CD56^dimCD57⁻NKG2C⁺ NK cells retaining proliferative potential are possible precursors of CD57⁺NKG2C⁺ memory-like NK cells

Abstract

Formation of the adaptive-like NK cell subset in response to HCMV infection is associated with epigenetic rearrangements, accompanied by multiple changes in the protein expression. This includes a decrease in the expression level of the adapter chain FcεRIγ, NKp30, and NKG2A receptors and an increase in the expression of NKG2C receptor, some KIR family receptors, and co-stimulating molecule CD2. Besides, adaptive-like NK cells are characterized by surface expression of CD57, a marker of highly differentiated cells. Here, it is shown that CD57-negative CD56^dimNKG2C⁺ NK cells may undergo the same changes, as established by the similarity of the phenotypic expression pattern with that of the adaptive-like CD57⁺NKG2C⁺ NK cells. Regardless of their differentiation stage, NKG2C-positive NK cells had increased HLA-DR expression indicating an activated state, both ex vivo and after cultivation in stimulating conditions. Additionally, CD57⁻NKG2C⁺ NK cells exhibited better proliferative activity compared to CD57⁺NKG2C⁺ and NKG2C⁻ NK cells, while retaining high level of natural cytotoxicity. Thus, CD57⁻NKG2C⁺ NK cells may represent a less differentiated, but readily expanding stage of the adaptive-like CD57⁺NKG2C⁺ NK cells. Moreover, it is shown that NK cells have certain phenotypic plasticity and may both lose NKG2C expression and acquire it de novo during proliferation, induced by IL-2 and K562-mbIL21 feeder cells.

1 |. INTRODUCTION

NK cells were first identified by their ability to kill tumor cell lines in vitro without prior activation.¹ Nowadays, it is shown that NK cells provide protection not only against viral, bacterial or fungal infections and tumors,^2–4 but also play a role in the formation and functioning of the adaptive immune response through production of cytokines.² Intriguingly, the findings of the past decade have revealed that NK cells possess some features typical for adaptive cells, despite the fact that traditionally NK cells were classified as innate immune cells. These adaptive-like NK cells have the ability to accumulate, similar to the clonal expansion of T cells during the formation of immunological memory, and demonstrate more effective and specialized response after re-stimulation.^5,6 The stimuli can be both specific (haptens in murine models)⁷ and unspecific (for instance, cytokines IL-12, IL-15, IL-18).^8,9 The mechanisms of NK cell memory development differ from those in T and B lymphocytes and are not fully understood yet. However, at the transcriptional level, a significant similarity between adaptive NK cells and memory T cells has been established.¹⁰

The most widely known and studied type of adaptive NK cells is associated with cytomegalovirus infection. Initially, these cells were discovered in mice: murine NK cells recognized the m157 viral protein through Ly49 receptor, actively proliferated, and showed an enhanced effector response after the re-injection of the virus.⁵ In humans, cytomegalovirus (HCMV) infection is also associated with the generation of adaptive-like NK cells characterized in the first place by the expression of an activating receptor NKG2C.^11,12 Typical adaptive-like NKG2C⁺ NK cells are highly differentiated, CD57-positive and NKG2A-negative, and express KIR family receptors.¹³ In these cells, a decrease in the expression of the FcεRIγ transmembrane adapter molecule and, as a result, of the NKp30 and NKp46 surface receptors is often observed.¹⁴ The CD7 and CD161 expression is also down-regulated.¹⁵ Additionally, the adaptive-like NK cells are characterized by up-regulation of CD2 – co-stimulatory molecule, providing “signal 2” during CD16-dependent recognition of the pathogen.¹⁶ According to the current data, main functional characteristics of adaptive-like NK cells are: a decrease in natural cytotoxicity and sensitivity to cytokines produced by the innate immune cells (IL-12, IL-18), an increase in the CD16-mediated Ab-dependent cellular cytotoxicity (ADCC) and IFN-γ production in response to appropriate stimuli, and also an increased lifetime (according to various estimates, from 4 months to 1 year).^6,11,17–20 Another feature of NKG2C⁺CD57⁺ adaptive NK cells is epigenetic remodeling, which includes the regulatory region of the IFNG gene, and, apparently, determines increased ability of these cells for IFN-γ production under CD16-dependent activation.²¹ Recently, adaptive-like NK cells were shown to express HLA-DR – MHC class II molecule, commonly expressed on the APCs.^13,22 The expression of this molecule is usually associated with NK cell activation and increased IFN-γ production.^23,24

It should be noted that, although the fraction of NKG2C⁺ NK cells is found in most of the HCMV-seropositive people, its size can vary significantly, and a small proportion of NKG2C⁺ NK cells can be detected in HCMV-seronegative individuals too.²⁰ The reasons for such a significant heterogeneity and its clinical significance, as well as the mechanisms driving the appearance and expansion of this subset, are still not fully understood. It is suggested that pro-inflammatory cytokines, such as IL-12, are necessary for the formation of NKG2C⁺ memory-like NK cells, along with HCMV-infected cells.²⁵ The number of gene copies encoding NKG2C also affects the amount of NKG2C-positive NK cells.²⁶ However, even in the absence of NKG2C gene, HCMV infection leads to formation of a special adaptive-like NK cells, characterized by the expression of activating KIR receptors.¹⁵ Besides, it has been recently shown that homeostasis of NKG2C⁺ NK cells during HCMV infection can at least partly be controlled by co-expression of certain inhibitory KIRs (KIR2DL1, KIR2DL2/3).^27,28 For instance, the strong interaction of KIR2DL1 and self-expressed HLA-C2 ligands seems to promote the vast and stable expansion of adaptive NK cells in an HCMV-infected individual.²⁷

At the moment, the subset of NKG2C⁺CD57⁺ adaptive NK cells is characterized quite well, but at the same time, the role of NKG2C expression in less differentiated CD57-negative NK cells remains unclear. It is unknown whether NKG2C⁺CD57⁻ NK cells are precursors to the NKG2C⁺CD57⁺ cells, and whether they possess any features characteristic of the adaptive NK cell subset. According to the widespread hypothesis, NK cells with the CD56^bright phenotype differentiate into CD56^dim, while losing the expression of NKG2A and up-regulating the expression of CD57 and KIR receptors.^29,30 It is possible that NKG2C⁺ NK cells acquire certain adaptive features when the differentiation is still incomplete, while retaining the ability to proliferate in response to activating stimuli. In this paper, NKG2C-expressing NK cells at different stages of differentiation from both HCMV-positive and HCMV-negative individuals were examined in detail, and their phenotypic and functional characteristics were compared.

Until recently, the process of NK cell differentiation, including the acquisition of adaptivity features, was considered irreversible.³¹ However, there is more and more evidence favoring the plasticity of NK cell phenotype under certain conditions.^29,32–34 Particularly, it was demonstrated that NK cells are able to up-regulate NKG2A expression and lose CD57 expression upon cytokine and/or feeder cell-mediated activation.^33,34 Exploring the ability of NK cells to exhibit phenotypic and functional plasticity is important for broadening our knowledge about the processes of NK cell differentiation and activation, which may help to improve the methods of NK cell expansion for clinical applications. Here, we investigated the proliferative activity, phenotypic stability, and functional properties of weakly and highly differentiated NKG2C⁺ NK cells upon in vitro stimulation with IL-2 and K562 feeder cells, expressing membrane-bound IL-21.

