BCL6 Mediates a Stress Tolerance Phenotype through Its BTB Domain

Diffuse large B-cell lymphomas (DLBCLs) arise from germinal center (GC) B-cells. Normal GC B-cells clonally expand and undergo somatic hypermutation of their immunoglobulin loci to produce high-affinity antibodies. Induction of BCL6, a transcription factor that represses genes involved in DNA damage sensing and checkpoint activation, is essential for GC B-cells to tolerate replicative and genotoxic stress without inducing cell cycle arrest. We previously showed that BCL6 forms a complex with tumor enriched HSP90 (TE-HSP90) to repress target genes in DLBCL cells.

Based on these facts, we hypothesized that BCL6 is a component of a conserved stress response program, required for GC formation and maintenance of established lymphoma cells. Along these lines, we report that treatment of mice with TE-HSP90-selective inhibitor PU-H71 completely abrogates GC formation after immunization by T-cell dependent antigen. Stress responses are coordinated by the transcription factor heat shock factor 1 (HSF1), which is activated by phosphorylation. Immunofluorescence of human tonsillar sections revealed HSF1-pSer326-positive cells within BCL6⁺ GC B-cells. We found that HSF1^-/- mice manifest a 40% decrease in GC B-cells after immunization and significantly (p=0.0073) decreased titers of high-affinity immunoglobulin compared to WT mice, indicating defective affinity maturation. Mixed chimera experiments revealed that the GC defect is exclusively due to malfunction of GC B-cells and not other cell types. Because of this GC B-cell defect, we reasoned that HSF1 might induce BCL6 expression. We identified three conserved heat shock elements (HSEs) in the BCL6 promoter. Quantitative ChIP analysis demonstrated HSF1 binding to these HSEs in human GC B-cells. Heat shock induced the BCL6 promoter in reporter assays and resulted in an increase in BCL6 nascent transcripts and protein. However this induction did not occur in HSF1^-/-B-cells or after HSF1 knockdown. These data suggest that BCL6 is a stress response gene downstream of HSF1.

To determine whether BCL6 is involved in mediating a stress tolerant phenotype, we performed serial stress response assays (using heat shock) in B220⁺ splenocytes from BCL6^+/+ or BCL6^-/- mice. Whereas BCL6^+/+ cells were able acquire stress tolerance if preconditioned with an initial heat shock, BCL6^-/- splenocytes failed to adapt to stress and died. To understand the mechanistic basis of this finding, we generated knockin mice with point mutations that disrupt the repressor activity of the BCL6 BTB (BCL6^BTB) or the BCL6 RD2 (BCL6^RD2) domain. Interestingly while splenocytes from BCL6^RD2 mutant mice displayed normal stress tolerance responses, BCL6^BTB mutant B-cells were completely deficient similar to BCL6^-/-B-cells. Likewise the BCL6 BTB domain inhibitor RI-BPI also abrogated the BCL6 stress tolerance function.

DLBCLs are dependent on many of the same molecular mechanisms as normal GC B-cells (e.g. BCL6). Indeed the lentiviral transduction of HSF1 shRNAs in DLBCL cell lines reduced BCL6 protein levels by more than 50% resulting in a 70%-90% loss in viability. HSF1 is known to help tumor cells survive exposure to chemotherapy drugs, and the BCL6 BTB domain is required for stress tolerance. Thus we hypothesized that BCL6 BTB domain targeted therapy (RI-BPI) would synergistically kill DLBCL when combined with chemotherapy. We treated 7 DLBCL cell lines with RI-BPI in combination with doxorubicin, vincristine, dexamethasone, mechlorethamine (in place of cyclophosphamide), and their combination CHOP. Almost all combinations resulted in an additive or synergistic effect on DLBCL growth inhibition. Moreover the combination of RI-BPI and doxorubicin in a DBLCL xenograft model was more potent and significantly (p<0.001) better at reducing tumor growth than either drug alone. Using an ex vivo coculture system for primary human DLBCL specimens, the combination of RI-BPI and CHOP had at least an additive effect on growth inhibition in 80% of primary human DLBCL cells with the majority demonstrating a synergistic anti-lymphoma effect.

Collectively we demonstrate that BCL6 is an HSF1-dependent stress tolerance factor and mediates this effect via its BTB domain. This phenomenon occurs during normal GC formation and in lymphoma cells. Thus targeting the BTB domain of BCL6 pharmacologically in combination with other chemotherapy is a viable strategy to eradicate lymphomas.