Significant subsets of patients with oral cancer fail to respond to single-agent programmed death (PD) blockade. Syngeneic models of oral cancer were used to determine if blocking oncogenic signaling improved in vivo responses to PD-L1 monoclonal antibody (mAb). Anti–PD-L1 enhanced durable primary tumor control and survival when combined with mTOR (rapamycin), but not in combination with MEK inhibition (PD901) in immunogenic MOC1 tumors. Conversely, PD-L1 mAb did not enhance tumor control in poorly immunogenic MOC2 tumors. Rapamycin enhanced expansion of peripheral antigen-specific CD8 T cells and IFNγ production following ex vivo antigen stimulation. More CD8 T cells infiltrated and were activated after PD-L1 mAb treatment in mice with immunogenic MOC1 tumors, which were stable or increased by the addition of rapamycin, but suppressed when PD901 was added. Rapamycin increased IFNγ production capacity in peripheral and tumor-infiltrating CD8 T cells. In vivo antibody depletion revealed a CD8 T-cell–dependent, and not NK cell–dependent mechanism of tumor growth inhibition after treatment with rapamycin and PD-L1 mAb, ruling out significant effects from NK cell–mediated antibody-dependent cellular cytotoxicity. Rapamycin also enhanced IFNγ or PD-L1 mAb treatment–associated induction of MHC class I expression on MOC1 tumor cells, an effect abrogated by depleting infiltrating CD8 T cells from the tumor microenvironment. These data conflict with traditional views of rapamycin as a universal immunosuppressant, and when combined with evidence of enhanced antitumor activity with the combination of rapamycin and PD-L1 mAb, suggest that this treatment combination deserves careful evaluation in the clinical setting. Cancer Immunol Res; 4(7); 611–20. ©2016 AACR.
Carcinogen-associated head and neck squamous cell carcinoma (HNSCC) portends poor disease-specific survival, and treatment often leaves patients functionally disabled (1, 2). A significant subset of patients with HNSCC appear to have immunogenic tumors (3), and preliminary results of response rates to single-agent checkpoint inhibitors, such as monoclonal antibodies (mAb) that block the programmed death (PD) pathway, have been promising (4). To enhance the proportion of patients that respond to checkpoint inhibitor therapy, combinations of checkpoint inhibitors with standard cytotoxic and targeted therapies are being considered (5). However, many standard cytotoxic and targeted therapies can also suppress the function of effector immune cells, making these combination approaches challenging (6).
Patients with HNSCC frequently have coactivation of the phosphoinositide 3-kinase/mammalian target of rapamycin (PI3K/mTOR) and mitogen-activated protein kinase kinase/extracellular related signal kinases 1 and 2 (MEK/ERK1/2) pathways (7), making these attractive targets for HNSCC treatment. Rapamycin is an FDA-approved mTOR inhibitor (8) with preclinical and clinical promise in the treatment of HNSCC (9, 10). The MEK1/2 inhibitor PD0325901 (PD901) is an investigational small molecule with clinical activity (11) that potentiates the antitumor effects of single-agent PI3K/mTOR inhibition in HNSCC xenografts (12). PI3K/mTOR and MEK/ERK signaling promotes the development of a myeloid-rich immunosuppressive tumor microenvironment through myeloid chemokine expression (13).
We previously showed in an immunogenic model of oral cavity cancer that tumor growth inhibition seen after rapamycin therapy is CD8 dependent (14). MEK inhibitors and immune therapies have combinatorial and synergistic effects in other solid tumor models (15). We hypothesized that mTOR inhibition with rapamycin and MEK1/2 inhibition with PD901 alone or in combination may enhance the antitumor effects of PD-L1 mAb treatment. The Mouse Oral Cancer (MOC) model is carcinogen-induced, fully syngeneic on a C57BL/6 genetic background, and consists of cell lines with genetic alterations that mirror that of human oral cancer (16, 17). MOC1 cells display a high somatic alteration rate and generate tumors with slow primary tumor growth that do not metastasize, have high PD-L1 and MHC class I expression, and show robust effector immune cell infiltration (18, 19). Conversely, MOC2 cells have fewer genetic alterations, generate aggressive tumors that metastasize early to draining lymph nodes, have very low PD-L1 and MHC class I expression, and demonstrate limited effector immune cell infiltration. Use of these cell lines allows the modeling of both highly and poorly immunogenic human oral cancer, and the presence of PD-L1 in the tumor microenvironment of both models provides a rationale for using PD-L1–targeting checkpoint inhibition.
