Uncovering global SUMOylation signaling networks in a site-specific manner

Abstract

SUMOylation is a reversible post-translational modification essential for genome stability. Using high-resolution MS, we have studied global SUMOylation in human cells in a site-specific manner, identifying a total of >4,300 SUMOylation sites in >1,600 proteins. To our knowledge, this is the first time that >1,000 SUMOylation sites have been identified under standard growth conditions. We quantitatively studied SUMOylation dynamics in response to SUMO protease inhibition, proteasome inhibition and heat shock. Many SUMOylated lysines have previously been reported to be ubiquitinated, acetylated or methylated, thus indicating cross-talk between SUMO and other post-translational modifications. We identified 70 phosphorylation and four acetylation events in proximity to SUMOylation sites, and we provide evidence for acetylation-dependent SUMOylation of endogenous histone H3. SUMOylation regulates target proteins involved in all nuclear processes including transcription, DNA repair, chromatin remodeling, precursor-mRNA splicing and ribosome assembly.

Fig. 1

A strategy for mapping SUMO-2 acceptor lysines in endogenous proteins. (a) Immunoblot confirming the low level expression of H10-S2-K0 in HeLa cells. Ponceau-S staining is shown as a loading control. Additionally, His10-pulldown was performed to enrich SUMOylated proteins, and Ponceau-S is shown to indicate high-specificity enrichment. The experiment shown was replicated in biological duplicate. (b) Confocal fluorescence microscopy image confirming the predominantly nuclear localization of H10-S2-K0. GFP; Green Fluorescent Protein. DIC; Differential Interference Contrast. Scale bars represent 25 μm. The experiment shown was replicated in biological duplicate. (c) Coomassie stain displaying the efficacy of a single-step His10-pulldown performed on approximately 50 million HeLa and H10-S2-K0 cells. The experiment shown was replicated in biological duplicate. (d) Schematic overview of the H10-S2-K0 SUMOylation site purification strategy. A direct purification step is followed by concentration of SUMOylated proteins, which are subsequently digested with endopeptidase Lys-C. The H10-S2-K0 bearing the SUMOylated peptide is re-purified, concentrated, digested with trypsin, and finally analyzed by high-resolution nanoscale LC-MS/MS. (e) Immunoblotting analysis was used to confirm the efficiency of the purification steps described in (d). The experiment shown was replicated in biological duplicate. (f) Immunoblotting analysis of total lysates from cells stably expressing H10-S2-K0 which were mock treated, or treated with MG-132, PR-619 or heat shock. Ponceau-S staining is shown as a loading control. The experiment shown was replicated in biological duplicate.

Fig. 2

Overview of mass spectrometry results. (a) Overview of the amount of SUMOylated peptides identified with their respective Andromeda peptide scores. In the range of 400 to 1200 SUMOylation sites were identified from single runs depending on treatment conditions. The control experiments were performed in biological hexuplicate, PR-619 experiments were carried out in biological quintuplicate, and MG-132 and heat shock experiments were performed in biological triplicate. Experiments were measured in technical duplicate, with one 15-cm plate (20 million cells) serving as input for a single run. (b) Schematic representation of the amount of SUMOylation sites discovered in relation to the cellular treatments used. Over 4,300 SUMOylation sites were identified in total. (c) Schematic representation of the amount of SUMOylated proteins discovered in relation to the cellular treatments used. Over 1,600 SUMOylated proteins were identified in total. (d) Overview of the SUMOylated peptides identified under control conditions, as compared to the SUMOylated peptides exclusively discovered after cellular treatments. The absolute and relative amount of SUMOylation sites are plotted against the peptide intensity. (e) Overview of the amount of SUMOylation sites identified per protein.

