Main

The successful eradication of an invading pathogen is dependent on the coordinated action of the innate and adaptive immune systems. Natural killer (NK) cells are able to quickly kill a wide range of pathogens such as viruses, bacteria and intracellular parasites1,2,3, as well as tumors4. The activity of NK cells is tightly controlled by both inhibitory and activating receptors5. Among these, NCR1 (also called Ly94 and Mar-1, or NKp46 in humans) is the only receptor reported so far to be expressed specifically on NK cells in all mammals tested, including humans6, monkeys7, mice8, rats9 and bovines10. However, the precise in vivo function of NCR1, or any other NK cell–activating receptor, in the killing of tumors and viruses has remained unclear.

NKp46 is a transmembrane type I glycoprotein containing two immunoglobulin domains and a positively charged arginine residue in the transmembrane domain, which associates with the T cell receptor-ζ signaling molecule6,11. The discovery of activating NK cell receptors slightly modified the 'missing-self' hypothesis, which originally stated that NK cells kill unless inhibited by expression of self major histocompatibility complex (MHC) class I proteins12. It is now known that a delicate balance between self MHC–mediated inhibition and stress-induced activation5,12 controls the cytotoxic activity of NK cells. Several in vitro studies have demonstrated that NKp46 is important in the recognition and destruction of various tumors13,14,15,16,17 and virus-infected cells18,19. Low concentrations of NKp46 have been noted in several clinical situations, including acute myeloid leukemia20, human immunodeficiency virus 1 infection21 and bare lymphocyte syndrome, type I (ref. 22). The presence of NK cells in tumors has been reported in renal cell carcinoma23, and NK cells have been identified in situ by staining with an NKp46-specific monoclonal antibody23. A cellular ligand for NKp46 or NCR1 has not yet been identified, but evidence suggests that heparan-sulfate proteoglycans are involved in NKp46 ligand recognition24. However, the in vivo function of NKp46 or NCR1 in the killing of tumor cells has not been investigated directly.

The importance of NK cells in virus defense in vivo has been demonstrated most precisely in an adolescent human who lacked NK cells and was therefore susceptible to virus infections25. In mice, depletion experiments have demonstrated increased mortality in the absence of NK cells after infection with influenza26. Finally, the critical involvement of NK cells has been established in various virus infections and in immune deficiencies27,28. The recognition of influenza virus–infected cells is mediated through direct interaction of NKp46, NKp44 and the hemagglutinin proteins of various strains of influenza in a sialic acid–dependent way18,19,29. The in vivo function of NKp46 in viral resistance has not yet been investigated.

Here we have generated mice in which a green fluorescent protein (GFP) cassette was 'knocked into' the Ncr1 locus, thereby rendering Ncr1 nonfunctional. There was substantial expression of the reporter gene only on NK cells. NCR1 is the only mouse receptor known to be expressed specifically on NK cells. Therefore, visualization of GFP allowed accurate identification of NK cells in vivo in unmanipulated mice. Influenza virus infection was lethal in Ncr1gfp/gfp 129/Sv and B57BL/6 mice, demonstrating that Ncr1 (NCR1 in humans) is essential for influenza eradication in vivo. We monitored the in vivo accumulation of the NK cells at the site of infection by tracking GFP expression. In addition, there was reduced clearance of the MHC class I–deficient RMAS tumor cells in vivo in Ncr1gfp/gfp mice on a 129/Sv background but not on a C57BL/6 background, demonstrating that NCR1 contributes to tumor eradication.

