ELLI-1, a novel germline protein, modulates RNAi activity and P-granule accumulation in Caenorhabditis elegans

Germ cells contain non-membrane bound cytoplasmic organelles that help maintain germline integrity. In C. elegans they are called P granules; without them, the germline undergoes partial masculinization and aberrant differentiation. One key P-granule component is the Argonaute CSR-1, a small-RNA binding protein that antagonizes accumulation of sperm-specific transcripts in developing oocytes and fine-tunes expression of proteins critical to early embryogenesis. Loss of CSR-1 complex components results in a very specific, enlarged P-granule phenotype. In a forward screen to identify mutants with abnormal P granules, ten alleles were recovered with a csr-1 P-granule phenotype, eight of which contain mutations in known components of the CSR-1 complex (csr-1, ego-1, ekl-1, and drh-3). The remaining two alleles are in a novel gene now called elli-1 (enlarged germline granules). ELLI-1 is first expressed in primordial germ cells during mid-embryogenesis, and continues to be expressed in the adult germline. While ELLI-1 forms cytoplasmic aggregates, they occasionally dock, but do not co-localize with P granules. Instead, the majority of ELLI-1 aggregates accumulate in the shared germline cytoplasm. In elli-1 mutants, several genes that promote RNAi and P-granule accumulation are upregulated, and embryonic lethality, sterility, and RNAi resistance in a hypomorphic drh-3 allele is enhanced, suggesting that ELLI-1 functions with CSR-1 to modulate RNAi activity, P-granule accumulation, and post-transcriptional expression in the germline.

Introduction P granules are required in the C. elegans adult germline to maintain fertility and protect germ cell fate [1]. Recent studies confirm the overlap between P-granule function and siRNA regulation. PGL-1, which nucleates P-granule formation, is RNAi defective (Rde) [2]; while VASA/ GLH-1, another constitutive P-granule component, interacts directly with DCR-1 to regulate P-granule structure [3]. The Argonaute CSR-1 is also recruited to P granules through the Dicer-Related Helicase DRH-3, the RNA-Dependent RNA Polymerase (RDRP) EGO-1, and a Tudor-Domain protein called EKL-1 [4]. DRH-3, EGO-1, EKL-1, and CSR-1 constitute the CSR-1 22G RNA complex and each are essential for fertility and efficient RNAi [5][6][7][8][9][10], though it is unknown whether this RNAi defect is direct or an indirect consequence of having enlarged P granules and compromised PGL-1 [4,11,12]. Determining the cause and the consequence of the Rde phenotype in P-granule assembly and CSR-1 pathway mutants is a challenge because both types of mutations affect P-granule structure and morphology.
Recently, mRNA-seq of dissected P-granule and csr-1 depleted germlines revealed a strong correlation in regulated transcripts, with the majority of upregulated genes in both P-granule and csr-1 depleted germlines found in sperm-enriched datasets [13]. This is accompanied by the distal expansion of sperm transcripts into what become intersexual germ cells that rarely fertilize. This partial masculinization of developing oocytes after depletion of csr-1 is likely explained by a 9.5-fold overexpression of fog-3, the effector of the germline sex determination pathway [13]. A separate more recent report profiled gene expression changes in whole worms where CSR-1's enzymatic RNA slicing activity was inactivated, which also found significant enrichment in sperm transcripts and a 13.9-fold increase in fog-3 expression; however, worms with inactive CSR-1 slicer activity have somewhat higher broods than csr-1 deletions and no reported P-granule phenotype [14]. These observations may suggest that CSR-1's slicing activities repress masculinization of developing oocytes while CSR-1's slicer independent functions maintain P-granule integrity. These slicer-independent functions could involve translational regulation, which cannot be directly assessed in genome-wide RNA expression studies [15]. In addition to antagonizing sperm-specific transcripts in developing oocytes [5,13], P-granule assembly [4,11,12], and small RNA biogenesis [4,[16][17][18], CSR-1 and its cofactors have also been implicated in mitosis and meiosis [7][8][9]19,20], transcription [21], chromatin condensation and kinetochore assembly [4,[22][23][24], H3K9me2 distribution [22,25], histone and histone mRNA maturation [26], alternative splicing [27], and the epigenetic licensing of germline transcripts for expression [28][29][30][31][32][33][34]. While some of these roles will likely be more direct than others, the importance of CSR-1 in the C. elegans germline cannot be understated, and finding additional factors that interact with the CSR-1 complex is imperative to parsing out its functions.
Here we report results of a forward mutagenesis for regulators of P-granule assembly. Using this unbiased approach, we identified a class of enlarged P-granule mutants harboring loss-of-function alleles of all four known CSR-1 complex components, in addition to loss-offunction mutations in a novel gene we've named elli-1 (enlarged germline granules). We also describe ELLI-1 expression in the germline, differential gene expression in elli-1 mutants, and elli-1's genetic interaction with the CSR-1 pathway.

