Soluble IL6R Expressed by Myeloid Cells Reduces Tumor-Specific Th1 Differentiation and Drives Tumor Progression

The usage of T-cell–mediated cancer immunotherapies has gained momentum for achieving successes in advanced malignancies and is currently being hailed as a promising therapeutic approach (1, 2). While the induction of tumor-specific CD8⁺ T cells has been the main focus in cancer immunotherapy, recent studies have suggested that the better prognosis is correlated with higher levels of IFNγ-expressing CD4⁺ T (Th1) cells in patients with cancer, implying an important role of Th1 cells in controlling tumor regression (3–5). However, Th1 responses tend to be abolished in patients with cancer (3, 5) and tumor-bearing animals (6, 7). An induction of tolerance against immunosurveillance in tumor microenvironment is suggested to be the cause for failure of tumor-specific T cells to eradicate malignant cells (1, 3, 6). It has been proposed that immunosuppression occurs through complex interactions among tumor cells and a variety of immune cells such as T cells, macrophages, dendritic cells (DC), and other myeloid lineage cells (1, 8, 9). Therefore, the identification of pathways involved in these interactions and an understanding of the mechanism underlying Th1 suppression will be beneficial in predicting patient's prognosis and augmenting antitumor responses with immunotherapy.

The role of IL6 signal in promoting tumor cell survival has been demonstrated in vitro and in vivo (10–13). Importantly, a higher level of IL6 was significantly associated with unfavorable prognosis in patients with cancer (10, 14, 15). Canonical IL6 signaling is transmitted through membrane-bound IL6 receptor (IL6R) ligation and gp130 transmembrane receptor dimerization, whereas an alternative way of IL6 signal transmission “IL6 trans-signaling” is mediated by soluble IL6 receptor (sIL6R) that forms a complex with IL6 and directly engages the gp130 even in cells without membrane-bound IL6R (16, 17). The importance of sIL6R is highlighted by the fact that approximately 70% of the secreted IL6 binds the sIL6R in blood (18). The sIL6R functions as a carrier molecule for IL6, thereby markedly prolonging its half-life in vivo, and stabilizing IL6 signaling (17). In addition to IL6, a higher level of sIL6R is associated with a pathologic grade of various cancers (12, 15, 16) and thus may have a diagnostic and prognostic significance in patients with cancer (19). The tumor-promoting effects of IL6, sIL6R, and STAT3 activation, a critical signaling component activated by IL6 (10), demonstrated in mouse models also supports their usefulness as rational therapeutic targets for controlling cancer progression (12, 13). In fact, the chimeric anti-IL6 antibody (Ab) has been used in clinical trials for multiple myeloma, metastatic renal cell carcinoma, prostate cancer, and other cancers (10, 14, 20) with an intention to abrogate its direct effect on tumor growth. A better understanding of the biological effect of IL6/sIL6R in patients with cancer will lead to improvement of their clinical management.

We have previously shown that priming of tumor-specific CD4⁺ T cells in combination with a treatment of anti-IL6 blocking Ab enhanced the antitumor CD4⁺ T-cell responses (7). However, the mechanistic action by which IL6 attenuates the T-cell–mediated antitumor responses remains to be elucidated. In this study, we demonstrated that activation of tumor-specific CD4⁺ T cells did not promote the tumor elimination due to an early skewing into the dysfunctional cells in an IL6/sIL6R-dependent manner. Furthermore, myeloid lineage–specific deletion of IL6R in tumor-bearing mice revealed that myeloid cell–derived sIL6R functioned as a dominant factor in IL6-mediated attenuation of Th1 differentiation, which subsequently dampened their antitumor activity. Our study proposes a possibility that monitoring and manipulating IL6/sIL6R signaling are novel approaches to improve the T-cell–mediated immunity against malignant tumors.

C57BL/6 mice were purchased from Nihon Clea. CD45.1⁺B6.SJL-PtprcaPep3b/BoyJ mice and IL6Rα-flox/flox (IL6R^fl/fl) mice were purchased from The Jackson Laboratory. The mice with ubiquitous deficiency of IL6Rα were generated by crossing of IL6R^fl/fl mice with the mice expressing CAG promoter–driven Cre transgene (21). To generate the IL6R-deficient ovalbumin (OVA)-specific CD4⁺ T cells, CAG-Cre/IL6R^fl/fl mice were crossed with CD45.1⁺ OT-II T-cell receptor (TCR) transgenic mice. Myeloid cell–specific IL6R conditional knockout (KO) mice were generated by crossing the IL6R^fl/fl mice with the mice expressing lysozyme M promoter–driven Cre recombinase (kindly provided by Dr. Irmgard Foester). C3H-background Maf^Ofl mice with a mutant c-Maf allele that lacks the ability to bind to its target genes (22) were provided by the Medical Research Council. Maf^Ofl/+OT-II transgenic mice were generated by backcrossing Maf^Ofl/+mice with CD45.1⁺ OT-II transgenic mice for 10 generations. All the mice including IFNγ-deficient (23) and IL6-deficient mice were housed at the Center for Animal Resources and Development, Kumamoto University (Kumamoto, Japan). All the experimental procedures were approved by the Institutional Animal Committee of Kumamoto University and performed in accordance with the guidelines.

