Insulin signaling promotes germline proliferation in C. elegans

Development of multicellular organisms requires the coordination of cell proliferation and specification. Although major cell cycle regulators are defined, less is known about the control of cell proliferation by growth factors within developmental contexts (e.g. Edgar and Lehner, 1996; Fichelson et al., 2005; Orford and Scadden, 2008). Furthermore, the control of proliferation within stem and transit amplifying cell populations is not well understood (Kohlmaier and Edgar, 2008). The intersection of cell proliferation and cell fate specification is important for understanding human diseases, particularly cancer, which is characterized by the dysregulation of both proliferation and fate (see Hanahan and Weinberg, 2000; Sancho et al., 2004).

The insulin/IGF receptor (IIR) family is conserved across metazoans and controls multiple aspects of cell growth and metabolism (Taguchi and White, 2008). Defects in IIR-associated pathways account for a host of human diseases, including diabetes and cancer. Invertebrate model organisms have contributed much towards understanding IIR signaling on the cellular and organismal levels, especially with respect to metabolism, developmental decisions and lifespan. The C. elegans genome contains one IIR-encoding gene daf-2 (Kimura et al., 1997), and 40 putative insulin-like peptide genes (Pierce et al., 2001). The IIR signaling cascade initiated by DAF-2 is highly conserved from C. elegans to mammals: receptor activation signals through a PI 3-kinase cascade that results in phosphorylation and nuclear exclusion of a FOXO transcription factor (C. elegans DAF-16).

C. elegans daf-2 is implicated in many genetically separable processes, including the dauer decision, lifespan control, and reproductive timing (see Baumeister et al., 2006; Fielenbach and Antebi, 2008; Gami and Wolkow, 2006; Hu, 2007; Li and Kim, 2008; Murphy, 2006). Mutants with reduced daf-2 activity enter dauer even under replete conditions, and have extended lifespan and reproductive timing (Gottlieb and Ruvkun, 1994; Kenyon et al., 1993; Larsen et al., 1995; Vowels and Thomas, 1992).

C. elegans germline development (Hirsh et al., 1976), like that of many animals, includes a phase in which undifferentiated germ cells proliferate extensively, forming a stem cell or progenitor pool that maintains gamete production throughout reproductive adulthood. In C. elegans, the GLP-1/Notch receptor pathway maintains germ cells in the undifferentiated (mitotic) state and/or prevents their differentiation (meiotic entry) (Austin and Kimble, 1987). Two DSL family ligands, LAG-2 and APX-1 are expressed in the distal tip cell (DTC) (Henderson et al., 1994; Nadarajan et al., 2009), and activate GLP-1 in neighboring germ cells. Loss of Notch signaling causes all germ cells to differentiate (enter meiosis), whereas constitutive Notch signaling prevents differentiation (Austin and Kimble, 1987; Berry et al., 1997). The C. elegans germ line is, therefore, a prototypical system in which signaling promotes the undifferentiated (proliferative) versus differentiated (non-proliferative) fate. In such systems, the relative contributions of signaling to mitotic cell fate specification and cell cycle control can be difficult to parse.

Independently of cell fate patterning by Notch pathway activity, the gonadal sheath regulates larval germline proliferation (Killian and Hubbard, 2005; McCarter et al., 1997). Therefore, other signaling contributes to the control of robust germline proliferation.

In this study, we uncover a novel role for DAF-2 in germline development. We find that IIR signaling promotes robust germline cell cycle progression during the crucial larval expansion phase. We demonstrate that IIR activity affects cell cycle kinetics in a manner distinct from that of Notch signaling, is separate from the dauer decision, does not account for the sheath signal, and occurs through the canonical IIR-PI3K-FOXO pathway primarily in the germ line. We further implicate somatic activity of two of the 40 putative insulin-like peptide ligands in this function.

Strains (see Table S1 in the supplementary material) were derived from N2 wild type (Bristol) and handled using standard methods (Brenner, 1974). Plasmids used are listed in Table S2 in the supplementary material.

A procedure modified from Gumienny et al. was used (Gumienny et al., 1999): worms were washed in M9, incubated in 200 μl of 33 mM SYTO12 (Invitrogen) in M9 for 5 hours at room temperature in the dark, destained for 1 hour in the dark on a fresh plate of OP50, and scored live while paralyzed in levamisole.

RNAi by bacterial feeding was performed as described by Timmons et al. (Timmons et al., 2001). Unless otherwise indicated, parental worms were grown on OP50 at 15°C, their progeny were synchronized by L1 hatch-off (Pepper et al., 2003) and grown at 20°C to the desired stage. Bacteria bearing the empty RNAi expression vector L4440 served as a negative control.

