Non-Neutralizing Antibodies Alter the Course of HIV-1 Infection in Vivo

SUMMARY

Non-neutralizing antibodies (nnAbs) to HIV-1 show little measurable activity in prevention or therapy in animal models, yet were the only correlate of protection in the RV144 vaccine trial. To investigate the role of nnAbs on HIV-1 infection in vivo, we devised a replication-competent HIV-1 reporter virus that expresses a heterologous HA-tag on the surface of infected cells and virions. Anti-HA antibodies bind to, but do not neutralize the reporter virus in vitro. However, anti-HA protects against infection in humanized mice and strongly selects for nnAb-resistant viruses in an entirely Fc-dependent manner. Similar results were also obtained with tier 2 HIV-1 viruses using a human anti-gp41 nnAb, 246D. While nnAbs are demonstrably less effective than broadly neutralizing antibodies (bNAbs) against HIV-1 in vitro and in vivo, the data show that nnAbs can protect against and alter the course of HIV-1 infection in vivo.

eTOC

Could the induction of non-neutralizing antibodies represent an alternative solution to the HIV vaccine problem?

INTRODUCTION

The HIV-1 virus encodes a single surface protein, gp160. The spike is a trimer of furin-cleaved gp160s, each of which is composed of two non-covalently associated subunits, gp41 and gp120. The cellular receptor CD4 binds to gp120, leading to dissociation of the trimer and exposing epitopes commonly targeted by non-neutralizing antibodies (nnAbs) against HIV-1 Env.

Nearly all HIV-1 infected individuals develop nnAbs early in the course of infection. The initial response is typically directed against gp41 and later evolves to include gp120 (Tomaras et al., 2008). After a variable period, usually measured in years, a fraction of infected individuals goes on to develop potent broadly neutralizing antibodies (bNAbs), which have been shown to target a number of epitopes on the trimer (Klein et al., 2013b). The unusual delay between infection and bNAb development, and the difficulty in eliciting them by immunization, is explained in large part by their extraordinary levels of somatic mutations (Andris et al., 1991; Escolano et al., 2017; Klein et al., 2013a; Moran et al., 1993; Scheid et al., 2009).

bNAbs neutralize a diverse collection of HIV-1 strains in vitro and can confer protection against infection in mice and macaques (Balazs et al., 2012; Bournazos et al., 2014; Gautam et al., 2016; Hessell et al., 2007; Moldt et al., 2016; Pegu et al., 2014) and reviewed in (Pegu et al., 2017). Moreover, they can suppress established viremia in experimental animals and in humans (Caskey et al., 2015; Caskey et al., 2017; Horwitz et al., 2013; Klein et al., 2012; Scheid et al., 2016) and reviewed in (Margolis et al., 2017).

In contrast, nnAbs do not neutralize HIV-1 in vitro, and have variable activity in antibody dependent cellular cytotoxicity (ADCC) assays in vitro (Bruel et al., 2017; Chung et al., 2014; Holl et al., 2006; Kramski et al., 2013; Lewis, 2013). In the RV144 trial, induction of nnAbs capable of mediating ADCC in vitro was associated with decreased transmission risk (Haynes et al., 2012; Liao et al., 2013). However, there is little direct evidence that nnAbs exert selective pressure on HIV-1 during infection in vivo (Burton et al., 2011; Chung et al., 2011; Santra et al., 2015). Vaccination studies in non-human primates (NHP) have suggested a role for non-neutralizing immune responses in protection against SIV or SHIV infection, but a causal link has not been proven experimentally (reviewed in (Lewis et al., 2017)). In these NHP studies, different vaccination strategies induced nnAb responses that correlated with protection from infection (Barouch et al., 2015; Barouch et al., 2013; Demberg et al., 2007; DeVico et al., 2007; Florese et al., 2009; Fouts et al., 2015; Hidajat et al., 2009). However, these studies did not characterize the mechanisms of protection, and virus resistance to nnAbs was not investigated. Finally, passive-transfer studies with high doses of potent monoclonal nnAbs failed to demonstrate protection (Burton et al., 2011; Santra et al., 2015). Thus, although nnAbs remain the only correlate of the modest level of protection against infection observed in the RV144 vaccine trial conducted in Thailand (Bonsignori et al., 2012; Haynes et al., 2012; Tomaras et al., 2013; Zolla-Pazner et al., 2014), how nnAbs might have contributed to protection remains unclear.

Here we report on experiments that explore whether nnAbs can influence the course of HIV-1 infection in vivo, and we define the mechanisms by which they do so.

RESULTS

Replication-Competent Indicator Virus

To examine the effects of nnAbs on HIV-1 infection, we engineered a recombinant HIV-1 reporter virus (HIVivoHA) expressing a heterologous surface marker, mCD24/heat-stable antigen (HSA) bearing a hemagglutinin tag (HA-tag), and adapted it to replicate in hu-mice (Figures 1A and S1). The HA-tag is anchored to the membrane of HIVivoHA virions and infected cells by the transmembrane domain of HSA and it is entirely independent of HIV-1 Env. Accordingly, a humanized IgG1 anti-HA-tag antibody (anti-HA) (Churchill et al., 1994) that stains HIVivoHA-infected cells (Figure S1C) exhibited no neutralizing activity against the virus in TZM-bl assays in vitro, even at concentrations as high as 50 μg/ml (Figure 1B). Finally, hu-mice infected with HIVivoHA sustain infection with the indicator virus at high viral loads for periods of at least 39 days (Figure S1H).

Figure 1. Fc-dependent clearance of HIVivoHA-infected cells.

(A) Schematic representation of HIVivoHA (see also Figure S1F). Numbering below corresponds to the nucleotide position in the HIVivoHA genome. Viral strains composing the chimeric genome are indicated above; pink shading indicates the region of env derived from strain Q23.17. An ‘x’ indicates a dysfunctional gene. IL16, minimal human IL-16 promoter; HA, mCD24/HSA gene encoding an HA-tag. (B) TZM-bl neutralization activity (IC₅₀ and IC₈₀) of 3BNC117 and anti-HA IgG against HIV-1_YU2 and HIVivoHA. Mean IC₅₀-values and IC₈₀-values are shown (μg/ml). (C) Scatter plot showing the percentage of Gag⁺ cells among CD3⁺CD8⁻ cells 5 hours after adoptive transfer of HIVivoHA-infected cells to mice treated with anti-HA (blue dots), anti-HA:GRLR (red dots) or to untreated mice (gray dots). Horizontal black line indicates the mean percentage of Gag⁺ cells in each group. Significant differences are indicated: ***p <0.001, two-tailed Mann-Whitney U test. (D) Plasma viremia in anti-HA-treated mice (left) and anti-HA:GRLR-treated mice (right). X-axis depicts days post HIV-1 infection (DPI); Y-axis depicts viral RNA copies/ml. Gray shading indicates the treatment period with the indicated antibodies. (E) Flow cytometry staining with anti-Gag antibody and anti-HA antibody of infected cells isolated from spleen of anti-HA-treated mice (top) or anti-HA:GRLR-treated mice. Red circles mark the position of Gag⁺HA⁺ cells. X-axis depicts anti-Gag antibody staining; Y-axis depicts anti-HA antibody staining. (F) Single genome sequencing of viruses from HIVivoHA-infected animals treated with anti-HA (top) or anti-HA:GRLR. Each horizontal gray bar represents the sequence of a single sequence aligned to HIVivoHA; black tick marks indicate nucleotide changes relative to HIVivoHA. Numbering at the bottom corresponds to the nucleotide position in the HIVivoHA genome. See also Supplementary Figure 1.

