DEPS-1 promotes P-granule assembly and RNA interference in C. elegans germ cells

Germ granules are large non-membrane bound organelles composed of RNAs and proteins important for germ cell specification. They are found in the cytoplasm of developing germ cells in many organisms, and are commonly perinuclear (Eddy, 1975). Germ granules in different organisms are now known to share a number of components,including proteins related to the Drosophila DEAD-box helicase VASA,but many organism-specific germ granule components have also been described that appear to be important for germ granule formation or function (reviewed by Seydoux and Braun, 2006; Strome and Lehmann, 2007). The molecular functions of germ granules are likely to be complex but, based on their compositions and subcellular localization, it has been argued that germ granules may post-transcriptionally regulate mRNAs, sort newly transcribed mRNAs as they leave germ cell nuclei, and/or facilitate the localization of mRNAs and proteins to primordial germ cells of embryos with maternally inherited germ plasm (Seydoux and Braun,2006). Furthermore, recent findings in Drosophila and mice suggest that germ granules might play a role in the biogenesis of small RNAs (Kotaja et al., 2006; Lim and Kai, 2007).

In the nematode Caenorhabditis elegans, germ granules are known as P granules. P granules are germ cell-specific at all stages of development(Strome and Wood, 1982) and are important for germ cell development, as removing some of the constitutive components of P granules both alters the structure of P granules in the adult germ line (Schisa et al.,2001), and causes sterility and germ cell underproliferation(Kawasaki et al., 2004; Kawasaki et al., 1998; Kuznicki et al., 2000). Specifically, the nematode-specific predicted RNA-binding protein PGL-1 and the VASA-like RNA helicase GLH-1 are both constitutively associated with P granules and required for proper germ cell proliferation in C. elegans at elevated temperatures. Immunofluorescence-based analyses of the relationships between PGL-1, GLH-1 and other P-granule proteins have suggested that pgl-1 and glh-1 define a pathway for P-granule assembly or stability (Amiri et al., 2001; Kawasaki et al.,2004; Kawasaki et al.,1998). For example, PGL-1 localizes poorly to P granules when GLH-1 function is compromised (Kawasaki et al., 1998), while GLH-1 appears to localize normally to P granules when PGL-1 is absent (Kawasaki et al.,2004).

We report here the identification of mutations in a new gene, called deps-1 (defective P granules and sterile),that is important for the assembly of PGL-1 onto P granules. Like PGL-1 and GLH-1, DEPS-1 protein associates with P granules and is required for proper germ cell proliferation at elevated temperatures. Consistent with these phenotypic similarities, deps-1 mutant germ lines display reduced levels of glh-1 mRNA and protein, and a diffuse distribution of PGL-1. However, DEPS-1 also appears to have novel functions that are not mediated by reduced GLH-1 or diffuse PGL-1. Specifically, DEPS-1 promotes the expression of rde-4 (for RNAi DEfective) mRNA and protein,and efficient RNA interference in the C. elegans germ line. In addition, DEPS-1 represses the expression of a subset of genes the expression of which is also repressed by RDE-3. We propose that DEPS-1 plays a role in some of the RNA regulatory processes mediated by P granules in the C. elegans germ line, and that those processes may include the generation of small RNAs that repress the accumulation of endogenously expressed transcripts.

Nematodes were cultured as described by Brenner(Brenner, 1974). Alleles and transgenes were: deps-1(bn113, bn121, bn124 or bn128) I, rde-3(ne298) I, bnIs1[pie-1::gfp::pgl-1 unc-119(+)] I(Cheeks et al., 2004), glh-1(ok439) I, unc-11(e47) I, rde-4(ne301) III, unc-119(ed3) III, pgl-1(bn101) IV and pgl-3(bn104)V. Unmarked deps-1 alleles were balanced by hT2[qIs48]I;III. Transformation rescue experiments used a deps-1(bn121)unc-11(e47) I; sDp2 (I,f) strain.

bnIs1[gfp::pgl-1] L4 stage hermaphrodites were mutagenized with 50μM EMS, allowed to self-fertilize and their F1 progeny picked to individual plates. F2 hermaphrodites were examined for diffuse GFP::PGL-1 in young,unlaid F3 embryos on a dissecting microscope. GFP::PGL-1 was examined in the germ lines and oocytes of candidate mutants and sterile hermaphrodites on a compound microscope at higher power. Worms were raised at 24-25°C to promote GFP::PGL-1 expression from bnIs1. Approximately 8700 haploid genomes were screened, and five recessive mutations causing diffuse GFP::PGL-1 were recovered.

