IL-33/regulatory T cell axis triggers the development of a tumor-promoting immune environment in chronic inflammation

Significance

Chronic inflammatory diseases are well-recognized causes of cancer and account for up to 20% of all cancer deaths worldwide. However, the mechanism that initiates the development of a tumor-promoting immune environment in chronic inflammation is not known. Using mouse models of chronic skin and colon inflammation and human samples, we show IL-33 triggers the transition from tumor-suppressive immunity to chronic, tumor-promoting inflammation through a regulatory T cell-dependent mechanism. Our findings demonstrate a generalized dependency of tumor-promoting immune environments on the IL-33/Treg axis both in the skin and colon. Based on these findings, IL-33/Treg axis blockade may be an attractive therapeutic strategy for the treatment and prevention of cancers associated with chronic inflammatory diseases and potentiating the antitumor immunity induced by cancer immunotherapeutics.

Abstract

Chronic inflammation’s tumor-promoting potential is well-recognized; however, the mechanism underlying the development of this immune environment is unknown. Studying the transition from acute, tumor-suppressive to chronic, tumor-promoting allergic contact dermatitis (ACD) revealed how tumor-promoting chronic inflammation develops. Epidermis-derived interleukin (IL)-33 up-regulation and its induction of regulatory T cell (Treg) accumulation in the skin preceded the transition from acute to chronic ACD and triggered the tumor-promoting immune environment in chronic ACD. Mice lacking IL-33 were protected from chronic ACD and its skin cancer sequela compared with wild-type controls (P = 0.0002). IL-33’s direct signaling onto Tregs was required for the development of the tumor-promoting immune environment in the skin. IL-33–Treg signaling was also required for chronic colitis and its associated colorectal cancer development in a colitis model (P < 0.0001). Significantly increased IL-33 and Tregs marked the perilesional skin and colon in patients with cancer-prone chronic inflammatory diseases. Our findings elucidate the role of the IL-33/Treg axis in creating a tumor-promoting immune environment in chronic inflammatory diseases and suggest therapeutic targets for cancer prevention and treatment in high-risk patients.

The association between chronic inflammation and cancer risk has been appreciated for over a century (1). Chronic inflammatory conditions contribute to 15–20% of all cancer-related deaths worldwide (2). Besides chronic inflammation-induced cancers, tumor-promoting immune environments are an essential component of a wide array of sporadic cancers (2). In the skin, chronic inflammation is associated with an aggressive form of squamous cell carcinoma (SCC), Marjolin’s ulcer (3). Previously, we described the development of Marjolin’s ulcer in chronic allergic contact dermatitis (ACD) secondary to an allergenic orthopedic metal implant (4). The extent of chronic inflammation-associated cancers highlights the need to elucidate the pathogenesis of cancer-prone, chronic inflammatory conditions.

Cancer-prone, chronic inflammation is commonly described as a type 2 immune environment containing multiple protumorigenic immune cells and factors, including T helper 2 (Th2) cells, regulatory T cells (Tregs), M2 macrophages, mast cells, eosinophils, myeloid-derived suppressor cells, interleukin (IL)-6, tumor necrosis factor (TNF)-α, transforming growth factor (TGF)-β, chemokines, and other growth factors (5). Although targeting these tumor-promoting factors in chronic inflammation may reduce the rate of cancer development, the complete reversal of tumor promotion by blocking all these effector pathways is impractical (5). Therefore, it is necessary to explore the upstream factors creating a tumor-promoting immune environment in the first place to identify targets for cancer prevention in chronic inflammation.

Chronic ACD is marked by type 2, tumor-promoting immune responses (4). In stark contrast, acute ACD (i.e., contact hypersensitivity) is well-known to be tumor-suppressive and is characterized by type 1 immunity involving Th1, cytotoxic T, and natural killer cells (6). The opposite immune environments of chronic versus acute ACD provide a unique opportunity to determine the mechanism of transition from a type 1, tumor-suppressing immunity to a type 2, tumor-promoting immune environment. Th2 cells, Tregs, group 2 innate lymphoid cells (ILC2s), and mast cells are potential candidates for initiating the development of chronic inflammation (2). The epithelium-derived alarmins, IL-25, IL-33, and thymic stromal lymphopoietin (TSLP), may contribute to the formation of the type 2 immune environment of chronic inflammation (7). These alarmins initiate type 2 allergic inflammation at barrier sites (7). They can activate Tregs, ILC2s, and mast cells upon epithelial damage in barrier organs (8, 9). It is unknown if alarmins play a determining role in the transition from acute to chronic inflammation.

