Study participants were identified among the attendees of the outpatient clinics of the National AIDS Research Institute who visited the clinic regularly for care and support. LTNPs were identified as individuals with asymptomatic HIV infection for more than 7 years who had stable CD4 counts above 500 cells/μl and who had not taken ART (3). Progressors were identified as ART-naive HIV-infected patients with CD4 counts <500 cells/μl. Thirty-four LTNPs and 58 progressors were enrolled in the study along with 14 HIV-uninfected healthy individuals. Demographic, immunological, and virological characteristics of 92 HIV+ subjects are shown in Table 1.

The study was approved by the institutional ethical committee. Participants were enrolled after obtaining written informed consent.

The ADCC response in the serum/plasma samples of the study participants was determined using NK cell activation assay as described previously (10). The assay uses CD107a and IFN-γ expression by NK cells as a final readout. We preferred this assay, although it does not measure the killing of infected target cells, as this assay was helpful in allowing us to map the linear epitopes recognized by ADCC-mediating antibodies and also captured cytokine secretion (IFN-γ), an another important aspect of antibody-mediated NK cell activation. Also, it has been recently shown that the ADCC activity observed in NK cell activation assay correlates significantly with functional killing assays such as the rapid fluorometric ADCC assay (11) and other assays (12–17). The anti-HIV ADCC responses were assessed against consensus envelope [both HIV-1 B and C], Gag (HIV-1C consensus), Pol (HIV-1 B consensus), and accessory proteins: Nef, Vpu, Rev, and Tat (HIV-1 B consensus) using pools of 15mer peptides overlapping by 11aa (National Institutes of Health AIDS reagent program). The detailed description of the peptides and the procedure for preparation of peptide pools is given in Supplementary Material. Briefly, 150 µl whole blood from healthy donor (as a source of NK cells) and 50 µl of heat-inactivated heparin-anticoagulated plasma or serum [final concentration as1:4] from the study participants was incubated at 37°C with HIV peptide pools (at a final concentration of 1 μg/ml/peptide) for 5 h in the presence of brefeldin A (10 μg/ml) (Sigma) and monensin (0.68 μl/ml) (BD Biosciences). After incubation, CD3−CD56dim NK cells were gated and analyzed for the expression of intracellular IFN-γ and CD107a using fluorescent tagged antibodies [CD3 Per Cp: (clone-SK7), CD56 PE-Cy7: (clone-HCD56), and CD107a APC H7: (clone-H4A3) (all from Biolegend), and IFN-γ APC: (clone-B27) (BD Biosciences)]. The gating strategy is shown in Figures 1A–C. Stimulation by purified anti-CD16 monoclonal antibody (clone-3g8, at 2.5 μg/ml concentration) was used as a positive control (Figure 1D). Plasma samples of all study groups incubated with HIV-negative donor blood in the absence of antigen was considered as a negative control for the respective sample. Mock control containing plasma, HIV-negative donor blood, and DMSO equivalent solution was used to negate the effect of DMSO that was used to dissolve peptides on NK cell activation (Figure 1E). The percentage of activated NK cells as a marker of ADCC activity was represented as the sum of the percentage of any activated NK cells, expressing only CD107a, both CD107a and IFN-γ and only IFN-γ (Figures 1F,G). The percent NK cell activation seen in mock stimulation (donor blood (NK cells) and test plasma incubated along with DMSO equivalent solution) for each sample was subtracted from this sum. The NK cell activation in mock control (plasma+ cells without stimulation) ranged from 0.124 to 4.41% with the mean of 1.48%. The response was considered to be positive if the following two criteria were fulfilled: (1) the total percentage of CD107a and IFN-γ expressing (activated) NK cells was more than three times the percentage of activated NK cells from the negative control of the respective sample and (2) the total percentage of activated NK cells was greater than the mean plus two SD of the percent activated NK cells from 14 HIV-negative donors tested for each of the HIV antigens. The interassay variation was found to be <10% as assessed by determining the % CV for positive control [Anti-CD16 stimulation] was 7.18% and negative control [donor blood (source of NK cells) without any stimulant] was 9.8%.

The samples of ADCC responders were further used to identify linear peptides recognized by ADCC antibodies using the peptide matrix pools in NK cell activation assay. The matrix pools were designed in such a way that each peptide is present in two different pools as described previously (18). The peptide that was common in two pools showing that ADCC response was identified as the peptide recognized by ADCC antibodies and the recognition of the particular peptide was further confirmed by repeating the NK cell activation assay using the single peptide. For mapping the epitopes, 30 pools for HIV-1 subtype C Env peptides and 10 pools for HIV-1 subtype B Tat peptides were used. Details of the pools are given in Figures S1A,B in Supplementary Material.

