L-selectin/CD62L is a marker for central memory T cells and promotes the homing of T cells to lymph nodes (Von Andrian et al., 1993; Sallusto et al., 2004). We previously showed that L-selectin shedding facilitated HIV-1 release (Kononchik et al., 2018). However, the proteases responsible for the viral-induced CD62L shedding remain to be defined. CD62L shedding is mainly studied in the context of neutrophil activation and has been shown to involve ADAM10 and ADAM17 metalloproteinases (Le Gall et al., 2009; Wang et al., 2010; Saftig and Reiss, 2011). Like neutrophils, activated CD4 T cells shed CD62L and release TNFα that can be detected by ELISA (Figures 1A,B). The release of both TNFα and soluble CD62L were inhibited by batimastat (BB-94), a matrix metalloproteinase inhibitor that inhibits both ADAM10 and 17 (Saftig and Reiss, 2011), suggesting the involvement of ADAM enzymes in the shedding of CD62L on CD4 lymphocytes. To further characterize the enzymes responsible for T cell shedding of CD62L and thus facilitate the design of protease specific inhibitors as potential antiviral reagents, we measured shedding of CD62L by ELISA in the presence of various protease inhibitors. As T cells express other proteases (Kido et al., 1990), we systematically evaluated the inhibition of CD62L shedding in both anti-CD3 stimulated and non-stimulated cell cultures by serine-, cysteine-, aspartyl-proteases as well as metalloproteinases inhibitors (Table 1). Stimulated or unstimulated peripheral blood mononuclear cells (PBMC) were treated with inhibitors specific against various proteases for 6–24 h. None of the serine-, cysteine-, or aspartyl-protease inhibitors affected the release of soluble CD62L from either stimulated or naïve PBMC (Figure 1C; Supplementary Figure 1A; Table 1). We then evaluated ~20 additional known inhibitors of Zn-dependent metalloproteinases for their ability to inhibit CD62L shedding from stimulated PBMC. Most of the metalloproteinase inhibitors did not inhibit CD62L shedding, except TAPI-2, GM6001, GI254023X, BB-94 and prinomastat (Figure 1D; Table 1; Le Gall et al., 2009; Moller-Hackbarth et al., 2013; Woods and Padmanabhan, 2013). These compounds target Zn-dependent metalloproteinases and are known to inhibit ADAM10 and 17 (Parsons et al., 1997; Hooper and Turner, 2002; Woods and Padmanabhan, 2013). Further dose titration of these compounds showed that the reductions in CD62L shedding from both stimulated and unstimulated T cells were dependent on the inhibitor doses with minimum cytotoxicity between 20–100 μM compound concentrations (Figures 1E,F; Supplementary Figure 1B). Interestingly, 3,4-dichloroisocouman (or 3,4-dichloro for short), a serine protease inhibitor enhanced CD62L shedding and this shedding enhancement was blocked by BB-94 (Figure 1E).

To address the potency and specificity of CD62L shedding inhibitors against ADAM10 and 17, we carried out an enzymatic cleavage assay using recombinant ADAM10 and 17 to cleave a fluorogenic TNF-α substrate in the presence of 2 μM individual compounds (Figures 2A–C; Supplementary Figure 2). While the broad specificity metalloprotease inhibitors BB-94 and prinomastat inhibited both ADAM10 and 17, an MMP-1 specific inhibitor, Dichloromethylenediphosphonic acid disodium salt (DMDP) inhibited neither ADAM10 nor 17 enzymatic activities (Supplementary Figures 2A,D), consistent with DMDP’s lack of CD62L shedding inhibition (Figures 2B,C; Table 1). As expected, GI254023X, a ADAM10 specific compound, preferentially inhibited the enzymatic activity of ADAM10 compared to ADAM17 (Figures 2B–E; Table 2). In addition, we also synthesized two 3-Cl-phenylpyrrolidine tartrate diamide analogs, 14 m and 15, that were developed specifically against ADAM17 to further investigate the involvement of ADAM17 in HIV infection (Li et al., 2010). Compounds 14 m and 15 differ only by one methyl-group (Supplementary Figure 3A; Li et al., 2010). The chemical synthesis resulted in 2 racemic isomers for compound 14 m (14ma and 14mb) and 4 racemic isomers for compound 15 (15a, b, c, d). Between the two compound 14 m isomers, 14mb exhibited better inhibition to both ADAM10 and ADAM17 enzymatic cleavage of the fluorogenic peptide (Supplementary Figures 2B,E; Figures 2B,C). In contrast, the best of compound 15 isomers, 15b, only marginally inhibited the enzymatic cleavage (Supplementary Figures 2C,F). Further titration inhibition resulted in inhibition constants (Ki) for BB-94 of 18 nM against both ADAM17 and 10, for prinomastat of 48 nM and 2 nM against ADAM17 and 10, respectively, and 0.5–1.1 μM for 14mb against ADAM17 and 10, respectively (Table 2; Figures 2D,E).