We report the results of our investigation into the combination of rapamycin and/or PD901 with PD-L1 mAb treatment in highly (MOC1) and poorly (MOC2) immunogenic oral cavity cancers. The addition of mTOR inhibition with rapamycin to PD-L1 mAb treatment enhanced durable tumor responses and survival in immunogenic MOC1 but not in poorly immunogenic MOC2 tumors. Survival, tumor growth, and immune correlative data suggested that MEK inhibition suppresses antitumor immunity in immunogenic MOC1 tumors. No treatment combination induced detectable CD8 T-cell or NK cell–mediated antitumor immunity in poorly immunogenic MOC2 tumors. Both peripheral and tumor-infiltrating CD8 T cells in rapamycin-treated tumor-bearing mice had a more robust IFNγ response when activated. Tumor-infiltrating CD8 T cells, but not NK cells, were mechanistically required for the tumor growth alteration and enhancement of tumor cell MHC class I expression observed following rapamycin and PD-L1 mAb treatment. These findings challenge traditional views of rapamycin being immunosuppressive in all clinical contexts and have important implications for the rational combination of targeted and immune-activating therapies in the clinical trial setting.
Materials and Methods
Mice and in vivo experiments
The National Institute on Deafness and Other Communication Disorders Animal Care and Use Committee (ASP1364-14) approved all in vivo studies. MOC cell lines were generated from DMBA-induced oral cavity tumors and have been validated and pathogen tested as described (16). Experiments were carried out using 8- to 10-week-old female C57BL/6 mice (Charles River) kept in a pathogen-free environment. MOC1 and MOC2 cells were maintained in media as previously described (14). MOC1 (1.5 × 106) or MOC2 (1 × 105) cells were transplanted subcutaneously and allowed to engraft to a volume of 0.1 cm3 before treatment. Different concentrations of MOC1 and MOC2 cells were used for tumor engraftment given the dramatic differences in primary tumor growth rate (14). In vivo treatments and cellular depletions were performed as detailed in the Supplementary Methods.
Tissue flow cytometry
Spleens were mechanically dissociated into single-cell suspensions with frosted histologic slides and a 70-μm strainer. Freshly resected tumor tissue was digested into a single-cell suspension using the mouse tumor dissociation kit from Miltenyi per protocol. Cell surface, intracellular, and tetramer staining was performed as detailed in the Supplementary Methods.
Ex vivo antigen-specific lymphocyte stimulation
For analysis of peripheral lymphocytes, spleen single-cell suspensions were plated in the presence of H2-Kb–restricted p15E604–611 (KSPWFTTL) peptide (1 μg/mL) for 7 days. Lymphocytes were then enriched via a histopaque gradient and stimulated with irradiated splenocytes (20 Gy) pulsed with 1 μg/mL of p15E604–611 or control OVA257–264 (SIINFEKL) peptide at a 10:1 ratio of antigen-presenting cell (APC) to T cell for 24 hours. A flow cytometry–based assay (IFNγ secretion assay, Miltenyi) was used per protocol to detect IFNγ-secreting CD8 T cells. For analysis of tumor-infiltrating lymphocytes (TIL), CD8+ TILs were sorted from tumor single-cell suspensions using a FACSAria to >99% purity and immediately stimulated for 3 hours with PMA/ionomycin (eBioscience, 10 ng/mL, 500 ng/mL, respectively) in the presence of brefeldin-A, followed by intracellular staining with an antibody to mouse IFNγ (eBioscience). Dead cells were excluded via LIVE/DEAD fixable viability dye. Data were collected and analyzed as described in the Supplementary Methods.
In vitro MOC cell treatments and flow cytometry
Cells (5 × 104) were plated into 6-well plates, allowed to adhere overnight, and treated for 48 hours with rapamycin or IFNγ (10 ng/mL) alone or in combination. Subconfluent cells were harvested with 1× TrypLE Select (Fisher Scientific) and immediately stained with antibodies as indicated and used for flow cytometric analysis as detailed in the Supplementary Methods. Dead cells excluded via 7AAD negativity.
For details, please refer to Supplementary Methods.
