The TRIM-NHL protein NHL-2 is a Novel Co-Factor of the CSR-1 and HRDE-1 22G-RNA Pathways

Proper regulation of germline gene expression is essential for fertility and maintaining species integrity. In the C. elegans germline, a diverse repertoire of regulatory pathways promote the expression of endogenous germline genes and limit the expression of deleterious transcripts to maintain genome homeostasis. Here we show that the conserved TRIM-NHL protein, NHL-2, plays an essential role in the C. elegans germline, modulating germline chromatin and meiotic chromosome organization. We uncover a role for NHL-2 as a co-factor in both positively (CSR-1) and negatively (HRDE-1) acting germline 22G-small RNA pathways and the somatic nuclear RNAi pathway. Furthermore, we demonstrate that NHL-2 is a bona fide RNA binding protein and, along with RNA-seq data point to a small RNA independent role for NHL-2 in regulating transcripts at the level of RNA stability. Collectively, our data implicate NHL-2 as an essential hub of gene regulatory activity in both the germline and soma.

In the C. elegans germline, a diverse repertoire of regulatory pathways promote the expression of 23 endogenous germline genes and limit the expression of deleterious transcripts to maintain genome 24 homeostasis. Here we show that the conserved TRIM-NHL protein, NHL-2, plays an essential role in 25 the C. elegans germline, modulating germline chromatin and meiotic chromosome organization. We 26 uncover a role for NHL-2 as a co-factor in both positively (CSR-1) and negatively (HRDE-1) acting 27 germline 22G-small RNA pathways and the somatic nuclear RNAi pathway. Furthermore, we 28 demonstrate that NHL-2 is a bona fide RNA binding protein and, along with RNA-seq data point to a 29 small RNA independent role for NHL-2 in regulating transcripts at the level of RNA stability. 30 Collectively, our data implicate NHL-2 as an essential hub of gene regulatory activity in both the 31 germline and soma. 32

NHL-2 Physically Interacts with CSR-1 and HRDE-1 Pathway Proteins 242
Because NHL-2 had previously been shown to physically interact with miRNA Argonautes, we 243 decided to test for any association with the 22G-RNA-associated AGOs CSR-1 and HRDE-1. We first 244 performed co-Immunoprecipitation of endogenous NHL-2 and probed for CSR-1 by western blotting. 245 Indeed, we found that CSR-1 associates with NHL-2 by co-IP in adult hermaphrodites (Fig. 4A). We 246 then moved on to test for interactions with other members of the CSR-1 pathway by similar 247 experiments, and found an interaction between NHL-2 and the RdRP complex helicase, DRH-3 (Gu, et 248 al., 2009) (Fig. 4B). To test for additional protein interactions by a separate method, we incubated 249 purified GST-tagged full-length NHL-2 with protein lysate and determined whether NHL-2 was able to 250 interact with components of the 22G-RNA biogenesis machinery by western blotting. In these 251 experiments, we observed that GST-NHL-2 associated with both CSR-1 and HRDE-1 from whole adult 252 worm lysates, but not WAGO-1, or other components of the EGO-1, RRF-1 or EKL-1 (Fig. 4C, data 253 not shown). These data indicate that NHL-2 associates with CSR-1 and HRDE-1 pathway proteins at 254 multiple points of pathway activity (with DRH-3 of the RdRP complex and with AGO effectors of the 255 RISC). 256 The C-terminal RING domain of TRIM-NHL proteins is often associated with E3 ubiquitin 257 ligase activity and proteasome-mediated protein turnover. To determine whether NHL-2 functions as 258 an E3 ubiquitin ligase that targets proteins of the CSR-1 pathway for degradation, we examined CSR-1 259 expression in wild-type and nhl-2(ok818) one-day-old adult animals. We observed comparable levels 260 of both CSR-1 and DRH-3 protein in wild-type and nhl-2(ok818) animals (Supplemental Fig. S4A and 261 data not shown for DRH-3), suggesting that the association of NHL-2 and CSR-1 is not related to 262 potential E3 ubiquitin ligase activity by NHL-2. 263 264

