Niche cells regulate primordial germ cell quiescence in response to basement membrane signaling

Stem cells must balance proliferation and differentiation in order to meet the developmental and regenerative needs of an organism. To accomplish such precise regulation, stem cells rely on information conveyed from a surrounding micro-environment called the niche. The niche is composed of both cells and extracellular matrix (ECM) (Jones and Wagers, 2008; Scadden, 2014). Niche cells and ECM can function by interacting directly with stem cells. In addition, ECM can impact stem cells indirectly, either by acting through niche cells or by modulating the local signaling environment. Indirect control of stem cells by niche ECM adds an additional, underappreciated layer of complexity to niche function.

Basement membrane (BM) is one common type of extracellular matrix contributing to stem cell niches (Kruegel and Miosge, 2010). For example, mammalian muscle, epidermal, neuronal, hematopoietic and intestinal stem cells are all closely juxtaposed to BMs (Gattazzo et al., 2014). BMs are thin flexible sheets of extracellular matrix that include four core components conserved in all bilaterian animals: sheet-like layers of laminin and collagen IV; the proteoglycan perlecan; and the crosslinking protein nidogen (Pastor-Pareja, 2020; Yurchenco, 2011). These four components form a matrix (the LNCP matrix) that cells sense using cell-surface receptors, including integrin, dystroglycan and glypican (Yurchenco, 2011). Numerous additional proteins associate with the core BM matrix, altering its composition in different regions of the organism and during various developmental stages (Jayadev et al., 2019; Yurchenco, 2011).

Although BMs function as physical barriers that support and separate adjacent tissues, they also provide crucial signaling inputs to surrounding cells. Core BM proteins can signal to adjacent cells by interacting with cell-surface BM receptors. For example, collagen IV and perlecan contain Arg-Gly-Asp (RGD) motifs that can activate a specific class of RGD-binding integrin receptors (Ludwig et al., 2021). Other types of integrins bind the laminin G-like (LG) domains present in laminin α subunits (Yurchenco, 2011). Such interactions can relay both biochemical and mechanical properties of the BM to cells. In addition, BM proteins can affect local signaling indirectly by binding to a wide variety of growth factors. For example, in addition to binding to integrin directly through its RGD motifs, perlecan binds directly to FGF, Wnt and TGFβ ligands, affecting output from these signaling pathways in nearby cells by either concentrating or sequestering ligands (Hayes et al., 2022). It is clear from a variety of model systems that the signaling capacity of BMs is essential for stem cell niche function. For example, BM signaling ensures self-renewal of embryonic stem cells and hair follicle stem cells in mammals, and follicular stem cells in the fly ovary (Fujiwara et al., 2011; O'Reilly et al., 2008; Suh and Han, 2011). Given that integrin signaling also controls proliferation in many types of cancer (Cooper and Giancotti, 2019), one possibility is that dysfunctional tumor BM could employ similar signaling mechanisms to regulate cancer stem cells.

One of the most important niche functions is balancing stem cell quiescence and proliferation. Failure to properly regulate proliferation can lead to developmental defects and contributes to aging and cancer growth (Kai et al., 2019; Oh et al., 2014). This regulation often occurs through cooperation between niche BM and growth factor signaling ligands. For example, proliferation of mouse neural stem cells is controlled by perlecan, which binds growth factors such as FGF2 (Kerever et al., 2014, 2007; Sirko et al., 2007). Alternatively, transduction of BM signals by integrin receptors can impact stem cell proliferation. For example, in mouse muscle stem cells β1 integrin potentiates FGF2 signaling, which drives expansion of muscle stem cells (Rozo et al., 2016). Indeed, β1 integrin regulates proliferation across multiple mammalian tissues, including the skin, mammary and intestinal epithelium as well as neural stem cells in the subventricular zone (Faraldo et al., 2001; Jones et al., 2006; Raghavan et al., 2000; Shen et al., 2008). Integrin receptors also regulate stem cell proliferation in flies, often by cooperating with other signaling pathways, as reported for intestinal stem cells, follicular stem cells and blood progenitors (Khadilkar et al., 2020; Lin et al., 2013; O'Reilly et al., 2008). Thus, niche BM, and integrin signal transduction in particular, are important regulators of stem cell proliferation.

In principle, information from the BM can influence stem cell behavior in two ways. First, signals may flow directly from the BM to stem cells, either by BM receptors on stem cells contacting the matrix or by modulation of secreted signals by the BM. Alternatively, niche cells can interact with the BM and relay information to stem cells. In both the Caenorhabditis elegans and Drosophila melanogaster germ lines, niche cells act as intermediaries between stem cells and BM. In fly ovaries, formation of the gonadal BM is needed for normal niche cell morphology and for BMP signaling between niche and stem cells. Disrupting collagen IV deposition results in ovaries with too many germline stem cells (Diaz-Torres et al., 2021; Van De Bor et al., 2015). Similarly, in the C. elegans embryonic gonad primordium, we showed previously that depleting laminin severely disrupts niche morphology and induces germ cell proliferation (McIntyre and Nance, 2020). The heparan sulfate proteoglycan syndecan also functions in C. elegans gonadal niche cells to modulate germline stem cell proliferation (Gopal et al., 2021). However, in these systems the mechanisms niche cells use to convey information from the BM to stem cells remain largely unknown.

