PRMT5 and the role of symmetrical dimethylarginine in chromatoid bodies of planarian stem cells

The replenishment, survival and proper differentiation of stem cells is necessary for development, continuous production of gametes and the repair of lost or damaged tissues. It is therefore important to understand the mechanisms responsible for regulating the maintenance, proliferation and proper differentiation of stem cells. The capabilities of stem cells are vividly displayed during post-embryonic development of planarian flatworms. A population of planarian adult stem cells, known as neoblasts (Dubois, 1948; Dubois and Wolff, 1947), are the only dividing cells in the planarian soma: they are the sole source of additional neoblasts and differentiated cells during regeneration, growth and homeostasis (Baguñà, 1976; Newmark and Sanchez Alvarado, 2000; Newmark and Sanchez Alvarado, 2002). Single cells within this population (clonogenic neoblasts; cNeoblasts) are sufficient to rescue planarians depleted of stem cells, further demonstrating the self-renewal and differentiation capacities of this cell type (Wagner et al., 2011). Thus, planarians provide a valuable opportunity for understanding the mechanisms involved in maintenance and function of stem cells in vivo.

Many studies have identified genes required for neoblast maintenance and function in planarians (Bonuccelli et al., 2010; Conte et al., 2009; Fernandez-Taboada et al., 2010; Guo et al., 2006; Oviedo and Levin, 2007; Palakodeti et al., 2008; Pearson and Sanchez Alvarado, 2010; Reddien et al., 2005a; Reddien et al., 2005b; Rouhana et al., 2010; Salvetti et al., 2005; Scimone et al., 2010; Solana et al., 2009; Tasaki et al., 2011a; Tasaki et al., 2011b). The vast majority of genes required specifically for neoblast function encode conserved factors involved in post-transcriptional regulation of gene expression (reviewed by Shibata et al., 2010). For example, homologs of Pumilio (Salvetti et al., 2005), Bruno (Guo et al., 2006) and Argonaute-2 (Rouhana et al., 2010) are required for maintenance of neoblasts and, consequently, regeneration. Conversely, RNA helicases (Vasa, Me31b and UPF1) and exoribonuclease 1 (XRN1) are required for regeneration but not for neoblast maintenance, suggesting a role in differentiation (Rouhana et al., 2010). Members of the PIWI subfamily of proteins, which promote genomic stability by repressing transposition in metazoan germ cells (reviewed by Siomi et al., 2008; Siomi et al., 2011), are required for neoblast function in planarians (Reddien et al., 2005b; Palakodeti et al., 2008).

Large cytoplasmic ribonucleoprotein (RNP) granules, named chromatoid bodies (CBs), were described in neoblasts decades ago (Coward, 1974; Hay and Coward, 1975; Hori, 1982; Morita et al., 1969). CBs contain conserved post-transcriptional regulators, such as the DEAD box RNA helicase CBC-1 (CB component-1/Me31b/Dhh1/DDX6) (Yoshida-Kashikawa et al., 2007), the small nuclear ribonucleoproteins (snRNP) core protein SmB (Fernandez-Taboada et al., 2010) and a Tudor homolog (Spoltud-1; Solana et al., 2009). All of these factors are required for regeneration in planarians (Fernandez-Taboada et al., 2010; Rouhana et al., 2010; Solana et al., 2009).

Tudor domain-containing proteins (TDRDs) control the assembly of polar granules, RNP organelles required for germ cell formation in Drosophila (Boswell and Mahowald, 1985; Mahowald, 1962; Thomson and Lasko, 2004), and CBs in mice (Yabuta et al., 2011). It has been proposed that TDRDs serve as docking platforms for the assembly of RNP granules involved in germ cell formation (Arkov et al., 2006). Physical interaction of TDRDs with snRNP proteins SmB and SmD3 is enabled by symmetrical dimethylarginine (sDMA) modification of RG motifs of Sm proteins (Anne et al., 2007; Brahms et al., 2001; Cote and Richard, 2005; Friesen et al., 2001; Sprangers et al., 2003). This post-translational modification is catalyzed by the Type II protein arginine methyl transferase PRMT5 in mammals (Meister et al., 2001) and its ortholog Capsuléen in flies (Anne et al., 2007).

sDMAs form the major epitope for recognition of Sm proteins by the widely used Y12 monoclonal antibody (Brahms et al., 2000; Lerner et al., 1981), which also recognizes sDMAs on mouse and fly PIWI (Kirino et al., 2009) and Vasa (Kirino et al., 2010a) proteins. Indeed, PIWI complexes containing PRMT5 and TDRDs have been isolated from mouse germline cells (Reuter et al., 2009; Vagin et al., 2009b; Vasileva et al., 2009; Wang et al., 2009). sDMA modifications on PIWI and Vasa are also mediated by PRMT5, are required for the physical interaction between PIWI and TDRDs, and drive localization of PIWI to cytoplasmic foci (Kirino et al., 2009; Liu, H. et al., 2010; Liu, K. et al., 2010; Nishida et al., 2009; Vagin et al., 2009b).