2 |. MATERIALS AND METHODS

2.1 |. Cell lines

K562 (human erythroblastoid leukemia), C1R, and Raji (human B-cell lymphoblastoid-like cells) lines were obtained from ATCC (Manassas, VA, USA). Genetically modified K562-mbIL21 cell line was kindly provided by Dr. Dean Lee (MD Anderson Cancer Center, USA). Besides membrane-bound IL-21, this line also expresses CD19, CD64, CD86, and CD137L (4-1BBL).³⁵ The cell lines were cultivated in RPMI-1640 (PanEco, Russia) supplemented with 10% FCS (fetal calf serum) (HyClone, USA), 2 mM of L-glutamine (PanEco, Russia) and Antibiotic Antimycotic Solution (Sigma-Aldrich, St. Louis, MO, USA). Surface expression of IL-21 during cultivation was tested periodically by flow cytometry using IL-21-PE Ab (clone 3A3-N2, BioLegend, USA). Before adding to NK cells, the K562-mbIL21 cells were irradiated with γ-radiation (100 Gy) and stored in FCS (HyClone, USA) supplemented with 10% DMSO (Sigma-Aldrich, St. Louis, MO, USA) at −135 or −150°C.

2.2 |. Blood donors

Blood samples were obtained from healthy volunteers of different age and sex (50 individuals). All participants gave their verbal informed consent prior to the study that was approved by local ethics committee (Pirogov Russian National Research Medical University, Moscow, Russia).

2.3 |. Human NK cells isolation

Peripheral blood mononuclear cells were isolated using gradient centrifugation with standard Ficoll solution (PanEco, Russia) (density 1.077). NK cells were obtained from PBMC by negative magnetic separation using NK cell isolation kit (Miltenyi Biotec, Germany) according to the manufacturer’s protocol. After negative magnetic separation the purity of NK cells reached 95–99%.

2.4 |. HCMV serology status

HCMV serology status was determined according to the titer of anti-HCMV Abs in sera samples of patients, measured using the appropriate ELISA kit (Vector-Best, Russia) in accordance with the manufacturer’s protocol.

2.5 |. Antibodies

For surface and intracellular staining of NK cells the following mouse anti-human fluorescent-labeled Abs were used: CD2-PE-Cy7 (clone TS1/8), CD56-BrilliantViolet421 (clone HCD56), CD56-AF488 (clone 5.1H11), CD56-PE-Cy7 (clone 5.1H11), CD57-PE (clone TB01), HLA-DR-FITC (clone LN3), HLA-DR-PE (clone L243), IFN-gamma-FITC (clone 4S.B3), KI-67-PE (clone Ki-67), NKp46-FITC (clone 9E2) (Sony Biotechnology, USA), CD16-PE (Sorbent, Russia), CD16-APC (clone REA423), CD56-APC (clone REA196), CD57-FITC (clone TB03), CD57-APC (clone TB03), CD57-PE-Vio770 (clone TB03) (Miltenyi Biotec, Germany), CD44-F (clone J.173, Immunotech, USA), CD56-PE (clone C5.9, Dako, USA), CD107a-PE-Cy5 (clone H4A3), NKp30-PE (clone P30-15) (eBioscience, USA), CD161-AlexaFluor647 (clone HP-3G10), Granzyme B-Alexa Fluor 647 (clone GB11), KIR2DL2/DL3-PE (clone DX27) (Biolegend, USA), FcεRIγ-FITC (polyclonal, Milli-Mark, Merck, Germany), NKG2A-PE (clone 131411), NKG2C-AlexaFluor 488 (clone 108724), NKG2C-PE (clone 134591) (R&D Systems, USA). Supernatant of hybridoma producing Abs to human KIR2DL1 was kindly provided by prof. Miguel Lopez-Botet (Universitat Pompeu Fabra, Barcelona, Spain). Sheep anti-mouse IgG-FITC and PE (Sigma-Aldrich, USA) were used as second Abs for anti-KIR2DL1. IgG1-PE (clone IS5-21F5, Miltenyi Biotec, Germany) was used as isotype control for NKG2C-PE Ab staining.

2.6 |. Cell staining and flow cytometry

For surface fluorescent immunostaining cells were incubated with Abs for 30 min on ice in PBA staining buffer (PBS containing 0.5% BSA (Serva, Heidelberg, Germany) and 0.01% sodium azide (AMRESCO, Inc. (VWR International, LLC), Aurora, CO, USA)) and then washed twice in the same buffer. Some samples were stained with Hoechst 33342 (FluoroPure^™ Grade, Thermo Fisher Scientific, USA) during 45 min on water bath (+37°C). Samples were analyzed in FACSCalibur flow cytometer (BD Biosciences, USA), equipped with 488 and 640 nm lasers. Samples labeled with Hoechst 33342 were analyzed on FACSVantage DiVa machine (BD Biosciences, San Jose, CA, USA) equipped with 405, 488, and 643 nm lasers. From 50,000 to 300,000 events for freshly isolated NK cells and from 5000 to 30,000 events for NK cell fractions and clones in FSC-SSC gate were recorded.

2.7 |. NK cell sorting and cultivation

Freshly isolated NK cells were labeled with mouse anti-human Abs according to their sorting scheme. Cells were labeled with mAbs in PBS containing 0.5% BSA and 2 mM EDTA. FACSVantage DiVa machine was used for cell sorting. NK cells were sorted into 12 × 75 mm tubes and then cultivated in 96-well U-bottom plates (50,000 cells per well) with 40,000 of K562-mbIL21 feeder cells per well in NK MAX cell medium (Milteniy Biotec, Germany) containing 100 units/ml of recombinant IL-2 (Sigma-Aldrich, St. Louis, MO, USA). CD57⁻ NKG2C⁻, CD57⁻NKG2C⁺, CD57⁺NKG2C⁻ and CD57⁺NKG2C⁺, or CD56^dimCD57⁻NKG2C⁻, CD56^dimCD57⁻NKG2C⁺, CD56^dimCD57⁺ NKG2C⁻, and CD56^dimCD57⁺NKG2C⁺ NK cell fractions were sorted for different experiment series. The culture medium was replaced each 3–4 days.

2.8 |. Generation of NK cell clones

Isolated NK cells were labeled with Abs in PBS containing 0.5% BSA and 2 mM EDTA and then sorted on the FACSVantage DiVa machine in the “single-cell” mode, one per well, into 96-well U-bottom plates containing 200 μl of complete medium for clones: DMEM medium (PanEco, Russia) with 20% ExVivo medium (Thermo Fisher Scientific, Carlsbad, CA, USA), 100 U/ml of recombinant human IL-2 (Sigma-Aldrich, St. Louis, MO, USA)) and 2000 K562-mbIL21 feeder cells per well. The following sub-populations were sorted: CD56^brightNKG2C⁻, CD56^brightNKG2C⁺, CD56^dimCD57⁻NKG2C⁻, CD56^dimCD57⁻NKG2C⁺, CD56^dimCD57⁺ NKG2C⁻, and CD56^dimCD57⁺NKG2C⁺. After 3 weeks of incubation (37°C, 5% CO₂) half of the medium was replaced. After 6 weeks K562-mbIL21 feeder cells (2000 per well) were added once more to growing clones.

2.9 |. Intracellular staining

Staining of NK cells with Abs to FcεRIγ, granzyme B, and IFN-γ was performed through fixation and permeabilization of cells with BD Cytofix/Cytoperm kit (BD, USA) according to the manufacturer’s protocol.

2.10 |. Proliferation assays

NK cell proliferation was determined by the expression of nuclear protein Ki-67 and direct cell counting by automatic cell counter TC20 (Bio-Rad, USA). For NK cell immunolabeling with Ki-67 70% EtOH fixation was used (for 1 h at −20°C). After fixation, cells were washed two times with staining buffer and were incubated with anti-human Ki-67-PE Ab for 30 min at room temperature.

