Transcriptional signature with differential expression of BCL6 target accurately identifies BCL6-dependent diffuse large

Jose M. Polo*, Przemyslaw Juszczynski†, Stefano Monti‡, Leandro Cerchietti*, Kenny Ye§, John M. Greally¶, Margaret Shipp†ʈ, and Ari Melnick*,**

Departments of *Developmental and Molecular Biology, §Epidemiology and Biostatistics, and ¶Medical Genetics, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461; †Department of Medical Oncology, Dana–Farber Cancer Institute, 44 Binney Street, Boston, MA 02115; and ‡Broad Institute, 320 Charles Street, Cambridge, MA 02141

Communicated by Matthew D. Scharff, Albert Einstein College of Medicine, Bronx, NY, December 21, 2006 (received for review November 20, 2006) Diffuse large B cell lymphomas (DLBCLs) often express BCL6, a resemble either B cell category (Other) (10). Although GCB transcriptional repressor required for the formation of normal DLBCLs had more abundant BCL6 transcripts, there was no germinal centers. In a subset of DLBCLs, BCL6 is deregulated by association between BCL6 genetic abnormalities and this tumor chromosomal translocations or aberrant somatic hypermutation; in subset. other tumors, BCL6 expression may simply reflect germinal center More recently, we applied consensus clustering methods to the lineage. DLBCLs dependent on BCL6-regulated pathways should transcriptional profiles of two large independent series of pri- exhibit differential regulation of BCL6 target genes. Genomic array mary DLBCLs to identify the dominant substructure a priori ChIP-on-chip was used to identify the cohort of direct BCL6 target (i.e., to classify DLBCLs in an unbiased manner) (11). The genes. This set of genes was enriched in modulators of transcrip- obtained consensus clusters were highly reproducible and in- tion, chromatin structure, ubiquitylation, cell cycle, and cluded three groups of DLBCLs, termed B cell / DNA damage responses. In primary DLBCLs classified on the basis proliferation (BCR), oxidative phosphorylation (OxPhos), and of expression profiles, these BCL6 target genes were clearly host response (HR) tumors; these DLBCL subsets were unre- differentially regulated in ‘‘BCR’’ tumors, a subset of DLBCLs with lated to the developmentally defined COO tumor groups (11). increased BCL6 expression and more frequent BCL6 translocations. HR tumors are defined, in part, by their brisk host inflammatory/ In a panel of DLBCL cell lines analyzed by expression arrays and immune response and histologic and clinical similarities to the classified according to their profiles, only BCR WHO pathologic subtype, T cell/histiocyte-rich LBCL. HR tumors were highly sensitive to the BCL6 peptide inhibitor, BPI. tumors rarely exhibit the genetic lesions seen in other DLBCLs These studies identify a discrete subset of DLBCLs that are reliant (11, 12). In contrast, OxPhos DLBCLs have increased expression on BCL6 signaling and uniquely sensitive to BCL6 inhibitors. More of genes involved in oxidative phosphorylation and mitochon-

broadly, these data show how genome-wide identification of drial function and more common structural abnormalities of CELL BIOLOGY direct target genes can identify tumors dependent on oncogenic intrinsic and extrinsic apoptotic pathway components (11, 12). factors and amenable to targeted therapeutics. BCR tumors have increased expression of cell cycle regulatory genes, components of the BCR signaling cascade, and certain B targeted therapy ͉ transcriptional repression ͉ ChIP on ChIP ͉ integrative cell-specific transcription factors such as BCL6; these DLBCLs analysis ͉ gene expression profiling also exhibit more frequent translocations of the BCL6 (11, 12). CL6 is a BTB/POZ domain transcription repressor that is We predicted that differential regulation of BCL6 target genes Brequired for normal germinal center (GC) development and would identify tumors specifically driven by BCL6. We postu- expressed by the majority of normal GC B cells and a subset of lated that among DLBCLs, BCR tumors would be more likely to rely on deregulated BCL6 expression and be uniquely sensitive diffuse large B cell lymphomas (DLBCLs) (1, 2). BCL6 favors to BPI treatment. For these reasons, we used a chromatin the survival and proliferation of GC B cells, which undergo immunoprecipitation (ChIP)-on-chip approach to identify BCL6 somatic hypermutation of Ig variable regions and Ig class switch target genes in a B cell cell line and asked whether recombination; down-regulation of BCL6 is necessary for these BCL6 target genes contributed to the signature of a specific post-GC B cell maturation (3–5). Deregulation of BCL6, by DLBCL subset. Here, we demonstrate that BCR DLBCLs chromosomal translocation or aberrant somatic hypermutation, exhibit coordinate regulation of the identified BCL6 target is the most common genetic abnormality in DLBCL (6). Con- clusive evidence for the oncogenic role of BCL6 comes from

