SOCS3 Suppression Promoted the Recruitment of CD11b+Gr-1−F4/80−MHCII− Early-Stage Myeloid-Derived Suppressor Cells and Accelerated Interleukin-6-Related Tumor Invasion via Affecting Myeloid Differentiation in Breast Cancer

Interleukin-6 (IL-6) is an important trigger for the expansion and recruitment of myeloid-derived suppressor cells (MDSCs), which are regarded to be major coordinators of the immunosuppressive tumor microenvironment. In this study, we constructed IL-6-knockdown breast cancer mice models to explore the molecular events involved in the IL-6-mediated effects on MDSC development. We defined a subset of early-stage MDSCs (e-MDSCs) with the phenotype of CD11b+Gr-1−F4/80−MHCII− in IL-6 high-expressing 4T1 mice mammary carcinoma models, which were the precursors of CD11b+Gr-1+ conventional MDSCs. Furthermore, sustained suppression of SOCS3 and aberrant hyperactivation of the JAK/STAT signaling pathway was exclusively detected in wide-type 4T1 tumor-bearing mice, which promoted the accumulation of e-MDSCs in situ and their immunosuppressive capability in vitro. After blocking the IL-6/STAT3 signaling pathway with the IL-6 receptor antibody or STAT3 antagonist JSI-124 in tumor-bearing mice, significant shrinkage of primary tumors and decrease in lung metastatic nodules were observed in vivo, accompanied by the dramatic decrease of e-MDSC recruitment and recovery of anti-tumor T cell immunity. Thus, SOCS3 suppression accelerated the IL-6-mediated growth and metastasis of mammary carcinoma via affecting myeloid differentiation in breast cancer. Moreover, the IL-6/STAT3 signaling pathway might be a promising candidate target in developing novel therapeutic strategies to eliminate e-MDSCs and improve breast cancer prognosis.

Interleukin-6 (IL-6) is an important trigger for the expansion and recruitment of myeloid-derived suppressor cells (MDSCs), which are regarded to be major coordinators of the immunosuppressive tumor microenvironment. In this study, we constructed IL-6-knockdown breast cancer mice models to explore the molecular events involved in the IL-6-mediated effects on MDSC development. We defined a subset of early-stage MDSCs (e-MDSCs) with the phenotype of CD11b + Gr-1 − F4/80 − MHCII − in IL-6 high-expressing 4T1 mice mammary carcinoma models, which were the precursors of CD11b + Gr-1 + conventional MDSCs. Furthermore, sustained suppression of SOCS3 and aberrant hyperactivation of the JAK/STAT signaling pathway was exclusively detected in wide-type 4T1 tumor-bearing mice, which promoted the accumulation of e-MDSCs in situ and their immunosuppressive capability in vitro. After blocking the IL-6/STAT3 signaling pathway with the IL-6 receptor antibody or STAT3 antagonist JSI-124 in tumor-bearing mice, significant shrinkage of primary tumors and decrease in lung metastatic nodules were observed in vivo, accompanied by the dramatic decrease of e-MDSC recruitment and recovery of anti-tumor T cell immunity. Thus, SOCS3 suppression accelerated the IL-6-mediated growth and metastasis of mammary carcinoma via affecting myeloid differentiation in breast cancer. Moreover, the IL-6/STAT3 signaling pathway might be a promising candidate target in developing novel therapeutic strategies to eliminate e-MDSCs and improve breast cancer prognosis.
In our previous study, we identified a subset of immature MDSCs in human breast cancer tissues with the phenotype of Lin − HLA − DR − CD45 + CD33 + CD13 + CD14 − CD15 − that exerted potent immunosuppressive effects on T cells in vitro and in vivo (7). These primary MDSCs significantly correlated with advanced clinical stage, higher lymph node metastasis, and poor prognosis (7,8), which indicated that these immature MDSCs were representatives of e-MDSCs in breast cancer. Furthermore, we found positive correlation between the level of tumor-derived interleukin-6 (IL-6) and the recruitment of e-MDSCs locally (9). IL-6 potently promoted the amplification of e-MDSCs and their T cell-suppressive capacity in vitro by activating the STAT/IDO signaling pathway and generating a tryptophan-starved microenvironment that facilitated the evasion of breast cancer cells (8,9). Our previous study also demonstrated that tumor-derived IL-6 might play a significant role in the development and accumulation of e-MDSCs in vivo; however, the regulatory mechanisms underlying IL-6-related myeloid differentiation blockage are less understood.
Interleukin-6 is a pleiotropic cytokine with significant functions in the regulation of pro-inflammatory and metastatic tumor microenvironments (10). Accumulating evidences showed signi ficant correlation between IL-6 and MDSC development in human and murine models (11-13). It is well established that IL-6 activates tyrosine kinases Janus kinase 1 (JAK1), Janus kinase 2 (JAK2), and tyrosine kinase 2 (TYK2) via IL-6 receptor (IL-6R) and gp130, which leads to the phosphorylation of signal transducers and activators of transcriptions 1 and 3 (STAT1 and STAT3) (14, 15). IL-6-dependent activation of the JAK/STAT signaling pathway is tightly regulated by members of the suppressor of cytokine signaling (SOCS) protein family (16), and quick feedback of SOCS1/SOCS3 upregulation efficiently inhibits the phosphorylation of STAT3 under physiologic conditions, thereby attenuates the activation of the JAK/STAT signaling pathway and expression of downstream functional genes (17,18).
