Tuning of Natural Killer Cell Reactivity by NKp46 and Helios Calibrates T Cell Responses

See allHide authors and affiliations

Science  20 Jan 2012:
Vol. 335, Issue 6066, pp. 344-348
DOI: 10.1126/science.1215621


Natural killer (NK) cells are lymphocytes involved in antimicrobial and antitumoral immune responses. Using N-ethyl-N-nitrosourea mutagenesis in mice, we identified a mutant with increased resistance to viral infections because of the presence of hyperresponsive NK cells. Whole-genome sequencing and functional analysis revealed a loss-of-function mutation in the Ncr1 gene encoding the activating receptor NKp46. The down-regulation of NK cell activity by NKp46 was associated with the silencing of the Helios transcription factor in NK cells. NKp46 was critical for the subsequent development of antiviral and antibacterial T cell responses, which suggests that the regulation of NK cell function by NKp46 allows for the optimal development of adaptive immune responses. NKp46 blockade enhanced NK cell reactivity in vivo, which could enable the design of immunostimulation strategies in humans.

Natural killer (NK) cells are cytolytic and cytokine-producing lymphocytes that can directly kill transformed and microbe-infected cells. They also participate in the shaping of adaptive immune responses (13). While screening for altered NK cell phenotypes in C57BL/6J mice carrying N-ethyl-N-nitrosourea (ENU)–induced mutations, we identified a mouse pedigree with NK cells that were hyperreactive in responses to the prototypic NK cell tumor target cell line, YAC-1, in vitro (Fig. 1A). These mice were bred for the establishment of a homozygous stock. The phenotype was named Noé and appeared to result from a single autosomal recessive mutation. Furthermore, resting Noé NK cells were more responsive when stimulated by monoclonal antibody (mAb) cross-linking of the NK1.1-activating receptor (fig. S1A). This higher reactivity of Noé NK cells was not associated with an increase in NK1.1 cell surface expression (fig. S1B) and was also observed after stimulation with a combination of interleukin 12 (IL-12) and IL-18 cytokines (fig. S1C). In addition, NK1.1-stimulated Noé NK cells were less sensitive to the phosphatidylinositol 3-kinase inhibitor wortmannin, which prevents signal transduction (fig. S1D). Thus, Noé NK cells displayed a broad increase in reactivity to various stimuli in vitro, compatible with a profound change in the threshold of NK cell responses. Mixed–bone marrow chimera experiments revealed that the hyperresponsiveness of Noé NK cells resulted from an intrinsic cellular modification (fig. S2).

Fig. 1

Noé mice are more resistant to MCMV infection because of the hyperresponsiveness of their NK cells. (A) Frequencies of IFN-γ–producing and CD107a+ cells among IL-2–activated WT and Noé NK cells after coculture with YAC-1 tumor targets. n = 9 to 10, P < 0.0001 and P = 0.0003, Mann-Whitney test. (B and C) Kaplan-Meier representation of WT (closed symbols) and Noé (open symbols) mouse survival after infection with 5300 plaque-forming units (PFU) of MCMV per gram of body weight (gBW). n = 15 to 20, P < 0.0001, log-rank Mantel-Cox test. (C) NK cells from WT and Noé mice were depleted by treatment with NK1.1-specific mAb on the day before MCMV infection. (D) Viral loads (means ± SD) in the spleen and liver of WT and Noé mice 4 days after MCMV infection. Dotted lines indicate the detection limit. n = 10, **P = 0.0079 and *P = 0.0112, Mann-Whitney test. (E) Frequencies of IFN-γ–producing and CD107a+ NK cells in the spleen of WT or Noé mice 1.5 days after MCMV infection. n = 5, **P = 0.0079 and *P = 0.0159, Mann-Whitney test. All data result from a pool of two to three independent experiments.

