Leaders in Pharmaceutical Business Intelligence (LPBI) Group

Autophagy

Writer and Curator: Larry H Bernstein, MD, FCAP

2.1.6      Autophagy

2.1.6.1 Cepharanthine induces apoptosis through reactive oxygen species

Hua P, Sun M, Zhang G, Zhang Y, Tian X, Li X, Cui R, Zhang X
Biochem Biophys Res Commun. 2015 Mar 5. pii: S0006-291X(15)00378-2
http://dx.doi.org/10.1016/j.bbrc.2015.02.131

Cepharanthine is a medicinal plant-derived natural compound which possesses potent anti-cancer properties. However, there is little report about its effects on lung cancer cells. In this study, we investigated the effects of cepharanthine on the cell viability and apoptosis in human non-small-cell lung cancer H1299 and A549 cells. It was found that cepharanthine inhibited the growth of H1299 and A549 cells in a dose-dependent manner which was associated with the generation of reactive oxygen species (ROS) and the dissipation of mitochondrial membrane potential (Δψm). These effects were markedly abrogated when cells were pretreated with N-acetylcysteine (NAC), a specific ROS inhibitor, indicating that the apoptosis-inducing effect of cepharanthine in lung cancer cells was mediated by ROS. In addition, cepharanthine triggered apoptosis in non-small lung cancer cells via the upregulation of Bax, downregulation of Bcl-2 and significant activation of caspase-3 and PARP. These results provide the rationale for further research and preclinical investigation of cepharanthine’s anti-tumor effect against human non-small-cell lung cancer.

2.1.6.2 Mitochondrial Shape Governs BAX-Induced membrane permeabilization and apoptosis

Renault TT, Floros KV, Elkholi R, Corrigan KA, Kushnareva Y, Wieder SY, et al.
Mol Cell. 2015 Jan 8;57(1):69-82
http://dx.doi.org/10.1016/j.molcel.2014.10.028.

Highlights

• A proapoptotic BCL-2 repertoire containing BIM, PUMA, and BAX initiates MOMP
  • BAX-dependent membrane permeabilization exhibits mitochondrial size requirements
  • Mitochondrial membrane shape directly regulates BAX alpha helix 9 to induce MOMP
  • Mitochondrial hyperfission can be pharmacologically reversed to promote apoptosis

Proapoptotic BCL-2 proteins converge upon the outer mitochondrial membrane (OMM) to promote mitochondrial outer membrane permeabilization (MOMP) and apoptosis. Here we investigated the mechanistic relationship between mitochondrial shape and MOMP and provide evidence that BAX requires a distinct mitochondrial size to induce MOMP. We utilized the terminal unfolded protein response pathway to systematically define proapoptotic BCL-2 protein composition after stress and then directly interrogated their requirement for a productive mitochondrial size. Complementary biochemical, cellular, in vivo, and ex vivo studies reveal that Mfn1, a GTPase involved in mitochondrial fusion, establishes a mitochondrial size that is permissive for proapoptotic BCL-2 family function. Cells with hyperfragmented mitochondria, along with size-restricted OMM model systems, fail to support BAX-dependent membrane association and permeabilization due to an inability to stabilize BAXα9·membrane interactions. This work identifies a mechanistic contribution of mitochondrial size in dictating BAX activation, MOMP, and apoptosis.

http://www.cell.com/cms/attachment/2023304675/2043672113/fx1.jpg

Figure 1. Terminal UPR Requires BAX

(A–D) WT and Bak/Bax/ MEFs were treated with b-ME (A), DTT (B), Tg (C), or Tun (D) for 18 hr. (E) WT, Bak/, and Bax/ MEFs were treated with b-ME (15 mM), DTT (5 mM), Tg (1.5 mM), or Tun (2.5 mg/ml) for 18 hr. (F) Lysates from ER stress-treated WT, Bak/, and Bax/ MEFs in (E) were analyzed by western blot. (G) CHAPS lysates from ER stress-treated WT MEFs (highest doses at 18hr) were subjected to 6A7 IP and western blot. Total cell lysates (5%) were analyzed as a loading control. (H) HM fractions isolated from ER stress-treated WT MEFs (highest doses at 18hr) were subjected to trypsinization and analyzed by western blot.Total cell lysates (5%) were analyzed as a loading control for BAX; VDAC is a pretrypsinization mitochondrial loading control. (I) WT and Bid/Bim/ MEFs were treated with DTT for 18 hr. (J) HM fractions isolated from ER stress-treated WT MEFs (highest doses at 18 hr) were analyzed by western blot. (K) HM fractions from ER stress-treated WT MEFs (highest doses at 18 hr) were incubated with ABT-737 (1 mM) for 30 min at 37 C, centrifuged, and the supernatants were analyzed by western blot. CHAPS (0.25%) lysed mitochondria indicates total cyto c within each lane. *Nonspecific band. (L) Same as in (J), but probed for PUMA and SMAC. (M) WT and Puma/ MEFs were treated with b-ME for 18 hr. (N) Puma/ MEFs were pretreated either with ABT-737 (1 mM) for 1 hr and then b-ME for 18 hr or with b-ME for 18 hr and then ABT-737 for an additional 6 hr. (O)A summary schematic of BCL-2 family interactions required for apoptosis to proceed.All data are representative of at least triplicate experiments and reported as ±SD, as required. See also Figures S1–S3.

Figure 3. Mitochondrial Network Shape Regulates tUPR (A) WT and Mfn1/ MEFs were treated with DTT for 18 hr. (B) WT, Mfn2/, and Mfn1/ MEFs were loaded with MitoTracker Green (50 nM) and Hoechst 33342 (20 mM) before imaging (4003). (C) Mfn1/ MEFs were treated with mDIVI-1 (25 mM) for 2 hr before imaging (4003). Further magnified regions (2.53) are shown in white boxes. The average length of 200 mitochondria is shown. (D) Mfn1/ MEFs were pretreated with mDIVI-1 (25 mM) for 2 or 8 hr and then DTT for 18 hr. (E and F) Mfn1/ MEFs were pretreated with mDIVI-1 (25 mM) for 8 hr, then Tg (0.25 mM) (E) or Tun (0.5 mg/ml) (F) for 18 hr. (G) Mfn1/ MEFs were pretreated with mDIVI-1 (25 mM) for 8 hr, then TNFa and CHX (10 mg/ml) for 18 hr. (H) HM fractions from ER stress-treated Mfn1/ MEFs were analyzed by western blot. (I) Mfn1/MEFs were pretreated withmDIVI-1 (25mM) for 2hr and ER stress agents for 18hr,and mitochondria were isolated and analyzed by western blot. High molecular weight complexes of BAX are indicated (*). VDAC is a loading control. (J) Mfn1/ MEFs were treated with mDIVI-1 (25 mM) for 8 hr, and lysates were analyzed by western blot. (K) WT MEFs were pretreated with mDIVI-1 (25 mM) for 8 hr and ER stress agents for 18 hr. (L) Mfn1/ MEFs were pretreated with mDIVI-1 (25 mM) for 8 hr and Paclitaxel or cisplatin for 18 hr. (M) Same as in (L), but A375. (N) Same as in (C), but A375.

Figure 4. Mitochondrial Size Dictates Sensitivity to BAX-Dependent MOMP (A) Schematic representation of measuring D(DcM) to detect MOMP. (BandC) Digitonin-permeabilized, JC-1-loaded Mfn1/MEFs (pretreated with 25mM mDIVI-1, or DMSO, for 8hr) were incubated with BAX (0.25mM) or OG-BAX (0.25 mM), and mitochondrial depolarization (DcM) was determined. Kinetic and endpoint measurements are shown in (B) and (C), respectively. (D and E) Same as in (B), but with BIM-S (25 nM). Kinetic and endpoint measurements are shown in (D) and (E), respectively. (F) JC-1-loaded WT liver mitochondria were fractionated by size, and the relationships between 0.5 and 0.05 mm LUVs are indicated on the same graph. (G and H) Larger (>0.5 mm; fractions 6–8) and smaller (<0.5 mm; fractions 11–15) WT mitochondria were treated with BAX (100 nM) or OG-BAX (100 nM) for 1 hr at 37^oC. Kinetic and endpoint measurements are shown in (G) and (H), respectively. (I) JC-1-loaded Bak/ liver mitochondria were fractionated by size. (J and K) Larger (>0.5 mm; fractions 8–10) and smaller (<0.5 mm; fractions 12–16) Bak/ mitochondria were treated with BAX (20 nM) ± BIM-S (20 nM) for 1 hr. Kinetic and endpoint measurements are shown in (J) and (K), respectively. All data are representative of at least triplicate experiments and reported as ±SD, as required. See also Figure S5.

Figure 5. BAX Preferentially Permeabilizes OMVs with Diameters Similar to Those of WT Mitochondria (A) Schematic representation of OMVs. (B) Unextruded OMVs were combined with BAX (40 nM) and N/C-BID (20 nM) or BIM BH3 peptide (2.5 mM) for 30 min at 37 C. (C) Kinetic traces of unextruded OMV permeabilization with BAX (40 nM) and BID (25 nM) or BIM BH3 (2.5 mM) for 30 min at 37 C. Triton X-100 solubilizes OMVs and establishes 100% release. An anti-FITC antibody is used to quench the FITC-dextran released during permeabilization. (D–F) DLS analyses of extruded (1, 0.2, and 0.05 mm) OMVs. The major peak was calculated as the area under the curve and is reported as a percentage. (G) OMVs were combined with BAX (0.25 mM) for 10 or 30 min. (H) OMVs were combined with BAX (40 nM) and BIM BH3 (2.5 mM) for 10 or 30 min. (I) Same as in (H), but with N/C-BID (20 nM). (J) OMVs were combined with BAX (40 nM), BIM BH3 (2.5 mM), BCL-xLDC (300 nM), and PUMA BH3 (5 mM) for 30 min. All data are representative of at least triplicate experiments and reported as ±SD, as required. See also Figure S6.

Figure 6. BAX Preferentially Permeabilizes LUVs with Diameters Similar to Those of WT Mitochondria (A) Schematic representation of LUVs. (B) Standard LUVs (1 mm) were combined with BAX (100 nM) and N/C-BID (20 nM) or BIM BH3 (2.5 mM) for 1 hr at 37^oC. (C–E) DLS analyses of LUVs extruded 1 (C), 0.2 (D), and 0.05 (E) mm. (F) LUVs were combined with BAX (0.25 and 0.75 mM) for 1 hr. (G) LUVs were combined with BAX (75 and 100 nM) and BIM BH3 (2.5 mM) for 1 hr. (H) Same as in (G), but with N/C-BID (20 nM). (I) LUVs were combined with BAX (100 nM) and N/C-BID (20 nM) or BIM BH3 (2.5 mM) for 30minat 37^oC prior to centrifugation, solubilization, and western blot for associated BAX. (J) LUVs were combined with BAX (100nM), BIMBH3 (2.5mM), BCL-xLDC (300nM), and PUMABH3 (5mM) for 1 hr. All data are representative of at least triplicate experiments and reported as ±SD, as required. See also Figure S7.

Figure 7. BAX a9 Displays Requirements for Membrane Shape (A) LUVs were combined with BAX or BAXOG (0.25 mM) for 15 min at 37^oC. (B) BAX (100 ng) was incubated in the presence of BIM BH3 (2.5 mM) and LUVs for 30 min prior to 6A7 IP and western blot. (C) LUVs were combined with BAXWT or BAXDC (0.25 mM) for 1 hr at 37^oC. The required incubation time is longer for BAXDC compared to BAXWT, which increases BAXWT activity. (D) LUVs were combined with BAXWT or BAXS184A for 30 min at 37^oC. (E) Same as in (D), but with BIM BH3 (2.5 mM). (F) LUVs were combined with BAXWT or BAXS184A (100 nM) for 30 min at 37^oC prior to centrifugation, solubilization, and western blot for associated BAX. (G) NBD-BAXWT or NBD-BAXS184A was incubated with 1 mm LUVs for 5 min ± BIMBH3 (2.5 mM). An increase in NBD fluorescence indicates BAX$LUV interactions and is reported as fold increase compared to NBD-BAXWT + LUVs. (H) LUVs (1 mm) were combined with BAXWT or BAXS184A (50, 75, 100 nM) with BIM BH3 (2.5 mM) for 30 min at 37^oC. (I) LUVs were combined with BAXWT or BAXS184A (50 nM) with BIM BH3 (2.5 mM) for 30 min at 37^oC. (J) OMVs were combined with BAXWT or BAXS184A (50 nM) and BIM BH3 (2.5 mM) for 30 min at 37^oC. (K) NBD-BAXWT or NBD-BAXS184A ± BIM BH3 (2.5 mM) was incubated with OMVs for 30 min at 37^oC. The interaction between NBD-BAXWT + BIM BH3 with 1 mm OMVs is reported as 100%. (L) Digitonin-permeabilized, JC-1-loaded Mfn1/ MEFs were incubated with BIM BH3 (0.1 mM), BAXWT (50 nM), and BAXS184A (50 nM), and DDcM was determined. (M) Mfn1/ MEFs expressing shBax were reconstituted with human BAXWT or BAXS184A and treated with DTT (1.5 mM), and the kinetics of tUPR were evaluated by IncuCyte. (N) A schematic summarizing the relationship between BAX, mitochondrial shape, and apoptosis. All data are representative of at least triplicate experiments and reported as ±SD, as required. See also Figure S7.

