Activating Ly-49 NK Receptors: Central Role in Cytokine and Chemokine Production

Murine NK cells express multiple Ly-49 receptors that are type II transmembrane receptors. These receptors either inhibit or activate NK cell functions such as cytolysis or cytokine secretion. A functionally similar family of molecules exists on human NK cells, i.e., the killer cell Ig-like receptors (KIRs).² However, the human KIRs are structurally dissimilar to the murine Ly-49 family of receptors because they belong to the Ig superfamily of receptors.

The inhibitory Ly-49 receptors, Ly-49A, C, G, and I, inhibit NK cell function upon binding of class I ligands on target cells (1, 2, 3). These Ly-49 inhibitory receptors, as well as inhibitory KIRs, contain cytoplasmic immune receptor tyrosine-based inhibitory motifs (ITIMs) that are phosphorylated upon stimulation, leading to the recruitment of Src homology 2 domain-containing protein tyrosine phosphatase (SHP-1) and attenuation of intracellular signals (1, 4, 5). In contrast, the predicted amino acid sequences for the activating receptors, Ly-49D and Ly-49H, do not contain any ITIMs in their cytoplasmic domains, confirming that these are not inhibitory receptors (6, 7, 8, 9). Furthermore, it has been demonstrated that the activating receptors do not become phosphorylated after pervanadate treatment or receptor cross-linking and do not recruit SHP-1 (10). However, Ly-49D has been shown to mobilize intracellular Ca²⁺ and to mediate reverse Ab-dependent cellular cytotoxicity in the presence of specific mAb (10, 11).

These activating Ly-49 and KIR molecules have been shown to associate with a 12-kDa homodimeric protein, DAP12, that contains an immunoreceptor tyrosine-based activation motif that is critical for positive signaling by these receptors (10, 11). The Ly-49D and H activating receptors contain an arginine residue in their transmembrane domain that serves as a required docking site for DAP12 binding to the receptor (9, 10, 11). Association of the homodimeric DAP12 with these activating receptors is essential for phosphorylation of DAP12 following receptor triggering, leading to intracellular calcium mobilization and cytokine secretion (8, 9, 11, 12).

Most studies to date have examined the regulation of cytotoxic activity and ligand specificity of Ly-49 molecules (8, 13, 14). However, triggering of both human (6, 7) and mouse (15, 16, 17, 18) NK receptors results in both positive and negative (9, 19, 20) regulation of cytokine gene expression. Because little is known about the variety of genes that may be regulated upon triggering of the mouse Ly-49D receptor, we have examined the breadth of gene expression via microarray analysis following Ly-49D cross-linking and have provided pharmacological analysis of the biochemical pathways that are involved in Ly-49D signaling.

Splenic NK cells were isolated from C57BL/6 mice and were grown for 7–10 days in 1000 IU/ml recombinant IL-2 (Chiron Therapeutics, Emeryville, CA) as previously described (12). Liver NK cells were isolated from IL-2-treated mice as previously described (21). Liver mononuclear cells were either used when fresh or after IL-2 expansion.

Liver NK cells were incubated for 0.5 h with the concentrations of inhibitors (Calbiochem, La Jolla, CA) as indicated in the manuscript, and, at 6 h, supernatants were assayed for cytokine/chemokine production. Vehicle controls of DMSO or alcohol were used for selected inhibitors that were not soluble in water.

The monoclonal 4E5 (Ly-49D), 4D11 (Ly-49G2), and 24G2 (Fc receptor) were used as previously described (22). Rat IgG (BD PharMingen, San Jose, CA) was used as a control for functional studies and immunoprecipitations. Rabbit F(ab′)₂ anti-rat IgG was used as a cross-linking reagent. NK1.1-PE, DX-5-PE, and CD3e-PcP (BD PharMingen), as well as 4E5-FITC, were used for flow cytometric analysis.

Cells were stained as previously described (12) and analyzed on a FACSort flow cytometer (BD Biosciences, San Jose, CA) and a MoFlo cytometer (Cytomation, Ft. Collins, CO). Cells were directly stained using PE- and FITC-labeled primary Abs or indirectly stained using a primary Ab followed by an isotype-specific FITC- or PE-conjugated secondary Ab or a biotinylated primary Ab followed by streptavidin PerCP (BD Biosciences).

