Tumor-Derived IL-35 Promotes Tumor Growth by Enhancing Myeloid Cell Accumulation and Angiogenesis

Interleukin-35 is a dimeric protein composed of an IL-12α-chain and an IL-27β-chain, which are encoded by the IL12A and EBI3 genes, respectively (1–3). A recent study revealed that IL-35 signals through a unique heterodimer of receptor chains IL-12Rβ2 and gp130 or the homodimers of each chain in target cells (4). Although the expression pattern of IL-35 may differ in humans (5, 6), IL-35 was shown to be secreted by Foxp3⁺CD4⁺CD25⁺ regulatory T cells (Tregs) in mice (3) or a regulatory T cell population induced by IL-35 (7). IL-35 was also shown to expand Foxp3⁺ Tregs (2). IL-35–deficient (either deficient for EBI3 or P35) Tregs have significantly reduced regulatory activity in vitro and fail to control homeostatic proliferation or cure inflammatory bowel disease in vivo (3). rIL-35 suppresses T cell proliferation (2, 3), Th17 (2), Th2 responses (2), experimental arthritis (2), and airway inflammation (8). Ectopic expression of IL-35 in pancreatic β cells prevents the development of diabetes mellitus in NOD mice (9). Thus, IL-35 is a novel regulatory cytokine that has potent inhibitory effects on T cell responses.

Although the expression and function of IL-35 have only been demonstrated in Tregs, gene-expression analysis revealed that IL-35 may have much broader tissue distribution (10). Reports indicate upregulation of EBI3 and IL-12 p35 expressions in placental trophoblasts (11), and EBI3 associates with p35 in the extract of the trophoblastic components of human full-term normal placenta (1). EBI3 is also expressed in Hodgkin lymphoma cells (12), acute myeloid leukemia cells (13), and lung cancer cells (14). IL-12p35 (12), but not IL-27p28 (15), was detectable in EBI3⁺ tumor cells; therefore, it is likely that some cancer cells can produce IL-35 but not IL-27. In the tumor microenvironment, Foxp3⁺ Tregs and other Tregs are common (16, 17); they can provide another source of IL-35. In addition, tumor-infiltrating dendritic cells were found to express EBI3 (12, 15), which could be an additional source of IL-35. Taken together, IL-35 could be an important factor in the tumor microenvironment that impacts tumor-specific T cell responses and tumor progression.

Treg-derived IL-35 was shown to inhibit antitumor T cell responses (7). Small interfering RNA silencing of EBI3 in lung cancer cells inhibits cancer cell proliferation, whereas stable expression of EBI3 in lung cancer cells confers growth-promoting activity in vitro (14). Moreover, high EBI3 gene expression in human lung cancer cells was shown to be associated with poor prognosis (14). However, it is unclear whether the observed effect was due to the production of the IL-35 heterodimer. Overall, little is known about the roles of tumor-derived IL-35 in tumorigenesis and the antitumor CTL response. Based on the known roles of IL-35, we hypothesized that IL-35 production in the tumor microenvironment could contribute to tumor progression. To test this hypothesis, we generated IL-35–producing cancer cells and found that expression of IL-35 significantly increased tumorigenesis. IL-35 in the tumor microenvironment significantly increased the numbers of CD11b⁺Gr1⁺ myeloid cells in tumors and subsequently promoted tumor angiogenesis. Although tumor-derived IL-35 inhibits T cell responses in tumors in immune-competent mice, it has no direct effect on the stimulation of tumor Ag–specific CD8⁺ T cells. However, IL-35 upregulates gp130 and renders cancer cells less susceptible to CTL destruction. Thus, our results indicate novel functions of IL-35 in the tumor microenvironment.

BALB/c, C57BL/6, and Rag1^−/−C57BL/6 mice were purchased from The Jackson Laboratory. Rag2^−/−BALB/c mice were purchased from Taconic Farms (Germantown, NY). Transgenic mice expressing a TCR specific for the tumor Ag P1A (P1CTL), whose TCR recognizes the H-2L^d:P1A35-43 complex, were described (18). All animal experiments were performed after approval by The Ohio State University Institutional Animal Care and Use Committee.

