Indeed, as discussed above, IL-15 DCs do not bear any other NK cell-associated surface markers, such as NKG2D or NCRs. The mechanism underlying the ability of NK cells to induce U937 cell death has been recently identified as being Gomisin-D NCR-mediated, likely explaining the absent cytotoxic activity of IL-15 DCs against U937 cells. Another striking dissimilarity between IL-15 DCs and NK cells that merits further discussion is their differential pattern of cytotoxicity against the K562 cell line. While NK cells are strong and rapid inducers of K562 cell death, the anti-K562 cytotoxic activity of IL-15 DCs occurs only in the higher E:T range and with much slower dynamics. Interestingly, this intrinsically lower lytic potential has also been reported in other ‘killer DC’ studies and thus appears to be a common feature that distinguishes killer DCs from “classical” cytotoxic effector cells such as NK cells. The observation that IL-15 DCs display a distinct lytic profile further supports our view that these cells, despite the non-conforming expression of CD56, should be regarded as bona fide DCs endowed will killing potential and not as NK cells with antigen-presenting function. An important finding from this study is that IL-15 killer DCs do not induce cell death of tumor antigen-specific T cells, suggesting 
that their cytotoxic action is tumor-selective. This is especially noteworthy in view of recent data from Luckey et al., who showed that murine killer DCs are capable of eliminating allergen-specific T cells through a TNF-a-dependent mechanism and, as such, of preventing mice from developing allergic contact dermatitis. In line with this, murine CD8 + DCs have been previously shown to be capable of inducing T cell apoptosis through the Fas/FasL pathway. DC-mediated killing of T cells has also been demonstrated in the context of HIV infection. Evidently, the possibility of T cell killing would represent a major obstacle to the exploitation of killer DCs for cancer immunotherapy. Our data, however, indicate that T cell-directed cytotoxicity is not a general feature of killer DCs. This is consistent with the emerging view that killer DCs are a heterogeneous population, containing subsets that are preferentially tumoricidal as well as others that appear to be more biased toward a tolerogenic profile. This heterogeneity also applies to the different cytotoxic effector mechanisms that can be used by killer DCs. FasL and TNF-a, previously described as key components of the lytic armamentarium of killer DCs, are not found to be expressed on the IL-15 DC surface, thus arguing against their possible involvement in IL-15 DC-mediated killing. Although they lack membrane expression of TRAIL, IL-15 DCs �C in particular the CD56 + fraction �C harbor an internal pool of TRAIL molecules. Nevertheless, TRAIL neutralization resulted only in a marginal reduction of the lytic activity of CD56+IL-15 DCs against K562 cells, indicating that TRAIL is not a major contributor to the cytotoxic action of these DCs. This is in contrast to several other studies that implied an important role for this death receptor ligand in DC-mediated cytotoxicity. Our results point to Benzethonium Chloride granzyme B-induced apoptosis as the main cell death pathway used by IL-15 DCs. The presence of intracellular granzyme B deposits in IL-15 DCs was ascertained by direct gating on the DC population on the basis of a combination of scatter profile and CD11c positivity. The functional importance of this expression was further supported by the capacity of IL-15 DCs to release granzyme B extracellularly and ultimately confirmed.