NK Cells Require Cell-Extrinsic and -Intrinsic TYK2 for Full Functionality in Tumor Surveillance and Antibacterial Immunity

This article is featured in In This Issue, p.1647

Natural killer cells are effector lymphocytes of the innate immune system and are characterized by their strong cytotoxic activity against infected and transformed cells. NK cell effector functions are tightly regulated by several mechanisms, including activating and inhibitory NK cell receptor and cytokine signaling (1). Most of the cytokines that act on NK cells signal through the JAK/STAT pathway (2). All STAT family members positively or negatively regulate NK cell activities, although underlying mechanisms are just beginning to emerge (3). Little is known about the impact of the individual JAK family members (JAK1-3 and tyrosine kinase 2 [TYK2]). Jak1^−/− and Jak2^−/− mice die soon after birth and during embryonic development, respectively (4–6). Conditional deletion of JAK2 in adult mice revealed a critical role of JAK2 in the maintenance of peripheral NK cell numbers and their maturation state (7). Treatment of mice with the JAK2-specific inhibitor BSK805 or the JAK1/JAK2 inhibitor ruxolitinib mimics NK cell defects upon conditional deletion of JAK2 and results in accelerated metastasis of transplanted breast cancer cells (7). Ruxolitinib treatment of patients suffering from myeloproliferative neoplasms impairs NK cell proliferation, maturation, and cytolytic capacity (8). Jak3^−/− mice and mice with a Jak3 loss-of-function mutation fail to develop NK cells (9–11), a phenotype that is recapitulated in patients bearing Jak3 mutations (12, 13). NK cells from Tyk2^−/− mice fail to produce IFN-γ in response to IL-12 and/or IL-18 and have an impaired early control of Leishmania major infections (14, 15). Defective IFN-γ production by NK cells in response to IL-12/IL-18 cotreatment has been described in TYK2-deficient patients (16), albeit with one exception (17). TYK2 is also involved in signaling by type I IFNs, IL-23, and IL-10, which directly act on NK cells but also on other cell types that might affect NK cell development and activity, such as dendritic cells (DCs), macrophages, or stromal cells (18). The first evidence for a role of TYK2 in NK cell–mediated tumor surveillance was provided by studies using the Abelson murine leukemia virus–induced B cell tumor model (19), which was later confirmed with tumor transplants (20). NK cells from Tyk2^−/− mice show reduced maturation and cytotoxicity and produce considerably less IFN-γ upon NK cell activating receptor (actR) stimulation than wild-type (WT) NK cells, although the underlying mechanisms remain unclear (19, 20). JAK/STAT signaling components are frequently altered in cancers, and considerable effort is directed toward the development of specific inhibitors for tumor therapy (21–24). TYK2 inhibitors are considered for the treatment of inflammatory diseases and tumor therapy (25–32). However, to realize their full potential and to identify harmful side effects, a better understanding of how TYK2 regulates NK cell function and antitumor activity is required.

In this study, we made use of mice that lack TYK2 selectively in NK cells to discriminate between direct and indirect effects of TYK2 on NK cell functionality. We demonstrate that TYK2 in CD11c⁺ and other non–NK cells regulates NK cell maturation, target cell cytotoxicity, and tumor surveillance, whereas TYK2 in NK cells is required for IL-12 signaling and an efficient immune defense against the intracellular bacterium Listeria monocytogenes. We furthermore provide evidence that NK cell–extrinsic, TYK2-dependent signals increase the competence of NK cells to produce high amounts of IFN-γ in response to actR stimulation and provide, to our knowledge, the first indication for an involvement of TYK2 in Ifng promoter demethylation.

All animal experiments were approved by the Ethics and Animal Welfare Committee of the University of Veterinary Medicine Vienna and the national authority (Austrian Federal Ministry of Science and Research) according to §§ 26ff. of Animal Experiments Act, Tierversuchsgesetz 2012: TVG 2012 (BMWF-68.205/0218-II/3b/2012, BMWFW-68.205/0032-WF/II/3b/2014, BMWFW-68.205/0103-WF/V/3b/2015, BMWFW-68.205/0212-WF/V/3b/2016).