NHL-2 Is Required for Maintaining the Steady-State Levels of 22G-RNAs 265
To explore the role of nhl-2 in the biogenesis or stability of the 22G-RNAs, we conducted small 266 RNA high throughput-sequencing of wild-type and nhl-2(ok818) animals at 20°C and 25°C 267 (Supplemental Table S1, S3). The overall size and first nucleotide distributions of small RNA species 268 in nhl-2(ok818) mutants are consistent with the wild-type controls at both temperatures (Supplemental 269 Fig. S5, S6). However, when we began to examine particular classes of small RNAs in each genotype 270 and at the different temperatures, we observed some interesting differences. 271 First, we observed overall changes in some small RNA populations in both wild-type and nhl-272 2(ok818) mutants at the higher temperature. There was an overall decrease in 21U-RNAs at 25°C 273 compared with 20ºC, for both wild-type and nhl-2(ok818) mutants (Supplemental Figs. S5, S4, S6). 274 Notably, many mutants in the piRNA and 26G-RNA pathways exhibit temperature sensitive fertility 276 defects that appear to be consistent with an overall decline in these small RNA populations at higher 277 temperatures. 21-23nt siRNA levels (which encompass the 22G-RNA category) increased slightly in 278 both wild-type and nhl-2(ok818) mutants at 25°C compared to 20ºC, while miRNA levels were 279 relatively unchanged in both strains between temperatures (Supplemental Figs. S5, S6, S7). 280 Next, we compared wild-type and nhl-2(ok818) mutant small RNA populations at each 281 temperature. We observed a consistent decrease in 22G-RNA and 26G-RNA populations in nhl-282 2(ok818) mutants relative to wild-type at 25°C (Supplemental Figs. S5, S6, S7). Concomitant with the 283 22G-RNA decreases, we observed an increase in the proportion of miRNAs in nhl-2(ok818) mutants. 284 While it is possible that this change in miRNAs is biologically meaningful, it is more likely due to a 285 "filling in" of the cloning space when small RNA populations are expressed as a proportion of the total 286 reads, as previous reports observed little change in miRNA levels in nhl-2(ok818) mutants (Hammell, 287 et al., 2009). Finally, we observed decreases in the 21U-RNA populations of nhl-2(ok818) mutants 288 relative to wild-type at 20°C, which are not evident at 25°C. 289 Because of their abundance and the link between NHL-2, CSR-1, and the RdRP complex that 290 synthesizes 22G-RNAs (via an interaction between NHL-2 and DRH-3), we examined this class of 291 small RNAs in greater depth (Fig. 5A, B, Supplemental Table S2). We observed that 573 and 724 292 genes display a two-fold or greater depletion of 22G-RNAs in nhl-2(ok818) mutants relative to wild-293 type at 20°C and 25°C respectively, with 381 genes in common between the two temperatures (the top-294 right Venn-pie diagram in Figure 5A). 295 and HRDE-1 target genes (Fig. 5A), comprising 36.5% (209/573; hypergeometric test q-value or q = 297 9.6E-80) and 36.6% (210/573; q = 2.1E-83), respectively, with 66 genes shared between the two 298 WAGOs (WAGO-1 targets are as defined in (Gu, et al., 2009), and HRDE-1 targets are as defined in 299 (Shirayama, et al., 2012). Consistent with this result, 74.5% (427/573; q = 1.6E-296) of these genes 300 were depleted of 22G-RNAs in a mutant strain that carries mutations in twelve wago (12-fold wago 301 mutant including mutations in hrde-1 and wago-1 (Gu, et al., 2009)). Notably, 33% (189/573; q = 3.7E-302 5) of these genes are CSR-1 targets, with only 17 of these genes overlapping with HRDE-1 or WAGO-303 1 targets (CSR-1 targets are as defined in (Tu, et al., 2015)). 304 Similar to the results at 20°C, 29.3% of the 724 genes depleted of 22G-RNAs in nhl-2(ok818) 305 mutants at 25°C are CSR-1 target genes (212/724; q = 3.7E-2), while 33.7% and 37.2% are WAGO-1 306 and HRDE-1 targets (244/724 and 269/724; q = 2.1E-85 and 4.7E-110; with 139 genes in common). 