Abstract

   Interorgan signaling events are emerging as key regulators of
   behavioral plasticity. The foregut and hindgut circuits of the C.
   elegans enteric nervous system (ENS) control feeding and defecation
   behavior, respectively. Here, we show that epithelial cells in the
   midgut integrate feeding state information to control these behavioral
   outputs by releasing distinct neuropeptidergic signals. In favorable
   conditions, insulin and noninsulin peptides released from midgut
   epithelia activate foregut and hindgut enteric neurons, respectively,
   to sustain normal feeding and defecation behavior. During food
   scarcity, altered insulin signaling from sensory neurons activates the
   transcription factor DAF-16/FoxO in midgut epithelia, which blocks both
   peptidergic signaling axes to the ENS by transcriptionally shutting
   down the intestinal neuropeptide secretion machinery. Our findings
   demonstrate that midgut epithelial cells act as integrators relaying
   internal state information to distinct parts of the ENS to control
   animal behavior.
     __________________________________________________________________

   Gut epithelium modulates output from distinct enteric circuits by
   altering secretion of insulin and noninsulin peptides.

INTRODUCTION

   The enteric nervous system (ENS), also referred to as the “second
   brain,” is an autonomously acting neuronal network present in the
   digestive tract of all bilaterians, whose primary function is to
   regulate motility in the alimentary tract ([28]1–[29]4). The mammalian
   ENS consists of millions of neurons that fall into at least 20 neuron
   types that regulate gut peristalsis required for digestion, absorption,
   and egestion of food contents ([30]5–[31]8). Although enteric circuits
   are able to generate peristaltic behavior independent of central
   nervous system (CNS) input ([32]4), factors such as nutritional,
   reproductive, or psychological status of the animal have been shown to
   influence motility in the gut ([33]9–[34]11). Understanding how
   internal and external factors are sensed and integrated to coordinate
   the output from enteric circuits will provide novel insights on
   gastrointestinal physiology and disease.

   The ENS in Caenorhabditis elegans has been extensively studied in terms
   of its anatomy, development, molecular topology, and behavioral output
   ([35]12–[36]20), which makes it an attractive model for investigating
   how ENS function is regulated in metazoans. The C. elegans ENS consists
   of 22 neurons, 20 that are present in the foregut (the pharynx), from
   here on termed as pharyngeal enteric neurons (PENs), and two that
   innervate the hindgut, from here on termed as hindgut enteric neurons
   (HENs) ([37]Fig. 1A). The 20 PENs are further classified into 14
   anatomically and molecularly distinct classes forming a densely
   interconnected network that controls rhythmic pharyngeal pumping
   ([38]14–[39]16). The two HENs, termed AVL and DVB, are GABAergic
   neurons that are activated by the neuropeptide NLP-40 released from the
   intestine, which generates all-or-none action potentials in these
   neurons to control the rate of defecation ([40]12, [41]18, [42]20,
   [43]21). The rhythmic activation of the HENs, which form an
   electrically coupled circuit, results in γ-aminobutyric acid–dependent
   contraction of enteric muscles leading to expulsion of contents of the
   gut, a behavior termed as the defecation motor program ([44]12, [45]22,
   [46]23).

Fig. 1. Starvation inhibits ENS function via cell nonautonomous activity of
DAF-16/FoxO in the intestine.

   [47]Fig. 1.
   [48]Open in a new tab

   (A) Schematic showing the anatomical locations of PENs and HENs in the
   alimentary tract of C. elegans; adapted from WormBook ([49]117). (B and
   C) Pharyngeal pumping rate and frequncy of expulsion on food in fed
   versus starved (4 hours) wild-type adults and in starvation-induced
   dauer stage animals. (D and E) Timing of pBoc and Exp events, and
   expulsion:pBoc ratio on food in fed vs starved (4 hours) wild-type
   adults. (F) Pharyngeal pumping rate on food in fed versus starved (4
   hours) wild-type and daf-16(mu86) adults. (G) Intestine-specific DAF-16
   depletion in daf-16(ot853); otSi2[ges-1p::TIR1(F79G)]; daf-2(e1370)
   dauer stage animals. Animals were treated with either solvent (ethanol)
   or 100 μM 5-Ph-IAA. DAF-16 was not detected in the midgut epithelial
   cells after 5-Ph-IAA treatment in 15 of 15 animals. Scale bars, 20 μm.
   (H) Pharyngeal pumping rate on food after intestine-specific DAF-16
   depletion in fed versus starved (4 hours) daf-16(ot853);
   otSi2[ges-1p::TIR1(F79G)] adults. Animals were treated with either
   solvent (ethanol) or 100 μM 5-Ph-IAA. (I and J) Frequency of expulsion
   and expulsion:pBoc ratio on food after intestine-specific DAF-16
   depletion in fed versus starved (4 hours) daf-16(ot853);
   otSi2[ges-1p::TIR1(F79G)] adults treated with 100 μM 5-Ph-IAA.
   Horizontal line in the middle of data points represents median value of
   biological replicates in (B), (C), (E), (F), and (H) to (J). Additional
   horizontal lines represent 25th and 75th percentiles in (B), (F), and
   (H). *P < 0.05, ***P < 0.001, ****P < 0.0001; ns, not significant. (B
   and C) Dunn’s multiple comparison test after Kruskal-Wallis test. (E, I
   and J) Mann-Whitney test. (F and H) Sidak’s multiple comparisons test
   after two-way analysis of variance (ANOVA).

   After completion of postembryonic development, C. elegans continuously
   feeds and defecates, and the rates of both of these behaviors remain
   highly consistent both within and across individuals ([50]23–[51]26).
   When an animal is transiently moved away from its food source, there is
   an immediate reduction in the frequency of foregut and hindgut
   contractions, which become highly irregular because the animal is no
   longer actively consuming food ([52]23–[53]25). ENS output in the
   presence of abundant food can also be altered if an animal has
   experienced starvation in the recent past ([54]25, [55]27, [56]28), and
   this effect is more pronounced in the dauer diapause stage, which the
   animal enters after experiencing acute nutrient deprivation during
   early larval development ([57]29–[58]31). It remains little understood
   how different neuronal types in the ENS sense the internal state of an
   animal to coordinate gut contractility at the systemic level during
   periods of altered nutritional requirement ([59]10, [60]32, [61]33).

   Here, we show how distinct neuropeptidergic signals are transmitted
   from midgut epithelial cells to communicate the internal state of the
   animal to the foregut and hindgut enteric circuits and show that this
   interorgan signaling is required to maintain normal ENS output in
   favorable conditions. Previous work has shown that the neuropeptide
   NLP-40 released from gut epithelia activates the HENs via G
   protein–coupled receptor signaling to generate normal defecation
   behavior ([62]20, [63]21). We find that two neuropeptides of the
   insulin family, INS-7 and INS-35, are also continuously secreted from
   midgut epithelial cells, and these peptides are sensed by the insulin
   receptor on neuromodulatory NSM neurons in the foregut to induce rapid
   foregut contractions in nutrient replete conditions. When an animal
   undergoes acute starvation, the absence of insulin signaling from
   sensory neurons activates the stress-responsive transcription factor
   DAF-16/FoxO in the midgut epithelial cells ([64]34, [65]35). Using
   transcriptomic profiling, cell type–specific genetic manipulations and
   CRISPR-Cas9–generated endogenous reporters, we show that DAF-16/FoxO
   activation in the midgut epithelia reduces the secretion of
   neuropeptides from the midgut by transcriptionally silencing not only
   insulin genes but also genes that constitute the neuropeptide secretion
   machinery of the midgut epithelial cells. Because of the absence of a
   neuropeptide relay signaling via the intestine to the two enteric
   circuits, ENS output is greatly reduced both in terms of foregut and
   hindgut contractions even after the animal returns to a favorable
   environment after experiencing starvation. Our findings highlight the
   role of the intestinal epithelia as an integrator of the metabolic
   status of the animal and unveil the mechanisms of how the non–cell
   autonomous activity of a stress-responsive transcription factor
   modifies enteric behaviors by disrupting gut-to-ENS signaling during
   periods of nutritional adversity.

RESULTS

Starvation inhibits output from the ENS via non–cell autonomous activity of
DAF-16/FoxO in gut epithelial cells

   To study how the nutritional state of an organism alters output from
   the ENS, we measured how the two enteric behaviors in C. elegans,
   feeding and defecation, are affected by starvation. The rate of
   feeding, controlled by PENs, is measured as the frequency of foregut
   contractions, while the rate of defecation, controlled by HENs, is
   measured as the frequency of expulsion events from the hindgut. The
   defecation motor program consists of three steps: contraction of
   posterior body muscles (pBoc), contraction of anterior body muscles
   (aBoc), and expulsion (Exp) ([66]23). The pBoc step is initiated by a
   pacemaker in the intestinal epithelial cells ([67]36), which acts
   upstream of HENs, while the Exp step requires activation of the two
   HENs that constrict hindgut enteric muscles to release the contents of
   the gut through the anal opening ([68]20, [69]21). We measured how
   feeding and defecation behaviors were affected by starvation after the
   animal was transferred back to food and allowed an acclimation period
   of 5 min. We used two different starvation paradigms: (i) a 4-hour
   period of total food deprivation during adulthood that is sufficient to
   induce risk-taking behavior during food search ([70]37) and (ii) food
   deprivation combined with animal crowding during early development that
   induces remodeling into the dauer stage of diapause ([71]29, [72]30).

   We found that starvation during adulthood reduces the rate of feeding
   even after the animal is returned to food ([73]Fig. 1B and fig. S1A).
   In contrast, animals in the dauer stage display no foregut (pharynx)
   contractions on food, except for a few sporadic pharyngeal pumps, as
   described previously ([74]Fig. 1B) ([75]29, [76]31). We observe a
   similar effect of starvation on the rate of defecation. The frequency
   of pBoc events was unaffected after starvation (fig. S1B), but the rate
   of expulsions was significantly reduced ([77]Fig. 1, C and D). Half of
   the pBoc events in starved adults did not result in any expulsion of
   gut contents ([78]Fig. 1, C and D), which indicates altered
   communication between the intestinal pacemaker and the HENs. Although
   these animals underwent a period of food deprivation before the
   measurement of defecation behavior, their gut is filled with food
   contents in the 5 min of acclimation before behavioral recordings,
   which can be observed during the expulsion events. In contrast, dauer
   stage animals do not initiate any pBoc or Exp events when returned to
   food ([79]Fig. 1C and fig. S1B), which is likely attributable to the
   fact that these animals are not actively feeding because their buccal
   cavity is sealed ([80]38).

   Starvation during adulthood or in dauer-inducing conditions during
   postembryonic development both involve activation of the
   stress-responsive protein DAF-16 ([81]30, [82]39, [83]40), which is the
   sole C. elegans homolog of mammalian FoxO transcription factors. In
   nutrient replete conditions, agonist insulin peptides signal via DAF-2,
   the sole member of the insulin/insulin-like growth factor 1 (IGF-1)
   receptor (InsR) family in C. elegans, to trigger a phosphorylation
   cascade that retains DAF-16/FoxO in the cytoplasm in an inactive state
   (fig. S1C) ([84]35, [85]41–[86]43). When the environment becomes
   unfavorable, the absence of agonist insulins and presence of
   antagonistic insulins inhibit signaling via the DAF-2/InsR receptor,
   and this results in nuclear translocation of unphosphorylated
   DAF-16/FoxO, where it activates the transcription of stress-responsive
   genes (fig. S1C) ([87]35, [88]42, [89]43). We tested whether starvation
   affects enteric functions in animals that lack the DAF-16/FoxO
   transcription factor. We found that daf-16 null adults do not reduce
   their rate of feeding when placed on food after the same period of
   starvation that reduces foregut contractions in wild-type animals
   ([90]Fig. 1F). Similarly, in contrast to wild-type adults ([91]Fig. 1,
   C to E), a 4-hour starvation period did not significantly alter the
   rate of defecation in adults that lack daf-16 (fig. S1, D to F).