Blood samples were obtained from patients with head and neck malignant tumor (HNT) enrolled in a randomized phase II trial (24) before and after receiving the peptides vaccine. This trial was approved by the Institutional Review Board of Kumamoto University and registered in the University Hospital Medical Information Network Clinical Trials Registry (UMIN-CTR; 000008379). Written informed consents were obtained before enrollment. Healthy donor samples were obtained as well. Peripheral blood mononuclear cells (PBMC) from Ficoll-separated blood and plasma were cryopreserved until experimental use.

Mice were inoculated subcutaneously with 5 × 10⁵ OVA-expressing MCA205, MCA-OVA (7), OVA-expressing EL4 thymoma (EG7), or 2.5 × 10⁴ Rauscher murine leukemia virus (MuLV)-induced lymphoma, RMA (25). Tumor size is expressed as tumor index, which is the square root of (length × width) as described previously (7). In the pulmonary metastatic model, mice were intravenously injected with 5 × 10⁵ OVA/firefly luciferase-expressing melanoma, MO4-Luc (7). To monitor their lung metastasis, luminescent images were analyzed using NightOWL (Berthold Technologies). A total of 250 μg anti-IL6R, anti-IL6 Ab (BioXCell), or control IgG Ab (Millipore) was injected 1 day before and after immunization. For in vivo cell depletion, mice were injected with anti-CD4 Ab (50 μg) before and after immunization or RMA inoculation. A total of 250 ng recombinant sgp130 (R&D Systems) was intravenously injected.

CD4⁺ T cells (8 × 10⁵) isolated from CD45.1⁺ OT-II transgenic mice using naïve CD4⁺ T-cell isolation kit (Miltenyi Biotec) were injected intravenously into tumor-free or tumor-bearing CD45.2⁺ mice. Eight hours later, mice were immunized with an intravenous injection of 4 × 10⁵ bone marrow–derived DC pulsed with peptide ISQAVHAAHAEINEAGR (referred to OVA-IIp) that is recognized by I-A^b-restricted OT-II cells, or with the MuLV Env-gp70 H13.3 peptide SLTPRCNTAWNRL (EnvH13.3) that is recognized by I-A^b-restricted CD4⁺ T cells (25). The peptide SIINFEKL (OVA-Ip) recognized by H2-K^b–restricted OVA-specific CD8⁺ T cells was also utilized for immunization.

Cells from spleen and lymph nodes were stained with the following antibodies for flow cytometric analyses: anti-CD11b, anti-Vβ5, anti-CD4 Abs, and PerCP-streptavidin (BD Biosciences); anti-Ly6C, anti-Gr-1, anti-CD45.1, anti-Ly6G, anti-CXCR3 Abs (eBioscience); anti-human HLA-DR, anti-CD16 (BD Biosciences); anti-CD14, anti-CD33, anti-CD4 Abs (BioLegend); and anti-IL6R Ab (Beckman coulter). The H-2K^b/SIINFEKL-tetramer-PE was from MBL. For cytokine staining, CD4⁺ T cells isolated using anti-CD4 beads (Miltenyi Biotec) or from in vitro cultures were restimulated with OVA-IIp–pulsed DC or PMA/ionomycin and stained with anti-IL2, anti-IL17A, anti-IL10 (eBioscience), or anti-IFNγ (TONBO), anti-TNF-α Abs (Biolegend) as described previously (7). For human samples, anti-human IL2, anti-IFNγ, or anti-TNF-α Abs (BD Biosciences) was used. Staining with anti-phospho-STAT3 Ab (BD Biosciences) was performed as described previously (26). Staining with anti-human/mouse T-bet, anti-RORγt, anti-c-Maf (eBiosciences), or anti-GATA3 Abs (BD Biosciences) was performed using the Transcription Factor Buffer (BD Biosciences). Immunofluorescent images were analyzed using the FACSVerse or FACSCalibur (Becton Dickinson). Data were analyzed using FlowJo software (TreeStar). IL6 and soluble IL6R (sIL6R) levels were measured by ELISA (R&D Systems). For the ELISPOT assay (BD Biosciences), 1 × 10⁵ draining lymph node cells and 3 × 10⁴ DCs pulsed with H13.3, D^b-binding MuLV GagL (LCCLCLTVFL) or K^b-binding Env peptides (SSWDFIT; ref. 25) were mixed and incubated for 12 hours. IFNγ spots were visualized and analyzed as previously described (27).

