A Critical Role for IL-18 in the Proliferation and Activation of NK1.1⁺CD3⁻ Cells¹

Natural killer cells have a critical role in the early phases of immune responses against various types of pathogens (1, 2). Namely, the capacities of NK cells to produce IFN-γ after microbial infection or to exhibit cytotoxicity against infected cells, before the clonal expansion and differentiation of Ag-specific T cells, is important in the innate immune response (3, 4).

Most facets of NK cell biology have been the expression of cytotoxicity, cytokine secretion, and IL-2-induced expansion (5). For example, NK cells exhibit enhanced cytolysis, IFN-γ secretion, and proliferative activity in response to IL-2 and represent the predominant IL-2-induced lymphokine-activated killer cell activity (6, 7, 8). However, in the last several years, additional cytokines have been found to dramatically regulate NK cell activity in vitro and in vivo. In particular, IL-12 has been shown to be a potent stimulator of mature NK and T cells (9, 10, 11, 12, 13, 14, 15, 16). The effects of this cytokine on both lineages of cells include: 1) production of IFN-γ and other cytokines (10, 13) requiring participation of accessory cells (13); 2) enhancement of cytotoxicity induced directly by IL-12 (14); and 3) proliferation of previously activated cells (15, 16). Thus, unlike IL-2, IL-12 induces little or no proliferation in resting NK and T cells. Moreover, IL-12 requires the presence of accessory cells for the stimulation of IFN-γ production. This may be consistent with the fact that IL-1β is required for IL-12 to induce IFN-γ production by NK cells (17). While IL-2 is a cytokine produced by activated T cells in an acquired immune response, little is known regarding how NK cell growth is supported in the innate immune response before T cell-mediated protective immunity develops. It also remains to be investigated whether additional cytokine(s) regulates the function of NK cells, alone or in combination.

Recently, another proinflammatory cytokine with the ability to markedly stimulate IFN-γ production by T cells and NK cells was discovered and designated IFN-γ-inducing factor (18), now termed IL-18. Our previous study showed a striking synergy between IL-12 and IFN-γ-inducing factor/IL-18 in enhanced production of IFN-γ by a cloned T cell line (19). The present study investigated the effect of IL-18 on the proliferation and activation of NK cells. The results show that IL-18, in combination with either IL-12 or IL-2, induced much higher levels of proliferation of NK1.1⁺CD3⁻CD4⁻CD8⁻ cells than did IL-2 or IL-12 alone, or the combination of IL-2 plus IL-12. Moreover, NK cells activated with either IL-12 + IL-18 or IL-2 + IL-18 expressed the capacity for strikingly high levels of IFN-γ secretion and exhibited potent cytotoxicity against various tumor cells. Such an enhanced NK activity was not induced by either the combination of IL-12 or IL-2 with various other cytokines including IL-1β or by various combinations of cytokines excluding IL-18. Thus, these results indicate an obligatory role for IL-18 in mediating the growth promotion and enhanced function of NK cells.

C57BL/6 (B6) mice and B6 (nu/nu) mice were purchased from Shizuoka Laboratory Animal Center, Hamamatsu, Japan, and CLEA JAPAN, Kanagawa, Japan, respectively. Animals were used at 6 to 7 wk of age.

The following recombinant mouse cytokines were obtained: rIL-12 and rIL-18 (previously described as IFN-γ-inducing factor) were expressed and purified at Hayashibara Biochemical Laboratories as previously described (20, 21); rIL-2 was kindly provided by Shionogi Research Laboratories, Osaka, Japan; rIL-1β was purchased from Genzyme, Cambridge, MA.

