Trans-generational epigenetic regulation of C. elegans primordial germ cells
© Furuhashi et al; licensee BioMed Central Ltd. 2010
Received: 11 May 2010
Accepted: 12 August 2010
Published: 12 August 2010
Skip to main content
© Furuhashi et al; licensee BioMed Central Ltd. 2010
Received: 11 May 2010
Accepted: 12 August 2010
Published: 12 August 2010
The processes through which the germline maintains its continuity across generations has long been the focus of biological research. Recent studies have suggested that germline continuity can involve epigenetic regulation, including regulation of histone modifications. However, it is not clear how histone modifications generated in one generation can influence the transcription program and development of germ cells of the next.
We show that the histone H3K36 methyltransferase maternal effect sterile (MES)-4 is an epigenetic modifier that prevents aberrant transcription activity in Caenorhabditis elegans primordial germ cells (PGCs). In mes-4 mutant PGCs, RNA Pol II activation is abnormally regulated and the PGCs degenerate. Genetic and genomewide analyses of MES-4-mediated H3K36 methylation suggest that MES-4 activity can operate independently of ongoing transcription, and may be predominantly responsible for maintenance methylation of H3K36 in germline-expressed loci.
Our data suggest a model in which MES-4 helps to maintain an 'epigenetic memory' of transcription that occurred in germ cells of previous generations, and that MES-4 and its epigenetic product are essential for normal germ cell development.
Chromatin structure is an important determinant of transcriptional activity, and is thought to influence accessibility of the transcriptional machinery to the DNA and/or modulate its productivity, as a component of regulation. The structure of chromatin and its influence on genetic regulation can be heritable, and this heritability forms the basis of epigenetic forms of genome regulation. As the eukaryotic genome is passed between generations, there occurs significant remodeling or re-programming of the gamete epigenomes as they merge in the zygote. An additional round of epigenetic reprogramming also occurs upon establishment of the embryonic germline in many species . The purpose of these events are not clear, but they are thought to be important for resetting an epigenetic 'ground state' that is compatible with developmental pluripotency in the zygote, and with maintaining or establishing totipotency in the germline. Although much of the research focus has been on epigenetic erasure events that occur in the zygote, it is important to note that significant epigenetic information is probably retained and/or re-established in the zygote and primordial germ cells (PGCs). How any epigenetic information is selected for erasure, retention or establishment is not yet understood.
Interestingly, a state of transcriptional quiescence also accompanies germline determination in many organisms . In Drosophila and C. elegans, this quiescence is achieved by interfering with RNA polymerase (Pol) II transcriptional activation. A key mechanism of transcriptional regulation is phosphorylation of serine residues (specifically serine 2 and serine 5) within a highly repetitious seven amino acid sequence in the C-terminal domain (CTD) of the largest subunit of Pol II . During the transition from initiation stages to productive elongation, Ser2 is phosphorylated by positive transcription elongation factor (P-TEFb), the predominant kinase complex that targets this residue in the CTD repeat [3, 4]. In C. elegans, in addition to P-TEFb, Tousled-like kinase (TLK-1) has also been implicated in regulating phosphorylation of the CTD of Pol II .
We were intrigued by the simultaneous appearance and disappearance of epitopes that are all correlated with active transcription, and sought to investigate further the transcriptional regulation in the PGCs. We show that Pol II phosphorylation is uniquely regulated and transient in PGCs, suggesting that transcriptional repression is continued in the PGCs after the PIE-1 mode of repression is lost. This repression requires a component called maternal effect sterile (MES)-4, a histone H3K36 methyltransferase that is essential for fertility [16, 17]. Genomewide chromatin immunoprecipitation followed by deep sequencing (ChIP-seq) analyses in embryos show that MES-4-mediated H3K36 methylation marks the bodies of genes that are expressed in germ cells, even those known to express only in post-embryonic stages. Our results are consistent with MES-4 contributing across generations to an epigenetic memory that marks germline-expressed genes and is required for normal germ cell development.
Remarkably, the H5 signal in wild type PGCs is transient. The strong H5 staining that initially appeared in PGCs soon after their birth (~200 minutes post-fertilization/~90 cells, at 22°C; Figure 1C) was not maintained in later stages and was significantly reduced by the 1.5-fold-stage (~450-500 minutes post-fertilization/~550 cells; Figure 1D). H5 staining remained very low in PGCs until hatching (~850 minutes post-fertilization) (Figure 1E). These observations suggest that Pol II may be transiently engaged in some stage of transcription elongation in the nascent PGCs, but that in later embryos the PGCs are largely transcriptionally inactive.
