Abstract
Certain cytotoxic chemotherapeutic drugs are immunogenic, stimulating tumor immunity through mechanisms that are not completely understood. Here we show how the DNA-damaging drug cisplatin modulates tumor immunity. At the maximum tolerated dose (MTD), cisplatin cured 50% of mice with established murine TC-1 or C3 tumors, which are preclinical models of human papillomavirus (HPV)-associated cancer. Notably, the curative benefit of cisplatin relied entirely upon induction of tumor-specific CD8+ T cells. Mechanistic investigations showed that cisplatin stimulated tumor infiltration of inflammatory antigen-presenting cells (APC) expressing relatively higher levels of the T-cell costimulatory ligands CD70, CD80, and CD86. Cell death triggered by cisplatin was associated with the release of at least 19 proteins in the tumor environment that could act as damage-associated molecular patterns and upregulate costimulatory molecules, either alone or in concert, but the responsible proteins remain unknown. Essentially, the curative effect of cisplatin was abrogated in mice lacking expression of CD80 and CD86 on APCs. Furthermore, cisplatin treatment was improved by CTLA-4 blockade, which increases the availability of CD80/86 to bind to CD28. In contrast, there was no effect of CD27 stimulation, which replaces CD70 interaction. At the cisplatin MTD, cure rates could also be increased by vaccination with synthetic long peptides, whereas cures could also be achieved at similar rates at 80% of the MTD with reduced side effects. Our findings reveal an essential basis for the immunogenic properties of cisplatin, which are mediated by the induction of costimulatory signals for CD8+ T-cell–dependent tumor destruction. Cancer Res; 76(20); 6017–29. ©2016 AACR.
Introduction
The involvement of the immune system in controlling tumor progression is now well established (1). The adaptive immune system is not only capable of recognizing tumor-associated antigens (TAA) and tumor-specific antigens (TSA), but also has the ability to eliminate cancerous cells and to generate long-term memory that protects against tumor recurrence. However, despite immune recognition many tumors can still progress due to various immune evasion strategies, which dampen or circumvent the tumor-specific immune response (2–4).
Chemotherapy is frequently used as primary care to treat cancer, to prevent metastases and to debulk the tumor mass. In spite of evidence from the late 1960s pointing at a role of the immune system in the action of chemotherapeutic drugs (5–7), it was only recently shown that the therapeutic efficacy of certain chemotherapeutic agents was partially mediated via immune system activation (8). This resurgence of the role of the immune system in chemotherapy thus revisited the notion that, rather than the original idea of neutral or immunosuppressive effects of chemotherapeutic agents, certain of these agents given as monotherapy have direct and indirect effects on the immune system, leading to improved antitumor responses (9).
Chemotherapeutic agents, such as anthracyclines, can cause cancer cell death that provokes an immunologic response, which relies on the induction of damage-associated molecular patterns (DAMP) including the exposure of endoplasmic reticulum chaperones such as calreticulin on the cell surface, the secretion of ATP, and the release of the non-histone chromatin-binding protein high mobility group box 1 (HMGB1; refs. 10–12). This so-called immunogenic cell death improves tumor antigen uptake, processing, and presentation by APCs to T cells, thereby polarizing T cells towards IFNγ-producing CD8+ T cells. Other chemotherapeutic agents such as the widely used platinum-containing cisplatin do not induce upregulation of calreticulin (13, 14). However, immune-stimulating properties have been associated with cisplatin including increased susceptibility of tumor cells to CD8+ T cells (15, 16) and synergy with TNF to induce tumor cell apoptosis (17). Thus, chemotherapeutic agents can trigger diverse immune-stimulating processes that contribute to enhanced antitumor immune responses and improved therapeutic effects. However, the immune-stimulating properties of chemotherapeutics are often researched in mouse models in which the MTD is not determined, while in the clinic chemotherapies are administered at the MTD (18). Potential effects may be therefore unexplored.
In this study, we aimed to explore in preclinical models of HPV-associated cancer the immune-associated mechanisms of cisplatin-mediated tumor destruction in settings in which the MTD was used to mimic the clinical setting. The MTD of cisplatin induces attraction of inflammatory APCs with increased expression of costimulatory molecules mediated through the release of DAMPs in the local tumor microenvironment upon cisplatin-mediated tumor cell death. This phenomenon leads to subsequent induction of tumor-specific CD8+ T cells and is an integral part of the mechanisms underlying the sustained curative action of this chemotherapeutic agent.