The composition of cytoplasmic RNP granules in somatic stem cells has been investigated less extensively than that of their germline-restricted counterparts. Here, we examine sDMA modifications in CB components of planarian neoblasts, identify the enzyme responsible for this modification and reveal novel CB components.

The clonal asexual strain CIW4 (Sanchez Alvarado et al., 2002) and a hermaphroditic strain (Zayas et al., 2005) of Schmidtea mediterranea were used and maintained as described by Wang et al. (Wang et al., 2007).

Asexual planarians 3-5 mm in length were exposed to 40 Gy of gamma irradiation using a Gammacell-220 Excel with a cobalt-60 source (Nordion, Ottawa, ON, Canada) in 2 ml of planarian salts (Cebria and Newmark, 2005) and processed at the indicated time points.

ISH was performed as described (Pearson et al., 2009), with modifications as per Wang et al. (Wang et al., 2010) for large hermaphrodites.

Animals for anti-PCNA (1:500; Orii et al., 2005) and anti-phospho-histone H3 (Ser10) D2C8 (1:1000; Cell Signaling Technology, Danver, MA, USA) analyses were killed in 2% HCl and fixed for two hours in 4% formaldehyde, 5% methanol in PBS. Y12 (NeoMarkers, Fremont, CA, USA) was used at 1:250. Goat anti-mouse Alexa-488 (1:1000) and goat anti-rabbit Alexa-568 (1:500) (Molecular Probes, Eugene, OR, USA) were used for detection of primary antibodies. Immunofluorescence was performed and visualized as described by Forsthoefel et al. (Forsthoefel et al., 2011) following overnight bleaching or fluorescent ISH.

Double-stranded RNA (dsRNA) feedings were performed as described by Rouhana et al. (Rouhana et al., 2010) with modifications (Collins et al., 2010). Briefly, dsRNA diluted to 0.1 μg/μl in 2:1 minced liver:ultra-pure water was presented to planarians twice a week (250 ng per animal). Non-planarian dsRNA encoded in pJC53.2 (Collins et al., 2010) served as negative control.

Polysciences (Warrington, PA, USA) provided chemicals unless otherwise stated. We processed and analyzed three animals per condition for EM and immuno-EM. Animals were fixed in iced 2% formaldehyde, 2.5% glutaraldehyde in EMBuffer (70 mM sodium cacodylate, 1 mM CaCl₂, pH 7.4), excised, fixed for four additional hours, washed twice in EMBuffer, post-fixed (1% OsO₄, 90 minutes, 4°C in dark), washed twice (EMBuffer), dehydrated (20%+2% uranyl acetate, 40%, 60%, 80%, 100% ethanol), gradually placed in acetone, and infiltrated (Mollenhauer, 1964). Sections (80 nm) were stained as described by Vennable and Coggeshall (Vennable and Coggeshall, 1965) and were imaged using a Phillips CM-200.

For immuno-EM, fixed samples were rinsed twice (EMBuffer), quenched with 0.1 M glycine, and dehydrated, infiltrated and cured using Lowicryl-K4M according to the manufacturer’s instructions. Sections collected on FORMVAR-coated grids were treated with IEMBlock (TBST, IgG-free 0.5% BSA, 0.45% fish gelatin), incubated for 1 hour in IEMBlock with or without Y12 (1:1), washed in TBST, incubated for 1 hour with 10 nm immunogold bead-conjugated anti-mouse antibodies (Molecular Probes) in IEMBlock (1:2), TBST-washed, rinsed in water, stained with 4% uranyl acetate, and imaged described as above.

Groups of four 3-4 mm asexual or two 1-1.2 cm sexual planarians (rinsed) were snap-frozen, resuspended in 50 or 100 μl SDS loading dye, respectively, homogenized with Kontes pestles, snap-frozen once again, boiled for three minutes, vortexed and centrifuged for 10 minutes (20,000 g), all at 4°C. The equivalent of half to three-fourths of an asexual planarian, or one-fifth of a sexual planarian was used for immunoblotting. Membranes blocked in 1% BSA in PBS, were incubated with Y12 (1:250, NeoMarkers, Fremont, CA, USA) or anti-actin A2066 (1:1000, Sigma, St Louis, MO, USA) in 0.05% Tween-20 PBS overnight at 4°C.