Cytokines were measured using IFN-γ and chemokine ELISA kits (R&D Systems, Minneapolis, MN). Cell stimulations were performed at cell concentrations of 1–5 × 10⁶/ml. Abs were added at a concentration of 1 μg/10⁶ cells for 30 min at 4°C. Cells were then washed and plated on 24-well Costar (Corning, Corning, NY) plates that were precoated with 2 μg/well rabbit F(ab′)₂ anti-rat IgG and blocked with medium containing 10% FCS. Unless stated otherwise, samples were collected after 5–6 h incubation (37°C, 5% CO₂) and were measured in duplicate against the standard curve of the assay and reported as pg/ml. In all assays, the SD of the cytokine measurement was <5 pg/ml.

Approximately 200 × 10⁶ C57BL/6 splenic NK cells expanded with 1000 U/ml IL-2 were stimulated with anti-Ly-49D (4E5) or control IgG for 6 h. These expanded NK cells were 85% CD3⁻NK1.1⁺. In one experiment, primary liver NK cells were isolated from animals treated for 3 days with IL-2, as previously described (21). Poly(A)⁺ RNA was isolated according to the manufacturer’s protocol using a FastTrack 2.0 mRNA isolation kit (Invitrogen, Carlsbad, CA). cDNA generation, hybridization, and data collection were performed by Incyte Genomics. In brief, alterations in gene expression were evaluated by reverse transcription of poly(A)⁺ RNAs in the presence of Cy3 or Cy5 fluorescent labeling dyes followed by hybridization to a mouse GEM 2 microarray chip. Each chip displays a total of 8734 elements of which 7854 are unique genes/clusters. These unique gene/clusters can be further defined as 3205 annotated and 4649 unannotated sequences. Subsets of genes were selected for further study based on differential Cy3/Cy5 expression ratios that were ≥2 in response to Ab-cross-linking treatment. Differential expression of representative selected genes was confirmed by RNase protection assay (RPA). Definition of terms for the gene chip can be found at http://reagents.incyte.com/support/ gem/unigem_leg.html.

The multiprobe RPA was performed using the mCK-1 or mCK-5 probe set (BD PharMingen) or a custom multiprobe set (Torrey Pines Biolabs, La Jolla, CA). Total cellular RNA was extracted using TRIzol (Life Technologies, Gaithersburg, MD), and 5 μg of the total was hybridized with a [³³P]UTP-labeled RNA probe (1–1.5 × 10⁶ cpm/sample) prepared according to the manufacturers directions (BD PharMingen) using the PharMingen RiboQuant in vitro transcription kit. Following hybridization, the samples were treated with RNase A and T1 according to the procedure provided by BD PharMingen. The RNase was inactivated and precipitated using a master mixture containing 200 μl RNase inactivation reagent (Ambion, Austin, TX), 50 μl ethanol, 5 μg yeast tRNA, and 1 μl GycoBlue coprecipitate (Ambion) per RNA sample. The samples were mixed well, incubated at −70°C for 30 min, and centrifuged at 14,000 rpm for 15 min at room temperature. The pellets were suspended in 3 μl of BD PharMingen sample buffer and subjected to PAGE as recommended by the manufacturer.

Murine NK cells express Ly-49 receptors that can initiate either inhibitory or activating signals to regulate lytic function. Ly-49A, C, and G2 have been shown to inhibit NK cell function upon recognition of class I ligands on target cells (265, 152, 229, 258, 231, 262). These inhibitory receptors contain ITIMs in their cytoplasmic domains that are phosphorylated upon stimulation leading to the recruitment of SHP-1 and attenuation of intracellular signals (5, 6). However, receptors exist in both systems that can activate NK cells. Murine Ly-49D activates NK cells by inducing reverse Ab-dependent cellular cytotoxicity of FcR⁺ targets. Upon activation, this molecule associates with and stimulates the rapid phosphorylation of a specific protein, DAP12. This associated phosphoprotein exists as a disulfide-linked dimer that becomes highly tyrosine-phosphorylated upon specific receptor cross-linking. The in vivo function of these receptors is not known, but the activating NK receptors have been shown to be functional in activation of cytotoxicity (5, 6) and cytokine production (20). In addition, Ly-49D was only found on CD3⁻NK1.1⁺ cells. Extensive analysis of T cells, including NK T cells, has indicated that the activating Ly-49D is not present on cells other than NK cells (8, 10, 13).

