Post-senescence reproductive rebound in Daphnia associated with reversal of age-related transcriptional changes

Abstract

A long-lived species of zooplankton microcrustaceans, Daphnia magna, sometimes exhibits late-life rebound of reproduction, briefly reversing reproductive senescence. Such events are often interpreted as terminal investments in anticipation of imminent mortality. We demonstrate that such post-senescence reproductive events (PSREs) neither cause nor anticipate increased mortality. We analyze an RNAseq experiment comparing young, old reproductively senescent, and old PSRE Daphnia females. We first show that overall age-related transcriptional changes are dominated by the increased transcription of guanidine monophosphate synthases and guanylate cyclases, as well as two groups of presumed transposon-encoded proteins, and by a drop in transcription of protein synthesis-related genes. We then focus on gene families and functional groups in which full or partial reversal of age-related transcriptional changes occur. This analysis reveals a reversal, in the PSRE individuals, of age-related up-regulation of apolipoproteins D, lysosomal lipases, and peptidases as well as several proteins related to mitochondrial and muscle functions. While it is not certain which of these changes enable reproductive rejuvenation, and which are by-products of processes that lead to it, we present some evidence that post-senescence reproductive events are associated with the reversal of age-related protein and lipid aggregates removal and apoptosis.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11357-024-01401-y.

Introduction

Scheduling the reproductive effort along the lifespan is one of the key life-history decisions organisms make to maximize their lifetime fitness. While generally, selection supports higher investment in early reproduction, as earlier produced offspring have a higher reproductive value [9, 22], in certain situations higher late-life reproduction output is observed. Typically, late-life outbursts of reproduction are described as a terminal investment—a switch of resource allocation to reproduction in the face of anticipated high mortality [13, 18, 24, 30]. Essentially, this strategy amounts to a late-life switch from iteroparous to semelparous life history. The expectation of increased mortality may be either due to general properties of actuarial aging (e.g., Gompertz law [34, 57]), or due to environmental changes, for example seasonal, or maybe a self-fulfilling prophecy, as redirecting resources for a bout of reproduction may take resources away from maintenance and repair [18]. An alternative explanation of late-life reproduction bouts is the exit from reproductive diapause or otherwise temporary reproductive quiescence caused by factors like, again, seasonality, current feeding conditions, or time necessary to accumulate resources for successful reproduction [50]. Such periods of reproductive quiescence are distinct from senescence and, in fact, may exist in order to postpone senescence [36, 70]. Perhaps the most striking example of such rejuvenation is the exit of Caenorhabditis elegans larvae from the long-lived, non-reproducing dauer stage [28].

Either way, it is of interest for aging research to find out what prior life-history events correlate with late-life reproductive “rejuvenation” and what gene expression changes occur during such a switch, given that one of the aging research and gerontology central goals is to investigate biological foundations of healthy and productive late-age life. Current knowledge about mechanisms of reproductive rejuvenation is largely built around three interrelated hypotheses: that reproductive rejuvenation occurs through the reversal of mitochondrial dysfunction in the germline [11, 37], or through attenuation of germ-line apoptosis [80], or through removal of oxidatively damaged lipid or protein aggregates [71] in both somatic and germline cells. These likely mechanisms of rejuvenation provide an a priori list of functions expected to be regulated to allow rejuvenation: mitochondria function and integrity, aggregate degradation mechanisms, including lysosome acidification [7], and apoptosis regulation such as the p53 pathway.

In this study we focus on the regularly occurring cases of late-life restoration of reproductive function in Daphnia—a freshwater planktonic crustacean that is a classic model organism in aging and longevity research [29], which is enjoying a revival due to contemporary genomic tools [52], generating a significant new data on aging [1, 3, 10, 12, 15, 27, 55, 65]. Daphnia are particularly suitable for cohort longevity and reproductive senescence studies because their reproduction mode, the cyclic parthenogenesis, allows the creation of cohorts of genetically identical and yet fully outbred individuals, thus providing an advantage over the use of inbred lined customary in Drosophila or C. elegans longevity studies. Additionally, Daphnia transparent bodies allow straightforward measurements of fecundity and lipid content, as well as of the in vivo use of various tissue-specific fluorescence dyes.

Our previous studies on aging hallmarks in Daphnia indicate that mitochondrial dysfunction is an unlikely cause of aging (and therefore its reversal an unlikely cause of reproductive rejuvenation), as little if any reduction of respiratory activity of mitochondrial membrane potential occurs at old age [2]. On the other hand, the accumulation of oxidized lipids and misfolded proteins in the forms of lipofuscins and amyloids has been shown to increase with age [41]. Little is known about the role of apoptosis in Daphnia aging.