Mouse plasmacytoma J558 cells (H-2L^d) were described (19). Mouse plasmacytoma J558 cells or B16F10 melanoma cells were cotransfected with an expression vector pORF9-mIL-35elasti (InvivoGen) and a selection vector (pcDNA3-neo) or the control expression vector pORF9 (InvivoGen) and pcDNA3-neo. Thereafter, stable cell lines resistant to G418 were generated. RT-PCR was used to screen IL-35⁺ cell lines, using the following primers: EBI3: 5′-ACGTCCTTCATTGCCACTTACAGGCT-3′ (forward), 5′-AGGGAGGCTCCAGTCACTTGGTTT-3′ (reverse) and IL12A: 5′-AGGTGTCTTAGCCAGTCCCGAAACC-3′ (forward), 5′-CTGAAGGCGTGAAGCAGGATGCAGA-3′ (reverse). RT-PCR was also used to determine the expression of IL-35R subunits (IL-12Rβ2 and gp130) in IL-35–treated or IL-35⁺ and IL-35⁻ tumor cells. The following primers were used: IL-12Rβ2: 5′-GTATGACCTTGTTTGTCTGCAAGC-3′ (forward), 5′-CTGTAAACGGTCTCAGATCTCGCA-3′ (reverse) and gp130: 5′-TGTCACGTTCACAGACGTGGTCCT-3′ (forward), 5′-CCAAGTTGAGGTATCTTTGGTCCT-3′ (reverse). HPRT gene was amplified for PCR loading control, and the primers used were 5′-GTCGTGATTAGCGATGATGAACCA-3′ (forward), 5′-CACCAGCAAGCTTGCAACCTTAAC-3′ (reverse). The generated J558-IL-35, J558-Ctrl or B16-IL-35, B16-Ctrl cells were maintained in RPMI 1640 medium (Life Technologies) supplemented with 100 μg/ml penicillin, 100 μg/ml streptomycin, and 5% FBS. To establish tumors in mice, 5 × 10⁶ J558-IL-35, J558-Ctrl or 0.1 × 10⁶ B16-IL-35, B16-Ctrl cells were inoculated s.c. into the flank. The length (a) and width (b) of each tumor were measured using a digital caliper every 2 or 3 d. The tumor volume was calculated according to the formula V = ab²/2, as described (18–21).

Immunohistochemistry (IHC) was used for the staining of formalin-fixed, paraffin-embedded cancer tissues. Immunostaining was performed on cancer tissues using 4 μg/ml mouse anti-human IL-35 IgG1 (15K8D10; Imgenex, San Diego, CA) for 120 min. Ab binding was detected using anti-mouse peroxidase–conjugated EnVision reagent and diaminobenzidine as chromogen. Sections were also counterstained with Mayer hematoxylin. Standard H&E staining was performed on each serial cancer tissue.

Immunofluorescence staining was performed on tissues of mouse tumors. Briefly, established mouse tumors were harvested and frozen in Tissue-Tek OCT media (Sakura Finetek), and 10-μm-thick slices were prepared. Tissue sections were fixed in ice-cold acetone for 30 s and then stained with the corresponding fluorescent Abs overnight at 4°C. The following Abs were used for fluorescence staining of tumor sections: FITC–anti-CD31, FITC–anti-VEGF, PE–anti-Gr1, and FITC–anti-CD11b. After washing with PBS, slides were mounted with DAPI-containing VECTASHIELD mounting medium (Vector Laboratories). For detection of apoptotic cells in tumor tissues, a TUNEL apoptosis detection kit (GenScript, Piscataway, NJ), containing biotin-11-dUTP and streptavidin-FITC, was used following the manufacturer’s labeling protocol. Immunofluorescence-labeled slides were examined and photographed on an inverted three-color fluorescence microscope system (Nikon Ti-U). Images were analyzed and quantified using ImageJ software (National Institutes of Health).