Mouse naïve T cells were stimulated with plate-coated anti-CD3 and anti-CD28 Abs (both TONBO) in the presence of IL12 (4 ng/mL; Wako) with or without IL6 (Peprotech) and sIL6R (Peprotech). After culture for 6 to 8 days, T cells were analyzed. For human samples, naïve CD4⁺ T cells (5 × 10⁴) were isolated with the naïve CD4⁺ T-cell isolation kit (Miltenyi Biotec) from PBMCs and stimulated with anti-CD3/CD28-coated beads (Gibco) for 4 days in the presence of IL12 (20 ng/mL; R&D Systems) together with or without IL6 (20 ng/mL; R&D Systems), sIL6R (Peprotech), or sgp130 (60 ng/mL; R&D Systems). Cells were restimulated for 2 days, and then following 1 day of resting culture, cells were analyzed. CD14⁺ cells were isolated from PBMCs using CD14 microbeads (Miltenyi Biotec), and their culture supernatant was added into T-cell culture simultaneously with IL6 (10 ng/mL).

siRNAs against Stat3 and negative control duplexes were purchased from Applied Biosystems. Transfection of 100 pmol siRNA into anti-CD3/CD28 antibody–stimulated T cells (1 × 10⁶) were conducted using HVJ-E vector (GenomeONE, Ihsihara Sangyo Ltd.) according to manufacturer's instruction. After 72 hours of transfection, cells were analyzed for cytokine production by flow cytometry.

RNA was extracted with the RNeasy Plus Mini Kit (QIAGEN) and reverse-transcribed with ReverTra Ace (TOYOBO). Real-time PCR was performed on ViiA7 Real-Time PCR System with TaqMan probes and Master Mix reagents (Applied Biosystems). Each gene expression was normalized to the Gapdh expression with the comparative 2^[−ΔΔCT] method.

The mice were transferred with OT-II cells and immunized with OVA-IIp-pulsed DC. Four days later, CD4⁺ T cells were isolated, and total RNA was extracted. Microarray analysis using Agilent SurePrint G3 Mouse Gene Expression 8 × 60 K Microarray Kit and data processing was performed by MBL. Data analysis was performed using GeneSpring GX (Agilent Technology). Background corrected intensity values between arrays were normalized using the 75th percentile shift method. Normalized log₂-transformed probe fluorescence intensities from independent samples showing a median fold change in expression >2.0 or <−2.0 was considered as differentially expressed genes. Microarray data are available under GEO accession number GSE93105.

Multiple comparisons were performed by one-way ANOVA followed by Tukey–Kramer post-hoc tests. A Kruskal–Wallis test was the nonparametric alternative to ANOVA. Data were also analyzed using unpaired Student t test when comparing 2 experimental groups. These analyses were performed using the Prism 4.0 (GraphPad).

For path analysis based on structural equation models, hypothetical pathways predicted from the mouse model were examined using 48 samples that were collected from 22 patients with HNT along with the vaccination and 25 healthy donor samples. For modeling the pathway, the concentration of sIL6R and IL6 in plasma, the frequencies of CD14⁺CD16⁻ cells and CD4⁺ T cells in PBMC, and c-Maf⁺CD4⁺ T cells were included. Correlations induced by multiple samples from each patient as well as effects of age and a number of peptides vaccinations were adjusted. After visual investigation of the distribution of measurements, some variables were log-transformed to approximate normal distributions. STATA (version 14.1) was used to fit the above models. P < 0.05 was considered significant.

We previously demonstrated that differentiation of CD4⁺ T cells expressing OVA-specific TCR (OT-II cells) into IFNγ-producing Th1 cells was attenuated in mice with tumors expressing the surrogate tumor-associated antigen, OVA, and that the defect was reversed by anti-IL6 Ab treatment (Fig. 1A and ref. 7). This finding was confirmed by the fact that the defective Th1 differentiation of OT-II cells primed with OVA-IIp-pulsed DC was observed in tumor-bearing wild-type (WT) but not in IL6-deficient mice, regardless of the type of tumor cells such as MCA-OVA or EG7 (Fig. 1B and C). IL6 did not affect the priming or expansion of donor T cells or the proportion of immunosuppressive components such as Gr-1⁺CD11b⁺ myeloid-derived suppressor cells (MDSC) or Foxp3⁺ regulatory T cells (Fig. 1B–D and ref. 7). We also found an IL6-dependent decrease of granulocyte macrophage-colony-stimulating factor (GM-CSF)–producing cells and a modest but significant increase of IL10-producing cells in tumor-bearing mice (Fig. 1E). This IL6-dependent Th1 suppression was mediated through the direct action of IL6 on CD4⁺ T cells because a substantial STAT3 activation was detected in donor OT-II cells in tumor-bearing mice, which was abrogated by in vivo administration of anti-IL6R Ab (Fig. 1F).

