Abstract
P granules are non-membrane-bound organelles found in the germ-line cytoplasm throughout Caenorhabditis elegans development. Like their “germ granule” counterparts in other animals, P granules are thought to act as determinants of the identity and special properties of germ cells, properties that include the unique ability to give rise to all tissues of future generations of an organism. Therefore, understanding how P granules work is critical to understanding how cellular immortality and totipotency are retained, gained, and lost. Here we report on a genomewide RNAi screen in C. elegans, which identified 173 genes that affect the stability, localization, and function of P granules. Many of these genes fall into specific classes with shared P-granule phenotypes, allowing us to better understand how cellular processes such as protein degradation, translation, splicing, nuclear transport, and mRNA homeostasis converge on P-granule assembly and function. One of the more striking phenotypes is caused by the depletion of CSR-1, an Argonaute associated with an endogenous siRNA pathway that functions in the germ line. We show that CSR-1 and two other endo-siRNA pathway members, the RNA-dependent RNA polymerase EGO-1 and the helicase DRH-3, act to antagonize RNA and P-granule accumulation in the germ line. Our findings strengthen the emerging view that germ granules are involved in numerous aspects of RNA metabolism, including an endo-siRNA pathway in germ cells.
GERM granules are large, non-membrane-bound, ribonucleoprotein (RNP) organelles found in the germ-line cytoplasm of most, if not all, animals (Eddy 1975; Saffman and Lasko 1999). The term “germ granule” encompasses what are known as P granules in Caenorhabditis elegans, polar granules in Drosophila melanogaster, germinal granules in Xenopus laevis, and the perinuclear nuage in mouse and human germ cells. These large RNP complexes contain a heterogeneous mixture of RNAs and proteins. To date, most of the known germ granule proteins across species, and all of the known P-granule components in C. elegans, are associated with RNA metabolism, which suggests that a main function of germ granules is post-transcriptional regulation (Strome 2005; Seydoux and Braun 2006). Germ cells are unique in their ability to give rise to all tissues of future generations of an organism. Consequently, germ cells are considered to be both totipotent and immortal. The widespread presence of germ granules in germ cells across species and the ability of germ granule transplantation to induce functional germ cells suggest that germ granules are key determinants of the identity and special properties of germ cells (Smith 1966; Illmensee and Mahowald 1974; Ephrussi and Lehmann 1992). As more components and regulators of germ granules are identified, we will better understand their role in conferring germ cell identity and properties.
One of the earliest identified constitutive components of C. elegans P granules is PGL-1 (Kawasaki et al. 1998). Identification of genes whose loss alters the level and/or distribution of PGL-1 offers an avenue to identify other P-granule components and components that regulate P-granule assembly and stability. For example, the C. elegans VASA homolog GLH-1, another constitutive P-granule component, acts “upstream” of PGL-1; in glh-1 loss-of-function mutants, PGL-1 is not properly localized to P granules and instead a significant portion of PGL-1 is diffusely distributed in the germ-line cytoplasm (Kawasaki et al. 1998; Spike et al. 2008a). More recently another constitutive P-granule component, DEPS-1, was identified in a forward genetic screen for mutants that display PGL-1 localization defects similar to those seen in glh-1(lf) mutants (Spike et al. 2008b). Targeted studies have demonstrated that mutation or RNAi of other genes also disrupts PGL-1 in the germ line. These include genes encoding the nuclear transportins IMB-2, IMB-3, IMB-5, and IMA-3 (J. Ahringer, personal communication; Geles and Adam 2001); the maternal zinc finger protein MEX-1 (Mello et al. 1992; Guedes and Priess 1997); many of the Sm spliceosome components (Barbee et al. 2002); the eIF-5A homolog IFF-1 (Hanazawa et al. 2004); the germ-line enriched protein MEG-1 (Leacock and Reinke 2008); and the oocyte maturation factors OMA-1 and OMA-2 (Shimada et al. 2006). Of these, MEX-1, MEG-1, OMA-1, and the Sm proteins are also P-granule components. This led us to predict that additional genes that affect the composition and behavior of P granules could be identified through a genomewide RNAi screen for PGL-1 accumulation and localization defects. Unlike the forward mutagenesis that identified deps-1, an RNAi screen would not require mutants to be viable and fertile. In addition, the variable expressivity of RNAi phenotypes could be used to discover P-granule defects accompanying the incomplete loss of essential genes.
In this article we report the identification of 173 genes required for the normal assembly and localization of PGL-1. Many of these components fall into specific gene classes with shared P-granule phenotypes, allowing us to better understand how cellular processes converge on P-granule assembly and function. We looked closely at the striking phenotype, accumulation of enlarged P granules, caused by RNAi depletion of CSR-1, an Argonaute that was previously shown to target its mRNA slicing activity through secondary siRNAs (Yigit et al. 2006; Aoki et al. 2007). Loss of other predicted members of a secondary siRNA pathway, the RNA-dependent RNA polymerase EGO-1 and the DExH helicase DRH-3, causes a similar phenotype. We propose that CSR-1(+), EGO-1(+), and DRH-3(+) antagonize the growth and accumulation of P granules through an endogenous siRNA pathway that functions in the germ line.
MATERIALS AND METHODS
Strains and culture:
C. elegans strains were maintained as described by Brenner (1974). Strains used for this study include N2(Bristol) as wild type (WT), SS747 bnIs1(pie-1∷GFP∷PGL-1; unc-119(+)) (Cheeks et al. 2004), ZT3 csr-1(fj54)IV/nT1[qIs51](IV:V), FX892 csr-1(tm892)/unc-24 IV, NL2098 rrf-1(pk1417)I, NL2099 rrf-3(pk1426)II, EL500 ego-1(om84)I/hT2G(I:III), and SS889 rde-3(r459)I/hT2G(I:III).
GFP∷PGL-1 RNAi screen:
SS747 worms carrying a GFP∷PGL-1 transgene inserted into LGI were maintained at 24°. The following screen protocol was followed for 10 weeks:
Day 1. Worms were washed off of 8 recently starved plates into 100 ml of S medium with 2 g of HT115 bacteria and grown at 24° with shaking (150 rpm) (Lewis and Fleming 1995). Three liters of NGM medium + 60 μg/ml Carbenicillin + 1 mm IPTG (Kamath et al. 2001) were poured into 80 24-well plates (1 ml/well).
Day 2. Five 384-well plates from the Ahringer RNAi library (Kamath et al. 2003) were replicated onto Nunc plates containing LB agar + 50 μg/ml tetracycline + 50 μg/ml ampicillin and incubated overnight at 37°.
Day 3. The 5 spotted Nunc plates were used to inoculate LB medium + 60 μg/ml Carbenicillin cultures in 5 384-well plates (50 μl/well), which were grown overnight at 37°. glh-1 RNAi feeding bacteria were inoculated into 4 empty wells of the 384-well plates each week as a positive control.
Day 4. Worms were collected from the liquid culture (started on day 1) by centrifugation. Embryos, prepared by bleach treatment of adults, were transferred to 3 NGM plates without food and allowed to hatch overnight at 24°. RNAi bacteria from the 50-μl cultures were seeded onto the 80 24-well plates and incubated overnight at 37°.
Day 5. The 80 seeded 24-well plates were placed at room temperature. Hatched L1s were washed off the unseeded plates into 100 ml of S media + 2 g of bacteria and incubated at 24° with shaking.
Day 6. After 30–33 hr of growth, the synchronized worms (mainly L3/L4 larvae) were washed and distributed onto the 80 24-well plates using a COPAS Biosort (Union Biometrica) to deliver 12–15 worms/well. Worms were incubated on RNAi plates at 24°.
