IL-10⁺ NK and TGF-β⁺ NK cells play negative regulatory roles in HIV infection

Sixty-three individuals participated in this study, including 31 antiretroviral treatment-naïve HIV-infected patients (ART-naïve), 17 ART-treated HIV-infected patients (ART-treated), and 15 HIV-negative HCs. The 31 ART-naïve patients had a median CD4⁺ T cell count of 363 cells/μl (interquartile range [IQR] 277–421 cells/μl) and median viral load of 24,760 copies/ml (range, 374–388,000 copies/ml). ART-naïve patients were further grouped according to CD4⁺ T cell counts (CD4⁺ T ≥ or < 350 cells/μl groups) or viral load (viral load > 10⁴ or ≤10⁴ copies/ml groups). The 17 ART-treated patients had received ART for ≥2 years, and had undetectable viral loads (HIV RNA, < 20 copies/ml); the median duration of their virologic suppression at the time of sampling was 33 months (IQR: 24–60 months), and they had a median CD4⁺ T cell count of 586 cells/μl (IQR: 453–684 cells/μl). HCs were HIV-negative, anti-hepatitis C antibody-negative, and hepatitis B surface antigen-negative, with normal liver and kidney functions, and without immune-system diseases.

Peripheral blood mononuclear cells (PBMCs) were isolated from HIV-infected patients and HCs by Ficoll™ density gradient centrifugation. PBMCs were cultured for 6 h with recombinant IL-12 (rIL-12, 10 ng/ml) and recombinant IL-15 (rIL-15, 50 ng/ml) in a 5% CO₂ incubator at 37 °C. Total NK cells and NK cell subsets were defined by their expression of CD3 and CD56, detected using the anti-CD3 PerCP and anti-CD56 PE-Cy7 antibodies (BD Biosciences, San Jose, CA, USA), respectively. Cells were stained with intracellular anti-TGF-β-PE and anti-IL-10-APC (Biolegend, San Diego, CA, USA) and fixed in 1% paraformaldehyde before detection using a flow cytometer (LSR II; Becton Dickinson, Franklin Lakes, NJ, USA) and analysis with FACSDiva™ software (Becton Dickinson); the gating strategy is shown in Fig. 1.

PBMCs were co-cultured with different concentrations of rIL-10 and/or rTGF-β for 5 h, then stimulated with rIL-12 (10 ng/ml, R&D) and rIL-15 (50 ng/ml, R&D) for 24 h at 37 °C, and anti-CD107a-PE antibody (Biolegend, San Diego, CA, USA) added simultaneously. Cells were co-cultured in GolgiStop™ ((Becton Dickinson) for ≥5 h before the end of stimulation. PBMCs were collected, and anti-CD3-Percp and anti-CD56-PE-cy7 were used for surface staining. Cells were then stained for intracellular IFN-γ with anti-IFN-γ-APC antibody (BD Biosciences, San Jose, CA, USA), washed with phosphate-buffered saline, and fixed in 1% paraformaldehyde, followed by analysis using the LSR II flow cytometer.

PBMCs were co-cultured with different concentrations of rIL-10 and/or rTGF-β for 5 h. Then, cells were stimulated with PMA (10 ng/μL) and ionomycin (1 μg/μL) (Sigma Chemical St. Louis, MO, USA) for 6 h at 37 °C. PBMCs were collected, and surface-stained with anti-CD3-PerCP and anti-CD56-PE-Cy7. Anti-granzyme B-PE and anti-perforin-fluorescein isothiocyanate (FITC) (BD Biosciences, San Jose, CA, USA) were used for intracellular staining, and cells detected by flow cytometry (LSR II).

CD4⁺ T cell counts were determined using a flow cytometer (FACS Calibur; Becton Dickinson). High, medium, and low Trucount™ Control Beads (BD Biosciences, San Jose, CA, USA) were used to confirm the accuracy and quality of CD4⁺ T cell counts. A single-platform lyse-no-wash procedure was conducted with TriTEST™ anti-CD3-PerCP/CD4-FITC/CD8-PE reagents and Trucount tubes (BD Biosciences, San Jose, CA, USA) according to the manufacturer’s instructions.

Plasma HIV RNA levels were determined with by reverse transcription-polymerase chain reaction (RT-PCR) using the COBAS®AmpliPrep®/COBAS TaqMan™ HIV test v2.0 (Roche Diagnostic Systems, Basel, Switzerland). The detection range of this method is 20–10,000,000 copies/ml. The number of HIV RNA copies was calculated according to the manufacturer’s reference standard.

The non-parametric Mann–Whitney U-test was used for evaluation of differences between two groups, and the Kruskall–Wallis test employed for comparisons of multiple groups. The Friedman’s test was used for multiple pairwise comparisons. Correlation analysis was performed using the Spearman’s rank correlation test. All analyses were undertaken using SPSS v19.0 (IBM, Armonk, NY, USA). p < 0.05 was considered significant.

The clinical characteristics of ART-naïve and ART-treated patients, and HCs, are presented in Table 1. The sex and age of HIV-infected patients were matched with those of HCs. The percentages of total NK cells and NK subsets among lymphocytes in peripheral blood samples from the three groups were measured by flow cytometry.