murine models in which constitutive BCL6 expression results in Author contributions: J.M.P. and P.J. contributed equally to this work; J.M.P., P.J., S.M., the development of a lymphoid malignancy resembling DLBCL J.M.G., M.S., and A.M. designed research; J.M.P., P.J., S.M., and L.C. performed research; (7, 8). Although deregulated BCL6 clearly plays a pathogenetic J.M.P., P.J., S.M., L.C., K.Y., J.M.G., M.S., and A.M. analyzed data; and J.M.P., P.J., S.M., M.S., role in a subset of human DLBCLs, other DLBCLs may simply and A.M. wrote the paper. express this transcriptional repressor because the lymphomas are The authors declare no conflict of interest. derived from normal BCL6ϩ GC B cells. Identification of Abbreviations: ABC, activated peripheral blood B cells; BCR, B cell receptor/proliferation; BCL6-dependent tumors has important therapeutic implications BPI, BCL6 peptide inhibitor; COO, cell of origin; DLBCL, diffuse large B cell lymphoma; ES, enrichment score; GC, germinal center; GC B cells, germinal center B cells; GO, Gene because a recently described specific BCL6 peptide inhibitor Ontology; GSEA, gene set enrichment analysis; HPRT, hypoxanthine-guanine phosphori- (BPI) inhibits the growth of some but not all DLBCLs (9). bosyl transferase; HR, host response; OxPhos, oxidative phosphorylation; Q-ChIP, quanti- To delineate functionally relevant DLBCL subsets, we and tative-PCR ChIP; SNR, signal-to-noise ratio. others have used gene expression signatures. In an earlier ʈTo whom correspondence may be addressed. E-mail: margaret࿝[email protected]. approach known as the cell of origin (COO) classification, **To whom correspondence may be addressed. E-mail: [email protected]. subsets of DLBCLs were associated with specific types of normal This article contains supporting information online at www.pnas.org/cgi/content/full/ B cells [GC B cells (GCB) or in vitro activated peripheral blood 0611399104/DC1. B cells (ABC)] or left unassigned if the tumors did not closely © 2007 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0611399104 PNAS ͉ February 27, 2007 ͉ vol. 104 ͉ no. 9 ͉ 3207–3212 Downloaded by guest on September 25, 2021 genes. In addition, the BCL6 signature and BCR subtype designation have important functional consequences because only BCR DLBCL growth is inhibited by BPI treatment. The BCL6 target gene signature provides important insights into the biology of BCR DLBCLs and identifies these tumors as candi- dates for rational targeted BCL6 inhibition. Results and Discussion Identification of BCL6 Target Genes. We predicted that BCL6- dependent DLBCLs would have a transcriptional signature that was defined, at least in part, by the differential expression of BCL6 target genes. To identify such genes, we performed high-throughput ChIP on chip in the Ramos B cell lymphoma cell line, which is frequently used to evaluate BCL6 function (13–15). Chromatin fragments were immunoprecipitated with an antibody directed against BCL6 or an irrelevant control (actin), and the resulting products were amplified by ligation-mediated PCR (LMPCR). Specific enrichment of BCL6 target genes was validated by single-locus quantitative-PCR ChIP (Q-ChIP) be- fore and after LMPCR. Thereafter, the resulting amplicons were labeled and cohy- bridized with input chromatin to high-density oligonucleotide arrays containing a 1.5-kb sequence of 24,275 gene promoters, each of which was represented by 15 consecutive 50-mer oligo- nucleotides. ‘‘Hits’’ were captured through a highly stringent approach employing random permutation analysis on a sliding window of oligonucleotide probes (i.e., on groups of three consecutive probes) (see Materials and Methods). The threshold of positivity was set at the enrichment level of the known BCL6-binding site in the CCL3 promoter (16), which corre- sponded to the 95th percentile confidence interval for this method [Fig. 1A and supporting information (SI) Fig. 4]. Only genes that were captured by all three replicates and that dis- played overlapping peak enrichment were considered positive. Fig. 1. Identification of BCL6 target genes. ChIP on chip was performed in Examples of BCL6 peak oligonucleotide enrichment at target triplicate on a 24,000 promoter array. (A) Selection of BCL6 target genes. (Left) gene regulatory regions are shown in Fig. 1B. Density plot of the normalized log ratio of fluorescence intensity (BCL6 vs. BCL6 was recruited to 436 promoters, potentially regulating input) of the oligonucleotide probes. The probes showing relative enrichment 485 target genes (SI Table 2), including known target genes such by BCL6 antibodies are clustered in the right tail of the distribution curve. as FCER2 and CCL3 (16, 17). To determine the accuracy of (Right) Density plot of the maximal enrichment peak of each promoter (black BCL6 target gene discovery, single locus quantitative ChIP was line). The gray line represents a similar plot generated by using a random distribution of probes. The indicated cutoff point for selection of positive hits performed on 54 of the candidate BCL6 target genes, using the (shaded) was set at the 95th percentile of the random probe curve (see also SI known targets CCL3 and FCER2 as positive controls. Eighty-one Fig. 4). The y axis for both panels represents the local relative frequency of percent of the examined candidate BCL6 target genes were events within each level of fluorescence intensity represented on the x axis, confirmed with this stringent approach (SI Fig. 5). corresponding to probe frequency (Right) and peak frequency (Left). (B) Representative BCL6 target genes. Shown are the peak BCL6 vs. input enrich- Functional Classification of BCL6 Target Genes. To gain insights into ment at negative and positive control promoters (CD20 and FCER2, respec- the functions of identified BCL6 target genes, we evaluated their tively) and four selected gene promoters (SUB1, CR1, CBX3, and MBD1) that associated (GO) Biological Process terms (www. met the following criteria: (i) inclusion in an enriched GO category; (ii) vali- geneontology.org/GO.doc.shtml). GO terms annotate genes and dation by single-locus Q-ChIP; (iii) up-regulation after BPI treatment; and (iv) inclusion in the leading edge of the BCL6 target gene set in GSEA (for details, their products based on described biological functions. The GO see Materials and Methods). In each graph, the y axis shows fold enrichment term frequency in a given gene set (i.e., BCL6 target genes) can by BCL6 antibodies (gray field) vs. a control IgG (black field). The x axis be compared with the global GO term frequency to identify indicates the relative position of the different oligonucleotides relative to the functional categories that are represented more frequently in the transcriptional start site as annotated in the National Center for Biotechnol- examined gene set. We specifically compared the representation ogy Information (NCBI) assembly version 35 (May 2004). of GO terms in the BCL6 target gene set with that in the total analyzed gene pool (i.e., all of the genes in the GO database) (18) (Table 1). The BCL6 target list was enriched in genes regulating enrichment analysis (GSEA) to determine whether the set of transcription, DNA damage responses, chromatin modification BCL6 target genes was expressed differentially in a specific of the cell cycle, and protein ubiquitylation (Table 1). Because DLBCL subtype (19). Because the signature of HR tumors is the mechanism(s) through which BCL6 mediates the GC reac- largely defined by genes expressed by the tumor-infiltrating tion and lymphomagenesis are largely unknown, these data normal inflammatory and immune cells, we focused the GSEA provide insights regarding BCL6 function in these processes. on BCR vs. OxPhos DLBCLs. Although BCL6 likely functions as a direct transcriptional repressor, the absolute levels of BCL6 Target Gene Expression in DLBCL Subtypes. We predicted that specific target genes may depend on BCL6 cooperation with differential expression of BCL6 target genes would identify other transcription factors, binding to different or DLBCLs in which BCL6 plays a dominant oncogenic role, so we additional epigenetic modifications of chromatin. For these assessed the relative abundance of BCL6 targets in the respective reasons, we ranked the genes discriminating between BCR and DLBCL consensus clusters (11). In this analysis, we used gene set OxPhos phenotypes according to absolute (rather than positive

