{"paper_id":"8d72c3c2-55fb-4e62-a3a2-86ac8c4749f2","body_text":"DRAFT\n1\n2\n3\n4\n5\n6\n7\n8\n9\n10\n11\n12\n13\n14\n15\n16\n17\n18\n19\n20\n21\n22\n23\n24\n25\n26\n27\n28\n29\n30\n31\n32\n33\n34\n35\n36\n37\n38\n39\n40\n41\n42\n43\n44\n45\n46\n47\n48\n49\n50\n51\n52\n53\n54\n55\n56\n57\n58\n59\n60\n61\n62\n63\n64\n65\n66\n67\n68\n69\n70\n71\n72\n73\n74\n75\n76\n77\n78\n79\n80\n81\n82\n83\n84\n85\n86\n87\n88\n89\n90\n91\n92\n93\n94\n95\n96\n97\n98\n99\n100\n101\n102\n103\n104\n105\n106\n107\n108\n109\n110\n111\n112\n113\n114\n115\n116\n117\n118\n119\n120\n121\n122\n123\n124\nLarval diapause slows adult epigenetic ageing in\nan insect model, Nasonia vitripennis\nErin E.B. Foleya, Christian L. Thomasa, Charalambos P. Kyriacoua, and Eamonn B. Mallona,1\nThis manuscript was compiled on May 22, 2025\nBRIEF\nREPORT\nEpigenetic clocks based on DNA methylation provide robust biomarkers of biological age,\nyet the mechanistic basis and functional significance of slowing these clocks remain unclear.\nProgress has been limited by the lack of short-lived, genetically tractable model organisms\nwith functional DNA methylation systems. The jewel wasp, Nasonia vitripennis, offers a\nunique solution. It combines a functional DNA methylation system with a short lifespan and\nestablished tools for experimental manipulation. We previously developed an epigenetic clock\nin Nasonia, but whether this clock reflects plastic, environmentally driven ageing processes\nwas unknown.\nHere, we test this directly by experimentally inducing larval diapause, a naturally occurring\ndevelopmental arrest triggered by environmental cues. Diapause extended median adult\nlifespan by 36% and significantly slowed the rate of epigenetic ageing. Using whole-genome\nbisulfite sequencing across multiple adult timepoints, we show that while diapaused adults\ninitially emerge epigenetically older, their subsequent epigenetic ageing proceeds 29% more\nslowly than non-diapaused controls.\nClock CpGs were enriched for gene ontology terms related to conserved nutrient-sensing\nand developmental pathways, including insulin/IGF signaling and mTOR, supporting the\nestablished mechanistic link between development and epigenetic ageing. These findings\ndemonstrate that epigenetic ageing is plastic and can be experimentally modulated by early-life\nenvironment, establishing Nasonia as a tractable system for dissecting the causal mechanisms\nof epigenetic ageing.\nSenescence plasticity | DNA methylation | Epigenetic clock | Hymenoptera\nU\nnderstanding the biology of ageing is a major scientific and societal challenge.\nEpigenetic clocks, biomarkers based on DNA methylation, have emerged\nas powerful predictors of biological age and healthspan that can outperform\nchronological age ( 1, 2). Yet despite their utility, the mechanistic basis of these\nclocks and the biological significance of slowing epigenetic ageing remain poorly\nunderstood (3).\nProgress in this area has been hindered by limitations in the current model\norganisms. While invertebrates like Drosophila melanogaster and Caenorhabditis\nelegans are invaluable for ageing research due to their genetic tractability and short\nlifespans, they lack detectable DNA methylation ( 4, 5), precluding their use for\nstudying the relevance of DNA methylation in ageing.\nThe jewel wasp, Nasonia vitripennis, overcomes this barrier by combining a short\nlifespan, a well-annotated genome and a functional DNA methylation system ( 6, 7).\nWe recently established an epigenetic clock in Nasonia, making it the first insect\nmodel with a methylation-based biomarker of ageing ( 8). However, whether this\nclock is plastic and responsive to environmental changes remains unknown.\nHere, we test the plasticity of the Nasonia epigenetic clock using diapause,\na naturally induced larval developmental arrest triggered by environmental cues.\nDiapause has been linked to extended adult lifespan and modulates conserved ageing\npathways such as insulin/IGF signalling and mTOR (\n9, 10), along with broad DNA\nmethylation reprogramming (11).