In this study, we investigate how niche BM and niche cells in the C. elegans gonad primordium collaborate to control quiescence of primordial germ cells (PGCs), which are the direct precursors of germline stem cells. The gonad primordium assembles during embryogenesis when two somatic gonadal precursor cells (SGPs), which function as niche cells, enwrap the two PGCs to form a four-cell gonad primordium (Fig. 1A) (Rohrschneider and Nance, 2013; Sulston et al., 1983). Soon after the gonad primordium assembles, SGPs nucleate a BM that surrounds the gonad primordium (Graham et al., 1997; Guo et al., 1991; Huang et al., 2003; Kao et al., 2006; Kang and Kramer, 2000; McIntyre and Nance, 2020; Sibley et al., 1993). Embryonic PGCs are quiescent (arrested in G2 phase) and do not re-enter the cell cycle unless the hatched larva encounters food (Fukuyama et al., 2006). We showed previously that depleting laminin from the gonadal BM activates Notch signaling in PGCs, causing them to exit from quiescence precociously and proliferate in the embryo (McIntyre and Nance, 2020). However, it remained unclear from that study how BM signaled to the PGCs. Here, we identify BM proteins that are required for regulating PGC quiescence, show that these signals are relayed to PGCs by gonadal niche cells using integrin receptors, and provide evidence that BM signals continue to regulate germ cell proliferation in larvae under physiological conditions. Our data support a model wherein niche cells induce stem cells to exit from quiescence and proliferate in response to signals from the gonadal BM.

Laminin, which is a heterotrimer composed of α, β and γ subunits, can be removed by inhibiting the sole laminin β-encoding gene lam-1 (Kang and Kramer, 2000). Eliminating laminin using a lam-1 null mutation causes only a fraction of PGCs to exit quiescence, and those that do typically divide only once (McIntyre and Nance, 2020). One potential explanation for this finding is that the gonadal BM regulates only the initial escape from PGC quiescence, after which point another pathway controls continued proliferation. Alternatively, given that lam-1 embryos arrest with major morphogenetic defects (Kang and Kramer, 2000), it is possible that the attendant deterioration of mutant embryos prevents additional rounds of PGC proliferation. To distinguish between these possibilities, we inhibited the laminin α gene epi-1, which encodes the primary laminin α detected in the gonadal BM (Fig. 1B) (Huang et al., 2003; McIntyre and Nance, 2020). epi-1 functions redundantly with the only other laminin α gene (lam-3) in embryonic development (Huang et al., 2003); consequently, in contrast to lam-1 mutant embryos, most epi-1(RNAi) embryos are able to hatch (95/103 embryos). epi-1(RNAi) L1 larva starved for 2 or 3 days showed significantly greater PGC proliferation than those starved for only 1 day (Fig. 1C), and ∼20% of larvae starved for 3 days contained more than four PGCs (up to ten PGCs; Fig. 1D,F). We observed comparable results using the epi-1(xn200) putative null mutation, which we created by inserting a stop cassette into the second exon of an endogenously tagged epi-1::mNG allele (Fig. S1A). Almost all epi-1(xn200) mutants died as embryos or young larva (Fig. S1D-F), and no EPI-1^mNG was visible in mutant animals (Fig. S1B,C). PGCs in starved epi-1(xn200) L1 larvae underwent a similar number of divisions as those in starved epi-1(RNAi) L1 larvae (Fig. 1C,D,G). We conclude that depleting laminin from the gonadal BM induces PGCs to exit from quiescence and to continue dividing, perhaps until they can no longer do so because of nutrient limitation.

SGP cell membranes enwrap the PGCs, making it unlikely that PGCs make significant contact with the gonadal BM (McIntyre and Nance, 2020; Rohrschneider and Nance, 2013). In both confocal reconstructions of the L1 gonad (Fig. 1B) and published transmission electron micrographs (Sulston et al., 1983), PGCs are separated from the BM by niche cell membranes. Therefore, we hypothesized that SGPs relay laminin-dependent, quiescence-regulating signals from the basement membrane to the PGCs. We tested this hypothesis using laser microsurgery to kill the SGP ancestors combined with a transgene, sygl-1p::GFP-H2B, that reports on Notch signaling activity in the PGCs (Kershner et al., 2014); because sygl-1p::GFP-H2B is activated in PGCs when laminin is depleted, and glp-1/Notch is required for PGC proliferation under these conditions, sygl-1p::GFP-H2B expression in PGCs is an indicator of exit from quiescence (McIntyre and Nance, 2020). In contrast to mock-ablated lam-1(RNAi) embryos, which showed robust activation of sygl-1p::GFP-H2B, PGCs in ablated lam-1(RNAi) embryos lacking SGPs expressed very low levels of sygl-1p::GFP-H2B (Fig. 1H-K). Although we do not know the cause of the low level of sygl-1p::GFP-H2B activation in lam-1(RNAi) embryos lacking SGPs (compared with mock control, P=0.004), we speculate that it could arise if other cells contributed in minor ways to PGC Notch activation. These results demonstrate that SGPs are required for robust activation of Notch signaling in PGCs when laminin is depleted, and are consistent with classic laser-ablation experiments showing that SGPs are needed for PGCs to proliferate in fed larva (Kimble and White, 1981). Based on these findings, we propose that SGPs relay signals from the BM to the PGCs, regulating their exit from quiescence (Fig. 1A).

