Natural killer (NK) cells have the potential to deliver both direct antimicrobial effects and regulate adaptive immune responses, but NK cell yields have been reported to vary greatly during different viral infections. Activating receptors, including the Ly49H molecule recognizing mouse cytomegalovirus (MCMV), can stimulate NK cell expansion. To define Ly49H's role in supporting NK cell proliferation and maintenance under conditions of uncontrolled viral infection, experiments were performed in Ly49h−/−, perforin 1 (Prf1)−/−, and wild-type (wt) B6 mice. NK cell numbers were similar in uninfected mice, but relative to responses in MCMV-infected wt mice, NK cell yields declined in the absence of Ly49h and increased in the absence of Prf1, with high rates of proliferation and Ly49H expression on nearly all cells. The expansion was abolished in mice deficient for both Ly49h and Prf1 (Ly49h−/−Prf1−/−), and negative consequences for survival were revealed. The Ly49H-dependent protection mechanism delivered in the absence of Prf1 was a result of interleukin 10 production, by the sustained NK cells, to regulate the magnitude of CD8 T cell responses. Thus, the studies demonstrate a previously unappreciated critical role for activating receptors in keeping NK cells present during viral infection to regulate adaptive immune responses.
Classical (non–T) NK cells are generally found at low frequencies in leukocyte populations (Biron et al., 1999). They have the potential to mediate antiviral and immunoregulatory functions through a variety of mechanisms (Orange et al., 1995; Su et al., 2001; Lee et al., 2007; Robbins et al., 2007; Strowig et al., 2008). By altering cell availability, in vivo conditions changing NK cell numbers may indirectly influence all of their effects. Activating receptors on NK cells are linked to stimulatory pathways overlapping with those used by TCRs to drive cell expansions (Murali-Krishna et al., 1998; Pitcher et al., 2003; French et al., 2006; MacFarlane and Campbell, 2006; Biron and Sen, 2007; Lee et al., 2007) and can induce NK cell proliferation (Dokun et al., 2001; French et al., 2006). Although particular activating receptors have been reported to recognize microbial products (Lanier, 1998; Vidal and Lanier, 2006; Jonjic et al., 2008), dramatic NK cell expansion has not been observed during infections. Except under rare experimental conditions (Caligiuri et al., 1991; Yamada et al., 1996; Fehniger et al., 2001; Huntington et al., 2007a; Sun et al., 2009), NK cell division is generally induced for limited periods of time as a consequence of transient innate cytokine exposure (Biron et al., 1984; Biron et al., 1999; Dokun et al., 2001; Nguyen et al., 2002; Yokoyama et al., 2004). Increasing proportions of NK cell subsets with activating receptors recognizing particular viral ligands can be detected during certain infections (Dokun et al., 2001; Gumá et al., 2006), but this is observed without dramatic increases in overall NK cell numbers, and many viral infections induce striking reductions in NK cell functions, frequencies, and yields (Biron et al., 1999; Tarazona et al., 2002; Lehoux et al., 2004; Reed et al., 2004; Azzoni et al., 2005; Vossen et al., 2005; Morishima et al., 2006). Thus, particular conditions of viral challenges must result in differential regulation of NK cell proportions and numbers, with consequences for the delivery of NK cell functions.
An NK cell activating receptor in the mouse is Ly49H (Lanier, 1998; Gosselin et al., 1999; Smith et al., 2000; Vidal and Lanier, 2006). This molecule recognizes a mouse CMV (MCMV) ligand (Arase et al., 2002; Smith et al., 2002), is expressed on NK cell subsets in strains of particular genetic backgrounds, including C57BL/6 (B6) mice, and is reported to be an exclusive marker for the classical NK cell subset (Smith et al., 2000). Through an associated molecule, Ly49H stimulates using signaling pathways overlapping with those used by the TCR (MacFarlane and Campbell, 2006; Biron and Sen, 2007). Additional markers for all NK cells include CD49b, expressed on other activated cell types (Arase et al., 2001); NKp46, selectively expressed on classical NK cells (Gazit et al., 2006; Walzer et al., 2007a; Walzer et al., 2007b); CD122, the IL-2Rβ chain, expressed on all NK cells and activated T cells (Huntington et al., 2007b); and NK1.1, exclusively expressed on C57BL6 (B6) NK and NKT cells (Lian and Kumar, 2002; MacDonald, 2002; Yokoyama et al., 2004; Huntington et al., 2007b). The TCR with associated CD3 molecules is not expressed on their cell surfaces (Biron et al., 1999). The mechanisms for NK cell–enhanced resistance to MCMV infection are incompletely characterized (Lee et al., 2007), but Ly49H contributes to their protective effects (Scalzo et al., 1990; Brown et al., 2001; Daniels et al., 2001; Lee et al., 2001; Lee et al., 2003). Engagement of the Ly49H receptor can lead to killing of target cells (Arase et al., 2002; Smith et al., 2002), and the correlation of increases in viral burdens resulting from the absence of Ly49H (Scalzo et al., 1990) as compared with those resulting from defects in cytotoxicity functions, such as mutation of the membrane pore-forming protein perforin 1 (Prf1; Tay and Welsh, 1997; Loh et al., 2005; van Dommelen et al., 2006), supports a role for Ly49H-dependent killing of virus-infected cells in the delivery of NK cell antiviral effects. The receptor may have other functions associated with its ability to stimulate proliferation, however, as the proportions of NK cells expressing Ly49H are increased during MCMV infection (Dokun et al., 2001).