2.11 |. Degranulation assay

Cytotoxic activity was measured by degranulation of NK cells according to the procedure described earlier.^24,36 Briefly, pre-stimulated (500 units/ml IL-2, 20 h) NK cells were mixed in ratio 1:1 with K562 cells or C1R cells, treated with 2.5 μg/ml Rituximab (CD20 mAb, Roche Holding, Basel, Switzerland), in the RPMI-1640 medium with anti-CD107a-PE-Cy5 Ab and 10 μg/ml of brefeldin A (Sigma, USA). Cells were precipitated at 240 g for 30 s and then incubated for 2.5 h at 37°C, 5% CO₂. Analysis was performed on the FACSCalibur flow cytometer; the percentage of CD107a⁺ NK cells was evaluated.

2.12 |. Cytokine and Ab-dependent IFN-γ production assay

IFN-γ production was estimated by intracellular staining. Freshly isolated NK cells were stimulated for 20 h by IL-12 and IL-18 (20 ng/ml) or for 5 h by Raji cells (in ratio NK cells: Raji cells - 4:1) in the presence of Rituximab (2.5 μg/ml) and brefeldin A (10 μg/ml).

2.13 |. Statistical analysis

To determine statistical significance ANOVA or Student’s t-test was used. P < 0.05 was considered significant. Cytometry data were processed by Flowing Software (version 2.5.1, Finland) and FlowJo (version vX.0.7, FlowJo LLC, Oregon, USA) and analyzed by Graph Pad Prism (version 7.00, GraphPad Software, USA). Data are shown as mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

3 |. RESULTS

3.1 |. Surface expression of NKG2C on circulating NK cells depends on their differentiation stage and donor HCMV status

As a first step, we quantified the proportion of CD57⁻NKG2C⁺ and CD57⁺NKG2C⁺ NK cells in the individuals with different HCMV serological status. NKG2C⁺ NK cells were present in the blood of all the studied volunteers (57 individuals), ranging from 1% to 78% of all NK cells. The age of donors ranged from 20 to 61 years (median was 26 years). According to the HCMV serological status, the donors were divided into 4 groups: with an undetectable Ab level (group 1, less than 0.25 active units (AU) /ml, age median 27), and with low (group 2, 0.25–4 AU/ml, age median 26), medium (group 3, 4–7 AU/ml, age median 27.5), and high titer of the anti-HCMV Abs (group 4, more than 7 AU/ml, age median 23). As expected, group 1 (HCMV-seronegative donors) had fewer circulating NKG2C⁺ NK cells compared with the other groups (HCMV-seropositive donors) (Fig. 1A). However, significant differences were registered only between groups 1 and 3. We did not reveal any significant differences between the groups of volunteers in the number of circulating highly differentiated CD57⁺ NK cells (Supplementary Fig. S1).

FIGURE 1. NKG2C⁺ NK cell fraction analysis from donors with different HCMV-status.

(A) Scatter plot illustrating proportion of NKG2C-positive NK cells in NK cell fraction of four donor groups with different HCMV serostatus. (B-D) Percentage of NKG2C⁺ cells in NK cell subsets at distinct differentiation stage: CD56^bright, CD56^dimCD57⁻ and CD56^dimCD57⁺ in (B) 4 donor groups with different HCMV serostatus, (C) HCMV-negative individuals (n = 13) and (D) HCMV-positive individuals (n = 44). Gr.1, Gr.2, Gr.3, Gr.4 stand for group 1 (no anti-HCMV Abs), group 2 (low titer), group 3 (medium titer), group 4 (high titer), and comprise 13, 21, 13, and 10 donors, respectively. To determine statistical significance 1-way ANOVA was used for (A), (C), and (D), 2-way ANOVA for (B)

NKG2C⁺ NK cells were analyzed depending on their differentiation stage according to the following scheme: CD56^bright < CD56^dimCD57⁻ < CD56^dimCD57⁺ subsets, from less to more differentiated.³¹ HCMV-seropositive donors (groups 2–4) had a higher percentage of NKG2C⁺ NK cells in the CD56^dimCD57⁺ subset compared with HCMV-seronegative (group 1) (Fig. 1B). Moreover, within the HCMV-seronegative group, the highest number of NKG2C⁺ NK cells was revealed in the intermediate subset CD56^dimCD57⁻ (Fig. 1B and C), and within the HCMV-seropositive individuals – in the terminally differentiated subset CD56^dimCD57⁺ (Fig. 1B and D). Interestingly, in all groups, including the group of seronegative donors, the number of NKG2C-positive cells was higher in the CD56^dimCD57⁻ subset compared with the less differentiated CD56^bright fraction (Fig. 1C and D). Thus, high amount of total NKG2C⁺ NK cells in HCMV-seropositive individuals is mostly the result of accumulation of the CD56^dimCD57⁺NKG2C⁺ fraction, which represent the classical phenotype of adaptive NK cells associated with HCMV.¹¹ At the same time, we observed an increase in the NKG2C⁺ cell number occurring after the transition of NK cells from the CD56^bright to CD56^dimCD57⁻ differentiation stage, which did not depend on the donor’s HCMV serostatus (Fig. 1C, D). We can suppose that further expansion and differentiation of these NKG2C⁺ NK cells accompanied by CD57 acquisition takes place only in the HCMV-infected individuals (Fig. 1D).

3.2 |. CD56^dimCD57⁻NKG2C⁺ cell fraction shares certain phenotypic features with the adaptive-like CD57⁺NKG2C⁺ NK cells

Next, we analyzed phenotypic differences between ex vivo NKG2C⁺ NK cells from distinct differentiation stages and compared them with NKG2C⁻ cells. Phenotype of freshly isolated NK cells was analyzed in the following 6 fractions separated by gating: CD56^brightNKG2C⁻, CD56^brightNKG2C⁺, CD56^dimCD57⁻NKG2C⁻, CD56^dimCD57⁻NKG2C⁺, CD56^dimCD57⁺NKG2C⁻, CD56^dimCD57⁺ NKG2C⁺ (Fig. 2). We have checked a number of markers, which are shown to be down-regulated (NKG2A, CD161, NKp30, and NKp46), or up-regulated (KIR2DL2/DL3, CD2) in the adaptive NK cells. We have also estimated the expression of Fc-receptor CD16, matrix adhesion molecule CD44, and activation marker HLA-DR (Figs. 3–5). According to the literature, the NKG2A receptor is mainly expressed on the surface of less differentiated cells, and KIR and CD16 receptors – on more differentiated NK cells.^31,37

FIGURE 2.

Scheme of NK cell sorting into subsets according to their differentiation stage (estimated by CD56 and CD57 expression) and NKG2C expression

FIGURE 3. Ex vivo expression of NKG2A, KIR2DL2/DL3, and KIR2DL1 receptors on NK cells differing in maturation stage and NKG2C expression.

Proportion of (A) NKG2A⁺ (n = 28), (C) KIR2DL2/DL3⁺ (n = 30), and (E) KIR2DL1⁺ NK cells (n = 19), expression level of (B) NKG2A, (D) KIR2DL2/DL3, and (F) KIR2DL1 receptors are represented. To determine statistical significance, 1-way ANOVA was used

FIGURE 5. Ex vivo expression of HLA-DR on NK cells differing in maturation stage and NKG2C expression.

