NK cells require IL-28R for optimal in vivo activity

Significance

Natural killer (NK) cells are naturally circulating innate lymphocytes that sense altered cells, including pathogen-activated and early-transformed cells. The signals that prime the NK cell to respond are not completely understood, but cytokines, such as IL-12, IL-18, and type I interferon (IFN-αβ) from antigen-presenting cells, are appreciated to be key to NK cell effector functions in response to bacteria, viruses, and tumors. In this light, another class of IFN, IFN type III (IFN-λ), has been described that shares some common functions with IFN-αβ, but with a more restricted cellular expression. Here, we demonstrate for the first time, to our knowledge, the ability of IFN-λ to directly regulate NK cell effector functions in vivo, alone and in the context of IFN-αβ.

Abstract

Natural killer (NK) cells are naturally circulating innate lymphoid cells that protect against tumor initiation and metastasis and contribute to immunopathology during inflammation. The signals that prime NK cells are not completely understood, and, although the importance of IFN type I is well recognized, the role of type III IFN is comparatively very poorly studied. IL-28R–deficient mice were resistant to LPS and cecal ligation puncture-induced septic shock, and hallmark cytokines in these disease models were dysregulated in the absence of IL-28R. IL-28R–deficient mice were more sensitive to experimental tumor metastasis and carcinogen-induced tumor formation than WT mice, and additional blockade of interferon alpha/beta receptor 1 (IFNAR1), but not IFN-γ, further enhanced metastasis and tumor development. IL-28R–deficient mice were also more susceptible to growth of the NK cell-sensitive lymphoma, RMAs. Specific loss of IL-28R in NK cells transferred into lymphocyte-deficient mice resulted in reduced LPS-induced IFN-γ levels and enhanced tumor metastasis. Therefore, by using IL-28R–deficient mice, which are unable to signal type III IFN-λ, we demonstrate for the first time, to our knowledge, the ability of IFN-λ to directly regulate NK cell effector functions in vivo, alone and in the context of IFN-αβ.

IFN-λ is a group of viral-related interferons (type III IFN) that, in humans, includes four isoforms [IFN-λ1, IFN-λ2, and IFN-λ3 (also known as IL-29, IL-28A, and IL-28B, respectively); and IFN-λ4 as a novel variant upstream of IFN-λ3 recently characterized by a genome-wide association study in association with impaired clearance of hepatitis C virus] whereas, in mice, only two isoforms exist [IFN-λ2 and IFN-λ3 (or IL-28A and IL-28B, respectively)] (1–4). Type III IFN was shown to display a similar signaling pathway downstream as type I IFN (IFN-αβ), via JAK1/TYK2 tyrosine kinases and IRF9. However, IFN-λ has an affinity for a unique heterodimeric IFN-λR composed of an IL-28R chain and an IL-10R2 chain (which is also shared with the IL-10, IL-22, and IL-26 receptors) (5). To date, IL-28R cellular expression is reported to be expressed mainly by plasmacytoid DCs, B cells, epithelial cells, and hepatocytes (2, 6) whereas IFN-λ is believed to be strictly expressed by plasmacytoid and conventional DCs and type II epithelial cells (7).

NK cells are naturally circulating innate lymphocytes that trigger cell death in target cells that are stressed or display altered self, including early transformed cells (8, 9). The role of type I IFN and IFN-γ in NK cell-mediated control of tumor initiation, growth, and metastasis has been well-documented (10, 11). As a major and rapid source of proinflammatory cytokines, such as IFN-γ, NK cells can also contribute to promoting overzealous and deleterious inflammation in bacterial infection and sepsis (12, 13). In contrast, the role of IFN-λ in NK cell-mediated immune responses is poorly understood. In mice, IFN-λ displayed a potential antiviral role in models of influenza A virus, herpes simplex virus 2 (HSV2), and hepatitis B and C virus (3, 14–16). In these models, the direct effect of IFN-λ on host NK cells was not explored. When a mouse MCA205 fibrosarcoma cell line was engineered to express IFN-λ, IFN-λ displayed substantial in vivo antitumor properties dependent upon host NK cells (17). In addition, Lasfar et al. and Sato et al. also showed that IFN-λ–expressing B16F10 cells were rejected in an NK cell-dependent fashion (18, 19). B16F10 melanoma cells express both IL-28R and IL-10R2 chains and respond to rIFN-λ by up-regulating MHC class I. Abushahba et al. demonstrated that IFN-λ and NK cells played a role in the rejection of IFN-λ–expressing hepatocellular carcinoma cells (HCCs). In that study, a marked tumor infiltration and cellular cytotoxicity mediated by NK cells were observed in HCC-expressing IFN-λ (20).

From these previous reports, it remained unclear whether NK cells could respond to IFN-λ directly or whether they were activated by secondary signals from other cells activated by IFN-λ (e.g., DCs). To this end, Ank et al. recently described an IL-28R gene-targeted mouse strain (21), and these mice showed an indistinguishable natural clearance of different viruses compared with WT mice. However, the TLR-induced anti-HSV2 response was abolished in IL-28R^−/− mice, similar to that observed in IFNAR1^−/− (interferon alpha/beta receptor 1) mice (21). Although HSV2 antigens were recently shown to directly activate NK cells (22), the role of NK cells was not addressed using the IL-28R^−/− mice. Furthermore, these mice have not been used to explore the role of IL-28R in NK cell-mediated control of tumors, nor has the relationship of host IL-28R and IFNAR1 been examined. In this study, we aimed to determine the importance of IL-28R in NK cell activation in vivo, with an emphasis on response to TLR activation and control of tumor initiation and metastasis.

