Global Identification of Small Ubiquitin-related Modifier (SUMO) Substrates Reveals Crosstalk between SUMOylation and Phosphorylation Promotes Cell Migration

Proteomics studies have revealed that SUMOylation is a widely used post-translational modification (PTM) in eukaryotes. However, how SUMO E1/2/3 complexes use different SUMO isoforms and recognize substrates remains largely unknown. Using a human proteome microarray-based activity screen, we identified over 2500 proteins that undergo SUMO E3-dependent SUMOylation. We next constructed a SUMO isoform- and E3 ligase-dependent enzyme-substrate relationship network. Protein kinases were significantly enriched among SUMOylation substrates, suggesting crosstalk between phosphorylation and SUMOylation. Cell-based analyses of tyrosine kinase, PYK2, revealed that SUMOylation at four lysine residues promoted PYK2 autophosphorylation at tyrosine 402, which in turn enhanced its interaction with SRC and full activation of the SRC-PYK2 complex. SUMOylation on WT but not the 4KR mutant of PYK2 further elevated phosphorylation of the downstream components in the focal adhesion pathway, such as paxillin and Erk1/2, leading to significantly enhanced cell migration during wound healing. These studies illustrate how our SUMO E3 ligase-substrate network can be used to explore crosstalk between SUMOylation and other PTMs in many biological processes. Proteomics studies have revealed that SUMOylation is a widely used post-translational modification (PTM) in eukaryotes. However, how SUMO E1/2/3 complexes use different SUMO isoforms and recognize substrates remains largely unknown. Using a human proteome microarray-based activity screen, we identified over 2500 proteins that undergo SUMO E3-dependent SUMOylation. We next constructed a SUMO isoform- and E3 ligase-dependent enzyme-substrate relationship network. Protein kinases were significantly enriched among SUMOylation substrates, suggesting crosstalk between phosphorylation and SUMOylation. Cell-based analyses of tyrosine kinase, PYK2, revealed that SUMOylation at four lysine residues promoted PYK2 autophosphorylation at tyrosine 402, which in turn enhanced its interaction with SRC and full activation of the SRC-PYK2 complex. SUMOylation on WT but not the 4KR mutant of PYK2 further elevated phosphorylation of the downstream components in the focal adhesion pathway, such as paxillin and Erk1/2, leading to significantly enhanced cell migration during wound healing. These studies illustrate how our SUMO E3 ligase-substrate network can be used to explore crosstalk between SUMOylation and other PTMs in many biological processes. The construction of comprehensive networks linking protein substrates to their respective modifying enzymes is critical to increasing our functional understanding of the role of posttranslational modifications (PTMs) 1The abbreviations used are: PTM, post translational modification; SUMO, small ubiquitin-related modifier; MAPK, mitogen-activated protein kinase; GST, glutathione S-transferase; PPI, protein-protein interaction; RIPA, radioimmunoprecipitation assay buffer; IP, immunoprecipitation; IB, immunoblot; PIAS, protein inhibitor of activated STAT; GO, gene ontology; KSR, kinase substrate relationship; ORF, open reading frame; WT, wild type. 1The abbreviations used are: PTM, post translational modification; SUMO, small ubiquitin-related modifier; MAPK, mitogen-activated protein kinase; GST, glutathione S-transferase; PPI, protein-protein interaction; RIPA, radioimmunoprecipitation assay buffer; IP, immunoprecipitation; IB, immunoblot; PIAS, protein inhibitor of activated STAT; GO, gene ontology; KSR, kinase substrate relationship; ORF, open reading frame; WT, wild type. in signal