Results

Generation of Ncr1gfp/gfp mice

We initially maintained the Ncr1gfp/gfp mice on the 129/Sv strain because gene targeting was done in R1 embryonic stem cells (Supplementary Figure 1 online), which are of the 129/Sv background. In addition, the DAP12 adaptor protein, which transduces signals from Ly49-activating receptors, is impaired in 129/Sv mice30. As NCR1 signals via the T cell receptor-ζ chain, its activity would not be impaired in 129/Sv mice. Therefore, the 129/Sv background would allow a distinct assessment of the in vivo function of NCR1. Offspring were born at a mendelian ratio of 105:208:95 (wild-type/heterozygous/homozygous) and developed normally. We genotyped mice by Southern blot and PCR (Supplementary Figure 1). As a monoclonal antibody recognizing NCR1 was not available, we used RT-PCR to verify the lack of Ncr1 transcripts (Supplementary Figure 1). We later deleted the phosphoglycerate kinase–neomycin-resistance (PGK-neor) cassette by crossing Ncr1gfp/gfp mice with transgenic mice expressing Cre recombinase in the germline and verified that the PGK-neor cassette was excised (Supplementary Figure 1). Therefore, Ncr1gfp/gfp mice lacked Ncr1 transcripts and instead transcribed the gene encoding GFP under control of the endogenous Ncr1 regulatory elements.

GFP expression only in NK cells

NK cells are distinguished from other cell types by their expression of CD56 (in human) or DX5 (in mice)31 and their lack of CD3 expression. Although those markers are also expressed by some T cell subsets, NCR1 expression is specific and restricted to NK cells8. As we had introduced a GFP reporter into Ncr1, we first tested whether green cells were detectable in Ncr1gfp/gfp mice. A population of GFP+ cells was distinctly visible in Ncr1+/gfp and Ncr1gfp/gfp mice but not in Ncr1+/+ littermates (Fig. 1). To verify that all GFP+ cells were NK cells, we stained the cells for DX5 and showed that all GFP+ cells also expressed DX5 (Fig. 1a). In agreement with published studies31, not all DX5+ cells expressed GFP. Most of the DX5+GFP− cells also expressed CD3, indicating that these were T cells (data not shown). Similar analysis of expression of the NK cell–activating receptor NKG2D showed that almost all GFP+ cells also expressed NKG2D, whereas not all NKG2D+ cells expressed GFP (Fig. 1b). Most of the NKG2D+GFP− cells also expressed CD3 (data not shown). Analysis of cells from various organs demonstrated GFP expression in bone marrow, blood and spleen cells (Fig. 1). There was a small percentage of NK cells in the lymph nodes, whereas almost no green cells were detected in the thymus (Fig. 1).

Figure 1: Ncr1gfp expression in NK cells.
figure 1

(a,b) Flow cytometry of GFP expression in NK cells from various tissues (left margins). Numbers in quadrants indicate percentages of gated cells in each. BM, bone marrow; LN, lymph node. Representative of three independent experiments.

Full size image

We next analyzed whether GFP expression could be detected in B cells or T cells by staining with antibody to B220 (anti-B220) or anti-CD3, respectively. Weak B220 staining was apparent in a small fraction of the GFP+ population (Fig. 2a) and no CD3+ T cells expressed GFP (Fig. 2b). These results are in agreement with published reports demonstrating that some NK cells express B220 (ref. 32). They also indicate that the GFP reporter faithfully 'marked' only NK cells in vivo.

Figure 2: NK cell–specific Ncr1gfp expression.
figure 2

Flow cytometry demonstrates a lack of GFP expression in B cells (a) or T cells (b) from various tissues (left margins). Representative of three independent experiments.

Full size image

Enhanced tumor spread in Ncr1gfp/gfp mice

The importance of human NKp46 in the in vitro lysis of many tumors is well established4,5,18,19,29. To test whether NCR1 recognizes classical mouse NK cell–sensitive targets, we fused the extracellular portion of the NCR1 receptor with the Fc portion of human immunoglobulin G1 (IgG1) to generate the NCR1-Ig fusion protein as described before18. 'Classical' NK cell–sensitive targets such as the RMA, RMAS and YAC-1 cell lines were recognized by NCR1-Ig, whereas other cell lines, including P815 or MS1, were not (Fig. 3a). Next we investigated whether the killing of targets recognized by NCR1-Ig would be affected in Ncr1gfp/gfp mice. As little or no killing of RMA, RMAS or YAC-1 cells was induced by freshly isolated splenocytes (data not shown), we activated NK cells in vivo by injecting poly(I)·poly(C) and, 18 h later, tested killing activity in vitro. There was no difference in the killing activity of NK cells from Ncr1gfp/gfp, Ncr1+/gfp and Ncr1+/+ mice (Fig. 3b). In general, there was poor killing when RMA cells were used (Fig. 3b), probably because RMA cells express H-2b MHC class I molecules, which are recognized by inhibitory NK cell receptors expressed in 129/Sv mice. We obtained similar results when we used other tumor lines such as MS1 or P815, when we activated splenocytes in vitro with interleukin 2 and when we purified NK cells from splenocytes using magnetic beads (data not shown). Thus, it seems that in vitro killing of tumor cells is not affected by a lack of Ncr1, despite their recognition by NCR1-Ig.

