Dynamic In Vivo Mapping of the Methylproteome Using a Chemoenzymatic Approach

Abstract

Dynamic protein post-translational methylation is essential for cellular function, highlighted by the essential role of methylation in transcriptional regulation and its aberrant dysregulation in diseases, including cancer. This underscores the importance of cataloging the cellular methylproteome. However, comprehensive analysis of the methylproteome remains elusive due to limitations in current enrichment and analysis pipelines. Here, we employ an l-methionine analogue, ProSeMet, that is chemoenzymatically converted to the SAM analogue ProSeAM in cells and in vivo to tag proteins with a biorthogonal alkyne that can be directly detected via liquid chromatography and tandem mass spectrometry (LC-MS/MS), or functionalized for subsequent selective enrichment and LC-MS/MS identification. Without enrichment, we identify known and novel lysine mono-, di-, and tripargylation, histidine propargylation, and arginine propargylation with site-specific resolution on proteins including heat shock protein HSPA8, the translational elongation factor eEF1A1, and the metabolic enzyme phosphoglycerate mutase 1, or PGAM1, for which methylation has been implicated in human disease. With enrichment, we identify 486 proteins known to be methylated and 221 proteins with novel propargylation sites encompassing diverse cellular functions. Systemic ProSeMet delivery in mice propargylates proteins across organ systems with blood-brain barrier penetrance and identifies site-specific propargylation in vivo with LC-MS/MS. Leveraging these pipelines to define the cellular methylproteome may have broad applications for understanding the methylproteome in the context of disease.

Figure 1

Chemoenzymatic approach for metabolic Methyltransferase labeling. a. ProSeMet can be converted to ProSeAM by MAT enzymes in live cells. ProSeAM can then be used by diverse methyltransferases to propargylate target protein. b. (top panel) Conversion of unmethylated lysine to mono, di, and trimethyl-Lys through the action of SAM. (bottom panel) The guanidine moiety on Arg can undergo monomethylation, then symmetric or asymmetric dimethylation. The imidazole ring of His can undergo monomethylation at the C1 or C3 position c. (left panel) Conversion of unpropargylated Lys to mono, di, and tripropargyl lysine could occur through the intermediacy of ProSeAM. (right panel) Predicted Arg monopropargylation, symmetric dipropargylation, and asymmetric dipropargylation as well as His propargylation at the C1 and C3 positions.

Figure 2

ProSeMet propargylates nuclear and cytosolic proteins in live cells. a. Workflow for gel- and IF-based profiling of ProSeMet target proteins. Cells were starved of l-Met by incubating for 30 min in l-Met-free media. Cells were then lysed or fixed, subjected to CuAAC to attach a fluorophore, and separated via SDS-PAGE or imaged via confocal microscopy. b. T47D cells treated with 100 μM ProSeMet or 100 μM l-Met for 16 h were lysed and subjected to the click reaction to attach a fluorescent picolyl azide (680 nm). Increasing protein concentration (5–20 μg) was loaded into each well and separated by SDS-PAGE. (n = 3). c. T47D cells were treated with 200 μM l-Met or increasing concentration (25–200 μM) ProSeMet 16 h. Cellular lysates were subjected to CuAAC and separated via SDS-PAGE. (n = 3). d. T47D cells were treated with 100 μM l-Met, ProSeMet, or AHA in the presence or absence of 10 μg/mL of cycloheximide (CHX). After 16 h, lysates were collected and subjected to CuAAC to attach a fluorescent picolyl azide (l-Met, ProSeMet) or fluorescent alkyne (AHA) then separated by SDS-PAGE (n = 3). e. G401 cells were starved of l-Met by incubating for 30 min in l-Met Free media. Cells were then treated with 1 μM tazemetostat or DMSO for 72 h, after which histones were acid extracted and used for CuAAC to attach a fluorescent picolyl azide (680 nm). Resulting lysates were separated via SDS-PAGE and directly imaged or immunoblotted with the indicated antibodies. *, indicated protein, (n = 2). f. Competition by increasing concentrations of l-Met during 16 h incubation of T47D cells with 100 μM ProSeMet reduces ProSeMet labeling across molecular weights and in a dose-dependent manner (n = 3). g. Cell fractionation of T47D cells treated with 100 μM ProSeMet or 100 μM l-Met for 16 h. (n = 3). h. T47D, LNCaP, and MCF10A cells were treated with 100 μM ProSeMet or 100 μM l-Met 16 h then fixed, permeabilized, subjected to CuAAC to attach a fluorescent picolyl azide (568 nm, pseudocolored green), and counterstained with Hoechst (blue). (n = 3).