In the present study, we sought to characterize the signals that are induced upon Ly-49D receptor cross-linking. For our initial assay, we examined both freshly isolated and IL-2-expanded NK cells (data not shown) (20) for the kinetics of anti-Ly-49D (4E5) receptor-induced IFN-γ gene transcription. In IL-2-expanded NK cells, IFN-γ gene expression could be observed by 3 h and became maximal by 6 h, whereas the kinetics of gene induction in primary cells was faster, as maximal mRNA expression was observed between 1 and 3 h (data not shown). IL-2-expanded NK cells activated for 6 h were used due to the inability to obtain sufficient poly(A) RNA from fresh cells. In addition, controls for nonspecific activation of NK cells by FcR and or other surface Ly-49, as well as specific activation of cells by F(ab′)₂ anti-Ly-49D (4E5), were previously reported (12). In addition, anti-FcR Ab (24G2) activated when fresh or IL-2-expanded NK cells did not produce IFN-γ or chemokines during the 6 h of treatment (Table I). Although CD16 cross-linking has been shown to induce cytokines, the contribution of CD16 to the Ly-49D induction was not evident under the conditions examined. Based on these results, IL-2-expanded activated NK cells (85% CD3⁻NK1.1⁺) were used to generate poly(A) mRNA for microarray analysis using the mouse GEM 2 microarray. Results of this experiment, highlighting genes that were decreased or increased by >2-fold or greater, are shown in Table II. As can be seen in Table II, only a few genes were significantly decreased (>2 balanced differential expression). In contrast, as shown in Table I, Ly-49D activation induced a number of genes that fell into two major categories, e.g., cytokine or chemokine genes and apoptosis-related genes. The latter is not unexpected, because our previous studies have shown that activating NK receptors can induce apoptosis-mediated events in both human and mouse NK cells (8, 23). The other genes that were activated were either regulatory proteins or secretory proteins. Interestingly, in addition to lymphotactin, four of the six genes most strongly induced were chemokine genes: MIP1α, MIP1β, MIP1-C10, and single C motif (SCM)1. It has been shown previously that NK cells both respond to and make chemokines, but the induction of chemokines via activating NK receptors suggests an important role for NK cells in attracting other immune effector cells upon interaction of this receptor with its target ligand.

The gene chip analysis shown in Table II is the result of a single analysis. Due to the amount of poly(A) RNA required for analysis and the cost, it was not possible to perform multiple experiments. However, verification of gene chip results was performed by subsequent measurement of mRNA by RPA and cytokine secretion. To verify the results of the microarray analysis, NK cells from two different experiments were stimulated with either IgG or 4E5 and plated on plates coated with anti-rat IgG. These cells were then evaluated for mRNA expression by RPA, and typical results are shown in Fig. 1. Consistent with the microarray analysis, MIP1α and MIP1β mRNA but not RANTES and monocyte chemoattractant protein (MCP)-1 mRNA were induced by activating the Ly-49D NK receptor. In addition, an increase of TNF-β and inducible protein-10 mRNAs was observed with RPA analysis. As indicated above, these treatments also induced a strong increase in IFN-γ mRNA (data not shown). To further verify that activating NK receptors could stimulate chemokine expression, a series of experiments were performed in which fresh NK cells treated with IgG or 4E5 were placed on plates coated with anti-rat IgG, supernatants collected after 6 h, and cytokine or chemokine protein production evaluated using specific capture ELISAs. The summary of four experiments is shown in Fig. 2. Parallel to that which was observed in the mRNA analysis, NK cells made significant quantities of IFN-γ, MIP1α, and MIP1β protein after anti-Ly-49D (4E5) cross-linking. Chemokines KC and MCP-1 were not produced, whereas low levels of MIP-2 were made spontaneously but not altered by NK receptor cross-linking. We were not able to analyze lymphotactin (SCM1) protein because an ELISA is not commercially available. Thus, in addition to mRNA, cytokine and chemokine protein expression was strongly increased upon Ly-49D receptor cross-linking.