transduction. Although many PTMs are carried out by individual enzymes (e.g. protein phosphorylation by protein kinases), some PTMs are regulated by complex enzymatic cascades (e.g. conjugation of small ubiquitin-related modifier (SUMO) to cellular proteins on lysine residues). SUMOylation is an essential PTM that controls a broad range of physiological processes, including DNA repair, transcriptional regulation, and nuclear import (1.Gareau J.R. Lima C.D. The SUMO pathway: emerging mechanisms that shape specificity, conjugation and recognition.Nat. Rev. Mol. Cell Biol. 2010; 11: 861-871Crossref PubMed Scopus (846) Google Scholar, 2.Guzzo C.M. et al.RNF4-dependent hybrid SUMO-ubiquitin chains are signals for RAP80 and thereby mediate the recruitment of BRCA1 to sites of DNA damage.Sci. Signal. 2012; 5: ra88Crossref PubMed Scopus (139) Google Scholar, 3.Psakhye I. Jentsch S. Protein group modification and synergy in the SUMO pathway as exemplified in DNA repair.Cell. 2012; 151: 807-820Abstract Full Text Full Text PDF PubMed Scopus (347) Google Scholar, 4.Silver H.R. et al.A role for SUMO in nucleotide excision repair.DNA Repair. 2011; 10: 1243-1251Crossref PubMed Scopus (33) Google Scholar, 5.Z. X. B G. A SUMO ligase is part of a nuclear multiprotein complex that affects DNA repair and chromosomal organization.Proc. Natl. Acad. Sci. U.S.A. 2005; 102: 4777-4782Crossref PubMed Scopus (345) Google Scholar, 6.Brunet Simioni M. et al.Heat shock protein 27 is involved in SUMO-2/3 modification of heat shock factor 1 and thereby modulates the transcription factor activity.Oncogene. 2009; 28: 3332-3344Crossref PubMed Scopus (64) Google Scholar, 7.Onishi A. et al.Pias3-dependent SUMOylation directs rod photoreceptor development.Neuron. 2009; 61: 234-246Abstract Full Text Full Text PDF PubMed Scopus (97) Google Scholar, 8.Matunis M.J. Coutavas E. Blobel G. A Novel Ubiquitin-like Modification Modulates the Partitioning of the Ran-GTPase-activating Protein RanGAP1 between the Cytosol and the Nuclear Pore Complex.J. Cell Biol. 1996; 135: 1457-1470Crossref PubMed Scopus (957) Google Scholar). The vertebrate genome encodes three distinct SUMO isoforms (SUMO1, SUMO2, and SUMO3), which are conjugated to substrate proteins via the SUMOylation enzymatic cascade. A single E1-activating enzyme (SAE1/SAE2 heterodimer) and E2-conjugating enzyme (Ubc9), and several E3 ligases mediate conjugation of SUMO to lysine residues on target proteins. Initially, it was unclear whether SUMO E3 ligases existed, because modification of many substrates did not require the presence of an E3 ligase in vitro (9.Johnson E.S. Protein modification by SUMO.Annu. Rev. Biochem. 2004; 73: 355-382Crossref PubMed Scopus (1384) Google Scholar). In contrast to the ubiquitylation cascade, which includes >600 E3 ligases, only ∼15 SUMO E3 ligases have been reported. These contrasts raise three key questions about the role of SUMO E3 ligases. First, do the E3 ligases determine global substrate specificity? Second, do individual SUMO E3 ligases show preference for SUMO isoforms? Finally, do individual SUMO E3 ligases selectively modify specific protein sub-families? Current techniques have not adequately provided answers to these questions. Proteomic studies have identified thousands of SUMOylated human proteins conjugated to SUMO1 and SUMO2, by affinity purification of SUMO conjugates followed by mass spectrometry (10.Vertegaal A.C. et al.Distinct and overlapping sets of SUMO-1 and SUMO-2 target proteins revealed by quantitative proteomics.Mol. Cell. Proteomics. 2006; 5: 2298-2310Abstract Full Text Full Text PDF PubMed Scopus (245) Google Scholar, 11.Matic I. et al.In vivo indentification of human small ubiquitin-like modifier polymerization sites by high accuracy mass spectrometry and an in vitro to in vivo strategy.Mol. Cell. Proteomics. 