Materials and Methods
Mice
C57BL/6 mice were obtained from Charles River Laboratories and were used as wild-type mice. Cd80/86−/− (19) and Cd70/80/86−/− mice (20) mice were bred inhouse. Tlr4−/− mice were obtained from The Jackson Laboratory and bred inhouse. Pmel-1 TCR transgenic mice containing CD8+ T cells expressing gp10025–33-specific TCRs (21) were a kind gift from Dr. N.P. Restifo (National Cancer Institute, Bethesda, MD) and were bred to express the congenic marker CD90.1 (Thy1.1). All used knockout and transgenic mice were on a C57BL/6 background. At the start of the experiments, mice were 8 to 14 weeks old. Mice were housed in individually ventilated cages under specific pathogen-free conditions in the animal facility of Leiden University Medical Center (LUMC, Leiden, the Netherlands). All animal experiments were approved by the Animal Experiments Committee of LUMC and were executed according to the animal experimentation guidelines of LUMC in compliance with the guidelines of Dutch and European committees.
Tumor challenge models and chemotherapy
The tumor cell line TC-1 was generated by retroviral transduction of C57BL/6 lung epithelial cells with the HPV16 E6/E7 and c-H-ras oncogenes (22) and cultured as described previously (23). The tumor cell line C3 was developed by transfection of mouse embryonic cells with the HPV16 genome and an activated ras oncogene (24). Cell line authentication was carried out by flow cytometry 4 months prior to the first submission of the manuscript. Treatment schedule of each experiment is indicated in the respective figures. In brief, female mice were inoculated subcutaneously with 1 × 105 TC-1 or 5 × 105 C3 tumor cells (both preclinical models of HPV-associated cancer) in 200-μL PBS containing 0.2% BSA on day 0. Tumor size (horizontal dimension × vertical dimension) was measured two times a week using a caliper. When a palpable tumor was present on day 7 (TC-1) or day 13 (C3) post-tumor injection, mice were divided into groups with comparable tumor sizes. On day 8 and 14 after TC-1 and C3 tumor challenge, respectively, mice were treated intraperitoneally (i.p.) with the indicated dosages of cisplatin. Treatments with 1, 2, 4, and 8 but not 10 mg/kg of cisplatin were followed by a second injection on day 15 in the TC-1 tumor model. Mice were routinely weighed 2–3 times per week. After cisplatin administration, mice were weighed 3–4 times per week until mice recovered. Exclusion criteria were ulceration of tumors and insusceptibility for cisplatin treatment as evidenced by complete lack of weight loss. Mice were euthanized when tumor size reached >2,000 mm3 in volume or when mice lost >20% of their total body weight (relative to initial body mass).
Statistical analysis
Statistical analyses were performed using Prism (GraphPad). Survival data were analyzed by Kaplan–Meier and the log-rank (Mantel–Cox) test. Statistical significance was determined using the Mann–Whitney test or Student t test as indicated in the legends. P values of ≤ 0.05 were considered statistically significant.
Additional methods are described in the Supplementary Information.
Results
Efficacy of the MTD of cisplatin to control tumor progression
To identify a clinically relevant dose of cisplatin in mice, the MTD was assessed. TC-1 tumor cells, expressing the HPV-16 oncoproteins E6 and E7, were injected into mice. When a palpable tumor was present on day 8, groups of mice were treated with increasing doses of cisplatin. Administration of 1, 2, and 4 mg/kg of cisplatin did not alter the weight of the mice significantly (Fig. 1A). However, substantial weight loss was observed after each administration of 8 or 10 mg/kg of cisplatin. Mice receiving 10 mg/kg of cisplatin displayed close to 20% weight loss, preventing a second administration of cisplatin because of ethical considerations. To analyze cisplatin-induced toxicity, the levels of aspartate transaminase (AST) and alanine transaminase (ALT) were measured 4 days after administration of cisplatin as clinical biomarkers for tissue and liver injury (Supplementary Fig. S1A). Even the highest dose (10 mg/kg) of cisplatin was not associated with significant changes in AST and ALT serum levels, indicating that cisplatin treatment at these doses is not accompanied with major tissue/liver damage in mice. In addition, blood counts also showed no significant difference in the number of white blood cells, red blood cells, and platelets (PLT; Supplementary Fig. S1B).
