Broadly neutralizing antibodies that inhibit HIV-1 cell to cell transmission

We initially assessed the efficacy of the selected bNAbs to inhibit HIV-1 cell to cell transmission in culture using flow cytometry (Sourisseau et al., 2007). Primary CD4⁺ T cells infected with either NLAD8 or NL4.3 HIV-1 strains were incubated with bNAbs before co-culture with autologous target cells labeled with FarRed. Infection of target cells was measured by Gag expression. In the absence of bNAbs, 15–50% of the recipient cells were Gag⁺ after 48 h (representative experiments are shown in Fig. 1 a). Under these conditions, the contribution of cell-free virus to infection was negligible because Gag expression by the recipient cells was abrogated by separation of donors and targets in a transwell chamber or when cultures were gently shaken to avoid prolonged contacts (Sourisseau et al., 2007). Moreover, Gag expression in target cells was caused by de novo synthesis because it was significantly reduced in the presence of reverse transcription inhibitor (nevirapine [NVP]; Fig. 1 a).

We tested a panel of 15 bNAbs to investigate their effects on viral cell to cell transmission and in parallel measured their neutralization activity in a conventional cell-free assay (Abela et al., 2012; Heyndrickx et al., 2012). As expected, all bNAbs blocked cell-free infection with either NLAD8 or NL4.3, with IC50s varying from 0.05 to 9 µg/ml (Table 1 and Fig. S1). However, only a subset efficiently inhibited cell to cell transmission (Table 1 and Fig. S1). Representative inhibition curves with one active (NIH45-46) and one inactive (8ANC195) bNAb are shown in Fig. 1 b. The most active antibodies in the cell to cell transmission assay were NIH45-46 and 3BNC60, which target the CD4bs, and clonally related anti-glycan/V3 antibodies, 10-1074, 10-996, and PGT121 (Table 1 and Fig. S1; IC50s < 0.9–2.5 µg/ml). Time of addition experiments showed that NIH45-46 and 3BNC60 remained strongly active when added up to 6 h after the beginning of the co-culture (not depicted). However, the antibody concentrations required to inhibit cell to cell transmission were 10–20-fold higher than cell-free transmission (Table 1 and Fig. S1). As previously reported (Abela et al., 2012), VRC01, a less active clonal relative of NIH45-46, was only partially effective in blocking cell to cell transmission, with 65 and 85% inhibition of NLAD8 and NL4.3, respectively, at concentrations of 15 µg/ml. NIH45-46 and 3BNC60 also inhibited viral spread when CD4⁺ T cells were co-cultured with MOLT cells chronically infected with BaL (not depicted). In contrast, most of the other antibodies tested were relatively ineffective, including the two anti-MPER antibodies and PG16. 10E8 was only partially active against NL43, with 75% inhibition at a concentration of 15 µg/ml, and PG16 showed only 60% inhibition at this concentration.

Some of the bNAbs were also tested for their ability to block cell to cell transmission of two transmitted/founder (T/F) HIV-1 viruses (Salazar-Gonzalez et al., 2009; Ochsenbauer et al., 2012). The IC50s of the antibodies against these viruses were generally similar to those observed with NLAD8 (Table 2). The more active bNAbs against T/F viruses were PG16, 10-1074, and 3BNC60 (Table 2).

The first step in cell to cell spread is the formation of conjugates between infected cells and targets, a process which is dependent on Env (Massanella et al., 2009). To examine the mechanism of inhibition of cell to cell transmission by bNAbs, we initially assayed their effects on conjugate formation. Preincubation of NL4.3- or BaL-infected lymphocytes with NIH45-46, 3BNC60, or VRC01 decreased the formation of conjugates, and 10-1074 and PGT121 inhibited the formation of clusters with cells expressing HIV-BaL (Fig. 2, a and b). In contrast, the other bNAbs did not show measurable inhibitory effects (Fig. 2, a and b). Of note, 10E8 increased formation of conjugates with NL4.3 and was ineffective against BaL (Fig. 2, a and b). It is possible that 10E8, which blocks Env-mediated fusion without affecting Env-CD4 interaction, may stabilize cell conjugates in the absence of cell fusion.