307 Nearly 78% (563/724, q < 1E-300) of the genes depleted of 22G-RNAs in nhl-2(ok818) mutants at 308 25°C overlap with genes depleted of small RNAs in the 12-fold wago mutant. Finally, the majority of 309 genes depleted of 22G-RNAs in nhl-2(ok818) mutants at either temperature significantly overlap with 310 genes depleted of 22G-RNAs in RdRP complex mutants, ekl-1(tm1599), ego-1(om97), and drh-311 3(ne4253) (all q-values < 7.3E-195) (Fig. 5A). The overlap between genes depleted of 22G-RNAs in 312 nhl-2(ok818) mutants and the three well characterized germline small RNA pathways, coupled with the 313 overlapping phenotypes of nhl-2(ok818) and ago mutants point to a role for NHL-2 in germline small 314 RNA pathways overall. 315 The glp-4(bn2) small RNA data set reveals the complete repertoire of germline genes targeted 316 by 22G-RNAs (Tu, et al., 2015;Gu, et al., 2009), as these mutants do not possess significant germline 317 NHL-2 in fertility and germ cell development (Fig. 1, 2), 83.8% (480/573, q = 2.6E-166) and 83.4% 319 (604/724 q = 3.9E-209) of the genes depleted of 22G-RNAs in nhl-2(ok818) mutants at 20°C and 25°C, 320 respectively, are also depleted of 22G-RNAs in glp-4(bn2) mutants (Fig. 5A). These data indicate that 321 the genes depleted of 22G-RNAs in nhl-2(ok818) mutants are targeted by 22G-RNAs in the germline. 322 Further comparison with the male-and female-specific gonad gene expression data indicated that 323 58.4% (423/724; q-value = 0.0128) and 61.6% (353/573; q-value = 1E-4) of the genes depleted of 22G-324 RNAs in nhl-2(ok818) mutants at 25°C and 20°C respectively overlap with genes generally expressed 325 in the gonad (meaning that they are expressed in both the spermatogenic or oogenic gonads) (Fig. 5B). 326 We observed a subtle, but statistically significant enrichment for oogenic genes in the nhl-2(ok818) 327 depleted gene sets (16.4%, 119/724 genes at 25°C, q = 2.2E-11, 22.5%, 129/573 genes at 20°C, q-value 328 = 2.7E-24). This observation may simply be reflective of the developmental stage from which the 329 samples were prepared (young adults undergoing oogenesis), but is also consistent with oogenesis 330 defects we observed in nhl-2(ok818) mutants. 331 332 NHL-2 Is Required for 22G-RNA Coverage at the 5´ Portion of CSR-1 Target Genes 333 The decrease in steady-state levels for a subset of 22G-RNAs in nhl-2(ok818) mutants could 334 indicate a defect in 22G-RNA synthesis or stability. Given the link between NHL-2 and the RdRP 335 complex, we hypothesized that NHL-2 could affect 22G-RNA synthesis to a greater extent than 336 turnover. Thus, we performed a metagene analysis to examine the distribution of 22G-RNAs across the 337 gene body (Fig. 5C-E, Supplemental Fig. S8). In wild-type worms, 22G-RNAs are distributed across 338 the entire length of WAGO target genes and are present in greater abundance than CSR-1 22G-RNAs. 339 bias in small RNA coverage toward the 5´ end. In nhl-2(ok818) mutants at 25°C relative to wild-type, 341 we noted that the distribution of 22G-RNAs at WAGO targets was generally unchanged over the entire 342 locus. 343 For CSR-1 target genes, the distribution of 22G-RNAs in nhl-2(ok818) mutants was distinct 344 from that in wild-type worms. At both 20°C and 25°C, there was a significant decrease in the 345 abundance of 22G-RNAs over the 5´ half of the gene in nhl-2(ok818) mutants relative to wild-type 346 controls (at 20°C, there was a 33.6% reduction in 22G-RNAs, t-test p-value = 1.4E-79; at 25°C, there 347 was a 15.5% reduction in 22G-RNAs, t-test p-value = 1.77E-9). Examined a different way, the centroid 348 of the 22G-RNA distribution for the group of CSR-1 target genes shifts by 5.77% of the metagene 349 length towards the 3´ end in nhl-2(ok818) mutants relative to the wild-type at 25°C (Wilcoxon rank-350 sum test p-value = 3.67E-29; Fig. 5C, Supplemental Fig. S8). These data suggest that NHL-2 may 351 influence the activity and/or processivity of the RdRP complex. 