3208 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0611399104 Polo et al. Downloaded by guest on September 25, 2021 Table 1. GO term analysis of BCL6 target genes GO term frequency in Global GO GO term BCL6 target gene set term frequency P value FDR

Transcription 37/418 (0.0885) 0.0386 0.0000 0.0003 Protein ubiquitylation 15/418 (0.0359) 0.0096 0.0000 0.0006 Cell cycle 14/418 (0.0335) 0.0103 0.0001 0.0035 Ubiquitin cycle 12/418 (0.0287) 0.0083 0.0002 0.0043 Chromatin modification 6/418 (0.0144) 0.0023 0.0003 0.0053 Response to DNA damage stimulus 3/418 (0.0072) 0.0004 0.0004 0.0053 Regulation of transcription, DNA-dependent 41/418 (0.0981) 0.0567 0.0005 0.006 Ubiquitin-dependent protein catabolism 7/418 (0.0167) 0.0039 0.0011 0.013

Enrichment was evaluated using the GeneMerge program (18), which compares the frequency of GO categories represented in the nonredundant list of SwissProt/TrEMBL accession nos. of BCL6 target genes (n ϭ 418) versus the global frequency of GO categories in the population gene file [19,168 nonredundant SwissProt/TrEMBL accession nos. corresponding to the known genes in NCBI human genome assembly version 36 (March 2006)]. GO enrichment analysis was carried out with GO biological process terms, and obtained P values were corrected for multiple-hypothesis testing by false-discovery rate (FDR) (24, 25). P values (E scores) for GO term overrepre- sentation in the BCL6 target genes were obtained using the hypergeometric distribution (18, 29). The hypergeometric distribution quantifies the overrepresentation of a given GO term in a sample of a specific size that is drawn from a larger population (18).

