The glutaminase inhibitor CB-839 (Telaglenastat) enhances the anti-melanoma activity of T cell mediated immunotherapies.

Cell lines, mice and inhibitors.

B16-F10 cells were from ATCC, Yummer 1.7 cells were kindly provided by Dr. Marcus Bosenberg (Yale University School of Medicine, New Haven, CT) (20). A375HG cells that constitutively express murine H-2D^b and GFP-tagged gp100 melanoma tumor antigen have been described earlier (21). Patient-derived melanoma cell lines and their autologous tumor-infiltrating lymphocytes (TILs) were isolated under an institutional review board–approved laboratory protocol (LAB06–0755), as previously described (22,23). All human subjects gave verbal and written consent to participate. Cell line authentication was performed by short tandem repeat (STR) profiling as previously described (24,25). Thy1.1 positive Pmel-1 T cells that express a TCR specific for murine H-2D^b–restricted epitope of gp100 were extracted from the spleens of TCR/Thy1.1 mice as described previously (26). All cell lines were grown in RPMI media containing 2mM glutamine and 10% fetal bovine serum. Pmel-1 cell culture media additionally contained 300 I.U./ml IL-2 and 0.3 µg/ml anti-mouse CD3 (BD Bioscience). For glutamine dose dependent studies, glutamine-deficient RPMI media and dialyzed fetal bovine serum were used. C57BL/6 mice were purchased from Charles River Frederick Research Facility and Pmel-1 TCR/Thy1.1 mice were from in-house breeding colonies. CB-839 was provided by Calithera Biosciences. For in vitro studies, DMSO stocks of CB-839 were freshly prepared and used at 1µM concentration, unless dose responses were studied. For in vivo treatments, CB-839 was prepared in a vehicle containing 25% hydroxypropyl-β-cyclodextrin in 10 mmol/L citrate (pH 2.0) solution. The anti-PD1 (clone 29F.1A12), anti-CTLA4 (clone 9H10) and isotype control antibodies were from BioXCell, and administered into mice via intraperitoneal injections.

Cellular Glutamine consumption analysis.

Tumor cells (2×10³) or T cells (1×10⁴) seeded in 96 well plates were incubated in serum-free RPMI media containing 2 mmol/L glutamine for 12 hours, in the presence of the indicated concentrations of CB-839 or vehicle. After incubation, the culture media was collected, and glutamine levels were quantified using the YSI 2900 Biochemistry Analyzer.

Metabolite analysis of co-cultured cells.

Targeted metabolite analysis and [U-¹³C]-glutamine tracing analysis were performed at the MDACC metabolomics core facility. Methodologies are described in supplementary methods.

Bioenergetics Stress tests.

Real-time oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were determined by the Seahorse XFe analyzer in 96 well plates, and data was normalized using cell numbers as previously described (2).

Cell proliferation and T cell division assays.

For the cell proliferation assay, tumor cells or TILs extracted from the same tumors were seeded in 96 well plates and treated with increasing doses of CB-839, and incubated for 72h, followed by addition of Cell Titer Blue reagent (Promega) and fluorescence measurement of viable cells as described previously (5). For the T cell division assay, T cells were purified by negative selection (Stemcell) from the spleen of C57BL/6 mice. The cells were labeled with Cell-Trace CFSE (Invitrogen) following the manufacturer’s instructions, stimulated with plate-bound anti-CD3 and soluble anti-CD28 antibodies in glutamine-free RPMI-1640 media supplemented with increasing doses of fresh glutamine. Cell division was assessed after 96 hours by CFSE dye-dilution using flow cytometry and compared to non-stimulated control cells.

Apoptosis and cell death assays.

A T-cell killing assay to detect tumor cell apoptosis by caspase 3 cleavage was performed as described previously (27,28). Induction of cell death of A375HG tumor cell death by T cells was evaluated by cytoplasmic histone-associated DNA fragmentation using the Cell Death ELISA Plus kit as described earlier (25). Methodologies are provided in supplementary methods.

Mouse vaccine model study.