352 Because our small RNA results indicated a possible role for NHL-2 in RdRP complex 353 processivity or activity, and we had observed genetic and physical interactions between NHL-2 and the 354 RdRP complex via DRH-3, we asked whether NHL-2 was required for the formation or stability of the 355 RdRP complex. Therefore, we tested whether the CSR-1 RdRP complex, as measured by an 356 association between the key components DRH-3 and EGO-1, could properly form in the absence of 357 NHL-2. To answer this question, we immunoprecipitated DRH-3 and probed for EGO-1 and EKL-1, 358 three key components of the CSR-1 RdRP complex, and found that the association between DRH-3, 359 EGO-1, and EKL-1 was maintained in the absence of NHL-2 (Supplemental Fig. S4). Thus, NHL-2 is 360 not required for the formation or maintenance of the RdRP complex, and may play a different role in 361 the biogenesis of 22G-RNAs, possibly in the handoff of newly synthesized 22G-RNAs to CSR-1 362 (because of its physical association with both CSR-1 and DRH-3), aiding the assembled RdRP complex 363 in moving along the RNA template, or in the selection of particular mRNAs as templates for 22G-RNA 364 synthesis. binding preference for U-enriched RNAs, with a core consensus of UUUU, and preference for U 380 residues 5' and 3' to the core (Fig. 6B). We next examined the binding affinity of the NHL domain for a 381 5'-Fluorescein labelled 17mer poly-U RNA oligonucleotide using fluorescence anisotropy. In these 382 experiments, GST-tagged NHL domain at a range of concentrations was mixed with poly-U RNA 383 6C). These data strongly suggest that NHL-2 is a bona fide RNA-binding protein with an ability to 386 bind U rich sequences. 387 388 Steady State mRNA Levels are Altered Independent of Small RNA Levels in nhl-2(ok818) 389

Mutants 390
Because of its link to multiple small RNA pathways and its capacity to bind RNA, we asked 391 whether NHL-2 plays a role in transcript regulation. To test this, we performed mRNA-seq in wild-type 392 (N2) and nhl-2(ok818) adult animals at both 20°C and 25°C, with three biological replicates each ( Fig.  393 7A, Supplemental Table S3). At 20°C, we observed less extensive changes in steady-state mRNA 394 levels in nhl-2(ok818) mutants, with 1,014 genes increased and 1,630 genes decreased by two-fold or 395 greater. This is in contrast to 25°C, where we identified 3,554 genes with two-fold or greater increases 396 in steady-state mRNA levels, and 4,370 genes with two-fold or greater decreased steady-state mRNA 397 levels in nhl-2(ok818) mutants relative to wild-type (Fig. 7A). There was a significant overlap of genes 398 up-regulated in nhl-2(ok818) between the two temperatures (643 genes; q-value = 6.2E-249). Similarly, 399 there was a significant overlap between down-regulated genes at both temperatures (1,326 genes; q-400 value < 1E-300). We then went on to examine the genes with altered expression in more detail. 401 Based on the previously described role for NHL-2 with let-7 and lsy-6 in the miRNA pathway, 402 we first asked whether predicted targets of these particular miRNAs were de-repressed upon loss of 403 nhl-2 (Fig. 7B). Although several lines of data previously suggested that NHL-2 functioned with 404 miRISC in the translational regulation of targets, no genome-wide transcriptome data in nhl-2(ok818) 405 20 mutants were available to test the possibility that NHL-2 and miRISC could impact targets via mRNA 406 stability or turnover. Using TargetScan (Worm Release 6.2, June 2012), we identified 162 predicted 407 targets of lsy-6, and 126 predicted targets of the let-7 family of miRNAs. Only 33 of the 162 predicted 408 lsy-6 target mRNAs and 29 of the 126 let-7 family predicted targets were up-regulated in nhl-2(ok818) 409 mutants at 25°C. Examining the data the other way around, we asked if the genes up-regulated in nhl-410 2(ok818) mutants were enriched for let-7 or lsy-6 predicted targets and found no correlation (data not 411 shown). Overall, these data indicate that miRNA target genes are not regulated by NHL-2 at the level 412 of transcript abundance or stability, and instead, NHL-2 is likely to exert a predominantly translational 413 mode of regulation on these genes. 