We hypothesized that SGPs express a receptor that transduces BM signals. Integrin BM receptors function as heterodimers comprised of α and β subunits. C. elegans has two integrin α genes (pat-2 and ina-1) and a single integrin β gene (pat-3) (Baum and Garriga, 1997; Gettner et al., 1995; Williams and Waterston, 1994). SGPs profiled by single-cell RNA sequencing contain transcripts for all three genes (Packer et al., 2019), indicating that SGPs could express the α/β integrin heterodimers INA-1/PAT-3 and PAT-2/PAT-3 at their surfaces. To confirm integrin protein localization and assess its function in the SGPs, we tagged the endogenous pat-3 gene with sequences encoding GFP and the ZF1 degron, which can be used to conditionally deplete proteins [pat-3(xn109[pat-3::gfp::zf1]); Fig. S2A] (Armenti et al., 2014). In addition to robust expression in other tissues contacting the BM, including the pharyngeal primordium and body-wall muscles, PAT-3^GFP-ZF1 was present on SGP cell membranes in both embryonic and L1 gonad primordia (Fig. 2A-B′). PAT-3^GFP-ZF1 remained on SGP cell surfaces in epi-1(RNAi) embryos, although at slightly reduced levels (Fig. 2G-I), indicating that integrin is appropriately positioned to transduce signals from the BM to SGPs following laminin depletion. We also examined the localization of both α integrin subunits using alleles endogenously tagged with mNeonGreen (Jayadev et al., 2019). INA-1^mNG and PAT-2^mNG are expressed by SGPs in both embryos and L1 larvae (Fig. 2C-F′). Whereas INA-1^mNG appeared fairly evenly distributed on SGP surfaces, like PAT-3^GFP-ZF1, PAT-2^mNG showed a more punctate distribution. Thus, SGPs contain both INA-1/PAT-3 and PAT-2/PAT-3 integrins at their surfaces, and the two heterodimers appear to have unique localization patterns.

To examine integrin function in PGC quiescence, we engineered a putative null allele of pat-3 by inserting a stop cassette into the pat-3(xn109[pat-3::gfp::zf1]) allele upstream of the sequence encoding the transmembrane domain [pat-3(xn113); Fig. S2B]. Like existing strong pat-3 mutations (Williams and Waterston, 1994), pat-3(xn113) embryos arrested at the two-fold stage of embryogenesis (54/55 embryos). We first assessed whether pat-3 mutant embryos, like those depleted of laminin, also showed inappropriate PGC proliferation. Contrary to our expectation if laminin signaled through integrin receptors, PGCs did not proliferate in starved pat-3 mutants (n=83/83≤2 PGCs). Existing alleles of the integrin α genes ina-1 and pat-2, which are likely null mutations, also caused no PGC proliferation phenotype [ina-1(gm86) n=59/59≤2 PGCs; pat-2(ok2148) n=33/33≤2 PGCs]. These experiments indicate that laminin does not inhibit PGC proliferation by engaging integrin receptors on SGPs.

We next considered an alternative hypothesis: laminin could regulate PGC quiescence by preventing activation of integrin receptors, for example by sequestering or inhibiting an integrin ligand. In this case, loss of integrin should suppress the PGC proliferation that occurs when epi-1 is depleted. Consistent with this model, we found that most PGCs did not proliferate in pat-3(xn113) epi-1(RNAi) mutants (Fig. 2J). One possible confounding explanation for these results is that the twofold arrest of pat-3 mutant embryos may not be conducive for PGC proliferation. To address this possibility, we examined myo-3(st386) mutants, which lack functional muscle myosin and also arrest at the two-fold stage (Williams and Waterston, 1994). myo-3(st386) mutants only modestly suppressed PGC proliferation in epi-1(RNAi) embryos (Fig. 2K). Because the mild suppression was equivalent in myo-3(st386) mutants that did not arrest because they carried a myo-3(+) extrachromosomal array (Fig. 2K), this small effect is likely explained by genetic background and is not a result of failure in embryo elongation. These findings show that integrin receptors regulate PGC proliferation.

To determine which integrin heterodimer transduces the BM signal, we depleted epi-1 in ina-1 and pat-2 mutants. Whereas PGC proliferation in ina-1 epi-1(RNAi) mutant larvae was similar to that in epi-1(RNAi) larvae, almost no PGC proliferation occurred in pat-2 epi-1(RNAi) mutants (Fig. 2J). Together, these findings suggest that EPI-1 laminin trimers in the BM normally inhibit PGC proliferation by preventing another BM component from activating PAT-2/PAT-3 integrin receptors. This model also agrees with the ligand specificities for the two α integrin subunits in C. elegans: INA-1/PAT-3 dimers are predicted to bind to laminin, which is depleted in these experiments, whereas PAT-2/PAT-3 dimers are predicted to bind Arg-Gly-Asp (RGD) sequences present in other BM proteins, such as collagen IV and perlecan (Gettner et al., 1995; Jayadev et al., 2019; Williams and Waterston, 1994).

