Natural killer (NK) cells are crucial in resistance to certain viral infections, but the mechanisms used to recognize infected cells remain largely unknown. Here, we show that the activating Ly49P receptor recognizes cells infected with mouse cytomegalovirus (MCMV) by a process that requires the presence of H2-Dk and the MCMV m04 protein. Using H2 chimeras between H2-Db and -Dk, we demonstrate that the H2-Dk peptide-binding platform is required for Ly49P recognition. We identified m04 as a viral component necessary for recognition using a panel of MCMV-deletion mutant viruses and complementation of m04-deletion mutant (Δm04) virus infection. MA/My mice, which express Ly49P and H2-Dk, are resistant to MCMV; however, infection with Δm04 MCMV abrogates resistance. Depletion of NK cells in MA/My mice abrogates their resistance to wild-type MCMV infection, but does not significantly affect viral titers in mice infected with Δm04 virus, implicating NK cells in host protection through m04-dependent recognition. These findings reveal a novel mechanism of major histocompatability complex class I–restricted recognition of virally infected cells by an activating NK cell receptor.
Humans (1) and mice (2) lacking NK cells are vulnerable to certain viral infections. NK cells patrol for aberrant cells, which they eliminate through cytotoxic activity and secretion of cytokines (3). NK cell activation depends on a complex array of inhibitory and activating receptors. Inhibitory receptors recognizing MHC class I molecules, including human killer cell immunoglobulin-like receptors (KIR) and mouse Ly49 receptors, play a predominant role. Both receptor families contain activating receptors, but the ligands are not well characterized. Viruses have devised numerous strategies to modulate the function of NK cells and have influenced the evolution of NK cell receptors.
The identification of mouse genes that protect from mouse cytomegalovirus (MCMV) infection has provided insights into the NK cell–mediated control of viral proliferation. In C57BL/6 mice, NK cells expressing the activating Ly49H receptor that binds directly to m157, which is a MCMV-encoded cell surface glycoprotein (3), control viral replication early after infection (4). The essential role of Ly49H was confirmed by the transfer of MCMV resistance to genetically susceptible mice by Ly49h transgenesis (5), whereas absence of Ly49h in C57BL/6 mice abolishes resistance (6). Moreover, C57BL/6 mice are susceptible to MCMV mutants lacking m157 (Δm157) (7). Collectively, these data demonstrated the importance of activating NK cell receptor–specific recognition of a virus-encoded ligand in host resistance.
Despite lacking Ly49h, MA/My mice are resistant to MCMV (8, 9), which depends on the specific combination of MA/My alleles at Ly49 and MHC loci. The activating receptor Ly49P in MA/My mice recognizes MCMV-infected cells, and Ly49P-dependent activation is abrogated by an anti–H2-Dk antibody (9). These results suggested that MA/My resistance is conferred by NK cell–mediated recognition of infected cells by a mechanism involving Ly49P, H2-Dk, and an additional molecule expressed during MCMV infection.
CMVs possess numerous genes dedicated to manipulating the host immune response. Three MCMV gene products alter host MHC class I expression: m04, m06, and m152 (10). m152 retains peptide-loaded MHC class I in the endoplasmic reticulum cis-Golgi intermediate compartment (11). m04 and m06 carry cytoplasmic (CT) motifs involved in cargo-sorting pathways; m06 redirects MHC class I to the late endosome–lysosome pathway for degradation, thus preventing antigen presentation (12). m04 associates with MHC class I in the ER, and these complexes travel to the cell surface (13, 14); however, m04 does not down-regulate MHC class I.
Here, we found that Ly49P recognition depends on the following: H2-Dk expression on the infected cell, the peptide-binding groove of H2-Dk, and m04. Infection of MA/My mice with a m04-deletion MCMV mutant resulted in increased susceptibility, suggesting that Ly49P-mediated NK cell recognition of H2-Dk complexed with m04 on the infected cell is necessary to attenuate virus replication.