We then investigated whether the Noé NK cells were also hyperresponsive in vivo. In the C57BL/6J background, NK cells play a key role in the early control of mouse cytomegalovirus (MCMV) infection by recognizing the viral m157 protein via the activating receptor Ly49H (4, 5). Both Noé and wild-type (WT) animals survived infection with low doses of MCMV (fig. S3A), but only Noé mice survived infection with intermediate doses (Fig. 1B). High doses were lethal for both strains, but Noé mice survived significantly longer than WT animals (fig. S3B). This greater resistance of Noé mice was dependent on NK cells, as the injection of NK cell–depleting NK1.1 monoclonal antibody (mAb) resulted in the death of Noé mice after infection (Fig. 1C). Consistent with their resistance to MCMV infection, Noé mice had MCMV loads in the spleen and liver that were four-fifths and five-sixths, respectively, of those in WT animals 4 days after infection (Fig. 1D). We detected no difference in the frequencies of Ly49H+ NK cell between Noé and WT mice (fig. S4), but the frequencies of interferon- γ (IFN-γ)–producing and degranulating CD107a+ NK cells ex vivo were 53% ± 11% and 33% ± 8% (means ± SD) higher, respectively, in Noé mice than in WT mice 1.5 days after infection (Fig. 1E). These results suggested that Noé mice were more resistant to MCMV infection because of a greater responsiveness of NK cells in vivo.

We then sought to identify the recessive mutation responsible for the broad hyperresponsiveness of Noé NK cells by resequencing the whole genome of a Noé mouse and a control WT mouse (table S1). We focused on a mutation of the Ncr1 gene that was present in all Noé mice. Ncr1 encodes the NK cell–activating receptor NKp46, which is conserved in mammals and expressed by all mature NK cells (6). This mutation resulted in the replacement of the tryptophan 32 residue with an arginine residue (W32R) (fig. S5). Ncr1 transcript levels were found to be similar in Noé and WT NK cells (fig. S6A), but we were unable to detect NKp46 at the surface of NK cells from Noé mice (Fig. 2A and fig. S6B). Flow cytometric analysis of human embryonic kidney (HEK) 293T cells transfected with the WT and Noé NKp46 sequences tagged at the N terminus (fig. S7A) revealed that, even though the NKp46 protein bearing the W32R substitution was detected intracellularly, it was not expressed at the cell surface (fig. S7B). The W32R mutation thus abolishes the cell surface expression of NKp46 on the NK cells of Noé mice.

Fig. 2

A point mutation in Ncr1 gene is responsible for the hyperresponsiveness of Noé NK cells. (A) Flow cytometric analysis of WT and Noé splenocytes stained with NK1.1 and NKp46 mAbs. Results are representative of the Noé phenotype. (B) Frequencies of IFN-γ–producing and CD107a+ cells among IL-2–activated WT, Ncr1Noé/Noé, and Ncr1Noé/Noé huNKp46 Tg NK cells after coculture with YAC-1 tumor targets. n = 6 to 8, *P < 0.05, **P < 0.01, Kruskal-Wallis test followed by Dunns’ test. (C) Kaplan-Meier representation of WT (closed circles), Ncr1Noé/Noé (open circles), and Ncr1Noé/Noé huNKp46 Tg (open triangles) mice survival after MCMV infection with 5300 PFU/gBW. n = 8 to 10, P < 0.0001, log-rank Mantel-Cox test. In (A to C), data result from a pool of two to three independent experiments. (D) Helios transcripts were quantified by reverse transcription polymerase chain reaction in sorted CD11b+ bone marrow NK cells from WT, Ncr1Noé/Noé, and Ncr1Noé/Noé huNKp46 Tg mice. Results are normalized with respect to Gapdh (glyceraldehyde phosphate dehydrogenase) and expressed as means ± SD in arbitrary units. The data correspond to two independent experiments with a total of six animals per group. (E) IL-2–activated WT and Ncr1Noé/Noé NK cells were transduced with Helios or EGFP control short hairpin RNA. Data show frequencies of IFN-γ–producing and CD107a+ NK cells (means ± SD) after coculture with YAC-1 tumor targets. The data correspond to two independent experiments with a total of six transductions per group.