We tested whether integrin function is required in SGPs by utilizing the ZF1 degron we engineered into the pat-3(xn109[pat-3::gfp::zf1]) allele. Proteins tagged with the ZF1 domain can be rapidly degraded by conditionally expressing the ZIF-1 E3 ligase substrate adaptor (Abrams and Nance, 2021; Armenti et al., 2014). To achieve SGP-specific degradation of PAT-3^GFP-ZF1, we expressed ZIF-1 from the ehn-3 promoter (ZIF-1^SGP), which is active only in SGPs from the time of gonad formation to late embryogenesis (Mathies et al., 2003). We tested the specificity of ZIF-1^SGP by examining ubiquitously expressed transgenic CDC-42^GFP-ZF1, which was specifically degraded in SGPs (Fig. S2C). Alone, pat-3^GFP-ZF1 did not suppress PGC proliferation induced by epi-1 RNAi (Fig. 2L). However, when pat-3^GFP-ZF1 was combined with ZIF-1^SGP in epi-1(RNAi) animals, PGC proliferation was suppressed to a similar extent as in epi-1(RNAi) pat-3(xn113) null mutants (Fig. 2L). Because pat-3^GFP-ZF1; zif-1^SGP animals do not arrest during embryogenesis like pat-3 and pat-2 mutants, this experiment further demonstrates that the suppression of PGC proliferation in epi-1(RNAi) larvae is not an indirect consequence of embryonic arrest, but rather results from specific loss of integrin function in the SGPs. Together, these findings indicate that SGP integrins are required to transduce quiescence-regulating signals from the BM to the PGCs.

The experiments above raise the possibility that a BM protein other than laminin activates PAT-2/3 integrin receptors on SGPs when epi-1 is depleted. To determine whether any of the other three core BM proteins in the LNCP matrix (nidogen, collagen IV and perlecan) could function as an integrin ligand to relay quiescence signals to the PGCs, we first examined their expression in the gonadal BM. Using a recently described collection of BM genes endogenously tagged to produce mNeonGreen (mNG) fusion proteins (Keeley et al., 2020), we imaged tagged laminin α (EPI-1^mNG), the sole nidogen (NID-1^mNG), one of the two subunits of collagen IV (α1 EMB-9^mNG; the α2 subunit is LET-2) and the sole perlecan (UNC-52^mNG) in live embryos during gonad formation. Because each of these proteins is fused with the same fluorescent protein, differences in their order of appearance likely reflect real properties of the gonadal BM. LNCP matrix proteins appeared to be added to the gonadal BM in three waves. Laminin and nidogen assembled first, surrounding the outer surfaces of the SGPs once they began wrapping the PGCs (Fig. 3B-I; yellow rectangles highlight gonadal BM accumulation). Collagen IV became visible in the gonadal BM ∼60 min after laminin and nidogen first appeared (Fig. 3J-M; yellow rectangles highlight accumulation beginning at the 1.5-fold stage). Finally, perlecan accumulated approximately 30 min later (Fig. 3N-Q; yellow rectangle highlights accumulation beginning at the 2-fold stage). This ordered accumulation (L+N>C>P; Fig. 3A) mirrors previous findings in other C. elegans BMs (Keeley et al., 2020), and demonstrates that all parts of the LNCP matrix are present in the gonadal BM by the twofold stage and thus could contribute to PGC proliferation.

We hypothesized that another BM component activates integrin receptors in SGPs when epi-1 is depleted. As a first test to determine which BM components could activate integrin in the SGPs following laminin knock down, we assayed whether type IV collagen, nidogen and perlecan were still present in the gonadal BM in epi-1(RNAi) L1 larvae. First, we examined collagen IV α1/EMB-9^mNG localization. In newly hatched, starved L1 larvae, collagen IV was present in the gonadal BM of both control and epi-1(RNAi) larva (Fig. 4A,B). Although collagen IV localization in epi-1(RNAi) L1 larva was discontinuous, quantification of fluorescence intensity showed no significant change in the amount of collagen IV present in the gonadal BM adjacent to SGPs compared with wild type (Fig. 4C). In epi-1(RNAi) larva, nidogen/NID-1^mNG also localized to the gonadal BM at levels similar to control larva and showed discontinuities similar to those we observed for collagen IV (Fig. 4D-F). Finally, we observed that in wild-type, newly hatched, starved L1 larva, perlecan/UNC-52^mNG surrounded the gonad primordium and was also present at low levels in the extracellular space between the PGCs (Fig. 4G, arrowhead). In epi-1(RNAi) larvae, perlecan still localized robustly to the gonadal BM at comparable levels to wild-type larvae (Fig. 4H,I). We conclude that epi-1 RNAi does not prevent accumulation of type IV collagen, nidogen or perlecan and that these components could engage integrin receptors when laminin is knocked down.

