SYGL-1 and LST-1 link niche signaling to PUF RNA repression for stem cell maintenance in Caenorhabditis elegans

Central questions in regenerative biology include how stem cells are maintained and how they transition from self-renewal to differentiation. Germline stem cells (GSCs) in Caeno-rhabditis elegans provide a tractable in vivo model to address these questions. In this system, Notch signaling and PUF RNA binding proteins, FBF-1 and FBF-2 (collectively FBF), maintain a pool of GSCs in a naïve state. An open question has been how Notch signaling modulates FBF activity to promote stem cell self-renewal. Here we report that two Notch targets, SYGL-1 and LST-1, link niche signaling to FBF. We find that SYGL-1 and LST-1 proteins are cytoplasmic and normally restricted to the GSC pool region. Increasing the distribution of SYGL-1 expands the pool correspondingly, and vast overexpression of either SYGL-1 or LST-1 generates a germline tumor. Thus, SYGL-1 and LST-1 are each sufficient to drive “stemness” and their spatial restriction prevents tumor formation. Importantly, SYGL-1 and LST-1 can only drive tumor formation when FBF is present. Moreover, both proteins interact physically with FBF, and both are required to repress a signature FBF mRNA target. Together, our results support a model in which SYGL-1 and LST-1 form a repressive complex with FBF that is crucial for stem cell maintenance. We further propose that progression from a naïve stem cell state to a state primed for differentiation relies on loss of SYGL-1 and LST-1, which in turn relieves FBF target RNAs from repression. Broadly, our results provide new insights into the link between niche signaling and a downstream RNA regulatory network and how this circuitry governs the balance between self-renewal and differentiation.


Introduction
The balance between stem cell self-renewal and differentiation is pivotal for normal development, adult homeostasis, and regeneration. Indeed, aberrant stem cell regulation can cause disease, including human degenerative disorders and cancers [1]. Stem cell daughters can exist in a "naïve" multipotent state or a "primed" state that has been triggered to differentiate, typically via transit-amplification [2][3][4]. Stem cells that divide asymmetrically rely on oriented cell division to generate one naïve and one primed daughter [e.g. 5], but the mechanism underlying stem cells that divide stochastically to generate pools of naïve and primed daughters [e.g. 6,7] remains largely unanswered. Challenges have included the complexity of their niches [8] and diversity of stem cell states (e.g. quiescent vs. proliferative) [9]. Thus, understanding how stem cell daughters are regulated to remain naïve or transition to a primed state can greatly benefit from a tractable model with well-defined niche and stem cells.
The Caenorhabditis elegans gonad provides a paradigm for analyzing regulation of a stem cell pool [10]. In this system, a single-celled mesenchymal niche maintains a pool of~225 stochastically-dividing germ cells in the "progenitor zone" (Fig 1A) [10]. That progenitor zone itself consists of a distal pool of 30-70 naïve germline stem cells (GSCs) and a more proximal pool of GSC daughters that have been triggered to begin differentiation and hence have been "primed" (Fig 1A) [11]. Central to GSC maintenance are two conserved regulators, Notch signaling and PUF (for Pumilio and FBF) RNA-binding proteins [12,13]. GLP-1/Notch signaling from the niche is essential for GSC maintenance [14] and two nearly identical PUF proteins, FBF-1 and FBF-2 (collectively FBF), act as broad-spectrum repressors of differentiation RNAs to promote GSC self-renewal (Fig 1B) [15,16]. FBF provides one regulatory hub in the stem cell regulatory network; other hubs rely on GLD translational regulators to drive differentiation [17]. However, key questions remain. Here we focus on how Notch signaling and FBF repression are coordinated to establish a naïve GSC pool and facilitate transition to the primed state.
Recently-identified GSC regulators are the sygl-1 and lst-1 genes, which are direct targets of niche signaling [18]. The lst-1 sygl-1 double mutant exhibits the same severe GSC loss as a GLP-1/Notch mutant while single mutants maintain GSCs, revealing functional redundancy [18]. However, the molecular functions of SYGL-1 and LST-1 have been a mystery. LST-1 harbors a single Nanos-like zinc finger, suggesting a possible role in post-transcriptional regulation. Yet both proteins are composed largely of low-complexity regions; neither is recognizable beyond Caenorhabditids; and the two amino acid sequences bear little resemblance to each other despite their redundancy [18]. Despite the novelty of these proteins, their striking GSC loss phenotype coupled with the restriction of their mRNAs to a region corresponding to the GSC pool [18,19] suggested that understanding their function and regulation would provide insights into regulation of a stem cell pool.
Here we investigate SYGL-1 and LST-1 proteins to understand their roles in stem cell regulation. We find that both are cytoplasmic proteins and spatially restricted to the GSC region. Intriguingly, modest SYGL-1 expansion increases size of the stem cell pool, and vast expansion of either SYGL-1 or LST-1 drives formation of a germline tumor. The SYGL-1 and LST-1dependent tumors form in the absence of GLP-1/Notch signaling, reinforcing their key roles in stem cell maintenance. However, SYGL-1 and LST-1 no longer drive tumor formation in the absence of FBF. Consistent with the idea that SYGL-1 and LST-1 drive stem cell selfrenewal in a complex with FBF, SYGL-1 and LST-1 interact physically with FBF and are required for repression of an FBF target RNA. We suggest that SYGL-1 and LST-1 are FBF partners and function to ensure repression of FBF target RNAs within the stem cell pool.

