Supporting Information

Impens et al. 10.1073/pnas.1413825111 13 15 13 15 SI Methods beling) (Silantes Gmbh), or C6 N2 L-lysine HCl and C6 N4 L- Plasmids. pSG5-His6-SUMO1 plasmid encodes the N-terminal arginine HCl (heavy labeling) (Silantes Gmbh). L-Lysine HCl was His6-tagged mature Small ubiquitin modifier 1 (SUMO1) isoform added at its normal concentration in DMEM (146 mg/L), but the (kind gift of A. Dejean, Institut Pasteur, Paris). The pSG5-His6- concentration of L-arginine HCl was reduced to 25 mg/L (30% of SUMO1 T95R mutat was derived from this plasmid using PCR the normal concentration in DMEM) to prevent metabolic con- mutagenesis. pSG5-His6-SUMO2 was obtained by inserting the version of arginine to proline (4). Cells were kept for at least six cDNA corresponding to the human mature SUMO2 isoform population doublings to ensure complete incorporation of the la- with an N-terminal His6 tag in the pSG5 vector (Stratagene). beled lysine and arginine. 2 The pSG5-His6-SUMO2 T91R mutant was derived from this For transfections, cells were seeded in 75-cm flasks or in 6- or plasmid by PCR mutagenesis. N-terminally HA-tagged human 24-well plates at a density of 2.7 × 106 cells per flask or 3 × 105 or cDNA of ZBTB20 (Zinc finger and BTB domain containing 0.5 × 105 cells per well, respectively. The next day cells were 20) isoform 2 (UniProt identifier Q9HC78-2), HMBOX1 (Ho- transfected with Lipofectamine LTX reagents (Invitrogen) (20 μg meobox containing 1) isoform 1 (HMBOX1A) (UniProt of DNA per flask, 3.5 μg per well in the six-well plates, or 0.75 μg identifier Q6NT76-1), NACC1 (Nucleus accumbens-associated per well in the 24-well plates) for 48 h. protein 1) (UniProt identifier Q96RE7), MAP7 (Microtubule-as- For listeriolysin O (LLO) treatment, cells were serum-starved sociated protein 7) isoform 1 (UniProt identifier Q14244-1), or for 2 h; then 3 nM of purified toxin (5) was added directly to the LMNB1 (Lamin-B1) (UniProt identifier P20700) and C-terminally culture medium for 20 min. Cells were lysed further for His HA-tagged human cDNA of TFAP2A (Transcription factor AP-2 pulldown assays or immunoblot analysis. alpha) isoform 1 (UniProt identifier P05549-1) were introduced into the pCDNA3 vector (Invitrogen). From these plasmids ex- Luciferase Assays. HeLa cells were transfected with different mixes pression vectors for HA-ZBTB20 K257R, HA-HMBOX1 K413R, of NRLuc- and CRLuc-encoding plasmids (ratio 1:1). Cells were HA-NACC1 K167R, HA-MAP7 K406R, HA-LMNB1 K241R, harvested 48 h after transfection, and luciferase activities were and TFAP2A K10R-HA were derived by PCR mutagenesis. The quantified on a Tristar LB491 luminometer (Berthold Technol- pCDNA3-NRLuc plasmid was obtained by inserting the coding ogies), using the Renilla Luciferase Assay System (Promega). sequence corresponding to N-terminal residues 1–229 from the green Renilla luciferase protein (RLuc; ThermoFisher Scientific), His Pulldown Assays. SUMOylated were isolated from cell fused to a GGGS flexible linker, into pCDNA3 vector (1). From lysates as described in ref. 6. Briefly, cells were washed in PBS this plasmid four expression vectors were derived with SUMO1 and lysed in lysis buffer [6 M guanidium HCl, 10 mM Tris, 100 WT, SUMO1 T95R, SUMO2 WT, or SUMO2 T91R fused in the mM sodium phosphate buffer (pH 8.0), 5 mM β-mercaptoetha- C-terminal luciferase fragment (residues 1–229). The pCDNA3- nol, 1 mM imidazole]. For MS analysis, 2 × 107 cells per SILAC CRLuc plasmid was obtained by inserting the coding sequence condition were lysed in 40 mL of lysis buffer without β-mer- corresponding to the C-terminal residues 230–311 from the RLuc captoethanol and imidazole. Equal volumes of each SILAC protein, fused to three GGGS flexible linkers, into the pCDNA3 condition were mixed, and proteins were reduced and alkylated vector (1). Expression vectors were derived from this vector for by the addition of 5 mM Tris(2-carboxyethyl)phosphine (TCEP) either a phosphorylation-mimic variant of the Daxx (Death do- and 10 mM chloroacetamide. After incubation for 30 min at 37 °C main-associated protein) SUMO-interacting motif (SIM) (KGG- in the dark, excess chloroacetamide was quenched by addition of KTSVATQCDPEEIIVLDDDD) (1) or the PIAS2 (Protein 20 mM DTT. Then cell lysates were incubated overnight at 4 °C inhibitor of activated STAT) SIM (KVDVIDLTIESSSDEEEDP- with 3 mL of packed NiNTA agarose beads (Qiagen) prewashed PAKR) (2), fused in the N terminus of the luciferase fragment in lysis buffer. After incubation, beads were washed once in lysis – (residues 230 311). All generated constructs were verified by buffer, once in wash buffer pH 8.0 [8 M urea, 10 mM Tris, 100 sequencing. mM sodium phosphate buffer (pH 8.0), 0.1% Triton X-100, 5 mM β-mercaptoethanol], and three times in wash buffer pH 6.3 Antibodies. The following primary antibodies were used for im- [8 M urea, 10 mM Tris, 100 mM sodium phosphate buffer (pH munoblot analysis: mouse anti-actin (R5441; Sigma-Aldrich) and β anti-HA tag (no. 2367; Cell Signaling Technology); rabbit anti- 6.3), 0.1% Triton X-100, 5 mM -mercaptoethanol, 10 mM im- RanGAP1 (no. R0155; Sigma-Aldrich) and anti-SUMO1 (no. 4930; idazole]. SUMOylated proteins then were eluted from the beads Cell Signaling Technology); and home-made rabbit polyclonal using elution buffer A [200 mM imidazole, 5% (wt/vol) SDS, 150 · β antibodies raised against recombinant SUMO3 protein produced in mM Tris HCl (pH 6.7), 30% (vol/vol) glycerol, 720 mM -mer- Escherichia coli (R206). Anti-mouse and anti-rabbit HRP-conju- captoethanol, and 0.0025% bromophenol blue] for immunoblot gated antibodies (AbCys) were used as secondary antibodies. analysis or elution buffer B [100 mM sodium phosphate buffer (pH 6.8), 200 mM imidazole] for MS analysis. The latter eluates Cell Culture, Stable Isotope Labeling by Amino Acids in Cell Culture, contained about 500 μg of protein in a volume of 1.5 mL for both and Transfections. HeLa cells (CCL-2; ATCC) were cultured in SUMO1 and SUMO2 analysis. MEM-GlutaMAX medium (Invitrogen), supplemented with 10% (vol/vol) FBS, MEM nonessential amino acids (Invitrogen), and Immunoblotting. Cells lysed in Laemmli buffer [0.125 M Tris (pH 1 mM sodium pyruvate (Invitrogen). 6.8), 4% (wt/vol) SDS, 20% (vol/vol) glycerol, 100 mM DTT] and For stable isotope labeling by amino acids in cell culture (SILAC) proteins eluted from His pulldown assays were separated on SDS- (3), cells were cultured in DMEM without L-lysine, L-arginine, or polyacrylamide gels. Proteins were transferred to PVDF mem- L-glutamine (Silantes Gmbh) supplemented with 10% (vol/vol) branes and incubated with primary and secondary antibodies. dialyzed FBS (Invitrogen), 2 mM GlutaMAX (Invitrogen), and Proteins were revealed using Pierce ECL 2 Western Blotting either natural L-arginine HCl and L-lysine HCl (light labeling) Substrate (Fisher Scientific). All displayed immunoblots are 13 (Sigma), D4 L-lysine HCl and C6 L-arginine HCl (medium la- representative of at least two independent experiments.

