High IL-1R8 expression in breast tumors promotes tumor growth and contributes to impaired antitumor immunity

Tumors develop numerous strategies to fine-tune inflammation and avoid detection and eradication by the immune system. The identification of mechanisms leading to local immune dysregulation is critical to improve cancer therapy. We here demonstrate that Interleukin-1 receptor 8 (IL-1R8 - previously known as SIGIRR/TIR8), a negative regulator of Toll-Like and Interleukin-1 Receptor family signaling, is up-regulated during breast epithelial cell transformation and in primary breast tumors. IL-1R8 expression in transformed breast epithelial cells reduced IL-1-dependent NF-κB activation and production of pro-inflammatory cytokines, inhibited NK cell activation and favored M2-like macrophage polarization. In a murine breast cancer model (MMTV-neu), IL-1R8-deficiency reduced tumor growth and metastasis and was associated with increased mobilization and activation of immune cells, such as NK cells and CD8+ T cells. Finally, immune-gene signature analysis in clinical specimens revealed that high IL-1R8 expression is associated with impaired innate immune sensing and T-cell exclusion from the tumor microenvironment. Our results indicate that high IL-1R8 expression acts as a novel immunomodulatory mechanism leading to dysregulated immunity with important implications for breast cancer immunotherapy.


INTRODUCTION
During tumorigenesis, cancer cells interact with the immune system and modify the tumor microenvironment (TME) to subvert the host immune response [1,2].
In advanced tumors, the immune balance is tilted towards pro-tumor inflammation and local dysregulation of the innate and adaptive immune response [3]. Inflammatory signals derived from tumor cells, stromal cells and immune cells act as critical factors regulating qualitative

Research Paper
and quantitatively pro-tumoral inflammation and antitumor immunity during the cancer-immunity cycle [3]. This complex crosstalk has been recognized as a hallmark of cancer [4], and understanding the molecular pathways involved in the induction of pro-tumoral inflammation and immune evasion could enable the development of new therapies and improve the efficacy of existing ones.
The activation of Interleukin-1 (IL-1) receptor family members (ILRs) and Toll-like receptors (TLRs) represents a critical early innate immune sensing event promoting immunosurveillance and antitumor immunity [5][6][7]. ILRs and TLRs activation ignites a signal transduction cascade with pro-inflammatory outcomes, including the activation of NF-κB and the secretion of pro-inflammatory cytokines such as Type I IFNs and TNFα, which are in turn necessary for natural killer (NK) and dendritic cell (DC) activation, together with CD8 + T cell priming against tumor antigens and trafficking into the TME [8][9][10]. Therefore, ILRs and TLRs activation plays an important role in inflammation, initiation and amplification of innate immunity and orientation of adaptive immunity [11,12].
IL-1R8 is a member of the ILRs family. IL-1R8 negatively regulates signaling from IL-1R1, IL-18R, ST2 and a number of TLRs [13,14] by acting as a decoy receptor for key components of the MyD88 signaling cascade, such as IRAK and TRAF6 and by interfering with ILRs dimerization through its Ig domain [13][14][15].
Recently, it has been demonstrated that, in humans, IL-1R8 can also bind to IL-37, an anti-inflammatory cytokine induced by TLRs and cytokines and a natural suppressor of innate inflammatory response [16,17]. Curiously, no IL-37 homologue has been identified in mice wherein.
Irrespectively of its mode of action, IL-1R8 acts as a brake for pro-inflammatory signals, and its expression is essential for regulating the detrimental effects of innate and adaptive immune responses in pathologic conditions such as infections, sepsis, chronic inflammation and autoimmune disease [18][19][20][21]. In addition, IL-1R8deficiency in mice results in more severe gut inflammation during dextran sulfate sodium colitis [22,23], increased susceptibility to colitis-associated colorectal cancer [23][24][25], and more severe and earlier onset of monoclonal B cell expansions in a murine model of chronic lymphocytic leukemia (CLL) [26].
In this study, we show that IL-1R8 is up-regulated during breast epithelial cell transformation and in primary breast tumors. Using clinical samples and in vitro and in vivo experiments, we also demonstrate that high expression of IL-1R8 in breast tumors modulates the expression of inflammatory mediators in the TME, affecting the mobilization and activation of immune cells and fostering tumor growth and metastasis. Collectively, our findings indicate that expression of IL-1R8 represents a novel immunomodulatory mechanism leading to impaired innate immune sensing and antitumor immunity and provides new insights to cancer immunotherapy.