However, sustained activation of the JAK/STAT signaling pathway was observed in breast cancer e-MDSCs because of significant SOCS3 suppression, which consequently induced the long-term activation of the NF-κB signaling pathway and suppression of T cell immunity (9). STAT3 has been reported to be essential in maintaining a well-differentiated and fully competent immune system (14). Therefore, SOCS3 deficiency-dependent sustained activation of the JAK/STAT signaling pathway might regulate the differentiation of myeloid progenitors. Multiple hemopoietic and immunological defects were also reported in SOCS1/SOCS3-deficient mice as a consequence of prolonged STAT3 activation (19)(20)(21). Croker et al. found that the differentiation of the SOCS3-deficient progenitor cells skewed toward macrophage production due to poor response to G-CSF (22). Furthermore, Yu et al. found that SOCS3 deletion in myeloid cells produced higher levels of CD11b + Gr-1 + MDSCs in prostate tumors (23). Therefore, it will be essential to clarify that if SOCS3 deficiency and sustained activation of the JAK/STAT signaling pathway blocked the differentiation of myeloid progenitors and thus promoted e-MDSC development in breast cancer.
In this study, we constructed IL-6-knockdown 4T1 murine mammary carcinoma-bearing models to study the effects of tumor-derived IL-6 on the development of e-MDSCs in vivo to determine whether SOCS3 deficiency and sustained activation of the JAK/STAT signaling pathway blocked the differentiation of myeloid linkage and promoted the recruitment of e-MDSCs locally. We defined a subset of e-MDSCs with a poorly differentiated phenotype of CD11b + Gr-1 − F4/80 − MHCII − in mice mammary carcinoma, which were the precursors of CD11b + Gr-1 + conventional MDSCs and exerted more potent suppression on T cell immunity. Tumor-derived IL-6 impaired the differentiation of myeloid cells and promoted the accumulation of e-MDSCs by inhibiting SOCS3 expression and persistently activating the JAK/STAT signaling pathway. Moreover, IL-6R blocking antibody and STAT3 antagonist JSI-124 effectively inhibited the growth of primary tumors and distance metastases in lungs while simultaneously reducing the recruitment of e-MDSCs in situ and reversing T cell immunosuppression in vitro. Thus, we concluded that the IL-6-related dysfunction of the SOCS3 feedback loop accelerated the growth and metastasis of mammary carcinoma by affecting myeloid differentiation and attenuating the T cell-based immune surveillance. Furthermore, the IL-6/STAT3 signaling pathway could be a promising candidate target in developing novel therapeutic strategies to eliminate e-MDSCs and improve breast cancer prognosis.

MaTerials anD MeThODs cell lines and Mice
BALB/c female mice were purchased from Beijing Vital River Laboratory Animal Technology, and 6-to 8-week-old mice were used in the experiments. The mouse mammary tumor cell line (4T1) used in this study was purchased from the Chinese Academy of Medical Sciences and cultured in RPMI-1640 medium (Gibco/ BRL, Grand Island, NY, USA) containing 10% fetal bovine serum (FBS) in a humidified 5% CO2 incubator at 37°C. By transfecting wild-type 4T1 cells (4T1 WT ) with IL-6-specific short hairpin RNA (shRNA) lentiviral vectors, 4T1 IL-6low cells were established as indicated in the Supplementary data 1 in Supplementary Material. Nonsense lentiviral shRNA vector-transfected 4T1 WT cell s were established as 4T1 NC controls. All transfected cells were screened using 3 µg/mL of puromycin (Gibco, Grand Island, NY, USA). Stably transfected cells were cloned by limiting dilution and validated using RT-PCR and enzyme-linked immunosorbent assay (ELISA) before expansion. And anti-IL-6R mAb (10 µg/mL; Bio X Cell, USA) was utilized to block IL-6 signaling in 4T1 IL-6low cells in vitro.
construction of 4T1 Breast cancer-Bearing Murine Models Cells (4T1 WT , 4T1 NC , or 4T1 IL-6low ; 1 × 10 6 cells in 200 µL PBS) were injected into the mammary fat pads of BALB/c mice or NOD/ SCID mice and monitored every day for 3 weeks. JSI-124 (1 mg/ kg day; Sigma, USA) or anti-IL-6R mAb (2.5 mg/kg week; Bio X Cell, USA) were administrated intraperitoneally into 4T1 WTbearing BALB/c mice for 13 days. Saline solution-treated mice were used as negative controls. Besides, CD11b + Gr-1 − MDSCs (2 × 10 6 /twice/week) isolated from 4T1 WT tumors were transferred to 4T1 IL-6low -bearing mice. The tumor volume was calculated using the formula V = (π × a × b 2 )/6, where a is the length and b is the width of the tumor. The number of metastatic nodules in the lungs was calculated as previously described (8). The experiment was approved by the Ethics Committee for Animal Experiments at the Tianjin Medical University Cancer Hospital and Institute and was performed in accordance with the Guide for the Care and Use of Laboratory Animals.