The tumor outgrowth in mice treated with 1 or 2 mg/kg of cisplatin was comparable with untreated mice. However, the groups of mice treated with 4 or 8 mg/kg displayed a better survival of 14% and 25%, respectively (Fig. 1B and C). Remarkably, a single dose of 10 mg/kg of cisplatin cured 50% of the mice. On the basis of the clinical outcome, the weight loss and the absence of changes in AST and ALT levels, a single dose of cisplatin at 10 mg/kg was defined as the MTD.
The antitumor effect of cisplatin treatment fully depends on the induction of tumor-specific CD8+ T-cell responses
Next, we asked whether the efficacy of the MTD of cisplatin to cure half of the mice is solely determined by direct killing of tumor cells or whether also immune-associated mechanisms are implicated. To address this, we ablated several types of immune cells: that is, NK cells, CD4+ T cells, and CD8+ T cells (Supplementary Fig. S1C and S1D) in vivo through administration of depleting antibodies during cisplatin chemotherapy in TC-1 tumor-bearing mice. Clodronate liposomes were used to deplete phagocytic cells including macrophages and dendritic cells. Whereas treatment with the MTD of cisplatin was able to cure 50% of the immunocompetent mice, cisplatin only delayed tumor progression until day 20 post-tumor challenge in mice depleted of CD8+ T cells or phagocytic APCs, after which all tumors progressed rapidly (Fig. 2A–C). Depletion of CD4+ T cells and NK cells had no effect. To validate the importance of CD8+ T cells in cisplatin-mediated tumor destruction in a second model, we used HPV16-transformed C3 tumors. Administration of cisplatin markedly delayed C3 tumor outgrowth and full eradication was observed in about half of the mice. Again the sustained clinical effect fully depended on the presence of CD8+ T cells (Fig. 2D). Thus, tumor regression is a direct effect of cisplatin on tumor cells but cisplatin-induced cure, which means full tumor eradication, requires the presence of CD8+ T cells, and phagocytic APCs.
Given the role of CD8+ T cells in cisplatin-mediated tumor destruction, we determined the tumor-reactivity of circulating CD8+ T cells after cisplatin treatment. Stimulation of splenic CD8+ T cells isolated from mice that previously eradicated TC-1 tumors (day 60) resulted in the production of IFNγ and TNF toward irradiated TC-1 tumor cells but not to unrelated tumor cells (Fig. 2E and Supplementary Fig. S2A and S2B). Splenic CD8+ T cells isolated from mice with established TC-1 tumors (day 21) produced more IFNγ and TNF in response to irradiated TC-1 tumor cells compared with untreated mice (Supplementary Fig. S2B), confirming the importance of tumor-specific CD8+ T cells in tumor clearance. As a control, splenic CD8+ T cells isolated from cisplatin-treated naïve (non tumor-bearing) mice stimulated with irradiated tumor cells were unable to produce IFNγ (Supplementary Fig. S2C), indicating that the observed tumor-specific T-cell responses depends on cisplatin-treated tumor cells and are not caused by effects of cisplatin on other cells. Moreover, splenic CD8+ T cells from cured mice showed higher production of IFNγ compared with noncured mice in the C3 tumor model (Supplementary Fig. S2C).
To examine whether cisplatin treatment induced protective tumor-specific CD8+ T-cell responses, we rechallenged mice that previously eradicated TC-1 tumors after cisplatin treatment and remained tumor free for at least one month. The majority of these mice were protected against TC-1 tumor rechallenge (Fig. 2F), indicating that cisplatin treatment can lead to the induction of tumor-specific immunologic memory. Thus, cisplatin treatment of tumor-bearing mice results in the induction of tumor-specific effector and memory CD8+ T-cell responses.
