The E3 Ubiquitin Ligase Asb2α in T Helper 2 Cells Negatively Regulates Antitumor Immunity in Colorectal Cancer.

The escape of cancer cells from host immunosurveillance involves a shift in immune responses, including an imbalance in Th1 and Th2 cells. A Th1-dominated immune response predicts positive outcomes in colorectal cancer. The E3 ubiquitin ligase, Asb2α, is expressed in Th2 cells, but its roles in T-cell maturation and cancer are unclear. We provide evidence that the Th2 master regulator, Gata3, induces Asb2 Loss of Asb2 did not affect Th differentiation ex vivo, but reduced IL4 production from Th2 cells. We found that high ASB2 expression was associated with poor outcome in colorectal cancer. Loss of Asb2 from hematopoietic cells promoted a Th1 response and attenuated colitis-associated tumorigenesis in mice. Diminished Th2 function correlated with increased IFNγ production and an enhanced type 1 antitumor immune response in Asb2-deficient mice. Our work suggests that Asb2α promotes a Th2 phenotype in vivo, which in turn is associated with tumor progression in a mouse model of colitis.


Introduction
The key roles played by the tumor microenvironment, which is mainly characterized by the various stromal and immune cell subsets and the cytokines they produce (1), in the development, evolution, and outcome of cancers is increasingly appreciated. Notably, the quality of the immune response is an important prognostic factor for cancer patients (2)(3)(4). This is particularly true for patients with colorectal cancer (CRC) (5). CRC, which includes hereditary, sporadic and colitis-associated colorectal cancer (CAC) is a leading cause of cancer and cancer-related death (6). CAC represents about 2% of CRC cases and develop in patients with a history of inflammatory bowel disease, including ulcerative colitis and Crohn's disease (7). The risk of developing CAC is approaching 20% in patients with prolonged and extensive colitis (8). In addition to somatic and epigenetic abnormalities, the inflammatory microenvironment surrounding intestinal epithelial cells plays a pivotal role in carcinogenesis. Indeed, inflammation-related epithelial cell injury and repair contribute to the progression of CAC (9)(10)(11). Therefore, deciphering the mechanisms linking inflammation and CRC may facilitate the development of new therapeutic approaches.
Recent studies have highlighted the fundamental role played by CD4 + helper T cell in the generation and control of cancer-related inflammation (12)(13)(14). CD4 + T cells support cytotoxic CD8 + T cell (CTL) response (15) but may also exert direct tumor suppressive or promoting effects, depending on the cytokines they produce (16). CD4 + T cells include interferon-gamma (IFN-γ) producing Th1 cells, interleukin (IL)-4, -5, and -13 producing Th2 cells, IL-17 producing Th17 cells and IL-10 and TGF-β producing T regulatory (Treg) cells. An immune response dominated by Th1 cells is associated with a positive outcome in CRC patients while the role of Th17 and Treg cells in CRC pathogenesis remains controversial (9,(17)(18)(19). The contribution of Th2 cells in CRC development remains poorly understood.
In adults, the ASB2 gene encodes two isoforms, the hematopoietic-type ASB2α and the muscletype ASB2β (20, 21). ASB2α is the specificity subunit of a Cullin 5-RING E3 ubiquitin ligase that triggers ubiquitylation and degradation by the proteasome of the actin-binding protein filamins A and B (22-25). Although ASB2α transcript and protein were initially identified as induced in differentiating human myeloid leukemia cells (20), they are also expressed in non-malignant immune cells including dendritic cells (DCs) (23, 24) and CD4 + T lymphocytes (26,27).
In the present study, we asked whether ASB2α is involved in the immune response to CRC and investigated the impact of ASB2α deficiency in tumor development using a pathological relevant mouse model of CRC.

Human dataset and relapse-free survival analysis
We used published microarray and clinical outcome data deposited in the Gene Expression Omnibus (GSE39582) (28) to perform correlation and survival analyses. Tumor samples were split into ASB2 high, intermediate and low expressors based on gene expression value using the R software. Relapse-free survival of patients was analyzed according to the Kaplan-Meier method and differences between relapse-free survival distributions were assessed with the log-rank test.

