Proteomics Analysis of Nucleolar SUMO-1 Target Proteins upon Proteasome Inhibition*

Many cellular processes are regulated by the coordination of several post-translational modifications that allow a very fine modulation of substrates. Recently it has been reported that there is a relationship between sumoylation and ubiquitination. Here we propose that the nucleolus is the key organelle in which SUMO-1 conjugates accumulate in response to proteasome inhibition. We demonstrated that, upon proteasome inhibition, the SUMO-1 nuclear dot localization is redirected to nucleolar structures. To better understand this process we investigated, by quantitative proteomics, the effect of proteasome activity on endogenous nucleolar SUMO-1 targets. 193 potential SUMO-1 substrates were identified, and interestingly in several purified SUMO-1 conjugates ubiquitin chains were found to be present, confirming the coordination of these two modifications. 23 SUMO-1 targets were confirmed by an in vitro sumoylation reaction performed on nuclear substrates. They belong to protein families such as small nuclear ribonucleoproteins, heterogeneous nuclear ribonucleoproteins, ribosomal proteins, histones, RNA-binding proteins, and transcription factor regulators. Among these, histone H1, histone H3, and p160 Myb-binding protein 1A were further characterized as novel SUMO-1 substrates. The analysis of the nature of the SUMO-1 targets identified in this study strongly indicates that sumoylation, acting in coordination with the ubiquitin-proteasome system, regulates the maintenance of nucleolar integrity.

Targeting of proteins by conjugation of Small Ubiquitin-like MOdifier (SUMO) 1 is a key mechanism for regulating many cellular processes (1,2), for example the activity of transcription factors (3). Other regulated processes are DNA repair, protein transport, protein-protein interaction, cell cycle progression, and RNA metabolism (4 -6).
SUMO proteins are ubiquitously expressed throughout the eukaryotic kingdom. Yeast, Caenorhabditis elegans, and Drosophila melanogaster carry a single SUMO gene, whereas plants and vertebrates have several SUMO genes (5). In particular, humans express four distinct SUMO family members: SUMO-1, SUMO-2, SUMO-3, and SUMO-4 (7,8). SUMO-1 is an 11.6 kDa protein. It shares about 47% homology with SUMO-2 and SUMO-3 that, on the contrary, differ from each other only by three amino-terminal residues and form a distinct subfamily known as SUMO-2/-3 (9). Despite the low sequence homology, SUMO-1 and SUMO-2/-3 share a similar protein size, tertiary structure, and a carboxyl-terminal diglycine motif (10,11). At the cellular level, different amounts of free SUMO-1 and SUMO-2/-3 are present. The majority of SUMO-1 in fact is conjugated to substrates, whereas the conjugation of SUMO-2/-3 is strongly induced in response to various stresses (10). Finally SUMO-1 and SUMO-2/-3 serve distinct functions as they modify different target proteins (5). Unlike SUMO-1, SUMO-2, and SUMO-3, which are ubiquitously expressed (7),  isoform has yet to be characterized. It seems to be expressed mainly in the kidney, lymph nodes, and spleen, but its role still remains unclear because its mature form has never been reported in vivo (7,12).
Several SUMO targets are known; they are mostly nuclear proteins presenting a consensus acceptor site: ⌿KXE (in which ⌿ is an aliphatic branched amino acid and X is any amino acid) (5). The mutation of this site abolishes sumoylation of substrates and is commonly used to understand the biological implication of the substrate modification. Also SUMO-2/-3 present a conserved lysine in this motif, and they form polymeric SUMO chains (13,14). SUMO-1, however, lacks this consensus site and is not thought to form chains even if recent studies demonstrate that SUMO-1 can be linked to the end of a poly-SUMO-2/-3, terminating the chain (11).