Polymerase (Pol II) phosphorylation dynamics and H3K4me2 erasure in primordial germ cells (PGCs)
H5 staining positive, % (n/total n)
H3K4me2 positive, % (n/total n)
Wild type (N2)
Interestingly, although mes-4 PGC chromatin initially lacked H3K36me3 (Figure 5B, middle of bottom row), this modification was observed to increase gradually in later stage PGC chromatin (Figure 5B, right of bottom row, arrow). This signal is dependent on MET-1, as it does not appear in the double mutant (Figure 5C, arrows). We cannot rule out that there is normally some MET-1-dependent H3K36 methylation in wild type late stage PGCs that is masked by the normal MES-4-dependent deposition of this marker. However, the H3K36 methylation appearing in late stage mes-4 PGCs coincides with the aberrant appearance of H3K4 methylation (Figure 4, Table 1), which suggests that the MET-1-dependent H3K36me3 is also aberrant. The aberrant presence in mes-4 PGCs of hyperphosphorylated Pol II, H3K4me and MET-1-dependent H3K36 methylation supports a conclusion that mes-4 mutant PGCs are precociously transcriptionally engaged. A summary of the dynamics for all of these modifications is shown (see Additional file 2).
We do not yet know if the precocious transcription that we observed in mes-4 PGCs represents premature activation of germline-expressed loci, ectopic activation of soma-expressed genes or some stochastic combination of both. We suspect that ectopic activation of genes expressed in soma is involved because (1) we often observed PGC chromatin structure changes that are reminiscent of somatic nuclei, such as a prominent nucleolus (Figure 1I; DAPI inset) and (2) unlike the absence of 8WG16 staining on the X chromosomes in late wild type PGCs, there is no obvious exclusion of H5 staining on any chromosome in late mes-4 PGCs (see Figure 3 vs. Figure 1I) (additional data not shown).
The above results show that the genomewide maintenance of H3K36me3 in very early embryos, in both somatic blastomeres and the P-cells, is strictly dependent on MES-4 (Figure 5A). The somatic blastomeres have little requirement for genomewide transcriptional activity at these stages, because of the substantial maternal load of RNA and protein, and Pol II transcription is repressed in the P-cells . Maintenance of H3K36me3 in these lineages despite little or no transcriptional activation suggests that MES-4 activity may be capable of operating independently of ongoing transcription. Indeed, previous studies have shown that MES-4-dependent H3K36 methylation in embryos is largely unaffected by RNAi knockdown of the Pol II large subunit AMA-1, whereas the MES-4-independent (MET-1-dependent) signal is strikingly decreased by ama-1 RNAi . Genomewide immunoprecipitation plus microarray (ChIP-chip) analyses also indicate that MES-4 is enriched in gene that lacking detectable Pol II . Thus, whereas MET-1 is required for H3K36me3 in transcriptionally engaged somatic nuclei, MES-4 can provide this modification in nuclei that are largely quiescent and/or those in which Pol II activity has been experimentally crippled.
The loss of both H3K4me3 and AMA-1 in the ama-1(RNAi) animals is evidence that both Pol II activity and its chromatin-associated consequences were strongly affected by the ama-1 RNAi treatment, yet this did not noticeably affect H3K36 methylation by MES-4 in the met-1 animals. This indicates that at least some MES-4-dependent H3K36me3 is uncoupled from ongoing transcription in all stages of the germline cycle tested. MET-1-dependent H3K36me3 signal was also present in adult germ nuclei of mes-4 M+Z- (homozygous null mutants produced from mes-4/+ heterozygous hermaphrodites). This was not observed in mes-4;met-1 double mutant germlines, indicating that MET-1 can provide H3K36 methylation in the adult germline, probably during ongoing germ cell transcription (data not shown). Interestingly, Bender et al. observed that H3K36me2 was not detectable in pachytene nuclei of mes-4 M+Z- adult germ cells , suggesting that MET-1-dependent H3K36 methylation is largely the trimethylated form in these cells.
