Cutting Edge: Viral Infection Breaks NK Cell Tolerance to “Missing Self”

Abstract

NK cells attack cells lacking MHC class I, yet MHC class I-deficient mice have normal numbers of NK cells with intact, albeit diminished, functions. Moreover, wild-type NK cells are tolerant of MHC class I-deficient cells in mixed bone marrow chimeras. Here, we investigated how the absence of MHC class I affects NK cells. NK cells from β2-microglobulin-deficient (B2m^−/−) and wild-type mice exhibit similar phenotypic and functional characteristics. Both B2m^−/− and wild-type Ly49H⁺ NK cells proliferated robustly and produced IFN-γ after infection with mouse cytomegalovirus (MCMV). NK cells in mixed wild-type:B2m^−/− chimeric mice were initially tolerant of MHC class I-deficient host cells. However, this tolerance was gradually lost over time and after MCMV infection was rapidly broken, with a pronounced rejection of host B2m^−/− hematopoietic cells. Thus, although NK cells can be held in check against “missing self”, acute inflammation driven by infection can rapidly break established self-tolerance.

Introduction

Karre's “missing self” hypothesis posits that NK cells can recognize or “sense” the loss of MHC class I (or “self”) on potential target cells and become activated (1). Over two decades of research has supported this hypothesis, with several studies convincingly showing that NK cells in wild-type mice are solely responsible for the rejection of MHC class I-deficient cells (2-4). Interestingly, MHC class I-deficient animals (such as B2m^−/−, TAP1^−/−, or H2-D^−/−K^−/− mice) contain relatively normal numbers of NK cells (4-7), yet do not succumb to autoimmunity. If NK cell-mediated rejection of MHC class I-deficient cells is indeed robust, why don't NK cells in the periphery of B2m^−/− mice target self-tissues for destruction? What are the regulatory mechanisms that keep NK cells in B2m^−/− mice in check?

Recent studies have suggested that NK cells undergo “licensing” or “disarming” in the bone marrow, a process that dictates the proper development of NK cells (8, 9). Immature NK cells expressing inhibitory receptors that can interact with their autologous MHC class I molecules go on to functional maturity. In the absence of these interactions, NK cells either fail to complete their maturation or are chronically stimulated, with either scenario resulting in hypo-responsive NK cells in the periphery (8, 9). Thus, these models predict that in mice lacking MHC class I, selective pressures ensure self-tolerance in developing NK cells that are unable to engage their inhibitory receptors.

Challenging the “licensing” and “disarming” models of NK cell unresponsiveness are studies showing robust NK cell activation and effector function in B2m^−/− mice following viral, bacterial, and parasite infections (8, 10-12). In this study, we addressed these issues by examining the phenotype and function of NK cells from B2m^−/− mice, either in non-infected or virus-infected animals. Furthermore, we generated mixed wild-type:B2m^−/− bone marrow chimeric mice in order to study the tolerance of wild-type NK cells to B2m^−/− host cells that are “missing self” in an environment where these NK cells were resting or activated by viral infection.

Materials and Methods

Mice and infections

C57BL/6 and congenic (CD45.1) mice were purchased from the National Cancer Institute. β2-microglobulin-deficient B6 mice were bred at UCSF. Experiments were done according to the UCSF Institutional Animal Care and Use Committee guidelines. Mice were infected by intraperitoneal injections of MCMV (Smith strain, 5×10⁴ PFU). NK cells were depleted by injection with 200 μg anti-NK1.1 mAb PK136.

Generation of mixed bone marrow chimeric mice

Mixed bone marrow chimeric mice were generated as described previously (13). Briefly, NK cells were depleted from B6 (CD45.1 × CD45.2) mice by injection with 200 μg anti-NK1.1 mAb, and then mice were irradiated with 1000 rad from a ¹³⁷Cs source. One day later, mice were injected intravenously with a 1:1 or 10:1 mixture of bone marrow cells isolated from wild-type congenic (CD45.1⁺) and B2m^−/− (CD45.2⁺) mice. Donor mice were also pre-treated by injection with 200 μg anti-NK1.1 mAb 24 hrs prior to harvest of bone marrow to eliminate donor NK cells from the bone marrow graft. Peripheral blood leukocytes from chimeric mice were analyzed at various time points after bone marrow reconstitution to assess the ratio of wild-type and B2m^−/− cells, and the animals were infected at approximately 10 weeks following bone marrow reconstitution.