The studies presented in this paper were undertaken to dissect the proliferative from the cytotoxic functions accessed through Ly49H, and to define the contribution of the regulation of NK cell numbers to protection during infection. To carry out the work, responses were evaluated, during MCMV infections, in wild-type (wt) B6 mice and mice deficient in Ly49h (Fodil-Cornu et al., 2008), Prf1 (Kägi et al., 1994), or both. As expected, single Ly49h−/− and Prf1−/− mice had profoundly increased viral burdens, but significant differences in NK cell expansion were discovered. The NK cell populations were decreasing in infected Ly49h−/− mice, whereas infected Prf1−/− mice had an unexpected dramatic proliferation of NK cells uniformly expressing Ly49H. The expansion was proven to be dependent on Ly49H. During uncontrolled infection in the absence of Prf1, Ly49H beneficially promoted effects for survival, because the sustained NK cells produced IL-10 to control the magnitude of the CD8 T cell response and limit immunopathology. The data suggest that Ly49H-dependent cytotoxicity acts to control viral infection and NK cell expansion, but that in the absence of the killing function, Ly49H promotes a continued NK cell expansion critical for supporting life over death because the NK cells are available to regulate adaptive immune responses.
Consequences of Ly49H− as compared with Prf1 deficiency for maintenance of NK cells
NK responses were investigated in wt, Ly49h-deficient (Fodil-Cornu et al., 2008), and Prf1-deficient B6 mice after infection with 2,500 PFU MCMV. As expected and in contrast to wt mice, high virus replication was consistently observed in the absence of Ly49H receptor (Fig. 1 A). Interestingly, Ly49h−/− mice had decreasing frequencies and numbers of NK cells, identified as either CD49b+TCRβ− (Fig. 1, B and C) or NK1.1+TCRβ− (Fig. 1, D and E), in spleens. The differences in Ly49h−/− as compared with wt were greater than twofold reductions in percentages and fivefold reductions in yields on day 5 of infection. The frequencies of cells expressing TCRβ were comparable in wt and Ly49h−/− B6 mice (Fig. 1 F), indicating a differential decrease of NK cells. Consistent with earlier studies (Tay and Welsh, 1997; Loh et al., 2005) and similar to Ly49h−/− mice (Fig. 1 A), Prf1−/− mice showed higher viral loads after infection (Fig. 1 G). Surprisingly, however, Prf1−/− mice contained higher frequencies, ∼25% or fivefold increases, of splenic CD49b+TCRβ− NK cells at day 5 as compared with wt mice (Fig. 1 H), and the absolute numbers of these NK cells were significantly higher (by approximately threefold) than those of wt mice (Fig. 1 I). The trends were the same for NK1.1+TCRβ− NK cells (Fig. 1, J and K). The proportions of cells expressing TCRβ were comparable in wt and Prf1−/− mice (Fig. 1 L), suggesting selective expansion of NK cells. Thus, the absence of Ly49H results in conditions failing to sustain NK cells but the absence of Prf1 results in profound NK cell expansion during MCMV infection. Because both Ly49h−/− and Prf1−/− mice had high viral burdens, the changes in NK cell frequencies and yields are not simply a consequence of secondary effects resulting from high viral replication.
The higher proportions and numbers of NK cells in Prf1−/− as compared with Ly49h−/− mice led us to examine expression of Ly49H on the CD49b cell subsets. After challenge with 2,500 PFU MCMV, ∼30–50% of splenic CD49b cells expressed Ly49H in uninfected wt and Prf1−/− mice as well as in day 1.5, 3, and 5 infected wt mice (Fig. 2 A). In contrast, the proportions of CD49b cells expressing Ly49H were increasing during infection of Prf1−/− mice such that >90% expressed Ly49H on day 5. To evaluate the effect of challenge dose, infections were initiated with 1,000, 2,000, or 5,000 PFU MCMV. The shift to NK cell subsets predominantly expressing Ly49H was observed under all conditions, but the proportions of NK cells increased with increasing challenge dose such that 36% of the splenic leukocytes were Ly49H+CD49b+ cells on day 5 after infection with 5,000 PFU (Fig. 2 B). Thus, the dramatic increases in NK cells expressing Ly49H are observed in response to different doses of MCMV, but higher doses lead to higher proportions of the cells.
The phenotype of the expanding NK cells was characterized further. Several development or activation status markers were compared for their expression on CD49b+TCRβ− NK cells from wt and Prf1−/− mice at day 4 of MCMV infection. Consistent with known NK cell lineage markers, NK cells from both mice showed high expression of CD122 (IL-2Rβ chain), and high expression of the NK cell activation markers CD69, CD11b, and CD11c (Fig. 2 C). Interestingly, markers thought to be specific for immature NK cells (Kim et al., 2002; Huntington et al., 2007b) were differentially expressed, with CD27 detected on both wt and Prf1−/− NK cell populations and CD51 (integrin αV) detected at high levels on NK cells from Prf1−/− mice. Investigation of other known lymphocyte activation markers, Ly6C and CD62L (L-selectin; Sinclair et al., 2008), demonstrated an unexpectedly higher expression of Ly6C and lower expression of CD62L on NK cells from Prf1−/− as compared with wt mice. Expression of another NK selective cell marker, NKp46, was also examined, and virtually all of the NKp46+ cells coexpressed Ly49H on day 4 in the Prf1−/− mice, but only half expressed Ly49H under the other conditions. The cells were double positive for CD49b and NKp46 under all conditions (Fig. 2 D). Mixed populations of NK cells expressing cytoplasmic CD3 have been reported, and certain activated T cells can down-regulate CD3 expression while maintaining expression of markers overlapping with NK cells (Lanier, 2007; Stewart et al., 2007). Because cytoplasmic staining of CD3ε can identify such T cell populations, experiments were performed to evaluate intracellular CD3ε expression in the highly expanding NK cells. Intracellular CD3ε expression was not significantly expressed in any of the Ly49H-positive cells isolated from wt or Prf1−/− mice on either day 0 or 4 (Fig. 2 E). Thus, the NK cells expanded in the Prf1−/− mice are bona fide NK cells but enriched in proportions expressing Ly49H.