2008; 7: 132-144Abstract Full Text Full Text PDF PubMed Scopus (225) Google Scholar, 12.Golebiowski F. et al.System-wide changes to SUMO modifications in response to heat shock.Sci. Signal. 2009; 2: ra24Crossref PubMed Scopus (385) Google Scholar, 13.Impens F. et al.Mapping of SUMO sites and analysis of SUMOylation changes induced by external stimuli.Proc. Natl. Acad. Sci. U.S.A. 2014; 111: 12432-12437Crossref PubMed Scopus (113) Google Scholar, 29.Hendriks I.A. et al.Site-specific mapping of the human SUMO proteome reveals co-modification with phosphorylation.Nat. Struct. Mol. Biol. 2017; 24: 325-336Crossref PubMed Scopus (195) Google Scholar, 43.Hendriks I.A. et al.Uncovering global SUMOylation signaling networks in a site-specific manner.Nat. Struct. Mol. Biol. 2014; 21: 927-936Crossref PubMed Scopus (337) Google Scholar, 59.Lamoliatte F. McManus F.P. Maarifi G. Chelbi-Alix M.K. Thibault P. Uncovering the SUMOylation and ubiquitylation crosstalk in human cells using sequential peptide immunopurification.Nat. Commun. 2017; 8: 141049Crossref Scopus (85) Google Scholar). Although these studies have considerably increased the number of known SUMO substrates, the lack of connection to their upstream E3 ligases remains a roadblock to our understanding of how protein SUMOylation is dynamically regulated in mammalian cells. In this study, we developed an activity-based method for elucidating the global SUMO E3 ligase substrate network, by employing a human proteome microarray (HuProt™) containing >17,000 individually purified proteins (14.Jeong J.S. et al.Rapid identification of monospecific monoclonal antibodies using a human proteome microarray.Mol. Cell. Proteomics. 2012; 11 (O111 016253)Abstract Full Text Full Text PDF Scopus (117) Google Scholar). Utilizing in vitro array-based SUMOylation reactions with purified recombinant E1, E2, and E3s (i.e. PIAS1–4, RanBP2, and TOPORS), we systematically identified >1,700 E3 ligase-dependent substrates that are selectively modified with SUMO1 and/or SUMO2. Gene ontology analysis revealed a significant enrichment of protein kinases as SUMO substrates. In cellulo validation of members of the mitogen-activated protein kinase (MAPK) family identified an essential role for SUMOylation in kinase signaling. Further in vivo characterization of SUMO modification of a nonreceptor tyrosine kinase PYK2 demonstrated novel intramolecular crosstalk, where SUMOylation promotes cell migration via activation of PYK2. E3 ligase purification: Full length PIAS1, PIAS3, and PIAS3ΔSUMO were expressed in bacterial as glutathione S-transferase (GST) fusions in the pDEST15 bacterial expression vector. PIASxβ and PIASγ were expressed in bacteria in the pQLink 6XHis plasmid. Fragments of TOPORS (268–644) and the RanBP2 IR region were subcloned and expressed in pDEST15. All constructs were expressed in Escherichia coli and purified with Glutathione Sepharose 4B (GE Healthcare, Wauwatosa, WI) or Nickel NTA agarose (Qiagen, Germantown, MD). E1 (standard 200 nm; low 35 nm), E2 (standard 600 nm; low 15 nm) were added to S35 radioactively labeled substrate (TNT Quick Coupled Transcription/Translation, Promega, Madison, WI). Assays performed with low concentrations of E1 and E2 were supplemented with E3 ligases (5–20 nm). Reaction mixtures were supplemented with energy mix buffer system (17). Reactions were incubated at 37 °C for 1 h, then quenched by addition of SDS-PAGE sample buffer and analyzed by SDS-PAGE and autoradiography. SUMO substrate open reading frames (ORFs) were expressed as GST fusion proteins in yeast. Cultures (6 ml) were grown at 30 °C to an optical density at 600 nm of 0.7 to 0.9 and induced