A BCL6 Target Gene Signature Predicts the Biological Behavior and Classification of Diffuse Large B-Cell Lymphoma.

Diffuse large B-cell lymphoma (DLBCL), the most common lymphoid malignancy, is a heterogeneous disease. These tumors are thought to arise from normal antigen-exposed B-cells that have migrated to or through the germinal center (GC). Structural abnormalities of the BCL6 locus (chromosomal translocation and aberrant somatic hypermutation) are the most common genetic abnormalities in DLBCL, occurring in over a third of these tumors. We recently developed a potent and specific BCL6 peptide inhibitor (BPI) that disrupts the interaction between BCL6 and the SMRT co-repressor complex. BPI was cytotoxic against some, but not all, BCL6 positive primary DLBCLs and DLBCL cell lines, indicating that a subset of DLBCLs was particularly dependent on BCL6 for their survival. We predicted that such cases might be identified through a specific BCL6-dependent gene signature and utilized ChIP on chip and a 24,000 promoter genomic microarray to identify BCL6 target genes. In these studies, BCL6 bound to 431 loci. Eighty percent of these candidate target genes contained a canonical BCL6 binding site and 85% of analyzed candidates were confirmed using quantitative single-locus CHIP. GO term enrichment revealed that BCL6 targets were significantly more likely to be genes associated with transcription, ubiquitylation, response to DNA damage, cell cycle and chromatin assembly/disassembly (FDR <.05). We predicted that coordinate regulation of the BCL6 targets would serve as a signature of BCL6 activity. For this reason, we asked whether the BCL6 target gene set was differentially expressed in the recently described DLBCL comprehensive clusters, “B-cell Receptor/Proliferation (BCR),” “Oxidative Phosphorylation (OxPhos),” and Host Response (HR),” using 2 large series of primary DLBCLs with available transcriptional profiles. Since HR tumors are largely defined by infiltrating host inflammatory cells, we focused on BCR and OxPhos DLBCLs. Of interest, BCR tumors more frequently exhibit BCL6 chromosomal translocations and increased BCL6 expression. Consistent with these observations, gene set enrichment analysis (GSEA) revealed highly significant differential expression of BCL6 target genes in BCR vs. OxPhos tumors (p <.0001). In contrast, GSEA of the BCL6 targets in the same DLBCLs sorted into developmental cell-of-origin groups was not significant (“GC” vs. “ABC”/”Other”, p =.25 and “GC” vs. “ABC” only, p =.082). To assess the functional significance of the BCL6 signature in the DLBCL consensus clusters, we generated a predictive algorithm based on the transcriptional profiles of a series of DLBCL cell lines to assign the lines to BCR, OxPhos or HR comprehensive clusters. Five BCR and 3 OxPhos DLBCL cell lines were selected for additional blinded functional analyses. BPI treatment upregulated BCL6 target gene expression in BCR, but not OxPhos, DLBCLs. Furthermore, BPI was significantly more effective in BCR lines than OxPhos DLBCLs (BPI IC50s of BCR vs. Ox Phos lines, 12.7±1.7 μM vs 48.9±2.6 μM, respectively, p<. 0001). Taken together, the data indicate that BCR DLBCLs are significantly more dependent on BCL6 transcriptional repression and more sensitive to targeted BCL6 inhibition. More generally, these studies suggest that DLBCL comprehensive cluster designation will help guide the targeted therapy of biologically discrete DLBCL subsets.