Abstract
Germline stem cells (GSCs) in Caenorhabditis elegans are maintained by GLP-1/Notch signaling from the niche and by a downstream RNA regulatory network. Loss of the GLP-1 receptor causes GSCs to precociously undergo meiotic differentiation, the “Glp” phenotype, due to a failure to self-renew. lst-1 and sygl-1 are functionally redundant direct targets of GLP-1 signaling whose gene products work with PUF RNA binding proteins to promote GSC self-renewal. Whereas single loss-of-function mutants are fertile, lst-1 sygl-1 double mutants are sterile and Glp. We set out to identify genes that function redundantly with either lst-1 or sygl-1 to maintain GSCs. To this end, we conducted forward genetic screens for Glp mutants in genetic backgrounds lacking functional copies of either lst-1 or sygl-1. The screens generated nine glp-1 alleles, two lst-1 alleles, and one allele of pole-1, which encodes the catalytic subunit of DNA polymerase ε. Three glp-1 alleles reside in Ankyrin (ANK) repeats not previously mutated. pole-1 single mutants have a low penetrance Glp that is enhanced by loss of either lst-1 or sygl-1. Thus, the screen uncovered one locus that interacts genetically with both lst-1 and sygl-1 and generated useful mutations for further studies of GSC regulation.
Introduction
Stem cells maintain a robust balance between self-renewal and differentiation to ensure tissue homeostasis despite physiological and environmental challenges. Failure to maintain that balance can lead to tissue dysfunction, disease, and death (Simons and Clevers 2011). Therefore, understanding the molecular circuitry governing stem cell regulation is critical. Yet biologically robust regulatory circuits are notoriously difficult to disentangle.
The C. elegans germline is a powerful system for the study of stem cell regulation (Hubbard and Schedl 2019; Gordon 2020). The adult hermaphrodite germline is contained in two U-shaped gonadal arms and produces oocytes; sperm are made during larval development and stored for later fertilization (Figure 1A, top). Germline stem cells (GSCs) are maintained at the distal end of each gonadal arm by a single-celled somatic niche, while GSC daughters differentiate as they move proximally away from the niche and ultimately undergo oogenesis (Figure 1A, middle)(Hubbard and Greenstein 2000).
GSC self-renewal depends on GLP-1/Notch signaling from the niche and on a downstream RNA regulatory network. In glp-1 null mutants, GSCs fail to self-renew and instead differentiate precociously into sperm—the “Glp” phenotype (Austin and Kimble, 1987) (Figure 1A, bottom). Downstream of GLP-1/Notch, a “PUF hub” is required for self-renewal (Figure 1B). This regulatory hub comprises four genes encoding PUF RNA binding proteins as well as two direct GLP-1/Notch target genes, lst-1 and sygl-1, that encode novel PUF interacting proteins (Crittenden et al. 2002; Kershner et al. 2014; Shin et al. 2017; Haupt et al. 2019a, 2019b; Qiu et al. 2019).
The PUF hub is characterized by pervasive genetic redundancy. For example, mutants lacking three PUF homologs are able to sustain some GSC self-renewing divisions, but animals lacking all four homologs phenocopy glp-1 null mutants (Haupt et al. 2019b). Moreover, single mutants lacking lst-1 or sygl-1 are fertile and similar to the wildtype, while lst-1 sygl-1 double mutants phenocopy glp-1 null mutants (Figure 1C) (Kershner et al. 2014). The highly redundant nature of the PUF hub has hampered the identification of its component parts. Indeed, the LST-1 and SYGL-1 were not identified using standard forward genetic approaches, but instead were discovered using a candidate gene approach (Kershner et al. 2014), leaving open the possibility that additional components remain unidentified. For example, the LST-1 or SYGL-1 proteins might work with other unknown redundant factors. Here we describe the results of mutagenesis screens designed to identify regulators that function redundantly with lst-1 or sygl-1.
