This study enrolled a total of 45 individuals, 22 of whom had a PHI and were being seen at the Red Ribbon Clinic in the First Affiliated Hospital of China Medical University. The HIV-infected individuals were all males, aged 19–52 years; 23 HIV-negative individuals served as NC subjects. The inclusion criteria for PHI subjects were as follows: aged >18 years; infected with HIV within 36–156 days at enrollment; antiretroviral therapy (ART) naïve; and followed-up for 720 days. The inclusion criteria for NC subjects were as follows: HIV-negative; normal blood routine examination and normal liver function; negative for hepatitis B and hepatitis C antibody; and no disease of the immune system (Table 1). The subjects had been tested for anti-CMV IgG antibody, anti-CMV IgM antibody, and CMV nucleic acid. The demographic, immunological, virological, and CMV infection data of all study subjects are shown in Table 1. The Research and Ethics Committee of The First Affiliated Hospital of China Medical University approved the protocol for this study, and each enrolled individual provided their written informed consent for participation in the study.

We used the following antibodies in this study: NKG2C-APC, NKG2C-PerCp, NKG2A-PE (R&D); CD3-FITC, CD4-APC-Cy7, CD16-PerCP-Cy5.5, CD56-PE-Cy7, CD107a-APC-H7 (BD Biosciences, San Jose, CA, USA); IFN-γ-BV421 (Biolegend, San Diego, CA, USA); CD4-FITC/CD8-PE/CD3-PerCP (BD Biosciences, San Jose, CA, USA).

A whole blood sample was collected from each subject by venipuncture and a cocktail of fluorescent antibodies was used against CD3, CD16, CD56, NKG2C, and NKG2A. Mixtures were incubated for 30 min at 4°C, and erythrocytes were lysed using red blood cell lysing solution. Three NK cell subsets were identified as CD3⁻CD56⁻CD16⁺, CD3⁻CD56^dimCD16⁺, and CD3⁻CD56^briCD16^+/−, and the total population of NK cells included these three subsets (Figure 1). The expression of NKG2A or NKG2C was detected on NK cell subsets, and total NK cells were analyzed by flow cytometry; digital data from scatter plots were analyzed with Diva software.

Freshly isolated peripheral blood mononuclear cells (PBMCs) were incubated in the presence of phorbol 12-myristate 13-acetate (0.1 µg/mL) and ionomycin (5 µg/mL) at 37°C for 6 h. The cells were then stained for a panel of cell-surface markers (CD3-PerCP, CD56-PE-Cy7, NKG2C-APC, and CD107a-APC-H7). Intracellular staining was performed by incubating cells in 1× Perm/Wash Buffer (Becton Dickinson) for 15 min in the dark, and then with BV421-conjugated anti-IFN-γ (Biolegend, USA) for 30 min at 4°C. After staining, the cells were fixed in 1% formaldehyde; CD107a and IFN-γ expression were detected by flow cytometry.

CD4⁺ T cells, NKG2C⁺ NK cells, and NKG2C⁻ NK cells were sorted from PBMCs of HIV-negative donors (N = 5) and HIV-positive donors (N = 7) by BD FACS Aria. CD4⁺ T cells were stimulated by PHA (50 ng/mL, Sigma) and IL-2 (20 U/mL, Sigma) for 36 h, while NK cells were stimulated by IL-12 (10 ng/mL, R&D), IL-15 (50 ng/mL, R&D), and IL-18 (100 ng/mL, R&D) for 36 h. CD4⁺ T cells were infected by HIV-1 NL4-3 strain and co-cultivated with autologous NKG2C⁺ NK cells, or NKG2C⁻ NK cells, at a target ratio (E:T) of 5:1 for 5 days. Supernatants were collected from each culture and tested for HIV-1 P24 antigen by ELISA (QuantoBio, Beijing, China).

CD4⁺ T cell counts were determined with a FACS Calibur flow cytometer (Becton Dickinson). A single-platform lyse-no-wash procedure was conducted with TriTEST anti-CD4-FITC/CD8-PE/CD3-PerCP reagents and Trucount tubes (Becton Dickinson).

Plasma HIV RNA levels were determined by reverse transcription polymerase chain reaction (RT-PCR) performed with the COBAS^®AmpliPrep^®/COBAS Taqman system (Roche Diagnostic Systems); this assay had a detection range of 20–10,000,000 copies/mL. HIV RNA copy numbers were calculated by using the manufacturer’s reference standards.

Levels of IgM and IgG against HCMV in plasma samples were detected with a chemiluminescence immunoassay (LIAISON^®CMV IgM II, LIAISON^® CMV IgG II, DiaSorin S.p.A., Saluggia, Italy). HCMV DNA was analyzed with an RT-PCR Kit (CMV Real Time PCR Kit, Liferiver, Shanghai, China). All samples were tested in the Department of Clinical Laboratory, at the First Affiliated Hospital of China Medical University.

