Transcriptome- and phenotype-based epistasis analysis in Caenorhabditis elegans reveals daf-16/FoxO-dependent and independent effects of daf-2/InsR in L1 starvation and recovery
Kinsey Fisher, Rojin Chitrakar, L Ryan Baugh
TL;DR
This study in C. elegans shows that reduced insulin signaling during starvation affects survival through DAF-16/FoxO but preserves reproduction via an unknown mechanism.
Contribution
Identification of 4,653 putative DAF-16/FoxO targets and evidence for DAF-16-dependent and -independent effects of insulin signaling during starvation.
Findings
Disruption of daf-2/InsR during starvation causes daf-16-dependent changes in gene expression and survival.
DAF-16 is not required for reproduction after starvation, and daf-2 loss preserves reproduction independently of DAF-16.
Reduced insulin signaling during L1 starvation is largely DAF-16-dependent but involves additional effectors for reproduction.
Abstract
Reduced insulin/IGF signaling (IIS) in Caenorhabditis elegans increases starvation resistance in a daf-16/FoxO-dependent fashion, but it is unclear whether the effects of reduced IIS are entirely dependent on daf-16/FoxO. We used RNA sequencing and phenotypic analysis of L1 starvation resistance to assess epistasis between daf-2/InsR and daf-16/FoxO. We identified 4,653 putative DAF-16/FoxO targets, many of which had not been previously identified, providing a valuable reference data set. Differential gene expression and increased survival caused by disruption of daf-2/InsR during starvation are daf-16-dependent. The effect of daf-2/InsR on growth following starvation is largely but not entirely daf-16-dependent. Notably, daf-16 is dispensable for reproduction following extended starvation, and daf-2 loss preserves reproductive success independent of daf-16. These results show that the…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Fig. 1
Fig. 2
Fig. 3| Comparison | Number of upregulated genes | Number of downregulated genes |
|---|---|---|
| wild type starved / wild type fed | 5,703 | 5,275 |
|
| 1,618 | 1,587 |
|
| 866 | 1,093 |
|
| 1,086 | 1,183 |
|
| 2,282 | 2,371 |
|
| 22 | 12 |
- —National Institutes of Health10.13039/100000002
- —NIH Office of Research Infrastructure Programs10.13039/100016958
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGenetics, Aging, and Longevity in Model Organisms · FOXO transcription factor regulation · Muscle Physiology and Disorders
Introduction
Nutrient availability has profound effects on development and metabolism. The nematode Caenorhabditis elegans consumes ephemeral microbial food sources in the wild, and it has evolved robust mechanisms to sense and respond to starvation (Baugh and Hu 2020). When C. elegans larvae hatch in the absence of food, they reversibly arrest development in the first larval stage and persist in a state known as L1 arrest (or L1 diapause) (Baugh 2013). Arrested larvae can survive for weeks and resume development upon feeding (Johnson et al. 1984). Several thousand genes are differentially expressed in fed vs. starved L1 larvae (Baugh et al. 2009), providing a valuable animal model for nutritional control of gene expression. L1 arrest and recovery also provides a powerful model to study starvation resistance, including immediate effects of starvation on survival as well as persistent effects on growth and reproduction upon feeding (Baugh and Hu 2020).
The widely conserved insulin/IGF-1 signaling pathway (IIS) coordinates nutrient status with development and metabolism. The sole known C. elegans insulin/IGF-1 receptor DAF-2/InsR signals through a PI3K/AKT kinase cascade to antagonize the activity of the FoxO transcription factor DAF-16 (Lin et al. 1997; Ogg et al. 1997). IIS is reduced during starvation, and DAF-16/FoxO activates transcription of genes that support survival and promote developmental arrest (Muñoz and Riddle 2003; Baugh and Sternberg 2006; Kaplan et al. 2015; Hibshman et al. 2017). These observations have established daf-16 as an essential effector of reduced IIS during L1 arrest (Baugh and Hu 2020). However, DAF-16 also promotes adult longevity (Murphy and Hu 2013), but DAF-16 is not the only effector of reduced IIS in this context. The transcription factor SKN-1/Nrf is also antagonized by PI3K/AKT signaling downstream of DAF-2, and skn-1 promotes longevity independently of daf-16 (Tullet et al. 2008). It is unclear whether DAF-16 is the sole effector of IIS during L1 arrest and recovery.
