Lysine methyltransferase SETD6 modifies histones on a glycine-lysine motif

ABSTRACT Although central to regulating the access to genetic information, most lysine methyltransferases remain poorly characterised relative to other family of enzymes. Herein, I report new substrates for the lysine methyltransferase SETD6. Based on the SETD6-catalysed site on the histone variant H2AZ, I identified similar sequences in the canonical histones H2A, H3, and H4 that are modified by SETD6 in vitro, and putative non-histone substrates. I herein expend the repertoire of substrates for methylation by SETD6.

The methyltransferase SETD6 mono-methylates the NFκB subunit RelA at lysine 310 (K310 me1 ) [14], the histone variant H2AZ at lysine 7 (K7 me1 ) [15], and the kinases PAK4 [16], and PLK1 [17]. The expression of SETD6 is amplified in about 10% of cases of breast cancer according to a study using a patient xenograft model [18] and is required for cellular proliferation in both ER + and ER − breast cancer cell models [19], suggesting an important role in driving breast cancer progression. Indeed, SETD6 was recently found to associate with the cytoskeleton protein VIM [20], which is involved in epithelial to mesenchymal transition (EMT), cellular attachment, migration, and signalling, suggesting a role in metastasis.
Much like classical signal transduction events involve phospho-dependent protein-protein binding, chromatin signalling events implicate posttranslational modifications in the regulation of macromolecules interactions. For example, lysine methylation of histones lysine residues serves as landing pads for chromatin proteins, which are referred to as histone mark readers or simply readers, thereby nucleating enzymatic complexes that modify and remodel chromatin to regulate access to genetic information.
Herein, I demonstrate that recombinant SETD6 methylates canonical histones H2A, H3, and H4, as well as linker histones H1 and the non-histone protein ING2 in vitro, and identify several putative novel substrates, including chromatin proteins and other lysine methyltransferases.

Results
I previously identified 2 mono-methylation sites on the histone variant H2AZ catalysed by SETD6 [15]. Interestingly, these modified sites, H2AZK4 me1 and H2AZK7 me1 , are similar. Both modified lysine residues are preceded by a small CONTACT Olivier Binda olivier.binda@mail.mcgill.ca Newcastle Cancer Centre at the Northern Institute for Cancer Research, Newcastle University, Newcastle upon Tyne, England This article has been republished with minor changes. These changes do not impact the academic content of the article.
Supplemental data for this article can be accessed here.
amino acid (alanine or glycine) residue at position −2 and a glycine at position −1 ( Table 1). Examination of canonical histone tails revealed similar sequences in histones H2A, H3, and H4, and highlighted a putative SETD6 methylation consensus motif A/G/RGK me1 A/GG (Table 1).
To test whether SETD6 could methylate these other histones, I used a mixture of purified calf thymus histones as substrates. Interestingly, SETD6 was capable of modifying the linker H1 histones as well as the canonical histones H2A, H3, and H4 ( Figure 1). As a positive control, I used SET7, which is known to methylate H3 [21], and H1 histones [22]. As a negative control, GST alone was used and as expected GST had no detectable methyltransferase activity on histones ( Figure 1).
To confirm the SETD6-catalysed methylation sites on canonical histones, the first 50 amino acid residues of H2A, H3, and H4 were fused to the amino terminus of GST to leave the histone tail free at the amino terminus and generate H2A 1-50 -GST, H3 1-50 -GST, and H4 1-50 -GST. Then the predicted sites (Table 1) were converted to arginine by site-directed mutagenesis. The affinity purified recombinant proteins were then used for in vitro KMT assays with SETD6. In agreement with previous experiments showing that SETD6 modifies canonical histones (Figure 1), SETD6 methylated recombinant histone tails from H2A, H3, and H4 (Figure 2(a)). Importantly, mutation of the GK motifs reduced drastically the methylation of H2A and H3 by SETD6 (Figure 2(a)). However, single mutation of H4 at K5 or K12 only minimally reduced methylation by SETD6 (Figure 2(a)), suggesting that both K5 and K12 are modified.  Since H4 methylation seemed stronger (Figure 1), and to further investigate the methylation of H4 by SETD6, methyltransferase assays on H4 peptides H4 1-20 and H4 10-31 were performed and confirmed that SETD6 methylates H4 (Figure 2(b)). Interestingly, SETD6 methylated H4 10-31 better than the H4 1-20 peptide, while methylation at K20 (K20 me1 , K20 me2 , or K20 me3 ) impaired this effect (Figure 2(b)), suggesting that SETD6 modifies H4K20 in addition to H4K12 or other site(s), such as H4K16 or H4K31 (Figure 2(c)). Alternatively, these results may suggest that there is a cross-talk between H4K20 me and the SETD6-catalysed methylation site(s).
Several KMT, including SETD6, modify nonhistone proteins. I thus searched for the GKDS motif in protein sequence repositories and identified several putative SETD6 substrates (Table 2 and S1), including the ATPase RUVBL1, which is modified by the H3K9 methyltransferases G9A and GLP [12]. Importantly, some putative SETD6 substrates (AHNAK2, ERICH3, and MDN1) were found in the PhosphoSitePlus mass spectrometric database to be methylated at the predicted site [23]  (Table S1). A similar search using the H4K5 and H4K12 motif GKGG also yielded several putative substrates for SETD6, such as the chromatin remodeller BRG1 (GK1029GG) and the HBO1 acetyltransferase subunit JADE2 (GK637GG).

