Transcriptionally repressed germ cells lack a subpopulation of phosphorylated RNA polymerase II in early embryos of Caenorhabditis elegans and Drosophila melanogaster

The separation of germ line and soma is an essential process in the development of multicellular organisms, yet the molecular properties that initially differentiate between these two cell types during embryogenesis are still poorly understood. One characteristic that distinguishes germ cells from somatic cells in early C. elegans and Drosophila embryos is the ability to produce new messenger RNAs. In both organisms, several studies have suggested that somatic blastomeres activate mRNA production earlier than do germ-line blastomeres. This difference was first discovered in Drosophila embryos where somatic nuclei, but not germ-line nuclei, were shown to incorporate [³H]UTP and to hybridize with [³H]poly-U in the blastoderm stage (Zalokar, 1976; Lamb and Laird, 1976; Kobayashi et al., 1988). More recent experiments have demonstrated that the introduction of potent transcription factors in germ-line nuclei is not sufficient to activate mRNA production in these cells prior to the onset of gastrulation (A. Williamson, M. Van Doren and R. Lehmann, personal communication). These observations have led to the hypothesis that early germ cells may be refractory to transcriptional activation (Williamson and Lehmann, 1996).

A similar conclusion was recently reached in C. elegans (Seydoux et al., 1996). In this organism, several newly transcribed mRNAs have been detected in somatic blastomeres as early as the 4-cell stage, but none to date have been detected in germ-line blastomeres (Seydoux and Fire, 1994; Seydoux et al., 1996). This soma-germ line difference was shown to depend on the germ-line-specific factor PIE-1. In the absence of PIE-1, newly transcribed mRNAs are detected in both somatic and germ-line blastomeres (Seydoux et al., 1996) and descendants of germ-line blastomeres eventually differentiate inappropriately into somatic tissues (Mello et al., 1992). These data have suggested that germ cell fate in C. elegans depends on a PIE-1-mediated inhibitory mechanism that generally blocks mRNA production in the early embryonic germ lineage. We have hypothesized that this inhibitory mechanism may serve to protect germ-line blastomeres from the activity of transcription factors that would otherwise promote somatic development in these cells (Seydoux et al., 1996).

How is mRNA production globally inhibited in germ cells? One possibility is that newly transcribed mRNAs are rapidly degraded in germ cells; another possibility is that mRNA transcription itself is blocked in germ cells. To begin to distinguish between these two possibilities, we have characterized the distribution of different isoforms of the large subunit of RNA polymerase II in C. elegans and Drosophila embryos. We find that germ cells lack a specific subpopulation of phosphorylated RNA polymerase II, consistent with the possibility that mRNA transcription is blocked in these cells. We also show that, unlike mRNA transcription, rRNA transcription is active in early germ cells.

C. elegans and D. melanogaster stocks were maintained at 20°C and 25°C, respectively, using standard methods (Brenner, 1974; Ashburner, 1989). Wild-type stocks were Caenorhabditis elegans N2 and Drosophila melanogaster Oregon R. C. elegans pie-1(−) embryos were generated by self-fertilization of pie-1(zu154) unc-25(e156) hermaphrodites derived from the strain pie-1(zu154) unc-25(e156)/qC1 (Mello et al., 1992). D. melanogaster nos( −) embryos were obtained from nos^bn/nos^bn females (Wang et al., 1994) crossed with Canton S males.

Antisense ama-1 or pes-10 (control) RNA was synthesized from plasmids pDB19 and pGS17.09, respectively, and injected into the gonad of wild-type adult hermaphrodites as described in Guo and Kemphues (1995) and Powell-Coffman et al. (1996). 18 to 24 hours after injections, injected hermaphrodites were squashed under coverslips to release embryos and processed for antibody staining.