isolation and Differentiation of Primary MDscs In Vivo
Primary MDSCs in tumors and spleens were isolated from the single cell suspension of tumor tissues and spleens via magnetic bead enrichment as described previously (12). Briefly, both tumor tissues and spleens were dissociated into single cell suspensions (24). After erythrocytolysis, CD11b + Gr-1 + MDSCs were isolated using beads conjugated with biotin anti-mouse Gr-1 and anti-biotin microbeads (Miltenyi Biotec, Germany), and CD11b + Gr-1 − MDSCs were isolated using anti-mouse CD11b microbeads after CD11b + Gr-1 + MDSCs were removed. CD11b + Gr-1 − F4/80 − MHCII − MDSCs were separated using the BD FACSAria™ II cell sorter (BD Biosciences, San Jose, CA, USA). The viability and purity of the recovered cells were determined using trypan blue staining assay and flow cytometry.
CD11b + Gr-1 − MDSCs isolated from tumors were labeled with CSFE (0.5 µM, Invitrogen, USA) for 20 min and transferred back to female BALB/c mice via tail vein. And 96 h later, spleen single cell suspensions were prepared, and the proportions of CSFE-labeled cells in CD11b + Gr-1 + subset were analyzed using flow cytometry.
T cells, isolated from normal BALB/c mice using the Pan T Cell Isolation Kit (Miltenyi Biotec), were cocultured with either spleen-or tumor-derived MDSCs from tumor-bearing mice at a 1:3 ratio in RPMI-1640 medium supplemented with 10% FBS in 24-well plates. Anti-CD3/CD28 beads (20 μL/10 6 cells; Gibco) were utilized to stimulate T cells in vitro. The proliferation and apoptosis of the T cells were examined using BrdU and Annexin V staining assay as described previously (7). Finally, the supernatants were collected to detect multiple cytokine levels using ELISA.

induction and Differentiation of MDscs In Vitro
Bone marrow (BM) cells were flushed from mice femurs using PBS. Erythrocytolyzed cells were added to multi-well plates and cocultured with 4T1 NC or 4T1 IL-6low breast cancer cells at 1:1 ratio or fresh supernatant of 4T1 NC or 4T1 IL-6low or recombinant IL-6 (rIL-6, 50 ng/mL; PeproTech, USA) to induce MDSCs (iMD-SCs). IL-6R Ab (10 µg/mL) was utilized to block IL-6 signaling pathway. Seventy-two hours later, the percentage of MDSCs was detected by flow cytometry. Besides, CD11b + Gr-1 − iMDSCs and BM cells stimulated by IL-6 (40 ng/mL) for 15 min were collected at different time point. Besides, MDSCs isolated from tumors were cultured in complete RPMI-1640 medium supplemented with 10% FBS and incubated in 5% CO2 at 37°C. GM-CSF (10 ng/mL; PeproTech, USA) was used to induce differentiation in MDSCs in vitro. Seventy-two hours later, the cells were collected to detect the expression of differentiation-related membrane molecules using flow cytometry and RT-PCR.

Quantitative real-Time rT-Pcr
Total cellular RNA was extracted using the Trizol-trichloromethane method, and RNA quantity was determined spectrophotometri cally. Synthesis of cDNA was performed using the TaKaRa PrimeScript ® RT reagent Kit (TaKaRa Bio, Japan). Quantitative real-time PCR was performed using the SYBR Premix Ex Taq TM system (TaKaRa Bio) and primers shown in Table 1. The relative mRNA levels were calculated based on the threshold cycle (Ct) values normalized to the Ct value of β-actin using the following formula: 2 −ΔCt , where ΔCt = Cttarget gene − Ctβ-actin. All tests were performed in triplicate.

