IL33 Is a Key Driver of Treatment Resistance of Cancer

Recurrence and treatment resistance are the major causes of cancer-associated death. The topics that most resonated has been epithelial–mesenchymal transition (EMT) that enables tumor escape by conferring mesenchymal and stem properties such as high motility and dormancy (1), and numerous genomic, proteomic, microenvironmental and immunologic mechanisms have been demonstrated (2, 3). However, little is known about the precise mechanisms underlying the metastatic colonization of cancer stem cells (CSC) after terminating EMT within the niche, and thus practical treatment strategies targeting CSCs have not been established yet. Although its reversal MET has been believed as the sequential step in the mechanism (4), relapsed and metastatic tumors appear more aggressive than primary tumors, implying a possible different mechanism.

We have been investigating the interplay between cancer EMT and host immunity, and previously demonstrated that tumor metastasis (lung and bone marrow) is a possible risk factor of resistance to immune checkpoint inhibitors (ICI) using mouse metastasis models (5). Particularly in the bone metastasis models, which were implanted with murine melanoma B16-F10 cells transduced with an EMT-governing transcriptional factor snail (designated F10-snail⁺), tumor growth and metastasis were adversely promoted by ICI treatment just like hyperprogression reported in the clinical settings (6). In this study, we harvested and biologically and immunologically analyzed B16-F10 cells obtained from the metastatic bone marrow of the implanted mice, and attempted to define the molecular mechanisms underlying the adverse effect on tumors.

Five-week-old female C57BL/6 mice were purchased from Charles River Laboratories, and were maintained under pathogen-free conditions. The mice were used according to the protocols approved by the Animal Care and Use Committee at the National Cancer Center Research Institute (Tokyo, Japan).

Human breast cancer cell lines (MCF7 and MDA-MB-231) were purchased from ATCC, and were authenticated by short tandem repeat profiling. Murine melanoma B16-B10 cells were purchased from the Cell Resource Center for Biomedical Research at Tohoku University in Japan. We used B16-F10 cells transfected with plasmid vector pcDNA3.1(+) encoding neomycin-resistant gene with or without murine snail (F10-snail⁺ or F10-mock) that we established before (7). All tumor cells were tested for Mycoplasma negativity using a Hoechst-staining Detection Kit (MP Biomedicals) and were expanded and frozen in liquid nitrogen to avoid changes occurred by a long-term culture until used in experiments.

F10-mock cells were harvested from subcutaneous tumors (F10-primary) and femur bone marrow (F10-BM) of mice 25 days after implantation and were cultured in 10% FBS/DMEM with Geneticin (Merck) to select tumor cells. The details were described in the Supplementary Methods.

B16-F10 cells and MCF7 cells were transfected with plasmid vector pCMV6-ENTRY (OriGene Technologies) encoding murine or human il33 by electroporation (0.4 kV, 25 μFD), and were cultured in 10% FBS/DMEM with Geneticin. The details were described in the Supplementary Methods.

We assessed cellular functions: cell proliferation (2 days) by cell count or WST-1 assay (Takara), cell adhesion (1 hour) using fibronectin-coated multiwell plates (Corning), and cell invasion (8 hours) using a transwell chamber with a Matrigel-coated membrane (Corning) as described before (7). To determine cell cycle, after fixation with ethanol, cells were treated with propidium iodide (PI; 50 μg/mL) and RNase A (200 μg/mL) for 30 minutes, and were analyzed by flow cytometry. Cells were stimulated with IL13 (1 ng/mL; PeproTech) for 2 days, and were tested for adhesion and invasion. To determine chemosensitivity, tumor cells were treated with paclitaxel (Wako), 5-fluorouracil (5-Fu; Wako) or gemcitabine (Wako) for 3 days (0–100 μg/mL, two-fold serial dilution), and the data were indicated as the percentage of untreated control (100%). Cell death was analyzed by flow cytometry after staining with PI and Annexin V. For tracking cell division, PKH67-labeled cells were used. In the in vivo experiments, tumor cells were subcutaneously (3 × 10⁵ cells) and intravenously (3 × 10⁵ cells) implanted into mice, and tumor volume (0.5 × length × width², mm³) was measured. For a convenience of observation and quantification, we assessed lung metastasis by counting the number of tumor metastatic nodules in lung, albeit tumor metastases in many tissue organs of the mice.