All steps were performed in medium salt buffer [MSB; 150 mM NaCl, 50 mM Tris-HCl (pH 7.5), 0.05% NP-40, Roche EDTA-free protease inhibitors] at 4°C. Extracts were prepared from eighteen 5-7 mm asexual planarians snap-frozen, homogenized in 200 μl MSB, snap-frozen once again, supplemented with 200 μl MSB, further disassociated by pipetting, snap-frozen, thawed, centrifuged for ten minutes (8000 g), re-cleared for five minutes (20,000 g) and split between 4 μl of control or Y12 mAb [courtesy of Drs J. Steitz (Yale University) and G. Dreyfuss (University of Pennsylvania)] (Lerner et al., 1981; Pettersson et al., 1984) bound to 30 μl of protein G-Dynabeads slurry (Invitrogen, Carlsbad, CA, USA). FE-J1 antibodies (Federson et al., 1984) (Developmental Studies Hybridoma Bank, University of Iowa, IA, USA) served as a negative control. Immunoprecipitation and RNA analysis were performed as per Nishida et al. (Nishida et al., 2009) and Kirino et al. (Kirino et al., 2011) with minor modifications. [γ-³²P] ATP-labeled RNA was RQ1 DNase-treated (Promega, Madison, WI, USA) or RNaseT1-treated (Ambion, Austin, TX, USA), and resolved by 12% denaturing polyacrylamide gel electrophoresis.

RNA from two-month-old control and PRMT5 RNAi-treated planarians was isolated using TRIzol (Invitrogen), DNase-treated, purified with RNA Clean and Concentrator kit (Zymo, Irvine, CA, USA) and submitted to the W. M. Keck Center for Comparative and Functional Genomics for Illumina sequencing. Reads were mapped to a reference transcriptome (https://www.ideals.illinois.edu/handle/2142/28689) using CLC Genomics Workbench and compared as per Marioni et al. (Marioni et al., 2008). Sequence Read Archive (SRA) data were deposited at NCBI under project accession PRJNA79031.

Transcript quantification in control and PRMT5(RNAi) animals was performed as described by Miller and Newmark (Miller and Newmark, 2012) using GoTaq qPCR reagent (Promega), with the following primers: 5′-CCCACCGAGATATTCCAAAA-3′, 5′-CCCCATCCTCTTTGACTTTG-3′ for Smed-PRMT5; 5′-GGGCTAATCCAAATCCTGGT-3′, 5′-TGCTGCAATACACTCGGAGA-3′ for Smedwi-1 (Eisenhofer et al., 2008); and 5′-TGGCTGCTTGTGATCCAAGA-3′, 5′-AAATTGCCGCAACAGTCAAATA-3′ for β-tubulin.

cDNA clones for gH4 and nanos (Wang et al., 2007), Smedwi-1 (Reddien et al., 2005b), Smedwi-2 (Palakodeti et al., 2008; Reddien et al., 2005b) and Smedwi-3 (Palakodeti et al., 2008), were obtained from an expressed sequence tag (EST) library (Zayas et al., 2005). PRMT homologs were amplified from planarian cDNA and cloned into pJC53.2 using the following: 5′-ATGTTGATGTTAGCTTTGCCCAAGTCTCCAGTCGATCAG-3′, 5′-GGACATGGAATAAGAGCGGCCATTCGGGTTATGAATCGACG-3′ for Smed-PRMT5 (GenBankID:JQ035529); 5′-ATGGAGCACGCATCAATCAATGGAGATGCTATTGAAGCATC-3′, 5′-CCAGCTCAAATTATCGCATTCGATAACTGAATGTTTCCTCC-3′ for Smed005105 (GenBankID:JQ035531); and 5′-CAATGGACCACAATAGCAATGCAATGGATACAAATGAC-3′, 5′-TCTCCCTTAAACTTCAGATCAATACTGACATCAATTGCTCTAGC-3′ for Smed001246 (GenBankID:JQ035530).

Previous studies of germline development in the sexual strain of S. mediterranea revealed the presence of Y12-immunoreactive material in the cytoplasm of spermatogonia (Fig. 1A) (Wang et al., 2010). Y12-immunoreactive material is also detected in neoblasts of asexual planarians (supplementary material Fig. S1). Y12 specifically recognizes sDMA (Brahms et al., 2000), suggesting that neoblasts express machinery necessary for this modification. Recent reports have shown that interactions between proteins containing sDMA and TDRDs are important for RNP aggregation in mouse and fly germlines (Anne, 2010; Gonsalvez et al., 2010; Kirino et al., 2010b; Vagin et al., 2009b). Given that neoblasts are enriched with expression of tudor homologs (Rouhana et al., 2010; Yoshida-Kashikawa et al., 2007) and that one homolog is a neoblast CB component (Solana et al., 2009), we analyzed sDMA-containing proteins in neoblasts.

Neoblasts are the only proliferating somatic cells in planarians and expression of proliferating cell nuclear antigen (PCNA) has been used as a marker for this cell type (Orii et al., 2005). We performed simultaneous immunofluorescence staining with Y12 and anti-PCNA antibodies. Indeed, PCNA-expressing cells contained Y12-immunoreactive material in their cytoplasm (Fig. 1B), confirming the presence of sDMAs in neoblast proteins. PCNA was absent from some Y12-labeled cells (∼31%; n=94), as is expected given the cyclic expression of this gene and the fact that not all neoblasts are in S-phase at any given time (Newmark and Sanchez Alvarado, 2000). The presence of sDMA-containing protein in the cytoplasm of neoblasts was confirmed further by performing simultaneous Y12 immunofluorescence and in situ hybridization (ISH) for germinal histone H4 (gH4) RNA, a known neoblast and germ-cell marker (Wang et al., 2007). Indeed, Y12-labeled cytoplasmic aggregates and gH4 transcripts were detected in the same cells (Fig. 1C; 98.6%; n=139), and often expressed in the same cytoplasmic foci (Fig. 1C). Together, these data show that Y12-immunoreactive material is present in neoblasts and preferentially localized to cytoplasmic RNP foci.