Results

IL-28R mRNA Is Expressed by Mouse NK Cells.

Human NK cells reportedly express IL-28R mRNA and potentially respond to IFN-λ (23, 24); however, mouse NK cells have not been examined in the same manner. To date, monoclonal antibodies (mAbs) specifically reactive with mouse IL-28R are not available. Thus, based on the published literature (19, 21), we first compared IL-28R mRNA expression in NK cells and dendritic cells (DCs) (CD11b⁺CD11c⁺) purified from the spleens of WT and IL-28R^−/− mice (as a negative control) and B16F10 melanoma cells (as a positive control) (SI Appendix, Fig. S1). Notably, naive NK cells expressed a significant level of IL-28R mRNA equivalent to B16F10 melanoma, but reduced levels compared with DCs. As expected, we failed to detect any IL-28R mRNA in NK cells or DCs from the IL-28R^−/− mice or RMAs tumors.

IL-28R–Deficient Mice Have a Normal NK Cell Repertoire.

Several studies have associated abnormalities in NK cell function with changes in their NK cell receptor repertoire (25, 26). To verify whether the differentiation of NK cells in IL-28R^−/− mice was normal, we assessed the NK cell receptor repertoire in blood, liver, lung, and spleen NK cells (based on the gating of CD3^negNK1.1⁺DX5⁺) as previously described (27). However, the expression of CD43, CD226, Ly49A, Ly49C/I, Ly49D, NKG2A/C/E, NKG2D, NKp46, and the cellular maturation markers CD27 and CD11b was similar between the NK cells of WT and IL-28R^−/− mice (SI Appendix, Fig. S2).

IL-28R Deficiency Sensitizes Mice to LPS-Induced Endotoxicosis.

Using the IL-28R^−/− and IFNAR1^−/− mice, we first analyzed the capacity of NK cells to respond in a model of lethal endotoxicosis after in vivo LPS challenge. Here, NK cells are significant producers of early IFN-γ and contribute to the inflammation and lethality (28–30). We first observed that, compared with WT mice and as previously reported (31), IFNAR1^−/− mice were profoundly resistant to LPS endotoxicosis (Fig. 1A). By contrast, IL-28R^−/− mice were partially resistant to LPS challenge, displaying an intermediate phenotype between WT and IFNAR1^−/− mice (Fig. 1A and SI Appendix, Fig. S3). All three strains of mice were completely resistant to LPS when NK cells were depleted using anti-asialoGM1 antibody (Fig. 1A). In concert, IFNAR1^−/− mice, and not the IL-28R^−/− mice, displayed a reduced level of the early activation marker CD69 on the surface of spleen NK cells compared with WT mice post LPS challenge (Fig. 1B). To address whether spleen NK cell IFN-γ production was also altered in the IL-28R^−/− mice, intracellular cytokine analysis was performed 6 h post LPS injection. Both IL-28R^−/− and IFNAR1^−/− mice displayed a significantly reduced proportion of IFN-γ⁺ NK cells, compared with WT mice (Fig. 1C). Serum cytokines were also assessed in the same mice to correlate the strength of the systemic inflammatory responses. Six hours after LPS challenge, cytokines characteristic of lethal endotoxicosis (32), such as IFN-γ, were decreased in mice deficient for IFNAR1 or IL-28R. The early TNF-α signal was significantly decreased in IFNAR1^−/− mice only whereas no differences were found in serum KC levels between WT and IL-28R^−/− and IFNAR1^−/− mice (Fig. 1D).

Fig. 1.

LPS stimulation in vivo. (A) IL-28R^−/− mice display resistance to endotoxicosis compared with WT mice. Groups of WT and various gene-targeted mice (IL-28R^−/−, IFN-γ^−/−, and IFNAR1^−/−) were inoculated i.p. with 1.25 mg LPS/30 g mouse. Groups received either 100 μg i.p. of cIg or anti-asialoGM1 (to deplete NK cells) on day −1 and 0 (where day 0 is LPS challenge). Statistical analysis was performed using Mantel–Cox Log-rank test, ***P < 0.001, n = 10 in two independent experiments. (B) Spleen NK cells from IL-28R^−/− mice, but not IFNAR1^−/− mice, display normal expression of CD69 in vivo, in response to LPS (i.p. injection of 0.1 mg/20 g – 6 h). Representative FACS plots are shown and mean ± SEM. CD69 mean fluorescence intensity (MFI) depicted in bar graphs, with n = 10 from two independent experiments. Statistical analysis was performed using Mann–Whitney test; *P < 0.05, **P < 0.01, ***P < 0.001. (C) IL-28R^−/− and IFNAR1^−/− spleen NK cells demonstrated defective IFN-γ production compared with WT mice at various time points (0–12 h) post-LPS. Representative FACS plots are shown and mean ± SEM. CD69 MFI depicted in bar graphs, with n = 9 from two independent experiments. Statistical analysis was performed using Mann–Whitney test; *P < 0.05, **P < 0.01. (D) Serum cytokines at various time points (0–12 h) post-LPS reveal decreased IFN-γ production in the IFNAR1^−/− and IL-28R^−/− mice. Results are expressed in mean ± SEM. Statistical analysis was performed using Mann–Whitney test; *P < 0.05, with n = 5 mice per time point.

Specific Deletion of IL-28R in NK Cells Renders Their IFN-γ Production Deficient.