Western Blot analysis
Cell lysates were resolved using SDS-PAGE and transferred to polyvinylidene difluoride membranes. The membranes were blocked with 5% BSA and then incubated overnight at 4°C with anti-SOCS1  presented as mean ± SD. One-way analysis of variance and least significant difference tests were used to compare the quantitative data. P values for each analysis are reported in the figure legends. The level of statistical significance was set at P < 0.05. resUlTs il-6 Knockdown repressed the growth and Metastasis of 4T1 Tumors In Vivo The murine mammary carcinoma cell line, 4T1, highly expressed IL-6 (13). Our preliminary study was consistent with previous reports and confirmed high expression levels of IL-6 at both the RNA and protein levels. In this study, the IL-6-knockdown in the 4T1 murine mammary carcinoma cell line (4T1 IL-6low ) was successfully established to investigate the effects of tumor-derived IL-6 on the growth and invasion of breast cancer cells by infecting lentiviral shIL-6 into mammary carcinoma 4T1 cells. Compared with the 4T1 WT and 4T1 NC cells, the 4T1 IL-6low cells expressed lower levels of IL-6 at both the RNA and protein levels (P = 0.0002, P = 0.0001; P < 0.001, P = 0.0002, Figure 1A). Furthermore, the 4T1 WT , 4T1 NC , and 4T1 IL-6low cells were implanted into BALB/c mice to generate mammary carcinoma-bearing models. We found that IL-6 knockdown strikingly inhibited tumor growth (P < 0.001, P < 0.001, Figure 1B). After the mice were sacrificed, the volume of primary tumors and number of lung metastatic nodules were calculated. The average volume of the 4T1 IL-6low tumors was lower than that of both 4T1 WT and 4T1 NC cells (585.2 ± 78.6 vs. 1,080.6 ± 115.7 vs. 1,122.6 ± 132.9 mm 3 , P = 0.0036, P = 0.0038, Figure 1C). Furthermore, the number of metastatic nodules in the lungs was also dramatically reduced in 4T1 IL-6low group than both the 4T1 WT and 4T1 NC groups (56.0 ± 6.4 vs. 151.0 ± 6.1 vs. 154.0 ± 14.3, P = 0.0004, P = 0.0033, Figures 1D,E). Furthermore, the 4T1 WT , 4T1 NC , and 4T1 IL-6low cells were implanted into NOD/SCID mice separately, and the results showed that IL-6 knockdown inhibited tumor growth to some extent in T-and B-cell immunity-deficient mice (P = 0.001). But less tumor suppression in NOD/SCID mice was observed compared with that in immunity competent mice which implied that except T and B cells, other immune cells, such as NK cells,

il-6 Knockdown Did not affect the growth and invasion of 4T1 cells In Vitro
To determine whether IL-6 promoted tumor growth and metastasis by directly affecting the biological behavior of 4T1 cells, we examined the effects of IL-6 knockdown on the proliferation, apoptosis, migration, and invasion of 4T1 cells in vitro. These cells normally exhibited spindle-like features, and no morphological changes were observed after IL-6 knockdown (Figure 2A).
We compared the proliferation of 4T1 WT , 4T1 NC , and 4T1 IL-6low cells using the CCK-8 staining assay and found no difference in their proliferation rates in vitro (P = 0.773, P = 0.461, Figure 2B). Furthermore, no disparity in colony numbers was observed between these three groups in the colony formation assay, which indicated that there was no detectable disparity in the colony formation capacity of 4T1 cells after IL-6 knockdown (P = 0.5686, P = 0.8369, Figure 2C). We also found that IL-6 knockdown had no effect on cell apoptosis in the Annexin V assay (P = 0.7780, P = 1.00, Figure 2D). Furthermore, cell migration and invasion of the 4T1 WT , 4T1 NC , and 4T1 IL-6low cells were compared using the wound healing and transwell assays, which demonstrated a similar migration pattern (P = 0.7953, P = 0.6537, Figure 2E) and invasive capacity (P = 0.8892, P = 0.7076, Figure 2F). These results implied that IL-6 knockdown did not affect the growth and invasion of 4T1 cells in vitro.
We also detected the IL-6 signaling pathway in 4T1 WT , 4T1 NC , and 4T1 IL-6low cells. And the results showed that JAK1, JAK2, TYK2, and STAT3 phosphorylation was observed, and no significant disparity was found among three groups ( Figure 2G). Besides, we detected CD126 and gp130 in three cells. And the results showed that there was no significant difference in CD126 and gp130 expression both at mRNA and protein level (Figures 2G,H). Furthermore, IL-6 signaling pathway was further blocked using IL-6R antibody (10 µg/mL) in 4T1 IL-6low cells. The results showed that IL-6R Ab inhibited 4T1 IL-6low cells growth (P < 0.001) and colony formation (P = 0.0025) but had no effect on apoptosis (P = 0.9106). Besides, cell migration (P = 0.0345, P = 0.0024) and invasion of the 4T1 IL-6low cells (P = 0.0084) were significantly suppressed by IL-6R Ab (Supplemental data 2 in Supplementary Material). Above results implied that merely reducing the level of IL-6 did not affect the downstream JAK/STAT signaling pathway, but fully blocking the interaction between IL-6 and IL-6R was sufficient to inhibit the growth and invasion of 4T1 cells in vitro.
il-6 Promoted the accumulation of cD11b

-MDscs in Tumors
In our previous study, we demonstrated that tumor-derived IL-6 positively correlated with MDSC infiltration in situ in human primary breast cancer tissues, which correlated with more aggressive tumor phenotypes and worse clinical outcomes (9). Therefore, we compared the relationship between IL-6 levels and MDSCs infiltration in primary 4T1 tumors, along with the size and number of tumors and metastatic lung nodules. Flow cytometry analysis revealed that CD11b + Gr-1 + MDSCs were higher in the tumor tissues and spleens of tumor-bearing mice when compared with the same in normal controls (44.13 ± 1.00 vs. 39.03 ± 2.38 vs. 6.03 ± 0.79%, P = 0.0002, P < 0.001). However, a significant increase in CD11b + Gr-1 − cells was exclusively observed in the tumor tissues and spleens of tumorbearing mice when compared with the same in normal controls (45.13 ± 0.85 vs. 7.20 ± 0.30 vs. 4.17 ± 0.30%, P < 0.0001, P = 0.0019) ( Figure 3A).