RESULTS AND DISCUSSION
H2-Dk is necessary for Ly49P recognition of MCMV-infected cells
In view of MCMV's interference with host MHC class I expression, we examined whether sufficient amounts of H2 are expressed on the surface of infected cells to permit recognition by Ly49P (Fig. 1 A). BALB.K mouse embryonic fibroblasts (MEFs) were infected with MCMV expressing GFP (15), followed by staining with anti–H2-Dk or anti–H2-Kk antibodies (Fig. 1 A). Flow cytometric analysis revealed populations of cells that were noninfected (GFP-negative), mildly infected (GFP-dim), and highly infected (GFP-bright). We observed increased MHC class I expression on the GFP-negative cells, likely caused by up-regulation by type I IFN secreted by virus-infected cells. Importantly, GFP-dim MCMV-infected cells expressed H2-Dk and -Kk on their surface (Fig. 1 A). Therefore, MHC class I expression is diminished, but not totally ablated, in MCMV-infected cells (16).
Activation of Ly49P-expressing NFAT-GFP reporter cells co-cultured with MCMV-infected H2k MEFs is abolished by anti–H2-Dk, but not anti–H2-Kk, blocking antibodies (9). We transduced NIH3T3 cells, which endogenously carry H2q2, with H2-Dk and/or -Kk, and assessed surface expression (Fig. 1 B). None of the transduced cells stimulated a human NK cell line transduced with Ly49P (NKL-Ly49P) to secrete INF-γ in the absence of infection (Fig. 1 C). However, when infected with MCMV, cells expressing H2-Dk stimulated NKL-Ly49P cells to produce IFN-γ (Fig. 1 C). Ly49H-expressing NKL (NKL-Ly49H) cells secreted INF-γ when co-cultured with MCMV-infected cells, irrespective of H2 expression (Fig. 1 D). Therefore, H2-Dk, but not H2-Kk, is necessary for Ly49P activation in the presence of infection.
α1α2 domains of H2-Dk are necessary for Ly49P recognition
MHC class I molecules consist of a heavy chain with α1, α2, and α3 domains associated with β2-microglobulin (β2m). The peptide-binding platform is composed of two parallel α helices positioned on top of a β-plated sheet, formed by the α1 and α2 domains. To determine which domains of H2-Dk are recognized by Ly49P, we generated chimeras by swapping domains between H2-Dk and H2-Db (Fig. 2 A). α1 and α2 from H2-Db were fused with the α3, transmembrane (TM), and CT domains of H2-Dk (Dbα1α2Dkα3). α1, α2, and α3 from H2-Db were fused with the TM and CT domains of H2-Dk (Dbα1α2α3Dk). α1 and α2 from H2-Dk were fused with the α3, TM, and CT domains of H2-Db (Dkα1α2 Dbα3). α1, α2, and α3 from H2-Dk were fused with the TM and CT domains of H2-Db (Dkα1α2α3Db). Chimeric proteins were expressed in high amounts on the surface of NIH3T3 cells (Fig. 2 B).
These cells, which were infected or not infected with MCMV, were used to stimulate NKL-Ly49H or NKL-Ly49P cells. None of the uninfected cells induced significant amounts of IFN-γ secretion. In contrast, MCMV-infected cells expressing H2-Dk, but not H2-Db, activated the NKL-Ly49P cells (Fig. 2 C). When α1 and α2 of H2-Dk were replaced with those of H2-Db, NKL-Ly49P cells were not stimulated (Fig. 2 C). Interestingly, when α1 and α2 of H2-Db were replaced by those of -Dk, infected cells stimulated significantly higher amounts of IFN-γ from NKL-Ly49P cells (P < 0.01; Fig. 2 C). Thus, the peptide-binding platform specified by α1 and α2 of H2-Dk is necessary for activating Ly49P recognition. The same region is important for contact between inhibitory Ly49 receptors and their MHC class I ligands (17).