To confirm whether these results might be explained by an intrinsic IFN-λ (IL-28R) signaling defect in NK cells, we transferred equal numbers of highly purified spleen NK cells from WT and IL-28R^−/− mice into immunodeficient RAG2^−/−γc^−/− mice. After 5 d of homeostatic reconstitution, the WT and IL-28R^−/− NK cell numbers and proportions in peripheral blood were equivalent, suggesting that IFN-λ (IL-28R) signaling is not required for NK cell homeostatic proliferation (Fig. 2A). On day 6, these mice were challenged with LPS, and, 6 h later, intracellular IFN-γ was measured in spleen NK cells. Clearly, spleen NK cell IFN-γ production was decreased in the mice reconstituted with IL-28R^−/− NK cells compared with WT NK cells (Fig. 2B). Serum cytokines from the IL-28R^−/− NK cell-reconstituted mice revealed decreased levels of IFN-γ as expected, but not KC or TNF-α (Fig. 2C). Because IL-28R is composed of the IL-28R and IL-10R2, one possible simple explanation for the lower IFN-γ secretion by IL-28R–deficient NK cells in vivo might have been a compensatory increase in IL-10R signaling permitted by the accumulation and pairing of otherwise free IL-10R2 with IL-10R1. Analysis of NK cells from IL-28R–deficient mice compared with WT NK cells revealed no significant increase in IL-10R nor enhanced direct IL-10 signaling (as measured by pSTAT1 and pSTAT3 in response to recombinant IL-10) in IL-28R^−/− NK cells (SI Appendix, Fig. S4).

Fig. 2.

(A) Sorted IL-28R^−/− or WT spleen NK cells display normal peripheral blood reconstitution in RAG2^−/−γc^−/− recipient mice 5 d post cell transfer (2 × 10⁵ NK cells injected i.v.). Representative FACS plots are shown, and all data from individual mice are depicted by symbols in bar graphs (Lower). (B) IL-28R^−/− NK reconstituted RAG2^−/−γc^−/− mice display decreased NK cell IFN-γ expression in spleen 6 h after LPS (0.1 mg/30 g) challenge. Representative FACS plots are shown, and all data from individual mice are depicted by symbols in bar graphs (Lower). Results are expressed as mean ± SEM; n = 15 pooled from three independent experiments. Statistical analysis was performed using Mann–Whitney test; **P < 0.01. (C) From the same mice as B, cytokines were assessed from serum after 6 h post LPS challenge. IL-28R^−/− NK cell reconstituted RAG2^−/−γc^−/− mice displayed decreased levels of IFN-γ. Results are expressed as mean ± SEM; n = 8 pooled from two independent experiments, and all data from individual mice are depicted by symbols in bar graphs. Statistical analysis was performed using Mann–Whitney test; *P < 0.05.

IL-28R^−/− Mice Are Resistant to Septic Shock.

To address whether the absence of IFN-λ signaling also regulated systemic polymicrobial sepsis and septic shock, a cecal ligation and puncture (CLP) model was performed as previously described (33). Because lethality was shown to be dependent upon NK cell IFN-γ production (32, 34, 35), not surprisingly the IL-28R^−/− mice were relatively resistant compared with WT mice (Fig. 3A). It is known that proinflammatory cytokines such as TNF-α are associated with worse prognosis in sepsis (36). In addition IL-17A is associated with excessive inflammation by increasing levels of macrophage inflammatory proteins (MIPs) (37) and by enhancing NK cell IFN-γ production (38). On the other hand, neutrophil recruitment-related cytokines (e.g., G-CSF and KC) are associated with a survival benefit because they enhance bacterial clearance (39). Beneficial serum inflammatory factors such as G-CSF and KC (increased 3 h post-CLP, and increased 3 and 12 h post-CLP, respectively) were found at higher levels post-CLP in the IL-28R^−/− mice. In concert, deleterious proinflammatory factors were found at lower levels post-CLP in the IL-28R^−/− mice (IL-17A decreased 12 h post-CLP; MIP1α decreased 3 and 12 h post-CLP; MIP1β decreased 3 h post-CLP; and TNF-α decreased 3 and 12 h post-CLP) (Fig. 3B). These data suggested that IFN-λ signaling also plays an important role in innate immunity to polymicrobial systemic infection.

Fig. 3.

(A) IL-28R^−/− mice display enhanced survival after CLP-induced septic shock. Statistical analysis was performed using Mantel–Cox Log-rank test; **P < 0.01, n = 18 pooled from three independent experiments. (B) Cytokine assessment from serum after 3 and 12 h post-CLP. IL-28R^−/− mice have increased levels of G-CSF and KC, and decreased levels of IL-17A, MIP1α, MIP1β, and TNF-α. Results are expressed as mean ± SEM, with all data from individual mice depicted by symbols in bar graphs. Statistical analysis was performed using Mann–Whitney test; *P < 0.05, **P < 0.01, and ***P < 0.001.

Ank et al. showed that IL-28R–deficient mice responded normally to a number of different viruses compared with WT mice, including the following: genital herpes HSV-2; an RNA virus, lymphocytic choriomeningitis virus (LCMV); an orthomyxovirus, influenza A virus (IAV); and a picornavirus, encephalomyocarditis virus (ECMV) (21). Given that production of IFN-γ by NK cells is dependent on IL28R signaling, we tested the relevance of this pathway in the control of murine cytomegalovirus (MCMV) infection. NK cells are essential for the control of acute MCMV infection in B6 mice, with NK cell-derived IFN-γ playing an important role in limiting viral replication (40). IL28R^−/− mice were infected with MCMV, and viral replication was assessed in target organs. Quantification of replicating virus by plaque assay demonstrated no difference in viral loads between WT and IL28R^−/− mice in spleen, liver, and lung during acute infection (SI Appendix, Fig. S5). By contrast, IFNAR1^−/− mice were more sensitive to MCMV infection. Thus, activities mediated by IL28R are not essential for the control of acute MCMV infection in B6 mice.