   A previous study has shown that in the dauer diapause stage, the
   removal of DAF-16/FoxO from the midgut epithelial cells can initiate
   foregut contractions in conditions that completely silence ENS output
   ([92]40), which suggests a non–cell autonomous role of this
   transcription factor in regulating ENS function. By visualizing an
   endogenously green fluorescent protein (GFP)–tagged DAF-16/FoxO
   protein, we found that this transcription factor localizes to the
   nuclei of midgut epithelial cells in starvation conditions that reduce
   the rates of feeding and defecation during adulthood (fig. S1G). To
   test whether removing DAF-16/FoxO only from the midgut epithelial cells
   can suppress the effects of starvation on enteric behaviors in adults,
   we performed spatially controlled depletion of DAF-16/FoxO protein
   using the AID2 technology, an improved version of the auxin-inducible
   degron (AID) system ([93]44–[94]46). These experiments were performed
   in a genetic background where daf-16 is endogenously tagged with AID*
   ([95]40), and tissue-specific heterologous expression of TIR1(F79G) was
   used for spatially controlled depletion of DAF-16/FoxO in the presence
   of 5-phenyl-indole-3-acetic acid (5-Ph-IAA), the synthetic auxin ligand
   for TIR1(F79G) ([96]44–[97]46). We find that AID2-mediated depletion of
   DAF-16/FoxO from all tissues using a ubiquitously expressed TIR1(F79G)
   transgene results in 99.8% of animals with the dauer-constitutive
   daf-2(e1370) mutation failing to arrest in the dauer diapause stage at
   25°C, thus reconstituting the dauer-defective phenotype of daf-16 null
   alleles (fig. S1H) ([98]35, [99]42, [100]43). In contrast, if
   DAF-16/FoxO is depleted from only the midgut epithelial cells using a
   TIR1(F79G) transgene driven by the midgut-specific ges-1 promoter, then
   98% of daf-2(e1370) animals arrest as SDS-resistant dauers at 25°C
   ([101]Fig. 1G and fig. S1I), which supports previous evidence that
   animals can attain dauer morphology even after depletion of endogenous
   DAF-16/FoxO protein from their intestine ([102]40). Using this
   intestine-specific AID2 system, we found that the effect of starvation
   on feeding rate can be completely suppressed if DAF-16/FoxO is
   selectively depleted from midgut epithelia of adult stage animals
   ([103]Fig. 1H and fig. S2A). Similarly, starvation does not affect
   defecation behavior in animals that have DAF-16/FoxO depleted only from
   their midgut epithelial cells ([104]Fig. 1, I and J, and fig. S2B).
   Unlike the effect of starvation on defecation behavior in wild-type
   animals ([105]Fig. 1, D and E), DAF-16/FoxO removal specifically from
   midgut epithelia resulted in the expulsion of gut contents immediately
   after most pBoc events even in starved animals ([106]Fig. 1J and fig.
   S2B). This establishes a non–cell autonomous role of DAF-16/FoxO to
   modify output from PENs and HENs to regulate the rates of feeding and
   defecation, respectively, after a period of nutrient deprivation during
   adulthood.

Evidence for two signaling axes from midgut epithelial cells to the ENS

   Our previous work has shown that DAF-16/FoxO also functions cell
   autonomously in the PENs to silence foregut contractions during acute
   starvation ([107]40). This suggests that when food is abundant, insulin
   peptides released from midgut epithelial cells might act on the PENs
   via interorgan signaling to prevent DAF-16/FoxO activation and generate
   normal feeding behavior, while during food deprivation, DAF-16/FoxO
   activity in the midgut likely alters the release of these gut-derived
   insulin peptides to activate DAF-16/FoxO in PENs via “FoxO-to-FoxO”
   signaling ([108]47). To test whether enteric neurons actively need
   insulin signaling to maintain normal ENS output in well-fed conditions,
   we inhibited DAF-2/InsR specifically in all enteric neurons by
   expressing a dominant negative form of the DAF-2 receptor using a
   driver that is expressed in both PENs and HENs (fig. S3A) ([109]40,
   [110]48). This dominant negative form of the protein, termed DAF-2(DN),
   has the intracellular kinase domain replaced by a fluorescent protein
   (fig. S3A), which allows DAF-2(DN) to bind insulin peptides on the
   extracellular side but does not allow the signal to be transmitted to
   its intracellular downstream effectors. Insulin/IGF-1 receptors need to
   be phosphorylated by their dimerization partner to achieve activation
   ([111]49), but the lack of a kinase domain in DAF-2(DN) results in
   inactivation of the wild-type DAF-2/InsR molecule it dimerizes with and
   thus producing a dominant negative effect.

   We found that expression of DAF-2(DN) in all neurons of the ENS (HENs
   and PENs) is sufficient to reduce the animal’s rate of feeding even in
   nutrient-rich conditions, but it has no effect on defecation behavior
   ([112]Fig. 2A and fig. S3, B to E). In terms of change in the rate of
   feeding, the effect of inhibiting DAF-2/InsR signaling in PENs of
   well-fed animals is identical to the effect of starvation on wild-type
   animals ([113]Fig. 2A). When animals expressing DAF-2(DN) in PENs were
   subjected to starvation, it does not further reduce their rate of
   feeding ([114]Fig. 2A), which indicates that starvation in wild-type
   animals reduces feeding rate likely via inhibition of DAF-2/InsR
   signaling in PENs. In contrast, each pBoc event is almost always
   followed by an expulsion of gut contents in animals that express
   DAF-2(DN) in HENs (fig. S3, D and E), which suggests that the
   communication between the intestinal defecation pacemaker and the HENs
   does not require active insulin signaling in the hindgut-innervating
   neurons. Hence, we conclude that the control of PENs and HENs by
   DAF-16/FoxO activation in the midgut involves independent signaling
   axes.

Fig. 2. DAF-16/FoxO silences transcription of several metabolic pathway genes
in the gut upon acute starvation.

   [115]Fig. 2.
   [116]Open in a new tab

   (A) Pharyngeal pumping rate on food in fed versus starved (4 hours)
   wild-type and otIs913[pha-4prom2::daf-2(DN)] adults. Horizontal line in
   the middle of data points and additional horizontal lines represent
   median of biological replicates, and 25th and 75th percentiles,
   respectively. *P < 0.05, ***P < 0.001; ns, not significant. Sidak’s
   multiple comparisons test after two-way ANOVA. (B) Schematic of RNA-seq
   strategy to determine gene expression changes mediated by intestinal
   DAF-16 in dauer stage animals. (C and D) Principal components analysis
   (PCA) plot and heatmap of sample-to-sample distance matrix for all
   samples used in RNA-seq analysis. (E) Volcano plots for expression
   changes due to auxin alone or auxin in the presence of intestinal
   TIR1(F79G). Dashed vertical lines indicate the fold change cutoff of
   1.5-fold. (F) Venn diagram for the number of differentially expressed
   genes for each of the four comparisons. (G) Heatmap of 1205 genes that
   are differentially expressed after depletion of intestinal DAF-16 in
   dauers. (H) Overlap of genes regulated by intestinal DAF-16 in dauers
   (this study) with all genes that undergo transcriptional changes during
   dauer formation as previously reported ([117]50). (I) Overlap of genes
   regulated by intestinal DAF-16 in dauers (this study) with known class
   I and class II transcriptional targets of DAF-16 as previously reported
   ([118]52). (J) Venn diagram showing overlap of genes regulated by
   intestinal DAF-16 in dauers (this study) with all genes that are
   differentially expressed in elt-2; elt-7 double mutant animals as
   previously reported ([119]53).

Identification of DAF-16/FoxO–induced gene expression changes in the
intestine during acute starvation

   We sought to identify these gut-to-ENS interorgan signaling axes by
   profiling DAF-16/FoxO–dependent gene expression changes in the
   intestine during nutrient deprivation. For this comparison, we chose
   animals arrested in the dauer diapause stage because: (i) the dauer
   stage is a more robust and stable state of nutritional stress response
   compared to starvation during adulthood ([120]39); (ii) the reduction
   in feeding rate is much stronger in the dauer stage, i.e., a complete
   silencing of feeding behavior ([121]Fig. 1B); and (iii) DAF-16/FoxO
   removal selectively from the dauer intestine using AID2 results in a
   strong reversal of the silenced state of the pharynx without affecting
   dauer morphology ([122]Fig. 1G, figs. S1I and S3, F and G). Because the
   AID2 system requires two components to achieve degradation of
   AID*-tagged DAF-16/FoxO, TIR1(F79G), and its synthetic auxin ligand
   5-Ph-IAA, we included three controls in our analysis: only TIR1(F79G)
   but no auxin, only auxin but no TIR1(F79G), and neither TIR1(F79G) nor
   auxin ([123]Fig. 2B). Since the AID2 system has not yet been
   extensively used for whole transcriptome level comparisons in animals,
   the use of these controls allows to estimate the gene expression
   changes due to TIR1(F79G) alone or auxin alone ([124]Fig. 2B).

   We find that at the whole transcriptome level, TIR1(F79G) alone or
   auxin alone do not produce noticeable gene expression changes
   ([125]Fig. 2, C and D). All five biological replicates of the
   experimental condition cluster separately from all the replicates of
   the three control conditions ([126]Fig. 2, C and D). The synthetic
   auxin ligand, 5-Ph-IAA, alters gene expression only when acting
   together with TIR1(F79G), but not alone ([127]Fig. 2E). On estimation
   of the number of differentially expressed genes for each comparison, we
   found that auxin alone or TIR1(F79G) alone altered the expression of
   less than 50 genes, while when both of the AID2 components were acting
   together, the expression of ~1000 genes was affected ([128]Fig. 2F). To
   find the genes that are affected by the removal of DAF-16/FoxO from the
   intestine, we selected the 1205 transcripts that were differentially
   expressed in the experimental condition in comparison to either the no
   TIR1(F79G) control or the no auxin control ([129]Fig. 2, F and G).
   Among these 1205 genes, 933 were up-regulated and 272 were
   down-regulated ([130]Fig. 2G and data S1).

   We compared the genes identified in our transcriptomic analysis to
   another study that has profiled the transcriptomes of the dauer and
   predauer L2D stages ([131]50). In the L2D stage, the animal is
   uncommitted to dauer formation and can resume reproductive growth if
   conditions become favorable ([132]51). We found that genes that are
   up-regulated during dauer formation are significantly enriched among
   genes that are down-regulated after DAF-16/FoxO removal from the dauer
   intestine (152 of 272, P = 2.53 × 10^−69) ([133]Fig. 2H). Similarly,
   genes that are down-regulated during dauer formation are significantly
   enriched among genes that are up-regulated after the intestine-specific
   removal of DAF-16/FoxO (382 of 933, P = 5.46 × 10^−110) ([134]Fig. 2H),
   which suggests that depleting DAF-16/FoxO only from the intestine can
   reverse several of the gene expression changes associated with dauer
   formation.

   The transcriptional targets of DAF-16/FoxO are classified into class I
   and class II, with class I targets being up-regulated by this
   transcription factor, and class II targets being down-regulated
   ([135]52). We found that genes up-regulated after DAF-16/FoxO removal
   from the dauer intestine have a significant enrichment of class II
   DAF-16 targets (304 of 933, P = 4.53 × 10^−96) ([136]Fig. 2I).
   Similarly, genes that were down-regulated after DAF-16/FoxO removal
   from the dauer intestine have a significant enrichment of class I
   DAF-16 targets (114 of 272 = 8.86 × 10^−50) ([137]Fig. 2I), which
   demonstrates that a large proportion of the genes identified in our
   study are direct transcriptional targets of DAF-16/FoxO.

