Physiological Starvation Promotes Caenorhabditis elegans Vulval Induction

Studying how molecular pathways respond to ecologically relevant environmental variation is fundamental to understand organismal development and its evolution. Here we characterize how starvation modulates Caenorhabditis elegans vulval cell fate patterning – an environmentally sensitive process, with a nevertheless robust output. Past research has shown many vulval mutants affecting EGF-Ras-MAPK, Delta-Notch and Wnt pathways to be suppressed by environmental factors, such as starvation. Here we aimed to resolve previous, seemingly contradictory, observations on how starvation modulates levels of vulval induction. Using the strong starvation suppression of the Vulvaless phenotype of lin-3/egf reduction-of-function mutations as an experimental paradigm, we first tested for a possible involvement of the sensory system in relaying starvation signals to affect vulval induction: mutation of various sensory inputs, DAF-2/Insulin or DAF-7/TGF-β signaling did not abolish lin-3(rf) starvation suppression. In contrast, nutrient deprivation induced by mutation of the intestinal peptide transporter gene pept-1 or the TOR pathway component rsks-1 (the ortholog of mammalian P70S6K) very strongly suppressed lin-3(rf) mutant phenotypes. Therefore, physiologically starved animals induced by these mutations tightly recapitulated the effects of external starvation on vulval induction. While both starvation and pept-1 RNAi were sufficient to increase Ras and Notch pathway activities in vulval cells, the highly penetrant Vulvaless phenotype of a tissue-specific null allele of lin-3 was not suppressed by either condition. This and additional results indicate that partial lin-3 expression is required for starvation to affect vulval induction. These results suggest a cross-talk between nutrient deprivation, TOR-S6K and EGF-Ras-MAPK signaling during C. elegans vulval induction.


LIN-3 mutational penetrance TOR-S6K vulval cell fate patterning
Organismal development is inherently sensitive to environmental variation, and specific environmental conditions, such as nutrient availability or temperature, may also reflect instructive cues controlling growth and other critical developmental decisions, such as developmental timing, diapause entry and production of alternative phenotypes. Developmental integration of environmental cues has become increasingly understood, primarily in the context of nutritional and metabolic regulation of developmental decision, growth and lifespan, such as in Drosophila and C. elegans (Flatt et al. 2013). In contrast to the understanding of such instructive environmental cues in development, very little is known about how environmental variation impacts other, highly diversified developmental processes, seemingly unaffected by the environment as they maintain their function and corresponding phenotypic outputs when facing such variation. However, such stability of the phenotypic output may go in hand with an underlying flexibility of developmental mechanisms in changing environmental (and genetic) conditions (Félix and Barkoulas 2015;True and Haag 2001;Paaby and Rockman 2014). C. elegans vulval development provides one clear example of a developmental process, which robustly generates an invariant phenotypic output (cell fate pattern) although activities and interactions of underlying molecular signaling pathways are environmentally sensitive Grimbert and Braendle 2014). C. elegans vulval cell fate patterning involves a network of conserved signaling pathways, EGF-Ras-MAPK, Delta-Notch and Wnt pathways that reliably establish a stereotypical cell fate pattern of hypodermal vulval precursor cells (VPCs) (Félix 2012;Sternberg 2005) (Figures 1A and 1B).
The VPCs represent a subset of Pn.p cells, P3.p to P8.p, competent to adopt vulval cell fates. Competence of vulval precursor cells is established during L1-L2 stages through expression of the Hox gene lin-39 and Wnt signals from the posterior (Pénigault and Félix 2011;Salser et al. 1993). The key events of the vulval patterning process take place from mid L2 to early L3 stage and involve intercellular signaling between the gonadal anchor cell (AC) and the VPCs (Figures 1A and 1B). In brief, LIN-3/EGF ligand released from the AC induces the primary (1°) vulval cell fate by activating the EGF-Ras-MAPK pathway in P6.p, which receives the highest dose of this signal (Hill and Sternberg 1992;Grimbert et al. 2016). EGF-Ras-MAPK activation induces production of a lateral signal via the Delta-Notch pathway, promoting 2°and inhibiting 1°cell fate in the neighboring cells, P5.p and P7.p (Greenwald et al. 1983;Sternberg and Horvitz 1986;Berset et al. 2001;Yoo et al. 2004). Moreover, a switch from the canonical LET-60/Ras-LIN-45/Raf pathway to a LET-60/Ras-RGL-1-RAL-1/RAL signaling pathway can promote the 2°cell fate in P5.p and P7.p (Zand et al. 2011). The remaining three VPCs, although competent, adopt nonvulval cell fates (3°for P4.p and P8.p, and 3°or 4°for P3.p) as they do not receive sufficient doses of either EGF-Ras-MAPK or Delta-Notch signal. In addition to EGF-Ras-MAPK and Delta-Notch pathways, the canonical Wnt signaling pathway may also be involved in vulval induction: overactivation of the Wnt pathway, e.g., through pry-1/Axin mutation, increases vulval induction of diverse hypoinduced mutants Gleason et al. 2002;Seetharaman et al. 2010). C. elegans vulval development thus involves a regulatory network of three key molecular cascades and their cross-talk contributes to a reliable and precise patterning output in the presence of both genetic and environmental perturbations (Braendle et al. 2010;Félix and Barkoulas 2012;Gleason et al. 2002;Hoyos et al. 2011;Milloz et al. 2008).