Both perlecan and collagen IV are ligands for RGD integrin receptors. Perlecan/UNC-52 is a ligand for PAT-2/PAT-3 integrin receptors in C. elegans body-wall muscle cells, and unc-52 mutants, like pat-2 and pat-3 mutants, arrest at the twofold stage (Gettner et al., 1995; Guo et al., 1991; Rogalski et al., 1993; Williams and Waterston, 1994). Likewise, collagen IV is recruited to pharyngeal BM by PAT-2/PAT-3 integrins (Jayadev et al., 2019). We hypothesized that either perlecan or collagen IV would be required for PGC proliferation when laminin is depleted. To test whether perlecan is required for PGC proliferation in laminin-depleted embryos, we examined unc-52(st549) mutants; unc-52(st549) contains a premature stop codon predicted to severely truncate the UNC-52 protein (Fig. 4J) (Williams and Waterston, 1994). Compared with control epi-1(RNAi) worms, PGC proliferation was suppressed in starved unc-52(st549) epi-1(RNAi) mutants (Fig. 4K) to nearly the same extent as pat-3 epi-1(RNAi) mutants (see Fig. 2F,J). Next, we investigated whether the collagen IV matrix is required for PGC proliferation by examining loss-of-function mutations of both of the collagen IV subunits: an engineered putative null allele of emb-9 (xn201; Fig. S3) and the let-2(b246) temperature-sensitive lethal allele (Wood et al., 1980). Starved emb-9 epi-1(RNAi) larvae had significantly fewer PGCs compared with epi-1(RNAi) alone (Fig. 4L). Likewise, let-2(ts) epi-1(RNAi) larva had significantly fewer PGCs when raised at the restrictive temperature (25°C) compared with larva raised at a permissive temperature (20°C) (Fig. 4L). However, in contrast to removing integrin (pat-2 or pat-3) or perlecan (unc-52), loss of collagen IV only partially suppressed PGC proliferation. Together, these results indicate that perlecan plays an essential role in promoting PGC exit from quiescence when laminin is depleted, whereas collagen IV contributes to PGC quiescence but plays a modifying role.

Collagen IV can bind to perlecan, leading us to speculate that the weaker suppression phenotype of mutations disrupting collagen IV (emb-9 and let-2) compared with those in perlecan (unc-52) could result from reduced levels of perlecan in the gonadal BM of collagen mutants. To test this hypothesis, we examined UNC-52^mNG levels in the gonadal BM of emb-9 larvae (Fig. 4I,M). In contrast to control and epi-1(RNAi) larvae, which contained similar levels of UNC-52^mNG in the gonadal BM, levels of UNC-52^mNG in emb-9 mutant larvae were moderately reduced. This finding suggests that collagen could contribute to PGC quiescence by promoting perlecan enrichment in the gonadal BM, although we cannot exclude a signaling role for collagen IV.

Because PAT-2/PAT-3 integrins are thought to interact with RGD peptides in target ligands, we tested whether a putative integrin-binding RGD sequence within UNC-52 is important for regulating PGC quiescence. Targeted deletion of the second of two RGD sequences, within domain IV of UNC-52 [unc-52(kq2020); Fig. 4J] causes embryonic arrest at the twofold stage (Qiu et al., 2022). This phenotype is shared with unc-52 and pat-3 putative null mutants, suggesting that the domain IV RGD functions as an essential integrin ligand. Deleting the domain IV RGD suppressed PGC proliferation in epi-1(RNAi) animals to a similar extent as the unc-52(st549) null mutation (Fig. 4K). Because previous antibody staining experiments indicate that UNC-52 protein is still produced and localizes in unc-52(kq2020) mutants (Qiu et al., 2022), these experiments suggest that the domain IV RGD of UNC-52 may function as an integrin-binding motif required to trigger PGC proliferation when laminin is depleted.

The data presented so far demonstrate that BM signaling can operate under conditions of starvation to regulate PGC quiescence. To test whether BM signaling also regulates germ cell proliferation under physiological conditions, we depleted unc-52 in fed larva and examined germ cell number. Because unc-52 null mutants are embryonic lethal, we examined germ cell proliferation in three different unc-52 mutants (e444, e669 and e998; Fig. 5A), which, because of alternative splicing, affect larval and adult but not embryonic unc-52 function (Brenner, 1974; Gilchrist and Moerman, 1992; Lundquist et al., 1996; Mullen et al., 1999; Rogalski et al., 1995); as a result, mutant animals develop and move normally as embryos and young larvae, but become severely paralyzed as adults. These unc-52 mutants were previously reported to have reduced brood sizes (Gilchrist and Moerman, 1992), a finding that we confirmed (Fig. 5B).

We imaged the adult germ line in unc-52(e444) mutants and found that the germ line itself was greatly reduced in size (Fig. 5C,D), suggesting that perlecan might promote germline development. Because the severe paralysis of unc-52 mutant adults could affect feeding and contribute to their smaller germ line, we measured proliferation in 1-day-old larva, when unc-52 mutants move and feed normally, and do not differ in size compared with wild-type larvae (Mackenzie et al., 1978; Waterston et al., 1980). All three unc-52 mutants had significantly fewer germ cells compared with wild-type controls (Fig. 5E-G). These findings support a model in which perlecan continues to promote germ cell proliferation under physiological conditions as the larval gonad grows.