\nWe show that exposure to larval diapause slows adult\nepigenetic ageing in Nasonia vitripennis, providing direct\nevidence that an invertebrate epigenetic clock is responsive to\nenvironmental inputs. This establishes Nasonia as a powerful\nnew model for dissecting the mechanisms of plastic epigenetic\nageing.\nA\nuthor affiliations: aDepartment of Genetics, Genomics\nand Cancer Sciences, University of Leicester, Univer-\nsity Road, LE1 7RH, U.K.\nEBF , CPK and EBM conceived the study. EBF\nconducted the experiments. EBF , CLT, and EBM\nperformed the data analysis. EBM wrote the initial\ndraft of the manuscript. All authors contributed to\nmanuscript revision and approved the final version.\nEBM and CLT are affiliated with Vitality Labs, a\nspinout exploring commercial applications of insect\nepigenetic ageing models. This work was conducted\nindependently of the company. All other authors\ndeclare no competing interests.\n1To whom correspondence should be addressed. E-\nmail: ebm3@le.ac.uk\nwww\n.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX PNAS — May 22, 2025 — vol. XXX — no. XX — 1–3\n.CC-BY 4.0 International licenseavailable under a \nwas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprint (whichthis version posted May 23, 2025. ; https://doi.org/10.1101/2025.05.22.655466doi: bioRxiv preprint \n\nDRAFT\n125\n126\n127\n128\n129\n130\n131\n132\n133\n134\n135\n136\n137\n138\n139\n140\n141\n142\n143\n144\n145\n146\n147\n148\n149\n150\n151\n152\n153\n154\n155\n156\n157\n158\n159\n160\n161\n162\n163\n164\n165\n166\n167\n168\n169\n170\n171\n172\n173\n174\n175\n176\n177\n178\n179\n180\n181\n182\n183\n184\n185\n186\n187\n188\n189\n190\n191\n192\n193\n194\n195\n196\n197\n198\n199\n200\n201\n202\n203\n204\n205\n206\n207\n208\n209\n210\n211\n212\n213\n214\n215\n216\n217\n218\n219\n220\n221\n222\n223\n224\n225\n226\n227\n228\n229\n230\n231\n232\n233\n234\n235\n236\n237\n238\n239\n240\n241\n242\n243\n244\n245\n246\n247\n248\nFig. 1. Diapause effect on lifespan and epigenetic age-\ning. A) Adults diapaused as larvae live longer than non-\ndiapaused conspecifics. Shaded areas represent 95%\nconfidence intervals. Dotted lines represent median survival.\nB) Diapause slows adult epigenetic ageing. Shaded areas\nrepresent 95% confidence intervals.\nMaterials and Methods\nDiapause was induced by maintaining Nasonia vitripennis\nmothers at 20°C under a 16:8 h light:dark (LD 16:8) cycle. After ten\ndays, their larval offspring are diapaused destined. Fourth instar\ndiapaused larvae were maintained at 4°C in constant darkness\n(DD) for three months before adult emergence. All experiments\nwere carried out on virgin males.\nPost-emergence, diapaused and non-diapaused (control) virgin\nmales were individually housed (25°C, 40% humidity, LD 12:12),\nfed daily with 20% sucrose solution, and their survival monitored\ndaily.\nFor whole-genome bisulfite sequencing (WGBS), individuals\nwere sampled at five timepoints post emergence (days 6, 12, 18,\n24, 30). Each of the 40 libraries (2 treatments x 5 timepoints\nx 4 replicates) represented pooled DNA from 10 individuals and\nincluded a 1% unmethylated lambda spike-in. Sequencing and\nbioinformatic processing, including read trimming, alignment,\nmethylation calling, and duplicate removal, followed our standard\npipeline ( 8). Age-related differentially methylated loci (DMLs)\nwere identified (12) and then filtered to retain CpGs strongly corre-\nlated (Pearson’s|r|≥0.3, uncorrected p≤0.05) with chronological\nage in control samples for epigenetic clock construction (13).\nAn elastic net regression was trained on age-associated CpG\nsites to predict chronological age providing an epigenetic clock.\nPredicted epigenetic age was validated against chronological age\nusing a linear model.\nFull methodological details are provided in the Supplementary\nInformation.