IL-28R^−/− Mice Are Susceptible to Tumor Metastases.

We next determined whether IL-28R was critical in NK cell-dependent control of tumor metastasis. The NK cell-mediated clearance of B16F10 experimental lung metastases after i.v. inoculation is a well-characterized and used metastasis assay (28, 41, 42). IL-28R^−/− mice displayed impaired control of lung metastases compared with WT mice at several different inoculated doses (Fig. 4A). Notably, IFNAR1^−/− mice displayed an even greater susceptibility to experimental B16F10 lung metastasis at the equivalent tumor doses (Fig. 4A). To determine whether IL-28R deletion in NK cells was critical for immune response against B16F10 lung metastases, we reconstituted immunodeficient RAG2^−/−γc^−/− mice with highly purified NK cells from WT and IL-28R^−/− mice. After 5 d of reconstitution, B16F10 cells were injected i.v., and then lungs were harvested after 14 d. The mice reconstituted with IL-28R^−/− NK cells demonstrated significantly higher numbers of lung metastases compared with those transferred with WT NK cells (Fig. 4B). This data corroborated the LPS experiments (Fig. 2) showing that the IL-28R^−/− NK cells are intrinsically defective in vivo. In addition, a series of depletion and neutralization experiments were performed in the B16F10 experimental lung metastasis model using WT, IFN-γ^−/−, IL-28R^−/−, and IFNAR1^−/− mice (Fig. 4C). This experiment revealed the combinatorial effects of IL-28R or IFN-γ and IFNAR1 deficiency. Clearly, by contrast, additional IFN-γ blockade did not offer any further increase in metastases in the IL-28R^−/− mice than that observed in IL-28R^−/− or IFN-γ^−/− mice alone or WT mice treated with anti–IFN-γ mAb (Fig. 4C). Experiments in mice deficient in both IFNAR1 and IL-28R supported these findings (SI Appendix, Fig. S6). To support data in the B16F10 melanoma model, we also showed that IL-28R^−/− mice were defective in controlling experimental metastasis of RM-1 prostate carcinoma cells and that, once again, additional IFNAR1 blockade further enhanced metastases (Fig. 4D).

Fig. 4.

IL-28R^−/− and IFNAR1^−/− mice have decreased control of B16F10 and RM-1 experimental lung metastases. (A) Groups of five WT or gene-targeted mice were injected i.v. with B16F10 melanoma cells (dose as shown). (B) Groups of seven to eight RAG2^−/−γc^−/− recipients reconstituted for 5 d with 2 × 10⁵ IL-28R^−/− or WT NKs (sorted by TCRβ^neg, NKp46⁺, NK1.1⁺) and injected i.v. with B16F10 melanoma cells (5 × 10⁴). (C) Groups of five WT or gene-targeted mice were injected i.v. with B16F10 melanoma cells (5 × 10⁴). Some mice received cIg (250 μg), anti-IFNAR1 (250 μg), anti-IFN-γ (250 μg), or anti-asGM1 (100 μg) i.p. on days −1, 0, and 7 relative to tumor inoculation as indicated. (D) Groups of 5–10 WT or gene-targeted mice were injected i.v. with RM-1 prostate carcinoma cells (5 × 10³). Some mice received cIg (250 μg), anti-IFNAR1 (250 μg), or anti-IFN-γ (250 μg) i.p. on days −1, 0, and 7 relative to tumor inoculation as indicated. In A–D, 14 d after tumor inoculation, the lungs of these mice were harvested and fixed, and the number of B16F10 or RM-1 colonies was counted under a dissection microscope. Symbols in scatter plots represent the number of B16F10 or RM-1 tumor colonies in the lung from individual mice (with mean and SEM shown by cross-bar and errors). Mann–Whitney test was used to compare differences between the groups of mice as indicated (**P < 0.01; ***P < 0.001; ns, not significant).

We next assessed the antimetastatic activity of IFN-αβ and IFN-λ [pegylated IL-28A (PEG-IL-28A)], alone and in combination, in the B16F10 experimental metastases model (Fig. 5 A and B). Although IFN-λ alone did not significantly suppress B16F10 lung metastases, IFN-λ did enhance the antimetastatic activity of IFN-αβ when in combination (Fig. 5A). IFN-αβ and IFN-λ increased the survival of tumor-inoculated mice, and prolonged survival was observed with the combination (Fig. 5B). To further examine the specific activity of IFN-λ on NK cells in vivo, RAG2^−/−γc^−/− recipients receiving no transfer or WT or IL-28R^−/− NK cell transfers were then challenged with B16F10 and treated with mock or IFN-λ. Critically, this experiment showed that IL-28R^−/− NK cells were less protective than WT NK cells and that IFN-λ was able to enhance the antitumor effect of WT NK cells but was without effect on IL-28R^−/− NK cells or in mice that received no NK cell transfer (Fig. 5C).

Fig. 5.

IFN-λ (PEG-IL-28A) suppresses B16F10 metastases in an NK cell-dependent fashion. (A and B) Groups of five WT mice in each panel were injected i.v. with B16F10 melanoma cells (2 × 10⁵). Mice received PEG-IL-28A or mIFN-αβ (25 μg i.p.) daily (days 0–5). (C) Groups of 4–12 RAG2^−/−γc^−/− recipient mice were reconstituted for 5 d with 2 × 10⁵ IL-28R^−/− or WT NKs (sorted by TCRβ^neg, NKp46⁺, NK1.1⁺) and injected i.v. with B16F10 melanoma cells (2 × 10⁵). Some mice received no NK cell transfer. Mice then received mock or PEG-IL-28A (25 μg i.p.) daily (days 0–5). In A, 14 d after tumor inoculation, the lungs of these mice were harvested and fixed, and the number of B16F10 colonies was counted under a dissection microscope. In B and C, survival of mice was plotted, and statistical analysis was performed by Mantel–Cox Log-rank test; *P < 0.05, **P < 0.01, ****P < 0.0001.