The following mAbs or polyclonal Ab were used: anti-I-A^d/b (34-5-3S) (22) and anti-CD16/32 (FcγR III/II) (23) were prepared from the culture supernatants of the relevant hybridoma cells; and anti-CD4 (ATCC clone GK1.5), anti-CD8 (ATCC clone 2.43), and anti-IL-2R α-chain (7D4) (24) were prepared from the ascitic fluid of the relevant hybridoma cells and purified in our laboratory; phycoerythrin (PE)³-conjugated anti-CD4, FITC-conjugated anti-CD8, FITC-conjugated anti-CD4, PE-conjugated anti-B220, FITC-conjugated anti-CD5, PE-conjugated anti-NK1.1, FITC-conjugated anti-Mac-1, FITC-conjugated anti-CD3ε, FITC-conjugated anti-Thy-1.2, and biotinylated anti-Sca-1 mAbs were purchased from PharMingen, San Diego, CA; anti-IL-2R α-chain mAb was biotinylated in our laboratory; rabbit anti-mouse asialo-GM₁ Ab was from Wako Pure Chemicals, Osaka, Japan; biotinylated goat anti-rabbit IgG (Wako Pure Chemicals), biotinylated mouse anti-rat IgG (Jackson ImmunoResearch, West Grove, PA), PE-conjugated streptavidin (Becton Dickinson, San Jose, CA), and RED670-conjugated streptavidin (Life Technologies, Gaithersburg, MD) were also purchased.

Spleen cells were depleted of CD4/CD8 T cells, B cells, and Ia⁺ APCs by immunomagnetic negative selection as described (25). Briefly, Ia⁺ APC in a splenic cell population were allowed to react with anti-I-A^d/b mAb. Spleen cells containing these labeled cells and surface Ig⁺ (sIg⁺) cells (B cells) were incubated with magnetic particles conjugated to goat anti-mouse IgG (Advanced Magnetic, Cambridge, MA). sIg⁻ and Ia⁻ cells (B cell- and APC-depleted population) were obtained by removing cell-bound magnetic particles with a rare earth magnet (Advanced Magnetic). CD4⁺ and CD8⁺ T cells were depleted by incubation with anti-CD4 and anti-CD8 mAb followed by magnetic particles conjugated to goat anti-rat IgG (Advanced Magnetic).

CD3⁻NK1.1⁺ and CD3⁻NK1.1⁻ cells were sorted from the above CD4⁻CD8⁻sIg⁻Ia⁻ splenic population by EPICS Elite (Coulter, Miami, FL). Briefly, cells were incubated with PE-conjugated anti-NK1.1 and FITC-conjugated anti-CD3ε, and the stained cells were sorted into CD3⁻NK1.1⁺ and CD3⁻NK1.1⁻ populations. The purities of these populations were >96 and 94%, respectively.

Splenic cell populations were cultured with various combinations of cytokines in 96-well flat-bottom microculture plates (Corning 25860, Corning Glass Works, Corning, NY) or in 24-well culture plates (Corning 25820). For determination of [³H]TdR uptake, cells were cultured in 0.2 ml of RPMI 1640 medium supplemented with 10% fetal bovine serum and 2-ME at 4.0 × 10⁴ (fresh spleen cells) or 1.0 × 10⁴ (blasts harvested from primary cultures) cells/well of 96-well microculture plates in a humidified atmosphere at 5% CO₂ at 37°C for various days. The cultures were conducted in triplicate and harvested after an 8-h pulse with 20 kBq/well [³H]TdR. Results were calculated from uptake of [³H]TdR and expressed as the mean cpm ± SD of triplicate cultures. For assays other than [³H]TdR incorporation, cells were cultured in 24-well culture plates in a volume of 2 ml at 2.5 × 10⁵ (fresh spleen cells) or 1.0 × 10⁵ (blasts harvested from primary cultures) cells/well. Culture supernatants (SNs) and cells were harvested after various days in culture.

IFN-γ concentration was measured by ELISA as described (19): mouse IFN-γ ELISA kits were purchased from Genzyme; and our own ELISA system was prepared using two types of anti-mouse IFN-γ mAbs (XMG1.2 (Endogen, Cambridge, MA)) and biotinylated R4-6A2 (R4-6A2 was purified from R4-6A2 hybridoma and biotinylated in our laboratory)) as well as mouse rIFN-γ provided from Shionogi, Osaka, Japan. One unit/ml of IFN-γ in our ELISA system corresponded to ∼100 pg/ml in Genzyme ELISA kits.

Target cells were ⁵¹Cr labeled and incubated with various effector populations for 4 h. Percent specific lysis was calculated as previously described (26). SE of the means were excluded from the figures, because they were consistently <5%.

Cells were first incubated with anti-CD16/32 (FcγR III/II) mAb and normal mouse serum to prevent the binding of staining mAbs with FcRs. These treated cells were then stained directly with FITC- or PE-conjugated reagents or sequentially with biotinylated mAb followed by PE- or RED670-conjugated streptavidin. Stained cells were analyzed with a FACSCalibur (Becton Dickinson, Mountain View, CA).

