrate dehydrogenase 2, OTC = ornithine transcarbomylase,TCA = tricaboxylic acid, ROS = reactive oxygen species, mPTP = membrane permeability transition pore, I–V = respiratory chain complex I–V

Fig. 3.(A) Schematic diagram showing different roles of Sirt4 in the regulation of various metabolic pathways. The diagram shows the Sirt4 regulated decrease in insulin level and the increase in availability of ATP inside mitochondria via upregulation of insulin degrading enzyme (IDE) and adenine translocator (ANT). The diagram also shows the Sirt4 regulated decrease in the efficiency of fatty acid oxidation and tricarboxylic acid cycle (TCA) via inhibition of glutamate dehydrogenase (GDH) and malonyl CoA decarboxylase (MCoAD). (B) Schematic diagram indicating the different roles of Sirt5 in regulation of various metabolic pathways. Sirt5 regulates urea production, fatty acid oxidation, tricarboxylic acid cycle (TCA), glycolysis, reactive oxygen species (ROS) metabolism, purine metabolism via regulating carbamoyl phosphate synthetase (CPS), hydroxyl-coenzyme A dehydrogenase (HADH), pyruvate dehydrogenase (PDH), pyruvate kinase (PK), succinate dehydrogenase(SDH) andurate oxidase (UO) respectively

Conclusion and future perspectives

Sirtuins are highly conserved NAD+-dependent protein deacetylases or ADP ribosyl transferases involved in many cellular processes including genome stability, cell survival, oxidative stress responses, metabolism, and aging. Mitochondrial sirtuins, Sirt3, Sirt4 and Sirt5 are important energy sensors and thus can be regarded as master regulators of mitochondrial metabolism. But it is still not known whether specific sirtuins can only function within particular metabolic pathways or two or more sirtuins could affect the same pathways. One of the mitochondrial sirtuins, Sirt3 is a major mitochondrial deacetylase that plays a pivotal role in the acetylation based regulation of numerous mitochondrial proteins. However, the question how mitochondrial proteins become acetylated is still unsolved and the identity of mitochondrial acetyltransferases is mysterious. Although the predominant function of the sirtuins is NAD+ dependent lysine deacetylation, but along with this major function another less characterized activity of these sirtuins includes ADP ribosylation which is mainly done by Sirt4. Moreover, in the case when the mitochondrial sirtuins exhibit both deacetylase and ADP ribosyl transferase activity, the conditions that determine the relative contribution of both of these activities in same or different metabolic pathways require further investigation. Sirt5 another mitochondrial sirtuin, was a puzzle until the recent finding as it possesses unique demalonylase and desuccinylase activities. However, most of the malonylated or succinylated proteins are important metabolic enzymes but as the significance of lysine malonylation and succinylation is still unknown thus it would be interesting to know how lysine malonylation and succinylation alter the functions of various metabolic enzymes. The mitochondrial sirtuins Sirt3, Sirt4 and Sirt5 serve as critical junctions and are required to exert many of the beneficial effect in mitochondrial metabolism. The emerging multidimensional role of mitochondrial sirtuins in regulation of mitochondrial metabolism and bioenergetics may have far-reaching consequences for many diseases associated with mitochondrial dysfunctions. However it is very important to fully elucidate the functions of mitochondrial sirtuins in different tissues to achieve the goal of therapeutic intervention in different metabolic diseases. Although several proteomic studies have provided detailed information that how mitochondrial sirtuin driven modification takes place on various targets in response to different environmental conditions, still the role of sirtuins in mitochondrial physiology and human diseases requires further exploration. Hopefully the progress in the field of sirtuin biology will soon provide insight into the therapeutic applications for targeting mitochondrial sirtuins by bioactive compounds to treat various human age-related diseases.


Ahn B.H.,et al.,2008. A role for the mitochondrial deacetylase Sirt3 in regulating energy homeostasis. Proc. Natl. Acad. Sci. U. S. A. 105 (38), 14447–14452.

Ahuja N.,et al., 2007. Regulation of insulin secretion by SIRT4, a mitochondrial ADP ribosyltransferase. J. Biol. Chem. 282 (46), 33583–33592.

Allison, S.J., Milner, J., 2007. SIRT3 is pro-apoptotic and participates in distinct basal apoptotic pathways. Cell Cycle 6, 2669–2677.