Despite the apparent robustness of C. elegans vulval cell fate patterning to environmental perturbations, multiple reports demonstrate that this process is nevertheless responsive to environmental and physiological inputs (Braendle and Felix 2009;Félix 2012;Félix and Barkoulas 2012;Grimbert & Braendle 2014;Sternberg 2005) (summarized in Figure 1C). In early mutagenesis screens, various Vulvaless mutations (lin-2, lin-3, lin-7, lin-24, lin-33 and let-23) were found to be suppressed by starvation and/or dauer passage (Ferguson and Horvitz 1985), suggesting that vulval inductive pathways, such as EGF-Ras-MAPK, are environmentally sensitive. Detailed studies later confirmed that specific chemical elements, such as zinc (Bruinsma et al. 2002;Yoder et al. 2004), and growth conditions indeed modulate vulval inductive signaling. Moghal et al. (2003) reported that liquid culture increases vulval inductive signals via the neuronally expressed heterotrimeric Gaq protein, EGL-30. The EGL-30/Gaq signal is transduced via the voltage-gated calcium channel, EGL-19, in muscle cells and its positive effect on vulval induction is mediated by Wnt signaling via BAR-1/b-Catenin (Moghal et al. 2003). Battu et al. (2003) further found evidence for a starvation signal that negatively affects vulval induction via the sensory system. In this study, starvation conditions were found to suppress vulval hyperinduction, for example, induced by let-60/Ras(gf). This negative starvation effect requires the G-protein-coupled receptor SRA-13 to modulate EGF-Ras-MAPK activity (Battu et al. 2003). Consistent with a sensory-mediated nutritional modulation of vulval signaling, compromised DAF-2/Insulin signaling recapitulates the results that starvation negatively affects vulval induction (Battu et al. 2003;Nakdimon et al. 2012). In addition, another study (Nakdimon et al. 2012) indicates that DAF-18/PTEN negatively regulates EGF-Ras-MAPK activity during vulval induction, further supporting the view that multiple key metabolic and sensory pathways interact with vulval signaling pathways.  examined the effects of diverse environmental conditions (different temperatures, starvation, liquid culture, dauer passage) on vulval induction using a diverse set of known mutations in EGF-Ras-MAPK, Delta-Notch and Wnt pathways. A majority of mutations showed variable penetrance depending on the environment, with starvation and dauer environments showing the most pronounced effects . Consistent with the results of Ferguson & Horvitz (1985), yet seemingly contrary to the results obtained by Battu et al. (2003),  found that starvation increased vulval inductive levels, leading to drastic suppression of the Vulvaless phenotypes caused by lin-3(rf) and let-23/egfr(rf) mutations. Genetic analyses reported in this study suggest that this starvation signal acts at the level or upstream of LET-23/EGFR . The positive effect of starvation on vulval induction was further confirmed by quantification of pathway activities in starved vs. fed wild type animals: starved animals showed a significantly higher EGF-Ras-MAPK pathway activity in the 1°cell, P6.p, whereas Delta-Notch activity was higher in 2°cells, P5.p and P7.p . In addition, the same study found that starvation potentially modulates vulval induction via the Wnt pathway as vulval induction was strongly compromised in starved bar-1/b-Catenin(0) animals.
In summary, starvation experienced during larval growth may have both positive Ferguson and Horvitz 1985) and negative (Battu et al. 2003) effects on vulval induction. While negative starvation effects have been shown to be mediated by the sensory system (Battu et al. 2003), it remains unknown how positive starvation effects are perceived and transduced to modulate vulval induction. In addition, how such, apparently antagonistically acting, starvation signals interact during vulval induction is also unknown.
Addressing these open questions, we have aimed here to quantify and characterize more precisely how starvation affects C. elegans vulval cell fate patterning. In particular, we ask how sensory perception of external starvation conditions vs. the (internal) animal's nutrient status affect vulval induction.

Strains and maintenance
Strains were maintained on NGM agar plates (55mm petri dishes, 1.7% agar) carrying a lawn of E. coli OP50 (Wood 1988;Brenner 1974). Animals were grown at 20°unless indicated otherwise and both mutant and wild type strains were freshly thawed prior to experiments. Our analysis included the C. elegans N2 wild type reference strain and the mutants listed below, all of which had been previously isolated and described.

Scoring of vulval phenotypes
The vulval phenotype was observed using Nomarski optics in early to mid L4 individuals, anesthetized with sodium azide (Wood 1988). We counted the Pn.p progeny and determined their fates as previously described Sternberg and Horvitz 1986).

Experimental environments
Experimental populations were age-synchronized by hypochlorite treatment and liquid arrest (27 hr) at the beginning of experiments. Individuals examined in different environments or of different strains were always scored in parallel and derived from populations kept in identical environmental conditions over at least two generations.
For starvation assays, L1 larvae were grown on standard NGM plates until they reached the mid L2 stage (23 hr after L1 transfer back on NGM seeded plates) unless mentioned otherwise. At this stage, animals were washed three times with sterile M9 buffer and transferred on starvation plates, i.e., unseeded NGM plates containing 1mg/ml of ampicillin to prevent bacterial growth. After 48 hr, starved animals were transferred back to regular NGM plates seeded with E. coli OP50 and the vulval phenotype was scored when animals had reached the early or mid L4 stage (approximately 15-20 hr later). Control animals were kept on NGM plates seeded with E. coli OP50 from L1 to L4. This starvation treatment drastically reduced, yet did not completely stop, developmental progression of worms: after 48 hr most animals had developed into the early to mid-L3 individuals. Note that this starvation treatment did not induce dauer formation.

