IL-1α Mediates Innate and Acquired Resistance to Immunotherapy in Melanoma

Key Points

• Presence of IL-1α in melanoma makes the disease resistant to immunotherapy.

• IL-1 pathway blockade enhances the efficacy of CD40 agonist therapy.

• IL-1 pathway blockade improves response to therapy through elimination of PMN-MDSCs.

Abstract

Inflammation has long been associated with cancer initiation and progression; however, how inflammation causes immune suppression in the tumor microenvironment and resistance to immunotherapy is not well understood. In this study, we show that both innate proinflammatory cytokine IL-1α and immunotherapy-induced IL-1α make melanoma resistant to immunotherapy. In a mouse melanoma model, we found that tumor size was inversely correlated with response to immunotherapy. Large tumors had higher levels of IL-1α, Th2 cytokines, polymorphonuclear myeloid-derived suppressor cells (PMN-MDSCs), and regulatory T cells but lower levels of IL-12, Th1 cytokines, and activated T cells. We found that therapy with adenovirus-encoded CD40L (rAd.CD40L) increased tumor levels of IL-1α and PMN-MDSCs. Blocking the IL-1 signaling pathway significantly decreased rAd.CD40L-induced PMN-MDSCs and their associated PD-L1 expression in the tumor microenvironment and enhanced tumor-specific immunity. Similarly, blocking the IL-1 signaling pathway improved the antimelanoma activity of anti–PD-L1 Ab therapy. Our study suggests that blocking the IL-1α signaling pathway may increase the efficacy of immunotherapies against melanoma.

Introduction

Melanoma is the most deadly form of skin cancer (1) and is estimated to have caused 60,712 deaths worldwide in 2018 (2). Currently, immunotherapies are the most promising treatments for metastatic melanoma (3). However, most patients with melanoma do not respond or only partially respond to available immunotherapies. Results of preclinical and clinical studies suggest that both innate and acquired resistance play roles in melanoma resistance to immunotherapy (4). Identifying the factors and mechanisms involved in this resistance could help us design new strategies to avoid resistance and improve therapeutic efficacy.

Inflammation has long been associated with cancer initiation and progression (5); however, how inflammation causes immune suppression in the tumor microenvironment is not well studied. Moreover, the role of inflammation in tumor resistance to immunotherapy is unknown.

IL-1 is a proinflammatory cytokine (6) that increases tumor invasiveness and metastasis. The two forms of IL-1, IL-1α and IL-1β, are derived from different genes but are functionally similar, and both forms bind to IL-1R type 1 (IL-1R1) (7). IL-1 has been associated with decreased survival of patients with melanoma (8), and endogenous IL-1 facilitates the growth of human melanoma (9). Some studies indicate that IL-1 induces accumulation of neutrophils (10) and regulatory T cells (Tregs) (11) at the site of inflammation, but the importance of this mechanism in the tumor microenvironment is unknown.

Inflammatory monocytes (iMO) play an important role in infection, inflammation, and tumor immunity (12). However, how the tumor microenvironment controls the modulation of inflammatory/regulatory monocytes and determines their function remains unclear.

In this study, we found that iMO were the key cells that produced IL-1α in immunotherapy-resistant tumors and IL-12 in immunotherapy-responsive tumors. Interestingly, IL-1α in the tumor microenvironment plays a critical role in tumor resistance to immunotherapy. Therefore, we studied the underlying mechanism of action and the effect of blocking the IL-1 pathway on tumor-specific immunity and changes in the tumor microenvironment.

CD40 is a TNF superfamily member expressed on APCs. We and others have reported that activation of tumor-associated APCs through agonist CD40 Ab or rAd.CD40L (adenovirus-encoded CD40L; ISF35) promotes antitumor immune response and suppresses tumor growth (13, 14).

Our studies revealed that anti–IL-1R1 Ab as monotherapy moderately suppressed tumor growth, blocking the IL-1 pathway with anti–IL-1R1 Ab during treatment with ISF35 (13), increased the antitumor activity of ISF35. In exploring the mechanism underlying this synergy, we found that ISF35 increased tumor levels of IL-1α and polymorphonuclear myeloid-derived suppressor cells (PMN-MDSCs) and that blocking the IL-1 signaling pathway with anti–IL-1R1 Ab significantly decreased the numbers of ISF35-induced PMN-MDSCs and their associated PD-L1 expression in the tumor microenvironment, resulting in enhanced tumor-specific immunity. Our results suggest that blocking the IL-1α signaling pathway with currently approved IL-1 pathway antagonists may increase the efficacy of immunotherapies against melanoma.

