GLP-1 Notch - LAG-1 CSL control of the germline stem cell fate is mediated by transcriptional targets lst-1 and sygl-1

Stem cell systems are essential for development and maintenance of polarized tissues. Intercellular signaling pathways control stem cell systems, where niche cells signal stem cells to maintain the stem cell fate/self renewal and inhibit differentiation. In the C. elegans germline stem cell system, GLP-1 Notch signaling specifies the stem cell fate. However, the downstream transcriptional targets of GLP-1 signaling that mediate the stem cell fate have not been fully enumerated. We employed a genome-wide approach to uncover transcriptional targets of GLP-1 signaling - the intersection of genes identified as directly bound by LAG-1, the C. elegans Notch pathway sequence-specific DNA binding protein, from ChIP-seq experiments, with genes identified as requiring GLP-1 signaling for RNA accumulation, from RNA-seq analysis. lst-1 and sygl-1, genes previously identified as transcriptional targets from a bioinformatic candidate gene approach, were bound by germline LAG-1 and their expression dependent on glp-1 and germline lag-1 activity. No additional genes were identified as both bound by LAG-1 and whose mRNA level was dependent on glp-1 and lag-1. Genes were identified as likely secondary effects of GLP-1 signaling with the properties that their glp-1 dependent mRNA accumulation could be explained by a requirement for lst-1 and sygl-1 activity and their lack of LAG-1 binding. Furthermore, glp-1 dependent peak accumulation of FBF-2, which promotes the stem cell fate, is explained by a requirement for lst-1 and sygl-1 activity. Finally, we showed that lag-1 is germline autonomously required for the stem cell fate and that elevated LAG-1 accumulation is spatially limited to the stem cell region by posttranscriptional regulation that, in part, requires lst-1 and sygl-1. These findings are consistent with the possibility that lst-1 and sygl-1 are the sole germline GLP-1 signaling mRNA transcriptional targets, which largely or completely mediate the stem cell fate.