Contact between Env⁺ cells and target cells can lead to formation of syncytia. Therefore, we assessed the effect of bNAbs on syncytia formation using a quantitative assay, in which HeLa cells expressing Env (from NL4.3) and Tat are mixed with HeLa cells expressing CD4 and HIV-1 LTR–β-gal reporter cassette (P4C5 cells; Schwartz et al., 1994). β-gal expression is induced upon fusion and transfer of Tat between the two cell types. NIH45-46, 3BNC60, and VRC01 blocked fusion as did the anti-gp41 MPER 10E8 (Fig. 2 c). In contrast, bNAbs that failed to inhibit cell to cell transmission of NL4.3 had little or no effect on fusion (Fig. 2 c).

After formation of cell conjugates, viral particles are transferred from infected cells to targets. To measure viral transfer, we examined Gag expression in target cells by flow cytometry. 4 h after initiation of the co-culture, 4–6% of target cells were Gag⁺ (Fig. 3 a). NIH45-46, 3BNC60, VRC01, and 10-1074 decreased transfer of NL4.3 and/or NLAD8, whereas the other antibodies were inactive (Fig. 3 a). Immunofluorescence and confocal imaging further demonstrated that preincubation of donor cells with NIH45-46 or 3BNC60 led to accumulation of the bNAb at the virological synapse and to colocalization with Gag (Fig. 3 b). Moreover, an examination of target cells that captured viral particles despite the presence of bNAbs demonstrated that these viruses were generally, if not always, coated with bNAbs (Fig. 3 c).

Plasmacytoid DC (pDCs) sense HIV-1–infected cells and react by producing type-I IFN (Lepelley et al., 2011). To determine whether bNAbs inhibit the recognition of infected lymphocytes by pDCs, infected donor cells were preincubated with bNAbs and co-cultured with target PBMCs. The levels of type-I IFN were measured in supernatants after 18 h of co-culture. NIH45-46, 3BNC60, VRC01, and 10-1074 inhibited IFN induction by lymphocytes infected with NL4.3 and/or NLAD8, whereas PG16 and 10E8 were inactive (Fig. 4 a). In addition, the role of antibodies in blocking transmission was examined using the pDC-like cell line Gen2.2 as target cells (Lepelley et al., 2011). NIH45-46, 3BNC60, VRC01, 10-1074, PG16, and 10E8 blocked HIV-1 infection of Gen2.2 cells (Fig. 4 b). Moreover, bNAbs that blocked Gen2.2 infection also inhibited induction of the IFN-induced protein MxA. Thus, in addition to inhibiting lymphocyte to lymphocyte viral spread, the most active bNAbs block transmission from lymphocytes to pDCs and impair innate immune sensing of the virus.

To date, only VRC01 among the second generation, more potent, bNAbs has been assayed for its ability to block HIV-1 cell to cell transmission, and in agreement with our own findings, it was reported to be poorly active (Chen et al., 2007; Massanella et al., 2009; Martin et al., 2010; Abela et al., 2012). Here, we show that many of the 15 bNAbs tested, including some of the most active antibodies to six different sites of vulnerability on the envelope, were ineffective in HIV-1 cell to cell transmission. However, five bNAbs, isolated from four different patients (NIH45-46, 3BNC60, PGT121, 10-1074, and PG16) blocked the intercellular spread of the tested viruses at IC50s of 0.05–2.5 µg/ml. NIH45-46 and 3BNC60 target the CD4-binding site, whereas 10-1074 and PGT121 bind to a complex epitope composed of glycans and the V3 loop epitope on HIV-1 gp120 (Scheid et al., 2009; Walker et al., 2011; Wu et al., 2011; Kwong and Mascola, 2012; Mouquet et al., 2012; Liao et al., 2013). Of note, 10-1074 and PGT121 efficiently inhibited cell-free and cell to cell transmission of all R5 viruses tested but were inactive against the X4 strain NL4.3. Further work will help determining whether V3-directed bNAbs inhibit other X4 isolates.

The bNAbs inhibiting cell to cell viral spread were those acting at a low concentration against free virus (IC50 < 0.2 µg/ml; Scheid et al., 2011; Walker et al., 2011). This suggests that very high levels of potency in cell-free virion assays are required for antibodies to block viral cell–cell spread. But potency alone may not be sufficient, as indicated by 10E8, which fails to reach 80% neutralizing activity in cell to cell transmission despite potencies that are nearly comparable with the effective antibodies in cell-free assays. It is noteworthy that the concentrations required to inhibit cell to cell transmission were at least 10–20-fold higher than for free virus. Similarly, the serum concentrations of bNAbs required to inhibit infection in mouse or monkey models of HIV-1 infection are also 1–2 logs higher than in cell-free assays (Balazs et al., 2012; Klein et al., 2012a; Moldt et al., 2012; Pietzsch et al., 2012). Cell to cell transmission may therefore provide a more reliable method for predicting the potency of bNAbs in vivo.