Herein, we demonstrate IL-33 up-regulation precedes the development of a tumor-promoting immune environment in chronic inflammation. We determined that IL-33/Treg axis was required for the transition to a type 2 immune environment and tumor development in chronic ACD and colitis-induced colorectal cancer. Finally, we demonstrated increases in IL-33 and Tregs are associated with human cancer development in chronic inflammatory diseases. We conclude the IL-33/Treg axis is essential for the initiation of a tumor-promoting immune environment in chronic inflammation.

Results

IL-33 Overexpression Precedes the Development of a Tumor-Promoting Immune Environment in Chronic ACD.

To study the transition from an acute, tumor-suppressing immunity to a chronic, tumor-promoting inflammation in the skin, we searched for the timepoint at which the switch from acute to chronic ACD occurs. One week after abdominal sensitization with 0.5% 1-fluoro-2,4-dinitrobenzene (DNFB), mice were challenged with 0.25% DNFB on their ear three times per week, and ear thickness was measured over time. Increased ear thickness, marking acute inflammation and dermal edema during the acute phase of ACD [i.e., contact hypersensitivity (6)], peaked at day 14–18 after first challenge in wild-type (Wt) mice, followed by a decrease, reaching a plateau at approximately day 22 (SI Appendix, Fig. S1A). Therefore, we identified day 22 as the transition timepoint from acute to chronic ACD (SI Appendix, Fig. S1A, arrow). Given the role for the cardinal epithelium-derived alarmins, TSLP, IL-25, and IL-33, in the initiation of a type 2 immune response at barrier sites (7), we examined the expression levels of TSLP, IL-25, and IL-33 in the back skin of Wt animals at day 22 post DNFB versus acetone (carrier alone) challenge (SI Appendix, Fig. S1B). Tslp expression was down-regulated (P < 0.05), and Il25 expression was not changed in DNFB- compared with acetone-treated skin at the transition timepoint (SI Appendix, Fig. S1C). In contrast, Il33 expression was significantly elevated in DNFB-treated skin at the transition timepoint (SI Appendix, Fig. S1C) and, to a lesser degree, at 24 h after the first challenge with DNFB (acute phase; Il33 expression at transition to acute ratio: 2.32; SI Appendix, Fig. S1D). IL-33 protein levels were also significantly elevated in DNFB-treated skin during the transition to chronic ACD (SI Appendix, Fig. S1E). IL-33 was detectable in keratinocytes at the transition timepoint, but absent at 24 h post first DNFB challenge or in acetone-treated skin (Fig. 1A and SI Appendix, Figs. S1F and S2). IL-33 up-regulation was also detectable in the skin of MRL/lpr mice that develop cancer-prone cutaneous lupus inflammation [SI Appendix, Fig. S3 (10)]. IL-33 up-regulation at the transition timepoint and its initiating role in the development of type 2 inflammation (7) strongly suggest its role in driving the transition from acute to chronic inflammation.

Fig. 1.

IL-33 is required for tumor promotion in chronic ACD. (A) Representative images of IL-33 immunostaining in the Wt skin at the transition timepoint relative to acute timepoint and acetone control. (B) Epidermal thickness measured across 10 random high power fields (hpf) per each back skin of IL-33KO and Wt mice treated with DNFB compared with acetone control at the transition timepoint. (C) Representative H&E images of back skin of IL-33KO, Wt, and acetone-treated mice at the transition timepoint. Note the epidermal thickness (brackets) and dermal inflammatory infiltrates in each group. (D) Experimental design for chronic ACD-induced skin carcinogenesis consisting of initial sensitization with DNFB allergen, treatment with DMBA carcinogen, and challenge with DNFB three times a week for 29 wk. (E) Epidermal thickness measured across 10 random hpf per back skin of each IL-33KO and Wt mice treated with DNFB compared with acetone control at the end of 30-wk carcinogenesis protocol. (F) Skin tumor onset in IL-33KO, Wt, and acetone-treated mice (log-rank test). (G) Final tumor counts per IL-33KO mice compared with Wt and acetone-treated controls. n = 10 per group; error bars represent the mean ± SD; *P < 0.005, **P < 0.0001 by two-tailed Mann–Whitney U test; epidermal thickness measurements were performed blindly. (Scale bars, 100 μm.)