(A) Proportion of HLA-DR⁺ NK cells in NK cell fractions in individuals with different HCMV serostatus. Data of 4 HCMV-seronegative and 22 HCMV-seropositive donors are shown. To determine statistical significance, 2-way ANOVA was used. (B) HLA-DR surface expression on NK cells from different subsets. Representative staining of cells from 1 donor out of 26 donors examined is shown

We have shown that all the NKG2C⁺ NK cell subsets had a significantly smaller percentage of NKG2A⁺ cells (Fig. 3A), but a larger percentage of KIR2DL2/DL3⁺ cells (Fig. 3С), compared with the NKG2C⁻ counterparts. Each NKG2C⁺ subset also had a significantly higher KIR2DL2/DL3 expression level per cell compared to the corresponding NKG2C⁻ subset. Interestingly, between NKG2C-negative fractions, the highest KIR2DL2/DL3 expression intensity was observed in the CD56^bright subset (Fig. 3D). Difference in the NKG2A expression intensity between NKG2C⁺ and NKG2C⁻ cells was found only in the CD56^bright subset, indicating more differentiated state of the CD56^brightNKG2C⁺ NK cells compared to the CD56^brightNKG2C⁻ cells (Fig. 3B). Patterns of KIR2DL1 expression were similar to KIR2DL2/DL3 (Fig. 3E, F). We observed a tendency to increased proportion of KIR2DL1⁺ NK cells in the CD57⁻NKG2C⁺ subset (Fig. 3E). In addition, CD56^brightNKG2C⁺ NK cells had the highest KIR2DL2/DL3 and KIR2DL1 expression intensity (Fig. 3D, F). To conclude, NKG2C⁺ NK cells at each differentiation stage (CD56^bright, CD56^dimCD57⁻ and CD56^dimCD57⁺) differed from the NKG2C⁻ cells in NKG2A and KIR2DL2/DL3 expression.

We have shown that CD161 expression was reduced in all the NKG2C⁺ fractions (Fig. 4A, Supplementary Fig. S2A). It has been previously described that epigenetic rearrangements during the formation of the CD57⁺NKG2C⁺ adaptive NK cell subset lead to a decrease in CD161 expression.³⁸ Additionally, both CD56^dimCD57⁻NKG2C⁺ and CD56^dimCD57⁺NKG2C⁺ subsets had lower expression of CD16 and a smaller proportion of CD16⁺ NK cells compared to the NKG2C⁻ counterparts, although generally CD16 expression was high in all СD56^dim cells (Fig. 4B; Supplementary Fig. S2B). At the same time, there were more CD16⁺ NK cells in the CD56^brightNKG2C⁺ subset compared to the CD56^brightNKG2C⁻ subset (Supplementary Fig. S2B). Apparently, lower CD16 expression in CD56^dimNKG2C⁺ NK cells may be a result of the observed down-regulation of FcεRIγ, the adapter molecule, visible in NKG2C⁺ cells at the CD56^dim differentiation stage (Fig. 4C). Decreased FcεRIγ expression, caused by epigenetic changes during the formation of the HCMV-associated adaptive pool, is an important feature of memory-like CD56^dimCD57⁺NKG2C⁺ NK cells.³⁹ NKp30 expression was also reduced in NKG2C⁺ cells from both CD56^dimCD57⁻ and CD56^dimCD57⁺ subsets (Fig. 4D; Supplementary Fig. S2C). At the same time, we did not reveal considerable differences in NKp46 expression between NKG2C⁺ and NKG2C⁻ NK cells at any of the stages: NKp46 expression decreased gradually through the differentiation process (Fig. 4E, Supplementary Fig. S2D). Altogether, the phenotypic changes characteristic of adaptive NK cells were observed not only in the most differentiated CD57⁺NKG2C⁺ subset, but also in the CD56^dimCD57⁻NKG2C⁺ subset, which indirectly indicates that CD56^dimCD57⁻NKG2C⁺ NK cells can be considered as an earlier stage in the development of adaptive NK cells.

FIGURE 4. Ex vivo phenotype of NK cells differing in maturation stage and NKG2C expression.

Expression level of (A) CD161 (n = 11), (B) CD16 (n = 27), (C) FcɛRIγ (n = 19), (D) NKp30 (n = 10), (E) NKp46 (n = 12), (F) CD2 (n = 17), (G) CD44 (n = 19) are represented. To determine statistical significance, 1-way ANOVA was used

It has been previously shown that adaptive NK cells are characterized by more intense expression of the co-stimulatory molecule CD2.¹⁶ In our donor cohort, about 70% of freshly isolated NK cells expressed CD2 (Supplementary Fig. S2E). CD56^bright cells expressed this molecule at a considerably higher level than CD56^dim cells. In all NKG2C⁺ subsets an increased level of CD2 expression was also observed (Fig. 4F), along with the increased expression of KIR receptors (Fig. 3C–F). Moreover, each NKG2C⁺ subset contained more CD2⁺ NK cells compared to the respective NKG2C⁻ subset (Supplementary Fig. S2E). CD44, a hyaluronic acid receptor that plays a role in intercellular adhesion and lymphocyte homing,⁴⁰ was expressed on more than 98% of freshly isolated NK cells (Supplementary Fig. S2F). CD44 expression per cell was slightly up-regulated in CD56^dimCD57⁻NKG2C⁺ and CD56^dimCD57⁺NKG2C⁺ NK cells compared to the NKG2C⁻ counterparts (Fig. 4G). The level of CD44 expression in CD56^bright NK cells was significantly higher than in CD56^dim, but lower in the CD56^brightNKG2C⁺ subset compared to CD56^brightNKG2C⁻ (Fig. 4G).

Thus, it can be concluded that NKG2C⁺ NK cells at all the observed differentiation stages possess the features of more mature cells compared to the NKG2C⁻ counterparts, including certain “adaptivity” characteristics, like, for example, the lower level of CD16 expression. The CD56^dimCD57⁻NKG2C⁺ subset appears to be closer to the CD56^dimCD57⁺NKG2C⁺ subset by a number of phenotypic features, than the CD56^dimCD57⁺NKG2C⁻ one. A decrease in the expression of FcεRIγ, CD16, NKp30, and CD161 indicates possible epigenetic rearrangements already occurring in NKG2C⁺ NK cells at the CD56^dimCD57⁻ differentiation stage.

3.3 |. The formation of specific adaptive features occurs along with NKG2C⁺ NK cell activation evaluated by HLA-DR expression ex vivo

HLA-DR is another marker, surface expression of which appeared to be increased in NKG2C⁺ NK cells (Fig. 5A and B). HLA-DR is considered a marker of NK cell activation, although no relationship with the expression of other activation markers, such as CD25 and CD69, has been found.²² We confirmed the previously obtained data²⁴ that the least differentiated CD56^bright NK cells have the highest level of HLA-DR expression. However, there were no significant differences in the proportion of HLA-DR⁺ NK cells and HLA-DR expression level between the CD56^brightNKG2C⁻ and CD56^brightNKG2C⁺ subsets (Fig. 5A; Supplementary Fig. S2G). Among the CD56^dim NK cells, the highest HLA-DR expression levels were observed in both NKG2C-positive subsets. Interestingly, HLA-DR expression pattern in NKG2C⁺ NK cells from the HCMV-seronegative donors repeated the trends observed for NK cells from the HCMV-seropositive donors (Fig. 5A; Supplementary Fig. S2G). Apparently, HLA-DR expression is associated with the expression of NKG2C in NK cells, and reflects an activated status of NKG2C⁺ NK cells at all differentiation stages. Moreover, among the CD56^dim NK cells, HLA-DR expression is a feature that distinguishes NKG2C-positive NK cells from the NKG2C-negative.

3.4 |. Functional activity of NK cells depends on NKG2C expression and differentiation stage

An important indicator of the NK cell functional activity is the expression of granzyme B (GrB), a serine protease by which NK cells carry out the killing of their targets.⁴¹ We analyzed the content of GrB in six cell fractions listed above. The level of GrB expression in NK cells increased according to the following scheme: CD56^bright < CD56^dimCD57⁻ < CD56^dimCD57⁺. Significant difference in GrB expression between NKG2C⁻ and NKG2C⁺ NK cells at the same differentiation stage was observed only in the CD56^dimCD57⁺ subset: CD56^dimCD57⁺NKG2C⁺ cells had the maximum level of GrB (Fig. 6A).

FIGURE 6. Ex vivo analysis of functional activity of NK cell subsets differing in NKG2C expression and differentiation stage.