IL-12 and IL-18 both have the capacity to stimulate both T and NK lineages of cells (9, 10, 11, 12, 13, 14, 15, 16, 18, 19, 20). We first examined whether these cytokines, alone or in combination, can induce proliferation of a splenic population obtained following elimination of sIg⁺ and Ia⁺ cells (B cells and APC) from B6 spleen cells (Fig. 1,A, left). Neither cytokine alone elicited [³H]TdR uptake by a B cell/APC-depleted splenic population during the entire culture period. However, simultaneous stimulation with IL-12 and IL-18 resulted in a striking proliferative response. To examine whether T cells are the responders to IL-12 + IL-18 stimulation, B cell + APC-depleted spleen cells from B6 athymic mice were stimulated with either cytokine or their combination (Fig. 1,A, right). These spleen cells also exhibited marked proliferation when stimulated with the two cytokines. Proliferating cells harvested from the above two cultures were stained doubly for CD4 and CD8. Lower panels of Figure 1 show that almost all proliferating cells from athymic mouse spleen cells are CD4⁻CD8⁻ (double negative (DN)) and DN cells are also the predominant blasts in cultures of B cell/APC-depleted B6 splenic populations. These results suggest that a splenic population depleted of CD4/CD8 T cells, B cells, and Ia⁺ APC responds directly to a combination of IL-12 + IL-18 with striking levels of proliferation.

B6 spleen cells were depleted of CD4⁺ and CD8⁺ T cells together with B cells and APC. As shown in Figure 2, the resultant population designated CD4⁻CD8⁻sIg⁻Ia⁻ consisted of NK1.1⁺CD3⁻, NK1.1⁻CD3⁻, NK1.1⁺CD3⁺, and NK1.1⁻CD3⁺ subsets. This CD4⁻CD8⁻sIg⁻Ia⁻ population was stimulated with various cytokines, alone or in combination (Fig. 3). The results show that stimulation with IL-12 + IL-18 again induced marked levels of proliferation, and comparable levels of proliferation were elicited when IL-18 was combined with IL-2 instead of IL-12. The doses of IL-12, IL-2, and IL-18 required to induce synergistic proliferation were titrated. These cytokines exerted their effects in the ranges of 1 to 1000 pg/ml for IL-12, 1 to 100 U/ml for IL-2, and 1 to 1000 ng/ml for IL-18, and their optimal doses were found to be 250 to 500 pg/ml (IL-12), 20 to 100 U/ml (IL-2), and 10 to 100 ng/ml (IL-18). Under the culture conditions used here (2.5 × 10⁵ cells/well in 24-well culture plates or 4 × 10⁴ cells/well in 96-well microplates), individual cytokines including IL-2 failed to generate sufficient numbers of blasts and to induce [³H]TdR uptake (data not shown). When IL-2 was combined with IL-12, significant levels of proliferation were detected at later time points (8 or 9 days after culture) (Fig. 3). However, such proliferative responses were much weaker compared with those induced by cytokine combinations involving IL-18. Considering the structural similarity between IL-18 and IL-1α or IL-1β, the stimulatory capacity of IL-12 + IL-1α or β or IL-2 + IL-1α or IL-1β was also examined. IL-1α or IL-1β could not induce detectable levels of proliferation even when combined with IL-12 and IL-2 (data not shown). Thus, these results indicate that IL-18 plays an obligatory role in inducing high levels of proliferation of a CD4⁻CD8⁻sIg⁻Ia⁻ splenic population in collaboration with either IL-12 or IL-2.

We next analyzed surface phenotypes of cells proliferating in response to IL-12 + IL-18 or IL-2 + IL-18. Figure 2, right, demonstrates that blasts generated in response to IL-12 + IL-18 from a CD4⁻CD8⁻sIg⁻Ia⁻ population containing various NK1.1^+/− CD3^+/− subsets are exclusively NK1.1⁻CD3⁻, whereas most blasts induced by stimulation with IL-2 + IL-18 are NK1.1⁺CD3⁻. Appreciable percentages of CD3⁺ (NK1.1⁺ or NK1.1⁻) cells were present in the population before stimulation, but only a small percentage of CD3⁺ blasts were generated when cultured in the presence of IL-2 + IL-18. Figure 4 shows additional phenotypes of these IL-12 + IL-18- and IL-2 + IL-18-induced blasts. Another NK marker, FcγR lIl/ll (CD16/32) was detected only on IL-2 + IL-18 blasts in parallel to the NK1.1 positivity. Thy-1.2 and asialo-GM₁ as well as lymphoid activation markers (Sca-1 and B220) were observed on both types of blasts, whereas both were negative for T cell markers including CD3, CD4, CD8, and CD5. Thus, stimulation with IL-12 + IL-18 or IL-2 + IL-18 generates phenotypically distinct blasts regarding the expression of NK markers.

