The histone H3K36 methyltransferase MES-4 acts epigenetically to transmit the memory of germline gene expression to progeny

Abstract

Methylation of histone H3K36 in higher eukaryotes is mediated by multiple methyltransferases. Set2-related H3K36 methyltransferases are targeted to genes by association with RNA Polymerase II and are involved in preventing aberrant transcription initiation within the body of genes. The targeting and roles of the NSD family of mammalian H3K36 methyltransferases, known to be involved in human developmental disorders and oncogenesis, are not known. We used genome-wide chromatin immunoprecipitation (ChIP) to investigate the targeting and roles of the Caenorhabditis elegans NSD homolog MES-4, which is maternally provided to progeny and is required for the survival of nascent germ cells. ChIP analysis in early C. elegans embryos revealed that, consistent with immunostaining results, MES-4 binding sites are concentrated on the autosomes and the leftmost approximately 2% (300 kb) of the X chromosome. MES-4 overlies the coding regions of approximately 5,000 genes, with a modest elevation in the 5' regions of gene bodies. Although MES-4 is generally found over Pol II-bound genes, analysis of gene sets with different temporal-spatial patterns of expression revealed that Pol II association with genes is neither necessary nor sufficient to recruit MES-4. In early embryos, MES-4 associates with genes that were previously expressed in the maternal germ line, an interaction that does not require continued association of Pol II with those loci. Conversely, Pol II association with genes newly expressed in embryos does not lead to recruitment of MES-4 to those genes. These and other findings suggest that MES-4, and perhaps the related mammalian NSD proteins, provide an epigenetic function for H3K36 methylation that is novel and likely to be unrelated to ongoing transcription. We propose that MES-4 transmits the memory of gene expression in the parental germ line to offspring and that this memory role is critical for the PGCs to execute a proper germline program.

Figure 1. Two C. elegans HMTs, MES-4 and MET-1, perform H3K36 trimethylation and display different dependence on Pol II.

H3K36me3 levels are reduced relative to wild type (A) in mes-4 mutant (B), met-1 mutant (C), and ama-1(RNAi) (E) ∼80–110-cell embryos. H3K36me3 is undetectable in mes-4; met-1 (D) and mes-4; ama-1(RNAi) (F), but is still present in met-1; ama-1(RNAi) (G) embryos. Embryos were stained in parallel and images acquired and processed identically. See Figure S1 for specificity of H3K36me3 antibody.

Figure 2. MES-4 is concentrated on the autosomes and the left tip of the X chromosome.

(A) The anti-MES-4 antibody used for ChIP stains the 10 autosomes and the left tip of the X (arrow; the other X chromosome is marked with an arrowhead), similar to previously characterized antibodies , . The MES-4 signal was over-exposed to show the dot of MES-4 on the left end of X and the absence of detectable MES-4 along the rest of the X. Additional staining, including absence of MES-4 signal in a mes-4 mutant, is shown in Figure S2B. (B) 5291 of the 5391 MES-4 binding sites are distributed along the five autosomes. Of the 100 MES-4 peaks on the X chromosome, 25 are in the leftmost 300 kb. The expected and observed numbers of MES-4 peaks on each chromosome are in Figure S2D. All of the ChIPs shown in Figure 2, Figure 3, Figure 4, Figure 5, and Figure 6 were performed in early embryo extracts.

Figure 3. MES-4 is mostly concentrated on Pol II–bound genes.

For a representative ∼180 kb autosomal (ChrIV) region of the genome, annotated genes (arrows; for genes encoding multiple transcript variants, the longest ones are shown) and ChIP z-scores (standardized log₂ ratios of ChIP/Input signals) for Pol II, MES-4, MES-4::FLAG, H3K36me2, H3K36me3, H3K4me3, H3K9me3, H3, and a control ChIP (Protein A beads + rabbit IgG) are shown. Pol II, MES-4, H3K36me2 and H3K36me3 generally colocalize over bodies of transcribed genes. H3K4me3 is mainly found in the promoter and 5′ coding region of transcribed genes. H3K9me3 is absent from expressed genes. See Figure S3 for analysis of reproducibility and concordance among ChIPs.

Figure 4. MES-4 and H3K36 methyl marks are concentrated over gene bodies of highly expressed genes.