DEPS-1 coding sequences from cDNA yk605c11 were PCR-amplified and cloned into modified pET or pGEX vectors (J. Suh, personal communication) using Gateway cloning technology (Invitrogen) to generate four fusion proteins:6XHIS::DEPS-1 (amino acids 1-619), GST::DEPS-1 (amino acids 1-619),GST::DEPS-1N (amino acids 1-316) and GST::DEPS-1C (amino acids 313-619). Fusion proteins were purified using Ni-NTA agarose (Qiagen) or as inclusion bodies. Antibodies were raised against the 6XHIS fusion protein in rabbits(Cocalico), immuno-affinity purified against a GST fusion protein coupled to CNBr-activated sepharose 4B (Amersham) and eluted in 0.2 M glycine and 0.15 M NaCl. Antibodies purified against GST::DEPS-1C had the least cross-reactivity with a ∼120 kDa nuclear antigen and are shown. All antibody preparations recognized DEPS-1 in wild-type animals but not in deps-1 mutants.

A 3.3 kb region from cosmid W03C10 was subcloned into pBluescript SK+ to make pCS306. Site-directed mutagenesis of pCS306 replaced the deps-1stop codon with an AgeI site, and a 0.9 kb AgeI fragment from pID3.01 (Pellettieri et al.,2003) was inserted to make pCS320 (deps-1::gfp). deps-1(coding)::gfp sequences were PCR amplified and inserted into pID2.02 (D'Agostino et al.,2006) using Gateway cloning technology (Invitrogen) to make pCS336(pie-1::deps-1::gfp). deps-1-specific primer sequences were:ATGTCAGAACGCCAATCCAA (5′) and TTTGGTTGGATAACGGGTAG (3′). pCS336 was introduced into worms by biolistic transformation of unc-119(Praitis et al., 2001), and an integrated line expressing DEPS-1::GFP was identified.

Embryos and germ lines were fixed using methanol/acetone(Strome and Wood, 1983). Antibody dilutions were 1:500-1:1000 anti-DEPS-1, 1:4000 anti-PGL-3(Kawasaki et al., 2004),1:10,000 anti-GLH-1 (Kawasaki et al.,2004), 1:1000 PA3 [a gift from M. Monestier(Monestier et al., 1994)] and 1:500 Alexa Fluor 488 goat anti-rabbit IgG, Alexa Fluor 594 goat anti-mouse IgG and Alexa Fluor 594 goat anti-rat IgG (Molecular Probes). Images were acquired with a Nikon Eclipse TE200 microscope and UltraVIEW LCI spinning-disk confocal laser using UltraVIEW software (Perkin Elmer).

Experimental conditional were as follows: (1) for DEPS-1 western, 30 deps-1 M-Z- or wild-type gravid adults (20°C) were loaded per lane; (2) for GLH-1 western, 30 deps-1 M+Z-, M-Z- or wild-type gravid adults (15°, 24.5°C) were loaded per lane; (3) for RDE-4 western, 50μg acetone-precipitated protein isolated from 1-2 ml deps-1 M-Z-or wild-type worms enriched for gravid adults (20°C) was loaded per lane. Protein was isolated by dounce homogenization of worms in 50 mM Tris pH 7.5,10 mM KOAc, 100 mM KCl, 1 mM EDTA, 1 mM DTT, 1% Triton X-100, 2.5×protease inhibitors (Roche) followed by centrifugation at 41 gin a Beckman JA20 rotor at 4°C to remove insoluble material. Primary antibodies were 1:2000 anti-DEPS-1, 1:5000 anti-GLH-1, 1:1000 anti-RDE-4(Tabara et al., 2002) and anti-α-tubulin (DM 1a, Sigma) at 1:250 for GLH-1 western, 1:1000 for DEPS-1 western or 1:10,000 for RDE-4 western. Secondary antibodies were 1:5000-1:10,000 horse radish peroxidase (HRP)-conjugated goat anti-rabbit IgG,1:5000 HRP-conjugated goat anti-mouse IgG, alkaline phosphatase(AP)-conjugated goat anti-rabbit IgG and AP-conjugated goat anti-mouse IgG(Jackson Labs). Antibody signals were detected using SuperSignal West Pico Chemiluminescent Substrate (Pierce) and autoradiography film (DEPS-1, RDE-4 westerns) or the Enhanced Chemifluorescence detection system and Typhoon 9200(Amersham) (GLH-1 western).

A deps-1(RNAi) clone was generated by subcloning a 1.4 kb genomic BamHI fragment from the deps-1 gene into L4440(Timmons and Fire, 1998) to generate pCS302. Other RNAi clones (Kamath et al., 2003) were purchased (Geneservice). RNAi was essentially as described (Kamath et al.,2001) but used plates containing 0.2% lactose to induce dsRNA expression (E. Lambie, personal communication).

RNA was isolated from 50 worms or 50-100 dissected gonads (50 for PCR, 100 for microarrays) as described (Chi and Reinke, 2006). Worms were subjected to three rapid freeze-thaw cycles in Trizol (Invitrogen) prior to RNA isolation. RT-PCR was performed in triplicate either as described (Bender et al., 2006) or using an Mx3000p QPCR system (Stratagene) and iQ SYBR Green Supermix (BioRad) containing 30 nM reference dye (Stratagene). All data were normalized to ama-1 and the Pfaffl method(Pfaffl, 2001) used to calculate relative fold changes. Gonad dissection and microarray analysis of linearly amplified polyadenylated RNA was as described by Bender et al.(Bender et al., 2006). Confidence levels were determined using a z test performed in Excel followed by a correction for multiple testing in which the P-value was multiplied by 17,539 (the number of genes on the microarrays). After this correction, all genes with a fold change of >1.8-fold, P<0.05 were selected for analysis. The GEO accession number for microarray data is GSE 9993.

