Cutting Edge: Stage-Specific Requirement of IL-18 for Antiviral NK Cell Expansion

Natural killer cells play a significant role in the control of infected, stressed, or transformed cells that may be detrimental to the host. Recent studies in mice and humans have demonstrated that NK cells possess adaptive immune qualities (1). In mice infected with mouse CMV (MCMV), Ly49H⁺ NK cells activated by the viral glycoprotein m157 undergo extensive proliferation and contract, resulting in the formation of a small pool of long-lived memory NK cells that can be recalled, and exhibit heightened effector function (1).

Proinflammatory cytokines strongly influence the NK cell response against MCMV infection (2). Although previous work has described the effect of proinflammatory cytokines on the general activation of NK cells during MCMV infection (2), their role in driving clonal-like expansion and memory in Ag-specific NK cells is largely unknown. We previously implicated IL-12, its signaling molecule STAT4, and the downstream transcription factor Zbtb32 as crucial signals in the generation of robust effector and memory NK cell responses against MCMV infection (3, 4). IL-18 has been suggested to “prime” resting NK cells for maximum IFN-γ production after ex vivo stimulation (5), and synergize with IL-12 during NK cell activation (6). Although IL-18 is produced early during MCMV infection (7), it is not known how IL-18 signals influence the virus-specific Ly49H⁺ NK cell response. In this study, we investigate the direct effects of IL-18 signaling on primary and recall NK cell responses to MCMV infection.

All mice used in this study were bred and maintained at Memorial Sloan-Kettering Cancer Center in accordance with Institutional Animal Care and Use Committee guidelines. Mixed bone marrow chimeric mice were generated, and adoptive transfer studies and viral infections were performed as previously described (8).

FcRs were blocked with 2.4G2 mAb before staining with the indicated surface or intracellular Abs (BD, Biolegend, or eBioscience). Flow cytometry was performed on an LSR II (BD). Cell sorting was performed on an Aria II cytometer (BD). All data were analyzed with FlowJo software (Tree Star). NK cell enrichment and adoptive transfers were performed as previously described (3).

Quantitative RT-PCR (qRT-PCR) and chromatin immunoprecipitation (ChIP) were performed as previously described (4). The following qRT-PCR primers were used: Myd88, Forward [For]: 5′-CACCTGTGTCTGGTCCATT-3′, Reverse [Rev]: 5′-AGGCTGAGTGCAAACTTG-3′; Actb, For: 5′-TGCGTGACATCAAAGAGAAG-3′, Rev: 5′-CGGATGTCAACGTCACACTT-3′. The following quantitative PCR primers were used for ChIP studies: Myd88 promoter, For: 5′-AAGTAGGAAACTCCACAGGCGAGC-3′, Rev: 5′-TTCAAGAACAGCGATAGGCGGC-3′; Gene desert 50 kB upstream of Foxp3, For: 5′-TAGCCAGAAGCTGGAAAGAAGCCA-3′, Rev: 5′-TGATACCCTCCAGGTCCAACCATT-3′; Zpf42 promoter, For: 5′-AGAGGGCGGTGTGTACTGTGGTG-3′, Rev: 5′-CTTCTTCTTGCACCCGGCTTGAG-3′; Utf1 promoter, For: 5′-AGTCGTTGAATACCGCGTTGCTG-3′, Rev: 5′-CTGTTGAGATGTCGCCCAAGTGC-3′; Ifng promoter, For: 5′-GCTCTGTGGATGAGAAAT-3′, Rev: 5′-GCTCTGTGGATGAGAAAT-3′.

Purified NK cells were stimulated for 4 (memory cells) or 18 h (for ChIP), as previously described (4). Negative and positive controls include NK cells incubated with media only, or with PMA (50 ng/ml) and ionomycin (1 μg/ml), respectively.

All graphs depict mean ± SEM. Two-tailed paired Student t test was used to derive statistical differences. A p value <0.05 was considered significant. Plots and statistical analyses were produced in GraphPad Prism.

Although IL-18 has previously been shown to activate NK cells during viral infection, these studies involve directly infecting IL-18– and IL-18R–deficient mice (7, 9). Therefore, to address whether IL-18 influences NK cell responses in a cell-intrinsic manner, we adoptively transferred equal numbers of wild-type (WT) and Il18r1^−/− NK cells into Ly49h^−/− mice, which harbor normal numbers of NK cells but are incapable of recognizing the MCMV-derived m157 protein (3, 8). Postinfection (PI) with MCMV, WT NK cells preferentially expanded during the first week of infection and were higher in frequency than Il18r1^−/− NK cells at day 7 PI (Supplemental Fig. 1A) and at later time points (Fig. 1A). Consistent with the adoptive transfer experiment, we observed a similar expansion defect by Il18r1^−/− Ly49H⁺ NK cells in WT:Il18r1^−/− mixed bone marrow chimeric mice infected with MCMV (Fig. 1B and Supplemental Figure 1B). Together, these studies confirm a cell-intrinsic requirement for IL-18 signaling in the antiviral NK cell response.