\nResults and Discussion\nDiapause treatment extended lifespan (Figure 1A), reducing\nthe hazard of death by approximately 65% compared to\ncontrols (Cox Proportional Hazards model; Hazard Ratio\n[HR] = 0.35, 95% CI: 0.26–0.49; Wald test,\np = 5×10−9).\nThis was also reflected in the median survival determined by\nKaplan-Meier analysis which was 30 days (95% CI: 28–32\ndays) for the diapause group (n=71), versus 22 days (95%\nCI: 22–23 days) for controls (n=101).\nFrom 715,987 CpG sites classified as methylated (SI\nDataset S1), the generalized linear model (12) identified 7,950\nCpGs with significant age-related differential methylation (SI\nDataset S2). We further prioritized 289 of these sites that\nwere strongly correlated with chronological age (Pearson’s\n|r|≥0.3, uncorrected p≤0.05) as input features for an\nElastic Net regression model (SI Dataset S3). The final\noptimised model (mixing parameter α= 0.5; regularization\nλ= 1.84207, determined by 10-fold repeated cross-validation\n[3 repeats] minimizing Root Mean Squared Error [RMSE])\nutilized a concise panel of 27 CpGs to estimate epigenetic\nage (SI Dataset S4).\nOur resulting epigenetic clock accurately predicted chrono-\nlogical age in control samples, explaining 91.7% of the variance\n(cross-validatedR2; RMSE = 2.44 days). Importantly, the\nclock also performed robustly when applied to diapause\nsamples, accounting for 78.0% of chronological age variance\n(R2; RMSE = 3.98 days), demonstrating its potential\napplicability across distinct physiological conditions.\nCuriously, at day 6 post-eclosion, diapaused adults were\nepigenetically older than age-matched controls by an esti-\nmated 2.8 days (diapaused: 11.32 days vs. control: 8.53\ndays, t = -3.22, d.f. = 36, p = 0.0027,\nemmeans post\nhoc). One possible explanation is that epigenetic ageing\noccurs during the diapause period, albeit at a markedly\nreduced rate compared to adult ageing. Alternatively, the\nobserved overshoot may reflect transient remodelling of DNA\nmethylation during emergence from diapause. Distinguishing\nbetween these scenarios will require direct measurement of\nmethylation dynamics during diapause itself.\nDespite this initial increase in epigenetic age, diapaused\nadults subsequently age epigenetically 29% more slowly than\ncontrols (Figure 1B; control slope = 0.78812, diapause slope\n= 0.55828; linear model interaction of day and treatment: t\n= -3.903, d.f. = 36, p = 0.000399). By day 18, both groups\nconverge on an epigenetic age of approximately 18 days.\nHowever, by day 30, diapaused individuals are epigenetically\n2.7 days younger than controls (24.71 days vs. 27.44 days;\nemmeans post hoc: t = 3.152, d.f. = 36, p = 0.0033).\nThis striking deceleration mirrors the observed extension in\nlifespan and 65% reduction in mortality hazard in diapaused\nindividuals. Together, these results suggest that the molecular\nchanges captured by the epigenetic clock are closely aligned\nwith the physiological mechanisms that promote survival\npost-diapause.\nOur findings place diapause-induced epigenetic decelera-\ntion within the broader context of early life programming,\nwhere environmental cues can reshape long-term molecular\nand physiological trajectories. The direction and magnitude\nof changes in the epigenome largely depend on early life\nenvironmental conditions (14). This is not unexpected, as the\nepigenome is highly plastic during early development. Such\nplasticity may be adaptive, allowing early life environments to\nreconfigure the epigenome in ways that enhance future fitness,\n2 — www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX Foley et al.