Day 7. Approximately 28 hr after sorting, worms were placed at 15° to slow growth. All observable P0 adults and F1 embryos in each well were screened for GFP∷PGL-1 phenotypes using a Zeiss SV11 fluorescence dissecting microscope fitted with a stereo and 10× compound objective. Secondary and tertiary screens were performed using a Leica MZ16F fluorescence dissecting microscope fitted with a stereo and 5× compound objective.
Immunocytochemistry:
Embryos and worms were fixed using methanol/acetone (Strome and Wood 1983). Antibody dilutions were 1:30,000 rabbit anti-PGL-1 (Kawasaki et al. 1998), 1:5000 mouse anti-NPC MAb414 (Covance) (Blobel 1985), 1:5000 rat anti-PGL-3 (Kawasaki et al. 2004), 1:1000 rabbit anti-GLH-1 (Gruidl et al. 1996; Kawasaki et al. 2004), 1:400 Alexa Fluor 488 goat anti-rabbit IgG, 1:400 Alexa Fluor 594 goat anti-mouse IgG, and 1:400 Alexa Fluor 594 goat anti-rat IgG (Molecular Probes, Eugene, OR). Images were acquired with a Volocity spinning disk confocal system (Perkin-Elmer/Improvision, Norwalk, CT) fitted on a Nikon Eclipse TE2000-E inverted microscope.
Quantitative RT–PCR:
Four biological replicates of N2 worms fed empty vector or csr-1 RNAi bacteria at 24° for 30 hr were washed and frozen prior to RNA isolation using Trizol (Invitrogen, Carlsbad, CA). cDNA was synthesized using Superscript III First Strand synthesis (Invitrogen). Quantitative PCR was performed using a 2× SYBR green mix (Roche, Indianapolis) on a Roche LightCycler 480. pgl-1, pgl-3, and glh-1 amplicons were normalized to act-2. Primers were as follows: pgl-1, f-tgttgttggagtcgcgaag; pgl-1, r-tccgcaatggctcgtctt; pgl-3, f-ctcgagcagtgcttttctca; pgl-3, r-tttcgttgttcaactcgcttt; glh-1, f-actctggttttggggaagga; glh-1, r-gtcactcgatcgatgtcctg; act-2, f-cgtcatcaaggagtcatggtc; and act-2, r-catgtcgtcccagttggtaa.
SYTO14 staining of RNA:
Gonads from N2 worms 3 days posthatching were dissected in 118 mm NaCl, 48 mm KCl, and 5 um SYTO14 (Schisa et al. 2001), incubated for 10 min, and then imaged with a confocal microscope.
RNAi conditions:
RNAi depletion of csr-1, ego-1, and drh-3 was performed by feeding bacteria from the Ahringer library (Kamath et al. 2003) seeded on agar containing 50 μg/ml carbenicillin and 1 mm IPTG. Worms at the L3/L4 stage were fed RNAi bacteria for 30 hr before imaging or fixation. Embryos or worms labeled as wild type were fed HT115 bacteria containing the L4440 empty vector as a control. To enhance gene product depletion, worms used for SYTO14 staining were fed either control or RNAi bacteria starting at the L2/L3 stage for 45 hr before dissection. On the basis of sequence data from the Ahringer lab (J. Ahringer, personal communication), we resequenced RNAi clones that had been flagged as targeting the wrong sequence. Correct gene targets are as follows: JA:B0464.7 targets atx-2, JA:Y55F3A_739.a targets cct-8, JA:Y65B4B_13.b targets imb-5, JA:ZK1058.2 targets rpn-1, JA:Y47D9A.d targets rps-3, JA:C24H12.6 targets an AT-rich area with multiple targets, JA:T05H10.3 targets F15E11.5, and JA:ZK675.1 targets F59E12.11.
Western blot analysis:
L3 larvae were fed control or csr-1(RNAi) bacteria for 30 hr. Forty adult worms boiled in SDS–PAGE sample buffer were loaded in each lane. Blots were probed with 1:50,000 rabbit anti-PGL-1 and 1:1000 mouse anti-tubulin [Sigma (St. Louis) DM 1a] for a loading control. Normalized ratios of PGL-1 were compared between six samples of control and csr-1(RNAi) lysates.
RESULTS
Multiple components are required for the proper organization and localization of PGL-1:
To identify effectors of P-granule assembly, stability, and function, we used the Ahringer RNAi feeding library (Kamath et al. 2003), which targets 16,757 genes (87% of the C. elegans genome), to screen for abnormal accumulation and/or localization of GFP-tagged PGL-1 in adult germ lines and their embryonic progeny (Figure 1A). Historically, PGL-1 antibody and GFP∷PGL-1 have been used to identify and characterize P granules, so often the terms P granules and PGL-1 granules are used interchangeably; however, the phenotypes observed in our screen rely solely on PGL-1 localization unless otherwise stated. Our visual GFP screen allowed us to observe patterns that deviated from normal GFP∷PGL-1 patterns, including differences in GFP intensity, subcellular and organismal distribution, and granule shape throughout adult germ-line and embryonic development (Table 1; Figure 1, B and C). We initially identified 234 genes whose depletion by RNAi caused GFP∷PGL-1 phenotypes. These initial positives were taken through two additional rounds of screening, which narrowed our list of positives to 173 genes whose depletion consistently caused aberrant GFP∷PGL-1 phenotypes (Table 1). GFP∷PGL-1 phenotypes observed under the fluorescence dissecting scope were placed in broad classes of high, low, or undetectable (und.) GFP∷PGL-1 accumulation (Figure 1B, Table 1). GFP∷PGL-1 dispersed in the cytoplasm instead of in perinuclear aggregates was classified as “diffuse”; this category included dispersed small granules and PGL-1 signal homogeneously distributed in the cytoplasm. Altered GFP distributions were scored as throughout the embryo or restricted to the germ-line blastomeres. Examples of wild-type GFP∷PGL-1 accumulation, diffuse GFP∷PGL-1 in glh-1(RNAi), and high GFP∷PGL-1 accumulation in csr-1(RNAi) are shown in Figure 1C.
Genomewide GFP∷PGL-1 screen. (A) GFP∷PGL-1 screen strategy. L3/L4 stage worms were placed on bacteria from the Ahringer RNAi feeding library for 30 hr at 24°. GFP∷PGL-1 phenotypes were examined in the adult (P0) germ line of living animals and their embryonic progeny (F1) under a fluorescence dissecting microscope. (B) RNAi depletion of 173 different genes caused GFP∷PGL-1 phenotypes in F1 embryos and/or phenotypes in P0 germ lines. The range of phenotypes included diffuse, high, low, and undetectable (und.) GFP∷PGL-1 accumulation. “Diffuse” GFP∷PGL-1 included both dispersed small granules and more homogeneously distributed GFP∷PGL-1. Diffuse GFP∷PGL-1 in embryos was either throughout the embryo (see glh-1) or restricted to germ-line progenitor (P) cells. (C) GFP∷PGL-1 accumulation in the P0 germ line (top) and F1 embryos (bottom) of wild-type, glh-1, and csr-1 RNAi worms showing wild-type, diffuse, and high-accumulation phenotypes, respectively. Wild-type and control embryos display small GFP∷PGL-1 granules in somatic cells from the ∼24- to ∼200-cell stage. Small somatic GFP∷PGL-1 granules are much more prevalent in glh-1 and csr-1 RNAi embryos. csr-1 RNAi embryos also often display large perinuclear GFP∷PGL-1 granules in multiple cells (also see Figure 7D). Bars, 25 μm.