Recently two different extensions of the simple consensus SUMO acceptor site have been identified. These motifs share a negative charge next to the basic SUMO consensus site: one involves a phosphorylated (p) Ser and a Pro residue (⌿KXEXXpSP), and the other contains a negatively charged amino acid close to the acceptor Lys residue (5). Although many targets contain the above mentioned motifs, there are examples of substrates that do not contain these acceptor sites. The presence of a phosphorylated residue in the motif indicates that regulatory mechanisms, which can enhance or decrease the sumoylation of specific targets, may occur at the level of the target itself (15). Indeed sumoylation often acts in coordination with other post-translational modifications like acetylation, methylation, and ubiquitination (16). As discussed above, SUMO proteins are similar in three-dimensional structures. Although they do not display high sequence homology, they share the same structure of ubiquitin and a common conjugation mechanism. In fact like ubiquitination sumoylation also requires the formation of an isopeptide bond between the carboxyl-terminal Gly residue of the modifier protein and the -amino group of a Lys residue in the acceptor protein (5). The enzymatic cascade that mediates SUMO conjugation is similar to that of ubiquitin. The immature precursor is first processed by a specific carboxyl-terminal hydrolase that exposes the diglycine motif, and then mature SUMO proteins are activated by an ATP-dependent heterodimer of SUMO-activating enzyme subunit 1 (SAE1) and SAE2. The above dimer transfers the activated SUMO protein to the ubiquitin-conjugating enzyme 9 (Ubc9) through a transesterification reaction. Ubc9 usually acts together with an E3 ligating enzyme that catalyzes SUMO conjugation to the substrate. In contrast to the ubiquitin pathway in which an E3 enzyme is essential for conjugation, SUMO modification just requires Ubc9, which is able to bind directly to the SUMO consensus sequence and substrates, aligning them for conjugation (5,10,11).
Despite the similarity between SUMO and ubiquitin, the molecular consequences of these two modifications are distinct (17,18). In some cases, such as IB␣ modification, SUMO plays an antagonistic role to ubiquitin, competing for the same lysine (19). In other cases, as for NFB essential modulator/IB kinase ␥, SUMO and ubiquitin are conjugated in a sequential manner in response to a toxic stress; in further cases SUMO may regulate protein localization, stabilizing substrate, independently from ubiquitination as for Smad4 (20 -22). Cross-regulation between SUMO and ubiquitin and the possible interchange of modifiers remain unclear (23,24). Several recent studies indicate that there is a cross-talk between ubiquitinated and SUMO-modified proteins in coordination with proteasome activity (25)(26)(27).
To gain insights into the interconnection of the SUMO and the ubiquitin-proteasome pathway, we investigated the effect of proteasome inhibition on SUMO-conjugated proteins. We analyzed the subcellular distribution of sumoylated proteins in HeLa cells upon MG132 treatment and identified SUMO-1 targets by mass spectrometric techniques. Moreover we measured the effect of MG132 on target modification by stable isotope labeling by amino acids in cell culture (SILAC), and we demonstrated that, upon proteasome inhibition, the amount of SUMO-1 species increases and accumulates in nucleolar structures (28 -30). This enrichment of SUMO-1 allowed the detection of sumoylated targets at endogenous levels, although usually the abundance of sumoylated proteins is relatively low in the cell, and they are difficult to detect. Based on these data, we focused our attention on the nucleolar compartment and identified nucleolar sumoylated proteins that accumulate after proteasome inhibition. The analysis of such proteins strongly indicates that sumoylation is involved in the regulation of nucleolar dynamics.

EXPERIMENTAL PROCEDURES
Cell Culture-HeLa cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% FCS and 100 units/ml penicillin and streptomycin (Invitrogen). Stable isotope labeling was carried out essentially as described previously (28)

cells each).
Immunofluorescence-HeLa cells were grown on sterile 13-mm coverslips and then treated with 10 M MG132 or DMSO overnight. A time course was performed, treating HeLa cells with 10 M MG132 for 1, 6, and 12 h using DMSO as control. Comparison of SUMO-1 staining under several stresses was carried out using 0.2 M actinomycin D (Sigma-Aldrich) for 12 h, 10 g/ml cycloheximide (Sigma-Aldrich) for 12 h, 1 M Velcade (bortezomib) (Millennium Pharmaceuticals, Cambridge, MA) for 12 h, 1 mM H 2 O 2 (Sigma-Aldrich) for 1 h, and 10 M MG132 for 12 h. Cells were fixed with PBS, 3.0% paraformaldehyde for 15 min at room temperature and then permeabilized with 0.2% Triton X-100, 300 mM sucrose, 20 mM Hepes, pH 7.4, 50 mM NaCl, 3 mM MgCl 2 for 3 min at 4°C. HeLa cells were incubated with the indicated antibodies in blocking buffer (0.2% bovine serum albumin in PBS) for 1 h at 37°C, rinsed with PBS, incubated with purified Alexa Fluor 488-conjugated goat anti-mouse immunoglobulin G antibodies, rinsed, and mounted with Immuno-Fluore Mounting Medium (ICN Biomedicals, Costa Mesa, CA). The nuclei were visualized by Hoechst 33258 (Sigma-Aldrich) staining for 3 min at room temperature, and then after several washes with PBS, the nucleoli were stained with 0.66 mM pyronin Y (Sigma-Aldrich). Fluorescence was visualized with an inverted fluorescence microscope (DM IRBE; Leica, Wetzlar, Germany) and captured with a TCS-NT argon/krypton confocal laser microscope (Leica).