In terms of antitumor responses triggered by CD4⁺ T cells, the treatment with anti-IL6R Ab augmented the antitumor effect of WT OT-II cells in MCA-OVA–bearing WT host mice (7) or even in IFNγ KO hosts (Fig. 1G). However, the effect of IL6R blockade was not observed in tumor-bearing mice that were transferred with IFNγ KO OT-II cells, suggesting that the beneficial effect of IL6 (R) blockade on CD4⁺ T-cell–mediated antitumor immunity was exerted through their Th1 responses. These results were consistent with the observation that when WT OT-II cells were transferred, their helper activity for endogenous tumor (OVA)–specific CD8⁺ T cells was recovered by IL6 blockade in tumor-bearing mice, which was not observed in IFNγ-deficient OT-II cells–transferred mice (Fig. 1H). Given that CD8⁺ T cells were essential for tumor rejection in this model (7), these data suggest an indispensable role of IFNγ derived from CD4⁺ T cells (Th1 responses) to provide help for the activation of tumor-specific CD8⁺ T cells, which was abrogated by IL6 signaling in CD4⁺ T cells.

We found that CD4⁺ T cells temporally lost the surface IL6R expression in response to antigenic stimulation in vivo (Fig. 2A), regardless of the presence of a tumor or IL6 activity (Fig. 2B), confirming the previous study (28). However, despite the lack of surface IL6R at priming phase, the IL6 signal could be transmitted in OT-II cells activated in tumor-bearing mice as indicated by STAT3 activation (Fig. 1E). This discrepancy might be explained by “IL6 trans-signaling” mediated through IL6/sIL6R complex (16, 17). To address this possibility, we analyzed the Th1 differentiation of IL6R-deficient CD4⁺ T cells under Th1-skewed condition in the presence of sIL6R in vitro. IL6-dependent inhibition of Th1 differentiation was recapitulated in vitro (Fig. 2C). Although an additive effect of sIL6R on attenuating the ability to produce IFNγ was observed in IL6-stimulated WT cells, sIL6R-dependent Th1 inhibition was more obvious in IL6R-deficient OT-II cells, although the dependency of IL6 trans-signaling in IL6R⁺ cells is still unclear. The synergistic effect of sIL6R on IL6-mediated attenuation of Th1 differentiation was also confirmed by human polyclonal CD4⁺ T cells (Fig. 2D). Furthermore, IL6-dependent Th1 inhibition was partially rescued by Stat3 knockdown (Fig. 2E), confirming the involvement of STAT3 in this event.

Next, we directly tested the involvement of IL6 trans-signaling in defective Th1 development in tumor-bearing mice using recombinant soluble gp130 (sgp130), which selectively captures the IL6/sIL6R complex and competitively antagonizes its ligation with membrane-anchored gp130 (16, 29). Th1 differentiation in both IL6R-sufficient and -deficient donor OT-II cells, but not their expansion, was significantly augmented by an administration of sgp130 in tumor-bearing mice (Fig. 2F). These results suggest that the surface expression of IL6R on T cells is not essential and that T-cell–extrinsic sIL6R contributed to the Th1 inhibition in tumor-bearing mice. This was supported by the systemic increase of sIL6R in tumor-bearing mice, which was abolished in IL6R-deficient hosts (Fig. 2G), despite comparable tumor sizes in both IL6R^−/− and IL6R^+/+ mice (Fig. 2H). The sIL6R-dependent attenuation of Th1 differentiation was further demonstrated in ubiquitous IL6R-deficient mice that were utilized as tumor-bearing hosts, where a subtle decrease in IFNγ-producing cells was observed at early phase (within 8 days) during tumor progression (Fig. 2I). Collectively, these results suggest that IL6 trans-signaling via sIL6R, rather than classical IL6 signaling, attenuates Th1 differentiation of tumor-specific CD4⁺ T cells.

The sIL6R can be released through a cleavage of membrane-bound IL6R via TNF-converting enzymes (TACE) such as a disintegrin and metalloproteinase domain (ADAM) 10 and ADAM17 (16, 28). In fact, PMA/ionomycin-stimulated sIL6R production from splenocytes of tumor-bearing mice was abrogated with the treatment of TACE inhibitor, TAPI-0 (Supplementary Fig. S1A). To determine potential sources of sIL6R in tumor-bearing mice, TACE activity and sIL6R production in isolated cell fractions were examined. To this end, Gr-1⁺ cells and CD11b⁺ cells exhibited higher TACE activity (Supplementary Fig. S1B). CD11b⁺Gr-1⁻ macrophages and CD11b⁺Gr-1⁺Ly6C⁺ MDSC cells from tumor-bearing mice exhibited an abundant production of sIL6R and higher expression of Adam10/17 (Supplementary Fig. S1C–S1E). Consistent with these results, in vivo depletion of MDSC or macrophages with anti-Ly6G Ab, anti-Gr-1 Ab, or clodronate, respectively, significantly reduced the sIL6R concentration in tumor-bearing mice but did not lead to its complete abrogation (Supplementary Fig. S1F and S1G).