FITC-, PE-, allophycocyanin-, or PerCP-labeled Abs to CD4, CD8α, Vα8.3, Foxp3, and IFN-γ, as well as isotype-matched control Abs, were purchased from Becton Dickinson Biosciences (San Diego, CA) or eBiosciences (San Diego, CA). For staining of cell surface markers, cells (cultured lymphocytes and single-cell suspensions of tumors) were stained with various Abs in staining buffer (PBS with 1% FCS) and incubated on ice for 30 min. After washing with staining buffer, cells were fixed in 1% paraformaldehyde in PBS. For intracellular cytokine staining, cells were stimulated in culture medium for 4 h with 100 ng/ml PMA and 500 ng/ml ionomycin in the presence of GolgiStop (1:1500; Becton Dickinson Biosciences). Viable cells were first stained for cell surface markers and then fixed in IC Fixation Buffer (eBioscience), permeabilized with 1× permeabilization buffer (eBioscience), and stained with the respective Abs. Staining for Foxp3 was performed according to the manufacturer’s protocol (Becton Dickinson Biosciences). Stained cells were analyzed on a FACSCalibur flow cytometer, and data were analyzed using FlowJo software (TreeStar, Ashland, OR).

A mouse IL-35 ELISA kit (MyBioSource, San Diego, CA) was used to detect IL-35 concentration in the supernatants of cell cultures or protein lysates of established IL-35⁺ or IL-35⁻ J558 tumors.

To determine proliferation and survival of IL-35⁺ or IL-35⁻ tumor cells, an MTT assay kit (American Type Culture Collection) was used, following the manufacturer’s instructions.

For measuring P1CTL cell proliferation, 0.3 × 10⁶/ml spleen and lymph node cells from P1CTL mice were cultured in Click’s Medium Eagle Hank's Amino Acids (Invitrogen) containing 100 μg/ml penicillin, 100 μg/ml streptomycin, 1 mM 2-ME, 5% FBS, and 0.2 μg/ml P1A35-43 peptide in the absence or presence of IL-35–containing culture supernatants from J558 cells in 96-well U-bottom plates. For detection of proliferation of P1CTL cells, [³H]thymidine was added to the culture for the last 12 h, and incorporation of [³H]tritium was quantified using a scintillation counter.

To determine the cytotoxicity of cultured P1CTL cells, splenocytes from P1CTL TCR-transgenic mice were stimulated with P1A peptide (0.1 μg/ml) for 5 d in the presence or absence of IL-35 and used as effectors. ⁵¹Cr-labeled tumor cells, with or without IL-35 incubation, were used as targets. The effector T cells and the targets were incubated together for 6 h, and the percentages of specific lysis were calculated based on the following formula: specific lysis (%) = 100 × (cpm_sample − cpm_medium)/(cpm_max − cpm_medium).

CD11b⁺ or Gr1⁺ cells were purified from spleens of J558 tumor–bearing Rag2^−/− mice or bone marrow of normal mice. Macrophage cell lines Raw264.7 (H-2^b) and P338D1 (H-2^d) were cultured in complete RPMI 1640 medium. A total of 2 × 10⁵ cells in 200 μl medium was added to the upper chamber of a Transwell system (3.0–8.0 μm pore size polyester; Becton Dickinson Biosciences) and inserted into a 24-well plate. The lower chambers of each Transwell were filled with 800 μl culture medium, with or without IL-35. The cells were incubated for 24 h at 37°C, medium was discarded, the transmembranes were washed, and the numbers of cells that had migrated to the bottom of the transmembranes were counted under the microscope.

Data are expressed as mean ± SD. The two-tailed Student t test was used for statistical analysis; p < 0.05 was considered significant.

