IL1α Antagonizes IL1β and Promotes Adaptive Immune Rejection of Malignant Tumors

Both IL1α and IL1β mediate their activities through binding to a single receptor (1). Binding of either IL1α or β to the type I IL1 receptor (IL1R1) activates MyD88 and TRAF6, leading to NF-κB activation and MAP kinase 3 activation with attendant JNK, P38, and ERK activation (2, 3). These activities are mediated via a “Toll Interleukin 1 Homology” domain which is common to the intracellular domains of both IL1R1 and most TLRs (4). This relationship to Toll-like receptors has led to IL1α and IL1β being considered primarily innate immune cytokines (5), inducing a robust inflammatory response upon binding to IL1R1 on target cells (6–8). Regardless of the focus on innate immunity, IL1 activity was once described as “lymphocyte-” or “thymocyte-activating factor” for its role in costimulating thymocyte cell division (9). Indeed, early assays for IL1 activity involved T-cell proliferation (10, 11). More recent studies have shown that IL1 is critical for Th17 cell maturation, is a growth and survival factor for naïve T cells, and enhances antigen driven CD8 responses (12–14). Evidence for a role of IL1 in adaptive immune responses involving T cells is abundant, although frequently overlooked.

Although IL1α and IL1β signal through the same receptor, they differ (15). IL1β is secreted from cells as a 17 kDa molecule cleaved by caspase 1 from an inactive 31 kDa precursor (pro-IL1β) through the inflammasome (16, 17). In contrast, IL1α is predominantly cell associated (18), and although a 17 kDa active form of IL1α can be generated by calpain cleavage, its 31 kDa precursor is also active (19, 20). IL1α can also be detected in a cell-associated form on the cytoplasmic membrane, where it resides as a biologically active molecule (18). IL1β, but not IL1α, can be detected in circulation in a number of diseases (3, 21, 22). IL1β has been a therapeutic target: canakinumab (a therapeutic antibody to IL1β) is used to treat a number of inflammatory and rheumatologic disorders (23–25).

The roles of IL1α and IL1β in cancer biology have been studied as well (26, 27). Innate immune inflammation is associated with tumor-promoting activities (27, 28). IL1β is thought to enhance tumor growth by inducing angiogenesis and blocking antitumor immunity in part by inducing myeloid-derived suppressor cells (MDSC; refs. 26, 27, 29). The role of IL1α in cancer is less clear, with some reports suggesting a protumorigenic role (30) and other reports suggesting tumor inhibition (31). Overexpression of IL1α in fibrosarcoma lines appears to induce antitumor immunity, although the mechanism is thought to involve innate immune cells (32). On the other hand, overexpression of IL1α is a negative prognostic factor in many human cancers (33, 34). Blockade of both IL1α and IL1β activity with antagonists or antibodies has been proposed as a general approach to therapy of cancer (35, 36).

To assess the relative roles of IL1α and IL1β in tumor growth, we chose mice deficient in these two molecules as well as mice deficient in the type I IL1R. This allowed us to assess functional deficiency in only IL1α (IL1α^−/−), only IL1β (IL1β^−/−), or both IL1s (IL1R1^−/−). In addition, we used neutralizing antibodies to both IL1α and IL1β. Our results indicate that complete blockade of IL1 activity through deficiency of IL1R1 has no effect on tumor growth; this was true across different cellular lineages and in two different anatomic sites (skin and lung). However, deficiency of IL1β, but not IL1α, reproducibly led to inhibition of tumor growth. This tumor growth inhibition was dependent on an intact adaptive immune system as well as the presence of IL1α. IL1β^−/− mice that rejected tumors retained durable antitumor immunity mediated by tumor-specific CD8⁺ resident memory T cells (T_RM).

Our observations suggest that IL1α may play a larger role than previously considered in adaptive immune antitumor responses, and that altering the balance between IL1α and IL1β may have a role in therapeutic antitumor immunity. Antibodies to IL1α, IL1β, and a soluble form of the receptor antagonist (IL1Ra, which blocks all IL1 signaling at the IL1R1) are already in clinical trials, and our results suggest that globally inhibiting IL1 activity may not be an optimal approach.