Figure 3: Reduced clearance of tumor cells in vivo.
figure 3

(a) NCR1-Ig recognizes NK cell targets, as demonstrated by its binding to various mouse cell lines (above histograms). Gray shading, background secondary antibody staining; dark lines, specific staining. Representative of three independent experiments. (b) Natural cytotoxicity of poly(I)·poly(C)-activated splenocytes from mice of various genotypes (key) against various target cell lines (above graphs). E:T, effector/target ratio. Data represent percent specific lysis ± s.d. and are representative of three independent experiments. (c) Reduced in vivo killing of MHC class I–deficient RMAS cells in Ncr1gfp/gfp mice. Labeled target cells and control HeLa cells in the lung were quantified by flow cytometry 5 h after injection into the tail vein. Representative of three independent experiments.

Full size image

However, although NK cells had to be activated to achieve meaningful results in vitro, NK cells in vivo are part of the innate immune response, ready to kill without prior stimulation. We therefore examined the NK cell killing function in vivo using a published lung tumor clearance assay33, with minor modifications because of the presence of GFP in our mice. We used fluorescence-labeled (Vybrant DiD) human HeLa cells, which are not killed by mouse NK cells (data not shown), as a reference internal control. We labeled target cells with Vybrant Dil and injected these cells together with equal numbers of labeled HeLa cells into the tail veins of mice. We explanted lungs 5 h later and quantified labeled cells by flow cytometry. The in vivo killing of RMAS cells was impaired in Ncr1gfp/gfp mice compared with that in Ncr1+/gfp or Ncr1+/+ mice (Fig. 3c). The RMAS/HeLa ratio was approximately 0.4 in Ncr1+/+ and Ncr1+/gfp mice but was 0.65 in Ncr1gfp/gfp mice (P = 0.005). RMAS cells are derived from RMA cells and express little to no surface MHC class I molecules12,33. These two cell lines were used to establish the 'missing-self' hypothesis and to demonstrate that NK cells efficiently kill tumors that have lost MHC class I expression12. The in vivo RMA/HeLa ratio was approximately 0.8 in all strains (Fig. 3c), confirming that RMA cells are killed less efficiently than RMAS cells by NK cells. Thus, in agreement with the 'missing-self' hypothesis, the killing of tumor cells by NK cells was influenced by dominant negative signals emanating from MHC class I–specific inhibitory receptors. However, in the absence of those inhibitory receptors, killing depended on the presence of activating receptors. The next issue was whether the killing of other cells would also be impaired in vivo in Ncr1gfp/gfp mice. YAC-1 cells were killed with an efficiency similar to that of RMAS cells in all strains (Fig. 3c). These experiments suggest that NCR1 is directly involved in the killing of tumors such as RMAS in vivo and that there is no direct correlation between the in vitro and in vivo killing of tumor cells (probably because after activation, additional NK cell receptors are involved in killing, similar to the NKp44 situation in humans34). When MHC class I molecules are present, killing is inhibited and NCR1 is no longer involved in residual cytotoxicity, and the killing of YAC-1 cells is probably mediated by other activating receptors such as NKG2D.