SYGL-1 and LST-1 are restricted to the GSC pool region
To visualize SYGL-1 and LST-1 proteins, we generated epitope-tagged versions of sygl-1 and lst-1, including single-copy transgenes using MosSCI [20][21][22] and endogenous alleles using CRISPR-Cas9 [23,24] (Fig 1C and 1D). Importantly, these epitope-tagged SYGL-1 and LST-1 proteins were functional: they maintain GSCs when tested in appropriate mutant backgrounds (S1D and S1E Fig). Therefore, they mimic their wild-type counterparts and we refer to them henceforth as SYGL-1 and LST-1. By immunostaining, both proteins were expressed in the cytoplasm of the distal-most germ cells within the progenitor zone: SYGL-1 was largely punctate while LST-1 was enriched in perinuclear granules (Figs 1E-1H and S1A-S1C). Using the conventional metric for position along the gonadal axis, germ cell diameters (gcd) from the distal end (Fig 1A), we found SYGL-1 enriched from 1-~12 gcd, and LST-1 from 1-~5 gcd (Fig 1K and 1L, see legend for how we determined extents). These protein extents correspond well to the distributions of their respective wild-type mRNAs, assayed by single-molecule FISH [19], and were reproducible regardless of epitope tag. We counted the number of germ cells stained for each protein and found SYGL-1 in~125 cells and LST-1 in~45 germ cells (Fig  1K and 1L). Strikingly, high SYGL-1 and LST-1 levels were correlated with low GLD-1 expression (Fig 1I and 1J), consistent with their opposing functions (see Introduction). We conclude that SYGL-1 and LST-1 are restricted within the progenitor zone to the GSC region, consistent with their pivotal roles in GSC self-renewal.
We first found that PZ size was 1.3-fold larger in tbb-2 3'UTR transgenic animals than in either sygl-1 3'UTR transgenic animals or wild type (Fig 2F). To test the idea that GSC pool size might also be enlarged, we used the emb-30 assay [11]. Briefly, this assay arrests cell divisions with a temperature-sensitive allele of emb-30 (tn377), which encodes a component of the anaphase promoting complex [29]. This arrest stops proximal movement of germ cells through the progenitor zone and resolves them into two discrete pools: cells in the distal GSC pool remain naïve and acquire an M-phase marker (PH3), while cells in the proximal pool are primed to differentiate and acquire a differentiation marker (GLD-1) [11] (Fig 2G). With this assay, we estimated GSC pool sizes in strains carrying emb-30 and either the wild-type sygl-1 locus (normal SYGL-1), the sygl-1 null mutant (no SYGL-1) or the tbb-2 3'UTR transgene (expanded SYGL-1). GSC pools with wild-type SYGL-1 contained~35 naïve cells; those with no SYGL-1 contained~21, and those with expanded SYGL-1 had~43 on average (Fig 2K). Indeed, the 1.4-fold increase in SYGL-1 extent (from~11 to~15 gcd, on average) corresponds well with the estimated 1.3-fold increase in GSC number (from 35 to 43, on average) and PZ germ cell number (from 229 to 298, on average). Importantly, the extent of LST-1 expression along the gonadal axis (gcd) and number of LST-1-expressing cells in the distal gonad were essentially the same in sygl-1(+) and sygl-1(0) germlines as well as those harboring the tbb-2 3'UTR transgene (S3B-S3F Fig). The simplest explanation is that LST-1 expression is likely independent of SYGL-1: The smaller LST-1 expression domain establishes a smaller GSC pool size in sygl-1 mutants, but that extent of SYGL-1 expression establishes GSC pool size in wildtype and tbb-2 3'UTR animals. We conclude that GSC pool size correlates with SYGL-1 extent and suggest that GSC pool size correlates with LST-1 extent in the absence of SYGL-1. To extend the idea that distributions of SYGL-1 and LST-1 govern GSC pool size, we tested the effect of expressing SYGL-1 or LST-1 throughout the germline. To this end, we made single-copy transgenes driven with a mex-5 germline promoter and the tbb-2 3'UTR, which supports ubiquitous expression throughout the germline [28] (Fig 3A). For brevity, we refer to the transgenes as sygl-1(ubiq) and lst-1(ubiq), respectively (Fig 3B and 3C). Because ubiquitous germline expression of SYGL-1 or LST-1 might render animals sterile, we created transgenes on sygl-1 or lst-1 feeding RNAi, and scored effects after RNAi removal, waiting 2-3 generations to minimize transgenerational RNAi inheritance (Fig 3A). Strikingly, ubiquitous germline expression of either SYGL-1 or LST-1 created extensive germline tumors (Fig 3E-3H). The penetrance of tumor formation depended on both temperature and number of generations after removal from RNAi, but was close to 100% for both sygl-1(ubiq) and lst-1(ubiq) after two or three generations at 15˚C (S4A and S4B Fig). About half of these tumors were proliferative throughout the gonad, while the other half included cells in the meiotic cell cycle, perhaps due to incomplete release from RNAi inheritance. Control animals harboring a GFP::H2B transgene driven with the same regulatory elements (Fig 3D) had no tumors (Fig 3I and 3J), demonstrating that the tumors are specific to SYGL-1 or LST-1.
We next used markers to determine the state of cells in sygl-1(ubiq) and lst-1(ubiq) tumors. REC-8 localizes to the nucleus of germ cells in the mitotic cycle [30] and REC-8 was nuclear throughout the tumor (Figs 3E and 3G, S4I and S4J); PH3 marks M-phase [31] and was seen in dividing cells throughout the tumor (Fig 3F and 3H); and PGL-1 marks germ cells [32] and also was found throughout the tumor (S4C and S4D Fig). Therefore, sygl-1(ubiq) and lst-1 (ubiq) tumors are composed of germ cells that are mostly in the mitotic rather than the meiotic cell cycle. In addition, FBF-1 was abundant and GLD-1 was low throughout the tumors, consistent with germ cells being in an undifferentiated state (S4F and S4G Fig). As expected, all markers behaved like wild type in the GFP::H2B control (Figs 3I and 3J, S4E, S4H and S4K).
In summary, GLP-1/Notch signaling from the niche is dispensable for SYGL-1 and LST-1 tumors, and SYGL-1 and LST-1 do not need each other for their activity (Fig 4J). In contrast, SYGL-1 and LST-1 rely on FBF to form tumors (Figs 4J and S5E-S5J). Therefore, our results are consistent with a genetic model in which sygl-1 and lst-1 act downstream of Notch but upstream or parallel to fbf (Fig 4K).