C57BL/6J mice, IL1R1^−/− (Il1r1tm1Imx/J) mice, Rag1^−/− (B6.129S7-Rag1tm1Mom/J) mice, B6 SCID (B6.CB17-Prkdcscid/SzJ) mice, and Lang-DTR (B6.129S2-Cd207tm3(DTR/GFP)Mal/J) mice were obtained from The Jackson Laboratory. IL1α^−/− and IL1β^−/− mice were obtained from Dr. Yoichiro Iwakura (Tokyo University of Science, Tokyo, Japan). IL1β^−/− mice were crossed with Rag1^−/− mice, B6 SCID mice, and Lang-DTR mice to generate IL1β^−/−Rag1^−/− mice, IL1β^−/−SCID mice, and IL1β^−/−Lang DTR mice. Mice were bred in a biosafety level 1 facility at Harvard Medical School (HMS, Boston, MA) and Brigham and Women's Hospital (BWH, Boston, MA). All mice were handled in accordance with guidelines set out by the Center for Animal Resources and Comparative Medicine (CCM) at HMS/BWH (Boston, MA). All procedures and protocols were advised and approved by an Institutional Animal Care and Use Committee as well as the veterinary staff at the CCM.

EL-4 T-cell lymphoma cells were received in 2014. B16-F10 melanoma cells were received in 2016. Lewis lung carcinoma cells (LLC) were received in 2011. YUMM1.7 melanoma cells were received in 2015. B16-F10^LUC melanoma cells were received in 2017. These cell lines were not authenticated in the past year. Mycoplasma testing was performed in 2019 and was negative for all cell lines. Cell lines used in these experiments were cultured for 7–14 days. EL-4 cells were grown in complete RPMI1640 medium. B16-F10, YUMM1.7, and LLC cells were grown in complete DMEM. Tumor cells were injected intradermally into the right flank and tumor development was monitored over time. Tumor volume was calculated as follows: (major circumference × minor circumference²)/2. Mice were euthanized when external necrosis was present or the tumor size reached 2 cm in any direction. For melanoma cell metastasis studies, 2–5 × 10⁵ tumor cells were intravenously injected into the mice. Lung tumor foci were enumerated at different time points. For the tumor imaging experiment: 10 days after B16-F10^LUC intravenous injection, mice were imaged twice a week for 2 to 3 weeks with IVIS Lumina III Imaging System (PerkinElmer). Ten microliter per gram body weight of firefly Luciferin (15 mg/mL) was given to the mice by intraperitoneal injection 10 minutes before imaging. Seven minutes after Luciferin administration, mice were placed in an induction chamber to be anesthetized with 3% isoflurane. After imaging, mice were allowed to recover in their home cages. Living image 4.2 Software (PerkinElmer) was used to optimize image display and analyze images.

Mice were injected intraperitoneally with 100–400 μg of anti-mIL1α and anti-mIL1β (clones ALF-161 and B122, BioLegend) one day prior to tumor cell injection and then every third day until the end of the experiment. To deplete T cells, mice were injected intraperitoneally with 250 μg of anti-CD4 (clone GK1.5) and (or) anti-CD8(clone 2.43) 2 days prior to tumor cell injection and then every third day until the end of the experiment. FTY 720 was used to prevent T-cell egress from lymph nodes (LN). Mice were injected intraperitoneally with 1 mg/kg FTY 720 2 days prior to tumor cell injection and then every 2 days until the end of the experiment.

Wild-type (WT) and IL1β^−/− mice were treated with 200 μL clodronate liposomes (5 mg/mL, Encapsula Nano Sciences) at four sites bordering the tumor implantation site by intradermal injection 2 days before and three times a week after tumor cell inoculation until the end of the experiment. PBS-containing liposomes were used as controls. IL1β^−/− Lang DTR mice were treated with 1 μg diphtheria toxin (DT) at the specified timing of administration to remove Langerhans cells (LC) and Lang⁺ dendritic cells (DC).