Lethal influenza infection in Ncr1gfp/gfp mice

Studies have indicated that viral hemagglutinin proteins interact with human NKp44 and NKp46 but not with NKp30 and that this interaction is dependent on receptor sialylation18,29. To determine whether those interactions are relevant in vivo, we first tested whether NCR1-Ig recognizes influenza-infected cells. Infection of RMAS by influenza A/PR8 (H1N1) virus enhanced NCR1-Ig binding (Fig. 4a), suggesting that similar to human NKp46, mouse NCR1 specifically recognizes influenza-infected cells. To test whether those interactions are important in vivo, we inoculated mice intranasally with influenza A/PR8. This assay directly addresses the function of NCR1 in a very complex in vivo physiological situation. In this assay, mice are subjected to life-threatening virus infection, and the NK cells in these mice express an array of activating and inhibitory receptors (except for NCR1 in Ncr1gfp/gfp mice). Most Ncr1+/+ and Ncr1+/gfp mice (66.6% and 58%, respectively) survived infection (Fig. 4b). In contrast, all of their Ncr1gfp/gfp littermates died (Fig. 4b). All mice were infected, as demonstrated by body weight reduction and hemagglutination-inhibiting antibodies to influenza (data not shown).

Figure 4: Lethal influenza infection in 129/Sv Ncr1gfp/gfp mice.
figure 4

(a) Enhanced binding of NCR1-Ig to infected cells. RMAS cells were infected with influenza PR8 virus and were stained with NCR1-Ig. Gray shading, secondary antibody staining; dark lines, specific staining. Representative of five independent experiments. (b) Decreased survival of Ncr1gfp/gfp mice after influenza infection. Viability was monitored daily after infection. Representative of three independent experiments. (c) Normal control of vaccinia infection in Ncr1gfp/gfp mice. β-galactosidase activity was quantified as absorbance at 570 nm (A570) in the ovaries of mice infected with vaccinia virus encoding β-galactosidase. Representative of two independent experiments.

Full size image

To test whether the virus susceptibility of Ncr1gfp/gfp mice is specific for influenza, we infected mice with a vaccinia virus expressing β-galactosidase. Vaccinia-infected cells showed no increase in the binding of NCR1-Ig (data not shown). We assessed β-galactosidase expression in the ovaries 3 d after infection. There was similar infection of Ncr1gfp/gfp, Ncr1+/gfp and Ncr1+/+ littermates (Fig. 4c), suggesting that NCR1 is required specifically for controlling influenza infection.

Because we detected GFP+ NK cells in the lung even before influenza infection (Fig. 5a), we monitored the in vivo accumulation of NK cells in influenza-infected mice. The number of NK cells in the lungs increased fourfold during illness (Fig. 5b). We stopped the experiment on day 5 because mice were already severely ill. NK cell accumulation in the lung was almost identical in influenza-susceptible Ncr1gfp/gfp and influenza-resistant Ncr1+/gfp littermates, indicating that NK cell trafficking in vivo is unaffected by the absence of NCR1 and that the death of Ncr1gfp/gfp mice was not due to defects in trafficking of NK cells to the lung.

Figure 5: NK cells accumulate at the site of influenza infection.
figure 5

(a) Flow cytometry of NK cells in the lungs of influenza-infected mice; boxed areas indicate DX5+GFP+ cells. Representative of ten independent experiments. (b–e) Quantification of NK cells in the lung (b), blood (c), spleen (d) and bone marrow (e) of mice of various genotypes (key) at various times after intranasal influenza infection (horizontal axes). All mice were infected on the same day. Representative of three independent experiments.

Full size image

We tracked NK cells in various organs during influenza infection and noted a salient reduction in the percentage of NK cells in the blood 2 d after infection (Fig. 5c). That reduction correlated with the increased percentage of NK cells in the lung (Fig. 5b), suggesting that early in the course of influenza infection, NK cells are recruited to the lung directly from the blood. Later, the percentage of NK cells in the blood increased back to the preinfection percentage, whereas the percentage of NK cells in the lung continued to increase (Fig. 5b,c). There were modest increases in NK cell percentages in the bone marrow and spleen throughout infection (Fig. 5d,e). NK cells are generated in the bone marrow1. It therefore seems possible that NK cell proliferation is increased 3 d after infection and NK cells emigrate from the bone marrow. These results suggest that at early time points after infection, NK cells rapidly migrate to the site of infection directly from the blood. Later, at the peak of infection, when NK cells are most needed to fight the invading virus, NK cells are found in large numbers in the infected organ. However, that accumulation does not leave other organs unprotected.