IL-18 has been suggested to enhance IL-12–induced effector functions of NK cells such as IFN-γ production (5, 6). To determine whether IL-18 might also “prime” NK cells for MCMV-driven expansion, we isolated resting NK cells from Il18^−/− mice and cotransferred them with equal numbers of WT NK cells into Ly49h^−/− hosts. Post MCMV infection, WT and Il18^−/− Ly49H⁺ NK cells exhibited comparable expansion and memory cell formation (Fig. 1C), indicating that previous exposure to IL-18 during development or homeostasis was not required for normal expansion in response to viral challenge as long as IL-18 is present during infection.

To determine whether the effect of IL-18 signaling was limited to Ag-specific NK cell responses, we used MCMV and VSV viruses engineered to lack or express m157, respectively. After adoptive transfer of WT and Il18r1^−/− NK cells into Ly49h^−/− hosts, we observed that in contrast with MCMV, infection with MCMV lacking m157 (MCMV-Δm157) elicited an equivalent (albeit modest) expansion of both WT and Il18r1^−/− NK cell populations (Fig. 1D), indicating that IL-18 signaling is not required for the low-level proliferation of “bystander” NK cells responding to viral infection. Consistent with this observation, infection with VSV-m157, but not with VSV-Ova, elicited a preferential Ag-specific expansion of WT NK cells over Il18r1^−/− NK cells (Fig. 1D). We next investigated whether IL-18 was required for NK cells to undergo homeostatic proliferation. We found that WT and Il18r1^−/− NK cells proliferated similarly when transferred into Rag2^−/− × Il2rg^−/− mice, generating comparable long-lived populations >4 wk later (Fig. 1E), demonstrating IL-18 signaling is not required for NK cells to proliferate in response to common γ-chain cytokines. Altogether, these data suggest that IL-18 signaling directly regulates Ag-driven expansion of NK cells during viral infection, but not during bystander or homeostatic expansion.

Because IL-18 is required for optimal proliferation of NK cells during viral infection, we investigated additional NK cell effector functions that may be compromised in the absence of IL-18 signaling. At day 1.5 PI, we observed that Il18r1^−/− NK cells exhibited a defect in IFN-γ production, but not granzyme B (Fig. 2A), consistent with a previous report (7). Although both NK cell populations became similarly activated at days 1.5 and 7 PI (Supplemental Fig. 1C, 1D), WT NK cells were found to be more mature, as measured by CD27 and CD11b staining, compared with Il18r1^−/− cells (Fig. 2B), suggesting IL-18 may have a role in promoting the maturation of activated NK cells.

Adoptive cotransfer and bone marrow chimeric studies revealed an expansion defect in the Il18r1^−/− NK cell response to MCMV infection, which could be a consequence of decreased proliferation or increased apoptosis of Il18r1^−/− NK cells relative to WT NK cells. We were not able to detect differences in BrdU incorporation, Ki67 staining, or CFSE dilution by WT and Il18r1^−/− NK cells in MCMV-infected bone marrow chimeric mice (data not shown), consistent with a prior study (10). In addition, FLICA incorporation studies did not reveal any differences in pan-caspase activity between Il18r1^−/− and WT NK cells on day 7 PI (Supplemental Fig. 1E), suggesting that Il18r1^−/− NK cells were not more susceptible to apoptosis. Lastly, when WT and Il18r1^−/− effector NK cells were sorted from MCMV-infected hosts on day 7 PI, and equal numbers cotransferred into naive recipients, the two populations contracted at similar rates (Fig. 2C), indicating that Il18r1^−/− NK cells are not undergoing increased cell death at later time points and that IL-18 signaling does not regulate the contraction phase or maintenance of memory NK cells.