Ashraf, N., et al., 2006. Altered sirtuin expression is associated with node-positive breast cancer. Br. J. Cancer 95, 1056–1061.

Bao, J.,et al.,2010. SIRT3 is regulated by nutrient excess and modulates hepatic susceptibility to lipotoxicity. Free Radic. Biol. Med. 49, 1230–1237.

Beal, M.F., 2005. Less stress, longer life. Nat. Med. 11 (6), 598–599.

Bell, E.L., Guarente,L., 2011. The SirT3 divining rod points to oxidative stress. Mol.Cell 42 (5), 561–568.

Bell,E.L., Emerling,B.M., Ricoult,S.J.H., Guarente,L., 2011. SirT3 suppresses hypoxia inducible factor 1α and tumor growth by inhibiting mitochondrial ROS production. Oncogene 30, 2986–2996.

Bellizzi,D.,Rose,G.,Cavalcante,P.,Covello,G.,et al., 2005. A novel VNTR enhancer within the SIRT3 gene, a human homologue of SIR2, is associated with survival at oldest ages. Genomics 85, 258–263.

7.8.11 Sirtuin regulation of mitochondria: energy production, apoptosis, and signaling

Verdin E1Hirschey MDFinley LWHaigis MC.
Trends Biochem Sci. 2010 Dec; 35(12):669-75.

Sirtuins are a highly conserved family of proteins whose activity can prolong the lifespan of model organisms such as yeast, worms and flies. Mammals contain seven sirtuins (SIRT1-7) that modulate distinct metabolic and stress response pathways. Three sirtuins, SIRT3, SIRT4 and SIRT5, are located in the mitochondria, dynamic organelles that function as the primary site of oxidative metabolism and play crucial roles in apoptosis and intracellular signaling. Recent findings have shed light on how the mitochondrial sirtuins function in the control of basic mitochondrial biology, including energy production, metabolism, apoptosis and intracellular signaling.

Mitochondria play critical roles in energy production, metabolism, apoptosis, and intracellular signaling [13]. These highly dynamic organelles have the ability to change their function, morphology and number in response to physiological conditions and stressors such as diet, exercise, temperature, and hormones [4]. Proper mitochondrial function is crucial for maintenance of metabolic homeostasis and activation of appropriate stress responses. Not surprisingly, changes in mitochondrial number and activity are implicated in aging and age-related diseases, including diabetes, neurodegenerative diseases, and cancer [1]. Despite the important link between mitochondrial dysfunction and human diseases, in most cases, the molecular causes for dysfunction have not been identified and remain poorly understood.

One of the principal bioenergetic functions of mitochondria is to generate ATP through the process of oxidative phosphorylation (OXPHOS), which occurs in the inner-mitochondrial membrane. Mitochondria are unique bi-membrane organelles that contain their own circular genome (mtDNA) encoding 13 protein subunits involved in electron transport. The remainder of the estimated 1000-1500 mitochondrial proteins are encoded by the nuclear genome and imported into mitochondria from the cytoplasm [56]. These imported proteins can be found either in the matrix, associated with inner or outer mitochondrial membranes or in the inner membrane space (Figure 1). Dozens of nuclear-encoded protein subunits form complexes with the mtDNA-encoded subunits to form electron transport complexes I-IV and ATP synthase, again highlighting the need for precise coordination between these two genomes. The transcriptional coactivator PGC-1α, a master regulator of mitochondrial biogenesis and function, is responsive to a variety of metabolic stresses, ensuring that the number and capacity of mitochondria keeps pace with the energetic demands of tissues [7].

Network of mitochondrial sirtuins

Network of mitochondrial sirtuins

Network of mitochondrial sirtuins. Mitochondria can metabolize fuels, such as fatty acids, amino acids, and pyruvate, derived from glucose. Electrons pass through electron transport complexes (I-IV; red) generating a proton gradient, which is used to drive ATP synthase (AS; red) to generate ATP. SIRT3 (gold) binds complexes I and II, regulating cellular energy levels in the cell [4355]. Moreover, SIRT3 binds and deacetylates acetyl-CoA synthetase 2 (AceCS2) [3940] and glutamate dehydrogenase (GDH) [3347], thereby activating their enzymatic activities. SIRT3 also binds and activates long-chain acyl-CoA dehydrogenase (LCAD) [46]. SIRT4 (light purple) binds and represses GDH activity via ADP-ribosylation [21]. In the rate-limiting step of the urea cycle, SIRT5 (light blue) deacetylates and activates carbamoyl phosphate synthetase 1 (CPS1) [4849].

