Multigenerational downregulation of insulin/IGF‐1 signaling in adulthood improves lineage survival, reproduction, and fitness in Caenorhabditis elegans supporting the developmental theory of ageing

Abstract Adulthood‐only downregulation of insulin/IGF‐1 signaling (IIS), an evolutionarily conserved pathway regulating resource allocation between somatic maintenance and reproduction, increases life span without fecundity cost in the nematode, Caenorhabditis elegans. However, long‐term multigenerational effects of reduced IIS remain unexplored and are proposed to carry costs for offspring quality. To test this hypothesis, we ran a mutation accumulation (MA) experiment and downregulated IIS in half of the 400 MA lines by silencing daf‐2 gene expression using RNA interference (RNAi) across 40 generations. Contrary to the prediction, adulthood‐only daf‐2 RNAi reduced extinction of MA lines both under UV‐induced and spontaneous MA. Fitness of the surviving UV‐induced MA lines was higher under daf‐2 RNAi. Reduced IIS increased intergenerational F1 offspring fitness under UV stress but had no quantifiable transgenerational effects. Functional hrde‐1 was required for the benefits of multigenerational daf‐2 RNAi. Overall, we found net benefit to fitness from multigenerational reduction of IIS and the benefits became more apparent under stress. Because reduced daf‐2 expression during development carries fitness costs, we suggest that our findings are best explained by the developmental theory of ageing, which maintains that the decline in the force of selection with age results in poorly regulated gene expression in adulthood.

support for trade-offs between somatic maintenance and reproduction, growing empirical work questions the universality of such resource allocation trade-offs in the evolution of ageing (Edward and Chapman 2012;Maklakov and Immler 2016;Maklakov and Chapman 2019;Flatt 2020). Several studies have shown that experimentally increased life span, often via the downregulation of genes in nutrient-sensing signaling pathways in adulthood, is not detrimental to reproduction Flatt 2011;Maklakov and Immler 2016;Lind et al. 2019;Maklakov and Chapman 2019). These results are in line with the hypothesis that a gradual decline in selection gradients with advancing age after reproductive maturity results in physiological processes that are optimized for successful early-life reproduction, but start causing harm in later life, contributing to somatic deterioration (de Magalhaes and Church 2005;de Magalhaes 2012;Blagosklonny 2013;Ezcurra et al. 2018;Wang et al. 2018;Maklakov and Chapman 2019;Sala et al. 2020;Carlsson et al. 2021).
However, it has been proposed that reduced offspring quality, for example, via reduced germline maintenance, can be a hidden cost of life span extension via reduced nutrient-sensing signaling (Maklakov and Immler 2016). Germline maintenance, the repair and surveillance of genomic and proteomic integrity in germline stem cells and gametes, is energetically expensive (Maklakov and Immler 2016;Berger et al. 2017) and germline signaling plays a key role in resource allocation to somatic maintenance (Kenyon 2010;Labbadia and Morimoto 2015;Maklakov and Immler 2016;Berger et al. 2017). Germline ablation results in increased somatic maintenance and life span in Drosophila melanogaster fruitflies (Flatt et al. 2008) and Caenorhabditis elegans nematodes (Hsin and Kenyon 1999;Arantes-Oliveira et al. 2002;Antebi 2013;Ermolaeva et al. 2013;Shemesh et al. 2013), although life span extension in C. elegans requires an intact somatic gonad (Hsin and Kenyon 1999). It is important to note that the requirement of the intact somatic gonad for life span extension in germline-less worms does not negate the possibility of a resource allocation trade-off as sometimes implied, but only shows that germline signaling is required to mediate the effect (Arantes-Oliveira et al. 2002;Flatt et al. 2008;Maklakov and Immler 2016). Similarly, recent work in Danio rerio zebrafish suggests that germline ablation increases somatic maintenance under stress (Chen et al. 2020). Furthermore, nutritional stress in C. elegans results in germline reduction and increased life span (Thondamal et al. 2014). When soma-togermline communication is disrupted, the number of germ cells is unaffected and life span extension is abolished (Thondamal et al. 2014). Taken together, these results suggest that germline maintenance is costly and can trade-off with life-history traits such as life span and offspring number (Maklakov and Immler 2016;Berger et al. 2017). The corollary is that increased investment in somatic maintenance and longevity can trade-off with offspring quality by increasing germline mutation load in offspring.
We tested this hypothesis by focusing on life span extension via reduced insulin/IGF-1 signaling (IIS) in C. elegans. IIS is an evolutionarily conserved pathway that regulates the physiological response of organisms to their environment (Regan et al. 2019). IIS mechanistically links nutrient intake with development, growth, reproduction, and life span across diverse taxa (Fontana et al. 2010;Kenyon 2010;Regan et al. 2019). Reduced IIS, via genetic and environmental interventions, consistently extends life span (Kenyon 2010). Mutations that reduce the function of daf-2 gene in C. elegans, the homolog of human insulin/IGF-1 receptor, increase life span and reduce fitness (Chen et al. 2007). However, previous work showed that reducing IIS in adulthood via daf-2 RNA interference (RNAi) knockdown in a single generation increases life span without a cost to offspring number , quality (Lind et al. 2019), or fitness in variable environments . It is possible, however, that multigenerational daf-2 knockdown will expose a hidden cost to offspring number and/or offspring quality via accumulation of germline mutations. Here, we tested this prediction via adulthood-only daf-2 RNAi knockdown for 40 consecutive generations in 400 spontaneous and UV-induced mutation accumulation (MA) lines. We found that multigenerational adulthood-only daf-2 knockdown reduced lineage extinction under both spontaneous and UV-induced MA and increased fitness of remaining lineages under UV-induced MA. Reduced IIS protected MA lines from infertility, egg hatching failure, and developmental abnormalities and increased offspring fitness under UV irradiation, but we found no evidence of transgenerational effects of adulthoodonly daf-2 RNAi.