Accumulating evidence suggests that human cancer tissues may produce IL-35. IHC studies revealed that EBI3 protein can be detected in Hodgkin lymphoma cells (12), nasopharyngeal carcinoma cells (12), and lung cancer cells (14). IL-12p35 (12), but not IL-27p28 (15), was detectable in EBI3⁺ tumor cells. Therefore, it is likely that some cancer cells can produce IL-35. Moreover, tumor-infiltrating dendritic cells (TIDCs) also express EBI3 protein (12, 15); thus, TIDCs could be an additional source of IL-35. Furthermore, Foxp3⁺ Tregs are frequently found in human cancer tissues (16, 17) and could be an additional source of IL-35. Accordingly, it is possible that IL-35 can be detected in human cancer tissues. To test this hypothesis, we performed IL-35 IHC analysis on human large B cell lymphoma (Fig. 1A, 1B) and nasopharyngeal carcinoma (Fig. 1C, 1D), two cancer types that are associated with EBV infection. In addition, we also examined IL-35 expression in non-EBV–related cancer tissues, such as melanoma. As shown in Fig. 1B, some large B cell lymphoma cells showed positive staining of IL-35. In other cancer types, IL-35⁺ staining was detected mainly in stromal cells rather than cancer cells (Fig. 1D, 1F, 1H). In addition, IL-35⁺ cells exhibited various shapes (Fig. 1D, 1F, 1H). Thus, IL-35 can be readily detected in various human cancer tissues and is likely of multiple cellular sources.

To determine the roles of IL-35 in the tumor microenvironment, we generated IL-35⁺ or IL-35⁻ mouse plasmacytoma J558 cells and B16 melanoma cells by transfecting J558 and B16 melanoma cells with an IL-35 expression vector or a control expression vector. RT-PCR analysis revealed that J558–IL-35 and B16–IL-35 cells expressed both EBI3 and IL35A mRNA (Fig. 2A, 2E). Moreover, RT-PCR assay using EBI3 forward primer and P35 reverse primer detected the rRNA encoding both EBI3 and IL-35A subunits (Fig. 2A, 2E). Thus, the IL-35 molecule was stably expressed in J558–IL-35 and B16–IL-35 cells. Immunocytochemistry staining of cells and ELISA assay of supernatants from cell cultures revealed that IL-35 protein was produced by J558 (Fig. 2B) and B16 (Fig. 2F) cells. Expression of IL-35 in cancer cells did not alter the expression of MHC class I (Fig. 2C, 2G) and tumor Ag P1A (Fig. 2A) in J558 cells. MTT assay revealed that IL-35 expression did not affect J558 and B16 cell growth in vitro (Fig. 2D, 2H).

To determine whether the expression of IL-35 in tumor cells affects tumor growth in vivo, 5 × 10⁶ IL-35⁺ or IL-35⁻ J558 cells were injected s.c. into each BALB/c mouse. As shown in Fig. 3A, IL-35⁻ J558 cells induced tumors in normal BALB/c mice 10–14 d after tumor cell injection. J558–IL-35 tumors grew progressively in BALB/c mice; within 2 wk the tumors grew to a size that required euthanasia. Our results indicated that ex vivo J558–IL-35 tumors were significantly bigger than ex vivo J558-Ctrl tumors by day 15 after tumor cell injection, and a high concentration of IL-35 was detected in protein lysates of IL-35⁺ tumors (Fig. 3B).

To determine whether IL-35 enhances tumor growth via inhibition of the adaptive immune response, we injected J558–IL-35 or J558-Ctrl cells into Rag2^−/−BALB/c mice that lack T and B lymphocytes. As shown in Fig. 3C, J558–IL-35 tumors grew much faster compared with J558-Ctrl cells in Rag2^−/−BALB/c mice. To determine whether the tumor enhancement was IL-35 specific, an IL-35–neutralizing mAb or an isotype-matched control mAb was coinjected with J558–IL-35 cells into Rag2^−/−BALB/c mice, and we found that IL-35–neutralizing mAb abrogated tumor enhancement (Fig. 3D).

Similarly, we found that B16–IL-35 tumors grew much faster than B16-Ctrl tumors in both C57BL/6 mice (Fig. 3E) and Rag1^−/−C57BL/6 mice (Fig. 3F). In addition, when injected i.v., B16–IL-35 cells established lung foci faster than did B16-Ctrl cells (Fig. 3G). Thus, IL-35 production in the tumor microenvironment stimulates tumor growth.