We analyzed epistasis between daf-2/InsR and daf-16/FoxO during L1 arrest and recovery. For L1 arrest, we analyzed gene expression using RNA-seq and starvation survival. For recovery from arrest, we analyzed growth rate and reproductive output. We assayed each of these phenotypes since we are interested in immediate effects of starvation and IIS during arrest as well as persistent effects of extended starvation, which are also important to fitness. Our primary objective was to determine whether the effects of reduced IIS are entirely dependent on alteration of daf-16 activity or if there are daf-16-independent effects of reduced IIS. Our RNA-seq analysis provides a valuable reference dataset for the investigation of the role of IIS in mediating nutritional control of gene expression. Our results suggest the effects of reduced IIS on gene expression and survival during L1 arrest are daf-16-dependent and that reduced IIS elicits daf-16-dependent and independent effects on growth and reproduction, with no detectable contribution of daf-16 to reproductive success following extended starvation.
Materials and methods
C. elegans maintenance and strains
C. elegans strains were maintained on nematode growth medium (NGM) plates at 20 °C, seeded with Escherichia coli OP50. The following strains were used: N2 (wild type; Sternberg Lab at Caltech); PS5150 [daf-16(mgDf47)] (Sternberg Lab at Caltech); CB1370 [daf-2(e1370)] (CGC); and GR1309 [daf-16(mgDf47); daf-2(e1370)] (CGC). daf-16(mgDf47) was named PS5150 after being backcrossed.
Sample preparation and collection for bulk RNA-seq
Seven L4 larvae were picked onto each of 5 to 7 10 cm plates seeded with OP50 and cultured at 20 °C for 96 h (wild type, daf-16(mgDf47), and daf-16(mgDf47); daf-2(e1370)) or 120 h (daf-2(e1370)) (the daf-2 mutant develops slower than the other genotypes). Plates were washed with S-basal and treated with hypochlorite solution to obtain sterile, developmentally synchronized embryos. A total of 20,000 embryos were placed in 20 mL of S-complete with 1 × E. coli HB101 in a 50 mL flask for the wild-type fed sample, and 20,000 embryos were placed in individual cultures of 20 mL S-basal with 0.1% EtOH in a 50 mL flask for starved samples. All flasks were incubated at 20 °C and 180 rpm for 24 h. With these conditions, it takes approximately 12 h for embryos to hatch following hypochlorite treatment. Samples were then washed 3 × with S-complete (fed) or S-basal with 0.1% EtOH (starved) before being pelleted and snap frozen with liquid nitrogen. Samples were stored at −80 °C until RNA isolation and library preparation. At least 3 biological replicates were collected for each condition.
RNA isolation and library preparation
RNA was isolated with TRIzol reagent (Invitrogen #15596026) as per the manufacturer's instructions, except 100 μL of acid-washed sand (Sigma-Aldrich #27439) was added to each sample at the beginning of the extraction protocol to aid with homogenization. RNA was eluted in nuclease-free water and stored at −80 °C until further use. Libraries were prepared for sequencing using the NEBNext Ultra II RNA Library Prep Kit for Illumina (New England Biolabs #E7775) starting with 200 ng of total RNA per sample as input and 10 cycles of PCR. Individually barcoded libraries were pooled and sequenced on the NovaSeq 6000 S-Prime flowcell to obtain 50 bp paired-end reads.
Read mapping and differential expression analysis
Version WS273 of the genome was used for mapping reads. Bowtie version 1.2.3 was used to map paired-end reads with the following settings: bowtie -I 0 -X 500 -k 1 -m 2 -S -p 2 (Langmead et al. 2009). HTSeq version 0.11.2 was used to count reads mapping to the WS273 canonical geneset (Putri et al. 2022). Count data was restricted to include only protein-coding genes. Exact tests in edgeR version 3.24.3 were used for differential expression analysis between pairs of conditions (Chen et al. 2025). Only genes with counts per million (CPM) > 1 in at least 4 libraries across all conditions were included in differential expression analysis.
Principal component analysis
Principal component analysis (PCA) was performed on log_2_ mean-normalized CPM values for the genes included in differential expression analysis. PC1 and PC2 were plotted for all samples using ggplot2 (Wickham 2016) with 95% confidence intervals.