Materials and Methods
Supplementary methods details are included in the Supporting Information.

NEMATODE STOCKS AND CULTURE
The nematode (roundworm) C. elegans N2 wild-type (Bristol) and heritable RNAi deficiency 1 (hrde-1) mutant strains were defrosted from stocks acquired from Caenorhabditis Genetics Center (University of Minnesota, USA, funded by NIH Office of Research Infrastructure Programs, P40 OD010440) and from the lab of Prof. Eric Miska (Gurdon Institute, University of Cambridge, UK), respectively, and stored at −80°C until use.
Defrosted C. elegans strains were reared through two generations prior to setup, on NGM agar supplemented with the fungicide nystatin and antibiotics streptomycin and ampicillin, to prevent infection (each at 100 µg/mL, as standard recipes, e.g., Lionaki and Tavernarakis 2013) and seeded with the antibioticresistant Escherichia coli bacterial strain OP50-1 (pUC4K, from J. Ewbank at the Centre d'Immunologie de Marseille-Luminy, France). We bleached eggs from the grandparents of experimental individuals, to standardize parental age and remove any infection or temperature effects from defrosting, prior to experiments.

ADULTHOOD
To downregulate adulthood expression of the insulin-like sensing signaling receptor homolog, daf-2, we fed late-L4 larvae with E. coli bacteria expressing daf-2 double-stranded RNA, which decreases mRNA levels of the complementary transcribed daf-2 systemically (Fire et al. 1998;Timmons et al. 2001). The daf-2 gene is upstream of and inhibits the action of master regulator daf-16/FOXO. RNase-III deficient, IPTG-inducible HT115 E. coli bacteria with an empty plasmid vector (L4440) was used as the control (as Timmons et al. 2001;Dillin et al. 2002;Lind et al. 2019). RNAi clones were acquired from the Vidal feeding library (Source BioScience, created by M. Vidal lab, Harvard Medical School, USA) and all clones were verified via sequencing prior to delivery.

ULTRAVIOLET WAVELENGTH C IRRADIATION
To induce mutations and stress, we UV-irradiated Day 2 adults, with a calibrated ultraviolet-C (UV-C) radiation dose of 46 J/m 2 (wavelength 254nm), via 20 s exposure to the UV-C radiation emitted from the lamp of a Thermo Scientific Heraguard ECO Safety Cabinet (calibration details in Supplementary Methods). This dose is in the range of previous UV-C irradiation doses for C. elegans adults or eggs (Meyer et al. 2007;Stergiou et al. 2007;Hyun et al. 2008;Boyd et al. 2010). Our pilot work showed that this dose reduced the fecundity of Day 2 adults laying at 20°C by 61% compared with unexposed sham controls (n = 30 worms/treatment, mean Day 2 offspring count = 35 for UV, 90 for non-UV; full data not shown). Nonirradiated control worms received a sham-irradiation, by being positioned in identical orientation under the UV-C lamp for 20 s, while it was switched off.