2.1.6.3 Stress-Independent Activation of XBP1s and/or ATF6 Reveals Three Functionally Diverse ER Proteostasis Environments

MD Shoulders, LM Ryno, JC Genereux, JJ Moresco, PG Tu, et al.
Cell Reports 25 Apr 2013; 3(4):1279–1292
http://dx.doi.org:/10.1016/j.celrep.2013.03.024

Highlights

► Orthogonal, ligand-dependent control of XBP1s and/or ATF6 in a single cell ► Proteomic and transcriptomic characterization of XBP1s and/or ATF6 activation ► XBP1s and/or ATF6 influences pathogenic protein fates, but not the endogenous proteome ► Arm-selective UPR activation reduces secretion of destabilized transthyretin variants

The unfolded protein response (UPR) maintains endoplasmic reticulum (ER) proteostasis through the activation of transcription factors such as XBP1s and ATF6. The functional consequences of these transcription factors for ER proteostasis remain poorly defined. Here, we describe methodology that enables orthogonal, small-molecule-mediated activation of the UPR-associated transcription factors XBP1s and/or ATF6 in the same cell independent of stress. We employ transcriptomics and quantitative proteomics to evaluate ER proteostasis network remodeling owing to the XBP1s and/or ATF6 transcriptional programs. Furthermore, we demonstrate that the three ER proteostasis environments accessible by activating XBP1s and/or ATF6 differentially influence the folding, trafficking, and degradation of destabilized ER client proteins without globally affecting the endogenous proteome. Our data reveal how the ER proteostasis network is remodeled by the XBP1s and/or ATF6 transcriptional programs at the molecular level and demonstrate the potential for selective restoration of aberrant ER proteostasis of pathologic, destabilized proteins through arm-selective UPR activation.

http://ars.els-cdn.com/content/image/1-s2.0-S2211124713001319-fx1.jpg

One-third of the human proteome is directed to the endoplasmic reticulum (ER) for partitioning between folding and trafficking versus ER-associated degradation (ERAD), a decision primarily dictated by the exact composition of the ER protein homeostasis (or proteostasis) network (Balch et al., 2008; Braakman and Bulleid, 2011; Hartl et al., 2011; McClellan et al., 2005). This partitioning protects the integrity of downstream proteomes by ensuring that only folded, functional proteins are trafficked from the ER (Brodsky and Skach, 2011; Smith et al., 2011b; Wiseman et al., 2007).

The folding, trafficking, and degradation capacity of the ER is dynamically adjusted to meet demand by the unfolded protein response (UPR)—a stress-responsive signaling pathway comprising three integrated signaling cascades emanating from the ER transmembrane proteins IRE1, ATF6, and PERK (Schröder and Kaufman, 2005; Walter and Ron, 2011). UPR signaling is activated by the accumulation of misfolded or aggregated proteins within the ER lumen. UPR activation causes transient, PERK-mediated translational attenuation and activation of the basic leucine zipper transcription factors ATF4, XBP1s, and the cleaved N-terminal fragment of ATF6 downstream of the ER stress sensors PERK, IRE1, and full-length ATF6, respectively. These transcription factors increase expression of distinct but overlapping sets of genes comprising both ER-specific and general cellular proteostasis pathways (Adachi et al., 2008; Lee et al., 2003; Okada et al., 2002; Yamamoto et al., 2004, 2007). The three mechanistically distinct arms of the metazoan UPR presumably evolved to provide cells with flexibility to adapt to tissue-specific environmental and metabolic demands, creating a mechanism to restore ER proteostasis in response to a wide array of cellular insults (Gass et al., 2008; Harding et al., 2001; Kaser et al., 2008; Wu et al., 2007).

Pharmacologic activation of the UPR offers the potential to adapt ER proteostasis and rescue misfolded, aberrantly degraded, or aggregation-prone ER client proteins without significantly affecting the healthy, wild-type proteome (Balch et al., 2008; Walter and Ron, 2011). For example, activation of a UPR signaling pathway that increases ER protein folding capacity could decrease the aberrant ERAD and increase the ER folding and export of destabilized, mutant proteins, thereby ameliorating loss-of-function diseases such as cystic fibrosis or lysosomal storage diseases (Chiang et al., 2012; Mu et al., 2008; Wang et al., 2006). Alternatively, increasing ERAD activity could attenuate the secretion of destabilized, aggregation-prone proteins that undergo concentration-dependent extracellular aggregation into amorphous aggregates and amyloid fibrils (Braakman and Bulleid, 2011; Brodsky and Skach, 2011; Luheshi and Dobson, 2009; Sitia and Braakman, 2003), providing a potential strategy to ameliorate amyloid disease pathology.

Concomitant pharmacologic activation of the PERK, IRE1, and ATF6 UPR arms can be achieved by the application of toxic small molecules such as tunicamycin (Tm; inhibits protein N-glycosylation) or thapsigargin (Tg; disrupts ER calcium homeostasis) that induce ER protein misfolding and aggregation, ultimately causing apoptosis (Schröder and Kaufman, 2005; Walter and Ron, 2011). These global UPR activators have proven useful for delineating the molecular underpinnings of UPR signaling pathways. Unfortunately, the pleiotropic effects and acute toxicity of global UPR activation complicate studies focused on understanding how UPR activation (either global or arm selective) remodels the ER proteostasis network in the absence of an acute ER stress or how the partitioning between folding and trafficking versus degradation of ER client proteins can be influenced by arm-selective UPR activation. Thus, despite the considerable effort focused on understanding the signaling mechanisms of IRE1, ATF6, and PERK activation, the functional implications of activating these pathways on ER proteostasis pathway composition and function remain poorly defined.

Herein, we introduce small molecule-regulated, genetically encoded transcription factors that enable orthogonal activation of UPR transcriptional programs in the same cell. Using our methodology, we characterize the three distinct ER proteostasis environments accessible by activating XBP1s and/or ATF6 to physiologically relevant levels in the absence of stress. We also evaluate the functional consequences of activating XBP1s and/or ATF6 on the folding and trafficking versus degradation of destabilized ER client proteins, including transthyretin (TTR). Ultimately, we demonstrate that arm-selective UPR activation selectively reduces secretion of a destabilized, aggregation-prone TTR variant without affecting the analogous wild-type protein and without globally altering the endogenous intracellular or secreted proteomes. Our results demonstrate, in molecular detail, how the XBP1s and/or ATF6 transcriptional programs integrate to adapt ER proteostasis pathways and highlight the capacity of functionally distinct ER proteostasis environments accessed by arm-selective UPR activation to restore the aberrant ER proteostasis of destabilized protein variants.

To characterize the ER proteostasis environments accessible by the selective or combined activity of the UPR-associated transcription factors XBP1s and ATF6, we required methodology for the small molecule-mediated, orthogonal regulation of two transcription factors in the same cell. Tetracycline (tet)-repressor technology can be applied to allow doxycycline (dox)-dependent control of XBP1s levels in the physiologic range (Lee et al., 2003). However, we have found that tet-repressor regulation of ATF6 activity within the physiologically relevant range is difficult. Even after careful optimization and single-colony stable cell selection of HEK293T-REx cells expressing constitutively active ATF6(1–373) (henceforth termed ATF6) under the tet repressor, we observed nonphysiologic levels of ATF6 target gene expression and significant off-target effects including strong upregulation of established XBP1s target genes, following ATF6 induction at all permissive dox doses (Figures S1A and S1B). We required, therefore, an alternative strategy to regulate the ATF6 transcription factor that would be dosable and orthogonal to tet-repressor technology.

Figure S1. Development and Characterization of HEK293^DAX and HEK293^DYG Cell Lines, Related to Figure 1

(A) qPCR analysis of clonal HEK293T-REx cells stably expressing dox-inducible ATF6 treated for 12 hr with vehicle, or 1 μg/mL dox. The effects of activating the global unfolded protein response with tunicamycin (Tm; 10 μg/mL for 6 h) or thapsigargin (Tg; 10 μM for 2 h) in HEK293T-REx cells expressing DHFR.YFP are shown for comparison. The dox-inducible ATF6 cell line was carefully selected for the lowest levels of ATF6 expression across multiple isolated single colonies. Note the non-physiologic levels of Hyou1 and HerpUD induction and the upregulation of the established XBP1s-selective target Erdj4 following dox-dependent ATF6 activation. qPCR data are reported relative to appropriate clonal HEK293T-REx cell lines stably expressing dox-inducible eGFP or DHFR.YFP. qPCR data are reported as the mean ± 95% confidence interval.

(B) qPCR analysis of HerpUD mRNA levels in clonal HEK293T-REx cells expressing dox-inducible ATF6 treated for 12 hr with increasing concentrations of dox. Note the lack of dox dose-dependence of HerpUD upregulation in these cells. qPCR data are reported as the mean ± 95% confidence interval.

(C) Time course for induction of HerpUD in HEK293T-REx cells expressing DHFR.ATF6 or tet-inducible ATF6 and treated with 10 μM TMP or 1 μg/mL dox, respectively. Data are presented as percentage of maximal induction and calculated relative to vehicle-treated DHFR.YFP- or eGFP-expressing cells. qPCR data are reported as the mean ± 95% confidence interval.

(D) Immunoblot of nuclear (top) and post-nuclear (bottom) fractions from HEK293^DAX and HEK293^DYG cells treated 12 hr with dox (1 μg/mL), TMP (10 μM) or both. The immunoblot of matrin-3 shows the efficiency of the nuclear extraction.

(E) qPCR analysis of ATF6 and XBP1s target genes in HEK293^DAX cells following 12 hr activation of XBP1s (dox; 1 μg/mL), DHFR.ATF6 (TMP; 10 μM), or both. qPCR data are reported relative to vehicle-treated HEK293^DYG cells. qPCR data are reported as the mean ± 95% confidence interval.

(F) Representative autoradiogram of cell lysates prepared from HEK293^DAX cells pretreated for 12 hr with dox (1 μg/mL), TMP (10 μM), or both. Following activation, cells were labeled with [³⁵S]-methionine/cysteine for 30 min then chased in non-radioactive media for 0 or 4 hr, as indicated. These lysates are prepared from the same experiments described in Figure 2J.

(G) Quantification of cell lysate autoradiograms prepared from [³⁵S]-labeled HEK293^DAX cells following a 12 hr activation of XBP1s and/or ATF6 as described in Figure S1F. The quantified results reflect the amount of [³⁵S] incorporated into the cellular proteome directly following a 30 min labeling period. Error bars indicate the standard error from biological replicates (n = 3).

(H) Immunoblot of lysates prepared from HEK293^DAX cells treated with dox (1 μg/mL), TMP (10 μM), or both for the indicated time. Tg (1 μg/mL) was added for 2 hr as a control.

(I) HEK293^DAX cells were plated at 5,000 cells/well in a translucent, flat-bottomed 96 well plate, and treated for 15 hr with vehicle, 10 μM TMP or 1 μg/mL dox. 2 hr before cell metabolic activity was assessed, Tg (10 μM) was added to untreated cells. Cell metabolic activity was measured using the 7-hydroxy-3H-phenoxazin-3-one 10-oxide (resazurin) assay, which reports on mitochondrial redox potential. Cells were incubated with a final concentration of 50 μM resazurin for 2 hr at 37°C, The fluorescence signal, which is proportional to cell metabolism and viability, was then measured (excitation wavelength 530 nm, emission wavelength 590 nm). Error bars indicate standard error from biological replicates (n = 3). ^∗∗ indicates p-value < 0.01.

We envisioned that destabilized domain (DD) technology (Figure 1A) (Banaszynski et al., 2006; Iwamoto et al., 2010) could be adapted to prepare a dose-dependent, ligand-regulated ATF6 transcription factor whose activity would be inducible to levels more consistent with those observed in human physiology. We fused a destabilized variant of E. coli dihydrofolate reductase (DHFR) to the N terminus of ATF6 via a short Gly-Ser linker. The poorly folded DHFR domain directs the entire constitutively expressed DHFR.ATF6 fusion protein to rapid proteasomal degradation. Administration of the DHFR-specific pharmacologic chaperone, trimethoprim (TMP), stabilizes the folded DHFR conformation, increasing the initially poorly populated folded DHFR population, attenuating proteasomal degradation, and inducing the ATF6 transcriptional program ( Figure 1A).

Figure 1. Orthogonal, Ligand-Dependent Control of XBP1s and ATF6 Transcriptional Activity

(A) Model illustrating the TMP-mediated, posttranslational regulation of DHFR.ATF6.

(B) Immunoblot of nuclear (top) and postnuclear (bottom) fractions from HEK293T-REx cells expressing DHFR.YFP or DHFR.ATF6 treated 12 hr with TMP (10 μM). The immunoblot of matrin-3 shows the efficiency of the nuclear extraction.

(C) qPCR analysis of Hyou1, HerpUD, and Erdj4 in HEK293T-REx cells expressing DHFR.YFP or DHFR.ATF6 following a 12 hr treatment with TMP (10 μM) or a 6 hr treatment with Tm (10 μg/ml). qPCR data are reported relative to vehicle-treated cells expressing DHFR.YFP. qPCR data are reported as the mean ± 95% confidence interval.

(D) TMP dose dependence of HerpUD upregulation in HEK293T-REx cells expressing DHFR.ATF6 (12 hr treatments with TMP). qPCR data are reported as the mean ± 95% confidence interval.

(E) qPCR analysis of the ATF6 target gene BiP in HepG2, Huh7, or primary fibroblast cells transiently transduced with DHFR.YFP- or DHFR.ATF6-expressing adenoviruses and treated for 12 hr with 100 μM TMP or vehicle. qPCR data are reported relative to the corresponding vehicle-treated cells. qPCR data are reported as the mean ± 95% confidence interval.

(F) qPCR analysis of BiP and Erdj4 in HEK293^DAX cells following a 12 hr activation of XBP1s (dox; 1 μg/ml), DHFR.ATF6 (TMP; 10 μM), or both. qPCR data are reported relative to vehicle-treated HEK293^DYG cells. qPCR data are reported as the mean ± 95% confidence interval.

See also Figure S1 and Table S4.