sygl-1 activity. Finally, we showed that lag-1 is germline autonomously required for the 48 stem cell fate and that elevated LAG-1 accumulation is spatially limited to the stem cell both germline LAG-1 binding and whose expression was dependent on glp-1 and 164 germline lag-1 activity. We used a time course following auxin-induced degradation of 165 LAG-1 to distinguish between primary versus secondary changes in RNA level in GLP-1 166 signaling. lst-1 and sygl-1 were the first genes (2hrs treatment) whose mRNA level were 167 dependent on LAG-1 activity, representing a primary effect. Five additional genes were 168 then identified at a later time point (4hrs treatment) whose RNA accumulation was 169 dependent of LAG-1 activity. Consistent with these changes being a secondary effect, Following high throughput sequencing, peaks were identified with greater than a two-353 fold enrichment compared to input control, and a false discovery rate (FDR <0.05). The 354 Homer suite (Heinz et al. 2010) was used to annotate the peaks to their nearest 355 transcription start site (TSS). Seventy-six genes were identified as binding to LAG-1 356 from the intersection of the FLAG-IP (one biological replicate) and GFP-IP (two 357 biological replicates) ChIP-seq (S4 Fig; S3 Table). Known Table). This maybe a result of our analysis being from only a single stage 374 (mid-L4) or because only a small number of cells are expressing the target gene under 375 LIN-12 or GLP-1 control, resulting in only a small amount of total LAG-1 bound DNA, 376 which is below the limit of detection in the ChIP-seq assay. We performed de novo 377 discovery of over-represented DNA sequence motifs among the 75 genes and found the 378 highest hit to be an 9-mer that contains the canonical CBF/LAG-1 binding site (p-value: 379 e-42 ) (S4H Fig). Thus, we believe that the bulk of the genes identified are bound by LAG- To identify genes whose transcription promotes the stem cell fate through direct LAG-1 384 binding, we performed germline specific LAG-1 ChIP-seq, where the approach was 385 guided by our expression and germline autonomy analysis above. LAG-1 is modestly 386 expressed in germline stem cells compared to much higher expression in late stage 387 pachytene and diplotene oogenic germ cells and somatic cells (Fig 2, 3). This 388 necessitated performing germline LAG-1 ChIP-seq on L4 stage hermaphrodites that 389 lack oogenic germ cells in pachytene and diplotene. We generated a transgenic strain 390 with the following components: (1) A fosmid transgene where the BioTag, a 23-amino 391 acid peptide that is recognized and biotinylated by the E. coli enzyme BirA biotin ligase 392 (Cronan 1990), was placed at the C-terminus of LAG-1 (ozIs43[lag-1::3xFLAG::BioTag, The lag-1 deletion allele, tm3052, which demonstrated that lag-1::BioTag in ozIs43 396 produced functional, rescuing, LAG-1. Germline specific expression of BirA should 397 result in germline restricted biotinylation of LAG-1::BioTag, which can be pulled down by 398 streptavidin beads (Fig 4A) we performed sequential ChIP, first with anti-FLAG and then with streptavidin beads, 402 followed by library construction directly on the beads due to ultra-low quantities of DNA 403 following sequential ChIP (Fig 4B). Germline specific LAG-1 ChIP-seq data was 404 analyzed as described for whole worm ChIP-seq. 405 406 One hundred and thirty seven genes with germline specific LAG-1 peaks were identified 407 in common between three biological replicates, including lst-1, sygl-1 and mir-61/250 408 (Fig 4C-D). The lag-1 gene was also found to contain multiple peaks. However, other 409 linked genes on the fosmid (e.g., zen-4) and the selectable marker used (unc-119) also 410 contained multiple peaks, which were absent in the whole worm ChIP-seq. Thus, it is 411 likely that there are multiple integrated copies of ozIs43, leading to an elevation of 412 background sequences being pulled down and detected in the germline ChIP-seq 413 experiments, therefore these genes were not considered to be specifically bound by 414 germline LAG-1 . De novo motif identification recovered the same 9-mer that contains 415 the canonical CBF/LAG-1 binding site (p-value: e-34 ) as found in the whole worm 416 experiment, consistent with LAG-1 binding many of these genes in vivo, in germ cells. 417 Thirty-six genes were identified in common between germline specific and whole worm 418 that show LAG-1 ChIP-seq peaks, supporting that LAG-1 binds to these genes in vivo 419 signaling OFF, using the null allele q175. As described above, we employed the meiotic 430 entry defective gld-2 gld-1 double null mutant to allow examination of the effect of glp-1 431 null in proliferating germ cells; glp-1 gf was also placed in this background so that 432 genotype was identical, except for glp-1 activity status (Fig 5A; Materials and  433 Methods). Further, we used dissected gonad preparations to significantly enrich for 434 expression changes that occur in the germline. In situ hybridization and qRT-PCR was 435 used to confirm that in GLP-1 ON, transcriptional target sygl-1 and lst-1 were expressed 436 throughout the germline at significantly elevated levels, compared to GLP-1 OFF where 437 expression was similar to background (Fig 5B-C Table). 445 Intersection of the 137 genes identified as bound by LAG-1 with the 94 genes whose 447 RNA accumulation was dependent on GLP-1 signaling yielded only two genes, lst-1 and 448 sygl-1 (Fig 4, 5). Three  transcriptional targets of GLP-1 signaling (Fig 4C-D). Together, the above experiments found that FBF-2::V5 displayed a peak of accumulation at 8 -13 CD from the distal tip, 479 followed by a somewhat gradual fall to low proximal levels by ~35 CD (Fig 6A-B), 480 similar to previously reported (Lamont et al. 2004), with peak accumulation ~4 fold 481 higher than base. In gld-2 gld-1 double null mutant germlines, the FBF-2::V5 peak was 482 ~80% of wild type, with a the fall more rapid than in wild type and a flat base from ~18 483 CD through 35 CD, with the peak also ~4 fold higher than base (Fig 6A, C). In the gld-2 484 gld-1; glp-1 triple null mutant germlines, FBF-2::xV5 levels are low throughout the 485 progenitor zone (Fig 6A, C-D), with the level ~4 fold lower than in the gld-2 gld-1 double progenitor zone (Fig 3), we next examined whether LST-1 and SYGL-1 promoted peak 491 FBF-2 accumulation. Analysis of gld-2 gld-1 lst-1 sygl-1 quadruple null mutant germlines 492 showed low FBF-2 accumulation throughout the progenitor zone (Fig 6), consistent with 493 LST-1 and SYGL-1 being required for peak FBF-2 accumulation. FBF-2 levels at the 494 peak position in wild type are almost 2 fold lower in the absence of lst-1 and sygl- 1,495 compared to the absence of glp-1; the reason for this difference is unknown. In the 496 proximal progenitor zone, GLD-1 appears to promote FBF-2 accumulation as the level 497 is lower in the gld-1 null mutant (S7 Fig). Since GLD-1 acts in translational repression, 498 the effect on FBF-2 accumulation is presumably indirect. In summary, GLP-1 dependent 499 peak FBF-2 accumulation in the progenitor zone appears to be completely explained 500 through post-transcriptional regulation by lst-1 and sygl-1 activity (Fig 6E). Drosophila, it is known that in the absence of Notch ICD, Su(H) CSL functions in 508 actively repressing Notch transcriptional target genes (Bray 2006). To test if LAG-1 has 509 a GLP-1 signaling independent transcriptional function in the germline, we performed 510 LAG-1 AID followed by gonad dissection and RNA-seq (Fig 7). We used a strain 511 containing lag-1::degron, glp-1(ar202) gf that at the restrictive temperature will have the 512 bulk of germ cells undergoing GLP-1 signaling, and the meiotic entry defective double 513 mutant gld-2 gld-1, to allow analysis of proliferating germ cells, with or without auxin 514 treatment. Germline specific degradation of LAG-1 was achieved with gld-515 1p::tir1::mRudy as described above (Fig 2D, Materials

