Thirteen [3 men/10 women; age 35 years (30.5–41.5) years] LTNPs [HIV-infected individuals who are asymptomatic and maintained a stable CD4 count of ≥500 cells/mm³ for a period of 7 or more years without antiretroviral therapy (ART)] (21) (viral load: median: 3 log copies/ml; IQR: 2.6–4 log copies/ml) were enrolled from the outpatient clinic of the institute. Based on their plasma ADCC activity observed earlier (7), they were categorized as R-ADCC (n = 10, showed anti-HIV-1 C Env ADCC) and NR-ADCC (n = 3, did not show ADCC against any HIV-1 antigen) LTNPs (7). Three (2 men/1 woman) [age: 32 years (27–34 years)] HIV-seronegative controls (HCs) were also enrolled to assess the generation of non-specific antibodies. Additionally, eight HIV-infected ART-naive individuals (stable CD4 count ≥500 cells/mm³ at least for 3 years) were enrolled as a source of HIV reservoir and also the effector cells in latent reservoir reduction assay.

Forty milliliters of whole blood sample was collected from all participants. The peripheral blood mononuclear cells (PBMCs) were separated by density gradient centrifugation and stored at −196°C until use. For all experiments, the cryopreserved PBMCs were revived, rested overnight at 37°C in 5% CO₂ in RPMI with 10% fetal bovine serum, and then used for various assays.

The study was reviewed and approved by the Institutional Ethics Review Board of ICMR-National AIDS Research Institute (NARI/EC/2017-20). The study participants provided their written informed consent to participate in this study.

We identified the HIV-Env-gp140-specific memory B cells using a biotinylated HIV-gp140 protein probe [HIV-1/clade B′/C (CN54)] (Immune Technology Corp. Lexington Ave., New York). Since the HIV envelope is also known to bind certain integrin and C-type lectin receptors expressed on B cells (22–24), the specificity of the gp140 probe was confirmed using unrelated antigen probes, viz., Influenza A H5N1 (A/chicken/India/NIV33487/06) Hemagglutinin/HA Protein (His Tag) (Sino Biological, Inc., China), Rubeola virus Native Protein (MyBioSource, USA), and tetanus toxoid C-fragment (TTCF) (MyBioSource, USA). These probes were biotinylated using EZ-Link™ Sulfo-NHS-LC-Biotinylation Kit (Thermo Fisher Scientific, Waltham, MA, USA) as per the manufacturer’s protocol and tagged with streptavidin BV510 (BD Biosciences, CA, USA). The biotinylated gp140 probe was conjugated with streptavidin PE-Cy7 (BD Biosciences). The revived PBMCs from an LTNP were incubated with the gp140 protein probe and either of the unrelated antigen probes. After washing and fixating, the cells were acquired on BD FACSAria™ Fusion. The antigen-specific memory B cells were gated from CD3−CD19+CD27+ total memory B cells (Figure 2A). The quadrant was set on CD3−CD19+CD27 total memory B cells to gate respective antigen-specific memory B cells as shown in Figure 2B(1), (2), (3), and (4), respectively, for each unrelated antigen probe. The gating showed that the antigen-specific memory B cells were present in separate and mutually exclusive quadrants (Q1 and Q4), and the cells showing binding to both the probes (Q2 quadrant) were negligible (0.01%–0.02%), confirming the specificity of the HIV-gp140 probe.

The HIV-specific memory B cells were identified using the HIV-gp140 protein probe as per the previously reported protocol (25, 26) with some modifications. In brief, after revival and resting, the PBMCs from the study participants were stained with CD3 BV786, CD19 FITC, and CD27 PE (all from BD Biosciences) and with biotinylated gp140 (5 µg/ml) for 30 min at room temperature and were then stained with streptavidin PE-Cy7 (BD Biosciences) for 30 min at room temperature and acquired on BD FACSAria™ Fusion (BD Biosciences). The total memory B cells were gated as CD3−CD19+CD27+ cells from lymphocytes (Figure 2A) which were further drilled down to identify HIV-gp140-specific memory B cells as CD3−CD19+CD27+gp140+ cells (Figure 2C). These cells were then sorted using BD FACSAria™ Fusion (BD Biosciences) and collected in the FACS tube. The purity of the sorted cells was >95% in all cases.