This study was performed as previously described (21). Briefly, Pmel-1 T cells extracted from the spleens of Pmel-1 receptor–transgenic mice were intravenously administered into C57BL/6 mice (10⁴ cells/mouse), and the mice were vaccinated with 100 µL of saline containing 100 µg/µL human gp100 and 4×10⁵ IU anti-CD40 antibody (BioXcell; clone FGK45; Cat# BP0016). The vaccinated mice also received intra-peritoneal IL2 (1 × 10⁵ IU) once on the day of vaccination and twice daily for the next two days, and were topically treated with 50 mg of 5% imiquimod cream (Fougera) on the vaccination site once after each vaccination. CB-839 treatments were performed as indicated. Mouse blood was collected from tail veins on the days shown, and complete blood counts were analyzed using a scilVet ABC hematology analyzer to determine the effect of CB-839 on white blood cell (WBC) numbers. Peripheral blood mononuclear cells (PBMC) were isolated using ACK lysis buffer and subjected to FACS analysis after appropriate antibody staining.

In vivo xenograft tumor growth studies.

The adoptive T cell therapy study was performed as described earlier (26,29), where sub-cutaneous xenograft tumors of B16-F10 were generated with 2×10⁵ cells/animal in the right flank of C57BL/6 mice. After 50–100mm³ size tumors were established, 1×10⁶ Pmel-1 T cells were adoptively transferred into tumor-bearing mice, followed by intraperitoneal administration of IL2 (5 × 10⁵ IU) twice daily for three days after T cell transfer. The mice were then treated with CB-839 (200mg/kg) by oral gavage twice daily for the indicated number of days, and tumor volumes and mice weight were recorded on the indicated days. For the immune checkpoint inhibitor studies, sub-cutaneous xenograft tumors of Yummer 1.7 were generated as above in C57BL/6 mice. Mice were administered twice daily treatments of CB-839 as above, or twice a week intraperitoneal injections of anti-CTLA4 (100 µg/animal) or anti-PD1 (500µg/animal) as suggested (20,30), either as single agents or in combination with each other and with CB-839. Tumor measurements were performed as above. Animals were excluded if they showed overt toxicity, or lost >15% body-weight. For FACS and other molecular analysis of treatment-related effects, tumors from three mice per group were harvested three hours after the specified treatments on the indicated days, and analyzed accordingly. For RNA and metabolite analysis, harvested tumors were rinsed in ice-cold normal saline and flash frozen in liquid nitrogen. All animal experiments were approved by IACUC.

Immunohistochemical analysis.

Tumors were harvested at the indicated times and immunostaining performed on tissue sections using standard methodology, scanned and analyzed as described earlier (31). Antibodies and analysis are described in supplementary methods.

FACS analysis.

Peripheral blood or tumors were harvested at the indicated time points, and processed for FACS analysis of Pmel-1 cells or tumor infiltrating lymphocytes as described earlier (21,26). Antibodies and analysis are described in supplementary methods.

Gene expression analysis.

RNA was extracted from tumors using the mirVana kit, and the expression of genes associated with the immune system were profiled using the NanoString mouse Cancer Immune Panel on the nCounter platform following the manufacturer’s instructions (NanoString Technologies).

Statistical analysis.

Statistical significance testing used in the study are described in supplementary methods.

CB-839 inhibits glutamine consumption in patient-derived melanoma cell lines and promotes their killing by autologous TILs.

Inhibition of GLS activity by CB-839 impedes cellular glutamine consumption and decreases its flux into the TCA cycle, resulting in decreased mitochondrial OxPhos (15). We assessed the effect of CB-839 on glutamine consumption in four different melanoma patient-derived cell lines by analyzing glutamine levels in extracellular media. None of these patients were previously treated with TIL or checkpoint inhibitors. Treatment of these cell lines with 1 µM CB-839 potently decreased up to 80% of glutamine consumption after 12 h treatment (Figure 1A). CB-839 at 1µM was earlier shown to cause sustained inhibition of GLS in cells grown in media containing 2 mmol/L glutamine (15). Bioenergetics stress tests using the Seahorse analyzer showed that CB-839 induced a small, but significant inhibition of cellular OCR, a surrogate of mitochondrial OxPhos, in all four cell lines (Basal OCR in Figure 1B and bioenergetics profiles in Figure S1). Thus, CB-839 decreased glutamine consumption and mitochondrial OxPhos in the melanomas. In contrast, CB-839 treatment caused minimal inhibition of both glutamine consumption and OCR in the autologous TILs derived from the same patient-derived tumors that gave rise to the above melanoma cell lines (Fig 1C and S2). CB-839 did not affect ECAR, a measure of glycolytic activity, in either the melanoma or TIL lines (Figures S1–S2).