We then examined ADAM10 and 17 cleavage of a fluorogenic CD62L peptide substrate. While ADAM17 cleaved TNF-α peptide better than ADAM10 (Figure 2A), the enzymes showed reverse activities against CD62L (Figure 2F), suggesting ADAM10 may be more effective in shedding of CD62L than ADAM17. Similar to the cleavage of TNF-α peptide, BB-94, GI254023X and prinomastat but not DMDP completely suppressed ADAM10 cleavage of CD62L at 2 μM compound concentration (Figure 2G). In comparison, the tartrate-based compounds (14ma, mb, 15a,b,c,d) were significantly less potent in inhibiting ADAM10 (Figure 2H), with the best inhibitory compound, 14mb, displaying a Ki of 1.6 μM against ADAM10 cleavage of CD62L (Figure 2I; Supplementary Figures 3B,C).

To investigate the contribution of individual ADAM to HIV-1 infection, we infected CD8-depleted PBMC with a replication competent R5 tropic HIV-1_BAL in the presence of either ADAM10 or 17 specific or broad specificity inhibitors and analyzed the intracellular p24 levels on day 7 of post infections (Figure 3). Both BB-94 and prinomastat, inhibited HIV-1_BAL infections without significant impact to cell viabilities (Figures 3A,B; Supplementary Figure 4A). As GI254023X distinguished ADAM10 and 17 enzymatic activities (Figures 2D,E), we investigated the effect of their specific inhibition to HIV-1 infection using either GI254023X or TAPI-0, a known ADAM17 inhibitor. In comparison, individual ADAM10 and 17 inhibitor, GI254023X and TAPI-0, respectively, fail to significantly suppress the viral infections, suggesting the involvement of both ADAM10 and 17 in HIV-1 infections. Importantly, DMDP, an MMP-1 inhibitor that did not inhibit ADAM10 and 17 enzymatic activities nor CD62L shedding (Figures 1E, 2B,C), failed to inhibit HIV-1_BAL infections (Figures 3A,B). To address the mechanistic connection between ADAM-mediated CD62L shedding and HIV-1 infection, we examined the loss of CD62L expression in infected (p24+) T cells and its dependence on ADAM inhibitions. Upon infection, HIV-1 down regulated cell surface expression of CD62L, as evident from the preferential accumulation of CD4 and CD62L double negative T cells in the p24⁺ compared to the p24⁻ populations or uninfected samples (Figure 3B; Supplementary Figure 4B; Kononchik et al., 2018). The presence of BB-94 or prinomastat reduced the accumulation of infected CD4⁻/CD62L⁻ T cells in the R5 HIV-1 infections, suggesting the loss of CD62L expression in infected T cells is ADAM dependent. Similar to their lack of inhibitions to the p24 levels, the individual ADAM10 or 17 inhibitors, GI254023X or TAPI-0, respectively, did not inhibit the accumulation of CD4⁻/CD62L⁻ T cells compared to BB-94 and prinomastat (Figure 3C), consistent with the involvement of both ADAM10 and 17 in down-regulation of CD62L. Importantly, the MMP-1 inhibitor, DMDP, failed to prevent the loss of CD62L on HIV-1_BAL infected cells (Figures 3B,C). Indeed, the effect of these inhibitors to HIV-1_BAL infection correlated with their ability to inhibit CD62L shedding (Figure 3D). Interestingly, when cells were treated with BB-94 on day 3 and 5 after the initiation of the infection, the compound showed a progressive reduction in its inhibition (Supplementary Figure 4C). This is likely due to the presence of multi-round replications of HIV-1_BAL during the experiment (Murray et al., 2011). In addition, a serine protease inhibitor, 3,4-dichloroisocouman, significantly increased soluble CD62L concentration in treated media compared to the control (Figure 1E; Supplementary Figure 1). The compound, however, did not enhance HIV-1 infection (Supplementary Figure 4D), suggesting either the viral-induced CD62L shedding is sufficient for its release or 3,4-dichloroisocouman did not affect CD62L shedding but merely reduced the degradation of soluble CD62L in treated cell culture supernatant. To address if X4 tropic HIV-1 infection also depended on CD62L shedding, we infected CD8-depleted PBMC with a replication competent X4 tropic HIV-1_LAI or the R5 tropic HIV-1_BAL in the presence of BB-94 or prinomastat. Both compounds inhibited the X4 tropic HIV-1_LAI infections similar to their inhibitions to the R5 tropic HIV-1_BAL (Figure 3E), suggesting ADAM10 and 17 are also involved in X4 tropic HIV-1 infections.

