Leaders in Pharmaceutical Business Intelligence (LPBI) Group

Posted in Biological Engineering, Biological Networks, Gene Regulation and Evolution, Biomarkers & Medical Diagnostics, Biomedical Measurement Science, Cell Biology, Cell Biology, Signaling & Cell Circuits, Cell Processing System in Cell Therapy Process Development, Epigenetics and Environmental Factors, Genomic Endocrinology, Human aging, Methylation, MicroEngineering Cell-Tissue & Systems, Molecular Genetics & Pharmaceutical, Population Health Management, Genetics & Pharmaceutical, Preimplantation Genetic Diagnosis and Reproductive Genomics, Reproductive Andrology, Embryology, Genomic Endocrinology, Preimplantation Genetic Diagnosis and Reproductive Genomics, Reproductive Biology & Bio Instrumentation | Tagged DNA methylation, Epigenetics, infertility, pregnancy, sperm | 1 Comment

One Response

1. on May 15, 2022 at 1:28 PM | Reply 2012pharmaceutical

   Dr. Saha,

   Thank you for posting. Which of the articles in https://pharmaceuticalintelligence.com/biomed-e-books/series-d-e-books-on-biomedicine/series-d-volume-4-human-reproductive-system-genomic-endocrinology-and-cancer-types/
   needs to be updated by this new article??

   Discussion
   Chronological age is a significant determinant of reproductive capacity and success among couples (Dunson et al., 2002; Mutsaerts et al., 2012; Sharma et al., 2015; Oluwayiose et al., 2021). However, chronological age does not encapsulate the cumulative internal (e.g. genetics) and external (e.g. environmental conditions) factors that ensue over the life-course, and thus it serves as a proxy measure of the ‘true’ biological age of cells. While semen quality outcomes utilizing WHO guidelines (Cooper et al., 2010) have been used for the assessment of male infertility for decades, they remain poor predictors of reproductive outcomes (Buck Louis et al., 2011; Jungwirth et al., 2012). Thus, the ability to capture the biological age of sperm may provide a novel platform to better assess the male contribution to reproductive success, especially among idiopathic infertile couples. Here, we report, for the first time to our knowledge, that our novel approach to estimate SEA is efficacious in strongly predicting couple fecundity, as measured by TTP, among couples from the general population. Our results indicate that higher SEA (both continuous and categorical) is associated with a longer TTP as well as shorter gestation among couples becoming pregnant. SEA was also higher among males who were current cigarette smoking. These results among pregnancy planners in the general population who are not seeking clinical fertility treatment are novel and hold promise to overcome the limitations of using conventional semen quality in predicting couples’ reproductive outcomes.

   In regard to epigenetic aging of sperm, we must consider the developmental stage whereby age has its greatest influence on DNA methylation patterns overtime. Male germ cells require dynamic epigenetic reprogramming for the progression from diploid spermatogonia to haploid spermatozoa (Godmann et al., 2009; Marcho et al., 2020). Although it is recognized that the final DNA methylation patterns are established during meiotic divisions (Oakes et al., 2007), the accumulation of aging-related methylation errors likely occur in the highly proliferative and self-renewing spermatogonia and are carried forward when cells are committed to differentiation during spermatogenesis. As such, SEA likely reflects that of spermatogonia, from which mature spermatozoa are generated during the 74-day process of spermatogenesis. In humans, it is estimated that spermatogonia divide every 16 days, which equates to 23 divisions a year (Goriely, 2016). Thus, the spermatogonia of the oldest participant in our study of 50 years would have undergone (taking into account approximate age of puberty) >800 divisions during his reproductive life-course. A consensus for the genomic context (intergenic versus gene regions) of these accumulated methylation errors remains to be resolved. Detailed nucleosome maps as well as 3D/4D architecture and dynamics of spermatogonia, not mature spermatozoa, in space and time hold promise to uncover why certain regions are more susceptible to age-related epigenetic dysregulation.

   Emerging data over the last few years demonstrate the profound effect of aging on the sperm methylome and their potential for constructing epigenetic clocks (Jenkins et al., 2018; Cao et al., 2020; Laurentino et al., 2020; Oluwayiose et al., 2021); however, the clinical relevance of these clocks has remained largely unexplored. Among 329 samples from infertile patients, sperm donors and individuals from the general population, Jenkins et al. (2018) observed that sperm methylation at 51 genomic regions (via Illumina’s 450K) reproducibly predicted an individual’s chronological age regardless of fertility status (r = 0.89; MAE = 2.04). Similar to our results, smokers had higher SEA compared to never smokers. Comparing young (n = 6; 18–24 years) and old men (n = 6; 61–71 years) men, Laurentino et al. identified 236 age-related sperm DMRs via shotgun sequencing (Laurentino et al., 2020). Six DMRs with the lowest P-value were selected to build an epigenetic clock in 42 additional samples and subsequently in an independent set of 33 samples; however, the clock yielded high errors (MAEs = 7.8 and 9.8 years, respectively), which is likely attributable to the smaller sample sizes and small number of DMRs used in their analyses. Most recently, a customized methyl-capture sequencing approach identified 798 age-associated sperm DMRs by categorizing men as either young (n = 48; 18–38 years) or old (n = 46; 46–71 years) (Cao et al., 2020). Elastic net analyses utilizing the top 5000 age-associated CpGs generated a sperm clock with an average error of 2.7 years. The authors note that their age prediction improved by increasing the number to thousands of CpGs; however, this puts into question the balance between assay efficiency (e.g. measuring thousands compared to 120 CpGs in our SEACpG clock) and incremental improvement in predictive value of SEA. A distinct advantage of our approach was to build a clock specifically to understand the impact of biological age on reproductive outcomes in the general population among couples who were discontinuing contraception for purposes of trying to become pregnant. Jenkins et al. (2018) combined samples from sperm donors and infertility patients; however, this approach may obscure differences in the biological aging patterns across these groups. Furthermore, the application of our clocks to predict biological age in an independent cohort of men seeking infertility treatment provided strong evidence of its relevancy for the general population.

