Abstract
The roundworm C. elegans transiently arrests larval development to survive extended starvation (1), but such early-life starvation reduces reproductive success (2, 3). Maternal dietary restriction (DR) buffers progeny from starvation, increasing reproductive success (4). It is unknown why early-life starvation decreases reproductive success and how maternal diet modifies this process. We show here that extended starvation in first-stage (L1) larvae followed by unrestricted feeding results in a variety of abnormalities in the reproductive system, including glp-1/Notch-sensitive germ-cell tumors and uterine masses that express neuronal and epidermal markers. We found that maternal DR reduces the penetrance of starvation-induced abnormalities, including tumors. Furthermore, we show that maternal DR reduces insulin/IGF signaling (IIS) in progeny, and that daf-16/FoxO and skn-1/Nrf, transcriptional effectors of IIS, are required in progeny for maternal DR to suppress abnormalities. daf-16/FoxO activity in somatic tissues is sufficient to suppress starvation-induced abnormalities, suggesting cell-nonautonomous regulation of reproductive system development. This work reveals complex inter- and intra-generational effects of nutrient availability mediated by IIS with consequences on developmental integrity and reproductive success.
One Sentence Summary Intergenerational effects of diet on IIS
Main Text
To determine how extended larval starvation compromises reproductive success, we compared early adult worms starved for 8 days as L1-stage larvae to control adults that were starved briefly, hereafter referred to as “starved” and “control”, respectively. Approximately half of the starved worms displayed prominent abnormalities in their reproductive system despite being well fed after the starvation period (Fig. 1A-C). These abnormalities were highly variable, both among worms and between gonads in an individual worm. The most common abnormalities fell on a spectrum ranging from gonads enlarged with proliferative germ cell nuclei (detected as Pmex-5::H2B::mCherry-positive, “mex-5+”) to uterine masses consisting of largely Pmex-5::H2B::mCherry-negative cells (“mex-5-”, Fig. 1C). The mex-5+ gonad enlargements resembled proximal germ cell tumors similar to those caused by “latent niche” signaling (5, 6), and the mex-5- uterine masses were disorganized and appeared to contain differentiated cells. Indeed, the majority of uterine masses expressed elt-1 and unc-119 reporter genes, embryonic markers of epidermis and neurons, confirming somatic differentiation (Fig. 1D and E). We observed similar abnormalities in three wild isolates subjected to 8 d L1 starvation (Fig. S1), suggesting that these abnormalities were not an artifact of the laboratory strain. Individuals with abnormalities produced smaller broods (Fig. 1F), consistent with developmental abnormalities limiting reproductive success.
The Notch receptor GLP-1 and the RNA-binding protein GLD-1 regulate germline differentiation (7). glp-1 activity maintains germline stem cells by preventing differentiation, and gain-of-function mutants develop germ cell tumors (8, 9). We found that the germline abnormalities we see in starved wild-type worms resemble proximal tumors seen in glp-1(ar202) gain-of-function mutants, and starvation increased penetrance of tumor formation in the mutant (Fig. 2A and B)(8). Interestingly, the presence of abnormalities was sensitive to glp-1 dosage, since few animals heterozygous for the glp-1(e2072) loss-of-function mutation (10) displayed germline abnormalities following L1 starvation (Fig. 2B). gld-1 promotes entry into meiosis and differentiation, and class A loss-of-function mutants develop germ cell tumors (11). In our control conditions, early-adult gld-1(q485) null mutants developed proximal germ cell tumors but heterozygous mutants did not, as expected. In worms subjected to 8 d L1 starvation, however, heterozygous mutants developed tumors and the size of the tumors in homozygous mutants was greatly enlarged (Fig. 2C and D). gld-1 mutants do not perform physiological apoptosis (12, 13). Thus, the absence of large tumors in control gld-1 mutants and our glp-1 results suggest that extended L1 starvation causes hyperproliferation of germ cells, disrupting coordination of germ line development. Based on our observation of proliferative germ cells in the proximal gonad (adjacent to the vulva), we believe that extended starvation causes heterochronically discordant somatic and germline development resulting in signaling from the “latent niche” (modeled in Fig. 2E)(5, 6).