GFP∷PGL-1 screen positives
We wanted to know if GFP∷PGL-1 defects reflect defects associated with endogenous PGL-1 or instead are specific to the GFP-tagged transgenic protein. The GFP∷PGL-1 transgene was integrated into the genome through particle bombardment (Cheeks et al. 2004) and as a consequence should be present at low copy number and shielded from transgene silencing effects. However, since the reporter uses the germ-line-specific pie-1 promoter (Cheeks et al. 2004), it is possible that some RNAi depletions specifically affect expression of the transgene and not endogenous PGL-1. It is also possible that the post-transcriptional processing, expression, or stability of transgenic PGL-1 is affected by the GFP tag. Therefore, we repeated RNAi against 126 screen positives in wild-type (not transgenic) worms, stained with antibody against PGL-1, and examined the P0 germ line and F1 embryos using both widefield fluorescence and spinning disk confocal microscopy (Table 1). In the majority of RNAi depletions that we tested (76 of 126), endogenous PGL-1 recapitulated the GFP∷PGL-1 phenotypes. Furthermore, the improved resolution of fixed and stained samples allowed us to examine the PGL-1 phenotypes in greater detail. However, subtle differences in PGL-1 levels were harder to see in fixed and stained embryos than by GFP imaging. Fifty of 126 RNAi depletions that caused reduced levels of GFP∷PGL-1 did not cause noticeably reduced levels of endogenous PGL-1. This set of 50 genes may specifically affect expression of the GFP transgene or may result in changes in PGL-1 protein levels or distribution that are more easily seen by GFP than by antibody staining.
As expected, the list of screen positives is significantly enriched for genes expressed in the germ line. Seventy-four of the 173 genes identified are in the set of 3144 genes with germ-line-enriched expression defined by Reinke et al. (2004), which is a 2.7-fold enrichment over the number expected by chance. One hundred twenty-seven of the 173 genes identified are in the set of 4699 genes expressed in dissected gonads as detected using SAGE (Wang et al. 2009), which is a 3.5-fold enrichment over the number expected by chance.
The majority of screen positives were grouped into large gene classes that display common phenotypes (Table 2). These classes are discussed below, followed by more in-depth analysis of an argonaute gene identified in our screen.
Most prominent PGL-1 phenotypes associated with each gene class
Cell cycle/polarity/cell division:
The largest group of screen positives can be classified by their role in the cell cycle, establishing cell polarity, and the segregation of cellular components during early cell divisions. This group includes 28 genes of the 173 total screen positives (Table 2). In wild-type animals, P granules are evenly distributed in oocytes and newly fertilized zygotes, but prior to each P-cell division the granules are segregated to the cytoplasm destined for the next P-cell daughter (Strome and Wood 1982). P granules left behind in somatic daughters are disassembled or degraded (Hird et al. 1996; Saffman and Lasko 1999; DeRenzo et al. 2003; Spike and Strome 2003). RNAi depletions that cause early cell cycle arrest or altered cell division patterns and eventually embryonic lethality may lead to a diffuse or “missegregated” P-granule pattern. In fact, P granules have been used as a marker to study cell polarity for >20 years, and some examples of P-granule missegregation have been intensively studied (e.g., Gonczy and Rose 2005; Cowan and Hyman 2007). In our secondary and tertiary screens, we more closely examined whether a defective GFP∷PGL-1 phenotype appeared to be correlated with abnormal nuclear morphology, embryo patterning defects, and/or early arrest (Table 1). PGL-1 patterning defects appeared to precede general embryo defects for only 8 of the 28 genes in this class (bir-1, cdc-25.1, cdc-42, let-99, mbk-2, par-1, par-5, and F22B5.10).
As shown in Figure 2, RNAi depletion of genes in the cell cycle/polarity/cell division class tended to show two types of phenotypes. As exemplified by par-1 and cdk-9, RNAi depletion caused PGL-1 to be present in all cells of early embryos and reduced or absent from all cells of later embryos. This pattern likely reflects both missegregation of P granules and failure to properly separate germ-line and somatic lineages, with their different abilities to retain (germ-line) vs. disassemble (somatic) P granules (Kemphues et al. 1988). As exemplified by let-99, par-5, and cdc-25.1, RNAi depletion caused PGL-1 to be present in all cells of early embryos and to persist in numerous cells of later embryos (Rose and Kemphues 1998; Morton et al. 2002). We suspect that the extra PGL-1-containing cells in later embryos are in most cases a consequence of the earlier segregation defects, but another interesting possibility is that they arise by extra divisions of the P-cell lineage.
Cell cycle/cell division/polarity class. Antibody-stained PGL-1 (grayscale, green in merge) and nuclear pores (red) and DAPI-stained DNA (blue) in fixed embryos at the 4-, ∼30-, and ∼150-cell stages of development are shown. PGL-1 segregation defects are shown at the 4-cell stage of par-1, cdk-9, cdc-25.1, par-5, and let-99 RNAi embryos. In later embryos, the absence of germ-line progenitors (par-1) or improper P-granule segregation accompanied by general embryo defects (cdk-9) caused PGL-1 to be low or undetectable. Embryos depleted of cdc-25.1, par-5, and let-99 have multiple P-granule-containing cells later in development. Bar, 10 μm.
Proteasome/ubiquitin:
The next largest group of screen positives contains 23 protein degradation components from the proteasome and ubiquitin pathways (Table 2). RNAi depletion of the majority of genes in this class resulted in a characteristic GFP∷PGL-1 phenotype: the uterus was filled with early embryos that exhibit a diffuse PGL-1 distributed throughout the embryo (Table 1, Figure 3A). Often the embryonic phenotype was accompanied by diffuse PGL-1 distribution in the P0 germ line. The PGL-1 phenotype displayed by this class is likely a result of compromised protein degradation mechanisms, which may stabilize PGL-1, and by extension P-granule proteins, in both the P0 germ line and in embryos (DeRenzo et al. 2003; Spike and Strome 2003). This phenotype is probably compounded by the early embryonic arrest associated with the depletion of all but 4 of the 23 genes. These 4 include ubl-1(RNAi), whose RNAi depletion caused PGL-1 granules to persist in multiple cells prior to embryonic arrest, and the F-box-containing genes fbxa-81, fbxb-51, and fbxb-60, whose depletion did not cause embryonic lethality under our experimental conditions. This class emphasizes the importance of protein degradation in regulating P-granule patterns in germ lines and embryos.
Proteasome/ubiquitin and ribosome classes. GFP∷PGL-1 (green) in live worms (A) and PGL-1 (green), nuclear pores (red), and DNA (blue) in fixed embryos (B) are shown. Dashed boxes indicate the location of zoomed images on the right. (A) Proteasome/ubiquitin class. Depletion of proteasome (e.g., pbs-5) and ubiquitin components often causes early embryonic arrest with diffuse PGL-1 throughout the embryo. Arrested embryos accumulate in the uterus (bracketed bar) after proteasome or ubiquitin depletion, which creates a distinctive GFP∷PGL-1 phenotype. (B) Ribosome class. Four-cell embryos are shown. Depletion of core ribosomal components such as rps-0, rpl-20, and rpl-36 results in reduced PGL-1 and smaller granules (arrowheads). Bars: (A) 50 μm and (B) 10 μm.
Ribosome:
We obtained 14 genes that encode ribosomal components in our screen (Table 2). In contrast to the groups mentioned above, most PGL-1 phenotypes in this class (11/14) preceded apparent cell/embryo death (egl-45, eif-3.f, rpl-12, rpl-20, rpl-36, rps-0, rps-3, rps-5, rps-9, rps-22, and rps-24) (Table 1). RNAi depletion of all 14 of these genes resulted in reduced PGL-1 signal and small PGL-1 aggregates (Figure 3B). While compromised translation of GFP∷PGL-1 or endogenous PGL-1 may account for this phenotype, it is also possible that the accumulation of P granules is dependent upon active translation. In comparison to somatic cell differentiation, which relies heavily on transcriptional regulation, this “ribosome” class emphasizes the crucial role of post-transcriptional regulation in the identity and development of the germ line. Indeed, recent transcript profiling from dissected germ lines revealed that a relatively large percentage of germ-line expressed mRNAs are involved in translation and ribosome structure and biogenesis (Wang et al. 2009). P-granule integrity and distribution may be sensitive to the translational status of germ cells and early embryos.