Furthermore, fewer CD3 + T cells infiltrated the 4T1 WT and 4T1 NC tumor tissues when compared with the 4T1 IL-6low tumors (81.50 ± 10.11 vs. 80.25 ± 10.43 vs. 165.0 ± 12.91, P = 0.0022, P = 0.0022, Figure 3I). To determine the T cell function in vivo, BrdU (50 mg/kg) was injected by tail vein separately to mice bearing 4T1 WT , 4T1 NC , or 4T1 IL-6low tumors, and proliferative T cells labeled with BrdU in tumors and spleens were detected. T cells in spleen of normal mice were used as control. The results showed that T cell proliferation was suppressed dramatically in 4T1 WT and 4T1 NC tumors compared with normal controls (12.50 ± 1.33 vs. 11.03 ± 0.73 vs. 29.53 ± 2.74%, P = 0.0050, P = 0.0029). But the proliferation of T cells in 4T1 IL-6low tumors was recovered compared with that in 4T1 NC tumors (18.30 ± 1.91 vs. 11.03 ± 0.73, P = 0.0236), which implied that IL-6 reduction in tumor microenvironment can relieve immunosuppression on T cell proliferation and function. Meanwhile, the proliferation of T cells in spleens was also detected. And the results showed that the proliferation of T cells in 4T1 WT and 4T1 NC spleens was dramatically inhibited compared with normal controls (14.30 ± 1.51 vs. 15.00 ± 0.93 vs. 29.53 ± 2.74%, P = 0.0082, P = 0.0074, Figure 3J). But IL-6 knockdown has less effect on T cell proliferation of spleens compared with that in 4T1 NC spleens (15.80 ± 1.44 vs. 15.00 ± 0.93%, P = 0.6649, Figure 3J). Above data suggested that tumor-derived IL-6 mainly affected the immune state in tumors, rather than the systematic environment.
To determine if CD11b + Gr-1 − cells could switch to CD11b + Gr-1 + MDSCs in vivo, CD11b + Gr-1 − cells were isolated from tumors and labeled with CSFE (0.5 µM) for 20 min in vitro. Then, the labeled cells were transferred back to normal BALB/c mice. Seventy-two hours later, spleen single cell suspensions were prepared and the proportions of CSFE-labeled cells in CD11b + Gr-1 + subset were analyzed using flow cytometry. The results showed that 3.0% of CD11b + Gr-1 + cells was labeled with CSFE. Furthermore, CD11b + Gr-1 + cells transformed from CD11b + Gr-1 − cells were rich in gated P4, since 71.3% of cells in gated P4 was CSFE-labeled. All above data indicated that transferred exogenous CD11b + Gr-1 − cells were capable of inducing CD11b + Gr-1 + cells in vivo ( Figure 4B).

sOcs3 suppression and sustained activation of the JaK/sTaT signaling Pathway in e-MDscs is il-6 Dependent
We assessed the expression and phosphorylation of multiple functional proteins induced by IL-6, including JAK/STAT pathway and MAPK signaling pathway using Western blotting. A comparable increase in p-JAK1, p-JAK2, p-TYK2, p-STAT1, and p-STAT3 was detected in tMDSC WT when compared with normal controls (Figure 6A). However, the levels of the phosphorylated proteins mentioned above were reduced significantly in tMDSC IL-6low when compared with those in tMDSC NC (Figure 6A), which implied that the enhanced phosphorylation of the STAT proteins in tMDSC WT is IL-6 dependent. The results about MAPK pathway showed that phosphorylated P38 mitogen-activated protein kinases (p-P38), extracellular regulated protein kinases (p-ERK), and c-Jun N-terminal kinase (p-JNK) increased significantly, though these changes have no relation to IL-6 levels ( Figure 6B). All above data showed that the JAK/STAT signaling pathway participated in IL-6-related e-MDSCs accumulation, rather than the MAPK signaling pathway.
Next, we compared the expression of SOCS1-3 in tMDSC WT and normal controls at the mRNA and protein levels since the deficiency of the SOCS proteins has been reported to induce an aberrantly sustained activation of the JAK/STAT signaling pathway in human breast cancer e-MDSCs (9). The results showed that the SOCS1 and SOCS2 mRNA levels increased, but the SOCS3 mRNA level significantly decreased in tMDSC WT when compared with the levels in normal controls. Moreover, the protein levels of SOCS1 decreased and SOCS3 increased in tMDSC IL-6low when compared with the levels in tMDSC NC (Figure 6C). Similar expression patterns of both SOCS1 and SOCS3 were confirmed at the mRNA level ( Figure 6D). Therefore, these results implied that tumor-derived IL-6 plays a significant role in the suppression of SOCS3 both at the mRNA and protein levels in tMDSCs, which consequently triggered the sustained activation of the JAK/STAT signaling pathway.