Previous studies suggested that mutations in pgl-1 and glh-1 define a genetic pathway for normal P-granule formation in C. elegans (Kawasaki et al.,2004; Kawasaki et al.,1998). For example, the P-granule-associated protein PGL-1 is required for proper P-granule morphology(Schisa et al., 2001) and the localization of PGL-1 to P granules is disrupted in glh-1(lf) mutants(Kawasaki et al., 1998)(C.A.S. et al., unpublished). To identify other participants in P-granule formation, we screened for mutations that phenocopy the PGL-1 localization defect of glh-1(lf) mutants. We identified five new mutations that partially disrupt the localization of GFP-tagged PGL-1 (GFP::PGL-1) to P granules (Fig. 1 and data not shown). These mutations include bn125, a new allele of glh-1that will be described elsewhere, and four independent mutations in the deps-1 (defective P granules and sterile)gene (bn113, bn121, bn124 and bn128). Like glh-1(lf) mutations, deps-1 mutations disrupt the localization of endogenous PGL-1 and the related protein PGL-3 to P granules(data not shown). This suggests that deps-1 is a new member of the P-granule formation pathway, and that it functions upstream of the PGL family of proteins.

deps-1 was mapped to the left arm of LGI using snip-SNP bulk segregant analysis (Wicks et al.,2001) and other standard genetic techniques (data not shown). The PGL-1 localization defect of deps-1(bn121) animals was rescued by germline transformation with genomic sequences containing the gene Y65B4BL.2,which by in situ hybridization appears to be strongly expressed in germ cells at all stages of development (Y. Kohara, personal communication). Furthermore,RNA interference (RNAi) of the Y65B4BL.2 gene disrupted the localization of GFP::PGL-1 to P granules and caused sterility at 24.5°C, two distinctive phenotypes that are characteristic of deps-1 mutants (Figs 1, 3; data not shown). Finally,sequence analysis of Y65B4BL.2 identified allele-specific lesions in all four deps-1 mutants (Fig. 2A), indicating that Y65B4BL.2 is deps-1.

deps-1 encodes a novel protein with a serine-rich C-terminal domain of low amino acid complexity (Fig. 2A). DEPS-1 is 45-51% identical to predicted proteins in the closely related nematodes C. briggsae and C. remanei, but does not resemble identified proteins in other organisms. Affinity-purified polyclonal antibodies raised against DEPS-1 stain both P granules and nuclei in germ lines, and developing embryos of all stages(Fig. 2C-F, 2I-N and data not shown). Several findings suggest that the P-granule stain is due to DEPS-1, but that the nuclear stain is due to cross-reactive material. First, in deps-1mutants, the P-granule stain and a DEPS-1-sized band (∼69 kDa) on western blots are absent, while nuclear stain and a band of ∼120 kDa persist(Fig. 2B,E). Second, a DEPS-1::GFP fusion protein expressed in the C. elegans germ line is cytoplasmic and concentrated on P granules in adult germ lines and late stage embryos; nuclear GFP is not observed (Fig. 2G,H). These imaging results reveal that DEPS-1 is a new constitutive component of P granules.

Most of the deps-1 mutations identified in our screen are predicted to generate truncated DEPS-1 proteins(Fig. 2A). However, truncated proteins were not detected on western blots of any of the deps-1mutants (Fig. 2B and data not shown), suggesting that all four deps-1 mutations are strong loss-of-function or null. Consistent with this observation, all four deps-1 mutants and deps-1/Df animals display similar defects at high temperature (24.5°C, data not shown). We consider deps-1(bn124 and bn121), the two alleles with the earliest premature stop codons (Fig. 2A), to represent likely deps-1(null) alleles and describe their phenotype in the following sections.

Lack of the two constitutive P-granule components PGL-1 and GLH-1 causes a germ cell proliferation defect that is sensitive to temperature and maternal genotype (Kawasaki et al.,2004; Kawasaki et al.,1998) (C.A.S. et al., unpublished). We examined the fertility of deps-1 mutants at multiple temperatures (15, 20 and 24.5°C) and found that lack of DEPS-1 causes similar defects.

deps-1 mutant embryos from heterozygous (deps-1/+)mothers inherit maternally provided DEPS-1 but are unable to synthesize zygotic DEPS-1. These animals, referred to as deps-1 M+Z- mutants,are typically >90% fertile at 15-24.5°C. The self-progeny of deps-1 M+Z- mutant hermaphrodites lack maternal DEPS-1 and are unable to synthesize DEPS-1. These deps-1 M-Z- mutants tend to be sterile at high temperature (24.5°C) and fertile at lower temperatures (15 and 20°C) (Fig. 3A and data not shown). For example, 93% of deps-1(bn121) M-Z- mutants raised at 24.5°C are sterile, while only 22-31% of deps-1(bn121) M-Z-mutants raised at 15-20°C are sterile(Fig. 3A). These results indicate that deps-1 sterility is both temperature-sensitive and maternal effect, similar to null mutations in pgl-1, which cause 75-85% sterility at 25°C and 7-19% sterility at lower temperatures in the M-Z- generation (Kawasaki et al.,1998).