Forward genetic screen identifies mutants with enlarged P granules
To discover regulators of P-granule accumulation, EMS mutagenesis was performed on a C. elegans strain with an integrated transgene expressing the constitutive P-granule component, PGL-1, tagged with GFP ( Fig 1A) [35]. The F2 generation was then screened for accumulation of large and more intense PGL-1::GFP granules (Fig 1B), similar to those previously observed with csr-1 RNAi [12]. Ten independent, fully-penetrant alleles were isolated, and single nucleotide polymorphisms mapped six alleles to three complementation groups on chromosome I, and four alleles to two complementation groups on chromosome IV (Table 1). PGL-1 intensity was imaged under fixed exposure conditions and quantified, with sam2 exhibiting the smallest increase in PGL-1 intensity to sam14 and sam19 exhibiting the highest intensity ( Fig 1C). A closer comparison of P-granule size between wild-type, sam3, and sam18 shows that P granules are two to three times larger in sam3 and sam18 mutants, suggesting that the observed increase in intensity reflects both larger P granules and more abundant PGL-1::GFP ( Fig 1D).
To determine if alleles mapping to chromosome IV contain mutations in csr-1/F20D12.1, complementation tests were performed with csr-1(fj54). Large and bright P granules were observed in both csr-1(fj54)/sam15 and csr-1(fj54)/sam18 cross progeny, showing that these alleles failed to complement, while P granules in csr-1(fj54)/sam3 and csr-1(fj54)/sam6 cross progeny were normal (n>15 progeny examined from each cross). Sequencing sam15 revealed a 169 base pair deletion in the PAZ domain of csr-1 that causes a frameshift starting at amino acid 384 of F20D12.1a, and an early stop codon 14 amino acids later (Fig 2A). Sequencing sam18 revealed a G to A point mutation in csr-1's PAZ domain that causes an early stop codon at amino acid 443 of F20D12.1a. Like other csr-1 alleles, sam15 and sam18 have very small broods of early arrested embryos and must be maintained over a balancer. The isolation of two csr-1 alleles from this screen demonstrates the specificity of the large P-granule accumulation phenotype, making it likely that this screen isolated alleles of known CSR-1 co-factors.
Hawaiian Variant Mapping was used in combination with genome-wide sequencing and the Cloudmap pipeline to identify the remaining alleles on chromosomes I and IV [35,36]. On chromosome I, mutations were identified in three known csr-1 cofactors. Both sam14 and sam16 contained point mutations that introduced stop codons in the RNA dependent RNA polymerase (RdRP) domain of ego-1, while sam7 contained an N-terminal missense mutation in the same gene (Fig 2B), and all three alleles are homozygous sterile. sam19 contains a point mutation that introduces an early stop codon in the first Tudor domain of ekl-1, and is also homozygous sterile (Fig 2C). sam5 contains a G to A point mutation that introduces an early stop codon in exon 10 of drh-3, while in sam2 a C to T missense substitution (Leu307Phe) was discovered in exon 5 of the same gene. This mutation lies directly before the DEAD/DEAH helicase coding domains of drh-3 and causes a relatively minor change, reflecting the subtler phenotype of sam2 homozygotes, which are only sterile when raised at a higher (25˚C) temperature ( Fig 2D). CRISPR was used to evaluate whether this simple substitution of hydrophobic residues was sufficient to cause a P-granule phenotype. A C to T base pair change recapitulating the original mutation plus a silent mutation to prevent Cas9 re-cleavage [drh-3(sam27)] was introduced in wild-type worms, which were then crossed into a new CRISPR-derived pgl-1(sam33[pgl-1::gfp::3xFLAG])line. drh-3(sam27) reproduced the enlarged PGL-1::GFP expression, confirming that this missense mutation is responsible for the P-granule phenotype of drh-3(sam2) (Fig 2E).
Because this screen identified multiple alleles of csr-1 and its cofactors ego-1, ekl-1, and drh-3, other mutations with identical P-granule phenotypes likely interact with the same complex. Only one complementation group consisting of sam3 and sam6 remained. Following the sequencing strategy above, it was discovered that both alleles contain a G to A point  Fig 3A). This mutation is not found in the parental strain or the csr-1, ego-1, ekl-1 or drh-3 mutant strains. Given that sam3 and sam6 were isolated from different F1s following mutagenesis, these two alleles may have arisen in pre-meiotic germ cells from the same mutagenized animal. However, because of a closely linked lethal mutation in sam6, it cannot be passaged as a homozygous strain like sam3. Targeting F20C5.3 with RNAi causes bright, enlarged P granules (in 124/140 RNAi fed worms) (Fig 3B), and injection of a fosmid carrying F20C5.3 rescues the P-granule phenotype in sam3 mutants (36/38 transgenic animals rescued) (Fig 3C). F20C5.3 has now been renamed elli-1, for its enlarged germline granules.
The PGL-1::GFP reporter used in the screen is likely a low copy transgene integrated on chromosome I; however, the actual copy number is unknown. To ensure that the enlarged and bright P-granule phenotype of elli-1(sam3) was not simply de-silencing the integrated transgene, PGL-1::GFP was crossed out of elli-1(sam3) worms and endogenous PGL-1 was imaged with a PGL-1 antibody ( Fig 3D). The size and intensity of PGL-1 granules were higher in elli-1(sam3) compared to the control (p = 0.0001) and indistinguishable from those in csr-1(fj54) germlines ( Fig 3D, representative images of 10 fixed and stained germlines for each condition). In both elli-1(sam3) and csr-1(fj54) germlines, some PGL-1 granules were observed to detach from the nuclear envelope (mAb414 stain in red), likely contributing to the accumulation of excess PGL-1::GFP and endogenous PGL-1 granules in the rachis of elli-1 mutants (Fig 3E, 8/8 elli-1 compared to 0/8 N2, p = 0.0002). This detachment may be caused by the spherical shape of larger granules, decreasing the surface in contact with the nuclear periphery. Worms lacking the GFP transgene were then crossed into a new CRISPRderived glh-1(sam24[glh-1::gfp::3xFLAG]) line. The size and intensity of endogenous GLH-1 tagged with GFP also increased in germ cells of elli-1(sam3) mutants, showing that the increase in P-granules size is not specific to PGL-1 ( Fig 3F). Constructs were generated to reintroduce the elli-1(sam3) point mutation into the GFP::PGL-1 transgenic strain using CRISPR. From this experiment, two lines were generated ( Fig 3A): one with the desired point mutation plus three silent mutations to prevent Cas9 re-cleavage [elli-1(sam21)], and one complex rearrangement near the mutated base [elli-1(sam22)]. The complex rearrangement in elli-1(sam22) causes a Pro!Leu mutation at 179aa of F20C5.3a, a frameshift at 180aa, and a stop codon 34 aa later. Both elli-1(sam21) and elli-1(sam22) exhibit a fully-penetrant and identical P-granule phenotype to elli-1(sam3) and elli-1(sam6). Taken together, these results demonstrate that elli-1/F20C5.3 loss-of-function causes the accumulation of large P granules in the C. elegans germline.