MUTATION ACCUMULATION LINES
To determine the effects of reduced adulthood IIS, via daf-2 RNAi, on spontaneous and UV-induced MA, we established 800 MA lines, across two genetic backgrounds. The eight experimental treatments were the full-factorial combinations of genotype (N2 or hrde-1 mutant), RNAi treatment (empty vector or daf-2 RNAi), and irradiation (UV or sham), with 100 MA lines per treatment. We ran each of the eight treatments in parallel, divided into two time-staggered independent blocks of 50 lines per treatment, for logistical reasons and to capture any between-block variation. We tracked MA lines until the majority of UVirradiated lines were extinct-40 generations for N2 and 25 generations for hrde-1 mutants. Every MA generation, daf-2 RNAi or empty vector was delivered from the start of adulthood onward, and all lines developed on empty vector. More details are provided in the Supporting Information.
Each MA line was propagated as a single individual hermaphrodite per generation, to create successive genetic bottlenecks, allowing de novo deleterious mutations to accumulate in the relative absence of selection (Keightley and Bataillon 2000;Vassilieva et al. 2000;Ajie et al. 2005), if they had no effect on developmental time or viability, and in the absence of mating. MA is a common approach used in several species, to study the evolutionary genetics of de novo mutation rates (Halligan and Keightley 2009). In wild-type C. elegans, the spontaneous mutation rate in MA lines is positively correlated with the rate of fitness decay (Baer et al. 2005).

RNAi
We quantified, via quantitative reverse transcriptase polymerase chain reaction (qRT-PCR), the extent of downregulation of daf-2 expression in Day 2 adults from the MA lines at generation 20, following adulthood-only daf-2 RNAi knockdown, relative to generation 20 empty vector control MA lines. A sample from all eight MA treatments above that had been frozen at generation 20 were used (mean n = 6 MA lines per treatment, with mean n = 3 individual worms per MA line; Table S1). Previous work has shown that adulthood daf-2 RNAi reduces daf-2 expression in Day 2 N2 adults by 50%, relative to controls , equivalent to the first generation of MA. The aim was to test the degree of daf-2 knockdown after 20 generations of daf-2 RNAi and MA in both N2 and hrde-1 mutant backgrounds. Worms for the gene expression assay were maintained under standard conditions (as the main experiment) and were reared for two generations on empty vector prior to the assay, to remove freezing effects and age-synchronize. RNAi was applied from late-L4 stage. Further method details are included in the Supporting Information.

SURVIVING MA LINES
To test for life-history differences between the N2 wild-type MA lines after 20 generations of MA and of adulthood daf-2 RNAi, we assayed age-specific reproduction, egg size, male production, and adult heat shock resistance (n = 100 grand-offspring from each of the four N2 MA treatments). Further details are provided in the Supporting Information.
Individual age-synchronized, late-L4 hermaphrodites were set up on separate empty vector plates, labeled with a unique identifier, to blind experimenters to treatment identity, and thus avoid bias. The two experimenters conducted the life-history assay in two simultaneous time blocks and each block had an identical and equal representation of individual worms from all four treatments (n = 50 per treatment per block).
For all experiments, we assayed age-specific offspring production (fecundity) over the first 6 days of adult life (the reproductive window for C. elegans hermaphrodites), by transferring to fresh plates every 24 h, to acquire daily reproduction counts, following the standard temporal resolution for C. elegans studies Lind et al. 2019;Scharf et al. 2021). Plates were scored for viable adult offspring 2.5 days later. Individual fitness (lambda) was calculated-a measure analogous to the intrinsic rate of population growth, which is weighted by early reproduction and fast development (Stearns 1992;Brommer et al. 2002;Lind et al. 2016;Lind et al. 2019). Total lifetime reproduction (lifetime reproductive success) was calculated for each individual as the sum of age-specific offspring counts across the first 6 days of adulthood.
To determine if reduced IIS influenced parental resource allocation into eggs under MA, we measured egg area (in mm 2 ), with 75 samples per treatment. Eggs measured were laid by the individuals in the generation 20 age-specific reproduction assay at peak reproduction (Day 2). Egg size is commonly used as a proxy for parental investment (Hibshman et al. 2016;Lind et al. 2019). We photographed one egg laid per individual, within 2 h of lay, under 12× magnification light microscopy (Leica M165C with Lumenera Infinity 2-7C digital microscope camera).

