Pharmaceutical targeting Th2-mediated immunity enhances immunotherapy response in breast cancer

The Cancer Genome Atlas (TCGA) and Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) gene expression data and clinicopathology data were extracted from cBioPortal (http://​www.​cbioportal.​org/​) [31], and other breast cancer cohorts’ data was accessed from Gene Expression Omnibus (GEO) datasets (https://​www.​ncbi.​nlm.​nih.​gov/​geo/​). We reanalyzed public datasets to show the functional gene expression signatures (Fges) scores (http://​science.​bostongene.​com/​tumor-portrait/​). Breast cancer was screened based on immunohistochemistry (ER, PR and Her2). The correlation analysis between GATA3 expression and CTL infiltration in 9 breast cancer datasets was administrated on Tumor Immune Dysfunction and Exclusion (TIDE) system (http://​tide.​dfci.​harvard.​edu/​faq/​) [32]. CIBERSORT algorithm was utilized to analyze 22 types of immune cells in breast cancer tissues, including 7 T cell subtypes, memory and naïve B cells, natural killer cells, and myeloid cells. After quality filtering (p < 0.05), breast cancer samples were included in the analysis. Additionally, six immune cell infiltration including CD8⁺ T cells, CD4⁺ effector T cells, B cells, macrophages, neutrophils, and dendritic cells were analyzed by Tumor Immune Estimation Resource (TIMER) (http://​timer.​cistrome.​org/​) [33]. MTABRIC breast cancer samples were used to asses correlating Fges in TME. Signature scores of TME were calculated by python implementation of the single-sample gene set enrichment analysis (ssGSEA), and the correlation analysis of signature was performed by pearson’s correlation [34]. Th2 cell groups in breast cancer samples were based on 2 separations using GATA3 expression. CD8⁺ CTL function was elucidated using a 15-gene signature (IL2, CD8A, CCL5, GZMA, PRF1, IFNG, PTPRC, GZMM, RAB27A, XCL1, ICOS, TBX21, GZMB, GNLY, and IL12A). To test the association between GATA3 gene expression level and patient overall survival, Kaplan–Meier survival analysis was conducted based on TIDE online. T cell dysfunction scores of Th2 cells in breast cancer were evaluated by TIDE.

MDA-MB-231 cell (human breast cancer cell line), EO771 cells (a C57BL/6 murine breast cancer cell line), EMT6 cells (a BALB/c murine breast cancer cell line), and 4T1 cells (a BALB/c murine breast cancer cell line) were obtained from the American Type Culture Collection. All cells were cultured in RPMI-1640 (Roswell Park Memorial Institute-1640) medium with 10% FBS.

For EMT6, 4T1 and EO771 models, BALB/c or C57BL/6 mice were injected respectively with 1.5 × 10⁵ or 2 × 10⁵ tumor cells orthotopically. Mice were intragastrical administered with PBS or IPD (S2015, Selleck; 100 mg/kg) at day four post tumor induction [35]. Anti-CTLA4 antibody (BioXCell, Clone:9D9, Catalog #: BP0164; 5 mg/kg) and anti- PD-1 antibody (BioXCell, Clone:29F.1A12, Catalog #: BP0273; 2 mg/kg) were administered intraperitoneally on days 7, 10, 13 after tumor inoculation. Tumor volume was measured 2 or 3 times per week by using electronic calipers, and calculated according to the following formula: volume = (length × width²)/2. Female BALB/c and C57BL/6 mice between 6 and 8 weeks old were applied in all experiments, and all mice were supplied by the Laboratory Animal Centre of Chongqing Medical University.

Tumor-bearing mice were sacrificed after tumor inoculation (EO771 and EMT6: 18 days, 4T1: 30 days). Tumor samples were harvested, chopped with scissors, and incubated with 2 mg/ml collagenase A (Roche) for 40–50 min at 37 ℃. The dissociated cells were filtered through 70 μm filters (BD Biosciences) to obtain single-cell suspensions. Afterward, erythrocytes were lysed with red blood cell lysis buffer for 1 min on ice, and single-cell suspension was washed and re-suspended with DMEM or PBS depending on further use.