Impens et al. www.pnas.org/cgi/content/short/1413825111 1of5 Immunocapture of Diglycine-Modified Peptides. Eluates from the cation of lysine residues (+114.042927 Da; light search), SILAC His pulldown were diluted further with 8.5 mL 50 mM ammonium modification of lysine residues, and GG modification of SILAC- bicarbonate, and proteins were digested with 20 μg trypsin labeled lysine residues (+4.025107 Da and +118.068034 Da, (Promega). Immunocapture of diglycine (GG)-modified pep- medium search; +8.014199 Da and +122.057126 Da, heavy tides then was performed using the PTMScan Ubiquitin Rem- search). SILAC modification of arginine residues was set as an nant Motif (K-e-GG) Kit (Cell Signaling Technology) according additional fixed modification for the medium (+6.020129 Da) to the manufacturer’s instructions. Briefly, peptides were puri- and heavy (+10.008269 Da) searches. For all searches mass fied on Sep-Pak C18 cartridges (Waters), lyophilized for 2 d, and tolerance of the precursor ions was set to 10 ppm, and mass redissolved in 1.4 mL of the 1× immunoprecipitation buffer tolerance of the fragment ions was set to 0.5 Da. The peptide without detergent supplied with the kit. Note that at this point an charge was set to 2+,3+, and 4+, and up to three missed tryptic aliquot corresponding to 4 μg of digested protein material was cleavage sites were allowed. Also, the C13 setting of Mascot was taken to analyze the input. Peptides were incubated with the set to 1. To identify the phosphorylated peptides listed in Da- antibody-bead slurry for 2 h on a rotator at 4 °C, and after taset S3, these searches were repeated with phosphorylation of μ several wash steps GG-modified peptides were eluted in 100 L serine, threonine, and tyrosine residues as additional variable 0.15% TFA and desalted on reversed-phase C18 OMIX tips modifications. Only peptides that were ranked first and scored ’ (Agilent) according to the manufacturer s protocol. Purified above the threshold score set at 99% confidence were withheld. GG-modified peptides were dried under vacuum in HPLC in- For processing of all MS data the ms_lims (version 7.7.7) soft- − serts and stored at 20 °C until LC-MS/MS analysis. ware platform was used (8). In total, 2,016 and 3,123 peptide spectrum matches (PSMs) were obtained for the SUMO1 and LC-MS/MS Analysis. Peptides were redissolved in 20 μL of solvent A SUMO2 analysis, respectively, with a false-discovery rate (FDR) [0.1% formic acid in water/acetonitrile (98:2, vol/vol)], of which <0.3% (9). For GG-modified peptides, peak intensities from the 6 μL was injected using an Ultimate 3000 HPLC system (Dionex) light, medium, and heavy SILAC label were looked up manually connected in line to an LTQ Orbitrap Velos mass spectrometer in the MS spectra (a few peptides that were identified with (Thermo Electron) operated as described previously (7). Briefly, modified lysine residues at the C terminus were discarded from peptides were loaded onto a C18 reversed-phase chromatogra- phy column and eluted with a linear gradient (2–55%) of solvent further analysis). Only peptides with no detectable light peak B [0.08% formic acid in water/acetonitrile (2:8, vol/vol)] over 120 were annotated to report true SUMOylation sites (625 PSMs for min at a constant flow rate of 300 nL/min. Separated peptides SUMO1 and 324 PSMs for SUMO2). To calculate the degree of were ionized by electrospray ionization and measured in the deSUMOylation upon LLO treatment, the ratio between the medium (M) and heavy (H) peptide envelope was normalized mass spectrometer that was operated in data-dependent mode, – automatically switching between MS and MS/MS acquisition for against the median M/H ratio from all non GG-modified pep- the 20 most abundant ion peaks per MS spectrum. Full-scan MS tides that were quantified by the MaxQuant software (1,427 for spectra (300–2,000 m/z) were acquired at a resolution of 60,000 SUMO1 and 2,954 for SUMO2). Initially, raw data files were in the Orbitrap analyzer after accumulation to a target value of analyzed using the MaxQuant software (version 1.4.0.1) with 1 million. The 20 most intense ions above a threshold value of search settings similar to those described above and allowing 5,000 were isolated for fragmentation by collision-induced dis- a maximum FDR of 1%. However, we observed that specifically sociation at a normalized collision energy of 35% in the linear GG-modified peptides without any detectable light signal were not ion trap (LTQ) after filling the trap at a target value of 5,000 for identified at the given confidence level, thus leading to very low a maximum of 50 ms. Peptides with unassigned charge states and numbers of identified SUMO sites. Unlike the Mascot workflow those with a charge state less than 2+ were excluded from described above, in MaxQuant only a single search is performed to fragmentation. identify peptides derived from the three SILAC conditions. For peptides in which three SILAC labels are expected, the complete Data Processing. From the MS/MS data in each LC-run, Mascot lack of one of the labels presumably results in lower confidence generic files (mgf) were created using the Mascot Distiller about their identification. The M/H ratio for every SUMO site was software (version 2.4.3.3; Matrix Science Ltd.) as described calculated by averaging the M/H ratio of the different corre- previously (7). Generated peak lists then were searched with sponding GG-modified peptides. Peptides that could not be Mascot using the Mascot Daemon interface (version 2.3.0; Ma- mapped precisely in their cognate protein were excluded from trix Science Ltd.) against the human proteins in the Uniprot/ analysis. Swiss-Prot database (database release version of March 6, 2013 The MS proteomics data have been deposited in the Proteo- containing 20,258 human protein sequences, www..org/). meXchange Consortium (http://proteomecentral.proteomexchange. Three independent searches were performed to identify peptides org) via the PRIDE partner repository (10) with the dataset iden- with light, medium, and heavy SILAC labeling. For all three tifier PXD000459. searches variable modifications were set to oxidation of methi- onine residues and pyroglutamate formation of N-terminal Ontology Terms Enrichment Analysis. terms glutamine residues. Carbamidomethyl formation of cysteine enrichment analyses were performed using Database for Anno- residues was set as a fixed modification. Depending on the tation, Visualization and Integrated Discovery (DAVID) bio- search, additional variable modifications included GG modifi- informatics resources (11).