   Our epigenetic clocks are the first to employ Super Learner, which uses state-of-the-art machine learning methods to improve predictive performance. Previous clocks have relied on penalized linear regression (Jenkins et al., 2018; Cao et al., 2020; Laurentino et al., 2020). This approach requires the strong assumption that chronological age is related to the DNA methylation in a simple linear fashion; in other words, there are no interaction terms or other non-linear effects. Additionally, the penalty terms in LASSO, elastic net or ridge regression aim to balance the total number of features with predictive performance; in so doing, some relevant features might be excluded to avoid overfitting. In contrast, our approach uses cross-validation to create an optimal weighted combination of multiple prediction algorithms, including both linear and non-linear approaches. Super Learner has improved performance in a variety of settings, including predicting mortality in intensive care units, violence in prisons and HIV risk in resource-limited settings (Pirracchio et al., 2015; Baćak and Kennedy, 2019; Balzer et al., 2020). Indeed, Super Learner is theoretically guaranteed to perform at least as well as the best algorithm in its ensemble (van der Laan and Dudoit, 2003). In this application, we considered both penalized regressions as well as multivariate adaptive regression splines, which is a highly non-linear and flexible approach. Our Super Learner SEACpG resulted in a low MAE of 1.6 and, as expected, outperformed prediction when relying solely on elastic net, which yielded a MAE of 2.2 (data not shown). Importantly, the weights assigned to each algorithm differed for the SEACpG and SEADMR clocks, underscoring the flexibility of the Super Learner algorithm.

   Previous epigenetic clocks in somatic tissue (Hannum et al., 2013; Horvath, 2013; Levine et al., 2018) and sperm (Jenkins et al., 2018; Cao et al., 2020; Laurentino et al., 2020) have relied on either individual CpGs or regional (DMR-based) approaches; however, the comparison of these two approaches is limited within the same study. In our approach, Super Learner utilized 22 397 Bonferroni age-associated CpGs and over 12 000 DMRs (q < 0.05) and selected 120 CpGs and 117 DMRs (comprising 318 CpGs) for our SEACpG and SEADMR clocks, respectively. In terms of out-of-sample performance, both SEACpG and SEADMR clocks performed well, yielding high accuracy of prediction and low error (r = 0.81; MAE = 2.2 years and r = 0.79; MAE = 2.3 years, respectively; all metrics cross-validated). Surprisingly, we found minimal overlap of methylation sites between our SEACpG and SEADMR clocks, such that only 10 CpG (<1%) of individual CpGs were present in the 318 CpGs of the 117 DMRs. This suggests that the two clocks harbor methylation sites in distinct genomic regions and may have independent utility for downstream reproductive outcome analyses. However, while both clocks were significantly associated with TTP, GA and male smoking, our SEACpG clock generated larger effect sizes in both linear and categorical analyses. Moreover, in regard to a practical application, our SEACpG clock requires ∼ 3x less CpGs compared to our SEADMR clock, and thus is more efficient and would minimize the cost of constructing a custom methylation array without any discernable differences in performance. Although it is recognized that regional methylation status may have more profound effects on downstream gene expression, our data suggest that our SEACpG clock is the preferred approach to estimate SEA to predict reproductive outcomes over our SEADMR approach. The strong and consistent relation between chronological age and DNA methylation at specific loci across individuals indicates that these age-associated changes in DNA methylation are likely not stochastic, but rather they occur at targeted regions that are more prone to epigenetic errors. The annotation of our CpG clock shows that the CpGs are enriched in CpG shores and intergenic regions, while depleted in CpG islands, exons and regions of known nucleosome retention in mature sperm. Taken together, our clock’s CpGs are distal from genes and regions of nucleosome retention, which have been linked with genes important for embryo development (Arpanahi et al., 2009; Hammoud et al., 2009). Our previous research has shown that sperm DNA methylation mediated the effect of male chronological age on poor reproductive outcomes such as fertilization rates, embryo development and live birth (Oluwayiose et al., 2021). Interestingly, consistent with our results here, we found that age-associated CpGs were depleted in retained nucleosome regions, suggesting that the effect of age via sperm DNA methylation on reproductive outcomes, such as TTP, may be independent of genic regions known to retain nucleosomes (Oluwayiose et al., 2021). Other groups have reported that age-related hypomethylated DMRs were enriched in gene regions, while hypermethylated DMRs were enriched in distal regions (Cao et al., 2020); however, we observed no such distinction upon stratification of our clock CpGs by change in methylation direction. While our study utilized gradient centrifugation to remove somatic cell contamination, future studies could employ flow cytometry to further isolate haploid sperm from leukocyte and non-haploid sperm contamination. Owing to the strong association between SEA and couples’ pregnancy probability, the slowing or reversal of SEA through lifestyle choices and/or pharmacological interventions warrants further investigation. Therefore, the characterization of the potential pathways driving the losses/gains of methylation with age offers innovative avenues of translational research to mitigate sperm aging. Moreover, as older fathers have an increased risk of offspring with a host of adverse neurological outcomes (Montgomery et al., 2004; Saha et al., 2009; Puleo et al., 2012), it is of clinical importance to understand the potential relation of SEA on offspring health and development, such as our significant findings with GA, and if it is a more precise predictor of risk of adverse offspring health. SOURCE
   https://academic.oup.com/humrep/advance-article/doi/10.1093/humrep/deac084/6583111?login=false

   Loading...