Primary EL-4 tumors and skin draining LNs from the tumor injected site were obtained and digested with collagenase D (Roche, 1108886600, final concentration 400 units/mL) at 37°C for 20–30 minutes to get single-cell suspensions. After cell staining, different subsets of cells were sorted with a FACS Aria III (BD Biosciences) or acquired on a FACS Canto (BD Biosciences) and analyzed using FlowJo Software (Version 6.4.7, Tree Star). The following mAbs were purchased from BioLegend: FITC-conjugated anti-CD3ϵ (clone 145-2C11/catalog no. 100306), anti-MHCII (clone M5/114.15.2/catalog no. 107606), anti-Ly6G (clone 1A8/catalog no. 127605), PE-conjugated anti-CD11C (clone N418/catalog no. 117308), anti-CD103 (clone 2E7/catalog no. 121406), anti-CD62L (clone MEL-14/catalog no. 104408), anti-CD69 (clone H1.2F3/catalog no. 104508), anti-MHCII (clone M5/114.15.2/catalog no. 107607), PerCP-conjugated anti-CD44(clone IM7/catalog no. 103036), anti-CD11b (clone M1/70catalog no. 101230), anti-CD3 (clone 145-2C11/catalog no. 100325), anti-CD11c (clone N418/catalog no. 117325), APC-conjugated anti-CD8 (clone 53-6.7/catalog no. 100712), MHCII (clone M5/114.15.2/catalog no. 107613), CD19 (clone 1D3/CD19/catalog no. 152409), Ly6C (clone HK1.4/catalog no. 128015), PE-Cy7 conjugated CD45 (clone 30F-11/catalog no. 103113), and APC-Cy7 conjugated CD8 (clone 53-6.7/catalog no. 100714).

For melanoma cell metastasis models, mice were euthanized and lungs were isolated and snap frozen in liquid nitrogen before being stored at −80°C. Total RNA was extracted using RNeasy Fibrous Tissue Mini Kit (Qiagen). For EL-4 intradermal injection models, primary tumors were harvested and total RNA was extracted with RNeasy Mini Kit (Qiagen). RNA Plus Mini Kit (Qiagen) was used to extract RNA from cultured 3T3, EL-4, B16, LLC cells, and sorted cells from tumor and LNs. RNA content in the samples was measured using a NanoDrop. RNA (0.5 μg per sample) was reverse transcribed into cDNA with iScript cDNA Synthesis Kit (Bio-Rad). Triplicate cDNA products were then mixed with Fast SYBR Green Mix and the primers specific for GP100, IL1α, IL1β, IL1R1, or GAPDH. GAPDH was used for normalization. These primers were used: GP100 forward: 5′GAGCTTCCTTCCCGTGCTT3′, GP100 reverse: TGCCTGTTCCAGGTTTTAGTTAC. IL1α forward: 5′CGAAGACTACAGTTCTGCCATT3′. IL1α reverse: 5′GACGTTTCAGAGGTTCTCAGAG3′. IL1β forward: 5′GCAACTGTTCCTGAACTCAACT3′, IL1β reverse: 5′ATCTTTTGGGGTCCGTCAACT3′. IL1R1 forward: 5′GTGCTACTGGGGCTCATTTGT3′, IL1R1 reverse: GGAGTAAGAGGACACTTGCGAAT. GAPDH forward: 5′AGGTCGGTGTGAACGGATTTG3′, GAPDH reverse: TGTAGACCATGTAGTTGAGGTCA. The real-time PCR was performed with a StepOne Plus Real-time PCR System (Applied Biosystems). The thermal cycle profile was as follows: 95°C for 10 seconds, 40× (95°C for 15 seconds, 58°C for 30 seconds). GP100 gene expression in the lungs from control WT mice was used as a baseline. For qRT-PCR analyses, transcript levels were normalized to GAPDH and represented as 2^–ΔC_t, where ΔC_t is C_t(target gene) – C_t(GAPDH).

Skin draining LNs were obtained from IL1α^−/− mice and WT mice, after collagenase D digestion, single-cell suspension was prepared and stained with DC markers. After cell sorting, DCs were pulsed with 1 μg/mL SIINFEKL peptide at 37°C for 2 hours, washed two times, and cocultured with OT-1⁺ T cells with or without exogenous IL1α cytokine (R&D Systems). Supernatants were collected 60 hours after incubation for IFNγ detection with commercial ELISA Kits (BioLegend).