1. Hirohama M, et al. (2014) Assay methods for small ubiquitin-like modifier (SUMO)- 6. Tatham MH, Rodriguez MS, Xirodimas DP, Hay RT (2009) Detection of protein SUMO-interacting motif (SIM) interactions in vivo and in vitro using a split-luciferase SUMOylation in vivo. Nat Protoc 4(9):1363–1371. complementation system. Anal Biochem 448:92–94. 7. Eskandarian HA, et al. (2013) A role for SIRT2-dependent histone H3K18 deacetylation 2. Hecker CM, Rabiller M, Haglund K, Bayer P, Dikic I (2006) Specification of SUMO1- and in bacterial infection. Science 341(6145):1238858. SUMO2-interacting motifs. J Biol Chem 281(23):16117–16127. 8. Helsens K, Martens L, Vandekerckhove J, Gevaert K (2007) MascotDatfile: An open-source 3. Ong SE, et al. (2002) Stable isotope labeling by amino acids in cell culture, SILAC, as a simple library to fully parse and analyse MASCOT MS/MS search results. Proteomics 7(3):364–366. and accurate approach to expression proteomics. MolCellProteomics1(5):376–386. 9. Elias JE, Gygi SP (2007) Target-decoy search strategy for increased confidence in large- 4. Ong SE, Kratchmarova I, Mann M (2003) Properties of 13C-substituted arginine in scale protein identifications by mass spectrometry. Nat Methods 4(3):207–214. stable isotope labeling by amino acids in cell culture (SILAC). J Proteome Res 2(2): 10. Vizcaíno JA, et al. (2013) The PRoteomics IDEntifications (PRIDE) database and 173–181. associated tools: Status in 2013. Nucleic Acids Res 41(Database issue):D1063–D1069. 5. Ribet D, et al. (2010) Listeria monocytogenes impairs SUMOylation for efficient 11. Huang W, Sherman BT, Lempicki RA (2009) Systematic and integrative analysis of infection. Nature 464(7292):1192–1195. large gene lists using DAVID bioinformatics resources. Nat Protoc 4(1):44–57.