nNAbs Clear Infected Cells and Protect Against Infection in Hu-Mice

To determine whether anti-HA enhances clearance of HIVivoHA-infected cells in vivo, we performed adoptive transfer experiments (Lu et al., 2016) of cells infected with the indicator virus in the presence or absence of anti-HA antibody. Anti-HA antibody directed near-complete clearance of HIVivoHA-infected cells from the mice within five hours (Figure 1C), as would be expected for antibody-driven, Fc-dependent clearance of cells bearing a highly expressed surface antigen (Smith et al., 2012). By contrast, a modified anti-HA that carries mutations which abrogate binding to activating Fc receptors (anti-HA:GRLR (Pietzsch et al., 2012)) failed to clear HIVivoHA-infected cells, and the percentage of HA⁺Gag⁺ cells was comparable to the percentage seen in untreated mice. Thus, anti-HA, which is a nnAb, can clear HIV-1-infected cells in vivo in an Fc receptor-dependent manner (Figure 1C).

To determine whether a nnAb can affect the course of viral infection in vivo, we established HIVivoHA infection in hu-mice and then treated with anti-HA or anti-HA:GRLR. Effects on viral load after anti-HA administration were modest (Figure 1D). However, flow cytometry showed that infected cells from anti-HA treated animals were devoid of surface HA expression, while anti-HA:GRLR treated animals retained HA (Figure 1E). Single-genome sequencing of plasma viruses revealed that loss of HA expression in anti-HA treated mice was due to mutation or deletion of the HSA-HA cassette. In contrast, the HA cassette remained intact in most viruses isolated from anti-HA:GRLR treated animals, indicating an Fc-dependent mechanism of selection (Figure 1F). Thus, anti-HA imposes selective pressure on HIV-1 by an Fc-dependent mechanism that leads to predominance of anti-HA-resistant viruses.

To determine whether nnAbs can also prevent HIV-1 infection in hu-mice, we administered anti-HA or anti-HA:GRLR to hu-mice one day before and one day after challenge with HIVivoHA. Whereas all mice receiving no treatment or anti-HA:GRLR became viremic, a majority (55%) of animals treated with anti-HA were protected from HIVivoHA infection (Figure 2). The 45% of anti-HA treated mice that did become viremic did so with delayed onset, after the serum antibody concentrations had waned, and the emerging viruses had lost the HA expression cassette (Figures 2A and 2B). Thus, anti-HA protected against HIVivoHA infection and blocked the outgrowth of HA-expressing viruses in hu-mice. We conclude that nnAbs can prevent infection when they target a highly expressed surface epitope, albeit to a lesser extent than bNAbs (Balazs et al., 2012; Burton et al., 2011; Gautam et al., 2016; Halper-Stromberg et al., 2014; Shingai et al., 2014), and do so entirely by an Fc receptor-mediated mechanism.

Figure 2. Anti-HA nnAb protects against HIVivoHA infection in Hu-mice.

(A–B) Plasma viremia in anti-HA-treated mice (A) or anti-HA:GRLR-treated mice (B). X-axis depicts days post-infection. Left Y-axis depicts viral RNA copies/ml; right Y-axis depicts antibody concentration in mouse plasma (μg/ml). Black lines depict the plasma viremia of each mouse; red line depicts the average antibody concentration in mouse plasma. Black filled circles depict mice that became viremic; open circles depict aviremic animals. Black arrows indicate antibody injection. The dashed line depicts the quantitation limit (800 copies/ml). Pie charts show the percentage of all sequences from all animals carrying intact (white) or mutated (black) HA expression cassettes at 37 DPI. (C) Proportion of animals that were viremic (black) or aviremic (white) in each indicated treatment group at 37 DPI. Percentage of aviremic hu-mice following anti-HA treatment is indicated.

A Patient Derived nnAb Alters the Course of HIV-1_YU2 Infection in Hu-Mice

nnAbs are among the first antibodies to arise, and are also the most prevalent antibodies found in infected individuals (Xu et al., 1991). Whether these antibodies influence the course of infection in vivo, and by which mechanism(s) they might do so, has not been studied directly. To determine whether a patient-derived nnAb can exert selective pressure on HIV-1 infection, we screened for nnAbs that bind to HIV-1_YU2-infected CD4⁺ T-cells (Figure 3A). Monoclonal Ab 246D (Xu et al., 1991), which is directed against a linear peptide in the immunodominant cluster I region of gp41, exhibited the strongest binding (Figure 3A). Antibodies to this region are known to be highly cross-clade reactive, binding to the surface of infected cells and virus particles (Nyambi et al., 2000; Zolla-Pazner et al., 1995). However, these antibodies do not bind to intact trimers because the epitope they recognize is not accessible. Instead, they recognize gp41 stumps that are found on the surface of infected cells and viruses due to shed gp120 (Moore et al., 2006).

Figure 3. Anti-gp41 nnAb 246D selects mutations in HIV-1_YU2-infected hu-mice.

(A) Flow cytometry staining of HIV-1_YU2-infeceted CD4⁺ T cells with a panel of anti-Env antibodies. The antibodies used are indicated on the X-axis. Y-axis depicts the mean fluorescence intensity (MFI) of Env staining by the indicated antibodies after gating on Gag⁺ cells. Columns are color-coded according to the regions of HIV-1 Env targeted by each antibody, indicated above. Data are represented as mean ± SEM. (B) Clearance of HIV-1_YU2-infected cells in vivo following adoptive transfer to NRG mice in the presence of 246D or a control antibody. Dots indicate the percentage of Gag⁺ cells among CD3⁺CD8⁻ cells 5 hours after transfer of HIV-1_YU2-infected cells to mice treated with 246D (blue dots) or the control (gray dots). Each dot reflects a single animal. Horizontal black line indicates the mean percentage of Gag⁺ cells in each group. Significant differences are indicated, *p <0.05, two-tailed Mann-Whitney U test. (C) Amino acid sequence alignment of gp120 sequences from HIV-1_YU2-infected animals treated with 246D, 246D:GRLR, or no antibody, as indicated above. Each horizontal gray bar represents a single gp120 sequence aligned to YU2. Amino acid substitutions are indicated as black ticks; R456K substitution is indicated by a red tick. Numbers and vertical black lines at right indicate the animal from which the corresponding sequences were obtained. Numbering at the bottom of each sequence alignment corresponds to the amino acid position in YU2 gp120 (HXB2 numbering). (D) Pie charts showing the proportion of all sequences from each corresponding treatment group in (C) that carry the R456K mutation (red) or are not mutated at this position (white). The numbers in the center denote the total number of R456K positive sequences over the total number of sequences obtained for each treatment group. See also Supplementary Figure 2.

