Antitumor NK activation induced by the Toll-like receptor 3-TICAM-1 (TRIF) pathway in myeloid dendritic cells

Retardation of Tumor Growth by NK Cells in polyI:C-Treated Mice.

PolyI:C is reported to induce type I IFN in vivo in mice and in vitro in mDCs (19, 20). Using a C57BL/6-B16 syngeneic tumor-implant model, we evaluated the antitumor activity of polyI:C, which was injected i.p. twice a week (Fig. 1A). Suppression of tumor growth, determined as reported in ref. 5, was observed in the group that received polyI:C compared with the saline-treated group. The retardation of tumor growth appeared to depend on polyI:C treatment but not on the level of MHC class I or a direct effect (such as apoptosis) on B16 cells (SI Figs. 8 and 9).

PolyI:C-mediated suppression of tumor growth was terminated when NK activity in mice was blocked by an injection of NK1.1 Ab (Fig. 1B) or asialo-GM1 Ab (data not shown). Thus, the polyI:C-mediated tumor suppression largely relies on the effector NK cells. Similar experiments by using another model (BALB/c vs. CT-26) and asialo-GM1 Ab confirmed the results obtained with the C57BL/6 vs. B16 model (data not shown).

Similar experiments by using the syngeneic mouse model were performed with CD8-eliminating Ab (Fig. 1C). Suppression of B16 tumor growth was not significantly changed by the treatment of mice with anti-CD8 Ab. CD8+ CTL-mediated tumor suppression is minimal, if any, in this model.

TICAM-1 and IFNAR Participate in NK Activation in Mice.

We then assessed the tumor cytotoxicity of NK cells isolated from polyI:C-administered C57BL/6 mice. PolyI:C was administrated i.p. to mice, and the spleen cells were collected as a source of NK cells. To eliminate contaminating NKT cells, negative selection was performed to isolate NK cells by using MACS beads. NK cells efficiently expressed tumoricidal activity against the same lot of B16 melanoma cells (Fig. 2A). YAC-1 (an NK target) cells were killed in a fashion similar to B16 cells (Fig. 2B), supporting the view that NK is the effector. B16 and YAC-1 were used to measure the cytotoxicity of spleen cells collected from the indicated polyI:C-treated knockout (KO) mice (Fig. 2C). A significant decrease of cytotoxicity was observed in spleen cells prepared from IFNAR^−/− or TICAM-1^−/− mice. The polyI:C-mediated YAC-1 cytotoxicity was reduced only slightly in spleen cells of MyD88^−/− mice. NK-mediated cytotoxicity increased in cells from PKR^−/− mice. Similar results were obtained with B16 cells (data not shown).

To further confirm the results of Fig. 2C, B16 cells were implanted and tumor growth was followed in vivo in these KO mice. Retardation of B16 tumor growth by polyI:C was observed in MyD88^−/−, IFN-β^−/−, and PKR^−/− mice at a level comparable to the control wild-type mice (Fig. 3). The antitumor activity of polyI:C diminished in TICAM-1^−/− or IFNAR^−/− mice (Fig. 3). Tumor regression by polyI:C was not affected by depletion of IFN-β, suggesting the participation of other IFNs (especially IFN-α) in the IFNAR-mediated antitumor response. These results infer that the antitumor activity by polyI:C is elicited through TICAM-1 and/or IFNAR in these mice. In the absence of IFNAR, no TLR3 was expressed by stimulation with polyI:C in mDCs (SI Fig. 10); therefore, the TLR3-TICAM-1 pathway does not function in IFNAR^−/− mice. Taken together, the final effector for tumor killing is evidently NK cells activated in association with TLR3-TICAM-1 in this mouse tumor-implant model (Figs. 1 B and C and 3).

Mechanisms of NK Activation by polyI:C-Stimulated mDCs.

The TICAM-1-dependent NK activation in polyI:C-treated mice was unexpected because polyI:C functions through multiple pattern-recognition receptors, i.e., PKR, RIG-I, and MDA5, in addition to TLR3 (21–23). All these receptors are induced by IFN-α/β in mDCs to recognize polyI:C in vitro (20). Although RIG-I and MDA5 were normally induced in TICAM-1^−/− mDCs (SI Fig. 11), polyI:C-mediated NK activation was abolished in TICAM-1^−/− mice (Fig. 3). We then aimed at clarifying the point that TICAM-1 in mDCs is responsible for the in vivo and in vitro antitumor NK activation by polyI:C.