Many studies have demonstrated that BM signals are essential for proper regulation of stem cell survival, self-renewal and differentiation (Fujiwara et al., 2011; O'Reilly et al., 2008; Suh and Han, 2011). However, BMs are dynamic, heterogeneous structures the composition of which varies, not just across time and space, but between healthy and diseased states (Ge et al., 2020; Kai et al., 2019; Keeley et al., 2020; Tsutsui et al., 2021). Comparatively little is known about how cells respond to changes in the composition of adjacent BMs, nor about how such changes are regulated during development. The data we have presented show that depleting laminin changes the composition, and ultimately the function, of the gonadal BM. Relying on endogenously tagged BM components and engineered mutant alleles, we find that knocking down epi-1 does not prevent assembly of collagen IV, nidogen or perlecan to the gonadal BM. Perlecan and collagen IV are in fact required for PGCs to exit quiescence when epi-1 is knocked down. Under this model, niche cells act as intermediaries that convey information from the BM to the PGCs (Fig. 6A). Laminin in the gonadal BM normally prevents perlecan from engaging integrin receptors in SGP niche cells (Fig. 6B). When this inhibition is relieved, perlecan activates integrin receptors on SGP niche cells (Fig. 6C), which then induce PGCs to exit from quiescence. Given that Notch is required for PGCs to exit quiescence (McIntyre and Nance, 2020), and laminin depletion activates Notch in PGCs, we consider it likely that integrin signaling functions by activating a Notch ligand in the SGPs. However, it remains possible that BM signaling could also regulate PGC quiescence via a parallel pathway. Finally, we demonstrate that perlecan promotes germ cell proliferation in fed larva. Although we do not yet know whether BM signaling operates through similar or distinct pathways in embryos and larvae, these findings support the conclusion that BM signaling regulates proliferation under physiological conditions.

There are several possible ways that altering gonadal BM composition could affect niche function. One way this may occur is if changes in the mechanical properties of the BM facilitate perlecan interactions with integrin. In both developmental and cancer models, increasing the amount of collagen IV relative to laminin stiffens BM and can drive cell proliferation (Kai et al., 2019). Alternatively, the BM could be so compromised in epi-1(RNAi) larvae that normal mechanical cues are lost. Similar mechanisms have been reported in mammalian cells, where integrin αV signaling is needed to maintain the quiescence of hematopoietic stem cells in the bone marrow and spleen, and integrin-mediated adhesion to osteopontin supports quiescence of disseminated acute lymphoblastic leukemia cells (Boyerinas et al., 2013; Horton et al., 2020; Khurana et al., 2016; Mehatre et al., 2021). A distinct possibility is that laminin knockdown liberates perlecan from the BM. In the Drosophila ovary, perlecan functions both within the BM and the interstitial matrix to maintain the germline stem cell pool (Diaz-Torres et al., 2021). We observed interstitial perlecan within the gonad in laminin-depleted worms, raising the possibility that BM-free perlecan could contribute to proliferation. Finally, altering niche BM composition could change the way that perlecan interacts with one of the numerous growth factors that it binds. Often this cooperation controls stem cell proliferation. For example, perlecan cooperates with FGF to activate signaling via the Ras/ERK or PI3K/AKT pathways, leading to increased proliferation of mammalian chondrocytes and neural stem cells (Ghiselli et al., 2001; Kerever et al., 2014, 2007; Patel et al., 2007; Smith et al., 2007a,b; Vincent et al., 2007). Changes to BM composition therefore could impact the availability of growth factors in the niche, leading to PGC proliferation. Given that perlecan interacts with Wnt and TGFβ signals during cell migration as the gonad grows in larva (Merz et al., 2003), this mechanism could explain the ability of perlecan to trigger PGC proliferation when laminin is knocked down.

Studies in other systems have indicated that integrin activity is used to tune Notch signaling to the local ECM environment. For example, integrin promotes Notch signaling in order to control endothelial-to-mesenchymal transition following blood vessel damage in human cells (Wang et al., 2018b), differentiation of hematopoietic stem cells in mice (Hadland et al., 2022), self-renewal of mouse embryonic stem cells (Suh and Han, 2011) and proliferation of mouse neural stem cells (Androutsellis-Theotokis et al., 2006; Rosa et al., 2016; Shen et al., 2004). Conversely, activation of integrin by extracellular MAGP2 (MFAP5) negatively regulates Notch in human endothelial cells (Deford et al., 2016). In the worm gonad primordium, integrin may also coordinate Notch signaling with the local ECM environment. However, modulation of Notch signaling in the gonadal niche differs from these examples because integrin signaling in SGPs, if it impinges on Notch activity in PGCs as predicted, must do so non-autonomously. How this occurs will be an important area of future study. For example, integrin signaling could promote the expression of a Notch ligand in SGPs, or alter its activity or localization. Compared with regulation of Notch receptors, less is known about how integrin signaling modulates the activity of Delta-like Notch ligands. One example of where this has been investigated is the branching morphogenesis of new blood vessels. Tip cells signal to surrounding endothelial cells via Delta-Notch, and Notch activation favors vessel elongation and inhibits new vessel branches from forming. Laminin in surrounding BMs contributes to Notch activation by inducing expression of Delta/DLL4 (in human and mouse endothelial tip cells) in an integrin β1-dependent pathway (Estrach et al., 2011; Stenzel et al., 2011). Our results identify PAT-2/PAT-3 integrin heterodimers as receptors that may be needed in niche cells in order to tune Notch signaling with BM composition. However, the signal transduction mechanism(s) downstream of integrin in the gonad have not yet been identified.