Methods
Strain Maintenance
Unless noted otherwise, strains were maintained as previously described (Brenner 1974), at a temperature of 15°C. Balancers used to maintain recovered alleles were hT2[qIs48] (Siegfried and Kimble 2002) and hIn1[unc-54(h1040)] (Zetka and Rose 1992). Table 1 lists the strains used and their genotypes.
Screen design and phenotype scoring
We screened for lst-1 or sygl-1 enhancers using a modified ethyl methanesulfonate (EMS) protocol (Brenner 1974). Fourth larval stage (L4) hermaphrodites were soaked in 25 mM EMS (Sigma: M0880) for 4 hours at room temperature, washed with M9, and placed on plates. F1 progeny were singled onto individual Petri dishes and allowed to self at 15°C. F2 adult progeny were scored for sterility by dissecting scope, and then L4 larvae were scored for a Glp phenotype using a Zeiss Axioskop compound scope equipped with DIC Nomarski optics, as described (Kershner et al. 2014). Each screen was done in two ways— first with single mutants lst-1(ok814) and sygl-1(tm5040) (Figure 1D, regimen 1) and then with each of the same mutants carrying a transgenic copy of wildtype glp-1 (Sorensen et al. 2020) in addition to an endogenous copy of wildtype glp-1 (Figure 1D, regimen 2).
Allele identification
Following isolation of a Glp mutant, the starting lst-1 or sygl-1 allele was crossed away to test whether the Glp phenotype depended on loss of lst-1 or sygl-1. Mutations were then mapped to a chromosome and tested for their ability to complement alleles of likely candidate genes. Mutants that were fertile as single mutants and mapped to chromosome I were tested for complementation with lst-1(ok814) I. Briefly, the double mutant (e.g. mut-x sygl-1) was balanced over the green balancer hT2[qIs48], crossed to lst-1(ok814) sygl-1(tm5040)/hT2[qIs48] males, and non-green L4 male progeny (e.g. mut-x sygl-1/ lst-1 sygl-1) scored for Glp. Mutants that were sterile as single mutants and mapped to chromosome III were tested for complementation with glp-1(q175) III. Briefly, unc-32 glp-1(q175)/ hT2[qIs48] males were mated to each suspected glp-1 allele and non-green male progeny scored for Glp. If an allele failed to complement either lst-1 or glp-1, then Sanger sequencing was used to identify the molecular lesion. The glp-1(q823) allele was sequenced 2382 bp upstream of the 5’ UTR and 927 bp downstream of the 3’ UTR in addition to the exons and introns, but no lesion was found.
Whole genome sequencing was used to identify the likely lesion in q831, which was sterile as a single mutant and mapped to the right arm of chromosome I. Briefly, we picked ∼570 adult homozygotes, isolated DNA with Puregene Core Kit A (Qiagen ID: 158667) following manufacturer’s directions and submitted the DNA (∼100 ng) to the Wisconsin Biotechnology Core for sequencing using an Illumina MiSeq. The genome sequence was uploaded to a Galaxy server and analyzed by CloudMap, as previously described (Minevich et al. 2012). A premature stop codon occurred in one gene, F33H2.5, which resides on the right arm of chromosome I. q831 failed to complement F33H2.5 (gk49) (Barstead et al. 2012), and the premature stop codon was confirmed by Sanger sequencing of DNA from q831 homozygotes.
Assay for temperature sensitivity of glp-1 alleles
Balanced strains carrying glp-1 alleles were maintained at 15°C, 20°C, or 25°C for at least one generation before homozygous glp-1 L4 progeny were scored for Glp.
pole-1 phenotype assay
Homozygous pole-1 (q831 or gk49) animals were distinguished from the balancer hIn1[unc-54(h1040)] by their kinked, uncoordinated movement. Homozygous mid-L4 hermaphrodites were raised at 20°C, anesthetized in levamisole, mounted on an agarose pad, and examined using a Zeiss Axioskop compound scope (Crittenden et al. 2017). Vulva formation—wildtype, multivulva, or vulvaless—was scored in addition to germline defects.