TICAM-1 in mDCs participates in NK activation because adoptive transfer of mDCs with TICAM-1 overexpression (SI Fig. 12) also retarded tumor growth in wild-type mice (Fig. 4A). In vitro cytotoxic assay was performed with B16 cells and NK cells treated with TICAM-1-transduced mDCs (Fig. 4B). B16 cells were damaged by NK cells cocultured with polyI:C-treated mDCs, and to a lesser extent, with TICAM-1-transduced mDCs. Thus, mDCs expressing TLR3-TICAM-1 contribute to antitumor NK activation.

We next tested whether IFNAR in mDCs or other cells is important for mDC-mediated NK activation. IFNAR is distributed across mDCs and lymphocytes, including NK cells. The IFNAR^−/− mice essentially failed to respond to polyI:C for tumor regression. When IFNAR-positive mDCs were transferred into IFNAR^−/− mice, polyI:C-mediated antitumor NK activation was recovered in the IFNAR^−/− mice (Fig. 4C). In contrast, no recovery of polyI:C-mediated tumor regression was observed by supplementing TICAM-1^−/− mDCs into IFNAR^−/− mice with the tumor burden (Fig. 4C). Thus, mDCs lead to tumoricidal NK activation in response to polyI:C in vivo and in vitro, where mDC TICAM-1 and IFNAR actively are involved.

The mechanism whereby TICAM-1 induces NK activation then was analyzed. Rae-1 proteins are up-regulated by TLR stimuli on murine macrophages and serve as ligands for NKG2D on NK cells (24). We checked the level of Rae-1 in mDCs. The Rae-1 level was not altered in response to polyI:C stimulation (SI Fig. 10). Next, we measured the cytokine levels in the culture supernatant (sup) of polyI:C-stimulated mDCs prepared from gene-disrupted mice (Fig. 5A and B). IFNAR^−/− mDCs did not release IFN-α, whereas TICAM-1^−/− mDCs and conventional (prepared from wild-type mice) and MyD88^−/− mDCs released it by stimulation with polyI:C (Fig. 5A). There was only marginal up-regulation of IL-12 p40 (Fig. 5B) or mDC maturation (SI Fig. 13) by polyI:C in TICAM-1^−/− and IFNAR^−/− mDCs. However, mDCs are not merely a cell type producing IL-12 and type I IFN, and these cytokine profiles were somewhat discrepant in the serum of polyI:C-treated KO mice (Fig. 5 C and D). Furthermore, polyI:C-mediated tumor regression was not abolished by functional blockade of IL-12 in the B16 tumor implant model (SI Fig. 14), nor was the in vitro B16 tumor killing by NK cells cocultured with polyI:C-primed mDCs cancelled by the addition of IL-12-neutralizing Ab (Fig. 6A). The IL-12/IFN-α/β profiles in the serum are not consistent with the degrees of NK activation.

B16 killing due to NK cells activated by polyI:C-primed mDCs was impaired if spleen NK cells were cultured with polyI:C-primed mDCs in the transwell (Fig. 6B), suggesting that tumoricidal activity NK cells acquire via mDCs is attributable largely to cell-cell contact.

RIG-I and MDA5 as well as TLR3 and IFN-α/β are IFN-inducible genes that were up-regulated in the response of mDCs to polyI:C within 6 h (SI Fig. 11). The up-regulation was impaired in IFNAR^−/− mice but not TICAM-1^−/− mice (SI Fig. 11). We tested whether TICAM-1^−/− mDCs stimulated with polyI:C still retain the activity to induce antitumor NK cells (Fig. 6C). Only weak antitumor NK activity was detected when the spleen NK cells were preincubated with TICAM-1^−/− mDCs. Costimulator CD86 up-regulation in response to polyI:C also was impaired in the mDCs prepared from TICAM-1^−/− as well as IFNAR^−/− mice (SI Fig. 13). Thus, NK cell-binding molecules, rather than cytokines, are induced in mDCs through the TLR3-TICAM-1 pathway in response to exogenic dsRNA stimuli, which may explain the functional link between mDC TICAM-1 and NK activation.

Mice, Cells, and Reagents.