Images were collected from a Zeiss Imager Z1 with an Apotome Axioimager (Carl Zeiss) using an AxioCamMRm digital camera and Zeiss AxioVision and NIH ImageJ software.

After synchronization by L1 hatch-off, vulval development was used as the primary developmental stage marker [see figure 4 of Seydoux et al. (Seydoux et al., 1993)]. Early L3 is characterized by an undivided P6p under a distinct anchor cell; L3/L4 by vulval morphology and initiation of the gonad turn. `Early adult' stage was 8-10 hours post-mid to mid/late L4 at 20°C, and characterized by adult vulval morphology but few if any embryos in the uterus.

The `number of nuclei in the proliferative zone' included all the germ nuclei between the distal tip and the transition zone, as determined after ethanol fixation and DAPI staining (Pepper et al., 2003). The distal edge of the transition zone border was defined as the first cell diameter in which two or more nuclei displayed the characteristic crescent shape (Crittenden et al., 2006). `Distance to transition zone' was measured in cell diameters from the distal tip to the transition zone border. The mitotic index is the number of metaphase and anaphase figures over the total number of nuclei in the proliferative zone (Maciejowski et al., 2006).

EdU-labeled bacteria were fed to L4 N2 and daf-2(e1370) worms for 30 minutes prior to fixation, and processed as described by Dorsett et al. (Dorsett et al., 2009), with the following modifications. Whole worms or dissected gonads (both yielded similar results) were fixed for 10 minutes in 3.7% paraformaldehyde (Electron Microscopy Sciences) in PBST, washed in PBST, incubated for 5 minutes in –20°C methanol, and washed three times in PBST. After the Click-IT EdU reaction (Molecular Probes), preparations were imaged. S-phase index is the percentage of the DAPI-labeled proliferative zone nuclei that are EdU positive. Nuclei with any EdU labeling (individual chromosomes or all chromosomes) were scored as positive.

The relative DNA content of proliferative zone nuclei was determined as described (Feng et al., 1999), with the following modifications. Worms were grown on OP50 at 20°C to the mid to mid/late L4 stage, ethanol fixed for 10 minutes, incubated for 1 hour at 37°C with 40 μg/ml RNAse A in PBS, then imaged on agar pads in propidium iodide (Vector Laboratories, H-1300). The area and average intensity was determined for each nucleus in each z-layer (0.5 μm intervals) using ImageJ. The sum of the products of area × the average intensity for each layer gave the total fluorescence per nucleus. Average background intensity (from four similarly sized circles within or adjacent to gonad) was subtracted from measurements of each slice. 2N DNA content was established from non-mitotic somatic cells of the vulva and uterus in the same animal and from sets of daughter chromosomes of anaphase germ nuclei, and was verified using 4N nuclei (metaphase figures and pachytene nuclei). To obtain the haploid equivalent, the total fluorescence from each germ cell nucleus was divided by one half of the 2N value obtained from the somatic cells. Every nucleus was measured from the distal tip to the first cell diameter within four cell diameters of the transition zone (to avoid meiotic S).

Parental GC678 tnIs6[lim-7::GFP];qIs19[lag-2::GFP] worms were grown on OP50 at 20°C; their progeny were synchronized by L1 hatch-off, washed and placed on plates with daf-16(RNAi)-inducing or L4440-bearing bacteria (empty RNAi expression vector, used as a negative control), and grown at 20°C until the early L3 stage. Laser microsurgery (Photonics Instruments Micropoint Laser System) of both SS cells in a single gonad arm was carried out as described (Killian and Hubbard, 2005). Unablated controls were the unoperated gonad arm and additional siblings mounted on the same slides as operated animals. Ablated worms were placed back onto daf-16(RNAi) or L4440 and allowed to develop until adulthood at 20°C. Adult gonad arms with a GFP-positive DTC but no GFP-positive cells at the normal sheath location were considered to be successful ablations. To control for daf-16 RNAi efficacy, the same bacterial cultures were tested for suppression of daf-2(e1370) dauer at 25°C by L1 feeding.

Experiments were performed as described (Dillin et al., 2002). Synchronized populations of L1 (1 hour post-hatch) daf-2 (e1370), ins-3(ok2488) and ins-33(tm2988) worms were grown on daf-16(RNAi) and L4440 until the mid to mid/late L4 stage at 20°C. Individual worms were placed on separate plates at 20°C, transferred to new plates every 12 hours, and live progeny counted.