Proliferation of Ly49H+ NK cells
To measure proliferation of NK cells during infections of wt, Ly49h−/−, and Prf1−/− mice, in vivo incorporation of the DNA analogue BrdU was analyzed. In wt mice, CD49b+TCRβ NK cell proliferation was increasing with a peak at day 3 of MCMV infection when 31% incorporated BrdU during a 2-h pulse in vivo (Fig. 3 A). In the case of Ly49h−/− mice, NK cell proliferation showed a severe defect at day 3. In sharp contrast to these two groups, greater and longer BrdU incorporation was observed in Prf1−/− mice during MCMV infection (Fig. 3 A). The peak was found at day 4, and at this time, the proportion of BrdU+ cells among the NK cells was 2.3-fold higher than that in wt mice. Because differential NK cell proportions (5% in B6, 2% in Ly49h−/−, and 15% in Prf1−/− mice) were found at day 4, the frequencies and numbers of proliferating NK cells among splenic leukocytes demonstrated even more dramatic differences. In Prf1−/− mice on day 4, >5- and 50-fold increases in the proportions of BrdU+ NK cells, respectively, were observed in comparison to wt and Ly49h−/− mice. The absolute numbers of proliferating NK cells in Prf1−/− mice were found to be increased similarly and showed the greatest difference at day 4 of MCMV infection. Hence, in comparison to wt NK cell responses, NK cells have reduced BrdU incorporation in the absence of Ly49H but increased incorporation in the absence of Prf1, and these changes are accompanied by respective differences in NK cell yields.
Because a severe defect in NK proliferation was observed in the absence of Ly49h, and nearly all NK cells in Prf1−/− mice were shown to be Ly49H expressing during MCMV infection, BrdU incorporation was specifically examined within the Ly49H-positive and -negative subsets (Fig. 3 B). As the peak of BrdU incorporation in Prf1−/− mice and the maximum difference in NK proliferation between wt and Prf1−/− mice were observed at day 4, this time point was chosen for the analysis. The Ly49H+ NK cells proliferated at a higher rate than did Ly49H− NK cells in wt mice, and these cells were absent in Ly49h−/− mice. Interestingly, at this time point in Prf1−/− mice, most NK cells were expressing Ly49H, and these NK cells had the highest level of BrdU incorporation. As analyzed in Fig. 3 B, the frequency of Ly49H− NK cells incorporating BrdU was similar among B6, Ly49h−/−, and Prf1−/− mice, whereas the frequency in Ly49H+ NK cells was significantly higher in Prf1−/− mice. Therefore, preferential proliferation of Ly49H+ NK cells accompanies the dramatic expansion of NK cell numbers in Prf1−/− mice.
Although the studies clearly demonstrated heightened proliferation of Ly49H+ cells, particularly in Prf1−/− mice, they did not evaluate a potential contribution of cell-intrinsic differences to division. To determine the intrinsic proliferative potential of Ly49H+ as compared with Ly49H− cells from wt, Prf1−/−, and Ly49h−/− mice, negatively enriched NK cells (40–50%) were stimulated with a growth factor capable of activating all cells, IL-2, and with control cells (BaF3) or cells expressing the Ly49H ligand m157 (BaF3/m157) ex vivo. The BrdU incorporation was evaluated after a 2-h pulse at the end of stimulation. After IL-2 treatment, BrdU incorporation in NK1.1+TCRβ− NK cells from wt, Ly49h−/−, and Prf1−/− mice was comparable and irrespective of Ly49H expression (Fig. S1 A). In contrast, stimulation with either γ-irradiated BaF3 or BaF3/m157 cells demonstrated that Ly49H+ NK cells from both wt and Prf1−/− mice specifically incorporated BrdU in response to the ligand and did so at comparable rates but that Ly49H− cells did not (Fig. S1 B). The populations from Ly49h−/− mice had extremely poor responses under these conditions. The results indicated that the intrinsic proliferative responsiveness of NK cells from wt, Ly49h−/−, and Prf1−/− mice are equivalent but that the expression of the Ly49H selects for subset expansion in either wt or Prf1−/− populations.
Requirement for the Ly49H receptor in sustaining NK cell proliferation and survival, and consequences for health
To directly test the hypothesis that the enhanced NK cell expansion in Prf1−/− mice is dependent on Ly49H, mice deficient in both Prf1 and Ly49h (hereafter Ly49h−/−Prf1−/−) were bred and investigated for NK cell proliferation at day 4 of MCMV infection. As compared with wt mice, the Ly49h−/−Prf1−/− and Prf1−/− mice were both highly susceptible to MCMV infection, and showed similar levels of uncontrolled virus replication (Fig. 4 A). The ablation of the Ly49h gene in Prf1−/− mice, however, resulted in dramatic reductions of both NK cell frequencies and numbers at day 4 (Fig. 4, B and C), with the absolute number of NK cells being lower by greater than two- and fivefold, respectively, as compared with wt and Prf1−/− mice (Fig. 4 C). The frequencies of cells expressing TCRβ were comparable among the groups (Fig. 4 D). To examine the consequences for proliferation in vivo, experiments were performed with 2-h BrdU pulses. The NK cells were identified as CD49b+TCRβ− (Fig. 4 E) and NKp46+TCRβ− (Fig. 4 F) populations. A severe reduction in the percentages of BrdU+ NK cells was apparent in the absence of Ly49H. When examined as total numbers of NK cells incorporating, a significant reduction (up to greater than eightfold) was observed in Ly49h−/−Prf1−/− as compared with Prf1−/− mice at day 4 of MCMV infection (Fig. 4, G and H). Thus, the Ly49H NK receptor is required for NK cell proliferation and maintenance of NK cells during MCMV infections in Prf1−/− mice.