TICAM-1^−/− mice were backcrossed more than eight times to adapt the C57BL/6 background. PKR^−/− (33), TLR3^−/− (19), MyD88^−/− (34), IFN-β^−/− (35), and IFNAR^−/− mice (36) were provided by T. Taniguchi (University of Tokyo, Tokyo, Japan) and S. Akira (Osaka University). All of the mice were maintained under specific pathogen-free conditions in the animal facility of the Osaka Medical Center and the Graduate School of Medicine Hokkaido University. They were backcrossed with C57BL/6 mice 8 to 10 times before use. Animal experiments were performed according to the guidelines set by the animal safety center, Japan.

B16D8 cell line was established in our laboratory as a subline of B16 melanoma (37). This subline was characterized by its low or virtually no metastatic properties when injected s.c. into syngeneic C57BL/6 mice (37). The mouse YAC-1 (BALB/c origin) cell line was provided by Sumitomo (Osaka, Japan), as described in ref. 32. These cell lines were cultured in RPMI medium 1640/10% FCS. Mouse NK cell was isolated with MACS Beads (Miltenyi Biotec, Auburn, CA). Bone marrow-derived mDCs were prepared as reported in ref. 38 with minor modifications.

Tumor Challenge and polyI:C Treatment.

Mice were shaved at the flank and injected s.c. with 300 μl of 6 × 10⁵ syngeneic B16D8 melanoma cells in PBS. After 1 week, tumor volumes were measured at regular intervals by using a caliper. Tumor volume was calculated by using the formula: Tumor volume (cm³) = (long diameter) × (short diameter) × (thickness) × 0.4. PolyI:C was injected i.p. at a concentration of 250 μg per head. The treatments were started on day 10–14 (when average of tumor volume reached at 0.5–0.8 cm³) and were repeated twice a week.

To deplete CD8+ T cells and NK cells in vivo, mice were i.p. injected with hybridoma ascites of anti-CD8β mAb and anti-NK1.1 mAb (39). All doses of antibodies and treatment regimens were determined in preliminary studies by using the same lots of antibodies used for the experiments. Treatment was confirmed to deplete completely the desired cell populations for the entire duration of the study.

TICAM-1 Gene-Transfected mDC Therapy.

mDCs were transfected with mouse TICAM-1 cDNA by Lenti-viral transfection system (Invitrogen, Carlsbad, CA). The preparation and propagation of lentivirus were performed as follows (31). A sequence of hrGFP with the multicloning site with hrGFP was cloned from pIRES-hrGFP-1a vector (Stratagene, La Jolla, CA) and placed into the cloning site of pLenti6/V5-D-TOPO vector (Invitrogen) by TOPO-cloning system to add optimal restriction enzyme site to the 3′ terminal end of the target gene. This vector was named as pLenti-IRES-hrGFP. We used this empty vector as control. More than 50% of mDCs were GFP-positive after treatment with pLenti-IRES-hrGFP (data not shown).

Mouse TICAM-1 cDNA was cloned as described in ref. 4. In most experiments, a cDNA encoding the N-terminal region of TICAM-1 (1–550 aa) (40) was used instead of the full-length TICAM-1 cDNA, because the full-length TICAM-1 expression led to cell apoptosis within 8 h after transfection. The cDNAs were subcloned and sequenced for confirmation. The cDNAs were ligated into pLenti-IRES-hrGFP vector. The virus dose was determined so as to reach 50% GFP expression in mDCs 36 h after infection (31). Transfection efficiency was checked by using mDCs. Briefly, percent GFP-positive cells was estimated by flow cytometry or counting the cells under illumination of a fluorescence microscope as reported in ref. 41. Virus particles were prepared for transfection according to the manufacturer's protocol.

Assessment of NK Cytolytic Activity.

Cytolytic activity of splenocytes and purified NK cells derived from polyI:C-treated mice was determined by ⁵¹Cr-release assay. Mice were i.p. injected with 250 μg of polyI:C. After 16 h, mice were killed and splenocytes were isolated by using Lympholyte-M. NK cells were purified with MACS-negative selection beads. Target cells were labeled with ⁵¹Cr for 3 h at 37°C, then washed and coincubated with effector cells at the indicated lymphocyte-to-target cell ratio in V-bottom 96-well plates in a total volume of 200 μl of RPMI medium 1640. Cytotoxicity was determined by measuring the ⁵¹Cr radioactivity released in 100 μl of the supernatant harvested from the plate after 8 h of incubation at 37°C (32). The percentage of specific lysis was calculated by using the formula: %Specific lysis = [(experimental release – spontaneous release)/(total release – spontaneous release)] × 100.

Supporting Information.

For additional details, see SI Materials and Methods and SI Table 1.