RNAi experiments
RNAi by bacterial feeding was performed as described by Timmons et al. (2001). The HT115 bacterial strain carrying the empty RNAi expression vector L4440 served as a negative control. We used RNAi plates composed of standard NGM with 50mg/ml of ampicillin and 1mM of IPTG. Late L2/ early L3 individuals were transferred on RNAi bacteria. The vulval phenotype was scored in L4 individuals of the F1 generation. RNAi clones were selected from Ahringer's RNAi library (Kamath et al. 2003) or from Vidal's RNAi library (Rual et al. 2004). The following RNAi clones Figure 1 Progression and environmental sensitivity of C. elegans vulval cell fate patterning. (A) A subset of six Pn.p cells (P3.p to P8.p) acquires competence to form vulval tissue through the expression of LIN-39; however, usually only P5.p, P6.p and P7.p adopt vulval cell fates in a stereotypical 2°-1°-2°pattern. The remaining competent cells, P4.p and P8.p, adopt a non-vulval 3°fate, while P3.p may either adopt a 3°fate or a 4°fate, also referred to as F(fused) fate. The induction of vulval cell fates is generated by the cross-talk of EGF-Ras-MAPK and Delta-Notch signaling pathways. Distinct vulval cell divisions according to cell fate occur during the L3 stage and generate 22 vulval cells. (B) The anchor cell (AC) releases the LIN-3 inductive signal. LIN-3 acts as a morphogen, with P6.p receiving the highest level of LIN-3 causing it to adopt a 1°Cell fate (blue). The expression of the Delta ligands in P6.p activates the Delta-Notch pathway in its neighbors, P5.p and P7.p. This activation causes them to adopt a 2°fate (red) and represses the primary fate. A switch from the canonical LET-60/Ras-LIN-45/Raf pathway to a LET-60/Ras-RGL-1-RAL-1 signaling pathway can also promote the 2°Cell fate in P5.p and P7.p (not detailed). (C) Schematic representation of previously reported environmental and physiological inputs affecting C. elegans vulval induction (for details, see introduction). For effects of high temperature on vulval induction, note that  found that 25°C culture had a positive effect on vulval induction (e.g., suppression of diverse Vul mutants) whereas Grimbert and Braendle (2014) found that 30°C consistently caused hypoinduced wild type animals.
We quantified reporter gene activity in VPCs as previously described Braendle et al. 2010). In brief, CFP or GFP signal quantification was performed when individuals had reached the stage of lethargus L2/L3 or early L3. Pn.p cells of live, anesthetized individuals were first identified using DIC imaging, followed by measurement of pixel signal intensity in P5.p, P6.p and P7.p for each individual. Images were acquired using an Olympus BX61 microscope at 40X magnification, equipped with a Coolsnap HQ2 camera. To quantify signal intensity of each cell, we first selected a fixed sub-region of the same size within nuclei of target VPCs and then measured the mean signal intensity of this region. After background subtraction, we used the mean signal intensity as a measure of the corresponding signaling pathway activity in each Pn.p.
Single Molecule FISH L3 stage animals were processed to quantify lin-3/egf mRNA levels by smFISH as described before (Barkoulas et al. 2013). In brief, they were permeabilized and hybridized with custom-made probes that bind specifically to lin-3/egf and lag-2 mRNAs, labeled with Cy5 and Alexa-594, respectively. The animals were stained with DAPI and placed in GLOX buffer for image acquisition. For each animal, we acquired Z-stacks with a Pixis 1024B camera (Princeton Instruments) mounted onto an upright Zeiss AxioImager M1 with a Lumen 200 metal arc lamp (Prior Scientific). Each Z-stack was composed of 30 images spanning 20.3 mm (0.7 mm in between two acquisitions) on each filter channel. Finally, spot quantification was performed using a custom-made Matlab routine optimized for the anchor cell. Only worms with gonad length .50 mm were included in the quantifications.

Statistical Analyses
Data were transformed (e.g., Box-Cox-or log-transformed) when necessary to meet the assumptions of ANOVA procedures (homogeneity of variances and normal distributions of residuals) (Sokal and Rohlf 1981). For post hoc comparisons, Tukey's honestly significant difference (HSD) procedure was used. Statistical tests were performed using the software programs JMP v13 or SPSS v19.

Data availability
All strains generated in this study are available upon request. The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article.

RESULTS AND DISCUSSION
Quantifying starvation effects on vulval induction in lin-3/egf(rf) mutants Previous studies have used different protocols to examine starvation effects on vulval induction, e.g., animals were exposed to starvation conditions for different periods of time and at different developmental time points (Battu et al. 2003;Ferguson and Horvitz 1985). We therefore first aimed to determine the starvationsensitive periods of vulval induction using lin-3(rf) mutants known to be affected by starvation conditions Ferguson and Horvitz 1985). Consistent with these earlier studies, a 48h starvation exposure consistently increased vulval induction from late L1/early L2 to early L3 in lin-3(n378) animals ( Figure 2A). Starvation sensitivity thus extends over the entire period of the vulval cell fate patterning process, with mid-L2 individuals being most sensitive. Next, we asked how the duration of starvation exposure modulates vulval induction. Exposing lin-3(n378) animals to different time periods of starvation (0-120h) at the mid L2 stage caused strong suppression of the Vulvaless phenotype when starved for 48 hr or more, low suppression after 19-24h of starvation, and no suppression after a brief (2h) exposure to starvation or early L1 starvation (12h) in liquid medium ( Figure 2B). Therefore, the extent of vulval induction depends on the duration of the starvation exposure during specific larval developmental stages.
To characterize how starvation of lin-3(rf) mutants affects vulval cell lineages, we further scored cell fate patterns in mid-L4 individuals that had been exposed to 48 hr of starvation at the mid-L2 stage. The Vulvaless phenotypes of lin-3(n378) and lin-3(e1417) were strongly suppressed by starvation ( Figures 2C and 2F), similar to previous results . Starvation resulted in an increased mean number of induced vulval cells and a higher proportion of individuals with a canonical 2°-1°-2°cell fate pattern (P5.p to P7.p), and these suppression effects were consistently stronger in lin-3(n378) compared to lin-3(e1417) ( Figures 2C-2H). These allelic differences in observed starvation effects suggest that vulval hypoinduction due to the n378 missense mutation is more easily suppressed by starvation than transcriptional reduction of lin-3 in the anchor cell caused by the e1417 allele. In lin-3(n378), the proportion of individuals with complete induction increased from 12% in food conditions to 50% after starvation, and most of these individuals showed a correct 2°-1°-2°pattern for P5.p to P7.p ( Figures 2D and 2E). Together, these results show that starvation can partly compensate for reduced LIN-3 signaling, confirming the previously reported positive effect of starvation on vulval induction .