WT, C57BL/6N mice were purchased from Janvier Labs. Tyk2^−/− (B6.129P2-Tyk2^tm1Biat and B6.129P2-Tyk2^tm1.1Biat), Tyk2^fl/fl (B6.129S2-Tyk2^{tm1(fl/fl)Biat}), Tyk2^ΔDC (B6.129P2-Tyk2^{tm1[fl/fl] Biat}-Tg[Itgax-cre]^1.1Reiz), and Tyk2^ΔM (B6.129P2-Tyk2^{tm[fl/fl]Biat}-Lyz2^tm[cre]Ifo/J) mice were described previously (33, 34). To generate mice that lack TYK2 in NK cells (Tyk2^ΔNK), Tyk2^fl/fl mice were crossed to Ncr1-iCre–Tg (7BL/6N-Tg[Ncr1-icre]^265Sxl) mice (35). B6.SJL-Ptprc^a (Ly5.1) mice were purchased from Charles River Laboratories, β2-microglobulin^−/− ([β2m^−/−]; B6.129-B2m^tm1Jae) mice (36) were kindly provided by Wilfried Ellmeier (Medical University of Vienna, Austria). All mice were on C57BL/6 background and bred at University of Veterinary Medicine Vienna under specific pathogen-free conditions. Age- and sex-matched (6–12 wk) mice were used for all experiments. For in vivo experiments, male mice (age 8–12 wk) were used. Murine lymphoma cell lines RMA-S (37) and RMA-Rae1 (38) were cultured in RPMI 1640 complete medium (R8758; Sigma-Aldrich) supplemented with 10% heat-inactivated FCS (Invitrogen), 100 U/ml penicillin, 100 μg/ml streptomycin (both from Sigma-Aldrich), and 50 μM 2-ME (Life Technologies).

A total of 1 × 10⁶ RMA-Rae1 tumor cells were s.c. injected into both flanks of mice depilated 3–4 d prior to tumor cell injection. Tumor growth was monitored every other day. After 10–13 d, the body and tumor weight were assessed.

Mice were infected i.p. with 5 × 10⁵ CFU L. monocytogenes strain EGD in 200 μl of PBS or were mock infected with PBS. Survival of mice was monitored for 2 wk. To determine bacterial burden, spleens and livers were harvested on day 5 postinfection (p.i.) and homogenized in PBS. Serial dilutions of homogenates were plated on Oxford agar plates (Biolife), and colonies were counted after 48 h growth at 37°C.

WT and Tyk2^−/− mice were injected i.p. with recombinant murine (rm) IL-15 and IL-15Rα–Fc (both R&D Systems), which were preincubated for complex formation, as previously described (39), or PBS as a control. Injections were given every 2–3 d for 2 wk (four doses). Two days after the last injection, splenic NK cells were analyzed for the expression of maturation markers, or isolated splenocytes were analyzed for IFN-γ in response to anti-NK1.1 Ab stimulation as described below.

NK cells from in vitro cultures and splenic single-cell suspensions were stained with the following Abs (all from eBioscience) against: CD16/CD32 (clone 93), CD49b (DX5), NK1.1 (PK136), NKp46 (29A1.4) CD3ε (145-2C11), CD3 (17A2), TCRβ (H57-597), CD8a (53-6.7), CD11c (N418), KLRG1 (2F1), CD27 (LG.7F9), CD11b (M1/70), MHC class II (M5/114.15.2), Ly6G (1A8), Ly6C (HK1.4), F4/80 (BM8), IFN-γ (XMG1.2), Ly5.2 (clone 104), and T-bet (eBio4B10). Biotinylated Ab to IL-15Rα and the isotype control were purchased from R&D systems. Intracellular T-bet and IFN-γ levels were analyzed using Foxp3/Transcription Factor Staining Buffer Set (eBioscience) according to manufacturer’s instructions. Analyses were performed on a FACSCanto II (BD Biosciences) and analyzed using BD FACSDiva software version 8.0 or on a CytoFLEX (Beckman Coulter) and analyzed using CytExpert version 2.2.0.97.

Ly5.1 recipient mice were lethally irradiated (9 Gy) 24 h prior to transplantation. Bone marrow (BM) of donor WT, Tyk2^−/−, and Ly5.1 mice was isolated and depleted of mature T and NK cells using CD5- and DX5-labeled MACS beads (Miltenyi Biotec) according to the manufacturer’s instructions. A total of 4 × 10⁶ of WT or Tyk2^−/− BM cells were mixed at a 1:1 ratio with Ly5.1 BM cells and i.v. injected into Ly5.1 recipient mice. Spleens of recipient Ly5.1 animals were isolated 6 wk after receiving the donor BM and NK cell maturation, and T-bet levels were analyzed by flow cytometry.