Splicing:
Our screen identified 12 genes that encode known or putative splicing factors (Table 2). The PGL-1 phenotypes of 9 of these genes appeared to precede cell/embryo death (cpf-2, mag-1, phi-6, phi-7, phi-10, phi-12, prp-8, rsp-7, and skp-1) (Table 1). Several PGL-1 phenotypes were shared by multiple members of the splicing class of genes (Figure 4). RNAi depletion of 5 genes (phi-6, phi-7, phi-10, prp-8, and skp-1) resulted in increased PGL-1 signal in the germ line of P0 adults and in F1 embryos. RNAi depletion of 9 genes (cpf-2, phi-6, phi-7, phi-10, phi-12, prp-8, rnp-4, rsp-7, and skp-1) resulted in some diffuse PGL-1 in the embryonic germ-line blastomeres and PGL-1 granules that are dissociated from the nuclear periphery. In wild-type embryos, between the 4- and 8-cell stages, P granules start to become perinuclear in the germ-line blastomere, while embryos depleted for splicing factors often contained PGL-1 granules dispersed in the cytoplasm, which persisted until the embryos arrested around the 100-cell stage. This arrest was accompanied by failure of the primordial germ cell (P4) to migrate internally.
Splicing class. Antibody-stained PGL-1 (green), nuclear pores (red), and DAPI-stained DNA (blue) in fixed embryos are shown. Dashed boxes indicate the location of zoomed images on the right. (A) Approximately 16-cell (left) and ∼150-cell (right) embryos. In wild-type embryos, PGL-1 granules are attached to the periphery of the nucleus in the primordial germ cell P4 (left embryo, ∼16 cells) and its daughters Z2 and Z3 (right embryo, ∼150 cells). P4, Z2, and Z3 nuclei are marked with asterisks. phi-12, phi-10, prp-8, phi-6, and skp-1 RNAi embryos contain P granules that are detached from the periphery of the nucleus (arrowheads). Higher levels of diffuse PGL-1 in the cytoplasm of Z2/Z3 cells of phi-12, phi-10, and prp-8 embryos can be observed in the areas circumscribed by dashed lines, and Z2/Z3 fail to migrate internally. (B) SF3b complex components phi-6, phi-11, and sap-49 RNAi embryos contain PGL-1 granules that are detached from the nucleus, and Z2/Z3 fail to migrate internally. Bar, 10 μm.
Sm components of the splicesome have been shown to localize to germ plasm in multiple species, and previous RNAi depletion of the C. elegans Sm proteins disrupted P-granule patterns in embryos (Moussa et al. 1994; Barbee et al. 2002; Chuma et al. 2003; Bilinski et al. 2004; Barbee and Evans 2006). However, previous RNAi depletion of other splicing components did not disrupt P-granule patterns in embryos (Barbee et al. 2002), suggesting that P-granule integrity and perinuclear localization in embryonic cells do not depend on pre-mRNA splicing. Our screen demonstrated that at least some pre-mRNA splicing factors are necessary for proper PGL-1 localization to granules at the nuclear periphery in embryos. Since the embryonic germ-line blastomeres are not actively engaged in transcription (Seydoux et al. 1996), the requirement for proper splicing regulation likely reflects that maternally generated mRNAs must be properly spliced for P granules to localize normally in embryos. Whether properly spliced mRNAs or the process of splicing affect P-granule patterning directly or indirectly remains to be determined.
We noted that several splicing factors obtained in our screen represent one subunit of multisubunit splicing complexes. This finding raised two possibilities: the other subunits in the complex do not participate in regulating P-granule distributions or the other subunits were missed in our single-pass RNAi screen. To distinguish between these possibilities, we tested all subunits of the SF3b complex, which is absolutely essential for pre-mRNA splicing (Will et al. 1999). This complex contains seven proteins, of which six have clear homologs in worms, encoded by the genes W03F9.10, phf-5, C46F11.4, sap-49, phi-11, and phi-6 (supporting information, Table S1). Only phi-6 was identified in our screen. Of the six worm genes, only phi-6, phi-11, sap-49, and C46F11.4 are targeted by the Ahringer RNAi library. We depleted these four genes by RNAi and found that in addition to phi-6, loss of both phi-11 and sap-49 resulted in diffuse PGL-1 in the embryonic germ-line blastomeres, PGL-1 granules that are dissociated from the nuclear periphery, and failure of P4 to migrate internally (Figure 4B, Table S1). Late embryonic arrest was used as a measure of RNAi effectiveness, and unlike phi-6, phi-11, and sap-49, depletion of C46F11.4 did not result in late embryonic arrest, diffuse PGL-1, or a P4 migration defect. Interestingly, the Ahringer RNAi clone targeting C46F11.4 contains a short sequence that is present in numerous copies throughout the genome, potentially diluting the effectiveness of this RNAi clone. Taken together, our results suggest that the SF3b complex and at least certain aspects of pre-mRNA splicing are essential for proper PGL-1 localization. These results exemplify the limitations of RNAi screens (incomplete gene coverage, ineffective RNAi, and missing phenotypes in single-pass screens) but also illustrate how RNAi screen positives can lay the foundation for more targeted approaches.
Nuclear pore/envelope/transport:
We identified 11 nuclear pore or nuclear pore/envelope associated factors in our screen (Table 2). RNAi depletion of 9 of these resulted in a distinctive phenotype that appeared to precede cell death in early embryos (imb-4, imb-5, lmn-1, npp-6, npp-7, npp-9, npp-10, ran-1, and ran-4). In contrast to the perinuclear association of PGL-1 in the more than four-cell stage wild-type embryos, most PGL-1-containing granules in embryos depleted for nuclear pore components remained detached from the nuclear envelope (Figure 5A). The finding that >75% of nuclear pores are overlaid by P granules in the adult germ line (Pitt et al. 2000) predicted that loss of nuclear pore components might compromise association of P granules with the nuclear periphery.
Nuclear pore/envelope/transport and MCM licensing classes. PGL-1 (green), nuclear pores (red), and DNA (blue) in fixed ∼30-cell embryos are shown. Dashed boxes indicate the location of zoomed images on the right. (A) Nuclear pore/envelope/transport class. Depletion of nuclear pore and nuclear pore-associated components results in the dispersal of PGL-1 granules from the nuclear periphery. (B) MCM licensing class. Depletion of MCM licensing factors also causes PGL-1 granules to be distributed throughout the cytoplasm; this phenotype is accompanied by severe embryonic defects. Bars, 10 μm.
Other labs have previously demonstrated a requirement for the nuclear transportins IMB-2, IMB-3, IMB-5, and IMA-3 to maintain P-granule integrity or perinuclear localization (J. Ahringer, personal communication; Geles and Adam 2001). There are three imb-5 RNAi clones in the Ahringer RNAi library (two correctly and one incorrectly annotated); our screen identified all three clones as causing PGL-1 granules to be detached from the nuclear periphery. However, our screen did not identify imb-2, imb-3, or ima-3. In targeted retests of those three RNAi clones, depletion of imb-2 and imb-3 resulted in detachment of PGL-1 from the nuclear periphery of embryos (Table S2), while depletion of ima-3 did not. We note that a stronger RNAi regime by Geles and Adam resulted in diffuse P granules in the adult germ line but, similar to our results, not in embryos (Geles and Adam 2001).
