Single-cell profiling aligns CD56bright and cytomegalovirus-induced adaptive natural killer cells to a naïve-memory relationship

Abstract

Development of antigen-specific memory upon pathogen exposure is a hallmark of the adaptive immune system. While natural killer (NK) cells are considered part of the innate immune system, humans exposed to the chronic viral pathogen cytomegalovirus (CMV) often possess a distinct NK cell population lacking in individuals who have not been exposed, termed "adaptive" NK cells. To identify the "naïve" population from which this "memory" population derives, we performed phenotypic, transcriptional, and functional profiling of NK cell subsets. We identified immature precursors to the Adaptive NK cells that are equally present in both CMV+ and CMV- individuals, resolved an Adaptive transcriptional state distinct from most mature NK cells and sharing a common gene program with the immature CD56^bright population, and demonstrated retention of proliferative capacity and acquisition of superior IFNγ production in the Adaptive population. Furthermore, we distinguish the CD56^bright and Adaptive NK populations by expression of the transcription factor CXXC5, positioning these memory NK cells at the inflection point between innate and adaptive lymphocytes.

Keywords: HCMV (human cytomegalovirus); human immunology; innate memory; lymphocyte development and function; natural killer cell (NK cells); single-cell RNA (scRNA) sequencing.

Figure 1

Phenotypic identification and characterization of an immature NKG2C+ precursor for Adaptive NK cells. (A) Left, expression of CD57 and CD62L on live CD3-CD56+NKG2C+ NK cells in a representative healthy donor. Right, frequency of CD57-CD62L+ cells among NKG2C+ NK cells; bar represents mean and SD. (B) Left, expression of NKG2A in NKG2C+ NK cell populations from a representative healthy donor. Right, frequency of NKG2A expression in NKG2C+ NK cell populations; lines connect paired donors, Wilcoxon matched pairs signed-rank test performed. (C) Left, frequency of A+C+ among indicated populations from either CMV-seronegative or seropositive healthy donors; Mann-Whitney Test performed. Right, frequency of A+C+ and A-C+ among CD56^dim NK cells from CMV-seronegative or seropositive healthy donors; Spearman correlation coefficient shown. (D) Left, mean frequency of populations among total NK cells in peripheral blood of HSCT recipients with and without CMV re-activation in this timespan; N=10 each. Right, timelines showing fraction of the same recipients reaching reconstitution threshold of 1%. (E) Proportions of CD57 and CD62L expression among NK cell subsets from 12 healthy donors; mean represented. (F) Expression of cell surface markers on NK cell populations from healthy donors Dunn’s multiple comparison test performed, comparing groups below tick marks to group below capped end. ns, not significant; * p <0.05, ** p <0.01, *** p <0.001, **** p<0.0001.

Figure 2

Distinct transcriptional profiles of CD56^bright, CD56^dim, and Adaptive NK cells. (A) Left, UMAP representation of peripheral blood NK cells from representative donor, color-coded according to phenotypic population hashtag. Right, individual phenotypic populations superimposed on total NK. (B) Left, UMAP representation of peripheral blood NK cells from representative donor, color-coded according to transcriptional cluster. Right, top 10 consistently upregulated (green), downregulated (red), and transcription factor (bold) genes among the top 100 DEGS of both donors for each transcriptional cluster; Central and T-Like clusters are unique to this donor.

Figure 3

Convergence of a common Bright-Adaptive gene program. (A) UMAP of representative donor with KLRC2 removed from analysis, color-coded according to transcriptional cluster (left) or phenotypic population (right). (B) Top DEGs between the CD56^bright A+C- and CD56^bright A+C+ phenotypic populations, and between the Classic Dim and Adaptive transcriptional clusters. Right, dot plot of selected DEGs in the indicated phenotypic populations in Donor 1. (C) Violin plots of selected gene expression in the indicated phenotypic populations in Donor 1; 25-75% IQR and median indicated by box and white dot, respectively. (D) Surface expression of proteins encoded by genes commonly expressed by Bright and Adaptive transcriptional clusters on NK cell populations from CMV+ donors. Dunn’s multiple comparison test performed, comparing groups below tick marks to group below capped end. * p <0.05, ** p <0.01, *** p <0.001, **** p<0.0001.

Figure 4

Bright and Adaptive NK Cells are distinguished from each other by expression of CXXC5. (A) Heatmap of CXXC5 expression among Donor 1 NK cells. (B) Flow cytometry staining of NK cells for CXXC5 in representative CMV+ (upper) and CMV- (lower) donors. (C) Flow cytometry staining of NK cells for CXXC5 in a CMV+ donor with no copies of KLRC2. (D) Flow cytometry staining of NK cells for PLZF in representative CMV+ (upper) and CMV- (lower) donors. (E) Expression of CXXC5 in NK and T cell populations from CMV+ donors; dashed line represents isotype-matched control Ig staining. Dunn’s multiple comparison test performed, comparing groups below tick marks to group below capped end. * p <0.05, ** p <0.01, *** p <0.001, **** p<0.0001.

Figure 5

Memory-like effector function and proliferative capacity is intrinsic to Adaptive NK cells. (A) IFNγ production in NK cell populations from CMV+ donors following five hour stimulation with PMA+Ionomycin; Wilcoxon test performed on indicated pairs. (B) IFNγ production among NK cell populations expressing any one or more of KIR2DL1/2DS1/2DL2/2DS2/2DL3/3DL1/3DS1 or negative for all of these following five hour stimulation; Wilcoxon test performed. (C) IFNγ production in NK cell populations expressing CD57 and at least one self-recognizing KIR (KIR2DL2 and/or KIR2DL3) following five hour stimulation; Wilcoxon test performed on indicated pairs. (D) IFNγ production among NK cell populations negative for CD57 and KIR2DL1/2DS1/2DL2/2DS2/2DL3/3DL1/3DS1 following five hour stimulation; Wilcoxon test performed. (E) IFNγ production in NK cell populations at timepoints indicated following stimulation, after overnight culture with 200 U/mL IL-2. (F) NK cell populations nine days after stimulation. Left, dilution of proliferation dye in representative donor shown. Right, proliferation kinetics with lines connecting samples from the same donor; Wilcoxon test performed. (G) Expression of CD57 and proliferation dye among A-C+ FcRγ- NK cells nine days after stimulation with PMA+Ionomycin, representative donor shown. (F, G) Total CD3-CD56+ NK cell population shown as the distinction between CD56^bright and CD56^dim is lost after prolonged stimulation. ns not significant, * p <0.05, ** p <0.01.