(A) Granzyme B expression levels in NK cell subsets (n = 13). (B and C) Proportion of degranulating cells in NK cell subsets in presence of stimuli (IL-2 500 U/ml for 20 h). (B) K562-mediated degranulation. “NK” – bulk NK cell population without target cells (K562), experimental samples included bulk NK cell population + K562 cells (n = 9). (C) Ab-induced degranulation. “NK+Rm” - bulk NK cell population + Rituximab without target cells (C1R), “NK+C1R” - bulk NK cell population + target cells (C1R) without Rituximab, experimental samples included bulk NK cell population + Rituximab + C1R cells (n = 6). NK cell subsets differing in NKG2C expression and differentiation stage were isolated by gating. To determine statistical significance, 2-way ANOVA was used for (A) and paired Student’s t-test for (B) and (C). (D) K562-induced degranulation of NK cells from different subsets. Representative staining of cells from 1 donor out of 9 donors examined is shown

We also evaluated natural and Ab-dependent degranulation responses in NK cells from different subsets, using standard K562 and anti-CD20 Ab-coated C1R target cells, respectively (Fig. 6B–D). As there is a direct correlation between CD107a expression and both cytokine secretion and NK cell-mediated lysis of target cells,³⁶ we regarded the proportion of CD107a⁺ NK cells as a marker of their cytotoxic (natural or Ab-dependent) activity. IL-2-stimulated CD56^bright NK cells demonstrated high level of natural cytotoxicity towards target K562 cells (Fig. 6B). We noticed a tendency toward an increased CD107a⁺ NK cell proportion in the CD56^dimCD57⁻NKG2C⁺ and CD56^dimCD57⁺NKG2C⁺ subsets compared to the NKG2C⁻ counterparts (Fig. 6B). The proportion of degranulating cells in the CD56^dimCD57⁻NKG2C⁺ subset was above 55% in 4 of the 5 donors. The ability to degranulate toward anti-CD20-coated C1R cells (Ab-dependent cellular cytotoxicity, ADCC) increased gradually throughout the NK cell differentiation stages. A tendency for higher degranulation activity of NKG2C⁺ cells compared to the NKG2C⁻ counterparts was shown for all the studied subsets (Fig. 6C). Thus, despite the fact that NKG2C⁺ NK cells have a reduced expression level of the natural cytotoxicity receptor NKp30, ADCC receptor CD16, and the adapter molecule FcɛRIγ, they exhibit high cytolytic activity after stimulation.

We also analyzed IFN-γ production by freshly isolated NK cells after cytokine and Ab-dependent contact stimulation. As it was expected, after pre-incubation with IL-12 and IL-18 CD56^bright NK cells had more IFN-γ⁺ cells compared to CD56^dim cells. Additionally, we noticed a tendency toward a decrease in IFN-γ production gradually through the NK cell differentiation process (Supplementary Fig. S3A). We did not find any significant differences in IFN-γ expression between NKG2C⁺ and NKG2C⁻ NK cells at any of the stages independently of stimulation, perhaps due to individual differences between the donors. However, we observed a tendency toward an increase in the “Ab-dependent” IFN-γ production induced by co-incubation with Raji cells coated with anti-CD20 Ab in more differentiating NK cells compared to less differentiating cells (Supplementary Fig. S3B). A significant difference between NKG2C⁺ subset and NKG2C⁻ counterparts on cell proportion producing IFN-γ in response to Ab-dependent stimulation was shown only within less differentiated CD56^bright subset (Supplementary Fig. S3B).

3.5 |. CD57⁻NKG2C⁺ NK cells demonstrate the best proliferative activity and survival among studied fractions during in vitro stimulation

Next, we compared proliferative capacity of the NKG2C⁺ and NKG2C⁻ NK cells at distinct stages of differentiation, pre-isolated by cell sorting, during in vitro cultivation in the presence of IL-2 and K562mbIL-21 feeder cells. The greatest expansion was obtained in cultures derived from the CD57⁻NKG2C⁺ NK cell fraction, which included CD56^bright and CD56^dimCD57⁻ cells (Fig. 7A). To evaluate the contribution of the CD56^dimCD57⁻NKG2C⁺ NK cells to the observed expansion, in a series of experiments CD56^bright NK cells were excluded from sorting, and four corresponding CD56^dim fractions were additionally evaluated (Fig. 7B). After 2 weeks of stimulation, the highest number of cells was observed in the CD56^dimCD57⁻NKG2C⁺-derived NK cell cultures. Lifespan of this subset was also the highest and reached 1 month and above.

FIGURE 7. Analysis of expansion and viability of isolated NK cell subsets, differing in NKG2C expression and stage of differentiation, and clones obtained from these subsets under stimulation with IL-2 and K562-mbIL21 feeder cells.

(A and B) Dynamics of cell number in NK cell subsets sorted from (A) total NK cells (n = 7, three representative donors are shown) or (B) CD56^dim NK cells (n = 2). (C) Cloning efficiency in NK cell subsets, differing in NKG2C expression and stage of differentiation (n = 5). (D) Percentage of NK cell clones obtained from different subsets, survived to the indicated time points of cultivation (n = 5). To determine statistical significance paired Student’s t-test was used for (C) and 2-way ANOVA for (D)

Even on day 21 of cultivation, >40% of NK cells from the CD56^dimCD57⁻NKG2C⁺ subset were in the cell cycle, which was measured by Ki-67 protein expression and Hoechst DNA staining (Supplementary Fig. S4A). Besides, CD57⁻NKG2C⁺-derived cultures maintained a high level of natural cytotoxicity after 3 weeks of cultivation. The proportion of degranulating cells in this subset reached 70%, however, a high level of spontaneous degranulation was observed, possibly resulting from cultivation in the presence of K562-mbIL21 feeder cells⁴² (Supplementary Fig. S4B).

NK cells from the fractions CD56^brightNKG2C⁻, CD56^bright NKG2C⁺, CD56^dimCD57⁻NKG2C⁻, CD56^dimCD57⁻NKG2C⁺, CD56^dimCD57⁺NKG2C⁻, and CD56^dimCD57⁺NKG2C⁺ were analyzed at the clonal level. Clonal cultures were obtained using IL-2 and K562-mbIL21 feeder cells for stimulation, as described previously.⁴² Volunteers with at least 10% of NKG2C⁺ NK cells were selected for cloning. Whereas the greatest cloning efficiency was observed in the CD56^bright subsets, in consistency with our earlier data,³⁴ the frequencies of clone formation in the CD56^dimCD57⁺ and CD56^dimCD57⁻ subsets did not differ from each other (Fig. 7C). However, the amount of clonal cultures with cell number >3 × 10⁵ cells (large clones) varied between subsets, depending on stage of differentiation, NKG2C expression, and cultivation period. At the sixth week of cultivation, the maximum number of large clones (more than 8 clones) was detected in a set of clones derived from the CD56^bright NK cell subsets. After the seventh week of cultivation, the maximum number of large clones was observed among CD56^dimCD57⁻NKG2C⁺-derived clonal cultures (Supplementary Fig. S5). There were more large clones in clone collections derived from the CD56^dimNKG2C⁺ NK cells compared to NKG2C⁻ counterparts. The clone survival after 5–8 weeks of cultivation (the percentage of alive clones from the initial clone number) was the highest for the clones derived from the CD56^dimCD57⁻NKG2C⁺ NK cells (Fig. 7D). The survival of clones obtained from all the NKG2C⁺ NK cell subsets was higher compared with the clones from the corresponding NKG2C⁻ subsets (Fig. 7D). Thus, NKG2C-expressing NK cells are prone to more intensive proliferation, especially at the CD56^dimCD57⁻ differentiation stage, and, possibly, have higher viability of the progeny.