Attempts to demonstrate a direct effect of IFN-λ (PEG-IL-28A) on NK cells in vitro did not yield any detectable phosphorylation of STAT1 or downstream cytokine production, even in the context of combinations of NK cell survival and activation factors such as IL-15, IL-12, and IL-18 (SI Appendix, Fig. S7). IFN-αβ was able to induce phosphorylation of STAT1 in the same assays; however, no additional effect on pSTAT1 induction was observed with the addition of both IFN-αβ and PEG-IL-28A (SI Appendix, Fig. S7). The PEG-IL-28A was active as demonstrated by its effect on Mx1 expression in B16F10 and RENCA tumor cells (SI Appendix, Fig. S8).

Lymphoma Growth Is Enhanced in IL-28R–Deficient Mice.

We then examined the capacity of NK cells to control target lymphoma cells in vivo. The in vivo tumor growth of MHC class I-deficient RMAs lymphoma cells in the peritoneum is controlled by NK cells (43). Transduction of RMAs lymphoma with a lentivirus containing luciferase-venus allows the detection of tumor progression in a kinetic manner (44). Using this model, we clearly demonstrated that IL-28R^−/− mice were defective in their in vivo control of RMAs tumor growth compared with WT mice (Fig. 6 A and B and SI Appendix, Fig. S9A). Control of RMAs lymphoma is often attributed to NK cell perforin and cytotoxicity (43). However, assessment of the cytotoxic capability of naive, in vitro IL-15–IL-15Rα complex-stimulated or in vivo Poly(I:C)-stimulated spleen NK cells from WT or IL-28R^−/− mice against classical targets (YAC-1 and B16F10) revealed no differences in cytotoxicity (SI Appendix, Fig. S10). NK cell incubation with PEG-IL-28A in vitro did not enhance NK cell-mediated target cell killing. In contrast, IFN-αβ was able to enhance killing of YAC-1 target cells in the same assays (SI Appendix, Fig. S10).

Fig. 6.

IL-28R is necessary for effective control of NK cell-sensitive lymphoma and MCA-induced sarcoma in vivo. (A) Enhanced RMAs (5 × 10⁴ Luc⁺ RMAs cells) lymphoma growth in vivo in IL-28R^−/− mice after i.p. inoculation is depicted by images taken every 5 d from one of these independent experiments in SI Appendix, Fig. S10A; ND, not determined. (B) IL-28R^−/− mice display decreased survival after i.p. challenge with 5 × 10² NK cell-sensitive RMAs lymphoma cells. Statistical analysis was performed using a Mantel–Cox test; ***P < 0.001, n = 10–13 per group. (C and D). Groups of 20 male C57BL/6 WT or gene-targeted mice as indicated were inoculated s.c. with 5 μg of MCA in corn oil and subsequently monitored for tumor development over 250 d. WT, IL-28R^−/−, or IFNAR1^−/− mice were treated with (C and D) control Ig (cIg) or (D) neutralized for IFNAR1 or IFN-γ (250 μg i.p. at day −1 and 0 and then weekly until week 8). Results are shown as survival curves defined as the percentage of tumor-free mice at each time point. Statistical differences in tumor incidence were determined by Mantel-Cox Log-rank test (*P < 0.05; **P < 0.01).

Carcinogen-Induced Tumor Initiation Is Prevented by IL-28R and IFNAR1.

Host protection from MCA-induced sarcoma is NK cell-dependent, and many host immune cell types and molecules, including IFNAR1, have been assessed in this mouse model (11, 28, 45). In concert with these findings, at a low dose of MCA carcinogen (5 μg), IFNAR1^−/− mice and IFN-γ^−/− mice treated with control Ig (cIg) displayed a significantly reduced survival after MCA inoculation compared with WT mice treated with cIg (Fig. 6C). IL-28R^−/− mice also demonstrated a strong trend toward reduced survival (P = 0.056) (Fig. 6D and SI Appendix, Fig. S9B). When the WT and IL-28R^−/− strains were additionally neutralized for IFNAR1 or IFN-γ, the additive effects of IL-28R deficiency and IFNAR1 deficiency were observed (Fig. 6D).

Discussion

NK cells play a key role in protecting against tumor initiation and metastasis and contribute to immunopathology during inflammation. Although the role of type I IFN in priming NK cells is well-recognized, the other signals that prime NK cells are not completely understood, and, in particular, the direct effect of type III IFN on NK cells is comparatively very poorly studied. We have taken advantage of the IL-28R–deficient mice, which cannot bind type III IFN (IFN-λ) to demonstrate in a series of in vivo experiments that both host IL-28R and, specifically, IL-28R expressed by NK cells contribute significantly to NK cell function in models of tumor control and bacterial-induced inflammation. IL-28R–deficient mice were resistant to LPS and cecal ligation puncture-induced septic shock, and important cytokines in these disease models were significantly altered in the absence of host IL-28R. IL-28R–deficient mice were more sensitive to experimental and spontaneous tumor metastasis, lymphoma growth, and carcinogen-induced tumor formation than WT mice, and additional blockade of IFNAR1, but not IFN-γ, further enhanced metastasis and tumor development. A combination of type I IFN and type III IFN was significantly more antimetastatic in the B16F10 model than either IFN alone. IFN-λ promoted the antimetastatic activity of WT NK cells, but not IL-28R^−/− NK cells. Importantly, specific loss of IL-28R in NK cells transferred into lymphocyte-deficient mice resulted in reduced LPS-induced IFN-γ levels and enhanced tumor metastasis. These data strongly suggest that IFN-λ can directly regulate NK cell effector functions in vivo, alone and in the context of IFN-αβ.