As high-energy electrons derived from glucose, amino acids or fatty acids fuels are passed through a series of protein complexes (I-IV), their energy is used to pump protons from the mitochondrial matrix through the inner membrane into the inner-membrane space, generating a proton gradient known as the mitochondrial membrane potential (Dψm) (Figure 1). Ultimately, the electrons reduce oxygen to form water, and the protons flow down their gradient through ATP synthase, driving the formation of ATP from ADP. Protons can also flow through uncoupling proteins (UCPs), dissipating their potential energy as heat. Reactive oxygen species (ROS) are a normal side-product of the respiration process [18]. In addition, an increase in Dψm, whether caused by impaired OXPHOS or by an overabundance of nutrients relative to ADP, will result in aberrant electron migration in the electron transport chain and elevated ROS production [1]. ROS react with lipids, protein and DNA, generating oxidative damage. Consequently, cells have evolved robust mechanisms to guard against an increase in oxidative stress accompanying ROS production [9].

Mitochondria are the primary site of ROS production within the cell, and increased oxidative stress is proposed to be one of the causes of mammalian aging [1210]. Major mitochondrial age-related changes are observed in multiple tissues and include decreased Dψm, increased ROS production and an increase in oxidative damage to mtDNA, proteins, and lipids [1114]. As a result, mitochondrial bioenergetic changes that occur with aging have been extensively reviewed [1517].

Silent information regulator (SIR) 2 protein and its orthologs in other species, termed sirtuins, promote an increased lifespan in model organisms such as yeast, worms and flies. Mammals contain seven sirtuins (SIRT1–7) that are characterized by an evolutionary conserved sirtuin core domain [1819]. This domain contains the catalytic activity and invariant amino acid residues involved in binding NAD+, a metabolic co-substrate. All sirtuins exhibit two major enzymatic activities in vitro: NAD+-dependent protein deacetylase activity and ADP-ribosyltransferase activity. Except for SIRT4, well-defined acetylated substrates have been identified for the other sirtuins. So far, only ADP-ribosyltransferase activity has been described for SIRT4 [2021]. Thus, these enzymes couple their biochemical and biological functions to an organism’s energetic state via their dependency on NAD+. A decade of research, largely focused on SIRT1, has revealed that mammalian sirtuins regulate metabolism and cellular survival. In brief, SIRT1–7 target distinct acetylated protein substrates and are localized in distinct subcellular compartments. SIRT1, SIRT6 and SIRT7 are found in nucleus, SIRT2 is primarily cytosolic and SIRT3, 4 and 5 are found in the mitochondria. The mitochondrial-only localization of SIRT3 is controversial and other groups have reported non-mitochondrial localization of this sirtuin [2223]. The biology and biochemistry of the seven mammalian sirtuins have been extensively discussed in the literature [2426] and is not the topic of this review. Instead, we focus on the mitochondrial sirtuins, their substrates, and their impact on mitochondrial biology.

The mitochondrial sirtuins, SIRT3–5 [212729], participate in the regulation of ATP production, metabolism, apoptosis and cell signaling. Unlike SIRT1, a 100 kDa protein, the mitochondrial sirtuins are small, ranging from 30–40 kDa. Thus, their amino acid sequence consists mostly of an N-terminal mitochondrial targeting sequence and the sirtuin core domain, with small flanking regions. Whereas, SIRT3 and SIRT5 function as NAD+-dependent deacetylases on well defined substrates, SIRT4 has no identified acetylated substrate and only shows ADP-ribosyltransferase activity. It is likely, however, that SIRT4 possesses substrate-specific NAD+-dependent deacetylase activity, as has been demonstrated for SIRT6 [30,31]. The three-dimensional structures for the core domains of human SIRT3 and human SIRT5 have been solved and reveal remarkable structural conservation with other sirtuins, such as the ancestral yeast protein and human SIRT2 (Figure 2) [3234]. Given its sequence conservation with the other sirtuins [18], it is likely that SIRT4 adopts a similar three-dimensional conformation.

