Safety and antiviral effect of a triple combination of HIV-1 broadly neutralizing antibodies: a phase 1/2a trial

Abstract

Human immunodeficiency virus type 1 (HIV-1)-specific broadly neutralizing monoclonal antibodies (bNAbs) have to date shown transient viral suppression when administered as monotherapy or as a cocktail of two antibodies^1,2,3,4. A combination of three bNAbs provides improved neutralization coverage of global viruses, which may more potently suppress viral escape and rebound^5,6,7. Here we performed an open-label, two-part study evaluating a single intravenous dose of HIV-1 bNAbs, PGT121, PGDM1400 and VRC07-523LS, in six adults without HIV in part 1 and a multicenter trial of up to six monthly infusions of these three bNAbs in 12 people living with HIV with an antiretroviral therapy (ART) interruption in part 2. The primary endpoints were safety, tolerability and pharmacokinetics, and the secondary endpoints in part 2 were antiviral activity following ART discontinuation, changes in CD4⁺ T cell counts and development of HIV-1 sequence mutations associated with bNAb resistance. The trial met its prespecified endpoints. The bNAb treatment was generally safe and well tolerated. In part 2, 83% of participants (10 of 12) maintained virologic suppression for the duration of antibody therapy for at least 28 weeks, and 42% of participants (5 of 12) showed virologic suppression for at least 38–44 weeks, despite the decline of serum bNAb concentrations to low or undetectable levels. In exploratory analyses, early viral rebound in two individuals correlated with baseline resistance to PGT121 and PGDM1400, whereas long-term virologic control in five individuals correlated with reduced immune activation, T cell exhaustion and proinflammatory signaling following bNAb therapy. Our data show the potential of a triple bNAb cocktail to suppress HIV-1 in the absence of ART. ClinicalTrials.gov registration: NCT03721510.

Main

Broadly neutralizing monoclonal antibodies (bNAbs) against human immunodeficiency virus type 1 (HIV-1) have been shown to reduce plasma viremia transiently in people living with HIV (PLWH) who are not on antiretroviral therapy (ART)^{1,2,5,8,9,10,11}. The combination of the CD4 binding site (CD4bs) antibody 3BNC117 and the V3-glycan antibody 10-1074 also provided partial maintenance of virological control in PLWH following ART discontinuation^3,4,12,13. However, the selection of resistant virus during bNAb therapy presents a challenge for bNAb-based therapeutic strategies and suggests that a two-bNAb cocktail may be insufficient for sustained virologic control^14,15,16. In this study, we evaluated the triple-bNAb combination involving the V3-glycan antibody PGT121, the V2-apex antibody PGDM1400 and the CD4bs antibody VRC07-523LS. This cocktail neutralizes in vitro 99% of global viruses, with 82% neutralized by at least 2 active antibodies with 80% inhibitory concentration (IC₈₀) < 10 µg ml⁻¹ and 65% neutralized by at least 2 active antibodies with IC₈₀ < 1 µg ml⁻¹ (ref. ⁷). We conducted a phase 1/2a clinical trial to evaluate the safety, pharmacokinetics and antiviral activity of up to six monthly infusions of PGT121, PGDM1400 and VRC07-523LS in PLWH.

Results

Study design and safety

Part 1 of this phase 1/2a study was a single-center, open-label study in Boston, MA, USA, to evaluate the safety and pharmacokinetics of a single intravenous (IV) dose of PGT121 and VRC07-523LS (group 1A: 30 mg kg⁻¹ per antibody; n = 3, 2 females (f) and 1 male (m)) or the combination of PGT121, PGDM1400 and VRC07-523LS (group 1B: 20 mg kg⁻¹ per antibody; n = 3, 2 f and 1 m) in adults without HIV-1 (Figs. 1 and 2a, and Supplementary Table 1). Part 2 of this phase 1/2a study was a multicenter, open-label trial of three monthly IV infusions of 20 mg kg⁻¹ each of PGT121, PGDM1400 and VRC07-523LS (group 2; n = 13, 3 f and 10 m) in PLWH with the option of three additional monthly IV infusions of this triple-antibody cocktail, as long as participants remained virologically suppressed, for six infusions in total (Figs. 1 and 2a, and Supplementary Table 1). Of the 13 volunteers enrolled into group 2, 1 volunteer received only 2 of the 3 bNAbs on day 0 owing to a preparation issue and was therefore excluded from further bNAb administrations and excluded from all analyses. All remaining participants in group 2 discontinued ART 2 days after the first antibody infusion. All participants were PLWH in the United States, with HIV-1 subtype B. Thirty-four participants were screened, and 15 were found to be ineligible or excluded for other reasons (Fig. 1). The first participant was enrolled on 15 January 2019, and the last participant was enrolled on 2 May 2022.