The addition of TMP stabilizes DHFR.ATF6 in nuclear fractions isolated from HEK293T-REx cells expressing DHFR.ATF6 (Figure 1B). DHFR.ATF6 is not detected in the absence of TMP. Furthermore, TMP induces expression of the ATF6 target genes HerpUD and Hyou1 ( Adachi et al., 2008) in cells expressing DHFR.ATF6 to levels consistent with those observed following global UPR-dependent activation induced by Tm ( Figure 1C). We observe no increased expression of these genes in untreated cells expressing DHFR.ATF6 or TMP-treated cells expressing DHFR.YFP. The TMP-dependent activation of DHFR.ATF6 is rapid, causing significant upregulation of HerpUD in <2 hr ( Figure S1C). Importantly, TMP treatment does not induce expression of the XBP1s-selective target gene Erdj4 ( Lee et al., 2003) (Figure 1C). Increasing concentrations of TMP reveal a linear dose-dependent upregulation of ATF6 target genes, demonstrating a significant dynamic range for activation of DHFR.ATF6 by TMP ( Figure 1D). Because DHFR.ATF6 is a single gene product, it similarly enables the straightforward, ligand-dependent activation of the ATF6 transcriptional program at physiologic levels in a wide variety of other cellular model systems ( Figure 1E).

In order to activate both XBP1s and ATF6 in the same cell, we incorporated DHFR.ATF6 and tet-inducible XBP1s into a HEK293T-REx cell line stably expressing the tet repressor. Selection of a single colony resulted in the HEK293^DAX cell line in which XBP1s is induced by dox, and DHFR.ATF6 is activated by TMP (TMP-dependent DHFR.ATF6 activation in HEK293^DAX cells will henceforth be referred to as ATF6 activation for simplicity). We confirmed ligand-dependent regulation of XBP1s and ATF6 by immunoblotting (Figure S1D). qPCR analysis of HEK293^DAX cells demonstrates the orthogonal, ligand-dependent activation of the XBP1s and/or ATF6 transcriptional programs (Figures 1F and S1E) (Lee et al., 2003). An analogous HEK293^DYG control cell line expressing tet-inducible EGFP and DHFR.YFP was also prepared as a control (Figure S1D).

The addition of activating ligands to HEK293^DAX cells neither alters the incorporation of [³⁵S]-labeled methionine into the cellular proteome (Figures S1F and S1G) nor increases eIF2α phosphorylation (Figure S1H), demonstrating that selective XBP1s and/or ATF6 activation within the physiologically relevant regime does not cause PERK-mediated translational attenuation through stress-induced global UPR activation. Independent activation of XBP1s or ATF6 also does not significantly reduce cellular viability (unlike global UPR activators such as Tm or Tg; Figure S1I). Thus, HEK293^DAX cells enable orthogonal control of the transcriptional programs regulated by XBP1s and ATF6 in the same cell independent of stress.

The activation of XBP1s or ATF6 results in the upregulation of overlapping but divergent gene sets (Figure 2A), reflecting two distinct ER proteostasis environments accessible by activating these transcription factors independently. The transcriptional targets induced by XBP1s or ATF6 largely overlap with those previously identified by Adachi et al. (2008), Lee et al. (2003), Okada et al. (2002), andYamamoto et al., 2004 and Yamamoto et al., 2007. Interestingly, activating both XBP1s and ATF6 affords a third, previously inaccessible, ER proteostasis environment that is not simply the sum of the transcriptional consequences of activating XBP1s or ATF6 independently. This third ER proteostasis environment includes genes upregulated to similar levels by activating either XBP1s or ATF6 in comparison to the combination (Figure 2B, red and blue, respectively). In addition, 31 genes display cooperative upregulation owing to combined XBP1s and ATF6 activation (Figure 2B, green). We have validated the cooperative induction of several of these genes by qPCR (Figure 2C). This cooperative induction likely reflects the binding of both XBP1s and ATF6 to promoter regions or the preferential binding of XBP1s/ATF6 heterodimers to select promoters (Yamamoto et al., 2007) and represents a unique transcriptional profile only accessible by our ability to activate both XBP1s and ATF6 in the same cell independent of stress.

http://ars.els-cdn.com/content/image/1-s2.0-S2211124713001319-gr2.jpg

Integration of Transcriptomics and Proteomics Reveals Three Distinct ER Proteostasis Environments Accessible upon Activation of XBP1s and/or ATF6

Our transcriptional and proteomic profiling of HEK293^DAX cells reveals how the composition of the ER proteostasis network is differentially remodeled by activation of the XBP1s and/or ATF6 transcriptional programs (Figure 3). Consistent with the IRE1-XBP1s signaling cascade being the only UPR pathway conserved from yeast to humans, XBP1s activation has a broader impact on the composition of ER proteostasis pathways than does ATF6. XBP1s activation upregulates entire ER proteostasis pathways, including those involved in ER protein import, N-linked glycosylation, and anterograde/retrograde vesicular trafficking (Figure 3, red). The induction of these pathways is similarly observed by enrichment analysis (Table S2). In contrast, although ATF6 is responsible for upregulating only a select subset of ER proteostasis network proteins, these ATF6-selective targets represent critical hub proteins in the ER proteostasis network, including BiP, Sel1L, and calreticulin (Figure 3, blue).

http://ars.els-cdn.com/content/image/1-s2.0-S2211124713001319-gr3.jpg

Figure 3. Predictive Pathway Analysis for Stress-Independent XBP1s- and/or ATF6-Mediated Remodeling of the ER Proteostasis Network

Cartoon depicting the impact of activating XBP1s, ATF6, or both XBP1s and ATF6 on the composition of ER proteostasis pathways obtained by integrating transcriptional, proteomic, and biochemical results. XBP1s (red) and ATF6 (blue)-selective genes are genes where activating either XBP1s (but not ATF6) or ATF6 (but not XBP1s) independently results in >75% of the induction observed when both XBP1s and ATF6 are activated (“max induction”). Genes induced >75% of the max induction by activating XBP1s in isolation and induced >75% of the max induction by activating ATF6 in isolation (i.e., lacking selectivity) are colored purple. Genes cooperatively induced >1.33-fold upon activation of both XBP1s and ATF6 relative to the activation of either transcription factor alone are colored green. Plain type indicates results from array data. Italicized type indicates results from proteomics data. Underlined type indicates results confirmed at both the transcript and the protein levels. Thresholds for transcriptional analyses were set at a FDR of <0.05. Thresholds for proteomic analyses were set at a FDR of 0.1.

Some proteins are upregulated to similar levels by activating XBP1s in isolation, ATF6 in isolation, or both XBP1s and ATF6 (Figure 3, purple). Alternatively, a number of proteins primarily involved in ER quality control and degradation are cooperatively upregulated when both XBP1s and ATF6 are activated (Figure 3, green). These results are consistent with the biological pathways predicted to be transcriptionally enhanced by XBP1s:ATF6 heterodimers (Yamamoto et al., 2007) and clearly demonstrate that the impact of the combined activation of XBP1s and ATF6 on the composition of the ER proteostasis network is greater than the sum of activating XBP1s or ATF6 individually.

http://ars.els-cdn.com/content/image/1-s2.0-S2211124713001319-gr4.jpg

Figure 4. XBP1s and/or ATF6 Activation Differentially Influences the Degradation of NHK-A1AT and NHK-A1AT^QQQ

(A) Representative autoradiogram of [³⁵S]-labeled NHK-A1AT immunopurified from transfected HEK293^DAX cells following a 15 hr preactivation of XBP1s (dox; 1 μg/ml), DHFR.ATF6 (TMP; 10 μM), or both. The metabolic-labeling protocol employed is shown.

(B) Quantification of autoradiograms in (A) monitoring the degradation of [³⁵S]-labeled NHK-A1AT. The fraction of NHK-A1AT remaining was calculated by normalizing the recovered [³⁵S] signal to the total amount of labeling observed at 0 hr. Error bars represent SE from biological replicates (n = 18).

(C) Bar graph depicting the normalized fraction of NHK-A1AT remaining at 3 hr calculated as in (B).

(D) Representative autoradiogram of [³⁵S]-labeled NHK-A1AT^QQQ immunopurified from transfected HEK293^DAX cells following a 15 hr preactivation of XBP1s (dox; 1 μg/ml), DHFR.ATF6 (TMP; 10 μM), or both. The metabolic-labeling protocol employed is shown.

(E) Quantification of autoradiograms in (D) monitoring the degradation of [³⁵S]-labeled NHK-A1AT^QQQ. The fraction of NHK-A1AT^QQQ remaining was calculated as in (B). Error bars represent SE from biological replicates (n = 6).

(F) Bar graph depicting the normalized fraction of NHK-A1AT^QQQ remaining at 4.5 hr calculated as in (B).

^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S3.

http://ars.els-cdn.com/content/image/1-s2.0-S2211124713001319-figs3.jpg

Figure S3. Influence of Dox and/or TMP Treatment on NHK-A1AT and NHK-A1AT^QQQ Degradation in HEK293^DAX and HEK293^DYG Cells, Related to Figure 4

(A) Quantification of immunoblots of lysates prepared from HEK293^DAX cells transfected with NHK-A1AT and treated for 15 hr with vehicle or TMP (10 μM; activates DHFR.ATF6). Cycloheximide (CHX, 50 μg/mL) was applied for the indicated time prior to harvest. Total NHK-A1AT at each CHX time point was normalized to the amount of NHK-A1AT observed in the absence of CHX. Error bars indicate the standard error from biological replicates (n = 4).

(B) Quantification of autoradiograms monitoring the degradation of [³⁵S]-labeled, NHK-A1AT in transfected HEK293^DYG cells following a 15 hr induction of GFP (dox; 1 μg/mL), DHFR.YFP (TMP; 10 μM), or both. The metabolic labeling protocol employed is identical to that used in Figure 4A. Fraction remaining was calculated as in Figure 4B. Error bars indicate the standard error from biological replicates (n = 3).

(C) Quantification of autoradiograms monitoring the degradation of [³⁵S]-labeled NHK-A1AT^QQQ in transfected HEK293^DYG cells following a 15 hr induction of GFP (dox; 1 μg/mL), DHFR.YFP (TMP; 10 μM) or both. The metabolic labeling protocol employed is identical to that used in Figure 4D. Fraction remaining was calculated as in Figure 4E. Error bars indicate the standard error from biological replicates (n = 3).

http://ars.els-cdn.com/content/image/1-s2.0-S2211124713001319-gr5.jpg

Figure 5. ATF6 Activation Selectively Attenuates the Secretion of Amyloidogenic TTR

(A) Autoradiogram of [³⁵S]-labeled ^FTTTR^A25T immunopurified from media and lysates collected from transfected HEK293^DAX cells following a 15 hr preactivation of XBP1s (dox; 1 μg/ml), DHFR.ATF6 (TMP; 10 μM), or both. The metabolic-labeling protocol employed is shown.

(B) Quantification of autoradiograms as shown in (A). Fraction secreted was calculated as previously described by Sekijima et al. (2005). Error bars represent SE from biological replicates (n = 4).

(C) Graph depicting the normalized fraction secreted of [³⁵S]-labeled ^FTTTR^WT (white bars) or ^FTTTR^A25T (orange bars) at 4 hr following a 15 hr preactivation of DHFR.ATF6 (TMP; 10 μM) in HEK293^DAX cells. Error bars represent SE from biological replicates (n = 8 for ^FTTTR^A25T, and n = 9 for ^FTTTR^WT).

(D) Graph depicting the total [³⁵S]-labeled ^FTTTR^A25T remaining in HEK293^DAX cells (combined media and lysate protein levels as in A). The fraction remaining was calculated as reported previously by Sekijima et al. (2005). Error bars represent SE from biological replicates (n = 8).

(E) Graph depicting the normalized fraction secreted of [³⁵S]-labeled ^FTTTR^A25T (orange bars) at 4 hr following preactivation of DHFR.ATF6 (TMP; 10 μM; 15 hr) in the presence or absence of tafamidis (10 μM; 15 hr) in HEK293^DAX cells. Error bars represent SE from biological replicates (n = 4).

(F) Bar graph depicting the normalized fraction secreted of ^FTTTR^A25T and endogenous TTR^WT at 4 hr following a 13 hr pretreatment with TMP (100 μM) in HepG2 cells stably expressing DHFR.ATF6. Error bars represent SE from biological replicates (n = 4).

(G) Bar graph depicting the normalized fraction secreted of [³⁵S]-labeled ^FTTTR^D18G at 4 hr following a 15 hr pretreatment with TMP (10 μM) from HEK293^DAX cells transfected with both ^FTTTR^D18G and TTR^WT. Error bars represent SE from biological replicates (n = 4).

(H) Immunoblot of α-FLAG M1 FTTTRA25T immunoisolations from DSP-crosslinked lysates prepared from HEK293^DAX cells expressing ^FTTTR^A25T following 15 hr activation of XBP1s (dox; 1 μg/ml), DHFR.ATF6 (TMP; 10 μM), or both. HEK293^DAX cells expressing GFP are shown as a negative control (Mock). The KDEL immunoblot shows BiP.

^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.005. See also Figure S4.

http://ars.els-cdn.com/content/image/1-s2.0-S2211124713001319-figs4.jpg

Figure S4.  ATF6 Activation Selectively Attenuates the Secretion of Amyloidogenic Transthyretin, Related to Figure 5

(A) Quantification of immunoblots measuring ^FTTTR^A25T and ^FTTTR^WT secreted into the media from transfected HEK293^DAX cells during a 15 hr activation of XBP1s (dox; 1 μg/mL), DHFR.ATF6 (TMP; 10 μM), or both. Error bars indicate standard error from biologic replicates (n = 10 for ^FTTTR^A25T and n = 7 for ^FTTTR^WT).

(B) Quantification of autoradiograms monitoring [³⁵S]-labeled ^FTTTR^A25T secreted from transfected HEK293^DYG cells following a 15 hr induction of GFP (dox; 1 μg/mL), DHFR.YFP (TMP; 10 μM) or both. The metabolic labeling protocol employed is identical to that used in Figure 5A. Fraction secreted was calculated as in Figure 5B. Error bars indicate the standard error from biological replicates (n = 3).