Materials and Methods

Mice and cell lines

All animal experiments were performed in accordance with National Institutes of Health guidelines and approved by the MD Anderson Cancer Center Institutional Animal Care and Use Committee. C57BL/6 mice were purchased from The Jackson Laboratory. Female mice were used at 6–12 wk of age. Mycoplasma-free B16.F10 and B16.OVA cells were obtained from the American Type Culture Collection and maintained in RPMI 1640 supplemented with 10% heat-inactivated FBS, l-glutamine, sodium pyruvate, nonessential amino acids, and penicillin–streptomycin (all from Invitrogen/Life Technologies).

Tumor induction, treatment, and monitoring

Mice were s.c. inoculated with 400,000 or 300,000 B16.F10 or B16.OVA melanoma cells or 1 million braf^V600E × pten^−/− (BP) gp100 melanoma cells (15, 16) for tumor development. Three days after tumor inoculation, mice were treated with i.p. injection of 200 μg of anti–CTLA-4 (9H10), anti-mouse Ly6G (1A8), anti–IL-1α (ALF-161), anti–IL-1β (B122), anti–IL-1R1 (JAMA-147), and/or 200 μg of anti–PD-L1 (10F.9G2) (all from Bio X Cell) or their isotype control Abs. Beginning on day 8 after tumor inoculation, mice were treated by intratumoral injection of ISF35 (13) or vehicle. All treatments were repeated every 3 d. In some experiments, mice bearing melanoma tumors from melanoma cells inoculated 3 d earlier were treated with tail vein injection of 10,000 CD11b⁺Ly6G⁺ cells isolated from large melanoma tumors in mice. Peptide vaccination was performed as described previously (17). Briefly, gp100 (a melanoma Ag)-specific Thy1.1⁺CD8⁺ T cells were adoptively transferred to tumor-bearing mice. These cells were isolated from Pmel-1 TCR-transgenic mice (18, 19). Thy1.1 is expressed in Pmel-1 mice to facilitate T cell tracking. Mice also received gp100 peptide, CD40 agonist, and TLR7/8 agonist to facilitate CD8 T cell stimulation. Tumor size was expressed as the product of perpendicular diameters of tumors measured with calipers. Mice were sacrificed when tumor size reached 200 mm².

Flow cytometry analysis

Leukocytes were isolated from mechanically disrupted tumors by using lymphocyte separation medium (Corning Cellgro). RBC lysis was performed on blood. Intracellular IFN-γ allophycocyanin (clone XMG1.2, 1:200 dilution; BD Biosciences) staining was performed using the Cytofix/Cytoperm kit (BD Biosciences). Transcription factor staining for Foxp3 PE (clone MF23, 1:200 dilution; Invitrogen) was performed using the Foxp3-staining buffer set (Affymetrix eBioscience). Briefly, the cells were first stained for surface Ag CD8 or CD4; washed, fixed, and permeabilized; and then stained for IFN-γ Foxp3. Cells were surface stained with Abs against CD45 allophycocyanin/pacific blue (clone 30-F11, 1:200 dilution; eBioscience), CD8 PE/PerCPCy5.5/FITC (clone 53-6.7, 1:200 dilution; BD Biosciences), CD44 FITC (clone MEL-14, 1:100 dilution; BD Biosciences), CD4 PerCP (clone GK1.5, 1:200 dilution; BD Biosciences), CD11b FITC (clone M1/70, 1:100 dilution; eBioscience), PD-1 PE-cy7 (clone RMP1-30, 1:400 dilution; BioLegend), PD-L1 brilliant violet 421 (clone 10F.9G2, 1:100 dilution; BioLegend), Ly6CPB (clone HK1.4, 1:300 dilution; BioLegend), Ly6G PE (clone 1A8, 1:200 dilution; BioLegend), and Ki-67 PE (clone 16A8, 1:300 dilution; BioLegend). Dead cells were excluded from analysis using a fixable live stain (LIVE/DEAD fixable aqua stain, 1:150 dilution; Thermo Fisher Scientific). Cells were stained with OVA-tetramer Ab (clone H-2Kb SIINFEKL; Beckman Coulter) per the manufacturer’s instructions. Data were acquired on a Canto II flow cytometer (BD Biosciences) and analyzed using FlowJo software (version 10).