Worms (GC865 and CF1553) were grown on RNAi bacteria targeting daf-2, ins-3 or ins-33 (or L4440) to the mid to mid/late L4, and images captured. All animals were examined within 10 minutes because, after 15 minutes on the agar pad (in 0.4 mM levamisole to immobilize worms), DAF-16::GFP translocated to the nucleus even in controls.

Several mutations in daf-2, the gene encoding the C. elegans insulin/IGF-like receptor (IIR), cause a sterile or a reduced-fecundity adult phenotype, suggesting a role for IIR signaling in gonadogenesis or germline development (Gems et al., 1998; Malone and Thomas, 1994; Patel et al., 2008; Tissenbaum and Ruvkun, 1998). We examined a well-characterized temperature-sensitive, dauer-constitutive allele, daf-2(e1370), under non-dauer-inducing conditions (well fed and at a semi-permissive temperature of 20°C). Compared with stage-matched wild-type (N2) animals, we found that daf-2(e1370) adult hermaphrodites and males have fewer germ cells in the distal proliferative zone (Fig. 1; see also Table S3 in the supplementary material). This region contains pre-meiotic stem or progenitor germ cells, the majority of which are in the mitotic cell cycle (Crittenden et al., 2006).

To further characterize the daf-2 germline phenotype, we conducted temperature-shift and time-course analyses (Fig. 1A). We observed severely reduced numbers of distal germ cells after a shift to the non-permissive temperature early in the third larval stage (L3; after the dauer entry decision). This shift reduced the average brood size to 17±1.9 progeny (n=10 broods; wild-type control broods averaged >200), underscoring the importance of robust larval proliferation for fecundity. These results also suggest that the germline and dauer phenotypes are separable.

Time-course analysis at 20°C (<1% dauer; see Table S4A in the supplementary material) revealed a defect in the amplification of proliferative zone germ cell numbers during the L3 and L4 stages. After the early L3 stage, the proliferative zone of daf-2(e1370) animals contained, on average, fewer germ cells than that of stage-matched N2 animals (Fig. 1A). The difference was first evident in the mid-L3 but became more pronounced at later stages, consistent with a cumulative effect. We found a similar but less-marked defect in germ cell numbers after L1-initiated daf-2(RNAi) feeding (see Table S3 in the supplementary material; see also Fig. 3B; at 27°C, the same RNAi yields 100% dauer; see Table S4C in the supplementary material).

In other systems, insulin/IGF signaling is associated with cell size control (Edgar, 2006). We measured cell volume in 40-50 total germ cells from multiple individuals. RNAi directed against daf-2, ins-3 or ins-33 (see below) had no effect on L4 germ cell volume, nor did a daf-2 mutation (see Table S5 in the supplementary material).

A reduction in the number of cells in the proliferative zone could result from elevated cell death, a change in the balance between proliferation and differentiation, and/or a cell-cycle defect within the proliferative zone. We examined each of these possibilities. We stained late-L4 and young adult daf-2 mutants with SYTO-12 to mark apoptotic nuclei, and found that although all animals contained SYTO-12-positive cells in the loop region, as expected (Gumienny et al., 1999), none contained SYTO-12-positive cells in the distal germ line (n>80 each stage; data not shown). Therefore, inappropriate distal zone apoptosis does not account for the phenotype.

The distance from the distal tip to the transition zone is a measure of the effective `reach' of the DTC signal to deter meiotic entry, and it often correlates well with cell numbers (e.g. Eckmann et al., 2004). We examined this parameter in multiple reduction-of-function (rf) glp-1 and daf-2 mutants at the L4 stage. Consistent with previous observations (Hansen et al., 2004), the distance from the distal tip to the transition zone in glp-1(rf) mutants raised at a semi-permissive temperature was considerably reduced relative to that measured in wild type (12-13 cell diameters in the mutant versus 23 in wild type; Fig. 2A,C). By contrast, the distance to transition was only slightly reduced in three different daf-2(rf) alleles (19-21 cell diameters versus 23; Fig. 2A,C). The effects of both the glp-1 and the daf-2 mutations were statistically significant, so a minor role for daf-2 in preventing differentiation cannot be ruled out. However, the effect of daf-2 was much smaller than that of glp-1. These data also suggest that cell number and distance to transition do not always correlate well.

