Single Cell Transcriptomics to Characterize the Features of Bone Marrow Natural Killer Cells in Different Relapse Modalities

Allogeneic hematopoietic cell transplantation (allo-HCT) represents the most effective treatment for high-risk malignancies such as acute myeloid leukemia (AML). Despite increasing understanding of the biology underlying allo-HCT success, its efficacy is dramatically hampered by disease relapse, and an important factor that complicates the management of this event is related to the observation that post-transplantation relapse may occur through different mechanisms, each obeying a different biology and possibly requiring different therapeutic approaches.

AML develops, and is sustained by, a specialized bone marrow (BM) microenvironment that comprises several immune cell subsets. In particular, Natural Killer (NK) cells represent the first lymphoid compartment to reconstitute following transplantation, and their successful recovery is associated with protection against relapse. NK cells are involved in many processes aimed at tumor control, but their role in mediating the Graft versus Leukemia (GvL) effect following allo-HCT has not yet been fully clarified.

Here we leveraged single-cell RNA-sequencing (scRNA-seq) to profile 24,529 NK cells from the BM of 25 adult AML patients at the time of disease relapse, 5 post-transplant patients in complete remission (CR), each sampled at two different timepoints (+90 and +365 days), and 6 healthy controls (HC). We identified 3 prominent NK cell clusters (CD56^dim, CD56^bright, transitional NK), further divided into a total 8 smaller subgroups, each characterized by a distinct transcriptomic profile. Notably, we described a small subpopulation of CD56bright cells characterized by a pronounced upregulation of IFN-stimulated genes, such as IFIT2, IFIT3, and ISG15. Through pathway enrichment analysis, we also highlighted a strong enrichment in IFN-γ and IFN-α responses, and TNF-α signaling via NFκB, leading us to the definition of “inflamed” NK cells. Interestingly, by looking at subtype distribution among different known type of relapses in our cohort, we evidenced an increase in inflamed NK cells in those cases marked by the upregulation of T cell inhibitory ligands on blast surface. Presence of these inflamed NK cells in this relapse type suggests that sustained immune activation, coupled with an IFN-rich inflammatory TME, might steer both T and NK cells towards an exhaustion trajectory. Further, we highlighted the presence of a terminally differentiated CD56^dim cluster, enriched in patients that experienced CMV reactivation following transplantation, and characterized by a higher expression of Killer-cell Immunoglobulin-like Receptors (KIRs), KLRC2 (NKG2C), and cytotoxic molecules (NKG7, GZMB, GZMH).

Finally, we focused our attention on the relationship between NK cells and HLA loss relapses. HLA loss patients, while losing the major target of T cell-mediated alloreactivity, are still, in principle, subjected to NK cell control, as loss of some HLA alleles might result in the loss of ligands for inhibitory KIRs, thus triggering NK cell alloreactivity. Given the later onset of HLA loss relapses, we hypothesized that in patients experiencing this mechanism of relapse the more differentiated and cytotoxic CD56^dim subset would assume a more dominant role. Using known NK cell activation markers, we generated a score to test the functional properties of CD56^dim NK cells in our cohort, and, strikingly, HLA loss relapses showed a much lower score compared to other relapses, but also to CR and HC, providing a possible explanation as to why NK cells in these relapses are unable to counteract leukemic cells.

In conclusion, our study provides insights into the functional diversity of NK cells in the context of AML relapses following allo-HCT. By exploiting the high resolution provided by scRNA-seq, we highlight the presence of an “inflamed” subset of NK cells, whose transcriptional state might be representative of a more generalized pro-inflammatory milieu, ultimately leading to effector cell exhaustion. We also document a reduced functionality of NK cells in the context of HLA loss relapses, that could represent a critical factor contributing to the inability of the immune system to effectively combat this recurrence mechanism. Our findings underscore the need for further research to exploit these NK cell characteristics for developing targeted therapies, ultimately aiming to improve relapse management and patient outcomes in AML.