The sorted gp140-specific memory B cells (2 × 10⁴/ml) from all LTNPs (10 R-ADCC, 3 NR-ADCC LTNPs) and total memory B cells from three HCs were cultured in a 96-well plate for 10 days at 37°C and 5% CO₂ in Iscove’s modified Dulbecco’s medium (Thermo Fisher) and 10% FBS along with human insulin (5 μg/ml) and human transferrin (50 μg/ml) (Sigma-Aldrich) and with different combinations of cytokines (detailed in Supplementary Table 1). On the 4th, 7th, and 10th day of culture, the supernatants were collected and stored at −20°C for further analysis. On the 10th day of culture, the cells were washed and stained for plasma cell markers: anti-human CD20 APC Cy7, CD38 PECF594, and CD138 PerCp Cy5·5 (BD Biosciences) and violet amine-reactive dye live-dead APC (Life Technologies) for 30 min at room temperature. The cells were acquired on BD FACSAria™ Fusion after fixing and analyzed by BD FlowJo (version 10.0) software. The gating strategy for the identification of plasma cells is depicted in Figure 2D.

To confirm the differentiation of memory B cells into plasma cells, we assessed the expression of transcription factors (TFs) of PAX-5, PRDM-1, BCL-6, and XBP-1 genes that regulate the differentiation of memory B cells using quantitative real-time PCR (RT-qPCR). The total RNA was extracted from cells at day 0 and day 10 culture using TRIzol^® Reagent method (Thermo Fisher) and quantitated using NanoDrop (Thermo Fisher). The cDNA was prepared using TaqMan^® Reverse Transcription Reagents (Thermo Fisher). The RT-qPCR was performed using SYBR^® Select Master Mix (Thermo Fisher) on ABI Thermal cycler HT 9700 (Applied Biosystems) according to the manufacturer’s protocol. The primer sequences are provided in Supplementary Table 2. The relative gene expression was normalized to β-Actin and calculated using the 2^−ΔΔCt method as described previously (27).

The 4- and 10-day culture supernatants were tested for the presence of IgG using Human IgG total Ready-SET-Go ELISA kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. Briefly, the ELISA plate was coated overnight with purified anti-human IgG monoclonal antibody at 4°C. The next day, the wells were blocked with blocking buffer (1× PBS with 1% Tween 20 and 10% BSA) for 2 h at room temperature. All further incubations were carried out at room temperature. After washing, the culture supernatants (1:1,000 diluted) and 2-fold serially diluted IgG standards were added to the respective wells, and the plate was incubated for 2 h on a microplate shaker at 400 rpm. After washing, the horseradish peroxidase (HRP)-conjugated anti-human IgG antibody was added and the plate was incubated for 1 h, washed, and developed using 3,3′,5,5′-tetramethylbenzidine substrate (TMB) solution for 15 min. The reaction was then quenched using the stop solution (2 N H₂SO₄). The optical density (OD) was measured at 450 nm with a reference wavelength of 650 nm. The OD values of serially diluted standards were plotted on the Y-axis and their respective concentrations were plotted on the X-axis to create a graph to measure the IgG concentration from the culture supernatants based on their OD values.

The IgG antibodies from the day 10 culture supernatants and the antibodies from the plasma of LTNPs showing ADCC activity were purified using NAb Protein G Spin Kit (Thermo Fisher) as per the manufacturer’s protocol and desalted using the Spin desalting columns (Thermo Scientific). The concentration was determined using NanoDrop (ND-1000). The concentration of antibodies purified from 10-day culture supernatants was found to be 40–80 μg/ml. The desalted purified IgG antibodies and their respective unbound fractions were stored at −20°C for further use.

The binding of purified antibodies (from 10 R-ADCC, 3 NR-ADCC LTNPs, and 3 HCs) to HIV-gp140 protein was assessed by gp140 ELISA as described previously (26). In brief, the high binding ELISA plate was coated overnight with 100 ng/well of purified gp140 protein at 4°C. Following washing with wash buffer (1× PBS with 0.05% Tween 20), the wells were blocked with the blocking buffer (1× PBS with 0.1% BSA and 1 mM EDTA) for 2 h. After washing, 50 µl of 4-fold serially diluted standards, i.e., anti-HIV immune globulin (HIVIG) (NIH AIDS Reagent Program) and purified antibodies (1 µg/ml), were added to the respective wells followed by incubation for 2 h at 37°C. The purified antibodies from the plasma of HIV-positive patients were used as a positive control (PC). The purified antibodies from the plasma of HIV-seronegative individuals were used as a negative control (NC) for the assay. The binding was then detected with the combination of HRP–anti-human IgG conjugate and TMB substrate. The OD values were measured at 450 nm with a reference wavelength of 650 nm. The cutoff value was calculated as (mean OD of PC + mean OD of NC)/6. The mean OD value of NC was 0.0685 and that of PC was 3.1965. The samples with OD values above the cutoff value of 0.544 were considered positive for gp140-binding antibodies. The OD values of diluted standards were plotted on the Y-axis and their respective concentrations were plotted on the X-axis to generate the reference graph. The concentrations of gp140-specific antibodies were determined by plotting the OD values on the reference graph.