IL-33 Is Required for the Development of a Tumor-Promoting Immune Environment in ACD.

We investigated whether IL-33 is necessary for the transition to a tumor-promoting immune environment in chronic ACD. IL-33 deletion had no impact on baseline T cell infiltrates in the skin (SI Appendix, Fig. S4 A and B) or the acute phase of ACD as measured by ear thickness in DNFB-treated IL-33 knockout mice (IL-33^tm1b/tm1b or IL-33KO) compared with Wt mice (SI Appendix, Fig. S5). However, epidermal hyperplasia and dermal inflammation were significantly blunted in DNFB-treated IL-33KO compared with Wt mice during the transition to chronic ACD (Fig. 1 B and C). To determine the role of IL-33 in regulating tumor promotion in chronic ACD, we sensitized IL-33KO and Wt mice to DNFB, treated their back skin with the carcinogen, 7,12-dimethylbenz(a)anthracene (DMBA), followed by treatment with DNFB or acetone three times per week on their back skin for 30 wk to measure their tumorigenesis potential (Fig. 1D). IL-33 loss blunted the epidermal hyperplasia in chronic ACD (P = 0.0021; Fig. 1E and SI Appendix, Fig. S6). DNFB-treated IL-33KO mice were protected from chronic ACD-induced skin tumors compared with DNFB-treated Wt mice (P = 0.0002, Fig. 1F and SI Appendix, Fig. S6). Further, IL-33KO mice developed significantly fewer tumors compared with Wt animals in response to chronic DNFB exposure (Fig. 1G). IL-33 loss did not impact tumor development in mice subjected to the standard DMBA/TPA skin carcinogenesis protocol (SI Appendix, Fig. S7), suggesting that IL-33’s protumorigenic effect is specific to inflammation-associated, but not sporadic forms of skin cancer. These findings highlight distinct pathways of skin carcinogenesis and IL-33’s essential role in the formation of a tumor-promoting chronic inflammatory state.

To determine the cellular target(s) of IL-33, we examined the immune environment of IL-33KO relative to that of Wt skin at day 22 post DNFB challenge (i.e., transition timepoint, SI Appendix, Fig. S1B). CD4⁺ and CD8⁺ T cells were significantly reduced in DNFB-treated IL-33KO compared with Wt skin, with CD4⁺ T cell numbers in IL-33KO skin approximating those in acetone-treated controls (SI Appendix, Fig. S8 A and B). We did not detect any differences between DNFB-treated IL-33KO and Wt mice in mast cell, basophil, or ILC2 accumulation in the skin, which are the other direct targets of IL-33 and can drive the development of a type 2 immune environment (SI Appendix, Figs. S8 C–E and S9). Among CD4⁺ T cell subsets, Foxp3⁺ Tregs showed a marked reduction in DNFB-treated IL-33KO compared with Wt skin at the transition timepoint (P = 0.0068; SI Appendix, Fig. S8 F and G). To determine the impact of IL-33 loss on the transition from type 1 immunity to type 2 inflammation in chronic ACD, we analyzed the expression of cytokines associated with Th cells (Ifng, Il4, and Il17) and Tregs (Il10) in acute (D1) versus transition (D22) timepoints in IL-33KO and Wt skin-draining lymph nodes. IFN gamma expression, a marker of type 1, acute ACD, was significantly decreased in Wt mice at transition relative to the acute timepoint; however, IL-33KO mice showed a reversed trend (SI Appendix, Fig. S8H). There was a significant reduction in percent T-bet⁺ Th1 cells in Wt skin at the transition timepoint, which was reversed in IL-33KO skin (SI Appendix, Figs. S8F and S10A). Il4 and Il17 expression levels did not change significantly from acute to transition phase (SI Appendix, Fig. S8H). GATA3⁺ Th2 cells were not increased in Wt or IL-33KO skin at the transition timepoint, suggesting Th2 cells may not play a role in the initiation of a type 2 inflammatory state in chronic ACD (SI Appendix, Fig. S10B). Il10 expression was significantly elevated at the transition timepoint in Wt mice and blunted in IL-33KO mice (SI Appendix, Fig. S8H). We determined that a significantly higher percentage of ST2⁺ Tregs express IL-10 compared with ST2⁻ Tregs in the skin during transition from acute to chronic ACD (P < 0.0001, SI Appendix, Fig. S11A). Accordingly, there were significantly fewer IL-10⁺ Tregs, ST2⁺ Tregs, and IL-10⁺ ST2⁺ Tregs in IL-33KO skin compared with Wt controls at the transition timepoint (SI Appendix, Fig. S11 B and C). In contrast, IFNγ⁺ CD4⁺ T cells and IFNγ⁺ CD8⁺ T cells were markedly increased in the skin of IL-33KO mice compared with Wt controls at the transition timepoint (SI Appendix, Fig. S11 B and D). These findings suggest an essential role for Treg and its associated cytokine, IL-10, downstream of IL-33 in mediating the transition from acute to chronic ACD.