with 2% galactose for 4 to 6 h. Harvested cells were lysed with glass beads in lysis buffer (100 mm Tris-HCl (pH 7.4), 100 mm NaCl, 1 mm EGTA, 0.1% 2-mercaptoethanol, 0.5 mm phenylmethylsulfonyl fluoride (PMSF), 0.1% Triton X-100 plus protease inhibitor mixture (Roche, Indianapolis, IN). GST fusion proteins were bound to glutathione beads (GE Healthcare) for 1 h at 4 °C and washed three times with wash buffer I (50 mm Tris-HCl, pH 7.4, 100 mm NaCl, 1 mm EGTA, 0.1% Triton X-100, 0.1% β-mercaptoethanol, and 0.5 mm PMSF) and three times with wash buffer II (50 mm HEPES, pH 7.4, 100 mm NaCl, 1 mm EGTA, 10% glycerol, 0.1% β-mercaptoethanol, and 0.5 mm PMSF) and eluted by glutathione competition elution buffer (100 mm Tris-HCl, pH 8.0, 100 mm NaCl, 10 mm MgCl2, 40 mm glutathione, and 30% glycerol). The eluate was collected through a filter unit and stored in a 384-well plate. Sixty-six SUMO substrate proteins were successfully purified, as determined by probing with anti-GST antibody. The purified substrates along with 16 recombinant control proteins, acquired through generous contributions, were printed in duplicate on modified glass (Full Moon Biosystems, Sunnyvale, CA) microscope slides using a 48-pin contact printer (Bio-Rad, Hercules, CA) employing four pins. Protein chips were incubated overnight in BSA blocking buffer at 4 °C. SUMOylation on the protein chips was performed under two assays conditions. Chips were incubated with high concentrations of E1 (2.3 μm) and E2 (6.25 μm) and a positive control for control SUMOylation. To evaluate the activity of E3 ligases, limiting concentrations of E1 (45 nm) and E2 (125 nm) were supplemented with recombinant E3 ligases (5–20 nm). All mixtures included 0.7 μm mature SUMO-Alexa555, 5 mm ATP in 20 mm HEPES, pH 7.4, 100 mm NaCl, 10 mm MgCl2, 0.1 mm dithiothreitol. The SUMOylation reactions were carried out in a humidity chamber for 90 min at 37 °C then washed with TBST, followed by 1% SDS solution at 55 °C, rinsed with water and spun dry. Negative controls containing SUMO reaction mixture without enzymes were run in parallel. All conditions were performed in triplicate. To normalize the signal, we assume that the real reaction/signal are rare and almost evenly dispersed in each block, thus we force each block on a chip to have a median signal intensity of one. To be included as a positive hit, the duplicate spots of each gene must both have signal intensities (Foreground/Background ratio) five standard deviations above the mean. Additionally, positive hits from the triplicates of each enzymatic reaction must be identified by at least two of the three replicates. Positive hits also identified on negative control protein microarrays were removed. The names of all genes were checked and nonofficial gene name were replaced by official gene symbol name. We used Cytoscape to create the SUMOylation substrate network. The E3 ligase/SUMO pairing and substrates are represented by large filled circles and corresponding small filled circles, respectively. The E3 ligase and ligase-specific substrates are marked by same ligase/SUMO-specific color with substrates circled around the E3 ligase. For example, PIAS3/SUMO2 and its specific substrates are marked by blue. The shared substrates are marked by gray color and connected to corresponding case by case-specific color lines. For example, the substrate sets shared by PIAS3/SUMO2 and PIAS1/SUMO1 are connected to PIAS3/SUMO2 by green colored lines, and connected to PIAS1/SUMO1 by gold colored lines. The amino acid sequences of kinase domain of all human kinase proteins have been annotated by Manning et al. with Hidden Markov Model (15.G M. et al.Evolution of protein kinase signaling from yeast to man.Trends Biochem. Sci. 