Although 3-Cl-phenylpyrrolidine tartrate diamide analogs were developed as ADAM17 inhibitors, two of the isoforms,14mb and 15b inhibited ADAM10 and 17 enzymatic activities at 2–10 μM concentrations (Figure 2; Table 2). They are 50–100 times less potent compared to 20–50 nM concentrations of BB-94 and prinomastat. Despite with dual ADAM inhibitory specificities, 14mb only partially inhibited HIV-1_BAL infection of PBMC and the loss of CD62L expression from some donors (Figures 4A,B; Supplementary Figure 4E). When 14mb and 15b failed to inhibit HIV-1 infection of PBMC, they also failed to suppress CD62L shedding (Figure 4C), suggesting both potency and broad specificity are needed to suppress HIV-1 infection. Consistently, those tartrate analogs failed to inhibit ADAM10 digestion of CD62L, including 14ma, 15a,b,c, also failed to suppress HIV-1_BAL infection (Figures 2H, 4A). Together, our data showed that both R5 and X4 HIV-1 infections as well as the loss of CD62L expression on infected T cells are inhibited by broad spectrum metalloproteinase inhibitors, suggesting HIV-1 infections depend on ADAM-mediated shedding of CD62L.

Currently, antiretroviral therapy includes several classes of compounds targeted at HIV entry, reverse transcriptase, integrase, and the viral protease (Lu et al., 2018). ADAMs are Zn²⁺-dependent metalloproteinases distinct from HIV protease, a retroviral aspartyl protease. Metalloproteinase inhibitors that target CD62L shedding constitute a potential new class of antiviral compounds. It is not clear, however, if the ART compounds affect CD62L shedding. To assess the potential overlap between ART compounds and CD62L shedding inhibitors, we evaluated the ability of HIV protease inhibitor (nelfinavir), reverse transcriptase inhibitor (nevirapine, AZT), and integrase strand transfer inhibitor (raltegravir) to inhibit L-selectin shedding from activated PBMCs (Vera et al., 2015). None of these ART compounds significantly inhibited CD62L shedding at their effective dose concentrations compared to BB-94, as measured by ELISA (Figure 5A; Supplementary Figures 5A–C). Conversely, to address if CD62L shedding inhibitors suppressed HIV-1 protease activity, we carried out recombinant HIV protease cleavage of its fluorogenic substrate peptide in the presence of various MMP or viral protease inhibitors (Vella, 1994). While the known viral protease inhibitors, saquinavir and nelfinavir, both suppressed the substrate cleavage by HIV protease, none of the MMP inhibitors, including BB-94, prinomastat, GM6001 and DMDP, inhibited the activity of HIV protease (Figure 5B; Supplementary Figure 5D), suggesting that inhibition of HIV infection by MMP inhibitors is independent of HIV protease. Thus, CD62L shedding inhibitors represent a new class of anti-HIV compounds distinct from the existing HIV ART regimen.