Loss of ST2 Expression on Tregs Blocks Chronic ACD-Induced Skin Carcinogenesis.

Having shown IL-33’s role in the transition to tumor-promoting chronic inflammation in association with Treg accumulation in the skin, we investigated whether the IL-33/Treg axis is necessary for the transition to a chronic tumor-promoting immune environment. We utilized Foxp3^Cre, ST2^flox/flox (TregST2CKO) mouse model in which IL-33 receptor [interleukin 1 receptor like 1 (IL1RL1) or ST2] was deleted specifically on Tregs. As observed in the IL-33KO mice, knocking out ST2 specifically on Tregs had no impact on baseline T cell infiltrates in the skin (SI Appendix, Fig. S4 A and B) or acute ACD inflammation (SI Appendix, Fig. S12A). However, ST2 deletion on Tregs blunted the epidermal hyperplasia in response to repeated DNFB challenge at the transition timepoint (SI Appendix, Fig. S12B). DNFB-treated TregST2CKO skin exhibited decreased CD4⁺ T cell accumulation, decreased Treg accumulation, and increased CD8/Treg ratio compared with DNFB-treated Wt skin at the transition timepoint (SI Appendix, Fig. S12 C–G). To test whether IL-33 signaling onto Tregs was required for the transition to a tumor-promoting immune environment in chronic ACD, we subjected TregST2CKO, IL-33KO, and Wt mice to a chronic ACD-associated skin carcinogenesis protocol (Fig. 1D). Compared with DNFB-treated Wt mice, TregST2CKO and IL-33KO mice had significantly reduced CD4⁺ T cell and Treg accumulation in the skin during chronic ACD (Fig. 2 A–C). TregST2CKO maintained a high number of CD8⁺ T cells in their skin leading to significantly elevated CD8/Treg ratio, a marker of antitumor immunity, in these animals compared with DNFB-treated Wt mice (Fig. 2D and SI Appendix, Fig. S13). Accordingly, epidermal hyperplasia was markedly blunted in DNFB-treated TregST2CKO mice compared with Wt controls (Fig. 2E and SI Appendix, Fig. S14). Significantly fewer TregST2CKO mice developed skin tumors in response to DMBA/DNFB treatment, and the number of tumors per mouse was markedly less among TregST2CKO mice compared with DMBA/DNFB-treated Wt mice (Fig. 2 F and G). Our findings demonstrate that blocking IL-33 signaling to Tregs recapitulates the tumor-protective effect of global IL-33 loss in chronic inflammation.

Fig. 2.

ST2 deletion in Tregs blocks chronic ACD-induced skin tumorigenesis. (A) Representative images of CD4⁺ T cells in DNFB-treated TregST2CKO, IL-33KO, Wt, and acetone-treated Wt skin at the conclusion of 30-wk DMBA/DNFB carcinogenesis protocol (CK: cytokeratin). (B and C) CD4⁺ T cells (B) and Tregs (C) quantified in 10 random hpf per skin of DMBA/DNFB-treated TregST2CKO and IL-33KO compared with Wt mice. (D) CD8/Treg ratio in DMBA/DNFB-treated TregST2CKO and IL-33KO skin compared with Wt skin. Cell counts were computed using immunofluorescence image-based quantification and flow cytometry. (E) Epidermal thickness measured across 10 random hpf per back skin of each DMBA/DNFB-treated TregST2CKO mice compared with IL-33KO, Wt and DMBA/acetone controls at the end of 30-wk carcinogenesis protocol. (F) Skin tumor onset in DMBA/DNFB-treated TregST2CKO, IL-33KO, Wt, and DMBA/acetone-treated groups (log-rank test). (G) Tumor counts per mouse in TregST2CKO mice compared with IL-33KO, Wt, and acetone-treated groups. n = 10 TregST2CKO, 13 Wt (including Foxp3^Cre, ST2^+/flox), 4 IL-33KO, and 4 acetone control mice; error bars represent the mean ± SD; *P < 0.05, **P < 0.01; ns, not significant by two-tailed Mann–Whitney U test; cell counts and epidermal thickness measurements were preformed blindly. (Scale bar, 100 μm.)