Although monoclonal Ab 246D does not neutralize HIV-1_YU2 in TZM-bl assays in vitro, even at concentrations of 50 μg/ml (Figure S2A), it can block macrophage infection in vitro in an Fc receptor-dependent manner (Holl et al., 2004; Holl et al., 2006). When tested in adoptive transfer experiments for its ability to accelerate the clearance of infected cells in vivo, nnAb 246D exhibited modest, but significant activity against HIV-1_YU2-infected cells (Figure 3B). To determine whether this nnAb can alter the course of HIV-1 infection, we first administered 246D or 246D:GRLR control antibodies to hu-mice with established HIV-1_YU2-infection. We found no detectable effects of either antibody on viral load (Figure S2B). However, single genome sequence analysis revealed a recurring mutation in the CD4 binding site of gp120 (Kwong et al., 1998), R456K, in 246D-treated animals, but not in 246D:GRLR-treated or untreated animals (Figures 3C and 3D). Thus, antibody therapy with the nnAb 246D selects for a recurring mutation in HIV-1_YU2 in an Fc-dependent manner.

To determine whether 246D can also mediate selection during infection with primary isolates, we tested it against viruses carrying patient-derived transmitted-founder Envs (HIV-1_9004SS (Gnanakaran et al., 2010) and HIV-1_DU156 (Li et al., 2006), Figure 4). 246D bound to cells infected with HIV-1_9004SS, but not to HIV-1_DU156 (Figure 4A). We therefore tested whether 246D treatment would have an effect on HIV-1_9004SS infection, but not on HIV-1_DU156 infection, in hu-mice. Similar to HIV-1_YU2 infection, 246D had no impact on plasma viral load in mice infected with HIV-1_9004SS or HIV-1_DU156 (Figure 4B). However, single genome sequencing revealed that 246D selected for recurring mutations in HIV-1_9004SS, but not in the control virus, HIV-1_DU156. Similar to 246D treatment in HIV-1_YU2-infection, HIV-1_9004SS infected mice developed recurring mutations in gp120 at residue G366, which is also in the CD4 binding site (Kwong et al., 1998) (Figures 4C and 4D). The observation that a nnAb to gp41 selected for mutations in the CD4bs in gp120 in both HIV-1_YU2 and HIV-1_9004SS suggested a common mode of viral escape, and prompted us to investigate the mechanisms by which such mutations might confer resistance to 246D.

Figure 4. 246D treatment selects CD4bs mutations in a transmitted/founder Env.

(A) Flow cytometry staining with anti-Gag antibody and 246D antibody of CD4⁺ T cells infected with HIV-1_9004SS (left) or HIV-1_DU156. Numbers at top-left indicate the percentage of Gag⁺Env⁺ cells out of all live cells. X-axis depicts anti-Gag antibody staining; Y-axis depicts 246D staining. (B) Plasma viremia in hu-mice infected with HIV-1_9004SS (left) or HIV-1_DU156. X-axis, days post-infection; Y-axis, viral RNA copies/ml. Gray shading indicates the period of 246D treatment. Black filled triangles depict untreated mice; circles depict treated mice. (C) Amino acid sequence alignment of gp120 sequences from HIV-1_9004SS- or HIV-1_DU156-infected hu-mice from (B). The infecting virus for each sequence group is listed at top. Each horizontal gray bar represents a single gp120 sequence aligned to YU2 gp120. Amino acid substitutions are indicated as black ticks, G366R substitution is indicated by a red tick and G366E is indicated by an orange tick. Numbers and vertical black lines at left indicate the animal from which the corresponding sequences were obtained; at right, treatment group is indicated. Numbering at the bottom of each sequence alignment corresponds to the amino acid position in YU2 gp120 (HXB2 numbering). (D) Pie charts show the proportion of all sequences from 246D-treated animals infected with the indicated virus that carry the G366R mutation (red), the G366E mutation (orange), or are not mutated at position 366 (white). The numbers in the center of the pies denote the number of G366R/E sequences over the total number of sequences obtained.

CD4bs mutation broadly alters nnAb and bnAb binding

To determine if the recurring 246D-induced mutation in HIV-1_YU2, R456K, alters virus recognition by the 246D antibody, we mutated R456 to lysine in HIV-1_YU2 (HIV-1_YU2(R456K)). The R456K mutant had reduced binding to 246D and all other anti-gp41 antibodies tested, as well as to nnAbs targeting the V3 crown (Figures 5A, 5B, 5E and 5F). By contrast, binding to bnAbs PG16 and 3BNC117, which target the V1/V2 apex and the CD4 binding site, respectively, was increased for HIV-1_YU2(R456K) compared with HIV-1_YU2 (Figures 5C and 5D). To determine whether the R456K mutation also alters virus sensitivity to neutralization by bNAbs, we compared the neutralizing activity of a panel of antibodies against HIV-1_YU2(R456K) and HIV-1_YU2 in TZM-bl assays (Figure 5G). The R456K mutation increased the neutralizing activity of several bNAbs, most notably antibodies targeting the V1/V2 apex and the glycan patch in the vicinity of the base of the V3 loop, while neutralization by bNAbs that target the gp41-gp120 interface and the gp41 membrane-proximal external region was reduced. Thus, the R456K mutation that arises in vivo as a result of pressure by gp41-specific nnAb 246D appears to produce significant alterations in antibody accessibility to gp41 and gp120 epitopes on the surface of infected cells and viruses.

Figure 5. R456K mutation alters antibody accessibility to Env epitopes.