Maternal DR buffers progeny reproductive success from the costly effects of L1 starvation (4), so we wondered whether maternal DR reduces the incidence of germline abnormalities caused by extended L1 starvation. We acquired progeny from parents fed ad libitum (AL) or DR using well-controlled liquid-culture conditions (4, 14) and starved them for 8 d as L1 larvae. Fewer DR progeny developed germline abnormalities after starvation than AL progeny (Fig. 3A), revealing an intergenerational effect of DR on pathological consequences of early-life starvation.
Analysis of transcript abundance by RNAseq revealed intergenerational effects of maternal DR on progeny gene expression. 114 genes were differentially expressed on the first day of L1 starvation in DR progeny compared to AL progeny (FDR < 0.1; Table S1). Gene Ontology (GO) term enrichments included terms related to IIS, including “innate immune response” and “regulation of dauer entry” (Table S2). In particular, expression of the only known IIS receptor daf-2 was decreased in DR progeny (Fig. 3B). DAF-2/InsR antagonizes function of the FoxO transcription factor DAF-16 in response to IIS (15), and regulatory targets of daf-16/FoxO define a transcriptional signature of reduced IIS (16). Despite the relatively small number of differentially expressed genes, genes up-regulated in DR progeny were significantly enriched for Class I daf-16 targets (activated by daf-16; p < 0.05), and down-regulated genes were significantly enriched for Class II targets (repressed by daf-16; p < 0.001) (Table S3), consistent with reduced IIS in DR progeny. DAF-2/InsR regulates localization of DAF-16, with DAF-16 shifting from cytoplasmic to nuclear when IIS is reduced (15). GFP::DAF-16 displays strong nuclear localization during L1 starvation (17), but appeared even more completely localized to nuclei in DR progeny (Fig. 3C). GFP::DAF-16 was relatively more cytoplasmic in fed than starved larvae, as expected, but it remained relatively more nuclear in DR progeny (Fig. 3C). These results collectively support the conclusion that maternal DR decreases IIS in progeny.
To determine whether intergenerational reduction of IIS caused by maternal DR has functional consequences, we examined the effects of reduced daf-16/FoxO and skn-1/Nrf, both effectors of IIS (15). Disruption of either gene by RNAi did not alter the incidence of abnormalities in AL progeny starved for 8 d as L1 larvae, but they were each required for maternal DR to reduce the frequency of abnormalities (Fig. 3D). These results further support the conclusion that maternal DR reduces progeny IIS, protecting progeny from pathological consequences of extended starvation by promoting the activity of daf-16/FoxO and skn-1/Nrf.
Fewer starved larvae that recover in low IIS conditions display germline abnormalities. Dramatically fewer starved larvae recovering in DR or in conditions that induce dauer diapause (18) (19) (followed by recovery from dauer in AL conditions) displayed abnormalities (Fig. 4A and Fig. S2A). These results are consistent with a protective effect of reduced IIS during development after starvation, but other pathways affected by these conditions could also be involved. Mutation of daf-2/InsR also suppressed abnormalities (Fig. 4B). Likewise, skn-1/Nrf gain-of-function mutants suppressed abnormalities in daf-16-independent fashion (Fig. 4C). Though consistent with reduced IIS suppressing starvation-induced germline abnormalities, it is unclear when during the lifecycle these mutations exert their effect. We therefore used RNAi to perturb gene function during development after starvation. RNAi of daf-2 (or double RNAi of daf-2 and gfp) suppressed the frequency of abnormalities (Fig. 4D and E), demonstrating that reducing IIS during recovery of previously starved larvae suppresses abnormalities. Suppression depended on daf-16/FoxO and skn-1/Nrf (Fig. 4E), as seen with maternal DR (Fig. 3D). To determine whether these results were due to RNAi depletion of IIS genes in the germline or soma, we performed RNAi in mutants that largely restrict RNAi to these tissues (20, 21). The effects of daf-2 and daf-16 RNAi were abrogated in a mutant background that restricts RNAi primarily to the germ line but were retained in a mutant background that restricts RNAi primarily to the soma (Fig. 4F), indicating that these genes function in the soma to suppress the observed abnormalities in the reproductive system. Tissue-specific transgenic rescue of a daf-16 null mutant showed that overexpression of daf-16 without daf-2 RNAi is sufficient to suppress abnormalities but that daf-2 RNAi enhances suppression (Fig. 4G). These results corroborated results using double RNAi (Fig. 4E and F), and they revealed that daf-16 can function cell-nonautonomously in the epidermis, intestine or neurons to regulate post-starvation germline development. These sites of action have been shown for daf-16 in regulating developmental arrest and organismal aging (22-25), but not germline proliferation where daf-2 acts germ cell-autonomously in late-larval stages to promote germ cell proliferation (26) but acts cell-nonautonomously in the somatic gonad during aging to antagonize germline progenitor cell loss (27). Thus, the role of IIS in germline development during recovery from L1 starvation involves a different mechanism. Nonetheless, multiple lines of evidence support the conclusion that reduction of somatic IIS during development suppresses starvation-induced germline abnormalities.