For il33 knockdown, we used two siRNAs targeting distinct il33 sequences or one scrambled sequence as a negative control (Invitrogen; Supplementary Methods). The siRNAs were complexed with jetPEI (PolyPlus) according to the manufacturer's instruction before transfection. The transfection efficacy was validated by RT-PCR 1 to 2 days after transfection.

After flow cytometric blocking, cells were stained with the immunofluorescence-conjugated antibodies (Supplementary Methods). For intracellular staining, cells were treated with Cytofix/Cytoperm solution (BD Biosciences) before the staining. In mouse study, data were acquired using the FACSCalibur cytometer (Becton Dickinson), and were analyzed by Cellquest software (BD Biosciences). In clinical study, data were acquired using a BD LSR Fortessa X-20 cytometer (Becton Dickinson), and were analyzed by FlowJo software (BD Biosciences). Before defining the specific molecular expressions, debris was firstly excluded by forward scatter/side scatter, followed by gating CD45⁺ leukocytes, and immunofluorescence intensity was compared to the isotype control (Supplementary Fig. S1).

IL17RB⁺ cells were sorted from spleen cells (SPC) of mice 7 days after 5-Fu treatment using a BD IMag System (BD Biosciences) with anti-mouse IL17RB mAb (R&D Systems) and the secondary anti-rat Ig magnetic beads (BD Biosciences). The purity (> 90%) of the IL17RB⁺ cells was validated by flow cytometry. The IL17RB⁺ cells (1 × 10⁶/mL) were cultured in 10%FBS/RPMI1640 at 37°C for 3 days, and the supernatant fluid was harvested and filtrated (ø = 0.22 μm) before stock at 4°C. IL13 in the supernatant was measured using an ELISA Kit (R&D Systems). The IL17RB⁺ supernatant or IL13 (1 ng/mL; PeproTech) were added to a CTL induction system with splenic CD3⁺ T cells, gp70 peptide (MBL), antigen-presenting cells (inactivated bulk SPCs) in the presence or absence of anti-mouse IL13 mAb (1 μg/mL; R&D Systems). Six days later, the sorted CD8⁺ T cells were tested for IFNγ production (24 hours) and cytotoxic activity (ET ratio = 25:1, 4 hours) as described before (7). The IL17RB⁺ cells (3 × 10⁵) were coinjected with B16-F10 cells (3 × 10⁵) in mice, and tumor volume was measured. In a setting, anti-mouse IL13 mAb (20 μg) was intratumorally (i.t.) injected in the mice on days 4 and 8 after coinjection.

To evaluate the antitumor efficacy on both subcutaneous tumor growth and metastasis mimicking metastatic cancer patients, tumor cells were both subcutaneously (3 × 10⁵ cells) and intravenously (3 × 10⁵ cells) implanted in mice. The mice were intraperitoneally treated with PBS or 5-Fu (20 mg/kg; Wako) on days 4 to 8 after tumor implantation, or were intratumorally treated with anti-mouse PD1 mAb (BioLegend), anti-mouse IL33 mAb (R&D Systems), or the isotype control (R&D Systems) on day 5 (n = 5–10 per experiment). Tumor volume was measured twice a week. Metastatic nodules in lung were counted, and subcutaneous tumors and spleens were harvested for assays on days 14 to 18.

