Cutting Edge: Induction of IFN-γ Production but Not Cytotoxicity by the Killer Cell Ig-Like Receptor KIR2DL4 (CD158d) in Resting NK Cells

The mAbs, cells, and reagents

BALB/c mice were immunized i.p. five times with 5 × 10⁶ NK3.3 cells and boosted once with 50 μg 2DL4-Ig fusion protein. Supernatants of hybridomas produced by fusion with P3×63Ag8 were screened by ELISA with 2DL4-Ig fusion protein (9). Positive hybridomas were further tested by flow cytometry on NK3.3 cells. Three hybridomas were obtained after subcloning: 33 (IgG1), 36 (IgM), and 64 (IgM). The hemaggluttinin (HA)-tag specific Abs (Covance, Richmond, CA) used were mAb 16B12 (IgG1) and polyclonal rabbit Ab HA.11. The mAbs directed against NK surface proteins were 2B4 (C1.7, IgG1; Beckman Coulter, Miami, FL); CD94 (HP3D9, IgG1; Ancell, Bayport MN), and CD16 (3G8, IgG1; Medarex, Princeton, NJ). Isotype-matched control Abs for IgG1, IgG2a, and IgM were obtained from Beckman Coulter. The following NK cell lines were used: NK92 (obtained from H. Klingemann, Rush Unversity, Chicago IL), NK3.3 (obtained from J. Kornbluth, St. Louis University School of Medicine, St. Louis, MO), and NKL (obtained from M. J. Robertson, Indiana University Cancer Research Institute, Indianapolis, IN). 293T/17 and P815 cells were obtained from American Type Culture Collection (Manassas, VA). Human NK cell populations were derived from the PBL of normal donors as previously described (10) using the MACS NK cell isolation kit (Miltenyi Biotec, Auburn, CA). Freshly isolated NK cells cultured in Iscove’s medium containing 10% human serum without any added cytokines were used within 24 h. NK cells were expanded in the presence of IL-2 as previously described (10).

Molecular construct transfections and immunoblotting

The 2DL4 cDNA cloned into pBlueScript was used as a PCR template using the following primers: sense primer (containing a BglII site), 5′-GGGGAGATCTCACGTGGGTGGTCAGGACAA-3′; and antisense primer (containing SalI and NheI sites), 5′-GACTGGTCGACGCTAGCTCAGATTCCAGCTGCTGGTA-3′, and cloned into the BglII and SalI sites of pDisplay in frame with a signal sequence and a HA tag (Invitrogen, Carlsbad, CA). To produce the HA-2DL4 (RY-GT) mutant, the pDisplay-HA-2DL4 was mutated using the QuikChange kit (Stratagene, La Jolla, CA). Primers used were: sense, 5′-CATGCTGTGATTGGGACCTCAGTGGCCATC-3′; and antisense, 5′-GATGGCCACTGAGGTCCCAATCACAGCATG-3′. Wild-type and mutant HA-2DL4 were cloned using KpnI and NheI into the vaccinia virus vector pSC65 with a modified polylinker, and recombinant viruses were produced as previously described (9). All constructs were verified by sequencing. The 293 T/17 cells were transiently transfected using LipofectAMINE (Life Technologies). After 48 h, cells were lysed in 0.3% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate, 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM PMSF, and 8 mM iodoacetamide. Samples were either used as total lysate or immunoprecipitated with mAbs followed by protein G-agarose (Life Technologies). Western blotting was done as previously described (10), and blots were developed using rabbit anti-HA Abs (Covance), followed by goat anti-rabbit IgG peroxidase (Amersham, Arlington Heights, IL) and Super Signal substrate (Pierce, Rockford, IL).