   We performed pathway enrichment analysis on genes that are up-regulated
   after gut-specific DAF-16/FoxO removal in dauers and found that these
   genes are involved in several metabolic functions of the gut, such as
   synthesis and processing of fatty acids and vitamins, and function of
   organelles such as lysosomes and peroxisomes (fig. S3H). To identify if
   DAF-16/FoxO cooperates with other transcription factors in the gut to
   alter the expression of genes related to intestinal function, we
   scanned the upstream promoter regions of these genes for enrichment of
   any known transcription factor binding sites. Genes that are
   up-regulated after DAF-16/FoxO depletion from the dauer intestine show
   an enrichment for the binding sites of GATA and nuclear hormone
   receptor (NHR) family transcription factors in their upstream sequences
   (data S2). Because the GATA transcription factors ELT-2 and ELT-7 and
   several members of the NHR family of transcription factors are required
   for the expression of metabolism-related genes in intestinal epithelial
   cells ([138]53–[139]56), we speculated that DAF-16/FoxO shuts down
   several functional processes of the gut during acute nutritional
   deprivation by antagonizing the ability of these gut-specific
   transcription factors to activate specific target genes. To test this,
   we compared the genes that we identified to be affected by
   intestine-specific DAF-16/FoxO depletion with genes that are
   differentially expressed in animals that lack ELT-2 and ELT-7, GATA
   transcription factors that contribute to the differentiation program of
   the C. elegans intestine ([140]Fig. 2J) ([141]53). We found that ~30%
   of genes (273 of 933, P = 8.23 × 10^−136) that are up-regulated after
   DAF-16/FoxO depletion from the dauer intestine are down-regulated in
   animals that lack both ELT-2 and ELT-7 ([142]Fig. 2J), suggesting that
   DAF-16/FoxO induces global down-regulation of several intestinal
   differentiation genes by antagonizing the activity of gut-specific GATA
   transcription factors during acute starvation.

   For genes that are down-regulated after DAF-16/FoxO depletion from the
   dauer intestine, their upstream sequences show the expected enrichment
   of the DAF-16 binding element (DBE) “GTAAACA,” which is similar to the
   binding sites of other forkhead box (FOX) family of transcription
   factors (data S2) ([143]52, [144]57). In addition, we see enrichment of
   “TGCACTT,” the binding site for DAF-12/VDR, which is an NHR family
   transcription factor that cooperates with DAF-16/FoxO to remodel
   tissues during dauer formation (data S2) ([145]40, [146]58). We also
   found enrichment of binding sites for GATA transcription factors in the
   upstream sequences of genes that are down-regulated after gut-specific
   DAF-16/FoxO depletion (data S2). A small but statistically significant
   proportion of these genes (25 of 272, P = 0.0045) are also
   down-regulated in animals that lack ELT-2 and ELT-7, and many of these
   genes (e.g., mtl-1, ftn-1, gst-38, hsp-12.3 and hsp-12.6) are involved
   in responding to stressful conditions ([147]Fig. 2J). This suggests
   that DAF-16/FoxO cooperates not only with another dauer-associated
   transcription factor DAF-12/VDR but possibly also with GATA
   transcription factors such as ELT-2 to up-regulate stress responsive
   genes in the intestine during acute nutritional deprivation, as
   previously shown in the context of life span regulation for some of the
   same genes (mtl-1 and hsp-12.6) ([148]59).

DAF-16/FoxO shuts down the expression of the gut-derived insulin peptide
INS-7 to silence foregut contractions during acute starvation

   Because the primary goal of our transcriptomic analysis was to identify
   changes in expression of secreted factors from the gut that act via the
   DAF-2/InsR receptor in PENs, we focused on the expression change of
   insulin family of genes after intestine-specific DAF-16/FoxO removal
   during starvation. The C. elegans genome encodes 39 members of the
   insulin family of peptides that show a complex and dynamic pattern of
   expression ([149]41, [150]60–[151]62). We found that the gut-specific
   removal of DAF-16/FoxO during acute starvation up-regulates transcript
   levels of ins-7 and ins-33 and down-regulates the transcript levels of
   ins-1, ins-18, and ins-24 ([152]Fig. 3A). Compared to the reported
   changes of insulin gene expression during dauer formation in the
   presence of DAF-16/FoxO ([153]50), we found that they are affected in
   the opposite direction, i.e., ins-1, ins-18, and ins-24 are
   up-regulated, while ins-7 and ins-33 are down-regulated in dauers vs
   predauer L2d ([154]Fig. 3B), which demonstrates that the expression
   change of these insulins during acute starvation is mediated via
   DAF-16/FoxO activity in the intestine.

Fig. 3. Reversing the intestinal DAF-16/FoxO–induced changes in the
expression of insulin peptide genes mildly reverses the silenced state of the
dauer pharynx.

   [155]Fig. 3.
   [156]Open in a new tab

   (A) Change in expression levels of insulin family genes that are
   differentially expressed after intestine-specific DAF-16 depletion in
   dauers. (B) Change in expression levels of insulin family genes shown
   in (A) from the previously reported dauer versus L2d dataset ([157]50).
   (C) Expression of the endogenously tagged ins-1 reporter allele
   syb5452[ins-1::SL2::gfp::his-44] in fed adult and starvation-induced
   dauer stage animals. (D) Pharyngeal pumping rate on food in
   starvation-induced dauer stage animals with single or combination of
   null mutations in ins-1, ins-18, and ins-24 genes. (E) Pharyngeal
   pumping rate on food in starvation-induced dauer stage animals with
   defective neuropeptide secretion from the intestine, i.e., aex-1(sa9)
   and aex-5(sa23) mutants. (F) Expression of the endogenously tagged
   ins-7 reporter allele syb5424[ins-7::SL2::gfp::his-44] in fed adult and
   starvation-induced dauer stage animals. (G) Pharyngeal pumping rate on
   food in starvation-induced dauer stage animals that constitutively
   express ins-7 in the intestine (otEx8240 and otEx8241 lines).
   Horizontal line in the middle of data points represents median value of
   biological replicates in (D), (E), and (G). ***P < 0.001,
   ****P < 0.0001; ns, not significant. (D, E, and G) Dunn’s multiple
   comparison test after Kruskal-Wallis test. Scale bars, 20 μm (C and F).

   Because members of the insulin peptide family can act as agonists or
   antagonists of DAF-2/InsR receptor to produce a combinatorial effect on
   insulin/IGF-1 signaling at the systemic level ([158]35,
   [159]62–[160]64), there are two possible scenarios for how insulin
   peptides made in the gut can affect ENS output: (i) agonist insulins
   are released from the gut in favorable conditions that act on
   DAF-2/InsR in the PENs to prevent DAF-16/FoxO activation and generate
   normal feeding behavior in the presence of abundant food, and (ii)
   antagonist insulins are released from the gut in unfavorable conditions
   that inhibit DAF-2/InsR signaling in the PENs resulting in DAF-16/FoxO
   activation, which inhibits ENS output during periods of nutritional
   adversity (fig. S4A). DAF-16/FoxO activation in the intestine during
   starvation conditions is expected to shut down the expression of the
   agonist insulins and increase the expression of the antagonist insulins
   (fig. S4A). Because the expression of ins-1, ins-18, and ins-24 are
   up-regulated by intestinal DAF-16 during acute starvation ([161]Fig. 3,
   A and B), we speculated that active release of these insulins in the
   dauer stage might be required to completely silence foregut
   contractions. We generated CRISPR-based endogenous expression reporters
   for these three insulin genes, and unexpectedly, none of these are
   expressed in the intestine in well-fed or starved conditions ([162]Fig.
   3C and fig. S4, B and C). All of these insulins are expressed in head
   neurons, while ins-1 and ins-18 are also expressed in tail neurons
   ([163]Fig. 3C and fig. S4, B and C), which suggests that intestinal
   DAF-16/FoxO activation in the gut epithelia non–cell autonomously
   increases the expression of ins-1, ins-18, and ins-24 in the nervous
   system during acute starvation. We generated CRISPR-based null alleles
   for all of these insulin genes and found that the deletion of ins-1,
   ins-18, and ins-24, either individually or in combination, do not
   reverse the silenced state of the dauer pharynx ([164]Fig. 3D),
   suggesting that up-regulation of these insulin peptides is not required
   to silence foregut contractions during acute starvation.

   We found that a large number of noninsulin neuropeptide genes are
   transcriptionally down-regulated when DAF-16/FoxO is depleted from the
   dauer intestine (fig. S4D). This is consistent with a previous study
   that showed that the entire family of FMRFamide-like neuropeptide (flp)
   genes is transcriptionally up-regulated during dauer formation
   ([165]50), which suggests that some of these neuropeptides might be
   involved in silencing ENS output during starvation. We tested whether
   blocking pathways related to neuropeptide processing or release affects
   the silenced state of the dauer pharynx. We found that disruption of
   unc-31/CAPS, a calcium-dependent activator protein required for the
   release of neuropeptides in dense-core vesicles ([166]65, [167]66),
   sbt-1/SCG5, a chaperone for proprotein convertases that cleave
   propeptides ([168]67), or trap-1/SSR1, an ER membrane protein required
   for biogenesis of antagonistic insulin peptides ([169]68), did not
   induce foregut contractions in dauer stage animals (fig. S4E). Because
   unc-31/CAPS, sbt-1/SCG5, and trap-1/SSR1 are involved in the processing
   and release of neuropeptides from the nervous system but the signal
   that silences ENS output is regulated by DAF-16/FoxO in the gut, we
   also manipulated the genes that are involved in secretion of
   neuropeptides from the intestinal epithelial cells. Disrupting the
   functions of aex-1/UNC13D or aex-5/PCSK5, constituents of the
   neuropeptide secretion machinery of the gut ([170]21, [171]69), did not
   generate contractions in the silenced dauer pharynx ([172]Fig. 3E),
   which indicates that active secretion of neuropeptides from the gut
   epithelial cells is likely not required to silence PEN activity during
   acute starvation.

   Next, we focused on ins-7 and ins-33, insulin genes that are
   down-regulated by intestinal DAF-16/FoxO in dauer stage animals
   ([173]Fig. 3, A and B). We generated CRISPR-based endogenous expression
   reporters for both these genes and found that ins-7 is only expressed
   in the anterior region of the midgut of well-fed adults, while ins-33
   is expressed in hypodermal cells, but not in the gut ([174]Fig. 3F and
   fig. S4F). During acute starvation, the expression of both ins-7 and
   ins-33 are reduced in the respective tissues ([175]Fig. 3F and fig.
   S4F). We found that constitutive expression of ins-7 from gut epithelia
   produces a mild but significant reversal of the silenced dauer pharynx,
   while the constitutive expression of ins-33 produces no effect
   ([176]Fig. 3G and fig. S4G). This indicates that release of the INS-7
   peptide from gut epithelia can induce foregut contractions in
   starvation conditions that completely silence ENS output. The role of
   INS-7 in regulating foregut contractions in nutrient replete conditions
   is described in a subsequent section.