\n.CC-BY 4.0 International licenseavailable under a \nwas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprint (whichthis version posted May 23, 2025. ; https://doi.org/10.1101/2025.05.22.655466doi: bioRxiv preprint \n\nDRAFT\n249\n250\n251\n252\n253\n254\n255\n256\n257\n258\n259\n260\n261\n262\n263\n264\n265\n266\n267\n268\n269\n270\n271\n272\n273\n274\n275\n276\n277\n278\n279\n280\n281\n282\n283\n284\n285\n286\n287\n288\n289\n290\n291\n292\n293\n294\n295\n296\n297\n298\n299\n300\n301\n302\n303\n304\n305\n306\n307\n308\n309\n310\n311\n312\n313\n314\n315\n316\n317\n318\n319\n320\n321\n322\n323\n324\n325\n326\n327\n328\n329\n330\n331\n332\n333\n334\n335\n336\n337\n338\n339\n340\n341\n342\n343\n344\n345\n346\n347\n348\n349\n350\n351\n352\n353\n354\n355\n356\n357\n358\n359\n360\n361\n362\n363\n364\n365\n366\n367\n368\n369\n370\n371\n372\nconsistent with the predictive adaptive response hypothesis\n(15). In this framework, insect larval diapause would serve as\na predictive adaptive response: an overwintering strategy that\nanticipates a more challenging adult environment. Adults\nthat have passed through diapause may be under selection\nto survive longer, facilitating reproductive success in harsher\npost-winter conditions. In Nasonia, our data suggest that\nthis is mirrored at the molecular level by a long-term slowing\nof the epigenetic clock.\nEpigenetic clocks across diverse species consistently high-\nlight key developmental gene sets as predictors of biological\nage (16). The CpGs comprising our Nasonia epigenetic clock\nare significantly enriched for gene ontology terms related\nto conserved developmental and nutrient-sensing pathways,\nincluding mTOR and insulin/IGF signaling (SI Dataset S5).\nThese pathways are central regulators of growth, metabolism,\nand lifespan. Theoretical links between development and\nageing have a long history in evolutionary biology ( 17), and\nexperimental studies, particularly in invertebrates such as\nC. elegans, have shown that developmental alterations, such as\nentry into the dauer stage, can dramatically extend lifespan.\nOur findings align with recent work in Drosophila, where\nKang et al. (18) demonstrated that delayed development in\nprothoracicotropic hormone (PTTH)-null mutants extends\nlifespan and postpones the onset of age-related transcriptional\nchanges. In Nasonia, we observe a similar phenomenon,\nlarval diapause, a naturally induced developmental delay,\nslows the progression of the epigenetic ageing clock. Notably,\nPTTH suppression has been implicated in diapause induction\nacross diverse insect species ( 19). Together, these findings\npoint to a conserved endocrine-epigenetic axis through which\ndevelopmental timing modulates ageing trajectories.\nEpigenetic ageing is influenced by inflammation, cell\ndivision, metabolic state, and early-life environment ( 3).\nWith its compact genome, short lifespan, and functional\nmethylation system, Nasonia enables experimental dissection\nof these processes in vivo ( 6). Our findings demonstrate\nthat epigenetic ageing in this model is not only measurable,\nbut developmentally modifiable. This positions Nasonia to\naddress a fundamental translational question: can targeted\nreductions in epigenetic age improve long-term health and\nresilience?\nACKNOWLEDGMENTS. EBF was supported by a BBSRC\nMIBTP DTP studentships. CT was funded by a Leverhulme Trust\naward RPG-2020-363. EM was funded by a BBSRC Pioneer Award\nAPP3335. For the purpose of open access, the author has applied\na Creative Commons Attribution (CC BY) licence to any Author\nAccepted Manuscript version arising\n1. K Seale, S Horvath, A Teschendorff, N Eynon, S Voisin, Making Sense of the Ageing\nMethylome. Nat. Rev. Genet. 23, 585–605 (2022) Publisher: Nature Publishing Group.\n2. L Drew, Turning back time with epigenetic clocks. Nature 601, S20–S22 (2022)\nBandiera abtest: a Cg type: Outlook Number: 7893 Publisher: Nature Publishing Group\nSubject term: Ageing, Society, Epigenetics.\n3. CG Bell, et al., DNA methylation aging clocks: challenges and recommendations. Genome\nBiol. 20, 249 (2019).\n4. F Lyko, R Maleszka, Insects as Innovative Models for Functional Studies of DNA\nMethylation. T rends genetics27, 127–131 (2011).\n5. CW Hu, JL Chen, YW Hsu, CC Y en, MR Chao, Trace analysis of methylated and\nhydroxymethylated cytosines in DNA by isotope-dilution LC-MS/MS: first evidence of DNA\nmethylation in Caenorhabditis elegans. The Biochem. J. 465, 39–47 (2015).