Cisplatin treatment triggers enhanced levels of costimulatory molecules on intratumoral inflammatory APCs
To dissect the underlying mechanisms of the cisplatin-mediated cure of tumor-bearing mice and the connection with the important role for APCs and the induction of tumor-specific CD8+ T-cell responses, we performed a series of in vitro experiments. Cisplatin-treated tumor cells were unable to directly cause significant activation of CD8+ T cells but when antigen-pulsed APCs such as BMDCs or D1 dendritic cells were present in the cultures, CD8+ T-cell proliferation was clearly increased, confirming the importance of APCs in the cisplatin-induced immunologic response (Fig. 3A and Supplementary Fig. S3A). Moreover, the IFNγ production in these cultures correlated with the level of T-cell proliferation (as evidenced by CFSE dilution; Fig. 3B and Supplementary Fig. S3B). Together, these data indicate that cisplatin treatment of tumor cells stimulate local APCs to activate CD8+ T cells.
Given the pivotal role of costimulatory signals provided by APCs in driving T-cell responses (25), we hypothesized that cisplatin treatment mediates induction of these molecules on APCs. To test this, we exposed dendritic cells to TC-1 tumor cells, which had been incubated with increasing amounts of cisplatin causing 20% to 50% cell death overnight (Supplementary Fig. S3C). In a dose-dependent fashion, cisplatin increased the expression of the costimulatory molecules CD70, CD80, and CD86 but not that of OX40L, 4-1BBL, and ICOSL (Fig. 3C and Supplementary Fig. S3D). In contrast, cisplatin treatment of D1 cells in the absence of tumor cells did not alter the expression of CD70 and CD86 while the expression of CD80 was slightly higher (Supplementary Fig. S3E). A similar result was observed in case of cisplatin treatment of TC-1 and C3 tumor cells without the presence of APCs. Together, these results indicate that cisplatin-mediated tumor cell death leads to upregulation of CD70, CD80, and CD86 on APCs.
To demonstrate the importance of the cisplatin-mediated induction of costimulatory molecule expression on APCs, the cytokine production of antigen-specific T cells was measured in cultures with cisplatin-treated tumor cells in the presence of dendritic cells from wild-type, Cd70−/−, Cd80/86−/−, and Cd70/80/86−/− mice. IFNγ production of T cells was markedly lower when stimulated with APCs lacking the expression of CD70, CD80, and CD86 costimulatory molecules (Fig. 3D).
We hypothesized that factors released during cisplatin-induced cell death might be involved in the costimulatory molecule upregulation on APCs. First, we used cytokine multiplex assays to identify whether such factors are present in the supernatant of TC-1 tumor cells that were treated with 1 or 2.5 μg/mL cisplatin. While most cytokines such as IL1, IL12, and TNF were not changed or below the detection limit, we were able to identify type I IFNs as factors that are elevated upon cisplatin treatment (Fig. 3E). Importantly, for our study, type I IFNs upregulated the expression of CD70, CD80, and CD86 on both BMDCs and D1 cells (Fig. 3F and Supplementary Fig. S3F). Preventing the effects of type I IFNs by using a blocking antibody for the IFNα/β receptor (IFNAR) abrogated the upregulation of these costimulatory molecules on dendritic cells treated with IFNα (Supplementary Fig. S3G), but failed to block the cisplatin-mediated upregulation of costimulatory molecules, indicating that the type I IFN pathway is not essential for this occurrence (Supplementary Fig. S3H).
To further identify factors, we took a global proteomics approach using mass spectrometry allowing comprehensive analysis of induced proteins. Mass spectrometric analysis was performed to detect and quantify proteins released in the supernatant of cisplatin-treated tumor cells. In total, 2,239 unique proteins were found, of which 526 proteins were increased with a ratio of >3 by cisplatin treatment (Supplementary Fig. S4A). Protein classification using PANTHER revealed a variety of proteins with different functions (Supplementary Fig. S4B). In particular, we focused on those proteins involved in the defence response, and could act as DAMPs. In total, we found 19 of such proteins, including heat-shock proteins and HMGB1 (Fig. 3G). The latter protein is known to be released upon tumor cell death and to be capable of mediating inflammation (26, 27). To validate the mass spectrometry results, we performed an ELISA for HMGB1, revealing a similar 5-fold increase of this DAMP in the supernatant of cisplatin-treated tumor cells (Fig. 3H). Essentially, incubation of BMDCs and D1 cells with HMGB1 caused significant cell-surface upregulation of CD70, CD80, and CD86 (Fig. 3I and Supplementary Fig. S5A). However, BMDCs that are deficient in the HMGB1 receptor TLR4 did not show the abrogation of upregulation of costimulatory molecules induced by cisplatin-treated tumor cells (Supplementary Fig. S5B). Similarly, neutralizing HMGB1 with antibody in cisplatin-treated wild-type mice or cisplatin treatment of Tlr4−/− mice did not decrease the survival of treated mice (Supplementary Fig. S5C and S5D). Collectively, these data indicate that the release of a cocktail of DAMPs in the local microenvironment by cisplatin-induced tumor cell death is responsible for the upregulation of costimulatory molecules on APCs, which are required for optimal T-cell activation, and blocking of one DAMP pathway is redundant.