To characterize the effect of ADAM inhibition on HIV infection, we performed transcriptome analysis by high throughput RNAseq using Illumina platform (Novogene Co. Ltd.) on HIV-1_BAL infected with and without BB-94 treatment as well as uninfected CD4 T cells on day 6 post infection (Table 3). Pathway analysis on differentially expressed genes performed using iDEP based on top 400 up- and 400 down-regulated genes showed that HIV-1 infection (samples B1 and B2) significantly up-regulated genes in type I interferon pathway and down-regulated those involved in cell cycle pathways compared to the uninfected (U1 and U2) samples (Sx et al., 2018; Figures 6A,B). This upregulation in type I interferon genes is not observed in the presence of BB-94. However, BB-94 alone does not suppress the transcription of type I interferon pathway genes (Supplementary Figure 6A), suggesting the lack of type I interferon gene upregulation in the BB-94 treated infection is primarily due to the reduction of viral infections in the presence of BB-94. In contrast to genes in type I interferon pathway, HIV-1 infections (B1 and B2) significantly downregulated the cell cycle pathway compared to the uninfected samples (U1 and U2) (Figure 6B). Interestingly, BB-94 alone decreased slightly the expressions of cell cycle genes without HIV-1 infection (Supplementary Figure 6B). Thus, BB-94 does not counter the viral infection in neither interferon nor cell cycle genes, suggesting the antiviral effect of BB-94 is not through regulating the expressions of the two major signaling pathways but rather through its inhibition to the enzymatic activity of metalloproteinases.

Human CD4 T cells express multiple members of ADAM family genes in addition to ADAM10 and 17 (Figure 6D). The higher expression of ADAM10 than 17 on CD4 T cells suggests ADAM10 contributes more to HIV infection than ADAM17. The expressions of ADAM8, 15 and 19 are also significant on CD4 T cells. The presence of all five ADAMs was confirmed by western blot analysis using antibodies specific to the extracellular domain of individual ADAMs (Figure 6E). Interestingly, while transcriptomic analysis showed no significant differences in ADAM mRNA levels among infected with or without BB-94 and uninfected samples on day 6 of the infection (Figure 6C), western blot analysis showed a marginal increase of ADAM10 and 17 in day 3 infected samples (Figure 6F). To further address if HIV-1 infection resulted in a transient increase in ADAM expressions, we used RT-PCR to quantify ADAM9, 10 and 17 mRNA in infected and uninfected cells from both day 3 and 6 post infection. Indeed, HIV-1_BAL infection elevated all three ADAM mRNA levels compared to the uninfected samples only on day 3 but not day 6 of post infections (Figures 6G,H). Their increase was not observed in the presence of BB-94 on both day 3 and 6 of infected samples. Thus, HIV-1 infection induced transient expressions of ADAM10 and 17 in the infected cells.

While CD62L shedding inhibitors suppressed HIV-1 viral infection and deficient budding virions were observed in the presence of BB-94 (Kononchik et al., 2018), it is not clear of the molecular mechanism linking CD62L to HIV-1 release. Recent evidence showed that the viral infection-induced CD62L shedding requires caspase activation and inhibition of caspases impeded the viral release (Segura et al., 2023). CD62L is a C-type lectin receptor capable of binding to HIV-1 envelope glycans (Kononchik et al., 2018). While this glycan binding enhanced viral attachment and entry, it may tether budding virions from release when CD62L shedding is prevented. To visualize CD62L-mediated virion tethering, we labeled X4 tropic HIV-1_LAI infected cells in the presence and absence of BB-94 with immunogold particle-labeled anti-CD62L on day 6 of the infection and examined the budding virions using transmission electron microscope (TEM). The CD62L labeling produced fewer than 50 gold-particles/cell to minimize non-specific labeling in the infected and uninfected cells (Figure 7; Supplementary Figure 7). While the majority of immunogold particles labeled cell surface CD62L that were not associated with HIV-1 virions (Supplementary Figure 7C, indicated by black arrows), many immunogold particles (25% of the total gold particles) were observed in contact with HIV-1 virions (Figure 7; Supplementary Figures 7D,E), illustrating the presence of virion-associated CD62L. We further separated the virion-associated immunogold particles into two categories, the ones tethered to budding virions on cell surface (non-cleaved CD62L, indicated by red arrows) and the ones bound to detached virions from cells (cleaved CD62L, indicated by blue arrows) (Supplementary Figures 7D,E). The two categories represent cell surface and shed CD62L. A survey of ~100 TEM images each from the DMSO and BB-94 treated HIV-1 infections showed the fraction of the virion-associated immunogold particles varied between the budding and released virions in DMSO versus BB-94 treated infections. In the absence of BB-94, the fraction of budding virion associated gold-particles (45%, red arrows) is slightly less than that bound to the released virions (55%, blue arrows) (Figure 7A). In the presence of BB-94, however, the fraction of CD62L-tethered virions increased to 79% while those on released virions decreased to 21%. Namely, the ratio of CD62L-tethered budding versus released virions increased from 1:1 to 4:1 in favor of cell surface tethering in the presence of BB-94. Many virions appeared in small aggregates tethered to cell membranes (Figure 7B). The increase in CD62L-tethered budding virions in the presence of BB-94 illustrates the hindrance of cell surface CD62L to the release of budding virions and supports importance of CD62L shedding in HIV-1 viral release.