Immunostaining
Strains were maintained at 20°C for immunostaining following published procedure (Crittenden et al. 2017). The SP56 polyclonal anti-sperm antibody (Ward et al. 1986), a gift from Susan Strome (UCSC, California), was diluted 1:200. The secondary antibody Alexa Fluor 555 donkey α-mouse (1:1000, Invitrogen #A31570) was added with DAPI (1 µg/mL) to mark DNA. Gonads were mounted in Vectashield (Vector Laboratories #H-1000), sealed with nail polish, and kept in the dark at 4°C until imaging.
Microscopy
DAPI/SP56 stained gonads were imaged with a Zeiss Axioskop compound microscope equipped with a Hamamatsu ORCA-Flash4.0 cMos camera and a 63/1.4 NA Plan Apochromat oil immersion objective. Carl Zeiss filter sets 49 and 43HE were used for the visualization of DAPI and Alexa 555. Images were captured using Micromanager (Edelstein et al. 2010, 2014).
lst-1 RNAi
The lst-1 RNAi clone from the Ahringer library (Fraser et al. 2000) was used. Briefly, lst-1 RNAi or empty vector control (pL4440) containing HT115 bacteria were grown overnight at 37°C in 2xYT media containing 25 μg/μl carbenicillin and 50 μg/μl tetracycline. Cultures were concentrated, seeded onto Nematode Growth Medium (NGM) plates containing 1mM IPTG, then induced overnight. L4 hermaphrodites were fed, allowed to self, and progeny were scored for the Glp phenotype by DIC.
GLP-1 protein conservation
Protein sequences for C. elegans glp-1 orthologs from other Caenorhabditis species were acquired from Wormbase. Sequences of the ANK repeats were aligned using M-Coffee to examine amino acid conservation (http://tcoffee.crg.cat/apps/tcoffee/do:mcoffee) (Notredame et al. 2000).
Results and Discussion
Screens for Glp mutants in lst-1 and sygl-1 single mutant backgrounds
To identify new GSC regulators and perhaps new components of the PUF hub, we conducted genetic screens for mutations that cause a Glp phenotype in a lst-1(lf) or sygl-1(lf) single mutant background (Figure 1D). Our initial screens simply mutagenized lst-1(lf) and sygl-1(lf) single mutants and scored their F2 progeny for the Glp phenotype (Figure 1D, regimen 1). We screened 8749 haploid genomes after mutagenesis of lst-1(lf) and 5504 haploid genomes after mutagenesis of sygl-1(lf) (Table 2). This first set of screens recovered ten mutants. However, outcrossing revealed that Glp phenotypes did not depend on either lst-1(lf) or sygl-1(lf); therefore, these mutant alleles were not of genes functionally redundant with lst-1 or sygl-1. Nine mutations, alleles q817-q825, caused a fully penetrant Glp phenotype and mapped to chromosome III (Table 3). Because the glp-1 locus is large (∼7.4 kb) and located on chromosome III, these nine mutations were likely glp-1 alleles. Indeed, all nine failed to complement glp-1(null) (Table 3). The 10th allele q831 caused a low penetrance Glp and was mapped to the right arm of chromosome I, at some distance from both sygl-1 and lst-1 loci. Therefore, this mutation must be a lesion in some other gene; its identity is described below.