The percentages of total NK cells in ART-naïve and ART-treated patient samples were lower than that in HCs (p = 0.014, p = 0.024) (Fig. 2a). In addition, the percentages of CD56^dim NK cells in ART-naïve patients were lower than those in HCs (p = 0.006) (Fig. 2b) and there was a tendency towards lower percentages of CD56^bright NK cells in ART-naïve patients compared with those of HCs (p = 0.067) (Fig. 2c).

PBMCs obtained from HCs, and ART-naïve, and ART-treated patients, were stimulated with rIL-12 and rIL-15. The percentage of IL-10⁺ or TGF-β⁺ NK cells was examined by intracellular staining and multi-color flow cytometry. Representative flow cytometry plots of IL-10 or TGF-β expression in NK cells from each participant group are shown in Fig. 3a. The percentage of IL-10⁺ NK cells in ART-naïve patients was higher than that in HCs (p = 0.015); moreover, the percentage of IL-10⁺ NK cells was also higher in the ART-treated HIV patient group than that in HCs (p = 0.002) (Fig. 3b). The percentage of TGF-β⁺ NK cells in ART-naïve patients was elevated compared with that of HCs (p = 0.022), while in ART-treated patients, the percentage of TGF-β⁺ NK cells still exhibited a tendency to be higher than that of HCs (p = 0.060) (Fig. 3c). There was no difference in the percentage of IL-10⁺ CD56^dim NK cells among the three groups (Fig. 3d), whereas the percentage of TGF-β⁺ CD56^dim NK cells in ART-naïve patients was higher than that in HCs (p = 0.024) (Fig. 3e). Moreover, there were no differences in the percentages of IL-10⁺ CD56^bright and TGF-β⁺ CD56^bright NK cells among the three groups (Fig. 3f, g). In ART-naïve patients, the percentage of IL-10⁺ NK cells was positively correlated with that of TGF-β⁺ NK cells (r = 0.388, p = 0.010) (Fig. 3h) and the percentage of IL-10⁺ CD56^dim NK cells correlated positively with that of TGF-β⁺ CD56^dim NK cells (r = 0.438, p = 0.003) (Fig. 3i).

CD4⁺ T cell counts and viral loads are important markers for HIV disease progression. We analyzed the correlation between the percentage of IL-10⁺ or TGF-β⁺ NK cells and CD4⁺ T cell counts and viral loads and found no significant correlations between them. Furthermore, we analyzed groups subdivided by CD4⁺ T cell count or viral load. The percentage of IL-10⁺ NK cells in the CD4⁺ T < 350 cells/μl group of ART-naïve patients was higher than that of the HC group (p = 0.032), and there was an increasing trend in the CD4⁺ T ≥ 350 cells/μl group of ART-naïve patients compared with HCs (p = 0.075) (Fig. 4a). There was no difference in the percentages of TGF-β⁺ NK cells among groups (Fig. 4b). When grouped according to the level of viral load, the percentage of IL-10⁺ NK cells in ART-naïve patients with viral load > 10⁴ copies/ml was higher than that in the HC group (p = 0.023) (Fig. 4c), and that in the group with viral load < 10⁴ copies/ml, while the percentage of TGF-β⁺ NK cells was higher than that in the HC group (p = 0.027) (Fig. 4d).

The expression of CD107a and IFN-γ secretion can be used to assess the functional anti-viral ability of NK cells. The inhibitory effects of rIL-10 or rTGF-β on NK cell functions were evaluated, and the percentages of IFN-γ⁺ and CD107a⁺ NK cells determined. Representative flow cytometry plots, demonstrating the effects of rIL-10 or rTGF-β on CD107a expression by NK cells are presented in Fig. 5a. CD107a expression in NK cells was inhibited by high concentrations of rIL-10 or rTGF-β in vitro (p = 0.037, p = 0.024), and the rate of inhibition increased gradually with increasing concentrations of these cytokines (Fig. 5b, c). Moreover, the effect of rIL-10 and rTGF-β on CD107a expression in NK cells was synergistic, demonstrating inhibitory effects much greater than those of either cytokine individually (p = 0.006) (Fig. 5d). Representative flow-cytometric plots demonstrating the effects of rIL-10 or rTGF-β on IFN-γ secretion from NK cells are shown in Fig. 6a. IFN-γ secretion by NK cells was also inhibited by rIL-10 or rTGF-β (p = 0.006, p = 0.016), and the rate of inhibition increased gradually with their concentrations (Fig. 6b, c). Again, the effect of both IL-10 and TGF-β was synergistic (p = 0.006) (Fig. 6d).

Granzyme B and perforin are important functional markers for NK cells, which indicate the killing capability of NK cells. The percentages of IL-10⁺ and TGF-β ⁺ NK cells were elevated in HIV infected patients; however, their influence on the release of granzyme B and perforin by NK cells is not known. Our results demonstrate that a low concentration of rIL-10 or rTGF-β did not inhibit the release of granzyme B, whereas high concentrations of either cytokine did inhibit the release of this factor (p = 0.014, p = 0.040) (Fig. 7a, b). A synergistic effect of rIL-10 and rTGF-β on inhibition of the release of granzyme B was also noted (p = 0.037) (Fig. 7c). Similarly, rIL-10 and rTGF-β inhibited perforin release by NK cells (Fig. 8a–c).