Profiling the epigenome at home

Steven Henikoff; Jorja G. Henikoff

A comparison between H3K27me3 and H3K9me3 CUT&Tag tracks suggests that these two silencing marks are non-overlapping (Fig. 6) . Interestingly, in the highlighted example, a strong peak of H3K9me3 is seen to correspond to an ERV-1 family Figure 3 were sequenced and tracks for a representative region were compared to an ENCODE dataset (GSM788088), to datasets produced using the original extraction protocol, and this single-tube protocol performed in the lab. Asterisks indicate CUT&Tag@home datasets produced using a commercial pAG-Tn5 preparation (Epicypher cat. no. 15-1017) . b) Same as (a) for H3K4me3 comparing results from CUT&Tag@home to those produced using the single-tube protocol in the lab and an ENCODE dataset (GSM733680). Tracks are autoscaled for clarity, except for the IgG negative control tracks, which were scaled the same as that for the 60-cell CUT&Tag@home sample. The copyright holder for this preprint (which was not peer-reviewed) is the . https: //doi.org/10.1101 //doi.org/10. /2020 endogenous retrovirus (ERV). To ascertain the generality of the observation, we extracted 20-kb segments centered over the middle of each ERV from the H3 lysine trimethylation tracks over Chromosome 1, which harbors 50,707 of the 695,067 human ERVs, from the UCSC Repeat Masker file. We stacked the segments ordered by decreasing length of the ERV (Fig. 7b) . For H3K4me3, H3K36me3 and H3K27me3 we observed essentially no signal over these elements, whereas for H3K9me3 we observed signal over elements of all sizes, including a cluster of heavily H3K9me3-marked ERV fragments spanning ~2 kb in length. This observation is consistent with studies showing that intact and active ERVs are among the most heavily H3K9me3-methylated elements in mammalian genomes (Bulut-Karslioglu et al., 2014; Ohtani et al., 2018; Walter et al., 2016) . To determine whether there are any differences among the ERV families in the propensity for gaining H3K9 trimethylation, we performed unsupervised k-means clustering over a ±1 kb span on the full set of Chromosome 1 ERVs (k=3). Cluster I comprises 5,536 heavily H3K9trimethylated ERVs (11%) and Cluster III comprises 31,177 ERVs (62%) with background levels of H3K9 trimethylation (Fig 7b, rightmost panel) . Among the ERV families, ERVK, which accounts for 8% of the total human ERVs, was on average 2.8-fold more highly represented in Cluster I relative to Cluster III (186:382) than were all other ERVs (5350:30795). The ERVK family is the youngest and most intact of the human endogenous retrovirus families (Hanke et al., 2016) , and in the mouse male germline, members of the ERVK family were specifically reactivated upon loss of a germline- Figure 6 : CUT&Tag@home segments the human epigenome with four H3 lysine trimethylation marks: Screenshot of a representative 1 Mb region of the human genome showing CUT&Tag@home profiles for histone H3 lysine-4 methylation of constitutive heterochromatin (K9me3), Polycomb-dependent silenced domains (K27me3) transcribed gene bodies (K36me3), promoters (K4me3), promoters and enhancers (K4me2) and accessible DNA (ATAC-seq, GSM269550). Two regions are expanded to illustrate the predominantly mutual exclusivity of the tri-methylation marks, also showing two SuRE autonomous regulatory elements annotated as an enhancer (above) and promoter (below). An ERV-1 retrotransposon is heavily marked by H3K9me3. The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.