Our screen identified 5 of the 20 NPP proteins in C. elegans. To determine if only a subset of NPPs participates in tethering PGL-1 granules to the nuclear periphery, or if some NPPs were missed by our screen, we more closely examined PGL-1 patterns after RNAi depleting individual NPPs. The results support both scenarios. Thirteen of the 15 remaining NPPs are covered in the Ahringer RNAi library. Of these 13, RNAi depletion of 4 (npp-1, npp-3, npp-8, and npp-19) caused PGL-1 to be detached from the nuclear periphery in a high enough proportion of embryos that they should have emerged as screen positives (Table S2). Thus, our screen was successful in identifying 5 of 9 of the genes whose depletion results in high penetrance detachment of PGL-1 from the nuclear periphery. We missed 4 genes. Interestingly, RNAi depletion of 3 genes (npp-14, npp-15, and npp-16) caused late embryonic lethality, demonstrating the effectiveness of RNAi, but did not cause any apparent defects in PGL-1 patterns. These findings suggest that the localization of PGL-1 granules to the nuclear periphery depends on many but not all NPPs.
Our results contribute to the emerging view that the localization of P granules to the nuclear periphery relies heavily upon underlying nuclear pores and nuclear transport. It will be intriguing to determine if this localization depends on a direct interaction between P-granule and nuclear pore components, mRNA trafficking through nuclear pores, or both. We are currently investigating the former possibility and the hypothesis that the GLH components of P granules interact via their phenylalanine–glycine (FG) domains with the FG domains of nuclear pore proteins (see discussion).
MCM licensing:
One class of screen positives contains all five MCM licensing factors represented in the RNAi library (Table 2). MCM licensing factors ensure that DNA replication occurs only once per cell cycle, so this group could have been included in the cell cycle class of positives. We chose to put the MCM licensing factors in their own class because depletion of all five factors caused a similar phenotype: PGL-1 granules were no longer perinuclear in the embryonic P cells (Figure 5B). RNAi of these five MCM factors caused severe abnormalities, so at this time we cannot deduce whether the PGL-1 phenotype is a secondary effect of embryonic death due to compromised DNA replication or whether the MCM licensing factors serve a DNA replication-independent function that facilitates P-granule attachment to the nuclear periphery. In yeast and mammalian cells, MCMs vastly outnumber replication origins and their function is not restricted to DNA synthesis (Blow and Dutta 2005). In Xenopus, MCMs associate with RNA polymerase II holoenzyme (Yankulov et al. 1999) and were recently found to be required for RNA polymerase II-mediated transcription (Snyder et al. 2009). Therefore, the PGL-1 phenotype in MCM RNAi embryos may be independent of replication and dependent on transcription or transcriptional regulation of other specific factors. It is worth noting that RNAi depletion of subunits of DNA polymerase δ and DNA polymerase α-primase can delay early cell divisions and cause missegregation of P granules (Encalada et al. 2000). We found that RNAi depletion of another DNA polymerase α-primase subunit, PRI-1, the large subunit of replication protein A, RPA-1, or the catalytic subunit of DNA polymerase alpha, Y47D3A.29, resulted in a diffuse cytoplasmic distribution of PGL-1 in early germ-line blastomeres. It would be interesting to investigate whether any treatment that slows the timing of cell divisions alters P-granule distribution and segregation or whether the factors discussed above have more direct effects on P granules.
Other small classes of screen positives:
Another small class of screen positives includes the actin/microtubule chaperonins cct-2, cct-3, cct-7, and cct-8 (Table 2). Depletion of these CCT proteins resulted in low and diffuse levels of PGL-1 in embryos. Severe embryonic defects prevented further characterization of the PGL-1 phenotype, but we predict that PGL-1 mislocalization (data not shown) stems from compromised actin-based P-granule segregation in these mutants. Interestingly, it was recently demonstrated that RNAi depletion of other components of this chaperonin complex, cct-4 and cct-6, causes ectopic expression of PGL-1 in the intestine and hypodermis of adult worms, which may contribute to their long-lived phenotype (Curran et al. 2009).
Remaining screen positives can be placed into a handful of smaller categories. Some genes encode components of subcellular organelles including vacuoles, Golgi, and ER. RNAi depletion typically resulted in decreased PGL-1 signal (data not shown). We were unable to assign 59 screen positives to a class larger than three members, and the functions of 8 positives are not yet predictable (Table 2).
Accumulation of P granules mediated through mRNA homeostasis:
In embryos, clusters of poly(A)+ RNAs colocalize with P granules (Seydoux and Fire 1994). In adult germ lines and embryos, SL1 sequences, which are trans-spliced onto the 5′ end of most mRNAs (Zorio et al. 1994), also colocalize with P granules (Pitt et al. 2000; Schisa et al. 2001). These and other results suggest that P granules transiently retain developmentally regulated mRNAs as they exit the nucleus (Schisa et al. 2001). P granules may regulate associated mRNAs in various ways, including control of mRNA stability and/or translation, and delivery of mRNAs to the germ-line blastomeres and eventually the primordial germ cells in embryos (Strome 2005). In somatic cells, mRNA deadenylation, decapping, and degradation occur in cytoplasmic RNP aggregates called P bodies, which share many components with P granules. For example, C. elegans P granules have been shown to contain the CCR4/Not1 deadenylase component CCF-1, as well as multiple decapping coactivators including PATR-1, DCAP-1, DCAP-2, and CGH-1 (Navarro et al. 2001; Lall et al. 2005; Squirrell et al. 2006; Gallo et al. 2008). Our screen identified several mRNA degradation components as being required for regulation of PGL-1 accumulation. Interestingly, their depletion caused different PGL-1 phenotypes (Figure 6). RNAi depletion of the Not1 component LET-711 resulted in increased PGL-1 accumulation in the adult germ line (Figure 6A) and in progeny embryos, while RNAi depletion of the decapping factors CGH-1 and ATX-2 and the predicted SKI2 exosome homolog F01G4.3 resulted in reduced PGL-1 levels (Table 1; Figure 6B). Depletion of CGH-1 and ATX-2 also caused PGL-1 to persist in numerous cells of later embryos (Figure 6B). These findings extend the list of mRNA degradation components that are likely to be shared between P bodies and P granules and support the view that the accumulation of P granules is closely connected to mRNA levels.
mRNA degradation components are required for proper P-granule accumulation. Antibody-stained PGL-1 (green) and nuclear pores (red) and DAPI-stained DNA (blue) in fixed germ lines and embryos are shown. (A) In the germ lines of animals depleted of the Not1 deadenylase LET-711, PGL-1 granules are larger and more numerous and more dispersed than in wild type. (B) After depletion of decapping coactivators ATX-2 and CGH-1, low levels of PGL-1 are observed in the adult germ line, and PGL-1 is often observed in multiple cells of late embryos (arrows). Bars, 10 μm.