Mice
All mice are specific pathogen free. Mice studies were handled according to the Centre National de la Recherche Scientifique ethical guidelines and were approved by the Comité d'éthique de la Fédération de Recherche en Biologie de Toulouse (C2EA-01). 300 µg polyinosinic-polycytidylic acid [poly(I·C); Sigma-Aldrich] was injected i.p. 3 times at 2-day intervals to 5-week-old Mx1-Cre;Asb2 fl/fl (23) and Mx1-Cre control mice. Female mice were injected i.p. 2 weeks after the last injection of poly(I·C) with a single 10 mg/kg dose of azoxymethane (AOM; Sigma-Aldrich). Dextran sodium sulfate (DSS; MP Biochemicals; MW 36,000-50,000 Da) was dissolved in drinking water, filtered and administered at 2% to mice for 8 days at 9, 13, and 16 week of age. Naïve groups that were not exposed to AOM/DSS were also included as noncancer controls. Mice were euthanized and analyzed 43 or 69 days after AOM injection. Bone marrow chimeras were generated by lethally irradiating CD45.1 C57BL/6 recipient mice (9 Gy) using a γ-irradiation system (Biobeam 8000). Irradiated mice were reconstituted by retro-orbital injection of 5X10 6 bone marrow cells of Mx1-Cre;Asb2 fl/fl or Mx1-Cre control mice previously flushed from femurs, subjected to red blood cell lysis, and filtered through a 30-µm filter. Chimera were kept on antibiotic-containing water (enrofloxacin) for two weeks. Eight weeks after transplantation, mice were injected with poly(I·C) and subjected to the AOM/DSS protocol as described above. For in vivo Th2 cell transfer, naïve T cells from CD45.1 C57BL/6 mice were differentiated for 6 days in Th2polarizing conditions and then were transferred into AOM/DSS treated cKO mice (2X10 6 Th2 cells for the first retro-orbital injection, 1X10 6 Th2 cells for the others). The reconstitution of bone marrow and the presence of Th2 cells in the colon after Th2 cell transfer were confirmed by flow cytometry analysis using anti-CD45.1 and anti-CD45.2. For antibody neutralization, mice were injected i.p. with 75 µg anti-IFN-γ neutralizing antibodies or isotype-matched control IgG1Ƙ (Ultra-LEAF™ purified antibodies, Biolegend) every four days after the second DSS cycle. Colons were dissected and measured. Colonic contents were removed and colons were cleaned with PBS. Tumors were counted and their area was measured using the AxioVision software (Zeiss). A portion of the colon was fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned with a Leica RM2245 microtome for alcian blue and nuclear fast red staining to examine tumor morphology. For the analysis of peripheral blood parameters, blood was collected retro-orbitally and analyzed with an ABX Micros 60 (Horiba) hematology analyzer.

Micro-array and ChIP-Seq Analysis
Micro-array ASB2 expression data and ChIP-Seq data showing global H3K4me3 and H3K27me3 histone modification status in the Asb2 locus in naïve CD4 + T cells or cells cultured under Th1, Th2, Treg and Th17 conditions were retrieved from the publicly available GSE14308 dataset (29). The patterns of GATA3 and Fli1 binding as well as H3K4me1, H3K4me2, H3K4me3, and H3K27me3 modifications at the Asb2 gene locus were retrieved from the GSE20898 (30).

Isolation of hematopoietic cells from organs
Isolation of bone marrow cells and splenocytes was performed as described (24). For isolation of immune cells from colonic mucosa, colon tissues were minced in 1-3 mm pieces and incubated at 37°C in RPMI medium containing 5 mM EDTA, 1% penicillin/streptomycin, 3% fetal calf serum (FCS; Biowest) for 20 min and then agitated three times 30 sec in RPMI, 2 mM EDTA, 1% Penicillin/Streptomycin. Colon pieces were then washed in PBS, cut in smaller pieces and digested with collagenase D (0.5 mg/ml, Roche) and DNase I (0.1 mg/ml, Roche) in RPMI for 60 min at 37 °C. Single cell suspensions were generated from digested tissues after filtration through a 40 µm-strainer.