Fig. 1: CONSORT diagram.

Participant recruitment, randomization and follow-up are depicted. Participants were enrolled at Beth Israel Deaconess Medical Center (BIDMC), Orlando Immunology Center (OIC) and Houston AIDS Research Team (HART), McGovern Medical School at The University of Texas Health Science Center). ^aNote that one volunteer from group 2 who received a partial infusion (one IP of the assigned three-IP combination was not given) at day 0 and no subsequent infusions had limited safety follow-up in the study. This volunteer will be left out of the safety data displays.

Fig. 2: Study design, pharmacokinetics and therapeutic efficacy of PGT121, PGDM1400 and VRC07-523LS.

a, Study design. Groups 1A and 1B included participants without HIV-1 that received a single IV dose of either PGT121 and VRC07-523LS (green triangle) or PGT121, PGDM1400 and VRC07-523LS (blue triangle). Group 2 PLWH that were virologically suppressed on ART received IV infusions of PGT121, PGDM1400 and VRC07-523LS monthly for 3 months (blue triangles), with the option to continue monthly infusions for another 3 months (light blue triangles) in the presence of ongoing virological suppression. ART was stopped on day 2 after the initial bNAb infusion on day 0 and was reinitiated if participants met predefined restart criteria or at the end of the study on day 308. D, day(s). b, Levels of VRC07-523LS, PGDM1400 and PGT121 in serum of participants in group 2 that received all six infusions of the three bNAbs (n = 8 participants), as determined by anti-idiotype-specific quantitative ELISAs (PGT121, PGDM1400) and electrochemiluminescence immunoassay (VRC07-523LS). Data are mean ± s.d. Mean t_1/2 of each bNAb is indicated in days. c, HIV-1 RNA levels. Plasma HIV-1 RNA levels (log₁₀(RNA copies per ml)) over the course of the study are shown for each of the 12 group 2 participants (color coded). The vertical dotted lines indicate the timepoints when the three bNAbs PGDM1400, PGT121 and VRC07-523LS were administered. ART was stopped 2 days after the first triple-bNAb infusion (as indicated by the gray box). One participant, 98063, was lost to follow-up. The horizontal dotted line indicates the lower limit of quantification for HIV-1 RNA levels. All participants received six infusions except for 65021 who received two and 65015, 98056 and 98059 who received three infusions.

All three antibodies were generally safe and well tolerated (Supplementary Tables 2–5). Four volunteers reported a serious adverse event (SAE) and/or a grade 3 or higher adverse event (AE) or reactogenicity that was determined to be unrelated to the bNAbs by the site investigators. CD4⁺ T cell counts did not substantially change after bNAb infusions in PLWH (Extended Data Fig. 1a,b).

Pharmacokinetics

Serum levels of PGT121, PGDM1400 and VRC07-523LS were assessed (Fig. 2b and Extended Data Fig. 1c,d). The median PGT121 and VRC07-523LS elimination half-lives when administered as a dual-bNAb combination in participants without HIV-1 (group 1A) were 20.4 days and 37.7 days, respectively, and when administered as a triple-bNAb combination in participants without HIV-1 (group 1B) half-lives were 21.1, 23.2 and 38.7 for PGT121, PGDM1400 and VRC07-523LS, respectively (Extended Data Fig. 1c,d). In PLWH (group 2) who received 6 monthly infusions with the triple-bNAb cocktail, the median elimination half-lives for these antibodies were 19.9, 23.9 and 44.9 days, respectively. These half-lives were longer than those previously reported in viremic PLWH⁵. Pharmacokinetic profiles of PGT121, PGDM1400 and VRC07-523LS were not altered by multiple administrations over a period of 20 weeks. No anti-drug antibodies were detected in these individuals (data not shown).