To identify the effect of SOCS3, BM cells were treated with 4T1 NC and 4T1 IL-6low separately to induce iMDSCs NC and iMDSCs IL-6low . Untreated BM cells were exposed to IL-6 (40 ng/ mL) for 15 min, and the activation status of the JAK/STAT pathway downstream of IL-6 signaling was detected as control at different time point. In IL-6-stimulated BM cells, the levels of phosphorylated STAT1 and STAT3 proteins were increased at 15 min but decreased at 30 min, disappearing entirely at 2 h. Sustained phosphorylation of STAT1 and STAT3 proteins was observed in iMDSCs NC , which was maintained for a longer time than in normal IL-6-stimulated BM cells (2 vs. 8 h). Meanwhile, the expression of SOCS3 and STAT3 phosphorylation was parallel in IL-6-stimulated BM cells. But in iMDSCs NC , SOCS3 expression was decreased significantly and did not response to IL-6 stimulation, which lead to sustained activation of STAT3. In iMDSCs IL-6low , phosphorylation levels of STAT1 and STAT3 were reduced at 1 h, and SOCS3 changed along with p-STAT3 ( Figure 6E). All above data showed that SOCS3 suppression and sustained activation of STAT3 occurred in e-MDSCs, which were correlated with tumor-derived IL-6. Therefore, the shortterm IL-6 simulation induced SOCS3, and SOCS3 in a feedback mechanism shows a negative effect on IL-6 signaling. But long-term IL-6 stimulation in local tumor microenvironment executed significant inhibitory effect on SOCS3 in myeloid cells and thus induced sustained activation of the JAK/STAT pathway.
We isolated primary e-MDSCs from the IL-6R Ab-and JSI-124-treated mice and found that the p-STAT3 protein levels in both groups significantly decreased when compared with the cells isolated from untreated controls ( Figure 7D). Moreover, the mRNA and protein levels of SOCS1 decreased while those of SOCS3 increased in IL-6R Ab-treated e-MDSCs. However, no significant difference in the expressions of SOCS1 and SOCS3 was observed in JSI-124-treated e-MDSCs (Figures 7D,E). Furthermore, the STAT and SOCS expression in e-MDSCs were also determined using intracellular flow cytometry, and mean fluorescence intensity was utilized to present the expression quantity. The results were consistent the WB results ( Figure 7F). Above data implied that tumor-derived IL-6 predominantly triggered SOCS3 suppression in e-MDSCs, and hence it could be reversed by using the IL-6R blocking antibody.

DiscUssiOn
Recent studies have demonstrated that the IL-6 plays an important role in tumor progression and metastasis (28) and is expressed in approximately 50% of breast cancers. IL-6 also correlated with poor prognosis and advanced disease in breast cancer patients (29). But the direct effect of IL-6 on breast cancer cell is not fully understood, and recent studies were paradoxical on IL-6 dependent effect on the proliferation and invasion of breast cancer cells. Studies showed both direct growth inhibitory effects (30)(31)(32), while other studies show growth promoting effects (33,34) or no effects at all (35,36). Asgeirsson et al. have demonstrated that IL-6 decreased cell adhesion of breast cancer cell lines T47-D, ZR-75-1, and MCF-7 to promote metastasis, but the effect was not observed in MDA-MB-231 (37). Ectopic stable IL-6 expressing MCF-7 breast adenocarcinoma cells exhibited an EMT phenotype characterized by impaired E-cadherin expression and induction of Vimentin, N-cadherin, Snail, and Twist (38). And researchers proposed that a pattern emerges from above studies, showing that ER-positive breast cancer cells are receptive to exogenous IL-6, while ER-negative breast cancer cells are mostly unresponsive, probably due to a high autocrine production of IL-6 in ER-negative breast cancer cells (39). And in present study, our results showed that IL-6 knockdown has no effect on 4T1 cells. And merely reducing the level of IL-6, rather than fully blocking the interaction between IL-6 and IL-6R, could not affect the downstream JAK/STAT signaling pathway and the proliferation and invasion of 4T1 cells in vitro. However, IL-6 knockdown exclusively displayed significant inhibition on 4T1 tumor growth and metastasis in vivo which indicated that microenvironment plays a vital role in breast cancer development and progression. Recent evidence showed that IL-6 played critical roles in cancer development and progression by regulating the tumor microenvironment (40), which threw new light on the molecular mechanism regulating the IL-6-induced growth and metastasis of mammary carcinoma in vivo. Several studies have confirmed the relationship between tumor-derived IL-6 and MDSCs in a wide variety of tumor types such as esophageal cancer, prostate cancer, and hepatocellular carcinoma (11,41,42). In breast cancers, IL-6-expressing 4T1 mammary tumor-bearing mice were reported to dramatically enhance the recruitment of CD11b + Gr-1 + MDSCs in the spleen, primary tumor mass, and metastasizing organs when compared with low IL-6-expressing EMT6 mice (13). Consistently, we observed the enhanced accumulation of CD11b + Gr-1 + MDSCs in the spleens and tumors of 4T1 WT mice in our study. Furthermore, we found a significant increase in the immature MDSC subset with the CD11b + Gr-1 − F4/80 − MHC-II − phenotype in the primary tumors of the 4T1 WT mice, which were regulated by tumor-derived IL-6. These immature MDSCs displayed a higher positive correlation with primary tumor size and the number of lung metastatic nodules in tumor-bearing mice than CD11b + Gr-1 + MDSCs, which is a major MDSC subset in primary tumors and more responsible for IL-6-stimulated tumor growth and metastasis.