Mass cytometry (cytometry by time of flight) for phenotypic analysis of tumor-infiltrating immune cells

To characterize differences in the tumor immune microenvironment between small, medium, and large tumors, we used cytometry by time of flight (CyTOF) to simultaneously detect different cell lineages, activation markers, cytokines, and transcription factors (Supplemental Table I). Tumors were mechanically disrupted, digested with a mixture of DNase (Sigma-Aldrich) and Liberase TL (Roche) in RPMI 1640 for 20 min, and dispersed through a 40-μm filter. CyTOF staining was performed as described by Subrahmanyam et al. (20), and data were acquired on a DVS CyTOF mass cytometer. CyTOF data were acquired and preprocessed in the MD Anderson Flow Cytometry Facility. CyTOF data were analyzed by using FlowJo software (version 10.6.0) and Spanning-tree Progression Analysis of Density-normalized Events software (version 3.0).

Cytokine/chemokine assay

Treated or untreated tumors were mechanically disrupted and centrifuged, and the supernatants were collected. An aliquot of 25 μl of a tumor supernatant from each sample was used to perform the assay. Cytokines/chemokines were measured using a Milliplex mouse cytokine/chemokine panel (Millipore) according to the manufacturer’s instructions. A fluorescence signal was measured on a Luminex 100/200 system, and data were analyzed using GraphPad Prism 8 software.

Microarray

Total RNA of small and large tumors was extracted using the RNAqueous kit (Ambion). The RNA quality was checked by using a Bioanalyzer 2100 instrument (Agilent). One hundred nanograms of total RNA was amplified and biotin labeled through a two-round Eberwine procedure using MessageAmp II and Illumina TotalPrep RNA amplification kits (Ambion) and hybridized to Illumina Ref-6 version 2 mouse whole-genome arrays.

Hierarchical clustering and heat mapping were performed using Cluster and Treeview software from Eisen et al. (21).

Statistical analysis

All results are expressed as mean ± SEM. For therapeutic experiments, 5 to 10 mice were assigned per treatment group. Data were analyzed using unpaired Student t test, and ANOVA and its differences were considered to be significant at p < 0.05. Results of survival experiments were analyzed using a log-rank Mantel–Cox test. All experiments were performed at least twice with comparable results.

Results

Tumor size is inversely correlated with response to immunotherapy

While working with a mouse melanoma model, we observed that smaller tumors responded better to immunotherapy than larger tumors did. Therefore, we stratified the mice by tumor size and treated them with ISF35 alone or with the combination of ISF35 and anti–CTLA-4 Ab. We measured response to therapy by calculating the fold change in tumor size (tumor size at the end of the treatment/tumor size at the beginning of the experiment), and we compared fold change in tumor size in the immunotherapy-treated mice with fold change in tumor size in PBS-treated mice. We found that small tumors (10–20 mm²) treated with monotherapy or combination therapy had significantly smaller fold change in size than PBS-treated small tumors (Fig. 1A). In contrast, medium tumors (30–50 mm²) treated with monotherapy or combination therapy had fold change in tumor size similar to that in PBS-treated medium tumors (Fig. 1B). These results suggested that response to immunotherapy was inversely correlated with tumor size.

FIGURE 1.

Tumor size is inversely correlated with response to immunotherapy. Mice bearing small (10–20 mm²) (A) or medium (30–50 mm²) (B) B16 tumors from cells inoculated 8 d earlier were treated as indicated, and fold change in tumor size was determined after 20 d of treatment. Data are representative of at least two independent experiments. n = 5–7 mice per group. **p < 0.01, unpaired Student t test. ISF35, adenovirus-encoded CD40L; α-CTLA-4, anti-CTLA-4 Ab.

Therefore, we sought to characterize differences in the tumor immune microenvironment between small and large tumors to determine which factors may cause resistance to immunotherapy in large tumors.