Most deps-1 M-Z- sterile hermaphrodites raised at 24.5-26°C fail to make embryos or oocytes and have an empty uterus(Fig. 3A and data not shown). We examined the germ lines of hermaphrodites raised at 24.5°C and found that they frequently lack gametes and have reduced numbers of germ cells compared with wild type (Table 1). For example, roughly half (56%, n=48) of the deps-1(bn121) germline arms examined had fewer than 200 germ cell nuclei and 63% lacked both sperm and oocytes. On average, deps-1(bn121) adult hermaphrodites raised at 24.5°C have 254 germ cells per gonad arm (s.d.=236, n=16, range 10-762), while wild-type adults have an average of 586 germ cells (s.d.=45, n=6, range 526-651). Furthermore, decreased numbers of germ cells are often observed in deps-1(bn121) M-Z- adult males and L2-L4 stage hermaphrodite larvae raised at 24.5°C when compared with wild-type animals (data not shown). Taken together, these observations suggest that deps-1 M-Z- mutant larvae raised at 24.5°C have partially penetrant defects in germ cell proliferation similar to those described for pgl-1 M-Z- mutants at 26°C (Kawasaki et al.,2004; Kawasaki et al.,1998).

Fertile M-Z- deps-1 mutants raised at 15-20°C often have very few progeny, and sterile hermaphrodites at 15°C typically lay eggs that fail to hatch (Fig. 3A and data not shown). To investigate whether DEPS-1 promotes embryonic viability, we counted the number of eggs laid by deps-1 M+Z- and M-Z- mutants, and determined whether they were able to complete embryonic development and hatch into larvae (Fig. 3B). deps-1 mutations cause variably penetrant embryonic lethality with no uniform stage of developmental arrest (data not shown). This lethality appears to be cold sensitive and influenced by maternal genotype: it is most penetrant in eggs laid by deps-1 M-Z- mutants at 15°C(Fig. 3B and data not shown). In addition, the total number of eggs laid by deps-1 mutants is also influenced by maternal genotype (Fig. 3B and data not shown). Notably, the amount of deps-1embryonic lethality observed in our experiments (44-74% in the M-Z- generation at 20°C) is considerably higher than that seen in pgl-1 M-Z-animals at the same temperature (Kawasaki et al., 2004), suggesting that DEPS-1 has functions in the C. elegans germ line and embryo in addition to promoting the localization of PGL-1 to P granules.

Loss-of-function mutations in glh-1, which encodes a P-granule-associated VASA-like RNA helicase(Gruidl et al., 1996), disrupt the localization of PGL-1 to P granules and cause germ cell proliferation defects similar to those described for pgl-1 and deps-1(Kawasaki et al., 1998)(C.A.S. et al., unpublished). To determine whether GLH-1 localizes to P granules in deps-1 mutants, we stained fertile deps-1hermaphrodites (M+Z- raised at 24.5°C) with an antibody that specifically recognizes GLH-1. GLH-1 localizes to P granules in deps-1 mutants but protein levels appear to be significantly reduced(Fig. 4A-D). We used western blot analysis to quantify the relative amounts of GLH-1 present in wild-type and deps-1 gravid adult hermaphrodites, and found that GLH-1 levels are reduced ∼5- to 10-fold in M+Z- deps-1 mutants raised at 24.5°C (Fig. 4E). The reduction in GLH-1 levels in deps-1 mutants is sensitive to both temperature and maternal genotype (data not shown). In the M+Z-generation, the reduction in GLH-1 levels in deps-1 relative to wild-type was ∼2-to 3-fold or less at 15°C, compared with ∼5- to 10-fold at 24.5°C. However, in the next generation (M-Z-), even 15°C deps-1hermaphrodites displayed a ∼5- to 10-fold reduction in GLH-1.

We used quantitative RT-PCR to examine whether DEPS-1 is important for glh-1 mRNA accumulation. The relative amount of glh-1 mRNA was reduced ∼5- to 10-fold in M+Z- deps-1 gravid adult hermaphrodites raised at 24.5°C (Fig. 4E and data not shown). In contrast to GLH-1 protein levels,reduction in glh-1 mRNA levels in deps-1 mutants was not sensitive to temperature or maternal genotype: glh-1 mRNA levels were reduced ∼5- to 10-fold in deps-1 mutants at all temperatures tested and in both the M+Z- and M-Z-generations (data not shown). This reduction is specific to glh-1, as mRNA levels of the germline-expressed genes pgl-1 and him-3 do not change significantly in deps-1 mutants by quantitative RT-PCR (data not shown). These results suggest that DEPS-1 is required to produce or stabilize glh-1 mRNAs.