In this study, we found that tumor cells blocked the differentiation of myeloid progenitors in CD11b + Gr-1 − e-MDSCs, but their capacity to differentiate was restored after a deprivation of tumor-derived factors. Consistently, a previous study reported that PMN-MDSCs could resemble neutrophils phenotypically and functionally after being stimulated by GM-CSF in vitro (43). After culturing tumor-derived MDSCs in the absence of tumor-derived factors or transferring MDSCs to tumor-free recipients, more mature macrophages and DCs were generated (44,45). All these reports showed that tumor-derived factors abrogated the differentiation of myeloid cells. Furthermore, we induced e-MDSCs by coculturing murine BM cells with 4T1 mammary cancer cells in vitro and found that the differentiation toward mature F4/80 + or MHCII + cells was obviously blocked upon GM-CSF stimulation. Blocking the IL-6 signaling pathway reversed the differentiation block and promoted the generation of F4/80 + or MHC-II + mature myeloid cells. A similar phenomenon was observed in primary tumors in which 70% of CD11b + Gr-1 − MDSCs were F4/80 − and MHC-II − immature myeloid cells. Knockdown tumor-derived IL-6 increased the percentage of mature myeloid cells significantly. All these findings suggested that tumor-derived IL-6 was the critical factor that blocked myeloid differentiation and the development of e-MDSCs.
It has been reported that the status of differentiation of myeloid progenitors partly influenced the immunosuppressive capacity of MDSCs. Pu et al. identified that both granulocyte/ macrophage progenitors and common myeloid progenitors, freshly isolated from tumor-bearing animals, were capable of inhibiting polyclonal stimuli-and alloantigen-induced T cell proliferation. Strikingly, early-stage myeloid progenitors displayed much stronger suppressive capacities than conventional MDSCs (46). Zhou et al. reported that CD115 + Ly6C − MDSCs were more immature and showed higher suppressive activities than CD115 + Ly6C + MDSCs (47). Another previous study showed that MDSCs with higher Gr-1 expression were more mature, CD11b + Gr-1 low cells were the most immunosuppressive, CD11b + Gr-1 int cells were less immunosuppressive, and CD11b + Gr-1 high cells (mostly granulocytes) were the least immunosuppressive (48). Consistent with these studies, we found that CD11b + Gr-1 − e-MDSCs isolated from primary 4T1 tumors were more immature but exerted more potent immunosuppressive capacities than conventional CD11b + Gr-1 + MDSCs. The robust suppression of T cells was recovered when IL-6R Ab or JSI-124 was administrated, which implied that tumor-derived IL-6 was the key trigger that regulated the development of competent e-MDSCs.
Interleukin-6 is a well-known trigger that activates the JAK/STAT signaling pathway, which plays crucial roles in the amplification and function of MDSCs in multiple tumors (49,50). Interestingly, we found an aberrantly activated JAK/STAT signaling pathway during the IL-6-initiated development of the CD11b + Gr-1 − e-MDSCs in this study, which demonstrated significant upregulation of the phosphorylated STAT1, STAT3, JAK1, JAK2, and TYK2 proteins. After inhibiting the phosphorylation of STAT3 via the specific antagonist JSI-124, a considerable decrease in the suppression of T cell immunity was confirmed. A consistent phenomenon was observed in human breast cancer in which the sustained activation of the JAK/STAT signaling pathway dominated the amplification and immunosuppressive capacity of the Lin − HLA-DR − CD45 + CD13 + CD33 + CD14 − CD15 − e-MDSCs (8, 9). The results in both human and mice implied that tumor-derived IL-6 involved in the hyperactivation of the JAK/STAT signaling pathway, which was the main cause of the development of competent e-MDSCs in breast cancer.