Statistical analysis
All experiments were repeated as indicated. n indicates the numbers of independent biological repeats. All p-values were calculated using the nonparametric Mann-Whitney t-test using the GraphPad Prism software, unless otherwise indicated. A p value <0.05 was considered significant (* p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001).

Asb2α is expressed in Th2 cells
To determine whether Asb2α contributes to T cell maturation, we evaluated its expression upon differentiation of wild-type naïve CD4 + T cells in vitro. By quantitative RT-qPCR with primers specific to the Asb2α isoform, we showed that Asb2α transcripts were highly expressed in Th2 cells compared to naïve, Th1, Treg or Th17 cells (Fig. 1A), in agreement with Affymetrix gene chip data (Fig. 1B). We therefore searched for the molecular mechanisms that may control Asb2 expression in Th2 cells. Global mapping of histone H3 lysine 4 and lysine 27 trimethylation (H3K4me3 and H3K27me3, respectively) in naïve and CD4 + effector T cells indicated that H3K4me3 active marks were associated with Asb2 in Th2 cells (Fig. 1C). Moreover, a genome-wide ChIP-seq analysis of Th2 master regulator Gata3 in naïve and CD4 + T cells (30) revealed Gata3 binding within the Asb2 locus in Th2 cells (Fig. 1D). The expression of Asb2 transcripts in Th2 cells was reduced following Gata3 deletion (ratio of the levels of Asb2 mRNA in wild-type versus Gata3-deleted Th2 cells = 2.24±0.12) (30). In addition, the genomic regions surrounding Gata3 binding sites in Asb2 were associated with H3K4me1, H3K4me2 and H3K4me3 active marks, but no notable signals of the H3K27me3 repressive mark in Th2 cells (Fig. 1D).
Bioinformatics analysis of the Asb2 locus further indicated that the three Gata3 binding sites (Fig. 1E) are surrounded by motifs for the Ets (AGGAAG), Runx (ACCACA) and AP1 (TGACTCA) families of transcription factors known to be involved in T cell differentiation and function. As shown in Fig. 1D, the Ets family member Fli1 colocalizes with Gata3 in the Asb2 locus as previously shown for the Il-4/Il-13 locus (30). Taken together, these results suggest that Gata3 positively modulates Asb2 expression by regulating histone methylation in Th2 cells.

High ASB2 expression in human CRC is associated with shorter relapse-free survival
To further investigate the Th1/Th2 balance in CRC and uncover a potential role for ASB2, we analyzed ASB2 expression in fresh-frozen primary tumor samples containing tumor, stromal and infiltrating cells, from a large multicenter cohort with stage I to IV CRC (28). As shown in Fig. 2A, transcript levels for ASB2 and for a marker indicative of total leukocytes (CD45/PTPRC) showed strong positive correlation. Further, mRNA expression levels of master Th2 transcription factors, GATA3, c-MAF and IRF4, showed strong positive correlation with those of ASB2 in samples from human CRC ( Fig. 2A). Strikingly, we found that ASB2 expression was higher in the C4 and C6 subtypes, which are associated with shorter relapse-free survival (28), than in the C1, C2, C3 and C5 subtypes (Fig. 2B). When patients were categorized according to ASB2 expression levels, high levels of ASB2 were associated with shorter relapse-free survival (Fig. 2C). Accordingly, patients of the C4 and C6 subtypes mainly clustered in the ASB2 high and ASB2 int groups (Fig. 2D). These results reveal that low expression of ASB2 in CRC biopsies is associated with a positive outcome for patients.