(C) Bar graph depicting the normalized fraction secreted of [³⁵S]-labeled ^FTTTR^A25T and ^FTTTR^WT at t = 4 hr following a 15 hr preactivation of DHFR.ATF6 (TMP; 10 μM) or global UPR activation by treatment with Tg (500 nM). Normalized fraction secreted was calculated as in Figure 5C and a representative autoradiogram is shown. The error bars represent standard error from biological replicates (n = 2).

(D) Concentration-dependent kinetics of recombinant TTR^A25T aggregation at pH 6.0 as monitored by turbidity at 400 nm. TTR was purified by gel filtration immediately prior to use. 150 μl of TTR in 10 mM phosphate buffer pH 7.0, 100 mM KCl, 1 mM EDTA, was added to 750 μl of 0.1 M citrate-phosphate buffer pH 6.0 in a plastic cuvette to yield final concentrations as indicated. The samples were incubated at 37°C without stirring, and agitated prior to transmittance measurement. Higher concentrations yield a higher extent and rate of TTR aggregation. Error bars indicate standard error from biologic replicates (n = 3).

(E) Representative immunoblot depicting detergent-soluble and insoluble ^FTTTR^A25T isolated from HEK293^DAX cells treated for 15 hr with vehicle or TMP (10 μM) and separated by SDS-PAGE. Detergent-insoluble protein was recovered by incubating the washed pellet from RIPA-lysed cells in 8 M urea in 50 mM Tris pH 8.0 at 4°C overnight, followed by shearing, dilution in RIPA, and centrifugation at 16000 × g for 15 min. The error bars represent standard error from biological replicates (n = 3).

(F) Representative autoradiogram and bar graph depicting the total [³⁵S]-labeled ^FTTTR^D18G remaining in HEK293^DAX cells following 15 hr TMP (10 μM) pretreatment. The fraction remaining of ^FTTTR^D18G following a 90 min chase was calculated as inFigure 5D. The error bars represent standard error from biological replicates (n = 4). ^∗indicates p-value < 0.05.

(G) Representative immunoblot and quantification of ^FTTTR^D18G protein levels in HEK293^DAX cells following a 15 hr incubation with TMP (10 μM) and/or MG-132 (10 μM). The error bars represent standard error from biological replicates (n = 9). ^∗∗indicates a p-value < 0.05

(H) Representative immunoblot and quantification of media collected from equal numbers of HEK293^DAX cells transiently transfected with ^FTTTR^A25T and pretreated with TMP (10 μM) and/or tafamidis (10 μM) for 15 hr, as indicated. The error bars represent standard error from biological replicates (n = 4). ^∗∗∗ indicates p-value < 0.01.

(I) Representative autoradiogram of [³⁵S]-labeled ^FTTTR^A25T in the media and lysates of HEK293^DAX cells treated with TMP (10 μM) and/or tafamidis (10 μM) for 15 hr, as indicated, and used to prepare Figure 5E. Cells were metabolically labeled using an identical protocol to that shown in Figure 5A. The bar graph shows the quantification of normalized total [³⁵S]-labeled ^FTTTR^A25Tremaining from autoradiograms as shown above. Error bars represent standard error from biological replicates (n = 4).

(J) qPCR analysis of clonal HepG2T-REx cells stably expressing DHFR.ATF6 treated overnight with vehicle or 100 μM TMP. TMP treatment leads to substantial expression of the ATF6 target BiP, but not the XBP1s target ERdj4, demonstrating the selective activation of the ATF6 transcriptional program in these cells. qPCR data are reported as the mean ± 95% confidence interval.

(K) Representative autoradiogram of [³⁵S]-labeled ^FTTTR^A25T and endogenous TTR^WT isolated from HepG2T-REx cells stably-expressing DHFR.ATF6 and pretreated with TMP (100 μM; 15 h). The cells were metabolically labeled using an identical approach to that shown in Figure 5A. ^FTTTR^A25T was immunopurified using the anti-Flag M1 antibody. Endogenous TTR^WT was immunopurified using an anti-TTR rabbit polyclonal antibody (Sekijima et al., 2005).

(L) Representative autoradiogram of [³⁵S]-labeled ^FTTTR^D18G isolated from HEK293^DAX cells transiently transfected with^FTTTR^D18G and TTR^WT. Cells were treated with TMP (10 μM; 14 h), as indicated. TTR was immunopurified from these cells using either the anti-Flag M1 antibody that selectively recognizes ^FTTTR^D18G (n = 2) or the anti-TTR rabbit polyclonal antibody, which immunopurifies both ^FTTTR^D18G and TTR^WT (n = 2). The arrows indicate ^FTTTR^D18G and TTR^WT, which are separated by SDS-PAGE. It is important to note that anti-Flag M1 immunoisolation of ^FTTTR^D18G co-purifies TTR^WT, reflecting efficient formation of heterotetrameric TTR containing both ^FTTTR^D18G and TTR^WT subunits.

Herein, we establish methodology that allows for the orthogonal, small molecule-mediated regulation of the UPR-associated transcription factors XBP1s and/or ATF6 in the same cell independent of stress. We employ our methodology to reveal the molecular composition of the three distinct ER proteostasis environments accessible by activating the XBP1s and/or ATF6 transcriptional programs. Furthermore, we show that selectively activating XBP1s and/or ATF6 differentially influences the ER partitioning of destabilized protein variants between folding and trafficking versus degradation. Our results provide molecular insights into how the XBP1s and/or ATF6 transcriptional programs remodel the ER proteostasis environment and demonstrate the potential to influence the ER proteostasis of destabilized protein variants via physiologic levels of arm-selective UPR activation.

Our quantitative transcriptional and proteomic profiling of HEK293^DAX cells provides an experimentally validated, conceptual framework to identify specific ER proteins and/or pathways that can be adapted to alter the fate of disease-associated ER client proteins (Figure 3). Critical pathways directly responsible for the partitioning of ER client proteins between folding and trafficking versus degradation are differentially impacted by XBP1s and/or ATF6 activation. For example, the levels of BiP and BiP cochaperones, which are known to modulate folding versus degradation decisions of client proteins in the ER lumen, are differentially influenced by XBP1s and/or ATF6 activation (Figure 3) (Kampinga and Craig, 2010). Considering the importance of BiP cochaperones in defining BiP function, these findings suggest that the fates of BiP clients are distinctly influenced by XBP1s and ATF6 activation. Consistent with this prediction, we show that BiP and HYOU1 have increased association with TTR^A25T only when ATF6 is activated, even though HYOU1 is also upregulated by XBP1s (Table 1).

Analogously, XBP1s- or ATF6-dependent remodeling of ER client protein folding pathways can be deconvoluted from our bioinformatic characterization of HEK293^DAX cells. For example, XBP1s-selective transcriptional upregulation of the ERAD-associated proteins ERMan1, ERdj5, and EDEM-3 may explain the enhanced degradation of NHK-A1AT upon XBP1s activation because overexpression of these three proteins enhances NHK-A1AT ERAD (Hosokawa et al., 2003, 2006; Ushioda et al., 2008). Alternatively, ATF6 selectively enhances the expression of the ERAD-associated protein Sel1L, which when overexpressed, accelerates degradation of the nonglycosylated protein NHK-A1AT^QQQ (Iida et al., 2011). Thus, our transcriptional and proteomic profiles of cells remodeled by XBP1s and/or ATF6 activation enable hypothesis generation to dissect the contributions of ER proteostasis proteins and/or pathways involved in altering the folding, trafficking, or degradation of ER client proteins.

We used HEK293^DAX cells to explore the potential for ER proteostasis environments accessed through arm-selective UPR activation to reduce the secretion of a destabilized, amyloidogenic TTR variant. We found that ATF6 activation selectively reduces secretion of the destabilized, aggregation-prone TTR^A25T, but not the secretion of TTR^WT or the global endogenous secreted proteome. Previously, we and others have demonstrated that the efficient secretion of destabilized TTR variants through the hepatic secretory pathway is a contributing factor to the extracellular aggregation and distal deposition of TTR as amyloid in the pathology of numerous TTR amyloid diseases (Hammarström et al., 2002, 2003a; Holmgren et al., 1993; Sekijima et al., 2003, 2005; Suhr et al., 2000; Susuki et al., 2009; Tan et al., 1995). Thus, our discovery that ATF6-dependent remodeling of the ER proteostasis environment selectively reduces secretion of destabilized TTR^A25T reveals a potential mechanism to attenuate the secretion and subsequent pathologic extracellular aggregation of the >100 destabilized TTR variants involved in TTR amyloid diseases (Sekijima et al., 2008). Furthermore, the establishment and characterization of the DHFR.ATF6 construct (which we demonstrate can be rapidly incorporated into any cellular model) and the HEK293^DAX cell line provide invaluable resources to evaluate the functional impact of arm-selective UPR activation to physiologic levels on the aberrant ER proteostasis of destabilized mutant proteins involved in the pathology of many other protein misfolding-related diseases. Consistent with the potential to correct pathologic imbalances in destabilized protein ER proteostasis, recent studies that employ global UPR activation using toxic small molecules or the unregulated overexpression of XBP1s or ATF6 have suggested that remodeling ER proteostasis pathways through arm-selective UPR activation could correct the aberrant ER proteostasis of pathologic destabilized protein mutants involved in protein misfolding diseases (Chiang et al., 2012; Mu et al., 2008; Smith et al., 2011a).

Finally, we note that despite clear functional roles for XBP1s and ATF6 in adapting the composition of ER proteostasis pathways highlighted herein, organisms have distinct dependencies on these transcription factors. XBP1s is critical for biological processes including plasma cell differentiation and development (XBP1s knockout mice are not viable; Reimold et al., 2000). Alternatively, mice lacking ATF6α, the primary ATF6 homolog involved in UPR-dependent remodeling of the ER proteostasis environment, develop normally, although deletion of both mammalian ATF6 homologs, ATF6α and ATF6β, is embryonic lethal (Adachi et al., 2008; Wu et al., 2007; Yamamoto et al., 2007). Thus, whereas XBP1s is required for organismal development, our results suggest that functional roles for ATF6 in remodeling the ER proteostasis environment are adaptive—adjusting ER proteostasis capacity to match demand under conditions of cellular or organismal stress. Therefore, modulation of ATF6 may provide a unique opportunity to sensitively “tune” the ER proteostasis environment without globally influencing the folding, trafficking, or degradation of the secreted proteome.

In summary, we show that the application of DD methodology to control ATF6 transcriptional activity provides an experimental strategy to characterize the impact of stress-independent activation of XBP1s and/or ATF6 on ER proteostasis pathway composition and ER function. Adapting the underlying biology of the proteostasis network through the activation of specific UPR transcriptional programs reveals emergent functions of the proteostasis network, including a window to alter the ER proteostasis of destabilized mutant proteins without significantly affecting the proteostasis of the vast majority of the endogenous, wild-type proteome. Our transcriptional, proteomic, and functional characterization of the ER proteostasis environments accessible by activating XBP1s and/or ATF6 in a single cell validates targeting specific pathways within the proteostasis network as a potential therapeutic approach for adapting the aberrant ER proteostasis associated with numerous protein misfolding diseases, strongly motivating the development of arm-selective small molecule activators of the UPR.

2.1.6.4 Modeling general proteostasis – proteome balance in health and disease

Roth, D. M., Balch, William E.
Current Opinion in Cell Biology 2011; 23(2): 126-134
http://vivo.scripps.edu/individual/endnote128101
http://dx.doi.org:/10.1016/j.ceb.2010.11.001

Protein function is generated and maintained by the proteostasis network (PN) (Balch et al. (2008) Science, 319:916). The PN is a modular, yet integrated system unique to each cell type that is sensitive to signaling pathways that direct development and aging, and respond to folding stress. Mismanagement of protein folding and function triggered by genetic, epigenetic and environmental causes poses a major challenge to human health and lifespan. Herein, we address the impact of proteostasis defined by the FoldFx model on our understanding of protein folding and function in biology. FoldFx describes how general proteostasis control (GPC) enables the polypeptide chain sequence to achieve functional balance in the context of the cellular proteome. By linking together the chemical and energetic properties of the protein fold with the composition of the PN we discuss the principle of the proteostasis boundary (PB) as a key component of GPC. The curved surface of the PB observed in 3-dimensional space suggests that the polypeptide chain sequence and the PN operate as an evolutionarily conserved functional unit to generate and sustain protein dynamics required for biology. Modeling general proteostasis provides a rational basis for tackling some of the most challenging diseases facing mankind in the 21st century.

Newly synthesized proteins must fold into a unique three-dimensional (3D) structure to become functionally active. We now appreciate that all proteins likely require the assistance of the “proteostasis network” (PN) to generate and maintain function. The PN comprises not less than a 1000 factors that regulate protein synthesis, folding, function, and degradation [1–3] (FIG. 1). These form the Yin and Yang environment that promotes what we have referred to recently as proteome balance in health [4]. Importantly, the composition of the PN is dynamically regulated by a variety of signaling pathways [5,6], and in response to developmental cues, genetic changes, epigenetic marks, environmental stress and aging; challenges that all cells encounter during their lifespan to maintain normal organismal physiology [3,7,8]. Of importance, is that a very large number of inherited diseases are caused by mutations in the sequence of a polypeptide chain, leading to loss of protein stability, misfolding and disease. While genetic changes often severely challenge the dynamics of proteostasis to retain proteome balance, the response of the PN to mutation can significantly contribute to organismal evolution [9]. Given the multiplicity of cellular PN stress responses, it is not surprising that the PN has evolved to be highly versatile in its capacity to maintain proteome balance. Herein, we discuss the role of protein energetics and kinetics in generating and maintaining proteome balance through the activity of the PN. We explore how modeling of proteostasis opens new avenues to the management of human health and disease.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3077458/bin/nihms258064f1.jpg

The PN

Shown are the interactions that comprise the PN, the composition of which is responsible for generating and maintaining the biological protein fold. Components comprising the PN outlined in the inner-most layer (in blue font) involve the synthesis module, the folding/unfolding module, and the degradation module (the GPC triad). A second layer shows the signaling transcriptional pathways (in green font) that influence the level and activity of the triad found in the innermost layer. The third layer (in red font) includes modifiers that influence and/or integrate the activities defined by the second and first layers. Modifiers and signaling pathways from both cell autonomous and cell non-autonomous origin. Modified figure reproduced with permission from Elsevier Press [2].