ELLI-1 is a novel germline component that functions in early development
ELLI-1 is a nematode-specific protein with no discernable domains. MobiDB predicts that most of ELLI-1 contains disordered regions (Fig 3A, green bars) [37,38]. RNA-seq of dissected germlines shows elli-1 transcripts are expressed more abundantly than 91% of genes [13], and expression profiling in germline-less animals previously showed that elli-1 is germline enriched [39,40]. To visualize elli-1 transcripts in whole worms, fluorescent in situ hybridization (FISH) was performed [13], showing that elli-1 mRNA is indeed enriched in the germline (Fig 4A). To observe the sub-cellular localization of ELLI-1 protein, CRISPR was used to place a C-terminal GFP-3xFLAG tag on endogenous elli-1 [41]. A diffuse GFP signal can be observed at low levels in the germline of these living worms, and M2 FLAG antibody shows diffuse ELLI-1 expression in germline cytoplasm with some small ELLI-1 foci (Fig 4B). While 4.3% of these foci can be found docked next to P granules ( Fig 4B, arrowheads), the majority of ELLI-1 foci accumulate in the rachis (averaged from 10 germlines). Within the rachis, ELLI-1 foci partially overlap with the P-body component CGH-1 (Fig 4C), suggesting these foci are associated with RNA. However, ELLI-1's RNA-binding affinity has yet to be examined and may be indirect. We sought to determine whether ELLI-1 colocalizes with PATR-1, a P-granule component found only in P bodes but not P granules, but an available PATR-1 antibody did not work in dissected germlines, and a CRISPR-derived PATR-1::GFP strain generated for this study was surprisingly diffuse in the adult germline and did not form P-body-like foci. Zygotic ELLI-1 begins to accumulate in the cytoplasm of primordial germ cells between the comma to 2-fold stage of embryogenesis, but again appears distributed throughout the germ plasm instead of concentrated on P granules ( Fig 4D).
In an RNAi screen of oocyte-enriched transcripts, it was previously discovered that depletion of elli-1 caused partially penetrant embryonic lethality, suggesting a role in early development [42]. An early embryonic arrest (30-100 cell stage) trend was observed in 5-8% of elli-1 worms, confirming this role ( Fig 4E). Embryonic lethality and P-granule phenotypes are not observed above background in the elli-1::GFP strain, suggesting that the GFP tag is not compromising ELLI-1 function.