Hierarchical clustering and heatmap
The glmQLFit and glmQLFTest functions in edgeR were used to identify 2,095 differentially expressed genes across all the starved samples for all 4 genotypes (FDR < 0.05). Log_2_ fold-change of mutant over wild type (starved) was plotted using the “pheatmap” package in R with hierarchical clustering on gene expression and genotype.
Comparison to published datasets
DAF-16 targets from our dataset consist of genes that were differentially expressed in daf-16(mgDf47); daf-2(e1370) vs daf-2(e1370) (FDR < 0.05). The overlap of these genes with class I and class II genes (Tepper et al. 2013) was performed using the “eulerr” package in R to create geometrically accurate depictions of the overlaps and list sizes (Larsson 2018). A similar analysis was used to compare our DAF-16 targets with those identified in Kaplan et al. (2015). Using their S1 dataset, genes were filtered to only include protein-coding genes using the “coding_status” column. Hypergeometric P-values for the overlaps were calculated using the phyper() function in R.
Gene Ontology-term analysis
Gene Ontology (GO) enrichment analysis for “biological process” (Ashburner et al. 2000; Thomas et al. 2022; Gene Ontology Consortium et al. 2023) was performed separately for genes that were significantly up- and downregulated in the daf-16(mgDf47); daf-2(e1370)/daf-2(e1370) comparison (2,282 and 2,371, respectively; FDR < 0.05) using the clusterProfiler R package (Xu et al. 2024) with a q-value cutoff of 0.2 for GO enrichment. GO terms were simplified to remove redundancy using clusterProfiler (simplify (cutoff = 0.4), by = “p.adjust”). Semantic similarity was calculated using the GOSemSim function (method = “Wang”) (Wang et al. 2007; Yu et al. 2010), reduced to 2 dimensions using cmdscale(), and visualized with ggplot2.
Tissue enrichment analysis
Up- and downregulated genes in daf-16(mgDf47); daf-2(e1370) vs. daf-2(e1370) were separately entered into the WormBase tissue enrichment analysis tool (Angeles-Albores et al. 2016; Sternberg et al. 2024) with all genes detected in RNA-seq (14,322 genes) used as the background set. The top 10 most significant terms by Q-value were plotted for Fig. 1e.
Differential gene expression due to reduced IIS during L1 arrest is daf-16/FoxO-dependent. a) Principal component analysis (PCA) separates samples based on nutrient availability and genotype, with daf-16(mgDf47) and daf-16(mgDf47); daf-2(e1370) being indistinguishable. Each point represents a biological replicate for a specific genotype/condition. Ellipses represent 95% confidence intervals. The experimental factor that correlates with each principal component is included with the axis labels. b) Hierarchical clustering suggests the effects of daf-2(e1370) on gene expression during starvation are daf-16-dependent. Log2 fold-change of each starved mutant relative to starved wild type is plotted for 2,094 differentially expressed genes (FDR < 0.05 across starved samples; GLM-QL). c) Overlap of differentially expressed genes in daf-16(mgDf47); daf-2(e1370) vs. daf-2(e1370) and class I and II targets from Tepper et al. (2013) and differentially expressed genes in daf-16(mgDf47) mutants during L1 arrest in Kaplan et al. (2015). All intersections are highly significant based on a hypergeometric test. Significant enrichments are indicated with a P-value, and significant depletions (1—enrichment P-value < 10−16) are indicated with an asterisk. d) GO-term enrichments for “biological process” for differentially expressed genes in the daf-16(mgDf47); daf-2(e1370)/daf-2(e1370) comparison (2,282 up and 2,371 down; FDR < 0.05) are plotted. The terms are plotted in semantic space using the Wang method (Wang et al. 2007). Each point represents a significantly enriched GO term, and point size reflects the −Log₁₀-adjusted P-value of the term enrichment. Color indicates gene expression direction (yellow for upregulated, blue for downregulated). e) Top 10 most significantly enriched tissues for genes that are differentially up- (“up genes”, yellow) or downregulated (“down genes”, blue) in daf-16(mgDf47); daf-2(e1370) vs. daf-2(e1370) are plotted.
Starvation cultures
Embryos were collected and synchronized by hypochlorite treatment as they were for the RNA-seq starved cultures (see above). A total of 5,000 embryos were placed into 5 mL of S-basal plus 0.1% ethanol (no cholesterol) in 16 mm glass test tubes at 20 °C in the dark on a tissue culture roller drum at ∼30 rpm. Embryos hatched and arrested as L1 larvae, and aliquots from these cultures were used for starvation survival and recovery assays.