Genes >2 kb long (9932 genes) were grouped into 5 equal sized bins based on expression in early embryos. Colors display bins with highest (green) to lowest (orange) transcript level in early embryo extracts generated in parallel with ChIP extracts. Average z-score profiles for each ChIP target are shown in 50 bp steps in the 2.5 kb around the transcript start site (TSS) and the transcript end site (TES). Error bars indicate 95% confidence intervals for the means. The bottom row shows four different control ChIPs: Protein A beads, Protein G beads, Protein A beads coated with rabbit IgG, and sheep anti-mouse beads coated with mouse anti-FLAG. GC content variation is also shown.

Figure 5. Distribution of RNA, Pol II, MES-4, H3K36me2, and H3K36me3 levels for various gene sets.

(A) Gene classes were defined as described in Results. The number of genes in each set is shown in parentheses. For each gene set, boxplots show the distribution of RNA level (standardized to z-score calculated from our microarray analysis) and Pol II, MES-4, H3K36me2 and H3K36me3 average z-scores from early embryo ChIP-chip analysis. Each box extends from the 25^th to 75^th percentile of the z-scores in the set, and the whiskers extending from a box indicate the 2.5^th and 97.5^th percentile. Wedges around the median indicate 95% confidence interval for the medians. See Figure S4C for scatter plots of the ubiquitous, germline-expressed, germline-specific and embryo-expressed categories. (B) Examples of Pol II- MES-4+ H3K36me+ and Pol II+ MES-4− H3K36me− genes. The germline-expressed gene csr-1 (left dashed box) contains high levels of MES-4, H3K36me2 and H3K36me3 but lacks Pol II and H3K4me3. The soma-expressed gene ceh-13 (right dashed box) contains Pol II and H3K4me3 but lacks MES-4, H3K36me2 and H3K36me3. Both csr-1 and ceh-13 are flanked by genes that show the typical high concordance between Pol II, MES-4 and H3K36me marks. See Table S2 for lists of genes in different temporal-spatial classes and Figure 6C for more examples of individual genes.

Figure 6. RNAi depletion of MES-4 substantially reduces H3K36me3 on germline-expressed genes.

(A) MES-4 and H3K36me3 ChIP-chip signal on the autosomes (gray) and the X chromosome (red), specifically on genes with the highest expression in wild type (top 20% in dark color) and genes with the lowest expression in wild type (lowest 20% in light color). MES-4 is dramatically reduced genome-wide in mes-4(RNAi). H3K36me3 on autosomes is reduced to the lower level observed on the X, from which MES-4 is largely absent in wild type. The remaining low level of H3K36me3 on transcribed genes in mes-4(RNAi) is most likely due to transcription-coupled methylation by MET-1. (B) MES-4 and H3K36me3 ChIP-chip signal on ubiquitously-expressed (orange), germline-expressed (green), germline-specific (blue), and embryo-expressed (red) gene sets in wild type and mes-4(RNAi). In mes-4(RNAi), H3K36me3 is dramatically reduced on genes expressed in the germline (ubiquitous and germline-expressed) and undetectable on germline-specific genes. The embryo-expressed class (and soma-specific class, not shown) lack MES-4 in wild type and show little change in mes-4(RNAi). (C) Examples of individual genes illustrating the patterns observed in (B).

Figure 7. The distribution of and requirement for MES-4-catalyzed H3K36 methyl marks.

(A) Germline-specific and soma-specific genes display different profiles of H3K36me3, MES-4, and Pol II in early embryos (emb) and H3K36me3 in L3 larvae (L3 data are from [37]). Gene profiles are as described for Figure 4, using the germline- and soma-specific genes described in Results. This figure highlights the reciprocal relationship between Pol II levels and MES-4/H3K36me3 levels on germline- and soma-specific genes in early embryos (also see Figure 5). On germline-specific genes, H3K36me3, like MES-4, is slightly more elevated in 5′ than 3′ gene bodies, whereas on soma-specific genes, H3K36me3 is more elevated in 3′ than 5′ gene bodies, especially in L3s, probably due to transcription-dependent methylation by MET-1. (B) Embryonic expression of MES-4(+) rescues fertility of mes-4 mutants only when maternal MES-4 retains some HMT activity. Analysis of mutant MES-4 proteins is from . M-Z− (lacking both maternal and zygotic expression of mes-4+) worms were mes-4/mes-4 hermaphrodites produced by mes-4/mes-4 mothers. M-Z+ worms were mes-4/+ hermaphrodites produced by mating mes-4/mes-4 mothers with wild-type males. At least 24 M-Z+ hermaphrodites were assessed for fertility. 75% of bn58/+ hermaphrodites produced viable progeny; 8% produced embryos that failed to hatch.