The tumor microenvironment shifts from immune active to immunosuppressive with increasing tumor size

To characterize differences in the tumor immune microenvironment between small and large tumors, we used mass cytometry (CyTOF) to analyze small (10–20 mm²), medium (30–50 mm²), and large (70–80 mm²) tumors for 70 molecules, including T cell, B cell, NK cell, and myeloid cell phenotypic and activation markers and their cytokines and chemokines. We also performed microarray and cytokine bead array assays to further evaluate the microenvironment of small and large tumors and confirm the CyTOF results. We found that numbers of iMO (CD11b⁺Ly6C⁺) did not differ by tumor size (Fig. 2A), but iMO produced more IL-1α in large tumors and more IL-12 in small tumors (Fig. 2B, 2C). As tumor size increased, the percentages of PMN-MDSCs (CD11b⁺Ly6G⁺) and Tregs (CD4⁺Foxp3⁺) increased (Fig. 2D, 2E), and the percentage of activated CD8⁺ T cells decreased (Fig. 2F). Consistent with our CyTOF data (summarized in Fig. 2G), our microarray data showed upregulation of genes related to T cells, activated APCs, and IFN-inducible chemokines in small tumors and upregulation of oncogenes and MDSC genes in large tumors (Fig. 2H). Multiplex cytokine bead array assay showed that large tumors contained significantly more Th2 cytokines, whereas small tumors contained significantly more Th1 cytokines (Fig. 2I).

FIGURE 2.

The tumor microenvironment shifts from immune active to immunosuppressive with increasing tumor size. Mice bearing small (10–20 mm²), medium (30–50 mm²), or large (70–80 mm²) tumors were killed, and tumors were removed and digested. Single-cell suspensions were prepared for CyTOF. (A–F) Percentages of iMO (A), IL-1α–producing iMO (B), IL-12–producing iMO (C), PMN-MDSCs (D), Tregs (E), and activated CD8⁺ T cells (F) in small (S), medium (M), and large (L) tumors. (G) Summary of the results of CyTOF analysis. (H and I) In another set of experiments, cells or tumor supernatant from only small tumors (S1, S2, and S3) and large tumors (L1, L2, and L3) was subjected to microarray assay (H) or multiplex cytokine bead array assay (I). n = 3–4 mice per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, unpaired Student t test.

iMO were the key cells that produced IL-1α in large (hypoxic) mouse melanomas (Fig. 2B). We explored whether monocytes/macrophages expressed IL-1α/β in human melanoma. In fact, we found that tumor-infiltrating macrophages expressed significantly more IL-1β (p = 0.001) than tumor cells or any other type of immune cell did in human melanoma when we analyzed single-cell RNA sequencing data from a previously published article (22) (Supplemental Fig. 1A).

Blocking the IL-1 signaling pathway prolongs mouse survival, enhances the efficacy of immunotherapy, and increases tumor-specific immunity

We found that in mice bearing s.c. B16.F10 tumors, anti–IL-1R1 Ab as monotherapy moderately suppressed tumor growth, but the combination of anti–IL-1R1 Ab with ISF35 (13) (Fig. 3A) or peptide vaccination (17) (Fig. 3B) significantly inhibited tumor growth and prolonged mouse survival more than ISF35 or peptide vaccination alone did. To investigate the effect of IL-1 pathway blockade on tumor-specific immunity, we treated B16.OVA tumors with anti–IL-1R1 Ab with or without ISF35. We found more OVA_257–264–specific CD8⁺ T cells in tumors treated with the combination of anti–IL-1R1 Ab and ISF35 than in tumors treated with ISF35 alone (Fig. 3C). Similarly, we found more tumor-specific and IFN-γ–producing CD8⁺ T cells in tumors treated with anti–IL-1R1 Ab plus peptide vaccination than in tumors treated with peptide vaccination alone (Fig. 3D, 3E). We also investigated the effect of IL-1 pathway blockade during ISF35 treatment of genetically engineered BP melanoma, derived from the BRafCA, PtenloxP, and Tyr::CreERT2 and engineered to express gp100 (melanoma protein). We found that treatment with anti–IL-1R1 Ab increased the efficacy of ISF35 against BP.gp100 melanoma (Fig. 3F). These findings suggested that IL-1 induces resistance to immunotherapy and that blocking the IL-1 pathway with anti–IL-1R1 Ab enhances the efficacy of immunotherapy and prolongs mouse survival.