DAF-16/FoxO shuts down the neuropeptide secretion machinery of the gut during
acute starvation

   In addition to altering the expression of neuropeptides, we found that
   the depletion of DAF-16/FoxO from the dauer intestine also up-regulates
   the expression of several genes that encode for components of the
   neuropeptide secretion machinery of the gut ([177]Fig. 4A). Among these
   genes, the strongest up-regulation is observed for aex-1/UNC13D, a
   SNARE regulator ([178]21, [179]69), while modest up-regulation is found
   for aex-4/SNAP25, a t-SNARE protein ([180]21), and aex-5/PCSK5, a
   proprotein convertase ([181]Fig. 4A) ([182]21, [183]69). All of these
   aex genes are significantly down-regulated during dauer formation
   ([184]Fig. 4B) ([185]50). We generated CRISPR-based endogenous
   reporters for these aex genes to profile their expression pattern in
   well-fed adults, starvation-induced dauers, and during the L3 larval
   stage, which is the developmental equivalent of the dauer stage in the
   reproductive phase of growth. For aex-1/UNC13D, we observe strong
   expression in all intestinal epithelial cells in adults and L3 larvae
   grown in nutrient replete conditions ([186]Fig. 4C). In well-fed
   adults, we also observe aex-1/UNC13D expression in uterine cells and
   dim expression in the hypodermis ([187]Fig. 4C). In dauers, we observe
   a strong reduction in aex-1/UNC13D expression in the intestinal
   epithelium ([188]Fig. 4, C and D). For both aex-4/SNAP25 and
   aex-5/PCSK5, we observe strong expression in the epithelial cells of
   the midgut during adulthood and L3 larval stage ([189]Fig. 4E and fig.
   S5A). In well-fed adults, aex-4/SNAP25 is also expressed in uterine
   cells, while aex-5/PCSK5 is also expressed in the body wall muscles of
   animals ([190]Fig. 4E and fig. S5A). In the dauer stage, both
   aex-4/SNAP25 and aex-5/PCSK5 undergo a strong reduction in expression
   in the intestinal epithelial cells ([191]Fig. 4, E and F, and fig. S5,
   A and B). To determine whether the reduction in expression of aex genes
   during starvation is mediated by DAF-16/FoxO, we quantified the
   expression of these genes in a daf-16 null background in both fed and
   starved conditions. We found that the expression of neuropeptide
   secretion machinery genes aex-1 and aex-4 and the gut-derived insulin
   peptide ins-7 is strongly down-regulated in midgut epithelial cells
   when adult animals were subjected to prolonged starvation, and this
   reduction in expression was suppressed in the absence of DAF-16/FoxO
   ([192]Fig. 4, G and H, and fig. S5C). In summary, acute starvation
   strongly down-regulates the expression of components of the intestinal
   neuropeptide secretion machinery in a DAF-16/FoxO–dependent manner.

Fig. 4. DAF-16/FoxO shuts down the intestinal neuropeptide secretion
machinery genes during prolonged starvation.

   [193]Fig. 4.
   [194]Open in a new tab

   (A) Change in expression levels of aex-1, aex-4, and aex-5 genes after
   intestine-specific DAF-16 depletion in dauers. (B) Change in expression
   levels of aex-1, aex-4, and aex-5 genes from the previously reported
   dauer versus L2d dataset ([195]50). (C and D) Expression of the
   endogenously tagged aex-1 reporter allele
   aex-1(ot1543[aex-1::SL2::gfp::his-44]) in fed adult and L3 larva and in
   starvation-induced dauer stage animals. Quantification of aex-1
   expression in midgut epithelial cells is shown in (D). (E and F)
   Expression of the endogenously tagged aex-4 reporter allele
   aex-4(ot1530[aex-4::SL2::gfp::his-44]) in fed adult and L3 larva and in
   starvation-induced dauer stage animals. Quantification of aex-4
   expression in midgut epithelial cells is shown in (F). (G) Expression
   of the endogenously tagged aex-1 reporter allele
   aex-1(ot1543[aex-1::SL2::gfp::his-44]) in the midgut epithelial cells
   of fed versus starved (24 hours) adults. (H) Expression of the
   endogenously tagged aex-4 reporter allele
   aex-4(ot1530[aex-4::SL2::gfp::his-44]) in the midgut epithelial cells
   of fed versus starved (24 hours) adults. Horizontal line in the middle
   of data points and additional horizontal lines represent median of
   biological replicates and 25th and 75th percentiles, respectively, in
   (D) and (F) to (H). *P < 0.05, ****P < 0.0001; ns, not significant. (D
   and F) Dunn’s multiple comparison test after Kruskal-Wallis test. (G
   and H) Sidak’s multiple comparisons test after two-way ANOVA. Scale
   bars, 20 μm (C and E) and 5 μm (G and H).

   From existing chromatin immunoprecipitation sequencing (ChIP-seq)
   datasets, we found that the upstream genomic regions of aex-1/4/5 genes
   contain DAF-16/FoxO binding peaks (fig. S6, A to C). These regions also
   contain binding peaks for ELT-2, a GATA transcription factor that is
   the master regulator of intestinal differentiation (fig. S6, A to C)
   ([196]54). This suggests that ELT-2 might be required for the
   expression of aex-1/4/5 genes for normal secretory functions of the
   intestinal epithelia in favorable conditions, while DAF-16/FoxO
   activation antagonizes the activity of this gut-specific GATA
   transcription factor to silence the expression of aex-1/4/5 genes
   during nutritional adversity.

   Because multiple components of the neuropeptide secretion machinery of
   the intestine are transcriptionally silenced during starvation, we
   generated a secretion reporter to directly test whether the release of
   neuropeptides from gut epithelia is hindered in adverse conditions. We
   generated a transgene with the insulin peptide INS-1 translationally
   fused to a TagRFP fluorescent protein and expressed it specifically in
   intestinal epithelial cells ([197]Fig. 5A). From the same construct, a
   GFP-tagged histone H2B protein is synthesized after an SL2
   trans-splicing event, which reports whether transcription of the
   transgene is affected in any condition ([198]Fig. 5A). In well-fed
   adults and L3 larvae, the INS-1::TagRFP peptide is synthesized only in
   gut epithelia, but most of this protein is secreted from the
   basolateral side of these cells, and the tagged INS-1 peptide
   accumulates in the coelomocytes, macrophage-like scavenger cells in the
   body coelom ([199]Fig. 5A). In the dauer diapause stage, we observe
   that instead of being released from the basolateral side of gut
   epithelia, the INS-1::TagRFP peptide is now released into the lumen of
   the gut from the apical side of intestinal cells ([200]Fig. 5, A and
   B), a phenomenon that has been reported previously, but whose
   physiological relevance was unknown ([201]70). In addition, we observe
   strong accumulation of the tagged INS-1 peptide in the cytoplasm of gut
   epithelial cells in dauers ([202]Fig. 5, A and B), suggesting that the
   secretion of this neuropeptide is severely reduced during acute
   starvation conditions. We tested whether this neuropeptide secretion
   block in the dauer intestine is a direct consequence of DAF-16/FoxO
   activation. After gut-specific depletion of DAF-16/FoxO in dauers, we
   observe a robust increase in the secretion of INS-1::TagRFP from the
   basolateral side of intestinal cells with reliably stronger
   accumulation in the coelomocytes of the animal ([203]Fig. 5, C and D).
   These findings suggest that DAF-16/FoxO cell autonomously silences the
   secretion of neuropeptides from the intestine into the body coelom
   during acute starvation.

Fig. 5. DAF-16/FoxO shuts down the release of insulin peptides from the
midgut epithelium during prolonged starvation.

   [204]Fig. 5.
   [205]Open in a new tab

   (A and B) Secretion of INS-1::TagRFP from the intestine into
   coelomocytes in fed adult and L3 larva and in starvation-induced dauer
   stage animals expressing
   otEx8059[ges-1p::ins-1::tagRFP::SL2::gfp::his-44]. Quantification of
   INS-1::TagRFP intensity in gut lumen and gut cytoplasm and the ratio of
   INS-1::TagRFP intensity in coelomocytes: Gut cytoplasm is shown in (B).
   (C and D) Secretion of INS-1::TagRFP from the intestine into
   coelomocytes in daf-16(ot853); otSi2[ges-1p::TIR1(F79G)]; daf-2(e1370);
   otEx8220[ges-1p::ins-1::tagRFP] dauer stage animals after
   intestine-specific DAF-16 depletion. Animals were treated with either
   solvent (ethanol) or 100 μM 5-Ph-IAA. Ratio of INS-1::TagRFP intensity
   in coelomocytes: Gut lumen is shown in (D). (E and F) Secretion of
   INS-1::TagRFP from the intestine into coelomocytes in fed versus
   starved (24 hours) adults expressing
   otIs904[ges-1p::ins-1::tagRFP::SL2::gfp::his-44]. Ratio of
   INS-1::TagRFP intensity in coelomocytes: Gut cytoplasm is shown in (F).
   (G and H) Secretion of INS-1::TagRFP from the intestine into
   coelomocytes in fed wild-type or aex-1(ot1357) adults expressing
   otIs904[ges-1p::ins-1::tagRFP::SL2::gfp::his-44]. Ratio of
   INS-1::TagRFP intensity in coelomocytes: Gut cytoplasm is shown in (H).
   Horizontal line in the middle of data points and additional horizontal
   lines represent median of biological replicates and 25th and 75th
   percentiles, respectively in (B), (D), (F), and (H). *P < 0.05,
   ***P < 0.001, ****P < 0.0001; ns, not significant. (B) Dunn’s multiple
   comparison test after Kruskal-Wallis test. (D, F, and H) Mann-Whitney
   test. Scale bars, 20 μm (A, C, E, and G).

   Next, we tested whether neuropeptide secretion from gut epithelial
   cells is blocked only during the dauer diapause state or also when
   adult animals are subjected to acute starvation. We found that extended
   food deprivation in adults strongly reduces the basolateral secretion
   of the tagged INS-1 peptide from the gut, which is apparent from a much
   reduced INS-1::TagRFP accumulation in the coelomocytes ([206]Fig. 5, E
   and F). The lower insulin signal in the coelomocytes of starved adults
   is not due to reduced expression of this construct in the gut because
   the intensity of GFP-tagged H2B in the intestinal nuclei remains
   unchanged during starvation (fig. S7A). Since aex-1/UNC13D shows a
   strong DAF-16/FoxO–induced silencing during prolonged starvation
   ([207]Fig. 4G), which is associated with a neuropeptide secretion block
   from the intestine ([208]Fig. 5, E and F), we asked whether
   aex-1/UNC13D depletion also inhibits neuropeptide release from the gut
   of adults. We generated a CRISPR-based deletion of the entire
   aex-1/UNC13D protein coding region and found much reduced secretion of
   the tagged INS-1 peptide from the intestinal epithelial cells into the
   coelomocytes in aex-1/UNC13D null animals even in nutrient rich
   conditions ([209]Fig. 5, G and H). These results collectively
   demonstrate a role of aex-1/UNC13D in neuropeptide release from gut
   epithelia in well-fed conditions, and DAF-16/FoxO–mediated silencing of
   aex-1/UNC13D strongly reduces the secretion of neuropeptides from the
   gut during acute starvation conditions.

Active secretion of INS-7 and INS-35 from gut epithelia is essential for
normal foregut contractility in favorable conditions

   Because starvation-induced DAF-16/FoxO activity in the gut reduces
   intestinal neuropeptide secretion and the output from PENs in terms of
   foregut contractions, we speculated that active neuropeptide release
   from the gut is required to generate normal feeding behavior in
   favorable conditions. We found that animals with previously described
   loss-of-function mutations in the intestinal secretion machinery genes
   ([210]21, [211]69), aex-1/UNC13D and aex-5/PCSK5, have a significantly
   reduced rate of foregut contractions compared to wild type even in
   nutrient replete conditions ([212]Fig. 6A). This phenotype was also
   observed in animals with the CRISPR-generated null allele of
   aex-1/UNC13D ([213]Fig. 6B). Because aex-1/UNC13D and aex-5/PCSK5
   mutants were initially identified in a genetic screen for animals with
   defective defecation behavior ([214]23), it can be postulated that
   these animals are feeding less because of their lower rate of expulsion
   from the hindgut. Although previous studies have observed no
   correlation between the rates of foregut and hindgut contractions
   ([215]71), we tested whether a defecation defective strain that does
   not affect neuropeptide secretion from the midgut can affect the rate
   of feeding. For this assay, we used a loss-of-function allele of
   nlp-40, a neuropeptide that is rhythmically released from the
   intestinal epithelia to activate the HENs and initiate expulsion of gut
   contents ([216]20). Unlike the aex-1/UNC13D null mutant, the depletion
   of nlp-40 does not affect the release of other neuropeptides from the
   intestinal cells, although both of these strains have similarly strong
   defects in expulsion behavior (fig. S7, B and C). We found that animals
   lacking NLP-40 show the same frequency of foregut contractions as
   wild-type animals (fig. S7D), indicating that inhibiting defecation
   behavior without disrupting the neuropeptide secretion machinery of the
   gut does not affect the output from PENs.