For IHC analysis, we purchased paraffin-embedded tissue sections (normal mammary tissues, and primary and metastatic tumor tissues) of stage II–III patients with breast cancer from SuperBioChips, and stained with immunofluorescence-conjugated anti-human IL33 mAb (R&D Systems), anti-human IL17RB mAb (R&D Systems), or the isotype IgG (BioLegend) as described before (5). The immunofluorescence intensity was automatically measured as pixel counts at two fields per section using a LSM700 Laser Scanning Microscope (Carl Zeiss), and the average was plotted in graphs. For flow cytometric analysis, EDTA-added peripheral blood was collected from healthy donors (n = 4) and patients with stage IV metastatic colorectal cancer (n = 9; age 57–79) after receiving written informed consent (November 2018 and June 2019), according to the protocol approved by the Institutional Review Board at the National Cancer Center (Tokyo, Japan). PBMCs were isolated by Ficoll (Nacalai), and were stained with the specific mAbs (Supplementary Methods). All activities were conducted in accordance with the ethical principles of the Declaration of Helsinki.

Data indicate mean ± SD unless otherwise specified. Significant differences (P < 0.05) were statistically evaluated using GraphPad Prism 7 software (MDF). Data between two groups were analyzed by the unpaired two-tailed Student t test. Data among multiple groups were analyzed by one-way ANOVA, followed by the Bonferroni post hoc test for pairwise comparison of groups based on the normal distributions. Nonparametric groups were analyzed by the Mann–Whitney test. Mouse survival was analyzed by the Kaplan–Meier method and ranked according to the Mantel–Cox log-rank test. Correlation between two factors was evaluated by the nonparametric Spearman rank test.

To compare tumor properties between primary sites and metastatic sites, B16-F10 cells were both subcutaneously and intravenously implanted into mice, and tumor cells were harvested from the subcutaneous tumors (F10-primary) and the metastatic bone marrow (F10-BM) on day 25 after implantation. The F10-BM cells slightly expressed an epithelial marker E-cadherin, but highly expressed a mesenchymal marker Snail in the cytoplasm, but not in the nucleus, implying deregulation state (Fig. 1A). The F10-BM cells showed significant lower proliferative activity and higher resistance to treatment with 5-Fu with a slight increase at low doses as compared with F10-primary cells and F10-snail⁺ cells derived from bone marrow (P < 0.05; Fig. 1B). In the in vivo setting, the F10-BM tumor also significantly slowly grew than F10-primary tumor (P = 0.002), but the growth was adversely promoted by 5-Fu treatment (days 4–8; P = 0.002 vs. PBS treatment), although F10-snail⁺ tumors were just unresponsive to the treatment (Fig. 1C). These suggest that the F10-BM cells acquire a distinct property from that of the F10-primary after metastasis. The F10-BM tumors showed significant increase in ST2/ST2L⁺ cells including c-kit⁺FceRIa⁺ mast cells (P = 0.002 vs. PBS treatment), CD11b⁺Gr1⁺ myeloid-derived suppressor cells (MDSC; P = 0.005), and IL17RB⁺GATA3⁺ type 2 innate lymphoid cells (ILC2s; P = 0.005) after the treatment (Fig. 1D). This observation raised a possibility of its ligand IL33 release from the progressing F10-BM tumors. Indeed, higher IL33 production was observed in the F10-BM cells compared with other clones, and IL33 knockdown with the specific siRNAs significantly reduced the chemoresistance (P < 0.05 vs. control siRNA; Fig. 1E). Similar results were observed using highly bone-metastatic breast cancer IL33⁺ MDA-MB-231 cells after siRNA-il33 transfection (P < 0.02 vs. control siRNA; Supplementary Fig. S2). We chose breast cancer cell lines because bone metastasis is most frequently seen in breast cancer in clinical settings (8, 9). These results suggest IL33 is involved in the treatment-induced tumor progressive mechanisms.