Starvation survival
Each day of L1 arrest, 100 to 150 μL of each starvation culture (150 μL used when survival was low) was plated onto the edges of a 6 cm NGM plate seeded with a small lawn of E. coli OP50 in the center and placed at 20 °C. Two days after plating, survival was scored as the number of worms that were crawling on the lawn divided by the total number of worms originally plated. Four biological replicates were used for each genotype. A quasi-binomial generalized linear model was used to fit the proportion of worms alive to the duration of starvation for each replicate. Based on the model, median survival for each replicate was calculated. Medians of each group of replicates per condition were tested for variance homogeneity using Bartlett's test. Two-tailed, unpaired t-tests were performed on the medians for each genotype with variance pooled across conditions. Medians were also used to test the additivity of effects of daf-2 and daf-16 in a 2-way ANOVA (formula: half-life ∼ loss of daf-16 × loss of daf-2). Plots were generated using ggplot2 in R.
Growth following recovery from starvation
At day 1 and day 6 of starvation, 1 and 1.5 mL aliquots (for day 1 and day 6, respectively) of the starvation culture were placed into 15 mL conical tubes and pelleted at 3,000 rpm for 1 min. L1 worm pellets were plated onto 10 cm NGM plates seeded with a lawn of E. coli OP50 and placed at 20 °C. After 48 h on food, worms were rinsed off the plates using S-basal and placed into 15 mL conical tubes and pelleted at 3,000 rpm for 1 min. Worms were then placed onto a 10 cm plate with no food and allowed to dry for ∼10 min. Images were taken on a Zeiss Discovery V20 stereomicroscope at 20× (day 1) or 40× (day 6) magnification. Worm length was determined using the WormSizer FIJI plugin as previously described (Moore et al. 2013). Four biological replicates were imaged and analyzed. Statistics for growth in control conditions (“1 d” L1 arrest, or ∼12 h after hatching without food) were performed by fitting a 1-factor linear model to the day 1 data for length with the fixed effect being genotype and random effect being replicate using the “nlme” package in R. A 2-factor model was used to assess genotype-dependent effects of starvation on growth, with interaction P-values determined by fitting a mixed-effect linear model with fixed effects for genotype and days of starvation and a random effect of replicate. Plots were generated using ggplot2 in R.
Fecundity following recovery from starvation
Eighteen individual worms for each strain and replicate were randomly selected from the plates used for image analysis of worm length and singled onto a 6 cm NGM plate with E. coli OP50 and kept at 20 °C. For each day of reproduction, individual worms were transferred onto a new plate with food. The number of progeny was counted 2 d after the removal of the parental worm from each plate for each day of reproduction. Total brood size was calculated as the sum of each day of reproduction for each individual worm. One-factor and 2-factor statistics were performed in the same way as they were for worm length, but total brood size was the dependent variable. Plots were generated using ggplot2 in R.
R-scripts for data analysis and reproducibility
All scripts for data analysis can be found at https://github.com/kinseyfish/daf-2_daf-16.
Results
The effects of daf-2/InsR on gene expression during L1 arrest completely depend on daf-16/FoxO
To interrogate the effects of daf-2 and daf-16 during L1 arrest, we performed bulk RNA-seq on starved wild-type, daf-2(e1370), daf-16(mgDf47) (null; (Lee et al. 2001)), and daf-16(mgDf47); daf-2(e1370) mutants. We used the daf-2 reference allele, e1370, a class II loss of function, because null alleles are inviable (Gems et al. 1998). We also included wild-type fed samples to compare the effects of reduced IIS to starvation. Principal component analysis (PCA) of 14,322 detected protein-coding genes (hereafter “genes”) revealed reproducibility of replicates, with samples clearly separating by genotype and condition (Fig. 1a). Most of the variation in gene expression (66.5% variance explained by PC1) was driven by the presence/absence of food, and modulation of IIS by mutation of daf-2 and/or daf-16 during starvation accounted for 15.6% of the variation in gene expression. daf-2 and daf-16 had opposite effects on gene expression, as expected, given that DAF-2 signaling antagonizes DAF-16 activity. Notably, daf-16 and daf-16; daf-2 are indistinguishable by PCA (Fig. 1a), suggesting the transcriptome-wide effects of daf-2 are daf-16-dependent.
