Global Identification of SUMO E3 ligase-dependent substrates using the HuProt™ arrays. A, Schematic describing the global SUMOylation assay design. SUMOylation reactions with high and low concentrations of recombinant E1/E2, paired with SUMO1 or SUMO2, were performed as controls. E3 ligases were spiked into the low E1/E2 conditions and performed with SUMO1 and SUMO2. All experimental and control conditions were performed in triplicate. B, E3 ligase activity was tested in pilot study performed on a specialized SUMOylation protein chip containing 82 control proteins. Low concentrations of E1/E2 allow for direct detection of E3 ligase activity compared with high E1/E2 controls reactions. C, PIAS3 SUMO2 substrates from the HuProt™ microarray are highlighted.

Specificity in E3 ligase mediated-SUMOylation. A, A network showing the connections between each E3 ligase/SUMO isoform pairing and the modified substrates was generated using Cytoscape. The colored edges depict the connection to an upstream E3 ligase. Many substrates are connected to more than one E3 ligase/SUMO pairing, thus revealing the overlap and redundancy between E3 ligases. B, The proportions of substrates shared by E3 ligases as compared with the unique substrates targeted by each E3 ligase/SUMO pairing. The colored fractions represent the number of substrates unique to the reaction, whereas the black fractions represent the targets modified in at least two reactions. C, Preferential modification with SUMO1 versus SUMO2. The orange fraction represents the percentage of total hits modified with SUMO1; the blue fraction represents substrates modified with SUMO2. Substrates modified with SUMO1 and SUMO2 are counted in both fractions.

Biological Classification of SUMOylated proteins. A, Gene Ontology analysis of significant biological processes and molecular functions. Overrepresented (p < 0.05) biological process and molecular function GO categories were identified. The composite data set and each E3 ligase/SUMO pairing were analyzed separately. B, Phylogenetic kinase tree overlaid with SUMOylation enrichment the amino acid sequences of kinase domain of all human kinase proteins were collected to build the phylogenetic tree with Mega 5, a program designed to construct phylogenetic trees from aligned sequences. Kinase families are designated by distinct color and SUMOylated kinases are indicated by red circles. C, Kinase SUMOylation substrates in PPI and KSR network. Protein-protein interaction analysis was performed on the subset of kinases identified as SUMO substrates in our array-based assays (orange edges). Kinase substrate relationships (KSRs) are represented by green directional arrows. Kinases encircled in red were selected for validation studies.

Validation of E3 ligase specificity of in the MAPK kinase family. A, Flow chart for validation studies. V5-tagged kinases were transfected into HeLa cells. In parallel each kinase is cotransfected with SUMO1 or SUMO2, and separately cotransfected with SUMO1 or SUMO2 and the indicated E3 ligase. V5-tagged substrates were immunoprecipitated, resolved by SDS-PAGE, then blotted with anti-MYC to detect SUMOylation. B, The fidelity of HuProt™ array E3 ligase results were tested both for the substrate and SUMO isoform specificity. C, Activity of PIAS E3 ligases was confirmed with four additional MAPK kinase substrates.

PYK2 is a SUMOylated tyrosine kinase. A, Bioactive peptide induced signaling pathway is enriched for SUMOylation substrates (p < 1.637 E10–4). Proteins that were recovered in our global SUMOylation assay are marked as SUMO modified. B, Validation of PYK2 SUMOylation and E3 ligase specificity in HeLa cells as indicated in A. SUMOylation of nonreceptor tyrosine kinase PYK2 was performed according to the conditions indicated in the global SUMOylation screen. C, Schematic representation of candidate attachment site lysines in PYK2. D, SUMOylation assay using PYK2 containing the four candidate lysine to arginine mutations (C) demonstrated impaired SUMOylation in the 4KR mutant.