SYGL-1 and LST-1 promote FBF activity rather than FBF expression
The reliance of SYGL-1 and LST-1 on FBF to promote tumor formation suggested two ideas for their molecular function. One possibility was that SYGL-1 and LST-1 regulate FBF expression. To test this notion, we compared FBF expression in germlines with and without SYGL-1 and LST-1, using a genetic background to circumvent the SYGL-1 and LST-1 requirement for GSC maintenance: gld-2 gld-1 mutants make germline tumors independently of sygl-1 and lst-1 [18]. To detect FBF-1 and FBF-2, we used epitope-tagged transgenes, which are expressed and function biologically like their endogenous counterparts [27]. By staining, FBF-1 and FBF-2 proteins were expressed robustly both with and without SYGL-1 and LST-1 (S6A-S6F An alternate idea posits that SYGL-1 and LST-1 act together with FBF, perhaps by enhancing FBF activity in a molecular complex. To ask if SYGL-1 and LST-1 physically interact with FBF, we first turned to the yeast two-hybrid assay (Fig 5A). Briefly, SYGL-1 or LST-1 was fused to the Gal4 activation domain (AD), and the PUF repeats of FBF-1 or FBF-2 were fused with the LexA DNA binding domain (BD). Binding was assayed by monitoring growth on minimal media lacking histidine, as a measurement of HIS3 gene expression level. We imposed a stringent threshold by adding a competitive inhibitor of the HIS3 enzyme (50 mM 3-AT) to minimize false positives. Robust growth was observed when either SYGL-1-AD or LST-1-AD was co-transformed with either FBF-1-BD or FBF-2-BD but not in controls (Fig 5B and 5C). We conclude that SYGL-1 and LST-1 both interact with FBF-1 and FBF-2 in yeast. We next set out to ask if SYGL-1 and LST-1 might associate with FBF in nematodes. Immunoprecipitation of SYGL-1 and LST-1 from nematodes had been technically difficult because both proteins are normally expressed at low abundance and in only a subset of cells. To circumvent this problem, we attempted immunoprecipitation from sygl-1(ubiq) and lst-1(ubiq) tumorous animals. Immunoprecipitation was successful with SYGL-1 (Fig 5D), and subsequent biochemistry therefore focused on SYGL-1.
To ask if SYGL-1 associates with FBF in nematodes, we generated strains harboring a sygl-1 (ubiq) transgene plus epitope-tagged 3xV5::FBF-2. Our experimental and control strains made germline tumors with 3xFLAG::SYGL-1 and 3xMYC::SYGL-1, respectively. The 3xV5::FBF-2 protein is functional and expressed (S7A-S7D Fig), as previously described [41]. We used FLAG antibodies to immunoprecipitate (IP) protein from both experimental and control strains; RNase A was added to all IPs to exclude RNA dependence of interactions. 3xFLAG:: SYGL-1 co-immunoprecipitated with 3xV5::FBF-2 from the experimental but not the control strain, and this interaction was not dependent on RNA (Fig 5D). We conclude that SYGL-1 and FBF-2 associate with each other in nematodes and suggest that they form a complex.
FBF regulates many target mRNAs (see Introduction). If SYGL-1 works in a complex with FBF, then SYGL-1 protein might co-IP with FBF targets. To test this idea, we used the same strains described above and performed quantitative PCR of two established FBF targets, gld-1 and fem-3 mRNAs [15,27,[42][43][44]. The experimental IP was enriched for both target mRNAs over the control IP, but it was not enriched for eft-3 mRNA (Fig 5E), an mRNA not detected as a potential FBF target in genomic studies [27,42]. We conclude that SYGL-1 associates specifically in nematodes with both FBF protein and with FBF target mRNAs.