m04 is critical for Ly49P recognition
MCMV genes involved in the manipulation of MHC class I might be involved in recognition by Ly49P. We tested MCMV deletion mutant viruses lacking members of the m145 (Δm144-158, Δm151-158, Δm151-165, Δm152, and Δm157) and the m02 (Δm1-22) gene families, including the m04 and m06 genes. Infected MEFs were used to stimulate Ly49P or Ly49H NFAT-GFP reporter cells. Absence of the m145 family did not affect Ly49P recognition (Fig. 3). As expected, Ly49H reporters did not respond to MEFs infected with mutants lacking m157 (Fig. 3) (9, 18). Although deletion of the m1-22 region did not affect Ly49H recognition, it completely abolished Ly49P reporter cell activation. We also assayed deletion mutants lacking m04 (Δm04), m06 (Δm06), and/or m152 (Δm152). Deletion of m04, but not m06 or m152, completely abrogated Ly49P-mediated activation (Fig. 3).
m04 complements Δm04 MCMV in Ly49P-mediated activation
To determine whether m04 protein is sufficient to activate Ly49P reporter cells, we transiently transfected m04 into immortalized MEF.K cells. When m04-transfected MEF.K cells were used to stimulate Ly49P-NFAT-GFP reporters, no significant activation was detected (Fig. 4 A). Ly49P-mediated recognition was rescued when m04-transfected cells were infected with Δm04 MCMV. m90 and m157 failed to complement Δm04 MCMV infection (Fig. 4 A). Thus, m04 is necessary, but not sufficient, for Ly49P recognition.
To verify m04 expression, we transduced MEF.K with a V5-tagged m04 and sorted V5-m04–expressing cells (V5-m04 MEF.K). As a control, we transduced MEF.K with a monomeric red fluorescent protein (mRFP-MEF.K). Uninfected cells did not activate the Ly49P-NFAT-GFP reporters (Fig. 4 B). In contrast to mRFP-MEF.K, Δm04 MCMV infection of V5-m04-MEF.K stimulated reporter cells at levels comparable to WT MCMV. Using FACS, we confirmed the presence of V5-m04 and H2-Dk on the cell surface, and distinguished infected cells by intracellular anti-m06 staining (Fig. 4 C). H2-Dk was expressed on the surface of ∼20% of the cells infected with mutant or WT virus. To examine V5-m04 expression during infection, we gated cells based on H2-Dk and m06 expression (Dk−m06+, Dk+m06+, and Dk+m06−; Fig. 4 D). Consistent with reports of the association of m04 with H2 (19), we observed a clear correlation between V5-m04 and H2-Dk (Fig. 4 D). The largest frequency of V5-m04–expressing cells and the highest amounts of V5-m04 were seen on Dk+m06+ cells, demonstrating coexpression of V5-m04 and H2-Dk at the surface of infected cells. These results suggest that a factor in addition to m04 protein and H2-Dk in the infected cell is required for Ly49P-dependent recognition.
m04 is physically associated with H2 molecules. Peptide-loading facilitates m04–H2 complex formation, and β2m is essential (19). Our results confirm that m04–H2 complexes are present on the cell surface and available for Ly49P recognition. However, m04 must play a role other than simply increasing the expression of MHC class I because Ly49P reporter cells were unresponsive to uninfected MEFs treated with type I IFN to increase H2 expression (9). m04 might cause an allosteric change in H2-Dk, thereby permitting interactions with Ly49P. However, expression of m04 in MCMV-infected fibroblasts does not affect stimulation of cytotoxic T cells, indicating native H2 folding and recognition (20). Alternatively, m04 might provide a specific peptide recognized by Ly49P in a H2-Dk–dependent manner. Although it remains to be determined whether m04 peptides are involved in Ly49P recognition, this might not be the case because MEF.K transfected with m04 failed to stimulate the Ly49P reporters in the absence of infection. m04 associates via its TM domain with H2 independently of MCMV infection. However, in uninfected cells, export of m04–H2 complexes to the cell surface is less efficient (Fig. 4 D) (19). Thus, other infection-induced proteins might be required for assembly or transport of H2-Dk–m04 complexes to the surface of infected cells for recognition by Ly49P.
m04 is critical for MCMV resistance in MA/My mice
Initially, m04 was proposed to inhibit NK cell activation, thereby favoring viral replication (13). Conversely, our results suggest that m04 activates Ly49P-bearing NK cells, benefiting host defense. To evaluate the role of m04 in vivo, we infected MA/My mice with WT or Δm04 MCMV and measured viral load (Fig. 5 A). 4 d after infection (a.i.), mice infected with WT MCMV had a 100-fold lower viral titer in the spleen and liver than mice infected with Δm04 MCMV. Thus, removal of m04 from MCMV genome diminishes MA/My resistance.