FIGURE 3.

IL-1 pathway blockade enhances the efficacy of immunotherapy. (A and B) Mice bearing s.c. B16.F10 tumors (400,000 cells per tumor) were treated with systemic anti–IL-1R1 (α-IL-1R1) Abs on days 3, 6, 9, and 12 after tumor implantation, intratumoral injection of ISF35 (rAd.CD40L), or both (A) or with anti–IL-1R1, pmel-1 T cell and gp100 peptide vaccination (gp100 peptide + CD40 agonist + TLR7/8 agonist), or both (B). Each set of experiments included a no-treatment (NoTx) control. Graphs show Kaplan–Meier survival curves. n = 10 mice per group. Data are representative of at least two independent experiments. **p < 0.01, ***p < 0.001, ****p < 0.0001, log-rank test. (C) Mice bearing B16. Ova tumors were treated as indicated. Graph shows absolute numbers of OVA-tetramer⁺ CD8⁺ T cells in tumor. **p < 0.01, One-way ANOVA. (D and E) Mice bearing B16.F10 tumors were treated as indicated. Graphs show percentage of tumor-specific CD8⁺ T cells (D) and tumor-specific IFN-γ–producing CD8⁺ T cells (E). n = 4–5 mice per group. *p < 0.05, unpaired Student t test. Vac, peptide vaccination as in (B). (F) Mice bearing BP.gp100 tumors were treated as indicated. Graph shows Kaplan–Meier survival curves. n = 8 mice per group. *p < 0.05, ****p < 0.0001, log-rank test.

CD40 activation induces IL-1α and PD-L1–expressing PMN-MDSCs in tumors

Next, we explored the reason for the synergy between ISF35 and anti–IL-1R1 Ab in terms of enhancing antitumor immunity. We analyzed ISF35-treated and control virus-treated tumors and found significantly increased levels of IL-1α (Fig. 4A) and PMN-MDSCs (Fig. 4B) in the ISF35-treated tumors, suggesting that CD40 pathway activation induced resistance through IL-1α pathway activation. In addition, the majority of the PMN-MDSCs induced in response to ISF35 had high expression of checkpoint molecule PD-L1 (Fig. 4C). However, we did not find a higher number of PD-L1–expressing CD11b⁺Ly6C⁺ cells in response to CD40 activation (Supplemental Fig. 1B).

FIGURE 4.

CD40 activation induces IL-1α and PD-L1–expressing PMN-MDSCs in tumors. Mice bearing B16.F10 tumors from cells inoculated 8 d earlier were treated with intratumoral injection of ISF35 (rAd.CD40L) or control virus every 3 d. (A–C) Eight days after treatment initiation, tumor supernatant was analyzed for IL-1α on a Luminex system (A), and tumor-infiltrating leukocytes were analyzed for the presence of PMN-MDSCs (B) and PD-L1 expression in PMN-MDSCs (C). n = 4–5 mice per group, *p < 0.05, ****p < 0.0001, unpaired Student t test. (D and E) Mice bearing B16.F10 tumors were treated with isotype and anti-Ly6G Ab (α-Ly6G). Graphs show tumor growth at indicated time points (D) and Kaplan–Meier survival curves (E). n = 10 mice per group. *p < 0.05, log-rank test, **p < 0.01, unpaired Student t test. Data are representative of at least two independent experiments.

Next, we sought to determine the effect of PMN-MDSC depletion on tumor growth. We found that depletion of PMN-MDSCs slowed tumor growth and prolonged mice survival (Fig. 4D, 4E).

IL-1 pathway blockade reverses the ISF35-induced immunosuppressive tumor microenvironment to immune active

We studied whether IL-1 pathway blockade reversed the ISF35-induced immunosuppressive tumor microenvironment to immune active. As expected, blocking the IL-1 pathway significantly decreased the number of ISF35-induced PMN-MDSCs (Fig. 5A), decreased PD-L1 expression (Fig. 5B), and increased IFN-γ, IL-7, IL-9, and IL-5 in tumors (Fig. 5C). However, we did not observe any change in the number of Tregs after IL-1 pathway blockade (Fig. 5D).