SYGL-1 and LST-1 repress gld-1 expression post-transcriptionally
The primary function of FBF in stem cell regulation is mRNA repression [16]. A crucial prediction of the idea that SYGL-1 and LST-1 work with FBF in a complex is that SYGL-1 and LST-1 should be required for repression of an FBF target mRNA. To test this idea, we examined gld-1 mRNA, a well-established FBF target required for differentiation [15]. Previous experiments detected a subtle increase in GLD-1 expression in GSCs of sygl-1 and lst-1 single mutants [25]. To explore this further, we again used gld-2 gld-1 mutants to remove both sygl-1 and lst-1 without changing cell fate. This time, however, we used gld-1(q361), a missense mutant that abrogates GLD-1 protein function but produces detectable gld-1 mRNA and GLD-1 protein [30,45,46] (Fig 6A). In this fashion, repression of gld-1 mRNA was uncoupled from complications of GLD-1 function in the germline.
We next assayed expression of gld-1(q361) mRNA using single molecule fluorescence in situ hybridization (smFISH). Our probe was specific to gld-1: transcripts were patterned as described previously in wild type [41,46] and cytoplasmic gld-1 mRNAs were undetectable in gld-1(q485), a deletion mutant that likely renders transcripts subject to non-sense mediated decay [45] (S8 Fig). Similar to the result with GLD-1 protein, gld-1 mRNAs were barely detectable distally when either sygl-1 or lst-1 was present, but became easily detectable distally when both sygl-1 and lst-1 were removed (Fig 6G-6J). By contrast, nascent transcripts were seen in distal germ cell nuclei regardless of sygl-1 and lst-1 (Fig 6K and 6L). We conclude that SYGL-1 and LST-1 function post-transcriptionally to repress gld-1 mRNA expression in the distal  [45] that generates mRNA and protein normally [30]. The smFISH probe set spanned the locus. See text for details. (B-E) GLD-1(q361) protein in distal gonads, stained with α-GLD-1 (green) and DAPI (cyan). Genotypes are: germline, a role that is strongly reminiscent of FBF activity. Collectively, our data support the idea that SYGL-1 and LST-1 partner with FBF to repress FBF target mRNAs in GSCs.

Discussion
The sygl-1 and lst-1 genes are targets of niche signaling and crucial for GSC self-renewal [18].
Here we investigate the functions of SYGL-1 and LST-1 proteins, which had been a mystery. Our results support three major conclusions: SYGL-1 and LST-1 are sufficient for stem cell maintenance and can be oncogenic when unregulated; the spatial restriction of SYGL-1 and LST-1 proteins governs GSC pool size; and SYGL-1 and LST-1 work with FBF to restrict its RNA repression to stem cells. Our discussion places these results in context with implications for stem cell biology more broadly.