NK cells were isolated from splenic single-cell suspensions from Tyk2^fl/fl, Tyk2^ΔNK, Tyk2^ΔM, Tyk2^ΔDC, and Tyk2^−/− mice using DX5-labeled MACS beads according to the manufacturer’s instructions (Miltenyi Biotec). NK cells were expanded in RPMI 1640 complete medium supplemented with 5000 U/ml IL-2 (Proleukin; Novartis) or 100 ng/ml IL-15 (eBioscience) for 3–7 d. Purity of NK cells was assessed by flow cytometry and was typically 70–85% CD3ε⁻NK1.1⁺ cells of all living cells. Expanded NK cells were stimulated with 5 ng/ml recombinant mouse IL-12 (R&D Systems) for 10 min and used for Western blot analysis. For the analysis of actR-induced IFN-γ production, test tubes were precoated with 10 μg/ml of anti-NK1.1 Ab (PK136) (BioLegend or purified from the HB-191 hybridoma cell line) 1 d prior to the experiment. Splenocytes or cultured NK cells were seeded at 5 × 10⁶ or 5 × 10⁵ cells per tube, respectively, in RPMI 1640 complete medium. For the analysis of IL-12–induced IFN-γ production, 5 × 10⁶ splenocytes or 5 × 10⁵ cultured NK cells were seeded on uncoated tubes in RPMI 1640 complete medium with 5 ng/ml IL-12 (R&D Systems) and 5000 U/ml IL-2. Brefeldin A (eBioscience) was added after 1 h of incubation at 37°C. After additional 4 h of incubation, cells were stained for CD3ε, DX5, NKp46, and IFN-γ, and the frequency of IFN-γ⁺ NK cells (CD3ε⁻DX5⁺ or CD3ε⁻NKp46⁺) was assessed by flow cytometry. For the analysis of IL-15Rα surface-levels, spleens were digested by collagenase D (1 mg/ml; Roche Applied Sciences) and DNase I treatment (20 μg/ml; Roche Applied Sciences).

In vitro cytotoxicity assays were performed as previously described (40) using RMA-S or RMA-Rae1 as target cells. For in vivo cytotoxicity assays, WT and β2m^−/− splenocytes were isolated, stained with different concentrations of CFSE, and mixed at a 1:1 ratio (input sample). Recipient mice were i.v. injected with input cells (1 × 10⁷ cells per mouse), and the spleens of recipient mice were isolated after 16 h. Specific lysis was assessed by flow cytometry by the analysis of the β2m^−/− (CFSE^high): WT (CFSE^low) cell ratio in input sample and splenocytes of recipient mice.

RNA from DX5-depleted splenocytes (flow) and DX5 MACS-purified NK cells was isolated using peqGOLD TriFast reagent (PEQLAB), and reverse transcription was performed using iScript First Strand cDNA Synthesis Kit (Bio-Rad). The quantitative PCR was performed with a Stratagene MX3000 instrument (Agilent Technology, Boebelingen, Germany) as described previously (41) using Ube2d2 as housekeeping gene. The following primers were used: Ube2d2-forward 5′-AGG TCC TGT TGG AGA TGA TAT GTT-3′, Ube2d2-reverse 5′-TTGGGAAATGAATTG TCA AGA AA-3′, Ube2d2-probe 5′-CCA AAT GAC AGC CCC TAT CAG GGT GG-3′, Tyk2-forward 5′-GGG TCA CCT TCA GCC AGA CA-3′, Tyk2-revers 5′-GAC CTT AGC CTG TGC ATT GTA GAG T-3′, Tyk2-probe 5′-CAC ATC GCA CAC AAA GTC GGC ATC A-3′.

Cell lysis, SDS-PAGE, and Western blots were performed as described previously (41). The detection of chemiluminescence was performed using ChemiDoc Touch Imaging System (Bio-Rad) and analyzed by Image Lab software version 5.2 (Bio-Rad). For the analysis of TYK2, cells were lysed in 1× Laemmli sample buffer. The following Abs were used: anti–phospho-STAT3 (Tyr705, no. 9131), anti-STAT3 (no. 9132S), and anti-STAT4 (clone C46B10, no. 2653) from Cell Signaling Technology; anti–phospo-STAT4 (Tyr693, clone 38/pSTAT4) and anti-panERK (clone 16/ERK) from BD Transduction Laboratories. The anti-TYK2 Ab was custom made and described previously (41).

NK cells were DX5 MACS purified and FACS sorted (CD3ε⁻NK1.1⁺) using FACS Aria II (BD Biosciences). Genomic DNA was isolated using the QIAamp DNA Mini Kit (Qiagen). The DNA methylation status of the Ifng locus was analyzed by Varionostic GmbH, Germany, using direct pyrosequencing. The following primers were used (see Fig. 6C): 400 bp Ifng promoter fragment: forward: 5′-AGAGAATTTTATAAGAATGGTATAGGTG-3′, reverse: 5′-Biotin-CATAAAAAAAACTACAAAACCAAAATACA-3′; sequencing primer (S)1 (for CpGs −212, −198 and −178): 5′-ATGGTATAGGTGGGTA-3′; S2 (for CpGs −58, −50, −39, and +12): 5′-AAAAAAAATTAAAAAAAAATTTGTG-3′; S3 (for CpGs +91 and +114): 5′-ATGGTATAGGTGGGTA-3′. Methylation levels were analyzed with ProMarkQ24 software. Annotation of the transcriptional start site (TSS) and the translational start site is according to Ensemble genome browser (42) release 90 (ENSMUST00000068592.4) and consistent with a previous report (43).