IL1-deficient chimeric mice were generated with WT, IL1β^−/−, and IL1R1^−/− mice by irradiating recipient mice at a sublethal dose and injecting 5 to 10 × 10⁶ bone marrow (BM) cells intravenously from age-matched donor mice. The level of blood chimerism was determined by flow cytometry. 96% to 98% chimerism was achieved 8 weeks after reconstitution.

Unpaired Student t test or repeated-measures two-way ANOVA followed by Bonferroni correction were performed with Prism software. P < 0.05 was considered significant.

To examine the role of the IL1 family in tumor immunity, we implanted EL-4 lymphoma cells intradermally in C57BL/6 (WT), IL1R1^−/− mice, IL1α^−/− mice, and IL1β^−/− mice. Our initial results demonstrated that tumors grew rapidly and with similar kinetics in WT, IL1R1^−/−, and IL1α^−/− mice. These mice had to be sacrificed between day 18 and 20 because of increasing tumor burden. However, EL-4 tumor growth was significantly reduced in IL1β^−/− mice. Many IL1β^−/− mice developed small tumors that subsequently disappeared, suggesting tumor rejection. More than 75% of IL1β^−/− mice eventually cleared tumor and became long-term survivors (Fig. 1A). To ensure that this was not a phenomenon unique to EL4 lymphoma cells, similar experiments were done with malignant tumors derived from different cellular lineages: YUMM1.7 melanoma cells (derived from a genetically defined BRAF^600EPten^−/− melanoma), B16-F10 melanoma cells, and LLC cells (Fig. 1B–D). In all cases, WT, IL1R1^−/−, and IL1α^−/− mice developed large tumors with similar kinetics, whereas tumor growth in IL1β^−/− mice was reduced. In many cases the tumors grew transiently and then regressed in IL1β^−/− mice, suggesting active rejection.

To determine whether reduced tumor growth in IL1β^−/− mice was specific to skin, B16-F10 melanoma cells were injected intravenously into WT, IL1R1^−/−, IL1α^−/−, and IL1β^−/− mice to simulate metastatic tumor spread. At day 11 and 16, metastatic tumor development was assessed in the lungs. Visual inspection showed diminished tumor growth in the IL1β^−/− lungs compared with WT, IL1R1^−/−, and IL1α^−/− lungs (Fig. 1E). The number of B16-F10 tumor foci was significantly decreased in the lungs of IL1β^−/− mice (Fig. 1F). When lungs were assayed for melanoma-specific GP100 by RT-PCR, there was a significant reduction of GP100 mRNA in the lungs of IL1β^−/− mice compared with WT, IL1R1^−/−, and IL1α^−/− mice (Fig. 1G).

We further explored whether EL4 or B16-F10 cells produced IL1α or β in vitro. IL1α, IL1β, and IL1R1 mRNA was undetectable in these cell lines. Furthermore, EL-4 tumors growing in IL1α^−/− mice did not express IL1α, tumors growing in IL1β^−/− mice did not express IL1β, and tumors growing in IL1R1^−/− mice did not express IL1R1 (Supplementary Fig. S1).

IL1α and IL1β both signal through the IL1R1, yet only deficiency of IL1β inhibited tumor growth. This suggested a role for IL1α in contributing to the tumor suppressive phenotype seen in IL1β^−/− mice. To test this possibility, WT mice were treated with neutralizing antibodies to IL1α or IL1β. Blocking antibody directed at IL1β inhibited EL-4 tumor growth reproducibly, whereas neutralizing antibody to IL1α had no effect, indicating that IL1β blockade could partially recapitulate the IL1β^−/− phenotype. Combining neutralizing antibodies to IL1α and IL1β in WT mice abrogated the effect of anti-IL1β alone, suggesting a role for IL1α in promoting tumor inhibition (Fig. 2A). In contrast to WT mice, neutralizing antibody to IL1β did not reduce tumor growth in IL1R1^−/− or in IL1α^−/− mice, suggesting that IL1α:IL1R1 interactions are critical for tumor inhibition (Fig. 2B). The requirement for intact IL1α signaling was highlighted when anti-IL1α–blocking antibody partially reversed the tumor suppression phenotype of IL1β^−/− mice. More than 80% of IL1β^−/− mice cleared EL-4 tumors and became long-term survivors, whereas all of the WT mice were sacrificed for progressive tumor growth by day 20. However, treatment of IL1β^−/− mice with neutralizing antibody to IL1α restored tumor growth; only 30% of these IL1β^−/− mice were able to permanently clear EL-4 tumors (Fig. 2C). These data suggested that the presence of both an intact IL1 signaling pathway (IL1R1) and IL1α were critical for the IL1β^−/− tumor suppressive phenotype.