We investigated whether or not NK1.1⁺ and NK1.1⁻ blasts generated by stimulation with different combinations of cytokines represent two subsets that have differentiated into mutually independent lineages of cells (Fig. 5). A CD4⁻CD8⁻sIg⁻Ia⁻ population was cultured in the presence of IL-12 + IL-18 or IL-2 + IL-18 for 3 days. Cells harvested at this time point were again exclusively NK1.1⁻CD3⁻ in the IL-12 + IL-18 group, and ∼50% NK1.1⁺CD3⁻ in the IL-2 + IL-18 group. These two groups of cells were restimulated with each combination of cytokines for an additional 3 days, resulting in four subgroups. When the IL-12 + IL-18 group of cells was again cultured in the same combination of cytokines, cells remained NK1.1⁻, and likewise, stimulating the IL-2 + IL-18 group of cells with IL-2 + IL-18 increased the proportion of NK1.1⁺ cells. In contrast, cultures of these two groups of cells in the presence of the alternative combination of cytokines resulted in the generation of populations that reversed back to the original phenotype in terms of NK1.1 positivity. These results suggest that NK1.1⁺CD3⁻ and NK1.1⁻CD3⁻ populations are cells of the same lineage, and their expression of the NK1.1 marker is reversible depending on the type of cytokine combined with IL-18.

We also investigated whether the two types of blasts (NK1.1⁺CD3⁻ and NK1.1⁻CD3⁻) are derived from the same or different subsets in the original CD4⁻CD8⁻sIg⁻Ia⁻ population (Fig. 6). To do this, the NK1.1⁺CD3⁻ and NK1.1⁻CD3⁻ subsets in a CD4⁻CD8⁻sIg⁻Ia⁻ population were isolated using a cell sorter. These two original subsets of cells were stimulated for 4 days with IL-12 + IL-18 or IL-2 + IL-18. No blast was generated from the NK1.1⁻CD3⁻ subset after stimulation with either combination of cytokines, whereas the NK1.1⁺CD3⁻ subset generated NK1.1⁻CD3⁻ or NK1.1⁺CD3⁻ blast cells depending on the type of the cytokine combination. These blasts harvested on day 4 were subjected to the following experiments. NK1.1⁻CD3⁻ blasts remained NK1.1⁻ after restimulation with IL-12 + IL-18, whereas a portion of these blasts became NK1.1⁺ after restimulation with IL-2 + IL-18, which is consistent with the results of Figure 5. A blast population generated 4 days after stimulation with IL-2 + IL-18 from the original CD3⁻NK1.1⁺ subset contained a large portion of NK1.1⁺ and a small portion of NK1.1⁻ blasts. These two blast subsets were resorted, and each isolated subset was restimulated with IL-12 + IL-18 or IL-2 + IL-18. From the NK1.1⁻ blast subset, similar patterns of blasts to those observed for the blasts induced initially with IL-12 + IL-18 were generated following restimulation. Restimulation of the NK1.1⁺ subset with IL-12 + IL-18 or IL-2 + IL-18 induced the generation of blasts that were predominantly NK1.1⁻ or NK1.1⁺, respectively. These results indicate that: 1) the NK1.1⁻CD3⁻ subset contained in a splenic CD4⁻CD8⁻sIg⁻Ia⁻ population is not the precursor for either NK1.1⁺CD3⁻ or NK1.1⁻CD3⁻ blasts; 2) both types of blasts are generated from the original NK1.1⁺CD3⁻ spleen cell subset; and 3) unlike NK1.1⁻CD3⁻ cells in the fresh spleen cell population, NK1.1⁻CD3⁻ blasts induced with IL-12 + IL-18 can convert back to NK1.1⁺ cells in the presence of IL-2 plus IL-18 or proliferate continuously without expressing NK1.1 in the presence of IL-12 plus IL-18.