IL33 is a member of the IL1 family expressed in some types of cells such as endothelial cells and fibroblasts, and is associated with diseases including allergy, infection, and cancer (10, 11). However, the role of IL33 in cancer is still controversial, albeit IL33 upregulation in many types of cancers (12). To address this issue, we established IL33 transfectants using B16-F10 cells (F10-TR; Fig. 2) and human breast cancer MCF7 cells (MCF7-TR; Supplementary Fig. S3). IL33 overexpression generated large cells with low proliferative property (P < 0.001 vs. F10-mock) and high 5-Fu–resistant property in vitro (Fig. 2A) and in vivo (Fig. 2B). The 5-Fu treatment remarkably promoted F10-TR tumor growth and metastasis, and significantly shortened mouse survival (P = 0.0001 vs. F10-mock; Fig. 2B). These suggest IL33 plays a key role in the treatment-induced tumor progression, albeit likely a better prognostic factor under tranquility. In the F10-TR–implanted mice, ST2⁺ cells, particularly ILC2s, were significantly expanded not only in the tumor tissues (P = 0.003 vs. PBS treatment) but also in the spleen (P < 0.0001) after the treatment (Fig. 2C; Supplementary Fig. S4A). ST2⁺ increase was also observed within F10-mock tumor tissues regardless of treatment. This might be potentially mediated by chemokines such as CXCL12 for CXCR4 that is commonly expressed in mast cells, MDSCs and ILC2s (13). CTLs were impaired in the F10-TR–implanted mice, and the CTL activities were further reduced by the treatment (IFNγ production, P = 0.002 vs. PBS treatment; cytotoxicity, P = 0.001; Fig. 2D). These suggest that IL33 induction confers a progressive property to tumor cells via not only biological self-transformation, but also immunologic damage possibly mediated by ST2⁺ cells.

We next examined whether IL33 could be associated with polyploidy because nuclear IL33 is known but its functional roles remain unclear (14), and IL33 overexpression transformed tumor cells into a giant size (Fig. 2A) that is a representative feature of polyploidy, by which giant cells are generated through misregulation of canonical G1–S–G2–M cell cycle without cell division followed by unrestrained propagation (15). F10-TR cells contained significantly more ≧ 4N DNA contents (P < 0.001 vs. F10-mock; Fig. 3A), and generated smaller progeny cells by stimulation with a low dose of 5-Fu albeit cell death in F10-mock cells (Fig. 3B). Aurka (16) and Aurkb (17) regulate polyploidization followed by inactivation of p53 (18). Indeed, Aurka/b increased and phosphorylated p53-Ser15 (p53-p) decreased in the F10-TR cells (P < 0.002 vs. F10-mock), and IL33 knockdown reduced Aurka/b expression but increased p53-p expression in the F10-TR cells (Fig. 3C). These suggest that IL33 expression, potentially in the nucleus, plays a key role in polyploidization followed by rapid proliferation in response to treatment stress.

ST2⁺ cells remarkably increased in the F10-TR–implanted mice particularly after treatment. We further pursued the observation because the ST2⁺ cells would likely play important roles in the IL33⁺ tumor progression mechanisms. Because ST2 is expressed not only in ILC2s (10, 11), mast cells, and MDSCs (19) but also in NK (20) and Tregs (21), we conducted antibody-mediated ablation experiments to determine the key effector cells using immunodeficient nu/nu mice treated with anti-asialo GM1, indicating absence of functional T cells and NK/NKT cells. 5-Fu–induced F10-TR tumor progression was partly retarded in the absence of FceRIa⁺ cells and CD11b⁺ cells, but not NK/NKT cells and T cells (Supplementary Fig. S4B). In coinjection experiments, however, greater impact on tumor progression was provided by IL17RB⁺ cells derived from F10-TR–implanted mice (designated TR-IL17RB cells) compared with FceRIa⁺ cells and CD11b⁺ cells (P < 0.001; Supplementary Fig. S4C). These suggest ST2⁺ cells, especially ILC2s, could play important roles in the IL33⁺ tumor progression. The TR-IL17RB cells highly produced IL13, suggesting these cells could be ILC2s (Fig. 4A). Stimulation with a low dose of IL13 that did not affect F10-mock cells remarkably enhanced F10-TR invasion (Fig. 4B). Anti-IL13 mAb injection into the TR-IL17RB–coinjected tumors significantly suppressed the growth (P = 0.010; Fig. 4C). These suggest IL13 is involved in the ILC2-induced tumor progressive mechanisms.