Fig. 6. Active intestinal secretion of INS-7 and INS-35 is essential for
normal pharyngeal pumping behavior in favorable conditions.

   [217]Fig. 6.
   [218]Open in a new tab

   (A) Pharyngeal pumping rate on food in fed wild-type, aex-1(sa9), and
   aex-5(sa23) adults. (B) Pharyngeal pumping rate on food in fed
   wild-type and aex-1(ot1357) adults. (C) DAF-16 depletion from enteric
   neurons of the foregut in daf-16(ot853);
   otIs908[pha-4prom2::TIR1(F79G)] adults. Animals were treated with
   either solvent (ethanol) or 100 μM 5-Ph-IAA. The anterior and posterior
   bulbs of the pharynx are shown. Nonpharyngeal neurons where otIs908 is
   expressed are labeled. (D) Pharyngeal pumping rate on food after DAF-16
   depletion from enteric neurons of the foregut in fed daf-16(ot853);
   otIs908[pha-4prom2::TIR1(F79G)] adults in wild-type or aex-1(ot1464)
   background. Animals were treated with either solvent (ethanol) or 100
   μM 5-Ph-IAA. (E) Expression of the endogenously tagged ins-35 reporter
   allele syb6236[ins-35::SL2::gfp::his-44] in fed adult and
   starvation-induced dauer stage animals. (F) Pharyngeal pumping rate on
   food in fed wild-type, ins-7(ot1427), ins-35(ot1443), and
   ins-7(ot1427); ins-35(ot1443) adults. (G) Pharyngeal pumping rate on
   food in starvation-induced dauer stage animals that constitutively
   coexpress ins-7 and ins-35 in the intestine (otEx8222, otEx8223, and
   otEx8224). Horizontal line in the middle of data points and additional
   horizontal lines represent median of biological replicates and 25th and
   75th percentiles, respectively in (A), (B), (D), (F), and (G).
   *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns, not
   significant. (A, F and G) Dunn’s multiple comparison test after
   Kruskal-Wallis test. (B) Mann-Whitney test. (D) Sidak’s multiple
   comparisons test after two-way ANOVA. Scale bars, 20 μm (C and E).

   Because the intestine-to-pharynx interorgan signal is likely an insulin
   peptide that acts via the DAF-2/InsR receptor to maintain DAF-16/FoxO
   in an inactive state in PENs to generate normal foregut contractions in
   nutrient-rich conditions ([219]Fig. 2A), we hypothesized that removing
   DAF-16/FoxO specifically from the ENS should reverse the reduced
   feeding rate in animals that have defective neuropeptide secretion from
   the midgut. To experimentally test this hypothesis, we expressed
   TIR1(F79G) specifically in the ENS of an animal that has its endogenous
   daf-16/FoxO locus tagged with AID* ([220]Fig. 6C). In the presence of
   the auxin ligand, AID*-tagged endogenous DAF-16/FoxO protein is
   specifically depleted from the ENS of the animal ([221]Fig. 6C). We
   found that ENS-specific DAF-16/FoxO depletion does not affect the rate
   of feeding in wild-type animals, but it significantly increases the
   rate of foregut contractions in the intestinal neuropeptide secretion
   defective aex-1/UNC13D null animals ([222]Fig. 6D). After depleting
   DAF-16/FoxO specifically from the ENS of aex-1/UNC13D null animals,
   their feeding rates become similar to that of wild-type animals grown
   in the abundance of food ([223]Fig. 6D). This increase in the rate of
   foregut contractions is not observed in aex-1/UNC13D null animals that
   are treated with the same dose of auxin in the absence of TIR1(F79G)
   (fig. S7E), ruling out the possibility that suppression of this
   phenotype is due to treatment with the synthetic auxin ligand. This
   suggests that a feeding defect due to the absence of neuropeptide
   secretion from midgut epithelial cells can be completely suppressed by
   the removal of a transcription factor from the PENs, indicating the
   presence of a midgut-to-ENS communication that is modulated during
   nutritional adversity.

   To identify the molecular nature of the interorgan signal from the gut,
   we interrogated a recent single-cell RNA sequencing (RNA-seq) dataset
   that captured gene expression from all major cell types in C. elegans
   ([224]72). We found that only two insulin genes, ins-7 and ins-35, are
   specifically enriched in the intestinal cluster (fig. S8A). We
   confirmed using a CRISPR-based endogenous expression reporter that
   ins-7 is only expressed in the anterior region of the midgut of
   well-fed adults, and its expression is strongly reduced during acute
   starvation ([225]Fig. 3F). We also generated a CRISPR-based reporter
   for ins-35 and found that this insulin peptide is only expressed in all
   of the midgut epithelial cells of the animal in well-fed conditions and
   in no other tissue type ([226]Fig. 6E). In the dauer diapause stage,
   ins-35 expression is restricted to the anterior cells of intestinal
   epithelia ([227]Fig. 6E), but at the functional level, INS-35 secretion
   from the basolateral side will be much reduced due to the neuropeptide
   secretion block in acute starvation conditions ([228]Fig. 5, A and B).

   To determine whether these intestinally produced insulin peptides
   constitute the gut-to-pharynx interorgan signal, we measured the rate
   of feeding in animals with loss-of-function mutations in the ins-7 and
   ins-35 genes. We found that the rate of foregut contractions in the
   abundance of food was significantly lower in both the ins-7 and ins-35
   mutant strains in comparison to wild-type animals (fig. S8, B and C).
   However, because the ins-7 loss-of-function allele ok1573 also removes
   a part of the neighboring gene htas-1, and the ins-35 loss-of-function
   allele ok3297 only removes the last exon of ins-35, we generated
   CRISPR-based null alleles for both ins-7 and ins-35 by precisely
   deleting the entire coding region of these insulin peptide genes. We
   found that the rate of foregut contractions in nutrient replete
   conditions is significantly reduced in animals with a full locus
   deletion of either ins-7 or ins-35 ([229]Fig. 6F). If both these
   insulin peptides are removed simultaneously, then we do not observe an
   additive effect ([230]Fig. 6F). These results indicate that INS-7 and
   INS-35 released from the gut epithelial cells are required for normal
   feeding behavior in favorable conditions.

   We found that similar to ins-7 ([231]Fig. 3G), the constitutive
   expression of ins-35 in midgut epithelial cells results in a mild but
   significant reversal of the silenced state of the dauer pharynx (fig.
   S8D). However, the constitutive expression of ins-7 and ins-35
   simultaneously in the intestine of dauers generates a much stronger
   output from the PENs, and we observe higher than 20 contractions per
   minute in the top quartile of these animals, which are still arrested
   in the dauer diapause stage ([232]Fig. 6G). Because the feeding
   behavior was recorded 5 min after these dauer stage animals were
   transferred to food, it could be argued that ectopic expression of
   ins-7 and ins-35 triggers a faster exit from the dauer stage when these
   animals encounter food. We consider this unlikely because: (i) these
   insulin peptides do not affect the dauer exit process ([233]63), and
   (ii) molecular and behavioral changes associated with dauer exit are
   observed at least an hour after the dauer stage animals encounter food
   ([234]73, [235]74). To rule out the possibility of a faster dauer exit
   in animals coexpressing ins-7 and ins-35 from the gut, we measured
   their feeding behavior in the daf-7(e1372) dauer-constitutive genetic
   background. In the presence of the daf-7(e1372) mutation, animals
   constitutively arrest in the dauer stage at 25°C independent of the
   presence of food ([236]75). We found that in daf-7(e1372) dauers, the
   constitutive coexpression of ins-7 and ins-35 from the intestine is
   sufficient to induce rapid foregut contractions (fig. S8E). These
   findings indicate that the gut-derived insulin peptides, INS-7 and
   INS-35, not only are required for normal feeding behavior in favorable
   environments but also are sufficient to induce foregut contractions in
   harsh environmental conditions that completely silence output from the
   PENs.

The serotonergic NSM neurons sense insulin signals from the gut to generate
normal feeding behavior

   We sought to identify the site of action of the gut-derived insulin
   peptides in the pharyngeal ENS. Previous studies have found that the
   bilaterally symmetric MC neuron pair is the pacemaker of the pharynx
   because their ablation strongly reduces the rate of foregut
   contractions ([237]24, [238]26). To test whether pacemaker activity of
   the MC neurons requires active DAF-2/InsR signaling, we generated a
   strain that expresses DAF-2(DN) only in the MC neurons in the entire
   ENS ([239]Fig. 7A). The inhibition of DAF-2/InsR in MC neurons did not
   affect the rate of foregut contractions ([240]Fig. 7B), suggesting that
   the MC neuron pair does not require insulin signals to generate normal
   feeding output in favorable conditions.

Fig. 7. NSM senses INS-7 and INS-35 from the gut to enhance pharyngeal
nervous system output during the fed state.

   [241]Fig. 7.
   [242]Open in a new tab

   (A) Expression of otIs937[ceh-19prom2::daf-2(DN)::eBFP2::SL2::tagRFP]
   in MC(L/R) neurons. (B) Pharyngeal pumping rate on food in fed
   wild-type and otIs937[ceh-19prom2::daf-2(DN)::eBFP2::SL2::tagRFP]
   adults. (C) Expression of
   otIs942[tph-1prom5::daf-2(DN)::eBFP2::SL2::tagRFP] in NSM(L/R) neurons.
   (D) Pharyngeal pumping rate on food in fed wild-type and
   otIs942[tph-1prom5::daf-2(DN)::eBFP2::SL2::tagRFP] adults. (E)
   Pharyngeal pumping rate on food in fed versus starved (4 hours)
   wild-type and otIs942[tph-1prom5::daf-2(DN)::eBFP2::SL2::tagRFP]
   adults. (F) Pharyngeal pumping rate on food in fed
   otIs942[tph-1prom5::daf-2(DN)::eBFP2::SL2::tagRFP] expressing adults in
   wild-type or ins-7(ot1427); ins-35(ot1443) background. (G) Pharyngeal
   pumping rate on food in fed versus starved (4 hours) wild-type and
   tph-1(ot1545) adults. (H) Pharyngeal pumping rate on food in
   starvation-induced dauer animals that constitutively coexpress ins-7
   and ins-35 in the intestine (otEx8222) in wild-type or tph-1(ot1545)
   background. (I and J) DAF-16 protein in NSM neuron of fed versus
   starved (4 hours) adults and in starvation-induced dauer daf-16(ot853);
   otIs625[cat-1::SL2::mCherry::his-44] animals. Yellow asterisks show the
   nuclei of neighboring cells in (I). Nucleus-to-cytoplasm ratio of
   DAF-16 protein is shown in (J). (K and L) Secretion of NLP-40::TagRFP
   from the intestine into coelomocytes in fed versus starved (8 hours)
   adults expressing otIs927[ges-1p::nlp-40::tagRFP::SL2::gfp::his-44].
   Horizontal line in the middle of data points represents median value of
   biological replicates in (B), (D to H), (J), and (L). Additional
   horizontal lines represent 25th and 75th percentiles in (B), (D to G),
   (J), and (L). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns,
   not significant. (B, D and L) Mann-Whitney test. (E to H) Sidak’s
   multiple comparisons test after two-way ANOVA. (J) Dunn’s multiple
   comparison test after Kruskal-Wallis test. Scale bar, 20 μm (A, C, and
   K) and 1 μm (I).