Functional assays with NK cells

NK cells (5 × 10⁵/well) were cocultured with or without P815 cells (1 × 10⁵/well) for 20 h with mAbs as indicated. Culture supernatants were tested for IFN-γ production by ELISA (R&D Systems, Minneapolis, MN). In parallel, NK cells were also tested for cytotoxicity against P815 cells in a 3-h ⁵¹Cr-release assay. Purified vaccinia viruses were used to infect IL-2-activated NK cells as previously described (9). After infection with 10 PFU/cell of each virus for 1.5 h, cells were washed and either monitored for receptor expression by flow cytometry or plated for standard 3-h ⁵¹Cr-release assays using P815 target cells. Treatment of NK cells with the extracellular signal-regulated (ERK) kinase (MEK)1 inhibitor PD098059 (Calbiochem, La Jolla, CA) or the p38 mitogen-activated protein kinase (MAPK) inhibitor SB203580 (Sigma-Aldrich, St. Louis, MO) was for 1–2 h at 37°C before addition of Abs or rIL-2 (5 U/ml). Inhibitors were present during stimulation with mAbs. Negative controls contained as much DMSO as the highest concentration of inhibitor. The inhibitors did not interfere with NK cell viability as assessed by trypan blue exclusion. After 20 h, culture supernatants were tested for IFN-γ by ELISA.

2DL4 induces cytotoxicity by activated NK cells but not by resting NK cells

The mAbs against 2DL4 were generated to characterize the expression and signaling properties of 2DL4. One IgG (#33) and two IgMs (#36 and #64) were identified that reacted specifically with 2DL4 and not other KIR family members. In an ELISA, mAb 33 bound to 2DL4-Ig but not to KIR-Ig fusion proteins of 2DL1, 2DL2, 2DL3, and 3DL2 (Fig. 1⇓A). This mAb also did not bind KIR-Ig fusion proteins of 2DS2 and 2DS4 (data not shown). Similar results were obtained with the mAbs 36 and 64 (data not shown). These mAbs also reacted with cells transfected with 2DL4 but not 2DL5, a closely related KIR (C. Chang and A. King, personal communication).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

The mAb 33 is specific for 2DL4 and stains NK cells. A, The KIR-Ig fusion proteins 2DL1 (•), 2DL2 (▴), 2DL3 (▾), 3DL1 (♦), and 2DL4 (▪) were tested for reactivity with mAb 33 in an ELISA. B, Total cell lysate (−) or immunoprecipitates (IP) using mAb 33 and DX27 from mock- or HA-2DL4-transfected 293T/17 cells were blotted with anti-HA. C, Surface 2DL4 expression of the indicated cell lines and of NK cells from a normal donor as detected with mAb 33. The thin lines indicate staining with control Abs.

Lysates of 293T/17 cells transfected with a cDNA encoding an epitope-tagged 2DL4 (HA-2DL4) were immunoprecipitated with mAb 33. Western blotting for the HA epitope revealed a broad 45- to 50-kDa band similar to that seen by blotting of the total cell lysate (Fig. 1⇑B). All three anti-2DL4 mAbs bound to the NK cell line NK3.3, but not to the B cell line 721.221 and the T cell line Jurkat as shown by flow cytometry (Fig. 1⇑C and data not shown). The NK cell lines NKL and NK92 displayed low levels of cell surface 2DL4. NK cells isolated from peripheral blood of different donors varied from low (similar to that seen with NK92) to intermediate (as shown in Fig. 1⇑C) reactivity with the different anti-2DL4 mAbs.

The 2DL4-specific mAb 33 was used to test whether 2DL4 induces activation or inhibition of lysis. Lysis of the FcR-positive P815 cells by IL-2-activated NK cells in the presence of mAb 33 was comparable to that obtained with mAbs to the activation receptors 2B4 and CD16 (Fig. 2⇓A). Lysis of P815 cells induced by mAb 33 was also obtained with several NK cell clones and with the cell lines NK92, NKL, and NK3.3 (data not shown). We conclude that 2DL4 has properties of an activation receptor.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Cross-linking 2DL4 induces cytotoxicity in IL-2-activated NK cells but not in resting NK cells. A, IL-2-activated NK cells were incubated with P815 cells (E:T ratio of 2) in the presence of the indicated mAbs (1 μg/ml). The ratio of lysis over that obtained with control Ig was 3.6 ± 0.23 for 2DL4 (n = 15), 3.3 ± 0.59 for 2B4 (n = 6), and 4.8 ± 0.39 for CD16 (n = 15). B, Cytotoxicity of resting NK cells against P815 cells (E:T ratio of 2) in the presence of the indicated mAbs. The ratio of lysis over that obtained with control Ig was 4.9 ± 0.47 for CD16 (n = 16), 3.1 ± 0.11 for 2B4 (n = 13), and 1.07 ± 0.02 for 2DL4 (n = 16).