CSR-1 downregulates RNA and P-granule accumulation through an endo-siRNA pathway in the germ line:
The most striking PGL-1 phenotype observed in the P0 germ line was caused by RNAi of csr-1. Depletion of csr-1 resulted in very intense PGL-1 staining and large PGL-1 aggregates distributed throughout the cytoplasm of the adult germ line and embryos (Figures 1C and 7A). To determine if the large PGL-1 aggregates in csr-1(RNAi) germ lines are bona fide P granules, we tested for whether they contain other constitutive components of P granules, such as the PGL-1 paralog PGL-3 (Kawasaki et al. 2004) and the VASA homolog GLH-1 (Gruidl et al. 1996). PGL-3 and GLH-1 colocalize with the large PGL-1 aggregates in csr-1(RNAi) adult germ lines (Figure 7C). Quantitative PCR of pgl-1, pgl-3, and glh-1 transcripts in adult hermaphrodites showed a modest increase after csr-1 RNAi (1.69-fold for pgl-1, P = 0.0053; 1.12-fold for pgl-3, P = 0.0390; 2.17-fold for glh-1, P = 0.0010). While GFP∷PGL-1 reporter and antibody-stained endogenous PGL-1 appear dramatically brighter in csr-1(RNAi) worms (Figures 1C and 7A), Western blot analysis revealed a modest 1.4-fold increase in PGL-1 protein in csr-1(RNAi) worms compared to wild type (Figure S1), consistent with the ∼1.7-fold change in pgl-1 transcript noted above. The brighter, larger P granules observed in csr-1(RNAi) embryos may reflect a combination of modestly elevated levels of P-granule proteins and enhanced aggregation of those proteins into granules. We noted an additional phenotype in csr-1(RNAi) embryos. In many RNAi embryos (40%, n = 70), PGL-1 and PGL-3 granules were found in somatic blastomeres (Figure 7D). Interestingly, GLH-1 did not colocalize with the PGLs in these ectopic granules (Figure 7D). This is reminiscent of the partial P granules (containing PGL-1 and PGL-3; lacking GLH-1) observed in mutant embryos defective in autophagy (Zhang et al. 2009), raising the possibility that CSR-1 participates in autophagocytic removal of P-granule proteins from somatic cells.
PGL-1 and RNA accumulation in animals depleted of the Argonaute CSR-1. (A) Endogenous PGL-1 staining (green) in wild-type and csr-1(RNAi) germ lines. Depletion of CSR-1 results in the accumulation of large PGL-1 aggregates. (B) RNA staining with SYTO14 shows that RNA accumulates in the syncytial cytoplasm of the adult germ line after CSR-1 is depleted. (C) Endogenous PGL-3 (green) and GLH-1 (red) colocalize in large P-granule aggregates after CSR-1 is depleted. (D) In csr-1(RNAi) embryos (∼150-cell stage shown), PGL-3 accumulates in somatic cells, whereas GLH-1 is restricted to germ-line blastomeres. Bars, 10 μm.
CSR-1 is an Argonaute protein with endonucleolytic mRNA cleavage (slicing) activity (Yigit et al. 2006; Aoki et al. 2007). The slicing activity of Argonautes mediates siRNA silencing of mRNAs, which are then degraded by exonucleases in the exosome and decapping pathways (Orban and Izaurralde 2005). To determine if the formation of large P granules in csr-1(RNAi) animals is accompanied by an increase in RNA accumulation, we dissected hermaphrodite gonads in the presence of SYTO14, a nucleic acid stain that is used to visualize nucleolar and cytoplasmic RNA (Schisa et al. 2001). Both control(RNAi) and csr-1(RNAi) worms showed comparable SYTO14 staining in germ-line nucleoli. However, we observed a much higher level of RNA accumulation in the central cytoplasmic “rachis” of csr-1(RNAi) animals than in control worms fed empty RNAi vector (Figure 7B, gonads with high levels of RNA accumulation: control, 1 of 16 gonad arms; csr-1, 12 of 18 gonad arms). Our several attempts to concomitantly image cytoplasmic RNA accumulations and P granules were unsuccessful. Thus, we do not know if they colocalize. The different locations and morphologies of cytoplasmic RNA accumulations and enlarged P granules in csr-1(RNAi) suggest that P granules are not the sites of excessive RNA accumulation. Previous studies have shown that depletion of sperm and arrested ovulation cause RNA to accumulate in the rachis; this phenotype is regulated by the major sperm protein pathway (Schisa et al. 2001; Jud et al. 2008). csr-1(RNAi) animals exhibited no obvious defects in fertilization or ovulation, suggesting that RNA accumulation in csr-1(RNAi) animals is not caused by loss of the major sperm protein pathway. We hypothesize that the accumulation of cytoplasmic RNA and of enlarged P granules observed after depletion of CSR-1 is due to reduced slicing and degradation of mRNAs.
If loss of CSR-1 slicing activity causes enlarged and disorganized P granules, then loss of proteins that function in the CSR-1 pathway should cause similar defects. CSR-1 preferentially binds to secondary siRNAs, siRNAs that are produced from primary siRNA-targeted mRNAs by RNA-dependent RNA polymerase (RdRP) (Aoki et al. 2007). We tested whether enlarged and disorganized PGL-1 granules are seen in mutants lacking the somatic RdRP RRF-1 or either of the two RdRPs that are known to be active in the germ line, EGO-1 and RRF-3. We observed an accumulation of large PGL-1 granules in ego-1 mutant germ lines (as observed by Vought et al. 2005) but not in rrf-1 or rrf-3 mutant germ lines (Figure 8A). The PGL-1 phenotype in ego-1 and csr-1 mutants was typically more severe than the phenotype caused by RNAi; in mutant germ lines, the large PGL-1 aggregates were less perinuclear than observed following RNAi. Mutant germ lines may have less EGO-1 or CSR-1 than can be achieved by RNAi, or alternatively mutant germ lines may have compromised health and show more severe defects. Enlarged PGL-1 granules were not observed in RNAi-defective mutants such as dcr-1 (Vought et al. 2005) or rde-3 (our observations, Figure 8A), suggesting that the phenotype is not due to a general defect in RNAi. Next we examined GFP∷PGL-1 in drh-3(RNAi) germ lines. drh-3 encodes a DExH-box helicase that is required for RdRP activity and shares several phenotypes with ego-1 and csr-1 (Duchaine et al. 2006; Yigit et al. 2006; Aoki et al. 2007; Rocheleau et al. 2008; She et al. 2009). Depletion of drh-3 phenocopied the PGL-1 phenotype observed in ego-1 and csr-1 RNAi germ lines (Figure 8B), providing further evidence that DRH-3 functions in a pathway with EGO-1 and CSR-1.
PGL-1 accumulates in numerous large granules in the germ line after an endogenous siRNA pathway is compromised. (A) Endogenous PGL-1 staining (green) resembles wild type in the germ lines of rde-1, rrf-1, and rrf-3 mutants. Large PGL-1 granules accumulate in the germ lines of ego-1 and csr-1 mutants. Most granules are dissociated from the nuclear periphery. (B) GFP∷PGL-1 accumulation in control(RNAi), drh-3(RNAi), and ego-1(RNAi) germ lines. RNAi depletion of EGO-1 and DRH-3 produces the same large PGL-1 granule phenotype as depletion of CSR-1. Bars, 10 μm.
We wanted to see if elevated RNA accumulation correlates with elevated levels of PGL-1 or if the effect is specific to the depletion of csr-1. We found that, like csr-1, depletion of both ego-1 and drh-3 also resulted in increased cytoplasmic RNA as detected by SYTO14 staining (Table 3). We also examined other factors whose depletion resulted in elevated PGL-1 in the adult germ line (eft-1, let-711, phi-6, phi-7, phi-11, prp-8, and skp-1). In general, increased PGL-1 levels correlated with increased cytoplasmic RNA. Depletion of csr-1, ego-1, and the splicing factors phi-7, phi-11, and skp-1 had particularly strong effects on RNA patterns (Table 3). At this time we do not know if the accumulation of cytoplasmic RNA results from loss of function of CSR-1, EGO-1, DRH-3, and the handful of RNA-binding proteins whose depletion results in elevated PGL-1 or if the accumulation of RNA results from displaced and elevated PGL-1 itself. A third possibility is that P-granule proteins and cytoplasmic RNAs interact with and stabilize each other.