We then used genetic complementation to assess formally the direct involvement of the NKp46W32R mutation in the hyperresponsiveness of Noé NK cells. Mice homozygous for the W32R mutation, henceforth referred as to Ncr1Noé/Noé mice, were crossed with transgenic mice in which the cell surface expression of human NKp46 protein is restricted on NK cells (huNKp46 Tg) (6). We analyzed F2 animals lacking mouse NKp46 expression but expressing the human NKp46 protein (Ncr1Noé/Noé huNKp46 Tg) and compared them with their nontransgenic Ncr1Noé/Noé or WT littermates (fig. S8). After coculture with YAC-1 tumor targets, the reactivity of IL-2–activated Ncr1Noé/Noé NK cells complemented with the human NKp46 protein was similar to that of WT NK cells (Fig. 2B). Human NKp46 also restored the sensitivity of Ncr1Noé/Noé mice to MCMV infection (Fig. 2C). Thus, the NKp46W32R mutation was responsible for the NK cell phenotype in Noé mice.

NKp46 is associated with immunoreceptor tyrosine-based activation motif (ITAM)–bearing polypeptides—such as CD3ζ and FcRγ, which transduce potent activating signals when triggered (7)—and has been seen as a stimulatory receptor involved in effector functions (811). To confirm the role of NKp46 in the down-regulation of NK cell responsiveness, we studied Ncr1iCre/iCre knock-in mice (12), in which the sequence of the NKp46 protein is WT but which also display a severe impairment of NKp46 expression on the plasma membrane of NK cells (fig. S9, A and B) because of a decrease in Ncr1 transcript levels (fig. S9C). Like Noé NK cells, Ncr1iCre/iCre NK cells were hyperresponsive to YAC-1 target cells and NK1.1 stimulation (fig. S9, D and E), and Ncr1iCre/iCre mice were more resistant to MCMV infection than WT animals (fig. S9F). Thus, mutation of Ncr1 resulted in enhanced NK cell reactivity and greater resistance to MCMV infection in two independent mouse genetic models.

We investigated whether this phenotype was restricted to MCMV infection or could be generalized to other models of infection. We studied H1N1PR8 influenza, as Ncr1 knockout mice have been reported to be highly susceptible to this infection (9). By contrast, Ncr1Noé/Noé and Ncr1iCre/iCre mice were more resistant to H1N1PR8 infection than the WT (fig. S10, A and B). This greater resistance was associated with a higher frequency of IFN-γ–producing NK cells in the lungs of Ncr1Noé/Noé mice 2 and 3 days after infection (fig. S10C). One possible explanation for this discrepancy is that the DNA deletions introduced in Ncr1 knockout mice removed three potential intronic microRNAs (miRNAs) (miR-1195, miR-1935, and miR-3470a). It thus remains to be tested whether these deletions are responsible for the phenotypes described in this model, as described for the deletion of miR-211 in the intronic region of the melastatin gene, which is entirely responsible for the tumor phenotype of melastatin knockout mice (13).

NKp46 is expressed early in NK cell differentiation—after the induction of NK1.1 expression but before CD11b up-regulation—and is then uniformly expressed by all CD11b+ NK cells (6, 12). We dissected the mechanisms by which NKp46 affected NK cell development by identifying genes displaying differential expression between NK1.1+CD11b and NK1.1+CD11b+ NK cells in WT mice, in a pan-genomic transcriptomic analysis (14). We identified the Ikzf2 gene (also known as Helios), which encodes a member of the Ikaros transcription factor family, as being down-regulated in NK1.1+CD11b+ cells as compared with NK1.1+CD11b NK cells (fig. S11, A and B). It is noteworthy that the ectopic expression of Helios in B cells modified their threshold of reactivity, which led to hyperresponsiveness to antigen (15). Helios transcripts were twice as abundant in the NK1.1+CD11b+ NK cells of Ncr1Noé/Noé mice as in those of WT mice (Fig. 2D). Genetic complementation of the Ncr1Noé mutation in Ncr1Noé/Noé huNKp46 Tg mice restored Helios transcript levels in NK1.1+CD11b+ NK cells to those in WT mice (Fig. 2D). Furthermore, Helios down-regulation was also impaired in NK1.1+CD11b+ NK cells from Ncr1iCre/iCre mice (fig. S11C). Of note, the silencing of Helios in NK cells from Ncr1Noé/Noé mice restored their reactivity to that of WT NK cells (Fig. 2E and fig. S12). These data are consistent with a model in which Helios down-regulation is involved in the regulation of NK cell reactivity via NKp46.