Nucleolar SUMO-1 target proteins were immunoprecipitated using anti-SUMO-1 monoclonal antibody that was incubated with protein G-Sepharose 4 Fast Flow beads (GE Healthcare/Amersham Biosciences) for 1 h at 4°C. The antibody was then linked to protein G by 3,3Ј-dithiobis(sulfosuccinimidylpropionate) cross-linker (Pierce) according to the manufacturer's instructions. Nucleolar lysates were incubated with the antibody bound to protein G beads at 4°C overnight. After extensive washing with RIPA buffer, immunoprecipitates were eluted in non-reducing Laemmli buffer and then in reducing buffer. SUMO-1 proteins were separated by 10% SDS-PAGE, stained with Coomassie Brilliant Blue (Bio-Rad), and excised in 24 slices for LC-MS/MS analysis.
Concerning immunoblotting experiments, proteins separated by SDS-PAGE were subsequently transferred onto nitrocellulose membranes (GE Healthcare/Amersham Biosciences). These membranes were incubated with specific antibodies as indicated.
For p160 Myb-binding protein 1A immunoprecipitation, NIH 3T3 cells were infected using a p160-FLAG retrovirus as described before (37). Then NIH 3T3 cell nuclear extracts (38) were adjusted to IBB buffer (10 mM Tris-HCl, pH 8, 0.2% Nonidet P-40, 150 mM NaCl) and precleared with protein G-Sepharose beads for 1 h at 4°C. The clarified supernatants were incubated with M2 anti-FLAG affinity resin (Sigma-Aldrich) overnight at 4°C. The beads were rinsed several times with IBB buffer, resuspended in Laemmli buffer, heated at 85°C, and centrifuged at 10,000 ϫ g.
In Vitro Reaction and Purification of His-SUMO-1 Target Proteins-The in vitro reaction was performed on HeLa extracts (38) as follows. 1.3 mg of HeLa nuclear extract and 6 mg of HeLa cytosolic extract were incubated with 100 g of His-SUMO-1 previously bound to Ni 2ϩ beads (Qiagen), 30 g of Ubc9, 0.5 units/ml inorganic pyrophosphatase, and 10 mM ATP in sumoylation buffer (10 mM MgCl 2 , 0.1 mM DTT, 50 mM Tris-HCl, pH 7.5) for 1 h at room temperature (39). The reaction mixture was incubated in the absence of SUMO-1 as a control. The sumoylation reactions were stopped by adding 10 mM N-ethylmaleimide and 50 mM imidazole in sumoylation buffer. After exhaustive washings, the His-SUMO-1-conjugated proteins were eluted from beads with 500 mM imidazole in 50 mM Tris-HCl, 150 mM NaCl. Proteins were separated by 10% SDS-PAGE, stained by silver staining (40), and excised in 34 slices for LC-MS/MS analysis. For histones, a mixture of calf thymus total histones (1 g) or purified histone H3 (1 g) (Sigma-Aldrich) was incubated in the presence (or absence as control) of His-SUMO-1 (1 g), Ubc9 (10 ng), Aos1/Uba2 (150 ng), 0.5 units/ml inorganic pyrophosphatase, and 10 mM ATP in sumoylation buffer for 1 h at room temperature.
Mass Spectrometry and Data Analysis-Mass Spectrometry analysis was performed using a hybrid quadrupole time-of-flight mass spectrometer (API QStar PULSAR, PE-Sciex, Toronto, Canada) equipped with a nanoelectrospray ion source (Proxeon Biosystems, Odense, Denmark). A total of 5 l of trypsin-digested sample was injected in a capillary chromatographic system Agilent 1100 Series equipped with a Nano Pump, Iso Pump, and Degasser (Agilent, Santa Clara, CA). Peptide mixtures were separated on a 10-cm fused silica capillary (75-m inner diameter and 360-m outer diameter; Proxeon Biosystems) filled with Reprosil-Pur C 18 3-m resin (Dr. Maisch GmbH, Ammerbuch-Entringen, Germany) using a pressurized "packing bomb." Peptides were eluted with a 60-min gradient from 92% buffer A (2% acetonitrile, 0.2% formic acid in water) to 80% buffer B (2% water, 0.2% formic acid in acetonitrile) at a constant flow rate of 200 nl/min. Analyses were performed in positive ion mode; the high voltage potential was set at around 1.6 -1.8 kV. Full-scan mass spectra ranging from m/z 350 to 1350 Da were collected, and for each MS spectrum, the two most intense doubly and triply charged ions peaks in the mass range were selected for fragmentation. Tandem mass spectra were extracted by Mascot.dll (version 1.6.0.21) through Analyst QS 1.1 (Applied Biosystems, Foster city, CA).