or negative) signal-to-noise ratios (SNRs) and assessed the analysis were: (i) validated by Q-ChIP; (ii) included in a signif- enrichment of BCL6 target genes in the ranked dataset. In our icantly enriched GO category; and (iii) most differentially ex- series of 176 primary DLBCLs, the BCR vs. OxPhos ranked gene pressed in BCR and OxPhos tumors (i.e., included in the list was significantly enriched for BCL6 target genes (P Ͻ leading-edge gene set). We specifically selected certain candi- 0.0001), indicating that the BCL6 signature contributes to the date BCL6 targets that were less abundant in BCR than in difference between BCR and OxPhos tumors. OxPhos cells (SUB1, ZNF443, CR1, CBX3) at baseline (shaded To validate these observations in an independent dataset, in blue in Fig. 2C) and others that were more abundant in BCR GSEA was performed in an additional large series of transcrip- tumors (CD74, CCN1, MBD1, FCER2) (shaded in red in Fig. tionally profiled primary DLBCLs with available COO and 2C). BPI treatment increased the expression of each of these consensus cluster designations (11, 20). In this independent BCL6 targets in the BCR DLBCL cell lines, but it did not alter series, BCL6 targets were similarly enriched in ranked genes the expression of these genes in OxPhos cell lines (Fig. 2C). discriminating between BCR and OxPhos signatures (SI Table These data suggest that BCL6 is biologically active in BCR but 3). In contrast, BCL6 target genes were not significantly enriched not OxPhos tumors and show that BCL6 represses its target in either dataset when the DLBCLs were sorted with respect to genes in BCR DLBCLs regardless of baseline target transcript

the GC B vs. ABC classification (SI Table 3). levels. CELL BIOLOGY To determine which BCL6 targets were more (or less) abun- dant in BCR vs. OxPhos DLBCLs, the BCL6 target genes were The BCR Signature Predicts for BCL6-Dependent DLBCL Survival. clustered with respect to these tumor types (SI Table 4). The Because BPI selectively increased BCL6 target expression in top-scoring BCL6 target genes [the ‘‘leading edge’’ (see Materials BCR DLBCLs, we predicted that these tumors would be more and Methods) (19)] are visually displayed in Fig. 2A, which also dependent on BCL6-regulated gene pathways than OxPhos included normal tonsillar GC B cells for comparison. Consistent DLBCLs. We had shown (9) that BPI specifically blocked BCL6 with the known role of BCL6 as a transcriptional repressor, a activities in vitro and in vivo and inhibited the growth of certain number of BCL6 target transcripts were less abundant in BCR BCL6-positive lymphomas. For this reason, we treated the five DLBCLs than in OxPhos tumors (Fig. 2 and SI Table 4); the BCR and four OxPhos DLBCL cell lines with BPI and subse- majority of these BCL6 targets were also less abundant in normal quently evaluated tumor cell proliferation. In these experiments, GC B cells (Fig. 2A). However, additional bona fide BCL6 cell line identity was blinded until after the functional data were targets were more abundant in BCR tumors and normal GC B analyzed independently. cells than OxPhos DLBCLs (Fig. 2A and SI Table 4). This BCR cell lines had significantly lower BPI IC50 values than unexpected observation prompted us to analyze directly the OxPhos lines, which were uniformly resistant to the peptide BCL6 dependence of candidate target genes in a panel of inhibitor (BCR vs. OxPhos DLBCL IC50, 12.7 Ϯ 3.49 ␮M vs. informative DLBCL cell lines. 50.15 Ϯ 4.43 ␮M, P Ͻ 0.0001; Fig. 3 A and B). To characterize further the differential sensitivity of BCR vs. OxPhos cell lines, BCL6 Actively Represses Its Target Genes in BCR but Not OxPhos we exposed the panel to 20 ␮M BPI for 48 h. BPI inhibited Tumors. We first identified representative BCR or OxPhos DLBCL cellular proliferation of BCR DLBCL cell lines by 65–90% but cell lines (BCR: Ly1, Ly7, SU-DHL4, SU-DHL6, and Farage; and had little effect on OxPhos tumors (Fig. 3C). Therefore, the OxPhos: Ly4, Toledo, Karpas 422, and Pfeiffer) based on their designation of BCR (vs. OxPhos) DLBCL accurately predicted transcriptional profiles (Materials and Methods and SI Methods). the response to BCL6 inhibition. Importantly, the consensus Thereafter, we performed GSEA for BCL6 targets by using the cell cluster designation was more effective in predicting response to line gene list, ranked according to absolute SNR values for the BCR BPI therapy than simple baseline BCL6 expression (SI Fig. 6). vs. OxPhos distinction. As was the case in primary DLBCLs, BCL6 Our approach of combining stringent genomic localization by target genes were highly enriched in the ranked cell line gene list ChIP on chip with large-scale functional genomics and the use (P Ͻ 0.001). In addition, certain BCL6 target transcripts were less of a specific inhibitor highlight the important abundant in BCR than in OxPhos cell lines, whereas other BCL6 contribution of an oncogenic transcription factor to the tran- targets were more abundant in BCR DLBCLs (Fig. 2B). scriptional programming of a human tumor. Specifically, our We then treated four of the BCR and OxPhos cell lines with studies identify a subset of DLBCLs, the BCR tumors, in which BPI and evaluated the transcript abundance of representative BCL6 plays a critical biological role. The BCR consensus cluster BCL6 targets (Fig. 2C). The BCL6 targets chosen for this designation was more accurate in predicting BPI sensitivity than