Tests of statistical significance between pairs of data are reported as P values, derived using a Student t test with a two-tailed distribution and calculated at 95% confidence. Comparison of multiple sets of data was achieved with one- or two-way analysis of variance (ANOVA). Survival curves were compared using the log-rank (Mantel–Cox) test. When present, error bars reflect standard error of measurement (SEM). Significance was set in each case to P < 0.05. All analyses were performed using GraphPad Prism v6.
mTOR inhibition and PD-L1 blockade in MOC tumor–bearing mice
Mice with immunogenic MOC1 tumors had durable antitumor effects and prolonged survival when mTOR, but not MEK, was inhibited, and the inhibition of primary tumor growth was CD8 T cell–dependent (14). We hypothesized that combining mTOR or MEK inhibition with PD-L1 mAb treatment could result in enhanced tumor control in immunogenic MOC1, but not poorly immunogenic MOC2, tumors. We combined the mTOR inhibitor rapamycin and the MEK inhibitor PD901 alone, or in combination with PD-L1 mAb, and assessed primary tumor growth and survival in both MOC1 and MOC2 tumor–bearing mice (Fig. 1). As shown in Fig. 1A, for mice bearing immunogenic MOC1 tumors, treatment with PD-L1 mAb alone afforded no survival advantage, although primary tumor growth had a statistically significant short-term delay, followed by tumor growth rebound (Supplementary Fig. S1A). MEK inhibition with PD901 did not enhance survival either alone or in combination with PD-L1 mAb, despite the presence of activating codon 61 Ras mutations (17). Conversely, mTOR inhibition with rapamycin alone prolonged survival of MOC1 tumor–bearing mice (P = 0.008), and this survival benefit was significantly enhanced with the addition of PD-L1 mAb (P = 0.04). The combination of rapamycin and PD901 alone enhanced survival, but the addition of PD901 partially negated the significantly improved survival and primary tumor growth delay achieved with rapamycin plus PD-L1 mAb. In poorly immunogenic MOC2 tumor–bearing mice, PD-L1 mAb monotherapy did not delay primary tumor growth or enhance survival (Fig. 1B and C; Supplementary Fig. S1B). Rapamycin and PD901 alone or in combination modestly improved survival in MOC2 tumor–bearing mice, but the addition of PD-L1 mAb did not further enhance survival or delay primary tumor growth with any combination. Thus, combination of mTOR inhibition and PD-L1 mAb enhanced tumor control over either treatment alone in immunogenic MOC1, but not in poorly immunogenic MOC2 tumor–bearing mice. However, any combination involving MEK inhibition did not demonstrate such responses in either model.
Enhancement of peripheral antigen-specific CD8 T cells by blocking PD-L1 and mTOR
Because a durable treatment effect was observed in immunogenic MOC1 but not in poorly immunogenic MOC2 mice, we investigated if correlative studies could support an immune-mediated mechanism of enhanced tumor control with combination of rapamycin and PD-L1 mAb treatment. We used the endogenous retroviral envelope protein p15E as a model antigen. It is significantly expressed in MOC1 and, to a lesser degree, in MOC2 cells (Supplementary Fig. S2). We first characterized the effect of these treatments on peripheral CD8 T cells by flow cytometry. Total peripheral CD8 T-cell counts were diminished following treatment with PD901, but not with rapamycin (Fig. 2A). Using a tetramer specific for T-cell receptors that recognizes H2-Kb–restricted p15E604–611 (KSPWFTTL), we observed that peripheral antigen-specific CD8 T cells expanded significantly with combination of rapamycin and PD-L1 mAb treatment compared with control or PD-L1 mAb alone. However, using CD107a cell-surface staining as a marker of T-cell degranulation, these antigen-specific peripheral CD8 T cells showed no evidence of activation.