The NK1.1⁺CD3⁻ cells present in a freshly prepared splenic population failed to proliferate in response to any single type of cytokine. We examined whether the blasts induced by IL-12 + IL-18 or IL-2 + IL-18 can proliferate by stimulation with cytokines, especially with a single cytokine. Both types of blasts were prepared 3 or 4 days after stimulation of a CD4⁻CD8⁻sIg⁻Ia⁻ population and restimulated with various cytokines. The results of Figure 7 show that both types of blasts respond to either IL-12 + IL-18 or IL-2 + IL-18 stimulation. Regarding stimulation with a single cytokine, IL-2 induced both types of blasts to proliferate, but IL-12 did not. IL-18 alone exhibits significant levels of growth-promoting effects, especially on IL-12 + IL-18 blasts.

Portions of the same blasts generated in Figure 7 were restimulated with various cytokines for 1–3 days and culture SNs were assayed for IFN-γ concentrations. The results are shown in Figure 8. Both IL-12 + IL-18 and IL-2 + IL-18 blasts produced enormously large amounts of IFN-γ exclusively when they were restimulated with IL-12 + IL-18. Although IL-12 + IL-18 blasts restimulated with IL-2 + IL-18 exhibited appreciably higher levels of IFN-γ production than groups stimulated with either cytokine alone, these levels were markedly low compared with those observed for both types of blasts restimulated with IL-12 + IL-18. Because either IL-12 or IL-18 alone induced marginal levels of IFN-γ production, these two cytokines can induce strikingly high levels of IFN-γ production only through their collaboration. We also investigated whether other cytokines such as IL-1α or IL-1β in combination with IL-2 or IL-12 can induce IFN-γ secretion from IL-12 + IL-18- or IL-2 + IL-18 blasts. However, such combined stimulation induced only weak (<100 U/ml) IFN-γ secretion compared with the IL-12 + IL-18 stimulation (data not shown).

To determine the distinct capacity of IL-12 + IL-18 to induce IFN-γ production, we examined IFN-γ production during primary stimulation of NK1.1⁺CD3⁻ cells with IL-12 + IL-18 or IL-2 + IL-18. Figure 9 shows that the combination of IL-12 + IL-18 elicited marked levels of IFN-γ production even during primary culture. Stimulation with IL-2 + IL-18 also induced appreciable levels of IFN-γ production on day 4, which may be consistent with the fact that this combination of cytokines had the capacity to stimulate IL-12 + IL-18-induced blasts for IFN-γ production (Fig. 8). However, the levels of IFN-γ production induced by IL-2 + IL-18 in both primary (Fig. 9) and secondary (Fig. 8) cultures were much lower than those induced by IL-12 + IL-18. Thus, IL-12 and IL-18 are the prominent combination in terms of stimulation of IFN-γ production by NK cells.

We finally examined the cytotoxic activity of IL-12 + IL-18 and IL-2 + IL-18 blasts. Blasts generated 4 days after stimulation of CD4⁻CD8⁻sIg⁻Ia⁻ cells with either IL-12 + IL-18 or IL-2 + IL-18 were assayed for their cytolytic activity against various tumor target cells (Fig. 10). Both types of blasts exhibited different degrees of cytotoxicity depending on the target cell types. However, there was no substantial difference in the cytolytic activity between two types of blasts. This was also the case when the cytotoxicity was compared between Fas-transfected and wild-type targets.

The results obtained in this study demonstrate that IL-18 plays a pivotal role in the proliferation and activation of NK cells. This cytokine alone produces almost no effect on NK1.1⁺CD3⁻ cells but allows them to proliferate vigorously when combined with IL-12 or IL-2. Both IL-12 + IL-18- and IL-2 + IL-18-induced blasts exhibit potent cytotoxicity on various tumor targets and produce enormous amounts of IFN-γ after restimulation with IL-12 + IL-18. The results also show that while IL-12 + IL-18- and IL-2 + IL-18-blasts derive from a NK1.1⁺CD3⁻CD4⁻CD8⁻ population, the former type of blast lacks the NK1.1 marker. However, the expression of the NK1.1 marker on the two types of blasts is reversible depending on whether they are restimulated with IL-12 or IL-2, together with IL-18. Thus, our present results provide several new findings and/or important implications, most of which are based on the roles of IL-18.