To determine the role of NK cells, MA/My mice were depleted of NK cells with anti-NK1.1 mAb. Splenic viral titers in untreated or NK cell–depleted MA/My mice infected with Δm04 MCMV were equivalent (Fig. 5 B). In contrast, untreated C57BL/6 mice infected with Δm04 MCMV showed a low viral load, equivalent to infection with WT MCMV, whereas NK cell–depleted C57BL/6 mice had a high viral burden when infected with either Δm04 or WT MCMV (Fig. 5 B). Although neutralizing antibodies against Ly49P are not available to formally prove that Ly49P-bearing NK cells are responsible for resistance, our studies strongly support a role for Ly49P–H2-Dk–m04 axis in the NK cell–mediated control of MCMV.
MCMV resistance in F2 crosses between MCMV-resistant MA/My and MCMV-susceptible BALB/c mice indicated that only mice homozygous for Ly49 and H2k MA/My-derived alleles were resistant to MCMV (9). We identified Ly49p as the most likely gene involved in resistance based on: (a) its genetic location, (b) its absence in H2k MCMV-susceptible mice, and (c) its recognition of MCMV-infected cells using Ly49P reporter cell assays. Although the importance of the H2k haplotype for MCMV resistance was established previously, the precise mechanism had not been determined (9, 21). Here, we demonstrate that the peptide-binding α1 and α2 platform of H2-Dk is an absolute prerequisite for Ly49P recognition. Moreover, this recognition and NK cell–dependent resistance in MA/My mice is dependent on m04.
MCMV isolates from wild mice indicate that the m04 gene is very polymorphic, like m157 (22, 23). Variation in m157 is viewed as a result of the strong selective pressure applied by Ly49H+ NK cells, as demonstrated by a high rate of mutations in m157 isolated from MCMV passaged in Ly49h-bearing mice (24, 25). Most m04 variation falls in the extracellular domain (26), which might influence the binding of Ly49P, suggesting that m04 polymorphisms emerge to escape NK cells.
Retention of a viral protein that enhances the killing of infected cells is counterintuitive, yet m04 might be beneficial for the virus. For example, through the stabilization of H2 on the surface of infected cells, m04 might render H2-Dk more accessible to inhibitory receptors and suppress NK cell activation in certain mouse strains. This has been proposed for m157, which is recognized not only by Ly49H but also by a closely related inhibitory receptor, Ly49I (18). Alternatively, m04 might confer host survival and thereby increase virus dissemination in the population, as proposed for B15R in vaccinia virus, which encodes a soluble IL-1 receptor. Deletion of B15R increases mortality in mice, suggesting that the viral blockade of IL-1β diminishes the acute phase response and modulates severity of the disease (27). Regulation of antigen presentation by m04, in concert with m06 and m152, might influence the adaptive immune response to the virus' benefit (28). Regardless of how m04 evolved, the increased viral titers observed in MA/My mice infected with Δm04 MCMV demonstrate that m04 is crucial for effective NK cell–mediated control of viral replication. Given the fact that intact or NK cell–depleted MA/My mice infected with Δm04 MCMV had similar viral titers, it seems that recognition of m04 is responsible for most of the NK cell–mediated protective effect.
The Ly49P–H2-Dk–m04 model system provides new perspectives in understanding the relationship between KIR and HLA in human disease. Genetic epidemiological studies have identified several combinations of KIR and MHC loci associated with resistance to HIV and hepatitis C virus (29); however, the molecular interactions responsible for these events remain unresolved. Our studies highlight the significance of activating NK cell receptor–MHC class I interactions in recognition of infection, warranting further investigation of their role in host resistance to other pathogens.