Mass spectrometry analysis was also performed by LC-MS/MS using an LTQ-Orbitrap mass spectrometer (ThermoScientific, Bremen, Germany). 5 l of tryptic digest for each band were injected in a capillary chromatographic system (EasyLC, Proxeon Biosystems). Peptide separations occurred on a homemade column as described above. A gradient of eluents A (distilled water with 2% (v/v) acetonitrile, 0.1% (v/v) formic acid) and B (acetonitrile, 2% (v/v) distilled water with 0.1% (v/v) formic acid) was used to achieve separation from 8% B (at 0 min, 0.2 ml/min flow rate) to 50% B (at 80 min, 0.2 ml/min flow rate).The LC system was connected to the orbitrap equipped with a nanoelectrospray ion source (Proxeon Biosystems). Full-scan mass spectra were acquired in the LTQ-Orbitrap mass spectrometer in the mass range m/z 350 to 1500 Da and with the resolution set to 60,000. The "lock-mass" option was used for accurate mass measurements. The four most intense doubly and triply charged ions were automatically selected and fragmented in the ion trap. Target ions already selected for the MS/MS were dynamically excluded for 60 s. All MS/MS samples were analyzed using the Mascot search engine (version 2.1.04; Matrix Science, London, UK) and X! Tandem (version 2007.01.01.1; The Global Proteome Machine Organization). X! Tandem and Mascot were set up to search the IPI_human_20081019 database (total, 73,994 sequences). QStar data were searched with a peptide mass tolerance of 100 ppm and 0.4 Da for precursor and fragment ions, respectively. Searches were performed with trypsin specificity, alkylation of cysteine by carbamidomethylation as a fixed modification, and oxidation of methionine as a variable modification. For LTQ-Orbitrap data mass tolerance was set to 5 ppm and 0.4 Da for precursor and fragment ions, respectively. Scaffold (version Scaf-fold_2_01_02, Proteome Software Inc., Portland, OR) was used to validate MS/MS-based peptide and protein identifications. Protein probabilities were assigned by the Protein Prophet algorithm (41). Protein thresholds were set to 99.0% minimum and a two-peptide minimum, whereas peptide thresholds were set to 95% minimum. The false positive rate was estimated consulting the IPI_human/decoy database (42). The estimated false positive rate was less then 4% in accordance with the Scaffold criteria selected in this work. 2

MS-
Quant, an open source program (SourceForge, Inc.), was used to extract quantitative information from the Mascot HTML database search files and to manually validate the certainty in peptide identification and peptide abundance ratio. Data analysis was also performed with the MaxQuant software (44). Mass spectra were analyzed by Mascot (version 2.2.2) against a concatenated forward and reversed version of the IPI human database (IPI.HUMAN.v3.52.decoy). The initial mass tolerance in MS mode was set to 7 ppm, and MS/MS mass tolerance was 0.5 Da. Cysteine carbamidomethylation was searched as a fixed modification, whereas N-acetyl protein and oxidized methionine were searched as variable modifications. Labeled arginine and lysine were also specified as variable modifications. The resulting Mascot ".dat" files were loaded into the MaxQuant software together with the raw data for further analysis. SILAC peptide and protein quantification was performed automatically with MaxQuant using default settings as parameters. Protein quantification was based on extracted ion chromatograms of contained peptides. Peptide assignments were statistically evaluated using a Bayesian model on the basis of sequence length and Mascot score. Peptides and proteins were accepted with a false discovery rate of less than 1%, estimated on the basis of the number of accepted reverse hits. The experiments were performed in biological triplicate (experiments I, II, and III).