The initial screens were heavily biased for the recovery of glp-1 alleles. To limit the isolation of more glp-1 alleles, we introduced a transgenic copy of wildtype glp-1 into the lst-1(lf) and sygl-1(lf) single mutants (Figure 1D; Table 2). The glp-1 transgene, qSi44 or glp-1(tg), is a single copy insertion of wildtype glp-1 on chromosome II that rescues a glp-1 null mutant (Sorensen et al. 2020). Using the same EMS mutagenesis procedure as before, we screened 7922 lst-1(lf); glp-1(tg) haploid genomes and 3868 sygl-1(lf); glp-1(tg) haploid genomes. No Glp mutants were isolated from lst-1(lf); glp-1(tg) but two were recovered from sygl-1(lf); glp-1(tg) (Table 2). These mutations were subsequently determined to be alleles of lst-1 (see below). Table 3 summarizes the genetic characterization of alleles recovered from the screen, and Table 4 summarizes their molecular lesions. Our failure to recover sygl-1 alleles in the lst-1(lf) background shows that our screens were not performed to saturation. However, we note that the sygl-1 locus is relatively small (621 bp coding region) and therefore likely a poor mutagenesis target.
Characterization of lst-1 alleles
The lst-1 locus generates two RNA isoforms – one longer, called lst-1L, and one shorter, called lst-1S (Figure 2A; Table 4). Most lst-1 alleles available prior to this work were isolated in deletion screens (Kershner et al. 2014) or engineered by CRISPR/Cas9 gene editing (Haupt et al. 2019a). In addition, one allele from these screens was previously reported, the nonsense mutant lst-1(q826)(Shin et al. 2017). Here we report a second allele obtained in the screen, lst-1(q827), which alters the 5’ splice site in lst-1L intron 2 (Figure 2A; Table 4). As previously reported for lst-1(q826), lst-1(q827) was confirmed by complementation tests and Sanger sequencing. Both alleles are phenotypically similar to previously characterized lst-1(lf) mutants: as a single mutant, they appear wildtype and as lst-1 sygl-1 double mutants they are Glp. These lst-1 alleles will prove useful in future studies focused on lst-1 function.
Characterization of glp-1 alleles
We identified the molecular lesions in the glp-1 alleles with Sanger sequencing: q818, q821, and q822 were nonsense mutants; q817, q819, and q820 were missense mutants and q825 altered a 5’ splice site (Figure 2B; Table 4). The q824 allele had a 2 bp change (AC→CA) in intron 4 that did not affect the 5’ or 3’ splice sites or the branch point (Figure 2B). We failed to determine the lesion in one allele, q823, despite sequencing all exons and introns plus 2382 bp upstream of the transcription start site and 927 bp downstream of the 3’ UTR. Nonetheless, the remaining eight alleles were all previously unreported glp-1 lesions.
The three glp-1 missense alleles—q817, q819, and q820 – all carry amino acid changes in the Ankyrin (ANK) repeats (Figure 2B and 2C). ANK repeats are conserved across eukaryotes with roles in protein interaction, cell signaling, and disease (Roehl et al. 1996; Mosavi et al. 2004). Many previously identified glp-1 alleles also have changes in this region. Mutations in ANK repeats 1, 2, 4, and 5 all cause a temperature sensitive Glp phenotype (Kodoyianni et al. 1992; Berry et al. 1997; Nadarajan et al. 2009; Dalfo et al. 2010). Our three newly identified missense alleles occur in different repeats, ANK 3 (q819 and q820) and ANK 6 (q817) and they are not temperature sensitive (Table 5). All three affect conserved residues (Figure 3). We conclude that the newly identified ANK missense mutations affect residues essential for GLP-1 function. These alleles should prove useful for investigating ANK repeats and their role in Notch signaling.
Characterization of pole-1(q831)
One mutant allele isolated in the sygl-1(lf) background, q831, mapped to the right arm chromosome I. Whole genome sequencing revealed a nonsense mutation R1899Stop in F33H2.5 (Table 4), which encodes a C. elegans ortholog of the catalytic subunit of DNA polymerase ε (Figure 4A). We confirmed q831 as an allele of F33H2.5 by Sanger sequencing, and by its failure to complement gk49, a deletion allele in F33H2.5 that had been generated by the C. elegans Knockout Consortium (Barstead et al. 2012). F33H2.5 has been named pole-1 for its DNA polymerase ε orthology.