3.6 |. CD57 and NKG2C expression in NK cells during in vitro cultivation is reversible

We have demonstrated in pre-sorted cultures that less differentiated CD57⁻ subsets begin to express CD57 upon NK cell stimulation with IL-2 and K562mbIL-21 feeder cells (Fig. 8A and B). The proportion of CD57⁺ NK cells in the CD57⁻NKG2C⁻ subset on the 4th day of cultivation ranged from 20% to 30%, and in the CD57⁻NKG2C⁺ subset – from 4% to 50%. An increased number of CD57⁺ NK cells in the CD57⁻NKG2C⁺ subset (> 40%) was observed in the individuals with high proportion of CD57⁺NKG2C⁺ NK cells ex vivo (22% and 37%). However, the percentage of CD57⁺ cells in initially CD57⁻ subsets was on average significantly lower than in initially highly differentiated CD57⁺ subsets. During further cultivation, there was registered a decrease in the proportion of CD57-positive cells in the cultures derived from the CD57⁻NKG2C⁻ and CD57⁻NKG2C⁺ subsets (Fig. 8A), apparently, due to relatively high proliferative activity of the CD57⁻ NK cells and gradual extinction of the CD57⁺ NK cells. In the initially CD57⁺ subsets, a gradual decrease in the percentage of CD57⁺ NK cells was observed (Fig. 8A and B), which suggests that NK cells are able to lose the expression of this molecule. In clonal cultures obtained from the NKG2C⁺ and NKG2C⁻ NK cells at distinct stages of differentiation, similar CD57 expression patterns were observed (Fig. 8C). Interestingly, clonal cultures derived from the CD57⁺NKG2C⁺ NK cells still consisted mainly of highly differentiated cells with CD57⁺ phenotype, in contrast to clones from the CD57⁺NKG2C⁻ subset, where the majority of proliferating cells have lost CD57 expression (Fig. 8C).

FIGURE 8. CD57⁺ cell dynamics in (A and B) isolated NK cell subsets, differing in NKG2C expression and stage of differentiation, and in (C) clones obtained from these subsets and cultivated under stimulation with IL-2 and K562-mbIL21 feeder cells.

(A) Data obtained from 12 donors are shown. (B) Representative staining of NK cells from 1 donor out of 12 donors (7 days of cultivation) is presented. Autofluorescence control is light grey, sample is dark grey. (C) Data obtained from 7 donors are presented. To determine statistical significance, unpaired Student’s t-test was used

We also analyzed the stability of NKG2C receptor expression in the bulk and clonal cultures derived from the NK cells with different initial phenotypes. As a rule, NK cells that were initially negative did not acquire NKG2C, although in two donors out of twelve a small fraction of NK cells weakly expressing NKG2C was registered during cultivation (up to 13% in CD57⁻NKG2C⁻ and up to 2% in CD57⁺NKG2C⁻ subsets) (Fig. 9A and B). In collections of clones derived from the CD57⁻NKG2C⁻ NK cells, up to 12% of clones started to express NKG2C at a low level after 5 weeks of cultivation, in contrast to clones from the CD57⁺NKG2C⁻ subset, none of which acquired NKG2C expression. Both in pre-sorted bulk cultures and clonal cultures derived from the NKG2C⁺ NK cells, a decrease in the NKG2C⁺ cell proportion was observed during cultivation (Fig. 9A–C). Moreover, among the NKG2C⁺-derived clones, the loss of NKG2C expression occurred most often in the offspring of less differentiated CD57⁻ cells (Fig. 9C). In clone collections obtained from the CD57⁺NKG2C⁺ subset, only 1 clone out of 23 did not express NKG2C after 5 weeks of cultivation.

FIGURE 9. NKG2C⁺ cell dynamics in (A and B).

isolated NK cell subsets, differing in NKG2C expression and stage of differentiation, and in (C) clones obtained from these subsets, cultivated under stimulation with IL-2 and K562-mbIL21 feeder cells. (A) Data obtained from 12 donors are shown. (B) Representative staining of NK cells from 1 donor out of 12 donors. (C) Data obtained from 7 donors are presented. To determine statistical significance, unpaired Student’s t-test was used

Thus, during in vitro cultivation, NK cells are able to exhibit certain plasticity of their phenotypic traits. Under stimulation with IL-2 and K562mbIL-21 feeder cells, NK cells are capable of both acquiring the expression of CD57 and NKG2C de novo and losing it. Highly differentiated CD57⁺NKG2C⁺ NK cells appeared to be the least pliant, and mostly retained their phenotypic properties.

3.7 |. NKG2C⁺ NK cells are better activated when cultured in vitro compared to NKG2C⁻ NK cells

According to our earlier publication,²⁴ NK cell stimulation with IL-2 and K562mbIL-21 leads to an increase in the proportion of HLA-DR⁺ cells. In this work, this phenomenon was studied in the fractions of NK cells already mentioned above (Fig. 10). Cultivation of NKG2C⁻ and NKG2C⁺ CD57^+/− NK cells and NK cell clones in the presence of IL-2 and K562-mbIL21 feeder cells lead to an up-regulation of HLA-DR expression (Fig. 10A–C). At the clonal level, clones derived from the CD56^dimNKG2C⁺ NK cell subsets had a significantly higher HLA-DR expression and more HLA-DR⁺ NK cells than clones from the corresponding NKG2C⁻ subsets (Fig. 10D; Supplementary Fig. S6). Moreover, the highest level of HLA-DR expression after 5 weeks of cultivation was observed in NK cell clones from the CD56^dimCD57⁻NKG2C⁺ subset (Supplementary Fig. S6).

FIGURE 10. HLA-DR⁺ cell dynamics in (A and B) isolated NK cell subsets, differing in NKG2C expression and stage of differentiation, and in (C ands D) clones obtained from these subsets, cultivated under stimulation with IL-2 and K562-mbIL21 feeder cells.

(A) Statistical difference within 1 subset between different days of cultivation was evaluated by paired t-tests. Statistical difference between subsets was evaluated by 2-way ANOVA multiple comparisons. Data obtained from 12 donors are shown. (B) Representative staining of CD57⁻NKG2C⁻ and CD57⁻NKG2C⁺ NK cell subsets from 1 donor out of 12 is shown. (C and D) Data obtained from 7 donors are represented. To determine statistical significance, unpaired Student’s t-test was used

Thus, the association between NKG2C expression and increased cellular activation, evaluated by HLA-DR expression, was confirmed both ex vivo and after in vitro stimulation.

4 |. DISCUSSION

A subset of NK cells expressing the NKG2C activating receptor is often found in people infected with cytomegalovirus. In some individuals, the proportion of NKG2C⁺ NK cells can reach 40–60% or more of all NK cells⁴³; in our study, the maximum value was 78%. Most of these cells comprise the adaptive NK cell subset, which differs from the conventional cells phenotypically and functionally due to the epigenetic changes that they undergo during activation, differentiation, and expansion.^6,38,44 The adaptive-like NK cells are phenotypically characterized by almost complete absence of NKG2A receptor, decreased expression of NKp30, NKp46, CD161, and increased expression of KIRs and CD57.^38,45,46 In this work, we focused on NKG2C⁺ NK cells, which lack surface CD57 expression. We examined these cells in two fractions – less differentiated CD56^bright and more differentiated CD56^dim, and compared them with the classical adaptive CD57⁺NKG2C⁺ NK cells.

Most interestingly, we have found that NKG2C⁺ NK cells in the CD56^dimCD57⁻ and CD56^dimCD57⁺ fractions demonstrated many similar phenotypic traits. Both NKG2C-expressing CD56^dim subsets had reduced proportion of NKG2A⁺ cells and increased proportion of cells expressing KIR2DL2/DL3; they were characterized by lower expression of NKp30, CD161, CD16, and higher expression of CD2 and CD44 receptors, compared to the corresponding NKG2C-negative fractions (Figs. 3 and 4). Additionally, a lower expression of FcɛRIγ adapter chain in CD56^dimCD57⁻NKG2C⁺ and CD56^dimCD57⁺NKG2C⁺ NK cells was registered compared to their NKG2C-negative counterparts (Fig. 4C). Down-regulation of FcɛRIγ is a typical consequence of epigenetic rearrangements during the formation of adaptive NK cell subset.⁴⁵ Based on similarities of the expression patterns, we hypothesized that CD56^dimCD57⁻NKG2C⁺ cells are mostly the precursors of the classical CD56^dimCD57⁺NKG2C⁺ adaptive NK cells or, more precisely, these are adaptive-like NK cells at an intermediate stage of their maturation.

