Functional Link Between Heat Shock Protein HSP90alpha and Sirtuin 1 (SIRT1) in the Pathogenesis of Diffuse Large B Cell Lymphoma

Diffuse large B-cell lymphomas (DLBCL) are heterogenous tumors. A large subset of DLBCL is reliant on B-cell receptor activity ('BCR'-DLBCLs). Additional subset of tumors that does not exhibit addiction to BCR signaling features transcriptional program is associated with increased oxidative phosphorylation ('OxPhos'-DLBCLs, Blood 2005; 105: 1851-61). These OxPhos tumros are resistant to BCR inhibition and remain functionally poorly defined. We and others found that OxPhos-DLBCLs, compared to BCR subtype, exhibit overexpression of heat shock protein HSP90alpha (Br J Haematol 2009; 144: 358-66). Expression of multiple heat shock proteins is regulated by the activity of a stress-responsive, acetylation-dependent transcription factor HSF1. Acetylation blocks, whereas deacetylation by sirtuins increases HSF1 activity. However, molecular mechanisms responsible for increased HSP90alpha expression in OxPhos-DLBCLs remain unknown.

We investigated the abundance of NAD-dependent deacetylase SIRT1 in a panel of DLBCL cell lines and found that SIRT1 protein level was higher in OxPhos- than BCR-DLBCLs and correlated positively with HSP90alpha gene expression (p=0.02, r=0.7857). To assess the role of SIRT1 in transcriptional regulation of HSP90alpha, we used a small molecule SIRT1 inhibitor tenovin-6 or interfering RNA, which dramatically reduced HSP90alpha induction after heat shock. We next characterized the mechanisms of SIRT1 overexpression in OxPhos cells. SIRT1 transcript abundance was similar in OxPhos and BCR cells, suggesting that changes in SIRT1 expression result from different protein half-life/stability. To test this assumption, OxPhos and BCR cells were incubated with a protein synthesis inhibitor, cycloheximide, and assessed for SIRT1 protein decays over time. Our experiments revealed increased stability of SIRT1 protein in OxPhos-DLBCLs, compared to BCR-DLBCLs. Addition of an HSP90 inhibitor 17-AAG accelerated SIRT1 protein degradation, suggesting that HSP90 might chaperone and thus protect SIRT1 against degradation. We further found that SIRT1 coimmunoprecipitated with HSP90alpha, confirming interaction between these proteins and highlighting a self-reinforcing feedback loop linking HSP90alpha and SIRT1 in OxPhos-DLBCLs.

We next assessed the therapeutic potential of SIRT1 inhibition in DLBCLs. SIRT1 knockdown with small interfering RNA or chemical SIRT1 inhibition (tenovin-6) decreased proliferation of OxPhos-DLBCL cells, but also of BCR-DLBCLs, indicating that SIRT1 inhibition exhibits broader, subtype-independent activity in DLBCL cell lines. Surprisingly, BCR-type DLBCLs were significantly more sensitive to SIRT1 inhibitor, tenovin-6. We thus sought to identify a mechanism responsible for SIRT1 inhibition in BCR-DLBCLs. Since SIRT1 deacetylase modulates the transcriptional activity of BCL6 oncogene, critical for survival of BCR-type DLBCLs (PNAS 2007; 104: 3207-12), we assessed the expression of BCL6-regulated genes in these cells. SIRT1 inhibition upregulated the abundance of multiple BCL6-dependent transcripts (including FCER2, CCL3, BLIMP1 and CD74), suggesting that toxicity of SIRT1 inhibition in BCL6-dependent DLBCLs is, at least partially, mediated by decreased BCL6 repressor activity.

Although OxPhos-DLBCLs were less sensitive to tenovin-6 than BCR-type DLBCLs, we hypothesized that, given the feedback loop between SIRT1 and HSP90, simultaneous inhibition of HSP90 would increase the cellular sensitivity to tenovin-6. Consistent with this, we found that combination of these two compounds exhibited synergistic activity in OxPhos-DLBCL. Taken together, our data demonstrate that SIRT1 and HSP90alpha are mutually linked and involved in the pathogenesis of OxPhos-DLBCL. Targeting SIRT1-HSP90alpha loop with combinatorial use of SIRT1 and HSP90 inhibitors might be a promising treatment strategy in OxPhos-DLBCLs.