We hypothesized that the prolonged developmental time in the starvation environment could affect the output of the signaling network. Applied starvation treatments caused a strong delay in development: animals remained in the L2 stage throughout the 48-hr starvation period (the generation time under normal conditions at 20°is 3.5 days). In addition, we observed that individuals starved for only short time period (2 hr) showed no suppression while longer starvation periods (19-24 hr vs. 48-120 hr) had different effects, with prolonged starvation periods causing increased suppression ( Figure 2B). If the vulval signaling network does not operate at steady state, such developmental delays may allow for accumulation of active downstream effectors inducing vulval fates. Previous experiments, using hydroxyurea treatment or anchor cell ablation, supported this hypothesis (Canevascini et al. 2005;Félix 2007;Wang and Sternberg 1999). However, a mutation in clk-1 (a gene necessary for the synthesis of ubiquinone) that causes a twofold increase in developmental time (EWBANK et al. 1997) did not increase vulval induction in lin-3(n378) ( Figure 2I). Thus, a prolonged developmental time alone may not be sufficient to affect vulval induction. However, this result should be interpreted with caution given that clk-1 has been reported to negatively affect vulval induction (Shibata et al. 2003), consistent with our observation that clk-1(e2519) reduces vulval induction of lin-3(e1417) in food conditions (2I).

Compromised sensory signaling does not abolish starvation effects on vulval induction
To understand how starvation stimuli are transduced to affect vulval induction, we first tested for an implication of the sensory system in relaying external starvation conditions. Battu et al. (2003) reported that vulval hyperinduction caused by let-60/ras(gf) was significantly reduced after starvation of late L1 larvae, indicative of a negative starvation effect on vulval induction. Moreover, this study showed that mutation of the G-protein-coupled receptor sra-13 or mutations causing sensory signaling defects, such as osm-5(p813) and che-3(e1124), abolished starvation sensitivity of let-60(gf) animals, implying that observed starvation effects acted on vulval induction via components of the sensory system. We therefore tested whether disruption of the same sensory signaling components also inhibits starvation effects on vulval induction in lin-3(rf) mutants. However, we observed that starvation still significantly increased vulval induction of lin-3(n378) in sra-13 (zh13), osm-5(p813) and che-3(e1124) backgrounds, comparable to the results of the lin-3(n378) single mutant (Table 1). Therefore, in contrast to the results of Battu et al. (2003), these sensory mutants did not abolish starvation suppression of the Vulvaless phenotype in lin-3(n378) mutants.
We next tested for an implication of TGF-b and Insulin signaling: two key signaling pathways involved in sensory transduction of environmental stimuli, which mediate diverse developmental and metabolic responses to changes in nutritional status, e.g., during dauer formation (Fielenbach and Antebi 2008) or germline proliferation (Dalfó et al. 2012;Michaelson et al. 2010). The activity of the TGF-b and insulin pathways are reduced upon starvation. Reducing TGF-b pathway activity did not mimic starvation effects on vulval induction in Figure 2 Quantifying starvation effects on vulval induction in lin-3/egf(rf) mutants. (A) Differences in vulval induction of lin-3(n378) animals exposed to starvation (48 hr) at different developmental stages. Starvation suppression of lin-3 (n378) differed significantly between exposed developmental stages (ANOVA, F 5,239 = 14.02, P , 0.0001), with strongest effects observed from mid L2 to early L3 stages. (B) Effects of starvation duration on vulval induction of lin-3(n378). Animals (age-synchronized by egg-laying windows) were exposed to starvation at the mid L2 stage for different time periods except for the first treatment where animals were starved in liquid (M9 buffer) for 12 hr directly after hatching (N = 80 per treatment). Starvation duration significantly affected levels of vulval induction (ANOVA, F 5,239 = 14.02, P , 0.0001) with strongest suppression of lin-3(n378) at 48 to 120 hr of starvation. (C) Starvation (48 hr food conditions as shown by the double mutant daf-7(e1372); lin-3 (e1417); and starvation still significantly increased vulval induction in this double mutant (Table 1). Similarly, reduced activity of the DAF-2/ Insulin receptor caused by the daf-2(e1370) mutation did not mimic starvation effects on lin-3(n378) mutants in food conditions (Table 1). However, vulval induction in daf-2(e1370) lin-3(n378) double mutants was significantly lower than in the single lin-3(n378) mutant, both in food and starvation conditions. This result is consistent with previous observations that reduced Insulin signaling may lower vulval induction in sensitized backgrounds, such as let-60/Ras(gf) (Battu et al. 2003;Nakdimon et al. 2012). To confirm that TGF-b and Insulin signaling are unlikely to mediate positive starvation effects on vulval induction, we further examined the role of a central genetic component, the DAF-12 steroid receptor, integrating downstream effects of these two signals (Fielenbach and Antebi 2008) using the Daf-d mutant, daf-12(rh61rh411) (Antebi et al. 2000). Again, as for daf-2 and daf-7, this mutant did not mimic starvation effects on lin-3(n378) vulval induction (Table 1).