The acquisition of NKG2C in CD56^bright NK cells was associated with a decrease in NKG2A and an increase in KIR expression levels (Fig. 3B, D, and F). The similar NKG2A and KIR expression changes are characteristic for the NK cell differentiation from CD56^bright to CD56^dim stage.⁴⁷ Based on this we can conclude that CD56^brightNKG2C⁺ NK cells are more differentiated compared to CD56^brightNKG2C⁻ cells. Perhaps higher expression level of KIR receptors on CD56^brightNKG2C⁺ NK cells is essential for their education. It was shown that NK cell become functionally competent only after recognition of self-MHC-I by inhibitory receptors, primarily by KIRs.⁴⁸ After licensing, NK cells may reduce the expression level of inhibitory KIRs. Also, we should not exclude the point of view that CD56^bright and CD56^dim subsets can be 2 separated lineages with different origin,⁴⁹ which can also explain the differences in the expression level of KIR receptors shown in our study. The expression intensity of other markers, such as NKp30 and CD16 was mostly similar in NKG2C-positive and negative CD56^bright NK cells, but changed significantly at more advanced stages of differentiation and began to differ between NKG2C⁺ and NKG2C⁻ subsets (Fig. 4). At the same time, expression pattern of CD161 and CD2 was apparently associated with NKG2C (Fig. 4A and F). It was previously shown that the level of CD44 expression is higher on memory-like NK cells.⁵⁰ Here, we observed that CD44 expression was down-regulated in CD56^brightNKG2C⁺ NK cells and up-regulated in both CD56^dimCD57⁻NKG2C⁺ and CD56^dimCD57⁺NKG2C⁺ NK cells, compared to their NKG2C⁻ counterparts, showing differently biased expression in CD56^bright and CD56^dim subsets (Fig. 4G). Thus, the observed differences in the marker expression between NK cell subsets are more or less linked to 3 interrelated but still different processes: “classical” differentiation (changes in NKG2A, KIR, NKp46, GrB expression), acquisition of the NKG2C receptor (changes in CD2, CD161 expression) and the formation of a pool of adaptive NK cells associated with HCMV infection (changes in CD16, FcɛRIγ, NKp30, CD44 expression).

In our previous work and in the publications of other research groups, it has been shown that the adaptive-like NKG2C⁺ NK cells associated with HCMV have an increased expression level of the HLA-DR activation marker.^13,22 This expression was not associated with the other activation markers (CD69 and CD25) or markers of the adaptive NK cell differentiation process (FcεRIγ, CD57), but remained stable over time at the individual level.²² In this work, we have demonstrated that not only CD56^dimCD57⁺NKG2C⁺, but also CD56^dimCD57⁻NKG2C⁺ NK cells have increased level of HLA-DR expression compared to the NKG2C-negative counterparts. Moreover, the HLA-DR expression pattern looked similar in HCMV-seropositive and seronegative individuals (Fig. 5A), suggesting that HLA-DR is associated with NKG2C expression rather than with HCMV infection. We have also shown previously, that NK cells up-regulate HLA-DR expression under stimulation with IL-2+K562-mbIL21.²⁴ In this work, we also registered an increase in the proportion of HLA-DR⁺ NK cells upon stimulation in all of the studied subsets, both in bulk and clonal cultures (Fig. 10). The highest levels of HLA-DR expression were observed in the CD57^+/−NKG2C⁺ subsets and clones derived from them. Thus, we confirmed ex vivo and in vitro that the expression of NKG2C on NK cells is associated with high level of cellular activation, as determined by HLA-DR expression.

In contrast to HLA-DR, CD16 was down-regulated in both CD56^dimCD57⁻NKG2C⁺ and CD56^dimCD57⁺NKG2C⁺ subsets. CD16 expression increased during the differentiation process, but decreased with the acquisition of NKG2C in CD56^dim cells (Fig. 4B). Interestingly, this decrease had no effect on Ab-dependent cytotoxicity (Fig. 6C). Currently, there are studies suggesting that conventional and adaptive NK cells do not differ in CD16 expression,⁵¹ and alternative studies that do not support this point of view.^14,52,53 According to our observations, the level of CD16 expression varied depending on compared subsets and studied donors. The reasons of such variations are unknown and require additional research. In earlier work, it was shown that an increase in HLA-DR surface expression and a concomitant loss of CD16 in the adaptive subset can be induced by incubation of purified NK cells with the HCMV-Ab complex.²² Suggestively, the similar processes may take place in vivo in HCMV-infected individuals.

NKG2C⁺ adaptive NK cells are believed to have reduced natural cytotoxicity due to a decrease in the expression level of the corresponding receptors.^13,14 However, we demonstrated that CD56^dimNKG2C⁺ NK cells had the same or even better degranulation activity compared to NKG2C⁻ counterparts towards K562 target cells (Fig. 6B and D). The CD56^dimCD57⁻NKG2C⁺ subset had a significantly higher level of degranulation compared to its NKG2C⁻ counterpart, and retained this high level of natural cytotoxicity even after prolonged cultivation in stimulating conditions (Fig. 6; Supplementary Fig. S4). The observed high level of degranulation among CD56^bright NK cells can be explained by their better responsiveness to preliminary cytokine stimulation. At the same time, the level of granzyme B expression was related mostly to the stage of differentiation and had little association with NKG2C expression, being the highest in the terminally differentiated CD57⁺NKG2C⁺ NK cell subset. Overall, natural cytotoxic activity of the NK cells had no direct correlation with granzyme B expression. Increased cytotoxic activity of the CD56^dimCD57⁻NKG2C⁺ and CD56^dimCD57⁺NKG2C⁺ NK cells could result from the increased CD2 expression in these subsets (Fig. 4F). It has been previously shown that adaptive NK cells highly express CD2: this co-stimulatory molecule is likely to provide “signal 2” during CD16-mediated recognition of the Ag-Ab complex.¹⁶ CD2 may act as a co-stimulatory molecule during natural cytotoxic responses too, generally enhancing degranulation activity.

Despite the fact that we showed clear changes in the production of IFN-γ in the process of differentiation of NK cells from CD56^bright to CD57⁺ stages for both cytokine-dependent and CD16-dependent IFN-γ production, we did not find significant differences between the NKG2C⁻ and NKG2C⁺ fractions on the IFN-γ production within highly differentiated CD57⁺ subset. It may be connected to the fact that in an earlier work an increased level of IFN-γ production in adaptive-like NK cells was shown compared with the general population of conventional NK cells, without dividing them by differentiation stages.⁵⁴ In addition, it should be noticed that some NKG2C-negative NK cells may also possess some features of adaptive-like cells,^16,20 and individual differences between donors may contribute to the average level of IFN-γ production in NKG2C⁺ and NKG2C⁻ fractions.