To evaluate the consequences of the Ly49H-dependent NK cell expansion for health, body weight changes and survival were evaluated for extended periods of time after infections in wt, Prf1−/−, and Ly49h−/−Prf1−/− mice. At a challenge dose of 2,500 PFU of virus, the wt mice started gaining weight at day 4 after infection. In contrast, both the Prf1−/− and the Ly49h−/−Prf1−/− mice were losing weight by day 4 of infection. The Prf1−/− mice had only ∼6% weight loss at maximum. This was sustained through intermediate times, and the mice were returning to their preinfection weights at day 12 after infection. However, the Ly49h−/−Prf1−/− mice continued losing weight after day 4 and reached weight losses of ∼15% at intermediate times (day 10) before starting to then gain weight on day 12 (Fig. 5 A). After infection with 5,000 PFU MCMV, the wt and Prf1−/− mice survived, but Ly49h−/−Prf1−/− mice were beginning to die at day 6 and 80% had succumbed by day 10 of infection (Fig. 5 A). Thus, the Ly49h−/−Prf1−/− mice are significantly more sensitive to disease during MCMV infection, and the presence of Ly49H makes the difference between life and death.
Because the separation of body weight loss and survival after day 4 of infection was observed in the Ly49h−/−Prf1−/− as compared with wt and Prf1−/− mice, viral titers were measured. As expected, wt mice were clearing a challenge dose of 2,500 PFU MCMV by day 4 and 5 after infection. In contrast, both the Prf1−/− and the Ly49h−/−Prf1−/− mice had substantially higher (>4 log) increases in viral titers at these times, but the groups did not differ significantly (Fig. 5 B). Thus, differences in disease susceptibility between the Prf1−/− and the Ly49h−/−Prf1−/− mice could not be explained by altered sensitivity to infection. Because pathology during viral infections can also be a result of the magnitude of immune responses, circulating levels of cytokines were measured on days 4 and 5 of infection. wt mice had extremely low levels of IFN-γ and TNF-α, whereas both the Prf1−/− and the Ly49h−/−Prf1−/− mice had elevated levels of these cytokines on days 4 and 5 of infection, with no statistically significant differences in TNF-α but higher IFN-γ levels on day 5 in the Ly49h−/−Prf1−/− mice (Fig. 5 B). The differences in disease parameters in these two groups of mice were, however, also accompanied by significant differences in circulating levels of IL-10, a cytokine known to inhibit T cell responses. Serum IL-10 levels were undetectable in wt mice. High circulating levels of the cytokine were detectable in the Prf1−/− mice; on day 5 of infection, it reached a level of >3,600 pg/ml in individual mice. However, Ly49h−/−Prf1−/− mice only had IL-10 levels reaching 984 pg/ml (Fig. 5 B). Therefore, under conditions of viral challenge in Prf1−/− mice, there is high Ly49H-dependent NK cell expansion and better protection against disease, whereas Prf1−/− mice also deficient for Ly49H lose their NK cells, have poor health, and succumb to death. High endogenous IL-10 production is present in the first but not the second condition.
IL-10 production by NK cells
To determine whether or not the expanded NK cells were producing IL-10, NK cells were purified and their ex vivo production of the factor was measured. For these studies, wt and Prf1−/− mice were infected for 4 d. In the first set of experiments (Fig. 6 A), total, NK and non–NK cells were prepared by positive selection using CD49b+ magnetic bead isolation to compare responses in wt and Prf1−/− mice. As reported in Fig. 5 B, the Prf1−/− but not the wt mice had high circulating levels of IFN-γ and IL-10 on day 4. Ex vivo production of IFN-γ but not IL-10 was detectable with NK but not non–NK cells enriched from the wt mice. High production of both of these cytokines, however, was enriched in the NK but not the non–NK cells from the Prf1−/− mice. As a second approach, highly purified NK cells were prepared from the Prf1−/− mice by FACS sorting of the CD49b+Ly49H+ cells on day 4 of infection (Fig. 6 B). Again, ex vivo production of IFN-γ and IL-10 was greatly enriched in the NK but not the non–NK cell populations. The total and sorted populations from day 4 MCMV-infected Prf1−/− mice were also examined for IL-10 expression at the mRNA level using RT-PCR, and the signaling for IL-10 was increased in the NK but not the non–NK cells (Fig. 6 C). Thus, high IL-10 is being produced in the Prf1−/− but not the Ly49h−/−Prf1−/− mice at intermediate times after infection, and the greatly expanded NK cells are the source of the factor.
The cytokine IL-21 has been reported to induce IL-10 mRNA expression within NK cells (Brady et al., 2004). It was not possible to detect IL-21 protein production in the serum (unpublished data), but the mRNA was detected using samples from the populations isolated on day 4 after infection of the Prf1−/− mice. The signal was increased in the samples prepared from non–NK rather than NK cells (Fig. 6 C). Many culture experiments evaluating NK cell responses are enhanced by the addition of IL-15, and this cytokine was being expressed at the mRNA level in samples prepared from non–NK as compared with NK cells (Fig. 6 C) on day 4 after infection of the Prf1−/− mice. To evaluate the pathways stimulating IL-10 on day 4 in the Prf1−/− mice, ex vivo production by the total, NK, and non–NK cells was evaluated after stimulation with antibodies against Ly49H or with cytokines. Treatment with plate-bound anti-Ly49H did induce increases in IL-10 production by the NK cells, with levels increased approximately from 230 to 307 pg/106 cells (Fig. 6 D). This was an extremely modest response, however, when compared with cytokine stimulation. In this case, IL-21 treatment alone increased the levels to ∼3,451 pg/106 cells, and IL-21 with IL-15 treatment increased the levels to ∼8,582 pg/106 cells (Fig. 6 D). Thus, NK cells activated during MCMV infection of Prf1−/− mice produce IL-10, and IL-21 is a potent inducer of elevated IL-10 production by these cells.