Taken together, these results indicate that Insulin and TGF-b pathways, and thus by implication, environmental signal transduction via these pathways, do not seem to play major roles in mediating positive starvation effects on vulval induction. However, we confirm previous observations (Battu et al. 2003;Nakdimon et al. 2012) that reduced DAF-2 Insulin activity may lower vulval induction in sensitized backgrounds, including lin-3(rf) mutants, in both food and starvation conditions. Yet, in our experiments, starvation nevertheless increased vulval induction of daf-2(e1370); lin-3(n378) animals, suggesting that the negative starvation effect (Battu et al. 2003;Nakdimon et al. 2012) on vulval induction is significantly weaker than the positive starvation signal. Therefore, while Insulin signaling and sensory perception seem to mediate a mild negative starvation signal (Battu et al. 2003;Nakdimon et al. 2012), we here provide evidence for an additional positive starvation signal acting independently of Insulin and sensory signaling. Moreover, starvation thus appears to induce multiple antagonistic effects on C. elegans vulval induction. Nutrient-deprivation caused by rsks-1 and pept-1 mutations strongly increase vulval induction in lin-3(rf) To test whether observed starvation effects on vulval induction might be mediated by internal perception of the organism's nutritional status, we focused on a conserved key regulator of cell growth and proliferation via nutrient sensing, the LET-363/TOR (target of rapamycin) pathway (Hietakangas and Cohen 2009;Jia et al. 2004). RNAi knock-down of let-363 and rsks-1/S6K (encoding the C. elegans ortholog of mammalian P70S6K, a known TORC1 target) (Laplante and Sabatini 2012), significantly increased vulval induction of lin-3(n378) animals in food conditions ( Figure 3A). Moreover, RNAi knockdown of pept-1 (encoding an intestinal oligopeptide transporter) (Fei et al. 1998), causing nutrient-deprived animals through amino acid starvation (Meissner et al. 2004), increased vulval induction of lin-3(n378) even more strongly ( Figure 3A). These RNAi effects thus mimiced starvation effects on lin-3(rf), consistent with the idea that TOR pathway components and targets mediate starvation effects on vulval induction. We further tested whether RNAi knock-down of an additional TOR pathway member daf-15 (RAPTOR) affects vulval induction using the lin-3(n378) mutant. RNAi of daf-15 did not cause any suppression of the Vulvaless phenotype of lin-3(n378) ( Figure 3A).
Given that reduced amino acid availability caused by pept-1(lg601) primarily acts through TOR signaling (Meissner et al. 2004;Korta et al. 2012) and that RNAi of rsks-1 similarly mimics starvation effects, we conclude that the observed positive starvation modulation of vulval induction likely involves TOR-S6K signaling.
n Table 1 Effects of starvation treatment on vulval induction level of different mutants of the sensory system. Statistical tests (One-Way-ANOVAs) were performed for each mutant separately testing for the effect of treatment (food vs. starvation) on the number of vulval cells induced ( Ã P < 0.05, ÃÃ P < 0.01, ÃÃÃ P < 0.001, ns: nonsignificant). To facilitate comparison, measurements for lin-3(e1417) and lin-3(n378) are also included (i.e., same data as in Figures  In both RNAi and mutant experiments, disruption of PEPT-1 activity caused a stronger increase in lin-3(n378) vulval induction compared to disruption of RSKS-1 activity ( Figures 3B and 3C). Therefore, nutrient deprivation caused by reduced PEPT-1 activity did not precisely recapitulate the effects of the rsks-1 mutant, which could indicate that factors other than rsks-1 mediate effects of nutrient deprivation on vulval induction. Consistent with this idea, we found that rsks-1 (ok1255); lin-3(n378) double mutants exposed to either pept-1 RNAi or starvation exhibited a slight, but significant increase in vulval induction compared to control animals ( Figure 3G). Positive effects of nutrient deprivation on vulval induction may thus involve additional factors other than TOR-S6K signaling.
Overall, nutrient deprivationcaused by external starvation through absence of food or by internal, physiological starvation due to pept-1 mutation or RNAihave very similar effects, generating an overall net increase in vulval induction.

pept-1 RNAi increases Ras and Notch pathway activities in vulval precursor cells
To address how physiological starvation caused by pept-1 RNAi affects the vulval signaling network, we focused on the main inductive pathway, the Ras pathway. Starvation conditions have previously been shown to increase EGF-Ras-MAPK activity in the 1°cell, P6.p, and Delta-Notch activity in the 2°Cells, P5.p and P7.p of the wild type strain N2 . We therefore tested whether pept-1 RNAi mimicked these starvation effects using the same fluorescent reporters, egl-17::cfp to quantify Ras pathway activity (Yoo et al. 2004) and lip-1::gfp to quantify Delta-Notch activity (Berset et al. 2001). pept-1 RNAi effects closely mirrored previously quantified starvation effects on reporter gene activities , with pept-1 RNAi increasing egl-17 expression in P6.p and increasing lip-1 expression in P5.p (but not P7.p) (Figures 4A and 4B; Table 2). Consistent with these changes, we also observed that pept-1 RNAi reduced lip-1 expression in P6.p while reducing egl-17 expression in P5.p and P7.p (Figures 4A and 4B; Table 2). pept-1 RNAi thus tightly recapitulated previously observed starvation effects on activities of the two key inductive vulval signaling pathways . Observed changes in pathway activities may be explained by distinct effects on Ras and Notch pathways, or by effects mediated solely by the Ras pathway, given the cross-talk between the two pathways (Yoo et al. 2004).