Ex vivo stimulation of NK cells

DOTAP-treated tissue culture plates were coated with antibodies specific for Ly49H (generously provided by Dr. Wayne Yokoyama) (or uncoated wells as negative control) and whole splenocytes were incubated for 5 hours at 37°C in the presence of Brefeldin A (BD Pharmingen), followed by staining for intracellular cytokines (14).

Flow cytometry

Single-cell suspensions of spleen and liver were prepared. Fc receptors were blocked by using anti-CD16+CD32 mAb (2.4G2) prior to surface staining with the indicated antibodies (purchased from BD or eBiosciences). Samples were acquired on a LSRII (BD) and analyzed using FlowJo software (TreeStar).

Statistical Analysis

Statistical differences in percentage of NK cells from wild-type and B2m^−/− mice making IFN-γ ex vivo after stimulation with plate-bound anti-Ly49H antibodies were determined by using the two-tailed unpaired Student's t test.

Results and Discussion

Phenotypic similarities between NK cells from wild-type and B2m^−/− mice

According to the “licensing” or “disarming” models of development, NK cells that mature in the absence of MHC class I might show phenotypic abnormalities. Others have noticed subtle or no differences in maturation markers (such as Mac-1 or CD43) on NK cell subsets between wild-type and B2m^−/− mice (8, 15). We examined survival and activation markers and expression of the activating receptors Ly49D and Ly49H on peripheral NK cells from wild-type and B2m^−/− mice. Although the percentage of CD8⁺ T cells was severely diminished in B2m^−/− mice compared to wild-type mice, overall NK cell numbers showed no statistically significant differences in the spleen (p>0.05) (Fig. 1a). The percentages of NK cells expressing the activating Ly49D, Ly49H, and NKG2D and the amounts of these receptors were also not statistically different on NK cells from wild-type and B2m^−/− mice (p>0.05) (Fig. 1a-b). Furthermore, the survival and activation marker phenotype of NK cells from both wild-type and B2m^−/− mice were comparable (Fig. 1b). In particular, the expression of CD122, CD25, CD62L, CD27, Ly6C, and CD69 markers, which have been correlated with NK cell homeostasis and activation, was comparable on wild-type and B2m^−/− NK cells. The notable exception is KLRG1, which was on fewer NK cells from B2m^−/− compared with wild-type mice (Fig. 1b), as previously reported (16, 17). Overall, other than the previously described effects of MHC class I on expression of certain inhibitory Ly49 receptors (18-20), there are few phenotypic differences between wild-type and B2m^−/− NK cells, demonstrating that NK cell development is largely intact and there is no evidence for chronic activation of NK cells in the absence of MHC class I.

Figure 1.

Phenotype of NK cells from wild-type and B2m^−/− mice. a, Percentages of total NK cells and NK cell subsets from the spleens of wild-type and B2m^−/− mice were compared. Left plots were gated on total live cells in the spleen. Right plots were gated on total NK cells (CD3⁻, NK1.1⁺) in the spleen. b, NK cells from wild-type (solid lines) and B2m^−/− (dotted lines) mice were compared for expression of different cell surface markers. Data are representative of 3 independent experiments with 2−3 mice per group analyzed in each independent experiment.

Functional similarities of wild-type and B2m^−/− NK cells

We compared the ability of NK cells from wild-type and B2m^−/− mice to produce IFN-γ ex vivo in response to plate-bound antibodies against the activating receptor Ly49H, the receptor for the MCMV m157 glycoprotein (21, 22). Both wild-type and B2m^−/− NK cells produced similar amounts of IFN-γ; however, about 2-fold more wild-type NK cells made IFN-γ than B2m^−/− NK cells (p < 0.01) (Fig. 2a). In concordance with these findings, prior studies have reported that B2m^−/− NK cells mediated slightly less lytic activity against NK-sensitive Yac-1 cell targets and produced less IFN-γ when stimulated in vitro with anti-NK1.1 or anti-Ly49D compared with wild-type NK cells (3, 4).

Figure 2.

