Broadly neutralizing anti-HIV-1 antibodies tether viral particles at the surface of infected cells

Abstract

Broadly neutralizing antibodies (bNAbs) targeting the HIV-1 envelope glycoprotein (Env) are promising molecules for therapeutic or prophylactic interventions. Beyond neutralization, bNAbs exert Fc-dependent functions including antibody-dependent cellular cytotoxicity and activation of the complement. Here, we show that a subset of bNAbs targeting the CD4 binding site and the V1/V2 or V3 loops inhibit viral release from infected cells. We combined immunofluorescence, scanning electron microscopy, transmission electron microscopy and immunogold staining to reveal that some bNAbs form large aggregates of virions at the surface of infected cells. This activity correlates with the capacity of bNAbs to bind to Env at the cell surface and to neutralize cell-free viral particles. We further show that antibody bivalency is required for viral retention, and that aggregated virions are neutralized. We have thus identified an additional antiviral activity of bNAbs, which block HIV-1 release by tethering viral particles at the surface of infected cells.

Introduction

The HIV-1 envelope (Env) glycoprotein is a trimer of gp41/gp120 heterodimers. It is the only viral protein present at the surface of viral particles. Env mediates entry into target cells, which makes it the target of neutralizing antibodies. Broadly neutralizing antibodies (bNAbs) have been isolated from patients called “elite neutralizers” and inhibit the majority of HIV-1 strains¹. They target conserved sites of vulnerability at the surface of Env: the CD4 binding site (CD4bs), the N-glycans associated with the V1/V2 and V3 loops, the silent face of gp120, the membrane proximal external region (MPER) of gp41 and a larger site spanning the interface between gp41 and gp120. In both non-human primates and humanized mice, infusion of bNAbs decreases viral loads^2,3, prevents infection^4,5,6 and delays viral rebound^7,8. Several bNAbs are under clinical evaluation^9,10. The anti-CD4bs bNAbs 3BNC117 and VRC01, and the anti-V3 10–1074 decrease viral loads in HIV-1-infected viremic individuals^11,12,13. 3BNC117, alone or in combination with 10–1074, delays viral rebound after antiretroviral therapy (ART) interruption^14,15.

In addition to neutralizing viral particles, bNAbs recognize infected cells and recruit immune effectors through their Fc domain. Such Fc-dependent activities include antibody-dependent cellular cytotoxicity (ADCC)^16,17,18, antibody-dependent phagocytosis (ADCP)¹⁹ or complement activation²⁰. The contribution of Fc-effector functions to antibody-mediated protection against simian-human immunodeficiency virus (SHIV) acquisition in animal models is debated^21,22,23,24. The discrepant results may depend on the capacity of the tested antibody to neutralize the virus used for challenge, with the contribution of Fc-effector functions increasing as neutralization potency decreases. Fc-effector functions boost bNAbs therapeutic efficacy in macaques²⁵ and are required to efficiently target the reservoir in humanized mice⁷. Therefore, the non-neutralizing activities of bNAbs contribute to their in vivo efficacy.

A few antibodies directed against other viruses inhibit the assembly or the release of viral particles. For instance, a neutralizing monoclonal antibody (mAb) directed against Chikungunya virus (CHIKV) blocks envelope-driven viral assembly, leading to the intracellular accumulation of immature viral-like particles and inhibition of viral release^26,27. A mAb against influenza virus inhibits viral egress by extracellularly aggregating mature viral particles²⁸. Similar inhibition of release was demonstrated for antibodies targeting Marburg (MARV) or Ebola (EBOV) viruses^29,30. The impact of anti-HIV-1 bNAbs on viral release has yet to be examined.

The steps of HIV-1 budding are well characterized. HIV-1 Gag p55 are anchored in the inner leaflet of the plasma membrane, forming a lattice that initiates particle formation³¹. In parallel, Env accumulates at the budding site through interactions between the gp41 cytoplasmic tail and Gag³¹. This is followed by ESCRT-mediated budding, viral release and particle maturation by the viral protease, which cleaves Gag into virion-associated proteins p24, p17, p7, p6, p2 and p1³². Whether bNAbs interfere with these processes is unknown.

Here, we show that a subset of bNAbs inhibits HIV-1 release from infected CD4 T cells. bNAbs tether mature viral particles as large extracellular immune complexes without inhibiting budding or maturation.

Results

bNAbs inhibit HIV-1 release from infected CD4 T cells

We first asked whether bNAbs impair HIV-1 release. To this aim, we infected primary CD4 T cells with HIV-1 for two days, washed the cells to replace the medium and then subjected infected cells to treatment with a panel of bNAbs or an isotype control (mGO53) for 24 h. Since the capacity of bNAbs to neutralize HIV-1 varies across antibodies and viral strains, we added antiretrovirals (azidothymidine [AZT] and lamivudine [3TC]) during bNAbs treatment (Fig. 1a). To further avoid any confounding effect of viral replication, we treated cells with antibodies at the time of peak viral replication (Supplementary Fig. 1a). This strategy allowed the normalization of the frequency of infected cells across isotype- and bNAb-treated conditions (Supplementary Fig. 1a). AZT and 3TC are reverse-transcriptase inhibitors that act early in the viral cycle, without interfering with p24 production by cells productively infected at the time of addition. Accordingly, p24 production is reduced but not halted by addition of AZT-3TC (Supplementary Fig. 1b). We analyzed viral release in supernatants by ELISA and assessed cell-associated Gag by flow cytometry and microscopy (Fig. 1a). We used three HIV-1 strains: the lab-adapted AD8 isolate, a transmitted/founder strain (CH058) and a clade B virus isolated from the reservoir of an ART-treated patient (vKB18)¹⁶. We first used two bNAbs, 10–1074 and 3BNC117, which target the V3 loop and the CD4bs, respectively. We normalized results to the condition without antibody. Both bNAbs, but not the isotype control, decreased p24 levels in the supernatant (Fig. 1b). This reduction reached 64% with 10–1074 in vKB18-infected cells. This effect was not due to a residual inhibition of viral spread by bNAbs, as the frequency of infected cells (as measured by a Gag-specific staining) was similar regardless of the antibody tested (Supplementary Fig. 1c). We then analyzed infected cells (defined as CD4^– Gag⁺) by flow cytometry (Supplementary Fig. 1d). Both 10–1074 and 3BNC117 increased the median fluorescence intensity (MFI) of cell-associated Gag (Fig. 1c, d). The most potent effect was again observed with 10–1074 in vKB18-infected cells (1.8-fold increase in Gag MFI compared to control antibody). A kinetic analysis performed with 10–1074 and CH058-infected cells revealed that Gag MFI increases over 24 h (Supplementary Fig. 1e). Of note, adding extra PBS washes or a proteolytic degradation of cell surface proteins by trypsin prior to bNAbs treatment had no impact on the increase in MFI induced by 10–1074 (Supplementary Fig. 2a). We then tested a panel of 17 anti-Env antibodies including bNAbs and non-neutralizing antibodies (nnAbs) (Fig. 1e, Supplementary Fig. 2b, c, and Supplementary Table 1). For the three viral strains, bNAbs targeting the V3 and V1/V2 loops or the CD4bs were the most efficient at decreasing viral release and increasing cell-associated Gag. Both activities were correlated and dose-dependent (Supplementary Figs. 2d, e and 3). In contrast, bNAbs targeting the MPER, the gp120/gp41 interface and nnAbs were inactive….