(A–F) Flow cytometry staining of CD4+ T-cells infected with HIV-1_YU2 or HIV-1_YU2(R456K). Cells were stained with anti-Gag antibody and nnAbs that target gp41 (A, B), bNAbs that target gp120 (C, D), or nnAbs that the gp120 V3 crown (E, F). (A, C, E) X-axes depict Gag staining; Y-axes depict staining with the indicated antibodies. (B, D, F) Ratio of the mean florescent intensity (MFI) between cells infected with HIV-1_YU2 and HIV-1_YU2(R456K). The dashed line denotes a ratio of 1, indicating equal staining for both viruses. Data are represented as mean ± SEM. (G) TZM-bl neutralization activity of bNAbs against HIV-1_YU2 relative to HIV-1_YU2(R456K). The ratio of the antibody concentration (μg/ml) required to achieve 50% (IC50, left) or 80% (IC80, right) inhibition of virus infection with HIV-1_YU2 versus HIV-1_YU2(R456K) is given. X-axis shows antibodies used; y-axis indicates the percent of cells infected on a log-scale. Horizontal line depicts a ratio of 1; values greater than 1 indicate greater neutralizing potency of a given antibody against the R456K mutant virus. Columns are color-coded according to the envelope epitopes targeted by the antibodies, indicated in the key at right. Asterisks indicate antibodies for which an IC80 was not achieved at any antibody concentration tested.

246D Selects for Mutations that Prevent CD4-induced gp120 Shedding

R456 and G366 are residues in the gp120 CD4bs that make direct contacts with CD4 (Kwong et al., 1998), and are far-removed from the 246D nnAb epitope in gp41 (Figure 6A). The backbone carbonyl of R456 forms a hydrogen bond with the K35 residue of CD4, and its guanidino group hydrogen-bonds to the backbone carbonyl and side chain of gp120 CD4bs residues F277 and Q466, respectively (Figure 6B) (Kwong et al., 1998). One explanation for the decreased susceptibility of the HIV-1_YU2 R456K mutant to binding by nnAb 246D is that this mutation may reduce gp120 shedding, and therefore limit the number of gp41 stumps available for binding by 246D.

Figure 6. R456K mutation reduces YU2 gp120 binding to CD4.

(A) Gp120 (light gray) and gp41 (dark gray) from the BG505.SOSIP.664 structure (RCSB: 5TZ3) are shown in ribbons. Gp120 residues R456 and G366 (blue and orange spheres) are distal to the 246 epitope (red ribbons) within gp41. (B) Hxb2 gp120 (gray) is shown in complex with soluble CD4 (sCD4) (green) in ribbons (RCSB: 1GC1). Gp120 residues R456 (navy) and G366 (orange), and CD4 residue K35 (cyan), are shown in stick representation. R456 makes H-bond contacts (red dashed lines) with K35 of CD4 via its backbone, and with gp120 residues F276 and E466 (gray sticks) in the D and V5 loops, respectively, via its side chain. (C) TZM-bl assay neutralization of HIV-1_YU2 (black lines/symbols) and HIV-1_YU2(R456K) (red lines/symbols) viruses by sCD4. X-axis reflects percent of viruses neutralized; y-axis, antibody concentration (μg/ml). (D) Binding of the monomeric phases of YU2.SOSIP.664 and YU2(R456K).SOSIP.664 at the indicated concentrations to immobilized soluble 2-domain CD4 (2dCD4) was assayed by biolayer interferometry. Colored curves give the arithmetic mean signal from three independent experiments at the indicated concentrations of SOSIP protein. See also Supplementary Figure 3.

Decreased shedding could be due to intrinsic stabilization of the trimer by the R456K mutation directly, or, alternatively, the mutation could indirectly reduce shedding by altering the ability of CD4 expressed on the surface of infected cells to interact with gp120 (Moore et al., 1990). To determine whether the R456K mutation directly stabilizes the trimer, we produced YU2 and YU2 (R456K) SOSIP trimers and analyzed them by size exclusion chromatography and negative stain electron microscopy (EM) (Figure S3). We found no difference in the proportion of aggregated trimers and monomers, and an equal amount of disorder in both preparations. Thus, the R456K mutation does not appear to stabilize the YU2 trimer directly.

To determine whether gp120 binding to CD4 is impaired by the R456K mutation, we measured HIV-1_YU2 or HIV-1_YU2(R456K) neutralization by soluble CD4 (sCD4) in TZM-bl assays. HIV-1_YU2(R456K) was far more difficult to neutralize with sCD4 than HIV-1_YU2, indicating that R456K reduces gp120 binding to CD4 (Figure 6C). We then measured CD4 binding to envelope trimers directly using biolayer interferometry (Figure 6D). The R456K mutation increased the dissociation constant (K_D) by ~2-fold, which was the result of a two-fold reduction in the on-rate of CD4. The off-rate was unchanged, suggesting that the mutation may have partially occluded the CD4bs, thereby limiting CD4 accessibility. This effect may be due to destabilization of the V5- and D-loop regions within the CD4bs of Env, and/or to a loss of a potential H-bond partner with K35 of CD4 (Figure 6B). We speculate that mutation of the glycine at position 366 to arginine or glutamate in HIV-1_9004SS likely confers resistance to 246D by a related mechanism, as the large side chain substitutions would occlude the CD4-binding pocket to limit CD4 accessibility in a manner similar to R456K for HIV-1_YU2. We conclude that the R456K mutation indirectly enhances the integrity of Env trimers on infected cells and virions by reducing gp120:CD4 interactions that cause gp120 shedding.

Increased Trimer Integrity by Decreased Interaction with CD4 on Infected Cells

To measure the effects of R456K on trimer integrity directly, we assayed the amount of shed gp120 bound to uninfected cells in HIV-1-infected cultures (Richard et al., 2016). CD4⁺ T-cells were infected with HIV-1_YU2 or HIV-1_YU2(R456K) and stained with the nnAb A32,(Wyatt et al., 1995) which binds to gp120 that has been shed from infected cells and is bound to cell surface CD4 (Bruel et al., 2017; Richard et al., 2016), and the V3 crown targeting antibody 10–188 (Mouquet et al., 2011). As expected, uninfected bystander p24⁻CD4⁺ T cells in cultures infected with HIV-1_YU2 showed high levels of A32 and 10–188 staining. In contrast, staining of p24⁻CD4⁺ bystander T cells was markedly lower in cultures infected with HIV-1_YU2(R456K) (Figure 7A). To confirm this result biochemically, we produced recombinant versions of HIV-1_YU2 and HIV-1_YU2(R456K) bearing a surface-expressed HA-tag (HIV-1_YU2HA and HIV-1_YU2(R456K)HA, respectively) to facilitate purification of intact infected cells and virions based on HA expression. We then purified infected and uninfected bystander cells from HIV-1_YU2HA- and HIV-1_YU2(R456K)HA-infected cultures by sorting HA⁺CD4^low or HA⁻CD4^high cells, respectively, and performed Western blotting for gp120 (Figures 7B, 7C, and S4). In accordance with cell staining experiments (Figure 7A), we observed more gp120 associated with uninfected bystander cells cultured in the presence of HIV-1_YU2HA relative to HIV-1_YU2(R456K)HA (Figure 7C), and we observed slightly higher levels of gp120 on HIV-1_YU2(R456K)HA infected cells (Figure S4). Thus, the R456K mutation decreases gp120 shedding by infected cells.