To determine whether the frequency of germ cell divisions was altered in daf-2(rf) and glp-1(rf) mutants, we measured the mitotic index. The germline mitotic index in daf-2 mutant larvae (but not in adult, data not shown) was reduced relative to that of wild type (Fig. 2B), consistent with our time-course analysis and with previous work (Pinkston et al., 2006). In striking contrast, the mitotic index of glp-1(rf) mutant larvae was not reduced relative to that of wild type (Fig. 2B,C). These results support the hypothesis that DAF-2/IIR signaling through DAF-16 primarily affects the cell cycle, whereas GLP-1/Notch signaling affects the mitosis/meiosis decision, rather than accelerating the larval germline cell cycle rate per se. If the effects of glp-1/Notch and daf-2/IIR on larval germline proliferation were indeed largely independent, reducing both should cause both phenotypes. Indeed, daf-2(e1370) glp-1(e2141) double mutants raised at a semi-permissive temperature for both alleles exhibited both the mitotic index and the meiotic entry defects, with no indication of synergy (Fig. 2A-C).

We further explored the effects of daf-2 on the cell cycle by examining S-phase index and DNA content. We compared the number of proliferative zone nuclei in S-phase in L4 wild-type and daf-2(e1370) worms. Although more than 50% of the nuclei labeled with EdU in both strains, the proportion of EdU-labeled nuclei was significantly lower in daf-2(e1370) worms compared with wild type (Fig. 2D; see also Fig. S1C in the supplementary material). We then examined the effects of daf-2(e1370) on the L4 germline cell cycle using a propidium iodide-based protocol (Fig. 2E). We observed a shift to higher haploid equivalents in germlines of daf-2 larvae relative to those of N2. In particular, the number of nuclei ∼2N was reduced, whereas the number ∼4N was elevated.

Taken together, several conclusions can be drawn from the results of the M and S phase indices and DNA content measurements. A decrease in both M and S phase indices suggests that the cells are cycling less frequently. Furthermore, because we see a reduction in the number of nuclei with ∼2N DNA content and a concomitant elevation in the number of nuclei with a ∼4N DNA content, we conclude that the delay is in G2.

To determine whether the conserved PI3K branch of the DAF-2 signaling pathway promotes larval germ cell proliferation, we examined epistasis relationships (Gil et al., 1999; Lin et al., 1997; Ogg et al., 1997; Ogg and Ruvkun, 1998; Rouault et al., 1999). If this pathway is responsible for the daf-2 germline phenotypes, reducing the activity of DAF-18/PTEN or DAF-16/FOXO should ameliorate the effect of reduced daf-2. We counted the number of adult proliferative zone germ cells in mutant/RNAi treatments that reduced the activity of both daf-2 and daf-18 or daf-16 (that is, daf-18 or daf-16 RNAi in the daf-2 mutant, daf-2 RNAi in the daf-18 or daf-16 mutants, or double mutants). In each case, reducing daf-18 or daf-16 restored normal germ cell numbers (Fig. 3A,B; see Table S3 in the supplementary material). Thus, the canonical pathway mediates the role of daf-2 in larval germline proliferation.

We then sought to determine whether the daf-2 pathway acts in the germ line or the soma to influence germline proliferation. Previous mosaic analysis indicated that for many phenotypes daf-2 acts cell non-autonomously, but with a possible germline-autonomous component for fertility (Apfeld and Kenyon, 1998). We first examined data from microarray studies in early embryos (Hill et al., 2000), and found that many DAF-2 pathway components (including akt-1, daf-18 and daf-16) are germline transcribed, as is hcf-1, a constitutively nuclear co-factor that inhibits DAF-16 largely independently of the insulin pathway (Li et al., 2008). Consistent with a daf-16-dependent positive role for HCF-1 in germline proliferation, hcf-1 mutants have a defect similar to that of daf-2 mutants (Fig. 2A,B; see also Table S3 in the supplementary material; Fig. 3B).

To investigate further the germline versus soma activity of pathway components, we tested the efficacy of RNAi knockdown of several pathway components in the rrf-1 mutant background that reduces RNAi effectiveness in the soma but retains it the germ line [rrf-1(pk1417) (Sijen et al., 2001)]. RNAi directed against genes encoding both positively and negatively acting components of the pathway, from daf-2 to daf-16, remains effective in the rrf-1 mutant (Fig. 3A,B; see also Table S3 in the supplementary material), suggesting that these components act in the germ line.