Loss of Asb2 in hematopoietic cells inhibits colitis-associated CRC in mice
To elucidate a potential functional relationship between ASB2 expression and CRC, we used the Mx1-Cre;Asb2 fl/fl mutant mouse model (23). In this model, Asb2 is conditionally inactivated upon treatment with poly(I·C) (hereafter referred to as cKO). Two weeks after the last injection of poly(I·C), we subjected these mutant mice and their Mx1-Cre control counterparts (hereafter referred to as ctrl) to a model of colitis-associated tumorigenesis by administration of a single dose of the carcinogen azoxymethane (AOM) followed by repeated cycles of dextran sulfate sodium (DSS) in the drinking water ( To evaluate whether these phenotypes are due to loss of Asb2 in hematopoietic cells, we generated mouse bone marrow chimeras. We irradiated C57Bl/6 mice, reconstituted their bone marrow with cells from Mx1-Cre;Asb2 fl/fl or Mx1-Cre animals, and treated them with poly(I·C) to generate hematopoietic-cKO and ctrl mice (Fig. 3L). After AOM/DSS treatment, the hematopoietic-cKO mice displayed substantially reduced colon tumor development compared to ctrl mice (Fig. 3L). Taken together, our results indicate that loss of Asb2α in hematopoietic cells dramatically reduced tumor burden in colitis-associated tumorigenesis in mice. These functional findings are consistent with our observation that low expression of ASB2 in CRC biopsies correlates with a better relapse-free survival of patients.

Colitis-associated tumorigenesis in mice triggers activation and myeloid-biased differentiation of hematopoietic stem cells (HSCs)
Similar to the human CRC disease (31, 32), the numbers of monocytes, granulocytes and platelets in the blood of AOM/DSS treated mice were increased whereas the numbers of red blood cells were decreased, but there were no differences between cKO and ctrl mice (Fig. 4A). In contrast, the lymphocyte blood counts were similar in untreated and AOM/DSS treated cKO or ctrl mice (Fig. 4A). The increased numbers of monocytes and granulocytes in the blood of tumor-bearing cKO and ctrl mice are likely due to an emergency myelopoiesis in the bone marrow (Fig. 4B) and an extramedullar myelopoiesis in the spleen (Fig. 4C). Of note, increases in the lineage -Sca-1 + c-Kit + (LSK) cell compartment (including long-term HSCs, short-term HSCs and multipotent progenitors (MPPs)), in the granulocyte/macrophage progenitors (GMP), in monocytes and in immature myeloid cells (CD11b + Ly6G + Ly6Cand CD11b + Ly6G low Ly6C high ) were observed in the bone marrow of tumor-bearing mice (Fig. 4B). Expansion of myeloid cells also occurred in the spleen of AOM/DSS treated mice (Fig.  4C). In contrast, the numbers of B and T cells in the spleen of cKO and ctrl mice were not affected by tumor development (Fig. 4C). Although a skewing of HSC differentiation toward the myeloid lineage has been previously described in colitis (33), our data provide the first evidences for myeloid biased differentiation of HSCs in colitis-associated tumorigenesis in mice. However, no differences in the numbers of hematopoietic stem progenitor cells between cKO and control mice were observed (Fig.  4), indicating that the activation of the most immature HSCs and their myeloid differentiation were similar and could not explain the reduced tumor development observed in cKO mice.

Mice lacking Asb2 display markedly reduced Th2 and Treg together with enhanced Th1, Th17 and CTL response
We next examined whether Asb2α controlled immune cell infiltration by analyzing immunocyte profiles in the colon of mice treated with AOM/DSS. More CD4 + T cells accumulated in the colonic mucosa of cKO mice than in their ctrl counterparts 43 days after AOM injection, when both genotypes bear similar tumor numbers (Fig. 5A). In contrast, no difference was found between ctrl and cKO mice in the infiltration by DCs, macrophages and B cells into the colonic mucosa of tumor-bearing mice, nor in the numbers of DCs in the draining lymph nodes (supplementary Fig. S1). Analysis of the differentiation profile of lymphocytes infiltrating the colonic mucosa of cKO vs control tumor-bearing mice revealed markedly altered differentiation programs. First, we observed a decrease in the number of Th2 (CD4 + IL-4 + or CD4 + IL-13 + ; Fig. 5B) and Treg (CD4 + FoxP3 + IL-10 + ; Fig. 5C) cells in the colonic mucosa of tumor-bearing cKO mice compared to in ctrl mice. In contrast, the numbers of Th1 (CD4 + IFN-γ + ; Fig.  5D) and Th17 cells (CD4 + IL-17A + ; Fig. 5E) in the colonic mucosa of tumor-bearing cKO mice were increased. Both γδ T and IFN-γ + γδ T cells as well as NK and IFN-γ + NK cells in the colon of tumor-bearing mice showed a non-significant increase in cKO mice (supplementary Fig. S2).
More CD8 + T cells tend to accumulate in the colonic mucosa of cKO mice than in their ctrl counterparts 43 days after AOM injection (Fig. 5F). Moreover, CD8 + T cells from tumors of cKO mice produced more IFN-γ when compared to tumor-associated CD8 + T cells from ctrl mice (Fig. 5G). Furthermore, increased levels of transcripts of the cytotoxicity-related marker perforin 1 and granzyme B in the colonic mucosa of cKO vs ctrl mice treated with AOM/DSS were observed (Fig. 5H). From these data, we infer that Asb2α controls the balance between Th1/Th17/CTL versus Th2/Treg, promoting the Th2/Treg response. The enhanced Th1/Th17/CTL response in cKO mice likely contributes to the protection against tumor progression.