Our experiments show that the most active bNAbs interfere with cell to cell transmission by efficiently impairing formation of clusters between infected cells and targets, the appearance of syncytia, and transfer of viral material to recipient cells and by accumulating at the virological synapse. Moreover, they also decorate any free viral particles that may still have been captured by target cells. Additional experimentation will be required to determine whether the various bNAbs display differing abilities to accumulate at sites of cell–cell contacts and to impair the formation of virological synapses. It will also be of interest to further examine how these bNAbs interfere with viral cell to cell transmission between myeloid DCs and lymphocytes.

HIV-1 cell to cell transmission is likely playing a predominant role in infected individuals (Murooka et al., 2012). Our results confirm that this mode of viral spread may represent a means to escape the selection pressure exerted by most of the bNAbs (Ganesh et al., 2004; Abela et al., 2012). Our observations may also help explain why some bNAbs like 3BC176 are ineffective in vivo (Klein et al., 2012b), whereas others like 10-1074 and a derivative of NIH45-46 (45-46W) suppressed viral loads below detection (Klein et al., 2012b). Consistent with this idea, 10E8, which does not efficiently inhibit HIV-1 cell to cell transmission, failed to suppress viremia in vivo and failed to select antibody-resistant HIV-1 variants (Fig. S2). We speculate that bNAbs that effectively interfere with cell to cell transmission in vitro will also display efficient and long-lasting therapeutic or prophylactic properties in vivo.

MOLT cells chronically infected with NL4.3 or BaL isolates and MT4C5 cells were grown as described previously (Massanella et al., 2009; Rudnicka et al., 2009). Primary CD4⁺ T cells were purified from human peripheral blood by positive selection (Miltenyi Biotec). About 98% of cells were CD4⁺CD3⁺. For activation, primary T cells were treated with 1 µg/ml PHA for 24 h and then cultured with 50 IU/ml IL-2 for 3–5 d before use. NL4.3, NLAD8, and T/F HIV-1 strains WITO and SUMA (Ochsenbauer et al., 2012) were obtained through the National Institutes of Health AIDS reagents program and prepared by transfection of 293T cells (Rudnicka et al., 2009). Primary cells were infected with HIV-1 as described previously (Lepelley et al., 2011). HIV-1 strains used are R5, except NL4.3 (X4). The env genes correspond to viruses with high (tier 1 for NL4.3 and BaL) or moderate (tier 2 for WITO) sensitivity to antibody-mediated neutralization (Seaman et al., 2010).

bNAbs were prepared as described previously (Scheid et al., 2009; Mouquet et al., 2012).

Donor cells were infected with the indicated HIV-1 strains and used a few days later, when ∼10–75% of the cells were Gag⁺. Target cells were labeled with FarRed or 2.5 µM CFSE (Molecular Probes). Donors were preincubated with the indicated doses of bNAbs. Donor and target cells were then mixed at a 1:2 ratio in 96-well plates at a final concentration of 1.5 × 10⁶/ml in 200 µl. After 48 h, cells were stained for intracellular Gag (KC57 mAb; Beckman Coulter) and analyzed by flow cytometry. When stated, 12.5 nM NVP was added 1 h before co-culture.

The neutralization activity of bNAbs was evaluated on P4C5 cells (HeLa CD4⁺CCR5⁺ cells carrying an HIV-1 LTR–β-gal reporter cassette). 1 d before infection, 7 × 10³ cells were plated in 96-well plates. Cells were infected in triplicate with 1 or 5 ng Gag p24. Viruses were incubated with the indicated bNAbs for 1 h before infection. After 36 h, cells were lysed in PBS, 0.1% NP-40, and 5 mM MgCl₂ and incubated with the β-gal substrate CPRG (Roche), before measurement of 570-nm OD.

Dose–response inhibition curves were drawn by fitting data from three to six independent experiments to sigmoid dose–response curves (variable slope) using Prism software (GraphPad Software). Percentage of inhibition was defined as (percent signal in nontreated target cells − percent signal in bNAb-treated cells)/(percent signal in nontreated target cells) × 100. The IC50 was calculated with Prism.

