Analysis of BCL6-interacting Proteins by Tandem Mass Spectrometry*S

B-cell lymphoma 6 (BCL6) is a 95-kDa nuclear phosphoprotein and member of the Pox virus zinc finger/bric-a-brac, tramtrack, broad complex (POZ/BTB) family of transcription factors. BCL6 is a transcriptional repressor required for germinal center formation, and the gene encoding it is frequently altered in diffuse large B-cell and follicular lymphomas. The dysregulation of BCL6 has therefore been implicated in lymphomagenesis. A limited number of proteins is known to interact with BCL6 and modulate its activity or participate in its role in transcriptional regulation. Identification of additional BCL6-binding proteins could reveal potential signaling targets and previously undescribed functional roles for BCL6. We used a functional proteomic approach to determine the identity of proteins that interact with BCL6. Proteins were isolated by co-immunoprecipitation with an anti-BCL6 antibody and identified using MS/MS. We identified 61 proteins in the BCL6 immunocomplex from the following Gene Ontology categories: transcription regulator activity (n = 18), binding activity (n = 11), signal transducer activity (n = 10), catalytic activity (n = 8), structural molecule activity (n = 3), enzyme regulator activity (n = 3), transporter activity (n = 2), motor activity (n = 2), chaperone activity (n = 1), and unknown function (n = 3). Importantly we identified BCL6 and several previously reported BCL6-interacting proteins in the BCL6 immunocomplex. The remaining proteins have not been shown previously to be associated with BCL6. MS/MS results were validated on four proteins using immunoprecipitation and Western blotting. Two of these protein interactions were further confirmed by reciprocal immunoprecipitation. This study demonstrates the utility of antibody immunoprecipitation and subsequent peptide identification by MS/MS for the elucidation of BCL6-binding proteins. Many of the novel proteins identified in this study suggest additional functional roles for BCL6 beyond transcriptional repression.

finger domains are likely to bind additional proteins that may further define and expand the role of BCL6 in the cell. Furthermore there are additional proteins that complex with BCL6 via indirect or secondary interactions. Identification of the complete interactome of BCL6 would provide additional information regarding the function and regulation of this oncogene.
In this study, we report the identification of multiple BCL6interacting proteins via immunoprecipitation followed by LC-MS/MS peptide sequencing (Fig. 2). In addition to confirmation of previously described BCL6-binding proteins, we expand the list of BCL6-interacting proteins to include additional transcription regulation factors. Furthermore we identified potential BCL6-interacting proteins from other functional activity groups including binding, signal transduction, catalysis, structural molecule activity, enzyme regulation, transport, motor, and chaperone activity.
Cell Lysis, Immunoprecipitation, and Western Blot Analysis-Immediately upon thawing, cells were lysed using 0.5 ml of lysis buffer (25 mM Tris-HCl, 1% Triton X-100, 1% sodium deoxycholate, 0.15 M NaCl, 1 mM EDTA) per pellet. Before each use, 0.1% protease inhibitor mixture (Sigma) was added to the required volume of lysis buffer. Protein concentrations were estimated using the Bradford colorimetric method against known concentrations of bovine serum albumin.