ELLI-1 modulates transcripts encoding core P-granule and RNAi components
To gain insight into the function of ELLI-1, tiling arrays were used to compare the wholeworm RNA expression profile of elli-1(sam3) to wild-type animals (four biological replicates each). This analysis showed significant changes (over 1.2-fold, q value < 0.05) in expression for 1079 genes, 43% of which are upregulated while 57% are downregulated (Fig 5A red, S1 Table). No significant overlap was observed when comparing these 1079 genes to previously published datasets of csr-1 regulated genes [13], pgl-1 regulated genes [13], CSR-1 target genes [4], or germline enriched genes [43] (Fig 5B). There was a statistically significant overlap of genes that are soma-enriched [43], but they were evenly distributed between genes that were up and downregulated in the elli-1 mutant (Fig 5A green, Fig 5B). Expression of transcripts encoding CSR-1 and its cofactors EGO-1, EKL-1, and DRH-3 showed modest but significant increases in elli-1 (Fig 5C). Consistent with the enlarged P-granule phenotype of elli-1, transcripts encoding the constitutive P-granule components PGL-1, GLH-1, GLH-2, and GLH-4 increased in the mutant (2.37, 1.54, 1.57, and 1.92 fold respectively); although, this increase was not seen with all P-granule transcripts as those encoding PGL-3 and GLH-3 changed very little (Fig 5C). Quantitative RT-PCR on pgl-1, pgl-3, and glh-1 in wild-type and an outcrossed elli-1(sam3) line validated the microarray results (Fig 5C, green). As P granules regulate translational silencing in the germline, increased expression of P-granule components suggest that one function of ELLI-1 may be to keep P-granule accumulation and translational silencing in check; however, an alternative interpretation is also discussed below.