Virologic control

Of 12 participants in group 2, 10 (83%) maintained virologic suppression (<200 copies per ml) for the duration of the bNAb dosing period for at least 28 weeks after ART discontinuation (Fig. 2c). Of note, participants had not been screened for bNAb sensitivity at baseline. Plasma ART drug levels (tenofovir and emtricitabine were chosen for testing because either drug was present in nearly all available combination ART regimens at that time) were below the limit of quantitation, confirming no surreptitious use of ART in individuals after ART discontinuation (data not shown).

Two participants showed early viral rebound at week 6 (65021) and week 10 (98059). Of the 10 participants who showed virologic suppression for the duration of bNAb therapy, 50% (5 of 10) rebounded when bNAb levels declined after week 28 (65010, 65015, 98050, 98058 and 98060), whereas 50% (5 of 10) showed virologic control for the duration of follow-up (38–44 weeks) even after bNAb levels declined to low or undetectable levels (98055, 98056, 09057, 98063 and 98064), which was at week 44 at the end of the study for 4 participants and week 38 for 1 participant who was subsequently lost to follow-up (98063). Thus, 5 of 12 (42%) participants in group 2 showed long-term virologic control for the duration of follow-up, which was more than 24 weeks after the final bNAb infusion. As this was an exploratory proof-of-concept trial, analyses were descriptive, and no formal null hypothesis was tested.

In the two participants who showed early viral rebound, virologic failure occurred in the presence of high levels of all three antibodies, 208.8, 178.3 and 422.1 μg ml⁻¹ of PGT121, PGDM1400 and VRC07-523LS, respectively (Fig. 3). By contrast, in the 5 participants who rebounded between weeks 28 and 44, the average serum concentrations of PGT121, PGDM1400 and VRC07-523LS were 3.1, 8.9 and 93.4 μg ml⁻¹, respectively, at the time of rebound. In the participants who did not rebound by week 44, the average serum concentrations were 3.4, 1.5 and 24.7 μg ml⁻¹, respectively. These data show that the triple-bNAb cocktail maintained virologic suppression for >28 weeks in the majority of PLWH in this trial, with viral rebound generally characterized by PGT121 and PGDM1400 concentrations below 10 μg ml⁻¹ and a VRC07-523LS concentration below 100 μg ml⁻¹. Although two participants rebounded early, no other participants rebounded in the context of therapeutic bNAb levels until bNAb levels declined to these low threshold levels.

Fig. 3: Virological and pharmacokinetic follow-up of individual participants.

Plasma HIV-1 RNA levels (red line; right y-axis) and bNAb serum concentrations (VRC07-523LS (green), PGDM1400 (blue) and PGT121 (red); left y-axis) of group 2 participants. Gray-shaded areas indicate time on ART. Participant 65021 resumed ART but was lost to follow-up (LTFU), and participant 98063 moved out of state and had to exit the study prematurely. The red dotted horizontal line indicates the limit of detection.

Notably, 5 participants maintained virologic suppression for the duration of follow-up for over 38–44 weeks, despite bNAb levels declining to low or undetectable levels. One possibility is that very low levels of bNAbs may, in some cases, be sufficient for virologic suppression, and tissue levels of bNAbs remain to be determined. The bNAb infusions did not detectably reduce the size of the peripheral HIV-1 reservoir, as measured by intact provirus detection assays (IPDAs) (Fig. 4a,b), although we cannot exclude minor changes in the viral reservoir by these assays and we did not collect tissues for reservoir measurements in this study. Moreover, the bNAb infusions did not lead to a detectable increase in autologous antigen-specific T cell responses at week 24, as measured by the IFNγ enzyme-linked immunospot (ELISpot) and intracellular cytokine staining assays (Fig. 4c–g), although we cannot exclude potential modulation of other T cell subsets or functions. High-resolution human leukocyte antigen (HLA) typing did not reveal protective HLA alleles in any participants in this study (Supplementary Table 6). As per protocol, all participants that did not develop detectable plasma viremia were advised to restart ART at the end of the trial.