We used cluster analysis to look more closely at the effects of daf-2 and daf-16 on gene expression during L1 arrest. We identified 2,094 genes that were differentially expressed (FDR < 0.05) across the 4 genotypes during starvation (Supplementary File 1). We performed hierarchical clustering of these 2,094 genes based on their log_2_ fold-change in each mutant compared to wild type, revealing that the changes in expression seen in the daf-2 mutant are largely reversed by daf-16 mutation. As seen with PCA (Fig. 1a), differential gene expression appears essentially identical between the daf-16 and daf-16; daf-2 mutants (Fig. 1b), further suggesting that the effects of reduced IIS during L1 arrest are daf-16-dependent.
We tallied the number of individual genes differentially expressed between genotypes to more closely examine the effects of daf-2 and daf-16. A total of 10,978 genes were differentially expressed in starved vs. fed wild type, and 3,205 genes were differentially expressed in the daf-2 mutant vs wild type during starvation (FDR < 0.05; Table 1; Supplementary File 1). The pervasive effects of nutrient availability evident here (differential expression of ∼77% of detected genes) are even larger than in previous studies (Baugh et al. 2009; Maxwell et al. 2012; Stadler and Fire 2013), and it has been shown that starvation affects more genes than IIS (Hibshman et al. 2017). The largest effect of genotype on differential expression is in daf-16; daf-2 vs. daf-2 (4,653 genes), revealing the importance of daf-16 in regulating gene expression (see below).
To directly determine whether daf-2 affects gene expression independently of daf-16, we considered daf-16; daf-2 vs. daf-16. When daf-2 is mutated in a daf-16 mutant background, only 34 genes are differentially expressed, indicating that daf-16 is required for nearly all gene expression changes in the daf-2 mutant. However, examination of these 34 genes across each of the comparisons does not provide robust support for the conclusion that they are in fact DAF-16-independent targets of DAF-2. If DAF-2 activates a gene, then we expect it to not be significantly downregulated in starved larvae compared to fed or in daf-2 mutants compared to wild type, and vice versa for genes repressed by DAF-2. However, 17 of the 34 genes display such an unexpected pattern (Supplementary File 1). In addition, if DAF-2 regulates a gene independently of DAF-16, then we expect it to not be affected by loss of daf-16 in other comparisons, such as daf-16 vs. wild type or daf-16; daf-2 vs. daf-2. However, 27 of the 34 genes display such unexpected differential expression. Between these 2 considerations, only 4 genes remain that could be regulated by DAF-2 independently of DAF-16 (nars-2/NARS2, pdcd-2/PDCD, T24H7.2/HYOU1, and lgc-8/HTR3). However, the largest change in expression in response to loss of daf-2 in a daf-16 mutant background for any of these 4 genes is 1.8-fold. These 4 genes may be false positives, or future investigation could reveal that they are in fact regulated by IIS independently of DAF-16. We conclude that the effects of daf-2 on gene expression during L1 arrest are essentially daf-16-dependent.
Identification of an expanded set of DAF-16/FoxO targets during starvation
IIS has received a great deal of attention in C. elegans, and many studies have characterized the effects of daf-2/InsR and daf-16/FoxO on gene expression. Most of these studies identified DAF-16 targets in a daf-2 mutant background, typically in fed adults. A meta-analysis of several studies identified DAF-16 class I and II targets (activated and repressed by DAF-16, respectively) (Tepper et al. 2013). A subsequent study identified DAF-16 targets during L1 arrest in a wild-type background—IIS is low, and DAF-16 is active during starvation, making this possible (Kaplan et al. 2015). Our analysis leveraged the powerful daf-16; daf-2 vs. daf-2 mutant comparison in starved larvae, identifying 4,653 differentially expressed genes (Table 1; Supplementary File 1), constituting an expanded set of putative DAF-16 targets.