To investigate if CD62L shedding inhibitors affected viral release in clinical specimens, we stimulated CD4 T cells from six HIV-1 infected individuals with anti-CD3 antibody in the presence of BB-94, 14 MB, GM6001 or prinomastat and measured the level of virion-associated HIV RNA released on day 2 and 4 post stimulation. The results showed that BB-94 inhibited the viral release from all individuals’ CD4 T cells (Figure 8), resulting in 2–50 fold reductions in released virion-associated RNA (Supplementary Figure 8). Prinomastat and GM6001 inhibited viral release in a portion of the samples, whereas 14mb fail to inhibit the viral release. It is worth noting that BB-94 was more effective in inhibiting the viral release in individuals with lower viral loads (Figure 8). Together, the results showed that HIV-1 viral release can be inhibited by the most potent metalloprotease inhibitor, suggesting an optimum therapy to inhibit HIV-1 viral release is in conjunction with the existing ART therapy.

Compounds used to inhibit CD62L shedding were purchased from Sigma-Aldrich, Santa Cruz Biotechnology and R&D systems, Inc. HIV protease inhibitors and reverse transcriptase inhibitors were purchased from Sigma-Aldrich, St. Louis, MO. MMP inhibitors were purchased from Sigma-Aldrich, AdooQ BioScience, Irvine, CA or MedKoo Biosciences, Inc., Morrisville, NC. All compounds were used at manufacture recommended effective concentrations or otherwise indicated. Western blot human anti-ADAM antibodies were purchased from R&D systems (R&D systems, Inc., Minneapolis, MN) as follows: anti-ADAM8 (MAB1031 and MAB10311), anti-ADAM10 (MAB1427), anti-ADAM15 (MAB935), anti-ADAM 17 (MAB9301), and anti-ADAM19 (AF5050). Anti-CD3 antibody okt3 was kindly provided by Dr. Gary Gilliland of Janssen Pharmaceuticals, Inc. Human CD62L and TNF-α DuoSet ELISA kits were purchased from R&D systems, Inc.

Compounds 14 m and 15 were synthesized as previously described (Li et al., 2010). Both compounds exist as 3-Cl-phenylpyrrolidine and tartrate racemic mixture and were further separated by high performance liquid chromatography (HPLC, Waters Corportion, Milford, MA) prior to linking them to pyrazole. As a result, a total of six conformation variants were obtained, 14ma, 14mb, 15a, 15b, 15c, and 15d. The compounds were not soluble in dimethyl sulfoxide (DMSO) but were dissolved in a 1:1:1 mixture of ethanol:isopropanol:tween 20, referred to as EIT. Among them, 14mb and 15b were the most active racemic form of 14 m and 15. The other conformation variants, including 14ma, 15a, 15c, and 15d did not inhibit the HIV infection.