To determine the mechanisms by which IL-35 expression promote tumor growth, we first examined tumor angiogenesis in established J558–IL-35 and J558-Ctrl tumors collected from Rag2^−/−BALB/c mice. Tumor angiogenesis, as determined by CD31 expression and expression of VEGF, was significantly increased in J558–IL-35 tumors in comparison with J558-Ctrl tumors (Fig. 4A, 4B). Increased tumor angiogenesis was also found in B16–IL-35 tumors (Fig. 4C, 4D).

IL-35⁺ tumors also contained increased numbers of myeloid cells in comparison with J558-Ctrl tumors. Fluorescence staining of sections from J558–IL-35 tumors grown in Rag2^−/− mice revealed accumulation of more myeloid-derived suppressor cells (MDSCs; CD11b⁺Gr1⁺) (Fig. 5A). In contrast, the numbers of CD11b single-positive cells were lower in J558–IL-35 tumors compared with control tumors (Fig. 5A). However, the numbers of Gr1 single-positive cells were low and did not differ between the two types of tumors (Fig. 5A). Flow cytometry analysis of disassociated tumor cells verified the fluorescence staining results obtained from tumor sections (Fig. 5B). Furthermore, increased numbers of CD11b⁺Gr1⁺ cells were also found in J558–IL-35 tumors grown in BALB/c mice (Fig. 5C) and B16–IL-35 tumors from C57BL/6 mice (Fig. 5D) in comparison with their controls. To determine whether increased MDSC accumulation was responsible for enhanced tumor growth, we injected three doses of anti-Gr1 mAb (250 μg/dose) into J558–IL-35 or J558-Ctrl cell–inoculated Rag2^−/−BALB/c mice. We previously showed that this regimen could completely deplete Gr1⁺ cells (22). We found that anti-Gr1 mAb treatment largely eliminated the tumor growth difference between the two groups of mice (Fig. 5E). Thus, MDSC accumulation is largely responsible for the enhanced growth of J558–IL-35 tumors in Rag2^−/−BALB/c mice.

The increased accumulation of myeloid cells in IL-35⁺ tumors suggests that IL-35 may play a chemotaxis effect on myeloid cells. To test this possibility, we performed a transmigration assay using a Transwell system in the presence or absence of IL-35. Different lineages of myeloid cells, including macrophage cell lines Raw264.7 (H-2^b) and P338D1 (H-2^d), MDSCs purified from the spleen of J558-tumor bearing mice, and bone marrow–derived CD11b⁺Gr1⁺ cells, were used in the transmigration assay. As shown in Fig. 6, similar numbers of Raw264.7 cells (Fig. 6A), P338D1 cells (Fig. 6B), spleen MDSCs (Fig. 6C), and bone marrow CD11b⁺Gr1⁺ cells (Fig. 6D) migrated to chambers containing IL-35 or the control medium. Thus, IL-35 does not directly increase migration of myeloid cells.

Because the IL-35⁺ tumors contained higher numbers of myeloid suppressor cells, which are known to suppress CTL responses in tumors (23, 24), we hypothesized that IL-35 production in the tumor microenvironment inhibits CTL responses. To test this hypothesis, we compared T cell responses in the J558–IL-35 and J558-Ctrl tumors grown in normal BALB/c mice. As shown in Fig. 7A, we found that total leukocyte numbers (CD45⁺) were increased in J558–IL-35 tumors. However, among the CD45⁺ leukocyte population, the numbers of CD8⁺ T cells were greatly reduced in J558–IL-35 tumors (Fig. 7A, 7B). CD8⁺ T cells from J558–IL-35 tumors also were less capable of producing IFN-γ in comparison with CD8⁺ T cells collected from J558-Ctrl tumors (Fig. 7C). However, the percentages of CD4⁺ T cells (Fig. 7D) and CD4⁺Foxp3⁺ Tregs (Fig. 7E) were not affected significantly. Similar to the J558 tumors, we also observed that B16–IL-35 tumors grown in C57BL6 mice contained significantly reduced numbers of CD8⁺ T cells (Fig. 7F, 7G). In contrast, the numbers of CD4⁺ T cells were not significantly affected (Fig. 7F, 7H).