IFN type III displays similar signal transduction pathways as IFN type I but has a more restricted receptor expression (46). The type III IFN receptors, like IFNAR1, activate the ISGF3 complex, but, unlike the IFNARs, they are restricted in their tissue distribution, are not highly expressed in hematopoietic cells, and act predominantly at epithelial surfaces. A recent report correlating genomic alterations in colorectal cancer patients has shown that patients without lymph node metastases had significantly more amplification of several genes, including IFNAR1, IFNAR2, and IL-10R2 (shared by IL-10, IL-22, IL-26, and IL-28), than those with lymph node invasion (47). The highest levels of gene loss mutations were found for IL-15 whereas the IL-28R was the most frequently deleted cytokine receptor in immune cells (47). IL-15 is a critical NK cell activating and survival factor for both human and murine cells (48). By contrast, IL-28R expression and signaling in NK cells have remained controversial. Although we were able to detect IL-28R mRNA by RT-PCR in purified NK cells from WT mice, but not IL-28R–deficient mice, we were unable to demonstrate any modulation of STAT phosphorylation or downstream effector cytokine secretion by purified mouse NK cells exposed to PEG-IL-28A. Previously, by RT-PCR, it was identified that human NK cells can express IL-28R (23, 24); however, they demonstrated no responsiveness to recombinant IFN-λ as measured by phosphorylation of STAT1 and STAT3 in vitro (49). Demonstrating early signal transduction events via IL-28R even in cells that express significant levels of IL-28R is not straightforward (46). The interaction of multiple cytokines (e.g., IL-2, IL-15, IL-12, IL-18, or IL-21) may be required to allow NK cell responsiveness to a single cytokine (50, 51); however, this lack of a demonstrable direct effect of PEG-IL-28A on mouse NK cells in vitro occurred despite the presence of various mixtures of other NK cell stimulating and survival factors (e.g., IL-12, IL-18, IL-15, etc.). It remains possible that an alternative signaling pathway is triggered via IL-28R in NK cells or that we have not reproduced the optimal cytokine milieu in vitro to promote IL-28 activity on NK cells. Certainly, the altered function of IL-28R–deficient NK cells compared with WT NK cells in RAG2^−/−γc^−/− mice suggests a direct effect of ligand on IL-28R on NK cells. The IL-28R–deficient NK cells were less responsive to LPS in vivo and were ineffective in clearing B16F10 lung metastases. Both of these phenotypes could potentially be attributed to a reduction in NK cell IFN-γ secretion. One potential explanation might be that the loss of IL-28R on NK cells causes a compensatory increase in free IL-10R2 and enhanced signaling of IL-10. However, we did not observe any evidence of enhanced IL-10R2 expression or IL-10 signaling in IL-28R–deficient NK cells. Other receptors that use IL-10R2, such as IL-26, could not be assessed due to a lack of suitable reagents. The creation of a mouse IL-28R–specific antibody that can detect low levels of IL-28R expression will shed light on this question.

Regardless of the direct or indirect effects of IL-28 on NK cells, the combined effect of IL-28R and IFNAR1 deficiencies created a mouse as defective in NK cell-mediated tumor control as one lacking NK cells. Notably, neutralization of IFN-γ was largely without effect in the IL-28R–deficient mouse but increased tumor metastases in mice deficient for IFNAR1. These data suggest that a combination of type I and III IFN signaling contributes almost all NK cell-mediated control of tumor initiation and metastases whereas the major role of IFN-γ may be downstream of IL-28R, IL-12R, and IL-18R. Type I IFN has been used successfully for the treatment of several types of cancer, including hematological malignancies (hairy cell leukemia, chronic myeloid leukemia, and some B- and T-cell lymphomas) and solid tumors (melanoma, renal carcinoma, and Kaposi’s sarcoma) (52, 53). The antitumor effect of type I IFN therapy is accompanied by severe side effects, including autoimmune and inflammatory symptoms, as well as direct tissue toxicity, that are probably responsible for the hematological and neurological side effects. The use of recombinant IFN-λ (PEG-IL-28A) in combination with IFN-αβ suggested potential additive benefit against experimental tumor metastases, and the activity and safety of this combination may be further explored preclinically. Previous approaches using gene transfer of IFN-λ into tumor cells (46) are of less translational value and might not say much about the mechanism of action of combined soluble cytokines in vivo. In particular, NK cells are critical endogenous IFN-αβ targets during the development of protective antitumor responses (11, 54), and NK cells may require a combination of IFNAR1 and IL-28R signaling (direct or indirect) to be completely antimetastatic. One alternate mechanism is inhibited angiogenesis in vivo because type I and type III IFNs up-regulate Mig and IP-10, both of which suppress neoangiogenesis within tumors. In addition, IFN-λ has been shown to augment the expression of MHC class I molecules, which subsequently increased the expression levels of putative tumor antigens (55). Alternatively, several experimental models showed that activated NK cells were primarily responsible for IFN-λ–mediated antitumor effects (17, 19). However, an antitumor role of type I and III IFN via additional mechanisms, such as regulating tumor cell proliferation, apoptosis, and autophagy, needs to be explored. These studies suggest that mechanisms of type III IFN-mediated antitumor effects are dependent on the tumor model used and that many factors influence the type III IFN-induced activities. Combinatorial therapy using IFN-αβ and IFN-λ may achieve antimetastatic activity by inducing complementary mechanisms and engaging both IFNAR1 and IL-28R.