PYK2 SUMOylation promotes its interaction with SRC. A, Comparison of phosphorylation at Tyr402/579/580/881 in WT versus 4KR mutant. V5-PYK2 constructs were cotransfected with MYC-tagged SUMO1 along with FLAG-tagged PIAS1 or PIAS4 as indicated. Lysates were analyzed for phosphorylation status at Tyr402, Tyr579/80, and Tyr881 using phosphospecific antibodies. B, SUMOylation stimulates autophosphorylation of PYK2 in vitro. Recombinant GST-PYK2 was SUMOylated by SUMO1 in the absence of an E3 ligase. Autophosphorylation was examined for the following conditions: untreated, autophosphorylated (kinase assay), SUMOylated, SUMOylated then autophosphorylated (kinase assay). The samples were resolved by SDS-PAGE and subjected to Western blotting with anti-pTyr402 antibody. C, SUMOylation of PYK2 promotes interaction with SRC. HeLa cells were transfected with plasmids encoding for V5-tagged PYK2, V5-PYK2 containing four lysine to arginine mutations (4KR), and V5-PYK2 Y402F. V5-PYK2 constructs were cotransfected with MYC-tagged SUMO1 along with FLAG-tagged PIAS1. V5-PYK2 was immunoprecipitated under nondenaturing conditions to assess the interaction with endogenous SRC. Phosphorylation status of PYK2 and SRC were also analyzed using phosphospecific antibodies. Inputs controls for MYC-SUMO1, endogenous SRC and Tubulin are also shown. D, Quantitative analysis of PYK2 autophosphorylation and PYK2-SRC interaction based on biological duplicates in C. E, SUMOylation of PYK2 promotes phosphorylation of paxillin. HeLa cells were transfected with plasmids encoding for V5-tagged PYK2, V5-PYK2 4KR, or V5-PYK2 Y402F. V5-PYK2 constructs were cotransfected with MYC-tagged SUMO1 along with FLAG-tagged PIAS1. Phosphorylation status of PYK2 adaptor protein paxillin was determined by Western blotting for pTyr118. F, PYK2 4KR mutant inhibits activation of ERK1/2 but not P38/MAPK. HEK293 cells were transfected with either WT or 4KR versions of PYK2, plus and minus SUMO1 coexpression. Whole cell lysates were subject to Western blotting to probe for activation of ERK1/2 (pERK1/2), total ERK1/2, activated P38 MAPK (pP38/MAPK), and PYK2 expression.

PYK2 SUMOylation promotes cell migration. A, 2D scratch assay. MDA-MB-231 breast cancer cells were infected with adenovirus constructs as indicated, at MOI 30. The monolayer was scratched 24 h following infection and cells were allowed to migrate into the wound for 24 h in low serum (0.1% FBS). The number of migratory cells was counted 24 h post scratch (n = 3 per condition). B, Quantification of cell migration. The number of migratory cells in the WT condition is significantly more than control and 4KR PYK2 (p < 0.029 and p < 0.036, respectively, two tailed, t test). WT PYK2 SUMO1 condition is significantly larger than 4KR PYK2 SUMO (p < 0.046, two tailed, t test). C, Paxillin activation under PYK2-SUMOylation and wounding. Cells were set up identical to panel A. 4 h post scratch, cells were fixed and immunofluorescence assays were performed to detect paxillin phosphorylatedTyr-118. Nuclei (Blue); p-Tyr-118 paxillin (Red). D, Crosstalk-mediated activation of PYK2 model. In step 1, SUMOylation of PYK2 occurs at Lys581 as well as other acceptor lysine sites. SUMOylation triggers autophosphorylation at Y402, which stimulates interaction with SRC through its SH2 domain as well as potential SIM-mediated interactions (step 2). SRC phosphorylates PYK2 at Tyr579, Tyr580, and Tyr881 resulting in full catalytic activity of PYK2 (step 3). PYK2-SUMO1 phosphorylates focal adhesion protein paxillin at Tyr118 and activates ERK1/2. pTyr118 is linked to cell migration likely through activation of the MAP kinase pathway (step 4). SUMOylation of PYK2 uncovers a novel crosstalk-mediated mechanism for kinase activation and function.