NK cell-mediated killing of Leukemia in humanized NSG mice

Many patients suffering from acute myeloid or lymphoid leukemia (AML and ALL, respectively) have increased survival and (or) more durable responses following haploidentical hematopoietic stem cell (HSC) transplantation.  Natural Killer (NK) cells are the immune effector cells that likely mediate these responses.  NK cells express highly polymorphic killer cell immunoglobulin receptors (KIR’s) that recognize specific HLA class I molecules.  Binding of KIR to “self” HLA acts as an inhibitory signal, preventing NK cell cytotoxic activity.  In the haploidentical setting, donor NK cells mismatched to the host leukemia HLA triggers cytotoxicity due to a lack of the self-inhibitory signal.  Unfortunately for leukemia patients, HLA-matched donors that carry useful alloreactive NK cells are often difficult to find.  NK cells also express another, non-polymorphic, self-inhibitory receptor called NKG2A that binds HLA-E.  HLA-E is widely expressed on both normal and cancerous cells. Researchers at Innate Pharma developed an antibody that binds NKG2A and blocks its binding to HLA-E.  In collaboration with investigators at the University of Perugia and the University of Marseille, they demonstrated that anti-NKG2A antibodies can stimulate human NK cell-mediated killing of AML and ALL in NSG mice co-engrafted with human HSCs.  The results of this investigation were published in the December issue of Haematologica (Ruggeri et al., 2015).

Anti-NKG2A antibody stimulates cytotoxicity in vitro

Initial experiments were conducted in vitro with two clonal populations of human NK cells derived from peripheral blood. One NK cell population was KIR⁺/NKG2A^- and the other was KIR^-/NKG2A⁺.  The NK cells were cultured with either HLA-matched or alloreactive, HLA-mismatched HLA-E⁺ cells.  Target cells were normal mature hematopoietic cells (B cells, T cells, monocytes) or cancer cells (EBV-infected B cells, or leukemia-derive (ALL, AML, etc.) cells.  As expected, KIR⁺/NKG2A^- NK cells were cytotoxic when co-cultured with HLA-mismatched cells, but not when co-cultured with syngeneic targets.  Further, KIR^-/NKG2A⁺ cells did not show cytotoxicity to the HLA-E⁺ cells, but did when incubated with anti-NKG2A antibody.  This experiment demonstrated that the anti-NKG2A antibody could block NKG2A binding to HLA-E to stimulate the NK cells’ cytotoxic activity and that the KIR^-/NKG2A⁺ cells had the same level of cytotoxic response as the KIR⁺/NKG2A^- NK cells when the NKG2A inhibitory pathway was blocked.

Anti-NKG2A antibody enables NK-mediated killing of leukemia in vivo

In vivo efficacy of the anti-NKG2A antibody was initially tested in immunodeficient NOD scid mice (NOD.CB17-Prkdc^scid/J , 001303).  In these experiments, mice were co-injected with human NKG2A⁺ NK cells and HLA-E⁺ EBV-infected B cells or AML cells (1:12 effector to target ratio).  The human NK cells were pre-treated with either an isotype control antibody or anti-NKG2A.  Disease progressed rapidly in the mice that received isotype control treated NK cells.  In contrast, 100% of the mice injected with anti-NKG2A treated cells survived to 150 days, and no neoplastic cells were found in the mice.  In a second experiment, NOD scid mice were initially injected with either EBV-infected B cells or AML alone.  When recipients showed 20-30% leukemia engraftment in peripheral blood, cohorts of mice received escalating doses of human NKG2A⁺ NK cells pretreated with either isotype control antibody or anti-NKG2A.  Mice engrafted with NK cells that had been treated with the control antibody all succumbed to disease.  When the leukemic mice were injected anti-NKG2A-treated NK cells, 80% survived to 150 days when injected with 3x10⁶ anti-NKG2A-treated cells and 100% survived when injected with 4x10⁶ cells.  These experiments showed that anti-NKG2A antibody initiated cytotoxicity in vivo by blocking the NKG2a-HLA-E inhibitory signal, prevented progression of disease following co-engraftment, and enabled NK cells to clear leukemia when introduced after disease had been established. 

Anti-NKG2A antibody induces endogenous NK cytotoxicity in humanized mice  

Following these successful experiments demonstrating the efficacy of the anti-NKG2a antibody, the next important question was could endogenously generated human NK cells be activated in vivo to destroy leukemic cells following systemic anti-NKG2a antibody administration?  This question was answered using the highly immunodeficient NSG mice (NOD.Cg- Prkdc^scid Il2rg^tm1Wjl/SzJ, 005557) engrafted with human CD34⁺ HSCs.  NSG mice are devoid of functionally mature mouse NK cells due to the Il2rg^tm1Wjl mutation, and allow efficient engraftment of human CD34⁺ HSCs.  The mice support multi-lineage differentiation and subsequent expansion of human HSCs into mature and functional human immune cells that repopulate the host’s blood, thymus, spleen, and bone marrow.    Twenty days after HSC engraftment, the “humanized” NSG mice received either EBV-infected B cells or human AML cells to establish mice with both a human immune system and human leukemia.  Ten days after engraftment with the leukemia, the mice were treated with 200, 250, or 300 µg of anti-NKG2A antibody or an isotype control antibody.  All of the isotype control-treated mice died from leukemia.  Humanized NSG mice engrafted EBV-infected B cell leukemia showed 60% survival when treated with 200 µg of anti-NKG2A antibody and 100% survival when treated with either 250 or 300 µg.  In the humanized mice engrafted with AML, none of the mice survived when treated with 200 µg of anti-NKG2A antibody, but 100% survived when treated with either 250 or 300 µg of anti-NKG2A.  Tracking the numbers of HLA-E⁺ leukemia cells and human CD45⁺ hematopoietic cells by flow cytometry during the first week following anti-NKG2a antibody treatment revealed that the leukemia cells steadily declined from bone marrow and spleen, and were eliminated by day 7.  The human hematopoietic cells showed reduced numbers at day 1 and day 3 following antibody treatment, but the cell numbers rebounded to pre-treatment levels by day 7.  These experiments clearly demonstrated that the humanized NSG mice produced endogenous NKG2A⁺ NK cells that when treated in vivo with anti-NKG2A antibody were capable of mounting cytotoxic responses towards HLA-E⁺ leukemia cells.  Although the treated NK cells also mounted a response to “self” HLA-E⁺ hematopoietic cells, the response only transiently affected human hematopoietic cell numbers.

The data presented by Ruggeri et al. provide strong, preclinical evidence that human NK cells can be released from self-inhibitory signaling using a monoclonal antibody that prevents interaction between NKG2A receptors and HLA-E.  The in vitro experiments showed that KIR^-/NKG2A⁺ NK cells have the same level cytotoxic response as NK cells acting through alloreactive KIRs.  This is important because blocking the NKG2A inhibitory pathway opens the door to treatment for those patients where a haploidentical donor with alloreactive NK cells is not available.  The co-injection and NK cell dose escalation experiments showed that antibody-treated NK cells can clear tumor cells in an in vivo environment.  More importantly, the studies in humanized NSG mice demonstrated that endogenous human NK cells can be induced to kill leukemia in vivo when anti-NKG2A antibody was delivered systemically, and that the NK cells were able to clear two different types of HLA-E⁺ leukemia in a dose responsive manner.  These studies highlight the increasing value of humanized NSG mice as preclinical models for developing new therapeutic strategies for treating cancers, particular those focused on immune-oncology.  These data also provide therapeutic optimism for treating patients whose cancers are refractory to traditional chemotherapy.