Deletion of Asb2 impedes Th2 response of mouse CD4 + T cells
To investigate how Asb2α loss altered CD4 + T cell polarization, we first investigated the impact of Asb2α loss on the ex vivo generation of Th2 cells from naïve CD4 + cells. As shown in Fig. 6A, deletion of Asb2 had no impact on the expression of the master regulator Gata3 mRNA in Th2 cells and had no impact on the expression of the master regulators Tbet/Tbx21 mRNA in Th1 cells after 6 days of differentiation. This was confirmed at the protein levels by flow cytometry (Fig. 6B-C). However, the percentage of IL-4 + cells and the geometric mean fluorescence intensity (geoMFI) for IL-4 in CD4 + cells were lower in Th2 cells generated from cKO mice than from ctrl mice (Fig. 6D). In contrast, Asb2 deletion had no impact on the percentage of IFN-γ + cells in CD4 + cells generated in Th1 conditions (Fig.  6B) and no impact on the ex vivo generation of Th17 and Treg cells from naïve CD4 + cells (Fig. 6E-F). Furthermore, the frequencies of CD4 + IL-17A + Th17 cells and CD4 + FoxP3 + IL-10 + Treg cells in the spleen, mesenteric lymph nodes and colons of untreated control and cKO mice were similar (supplementary Fig. S3). Altogether, our results indicate that the stronger Th1/Th17/CTL response in AOM/DSS treated cKO mice is secondary to diminished functions of Th2 cells, consistent with their cross-regulation.

Enhanced expression of IFN-γ upon loss of Asb2 in Th2 cells suppresses CRC progression in mice
To determine whether the reduced tumor burden observed in AOM/DSS treated cKO mice was due to Asb2 deficiency in Th2 cells, we transfer wild-type Th2 cells into cKO mice. The transfer of wild-type Th2 cells into cKO mice increased the tumor growth to a level similar to that observed in control mice (Fig. 7A-C and 3G-H). We then questioned the mechanisms underlying the Th1/Th17/CTL response and the role of IFN-γ in anti-tumor immunity. Indeed, we observed increased numbers of IFN-γ + cells and increased levels of Ifn-γ transcripts in the colonic mucosa of cKO mice treated with AOM/DSS (Fig. 7D-E) that are likely to contribute to the reduced tumor burden following Asb2 deletion. In order to evaluate the role of IFN-γ in the reduced tumor progression in cKO mice, anti-IFN-γ neutralizing antibodies were injected to cKO mice. As shown in figure 7F-G, the tumor load was increased when cKO mice were administered anti-IFN-γ neutralizing antibodies. Tumor numbers were also increased in colons of cKO mice treated with anti-IFN-γ compared to colons of cKO mice treated with control antibodies (Fig. 7F and 7H). Furthermore, large tumors were more numerous in mice treated with anti-IFN-γ antibodies than in mice treated with control antibodies (Fig. 7F and 7H-I). Together, these results indicate that IFN-γ signaling is essential for protection of cKO mice against CAC development.