Strain specificity of the function of NCR1

PGK-neor cassettes can affect the function of other proximal genes. Therefore it was possible that the effects reported here could have been attributed to the presence of the PGK-neor cassette. In addition, as stated above, the Ly49-activating receptors that signal via DAP12 are not functional in 129/Sv mice30. Therefore, the functional consequences of Ncr1 deficiency noted here might differ in mice expressing a 'normal' repertoire of activating NK cell receptors. To address the former issue, we excised the PGK-neor gene by breeding Ncr1gfp/gfp mice with 129/Sv (Prm-Cre) or C57BL/6 (Zp3-Cre) mice expressing Cre in the germline. To address the latter issue, we backcrossed Ncr1gfp/gfp mice to the C57BL/6 background (which expresses functional DAP12) for ten generations (Fig. 6).

Figure 6: Strain-specific NCR1 functions.
figure 6

(a,b) GFP expression in NK cells of Ncr1gfp/gfp and Ncr1+/gfp mice before and after excision of the PGK-neor cassette: 129neor, 129/Sv before excision; 129, 129/Sv after excision; C57, C57BL/6 after excision and ten generations of backcrossing. Representative of three independent experiments. (c) Ncr1gfp/gfp 129/Sv mice remain susceptible to influenza infection after excision of the PGK-neor cassette. Viability was monitored daily after infection. Representative of two independent experiments. (d) Reduced RMAS clearance is maintained in Ncr1gfp/gfp 129/Sv mice after PGK-neor excision. Labeled RMAS target cells and control HeLa cells in the lung were quantified by flow cytometry 5 h after injection into the tail vein. Representative of two independent experiments. (e) Influenza is lethal in Ncr1gfp/gfp C57BL/6 mice. Viability was monitored daily after infection. Representative of three independent experiments. (f) Normal RMAS clearance in Ncr1gfp/gfp C57BL/6 mice, assessed as described in d. Representative of two independent experiments.

Full size image

In agreement with our results reported above (Fig. 1), the only GFP+ cells found in the 129/Sv mice after PGK-neor excision or in the C57BL/6 mice were NK cells (Supplementary Figs. 2, 3, 4, 5 online). The intensity of GFP expression was greater after excision of the PGK-neor cassette (Fig. 6a,b; median fluorescence intensity of 110 or 180 for Ncr1gfp/gfp mice of the 129/Sv background before or after PGK-neor excision, respectively). There was even higher GFP intensity in the C57BL/6 mice after PGK-neor excision and ten generations of backcrossing, and all GFP+ cells expressed DX5 (Fig. 6a,b; median fluorescence intensity of 420 for Ncr1gfp/gfp mice of this background).

Regardless of whether the PGK-neor cassette was deleted or whether Ncr1 deletion was assessed on a 129/Sv or a C57BL/6 background, Ncr1gfp/gfp mice died from influenza infection, whereas approximately 60% of Ncr1+/+ and Ncr1+/gfp littermates survived the infection (Fig. 6c,e). Thus, even in the presence of a full spectrum of NK cell–activating receptors, NCR1 is critical for the survival of influenza-infected mice. Also regardless of whether the PGK-neor cassette was deleted, in vivo clearance of RMAS tumor cells was impaired in Ncr1gfp/gfp 129/Sv mice compared with that in Ncr1+/+ and Ncr1+/gfp littermates (P < 0.01; Fig. 6d). In contrast, clearance of RMAS cells was similar in Ncr1gfp/gfp, Ncr1+/gfp and Ncr1+/+ C57BL/6 littermates (Fig. 6f). Thus, the killing of tumor cells by NK cells in vivo is efficient in some strains such as the C57BL/6, even in the absence of NCR1, as other NK cell–activating receptors probably compensate for the lack of NCR1 in those strains.