IL-1R8 is up-regulated in transformed breast epithelial cells and in primary breast tumors
IL-1R8 was identified as an up-regulated gene in transformed breast epithelial cells by comparing gene expression profiles from a parental, non-transformed, conditionally immortalized human mammary luminal epithelial cell line (HB4a), and a HER2 overexpressing variant (HB4a-C5.2, designated HB4a HER2+ for the purpose of this work) [27]. Transcriptional changes associated with breast epithelial cell transformation were measured using Massively Parallel Signature Sequencing (MPSS) and IL-1R8 ranked among the top 50 differentially expressed genes (unpublished results). Reliable MPSS tags (5ʹGATCATAGGGACAGCGG3') assigned to IL-1R8 were more frequently found in the HB4a HER2+ library than in the HB4a library (36 tpm vs. 4 tpm, P < 0.001), indicating that IL-1R8 gene expression is up-regulated in the transformed breast epithelial cells. IL-1R8 differential expression in the HB4a HER2+ variant was confirmed both at the mRNA and protein levels. A 4-fold induction of IL-1R8 mRNA and a 2-fold induction of IL-1R8 protein expression were observed in HB4a HER2+ cells when compared to HB4a ( Figure 1A).
IL-1R8 up-regulation in primary breast tumors was confirmed by analyzing RNA-seq expression data obtained from The Cancer Genome Atlas (TCGA). We observed that IL-1R8 gene expression is significantly higher in primary breast tumors compared to normal breast tissue (median 701.1 vs. 358.8 RSEM normalized expression values, P < 0.0001, Figure 1B) and higher levels of IL-1R8 mRNA were observed across all molecular breast cancer subtypes, except in the basal-like breast cancer subtype (HER2 + subtype median 563.4 RSEM normalized expression values, P = 1.13e -05 , Luminal A subtype median 830.2 RSEM normalized expression values, P < 2.2e-16, Luminal B median 823.9 normalized expression values, P < 2.2e-16 and basal-like subtype median 360.9 normalized expression values, P = 0.83) ( Figure 1B).
Collectively, these results indicate that IL-1R8 is upregulated during breast epithelial cell transformation and across all molecular breast cancer subtypes, except in the basal-like subtype.