P granules contain mRNAs and may be involved in regulating mRNA transport as well as mRNA stability or translation(Pitt et al., 2000; Schisa et al., 2001; Seydoux and Braun, 2006). Because DEPS-1 localizes to P granules and promotes the accumulation of glh-1 mRNA, DEPS-1 is a potential link between P granules and mRNA stability. Mutations in deps-1 probably also alter the structure of P granules, as DEPS-1 promotes the localization and accumulation of PGL-1 and GLH-1, respectively, and both PGL-1 and GLH-1 appear to be important for proper P-granule morphology in the adult germ line(Schisa et al., 2001).

We performed a genome-wide analysis of mRNA levels to determine whether the P-granule defects described above generally alter mRNA levels in deps-1 mutant germ lines. Total RNA was isolated from wild-type and deps-1(bn124) dissected gonads (M-Z-generation) raised at 20°C. Polyadenylated mRNAs were linearly amplified, labeled with Cy3 and Cy5 and hybridized to microarrays containing ∼17,600 of the 20,000 predicted genes in the C. elegans genome (Chi and Reinke, 2006). These experiments identified only 13 genes,including glh-1 and deps-1, that are downregulated and 32 genes that are upregulated at least 1.8-fold (P<0.05) in deps-1 mutant germ lines compared with wild type(Table 2). Thus, despite the multiple P granule-related defects identified in deps-1 mutants,DEPS-1 regulates the expression of relatively few genes, suggesting that GLH-1, localized PGL-1, and possibly P granules themselves do not play a global role in controlling mRNA levels in germ cells.

Although relatively few genes are mis-regulated in deps-1 germ lines, we noted that many of the genes are members of gene families (Tables 2, 3). glh-1 and pgl-1 are both members of gene families and are partially redundant with other gene family members (Kawasaki et al., 2004; Kuznicki et al.,2000), suggesting that many genes related to P-granule function may have multiple genomic copies with redundant functions. Some cases of apparent mis-regulation in deps-1 mutants of multiple members of a gene family may be due to cross-hybridization (see Table 3). However, analyses of the mRNA levels of specific gene family members by quantitative RT-PCR so far confirm the microarray results [Table 4; genes C38D9.2, F15D4.5, Y55H10B.1 and T21G5.3(glh-1)]. In addition to gene families, several DEPS-1 regulated genes are clustered and either co-transcribed or possibly co-regulated(T20G5.2 and T20G5.11; F54H12.1 and F54H12.6; Y19D10A.4, Y19D10A.12 and Y19D10A.16; C01B4.6 and C01B4.7; K02B7.1 and K02B7.2; W09B7.1 and W09B7.2). Most of the clustered genes with decreased expression in deps-1 germ lines (8/13 downregulated genes) appear to have functions related to carbohydrate metabolism (C01B4.6 and C01B4.7; Y19D10A.4, Y19D10A.12 and Y19D10A.16), the tricarboxylic acid (TCA) cycle (T20G5.2; F54H12.1) or RNA interference [T20G5.11 (rde-4); F54H12.1 (RNAi depletion causes RNAi resistance according to Kim et al. (Kim et al., 2005))], suggesting that deps-1 germ lines may have defects in these processes. The idea that deps-1 germ lines have RNAi-related defects is strengthened by the observation that many of the genes upregulated in deps-1 germ lines are also upregulated in rde-3 mutant worms, which are resistant to RNAi(Table 2). Connections between deps-1 and RNAi are examined in the following sections.

As described, our microarray experiments identified rde-4 as a gene that is downregulated nearly as much as glh-1 in deps-1germ lines (Table 2). rde-4 mRNA appears to be abundant in adult germ lines and may be maternally provided to early embryos (Baugh et al., 2003; Reinke et al.,2004; Y. Kohara, personal communication). We used quantitative RT-PCR to determine that rde-4 mRNA levels are reduced 7- to 10-fold in deps-1 gravid adult hermaphrodites (M-Z-generation) relative to wild-type animals (Fig. 4F, Table 4B and data not shown). Additional experiments indicated that rde-4 mRNA levels are not significantly altered in pgl-1 or glh-1 mutants (data not shown), suggesting that this change in rde-4 mRNA levels is specific to deps-1.

To investigate whether reduced rde-4 mRNA levels in deps-1 mutants result in reduced RDE-4 protein levels, we performed western blot analysis with an affinity purified anti-RDE-4 antibody(Tabara et al., 2002) on extracts enriched for gravid adults. RDE-4 protein levels are decreased∼10-fold in deps-1 M-Z- mutants relative to wild-type adults(Fig. 4F). RDE-4 is a dsRNA-binding protein that is essential for RNAi in C. elegans(Tabara et al., 2002). At the molecular level, RDE-4 probably functions in the target tissue to recognize the long dsRNA molecules that initiate RNAi(Parker et al., 2006; Parrish and Fire, 2001; Tabara et al., 1999; Tabara et al., 2002), leading us to predict that a reduction in RDE-4 protein levels would cause measurable defects in RNAi in deps-1 germ lines.