To investigate whether peripheral antigen-specific CD8 T cells in MOC1 tumor–bearing mice could be activated in the presence of specific peptide, we measured IFNγ production following stimulation with OVA peptide p15E604–611 (KSPWFTTL)-pulsed APCs. Compared with mice stimulated with a control OVA peptide, baseline production of IFNγ by IFNγ-producing antigen-specific CD8 T cells in MOC1 tumor–bearing control mice was elevated and modestly, but significantly, enhanced with rapamycin treatment alone. Treatment with PD-L1 mAb significantly elevated both the percentage of IFNγ-producing CD8 T cells and their IFNγ mean fluorescence, which was then further enhanced by adding rapamycin (Fig. 2B and C, quantified in Fig. 2D). Taken together, these data suggest that expanded, antigen-specific CD8 T cells from the periphery could be activated in an antigen-specific fashion, and that the significantly elevated IFNγ responsiveness in mice treated with PD-L1 mAb was further enhanced with the addition of rapamycin.
mTOR preserves recruitment and enhances activation of infiltrating immune cells
To evaluate the effects of treatment on tumor-infiltrating immune cells, we performed flow cytometry on freshly isolated MOC1 and MOC2 tumor single-cell suspensions following treatment. In MOC1 tumors, total CD3+ TILs increased following PD-L1 mAb treatment (P = 0.009). This increase was largely preserved following the addition of rapamycin, but reduced to levels below baseline after adding PD901 (Fig. 3A). This change was not due to alterations in the number of CD4 TILs. After PD-L1 mAb treatment, MOC1 tumors had enhanced infiltration of total (P = 0.01) and p15E antigen-specific (P = 0.007) CD8 TILs that was again largely preserved with the addition of rapamycin, but reversed to equal to or less than control numbers after the addition of PD901 (P < 0.01). The absolute number of activated antigen-specific CD8 TILs positive for CD107a followed the same pattern of suppression following the addition of PD901, indicating that PD901 inhibits, but rapamycin preserves, PD-L1 mAb–induced activation of antigen-specific CD8 TILs in MOC1 tumors. Figure 3B demonstrates representative cytometry dot plots of total and antigen-specific, p15E tetramer-positive CD8 TILs from MOC1 tumors following treatment. Activation markers CD44 and PD-1 showed variable preservation of PD-L1 mAb–induced expression as indicated (Fig. 3C). Analysis of treated MOC2 tumors revealed that although the same trend of decreased total TILs following PD901 treatment was observed, the majority of TILs were CD4, and the low baseline numbers of CD8 TILs were not enhanced by any treatment (Supplementary Fig. S3A).
As peripheral CD8 T-cell activation (as investigated in Fig. 2) may not be representative of TIL functional status, we sorted CD8 TILs from treated MOC1 tumors and assessed their ability to secrete IFNγ after PMA/ionomycin stimulation (Fig. 3D). Rapamycin treatment alone enhanced CD8 TIL activation potential compared with control (P < 0.001). Treatment with PD-L1 mAb dramatically increased CD8 TIL IFNγ secretion potential, and this was further enhanced with the addition of rapamycin (P = 0.03). Thus, modulation of the tumor microenvironment after rapamycin treatment alone enhanced the activation potential of CD8 TILs. Also, whereas expression of surface activation markers CD107a, CD44, and PD-1 on CD8 TILs was unaffected by the addition of rapamycin to PD-L1 mAb treatment in MOC1 tumor–bearing mice, IFNγ secretion potential was modestly but significantly enhanced.
Effects of mTOR, MEK, and PD-L1 inhibition on tumor-infiltrating NK cells, MDSCs, and Tregs
Given the established role of NK cells in controlling malignant progression, and the potential for PD-L1 mAb–mediated, NK cell–dependent, antibody-dependent cell-mediated cytotoxicity (ADCC), we measured infiltration and activation of tumor infiltrating NKs (Fig. 4A). In MOC1 tumors, NK infiltration did not increase with PD-L1 mAb treatment alone, and this baseline was significantly reduced with PD901, but not rapamycin, treatment. Similar to the case with CD8 TILs, the number of CD107a+ NK cells was stable with the addition of rapamycin to PD-L1 mAb treatment, but decreased with PD901 alone or in combination. Low baseline NK infiltration in MOC2 was not enhanced with any treatment (Supplementary Fig. S3B).
We also measured infiltration of immunosuppressive myeloid-derived suppressor cells (MDSC) and Tregs into the tumor microenvironment following treatment. Changes in MOC1 tumor–infiltrating MDSCs and Tregs were heterogeneous and did not reach statistical significance for any treatment group (all MDSC and Treg changes between groups P > 0.05), though a trend toward increased Tregs in the rapamycin plus PD-L1 mAb treatment group was present (Fig. 4B and C). Largely, similar results were observed in MOC2 tumors with several modest but statistically significant observed changes in either MDSC or Treg infiltration following treatment (Supplementary Fig. S3C and S3D).