Figure 2 Structure and alignment of sirtuins

Role of mitochondrial sirtuins in metabolism and energy production

The NAD+ dependence of sirtuins provided the first clue that these enzymes function as metabolic sensors. For instance, sirtuin activity can increase when NAD+ levels are abundant, such as times of nutrient deprivation. In line with this model, mass spectrometry studies have revealed that metabolic proteins, such as tricarboxylic acid (TCA) cycle enzymes, fatty acid oxidation enzymes and subunits of oxidative phosphorylation complexes are acetylated in response to metabolic stress [3537].

Fatty acid oxidation

Consistent with the hypothesis that nutrient stress alters sirtuin activity, a recent report identified significant metabolic abnormalities in Sirt3-/- mice during fasting [38]. In this study, hepatic SIRT3 protein expression increased during fasting, suggesting that both its levels and enzymatic activity are elevated during nutrient deprivation. SIRT3 activates hepatic lipid catabolism via deacetylation of long-chain acyl-CoA dehydrogenase (LCAD), a central enzyme in the fatty acid oxidation pathway. Sirt3-/- mice have diminished fatty acid oxidation, develop fatty liver, have low ATP production, and show a defect in thermogenesis and hypoglycemia during a cold test [38].

Surprisingly, many of the phenotypes observed in Sirt3-/- mice were also observed in mice lacking acetyl-CoA synthetase 2 (AceCS2), a previously identified substrate of SIRT3 [3940]. For example, fasting ATP levels were reduced by 50% in skeletal muscle of AceCS2-/- mice, in comparison to wild type (WT) mice. As a result, fasted AceCS2-/- mice were hypothermic and had reduced capacity for exercise. By converting acetate into acetyl CoA, AceCS2 provides an alternate energy source during times of metabolic challenges, such as thermogenesis or fasting. Interestingly, Acadl-deficient mice (Acadl encodes LCAD) also show cold intolerance, reduced ATP, and hypoglycemia under fasting conditions [41]. These overlapping phenotypes between Sirt3-/-AceCS2-/- and Acadl-/- mice indicate that the regulation of LCAD and AceCS2 acetylation by SIRT3 represents an important adaptive signal during the fasting response (Figure 2).

Electron transport chain

Of all mitochondrial proteins, oxidative phosphorylation complexes are among the most heavily acetylated. One study reported that 511 lysine residues in complexes I-IV and ATP synthase are modified by acetylation [37], hinting that a mitochondrial sirtuin might deacetylate these residues. Indeed, SIRT3 interacts with and deacetylates complex I subunits (including NDUFA9) [42], succinate dehydrogenase (complex II) [43]. SIRT3 has also been shown to bind ATP synthase in a proteomic analysis [44]. SIRT3 also regulates mitochondrial translation, a process which can impact electron transport [45]. Mice lacking SIRT3 demonstrate reduced ATP levels in many tissues [42 46]; however, additional work is required to determine if reduced ATP levels in Sirt3-/- mice is a direct result of OX PHOS hyperacetylation or an indirect effect, via decreased fatty acid oxidation, or a combination of both effects.

Less is known about the roles of SIRT4 and SIRT5 in electron transport. SIRT4 binds adenine nucleotide translocator (ANT), which transports ATP into the cytosol and ADP into the mitochondrial matrix, thereby providing a substrate for ATP synthase [20]. SIRT5 physically interacts with cytochrome C. The biological significance of these interactions, however, remains unknown [21].

TCA cycle

Enzymes for the TCA cycle (also called the Kreb’s cycle) are located in the mitochondrial matrix; this compartmentalization provides a way for cells to utilize metabolites from carbohydrates, fats and proteins. Numerous TCA cycle enzymes are modified by acetylation, although the functional consequences of acetylation have been examined for only a few of these proteins. SIRT3 interacts with several TCA cycle enzymes, including succinate dehydrogenase (SDH, see above [43]) and isocitrate dehydrogenase 2 (ICDH2) [33]. ICDH2 catalyzes the irreversible oxidative decarboxylation of isocitrate to form alpha-ketoglutarate and CO2, while converting NAD+ to NADH. Although the biological significance of these interactions is not yet known, it seems possible that SIRT3 might regulate flux through the TCA cycle.

