MES-4 has emerged as a critical germline regulator in several studies. Its maternal-effect sterile (mes) phenotype demonstrates that the action of MES-4(+) in the maternal germ line and/or a maternal supply of MES-4(+) is required for the PGCs in offspring to thrive and survive 
. Germline functions that require MES-4 are repression of genes on the X chromosome, transgene silencing, cosuppression, and RNAi 
. In addition to all of these repressive roles, MES-4 is required to promote ectopic expression of germline genes in somatic cells of synMuv B mutant larvae 
. The current study identified MES-4 binding targets as being genes expressed in the adult germ line. The targeting and function of MES-4 appear to be distinct from the well-studied Set2 family of H3K36 HMTs. We speculate that MES-4 transmits the memory of germline gene expression patterns from the parental germ line to the PGCs in offspring, and that loss of this function causes PGCs to degenerate in young larvae.
By both immunofluorescence and ChIP-chip analysis, MES-4 is concentrated on the five autosomes and the very left tip of the X (
and this study). The strong autosomal bias of MES-4 ChIP signal is not due to an X chromosome structure that is recalcitrant to ChIP, as ChIP analysis of the dosage compensation complex preferentially targets the X 
. Instead, the autosomal enrichment of MES-4 is likely explained by the association of MES-4 with germline-expressed genes, which are significantly under-represented on the X 
. The left end of X has several distinquishing properties, including a relatively high concentration of H3K9me3 and low density of periodic AA/TT clusters compared to the rest of the X 
and the presence of the X chromosome “pairing center” for meiosis 
. The autosome-like higher density of MES-4 binding sites on the left end of X is consistent with its being “pseudo-autosomal” in nature. Perhaps it arose by translocation of an autosomal segment to the end of the X. In fact, comparison of the genomes of C. elegans
and Pristionchus pacificus
suggests that they evolved from a common ancestor in which chromosomes X and V were fused (
and R. Sommer, personnal communication). The autosomal concentration of MES-4 is regulated by the other three MES proteins, MES-2, MES-3, and MES-6, which form the C. elegans
Polycomb Repressive Complex 2 and like PRC2 catalyze methylation of H3K27 
. Loss of any component of the MES-2/3/6 complex leads to loss of silencing of the X chromosomes in the germ line and significant MES-4 immunofluorescence signal along the full length of the X in oocytes and embryos 
. Our model predicts that in mes-2
, and mes-6
mutant embryos, elevated MES-4 signal will be observed on X-linked genes whose expression was up-regulated in the maternal germ line.
Our studies demonstrate that in early embryos MES-4 associates with genes that were expressed in the maternal germ line. Genes expressed exclusively in the maternal germ line have high MES-4 and low Pol II. Some individual genes are MES-4+ Pol II−. These genes strongly argue that Pol II is not required to maintain MES-4 association with genes. Conversely, genes newly expressed in embryos and genes expressed specifically in somatic cells have high Pol II and low MES-4. Indeed, numerous individual genes are Pol II+ MES-4−. This is consistent with Pol II not being sufficient to recruit MES-4 to expressed genes.
Additional evidence supports the notion that MES-4 is capable of associating with gene bodies and methylating H3K36 independently of Pol II. First, RNAi depletion of the large subunit of RNA Pol II does not impair MES-4 binding to chromosomes or H3K36 di− or trimethylation, as detected by immunofluorescence (
and this study). Second, the slightly 5′ enriched distribution of MES-4 across genes bodies is quite different from the 3′ enriched distribution of Set2 homologs, the latter driven by association of Set2 with elongating Pol II 
. Recent studies in yeast have revealed that Set2 can be recruited to genes independently of Pol II, but is impaired in H3K36 di− and trimethylation 
. In contrast, MES-4 appears capable of di− and trimethylating H3K36 in the absence of Pol II. Taken together, our findings suggest that MES-4 and perhaps the related mammalian NSD proteins provide another layer of function for H3K36 methylation that is novel and likely to be unrelated to ongoing transcription.