Regarding the proliferation of NK cells (CD3⁻NK1.1⁺), IL-2 has been proposed to be the principal growth factor. Although IL-2 by itself can stimulate NK cell proliferation (6, 8), IL-2 responsiveness of freshly isolated human NK cells varies greatly depending on the NK cell subset divided by the CD56 expression. Whereas CD56^bright NK cells, a minor population of peripheral blood NK cells, respond potently to IL-2, IL-2 alone induces little proliferation for the CD56^dull subset comprising ∼90% of human NK cells (27, 28, 29). Consistent with this, freshly isolated CD3⁻NK1.1⁺ cells in the mouse proliferate poorly in response to IL-2 alone (30). Thus, costimulatory signals have been shown to be required for optimal proliferation of human (CD56^dull) and mouse (CD3⁻NK1.1⁺) NK cells, including a CD28-mediated signal (30) and soluble factors such as IL-1 and TNF (31). Nevertheless, the levels of NK cell proliferation enhanced by such costimulatory signals were not high. In fact, we failed to obtain sufficient numbers of blasts for analyses when a CD4⁻CD8⁻sIg⁻Ia⁻ population as a source of CD3⁻NK1.1⁺ cells was stimulated with IL-2 + IL-1α or β instead of IL-2 + IL-18. Moreover, vigorous proliferation of NK1.1⁺CD3⁻ cells induced by IL-12 + IL-18 was not prevented by anti-IL-2 mAb (our unpublished observations), excluding the possibility that endogenous IL-2 is induced by IL-12 + IL-18 stimulation to act as the major growth factor. Thus, by showing that high levels of NK1.1⁺CD3⁻ cell proliferation are induced by the IL-18-based cytokine combinations, the present study provides the first evidence for an obligatory role of IL-18 in the proliferation of NK1.1⁺CD3⁻ cells.

In the mouse, NK cells of appropriate strains has been identified as CD3⁻NK1.1⁺ (32). Two sets of the IL-18-based cytokine combinations (IL-12 + IL-18 and IL-2 + IL-18) induced the proliferation of this subset present in spleens. Interestingly, NK cells once activated with IL-12 + IL-18 lost the NK1.1 marker, whereas most of IL-2 + IL-18-activated NK cells still expressed the NK1.1 marker. However, the expression of NK1.1 was reversed when blasts activated with these two cytokine combinations are restimulated with the alternative combination of cytokines. Both types of activated NK cells exhibited almost the same levels of functions including cytotoxicity and IFN-γ production. Therefore, it appears that the expression of the NK1.1 marker correlates neither with the differentiation/activation stages of NK cells nor with the expression of its function.

The most prominent function of NK cells stimulated with the IL-18-based cytokine combinations would be to produce enormous amounts of IFN-γ during the primary or secondary stimulation with IL-12 + IL-18. Previously, it was shown that IL-2 stimulates IFN-γ production by T and NK cells (5, 6, 33). However, in the last several years, IL-12 has been found to dramatically promote IFN-γ secretion from both mouse and human NK cells (10, 11, 12, 13). IL-12 by itself induces IFN-γ secretion and can also act in synergy with IL-2 and TNF-α (34). While IL-12 has been regarded as the strongest inducer of IFN-γ secretion, another potent inducer designated IL-18 was more recently discovered (18). Thus, two powerful reagents capable of stimulating IFN-γ production had still to be investigated for their collaborative actions.

Unlike the up-regulating effect on cytotoxicity, the effect of IL-12 on IFN-γ induction requires the presence of accessory cells (13). This may be compatible with the observations that IL-1β is required for IL-12 to induce IFN-γ production by human (35) and mouse (17) NK cells. Thus, like the induction of NK cell proliferation, costimulatory signals derived from accessory cells are necessary for IFN-γ production by NK cells. In this context, it should be noted that IL-12 induced a dramatically high level of IFN-γ secretion in combination with IL-18 that is derived from a macrophage lineage of cells such as Kupffer cells (18). Because a combination of IL-12 + IL-1β induced much lower levels of IFN-γ secretion by NK cells compared with the combination of IL-12 + IL-18, the latter would represent the strongest cytokine combination for stimulating IFN-γ secretion.