2002; 27: 514-520Abstract Full Text Full Text PDF PubMed Scopus (739) Google Scholar). We collected these sequences from kinase.com and built the phylogenetic tree by Mega 5 (16.Tamura K. et al.MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods.Mol. Biol. Evol. 2011; 28: 2731-2739Crossref PubMed Scopus (34916) Google Scholar). We marked different kinase family by distinct color shadow and marked the SUMOylated kinases with a red circle. Seventy-one of the 2150 SUMOylation substrates (not including the substrates of high concentration E1 and E2) are kinases. To analyze the relationship of these 71 SUMOylation kinases, we built their functional relationship network. In this network, two kinases are connected by an undirected orange line if there is protein-protein interaction (PPI) between them. In addition, two kinases are connected by a directed green line with an arrow pointing to the substrate if there is a known kinase-substrate relationship between them. The kinases that do not have phosphorylation or PPI relationships are represented by orphan node. Candidate substrates were subcloned into the PCAGIG-V5 mammalian expression vector, SUMO1 and SUMO2 were subcloned into PCAGIG-MYC mammalian expression vector, and E3 ligases were subcloned into PSG5-FLAG mammalian expression vector. V5-substrate and MYC-SUMO were transfected together with and without FLAG-E3 ligases constructs with Fugene 6 transfection reagent (Promega) into HeLa cells seeded at 2E105 cells per well. After 48 h, the cells were washed with phosphate buffered saline and lysed in RIPA buffer containing 20 mm N-ethylmaleimide and 1% SDS, to inhibit deSUMOylation and dissociate noncovalent protein complexes. Anti-V5-agarose beads were added (Sigma-Aldrich, St. Louis, MO) for 2 h to immunoprecipiate the substrate. Immunoprecipitates were resolved by SDS-PAGE and subject to immunoblotting to detect SUMOylation. 4E107 HeLa cells were cotransfected with V5-PYK2, MYC-SUMO1, and FLAG-PIAS1 using Fugene 6 transfection reagent (Promega). After 48 h, cells were rinsed with warm PBS then lysed with RIPA buffer (50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 2 mm EDTA, 1% Triton-X-100) containing 0.5 mm PMSF, Roche Protease Inhibitor Mixture, 20 mm N-ethylmaleimide and 1% SDS. The concentrated lysate was spun down to clear debris and the supernatant was diluted with RIPA without SDS, to a final concentration of 0.2% SDS. V5-PYK2 was immunoprecipitated with Anti-V5-agarose (Sigma Aldrich), washed three times with RIPA containing 0.1% SDS. PYK2 was eluted using V5 peptide (0.5 mg/ml). Eluate was incubated with anti-MYC agarose to isolate PYK2-SUMO1. Beads were again washed three times with RIPA containing 0.1% SDS. PYK2-SUMO1 was eluted from beads in 0.4 m ammonium bicarbonate and 8 m urea. TCA precipitation was used to concentrate the eluate and the pellet was resuspended in 8 m urea. The sample was heated to 100 °C for 10 min in LDS with β-mercaptoethanol, then fresh iodoacetamide was added to 50 mm concentration, the mixture was incubated in the dark at room temperature for 30 min. The sample was resolved by SDS -PAGE using a 12% NuPage Bis-Tris gel (Life Technologies, Carlsbad, CA) then silver stained with the SilverQuest kit (Life Technologies). Four bands corresponding to PYK2-SUMO1 were excised from the gel, destained, and subject to tryptic digestion (1:20 trypsin/substrate ratio). The resulting peptides were separated on a Dionex Ultimate 3000 RSLCnano system (Thermo Scientific, Wilmington, DE) with a 75 μm x 15 cm Acclaim PepMap100 separating column (Thermo Scientific) protected by a 2 cm guarding column (Thermo Scientific). Mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid 95% acetonitrile (B). The gradient