Figure 7. R456K decreases CD4-induced gp120 shedding to mask nnAb targets.

(A) Flow cytometry staining of CD4+ T-cells infected with HIV-1_YU2 or HIV-1_YU2(R456K). X-axis depicts anti-Gag staining; Y-axis depicts staining with the anti-C1 nnAb A32 (top panels) or the anti-V3 crown nnAb 10–188 (bottom panels). The infecting virus is shown at top. (B) Sorting of CD4⁺ T cells from cultures infected with HIV-1_YU2HA (top) or HIV-1_YU2(R456K)HA. Cells were stained with anti-CD4 (Y-axis) and anti-HA (X-axis) and uninfected cells (blue squares) were sorted for Western blot analysis. (C) The sorted cells from (B) were blotted with anti-gp120 antibody, anti-gp41 antibody, and anti-GAPDH antibody. The blotting antibody is indicated at right of each gel image. The black frames depict that the irrelevant parts of the blots were removed from the figure. The blots’ background was adjusted for better clarity. The gp120/gp41 band intensity ratio (arbitrary units) is shown in the column graph at far-right. (D) Flow cytometry staining with A32 and 246D antibody of CD4+ T-cells infected with HIV-1_YU2 or HIV-1_YU2(R456K). Cells were mock-treated (Gray, filled histogram) or treated with ibalizumab (black, empty histogram) before cell staining. The antibody used for staining is indicated on the X-axis; the infecting virus is indicated above. (E) Mean florescent intensity (MFI) of the histograms shown in (D). Black columns depict mock-treated cells; gray columns depict ibalizumab-treated cells. The antibody used for staining and the virus used for infection are indicated on the X-axis. Data are represented as mean ± SEM. *p <0.05, **p <0.005, one-tailed t-test. See also Supplementary Figures 4 and 5.

To confirm the requirement for CD4 on the effects of R456K on gp120 shedding, we co-transfected HEK 293T cells (ATCC) with a CD4 expression vector and either HIV-1_YU2 or HIV-1_YU2(R456K) (Figure S5). Uninfected bystander cell staining by the anti-V3 antibody 10–188 was detectable mainly on CD4-expressing 293T cells, and was markedly reduced in CD4-negative cells (Figure S5). Finally, to confirm the role of Env-CD4 interactions on the gp41 and V3 antibody evasion seen in Figure 5, we measured A32 and 246D binding to CD4⁺ T cells infected with HIV-1_YU2 or with HIV-1_YU2(R456K) in the presence or absence of ibalizumab (Song et al., 2010), which binds to CD4 and interferes with trimer-receptor interactions. Ibalizumab reduced A32 and 246D binding to cells infected with either virus (Figures 7D and 7E), reinforcing the hypothesis that limiting CD4:gp120 interactions enhances trimer integrity and reduces accessibility of 246D to its epitope in gp41.

We conclude that the gp41-specific nnAb 246D selects for CD4bs mutants like R456K, which produce significant alterations in gp120 epitope accessibility to bNAbs on the surface of infected cells and viruses, and which decrease CD4-induced gp120 shedding to limit the accessibility of nnAbs including 246D.

DISCUSSION

We have examined the role of nnAbs in HIV-1 infection in vivo using hu-mice. To do so, we produced an HA indicator virus that is replication competent in vivo, and confirmed the results with HIV-1 viruses bearing tier 2 Envs. The data indicate that nnAbs can alter the course of HIV-1 infection, apply immune pressure on the infecting virus, and even achieve modest levels of protection by a mechanism that is entirely Fc receptor dependent.

The role of nnAbs in HIV-1 infection in vivo has been the subject of extensive debate. In vitro studies with laboratory-adapted, tier 1 HIV-1 strains showed that Env-specific nnAbs can direct ADCC/ADCVI (reviewed in (Ferrari et al., 2017)). However, little evidence exists to demonstrate Fc-mediated nnAb activity against primary HIV-1 isolates and transmitted/founder viruses (Bruel et al., 2017). Passive transfer experiments in macaques have associated nnAbs with reduced numbers of transmitted/founder variants and, in some cases, with lower acute viremia, but not with protection from infection (Burton et al., 2011; Mascola et al., 1999; Moog et al., 2014; Santra et al., 2015). In these experiments, there was no evidence of nnAb-specific viral escape mutations, and the role of Fc receptors in nnAb activity in vivo was not evaluated. To specifically interrogate how nnAbs might affect HIV-1 infection in vivo, we screened for nnAbs that bind to cells infected with tier 2 viruses. Despite binding to surface-expressed Env on infected cells and virions, the selected antibodies do not neutralize HIV-1 in TZM-bl assays in vitro. Nevertheless, we show that such nnAbs mediate infected cell clearance and alter the course of HIV-1 infection in hu-mice.

These antibodies do so in an Fc receptor-dependent manner by selecting for viral escape mutations that confer resistance to nnAb binding. We also show that nnAbs have activity in hu-mice infected with a tier 2 reference strain, HIV-1_YU2, and with a transmitted/founder Env, HIV-1_9004SS, demonstrating that such nnAbs are indeed active against clinically relevant viruses. Although anti-HA lowered viremia in HIVivoHA-infected cells, we did not observe these effects in HIV-1_{YU2 or} HIV-1_9004SS infected animals treated with the nnAb 246D. Since all antibody concentrations were near saturating (Horwitz et al., 2013; Klein et al., 2012), this is likely due to the low levels of Env expressed on HIV-1_{YU2 or} HIV-1_9004SS infected cells relative to HA expression on HIVivoHA-infected cells.

Antibody-specific escape mutations in HIV-1 typically occur within the epitope targeted by the antibody (Andrus et al., 1998; Diskin et al., 2011; Horwitz et al., 2013; Klein et al., 2012; West et al., 2013), although allosteric mutations may also alter antibody binding in some cases (Bradley et al., 2016; Wei et al., 2003). Accordingly, hu-mice infected with HIVivoHA and treated with anti-HA IgG developed viral escape mutations that inactivated the HA cassette. In contrast, the anti-gp41 cluster I nnAb 246D failed to select for mutations in its target epitope within gp41. Unexpectedly, we found strong selection for mutations in gp120 at residues that make contacts with CD4, most notably R456K for HIV-1_YU2 and G366E/R for HIV-1_9004SS. We show that the R456K mutation in HIV-1_YU2 confers resistance to nnAb 246D by preventing CD4-induced gp120 shedding and trimer decay, thereby concealing the 246D epitope on gp41 stumps. Similar indirect effects on shedding are mediated by the virus-encoded proteins Nef and Vpu. These two proteins decrease virus susceptibility to nnAbs in vitro by reducing surface levels of CD4 and/or antagonizing tetherin, which in turn decreases gp120 shedding from infected cells (Alvarez et al., 2014; Arias et al., 2014; Pham et al., 2016; Pham et al., 2014). We conclude that nnAbs elicit specific viral escape mutations that may be distal to the target epitope, yet nonetheless confer resistance.