To determine whether deps-1 mutants have defects in germline RNAi,we tested the effectiveness of RNAi against target genes known to be expressed in the maternal germ line and target genes known to be zygotically expressed. RNAi was triggered by feeding L4 stage worms and their progeny E. coli expressing gene-specific dsRNAs(Timmons et al., 2001). RNAi against the maternally provided genes pos-1, skn-1 and pie-1caused essentially all of the progeny of wild-type animals to arrest and die during embryogenesis, but did not cause a similar phenotype in deps-1mutants (Fig. 5A and data not shown). By contrast, RNAi against zygotically expressed genes required for proper locomotion (unc-52, unc-22, pat-4) or viability(lin-26) was not obviously different between wild type and deps-1 mutants (Fig. 5B and data not shown).

Robert et al. previously described that mutations in the pgl-1gene also cause a defect in germline RNAi(Robert et al., 2005). We confirmed that a pgl-1(null) mutant is resistant to pos-1(RNAi) in our assay (Fig. 5A). We also performed pos-1(RNAi) on pgl-3(null) and glh-1(lf) M-Z- mutants to determine whether the P-granule defects of deps-1 mutants (reduced GLH-1, mis-localized PGL-1 and PGL-3) might promote resistance to germline RNAi. If this were the case, we would predict that (1) pgl-3 mutants, which have no obvious P-granule or germline defects when pgl-1 is wild type(Kawasaki et al., 2004), would not be resistant to RNAi, and (2) glh-1(lf) mutations, which disrupt the localization of both PGL-1 and PGL-3 to P granules (C.A.S. et al.,unpublished), would be resistant to RNAi. We found that glh-1 and pgl-3 mutants both exhibit a strong RNAi response to pos-1dsRNA (Fig. 5A) and conclude that reduced GLH-1 levels and mis-localized PGL-1 and PGL-3 are not likely to be the cause of the deps-1 germline RNAi defect. Furthermore, these findings support the emerging view that P-granule components serve diverse roles in the germ line.

The rde-3 (formerly called mut-2) gene encodes a potential poly-A polymerase and, like rde-4, is essential for RNAi. RDE-3 probably functions downstream of RDE-4 in the RNAi pathway, and mutations in rde-3 (but not rde-4) cause numerous germline-related phenotypes, including partially penetrant embryonic lethality, temperature-sensitive sterility and transposon activation(Chen et al., 2005; Tabara et al., 1999). Lee et al. recently identified 257 genes whose expression is at least twofold upregulated in rde-3(ne298) worms(Lee et al., 2006). RDE-3 may repress the expression of some of these genes in normal worms by promoting the accumulation of gene-specific endogenous short interfering RNAs (endo-siRNAs)antisense to the coding strand of mRNAs(Ambros et al., 2003; Lee et al., 2006). Endo-siRNAs resemble the short interfering RNAs (siRNAs) generated during RNAi, and may regulate the levels of specific mRNAs in wild-type worms by RNAi-related mechanisms (Duchaine et al.,2006; Lee et al.,2006; Ruby et al.,2006). Interestingly, nearly 30% of the genes upregulated in deps-1 germ lines (9/32) are also upregulated in mixed-stage mRNA preparations made from rde-3(ne298) worms(Table 2), and many of these are strongly upregulated in rde-3 worms (5/9 increased >9-fold)(Lee et al., 2006). This degree of overlap is statistically significant(P<2.2×10^-9) and striking considering that the deps-1 and rde-3 experiments used different stages (adult germ line versus mixed stage worms) and microarray platforms (amplicon versus oligo probes). rde-3(ne298) mutant worms do not display altered levels of deps-1, glh-1 or rde-4 mRNAs(Table 4B), and deps-1mutant germ lines and worms do not have altered levels of rde-3 mRNA(0.91 deps-1/wild-type ratio by microarray analysis; 1.1 deps-1/wild-type ratio by quantitative RT-PCR on gravid adults),suggesting that RDE-3 and DEPS-1 do not regulate each other's expression. Instead, the two proteins might work together to regulate the expression of several genes in the C. elegans germ line.