To determine whether IL-35 production directly inhibits tumor Ag–specific CD8⁺ T cell proliferation and effector function, we cultured P1CTL transgenic T cells that recognize tumor Ag P1A in the presence of IL-35 produced by J558 cells. There were no differences in the proliferation and growth of P1CTL cells either in the presence or absence of IL-35, as determined by [³H]thymidine-incorporation assay (Fig. 8A) and MTT assay (Fig. 8B). IL-35–stimulated P1CTL cells also expressed similar levels of IFN-γ and granzyme B (Fig. 8C) and exhibited similar cytotoxicity to P1A⁺ P815 target cells (Fig. 8D).

Despite expressing similar levels of P1A Ag and MHC class I (Fig. 2), J558–IL-35 cells were more resistant to P1CTL destruction compared with J558-Ctrl cells (Fig. 9A). Similarly, IL-35–stimulated P815 cells were also found to be more resistant to P1CTL lysis (Fig. 9B) compared with control P815 cells. Consistent with this observation, established J558–IL-35 tumors in BALB/c mice contained less apoptotic tumor cells compared with J558-Ctrl tumors (Fig. 9C). Thus, IL-35 renders tumor target cells more resistant to CTL destruction. To determine whether cancer cell resistance to CTL destruction was due to IL-35R signaling, we determined the expression of IL-35R subunits (IL-12Rβ2 and gp130) in IL-35⁺ and IL-35⁻ tumor cells by RT-PCR. As shown in Fig. 9D, although expression of IL-12Rβ2 and gp130 was barely detectable in B16 cells, increased gp130 expression was detected in J558–IL-35 cells compared with J558-Ctrl cells. To determine whether induction of gp130 was IL-35 specific, J558 and P815 cells were cultured or not with IL-35 for 24 h, and we determined gp130 expression by RT-PCR. As shown in Fig. 9E, IL-35 increased gp130 expression in both J558 and P815 cells.

Our study validates the hypothesis that IL-35 is produced in human cancer tissues, and tumor-derived IL-35 plays important roles in tumor progression and tumor immune surveillance.

First, we demonstrated that IL-35 is produced in human cancer tissues. Among the three types of human cancer tissues examined, IL-35 is mainly found in tumor stromal cells. Thus, it will be interesting to determine the types of cells in cancer stromas that produce IL-35. Previous studies revealed that TIDCs (12, 15) express EBI3; therefore, TIDCs could be a source of IL-35. Additionally, tumor-infiltrating FOXP3⁺ Tregs could be another source of IL-35. EBI3 is known to be expressed in EBV-associated Hodgkin lymphoma, diffuse large B lymphoma, nasopharyngeal carcinoma (12, 25), and EBV⁻ malignant tumors (14). In this study, we observed that some diffuse large B lymphoma cells are also positive for IL-35. Therefore, IL-35 is abundantly produced in many human cancer tissues and multiple cell types, including some cancer cells.

We next observed that tumor-derived IL-35 considerably increased tumorigenesis in both immune-competent and immune-deficient mice. The protumor effect is very rapid, which necessitates sacrifice of mice injected with IL-35⁺ cancer cells within 2–3 wk after tumor cell injection. This protumor effect of IL-35 was confirmed in two tumor models: plasmacytoma J558 and B16.F10 melanoma. Previous studies reported that IL-35–producing Tr35 cells can inhibit tumor growth via suppression of antitumor immune responses (7). Because the protumor effect of IL-35 was also observed in Rag1/2-deficient mice in this study, the rapid tumor growth of IL-35–producing tumors is not solely due to suppression of adaptive immunity. The IL-35 effect in tumors also differs from another EBI3-containing cytokine, IL-27, which has potent tumor-inhibiting effects (26–31). Expression of EBI3 in human lung cancer cells was reported to promote lung cancer cell growth in vitro (14). However, in this study we showed that IL-35 expression did not affect tumor growth and proliferation in vitro. It is not known whether the discrepancy of the results is due to a lack of IL-35 heterodimer formation in lung cancer cells. To our knowledge, this is the first study to show that tumor-derived IL-35 has a protumor effect in vivo.