We used a gld-1 mutant to further investigate the effect of IIS on germ cell tumor proliferation. 8 d L1 starvation increased the size of proximal tumors in this differentiation-defective mutant (Fig. S2C and D), as before (Fig. 2C and D). daf-2 RNAi suppressed the effect of starvation, decreasing tumor size, in daf-16-dependent fashion (Fig. S2C and D). These results suggest that reduction of IIS after extended L1 starvation inhibits the hyperproliferative state of germ cells induced by starvation.
In addition to reduced brood size, extended L1 starvation results in production of abnormally shaped, small embryos with reduced hatching efficiency (2). Reduction of IIS by daf-2 mutation or RNAi suppressed production of abnormal embryos, and it increased brood size following 8 d L1 starvation (Fig. 4H and I). These results further demonstrate that the starvation-induced germline abnormalities we have described compromise reproductive success, presumably reducing organismal fitness, and they show that reducing IIS following starvation increases reproductive success.
We have shown that early-life starvation leads to developmental abnormalities including tumors in the reproductive system of C. elegans, and that these abnormalities reduce reproductive success (Fig. 4J). Our results suggest that extended starvation followed by AL feeding causes hyperproliferation of germ cells, disrupting regulation of germ cell development and leading to tumor formation (5, 6). Remarkably, maternal DR mitigates these pathological consequences of starvation. We show that maternal DR reduces progeny IIS, and that somatic reduction of IIS following starvation suppresses germ cell hyperproliferation and development of abnormalities via activation of daf-16/FoxO and skn-1/Nrf, increasing reproductive success. These results add to an expanding body of work in C. elegans demonstrating intergenerational effects of the environment mediated by IIS (4, 28-30). Insulin signaling is frequently altered in mammalian models of maternal dietary effects on offspring (31), suggesting conservation of a central role of IIS in mediating physiological trade-offs between generations.
Materials and Methods
Worm maintenance
Worms were maintained under standard laboratory conditions at 20°C unless otherwise noted. Animals were faithfully passaged without dietary restriction (thinning of the E. coli lawn) or starvation for many generations (greater than five) prior to commencing experiments unless otherwise noted.
Strains used in this study
N2 (Bristol), GC1171 naSi2[pGC550(Pmex-5<mCherry::H2B::nos-2 3’UTR<GFP::H2B::nos-2 3’UTR - unc-119(+))] unc-119(ed3), DP132 edIs6, OP354 unc-119(tm4063) III; wgIs354, CB1370 daf-2(e1370), SPC167 dvIs19; skn-1(lax120), SPC168 dvIs19; skn-1(lax188), CF1038 daf-16(mu86), NK1228 qyIs288; daf-16(mu86); unc-119(ed4), NK1229 qyIs290; daf-16(mu86); unc-119(ed4), NK1231 qyIs291; daf-16(mu86); unc-119(ed4), NK1233 qyIs293; daf-16(mu86); unc-119(ed4), NL2550 ppw-1(pk2505), NL2098 rrf-1(pk1417), JK3025 gld-1(q485)/hT2 [bli-4(e937) let-?(q782) qIs48], GC833 glp-1(ar202), JK1505 unc-32(e189) glp-1(e2072)/eT1, and wild-type isolates: CB4856, ED3077, and JU561.