Endogenous SUMO-1 Conjugates Accumulate in Nucleolar
Structures upon Proteasome Inhibition-To examine the link between SUMO-1 and the proteasome pathway, the effect of the proteasome inhibitor MG132 on the subcellular localization of SUMO-1 was analyzed. Immunofluorescence was performed on HeLa cells treated with 10 M MG132 overnight or DMSO as control. In the control cells, SUMO-1-positive staining was seen as dots dispersed throughout the nucleus; whereas after MG132 treatment, the staining accumulated in well defined structures (Fig. 1A) in agreement with previous studies (25,26). To verify the nature of such structures, a double staining of SUMO-1 and pyronin Y, a nucleolus-specific marker, was performed. As evident from Fig. 1B, in treated cells SUMO-1 appears to accumulate into nucleoli. The dynamic behavior of SUMO-1 structures during proteasome inhibition was followed in HeLa cells at different times. As shown in Fig. 2, during the treatment with MG132, SUMO-1 structures increased in size and number. These new structures appear to be highly dynamic: as the treatment progressed, larger portions of SUMO-1 structures had the propensity to fuse with each other moving into the nucleolar compartment. These events resemble the fission and fusion processes of the promyelocytic leukemia bodies during chromatin organization (45). The changes of SUMO-1 structures in HeLa cells were also observed upon several stress conditions. The immunofluorescence analysis in supplemental Fig.  1 reveals that cycloheximide (26), actinomycin D (45), and oxidative stress (46) had no effect on the nucleolar accumulation of SUMO-1 particles. Conversely the behavior of SUMO-1 structures after Velcade treatment, a potent and selective proteasome inhibitor (47), was the same as that observed upon MG132 treatment, confirming the specificity of the stimulus. This observation was confirmed by biochemical analysis of purified nucleoli. Immunoprecipitation with anti-SUMO-1 antibody was used to enrich the sumoylated proteins. The Western blot analysis of treated and untreated samples shows that SUMO-1 conjugated proteins accumulated in nucleolar extract in treated cells (Fig. 1C). A Western blot with anti-ubiquitin was performed on the same nucleolar extracts (Fig. 1D) to show that, after treatment with MG132, the level of ubiquitinated protein increased, confirming the effectiveness of the inhibition.
Proteomics Analysis of Nucleolar Sumoylated Proteins-To study the effect of proteasome inhibition on endogenous SUMO-1 targets, a quantitative proteomics analysis (SILAC) on HeLa nucleolar sumoylated proteins was performed. Two different SILAC experiments were performed: one using the isotopes Arg0 and Arg10 (the two biological replicates are named SILAC I and SILAC II) and the other using Arg0, Lys0 and Arg10, Lys8 (SILAC III). In all experiments, cells grown with the light isotopes were treated with MG132, whereas the other cell population, grown with the heavy isotopes, was treated with DMSO (Fig. 3). The two HeLa populations from each SILAC experiment were mixed in a 1:1 ratio, and nucleoli were isolated. To verify the purity of the nucleolar fraction, equal amounts of proteins from cytoplasmic, nuclear, nucleoplasmic, and nucleolar fractions were immunolabeled with specific antibodies against ␣-tubulin, lamin A/C, and nucleophosmin. The presence of the signal of nucleophosmin and the absence of lamin A/C and ␣-tubulin signals in the nucleolar fraction attest the quality of the purification (supplemental Fig. S2). To purify SUMO-1-specific targets, nucleoli were lysed and immunoprecipitated with anti-SUMO-1 antibody (Fig. 4). 10 mM N-ethylmaleimide was added in the lysis buffer to inhibit SUMO proteases that remove SUMO from target proteins, and 0.5% SDS, a strong ionic detergent, was used for the immunoprecipitation to break up protein complexes and enhance the specificity of the purification. Moreover a cross-linker carrying a disulfide bridge, 3,3Ј-dithiobis(sulfosuccinimidylpropionate), was used to link SUMO-1 antibody to

FIG. 2. Subcellular distribution of SUMO-1 after DMSO and MG132 treatment at different time points.
Shown is immunofluorescence analysis of HeLa cells cultured in the presence of DMSO as control (a, e, and i) and with 10 M MG132 for 1 h (b, f, and l), 6 h (c, g,  and m), and 12 h (d, h, and n).  protein G. The advantage of this step is the possibility to elute specific proteins and immunoglobulins in differential ways using non-reducing and reducing denaturing conditions (Fig.  3A). As expected, SUMO-1 targets were enriched after immunoprecipitation only in non-reducing conditions, indicating the validity of the purification strategy.