We demonstrated that IL-35 production in the tumor microenvironment increased CD11b⁺Gr1⁺ myeloid cell accumulation and tumor angiogenesis (reflected by increased density of CD31⁺ blood vessels and VEGF production in the tumor bed). Because IL-35 production by cancer cells does not affect their growth and proliferation in vitro, it is likely that IL-35–producing cancer cells are capable of inducing host cells to promote tumor growth. In this regard, increased numbers of CD11b⁺Gr1⁺ MDSCs were present in IL-35⁺ tumors. MDSCs are a known cell type that can promote tumor angiogenesis via production of VEGF and other proangiogenesis factors (32–34). Thus, it is likely that increased angiogenesis and tumor growth in IL-35⁺ tumors are due to increased accumulation of CD11b⁺Gr1⁺ myeloid cells. Indeed, depletion of Gr1⁺ myeloid cells abrogated tumor growth enhancement by IL-35. Our in vitro migration assay revealed that tumor-produced IL-35 does not have a direct role in MDSC chemotaxis. Quantitative RT-PCR analysis also revealed that VEGF expression was not altered in J558–IL-35 and B16–IL-35 cells in comparison with their relative control cells (data not shown). Thus, it remains to be determined what signal pathways are activated in tumor cells by IL-35, which are responsible for initially accumulating CD11b⁺Gr1⁺ myeloid cells into tumors and, thereby, inducing tumor angiogenesis.

Finally, in this study we demonstrated that tumor-derived IL-35 induces a suppressive tumor microenvironment, which, in turn, inhibits tumor immunity. MDSCs induce immune suppression and inhibit CTL responses (23, 24). Consistent with this observation, we found that spontaneous CTL responses in IL-35⁺ tumors were inhibited in immune-competent mice. However, IL-35 does not directly inhibit CTL proliferation, differentiation, and effector functions in in vitro assays, and this is in contrast to other IL-12 family cytokines, such as IL-12 and IL-27, which can significantly affect CTL differentiation (35–37). Thus, it is likely that the inhibition of CTL responses in IL-35⁺ tumors is due to the accumulation of MDSCs therein, which subsequently inhibits CTL responses.

Although IL-35 does not have direct effects on CTL activation and effector functions, we showed that IL-35–treated target cells were less susceptible to CTL-mediated destruction. Consistent with this effect, we found that less apoptotic tumor cells were observed in IL-35⁺ tumors than in IL-35⁻ tumors grown in immune-competent mice. This effect is unlikely mediated by downregulation of MHC class I or Ag expression, because IL-35–expressing tumor cells have similar levels of MHC class I and Ag. Our RT-PCR results suggest that IL-35 upregulates gp130 expression in tumor cells. Interestingly, gp130 signaling was shown to mediate cancer cell resistance to chemotherapy (38). Because IL-35 is known to signal through the gp130 dimer (4), it is highly suggestive that IL-35 signaling via gp130 induces cancer cell resistance to CTL destruction. Thus, IL-35–induced cancer cell resistance to CTL destruction could be one mechanism by which cancer cells escape CTL destruction in tumors.

Taken together, our investigation determining the role of tumor-derived IL-35 on tumor growth and immunity revealed novel functions for IL-35 in promoting tumor growth and inhibition of antitumor CTL responses. Because IL-35 can be produced by both cancer cells and tumor-infiltrating stromal cells, further investigations elucidating the cross-talk among cancer cells, tumor-infiltrating stromal cells, and MDSCs via IL-35 may help us to better understand tumor progression and immune evasion. Accordingly, targeting IL-35–mediated cross-talk may be a novel immunotherapeutic approach for the treatment of cancer patients.

The authors have no financial conflicts of interest.