Robust TIL and NK infiltration into MOC1 but not MOC2 tumors, a PD-L1 mAb monotherapy response in MOC1 but not MOC2 tumor–bearing mice, and enhanced MOC1 peripheral and TIL CD8 IFNγ production capacity following rapamycin and PD-L1 mAb treatment alone or in combination provide support for an immune mechanism of tumor control in responsive MOC1 tumors. The lack of tumor growth suppression or statistically significant changes in immune correlates in MOC2 is consistent with the known poorly immunogenic status of these tumors. Consistent with previous experiments (14), MEK inhibition significantly reduced the presence of total and antigen-specific CD8 TILs and NKs in MOC1 tumors. These data, along with preserved NK-cell tumor infiltration and CD107a staining in MOC1, suggest that CD8 TILs, NK cells, or both could participate in the enhanced, durable MOC1 tumor control following combination of rapamycin and PD-L1 mAb treatment.
Depletion of CD8 but not NK cells in vivo abrogates antitumor responses
To differentiate between a CD8 TIL and NK cell–mediated ADCC mechanism for the observed treatment response, we depleted in vivo CD8 T cells or NK cells after treatment with rapamycin and PD-L1 mAb alone or in combination (Fig. 5). CD8 but not NK-cell depletion completely abrogated the antitumor effect observed in MOC1 tumor–bearing mice following PD-L1 mAb alone or in combination with rapamycin, and partially abrogated the effect observed with rapamycin monotherapy. Depletion of NK and CD8 cells from the periphery and from within the tumor microenvironment was validated at multiple time points (Supplementary Fig. S4). Thus, CD8 T cells were the effector immune cells mediating the enhanced survival and durable tumor growth suppression following withdrawal of treatment in MOC1 tumor–bearing mice.
Rapamycin enhances IFNγ production or PD-L1 mAb-inducible MHC class I expression
To explore putative mechanisms of enhanced antitumor immunity in MOC1 tumor–bearing mice treated with rapamycin and/or PD-L1 mAb, we measured cell-surface MHC class I expression on tumor cells in treated mice (Fig. 6). Although rapamycin alone did not enhance tumor cell-surface class I expression, it further increased class I expression over PD-L1 mAb treatment alone (Fig. 6A). Thus, the enhanced class I expression on MOC1 tumor cells in vivo may have been in response to increased IFNγ in the tumor microenvironment, so we modeled MOC1 cell exposure to IFNγ in vitro. Whereas rapamycin treatments from 10 to 100 nmol/L did not alter MOC1 cell-surface class I expression, lower dose rapamycin (10 or 20 nmol/L) significantly enhanced IFNγ-inducible H2-Kb and H2-Db expression (Fig. 6B). Given that IFNγ can be produced by TILs and NK cells in the tumor microenvironment, we explored whether depletion of CD8 TILs or NK cells from MOC1 tumors could reverse this enhanced class I expression on MOC1 tumor cells (Fig. 6C). Depletion of CD8 T cells, but not NK cells, reduced tumor-cell H2-Kb and H2-Db expression to near baseline, suggesting that CD8 TILs are a significant source of tumor microenvironment IFNγ that drives tumor cell class I expression in treated MOC1 tumors.