ELLI-1 enhances sterility and RNAi-defective phenotypes when the CSR-1 complex is partially compromised
Similarities between the P-granule phenotypes of csr-1, ego-1, ekl-1, drh-3 and elli-1 suggests ELLI-1 functions with or in parallel to CSR-1; however, unlike elli-1, null mutations in CSR-1 complex components are RNAi defective and have no viable progeny [4]. This difference, and the lack of correlation between differentially expressed genes in elli-1 and csr-1 worms, may suggest the P-granule phenotypes are coincidental; however, given the role of P granules in translational regulation, mutations in CSR-1 complex components may still interact genetically with elli-1. To see if there is a genetic interaction, the homozygous viable drh-3(sam27) allele was utilized. Hypomorphic alleles of drh-3 previously isolated from an RNAi-deficiency screen have viable progeny at permissive temperature, but are sterile at elevated temperatures [10]. First, broods were counted from individual drh-3(sam27) animals grown at 20˚C and 25˚C, showing this new allele is also temperature sensitive sterile (Fig 6B). To see if elli-1(sam3) enhanced sterility of drh-3(sam27), brood sizes from individual worms were then counted in the drh-3(sam27); elli-1(sam3) double mutant. At the permissive temperature of 20˚C, drh-3 (sam27) and elli-1(sam3) single mutants are fertile, but the majority of double mutants are not. This double mutant line must be maintained at 15˚C, as the small percentage of fertile animals could not be passaged beyond three generations at 20˚C. RNAi sensitivity was also tested in the double mutant (Fig 6A). drh-3(sam27) animals suppress larval arrest when fed his-44 RNAi at 25˚C, but only partially suppress larval arrest at 20˚C. In contrast, the drh-3(sam27); elli-1 (sam3) double mutant suppresses his-44 RNAi larval arrest at both permissive and restrictive temperatures. Thus, elli-1(sam3) enhances both sterility and RNAi defective phenotypes of the hypomorphic drh-3 allele, suggesting that ELLI-1 functions with or in parallel to components of the CSR-1 complex.
CSR-1 pathway components share both Ego (Enhancer of Glp-One sterility) and Ekl (Enhancer of Ksr-1 Lethality) phenotypes [5,11,22,25,50]. elli-1 RNAi failed to enhance glp-1 (bn18ts) sterility at the permissive temperature of 20˚, and there was no difference in rod-like larval lethality with elli-1 RNAi in ksr-1 mutants (S2 Table). Lethality in ksr-1 mutants is attributed to a defect in excretory duct cell specification, a maternally contributed but somatic cell fate. While it is easy to over interpret negative results showing the lack of glp-1 and ksr-1 enhancement, these observations suggest that ELLI-1's function with the CSR-1 complex is limited to a subset of CSR-1's targets.
Another characteristic of csr-1 is an RNA pooling defect in the germline of adults. This phenotype was proposed as a consequence of losing csr-1 mRNA slicing activity, causing transcripts to accumulate that would normally be degraded [12]. Later, RNA expression in germlines dissected from csr-1 RNAi-depleted animals failed to support this hypothesis; in contrast, mRNA targets of CSR-1-22G siRNAs in adult hermaphrodites were the least likely transcripts to change after a six-fold depletion of csr-1 [13], correlating with previous findings from tiling array expression in csr-1 mutants [4]. While some fine tuning of a fraction of CSR-1's >4000 target mRNAs has now been reported in whole worms [14], the bulk of these targets show no change in dissected germlines when csr-1 is compromised [13]. To see if elli-1 mutants also have this same RNA pooling defect, ten dissected germlines from wild-type, csr-1 (fj54), and elli-1(sam3) animals were stained with the RNA dye SYTO14. While only 1/10 wild-  -1(sam3) enhances sterility (no brood) of drh-3(sam27) at permissive temperature. C) STYO14 staining of dissected germlines shows RNA pooling around the nuclear periphery and in the rachis of both csr-1 and elli-1 mutants. D) Model for CSR-1 and ELLI-1 function. P granules (yellow) reside on the nuclear periphery where they receive transcripts coming through the nuclear pore complex. The CSR-1 complex functions in P granules to recognize germline abundant or licensed transcripts. Our model is that CSR-1 and ELLI-1 function to move type germlines showed RNA pooling, 10/10 csr-1 (p = 0.0001) and 8/10 elli-1 (p = 0.0054) germlines showed pooling in the central rachis and around the nuclear periphery of germ cells, although pooling in csr-1 is more pronounced (Fig 6C). Since ELLI-1 does not contain mRNA slicing domains like CSR-1, the pooling defect in the germline of elli-1 mutants, and potentially csr-1 mutants, may instead reflect defects in mRNA shuttling and dispersal rather than changes in mRNA levels.