Cells were stained for Live/Dead with Fixable Viability Dye eFluor 450 (eBioscience) for 30 min at 4 ℃. After being washed once with flow cytometry buffer, cells were stained with dilutions of various combinations of primary antibodies to cell surface marker, including CD45 (APC-CY7), CD11b (BV510), CD3 (FITC), CD4 (PE-CY5.5), CD8 (PE-CY7/FITC), Gr-1 (PE), F4/80 (APC/BV605), Tim-3 (PE-CY7), PD-1 (PerCP-CY5.5) and I-A/I-E (PerCP-CY5.5). After that, cells were washed with flow cytometry buffer. Intracellular staining was administrated according to the manufacturer’s protocol. Briefly, cells were first fixed and permeabilized using the Foxp3 staining buffer kit (eBioscience), and then incubated with fluorochrome-conjugated antibodies to CD206 (APC), FOXP3 (FITC) and Ki67 (PE) from BioLegend. For intracellular cytokine staining, cells were stimulated with Cell Stimulation Cocktail (eBioscience) for 4 h at 37 ℃. Cells were then stained with anti-IFN-γ (APC) and anti-TNF-α (PE/Cyanine7). The dilution ratios for all antibodies were 1:100. After all staining procedures, cells were acquired using BD FACS Canto II and BD FACSDiva software (BD Biosciences). FlowJo software was used in subsequent analysis. The representative gating strategy for the flow cytometry was shown in the Additional file 8: Figure S8.

Mouse tumor tissues were mechanically processed and digested as described above. After filtration using 70 μm (BD Biosciences), cells were incubated with 25 mM cisplatin for 1 min (viability staining) and then stained with mAb cocktails against intracellular proteins. Metal tagged antibodies utilized in mass cytometry analysis were purchased from Fluidigm. Cells suspension was diluted to approximately 10⁶ cells per ml using ddH₂O containing bead standards and then analyzed on a CyTOF 2 mass cytometer (Fluidigm). CyTOF data was normalized and manually gate using Cytobank software. After CD45⁺ immune cells gated, data were transformed using cytofAsinh function before applying it to the downstream analysis. The immune subsets were generated during Phenograph clustering analysis through R cytofkit package. FlowSom, an unsupervised automated algorithm, was used to order cells based on their phenotypic similarities.

Based on the mean value of each marker in clusters, heatmaps were generated. Cell frequency of each cluster was calculated as a percent of each cell type divided by the total CD45⁺ cells in same sample. Antibodies used in the mass cytometry analysis were purchased from Fluidigm: 89Y-anti-CD45, 175Lu-anti-CD4, 141Pr-anti-PD1, 143Nd-anti-CD11b, 144Nd-anti-Siglec F, 145Nd-anti-CD69, 146Nd-anti-CD206, 148Nd-anti-Tbet, 114Sn-anti-CD103, 151Eu-anti-CD68, 152Gd-anti-CD3e, 156Gd-anti-CD14, 159 Tb-anti-F4/80, 160Dy-anti-CD62L, 161Dy-anti-Ki67, 162Dy-anti-Ly-6C, 165Ho-anti-Foxp3, 149Sm-anti-CD19, 167Er-anti-GATA3, 142Ce-anti-NK1.1, 168Er-anti-CD8a, 172Yb-anti-CD86, 173Yb-anti-CD117, 174Yb-anti-ly-6G/Ly-6C(Gr-1), 209Bi-anti-I-A/I-E, 150Sm-anti-CD11c, 169Tm-CTLA4.

Standard IF staining was performed as described previously [36]. The primary antibodies used for IF staining includes anti-granzyme B (ab4059, Abcam, 1: 800), and anti-CD8 (ab22378, Abcam, 1: 200); DyLight 488- or DyLight 594-conjugated secondary antibodies against rabbit or mouse IgG were purchased from Thermo Fisher Scientific.

Flow cytometric data was analyzed using FlowJo software (v10; Tree Star). GraphPad Prism (v8) was used to generate graphs and for statistical analysis. Student’s t test and ANOVA were used to determine statistical significance. P-value < 0.05 was considered statistically significant. All the above analysis methods and R package were implemented by R foundation for statistical computing (2020) version 4.0.3 and software packages ggplot2 and pheatmap. Sanguini diagram was built based on R software package ggalluval. All the above analysis methods and R package were implemented by R foundation for statistical computing (2019) version 4.0.3. P < 0.05.