To confirm a germline requirement for DAF-16, we took two additional approaches: mosaic analysis and tissue-specific gene expression, using daf-16(mu86) suppression of the daf-2(e1370) germline proliferation defect as a read-out (Fig. 2C). Simple transgenic arrays are not normally expressed in the germ line (Kelly et al., 1997), but we found that two independent transgenic arrays, muEx108 (Lin et al., 2001) and naEx148 [based on Henderson and Johnson (Henderson and Johnson, 2001)], expressed DAF-16::GFP in the germ line (see Fig. S3 in the supplementary material) and reversed the daf-16(–) suppression of the daf-2 germline defect (Fig. 3C; see also Table S3 in the supplementary material), although GFP was not visible in the germ line. We speculate that only low levels of DAF-16 are required to rescue the germline phenotype, and that these are insufficient to visualize GFP. We then used the daf-16(mu86);daf-2(e1370); muEx108[daf-16(+)] strain for a simple germline/soma mosaic analysis (see Table S8 in the supplementary material). Briefly, we shifted individual animals of the genotype daf-16;daf-2;Ex[daf-16::GFP] and determined (1) the brood size and (2) whether the transgene passed through the germ line to the progeny. The results are consistent with a germline requirement for daf-16, but also indicate a possible somatic requirement. Specifically, 97% (475/487) of worms that transmitted the array to their progeny had small broods (equivalent to those of the daf-2(e1370) single mutant). Conversely, of those that failed to transmit the array, 74% (17/23) had brood sizes comparable to those of the daf-16;daf-2 double mutant. Therefore, loss of daf-16(+) from the germ line correlates with a considerable restoration of fecundity in daf-2(e1370).

Finally, we assayed DAF-16::GFP expression from heterologous promoters. Neuronal and intestinal DAF-16 expression largely reverses daf-16 mutant suppression of the daf-2 dauer and lifespan phenotypes, respectively (Libina et al., 2003) (see Table S4B in the supplementary material). Expression of DAF-16::GFP in neurons, intestine, or the DTC, did not reverse daf-16(–) suppression of the daf-2 germline defect, while expression from the myo-3 promoter (muscle/proximal sheath) somewhat reduced the number of proliferative germ cells (Fig. 3C; see Table S3 in the supplementary material), suggesting a minor contribution from myo-3-expressing tissues to germline proliferation (see also Table S8 in the supplementary material). Importantly, the number of proliferative germ cells was even lower upon germline-specific DAF-16::GFP expression (Prpl-11.1; Fig. 3C; see also Table S3 in the supplementary material) and was comparable to that in the daf-2 single mutant.

Taken together, our studies are consistent with a primary germline requirement for daf-2, daf-16, daf-18 and hcf-1 in controlling germline proliferation, with a possible additional minor contribution from the soma.

To determine which of the 40 putative insulin-like ligands (Li et al., 2003; Pierce et al., 2001) promote larval germline proliferation, we used a genetic assay based on the observation that a reduction of robust larval germline proliferation can enhance the penetrance of proximal germline tumor formation (Pro phenotype) in glp-1(Pro) mutants (Killian and Hubbard, 2004; Killian and Hubbard, 2005; McGovern et al., 2009) (see legend to Fig. S4 in the supplementary material for additional explanation). The Pro phenotype of two glp-1(Pro) mutants (ar202 and ar218) was enhanced by daf-2(e1370) (data not shown). We individually targeted each of the 40 predicted insulin genes by RNAi feeding and tested them for enhancement of the glp-1(ar202) Pro phenotype. Out of the 40 genes tested, two putative ligands, ins-3 and ins-33, enhanced the glp-1(ar202) mutant phenotype (see Fig. S4 in the supplementary material).

Further investigation substantiated the role of ins-3 and ins-33 in larval germline proliferation. Treatment with ins-3 or ins-33 RNAi significantly reduced the number of proliferative zone germ cells compared with L4440 or ins-1 RNAi negative controls in both males and hermaphrodites (Fig. 4; see also Table S3 in the supplementary material). Interestingly, RNAi targeting either ins-3 or ins-33 mRNA gave nearly identical results at all developmental stages, suggesting that they act together or in series to promote robust larval germline proliferation. Time-course analysis indicated that ins-3 or ins-33 RNAi defects (Fig. 1A, Fig. 4A) were less severe than those of daf-2 mutants shifted to the restrictive temperature, suggesting that additional redundantly-acting ligands might contribute. Results of timed shifts to and from RNAi bacteria indicate a requirement for ins-3 and ins-33 after the mid-L3 stage (see Table S7 in the supplementary material), consistent with that of daf-2. RNAi targeting ins-3 and ins-33 in rrf-1(pk1417) resulted in near-normal germline proliferation, suggesting that ins-3(+) and ins-33(+) are required in the soma (Fig. 4B). The canonical IIR pathway acts downstream of ins-3 and ins-33, as the phenotype was suppressed in daf-18 and daf-16 mutant backgrounds (Fig. 4B).