The generated purified antibodies showing gp140 specificity and their respective unbound fractions were tested for the ADCC against the Env-coated targets using rapid fluorometric antibody-dependent cellular cytotoxicity (RF-ADCC) assay as reported previously (28, 29). Briefly, the CEM.NKr-CCR5 target cells were coated with HIV-1 96ZM651 gp120 (HIV-1 C Env) and HIV-1 BaL gp120 (HIV-1 B Env) recombinant proteins (NIH AIDS Reagent Program) for 1 h at room temperature. Both coated and uncoated cells were then stained with the membrane dye PKH26 (Red Fluorescent Cell Linker, Sigma Aldrich, St. Louis, MO, USA) and carboxyfluorescein succinimidyl ester (CFSE) dye (Molecular Probes, UK) and incubated with purified antibody (2 µg/ml) or the unbound fraction at 37°C in 5% CO₂ for 30 min. Subsequently, fresh PBMCs from a healthy donor were added as a source of effector cells in 1:10 target to effector ratio and incubated for 4 h at 37°C in 5% CO₂ and then surface stained with CD3 PerCP and CD14 APCCy7 (both from BD Biosciences). The cells were acquired on BD FACSAria™ Fusion and analyzed using FlowJo software (version 10). The ADCC response was measured in terms of the uptake of the PKH26 dye from lysed target cells by CD3−CD14+ monocytes. The control of only effector cells + gp120-coated target was kept to check the background response. The gating strategy used for this assay is shown in Figure 3A.

The potential of in-vitro-generated ADCC antibodies to facilitate the killing of reactivated HIV-1 latent reservoirs was assessed in a flow-based latency reduction assay as used in our previous study (16). The reduction in the frequencies of intracellular p24-expressing resting memory CD4+ T cells was used as an indicator of the killing of reactivated HIV latent reservoirs by the NK cells in the presence of the generated purified antibodies. Briefly, CD4+CD45RO+ resting memory T cells (target) and CD4−CD8− effector cells (NK cells) were FACS sorted from the PBMCs of ART-naive HIV-infected individuals using CD4 PECF594, CD8 PE, and CD45 RO PEcy5.5 (all from BD Biosciences) as described previously (16). The sorted effector and target cells (1:1) were then incubated together in the presence of consensus HIV-1C envelope peptide pool (NIH AIDS Reagent Program) (5 µg/ml) for 3 h at 37°C in 5% CO₂ and were then washed to remove peptides and incubated for 5 h with the antibodies generated from 5 R-ADCC and 2 NR-ADCC LTNPs in the presence of CD107a APCCy7 (BD Biosciences), brefeldin A (10 μg/ml) (Sigma), and monensin (0.68 μl/ml) (BD Biosciences) at 37°C in 5% CO₂. Then, cells were surface stained with fixable viability stain 510, CD3 BV786, CD4 PECF594, CD45RO PE-Cy5.5, and CD56 PECy7 (all from BD Biosciences) for 30 min at room temperature. After permeabilization, the cells were stained for intracellular p24 (KC57-FITC) (Beckman Coulter, USA) and IFNγ APC (BD Biosciences) and incubated for 45 min at room temperature. After washing and fixing, the cells were acquired on BD FACSAria™ Fusion (BD Biosciences, USA) and analyzed using FlowJo software (Version 10). The gating strategy for identifying frequencies of p24-expressing cells in different experimental conditions is shown in Figure 4A. The gate for % p24-expressing cells was set using the stained uninfected cells. The percentages of p24 cells in unstimulated control (background response) were subtracted from the frequencies of p24+ cells in the respective Env-stimulated wells and used to calculate the percent p24 reduction using the formula: % p24 reduction = % p24+CD4+CD45RO+ cells in [(Env-stimulated well) − (Env+IgG well)]/(Env-stimulated well) × 100. Simultaneously, the activation (secretion of cytokine IFNγ and/or expression of degranulation marker CD107a) of CD3−CD56^dim NK cells was measured before and after the addition of purified antibodies as described previously (16) (Figure 4B).