For immunoprecipitation, lysate containing 800 g of total protein was mixed with either 2 g of primary anti-BCL6 (C-19) polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) or normal rabbit IgG in a total volume of 1 ml. The mixtures were incubated with gentle shaking for 1 h at 4°C. Protein G-agarose beads were then added to each tube, and the samples were incubated in a similar fashion. The immunocomplex was washed five times with cold lysis buffer, and the antibody-selected proteins were eluted from the agarose beads by boiling in SDS loading buffer (0.1 M Tris-HCl, 10% glycerol, 2% SDS, FIG. 2. Experimental design. The BCL6 protein complex was isolated by antibody immunoprecipitation, and the proteins were separated by SDS-PAGE. The entire lane was submitted, in 12 separate gel slices, for trypsin digestion and analysis by ion trap tandem mass spectrometry. FIG. 3. Immunoprecipitation of BCL6. Proteins immunoprecipitated from total cell lysates of SUDHL-4 and SUDHL-9 cell lines using a polyclonal anti-BCL6 antibody or rabbit IgG control were divided and resolved on parallel 8% SDS-polyacrylamide gels. A, silver staining of one gel demonstrated a 95-kDa band corresponding to BCL6 that was present in the anti-BCL6 immunoprecipitate of SUDHL-4 lysates but not in the IgG immunoprecipitation control or the anti-BCL6 immunoprecipitate of SUDHL-9 lysates. B, after electrophoresis, the proteins in the second gel were transferred to nitrocellulose membrane and immunoblotted using anti-BCL6 monoclonal antibody. A band at 95 kDa corresponding to BCL6 was identified in the SUDHL-4 total cell lysate but not the SUDHL-9 or IgG control lanes. FIG. 1. BCL6-interacting proteins. The diagram shows the conserved motifs that mediate BCL6 protein interactions with the proteins listed under each domain. BCL6 directly interacts with several proteins via the POZ/BTB domain, the zinc finger domain, or both. p300 binds and acetylates BCL6, targeting it for degradation. 0.05% bromphenol blue, and 0.1 M DTT) for 10 min. Each sample was divided equally, and each half was resolved on separate 8% SDSpolyacrylamide gels. On one gel, the proteins were visualized with mass spectrometry-compatible silver staining (Invitrogen). Separated proteins were transferred from the other gel to a nitrocellulose membrane (Millipore, Billerica, MA) for immunoblotting using semidry transfer. After blocking overnight at 4°C with 5% milk, the membrane was incubated at room temperature with a 1:200 dilution of monoclonal anti-BCL6 antibody (DakoCytomation, Carpinteria, CA) for 1 h, then washed, and incubated with horseradish peroxidase-conjugated secondary antibody (Santa Cruz Biotechnology). Western blot targets were visualized using chemiluminescence (ECL Plus kit, Amersham Biosciences).
Validation by Western Blot and Reciprocal Immunoprecipitation-For validation experiments, immunoprecipitations were performed using SUDHL-4 lysates containing 800 g of total protein, anti-BCL6 polyclonal antibody, and normal rabbit IgG control. Immunoprecipitated proteins were resolved and transferred, and Western blots were performed using antibodies against HDAC4, SMRT, PKC, and WNK4 (Santa Cruz Biotechnology). Reciprocal immunoprecipitations were performed using SUDHL-4 lysates (800 g of total protein) and antibodies against HDAC4 and PKC. Immunoprecipitated proteins were resolved by SDS-PAGE, and BCL6 was detected by immunoblotting.
Digestion, Peptide Extraction, and Mass Spectrometric Analysis-Each lane of interest on the silver-stained gel was isolated and divided into 12 equal slices. The proteins in each slice were digested, and the peptides were extracted using the Invitrogen protocol. Briefly each gel slice was destained, washed, crushed, dried, and then rehydrated in ammonium bicarbonate buffer. Freshly prepared sequencing grade lysine-C endopeptidase (Princeton Separations, Ad-FIG. 4. Protein identification by tandem mass spectrometry. MS/MS sequencing of peptides identifying the proteins BCL6 (A), BCL11 (B), and PKC-(C). The MS full scan is shown for each sample followed by data-dependent MS/MS sequencing scans of a top peptide candidate. elphia, NJ) was added (1:50) to each tube, and the tubes were incubated at 37°C for 4 h. Sequencing grade, modified trypsin (Princeton Separations) was then added (1:50), and the tubes were incubated at 37°C overnight.
The digested peptides were extracted from the gel pieces using 50% acetonitrile with 0.1% TFA and evaporated to a final volume of 30 l. Duplicate 15-l aliquots of each sample were analyzed by automated nanoflow reverse-phase LC-MS/MS using the LCQ Deca XP ion trap mass spectrometer (ThermoFinnigan, San Jose, CA). Digested peptides were injected by an autosampler using an acetonitrile gradient (100% Buffer A to 40% Buffer A, 60% Buffer B in 90 min; Buffer A ϭ 5% acetonitrile with 0.4% acetic acid and 0.005% heptafluorobutyric acid; Buffer B ϭ 95% acetonitrile with 0.4% acetic acid and 0.005% heptafluorobutyric acid) through a reverse-phase column (75-m inner diameter fused silica packed with 12 cm of 5-m C 18 particles) to elute the peptides at a flow rate of 200 nl/min into the mass spectrometer. An electrospray voltage of 1.8 kV was used with the ion transfer tube temperature set to 200°C. Peptide analysis was performed using data-dependent acquisition of one MS scan (600 -2000 m/z) followed by MS/MS scans of the three most abundant ions in each MS scan. Dynamic exclusion was used to acquire a more complete survey of the peptides contained in the BCL6 antibody pull-down complex by automatic recognition and temporary exclusion of ions from which definitive mass spectral data

FIG. 4-continued
Analysis of BCL6-interacting Proteins had been acquired previously. Dynamic exclusion was set to a repeat count of two with the exclusion duration of 3 min.
