**9. Adoptive transfer of donor natural killer cells and natural killer cell expansion**

**7. The effects of natural killer cells on graft-versus-host disease**

of donor CD4<sup>+</sup>

140 Natural Killer Cells

**receptor typing**

ligands on normal nonhematopoietic cells [155].

Allo-HSCT has curative potential through donor immune effectors-mediated GvL effects. However, the donor versus host direction effect of alloreactive T cells also causes GvHD which is a major complication of allo-HSCT [2]. Depletion of donor T cells in grafts and application of immunosuppressive pretreatment are often used to eliminate GvHD; however, these treatments also hampers GvL effects at the same time. NK cells have also been shown to play an important role in GvHD. In murine models, alloreactive NK cells are shown to kill antigen presenting cells (APCs) thereby prevent the initiation of aGvHD [100]. Correspondingly, another study demonstrated that alloreactive NK cells can be recruited into the lymph nodes and suppress T cells activation through cytotoxicity toward allogeneic dentritic cells (DCs) [149]. Donor NK cells may also mediate suppression of aGvHD through the direct killing of activated donor alloreactive T cells [150]. The possible mechanism involves the interactions between activating receptor NKG2D and its ligand highly expressed on the surface of activated T cells, which help NK cells to discriminate activated T cells from normal T cells and initiate killing [151, 152]. NK cells were also found to inhibit aGvHD through cytokine production such as TGF-β [153]. Furthermore, NK cells were able to inhibit the proliferation

T cells and reduce the risk of cGvHD [154]. Remarkably, donor NK cells are

believed not to attack normal host tissues mainly due to the absence of activating receptor

Although these previous experimental studies suggested that donor NK cells may have potential role in modulating GvHD, the clinical results are variable. Ruggeri L et al. reported NK cell alloreactivity was favorably associated with the lower risk of aGvHD [6]. However, this benefit effect cannot be observed in most groups, some groups even found KIR mismatch may worsen GvHD [156, 157]. Notably, in a recent clinical trial, IL-15/4-1BBL-activated donor NK cell infusion was found to contribute to aGvHD in T-cell depleted HSCT [158]. Collectively, these contradictory results indicated the role of alloreactive NK cells in GvHD may be variable. They could either aggravate or attenuate GvHD probably depended on different immune microenvironment in HSCT. Correspondingly, in our study of murine acute GvHD model, IL-12/15/18-preactivated donor NK cells significantly inhibited severe aGvHD, whereas they

accelerated the development of aGvHD in a mild aGvHD model (unpublished data).

Since alloreactive NK cells play a critical role in the successful treatment of leukemia patients receiving alloreactive grafts, it is crucial to ensure sufficient number of NK cells in the best possible donor. As we have discussed above, the polymorphisms of KIR and its ligand HLA obviously affect NK cell alloreactivity. Therefore, genotyping of these two families of genes provides the probability to identify the size of alloreactive NK cell populations in donor. Up to now, there are three levels of KIR typing. The first level is genotyping to test the gene content of KIR families [159]. Based on these results, the KIR haplotype of donor is confirmed and scored.

**8. Donor selection based on killer cell immunoglobulin-like** 

Despite their potential beneficial role in allo-HSCT, donor NK cells derived from the hematopoietic stem cells may be functionally impaired due to their immaturity and insufficient education. Thus, adoptively transfer of mature donor NK cells has been employed to provide the immunotherapeutic benefits for allo-HSCT. The first trial of utilizing allogeneic NK cells as immunotherapy in humans was published in 2005 focused on the impact of different chemotherapy preparative regimens on infused donor NK cells [163]. Three different chemotherapy preparative regimens were used in their study: high cyclophosphamide and fludarabine (Hi-Cy/Flu), low cylcophosphamide and methylprednisone, or fludarabine alone. Their results demonstrated that the infusion of NK cells was well tolerated by patients, donor-derived NK cells expansion can only be detected in patients receiving Hi-Cy/Flu. Furthermore, long-term remission was observed in 5 of 19 patients. An obvious increase in endogenous IL-15 concentration was detected in patients receiving high intensity of the conditioning regimen and was associated with the expansion of adoptive transferred donor NK cells. These results indicated the Hi-Cy/Flu treatment may eliminate recipient's lymphocytes thereby provide "space" and adequate cytokines for expansion of transferred donor NK cells. These results also suggested besides IL-2, IL-15 might be a good choice to promote NK cell expansion *in vivo*. In the follow-up study, the criteria for successful donor NK cell expansion were defined as a measurement of more than 100 donor NK cells/μl of peripheral blood 12–16 days after infusion [113].