FIGURE 5.

The tumor microenvironment changes from immunosuppressive to immune active after IL-1 pathway blockade during ISF35 treatment. Mice bearing B16.F10 tumors were treated as indicated. Graphs depict percentages of PMN-MDSCs (A) and Ly6G⁺PD-L1⁺ cells (B), cytokine levels (C), percentages of Tregs (D) and CD8⁺ T cells (E), and CD8⁺ T cell proliferation (F) in tumors. n = 4–10 mice per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, unpaired Student t test or one-way ANOVA. (G) Mice bearing s.c. B16.F10 tumors were treated with anti–IL-1R1 Abs or ISF35 (rAd.CD40L) plus anti–IL-1R1 Abs and received CD11b⁺Ly6G⁺ cells. Graph shows Kaplan–Meier survival curves. n = 5 mice per group. **p < 0.01, log-rank test. α-IL-1R1, anti-IL-1R1 Ab; α-Ly6G, anti-Ly6G Ab; NoTx, no treatment.

Next, we determined the effect of PMN-MDSC depletion on CD8⁺ T cell infiltration and proliferation. As expected, we found more CD8⁺ T cells and greater CD8⁺ T cell proliferation in mice treated with anti-Ly6G Ab (which depleted PMN-MDSCs) and ISF35 than in mice treated with ISF35 alone (Fig. 5E, 5F).

To further confirm our findings that CD11b⁺Ly6G⁺ cells were the key immune cells that were induced through the IL-1α pathway and played an important role in immunotherapy-induced resistance in melanoma, we administered purified CD11b⁺Ly6G⁺ cells (PMN-MDSCs) in mice during treatment with anti–IL-1R1 Ab with or without ISF35 and monitored their effect on treatment efficacy. As expected, we found that the adoptive transfer of CD11b⁺Ly6G⁺ cells abrogated the effect of IL-1 pathway blockade, resulting in shorter survival of tumor-bearing mice (Fig. 5G). These data suggested that blocking the IL-1 pathway causes melanoma suppression and improves response to immunotherapy through elimination of CD11b⁺Ly6G⁺ cells (PMN-MDSCs).

Next, we analyzed Ly6C⁺ monocytes (iMO) in small and large tumors for any epigenetic modification that might switch them from IL-12 to IL-1α production. We found greater H3K27 methylation in iMO from small tumors than in iMO from large tumors (Supplemental Fig. 2A), indicating that H3K27 methylation might suppress IL-1α expression in small tumors. Consistent with this result, we found that EZH2 inhibition by GSK126 to inhibit histone methylation significantly enhanced melanoma growth (Supplemental Fig. 2B).

Blocking the IL-1 signaling pathway enhances the efficacy of anti–PD-L1 Ab therapy

To determine whether blocking the IL-1 signaling pathway enhances the efficacy of anti–PD-L1 Ab therapy, we treated B16 melanoma–bearing mice with anti–PD-L1 Ab alone or both anti–PD-L1 and anti–IL-1R1 Abs. As expected, the combination of anti–PD-L1 and anti–IL-1R1 Abs delayed tumor growth (Fig. 6A) and prolonged mouse survival (Fig. 6B) significantly more than anti–PD-L1 Ab monotherapy did.

FIGURE 6.

Blocking the IL-1 pathway enhances the efficacy of anti–PD-L1 Ab therapy. Mice bearing B16 melanoma tumors from B16.F10 cells inoculated 3 d earlier were treated with anti–IL-1R1 (α-IL-1R1), anti–PD-L1 (α-PD-L1), or both and their isotype controls twice a week. Shown are tumor growth (A) and Kaplan–Meier survival curves (B). n = 5–10 mice/group. **p = 0.001, one-way ANOVA, *p < 0.05, log-rank test.

To evaluate whether IL-1 induced resistance in response to anti–PD-1 therapy, we analyzed tumor RNA sequencing data in patients treated with anti–PD-1 (23) for IL-1R1 transcripts. Interestingly, we found a trend toward an inverse correlation between IL-1R1 upregulation and response to anti–PD-1 therapy (Supplemental Fig. 3).