Discussion

The ability to directly and quickly destroy hazardous cells without the need for prior stimulation positions NK cells at the front line of innate immune defense. Two mechanisms regulate NK cell cytotoxicity. Inhibition is achieved mostly by inhibitory NK cell receptors that interact with MHC class I proteins12,35, and activation is achieved by activating NK cell receptors that recognize stress-induced proteins36, viral proteins18 or unknown tumor ligands29. Here we have demonstrated that the NCR1 activating receptor was directly involved in the in vivo killing of RMAS tumor cells that lack MHC class I molecules. However, in the presence of MHC class I molecules (on RMA cells) or perhaps other activating ligands and/or receptors (on YAC-1 cells and in C57BL/6 NK cells), or when NK cells were activated, the specific function of NCR1 in tumor clearance was diminished, suggesting that NCR1 participates in the delicate balance of activating and inhibitor signals regulating NK cell cytotoxicity. Indeed, it has been shown that the killing of YAC-1 but not RMAS cells is mediated via the NKG2D receptor37.

Studies have shown in vitro that various viral hemagglutinin proteins interact with the sialic acid residues on human NKp44 and NKp46 (ref. 19). We have demonstrated here that despite the expression of MHC class I proteins and other activating receptors on NK cells and despite the fact that NK cells accumulate in vivo at the site of infection, the activity of NCR1 against influenza was crucial and rescued mice of two different genetic backgrounds from influenza-induced death. Tumor spread is limited to the lifespan of each person, but influenza infection is a threat to an entire population. Therefore, NKp46 (NCR1 in mice) might have evolved first to fight influenza infection and then to fight tumor spread in specific situations (such as in the absence of MHC class I–mediated inhibition).

We first did experiments before excision of the PGK-neor cassette. However, there was no difference functionally in the Ncr1gfp/gfp 129/Sv mice before or after PGK-neor excision; all died from influenza and RMAS clearance was impaired in both. However, there was more GFP expression after PGK-neor excision. Thus, as commonly appreciated, we suggest that careful attention should always be given to the presence of the PGK-neor gene.

Here we have directly investigated the function of a single gene in vivo and have devised a way to track cells expressing this gene. As the detection of NK cells has always been complicated by the expression of 'classical' NK cell markers on other cell types, Ncr1gfp/gfp mice provide a unique ability to isolate and track NK cells in vivo without experimental manipulation or complicated staining protocols. The GFP reporter 'marks' only NK cells in both 129/Sv and C57BL/6 mice.

The differentiation of NK cells might have been impaired in Ncr1gfp/gfp mice, as activating signals are essential for the maturation of the closely related T cell linage38 and NKp46 is known to be expressed in humans from an early stage of cell maturation39. The identification of normal NK cell populations in both 129/Sv and C57BL/6 Ncr1gfp/gfp mice indicated that multiple activating pathways are redundant or that NCR1signaling is not essential during NK cell development. The former option is supported by the fact that NK cells develop normally even in the absence of the tyrosine kinases Zap70 and Syk40. The latter option is supported by the observation that in vitro maturation of human NK cells from progenitors has been achieved without any reported stimulation of activating receptors39. Results obtained with the Ncr1gfp/gfp mice indicate that despite its critical involvement in cytotoxicity, Ncr1 is not essential for the generation of a normal NK cell population in vivo.

In summary, Ncr1gfp/gfp mice demonstrate the critical involvement of NCR1 in the control of influenza and some tumors. These mice will enable direct investigation of the function of NCR1 in other virus infections and tumor models, as well as in bacterial and parasite infections and in autoimmune disorders. Furthermore, the NK cell–specific expression of GFP in these mice will allow study of NK cell trafficking in vivo and imaging of NK cells in various pathological conditions.

Methods

Mice.