The observations that tumors initially grew and then subsequently regressed in IL1β^−/− mice led us to suspect a process mediated by adaptive immunity. To investigate further, we treated IL1β^−/− and WT mice with either control antibodies or antibodies to both CD4 and CD8 to deplete T cells from the circulation prior to intradermal injection of EL-4 cells. As shown in Fig. 3A, tumors grew in WT mice, whether treated with control antibody or antibodies to CD4 and CD8. In IL1β^−/− mice treated with the control antibody, tumors grew briefly and then regressed, consistent with previous observations. This phenotype was reversed in IL1β^−/− mice treated with antibodies to CD4 and CD8: in such mice, tumors grew as rapidly as in WT mice without evidence of rejection. Similar results were obtained in mice implanted with B16-F10 melanoma cells (Supplementary Fig. S2).

To further elucidate the roles of CD4⁺ and CD8⁺ T cells in this T-cell–mediated tumor regression in IL1β^−/− mice, we treated IL1β^−/− and WT mice with either anti-CD4 or anti-CD8. Anti-CD4–treated IL1β^−/− mice could still clear EL-4 tumors and became long-term survivors, similar to IL1β^−/− mice treated with control antibody. In contrast, EL-4 tumor growth was restored in anti-CD8–treated IL1β^−/− mice (Fig. 3B). We also compared tumor-infiltrating CD8⁺ T cells among WT and IL1-deficient mice. We found that tumors from IL1β^−/− mice contained significantly more CD8⁺ T cells compared with tumors from WT, IL1α^−/−, and IL1R1^−/− mice assessed at the same time point (Fig. 3C).

To explore this observation in the B16-F10 lung metastasis model, we treated IL1β^−/− and WT mice with either control antibodies or antibodies to both CD4 and CD8 prior to intravenous injection of B16-F10^LUC cells. Diminished tumor growth by bioluminescence imaging was observed in IL1β^−/− mice treated with control antibody at different time points. In contrast, after T-cell depletion, B16 melanoma tumor growth was restored in IL1β^−/− mice (Fig. 3D). Lungs were harvested from these mice at day 18 for tumor foci numeration and GP100 analysis. Consistent with the imaging results, control IL1β^−/− mice manifested fewer tumor foci and decreased expression of GP100 compared with control WT mice, whereas T-cell–depleted IL1β^−/−mice demonstrated comparable numbers of tumor foci and GP100 expression in the lungs compared with T-cell–depleted WT mice (Fig. 3E).

To further investigate the contribution of the adaptive immune system to tumor immunity in IL1β^−/− mice, we bred these mice with immunodeficient Rag 1^−/− and SCID backgrounds. After the requisite number of backcrosses, we subjected IL1β^−/−Rag 1^−/− and IL1β^−/−SCID mice to intradermal EL-4 tumor implantation. Although IL1β^−/− mice showed tumor resistance, both IL1β^−/−Rag 1^−/− and IL1β^−/−SCID mice showed tumor growth indistinguishable from WT mice (Fig. 4A). Thus, in the absence of adaptive immune cells, the IL1β^−/− tumor suppressive phenotype is abolished.

A primary adaptive immune response to tumor involves recognition of tumor antigen in draining LNs, followed by proliferation of tumor-specific T cells and migration of effector T cells from LN to tumor-bearing tissue. Effector T-cell migration out of LNs requires downregulation of sphinogosine-1-phosphate receptor activity and can be blocked by FTY720, an S1PR1 agonist that traps T cells within LNs. Treatment of mice with FTY720 allowed for rapid tumor growth IL1β^−/− mice, indicating that exit of T cells from the LN and their recruitment to tumors was required to mediate protective immunity in the IL1β^−/− setting (Fig. 4B).