IL-1α but not IL-1β mediates resistance to immunotherapy in murine melanoma

Our data indicated that IL-1α mediated immune suppression in murine melanoma (Fig. 3). However, IL-1β, not IL-1α, was expressed by macrophages in human melanoma (Supplemental Fig. 1). Because IL-1α and IL-1β use the same receptor for their activity, IL-1R1, we used anti–IL-1R1 Ab in our preclinical experiments. To confirm that only IL-1α, not IL-1β, is responsible for resistance to immunotherapy in murine melanoma, we compared the therapeutic efficacy of IL-1α– and IL-1β–neutralizing Abs as monotherapy because we found insignificant differences between the survival of untreated mice and mice treated with anti–IL-1R1 Ab as a monotherapy (Fig. 3). To facilitate observation of any effect on survival, we used only 300,000 B16 cells in this experiment instead of the 400,000 cells that we used to test the efficacy of combination therapy. Interestingly, we found that only anti–IL-1α Ab, not anti–IL-1β Ab, prolonged mouse survival (Fig. 7A, 7B). These results indicated that only IL-1α is responsible for resistance to immunotherapy in murine melanoma.

FIGURE 7.

IL-1α but not IL-1β mediates resistance to immunotherapy in murine melanoma. Mice bearing B16 melanoma tumors from B16.F10 cells inoculated 3 d earlier (300,000 cells per tumor) were treated with anti–IL-1α (A) or anti–IL-β (B) Abs or isotype controls every 3 d. Shown are Kaplan–Meier survival curves. n = 5 mice per group. Anti–IL-1α: p = 0.0074, log-rank test. Anti–Il-1β: p = 0.1434, log-rank test.

Discussion

In this study, we investigated the mechanisms underlying inflammation-mediated immunosuppression and resistance to immunotherapy in melanoma. Our results implicate IL-1R1 and suggest that blockade of the IL-1 signaling pathway is a new strategy to inhibit PMN-MDSCs and/or convert an immunosuppressive tumor microenvironment to an immune active tumor microenvironment. This strategy would be more feasible, more effective, and less toxic than direct elimination of PMN-MDSCs [also known as granulocytic GMDSCs (24)] or Th2 cytokine neutralization.

Our preclinical mouse study with established B16 melanoma showed that tumor size was inversely correlated with response to immunotherapy. Our findings agree with those of Nishino et al. (25), who reported that patients with metastatic melanoma with larger tumor size on computed tomography had worse response to immunotherapy and shorter survival.

Many studies have shown that IL-1 mediates melanoma progression and angiogenesis (7, 9, 26); however, it has not been clear whether tumor or immune cells produce this cytokine in the tumor microenvironment. Our results indicate that the main source of IL-1α/β is immune cells (monocytes/macrophages); therefore, our findings should be applicable not only to melanoma but more broadly to the majority of solid tumors. A recent study showed that the combination of IL-1β blockade with either anti–PD-1 or a tyrosine kinase inhibitor had greater antitumor activity than either monotherapy alone against mouse renal cell carcinoma (27). However, in contrast to our present IL-1α study, the previous study did not show whether IL-1β was induced in response to the therapies used in the study. Therefore, it may be that renal carcinoma produces IL-1β inherently and that in contrast to IL-1α, IL-1β causes only innate resistance but not acquired resistance to cancer therapies. Consistent with an IL-1α–mediated antitumor immune suppression mechanism, the previous study also showed the role of PMN-MDSCs in IL-1β–mediated immune suppression. However, in contrast to our study, the previous study did not show an increase in tumor-specific CD8⁺ T cells after IL-1β blockade. It may be that IL-1α and IL-1β use related but distinct mechanisms to induce immune suppression.

Agonistic CD40 Abs are a promising immunotherapy for generating long-lasting tumor-specific immunity, and many clinical trials are being conducted with such Abs (28). However, many patients do not respond to them. In this study, we found that rAd.CD40L (ISF35) induces IL-1α production in tumors and that IL-1α induces resistance to CD40L therapy. Our findings suggest that we can overcome resistance to agonistic CD40 Abs by blocking the IL-1α pathway during treatment with agonistic CD40 Abs in the clinic. In contrast to tumor-intrinsic activation of β-catenin and loss of PTEN, another mechanism of immune resistance in melanoma (29), IL-1α/β–mediated immune resistance could be used by a high proportion of melanomas and may also contribute to immune evasion in other tumors beyond melanoma because IL-1α is produced from iMO independent of melanoma cell–intrinsic oncogenic pathways.