   Next, we targeted the serotonergic NSM neuron pair in the ENS, which
   has a known neuromodulatory role in increasing the rate of feeding in
   the presence of food ([243]76). To inhibit DAF-2/InsR signaling in the
   NSM neurons, we expressed DAF-2(DN) using a cis-regulatory element of
   the tph-1/TPH gene that drives expression exclusively in the NSM
   neurons in the entire nervous system ([244]Fig. 7C) ([245]77). We found
   that inhibition of DAF-2/InsR signaling only in the NSM neurons is
   sufficient to reduce the rate of feeding in nutrient replete conditions
   ([246]Fig. 7D). NSM-specific DAF-2(DN) expression also strongly
   suppresses the effect of starvation on the rate of feeding ([247]Fig.
   7E), suggesting a role of insulin signaling in NSM neurons to mediate
   an internal state-dependent change in feeding behavior. Compared to
   wild-type animals, the reduction in feeding rate due to NSM-specific
   DAF-2/InsR inhibition is similar to that in animals that lack INS-7 and
   INS-35 ([248]Fig. 7F). To confirm that the gut derived insulins, INS-7
   and INS-35, act via DAF-2/InsR in NSM to regulate the rate of foregut
   contractions in favorable environments, we performed an epistasis
   experiment. We found that inhibiting DAF-2/InsR in the NSM neurons of
   animals that lack INS-7 and INS-35 results in no further decrease in
   their rate of feeding in nutrient rich conditions ([249]Fig. 7F). These
   results indicate that the neuromodulatory NSM neurons in the ENS sense
   the insulin peptides released from the midgut epithelial cells to
   generate rapid foregut contractions when food is abundant.

   Since the NSM neurons increase the rate of foregut contractions in a
   serotonin (5-HT)–dependent manner ([250]78), we asked whether 5-HT
   plays a role in altering the rate of feeding in starved animals.
   Previous studies have shown that animals lacking tph-1/TPH, the
   tryptophan hydroxylase enzyme essential for neuronal 5-HT biosynthesis,
   are unable to increase foregut contractions when they encounter food
   ([251]79). Using a CRISPR-generated null allele of tph-1/TPH, we show
   that animals lacking the ability to synthesize 5-HT in neurons have a
   much lower rate of feeding than wild-type animals ([252]Fig. 7G).
   Unlike wild-type animals, starvation does not reduce the rate of
   foregut contractions in tph-1/TPH null animals ([253]Fig. 7G), which
   supports the role of serotonergic signaling in modulating feeding
   behavior in the context of nutritional status of the animal. We also
   found that constitutive coexpression of the insulin peptides ins-7 and
   ins-35 from the intestine cannot reverse the silenced state of the
   dauer pharynx in a tph-1/TPH null background ([254]Fig. 7H), which
   indicates that the gut-derived insulin peptides require neuronal
   serotonergic signaling to induce foregut contractions.

   Since inhibiting DAF-2/InsR signaling in the serotonergic NSM neurons
   mimics the effect of starvation on feeding behavior ([255]Fig. 7E), we
   asked whether reduced DAF-2/InsR signaling during starvation results in
   nuclear translocation of DAF-16/FoxO in the NSM neurons (fig. S1C).
   Using a strain where mCherry-tagged histone H2B marks the nuclei of all
   monoaminergic neurons, we show that endogenous DAF-16/FoxO protein
   translationally fused to mNeonGreen predominantly localizes to the
   cytoplasm of NSM neurons in well-fed adults ([256]Fig. 7, I and J). In
   contrast, DAF-16/FoxO translocates to the nucleus of NSM neurons in
   starved adults or in animals arrested in the dauer diapause state
   ([257]Fig. 7, I and J). These results collectively indicate that
   reduced insulin signaling during nutrient deprivation triggers
   DAF-16/FoxO activation in the serotonergic NSM neurons of the foregut
   enteric circuit to inhibit feeding behavior.

Intestinal DAF-16/FoxO controls the second neuropeptidergic signaling axis to
HENs

   Previous analysis has shown that the intestinally expressed AEX-1/4/5
   peptidergic release machinery acts in the same pathway as NLP-40 to
   activate AEX-2, the NLP-40 receptor in HENs to control defecation
   behavior ([258]20, [259]21). The DAF-16/FoxO–dependent down-regulation
   of aex-1/4/5 under starvation conditions, described above, would
   therefore be expected to affect NLP-40 release from intestinal cells.
   To test this prediction, we generated a strain that expresses the
   NLP-40 neuropeptide fused with TagRFP specifically in the midgut
   epithelial cells. In well-fed adults, NLP-40::TagRFP synthesized in the
   intestine is secreted from the basolateral side of these cells into the
   body coelom, which is visualized by the accumulation of this
   neuropeptide in the coelomocytes ([260]Fig. 7K). The secretion of
   NLP-40 from the midgut epithelium is indeed strongly reduced after
   prolonged starvation in adults ([261]Fig. 7, K and L). Our findings
   collectively show that DAF-16/FoxO shuts down the release of both
   insulin and noninsulin peptidergic signals from the midgut epithelial
   cells to enteric neurons in the foregut and hindgut circuits during
   periods of nutrient scarcity.

DISCUSSION

   How the internal state of an animal affects CNS functions has received
   extensive attention ([262]32, [263]80), but how internal states affect
   the ENS remains much less studied ([264]81). Here, we have shown that
   the epithelial cells in the C. elegans midgut coordinate the function
   of two branches of the ENS by controlling the release of distinct
   peptidergic signals ([265]Fig. 8). These two distinct signaling
   axes—insulins from the midgut to the PENs and NLP-40 from the midgut to
   HENs—are coordinated via DAF-16/FoxO transcriptionally regulating key
   nodes of the peptidergic secretion machinery ([266]Fig. 8). On the
   basis of available ChIP-seq data for DAF-16/FoxO and the master
   regulator of intestinal differentiation, ELT-2 (fig. S6, A to C), we
   propose that ELT-2 normally activates aex-1/4/5 and that this
   activation is antagonized by DAF-16/FoxO under food-deprived
   conditions.

Fig. 8. Model for regulation of both feeding and defecation behavior by
neuropeptides released from midgut epithelial cells in favorable or starved
conditions.

   [267]Fig. 8.
   [268]Open in a new tab

   On sensing food-related cues in the environment, C. elegans
   chemosensory neurons release insulin peptides to modulate DAF-16/FoxO
   activity in the intestine ([269]35), which, as we have shown here,
   regulates the release of other insulins to control pharyngeal pumping
   and of a noninsulin neuropeptide to control enteric muscle
   contractions. We hypothesize that the relay of insulin signals from
   sensory neurons to midgut back to enteric neurons permits midgut
   epithelia cells to serve as an integrator of various types of
   nutritional signals both from the external environment and from
   internal nutritional signals from digested food within the alimentary
   tract. The nutrient sensing target of rapamycin complex 1 (TORC1) and
   TORC2 pathways have been shown to regulate DAF-16/FoxO activity
   specifically in the intestine ([270]82–[271]84), and the AAK-2/AMPK
   kinase directly phosphorylates DAF-16/FoxO to increase its activity
   during periods of extended starvation ([272]85). Reduced TORC2
   signaling in the gut in adverse environments inhibits the expression of
   insulin peptides in chemosensory neurons, which suggests a role of the
   intestine as an amplifier of a starvation signal since it can further
   reduce the release of agonist insulin peptides from the nervous system
   via a feed-forward loop ([273]86). Hence, intestinal DAF-16/FoxO may
   integrate signals from multiple stress response pathways to modify
   nervous system function during periods of extended nutritional
   adversity.

   It has been well documented that peptides of the insulin family can
   cross the blood-brain barrier to affect CNS function in mammals
   ([274]87). Research in invertebrate models has shed light on the
   mechanistic bases by which the release of specific insulin peptides is
   actively regulated from non-neuronal tissues in the gut to modulate
   behaviors during different internal states of the animal. Neuropeptides
   released from intestinal epithelia have been shown to affect sensory
   behaviors such as thermotaxis and chemotaxis during prolonged
   starvation ([275]37, [276]88). Altered insulin signaling from the gut
   during periods of food scarcity can also modify secretion of
   neuropeptides from specific sensory neurons ([277]89). Our findings
   show that in addition to the transcriptional regulation of individual
   insulin genes during starvation conditions, a remodeling of the entire
   secretion machinery of the gut occurs during acute nutritional
   deprivation, indicating a broader role of the gut epithelia as a
   secretory organ that controls animal behavior. This expands on the
   previously described “FoxO-to-FoxO” signaling mechanism by
   demonstrating that intestinal DAF-16/FoxO modifies the function of
   other organs not only by orchestrating insulin-mediated cross-tissue
   FoxO activation ([278]47) but also via coordinating other interorgan
   signaling axes that do not directly involve insulin signaling. An
   internal state-induced global block of all gut-to-brain signaling
   during adverse environments would affect the release not only of
   neuropeptides but also of yolk proteins, metabolites, and other
   signaling molecules that are released from the gut in secretory
   vesicles to modulate several aspects of the animal’s physiology
   ([279]90).

   We found that the pacemaker neurons in the ENS of the C. elegans
   foregut are not receptive to the gut-to-brain insulin signaling axis,
   which agrees with the previously proposed autonomous nature of the MC
   neurons ([280]24, [281]26). Instead, the neuromodulatory NSM neurons
   receive insulin cues from the midgut epithelial cells to modify the
   output from the foregut ENS circuit depending on the nutritional status
   of the animal. We hypothesize that activation of DAF-16/FoxO in NSM
   during starvation conditions alters 5-HT release from these neurons.
   This is supported by a previous study that found an altered response of
   the NSM neurons to food stimuli in animals that have experienced
   extended food deprivation ([282]91). Altered 5-HT release from NSM
   neurons may modulate the rate of foregut contractions by acting on
   known serotonergic receptors that are expressed in both neurons and
   muscles of the foregut, as described previously ([283]76, [284]78,
   [285]92). For the noninsulin peptidergic signal NLP-40 from the midgut
   that regulates defecation behavior by activating HENs ([286]20), our
   results demonstrate a physiological regulation of the release of this
   neuropeptide during nutrient scarcity. A reduced rate of expulsion from
   the hindgut of a starved animal would increase the residence time of
   food in its intestinal lumen, which will likely support increased
   nutrient absorption, a phenomenon that has been observed when
   Drosophila is exposed to low nutrient conditions ([287]9).

   We observe different feeding responses in starved adults and dauer
   diapause arrested larvae on encountering food. This difference is
   likely due to the activity of an additional stress-responsive
   transcription factor, DAF-12/VDR, during dauer remodeling ([288]30),
   which cell autonomously shuts down pacemaker activity in the ENS of the
   foregut in dauers ([289]40). Reconstituting the gut-to-pharynx insulin
   signaling during the dauer diapause stage activates this silenced
   circuit likely via the release of 5-HT from NSM neurons ([290]Fig. 7H),
   which is supported by a previous study that induced foregut
   contractions in dauer stage animals by exposing them to exogenous 5-HT
   ([291]93). While the inhibitory effect on foregut contractions is
   unambiguous in the starvation-induced dauer diapause stage, the effect
   of prolonged starvation on the feeding rate of adults is dependent on
   the conditions in which the behavior is recorded. Initial studies
   reported an increase in foregut contractions in starved animals, but
   this behavior was observed in the absence of food ([292]27, [293]94). A
   subsequent study found an increase in the rate of feeding immediately
   after starved adults were returned to food ([294]28). This is
   consistent with strong calcium activity in the NSM neurons after a
   starved animal encounters food and this activity lasts for less than 5
   min after food exposure ([295]91). These findings indicate a heightened
   sensitivity of the animal to food-derived cues after an extended period
   of starvation. Instead, if a starved animal is allowed to habituate on
   food for 5 min, then its rate of feeding is consistently lower than
   that in the fed state, which we have validated in blinded experiments
   (fig. S1A). Two recent studies that quantified feeding behavior over
   longer timescales, one using automated pharyngeal pumping measurements
   and another that quantified feeding rate independent of foregut
   contractions, have not observed an increase in the rate of feeding
   after starved animals are returned to food ([296]25, [297]95). This
   emphasizes the importance of separating the immediate response to the
   introduction of a new stimulus from the persistent effect of a change
   in the internal state of the animal in studies that measure behavioral
   output from autonomous circuits.