Deletion mutant alleles of ins-3 and ins-33 cause defects similar to those caused by RNAi with respect to enhancement of the glp-1(ar202) Pro phenotype and reduction of germ cell numbers (see Tables S3, S6 in the supplementary material). Their effects on germ cell number and mitotic index were daf-18 and daf-16 dependent (Fig. 4C). Reintroducing ins-3(+) and ins-33(+) into the respective mutant strains on simple transgenic arrays rescued these defects (see Table S3 in the supplementary material; Fig. 4D), verifying that the mutant phenotype is due to the deletion of these genes. Overexpression of ins-3 or ins-33 did not elevate the number of proliferative germ cells, suggesting that their activity might not be sufficient to drive the germline cell cycle (see Table S3 in the supplementary material). Alternatively, overexpression might not cause increased levels of active ligand because of post-transcriptional regulation (e.g. RNA stability, translation, or ligand processing, secretion or stability).

Surprisingly, the cumulative effect on the number of adult germ cells in the proliferative zone in each deletion mutant (ins-3 and ins-33) was less severe than that caused by the corresponding RNAi treatments, whereas the defect in larval mitotic index was similar to that caused by the RNAi. Genome scanning revealed no obvious off-target candidates (ins-3 matches 19 bp in T10G3.4, and ins-33 matches 21 bp in C44C10.2, a likely pseudogene). Furthermore, neither ins-3(RNAi) in ins-3(0) nor ins-33(RNAi) in the ins-33(0) mutant produced a more severe phenotype than did the mutants alone, suggesting that the mutations and RNAi are targeting identical genes (Fig. 4C). Possibly, reducing the activity of these genes may cause a stronger phenotype than removing them; the germ line may compensate for the loss (but not the reduction) of ins-3 or ins-33 over time. Moreover, neither mutant-RNAi combination (Fig. 4C) nor the ins-33;ins-3 double mutant (see Table S3 in the supplementary material) exacerbated the proliferation phenotype.

In summary, ins-3 and ins-33 act similarly, upstream of the daf-2 pathway, and largely account for the effects of daf-2-mediated signaling on the larval germline cell cycle.

Previous results indicated that the somatic gonadal sheath (especially the distal-most pair of sheath cells) is required for robust larval germline proliferation (Killian and Hubbard, 2005; McCarter et al., 1997). We considered the possibility that ins-3 or ins-33 could be the sheath signal. Using reporters, we did not detect ins-3 or ins-33 in the sheath/spermatheca lineage of the somatic gonad. Rather, ins-3 expression was largely neuronal, whereas ins-33 expression was largely hypodermal (see Fig. S6 in the supplementary material). However, it remains possible that expression in other tissues is below the level of detection.

To test directly whether IIR promotion of germline proliferation requires the gonadal sheath, regardless of the anatomical source of the ins-3 and ins-33 ligands, we ablated the sheath in the presence and absence of daf-16 activity. Because reduction of daf-16 restores germ cell proliferation even when daf-2 is reduced, we reasoned that, if the sheath cells mediate IIR signaling, then unless the sheath produces another essential factor for germline proliferation, the effects of ablating the sheath should be similarly abrogated when daf-16 is reduced. We ablated both SS cells (sheath/spermatheca precursors) and found that the ablations had similar effects in both daf-16(+) and daf-16(RNAi) animals (Fig. 5). Thus, unlike the defect in germline proliferation caused by reduced IIR signaling, the effects of ablating the sheath are not dependent on daf-16, which suggests that IIR signaling is not the sole essential sheath proliferation-promoting mechanism.

The anatomical focus of activity for IIR signaling for dauer control is largely neuronal (Apfeld and Kenyon, 1998; Wolkow et al., 2000), whereas the anatomical focus for lifespan is predominantly intestinal (Libina et al., 2003). Therefore, the different phenotypical effects of IIR pathway mutants might correlate with tissue-specificity of the IIR response. Another, not mutually exclusive, hypothesis is that different ligands elicit different responses – either by virtue of their source, sequence of action, or interactions with other ligands or co-receptors (Murphy, 2006). We tested both of these ideas. Because previous results connect IIR signaling components with germline proliferation control in dauer (Narbonne and Roy, 2006) and with reproductive timing (Dillin et al., 2002), and because additional studies link germline proliferation with lifespan, we particularly wished to establish whether the effect of ins-3, ins-33 and IIR signaling on larval germline proliferation is related to daf-2 activities associated with the dauer decision, reproductive timing, and lifespan regulation.