To investigate whether the IL1β^−/− mice that previously cleared EL-4 tumor developed long-lasting antitumor immune memory, we injected EL4 cells (5 × 10⁶) into these IL1β^−/− tumor survivor “memory” mice. Age-matched naïve IL1β^−/− mice grew tumors with this EL4 cell inoculum. However, memory IL1β^−/− mice that had previously rejected lower dose intradermal implantation of EL4 cells showed no tumor growth, suggesting durable immune memory. In contrast to naïve IL1β^−/− mice, FTY720-treated memory IL1β^−/− mice did not develop tumors, indicating that mobilization of T cells from LNs was not required, consistent with a protective role mediated by tissue T_RM cells (Fig. 4C). Naïve IL1β^−/− mice had few skin CD8⁺ T cells, but a distinct population of CD103⁺CD62L⁻CD69⁺ CD8⁺ T cells could be found in tumor survivor memory mice, in both tumor-exposed and -distant skin, consistent with the presence of T_RM cells (Fig. 4D). Finally, we asked whether MDSCs could be identified in IL1β^−/− mice. We found that both monocyte MDSC (M-MDSC) and polymorphonuclear MDSC (PMN-MDSC) were decreased in numbers in IL1β^−/− mice in tumors, blood, and lymphoid tissue (Supplementary Fig. S3).

To determine the contribution of antigen-presenting cells (APC) in the antitumor immune response observed in IL1β^−/− mice, IL1β^−/− and WT control mice were injected at the site of tumor implantation with either PBS or clodronate-encapsulated liposomes 2 days before EL-4 cell intradermal injection and then every 4 days until the end of the experiment. Intradermal injection of clodronate liposomes at the site of tumor implantation in WT mice had little effect. However, clodronate liposome injection in IL1β^−/− mice reversed the tumor suppressive phenotype and tumors grew progressively. Control liposome injection had no effect on tumor growth or survival in either IL1β^−/− or WT mice (Fig. 5A). The efficacy of the clodronate-mediated depletion of APCs was confirmed by significant reduction of migratory DCs (mDC, MHCII^high CD11c⁺ cells) and partial depletion of MHCII⁺CD11b⁺ cells in the skin draining LNs of treated mice (Fig. 5B).

Because macrophages are also depleted by clodronate, we investigated the role of epidermal Langerhans cells and dermal CD207⁺ DCs in the tumor-suppressive response by crossing IL1β^−/− mice with Lang-DTR transgenic mice expressing the diphtheria toxin receptor under the control of the murine CD207 promoter (37). Administration of a single dose of DT can deplete both LC and CD207⁺dDC within 48 hours. Epidermal LC repopulation is slow and does not reach completion for 8 weeks, whereas CD207⁺dDC repopulate dermis within a few days. Systemic DT injections 13 days before tumor implantation allow repopulation of CD207⁺dDC by the first week of tumor growth, while systemic DT injections every 48 hours throughout the time course of the experiment would deplete both LC and CD207⁺dDC. Most of the PBS-treated Lang DTR IL1β^−/− mice cleared EL-4 tumor and became long-term survivors, as observed in the IL1β^−/− mice. Similar results were found in Lang DTR IL1β^−/− mice that received systemic treatment with DT 13 days before tumor injection (these mice were defined as “Lang DTR IL1β^−/−, DT day -13”). In contrast, all Lang DTR IL1β^−/− mice treated with systemic DT 1 day before EL-4 tumor cell injection and then every 48 hour intervals (these mice were defined as “Lang DTR IL1β^−/−, DT 48 h int”) developed large tumors and had to be sacrificed at day 20 due to high tumor burden (Fig. 5C). Thus, repetitive DT treatment during the tumor growth phase restored tumor growth in Lang DTR IL1β^−/− mice although a single dose prior to tumor implantation did not. These results suggest that CD207⁺dDC are required to cross-present tumor antigen to T cells in IL1β^−/− mice and inhibit tumor growth. We next asked whether APCs expressed IL1α and β. We determined that both MHCII⁺CD11c⁺ DCs and MHCII⁺CD11b⁺CD11c⁻ myeloid APCs expressed IL1α, β, and IL1R1. Moreover, IL1α^−/− DC were less effective at stimulating CD8⁺T cells to produce IFNγ, whereas exogenous IL1α enhanced IFNγ production (Supplementary Fig. S4).