Uncovering these mechanisms will also illustrate how BM cues are integrated with nutritional signals to control stem cell quiescence and the onset of proliferation. Independent nutritional signaling through PI3 kinase (PI3K)/AGE-1, PTEN/DAF-18 and AKT-1 also controls PGC escape from quiescence in response to food (Fry et al., 2021; Fukuyama et al., 2006; Hubbard and Schedl, 2019). Given that disruption of gonadal BM bypasses nutritional regulation and operates independently of daf-18 (McIntyre and Nance, 2020), niche cells may coordinate these two mechanisms in order to determine when PGCs exit quiescence. Coordination could happen at the level of signal transduction, similar to many cancers in which integrin cooperates with receptor tyrosine kinases to drive proliferation though PI3K and AKT (Cooper and Giancotti, 2019). BM and nutritional signals could also act in parallel to regulate a common target ligand. In fly neuroblasts, nutritional signals (via PI3K and AKT) regulate stem cell exit from quiescence when larva begin to feed (Chell and Brand, 2010; Otsuki and Brand, 2018). However, in mushroom body neuroblasts, Pax-6/eyeless expression decouples stem cell proliferation from nutrition, similar to loss of laminin in the worm gonad (Sipe and Siegrist, 2017). Perhaps integrin signaling activates a transcription factor like VAB-3/Pax-6, which does function in C. elegans distal tip migration (Meighan and Schwarzbauer, 2007), to induce Notch ligand expression.

Finally, a significant implication of our findings is that niche cells interpret environmental cues, including BM composition, in order to modulate stem cell behavior. Laser ablation studies show that PGCs do not proliferate without somatic niche cells (Kimble and White, 1981; McIntyre and Nance, 2020). In addition, our findings using tissue-specific degradation demonstrated that integrin is needed in SGP niche cells to trigger PGC proliferation when epi-1 is depleted. Therefore, in contrast to many other types of stem cells that either respond directly to nutritional signals (such as Drosophila neuroblasts) (Chell and Brand, 2010) or physically interact with the BM (such as Drosophila intestinal stem cells) (You et al., 2014), C. elegans niche cells sense the extracellular environment and modulate signaling to the PGCs accordingly. Because of its ability to interact with both integrin and various growth factor signaling pathways, perlecan may function as a crucial node in this process. Its necessity, along with integrin, demonstrates the importance of changes to BM composition in determining when quiescent stem cells are permitted to proliferate.

See Table S1 for a list of key resources.

Worms were cultured at 23°C under standard experimental conditions (Brenner, 1974) unless otherwise described. Strain N2 (Bristol) was used as the wild type. Genotypes of strains utilized in this study are listed in Table S1.

ehn-3p::cfp::PH::sl2::zif-1::tbb-2 3′ UTR (pJN640) was created by multi-fragment Gibson assembly using the vector backbone pCFJ151 (Frøkjaer-Jensen et al., 2008), which contains the unc-119(+) co-transformation marker. Plasmid DNA was injected into unc-119(ed3) worms to obtain an extrachromosomal array, which was integrated using trioxsalen and UV irradiation as described (Armenti et al., 2014). The resulting multicopy transgene insertion, xnIs559, was outcrossed twice before use.

CRISPR/Cas9-mediated genome editing was used to engineer mutations into the genome, as described (Paix et al., 2017). Purified Cas9 protein was pre-incubated with crRNA and tracrRNA, mixed with ssDNA oligonucleotide or dsDNA PCR repair templates containing >35 bp homology arms, and injected into worms together with dpy-10 co-CRISPR reagents. F1 Roller worms were singled and genotyped by PCR to identify edited worms. crRNA, repair template sequences and genotyping primers are listed in Table S1.

To make an epi-1 null allele, the STOP-IN cassette was inserted into the second exon of the previously described epi-1(qy31[epi-1::mNG+loxP]) allele (Keeley et al., 2020). epi-1(qy31) was cut with crRNA ocrDCM_001 and repaired with an ssDNA oligo (oDCM_247) that contained the STOP-IN cassette. epi-1(xn200) was maintained over the tmC5 balancer (Dejima et al., 2018). See Fig. S1.

To create the repair template, gfp::zf1 lacking the initiator methionine was amplified by PCR from a plasmid template using primers oDCM_262 and oDCM_263, which contained homology arms to the pat-3 3′ end. pat-3 was cut with crRNA ocrDCM_003 and repaired with the dsDNA PCR repair template, resulting in gfp::zf1 insertion immediately after the final pat-3 codon without a linker sequence. Insertion pat-3(xn109) was verified by PCR, sequencing and GFP expression. pat-3(xn109) animals were viable (n=94/99 embryos developed to adult). See Fig. S2.

To create a putative pat-3 null allele, we used the STOP-IN method (Wang et al., 2018a) to introduce a stop codon (in any reading frame) into the sixth exon, upstream of sequences encoding the transmembrane domain, in pat-3(xn109[pat-3::gfp::zf1]). pat-3(xn109) was cut with crRNA ocrDCM_004 and repaired with ssDNA oligo (oDCM264), which contained the STOP-IN cassette. pat-3(xn113) was maintained over the qC1 balancer. See Fig. S2.