Impens et al. www.pnas.org/cgi/content/short/1413825111 2of5 ain UO n Ru oDx rPA2SMdmis eaclswr rnfce ihteidctdpiso lsis n uieaeatvte were activities associ SIMs. are luciferase tested SIMs, the and PIAS2 with or plasmids, similarly Daxx of with interact pairs coexpressed forms when indicated SUMO wild-ty variants, different the to SUMO prox these and with NRLuc into that wild-type transfected fused indicating Both brought we were activity. signal, are luciferase Here, cells luciferase mean fragments low strong substrate. to HeLa in luciferase correspond a adequate resulting domains. Bars two other, an SIM transfection. each the of after PIAS2 for h interact, presence affinity or 48 indeed the lysates Daxx proteins cell in to in the bioluminescence quantified CRLuc If emits and proteins. that SUMOs protein interacting variant luciferase potentially functional to a fused reconstituting are RLuc of fragments S1. Fig. mese al. et Impens eeto fSUMO of Detection www.pnas.org/cgi/content/short/1413825111 – I neatosuigsltlcfrs opeetto sas nsltlcfrs opeetto sas -adC-terminal and N- assays, complementation luciferase split In assays. complementation luciferase split using interactions SIM

Luciferase activity (fold change) 50 10 20 30 40 0

+ NRLuc (empty)