Figure 3

Protein propargylation is mapped with site-specific resolution. All experiments and analyses were performed with two biological replicates (n = 2). a. Approach schematic. G401 cells treated with 100 μM ProSeMet with 100 μM l-Met for 16, 24, 36, or 48 h were lysed and processed for LC-MS/MS. MS raw files were searched against the human Swiss-Prot database (20,456 entries), with variable mass shifts of (+38.0157 Da for monopropargyl, +76.0314 Da for dipropargyl, and +114.0471 Da for tripropargyl) on lysine, arginine, and histidine, with a maximum propargylation state of 3 on lysine, 2 on arginine, and 1 on histidine. Proteomics analysis identified a total of 376 peptide spectral matches (PSMs) corresponding to 149 total proteins. Of these, 123 (82.5%) unique propargylated proteins were defined in ProSeMet-treated samples and 27 PSMs (26 proteins, 17.5%) for l-Met. b. Total number of propargylated peptides identified via LC-MS/MS across all time points, compared to l-Met. Data are represented as mean+SD c. Density plot of peptide propargylation states over the indicated time course. d. Sequence motif analysis of propargylated lysine, arginine, and histidine residues. Sequences containing 5 residues from the left and 4 residues from the right of modified lysine and arginine sites were utilized, with lysine or arginine as the fixed positions where p < 0.05. For sequence motif analysis of histidine, sequences containing 4 residues from the left and 4 residues from the right of modified histidine sites were utilized. The sequence motif was generated using the “probability logo generator for biological sequence motif” pLogo v1.2.0 e. Representative proteins propargylated in response to ProSeMet incubation and corresponding mapped sites of propargylation. Green, lysine propargylation events; red, arginine propargylation events; blue, histidine propargylation events. f. Gene ontology (GO) and pathway-process enrichment analysis of propargylated proteins in response to ProSeMet treatment. Gene list of propargylated proteins were utilized as input in metascape, with input and analysis species set to Homo sapiens. Pathway and process enrichment analysis was carried out with the following ontology sources: KEGG pathway, GO biological processes, Reactome Gene Sets, Canonical pathways, CORUM, WikiPathways, and PANTHER pathway. All genes in the human genome were used as the enrichment background. Terms with p < 0.01, a minimum count of 3, and an enrichment factor >1.5 were utilized. p-values were calculated based on the cumulative hypergeometric distribution, and q-values are calculated using the Benjamini–Hochberg procedure.

Figure 4

ProSeMet-mediated chemoproteomics identifies known and novel propargylation events. a. Schematic of HSPA8 propargylation states defined from G401 cells treated with ProSeMet or l-Met for 16, 24, 36, or 48 h, using LC-MS/MS. Propargylation of HSPA8 K3, K384, and R469 were observed across time points. LC-MS/MS experiments and analyses were performed with two biological replicates (n = 2). Propargylation sites overlaid on PDB file 2V7Z. b. PSM distribution of propargylated HSPA8 sites with 8 PSMs observed for HSPA8 R469me1, over 3 PSMs for HSPA8 K3me1, and HSPA8 K384me3. Analyses were performed with two biological replicates (n = 2). c. Model MS2 peptide spectrum map of HSPA8 monopropargylated R469. d. HSPA8-V5 immunoprecipitation with lysate from HEK293T cells treated with ProSeMet or equimolar l-Met for 24 h, followed by CuAAC to attach a fluorophore azide. e. Quantification of (d). (n = 3), *, p ≤ 0.05. f. Schematic of PGAM1 propargylation states defined from G401 cells treated with ProSeMet or l-Met for 16, 24, 36, or 48 h, using LC-MS/MS. Propargylation of PGAM1 K222 was observed across time points. LC-MS/MS experiments and analyses were performed with two biological replicates (n = 2). Propargylation sites overlaid on PDB file 5Y2I. g. Model MS2 peptide spectrum map of LC-MS/MS validated PGAM1 K222 propargylation. Ectopic expression of a PGAM1-V5 vector in HEK293T cells, followed by treatment with l-Met or ProSeMet for 24 h, and subsequent cell lysis, V5 IP, and LC-MS/MS. h. Schematic of eEF1A1 propargylation states defined from G401 cells treated with ProSeMet or l-Met for 16, 24, 36, or 48 h, using LC-MS/MS. Propargylation of eEF1A1 K273me1, and K392me1 was observed across time points. LC-MS/MS experiments and analyses were performed with two biological replicates (n = 2). Propargylation sites overlaid on PDB file 1G7C. i. eEF1A1-mCherry immunoprecipitation with lysate from HEK293T cells treated with ProSeMet or equimolar l-Met for 24 h, followed by CuAAC to attach a fluorophore azide. j. Quantification of (i). (n = 3), *, p ≤ 0.05. k. Schematic of CALM1 propargylation states defined from G401 cells treated with ProSeMet or l-Met for 16, 24, 36, or 48 h, using LC-MS/MS. Propargylation of CALM R38me1 was observed across time points. LC-MS/MS experiments and analyses were performed with two biological replicates (n = 2). Propargylation sites overlaid on PDB file 6YNS. l. CALM1-mCherry immunoprecipitation with lysate from HEK293T cells treated with ProSeMet or equimolar l-Met for 24 h, followed by CuAAC to attach a fluorophore azide. m. Quantification of (l). (n = 2).