DX5 MACS-purified IL-2–expanded Tyk2^−/− and Tyk2^ΔNK NK cells were treated with 5-aza-2-deoxycytidine (5-Aza) (Sigma-Aldrich) at 10 μM daily for 72 h with an untreated control, as described before (44). Cells were stimulated with anti-NK1.1 or IL-12 as described above, and frequency of IFN-γ⁺ NK cells was assessed by flow cytometry. Cell viability was determined with flow cytometry using the LIVE/DEAD fixable dye eFluor 780 (eBioscience).

One-way ANOVA with Bonferroni post hoc test, two-way ANOVA with Tukey post hoc test, unpaired t test, or log-rank (Mantel–Cox) test (for Kaplan–Meier plot) were performed using GraphPad Prism version 7.0 for Mac (GraphPad Software). Statistical significances are indicated for each experiment (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

To analyze NK cell–intrinsic function of TYK2, we generated mice with an NK cell–specific deletion of TYK2 (Tyk2^ΔNK) by crossing Tyk2^fl/fl mice (34) with Ncr1-iCre–Tg mice (35). Deletion of TYK2 in NK cells from Tyk2^ΔNK mice was analyzed at both protein and mRNA level in MACS-enriched NK cells (Fig. 1A, 1B). TYK2 protein was barely detectable in NK cells from Tyk2^ΔNK mice (Fig. 1A), and mRNA levels were reduced to around 15% compared with NK cells from Tyk2^fl/fl mice (Fig. 1B). NK cells develop in the BM from NK precursors to immature NK cells and mature NK cells before migrating to the periphery where they undergo full maturation and acquire functional competence (45–47). Four distinct NK cell maturation stages are defined by the cell surface markers CD27 and CD11b: CD27⁻CD11b⁻, CD27⁺ CD11b⁻, CD27⁺ CD11b⁺, and CD27⁻CD11b⁺ (from least to most mature NK cells). Complete TYK2 deficiency does not affect NK cell development and peripheral NK cell numbers but results in a reduced frequency of the most mature NK cell population (20). We did not observe differences in the maturation of NK cells from Tyk2^ΔNK and Tyk2^fl/fl mice (Fig. 1C). In line with this, NK cells from Tyk2^ΔNK mice had similar levels of the inhibitory lectin-like receptor KLRG1 as NK cells from Tyk2^fl/fl mice (Fig. 1D, Supplemental Fig. 1A) and only modestly reduced levels of the transcription factor T-bet (Fig. 1E), both of which are highly expressed in the most mature NK cell population (48). Similar to Tyk2^−/− mice (20), Tyk2^ΔNK mice did not differ from control mice with respect to the total abundance of splenic NK cells (Supplemental Fig. 1B).

Ncr1-driven Cre recombinase expression starts at the immature NK cell stage (Lin⁻CD122⁺DX5⁻NK1.1⁺) and remains thereafter (35). To exclude the possibility that TYK2-dependent signals prior to Cre recombinase–mediated deletion impact peripheral NK cell maturation, we performed mixed BM chimera experiments. WT or Tyk2^−/− Ly5.2 BM cells were mixed at a 1:1 ratio with WT Ly5.1 BM and adoptively transferred into lethally irradiated Ly5.1 mice. Splenic NK cell maturation was analyzed before (input) and 6 wk after transplantation (Fig. 2A). Tyk2^−/− BM cells repopulated with the same efficiency as WT cells (Supplemental Fig. 2A). Transfer of Tyk2^−/− BM into WT mice restored the frequency of Tyk2^−/− CD27⁻CD11b⁺ and KLRG1⁺ NK cells (Fig. 2B, 2C, Supplemental Fig. 2B) and T-bet levels in Tyk2^−/− NK cells to those observed in WT NK cells (Fig. 2D). Thus, TYK2 in the environment rather than in NK cells themselves promotes NK cell maturation.