Statistical analyses were performed using GraphPad Prism, version 5.0 (GraphPad Software, San Diego, CA, USA). Data were analyzed by Mann–Whitney U test to compare the magnitude of ADCC responses between LTNP and progressor groups. Correlation analyses were performed using Spearman correlation between CD4 count, viral load, and magnitude of ADCC responses. Fisher’s exact test and chi square tests were used to analyze the significance of ADCC responses in different categories.

Thirty-four HIV-infected individuals who fulfill criteria for LTNP were enrolled in the study. As a comparator, we selected 58 progressor subjects with CD4 T cell counts of <500/μl. The median age of LTNPs (median 39 years [IQR 35–45]) and progressors (median 35 [IQR 31–39]) was not significantly different. Of the 34 LTNPs, 23 were female and 11 were male, versus 24 and 34 among the 58 progressors, respectively. The mean seropositivity in LTNP cohort was 10.8 years with a range from 7 to 19 years. The CD4 counts of LTNPs at the study visit (median 699 cells/μl [IQR 632–942]) were significantly higher than in the progressors (median 409 cells/μl [IQR 315–456]) (p < 0.001). As expected, the plasma viral load values at the study visits were significantly lower in LTNPs (median 2,174 HIV RNA copies/ml [IQR 422–9,799]) when compared to the values from progressors (median 17,148 [IQR 5,543–87,100]) (p = 0.0002) (Table 1).

A total of 20 of 34 LTNPs (58.8%) and 20 of 58 progressors (34.4%) showed antibody-dependent NK cell activation against at least one of the HIV overlapping peptide pools. The trend of higher magnitude in LTNP was observed in the case of Env-C [LTNP mean 18.2%; range 8.4–35%, progressors mean 13.72%; range 5.6–24.5%], Tat B [LTNP mean 10.8%; range 5.8–16.4%, progressors mean 4.3%], Nef B [LTNP mean 7.6%; range 4.6–15.87%, progressors mean 5.6%; range 3.5–10.6%], and Vpu [LTNP mean 13%; range 5.1–20.8%, progressors mean 11%; range 3.6–21%]. Whereas a trend of higher magnitude in progressors was observed in the case of Env B (LTNP mean 9.36%; range 2.8–21%, progressors mean 13.01%; range 2.7–43%) and Rev B [LTNP mean 6.2%; range 4–9.2%, progressors mean 9%; range 3.7–16.6%]. In the case of Gag C, ADCC responses were seen only in the LTNP cohort and only one from progressor group showed the ADCC response against Pol B peptide pool (Figures 2A,B). The number of LTNPs showing Env-C and Gag-specific ADCC responses was also significantly higher than among progressors (p < 0.007 and p < 0.016, respectively). The number of responders against Tat antigen was also higher in LTNPs, but the difference did not reach significance (p = 0.06, Fisher’s 2 × 2 exact test), whereas the number of responders against HIV-1 subtype B Env peptides, Pol, Nef, Rev, and Vpu antigens was similar in both groups (Figure 2C). Although the magnitude of ADCC response is shown as % total NK cell activation (sum of % NK cells expressing CD107a and/or IFN-γ), overall the distribution of CD107a and IFN-γ expression of activated NK cells by the NK cell-activating antibodies within the plasma of the LTNPs and progressors was found to be skewed toward higher IFN-γ expression in both groups (Table S2 in Supplementary Material).

We then analyzed whether the magnitude of the Env subtype C, Tat, and Gag-specific ADCC responses in the HIV-infected individuals correlated with CD4 T cell count and viral load; standard markers of HIV disease progression. Anti-Env ADCC responses were modestly correlated with CD4 T cell counts (r = 0.235, p = 0.02, Figure 3A). There was a trend toward higher CD4 T cell counts in subjects with higher Tat-specific ADCC responses (Spearman = 0.190, p = 0.07, Figure 3B). The plasma viral load values were negatively correlated with ADCC responses against Env-C specific ADCC (r = −0.367, p = 0.0006, Figure 3C) and Tat (r = −0.263, p = 0.016, Figure 3D) antigens. Also, a significant association between Env-C ADCC responses and lower plasma viral load was observed when only LTNP group was separately analyzed (Spearman r = −0.377; p = 0.04). Besides this, the median plasma viral load values (calculated based on annual viral load values obtained during the follow-up period) among LTNPs were also negatively correlated with ADCC responses against Env-C peptide pool (r = −0.505, p = 0.004). There was no significant correlation between CD4 T cell counts or viral load values with the magnitude of Gag-specific ADCC responses (data not shown). Our definition of LTNPs was immunological (preserved CD4 T cell counts for at least 7 years) rather than virological. Hence, to further study the role of Env-specific ADCC in control of viral load, we divided the LTNP cohort into those with low-level viremia (median VL load <2,000 RNA copies/ml during follow-up) and with higher viral loads (median >2,000 RNA copies/ml). The magnitude of ADCC response against envelope C peptide pool was similar in the viremic controllers and non-controllers within LTNP (p = 0.15). Out of five responders to V3 loop epitope from LTNP, four were viremic controller subgroup as against only one was from non-controller (Figure 3E).