Materials and Methods

Mice.

C57BL/6J WT mice were purchased from the Walter and Eliza Hall Institute for Medical Research and housed at the QIMR Berghofer Medical Research Institute. C57BL/6 IL-28R^−/− mice, described by Ank et al. (21), were kindly provided by Bristol-Myers Squibb. C57BL/6 IFNAR1^−/−, Ifnγ^−/−, and RAG2^−/−γc^−/− mice have been previously described (11, 28, 56) and were bred at the QIMR Berghofer Medical Research Institute. IL-28R^−/− x IFNAR1^−/− mice were generated at the QIMR Berghofer Medical Research Institute by crossing the strains as above. These mice were maintained on a C57BL6 background at the QIMR Berghofer Medical Research Institute. All mice were used between the ages of 6 and 14 wk. All experiments were approved by the QIMR Berghofer Medical Research Institute animal ethics committee.

Cell Culture.

B16F10 melanoma and RM-1 prostate adenocarcinoma cell lines were cultured as previously described (42, 57) in Dulbecco's modified Eagle medium supplemented with 10% (vol/vol) heat-inactivated FCS (Thermo), glutamax (Gibco), and penicillin-streptomycin (Gibco). B16F10 were sourced from the American Type Culture Collection whereas RM-1 was obtained from Pamela Russell, Queensland University of Technology, Brisbane, Australia. YAC-1 (a Moloney murine leukemia virus-induced T-cell lymphoma of the A/Sn strain) and RMAs [a TAP2^neg/H-2b^neg variant of RMA cells (a Raucher virus-induced T-cell lymphoma RBL-5, H-2b⁺)] cell lines were cultured as previously described (58) in RPMI medium 1640 supplemented with 10% heat-inactivated FCS (Thermo), glutamax (Gibco), and penicillin-streptomycin (Gibco). The generation of RMAs stably transduced with luciferase was performed in the same growth medium with 8 μg/mL polybrene at 75% confluency with 10 multiplicity of infection of lentivirus carrying the venus-luciferase (v2luc) expression plasmid. V2luc was generated by inserting the luciferase coding sequence into the LeGO-iV2 parent vector and was kindly provided by Michael Milsom, German Cancer Research Center, Heidelberg, Germany. After 4 h of incubation at 37 °C, virus- and polybrene-containing medium was replaced with fresh complete growth medium. Cells were kept for an additional 48 h in culture and were subsequently fluorescence-activated cell sorted on the basis of venus expression. All cell lines were tested for Mycoplasma detection by the QIMR Berghofer Medical Research Institute’s scientific services.

In Vivo LPS Challenge.

As previous described (28), LPS (from E. coli 0127:B8; Sigma-Aldrich) suspended in PBS was injected intraperitoneally into mice at the described doses (0.10, 0.75, 1.00, or 1.25 mg/30 g mouse). For survival experiments, mice were checked hourly for symptoms of endotoxicosis. Serum from these mice was taken for cytokine analysis by retroorbital or cardiac bleeding. Spleens were also taken from mice after 6 h post-LPS injection to analyze CD69 and intracellular IFN-γ expression by NK cells.

In Vivo CLP-Induced Septic Shock.

CLP was performed as previously described (33). Briefly, mice were individually anesthetized by isoflurane, the abdomen was shaved and disinfected by betadine antiseptic spray, a midline incision was made, and 1 mL of saline was injected to prevent tissue dehydration. Cecum was externalized, and a 75% portion was ligated and punctured once using a 25-gauge needle to extrude a small amount of cecal content and induce a high-grade sepsis (100% mortality within 10 d). The cecum was returned to the abdomen, the peritoneum was closed via suture, and the skin was sealed using an auto clip wound clip applier (Becton Dickinson). Buprenorphin (Reckitt Benckiser Pharmaceutical) was applied at 0.05 mg per kg body weight at the incision site for postoperative analgesia.

NK Cell Activation in Vitro.

Spleens from the indicated strains of mice were stained with anti-NK1.1, anti-NKp46, and anti-TCRβ mAbs, and NK cells were sorted by FACS (BD FACSAria II; BD Biosciences). Two hundred thousand freshly purified NK cells were plated in 96-well U bottom plates in NK cell media (RPMI supplemented with 10% FCS, non essential amino acids, Pyruvate, Hepes, glutamax, 2-mercaptoethanol, penicillin/streptomycin) in the presence of rIL-10 (Biolegend), rIL-12 (eBiosciences), rIL-15/IL-15Rα complex (eBiosciences), rIL-18 (R&D Systems), and PEG-IL-28A (kindly donated by Sean Doyle, Zymogenetics, Seattle) for 24 h. For NK cell-mediated cytotoxicity assays, sorted NK cells were either cultured for 5 d in NK cell media with supplementation of 10 ng/mL of rIL-15/IL-15Rα complex (in vitro priming) or sorted from mice post 24 h Poly I:C (100 μg per mice) i.p. injection (in vivo priming). Target B16F10 or YAC-1 cells, labeled with 100 μCi/1 × 10⁶ cells of ⁵¹Cr, were cocultured for 4 h with the indicated ratio of primed NK cells.

Flow Cytometry Analysis.