Early embryos provide an informative window into the maternal to zygotic transition in control of K36 methylation. As illustrated in Figure 5A and described above, in mes-4 M-Z- embryos (lacking both maternal and zygotic MES-4 activity) H3K36me3 is clearly detectable in the pronuclear and early zygotic chromatin. This is presumably the result of MET-1 activity in the parental germline, but this signal is not maintained and disappears by the 4 to 8-cell stage (Figure 5A). The loss of this signal in mes-4 embryos suggests that MES-4 is either essential for maintaining pre-existing H3K36 methylation in embryonic chromatin through cell divisions or must provide this marker de novo after replication-coupled depletion.
We next wished to determine the genomic distribution of MES-4-dependent H3K36 methylation in embryos. H3K36 methylation mediated by Set2-type methyltransferases occurs as a consequence of ongoing transcription, so the distribution of this modification observed in mixed cell populations is likely to be dominated by its cotranscriptional addition. Indeed, the total amount of H3K36me3 detectable by western blot analysis in mixed-stage met-1 embryos is reduced by ~90% compared with wild type levels . To observe only MES-4-dependent methylation, we examined H3K36me3 patterns at high resolution in met-1 mutant embryos using ChIP-seq. In contrast to mes-4 mutants, which exhibit maternal effect sterility, the met-1(n4437) deletion strain is homozygous viable and fertile . We isolated mid to late stage met-1 embryos, which are transcriptionally fully active and within which somatic developmental pathways should be fully engaged. Although both MES-4 protein and MES-4-dependent H3K36me3 become enriched in PGCs at late stages (Figure 5B), the embryo samples used for both met-1 and wild type ChIP-seq contained a mixture of embryos with significant somatic levels of both MES-4 protein and its histone marker (see Additional file 3). We therefore believe that the vast majority of H3K36me3 material we obtained by this method is from the 300-500 somatic nuclei, rather than the two PGC nuclei, in these embryos.
Interestingly, a significant number of genes annotated as having somatic expression also showed H3K36me3 enrichment in the met-1 experiment (Figure 10D). We further analyzed the expression pattern of these genes in a public in situ database (NEXTDB; http://nematode.lab.nig.ac.jp/. Of those genes with a discernable expression pattern, almost all (17/18) showed expression in adult germ cells, indicating that either these genes may be ubiquitously expressed or that their somatic designation is in error (not shown). Collectively, these data suggest that MES-4 maintenance of H3K36me3 in met-1 embryos is largely, if not solely, devoted to genes whose expression occurs in postembryonic larval and adult germ cells. This is again consistent with results obtained by MES-4 ChIP-chip .
Genes that are specifically transcribed in the larval/adult germline but not during embryogenesis would not exhibit transcription-dependent H3K36me3 in embryonic chromatin. We therefore determined if these genes are still marked by MES-4 in met-1 embryos. We selected the intersection of germline-expressed genes (that is, data published in [34, 35]) and 'strictly maternal' genes identified in C. elegans embryonic transcriptome analyses (see experimental procedures). Scatter plot analysis using the 'larval/adult germline-specific' gene set showed obvious enrichment of MES-4-mediated H3K36me3 on most of these genes (Figure 10E). Profiling of H3K36me3 signals across the transcription start and transcription end sites revealed that MES-4-mediated H3K36me3 is enriched in the bodies of these genes, whereas MES-4-mediated H3K36me3 ChIP-seq reads are detected at very low frequencies in the bodies of genes known to be expressed in somatic cells in the embryonic stages examined (Figure 9B). Interestingly, the MES-4-dependent pattern of H3K36 methylation in germline-expressed loci shows a slight enrichment at the 5' end, whereas transcription-coupled H3K36me3 is usually more enriched toward the 3' end of gene bodies . These results indicate that MES-4 activity in met-1 embryos is maintaining H3K36 methylation within the transcribed portion of germline-expressed genes, even though transcription of these genes last occurred in the adult germ cells of the previous generation.
In this study, we show that Pol II phosphorylation and transcriptional activity is uniquely regulated in the primordial germ cells of C. elegans, and that this regulation requires the H3K36-specific methyltransferase MES-4. The unique retention of Pol II phosphorylation in the PGCs that we observe with cdk-9(RNAi) indicates that Pol II activity is regulated differently in these cells. Indeed, the transience of this phosphorylation state followed by a prolonged association of the hypophosphorylated form with only autosomal chromatin is clearly unusual. This autosomal restriction is probably related to the paucity of germline-expressed loci on the X chromosome, and may indicate that Pol II is 'poised' at such loci, held ready at the gate in advance of larval germ cell activation at hatching. The requirement for MES-4 to maintain this status suggests that H3K36 methylation contributes to this germ cell-specific form of Pol II regulation.