Our ELISA systems using recombinant IFN-γ as the standard preparation have detected surprisingly high levels of IFN-γ production by NK cells following stimulation with IL-12 + IL-18. The possibility exists that the titers detected may be inflated in the present assays. Nevertheless, the comparison of the present results with the following observations supports the distinguished capacity of NK cells to produce IFN-γ. In our preceding paper (36), we measured the titers of IFN-γ produced by T cells using the same ELISA systems as those used here. We found that T cells activated with anti-CD3 plus anti-CD28 mAbs produce 50 to 100 U/ml of IFN-γ in response to IL-12 and IL-18. Therefore, it is obvious that the levels of IFN-γ produced by NK cells in response to IL-12 + IL-18 are extraordinarily high compared with those produced by T cells.

Synergy between IL-12 and IL-18 for enhanced production of IFN-γ was shown for a cloned T cell line, 2D6, in our previous study (19). The results demonstrated that 1) the 2D6 T cell clone constitutively expressed IL-12R; 2) IL-12 induces the expression of IL-18R on the T cell clone; and 3) enhanced IFN-γ production is elicited when the IL-18R-induced 2D6 clone is stimulated with IL-18. Because IFN-γ secretion by this clone induced following stimulation with IL-12 alone was much weaker than that with IL-12 + IL-18 (19), the induction of IL-18R by IL-12 was considered to represent a mechanism underlying the synergy between IL-12 and IL-18 in enhanced IFN-γ production. Our previous study further investigated whether this synergy is also observed for naive T cells (19). In contrast to the 2D6 clone, naive T cells express neither IL-12R nor IL-18R. Therefore, freshly isolated naive T cells fail to respond to IL-12 and/or IL-18. However, stimulation of purified T cells with anti-CD3 plus anti-CD28 mAb induced IL-12R expression, and these activated T cells were stimulated with IL-12 and IL-18 to exhibit the above mechanism of synergy for enhanced IFN-γ secretion.

In the present study, NK1.1⁺CD3⁻ cells could proliferate in response to the combination of IL-12 + IL-18 or IL-2 + IL-18. However, strikingly high levels of IFN-γ secretion was induced only when freshly prepared NK1.1⁺CD3⁻ cells or blasts activated with either of the two cytokine-combinations were stimulated with IL-12 + IL-18. The fact that NK1.1⁺CD3⁻ cells can respond to IL-12 + IL-18 without activation with any stimulus such as anti-CD3 differs from the failure of naive resting T cells to respond to these cytokines. Neither IL-12R nor IL-18R was detected on preactivated and activated NK cells in our previously described assay system (19) using ligands (IL-12 and IL-18) and anti-ligand Abs (our unpublished observations). While fresh NK1.1⁺ CD3⁻ cells failed to show IL-18 responsiveness, day 4 blasts obtained after stimulation with IL-12 + IL-18 exhibited the proliferative response to IL-18 alone. Although these results suggest that the expression of IL-18R is up-regulated on the activated (blast) NK population, IL-18R was not detected even on such a blast population (our unpublished observations). Regarding IL-12R expression, it is possible that at least IL-12R is expressed on freshly isolated NK1.1⁺CD3⁻ cells as has been described for resting human NK cells (37). By combining this possibility with the fact that IL-12 has the capacity to induce the expression of IL-18R on T cells (19), it may be assumed that IL-12 first induces NK1.1⁺CD3⁻ cells to express IL-18R and then these cells can respond vigorously to the combined stimulation of IL-12 + IL-18. This is consistent with the concept that IL-12 + IL-18 is the most potent cytokine combination for IFN-γ production.

Our present results illustrate that IL-18 exerts its striking effects on the proliferation and activation of NK1.1⁺CD3⁻ cells when combined with another proinflammatory cytokine, IL-12. Considering that significant numbers of NK1.1⁺CD3⁻ cells reside in the liver and IL-18 is produced efficiently by Kupffer cells, NK cell activation induced by IL-12 + IL-18 may represent a molecular mechanism underlying the development of inflammatory responses in the liver. It may also be possible to control various diseases by regulating the collaborative action of IL-12 and IL-18.

We thank Dr. M. Micallef for critical reading of the manuscript and Miss T. Katsuta and Miss M. Yamanaka for secretarial assistance.