In the in vitro CTL induction, addition of the TR-IL17RB supernatant generated exhausted and dysfunctional CTLs expressing PD1, Tim3, and Tigit (Fig. 4D). However, anti-IL13 mAb addition to the culture-generated potent CTLs, suggesting ILC2-induced IL13 could impair CTLs (P < 0.004; Fig. 4D). Because IL13R is not generally expressed on T cells except some cases (22), the CTL dysfunction observed might be indirectly induced by other IL13R⁺ cells such as M2 macrophages and MDSCs contained in the antigen-presenting cells used. Anti-IL13 mAb injection significantly suppressed 5-Fu–induced F10-TR tumor progression and metastasis (P < 0.02 vs. 5-Fu only; Fig. 4E). These suggest that ILC2s contribute to IL33⁺ tumor progression via immune exhaustion and dysfunction partly mediated by IL13.

Unexpected hyperprogression is a serious problem in cancer therapy with ICIs (6). In the mouse models implanted with F10-BM tumors, which is similar to the F10-TR tumors (Supplementary Fig. S5), anti-PD1 treatment significantly promoted tumor growth, and shortened mouse survival (P < 0.0001 vs. control), albeit significantly effective in the F10-primary models (P = 0.0017; Fig. 5A). In the anti-PD1–treated mice, cell infiltration was hardly seen in the F10-BM tumors (Fig. 5B and C), and splenic CTLs were still dysfunctional (Fig. 5D and E). In contrast, anti-IL33 treatment did not cause tumor progression, and the mouse survival was slightly but significantly prolonged (P = 0.004 vs. control; Fig. 5A). In the anti-IL33–treated mice, tumor-specific CD8⁺ T cells increased within tumors, and splenic CTL activities were significantly elevated (Fig. 5). Combination of anti-PD1 treatment synergistically enhanced the anti-IL33 efficacy on tumor growth (P = 0.0003 vs. anti-IL33 monotherapy) and mouse survival (P = 0.0036). The combinatory regimen was also significantly effective in the F10-primary models (P = 0.047 vs. anti-PD1 monotherapy). These results provide a potential of blocking IL33 for successfully eliciting antitumor immunity in combination with other therapeutics in the treatment of IL33⁺ tumors.

We next validated the findings using clinical samples. We firstly conducted IHC analysis using tumor tissues obtained from patients with breast cancer. IL33 expression was upregulated in primary and metastatic tumor tissues compared with noncancerous portions (P < 0.05; Fig. 6A), and the IL33 intensity was correlated with IL17RB⁺ cell infiltration particularly in the stage III patients with tumor metastasis (P = 0.0005) rather than the stage II patients (P = 0.0886). IL17RB⁺ cells were hardly seen in the tumors if IL33 was expressed only in the nuclei. These suggest a potential causal relationship between IL33 positivity in tumors and increase in IL17RB⁺ cells within the tumor microenvironment.

We further analyzed PBMCs for ST2⁺ cells and IL17RB⁺ cells by flow cytometry, because the cells could potentially play critical roles in systemic immunity as well as local immunity. We used PBMCs obtained from patients with metastatic colorectal cancer because of only one available for us. It has been demonstrated that IL33 upregulation is associated with colorectal cancer pathogenesis (23), and hyperprogression is seen in patients with colorectal cancer after anti-PD1/PDL1 treatment albeit a few in number (24). ST2⁺ cells (P = 0.008) and IL17RB⁺ cells (P = 0.021) were significantly increased in patients with metastatic colorectal cancer compared with healthy donors, and the increase of these cells, particularly ST2⁺ cells (P = 0.005), was correlated with decrease in CD3⁺CD8⁺ T cells (Fig. 6B). These suggest a potential causal relationship between ST2/IL17RB⁺ expansion and CD8⁺ T-cell reduction in systemic immunity. Taken together, targeting IL33 and the consequent ST2/IL17RB⁺ cells may be a promising strategy of diagnosis and treatment of patients who are likely resistant to treatments in the clinical settings.