Concepts of how best to combine different cytotoxic modalities with immunotherapy to maximize antitumor effects are evolving rapidly. The use of either standard anticancer treatments, such as platinum-based agents and radiation (20), or small-molecule targeted therapies, such as tyrosine kinase inhibitors (5), can damage tumor cells and lead to enhanced antigen release, APC activation, and adaptive antitumor immunity. Yet, many of these same agents can potently suppress cells of innate and adaptive immunity (6). Here, we demonstrate that the addition of the mTOR inhibitor rapamycin to PD-L1 mAb treatment enhances both immediate and durable tumor control in an immunogenic model of oral cavity cancer. This durable response was not observed in the same model when PD-L1 mAb is combined with a MEK inhibitor, despite the fact that this mouse carried a DMBA-induced codon 61 Ras-mutant tumor. In support of our previous work demonstrating suppressed effector immune cell infiltration following MEK inhibition, we showed that PD901 treatment reverses the enhanced infiltration and activation of antigen-specific CD8 TILs observed following PD-L1 mAb treatment alone. Conversely, the addition of mTOR inhibition to PD-L1 mAb treatment preserved antigen-specific CD8 TIL and NK-cell infiltration in immunogenic MOC1 tumors and enhanced IFNγ secretion capacity in both peripheral CD8 T cells and CD8 TILs. The different responses of rapamycin- and PD901-treated mice did not appear to be due to significant changes in the infiltration of MDSCs or Tregs, and our previous work indicates that rapamycin itself had no direct cytotoxic or antiangiogenic effect on MOC tumors, despite high baseline activity of the PI3K/mTOR pathway in MOC cells (14). Mechanistically, primary tumor growth suppression following treatment with rapamycin and PD-L1 mAb alone or in combination was dependent upon the presence of CD8 TILs but not NK cells in the tumor microenvironment. Enhanced IFNγ and PD-L1 mAb-inducible MHC class I expression on MOC1 tumor cells was reversed again with depletion of CD8 but not NK cells, indicating that enhanced ability of tumor cells to present antigen is likely to play a role in the durable antitumor immunity induced with this treatment combination.
These results support emerging concepts that immune-activating treatments such as checkpoint inhibitors are effective in immunogenic tumors with high genetic alteration rates and not effective in tumors with low baseline immunogenicity and effector immune cell infiltration (21–23). Our use of both MOC1 “T-cell–inflamed” and MOC2 “non–T-cell-inflamed” tumors allows the modeling of both immunogenic and poorly immunogenic malignancies, each of which is represented roughly equally in oral cavity cancers (3, 24). MOC1 tumors represent immunogenic tumors that are likely to respond to immune-activating therapies, whereas MOC2 tumors represent very aggressive, poorly immunogenic tumors likely to be resistant to immunotherapy. Studying both types of tumors will be critical moving forward as we aim to expand the number of patients that have a durable response to checkpoint inhibition.
Our work also suggests that in this model PD-L1 mAb treatment works primarily through a CD8 T-cell and not an NK-cell mechanism, such as ADCC. ADCC against human tumor cells in vitro mediated by an anti-human PD-L1 mAb, demonstrated by Boyerinas and colleagues (25), indicates that PD-L1 mAb–induced ADCC may be model or target cell dependent. These data add to a growing body of literature indicating that MEK1/2 inhibition suppresses effector immune cell function, despite the activating Ras mutations in MOC cells. Other preclinical studies have demonstrated suppression of antigen-specific T-cell responses in vitro with MEK, but not upstream BRAF inhibition (23, 26), which is in direct contrast with work showing enhancement of on-treatment responses to PD-1 mAb monotherapy with the addition of trametinib, a MEK1/2 inhibitor similar to PD901 (15). These authors demonstrate decreased tumor infiltration of total CD8 T cells following MEK inhibition alone, but dramatically enhanced infiltration following concurrent PD-1 mAb and trametinib treatment. These discrepancies may be due to differences in sensitivity to checkpoint inhibitor or MEK inhibitor monotherapy between models and highlight the remarkable heterogeneity in therapeutic responses that can exist between different syngeneic models of murine carcinoma.
Our findings that rapamycin enhanced both the IFNγ production capacity of peripheral and tumor-infiltrating CD8 T cells and the induced expression of MHC class I expression on MOC tumor cells suggest that rapamycin has multiple and diverse effects on the MOC tumor microenvironment. The ability of IFNγ to upregulate MHC class I expression is well established (27, 28). The use of rapamycin to enhance an immune-activating treatment seems counterintuitive, given its use clinically to suppress rejection of transplanted solid organs. However, patients receiving mTOR inhibitors following solid organ transplantation have a significantly reduced incidence of developing squamous cell carcinoma (reviewed in ref. 29). Mechanistically, mTOR inhibition enhances memory T-cell development upon viral challenge (30) and the antitumor effects of therapeutic peptide vaccines (31, 32). Combining agonist CD40 mAbs with AZD8055, a potent inhibitor of both mTORC1 and mTORC2, led to synergistic antitumor effects in a model of metastatic renal carcinoma (33). Given that mTORC2 inhibition is required for the development of memory T cells (34), it is possible that dosing and duration of rapamycin treatment are factors affecting whether immune responses are enhanced or inhibited. These data, along with ours demonstrating enhanced antitumor immunity with the combination of rapamycin and PD-L1 mAb treatment, contrast with a report showing impairment of an HPV-peptide–based therapeutic vaccine after rapamycin treatment (35). Here, regression of a subset of HPV-E7+ TC-1 tumors following administration of a therapeutic vaccine was reversed following administration of rapamycin at doses similar to those of our experiments. Rapamycin treatment alone induced no regression of TC-1 tumors, suggesting significant differences in intrinsic sensitivity to rapamycin in the TC-1 and MOC models.