profile was set as following: 4–30% B for 40 min, 30–45% B for 10 min, 45–95% B for 10 min. MS analysis was performed using an Orbitrap Velos Pro mass spectrometer (Thermo Scientific). The spray voltage was set at 2.2 kV. Spectra (AGC 1 × 106) were collected from 400–1800 m/z at a resolution of 60,000 followed by data-dependent HCD MS/MS (at a resolution of 7500, collision energy 35%, activation time 0.1 ms) of the 10 most abundant ions. MS/MS spectra were searched against a human IPI reference database (V3.87) using the SEQUEST engine in Proteome Discover 1.3. Searching parameters included mass tolerance of precursor ions (± 20 ppm) and product ion (± 0.06 Da), dynamic modification of carboxyamidomethylated Cys (+ 57.0215 Da), dynamic mass shifts for oxidized Met (+ 15.9949 Da), and dynamic modification of SUMO1 C-terminal peptide (ELGMEEEDVIEVYQEQTGG) or target peptide terminal attached to the modified K. Only b and y ions were considered during the database match. GST-PYK2 WT was purified from yeast using glutathione Sepharose 4B (GE Healthcare). Following washes PYK2 was left on beads and separated into two aliquots. One aliquot was SUMOylated under standard conditions (reaction 1) with 90 nm E1, 300 nm E2, 5 mm ATP in 50 mm HEPES, pH 7.4, 100 mm NaCl, 10 mm MgCl2, and 0.1 mm dithiothreitol for 1 h at 37 °C with gentle shaking. The reaction mixture was removed, the beads were washed two times with SUMOylation buffer without enzymes or ATP, a final wash was performed with kinase buffer lacking ATP. The second aliquot served as a control (reaction 2) and was incubated with SUMOylation reaction buffer lacking enzymes, SUMO, and ATP. In the second phase of this reaction, the beads were again divided in two, creating four reaction conditions. Kinase buffer (50 mm HEPES, pH 7.4, 10 mm MgCl2, 10 mm MnCl2, 300 mm KCl, and 0.5% Nonidet P-40) with 1 mm cold ATP was added to one aliquot that had previously undergone SUMOylation (reaction 1.1) and one aliquot that had only been incubated with buffer (reaction 2.1). The paired aliquots were incubated with kinase buffer lacking ATP (reaction 1.2 and 2.2). The autophosphorylation and control reactions were allowed to proceed for 1 h at 30 °C with gentle shaking then all reactions were washed three times with kinase buffer lacking ATP. The beads were heated in 2X LDS sample buffer containing β-mercaptoethanol at 100 °C for 10 min, then resolved by SDS-PAGE and immunoblotted with anti-GST (Millipore), anti-SUMO1 (21C7) (Matunis laboratory), and anti-PYK2 pTyr402 (Life Technologies) antibodies. V5-tagged WT PYK2, 4KR PYK2, and PYK2 Tyr402F were all transfected with and without MYC-SUMO1 in HeLa cells. V5-tagged WT PYK2 and 4KR PYK2 were also cotransfected with flag-PIAS1 and MYC-SUMO1. 4KR PYK2 and Tyr402F were generated using the QuikChange II site directed mutagenesis kit (Agilent Technologies, Wilmington, DE). Cells were rinsed with warm phosphate buffered saline 48 h after transfection and lysed with Kamiya buffer (50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 5 mm MgCl2, 1 mm EDTA, 1% Triton-X-100). The lysate was incubated with anti-V5-agarose for two hours. Beads were washed three times with TBST then boiled with 2× LDS containing β-mercaptoethanol. The immunoprecipitates were resolved by SDS-PAGE followed by immunoblotting with antibodies to SRC (Cell Signaling, Danvers, MA) and paxillin (Santa Cruz, Dallas, TX). Phosphospecific antibodies for paxillin pTyr118 and SRC pTyr416 (Cell Signaling) were used to assess phosphorylation status. MDA-MB-231 cells were plated in 6 well format and infected with adenovirus constructs for WT PYK2, 4KR PYK2, and SUMO1 (generated using ViraPower™ Adenoviral Gateway™ Expression Kit, Life Technologies). When