The distribution of Y12-labeled cells corresponds to that described for undifferentiated neoblasts (supplementary material Fig. S1) (Eisenhoffer et al., 2008; Newmark and Sanchez Alvarado, 2000; Orii et al., 2005; Shibata et al., 1999). Neoblasts reside in the mesenchyme of intact planarians posterior to the photoreceptors and accumulate in the pre-existing tissue near the wound site in regenerating animals; neoblast progeny reside inside the emerging stump (the blastema) of regenerating planarians. Whole-mount immunofluorescence staining of intact and regenerating asexual planarians using Y12 and antibodies against the mitotic marker anti-phospho-histone H3 (Ser10) (PH3), showed that the distribution of Y12-immunoreactive material reflected that of undifferentiated neoblasts (supplementary material Fig. S1). Y12 staining was most intense in the region immediately posterior to the blastema, as originally observed by EM analyses of neoblasts (Morita et al., 1969). Nevertheless, a relatively weak and much less aggregated signal was also observed in the cytoplasm of cells in the blastema during regeneration (supplementary material Fig. S1), suggesting the persistence of some methylated material in neoblast progeny.

Among planarian cell types, CBs are an unambiguous characteristic of neoblasts and germ cells (Coward, 1974; Hori, 1982; Morita et al., 1969). We hypothesized that epitopes recognized by Y12 are CB components based on their perinuclear localization in neoblasts and colocalization with RNA (Fig. 1B,C). To test this idea, we performed immuno-EM on regenerating planarians three days post-amputation, because neoblasts accumulate at the base of the blastema. We found strong Y12 immunoreactivity in neoblasts, particularly in CBs (Fig. 1D), which did not appear in control sections prepared without primary antibody. We counted the number of colloidal gold-particles present in areas containing CBs (∼564.3 counts/μm²; n=8 cells), compared with nuclear (∼36.4 counts/μm²) or cytoplasmic neighboring areas (∼8.4 counts/μm²). This 15-fold higher Y12 signal in CBs than elsewhere in cells reveals that CBs are indeed enriched with sDMA-containing protein.

Immunoblot analysis of planarian protein extracts revealed Y12-reactive proteins of approximately 13, 24, 28, 85 and 110 kDa, and a minor polypeptide migrating at ∼15 kDa (Fig. 1E). The estimated sizes of some of these polypeptides correspond to planarian homologs of sDMA-containing proteins [such as SmB, SmD and PIWI (Table 1)], recognition of which by Y12 is evolutionarily conserved between flies and mammals (Anne et al., 2007; Gonsalvez et al., 2006; Kirino et al., 2009; Nishida et al., 2009; Vagin et al., 2009b). The conservation of canonical sDMA methylation motifs in planarian Sm and PIWI homologs was confirmed by sequence analysis (Table 1). Exposure to gamma irradiation, which leads to neoblast loss within 24 hours and ablation of differentiating neoblasts within four days (Eisenhoffer et al., 2008), caused a noticeable reduction of ∼85 and ∼110 kDa Y12 immunoreactive material within 24 hours of treatment, and complete loss of ∼13, ∼85 and ∼110 kDa immunoreactive material within four days of treatment (Fig. 1E). The detection of two other major polypeptides by Y12 (∼24 kDa and 28 kDa) was reduced, although not completely eliminated, in irradiated animals (Fig. 1E). The location of residual protein recognized by Y12 in differentiated cells is not clear; we cannot detect Y12 signal by immunofluorescence after irradiation (supplementary material Fig. S2A). From these observations, we conclude that Y12 labels material enriched in CBs of planarian neoblasts, suggesting that they contain sDMA-modified ribonucleoproteins.