While small, single domain proteins can fold efficiently in the test tube, we now appreciate that these and multi-domain proteins generated in the crowded environment of the cell often fail to do so. This is because there are energy barriers in the landscape model (Fig. 2a, peaks and troughs) [10,11], that dictate the kinetics and thermodynamics of folding intermediates in the path(s) required to achieve the native folded state (Fig. 2a, red circles and ‘N’ in figure). The native state here is defined as the state with the lowest energy- which may or may not be the biologically important state [4]. To avoid off-pathway misfolding, degradation and/or aggregation that can occur during progress through intermediate folding steps in biology (Fig. 2a, red circles and white arrows), PN components are thought to interact with the polypeptide chain to generate and protect biological function (Fig. 2a, black and gray arrows) [1,2].

Figure 2 Coupling of the folding energy landscape with the PN

(a) Illustrated is a bumpy energy landscape funnel (http://www.dillgroup.ucsf.edu/) in which an unfolded protein proceeds along various intermediate steps (red circles) that can pose energetic barriers to achieve the native state (N) at the base of the funnel. The white arrows indicate potential pathways through various folding intermediates that the nascent protein may take to reach the native state. The solid black and blue (synthesis and folding modules) and the dashed gray (degradative module) arrows illustrate how different components of the PN may influence pathway choice. The dashed oval indicates that all intermediates are potential steps in which a protein can be targeted for degradation. Hsp90 is thought to principally facilitate late folding events (solid blue lines). (b) The energy landscape (a slice through the funnel illustrated in panel a) illustrates the central role of the ATPase cycle in synthesis, folding and degradative modules in managing the biology of the protein structure in the cell to achieve function. By coupling the energy of ATP hydrolysis (X-axis) by PN ATPase machines with the energetics imparted in the chemistry of the amino acid sequence of the polypeptide chain (Y-axis), PN ATPases maintain the protein fold in a dynamic state- essential for biology. Right side of panel illustrates energy barriers necessary to achieve a functional fold (black arrow) relative to the energetics associated with misfolding (red and orange arrows). The additional energetic demands challenging GPC through misfolding (red line) or aggregation (short orange line) are illustrated. Purple arrows indicate potential steps for GPC to alter folding kinetics and energetics to promote proteome balance and cell health. Abbreviations: sHsp (small heat shock proteins); TRiC (TCP1-ring complex)

The PN is an integrated system consisting of chaperones, folding enzymes, degradation components, and regulatory pathways that control the composition and concentration of the general proteostasis system [5,12] (Fig. 1). PN components include the molecular chaperones/co-chaperones belonging to the Hsc/Hsp (Hsc/p) 70 and 90 families [13], the GroEL/TCP1-ring complex (TRiC)/chaperonin family of folding machines [14], tetratricopeptide repeat (TRP)-domain containing proteins, proteins that modulate oxidative folding (e.g., protein disulfide isomerases [15]), and degradation components comprising both the cytosolic ubiquitin-proteasome and membrane-linked autophagy-lysosome systems [6,16] (Fig. 1). Some PN components are highly abundant (e.g., ribosome, Hsc/p70-Hsp90, proteasome/lysosome) and provide a cellular ‘buffer’ for synthesis, folding and/or degradation [12]. Most others function as specialists, either alone or together with the Hsc/p70 and Hsp90 systems, to synthesize and/or maintain specific folds for the highly evolved dynamic functions dictating extant organismal physiology. Regulation of the composition of the PN occurs through a number of signaling pathways including the unfolded protein response (UPR) [17], the heat shock response (HSR) [7,18,19], oxidative stress pathways [20], and growth factor and diet sensitive pathways [1–3], among others. A simplistic view of the PN is that components directing synthesis, (un)folding, and degradation could be considered as a triad of modules (Fig. 1, dark black lines). Triad modules recognize the chemical properties of polypeptide folding intermediates (Fig. 2), yet are integrated by the overall composition of the PN to maintain normal biological function.

It is important to recognize that the PN is unique for each type of cell and the numerous subcellular compartments within a eukaryotic cell. These environments change differentially during development, aging and in response to physiological stress [2,21]. Moreover, the PN is constantly challenged by changes in the composition of the amino acid pool, metabolites/co-factors, ion balance, genetic-epigenetic-environmental triggers, and viral/bacterial pathogens. These factors not only affect the inherited capacity of the proteostasis program, but are readily sensed by the above regulatory signaling pathways that attempt to rebalance the proteome to preserve healthspan [1,3,4]. Thus, the PN is dynamically tuned to cellular function as prescribed by cell autonomous processes and cell non-autonomous signals that optimize folding for function in complex organismal environments [7,19].

Role of ATP in maintaining proteome balance

The capacity of the PN to maintain proteome balance in the cytosol and exocytic/endocytic trafficking compartments, that is, the Yin-Yang relationship between generating and maintaining a functional fold or targeting a protein for degradation [4], is referred here as general proteostasis control (GPC). GPC emphasizes that function of a polypeptide chain is tightly linked to the local composition of the PN triad- the environment being the ultimate arbitrator of biological folding for function. What is a wild-type protein fold in one PN environment becomes a ‘mutant’ in another and can be removed and/or challenge the health status of the cell. The former is evident in the cyclical stability of proteins during cell cycle, or the transient stability observed in developmental programs. The latter is observed in, for example, numerous sporadic aggregation diseases, type II diabetes and cancer [2].

Both subtle and global challenges to protein folding energetics directly challenge the dynamics of the kinetics of protein folding and its thermodynamic stability. Thus, there is a close link between PN folding for function (Fig. 2a) and ATP-based proteostasis machines that manipulate the energy landscape dictated by the unique chemistries of amino acid sequence of each polypeptide chain (Fig. 2b). For example, during protein biogenesis, newly synthesized polypeptides are generated by the ribosome at a very high energy cost and in response to protein specific translational control programs. They emerge from the ribosome with exposed hydrophobic residues that are recognized by the folding module of the PN to prevent protein aggregation. This first step of GPC faced by nascent proteins is often regulated by members of the Hsc/p70 family [12] and/or the TRiC/chaperonin ATPase machines [22]. While TRiC ATPases appear to be dedicated to folding, the Hsc/p70 ATPases function to either promote folding/assembly of newly synthesized proteins or direct ‘non-native’ polypeptides to degradation [23], serving as a key linker between the various PN modules (Fig. 1). Thus, the Hsc/p70 system plays a critical role in proteome balance in response to energetics of the polypeptide chain [4].

Hsc/p70 family of chaperones, utilize ATP-dependent cycles of client binding and release in response to a plethora of accessory proteins, called co-chaperones. In the case of Hsc/p70 these include nucleotide exchange factors (NEFs) composed by the Bag (BCL2 associated athanogene) family of proteins, which facilitate ADP release and ATP binding to promote Hsc/p70 client substrate release, and a large Hsp40/DnaJ family of co-chaperones that stimulate the ATPase activity of Hsc/p70 and stabilize protein client-chaperone interactions [24]. Thus, Hsc/p70 co-chaperones will not only fine-tune Hsc/p70 client substrate specificity but dictate the cellular fate of the protein client [13,25,26].

Proteins that interact with the Hsc/p70 arm of GPC are in many cases subject to a second level of maintenance by the Hsp90 system [27]. The Hsp90 ATPase machinery appears to recognize dynamic facets of more folded substrates to modulate their activity(s) (Fig. 2a, blue lines), [27,28] (www.picard.ch/downloads/Hsp90interactors.pdf). As is seen for the Hsc/p70 system, a unique collection of co-chaperones also regulate Hsp90 ATPase activity. Hsp90 co-chaperones include the ATPase activator Aha1 and the ATPase inhibitor p23, as well as Cdc37, HOP, protein-disulfide isomerases (PPIases), and a large family of TRP-domain containing proteins. Depending on the local activity/composition of the co-chaperone environment, Hsp90 can promote either protein stability or degradation of folding intermediates- dynamically altering the proteome balance [4] (Fig. 2b).

Like the folding GPC module (Fig. 1), the degradation module involving the proteasome and the autophagy-lysosome pathways are intensely ATP-dependent [29]. While many of the regulatory factors that control function of these degradative machines remain to be determined, client targeting to multiple degradative pathways is generally regulated by the ubiquitin/sumoylation system. Targeting for degradation utilizes a highly diverse (>300) set of client specific ligases that utilize the energy of ATP to prime polypeptide targets for destruction [23]. Thus, ATP-dependent cycling of synthesis, folding, and degradation modules provides an energetic link between the functional and degradation prone states of a target protein in biology.

While GPC has a universal high level of energy demand, it is important to realize that folding/function is highly compartmentalized. For example, the cytosol is a reducing environment maintaining folded proteins in the absence of disulfide bonding. In contrast, the endoplasmic reticulum (ER), the first step in the secretory pathway, is an oxidative environment where protein folding is driven by an evolutionarily related, but distinct set of luminal PN folding components. The folding module in the ER is tightly coupled by membrane translocation pathways to the reducing cytosolic proteasomal degradation module [23,30,31]. Recent studies [32] have shown that cytosolic GPC components important for generation of newly synthesized transmembrane polypeptides in the ER also modulate protein stability at the cell surface. Likewise, the lysosome not only handles internalized cargoes from the cell surface, but is a critical partner of the autophagosome pathway that engulfs a wide range of misfolded cytosolic proteins and dysfunctional organelles [29,33]. These results suggest the importance of as yet unknown cellular proteostasis sensors that unify and balance folding throughout the cell.

In summary, it is now clear that GPC may define energetic standards for each cell type that is linked in as yet to be determined ways by the activity of PN-linked ATPase machines. By coupling the chemistry of the polypeptide chain sequence and its associated folding energetics with the ATPase activity of PN modules (Fig. 2b), a biologically dynamic GPC standard generates the proper balance between the triad of synthesis, (un)folding, and degradation modules (Fig. 1). This standard defines the proteome balance in a healthy cell and its response to stress, disease, injury and aging programs.

Modeling proteostasis

An understanding of the rules guiding GPC to achieve protein function involves integrating the chemistry of the polypeptide sequence with the activity of PN components (Fig. 2). For this purpose, we applied Michaelis-Menten formalism in the FoldEx model to describe how the inherent chemistry and energetics of the polypeptide chain can be read and manipulated by the PN for proteins trafficking through the exocytic pathway [34]. The concepts stemming from the FoldEx model were extended to describe a more encompassing model of how folding energetics and the PN work together. We refer to this new model asFoldFx [2]. FoldFx is applicable to folding of proteins in all compartments of the cell and the extracellular space in response to the composition of the local PN (Fig. 1). In FoldFx, the operational goal of the triad of protein synthesis, (un)folding, and degradation modules through GPC is to achieve ‘function’.

A key feature of the FoldFx model which rigorously defines the activity of GPC is the concept of the ‘proteostasis boundary’ or PB [2] (Fig. 3a). The PB can be used to define the minimal energetic properties that a protein must have to achieve normal function in response to the local PN. The PB is best illustrated in a 3-dimentsional (3D) space diagram as a curved surface. The position of a protein in 3D is determined by its inherent folding kinetics, misfolding kinetics, and thermodynamics (Fig. 3a). The curved shape of the PB is dictated by the variable concentration of proteostasis components. These are, in turn, defined by the genetic, epigenetic, and intrinsic and extrinsic factors that regulate PN pathways and thereby tune the PN for specific client functionality. Beneath the boundary is a normal biological network, defined by nodes (the proteins) and edges (their links to other proteins within the network) (Fig. 3a). Each node is positioned according to its folding energetics (its unique folding and misfolding rate and stability). In a healthy cell, each node and its link (the edges in Fig. 3a) are embraced by the curved space of the PB, indicating that the PN is sufficient to maintain normal function (Fig. 3a). In disease, a node falls outside the embrace of the PB, resulting in misfolding, aggregation and/or degradation (Fig. 3b).

Fig 3. The proteostasis boundary (PB)

The position of each node (protein) relative to the PB (curved surface) responsible for biological function is defined by a protein’s folding properties: folding kinetics (Z-axis), misfolding kinetics (Y-axis) and thermodynamic stability (X-axis). Each line defines a physical or functional interaction between two proteins in the system. The location of the PB in 3D space is established by the composition of the PN and modulated by the GPC triad. (a) All of the nodes are within the PB boundary in a healthy cell. (b) Mutations or aberrant post-translational modifications can alter folding kinetics and energetics, making their corresponding nodes and edges fall outside (above the curved surface) of the PB. This space in the 3D plot does not support function of the energetically destabilized variant and can lead to either degradation (red node) or protein aggregation (black node). The loss of connectivity to proteins within the embrace of the PB can challenge the entire PN leading to cell, tissue, and organismal disease. Reproduced with permission from Elsevier Press [2].