and Methods). 516
We performed RNA-seq following 48hrs of auxin treatment from the L1 stage at 25 o C, a 517 time where LAG-1 is undetectable, and the same time point where GLP-1 ON and OFF 518 RNA-seq was performed (Fig 5). Heatmap and principal component analysis 519 demonstrated that changes in RNAs from LAG-1 ON (minus auxin) and LAG-1 OFF 520 (plus auxin) were significantly different, while biological replicates were similar (S8A-B; 521 S6 Table). Ninety-four genes were identified whose expression was dependent on LAG-522 1 (activated genes), where ~70% (64) of these were also GLP-1 dependent genes (S8C 523 The LAG-1 AID system described above provides a route to distinguish between genes 548 whose expression is a primary or a secondary effect of GLP-1 signaling/LAG-1 activity. 549 By performing a time course of LAG-1 degradation, using RNA-seq as the readout, 550 genes whose RNA levels are transcriptionally controlled by GLP-1 signaling should 551 change expression earlier, while genes whose RNA level is indirectly controlled/a 552 secondary effect of GLP-1 signaling, should change expression later. 553

554
We performed a LAG-1 AID time course, harvesting RNA following 0.5, 1, 2, 4 and 555 48hrs of auxin treatment (Fig 7). sygl-1 and lst-1 mRNA levels, as assessed by qRT-556 among the GLP-1 dependent genes, only lst-1 and sygl-1 were identified as LAG-1 568 mRNAs targets (Fig 7B-D). By 4hrs of auxin treatment, 5 additional genes were 569 identified whose RNAs displayed LAG-1, as well as GLP-1, dependent accumulation 570 (Fig 7D-E). The new 5 genes from the 4hr time point could represent additional GLP-1 -571 LAG-1 transcriptional targets, which have longer RNA half-lives than lst-1 and sygl-1, or 572 they may represent changes in RNA level that are secondary effects of GLP-1 signaling. 573 We did not detect germline LAG-1 ChIP-seq peaks or canonical LAG-1/CSL binding 574 sites in these 5 genes, indicating that they are not GLP-1 transcriptional targets. From 575 these results, and our previous finding that lst-1 and sygl-1 promote LAG-1 and FBF-2 576 protein accumulation, we reasoned that the accumulation of RNA for these 5 genes 577 maybe dependent on lst-1 and sygl-1 activity. 578

579
We first examined the kinetics of loss of LST-1 and SYGL-1 following LAG-1 AID. Peak 580 LST-1 falls ~5 fold at 2hrs and ~10 fold at 4hrs, while SYGL-1 falls ~3 fold at 2hrs and 581 10-fold at 4hrs auxin treatment (S10 its mRNA read count was below the 2 CPM cut-off for reliable RNA level assessment. 635 From transcriptomics analysis, while we identified genes whose RNA accumulation was 636 GLP-1 signaling dependent (94) or germline LAG-1 dependent (94), other than lst-1 and 637 sygl-1, they lacked germline or whole worm LAG-1 ChIP-seq peaks. From the LAG-1 638 AID time course, five genes were identified where their RNA level was dependent of 639 LAG-1 at 4hrs of auxin treatment, raising the possibility that they were direct/primary 640 targets of LAG-1 activity. However, in addition to the five genes lacking LAG-1 ChIP-641 seq peaks, RNA accumulation for three of the genes was dependent on lst-1 and sygl-1. 642 Thus, our genome-wide studies did not identify any new candidate GLP-1 signaling 643 mRNA transcriptional target genes. However, the RNA-seq approach employed cannot 644 assess the level of small RNAs (less than ~100nt), and a number of small RNA genes 645 were identified as containing germline LAG-1 ChIP-seq peaks (e.g, mir-61/250, five 21 646 U RNA genes). Thus, it is possible that there are small RNA genes that are germline 647 GLP-1 signaling transcriptional targets. We note that because of the low amount of DNA 648 that was obtained for the germline LAG-1 ChIP-seq experiments, it is possible that we 649 missed some weak ChIP-seq peaks. We also note that all of the transcriptomics 650 analyses were performed using dissected gonads from the gld-2 gld-1 meiotic entry 651 is not controlling lag-1 mRNA level, we nevertheless found that 61% of peak LAG-1 675 32 accumulation was GLP-1 signaling dependent, suggesting a posttranscriptional 676 mechanism (Fig 3E). This is consistent with the observation that many genes in the reported that distal peak FBF-2 accumulation requires GLP-1 signaling. We found that 718 the glp-1 dependent peak FBF-2 accumulation can be fully explained by the activity of 719 lst-1 and sygl-1 (Fig 6). FBF-2 activity functions in repression of the GLD-1 and GLD-2 720 meiotic entry pathways. As described above, ~60% of peak LAG-1 accumulation is 721 GLP-1 dependent, about 50% of which can be attributed to lst-1 and sygl-1 activity. 722 From the LAG-1 AID time course, we identified 5 genes whose RNA accumulation 723 depended on LAG-1, as well as glp-1 activity. For epg-5, tbx-2 and C17H12. 36