We next subjected mice to lethal irradiation and reconstitution with BM to prepare chimeric mice. When WT mice were irradiated and reconstituted with IL1β^−/− BM, tumor growth was rapid, even more rapid than when WT BM was used for reconstitution. In contrast, when IL1β^−/− mice were irradiated and reconstituted with either IL1β^−/− or WT BM, significant tumor suppression was observed, recapitulating the IL1β^−/− tumor-suppressive phenotype (Fig. 6, left). However, this protection against tumor growth was abrogated if BM from IL1R^−/− mice was transferred to IL1β^−/− mice (Fig. 6, right), suggesting that IL1R1 expression on effector T cells was required for the tumor-suppressive effect.

Our previous data (Fig. 2A) indicates that administration of blocking antibodies specific for IL1β can suppress EL-4 T-cell lymphoma tumor growth in WT mice and partially mimic the antitumor phenotype seen in IL1β^−/− mice. These findings prompted us to test the effect of anti-IL1β on tumor growth in lung using our model. B16-F10^LUC melanoma cells were injected intravenously in WT mice and tumor growth was compared using bioluminescent imaging. Anti-IL1β significantly slowed growth of melanoma cell tumors in lung, as compared with isotype control (Fig. 7A). Lungs were harvested from control antibody–treated and anti-IL1β–treated mice at day 10 and day 19 for tumor foci numeration. Consistent with the imaging results, anti-IL1β–treated mice manifested fewer tumor foci compared with control antibody–treated mice (Fig. 7B).

Because our experiments showed that the tumor resistance of IL1β^−/− mice is dependent on CD8⁺ T-cell–mediated tumor rejection, we hypothesized that anti-IL1β might synergize with immune checkpoint inhibitor antibodies directed at PD-1. In mice treated with anti-IL1β combined with anti-PD-1, T lymphoma tumors in the skin of WT mice grew at a slower rate and to a smaller final volume than with either antibody treatment alone (Fig. 7C). In conclusion, neutralizing antibodies to IL1β, alone or in combination with anti-PD-1, show promise for cancer immunotherapy.

IL1α and IL1β arose from a gene duplication event less than a million years ago and have only 25% homology at the amino acid level (1, 38, 39). Although IL1β is found in all vertebrate species studied, IL1α is only found in mammals (39). Both ligands, however, signal exclusively through the IL1R1 (1). Although IL1α and IL1β do bind equally to IL1R1, there is evidence that the soluble “decoy” IL1R2 binds preferentially to IL1β (40). The relative effect of this phenomenon on tumor growth in IL1β^−/− and IL1α^−/− mice was not explored in this study but may be worth investigating. We have demonstrated that IL1α and IL1β have distinct roles in tumor biology. Deficiency in IL1β, whether caused by genetics or by antibody blockade, leads to inhibition of tumor growth and often immune-mediated tumor rejection. However, complete blockade of IL1 signaling in IL1R^−/− mice abrogates this tumor suppressive effect, suggesting a possible antitumor role for IL1α. This activity of IL1α was further demonstrated by the use of IL1α-neutralizing antibodies, which blocked the immune-mediated rejection of tumors in IL1β^−/− mice. In addition, the slower tumor growth in WT mice induced by anti-IL1β is reversed by the simultaneous administration of anti-IL1α. Further indirect evidence for an immunostimulatory role of IL1α comes from the lack of a tumor-suppressive phenotype when anti-IL1β was given to either IL1α^−/− or IL1R1^−/− mice.