Polo et al. PNAS ͉ February 27, 2007 ͉ vol. 104 ͉ no. 9 ͉ 3209 Downloaded by guest on September 25, 2021 Fig. 2. BCL6 target genes in primary BCR and OXP DLBCLs and DLBCL cell lines. The top-scoring BCL6 target genes from the GSEA leading edge were Fig. 3. BCR and OxPhos DLBCL cell lines exhibit differential sensitivity to BPI. clustered with respect to the DLBCL BCR and OxPhos phenotypes and repre- (A) BPI IC50 for BCR and OxPhos DLBCL cell lines. BCR and OxPhos cell lines were sented visually. Each individual column represents a tumor, and each individ- exposed to increasing doses of BPI, and cellular proliferation was assessed at ual row corresponds to a gene. For comparison, the relative expression of 48 h. The IC50 Ϯ SEM for triplicate samples of each cell line in a representative these BCL6 target genes in normal GC B cells is also shown. The color scale at experiment are shown. (B) Mean BPI IC50 (ϮSD) for BCR and OxPhos DLBCL cell the bottom indicates relative expression. (A) Primary DLBCLs (11). (B) DLBCL lines. (C) Proliferation of BPI-treated BCR and OxPhos DLBCL cell lines after BPI cell lines (BCR lines: Ly1, Ly7, SU-DHL4, SU-DHL6, and Farage; OxPhos lines: Ly4, treatment. Cell lines were exposed to 20 ␮M BPI for 48 h, and cellular Toledo, Kaspas 422, and Pfeiffer). (C) BCL6 target gene abundance in BCR and proliferation was evaluated thereafter. OxPhos cell lines after BPI treatment. BCR (SU-DHL6, SU-DHL4) and OxPhos (Toledo, Ly4) cell lines were treated with 20 ␮M BPI or control peptide for 8 h, and the transcript abundance of the indicated BCL6 targets was evaluated regulatory programming by oncogenic transcription factors, and with real-time PCR thereafter. The y axis indicates fold activation of genes direct selection of tumors for targeted therapeutic agents. after treatment with BPI vs. control peptide based on the ⌬⌬CT normalized to the expression of hypoxanthine-guanine phosphoribosyl transferse (HPRT). Materials and Methods BPI treatment increased the expression of each BCL6 target gene in the BCR Q-ChIP. ϫ 6 cell lines but did not alter the expression of these genes in OxPhos lines. BPI ChIPs were performed as described in ref. 9 with 5 10 treatment increased the abundance of BCL6 targets that were less abundant Ramos cells (13–15) and rabbit antisera directed against BCL6 in BCR than OxPhos cells at baseline (SUB1, ZNF443, CR1, and CBX3; shaded in (N3 antibody; Santa Cruz Biotechnology, Santa Cruz, CA) or blue) and others that were more abundant in BCR tumors at baseline (CD74, actin (Sigma–Aldrich, St. Louis, MO). DNA fragments enriched CCN1, and MBD1; shaded in red). A known BCL6 target gene FCER2 was used by ChIP were quantified by real-time PCR using a SYBR green as a positive control (shaded in pink). kit (Applied Biosystems, Foster City, CA) and an Opticon Engine 2 (MJ Research, Waltham, MA). The known BCL6- binding sites in the CCL3 and FCER2 genes (16, 17) were used either BCL6 protein expression alone or the absolute levels of as positive controls for BCL6 target gene enrichment, whereas BCL6 target genes. This result is not surprising because the the CD20 gene, which is not a BCL6 target, was used as a consequences of BCL6 binding to its target genes depend on negative control. ⌬CT values (antibody Ϫ input) were expressed which corepressors, additional transcription factors, and epige- relative to control antibody by using the ⌬⌬CT method (21). The netic modifications are present at those genes at a given time. same approach was used to validate an additional 44 (of 54 Rather than absolutely repressing all of its targets, BCL6 likely promoters tested) target genes identified by ChIP on chip. modulates target gene expression in specific cellular contexts. In Primers used in these experiments are listed in SI Table 5. this regard, there may be differences in the BCL6 target gene signature in BCL6-dependent tumors and normal GC B cells. ChIP on Chip and Data Processing. After validation of enrichment Such differences could explain why the consensus clusters more by real-time PCR, BCL6, or actin, ChIP products and their accurately identify BCL6-dependent tumors than the COO respective input genomic fragments were amplified by ligation- categories, which relate DLBCLs to subsets of normal B cells. mediated PCR (22). Q-ChIP was repeated after amplification to From a clinical standpoint, our data indicate that patients with verify that the enrichment ratios were retained. The genomic BCR DLBCLs may represent the best candidates for therapeutic products of three biological ChIP replicates were labeled with trials of BCL6 inhibitors. Standard diagnostic methods will not Cy5 (for ChIP products) and Cy3 (for input) and cohybridized delineate these patients. Development of methods to identify on a NimbleGen human promoter array representing 1.5 kb of tumors most likely to respond to targeted therapy is an important promoter sequence from 24.275 genes (human genome version advance because it allows for molecular stratification of patients 35, May 2004) according to manufacturer’s protocol (Nimble- to therapeutic arms most likely to be of benefit. More broadly, Gen Systems, Madison, WI). The enrichment for each promoter these data show how integration of genome-wide transcription was calculated by computing the log ratio between the probe factor binding and gene expression profiling can provide impor- intensities of the ChIP product and input chromatin, which are tant insights into tumor biology, identify the presence of gene cohybridized on the same array. Thereafter, for each of the