Next, we sought to determine which cell type requires TYK2 to enable full NK cell maturation. Macrophages/monocytes (49–51), CD11c⁺ DCs (52), and neutrophils (53) have been implicated in the regulation of NK cell maturation. Hence, we tested a potential role of TYK2 in these cell types by using mice that lack TYK2 in LysM-expressing cells (i.e., monocytes, macrophages, and neutrophils [Tyk2^ΔM]) or in CD11c⁺ cells (i.e., mainly DCs [Tyk2^ΔDC]) (34). NK cells from Tyk2^ΔM mice did not display significant differences as compared with those derived from Tyk2^fl/fl mice (Fig. 3A, 3B). In contrast, the frequency of CD27⁻CD11b⁺ (Fig. 3A) and KLRG1⁺ (Fig. 3B) NK cells was significantly reduced in Tyk2^ΔDC mice, albeit to a lesser extent than in Tyk2^−/− mice. NK cell homeostasis and maturation critically depends on IL-15 (54, 55). As it has long been suggested that accessory cells use IL-15Rα to trans-present IL-15 to developing NK cells (50, 51, 56), we analyzed IL-15Rα surface levels in Tyk2^fl/fl, Tyk2^ΔM, Tyk2^ΔDC, and Tyk2^−/− mice. IL-15Rα levels were strongly reduced on macrophages and DCs but not on monocytes, NK cells, and CD8⁺ T cells from Tyk2^−/− compared with Tyk2^fl/fl mice (Fig. 3C–E, Supplemental Fig. 3A–D). DCs from Tyk2^ΔDC mice had a similar reduction in IL-15Rα levels as Tyk2^−/− mice, suggesting that DC-intrinsic TYK2 promotes IL-15Rα surface expression (Fig. 3C, Supplemental Fig. 3A, 3B). In contrast, macrophages from Tyk2^ΔM mice had similar IL-15Rα surface levels as Tyk2^fl/fl mice, suggesting that cell-extrinsic TYK2 signaling regulates IL-15Rα on macrophages (Fig. 3E). Having established that Tyk2^−/− mice show reduced IL-15Rα surface levels on macrophages and DCs, we next tested whether treatment with IL-15 rescues the NK cell maturation defects in Tyk2^−/− mice. Sustained treatment of Tyk2^−/− mice with IL-15/IL-15Rα complexes restored the abundance of both CD27⁻CD11b⁺ (Fig. 3F) and KLRG1⁺ NK cells to WT levels (Fig. 3G). Taken together, these data indicate that TYK2 in DCs and other cell types drives NK cell maturation and that this function of TYK2 can be compensated for by IL-15/IL-15Rα treatment.

Absence of TYK2 in humans and mice results in impaired IL-12 signaling and IFN-γ production in NK cells and T cells (16, 18). In vitro expanded NK cells from Tyk2^ΔNK mice showed strongly reduced phosphorylation of STAT3 and STAT4 in response to IL-12 compared with NK cells from Tyk2^fl/fl mice (Fig. 4A). The defect was less pronounced than in NK cells from Tyk2^−/− mice (Fig. 4A), which most likely reflects the presence of minor amounts of non–NK cells in MACS-purified and IL-2–expanded NK cell cultures. Ex vivo treatment of NK cells with IL-12 showed a similar impairment of IFN-γ production of NK cells from Tyk2^ΔNK and Tyk2^−/− mice compared with NK cells from Tyk2^fl/fl mice (Fig. 4B), confirming that IL-12 signaling requires TYK2 in NK cells themselves. To test IL-12 responsiveness of NK cells in vivo, we infected mice with L. monocytogenes and analyzed IFN-γ production at day 3 p.i. The frequency of splenic IFN-γ–producing NK cells (CD3ε⁻NK1.1⁺NKp46⁺) was profoundly reduced in Tyk2^ΔNK and Tyk2^−/− compared with Tyk2^fl/fl mice (Fig. 4C). Tyk2^−/− mice have a strongly increased susceptibility to L. monocytogenes, which has been associated with an impaired CD8⁺ T cell cytotoxicity and IFN-γ production (57). Tyk2^ΔNK mice succumbed to L. monocytogenes significantly earlier and showed a lower survival rate than Tyk2^fl/fl mice (Fig. 4D). Bacterial burden in spleen and liver was significantly higher in Tyk2^ΔNK than in Tyk2^fl/fl mice, albeit lower than in Tyk2^−/− mice at day 5 p.i. (Fig. 4E). In line with the differences in bacterial burden, the survival rate of Tyk2^−/− and Tyk2^ΔNK mice was 25 and 75%, respectively, compared with Tyk2^fl/fl mice at day 5 p.i. (Fig. 4F). Collectively these data indicate that TYK2 in NK cells is required for IL-12 signaling and IFN-γ production by NK cells during L. monocytogenes infection and contributes to an efficient immune defense against L. monocytogenes.