Post-hoc analyses of the only HIV-1 vaccine trial to show modest efficacy, RV144 (Rerks-Ngarm et al., 2009), associated protection with nnAbs directed against the V1/V2 loops of gp120 (Haynes et al., 2012; Zolla-Pazner et al., 2013) and with ADCC responses (Bonsignori et al., 2012). Protection from HIV-1 infection was most evident at early time points after vaccination, when antibody levels were highest (Robb et al., 2012). However, the question of whether nnAbs were indeed responsible for the observed protection remains unanswered. Our experiments using a recombinant HIV-1 virus expressing high levels of a heterologous surface marker (HA-tag) show that passive infusion of a nnAb can protect against infection in an exclusively Fc receptor-dependent manner. While these experiments made use of a recombinant virus, high doses of a monoclonal nnAb, and a highly expressed heterologous surface marker, Fc-mediated protection was achieved against an extremely high-titer inoculum that consistently infects 100% of animals (Figure S1G). Our data are consistent with the hypothesis that high titer polyclonal anti-Env nnAb responses could be sufficient to afford low levels of protection against HIV-1 infection at the low multiplicity infectious doses characteristic of human transmission.

Fc-mediated immune responses contribute to protection and therapy by bNAbs in humice and macaques (Bournazos et al., 2014; Halper-Stromberg et al., 2014; Hessell et al., 2007), and correlate with slower disease progression in humans (Baum et al., 1996; Lambotte et al., 2009). In addition, Fc receptor engagement accelerates the clearance of infected cells in humans treated with the bNAb 3BNC117 (Lu et al., 2016). However, whereas potent bNAbs can neutralize virions directly, nnAbs are entirely reliant upon Fc receptor engagement to protect against or alter the course of HIV-1 infection. Our results show that nnAbs can clear HIV-1-infected cells in vivo, and it is likely that they also accelerate the clearance of virions (Igarashi et al., 1999). Thus, the high levels of nnAbs found shortly after vaccination in the RV144 trial may have been protective against HIV-1, despite their inability to neutralize the virus directly, by accelerating virus and/or infected cell clearance by Fc mediated mechanisms.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Michel C. Nussenzweig (nussen@rockefeller.edu)

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Human Subjects

Samples of peripheral blood were obtained as de-identified samples from the New York Blood Center (New York, NY) and were used for isolation of human peripheral blood mononuclear cells (PBMC) for virus infection experiments. Samples of human fetal liver tissue were obtained from Advanced Biosciences Resources (Alameda, CA) and were used to generate humanized mice as described below. All human tissue samples were obtained in accordance with protocols approved by the institutional review board of The Rockefeller University.

Humanized Mice

NOD Rag^1−/− IL2Rγ^null (NOD.Cg-Rag1^tm1Mom Il2rg^tm1Wjl/SzJ, NRG) mice were obtained from The Jackson Laboratory and bred and maintained in the AAALAC-certified facility of the Rockefeller University. Mice were maintained under a 12 hr light/dark cycle with amoxicillin-containing chow diet under pathogen-minimizing care conditions. Both male and female mice (8–20 week old) were used for all experiments and were equally distributed within experimental and control groups. Animal protocols were in agreement with NIH guidelines and approved by the Rockefeller University Institutional Animal Care and Use Committee. Humanized mice were generated as previously described (Klein et al., 2012b). Human fetal livers were homogenized and hematopoietic stem cells (HSCs) were isolated from digested liver using a CD34⁺ HSC isolation kit (Stem Cell Technologies). NOD Rag^1−/− IL2Rγ^null (NOD.Cg-Rag1^tm1Mom Il2rg^tm1Wjl/SzJ, NRG) mice, 1–5 days old, were irradiated with 100 cG and were injected intrahepatically with 2×10⁵ human CD34⁺ HSCs. Injection of HSCs was done 6 hours after irradiation. After eight weeks of age, male and female hu-mice were screened for successful engraftment of human immune cells and enrolled in experiments.

Cell lines

Human embryonic kidney HEK 293T cells were obtained from ATCC (Manassas, VA) and cultured in 5% C0₂ at 37°C in DMEM supplemented with 10% FBS. CEM.NKR CCR5+ cells were obtained from the NIH AIDS Reagent Program and cultured in 5% C0₂ at 37°C in RPMI supplemented with 10% FBS.

Viruses

Virus stocks of replication-competent HIV-1 strains were prepared from transient transfection of plasmid DNA into 239T cells with linear polyethyleneimine (PEI). All virus preparations were handled in a Biosafety Level 2 (BSL-2) containment laboratory in accordance with proper BSL-2 safety procedures. Virus infections of hu-mice were carried out in a BSL-2 certified animal housing unit within the Comparative Bioscience Center at The Rockefeller University.

METHOD DETAILS

Hu-mice experiments

Male and female NRG mice, 5–12 weeks old, were infected by intraperitoneal injection of HIV-1_YU2 (510 ng p24). For 246D treatment, experimental mice were intraperitoneally injected with 2mg of 246D or 246D:GRLR starting at 10 days post infection (DPI) followed by three injections with 1 mg at 14, 17 and 21 DPI. Viral load in mouse plasma was measured 10, 14, 17, 21 and 24 DPI. For single genome sequencing of viruses from infected mice, 600 μl of blood was collected from each mouse by cardiac puncture and all mice were sacrificed at 24 DPI. For anti-HA IgG1-treatment experiment mice were infected with HIVivo-HA (510 ng p24) and intraperitoneally injected with 2mg of anti HA-IgG1starting at 15 DPI followed by three injections with 1 mg at 17,19 and 23DPI. Viral load in mice plasma was measured 14, 18, 21, 25 and 28 DPI. For single genome sequencing of viruses from infected mice, 600 μl blood was collected from each mouse by cardiac puncture and all mice were sacrificed at 28DPI. For 246D prophylaxis experiment, 1mg of 246D was intraperitoneally injected one day before and one day after infection with HIV-1_YU2 (17 ng p24). Viral load in mice plasma was measured 5, 9, 12, 16, 19 and 22 days post infection (DPI). For anti-HA-IgG1 prophylaxis experiment, 1mg of HA-IgG1 was intraperitoneally injected one day before and one day after infection with 1mg of HA-IgG1 was intraperitoneally injected one day before and one day after infection with HIVivo-HA (510 ng p24). Viral load in mice plasma was measured 3, 6, 10, 13, 17, 20, 24, 37 and 50 days post infection (DPI). For single genome sequencing of viruses from infected mice, 600 μl blood was collected from each mouse by cardiac puncture and all mice were sacrificed at 37 DPI.