To make an emb-9 null allele, the STOP-IN cassette was inserted into the first exon of the previously described emb-9(qy24[emb-9::mNG]) allele (Keeley et al., 2020). emb-9(qy24) was cut with crRNA ocrDCM_002 and repaired with an ssDNA oligo (oDCM_261) that contained the STOP-IN cassette. emb-9(xn201) was maintained over the qC1 balancer. See Fig. S3.

Escherichia coli bacterial strain HT115 expressing dsRNA from vector pPD129.36 or its derivatives was fed to L4 C. elegans larvae. Plates contained 0.2% β-lactose to induce dsRNA production. L4 larval worms were fed for 2 days at 23°C before progeny were analyzed. To collect progeny, adult hermaphrodites were washed twice, dissected in egg buffer [25 mM HEPES (pH 7.3), 118 mM NaCl, 48 mM KCl, 2 mM CaCl, 2 mM MgCl₂], and early-stage embryos were collected into fresh buffer. Embryos were then aged at 23°C for 2 days in egg salts (unless a different length of times is specifically indicated). Results shown include data from at least three independent trials. As a negative control, worms were fed empty pPD129.36 vector. The RNAi feeding plasmid targeting epi-1 was Ahringer RNAi library clone K08C7.3 (Kamath et al., 2003), and the RNAi feeding plasmid targeting lam-1 was previously described (McIntyre and Nance, 2020). The effectiveness of epi-1 RNAi was confirmed by comparison with the epi-1(xn200) putative null allele, as shown in Fig. 1.

For live imaging, embryos or larvae were mounted on 4% agarose pads in egg salts. Young larvae were immobilized with 1-2 μl of 5 mM levamisole applied to the coverslip before it was placed on the agarose pad. PGCs were counted using a Zeiss AxioImager A2 equipped with an AxioCam 503 mono CCD camera, Uniblitz shutter and 63×1.4 NA or 40×1.3 NA objective. Embryos and larva were imaged live using a spinning disk confocal microscope (Nikon Eclipse Ti2, CSU-W1 spinning disk, 100×1.35 NA Silicone or 60×1.49 NA oil-immersion objective, 488 nm and 561 nm lasers, matched Andor 888 Live EMCCD or single Hamamatsu Fusion BT cameras). A Leica SP8 confocal microscope (63×1.4 NA oil-immersion objective, Argon tunable and HeNe 594 nm lasers, and HyD detectors) was used for quantifying CDC-42::GFP::ZF1 degradation. Unless specified otherwise, the distance between slices in z-stacks was 500 nm. Images were acquired using Zeiss Zen software, Nikon Elements or Leica Application Suite X before being processed in ImageJ and Adobe Photoshop.

Ancestors of the SGPs (MSapp and MSppp) were ablated as described (McIntyre and Nance, 2020). Control unablated embryos were mounted on the same slides as experimental embryos and allowed to develop simultaneously. Rarely, laser ablation caused damage to nearby control unablated embryos. Such experiments were discarded and repeated.

To compare the fluorescence intensity of BM components and receptors, confocal images of control and perturbed animals were first acquired using equal laser power and exposure times and comparable focal depths. The maximum pixel intensity (averaged over a line 10 pixels in width) was measured across the gonadal BM on the side of the somatic gonad facing the endoderm. Because background fluorescence (as measured in the PGC nucleus) was minor and very consistent across the data, background fluorescence was not subtracted prior to image intensity quantification.

GFP::H2B expression from sygl-1p::gfp::H2B was quantified in the PGC nucleus by capturing images of control and SGP-ablated embryos with equal exposure settings using a Zeiss AxioImager A2 described above. Total pixel intensity was measured in a 4 µm diameter circle centered on the PGC nucleus and background fluorescence was subtracted by measuring GFP intensity in a region of the same size outside of the PGC.

Statistical tests used are indicated in figure legends. For categorical data, such as the fraction of worms with more than two PGCs, P-values were calculated using Fisher's Exact Test. Other datasets, such as fluorescence intensity measurements, were analyzed using a two-tailed Student's t-test. For t-test comparisons between two or more samples, normality was assessed according to the Shapiro–Wilk normality test. If multiple comparisons were made within a dataset, P-values were adjusted using the Bonferroni correction. Datasets presented represent at least three independent experiments with sample size indicated in each figure. Damaged embryos were excluded from analysis.

We thank David Sherwood and Myeongwoo Lee for sharing C. elegans strains with our lab. We thank Ariel Pani for generously sharing microscopes and reagents from his lab at the University of Virginia. We thank members of the Nance lab as well as E. Jane Hubbard and Niels Ringstad labs for discussion of this project and comments on the manuscript. We also thank Michael Cammer and Yan Deng from the NYU Langone Microscopy Laboratory, which is partially supported by the Cancer Center Support Grant P30CA016087 at the Laura and Isaac Perlmutter Cancer Center, for their advice and discussions of live imaging. Some strains were provided by the Caenorhabditis Genetics Center, which is funded by NIH Office of Research Infrastructure Programs (P40OD010440).

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201640.reviewer-comments.pdf.