Thiazolidinedione enhances the efficacy of anti-PD-1 monoclonal antibody in murine melanoma

Abstract

Several clinical studies observed a surprising beneficial effect of obesity on enhancing immunotherapy responsiveness in patients with melanoma, highlighting an as-yet insufficiently understood relationship between metabolism and immunogenicity. Here, we demonstrate that the thiazolidinedione (TZD) rosiglitazone, a drug commonly used to treat diabetes by sequestering fatty acids in metabolically inert subcutaneous adipose tissue, improved sensitivity to anti-programmed cell death protein 1 (PD-1) treatment in YUMMER1.7 tumor-bearing mice, an initially immunotherapy-sensitive murine melanoma model. We observed a transition from high to intermediate PD-1 expression in tumor-infiltrating CD8+ T cells. Moreover, TZD inhibited PD-1 expression in mouse and human T cells treated in vitro. In addition to its direct impact on immune cells, TZD also decreased circulating insulin concentrations, while insulin induced T cell exhaustion in culture. In TZD-treated mice, we observed higher fatty acid concentrations in the tumor microenvironment, with fatty acids protecting against exhaustion in culture. Together, these data are consistent with an indirect mechanism of TZD inhibiting T cell exhaustion. Finally, we analyzed imaging data from patients with melanoma before and after anti-PD-1 treatment, confirming the beneficial effect of increased subcutaneous fat on anti-PD-1 responsiveness in patients. We also found that the expression of peroxisome proliferator-activated receptor gamma (PPARγ), the canonical activator of lipid uptake and adipogenesis activated by TZD, correlated with overall survival time. Taken together, these data identify a new adjuvant to enhance immunotherapy efficacy in YUMMER1.7 melanoma mice, and discover a new metabolism-based prognostic marker in human melanoma.NEW & NOTEWORTHY Zhang et al. demonstrate that the diabetes drug rosiglitazone improves the efficacy of immunotherapy in mouse melanoma. This effect is both direct and indirect: TZD directly reduces PD-1 expression in CD8+ T cells (i.e., reduces exhaustion), and indirectly reduces exhaustion by lowering insulin levels and increasing local fat. Finally, they demonstrate that hallmarks of TZD action (such as PPARγ expression and subcutaneous fat content) correlate with improved immunotherapy efficacy in humans with melanoma.

Keywords: immunotherapy; melanoma; subcutaneous fat; thiazolidinedione.

Graphical abstract

Figure 1.

TZD increases the efficacy of anti-PD-1 immunotherapy in murine melanoma. A: tumor size of YUMMER1.7-bearing mice treated with/without anti-PD1 and/or TZD (n = 5 per group). B: cell number of YUMMER1.7 cultured with/without TZD (n = 3 replicates per group). C: tumor size of YUMM1.7-bearing mice treated with/without anti-PD1 and/or TZD (n = 5 per group). D: cell number of YUMM1.7 cultured with/without TZD (n = 3). SD is shown; *P < 0.05, **P < 0.01, ****P < 0.0001. anti-PD-1, anti-programmed cell death protein 1; TZD, thiazolidinedione.

Figure 2.

TZD directly inhibits PD-1 expression in T cells. A: percentage of infiltrated CD8+ T cells in YUMMER1.7 tumor (n = 4). B: representative PD-1 (y-axis) and Tim3 (x-axis) expression level in infiltrated CD8+ T cells in YUMMER1.7 tumor measured by flow cytometry. C: mean fluorescence intensity of PD-1 expression in infiltrated CD8+ T cells in YUMMER1.7 tumor measured by flow cytometry (n = 4). D: representative PD-1 (y-axis) and Tim3 (x-axis) expression level in cultured primary mouse T cells with/without TZD treatment, measured by flow cytometry. E: percentage of PD-1+ T cells in cultured primary mouse CD8+ T cells with/without TZD treatment (n = 5). F: percentage of PD-1+ T cells in cultured human CD8+ T cells with/without TZD treatment (n = 5). SD is shown; *P < 0.05, ***P < 0.001, ****P < 0.0001. anti-PD-1, anti-programmed cell death protein 1; TZD, thiazolidinedione.

Figure 3.

TZD indirectly inhibits PD-1 expression in T cells. A: circulating insulin concentration in YUMMER1.7-bearing mice treated with/without TZD (n = 6). B: percentage of PD-1-expressing cells in primary mouse CD8+ T cells cultured with/without insulin (n = 5). C: NEFA concentration in YUMMER1.7 tumor obtained from YUMMER1.7-bearing mice treated with/without TZD (n = 10). D: percentage of PD-1-expressing cells in primary mouse CD8+ T cells cultured with/without palmitic acid (n = 5). E: YUMMER1.7 cell proliferation with/without fatty acid (palmitic acid + oleic acid) treatment (n = 5). F: YUMMER1.7 tumor volume after 21 days of injection treated with/without high-fat diet. SD is shown; *P < 0.05, **P < 0.01. anti-PD-1, anti-programmed cell death protein 1; TZD, thiazolidinedione.

Figure 4.

Subcutaneous fat and PPARγ expression are positive prognostic factors in patients with melanoma. Comparative analysis of patients’ responding ratio to anti-PD-1 immunotherapy by obesity status determined by subcutaneous fat area (A) and subcutaneous fat thickness (B) (n = 20). Linear correlation between efficacy of anti-PD-1 immunotherapy and subcutaneous fat area (C) and thickness (D) in patients with melanoma (n = 20). Linear correlation between BMI and PPARγ expression (E), BMI and overall survival time (F), and PPARγ expression and lymphocyte infiltration (G) in melanoma patients with obesity (BMI > 28). All data were obtained from the Cancer Genome Atlas (TCGA) database. anti-PD-1, anti-programmed cell death protein 1; BMI, body mass index; PPARγ, peroxisome proliferator-activated receptor gamma.