We identified IL33 as a determinant of treatment resistance of cancer. IL33 induction in tumor cells transforms into polyploid giant cells showing abnormal cell cycle without cell division accompanied by Snail deregulation and p53 inactivation, and its small progeny cells are generated in response to treatment stress. Simultaneously, soluble IL33 is released from the tumor cells, and expands ST2⁺ cells including IL17RB⁺GATA3⁺ ILC2s that promote tumor progression and metastasis directly and indirectly through induction of immune exhaustion and dysfunction partly mediated by IL13. In the mouse IL33⁺ metastatic tumor models, however, blocking IL33 with the specific mAb abrogates the negative consequences, and successfully elicits antitumor efficacy induced by other treatment combined. In clinical samples, IL33 positivity in tumors is significantly correlated with increase of the IL17RB⁺ cells within the tumor milieu, and increase of ST2⁺ cells and IL17RB⁺ cells is correlated with reduction of CD8⁺ T cells in PBMCs. This suggests clinical relevancy of the basic findings. Thus, this study reveals the functional role of IL33 in cancer polyploidy that could be intrinsic tumor biological and extrinsic immunological mechanisms underlying treatment failure.

Although many studies demonstrated the roles of IL33 in cancer, it is still controversial (12). For example, treatments with mAbs specific for IL33 (25), ST2 (26), and IL1RAP (27) significantly suppressed tumor progression through reduction of tumor-associated macrophages and Tregs in mouse tumor models. On the other hand, IL33 induction in melanoma suppressed tumor progression (28). We found that IL33 expression in tumor cells retards proliferation and growth, whereas aberrant mitotic progression is caused by treatment stress. In contrast, stable silencing of IL33 in tumor cells restricts cell proliferation, resulting in poor engraftment and growth of the tumors in mice probably because of damaging cells not only internally via reduction of Aurka/b that is required for mitotic control, but also externally via loss of ST2⁺ cell support. Therefore, the tumor suppressive activity reported might be explained by the low proliferative property due to polyploidization. The authors did not conduct therapeutic experiments, in which tumor progression might be seen as shown in our study. IL33 overexpression simultaneously induced ST2 and IL1RAP expressions (Fig. 2A; Supplementary Fig. S3), and IL1RAP knockdown enhanced chemosensitivity of IL33⁺ tumor cells (Supplementary Fig. S2). The receptor coexpression may intensify the treatment resistance of IL33⁺ tumor cells in an autocrine manner.

Cancer polyploidy has been demonstrated as cancer stemness because of producing unrestrained propagation that undermines genomic stability (15, 29). Polyploid giant cancer cells are pathologically observed in clinical tumor tissues, and the incidence is associated with chemoresistance (30) and poor prognosis of patients with cancer (31). However, there is no practical management of the polyploidization and the rapid proliferation. Our findings that IL33 plays a critical role in cancer polyploidy provided a possible biomarker and a potentially druggable target in cancer therapy. Combination with polyploidy-associated Aurka/b inhibitors (32) may potentially enhance the anti-IL33 therapeutic efficacy.

This study provides new insights into the IL33 roles in recurrence and treatment resistance of cancer. Targeting IL33-driven polyploidy could, at least in part, overcome treatment failure in clinical settings.

No potential conflicts of interest were disclosed.

Conception and design: C. Kudo-Saito

Development of methodology: C. Kudo-Saito

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Kudo-Saito, T. Miyamoto, H. Imazeki, N. Boku

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Kudo-Saito, T. Miyamoto

Writing, review, and/or revision of the manuscript: C. Kudo-Saito

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Imazeki, H. Shoji, K. Aoki

Study supervision: C. Kudo-Saito, N. Boku

This work was financially supported by Grants-in-Aid for Scientific Research KAKENHI (21590445 and 26430122 to C. Kudo-Saito) and Japan Agency for Medical Research and Development AMED-P-CREATE (106209 to C. Kudo-Saito).

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.