Role of mitochondrial sirtuins in signaling

During cellular stress or damage, mitochondria release a variety of signals to the cytosol and the nucleus to alert the cell of changes in mitochondrial function. In response, the nucleus generates transcriptional changes to activate a stress response or repair the damage. For example, mitochondrial biogenesis requires a sophisticated transcriptional program capable of responding to the energetic demands of the cell by coordinating expression of both nuclear and mitochondrial encoded genes [4]. Unlike anterograde transcriptional control of mitochondria from nuclear transcription regulators such as PGC-1α, the retrograde signaling pathway, from the mitochondria to the nucleus is poorly understood in mammals. Although there is no evidence directly linking sirtuins to a mammalian retrograde signaling pathway, changes in mitochondrial sirtuin activity could influence signals transmitted from the mitochondria. Interestingly, the nuclear sirtuin SIRT1 deacetylates and activates PGC-1α, a key factor in the transcriptional regulation of genes involved in fatty acid oxidation and oxidative phosphorylation (Figure 3) [5051]. Thus, mitochondrial and nuclear sirtuins might exist in a signaling communication loop to control metabolism.



Mitochondria at nexus of cellular signaling. Mitochondria and mitochondrial sirtuins play a central role in intra- and extra-cellular signaling. Circulating fatty acids and acetate provide whole body energy homeostasis. The mitochondrial metabolites NAD+, NADH, ATP, Ca2+, ROS, ketone bodies, and acetyl-CoA participate in intracellular signaling.

Numerous signaling pathways are activated by changes in mitochondrial release of metabolites and molecules, such as Ca2+, ATP, NAD+, NADH, nitric oxide (NO), and ROS (Figure 3). Of these, Ca2+ is the best studied as a mitochondrial messenger. Mitochondria are important regulators of Ca2+ storage and homeostasis, and mitochondrial Ca2+ uptake is directly tied to the membrane potential of the organelle. Membrane potential serves as a gauge of mitochondrial function: disruption of OXPHOS, interruption in the supply or catabolism of nutrients or loss of structural integrity generally result in a fall in membrane potential, and, in turn, decreased mitochondrial Ca2+ uptake. Subsequent increases in cytosolic free Ca2+ will activate calcineurin and several Ca2+-dependent kinases [52] and affect a wide variety of transcription factors to produce appropriate cell-specific transcriptional responses [53]. Through regulation of nutrient oxidation and electron transport or yet to be identified target(s), mitochondrial sirtuins could influence mAlthough the effect of sirtuins on intracellular calcium signaling has not been studied directly, sirtuin effects on ATP production have been shown. ANT facilitates the exchange of mitochondrial ATP with cytosolic ADP. As a result the cytosolic ATP:ADP ratio reflects changes in mitochondrial energy production. A fall in ATP production activates AMP-activated protein kinase (AMPK), which directly stimulates mitochondrial energy production, inhibits protein synthesis through regulation of mammalian target of rapamycin (mTOR), and influences mitochondrial transcriptional programs [54]. SIRT3 regulates ATP levels in a variety of tissues, suggesting that its activity could have an important role in ATP-mediated retrograde signaling [46,55]. Indeed, recent studies have shown that SIRT3 regulates AMPK activation [5658]. Furthermore, SIRT4 interacts with ANT [20], raising the possibility that SIRT4 activity also influences the ATP:ADP ratio or membrane potential and modulates important mitochondrial signals.

NAD+ and NADH levels are intimately connected with mitochondrial energy production and regulate mitochondrial sirtuin activity. Unlike NAD+, however, NADH is not a sirtuin co-substrate. Indeed, changes in the NAD+:NADH ratio can change the redox state of the cell and alter the activity of enzymes such as poly-ADP-ribose polymerases and sirtuins, with subsequent effects on signaling cascades and gene expression [5961]. Changes in mitochondrial sirtuin activity could change the balance of these metabolites within the mitochondria. For example, fatty acid oxidation reduces NAD+ to NADH, which is oxidized back to NAD+ by OXPHOS. However, it is unclear whether changes in NAD+/NADH can be transmitted outside the organelle. The inner mitochondrial membrane is impermeable to NAD+ and NADH; however, the mitochondrial malate-aspartate shuttle could transfer reducing equivalents across the mitochondrial membranes.