Figure 4 Modeling the Yin-Yang of proteome balance in health, disease, and in response to proteostasis therapeutics

(Panel a) Proteome balance [4] in a healthy cell is determined by the composition of the synthesis and folding modules (FM) (the Yang on the left side of diagram) and degradative module (DM) (the Yin on the right side of diagram). GPC1 determines the position of the PB (the S-shaped curve) and healthspan. The dashed lines illustrate that misfolding and aging can challenge the position of the PB. (Panel b) Aging and unfolding move the PB to the left resulting in compromised proteostasis function (GPC2) and an unhealthy cell by triggering increased degradation and/or accumulation of protein aggregates. (Panel c) Biological signaling pathways including the HSR (HSF1 and IGF1-R/FOXO pathways), UPR, oxidative stress response (OSR), diet, IGF1-R and/or proteostasis targeted therapeutics can move the Yin-Yang balance defined by the PB to the right generating GPC3. GPC3 provides an environment that protects the cell from physiological stress, misfolding and aging, allowing the cell to return to GPC1.

Therapeutics in modeling of the GPC triad

Multiple lines of evidence suggest that protection to misfolding disease and aging can be boasted through multiple pathways that regulate the expression of PN components (e.g., HSR, IGF1-R signaling, diet restriction and pathways that protect against oxidative stress mentioned above) [1,3] (Fig. 1, Fig. 4c -GPC3). Modulation of components of the PN biologically by targeting individual PN components with siRNA implicated in these pathways can dramatically affect the outcome of disease. For example, depletion of Aha1 (an Hsp90 ATPase regulator) or E3ligase RMA1/CHIP, partially restores functionality in cystic fibrosis (CF) models [40,41]. These represent changes in distinct arms of the Yin-Yang balancing act, involving both cytosolic and exocytic/endocytic membrane trafficking pathways managed by the GPC triad (Fig. 4c). Moreover, overexpression of Hsc/p70 and its co-chaperones has been shown to reduce aggregation and toxicity in models of neurodegenerative/misfolding diseases, such as AD [42], prion disease [43], and HD [24]. Overexpression of Hsp40 reduces polyQ inclusion formation and toxicity [24] while the co-chaperone CHIP suppresses the toxicity of α-synuclein and polyQ proteins [44], and reduces accumulation of tau and Aβ [45], possibly through removal of aggregated misfolded proteins via the proteasome. The cofactor Bag1 also has been shown to reduce toxicity caused by polyQ Huntington aggregates [46]. Indeed, the FoldFx model predicts that bolstering the operation of the PN along specific axis’s is likely to not only improve healthspan, but also simultaneously improve longevity- the ultimate test of a therapeutic approach [1,2,7,8,47].

An increasing number of small molecules are now recognized to impact these pathways and provide protective function to human disease [4,48,49] (Fig. 4c). For example, the inhibition of the proteasome arrests myeloma disease [50], kinetic stabilizers arrest onset of TTR [51], histone deacetylases function to correct CF [52], Friedreich Ataxia [53], HD [54] and poly-glutamine (polyQ) disease [26] and have a strong like to GPC [18].

It is now clear that the ultimate goal for FoldFx modeling will be to utilize it as framework for further understanding of human biology and for development of small molecule therapeutics that manipulate GPC triad to maintain and restore human health.

2.1.6.5 To Be or Not to Be. How Selective Autophagy and Cell Death Govern Cell Fate

Douglas R. Green, Beth Levine
Cell    Mar 2014; 157(1):65–75
http://dx.doi.org:/10.1016/j.cell.2014.02.049

The health of metazoan organisms requires an effective response to organellar and cellular damage either by repair of such damage and/or by elimination of the damaged parts of the cells or the damaged cell in its entirety. Here, we consider the progress that has been made in the last few decades in determining the fates of damaged organelles and damaged cells through discrete, but genetically overlapping, pathways involving the selective autophagy and cell death machinery. We further discuss the ways in which the autophagy machinery may impact the clearance and consequences of dying cells for host physiology. Failure in the proper removal of damaged organelles and/or damaged cells by selective autophagy and cell death processes is likely to contribute to developmental abnormalities, cancer, aging, inflammation, and other diseases.

As in all living things, each of our cells suffers the slings and arrows of outrageous fortune, facing damage from without and within. And, like the Prince of Denmark, each decides whether to be or not to be. To be, the cell must monitor and repair the damage. If not, it will “melt, thaw, and resolve itself into a dew,” dying and cleared from the body by other cells (with apologies to Shakespeare for scrambling his immortal words).

Here, we consider how the molecular pathways of autophagy and cell death and, ultimately the clearance of dying cells, function in this crucial decision. Although autophagy and cell death occur in response to a wide variety of metabolic and other cues, our focus is restricted here to those aspects of each that are directly concerned with the quality control of cells—the “garbage” (cellular or organellar) that must be managed for organismal function. And although there are many important functions of quality control mechanisms (e.g., DNA and membrane repair, cell growth and cell-cycle control, unfolded protein and endoplasmic reticulum (ER) stress responses, innate and adaptive immunity, and tumor suppression), our discussion is limited to the selective disposal of damaged or otherwise unwanted organelles and, when necessary, damaged or excess cells and how the autophagic and cell death mechanisms function in these processes. Overall, we focus on the overriding theme of waste management, but as we will see, many of the links between these elements remain largely unexplored. Further, although a great deal of what we know was delineated in yeast and invertebrate model systems, we largely restrict our consideration to what is known in mammals.

Engaging Autophagy

The process of macroautophagy (herein, autophagy) is best understood in the context of nutrient starvation (Kroemer et al., 2010 and Mizushima and Komatsu, 2011). When energy in the form of ATP is limiting, AMP kinase (AMPK) becomes active, and this can drive autophagy. Similarly, deprivation from growth factors and/or amino acids leads to the inhibition of TORC1, which, when active, represses conventional autophagy. As a result of AMPK induction and/or TORC1 inhibition, autophagy is engaged, although other signals may bypass AMPK and TORC1 to engage autophagy (Figure 1).

http://ars.els-cdn.com/content/image/1-s2.0-S0092867414002943-gr1.jpg

Figure 1. Overview of the General Autophagy Pathway

Cellular events and selected aspects of the molecular regulation involved in the lysosomal degradation pathway of autophagy in mammalian cells are shown. Several membrane sources may serve as the origin of the autophagosome and/or contribute to its expansion. A “preinitiation” complex (also called the ULK complex) is negatively and positively regulated by upstream kinases that sense cellular nutrient and energy status, resulting in inhibitory and stimulatory phosphorylations on ULK1/2 proteins. In addition to nutrient-sensing kinases shown here, other signals involved in autophagy induction may also regulate the activity of the ULK complex. The preinitiation complex activates the “initiation complex” (also called the Class III PI3K complex) through ULK-dependent phosphorylation of key components and, likely, other mechanisms. Activation of the Class III PI3K complex requires the disruption of binding of Bcl-2 antiapoptotic proteins to Beclin 1 and is also regulated by AMPK and a variety of other proteins not shown in the figure. The Class III PI3K complex generates PI3P at the site of nucleation of the isolation membrane (also known as the phagophore), which leads to the binding of PI3P-binding proteins (such as WIPI/II) and the subsequent recruitment of proteins involved in the “elongation reaction” (also called the ubiquitin-like protein conjugation systems) to the isolation membrane. These proteins contribute to membrane expansion, resulting in the formation of a closed double-membrane structure, the autophagosome, which surrounds cargo destined for degradation. The phosphatidylethanolamine-conjugated form of the LC3 (LC3-PE), generated by the ATG4-dependent proteolytic cleavage of LC3, and the action of the E1 ligase, ATG7, the E2 ligase, ATG3, and the E3 ligase complex, ATG12/ATG5/ATG16L, is the only autophagy protein that stably associates with the mature autophagosome. The autophagosome fuses with a lysosome to form an autolysosome; inside the autolysosome, the sequestered contents are degraded and released into the cytoplasm for recycling. Late endosomes or multivesicular bodies can also fuse with autophagosomes, generating intermediate structures known as amphisomes, and they also contribute to the formation of mature lysosomes. Additional proteins (not depicted in diagram) function in the fusion of autophagosomes and lysosomes. The general autophagy pathway has numerous functions in cellular homeostasis (examples listed in box labeled “physiological functions”), which contribute to the role of autophagy in development and protection against different diseases.

The “goal” of the autophagy machinery is to deliver cytosolic materials to the interior of the lysosomes for degradation, thereby recovering sources of metabolic energy and requisite metabolites in times of starvation (general autophagy). Autophagy can similarly function to target damaged or otherwise unwanted organelles to lysosomes for removal (selective autophagy). Although we focus here primarily on selective autophagy, it is useful to also consider general autophagy to highlight similarities and distinctions between the two processes.

In both cases a double-membrane structure, the autophagosome, fuses with lysosomes to deliver the contents for degradation, and this involves a proteolipid molecule, LC3-II, a component of the autophagosome composed of a protein, LC3, and a lipid, phosphatidylethanolamine. LC3-II is generated by a process resembling ubiquitination, involving E1, E2, and E3 ligases (Figure 1). The parent molecule, LC3-I, is generated by the action of a protease, ATG4, which cleaves LC3 to produce LC3-I. This is bound by the E1, ATG7, and transferred to the E2, ATG3. The E3 ligase is a complex composed of ATG16L and ATG12-5; the latter is produced by another reaction in which ATG12 is bound by the E1, ATG7, transferred to a different E2, ATG10, and from there to ATG5. The process by which ATG12-5 is formed—and, subsequently, LC3-II (also known as LC3-PE) is generated—is referred to as the elongation reaction and is required for the formation of the autophagosome.

As discussed in the Introduction, autophagy can function to remove damaged or otherwise unwanted organelles in a cell. By “unwanted,” we mean organelles that are removed during differentiation (e.g., in maturing erythrocytes) or when environmental factors (e.g., hypoxia) disfavor some organelles in the cell. We refer to this process as selective autophagy. When considering selective autophagy, we are faced with two problems. First, how does the process “know” which structures or organelles to target for removal? And second, how does this occur even when the conventional autophagy machinery is suppressed (at least partially), such as in nutrient-rich conditions? With regard to the latter, the problem is confounded by the simple fact that lysosomal digestion of organelles will itself provide amino acids and other metabolites, presumably activating TORC1 and suppressing AMPK. As we have seen, such conditions inhibit the function of the preinitiation complex. Nevertheless, animals lacking Ulk1 display a defect in at least one selective autophagic process, that of efficient removal of mitochondria during erythrocyte development (Kundu et al., 2008). Presumably, there are ways to bypass conventional inhibitory mechanisms to engage Ulk1 activity and promote selective autophagy in some settings. Alternatively, the preinitiation complex may be bypassed in some situations. We will not fully resolve this paradox here but perhaps provide clues as we consider the first problem—how specific cargoes are marked for clearance.

Before considering this issue, it may be useful to note that, even in nutrient-starved conditions, autophagy may be selective. Ribosomes represent a major portion of the biomass of many cells, and upon starvation, these are more rapidly removed than other structures in the cell (Cebollero et al., 2012). Similarly, there appears to be selective removal of peroxisomes during starvation (Hara-Kuge and Fujiki, 2008). The same may be the case for ER (reticulophagy), although it remains possible that, in this case, this is a consequence of developing the requisite autophagosomes for nutritional supplementation using the ER membrane (see above). Another possible selection during starvation is the preservation of functional mitochondria; because these are necessary for the catabolism of free fatty acids or amino acids and for the optimal generation of energy from glucose (all generated by lysosomal digestion), it simply does not make sense that inadvertent removal of mitochondria during starvation would be permitted. Such possible “antiselection,” however, has not been fully documented or adequately explored.

Targeting in selective organellar autophagy is perhaps best analyzed in the clearance of mitochondria (mitophagy) and peroxisomes (pexophagy). Tissues or cells lacking requisite components of the autophagy elongation machinery (e.g., ATG5 and ATG7) often display greatly increased numbers of apparently damaged mitochondria (Mizushima and Levine, 2010) and peroxisomes (Till et al., 2012). That said, there is evidence that, even in the absence of ATG5 and ATG7, some selective mitophagy continues via unknown mechanisms, perhaps via vesicular trafficking between mitochondria and lysosomes (Soubannier et al., 2012). Nevertheless, accumulated observations indicate that the autophagy elongation machinery and autophagosome formation are important for selective autophagy of damaged or otherwise unwanted organelles.

One way in which damaged mitochondria are removed by autophagy involves the action of two proteins, PINK1 and Parkin (Figure 2). PINK1 is a kinase that is constitutively imported into functional mitochondria and degraded by the rhomboid protease, PARL. As with most mitochondrial import, this requires the transmembrane potential of the inner mitochondrial membrane, ΔΨm. Loss of this potential, which can occur when the electron transport chain is damaged or if protons are allowed to pass freely across the inner membrane (i.e., due to the expression of uncoupler protein [UCP], presence of environmental protonophores, or as a consequence of the mitochondrial permeability transition) causes active PINK1 to accumulate on the cytosolic face of the outer mitochondrial membrane. This then recruits and activates Parkin, which is a ubiquitin E3-ligase, which then ubiquitinates proteins on the mitochondria.

http://ars.els-cdn.com/content/image/1-s2.0-S0092867414002943-gr2.jpg

Figure 2. Roles of Autophagy Proteins in the Removal of Unwanted Organelles and in the Removal of Cells

The left panel shows Parkin-dependent and Parkin-independent mechanisms involved in the selective degradation of mitochondria by autophagy (mitophagy). In Parkin-dependent mitophagy, mitochondrial damage and loss of mitochondrial membrane potential (ΔΨm) lead to localization of the kinase, PINK1, on the cytoplasmic surface of the mitochondria, resulting in recruitment of the E3 ubiquitin ligase, Parkin, to the mitochondria, followed by the ubiquitination of mitochondrial proteins and the formation of an isolation membrane that surrounds the damaged mitochondria. In Parkin-independent mitophagy, proteins such as Nix (shown in figure), BNIP3, and FUNDC1 (not shown in the figure) bind to LC3. Other autophagy proteins may be involved in Parkin-dependent and Parkin-independent mitophagy (discussed in the text). The precise details of how an isolation membrane is formed around specific mitochondria earmarked for degradation are unclear. Other damaged/unwanted organelles such as ER, peroxisomes, and lipid droplets can also be degraded by selective autophagy; the molecular mechanisms of these forms of selective autophagy are not well understood in mammalian cells. The right panel depicts roles of LAP of apoptotic corpses and of live cells (entosis). In LAP, components of the autophagy initiation complex (Beclin 1 and VPS34) are recruited to the phagosome, which leads to recruitment of LC3-PE and facilitation of phagolyosomal fusion. This process requires other components of the elongation machinery, but—in contrast to general autophagy or selective autophagy—proceeds independently of the ULK preinitiation complex.