C. elegans embryos obtained by Clorox treatment or squashing of gravid hermaphrodites were freeze-cracked in PBS between slide and coverslip, immersed in −20°C MeOH for 10 seconds to 1 minute, fixed in 1× PBS, 0.08 M Hepes (pH 6.9), 1.6 mM MgSO₄, 0.8 mM EGTA, 3.7% formaldehyde for 30 minutes, washed 3 times in PBT (1× PBS, 0.1% Triton, 0.1% BSA) and blocked for 30 minutes in PBT. 1° antibodies were applied overnight at 4°C and 2° antibodies for 2 hours at room temperature. In a control experiment presented in Fig. 3, fixed embryos were incubated with alkaline phosphatase (0.125 units/μl, Boehringer Mannheim) for 2 hours at 37°C, and washed several times in PBT, prior to incubation with the primary antibody.

Drosophila embryos collected on molasses or apple juice plates were dechorionated, fixed and permeabilized for 30 minutes in 50% heptane, 3.7% formaldehyde, 1× PBS, devitellinized in MeOH and blocked in PBT as described in Reuter et al. (1990). 1° and 2° antibodies were applied overnight at 4°C.

Primary antibodies used were: mAbs H5 and H14, two mouse IgM monoclonal antibodies that bind to distinct phosphorylated epitopes on the carboxy-terminal domain (CTD) of the large subunit of RNA polymerase II (Warren et al., 1992; Bregman et al., 1995; Kim et al., 1997; M. Patturajan and J. Corden, personal communication); P4G5, a mouse monoclonal IgG antibody raised against a PIE-1 peptide and received as a gift from C. Schubert, C. Mello and J. Priess; OIC1D4, a mouse monoclonal IgG antibody that recognizes P granules (Strome, 1986); a rabbit polyclonal serum raised against C. elegans RNAP II (Sanford et al., 1985); a goat polyclonal serum raised against Drosophila RNAP II (Weeks et al., 1982) and rabbit polyclonal serum raised against Drosophila VASA received as a gift from A. Williamson.

Secondary antibodies used were: FITC-conjugated goat anti-mouse IgM, rhodamine-conjugated goat anti-mouse IgG, FITC-conjugated goat anti-rabbit IgG, FITC-conjugated rabbit anti-goat IgG (Jackson Immuno Research) and Cy3-conjugated goat anti-rabbit IgG (Amersham Life Science). DAPI (1 μg/ml) was added to secondary antibody dilutions to visualize nuclei.

In each experiment, a minimum of 10 embryos was analyzed for each developmental stage. Photographs were taken using Ektachrome 400ASA color slide film (Kodak) for immunofluorescence, or Ektachrome 160T color slide film (Kodak) for Nomarski. Images were digitized using a Vision 35 scanner (Agfa) and the digitized images were assembled using Adobe Photoshop 3.0 and Page Maker 6.0 programs (Adobe).

Drosophila ITS1 sequence (Tautz et al., 1988) was PCR amplified from Drosophila genomic DNA and cloned into pBluescript KS+ to generate digoxigenin-labeled, single-stranded DNA sense and antisense probes as described in Seydoux and Fire (1995). In situ hybridization to Drosophila embryos was carried out as described in Tautz and Pfeifle (1989), except that formaldehyde was used instead of paraformaldehyde and levamisole was omitted from the staining reaction.

C. elegans ITS1 and ITS2 sequences were PCR amplified from ribosomal DNA clone pCe7 (Emmons et al., 1979; Ellis et al., 1986) and cloned into pBluescript KS+ to generate digoxigenin-labeled, single-stranded DNA sense and antisense probes used for in situ hybridization as described in Seydoux and Fire (1995). Both ITS1 and ITS2 antisense probes hybridized to 1 or 2 foci per interphase nucleus, consistent with the fact that all rDNA genes have been mapped to a single region on LGI (Albertson, 1984). No staining was detected using ITS1 and ITS2 sense probes.