Data Analysis-The MS/MS-acquired data were searched using the SEQUEST algorithm in Bioworks 3.1 (ThermoFinnigan) against amino acid sequences in the UniProt protein database (September 27, 2004 download;190,183 entries). Protein search parameters included a precursor peptide mass tolerance of Ϯ0.7 amu, fragment mass tolerance of Ϯ0.1 amu, and carboxymethylation for cysteine residues. The search was constrained to tryptic peptides with one missed enzyme cleavage allowed. The peptide matching criteria of a cross correlation score (X corr ) Ͼ1.2 for ϩ1 peptides, Ͼ2.2 for ϩ2 peptides, and Ͼ3.2 for ϩ3 peptides and a ⌬ correlation score (⌬C n ) Ͼ0.100 were used as a threshold of acceptance. Other factors were also taken into consideration such as the number of unique peptides found per protein, the total number of matching y and b series ions, the ⌬ correlation score from the next nearest hit, any known conserved domain protein interactions, and feasible molecular signal transduction interactions.
Finally the SEQUEST data (.dta) and output (.out) files from duplicate aliquots were combined and analyzed using INTERACT and ProteinProphet (Institute for Systems Biology, Seattle, WA) (31,32). Data analysis using INTERACT and ProteinProphet improved the confidence of protein identification by best fit distribution of probability scores specific to each data set and reduced the risk of false positive protein identifications. All proteins passing a threshold of less than 5% predicted error (false positive rate) were summarized.

FIG. 4-continued
In Silico Analysis-We used several search methodologies in publicly available protein interaction databases (Online Mendelian Inheritance in Man (OMIM) and Human Protein Reference Database (HPRD)) to create a list of primary (direct) and secondary (indirect) protein-protein interactions centered on BCL6. The OMIM is hosted by the National Center for Biotechnology Information (NCBI) and contains information on 10,281 genes specifically associated with diseases in man as summarized from published literature. The OMIM search page is found at www.ncbi.nlm.nih.gov/entrez/query. fcgi?db ϭ OMIM. The HPRD (33) is a joint project combining the efforts of Bangalore's Institute of Bioinformatics and PandyLab. As of August 2005, the database contained 18,288 human protein entries and 27,330 documented protein-protein interactions with nearly 171,677 links to PubMed literature. HPRD can be found at www. hprd.org.
By searching each database first for direct BCL6 protein interactions then for any secondary BCL6 interactions, a summary of the BCL6 protein "interactome" was compiled. In addition, primary and secondary protein interactions with BCL6 were further screened by a literature survey using PubMed. Finally these in silico findings were compared with the protein interactions identified by BCL6 pull-down purification and tandem mass spectrometry.

Identification of BCL6-interacting Proteins-
To determine the identity of proteins associated with BCL6, we performed immunoprecipitation of total cell lysates from the BCL6-expressing cell line SUDHL-4 using anti-BCL6 polyclonal antibody. Immunoprecipitation reactions using the non-BCL6expressing SUDHL-9 cell line and normal rabbit IgG served as separate negative controls. A band at 95 kDa corresponding to BCL6 was identified in the SUDHL-4 BCL6 pull-down lane but not in the SUDHL-9 or IgG control lanes on a silverstained gel (Fig. 3A) and a Western blot (Fig. 3B).
Duplicate MS/MS analysis of the SDS-PAGE lanes obtained from BCL6 immunoprecipitation resulted in a total of 1183 peptides, which were matched to 726 known database entries and 139 proteins. The average size of identified peptides was 12 amino acid residues (range, 7-29). The MS full scan and MS/MS scan are shown for peptides identifying BCL6, BCL11A (previously reported to interact with BCL6), and PKC-(previously unreported interaction), although the predicted error rate for PKC-was above the 5% cut-off (Fig. 4). The peptide sequence identified in each scan is underlined in the complete amino acid sequence of the protein.