\n6. JH Werren, et al., Functional and evolutionary insights from the genomes of three parasitoid\nNasonia species. Sci. (New Y ork, N.Y .)327, 343–348 (2010).\n7. X Wang, et al., Function and Evolution of DNA Methylation in Nasonia vitripennis. PLOS\nGenet. 9, e1003872 (2013).\n8. K Brink, CL Thomas, A Jones, TW Chan, EB Mallon, Exploring the ageing methylome in the\nmodel insect, Nasonia vitripennis. BMC Genomics 25, 305 (2024).\n9. DL Denlinger, Why study diapause? Entomol. Res. 38, 1–9 (2008) eprint:\nhttps://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1748-5967.2008.00139.x.\n10. X Karp, Hormonal Regulation of Diapause and Development in Nematodes, Insects, and\nFishes. Front. Ecol. Evol. 9 (2021) Publisher: Frontiers.\n11. M Pegoraro, A Bafna, NJ Davies, DM Shuker, E Tauber, DNA methylation changes induced\nby long and short photoperiods in Nasonia. Genome Res. 26, 203–210 (2016).\n12. Y Park, H Wu, Differential methylation analysis for BS-seq data under general experimental\ndesign. Bioinformatics 32, 1446–1453 (2016).\n13. AE Teschendorff, S Horvath, Epigenetic ageing clocks: statistical methods and emerging\ncomputational challenges. Nat. Rev. Genet. 26, 350–368 (2025) Publisher: Nature\nPublishing Group.\n14. A Vaiserman, Developmental Tuning of Epigenetic Clock. Front. Genet. 9 (2018).\n15. P Bateson, P Gluckman, M Hanson, The biology of developmental plasticity and the\nPredictive Adaptive Response hypothesis. The J. Physiol. 592, 2357–2368 (2014).\n16. D Gems, RS Virk, JP de Magalh ˜aes, Epigenetic clocks and programmatic aging. Ageing\nRes. Rev. 101, 102546 (2024).\n17. JP de Magalh ˜aes, Programmatic features of aging originating in development: aging\nmechanisms beyond molecular damage? The FASEB J. 26, 4821–4826 (2012).\n18. P Kang, et al., NF-κB-mediated developmental delay extends lifespan in Drosophila. Proc.\nNatl. Acad. Sci. 122, e2420811122 (2025) Publisher: Proceedings of the National Academy\nof Sciences.\n19. Q Wang, AAM Mohamed, M Takeda, Serotonin Receptor B May Lock the Gate of PTTH\nRelease/Synthesis in the Chinese Silk Moth, Antheraea pernyi ; A Diapause\nInitiation/Maintenance Mechanism? PLOS ONE 8, e79381 (2013).\nFoley et al. PNAS — May 22, 2025 — vol. XXX — no. XX — 3\n.CC-BY 4.0 International licenseavailable under a \nwas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprint (whichthis version posted May 23, 2025. ; https://doi.org/10.1101/2025.05.22.655466doi: bioRxiv preprint \n\n1\nSupporting Information for2\nLarval diapause slows adult epigenetic ageing in an insect model, Nasonia vitripennis3\nErin B. Foley, Charalambos P. Kyriacou, Christian L. Thomas & Eamonn B. Mallon4\nEamonn B. Mallon.5\nE-mail: ebm3@le.ac.uk6\nThis PDF file includes:7\nSupporting text8\nLegends for Dataset S1 to S59\nSI References10\nOther supporting materials for this manuscript include the following:11\nDatasets S1 to S512\nErin B. Foley, Charalambos P. Kyriacou, Christian L. Thomas & Eamonn B. Mallon 1 of 4\n.CC-BY 4.0 International licenseavailable under a \nwas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprint (whichthis version posted May 23, 2025. ; https://doi.org/10.1101/2025.05.22.655466doi: bioRxiv preprint \n\nSupporting Information Text13\nMethods14\nRearing. Nasonia vitripennis used in this study were from the Leicester strain, a laboratory colony maintained at the University15\nof Leicester for over nine years. This strain derives from AsymC, originally isolated in 1989 and subsequently cured ofWolbachia16\nvia heat shock treatment (1,2).17\nTo generate diapaused offspring, virgin females were housed at 20°C under a 16:8 h dark:light photoperiod. Host pupae18\nwere collected from Day 10 post-oviposition onwards to ensure recovery of diapaused larvae. These larvae developed to the19\nfourth instar before being transferred to continuous darkness at 4°C for three months. After the diapause period, larvae were20\nreturned to standard rearing conditions and allowed to complete development