First, we tested whether reducing ins-3 or ins-33 could cause dauer under conditions that elicit dauer by daf-2(RNAi). We found that neither mutation nor RNAi treatment elevated the percentage of animals entering dauer at 27°C (see Table S4C in the supplementary material), suggesting that ins-3 and ins-33 do not act individually as daf-2 agonists for the dauer decision.

Next, we investigated whether loss of ins-3 or ins-33 prolongs the reproductive period. We speculated that the reproductive timing phenotype of daf-2 could be a consequence of reduced larval germline proliferation, consistent with a sensitive period for this reproductive phenotype during larval stages (Dillin et al., 2002). We measured progeny production in daf-2, ins-3 and ins-33 mutants in the presence or absence of daf-16. We found that unlike daf-2, neither ins-3 nor ins-33 mutants displayed delayed or prolonged reproduction (Fig. 6A). Therefore, extension of the reproductive period in daf-2 mutants is not likely to be a secondary consequence of reduced germline proliferation in larval stages, and these phenotypes are separable.

If ins-3 and ins-33 act globally to activate daf-2, their removal should cause similar changes in daf-16 activity in the intestine as does removal of daf-2. Similar to previous results (Libina et al., 2003; Lin et al., 2001), we found that reducing daf-2 activity by RNAi caused nuclear enrichment of DAF-16::GFP and induction of sod-3::GFP expression. By contrast, neither effect was observed when worms were treated with ins-3 or ins-33 RNAi (Fig. 6B). These results suggest that ins-3 and ins-33 do not act individually on the daf-2 pathway in the intestine.

Taken together with the results of temperature-shift experiments (Fig. 1A), these results are consistent with the hypothesis that the requirement for the IIR pathway regulating larval germline proliferation is temporally and anatomically distinct from its role in dauer and other phenotypes, and that different ligands preferentially regulate different daf-2-dependent processes.

We uncovered a role for the canonical DAF-2/IIR pathway in promoting robust larval germline proliferation, and found that this role is distinct from that of GLP-1/Notch. These results contrast with a previous study on the neuronal role of Notch in dauer recovery that suggested that IIR signaling may be epistatic to LIN-12/Notch activity (Ouellet et al., 2008).

We identified and characterized two putative DAF-2 agonists, ins-3 and ins-33, that are required in the soma to promote robust larval germline proliferation in a daf-18- and daf-16-dependent manner. Our data are consistent with a model in which activation of daf-2 signaling in response to specific ligands (ins-3, ins-33) at a distinct period (the L3 and L4) and in a distinct tissue (the germ line) underlies the effect of this pathway on larval germline proliferation. These findings suggest the existence of special regulatory controls at this crucial time of germline proliferation. In support of this model, we found that ins-3 and ins-33 do not individually affect the dauer decision, reproductive timing, or intestinal DAF-16 nuclear localization or sod-3 expression, that the effect of IIR signaling on the germline is after the dauer decision and before adulthood, and that its tissue requirement is distinct from its other roles (mainly germ line and neither neuronal nor intestinal).

ins-33 has been identified previously in two other contexts: as a direct target of the lin-14 transcriptional repressor in the L1 (Hristova et al., 2005); and as a target of TGFβ signaling by microarray analysis. Mutations in lin-14 cause precocious dauer programs, suggesting a possible role for ins-33 in dauer timing or development. However, neither the results of Hristova et al. nor those of this study indicate a dauer role for ins-33 alone. Liu et al. found that ins-33 was upregulated together with ins-18, a gene proposed to act as an antagonist of daf-2 signaling (Liu et al., 2004). Our studies, however, implicate ins-33 as an agonist. It is possible that ins-33 has different roles in different developmental times and contexts.

Reducing the activity of either ins-3 or ins-33 causes very similar germline proliferation phenotypes, and reducing both does not further exacerbate the phenotype. One interpretation of these results is that the two function together. Alternatively, they could act in series, triggering successive signaling pathways. However, exclusive DAF-2 pathway feedback is not easily reconciled with a predominantly germline-autonomous role. Alternatively, these two ligands could bind non-DAF-2 receptors and act in a signal relay. The family of insulin-related ligands in humans, some of which do not bind the insulin receptor or IGFRs [e.g. relaxins (Bathgate et al., 2005)], are structurally related to only one class of the putative C. elegans insulin-like peptides (Pierce et al., 2001). Also, a bioinformatics study suggested the existence of C. elegans genes with possible similarity to the extracellular domain of IIRs (Dlakic, 2002). Although there may be considerable redundancy among the 40 insulin-like peptides in the C. elegans genome, it will be interesting to resolve their roles in relation to daf-2 activity. Regardless of the precise mechanism, taken together with previous studies (Apfeld and Kenyon, 1998; Libina et al., 2003; Murphy et al., 2007; Wolkow et al., 2000), our data support a model in which different phenotypes are associated with specific ligands and target tissues.