H3K36 methylation is intriguing because although its addition has been generally correlated with actively transcribed genes, it has been shown also to negatively affect transcription when ectopically recruited to promoters, and can prevent initiation at cryptic promoters . Interestingly, this dichotomy has been reported to depend on the local H3K4 methylation state. In budding yeast, histone acetylation is promoted by coincidence of H3K36 and H3K4 methylation [38–40]. H3K36 methylation can recruit the Rpd3 S HDAC complex and repress initiation from cryptic intragenic promoters, and this occurs most often in the 3' regions of gene bodies, where Set2-dependent H3K36me is high and Set1-dependent H3K4me is low [41–45]. In C. elegans, H3K4 methylation is extensively erased from the chromatin in PGCs shortly after they are born . By contrast, H3K36me3 is actively maintained by MES-4 in these cells. Assuming the pattern of MES-4 dependent H3K36me3 is consistent in the PGCs, this would create a chromatin status in the 5' region of germline-expressed genes similar to that described above for gene bodies in yeast-- that is, enriched in H3K36me3 and low in H3K4me (Figure 1A, Figure 9B). It is thus possible that the presence of MES-4-mediated H3K36 methylation combined with an absence of H3K4me may present a repressive chromatin signature that prevents or suppresses sustained Pol II activation at most genes in the PGCs. When the PGCs activate after hatching and H3K4 methylation returns, this repressive combination no longer exists and active germline transcription ensues, perhaps through reactivation of Pol II already present (poised) at these loci.
Importantly, the defective PGC repression we observed in mes-4 PGCs was independent of the allele used. The bn67 allele is a point mutant converting a histidine to tyrosine in one of the three PHD domains of MES-4, whereas the bn85 allele is a deletion that disrupts the SET domain. Neither mutation creates a protein null mutant, but both abrogate the localization of MES-4 to DNA . It is thus possible that MES-4 activities other than addition of H3K36me2/3 to histones are required for Pol II repression in the PGCs, and that these activities require recruitment to the DNA for repression. However, the PHD and SET domains (including AWS-like, SET and post-SET motifs) are the only recognizable domains of MES-4: both are known to interact with histones and both are required for H3K36 methylation by MES-4. A direct role of H3K36 methylation in Pol II repression in the PGCs awaits future analysis using a mes-4 methyltransferase catalytic mutant.
MES-4 appears to 'maintain' rather than establish H3K36 methylation in genes, independently of their transcriptional status. We base this model on the following observations: (1) In the absence of MES-4, the H3K36me3 arriving in gamete chromatin is quickly diluted by replication and cell division (Figure 5A); (2) zygotically provided MES-4 cannot contribute detectable H3K36 methylation on its own in larval or embryonic germ cells (Figure 7); (3) neither normal transcriptional quiescence (for example, P cells) nor experimental disruption of Pol II activity in germ cells noticeably affects H3K36 methylation (Figure 5, Figure 6); (4) in embryos, MES-4-dependent H3K36me3 is enriched within the body of genes that are known to be expressed only in post-embryonic germ cells (Figure 9); (5) the gene-body distribution shows an unusual 5' enrichment, which does not overlap with the reported distributions of either H3K36me3 or Pol II in other organisms and, (6) also in embryos, MES-4-dependent H3K36me3 is not enriched within the body of somatically-expressed genes that are transcriptionally active in the stages analyzed (Figure 9, Figure 10). A recent ChIP-chip analysis of the genomewide distribution of MES-4 in embryos confirms that MES-4 is enriched in germline-expressed loci and that this enrichment is independent of Pol II occupancy .
There has been considerable controversy regarding the ability of histone modifications to provide stable and heritable epigenetic information, given the extensive nucleosome dynamics throughout the cell cycle. However, recent studies have identified molecular mechanisms that allow epigenetic markers in both DNA and chromatin to be actively maintained despite such dynamics. The maintenance of DNA methylation is mediated by recognition of hemimethylated CpG sequences by the SRA domain of UHRF1, which recruits Dnmt1 to the DNA [46–50]. Similarly, the PRC2 complex, which catalyzes H3K27me3, was shown to bind to pre-existing H3K27me3, and this binding appears to be crucial for the maintenance of this modification in proliferating cells [51, 52]. MES-4 may be similarly stabilized on nucleosomes through its three PHD domains, which have been shown to help recruit the Rpd3 S complex to nucleosomes marked by methylated H3K36 in yeast . Interestingly, both the mes-4 bn67 allele studied here and another allele with mutations in one of the PHD domains (bn50) cause dissociation of MES-4 from chromatin and the same germ cell degeneration phenotype observed in null alleles . Mutations in mrg-1, which encodes a germline-enriched chromodomain protein, cause PGC proliferation defects similar to those in mes-4 mutants and misregulation of genes that are also misregulated in mes-4 mutants. The MRG-1 protein, like MES-4, is also excluded from the X chromosome . MRG-1 may thus also participate in PGC transcriptional repression by MES-4.