The systemic increase of sIL6R and subsequent Th1 suppression was not prominent in tumor-bearing IL6R^−/− hosts even when sIL6R was produced by tumor cells (Fig. 2F and G and Supplementary Fig. S2A). Indeed, despite a higher TACE activity and ADAM-dependent sIL6R production in tumor cells (Supplementary Figs. S1B and S2A), Th1 inhibition was not restored in mice with ADAM10/17-knockdown tumor cells that reduced the ability to produce sIL6R (Supplementary Fig. S2A–S2C), suggesting that tumor-derived sIL6R did not have a significant impact on Th1 inhibition during the early phase of tumor progression.

To further assess the involvement of myeloid cell-derived sIL6R, we generated mice with myelomonocytic lineage–specific deletion of IL6R (LysM-Cre x IL6R^fl/fl; referred to as IL6R mKO), resulting in a failure to produce sIL6R in myeloid cells. As shown in Fig. 3A, during the early phase of tumor expansion, the level of sIL6R in serum rapidly increased in WT mice. In contrast, conditional deletion of sIL6R in myeloid cells resulted in a significant delay of sIL6R induction, despite the comparable increase of IL6 concentration and tumor size in both mice when the mice did not receive any treatments (Fig. 3A and B).

Next, IL6R-deficient OT-II cells were primed in tumor-bearing IL6R mKO mice, and their ability to produce IFNγ and GM-CSF was assessed. As a result, the detrimental effect of tumor-bearing mice on Th1 differentiation was abrogated in IL6R mKO mice (Fig. 3C). Consistent with this, we observed a marked increase of CXCR3 expression, a marker for Th1 cells when OT-II cells were primed in IL6R mKO mice, whereas the expression levels of co-inhibitory or co-stimulatory molecules, such as LAG3, PD-1, and ICOS, were comparable in both IL6R mKO and control mice (Supplementary Fig. S3A and S3B). Furthermore, tumor-specific CD8⁺ T cells in tumor-draining lymph nodes were dramatically increased by the helper activity of OT-II cells primed in IL6R mKO mice (Fig. 3D).

We next examined the antitumor effect when IL6R-deficient OT-II cells were transferred into litter control or IL6R mKO mice with more aggressive OVA-expressing melanoma, MO4. As evidenced by the metastatic burden in the lung, significant therapeutic effect of tumor-specific CD4⁺ T cells was observed in IL6R mKO but not in WT mice (Fig. 3E). This effect was comparable to that of IL6 blockade in vivo. In line with this, recruitment and infiltration of tumor-specific CD8⁺ T cells into tumor-draining lymph nodes and lungs were augmented in IL6R mKO mice (Supplementary Fig. S3C and S3D). Conversely, immunization with OVA-Ip-pulsed DC to directly prime tumor-specific CD8⁺ T cells elicited comparable antitumor effects in both IL6R mKO and WT mice, circumventing help from cognate CD4⁺ T cells (Fig. 3F). This was consistent with that the response of tumor-specific CD8⁺ T cells was not altered by sIL6R blockade when they were directly primed (Fig. 3G and Supplementary Fig. S3E).

We also investigated the suppressive effect of sIL6R on more physiological responses of endogenous tumor-specific CD4⁺ T cells in mice inoculated with MuLV-induced lymphoma, RMA (25). Only in conjunction with an administration of sgp130, immunization with tumor-specific MuLV peptide EnvH13.3 presented by I-A^b (25) induced IFNγ-producing endogenous tumor-specific CD4⁺ T cells and tumor (MuLV)-specific CD8⁺ T cells (Fig. 4A). Abrogation of specific CD8⁺ T-cell responses by depletion of CD4⁺ T cells revealed that sgp130 administration improved CD4⁺ T-cell–mediated helper activity toward CD8⁺ T cells rather than having a direct effect on CD8⁺ T-cell activation. Furthermore, CD4⁺ T-cell–mediated antitumor effects primed with H13.3 peptide–pulsed DC was profound in RMA-bearing IL6R mKO mice as compared with those in WT counterparts, which was also abolished by CD4 depletion (Fig. 4B). These results demonstrated that the immunosuppressive effect of sIL6R was mediated through CD4⁺ T cells. Collectively these data suggest that myeloid cell–derived sIL6R represents a critical factor for the attenuation of Th1 responses and subsequent impairment of CD4⁺ T-cell–mediated antitumor immunity.