To examine whether material recognized by Y12 in neoblasts is indeed sDMA, we identified homologs of Type I (asymmetrical) and Type II (symmetrical) protein arginine methyltransferases in the S. mediterranea genome (supplementary material Table S1) (Robb et al., 2008). Among these genes was Smed-PRMT5 (S. mediterranea protein arginine methyltransferase 5) mammalian and fly orthologs of which methylate PIWI (Kirino et al., 2009; Nishida et al., 2009), Vasa (Kirino et al., 2010a) and Sm proteins (Anne et al., 2007; Gonsalvez et al., 2006; Meister et al., 2001). Methylation by PRMT5 is required for recognition of these proteins by Y12 (Anne et al., 2007; Kirino et al., 2009; Kirino et al., 2010a). Thus, we suspected that PRMT5 could be responsible for the methylation of at least some of the sDMA-containing proteins present in planarians. To test this, we inhibited the expression of PRMT5 by RNA interference (RNAi) and tested for levels of sDMA recognized by Y12. In PRMT5(RNAi) asexual planarians, we found a dramatic reduction of Y12-immunoreactive material in neoblasts (supplementary material Fig. S2B) and in protein extracts (Fig. 2A). Similarly, sDMA modifications were prominent in spermatogonia of sexual control animals (Fig. 2B), but were undetectable in PRMT5(RNAi) animals (Fig. 2C). Disrupting expression of homologs of PRMT1, the predominant mammalian Type I arginine methyltransferase (supplementary material Table S1) (Tang et al., 2000), did not alter sDMA detection by Y12 (Fig. 2D,E), validating the specificity of effects seen after PRMT5 RNAi. From these results, we conclude that material recognized by Y12 in planarian cells and CBs is indeed sDMA-containing protein and that the enzyme responsible for this modification in S. mediterranea is PRMT5.

To identify the cell types in which PRMT5 is expressed, we performed ISH on sexual and asexual planarians. PRMT5 transcript was detected in the brain, neoblasts and testes of sexual planarians, predominantly in the outer layer spermatogonia (Fig. 2F). In asexual animals, PRMT5 expression was detected in mesenchymal cells and the cephalic ganglia (Fig. 2G-I). As expected for genes expressed in neoblasts, we observed reduction of mesenchymal PRMT5 transcripts in animals fixed 24- and 96-hours post-irradiation (Fig. 2H,I). PRMT5 expression was also detected in the ovaries of sexually mature planarians (Fig. 2J), leading us to examine sDMA-containing proteins in ovaries. We found that Y12 recognized ooplasmic material present in large granules (Fig. 2J). However, these granules seemed to be lost during progression through oogenesis and were not detected in more mature oocytes (Fig. 2J). Expression of PRMT5 in neoblasts and germ cells is consistent with a role for PRMT5 in generating sDMA epitopes in these tissues.

The detection of PRMT5 activity in planarian neoblasts and the enrichment of its product (sDMA) in CBs suggest an involvement of PRMT5 in neoblast function. To test the role of PRMT5 in planarian neoblasts, we inhibited PRMT5 expression by RNAi. As shown above, PRMT5 RNAi specifically reduces Y12-detectable sDMA modifications (Fig. 2A-C). Animals were subjected to a long-term regimen of biweekly PRMT5 and control double-stranded RNA (dsRNA) treatments (see Materials and methods). After two- to three-months of RNAi, asexual PRMT5(RNAi) animals were ∼35% smaller than control animals subjected to the same diet and amounts of dsRNA [average 5.86±0.65 mm (n=19) vs 3.77±0.56 mm (n=15)] (Fig. 3A). At this point, anterior blisters and head defects were observed in >25% of asexual PRMT5(RNAi) (n=21/76), some of which exhibited complete loss of the head region (n=6/76; Fig. 3B; supplementary material Fig. S3A). Head loss was also observed in PRMT5(RNAi) sexual animals (n=9/11; supplementary material Fig. S3B).

To analyze whether the defects observed after PRMT5 RNAi could be due to changes in neoblasts, we analyzed neoblast distribution by gH4 ISH. A substantial decrease in the number of gH4-expressing cells was found in severely affected PRMT5(RNAi) animals, which was not observed in controls (Fig. 3C-E). However, this reduction in neoblast number was not observed in all individuals nor distributed evenly throughout the entire body of single individuals subjected to PRMT5 RNAi (supplementary material Fig. S3C), complicating the quantification of neoblasts by whole-mount gH4 ISH (Fig. 3C-E). Thus, we analyzed the expression of another well-characterized neoblast marker, Smedwi-1 (Reddien et al., 2005b), in PRMT5(RNAi) and control animals. Total RNA from groups of PRMT5(RNAi) and control animals was extracted and subjected to reverse transcription-coupled quantitative PCR analysis (qRT-PCR; Fig. 3F). PRMT5 RNAi resulted in a modest yet reproducible reduction of Smedwi-1 RNA levels compared with controls (30-45% reduction; Fig. 3F), confirming an overall reduction of neoblasts after PRMT5 RNAi. The range of observed phenotypes (Fig. 3B-E; supplementary material Fig. S3) is unlikely to be due to insufficient knockdown, as PRMT5 transcript levels (Fig. 3F) and detection of sDMA by Y12 (Fig. 2A) are both dramatically reduced by this RNAi regimen. The compromised ability of PRMT5(RNAi) animals to sustain homeostatic size and growth, as well as the development of blisters, head loss and reduction of neoblasts, raised the possibility that PRMT5 could be required for proper neoblast function.