Mitochondrial sirtuin control of apoptosis

Apoptosis is a cellular process of programmed cell death. Mitochondria play an important role in apoptosis by the activation of mitochondrial outer membrane permeabilization, which represents the irrevocable point of no return in committing a cell to death. Outer membrane permeabilization leads to the release of caspase-activating molecules, caspase-independent death effectors, and disruption of ATP production. Despite the central role for mitochondria in the control of apoptosis, surprisingly little is known about how mitochondrial sirtuins participate in apoptotic programs. SIRT3 plays a pro-apoptotic role in both BCL2-53- and JNK-regulated apoptosis [63]. Additionally, cells lacking SIRT3 show decreased stress-induced apoptosis, lending further support for a pro-apoptotic role for SIRT3 [62]. Furthermore, recent work points to a tumor suppressive role for SIRT3: SIRT3 levels are decreased in human breast cancers and Sirt3 null mice develop mammary tumors after 12 months [62]. The mechanism for the tumor suppressive function of SIRT3 is incompletely understood, but involves repression of ROS and protection against DNA damage [62]. In conflicting studies, SIRT3 has been shown to be anti-apoptotic. For example, in the cellular response to DNA damage when mitochondrial NAD+ levels fall below critical levels, SIRT3 and SIRT4 display anti-apoptotic activity, protecting cells from death [64]. SIRT3 has also been shown to be cardioprotective, in part by activation of ROS clearance genes [65]. In future studies, it will be important to elucidate the balance achieved by SIRT3 between stress resistance (anti-apoptosis) and tumor suppression (pro-apoptosis). Additionally, the role of SIRT4 and SIRT5 in regulating metabolism suggests that these mitochondrial sirtuins could also contribute to apoptosis in tumor suppressive or stress resistant manners.

Concluding remarks

An elegant coordination of metabolism by mitochondrial sirtuins is emerging where SIRT3, SIRT4 and SIRT5 serve at critical junctions in mitochondrial metabolism by acting as switches to facilitate energy production during nutrient adaptation and stress. Rather than satisfy, these studies lead to more questions. How important are changes in global mitochondrial acetylation to mitochondrial biology and is acetylation status a readout for sirtuin activity? What are other substrates for SIRT4 and SIRT5? What molecular factors dictate substrate specificity for mitochondrial sirtuins? Moreover, further studies will provide insight into the therapeutic applications for targeting mitochondrial sirtuins to treat human diseases. It is clear that many discoveries have yet to be made in this exciting area of biology.

Body of review in energetic metabolic pathways in malignant T cells

Antigen stimulation of T cell receptor (TCR) signaling to nuclear factor (NF)-B is required for T cell proliferation and differentiation of effector cells.
The TCR-to-NF-B pathway is generally viewed as a linear sequence of events in which TCR engagement triggers a cytoplasmic cascade of protein-protein interactions and post-translational modifications, ultimately culminating in the nuclear translocation of NF-B.
Activation of effect or T cells leads to increased glucose uptake, glycolysis, and lipid synthesis to support growth and proliferation.
Activated T cells were identified with CD7, CD5, CD3, CD2, CD4, CD8 and CD45RO. Simultaneously, the expression of CD95 and its ligand causes apoptotic cells death by paracrine or autocrine mechanism, and during inflammation, IL1-β and interferon-1α. The receptor glucose, Glut 1, is expressed at a low level in naive T cells, and rapidly induced by Myc following T cell receptor (TCR) activation. Glut1 trafficking is also highly regulated, with Glut1 protein remaining in intracellular vesicles until T cell activation.

Dr. Aurel,
Targu Jiu

  1. sjwilliamspa

    Wouldn’t then the preferred target be mTORC instead of Sirtuins if mTORC represses Sirtuin activity?

  2. The answer may not be so simple, perhaps a conundrum.

    In conflicting studies, SIRT3 has been shown to be anti-apoptotic. For example, in the cellular response to DNA damage when mitochondrial NAD+ levels fall below critical levels, SIRT3 and SIRT4 display anti-apoptotic activity, protecting cells from death [64].

    For anti-cancer activity, apoptosis is a desired effect. This reminds me of the problem 15 years ago with the drug that would be effective against sepsis, the best paper of the year in NEJM. It failed.

    We tend to not appeciate the intricacies of biological interactions and fail to see bypass reactions. Pleotropy comes up again and again. The seminal work from Britton Chances lab on the NAD+/NADH ratio have been overlooked.

Read Full Post »

« Newer Posts - Older Posts »