Statistical comparisons of quantitative data between two groups were performed using the Mann–Whitney U test. The Kruskal–Wallis test was additionally used to compare significant differences across three groups. Associations between two groups were determined by the Spearman’s rank test. The non-parametric Wilcoxon matched-pairs signed-rank test was used to compare between two groups. Chi-square tests were conducted to examine differences in qualitative analysis. Receiver operating characteristic (ROC) curves and Kaplan–Meier survival analysis were used to evaluate the influence of NKG2C and NKG2A expression on NK cells upon the decline of CD4⁺ T cell counts. A p-value < 0.05 (two-tailed test) was considered to be statistically significant. All data analysis was performed using SPSS 20.0 (for three groups) and GraphPad Prism Version 5.0 software (for rest of data).

To investigate the alteration of NKG2A and NKG2C expression on NK cells during HIV infection, we conducted the assessment with flow cytometry, and the representative flow cytometry plots were showed in Figure 2A. First, we detected the single expression of NKG2A and NKG2C on NK cells of PBMCs from PHI and age-matched NC subjects. Results showed that the proportions of NKG2A expression were comparable between NC and PHI subjects (p > 0.05; Figure 2B). However, the proportions of NKG2C expression were higher in PHI subjects than in the NC subjects (p = 0.0039; Figure 2B). Most of the literature only tested the single expression of NKG2A or NKG2C on NK cells, but not the co-expression of these receptors. NKG2A⁺ NK cells can be either NKG2A⁺NKG2C⁻ NK or NKG2A⁺NKG2C⁺ NK cells. However, NKG2A and NKG2C play opposing roles in terms of NK cell function, and the function of NKG2A⁺ NK cells is likely to differ from that of NKG2A⁺NKG2C⁻ NK cells. Therefore, in this study, we deliberately evaluated the co-expression of NKG2A and NKG2C on NK cells. We found that the proportion of NKG2A⁺NKG2C⁻ NK cells was significantly lower in PHI subjects than that in NC subjects (p = 0.0143; Figure 2C), although there was no significant difference in terms of the expression of NKG2A alone. In addition, we also found that the proportion of NKG2C⁺NKG2A⁻ NK cells was significantly higher in PHI subjects than that in NC subjects (p = 0.0005; Figure 2C). Similar results were observed for three NK cell: CD56^dimCD16⁺, CD56^briCD16^+/−, and CD56⁻CD16⁺ (Figures S1A–C in Supplementary Material). We also tested the mean fluorescence intensity of NKG2C or NKG2A expression and found that there were no significant differences when compared between PHI and NC subjects (data not shown). As most people are infected by HCMV in a latent manner, we investigated the association between the proportion of NKG2C⁺ NK cells and HCMV infection. We tested anti-HCMV IgG, IgM, and nucleic acid of HCMV; and the data showed that there was no significant difference on the positive rates between NC and PHI subjects (Table 1). However, we found that the titers of anti-HCMV IgG were significantly higher in PHI subjects compared with NC subjects (p = 0.0035), and the proportions of NKG2C⁺ NK cells were upregulated in PHI subjects along with a higher titers of anti-HCMV IgG (p = 0.001; Figure 2D). We further removed the PHI subjects with current HCMV infection (positive for anti-HCMV IgM antibody) and compared the proportion of NKG2C⁺ NK cells between NC subjects and the remaining PHI subjects. This analysis confirmed that the proportions of NKG2C⁺ NK cells were significantly higher in PHI subjects without current HCMV infection (p = 0.0066; data not shown). Overall, these data showed that NKG2C⁺ NK cells expanded during PHI, and we further investigated their potential roles in HIV infection.

Because the proportion of NKG2C and NKG2A expression on NK cells was upregulated in PHI subjects compared with NC subjects, we next investigated whether NKG2C and NKG2A might exert influence on the viral set point, defined as the viral load about 120 days after infection (24, 25). PHI subjects were divided into two groups based on the viral set point: a low set point group (<10⁴ copies/mL) and a high set point group (>10⁴ copies/mL). Our results showed that the proportion of NKG2C⁺NKG2A⁻ NK cells was higher in the low set point group (p = 0.0355), while the proportion of NKG2A⁺ NK cells was lower in the low set point group (p = 0.0257) (Figure 3A).