NK cell cytotoxicity and cytokine production help to shape adaptive T cell responses (2, 16, 17). During MCMV infection, NK cells can limit T cell responses by decreasing the amount of viral antigen available and by killing activated T cells (18, 19). We therefore investigated the effect of the hyperresponsiveness of the NK cells of Ncr1Noé/Noé mice on T cell responses. In Ncr1Noé/Noé mice, the frequency of splenic CD8+ T cells specific for the immunodominant MCMV peptide m45 presented by H-2Db was lower than that in the WT, by a factor of two 7 days after infection and by a factor of three 10 days after infection (Fig. 3A). At the peak of the response, Ncr1Noé/Noé mice had frequencies of IFN-γ–producing CD8+ and CD4+ T cells lower than those of the WT by a factor of 2 to 3 (Fig. 3, B and C). In Ncr1Noé/Noé huNKp46 Tg mice, the frequency of H-2Db/m45+CD8+ T cells (Fig. 3D) and of IFN-γ–producing cells among CD8+ (Fig. 3E) and CD4+ T cells (Fig. 3F) was restored to that in WT mice. Thus, the hyperresponsiveness of NK cells improved virus control but limited subsequent T cell responses.

Fig. 3

Hyperresponsive NK cells limit T cell immunity. (A to E) Mice were infected with a dose of 1600 PFU of MCMV/gBW. (A) Frequencies of MCMV-specific CD8+ T cells among total CD8+ T cells (means ± SD) were measured by H-2Db/m45 pentamer staining in spleens from WT (closed symbols) and Ncr1Noé/Noé (open symbols) mice at the indicated time points after infection. n = 5 to 7, *P < 0.05 and ***P < 0.001, two-way analysis of variance (ANOVA) with Bonferroni correction. (B and C) Ex vivo restimulation of (B) spleen H-2Db/m45-specific CD8+ T cells or (C) liver CD4+ T cells. Frequencies of IFN-γ–producing cells among total CD8+ or CD4+ T cells are shown. n = 5 to 7, **P < 0.01 and ***P < 0.001, Two-way ANOVA with Bonferroni correction. (D) Frequencies of H-2Db/m45+CD8+ T cells among total CD8+ T cells in the spleen of WT, Ncr1Noé/Noé, and Ncr1Noé/Noé huNKp46 Tg littermates 7 days after infection. (E and F) Ex vivo restimulation of (E) spleen H-2Db/m45-specific CD8+ T cells or (F) liver CD4+ T cells. The frequencies of IFN-γ–producing cells are shown for WT, Ncr1Noé/Noé, and Ncr1Noé/Noé huNKp46 Tg mice 7 days after infection. n = 5, **P < 0.01 and ***P < 0.001, Kruskal-Wallis test followed by Dunn’s test. (G) Mice were injected with 104 Lm-OVA bacteria, and the frequencies of IFN-γ–producing cells were determined in the spleen after 24 hours. n = 5, *P = 0.0119, Mann-Whitney test. (H and I) Mice were injected with 104 Lm-OVA bacteria and challenged 1 month later with 5 × 105 bacteria. (H) Five days after secondary challenge, the frequencies of H-2Kb/OVA+CD8+ T cells and IFN-γ–producing CD8+ T cells were determined in the spleen of WT, Ncr1Noé/Noé and Ncr1Noé/Noé huNKp46 Tg littermates. Data show the frequencies of IFN-γ–producing cells among H-2Kb/OVA+ CD8+ T cells. n = 4 to 10, *P < 0.05 and **P < 0.01, Kruskal-Wallis test followed by Dunn’s test. (I) Bacterial loads (means ± SD) in the spleen of WT, Ncr1Noé/Noé, and Ncr1Noé/Noé huNKp46 Tg littermates 2 days after secondary challenge. n = 5 to 10, **P < 0.01 and ***P < 0.001, Kruskal-Wallis test followed by Dunn’s test. All data result from a pool of two to four independent experiments.

