HLA-E up-Regulation on IFN-γ Activated AML Blasts Impaired CD94/ NKG2A-Dependent NK Cytolysis Following Haplo-Mismatched Haematopoietic Stem-Cell Transplantation

We previously reported that NK-cell generated after haploidentical hematopoietic stem-cell transplantation (SCT) in high risk AML patients are characterized, during the first 6 months post-SCT, by specific phenotypic features and impaired functioning having potential impact for transplantation outcome (Nguyen et al, Blood 2005). In the present study, we examined the impact of INF-γ on the impaired recognition of AML blasts by NK cells after haplo-SCT. We studied NK cell reconstitution at 1 to 6 months (M1–M6; n=21patients) or at 24 months (M24; n=3 patients) post transplant. Patients received SCT from haplo-mismatched related donor after a myeloablative conditionning regimen for high risk malignant hematological diseases. As previously described, the level of CD56^brightCD94/NKG2A⁺ expression on NK cells was increased, around 3.5 higher at M1 and M6 as compared to the donors, and NK cytotoxicity against AML blasts was poorer than in the donors, suggesting an immature status of NK cells post haplo-SCT. However, at M24, similar level of CD56^brightCD94/ NKG2A⁺ was observed in donors and patients and NK cells recovered normal cytolytic function. Blockading, ex vivo, the interaction between the inhibitory CD94/NKG2A receptor on NK cells at M3 and HLA-E on target cells restored the lysis. Higher level of intra-cellular IFN-γ production was detected by flow cytometry in CD94/NKG2A⁺NK from patients at M1 and M6 as compared to the donors and at M24 (mean 70.6%± 5.2% of CD94/NKG2A+NK cells produced INF-γ after incubation with IL-12+IL-18 at M6, versus 14.6%±5% in donors and at M24). These data were confirmed by Elisa. During the first 6-months, NK cells from the SCT patients, stimulated by IL2 or IL12 plus IL18, produced roughly 14.8 or 8.4 fold more IFN-γ than healthy controls, respectively. However at M24, the level of IFN-γ production was similar to the control, regardless the conditions. IFN-γ treatment led to higher surface expression of HLA-E on AML blasts (the mean Multiplicity Fluorescence Intensity (MFI) of HLA-E expression on AML blasts from 10 patients was 9 fold higher after adjonction of INF-γ), resulting in decreased killing by NK cells from healthy donors. Actually, adjonction of IFN-γ 1000 IU/ml on AML blasts induced an inhibition of lysis by healthy donor NK cells (12% ±2% specific lysis at a ratio E/T:20/1 with INF-γ, versus 37.5%±5% without INF-γ). Blockading CD94/NKG2A restored the lysis previously inhibited by INF-γ. Neutralization of INF-γ during the interaction of NK cells generated during the first months post haplo-SCT with mismatched AML blasts led to enhanced lysis (16.5%±4.1% specific lysis at a ratio E/T:20/1 without INF-γ neutralization versus 44.1%±10% after IFN-γ neutralization at M1). This effect was no more observed at M24 or in healthy donors. These data indicate that the increased secretion of INF-γ by immature CD94/NKG2A+/ CD56^bright NK cells generated after haplo-SCT upregulate the cell surface expression of HLA-E on leukemic blasts. Increased HLA-E expression protects leukemic targets from NK-mediated lysis through the inhibitory CD94/NKG2A receptor which is overexpressed on NK cells during the first months post haplo-SCT. By contrast, at 2 years post-SCT, maturing NK cells were functionally active, as evidenced by high cytotoxicity and poor IFN-γ production. This suggests that HLA-E could be considered like a key component of an immune escape strategy against NK attack on AML blasts currently observed after haplomismathed SCT. Better understanding of how NK cells mature after SCT is necessary to improve immune response and transplatation outcome. Furthermore, the observation that mature NK cells are functionnally active against leukemic blasts is encouraging to develop clinical trials using adoptive transfer of healthy donor alloreactive NK cells.