Next, we assessed whether similar mechanisms regarding the cisplatin-mediated induction of tumor-specific CD8+ T-cell responses operate in vivo. First the potential effect of cisplatin on the composition of the intratumoral APCs was addressed. Tumors at day 4 after systemic treatment with 10 mg/kg cisplatin were dissected (Fig. 4A and B). This moment was chosen as the tumor sizes between cisplatin-treated and nontreated mice are comparable (Supplementary Fig. S6A), thereby excluding tumor size–related differences in immune parameters. We observed an increase in the percentage of the total CD45+ leukocyte tumor-infiltrating population upon cisplatin treatment in TC-1 and C3 tumor models (Fig. 4A and B and Supplementary Fig. S6B). Especially the percentage of the CD11b+ myeloid cells was elevated while the percentage of infiltrated T cells was not affected (Fig. 4A and B). Cisplatin treatment did not decrease the percentage of intratumoral regulatory CD4+CD25+Foxp3+ T cells (Supplementary Fig. S6C).
Next, we gated on the CD3−CD11b+ myeloid cells by using our previously described gating strategy to delineate four subsets (28): Ly6ChiF4/80hi, Ly6CintF4/80int, Ly6CintF4/80hi, and Ly6ChiF4/80low myeloid cells (Fig. 4C). The Ly6CintF4/80hi and Ly6ChiF4/80low populations express high levels of CD11c and Ly6G, respectively, resembling DC-like and granulocytic APCs. In both the TC-1 and C3 tumor models, cisplatin treatment induced a predominant increase in the inflammatory Ly6ChiF4/80hi APC population (Fig. 4C and D). Remarkably, cisplatin treatment not only increased the number of Ly6ChiF4/80hi APCs but also increased the percentage of CD11chi cells within this subset (Supplementary Fig. S6D and S6E). Moreover, cisplatin treatment enhanced the levels of monocyte chemotactic protein 1 (MCP-1/CCL2) in the serum as compared with untreated and naïve mice (Supplementary Fig. S6F).
To analyze the effect of cisplatin on myeloid-derived suppressor cells (MDSC), we used a gating strategy based on CD11b and Gr-1, a myeloid differentiation antigen that consists of Ly6C and Ly6G. Cisplatin treatment did not alter the percentage of the CD11b+Gr-1hi MDSCs, nor that of CD11b+Gr-1neg cells while there was a significant increase in the CD11b+Gr-1int population upon cisplatin treatment (Supplementary Fig. S6G). The increase of the latter population reflects the Ly6Chi F4/80hi cells (population 1) (29) shown in Fig. 4C and D.
To dissect whether chemotherapy altered the costimulatory activation status of the intratumoral myeloid cells in vivo, the expression of the costimulatory molecules CD70, CD80, and CD86 was analyzed. A cisplatin-mediated increase in the expression of the costimulatory molecules CD80 and CD86 was observed for myeloid cells isolated from both TC-1 and C3 tumors, while CD70 was increased on myeloid cells from TC-1 tumors treated with cisplatin (Fig. 5A and B). When analyzed per subset, CD80 and CD86 were specifically increased on the inflammatory Ly6ChiF4/80hi APCs and not on the other subsets (Fig. 5C and D). The expression of CD70 increased on both inflammatory Ly6ChiF4/80hi and DC-like Ly6CintF4/80hi APCs in TC-1 tumors upon cisplatin treatment (Fig. 5D). Detailed analysis of the Ly6ChiF4/80hi population based on CD11cint/hi indicated that CD70, CD80, and CD86 were highly expressed on the CD11chi cells of this population (Fig. 5E). Notably, the expression of these costimulatory molecules increased significantly on the CD11chiLy6ChiF4/80hi cells upon cisplatin treatment (Fig. 5E and Supplementary Fig. S7A). Importantly, cisplatin did not affect the expression of CD70, CD80, and CD86 costimulatory molecules on myeloid cells in spleen and tumor-draining lymph node (TDLN; Supplementary Fig. S7B and S7C), corroborating that upregulation of these costimulatory molecules is not a direct effect of cisplatin on myeloid cells but that the effect of cisplatin acts strictly locally in the tumor microenvironment where both tumor cells and APCs are present. Moreover, cisplatin treatment did not alter the expression of these costimulatory molecules on either TC-1 or C3 tumor cells in vivo (Supplementary Fig. S7D and S7E). Thus, cisplatin treatment specifically upregulates the expression of the costimulatory molecules CD70, CD80, and CD86 on intratumoral inflammatory APCs with a CD11chiLy6ChiF4/80hi phenotype.