cells were confluent (24 h post infection), a scratch was made using a fine pipette tip and cells were gently washed three times with room temperature phosphate buffered saline. Cells were then placed in low serum (0.1% FBS) medium for 24 h. Phase contrast images of the same fields were taken at 0 h and 24 h following the scratch and cells migrating beyond the wound's edge were manually counted (n = 3 per condition). Cells were plated in a glass slide chamber (Thermo Fisher, Westminster, MD) and infected with the designated adenoviral constructs. A scratch was made as previously described (2D scratch assay) and cells were placed in low serum medium for 4 h followed by fixation with 4% formaldehyde. Cells were permeabilized with 0.5% Triton-X-100 and blocked with normal goat serum. Cells were probed with anti-phospho-paxillin (Cell Signaling) followed by Cy3-conjugated goat-anti-rabbit 2nd antibody. Following PBS washes, slides were counter stained with DAPI and fluorescence images were taken using Axivision apotome microscopy (Zeiss, Peabody, MA) with the same exposure time. To identify SUMO E3 ligase-dependent substrates, we employed an in vitro assay using the human proteome microarray (i.e. HuProt™), in combination with bioinformatics analysis to determine the enzyme-substrate relationships. First, we fabricated a pilot protein microarray containing multiple known SUMO substrates expressed and purified from bacteria, in order to optimize the reaction conditions for detecting SUMO E3 ligase-dependent substrates. The pilot microarray contained 60 previously identified substrates, SUMO1–4, E1 and E2 enzymes involved in SUMOylation, and GST and BSA as negative controls. To verify detectable substrate SUMOylation, we incubated the pre-blocked pilot microarray with high concentrations of SAE1/UBA2 (i.e. E1; 2.3 nm), Ubc9 (i.e. E2; 6.25 nm), in the standard SUMOylation reaction buffer in the presence of fluorescently labeled SUMO1 or SUMO2. Following incubation with this enzyme mixture, the pilot microarray was washed under denaturing conditions, and signals from covalently SUMOylated substrates were detected via fluorescence imaging. Under this condition of high E1/E2 (defined as 50X in the rest of the text; see Experimental Procedures for more details) concentrations, we observed significant SUMOylation signals from 41% of the known substrates (supplemental Fig. S1), consistent with previous observations that E3s are not required for in vitro SUMOylation of many substrates when carried out in the presence of high E1 and E2 concentrations (9.Johnson E.S. Protein modification by SUMO.Annu. Rev. Biochem. 2004; 73: 355-382Crossref PubMed Scopus (1384) Google Scholar, 17.Rogers R.S. Horvath C.M. Matunis M.J. SUMO modification of STAT1 and its role in PIAS-mediated inhibition of gene activation.J. Biol. Chem. 2003; 278: 30091-30097Abstract Full Text Full Text PDF PubMed Scopus (127) Google Scholar, 18.Yunus A.A. Lima C.D. Structure of the Siz/PIAS SUMO E3 ligase Siz1 and determinants required for SUMO modification of PCNA.Mol. Cell. 2009; 35: 669-682Abstract Full Text Full Text PDF PubMed Scopus (92) Google Scholar). To determine the minimal detectable SUMOylation signal induced by E1 and E2 alone, we repeated SUMOylation reactions on our pilot protein microarrays using a dilution series of E1 and E2 enzymes until SUMOylation signals could be barely detected and therefore, defined as 1X E1/E2 condition (supplemental Fig. S1). In parallel, we purified the 11 reported SUMO E3 ligases, including PIAS1–4, RanBP2, TOPORS, MAPL, Pc2, Rhes, HDAC, and HSP27, as recombinant proteins and tested their E3 ligase activity against their reported substrates under standard in vitro conditions (17.Rogers R.S. Horvath