To explore the molecular mechanism dictating the IL6/sIL6R-mediated Th1 inhibition, we examined the gene expression profile of CD4⁺ T cells isolated from tumor-free or -bearing mice (Fig. 5A). We found that Il4, Ccr4, and c-maf, which were Th2-associated gene signature, were substantially increased when the tumor existed, and their upregulation was reversed by anti-IL6R Ab treatment. However, the other Th2-related genes such as Gata3, Irf4, Il5, and Il13 or key transcription factors that regulate the commitment to other effector lineages such as Tbx21(T-bet), Rorc, Bcl6, or Foxp3 (30) was not dramatically altered. In contrast, Ccl3(MIP-1α) and Wnt5a expression were decreased in tumor-bearing mice in an IL6-dependent manner. IL6-dependent alterations of these genes were recapitulated by in vitro TCR stimulation of naïve CD4⁺ T cells in the presence of IL6/sIL6R under Th1-skewed conditions (Fig. 5B). Importantly, consistent with previous in vitro studies (27, 31), IL6/sIL6R stimulation induced robust upregulation of c-maf, which was one of the most prominently altered transcription factors (Fig. 5A and B), and was abrogated by Stat3 knockdown (Supplementary Fig. S3F and S3G). Intracellular staining of transcription factors in OT-II cells primed in anti-IL6R Ab–treated mice confirmed the IL6-mediated c-Maf expression in tumor-bearing mice (Fig. 5C and D).

To further investigate the role of c-Maf, we generated OT-II cells with c-Maf mutation (designated as Ofl) that was located in DNA-binding domain (22) and thereby resulted in its dysfunction. In vitro stimulation of Ofl cells revealed that c-Maf function affected the expression of many genes that were identified as IL6-dependent altered genes (Fig. 5B). Actually, Il4, Il21, IL10, and Gata3 were greatly downregulated by c-Maf inactivation. When Ofl OT-II cells were transferred and primed in tumor-bearing mice, their ability to produce IFNγ and GM-CSF was partially rescued as compared to those in WT OT-II cells (Fig. 5E), suggesting the requirement of IL6-induced c-Maf activity for Th1 suppression. In addition, c-Maf mutant OT-II cells significantly retarded the growth of OVA-expressing melanoma, which was not observed in WT OT-II cells (Fig. 5F). These results suggest that IL6/sIL6R-mediated signaling intrinsically inhibited Th1 differentiation through c-Maf induction, resulting in a dysfunction of CD4⁺ T-cell–mediated antitumor activity.

As well as other type of cancers (10, 12, 15), the levels of IL6 and sIL6R in plasma were significantly higher in patients with HNT than in healthy donors (Fig. 6A). The detailed subject information is summarized in Supplementary Tables S1 and S2. Next, we assessed the expression of surface IL6R on myeloid cells from patients with HNT and healthy donors because the decreased level of surface IL6R expression indicates the shedding of membrane-bound IL6R in sIL6R-producing cells (16, 28). A profound decrease of IL6R expression was observed in CD14⁺CD16⁻ classical monocytes from patients with HNT (Fig. 6B and E), whereas no difference was detected in CD4⁺ T cells as compared with those from healthy donors (Supplementary Fig. S4A), suggesting that myeloid population was a likely source of sIL6R in patients with cancer.

On the basis of the mouse model, we hypothesized that the shedding and the release of sIL6R from myeloid cells contributed to a systemic increase of sIL6R in patients with cancer, resulting in the upregulation of c-Maf in CD4⁺ T cells. Therefore, we analyzed c-Maf expression in CD4⁺ T cells from HNT and healthy donors. Although CD4⁺ T-cell population was decreased in HNT significantly, a higher frequency of c-Maf⁺CD4⁺ T cells was observed in HNT as compared with healthy donor group (Fig. 6C). To investigate the directed dependency and correlation among a set of these variables and to examine the difference and significance of this pathway between HNT and healthy donor cohorts, we performed a path analysis on the basis of structural equation modeling. As shown in Fig. 6D, there was a substantial positive correlation between the frequency of IL6R^lowCD14⁺ cells in PBMC and the plasma levels of sIL6R/IL6 in HNT but not in healthy donors. Furthermore, in the healthy donor group, the frequency of c-Maf⁺ cells decreased along with an increase in the CD4⁺ T-cell population; however, such relationship was not observed in the HNT group (Fig. 6D and Table 1). The comparative analysis of this pathway suggests that there is an obvious difference in the regulatory pathway inducing c-Maf⁺ cells in CD4⁺ T cells between HNT and healthy donor cohorts (Supplementary Table S3, P < 0.002). These results imply a functional linkage between myeloid cell–derived sIL6R and c-Maf induction in CD4⁺ T cells only in patients with cancer but not in healthy donors. An adjustment and integration of the age and the frequency of peptides vaccination as additional factors in the structural equation modeling did not attenuate each association among the variables (Supplementary Fig. S4 and Supplementary Table S4).