Cisplatin-mediated tumor destruction requires T-cell costimulation
In view of the upregulation of these costimulatory molecules on tumor-infiltrated APCs in vivo, their requirement for cisplatin-mediated tumor eradication was next determined. Wild-type, Cd80/86−/−, and Cd70/80/86−/− mice were challenged with TC-1 tumor cells and subsequently, when tumors were palpable, treated with 10 mg/kg cisplatin, or left untreated. When left untreated, the tumors in the wild-type and costimulation-deficient mice progressed rapidly with similar kinetics (Fig. 6A). Treatment with cisplatin resulted in significant tumor delay and eradication in 50% of the wild-type mice. In contrast, cisplatin treatment in Cd80/86−/− and Cd70/80/86−/− mice resulted only in a delay of tumor outgrowth and the vast majority of the cisplatin-treated costimulation-deficient mice did not survive (Fig. 6A and B). The survival of cisplatin-treated Cd80/86−/− and Cd70/80/86−/− mice is similar, suggesting that CD70, although upregulated on myeloid cells after cisplatin treatment, seems to play a minor role in comparison with the B7 molecules CD80 and CD86. To examine whether increment of CD70- or CD80/86-mediated costimulation could affect cisplatin-mediated tumor eradication, we used agonistic CD27 antibodies or CTLA-4–blocking antibodies, respectively. No increase in survival of cisplatin-treated TC-1 tumor-bearing mice was observed when cotreated with CD27 agonistic antibody (Fig. 6C). On the other hand, blocking the inhibitory signal via CTLA-4, leading to enhanced capacity of CD80 and CD86 to bind CD28, slightly improved cisplatin-mediated tumor eradication (Fig. 6D). Taken together, these data show that T-cell costimulation via CD80/86 on host cells is required for the full efficacy of the cisplatin induced antitumor response.
The combination of cisplatin with therapeutic vaccination strongly improves clinical outcome and depends on costimulation
Next, we assessed whether the clinical effect of tumor-specific vaccination with HPV16 SLP could be improved by combination treatment with cisplatin at different dosage levels, including the MTD. Wild-type mice were treated with HPV16 SLP vaccination on day 8 and day 22 combined with 0, 1, 2, 4, or 8 mg/kg of cisplatin on day 8 and 15 or once with 10 mg/kg only on day 8. SLP vaccination alone resulted in a delay of tumor outgrowth but after day 40 all mice succumbed. The combination of SLP vaccination with cisplatin enhanced the overall survival of mice, albeit with varying efficacy (Fig. 7A). Interestingly, the strong clinical effect (90% cure) obtained by the combination of SLP vaccination and the MTD of cisplatin, was retained in animals given vaccination and twice 80% of the MTD (8 mg/kg; Fig. 7A) while the side effects induced by cisplatin are reduced at this dose (Fig. 1A). Notably, lowering the dose of cisplatin to 4 mg/kg in combination with SLP vaccination resulted only in 50% tumor eradication, similar to what otherwise could be achieved only with cisplatin used at the MTD (Fig. 7A).
To show that also in this combined therapeutic setting costimulation is a central component, wild-type and Cd70/80/86−/− mice were treated by SLP vaccination and the MTD of cisplatin (Fig. 7B). When compared with the clinical response in wild-type mice, the absence of costimulation clearly abrogated the delay in tumor progression despite chemoimmunotherapy, and only 30% of the mice survived tumor challenge. Thus, chemoimmunotherapy strongly depends on T-cell costimulation.