The Ncr1 targeting vector was generated by PCR amplification from ES-R1 embryonic stem cell line genomic DNA with the following primers: left arm forward, 5′-GGAGAGCTCTGGTCAAACTCGAGCAAC-3′, and left arm reverse, 5′-GAAGAATTCGGGTCGGTAGGTGGAAGG-3′ (including restriction sites for XhoI and EcoRI, respectively); and right arm forward, 5′-AAGCGGCCGACAGATTAACAAATTGGG-3′, and right arm reverse, 5′-CTGCCGCGGTAGCTGGAAACAGCGCAG-3′ (including restriction sites for EagI and SacII, respectively). IRES-GFP and PGK-neor cassettes were subcloned in pBluescript-KS II using EcoRI plus SalI or SalI plus XbaI, respectively. The targeting construct was electroporated into R1 embryonic stem cells41 and targeted clones were selected with G418 and analyzed by Southern hybridization. The generation of founder chimeras was done by aggregation at the Transgenic & Knockout Facility of the Weizmann Institute (Rehovot, Israel). The PGK-neor cassette was deleted by breeding of targeted mice with 129/Sv (Prm-Cre) or C57BL/6 (Zp3-Cre) mice expressing Cre in the germline (stocks 003328 and 003651, respectively; The Jackson Laboratory). Targeted mice were backcrossed for ten generations onto the C57BL/6 background. All experiments were done in the specific pathogen–free unit of Hadassah Medical School (Ein-Kerem, Jerusalem) according to guidelines of the ethical committee.

Flow cytometry.

Peripheral blood was obtained from the tail, single-cell suspensions of organs were obtained with a cell strainer and bone marrow was obtained from the tibia and femur. Antibodies specific for NK cells (DX5; Caltag), NKG2D (CX5; eBioscience) and B220 and CD3 (RA3-6B2 and 145-2C11, respectively; BD Bioscience) were conjugated to phycoerythrin. A monoclonal antibody specific for CD16 and CD32 (2G.4.2; Pharmingen) was used for blocking of Fc receptors before staining. The NCR1-Ig fusion protein was generated as described18,29, and staining of cell lines was visualized with secondary phycoerythrin-conjugated goat anti-human (Jackson ImmunoResearch).

Cytotoxicity assays.

For the in vitro assay, 200 μg poly(I)·poly(C) (Sigma) was injected and splenocytes were removed 18 h later. Target cells were labeled with [35S]methionine and were incubated for 5 h at 37 °C with effector splenocytes at various effector/target ratios. Specific killing was calculated as described18,29. For the in vivo assay, we used a published fluorescence labeling method33 with a modification to avoid possible influence of the intrinsic GFP present in Ncr1gfp/gfp mice. Cells were labeled with Vybrant cell labeling solutions (Molecular Probes). HeLa cells were labeled with Vybrant DiD (Molecular Probes), and RMAS, RMA or YAC-1 cells were labeled with Vybrant Dil (Molecular Probes). Cells were mixed at a concentration of 20 × 106 cells of each population/ml in PBS, and 200 μl was injected into the tail vein (average of three mice in each experiment). Lungs were collected 5 h later, single-cell suspensions were obtained with cell strainers and fluorescence was analyzed by flow cytometry. The ratio of the tested targets to the internal control HeLa cells was calculated.

Viral infection.

Influenza A PR8 virus was grown in hen eggs. Mice (6–8 weeks of age) were anesthetized and were inoculated intranasally with 50 μl of diluted virus (3 × 103 plaque-forming units; a dilution of 1/31,622 from 4096 HAU). Each experiment included at least five mice of each genotype. Viability was determined daily. Antibodies to influenza were assessed in the serum of mice that survived infection. For analysis of NK cell trafficking during the disease, mice were infected with influenza as described above and were killed at various times after infection, and lungs were analyzed as described above. Vaccinia infections were as described42; mice were infected with 2 × 107 plaque-forming units and, 3 d later, β-galactosidase activity was assayed with chlorophenol red-β-D-galactopyranoside substrate, measured at a wavelength of 570 nm.

Note: Supplementary information is available on the Nature Immunology website.