One pattern of Daphnia's life history that became apparent from a number of longevity studies in this organism is the lack of genetic trade-off between survival and fecundity [14, 17]. Rather, differences in longevity and fecundity among genotypes are readily explained by differences in genetic load [14, 42], while the differences observed within a single genotype reared under different resource availability often emerge through higher, not lower investment into reproduction, relative to growth under limited resources [44, 55]. Biochemical interventions known to extend lifespans, such as exposure to beta-hydroxybutyrate or nicotinamide mononucleotide, also tend to positively affect fecundity rather than longevity (Pearson et al., in preparation). Thus, it appears that Daphnia has been under selection to invest in reproduction whenever such an opportunity presents itself, at the expense of aging-related investments and life-history consequences. Indeed, investments into reproduction reach nearly 100% of total energy expenditures with Daphnia body weight [44, 45] and with the increase of food concentration [46]. In the largest, i.e., oldest Daphnia reproductive effort appears to be limited by the cost of carapace replacement during each molt [45]. With respect to diet composition, Daphnia reproduction is limited by dietary polyunsaturated fatty acids and sterols [48]. These constraints and limited resources are likely to play a role in the reproductive senescence, as they may become increasingly more limiting as the organism ages. Conversely, the accumulation of dietary polyunsaturated fatty acids and sterols during reproductive quiescence may contribute to subsequent renewed reproductive efforts.

To focus on possible mechanisms of old-age reproductive rejuvenation in Daphnia, we first demonstrate that the previously sporadically observed cases of old-age restoration of reproductive function (which we term post-senescence reproductive events, PSRE) represent a ubiquitous phenomenon in the long-lived species D. magna. We also show that while the reproductive rebound events do occur on the background of exponentially increasing late-life mortality, they neither cause nor anticipate elevated mortality or shorter lifespan in the individual undergoing such reversal of reproductive senescence. We then report the results of an RNAseq differential expression (DE) experiment comparing transcriptomes of young reproducing, senescent post-reproductive, and PSRE females. We discuss the functionality of genes with significant differential expression, both in the light of the a priori predicted reproductive senescence reversal pathways discussed above, and focusing on enriched gene ontologies not a priori predicted to play a role in rejuvenation.

Materials and methods

Four geographically distant Daphnia clones were obtained from Basel University Daphnia stock collection and cultivated in the laboratory for at least 5 years. Clones’ identity and geographic origin are listed in Supplementary Table 1. These clones were chosen to represent (two of each) the long-lived clones from permanent water bodies and short-lived clones from intermittent water bodies [17]. Briefly, clones were maintained at the density of 1 adult per 20 mL of water, fed with Tetradesmus obliquus (formerly Scenedesmus acutus) culture added daily at 100,000 cells/mL. Further details on clones’ provenance and maintenance in  [1].

Experimental animals were collected as neonates (less than 24 h old, all females) from 15- to 25-day-old mothers and placed individually in 35-mL shell vials containing 20 mL of water. The vials were maintained daily, and age and body size at maturity were recorded. Subsequently, neonates were counted and removed, and water was changed at the end of each ovary cycle (4 days at 20 °C, 2–3 days at 24 °C). Each vial was maintained until the death of the individual. Two separate experiments were conducted, referred to as Experiments 1 and 2 thereafter. Experiment 1 consisted of 94 individuals kept in a COMBO medium [33] at 20 °C. Experiment 2 included three treatments: the COMBO medium at 20 °C as in Experiment 1, the ADaM medium [35] at 20 °C, and the ADaM medium at 24 °C, with 89, 91, and 88 individuals in each of these three cohorts. The two media are similar, both imitating the ionic composition of natural pond water, but ADaM contains significantly higher concentrations of Ca²⁺ and Mg²⁺ ions, modeling water with higher hardness.

Reproductive “rejuvenation” (post-senescence reproductive events, PSRE) was defined as the production of a clutch of at least six live neonates such that the number of neonates produced was at least three standard deviations above the mean number of neonates produced in the previous five ovary cycles (including any cycles in which no neonates were produced) by a female over 60 days of age. Subsequent large clutches meeting the same criterion were considered a continuation of the same PSRE. In a few cases, two or even three PSRE events occurred in the same individual, separated by several low- or no reproduction cycles; these were counted as separate events. Differences in among clones and treatments as well as in post-PSRE survival between individuals showing a PSRE event and matching non-rejuvenated individuals were analyzed by means of Kaplan–Meier and proportional hazards methods implemented in the JMP statistical platform (ver. 17; SAS [64]).

Starting with age of 30 days random samples of neonates born to mothers in the ADaM medium, 24 °C treatment were taken for neonate size and lipid content measurement. Neonates were sampled < 24 h after birth and stained with Nile Red (1 mg/mL) for 1 h. After staining neonates were photographed individually using a Leica DM3000 microscope with a × 10 objective (0.22 aperture) equipped with a Leica DFc450C color camera, with a 488-nm excitation/broadband (> 515 nm) emission filter (for Nile Red fluorescence) and in bright field (for body length measurement). Images were analyzed using ImageJ [60] with the portion of intensity emitted from areas with unweighted intensity over 150 recorded as the measure of lipid content (see example image in the “Results” section). An additional cohort was started on day 30 of the experiment to allow a common garden comparison of neonates born to younger and older individuals.

Four 130- to175-days-old GB-EL75-69 females from the ADaM medium, 20 °C treatment that has just produced a clutch meeting the PSRE criterion described above were removed from the experiment and flash-frozen in liquid nitrogen for the RNAseq experiment. Four randomly chosen reproductively senescent females from the same clone and treatment were selected as the senescent controls. These females were carrying no clutch or, in one case, carrying a clutch of just two eggs. These two samples will be hereon referred to as Old, PSRE, and Old, non-PSRE. For the Young sample, four randomly chosen individuals from the same clone from a separate cohort of the age of 20 days were treated in the same manner. In all cases, the individuals were sampled within 24 h of producing a clutch or from the last molting if no clutch was produced. The eggs were removed from the brood chamber prior to freezing and frozen separately.