Discussion
In the current study, we show that the MTD of cisplatin results in complete tumor eradication in about half of the animals. While tumor regression was observed in all mice treated with the MTD of cisplatin, full eradication of established tumors requires the induction of tumor-specific CD8+ T cells and the expression of the costimulatory molecules CD80 and CD86 on host APCs. By performing cytokine multiplex assays and mass spectrometry, we found that a cocktail of factors, which are released upon cisplatin-induced tumor cell death, have the capacity to enhance costimulatory molecule expression on inflammatory APCs. Cisplatin-mediated tumor destruction resulted not only in the induction of cytokine-producing tumor-specific effector CD8+ T cells reactive to TC-1 tumor cells but also memory CD8+ T-cell formation occurred, as cured mice were able to resist a secondary tumor challenge. Thus, the MTD of cisplatin mediates tumor cell death that both directly delays tumor growth and indirectly stimulates tumor-specific immunity to achieve full tumor regression and cure.
Chemotherapeutic agents such as vinblastine, paclitaxel, and etoposide are known to exert direct effects on the APC maturation including the increased expression of CD80 and CD86 (30). We showed that the enhanced expression of costimulatory molecules on APCs upon cisplatin treatment was only observed in the presence of tumor cells, indicating a strict requirement of cisplatin-induced tumor cell death for this type of APC maturation. Moreover, we found that cisplatin treatment in vivo resulted in an enhanced expression of costimulatory molecules exclusively on intratumoral APCs while it did not affect the expression of CD70, CD80, and CD86 in spleen and TDLN, suggesting that the costimulation-inducing factors released by cisplatin-mediated tumor cell death act only locally on neighboring APCs. Certain types of chemotherapeutic agents can induce immunogenic cell death that is demarcated by calreticulin exposure, HMGB1 release, and ATP secretion, thereby stimulating antitumor responses. Moreover, the immune-stimulatory effect of immunogenic cell death–inducing chemotherapy is abolished in mice deficient in TLR4 (31). Although cisplatin fails to meet all the proposed criteria of this immunogenic cell death as it does not induce the preapoptotic exposure of calreticulin (12), it induces HMGB1 release of tumor cells as shown by us and others (13). Notably, our data also demonstrated that HMGB1 induces the upregulation of costimulatory molecules on APCs. Consistent with this notion, it has been shown that in a type I diabetes model, HMGB1 potentially could induce CD11c+CD11b+ DC maturation and macrophage activation in NOD mice (32). In addition to HMGB1, we found production of type I IFNs after cisplatin treatment of tumor cells. Reportedly, such type I IFN induction by cisplatin is important for the maturation and migration of DCs and subsequent T-cell priming (33). Like HMGB1, we also showed that type I IFNs can upregulate costimulatory molecules on APCs, indicating that multiple factors play a role in enhancement of costimulatory molecules on APCs upon cisplatin treatment. Moreover, using mass spectrometry analyses, we found that cisplatin induces the release of a cocktail of potential immune-stimulatory proteins (e.g., HSPs), which next to type I IFNs and HMGB1, can activate APCs and contribute to T-cell stimulation (34). This would fit with recent reports that type I IFNs can act in cis with other stimuli to induce DC maturation (35). Indeed, TLR4-deficient DCs still upregulated costimulatory molecules following stimulation with cisplatin-treated tumor cells and similarly blocking of type I IFN signaling did not affect the upregulation of costimulatory molecules by cisplatin. Moreover, cisplatin treatment of TLR4-deficient mice or during neutralization of HMGB1 was as efficient as in wild-type mice.