Quantitative RT-PCR was used to verify that five of the genes upregulated in both deps-1 and rde-3 mutants by microarray analysis are indeed upregulated in deps-1 mutant germ lines. Multiple cDNAs that may represent endogenous siRNAs were previously isolated for four of these genes (Table 4A), making them plausible targets of endo-siRNA-mediated repression(Ambros et al., 2003; Lee et al., 2006). All five genes were strongly upregulated (∼4- to 326-fold) in deps-1 germ lines compared with wild type (Table 4A). Strikingly, one of the genes examined (C38D9.2) is upregulated several hundred-fold in deps-1 germ lines(Table 4A) and in rde-3(ne298) and deps-1 gravid adults(Table 4B). Because RDE-3 and RDE-4 may both be required to generate specific endo-siRNAs(Lee et al., 2006), and RDE-4 levels are decreased in deps-1 mutants, we also examined the expression of all five genes in rde-4 mutant germ lines; none of them was strongly upregulated (Table 4A). We conclude that the upregulation of these genes in deps-1 germ lines is not due to the defect in rde-4expression, and that RDE-3 and RDE-4 may generally repress the expression of different genes in the C. elegans germ line. These conclusions are consistent with the idea that RDE-4 is essential for RNAi initiated by long dsRNA molecules, but not required for the initiation of other RNAi-related pathways that require RDE-3 (Grishok,2005).

DEPS-1 is a P-granule-associated protein important for the localization of PGL-1 and the accumulation of glh-1 and rde-4 mRNA and protein. It is also required for fertility, embryonic viability, germline RNA interference and to repress the expression of genes also repressed by RDE-3 in the C. elegans germ line. It seems likely that many or all of these defects are the result of RNA accumulation defects in deps-1 mutant germ lines: (1) reduced levels of glh-1 mRNA and protein probably cause or contribute to the PGL-1 localization and fertility defects; (2)reduced levels of rde-4, and possibly F54H12.1 (aco-2), mRNA and protein probably cause or contribute to the germline RNAi defect; and (3)reduced levels of specific endo-siRNAs could cause the overexpression of genes repressed by RDE-3. Furthermore, localization of DEPS-1 to cytoplasmic P granules suggests that if DEPS-1 is directly involved in regulating mRNA or endo-siRNA accumulation, it does so post-transcriptionally. Although many RNAs are thought to pass through and be at least transiently concentrated in P granules in the adult germ line (Schisa et al., 2001), the loss of DEPS-1 does not dramatically affect the mRNA accumulation profile in germ lines. This suggests either that the compromised P granules present in deps-1 mutants are largely functional or that P granules do not play a major role in stabilizing mRNAs in the C. elegans germ line.

More than 20 P-granule-associated proteins have been identified, and most are predicted to interact physically with RNA or are clearly implicated in RNA-related processes (Strome,2005). These processes include translation (GLD-1, IFE-1),polyadenylation (GLD-2, GLD-3), splicing (Sm proteins), 5′ decapping and decay (CGH-1, DCAP-2), and P-granule assembly or stability (PGL-1, GLH-1)(reviewed by Strome, 2005; Seydoux and Braun, 2006). DEPS-1 does not have an obvious RNA-binding domain or motif but has a C-terminal serine-rich domain that also contains several arginines. This domain is distinct from the C-terminal Arg-Ser (RS) domains of splicing-related SR proteins, but shares at least one unusual characteristic with RS domains: both are predicted to lack defined structures(Haynes and Iakoucheva, 2006)(data not shown). Such `intrinsically unstructured' regions are frequently important for protein function (Tompa,2002). RS domains are important functional motifs that interact with other RS domains as well as RNA (Shen et al., 2004). We speculate that the serine-rich C-terminal domain of DEPS-1, also present in other Caenorhabditis DEPS-1-like proteins,is a protein or RNA interaction domain important for DEPS-1 function.

One of the most intriguing observations about deps-1, pgl-1 and glh-1 is that animals with null or strong loss-of-function mutations in these genes have temperature-sensitive defects in germ cell proliferation(Kawasaki et al., 1998)(C.A.S. et al., unpublished). Studies of pgl-1 and glh-1suggest that the conditional nature of these defects is due to functional redundancy with other members of their respective gene families. For example, pgl-1; pgl-3 M-Z- and glh-1 glh-4 M-Z-animals are sterile even at low temperatures (Kawasaki et al.,2004; Kuznicki et al.,2000) (C.A.S. et al., unpublished). Is a reduction in GLH-1 levels the cause of deps-1 sterility? We think it is not the sole cause,because deps-1 M-Z- animals display high sterility at 24.5°C,while glh-1(lf) M-Z- animals require higher temperatures(25-26°C) to display high sterility (C.A.S. et al., unpublished). DEPS-1 may therefore promote fertility and germ cell proliferation by regulating factors in addition to GLH-1. GLH-4 is an obvious candidate for such a factor;it appears to be well-expressed in deps-1 mutants but localizes poorly to P granules (data not shown). It is possible that the function of GLH-4 is mildly impaired in deps-1 mutants and this impairment enhances the sterility defect caused by reduced levels of GLH-1, resulting in highly penetrant sterility at 24.5°C but not at lower temperatures.