Correlation of elevated PGL-1 and elevated RNA in RNAi germ lines
CSR-1, EGO-1, and DRH-3 are thought to work together in a germ-line-specific endogenous siRNA pathway (Maine et al. 2005; Robert et al. 2005; Duchaine et al. 2006; Yigit et al. 2006; Aoki et al. 2007; Rocheleau et al. 2008; She et al. 2009). Argonautes, like CSR-1, have been identified as germ granule components from worms to mammals (Kotaja et al. 2006; Batista et al. 2008; Wang and Reinke 2008). In addition, an RGG domain, like that found in PGL-1 and PGL-3, in the N terminus of CSR-1 also suggests that CSR-1 is likely to associate with P granules. To test that prediction, we attempted to generate good anti-CSR-1 antibodies and informative GFP∷CSR-1 transgenic worms. Although both antibodies and transgenic worms showed P-granule localization, neither passed our tests for specificity. Nevertheless, on the basis of the specific PGL-1 phenotype of CSR-1 pathway mutants and the predicted P-granule localization of CSR-1, we hypothesize that P granules function as a center for endogenous siRNA silencing in the germ line and that disruption of this endo-siRNA pathway results in not only the accumulation in the germ-line syncytium of RNA transcripts that in wild type are silenced and degraded, but also the accumulation of large P granules.
DISCUSSION
Our RNAi screen proved to be an effective method to identify genes whose depletion causes aberrant PGL-1 phenotypes. One advantage of performing an RNAi screen over a forward mutagenesis screen is that mutants are not required to be viable and fertile. This allowed us to identify multiple gene classes, including some essential genes that are required for PGL-1 assembly and localization. One disadvantage of RNAi screens is missing some genes due to ineffective RNAi or incomplete RNAi library coverage. To expedite our screen, RNAi was performed for only 30 hr, and only P0 and F1 progeny were examined. Because of the maternal contribution of many P-granule proteins, some phenotypes, like that caused by DEPS-1 loss, are not observable until the F2 generation after RNAi is started. Our P0/F1 screen did not identify any additional genes whose depletion causes the PGL-1 phenotypes of deps-1 and glh-1 mutants where PGL-1 is dispersed in all cells of normally developing embryos. We do not know if DEPS-1 and GLH-1 are the only proteins whose loss causes that specific phenotype or if an F2 generation screen could identify more genes with that phenotype.
Despite the limitations of RNAi screens, this type of genomewide screen serves as an excellent starting point for identification and in-depth analysis of gene families in particular cellular processes. For example, the 5 NPPs identified in our screen drew our attention to nuclear pores as being involved in P-granule localization. Targeted retests of the remaining 13 NPP genes represented in the RNAi library identified 4 NPP genes that our screen should have identified, 5 NPP genes that have low-penetrance RNAi phenotypes and therefore require other analysis strategies, and 4 NPP genes that may not participate in regulating P-granule localization. Another example is our identification of one subunit of the SF3b splicing complex. Targeted retests identified two additional subunits. Loss of the remaining three SF3b components, which are not covered or inefficiently covered in the RNAi library, is likely to result in PGL-1 phenotypes as well. Finally, a recent report used a similar RNAi screen and identified a single autophagy component, lgg-1, which is required for clearing small P granules from somatic cells. Targeted tests of other autophagy components, by injection of double-stranded RNA to achieve stronger RNAi, revealed their involvement in clearing P granules from somatic cells (Zhang et al. 2009). Certainly, the 173 genes identified in our screen are an underestimate. We expect more stringent RNAi procedures, such as injecting or soaking or feeding for longer times, will identify additional genes that affect P granules in C. elegans.
The majority of genes identified in our screen were grouped into large gene classes. While some classes were likely identified because of P-granule segregation or degradation defects (cell cycle/polarity/cell division and proteasome/ubiquitin), most classes associate P granules with post-transcriptional regulation in the germ line. For example, splicing, nuclear pore, and ribosomal classes emphasize the importance of mRNA maturation, nuclear transport, and translational regulation in controlling P-granule size and location.
Nascent transcripts are processed in the nucleus by spliceosomes prior to their export as mature mRNAs. Sm proteins are core components of nuclear spliceosomes and are also concentrated in the cytoplasmic germ granules of multiple species (Moussa et al. 1994; Chuma et al. 2003; Barbee et al. 2002; Bilinski et al. 2004). In C. elegans, Sm proteins are required for proper P-granule localization to the nuclear periphery; it has been proposed that this role is independent of pre-mRNA splicing (Barbee and Evans 2006). We, too, identified the Sm protein, SNR-2, as required for proper PGL-1 granule assembly. We also identified other splicing and mRNA maturation components. These include the cleavage and polyadenylation factor CPF-2 and the putative integrator complex subunit C06A5.1, both of which are involved in maturation of 3′ ends of mRNA; the putative exon–exon junction factors MAG-1 and RNP-4; possible snRNP associated factor PRP-8 along with SF3b splicesome components PHI-6, PHI-11, and SAP-49; and other factors whose homologs are required for pre-mRNA splicing, including RSP-7, SKP-1, PHI-7, PHI-10, and PHI-12. While the role of Sm proteins in germ granules may still be independent of their role in splicing, our results suggest that proper mRNA maturation in the nucleus affects the assembly and localization of P granules. It will be interesting to determine whether properly assembled P granules require specific features of spliced mRNAs, splicing factors associated with mRNAs, or mRNA export through nuclear pores.
The integrity of nuclear pores appears to be vital for the perinuclear distribution of P granules. At least 14 of 20 nuclear pore components are required for PGL-1's perinuclear distribution, as are nuclear transport factors such as IMB-1, IMB-5, RAN-1, and RAN-4. Nuclear lamin LMN-1 is also required, as is dynein light chain-1, DLC-1. DLC-1's yeast homolog DYN2 is a nucleoporin (Stelter et al. 2007), and interestingly C. elegans DLC-1 was found to interact with PGL-3 in a high-throughput binding assay (Li et al. 2004). Therefore, it is possible that some nucleoporins, like the putative nucleoporin DLC-1, provide a physical linkage between P granules and nuclear pores. On that note, we find it interesting that the P-granule components GLH-1, GLH-2, and GLH-4 contain multiple Phe–Gly (FG) repeat domains, domains that are found in many nucleoporins (Rexach and Blobel 1995). Some FG domains form cohesive hydrophobic interactions with other FG domains in the nuclear pore (Patel et al. 2007). The close association between P granules and nuclear pores, as well as similarities between GLH proteins and nucleoporins, has led to the prediction that P granules serve as an extension of nuclear pores in the germ line (Kuznicki et al. 2000; Pitt et al. 2000; Schisa et al. 2001). We are currently testing whether the FG domains of GLH-1, GLH-2, and GLH-4 interact with each other and with other FG-containing nucleoporins and whether FG proteins in P granules do indeed serve to extend the nuclear pore environment.
As RNAs transcribed in germ nuclei exit through nuclear pores, many transcripts will encounter P granules. Poly(A)+ and developmentally regulated mRNAs have been shown to accumulate in P granules (Seydoux and Fire 1994; Schisa et al. 2001). P granules may act as a scaffold for multiple RNA processing activities, possibly by providing an environment that facilitates germ-line-specific binding interactions and regulates the translation, transport, modification, storage, and/or degradation of RNA. Our screen results support many of these possibilities. For example, we found that multiple ribosomal components are required for proper P-granule accumulation and stability. This finding could reflect a need for synthesis of P-granule proteins or a requirement for ongoing translation during the assembly of P granules.