PBMC from de-identified healthy donors were obtained with informed consent from the Department of Transfusion Medicine, National Institutes of Health under Biological reagent Registry RD-14-IX-12 approved by NIH biological safety board. CD8-depleted PBMCs, either stimulated or non-stimulated with 10 μg/mL phytohaemagglutinin (PHA) or anti-CD3 antibody (with or without conjugated beads), were plated at 2×10⁶ cells/ml in 48 well plates in RPMI 1640 + 10%FBS (R10) to a final volume of 1 mL in each well. IL-2 was added at a final concentration of 20 U/mL. CD3 antibody okt3 was used at ~0.2 μg/mL final concentration. Compounds were prepared in appropriate solvents according to the manufacturer’s instructions, diluted to working concentrations using RPMI 1640 + 10%FBS (R10), and added to the cell cultures for 6–24 h for CD62L shedding, and 48 h for TNF-α release. Supernatants were collected and assessed using either the CD62L or TNF-α Duo Set ELISA kit according to manufacturer’s instructions (R&D Systems, Inc.). The viability of cells treated with compounds was assessed by acridine orange / propidiom iodide (AO/PI Logosbio catalog F23001) staining using Luna-FL cell counter following the manufacture’s instruction.²

A fluorescence based enzymatic assay was used to measure the cleavage of fluorogenic TNF-α and L-selectin (CD62L) peptides by either recombinant ADAMs or PBMC from healthy donors. The fluorogenic substrate, synthesized by Biomatik³, was a Dabcyl (4-(4-dimethylaminophenylazo)benzoyl)- and Edans (5-[(2-aminoethyl)amino]naphthalene-1-sulfonic acid)- conjugated 12 amino acid peptide: Dabcyl-KLDKSFSMIKEG-Edans that corresponds to the extracellular membrane proximal region of CD62L. The fluorescence of the Edans group is quenched by the Dabcyl group in the intact substrate and the cleavage of the CD62L substrate peptide results in a fluorescence emission at 490 nm. The cleavage of fluorogenic TNF-α peptide (ES003, R &D systems, Inc) was detected at emission of 405 nm. The cleavage was measured in a kinetic assay every 3 min for 4 h in black 96-well plates using a Synergy_h1 fluorescent plate reader (BioTek) with 340 nm excitation and 490 nm emission wavelengths. For cleavage using recombinant ADAMs, each enzymatic reaction contained 50–200 ng of recombinant ADAM17 or 10 (R&D systems, Inc) mixed with 5–10 μM fluorogenic TNFα peptide ES003 (R&D systems, Inc) or CD62L peptide with or without indicated inhibitors in 100 μL assay buffer of 25 mM Tris at pH 9.0, 2.5 μM ZnCl2, and 0.005% Brij-35 (w/v). For cell-based cleavage of the fluorescent CD62L-peptide substrate, 300,000 PBMC at 2×10⁶/ml density were mixed with 10 μM fluorescent CD62L-peptide and various inhibitors in a black 96-well plate in 150 μL culture media (RPMI 1640, 10% FBS, 1% penicillin streptomycin). PBMC were stimulated with anti-CD3 antibody for 2–4 days as described above.

HIV-1 protease enzymatic activity was assayed in black 96-well plates at room temperature using a Synergy_h1 fluorescent plate reader with 340 nm excitation and 490 nm emission wavelengths (BioTek). Each enzymatic reaction consisted of a mixture of 100 ng of recombinant HIV-1 protease (Sigma-Aldrich catalog SRP2151), 5 μM concentration of a fluorogenic substrate Arg-Glu(EDANS)-Ser-Gln-Asn-Tyr-Pro-Ile-Val-Gln-Lys(DABCYL)-Arg (Sigma-aldrich catalog H6660), and various concentrations of inhibitor compounds in 100 μL of assay buffer (0.15 M sodium chloride, 1.0 mM EDTA, 1.0 mM DTT, 0.1 M sodium acetate, pH 5.0). The fluorescence of the cleavage reaction was continuously read at every 30 s for a total of 40 min.