Whole organism RNAs were extracted from these individuals using the Qiagen RNeasy Plus Mini kit. RNA library preparation was performed using the NEBNext Ultra II Directional RNA Library Prep Kit (NEB, Lynn, MA) following the manufacturer’s protocols. The libraries were sequenced with Illumina Novoseq 6000, S4 flow cell, PE100. The quality check of Indexed sequences was performed by Fastqc, and indexed sequences were trimmed using an adaptor sequence by TrimGalore-0.4.5.

Reads were mapped to D. magna Xinb3 reference transcriptome (BioProject ID, PRJNA624896; D. Ebert and P. Fields, personal communication; Fields et al., in preparation) using STAR [16] and genes with differential expression (DE) either between young and old Daphnia or between PSRE- and non-PSRE old Daphnia were identified using DEseq2 [43] using sense reads only. To analyze Gene Ontology (GO) enrichment among genes with a significant DE by the list of such genes, each comparison, separately, was matched to the list of GOs and pathways identified in reference D. magna transcriptome obtained by PANTHER (Ver. 16; [51, 73]). Counts of genes representing each GO or pathway in a list consisting of unique combinations of genes and GOs were then tested for being heterogeneous (enriched or depleted) relative to the numbers of other genes in a given gene list and the numbers of other GOs in the reference using Fisher’s Exact Test (FET). The four counts used in the FET for each GO category were the number of genes in this GO category with DE,the number of genes in this category without DE, the number of genes in all other GO categories in the reference set showing DE, and the number of genes in all other GO categories in the reference set without DE. False Discovery Rate adjustment for multiple testing was applied and only the innermost nest GO category showing a significant enrichment or depletion is reported. Differential expression of a select subset of genes was confirmed by qPCR using the X-box binding protein Xbp-1 gene as a reference [67].

Results

Occurrence of post-senescence reproductive events

This section reports the results of two lifetable experiments in which Daphnia were maintained individually in 20 mL of medium, in COMBO medium at 20 °C in Experiment 1 and either in COMBO medium at 20 °C, in ADaM medium at 20 °C, or in ADaM medium at 28 °C. Overall lifespan and age-specific survival differed significantly between identical treatment cohorts of the two experiments, between two temperatures within experiment 2, and among clones in experiment 2 (Supplementary Fig. S1, Supplementary Table S2). There were, however, no significant differences between the two media tested in Experiment 2. Despite these overall differences, PSREs occurred in nearly all combinations of clones and treatments with frequencies of occurrence ranging from 10 to 60% of survivors to the minimal age of PSRE (Table 1).

Table 1.

Frequency of post-senescence reproduction events (PSRE) in Daphnia from four clones in life-history experiments conducted in various conditions

Experiment	Medium	T °C	Clone ID	Number in cohort	Number surviving to the earliest PSRE event	Number that had PSRE	Portion with PSRE among survivors (95% CI)
1	COMBO	20	FI-FSP1-16–2	29	23	5	0.217 (0.075, 0.437)
1	COMBO	20	GB-EL75-69	20	19	9	0.474 (0.244, 0.711)
1	COMBO	20	HU-K-6	32	30	12	0.400 (0.227, 0.594)
1	COMBO	20	IL-MI-8	13	13	4	0.308 (0.091, 0.614)
2	COMBO	20	FI-FSP1-16–2	18	3	0	0 (0, 0.708)
2	COMBO	20	GB-EL75-69	27	21	10	0.476 (0.257, 0.702)
2	COMBO	20	HU-K-6	24	18	4	0.222 (0.064, 0.476)
2	COMBO	20	IL-MI-8	20	10	1	0.1 (0.003, 0.445)
2	ADaM	20	FI-FSP1-16–2	20	15	0	0 (0, 0.218)
2	ADaM	20	GB-EL75-69	27	25	12	0.480 (0.278, 0.687)
2	ADaM	20	HU-K-6	21	18	11	0.611 (0.357, 0.827)
2	ADaM	20	IL-MI-8	23	14	3	0.214 (0.047, 0.508)
2	ADaM	24	FI-FSP1-16–2	16	4	0	0 (0, 0.602)
2	ADaM	24	GB-EL75-69	26	22	10	0.455 (0.244, 0.678)
2	ADaM	24	HU-K-6	21	21	8	0.381 (0.181, 0.616)
2	ADaM	24	IL-MI-8	25	19	3	0.158 (0.034, 0.396)

Two-way nominal logistic fit p-value for the difference among conditions (combination of medium and temperature), p < 0.01; for the difference among clones, p < 0.001; and for the interaction, p > 0.45. CI, Clopper–Pearson CI

Trade-offs with late-life mortality and investments per offspring

While the PSRE events occurred during periods of high mortality and preceded further increase in mortality overall (Fig. 1), we did not detect any increase in mortality following a PSRE event specifically in individuals experiencing such events. In fact, the life expectancy of such individuals was higher than that of non-PSRE individuals of comparable age (Fig. 2, Table 2). Thus, late-age reproduction neither anticipates nor causes elevated mortality. However, trade-offs were observed between the late-life number of offspring (clutch size) and body size and lipid content of offspring (Fig. 3). Offspring body length in PSRE individuals was consistently lower than in their non-PSRE counterparts, both significantly decreasing with age (Fig. 3A), although this difference was small in magnitude (approximately 20 mm or 2% of body length). A different pattern of trade-off was observed for the lipid content of the neonates (Fig. 3B). The PSRE females were producing neonates with a higher lipid content than their non-PSRE counterparts at ages 75–90 days, but rapidly reduced lipid investment per offspring at ages over 100 days, indicating higher cost of producing PSRE clutches at older age, as indicated by a significantly higher slope of neonate lipid content over mother’s age in PSRE than in non-PSRE mothers (Fig. 3B).