BACKGROUNDS
To determine whether daf-2 RNAi was transferred transgenerationally via the germline, we assayed, in a separate experiment, whether offspring fitness benefits persisted across three generations of offspring (F1, F2, and F3) from Day 2 parents treated with daf-2 RNAi or empty vector control throughout adulthood, in N2 wild-type and hrde-1 mutant backgrounds. Offspring of each generation were maintained on empty vector (n = 30 individuals per treatment and per genotype) and assayed for daily reproduction.

FASTING
To determine whether there was a reproductive cost of adulthoodonly daf-2 RNAi under resource limitation, we exposed N2 wild-type adults to either an intermittent fasting (IF) regime consisting of 9 h of starvation on Days 1, 3, 5, and 7 of adulthood, or ad libitum (AL) feeding throughout life, and assayed daily reproduction. Full details are in the Supporting Information.

STATISTICAL ANALYSES
All analyses were conducted in R version 4.0.0 (R Core Team 2020). Fitness and total lifetime reproduction data were plotted using the R package "dabestr" (data analysis using bootstrapcoupled estimation [Ho et al. 2019]), for visualization.
Parental survival in the intergenerational (first) experiment was analyzed using a Cox proportional hazards mixed-effects model (Therneau 2019) ("coxme"), fitting RNAi, UV, and their interaction as fixed effects and plate as a random effect. Separate analyses were conducted with matricides either classed as deaths or as censors. The age at death response variable contained a coding variable to distinguish deaths from censored individuals (due to accidental losses). Extinction (survival trajectories) under MA was analyzed using Cox proportional hazards regression analysis ("coxph" function in "survival" package). The generation number at which each MA line went extinct was the response variable (instead of age at death) and this contained a coding variable to distinguish extinctions from censors, as above. To test the effect of experimental block or genotype (either N2 wildtype or hrde-1 mutant) on extinction and on the interaction between RNAi and UV effects under MA, maximal models were fitted with either a three-way RNAi × UV × block interaction or a RNAi × UV × genotype interaction, and stepwise model simplification was conducted (Bolker et al. 2009). Block was included as a fixed effect to test for repeatability between the two blocks. For all survival analyses, the z-scores and P-values were reported, with the z-score defined as the log of the hazard ratio (risk of death, exp(coef)) divided by the standard error of the log hazard ratio.
Age-specific reproduction was analyzed using generalized linear mixed-effects models to account for temporal pseudoreplication of repeated measures on the same individuals across lifetime, with the template model builder package ("glmmTMB" in R [Brooks et al. 2017;Brooks et al. 2019]). Models fitted with Poisson, Zero-Inflated Poisson, Generalized Poisson, and Zero-Inflated Generalized Poisson error structure were compared using AIC values of model fit ("AICtab" function in "bbmle" package), following Brooks et al. (2019), to account for under-or overdispersion and for zero-inflation, when it was found to occur in simulated residuals generated with the "DHARMa" package (Hartig 2020). Age and a quadratic form of age (age 2 ) were fitted as fixed effects in both the conditional and zero-inflation model formula, and significance assessed in each case. The age 2 term controlled for the curved (nonlinear) trajectory of reproduction across age (Bates et al. 2015). UV irradiation, RNAi treatment, and their three-way interaction with age or age 2 were fitted as fixed effects, and experimental block, observer, and a unique plate identifier as random effects, the latter to control for repeated measures of fecundity from the same individuals across lifetime. Genotype was substituted with UV as a fixed effect to assess the transgenerational effects of daf-2 RNAi on N2 versus hrde-1 mutants. Models that did not converge were not included in model comparison, and the converging model with best AIC fit was presented (as Brooks et al. 2019).
Fitness (lambda) was calculated using the "lambda" function in the "popbio" package. More details are provided in the Supporting Information. Fitness, total reproduction, and egg size data were analyzed using a generalized linear model (GLM) with Gaussian error structure ("lm" function in "stats" package). Locomotion, a measure of stress resistance to heat shock, was coded as a binary response variable (normal or abnormal movement) and analyzed using a GLM with binomial error structure.
However, daf-2 RNAi significantly increased offspring fitness both under benign conditions and when parents were UV irradiated (lambda, GLM: RNAi: t = −2.305, df = 1, P = 0.0222; Fig. 1b). This absence of fitness costs for the offspring of UV-irradiated parents, despite the costs of irradiation for parental reproduction, suggests that irradiated parents may have invested in germline repair mechanisms, which improved the fitness of their offspring (Smelick and Ahmed 2005;Maklakov and Immler 2016;Chen et al. 2020).
Our findings show that under adulthood-only daf-2 RNAi, parents increase investment into somatic maintenance resulting in increased life span with no survival or reproductive cost to themselves or their offspring under benign conditions validating previous work in C. elegans Lind et al. 2019). Importantly, we reveal that the absence of a longevity-fecundity trade-off in parents persisted under stressful conditions, when organisms have to invest into repairing UV-induced damage. Previous work in daf-2 mutants has also shown them to be more resistant to UV irradiation (Hyun et al. 2008).
Furthermore, we show that daf-2 knockdown in adult parents primarily influenced the timing of reproduction in their offspring. Specifically, it caused a shift to increased early-life reproduction, improving offspring individual fitness, rather than an increase in total reproduction.
Timing of reproduction in both laboratory and natural C. elegans populations is important as C. elegans reproduces very fast and so the strength of natural selection (selection gradient) declines very rapidly with advancing age Frezal and Felix 2015). Under standard lab conditions, self-fertilizing C. elegans hermaphrodites complete most of their reproduction within the first 3 days of adulthood and in nature, C. elegans exhibit boom-and-bust population dynamics involving periods of very rapid population growth, driven by an ephemeral food supply (Frezal and Felix 2015;Schulenburg and Felix 2017). As individual fitness is heavily weighted by early reproduction and fast development (Supporting Information), even relatively small increases in Day 1 and Day 2 reproduction (Fig. S2) can improve individual fitness, despite having no effect on total lifetime reproduction, if later reproduction is correspondingly reduced.
Although an earlier study found increased total reproduction in the first generation of offspring from C. elegans parents treated with daf-2 RNAi, across N2 wild-type and two other genetic backgrounds, this was also accompanied by increased offspring fitness (Lind et al. 2019), in agreement with our results here. Under UV-induced mutagenesis, the costs of germline maintenance would be expected to become more apparent (Sniegowski et al. 2000;Berger et al. 2017;Chen et al. 2020). However, we show that adulthood-only daf-2 RNAi improves fitness of F1 offspring even when parents are under stress. We may have expected a reproductive cost of adulthood daf-2 RNAi to become apparent under nutritional or UV irradiation stress, but instead we show no cost to fitness or total lifetime reproduction of adulthood daf-2 RNAi under IF (Fig. 2) or UV irradiation (Fig. S1). This is in agreement with previous work that showed daf-2 RNAi in adult N2 C. elegans did not reduce lambda fitness under a variable environment of light and temperature stress ).