During in vitro stimulation and clonal expansion experiments less differentiated CD57⁻NKG2C⁺ NK cells demonstrated the best proliferative activity and were also the most long-lived subset due to long-term proliferation (Fig. 7). When CD56^bright NK cells were excluded from cell sorting, the CD56^dimCD57⁻NKG2C⁺ subset still demonstrated the greatest expansion (Fig. 7B). At the same time, no differences in cloning efficiency were registered between NKG2C⁺ and NKG2C⁻ NK cells from both the CD56^bright and CD56^dim subsets (Fig. 7C). Still, we confirmed the previously obtained results that the CD56^bright NK cells demonstrate greater cloning efficiency than CD56^dim.⁴² Further analysis showed that the maximal number of highly expanded clones was obtained from the CD56^dimCD57⁻NKG2C⁺ subset (Fig. S5). Thus, despite all the expression pattern similarities, the critical difference between CD56^dimCD57⁻NKG2C⁺ NK cells and classical adaptive CD56^dimCD57⁺NKG2C⁺ NK cells is high proliferative potential of the first subset, presumably related to the lack of CD57 expression. The association between high level of CD57 expression in NK cells and their low proliferation activity has been reported earlier.⁵⁵

When investigating the stability of CD57 and NKG2C expression on NK cells during cultivation, we observed the loss of CD57 by highly differentiated CD57⁺ NK cells, both in pre-sorted subsets and in clones (Fig. 8). Such a phenomenon has been previously described by us and other groups.^33,34 Also, under stimulation, less differentiated CD57⁻ subsets partly acquired CD57 expression (Fig. 8), which may reflect the effect of “aging” observed previously in primary NK cell cultures.⁵⁶ Next, we have shown for the first time that NK cells are also able to lose NKG2C expression. Besides, NK cells partly acquired NKG2C expression de novo in our stimulating conditions (Fig. 9). Phenotypic plasticity in response to stimulation varied among the studied subsets. The most stable expression of CD57 and NKG2C was observed in the clones derived from the highly differentiated CD56^dimCD57⁺NKG2C⁺ subset (Figs. 8 and 9), regardless to their expansion rate (Supplementary Fig. S5). This fact suggests that CD57 does not unconditionally mark matured NK cells with reduced proliferative activity, although CD57 expression is believed to appear at the final differentiation stage and be irreversible.³¹

We assume that the induction of NKG2C expression de novo in the progeny of CD57⁻ NK cells demonstrated in our experiments can be compared to the processes occurring in the HCMV-seronegative donors. We have shown in the first part of the study that such people have a small subset of NKG2C⁺ NK cells, although, apparently, they have never encountered the virus. Moreover, the highest proportion of NKG2C⁺ NK cells in such people was observed in the CD56^dimCD57⁻ subset, compared to both CD56^bright and CD56^dimCD57⁺ subsets (Fig. 1C). It is still unknown which signal induces the initial appearance of NKG2C⁺ NK cells. Perhaps, NKG2C expression emerges sporadically at certain steps of NK cell differentiation in response to activating conditions. On the one hand, high percentage of CD56^dimCD57⁻NKG2C⁺ NK cells in HCMV-seronegative individuals might result from an ongoing immune response to some other viral infection. On the other hand, these cells may appear independently of the viral load and carry out a supervisory function in the body: in case of active HCMV infection they rapidly expand and differentiate into “classical” adaptive CD57⁺NKG2C⁺ NK cells for the effective elimination of the virus. It has been shown recently that NKG2C⁺ NK cells are able to recognize certain HCMV peptides in the context of HLA-E, resulting in accumulation and differentiation of NKG2C⁺ NK cells even in the cell cultures from HCMV-seronegative donors.⁵⁷ This mechanism can be used by CD56^dimCD57⁻NKG2C⁺ NK cells for “sensing” the virus. Apparently, expanded CD56^dimCD57⁻NKG2C⁺ NK cell subset may be encountered in the blood of primary infected individuals during the active development of infection or in individuals with HCMV reactivation. Verification of this hypothesis is a matter for further investigation. A hypothetical outcome of such immune response in HCMV-seropositive individuals, after the active phase of HCMV infection has finished, is the formation of a highly expanded persistent subset of CD56^dimCD57⁺NKG2C⁺ cells with the increased expression of granzyme B and still detectable subset of CD56^dimCD57⁻NKG2C⁺ NK cells.

A decrease in NKG2C expression in NK cells during cultivation in the absence of the HCMV-derived stimulus, observed in our in vitro experiments, gives a hint why seronegative individuals possess lower rate of CD57⁺NKG2C⁺ NK cells compared to CD57⁻NKG2C⁺ cells (Fig. 1C and D). CD57⁺NKG2C⁺ subset may seem comparably low because of preferential proliferation of CD57⁻ NK cells and/or transition of CD57⁺NKG2C⁺ NK cells back to CD56^dimCD57⁻NKG2C⁺ or CD56^dimCD57⁺NKG2C⁻ stage in the absence of the appropriate stimulus.

In conclusion, in this work we characterized a special subset of adaptive-like NK cells with the CD56^dimCD57⁻NKG2C⁺ phenotype, showing high proliferative activity during cultivation in the presence of IL-2 and K562-mbIL21. Presumably, CD57⁻NKG2C⁺ NK cells represent an earlier, transient stage of formation of the “classical” adaptive NK cells CD56^dimCD57⁺NKG2C⁺. CD56^dimCD57⁻NKG2C⁺ NK cells demonstrate activated phenotype and significant functional activity.

Supplementary Material

Supple fig S3

Figure S3. IFN-γ production by NK cells differing in maturation stage and NKG2C expression. A. Cytokine induced IFN-γ production. Freshly isolated NK cells were stimulated by IL12+IL18 (20 ng/ml for 20h). Data of eight donors is represented. B. Antibody-dependent IFN-γ production. Freshly isolated NK cells were stimulated by Raji cells (in ratio 4:1) in addition with Rituximab (2,5 μg/ml) and brefeldin A (10 μg/ml) for 5 h. Data of eight donors is represented. NK cell subsets differing in NKG2C expression and differentiation stage were isolated by gating. To determine statistical significance one-way ANOVA was used.

Supple fig S4

Figure S4. Proliferation and cytotoxic activity of CD57⁻NKG2C⁺ NK cell subset from one representative donor measured at day 21 of cultivation with IL-2 and K562-mbIL21 feeder cells. A. Proliferation activity was measured by Ki-67 expression and Hoechst staining. Autofluorescence control is light grey, sample is dark grey. B. Percentage of degranulating NK cells in CD57⁻NKG2C⁺ subset. “Unstimulated” - NK cells without target K562 cells; “stimulated” - NK cells with K562 cells.

Supple fig S5

Figure S5. Petal chart demonstrating changes in the number of highly proliferating NK cell clones (cell number over 300′000), obtained from different subsets, under stimulation with IL-2 and K562-mbIL21 feeder cells. Average values of NK cell clone number are plotted on the axis (n = 4).

Supple fig S1

Figure S1. CD57⁺ NK cell fraction analysis from donors with different HCMV serostatus. Proportion of CD57-positive cells in NK cell fraction of four different donor groups. Gr.1, Gr.2, Gr.3, Gr.4 stands for group 1 (no anti-HCMV Abs), group 2 (low titer), group 3 (medium titer), group 4 (high titer), and comprise 13, 21, 13 and 10 donors, respectively. To determine statistical significance one-way ANOVA was used.

Supple fig S6

Figure S6. Expression level of HLA-DR in clones, obtained from different subsets, at 5th week of cultivation with IL-2 and K562-mbIL21 feeder cells. Data obtained from 7 donors are presented. Statistical significance was determined by unpaired Student’s t-test.

Supple fig S2

Figure S2. Ex vivo phenotype of NK cells differing in maturation stage and NKG2C expression. Proportions of A. CD161⁺ (n = 11), B. CD16⁺ (n = 27), C. NKp30⁺ (n = 10), D. NKp46⁺ (n = 12), E. CD2⁺ (n = 17), F. CD44⁺ (n = 19) cells were measured. G. Expression level of HLA-DR in NK cell fractions in individuals with different HCMV serostatus. Data of 4 HCMV-seronegative and 22 HCMV-seropositive donors are shown. Statistical significance was determined by one-way ANOVA for A., B., C., D., E., F. and two-way ANOVA for G.