The studies about MDSCs in prostate cancer demonstrated that SOCS3 negatively regulated the development and function of MDSCs by inhibiting STAT3 activation. Moreover, SOCS3deficient mice elevated Gr-1 + CD11b + MDSCs in tumors and exhibited heightened STAT3 activation (23). Besides, we previously demonstrated that significant suppression of SOCS3 in human e-MDSCs induced persistent activation of the JAK/ STAT signaling pathway and expression of downstream functional genes, such as IDO, IL-10, and TGF-β, which produce immunosuppressive microenvironments locally (8,9). In this study, significant suppression of SOCS3, both at the mRNA and protein levels, was identified in primary e-MDSCs isolated from 4T1 xenografts and induced e-MDSCs by coculturing murine BM cells with 4T1 cells. All above results suggested that SOCS3 plays critical role in MDSC development. And in this study, IL-6R Ab effectively reversed SOCS3 suppression, recovered T cell immunity, and abolished IL-6-related differentiation block of myeloid cells in vitro and vivo. Consistently, chemokine receptor 5 (CCR5) blockade inhibited IL-6-STAT3 pathway via SOCS3, and anti-CCR5 antibody treatment could inhibit the growth of tumor via upregulated SOCS3 to decrease MDSCs accumulation and immunosuppressive capacity in vivo (51). Therefore, it may be a practical scheme to upregulate the SOCS3 expression to inhibit MDSCs accumulation and reverse immunosuppression.
This study showed that IL-6 alone showed less effect on e-MDSCs development compared with 4T1 supernatant or 4T1 cells. The remarkable difference between IL-6-and high IL-6-expressing tumor microenvironment suggested that it was the IL-6-related tumor-derived factors, rather than IL-6 directly, that influenced e-MDSCs development. Besides, IL-6 was strong inducer of SOCS3 expression via STAT3 phosphorylation (52). But surprisingly, we found that e-MDSCs exposed in high IL-6 expressing tumor microenvironment presented SOCS3 suppression and sustained activation of JAK/STAT, while IL-6 short stimulation caused temporary SOCS3 regulation and JAK/STAT activation. The quite the opposite effect of rIL-6 and tumor microenvironment with high IL-6 on both e-MDSCs accumulation and SOCS3 expression also suggested that IL-6-related tumor-derived factors, but not IL-6 directly influenced SOCS3 suppression. It has been reported that promoter methylation and microRNAs were two main mechanisms of SOCS3 suppression. The CpG island in the human SOCS3 genomic sequence was identified, which encompasses the promoter and coding sequence of human SOCS3 gene (53). And hypermethylation of SOCS3 promoter region was observed in H460 cells (54). miR-NAs are about 23 nucleotides long, single-stranded molecules and regulate gene expression controlling various biological processes. Researchers reported that many miRNAs participated in the regulation of SOCS3 expression, such as miRNA-127, miRNA-124a, miR-122, and miR-483-5p (55)(56)(57)(58). It is reported that increased miR-30a-mediated SOCS3 decrease was involved in MDSC function and differentiation, and miR-30a antagomir could inhibit the growth of tumor via upregulated SOCS3 to decrease MDSCs accumulation and immunosuppressive capacity in vivo (59). Therefore, further study on the transcriptional and post-transcriptional regulatory mechanisms involved in IL-6related SOCS3 suppression will be conducted.
The fact that MDSCs play an important role in the regulation of tumor growth has stimulated the search for potential target therapeutic strategies (60). In this study, we confirmed the efficacy of target therapy against IL-6 signaling and STAT3 phosphorylation in the treatment of breast cancer. IL-6R Ab and a specific STAT3 antagonist inhibited the growth and metastasis of IL-6 + mammary cancer in vivo and dramatically reduced the accumulation of e-MDSCs. Recently, tocilizumab, a humanized recombinant mAb against the IL-6R, showed the potential to improve prognosis in IL-6-expressing lung cancer and recurrent epithelial ovarian cancer (61,62). Our finding is consistent with a previous report on MR16-1, a rodent analog of tocilizumab and a commercial humanized recombinant IL-6R mAb, that enhanced anti-tumor activity by eliminating MDSCs (63). STAT3 antagonist JSI-124 suppressed both CD11b + Gr-1 − e-MDSCs and conventional CD11b + Gr-1 + MDSCs simultaneously and induced considerable recovery of T cell immunity in vivo and in vitro, thereby displaying more significant anti-tumor efficiency than IL-6R Ab. Furthermore, another previous study demonstrated that the STAT3 antagonist JSI-124 induced MDSCs to differentiate into immunogenic DCs (64). However, considering that the STAT3 antagonist JSI-124 has more severely toxic adverse effects and is limited in animal experiment. IL-6R Ab might be a more promising therapeutic strategy for breast cancer.
In conclusion, this study provided new insights into investigating the crosstalk between breast cancer cells and regulatory immunocytes in tumor microenvironments. In murine mammary carcinoma, tumor-derived IL-6 impaired the differentiation of myeloid progenitors, promoted the development of CD11b + Gr-1 − F4/80 − MHC-II − e-MDSCs, and potently suppressed T cell immunity, in which IL-6-dependent SOCS3 suppression and the consequential activation of the JAK/STAT signaling pathway are the most crucial molecular events. Targeting the IL-6/STAT3 signaling pathway can significantly restore the differentiation of myeloid progenitors and reverse the immunosuppressive capacity of e-MDSCs. Therefore, blocking the IL-6 signaling pathway might be a promising therapeutic strategy for breast cancer treatment.

eThics sTaTeMenT
The experiment was approved by the Ethics Committee for Animal Experiments at the Tianjin Medical University Cancer Hospital and Institute and was performed in accordance with the Guide for the Care and Use of Laboratory Animals.