414 We next asked whether the genes with altered 22G-RNA levels in nhl-2(ok818) mutants were 415 differentially expressed (Fig. 7C, Supplemental Fig. S7). First, we overlapped the sets of genes 416 depleted of 22G-RNAs in nhl-2(ok818) mutants at either temperature and were surprised to find only 417 modest effects overall. Of the 573 genes depleted of 22G-RNAs at 20°C, 78 displayed increased 418 steady-state mRNA levels (13.6%; 78/573 genes; q-value = 3.3E-14) and 11 had decreased steady-state 419 mRNA levels (1.9%; 11/573: not significant). At 25°C, 724 genes showed depleted 22G-RNA levels, 420 and of these 210 were up-regulated (29%; 210/724, q-value = 1.8E-13), while 81 were down-regulated 421 (11.2%; 81/724: not significant). Thus, overall, genes displaying altered 22G-RNA levels were not 422 extensively affected at the mRNA level by the loss of nhl-2, and the fraction that were affected at the 423 mRNA level tended to be repressed by NHL-2 under wild-type conditions. Overall, these data are 424 consistent with a role for NHL-2 in the biogenesis, but not necessarily the effector steps, of a subset of 425 22G-RNAs. These data could also point to a role for NHL-2 in regulating the translation of this subset 426 of germline small RNA target genes for which the 22G-RNA levels are altered in nhl-2(ok818) 427 mutants, perhaps in a manner similar to the miRNA pathway. 428 Next, we examined the steady state levels by evaluating the levels of CSR-1 or WAGO pathway 429 target genes in nhl-2(ok818) mutants, independent of any changes in 22G-RNA levels. In accordance 430 with the opposing roles of the WAGO and CSR-1 pathways in germline gene regulation, we anticipated 431 that the WAGO targets with decreased 22G-RNAs would display increased mRNA levels, while the 432 CSR-1 targets with decreased 22G-RNAs would display decreased mRNA levels in nhl-2(ok818) 433 mutants. We observed that a statistically significant subset of WAGO-1 target genes (18%; 141/1718 434 genes; q-value = 9.9E-8) were de-repressed at 20°C, and 291 out of 1718 WAGO-1 target genes were 435 de-repressed at 25°C (16.9%; q = 0.6790). HRDE-1 target genes also tended to be up-regulated in nhl-436 2(ok818) mutants at both temperatures (141/1661; 8.5% q-value = 9.9E-9 at 20 o C, 335/1661; 20.2%, q-437 value = 9.9E-2 at 25 o C). These results are consistent with a cooperative role for NHL-2 in the 438 repression of WAGO target genes, albeit in a small RNA-independent manner. 439 Unexpectedly, CSR-1 target mRNAs were significantly up-regulated nhl-2(ok818) mutants, and 440 this effect occurred specifically at 25 o C (2,025/4932; 41%, q-value < 1E-300). Similarly, when 441 examined the behavior of total set of genes targeted by germline small RNAs using the glp-4(bn2) 442 mutant, we found a strong overlap between genes depleted of small RNAs in the glp-4(bn2) mutant and 443 genes up-regulated in nhl-2(ok818) mutants specifically at 25°C (60%; 2155/3554, q-value < 1E-300) 444 ( Fig. 7C). These results are consistent with a small RNA-independent role for NHL-2 in the repression 445 of CSR-1 22G-RNA targets, and this activity is antagonistic to the action of CSR-1. Moreover, the 446 functions of NHL-2 in regulating germline small RNA pathway target genes, especially those of the 447 CSR-1 pathway, appear to be particularly important at high temperature (25°C) or perhaps other 448 22 stressful conditions. Overall, the changes in steady-state mRNA levels for germline 22G-RNA pathway 449 target genes point to a role for NHL-2 in regulating transcripts via its intrinsic RNA binding capacity. 450 451

Loss of nhl-2(ok818) Impacts Genes Involved in Gametogenesis, Signaling, and Chromosome 452