Discussion
Based on EMS mutagenesis frequencies [51], we expected to get up to two loss-of-function mutations in each known component of the CSR-1 complex. This expectation came very close to our observation. A more saturated screen is likely to identify one or two more genes, but the ten alleles obtained in these five loci demonstrate both the specificity of the enlarged P-granule phenotype and the central role of the CSR-1 complex in the C. elegans germline.
We found that elli-1 enhances RNAi efficacy and sterility in a hypomorphic drh-3 allele, suggesting that ELLI-1 functions as part of or in parallel to the CSR-1 complex. The lack of overlap between elli-1 and csr-1 gene expression profiles better supports the latter possibility, and/or suggests that ELLI-1 and CSR-1 impinge on common targets post-transcriptionally to promote brood size, fertility, and RNAi efficacy. By residing on the nuclear periphery and extending the nuclear pore complex environment into the cytoplasm [52][53][54][55], P granules are well positioned to receive transcripts as they exit the nucleus, most of which then transit through P granules to reach the cytoplasm [56]. ELLI-1 is diffuse in the germ plasm (not enriched or excluded from P granules), but some ELLI-1 foci appear to dock next to P granules. One possibility is that CSR-1 shuttles its targets through P granules and into the cytoplasm to be translated (Fig 6D). ELLI-1 may be assisting in this process, although ELLI-1 does not contain known RNA-binding domains so a potential interaction with RNA may be indirect or through its predicted disordered region. The RNA pooling defects observed in elli-1 and csr-1 mutants could be a consequence of mRNA getting stuck in P granules and detached cytoplasmic RNPs. Cytoplasmic RNA distribution has recently been shown to drive phase separation of P-granule components [57][58][59], and it is worth further testing whether ELLI-1 and the CSR-1 complex function to disperse germline transcripts to antagonize phase separation, keeping P-granule size in check.
There are notable differences between elli-1 and the four other genes obtained in this screen despite the indistinguishable P-granule phenotype that suggest ELLI-1 impinges on just a subset of CSR-1 targets. First, elli-1 does not appear to enhance glp-1 or ksr-1 lethality. Second, elli-1 mutants can be maintained as homozygotes with only minor impacts on brood size, while null alleles of csr-1, drh-3, ego-1 and ekl-1 must be maintained over a balancer. We expect that truncations in the two elli-1 alleles obtained from the screen, as well as the two generated using CRISPR/Cas9, cause a complete loss-of-function as elli-1 RNAi phenocopies the enlarged P-granule defect and low frequency of embryonic lethality. The hypomorphic drh-3 allele obtained in this screen can also be maintained as a homozygote at permissive temperatures while exhibiting enlarged P granules. These results suggest that P-granule defects in CSR-1 pathway mutants can be disassociated from what is causing their embryonic lethality. Interestingly, the enzymatic slicing activity of CSR-1 was recently shown to fine tune expression in oocytes to support early embryogenesis; however, mutations introduced into CSR-1 to inhibit its slicing activity still cause embryonic lethality but were reported not to disrupt P germline abundant or licensed transcripts through and away from P granules and into the cytoplasm. In mutants this dispersal of RNA is blocked, RNA and P granules accumulate, and fertility and RNAi efficacy is decreased. It remains unclear whether ELLI-1 directly interacts with RNA.
A final difference is that elli-1 alone does not have RNAi defects like CSR-1 pathway mutants. This is important because while CSR-1 binds 22G RNAs, most evidence suggests that these siRNAs are not the product of exogenous RNAi; therefore, it has been proposed that defective transgene silencing and the Rde phenotypes of csr-1 and ego-1 are attributed to compromised P granules and the Rde phenotype of pgl-1 [4]. Our results make this less likely as the P-granule phenotypes of elli-1 and csr-1 are indistinguishable in the adult germline, suggesting that components of the CSR-1 complex play a more direct role in exogenous RNAi. Because germline defects resulting from partial loss of the CSR-1 pathway are enhanced by elli-1, we conclude that ELLI-1 functions with the CSR-1 pathway to modulate RNAi activity, P-granule accumulation, and post-transcriptional expression in the germline.

Strain maintenance
C. elegans strains were maintained as per standard protocols [60]. See S3 Table for list of strains used. N2, ZT3, TH206, MT8677 and the CB4856 Hawaiian isolate were obtained from the Caenorhabditis Genetics Center (CGC). EL44 was a gift from Eleanor Maine at Syracuse University. Remaining strains were generated in this study and are available upon request.