Although remission of patients was observed in the above studies, this benefit effect was not permanent as patients ultimately relapsed. To test whether host-mediated rejection affected the expansion of donor NK cells thereby resulted in remission, total body irradiation was added to Hi-Cy/Flu to deplete recipient cells [138]. The results demonstrated that measurable NK cell expansion was obviously increased in patients received total body irradiation, which was associated with better leukemia clearance effect. These data again indicated the importance of enough "space" for successful expansion of donor NK cells during HSCT. In a pediatric study, Rubnitz et al. reported after high intensity of lymphodepeleting chemotherapy, infused donor NK cells could successfully expand in the patients of AML [164]. Remarkably, among 10 patients, three patients were observed to have detectable NK cells up to 1 month after infusion and have a disease-free survival rate of 100% at a median of 2 years. This promising outcome may due to the higher cell dose of transferred donor NK cells used in this study; thus, higher doses of infused NK cells may be important for better HSCT outcome.

To produce sufficient number of donor NK cells and achieve the optimal expansion of NK cells *ex vivo*, numbers of studies were performed with variable approaches, such as cytokine stimulation and utilizing engineered feeder cells. Up to now, clinical NK cell doses can be reached even up to 10<sup>8</sup> NK cells/kg [165]. However, there are numerous factors that may affect the effector function of expanded NK cells. A crucial factor among these complicated production protocol is how to activate NK cells *ex vivo*. Addition of cytokines such as IL-2 or IL-15 was used to promote survival and expansion of NK cells. Although IL-2 stimulation may result in activation-induced cell death (AICD) of NK cells *in vivo*, IL-15 stimulation does not have this side effect [166, 167]. Subsequently, human-derived feeder cells were used to develop approaches in expanding NK cells. Fujisaki et al. first reported the use of genetically modified APCs that could expand NK cells from peripheral blood 500- to 1000 fold [13]. These artificial APCs can be further modified with costimulatory molecules and membrane-bound cytokines [168, 169]. The expanded NK cells express high level of activating receptors and CD16 molecule and are proved to be potent mediators of cytotoxicity. However, these activated NK cells often become "exhausted" and cannot maintain their effector function and expansion *in vivo* [170]. Denman CJ et al. used APCs modified with membrane-bound form of IL-21 to break this barrier [168]. However, after cytokine withdrawal, these *ex vivo* expanded NK cells have been shown to have shorter survival *in vivo* compared with freshly activated NK cells [171].

Another alternative approach is to generate "memory-like" NK cells *ex vivo*. Cooper et al. reported that IL-12/15/18-preactivated NK cells obtained the ability to produce increased IFN-γ upon restimulation for up to 4 months after adoptive transfer [24]. Therefore, cytokine preactivation before infusion may amplify and sustain the beneficial effects of NK cells during allo-HSCT. The enhanced antitumor activity of murine IL-12/15/18-preactivated NK cells was first demonstrated in a murine tumor model [172]. Similar to murine NK cells, human IL-12/15/18-preactivated NK cells have been recently shown to have memorylike properties, including increased IFN-γ production, enhanced proliferation, and high expression of IL-2 receptor [26]. Recently, a phase I study of adoptively transferred cytokine-induced memory-like NK cells has demonstrated sustained anti-leukemia responses in patients with relapsed or refractory AML [173]. Therefore, cytokine-induced memory-like NK cells may possess long-lasting effector function *in vivo*, which can be harnessed in the treatment of leukemia patients. However, the safety of such therapy could be complicated by the induction of aGvHD after allo-HSCT. Adoptively transfer of donor-derived IL-15/4- 1BBL-activated donor NK cells contributed to aGvHD [158], likely through upregulation of activating receptor expression and inflammatory cytokine production. Therefore, the effect of IL-12/15/18-preactivated NK cell infusion on GvL and GvHD after allo-HSCT needs further investigation.