SYGL-1 and LST-1 are sufficient for stem cell maintenance
We have found that ubiquitous expression of either SYGL-1 or LST-1 protein drives formation of extensive germline tumors, and that their tumor-forming activities do not require GLP-1/ Notch signaling from the niche. The significance of this result is three-fold. First, SYGL-1 and LST-1 are not only required for GSC maintenance, albeit redundantly [18], but each on its own also drives stemness in the form of a tumor when ubiquitously expressed. This sufficiency underscores the importance of SYGL-1 and LST-1 as key stem cell regulators. Second, SYGL-1 and LST-1 are the primary targets of niche signaling for GSC maintenance: GLP-1/Notch signaling does not induce other regulators that must work with either SYGL-1 or LST-1 to maintain GSCs. Third, spatial restriction of SYGL-1 and LST-1 prevents tumor formation, making them prototypes for a new class of oncogenes.
Central to understanding the niche regulation of stem cells is the identification and characterization of key downstream effectors. Advances have been made in several model systems [e.g. [47][48][49], but examples of niche effectors with validated in vivo significance are rare. Perhaps the most striking parallels to SYGL-1 and LST-1 are Ascl2 and LgR5, which encode niche signaling effectors in Wnt-regulated intestinal stem cells. Similar to SYGL-1 and LST-1, Ascl2 and LgR5 expression is limited to stem cells [50,51], and ectopic expression promotes hyperplasia [52]. However, in stark contrast to SYGL-1 and LST-1, Ascl2 and LgR5 functions are not independent of niche signaling: LgR5 enhances Wnt signaling and Ascl2 works with Wntdependent transcription factors to induce a stem cell transcriptional signature [53]. Therefore, SYGL-1 and LST-1 stand out as direct targets of niche signaling that promote self-renewal by an intrinsic signaling-independent mechanism. Spatial restriction of SYGL-1 and LST-1 governs GSC pool size Normally, SYGL-1 and LST-1 are spatially restricted to a region that correlates with estimates of the GSC pool (Fig 7A). We confirmed the biological significance of this spatial restriction in two ways. First, a moderate expansion of SYGL-1 expression led to a similar moderate expansion of pool size. Second, a major expansion of either SYGL-1 or LST-1 led to the formation of massive germline tumors. The simple conclusion is that the presence of either SYGL-1 or LST-1 promotes the stem cell fate, while their absence is critical for the transition towards differentiation. Logical corollaries are that spatial distributions of SYGL-1 and LST-1 govern the size of the GSC pool and that their loss facilitates the transition to a cell state primed for and (L) show a single z-slice. Conventions as in Fig 1E-1J; scale bar is 20 μm in all images, except 2 μm in (K) and (L). n, number of gonadal arms.
https://doi.org/10.1371/journal.pgen.1007121.g006 differentiation. A key question is how their spatial restriction is regulated. GLP-1/Notch signaling from the niche activates sygl-1 and lst-1 transcription in distal germ cells [18], but what regulates their disappearance? A partial answer is RNA regulation: the sygl-1 3'UTR restricts SYGL-1 protein expression compared to the tbb-2 (tubulin) 3'UTR. In addition to RNA regulation, we suggest that SYGL-1 and LST-1 protein stabilities are also regulated. Despite the rapid kinetics of germ cell movement (~1 gcd per hour [54]), the distributions of sygl-1 mRNA and protein are similar, as are those of lst-1 mRNA and protein [19; this work]. Therefore, the SYGL-1 and LST-1 proteins must turn over as germ cells move proximally within the progenitor zone. Others have found that the proteolytic machinery is critical for progression from a stem cell state to a differentiated state in the progenitor zone [55,56]. We suggest that SYGL-1 and LST-1 are likely targets of such proteolysis.
The C. elegans gonad therefore provides a new paradigm for how niche signaling can act through spatially restricted regulators to not only ensure the existence of stem cells but also to govern the size of a stem cell pool and facilitate the transition to a primed state. Spatial regulation is a common theme in animal development [57,58] and extends to stem cell regulators. In addition to Lgr5 and Ascl2 (described above), the Escargot/Snail transcription factor follows a similar principle in intestinal stem cells in Drosophila and mouse models [59,60]. More relevant to this work is the Drosophila PUF protein, Pumilio, which promotes GSC self-renewal [61,62]. Pumilio is spatially restricted to GSCs and its ectopic expression generates germline tumors [63]. The clarifying advances of our work are an application of this theme to the maintenance of a stem cell pool, which is likely a broadly-used mechanism, and to a PUF protein partner rather than a PUF protein per se (see below).