Hyperresponsive NK cells might therefore be advantageous initially but become disadvantageous during a secondary challenge if the capacity to mount a memory immune response was impaired. We tested this hypothesis by analyzing the CD8+ T cell protective immunity generated in response to intracellular bacteria Listeria monocytogenes (Lm)–expressing ovalbumin (OVA), which is completely cleared after primary infection, unlike MCMV. During Lm infection, NK cells are activated by cytokines, such as IL-12 (20). Consistent with their broad hyperresponsiveness, the frequencies of IFN-γ–producing NK cells were 36% ± 3.2% (mean ± SD) higher in Ncr1Noé/Noé mice than in the WT 24 hours after infection with Lm-OVA (Fig. 3G). After primary Lm-OVA infection and rechallenge 30 days later, the percentages of memory Lm-OVA–specific CD8+ T cells capable of producing IFN-γ were 35% ± 4.4% and 36% ± 4% (means ± SD) lower in Ncr1Noé/Noé mice than in WT mice and Ncr1Noé/Noé huNKp46 Tg mice, respectively (Fig. 3H). This alteration in the quality of Lm-OVA–specific CD8+ memory T cells in Ncr1Noé/Noé mice was associated with a bacterial load in the spleen 12 times that in WT mice and Ncr1Noé/Noé huNKp46 Tg littermates (Fig. 3I). Thus, the hyperresponsiveness of NK cells during the T cell–priming phase affected the generation of fully protective memory T cells. Evolution in mammals may have resulted in the counterselection of hyperreactive innate immunity mechanisms despite their efficiency during the first encounter with a pathogen, to favor the emergence of adaptive immune responses, which efficiently control high doses of pathogens upon reexposure.

The dissection of NK cell education and tolerance has, to date, focused on the role of inhibitory receptors recognizing self-MHC (major histocompatibility complex) class I molecules. The engagement of these inhibitory receptors increases NK cell responsiveness (2127). Our analysis of Ncr1Noé/Noé mice implicated the activating receptor NKp46 as a checkpoint in NK cell tuning and revealed another aspect of NK cell education in which the engagement of NKp46 down-regulates NK cell responsiveness. These results are consistent with observations made for other activating receptors (2831).

Apart from the interaction between NKp46 and viral hemagglutinins reported in a previous study (32), no cellular ligand for NKp46 has yet been identified. We found that NKp46 mAbs blocked the activation of NK cells by autologous CD11c+-enriched spleen cells (fig. S13). In order to block NKp46 during NK cell development, we selectively depleted NK cells by injecting diphtheria toxin into NK-DTR/EGFP (NDE) transgenic mice, which express the diphtheria toxin receptor and enhanced green fluorescent protein (6), and then treated mice during NK cell repopulation with either a blocking NKp46 mAb or with an isotype control mAb for 13 days (Fig. 4A and fig. S14). After coculture with YAC-1 tumor targets, NK cells isolated from NKp46 mAb-treated animals had 60% ± 8% and 38% ± 11% (mean ± SD) more IFN-γ–producing and CD107a+ NK cells, respectively, than NK cells from control mice (Fig. 4A). Thus, the treatment of NKp46-sufficient mice with blocking NKp46 mAbs mimicked the phenotype observed in Ncr1Noé/Noé and Ncr1iCre/iCre mice, which supports a model in which NK cell tuning via NKp46 occurs after interaction with endogenous ligands. The nature and regulation of the endogenous NKp46 ligands remain to be determined. Our findings suggested that NKp46 blockade could be used as an immunotherapeutic strategy to enhance NK cell effector functions. We thus evaluated whether we could modify the responsiveness of NK cells by injecting the NKp46 mAb, at steady state, in WT mice. Short treatments lasting 24 to 72 hours were sufficient to saturate NKp46 receptors (fig. S15, A to D), but did not increase the reactivity of NK cells (fig. S15, E and F). By contrast, in vivo blockade of NKp46 by the mAb for 13 days was sufficient to enhance NK cell responsiveness to YAC-1 tumors (Fig. 4B). Altogether, our results reveal the role of the conserved activating NK cell receptor NKp46 in NK cell function. They also pave the way for the counterintuitive use of blocking NKp46 mAbs to boost NK cell activity, as an immunostimulation strategy of particular relevance for patients with T cell deficiencies, such as those occurring after hematopoietic stem cell transplantation.