Eluted proteins were then separated by SDS-PAGE (Fig.  4B), and the top part of the Coomassie gel lane, which contains the majority of the SUMO conjugates, was cut in several slices. Proteins were in-gel digested by trypsin. The resulting peptide mixtures were analyzed by nLC-MS/MS, and only the proteins identified in at least in two of three biological replicates (193 proteins) were selected. A complete list of all identified proteins is available in supplemental Table S1. Because the SILAC technique can quantify changes as small as 10%, we chose 0.8 as a conservative cutoff ratio of heavy and light peptides, obtained by using MSQuant and MaxQuant softwares. The presence of structural proteins, potential contaminants of the mixture, such as Plectin 1, with an Arg10/ Arg0 or Lys8/Lys0 ratio equal to 1 attested the validity of the quantitative analysis and excluded the implication of these proteins in the SUMO-ubiquitin pathway. In Table I are listed the proteins identified by at least two arginine-or lysinecontaining peptides and enriched at least 0.8-fold in the heavy arginine or lysine form. An average of the three independently measured ratios is shown with S.D. For most potential sumoylated proteins the heavy and light peptide ratio is 0.6 -0.7 with an average S.D. of 0.06 indicating that two-peptide pairs lead to reasonably accurate quantification and attesting the good reproducibility of the experiment. Moreover the potential SUMO-1 targets are listed with their subcellular localization obtained consulting the ExPASy Proteomics Server and nucleolar database (NOPdb; Ref. 36). Several potential SUMO-1 targets are nucleolar proteins, whereas the others are enriched in nucleoli after proteasome inhibition (Fig. 4C). Interestingly we found the ubiquitin peptide containing Lys-48 modified by the Gly-Gly moiety characteristic of polyubiquitin chain formation (supplemental Fig. S3). Ubiquitin peptides of these chains notably increase (heavy/light ratio of 0.2) in high molecular weight bands, demonstrating the accumulation of polyubiquitinated proteins after proteasome inhibition (Table  I). Gene Ontology analysis of biological process was performed with the on-line software PANTHER using the data reported in Table I. As shown in Fig. 5, sumoylated targets were significantly enriched in proteins involved in ribosome biogenesis, RNA splicing and metabolism, and chromatin remodeling.
Confirming SUMO-1 Target Proteins-To validate the in vivo identified SUMO-1 targets, we performed an in vitro sumoylation reaction. In particular, we used HeLa nuclear extract as the source of SUMO-1 targets, HeLa cytosolic extract as the source of sumoylation enzymes, and an excess of His-SUMO-1, Ubc9, inorganic pyrophosphatase, and ATP as the components of the sumoylation buffer. As a control, the reaction mixture was incubated in the absence of SUMO-1. His-SUMO-1-tagged proteins were affinity purified on Ni 2ϩ beads. 10% of the eluted sumoylated targets were separated by SDS-PAGE and immunoblotted using anti-SUMO-1 antibody. As shown in Fig. 6A, the in vitro reaction sharply increased the level of sumoylated proteins. The remaining part of the purified His-SUMO-1 targets was separated by SDS-PAGE. Comparing the elution with the control, the recovery of SUMO target proteins was highly specific (Fig. 6B). The top part of the silver-stained gel lane, which contains the majority of the SUMO-1 conjugates, was cut in several slices, and proteins were in-gel digested with trypsin. The peptide mixtures were analyzed by nLC-MS/MS, identified by database search with the Mascot search engine, and validated by Scaffold software. In Table I (Table I), have been found previously in other SUMO-1 target protein screens supporting the validity of our strategy (48 -51).
Examples of known SUMO-1 target proteins are heterogeneous nuclear ribonucleoproteins such as hnRNPs K and M and the DEAD box family of RNA helicases, such as Ddx5, an interactor of transcription factors, that in its sumoylated form favors the recruitment of co-repressor thus repressing transcription (52,53). Other known SUMO-1 targets, also identified in this study, are DNA topoisomerase II, SWI/SNF protein, and histone H4 (54 -56). The presence of a sumoylation consensus motif was determined in the 78 putative SUMO-1 target proteins using SUMOPlot as shown in Table I. A total of 330 consensus sumoylation motif were found in 70 of 78 (90% of total) identified proteins. Considering the frequency of SUMO-1 sites in the Swiss-Prot database (57), the p value was estimated to be 6.13eϪ10 by the equation of hypergeometric distribution (58) indicating that proteins listed in Table I are indeed enriched in the sumoylation consensus motif, thus supporting the validity of the strategy adopted to identify endogenous SUMO-1 target proteins.