The maintenance of epigenetic information is particularly relevant to transmission through the germline, as this information has the potential to affect gene regulation across multiple generations. The maintenance of H3K36me by MES-4 in embryos, which we observed in the bodies of genes transcribed in adult germ cells, suggests that this marker is important for such gene regulation. Indeed, a transgenerational requirement for this marker is revealed by the mes-4 (maternal effect sterile) phenotype; the dysregulation that occurs in mes-4 PGCs is not observed until a full generation after the actual loss of MES-4 protein activity. In mes-4 M+Z- embryos, the maternally provided MES-4 protein becomes focused in the PGCs as in wild type embryos and maintains H3K36 methylation status in these cells. When the embryo hatches and germline development progresses, the thousand or so germ cells arising from these PGCs are functional and can develop into normal gametes, despite having only transcription/MET-1-dependent H3K36 methylation. However, in the next (M-Z-) generation, the MET-1-dependent H3K36me3 arriving within the gamete chromatin is not maintained and the 'information' is lost in the embryonic germline. The PGCs lose stable Pol II repression, and their few descendents degenerate after postembryonic activation . MES-4 is thus important for maintaining the H3K36m3 marker in germline genes, which may be essential for their normal regulation in the germ cells of subsequent generations.
The proposed maintenance histone methylation activity can also potentially explain, at least in part, the strict maternal effect sterility observed for mes-4 mutants, in which zygotically supplied MES-4 cannot rescue the germ cell degeneration phenotype (for example, M-Z+ embryos). The absence of maternal MES-4 maintenance activity in the embryonic germ cells would result in the loss of inherited H3K36me template through replication-coupled dilution (Figure 5A), yielding little 'template'/substrate for the zygotic MES-4 activity, which is not synthesized until later in development (~300-400 cells) (H. Furuhashi, unpublished results). Therefore, no rescue by zygotic MES-4 is observed.
Curiously, the marking of germline-expressed genes by MES-4 is also crucial for allowing ectopic activation of germline-expressed loci in somatic cells of mutant backgrounds that are defective in global transcriptional repression mechanisms [54, 55]. The reasons for this ectopic expression and for the requirement for MES-4 activity to allow the expression to occur in post-embryonic stages are currently unknown. The requirement for MES-4 marking to yield ectopic germline gene expression when somatic modes of repression are defective may indicate that H3K36 methylation plays a prominent role in marking genes for default expression when somatic repression mechanisms are absent (as in germ cells) or defective (as in the mutants).
The noted lack of association of MES-4 with the X chromosome in embryos can be readily explained by the paucity of germline-expressed loci on the X chromosome . The methylation pattern of MES-4 in the embryo may represent the maintenance of H3K36me3 at loci that were originally marked by transcription-coupled H3K36 methylation in the adult germ cells of each preceding generation. The only genes within which such 'transcriptional memory' could be transmitted across generations would be those that are expressed in adult germ cells--that is, the pattern we observed in met-1 embryos. An intriguing possibility is that the pattern of H3K36 methylation in met-1 embryos is evidence that this memory can be highly stable-that is, that loci originally marked by transcription-coupled MET-1 activity when present in the strain many, many generations previously is still being faithfully maintained by MES-4. The gene-body distribution of MES-4-dependent H3K36me3 in embryos indicates that it can be concentrated in areas where germ cell transcription occurred in the parental germline, yet where there is no evidence of ongoing transcription in the embryo. Indeed, a recent ChIP-chip analysis of MES-4 protein in wild type embryos showed that MES-4 protein is found within germline-expressed loci that lack detectable Pol II . However, although we have strong evidence that MES-4 activity can operate independently of transcription, we cannot know with certainty that this is always the case. Nevertheless, it is clear that MES-4 can provide stable maintenance across generations of H3K36me in germline genes, regardless of their transcriptional status.