Fig. 4: Viral reservoir and T cell immunity.

a, Baseline (before the first bNAb infusion) frequencies of CD4⁺ T cells carrying intact proviral HIV DNA by the IPDA assay in study participants that had either early, late or no rebound of plasma viremia during study follow-up. No statistically significant differences were observed. b, Dynamics of IPDA reservoir virus in the CD4⁺ T cells of participants with late plasma viral rebound (left panel) at baseline and at month 6 (before plasma viremia) and in participants with no viral rebound (right panel) at baseline and at the end of study follow-up. c, Frequency of T cell responses to HIV PTE peptide pools (Gag1, Gag2, Env1, Env2 and Env3) at baseline as measured by the IFNγ ELISpot assay. Shown are the means of responses from each participant group to each peptide pool. d, Frequencies of HIV-specific IFNγ⁺ CD8⁺ T cell subsets across early, late and no rebounders at baseline as determined by flow cytometry. e, Dynamics of HIV-specific T cell responses at baseline and 6 months in participants with late viral rebound and no viral rebound as quantified by IFNγ ELISpot assay. f,g, Dynamics of HIV-specific IFNγ⁺ CD8⁺ T cells (top panel) and CD4⁺ T cells (bottom panel) in participants with late viral rebound (f) and no viral rebound (g) as determined by intracellular cytokine staining. SFC, spot-forming cells; f/u, follow-up.

Proteomics

We next performed a plasma proteomics pilot analysis to provide additional insight into potential mechanisms of long-term virologic control following triple-bNAb therapy. We focused on the 10 participants who achieved virologic control for the duration of bNAb therapy and either rebounded when bNAb levels declined (rebounders; n = 5) or showed virologic control for the duration of the 38–44-week follow-up period (controllers; n = 5). We assessed plasma proteomic profiles at baseline before the initiation of bNAb infusions and at week 24 following the completion of bNAb infusions but before viral rebound. Comparing week 24 with the baseline, rebounders showed activation of pathways involving immune cell signatures, proinflammatory and interferon responses, apoptosis and epigenetic modifications involving histones and chromatin, suggesting immune activation and possible low levels of viral activation even before plasma viremia was detectable (Fig. 5a). By contrast, controllers showed an increase of metabolic pathways and a decrease of certain proinflammatory pathways (Fig. 5a). Comparing controllers with rebounders, we observed minimal differences in pathways and key genes at baseline before initiation of bNAb infusions between these groups (Fig. 5b). Following completion of bNAb infusions, controllers showed decreased inducible T-cell co-stimulator (ICOS) signaling, macrophage activation, T cell activation and exhaustion, apoptosis and proinflammatory signaling compared with rebounders (Fig. 5b). These data suggest a potential mechanism in which bNAb infusion was associated with immune activation, T cell exhaustion and proinflammatory signaling in participants who subsequently showed long-term virologic control.

Fig. 5: Proteomic profiles.

a, Proteomic pathways at week 24 compared with the baseline in the 10 participants who showed virologic control for the duration of bNAb therapy and who showed long-term virologic control for the duration of follow-up (controllers; n = 5) or who showed viral rebound (rebounders; n = 5). The size of the circle indicates the level of pathway enrichment; red refers to increased and blue refers to decreased levels for each pathway. b, Proteomic pathways modulated in controllers compared with rebounders at baseline before the initiation of bNAb infusions and at week 24 following completion of bNAb infusions but before viral rebound. Protein levels were averaged across individuals for each group. The color gradient indicates the log₂ transformation of the fold change, ranging from blue (decreased) to red (increased) protein levels.

