Human SETMAR is a DNA sequence-specific histone-methylase with a broad effect on the transcriptome

Abstract

Transposons impart dynamism to the genomes they inhabit and their movements frequently rewire the control of nearby genes. Occasionally, their proteins are domesticated when they evolve a new function. SETMAR is a protein methylase with a sequence-specific DNA binding domain. It began to evolve about 50 million years ago when an Hsmar1 transposon integrated downstream of a SET-domain methylase gene. Here we show that the DNA-binding domain of the transposase targets the enzyme to transposon-end remnants and that this is capable of regulating gene expression, dependent on the methylase activity. When SETMAR was modestly overexpressed in human cells, almost 1500 genes changed expression by more than 2-fold (65% up- and 35% down-regulated). These genes were enriched for the KEGG Pathways in Cancer and include several transcription factors important for development and differentiation. Expression of a similar level of a methylase-deficient SETMAR changed the expression of many fewer genes, 77% of which were down-regulated with no significant enrichment of KEGG Pathways. Our data is consistent with a model in which SETMAR is part of an anthropoid primate-specific regulatory network centered on the subset of genes containing a transposon end.

Figure 1.

Hsmar1 remnants in the human genome and expression of ectopic SETMAR. (A) The SETMAR domain and exon structure are illustrated together with the helix-turn-helix (HTH) DNA binding motifs and key active site residues in the methylase and transposase active sites. The third D residue (aspartate) that coordinates the catalytic metal ion is an N (asparagine) in SETMAR. All 260 full-length copies of Hsmar1 have inactivating point mutations and indels. Made1 elements comprise of six bp flanked by a pair of ITRs. (B and C) Distribution of the 6,334 ITRs in the human genome. (D) A Western blot for the FLAG-tagged codon optimized SETMAR in the U2OS, SMF and SMFN cell lines. SMFN has the N210A substitution of an essential residue in the SETMAR methylase domain active site. (E) qRT-PCR of the endogenous and the transgenic SETMAR in the SMF and SMFN cell lines. Figures above rightmost columns are the ratios of expression derived from the RNA-seq experiments presented in Figure 3. (F) Western blot of the indicated cell lines using an H3K36me2 antibody and a histone H3 antibody as a loading control. The lanes probed for H3 and H3K36me2, respectively, were from a single gel loaded with the same amount of protein. The same result was obtained from three biological replicates (R1 to R3) performed on different batches of cells.

Figure 2.

SETMAR binds Hsmar1 ITRs and other sequences in vivo. (A) SETMAR binding in vivo was assessed by ChIP-exo and enriched motifs were identified with the MEME-ChIP software. The three most enriched motifs are presented: Motif 1 corresponds to the SETMAR binding site on the Hsmar1 ITR; Motif 2 is a degenerate ITR sequence associated with a subset of Made1 elements; Motif 3 resembles an ITR and is enriched in centromeric regions of the genome. CENP-B box and the consensus Hsmar1 ITR are shown below. (B) Distribution of the ChIP-exo peaks with respect to annotated Hsmar1 ITRs. (C) Metaprofile of ChIP-exo reads around the center of the 419 bound ITRs. (D) MNase-seq nucleosome metaprofiles around annotate Hsmar1 ITRs were generated from the ENCODE data for U2OS, a lymphoblastoid cell line (GM18508), and two human embryonic stem cell lines (H1 and H9). (E and F) Distribution of SETMAR ChIP-exo peaks and Motif 1 with respect to annotated protein coding genes. Peak were called with MACS2.

Figure 3.

SETMAR regulates gene expression and is dependent on its methyltransferase activity. (A and B) Smear plots of RNA-seq data showing average gene expression versus log2-fold change in gene expression. Fold-change and p values are color coded as indicated. The number of genes up-regulated or down-regulated are indicated at the top and bottom of each graph, respectively. The number of genes with an ITR differentially expressed is indicated between brackets. P-value were determined using a hypergeometric distribution. SETMAR expression is 1372 making it the 9098th most express gene out of 16 776 genes. (C–E) Venn diagrams for the genes significantly differentially expressed more than ±2-fold from parts A and B.

Figure 4.

Distribution of SETMAR ChiP peaks and mode of action. (A) The distribution of intragenic ChIP-exo peaks with respect to the presence of intronic ITRs. 42% of genes have a ChIP-exo peak within 500 bp of an ITR. 12% of genes have a ChIP-exo peak >500 bp distant. 46% of genes with a ChIP-exo peak have no intronic copies of the ITR. P-values were determined using a hypergeometric distribution. (B) The distribution of intronic SETMAR-bound copies of the ITR with respect to non-coding and coding genes and the degree of differential expression in the SMF and SMFN cell lines. A third of the genes have their ITR bound in the ChIP-exo experiments are significantly differentially expressed in the SMF cell line. A third of the non-differentially expressed genes are non-coding and therefore have not been detected by the RNA-seq protocol used. DE, differentially expressed. (C) Boxplot showing the expression level of the 97 protein coding genes with a bound ITR (ChIP-exo peak <500 bp distant) compared to the vast majority of genes that have no ITR and no ChIP-exo peak. The rightmost three plots are subclasses of the 97 genes with a bound ITR. P-values were determined using a Mann-Whitney U test. DE, differentially expressed; Dn., down; reg., regulated. (D) Metaprofiles for the presence of H3K9ac marked-nucleosomes were generated for the 50 differentially expressed (DE) and 47 non-DE genes from part C using the ENCODE data for U2OS cells. TSS, transcriptional start site. (E) We present a model for transcriptional regulation by SETMAR. A gene is undergoing a moderate amount of transcription. An intronic ITR has positioned nucleosomes on either side owing either to SETMAR binding or the intrinsic positioning effect of the ITR sequence itself. SETMAR methylates nearby nucleosomes and/or RNA polymerase II associated factors such as snRNP70. Up-regulation of gene expression by SETMAR is dependent on its methyltransferase activity and is mediated by an increase in RNA polymerase II transcription and by a possible change in mRNA stability. Blue hexagons, methyl groups.