Fig. 1.

Fecundity (A, C, D, and E) and mortality ( $\mu =-\frac{1}{t}\text{ln}(\frac{N_t}{N_0})$; B, G, H, and I) in life-table experiments 1 (A and B) and 2 (C–I). Colors correspond to experimental treatments: COMBO medium, 20 °C (green); ADaM medium, 20 °C (blue); and ADaM medium, 24 °C (red). Inserts show distribution of PSRE events over age. Symbols correspond to four different clones used. See Supplementary Fig. S1 for clone-specific survival data

Fig. 2.

Survival of Daphnia past a late-life reproduction bout event (blue) vs. survival of individuals not showing a late-life reproduction bouts, past average age of late-life reproduction bouts in the same clones in the same experiment (red). P-values from Kaplan–Meyer survival analysis: p_L-R, log-rank test; p_W, Wilcoxon test. See Table 2 for more detailed analysis

Table 2.

Proportional Hazard analysis of late-life survival of Daphnia with and without post-senescence reproductive events (PSRE)

Source	DF	Wald chi-square	Prob > chi-square
Treatment	3	5762.6	< 0.0001
Clone	3	6083.8	< 0.0001
Treatment * clone	6	8554.1	< 0.0001
PSRE	1	5.17	0.023
Treatment * PSRE	3	12.16	0.007

Overall survival is shown on Fig. 2

Fig. 3.

Body length (A) and lipid content visualized by Nile Red staining (B) in neonates born to Daphnia females of different age who either have or have not experienced a PSRE while producing these neonates (blue and red, respectively). P-values represent heterogeneity of slopes test (A, differences in intercept and slope; B, differences in slope). Neonate parameters were determined only in AdaM 20 °C treatment. Example images of neonates born to parents of 25, 90, and 160 days of age, top to bottom

Correlations with environment and early-life events

In order to determine if early-age life history events predetermine late-life propensity to exhibit PSRE events, we correlated early body size at maturity and fecundity (mean clutch size) in age classes < 10, 10–30, 30–50, and 50–70 days with future PSRE event in Experiment 2 (Fig. 4). In two out of the three clones that showed PSRE events in Experiment 2 individuals were smaller at maturity than those that survived to the minimal PSRE age but never experiences a PSRE (Fig. 4A); overall effect was not significant, but Clone-by-body size interaction was (Supplementary Table S3). Clutch sizes in ages 10–70 were not a predictor of future PSRE (Fig. 4B, Supplementary Table S4), except for a border-line significant effect of fecundity in the age class 10–30 days (non-significant after multiple test correction) despite apparently slightly higher fecundity in future PSRE females than their same-age counterparts (Fig. 4B, Supplementary Fig. S2).

Fig. 4.

Body length at maturity (A) and early-to-mid life fecundity (clutch size) (B) in Daphnia females who either would (blue) or would not (red) experience a PSRE later in life. Only females who survived to the earliest age of PRSE are included into this analysis. P-values represent the PSRE status effect in a three-way nominal logistic regression with treatments, clones, and either body size or early fecundity on future PSRE (Supplementary Tables S3 and S4). Non-PSRE clone FI included for comparison. See Supplementary Fig. S2 for combined data across both experiments

RNAseq results

Transcriptional profiles differed between young, old non-PSRE, and old PSRE individuals. Principal component analysis revealed a good separation between the three groups along the 1st PC that explained 50.7% of the variance, with the old PSRE group positioned between the young and old non-reproducing individual samples (Fig. 5). PC 2 (14.7%) separated, with one sample exception, the reproducing individuals from the non-reproducing ones, indicating either overcompensation in PSRE individuals or otherwise PSRE-specific transcription. Much of the differential expression manifested through overexpression of numerous transcripts in old, relative to young individuals and in non-PSRE vs PSRE individuals (Supplementary Fig. S3).

Fig. 5.

Principal components analysis of young (green), old non-PSRE (orange), and old, PSRE (purple) Daphnia based on 3331 transcripts with a significant DE (p_adj < 0.1) between age classes

GO enrichment analysis results are shown in Supplementary Table S5. Many of the GO or pathway terms listed were enriched among differentially expressed due to the enrichment of a particular gene family, rather than unrelated genes with different functionality within the same GO term. Expression profiles of several such gene families are shown in Figs. 6 and 7. For the genes up-regulated in old Daphnia (Fig. 6), these gene families included guanidine monophosphate synthases (GMPSs), guanylate cyclases (GCs), a family of DNA helicases, as well as two putative transposon-related gene families (see below). In contrast, genes showing down-regulation in old Daphnia relative to the young ones were enriched in GO terms related to RNA processing, ribosome structure, protein synthesis and transport, and muscle development (Supplementary Table S5).

Fig. 6.