(empty) CRLuc + NRLuc-SUMO1 WT + NRLuc-SUMO1 T95R + NRLuc-SUMO2 WT

± + NRLuc-SUMO2 T91R Dfo w needn xeiet.Sltlcfrs rget ln aelow have alone fragments luciferase Split experiments. independent two from SD + NRLuc (empty) SIM (Daxx)

-CRLuc + NRLuc-SUMO1 WT + NRLuc-SUMO1 T95R + NRLuc-SUMO2 WT + NRLuc-SUMO2 T91R

SIM (PIAS2) + NRLuc (empty)

-CRLuc + NRLuc-SUMO1 WT + NRLuc-SUMO1 T95R + NRLuc-SUMO2 WT + NRLuc-SUMO2 T91R tdwith ated imity, 3of5 eor pe a SUMO1 sites SUMO1 sites analysis #1 analysis #2 (295) (132)

180 115 17

b * 80 * * 60 * * * % 40 *

20 37

0 nucleus dna- transcription zinc finger binding regulation

All SUMO1 conjugated proteins (analysis #1) Highly deSUMOylated SUMO1 targets (analysis #1) All SUMO1 conjugated proteins (analysis #2) Highly deSUMOylated SUMO1 targets (analysis #2)

Fig. S2. Evaluation of the reproducibility of the approach. The large-scale analysis performed for SUMO1 (analysis #1) was repeated on a smaller scale (analysis #2). (A) Overlap between SUMO1 sites identified in both experiments. Eighty-seven percent of sites identified in analysis #2 were also found in analysis #1, indicating that our approach has a high degree of reproducibility. (B) For each analysis, the functional enrichment of highly deSUMOylated proteins after LLO treatment compared with all identified SUMO-conjugated proteins is indicated. Bars correspond to the percentage of proteins annotated with each Splice Pattern-Protein Information Resource (SP-PIR) keyword. Asterisks indicate groups with significant enrichment (modified Fisher exact P value < 0.05). In both analyses, nuclear proteins involved in transcription regulation are enriched among highly deSUMOylated targets.

LLO-induced All SUMOylated proteins deSUMOylated proteins

consensus motif 17% [FILMV] K x [DE] 71% other KxD/E motifs SUMO 13% SUMO 48% no [FILMV] K x [DE] sites sites 15% inv. consensus motif 22% 7% [DE] x K x no [DE] 7% others

Fig. S3. Analysis of SUMO sites highly deSUMOylated after LLO treatment. The distribution of sites over different types of SUMOylation motifs in proteins highly deSUMOylated after LLO treatment (Right) and in all identified SUMOylated proteins (Left) is indicated.

Impens et al. www.pnas.org/cgi/content/short/1413825111 4of5 Table S1. Bacterial strains Name E. coli transformed with:

BUG 3127 pSG5-His6 SUMO1 WT

BUG 3459 pSG5-His6 SUMO1 T95R

BUG 3128 pSG5-His6 SUMO2 WT

BUG 3460 pSG5-His6 SUMO2 T91R BUG 3509 pCDNA3-HA ZBTB20 WT BUG 3510 pCDNA3-HA ZBTB20 K257R BUG 3511 pCDNA3-HA HMBOX1 WT BUG 3512 pCDNA3-HA HMBOX1 K413R BUG 3513 pCDNA3-HA NACC1 WT BUG 3514 pCDNA3-HA NACC1 K167R BUG 3515 pCDNA3-HA MAP7 WT BUG 3516 pCDNA3-HA MAP7 K406R BUG 3517 pCDNA3-HA LMNB1 WT BUG 3518 pCDNA3-HA LMNB1 K241R BUG 3519 pCDNA3-TFAP2A WT-HA BUG 3520 pCDNA3-TFAP2A K10R-HA BUG 3623 pCDNA3-NRLuc BUG 3624 pCDNA3-NRLuc-SUMO1 WT BUG 3625 pCDNA3-NRLuc-SUMO1 T95R BUG 3627 pCDNA3-NRLuc-SUMO2 WT BUG 3628 pCDNA3-NRLuc-SUMO2 T91R BUG 3630 pCDNA3-CRLuc BUG 3631 pCDNA3-SIM DaxxD-CRLuc BUG 3632 pCDNA3-SIM PIAS2-CRLuc

Other Supporting Information Files

Dataset S1 (XLS) Dataset S2 (XLS) Dataset S3 (XLS) Dataset S4 (XLS)

Impens et al. www.pnas.org/cgi/content/short/1413825111 5of5