Figure 5

ProSeMet propargylates proteins in vivo. a. Schematic of the propargylation strategy in model organisms. b. Lysates extracted from A. thaliana root tips from 5-day-old seedlings treated with 200 μM or 1 mM ProSeMet or equimolar l-Met for 24 or 48 h. Lysates were subjected to CuAAC to attach a fluorescent picolyl azide (680 nm), (n = 2). c. Lysates extracted from C. elegans treated with 200 μM ProSeMet or equimolar l-Met for 4, 18, or 48 h. Lysates were subjected to CuAAC to attach a fluorescent picolyl azide (680 nm), (n = 2). d. Lysates extracted from S. cerevisiae starved for 30 min in Met, Cys-media followed by treatment with 10 μM ProSeMet or equimolar l-Met for 5 h. Lysates were subjected to CuAAC to attach a fluorescent picolyl azide (680 nm), (n = 2). e. Workflow for in vivo administration of ProSeMet or l-Met and subsequent analysis of organs for propargylation. f. Organs extracted from mice treated with 15 mg of ProSeMet or equimolar l-Met via IP injection while being fed or starved for 12 h were fixed in neutral buffered formalin and paraffin embedded. Tissue sections were subjected to CuAAC to attach a fluorescent picolyl azide (568 nm, pseudocolored green) and counterstained with DAPI (blue). In-tissue fluorescence analysis of the lung, brain, heart, kidneys, and intestine demonstrates successful blood–brain barrier penetrance of ProSeMet, as well as pan-organ labeling. Scalebar represents 50 μM (n ≥ 3). g. Quantification of in-tissue fluorescence in the brain, heart, and lungs (n ≥ 3). h. Immunoblot densitometry normalized to β-actin and relative to the l-Met control (n ≥ 3). All tissues show increased protein labeling compared to l-Met; no significant difference was observed between mice that were fed or starved prior to ProSeMet administration. *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001.

Figure 6

In vivo propargylation is defined with site-specific resolution. Experiments and analyses were performed with two biological replicates (n = 2). a. Proteomics analyses of perfused brain, heart and lung tissues from mice treated with l-Met or ProSeMet. Peptide spectral matches (PSMs) of identified propargylated amino acids in ProSeMet fed and starved samples when compared to l-Met control. All samples were filtered for contaminant hemoglobin PSMs. b. Gene ontology (GO) and pathway-process enrichment analysis of propargylated proteins. MS raw files were searched against the M. musculus Swiss-Prot database, with variable mass shifts of (+38.0157 Da for monopropargyl, +76.0314 Da for dipropargyl, and +114.0471 Da for tripropargyl) on lysine, arginine, and histidine, with a maximum propargylation state of 3 on lysine, 2 on arginine, and 1 on histidine. MF, BP, and CC are GO subontologies. GO-MF (Molecular Function, red), GO-BP (Biological Processes, yellow), GO–CC (Cellular Compartment, green), KEGG (Kyoto Encyclopedia of Genes and Genomes, pink), REACTOME (blue), and WikiPathways (cyan). Dots represent protein clusters (terms) associated with GO subontologies and circle size corresponds to term sizes. Gene list of propargylated proteins was utilized as input in metascape and gProfiler, with input and analysis species set to M. musculus. Pathway and process enrichment analysis was carried out with the following ontology sources: KEGG pathway, GO biological processes, reactome gene sets, canonical pathways, CORUM, WikiPathways, and PANTHER pathway. All genes in M. musculus genome were used as an enrichment background. Terms with p < 0.01, a minimum count of 3, and an enrichment factor >1.5 were utilized. p-values were calculated based on the cumulative hypergeometric distribution, and q-values were calculated using the Benjamini–Hochberg procedure.