Strain maintenance 748
Unless otherwise noted, C. elegans strains were maintained at 20°C through 749 conventional methods (Brenner 1974). The animals were grown on NGM plates seeded 750 with OP50 bacteria. glp-1(ar202) is a temperature sensitive (ts) allele and strains with 751 this allele were maintained at 15°C. A complete list of strains used in this study is 752 provided in S1 Table.  protein function in the germ cells (Fig 2A). This allele was generated through the Self 774 injected animals were raised at 20°C for three days prior of adding 500 µl of 5 mg/µl 784 hygromycin. Six days later, animals that survived the antibiotic treatment and without 785 myo-2p::gfp injection marker were selected for PCR to screen for inserts (see S2 Table  786 for oligonucleotide information). lag-1(oz536)(rollers, with SEC in) was verified through 787 BS1193 (S1 Table) and used for germline LAG-1 ChIP-seq analysis (Fig 4). 799 800

Auxin treatment 801
The auxin treatment was used to degrade endogenous LAG-1 protein in the germline. were done in NGM plates supplemented with 1 mM auxin (auxin plates). The un-seeded 806 auxin plates were stored at 4°C in the dark, and used within two weeks. The auxin 807 plates were seeded with OP50 bacteria and kept at 20°C for 24 hours before use. The 808 auxin treatment was conducted either at 20°C (Fig 2E) or at 25°C (Fig 7, S8- Apart from the protein to be quantified, gonads were co-stained with DAPI and antibody 847 against WAPL-1. DAPI is used to count and mark cell diameter (cd) and WAPL-1 was 848 used to measure progenitor zone length. WAPL-1 staining was also used to distinguish 849 between different genotypes that were dissected and stained together. Hyperstack 850 images were captured using a 63X objective lens on a spinning disk confocal 851 microscope (PerkinElmer-Cetus, Norwalk, CT). Exposure time, which is kept constant 852 for an individual experiment, is set by using auto exposure in Volocity software (Perkin-853 Elmer) for each experiment using an epitope-tagged strain in wild type background. To The procedure for In situ hybridization is adopted from Jones et al., 1996. Briefly, young 878 adult animals were dissected in a glass dish. The dissected gonads were fixed with 3% 879 paraformaldehyde / 0.25% glutaraldehyde / 0.1 M K2HPO4 (pH7.2) for 2 hours at RT, 880 followed by post-fixed with 100% methanol at -20°C. After three washes in PBST to 881 remove residual methanol, the gonads were incubated with 50 µg/ml protease K in 882 PBST for 30 mins, followed by 15 mins re-fixation in 3% paraformaldehyde / 0.25% lacks LAG-1 from late stage of oogenesis that is not involved in GLP-1 signaling (Fig 2). 961 However, direct ChIP with streptavidin beads was not successful, likely due to issue. First, 300 µl FLAG M2 beads/6 grams of worm powder per replicate (Sigma, 965 #M8823) was first used to precipitate all LAG-1 proteins from both germline and somatic 966 tissues, following the same procedure from whole worm ChIP. After eluting in 1% SDS, 967 the DNA-protein complexes was diluted five times and re-immunoprecipitated with 968 streptavidin beads to specifically pull-down biotinylated LAG-1. Since the binding 969 between biotin and streptavidin is very stable, the precipitated DNA-protein complex 970 was washed in 2% SDS, 1% SDS for 10 mins each to remove any none specific 971 binding. Unlike the whole worm ChIP, the DNA amount after sequential ChIP was ultra 972 low, so the NGS libraries were prepared directly with DNA on the streptavidin beads 973 following manufacturer's instructions (Kapa Biosystem, #KK8500).