The pole-1(q831) mutation was isolated because sygl-1(lf) pole-1(q831) double mutants were Glp. During outcrossing, we found that pole-1(q831) single mutants were 100% sterile (Figure 4D-F). To ask if pole-1 sterility was due to a Glp defect, we examined L4 larvae under DIC/Normaski and also stained dissected gonads with a sperm-specific antibody (SP56) (Ward et al. 1986) and DAPI (Figure 4B-F) (see Methods). Wildtype L4 gonads contain several hundred germ cells, with undifferentiated cells at the distal end and differentiated sperm at the proximal end (Figure 4B). glp-1(null) L4 gonads, by contrast, contain only a few germ cells, all of which have differentiated into SP56-positive sperm extending to the distal end (Figure 4C). Similar to glp-1(null) gonads, the pole-1(q831) gonads were physically smaller than wildtype; however only ∼30% had differentiated sperm extending to the distal end and thus were Glp (Figure 4D and F). The other ∼70% did not have sperm extending to the distal end and were designated nonGlp steriles (Figure 4E and F). We also observed a low penetrance Glp phenotype in the deletion strain pole-1(gk49)(Figure 4A, 4F). In addition to germline defects, pole-1 mutants had a range of other defects, consistent with a broad role in development. For example, pole-1 mutants had vulval defects (Figure 4F) and were uncoordinated.
We next asked if the pole-1 Glp phenotype was enhanced by loss of either lst-1 or sygl-1. Whereas pole-1(q831) single mutants were 30% Glp, pole-1(q831) lst-1(RNAi) animals were 80% Glp and pole-1(q831) sygl-1(lf) double mutants were 65% Glp (Figure 4F). Thus, loss of either lst-1 or sygl-1 enhanced the pole-1 Glp defect. However, pole-1 vulval defects were not similarly enhanced (Figure 4F). DNA polymerase ε pole-1 had not been recognized as having an effect on GSC regulation though other components of the DNA replication machinery have been implicated in germ cell proliferation (Yoon et al. 2018). We conclude that sygl-1 and lst-1 are germline enhancers of pole-1.
Conclusions and future directions
The goal of the mutant screens in lst-1 and sygl-1 mutant backgrounds was to identify new regulators of GSC self-renewal. In particular, we sought to test the idea that the LST-1 and SYGL-1 proteins might work with other factors that were similarly redundant. The screens identified nine alleles of glp-1, two alleles of lst-1 and one allele of pole-1—the C. elegans ortholog of DNA polymerase ε. Although the screens were not saturated, identification of pole-1 with a low penetrance Glp phenotype demonstrates that additional genes likely await discovery. Any additional screens in lst-1 or sygl-1 mutant backgrounds should focus on the modified design with transgenic glp-1 to avoid isolation of more glp-1 alleles. Alternatively, one might seek suppressors of lst-1 or sygl-1 tumors (Shin et al. 2017) or enhancers of the low penetrance pole-1 Glp phenotype.
Author Contributions
A.K, H.S, K.H and J.K designed screens and methods for mutant characterization; A.K, H.S, K.H., PK-C and J.K. performed screens; H.S. and K. H. characterized lst-1 alleles; SR-T characterized glp-1 alleles; A. K. and SR-T characterized pole-1 alleles; SR-T, A.K, H.S, K.H, and J.K wrote the paper.
Acknowledgements
We thank past and present members of the Kimble and Wickens labs for thoughtful discussions during the screens. We thank Erika Sorensen for sharing glp-1(tg) prior to publication, and Jadwiga Forster for technical support. The gk49 allele was provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). SR-T was supported by the NSF Graduate Research Fellowship under Grant DGE-1256259 and NIH Predoctoral Training Grant in Genetics 5T32GM007133. JK was an Investigator of the Howard Hughes Medical Institute and is now supported by NIH R01 GM134119.