The use of healthy human peripheral blood mononuclear cells (PBMC) is approved by the Department of Transfusion Medicine at the Clinical Center of National Institutes of Health. The use of PBMC from HIV-1 infected individuals for the study was approved under protocol #02-I-0202 by the Institutional Review Board (IRB) of National Institutes of Health with written informed consent from all donors. Peripheral blood mononuclear cells (PBMCs) from anonymous healthy donors were isolated by Ficoll-Paque gradient, and cultured in RPMI 1640 supplemented with 10% FBS, 1%Penicillin/streptomycin, 1% HEPES buffer and 20 U/mL IL-2 (culture media); and stimulated with 0.2 μg/mL anti-CD3 antibody (okt3) for 2 days. CD8⁺ T cells were depleted prior to infection using the Stemcell CD8 positive selection kit according to manufacturer’s instructions. CD8-depleted PBMCs were plated at 2×10⁶ cells/ml in 48-well plates in 1 mL culture media with various inhibitors for 60 min prior to the addition of ~80 TCID₅₀ HIV-1_BAL or HIV-1_LAI. Inhibitors were replenished during media change every 3 days. The infections were proceeded with or without inhibitors for 6 days or otherwise specified number of days and then analyzed by FACS staining using fluorescent labeled CD4-BV510, CD62L-APC and CD3-PEC7 antibodies (BD Biosciences). Cells were then washed and permeabilized with CytoFix/CytoPerm (BD Biosciences) solution before intracellular staining for HIV p24 antigen using FITC conjugated anti-p24 antibody clone KC57 (Beckman Coulter, catalog 6604665). All assays were performed with cell viabilities >80%. Replication competent HIV-1 viruses were obtained from the NIH AIDS Reagent Program⁴, and expanded by infecting activated CD8-depleted PBMCs, and TCID₅₀ was determined by p24 ELISA.

RT-PCR using ADAM9, 10 and 17 primer probes was assayed using TaqMan primer/probe sets from ThermoFisher with a 50 ng cDNA template as described previously (Chun et al., 2013; Kononchik et al., 2018). The following TaqMan 20X Gene Expression Assay probes (Dye: FAM-MGB) were used: ADAM9 (Hs00177638_m1), ADAM10 (Hs00153853_m1), ADAM17 (Hs01041915_m1). Expression Assay standard probes (Dye: VIC-MGB) for β-Actin (Hs01060665_g1) or RNase P/RPPH1 (4403328) were used as gene expression controls. Briefly, HIV infected or uninfected cells from triplicate experiments were washed 3x in PBS and lysed for RNA isolation using RNeasy mini kit (QIAGEN corp). The corresponding cDNA samples were prepared using either TaqMan microRNA reverse transcription kit or Invitrogen High-Capacity cDNA reverse transcription kit (Thermo Fisher). Samples were run in duplicate across triplicate experiments on an ABI 7300 Real-Time PCR System (Applied Biosystems) with a threshold of 0.2. Relative gene expression was calculated using the 2^-ΔCt method and normalized to β-Actin or RNaseP as the endogenous controls. All statistical analyses were carried out using the software Prism 8 (GraphPad Software, Inc). Western blot analyses were used to detect the mature active form of ADAMs (Pereira Vatanabe et al., 2021). In brief, total protein was extracted using RIPA lysis buffer (catalog 89900, Thermo USA) either from 60×10⁶ million PBMC for the expression of ADAM8, 10, 15, 17 and 19, or from 20 ×10⁶ HIV-1_BAL infected or uninfected PBMC for ADAM10 and 17 on day 3 of post infection. SDS samples of protein extracts were electrophoresed, transferred to PVDF (IB34002, Thermo, USA) membranes, blocked with 5% non-fat dry milk (catalog 1,706,404, BioRad, USA) containing 0.1% Tween20 (catalog P1379, Sigma, USA) for 2 h at RT, and incubated overnight at 4°C with antibodies against ADAM8 (1:1000 dilution, catalog MAB10311, R&D systems), ADAM10 (1:1000 dilution, catalog ab1997, Abcam, USA), ADAM15 (1:1000 dilution, catalog MAB935, R&D systems), ADAM17 (1:1000 dilution, catalog PA5-11572, Invitrogen, USA), ADAM19 (1:1000 dilution, catalog ab191457, Abcam), or beta-Actin (1:1000 dilution, catalog MA5-15739, Invitrogen, USA), respectively. The membranes were washed three times and incubated for 1 h at room temperature with the appropriate HRP-conjugated secondary antibodies: rabbit anti-mouse IgG (Abcam catalog 6728), goat anti-rabbit IgG heavy chain (Invitrogen catalog A271036). Detection of protein expression levels by enzyme-linked chemiluminescence (catalog 34580, Thermo, USA) was performed according to the manufacturer’s protocol.