Role for Ly49H and IL-10 in regulating CD8 T cell responses
To examine the T cell responses in the wt, Prf1−/−, and Ly49h−/−Prf1−/− mice, mice were infected with MCMV for 7 d. Although the proportions of NK cells (CD49b+TCRβ−) in Prf1−/− mice were increased and those in Ly49h−/−Prf1−/− mice were severely reduced, the frequencies of CD8α+TCRβ+ cells reached 25–30% in all three groups of mice (Fig. 7 A). The activation states, however, were higher in the Prf1−/− mice and highest in the Ly49h−/−Prf1−/− mice. The proportions of CD8 T cells from these mice expressing IFN-γ were <1, 6, and 22%, respectively, and those expressing granzyme B were <4, 47, and 71%, respectively. When the proportions and yields of CD8α+IFN-γ+ cells in splenic populations were calculated, the order and differences between the groups were similar. In terms of overall yields of CD8 T cells expressing IFN-γ, however, the Ly49h−/−Prf1−/− mice had >45-fold increases as compared with wt mice and >4.2-fold increases as compared with Prf1−/− mice (Fig. 7 B). The MCMV titers were higher in the Ly49h−/−Prf1−/− as compared with the Prf1−/− mice, but the difference was not significant (Fig. 7 C). In contrast, the circulating levels of IFN-γ, TNF-α, and IL-10 were all substantially and significantly increased in the Ly49h−/−Prf1−/− mice, with values at least 3.4-fold higher. Thus, the Ly49h−/−Prf1−/− mice lacking NK cell expansion and dramatically reduced NK cell production of IL-10 at intermediate times after infection have profoundly increased CD8 T cell responses to MCMV infection with only modest increases in viral burdens.
IL-10 has a wide range of immunoregulatory functions including controlling T cell responses (Maynard and Weaver, 2008). To examine its role in the regulation of CD8 T cell responses in the Prf1−/− mice, the consequences of neutralizing the cytokine were evaluated. Mice were given control antibody or antibody-neutralizing IL-10. The treatment protocol used was developed to deliver maximum effects while allowing reappearance of day 7 circulating levels of IL-10 equivalent to those in the Prf1−/− mice, i.e., concentrations of 200–300 mg/ml (Fig. 7 C, bottom right). It required antibody delivery on days 2 and 4 of infection (see Materials and methods). Both control and anti–IL-10 antibody groups of mice still had increasing proportions of NK cells (CD49b+TCRβ−; Fig. 8 A). They were very different, however, in their CD8 T cell responses. Neutralization of IL-10 resulted in elevated endogenous CD8 T cell responses (Fig. 8, A and B). In this case, the viral burdens were increased by ∼1.5 logs, and this difference was significant (Fig. 8 C). The circulating levels of IFN-γ, TNF-α, and IL-10 were, however, increased by 12.6-, 10-, and 5.5-fold, respectively (Fig. 8 C). Overall, the differences in the parameters examined after IL-10 neutralization during MCMV infection of Prf1−/− mice paralleled those observed in the Ly49h−/−Prf1−/− mice. Therefore, early IL-10 production by the Ly49H-sustained NK cells regulates CD8 T cell responses.
To examine IL-10's role in controlling detrimental effects during infections in the Prf1−/− mice, the effects of IL-10 neutralization were evaluated under conditions of infection with 5,000 PFU MCMV. The first experiments examined the consequences of blocking IL-10 with antibodies against the IL-10 receptor (anti–IL-10R). Although all wt mice treated with anti–IL-10R survived the infection, Prf1−/− mice treated with anti–IL-10R all died by day 11 of infection (unpublished data). As an independent approach, Prf1−/− mice were treated with control antibody or antibody against the cytokine itself (αIL-10) on days 2 and 4 of infection, as described in the previous paragraph. Administration of αIL-10 resulted in death of >80% of the Prf1−/− mice (Fig. 8 D). Thus, blocking early IL-10 effects in Prf1−/− results in a increased sensitivity to infection and a phenotype comparable to Ly49h−/−Prf1−/− mice.
Role for CD8 T cell responses in late cytokine responses and death during infections of Ly49h−/−Prf1−/− mice
To conclusively demonstrate the contribution of the CD8 T cell response to cytokine production and death after MCMV challenge in the Ly49h−/−Prf1−/− mice, mice were control treated or treated with antibodies against CD8 (αCD8). After challenge with 2,500 PFU of virus, blocking the CD8 T cell response had no effect on the day 7 viral titers in the Ly49h−/−Prf1−/− mice (Fig. 9 A) but did significantly inhibit the elevated circulating levels of cytokines (Fig. 9 A). Moreover, the αCD8α treatment protected against the lethal consequences of infection with 5,000 PFU MCMV (Fig. 9 B). Thus, the presence of the CD8 T cell responses was not effective at controlling the viral infection but did result in elevated cytokine responses and death. Collectively, the studies demonstrate that expanded NK cells and early IL-10 production is regulating the magnitude of endogenous CD8 T cell responses and protecting from a CD8 T cell–dependent death under these conditions of sustained viral infections.
These studies demonstrate an unexpected role for an activating receptor in sustaining NK cells and, as a result, protecting against detrimental consequences during viral infection. In particular, they define the Ly49H receptor's contribution to the proliferation and maintenance of NK cells during MCMV infections. In wt B6 mice, viral replication was controlled. NK cell proliferation peaked at day 3 after challenge, with the proportions of Ly49H+ populations increasing without profound cell expansion. Deficiencies in Ly49h, Prf1, or both all resulted in significant increases in MCMV burdens, but the consequences for NK cell expansion were profoundly different; NK cell proportions, numbers, and division were decreased in the absence of Ly49h but dramatically increased in the absence of Prf1. The Ly49H receptor was required for the responses in infected Prf1−/− mice because (a) nearly all NK cells expressed Ly49H, (b) the Ly49H+ cells were preferentially proliferating, and (c) ablation of the Ly49h gene in addition to the Prf1 gene (Ly49h−/−Prf1−/−) abolished NK cell proliferation. The Ly49H-dependent effects were important because the absence of Ly49h in the Prf1−/− mice resulted in increased sensitivity to disease and death. The expanded Ly49H+ NK cells in Prf1−/− mice were making IL-10, and CD8 T cell responses were being regulated by the NK cells and IL-10; CD8 T cell responses were dramatically elevated in Ly49h−/−Prf1−/− and Prf1−/− mice having had IL-10 function neutralized. Elevated CD8 T cell responses in the Ly49h−/−Prf1−/− mice were responsible for death after MCMV challenge. Thus, the Ly49H activating receptor contributes to the proliferation and maintenance of NK cells, the NK cells are induced to produce IL-10 under these conditions, and the NK cells and IL-10 are important for limiting the magnitude of CD8 T cell responses to make the difference between life and death (Fig. 10).