Chronically infected MOLT cells (2 × 10⁵/well) were stained with CFSE and preincubated in 96-well plates with 15 µg/ml bNAbs for 1 h at 37°C, before adding 2 × 10⁵ FarRed-labeled CD4 T cells for 2 h (Massanella et al., 2009). Cellular contacts were measured by flow cytometry. CFSE⁺FarRed⁺ cells were considered as cellular conjugates. The percentage of conjugates was calculated as follows: (conjugates/total conjugates + MOLT cells + CD4 T cells) × 100.

Primary donor cells were infected with HIV-1 and used a few days later, when 10–75% of the cells were Gag⁺. The indicated target cells were labeled with FarRed or 2.5 µM CFSE (Molecular Probes). Donors were incubated with bNAbs for 1 h. Donor and target cells were then mixed at a 1:2 ratio in 96-well plates at a final concentration of 1.5 × 10⁶/ml in 200 µl. After 4 h, target cells were stained for Gag and analyzed by flow cytometry.

HeLa 243 cells, which stably express the HIV-1 NL4.3 Env and Tat proteins (Schwartz et al., 1994), seeded at 8 × 10³ cells in 96-well plates, were preincubated for 1 h with 15 µg/ml bNAbs, before addition of 8 × 10³ P4C5 cells for 8 h. Upon cell fusion, Tat transactivates the HIV-1 LTR and drives expression of β-gal, at levels which correlate with the amount of syncytia (Schwartz et al., 1994).

HIV-infected primary donor CD4⁺ T cells were treated with the indicated bNAbs for 1 h before addition of autologous targets. After 1 h of co-culture, cells were fixed and double Gag/bNAb stainings were performed with a mouse mAb anti-p24 and a secondary anti–human antibody to visualize the bNAb. Confocal microscopy analysis was performed on an LSM700 (Carl Zeiss) using a 40 or 63 objective. Z-series of optical sections were performed at 0.2–0.5-µm increments.

The indicated infected donor cells were incubated for 1 h with the bNAb and mixed with target cells (PBMCs or Gen2.2) at a ratio of 1:2 in 96-well plates at a final concentration of 2–3 × 10⁶/ml in 100 µl for PBMCs, or 10⁶/ml in 200 µl for Gen2.2, as described previously (Lepelley et al., 2011). After a co-culture of 18 h (PBMCs) or 48 h (Gen2.2), type I IFN levels and expression of the protein MxA were measured as described previously (Lepelley et al., 2011; Puigdomènech et al., 2013).

HIV-1 infection of humanized mice, bNAb treatments, and analysis of viral loads and sequences were performed as described previously (Klein et al., 2012b). Animal experiments were performed with authorization from the Institutional Review Board and the Institutional Animal Care and Use Committee at the Rockefeller University.

Wilcoxon-matched paired Student’s t tests (and Mann–Whitney unpaired Student’s t test in Fig. 3 b) were performed using Prism software.

Fig. S1 shows dose–response analysis of the antiviral effect of bNAb on cell-free and cell-associated infections. Fig. S2 shows that 10E8 treatment does not impose selective pressure on HIV-1 infection in humanized mice.

We thank members of the Virus and Immunity Unit for help and critical reading of the manuscript. We thank the National Institutes of Health AIDS reagents program for the gift of reagents.

O. Schwartz was supported by grants from the Agence Nationale de Recherche sur le Sida, Sidaction, AREVA Foundation, the Vaccine Research Institute, the Labex IBEID program, the FP7 program HIT Hidden HIV (Health-F3-2012-305762), and Institut Pasteur. M. Malbec was supported by fellowships from Sidaction, Fonds de dotation Pierre Bergé, and Fondation pour la Recherche Médicale. H. Mouquet was supported by the Milieu Intérieur program (ANR-10-LABX-69-01). M.C. Nussenzweig was supported by grants from the Bill and Melinda Gates Foundation (Collaboration for AIDS Vaccine Discovery grant 38619s) and the Center for HIV/AIDS Vaccine Immunology and Immunogen Discovery grant AI 100663-01. M.C. Nussenzweig is a Howard Hughes Medical Investigator.

The authors declare no competing financial interests.

Author contributions: M. Malbec, H. Mouquet, M.C. Nussenzweig, and O. Schwartz conceived of the study. M. Malbec, F. Porrot, C. Eden, J.F. Scheid, F. Klein, R. Rua, J. Horwitz, A. Halper-Stromberg, H. Mouquet, M.C. Nussenzweig, and O. Schwartz performed the experiments and/or participated in the experimental design. M. Malbec, H. Mouquet, M.C. Nussenzweig, and O. Schwartz wrote or edited the manuscript. All authors approved the final manuscript.