As validation, the genes down- and upregulated in daf-16; daf-2 vs. daf-2 overlap significantly with class I and II genes, respectively, as well the DAF-16 target sets identified by Kaplan et al. (Fig. 1c). Gene Ontology (GO) enrichment analysis identified “immune system process” as the most significantly enriched term among upregulated genes and “organic acid metabolic process” for downregulated genes (Supplementary File 1). After plotting non-redundant terms based on semantic similarity, we identified 11 enriched GO terms between the upregulated and downregulated genes (Fig. 1d). These terms suggest that DAF-16/FoxO targets are involved in metabolism (particularly lipid, amino acid, and organic acid metabolism), metabolite transport, immune-like defense responses to other organisms, and homeostasis. These results are consistent with known functions of DAF-16 and IIS (Garsin et al. 2003; McElwee et al. 2003; Murphy et al. 2003; Kim 2013; Murphy and Hu 2013; Hibshman et al. 2017), but we anticipate that the expanded DAF-16 target sets will suggest novel hypotheses about how DAF-16 exerts its potent effects on animal physiology.
We also examined our expanded DAF-16 target sets for enriched expression in specific tissues and cells. Surprisingly, the sites of regulation highlighted by this analysis largely differed between genes up- and downregulated by loss of daf-16 in a daf-2 mutant background (Fig. 1e, Supplementary File 1). The intestine was identified as an enriched tissue for both gene sets, suggesting it is a prominent site of DAF-16 action, but otherwise the results were distinct. The digestive system was prominent among genes upregulated by loss of daf-16, including multiple pharyngeal cell types, and downregulated genes were enriched in a variety of tissues/cell types. These results do not provide the full picture of DAF-16 activity throughout the animal since they include only the most significantly enriched tissues/cell types in this bulk RNA-seq analysis, but they do reflect the anatomical complexity of DAF-16 function.
daf-16/FoxO is completely epistatic to daf-2/InsR for L1 starvation survival
We characterized epistasis between daf-2 and daf-16 for the L1 starvation survival phenotype. daf-2(e1370) worms were long-lived compared to wild type, and daf-16(mgDf47) worms were short-lived (Fig. 2), as expected (Muñoz and Riddle 2003; Baugh and Sternberg 2006). Muñoz and Riddle used a non-null allele of daf-16, and it was only partially epistatic to daf-2(e1370). Baugh and Sternberg used a null allele of daf-16, but they scored a single time point rather than a survival curve, and it was unclear whether daf-16 is fully epistatic to daf-2. We found that the daf-16 and daf-16; daf-2 worms were indistinguishable from each other (P = 0.63) (Fig. 2). A 2-way ANOVA also showed non-additivity of the effects of daf-16 and daf-2 (interaction P = 0.016). Consistent with gene expression analysis (Fig. 1, a and b; Table 1), these results show that daf-16 is epistatic to daf-2 for L1 starvation survival.
*daf-16/FoxO is epistatic to daf-2/InsR for survival during L1 arrest. Proportion alive is plotted for each day of L1 starvation. Points represent samples of 50 to 150 worms from independent replicates. A survival curve was fit by logistic regression to all replicates for plotting. **<0.001, n.s. not significant (pairwise t-test on median survivals of 4 biological replicates).
daf-2/InsR has daf-16-dependent and independent effects on growth with and without extended L1 arrest
We used image analysis to measure worm length after 48 h recovery from L1 arrest, comparing “control” worms that had been arrested briefly for synchronization (1 d, which corresponds to ∼12 h after hatching in the absence of food given ∼12 h to complete embryogenesis after hypochlorite treatment) or subjected to extended starvation (6 d). daf-2(e1370) control worms were significantly smaller than wild type, and daf-16(mgDf47) control worms were indistinguishable from wild type (Fig. 3a). Double-mutant control worms were substantially longer than daf-2 but significantly shorter than daf-16 and wild-type control worms, revealing incomplete suppression. These results suggest that the effects of daf-2 on growth in control conditions are largely but not entirely dependent on daf-16.
*daf-2/InsR has daf-16-dependent and independent effects on growth and reproduction after extended L1 arrest. a) Length following 48 h of recovery from 1 or 6 d L1 starvation is plotted. b) Total brood size following 1 or 6 d L1 starvation is plotted. a and b) Points represent individual worms from 3 biological replicates. The red lines connect the mean values from 1 to 6 d of starvation, plotting the reaction norm for extended starvation for each genotype. Δ indicates the difference in means between day 1 and day 6 within a genotype. The black asterisks depict statistical significance for the difference between genotypes in the control condition (1 d L1 starvation; pairwise comparisons from a 1-factor linear mixed-effect model), and red pound signs depict statistical significance for the difference in reaction norms between genotypes (interaction P-value from a 2-factor linear mixed-effect model). **, ###<0.001; #<0.05; n.s. not significant.