The 7 kb BamHI rDNA insert from pCe7 (Emmons et al., 1979) was cloned into pBluescript KS+ to generate a rDNA repeat where the 18 S gene is interrupted by full-length Bluescript sequence. This construct (100 μg/ml) was injected with pRF4 (10 μg/ml; Mello et al., 1991) to generate an heritable extrachromosomal array. Roller males carrying the array were mated with tra-2(q122) females and their progeny processed for in situ hybridization using a pBluescriptspecific probe. We observed variable expression of the transgene in embryos, with a fraction of embryos showing expression in all blastomeres, including the germ-line blastomere. Because the transgene was on an extrachromosomal array with pRF4 DNA (encoding a RNAP II transcript), it may have had difficulty localizing to the nucleolus for efficient RNAP I-dependent expression.

In C. elegans and Drosophila, germ cells are formed during early embryogenesis (Fig. 1). To visualize the distribution of RNA polymerase II (RNAP II) in early C. elegans and Drosophila embryos, we used two polyclonal antibodies raised against RNAP II purified from C. elegans and Drosophila extracts, respectively (Sanford et al., 1985; Weeks et al., 1982). These antibodies recognize several subunits of RNAP II, and do not distinguish between the phosphorylated and non-phosphorylated forms of the large subunit (Sanford et al., 1985; Weeks et al., 1982). We found that these antibodies label the nucleoplasm of both somatic and germ-line interphase nuclei in C. elegans and Drosophila embryos (Fig. 2). In many nuclei, these antibodies also highlighted two subnuclear foci, which were often barely visible above the strong nucleoplasmic staining (data not shown). These results indicate that RNAP II is present in both somatic and germ-line nuclei in early embryos of C. elegans and Drosophila.

Phosphorylation of the carboxy-terminal domain (CTD) of the large subunit of RNAP II (RNAP II LS) has been linked with the process of transcriptional elongation (Dahmus, 1996 for review). To visualize phosphorylated RNAP II LS, we have used two monoclonal IgM antibodies, mAbs H5 and H14. By western analysis, these antibodies have been shown to bind to phosphorylated RNAP II LS in a wide range of eukaryotes, including C. elegans and Drosophila (Warren et al., 1992; Bregman et al., 1995; M. A. D. and G. S., unpublished data). Recently, these antibodies were shown to recognize distinct phosphorylated epitopes on the CTD of RNAP II LS (Kim et al., 1997; M. Patturajan and J. Corden, personal communication; also see Discussion).

To verify the specificity of mAbs H5 and H14 binding to RNAP II LS in fixed tissues, we have compared mAbs H5 and H14 in situ staining patterns in wild-type embryos and embryos lacking RNAP II LS. C. elegans embryos lacking RNAP II LS [ama-1(as) embryos] were generated by injecting the gonad of adult hermaphrodites with antisense ama-1 RNA, which encodes RNAP II LS (Bird and Riddle, 1989). This treatment causes hermaphrodites to produce embryos lacking RNAP II LS, which fail to activate mRNA transcription and arrest around the 100-cell stage (Powell-Coffman et al., 1996). Whereas high levels of mAbs H5 and H14 staining were detected in embryos derived from hermaphrodites injected with a control antisense RNA (Fig. 3A,C), no mAb H5 or mAb H14 staining was detected in interphase nuclei of ama-1(as) embryos (Fig. 3B,D). These results confirm that mAbs H5 and H14 can recognize RNAP II LS specifically in fixed interphase nuclei of C. elegans. Occasionally, some mAbs H5 and H14 immunoreactivity was detected on condensed mitotic chromosomes of ama-1(as) embryos (Fig. 3B), raising the possibility that this aspect of mAbs H5 and H14 staining may not be specific to RNAP II LS. For this reason, mitotic nuclei were not considered further in our analysis of mAbs H5 and H14 staining.

To verify that mAbs H5 and H14 recognize phosphorylated epitopes, we treated fixed C. elegans embryos with alkaline phosphatase prior to immunostaining (see Materials and methods). We found that this treatment completely abolishes staining with mAbs H5 and H14 (Fig. 3F,G), but does not affect staining with a polyclonal antibody that recognizes both phosphorylated and non-phosphorylated forms of RNAP II (Fig. 3E). We conclude that mAbs H5 and H14 bind specifically to phosphorylated RNAP II LS in fixed C. elegans embryos.