Previous studies implicate daf-2 pathway components in other aspects of germline proliferation. Narbonne and Roy showed that DAF-2/IIR and DAF-7/TGFβ converge on DAF-18/PTEN to prevent inappropriate germline proliferation under dauer-inducing conditions (Narbonne and Roy, 2006). Fukuyama et al. demonstrated daf-18-dependent germline cell cycle arrest in the starvation-induced L1 diapause (Fukuyama et al., 2006). In both cases, cell cycle arrest is daf-16 independent and occurs in G2 (Fukuyama et al., 2006; Narbonne and Roy, 2006). We found that under non-starvation conditions, the effect of reducing IIR signaling on larval germline proliferation is dependent on both daf-18 and daf-16, and that it also affects the G2.

Our studies suggest that the DAF-2 pathway acts independently of both GLP-1/Notch and the sheath to promote robust larval germline proliferation. Because GLP-1 signaling is required to maintain germ cells in an undifferentiated (mitotic) state as opposed to a differentiated (meiotic) state (Austin and Kimble, 1987), it is difficult to separate its contribution to cell cycle from its role in the mitosis/meiosis decision. Our results suggest that the role of GLP-1 in promoting larval germline proliferation is primarily to prevent differentiation. Although we found that reducing glp-1 activity did not reduce the mitotic index of remaining cycling larval germ cells, elevating GLP-1 signaling elevates the mitotic index of germ cells that are cycling in adults (Berry et al., 1997; Maciejowski et al., 2006). Therefore it is likely, as suggested by Berry et al. (Berry et al., 1997), that positive feedback exists between mitotically active cells and glp-1. Moreover, additional cell cycle controls are likely to act in conjunction with GLP-1 in the adult.

Recent studies suggest a role for IIR signaling in tumor growth. In C. elegans, Pinkston et al. observed that reducing daf-2 activity lowered the mitotic index in germline tumors in adult gld-1-mutants (Pinkston et al., 2006). Strikingly, mitotic index was reduced within the tumor, but not in the distal proliferative zone. Our data are consistent with these results, as we do not observe a defect in adult germline proliferation in daf-2 mutants. Taken together with the Pinkston et al. study (Pinkston et al., 2006), our results suggest that germline tumor growth control in gld-1 mutants may more closely resemble that of the larval germ line.

Our data further suggest that a role for IIR in nutrition-sensitive cell proliferation control might be widely conserved. In mammals, IIR signaling through PTEN can influence tumor growth sensitivity to short-term dietary restriction (Kalaany and Sabatini, 2009). In Drosophila, insulin-FOXO signaling controls germline stem cell proliferation in response to rich dietary conditions. Similar to our findings, receptor signaling is required in the germ line and impinges on G2 (Drummond-Barbosa and Spradling, 2001; LaFever and Drummond-Barbosa, 2005; Hsu et al., 2008). These parallels suggest that IIR signaling may be a conserved mechanism to tie nutrition to cell cycle control. We speculate that C. elegans larvae, having embarked on a reproductive (as opposed to dauer) developmental path, make an assessment of nutritional sufficiency, and that this information influences germline proliferation. It will be of interest to further test this hypothesis, as it would provide a simple model system in which to study the link between nutrition, metabolism and cell proliferation.

We thank Tim Schedl, Paul Fox, Ann Wehman, Jeremy Nance, the C. elegans Gene Knockout Consortium (Oklahoma City and Vancouver, Canada), the Caenorhabditis Genetics Center (CGC, Minneapolis, MN; funded by NIH NCRR), and the National Bioresource Project for the Nematode C. elegans (Tokyo, Japan) for strains and/or advice; Cathy Wolkow, Monica Driscoll, Michael Glotzer, and Thomas Johnson for plasmids; Shai Shaham for help with expression pattern analysis; Bernard Lakowski and Dan Tranchina for help with statistics; and John Maciejowski for excellent technical assistance. We especially thank Roumen Voutev for input on many aspects of the work and all members of Hubbard and Nance labs for helpful discussions. Funding was provided by NIH grant R01GM61706 to E.J.A.H. Deposited in PMC for release after 12 months.