Plasma Viral Load Measurements

300–500 μl of whole blood was collected by facial vein bleed. Plasma was separated by centrifugation from whole blood samples at 2000 x g centrifugation for 5 min. Total RNA was extracted from 100 ml plasma using QIAmp MinElute Virus Spin Kit (QIAGEN) in combination with RNase-free Dnase (QIAGEN) and eluted in a 55 μl AVE elution buffer (QIAGEN). HIV-1 RNA was quantified by qRT-PCR as previously described (Klein et al., 2012). The following forward and reverse primer were used for qRT-PCR: 5′ TAATGGCAGCAATTTCACCA-3′ (HXB2 nt 4577–4596) and 5′ GAATGCCA AATTCCTGCTTGA 3′ (HXB2 nt 4633 to 4653), respectively. The internal probe had the sequence 5′-/5HEX/CCCACCAAC/ZEN/ARGCRGCCTTAACTG/3IABkFQ/-3′. The quantitation limit was determined to be 800 copies/ml (Klein et al., 2012).

In vivo adoptive cell transfer

In vivo adoptive cell transfers were done as previously described (Lu et al., 2016). In brief, each mouse was injected intrasplenically with cultured cells containing 0.5 million Gag+ cells in a total volume of 50 μl. Mice were pre-treated with 1 mg of 246D or 1 mg of anti-HA-IgG1 12 hours before cell transfer. Mice were sacrificed five hours after cell transfer and splenocytes were collected and filtered throw 40-μm mesh into complete RPMI media (Gibco) with 6% serum. Enriched human cells were analyzed using the following antibodies: LIVE/DEAD® Fixable Aqua Dead Cell Stain Kit, mouse CD45 (30-F11, BD), human CD3 (UCHT1, BD), human CD4 (OKT4, ebioscience), human CD8 (RPA-T8, BioLegend), and Gag (KC57, Coulter).

Virus cloning

HIVivoHA is an in vivo-adapted reporter virus derived from HIV-1_YU2 that was developed as described in Figure S1. Briefly, a cassette containing the minimal human IL-16 promoter driving expression of mCD24/heat-stable antigen (HSA) bearing an HA-tag in place of the native HSA ectodomain (Ali and Yang, 2006) was inserted between env and nef in HIV-1_YU2. To improve replicative fitness of the recombinant, a portion of gp41 from the HIV-1 strain Q23.17 was grafted into the YU-2.DG env, yielding a reporter virus with identical replicative fitness to HIV-1_YU2 in vitro (HIV-1_YU2/Q23IL16HA, Figure S1). Further replicative adaptation through selective passaging in hu-mice was required to generate a stable recombinant that achieved sufficiently high viral loads for experimentation. Virus sequencing revealed ten nucleotide mutations throughout the entire genome, which were introduced into the initiating recombinant to generate the in vivo-adapted infectious molecular clone, HIVivoHA. For biochemistry experiments, the viruses HIV-1_YU2HA and HIV-1_YU2(R456K)HA were generated simply by insertion of the IL16HA cassette between env and nef in HIV-1_YU2. HIV-1_9004SS and HIV-1_DU156 were generated by cloning the 9004SS env gene (kindly provided by Michael Seaman (BIDMC; Boston, MA USA)) and the Du156 env gene (Du156, clone 12 (SVPC3), NIH AIDS Reagent program) into the HIV-1_NL4/3 vector backbone and produced by transfection in 293T cells, as previously described (Klein et al., 2012). HIV-1 virus preparations were quantified by p24 ELISA (Lenti-X p24 Rapid Titer Kit, Clontech) according to the manufacturer’s recommendations.

Expression and purification of 2dCD4

Soluble 2dCD4, with a noncleavable 6x histidine tag, was expressed using transient transfection of HEK 293T cells using linear polyethylenimine (PEI) following the manufacturer’s suggested protocol. After 5 d of expression, supernatants were clarified by centrifugation and loaded onto Ni-NTA superflow resin (Qiagen) preequilibrated with Buffer A (10mM Tris, pH 7.5, 100 mM NaCl), washed with Buffer A + 10 mM imidazole, and eluted with Buffer A + 350 mM imidazole. 2dCD4 was then purified by gel filtration chromatography in Buffer A using a superdex 200 analytical column (GE Healthcare).

Purification of Env SOSIP proteins

The YU2 SOSIP.664 constructs were transfected along with a plasmid encoding the cellular protease furin at a 3:1 Env:furin ratio in 293F cells. The cells were allowed to express soluble SOSIP trimers for 5–7 days. Culture supernatants were collected and cells were removed by centrifugation at 3800 x g for 20 min, and filtered with a 0.2 μm pore size filter. SOSIP proteins were purified by flowing the supernatant over a lectin (Galanthus nivalis) affinity chromatography column overnight at 4°C. The lectin column was washed with 1x PBS, followed by a second wash with 1x PBS and 0.5M NaCl and proteins were eluted with 0.5 M methyl-α-D-mannopyranoside and 0.5 M NaCl. The eluate was concentrated and loaded onto a Superdex 200 10/300 GL column (GE Life Sciences) prequilibrated in a buffer of 10 mM Hepes, pH 8.0, 150 mM NaCl and 0.02% sodium azide, to separate the trimer-size oligomers from aggregates and gp140 monomers.

Electron Microscopy

A 3 μL aliquot containing 0.01 mg/ml of YU2.SOSIP.664 or YU2(R456K).SOSIP.664 protein was applied for 15 s onto a carbon coated 400 Cu mesh grid that had been glow discharged at 20 mA for 30 s, followed by negative staining with 0.7% uranyl formate for 30 s. Samples were imaged using a Philips CM10 microscope operating at 80kV, at a magnification of 59,000x. Images were acquired with a Gatan 2K CCD camera using a nominal defocus of 1500 nm.

Biolayer Interferometry

Kinetic measurements of soluble 2dCD4 binding to the YU2 SOSIP.664 constructs were carried out using the Octet QK^e system (ForteBio); 0.2 mg/mL of each His-tagged 2dCD4 was immobilized onto a Nickel-NTA biosensor until it reached saturation. The SOSIP.664 trimers were tested at concentrations of 0.1 μM to 1 μM. A reference sample of buffer alone was used to account for any signal drift that was observed during the experiment. Association and dissociation were each monitored for 5 min. All experiments were conducted in the Octet instrument with agitation at 1,000 rpm. Analyses were performed using nonlinear regression curve fitting using the Graphpad Prism software, version 7.