To determine the pattern of mAb H5 immunoreactivity in C. elegans embryos, we co-immunostained embryos with mAb H5, and with an antibody specific for the germ-line factor PIE-1 (Fig. 4). As reported previously (Mello et al., 1996), PIE-1 was detected in germ-line blastomeres from the 1-cell stage to the 100-cell stage (Fig. 4, second column). In contrast, mAb H5 staining was restricted to somatic blastomeres (Fig. 4, third column). We first detected mAb H5 staining in the interphase nucleus of the somatic blastomere EMS in the 4-cell stage (Fig. 4I, arrowhead). In this stage, somatic blastomeres are known to initiate mRNA transcription (Seydoux and Fire, 1994, Seydoux et al., 1996; the somatic blastomeres ABa and ABp are transcriptionally active for a very short time in the early part of the 4-cell stage before they resume mitosis, which may explain our inability thus far to detect mAb H5 staining in these cells). From the 8-cell stage to the 100-cell stage, mAb H5 immunoreactivity was detected in the nuclei of all somatic blastomeres in interphase; in contrast, no staining was detected in germ-line blastomeres (Fig. 4L,O). Occasionally, two faint foci of staining were detected in the germ-line blastomere P₄ in 50-cell and older embryos, but this aspect of staining was not always reproducible (data not shown). In 100-cell and older embryos, mAb H5 staining began to be detected in the nucleoplasm of the 2 daughters of P₄, the primordial germ cells Z2 and Z3 (Fig. 4R). By this stage, PIE-1 staining was no longer detected in the germ lineage (Fig. 4Q; the primordial germ cells were identified here using another germ cell marker, the P granule antibody OIC1D4).

To determine the pattern of mAb H5 immunoreactivity in Drosophila embryos, we co-immunostained embryos with mAb H5, and with an antibody against the germ cell protein VASA (Lasko and Ashburner, 1990; Hay et al., 1990; A. Williamson and R. Lehmann, personal communication) (Fig. 5A-C). mAb H5 staining was first detected weakly in somatic nuclei of stage 3 embryos and with increased intensity from stage 4 onward (Fig. 5C). During this time, somatic nuclei acquire general competence for transcription (Edgar and Schubiger, 1986). In contrast, no staining was detected in germ cell (pole cell) nuclei until stage 7, when these cells began to show faint mAb H5 staining (data not shown).

We conclude that the phosphoepitope on RNAP II that is recognized by mAb H5 is present in somatic nuclei, but not in germ-line nuclei, in pregastrulation C. elegans and Drosophila embryos.

We determined the pattern of mAb H14 staining using the same approach used for mAb H5; embryos were co-stained with mAb H14 and with antibodies specific for germ cells (anti-PIE-1 or anti-P granule antibodies for C. elegans and anti-VASA antibody for Drosophila) (Figs 5, 6 and data not shown). In both C. elegans and Drosophila, mAb H14 immunostaining pattern differed in two respects from that of mAb H5. First, unlike mAb H5, which only stained nuclei, mAb H14 also stained the cytoplasm of dividing cells throughout embryogenesis, even prior to the onset of mRNA transcription (Fig. 6B). Second, mAb H14 stained both somatic and germ cell nuclei from early embryogenesis (4-cell stage in C. elegans, stage 3 in Drosophila) onward. In both cell types, mAb H14 labeled the nucleoplasm and two prominent foci per nucleus (Figs 5F, 6D,F,H; similar foci were also visible in early somatic nuclei with mAb H5, data not shown). These foci became more difficult to detect in somatic nuclei of later stage embryos, perhaps due to increased nucleoplasmic staining. In contrast, in germ cell nuclei, nucleoplasmic staining intensity remained low and the two foci of staining could be detected throughout early embryogenesis (4- to 100-cell stage in C. elegans, stage 4 to stage 6 in Drosophila). After the onset of gastrulation, (100-cell stage in C. elegans, stage 7 in Drosophila), staining in germ cells began to resemble that observed in somatic cells (diffuse nucleoplasmic staining; data not shown).