3210 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0611399104 Polo et al. Downloaded by guest on September 25, 2021 24,175 promoter regions, the maximum average log ratio of three primary DLBCL patients with available COO designations neighboring probes in a sliding window was calculated and (llmpp.nih.gov/DLBCL/DLBCL patient data NEW.txt; compared with random permutation of the log ratios of all ref. 20) and consensus cluster assignments (11). Affymetrix IDs probes across the entire array. The positive threshold was of BCL6 target genes were translated to Lymphochip IDs with defined by using the CCL3 signal that corresponds to the 95th current and archival UniGene cluster IDs and used as the query percentile in random permutation of the log ratios. gene set. Enrichment was assessed as described above by ranking The putative BCL6-binding regions were calculated from the genes with respect to the absolute SNR values for the triplicate experiments, represented as enrichment peaks of comprehensive cluster phenotypes BCR vs. OxPhos or COO BCL6 over control antibody signal and aligned with phenotypes GC vs. ABC. positions (NCBI human genome assembly version 35, May 2004). The top-scoring BCL6 target genes, described as the leading Thereafter, by using the NimbleGen 24K promoter array anno- edge genes, appear in the ranked list at or before the point where tation file, the peak signals of BCL6 binding were assigned to the the ES running sum reaches its maximum deviation from zero respective regulatory regions of candidate BCL6 target genes. In (26). The leading-edge genes can be interpreted as the core of addition, all peaks were inspected by using BLAT (The BLAST- a gene set that accounts for the enrichment signal (19). These like Alignment Tool) to identify genes on opposite strands that top-scoring BCL6 target genes were clustered with respect to the could be regulated from the same bidirectional promoter. Two BCR vs. OxPhos tumor phenotypes and represented on heat genes were considered to be bidirectional partners when they maps by using the dChip 2006 program. For comparison, the heat were located on the opposite strands in a ‘‘head-to-head’’ maps also included normal CD19ϩ sIgDϪ CD38ϩ GC B cells that orientation and their transcription start sites were separated by were isolated as described (27) and transcriptionally profiled at Ͻ1 kb (23). In previous studies, 90% of promoters meeting these the same time as the primary DLBCLs (11). criteria were bidirectionally active in functional assays (23). Cell Line Culture. DLBCL cell lines Ly1 and Ly7 were grown in GO Term Enrichment Analysis. GO term enrichment analysis was medium containing 90% Iscove’s medium, 10% FBS (Gemini performed with the online version of GeneMerge program (18). Bio-Products, Woodland, CA), and penicillin G/streptomycin. Enrichment was assessed by comparing the frequency of GO DLBCL cell lines Farage, Toledo, SU-DHL4, SU-DHL6, Karpas Biological Process categories represented in the nonredundant 422, and Pfeiffer were cultured in medium containing 90% list of SwissProt/TrEMBL accession numbers of BCL6 target RPMI medium, 10% FCS, 2 mM glutamine, 10 mM Hepes (ptt genes (n ϭ 418) versus the global frequency of GO categories in 7.2–7.5), and penicillin G/streptomycin. the population gene file containing 19,168 nonredundant SwissProt/TrEMBL accession numbers that corresponded to GSEA in Cell Lines. Total RNA was extracted from a panel of known genes in NCBI human genome assembly version 36 DLBCL cell lines, processed, hybridized to U133A and B (March 2006). SwissProt/TrEMBL ID codes of remaining BCL6 Affymetrix oligonucleotide microarrays, scanned, and analyzed target loci were not available. All SwissProt/TrEMBL ID codes as described (11). Cell lines were then assigned to consensus were obtained from the Affymetrix genome annotation file clusters by using an ensemble classifier incorporating multiple supporting U133 Plus 2 GeneChip (version July 2006). Obtained independent predictors (SI Methods, SI Table 6, and SI Fig. 7). P values were corrected for multiple hypothesis testing by Cell lines that were assigned to BCR or OxPhos categories with CELL BIOLOGY false-discovery rate analysis (24, 25). the highest probability were selected for GSEA and additional functional analyses. GSEA was performed as described above, GSEA. GSEA was performed by using the GSEA version 1.0 using the top 12,666 genes that met threshold and variation index program (19), the BCL6 target gene set, and two independent series criteria (28); genes were ranked according to absolute SNR of primary DLBCLs with gene expression profiles and consensus values for the phenotype BCR vs. OxPhos. The proximity of the cluster and COO designations (11, 20). Because the signature of HR BCL6 target gene set to the top of the ranked list was measured tumors is largely defined by normal tumor-infiltrating host inflam- with an ES, and the significance of the ES was determined by matory and immune cells, the analysis was focused on BCR and using 1,000 gene tag permutations, as described above. OxPhos DLBCLs. GSEA was performed as described previously, with minor modifications. The top 15,000 genes selected with a Treatment with BPI. Peptides (BPI and control) were obtained median absolute deviation-based variation filter were first ranked from Bio-Synthesis, Inc. (Lewisville, TX) and stored at Ϫ20°C with respect to the phenotype, BCR vs. OxPhos, by using an until reconstituted with sterile pure water immediately before absolute value (rather than positive or negative) SNR. With this use. BPI purity was determined by HPLC-MS to be 98% or approach, the final position in the ranked gene list depended only higher. We exposed 25 ϫ 104 DLBCL cells to BPI or control on the strength of the gene in discriminating between phenotypes peptide (0, 1, 2.5, 5, 10, 20, 40, and 80 ␮M) for 48 h. Cellular rather than specific up- or down-regulation in a given phenotype. proliferation was assessed by MTS assay (Cell Titer 96 Aqueous Represented members of the BCL6 target gene set were then One; Promega, Madison WI) according to the manufacturer’s located within the ranked gene list, and the proximity of the BCL6 instructions, by using eight replicates per treatment condition. target gene set to the most differentially expressed BCR vs. OxPhos The proliferation of BPI-treated cells (T) was normalized to their genes (i.e., those with the highest absolute SNR value) was mea- respective peptide concentration controls (C) as follows: (T/ sured with a weighted Kolmogorov–Smirnov statistic [ES, enrich- C)corr (%) ϭ (T/C)/UT ϫ 100. The growth inhibition (IC50) ment score (ref. 19)], with a higher score corresponding to a higher values were estimated by a linear least-squares regression of the proximity. The observed ES score was then compared with the (T/C)corr values versus the concentration of BPI (or control) distribution of 1,000 permuted ES scores (gene tag permutations) peptide; T/Ccorr values of 50% were extrapolated. The difference to assess statistical significance. Similar results were observed with in BPI IC50 values of BCR and OxPhos cell lines was assessed the permutation of the class template (data not shown). The query with a two-sided Student’s t test. Previous studies confirmed that gene set included the 309 (of a total 485) BCL6 target genes present BPI efficiently enters lymphoma cell lines that were subsequently in the 15,000 ranked genes; these 309 BCL6 targets were repre- identified as OxPhos DLBCLs (9). sented by 477 Affymetrix probe sets. BCL6 target gene enrichment was also assessed in the gene list ranked for the positively defined BCL6 Target Gene Expression. After treatment with 20 ␮M BPI or COO phenotypes GC B vs. ABC (10, 11) sorted by absolute SNRs. control peptide for 8 h, RNA was extracted from 104 DLBCL GSEA was also performed in an independent dataset of 218 cells by using the RNeasy kit (Qiagen, Valencia, CA). cDNA was