Expression level (RPKM, log scale) of select gene families or GOs of interest in young (green), old PSRE (purple), and old non-PSRE (orange) Daphnia showing DE with at least p_adj < 0.01 and at least |log₂(FC)|> 1 between young and old Daphnia, but not between PSRE and non-PSRE categories. Genes sorted by mean RPKM in the Young sample. Data not shown are for RPKM < 0.01, including cases of no reads detected (many transposon-related genes in the Young sample). Bars represent SE among the four samples and do not correspond to p_adj values estimated by DESeq2. GO was significantly enriched among genes with DE (***p < 0.0001; Supplementary Table S5). #RPKM = 0 (no reads observed)

Fig. 7.

Expression level (RPKM, log scale) of select gene families or GOs of interest in young (green), old PSRE (purple), and old non-PSRE (orange) Daphnia showing DE with at least p_adj < 0.01 and at least |log₂(FC)|> 1 between young and old non-PSRE Daphnia, with PSRE individuals showing full or partial reversal of transcription to youthful levels. Bars represent SE among the four samples and do not correspond to p_adj values estimated by DESeq2. Note selected gene families with opposite directions of age-related DE in different paralogs

GOs and pathways enriched among genes showing DE between PSRE and non-PSRE old Daphnia are shown in Fig. 7. These include apolipoproteins D (ApoD), putative peroxidases homologous to chorion peroxidase of mammals and curlySu peroxidase in Drosophila, serine and cysteine proteases, and a family of lysosomic lipases. Notably, in PSRE old Daphnia, all of the ApoDs and several lipases and endopeptidases show full or partial reversal back to a low level of expression characteristic of young Daphnia. The same pattern is observed in several gene families and functional groups not detected as enriched by the GO enrichment analysis (see below). DE of a subset of ApoDs, GMPS, and GS has been confirmed by qPCR (Supplementary Fig. S4, Table S7).

Differentially expressed members of each gene family included both tandemly or segmentally duplicated and unlinked paralogs. However, all old-age-specific ApoD’s are tightly linked, mapping DM3 assembly scaffold 0014F, scaffold positions 14.86–14.92. The greatly expanded GMPS gene family includes a series of paralogs mapping to scaffold 0000F, with scaffold positions between 30.73 and 73.16 and a pair of linked paralogs located in the scaffold 0013F, positions 23.8 and 24.11. Old-age-specific lysosomal lipases include a pair of closely linked paralogs in scaffold 001F positions 55.95 and 55.106 and several gene family members located elsewhere.

It is important to notice that the lists of differentially expressed genes are rich in genes with no GO or pathway terms that could be assigned to them, with many of such genes being Daphnia-specific. Unexpectedly, the lists of up- and down-regulated genes (DESeq2 adjusted p < 0.01) differed with respect to the abundance of such genes. For the young vs. old, nonreproducing Daphnia comparisons the genes up-regulated in the old Daphnia were significantly enriched with uncharacterized genes (expected, 960.8; observed, 1584; two-tailed uncorrected Fisher exact test p-value < 0.0001), while in the list of genes down-regulated in the old Daphnia, such genes were depleted (expected, 1163.2; observed, 459; p < 0.0001). In the old, non-reproducing Daphnia vs. old, reproducing (PSRE) ones, the list of up-regulated genes was also enriched in the uncharacterized genes (expected, 190.6; observed, 408; p < 0.0001), while the much smaller list of the down-regulated genes was neither depleted nor significantly enriched (expected, 5.8; observed, 8; p > 0.17).

All genes and gene families with significant DE between young and old non-reproducing Daphnia are listed in Supplementary Tables S6–S8. They appear to represent a variety of metabolic and regulatory pathways with little to no correlation to a priori expectations. One exception is a noticeable up-regulation of putative transposon-related genes in old, non-reproducing Daphnia. Both retro-transposons and DNA transposons are represented (35 copies of retrotransposon-related homologs of pol protein and 9 transposases, respectively, Supplementary Table S6, those with p_adj < 0.0001 also shown in Fig. 7). Another group of genes with predictable age effect are those encoding muscle- and motility-related proteins, several of which are down-regulated in old non-reproducing flies, including troponins, kinesins, actin-related proteins, and myosin light chain, as well as mitochondrial creatine kinase, (Supplementary Table S7). This is consistent with the enrichment of muscle development GOs among genes down-regulated in old non-reproducing Daphnia (Table S5). Likewise, numerous translation-related genes show down-regulation with age, including several cytoplasmic and mitochondrial aminoacyl-tRNA synthetases (Table S7).

Finally, several groups of paralogs showed an apparent subfunctionalization by age, with some paralogs being up- and others down-regulated in old, non-reproducing flies (Supplementary Table S8). Notably, several of these paralogs, including mitochondrial cytochromes P450, chorion peroxidases, and zinc finger proteins also show DE between reproducing and non-reproducing old Daphnia, indicating that age-specialization of paralogs may be not only related to absolute age but also to retention of old-age reproductive ability.

Genes and gene families with significant DE between PSRE and old non-reproducing Daphnia are listed in Supplementary Tables S9–S11. These include, in addition to gene families shown in Fig. 7, a plasma membrane “organizer” tetraspanin, a heat shock protein, two of several Daphnia neurotrophin paralogs, melanization proteases, and an NADPH oxidase, among others (all up-regulated in old non-reproducing Daphnia; Table S9). Fewer genes showed down-regulation in non-reproducing individuals, none being consistent across gene families, i.e., only individual genes showed this pattern (Table S10). These included, predictably, a vitellogenin precursor, one of several in the Daphnia genome. Finally, serine protease and chorion peroxidase gene families appeared to contain paralogs with opposite directions of DE between PRSE and non-PRSE old Daphnia (Table S11).