We conclude that the phosphoepitope on RNAP II that is recognized by mAb H14 is present in two subnuclear domains in both somatic and germ-line nuclei in early embryos. This phosphoepitope is also present in the nucleoplasm, but is detected there at a higher level in somatic nuclei compared to germ-line nuclei in pregastrulation embryos.

In C. elegans embryos, the PIE-1 protein is detected in germ-line blastomeres from the 1-cell stage to the 100-cell stage (Mello et al., 1996). As described above, this period corresponds to the interval during which germ cells do not stain with mAb H5 (Fig. 4) and stain with lowered intensity with mAb H14 (Fig. 6). We have previously shown that pie-1 activity is required to keep mRNA production off in germ cells of early embryos (Seydoux et al., 1996). To test whether pie-1 is also required for the germ-line-specific patterns of mAbs H5 and H14 staining, we stained embryos derived from pie-1(−) mothers with mAbs H5 and H14. We find that, in the absence of pie-1 activity, germ cells stain with mAbs H5 and H14 in a pattern identical to that observed in somatic cells (Fig. 7). These results indicate that pie-1 activity is required for the soma-germ line differences observed in mAbs H5 and H14 staining.

The nanos (nos) gene is required for the correct temporal expression pattern of several germ cell-specific transcripts in Drosophila (Kobayashi et al., 1996). In the absence of nos activity, expression of these transcripts, which normally initiates when the germ cells have reached the gonads (stage 14), begins prematurely during germ cell migration (stage 7-8) (Kobayashi et al., 1996). To test whether nos activity is also required for the proper onset of mAb H5 immunoreactivity in germ cells (stage 7), we stained embryos derived from nos(−) mothers with mAb H5. We found that the temporal pattern of mAb H5 staining was unaffected in nos(−) embryos (mAb H5 staining was detected in germ cells of nos(−) embryos no earlier than stage 7, as is observed in wild type; data not shown). These results indicate that nos activity is not required for the proper onset of mAb H5 immunoreactivity in germ cells and suggests that nos regulates the expression of specific germline transcripts after transcription has been generally activated in the germ line.

Our analysis of gene expression in embryonic germ cells has focused so far on RNA polymerase II transcription. To begin to test whether other polymerases are similarly regulated, we have analyzed the transcriptional pattern of ribosomal RNAs in early C. elegans and Drosophila embryos. To distinguish newly transcribed ribosomal RNAs from maternally inherited mature rRNAs, we used probes complementary to internal spacer sequences (ITS1 and ITS2 in C. elegans and ITS1 in Drosophila). These sequences are present in the nascent rRNA transcript, but are eliminated during rRNA processing in the nucleolus, and thus are not expected to be present in maternally inherited, mature rRNAs.

In Drosophila, nascent rRNA transcripts were first detected in stage 4 embryos, consistent with the observations of Edgar and Schubiger (1986), which showed that rRNA transcription begins during this stage. Initially, nascent rRNA transcripts were detected in all somatic nuclei, and in a subset of germline nuclei (Fig. 8A,B). By stage 5, nascent rRNAs were detected consistently in all somatic and germ-line nuclei (Fig. 8C,D).

In C. elegans, nascent rRNA transcripts were first detected in the decondensed maternal and paternal pronuclei of 1-cell stage embryos (Fig. 8F). Subsequently, these transcripts were detected in all nuclei (both somatic and germ line; Fig. 8G,I), with the exception of nuclei undergoing mitosis (Fig. 8H). Nascent rRNA transcripts were observed continuously throughout interphase in all blastomeres, except in the germline blastomere P₄, where they were transiently absent from the 28-cell stage to the 50-cell stage (data not shown). During this time, the chromosomes of P₄ adopt a unique morphology, which is retained in P₄ and its descendants throughout embryogenesis (G. S., unpublished observations); the significance of this morphology and of the apparent transient lack of rRNA transcription in P₄ is not known.