CB-839 treatment induced a modest dose-dependent inhibition of proliferation of the four melanoma cell lines, which was similarly modest in their autologous T cells (Figure 1D, Figure S3A). We then assessed the effects of CB-839 on melanoma cell killing by their autologous T cells in vitro by evaluating cleaved caspase 3 levels in the melanoma cells. Treatment with CB-839 for 24 h did not increase cleaved caspase 3 levels in any of the cell lines. However, co-culturing the melanoma cells with their autologous TILs induced a sharp increase of cleaved caspase 3, which was further increased in the cells pre-treated with CB-839 (Figure 1E–H). These results suggest that CB-839 treatment sensitizes tumor cells to apoptotic killing by TILs.

CB-839 treatment induces differential glutamine metabolism in tumors and T cells.

Glutamine is an important fuel that supports immune cell functions. Consistent with that concept, we observed that glutamine induced a dose-dependent increase in the proliferation of anti-CD3- and anti-CD28-stimulated T cells in vitro (Figure 2A). Treatment with CB-839 did not inhibit this effect (Figure 2B). We next measured the effect of CB-839 on intracellular glutamine and its GLS-catalyzed products using LC-MS in mono-cultures of gp100 antigen-specific Pmel-1 T cells; monocultures of A375HG melanoma cells which express murine MHC class I and gp100, allowing recognition by the Pmel-1 T cells (32); or the co-cultures of Pmel-1 and A375HG. In A375HG, CB-839 caused a >3-fold accumulation of intracellular glutamine (Figure 2C), but conversely caused a >2-fold decrease of glutamate and its TCA metabolite αKGA compared to vehicle controls (Figures 2D and 2E), consistent with potent inhibition of GLS activity. The non-TCA metabolites of glutamine, proline and glutathione (GSH) also underwent similar alterations (Figures 2F and S3B). Co-culturing A375HG with Pmel-1 cells interestingly decreased αKGA levels (Figure 2E). Proline and GSH levels were however not significantly altered by the Pmel-1 cells (Figure 2E and S3B). Combination of Pmel-1 and CB-839 treatments had the most significant impact on glutamate and αKGA, but not on proline and GSH in the A375HG cells (Fig 2D–2F, S3B). Similar analysis in the Pmel-1 mono-cultures showed that CB-839 treatment decreased glutamate levels, but minimally altered αKGA, proline and GSH levels compared to vehicle treatment (Figure 2G–J, S3C). Interestingly, co-culturing with A375HG induced an increase of glutamine and all four of its metabolites in the Pmel-1 cells. Treatment of the co-cultures with CB-839 increased glutamine levels, but did not significantly alter its four metabolites (Figure 2G–J, S3C). Taken together, the results suggest that co-culture with Pmel-1 cells alters glutamine metabolism in tumor cells by an unknown mechanism that inhibits αKGA levels, which is further augmented by CB-839 treatment. In contrast, CB-839 appears to have minimal effect on αKGA and in in mono- or co-cultured T cells.

As αKGA levels may also be influenced by other pathways like glycolysis, we performed ¹³C isotope tracing analysis to determine if CB-839 and Pmel-1 induced alterations in tumor αKGA levels were a direct result of glutamine metabolism. In this analysis, we performed the same experiment as above by replacing glutamine in the culture media with [U-¹³C]-glutamine. Incorporation of [U-¹³C]-glutamine into glutamate and αKGA was determined by LC-MS, which showed that the levels of these glutamine-derived metabolites were altered in a similar manner as the total levels of these metabolites in the A375HG (Figures S3D–S3G), and in the Pmel-1 cells (Figure S3G–S3K).

To evaluate the role of glutamine on T cell-mediated apoptotic killing of melanoma cells, we performed the caspase 3 activation assay in A375HG tumor cells co-cultured with Pmel-1 T cells in glutamine-free and glutamine-rich conditions. Tumor cells were cultured in glutamine-free media for 24 h, while Pmel-1 were simultaneously activated with IL2 in 2 mmol/L glutamine-containing media. Pmel-1 cells were then washed and added to the tumor cells at a cell number ratio of 5:1, with final glutamine concentrations of 0 or 2 mmol/L in the co-cultures. The cultures were then incubated for 12 h in the presence or absence of CB-839. Caspase 3 activity in the A375HG cells treated with CB-839 and Pmel-1 cells was assessed by FACS analysis. CB-839 treatment had no impact on caspase 3 cleavage in the absence of glutamine, either as a single agent or in the presence of Pmel-1 cells (Figure 2K, left panel). In the presence of glutamine however, CB-839 alone induced a significant increase of caspase 3 cleavage, Pmel-1 cells induced a larger increase, and their combination induced the highest increase (Figure 2K, right panel). To confirm that increased caspase 3 cleavage was leading to cell death in A375HG, we performed the co-culturing and CB-839 treatments as above. After 24h, Pmel-1 cells were washed away, and cell death in the adherent A375HG cells was measured as a function of cytoplasmic histone-complexed DNA fragments (mono- and oligonucleosomes) using the cell death ELISA plus assay. The results show the levels of cell death proportionately matched the percentage of cleaved caspase 3 following the various treatments (Figure S4A). These results demonstrate that a glutamine-rich microenvironment promotes cytotoxic activity of T cells against tumor cells, and CB-839 treatment significantly enhances this effect.