Discussion

We have observed nearly ubiquitous cases of renewal of asexual reproduction in a cyclic parthenogen Daphnia after complete or nearly complete reproductive senescence. Most of these cases occur at the age of 100–150 days, which is 70–80% of the maximal lifespan observed in our experiments and the age to which about 20% of a typical cohort survives. We did not observe any evidence that such renewal of reproduction represents the terminal investment strategy, as individuals with a post-senescence reproduction event show a higher, not lower remaining lifespan. Such events neither anticipate nor cause imminent mortality. We did observe a trade-off with offspring size, however, as the offspring of the oldest mothers producing unusually large clutches are smaller than their small-clutch counterparts. This can be interpreted as a shortage of resources necessary for oocyte provisioning that older Daphnia are experiencing even when their ovaries reactivate. These resources are, however, different from the main nutrient Daphnia mothers provision their offspring with—lipids, as there is no difference in lipid content between late-life small and large clutches.

Can the observed late-life reproductive rebound reported here be interpreted as an overcompensating exit from a reproductive diapause-like state? Not if one defines diapause and diapause-like states as an adaptive response to detrimental environments [19, 72], as we observe it at constant food, light, and temperature conditions. Such adaptation does not exist in adult Daphnia and other cladocerans, as they diapause as resting eggs [66]. However, if one assumes that there is an endogenic, environment-independent period of reproductive quiescence unrelated to senescence, then our observations may in fact be interpreted as an adaptation to a seasonal environment and not as “rejuvenation.” Perhaps, Daphnia possesses an internal clock that signals reproductive quiescence at a moderate age (as Daphnia of that age are likely to experience winter conditions) and signals the anticipation of improved environmental conditions at a more advanced age (indicative of having survived the winter). It should be noted though that post-senescent reproductive events have been observed not only in the D. magna clone GB-EL75-69 extracted from a permanent pond in London, UK, where the clone’s recent evolutionary history very likely included overwintering adults, but also in clones IL-MI-8 and FI-FSP1-16–2 originating from temporary pools in which surviving unfavorable seasons occurs only in the form of resting eggs.

We have observed a profound change in transcriptome-wide and GO- and gene family-focused transcriptional changes between young and old Daphnia, as well as between old non-reproducing and old, reproducing ones. In many cases the old, reproducing (PSRE) individuals showed a partial or even complete reversal of transcriptional profile back to youthful values. It may be noted, that in the multidimensional analysis, the biological replicates representing this condition were more similar to each other than those representing young and old, non-reproducing individuals, indicating a coordinated transcriptional state.

On the level of individual genes, gene families, and GOs, the transcripts showing aging-related DE either fit into a priori predictions (such as mitochondrial proteins, proteins involved in apoptosis, or lipid and protein aggregate removal) or showed significant enrichment in the GO enrichment analysis. Functionally, the GOs and gene families implicated here as transcriptional markers of aging and of reproductive reversal represented of mix of functionalities that could and could not be easily expected a priori, based on our knowledge of functional genomics of aging and reproduction in better-understood models such as Drosophila and Caenorhabdites. A graphic summary of the observed changes and their hypothesized consequences for physiology and life history is presented in Fig. 8.

Fig. 8.

Summary of hypotheses about age-related transcriptional changes, the reversal during post-senescence reproductive events, and consequences of these changes for cellular physiology and life history

It should be noted that some of the observed transcriptional differences may be natural random variants of transcription level present constitutively, creating a possibility for reproduction renewal. Others may be a result of active up- or down-regulation, serving as proximal causes of reproduction renewal. Others may be not necessary for it but happen to be downstream by-products of pathways leading to such events. It is not known presently which of the observed differences allow, enable, or are the consequences of the reproduction renewal.

Four groups of paralogs stood out as those showing a strong increase in expression in old Daphnia without any indication of reversal in the “reproductively rejuvenated” PSRE ones: guanidine monophosphate synthases, guanylate cyclases, and two transposon-related protein families. A more diverse list of functionalities showed either the opposite trend in different paralogs or consistent reversal or transcription activity back to “youthful” levels in the PRSE Daphnia. We will consider these functional groups separately.