The results have a broad significance to our understanding of the regulation and function of NK cells during viral infections. Before the clear demonstration of a role for NK cells in defense against herpes group viruses, there were numerous reports of defects in NK cell functions after a wide range of viral infections (Biron et al., 1999; Lee et al., 2007). More recently, decreases in NK cell numbers and maturation states have been reported during infections of humans with HIV (Tarazona et al., 2002; Azzoni et al., 2005), hepatitis C virus (HCV; Morishima et al., 2006), and varicella zoster virus (Vossen et al., 2005); of monkeys with Ebola virus (Reed et al., 2004); and of mice with mouse hepatitis virus 3 (Lehoux et al., 2004). It is important to note, however, that there are genetic differences between individuals and strains, and that there are also reports of increases in proportions of NK cells expressing particular activating receptors during viral infections (Dokun et al., 2001; Gumá et al., 2006; De Maria et al., 2007; Alter et al., 2009). In our study, an activating receptor is shown to play a critical role in sustaining NK cells. In fact, the extreme expansion of these “MCMV-specific” NK cells in the absence of cytotoxic function is demonstrated. Analogues of Ly49H recognition of virus-infected cells have been proposed to exist for viruses in humans. One good candidate identified during human CMV (HCMV) infections is the NKG2C/CD94-activating NK receptor. Significantly increased proportions of NK cells expressing NKG2C are observed in HCMV seropositive individuals compared with HCMV seronegative individuals (Gumá et al., 2006). Increases in the proportions of NK cells expressing the NKp30 activating receptor have been observed during chronic HCV infection (De Maria et al., 2007), and in the proportions of NK cells expressing the KIR3DS1 activating receptor have been observed during acute HIV infection (Alter et al., 2009). Thus, studies in both mice and men indicate that activating receptors may be helping to select NK cell subsets during infections and that as a consequence, the frequencies of NK cells expressing particular activating receptors are changing. Our study also demonstrates, however, that activating receptors are required for sustaining NK cells because the absence of Ly49H results in decreasing NK cell yields during MCMV infections. Collectively, the studies show that activating receptors promote the expansion of particular NK cell subsets and that without such a pathway, NK cells are not sustained during prolonged infection.
The work adds to the growing literature characterizing conditions supporting NK cell expansion (Biron et al., 1984; Caligiuri et al., 1991; Yamada et al., 1996; Lanier, 1998; Biron et al., 1999; Dokun et al., 2001; Fehniger et al., 2001; Nguyen et al., 2002; Yokoyama et al., 2004; French et al., 2006; Huntington et al., 2007a; Sun et al., 2009). In this case, the importance for the expansion is clearly demonstrated. Keeping the NK cells around is likely to promote defense and immunoregulatory mechanisms mediated through the activating receptor supporting expansion, and allow access of those induced through stimulation of other activating receptors, cytokine receptors, and/or engagement of the TNF-related apoptosis-inducing ligand expressed on activated NK cells (Lee et al., 2007). The latter three mechanisms have the potential to promote effects even if stimulation of the original activating receptor through its ligand is blocked. The NK cells expanded in the Prf1−/− mice during MCMV infection are shown in this paper to be producing IL-10. Although stimulation through Ly49H itself could modestly (<1.5-fold) enhance IL-10 production, mRNA expression of IL-21 is also apparent, and addition of this cytokine causes dramatic (>15-fold) increases in IL-10 production by the NK cells. Thus, the system demonstrates a mechanism using the activating NK cell receptor to promote the continued presence of the cells for activation by other cytokines. Keeping the cells around is likely to enhance resistance and health under a wide range of viral infections, and learning how to manage this during infections failing to directly stimulate an activating receptor to support NK cell expansion will have important therapeutic applications.
The cytokine IL-10 was first shown to be a product of T cells, but the list of cellular sources is growing (Brooks et al., 2006; Ejrnaes et al., 2006; Maris et al., 2007; O'Garra and Vieira, 2007). Before our study, the characterization of NK cell IL-10 production had been largely limited to expression at the mRNA level during Toxoplasma infection of mice (Maroof et al., 2008), after treatment of mouse cells with IL-21 (Brady et al., 2004), or in cells prepared from HIV-infected individuals (Brockman et al., 2009). There are recent reports, however, that human regulatory NK cells can be mitogen stimulated to produce IL-10 (Deniz et al., 2008) and that NK cells recovered during chronic HCV infections produce IL-10 (De Maria et al., 2007). Although the factor can also enhance inflammatory responses under particular conditions (Herrero et al., 2003; Sharif et al., 2004), IL-10 has largely been appreciated for its direct and indirect inhibitory effects on several T cell responses (Deniz et al., 2008; Maynard and Weaver, 2008). Therefore, the production of IL-10 during infections has been thought to present a situation for interfering with viral defense (Brooks et al., 2006; Ejrnaes et al., 2006; De Maria et al., 2007; Maris et al., 2007). However, IL-10 has also been shown to contribute to regulation of immunopathology (Gazzinelli et al., 1996), and the expansion of NK cells and their early production of IL-10 reported in this paper is demonstrated to be important for limiting CD8 T cell response and CD8 T cell–dependent disease. Thus, an NK cell contribution to regulating adaptive immunity for protection against immune-mediated disease is demonstrated for the first time. The pathway may be in place at some level during infections of immunocompetent mice because our laboratory has previously shown NK cell–dependent regulation of T cell responses to MCMV in wt mice (Su et al., 2001), and challenges with extremely high doses of the virus have been reported to induce low levels of IL-10 and IL-10–dependent protection against disease (Oakley et al., 2008).