CB-839 treatment increases in vivo activation status of CD8 T cells.

As our in vitro results suggested that inhibition of GLS may benefit the activity of tumor-reactive T cells, we explored the effect of CB-839 on T cell activity in vivo using a murine melanoma vaccine model (21). As shown schematically in Figure 3A, Pmel-1 T cells were injected into mice with concurrent systemic IL2 and gp100 vaccination. Mice were randomized to treatment with vehicle or CB-839 (administered days 1–5 and 27–32). Blood was drawn at multiple time points to characterize the effects of effects of CB-839 on complete blood cell counts, plasma glutamine concentrations, effector T cell viability and their activation status. We found that CB-839 significantly increased plasma glutamine levels (Figure 3B), but did not alter the levels of total WBCs, including its component lymphocytes and monocytes (Figure S4B). PBMCs were isolated and analyzed by FACS to assess Pmel-1 effector T cell numbers using antibodies against CD3, CD8 and Thy1.1 (congenic Pmel-1 marker), and their activation using anti-CD107a. CB-839 treatment induced a small increase of total Pmel-1 cell numbers, and a more significant increase in their activation status, following initial treatment on day 8, and post-booster dose treatments on day 33 (Figures 3C and 3D). These results suggest that CB-839 treatment contributes to the proliferation and activation of melanoma antigen-specific CD8 T cells in vivo.

CB-839 treatment improves the anti-tumor activity of adoptive T cell therapy.

We used an established adoptive T cell therapy model (26,29) to determine if in vivo T cell activation by CB-839 improves the anti-melanoma activity of Pmel-1 T cell therapy. In this study schematically shown in Figure 4A, IL2 stimulated Pmel-1 T cells were intravenously transferred into C57BL/6 mice growing sub-cutaneous B16 tumors. After three days of in vivo IL2 stimulation, the mice were randomized into treatment groups as shown in Figure 4B and treated for up to a month, and tumor growth was measured. Single agent CB-839 induced a small decrease of tumor growth compared to vehicle, while Pmel-1 TILs induced substantial inhibition, which was further enhanced by combining CB-839 and Pmel-1 treatments (Figure 4B). Kaplan Meier survival analysis showed that while both Pmel-1 and CB-839 treatments improved survival compared to vehicle (p=0.0004 and p=0.0004), their combination completely prevented mortality (p<0.0001) over the same time period (Figure 4C).

LC-MS metabolite analysis was performed on tumors collected on day 27, which showed that CB-839 induced a decrease of tumor αKGA and its downstream intermediates within the TCA cycle, succinate, fumarate and malate (Figure 4D). This is indicative of GLS inhibition by CB-839. Similar to effects observed in vitro (Figure 2C and 2E), Pmel-1 treatment of tumors also decreased TCA cycle intermediates. The combination of CB-839 and T cells however did not further inhibit their levels (Figure 4D). IHC analysis of the tumors showed that Pmel-1 and Pmel-1 + CB839 treatments significantly increased the infiltration of CD4+ and CD8+ cells (Figures 4E and S5); and decreased Ki67 proliferation marker staining of tumor cells, while increasing caspase 3 cleavage in them (Figure 4F). FACS analysis of TILs isolated from tumors as above showed a significant increase of IFNγ positive Pmel-1 cells in mice treated with combination of Pmel1 + CB839 versus Pmel1 alone (Figure 4G). MCP-counter was applied to the expression profiles of 36 genes (Table S1) to predict the abundance of tumor-intrinsic immune cell types, which identified >2-fold increases in intra-tumor CD8 T cells, NK cells and B cells after treatment with single agent CB-839 and Pmel-1 cells, while their combination increased only the CD8 T and NK cells (Figure S7A). Together, these results indicate that CB-839 treatment improves the efficacy of Pmel-1 cells in the B16 adoptive T cell therapy model.

Combination of CB-839 with immune checkpoint inhibitor therapy.