The evidence for adaptive immune–mediated tumor protection in IL1β^−/− mice is compelling. First, tumors grew initially in most animals, then subsequently diminished in size and became undetectable. These resolving tumors contained abundant CD8⁺ T cells, in contrast to tumors from WT, IL1α^−/−, and IL1R1^−/− mice. Second, depletion of CD8⁺ T cells from tumor-bearing mice reversed the suppressive phenotype conferred by the IL1β^−/− background. Depletion of CD4⁺ T cells had little effect. Third, crossing IL1β^−/− mice to immunodeficient mice, SCID and Rag1^−/−, also abrogated the tumor-suppressive phenotype. Fourth, blocking the migration of effector T cells from LN to peripheral tissue with FTY720 also blocked tumor immunity in IL1β^−/− mice. Fifth, depletion of CD207⁺ dermal DCs in IL1β^−/− mice abrogated the tumor suppressive effect, suggesting that these cross-presenting DC interacted with CD8⁺ T cells to mediate tumor immunity. Finally, the >70% of IL1β^−/− mice that reject tumors develop durable immunity and are resistant to subsequent implantation of high numbers of tumor cells. This immunity could not be abrogated by FTY720, indicating that it resided within skin. When skin was examined for the presence of tissue T_RM cells, CD8⁺ T_RM that also expressed CD69 and CD103 could be identified.

We demonstrated the IL1β^−/− antitumor immune effect in mice bearing tumors of multiple cellular lineages: EL4 lymphoma cells, B16-F10 and Yumm1.7 melanoma cells, and LCC cells. The effect was reproducible whether the cells were injected intradermally or intravenously. Thus, tumor growth in both the skin and the lung was reduced in IL1β^−/− mice. These data suggest that this observation is not an artifact of a single-tumor model system but is more broadly applicable. Our BM chimera experiments demonstrate that radioresistant cells from the IL1β^−/− host are critical for the antitumor immune effect. The effect was strongest when IL1β^−/− BM was transferred, but was still evident when WT BM was transferred. However, the effect was abrogated when IL1R1^−/− BM was transferred, suggesting that that the putative immune effector cell requires intact IL1 signaling (Fig. 6). We speculate that T cells activated by IL1α (e.g., on the cell surface of APCs) are responsible for the observed antitumor immunity.

Our study suggests that inhibiting IL1 signaling indiscriminately (e.g., with anakinra or rilanocept) may not be an optimal approach in cancer immunotherapy, as the beneficial effects of IL1α signaling on antitumor immunity would be lost. In contrast, blockade of IL1β appears to induce antitumor immunity, provided IL1α and IL1R1 are present. Whether it is blockade of angiogenesis or inhibition of the generation of MDSC's (or both) that are responsible for the tumor-promoting effects of IL1β is unclear. Other variables in the tumor microenvironment may also be at play. We also showed an additive, possibly synergistic, effect of IL1β antibody blockade and PD-1 blockade. Three therapeutic antibodies against PD-1 and one against IL1β (canakinumab) are FDA approved. The clinical trial that used canakinumab to prevent complications of atherosclerotic disease had the unexpected effect of reducing the incidence of lung cancer in recipients. This circumstantial evidence is consistent with our results and suggests that clinical trials aimed at blocking IL1β activity in cancer are warranted.

No potential conflicts of interest were disclosed.

Conception and design: T. Tian, R.C. Fuhlbrigge, T.S. Kupper

Development of methodology: T. Tian, Y. Pan, R. Lock, R.C. Fuhlbrigge, T.S. Kupper

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Tian, S. Lofftus, Y. Pan, C.A. Stingley, J. Zhao, T.Y. Pan, R. Lock, J.W. Marglous, K. Liu, H.R. Widlund, T.S. Kupper

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Tian, S. Lofftus, Y. Pan, K. Liu, R.C. Fuhlbrigge, T.S. Kupper

Writing, review, and/or revision of the manuscript: T. Tian, C.A. Stingley, R.C. Fuhlbrigge, T.S. Kupper

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Lofftus, C.A. Stingley, J. Zhao, T.Y. Pan, K. Liu, H.R. Widlund, T.S. Kupper

Study supervision: R.C. Fuhlbrigge, T.S. Kupper

Other (assisted T. Tian with editing the article): S.L. King

Other (imaging expertise): K. Cichowski

This work was funded in part by grants from the NIH (R01 AI127654 to T.S. Kupper).

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.