Function of NK cells from wild-type and B2m^−/− mice. a, Plots show percentage of NK cells from wild-type and B2m^−/− mice that produce IFN-γ in response to stimulation by anti-Ly49H antibodies. b, Wild-type and B2m^−/− mice were infected with MCMV and NK cells were analyzed at 36 hours post-infection (PI) for intracellular IFN-γ production and LAMP-1 expression. Top panels show percentages of IFN-γ production within the Ly49H⁺ and Ly49H⁻ NK cell populations. c, Wild-type and B2m^−/− mice were infected with MCMV and NK cells were analyzed at 8 days PI (compared to uninfected mice) for expression of Ly49H and KLRG1. d, Graph shows the percentage of Ly49H⁺ NK cells in the spleens and livers of uninfected and day 8 PI wild-type and B2m^−/− mice. Error bars on graph display s.e.m. (n = 3−5) and data are representative of 3 independent experiments with 3−5 mice per group in each independent experiment.

Within 48 h after infection of B6 mice with MCMV, all NK cells, including both Ly49H⁻ and Ly49H⁺ NK cells, become stimulated, as documented by upreguation of activation antigens and production of IFN-γ. This global and non-specific MCMV-induced activation, likely a consequence of the production of type I IFN and pro-inflammatory cytokines such as IL-12 and IL-18 (23), is followed by the specific expansion of Ly49H⁺ NK cells. In order to evaluate the function of B2m^−/− NK cells in a more physiological context, we infected wild-type and B2m^−/− mice with MCMV and measured IFN-γ and LAMP-1 (a measure of degranulation) in NK cells ex vivo at 36 hours post-infection. Again, both wild-type and B2m^−/− NK cells responded robustly and to a very comparable magnitude (Fig. 2b). In concordance with our in vivo findings, Yokoyama and colleagues previously reported that culturing NK cells in vitro in IL-12 permitted B2m^−/− NK cells to make IFN-γ at levels comparable to wild-type NK cells (15). The expansion of Ly49H⁺ NK cells following MCMV infection was similar between wild-type and B2m^−/− mice in the spleen and liver (Fig. 2c-d). Furthermore, although KLRG1 was initially expressed on a lower frequency of B2m^−/− NK cells, after MCMV infection KLRG1 was comparably up-regulated on both B2m^−/− and wild-type NK cells in liver and spleen (Fig. 2c). Together, these data suggest that NK cells from B2m^−/− mice are capable of robustly responding and functioning during a viral infection.

Our findings that Ly49H⁺ NK cells from B2m^−/− mice can produce cytokine and expand provide a mechanism for prior studies demonstrating that NK cell-mediated control of MCMV replication is equivalent in wild-type and β2m-deficient mice (11, 12). Thus, although NK cells from mice lacking MHC class I are less responsive when assayed in vitro, these NK cells developing in the absence of MHC class I respond robustly and efficiently when challenged with a relevant pathogen in vivo.

B2m^−/− cells gradually decline in chimeric wild-type:B2m^−/− mice

Raulet and colleagues previously reported that tolerance to “missing self” was induced in wild-type NK cells against B2m^−/− hematopoietic cells in mixed bone marrow chimeras reconstituted with a mixture of wild-type and B2m^−/− hematopoietic stem cells (24). Similarly, we generated chimeras by NK cell-depleting and lethally irradiating congenic B6 mice (CD45.1⁺ × CD45.2⁺), followed by reconstitution with a mixture of wild-type (CD45.1⁺) and B2m^−/− (CD45.2⁺) bone marrow stem cells, which had been depleted of NK cells, at ratios of 1:1 and 10:1 (Fig. 3a). Consistent with previous findings (24), we observed that both wild-type and B2m^−/− hematopoietic cells populated the irradiated recipients at 5 weeks post-reconstitution. However, the wild-type cells were detected at a higher percentage than the original input frequency (Fig. 3a). Surprisingly, by 3 months post-reconstitution, the 1:1 chimeras were found at a 3:1 frequency in favor of wild-type hematopoietic cells, and the 10:1 chimeras contained almost no B2m^−/− cells (Fig. 3a). Thus, while tolerance to “missing self” was initially established in these mixed bone marrow chimeric mice, longitudinal studies of these mice revealed a slow, but steady loss of cells lacking MHC class I (Fig. 3b). These findings suggest the possibility that either the B2m^−/− hematopoietic cells were less fit than the wild-type hematopoietic cells or, alternatively, that these B2m^−/− cells were being actively eliminated by the wild-type NK cells.