We next assessed the ability of NK cells from Tyk2^ΔNK mice to lyse tumor cells in vitro. IL-2–expanded NK cells from Tyk2^ΔNK mice killed RMA-S and RMA-Rae1 cells as efficiently as those from Tyk2^fl/fl littermate controls, whereas, consistent with previous reports (19, 20), NK cells from Tyk2^−/− mice had strongly reduced cytotoxic activity (Fig. 5A, 5B). Cultivation in the presence of IL-15 restored cytotoxic activity of NK cells from Tyk2^−/− mice (Fig. 5C), indicating that IL-15 not only restores maturation but also functional defects of Tyk2^−/− NK cells. To test NK cell cytotoxic activity in vivo, we injected splenocytes from mice deficient for β2m^−/− mice, which are recognized and lysed by NK cells because of low MHC class I (MHC I) expression (58, 59). β2m^−/− cells were eliminated in Tyk2^ΔNK mice as efficiently as in Tyk2^fl/fl mice, whereas Tyk2^−/− showed severely impaired rejection (Fig. 5D). In line with the cytotoxicity assays, Tyk2^ΔNK mice controlled the growth of transplanted RMA-Rae1 tumor cells as efficiently as Tyk2^fl/fl mice (Fig. 5E). Consistent with our previous report (20), tumor growth was significantly increased in Tyk2^−/− mice (Fig. 5E). NK cell activity is controlled by the balance of signals derived from actR and inhibitory NK cell receptors. Although TYK2 is not involved in signaling downstream of actR (60), such as NK1.1 and Ly49H, Tyk2^−/− NK cells have impaired IFN-γ production in response to anti-NK1.1 Ab stimulation (20). Unlike NK cells from Tyk2^−/− mice, those from Tyk2^ΔNK mice showed unimpaired anti-NK1.1 Ab–induced IFN-γ production (Fig. 5F). Thus, in contrast to IL-12, actR-induced IFN-γ production does not depend on TYK2 in NK cells themselves. Reminiscent of NK cell maturation and cytotoxicity, IFN-γ production in response to NK1.1 activation was rescued by treatment of Tyk2^−/− mice with IL-15/IL-15Rα complexes (Fig. 5G). Having established that TYK2 in NK cells themselves is of little importance for NK cell–dependent target cell killing and tumor surveillance, we next assessed the ability of Tyk2^ΔM and Tyk2^ΔDC mice to kill target cells in vivo and control RMA-Rae1 tumor growth. Tyk2^ΔM and Tyk2^ΔDC mice showed comparably impaired killing of β2m^−/− splenocytes, albeit the defect in Tyk2^ΔDC mice was less pronounced than in Tyk2^−/− mice (Fig. 5H). Tyk2^ΔM mice were able to control RMA-Rae1 tumor cells as efficiently as Tyk2^fl/fl controls, whereas Tyk2^ΔDC mice showed an intermediate tumor growth between Tyk2^fl/fl and Tyk2^−/− mice (Fig. 5I). We did not observe differences in the production of IFN-γ in response to actR stimulation between NK cells from Tyk2^fl/fl, Tyk2^ΔM, and Tyk2^ΔDC mice (Fig. 4J).

Taken together, these data show that cell-extrinsic TYK2 promotes NK cell cytotoxicity, tumor surveillance, and actR-induced IFN-γ production and that TYK2 signaling in myeloid cells and DCs augments NK cell–dependent target cell killing in vivo.

Human NK cells acquire the competence to produce IFN-γ in response to actR engagement during maturation through progressive demethylation at the Ifng locus (61). To test whether differences in DNA methylation cause the difference in IFN-γ production upon actR stimulation between WT and Tyk2^−/− NK cells, we expanded NK cells for 72 h in the presence of 5-Aza, a drug that blocks DNA methylation during cell proliferation. WT and Tyk2^−/− NK cells showed similar survival under these conditions (Supplemental Fig. 4). 5-Aza treatment restored anti-NK1.1 Ab–induced IFN-γ production by Tyk2^−/− NK cells to levels produced by WT NK cells (Fig. 6A). Importantly, IL-12–induced IFN-γ production remained strongly reduced (Fig. 6B), arguing against a general enhancement of Ifng gene induction upon global demethylation. We next analyzed the methylation status of nine CpGs located around the TSS of the Ifng gene in NK cells from Tyk2^fl/fl, Tyk2^ΔNK, and Tyk2^−/− mice (Fig. 6C). We found a significant increase in methylation of four CpGs (−212, −58, +91, and +114) in NK cells from Tyk2^−/− compared with those from Tyk2^fl/fl and Tyk2^ΔNK mice (Fig. 6D). Although not conserved in humans, CpGs +91 and +114 are in the same region as CpGs +122 and +128 at human Ifng promoter, which undergo demethylation during NK cell maturation (61). Consistent with the maturation-independent low level of methylation of the CpG at position −53 observed in human NK cells (61), CpGs in the proximal Ifng promoter (−198, −178, −58, −50, −39) were mostly demethylated in NK cells from Tyk2^fl/fl, Tyk2^ΔNK, and Tyk2^−/− mice (Fig. 6D). Collectively, these data suggest that impaired actR-induced IFN-γ production by NK cells from Tyk2^−/− mice is due to hypermethylation at the Ifng gene locus and most likely reflects defective terminal maturation.