It is self-evident that the selective removal of damaged or excess organelles is a critical homeostatic process, but beyond this, our information on what happens when this goes wrong is somewhat limited. There is an accumulation of damaged organelles (including mitochondria, perixosomes, and ER) and organ degeneration in mice with tissue-specific knockout of core autophagy genes such as Atg5 and Atg7in liver, neurons, heart, pancreatic acinar cells, muscle, podocytes, adipocytes, and hematopoietic stem cells ( Mizushima and Levine, 2010). Although it may not be possible to dissociate the effects of general autophagy from those of selective autophagy, it is reasonable to postulate that these phenotypes are partly related to defects in selective organellar autophagy, and at a minimum, such studies unequivocally establish a role for autophagy genes in the removal of damaged organelles in vivo.

The proper removal of excess or unwanted mitochondria is likely necessary for certain key aspects in development. As discussed above, the mitophagy factor Nix is required for mitochondrial clearance during erythroid maturation in vivo (Sandoval et al., 2008), and mouse erythrocytes lacking general autophagy factors such as Ulk1 and Atg7 fail to clear mitochondria ( Kundu et al., 2008 and Mortensen et al., 2010). Reduction in mitochondrial number may also contribute to the role of core autophagy genes, such as Atg7, in white adipocyte differentiation ( Zhang et al., 2009). An intriguing question is whether selective mitophagy—of paternal mitochondria—during embryonic development underlies mammalian maternal mitochondrial DNA (mtDNA) inheritance ( Levine and Elazar, 2011). In C. elegans, several studies showed that paternal mitochondria and mtDNA are eliminated from the fertilized oocyte by autophagy (with surrounding membranous organelles, but not the mitochondria themselves, marked by ubiquitin) ( Al Rawi et al., 2011, Sato and Sato, 2011 and Zhou et al., 2011). In one of these studies ( Al Rawi et al., 2011), p62 and LC3 were also found to colocalize with sperm mitochondria after fertilization in mice. However, a more recent study confirmed that sperm mitochondria colocalized with p62 and LC3 in mouse embryos but concluded that this was not involved in their degradation ( Luo et al., 2013). Thus, the question of whether selective mitophagy explains why our mitochondrial DNA comes mainly from our mothers remains to be resolved.

An emerging far-reaching biomedical paradigm is that defects in mitophagy—presumably through resulting abnormal mitochondrial function, abnormal mitochondrial biogenesis, and/or increased mitochondrial generation of reactive oxygen species (leading to genomic instability and enhanced proinflammatory signaling) —contribute to cancer, neurodegenerative diseases, myopathies, aging, and inflammatory disorders (reviewed in Ding and Yin, 2012, Green et al., 2011, Lu et al., 2013 and Narendra and Youle, 2011). This paradigm intuitively makes sense and is consistent with a large body of literature in autophagy-deficient mice. Yet, it is difficult to establish a direct causal relationship between mitophagy defects and disease in mice lacking general autophagy factors. Presumably, phenotypes observed in mice lacking selective mitophagy factors may be more informative. For example, Parkin-deficient mice have cancer-prone phenotypes, including accelerated intestinal adenoma development (in the background ofApc mutation) ( Poulogiannis et al., 2010) and the development of hepatocellular carcinoma ( Fujiwara et al., 2008). However, these studies also do not provide direct evidence that Parkin-mediated mitophagy, rather than other potential effects of Parkin, contribute to its role in tumor suppression. Moreover, mice lacking Ulk1 ( Kundu et al., 2008) or Nix ( Sandoval et al., 2008) have progressive anemia with mature erythrocytes containing mitochondria but no other obvious cancer-prone defects. In addition, Parkin-null mice clear defective mitochondria normally in dopaminergic neurons in the substantia nigra ( Sterky et al., 2011), even though PARKIN and PINK1 mutations in humans lead to overt degeneration of these neurons and Parkinson’s disease. It is not unlikely that there are several overlapping mechanisms for selective autophagy that compensate for such deficiencies. Another possible explanation for the lack of more striking phenotypes in mice lacking selective autophagy factors is that other processes help to mediate the damage that should accrue when damaged organelles are not effectively cleared from cells, including perhaps the removal of the cells themselves.

Figure 3. Cell Death Pathways Engaged by Cellular Damage

Cellular damage induces cell death by inducing expression and/or modification of proapoptotic BH3-only proteins of the Bcl-2 family (inset), which engage the mitochondrial pathway of apoptosis, in which MOMP releases proteins of the mitochondrial intermembrane space. Among these is cytochrome c, which activates APAF1 to form a caspase-activation platform (the apoptosome) that binds and activates caspase-9. This then cleaves and thereby activates executioner caspases to promote apoptosis. Cellular damage can also induce the expression of death ligands of the TNF family, which bind their receptors to promote the activation of caspase-8 by FADD. The latter is antagonized by expression of c-FLIP_L, and the caspase-8-FLIP heterodimer does not promote apoptosis but instead blocks another cell death pathway engaged by death receptors, necroptosis. Necroptosis involves the activation of RIPK1 and RIPK3, resulting in phosphorylation and activation of the pseudokinase, MLKL, which promotes an active necrotic cell death.

Noncanonical Autophagy Pathway and Clearance of Dying Cells

When dying cells are engulfed by a macrophage or other cell, the corpse-containing phagosome is rapidly decorated with the autophagic protein, LC3, which facilitates fusion with lysosomes and destruction of the cargo (Sanjuan et al., 2009). This LC3-associated phagocytosis (LAP) is dependent on the Beclin 1-VPS34 complex and the elongation machinery, but rather than generating a double-membrane autophagosome, LC3-II is generated on the single-membrane phagosome itself (Figure 2). In contrast to general or selective organellar autophagy, however, LAP appears to proceed independently of the ULK1 preinitiation complex in mammalian cells (Henault et al., 2012). If LAP is defective (due, for example, to lack of the requisite autophagy machinery), the corpse is not digested, and macrophages produce high levels of proinflammatory cytokines (Martinez et al., 2011). This may have implications for disease. For example, systemic lupus erythematosis is often characterized by circulation of “LE cells,” which have been identified as macrophages containing an undigested corpse. It is possible that proinflammatory signals emitted by such macrophages contribute to the disease.

The Interface of Autophagy and Cell Death in Tissue Homeostasis

LAP (see above) may also represent a link between autophagy components and a cell death process (and not only the clearance of dying cells), as engulfment of cells may restrict oncogenesis naturally. Immortalized mammary epithelial cells, upon loss of anchorage to basement membranes, engulf each other in a process called “entosis” (Florey et al., 2010). The engulfed cell dies by apoptosis due to nutrient deprivation. If the cell expresses an antiapoptotic signal, such as Bcl-2, it is nevertheless killed as LAP in the engulfing cell promotes fusion with lysosomes (Figure 2). However, if cells resist apoptosis and fail to engage LAP (e.g., due to ablation of the autophagic machinery), such immortalized cells escape entosis and grow in an anchorage-independent manner. The implications of such an interplay between apoptosis and autophagy at the cellular level has obvious consequences for understanding oncogenesis.

In thinking about general autophagy (in response to metabolic stress), selective autophagy (in response to damaged organelles), and cell death (as a consequence of excessive damage), it is obvious that the pathways crosstalk at a superficial level. That is, a cell that is defective for autophagy will necessarily be more prone to die if faced with nutrient deprivation. Cells lacking selective autophagy will accumulate damaged organelles such as mitochondria, which can generate signals (e.g., ROS) that promote further damage and ultimately cell death. If cell death does not occur, the ensuing dysfunction may promote severe effects in the form of oncogenesis. And, of course, a cell that engages death pathways will circumvent any benefit that might arise from autophagy. In a recent study, the extent of autophagy of individual cells in a population inversely correlated with the likelihood that a cell would die in response to engagement of the death receptor pathway of apoptosis (Gump et al., 2014).

There are more fundamental molecular interactions between these pathways, but it is difficult to parse how specific interactions contribute to cross-regulation in the face of the overarching effect of apoptotic defects on cellular health. For example, Beclin 1 is bound and inhibited by the antiapoptotic Bcl-2 proteins (Pattingre et al., 2005), and proapoptotic BH3-only proteins appear to be capable of disrupting this interaction to promote autophagy (Maiuri et al., 2007). Other autophagy components also interact with apoptotic players, but again, it is unclear whether these interactions, per se, influence cell fate.

We began our discussion with what may be the most fundamental dichotomy in biology, to be or not to be. At the cellular level, the question might be less elegantly posed along these lines: should we (the cell) be functional or, if not, die? Or, if we are dysfunctional and survive, do we risk compromising the life of the organism? Do we unite the fundamental pathways of garbage disposal, selective autophagy, and active cell death through complex (and largely unexplored) molecular interactions, or do we let the thresholds of damage dictate which pathway holds sway against the thousand natural shocks that flesh is heir to (again, with apologies to the Bard)? Those are the questions.

 

2.1.6.6 Signaling cell death from the endoplasmic reticulum stress response

Shore GC, Papa FR, Oakes SA.
Curr Opin Cell Biol. 2011 Apr; 23(2):143-9
http://dx.doi.org:/10.1016/j.ceb.2010.11.003

Inability to meet protein folding demands within the endoplasmic reticulum (ER) activates the unfolded protein response (UPR), a signaling pathway with both adaptive and apoptotic outputs. While some secretory cell types have a remarkable ability to increase protein folding capacity, their upper limits can be reached when pathological conditions overwhelm the fidelity and/or output of the secretory pathway. Irremediable ‘ER stress’ induces apoptosis and contributes to cell loss in several common human diseases, including type 2 diabetes and neurodegeneration. Researchers have begun to elucidate the molecular switches that determine when ER stress is too great to repair and the signals that are then sent from the UPR to execute the cell.

Connections from the UPR to the Mitochondrial Apoptotic Pathway

Under excessive ER stress, the ER transmembrane sensors IRE1α and PERK send signals through the BCL-2 family of proteins to activate the mitochondrial apoptotic pathway. In response to unfolded proteins, IRE1α oligomerizes and induces endonucleolytic decay of hundreds of ER-localized mRNAs, depleting ER protein folding components and leading to worsening ER stress. Phosphorylated IRE1α also recruits TNF receptor-associated factor 2 (TRAF2) and activates apoptosis signaling kinase 1 (ASK1) and its downstream target c-Jun NH2-terminal kinase (JNK). JNK then activates pro-apoptotic BIM and inhibits anti-apoptotic BCL-2. These conditions result in dimerization of PERK and activation of its kinase domain to phosphorylate eukaryotic translation initiation factor 2α (eIF2α), which causes selective translation of activating transcription factor-4 (ATF4). ATF4 upregulates expression of the CHOP/GADD153 transcription factor, which inhibits the gene encoding anti-apoptotic BCL-2 while inducing expression of pro-apoptotic BIM. ER stress also promotes p53-dependent transcriptional upregulation of Noxa and Puma, two additional pro-apoptotic BH3-only proteins. Furthermore, high levels of UPR signaling induce initiator caspase-2 to proteolytically cleave and activate pro-apoptotic BID upstream of the mitochondrion. In addition to antagonizing pro-survival BCL-2 members, cleaved BID, BIM and PUMA activate Bax and/or Bak. Hence, in response to excessive UPR signaling, the balance of BCL-2 family proteins shifts in the direction of apoptosis and leads to the oligomerization of BAX and BAK, two multi-domain pro-apoptotic BCL-2 family proteins that then drive the permeabilization of the outer mitochondrial membrane, apoptosome formation and activation of executioner caspases such as Caspase-3. Figure adapted with permission from the Journal of Cell Science [58].

The lumen of the ER is a unique cellular environment optimized to carry out the three primary tasks of this organelle: calcium storage and release, protein folding and secretion, and lipid biogenesis [1]. A range of cellular disturbances lead to accumulation of misfolded proteins in the ER, including point mutations in secreted proteins that disrupt their proper folding, sustained secretory demands on endocrine cells, viral infection with ER overload of virus-encoding protein, and loss of calcium homeostasis with detrimental effects on ER-resident calcium-dependent chaperones [2–4]. The tripartite UPR consists of three ER transmembrane proteins (IRE1α, PERK, ATF6) that alert the cell to the presence of misfolded proteins in the ER and attempt to restore homeostasis in this organelle through increasing ER biogenesis, decreasing the influx of new proteins into the ER, promoting the transport of damaged proteins from the ER to the cytosol for degradation, and upregulating protein folding chaperones [5]. The adaptive responses of the UPR can markedly expand the protein folding capacity of the cell and restore ER homeostasis [6]. However, if these adaptive outputs fail to compensate because ER stress is excessive or prolonged, the UPR induces cell death. The cell death pathways collectively triggered by the UPR include both caspase-dependent apoptosis and caspase-independent necrosis. While many details remain unknown, we are beginning to understand how cells determine when ER stress is beyond repair and communicate this information to the cell death machinery. For the purposes of this review, we focus on the apoptotic outputs trigged by the UPR under irremediable ER stress. While the ER contains numerous additional signaling platforms and targets that respond to diverse apoptotic stimuli (eg., those associated with the Bap31 complex [7,8]), their formal link to UPR-driven apoptosis remains to be determined.