PROTECTS AGAINST EXTINCTION
We found that daf-2 RNAi reduced extinction under both UVinduced and spontaneous MA across 40 generations (Cox proportional hazards regression analysis, coxph, RNAi: z = 2.159, P = 0.031; UV: z = 11.469, P < 0.001; RNAi × UV: z = −0.537, P = 0.591; Fig. 3a). Extinction results did not differ significantly between two independent experimental blocks (coxph, z = −0.942, P = 0.346; Fig. S3). The major causes of MA line extinction (Fig. 3b) were infertility (failure to lay eggs) and sterility (the production of eggs that did not hatch), indicative of underlying germline damage. Infertility and sterility were likely associated with observed reproductive abnormalities such as a deformed vulva, abnormal external growth close to the vulva (possible tumor), and cavities in the reproductive tract in place of embryos or oocytes (Fig. 3c). It is known that vulva-less C. elegans mutants are unable to lay eggs and can die from internal hatching (Ferguson and Horvitz 1985;Sternberg and Horvitz 1989).
It is possible that UV irradiation may also have caused nongenetic damage (to proteins or other macromolecules) that was not inherited but may also have contributed to sterility or infertility, and that daf-2 RNAi additionally may have protected against. Furthermore, daf-2 RNAi may have provided enhanced stress resistance to protect against stress responses induced by UV irradiation. Despite this, previous work in C. elegans has shown that UV irradiation most commonly induces base substitutions, which are subject to the nucleotide excision repair pathway (Meier et al. 2014).
The reduction in extinction observed in our MA lines with daf-2 RNAi shows the benefits of increased investment into germline maintenance that became more pronounced across multiple generations of UV-induced and spontaneous MA.