The work also conclusively separates, for the first time, the beneficial functions of Ly49H-dependent expansion from Ly49H-dependent cytotoxicity. The activating receptor has been reported to induce NK cell killing and proliferation in culture, but only delivery of cytotoxic function has been thought to be important in protection against MCMV (Scalzo et al., 1990; Tay and Welsh, 1997; Loh et al., 2005; van Dommelen et al., 2006). Although other factors such as cytokines may contribute (Nguyen et al., 2002; French et al., 2006), the requirement for Ly49H in supporting NK cell proliferation and health are definitively established in the absence of cytotoxic function. The results suggest a model whereby NK cells are sustained during periods of intense receptor stimulation but delivery of cytotoxic function to regulate viral replication acts to limit expansion by eliminating virus-infected cells. In the absence of cytotoxicity or in the presence of overwhelming viral replication failing to be limited by cytoxicity, however, NK cell expansion is increased through an Ly49H-dependent pathway to promote alternative beneficial effects. This identification of the importance of the activating receptor in the context of infection is distinct from another report investigating the role of these molecules in NK cell maturation and tolerance (Sun and Lanier, 2008; Tripathy et al., 2008). In this study, the activating receptor is shown to be critical for maintaining NK cells during acute infectious challenge.
Given these results, it is interesting to consider CD8 T cell expansion during infections with lymphocytic choriomeningitis virus (LCMV), a virus inducing dramatic expansion of CD8 T cells in Prf1−/− mice (Matloubian et al., 1999). Collectively, the responses to MCMV and LCMV indicate that linking stimulation for killing with stimulation for proliferation may be important for regulating proportions of both NK and CD8 T cells. It is remarkable, however, that in contrast to the known MCMV expression of a ligand for the NK cell activating receptor Ly49H, there is no evidence for NK cell activating receptor recognition of LCMV. Thus, LCMV infections of Prf1−/− mice may present a condition where NK cells are not expanded for maintenance, and as a result, they are not available to critically contribute to the regulation of the CD8 T cell response. This model would explain the dramatic magnitude of CD8 T cell responses during LCMV infections of both wt and Prf1−/− mice (Matloubian et al., 1999), as well as the apparent major contribution of the CD8 T cell responses to hemophagocytic lymphohistocytosis diseases under conditions of LCMV (Jordan et al., 2004) but not MCMV (Crozat et al., 2007) infections in systems with blocked cytotoxic functions. The model also suggests that the balance of NK and CD8 T cell responses might be controlled in a wide range of infection by the presence or absence of a ligand for an NK cell activating receptor.
In summary, this paper discovers a novel and important new pathway for sustaining NK cells during infection, proves its critical role in protection against disease, and defines the mechanism for the beneficial effects dependent on this pathway as NK cell regulation of adaptive immunity.
MATERIALS AND METHODS
Specific pathogen-free C57BL/6 (C57BL/6NTac) mice were purchased from Taconic. Ly49h-deficient (Ly49h−/−) mice with a B6 background were generated by a marker-assisted congenic approach (Fodil-Cornu et al., 2008). Perforin-deficient (Prf1−/−) mice (Kagi et al., 1994) were originally obtained from the Jackson Laboratory to establish breeding colonies at Brown University. Mice deficient in both Ly49h and Prf1 (Ly49h−/−Prf1−/−) were generated by crossing Ly49h−/− and Prf1−/− mice. All mouse colonies were kept and maintained in strict isolation under specific pathogen-free conditions. Mice used in experiments were 6–14 wk old. Animals obtained from sources outside of Brown University were housed in the animal care facility for at least 1 wk before use. Handling of mice and experimental procedures were conducted in accordance with institutional guidelines for animal care and use, and protocols were approved by the Brown University Institutional Animal Care and Use Committee.
In vivo manipulation.
Unless otherwise indicated in the figures, infections were initiated on day 0 by i.p. injection of 2,500 PFU of salivary gland–derived MCMV prepared, as previously described (Orange et al., 1995; Dalod et al., 2002). Experiments were repeated with independently derived Smith strain MCMV from American Type Culture Collection, as previously described (Depatie et al., 1997). The infectious virus in tissue homogenates was quantified by plaque assay using mouse embryonic fibroblasts (Orange et al., 1995). To evaluate in vivo DNA synthesis, mice were injected with 2 mg BrdU (BD) on the days indicated in the figures and sacrificed 2 h later. For CD8 T cell blocking, 1 mg anti-CD8α antibody (clone 53.6.72; Bio-Express) was administered by i.p. injection at day 5. For controls, an equal amount of rat IgG (Sigma-Aldrich) was given. For blocking of the IL-10 receptor, 500 µg of blocking anti–IL-10R antibody (clone 1B1.3A; Bio-Express) was administered by i.p. injection at day 3. For neutralization of IL-10, 600 µg anti–IL-10 antibody (clone JES5-2A5; Bio-Express) was administered by i.p. injection at days 2 and 4. An equal amount of isotype control rat IgG1 mAb (clone HRPN; Bio-Express) was used as a control for anti–IL-10R antibody and anti–IL-10 antibody. The anti–IL-10 protocol was developed by modifying treatment conditions to maximize effects, ensure neutralization of the IL-10 produced at intermediate times after infection, and allow circulating levels of the cytokine on day 7. For the survival assay, mice were infected with 5,000 PFU MCMV.
Preparation of leukocyte populations.
For preparation of splenic leukocytes, spleens were ground between the rough surfaces of glass slides and treated with Red Blood Cell Lysing Buffer (Sigma-Aldrich) to remove red blood cells. Viable cell yields were determined by Trypan blue exclusion. The frequencies of NK and CD8 T cells among leukocytes were determined by flow cytometric analysis, and the absolute numbers were calculated by multiplying the frequencies by the total number of leukocytes.
Flow cytometric and immunofluorescence analyses.