IL-1R8 up-regulation in transformed breast epithelial cells fine-tunes IL-1-dependent NF-κB activation and the expression of proinflammatory cytokines
IL-1R8 negatively regulates the innate inflammatory response by acting as a decoy receptor for TLRs and ILRs signaling. NF-κB activation and the production of pro-inflammatory cytokines are important endpoints of TLR and IL-1R family signaling [28]. Gene transfer experiments have shown that IL-1R8 up-regulation inhibits NF-κB activation and the production of pro-inflammatory cytokines in HeLa and hepG2 cells after exposure to IL-1 and TLR ligands [14]. Therefore, to address the role of IL-1R8 up-regulation in transformed breast epithelial cells, we treated HB4a and HB4a HER2+ cells with IL-1β and analyzed IκB expression levels and NF-κB activation by Western blot and EMSA, respectively. IκB is negative regulator of NF-κB and is phosphorylated by specific kinases in response to inflammatory signals. The phosphorylated IκB protein is then ubiquitinated and degraded leading to a decrease in IκB expression levels and to NF-κB activation. As expected, IκB expression decreases shortly after IL-1β treatment in both cell lines, but the decrease was less pronounced in HB4a HER2+ cells compared to HB4a ( Figure 1C). To verify if the observed differences in IκB expression levels were due to IL-1R8 up-regulation in HB4a HER2+ cells, we generated HB4a HER2+ cell variants stably expressing IL-1R8 shRNAs (HB4a HER2+/IL1R8KD ) (Supplementary Figure 1). No differences in the growth rate between HB4a HER2+ and HB4a HER2+/IL1R8KD cells studied herein were observed and IL-1R8 knockdown remained stable during the course of this work (data not shown). After IL-1R8 knockdown, the differences in IκB expression levels observed between HB4a and HB4a HER2+ disappeared ( Figure 1C). In addition, NF-κB activation was significantly enhanced in HB4a HER2+/ IL1R8KD cells compared to HB4a HER2+ cells ( Figure 1D).
Gene-expression of different pro-inflammatory cytokines in HB4a and HB4a HER2+ cells was then analyzed after stimulation with IL-1β. Shortly after stimulation, a significant increase in IL-6, IL-8, CSF2, CSF3, IFN-β1 and TNFα expression was observed in both cell lines, although the induction of these inflammatory mediators was significantly lower in HB4a HER2+ cells compared to HB4a cells. Noteworthy, IL-1R8 knockdown in HB4a HER2+ cells reverted the expression of these cytokines after IL-1β stimulation to the levels observed in parental HB4a cells ( Figure 1E).
Taken together, these results indicate that IL-1R8 up-regulation in transformed breast epithelial cells attenuates IL-1-dependent NF-κB activation and the expression of pro-inflammatory cytokines.
Bone marrow chimeras were then used to assess the relative importance of IL-1R8 expression in the nonhematopoietic/tumor cells in the reduced tumor burden and growth observed in MMTV-neu/IL-1R8 −/− mice. No significant differences in tumor onset, incidence and burden were observed when MMTV-neu/IL-1R8 −/− mice were transplanted with IL-1R8 +/+ bone-marrow cells, supporting an important role for IL-1R8 expression by non-hematopoietic/tumor cells during breast tumorigenesis (Supplementary Figure 2).
These results reveal that IL-1R8 expression in breast tumor cells contributes to tumor formation, progression and metastatic dissemination.

IL-1R8-deficiency in a transgenic mouse model of breast cancer promotes the mobilization and skews the activation of immune cells
Pro-inflammatory cytokines are critical factors regulating the mobilization and activity of immune cells in the TME [3]. Since IL-1R8 can negatively modulate the  expression of inflammatory mediators in IL-1R-stimulated transformed breast epithelial cells in vitro, we sought to analyze its role on the immune composition in vivo using our transgenic mouse model of breast cancer.

IL-1R8 expression in clinical samples is associated with impaired mobilization of NK, DC and CD8 + T cells
The relative abundance of tumor-infiltrating leukocytes in clinical samples can be indirectly quantified by measuring intratumoral transcript levels of coordinately expressed immune cell-specific genes (immune metagenes) [36]. To determine the clinical relevance of our in vivo and in vitro findings, we analyzed RNA-seq expression data from 1102 primary breast tumors, including all four major molecular subtypes obtained from TCGA. Breast tumors were classified according to IL-1R8 expression levels irrespectively of their molecular subtype (see Materials and Methods) and were analyzed for the expression of immune metagenes.
A similar analysis was carried out after classifying primary breast tumors according to their molecular subtype. Lower expression levels for most of these immune metagenes were also observed in IL-1R8 high luminal A and B molecular subtypes. However, clear differences in the expression level for these immune metagenes were not observed for IL-1R8 high Her2+ and basal-like tumors, possibly due to a small sample size in the case of Her2+ tumors (n = 65) and to the lack of IL-1R8 up-regulation in basal-like tumors (Supplementary  Tables 3-6).
These results corroborate our in vivo and in vitro observations and indicate that high IL-1R8 expression in primary human breast tumors is associated with a skewed inflammatory response and impairment in the mobilization of protective leukocytes in the TME.