To select for proteins that interact specifically with the BCL6 protein and not those that associate with the IgG in a nonspecific manner, we subtracted those proteins that were common between the IgG control immunoprecipitation lane (Supplemental Table 1) and the BCL6 immunocomplex lanes. One protein was identified as unique to the SUDHL-9 anti-BCL6 immunocomplex (Supplemental Table 2) and was also subtracted from the final summary of BCL6-interacting proteins. Thus, a total of 139 unique proteins were identified by BCL6 antibody pull-down and tandem mass spectrometry. After subtraction of proteins identified in control lanes, 61 BCL6-interacting proteins were identified with less than 5% error (Table I). These 61 proteins, specific to the BCL6 immunocomplex, are from the following Gene Ontology functional groups: transcription regulator activity (n ϭ 18), binding activity (n ϭ 11), signal transducer activity (n ϭ 10), catalytic activity (n ϭ 8), structural molecule activity (n ϭ 3), enzyme regulator activity (n ϭ 3), transporter activity (n ϭ 2), motor activity (n ϭ 2), chaperone activity (n ϭ 1), and unknown function (n ϭ 3) (Table I). Importantly BCL6 was identified among these proteins along with the previously reported BCL6-interacting proteins BCL11A (29), HDAC5, (15), HDAC9, N-CoR2/SMRT (18 -20), and mSin3A (19,22). The remaining 55 proteins have not been reported previously in the literature to interact with BCL6. A complete listing of identified peptides and corresponding X corr scores is provided in Supplemental Table 3. Interestingly several additional proteins known to interact with BLC6 were identified (HDAC2 (23), HDAC4, HDAC7 (15), and N-CoR1 (18)), but the predicted error rates for their identifications exceed the 5% false positive rate threshold.
Validation of MS/MS Data by Western Blotting and Reciprocal Immunoprecipitation-To further validate the authenticity of the proteins identified by MS/MS, we confirmed the presence of some selected proteins within the BLC6 immunocomplex using Western blotting analysis. After anti-BCL6 immunoprecipitation of SUDHL-4 lysates, Western blotting demonstrated bands correlating with two proteins reported previously to interact with BCL6 (HDAC4 and N-CoR2/SMRT) and two novel interactions (PKCand WNK4) (Fig. 5A). The bands were similar to those present in total cell lysates but were not seen in IgG immunoprecipitation controls. In addition, we performed reciprocal immunoprecipitations using SUDHL-4 lysates and polyclonal antibodies to HDAC4 and PKC. Immunoprecipitated proteins were resolved using SDS-PAGE, and bands correlating to BCL6 were identified by Western blotting in both pull-downs (Fig. 5B). Although HDAC4, PKC-, and WNK4 are below the error rate cut-off, these validation experiments suggest that they do interact with BCL6, and they are therefore included in Table I.
In Silico Analysis of BCL6-interacting Proteins-To validate potential BCL6-interacting proteins, we carried out in silico proteomic analysis of proteins that have been documented to interact, either directly or indirectly, with BCL6. Using a combination of the publicly available databases OMIM, HPRD, and PubMed, a total of 196 proteins were identified that have been reported to interact either directly or indirectly with BCL6 (Table II). These in silico protein interactions included 19 primary, or direct, interactions with BCL6. The remainder represent indirect interactions with BCL6. For example, HDAC4 is known to interact directly with BCL6, whereas ERK2 has not been reported to interact with BCL6 but does have interactions with HDAC4. Thus, both HDAC4 and ERK2 were summarized as potentially belonging to the BCL6 interactome as primary and secondary interactions, respectively.
Combining in Vitro and in Silico Strategies-A total of five proteins identified by BCL6 antibody pull-down also had reported direct or indirect interactions with BCL6. The previously reported direct BCL6 protein interactions identified in our experiment and BCL6 are bolded in Table I. The remaining 55 proteins not found in the database entries or reported literature may represent novel direct or indirect BCL6 protein interactions.

DISCUSSION
This study represents the first global, high throughput functional proteomic approach to determine the identity of proteins that may interact directly or indirectly with BCL6 in multimeric protein complexes. Our analysis of the anti-BCL6 pull-down revealed a total of 61 proteins that were not seen in pull-downs from non-BCL6-expressing cells (SUDHL-9) or using rabbit isotype antibody controls. Importantly BCL6 was identified along with several proteins reported to bind to BCL6, including BCL11A (29), HDAC5 (15), HDAC9, mSin3A (19,22), and SMRT/N-CoR2 (18,20). In addition, numerous proteins not known previously to be associated with BCL6 were identified.
Although we report a large number of proteins, we did identify several proteins whose interaction with BCL6 has already been described. Interestingly these known BCL6-interacting proteins were distributed throughout our list, not just among the proteins with the lowest predicted error. In addition, we confirmed using Western blotting, immunoprecipitation, and reciprocal immunoprecipitation our findings of BCL6 interactions with three proteins that fall below the 5% cut-off.