The highest demand for neoblasts is during regeneration, when amputation-induced signals drive an increase in proliferation; differentiation of newly born neoblasts will help replace missing body parts (Baguñà, 1976; Best et al., 1968; Saló and Baguñà, 1984; Wenemoser and Reddien, 2010). The homeostatic defects observed as a result of PRMT5 RNAi suggest that regeneration might also be compromised in these animals. Thus, after one month of RNAi, we challenged fragments of PRMT5(RNAi) animals to regenerate their heads and tails after amputation anterior and posterior to the pharynx. Under these conditions, the effect upon neoblasts should be less severe than those observed in two-month knockdowns (Fig. 3D-F). Indeed, 11% and 10% of PRMT5(RNAi) trunk and tail fragments (n=36 and n=29, respectively) failed to regenerate normal heads, as judged by the presence of photoreceptors on the seventh day after amputation (Fig. 4A-D). Additionally, 24% of tail fragments showed abnormal head regeneration with only a single photoreceptor, and 21% of PRMT5(RNAi) head fragments (n=34) failed to regenerate a tail (Fig. 4A). Conversely, all control fragments (n=38 and n=32, respectively) displayed successful head regeneration and only one of 38 head fragments failed to regenerate a tail (Fig. 4A,B). These results demonstrate that PRMT5 promotes proper regeneration.

Ultrastructural analysis of trunk fragments three days after amputation revealed additional differences between head-regenerating regions of control and PRMT5(RNAi) samples. The organized architecture of the blastema and postblastema region observed in controls (Fig. 5A) seemed distorted in PRMT5(RNAi) samples (Fig. 5B). Qualitative analyses revealed that neoblasts of similar shape and size covered the area close to the site of amputation in control animals (Fig. 5A). However, irregularities in shape, size and distribution of neoblasts were observed in PRMT5(RNAi) fragments (Fig. 5B). Differentiated structures, such as ciliated ducts of the protonephridial system (not shown) and secretory cell structures (Fig. 5A, arrows), were observed more often in sections of control animals by this time point, than in PRMT5(RNAi) sections (Fig. 5B). Similarly, germ cell clusters, which arise from neoblasts in adult planarians (Wang et al., 2007; reviewed by Newmark et al., 2008), were reduced in regenerating PRMT5(RNAi) head fragments relative to comparatively sized control fragments (P<0.05, two-tailed unpaired t-test; supplementary material Fig. S4). Furthermore, a 52% reduction in the number of CBs per neoblast was observed in sections of PRMT5(RNAi) animals (P<0.05, two-tailed unpaired t-test, n=82 controls and n=60 PRMT5(RNAi) neoblasts; Fig. 5C). These results suggest that PRMT5 plays a role in promoting CB stability or nucleation.

Examples of Sm and PIWI protein family methylation by PRMT5 have been observed in flies and mice, and primary sequence analysis suggested that the same could be true of planarian homologs (Table 1). To verify that planarian PIWI proteins are PRMT5 substrates, we analyzed the expression of polypeptides recognized by Y12 after knockdown of three PIWI homologs. Detection of the 110 kDa polypeptide by Y12 was lost after Smedwi-3 RNAi (Fig. 6A), and was often reduced (but not eliminated entirely) in extracts of Smedwi-1(RNAi) animals (Fig. 6A). The designation of this polypeptide as sDMA-SMEDWI-3 is corroborated by the predicted molecular weight of SMEDWI-3 and the presence of multiple PRMT5 substrate motifs in the primary sequence of SMEDWI-3 (but not SMEDWI-1 or SMEDWI-2; Table 1). However, indirect effects of Smedwi-3 RNAi on some other sDMA-modified protein are still possible.

Recognition of sDMA-SMEDWI-3 by Y12 suggests that CBs might be associated with piRNAs. Indeed, RNAs within the size range of piRNA (26-32 nts) (Palakodetti et al., 2008; Friedlander et al., 2009) were enriched in material from planarian extracts specifically co-immunoprecipitated with Y12, but not with control antibodies (Fig. 6B, single asterisk). RNA species of higher molecular weight were also detected in Y12 immunoprecipitates, as expected from this antibody’s likely affinity for methylated Sm proteins (Fig. 6B, double asterisks). These results strongly suggest that sDMA-SMEDWI-3 is one of the CB factors recognized by Y12 in planarian neoblasts, and that methylated planarian PIWI proteins are capable of direct or indirect association with PIWI-interacting RNAs (piRNAs).

The identification of sDMA-SMEDWI-3 raised the possibility that PRMT5 might influence repression of transposon activity by SMEDWI-3. Indeed, Illumina-based sequencing analyses revealed that the most over-represented transcripts in PRMT5(RNAi), compared with control planarians, corresponded to transposon and repetitive-element sequences (supplementary material Tables S2 and S3). This result suggests that RNAs localized to CBs might be substrates or components of the piRNA pathway, rather than representatives of the translatable pool of neoblast mRNAs. To test whether localization of RNAs to CBs was a general phenomenon, we analyzed the subcellular localization of a number of neoblast-enriched transcripts by ISH followed by Y12 immunofluorescence. Surprisingly, only gH4 RNA seemed to localize to CBs (Fig. 6C). Transcripts that encode proteins known to be required for neoblast function, such as SMEDWI-2 and SMEDWI-3 (Palakodeti et al., 2008; Reddien et al., 2005b), were not detected in CBs (Fig. 6C). Similarly, transcripts of Smed-nanos, function of which is required for germ cell maintenance, did not colocalize with Y12 in presumptive germ cells of asexual animals (Fig. 6C). These results suggest that mRNAs actively translated in neoblasts and presumptive germ cells (in the case of nanos) are not retained in CBs.