to adulthood.21\nLifespan Experiments.Diapaused and non-diapaused virgin maleNasonia were collected in batches within 24 hours of adult22\nemergence (Day 0) and housed individually in plastic tubes. All individuals were maintained at 25°C and 40% relative humidity23\nunder a 12:12 h light:dark cycle. Each wasp received a daily feeding of 20µL of 20% sucrose solution on filter paper, and24\nmortality was recorded daily. Lifespan was defined as the number of days from adult emergence to death.25\nKaplan–Meier survival analyses were performed using the survival package (v3.7) (3) and visualized with the survminer26\npackage (v0.4.9) (4). Cox proportional hazards models were fitted using the survival package. All analyses were conducted in27\nR v4.4.1 (5).28\nLifespan data were used to select sampling time points for whole-genome bisulfite sequencing (WGBS). Diapaused and29\nnon-diapaused virgin males were sampled at Days 6, 12, 18, 24, and 30, always at the same time of day. Upon collection,30\nindividuals were snap-frozen in liquid nitrogen and stored at –80°C until DNA extraction.31\nDNA Extraction.DNA was extracted using an adapted protocol based on the AllPrep® DNA/RNA Micro Kit (Qiagen). Each32\nsample consisted of ten whole-body adult virgin maleNasonia. Four biological replicates were processed per time point (Days33\n6, 12, 18, 24, and 30).34\nDNA quality and concentration were assessed using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific), a35\nQubitTM dsDNA BR Assay Kit (Thermo Fisher Scientific), and electrophoresis on a 1% agarose gel run for 40 minutes at36\n100V. Samples were sent to Novogene (Beijing) for whole-genome bisulfite sequencing (WGBS). A 1% unmethylated lambda37\nDNA spike-in was included in each sample to assess bisulfite conversion efficiency.38\nWhole-Genome Bisulfite Sequencing and Bioinformatic Processing. Raw sequencing data were provided in FASTQ format39\n(submitted to Sequence Read Archive (SRA)). For samples sequenced across multiple lanes, read files were concatenated using40\nthe Unix cat command. Read quality was assessed using FastQC v0.12.1 (6).41\nA custom Snakemake pipeline (Snakemake v7.32.4; Python v3.12.1) (7) was used for preprocessing. Adapter sequences were42\ntrimmed and the first 10 bases removed from each read using Cutadapt v4.4 (8). Paired-end reads were aligned to theNasonia43\nvitripennis reference genome (Nvit_PSR1.1) (9) using Bowtie2 v2.5.1 with default parameters (10). Reads were also aligned to44\nthe unmethylated lambda genome (RefSeq accession: GCF_000840245.1) to assess bisulfite conversion efficiency.45\nAligned reads were deduplicated and cytosine methylation calls extracted using Bismark v0.22.3 (11). Strand ambiguity was46\nresolved using thecoverage2cytosine utility in Bismark to generate destranded coverage files, which were used for downstream47\nanalyses.48\nMethylation Analysis.All downstream analyses were performed in R v4.4.1 (5). Destranded CpG coverage files were imported49\ninto the MethylKit package v1.30.0 (12) using themethRead() function. CpG sites with less than 10× coverage or coverage50\nabove the 99th percentile were filtered out. A binomial test was applied to each sample using the lambda genome conversion51\nrate as the null probability of success, with a false discovery rate (FDR) threshold ofp < 0.05 (SI Dataset S5). Only CpG sites52\nshowing significant methylation in at least one sample were retained. Percentage methylation at each site was calculated using53\nthe percMethylation() function.54\nDifferential Methylation Analysis.Differential methylation analysis was performed using theDSS package v2.52.0 (13), which55\nmodels methylation proportions using a beta-binomial generalized linear model (GLM) with an arcsine link function. A design56\nmatrix was constructed incorporating time point and treatment (diapause vs. non-diapause) as experimental factors. Linear57\nmodels were fitted using theDMLfit() function to evaluate main effects and interactions. Differentially