A second interesting aspect of the deps-1 phenotype is the fact that, at 15°C, GLH-1 protein levels are reduced only ∼2- to 3-fold in the M+Z- generation, even though glh-1 mRNA levels are decreased 5-to 10-fold. This is different from 24.5°C, where glh-1 mRNA and protein levels both are reduced 5- to 10-fold in the M+Z- generation. Several groups have noted that GLH-1 protein is more abundant at low temperatures than high temperatures (Orsborn et al.,2007; Walstrom et al.,2005) (data not shown), suggesting that GLH-1 is thermolabile. If GLH-1 is very stable at 15°C, perdurance of GLH-1 might obscure the effects of a reduction in glh-1 mRNA levels in the M+Z- generation. In that case, we would expect GLH-1 levels to decrease 5- to 10-fold in the next generation, as we have observed for deps-1 M-Z- animals at 15°C.

Extensive genetic screens have been performed looking for mutants that are resistant to germline RNAi (Tabara et al.,1999) (C. Mello, personal communication). None of these screens identified mutations in deps-1 as RNAi resistant. Two explanations seem likely. (1) deps-1 mutations cause highly penetrant embryonic lethality in the M-Z- generation at low temperatures. This phenotype makes it difficult to maintain deps-1 mutants as homozygotes, and also makes the RNAi resistance phenotype of deps-1 mutants look less dramatic when a germline gene that causes embryonic lethality is targeted. (2) deps-1 mutants may have a mild or hypomorphic defect in germline RNAi. In the course of our RNAi experiments with deps-1, we performed RNAi on the housekeeping gene ama-1, which is required maternally for embryonic viability. We found that deps-1 M+Z- mutants produced viable eggs and were clearly resistant to ama-1(RNAi), as deps-1(+) ama-1(RNAi)-treated control animals produced 100% dead eggs. We allowed the M-Z-progeny of these RNAi-treated M+Z- animals to grow up and found, to our surprise, that they produced increased proportions of dead eggs (data not shown). This observation suggests that deps-1 germ lines may initiate RNAi after prolonged exposure to a dsRNA trigger and is consistent with the idea that RDE-4 levels are decreased, but not absent, in deps-1 germ lines. Both deps-1 phenotypes would make deps-1 mutants unlikely to emerge in screens for homozygous viable mutants with strong defects in germline RNAi.

P granules are likely to be involved in multiple RNA-related processes in the C. elegans germ line (Pitt et al., 2000; Seydoux and Braun,2006), so it is reasonable to think that P-granule components like DEPS-1 and PGL-1 could be directly involved in RNAi. Indeed, several connections between RNAi and germ granules have recently become apparent. Proteins involved in RNAi-related processes localize to germ granules (nuage or polar granules) in Drosophila(Lim and Kai, 2007; Megosh et al., 2006; Pane et al., 2007) and to P bodies, RNP particles in somatic cells that may be related to germ granules,in several organisms (Eulalio et al.,2007). Dicer, the endonuclease that processes precursor RNAs into siRNAs and miRNAs, localizes to germ granules in the male germ cells of mice and interacts with MVH, a mouse VASA homolog(Kotaja et al., 2006). Interestingly, the ectopic expression of at least some components of P granules in C. elegans somatic cells in retinoblastoma (Rb) pathway mutants is correlated with an enhanced response to RNAi(Wang et al., 2005). Despite these connections, we find that P-granule components are not generally required for RNAi (Fig. 5 and data not shown; D. Conte, personal communication). We think the simplest explanation for deps-1 RNAi resistance is that DEPS-1 promotes the accumulation of RDE-4, a protein with a well-established role in RNAi(Tabara et al., 2002). The basis for pgl-1 RNAi resistance is not currently known.

The observation that genes upregulated in a second mutant required for germline RNAi (rde-3) are frequently upregulated in deps-1germ lines suggests that deps-1, and possibly other P-granule components, might be involved in a second RNAi-related process: the accumulation or function of specific endogenously expressed short interfering RNAs (endo-siRNAs). Endo-siRNAs appear to be a diverse group of small RNAs generated by multiple pathways. Their accumulation can depend on RDE-3, RDE-4 and other identified components of the RNAi machinery, suggesting that they repress gene expression by an RNAi-related mechanism(Duchaine et al., 2006; Lee et al., 2006; Ruby et al., 2006). Intriguingly, multiple components of Drosophila nuage promote the accumulation of a distinct group of short interfering RNAs, known as rasiRNAs,from repeated elements; rasiRNAs are thought to repress the expression of selfish genetic elements in the Drosophila germ line(Lim and Kai, 2007; Pane et al., 2007). DEPS-1 might function in an analogous manner and promote the accumulation of specific germline-expressed endo-siRNAs. Experiments are currently under way to determine whether endo-siRNAs associated with rde-3/deps-1upregulated genes accumulate normally in deps-1 mutants (D. Conte,personal communication).

We thank D. Conte and C. Mello for sharing unpublished results, T. Duchaine for a protocol for worm extract preparation and C. Carroll for assistance with germline dissections. This work was supported by Ruth Kirchstein NRSA postdoctoral fellowship GM69084 (C.S.), American Cancer Society postdoctoral fellowship PF-04-034-01-DDC (C.S.), and NIH grants GM34059 (S.S.) and GM65682(V.R.). Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR).