Involvement of CD56brightCD11c+ Cells in IL-18–Mediated Expansion of Human γδ T Cells

γδ T cells are considered to be innate lymphocytes that play an important role in host defense against tumors and infections. We recently reported that IL-18 markedly amplified γδ T cell responses to zoledronate (ZOL)/IL-2. In an extension of this finding, we analyzed the mechanism underlying the IL-18–mediated expansion of γδ T cells. After incubation of PBMCs with ZOL/IL-2/IL-18, the majority of the cells expressed γδ TCR, and the rest mostly exhibited CD56brightCD11c+ under the conditions used in this study. CD56brightCD11c+ cells were derived from a culture of CD56intCD11c+ cells and CD14+ cells in the presence of IL-2 and IL-18 without the addition of ZOL. They expressed IL-18Rs, HLA-DR, CD25, CD80, CD83, CD86, and CD11a/CD18. In addition, they produced IFN-γ, TNF-α, but not IL-12, when treated with IL-2/IL-18, and they exerted cytotoxicity against K562 cells, thus exhibiting characteristics of both NK cells and dendritic cells. Incubation of purified γδ T cells with CD56brightCD11c+ cells in the presence of ZOL/IL-2/IL-18 resulted in the formation of massive cell clusters and led to the marked expansion of γδ T cells. However, both conventional CD56−/intCD11chigh dendritic cells induced by GM-CSF/IL-4 and CD56+CD11c− NK cells failed to support the expansion of γδ T cells. These results strongly suggest that CD56brightCD11c+ cells play a key role in the IL-18–mediated proliferation of γδ T cells.

H uman gd T cells exhibit a rapid response to microbial infections and tumors and serve as a bridge between innate and adaptive immunity. Although the precise mechanism underlying stress-surveillance responses has not been fully clarified, gd T cells are likely to be activated repeatedly by both common pathogens and autologous stress Ags (1)(2)(3)(4)(5)(6)(7)(8). In humans, Vg9Vd2 (also termed Vg2Vd2) gd T cells represent a major subset of circulating gd T cells and constitute 1-10% of total peripheral blood T cells (9). Vg9Vd2 T cells recognize phosphoantigens, like microbial (E)-4-hydroxy-3-methylbut-2-enyl pyrophosphate, an intermediate in the 2-C-methyl-D-erythritol-4-phosphate pathway (also known as the 1-deoxy-D-xylulose-5-phosphate pathway) (10) and isopentenyl pyrophosphate (IPP) in the self-mevalonate pathway (11). They are also activated in a NK receptor-mediated manner, but in a MHC-unrestricted manner (12). Recently, it was demonstrated that nitrogencontaining bisphosphonates (N-BPs) can stimulate peripheral blood gd T cells (13). The pharmacological agents inhibit farnesyl pyrophosphate synthase, leading to the accumulation of IPP in monocyte/dendritic cells (DCs), by which gd T cells can be efficiently activated (13)(14)(15). Moreover, the majority of tumor cells pretreated with N-BPs were demonstrated to stimulate gd T cells in a species-specific manner (16). Although the molecular mechanism of recognition has not yet been clarified, N-BPtreated monocytes and tumor cells appear to express membraneassociated antigenic determinants on their cell surface. In contrast, (E)-4-hydroxy-3-methylbut-2-enyl pyrophosphate and IPP may be presented to gd T cells in an extracellular pathway that requires neither Ag uptake nor Ag processing (17).
In immune/inflammatory responses, individual immune cells are activated in the context of a complex network, not in isolation. For instance, the activation and regulation of gd T cells are dependent on signals mediated by TCRs, as well as cosignaling molecules such as NKG2D, CD28, ICAM, and CD40L (18,19). These signals are usually provided by accessory cells such as monocytes and DCs and the various cytokines secreted by them (20,21). Although various subsets of DCs are present in peripheral blood and may play a role in the activation of gd T cells, DCs involved in gd T cell responses have not yet been extensively characterized (22,23).
In murine, a distinct subset of DCs exhibits a phenotype of NK cells and is called NKDCs (24)(25)(26)(27). The other subset of murine DCs exerts lytic activity and produces IFN-g (28). Human cells that correspond exactly to murine NKDCs have not yet been identified. Human DCs with NK-like character and other phenotypes can be generated from PBMCs in vitro (29)(30)(31)(32)(33). Recently, it was demonstrated that a subset of DCs exhibiting CD14 + CD56 + CD86 + HLA-DR + activated Th1-type gd T cells in human PBMCs (34), although the ontogeny of these cells (if they originated from NK cells, DCs, or monocytes) was not determined. A number of studies have demonstrated that DCs and innate immune cells activate reciprocally, suggesting that DCs with NK phenotypes may be generated by interaction among DCs, NK cells, and monocytes (35).
The cross-talk between DCs and innate lymphocytes, including gd T cells, might be carried out through cell-cell interaction and cytokine signaling. Various cytokines are involved in the development of DCs, and growth factors such as IL-1, IL-2, IL-7, IL-15, and IL-23 are required for the expansion of gd T cells (36,37). Other cytokines, including IL-12 (38), TNF-a (39), and TGF-b (40), are also responsible for the activation and regulation of gd T cells.
Recently, we demonstrated that IL-18 markedly promoted the expansion of gd T cells in a culture of human PBMCs (41). IL-18 was originally discovered as an IFN-g-inducing factor (42), and subsequent studies revealed that it also plays essential roles in host defense against infections and tumors, as well as in the pathogenesis of various inflammatory diseases through the upregulation of IFN-g production (43,44). Several recent studies have suggested that IL-18 plays a novel role in cellular events such as proliferation, differentiation, and survival (29,30,(45)(46)(47). In these reports, IL-18 was demonstrated to activate various antiapoptotic signals such as PI3K/Akt and Bcl-x L in both immune and nonimmune cells. The molecular mechanisms of IL-18-mediated signaling, however, remain to be clarified in various cells. IL-18 has also been demonstrated to function as an epigenetic regulator that enhances the expression of selected genes by methylating or acetylating chromatins (48). This may provide an explanation for the apparently paradoxical actions sometimes observed in IL-18.
In the current study, we demonstrate that IL-2/IL-18 efficiently generates and expands CD56 bright CD11c + cells, which functionally and phenotypically overlap with NK cells and DCs and are essential in the amplification of gd T cell responses to zoledronate (ZOL) or 2-methyl-3-butenyl-1-pyrophosphate (2M3BPP), an IPP analog. In addition, we characterize CD56 bright CD11c + cells and discuss a possible therapeutic use of ZOL/IL-2/IL-18 in the treatment of patients with cancer.

Cell migration assay
Chemotaxis was assayed by allowing gd T cells to pass through a polycarbonate filter with 5 mm pore size in 24-well Transwell chambers (Corning, Lowell, MA). CD3-depleted PBMCs (2 3 10 5 cells/0.6 ml) were precultured in the presence of GM-CSF (5 ng/ml)/IL-4 (20 ng/ml) or IL-2 (10 ng/ml)/IL-18 (100 ng/ml) for 8 d; they were then purified and placed in lower chambers. The purified gd T cells (2 3 10 5 cells/0.1 ml) were loaded into the upper chambers in the presence of ZOL/IL-2/IL-18 in culture. After incubation overnight at 37˚C, the cell suspension in the upper well was removed. The total number of cells in the lower chamber was counted by the trypan blue dye exclusion method, and the proportion of migrated gd T cells was determined by flow cytometry. Chemokine (CCL21; 20 ng/ ml)-induced cell migration was included as a positive control for migration in accordance with previously described methods (29). The number of spontaneously migrated cells was subtracted from the total number of migrated gd T cells, and the ratio of migrated cells to total cells was calculated as follows: migration (%) = ([migrated gd T cells 2 spontaneously migrated gd T cells]/total gd T cells) 3 100 (%).

Cytotoxicity assay by the DELFIA method
Cytotoxic activity was assayed using DELFIA EuTDA cytotoxicity reagents (PerkinElmer, Foster City, CA). NK-susceptible K562 cells were pretreated with a fluorescence-enhancing ligand BATDA, which was hydrolyzed by intercellular esterases to form a hydrophilic ligand, TDA, inside the cells. The labeled target cells were placed in 96-well plates at 1 3 10 4 cells/well in a RPMI 1640 medium with 2% FCS without phenol red. Effector cells, which had been induced by IL-2/IL-18 from CD3-depleted PBMCs, were added to the culture at an E:T ratio of 3:1, 10:1, and 30:1, and further incubated for 3 h. The culture supernatant was harvested and mixed with DELFIA europium solution. The europium and TDA in the supernatant formed a highly fluorescent and stable chelate (EuTDA), which was quantified by measuring the fluorescence signal. Specific cytotoxicity was determined using a fluorometer as follows: specific cytotoxicity (%) = ([sample release 2 spontaneous release]/[maximum release 2 spontaneous release]) 3 100 (%).

Assay of cytokines
IFN-g, TNF-a, and IL-12 were determined using ELISA kits obtained from BioLegend (San Diego, CA), according to the manufacturer's instructions.

Statistical analysis
Data are expressed as means 6 SD and analyzed using the Student t test or Bonferroni multiple comparisons test. A p value , 0.05 was considered significant.

Accumulation of CD56 bright CD11c + cells during IL-18-mediated expansion of gd T cells
When PBMCs were incubated in the presence of ZOL/IL-2 for 14 d, the number of gd T cells increased by .1000-fold, consistent with previous reports (1) (Fig. 1A). Moreover, the inclusion of IL-18 amplified the expansion of gd T cells, as we recently reported (41), whereas the addition of IL-18Ra blocking mAb strongly suppressed the proliferation of gd T cells, demonstrating the essential role of IL-18 in gd T cell expansion (Fig. 1A). It is of note that the addition of IL-18 promoted the formation of larger cell clusters compared with those triggered by ZOL/IL-2 (Fig. 1B). In addition, IL-18 also led to augmented gd T cell responses to 2M3BPP/IL-2 (Fig. 1A). Although there is an intrinsic dichotomy between N-BPs and pyrophosphomonoester compounds in the recognition mechanism (14,15), these results clearly illustrate that IL-18 can promote the expansion of gd T cells irrespective of the manner of Ag presentation. Fig. 1C presents representative dot plot diagrams of PBMCs from a healthy volunteer before and after stimulation with ZOL/ IL-2/IL-18. Whereas gd T cells were only 2% on day 0, a significant proportion of the cells was positive for gd TCR on day 14. It is intriguing that non-T cells occupied a significant proportion of the non-gd T cell population. We focused on the characterization of this non-T cell population to explore it further.
Freshly isolated PBMCs from a representative donor were divided into CD11c + and CD11c 2 cells. The CD11c + cells were further divided into two subpopulations based on the intensity of CD56: one expressing an intermediate level of CD56 (designated as CD56 int ), and the other lacking the expression of CD56. Thus, freshly prepared PBMCs contained CD56 int CD11c 2 , CD56 int CD11c + , and CD56 2 CD11c + cells (Fig. 1D, left panel). After culture with ZOL/IL-2, the number of CD56 2 CD11c + cells was significantly reduced, and the number of CD11c + cells expressing a high level of CD56 (designated as CD56 bright ) increased significantly ( Fig. 1D, middle panel). When PBMCs were incubated PBMCs incubated with ZOL/ IL-2 or ZOL/IL-2/IL-18 were analyzed for expression of CD11c, CD3, and CD56 on days 0 and 14. CD56 int CD11c 2 , CD56 int CD11c + , CD56 bright CD11c 2 , and CD56 bright CD11c + cells were gated, respectively. E, Time course of the expansion of gd T cells and CD56 bright CD11c + cells. PBMCs were incubated with ZOL/IL-2/IL-18, and the numbers of living gd T cells and CD56 bright CD11c + cells were quantified at indicated time points, as described above.
in the presence of ZOL/IL-2/IL-18 for 14 d, the majority of nongd T cells exhibited CD56 bright , half of which exhibited CD11c + (Fig. 1D, right panel). We designated this population as CD56 bright CD11c + cells. It was notable that the CD56 bright CD11c + cells increased in number in parallel with gd T cells after stimulation of PBMCs with ZOL/IL-2 or ZOL/IL-2/IL-18 under the conditions used in this study (Fig. 1E). The addition of blocking mAb instead of IL-18 reduced the expansion of both CD56 bright CD11c + cells and gd T cells (41) (data not shown).
Derivation of CD56 bright CD11c + cells in CD3-depleted PBMCs by IL-2/IL-18 We next analyzed the mechanism underlying the generation of the CD56 bright CD11c + cells that synchronously expanded with gd T cells in response to ZOL/IL-2/IL-18. First, the effect of the depletion of T cells on the expansion of CD56 bright CD11c + cells was examined. When CD3 + cells were depleted from PBMCs, 2-4% exhibited the phenotype of CD56 int CD11c + (Fig. 3A, left  panel). After stimulation with ZOL/IL-2 or ZOL/IL-2/IL-18 for 8 d, CD56 bright CD11c + cells occupied ∼60% of the total cells (Fig.  3A, middle and right panels). In addition, CD56 2 CD11c + cells were almost negligible in the expanded cells, whereas CD56 bright CD11c 2 cells comprised 20-26% of the total. This clearly demonstrates that the expansion of CD56 bright CD11c + cells in response to ZOL/IL-2 or ZOL/IL-2/IL-18 does not require the existence of gd T cells. Because ZOL is known to be a synthetic stimulant of gd T cells, CD3-depleted PBMCs were treated with various non-ZOL stimulants, including IL-2, IL-2/IL-18, IL-18, and GM-CSF/ IL-4. As expected, CD3-depleted PBMCs proliferated strongly in response to IL-2 and IL-2/IL-18 (Fig. 3B). In particular, IL-2/IL-18 elicited the vigorous proliferation of CD3-depleted PBMCs, in which the total cell number increased by ∼5-fold (Fig. 3B), and the number of CD56 bright CD11c + cells increased by 150-fold in 8 d (Fig. 3C).
As indicated in Fig. 2, CD14 + plus CD56 int CD11c + and/or CD14 + plus CD56 int CD11c 2 cell combinations are likely to play critical roles in the expansion of gd T cells in response to ZOL/IL-2/IL-18. We thus examined whether CD56 bright CD11c + cells can be derived from these cell combinations. CD14 + , CD56 int CD11c + , and CD56 int CD11c 2 cells were purified from PBMCs, and the mixed cell populations comprising CD14 + plus CD56 int CD11c + cells or CD14 + plus CD56 int CD11c 2 cells were incubated with IL-2/IL-18. As illustrated in Fig. 3D, CD56 bright CD11c + cells were effectively generated in the culture of CD14 + plus CD56 int CD11c + cells, but not in the combination of CD14 + plus CD56 int CD11c 2 cells. It may be of note that intensity of CD11c was decreased, whereas that of CD56 became brighter during culture (Fig. 3D, middle panel). We then analyzed CD14-bearing cells during the culture of PBMCs. In the freshly prepared CD3depleted PBMCs, cells expressing a high level of CD11c also expressed a high level of CD14 (Fig. 3E, left panel). These CD11c high CD14 bright cells occupied ∼30% of the initial culture of C, Expansion of CD56 bright CD11c + cells in CD3depleted PBMCs. The total number of CD56 bright CD11c + cells was determined using trypan blue dye exclusion and a FACS flow cytometer. *p , 0.05, **p , 0.001. D, Generation of CD56 bright CD11c + cells from the mixture of CD14 + and CD56 int CD11c + cells. CD14 + , CD56 int CD11c + , or CD56 int CD11c 2 cells were purified using a FACSAria sorting system. CD14 + cells were incubated alone or in combination with CD56 int CD11c + cells or CD56 int CD11c 2 in the presence of IL-2/IL-18. After 7 d, cells were analyzed for surface markers by flow cytometry. E, CD14 expression profiles during in vitro culture of CD3-depleted PBMCs. Freshly prepared CD3-depleted PBMCs and those incubated for 8 d with IL-2 or IL-2/ IL-18 were examined for the expression of CD11c, CD56, and CD14 using a FACSCalibur flow cytometer. CD56 int CD11c + , CD56 2 CD11c + , and CD56 bright CD11c + cells were gated and examined for expression of CD14. F, Effects of GM-CSF/IL-4 on CD3depleted PBMCs. CD3-depleted PBMCs were treated with GM-CSF (5 ng/ml)/IL-4 (20 ng/ml) and analyzed for the expression of CD14, CD80, CD86, and CD83 by a FACSCalibur flow cytometer. Whereas two CD11c high cell populations were gated separately, as indicated, essentially the same staining patterns were observed. The cells in both gates were thus analyzed together. CD3-depleted PBMCs. After culture for 8 d in the presence of IL-2 or IL-2/IL-18, however, cells expressing a high level of CD14 disappeared (49) (Fig. 3E, right panels). Thus, in the present experimental system, we failed to detect CD56 + CD14 + cells, which were recently reported to play a role in the expansion of gd T cells (34). Because it has been well established that conventional DCs can be induced by GM-CSF/IL-4, we analyzed the phenotype of conventional DCs as reference cells. When CD3-depleted PBMCs were treated with GM-CSF/IL-4, CD56 2 CD11c high CD14 + and CD56 int CD11c high CD14 + cells, but not CD56 bright CD11c + cells, were observed (Fig. 3F). Both CD56 2 CD11c high CD14 + and CD56 int CD11c high CD14 + cells expressed molecules such as CD80, CD83, and CD86, which were commonly observed in conventional DCs (Fig. 3F).

Characterization of CD56 bright CD11c + cells
Freshly isolated peripheral blood CD56 int CD11c + cells were positive for IL-18R aand b-chains (Fig. 4A). They also expressed NK cell-related molecules such as CD122/IL-2Rb (but not CD25/IL-2Ra) and NKG2D, and DC-related Ags such as HLA-DR, CD11a/CD18 (LFA-1). In addition, they expressed a low level of CD80 and CD83; CD86 expression was negligible (Fig. 4A). CD56 bright CD11c + cells that had been induced by IL-2 with or without IL-18 from CD3-depleted PBMCs expressed IL-18R aand b-chains (Fig. 4B, 4C). IL-2/IL-18 induced CD25 expression, but CD122 was significantly downregulated compared with freshly isolated CD56 int CD11c + cells. They also expressed molecules commonly displayed in mature DCs, such as CD83, CD80, and CD86, as well as HLA-DR and CD11a/CD18 (Fig.  4B). It is noteworthy that a high level of CD86 was detected in CD56 bright CD11c + cells stimulated with IL-2/IL-18, whereas only a marginal level of CD86 was detected in those incubated in the presence of IL-2 without exogenous IL-18 (Fig. 4B, 4C). As described above, the phenotypes of CD56 bright CD11c + cells were noticeably different from those of DCs induced by GM-CSF/IL-4 ( Fig. 3E), which exhibited CD11c high CD14 + . It is of note that the CD56 bright CD11c + cells were essentially negative for CD14. Although IL-2 or IL-2/IL-18 stimulation without ZOL was sufficient for the induction of CD56 bright CD11c + cells, the effect of ZOL on the expression of costimulatory molecules was also examined. As illustrated in Fig. 4D, ZOL enhanced the expression of DC-related molecules, including CD40, CD80, CD83, and CD86 in the cells stimulated by IL-2 alone or IL-2/IL-18, suggesting that it can act directly on CD56 bright CD11c + cells to induce cosignaling molecule expression.
Profiles of cytokine production and cytotoxic activity in CD56 bright CD11c + cells To further characterize CD56 bright CD11c + cells, the profiles of cytokine secretion were determined. Purified CD56 bright CD11c + and CD56 bright CD11c 2 cells were restimulated by IL-2 alone or IL-2/IL-18 for 48 h, and the cytokines in the culture supernatants were measured using ELISA. As illustrated in Fig. 5A, IL-2/IL-18 elicited a high level of IFN-g and TNF-a production compared with other stimulants, although none of the stimulants induced IL-12 production. These results indicate that CD56 bright CD11c + cells produce cytokines that are characteristic of NK cells.
Because the CD56 bright CD11c + cells exhibited NK-like phenotypes, including NKG2D expression (Fig. 4B, 4C) and perforin accumulation (data not shown), we next examined the cells for NK activity. After the incubation of CD3-depleted PBMCs in the presence of IL-2 and IL-18 for 6 d, CD56 bright CD11c + cells were purified and tested for cytotoxicity against K562. As illustrated in Fig. 5B, the cultured cells exerted a high level of cytotoxicity in an E:T ratio-dependent manner, demonstrating that CD56 bright CD11c + cells possess NK cell-like characteristics phenotypically and functionally.

Effect of CD56 bright CD11c + cells on gd T cell migration and expansion
It is thus clear that IL-2/IL-18 facilitates the expansion of CD56 bright CD11c + cells and that ZOL/IL-2/IL-18 elicits the vigorous expansion of gd T cells. In this context, we examined the possible role of CD56 bright CD11c + cells in the augmented proliferation of gd T cells in the presence of IL-18. After treatment with ZOL/IL-2 or ZOL/IL-2/IL-18, CCR5 and CCR7 (the receptors for CCL21) were induced in gd T cells (Fig. 6A). We then determined whether the chemokine/chemokine receptor system is operative between CD56 bright CD11c + cells and gd T cells. CD56 bright CD11c + and CD56 2/int CD11c high CD14 + cells were prepared from CD3-depleted PBMCs and placed in the lower wells, and then gd T cells were placed in the upper wells together with ZOL/IL-2/IL-18. As depicted in Fig. 6B, gd T cells migrated toward CD56 bright CD11c + cells generated in the culture with IL-2/ IL-18 (Fig. 6B), but not toward CD56 2/int CD11c high CD14 + cells, which were induced by GM-CSF and IL-4 (Fig. 3F). The migration rate was comparable to a positive control Transwell containing CCL21 instead of CD56 bright CD11c + cells. This finding indicates that it is likely that a chemokine/chemokine receptor system greatly promotes the direct interaction between CD56 bright CD11c + cells and gd T cells. Consistent with this finding, microscopic analysis revealed that coculture of CD56 bright CD11c + cells with gd T cells led to the formation of large cell clusters in the presence of ZOL/IL-2/IL-18 (Fig. 6C), both on the size and number of the cell clusters. In contrast, ZOL/IL-2/IL-18 failed to induce the cell aggregation in the culture of gd T cells plus conventional DCs exhibiting CD56 2/int CD11c high that had been induced by GM-CSF/IL-4. gd T cells alone also failed to form clusters in the milieu of ZOL/IL-2/IL-18.
We next investigated whether CD56 bright CD11c + cells can support the expansion of gd T cells through direct interaction.
When either CD56 bright CD11c + cells or gd T cells alone were cultured in the presence of ZOL/IL-2/IL-18, only a weak or absent proliferative response was observed by day 10. In contrast, vigorous expansion of cells was noted in the culture of gd T cells together with CD56 bright CD11c + cells, which was more prominent than that with CD56 bright CD11c 2 cells (Fig. 6D). On day 10, the majority of the expanded cells were gd T cells (88%). Although the initial proportion of the CD56 bright CD11c + cells was 30%, they decreased to ,10% after the culture (Fig. 6D). In contrast, CD56 bright CD11c 2 cells failed to induce the proliferation of gd T cells, and their proportion in the culture remained unchanged (Fig. 6D). Although the data are not shown, both CD56 int CD11c + cells and CD56 int CD11c 2 cells failed to induce the proliferation of gd T cells significantly, and IL-2 or IL-2/IL-18 induced neither proliferative responses nor cluster formation in the culture of FIGURE 5. Cytokine profiles and cytotoxic activity exhibited by CD56 bright CD11c + cells. A, Production of cytokines by purified CD56 bright CD11c + cells induced by IL-2/IL-18. CD3-depleted PBMCs were incubated with IL-2/IL-18 for 6 d, allowing CD56 bright CD11c + cells to occupy 70% or more of the cells. The CD56 bright CD11c + cells were further purified by a FACSAria sorting system and restimulated by IL-2 or IL-2/ IL-18 for 48 h. The culture supernatants were assayed for IFN-g, TNF-a, and IL-12. Assay sensitivity, as reported by the manufacturer, was 10 pg/ ml for IFN-g and TNF-a and 4 pg/ml for IL-12. B, NK-like activity exhibited by CD56 bright CD11c + cells. T cell-depleted PBMCs were cultured with IL-2/IL-18 for 6 d, as in A, and the resulting cells were assayed for cytotoxic activity against K562 cells, a human NK target. CD56 bright CD11c + cells plus gd T cells in the absence of ZOL or 2M3BPP. Taken together, these results indicate that CD56 bright CD11c + cells strongly facilitate the IL-18-mediated expansion of gd T cells stimulated by ZOL or 2M3BPP.

Discussion
IL-18 was originally discovered as a factor that induces IFN-g in immune cells (43). Recent studies have demonstrated that IL-18, like other members of IL-1 family cytokines, is converted to an active form in a stress-induced manner (49), and plays a critical role in the differentiation, proliferation, and survival of various cells (45)(46)(47)(48). The molecular mechanisms underlying the IL-18associated modulation of target immune cells have not yet been fully clarified, however. We previously reported that IL-18 directly activated various antiapoptotic signals in gd T cells and led to the expansion of gd T cell population (41). IL-18 is thus likely to be a cellular protective factor in gd T cells. Because evidence has been accumulating that gd T cells play a key role in the surveillance and first line of defense against infections and malignancies, IL-18 seems to play an important role in the modulation of stresssurveillance responses against infections and malignancies. Little is known, however, about the cellular and molecular mechanisms by which IL-18 indirectly, rather than directly, amplifies the expansion of gd T cells in response to phosphoantigens. In the current study, we demonstrated that CD56 bright CD11c + cells are closely linked with the IL-18-mediated expansion of gd T cells.
As reported previously, IL-18 greatly enhanced gd T cell responses to ZOL/IL-2 (41). It is of note that the addition of anti-IL-18Ra mAb to the culture of PBMCs with ZOL/IL-2 or ZOL/ IL-2/IL-18 strongly inhibits the proliferation of gd T cells. This . Involvement of CD56 bright CD11c + cells in the migration and expansion of gd T cells. A, Induction of costimulatory molecules in gd T cells. PBMCs were stimulated with ZOL/IL-2 or ZOL/IL-2/IL-18, and the expanded gd T cells were examined for costimulatory molecules by flow cytometry. B, Effect of CD56 bright CD11c + cells on migration of gd T cells. CD3-depleted PBMCs were incubated with GM-CSF/IL-4 or IL-2/IL-18 for 8 d. Purified cells (1 3 10 5 /well) were loaded in the lower wells. PBMCs incubated with GM-CSF/IL-4 exhibited CD56 2/int CD11c high CD14 + , and those stimulated with IL-2/IL-18 contained CD56 bright CD11c + . The purified gd T cells (1 3 10 5 /well) were loaded in the upper wells. As a positive control, CCL21 (20 ng/ml) was placed in the lower wells, as described (29). The number of migrated gd T cells from the upper well to the lower well was counted. The number of spontaneously migrated cells was subtracted from the total number of migrated gd T cells, and the ratio of the migrated cells to the total cells was calculated as follows: migration (%) = ([migrated gd T cells 2 spontaneously migrated gd T cells]/total gd T cells) 3 100 (%). **p , 0.001. C, Induction of cell cluster formation in the mixed culture of CD56 bright CD11c +enriched cells and gd T cells. CD56 bright CD11c + -enriched cells were combined with gd T cells that had been purified from freshly prepared PBMCs at a ratio of 1:1. After incubation for 18 h in the presence of ZOL/IL-2/ IL-18, the extent of cell aggregation was examined under a microscope. D, Effect of CD56 bright CD11c + cells on proliferation of gd T cells. T cell-depleted PBMCs were cultured with IL-2/IL-18 for 8 d, from which CD56 bright CD11c + cells were isolated by a FACSAria sorting system to a purity of .98%. Freshly isolated gd T cells, sorted CD56 bright CD11c + cells, and a mixture of freshly isolated gd T cells plus CD56 bright CD11c + cells were incubated in the presence of ZOL/IL-2/IL-18. At indicated time points, the number of cells was counted by trypan blue dye exclusion, and the living cells were analyzed by flow cytometry. clearly shows that IL-18 signaling is essential in the expansion of peripheral blood gd T cells in response to ZOL even in the absence of exogenous IL-18. Interestingly, the expansion of gd T cells triggered by a synthetic pyrophosphomonoester Ag, 2M3BPP, was also notably enhanced by IL-18, and blockade of the cytokine signaling abrogated the responses. It has been well established that ZOL permeates cell membranes, inhibits farnesyl pyrophosphate synthase, and consequently upregulates intracellular IPP in accessory cells (13). Such accessory cells may serve as APCs for gd T cells in this culture system. In contrast, pyrophosphomonoester Ags are presented to gd T cells through an as yet unidentified extracellular pathway even in the absence of third-party cells like monocytes and DCs, but adherent cells might be essential for the sustained expansion of gd T cells. These findings raise the possibility that IL-18 generates or stimulates CD56 bright CD11c + cells that can efficiently support the expansion of gd T cells. In fact, the number of CD56 bright CD11c + cells increased between 500-and 1000-fold in 10 d to 2 wk; this was paralleled by the expansion of gd T cells. In addition, most of the non-gd T cell populations were composed of CD56 bright CD11c + cells on day 14 in this culture system. It remains unclear whether ZOL-sensitized CD56 bright CD11c + cells can initiate gd T cell responses.
When CD14-depleted PBMCs were stimulated with ZOL/IL-2/ IL-18, a proliferative response was observed in neither CD56 bright CD11c + nor gd T cells (Fig. 2). This indicates that a relatively small number of naturally occurring or naive CD56 int CD11c + cells in freshly isolated PBMCs is insufficient for triggering gd T cell responses. In the current study, CD14 + cells were demonstrated to play an essential role in the derivation of CD56 bright CD11c + cells (Fig. 3C). Thus, it remains to be determined whether CD14 + cells indirectly support the proliferation of gd T cells by generating a more efficient initiator for gd T cell activation or whether they directly initiate the activation of gd T cells.
As depicted in Fig. 3E, CD14 is strongly expressed on CD56 2 CD11c + cells, whereas freshly prepared CD56 int CD11c + cells and expanded CD56 bright CD11c + cells exhibit CD14 2 . Thus, we could not detect a recently reported population of CD56 + CD14 + cells (34) in the present experimental system. Although the exact origin of the mature CD56 bright CD11c + cells remains speculative, the present results suggest that CD56 int CD11c + cells develop into CD56 bright CD11c + cells. If this is the case, CD14 + cells may render CD56 int CD11c + cells immunologically active and mature through IL-18 signaling and other as yet unidentified signaling, allowing gd T cells to proliferate vigorously. As supportive data, IL-18 promoted the expression of costimulatory and adhesion molecules such as CD56 and CD86 in CD56 bright CD11c + cells. In addition, CD56 bright CD11c + cells produced CCL21 in the presence of IL-18 (data not shown), and gd T cells expressed its receptors, CCR5 and CCR7. Thus, IL-18 appears to promote gd T cell proliferation in multiple ways, as follows: forming growth centers, rendering signals through costimulatory molecules, and directly activating survival signals. It is of note that IL-18 can promote the expansion of CD56 bright CD11c + cells even in the culture of CD3depleted PBMCs, suggesting that the proliferation of CD56 bright CD11c + cells is apparently independent of gd T cells. In addition, both CD56 int CD11c + and CD56 bright CD11c + cells express IL-18Rs (Fig. 4). It is therefore most likely that IL-18 acted directly on these cells. This finding, however, does not preclude the possible involvement of gd T cells in the maturation of CD56 bright CD11c + cells under other circumstances.
The present finding of the generation of CD56 bright CD11c + cells with immunological properties of both NK cells and DCs in humans was unprecedented. There have been several reports of NKDCs in mice that exhibit both NK-like and DC-like charac-teristics. The murine NKDCs seem to be heterogeneous, and some subsets were shown to be generated in the presence of IL-18 and CpG (25)(26)(27). In humans, however, the concept of NKDCs has not been generally accepted because researchers have failed to induce and find human counterparts of murine NKDCs (29)(30)(31). The present results unambiguously demonstrated the existence of the cells that phenotypically and functionally overlap with NK cells and DCs in humans and the immunological roles of this subset of cells in the efficient expansion of gd T cells. Whereas CD56 bright CD11c + cells express perforin and Fas (data not shown) and exhibit potent NK-like activity (Fig. 5), the physiological roles of the death molecule and the death receptor remain to be determined.
The mechanism by which accessory cells activate gd T cells has not yet been clarified (1). Generally, the activation of gd T cells is independent of conventional MHC class I, MHC class II, CD1a, CD1b, and CD1c molecules, which allow gd T cells to respond rapidly to infections and malignancies. Recently, CD14 + monocytes were demonstrated to play crucial roles in the activation of gd T cells. ZOL specifically inhibited the isoprenoid synthetic pathway and increased the intracellular level of IPP, as mentioned above (13). It was also reported that CD56 + CD14 + DCs were essential for the proliferation of Th1-type gd T cells (34). The CD56 + CD14 + cells expressed CD80, CD83, and CD86, and swiftly increased in size. In the current study, however, we failed to identify CD56 + CD14 + cells in our culture system. These findings raise the interesting possibility that several different accessory cells may be responsible for the expansion of gd T cells under different culture or physiological conditions.
Recently, considerable attention has been given to the potent tumoricidal activity of human Vg9Vd2-bearing gd T cells. Several attempts have been made to develop novel cancer immunotherapy using gd T cells. It has been reported that the adoptive transfer of gd T cells together with IL-2 and/or ZOL can have beneficial effects in patients with cancer (12). Although the precise mechanism has not been fully clarified, the addition of ZOL to endocrine therapy in the treatment of premenopausal patients with breast cancer significantly improved the disease-free survival rate (50). Because the number of effector cells is critical in cancer immunotherapy, it is imperative that a practical strategy to induce a large number of gd T cells in vitro as well as in vivo be developed. The present study demonstrates that the use of N-BPs/IL-2/IL-18 greatly facilitates the expansion of gd T cells. As such, future examination of the precise molecular and cellular mechanisms underlying the IL-18-mediated expansion of gd T cells is essential in the development of effective cancer treatment regimens.