CD11b⁺Ly6C⁺ cells are sometimes described as monocytic MDSCs that cause immune suppression. However, these cells have also been described as iMO that suppress melanoma growth (30). We chose to call the cells iMO because they produced IL-1α and we did not find higher numbers of these cells in immunotherapy-resistant tumors, suggesting that these cells might not play a key role in direct immune suppression in tumors. Furthermore, in our study, iMO that produced IL-1α were protumor immune cells, and iMO that produced IL-12 were antitumor immune cells.

Our findings from this study suggest several promising avenues for further research. Our finding of an association between IL-1α and/or IL-12 expression and response to immunotherapy suggests that these molecules might be biomarkers for response to immunotherapy. Our finding of increased H3K27 methylation in iMO of small tumors suggested an association between methylation and IL-1α gene silencing; however, further study is required to test this concept. By identifying the epigenetic and/or environmental factors that induce conversion of IL-12–producing iMO to IL-1α–producing iMO as tumor size increases, we may be able to prevent such conversion, leading to inhibition of tumor progression.

It is well documented that PMN-MDSCs suppress antimelanoma immunity through various mechanisms (31–33). Similarly, we found that depletion of PMN-MDSCs slowed melanoma growth and prolonged the survival of tumor-bearing mice (Fig. 4D, 4E). Monocytic MDSCs and PMN-MDSCs are the key immune cells that express PD-L1 negative regulators in the tumor microenvironment. Our results showed high PD-L1 expression in IL-1α–induced PMN-MDSCs and lower PD-L1 expression in anti–IL-1R1 Ab–treated mice. Therefore, the therapeutic efficacy of anti–IL-1R1 blockade may result not only from lowering the numbers of PMN-MDSCs but indirectly from checkpoint elimination. We found that blocking the IL-1 signaling pathway significantly decreased the number of ISF35-induced PMN-MDSCs (Fig. 5A) and decreased PD-L1 expression (Fig. 5B); however, the frequency of PMN-MDSCs after IL-1 signaling pathway blockade alone without ISF35 treatment was not decreased compared with the frequency in nontreated mice. We saw clear reduction in CD11b⁺Ly6G⁺ cells in response to anti–IL-1R treatment when they were present in higher numbers in tumors (induced by ISF35). It may be that IL-1 acts on PMN-MDSCs in an autocrine manner and that higher frequency of PMN-MDSCs yields more receptors for anti–IL-1R1 to target. Along with the Th1 cytokines, IL-5, a Th2 cytokine, is induced in response to IL-1 signaling pathway blockade. However, IL-5 is produced not only by Th2 cells but also by mast cells, eosinophils, and basophils, and it has been reported to show antitumor activity (34). Therefore, IL-5 level would be expected to increase, possibly by innate cells.

Our findings may have an immediate impact on cancer treatment because an IL-1R antagonist, anakinra (Kineret), is commercially available and has already been approved for the treatment of rheumatoid arthritis (35).

In this article, we report for the first time (to our knowledge) that IL-1–mediated signaling inhibits tumor-specific and Th1 immunity and thus promotes resistance to immunotherapy. Our study suggests that combination therapies that include CD40 agonists, many of which have already been tested in clinical trials (28, 36), could become more effective with the addition of IL-1 blockade. Furthermore, confirmation of such a role of IL-1α would show that it is critical to determine whether U.S. Food and Drug Administration–approved immunotherapies for cancer induce IL-1α and, in the case of drugs that do induce IL-1α, to determine whether blocking the IL-1α pathway can increase the efficacy of the immunotherapies.

Disclosures

M.J.C. is an employee of Memgen and inventor on patents and patent applications concerning the composition of matter and use of ISF35 for cancer therapy. W.W.O., M.S., P.H., and M.J.C. are the authors and inventors on U.S. patent application “Methods and therapeutic combinations for treating tumors” No. 15/500,618 filed on July 31, 2015, concerning the use of ISF35 for cancer therapy. The remaining authors declare no competing financial interests.