Overcoming NK Cell-Mediated Allograft Rejection By Anti-3rd Party Central Memory Veto CD8 T Cells through Down-Regulation of the Activating Receptor DNAM-1 on Alloreactive NK Cells

Allograft rejection is a major challenge for allogeneic chimeric antigen receptor (CAR) cell therapy, particularly due to host T and NK cell responses. Previous studies demonstrated that anti-third-party central memory CD8 veto T cells (Tcm) can overcome graft rejection under mild conditioning without causing significant GvHD, suggesting these cells as an ideal platform for allogeneic CAR T cell therapy. We demonstrated that veto Tcm can effectively delete anti-donor T cell clones through a Fas-FasL mechanism, whereas the mechanism by which they neutralize alloreactive NK cells remains unknown.

To address this question, we established a model for NK-mediated rejection, in the absence of T cell activity. GFP+ T depleted bone marrow (TDBM) cells from B6 donors (H2K^b) were transplanted into BALB/c nude mice (H2K^d) pre-conditioned with 3Gy TBI, in the presence or absence of anti-asialo GM1 antibody pretreatment that specifically eliminates NK cells. FACS chimerism analysis of the blood and spleen at 30 days post-transplant revealed significant differences between anti-asialo GM1 treated or untreated mice (unpaired t-test, blood p=0.0035 and spleen p=0.0018, n=5). Notably, a similar ability to overcome NK-mediated rejection was observed upon transplantation of TDBM plus B6-Tcm veto cells, leading to marked enhancement of chimerism at 30 days post-transplant (average GFP+ cells in blood 4.26%, and spleen 3.18%), compared to mice treated with TDBM alone (average GFP+ in blood 0.22%, and spleen 0.18%) (unpaired t-test, blood p=0.0035, and spleen p=0.0018, n=5 ).

However, in contrast to the ablation of NK cells using anti-asialo GM1 antibody, FACS analysis revealed that administration of veto cells was not associated with significant reduction of host NK cell levels. Based on this observation, we hypothesized that the mechanism by which Tcm veto cells neutralize NK cell-mediated rejection might involve suppression of host NK cytotoxicity, as opposed to NK cell deletion. To address this possibility, we first interrogated the impact of Tcm veto cells on NK cell proliferation in mixed lymphocyte reaction (MLR). CFSE-labeled BALB/c splenocytes were co-cultured with irradiated (20 Gy) C57BL/6 splenocytes, in the presence or absence of B6-derived Tcm veto cells. While in the absence of Tcm veto cells, NK cells exhibited marked proliferation starting at 96 hours of culture, the addition of B6 Tcm veto cells led to a substantial arrest of NK cell proliferation (unpaired t-test, n=5, p<0.0001). Furthermore, as alloreactive BALB/c NK cell subpopulations directed against B6 MHC class-I express the Ly49G2+ receptor, it was possible to further analyze the expression of CD107a on Ly49G2+ alloreactive NK cells. Indeed, we found in MLR culture significant suppression of CD107a expression upon addition of veto Tcm (unpaired t-test, n=6, p=0.0014).

Considering that NK cell responses primarily depend on the balance between activating and inhibitory signals, one mechanism that could lead to NK anergy might be mediated by reduced expression of activating receptors such as DNAM-1. Thus, using the MLR culture described above, we found that DNAM-1 expression on the alloreactive Ly49G2+ BALB/c NK subset was reduced upon addition of Tcm (unpaired t-test, n=8, p=0.0032). This loss of expression was associated with enhanced internalization of DNAM-1 and downstream activation of phosphorylated Cbl, which in turn leads to endo-lysosomal degradation. Notably, compared to control spleen T cells that lack veto activity, the veto Tcm exhibit marked elevation of the DNAM-1 ligand, CD155; blockade of this ligand with anti-CD155 anybody inhibited DNAM-1 internalization. This finding is consistent with previous reports demonstrating that increased expression of CD155 on tumor cells is responsible for DNAM-1 down-modulation on NK cells (Carlsten, et al.; J. Immunol. 2009, 4921). Likewise, blocking antibodies against Ly49G2 or MHC-class I also inhibited DNAM-1 down regulation, suggesting that the entire cascade leading to DNAM-1 down regulation is initiated by ligation of Ly49G2 on NK cells with MHC-class I on the veto cells.

Taken together, our results show for the first time that veto Tcm can overcome NK-mediated allograft rejection in-vivo. Our studies in MLR demonstrate that this unique inhibitory activity is likely mediated thorough down-regulation of the DNAM-1 activating receptor on alloreactive NK cells.