Targeting HSP90 sensitizes pancreas carcinoma to PD-1 blockade.

Interferon gamma (IFNG/IFNγ)-induced adaptive immune resistance remains a challenge for tumor therapy. We observed that the chaperone heat shock protein 90 (HSP90) stabilizes the transcription factor signal transducer and activator of transcription 1 (STAT1), resulting in IFNγ-induced expression of immunosuppressive indoleamine 2,3-dioxygenase 1 (IDO1) and programmed death-ligand 1 (PD-L1/CD274). Pharmacological inhibition of HSP90 enhances the efficacy of programmed cell death 1 (PDCD1/PD-1) targeting immunotherapy in suitable mouse models without any toxicity.

Interferon gamma (IFNG, best known as IFNγ) is a pleiotropic cytokine that mediates antiviral effects and acts as a central coordinator of antitumor immune responses.[1] In addition to activating the cytotoxic function of CD8+ T cells, IFNγ is a strong inducer of the expression of multiple immune checkpoint molecules, presumably by activating the signal transducer and activator of transcription 1 (STAT1) pathway, leading to adaptive immune resistance.[2] Our recent study identifies the therapeutic vulnerability of pancreatic ductal adenocarcinoma (PDAC) by showing that heat shock protein 90 (HSP90) plays a new role in mediating IFNγ-induced adaptive resistance to immunotherapies targeting programmed cell death 1 (PDCD1, best known as PD-1) (Figure 1).[3]

Figure 1.

The HSP90 chaperone machinery modulates IDO1 and PD-L1 expression. IFNγ-induced the expression of immune checkpoint molecules (IDO1 and PD-L1) requires increased protein stability of the transcription factor STAT1 mediated by the HSP90-SUGT1 chaperone complex. Abbreviations: HSP90, heat shock protein 90; IDO1, indoleamine 2,3-dioxygenase 1; IFNγ, interferon gamma; PD-L1, programmed death-ligand 1; STAT1, signal transducer and activator of transcription 1; SUGT1, SGT1 homolog, MIS12 kinetochore complex assembly cochaperone.

PDAC is one of the deadliest gastrointestinal cancers driven by KRAS mutations, with a 5-year overall survival rate of 5% to 10% that has not been ameliorated over the past 30 years. PDAC is resistant to therapy with immune checkpoint inhibitors (ICI), but the mechanisms underlying this resistance are largely unknown.[4] We implanted KPC cells (which are PDAC cells derived from Pdx1-Cre;Kras mice) into C57BL/6 J mice and examined the expression of 16 common immune checkpoint molecules in isolated tumor after treatment with anti-PD-1 antibody (αPD-1). Compared with the minority of responder mice (~25%), animals that did not respond to PD-1 blockade (~75%) selectively upregulated the expression of indoleamine 2,3-dioxygenase 1 (IDO1), rather than other immune checkpoint molecules, in cancer cells in an IFNγ-dependent manner.[3] Given that the functions of different immune checkpoint molecules are complementary, dynamic monitoring of their expression in the tumor microenvironment may be important for identifying suitable therapeutic targets.[5]

Next, we used the KRAS and tumor protein p53 (TP53, best known as p53)-mutated human PDAC cell line CFPAC1 to screen for compounds that inhibit IFNγ-induced IDO1 as well as expression of CD274 molecule (best known as programmed death-ligand 1 [PD-L1]). Of note, we found that 24 compounds (used at 10 µM) that blocked IFNγ-induced IDO1 and PD-L1 expression in CFPAC1 cells. The vast majority (71%) of these agents were HSP90 inhibitors.[3] In addition, nanomolar amounts of six HSP90 inhibitors (luminespib, ganetespib, SNX-2112, PF-04929113, HSP990 and XL888) suppressed IFNγ-induced IDO1 and PD-L1 expression in 16 human tumor cell lines (corresponding to 11 different types of cancer) and primary PDAC cells from patients and KPC mice.[3] Mechanistically, we demonstrated that the binding of HSP90 to its partner SGT1 homolog, MIS12 kinetochore complex assembly cochaperone (SUGT1) resulted in increased protein stability of STAT1, a key transcription factor for the expression of immune checkpoint molecules. Expression of dominant-negative HSP90 (D88N) led to inhibition of STAT1-mediated IFNγ signaling, suggesting that the aforementioned HSP90 inhibitors act on target.[3] Altogether, these results point to a broad role for HSP90 in mediating the expression of inducible immune checkpoint molecules.

IDO1 is a rate-limiting metabolic enzyme that converts the essential amino acid tryptophan to a downstream immunosuppressor, kynurenine, thereby inhibiting the proliferation and function of cytotoxic CD8+ T cells. However, the mechanism of action of IDO1-mediated kynurenine production and secretion remains poorly understood. We found that tetraspanin 5 (TSPAN5), a transmembrane protein of the tetraspanin family, plays a critical role in mediating IFNγ-induced kynurenine secretion, but not IFNγ-induced IDO1 expression.[3] We also verified that the enzymatic activity of IDO1 requires iron (but not other metal ions) to trigger kynurenine production. Consequently, iron-enhanced kynurenine release (but not kynurenine synthesis) was inhibited in TSPAN5-deficient CFPAC1 cells.[3] These findings reveal a role for iron in promoting IDO1-dependent kynurenine production and subsequent TSPAN5-mediated kynurenine release.

Finally, we evaluated the efficacy and safety of combinations of αPD-1 with the HSP90 inhibitor ganetespib, the IDO1 inhibitor BMS-986205, or the iron chelator desferoxamine in the treatment of transplanted or a transgene-induced PDAC. Compared with the αPD-1 alone group, the combination of αPD-1 with ganetespib, BMS-986205, or desferoxamine was more efficient in reducing tumor growth and enhancing the infiltration of PDAC by CD8+ T cells and dendritic cells (but not CD4+ T cells, macrophages, and natural killer cells). Depletion of CD8+ T cells abolished deferoxamine and αPD-1-mediated tumor suppression, demonstrating that this combination therapy relies on the contribution of tumor-specific cytotoxic T lymphocytes. This new combination therapy had an acceptable safety profile and did not affect liver and kidney function in mice.

In summary, our study uncovers a new strategy through which PDAC cells hide from the immune system. Moreover, IDO1 emerges as a potential biomarker for predicting treatment responses to anti-PD-1. We provide proof-of-concept for future clinical applications of HSP90 inhibitors, IDO1 inhibitors, or iron chelators to enhance the anticancer activity of PD-1 blockade. In addition to STAT1, the expression of immune checkpoint molecules is controlled by other transcriptional factors, such as TP53,[6] hypoxia inducible factor 1 subunit alpha,[7] and MYC.[8] It is important to further define the crosstalk of gene transcription and protein degradation pathways in coordinating the expression of immune checkpoint molecules by PDAC cells.[9] Potentially, distinguishing different clients of the HSP90 chaperone machinery in tumor immunity remains a challenge. Regardless, the hypothesis that inducers of immunogenic stress and death,[10] including HDP90 inhibitors, may sensitize PDAC cells to immunotherapy should be explored in future clinical assays.