Screen design
EMS mutagenesis was performed on TH206 worms using the standard protocol [61]. Twothousand F1 progeny were cloned to individual plates, and F2 grandchildren were screened under a Leica M165FC fluorescence stereomicroscope for enlarged, bright PGL-1::GFP granules. This screen was done in parallel with a now published screen looking for somatic PGL-1:: GFP expression [35]. To quantify P-granule size, cross-sections of the surface of ten germlines/ strain were acquired with a 63X objective and fixed exposure conditions, and a fixed threshold was used to outline and measure the area of P granules using ImageJ.
Mapping CB4856 (Hawaiian) males were crossed into mutant strains. F1 cross progeny were picked to new plates, and on average 500 F2s worms with the enlarged P-granule phenotype were handpicked from each cross and pooled for whole genome sequencing [35,62]. Samples were sequenced on Illumina HiSeq2500 and NextSeq systems. The CloudMap pipeline was used to analyze mutant genome sequences, obtain map data, and find mutations as previously described [36].

RNAi feeding
RNAi feeding constructs were obtained from the Ahringer library. The L4440 plasmid in HT115 bacteria was used as the RNAi control. elli-1 RNAi was performed on L4 worms in three biological replicates, and the PGL-1::GFP phenotype was observed in F1 progeny.
To assay RNAi enhancement, approximately 60 L1s were placed on each of five his-44 RNAi feeding plates for each strain, and animals arrested during larval development were scored 3 days later [35,64].

Fluorescent in situ hybridization (FISH)
A Stellaris FISH probe (Biosearch Technologies) was designed for elli-1 with CAL Fluor Red 610. Whole-worm FISH was performed as previously described [13,66]. Images were acquired and deconvolved using Leica AF6000 acquisition software on an inverted microscope (Leica DMI6000B) with a cooled CCD camera (Leica DFC365FX) and a Leica 40x air objection with a 0.5 um z-stack. All white scale bars throughout the paper are 20 microns in length.

Immunocytochemistry
Dissected germlines were fixed using methanol/acetone [67], and stained with rabbit anti-PGL-1 or rabbit anti-GLH-1 polyclonal antibodies, and co-stained with monoclonal mAb414 as previously described [68]. To examine elli-1::gfp::3xFLAG expression and colocalization with PGL-1, dissected germlines and embryos were fixed with methanol acetone and costained with rabbit anti-PGL-1 [68] or rabbit anti-CGH-1 [69] and a 1:1000 dilution of M2 anti-FLAG monoclonal antibody (Sigma-Aldrich-F1804), followed by staining with anti-rabbit Alexa 594 and goat anti-mouse Alexa 488 conjugated secondaries and DAPI. To quantify PGL-1 expression in Fig 1C, dissected germlines were fixed and stained with rabbit anti-PGL-1, and a pixel intensity was measured in a fixed region on interest in pachytene germ cells; then normalized to the parental control (n>18 germline images per strain).

Brood size assay
For each strain, brood sizes were counted from each of ten L4 animals after placing them at 20˚C and 25˚C.

Embryonic lethality
For each strain, six adult worms were placed on a plate and allowed to lay for 5 hours. The number of embryos on each plate was counted, and unhatched embryos were counted the following day.

Microarray
For each of the four biological replicates of TH206 and DUP56 strains, synchronized L1 larvae were seeded on 10 plates, grown at 20˚C, and young adults were collected (corresponding to 90% with a vulva and no or few eggs). Young adults were washed three times, pelleted, and frozen. Total RNA was extracted with Trizol Reagent and ran through an RNA Stable kit (Biomatrica), which was submitted for expression analysis. OakLabs (www.oak-labs.com) prepared samples and hybridized them on the 8x60K Agilent ArrayXS C. elegans tiling array, and performed gene expression analysis (GEO accession number GSE82322). These microarrays also contain several thousands of piRNA probes, but as there was no significant change in piRNA levels, these probes were excluded from S1 Table. qRT-PCR validation of pgl-1, pgl-3, and glh-1 expression in N2 and DUP67 worms as described [12].
Supporting information S1