What dictates the localization of gH4 transcripts to CBs remains unclear. Processing or translational regulation of gH4 mRNA could occur in CBs. Alternatively, a subset of gH4 RNA could be a substrate for piRNA machinery localized to CBs. In support of this latter idea, the 13.7 kbp genomic region containing the gH4 locus harbors several piRNA islands, a recent inverted duplication of gH4 with >97% nucleotide sequence identity, and a Ty3/gypsy-type transposon (supplementary material Fig. S5A-B). Transcription from regions neighboring gH4 sequences could result in imperfect but detectable substrates of the piRNA machinery. Transposon-derived transcripts were not localized to CBs and were barely detected in neoblasts (supplementary material Fig. S6A). However, transcripts of one transposon (PL06017B1E01) were readily detected in a number of differentiated cell types (supplementary material Fig. S6B), raising the possibility that transposons are not transcribed and/or are effectively degraded in neoblasts. The identification of gH4 RNA as a CB component warrants future experiments addressing the nature, metabolism and regulation of this RNA to understand better the events that occur in CBs.

In this article, we demonstrate a role for the conserved methyltransferase PRMT5 in promoting stem cell function. We show that PRMT5 methylates CB components in planarian neoblasts. PRMT5 depletion leads to loss of protein sDMA methylation in neoblasts, reduced CB number and defects in regeneration, growth and homeostasis. Neoblast loss was observed in severely affected individuals after PRMT5 RNAi. However, complete loss of neoblasts does not seem to be required for manifestation of PRMT5 deficiency, as homeostatic defects in both size and blistering were seen in animals without a complete loss of neoblasts. Expression of PIWI homologs is characteristic of planarian neoblasts (Palakodeti et al., 2008; Reddien et al., 2005b; Rossi et al., 2006) and, similar to PRMT5(RNAi), defects in regeneration and homeostasis of PIWI knockdowns are not based upon complete loss of neoblasts (Reddien et al., 2005b), although decreased neoblast numbers are observed in Smedwi-2, Smedwi-3 and Smedwi-2/-3 RNAi-treated animals (Palakodeti et al., 2008). One possible explanation for this observation is that neoblasts in PIWI knockdown animals have decreased piRNA function and increased transposable element activity. Although neoblasts might be lost owing to disruptive activity of transposable elements in genes required for neoblast maintenance, transposition could also lead to mutation of genes involved in differentiation, repression of neoblast proliferation, or some other neoblast function. The increase in transposon and repetitive element transcripts in PRMT5(RNAi) and the detection of sDMA on SMEDWI-3, links PRMT5 and CBs to the piRNA pathway. Nevertheless, PRMT5 might also influence piRNA-independent processes in neoblasts.

Several ribonucleoprotein granules have been studied intensively in the last decade, and the mechanism by which they maintain structural stability continues to be the focus of much research. Interactions within prion-like domains of some RNP granule components might be important for formation of P-bodies and neuronal RNP granules (Decker et al., 2007; Mazzoni et al., 2007; Reijns et al., 2008; Salazar et al., 2010). Mechanisms involved in RNP granule assembly in the nematode germline include hydrophobic interactions between the phenylalanine-glycine repeat domains of VASA-related proteins GLH-1/2/4 (Updike et al., 2011), and self-interaction domains of P-granule abnormality family member (PGL) proteins (Hanazawa et al., 2011). A conserved interaction between TDRDs and sDMA on RNA-binding proteins appears to be responsible for the formation of germ granules in mice and flies (reviewed by Arkov and Ramos, 2010; Siomi et al., 2010). Our data suggest that planarian CBs are stabilized by this same mechanism and that PRMT5 function in RNP-granule formation is also conserved. The identification of SpolTud-1, a planarian gene encoding a protein that contains three Tudor domains and localizes to CBs, supports this idea (Solana et al., 2009).

