Connectivity Mapping of BCL6 Targeted Therapy Guides Rational Design of Potent and Specific Non-Chemotherapy Combinatorial Regimens in DLBCL.

DLBCL is the most common form of non-Hodgkin’s lymphoma. Combinations of untargeted chemotherapeutic agents cure between 40–60% of DLBCL patients. We are interested in the rational design of targeted combinatorial therapy for DLBCL using non-chemotherapy agents. Towards this goal we developed an inhibitor of the BCL6 transcriptional repressor, the most commonly involved oncogene in DLBCL. This BCL6 peptide inhibitor (BPI) causes de-repression of BCL6 target genes and kills DLBCL cells. Since single agent targeted therapy is unlikely to cure tumors, we hypothesized that identification of survival pathways triggered by BPI would facilitate rational design of combinatorial biological therapy for DLBCL. In order to identify such pathways we performed gene expression (GE) microarray studies in ten DLCBL cell lines treated with BPI vs. control. Six cell lines were BCL6 positive and four were BCL6 negative. Only the BCL6 positive cells yielded differences in gene expression. Among BPI induced genes was the p300 histone acetyl-transferase. The overlapping genes among the six cell lines were used to generate a BPI response signature. We used this signature to query the Broad Institute Connectivity Map, which contains the GE signature of 164 distinct small-molecule perturbagens. The top scoring classes of drugs were the histone deacetylase inhibitors (HDIs) and HSP90 inhibitors. Considering that BPI is chemically un-related to HDIs or HSP90 inhibitors and that BPI induces p300, we hypothesized that a major biological effect of BPI is to cause the acetylation of HSP90 (which inhibits Hsp90 pro-survival activity) and p53, (which enhances its pro-apoptotic activity). We verified that p300 is a direct BCL6 target gene by ChIP assays, that BPI induces p300 mRNA and protein by QPCR and western blot, and that p300 is silenced in most primary DLBCLs at both the mRNA and protein levels. Accordingly, BPI induced acetylation of Hsp90 and inhibited its function, as demonstrated by the decrease in the abundance of Hsp90 client proteins (AKT/PKB and c-raf). BPI also induced acetylation and functional activity of p53 in a p300-dependent manner (and also induces p53 expression). The importance of p300 was confirmed since a p300-dominant negative construct and the specific p300(HAT) inhibitor Lys-CoA-TAT could block BPI antilymphoma activity. Remarkably, we observed a dose-sequence dependent synergistic effect of BPI followed by Hsp90 inhibitors in killing DLBCL cells. Hsp90 is a relevant target in DLBCL since HSP90?/? protein was expressed in ∼90% of DLBCL patients (n=70). HDIs also increase acetylation of Hsp90 and p53. The HDI drugs SAHA, valproic acid and TSA all profoundly synergized with BPI to specifically eradicate BCL6 positive DLBCL cell lines. In conclusion, we discovered an unexpected mechanistic link between BCL6 and suppression of protein acetylation in lymphomagenesis. This information was harnessed for the rational design of synergistic targeted therapy with BCL6 inhibitors followed by Hsp90 or HDAC inhibitors to target cellular pathways induced by BPI. We anticipate that these drug combinations will result in more potent and less toxic therapeutic treatment of DLBCL, possibly with less or no added chemotherapy.