Organization 453 In an attempt to link alterations in gene expression to changes in phenotype for nhl-2(ok818) 454 mutants, we next compared the nhl-2(ok818) transcriptome data to spermatogenic versus oogenic Given that NHL-2 and CGH-1 have been shown to physically interact, we next compared the 470 CGH-1 RNA-IP/microarray data with the nhl-2(ok818) transcriptome data (Fig. 7D). In the germline, 471 CGH-1 is required to protect specific maternal mRNAs (which overlap with those enriched in the 472 oogenesis and gender neutral transcriptomes) from degradation, and is also involved in translational 473 regulation of some transcripts (Boag, et al., 2008). We observed a small, but significant, enrichment 474 between CGH-1-associated and up-regulated transcripts in nhl-2(ok818) mutants at 25 o C, (12. transcripts in nhl-2(ok818) mutants at 25 o C. This is consistent with the strong overlap between genes 484 expressed during oogenesis and those upregulated in nhl-2(ok818) mutants at 25 o C. These data suggest 485 that NHL-2 does not regulate the majority of mRNAs found in the key pathways governing the oocyte-486 to-embryo transitions, however, we cannot rule out any translational effects of NHL-2 in this context. 487 When we performed Gene Ontology (GO) analysis, we found that the genes that are down-488 regulated in nhl-2(ok818) mutants at 20°C and 25°C shared consistent sets of GO terms, and are 489 strongly enriched in cuticle/collagen proteins, kinases and phosphatases, and spermatogenesis proteins 490 (FDR = 6.1E-20 protein kinase, core, 7.9E-25 phosphatase activity, and 1.5E-31 major sperm protein, 491 24 respectively). Genes that are up-regulated in nhl-2(ok818) mutants at 20°C are weakly enriched in 492 signaling molecules and oxidative metabolism (FDR = 1.4E-12 signal peptide, and 2.8E-3 oxidation 493 reduction, respectively), and differ from the GO terms observed for the genes up-regulated at 25°C, 494 which were enriched for cell cycle, kinetochore, RNA-binding and DNA replication and damage repair 495 (FDR = 5.5E-35 cell cycle, 3.7E-12 kinetochore, 1.1E-16 RNA binding, and 3.7E-14 DNA replication 496 and 9.6E-10 damage repair GO terms. In addition to analyzing the complete sets of up-regulated or 497 down-regulated genes in the nhl-2(ok818) mutants, we also performed GO analysis on sets of 498 transcripts that were expressed in the gonad, and observed comparable results (data not shown). 499 Collectively, our data point to a role for NHL-2 in regulating the stability of a large fraction of 500 gonad/germline transcripts that are involved in spermatogenesis, cellular signaling cascades, cuticle 501 formation, and kinase/phosphatase activities, and chromosome organization. NHL-2 impacts these 502 mRNAs both positively and negatively, and likely utilizes the intrinsic RNA binding properties of its 503 NHL domain, perhaps in association with other protein binding partners, to do so. and NCL-1). NHL-2 has been shown to modulate miRISC via two specific miRNAs, let-7 and lys-6, 516 which act in the soma to regulate developmental timing and cell fate transitions in multiple tissues. 517 Recently it was also shown that NHL-2 is also required for sex determination, although the mechanism 518 is unclear (McJunkin and Ambros, 2017). In spite of these intriguing roles for NHL-2 in the embryo 519 and soma, little is known about its functions in germline development. Here, we set out to explore a 520 role for NHL-2 in the germline and in germ cell development. We found that NHL-2 is required for 521 proper germline chromatin organization and wild-type levels of fertility at high temperatures and for 522 the somatic nuclear RNAi pathway. We also identified the AGOs CSR-1 and HRDE-1 and the RdRP 523 component DRH-3 as genetic and physical interactors of NHL-2. High throughput sequencing of small 524 RNA populations in nhl-2(ok818) mutants revealed an additional role for NHL-2 in the WAGO-1, and 525 HRDE-1 22G-RNA pathways, but as previous data suggested, little biologically meaningful 526 perturbation in the overall miRNA population. Binding assays confirm that NHL-2 is a bona fide RNA 527 binding protein, and examination of the mRNA transcriptome by mRNA-seq points to NHL-2 as a 528 post-transcriptional regulator of a substantial set of mRNAs involved in signaling, phosphorylation and 529 transcription independent of its small RNA activities. Together, our data, implicate NHL-2 as a 530 regulator of mRNA stability for a significant portion of the genome, a likely translational regulator of 531 miRNA targets, and a biogenesis factor and/or possible translational regulator of targets in the CSR-1 532 and WAGO 22G-RNA pathways. 