Starvation sensitivity of vulval induction requires minimal EGF pathway activity
To further characterize how physiological starvation interacts with vulval induction to modulate inductive levels, we characterized effects of pept-1 RNAi on several components of the core EGF-Ras-MAPK cascade using various reduction-of-function mutants (Table 3). In addition to lin-3(rf) mutations, pept-1 RNAi suppressed, to different degrees, the Vulvaless phenotypes of let-23(sy1), sem-5(n2019) and mpk-1(ku1) ( Table 3). Although these results suggest that nutrient deprivation due to pept-1 RNAi may act downstream of MAPK, interpretation of these effects is limited given that these mutations do not represent null alleles, most of which are homozygous lethals. We therefore used a double mutant let-23(sy1); lin-3(n378) that exhibits a fully penetrant Vulvaless phenotype, indicating absence or very low levels of basal Ras activity . Vulval induction of let-23(sy1); lin-3(n378) animals treated with pept-1 RNAi remained virtually unaltered compared to controls (Table 3), with only 4/60 individuals showing some induced vulval cells (vs. 0/45 individuals in food conditions). Very similarly, a recently constructed tissue-specific null allele of lin-3/EGF, mf75 , did not show any sensitivity to pept-1 RNAi, suggesting that loss of basal EGF-Ras-MAPK activity due to complete lack of LIN-3 abolishes starvation effects. These results indicate that nutrient deprivation via PEPT-1 modulates vulval induction upstream of LIN-3 or act in parallel to reinforce expression of downstream target genes of the Ras pathway (as seen with transcriptional reporters).
Given the above results, we next tested whether nutrient deprivation caused by the pept-1 mutation may alter levels of lin-3 expression in the anchor cell. Using single molecule FISH, we therefore compared lin-3 transcript number in wild type, pept-1(lg601), lin-3(e1417) and lin-3 (e1417); pept-1(lg601) animals ( Figure 4C). However, there was no effect of the pept-1 mutation on lin-3 transcript numbers in either a wild type genetic background or in a background with low transcriptional level of lin-3. Therefore, increased vulval induction of lin-3 (e1417) ( Figure 4C) caused by pept-1 mutation is not due to increased transcription of lin-3 in the anchor cell. Consistent with this, starvation effects on vulval induction of lin-3(e1417), which specifically reduces transcription in the anchor cell, are weaker than of lin-3(n378) (Figure 2). These results, together with the observation that vulval induction of lin-3(mf75), a genetic background with no expression of lin-3 in the anchor cell, is insensitive to pept-1 RNAi, suggests that starvation effects on vulval induction can only manifest under minimal EGF-Ras-MAPK activity, corresponding to the situation where one or more lin-3 transcripts are present in the AC. One possible scenario is that the effect of such low doses of lin-3 on vulval induction can then become amplified under nutrient deprivation, e.g., at the level of the LIN-3 protein and its secretion, or at the level or downstream of the EGF receptor (LET-23)
hypoinduction phenotype of the null mutant of bar-1/b-Catenin (allele ga80) (Eisenmann et al. 1998) was strongly aggravated under starvation conditions. Moreover, the highly penetrant Vulvaless phenotype of lin-3 (n378); bar-1(ga80) double mutants was not suppressed by starvation, implying that Wnt pathway activity acting via bar-1/b-Catenin is required for starvation suppression of lin-3(rf). However, it remained unclear to what extent starvation compromised induction vs. competence of vulval cells in bar-1(ga80) because in the study of  animals were exposed to starvation from the late L1 stagea stage during which competence of vulval precursor cells is acquired, yet induction does not yet take place. To distinguish these two scenarios, we exposed bar-1(ga80) animals to starvation in late L1 vs. mid-L2 stage and found that mid-L2 starvation did not alter vulval induction whereas late L1 starvation caused strong hypoinduction, primarily due to Pn.p fusion, i.e., loss of competence of vulval precursor cells, including P5.p to P7.p (Figures 5A-E). Late L1 starvation therefore aggravated bar-1(ga80) fusion defects prior to vulval induction, so that this starvation treatment cannot be used to evaluate starvation effects on vulval inductive levels. In contrast, starvation at a later time point, at mid-L2covering the time period during which vulval induction is most sensitive to starvationslightly, although not significantly, increased vulval induction of bar-1(ga80) animals ( Figure 3A). Yet, pept-1 RNAi treatment of bar-1(ga80) clearly and significantly increased vulval induction ( Figure 5F), indicating that such a prolonged starvation stimulus can indeed increase vulval induction in the absence of BAR-1 activity. However, when we exposed lin-3(n378); bar-1(ga80) double mutants to pept-1 RNAi treatment (Control: 0.00 6 0.00 cells induced, N = 44; pept-1 RNAi: 0.03 6 0.01 cells induced, N = 59), or 48 hr of starvation at the mid-L2 stage (Food: 0.00 6 0.00 cells induced, N = 45; Starvation: 0.01 6 0.01 cells induced, N = 47), we observed no starvation suppression of this highly penetrant Vulvaless phenotype. This result suggests, similar to the results obtained for other highly penetrant Vulvaless mutants [e.g. let-23(sy1); lin-3(n378) or lin-3(mf75)], that a very strong reduction in inductive signals (including contribution from BAR-1/Wnt) cannot be rescued by starvation effects. Nevertheless, given that vulval induction of the bar-1 null allele ga80 increases in response to starvation, we conclude that Wnt pathway/BAR-1 activity is not essential to mediate positive starvation effects on vulval inductive levels.