SYGL-1 and LST-1 partner with FBF to repress mRNA in stem cells
When this work began, the molecular functions of SYGL-1 and LST-1 were unknown (see Introduction). A first clue from this work was their cytoplasmic localization, which is consistent with a role in post-transcriptional regulation but can be explained in other ways. A more significant clue was that SYGL-1 and LST-1 cannot drive germline tumors in the absence of the FBF RNA-binding protein. One explanation might have been that SYGL-1 and LST-1 promote FBF expression, but that possibility was not confirmed: FBF-1 and FBF-2 were expressed in the absence of SYGL-1 and LST-1. An alternative idea was that SYGL-1 and LST-1 might work with FBF to promote mRNA repression. In support of that explanation, SYGL-1 and LST-1 interact with FBF-1 and FBF-2 in yeast two-hybrid assays; SYGL-1 co-immunoprecipitates from nematodes with both FBF-2 protein and with FBF target mRNAs; and SYGL-1 and LST-1 post-transcriptionally repress expression of one of those FBF targets in GSCs. These multiple lines of evidence support the model that SYGL-1 and LST-1 partner with FBF to repress mRNAs in GSCs (Fig 7B). We emphasize that SYGL-1 and LST-1 must also have FBFindependent functions, because the lst-1 sygl-1 phenotype is more severe than the fbf-1 fbf-2 phenotype [15,18], and because single sygl-1 and lst-1 mutants enhance the fbf-1 fbf-2 phenotype [25; this work]. The fog-1 gene, which encodes a cytoplasmic polyadenylation element binding (CPEB) related protein [64,65], redundantly promotes GSC self-renewal with FBF in that fog-1 fbf-1 fbf-2 triple mutants contain a GSC loss similar to that of glp-1 null [66]. We speculate that the FBF-independent functions of SYGL-1 and LST-1 may involve regulation of FOG-1 protein or key FOG-1 mRNA targets. But of course, other possibilities exist. Regardless, this work shows conclusively that SYGL-1 and LST-1 have an FBF-dependent function and that they likely operate with FBF in a complex.
SYGL-1 and LST-1 stand out among PUF partners as the first to be essential for GSC maintenance, the first to be spatially restricted to the stem cell region, the first to affect size of a stem cell pool, the first to be tumorigenic when overexpressed, and the first to be essential for mRNA repression in GSCs. Previously identified FBF partners include NOS-3, a Nanos homolog which is expressed throughout the germline [67,68], and CPEB/CPB-1, which is expressed and functions in spermatocytes [64,69]. Two other FBF partners, GLD-2 and GLD-3, activate mRNAs and promote germ cell differentiation [70][71][72], a function opposite that of SYGL-1 and LST-1. The molecular mechanisms by which SYGL-1 and LST-1 repress RNAs await future studies. The simplest possibility is that they enhance FBF recruitment of the Not1 deadenylase complex, a conserved mode of PUF repression from yeast to humans [73][74][75]. Another idea is that SYGL-1 or LST-1 influences the sequence specificity and kinetics of FBF binding to target mRNAs, analogous to reports for other PUF partners such as CPB-1 for FBF [76] and Nanos or Brat for Drosophila Pumilio [77,78]. A third thought is that SYGL-1 and LST-1 repress RNAs by recruiting them to sites of repression in RNP granules. The emerging view of low complexity proteins as RNA granule scaffolds [e.g. 79, 80] coupled with the punctate or granular appearance of SYGL-1 and LST-1 make this third possibility attractive, but it remains speculative. Given that several mechanisms remain plausible, we note that SYGL-1 and LST-1 may employ distinct biochemical mechanisms, despite their biological redundancy in GSC maintenance and their molecular redundancy in gld-1 mRNA repression. A tantalizing future direction is to ask if similar counterparts of SYGL-1 or LST-1 exist in other vertebrate stem cell models to enhance the repressive activity of PUF proteins, Pum1 and Pum2.

Molecular model for governing the naïve state and size of a stem cell pool
Our findings together with previous studies support a model for how niche signaling is coordinated with intrinsic stem cell regulators to establish a GSC pool with stem cells in their naïve state and then facilitate the transition to a state primed for differentiation (Fig 7C). Essentially, Notch signaling localizes the GSC pool by activating expression of key intrinsic stem cell regulators, SYGL-1 and LST-1, which partner with FBF to repress differentiation mRNAs and thereby promote the naïve state (Fig 7C, left) [14, 15, 18, 19; this work]. Pool size is established roughly by Notch signaling, which activates sygl-1 transcription in a steep gradient across the pool [18,19]. However, sygl-1 mRNAs are less graded and therefore transform the steep transcriptional gradient into a markedly less steep RNA gradient [19]. Here, we show that SYGL-1 protein abundance disappears in a pattern closely mirroring loss of its mRNAs. We propose that removal of these key FBF partners drives the transition from a naïve to a primed state (Fig  7C, middle), and that loss of SYGL-1 and LST-1 triggers entry into a primed state by releasing gld-1 and likely other RNAs from repression (Fig 7C, right). We note that FBF is present not only in the GSC pool but also in primed cells and cells beginning overt differentiation (entry into meiotic prophase) [15,41,81]. However, repression of FBF target mRNAs occurs in the distal germline [15,40,75,[82][83][84] and is strongest in the distal-most region or the naïve GSC pool [11]. This pattern suggests that FBF in primed cells is becoming less repressive as SYGL-1 and LST-1 are lost; indeed, FBF may be transitioning to an activating mode in this primed region [10,75]. Two other FBF partners, GLD-2 and GLD-3, activate FBF-bound RNAs [75], suggesting the possibility of a partner exchange during the transition in primed cells. One can imagine that SYGL-1 and LST-1 might be displaced by competition of other FBF partners or they might be removed by spatially regulated proteolysis. Although our model is surely oversimplified, it provides a heuristic framework for future explorations of stem cell pool regulation. For example, the model poises our thinking for analysis of both the mechanism and kinetics of transition from a naïve state to a primed state, which are likely to have profound consequences on pool regulation. Regardless, this model provides critical insights into how niche signaling is coordinated with downstream intrinsic effectors to govern the existence of a stem cell pool and its size.
To obtain lst-1(q826), a sygl-1 enhancer screen was performed with EMS mutagenesis as described [85], with minor modifications. Briefly, sygl-1(tm5040) hermaphrodites of the fourth larval stage (L4) were mutagenized with 25 mM Ethyl methanesulfonate (Sigma #M0880) for 4 hours at room temperature. F1 progeny were singled and maintained at 15˚C, and F2 selfprogeny were screened for germline proliferation defective (Glp) [14] mutants. Details of this mutagenesis screen are available upon request. The lst-1 locus was sequenced from DNA extracted from Glp animals to identify the lst-1(q826) allele, which was outcrossed 10 times with wild type prior to analysis.

