Pharmacological Control of Recipient NK Cell Function By Venetoclax Promotes Engraftment and Limits GVHD Following Reduced Intensity Conditioning without Impacting the Graft Versus AML Effect

Background

The outcome of allogeneic haematopoietic stem cell transplantation (alloHSCT) is a complex interaction between conditioning regimen intensity (efficacy vs toxicity), recipient immunology (rejection vs engraftment) and donor immunology (rejection vs engraftment, and GVHD vs GVL). We must find means to lower conditioning toxicity, promote donor engraftment and limit GVHD in order to improve alloHSCT outcomes.

Aims: Optimize the use of BCL2 inhibitors to modify recipient NK cell function in mouse models of alloHSCT in order to minimize GVHD severity and onset, and retain GVL responses.

Hypothesis: Therapeutic targeting of recipient NK cell frequency or function pre-transplant will allow reduced intensity conditioning (RIC) and promote both donor T cell engraftment and GVL whilst reducing the risks of GVHD.

Methods

MHC-mismatched mouse model of alloHSCT: Donor BM and T cells from BALB/c (H2K^d) mice were injected into irradiated C57BL/6 (H2K^b) recipients. On day 0, C57BL/6 WT, or Bcl2^fl/fl (Bcl2 deleted in NKp46⁺NK1.1⁺ cells) recipients were irradiated with a RIC dose of 2 x 400 rad, before i.v. injection of 7.5^e6 BALB/c BM + 1^e6 T cells. C57BL/6 WT mice were also treated on day -2 and -1 by oral gavage with 100 mg/kg Venetoclax (ABT-199) or vehicle, before receiving alloHSCT. Mice were monitored for onset of GVHD and tested for early engraftment (day 7-14 post-transplant), and late engraftment (up to day 50).

Graft-versus leukaemia (GVL) AML model:8^e5mixed lineage leukaemia cells (MLL-AF9) were injected i.v. on day 0 into C57BL/6 WT recipients. On days 8 and 9 mice were treated by oral gavage with 100 mg/kg Venetoclax or vehicle, before receiving alloHSCT on day 10. Mouse survival and AML burden in the BM was measured on day 30.

Results

We found that at both a myeloablative radiation dose (1200 rad) and a RIC non-myeloablative dose (800 rad), recipient NK cells were significantly more radio-resistant than either CD8⁺ T or myeloid cells. However, when a RIC dose of 800 rad was used during alloHSCT, the rejection threshold was not overcome and engraftment failed. We investigated both genetic and pharmacological models of recipient NK cell suppression during alloHSCT with RIC, to promote donor cell engraftment, reduce GVHD, and retain GVL.

Conditional deletion of Bcl2 in NK cells results in a 90% loss of NK cells in vivo . Bcl2^fl/flalloHSCT RIC recipients showed robust donor engraftment, but absence of the pro-inflammatory cytokine storm and substantially less GVHD as determined by clinical scores and gut histology, compared to WT recipients. Pharmacological inhibition of BCL2 in WT recipients recapitulated the transplant findings in Bcl2^fl/flrecipients. We found that BCL2 inhibition by Venetoclax, a BCL2 antagonist approved in the treatment of AML, resulted in NK cell apoptosis in human cells . We extended our observations in Bcl2^fl/flrecipients to show that pharmacological inhibition of BCL2 in WT mice with just two doses of Venetoclax resulted in rapid depletion of NK cells. Over 35% of alloHSCT WT recipient mice pre-treated with Venetoclax developed long-term full donor engraftment even in the setting of RIC, with minimal GVHD, and retained potent GVL effects against pre-established AML.

Conclusions

Venetoclax depletion of NK cells in WT alloHSCT recipients in combination with RIC was: 1) well-tolerated, 2) associated with low rates of GVHD, 3) resulted in long-term donor haematopoietic cell engraftment, and 4) maintained a GVL response. Recipient NK cell inhibition may therefore represent a means by which to deliver alloHSCT more safely.