We further analyzed the correlation of NKG2C and NKG2A expression with plasma levels of HIV RNA. NKG2C and NGK2A expression and viral load were detected simultaneously in the same samples; results showed that the proportion of NKG2C⁺NKG2A⁻ NK cells was negatively correlated with viral load (r = −0.567; p = 0.009), and the proportion of NKG2A and NKG2A⁺NKG2C⁻ NK cells was positively correlated with viral load (r = 0.650, p = 0.002; r = 0.457, p = 0.043) (Figure 3B). The proportion of NK cells expressing NKG2A was negatively correlated with the proportion of NK cells expressing NKG2C (r = −0.662; p = 0.002). Moreover, the proportion of NKG2A⁺NKG2C⁻ NK cells was negatively correlated with the proportion of NKG2C⁺NKG2A⁻ NK cells (r = −0.805; p < 0.0001; Figure 3C).

Our data showed that during PHI, the elevated proportion of NKG2C⁺NKG2A⁻ NK cells gathered in the group of low HIV set point and was negatively correlated with plasma level of HIV RNA. Therefore, NKG2C could play an important role in anti-HIV activity. We next investigated the secretion of IFN-γ and the expression of CD107a of NKG2C⁺ NK cells in HIV-negative healthy donors. NKG2C⁺ NK cells showed higher IFN-γ secretion and expression of CD107a than NKG2C^– NK cells (p = 0.043 and p = 0.043, respectively; Figure 4A). These results indicated that NKG2C⁺ NK cells possessed strong functional activities. However, whether these highly functional cells could also play roles in the control of HIV infection remains unknown. We further investigated the anti-HIV activity of NKG2C⁺ NK cells. NKG2C⁺ NK cells from HIV-negative donors showed only weak anti-HIV activity (measured by levels of HIV-1 P24 antigen). Remarkably, strong anti-HIV activity was observed in NKG2C⁺ NK cells from HIV-positive donors (p = 0.01 and p = 0.033, respectively; Figure 4B). Because of the high level of anti-HIV activity exhibited by NKG2C⁺ NK cells, the proportion of NKG2C expression on the surface of NK cells might be used as an important biomarker with which to predict the progression of HIV disease.

We followed 19 PHI subjects for 720 days and determined their CD4⁺ T cell counts and viral load every 3–6 months. We analyzed the relationship between NKG2A and NKG2C receptor expression on the surface of NK cells from PHI subjects and CD4⁺ T cell counts, or viral loads, at the time of 360-day infection. We found that PHI subjects with a CD4⁺ T cell count >500 cells/μL were associated with a higher proportion of NKG2C⁺NKG2A⁻ NK cells and a higher ratio of NKG2C/NKG2A (p = 0.022 and p = 0.045, respectively). Furthermore, individuals with a viral load <10⁴ copies/mL were associated with a higher proportion of NKG2C⁺NKG2A⁻ NK cells and a higher ratio of NKG2C/NKG2A (p = 0.007 and p = 0.022, respectively; Figure 5A). We then analyzed the relative power of the NKG2C⁺NKG2A⁻ NK cell proportion, and the NKG2C/NKG2A ratio, for predicting the progression of HIV disease (CD4⁺ T cell counts), as calculated by ROC analysis. Our results indicated that the NKG2C⁺NKG2A⁻ NK cell proportion had a predictive accuracy of 82.1% for disease progression (CD4⁺ T cell counts), as determined by the area under ROC curve (AUC), while the NKG2C/NKG2A ratio had a predictive power of 78.6% (p = 0.022 and p = 0.043, respectively; Figure 5B). The point with the largest sum of sensitivity plus specificity was taken as the cutoff point. Based on the ROC curve for the proportion of NKG2C⁺NKG2A⁻ NK cells, and the ratio of NKG2C/NKG2A, cutoff points of 35.45% and 1.7 were selected. Kaplan–Meier survival analysis further showed that the mean time required for CD4⁺ T cell counts to reach <500 cells/μL in the low NKG2C⁺NKG2A⁻ NK cell group (<35.45%) was significantly shorter than the mean time taken for the high NKG2C⁺NKG2A⁻ NK cell group (>35.45%; p = 0.015). In addition, the mean time required for CD4⁺ T cell counts to reach <500 cells/μL in the group with a low NKG2C/NKG2A ratio (<1.7) was significantly shorter than in the group with a high NKG2C/NKG2A ratio (>1.7) (p = 0.008; Figure 5C).

There was one particular limitation in our study which should be considered when interpreting our results. In our study, the choice of 10⁴ copies/mL, as a limit between low and high viral loads, to compare the proportion or ratio of expression of NKG2A or NKG2C on NK cells, may appear somewhat arbitrary. However, thus far, there are no commonly accepted criteria for viral load grouping, and we therefore selected 10⁴ copies/mL as a limit to group our subjects according to the specific characterization of the patients recruited.