mex-5 reporter construction
The naSi2 allele results is an insertion of sequences from the plasmid pGC550 by MosSCI into strain EG4322(32). The insertion contains C. briggsae unc-119 sequences and an FRT-flanked mex-5p::mCherry::H2B:nos-2 3’ cassette followed by H2B::GFP. naSi2 expresses mCherry::H2B in all germ cells. pGC550 was generated by a multisite LR reaction performed between pJA252 (gift from Julie Ahringer, Addgene plasmid # 21512, (33)), pGC544 [containing an FRT site (“<”) <mCherry::H2B:: nos-2 3’UTR sequences and inserted into pDONR-2221 by BP reaction], pGC545 [containing Pmex-5<mCherry:: nos-2 3’UTR<GFP::nos-2 3’UTR], and pCFJ150 - pDESTttTi5605[R4-R3] (gift from Erik Jorgensen, Addgene plasmid # 19329, (32)). All plasmids and sequences are available from Addgene or upon request.
Recovery from Starvation Assay
C. elegans’ embryos were prepared with sodium hypochlorite treatment, washed several times, and allowed to hatch in S-basal without additions. Animals were arrested for one or eight days (unless otherwise noted) prior to being plated on lawns of OP50 or HT115 E. coli. Animals were incubated at 20°C unless otherwise noted and then assayed after three or four days of development.
Germline Microscopy and Scoring of Germline Abnormalities
Following recovery, adults were picked at random into 25 mM sodium azide on 2% noble agar pads. The worms were then examined using Nomarski microscopy on an AxioImager compound microscope (Zeiss) at 200-400x magnification. Animals with atypical gonads were scored as abnormal. A majority of the animals scored as abnormal had variations of proximal tumors that varied in the extent to which the germ cells were differentiated ranging from completely undifferentiated and mex-5+ to highly differentiated and mex-5-. However, starved animals also developed a variety of other abnormalities at varying rates including multiple, protruded, and extruded vulvas; gonad migration defects; egg-laying defects; endomitotic oocytes; and oocyte formation failures similar to those seen in gld-2 and gld-3 mutants (data not shown). When scoring glp-1gf(ar202), dumpy individuals were not scored. Animals subjected to treatments and mutations which delayed development were scored later (up to 5 d after initiating recovery) to maximize chances of visualizing abnormal germline development. Animals were censored if they were delayed to the extent that the normality of their germline could not be determined at time of scoring. Homozygous gld-1 mutant tumors were classified based on germ cell morphology for Figure 2D. For Figure 1E, each uterine mass was assessed for the presence or absence of Pelt-1 or Punc-119::GFP at 400-1000x.
Gonad imaging
Worms were imaged at 400x magnification, using an AxioCam camera equipped with Zen software, on an AxioImager compound microscope (Zeiss). Images were merged using Fiji. Images were cropped and features were added using Inkscape.
Determination of brood size
Animals were prepared as described above and plated onto the indicated bacterial lawn. After 48 hours of recovery, animals were singled onto plates seeded with the corresponding bacteria. Worms were transferred daily onto fresh plates until egg laying ceased and progeny were scored. Animals that could not be found were censored. For Fig. 1F, worms were inspected for germline abnormalities after 72 hours of recovery via dissection scope and otherwise scored as described.
Abnormalities in progeny of AL and DR parents
Progeny from AL and DR parents were generated as previously described for use in this study (4). Briefly, following synchronization by brief L1 arrest, worms were cultured in S-complete at a density of 10 per mL with 25 mg/ml (AL) or 3.13 mg/ml (DR) E. coli HB101 at 20°C and 180 rpm for 96 hr. Progeny embryos were prepared by hypochlorite treatment.
RNAseq in progeny from AL and DR parents
Progeny from AL and DR parents were cultured in S-basal for 24 hours so they hatch and enter L1 arrest, and then they were flash frozen in liquid nitrogen. RNA was prepared with Trizol (Invitrogen) according to the manufacturer’s instructions except that sand was included in homogenization. Libraries were prepared from 500 ng of total RNA and 12 PCR cycles using NEB Next Ultra RNA Library Preparation Kit (New England Biolabs). Sequencing was performed on an Illumina HiSeq 4000. Bowtie was used to map reads to the WS210 genome, including transcripts annotated in WS220 mapped back to WS210; (34, 35). HTSeq was used to generate count tables for each library (36). edgeR was used on count tables to determine differentially expressed genes. Detected genes were considered those with counts per million (CPM) > 1 across 3 libraries. This resulted in 13,439 genes for differential expression analysis. The “calcNormFactors” function was used and tagwise dispersion estimate was used for differential expression. Genes with a false-discovery rate of less than 0.1 were considered differentially expressed.