Discussion
Our study reveals an unexpected role for the hematopoietic E3 ubiquitin ligase Asb2α in tumor development. Analysis of ASB2 expression in tumor samples of patients with CRC indicated that ASB2 expression was higher in the subtypes associated with shorter relapse-free survival (28). In addition, Asb2 deficiency in hematopoietic cells dramatically reduced tumor development in a mouse model of colitis-associated tumorigenesis. The third most striking aspect of this study is that these are likely due to altered functions of Th2 cells.
It is now widely accepted that the CD4 + T cell subsets play differential roles during tumor development. Among them, Th1 cells can repress tumor growth by secreting IFN-γ and supporting the activity of cytotoxic T lymphocytes (34). Here, we show that cKO mice develop fewer and smaller tumors than control mice with increased levels of IFN-γ indicative of a higher Th1 response, and enhanced expression of the cytotoxic markers perforin and granzyme B in the tumor area. In line with this finding, IFN-γ neutralization increases tumor development in cKO mice, further supporting a key role for IFN-γ-dependent signaling in protection of CAC. Interestingly, high densities of Th1 cells in the tumor were shown to have better prognostic value in colorectal cancer than the classic Tumor Node Metastasis classification (35,36).
We and others have reported that Asb2 mRNAs are expressed in Th2 cells (26, 27, 30). Consistent with these findings, Asb2 is a target gene of the Th2 master transcription factor Gata3, and Gata3 binding to WGATAA motifs in the Asb2 locus in Th2 cells likely facilitates H3K4me1-, H3K4me2-and H3K4me3-mediated activation of Asb2. It has been proposed that Gata3-mediated gene regulation in Th subsets depends on specific cofactors (30). Indeed, the authors showed that in addition to the primary WGATAA motif, GATA binding sites in Th2 cells contained secondary motifs for other transcription factors known to be involved in T cell differentiation such as members of the Ets or Runx families. Not surprisingly, this is also the case for Asb2. It is therefore tempting to speculate that Asb2 belongs to the genes that play functional roles in Th2 cells.
The fact that the number of IL-4 positive Th2 cells is reduced in the colonic mucosa of tumorbearing cKO mice and following ex vivo Th2 polarization of naïve CD4 + T cells of cKO mice, strongly suggests that the balance in Th1/Th2 responses of tumor-bearing mice is affected by Asb2 deletion and has functional consequences in the development of the disease. Indeed, it is well-established that cross regulation by IL-4 and IFN-γ suppresses Th1 and Th2 differentiation, respectively (37)(38)(39)(40)(41). Although Th2 cells represent a major component in the activation and regulation of humoral immunity and allergic inflammatory responses, their roles in tumor immunity are yet to be investigated. Indeed, Th2 cells have pro-or antitumor activity depending on the type of cancer. Th2 cells have antitumor activity in a model of melanoma resistant to cytotoxic T lymphocytes (42). In contrast, Th2 cells are associated with tumor progression in renal cell carcinoma (43), in breast cancer (44), in melanoma (45) and in pancreatic carcinogenesis (46,47). In addition, depletion of IL-4 in a mouse model of metastatic melanoma limits tumor growth suggesting a pro-tumor effect for IL-4 (48). In agreement with our data in the AOM/DSS mouse model of CAC, tumor development is increased in the absence of IFN-γ in the inflamed colon in another mouse model of CAC (49). We here provide evidence that in a mouse model of colitis-associated tumorigenesis in which Th2 cells promoted tumor growth, Asb2 deficiency in Th2 cells blunted Th2 cytokine production leading to enhanced type 1 anti-tumor immune response.
Altogether, our results not only reveal that Asb2α is an unsuspected negative regulator of antitumor immune response in CRC, but also demonstrate that Asb2α plays functional roles in Th2 cells. We propose that Asb2α should be added to the growing list of E3 ubiquitin ligases involved in the regulation of CD4 + T cell identity and function. Of note, Asb2α was recently shown to induce IκBα degradation in T-ALL (acute lymphoblastic leukemia) cell lines (50). Whether IκBα or other proteins are substrates of Asb2α in Th2 cells and whose substrate degradation mediates Asb2α effects in these cells remain to be determined. Since E3 ubiquitin ligases are druggable, Asb2α might be considered as a promising pharmacological target to modulate Th2 response in cancer.