Fig. 1

The three H3K36me states have distinct genome-wide distributions. a Genome-browser tracks of H3K36me states in wildtype mMSCs, illustrating broad distribution of H3K36me2 in genic and intergenic regions (IGRs), while H3K36me1/3 are predominantly distributed within genes. b Barplots of global H3K36me1/2/3 abundance as quantified by MS: H3K36me2 shows the highest abundance, followed by H3K36me1/3. Error bars indicate mean ± standard deviation. Individual data points represent biological replicates (n = 3). c Heatmaps of enrichment patterns ± 20-kb flanking IGRs (n = 9551), indicating high levels of H3K36me2, low H3K36me1 and absence of H3K36me3 at IGRs. H3K36me signals were input- and depth-normalized to visualize distribution at IGRs despite large differences in total abundances. d Genome-browser tracks demonstrating H3K36me1/2 are highest in 5′ and decrease in 3′ genic regions where H3K36me3 is more prominent. e Meta-genebody plots of H3K36me1/2/3 distributions, highlighting 5′ to 3′ distribution patterns, exonic versus intronic signal, and dependence on gene expression. Transcripts with minimum 50,000 bp and 6 exons were used. Aggregate of H3K36me signals at the first three and last three exons is shown. Expression quantiles were computed based on a normalized average of parental replicates. Transcripts in expression quantile 4 had the highest expression, quantile 1 with the lowest expression greater than zero and quantile 0 with zero counts. Three wildtype replicates (n = 3) were merged per each methylation state. For a, d, and e, ChIP-seq signals were MS normalized and represent mean local frequency of the relevant modification. Normalization factors were computed by multiplying MS genome-wide modification percentage values (averaged per condition) by total number of bins and dividing by total signal for a given viewer track. This normalization factor was multiplied to the depth-normalized signal (CPM) for each merged track to generate MS-normalized tracks