Although we have established a CD8 T-cell mechanism for an antitumor effect that includes increased IFNγ production capacity and MHC class I expression, other mechanisms may also play a role. Rapamycin-induced metabolic perturbations or changes in tumor vascularity could enhance oxygenation and immune cell access to tumor antigens (36). Inhibition of tumor cell PD-L1 can directly suppress mTOR activity, thus enhancing tumor microenvironment glucose levels and lifting tumor cell–induced metabolic restrictions on TILs (37). Rapamycin and PD-L1 mAb could work in an additive fashion to block tumor cell mTOR signaling to achieve similar results in MOC1 tumors. Signaling through expression of the PD-1 receptor on tumor cells can directly contribute to tumor cell growth and survival (38), but MOC cells have no detectable expression of PD-1 (data not shown). Different doses or sequencing of rapamycin with PD-L1 mAbs could have profound effects on the development of antitumor immune responses, given the known modulation of effector and memory T-cell phenotypes by rapamycin (30, 34). Additionally, combined rapamycin and PD-L1 mAb treatment did not cure T-cell–inflamed MOC1 tumors, suggesting that mechanisms of resistance to immune-mediated tumor elimination, such as the persistence of tumor-infiltrating immunosuppressive MDSCs and Tregs, may need to be addressed to maximize the potential of this therapy. Because mTOR is a key node in regulating the development and suppressive function of MDSCs in tumors (39), how rapamycin may or may not alter the T-cell–suppressive capacity of MDSCs needs to be explored in our model.
In summary, we demonstrate antitumor immune responses to PD-L1 mAb treatment that are enhanced following mTOR inhibition, but suppressed following MEK inhibition, in T-cell–inflamed MOC1 tumors. This antitumor effect was CD8 T-cell–, and not NK cell–, dependent and was associated with more IFNγ production capacity and increased expression of MHC class I on MOC cells. Noninflamed MOC2 tumors, in contrast, did not induce CD8 T-cell– or NK cell–mediated antitumor immunity when treated with combinations of targeted and checkpoint inhibitors. This report demonstrates enhanced antitumor immunity following combination of mTOR and PD-L1 mAb checkpoint inhibition in a syngeneic carcinoma model. Given that both of these agents are FDA approved for the treatment of solid tumors and have acceptable safety profiles, the combination of mTOR and checkpoint inhibition deserves careful investigation in the clinical setting.
Disclosure of Potential Conflicts of Interest
R. Uppaluri reports receiving commercial research support and service as a consultant/advisory board member for Merck. No potential conflicts of interest were disclosed by the other authors.
Conception and design: H.A. Cash, C. Van Waes, C.T. Allen
Development of methodology: E.C. Moore, H.A. Cash, C.T. Allen
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.C. Moore, H.A. Cash, A.M. Caruso, C.T. Allen
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.C. Moore, H.A. Cash, A.M. Caruso, J.W. Hodge, C. Van Waes, C.T. Allen
Writing, review, and/or revision of the manuscript: E.C. Moore, H.A. Cash, R. Uppaluri, C. Van Waes
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.C. Moore, H.A. Cash, R. Uppaluri, C.T. Allen
Study supervision: H.A. Cash, C.T. Allen
This work was supported by the Intramural Research Program of the NIH, NIDCD, project number ZIA-DC000087. H.A. Cash was supported through the NIH Medical Research Scholars Program, a public–private partnership supported jointly by the NIH and generous contributions to the Foundation for the NIH from Pfizer, Inc., the Doris Duke Charitable Foundation, the Newport Foundation, the American Association for Dental Research, the Howard Hughes Medical Institute, and the Colgate-Palmolive Company, as well as other private donors. C.T. Allen received further support through the American Academy of Otolaryngology/American Head and Neck Society Duane Sewell Young Investigators Combined Award.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.