Like PRMT5(RNAi), Spoltud-1(RNAi) animals display regeneration defects and a reduction in neoblast number; however, these phenotypes are only observed several weeks after protein depletion (three weeks for protein depletion versus seven for phenotypic manifestation) (Solana et al., 2009). A similar long-term delay between loss of gene product and phenotypic manifestation occurs after PRMT5 RNAi. Loss of methylation indicative of PRMT5 depletion is observed within three weeks of dsRNA treatment (Fig. 2A); however, most homeostatic defects were not observed until two months after the start of RNAi treatment (Fig. 3). The phenotypic similarities between PRMT5 and Spoltud-1 knockdowns suggest that both factors could be working together to promote CB function and neoblast stability. How might these factors act together? One model proposed from work in mice and flies posits that interactions between PIWI proteins create an amplification loop of piRNA production (reviewed by Siomi et al., 2011). PIWI protein aggregates are stabilized by bridging interactions between the Tudor domains and sDMAs of different PIWI proteins (reviewed by Vagin et al., 2009a). Given that many TDRDs contain two or more Tudor domains, their interaction with multiple sDMA-containing proteins would lead to macromolecular aggregation and compartmentalization of piRNA processing (Fig. 6D) (Aravin et al., 2009; reviewed by Arkov and Ramos, 2010; Chuma and Pillai, 2009).

The effect of methylation on protein-protein interactions in RNPs came from studies of Sm proteins and the survival of motor neuron protein, a TDRD assembly factor for snRNPs and other ribomacromolecules (Brahms et al., 2001; Friesen et al., 2001). Tudor domains mediate binding to sDMA-modified RG repeats of SmB/B′, SmD1, SmD3 (Brahms et al., 2001; Cote and Richard, 2005; Selenko et al., 2001; Sprangers et al., 2003) and PIWI proteins (Kirino et al., 2010b; Liu, H. et al., 2010; Nishida et al., 2009; Vagin et al., 2009b). Functional analyses have shown that the PRMT5/PIWI/TUDOR network plays a role in the piRNA processing pathway, retrotransposon silencing and, consequently, genomic stability (Nishida et al., 2009; Reuter et al., 2009; Shoji et al., 2009; reviewed by Siomi et al., 2011). By contrast, the PRMT5/Sm/TUDOR network present in fly oocytes is involved in localization of oskar mRNA and specification of germ cell fate (Anne et al., 2007; Anne, 2010; Boswell and Mahowald, 1985; Gonsalvez et al., 2006; Gonsalvez et al., 2010). Our biochemical and sequence analyses suggest that both SMEDWI-3 and Sm protein family members are methylated in planarians. Recent work has shown that both SmB and PIWI are required for neoblast function (Fernandez-Taboada et al., 2010; Palakodeti et al., 2008; Reddien et al., 2005a). Furthermore, SmB was shown to localize to planarian CBs (Fernandez-Taboada et al., 2010) and our results suggest that the same is true for sDMA-SMEDWI-3. Thus, planarian CBs might be sites of piRNA processing, as well as other RNA-processing events. However, further characterization of gH4 RNA and identification of additional CB RNA components are necessary to fully elucidate the post-transcriptional processes that take place in CBs and their influence on planarian stem cell stability and function.

Given their transcriptionally quiescent state, germ cells are hubs of post-transcriptional regulation. The establishment, proliferation, differentiation and function of the germline are centered around RNA-binding proteins and RNA-modifying enzymes that regulate the timing, location and translational activity of specific transcripts (reviewed by Seydoux and Strome, 1999; Seydoux and Braun, 2006; Thompson et al., 2007; Voronina et al., 2011). nanos function is required for maintenance of the germline in asexual and sexual strains of S. mediterranea (Wang et al., 2007). To our knowledge, gH4 RNA is the first transcript reported to be abundant in CBs of asexual planarian neoblasts. We failed to see accumulation of nanos transcripts in CBs of asexual S. mediterranea presumptive germ cells (Fig. 6C). nanos transcripts accumulate in spermatogonial CBs of Dugesia japonica, but their localization to CBs in sexually immature animals is less clear (Sato et al., 2006). Interestingly, gH4 colocalization with Y12 was observed only infrequently in testes lobes (supplementary material Fig. S7). Thus, it is likely that regulation of piRNA and mRNA by CBs depends on sexual specification.

Recent studies in non-traditional model organisms have shown that a number of RNA-binding proteins often considered to be restricted to germline functions, such as Vasa, Nanos and PIWI, operate in multipotent progenitors and stem cells of many metazoans (reviewed by Juliano and Wessel, 2010; Juliano et al., 2010). These studies led to the suggestion of a ‘germline multipotency program’ present in multipotent stem cells that generate somatic cell types. Indeed, PIWI homologs are expressed in stem cells of sponge (Funayama, 2010; Funayama et al., 2010), acoels (De Mulder et al., 2009), sea urchin (Juliano et al., 2006), jellyfish (Seipel et al., 2004), annelids (Rebscher et al., 2007) and ctenophores (Alie et al., 2011). Methylation of RNP components by PRMT5 may also be considered a constituent of such a ‘germline multipotency program’.

We thank John Brubacher, Jim Collins, Ryan King and James Sikes for helpful comments on the manuscript; Alvaro Hernandez for guidance with next generation sequencing; Yuying Wang, Frank Echtenkamp, Brian Freeman and members of the Siomi and Agata laboratories for helpful discussions; Stoyan Toshkov for help with gamma-irradiation; and Drs Joan Steitz, Gideon Dreyfuss, Hidefumi Orii and Kenji Watanabe for antibodies.