GATA3 conducts as a master transcription factor for the differentiation of Th2 cells to activate Th2 cytokine expression in IL-4 dependent or independent pathway [37], and has been used to define Th2 cell population [38].To investigate the correlation between Th2 cells and tumor-immune landscape, we first divided the METABRIC samples into Th2-high and Th2-low groups based on GATA3 expression, and delineated the pattern of TILs in the two groups based on gene expression profile with CIBERSORT algorithm. Compared with Th2-high group, the proportions of antitumor cell population including CD8⁺ T cells, activated CD4⁺ memory T cells, M0/M1 macrophages, follicular helper T cells, and activated natural killer (NK) cells significantly increased in Th2-low group. In contrast, the proportions of suppressive immune cells including regulatory T cells (Tregs) and M2 macrophages were noticeably decreased in Th2-low group (Fig. 1A). Similarly, it also showed that antitumor immune cells such as B cells, CD8⁺ T cells, CD4⁺ effector T cells and dendritic cells were enriched in Th2-low group based on TCGA breast cancer database (Additional file 1: Fig. S1A). Furthermore, we classified breast cancer TME using curated list of Fges, which characterized the presence of immune-active or immunosuppressive microenvironment and tumor stroma. Their expression patterns were evaluated across breast cancer (METABRIC) based on ssGSEA scores. Our analysis revealed that breast cancer patients were clustered into Th2-high and Th2-low groups, which had varied significantly in immune landscape based on the expression of Fges. The Th2-low group demonstrated active immune microenvironment compared with Th2-high group (Fig. 1B). Next, we found that Th2 cells were negatively correlated with CTL infiltration in multiple breast cancer datasets based on TIDE system (Fig. 1C). To better understand the correlation among Th2 cell and CTL function, we calculated 15 genes signature expression of CTLs. The result showed that Th2-high versus Th2-low group had a different expression pattern of genes which were involved in CTLs function on METABRIC database. Clearly, GATA3 was negatively correlated with CTL signature gene (IL2, CD8A, CCL5, GZMA, PRF1, IFNG, PTPRC, GZMM, RAB27A, XCL1, ICOS, TBX21, GZMB, GNLY, and IL12A) in breast cancer (Fig. 1D). Importantly, based on TIDE system, we found that the breast cancer patients with high CTL infiltration (CTL-Top) had prolonged survival compared with low CTL infiltration (CTL-Bottom) in Th2-low group, which was not recapitulated in Th2-high group (Fig. 1E). This observation suggested that Th2 cells had positive association with T cell dysfunction, which was further supported by TCGA database (Additional file 1: Fig. S1B). Taken together, all these data demonstrated that Th2 cell was closely correlated with immunosuppressive immune landscape in breast cancer.

We next assessed whether inhibiting Th2-mediated immunity affected breast cancer progression using the Th2 cytokine inhibitor IPD. Firstly, the effect of IPD on Th1 and Th2 differentiation in the tumors of three syngeneic mouse breast cancer models (EO771, 4T1, and EMT6 cell lines) was detected by flow cytometry. We noticed that IPD markedly increased the infiltration of total CD4⁺ T cells in tumor tissues (Fig. 2A and Additional file 2: Fig. S2A). To be noted, IPD significantly enhanced Th1 (IFN-γ⁺CD4⁺) cell infiltration and inhibited Th2 (IL-4⁺CD4⁺) cell infiltration, indicating IPD promoted severe shift from Th2 to Th1 (Fig. 2B and Additional file 2: Fig. S2B-D). The similar results were also confirmed by CyTOF (Fig. 2C, D). Furthermore, the alteration of Th2 cells in peripheral blood was also investigated. Consistently, flow cytometric analysis showed that IPD dramatically reduced circulating Th2 subpopulation (Fig. 2E and Additional file 2: Fig. S2E). Collectively, these data supported that IPD indeed inhibited Th2 cell infiltration in breast cancer. Next, we evaluated the effect of IPD on breast cancer growth in vivo. We found IPD significantly suppressed tumor growth in the three immunocompetent mouse models (Fig. 2F and Additional file 2: Fig. S2F). However, cell proliferation assay suggested that the tumor cell proliferation was not affected by IPD in vitro (Additional file 3: Fig. S3A, B), while transwell migration assay also showed no significant change about metastatic potential of tumor cells after IPD treatment (Additional file 3: Fig. S3C). These observations indicated that IPD did not directly affect tumor cells and might exert the antitumor efficacy by regulating tumor immunity.