CD4⁺ T cell isolation and HIV-1 infection

CD4⁺ T cells were isolated from peripheral blood mononuclear cells according to manufacturer’s protocol (Miltenyi Biotec). The percentage of CD4⁺ T cells was determined by staining the isolated cells with anti-human CD3 (BioLegend #300417), CD45 (Invitrogene # MHCD4539), CD4 (eBioscience # 12-0049-42), HLA-DR (ebioscience #25-9952-42), CD25 (eBioscience #302612) and CD69 (BioLegend #310912). Purified CD4⁺ T lymphocytes were cultured with of RPMI 1640 (Gibco) supplemented with 10% (vol/vol) FBS (HyClone; Thermo Scientific), 1% penicillin/streptomycin (Gibco), 2mM L-glutamine, 1 μg/mL phytohemagglutinin (Life Technologies), and 100 U/mL IL-2 (Peprotech) at 37 °C and 5% CO2. For HIV-1 infection CD4⁺ T cells were cultured at 0.3×10⁶ cells per well in a 24-well plate. Cells were centrifuge at 400g at 4 °C. Cells media was removed and cells were resuspended with HIV-1_YU2, HIVivo-HA, HIV-1_YU2HA or HIV-1_YU2(R456K)HA (85ng p24) diluted in RPMI 1640 (Gibco) supplemented with 10% (vol/vol) FBS (HyClone; Thermo Scientific), 1% penicillin/streptomycin (Gibco), 2mM L-glutamine, 1 μg/mL phytohemagglutinin (Life Technologies), 100 U/mL IL-2 (Peprotech) and 5 μg/mL polybrene in a total volume of 200 μl per well. CD4⁺ T cells were then incubated with the virus for 48–72 hours in 37 °C at 37 °C and 5% CO2.

Single genome sequencing of HIV-1 env

HIV-1-RNA was extracted from 200 μl mice plasma using the Qiagen MinElute Virus Spin Kit according to manufacturer’s instructions. cDNA was synthesis from isolated RNA using SuperScript III Reverse Transcriptase and a poly-T primer. To remove RNA remains, RNA was digested using RNAseH for 20 minutes at 37 °C. cDNA was diluted and was subjected to two-rounds of nested PCR with gp160-specific primers as previously described (Salazar-Gonzalez et al., 2008) using the following primers for the first round PCR 5′ TTTTTTTTTTTTTTTTTTTTTTTTRAAGCAC 3′ and 5′ GGCTTAGGCATCTCCTATGGCAGGAAGAA3′. For the second-round PCR the following primers were used: 5′ TAGAAAGAGCAGAAGACAGTGGCAATGA 3′ and 5′ TCATCAATGGTGGTGATGATGATGTTTTTCTCTCTGCACCACTCTTCT 3′. PCRs products with amplification efficiencies lower than 30% were used for library preparation by the Illumina Nextera DNA Sample Preparation Kit as Previously described (Caskey et al., 2015). Sequencing was performed using the Illumina Miseq Nano 300 cycle kits at a concentration of 15 pM. Env alignments were generated using Geneious 8.1 (Biomatters).

Flow Cytometry

Infected-CD4⁺ T cells were harvested 48–72 hours post infection and were incubated with the indicated anti-Env antibodies (3–20 μg/ml) for 1.5 hours at 37 °C. The cells were then washed with FACS medium (1% BSA vs 5% FBS) and incubated at 4 °C for 30 min with anti-human CD3 (BioLegend #300417), CD45 (Invitrogen # MHCD4539), CD4 (eBioscience # 12-0049-42), HLA-DR (eBioscience #25-9952-42), CD25 (eBioscience #302612), CD69 (BioLegend #310912), CD8a (eBioscience #47-0087-42), Viability Dye (eBioscience #65-08651-14) and anti-human IgG1 (Jackson ImmunoResearch #109-136-088). Cells infected with HIVivo-HA, HIV-1_YU2HA or HIV-1_YU2(R456K)HA were also stained with anti-HA antibody (Miltenyi # 130-098-404). To detect intracellular p24, cells were washed twice and were resuspended in 200 μL of BD Cytofix/Cytoperm solution per well for 20 minutes at 4°C. To detect p24 expression fixed cells were washed twice with BD Perm/Wash and incubated with anti-p24 antibody (Coulter, #6604665) for 30 min at 37 °C. Cells were washed twice with BD Perm/Wash and analyzed with a LSRFortessa (BD) and FlowJo software (Tree Star). For CD4 blocking 2.5 μg/ml Ibalizumab was added to the media of 0.3×10⁶infected CD4 ⁺ cells cultured in 96 well plate 24, 36 and 48 hours post infection.

Western Blotting

For detection of gp120 on bystander cells, CD4⁺ T cells infected with HIV-1_YU2HA or HIV-1_YU2(R456K)HA were sorted by FACSorting on a BD Aria by staining with anti-human CD3 (BioLegend #300417), CD45 (Invitrogene # MHCD4539), CD4 (eBioscience # 12-0049-42), HLA-DR (eBioscience #25-9952-42), CD25 (eBioscience #302612), CD69 (BioLegend #310912), CD8a (eBioscience #47-0087-42), Viability Dye (eBioscience #65-08651-14) and anti-HA antibody (Miltenyi # 130-098-404). Uninfected cells were sorted by and gating on CD4 high HA negative cell population and infected cells were sorted by gating on CD4 medium HA high cell population. Cells were sorted into 96 well plate, centrifuged at 400g and were lysed with RIPA buffer (50 μL/500,000 cells). Cell lysate were sonicated, centrifuged at 12,000 g and supernatant was collected to a fresh tube with LDS sample buffer and sample reducing agent (Invitrogen). Samples were then incubated at 90 °C for 10 minutes and 40 μL of uninfected cell lysate or 60 μL infected cells lysate were loaded into 4–12% gel (Invitrogen) in reducing conditions and blotted overnight with anti-GAPDH antibody (Abcam #ab9482), anti gp41 antibody (NIH #1475) and anti gp120 antibody (Abcam #ab21179). Infected cells were also stain with anti p24 antibody (Abcam #ab9071). Membrane was then stained with Streptavidin-HRP(Abcam). For band detection Chemiluminescent HRP was added to the membrane (Denville Scientific). ImageQuant LAS 4000 and ImageJ were used to analyze band intensity.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical analyses were performed using GraphPad Prism version 6.0 or 7.0. Statistical tests used are stated in the figure legends where they appear, including the exact values of n and what n signifies for each test. The two-tailed Mann-Whitney U test was used to determine statistical significance in the reduction of gag+ cells between treatment conditions in adoptive transfer experiments (Figures 1C and 3B). A one-tailed t-test was used to determine the confidence in reduced levels of staining of Env on infected PBMC in the presence of ibalizumab (Figure 7E).

Highlights.

• Non-neutralizing antibodies (nnAbs) clear HIV-1-infected cells in vivo

• Antibody Fc-mediated immune responses can block HIV-1 infection

• nnAbs select for resistant HIV-1 viruses in vivo in an Fc-dependent manner

• Mutations in the gp120 CD4bs modulate trimer integrity and nnAb sensitivity