To confirm the functional relevance of sIL6R derived from cancer-associated myeloid cells in c-Maf upregulation and Th1 inhibition, we analyzed sIL6R production in CD14⁺ cells from patients with cancer and its ability to attenuate Th1 differentiation in the presence or absence of sgp130. Consistent with the hypothesis, lower IL6R expression and higher sIL6R production in HNT-derived CD14⁺ cells were reversed by the inhibition of IL6R shedding with TAPI-0 treatment in vitro (Fig. 6E and F). Furthermore, culture supernatant of CD14⁺ cells from HNT but not healthy donors suppressed Th1 differentiation, which was rescued by sIL6R neutralization with sgp130 treatment (Fig. 6G). Conversely, the culture supernatant of HNT-derived CD14⁺ cells specifically upregulated c-Maf expression (Fig. 6H). As with the case of Th1 inhibition, c-Maf expression was inhibited by sgp130, substantiating the sIL6R-dependent c-Maf induction. Collectively, the path diagrams and in vitro results suggest that CD14⁺ cell–derived sIL6R leads to c-Maf upregulation and defective Th1 differentiation in patients with cancer.

The surface IL6R as a source of sIL6R is mainly expressed by monocytes/macrophages, hepatocytes, neutrophil, and lymphocytes in addition to tumor cells (12, 13, 16). Our study demonstrated that while either impaired release of sIL6R from tumor cells or genetic ablation of IL6R in T cells had only a marginal effect in Th1 suppression in tumor-bearing mice, depletion of myeloid cells or myeloid cell–specific deletion of IL6R diminished the systemic increase of sIL6R in tumor-bearing mice, leading to improved Th1 development and subsequent antitumor responses. These findings are consistent with the fact that the presence of myeloid cells, including macrophages and MDSC, is associated with enhanced inflammatory status and poor clinical outcome in patients with cancer (1, 8). Preferential interaction between tumor-specific CD4⁺ T cells and myeloid cells rather than tumor cells because of their MHC class II expression may provide the dominant effect of myeloid cell–derived sIL6R on Th1 inhibition at the priming in the microenvironment such as tumor-draining lymph nodes. Thus, it is plausible that myeloid cells promote tumor progression partly through sIL6R-mediated dysregulation of tumor-specific Th1 cells.

Contrary to Th1, Th2-biased differentiation of CD4⁺ T cells may in turn enhance the skewing toward the tumorigenic myeloid cells through IL4 production (9, 32). In addition, cancer burden is likely to initiate qualitative changes in myeloid cells, as indicated by an increase in Adam10/17 expression and subsequent vigorous release of sIL6R. Collectively, sIL6R appears to serve as one of the key molecules that mediate a mutual activation loop among T cells, myeloid cells, and tumor cells to exacerbate tumor progression. This proof-of-concept based on the mouse model was supported by the results that sIL6R released from CD14⁺ cells in patients with cancer exerted an immunosuppressive function that modified the human CD4⁺ T-cell property.

An imbalance in quantity and quality of tumor-associated CD4⁺ T cells including regulatory T cells is highly relevant to the magnitude of antitumor immune responses and clinical outcomes (3, 5, 32, 33). We also demonstrated that IL6/sIL6R signaling in tumor-specific CD4⁺ T cells is uniquely predisposed to redirect Ccr4, Il4, Il21-expressing non-classical Th2-like c-Maf⁺ cells rather than functional Th1-skewed responses, which was possible cause for defective T-cell–mediated tumor regression, as demonstrated using IFNγ-deficient CD4⁺ T cells. In patients with cancer, dominant Th2 cytokine profiles resulted in unfavorable outcomes (3, 5, 33), whereas inhibition of IL4 in conjunction with immunotherapy improved the survival rate of tumor-bearing hosts (9, 34). Although little is known about the detailed transcriptional targets of IL6/sIL6R that dictate such unfavorable T-cell differentiation, in vitro and in vivo experiments elucidated that loss-of-function in c-Maf conferred resistance to IL6/sIL6R-mediated Th1 inhibition and significantly abrogated Il4 and Il21 expression, which were partly responsible for Th1 inhibition (27, 31). These findings provided the evidence for an immunosuppressive role of IL6/sIL6R–STAT3–c-Maf axis in intrinsically dictating the biased differentiation of tumor-specific CD4⁺ T cells into tolerogenic Th2 rather than the beneficial Th1 cells in tumor microenvironment, resulting in improved antitumor effects. However, Th1 differentiation restored by c-Maf inactivation was less than that in IL6-targeted mice, suggesting the additional target molecules or cells of IL6/sIL6R for dampening Th1 differentiation. Actually, the expression of Ccr4, Ccl3, Wnt5a, or Bcl11a was affected by IL6/sIL6R signaling but not by c-Maf inactivation.