MATERIALS AND METHODS
Mice were purchased at The Jackson Laboratory. Animal protocols or experiments were approved by Canadian Council on Animal Care (CCAC) and McGill University Animal Resources Center.
Δm04Δm06, Δm04Δm152, Δm06Δm152, Δm04Δm06Δm152, Δm144-158, Δm01-22, Δm151-m165 (DMS94.5), Δm151-m158 (MC96.73), and WT BAC-derived (MW97.01) viruses were previously described (16, 30–33). 6–8-wk-old mice were infected i.p. with 5 × 103 PFU/mouse. Viruses were quantified by plaque assays (9). For NK cell depletion, mice were injected i.v. with 150 µg anti-NK1.1 (PK136) mAb 48 h before infection.
Primary BALB.K MEFs were prepared as previously described (9). BALB.K MEFs were immortalized (MEF.K) with polyoma large T antigen (gift from M. Fried, University of California, San Francisco, San Francisco, CA). H2-Dk, H2-Db, or recombinant H2 cDNA in pMx-puro were used to transduce NIH3T3 cells. MEF.K were transiently transfected with pMx-puro containing m04 (35), m157, or m90. m04 lacking its endogenous leader sequence was cloned into pMx-neo that contains a human CD8 leader, followed by a V5 tag. mRFP cDNA was cloned into pMx-puro. MEF.K or NIH3T3-Dk cells were infected and sorted for high expression by flow cytometry.
MEF.K infected by GFP-MCMV (MOI 0.5 or 0.1) for 48 h were stained with biotin-conjugated anti–H2-Dk (15–5-5, ATCC) or anti-H2-Kk (16–3-22, ATCC), followed by APC-conjugated streptavidin (BD). NIH3T3 cells stably expressing H2-Dk, -Kk, -Db, or chimeras were stained with FITC-conjugated anti–H2-Dk, –H2-Kk, or –H2-Db (28–14-8; BD). Infected (MOI 0.5) or uninfected MEF.K, V5-m04 MEF.K, or V5-m04 NIH3T3-Dk cells were detached 24 h a.i., blocked with anti-CD16/CD32 mAb (eBioscience), and stained with anti-V5 (V5-10; Sigma-Aldrich), PE-conjugated donkey anti–mouse IgG (eBioscience), biotin-conjugated anti-H2-Dk, and APC-conjugated streptavidin. Cells were fixed, permeabilized, and stained with Alexa Fluor 488–conjugated anti-m06 antibody (CROMA 229) (36). Cells were analyzed on a FACSCalibur (BD).
Stimulator cells were infected (MOI 0.5 or 1) with MCMV or MCMV-deletion mutants for 22 h. NFAT-GFP reporter cells were co-cultured overnight with stimulators in 48-well plates at a 1:1 ratio and analyzed by flow cytometry. NKL reporter cells were co-cultured overnight with stimulators at 5:1 ratio in 96-well plates, and IFN-γ was measured by ELISA (eBioscience).
Differences between groups were calculated with two-way ANOVA analysis assuming not-repeated measures, followed by Bonferroni after tests. Results with P < 0.05 were considered significant.
Supported by Canadian Institutes of Health Research MOP-7781 (S.M. Vidal); National Institutes of Health AI068129 (L.L. Lanier), Croatian Ministry of Science, Education, and Sport 062-0621261-1263, and National Institutes of Health HD044721 (S. Jonjic). T. Sun is a Cancer Research Institute fellow. A. Kielczewska is supported by McGill Majors Fellowship. M. Pyzik is supported by CIHR Doctoral Award. A. Krmpotic is supported by the Howard Hughes Medical Institute Scholars grant. L.L. Lanier is an American Cancer Society Research Professor. S.M. Vidal is a Canada Research Chair.
The authors have no conflicting financial interests.
A. Kielczewska and M. Pyzik contributed equally to this paper. S. Jonjic, L.L. Lanier, and S.M. Vidal contributed equally to this paper.