A functional proteomics platform to reveal the sequence determinants of lysine methyltransferase substrate selectivity

Abstract

Lysine methylation is a key regulator of histone protein function. Beyond histones, few connections have been made to the enzymes responsible for the deposition of these posttranslational modifications. Here, we debut a high-throughput functional proteomics platform that maps the sequence determinants of lysine methyltransferase (KMT) substrate selectivity without a priori knowledge of a substrate or target proteome. We demonstrate the predictive power of this approach for identifying KMT substrates, generating scaffolds for inhibitor design, and predicting the impact of missense mutations on lysine methylation signaling. By comparing KMT selectivity profiles to available lysine methylome datasets, we reveal a disconnect between preferred KMT substrates and the ability to detect these motifs using standard mass spectrometry pipelines. Collectively, our studies validate the use of this platform for guiding the study of lysine methylation signaling and suggest that substantial gaps exist in proteome-wide curation of lysine methylomes.

Fig. 1. K-OPL platform for mapping KMT substrate selectivity.

(A) Composition and design of the K-OPL. (B) Cartoon depiction of the SPA developed for screening the activities of KMTs with K-OPL. TEG, triethylene glycol; SAM, S-adenosylmethionine.

Fig. 2. K-OPL reveals the substrate selectivity of G9a, SET7/9, and SMYD2.

K-OPL substrate selectivity profiles for G9a (A), SET7/9 (B), and SMYD2 (C). Mean results of two independent K-OPL SPA screens for each enzyme are reported as position-normalized heat maps (see fig. S1 for global normalized heat maps and raw K-OPL data). The color code is proportional to the creation of enzyme product, where red (1) is most active and blue (0) is least active. Rows show the identity of each fixed residue, and columns show the position within the sequence. Initial rate measurements with peptides corresponding to known and newly identified substrates for G9a (D), SET7/9 (E), or SMYD2 (F). cpm, counts per minute. Point mutations predicted to decrease or increase the rate of methylation are indicated in red or green, respectively. Data points are shown as the mean of three independent measurements, and error is presented as ±SEM. For some data points, error bars are masked by the symbol weight.

Fig. 3. MS analysis of methylation products.

The products from reactions of G9a (A), SET7/9 (B), and SMYD2 (C) with their corresponding peptide substrates were analyzed by MS. Mass spectra are shown in the absence (top) or presence (bottom) of enzyme treatment, as indicated.

Fig. 4. Structural and kinetic analysis of SMYD2.

(A) Hybrid ribbon-surface representation of SMYD2 (white) bound to SAH and GWKLNleSKRG (Nle, norleucine) (blue sticks). Costructure has been deposited in the Protein Data Bank (PDB) as PDB: 6MON. (B) Overlay of peptide substrates from SMYD2-GWKLNleSKRG (PDB: 6MON) and SMYD2-p53K370 (PDB: 3TG5) structures. (C) Scatterplot comparing the relationship between the methylation rate calculated from Fig. 2F and the dynamics of substrate coordination by SMYD2. Root mean square displacement (RMSD) of the Cα atoms of the indicated peptides was calculated from 500-ns whole-atom MD simulations. Peptide orientations at several time points over the course of the MD simulations are shown, indicated by color with the corresponding color scale (top left). (D) Kinetic analysis of SMYD2 methylation of p53, PER2, and PER2 derivative substrates. Data points are the mean of three independent measurements, and error is presented as ±SEM. PER2 and WKLKSKR were fit to a substrate inhibition kinetic model. DKLKSKR and SHLKSKK were fit to a standard Michaelis-Menten (MM) model. (E) IC₅₀ (median inhibitory concentration) measurements of Nle peptide inhibitors of SMYD2 using PER2 as a substrate. Data points are the mean of three independent measurements, and error is presented as ±SEM.

Fig. 5. Novel SMYD2 substrates identified with K-OPL.

(A) Representative in vitro SMYD2 methyltransferase assay with known and predicted protein substrates. K to R refers to a missense mutation (lysine to arginine) introduced at the target lysine. Coomassie-stained gel is shown on top in blue, and ³H fluorography is shown on the bottom. (B) Representative in vitro SMYD2 methyltransferase assay with mutant forms of PRDM11 and MAPKAPK3 substrates predicted to decrease or increase substrate efficiency, respectively. WT, wild-type. (C) Scatterplot of the LoB score for SMYD2 methylation motifs that are created (blue), weakened (red), strengthened (green), or unchanged (black) by missense mutations found in primary human cancer sequencing data.

Fig. 6. K-OPL analysis reveals a gap in MS-based lysine methylation datasets.

(A) Comparison of the arginine and lysine content between the top 50 K-OPL–predicted SMYD2 substrates (blue) and the 35 substrates identified using MS (red) (Olsen et al.). (B) Lysine and arginine content in the entire lysine methylome as curated by PhosphoSitePlus (8).