For example, HDAC4 has been shown previously to interact with BCL6, and we identified it within the BCL6 interactome in our studies albeit at a false positive rate of 17.1%. We further validated the presence of HDAC4 in the BCL6 pull-down by FIG. 5. Western blot and reciprocal immunoprecipitation validation. A, proteins immunoprecipitated from SUDHL-4 lysates using anti-BCL6 were subjected to Western blotting using antibodies to the previously reported BCL6-interacting proteins HDAC4 and N-CoR2/ SMRT and the novel interacting proteins PKC-and WNK4. The bands were similar to those present in total cell lysates but were not seen in IgG immunoprecipitation controls. B, proteins immunoprecipitated from SUDHL-4 lysates using polyclonal antibodies to HDAC4 (previously reported BCL6-interacting protein) and PKC (novel interacting protein) were subjected to Western blotting using anti-BCL6. Bands correlating to BCL6 were identified by Western blotting in both pull-downs but not in the IgG control. IP, immunoprecipitation; WB, Western blot.
BCL6 is known to associate with the transcription factors c-Jun, JunB, and JunD (27); Sp1 (28); and BCL11A (29), which was identified in the current study. Interestingly JunB was identified but with a predicted false positive rate exceeding 20%. Previously unreported BCL6-interacting transcription factors identified in this study include ATF-7, early B-cell factor, Elf-1, heat shock factor protein 4, hepatocyte nuclear factor-1 ␣, and YY1. In addition to sequence-specific transcriptional repression, BCL6 can mediate repression at promoters that do not contain BCL6 binding sites via interactions with transcription factors that do bind the promoter. For example, the interaction of BCL6 with c-Jun, JunB, and JunD is believed to explain BCL6-mediated repression at AP-1 sites (27). Therefore, the identification of additional transcription factors associated with BCL6 may lead to identification of additional targets of BCL6 in transcriptional regulation.
Although BCL6 is known to interact with transcription factors and play a role in chromatin remodeling, this study identified proteins with more diverse cellular functions within the BCL6 immunocomplex (Fig. 6). These included proteins with signal transducer activity such as NF-B-repressing factor, LPA2, P2Y9/LPA4, EphB6 precursor, and Smoothened homolog precursor. The kinases identified here could potentially participate in regulation of BCL6 similar to mitogen-activated protein kinase, which phosphorylates and targets BCL6 for   II-continued   JunD   ATF3, B-ATF, BCL6, BRCA1, protein fosB, c-Fos, CHOP, COP9, FRA2, JNK1, JNK2, Menin, NFE2L1,  NFE2L2, S100 CBP A7, SMAD3, SMAD4, TATA box-binding protein, TRC8, ubiquitin carboxyl-terminal  esterase L1   mSin3A   CBF1-interacting co-repressor, CD8, eukaryotic translation initiation factor 3, HDAC1, HDAC2, histone 2B,  histone 3, inhibitor of growth 1, methyl-CpG binding domain protein, MTA2, N-CoR1, N-CoR2, RBP1,  RBP4, RBP7, SAP18,  degradation (30). Because BCL6 is known to associate with proteins that modify chromatin, it was interesting to identify proteins that directly modify DNA (XP-G) as well as RNA (RNA helicase, ribonuclease III, and LSm1). Our proteomic analysis of BCL6-interacting proteins identified several proteins previously shown to interact with BCL6. Of additional significance, our study also identified a number of proteins that have not been reported previously or found by in silico analysis. These may represent novel proteins that interact with BCL6 and may be regulators of its function. Alternatively many of these proteins may not form direct interactions with BCL6 but may participate in multimeric complexes with BCL6. Many of the database interactions found by in silico analyses were compiled from yeast two-hybrid systems, which report interactions between pairs of proteins with no account for weaker interactions between proteins that can be stabilized by the presence of additional binding proteins in multimeric protein complexes. Our results indicate that a combinatorial approach involving immunoprecipitation-LC-MS/MS followed by validation using orthogonal and in silico methods is useful for identifying direct and indirect protein interactions.
In summary, we identified multiple novel protein interactions with BCL6. This should lead to a more complete understanding of transcriptional repression by BCL6. More importantly, we identified BCL6-interacting partners from diverse classes of proteins that may lead to an expansion of the known cellular roles and/or regulation of BCL6. This information will improve the understanding of BCL6 function and could lead to the identification of potential therapeutic targets in BCL6-associated lymphomas.