Discussion
Proteomic studies have identified posttranslational modifications on histones and nonhistone proteins, but the enzymatic activities depositing these modifications remain largely unknown. There is a dire need to identify PTMs to understand how proteins are regulated, but more importantly to identify the enzymes catalysing these biochemical events. To this end, I have in the past designed an unbiased chemical-biology approach to tag novel KMT substrates [24]. However, traditional biochemical studies are still required to investigate and validate novel posttranslational modifications.
The H3K14 me1 mark was reported to occur in both human and mouse [25,26], supporting the existence of the modification in cells. I have herein identified the first KMT capable of modifying this site in vitro. Further work will be required to validate the role of SETD6 in the catalysis of H3K14 me1 in cells and the function of this mark.
Structurally, the GK motif on AHR is not within a particular functional domain, while the putative methylation sites on CDKN2C and KDM5A are found at the very end of their unstructured carboxy terminus, while the GK motif of SMYD4 is found near the amino terminus. These observations lead to conclude that the GK motifs found on the putative SETD6 non-histone substrates could be available for modification.
Together, the in vitro data provided here identifies SETD6 as a likely candidate for the methylation of reported events on H2AK5 [30], H3K14 [25,26], H4K5 [28,31], H4K12, H4K16, and H4K31. In addition, based on sequence similarities I have identified several putative non-histone protein substrates for SETD6.

Plasmids
The modified pGEX plasmid with an engineered multi-cloning site at the N-terminus of the GST coding sequence was described previously [15]. The cDNA of histones H2A, H3, and H4 was amplified by PCR from reverse transcribed total RNA and inserted in frame with GST using restriction endonucleases and T4 DNA ligase (NEB).

Recombinant protein expression and purification
Essentially, BL21 DE3 competent bacteria (Stratagene) were transformed with pGEX plasmids. Single colonies were picked and grown in 2YT media. Expression of GST-fusion proteins was induced with 0.01 mM IPTG for 2.5 hours at 37°C, cells were collected and lysed in buffer (50 mM Tris-Cl pH 7.5, 150 mM NaCl, 0.05% NP-40). Recombinant protein were batch purified using glutathione-sepharose beads. GST-SETD6 was purified similarly, but from Sf9 insect cells as described [15].

Motif search
The GKDS sequence was used in a motif search using PHI-BLAST against the Non-redundant protein sequences (nr) database, restricted to Homo sapiens (taxid:9606).