An attractive model is that MES-4 serves as a maintenance HMT to mark germline-expressed genes and pass the memory of gene expression from one generation of germ cells to the next. In this model, the H3K36 HMT MET-1 serves a Set2-like role and tracks with Pol II to methylate H3K36 during gene expression in the adult germ line; during embryogenesis, MES-4 maintains H3K36 methylation on those genes independently of Pol II. We think that MES-4 can serve this maintenance role even in transcriptionally repressed cells (e.g. in the germline blastomeres) and potentially for generations (e.g. in met-1
mutant worms). In support of this model, MES-4 does not appear to be capable of de novo H3K36 methylation in the soma and embryonic PGCs: embryonic expression of MES-4 in embryos that lack maternal H3K36me3 (mes-4
double mutant mothers) does not generate detectable H3K36me3 signal (Furuhashi et al., unpublished). In contrast, embryonic expression of MET-1 generates robust H3K36me3. Thus, our model posits that MES-4 is a specialized maintenance HMT that in embryos can only methylate chromatin with pre-existing H3K36 methyl marks. Consistent with this, embryonic expression of MES-4(+) can rescue the fertility of embryos from mothers that produce mutant MES-4 with weak HMT activity, but not embryos from mothers that lack MES-4 or that produce MES-4 that lacks HMT activity. The maintenance HMT activity of MES-4 enables MES-4 to serve a truly epigenetic role, propagation of a particular chromatin state through meiosis and mitosis. Recent studies demonstrate that the Polycomb Repressive Complex 2 both initiates and maintains a repressed chromatin state, the latter by binding the chromatin marks that it generates 
How is MES-4 initially recruited to germline-expressed genes? Our findings that MES-4 can associate with genes and generate H3K36 methylation independently of Pol II do not rule out the possibility that MES-4 has a Pol II-dependent mode as well. In the adult germ line, MES-4 may be initially recruited to expressed genes in a Pol II-dependent manner. Another possibility is that MES-4, perhaps via its PHD fingers, binds H3K36 methyl marks generated by MET-1 and/or MES-4. Yet a third possibility is that the chromodomain protein MRG-1, like its counterparts in other systems 
, associates with methylated H3K36 and helps recruit MES-4. In fact, MRG-1, like MES-4, displays autosomal enrichment by immunofluorescence, and mrg-1
mutants display a suite of mes-4
-like defects, including maternal-effect death of PGCs 
. MES-4 association with autosomes is not lost in mrg-1
mutants and vice versa 
. Determining whether MRG-1 participates in recruiting MES-4 and/or is a downstream effector of MES-4-mediated H3K36 methylation is in progress.
An important question for future investigation is why the PGCs in embryos from mes-4
mutant mothers die. An attractive scenario is that MES-4 marking of genes expressed in the maternal germ line identifies genes to be expressed in the progeny's germ line. A related scenario, which might temporally precede the first, is that MES-4 marking of genes helps keep those genes repressed in the PGCs during embryogenesis. Indeed, wild-type PGCs generally do not acquire marks of active transcription until after embryos hatch into L1s, while mes-4
mutant PGCs acquire such marks prematurely, during mid-embryogenesis (
and Furuhashi et al., unpublished). In addition to a potential role in up- or down-regulating transcription in the PGCs, MES-4 may influence gene expression at the level of regulation of splicing. Recent papers report that H3K36me3 is enriched in expressed exons relative to introns 
and that the level of H3K36me3 influences alternative splicing 
. Like H3K36me3, MES-4 appears to be exon-enriched ( and Figure S5
), raising the exciting possibility that MES-4 preferentially associates with exons and methylates them in a manner that facilitates or regulates splicing. Experiments are planned to isolate PGCs from wild-type and mes-4
mutant embryos and compare their RNA accumulation and splicing patterns. In the meantime, we analyzed the overlap between MES-4-bound genes in embryos and genes mis-regulated in mes-4
mutant adult germ lines, and found it to be small (Figure S6
). The small overlap may be due to the stage difference (i.e. embryo versus adult germ line) or to technical and biological effects that cause even well documented transcription regulators to display a low overlap between factor-bound genes and genes mis-regulated when the factor is absent [e.g. 58]
. Despite the absence of gene expression data in PGCs, the strong dependence of early PGC development on maternal MES-4 demonstrates the functional importance of MES-4.
MES-4 and its HMT activity are detected in all cells of early embryos and become restricted to the PGCs during mid to late embryogenesis (
and our unpublished results). Our model that MES-4 propagates the memory of germline gene expression through embryogenesis raises the question how somatic cells in the embryo deal with that signal. Our current view is that the synMuv B chromatin regulators antagonize germline fate in somatic cells. Loss of synMuv B proteins causes somatic cells to express germline-specific genes, and concomitant loss of maternal MES-4 suppresses the germline potential of somatic cells 
. Further tests of our model and of the interplay between MES-4, MET-1, MRG-1 and the synMuv B chromatin regulators will shed light on how germline identity is passed from generation to generation and how germline gene expression patterns are controlled.