533 studies, in which loss of CSR-1 pathway factors enhanced the aggregation of diakinetic oocyte 535 chromosomes in nhl-2(ok818) mutants. Loss of nhl-2 also led to increased levels of H3K9me2 in 536 pachytene germline nuclei and a spreading of this heterochromatin modification onto autosomes, where 537 it is not normally observed. This phenotype is consistent with loss of CSR-1 pathway members, 538 providing another phenotypic link between NHL-2 and the CSR-1 pathway. At this time, we do not 539 entirely understand why this phenotype emerges in nhl-2 or csr-1 pathway mutants. It is possible that 540 CSR-1 is not properly recruited to its target genes, due to mis-regulation of CSR-1 target transcripts in 541 nhl-2(ok818) mutants. This, in turn, could disrupt the formation or maintenance of euchromatin at these 542 loci and allow for the mis-direction of chromatin modifiers throughout the genome, as observed in csr-543 1 mutants (Christopher Wedeles and Julie Claycomb, unpublished results). This leads to the aberrant 544 accumulation of histone modifications throughout the genome, which could impact chromosome 545 structure. Future ChIP-seq studies for histone modifications and CSR-1 recruitment in nhl-2(ok818) 546 mutants will enable us to address this possibility. Alternatively, and based on the GO analysis, this 547 chromosome organization defect could result indirectly from alterations in the levels of key transcripts 548 associated with chromosome organization and metabolism, as has been proposed for CSR-1 (Gerson-549 Gurwitz, et al., 2016). 550 It was somewhat surprising to observe little correlation between the genes depleted of 22G-551 RNAs and genes with altered mRNA levels in nhl-2(ok818) mutants. Based on the known regulatory 552 functions of these pathways we expected there would be a slight decrease in the level of CSR-1 target 553 genes for which the 22G-RNAs were depleted, and an increase in the set of genes targeted by WAGO-1 554 or HRDE-1 for which the 22G-RNAs were depleted. Instead, we observed little change in the steady-555 state levels of transcripts with depleted 22G-RNAs, indicating that NHL-2 is involved in the translation 556 of these genes, or that there is a different role for NHL-2 with regard to these genes in the 22G-RNA 557 pathways. 558 The lack of correlation between mRNAs with altered levels and changes in 22G-RNAs raises 559 another possibility: that NHL-2 is mainly involved mainly in the biogenesis of a subset of the 22G-560 RNAs. This model seems plausible for several reasons. First, DRH-3 and NHL-2 physically interact by 561 co-IP. Second, our metagene analysis of the distribution of 22G-RNAs along the length of target 562 mRNA transcripts is similar to the pattern observed for drh-3(ne4253) mutants, in which the 22G-563 RNAs are reduced along the length of the gene body, with most significant decreases present at the 5ʹ 564 end of the transcript. This pattern is consistent with a role for NHL-2 in the processivity or activity of 565 the RdRP complex on a subset of CSR-1 targets (Fig. 8). The RNA binding activity of NHL-2 points to 566 a model whereby NHL-2 could help to identify particular mRNAs as candidates for 22G-RNA 567 synthesis. Furthermore, because NHL-2 also associates with CSR-1 and HRDE-1 as well as DRH-3, it 568 could also act as a chaperone required efficient handoff of 22G-RNAs from the RdRP complex to the 569 Argonaute (Fig. 8). 