CD4+ T cells were enriched from PBMC using EasySep^™ human CD4 T cell enrichment kit (STEMCELL Technologies Inc., MA 02142). To prepare samples for RNA sequencing, 5-10×10⁶ PBMC or CD4+ T cells from either infected or uninfected samples were lysed for RNA isolation using RNeasy mini kit from QIAGEN. Samples of 2–10 μg RNA with concentration > 25 ng/μl and OD260/280 > 2.0 were used for cDNA library preparation of 250–300 base pair fragments in length, and high throughput sequencing using Illumina Miseq platform by Novogene (Novogene Corp Inc., CA 91914). A total of ~20×10⁶ reads were recorded for each sample (Table 3). The mapping and assembly of raw sequencing reads, statistical and preliminary bioinformatic analyses were performed by Novogene. Subsequent gene expression analyses were performed based on expected Fragments Per Kilobase of transcript sequence per Millions base pairs sequenced (FPKM) using sequences of 20 reads or higher. All differential gene expression analyses were done between infected samples in the presence or absence of treatment and their donor matched uninfected controls. Cellular pathway analyses in response to HIV-1 infection in the presence and absence of BB-94 treatment were analyzed by inputting the top 400 upregulated and 400 down regulated genes from the RNAseq data to a bioinformatics software iDEP.96 (bioinformatics.sdstate.edu/idep/) using generally applicable gene set enrichment (GAGE) method (Luo et al., 2009; Sx et al., 2018). The RNAseq data is available in GEO database under accession number GSE218175.

CD8-depleted PBMC were infected with HIV-1_LAI in the presence and absence of 100 μM BB-94 as described in the methods. For immunogold labeling, ~10⁶ cells were washed with 2 mL PBS, gently centrifuged at 250 g for 5 min, resuspended in 100 μL of labeling buffer (PBS with 1% BSA), and incubated with 10 μg of mouse anti-human CD62L (clone FMC46, Thermofisher) for 1 h on ice. After removal of the primary antibody and wash, cells were incubated with 10 μg of secondary goat anti-mouse IgG H&L conjugated to 10 nm gold nanoparticles (Electron Microscopy Sciences) for 1 h on ice, then washed and fixed in 2.5% glutaraldehyde with 0.1 M sodium cacodylate at pH 7.4. Specimens for transmission electron microscopy (TEM) were fixed with 2.5% glutaraldehyde in 0.1 M Sorenson’s buffer. Samples were post-fixed 1 h with 0.5% osmium tetroxide/0.8% potassium ferricyanide, 1 h with 1% tannic acid and overnight with 1% uranyl acetate at 4°C. Samples were dehydrated with a graded ethanol series and embedded in Spurr’s resin. Thin sections were cut with a Leica UCT ultramicrotome (Vienna, Austria) stained with 1% uranyl acetate and Reynold’s lead citrate prior to viewing at 80 kV on a Hitachi 7500 transmission electron microscope (Hitachi-High Technologies, Tokyo, Japan). Approximately 100 images were acquired with an AMT digital camera system (AMT, Chazy, NY) from each sample on sections containing HIV virions. HIV virions are identified as ~100 nm sized particles with visible membrane enclosed capsid. The immunogold labeling resulted in less than 50 gold-particles/cell to minimize non-specific labeling.

PBMCs were collected from the study subjects in accordance with a clinical protocol approved by the Institutional Review Board of the National Institute of Allergy and Infectious Diseases, National Institutes of Health under protocol #02-I-0202. CD4⁺ T cells were isolated from the PBMCs using a human CD4 T cell separation system (StemCell Technologies, Catalog 19052) and were cultured with medium alone or with plate-bound anti-CD3 and soluble anti-CD28 antibody in the absence or presence of 50 μM of indicated compounds for 48 h. The copy number of cell-free viron-associated HIV RNA in the culture supernatants was quantified using the Cobas Ampliprep/Cobas Taqman HIV-1 Test, Version 2.0 (Roche Diagnostics) with droplet digital PCR (ddPCR, BIO-RAD) on day 2 and 4 of stimulation.

All experiments have been repeated at least two times and the results presented here are representative of the duplicates without data rejection. All statistics are calculated using multiple t-tests with p-values either indicated above their data or expressed as * = <0.05, ** = <0.005, *** = <0.0005, **** = <0.00005. Normalized inhibition data in Figure 6 are fitted with a non-linear regression model using GraphPad Prism 8.0.1.