Furthermore, we compared the potency of in-vitro-generated ADCC antibodies (from five R-ADCC LTNPs) to mediate the killing of reactivated latent reservoirs with the antibodies purified from the plasma samples of the same LTNPs.

The GraphPad Prism software (version 7.0) was used for statistical analyses. The Wilcoxon matched-pairs t-test was used to analyze the paired data of relative gene expression, ADCC activity, percent p24 expression, and NK cell activation. The comparison of the potency of generated antibodies with the plasma antibodies was performed using paired t-test. Throughout the text, the data are reported in the median (IQR) format.

All LTNPs (R-ADCC and NR-ADCC) (median: 0.79%, IQR: 0.54%–1.225%), but not HCs, showed the presence of gp140-specific memory B cells (Figure 2E). We used a 10-day culture system representing the in-vivo antibody generation process to generate antibody-secreting plasma cells. On day 10, all cultures (10 R-ADCC and 3 NR-ADCC LTNPs and 3 healthy individuals) showed differentiation of memory B cells into CD20−CD38+ plasmablasts and CD20−CD38+CD138+ antibody-secreting plasma cells. The median frequency of plasmablasts was 80.2% (IQR: 75.5%–90%), and of these plasmablasts, 72.0% (IQR: 68.7%–82.2%) were CD20−CD38+CD138+ antibody-secreting plasma cells.

The transformation of memory B cells into plasma cells was also confirmed by assessing the upregulation of transcription factors required for plasma cell development. For that, the cells at day 0 and day 10 of cultures of eight samples (six R-ADCC and one NR-ADCC LTNPs and one HCs) were analyzed for the expression of four major transcription factors—BCL-6, PAX-5, XBP-1, and PRDM-1—that regulate memory B cell to plasma cell differentiation (30–32).

The expression of PRDM-1 (p = 0.0078) and XBP-1 (p = 0.0078) genes (required for the development and survival of plasma cells) was significantly increased in day 10 cells than day 0 cells. On the other hand, the expression of genes that maintain B-cell phenotype, i.e., BCL-6 and PAX-5, was found to be decreased in day 10 cells than day 0 cells (p = 0.0142 and p = 0.0078, respectively) (Figure 2F), suggesting that memory B cells were successfully differentiated into plasma cells in the 10-day culture system.

The culture supernatants of the 4th and 10th day of culture were tested for the presence of IgG. The 4-day culture supernatants of all the samples were negative for IgG (median: 0.00, IQR: 0.00–0.005), whereas the 10-day culture supernatants of all the samples showed the presence of IgG (OD: median: 0.253, IQR: 0.205–0.274). The median concentration of IgGs in the 10-day culture supernatants was 178.7 µg/ml (IQR: 133.9–211.8 µg/ml) (Figure 2G).

Of the 13 (10 R-ADCC and 3 NR-ADCC LTNPs) purified antibodies tested, 11 [(9 from R-ADCC (OD: range: median: 2.246, IQR: 1.684–3.014) and 2 from NR-ADCC (OD: 2.971 and 1.72)] showed specificity to gp140 protein. The purified antibodies generated from three HCs were used as a non-specific control. The concentration of gp140+ antibodies was 122.5 µg/ml (IQR: 88.52–169.4 µg/ml) with no difference in the concentration of gp140-specific antibodies from R-ADCC (median: 122.5 µg/ml, IQR: 86.16–172.2 µg/ml) and NR-ADCC (median: 129 µg/mL, IQR: 88.52–169.4) LTNPs (Figure 2H).

The antibodies showing specificity to gp140 (nine R-ADCC and two NR-ADCC LTNPs) along with the antibodies from three HCs as background control were tested for their ability to mediate HIV-Env C-specific ADCC by the RF-ADCC assay. No background ADCC activity was observed. The ADCC activity shown by the nine antibodies generated from R-ADCC LTNPs (median: 37.6%, IQR: 32.95%–51%, range: 27.9–62.1) was higher as compared to antibodies generated from two NR-ADCC LTNPs (median: 8.85%, IQR, range: 8%–9.7%) (Figure 3B, left panel) (Table 1) indicating that gp140-specific memory B cells from LTNPs with plasma ADCC could generate potent gp140-specific ADCC antibodies. Also, the ADCC was observed only with the antibody fraction in all samples (median: 37.6%, IQR: 32.95–51, range: 27.9–62.1) as compared with their respective unbound fraction (median: 3.0%, IQR: 1.05%–7.9%, range: 0.2–8.6) (p < 0.0001) (Figure 3B, right panel) indicating that cytotoxic activity was linked to the antibodies only.