FITNESS OF SURVIVING MA LINES
We next tested for the effects of daf-2 RNAi on life-history traits of surviving MA lines. We were interested in how IIS influences potentially detrimental effects of MA that are not sufficiently severe to cause line extinction. We assayed age-specific reproduction, egg size, male production, and adult heat shock resistance in grand-offspring from the spontaneous and UV-induced MA lines on daf-2 RNAi versus control treatments, at generation 20 of MA, following two generations of rearing under common garden conditions (no irradiation, on empty vector control) to attenuate direct effects of irradiation and RNAi. Generation 20 was a point at which there was a pronounced benefit of daf-2 RNAi for protection against extinction in the UV-induced MA lines, but no clear difference in extinction trajectories for the spontaneous (nonirradiated) MA lines (Fig. 3a).
We suggest that the fitness benefits of daf-2 RNAi for irradiated MA lines were most likely due to genetic differences between the treatments, as the benefits persisted after two generations of common garden rearing and therefore were neither due to the direct exposure of offspring to RNAi nor acute effects of UV (neither as adults nor as eggs). This conclusion is reinforced by the finding that parental daf-2 RNAi effects on offspring fitness do not persist beyond F1 (see below, Fig. 6). Variation in fitness was considerably greater across irradiated than across nonirradiated MA lines (Fig. 4a), suggesting that irradiation-induced de novo mutations generated greater genetic and phenotypic variation compared to spontaneous MA. The spontaneous mutation rate in N2 wild-type C. elegans is estimated as one de novo mutation per individual per generation, under standard conditions (Denver et al. 2012;Meier et al. 2014). The close association between extinction trajectories during the first 20 generations of MA (Fig. 3a) and the fitness of the surviving MA lines (Fig. 4a) suggests that a threshold of accumulating deleterious mutations needs to be crossed before the effect size is sufficiently severe to result in extinction.
The most parsimonious explanation for the reduced extinction and increased fitness of daf-2 RNAi-irradiated lines would therefore be due to genetic differences in the rate of MA. To directly distinguish between genetic and nongenetic effects associated with MA line extinction, future work could use whole genome sequencing to quantify mutation rates between the MA treatments.
To determine if increased offspring fitness in the surviving daf-2 RNAi-treated MA lines was associated with greater parental resource allocation into their eggs, we measured egg size, as a proxy for parental investment (n = 75 eggs measured, one per individual taken from the Generation 20 fitness assay).  Previous work has shown that reduced IIS, either via dietary restriction or via daf-2 RNAi, increases mean egg size (Hibshman et al. 2016;Lind et al. 2019). Egg size has been correlated with offspring fitness (Lind et al. 2019) and larval body length  in C. elegans.
We found that grandparental daf-2 RNAi resulted in grandoffspring (F2) that laid smaller eggs if their grand-parents from the MA lines had been irradiated, but there was no significant effect on the size of eggs laid by F2 offspring descended from nonirradiated grandparents treated with daf-2 RNAi following 20 generations of MA (GLM, UV lines, RNAi: t = 2.362, df = 1, P = 0.0195; non-UV lines, RNAi: t = −1.246, df = 1, P = 0.215; all data, RNAi × UV: t = 2.642, df = 1, P = 0.00869; RNAi × UV × Block: t = −0.774, df = 1, P = 0.440; Fig. 5). Our results indicate that even though F2 offspring from irradiated daf-2 RNAi MA lines laid smaller eggs, this did not seem to impact negatively on the quality (fitness) of offspring they produced. It is possible that a smaller egg size could be a phenotypically plastic response to UV radiation, perhaps to improve stress resistance to irradiation. Alternatively, it could be the result of a trade-off between improved investment into germline genetic quality and egg size. At present, it is unclear why grandparental daf-2 RNAi resulted in smaller eggs in F2.
Exposure to certain environmental stresses (e.g., high temperature [Nigon and Dougherty 1949], starvation [Morran et al. 2009], increased mutation rate [Cutter 2005]) increases male production and outcrossing in C. elegans. However, we found no increase in the proportion of males produced by the grand-offspring in irradiated MA lines. Only one male developed from the 293 eggs assayed.
Reduced IIS increases the activity of heat shock factor 1 that mediates life span extension (Hsu et al. 2003;Seo et al. 2013). Heat shock responses are conserved across diverse taxa and positively associated with life span (as reviewed by Muñoz 2003;Åkerfelt et al. 2010). To test the stress resistance of postreproductive (Day 7) adults derived from the Generation 20 MA lines, we assayed survival and locomotion under acute heat shock for 1 h and 45 min at 37°C (following Zevian and Yanowitz 2014). We found no effect of MA line treatment (neither UV irradiation nor daf-2 RNAi) on the survival of untreated Day 7 F2 offspring following heat shock. Only 4% of the 371 individuals were dead by 24 h after heat shock. This high post-heat-shock survival of Day 7 adults is in line with previous work, which found survival after heat shock to more than double with age, from the first to the fourth day of adulthood in C. elegans (Zevian and Yanowitz 2014). However, F2 offspring from nonirradiated daf-2 RNAi MA lines recovered normal locomotion faster following heat shock than F2 from nonirradiated MA control lines, but this benefit was not seen in descendants from irradiated MA lines (Binomial GLM, 3 h post-heat shock, RNAi: z = −2.274, df = 1, P = 0.0229; UV: z = −2.078, df = 1, P = 0.0377; RNAi × UV: z = 2.304, df = 1, P = 0.0212; 24 h post-heat shock, UV: z = −0.206, df = 1, P = 0.837; RNAi: z = −0.045, df = 1, P = 0.964; RNAi × UV: z = 1.347, df = 1, P = 0.178; Fig. S5).