Reannotation of sygl-1 and lst-1 gene structures
The sygl-1 and lst-1 gene structures reported here are based on 5' rapid amplification of cDNA ends (RACE), genome-wide mRNA sequencing data (WormBase release 255), and ribosome profiling data [94]. Specifically, the sygl-1 5'UTR, the lst-1 5'UTR, and the lst-1 start codon have been re-annotated. Most importantly, our reported lst-1 start codon removes 70 amino acids from the previously mis-annotated versions [18] and is consistent with evolutionary data from C. briggsae.
Phenotype analyses: Brood counts, sterility and embryonic lethality L4 hermaphrodites were placed onto individual plates at 20˚C. At 6 to 12 hour intervals, the hermaphrodite was moved to a new plate and the embryos were counted for sterility and brood counts. Several days later, hatched progeny on each plate were counted to determine embryonic lethality.

Assessment of progenitor zone size
All characterization of progenitor zone (PZ) size was done in animals raised at 20˚C until 24 hours after L4, except in S5J Fig where animals were raised at 25˚C until 16-18 hrs after L4. To visualize nuclear morphology, gonads were dissected, fixed, and stained with DAPI (see immunostaining and DAPI staining section below for dissection and fixation methods). To determine the PZ size, gonads were imaged using a confocal microscope with a z-stack depth of 0.4-0.5 μm. Next, the boundary between PZ and Transition Zone (TZ) was determined by conventional criteria [95]. Briefly, many germ cells in the TZ have entered meiotic prophase and hence have a crescent-shaped nuclear morphology. The PZ/TZ boundary was scored as the distal-most cell row with at least two crescent-shaped nuclei. Finally, the cells within the progenitor zone were counted manually using the cell-counter plug-in in FIJI/Image J, with each DAPI-stained nucleus scored as a single cell.

Germ cell number estimation in fbf gonads
To estimate the number of germ cells in fbf-1 fbf-2 gonads reported in S5D Fig, compact nuclei typical of mature sperm in a gonadal arm were counted manually using the cell counter tab in Openlab 5.5.2 (PerkinElmer). Next, the number of sperm was converted to the number of germ cells (four sperm are made from one germ cell).

Estimation of SYGL-1 or LST-1 positive germ cells
To estimate the number of distal germ cells expressing SYGL-1 or LST-1, JK4996, JK5073, JK5205, JK5263, JK5893, JK5929 and JK6002 were raised at 20˚C until adulthood (24 hours after L4), along with appropriate wild-type controls. Gonads were dissected, fixed, and stained with anti-FLAG, anti-OLLAS, anti-V5, or anti-HA (see immunostaining section below) and imaged using the confocal microscope. Next, the number of distal germ cells that contained positive V5 or OLLAS signal (SYGL-1) or positive HA, FLAG, or V5 signal (LST-1) above the background level was manually scored, using the cell-counter plugin in FIJI/Image J.

emb-30 assay
The assay was performed as described [11] with minor modifications. DG627, JK5233, JK5235 animals were raised at 15˚C until 36 hours past mid-L4, then moved to plates pre-incubated at 25˚C and maintained at 25˚C for 12.5 hours. We chose 12.5 hours because germ cell counts became unreliable with longer times (nuclear morphology became increasingly compromised after incubations of 13 hours and longer). Next, gonads were dissected, fixed, and stained for anti-PH3, anti-GLD-1 and DAPI (see staining section below). To estimate the number of cells within the distal pool, we manually counted the number of M-phase arrested germ cells distal to the GLD-1 boundary (as assessed by DAPI morphology and PH3 staining) using the cell counter tab in Openlab 5.5.2 (PerkinElmer). Scoring was done blind to genotype. We excluded samples with abnormal, fragmented nuclei that made cell counting unreliable (22-49% per genotype). We note that not every nucleus distal to the GLD-1 boundary was arrested in Mphase in some gonads but these few nuclei were included in the "distal pool" counts.

In situ hybridization
Single molecule FISH (smFISH) was performed as described [19,41,100]. Custom Stellaris FISH probes were designed by utilizing the Stellaris FISH probe designer (Biosearch Technologies, Inc) available online at www.biosearch.com/stellarisdesigner. The gld-1 probe set contains 48 unique probes labeled with CAL Fluor Red 610 and was used at a final concentration of 0.25 μM.