To date, detailed analysis of the biochemical signaling pathways downstream of Ly-49 activating receptors has not been reported. Previous studies have demonstrated that Syk kinases (24) are involved in downstream phosphorylation events that are mediated through mouse and human activating receptors. Therefore, we analyzed the effects of a series of pharmacological inhibitors to identify possible pathways activated by these receptors. These inhibitors were tested for their effects on cytokine/chemokine protein expression by inclusion into assays in which fresh NK cells were cross-linked with anti-Ly-49D (4E5) for 6 h. Inhibitors were tested at a two-log range above and below the published IC₅₀. Results are shown in Table III for IFN-γ, MIP1α, and MIP1β and are derived from three experiments. Values represent the observed IC₅₀ determined by regression analysis. It is compared with the published IC₅₀ from a chemical source. IFN-γ expression was very sensitive to Src family kinase and calcium-dependent phosphatase inhibitors as well as phosphatidylinositol (PI)-3 kinase inhibitors. These results are all consistent with the tyrosine phosphorylation of the DAP12 molecule upon Ly-49D cross-linking and subsequent signaling through Syk in a calcium-dependent pathway. Close analysis of the IC₅₀s for the different compounds indicated that the PI-3 kinase inhibitor Ly204002 and Src family inhibitor 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d] pyrimidine (PP2) required significantly higher doses (20- to 100-fold) to inhibit MIP1α and MIP1β (Table III) than IFN-γ. These doses were considered nonspecific, and these results suggest that the pathway(s) involved in the activation of these genes may be distinct from that of IFN-γ. Examination of p38 kinase (SB 203580) and mitogen-activated protein kinase kinase (PD 98059) inhibitors demonstrated mitogen-activated protein kinase kinase involvement in Ly-49D signaling, but not p38 kinase.

When the kinetics of inhibition of IFN-γ was compared with MIP1α and MIP1β (data not shown), both IFN and MIP were rapidly induced in the first 3 h. Although this data is consistent with rapid induction of new mRNA, we wanted to determine whether some of the gene expression might have been due to the presence of preexisting protein stored in granules that was then released upon Ly-49D activation. In an attempt to evaluate this possibility, we inhibited new mRNA synthesis by pretreatment of the NK cells with actinomycin D and then evaluated both MIP1α and MIP1β and IFN-γ production. Both IFN-γ and MIP1α production (data not shown) were completely blocked by actinomycin D, indicating that virtually all gene expression is the result of new mRNA expression upon receptor cross-linking.

In the present study, in an attempt to understand potential novel functions of receptors in vivo, we evaluated gene expression after cross-linking Ly-49D, an activating mouse NK receptor. Analysis of genes induced upon receptor cross-linking indicated that most of the strongly induced genes fell into two categories: 1) soluble factors and 2) apoptotic genes. Although other genes were identified, most were only modestly induced or reduced compared with control mRNA levels (see Table I). The majority of the mRNAs that were strongly induced as analyzed by microarray were chemokine genes. RPAs and chemokine protein production analysis validated the microarray results, as cross-linking the Ly-49D mouse NK receptor induced high levels of IFN-γ, lymphotactin, MIP1α, and MIP1β. This gene induction was specific because other chemokines such as KC and MIP-2 were unaffected. Although NK cells have previously responded to chemokines (25, 26), our data demonstrates that these cells can be strong producers of these inflammatory mediators. Although the results presented here indicate that Ly-49D is a potent inducer of IFN-γ and chemokines, our findings do not rule out the possibility that activation through other NK receptors (e.g. NKG2, FcR, etc.) also results in the induction of these potent immune modulators.

To gain more in insight into the activation pathways of both cytokine (IFN-γ) and chemokine (MIP1α, MIP1β) gene expression, a series of pharmacological inhibitors was used to identify the key signaling pathways involved in the cellular response. Previous studies with Ly-49D have indicated that membrane proximal tyrosine phosphorylation of DAP12 was a tysosine kinase-mediated event that leads to calcium mobilization (20, 24) and was proposed to be mediated by Src family kinases. This hypothesis was confirmed, as complete inhibition of IFN-γ expression was observed with the p72 Syk and Src kinase family inhibitors, Picannetol and PP2, respectively. Interestingly, the Src kinase inhibitor PP2 did not completely block MIP1α and MIP1β production at the reported IC₅₀ levels. In contrast, protein kinase C, Ca²⁺ phosphatase, and calmodulin phosphatase inhibitors completely blocked both MIP1α and MIP1β as well as IFN-γ production at IC₅₀ levels. These data would suggest that the primary Ly-49D signaling for IFN-γ production is predominately mediated through Src kinase pathways, whereas MIP1α and MIP1β gene induction is more complex and may involve multiple biochemical pathways.