OFFSPRING FITNESS TRANSGENERATIONALLY
Transgenerational epigenetic inheritance of RNAi can occur in C. elegans via transmission of small interfering RNAs (Fire et al. 1998;Posner et al. 2019). This transgenerational inheritance of RNAi can last for several generations (Rechavi et al. 2014;Ni et al. 2016;Posner et al. 2019) and requires the germline argonaut protein, HRDE-1, that is absent in hrde-1 (heritable RNAi defective-1) mutants (Buckley et al. 2012). RNAi is still functional in somatic cells, as the hrde-1 mutation only affects the inheritance of RNAi via the germline. To determine whether there was direct transgenerational transfer of daf-2 RNAi via the germline and thus for how many generations the effects of parental daf-2 RNAi could persist, we assayed the fitness effects on three generations of untreated offspring (F1, F2, and F3) from daf-2 RNAi-treated versus control parents. Two genetic backgrounds were used: C. elegans N2 wild-types, and hrde-1 mutants that did not transfer RNAi transgenerationally.
The absence of fitness benefits in the second and third generations of offspring from daf-2 RNAi parents implies that daf-2 RNAi is not transgenerationally inherited beyond the first generation of offspring in the N2 background. This strongly suggests that the life-history differences between the daf-2 RNAi and control irradiated MA lines at generation 20 in the common garden experiment were due to genetic differences and not the direct effects of inherited RNAi. Interestingly, the heritable RNAi deficiency 1 (hrde-1) gene was required for the fitness benefits of parental daf-2 RNAi in F1 offspring, as these offspring fitness benefits were absent in the hrde-1 mutant background. The hrde-1 mutants had overall lower fitness and total reproduction than N2 wild type.
The major cause of extinction in irradiated hrde-1 mutant MA lines was developmental arrest, followed by infertility (Fig. S8). This strongly suggests that functional RNAi, mediated by the HRDE-1 germline argonaut protein, is required for daf-2 RNAi protection against extinction, and for normal development and reproduction under UV-induced MA.
We show the important role of hrde-1 in the daf-2 RNAimediated protection of irradiated offspring under mutation. Our results suggest that the interaction between UV-induced germline damage and a deficiency in transgenerational inheritance of RNAi reverses the protective benefits of daf-2 RNAi. This supports previous work suggesting that hrde-1 mutants suffer from progressive sterility under high temperatures, implying their increased sensitivity to environmental stresses driven in part by defects in gametogenesis (Buckley et al. 2012). We confirmed, using qRT-PCR analysis, that feeding nematodes bacteria that expresses double-stranded RNA for daf-2 significantly downregulated daf-2 expression in the MA lines at gen-eration 20 (Table S3). The extent of daf-2 downregulation at generation 20 varied between the MA treatments (Table S3; Fig. S9). For N2 wild types, daf-2 expression was downregulated by 23% in UV-induced MA lines and by 12% in spontaneous MA lines (based on mean gene expression fold change, 2 − CT ). For hrde-1 mutants under spontaneous MA, daf-2 expression was more subtly downregulated (by 6% fold change; Table S3; Fig. S9). We were unable to analyze data from UVinduced MA hrde-1 mutant lines as they did not recover well from freezing (only n = 1 MA line for the daf-2 RNAi treatment; Fig. S9).
The degree of downregulation of daf-2 expression between MA treatments interestingly matched the different extent of lifehistory divergence between daf-2 RNAi and empty vector treatments for the respective MA treatments at generation 20, in both N2 and hrde-1 mutant extinction trajectories (Figs. 3a, 7) and in N2 fitness (Fig. 4a). The relatively modest degree of daf-2 downregulation in the spontaneous MA lines for N2, and particularly for hrde-1 mutants, in comparison to the 50% knockdown of daf-2 expression from adulthood daf-2 RNAi found in previous work on Day 2 N2 adults ), could be due to an interaction between RNAi knockdown and genetic background, or a possible "acclimatization" to the multigenerational application of daf-2 RNAi in adulthood, but the exact reasons would require further investigation.
Technical replicates were highly repeatable for all samples (coefficient of variation, CV < 2.4% and CV < 1.5% for 98% of samples). Biological replicates showed more variation in relative gene expression within each MA line (mean CV = 7%; maximum CV = 8%), but within the expected range for C. elegans based on previous qRT-PCR analyses for different genes (Chauve et al. 2020;Lind et al. 2021).