To rule out the possibility that the unprocessed transcripts that we observed were actually maternally derived, we used an additional strategy to visualize patterns of rRNA transcription in C. elegans embryos. We inserted pBluescript sequences in the 18S gene of a ribosomal DNA repeat (see Materials and Methods) and transformed this tagged rDNA construct into worms to establish a heritable line. Males carrying the tagged rDNA construct were mated to untransformed females and the resulting cross-progeny were analyzed by in situ hybridization for expression of the tagged 18S rRNA. We detected tagged 18S rRNA in all nuclei as early as the 2-cell stage (with the exception of P₄ in 28-cell to 50-cell embryos as described above) (Fig. 8J). Because the tagged transgene was not present in the maternal germ line, expression of the tagged rRNA must have derived from the paternal copy of the transgene and therefore is most likely to be due to embryonic transcription.

Taken together, these results indicate that rRNAs, unlike mRNAs, are transcribed in both somatic and germ-line cells in early C. elegans and Drosophila embryos.

In this study, we describe the distribution of different RNA polymerase II isoforms in C. elegans and Drosophila embryos. These observations have led to three major conclusions. First, germ cells lack a specific RNAP II phosphoepitope (RNAP II-H5), which is present in transcriptionally active somatic cells. Second, a distinct RNAP II phosphoepitope (RNAP II-H14) is present in both somatic and germ cells. Third, in C. elegans, the germline-specific pattern of RNA polymerase II phosphorylation is dependent on pie-1 activity. We also describe the distribution of newly transcribed rRNAs in early C. elegans and Drosophila embryos and show that, unlike mRNA transcription, rRNA transcription is active in both somatic and germ-line blastomeres. We discuss the implications of each of these findings below.

To test whether the lack of newly synthesized mRNAs in embryonic germ cells is due to a block in RNAP II activity, we have characterized the distribution of phosphorylated RNAP II in early C. elegans and Drosophila embryos. Towards this purpose, we used two monoclonal antibodies, mAbs H5 and H14, which recognize distinct phosphorylated epitopes on the carboxy-terminal domain (CTD) of the large subunit of RNAP II (Bregman et al., 1995; Kim et al., 1997; M. Patturajan and J. Corden, pers. comm.). Phosphorylation of the CTD has been linked temporally with transcriptional elongation (Dahmus, 1996 for review); indeed, we find that mAbs H5 and H14 begin to label somatic nuclei at stages coincident with the onset of embryonic transcription (4-cell stage in C. elegans, stage 3/4 in Drosophila). In early somatic nuclei, mAbs H5 and H14 stained the nucleoplasm and two discrete subnuclear domains. This pattern is reminiscent of the staining pattern of mAbs H5 and H14 in mammalian cells, where these antibodies were shown to label the nucleoplasm and several discrete subnuclear sites closely linked to ‘speckle’ domains (Bregman et al., 1995). These observations have suggested that phosphorylated RNAP II may exist in two forms in vivo: a storage form sequestered in discrete subnuclear domains and an active form engaged in transcription in the nucleoplasm (Bregman et al., 1995; Kim et al., 1997).

Remarkably, we detected different mAbs H5 and H14 staining patterns in early embryonic germ cells. mAb H5 failed to stain germ cells entirely in both C. elegans and Drosophila pregastrulation embryos. With mAb H14, some staining could be detected in early C. elegans and Drosophila germ cells, but this staining was concentrated in two subnuclear foci and was often barely detectable in the nucleoplasm. These observations suggest that early embryonic germ cells lack the phosphoepitope recognized by mAb H5 and have reduced nucleoplasmic levels of the phosphoepitope recognized by mAb H14. This germ-line-specific staining pattern persisted until early gastrulation (100-cell stage in C. elegans, stage 7 in Drosophila); at that time, the germ cells associate with the gut primordium inside the embryo and begin to stain with mAbs H5 and H14 in patterns similar to those observed in somatic nuclei. This stage is likely to correspond to the onset of mRNA transcription in germ cells; indeed, embryonically transcribed vasa transcripts have been detected in Drosophila germ cells as early as stage 9-10 (Williamson and Lehmann, 1996).