We observed PGL-1 phenotypes when RNA deadenlyase, decapping, and degradation components were depleted. Depletion of LET-711, the NOT1 ortholog of the CCR4/NOT1 mRNA deadenylation complex, had previously been reported to cause PGL-1 localization outside of the embryonic germ line (Gallo et al. 2008). We identified LET-711 primarily on the basis of the elevated accumulation of transgenic and endogenous PGL-1 in the adult germ line. This phenotype was observed within 30 hr after initiation of RNAi and before germ-line morphology became altered, suggesting that the PGL-1 phenotype was a direct result of LET-711 depletion. The PGL-1 aggregates in let-711(RNAi) germ lines are smaller than those observed in csr-1 mutants, but as in the case of csr-1, the enhanced accumulation may be correlated with RNA stabilization. The accumulation of P granules may be sensitive to cytoplasmic RNA levels, requiring RNA destabilizing factors like LET-711 and CSR-1 to control RNA levels through exo- or endonucleolytic activity.
One exciting gene that emerged from our screen was csr-1. CSR-1 is an Argonaute that exhibits endonucleolytic or slicing activity on RNA targets, showing a preference for using secondary siRNAs generated by RdRP activity rather than primary siRNAs generated by Dicer (Aoki et al. 2007). Thus far, C. elegans is unique among animals in its large number of Argonaute-encoding genes: 27 in C. elegans compared to 5 in Drosophila and 8 in humans (Hutvagner and Simard 2008). The diversity of Argonautes in C. elegans has proven advantageous for dissecting out the multiple roles of these proteins. C. elegans Argonautes ALG-1 and ALG-2 are essential for miRNA-mediated gene silencing, RDE-1 for exogenous RNAi, ERGO-1 for endogenous RNAi, and PRG-1 and PRG-2 for piRNA-mediated gene silencing, and a handful of other Argonautes are thought to bind secondary siRNAs but lack the ability to slice RNA targets (Grishok et al. 2001; Yigit et al. 2006; Batista et al. 2008; Das et al. 2008; Wang and Reinke 2008). csr-1 was first obtained in a germ-line cosuppression screen, which suggested that it may be involved in silencing repetitive transgene arrays in the germ line (Robert et al. 2005). csr-1 worms are also partially deficient in RNAi (Yigit et al. 2006). It was recently discovered that csr-1 shares multiple phenotypes with ego-1 and drh-3, which encode an RdRP and an RdRP interacting factor required for small endogenous RNAi in the germ line (Aoki et al. 2007; Rocheleau et al. 2008; She et al. 2009).
We discovered that csr-1 worms accumulate large P granules in the adult germ line and ectopic granules in somatic cells of embryos. This phenotype is accompanied by a dramatic increase in RNA in the central cytoplasm of the germ line. While it was not obtained in our original screen, our targeted experiments demonstrated that depletion of EGO-1 and DRH-3 phenocopy the very specific PGL-1 phenotype of csr-1, supporting a role for CSR-1 in a germ-line-specific endo-RNAi pathway. One model proposed previously to explain the irregular P granules in ego-1 mutants is that EGO-1 may promote a chromatin state that regulates nuclear pore distribution and that the clumping of nuclear pore material leads to perturbed P-granule assembly (Vought et al. 2005). This model is strengthened by the observation that EGO-1 is found in nuclear fractions and is required for unpaired chromosomal regions to accumulate high levels of histone H3 lysine K9 dimethylation, a mark associated with heterochromatin assembly and transcriptional silencing (Maine et al. 2005; She et al. 2009). However, we observed a different PGL-1 phenotype when nuclear pore components were depleted (PGL-1 granules dissociated from the nuclear periphery in germ-line blastomeres of developing embryos) than we did when CSR-1, EGO-1, and DRH-3 were depleted (enhanced accumulation of large PGL-1 granules in the adult germ line). Loss of the nuclear envelope component LMN-1, which also results in the clustering of nuclear pores (Liu et al. 2000), causes a P-granule phenotype that is more similar to that caused by depletion of core nuclear pore proteins than to the large P-granule phenotype caused by depletion of CSR-1, EGO-1, and DRH-3. Another simple model is that genes typically silenced by endo-RNAi in the germ line are desilenced when csr-1 is depleted, allowing these genes' transcripts to accumulate in the cytoplasm. In normal germ lines, these transcripts could be silenced at the level of chromatin and regulation of transcription or at the level of exiting the nucleus into P granules where they are sliced by CSR-1 and subsequently degraded. As all known P-granule components are thought to function in RNA metabolism, the accumulation of large P granules in csr-1 animals may be due to increased levels of cytoplasmic RNA in csr-1 germ lines. Another possibility is that transcripts encoding P-granule proteins are normally downregulated by endo-RNAi, and the loss of this downregulation in csr-1 animals leads to the accumulation of large P granules. Profiling germ-line transcripts in csr-1 mutants may reveal which, if any, of these models are correct.
The P-granule components DEPS-1 and PGL-1 are both required for a robust exogenous RNAi response (Robert et al. 2005; Spike et al. 2008b). While PGL-1 has not been implicated in a specific RNAi pathway, deps-1 mutants are thought to resist the effects of RNAi due to a 7- to 10-fold decrease in rde-4 expression (Spike et al. 2008b). RDE-4 binds dsRNA and forms a complex with Dicer to produce siRNAs, and rde-4 mutants are RNAi defective (Tabara et al. 2002). Another component identified in screens for RNAi-defective mutants, RDE-3, is thought to act downstream of RDE-4 and Dicer. Comparison of transcriptional profiles of rde-3 and deps-1 mutants showed that nearly 30% of genes upregulated in deps-1 mutant germ lines are also upregulated in rde-3 worms, suggesting that RDE-3 and DEPS-1 function together to regulate expression of at least their shared target genes (Spike et al. 2008b). As DEPS-1 is a constitutive P-granule component, it likely participates in regulation at a post-transcriptional level. Our results with CSR-1, EGO-1, and DRH-3 strengthen the emerging view that P granules and RNAi are intimately related: P-granule components participate in regulating RNAi, and RNAi factors influence the structure, localization, and perhaps function of P granules. Better defining the RNAi–P-granule relationship will shed light on mechanisms of post-transcriptional gene regulation used by germ cells.
Acknowledgments
We thank Colin Thacker for training provided during the screen, use of equipment including the Copas Biosort, and preparation of RNAi colonies from the Ahringer RNAi library; Robb Cundick for the development and use of a C. elegans RNAi database; Susan Mango for the use of lab space and equipment during the first pass of screening; Julie Ahringer for unpublished information on the imb-2, imb-3, and imb-5 PGL-1 phenotypes; Xingyu She and Eleanor Maine for providing ego-1 strain EL500; Andreas Rechtsteiner for comparison of germ-line data sets; and the National Bioresource Project at Tokyo Women's Medical College for providing the FX892 strain. This work was supported by Ruth Kirchstein National Research Service Award postdoctoral fellowship GM084673 (to D.U.) and National Institutes of Health (NIH) grant GM34059 (to S.S.). Some C. elegans strains were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources.
Note added in proof: Since the acceptance of this article for publication, an additional study was published that shows abnormal P granules in csr-1, drh-3, and ego-1 mutants (J. M. Claycomb, P. J. Batista, K. M. Pang, W. Gu, J. J. Vasale et al., 2009, The Argonaute CSR-1 and its 22G-RNA cofactors are required for holocentric chromosome segregation. Cell 139: 123–134). Claycomb et al. also show that CSR-1, DRH-3, and EGO-1 colocalize with P granules.
Footnotes
Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.109.110171/DC1.
Communicating editor: B. J. Meyer
- Received September 22, 2009.
- Accepted September 28, 2009.
- Copyright © 2009 by the Genetics Society of America