Regardless, these results collectively suggest the primary role for the activating NK receptors is cytokine and chemokine production. Previous studies have implicated Ly-49D as a positive regulator of cytotoxicity of D^d-expressing targets (13, 14). Although these results are clear, they are difficult to demonstrate unless high-sorted subsets of NK cells (>95% Ly-49D⁺) (13) or transfected effectors (14) are used. Demonstration of a direct cytotoxic role for Ly-49D against H-2D^d targets using unseparated NK cells has been very difficult, and considerable cytotoxic activity can be demonstrated in Ly-49D-negative NK cells. In contrast, strong IFN-γ production has been shown upon target interaction with unseparated NK cells as well as by receptor cross-linking in this report and previous (18, 20, 27) studies. Our current study extends our findings with regard to chemokine genes (lymphotactin, MIP1α, MIP1β) and suggests that a primary function of these activating NK receptors in immune regulation is one of immunomodulatory factor production. Recent reports of the phenotype of activating receptor DAP12-knockout mice (28) and receptor signaling disruption (29) demonstrated that the cytotoxic phenotype of NK cells was essentially intact. The DAP12-knockout mice (28) lacked or demonstrated a significant reduction of Ly-49D expression but exhibited little or no difference in direct cytotoxicity against a variety of NK targets. These mice lacking this signaling moiety did demonstrate a lack of experimental allergic encephalomyelitis (EAE). Several studies have indicated that chemokines and cytokines can dramatically alter the extent and character of EAE (30, 31). IFN-γ has been implicated as a direct regulator in EAE by acting on T cell proliferation and directing chemokine production, with profound effects on the onset and progression of the disease (30). Other studies (31) have suggested that cytokine and chemokine expression correlates with the Th1/Th2 paradigm contributing to the genetic basis of the EAE immune response. Thus, NK cells from the the DAP12-knockout mice should lose their ability to express IFN-γ and chemokines through activating Ly-49. The NK cells from the DAP12-disruption (29) mice also demonstrated minor effects on myeloid target killing, and there were profound defects on dendritic cell migration. These DAP12-disruption mice exhibited a dramatic accumulation of dendritic cells in mucocutaneous epithelia, associated with an impaired hapten-specific contact sensitivity. These data strongly suggest a unique role for DAP12 in innate immunity through the expression of secreted factors that regulate other hematopoietic cells. A defect in the initial production of cytokines and chemokines by NK cells might explain the defect in dendritic cell localization, as it is well established that macrophage and dendritic cells’ migration is controlled by chemokine production. DAP12 was shown (32) to be present in nonlymphoid cells including dendritic cells and monocytes. More recently (33, 34), DAP12 has been demonstrated as a signal-regulatory molecule with signal-regulatory protein β1 and MDL-1, receptors that are commonly found on either monocytes and/or dendritic cells. These reports suggest that DAP12 transduce immunoreceptor tyrosine-based activation motif-mediated activation signals that will regulate monocyte and dendritic cell functions. Chemokines (35) and chemokine receptors (36, 37, 38) have been shown to regulate dendritic cell localization in tissues and to regulate trafficking via the lymph or blood to lymphoid organs. Gangur and Oppenheim predict (38) that suppression of chemokines would interrupt the sequence of signals that culminate in an allergic response.

In addition, in vivo transplantation studies in perforin-null mice have shown that NK-mediated hybrid resistance is intact (39). This result suggests a critical role for the expression of soluble factors by NK cells, as opposed to direct cytotoxicty. Thus, one might conclude that activating NK receptors are more involved in soluble immune regulation than in cytotoxic regulation. Although the inhibitory NK receptors have been shown to strongly regulate both cytotoxic and secretory functions in vitro, their in vivo role has not been thoroughly defined. Thus, we conclude that a primary role for the activating NK receptors in vivo is to trigger soluble factor production and regulation of the immune response at the site of receptor activation. This would place NK cells and their activating Ly-49 receptors as important initiators of microbial immunity and key elements of the innate immune system.