Single splenic leukocytes were prepared as described in the previous section and incubated for 20 min with 2.4G2 antibody to reduce nonspecific staining. Cell-surface staining was performed as previously described (Dalod et al., 2002), using antibodies directed against the following: CD49b-FITC, CD49b-PE, TCRβ-PE, TCRβ-allophycocyanin (APC), Ly6C-FITC, CD122-PE, CD69-PE, CD11c-APC, CD27-PE, CD51-PE, CD62L-APC, CD11b-PE, NK1.1-PE, NKp46-PE, and CD8α-PE. Antibodies were purchased from BD and eBioscience. Polyclonal anti–NKp46-FITC antibody was purchased from R&D Systems. The specific monoclonal antibody, directed against Ly49H (a gift from W. Yokoyama, Washington University School of Medicine, St. Louis, MO; Smith et al., 2000), was biotinylated and used with streptavidin fluorochromes (PE or APC) to identify Ly49H-expressing cells. For the detection of incorporated BrdU, cells were first stained for surface antigens and then fixed, permeabilized, treated with DNase I, and stained with FITC–anti-BrdU antibody (clone 3D4; BD) according to the manufacturer's protocol. For intracellular staining of IFN-γ and granzyme B, the cells were incubated with 5 µg/ml brefeldin A for 6 h and stained for surface markers. The cells were fixed and permeablized (Fix/Perm Buffer; BD) and labeled with either IFN-γ–APC or granzyme B–APC. Samples were acquired using a FACSCalibur with the CellQuest Pro software package (both from BD). Laser outputs were 15 mW at 488- and 635-nm wavelengths.
Cytokine production analysis.
Cytokine production of IFN-γ, TNF-α, and IL-10 in serum or conditioned media was measured using cytometric bead assays (mouse inflammation kit; BD). Cytokine production was presented as pg/ml for serum and pg/106 cells for conditioned media.
NK cell isolation.
For isolation of NK cells expressing CD49b, splenic leukocytes from wt and Prf1−/− mice at day 4 MCMV infection were processed using anti-CD49b magnetic beads and positive selection with the Possel_s program on the AutoMACS according to the manufacturer's instructions (Miltenyi Biotec). The enriched populations contained >80% CD49b+ cells and >85% CD49b− cells. For FACS sorting of the NK cell subsets expressing CD49b and Ly49H and the non–NK cell subset not expressing CD49b and Ly49H, splenic leukocytes from Prf1−/− mice at day 4 of MCMV infection were stained with PE-CD49b and APC-Ly49H. Cell sorting was performed on a cell sorter (FACSAria; BD) at the Flow Cytometer Facility at Brown University. The purity of the subsets was >98%.
Ex vivo NK cell manipulations.
To evaluate IL-10 and IFN-γ production from isolated subsets, 105 cells were incubated for 24 h at 106 cells/ml in complete RPMI 1640 medium. For further stimulation, 105 highly purified cells were incubated either on control antibody (mouse IgG1)– and anti-Ly49H antibody–coated plates or with 100 ng/ml of mouse recombinant IL-21 (>97% purity; R&D Systems) in the presence or absence of 50 ng/ml IL-15 (>95% purity; R&D Systems).
Total RNA was isolated from total splenic leukocytes at days 0 and 4, and highly purified subsets at day 4 (>98%) using RNeasy (QIAGEN). 1 µg RNA was reverse-transcribed as described previously (Dalod et al., 2002). One tenth of the product from the RT reaction was used as a template for PCR amplification using primers specific for IL-10, IL-21, and IL-15 synthesized by Operon in a programmable thermal cycler (PTC-200; MJ Research). Sequences of oligonucleotides are as follows: IL-10 (forward, 5′-TGCTATGCTGCCTGCTCTTACTGA-3′; reverse, 5′-CCTGCTCCACTGCCTTGCTCTTAT-3′; Brady et al., 2004), IL-21 (forward, 5′-CCCTTGTCTGTCTGGTAGTCATC-3′; reverse, 5′-ATCACAGGAAGGGCATTTAGC-3′; Brandt et al., 2003), and IL-15 (forward, 5′-CATATGGAATCCAACTGGATAGATGTAAGATA-3′; reverse, 5′-CATATGCTCGAGGGACGTGTTGATGAACAT-3′; Zhang et al., 1998). Gapdh (forward, 5′-ACCACAGTCCATGCCATCAC-3′; reverse, 5′-TCCACCACCCTGTTGCTGTA-3′) was used as an internal control. The data shown in the figures were obtained before the amplification products reached saturation.
Evaluation of the body weight and survival.
Mice were weighed and monitored for death once a day at the same time on every day during infection from day 0 (before infection) to day 16. For clarity, body weight changes at 2-d intervals are shown in the figures.
Statistical analyses were performed using an unpaired Student's two-tailed t test. Means ± SD are shown. The Kaplan-Meier log-rank test was used to compare survival between groups of mice using Prism 5 software (GraphPad Software, Inc.).
Ex vivo NK cell proliferation assay.
For evaluation of intrinsic proliferative activity of NK cells, NK cells were enriched (30–50% of NK1.1+TCRβ− cells) from splenocytes of wt, Ly49h−/−, and Prf1−/− mice using negative selection with magnetic beads (Miltenyi Biotec). 5 × 104 enriched NK cells were stimulated on 96-well plates with either 150 ng/ml IL-2 or γ-irradiated Ba/F3 cells and Ba/F3 cells expressing m157 (effector/target ratio = 2:1; a gift from L. Lanier, University of California, San Francisco, San Francisco, CA). At 40 h after stimulation, cells were incubated with BrdU at a 20-µM concentration for an additional 2 h and were analyzed for BrdU incorporation.
Online supplemental material.
Fig. S1 shows intrinsic NK cell proliferative potentials.
The authors thank W. Yokoyama for providing anti-Ly49H antibody, L. Lanier for BaF3 cells expressing m157, and H. Cun and D.A. Feldman for technical assistance and reading the manuscript.
S.-H. Lee is a recipient of a Canadian Institutes of Health Research (CIHR) fellowship. This work was supported by National Institutes of Health grant CA041268 and CIHR grant MOP-7781.
The authors have no financial conflict of interest.