To test the outcome of 2DL4 coligation with an inhibitory receptor, NK cells that express mainly the CD94/NKG2A inhibitory receptor were selected among different donors (7). Coligation of CD94/NKG2A inhibited the lysis induced by 2DL4, 2B4, and CD16 (Fig. 2⇑A and data not shown). Thus, the activation signal delivered by 2DL4 in activated NK cells is typical of NK activation receptors that are sensitive to inhibition by ITIM-containing receptors.

Resting NK cells were incubated with P815 cells in the presence of mAbs for 2DL4, CD16, and 2B4 (Fig. 2⇑B). Enhancement of lysis by CD16 and by 2B4 was similar to that obtained with activated NK cells, except for the weaker lytic potential of freshly isolated NK cells. In contrast, 2DL4 did not enhance cytotoxic activity in resting NK cells. Thus, 2DL4 induction of killing was restricted to activated NK cells, distinguishing its activity from that of CD16 and 2B4. Coligation of 2B4 or CD16 with CD94 did not decrease the level of lysis (Fig. 2⇑B and data not shown), even though resting NK populations were chosen that yielded activated NK cells in which CD94 was inhibitory. Time course (6–72 h) and titration studies with 0.1–16 μg/ml of the Abs against 2DL4, 2B4, and CD16 were done to ensure that the unique induction of IFN-γ but not cytotoxicity upon 2DL4 activation was not a function of the intensity of mAb triggering (data not shown).

The transmembrane region of 2DL4 is necessary for the activation signal

Recombinant vaccinia viruses encoding 2DL4 and a mutated 2DL4 in which the amino acids arginine-tyrosine in the transmembrane region were replaced by the glycine-threonine (mutant RY-GT) conserved in all other KIR receptors were generated. HA-2DL4 was used to distinguish recombinant from endogenous 2DL4. Homogenous surface expression of these receptors was obtained by infection of NK cell populations (Fig. 3⇓A). Ligation of wild-type HA-2DL4 receptor by anti-HA mAb resulted in activation of lysis of P815 cells (Fig. 3⇓B). In contrast, ligation of the RY-GT mutant did not result in activation despite higher expression of mutant RY-GT than wild-type HA-2DL4 and even though P815 lysis was induced by ligation of 2B4 in the same infected cells. The small reduction in lysis of P815 cells as compared with control Abs may be a function of the inhibitory effect mediated by the ITIM in the cytoplasmic tail. Mutation of the tyrosine in the ITIM to a phenylalanine did not diminish the ability of 2DL4 to activate lysis (M. Faure, S. Rajagopalan, and E. O. Long, unpublished observations). We conclude that 2DL4-mediated activation requires an intact transmembrane domain but no ITIM. The unique arginine-tyrosine motif is conserved in the transmembrane region of 2DL4 in humans, chimpanzees, and rhesus monkeys (11, 12). It is likely that the arginine in the transmembrane domain mediates association with a partner chain responsible for the activation signal, as seen with other NK-activating receptors that associate with either FcγR, DAP12, or DAP10 (4). Transfection experiments in the Ba/F3 and 293T cell lines have also indicated that 2DL4 does not pair with FcRγ chain, DAP12, or DAP10 chains (J. Wu, T. Pertel, and L. Lanier, unpublished observation). Studies are underway to identify the putative partner chain of 2DL4.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

An intact transmembrane region of 2DL4 is required for the activation signal. Recombinant vaccinia viruses encoding HA-2DL4 or (RY-GT) mutant of HA-2DL4 were used to infect IL-2-activated NK cells. A, Flow cytometry analysis after infection using either anti-HA mAb or anti-2DL4 mAbs. B, Infected cells were used in a lysis assay with P815 cells (E:T ratio of 2) in the presence of mAbs as indicated. Fold activation: HA-2DL4/control Ig = 2.9 ± 0.18 (n = 4); RY-GT mutant/control Ig = 0.56 ± 0.03 (n = 4).