Microscopy
For the compound microscopy data shown in Fig 4, images were taken using a Zeiss Axioskop with Hamamatsu CCD or ORCA cMOS camera equipped with 63x 1.4NA Plan Apochromat oil immersion objective. Carl Zeiss filter sets 49, 38, and 43HE were used for the visualization of DAPI, Alexa 488, and Alexa 555 respectively. An X-Cite 120Q lamp (Lumen Dynamics) was used as the fluorescence light source. Openlab 5.5.2 (PerkinElmer) and Micromanager [101,102] were used as acquisition software. For all other figures, a Leica TCS SP8 confocal microscope driven by LAS software version 3.3.1 or X was used. This laser scanning confocal microscope was equipped with Photomultiplier (PMT) and Hybrid detectors (HyD). For all images, a 63x 1.4NA HC Plan Apochromat oil immersion objective was used with 100-200% zoom for immunostaining, and 300% zoom for single molecule FISH, using the standard scanner with 400Hz scanning speed. For figure preparation, contrast was linearly adjusted in Adobe Photoshop identically across all samples. In some cases, images were merged using the stitching plugin in FIJI/Image J [103] to generate whole gonad images.

Fluorescence quantitation
All images used for quantitation were acquired using the sequential scan mode on the Leica TCS SP8, under the same conditions across all samples. Next, average intensity of multiple zslices was projected onto a single plane. To eliminate signal intensities outside of the gonad (i.e. intestine), a separate binary mask was created by thresholding Nomarski images of the gonad taken at the same time; the binary mask was then multiplied to other channels such that only signals within the gonad would be considered for quantitation. Next, intensity at a given distance "x" from the distal tip of the gonad was averaged over five-micron intervals ("moving average"). For simplicity, distance from the distal end was converted to conventional germ cell diameters, using a conversion ratio of 4.55 μm for one germ cell diameter [19]. A custom MATLAB script was used to process steps described above.

Immunoprecipitations and Western blots
JK5366, JK5574, JK5783, and JK5844 animals were raised at 15˚C until they developed germline tumors as young adults (12 hours after L4) (see tumor assay above). Animals were washed twice with M9 buffer [3 g/L KH2PO4, 6 g/L NaHPO4, 5 g/L NaCl, and 1 mM MgSO4] and cross-linked with 1% (w/v) formaldehyde for 10 minutes at room temperature (RT). Pellets were resuspended in 1 ml lysis buffer [50 mM HEPES pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% (v/v) Triton-X, complete Protease inhibitor cocktail (Roche)], frozen in liquid nitrogen, and pulverized with mortar and pestle for 10 minutes. Lysates were cleared twice by centrifugation (12,000g, 10 minutes), and the total protein concentration was measured by Direct Detect Spectrophotometer (Millipore). To prepare antibody conjugated beads, 30 μg anti-FLAG (M2 clone, Sigma #F3165) was incubated with 4.5 mg protein G Dynabeads (Novex, Life Technologies, #10003D) for 30 minutes at RT. Next, 20 mg lysates were incubated with the antibody-bead mixture for 4 hours at 4˚C, with the presence of RNase A at 10 μg/ml. RNA degradation was confirmed by isolating total RNA from post-IP lysates using TRIzol LS (Invitrogen #10296028) and analyzing on agarose gels. Beads were pelleted, washed four times with lysis buffer, and then two times with wash buffer [50 mM HEPES pH 7.5, 0.5 M NaCl, 1 mM EDTA, 1% (v/v) Triton X-100]. Samples then were eluted with elution buffer [1% (w/v) SDS, 250 mM NaCl, 1 mM EDTA, 10 mM TRIS pH 8] for 10 minutes at 65˚C, and analyzed using SDS-PAGE on an 8% or 12% acrylamide gel.

RNA immunoprecipitation (IP) and quantitative PCR
JK5366 and JK5574 were raised at 15˚C until they developed germline tumors as young adults (see tumor assay above). Immunoprecipitation was done as above except that formaldehyde cross-linking and RNase treatment of lysates were omitted. Instead, lysis buffer contained 1 U/ μl SUPERaseÁIn RNase inhibitor (Ambion #AM2694). Successful IP was confirmed by analyzing 10% of elution by Western blot, and RNA was eluted from the rest of the beads with 0.5 ml TRIzol (Invitrogen #15596026). RNA was purified by RNeasy Micro kit (Qiagen #74004) including DNase I treatment on column. Purified RNA was checked for integrity, and converted to cDNA with Superscript III first strand synthesis kit (Invitrogen #18080051) using random-hexamers as primers. Quantitative PCR was carried out using a Roche Lightcycler 480 with TaqMan gene expression assays (Applied Biosystems). Enrichment was calculated by ΔΔ C T method [106]. Taqman probes used are as follows: gld-1, Ce02409901_g1; eft-3, Ce02448437_gH; rps-25, Ce02464216_g1; fem-3, Ce02457444_g1.

Statistical analysis
Statistical tests are indicated in figure legends with sample sizes. In most cases, one-way ANOVA and post-hoc Tukey multiple comparison tests were performed to calculate p-values. In cases where equal variance assumption of ANOVA was not established at p<0.01 (Levine's test), Welch's one-way ANOVA (modified ANOVA with heteroskedastic data) and post-hoc Games-Howell multiple comparison tests were performed to calculate p-values. All statistics were performed in R.