Differential expression between young vs. old Daphnia

GMP synthases

Overexpression of GMP synthases (GMPSs) in old Daphnia, regardless of reproductive status, is surprising, as is the massive duplication of SMPS genes in the Daphnia genome. It is not immediately obvious why the main function of GMPS, de novo synthesis of guanine nucleotide, should be up-regulated in an old Daphnia in which little growth and a few cell divisions occur. However, GMPSs are multifunctional proteins with additional functions far beyond the traditional purine biosynthesis function. GMPS has been shown to play a role in the deubiquitylation of regulatory proteins, most importantly of one of histones and p53, by interacting with the epigenetic silencer USP7 [74]. By facilitating p53 deubiquitylation the GMPS/USP7 complex results in stabilization of p53 content in cells, both in Drosophila and human cells [61, 74]. The p53 stabilization function is thought to be related to the nucleotide deficiency caused by or signaling the presence of replication or DNA-repair-related stress [4, 61]. Thus, elevated expression of GMPSs in old Daphnia may have nothing to do with purine biosynthesis, but rather with regulation of downstream targets of p53, including cell proliferation and apoptosis. Perhaps curbing the age-related accumulation of senescent or otherwise dysfunctional cells is the central target of the up-regulation of GMPSs. By deubiquitylation of histone H2B in Drosophila, GMPS is contributing to epigenetic silencing of homeotic genes by Polycomb. Furthermore, and, perhaps consequential for all arthropods, the GMPS/USP7 complex has been shown to interact with the ecdysone (molting hormone) receptor (EcR), acting as a transcriptional corepressor of ecdysone target genes [74]. This is consistent with observations of molting in old-age Daphnia, in which molting becomes less frequent and often causes prolonged difficulties and even death. Finally, this remarkably versatile protein has also been shown to regulate the accumulation of cytoplasmic triglyceride droplets in zebrafish. This function is related to de novo purine synthesis, which appears to activate ROS production, in turn upregulating the triglyceride hydrolase gene, ultimately triggering the utilization of stored triglycerides [56]. This is consistent with reduced lipid content in very old Daphnia (Yampolsky et al. in preparation) and with other lipid metabolism-related transcriptional changes observed in this study (see below). At present, it is not known which of these diverse functions is the reason for the drastic upregulation of GMPSs in aging Daphnia, whether this upregulation is the cause or a consequence of age-related changes, or why does Daphnia genome contains so many GMPS paralogs.

Guanylate cyclases

It is perhaps not surprising, given the upregulation of GMPSs, that this family of enzymes utilizing GMPS products also showed age-dependent transcription. Four paralogs of guanylate cyclase (GC) showed up-regulation in old Daphnia, one of which also appeared to partly reverse this up-regulation in the PSRE females. In the case of GCs, however, the explanation for late age upregulation is readily available: the product of GC, cyclic GMP (cGMP) is a second messenger with a variety of functions most likely operating through activation of protein kinase G (PKG). Many of the downstream pathways have long been known to regulate apoptosis. Interestingly, accumulation of cGMP can either induce [20, 68] or suppress [49, 77] apoptosis, depending on the cellular environment. Relevant to the GMPS functions discussed above, cGMP induces apoptosis under oxidative stress by activating p53 [6]. Partial reversal of one of GC paralog’s expression in the PSRE females is intriguing, but at present, we can only hypothesize that it may have to do with suppression of apoptosis or activation of cell proliferation in reproductive tissues. It did not escape our attention that there was a pair of paralogs showing opposite directions of age-related transcription rates are serine-threonine protein kinases, although there is no evidence that these two kinases are in fact cGMP-dependent (see below).

These results taken together suggest that there is a profound upregulation of cGMP- and p53-related apoptosis-enhancing pathways in aging Daphnia likely aiming at the elimination of senescent, damaged, or transcriptionally aberrant cells.

Transposon-related transcripts

Another group of genes we found to consistently show higher transcription levels in aging Daphnia were those showing sequence similarity to transposons-encoded genes, namely to the gag-pol polyprotein present in retroviruses and to DNA transposons’ transposase. It should be noted that there is less confidence in the annotation of these proteins, as, for example, the existing genome annotation could not be confirmed by PANTHER analysis. Our assumption that these proteins are indeed transposon-encoded relies on 29% sequence identity with a retrovirus-related Pol polyprotein from transposon TNT 1–94 present in nematode Trichinella spiralis and 43% sequence identity with the trigger transposable element-derived protein from the bumblebee (Bombyx mori), respectively. Yet, assuming that these transcripts are indeed transposon-encoded, their upregulation in the old Daphnia is interesting. There has been recently an explosion of studies demonstrating that reactivation of transposons, retrotransposons in particular, as a consequence of the erosion of epigenetic regulation that occurs with age, significantly contributes to downstream damage [8, 26, 40, 79]. If confirmed, the increased transposon activity in aging Daphnia may make this organism an attractive model for the research of transposon’s role in aging.

Differential expression between PSRE and non-PSRE old Daphnia

In nearly all cases, whenever DE was observed between these categories, there was also DE between young and old, reproductively senescent Daphnia, with the PSRE ones partly or fully reversing these changes. It appears that, consistent with the predictions, the common themes of gene families and protein functional groups showing this pattern are the removal of lipid and protein aggregates, apoptosis, and mitochondrial functionality. Perhaps the most striking reversals have been observed in apolipoprotein D paralogs (ApoD’s) and in several lysosomal lipases. Human and Drosophila ApoDs have long been known to play a key role in counterbalancing oxidative stress and other damages in aging cells and, at least in Drosophila, increase lifespan [25, 31, 53, 54, 63, 75]. In flies, overexpression of either human ApoD and native Drosophila orthologs, GLaz, and Nlaz reduces age-associated lipid peroxide accumulation and increases lifespan [54, 75], while loss of function GLas mutants have reduced lifespan [63]. This regulation is paralog- and sex-specific [62], indicating a possible connection to female reproductive function.