IL-1R8 expression in clinical samples is associated with a non-T cell-inflamed TME
Solid tumors can be classified into two broad categories according to cellular and molecular characteristics of the TME [37]. T cell-inflamed tumors are characterized by a spontaneous T cell infiltrate, a broad chemokine profile, which supports influx of CD8 + effector T cells, and a type I IFN signature indicative of innate immune activation. In contrast, non-T cell-inflamed tumors lack all these characteristics and present an immunosuppressive microenvironment that inhibits the activation of antigen presenting cells and the priming and trafficking of effector T cells [37,38] . Immune gene signatures based on the expression of T cell specific markers, different chemokines associated with CD8 + T cell recruitment [39] and IFN-induced genes have also been used to classify tumors according to their T cell-inflamed phenotype. Since IL-1R8 expressing MMTV-neu mammary tumors present a lower infiltration CD8 + T cells and lower levels of intratumoral IFN-γ, we used immune gene signature analysis to investigate whether IL-1R8 expression was associated with a non-T cell-inflamed TME in clinical samples.
Accordingly, 1,102 primary breast tumors including all major molecular subtypes, were classified into those with a non-T cell-inflamed and those with T cell-inflamed phenotype based of the expression level of T cell specific markers (FYB, LCP2, CD3E, CD8A, CD28), different chemokines associated with CD8 + T cell recruitment (CCL5, CCl2, CCl3, CCL4, CXCL9, CXCL10) and IFN-induced genes (IFI16, IFIH1, GBP1, OAS2, MX1). Approximately, 64% of all breast tumors expressed lower expression levels of the 16 T-cell inflamed signature genes (709/1,102) and, noteworthy, the majority of IL-1R8 high tumors were clustered in the group of tumors expressing lower levels of T-cell inflamed signature genes (394/551, 72%, P = 6.758e-07, Figure 5A). In addition, a significant inverse correlation between IL-1R8 expression and the expression of all, except one, T cell signature genes was observed ( Figure 5B), indicating that IL-1R8 expression is associated with a non-T cell-inflamed TME in breast tumors.
A similar analysis was carried out after classifying primary breast tumors according to their molecular subtype. Lower expression levels for most of these T cell signature genes were also observed for in IL-1R8 high luminal A and B molecular subtypes. However, clear differences in the expression level for these immune metagenes were not observed for IL-1R8 high Her2+ and basal-like tumors, possibly due to a small sample size in the case of Her2+ tumors (n = 65) and to the lack of IL-1R8 up-regulation in basal-like tumors (Supplementary Figure 8).
Altogether, these results indicate that high IL-1R8 expression in human breast tumors is associated with a non-T cell inflamed phenotype and may represent a new immune escape mechanism contributing to T cell exclusion and impaired antitumor immunity in this class of tumors.