Neutralization sensitivity of reservoir and rebound virus

We generated HIV-1 sequences via single genome amplification of proviral DNA in peripheral blood mononuclear cells (PBMCs) at baseline and where viral rebound occurred of plasma viral RNA in ten participants in group 2 (Fig. 6). We constructed HIV-1 pseudoviruses to assess bNAb sensitivity using TZM-bl neutralization assays and analyzed matched env sequences to identify the mutations that may contribute to underlying resistance patterns. In total, we generated 83 sequences from baseline PBMC proviral DNA and 227 sequences from plasma viral RNA from rebound viruses (with an average of 10 baseline and 32 rebound sequences per participant). We were unable to obtain viral sequences from 2 of the 12 participants (98057, 98064) who did not rebound owing to limited available PBMCs.

Fig. 6: Rebound virus sequences.

Phylogenetic tree of participant baseline and viral rebound Envs. Baseline Envs are colored blue and rebound Envs colored red. Orange indicates Env sequences derived from PBMCs during the absence of plasma viral rebound. Participant identification numbers are shown at the root of each participant Env cluster. Envs that were tested for neutralization are shown by filled circles or triangles, and their neutralization IC₈₀ titers for each of the three bNAbs are shown on the right using colored boxes, with color coding per legend below.

In the two participants who showed early viral rebound (65010, 98059), partial or complete neutralization resistance to both PGDM1400 and PGT121 was detected at baseline in provirus clones in CD4⁺ T cells, and rebound virus clones were fully resistant to both PGT121 and PGDM1400 but had variable sensitivity to VRC07-523LS (Extended Data Figs. 2 and 3). In 65021, baseline and rebound virus clones were virtually identical at VRC07-523LS contact sites, suggesting a lack of selection pressure by VRC07-523LS in this individual, despite high VRC07-523LS serum concentrations of 300–1,000 µg ml⁻¹. In contrast, rebound virus clones from 98059 showed resistance to VRC07-523LS that could be attributed to the G459D mutation, a previously known resistance signature¹⁷.

In the five participants who showed late viral rebound, two different susceptibility patterns in the rebound viruses were observed. In three participants (98050, 98058, 98060), most rebound viruses were resistant to PGT121 and PGDM1400 but remained sensitive to VRC07-523LS (Extended Data Figs. 4 and 5). These participants showed low serum concentrations of PGT121 and PGDM1400 but detectable levels of VRC07-523LS at the time of viral rebound. In the two remaining participants with late rebound (65010, 65015), rebound virus remained fully sensitive to all three bNAbs (Extended Data Figs. 6 and 7). We also obtained proviral sequences from peripheral CD4⁺ T cells from three of the five participants who did not rebound (Extended Data Figs. 8 and 9). Based on bNAb signature analysis, all sequences from these three participants showed enrichment of sensitivity signatures at the core PGT121 and VRC07-523LS contact sites while resistance signatures in the core C-strand contact region of PGDM1400 (161, 165 and 169) were observed (Extended Data Figs. 8 and 9). As pseudovirus generation failed for these sequences, the impact of these on neutralization sensitivity to PGDM1400 could not be ascertained. For the remaining two participants without viral rebound, we were unable to generate baseline proviral sequences.

Resistance to PGT121 was associated with the loss or absence of the 332 glycan (in participants 65010, 98050, 98058 and 98060), H330Y (in participant 65021) or G324R (in participant 98059) in the key ³²⁴GDIR³²⁷ motif contact site¹⁸, which are previously described resistance signatures¹⁷ (Extended Data Figs. 3–7). Resistance to PGDM1400 was associated with either previously reported resistance mutations N130 (in participants 65021, 98050 and 98058) and V16917 (in participant 98050), or due to mutation K168E or R190S (in participant 65015) that reduced the net positive charge of the V2-apex epitope that is preferred by PGDM1400 or loss of the paired Cys in the hypervariable V1 loop that may potentiate resistance by destabilizing the epitope–proximal loop¹⁹.

