Figure 2

Parallel Functional Annotation (PFA) of cancer-associated histone methyltransferase variants of uncertain significance (VUS). (A) Venn diagram of FATHMM and PolyPhen-2 functional prediction of Mixed Lineage Leukemia (MLL) VUS: 5 of 99 mutations had overlapping predictions. (B) Top, catalytic SET domain secondary structure map from PDBsum based on PDB 5F59, amino acids 4754–4911. Shown are alpha helices (H1-3), beta sheets (β1-10) beta hairpin sturns (red coloured hairpin sturns), and ligand/metal binding residues: H3/SAH binding (red filled square), SAH binding (blue filled triangle), zinc ion binding (blue filled square/green filled triangle). Bottom, representative PFA of MLL3 VUS mutations by scintillation counting. Quenching was after 30 min with data normalized against WT. Pink and purple, assays initiated with H3K4me0, H3K4me1. Dashed lines and corresponding shaded regions, average and standard deviation (1σ), respectively, for all variants with activity > 50% of WT. Error bars, standard deviation from 2 independent experiments. (C) Representative results from PFA for MLL3 VUS mutations by fluorography of SDS-PAGE. Upper, Coomassie-stained gel of quenched enzymatic reactions; middle, signal from reactions with H3K4me0 (unmethylated) or H3K4me1 (monomethylated) peptides; bottom, expression of MLL3 variants by Coomassie-stained SDS-PAGE. Assays were as described for Fig. 1, limiting the recombinant subunits required for full enzymatic activity^31,32,33 to minimize activity variation from differing MLL expression. Rates of monomethylation and dimethylation were determined using unmodified or monomethylated substrates. Activity depended on recombinant expression (no activity in uninduced control, UIC, lane 1). Lanes 2–11 show representative wild-type (WT) and variant MLL3 complexes, demonstrating that activity variation cannot be explained by differential expression. An uncropped version of Fig. 2C is shown in Fig. S11.