Fig. 4

NKp46 mAbs boost NK cell responsiveness. (A) NDE mice were treated with diphtheria toxin to deplete NK cells. Upon reconstitution of the NK cell compartment, mice were treated with NKp46 or control mAbs every 2 to 3 days for 9 days. Spleen cells from NKp46-treated and control antibody–treated mice were analyzed on day 13 (see fig. S13). Data show frequencies of IFN-γ–producing and CD107a+ cells among IL-2–activated NK cells from NDE mice after a coculture with YAC-1 tumor targets. n = 8, ***P = 0.0009, Mann-Whitney test. (B) The same experiments were carried out in WT mice. n = 8 to 9, *P = 0.03 and **P = 0.0081, Mann-Whitney test. All data result from a pool of two independent experiments.

Supporting Online Material

Materials and Methods

Figs. S1 to S15

Table S1

References (3338)

References and Notes

  1. Acknowledgments: We thank J. Ewbank, M. Bléry, J.-C. Andrau and P. Kruse for advice; D. Dilg for bioinformatics analysis; M. Dalod for MCMV reagents; M. C. Carroll for H1N1; G. Lauvau for Lm-OVA; M. Roger for mice handling; and the Centre d'Immunologie de Marseille-Luminy (CIML) mouse house and cytometry core facilities. This work was supported by an European Research Council advanced grant (E.V. and S.U.); Functional Genomics in Mutant Mouse Models as Tools to Investigate the Complexity of Human Immunological Disease (MUGEN), Network of Excellence and Mechanisms to Attack Steering Effectors of Rheumatoid Syndromes with Innovated Therapy Choices (MASTERSWITCH), Integrating Project from European Union (B.M. and M.M.); University of Manitoba Dean of Medicine Strategic Fund and Natural Sciences and Engineering Research Council (S.K.); Agence Nationale de la Recherche (E.V. and S.U.); Equipe labellisée “La Ligue,” Ligue Nationale contre le Cancer (E.V. and S.U.); Agence pour la Recherche sur le Cancer (E.M.N.); Axa research fund (B.N.J.); and institutional grants from INSERM, CNRS, and Aix-Marseille University to the CIML. E.V. is a cofounder of and shareholder in Innate-Pharma. E.N.M., E.V., and S.U. designed, analyzed the experiments, and wrote the paper; C.B. identified Noé mouse during ENU screen; E.N.M. and A.F. performed and analyzed the experiments; S.K. and S.M. provided lentiviral vectors and protocols to transduce NK cells; A.D.G. generated Helios-expressing vectors; B.N.J., F.V., M.G., I.G.G., J.E., and S.C.H. performed the sequencing and the bioinformatics analysis; L.N.G. generated the model for NKp46W32R; M.M. and B.M. initiated and conducted the early phases of the ENU screen; and B.B. and E.B. helped in the ENU screen. The data reported in this paper are tabulated in the main paper and in the Supporting Online Material. Microarray data have been deposited at the National Center for Biotechnology Information GEO repository under accession no. GSE13229 and have been reported elsewhere (13). Ensembl accession number for Ikzf2 is ENSMUST00000027146. Material Transfer Agreements are required for use of the following reagents: Noé mice, NKp46-specific mAb, NKp46iCre/iCre mice, Helios short hairpin RNA, and control lentiviral vectors. The invention (U.S. Patent Application 61/499,485; NKp46-mediated NK cell tuning; E.N.M., S.U., and E.V.) relates to compounds that inhibit NKp46.

Stay Connected to Science

Navigate This Article