The ability of ADCC responses to recognize diverse HIV strains is likely to improve the effectiveness of ADCC antibodies as a prevention strategy. Hence, to study the breadth of the ADCC antibody specificities, we analyzed whether LTNPs and progressors recognized both subtype C and B Env peptide pools. We found that of the 34 LTNPs, 20.6% (N = 7) had ADCC responses against both HIV-1 C and B Env peptides, whereas only 3.4% (2/58) of the progressors showed response against both B and C Env peptides (p < 0.01 by Fisher’s 2 × 2 exact test, Figure 4A).

Combinations of broad ADCC responses might also contribute to slow HIV progression given our findings on Env- and Tat-specific ADCC responses as mentioned above. To study this, we analyzed both the cohorts that had positive ADCC responses to 0, 1, 2, 3 or >3 of the HIV antigens represented by peptide pools. We found that the ADCC responses in the LTNPs showed significantly greater breadth compared to the progressors. A total of 21% of LTNPs showed responses against more than three antigens, compared to 7% in progressors (chi square test, p < 0.0001, Figure 4B).

ADCC responses targeting specific epitopes may be beneficial in HIV control, similar to that observed with specific CD8+ CTL responses. To identify epitopes recognized by the ADCC antibodies from the Indian LTNPs, we used a matrix approach using subpools of peptides. The peptide pools in the matrix were designed in such a way that a single peptide was common in both the pools. ADCC epitopes identified in the matrix approach were then confirmed by NK cell activation assay using the single peptide present in both the pools to which the response was observed. We studied epitopes from Env-C and Tat antigens, as the responses to these antigens were associated with lower plasma viral load and also were higher in frequencies and magnitudes in the LTNP cohort. Due to the scarcity of the samples from the progressor group, we mapped the potential ADCC epitopes only in LTNPs and only the frequently recognized peptides (when recognized by 3 or more than 3 LTNPs) were tested in progressors.

The consensus Env-C overlapping peptides were 212 in number and the number of consensus Tat B peptides were 23. Out of 212 Env-C peptides, 153 peptides were recognized by ADCC antibodies from 10 LTNPs and 16 out of 23 Tat B peptides were recognized by 3 LTNPs (Table 2). One LTNP showed response to Env and Tat peptides. It was observed that the conserved C1, C2, C4, and C5 region and the variable V3, V4, and V5 and a few CD4 binding sites were recognized by the ADCC antibodies from Indian LTNPs (Table 2). In the gp41 region, both transmembrane and cytoplasmic tail regions were also identified. The Env-C V3 region (aa 288–330: LNESVEIVCTRPNNNTRKSIRIGPGQTGDIIGDIRQAHC) was preferentially identified by Indian LTNPs (5/10 Env-C responders) (Table 2). This region has also been previously identified by the monoclonal antibodies (694/98D, 4117C, 41148D, CH22, CH23) generated from the HIV-1 B-infected ADCC responders (19–22). None of the progressors identified this peptide (p = 0.0002 by Fisher’s exact test) (Table 3), indicating a potential role of ADCC antibodies recognizing this region in virus control. Additionally, Indian LTNPs also showed recognition of novel antigenic stretches from the signal peptide region (aa 1–15), CD4 binding sites (aa 113–117 and aa 361–366), C2 region (aa 250–288), aa C3 (aa 331–360), V4, C4, and V5 region (aa 381–470), and fusion peptide region of gp41 (aa 510–570) (Table 2). Three “Tat” regions (peptide no. 8, 12–13, 20–22) were recognized by three (8.82%) LTNPs. None of the progressors identified these regions (Fisher’s exact test: p < 0.05 for all) (Table 3).