We used a 2-factor model to investigate the starvation-dependent effects of daf-2 and daf-16 on growth. The interaction P-value from this model indicates whether a difference in the reaction norm between 1 and 6 d of starvation for a pair of genotypes is significant. daf-2 mutants displayed robust starvation resistance, with a substantially smaller reaction norm than wild type (Fig. 3a). In contrast, daf-16 mutants displayed sensitivity, with a significantly larger reaction norm than wild type, as expected (Webster et al. 2022a). Mutation of daf-16 in the daf-2 background rescued the reaction norm, with the double mutant's reaction norm being indistinguishable from wild type, indicating epistasis. However, the double mutant's reaction norm is significantly smaller than daf-16, suggesting incomplete epistasis. These results suggest that the effects of daf-2 on growth after extended L1 starvation are largely but not entirely dependent on daf-16.
daf-16 does not contribute to reproductive success following extended L1 arrest
We scored total brood size as a measure of reproductive success after recovery from L1 arrest, comparing “control” worms that had been arrested briefly for synchronization (1 d) or subjected to extended starvation (6 d). daf-2 control worms produced significantly fewer progeny than wild type, and daf-16 control worms were indistinguishable from wild type (Fig. 3b). Loss of daf-16 in the daf-2 background mostly suppressed the reduction in brood size in control worms, but double-mutant control worms produced significantly fewer progeny than wild-type control worms. These results suggest the effect of daf-2 on reproductive success in control conditions depends on daf-16, but not entirely.
Strikingly, daf-16 did not affect reproductive success following extended L1 arrest. As for growth (Fig. 3a), we used the interaction P-value from a 2-factor model to assess the significance of the starvation-dependent effects between genotypes. daf-2 mutants were again remarkably resistant to starvation, with a significantly smaller reaction norm than wild type (Fig. 3b). The daf-16 reaction norm was indistinguishable from wild type, suggesting that daf-16 does not support reproductive success following extended starvation. Furthermore, the double-mutant reaction norm was small and indistinguishable from the daf-2 reaction norm, suggesting a lack of epistasis in this 1 case. Furthermore, the double-mutant reaction norm was significantly smaller than the reaction norms for wild type and daf-16. These results show that daf-16 is dispensable for reproductive success following extended L1 arrest.
Discussion
We provide a valuable RNA-seq dataset for the investigation of IIS in mediating nutrient-dependent gene regulation. Since this dataset was generated from arrested L1 larvae, it is not confounded by differences in developmental stage or secondary consequences of altered behavior. Moreover, by analyzing a daf-2/InsR mutant during starvation, we identified 4,653 putative DAF-16/FoxO target genes, many of which had not been identified.
We performed epistasis analysis between daf-2/InsR and daf-16/FoxO, focusing on L1 arrest and recovery, with the objective of determining whether DAF-16 is the sole effector of IIS. Based on RNA-seq and starvation survival during arrest, daf-16 is epistatic to daf-2 (Figs. 1 and 2), suggesting DAF-16 is the sole effector of IIS in regulating nutrient-dependent gene regulation and survival. mTOR signaling promotes anabolic metabolism and is activated by PI3K/AKT signaling (Zoncu et al. 2011), suggesting mTOR could mediate the effects of DAF-2. DAF-18/PTEN antagonizes PI3K signaling (Murphy and Hu 2013), it is required for cell-cycle arrest in starved L1 larvae (Fukuyama et al. 2006), and aberrant divisions of the primordial germ cells (PGCs) depend on the TORC1 mTOR complex (Fukuyama et al. 2012), further suggesting that mTOR could function as an effector of IIS to affect gene expression or survival. However, our results suggest otherwise, consistent with mTOR signaling being dispensable for PGC transcriptional quiescence during L1 starvation (Fry et al. 2021). SKN-1/Nrf is another potential effector of IIS. In addition to promoting adult longevity downstream of DAF-2 and in parallel to DAF-16 (Tullet et al. 2008), SKN-1 contributes to the starvation response and promotes L1 starvation survival (Paek et al. 2012). However, our results suggest that such regulation occurs independently of IIS.
daf -2/InsR* promotes growth with and without extended L1 arrest (Fig. 3a). These effects largely depend on daf-16/FoxO, although not entirely. These results suggest DAF-16 is the major effector of IIS-mediated growth control but that 1 or more additional effectors contribute. The observation that the effects of daf-2 on L1 starvation survival and gene expression during L1 arrest are daf-16-dependent but that the effects of daf-2 on growth appear to not entirely depend on daf-16 suggests that the time of action for daf-2 may differ for the phenotypes, with starvation survival and gene expression depending on daf-2 function during L1 arrest and growth depending on daf-2 function during L1 arrest and recovery.
