Insights into Clonal Relationships of Putative Adaptive Natural Killer Cells (NK) in Humans, Via Mapping of Somatic Piga Mutations in Patients with Paroxysmal Nocturna Hemoglobinuria (PNH)

NK cells play a central role in innate immunity, specifically in tumor surveillance and microbial pathogen control. Recent murine models and human studies have identified subsets of NK-cells with apparent memory cell function, strongly linked to CMV infection in humans and termed “adaptive” NK. Our recent clonal tracking studies following autologous hematopoietic stem cells transplantation (HSCT) of genetically-barcoded CD34+ cells in macaques revealed distinct clonal ontogeny of a subset of NK cells within the primate equivalent of the human CD56 ^dim population (Wu et al., Cell Stem Cell, 2014), with little clonal overlap with T-, B-lymphoid or myeloid cells, suggesting a separate precursor pool for this NK subtype. Peripheral blood CD 56^bright-NK cells have been previously hypothesized to be precursors for the main population of circulating cytotoxic CD56^dim cells. To further investigate NK-cell ontogeny and clonal relationships in humans we took advantage of naturally-occurring somatic mutations in the X-linked phosphatidylinositol glycan class A (PIGA) gene in patients with the hemolytic disorder paroxysmal nocturnal hemoglobinuria. This gene codes for an enzyme required for cell surface localization of glycosylphophatidylinositol (GPI)-anchored proteins, and thus loss of function mutations result in hematopoietic cells lacking GPI-anchored proteins, and red cell hemolysis. PNH patients have not been reported to have immune dysfunction and can have stable disease for many years. Membrane bound GPI anchors can be detected on any cell via flow cytometry using a labeled inactive aerolysin (FLAER), and serves as a marker for the fraction of cells comprising the PIGA mutant (GPI negative) clonal compartment.  The PNH clone sizes contributing to peripheral blood cells are variable in but can reach almost 100% in some patients, and can be stable over decades. We selected 9 PNH patients with GPI negative granulocytes ranging from 5% to 98% and a median time from diagnosis of 43.7 months (15-100) for this study.  NK cells were defined as CD56+/CD16+/CD3-/CD20- lymphocytes.  We observed disproportionally fewer GPI negative NK cells compared to granulocytes (Fig 1), with the discrepancy most marked the major peripheral blood CD56 ^dim population (mean 65% vs 25% GPI negative granulocytes, p = 0.0028, paired t-test), in contrast to 46% GPI negative cells in the CD56^bright population (p=0.057). Due to the prolonged life span of memory T and B cells, fewer GPI negative B and particularly T-lymphocytes have been reported in PNH patients. In our cohort 3.4% of CD3+ T-cells and 13.2% of CD20+ B-cells were GPI negative (p=0.0005 and 0.0014, respectively versus granulocytes). Compared to the NK subsets, the CD56 ^bright population showed the most significant differences (p = 0.0063 versus CD3 and p=0.0151 versus CD20).  To further characterize the phenotype of the GPI positive versus negative CD56^dim cells, we analyzed cells co-expressing either the terminal differential marker CD57, the inhibitory receptor NKG2A, or activating receptor NKG2C for FLAER positivity. Prior studies have suggested that the human CMV-linked adaptive NK subset is CD57+, NKG2A- and NKG2C+.  The NKG2C+ CD 56^dimpopulation was highly enriched for GPI positive cells (p=0.0024 vs granulocytes, Fig 1). This profile was most prominent in CMV-IgG positive patients who had also significantly more GPI positive CD 56^dim/CD57+ cells compared to granulocytes (p=0.008). Interestingly, one CMV positive patient (#5) had a complete lack of NKG2C expression, most likely due to homozygous loss of function mutation, and this patient had almost 100% GPI negative NK cells, matching his neutrophil pattern. Compared to granulocytes, NKGA2A+ or CD57+ positive CD56^dim cells were also mostly GPI negative (p=0.0081 and 0.028).

Circulating NK cell turnover has been estimated to be about 14 days.  The PNH patients studied had documented clonal PIG-A mutations for many years.  Our observation that the majority of CD56^dim NK cells, specifically the NKG2C subset, are not progeny of the same progenitors producing CD56bright NK cells or myeloid cells based on clonal disparity regarding the PNH clone is suggestive of an independent, very long-lived or self-renewing NK cell progenitor for a CMV-linked CD56dim/CD57+/NKG2C+ memory NK cell compartment. These observations provide novel further insights into the human adaptive NK cell subset.

 

Figure 1