Polo et al. PNAS ͉ February 27, 2007 ͉ vol. 104 ͉ no. 9 ͉ 3211 Downloaded by guest on September 25, 2021 Ϫ⌬⌬CT synthesized by using a SuperScript III first-strand cDNA syn- by the expression: 2 with ⌬⌬CT ϩ s and ⌬⌬CT Ϫ s where thesis kit (Invitrogen, Carlsbad, CA). The mRNA levels of s is the standard deviation of the ⌬⌬CT value for triplicates. SUB1, CBX3, CR1, ZNF433, CCN1, MBD1, CD74, FCER2, and Results were represented as fold expression Ϯ SD. HPRT (housekeeping control) were detected by using a SYBR green kit and an Opticon Engine 2 thermal cycler (MJ Research, We thank Riccardo Dalla-Favera and Andrea Califano for helpful Waltham, MA). Primer sequences for real-time PCR are listed discussions. J.M.P. was supported by the National Cancer Center. A.M. was supported by the G&P Foundation, Leukemia and Lymphoma in SI Table 7. The CT values of the genes of interest were ⌬ ⌬ Society Grant 7032-04, the Samuel Waxman Cancer Research Founda- normalized to HPRT ( CT). CT values of the BPI-treated cells tion, Chemotherapy Foundation Grant 95269247, and National Cancer were expressed relative to control peptide-treated cells by using Institute (NCI)/National Institutes of Health (NIH) Grant NCI R01 the ⌬⌬CT method. The fold change in expression of each gene CA104348. M.S. was supported by NCI/NIH Grant P01CA092625, the in BPI-treated vs. control peptide-treated cells was determined Doris Duke Charitable Foundation, and the Kittredge Foundation.