DAF-16/FoxO does not affect reproduction following extended L1 arrest. daf-2/InsR promotes reproduction with and without extended L1 arrest (Fig. 3b). daf-16 largely but not entirely mediates the effect without extended starvation; that is, in worms that were only briefly starved (∼12 h; “control”). This result suggests, as for growth control, that DAF-16 is the major effector of IIS-mediated effects on reproduction without extended L1 arrest but that 1 or more additional effectors contribute. In contrast, loss of daf-16 in wild-type or daf-2 mutant backgrounds had no effect on the impact of extended L1 arrest on reproduction (Fig. 3b). These results suggest that the preservation of reproductive success following extended starvation promoted by reduced IIS depends on an effector other than DAF-16. Furthermore, these results suggest that the site of action for daf-2 differs for starvation survival and gene expression, which are *daf-16-*dependent, compared to reproduction. Notably, daf-16 is not required for arrest of PGC divisions during L1 starvation though it is required for somatic cell arrest (Baugh and Sternberg 2006; Fukuyama et al. 2006), and PGC function is dispensable for L1 starvation survival (Webster et al. 2022b), supporting the possibility that DAF-2 functions independently of DAF-16 in the PGCs during L1 arrest to preserve reproductive potential. Identification of the IIS effector(s) in this context will improve understanding of how animals maintain fitness following early life starvation.
Supplementary Material
jkaf309_Supplementary_Data
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Angeles-Albores D, N Lee RY, Chan J, Sternberg PW. 2016. Tissue enrichment analysis for C. elegans genomics. BMC Bioinformatics. 17:366. 10.1186/s 12859-016-1229-9.27618863 PMC 5020436 · doi ↗ · pubmed ↗
- 2Ashburner M et al 2000. Gene ontology: tool for the unification of biology. The gene ontology consortium. Nat Genet. 25:25–29. 10.1038/75556.10802651 PMC 3037419 · doi ↗ · pubmed ↗
- 3Baugh LR . 2013. To grow or not to grow: nutritional control of development during Caenorhabditis elegans L 1 arrest. Genetics. 194:539–555. 10.1534/genetics.113.150847.23824969 PMC 3697962 · doi ↗ · pubmed ↗
- 4Baugh LR, Demodena J, Sternberg PW. 2009. RNA Pol II accumulates at promoters of growth genes during developmental arrest. Science. 324:92–94. 10.1126/science.1169628.19251593 · doi ↗ · pubmed ↗
- 5Baugh LR, Hu PJ. 2020. Starvation responses throughout the Caenorhabditis elegans life cycle. Genetics. 216:837–878. 10.1534/genetics.120.303565.33268389 PMC 7768255 · doi ↗ · pubmed ↗
- 6Baugh LR, Sternberg PW. 2006. DAF-16/FOXO regulates transcription of cki-1/cip/kip and repression of lin-4 during C. elegans L 1 arrest. Curr Biol. 16:780–785. 10.1016/j.cub.2006.03.021.16631585 · doi ↗ · pubmed ↗
- 7Chen Y, Chen L, Lun Aaron TL, Baldoni Pedro L, Smyth Gordon K. 2025. Edger v 4: powerful differential analysis of sequencing data with expanded functionality and improved support for small counts and larger datasets. Nucleic Acids Res. 53:gkaf 018. 10.1093/nar/gkaf 018.39844453 PMC 11754124 · doi ↗ · pubmed ↗
- 8Fry AL et al 2021. DAF-18/PTEN inhibits germline zygotic gene activation during primordial germ cell quiescence. P Lo S Genet. 17:e 1009650. 10.1371/journal.pgen.1009650.34288923 PMC 8294487 · doi ↗ · pubmed ↗