In this study, we provide evidence that NK cells require at least two distinct TYK2-dependent signaling inputs to be fully functional in both tumor surveillance and antibacterial immunity: an NK cell–extrinsic input that drives NK cell maturation, cytotoxicity, and IFN-γ production in response to actR stimulation and an NK cell–intrinsic input that drives IFN-γ production in response to IL-12 and contributes to the immune defense against bacterial infection.

NK cells from Tyk2^−/− mice develop normally in the BM but show impaired terminal maturation in the spleen (20). Using mice that lack TYK2 specifically in NK cells (Tyk2^ΔNK) and mixed BM chimera experiments, we show that TYK2 in accessory cells rather than in NK cells themselves drives splenic NK cell maturation. Several lines of evidence support the idea that IL-15 trans-presentation by CD11c⁺ DCs and LysM⁺ cells (i.e., mainly monocytes, macrophages, and neutrophils) is required for peripheral NK cell homeostasis and terminal maturation into the CD27⁻CD11b⁺ NK cell stage (49–52). Our finding that lack of TYK2 in CD11c⁺ cells results in a reduction of CD27⁻CD11b⁺ splenic NK cells further underscores the importance of DCs for peripheral NK cell maturation. In contrast, we did not find evidence for a role of TYK2 in LysM⁺ cells in this process. It is currently unclear which signals drive IL-15/IL-15Rα expression under homeostatic conditions. We show that the absence of TYK2 results in reduced IL-15Rα levels on macrophages and DCs in naive mice. In response to immune challenge, type I IFNs induce IL-15 production and trans-presentation from DCs to NK cells (62). It is generally accepted that low levels of type I IFNs are produced under homeostatic conditions (63), prompting the hypothesis that IL-15 trans-presentation to NK cells may be driven by type I IFNs even in the absence of immune challenge. This idea is supported by the findings that naive Stat1^−/− and Ifnar1^−/− mice, but not Ifnar1^ΔNK mice, have less CD27⁻CD11b⁺ NK cells and that Stat1^−/− NK cells fully mature upon transfer into WT mice (40, 64). However, further studies will be required to test this possibility and to confirm that TYK2 drives NK cell maturation through its role in the type I IFN signaling cascade.

IFN-γ is a major NK cell effector molecule that is critical for tumor surveillance and immunity against infections. IFN-γ production by NK cells can be triggered by cytokines, such as IL-12 and IL-18, or actR stimulation (65). In support of the well-established role of TYK2 in the IL-12 signaling cascade (18), IFN-γ production in response to IL-12 was similarly impaired in NK cells from Tyk2^ΔNK and Tyk2^−/− mice. IFN-γ production by NK cells was also severely impaired in Tyk2^ΔNK mice upon infection with L. monocytogenes. Our previous study revealed defects in anti-NK1.1 and anti-Ly49D Ab–induced IFN-γ production in Tyk2^−/− NK cells, although TYK2 is not involved in the signaling cascade downstream of NK cell actR (60). Unlike NK cells from Tyk2^−/− mice, splenic NK cells from Tyk2^ΔNK mice showed similar IFN-γ production as those from Tyk2^fl/fl mice upon stimulation with anti-NK1.1 Ab, indicating that the impact of TYK2 is indirect. Recently the most mature CD27⁻CD11b⁺ NK cells have been identified as the main producers of IFN-γ in response to actR stimulation (52), suggesting that the impaired anti-NK1.1–induced IFN-γ production in the absence of TYK2 may reflect defects in terminal maturation. However, our finding that NK cells from Tyk2^ΔDC mice have a lower frequency of the CD27⁻CD11b⁺ cells but show unimpaired actR-induced IFN-γ production argues against this possibility. In human NK cells, terminal differentiation of peripheral NK cells is accompanied with demethylation of the Ifng promoter at three CpG sites in the transcribed region (+122, +128, +171), which enables high IFN-γ production in response to actR engagement (61). Our study suggests that such a mechanism is also important in mice, as the impaired maturation of Tyk2^−/− NK cells correlated with an increased methylation at CpGs +91 and +114 and impaired IFN-γ production upon actR stimulation. Moreover, treatment of NK cells from Tyk2^−/− mice with 5-Aza restored anti-NK1.1 Ab– but not IL-12–induced IFN-γ production. However, more detailed studies will be required to test whether these CpGs are indeed demethylated during murine NK cell maturation and whether this increased methylation in Tyk2^−/− NK cells reflects the more immature phenotype.