methylated loci (DMLs)58\nand regions (DMRs) were identified accordingly.59\nGenomic features were assigned to CpG sites using a custom annotation file (GFF format) generated by Dr. Hollie Marshall60\nusing AGAT v0.10.0 (14).61\nEpigenetic Clock Construction.An elastic net regression model was trained to predict chronological age from DNA methylation62\nlevels across CpG sites, using data from virgin maleNasonia vitripennis sampled at five time points (Days 6, 12, 18, 24, and63\n30). Methylation data (percentage values) were filtered to retain CpG sites previously identified as differentially methylated64\nacross time usingDSS (13).65\nTo enhance model sparsity and robustness, CpG sites were pre-selected based on univariate Pearson correlation with age in66\nthe non-diapaused (control) group. Sites with absolute correlation≥ 0.3 and uncorrectedp-value≤ 0.05 were retained. CpG67\nfeatures were then centered and scaled.68\n2 of 4 Erin B. Foley, Charalambos P. Kyriacou, Christian L. Thomas & Eamonn B. Mallon\n.CC-BY 4.0 International licenseavailable under a \nwas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprint (whichthis version posted May 23, 2025. ; https://doi.org/10.1101/2025.05.22.655466doi: bioRxiv preprint \n\nThe final model was implemented using theglmnet algorithm with elastic net regularization (α = 0.5), as part of thecaret69\nframework (15). Ten-fold cross-validation with three repeats was used for model tuning and performance estimation. The70\ntreatment group (diapause vs. control) was included as a categorical predictor, encoded as a dummy variable, and entered71\nalongside the CpG methylation features. The optimal model was selected based on minimum root mean squared error (RMSE).72\nPredictions were generated for all samples, and model performance was assessed using RMSE andR2, stratified by treatment73\ngroup.74\nPost-hoc analysis of predicted age trajectories was conducted using linear models with interaction terms, and estimated75\nmarginal means were compared between treatment groups at multiple time points using theemmeans package (16).76\nGene Ontology Enrichment Analysis.Gene Ontology (GO) enrichment analysis was performed to assess the functional signifi-77\ncance of genes associated with CpG sites included in the final epigenetic clock model. The 27 clock CpGs (SI Dataset S4)78\nwere mapped to nearby genes, which were then tested for enrichment against a background set comprising all genes associated79\nwith differentially methylated loci across time (SI Dataset S1). GO annotations were derived fromNasonia vitripennis and80\nformatted for compatibility withGOstats (17).81\nThe gene list was tested for overrepresentation of GO terms in the Biological Process (BP), Cellular Component (CC), and82\nMolecular Function (MF) ontologies using a conditional hypergeometric test. Analyses were performed using theGOstats and83\nGSEABase packages. For each ontology, both over- and under-representation were tested, and GO terms with an adjusted84\nFDR of< 0.05 (Benjamini-Hochberg) were considered significant.85\nTo visualise and summarise redundant GO terms, semantic similarity clustering was performed using therrvgo package (18).86\nPairwise GO term similarities were calculated using theorg.Dm.eg.db annotation database and the “Rel” semantic similarity87\nmeasure. Representative GO terms were identified by reducing the similarity matrix with a similarity threshold of 0.7. Enriched88\nterms were visualised using treemaps, heatmaps, and scatter plots. Final results were exported in tabular form (SI Dataset S5).89\nAll scripts used are available athttps://tinyurl.com/5n6vcvsk90\nSI Dataset S1 (SI_Datasets.xlsx/S1_Methylated_CpGs)91\nCpG sites classified as methylated using a binomial test with the lambda genome conversion rate as the null probability of92\nsuccess. Sites passing a false discovery rate (FDR) threshold ofp< 0.05 and significantly methylated in at least one sample93\nwere retained.94\nSI Dataset S2 (SI_Datasets.xlsx/S2_Age_DMPs)95\n7,950 CpG sites showing significant age-associated differential methylation.96\nSI Dataset S3 (SI_Datasets.xlsx/S3_Age_Correlated)97\n289 age-associated CpG sites with strong correlation to chronological age (Pearson’s|r|≥ 0.3, uncorrectedp≤ 0.05).98\nSI Dataset S4 (SI_Datasets.xlsx/S4_Clock_Coefficients)99\nElastic net model coefficients for the 27 CpG sites and intercept used to estimate epigenetic age.100\nSI Dataset S5 (SI_Datasets.xlsx/S5_GO_terms)101\nEnriched gene ontology terms associated with the 27 CpG sites comprising the epigenetic clock.102\nReferences103\n1. JH Werren, DW Loehlin, The parasitoid wasp Nasonia: an emerging model system with haploid male genetics.Cold104\nSpring Harb. Protoc. 2009, pdb–emo134 (2009) Publisher: Cold Spring Harbor Laboratory Press.105\n2. DC Darling, JH Werren, Biosystematics of Nasonia (Hymenoptera: Pteromalidae): two new species reared from birds’106\nnests in North America.Annals Entomol. Soc. Am. 83, 352–370 (1990) Publisher: Oxford University Press Oxford, UK.107\n3. Terry M. Therneau, Patricia M. Grambsch,Modeling Survival Data: Extending the Cox Model. (Springer, New York),108\n(2000).109\n4. A Kassambara, M Kosinski, P Biecek,survminer: Drawing Survival Curves using ’ggplot2’ . (2021).110\n5. R Core Team,R: A Language and Environment for Statistical Computing. (R Foundation for Statistical Computing,111\nVienna, Austria), (2024).112\n6. S Andrews, et al., FastQC (2012) Place: Babraham, UK Published: Babraham Institute.113\n7. J Köster, S Rahmann, Snakemake—a scalable bioinformatics workflow engine.Bioinformatics 28, 2520–2522 (2012)114\n_eprint: https://academic.oup.com/bioinformatics/article-pdf/28/19/2520/48879301/bioinformatics_28_19_2520.pdf.115\n8. M Martin, Cutadapt removes adapter sequences from high-throughput sequencing reads.EMBnet.journal 17, 10–12116\n(2011).117\n9. E Dalla Benetta, et al., Genome elimination mediated by gene expression from a selfish chromosome.Sci. Adv. 6, eaaz9808118\n(2020) Publisher: American Association for the Advancement of Science.119\n10. B Langmead, SL Salzberg, Fast gapped-read alignment with Bowtie 2.Nat. methods 9, 357–359 (2012) Publisher: Nature120\nPublishing Group.121\nErin B. Foley, Charalambos P. Kyriacou, Christian L. Thomas & Eamonn B. Mallon 3 of 4\n.CC-BY 4.0 International licenseavailable under a \nwas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprint (whichthis version posted May 23, 2025. ; https://doi.org/10.1101/2025.05.22.655466doi: bioRxiv preprint \n\n11. F Krueger, SR Andrews, Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications.bioinformatics122\n27, 1571–1572 (2011) Publisher: Oxford University Press.123\n12. A Akalin, et al., methylKit: a comprehensive R package for the analysis of genome-wide DNA methylation profiles.124\nGenome biology 13, 1–9 (2012) Publisher: Springer.125\n13. H Feng, H Wu, Differential methylation analysis for bisulfite sequencing using DSS.Quant. Biol. 7, 327–334 (2019)126\nPublisher: Springer.127\n14. J Dainat, AGAT: Another Gff Analysis Toolkit to handle annotations in any GTF/GFF format.(Version v0. 7.0).Zendo.128\ndoi 10 (2023).129\n15. M Kuhn, Building Predictive Models in R Using the caret Package.J. Stat. Softw. 28, 1–26 (2008).130\n16. RV Lenth, emmeans: Estimated marginal means, aka least-squares means, manual (2022).131\n17. S Falcon, R Gentleman, Using GOstats to test gene lists for GO term association.Bioinformatics 23, 257–258 (2007)132\nPublisher: Oxford University Press.133\n18. S Sayols, rrvgo: a Bioconductor package for interpreting lists of Gene Ontology terms.microPublication Biol. (2023)134\nPublisher: Caltech Library.135\n4 of 4 Erin B. Foley, Charalambos P. Kyriacou, Christian L. Thomas & Eamonn B. Mallon\n.CC-BY 4.0 International licenseavailable under a \nwas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprint (whichthis version posted May 23, 2025. ; https://doi.org/10.1101/2025.05.22.655466doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}