Because the IL-18R requires the adapter molecule MyD88 for downstream signaling (11), we investigated the contribution of MyD88 to the NK cell response to MCMV infection. WT:Myd88^−/− mixed bone marrow chimeric mice were generated and both NK cell populations reconstituted similarly (data not shown). Post MCMV infection, Myd88^−/− Ly49H⁺ NK cells exhibited a cell-intrinsic expansion defect compared with WT NK cells (Fig. 3A), similar to Il18r1^−/− NK cells (Fig. 1C). In addition, fewer Myd88^−/− NK cells produced IFN-γ after MCMV infection despite comparable upregulation of CD69 (Fig. 3B). Similar to Il18r1^−/− NK cells, Myd88^−/− NK cells failed to mature as efficiently as WT NK cells at day 7 PI (Fig. 3C). When equal numbers of WT and Myd88^−/− NK cells were transferred into Ly49h^−/− hosts followed by infection with MCMV, we observed preferential expansion and memory cell formation of the WT NK cells compared with Myd88^−/− NK cells (Fig. 3D). The IL-1R, which is expressed by NK cells, is also known to use MyD88 for signaling (11); however, unlike Il18r1^−/− or Myd88^−/− NK cells, Il1r^−/− NK cells expanded comparably with WT NK cells at day 7 PI (Fig. 3E) and showed similar effector function and phenotype as WT NK cells at days 1.5 and 7 PI (Supplemental Fig. 2A–C). These findings support a mechanism for IL-18 signaling on NK cells through a MyD88-dependent signaling axis.

During MCMV infection, we observed an increase in Myd88 expression in sorted NK cells by microarray and qRT-PCR (Fig. 4A, data not shown). Given the important early role of IL-12 in activating NK cells, and a report that proinflammatory cytokines induce IL-18 signaling components in human NK cells (12), we investigated whether IL-12 directly influences the expression of the IL-18 signaling machinery. Overnight stimulation of sorted NK cells with IL-12 and IL-18 also showed an increase in the expression of Myd88 by qRT-PCR (Fig. 4B). STAT4 is a transcription factor that mediates signals downstream from the IL-12R (13), and analysis of the Myd88 promoter revealed one putative STAT4 binding site upstream of the transcriptional start site (Fig. 4C). STAT4 ChIP followed by quantitative PCR identified a significant enrichment of STAT4 binding directly upstream of the Myd88 transcriptional start site (Fig. 4D). This increase in STAT4 binding was dependent on IL-12 stimulation, suggesting Myd88 is a target gene of IL-12 signaling. Finally, we compared the expression of Myd88 in WT and Stat4^−/− NK cells during MCMV infection. Activated Stat4^−/− NK cells exhibited lower levels of Myd88 expression compared with WT NK cells. Altogether, these data indicate that IL-12 signaling acts through STAT4 to increase the expression of Myd88, identifying a novel role for IL-12 in potentiating IL-18 signaling early during MCMV infection.

To assess whether IL-18 signaling regulates the functional capabilities of memory NK cells, we stimulated WT and Il18r1^−/− memory NK cells with Ly49D, Ly49H, or IL-12 and IL-18 and measured their ability to degranulate and to make IFN-γ. Similar to resting NK cells, Il18r1^−/− memory NK cells produced less IFN-γ than their WT counterparts only when stimulated with proinflammatory cytokines, but not with Ly49D, Ly49H, or PMA and ionomycin (Fig. 5A and data not shown). Il18r1^−/− memory NK cells degranulated similarly to WT (Fig. 5B). Thus, like resting NK cells (5), memory NK cells continue to depend on IL-18 for IL-12–mediated IFN-γ production.

Given the importance of IL-18 signaling on the optimal expansion of naive NK cells during MCMV infection, we investigated its role during a recall response. Equal numbers of effector WT and Il18r1^−/− NK cells were cotransferred and “parked” in a naive host for 3 wk, followed by infection with MCMV. Surprisingly, Il18r1^−/− memory NK cells expanded similarly to their WT counterparts during secondary challenge by MCMV (Fig. 5C), demonstrating a stage-specific requirement for IL-18 signals for proliferation. After the peak of expansion, both cell populations contributed to equal frequencies of secondary memory NK cells. Thus, IL-18 signaling is specifically required for the primary expansion of virus-specific NK cells but is dispensable in the subsequent recall response.

NK cells possess adaptive immune qualities in response to cytokine treatment (14) or exposure to pathogen and nonpathogen Ags (15). It is poorly understood what the cytokine signal requirements are in resting and memory NK cells. In this study, we find a stage-specific requirement for IL-18, where IL-18 promotes the Ag-specific primary expansion of resting NK cells, but not the recall response of memory NK cells. Furthermore, we have identified a previously unknown role for IL-12 in promoting the expression of Myd88 to enhance the IL-18 signaling cascade and possibly sensitize NK cells to lower levels of IL-18 cytokine.

Our current findings suggest that as naive NK cells differentiate into Ag-experienced memory cells, they become “specialized” and may rely more on Ag-specific signals and less on proinflammatory cytokine signals for clonal-like expansion (16). This mechanism would allow previously Ag-experienced NK cells to respond more robustly during subsequent pathogen encounter. Because memory and recall NK cell responses have also been documented in humans post viral infection (1), determining whether a homologous role for IL-18 in the recall response of human NK cells may impact the use of this cytokine in clinical settings.

The authors have no financial conflicts of interest.