Conclusion
We found that reduced insulin signaling in adulthood, via daf-2 RNAi, protects against extinction under both UV-induced and spontaneous MA in C. elegans. Furthermore, the fitness of the surviving UV-irradiated MA lines was higher under daf-2 RNAi. Most extinctions occurred because of infertility, egg hatching failure, and developmental abnormalities suggesting that germline MA directly contributed to the observed differences between the RNAi treatments. Future assessment of mutation rates in the MA lines could be useful to distinguish between genetic and nongenetic effects on extinction. Germline protection under daf-2 RNAi requires nuclear argonaut hrde-1 because fitness of hrde-1; daf-2 RNAi worms was reduced both in onegeneration and in multigeneration experiments. This is in line with previous work suggesting that hrde-1 is required for transgenerational inheritance and germline immortality (Buckley et al. 2012;Spracklin et al. 2017).
We set out to test whether adulthood-only daf-2 RNAi, known to extend life span without an obvious cost to parental fecundity Lind et al. 2019), trades off with detrimental fitness effects when applied across multiple generations. However, we found that multigenerational downregulation of daf-2 via adulthood-only RNAi has positive effects on the germline and protects MA lines from extinction. This was particularly so when germline mutation rate was increased by low-level UV radiation. Our results therefore suggest that although wildtype levels of daf-2 expression promote development, growth, and early-life reproduction, they are costly for late-life fitness and this cost accumulates across generations leading to lineage extinction.
Antagonistic pleiotropy theory of ageing (AP) maintains that genes that increase fitness in early life at the expense of fitness late in life can be overall beneficial for fitness and go to fixation (Williams 1957). In line with AP, downregulation of daf-2 expression during development reduces fitness Chen et al. 2007), but what is the physiological basis of the putative trade-off? Our findings that adulthood-only downregulation of daf-2 expression protects both the germline and the soma under benign conditions and under UV-induced stress argue against the idea that resource allocation underlies this tradeoff. However, our results are in line with the hypothesis that selection optimizes physiological processes for development and early-life reproduction but fails to optimize the regulation of gene expression associated with these same physiological processes later in life (de Magalhaes and Church 2005; de Magalhaes 2012; Blagosklonny 2013; Gems and Partridge 2013;Maklakov and Chapman 2019). This is likely because selection on insulin signaling in adulthood is too weak in C. elegans . Future studies should focus on investigating the contribution of weakening selection on gene expression with age to the evolution of ageing across taxa, on testing the effects of downregulated IIS on survival and fitness under wide range of ecologically relevant stresses, and on developing a comprehensive theoretical framework that links genetic and physiological theories of ageing.