MES-4 appears to be a metazoan-specific H3K36 HMTase. MES-4-related proteins in other systems (dMES-4 in Drosophila and the NSD family of proteins in mammals) are crucial for normal development and are implicated in various cancers and developmental disorders such as Sotos and Wolf-Hirschhorn syndromes [56–59]. The molecular mechanisms underlying the developmental requirement for these proteins in these organisms are not clear. Indeed, the role of MES-4 in C. elegans is somewhat paradoxical. For instance, although MES-4 is largely absent from the X chromosome, genes on the X chromosome are the major targets of dysregulation in mes-4 (M+Z-) mutant adult germ cells . It has been proposed that MES-4 indirectly affects X chromatin structure by preventing repressive factors from accessing autosomal chromatin, thereby focusing their action on the X chromosome . In mes-2/3/6 mutants, MES-4 is observed to localize ectopically to the X chromosome in oocytes . This apparent connection between MES-4 and the other MES complex in adult germ cells may be separable from PGC-specific processes, because mes-2 mutations do not affect the PGC processes we studied.
Our results indicate that H3K36 methylation can serve as an important component of epigenetic memory, and that this memory is required for germline continuity in C. elegans. Although H3K36 methylation has been generally correlated with ongoing transcription elongation [22, 28], the H3K36me3 that is enriched in the C. elegans embryonic germline chromatin is dependent on MES-4, a methyltransferase whose activity can be independent of Pol II, as detailed above [17, 26]. H3K36 methylation on germline-expressed genes is stably maintained by MES-4 across generations, and our data suggest that MES-4 and/or the information transmitted by its histone modification product play a key role in preventing abnormal transcriptional activation in the PGCs of subsequent generations. These results provide new insights and identify additional modes of chromatin-based, transgenerational transcriptional regulation in metazoan development and germline specification.
C. elegans N2 Bristol strain was used as the wild type. The following mutations, balancers and translocations were used: LGI: met-1(n4337); LGII: mes-2(bn11), unc-4(e120), mnC1; LGIV: DnT1[qIs51](IV;V); LGV: dpy-11(e224), mes-4(bn67, bn73, bn85), DnT1[qIs51](IV;V)
met-1(n4337) is a deletion mutant that lacks the SET domain . mes-2(bn11) is a point mutation that results in a premature stop codon before the SET domain; it produces no detectable protein and very little to no H3K27me3 staining in the PGCs [23, 60]. mes-4(bn85) has an inframe deletion that disrupts the SET domain. mes-4(bn73) has a premature stop codon after amino acid 593 in the middle of the SET domain. mes-4(bn67) has a point mutation in its first PHD finger, which leads to complete dissociation of MES-4 protein from chromosomes . unc-4(e120) and dpy-11(e224) are visible genetic markers.
RNAi was performed to deplete RNA Pol II/AMA-1, CDK-9 and TLK-1 from embryos by microinjection into parent worms as described previously . Embryos in the injected animals were dissected out after ~24 h at 20°C, and prepared for whole-mount fixation and immunofluorescence analyses (see below). To deplete Pol II/AMA-1 from adult germlines, L4 larvae were soaked in 0.5 μg/μl ama-1 dsRNA for 24 hours at 20°C. The animals were then transferred onto NGM/Amp/IPTG plates (3g NaCl + 17g agar + 2.5g Bacto-peptone in 1L dH20, made 50 μg/ml ampicillin and 1 mM IPTG) seeded with HT-115 bacteria that had been transformed with an ama-1 dsRNA expression plasmid or control L4440 plasmid. The feeding RNAi was performed at 20°C for ~55 hours, during which the feeding plate was exchanged once after the first 24 hours.