The proximal unfolded protein response sensors

UPR signaling is initiated by three ER transmembrane proteins: IRE1α, PERK, and ATF6. The most ancient ER stress sensor, IRE1α, contains an ER lumenal domain, a cytosolic kinase domain and a cytosolic RNase domain [9,10]. In the presence of unfolded proteins, IRE1α’s ER lumenal domains homo-oligomerize, leading first to kinase trans-autophosphorylation and subsequent RNase activation. Dissociation of the ER chaperone BiP from IRE1α’s lumenal domain in order to engage unfolded proteins may facilitate IRE1α oligomerization [11]; alternatively, the lumenal domain may bind unfolded proteins directly [12]. PERK’s ER lumenal domain is thought to be activated similarly [13,14]. The ATF6 activation mechanism is less clear. Under ER stress, ATF6 translocates to the Golgi and is cleaved by Site-1 and Site-2 proteases to generate the ATF6(N) transcription factor [15].

All three UPR sensors have outputs that attempt to tilt protein folding demand and capacity back into homeostasis. PERK contains a cytosolic kinase that phosphorylates eukaryotic translation initiation factor 2α (eIF2α), which impedes translation initiation to reduce the protein load on the ER [16]. IRE1α splices XBP1mRNA, to produce the homeostatic transcription factor XBP1s [17,18]. Together with ATF6(N), XBP1s increases transcription of genes that augment ER size and function[19]. When eIF2α is phosphorylated, the translation of the activating transcription factor-4 (ATF4) is actively promoted and leads to the transcription of many pro-survival genes [20]. Together, these transcriptional events act as homeostatic feedback loops to reduce ER stress. If successful in reducing the amount of unfolded proteins, the UPR attenuates.

However, when these adaptive responses prove insufficient, the UPR switches into an alternate mode that promotes apoptosis. Under irremediable ER stress, PERK signaling can induce ATF-4-dependent upregulation of the CHOP/GADD153 transcription factor, which inhibits expression of the gene encoding anti-apoptotic BCL-2 while upregulating the expression of oxidase ERO1α to induce damaging ER oxidation [21,22]. Sustained IRE1α oligomerization leads to activation of apoptosis signal-regulating kinase 1 (ASK1) and its downstream target c-Jun NH₂-terminal kinase (JNK) [23,24]. Phosphorylation by JNK has been reported to both activate pro-apoptotic BIM and inhibit anti-apoptotic BCL-2 (see below). Small molecule modulators of ASK1 have been shown to protect cultured cells against ER stress-induced apoptosis, emphasizing the importance of the IRE1α-ASK1-JNK output as a death signal in this pathway [25]. In response to sustained oligomerization, the IRE1α RNase also causes endonucleolytic decay of hundreds of ER-localized mRNAs [26]. By depleting ER cargo and protein folding components, IRE1α-mediated mRNA decay may worsen ER stress, and could be a key aspect of IRE1α’s pro-apoptotic program [27]. Recently, inhibitors of IRE1α’s kinase pocket have been shown to conformationally activate its adjacent RNase domain in a manner that enforces homeostatic XBP1s without causing destructive mRNA decay [27], a potentially exciting strategy for preventing ER stress-induced cell loss.

When deciding whether to switch into apoptotic mode, cells might use one or more “timers” to indicate if UPR signaling remains continuously active under high or chronic ER stress. For example, sustained PERK activity could result in protracted translation attenuation, which should be incompatible with survival, as well as high levels of pro-apoptotic CHOP. Similarly, high-level mRNA degradation mediated by IRE1α may deplete ER protein folding capacity further, and along with JNK signaling push the cell towards apoptosis. Thus, the continuous activation of the proximal sensors IRE1α and PERK may constitute a “timer” that triggers the switch to apoptosis under irremediable ER stress. Moreover, the ultimate response may depend on cell context. For example, the ability of IRE1α to complex with regulators such as BAX/BAK and Bax Inhibitor 1 (BI-1) at the ER may influence its ability to remediate ER stress and/or potentially signal apoptosis [28,29].

The BCL-2 family and the Mitochondrial Apoptotic Pathway

A wealth of genetic and biochemical data argues that the intrinsic (mitochondrial) apoptotic pathway is the major cell death pathway induced by the UPR, at least in most cell types. This apoptotic pathway is set in motion when several toxic proteins (e.g., cytochrome c, Smac/Diablo) are released from mitochondria into the cytosol where they lead to activation of downstream effector caspases (e.g., Caspase-3) [30]. The BCL-2 family, a large class of both pro- and anti- survival proteins, tightly regulates the intrinsic apoptotic pathway by controlling the integrity of the outer mitochondrial membrane [31]. This pathway is set in motion when cell injury leads to the transcriptional and/or post-translational activation of one or more BH3-only proteins, a structurally diverse class of pro-apoptotic BCL-2 proteins that share sequence similarity only in a short alpha helix (~9–12 a.a.) known as the Bcl-2 homology 3 (BH3) domain [32]. Once activated, BH3-only proteins lead to loss of mitochondrial integrity by disabling mitochondrial protecting proteins (e.g., BCL-2, BCL-XL, MCL-1) and (for a subset) directly triggering the oligomerization of the multidomain pro-apoptotic BAX and BAK proteins that drive the permeabilization of the outer mitochondrial membrane.

ER stress has been reported to activate at least four distinct BH3-only proteins (BID, BIM, NOXA, PUMA) that then signal the mitochondrial apoptotic machinery (i.e., BAX/BAK) [33–35]. Each of these BH3-only proteins is activated by ER stress in a unique way. For example, BID is proteolytically cleaved at a caspase recognition site into a potent apoptotic signal [33]. The Bim gene is transcriptionally upregulated and its protein product stabilized through dephosphorylation in response to ER stress [34]. Cells individually deficient in any of these BH3-only proteins are modestly protected against ER stress-inducing agents, but not nearly as resistant as cells null for their common downstream targets BAX and BAK [36]—the essential gatekeepers to the mitochondrial apoptotic pathway. Moreover, cells genetically deficient in both Bim andPuma are more protected against ER stress-induced apoptosis than Bim or Puma single knockout cells [37], arguing that several BH3-only proteins are necessary for efficient activation of BAX/BAK-dependent apoptosis under conditions of irremediable ER stress.

The ER stress sensor that signals these BH3-only proteins is known in a few cases (i.e., BIM is downstream of PERK); however, we do not yet understand how the UPR communicates with most of the BH3-only proteins. Moreover, it is not known if all of the above BH3-only proteins are simultaneously set in motion by all forms of ER stress or if a subset is activated under specific pathological stimuli that injure this organelle. Understanding the molecular details of how ER damage is communicated to the mitochondrial apoptotic machinery is critical if we want to target disease specific apoptotic signals sent from the ER.

Initiator and Executor Caspases

Caspases, or cysteine-dependent aspartate-directed proteases, play essential roles in both initiating apoptotic signaling (initiator caspases- 2, 4, 8, 12) and executing the final stages of cell demise (executioner caspases- 3, 7, 9) [38]. The executioner caspases are proteolytically activated through either mitochondrial-dependent apoptosome formation or death receptor activation of upstream initiator caspases (i.e., caspase- 8, 10). Given the promiment role of the mitochondrial apoptotic pathway in ER stress-induced death, it is not surprising that the executioner caspases (casp-3,7,9) are critical for cell death resulting from damage to this organelle. On the other hand, there has been much controversy regarding the role of initiator caspases in ER stress-induced apoptosis. Caspase 12 was the first caspase reported to localize to the ER and become activated by UPR signaling in murine cells [39]. However, Caspase 12 was subsequently shown to be downstream of BAX/BAK-dependent mitochondrial permeabilization and executioner caspase activation in this pathway [40], arguing that its role is probably limited to amplifying rather than initiating ER stress-induced apoptosis. Moreover, most humans fail to express a functional CASP12 due to a polymorphism that creates a nonsense mutation in the coding region [41], which rules out an essential role for this protease in human ER stress signaling. More recently, caspase-2 was found to be the premitochondrial protease that proteolytically cleaves and activates the BH3-only protein BID in response to ER stress [33]. Genetic knockdown or pharmacological inhibition of caspase-2 confers resistance to ER stress-induced apoptosis [42]. How the UPR activates caspase-2 and whether other initiator caspases, such as caspase 4, are also involved remains to be determined.

Calcium and Cell Death

Although an extreme depletion of ER luminal Ca²⁺ concentrations is a well-documented initiator of the UPR and ER stress-induced apoptosis or necrosis, it represents a relatively non-physiological stimulus. Given that Ca²⁺ signaling from the ER is likely coupled to most pathways leading to apoptosis, however, it is not surprising that this also extends to UPR overload. For example, recent evidence in macrophages indicates that UPR-induced activation of ERO1-α via CHOP results in stimulation of inositol 1,4,5-triphosphate receptor (IP3R) [43], the major release channel for luminal Ca²⁺ from the ER. Although pathways may exist for ER Ca²⁺ release independently of IP3 receptors, many seemingly disparate pathways appear to converge on the IP3R platform. Consistent with this, all three sub-groups of the Bcl-2 family at the ER regulate IP3R activity. Mechanistically, this might ultimately result from titrations of pro-survival Bcl-X_L, Bcl-2, and Mcl-1 that physically associate with IP3R [44]. Release of ER Ca²⁺ via IP3R into the cytoplasm could of course influence multiple pathways upstream of the core apoptosis machinery. However, a significant fraction of IP3R is a constituent of highly specialized tethers that physically attach ER cisternae to mitochondria (mitochondrial-associated membrane) and regulate local Ca²⁺ dynamics at the ER-mitochondrion interface [45–46]. This results in propagation of privileged IP3R-mediated Ca²⁺ oscillations into mitochondria, which can influence cell survival in multiple ways. In an extreme scenario, massive transmission of Ca²⁺ into mitochondria results in Ca²⁺ overload and cell death by caspase-dependent and –independent means [46], particularly via the pathway involving the permeability transition pore/cyclophilin D complex [47]. More refined transmission regulated by the Bcl-2 axis at the ER can influence cristae junctions and the availability of cytochrome c for its release across the outer mitochondrial membrane [48]. Finally, such regulated Ca²⁺transmission to mitochondria is a key determinant of mitochondrial bioenergetics, which is linked not only to potential apoptotic responses, but importantly to survival/death mechanisms dependent on macroautophagy [49].

ER Stress-Induced Cell Loss and Disease

Mounting evidence suggests that ER stress-induced apoptosis contributes to a range of human diseases of cell loss, including diabetes, neurodegeneration, stroke, and heart disease, to name a few (reviewed in REF [50]). The cause of ER stress in these distinct diseases varies depending on the cell type affected and the intracellular and/or extracellular conditions that disrupt proteostasis. For example, some cases of inherited amyotrophic lateral sclerosis (ALS) are caused by toxic, gain-of-function point mutations in superoxide dismutase-1 (SOD1). Other neurodegenerative diseases, such as Huntington, result from mutant proteins (e.g., huntingtin) containing expanded glutamate repeat sequences. Both mutant SOD1 and mutant huntingtin proteins aggregate, exhaust proteasome activity, and result in secondary accumulations of misfolded proteins in the ER [51–52]. In the early stages of type 2 diabetes, peripheral insulin resistance challenges pancreatic beta cells to secrete greater amounts of insulin in order to maintain euglycemia. This increased secretory demand can lead to ER stress, beta cell loss, and hyperglycemia [53]. Mutations in PERK result in massive pancreatic beta cell death and infant-onset diabetes in patients with Wolcott-Rallison syndrome [54], an autosomal recessive inherited disorder that illustrates the importance of a properly functioning UPR for beta cell health. An association between ER stress and heart disease has been implicated on a number of levels. Oxidative stress, high levels of cholesterol, and fatty acids can all cause ER stress-induced apoptosis of macrophages and endothelial cells associated with atherosclerotic plaques, leading to progression of atherosclerosis [55]. Myocardial infarction activates the UPR in cardiac myocytes; and Ask1^−/− mice show preservation of left ventricular function compared to wild-type controls after coronary artery ligation [56]. Stroke (ischemia-reperfusion injury) has also been shown to induce ER stress-induced apoptosis, and Chop^−/− mice are partly protected from neuronal loss after stroke injury [57].

While by no means exhaustive, these examples illustrate the therapeutic potential for novel drugs that block ER stress-induced apoptosis. While chronic UPR-targeted therapies may be problematic for the many tissues that require this pathway to maintain proteastasis, acute modulation of the UPR during stroke or myocardial infarction could be an effective strategy to prevent cell loss. In the case of IRE1α, it may be possible to use kinase inhibitors to activate its cytoprotective signaling and shut down its apoptotic outputs [27]. Whether similar strategies will work for PERK and/or ATF6 remains to be seen. Alternatively, blocking the specific apoptotic signals that emerge from the UPR is perhaps a more straightforward strategy to prevent ER stress-induced cell loss. To this end, small molecular inhibitors of ASK and JNK are currently being tested in a variety preclinical models of ER stress [52–53,56–57]. This is just the beginning, and much work needs to be done to validate the best drugs targets in the ER stress pathway.

Conclusions

The UPR is a highly complex signaling pathway activated by ER stress that sends out both adaptive and apoptotic signals. All three transmembrane ER stress sensors (IRE1α, PERK, AFT6) have outputs that initially decrease the load and increase capacity of the ER secretory pathway in an effort to restore ER homeostasis. However, under extreme ER stress, continuous engagement of IRE1α and PERK results in events that simultaneously exacerbate protein misfolding and signal death, the latter involving caspase-dependent apoptosis and caspase-independent necrosis. Advances in our molecular understanding of how these stress sensors switch from life to death signaling will hopefully lead to new strategies to prevent diseases caused by ER stress-induced cell loss.

​