Identification of Novel SUMO-1 Target Proteins-To assess the sumoylation of novel SUMO-1 targets, we performed in vitro sumoylation reactions. It is known that several histones are sumoylated in yeast, but there is no evidence of this modification in mammals except for histone H4 (58,59). An enriched fraction of histones was incubated in the presence of His-SUMO-1, Ubc9, Aos1/Uba2, inorganic pyrophosphatase, and ATP for 1 h at room temperature. The same reaction buffer devoid of His-SUMO-1 was used as control. Moreover the mixture was incubated without histones as a further control. Proteins were separated by SDS-PAGE and analyzed by Western blot with anti-SUMO-1 antibody. As shown in Fig.  7A, there is an evident band at 45 kDa appearing only after the reaction with SUMO-1. The corresponding band was in-gel digested and identified by LC-MS/MS as SUMO-1 and histone H1 (supplemental Table S3). This is in accordance with the molecular mass of the complex. Moreover purified histone H3 was also incubated in the presence (or absence as control) of the reaction buffer described above for 1 h at 37°C, and the mixture was incubated without histone H3 as control of the reaction. Proteins were size-separated by SDS-PAGE and blotted to membranes, and sumoylated histone H3 was detected by anti-histone H3 antibody. As shown in Fig. 7B, there is an evident band at about 35 kDa due to histone H3 bound to one molecule of SUMO-1 that is not present in the control. The analysis of the Western blot with anti-SUMO-1 confirmed the transfer of the SUMO-1 moiety from Ubc9 to the sub-

Nucleolar SUMO-1 Targets
strate. The presence of sumoylated histone H3 was also confirmed by LC-MS/MS analysis of the band described above (supplemental Table S3).
Another novel SUMO-1 substrate, identified for the first time in this study, is the p160 Myb-binding protein 1A. p160 is mainly a nucleolar protein and, as reported in recent studies, may regulate ribosome biogenesis and Myb-dependent transcription (37,60). To analyze the endogenous sumoylation of p160, NIH 3T3 cells were infected with p160-FLAG. Immunoprecipitation with anti-FLAG antibody and immunoblotting with anti-SUMO-1 indicated that p160 is indeed sumoylated (Fig. 7C). DISCUSSION As emerging from recent studies (61), SUMO-2/-3 modification may regulate the ubiquitin-proteasome system. One explanation is based on the discovery of novel ubiquitin ligases that mediate the targeting of sumoylated proteins to the proteasome (27). In this study we demonstrated that a crosstalk also exists between SUMO-1 and the ubiquitin-proteasome system. In contrast to SUMO-2/-3 proteins and ubiquitin itself, SUMO-1 does not form polychains. Such a peculiarity together with its endogenous low level in the cell makes this modification more difficult to detect. As a consequence, in several conditions, the effect of any stimulus, such as the inhibition of the proteasome system, seems to affect SUMO-1 much less than SUMO-2/3 (61). Interestingly we observed that upon MG132 treatment there is a complete redistribution of SUMO-1 targets from nuclear dots into nucleolar structures. This process was highly specific with respect to SUMO-1 behavior under several stress conditions suggesting that SUMO-1 may play a major role in the nucleolar compartment that is linked to the inhibition of the proteasome system. To better understand this biological event and overcome the problem of the low detection of SUMO-1, we focused our analysis on the nucleolus. Although a large frac- FIG. 5. Biological process analysis. Analysis was performed with the on-line software PANTHER using the data set reported in Table I. The p value was set at Ͼ0.05. The Bonferroni correction for multiple testings was used. Only categories with significant differences are shown.