Fig. 3

SUMO is extensively involved in PTM crosstalk. (a) Schematic representation of the overlap of the identified SUMOylated lysines as compared to the other lysine post-translational modifications ubiquitylation, acetylation and lysine-methylation. Enrichment ratio between observed overlaps and expected overlaps are indicated, and were significant by Fisher Exact Testing with P < 1E-10. (b) Similar to (a), overlap between SUMOylation, ubiquitylation, acetylation and lysine-methylation. (c) Schematic overview of phosphorylation sites adjacent to SUMOylated lysines, both identified by mass spectrometry in our screen, and their amino acid spacing in relation to the SUMOylated lysine. Some peptides were exclusively found to be SUMOylated in combination with phosphorylation (blue). Non-unique phosphorylation sites on SUMOylated peptides were also discovered (red). (d) Schematic representation of the identified SUMOylation sites on ubiquitin. (e) Schematic representation of the Histone H3 peptide as identified by MS/MS, simultaneously modified by SUMOylation on lysine 19 (H3K18) and acetylation on lysine 24 (H3K23). Identified fragment ions are indicated. A fully annotated high resolution MS/MS spectrum is available as part of Supplementary Data Set 1. (f) Immunoblot analysis of total lysates and His10-pulldown samples from HeLa cells stably expressing His10-SUMO-2 wild-type or K0-mutant, which were either mock treated, treated with the histone de-acetylation inhibitor Trichostatin A (TSA) at the indicated dose in nM, or treated with the histone acetylation inhibitor curcumin (Cur) at the indicated dose in μM. Ponceau-S staining is shown as a loading control. The experiment shown was replicated in biological duplicate.

Fig. 4

Novel insight in the SUMOylation consensus motif. (a) IceLogo of all SUMOylation sites identified under control conditions. The height of the amino acid letters corresponds to fold-change. All amino acid changes were significant with P < 0.05 by two-tailed Student’s t test. (b) SubLogos of various consensus motifs. The height of amino acid letters represents the %-change enrichment or depletion of the motif set as compared to the reference set. All amino acid changes were significant with P < 0.05 by two-tailed Student’s t test. (c) As (a), but in heatmap format. Green is indicative of a statistical enrichment as compared to randomly expected, and red is indicative of a depletion. All amino acid changes were significant with P < 0.05 by two-tailed Student’s t test. (d) Overview of the amount of SUMOylation sites matching the short consensus motif K×E in different subsets of sites corresponding to different cellular treatments. Additionally, per subset, the top 25% intense, the top 50% intense, or all sites are shown. Matching of ubiquitin sites to the motif and the randomly expected frequency are also shown. (e) As (d), but for the short consensus motif [IVML]K. (f) Schematic representation of the cysteine frequency close to all identified control SUMOylation sites, as well as other PTMs, ranging from −10 to +10 amino acids of the modified lysine. For all PTMs, a 2^nd order polynomial trend line was calculated. The background cysteine frequency is indicated.

Fig. 5

SUMOylation is a predominantly nuclear event, and is involved in many biological processes and protein complexes. (a) Overview of protein domain families overrepresented or underrepresented in all identified SUMOylated proteins, ranked by −log₁₀ P-value. SUMOylated proteins as compared to total proteins are indicated. (b) Overview of the amount of SUMOylated proteins and sites in relation to their subcellular localization. Gene Ontology Cellular Compartments (GOCC) enrichment ratios are indicated above categories, and average SUMO sites per protein are indicated below. (c) Overview of the predicted secondary structure of different subsets of SUMOylated lysines. The random set corresponds to lysines randomly selected from SUMOylated proteins. Differences indicated with an asterisk (*) were significant with P < 0.001. Error bars represent s.e.m. and are based on structural predictions on a per-site basis. For “SUMO” n=4,361. For “Random” n=5,725. For “Control” n=1,070. For “Treatments” n=3,291. For “K×E” n=1,300. For “Non-K×E” n=3,061. (d) Overview of the predicted solvent exposure of different subsets of SUMOylated lysines, including either −5 to +5 amino acids, or −15 to +15 amino acids. Significance and error bars are synonymous to (c). (e) All identified SUMOylated proteins were annotated with Gene Ontology Biological Processes terms, and compared against the annotated human proteome. Categories were scored by a combination of enrichment ratio and P-value. The amount of hits as compared to the category size is indicated. (f) As (e), but for Gene Ontology Molecular Functions. (g) As (e), but for CORUM complexes. (h) As (e), but for Keywords.

Fig. 6

SUMO modifies highly interconnected functional networks of proteins. (a) STRING network analysis of all identified SUMOylated proteins, with a STRING interaction confidence of 0.7 or greater. MCODE was used to extract the most highly interconnected functional clusters from the network, which are indicated in different colors. (b) Overview of relative STRING network score corresponding to Table 1. This score was computed through multiplication of the interaction enrichment ratio, protein network connectivity, and the average STRING confidence of all interactions. (c) Schematic overview of the three highest scoring MCODE sub-clusters from (a). The size and color of the individual proteins corresponds to the amount of SUMOylation sites identified in the protein. The six additional MCODE clusters are available as Supplementary Fig. 7.