   The NSM neurons in the foregut enteric circuit can modulate behavioral
   outputs by altering overall brain state, which involves volumetric
   release of 5-HT to modify neural activity across multiple circuits via
   extra-synaptic communication ([298]78, [299]96). Axonal arborization
   patterns of serotonergic neurons in the mammalian brain suggest that
   5-HT release from individual neuron types can potentially have
   widespread effects on brain activity ([300]97). In the context of
   enteric circuits, the role of neuronally released 5-HT in enhancing gut
   contractions is evolutionarily conserved in the Drosophila and
   mammalian gut ([301]98, [302]99), but how these regulatory mechanisms
   are influenced during different physiological states of the animal need
   further exploration. Altered 5-HT signaling has been linked to
   functional gastrointestinal disorders in several clinical studies, but
   the internal or external factors that lead to 5-HT dysregulation in
   these patients remain poorly understood ([303]100). Understanding how
   impairment of insulin signaling affects the function of serotonergic
   neurons in the ENS might shed new light on this unexplored aspect of
   gastrointestinal physiology.

Limitations of the study

   In this study, we have used two types of starvation paradigms, (i)
   adult starvation on solid media and (ii) food deprivation and crowding
   during larval development that induces dauer formation. It is possible
   that additional pathways are involved in responding to other forms of
   starvation, such as bacterial dilution or food deprivation in liquid
   culture. We have elucidated the role of insulin and noninsulin
   peptidergic signaling between the gut epithelial cells and the enteric
   circuits in the foregut and hindgut of C. elegans, but it is possible
   that other forms of interorgan signaling involving small molecules
   (e.g., metabolites, short peptides, etc.) transported in secretory
   vesicles also contribute to communicating the starvation status of the
   animal from the intestinal epithelia to enteric neurons. Last, we have
   identified three components of the intestinal neuropeptide secretion
   machinery, the aex-1/4/5 genes, which are down-regulated by DAF-16/FoxO
   during starvation. It is likely that many other uncharacterized genes
   identified in our transcriptomic analysis are also essential components
   of the secretion machinery of the intestine that have not yet been
   functionally implicated in processing and release of signaling
   molecules from the gut.

MATERIALS AND METHODS

Maintenance of C. elegans strains

   All C. elegans strains used in this study and their genotypes are
   listed in data S3. All strains were maintained at 20°C on nematode
   growth medium (NGM) plates seeded with Escherichia coli (OP50 strain),
   unless otherwise specified. Strains with temperature-sensitive
   daf-2(e1370) and daf-7(e1372) alleles were maintained at 15°C. All
   strains were grown in optimum conditions in the abundance of food for
   at least three generations before using them in any experiment. All
   experiments in this study were performed using animals that were grown
   at 25°C for their entire developmental period (from unhatched embryos
   to adulthood) to be consistent with the temperature used for
   starvation-induced dauer formation assays.

DNA constructs and microinjections

   All plasmids used in this study were generated via NEBuilder HiFi DNA
   assembly [New England Biolabs (NEB)] using the manufacturer’s protocol.
   All plasmid sequences were validated using Sanger (Azenta) and/or
   Oxford Nanopore (Plasmidsaurus) sequencing. The pha-4prom2 sequence in
   pSS35 [pha-4prom2::TIR1(F79G)::mTur2::tbb-2 3′UTR] and pSS36
   [pha-4prom2::daf-2(DN)::ebfp2::tbb-2 3′UTR] is a 2393–base pair (bp)
   fragment upstream of isoform b of the pha-4 gene (−2387 to +6 position
   with respect to the start codon of pha-4 isoform b), which drives
   expression in all 22 enteric neurons (20 pharyngeal neurons and the two
   hindgut-innervating neurons, AVL and DVB) and also in RIS and PVT
   neurons ([304]48). The TIR1(F79G) sequence in pSS35 was generated by
   introducing the F79G [Phe(TTC) to Gly(GGA)] mutation in the Arabidopsis
   thaliana TIR1 sequence, as described previously ([305]45). The
   daf-2(DN)::ebfp2 sequence in pSS36 is from a previous study that
   replaced the tyrosine kinase and carboxyl-terminal domains of the cDNA
   of daf-2 isoform a (corresponding to amino acids 1246 to 1846) with the
   blue fluorescent protein ebfp2 sequence ([306]40). The tbb-2 3′UTR
   sequence is a 156-bp sequence immediately downstream of the stop codon
   of the tbb-2 gene. The ceh-19prom2 sequence of pSS43
   [ceh-19prom2::daf-2(DN)::eBFP2::SL2::tagRFP-T::tbb-2 3′UTR] is a
   1522-bp fragment upstream of isoform b of the ceh-19 gene (−1516 to +6
   position with respect to the start codon of ceh-19 isoform b). The
   tph-1prom5 sequence of pSS44
   [tph-1prom5::daf-2(DN)::eBFP2::SL2::tagRFP-T::tbb-2 3′UTR] is a
   previously described 232-bp fragment upstream of isoform a of the tph-1
   gene (−232 to −1 position with respect to the start codon of tph-1
   isoform a) ([307]77).

   The pSS29 [ges-1p::ins-1::tagRFP-T::SL2::gfp::his-44::tbb-2 3′UTR] and
   pSS34 [ges-1p::ins-1::tagRFP-T::SL2::ebfp2::his-44::tbb-2 3′UTR]
   constructs were generated by cloning the entire genomic locus (start
   codon to stop codon) of the ins-1 gene downstream of a 1999-bp ges-1
   promoter fragment (−1999 to −1 position with respect to the start codon
   of ges-1). The pSS31 [ges-1p::nlp-40::tagRFP-T::SL2::gfp::his-44::tbb-2
   3′UTR] construct was generated by cloning the cDNA of isoform a of the
   nlp-40 gene downstream of the ges-1 promoter fragment described above.
   pZW1 [ges-1p::ins-7::tbb-2 3′UTR], pZW3 [ges-1p::ins-33::tbb-2 3′UTR],
   and pZW24 [ges-1p::ins-35::tbb-2 3′UTR] were generated by cloning the
   entire genomic locus (start codon to stop codon) of ins-7 isoform b,
   ins-33, and ins-35, respectively, downstream of the ges-1 promoter
   fragment described above.

   Before microinjection, all plasmids were linearized using a single
   cutter restriction enzyme that cuts in the plasmid backbone. Linearized
   plasmids were injected at the concentrations listed in data S3 into
   both gonadal arms of young adult animals. F1 progeny was picked on the
   basis of the expression of the co-injection marker, and animals that
   transmitted the array to F2 progeny were used to generate transgenic
   lines. At least two independent lines (obtained from different injected
   P0 animals) for each injection type were used in subsequent
   experiments.

CRISPR-Cas9 genome editing

   CRISPR-Cas9 genome editing was performed using a modified version of a
   previously described protocol ([308]101). All CRISPR injection mixes
   constituted of Streptococcus pyogenes Cas9 nuclease [250 ng/μl;
   Integrated DNA Technologies (IDT)], trans-activating CRISPR RNA
   (tracrRNA; 100 ng/μl; IDT), all crRNAs [(combined); 56 ng/μl; IDT], and
   each single-stranded oligodeoxynucleotide (ssODN; 100 ng/μl) repair
   template (IDT). First, tracrRNA and all crRNAs were incubated at 95°C
   for 5 min followed by at 10°C for 5 min. Cas9 was added to this mix,
   pipetted several times, and incubated at 25°C for 10 min. All ssODN and
   plasmids were subsequently added to this mix, and the volume was filled
   to 20 μl using nuclease-free water. The mixture was centrifuged at
   17,900g for 2 min, and the supernatant was used for microinjections.
   For all CRISPR edits on chromosome II, the pRF4 [rol-6p::rol-6(su1006)]
   plasmid was co-injected at a concentration of 60 ng/μl. For CRISPR
   edits on all other chromosomes, dpy-10 co-CRISPR was performed using
   crRNA and ssODN listed in data S4. For all CRISPR edits, F1 animals
   were picked on the basis of roller phenotype and were genotyped by
   polymerase chain reaction (PCR) amplification of the edited locus,
   followed by Sanger (Azenta) or Oxford Nanopore (Plasmidsaurus)
   sequencing. For confirmed heterozygous edits, nonroller progeny was
   singled from F1 plates and genotyped by PCR to obtain homozygous lines.

   The genomic sequences for all genes edited in this study were obtained
   from WormBase ([309]102). The otSi2[ges-1p::TIR1(F79G)::mRuby::unc-54
   3′UTR] single-copy transgene was generated using a previously described
   technique by introducing the F79G [Phe(TTC) to Gly(GGA)] mutation in
   the TIR1 sequence of ieSi61[ges-1p::TIR1::mRuby::unc-54 3′UTR] using a
   crRNA and an ssODN listed in data S4 ([310]45). The ins-18(ot1326),
   ins-18(ot1328), ins-1(ot1360), ins-1(ot1363), aex-1(ot1357),
   aex-1(ot1464), daf-16(ot1705), ins-7(ot1427), ins-35(ot1443), and
   tph-1(ot1545) alleles were obtained by deleting the entire coding
   sequence (start codon to stop codon) of the corresponding gene using
   two crRNAs and an ssODN listed in data S4. The otDf2 allele was
   generated using two crRNAs and an ssODN listed in data S4 to delete a
   25.5-kb cluster of insulin genes, many of which (ins-24, ins-28,
   ins-29, and ins-30) are up-regulated in dauers ([311]50).

   The aex-1(ot1543[aex-1::SL2::gfp::his-44]),
   aex-4(ot1530[aex-4::SL2::gfp::his-44]), and
   aex-5(ot1532[aex-5::SL2::gfp::his-44]) alleles were generated using a
   crRNA (listed in data S4) and a long single-stranded repair template to
   insert the SL2::gfp::his-44 sequence immediately after the stop codon
   of the corresponding gene. For aex-1(ot1543[aex-1::SL2::gfp::his-44]),
   the SL2::gfp::his-44 sequence was inserted after the stop codon of
   isoform a of aex-1. The long single-stranded repair templates were
   generated using a previously described method ([312]101). The
   SL2::gfp::his-44 sequence was PCR amplified with primers containing
   35-bp homology sequence corresponding to each end of the genomic locus
   to be edited. One of the two primers used for PCR amplification had a
   5′ phosphate modification. The PCR product (10 μg) was digested with
   lambda exonuclease (NEB) to digest the DNA strand with 5′
   phosphorylation. The unphosphorylated strand of DNA was purified using
   the Monarch PCR & DNA Cleanup Kit (NEB) and was used as a repair
   template at a final concentration of 100 ng/μl in the CRISPR injection
   mix.

   Suny Biotech generated the ins-1(syb5452[ins-1::SL2::gfp::his-44]),
   ins-18(syb5462[ins-18::SL2::gfp::his-44]),
   ins-24(syb5447[ins-24::SL2::gfp::his-44]),
   ins-7(syb5424[ins-7::SL2::gfp::his-44]),
   ins-33(syb5561[ins-33::SL2::gfp::his-44]), and
   ins-35(syb6236[ins-35::SL2::gfp::his-44]) alleles by inserting the
   SL2::gfp::his-44 sequence immediately after the stop codon of the
   corresponding gene.