We next determined if CB-839 could also improve the efficacy of immune checkpoint inhibitor antibodies. We tested the combination of CB-839 with anti-PD1 and anti-CTLA4 in the Yummer 1.7 mouse melanoma model, which mirrors human melanoma with a BRAF^V600E mutation and a UV radiation-induced mutation burden (20). Treatments were performed as shown in Figure 5A, and tumor growth was analyzed. While single agent CB-839 had no effect (Figures 5B–D), single agent anti-PD1 and anti-CTLA4 antibodies significantly inhibited growth of this tumor model, which was further inhibited by their combinations with CB-839 (Figure 5B and 5C). The combination of CB-839 with anti-PD1+anti-CTLA4 co-treatments induced a marked improvement in tumor growth inhibition (Figure 5D).

IHC analysis of tumors harvested after three weeks of treatment showed that combinations of CB-839 with anti-PD1 or anti-PD1+anti-CTLA4 induced the most potent infiltration of CD8+ and CD4+ cells, potently decreased Ki67 staining of tumor cells, and increased caspase 3 cleavage in them (Figure 5E and 5F; Supplementary Fig. S6). TILs were extracted from the harvested tumors and analyzed by FACS to quantify CD107+ effector cells in the CD4+ and CD8+ T cell fractions. The results showed that CB-839, anti-PD1, anti-CTLA4, and anti-PD1+anti-CTLA4 each increased the prevalence of CD107a+ CD4+ effector T cells, and all but anti-CTLA also increased the CD8+ effector T cells. Anti-PD1 induced the most increase of CD4+ effector TILs among the single agents, and its combination with CB-839 induced the largest increase among all treatments. Interestingly, in the tumors from mice treated with anti-CTLA4 + anti-PD1 or their triple combination with CB-839, the increase of effector TILs was not as pronounced as single agent anti-PD1 or anti-PD1 + CB-839 (Figure 5G). Anti-CTLA4 and anti-PD1 have earlier been shown to potently increase proliferation of CD4+ T cells (33). It is unclear why the combination of anti-CTLA-4 with anti-PD1 slightly dampened CD107+ effector TILs compared to the single agent treatments, and whether this may be unique to the Yummer 1.7 tumor model. As the observed effector TIL status may only be a snapshot of the specific time of measurement, and potentially influenced by factors such as effector cell exhaustion or alterations in myeloid/regulatory T cells (34), longitudinal studies exploring the numbers of effector and regulatory T cells at different time intervals could be performed in future studies to understand this result in conjunction with inhibition of tumor growth. MCP-counter analysis showed an increase of CD8 T cells, cytotoxic lymphocytes, B cells and monocyte lineage with the combination treatments (Figure S7B), suggesting an overall increase of immunogenicity in tumors from respectively treated animals.

CB-839 and checkpoint inhibitor treatments increase IFNγ-associated T cell inflammatory gene expression in tumors.

Based on an earlier study that showed increased expression of the IFNγ-associated “T cell inflamed signature” genes in anti-PD1 responsive tumors (35), we analyzed these genes, PD1 (Pdcd1) and three additional genes that promote T cell cytotoxicity (Figure S5) in the tumor RNA samples, and presented the results as ratios of treatments to vehicle controls. In the B16 tumor model, CB-839 treatment of gp100 vaccinated mice (minus Pmel-1 treatment) increased the expression of seven genes, including two of the three T cell cytotoxicity genes, Gzma and Prf1, (Figures 6A and 6B). Pmel-1 treatments of the vaccinated B16 tumor-bearing mice increased the expression of only five genes, suggesting that CB-839 treatment of gp100 vaccinated mice induced a more potent immune cell activation than Pmel-1 cells at the time tested. The combination of Pmel-1 and CB-839 upregulated all but two of the T cell inflamed signature genes, and also upregulated all three T cell cytotoxicity genes (Figure 6A and 6B), which could be attributable to the higher potency of anti-tumor activity compared to the anti-PD1 + CB-839 combination.

In the Yummer1.7 model which is insensitive to CB-839, but is responsive to anti-PD-1 and anti-CTLA4 treatments, CB-839 increased the expression of only three genes in the T cell inflamed signature, but the checkpoint inhibitors induced a potent increase in almost all genes either as single agents or in combination (figure 6C and D). The largest increase was observed with the triple combination treatment of CB-839 + anti-PD1 + anti-CTLA4, with 20 of 21 genes upregulated, which correlated with enhanced anti-tumor activity. Taken together, these results suggest that CB-839 treatment synergizes with TIL and immune checkpoint inhibitor therapies to activate anti-tumor T cell functions (Figure 6E).