A previous study has further shown that liquid culture of animals suppressed the Vulvaless phenotype of lin-3(rf) and other EGF-Ras-MAPK mutations through the Wnt pathway (Moghal et al. 2003) ( Figure 1C). This environmental effect on vulval induction is based on activation of the heterotrimeric Gaq protein, EGL-30, which acts with muscle-expressed EGL-19 to promote vulval induction downstream or in parallel to LET-60/Ras in a Wnt-dependent manner (Moghal et al. 2003). The liquid effect on vulval induction is abolished when the Wnt pathway is mildly compromised (Moghal et al. 2003), i.e., in a bar-1(mu63) context (Maloof et al. 1999). We therefore asked whether starvation effects on lin-3(rf) could be mediated in the same fashion. However, starvation did not alter vulval induction of lin-3 (n378) in a bar-1(mu63) background ( Figure 5G). Furthermore, loss of egl-30/Gaq (Brundage et al. 1996) did not abolish starvation effects on vulval induction of lin-3(n378) ( Figure 5H). Therefore, the positive starvation effects on vulval induction seem to be distinct from previously reported effects of liquid culture promoting vulval induction (Moghal et al. 2003).

Conclusions
Analysis of starvation sensitivity of C. elegans vulval induction using mutant analysis and quantification of pathway activities allowed us to clarify how a specific environmental condition modulates C. elegans vulval induction ( Figure 6). We present quantitative analyses of starvation effects on vulval induction using lin-3(rf) mutants, demonstrating that starvation exposure of .19h exerts a strong positive effect during the entire period of vulval induction, spanning from mid-L2 to early L3 stages. By testing various candidate mechanisms that could transduce and elicit observed starvation effects, we find that reduced activity of DAF-2 (Insulin) or DAF-7 (TGF-b) does not mimic lin-3(rf) starvation suppression. Instead, nutrient-deprived animals induced by n Table 2 Effects of pept-1 RNAi on EGF-Ras-MAPK reporter egl-17::cfp (A) and Delta-Notch reporter lip-1::gfp (B). Results of statistical tests for transcriptional reporter assays ( Figure 5). We used an ANOVA testing for the fixed effects of RNAi treatment, Individual (nested in RNAi treatment), Pn.p cell and the interaction between RNAi treatment and Pn.p cell using mean signal intensity as a response variable. Including the effect Individual (RNAi treatment), allowed controlling for the non-independence of signal intensity in P5.p, P6.p, and P7.p   mutation of the intestinal peptide transporter pept-1 or by the TOR pathway component rsks-1 (in a food-rich environment) strongly mimicked suppression of lin-3(rf) by external starvation. Moreover, we find that pept-1 RNAi is sufficient to increase both EGF-Ras-MAPK and Delta-Notch pathway activities in Pn.p cells. These results provide evidence for a link between nutrient deprivation, TOR-S6K and EGF-Ras-MAPK signaling during C. elegans vulval induction. It remains unclear how nutrient deprivation feeds into genetic pathways regulating vulval development, and we do not know whether this phenomenon represents a specific mechanism to fine-tune levels of vulval induction in response to the environment. In any case, the observed starvation modulation of C. elegans vulval induction provides an ideal study system to understand how organismal nutrient status can alter activities of major signaling pathways and penetrance of mutant phenotypes.
A potentially important factor that could partly explain observed starvation effects on vulval induction is prolonged developmental time cause by either starvation exposure and mutation of pept-1 and rsks-1. Such slowing of development could potentially result in the accumulation of downstream effectors acting to promote vulval fates. Applied starvation treatments caused a strong delay in development: animals remained in the L2 stage throughout the 48-hr starvation period (the generation time under normal conditions at 20°is 3.5 days); and pept-1 and rsks-1 mutants also showed important developmental delays relative to the wild type, with pept-1 showing a stronger developmental delay than rsks-1. Moreover, longer starvation caused increased suppression of lin-3(rf) phenotypes (starvation of 19h to 48 h) ( Figure 2B). Importantly, however, vulval induction of lin-3(rf) was identical when starved for 48h to 120h) ( Figure 2B), which is somewhat contrary to the hypothesis that observed starvation effects on vulval induction are due to prolonged developmental time. In addition, a mutation in clk-1 which dramatically increases developmental time (Ewbank et al. 1997;Shibata et al. 2003) does not increase vulval induction in a lin-3(n378) background ( Figure 2I). Thus, a prolonged developmental time alone may not be sufficient to affect vulval induction.
Our results clarify some seemingly contradictory studies reporting both positive Ferguson and Horvitz 1985) Figure 5 Role of BAR-1 activity in starvation modulation of vulval induction. (A) Differences in vulval induction of bar-1(ga80) animals exposed to 48 hr of starvation in late L1 vs. mid L2 (and vs. control food conditions). Starvation at the late L1 stage, but not at the mid L2 stage, significantly reduced vulval induction of bar-1(ga80) animals (ANOVA, F 2,121 = 50.27, P , 0.0001). (B-D) Individual bar-1(ga80) fate patterns of P3.p to P8.p in (B) food vs. starvation conditions: late L1 (C) or mid L2 starvation (D). Vulval cell fate patterns of P3.p to P8.p (B-D) were, whenever feasible, separately inferred for Pn.pa and Pn.pp in cases of half-induced fates. Each line represents the vulval pattern of a single individual, and individuals are ordered from highest to lowest index of vulval induction. Color coding of vulval cell fates (1°: blue, 2°: red) and non-vulval cell fates (3°: yellow, 4°: gray). Induced vulval cells that could not be clearly assigned to either a 1°or 2°fate are coded in purple; non-induced cells that could not be clearly assigned a 3°or°fate are coded in white. (E) Proportion of Pn.p cells with a 4°(Fused) fate in control, starvation-exposed late L1, starvationexposed mid L2 bar-1(ga80) individuals (from the same experiment as shown in (A)). The proportions of 4°fates for each of P5.p to P7.p cell were higher after L1 starvation (P5.p: 37%, P6.p: 23%, P7.p: 44%) compared to control (P5.p: 2%, P6.p and P7.p: 0%) and L2 starvation (P5.p: 12%, P6.p: 0%, P7.p: 3%). (F) pept-1 RNAi increases vulval induction of bar-1(ga80) animals (ANOVA, F 1,81 = 14.47, P = 0.0003). (G) Starvation increased vulval induction of lin-3(n378); bar-1(mu63) relative