2DL4 induces strong IFN-γ production by resting NK cells

IFN-γ secretion in both activated and resting NK cells was measured after coculture with P815 cells in the presence of mAbs to NK receptors. In activated NK cells, the ability of 2DL4 to induce IFN-γ was greater than that of the activation receptors CD16 and 2B4 (Fig. 4⇓A). As seen in cytotoxicity assays (Fig. 2⇑A), coligation of 2DL4 with CD94 resulted in inhibition of IFN-γ production (Fig. 4⇓A). Activation of IFN-γ production by 2DL4 cross-linking was even more striking in resting NK cells (Fig. 4⇓B). By comparison, ligation of 2B4 and CD16 induced a much smaller response. Coligation of 2DL4 with CD16 had no additional effect on IFN-γ secretion (Fig. 4⇓B). Therefore, in resting NK cells, 2DL4 is unique in its ability to induce efficient IFN-γ production in the absence of any discernible lytic function. Intracellular staining of IFN-γ in both activated and resting NK cells stimulated with 2DL4 revealed that expression occurred in the bulk of the NK cell population (data not shown) in contrast to the selective expression of IFN-γ by CD56^bright NK cells in response to IL-12 (13).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

2DL4 induces IFN-γ production in resting NK cells. A, IFN-γ produced by IL-2-activated NK cells incubated with P815 cells for 20 h in the presence of receptor-specific mAbs (1 μg/ml). Increase in IFN-γ release over that induced by control IgG was 35.4 for 2DL4, 1.55 for 2B4, and 20.15 for CD16 (n = 2). B, IFN-γ produced by resting NK cells incubated with P815 cells in the presence of the indicated mAbs. Increase in IFN-γ release over control IgG was 57 ± 5.1 for 2DL4 (n = 18), 2.1 ± 0.15 for 2B4 (n = 17), and 4.2 ± 0.5 for CD16 (n = 18). C, IFN-γ production by the same resting NK cells as in B incubated with either isotype-matched control mAb or mAb against CD16 or 2DL4 (33) in the absence of secondary Ab. On the right, resting NK cells from a different donor were incubated with isotype-matched controls and with Abs 33 (IgG), 36 (IgM), and 64 (IgM) for 20 h. Enhancement of IFN-γ production over control Ig was 66 ± 5.9 for 33 (n = 16), 150 ± 37 for 36 (n = 5), and 92 ± 22 for 64 (n = 5). D, IFN-γ secretion by resting NK cells treated with MEK1 and p38 MAPK inhibitors. Top, Cells were incubated with SB203580 or PD098059 for 1 h at 37°C before activation with anti-2DL4 mAb for 20 h in the continuing presence of inhibitor. The four curves represent resting NK cells from four different donors. The IFN-γ release in the absence of inhibitor with resting NK cells from the different donors ranged between 660 and 1490 pg/ml. Bottom, IFN-γ release by resting NK cells incubated with inhibitors for 2 h before activation with rIL-2 (5 U/ml) for 20 h in the continuing presence of inhibitor.

Co-cross-linking of CD94 with 2DL4 on activated NK cells inhibited lysis and IFN-γ production (Figs. 2⇑ and 4⇑, A). In contrast, similar co-cross-linking on resting NK cells did not inhibit IFN-γ secretion induced by 2DL4 (Fig. 4⇑B) or lysis induced by 2B4 (Fig. 2⇑B). Expansion of NK cells from this resting NK population yielded cells with inhibitory CD94 (data not shown). The lack of 2DL4 sensitivity to CD94-mediated inhibition in resting cells could be due to different repertoires of CD94/NKG2 on resting vs activated NK cells. Alternatively, the ITIM-based inhibitory pathway may not function in resting NK cells. In any case, it is significant that activation through 2DL4 in resting NK cells is not inhibited by the receptor for HLA-E.