Previously, results from head and neck cancer patients receiving radiochemotherapy (5-FU and cisplatin) indicated that one of the immune-stimulatory mechanisms could be the reduction of immune-suppressive cells in the tumor microenvironment (36). Study of the tumor microenvironment in cisplatin-treated animals, however, showed that cisplatin did not act via the reduction of intratumoral Tregs or MDSCs. Instead, an increase in a particular subset of APCs was found, which was important for the clinical effect of cisplatin as depletion of APCs by clodronate liposomes prevented the cisplatin-induced cure of tumor-bearing mice. In-depth analyses of the different subsets of APCs after cisplatin treatment showed alteration in number and composition of myeloid cells toward inflammatory Ly6ChiCD11c+ APCs. Analysis of the APC status revealed that cisplatin indirectly stimulated the expression of the costimulatory molecules CD70, CD80, and CD86 on intratumoral inflammatory APCs with a CD11chiLy6ChiF4/80hi phenotype. The enhanced presence of these APCs in the tumor microenvironment upon cisplatin treatment may be related to enhanced production of MCP-1/CCL2. These myeloid cells are well equipped to engulf apoptotic tumor cells and subsequently present tumor-specific peptides to T cells, and our data suggest that this is one of the mechanisms. In this respect, it has been demonstrated that following anthracycline therapy, intratumoral CD11c+CD11b+Ly6Chi cells have such properties as well (37), suggesting that this effect is not limited to cisplatin only. Therefore, it will be of interest to further investigate the particular characteristics of chemotherapy-induced APCs in antitumor responses.
To improve cisplatin treatment, we combined it with therapeutic vaccination. This almost doubled the response rate, leading to cure of almost all animals treated with the combination. In the past, cisplatin was shown to increase the cytotoxicity of lymphocytes against tumor cells (15, 16) and we recently revealed that T-cell–derived TNF, through upregulation of several proapoptotic proteins, sensitized tumor cells to cisplatin-induced killing (17). In clinical settings, the side effects of cisplatin treatment often results in the administration of reduced doses as part of the standard clinical treatment. Our experiments revealed that at 80% of the MTD, chemotherapy can be given at least twice, without the loss in body weight seen when the MTD is used. Importantly, while the side effects are reduced, the clinical efficacy was retained. This provides ample opportunities to combine cisplatin chemotherapy at clinically relevant doses with immunotherapy.
In conclusion, cisplatin as single-agent chemotherapy results in an enhanced attraction of myeloid cells into the tumor, increases costimulatory molecules on the intratumoral myeloid APCs, and thereby fosters the stimulation of tumor-specific CD8+ T cells. Because immune mechanisms are obviously essential for the antitumor effects by cisplatin, a cancer cell centric view of tumor cell destruction has to be abandoned. The powerful immune-stimulatory effects of cisplatin are important underlying mechanisms for the sustained clinical antitumor effect of cisplatin monotherapy and provide opportunities for new treatment options including in combination with other immunotherapy modalities such as therapeutic vaccination or immunomodulating antibodies.
Disclosure of Potential Conflicts of Interest
C.J.M. Melief is a chief scientific officer at ISA Pharmaceuticals and is a consultant/advisory board member for Immatics. S.H. van der Burg reports receiving a commercial research grant and is a consultant/advisory board member for ISA Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: E. Beyranvand Nejad, T.C. van der Sluis, C.J.M. Melief, S.H. van der Burg, R. Arens
Development of methodology: E. Beyranvand Nejad, T.C. van der Sluis, G.M. Janssen, C.J.M. Melief, S.H. van der Burg, R. Arens
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Beyranvand Nejad, T.C. van der Sluis, S. van Duikeren, G.M. Janssen, P.A. van Veelen
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Beyranvand Nejad, T.C. van der Sluis, G.M. Janssen, P.A. van Veelen, C.J.M. Melief, S.H. van der Burg, R. Arens
Writing, review, and/or revision of the manuscript: E. Beyranvand Nejad, T.C. van der Sluis, H. Yagita, P.A. van Veelen, C.J.M. Melief, S.H. van der Burg, R. Arens
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. van Duikeren, H. Yagita
Study supervision: C.J.M. Melief, S.H. van der Burg, R. Arens
Acknowledgments
The authors acknowledge Suzanne Welten, Marjolein Sluijter, Thorbald van Hall, and Margreet de Vries for providing transgenic and knockout mice. The authors would like to thank Willemien Benckhuijsen, Jan Wouter Drijfhout, and Kees Franken for constructing peptides and tetramers. The authors greatly appreciate the help from Els van Beelen for cytokine multiplex assays.
Grant Support
This work was supported by a grant from the Dutch Cancer Society (KWF 2009-4400 to C.J.M. Melief and S.H. van der Burg), a grant from Leiden University Medical Center (LUMC; E. Beyranvand Nejad), a Gisela Thier grant from LUMC (R. Arens), and a grant from the Macropa Foundation (R. Arens).
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.