Figure 3.

Gradual loss of B2m^−/− cells in mixed wild-type:B2m^−/− chimeric mice. a, 1:1 (top panels) and 10:1 (lower panels) wild-type:B2m^−/− chimeric mice were generated. Left density plots show percentages of wild-type (CD45.1⁺) and B2m^−/− (CD45.2⁺) bone marrow cells prior to transfer into irradiated recipient mice. Histogram plot shows H-2K^b staining performed on wild-type (solid line) and B2m^−/− (dotted line) cells. Middle and right density plots show percentages of wild-type and B2m^−/− cells at 5 weeks and 3 months post-irradiation, respectively. b, Graphs show percentage of B2m^−/− cells (CD45.2⁺) at various time points post-irradiation in 1:1 chimeras (left graph) and 10:1 chimeras (right graph). Error bars on graphs display s.e.m. (n = 3−5 at each time point) and data are representative of 2 independent experiments with 3−5 mice per time point.

B2m^−/− cells are rapidly rejected by wild-type NK cells in chimeric mice following viral infection

When we infected the mixed wild-type:B2m^−/− bone marrow chimeric mice with MCMV, we observed a precipitous drop in B2m^−/− cell numbers (CD45.2⁺) over the course of one week (Fig. 4a-b). NK cells mediated this rapid rejection of B2m^−/− cells because when we treated chimeric mice with a NK cell-depleting antibody, B2m^−/− cells were maintained (Fig. 4a-b). Thus, although tolerance to “missing self” can be established by generating unresponsive or “disarmed” NK cells that develop in the presence of MHC class I-deficient hematopoietic cells, this tolerance is easily broken during inflammation, leading to the rapid destruction of hematopoietic cells lacking MHC class I.

Figure 4.

Rapid rejection of B2m^−/− cells in chimeric mice following infection. a, Upper plots show percentages of wild-type (CD45.1⁺) and B2m^−/− (CD45.2⁺) cells for one week in mixed wild-type:B2m^−/− chimeric mice following MCMV infection. Lower plots show percentages of wild-type and B2m^−/− cells for 2 months in mice treated with NK cell-depleting antibodies prior to MCMV infection. b, Graphs show percentage of B2m^−/− cells (CD45.2⁺) at various time points in mice infected with MCMV (left graph) and mice treated with NK cell-depleting antibodies prior to MCMV infection (right graph). Insets show percentages of CD3⁻, NK1.1⁺ cells at day 7 PI, demonstrating the efficacy of NK cell-depleting antibodies. Error bars on graphs display s.e.m. (n = 3−5 at each time point) and data are representative of 2 independent experiments with 3−5 mice per time point.

These observations suggest that the NK cell tolerance of “missing self” is actively maintained, not permanent, and can be readily overcome by activation of the NK cells by a physiological stimulus, i.e. viral infection. Similarly, Karre and colleagues reported that NK cells developing in a H-2^d transgenic B6 mouse, where H-2^d was expressed in a mosaic pattern in these animals, are tolerant (25). Although the NK cells in these mice expressing the H-2^d transgene were tolerant of surrounding cells not expressing H-2^d in vivo, the tolerance was broken when the transgene-bearing NK cells from these animals were cultured with IL-2 in vitro and assayed for their ability to kill targets lacking H-2^d (25). Thus, as in the mixed bone marrow chimeric mice, NK cell tolerance to “missing self” in these H-2^d transgenic B6 mice was not permanent and could be broken by NK cell activation.

Concluding Remarks

These findings raise several important questions. What are the molecular mechanisms of peripheral NK cell tolerance to “missing self” in the steady-state (non-inflammatory environment)? What accounts for NK cell-mediated rejection of MHC class I-deficient cells during inflammation, when tolerance is broken? When wild-type NK cells in MHC class I-deficient animals do become activated during inflammation, do these animals become autoimmune? Lastly, what are the implications to the “licensing” or “disarming” of NK cells, if self-tolerance can be easily broken during inflammation? Determining the precise developmental and peripheral tolerance mechanisms at work in NK cells will aid in our understanding of how immune responses are regulated, and the potential consequences of inducing inflammation.