Our data suggest that CTD phosphorylation is transiently blocked (or reversed by dephosphorylation) in early C. elegans and Drosophila germ cells. Since CTD phosphorylation has been linked to the transcriptional cycle, these results support the hypothesis that RNAP II activity is transiently inhibited in C. elegans and Drosophila embryonic germ cells. Interestingly, this inhibition appears to eliminate completely mAb H5- specific CTD phosphorylation, but only reduces H14-specific CTD phosphorylation in the nucleoplasm. mAbs H5 and H14 recognize different phosphorylated serines on the CTD (serine at position 2 of each CTD repeat for mAb H5, and serine at position 5 of each CTD repeat for mAb H14; M. Patturajan and J. Corden, pers. comm.). Suppression analysis in yeast have shown that these two serines are genetically distinct, raising the possibility that phosphorylation at these two sites may have different functional consequences (Yuryev and Corden, 1996). Indeed, whereas mAb H5-specific CTD phosphorylation was detected only in transcriptionally active nuclei, mAb H14- specific CTD phosphorylation was also detected in the cytoplasm of mitotic cells, indicating that this latter pattern of phosphorylation is not necessarily linked to transcription. We do not know, however, whether residual RNAP II-H14 present in germ cell nuclei is inactive and represents only a storage form of the polymerase, or is active and is transcribing those few (so far undescribed) germ-line mRNAs that may escape transcriptional repression. In addition, our observations do not address whether reduction in CTD phosphorylation is a cause or a consequence of the lack of RNAP II activity in germ cells.

In the discussion above, we have assumed that the absence of mAb H5 immunoreactivity in germ cells reflects a lack of H5-specific CTD phosphorylation in these cells. An alternative interpretation of our results, however, is that the mAb H5 epitope is present in germ cells, but is not accessible to the antibody (i.e. the mAb H5 epitope is ‘masked’). In this hypothesis, H5-specific CTD phosphorylation would still occur in germ cells, but in a context no longer recognizable by mAb H5. Although we know of no precedent for masking of the CTD, our observations so far have not ruled out this possibility formally. However, because this hypothesis implies that RNAP II-H5 exists in a unique context in early germ cells, this alternative interpretation is still consistent with our main conclusion, namely that phosphorylated RNAP II exists in a different form in germ cells compared to somatic cells.

The observation that mRNA production is repressed in germ cells led us to investigate whether rRNA production is similarly regulated. Surprisingly, we found that rRNA transcription occurs in both somatic and germ-line cells in early C. elegans and Drosophila embryos. In Drosophila, nascent rRNAs were detected in both somatic and germ-line nuclei starting in stage 4, although consistent expression in all germ-line nuclei was not detected until stage 5. In C. elegans, rRNA transcription was detected as early as the 1-cell stage, and continued in all somatic and germ-line interphase nuclei throughout early embryogenesis. The only exception was the germline blastomere P₄ where nascent rRNA transcripts were transiently absent from the 28-cell stage to the 50-cell stage. These observations indicate that, unlike mRNA transcription, rRNA transcription is not subject to continuous repression in the early germ lineage. We do not know whether differences between mRNA and rRNA transcription are due to the nucleolar localization of rRNA synthesis, or to the fact that rRNAs and mRNAs are transcribed by different polymerases, or both. In any case, these observations indicate that early germ cells in C. elegans and Drosophila are not completely transcriptionally repressed, and that at least a portion of their genome is accessible to transcription factors.

We have previously shown that the germ-line factor PIE-1 is required to repress mRNA production in the early embryonic germ lineage of C. elegans; in pie-1(−) embryos, germ-line blastomeres activate mRNA production (Seydoux et al., 1996). In this study, we show that pie-1 is also required for the germ-line-specific pattern of RNAP II phosphorylation; in pie-1(−) embryos, germline blastomeres stain with mAbs H5 and H14 in patterns identical to those observed in somatic blastomeres. This result is particularly significant since pie-1 activity is not required for other aspects of early germ-line blastomere identity, such as maintenance of maternal RNAs or segregation of P granule-associated RNAs (Seydoux et al., 1996). The fact that pie-1 activity is required specifically for both repression of mRNA production and repression of RNAP II phosphorylation suggests that these two phenomena are linked, and supports the hypothesis that the lack of newly transcribed mRNAs in germ cells is due to a block in RNAP II-dependent transcription.