Likewise, what we know about lysosomal lipases and their role in aging is consistent with the above hypothesis. Overexpression of lysosomal lipases increases lifespan in C. elegans and this occurs through activation of autophagy [23, 31, 38, 39, 58, 76]. Although this mechanism can operate also in germline-less worms [38], there is compelling evidence that it occurs through interaction between somatic maintenance and the germ line [76], again indicating a connection with reproduction function and a role reproduction-longevity trade-off.

The idea that DE of ApoD’s and lysosomal lipases in the old Daphnia and its reversal in PSRE ones may be interrelated is further corroborated by the fact that the role of ApoD in ameliorating severity of the neurogenerative Niemann–Pick disease occurs through preservation of lysosomes function [59]. Because this rescuing of lysosome function occurs through restoring pH gradient across the lysosome membrane, this may be a mechanism compensating the age-related loss of transcription of V-type proton ATPases also observed in this study. Furthermore, more recently, human ApoD has been identified as not just a marker of aging, but also a potential tool for achieving rejuvenation in human skin tissue [69].

With this in mind, the up-regulation of ApoD’s and lysosome lipases in aging Daphnia can be interpreted as the response to age-related oxidative stress operating through activation of liposomal removal of lipid aggregates, TAG hydrolysis, and autophagy. Then, the across-the-board reversal of ApoDs’ and lysosomal lipases’ expression in the PSRE Daphnia can be interpreted in several possible ways. Are the PSRE individuals simply those who, due to random developmental variation, have not accumulated products of oxidative stress, making it both unnecessary to up-regulate the ApoDs, and possible to resume reproduction? Or did previous overexpression of these proteins during the non-reproductive phase reduce such accumulation, thus being causative for the future reproductive “rejuvenation”? Or is the anti-aging, oxidative stress-reducing activity of ApoDs and lipases not compatible with active reproduction, requiring the PSRE Daphnia to bypass the anti-aging protective plasticity? If the latter, then is perhaps ApoDs’ and lysosomal lipases’ activity related to a fundamental trade-off between maintenance and reproduction? It is not immediately obvious what experiments might help differentiate between these possibilities. A knock-out study might, but it would be necessary to knock out several paralogs in each gene family.

Similar reasoning, perhaps related to both apoptosis and lysosomal removal of protein aggregates, applies to the same pattern observed in serine and cysteine proteases—upregulation in aging Daphnia reversed in the PSRE ones. We know that various serine proteases participate in a variety of aging-related aggregate removal [78], apoptosis [21, 47], and mitochondrial homeostasis ([32]; see below) pathways in yeast and mammals. However, the extreme diversity of functions of serine- and cysteine proteases makes it difficult to formulate specific hypotheses.

It is more difficult to interpret the transcriptional pattern observed in V-type proton ATPases, whose central function is maintaining acidic pH in lysosomes, critical for lysosomes’ lipid and protein hydrolysis function [5]. On the one hand partial restoration of their RNA abundancies in the PSRE Daphnia is to be expected (cf. [7]). However, the reduced transcription in senescent Daphnia does not fit the above damage-control reasoning applied to ApoDs and lipases. Perhaps here we observe the situation in which age-related transcriptional changes are caused not by mobilization of protective mechanisms, but by their decline. Either maintaining low pH in lysosomes becomes too expensive (perhaps due to reduced ATP production, see below) or unnecessary due to reduced reproduction, if the role of these ATPases is similar to that described by Bohnert and Kenyon [7], i.e., crucial for mainlining rejuvenated germline.

Likewise, the down-regulation of several myocyte-specific proteins is consistent with the observed reduction of locomotory and filter-feeding activity in aging Daphnia [2], Pearson et al., in preparation). Because Daphnia, like any other zooplankton filter-feeding organisms, has to constantly swim to stay in the water column and filter water to obtain food, reduced muscular activity in nature is likely to lead to mortality.

One other pattern of transcriptional changes associated with aging and PSREs we have observed was the opposite direction of DE changes in different paralogs, suggestive of possible aging-related specialization, and, in some cases, possibly paralog retention by subfunctionalization. There were several gene families showing this pattern, some of which with functionality without any a priori connection to aging or rejuvenation. These include ATP-dependent RNA helicases, chorion peroxidases, cuticle proteins, solute carriers, T-complex proteins, and zinc finger proteins. Two other functional groups, namely G-protein coupled receptors and serine/threonine-protein kinases, have known aging-related functionalities, but again, the large number of paralogs within these groups in most genomes and the great diversity of their functions make it difficult to formulate specific hypotheses about the roles of these particular paralogs in aging and rejuvenation.

Conclusions

Post-senescence reproductive evens in Daphnia are not followed by elevated mortality as predicted by the terminal investment hypothesis. Rather, they represent a transcriptionally unique state in which transcription of several gene families and functional groups of genes showing age-related decline or increase in transcription is reversed to the “youthful” levels. Functionally, this pattern is observed in genes implicated in lipid metabolism, lysosomal functions, removal of lipid and protein aggregates, apoptosis, autophagy, and mitochondrial and muscle functions.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

Impetus Foundation grant to LYY.

Data availability

All life-history data necessary to recreate the results reported here are available from the Supplementary Data file. Raw and mapped reads from the RNAseq experiment are available from NCBI, BioProject PRJNA1098345; SRA accession numbers SRR28604180–SRR28604191; processed data from NCBI GEO GSE263788, accession numbers GSM8199167–GSM8199178.

Declarations

Competing interests

The authors declare no competing interests.