IL-33 and Tregs Are Increased in Cancer-Prone Chronic Inflammatory Diseases of Human Skin.

We next investigated whether the IL-33/Treg axis was associated with skin cancer development in a clinical case of chronic ACD-associated skin cancer (4). IL-33 immunostaining of the patient’s skin cancer and perilesional skin, distant from the cancer site, revealed IL-33 overexpression in the keratinocytes compared with normal gender- and age-matched control skin from the same anatomical site (Fig. 3A). We also identified increased numbers of CD3⁺ Foxp3⁺ Tregs in the SCC and the perilesional skin of our patient compared with the normal skin (Fig. 3B). Importantly, we observed a similar pattern of IL-33 up-regulation in SCC and perilesional skin of two skin cancer-prone discoid lupus patients, which was absent in control discoid lupus lesions with no skin cancer history and normal skin [SI Appendix, Fig. S15 (10)]. Therefore, the IL-33/Treg axis may play an important role in skin cancer development in cancer-prone inflammatory skin diseases.

Fig. 3.

IL-33 overexpression and Treg accumulation are associated with cancer-prone chronic ACD in human skin. (A) Representative images of IL-33 immunostained squamous cell carcinoma (SCC) and perilesional skin of a patient with chronic ACD due to allergenic metal implant compared with age- and gender-matched normal skin (brown nuclear stains). (B) Representative images of Foxp3 (brown) and CD3 (red) highlighting Treg infiltrates in skin cancer and perilesional skin relative to normal skin. Arrows point to Foxp3⁺ Tregs. (Scale bars, 100 μm.)

ST2 Expression by Tregs Is Critical for Colitis-Induced Colorectal Carcinogenesis in Mice.

We extended our studies to chronic inflammatory bowel disease (IBD)-associated colorectal cancer. At baseline, TregST2CKO mice had more CD4⁺ T cells in their colon compared with IL-33KO and Wt mice (SI Appendix, Fig. S4 A and C). However, there were no differences in Treg and CD8⁺ T cell numbers in the colon between the three groups (SI Appendix, Fig. S4 A and C). An AOM/DSS model was used as an established model for colitis-induced colorectal cancer in which mice received four treatment cycles of a carcinogen, azoxymethane (AOM), injected intraperitoneally followed by a colitis-causing agent, dextran sodium sulfate (DSS), added to the drinking water over 5 d (Fig. 4A). This carcinogenesis protocol led to a massive IL-33 induction in Wt colon (SI Appendix, Fig. S16). IL-33KO mice did not experience weight loss and developed significantly fewer colorectal tumors compared with Wt mice (SI Appendix, Fig. S17). TregST2CKO mice subjected to this protocol were also resistant to weight loss, a marker of disease severity, compared with Wt controls (Fig. 4B). Likewise, colon length was significantly greater in the TregST2CKO mice compared with Wt mice (Fig. 4 C and D). TregST2CKO mice developed significantly fewer colorectal tumors compared with Wt controls (P < 0.0001; Fig. 4 E and F). Tumors were smaller in TregST2CKO compared with Wt colon, and we observed better preserved crypt architecture in TregST2CKO relative to Wt colon (Fig. 4 F and G). Flow cytometry demonstrated increased CD8⁺ T cell frequency in TregST2CKO relative to Wt colon and increased PD-1⁺ CD8⁺ T cells in the mesenteric lymph nodes of TregST2CKO compared with Wt mice (Fig. 4H and SI Appendix, Fig. S18). We identified a significant increase in CD8⁺ T cell counts in TregST2CKO relative to Wt colon, while CD4⁺ T cell counts remained the same (Fig. 4 I and J). Among CD4⁺ T cells, Treg percentage was significantly reduced in TregST2CKO compared with Wt colon (P < 0.0001; Fig. 4K). Finally, the CD8/Treg ratio in TregST2CKO colon was significantly higher compared with Wt controls (SI Appendix, Fig. S19). These findings support the generalized role of the IL-33/Treg axis as a driver of the tumor-promoting immune environment in chronic inflammation.