How might PIE-1 function to repress RNAP II activity in the germ lineage? The PIE-1 protein is translated from maternally derived RNA starting in the 1-cell stage, and accumulates in germ-line nuclei from the 2-cell stage to the 100-cell stage (Mello et al., 1996). Strikingly, PIE-1’s disappearance from the germ lineage in the 100-cell stage coincides with the appearance of mAb H5 immunoreactivity in that lineage. Thus, both the spatial and temporal distributions of PIE-1 are consistent with the possibility that PIE-1 itself functions directly to block RNAP II activity in the germ lineage. One possibility is that PIE-1 interacts with the transcriptional machinery to prevent transcriptional activation and RNAP II phosphorylation. Equally plausible, however, are more complex models where PIE-1 effects on transcription and RNAP II phosphorylation are indirect. For example, PIE-1 could function to modify chromatin structure in germ cells in such a way as to prevent access of the transcriptional machinery to DNA. In this model, absence of RNAP II phosphorylation in germ cells would be a secondary consequence of a lack of template available for transcription. Such indirect models, however, must account for the fact that, unlike mRNA transcription, rRNA transcription is not repressed in early germ cells, indicating that rRNA genes are accessible to RNA polymerase I.

So far, the structure of the PIE-1 protein has not helped distinguish among different models for PIE-1 action. PIE-1 contains 2 zinc-finger-like motifs of the CCCH class (Mello at al., 1996). These motifs have been found in a number of proteins including the mammalian growth-factor response protein TIS11/Nup475 (Dubois et al., 1990, Varnum et al., 1991), the splicing factor U2AF³⁵ (Zhang et al., 1992), and the Drosophila proteins Suppressor of Sable (Voelker et al., 1991), Unkempt (Mohler et al., 1992) and Clipper (Bai and Tolias, 1996). Recently, a region in Clipper that contains 5 CCCH fingers was shown to have RNA endonuclease activity in vitro (Bai and Tolias, 1996). It remains to be determined whether the PIE-1 fingers, which have diverged significantly from those of Clipper, will have the same function.

Perhaps the most significant finding in this study is the observation that embryonic germ cells of both C. elegans and Drosophila show an identical lack of a specific phosphoepitope of RNAP II. Since C. elegans, a nematode, and Drosophila, an insect, are separated by more than 1,000 million years of evolution (Vanfleteren et al., 1994), it is tempting to speculate that this similarity reflects the existence of a very ancient mechanism that inhibits RNAP II function in germ cells and thus insulates germ line from soma during early embryogenesis. We have shown that, in C. elegans, this mechanism depends on PIE-1, a germ-line factor essential for maintenance of the germ-line fate. Future studies will determine whether a PIE-1 homologue performs a similar function in Drosophila, and whether similar mechanisms are operating in other metazoans.

We are especially grateful to Jeff Corden and Andy Fire for many stimulating discussions and insightful suggestions during the course of this work. We thank Deborah Andrew and the Andrew laboratory for generously sharing their Drosophila expertise and protocols; Steve Warren; Miriam Golomb; Arno Greenleaf; Charlotte Schubert, Craig Mello and Jim Priess; Meera Patturajan and Jeff Corden; and Anne Williamson and Ruth Lehmann for antibodies; Jo-Anne Powell-Coffman and Bill Wood for the DB19 plasmid and antisense injection protocol; Barbara Sollner-Webb for advice on rDNA tagging; Liz Gavis for nos^bn flies; Randy Reed for the use of his scanner; Meera Patturajan and Jeff Corden, Anne Williamson and Ruth Lehmann for sharing data prior to publication; and Andy Fire, Bill Kelly, Jeff Corden, Dan Nathans and Deborah Andrew for critical reading of the manuscript. This work was supported by the March of Dimes and the American Cancer Society. G. S. is a Packard fellow.