to control food conditions (ANOVA, F 1,98 = 18.84, P , 0.0001). (H) Starvation increased vulval induction of egl-30(ad805); lin-3(n378) relative to control food conditions (ANOVA, F 1,59 = 7.28, P = 0.0091). and negative effects (Battu et al. 2003) of starvation on vulval induction. We confirm here that starvation results in a strong and consistent increase in vulval induction, e.g., illustrated by the strong starvation suppression of lin-3(rf) mutations ( Figure 2). This positive starvation effect appears to act independently of previously reported transmission of negative starvation effects via specific chemosensory components, Insulin or TGF-b signaling (Battu et al. 2003). The study by Battu et al. (2003) used sensitized backgrounds causing vulval hyperinduction (Multivulva phenotype), such as a gain-of-function mutation of let-60/Ras, to assess starvation effects: starvation decreased vulval induction in strains with overactivated LET-60 and MPK-1/MAPK signaling, yet had no effect on strains producing excessive LIN-3 or LET-23 (Battu et al. 2003). These results were interpreted to suggest that overstimulation of EGF/EGFR may overcome the negative starvation signal, however, it remained ambiguous at which level of the EGF-Ras-MAPK cascade the signal may integrate. The negative starvation effects on strains with excessive LET-60 or MPK-1 activation were abolished in chemosensory-defective mutants (osm-5, che-3, sra-13), implying starvation signal transduction via the sensory system (Battu et al. 2003). Moreover, daf-2(rf) suppression of the Multivulva phenotype caused by let-60/Ras(gf) in food conditions was interpreted to suggest mimicing (of observed) starvation effects via reduced Insulin signaling (Battu et al. 2003), but it was not tested how vulval induction of daf-2(rf); let-60(gf) responds to starvation. Our results reporting an overall positive effect of starvation on vulval induction suggest that sensory perception is not required for this effect to occur. Importantly, however, and in line with the results by Battu et al. (2003), we also find that a compromised DAF-2 activity reduces vulval induction in lin-3(rf) to similar extents in both food and starvation conditions, yet without abolishing starvation suppression of lin-3/egf(rf). Together, these different results suggest that starvation has both positive and negative effects on C. elegans vulval induction, with a positive starvation signal acting independently of external sensory inputs, likely mediated by nutrient sensing via TOR-S6K (this study), and a sensory-system-mediated negative starvation signal, likely involving DAF-2 signaling (this study; Battu et al. 2003). Despite the presence of two antagonistic starvation signals acting in parallel, there is a strong net positive starvation effect, suggesting that negative starvation effects are significantly weaker.
Our study and several others (Battu et al. 2003;Braendle et al. 2010;Ferguson and Horvitz 1985;Moghal et al. 2003) show that C. elegans vulval cell fate patterning is sensitive to diverse environmental inputs, which may act through distinct mechanisms to affect vulval signaling pathways. While these results show that specific developmental processes and underlying genetic pathways are responsive to the environment, several key questions remain: What are the consequences of this environmental sensitivity? How does the observed modulation of vulval signaling impact function and precision of this patterning process? Fundamentally, two opposed hypothetical scenarios can be considered to address these questions. First, environmental sensitivity of vulval development may be inevitable, so that environmental effects represent inherent environmental sensitivity of involved mechanisms. If this scenario holds true, the question is whether such environmental sensitivity translates into deleterious effects, such as patterning defects. In a second scenario, environmental sensitivity may reflect a specific developmental modulation to maintain or to enhance functioning and precision in different environmental conditions. This scenario would imply a vulval developmental system whose environmental flexibility is partly of adaptive origin. It is currently not known which of these Figure 6 Starvation modulation of C. elegans vulval induction. Schematic representation of (antagonistic) starvation effects reported in this and earlier studies (Battu et al. 2003;Ferguson and Horvitz 1985). scenarios apply to the observed environmental sensitivity of C. elegans vulval development. However, a previous study  has quantified functioning and precision of vulval cell fate patterning in different environmental conditions: wild type animals of multiple C. elegans isolates showed a very low rate of patterning defects in diverse, harsh environments, including starvation. Although starvation significantly increased levels of vulval induction , vulval patterning errors indicative of such increased inductive levels (e.g., hyperinduction) remained very rare (N2 strain, starvation: 3/1000 vs. control: 2/1000 individuals) . Therefore, this type of starvation modulation of vulval signaling pathways does not translate into an increased rate of corresponding errors, indicating that this process tolerates considerable changes in pathway activities. These observations reinforce the notion that the C. elegans vulval developmental system is robust to extensive signal fluctuations of involved pathways (Barkoulas et al. 2013;Braendle et al. 2010;Félix and Barkoulas 2012;Hoyos et al. 2011;Milloz et al. 2008). Nevertheless, although current results clearly demonstrate robustness, i.e., tolerance of the vulval developmental system to environmental variation, it remains to be evaluated whether specific environmental modulation of signaling pathways enhances the fidelity and precision of the vulval patterning output.

ACKNOWLEDGMENTS
We thank Marie-Anne Félix and Joao Picao-Osorio for discussion and comments on previous versions of the manuscript. We also thank the reviewers for helpful comments, which improved the manuscript. Strains were provided by the lab of Marie-Anne Félix and the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). CB acknowledges financial support by the Fondation ARC pour la Recherche sur le Cancer and the Centre National de la Recherche Scientifique (CNRS). SG was supported by fellowships from the Ministère de l'Enseignement Supérieur et de la Recherche and the Fondation ARC pour la Recherche sur le Cancer. AMVV was supported by the French government through ANR10-LABX-54 MEMOLIFE.