570 This potential role for NHL-2 with the RdRP complex is noteworthy for several reasons. First, 571 because TRIM/NHL proteins have thus far only been implicated in the effector step of miRNA 572 pathways, this is the first indication that NHL-2 (and thus TRIM/NHL proteins) could also be involved 573 in the biogenesis of endo-siRNAs. Second, we still have relatively little insight into the factors that 574 route particular transcripts into the 22G-RNA pathways, and NHL-2 provides an attractive candidate 575 for one such factor. Third, the role for NHL-2 in biogenesis of a subset of germline 22G-RNAs and the 576 effector steps of somatic miRNAs demonstrate differential roles for this intriguing protein in the 577 28 germline versus soma, and points to differences in protein binding partners in each of these tissues that 578 should be examined further. 579 Notably, many of the up-regulated genes in nhl-2(ok818) mutants were the targets of 22G-580 RNAs, yet these genes did not display alterations in the levels of 22G-RNAs. These data point to a role 581 for NHL-2 in regulating transcript stability, in cooperation with the WAGOs, and in opposition to 582 CSR-1, and suggest a combinatorial regulatory mechanism that engages both small RNA pathways and 583 bona fide RBPs such as NHL-2. The fact that nhl-2(ok818) mutants display temperature dependent 584 fertility defects is consistent with several small RNA pathway factors, including both the 585 piRNA/WAGO and the CSR-1 pathways and points to NHL-2 as a co-factor and/or co-regulator 586 required for optimal pathway activity under stressful conditions. Although NHL-2 was not identified 587 previously as a factor in the piRNA or 22G-RNA pathways, our results exemplify the power of 588 synthetic genetic screens to identify accessory factors involved in the optimal function these pathways. 589 In light of our data, we propose that NHL-2 acts as a hub of gene regulation, where it works 590 cooperatively with core factors in a diverse set of pathways that are central to both somatic and 591 germline gene regulation (Fig.8). NHL-2 localizes to several ribonucleoprotein structures involved in 592 RNA regulation, including P granules in the germline, CGH-1 granules in the gonad core, and In conclusion, we characterized the roles of NHL-2 in the germline and showed it localizes to P 612 granules and impacts a subset of 22G-RNAs in both the CSR-1 and WAGO/HRDE pathways. This 613 germline role in small RNA biogenesis is distinct from its role in the miRNA pathway in the soma, and 614 implicates NHL-2 in RdRP activity in the germline. Interestingly, also NHL-2 was required for the 615 nuclear RNAi pathway, suggesting that NHL-2 is a promiscuous co-factor of multiple distinct, but 616 related small RNA pathways. NHL-2 displays intrinsic RNA binding ability via its NHL domain and 617 thus is capable of binding and regulating the stability or translation of a large number of germline 618 transcripts in a small RNA-independent manner. NHL-2 may exemplify a new class of co-factor that is 619 required for optimal activity of small RNA pathways (both miRNA and 22G-RNA pathways). 620 Although this type of co-factor appears to be extremely important for the fidelity and robustness of nhl-2(ok818) mutants at 20°C vs. 25°C. 798 Figure S7. The distribution of 22G-RNAs is reduced at the ´5-end of CSR-1 targets in nhl-2(ok818) 799 relative wild-type control at 25°C. Related to Figure 4. 800 Figure S8. The RdRP complex is intact in nhl-2(ok818) mutants, Related to Figures 2, 4. 801  Quantification of chromosomal defects in diakinetic oocytes in wild-type and nhl-2(ok818) animals at 986 25°C. (E) NHL-2 co-localizes with CGH-1 in P-granules (top panels, Surface) and gonadal core CGH-987 1 granules (bottom panels, Core). DAPI (blue), NHL-2 (red), CGH-1 (green). Scale bar 10µm. 988 Immunostaining with affinity purified NHL-2 antibody was reduced to background in nhl-2(ok818) 989 germlines (Fig. S2A).  Table S2: Genes altered in small RNA sequencing datasets generated and investigated in this paper. 1147