Samples were fixed using methanol/formaldehyde  for H5 staining, and methanol/acetone  for H3K4me2 and H3K36me3 staining. For 8WG16 and MES-4 staining, embryos were fixed in 2.5% paraformaldehyde for 2 minutes, followed by a 2 minute post-fix in -20°C methanol. Primary antibodies used were: affinity-purified rabbit anti-MES-4  (1:10), mouse monoclonal CMA333 to H3K36me3 (0.25 μg/ml, rabbit anti-PGL-1  (1:10,000), mouse monoclonal OIC1D4 to P-granules  (1:4), mouse monoclonal H5 to hyper-phosphorylated Pol II CTD (1:50)  and mouse monoclonal 8WG16 to Pol II (Covance) (1:100). We extensively characterized the specificity of monoclonal H3K36me3 (CMA333) by immunostaining (Figure 3; see Additional file 3), immunoblotting and ELISA (data not shown). For H3K4me2 staining, we used both a fully characterized monoclonal antibody CMA303  and a rabbit polyclonal antibody (Millipore Corp., Billerica, MA, USA) Both antibodies yielded essentially the same results for the staining of PGC nuclei, although the polyclonal antibody tended to produce slightly higher background staining (data not shown). The data shown were obtained using the monoclonal anti-H3K4me2 antibody. Secondary antibodies used were Alexa Fluor 488-conjugated goat anti-mouse and Alexa Fluor 594-conjugated anti-rabbit antibodies (1:500) (Molecular Probes, Eugene, OR, USA). Samples were mounted in anti-fade reagent (ProLong Gold; Molecular Probes) and observed under a fluorescence microscope (Leica DMRXA; Hamamatsu Photonics, Hamamatsu, Japan) with Simple PCI software (Hamamatsu Photonics).
Synchronized, gravid young met-1 or wild type N2 adults were rinsed with M9 buffer and lysed in freshly prepared egg isolation solution (20% fresh commercial bleach, 500 mM NaOH) to collect embryos. Isolated embryos were rinsed with M9 buffer and separated from lysed worms and other debris by sucrose flotation as described previously . To allow embryos to develop to later stages, collected embryos were incubated in M9 buffer at 20°C for 5 hours. An image showing a representative field of embryos after isolation is provided (see Additional file 3). After incubation, eggshells were digested by treatment with egg buffer (25 mM HEPES pH7.4, 118 mM NaCl, 48 mM KCl, 2 mM MgSO4, 2 mM CaCl2) containing 1U/ml of chitinase (C6137; Sigma Chemical Co., Sigma, St Louis, MO, USA) at 20°C for 40 minutes. The embryos were rinsed with egg buffer and frozen in liquid nitrogen. The frozen embryo pellet (100-150 ul vol) was resuspended in 1 ml phosphate-buffered saline (PBS) containing protease inhibitor cocktail (Roche Applied Sciences, Indianapolis, IN, USA) and homogenized with a Dounce homogenizer. To fix chromatin, formaldehyde was added to the homogenate to a final concentration of 1%, and the tube rocked for 10 minutes at 25°C. After quenching with 125 mM glycine, the nuclei/chromatin fraction was collected by centrifugation at 900g for 1 minute, washed with 1ml PBS three times, and resuspended in 400 μl SDS lysis buffer (included in ChIP Assay Kit; Millipore Corp., Billerica, MA, USA) containing protease inhibitor cocktail. The fixed material was placed in a sonicator to give sheared chromatin preparations with an average DNA size of ~300 bp. After centrifugation at 13,000g for 10 minutes, the supernatant was collected and diluted 10-fold in ChIP dilution buffer (16.7 mM Tris-HCl pH7.5, 167 mM NaCl, 0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA). The DNA concentration of the chromatin preparations was determined and adjusted with LSW buffer (20 mM Tris-HCl pH7.5, 1 mM EDTA, 1% Triton X-100, 0.1% SDS, 150 mM NaCl) to ~20 μg/ml for H3K36me3 ChIP in met-1, ~10 μg/ml for H3K36me3 ChIP and ~5 μg/ml for pan-H3 ChIP in both genotypes. Aliquots of 2 ml (for H3K36 ChIP) or 1 ml (for pan-H3 ChIP) of chromatin preparation were mixed with 5 μg of monoclonal anti-H3K36me3 (CMA333 (supplied by co-author, H. Kimura)) or 2.5 μg of monoclonal anti-H3 (ab10799; Abcam Inc., Cambridge, MA, USA) overnight at 4°C on a rotator. Magnetic beads (150 μl) coated with anti-mouse IgG (Dynabeads; Dynal Bead Based Separations (Invitrogen Group) Carlsbad, CA, USA) were then added to each 1 ml reaction and rotated for 3 hours at 4°C. Beads were washed for 3 minutes with 1 ml of each of the following buffers in succession: LSW buffer, HSW buffer (20 mM Tris-HCl pH7.5, 1 mM EDTA, 1% Triton X-100, 0.1% SDS, 500 mM NaCl) and TEL buffer (0.25 M LiCl, 1% NP-40, 1% sodium deoxycholate, 1 mM EDTA, 10 mM Tris-HCl pH8.1), followed by two final washes with TE buffer (10 mM Tris-HCl pH8.0, 1 mM EDTA). Samples were eluted twice with elution buffer (10 mM Tris-HCl pH8.0, 1 mM EDTA, 1% SDS, 250 mM NaCl) for 15 minutes at 65°C, treated with proteinase K at 55°C for 1-2 hours, then transferred to 65°C overnight to reverse-crosslink. DNA was purified with a commercial kit (PCR Purification Kit; Qiagen, Valencia, CA, USA) and eluted in 40 μl 10 mM Tris-HCl pH 8.0. Eluate DNA was quantified (PicoGreen; Invitrogen) and adjusted to 1 ng/μl.