The cross-clade activity would be an important characteristic of the generated antibodies in their probable use as immune therapy. The antibodies generated from R-ADCC LTNPs showed similar activity against both Env C (median: 37.6%, IQR: 32.95%–51%, range: 27.9–62.1) and Env B (median: 36.5%, IQR: 32.65%–50.75%, range: 26.9–67.8) (Figure 3C; Table 1) indicating the cross-clade nature of these antibodies.

A flow-based latency reduction assay was used to determine the reduction in the percentages of Env C-reactivated p24-expressing CD4+ memory cells in the presence of generated antibodies and effector cells. Stimulation with HIV-1 Env C significantly increased the frequencies of p24+CD4+CD45RO+ memory cells (median: 4.83%, IQR: 4.42%–5.21%) as compared to the unstimulated cells (median: 0.64%, IQR: 0.59%–0.77%) (p = 0.0211) indicating the reactivation of resting memory CD4+ T cells. These frequencies were significantly reduced when antibodies generated from R-ADCC LTNPs (n = 5) were added (median: 1.6%, IQR: 1.54%–1.815%) (p = 0.0109), whereas marginal reduction in p24-expressing Env-stimulated cells was observed when antibodies generated from NR-ADCC LTNPs (n = 2) (median: 3.25%, IQR: 3.13%–3.37%) were added (Figure 5A). The percent p24 reduction of 62.5% (IQR: 58.71%–64.92%, range: 56.37–64.99) observed in the case of antibodies generated from R-ADCC LTNPs was 4.5-fold higher as compared to the reduction observed in the case of antibodies generated from NR-ADCC LTNPs (median: 13.84%, IQR, range: 9.863%–17.81%) (Figure 5B; Table 1).

Simultaneously, the percent NK cell activation (measured as IFNγ secretion and CD107 expression) was also increased significantly after the addition of antibodies generated from R-ADCC LTNPs (median: 28.27%, IQR: 25.11%–43.15%) as compared to the respective unstimulated control (median: 2.04%, IQR: 2.04%–2.94%) (p = 0.0041). However, the frequencies of activated NK cells after the addition of antibodies generated from two NR-ADCC LTNPs were increased marginally than their respective unstimulated controls (5.8% versus 12.94% and 5.8% versus 18.95%) (Figure 5C). Taken together, these observations indicate that HIV-specific antibodies generated from LTNPs with plasma ADCC activity showed higher potency to mediate ADCC and to effectively lyse the reactivated latent reservoirs as compared with the antibodies generated from the LTNPs without plasma ADCC activity.

The characteristics of the individual antibodies are compiled in Table 1. It was observed that the HIV-specific antibodies generated from R-ADCC LTNPs (R1 to R9) showed higher potency to mediate ADCC activity and also showed cross-clade ADCC activity as compared with the antibodies generated from NR-ADCC LTNPs (NR1 and NR2). Five of the nine antibodies generated from R-ADCC LTNPs (R1, R3, R4, R6, and R7) and antibodies generated from both NR-ADCC LTNPs (NR1 and NR2) were tested for their ability to lyse reactivated latent reservoirs. Four antibodies (R2, R5, R8, and R9) could not be tested in latency reduction assay due to the limitation of sample volume. All the five antibodies (R1, R3, R4, R6, and R7) showed higher potency to lyse Env-reactivated resting memory CD4+ T cells (% p24 reduction: median: 62.5%, IQR: 58.71%–64.92%, range: 56.37–64.99) as compared with the antibodies generated from NR-ADCC LTNPs (% p24 reduction: median: 13.84%, IQR/range: 9.863%–17.81%). Also, the simultaneous NK cell activation observed was higher in R-ADCC LTNPs (median: 26.23, IQR: 22.62–40.66, range: 21.56–50.27) than NR-ADCC LTNPs (median: 10.15, IQR, range: 7.14–13.15) (Table 1).

The in-vitro-generated antibodies should have good potency. The potency to mediate lysing of reactivated latent reservoirs of the generated ADCC antibodies and the antibodies purified from the plasma of the corresponding LTNPs was compared. The potency of the generated antibodies to lyse the reactivated reservoir (median: 62.5%, IQR: 58.71%–64.92%) was two-fold higher than the antibodies from the plasma of the respective ADCC responder LTNPs (median: 31.93, IQR: 21.33–40.37) (p = 0.0016) (Figure 6).