Cells harvested from in vitro cultures or single cell suspensions from various organs were incubated for 15 min in Fc blocking buffer (2.4G2 antibody). Cells were then stained with the following antibodies: anti-mouse-CD3 (17A2), -CD11b (M1/70), -CD27 (LG.3A10), -CD43 (eBioR2/60), -CD69 (H1.2F3), -DNAM-1 (480.1), -IFN-γ (XMG1.2), -Ly49A (A1), -Ly49C/I (14B11), -NKG2D (CX5), -NKG2A/C/E (20d5), -NKp46 (29A1.4), -NK1.1 (PK136), and -TCRβ (H57-597). All of the mAbs were purchased from eBiosciences, BD Biosciences, or Biolegend. A Zombie Yellow or Zombie UV Fixable Viability Kit (Biolegend) was used to assess viability. Acquisition was performed using an LSR II Fortessa Flow Cytometer (BD Biosciences). Analysis was achieved using Flowjo (Treestar) software. For NK cell purification, spleen homogenates were first stained with Mouse NK Cell Isolation Kit II (Miltenyi Biotec) and enriched by a depleting program by an AutoMACS-Pro (Miltenyi Biotec). NK cell-enriched samples were then stained with NK1.1, NKp46, TCR-β, or viability stain and sorted with high purity (viable, NK1.1⁺. NKp46⁺, TCRβ^neg) using a FACS Aria II (BD Biosciences).

Cytokine Detection.

All cytokines from in vivo assays were detected using Cytometric Bead Array (CBA) technology (BD Biosciences) according to the manufacturer’s instructions. IFN-γ detection from purified NK cell supernatants from in vitro assays was measured by ELISA with the IFN-γ Duoset Kit (R&D Systems) according to the manufacturer’s instructions. For intracellular cytokine detection, isolated splenocytes from LPS-injected mice or in vitro-activated NK cells were stained for the indicated surface markers, fixed, and permeabilized using BD cytofix/cytoperm (BD Biosciences) and then stained with an anti–IFN-γ antibody (XMG1.2).

In Vivo Tumor Imaging.

Single cell suspensions of RMAs (5 × 10² to 5 × 10⁴) or RMAs-Luciferase⁺ cells (5 × 10⁴) were injected i.p. into the indicated strains of mice at day 0 (D0). Tumor burden was measured by bioluminescence imaging and expressed as photon flux (photons per second) as previously described (44). Luminescence was assessed at 5-d intervals by injection of 0.5 mg/mL d-luciferin (Everest) per mouse for 5 min, and luminescence measure for 1 min in a Xenogen IVIS Caliper (Perkin-Elmer). Overall survival was calculated in parallel to the imaging kinetics.

Tumor Metastasis and MCA-Induced Fibrosarcoma.

Single cell suspensions of RM-1 (5 × 10³) or B16F10 melanoma cells (5 × 10⁴ to 2 × 10⁵) were injected i.v. into the tail vein of the indicated strains of mice. Lungs were harvested on day 14, and tumor nodules were counted under a dissection microscope. On day −1, 0, and 7 after tumor inoculation, some mice were treated with i.p. injections of control Ig (cIg) (2A3, 250 μg i.p), anti–mIFN-γ mAb (H22, 250 μg i.p), anti-mIFNAR1 mAb (MARI-5A3, 250 μg i.p), or asGM1 mAbs (100 μg i.p each). For MCA carcinogen-induced fibrosarcoma formation, groups of 8–20 male WT, IFNAR1^−/−, IFN-γ^−/−, or IL-28R^−/− mice were injected s.c. on the right flank with 5 μg, 25 μg, or 300 μg of MCA and were monitored over 250 d for fibrosarcoma development. Data were recorded as the percentage of mice tumor-free. Some mice in these experiments were treated with control Ig (cIg) (2A3, 250 μg i.p), anti-mIFNAR1 mAb (MARI-5A3, 250 μg i.p), or asGM1 mAbs (100 μg i.p each) as indicated in the legends.

NK Cell Adoptive Transfer.

NK cells (2 × 10⁵) freshly purified (NK1.1⁺, NKp46⁺, TCRβ^neg) from WT or IL-28R^−/− mice were injected via the tail vein into RAG2^−/−γc^−/− mice. Five days later, mice were injected either i.p with LPS (0.1 mg/30 g mouse) or i.v with B16F10 melanoma cells (5 – 10⁴). The NK cell reconstitution of each mouse was analyzed by flow cytometry in the peripheral blood before and after the completion of each experiment, and we observed no differences between IL-28R^−/− and WT NK cell injected mice.

In Vivo Treatment with IFN-αβ and IFN-λ.

Single tumor cell suspensions of 2 – 10⁵ B16F10 were injected i.v. into the indicated strains of mice at day 0. Treatment groups consisted of rMOCK IFN-αβ, rIFN-αβ (50,000 U per mouse per day), PEG-IL-28A (25 μg per mouse per day), or both rIFN-αβ + PEG-IL-28A (same concentration per mouse per day). Treatments were applied i.p. daily from day 0 to day 5 post tumor inoculation, and overall survival was calculated. PEG-IL-28A was kindly provided by Sean Doyle, Zymogenetics, Seattle, and rMOCK IFN-αβ and rIFN-αβ were kindly provided by Antonella Sistigu, Institut Gustave Roussy, Paris.

Statistical Analysis.

Statistical analysis was achieved using GraphPad Prism Software V6. Data were considered to be statistically significant where the P value was equal to or less than 0.05. Statistical tests used were the unpaired Student’s t test, Mann–Whitney test, and the Mantel–Cox Log Rank test for survival.

Supplementary figures and legends are detailed in SI Appendix.