Fig. 4.

ST2 loss on Tregs suppresses colitis-induced colorectal cancer in mice. (A) Schematic diagram for colitis-induced colorectal tumorigenesis in mice, consisting of four cycles of AOM/DSS administration over 15 wk. (B) Weight loss pattern in TregST2CKO and Wt mice during the final DSS treatment cycle. (C) Colon length measurement, from cecum to rectum, in TregST2CKO and Wt mice at the completion of AOM/DSS carcinogenesis protocol. (D) Representative images of TregST2CKO and Wt colon at the completion of AOM/DSS carcinogenesis protocol. (E) Colorectal tumors quantified per TregST2CKO and Wt mice at the completion of AOM/DSS carcinogenesis protocol. (F) Representative macroscopic images of exposed TregST2CKO and Wt colonic lumen. (G) Representative H&E image of distal colon of TregST2CKO and Wt mice. (Scale bar, 1 cm.) (H) Flow cytometric analysis of T cells isolated from colon showing CD8⁺ T and CD4⁺ T cell frequencies in TregST2CKO and Wt colon. Percent cells in each gate are shown on the flow dot plots. (I) CD4⁺ and CD8⁺ T cell infiltrates in TregST2CKO and Wt colons quantified in 10 random hpf per colon section and averaged across the animals in each group. CD8⁺ T cell abundance was determined using CD3-stained images and flow cytometry data (SI Appendix, Fig. S20). Stained cells were counted blindly. (J) Representative images of CD4⁺ T cell in TregST2CKO and Wt colons. (CK, cytokeratin; Scale bars, 100 μm.) (K) Treg frequencies of total CD4⁺ T cells in TregST2CKO and Wt colon. Treg frequency was determined by flow cytometry. n = 16 TregST2CKO and 14 Wt (including Foxp3^Cre, ST2^+/flox) mice; error bars represent the mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.0001; ns, not significant by two-tailed Mann–Whitney U test.

IL-33 and Treg Are Increased in Cancer-Prone IBD in Humans.

We next investigated whether the IL-33/Treg axis plays a role in human colons affected by cancer-prone IBD preceding the development of colorectal cancer. We collected 19 colorectal cancer specimens (12 sporadic and 7 IBD-associated), 13 colitis specimens (7 ulcerative colitis and 6 Crohn’s disease) from patients with concurrent diagnosis of colorectal cancers, and 20 normal colon specimens. We stained for IL-33 and CD4/Foxp3 using adjacent sections and quantified IL-33⁺ and Foxp3⁺ CD4⁺ T cells within 10 randomly selected high-power fields from each specimen in a blinded manner. We observed a significant increase in IL-33⁺ cells in colorectal cancers and colon tissues with colitis, away from any cancer, compared with the normal colon (P < 0.0005; Fig. 5A). In addition, the number of Tregs in colitis tissues was higher than normal colon (P < 0.001) and comparable to Treg numbers in colon cancers (Fig. 5B). Correlation analysis between IL-33⁺ cell counts and Treg counts across all specimens revealed a positive correlation (r = 0.321) between IL-33⁺ cell number and Treg accumulation in the colon (P = 0.0202; Fig. 5C). These outcomes demonstrate IL-33 up-regulation and its associated Treg accumulation precede the development of colorectal cancer in cancer-prone IBD.

Fig. 5.

IL-33 induction and Treg accumulation in the colon mark high-risk human IBD. (A) Representative IL-33 immunostained (brown) tissue sections from human colorectal cancer, colitis, and normal colon and quantification of IL-33⁺ cells in 10 random hpf per specimen (***P < 0.0005; ns, not significant by two-tailed Mann–Whitney U test). (B) Representative Foxp3 (brown) and CD4 (red) immunostained tissue sections from human colorectal cancer, colitis, and normal colon and quantification of Foxp3⁺ Tregs in 10 random hpf per specimen. (Insets, Right) show higher magnification of the boxed areas of the images. Arrowheads point to CD4⁺ T cells with positive Foxp3 nuclear stain (i.e., Treg; ***P < 0.001; ns, not significant by two-tailed Mann–Whitney U test). (C) Correlation between IL-33⁺ cell and Treg count in colon cancer, colitis, and normal colon (Student’s t test for the Pearson correlation coefficient). n = 19 colorectal cancer specimens (12 sporadic and 7 IBD-associated), 13 cancer-prone colitis specimens (7 ulcerative colitis and 6 Crohn’s disease), and 20 normal colons; cells were counted blindly; error bars represent the mean ± SD. (Scale bars, 100 μm.)