Given that CD8⁺ T cells play a central role in mediating antitumor immunity, we next analyzed the effect of IPD on the infiltration and cytotoxic activities of CD8⁺ T cells [39, 40]. It is demonstrated that CD8⁺ T cell infiltration in breast cancer tissues and circulating CD8⁺ T cells significantly increased after IPD treatment (Fig. 3A and Additional file 4: Fig. S4A, B). The release of effector cytokines including IFN-γ and TNF-α from CD8⁺ T cells was also enhanced by IPD in tumors (Fig. 3B, C and Additional file 4: Fig. S4C, D). The above findings were further validated by multiplex IF staining which showed more CD8⁺ T cell infiltration and granzyme B release in IPD-treated group than in vehicle-treated group (Fig. 3D). Furthermore, the infiltrating CD8⁺ T cells displayed higher ki67 expression in IPD-treated group compared with vehicle-treated group, indicating that CD8⁺ T cells had been activated to exert antitumor activities in response to IPD treatment (Fig. 3E, F and Additional file 4: fig. S4E). T cell exhaustion is a state of T cell dysfunction that arises during cancer [41]. We found that the percentage of exhausted CD8⁺ T cells (PD-1⁺TIM3⁺) was lower in IPD-treated group compared to vehicle-treated group as determined by flow cytometry analysis, suggesting IPD effectively reversed the exhausted state of CD8⁺ T cells (Fig. 3G, H and Additional file 4: fig. S4F). To summarize, these findings suggested that IPD treatment markedly improved the antitumor efficacy of CD8⁺ T cells in breast cancer.

To deeply elucidate the effect of IPD on tumor immunity, we next investigated the alteration of tumor immune landscape after IPD treatment by flow cytometry. We found that immune-suppressive cell population MDSCs was significantly decreased in IPD-treated groups of three mouse breast cancer models, compared with vehicle-treated groups. Additionally, the infiltration of TAMs and Tregs was also markedly inhibited by IPD in EO771 and EMT6 models (Fig. 4A, B and Additional file 5: fig. S5A). To evaluate the general impact of IPD on the tumor immune microenvironment, we profiled CD45⁺ immune cells isolated from EO771 tumors grown in vehicle- and IPD-treated mice using CyTOF, which revealed 13 distinct cell clusters (Fig. 4C–E, and Additional file 5: fig. S5B–D). EO771 tumors from IPD-treated mice had significantly increased CD8⁺ T cells (cluster 1), effector CD4⁺ T cells (cluster 2) and B cells (cluster 5) and fewer M2-like macrophages (cluster 8), suggesting enhanced antitumor immunity by IPD. Th2 cells have been shown to induce the differentiation of M2 macrophages through the release of IL-4 [42]. The expression ratio of major histocompatibility complex class II (MHCII, an M1 marker) versus CD206 (an M2 marker) was used to evaluate M1/M2 polarization status of macrophages, which generally represents antitumor versus protumor activities of TAMs [26] We validated that the MHCII:CD206 ratio of TAMs was increased in IPD-treated groups by CyTOF and flow cytometry, indicating IPD could induce M2 to M1 polarization of TAMs (Fig. 4F–H and Additional file 5: fig. S5E).

To test if inhibition of Th2-mediated immunity would enhance antitumor activity of ICB, the combinatorial treatments of IPD plus ICB (anti-PD-1 or anti-CTLA-4) were used in EO771, 4T1 and EMT6 models. Single treatment with IPD or ICB therapy delayed tumor growth, and combinatorial treatment more effectively inhibited tumor growth in all three mouse models (Fig. 5A and Additional file 6: fig. S6A). Moreover, compared with anti-PD1 treatment, the combination therapy dramatically prolonged mice survival in 4T1 tumor model (Fig. 5B). 4T1 lung metastases were also significantly less in combinatorial treatment group, compared with other groups, indicating that inhibiting Th2-mediated immunity could enhance anti-metastasis immune response in vivo (Fig. 5C). Consistently, further flow cytometry analysis showed that the combination of IPD and ICB significantly enhanced the infiltration and antitumor activity of CD8⁺ T cells, compared with ICB or IPD alone (Fig. 5D-F and Additional file 6: fig. S6B, C). Total CD4⁺ T cell was further increased and immunosuppressive cells including MDSC and TAM were further decreased after the combination treatment (Additional file 6: fig. S6D-F), suggesting IPD treatment combined with ICB drastically reversed the immunosuppressive microenvironment (Fig. 6).

Considering adverse effects may cause intolerance in therapy procedures and negatively affect its clinical efficacy, we next evaluated the adverse effects of IPD in the mouse models [43, 44]. The H&E staining showed that solid organ injury of breast cancer was not enhanced after IPD treatment. Compared with ICB therapy, IPD would not increase the immune-related adverse events associated with anti-CTLA-4 in combination therapy (Additional file 7: fig. S7A). Moreover, compared with vehicle-treated group, IPD treatment did not exacerbate the liver damage and impair renal and cardiac functions, and had no haematotoxin effect and bone marrow suppression based on blood test (Additional file 7: Fig. S7B, C). Together, our results implied that related adverse event was no presented after administration of IPD in vivo.