DISCUSSION
The activation of innate immune sensors, such as ILRs and TLRs, is an essential mechanism of sterile immunity, playing a critical role in the promotion of pro-tumoral inflammation and mediation of immunosurveillance [10,40]. Therefore, molecules that negatively regulate their signaling may be exploited by tumors to induce immune tolerance and mitigate host antitumor immune response. Here, we demonstrated for the first time that IL-1R8, a negative regulator of proinflammatory signals, is up-regulated in transformed breast epithelial cells and in HER2+ and luminal primary breast cancer molecular subtypes. Most importantly, we demonstrated that its expression contributes to an impaired innate immune sensing and the development of an antitumor immune response.
IL-1R8 negatively regulates ILRs and TLRs signaling by acting as a decoy receptor and by interfering with ILRs dimerization through its Ig domain [13][14][15].
Recently, it has been demonstrated that, in humans, IL-1R8 can also bind to IL-37, an anti-inflammatory cytokine induced by TLRs and cytokines and a natural suppressor of innate inflammatory response [16,17]. The binding produces the formation of the tripartite complex involving IL-37/IL-18Rα/IL-1R8, which is necessary for the activation of an anti-inflammatory response. Therefore, in humans, IL-1R8 negatively regulates the innate inflammatory response by acting both as a decoy for TLRs and ILRs signaling and as a co-receptor for IL-37 in the activation of an anti-inflammatory program. In the present work, we have primarily explored the role of IL-1R8 up-regulation in the modulation of ILRs and TLRs signaling. Further studies will be necessary to properly evaluate the implications of IL-37 and IL-1R8 interaction in cancer biology.
IL-1R8 has been previously shown to act as a decoy receptor for ILRs and TLRs in human colon and bladder epithelial cells [23,41]. In the present work we confirmed IL-1R8 decoy activity in human breast epithelial cells. Using a mammary epithelial cancer cell model, we showed that IL-1R8 expression in transformed cells fine-tuned IL-1-dependent NF-κB activity and the expression of several pro-inflammatory cytokines, including some involved in macrophage polarization (CSF2, IFN-β1 and TNFα) and NK cell activation (IFN-β1 and TNFα) [42][43][44]. Indeed, using co-culture experiments, we observed that IL-1R8 expression in transformed cells skews macrophage and NK cell activation in vitro.
Furthermore, using a murine breast cancer model, we showed that MMTV-neu/IL-1R8 +/+ mammary tumors grew faster and were more metastatic compared to MMTV-neu/IL-1R8 −/− tumors. Although HER2 overexpression is sufficient to drive mammary lesions in MMTV-neu animals, the pace of tumor progression is known to be significantly influenced by tumor-elicited mechanisms of immunosuppression [45,46]. In line with this evidence, there was a significant increase of tumor-infiltrating leukocytes associated with antitumor immunity, including mature NK cells, DCs and CD8 + T cells, in MMTV-neu/IL-1R8 −/− mice.
Our in vivo results however are in contrast with those obtained with colitis-associated colon-cancer models and lymphocytic leukemia where IL-1R8 has been shown to play a tumor-suppressive role [24][25][26]. In addition, the aberrant expression of an IL-1R8 splicing isoform, with dominant-negative function in colon tumors, has been recently shown to promote inflammation and colitisassociated colon cancer in mouse and humans [47]. The dual role of inflammation in cancer is well known and several lines of evidence indicate that inflammatory and immune mechanisms in cancer are tissue-and organ specific [48,49]. Therefore, it is not surprising that IL-1R8 can act as a tumor suppressor in models of cancer strongly dependent on inflammation, while having a tumor promoting effect by fine-tuning inflammation and avoiding detection and eradication by the immune system in other tumor types. In line with our results, IL-1R8 up-regulation in prostate cancer has been recently associated with worse prognosis and can predict biochemical recurrence after prostatectomy in low-grade prostate cancer patients [50].
In accordance with our in vitro experiments using IL-1R8-knockdown cancer cells, we observed that MMTV-neu/IL-1R8 −/− TAMs showed reduced expression of markers associated with an M2-like phenotype, in particular CD206, YM1, IL-10 [43]. TAMs usually acquire a skewed M2-like phenotype oriented to tumor promotion and it has been reported that Th2 CD4 + T cells promote breast tumor progression and metastasis by educating TAMs to produce pro-angiogenic and pro-metastatic factors [51]. Therefore, the results showing that M1macrophage polarization, directly induced by MMTVneu/IL-1R8 −/− tumors or indirectly induced by leukocytes recruited in the tumor, such as NK cells, DCs and CD8 + T lymphocytes, may also account for reduced tumor growth and metastasis observed in MMTV-neu/IL-1R8 −/− mice.
Notably, the protected phenotype observed in MMTV-neu/IL-1R8 −/− mice was not reversed when mice were transplanted with IL-1R8 +/+ bone-marrow cells, supporting an important role played by IL-1R8 expression in non-hematopoietic cells. Since besides leukocytes, cells which express highest levels of IL-1R8 are epithelial cells [13], we hypothesize that the observed effects on tumor growth and progression can be directly attributed to IL-1R8 expression in tumor cells, which might be sufficient to influence the inflammatory and immune responses occurring within the TME.
Using immune gene signature analysis, we confirmed that high IL-1R8 expression in primary breast tumors is significantly associated with impaired innate immune sensing and the mobilization of protective leukocytes. Using the same approach, we have also shown that the majority of primary breast tumors displayed a non-T cell inflamed phenotype and that high IL-1R8 expression in primary breast tumors is associated with lower expression levels of T-cell inflamed signature genes.
Increasing evidence support that current immunotherapies such as checkpoint blockade are predominantly effective in patients with a pre-existing T cell-inflamed tumor microenvironment [52][53][54].
Understanding the molecular mechanisms leading to T cell exclusion and to resistance to T cell-based immunotherapies would improve patient selection and allow the development of novel treatment modalities, expanding the fraction of patients benefiting from current immunotherapies. In this context, further studies to address a putative role for IL-1R8 expression as a prognostic and predictive marker for immunotherapy in breast cancer are necessary and are worth undertaking.
Interestingly, pre-clinical studies have demonstrated that the use of different agents to induce tissue-based inflammation and boost innate immunity, including activation of the Stimulator of Interferon Genes (STING) and local irradiation, can restore T cell trafficking and favorably alter the TME [55,56]. Based on our findings, we propose that IL-1R8 expression in breast tumors represents a new immunomodulatory mechanism leading to a dysregulated inflammatory response and impaired antitumor immunity. Our findings therefore have important therapeutic implications and strategies to block IL-1R8 activity directly or through combined therapies as a way to restore innate immune system activation and T cell trafficking in breast cancer TME should be further explored.