It has been shown that the immunosuppression in tumor microenvironment hinders the development of de novo CTL responses to tumor-associated (neo-)antigens (1). Considering the inherent ability of CD4⁺ T cells to potentiate CTLs (1, 4, 35, 36), a part of defective CD8⁺ T-cell activation appears to be a consequence of the IL6/sIL6R-induced failure in CD4⁺ T cells to provide a help for CD8⁺ T cells because the IL6/sIL6R blockade in tumor-bearing hosts allowed CD4⁺ T cells to promote the recruitment of tumor-specific CD8⁺ T cells into draining lymph nodes and tumor sites. These ideas are supported by the clinical finding in patients with cancer that the presence of CD8⁺ T-cell response and/or IFNγ-producing CD4⁺ T cells specific for NY-ESO-1 was associated with favorable antitumor responses and prolonged survival, whereas the presence of IL4-producing CD4⁺ T cells was not associated with a survival benefit and even abolished the favorable effect of CTL responses (33).

Our findings not only provide a plausible explanation for the failure of CD4⁺ T cells to regress tumors but also may have direct implications for practical applications in cancer immunotherapy. Taking into account the role of sIL6R in preventing a clearance of relatively lower level of IL6 and in amplifying its bioactivity (16, 17), it is postulated that the proceeding increase of sIL6R in the early phase of tumor progression is the rate-limiting step for the attenuation of the antitumor T-cell responses and thereby may provide an earlier prognostic value over that of IL6 for monitoring the immunosuppressive status in patients with cancer to predict their susceptibility to T-cell–mediated immunotherapies. However, more intensive investigations with a larger number and longitudinal follow-ups of patients will be required to determine their prognostic values in cancer progression because the significant correlation between overall survival of patients with HNT and their levels of sIL6/IL6R was not obtained because of the small size and advanced tumor stages of our current cohort.

Human CD4⁺ T cells still exhibit the plasticity with flexible genetic programs rather than irreversibly differentiated status even after antigen experiences and can acquire different properties in the context of secondary responses (37, 38). This supports a possibility that IL6/sIL6R converts the transferred or in vivo activated CD4⁺ T cells into dysfunctional T cells in patients with cancer when T-cell–mediated immunotherapy such as adoptive T-cell transfer including chimeric antigen receptor–expressing T cells (CAR-T cells) therapy, vaccination with tumor-associated antigenic peptides, or immune checkpoint blockade, were given. Indeed, recent trials demonstrated that the level of IL6 correlated with overall survival rate of patients treated with immunotherapies (39–41). It is particularly intriguing that following CAR T-cell infusion, the humanized anti- IL6R Ab tocilizumab is widely used to lessen cytokine release syndrome (CRS)–related toxicities (2, 42). Patients with CRS exhibited higher levels of sIL6R/IL6 and the profiles that mirror macrophage activation syndrome (42). Although whether tocilizumab improves the Th1 responses of CD4⁺ T cells within the CAR-T cells remains to be investigated, a combined IL6/sIL6R blockade in immunotherapeutic regimens not only reduces toxicity risks such as CRS but also may be a promising strategy for further enhancement of their therapeutic efficacy through improving the quality of tumor-specific T cells, in addition to the approaches that compensate for their quantitative decrease.

No potential conflicts of interest were disclosed.

Conception and design: H. Tsukamoto, K. Fujieda

Development of methodology: H. Tsukamoto, K. Fujieda

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Tsukamoto, K. Fujieda, M. Hirayama, A. Yuno, K. Matsumura, K. Araki, H. Nakayama

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Tsukamoto, K. Fujieda, T. Ikeda

Writing, review, and/or revision of the manuscript: H. Tsukamoto, K. Fujieda, M. Hirayama, T. Ikeda, A. Yuno, K. Araki, H. Mizuta, H. Nakayama, S. Senju, Y. Nishimura

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Tsukamoto, K. Fujieda, D. Fukuma, H. Nakayama

Study supervision: H. Tsukamoto, Y. Nishimura

We thank Drs. Youichiro Iwakura, Shigeo Koyasu, Irmgard Foester, and Akira Shibuya for the generous supply of IFNγ-deficient embryos, OT-II TCR transgenic embryos, lysozyme M-Cre knock-in embryos, and RMA cells, respectively.

This work was supported by JSPS KAKENHI No. 26430165 to H. Tsukamoto, and JSPS KAKENHI No. 15H04311 to Y. Nishimura. H. Tsukamoto was also supported by The Takeda Science Foundation, and the Kumamoto University Advance Research Project A “International Research Center for Cancer and Metabolism”.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.