Delta One T cells recognize AML via DNAM-1

In this issue of Blood, Mensurado et al¹ demonstrate that Delta One T (DOT) cells recognize acute myeloid leukemia (AML) cells primarily by interaction of their natural killer (NK) receptor DNAM-1 with the ligand polio virus receptor (PVR)/CD155 on AML cells.

The majority of human γδ T cells express either Vδ1 or Vδ2 as variable delta chain element of the T-cell receptor. Although the majority of circulating γδ T cells carry Vδ2 (usually together with Vγ9), Vδ1 T cells are more prominent in tissues.² γδ T cells are of substantial interest for immunotherapy of leukemias and solid cancers for 2 reasons: (1) in contrast to αβ T cells, γδ T cells do not recognize antigenic peptides presented by HLA molecules but rather γδ T cells recognize a range of stress- and transformation-inducible ligands, and (2) γδ T cells are not HLA restricted and thus can be applied in allogeneic settings across HLA barriers.^3,4 As a consequence, γδ T cells should be applicable in a wide range of hematological and nonhematological malignancies. Both Vδ1 and Vδ2 T cells are in clinical development. Vδ2 T cells recognize via their T-cell receptor small pyrophosphate molecules (“phosphoantigens”), which are overproduced in malignant cells. Activation of Vδ2 T cells by such phosphoantigens requires the presence of butyrophilin (BTN) molecules, specifically BTN3A1 and BTN2A1.² Vδ2 T cells can be easily and selectively activated by aminobisphosphonates like zoledronic acid, which stimulate the accumulation of phosphoantigens in target cells, and tumor cells are sensitized to Vδ2 T-cell recognition and destruction by pretreatment with zoledronic acid.^3,4 Vδ2 T cells expanded from healthy donors have been adoptively transferred into cancer patients with no major adverse effects, indicating that adoptive transfer of Vδ2 T cells is a promising strategy.⁵ Here, expression of the Vδ2 T-cell receptor and BTN-dependent recognition of phosphoantigens in malignant cells are thought to be key features accounting for the effective antitumor activity of Vδ2 T cells.

Vδ1 T cells also display impressive antitumor and antileukemia activity.⁴ The group of Silva-Santos has pioneered the development of protocols for large-scale in vitro expansion of clinical grade Vδ1 T cells, which they have termed Delta One T cells.⁶ DOT cells also efficiently target AML. Intriguingly, DOT cells recognizing AML express a polyclonal T-cell receptor repertoire suggesting that T-cell receptor-dependent ligand recognition does not play a major role. This was further corroborated by the observation that DOT cell activity was not significantly inhibited by antibodies directed against the T-cell receptor.⁷ 

If the T-cell receptor is not involved, how do DOT cells then recognize AML cells? Obvious candidates are activating NK receptors. Previous studies had already suggested that the interaction of NKp30 on DOT cells with the cognate ligand B7-H6 is important but cannot account for all AML target cell recognition.⁷ In the current study, Mensurado et al have now systematically investigated the role of NK receptors in the sensing of AML by DOT cells. They show that myeloid cell lines express upregulated levels of ligands for DNAM-1 (CD155/PVR, CD112 [nectin-2]) and NKp30 (B7-H6) but little if any ligands for NKG2D. Using CRISPR-Cas9 editing, they went on to generate AML cell lines that lacked 1 or a combination of PVR, nectin-2, and B7-H6 ligands. When these mutants were tested for sensitivity to DOT cells, it was found that PVR and B7-H6, but not nectin-2, synergistically contributed to DOT-cell targeting of AML cells. This was corroborated by reciprocal experiments where they ectopically expressed the ligands in murine cells. These experiments confirmed that PVR and B7-H6 but not nectin-2 conferred susceptibility to DOT-cell killing. Although the experiments with gene-edited target cells were informative to dissect the relative contribution of DNAM-1 and NKp30 ligands for DOT-cell activation, it was important to delineate the clinical significance of these findings. Therefore, Mensurado et al analyzed primary AML samples from 13 patients, of which 9 did and 4 did not express any of the 3 previously mentioned ligands. The susceptibility to DOT-cell cytotoxicity clearly segregated with the expression of at least 1 ligand. It appeared that the DNAM-1 ligand PVR was the most relevant because only PVR-positive but not PVR-negative AML patient samples were able to activate DOT cells as determined by CD69 upregulation. Therefore, the authors suggest that expression of PVR should be considered as a potential biomarker to determine the response to DOT-cell therapy in AML patients.

This study broadens our knowledge of how Vδ1-expressing DOT cells recognize AML target cells: it is not T-cell receptor-mediated antigen recognition but rather activation via activating NK receptors, specifically the interaction of DNAM-1 with its cognate ligand PVR (but not nectin-2!) on the AML cells, with additional contribution of the NKp30/B7-H6 interaction (see figure, left). This is in striking contrast to the way Vδ2 T cells recognize malignant cells (ie, through T-cell receptor-dependent recognition of phosphoantigens in the context of BTN molecules) (see figure, right). Like Vδ1/DOT cells, Vδ2 T cells can also be targeted to AML, for instance by sensitizing AML cells with agonistic anti-BTN antibodies⁸ or by selecting high-affinity Vδ2 T-cell receptors.⁹ Although future clinical studies will reveal which γδ T-cell subset might be superior for cellular immunotherapy of AML, it is important to emphasize that both γδ T cells can be applied across HLA barriers and thus are ideally suited for off-the-shelf production.

Conflict-of-interest disclosure: D.K. declares no competing financial interests.