FIG. 6. Purification of in vitro His-SUMO-1 target proteins. An in vitro reaction of HeLa nuclear extract was performed with His-SUMO-1, which was previously bound to Ni 2ϩ beads. The reaction mixture was obtained by incubating Ubc9, inorganic pyrophosphatase, and ATP in sumoylation buffer with and in the absence of SUMO-1 as control. A, anti-SUMO-1 Western blot (WB) of purified His-SUMO-1-conjugated proteins shows an enrichment of conjugated species. B, SDS-PAGE of purified His-SUMO-1 proteins stained by silver shows a very specific signal in the reaction mixture compared with the control (ctrl). The gel was excised in 34 slices for nLC-MS/MS analysis. tion of SUMO-1 targets is supposed to reside in the nucleolus, no characterization of these proteins has been performed until now. In this work, we identified 193 SUMO-1 nucleolar targets by a proteomics approach. In addition, by quantitative analysis, we found that 78 of these substrates change their level of sumoylation in response to proteasome inhibition. These results confirm that there is a relationship between SUMO-1 and the ubiquitin-proteasome system suggesting that SUMO-1, together with ubiquitin, may ensure the integrity of nucleolar organization, acting as a second level of quality control in the regulation of ubiquitin-dependent proteolysis. To increase our knowledge of SUMO-1 in this compartment, we performed an analysis of the nature of the substrates influenced by proteasome inhibition and found proteins involved in ribosome biogenesis, protein complex assembly, RNA splicing and metabolism, chromatin packaging and remodeling, and DNA replication (Fig. 5). As already supposed in budding yeast (62), we found clustering of SUMO modification among subunits of multiprotein complexes indicating that this modification may have a specific cooperative activity. For example, concerning the regulation of ribosome biogenesis, we found several sumoylated elongation factors, such as hnRNP proteins, RNA helicases, and ribosomal subunits, suggesting that SUMO-1 modification may regulate the assembly of these macromolecular complexes. In particular, these substrates changed upon MG132 treatment, suggesting that SUMO-1 may target unassembled ribosomal proteins to the ubiquitin-proteasome system in analogy to what has been demonstrated in yeast (63). This is supported by the observation that a significant fraction of ribosomal proteins imported in the nucleolus is degraded and not assembled into the ribosome subunits (64). SUMO-1 could ensure the correct building of such large complexes by inhibiting incorrect interactions between proteins. The potential link between SUMO-1 and ubiquitin is further supported by the observation that among identified SUMO-1 targets ubiquitin was present with a high confidence score and high value of fold change after MG132 treatment in concordance with previous studies that have shown the presence of ubiquitin within nucleoli (65).
Another evidence of multisumoylated complexes comes from the identification of SUMO-1 targets such as lamin A, nucleophosmin, topoisomerase II, histone H1, p160 Mybbinding protein, and several ribosomal subunits, which belong to a macrocomplex containing CTCF (CCCTC-binding factor) protein (54, 60, 66 -68). Some of these proteins are known to be sumoylated, whereas in this work we demonstrated that FIG. 7. Novel SUMO-1 target identification. A, histones mixture in vitro reaction. A mixture of calf thymus total histones was incubated in the presence or absence (as control) of His-SUMO-1, Ubc9, Aos1/Uba2, inorganic pyrophosphatase, and ATP in sumoylation buffer. The reaction mixture was analyzed by anti-SUMO-1 Western blot (WB) and SDS-PAGE. When SUMO-1 is added to the mixture, a new band of 45 kDa appears, and the amount of free SUMO-1 decreases indicating that SUMO-1 has been conjugated to one of the histones of the mixture. The Coomassie-stained band possibly corresponding to sumoylated histone was cut, trypsin-digested, and analyzed by nLC-MS/MS. This analysis identified histone H1.3 (supplemental Table S3) as the target of the sumoylation reaction. The identification of SUMO-1 protein confirms that the histone is sumoylated because the molecular mass of the band corresponds to the mass of the histone H1 plus SUMO-1. B, histone H3 in vitro reaction. Purified histone H3 was incubated in the presence or absence of SUMO-1 in the same reaction buffer described above, and the mixture was incubated without histone H3 as a control of the reaction. Proteins were separated by SDS-PAGE. Sumoylated histone H3 was detected by specific antibody as a band migrating at about 35 kDa, which was not present in the control. Western blot with anti-SUMO-1 confirms the transfer of SUMO-1 to the substrate. The sumoylation of histone H3 was also confirmed by nLC-MS/MS analysis of the band described above. * indicates a probable histone H3 dimer; ** indicates the complex Ubc9-SUMO-1. C, endogenous sumoylation of p160. NIH 3T3 cells were infected with p160-FLAG retrovirus. Immunoprecipitation (IP) with anti-FLAG affinity resin and immunoblotting with anti-SUMO-1 indicate that p160 is indeed sumoylated. Immunoprecipitation of not infected cells was used as control.
p160 Myb-binding protein and Histones H1 and H3 are indeed modified by SUMO-1. These findings suggest that SUMO-1 modification may regulate the organization of this complex, participating in the transcription of rRNA genes, ribosome maturation, assembly, and transport. Recently it has also been demonstrated in Drosophila that ribosomal proteins interact with histone H1 on condensed chromatin, supporting the possibility that the association of ribosomal proteins on chromatin may be part of their assembly/maturation process (43).
In summary, we suggest that sumoylation may be one of the key regulators linking these cellular processes and that this post-translational modification plays an important and specific role in the ubiquitin-proteasome system. Further investigations on the biochemistry and cell biology of SUMO-1 target proteins identified in this work should help to elucidate further aspects of the inter-relationship of SUMO-1 target proteins and proteasome activity.