daf-16-target gene signature analysis
A consensus background gene set was produced by overlapping all detected genes in the progeny of DR parents dataset with those reported by Tepper et al (16) using the R package gplots. Next, differentially expressed genes up- or down-regulated in the progeny of DR parents compared to AL parents (FDR < 0.1) were overlapped with high confidence (FDR < 0.001) daf-16-target genes as determined by Tepper et al (16). Hypergeometric p-values were determined using the Graeber lab calculator (http://systems.crump.ucla.edu/hypergeometric/index.php).
Gene ontology analysis
Genes differentially expressed in the progeny of DR parents (114) were analyzed for enrichment of gene ontology terms against the 13439 detected genes (criteria detailed above) using GOrilla (37).
GFP::DAF-16 localization in the progeny of AL and DR parents
Progeny of AL and DR parents were prepared as previously described for an N-terminal DAF-16 translational fusion reporter strain (NK1228). The extent of intestinal GFP localization was quantified at 400x or 1000x depending on stage using an AxioImager compound microscope (Zeiss).
Starvation and DR recovery
Animals were starved as described above, except instead of recovery on plates animals were recovered in DR liquid culture as previously described prior to gonad assessment at 96 hr.
Starvation to Dauer to AL recovery
Wild-type animals were starved and then recovered in dauer-forming conditions (18) (5 worms per µl and 1 mg/ml HB101 in S-complete at 20°C and 180 rpm) for 5 d then recovered on plates (AL). After animals reached early adulthood, their gonads were assessed for the presence or absence of germline abnormalities as described above.
RNAi
Single colonies, grown on 100 µg/ml carbenicillin and 12.5 µg/ml tetracycline LB plates, were inoculated into LB starter cultures with the same antibiotics. These were used to inoculate larger cultures grown in 50 µg/ml carbenicillin. Cells from large cultures were spun down at 4°C and resuspended at a concentration of 25 mg/ml in 15% glycerol S-complete medium. Aliquots were frozen and thawed only once and seeded onto NGM plates containing 25 µg/ml carbenicillin and 1mM IPTG. For double RNAi experiments, equal volumes of RNAi suspension were mixed. Lawns were allowed to grow overnight at room temperature before adding worms. Empty vector (pAD12), daf-2 (pAD48), and daf-16 (pAD43) clones were gifts from Coleen Murphy. skn-1 clones came from the Ahringer library.
Data analysis
t-tests were performed using R or Microscoft Excel. Bartlett’s tests were performed to determine if variance could be pooled. Two-way ANOVA was performed using R. Plots were generated using the R package ggplot2 or Microsoft Excel. The R package plotrix was used to calculate SEM. The R package Linear and Nonlinear Mixed Effects Models “nmle” was used to compute the brood size interaction p-value (Fig. 4I).
Table S1. Genes differentially expressed in the progeny of DR parents compared to AL parents during L1 arrest.
[See accompanying .csv file]
Acknowledgments
We thank the Duke University School of Medicine and the Center for Genomic and Computational Biology for use of the Sequencing and Genomic Technologies core resource, which provided RNA sequencing service. Yossi Capua, Amanda Fry, and Theadora Tolkin for generating and validating naSi2. Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). This work was funded by the National Institutes of Health (LRB, R01GM117408; EJAH, R01GM061706). JMJ, JDH, and LRB conceived and interpreted the experiments. JMJ and LRB wrote the manuscript. JMJ, JDH, REWK, AL, RG, and EAB conducted the experiments. JMJ, AKW, and CSM analyzed RNAseq data. EJAH provided an invaluable reagent and germline biology expertise. Authors declare no competing interests. RNAseq data set can be found at GEO using accession number GSE114271. Figure 1A adapted from Caenorhabditis elegans hermaphrodite adult-en.svg from Wikimedia Commons by K. D. Schroeder (License: CC-BY-SA 3.0).