C.M. Matunis M.J. SUMO modification of STAT1 and its role in PIAS-mediated inhibition of gene activation.J. Biol. Chem. 2003; 278: 30091-30097Abstract Full Text Full Text PDF PubMed Scopus (127) Google Scholar). Six of the eleven tested E3 ligases - namely PIAS1–4, RanBP2, and TOPORS - showed robust SUMOylation activity to their previously reported substrates and were thus selected for further studies using protein microarrays (supplemental Fig. S2). To identify E3 ligase-dependent substrates, we supplemented the reaction mixture (containing 1× E1/E2 enzymes) with active E3 ligases to conduct the SUMOylation reactions on the pilot array (Fig. 1A). We detected SUMO modification signals in all substrates previously identified under the 50X E1/E2 condition. Importantly, we also identified additional E3 ligase-dependent SUMOylation signals that could not be detected under low E1/E2 conditions. For certain substrates, such as CRIP, we detected robust SUMOylation signals in the presence of the E3 ligase RanBP2, even though these substrates were not found to be SUMOylated under high E1/E2 conditions (Fig. 1B). For the remainder of our experiments, the minimal amounts of E1 and E2 enzymes required for detecting E3-dependent substrates on protein microarrays were defined as 45 nm E1 and 12.5 nm E2 (referred to as 1× E1/E2). We employed the HuProt™ array (14.Jeong J.S. et al.Rapid identification of monospecific monoclonal antibodies using a human proteome microarray.Mol. Cell. Proteomics. 2012; 11 (O111 016253)Abstract Full Text Full Text PDF Scopus (117) Google Scholar), comprised of ∼17,000 unique proteins, to systematically identify potential E3 ligase-dependent SUMO substrates in the human proteome. The optimized 1X E1/E2 assay condition was supplemented with the six E3 ligases and either SUMO1 or SUMO2, and then applied to the HuProt™ array (Fig. 1C). HuProt™ arrays incubated with 50X E1/E2 paired with SUMO1 or SUMO2 were used as a SUMOylation positive control. Conversely, 1× E1/E2 paired with either SUMO1 or SUMO2 were incubated on the HuProt™ array as a negative control. To ensure reproducibility of the assay, each SUMOylation reaction was performed in triplicate. A total of eighteen SUMOylation reactions were performed on 54 HuProt™ arrays under the conditions specified in Fig. 1A (Data set 1). Only those proteins that were found SUMOylated in at least two of the triplicate assays were recognized as reproducible hits. Under the 50× E1/E2 reaction condition, we detected 2346 and 1933 SUMOylated proteins in the presence of SUMO1 and SUMO2, respectively. Unfortunately, despite numerous attempts, the PIAS2-SUMO2 reactions yielded saturated images that could not be analyzed by the image acquisition software and therefore PIAS2-SUMO2 data is not incorporated in our data set. Under the 1X E1/E2 concentrations, addition of the six E3 ligases yielded a variable number of modified substrates, ranging from 3 to 1092 (Data set 3) (Fig. 2A). For example, the PIAS3-SUMO2 reactions identified 478 unique targets; whereas TOPORS-SUMO2 only modified two proteins specifically (Fig. 2B; Table I). Except for two reactions (PIAS3/SUMO1 and PIAS4/SUMO1), all assays identified specific targets. Collectively, a total of 2149 substrates were SUMOylated by all E3 ligases tested (Table I). By removing the hits in the 50× E1/E2 experiment from those obtained in the E3 ligase reactions under 1× E1/E2 condition, we revealed the E3 ligase-dependent targets. The known SUMOylation consensus motif occurred in 27.27–54.61% (M3 algorithm) or 20.2–56.62% (GPS-SUMO) (19.Zhao Q. et al.GPS-SUMO: a tool for the prediction of sumoylation sites and SUMO-interaction motifs.Nucleic Acids Res. 2014; 42: W325-WW30Crossref PubMed Scopus (299) Goo