Libraries were prepared from 30 ng of purified immunoprecipitated DNA or input DNA and analyzed on a genome analyzer (Illumina Inc., San Diego, CA, USA) at the UNC sequencing facility. Experiments were carried out using two biologically independent samples.
The sequenced reads were mapped to the C. elegans genome (WormBase version WS170, ce4) using MAQ http://www.maq.sourceforge.net using default parameters. The mapped 36 bp reads were extended to 200 bp. For each base pair in the genome, the number of overlapping sequence reads was determined. For all replicates of Input and H3 and H3K36me3 ChIP experiments, the read count per base pair was scaled so that the median read count across the genome was the same. Read counts per base pair were averaged across replicates and visualized in the UCSC genome browser http://genome.ucsc.edu/.
Both met-1 ChIP-seq experiments displayed robust H3K36me3 signal over numerous open reading frames with significantly lower signal in most non-coding regions; we consider the latter to be 'background' or 'noise'. Both individual N2 ChIP-seq experiments had a somewhat lower signal to background ratio. Despite the difference in signal to background and the fact that read counts per base pair are not directly comparable between met-1 and N2, we could readily identify a large number of genes with clear enrichment of H3K36me3 in both samples. This allowed us to compare H3K36me3 on genes in particular categories (for example, those expressed specifically in germ cells or expressed in somatic cells) between the two genotypes.
Scatter plots (Figure 10) were generated by averaging the read counts of base pairs within the transcription start and end sites for each gene. Log10 of the average read counts per gene were plotted for H3K36me3 in met-1 versus N2.
Average profiles of H3K36me3 in met-1 around transcription start sites (TSS) and transcription end sites (TES) for larval/adult germline-specific and soma-expressed genes are shown in Figure 10. Read counts per base pair were averaged in 50 bp intervals for each gene from 1 kb upstream to 1.5 kb downstream of the TSS and 1.5 kb upstream to 1 kb downstream of the TES. Genes were aligned at the TSS and TES, and the average read count calculated for germline-specific and soma-expressed genes in 50 bp steps. Error bars indicate the 95% confidence interval for the mean of each 50 bp interval.
The gene sets (Figure 9) were defined based on various expression data sets. In total, 2243 germline-enriched genes were obtained from previous work ; spermatogenesis genes were excluded from this set. Ubiquitously expressed genes (n = 2580) were defined as genes expressed in germline, muscle, nerve and gut, according to SAGE analysis (SAGE tag count ≥ 1) [35, 65]. Somatic genes (n = 1273) were defined as genes expressed in at least one of the three somatic SAGE data sets (≥ 8 tags) and not expressed in the germline SAGE data set (≥ 1 tag). Larval/adult germline-specific genes (n = 675) were defined as the intersection of the germline-enriched genes , the germline expressed genes from SAGE analysis (≥ 1 tag)  and the class of 'strictly maternal' genes identified in the C. elegans embryonic transcriptome analysis by Baugh et al. .
A file containing the sequences of the primers used is available (see Additional file 6).
Freter et al.  recently reported that RNA Pol II pSer2 is significantly reduced in a number of adult somatic and germline stem cell types, suggesting that the global repression of pSer2 that we observe in late stage C. elegans PGCs may be a conserved and common feature of pluripotent cells in many organisms.
We thank Robert Horvitz, Geraldine Seydoux and Taryn Phippen for reagents, and Jason Lieb, Kohta Ikegami and members of the Kelly and Strome laboratories, especially Taryn Phippen, for helpful discussions. This work was supported by NIH grants GM077600 (W.G.K) and GM34059 (S.S.) and NHGRI modENCODE grant U01 HG004270.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.