Discussion

Our clinical and experimental findings demonstrate that IL-33–driven stimulation of Tregs is an essential trigger for the development of a tumor-promoting immune environment in chronic inflammation. Genetic studies demonstrated that IL-33 leads to the expansion of IL-10⁺ Tregs in the skin during the transition from acute to chronic inflammation, which can suppress type 1 immunity and promote the development of a tumor-promoting type 2 inflammation in the skin (11, 12). Deletion of ST2 specifically on Tregs blocked tumor development in cancer-prone, chronic inflammation of the skin and colon. IL-33–expressing cells and Tregs were significantly and specifically increased in cancer-prone, chronic inflammatory diseases of skin and colon in humans. Together, our findings have major implications for cancer prevention in chronic inflammatory diseases, cancer treatment in a large array of malignancies in which IL-33 and Tregs are increased, and enhancement of antitumor immunity and the efficacy of immunotherapeutics by blocking the detrimental effects of the IL-33/Treg axis.

Study of ACD and its distinct acute, chronic, and transition phases enabled us to identify the IL-33/Treg axis as a fundamental node in regulating the development of a complex tumor-promoting immune environment in the skin and colon. While carcinogenesis in chronic ACD is rare, the rising prevalence of contact allergies to nickel (13) and nickel-containing implants suggests that chronic ACD-induced carcinogenesis is likely underreported (4). As we showed in discoid lupus, the IL-33/Treg axis may also play a critical role in skin cancer development in other cancer-prone inflammatory diseases of the skin including hypertrophic lichen planus, chronic wound, and chronic scar (3, 14–16). In colon, IBD is a major risk factor of colorectal cancer with 23-fold increased risk among patients with pancolitis (17). In addition, IL-33 up-regulation has been detected in established invasive cancers, which may suggest a role for IL-33/Treg axis in tumor progression in response to damaged tissue homeostasis (9, 18–20). These observations suggest IL-33 blocking antibodies developed for atopic disorders may show efficacy in cancer treatment. However, it is critical to emphasize the complexity of IL-33 function and its potentially distinct effects on carcinogenesis. While our data demonstrate that IL-33/Treg axis has protumorigenic effects, IL-33 signaling onto other immune cell types, such as CD8⁺ T cells, can be tumor-protective (21). Importantly, while the IL-33/Treg axis serves as a promising therapeutic target in cancer, more work needs to be done to understand the IL-33 signaling pathway and determine the potentially distinct impact of IL-33 on Tregs versus other immune cell types in these contexts (22), which could dictate the outcome of IL-33 blockade for cancer therapy.

While the mechanism underlying the transition from acute to chronic immune activation has not been previously elucidated, the existence of such a transition has been long reported (23). Recently, it is appreciated that the transition from acute, antitumor immunity to chronic inflammation is a common detriment to the efficacy of cancer immunotherapeutics (5). This issue is evident in the ACD paradigm itself: the tumor-suppressive effect of acute ACD has been used for decades to treat skin cancers (24). However, the repeated application of contact allergens may create a tumor-promoting immune environment (4). Therefore, blocking the IL-33/Treg axis may provide an approach to maximize the efficacy of cancer immunotherapeutics by sustaining long-term antitumor immunity. Further studies are required to determine whether IL-33/Treg axis blockade would be effective in reverting to a tumor-suppressing immunity after a chronic inflammatory environment has been established.

Materials and Methods

Anonymized human tissue samples were used in the study. Massachusetts General Hospital Institutional Review Board approved the anonymized tissue study. Formalin-fixed paraffin-embedded tissue sections (4 μm) were used for manual (IL-33) or automated immunostaining (CD3/Foxp3 and CD4/Foxp3) using a Ventana Ultra automated staining system (Ventana Medical Systems). Stained cells were counted blindly. Refer to SI Appendix, Materials and Methods for detailed description of the methods.