1. Cattoretti G, Chang CC, Cechova K, Zhang J, Ye BH, Falini B, Louie DC, Offit 15. Niu H, Ye BH, Dalla-Favera R (1998) Genes Dev 12:1953–1961. K, Chaganti RS, Dalla-Favera R (1995) Blood 86:45–53. 16. Shaffer AL, Yu X, He Y, Boldrick JC, Chan EP, Staudt LM (2000) Immunity 2. Ye B, Cattoretti G, Shen Q, Zhang J, Hawe N, de Waard R, Leung C, 13:199–212. Nouri-Shirazi M, Orazi A, Chaganti R, et al. (1997) Nat Genet 16:161–170. 17. Dent A, Shaffer A, Yu X, Allman D, Staudt L (1997) Science 276:589–592. 3. Fujita N, Jaye DL, Kajita M, Geigerman C, Moreno CS, Wade PA (2003) Cell 18. Castillo-Davis CI, Harti DL (2003) Bioinformatics 19:891–892. 113:207–219. 19. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, 4. Tunyaplin C, Shaffer AL, Angelin-Duclos CD, Yu X, Staudt LM, Calame KL Paulovich A, Pomeroy SL, Golub TR, Lander ES, Mesirov JP (2005) Proc Natl (2004) J Immunol 173:1158–1165. Acad Sci USA 102:15545–15550. 5. Fearon DT, Manders PM, Wagner SD (2002) Adv Exp Med Biol 512: 20. Rosenwald A, Wright G, Chan WC, Connors JM, Campo E, Fisher RI, 21–28. Gascoyne RD, Muller-Hermelink HK, Smeland EB, Staudt LM (2002) N Engl 6. Ye B, Lista F, Lo Coco F, Knowles DM, Offit K, Chaganti R, Dalla-Favera R J Med 346:1937–1947. Science (1993) 262:747–750. 21. Chakrabarti SK, James JC, Mirmira RG (2002) J Biol Chem 277:13286– 7. Cattoretti G, Pasqualucci L, Ballon G, Tam W, Nandula SV, Shen Q, Mo T, 13293. Murty VV, Dalla-Favera R (2005) Cancer Cell 7:445–455. 22. Oberley MJ, Tsao J, Yau P, Farnham PJ (2004) Methods Enzymol 376:315– 8. Baron BW, Anastasi J, Montag A, Huo D, Baron RM, Karrison T, Thirman 334. MJ, Subudhi SK, Chin RK, Felsher DW, et al. (2004) Proc Natl Acad Sci USA 23. Trinklein ND, Aldred SF, Hartman SJ, Schroeder DL, Otillar RP, Myers RM 101:14198–14203. 9. Polo JM, Oso TD, Ranuncolo SM, Cerchietti L, Beck D, Da Silva GF, Prive (2004) Genome Res 14:62–66. GG, Licht JD, Melnick A (2004) Nat Med 10:1329–1335. 24. Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I (2001) Behav Brain Res 10. Wright G, Tan B, Rosenwald A, Hurt E, Wiestner A, Staudt L (2003) Proc Natl 125:279–284. Acad Sci USA 100:9991–9996. 25. Reiner A, Yekutieli D, Benjamini Y (2003) Bioinformatics 19:368–375. 11. Monti S, Savage KJ, Kutok JL, Feuerhake F, Kurtin P, Mihm M, Wu B, 26. Li C, Wong WH (2001) Proc Natl Acad Sci USA 98:31–36. Pasqualucci L, Neuberg D, Aguiar RC, et al. (2005) Blood 105:1851–1861. 27. Aguiar R, Yakushijin Y, Kharbanda S, Tiwari S, Freeman G, Shipp M (1999) 12. Takahashi H, Feuerhake F, Kutok JL, Monti S, Dal Cin P, Neuberg D, Aster Blood 94:2403–2413. JC, Shipp MA (2006) Clin Cancer Res 12:3266–3271. 28. Juszczynski P, Kutok JL, Li C, Mitra J, Aguiar RCT, Shipp MA (2006) Mol Cell 13. Phan RT, Dalla-Favera R (2004) Nature 432:635–639. Biol 26:5348–5359. 14. Fujita N, Jaye DL, Geigerman C, Akyildiz A, Mooney MR, Boss JM, Wade PA 29. Ashburner M, Ball CA, Bloke JA, Botstein D, Butler H, Chessy JM, Davis AP, (2004) Cell 119:75–86. Dolinski K, Dwight SS, Eppig JT, et al. (2000) Nat Genet 25:25–29.

3212 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0611399104 Polo et al. Downloaded by guest on September 25, 2021