FLInt-mediated transgene integration

   The otIs904, otIs908, otIs911, otIs912, otIs913, otIs927, otIs937, and
   otIs942 transgenes were inserted at the intergenic
   oxTi553[eft-3p::tdTomato::H2B::unc-54 3′UTR + Cbr-unc-119(+)] V. locus
   using the fluorescent landmark interference (FLInt) method ([313]103).
   The FLInt injection mixes were prepared similar to CRISPR mixes, i.e.,
   first, tracrRNA (final concentration of 100 ng/μl) and a crRNA (final
   concentration: 56 ng/μl) targeting the tdTomato locus (listed in data
   S4) were incubated at 95°C for 5 min, followed by at 10°C for 5 min.
   Streptococcus pyogenes Cas9 nuclease (final concentration: 250 ng/μl)
   was added to this mix, pipetted several times, and incubated at 25°C
   for 10 min. After this, all plasmids along with GeneRuler 1 kb plus DNA
   ladder (Thermo Fisher Scientific) were added to the mix at the
   concentrations listed in data S3. The volume was filled to 20 μl using
   nuclease-free water and centrifuged at 17,900g for 2 min, and the
   supernatant was used for microinjections. All injected animals were
   maintained at room temperature, and F1 progeny expressing the
   co-injection marker was singled onto plates. F2 progeny was singled
   from F1 plates that had at least 75% of the animals expressing the
   co-injection marker. F2 plates that transmitted the co-injection marker
   to 100% of the progeny were used to generate lines.

5-Ph-IAA treatment for AID2

   For all AID2 experiments except the RNA-seq protocol, the animals were
   treated with the desired concentration of 5-Ph-IAA (BioAcademia) as
   described previously with some modifications ([314]104). A 100 μl of
   freshly prepared solution of 5-Ph-IAA in 25% ethanol (v/v) was added on
   top of 60-mm NGM plates seeded with OP50 bacteria. For control plates,
   an equal volume of 25% ethanol (v/v) was added such that the final
   ethanol concentration on the plate is 0.25% (v/v). All plates were
   stored in the dark for the entire duration of the experiment.
   Age-synchronized embryos obtained via the alkaline bleaching method
   were transferred to the plates 1 day after addition of 5-Ph-IAA, and
   the animals were grown at 25°C until they reached the desired
   developmental stage for the experiment.

Dauer formation assays

   A previously described protocol was used to obtain starvation-induced
   dauers ([315]105). Eight gravid adult worms were transferred to NGM
   plates seeded with OP50 bacteria at 25°C. After 6 to 8 days when all
   animals on the plates were starved due to lack of bacterial food,
   animals were collected in 1% SDS and incubated for 20 min with gentle
   agitation to kill all non-dauer stages. Animals were centrifuges at
   1150g for 2 min and washed twice with M9 buffer to remove all SDS. At
   this point, the tube contained living dauer stage animals and the
   carcasses of non-dauer stages. Animals were pipetted on a fresh NGM
   plate seeded with OP50 bacteria and live dauer animals were used for
   experiments within 40 min of removing the 1% SDS solution.

   Dauers for strains with the daf-2(e1370) or daf-7(e1372) mutation were
   generated by transferring age-synchronized embryos obtained via the
   alkaline bleaching method to NGM plates seeded with OP50 bacteria.
   Animals were maintained at 25°C for 3 days, after which all animals
   carrying the daf-2(e1370) or daf-7(e1372) mutation were arrested in the
   dauer stage. For dauer formation assays in the daf-2(e1370) genetic
   background after tissue-specific depletion of DAF-16 using AID2,
   age-synchronized embryos obtained via the alkaline bleaching method
   were maintained at 25°C for 3 days, after which the proportion of
   animals arrested in the dauer stage was estimated on the basis of dauer
   morphology and resistance to 1% SDS. The experiments were performed on
   at least three independent days.

Adult starvation assays

   To starve adult animals, age-synchronized embryos obtained via the
   alkaline bleaching method were transferred to NGM plates seeded with
   OP50 bacteria at 25°C. After 48 hours, adult stage animals were washed
   off the plate using M9 buffer and centrifuged at 300g for 1 min.
   Animals were then washed three times with M9 buffer to get rid of all
   bacteria. For the first two washes, the supernatant was removed after
   centrifugation at 300g for 1 min. For the final wash, animals were
   allowed to settle down by gravity for 5 min after which the supernatant
   was removed, and animals were pipetted on unseeded NGM plates for the
   desired amount of time. At the end of the starvation period at 25°C,
   animals were transferred to NGM plates seeded with OP50 bacteria and
   were used for experiments within 15 min of transferring them on food.

Pharyngeal pumping assays

   All pharyngeal pumping assays were performed on freely moving animals.
   The required number of animals was transferred to an NGM plate seeded
   with a uniform thin layer of OP50 bacteria. Animals were allowed to
   settle down for 5 min, and the movement of the grinder of the pharynx
   was recorded using a hand-held tally counter by observing the animals
   under a Nikon Eclipse E400 upright microscope equipped with
   differential interference contrast (DIC) optics. For age-synchronized
   adults, the number of grinder movements in a 20-s period was recorded
   using a 20× air objective lens of the Nikon Eclipse E400 microscope and
   was multiplied by three to obtain pharyngeal pumps per minute. For
   dauer stage animals, the number of grinder movements in a 1-min period
   was recorded using a 50× air objective lens of the Nikon Eclipse E400
   microscope. The rate of pharyngeal pumping was recorded from at least
   eight animals on each day and at least on two independent days.

Defecation assays

   All defecation assays were performed on freely moving animals. The
   required number of animals was transferred to an NGM plate seeded with
   a uniform thin layer of OP50 bacteria. Animals were allowed to settle
   down for 5 min, after which the posterior end of animals is observed
   using a 20× air objective lens of a Nikon Eclipse E400 upright
   microscope equipped with DIC optics. Each animal was observed for a 6
   min period, during which the timings of pBoc and Exp events (that
   release food contents from the anal opening) were recorded using the
   Behavioral Observation Research Interactive Software software version
   8.23 ([316]106). Defecation behavior was recorded from at least five
   animals on each day and at least on two independent days.

RNA isolation and library preparation

   OH14654: daf-16(ot853[daf-16::mNG::AID]); daf-2(e1370) and OH17582:
   daf-16(ot853[daf-16::mNG::AID]);
   otSi2[ges-1p::TIR1(F79G)::mRuby::unc-54 3’ UTR]; daf-2(e1370) strains
   were used for this experiment. Before growing animals, 250 μl of a
   freshly prepared solution of 5-Ph-IAA in 25% ethanol (v/v) was
   uniformly spread on top of each unseeded 100-mm NGM plate to obtain a
   final 5-Ph-IAA concentration of 100 μM on the plates. For control
   plates, an equal volume of 25% ethanol (v/v) was added such that the
   final ethanol concentration on the plate is 0.2% (v/v). All plates were
   stored in the dark for the entire duration of the experiment. After 1
   day of adding 5-Ph-IAA to the plates, a uniform thin layer of OP50
   bacteria was spread on top of each plate and incubated at 25°C for an
   additional day. Subsequently, ~8000 age-synchronized embryos obtained
   via the alkaline bleaching method were transferred to each plate and on
   three plates per condition. All plates were maintained at 25°C for 3
   days after which all animals were arrested in the dauer stage due to
   presence of the daf-2(e1370) mutation. Any animals that were not
   arrested in the dauer stage (less than 10 animals per plate) were
   picked and removed using a worm pick. A total of ~24,000 animals from
   three 100-mm NGM plates for each condition, which constituted a
   biological replicate, were collected in M9 buffer. Animals were washed
   three times using M9 buffer with centrifugation at 1300g for 2 min
   after each wash to remove all bacteria. A 200 μl of TRIzol reagent
   (Thermo Fisher Scientific) was added to the worm pellet, and samples
   were homogenized using a Kontes Pellet Pestle Motor (Thermo Fisher
   Scientific) for 30 s. Samples were immediately frozen on dry ice for 5
   min, and the homogenization-freezing cycles were repeated for a total
   of five times. All homogenized samples were stored at −80°C.

   After five biological replicates were collected for each condition,
   total RNA was isolated from all samples using the RNeasy Micro Kit
   (QIAGEN) using the manufacturer’s protocol. A 100 ng of total RNA per
   sample was used for library preparation using the Universal RNA-seq
   with NuQuant kit (Tecan Genomics) using the manufacturer’s protocol
   with custom AnyDeplete (IC0149S) to deplete ribosomal RNA. All 20
   libraries (4 conditions × 5 biological replicates) were pooled at a
   final concentration of 2 nM and were multiplexed for 75-bp single-end
   sequencing on a NextSeq 550 system (Illumina).

RNA-seq data analysis and visualization

   All sequencing analyses were performed on the Galaxy web platform
   ([317]107). The quality control of raw RNA-seq reads was performed
   using FastQC. Adapter sequences were trimmed using Trimmomatic
   ([318]108). The trimmed reads were aligned to the C. elegans genome
   (WS284 release) using STAR ([319]109). Read counts were assigned to
   genes using featureCounts ([320]110). The differential expression of
   genes across the four conditions was determined using DESeq2
   ([321]111). The principal components analysis (PCA) and
   sample-to-sample distance matrix plots were generated using DESeq2. The
   heatmap for differentially expressed genes was generated on the Galaxy
   web platform using the heatmap.2 function from the ggplot2 package on
   R.

   The overlap between different gene lists was identified using Venny
   2.1.0. Proportioned Venn diagrams were generated using Venn Diagram
   Plotter (Pacific Northwest National Laboratory). Kyoto Encyclopedia of
   Genes and Genomes (KEGG) pathway enrichment analysis and enrichment of
   transcription factor binding motifs in upstream sequences of
   differentially expressed genes were performed using ShinyGO
   ([322]112–[323]114). ChIP-seq peaks for DAF-16 and ELT-2 were
   visualized using JBrowse2 on Wormbase ([324]102).

Microscopy and image quantification

   Animals were mounted on a 5% agarose pad on top of a glass slide and
   were anesthetized using 50 mM sodium azide. Strains expressing
   daf-16(ot971[daf-16::gfp]) or daf-16(ot853[daf-16::mNG::AID]) were
   anesthetized using 4 mM levamisole (instead of sodium azide). Animals
   expressing daf-16(ot853), daf-16(ot971), otEx8059, otIs937, and otIs942
   were imaged on an inverted Zeiss LSM 980 laser scanning confocal
   microscope using a 40× water immersion objective lens. All other
   strains were imaged on a Zeiss Imager Z2 upright microscope equipped
   with Colibri 7 light-emitting diodes (Zeiss) using a 40× oil immersion
   objective lens. A full z-stack was acquired for all animals to cover
   their entire body width.

   All images were processed and quantified on the Fiji platform on ImageJ
   ([325]115). The mean background subtracted signal intensity from the
   two coelomocytes in the anterior region of the body was quantified for
   INS-1::TagRFP and NLP-40::TagRFP secretion assays. For the same assays,
   the signal intensity from the intestinal lumen and intestinal cytoplasm
   were calculated as the mean background subtracted intensity from two
   different regions of the intestinal lumen or cytoplasm, respectively.
   For quantification of signal intensity from the intestinal nuclei, the
   mean background subtracted intensity was calculated from two intestinal
   nuclei, one each from the first two rows of the intestinal cells.

Statistical analysis

   Data points for all biological replicates and sample sizes (N) are
   displayed in the figures. Statistical tests used in each figure and the
   corresponding P values are listed in the figure legends. P < 0.05 was
   considered to be statistically significant. To calculate the P value
   for overlap between two gene lists, hypergeometric tests were performed
   on R. All other statistical tests and plotting were performed on
   GraphPad Prism 10. For each type of quantitative assay described in
   this study, sample sizes were estimated by performing a pilot
   experiment using eight animals per condition to determine the mean and
   SD. Using these values, the required sample size was estimated using
   G*Power for a type I error rate of 0.05 and a statistical power of 0.95
   ([326]116). If the estimated sample size per condition was less than
   35, the experiment was performed again using the estimated total number
   of animals. For all subsequent assays of the same type, the same sample
   size per condition was used.

Acknowledgments