Cell lines and IL-1R8 knockdown
The human mammary epithelial cell line HB4a and its variant HB4a -C5.

Co-culture experiments
Cells were treated with 5 ng/mL of IL-1β (R&D Systems, Minneapolis, MN, USA). After 24 h, the culture supernatant was collected and stored at -80ºC. THP-1 cells were cultured for 24 h with supernatants or 25 ng/mL LPS (M1-like control) or 30 ng/mL IL-4 (M2-like control). Cells were stained with anti-CD206 (eBioscience, San Diego, CA, USA), anti-CD86 (BD Biosciences, San Jose, CA, USA), anti-CD14 (BD Biosciences) and analyzed using FACScantoII (BD Biosciences). For co-culture assays with NK cells, human transformed mammary cell lines were seeded and allowed to adhere to the plate for 24 h. NKL cells were added in their complete medium and cells were co-cultured for 4h. IFN-γ secretion was measured by ELISA.

Animal model
F3 MMTV-neu/IL-1R8 +/+ and MMTV-neu/IL-1R8 −/− female mice were sacrificed at 24 weeks of age or once their biggest mammary tumor reached 500 mm 3 . For bone marrow transplantation, 3-weeks old MMTV-neu/ IL-1R8 −/− mice were lethally irradiated with a total dose of 900 cGy. Mice were then injected in the retro-orbital plexus with 5 × 10 6 nucleated bone marrow cells from IL-1R8 +/+ or IL-1R8 −/− donors. All experimental procedures were performed according to the Italian animal care and ethics legislation and had been approved by the Italian Ministry of Health.

In silico analysis of gene expression, metagenes and immune gene signature analyses
Breast invasive carcinoma Level 3 RNA-Seq data were downloaded from TCGA Portal (https://tcgadata.nci.nih) and molecular subtypes were classified as described [57]. Upper quartile normalized RSEM counts were used to estimate expression levels of IL-1R8 across 1,102 tumor samples and 113 normal samples. 792 out of those 1,102 samples, for which molecular classification was available, were used to compare IL-1R8 expression levels between each subtype. For the metagenes and immune gene signature analyses, tumor samples were classified into: "IL-1R8-high", if presenting IL-1R8 expression greater than the median; and "IL-1R8-low" if presenting IL-1R8 expression equal or less than the median.