Furthermore, we show that TYK2 in NK cells is not required for target cell cytotoxicity and NK cell–dependent tumor surveillance. In vitro expanded NK cells from Tyk2^ΔNK mice showed unimpaired cytotoxic activity against RMA-S and RMA-Rae1 target cells, whereas NK cells from Tyk2^−/− mice had strongly reduced cytotoxic activity. Similar to what has been shown for NK cells from Stat1^−/− mice (66), NK cells from Tyk2^−/− mice had restored cytotoxicity against RMA-Rae1 cells when grown in the presence of IL-15 instead of IL-2. It remains to be investigated whether this reflects differences in cellular responses to IL-2 and IL-15 (67) or in signal strength. However, these data further underscore that IL-15 not only restores maturation but also functional competence of NK cells from Tyk2^−/− mice. Tyk2^ΔNK mice rejected cells that lack MHC I molecules (β2m^−/− splenocytes), and controlled s.c. transplanted RMA-Rae1 cells as efficiently as littermate controls. We show that TYK2 in LysM⁺ cells and DCs is required for in vivo killing of β2m^−/− splenocytes and that TYK2 in DCs contributes to RMA-Rae1 growth control. It remains an open question which external TYK2-dependent signal augments NK cell cytotoxicity and tumor surveillance. Defects in cytotoxicity may, at least in part, be causally linked to defects in terminal maturation or may result from impaired MHC I molecule–dependent NK cell education, although these two processes might not be completely independent (61, 68, 69). Further studies are required to define how TYK2 regulates the interplay between NK cells and other immune and nonimmune cells in both the differentiation of a functionally balanced NK cell pool and in antitumor immunity.

Although not in the focus of our study, we found that TYK2 signaling in NK cells is required for an efficient immune defense against L. monocytogenes. This is surprising, as several studies established that NK cells act deleterious rather than protective during systemic L. monocytogenes infections (70–72). It seems possible that despite their predominant immunosuppressive phenotype, NK cells have low level of antibacterial activity. This idea is supported by the findings that NK cells produce considerable amounts of IFN-γ upon i.p. L. monocytogenes infection (73–76) and modestly reduce bacterial burden upon transfer into Ifng^−/− mice (77) and upon local infections (62, 78). Increased sensitivity of Tyk2^ΔNK mice to L. monocytogenes infection correlated with an impaired IFN-γ production by NK cells, supporting the notion that the balance between IFN-γ– and IL-10–producing NK cells critically determines their role in antibacterial immunity (79, 80), thus arguing against a previously suggested negative impact of NK cell–derived IFN-γ (72). However, further investigations are required to characterize the impact of TYK2 on NK cell responses other than the production of IFN-γ to understand how NK cell–intrinsic TYK2 improves survival upon L. monocytogenes infection.

In summary, we provide evidence that NK cell–extrinsic TYK2 promotes NK cell maturation, cytotoxicity, and tumor surveillance, whereas NK cell–intrinsic TYK2 is needed for an efficient immune defense against L. monocytogenes infection. Moreover, we show that TYK2 not only facilitates IFN-γ production through its role in the IL-12 signaling cascade but also through NK cell–extrinsic mechanisms that specifically impact on IFN-γ production in response to NK cell actR stimulation and may involve Ifng promoter demethylation. Our findings that cytotoxic defects of NK cells from Tyk2^−/− mice can be rescued by IL-15 treatment suggest that unwanted effects of TYK2 inhibitors in tumor therapy may be overcome by boosting NK cell activity. However, more detailed experimentation will be required to better understand how TYK2 regulates NK cell–dependent immunity in vivo and to identify signaling cascades and cellular networks involved.

We thank Claus Vogl for help with the statistical analysis and Therese Lederer, Sabine Fajmann, Sabine Macho-Maschler, Daniela Prinz, and Dominika Gogová for excellent technical support. We are grateful to Michaela Prchal-Murphy, Dagmar Gotthardt, and Eva Maria Putz for valuable discussions and, additionally, Eva Maria Putz for critically reading of the manuscript.

The authors have no financial conflicts of interest.