Robust IFN-γ production was obtained by incubation with the IgG1 33 or either one of the two IgM mAbs in the absence of further cross-linking (Fig. 4⇑C). The use of the two IgM mAbs, 36 and 64, excluded a role of FcγR in the activation by 2DL4. Therefore, the signal transduced by 2DL4 alone induces IFN-γ production by resting NK cells in the absence of cytokines or signals from other receptors. In contrast, engaging the receptors CD16 and 2B4 with soluble mAbs is not sufficient, and additional accessory interactions are needed to trigger IFN-γ release (Refs. 10 and 14 and data not shown).

The IFN-γ secretion by NK cells induced by IL-2 depends on an ERK mitogen-activated protein kinase pathway (15). The ERK-dependent IFN-γ secretion was also observed upon cross-linking of CD16 or β₁ integrin on activated NK cells or by mixing NK cells with the sensitive target cell line K562 (16, 17). Inhibitors of MEK1 (PD98059) and of p38 MAPK (SB203580) were added to resting NK cells during stimulation with anti-2DL4 mAbs to test whether 2DL4-induced production of IFN-γ required MAPK (Fig. 4⇑D, top). Complete inhibition occurred at 1 μM of the p38 inhibitor SB203580. The ERK pathway inhibitor, PD98059, had only a partial inhibitory effect. In contrast, the IL-2-mediated induction of IFN-γ production was severely inhibited by 50 μM PD98059 and unaffected by SB203580 (Fig. 4⇑D, bottom), as previously reported (15). Thus, the p38 MAPK-dependent 2DL4 signal for IFN-γ secretion is different from signals delivered to activated NK cells by CD16 and β₁ integrin and to resting NK cells by the IL-2R, which are all sensitive to an ERK MAPK inhibitor. Further biochemical analysis of the 2DL4 signaling pathway has been hampered thus far by the limited number of freshly isolated resting NK cells available. Polarization of cytotoxic granules in NK cells, leading to target cell killing, is also erk dependent (18). The use of a p38 rather than an ERK pathway for activation of resting NK cells by 2DL4 may be designed to avoid cytotoxicity while maintaining the IFN-γ response. In this regard, IFN-γ induction by 2DL4 is similar to IFN-γ production by Th1 cells and to IFN-γ gene transcription in T cells stimulated by IL-12 and IL-18 that are also regulated by p38 (19, 20).

To fully understand the physiological relevance of 2DL4-mediated IFN-γ production, the ligands that recognize this receptor need to be identified. One ligand that interacts with 2DL4 is HLA-G, because 2DL4 binds to cells expressing HLA-G (8, 9, 21). HLA-G is expressed by fetal trophoblast cells that invade maternal decidua, where they encounter NK cells during early pregnancy (22). Uterine NK cells are a major source of IFN-γ in pregnant mice, in which IFN-γ has an important role in vascularization at the implantation site (23). We have observed enhanced IFN-γ production by resting NK cells in the presence of HLA-G-expressing cells (S. Rajagopalan and E. O. Long, unpublished observations). The potential role of 2DL4 in stimulating IFN-γ production during pregnancy warrants further study with decidual NK cells and trophoblast cells.

This study has identified 2DL4 as a receptor with the unique and autonomous ability to induce rapid IFN-γ secretion but not cytotoxicity by resting NK cells in the absence of cytokines. Signals that can activate resting NK cells have physiological relevance to the in vivo induction of NK responses. Furthermore, we show that resting NK cells behave differently than activated NK cells in terms of the outcomes of receptor activation. This is noteworthy because most studies on NK receptor function are conducted using IL-2-activated NK cell populations or clones. The polarized response of resting NK cells to signals received from 2DL4 vs other activation receptors is analogous to the distinct NK responses induced by different cytokines and reveals another facet in the complex regulation of these effector cells.