Discussion

In this study, we show that the triple-bNAb cocktail PGT121, PGDM1400 and VRC07-523LS resulted in virologic control following ART discontinuation for the duration of bNAb therapy in 83% (10 of 12) of participants who were not screened for bNAb sensitivity at baseline. Of the 10 participants who showed virologic suppression for the duration of bNAb therapy, 50% (5 of 10) rebounded when bNAb levels declined, whereas 50% (5 of 10) showed virologic control for the duration of the follow-up period (38–44 weeks) even after bNAb levels declined to low or undetectable levels. Thus, 42% (5 of 12) of the total participants in group 2 showed long-term virologic control for the duration of follow-up.

Previous studies have shown that bNAbs given alone or in combination have antiviral activity in viremic PLWH^1,2,5,8,9,10. However, baseline neutralization resistance in the diverse within-host HIV-1 quasispecies resulted in rapid viral rebound, posing a challenge for bNAb therapy¹⁴. Reliable virological control without ART thus requires a combination of bNAbs with complementary epitope specificities⁷. Our data suggest that the triple-bNAb cocktail can serve as an effective ART replacement in PLWH for a prolonged period of time following ART discontinuation, except in individuals with baseline resistance to both PGT121 and PGDM1400.

Previous studies have shown limited duration of viral control with single and dual combinations of bNAbs. A previous study of 10-1074 and 3BNC117 in PLWH showed median virologic suppression of 20 weeks following ART discontinuation, with 2 of 17 (12%) participants showing long-term virologic control until the completion of the study at week 48 (ref. ³). Another study showed long-term virologic control (up to 30 weeks) with this dual-bNAb cocktail in two of seven PLWH who initiated ART during acute infection⁴, although these individuals may have had less diverse virus in their reservoirs than typical PLWH who initiated ART during chronic infection. A concern with current options for dual-bNAb therapy is that a substantial portion of circulating viruses are sensitive to only one antibody in the cocktail (Extended Data Fig. 10). By contrast, triple-bNAb therapy can achieve more robust coverage of global viruses with two or more antibodies in the cocktail^5,7.

Viral rebound was generally characterized by resistance to the two glycan-dependent antibodies, PGT121 and PGDM1400, but only limited resistance to the CD4bs antibody VRC07-523LS, suggesting that the glycan antibodies provided more selection pressure in vivo or had lower bars to resistance. Excluding the two individuals who rebounded early, participants generally rebounded when bNAb levels declined to approximately 10 μg ml⁻¹ for PGT121 and PGDM1400 and 100 μg ml⁻¹ for VRC07-523LS. The 80% inhibitory dose (ID80) titers of VRC07-523LS at the time of viral rebound were consistent across participants (approximately 100 μg ml⁻¹) and were higher compared with those of PGT121 and PGDM1400. The emergence of rebound virus resistant to VRC07-523LS in one participant confirms that VRC07-523LS can exert selective pressure in certain cases. Defining the precise bNAb levels for therapeutic efficacy in PLWH will require larger studies.

We did not observe that bNAb therapy directly impacted the viral reservoir or host T cell responses, although a proteomic analysis suggested that bNAb administration may have reduced immune activation, T cell exhaustion and proinflammatory signaling in the participants who subsequently showed long-term virologic control. Consistent with these observations, studies in macaques have suggested that additional immune activation or latency reversal agents are necessary for bNAbs to target the viral reservoir^20,21,22. Moreover, administration of these bNAbs in macaques and humans has not resulted in augmented autologous antigen-specific CD4 or CD8 T cell responses^1,5,20,21, suggesting that a classic vaccinal effect²³ was probably not responsible for the virologic control observed in our study. Taken together, we speculate that innate and cellular immune modulation of the host–virus balance following bNAb infusion may have contributed to the observed long-term virologic control after bNAb levels declined to low or undetectable levels.

In conclusion, our data show that a triple-bNAb cocktail achieved virologic control in PLWH in the absence of ART, except when baseline reservoir provirus contained virus resistant to the two glycan-dependent antibodies PGT121 and PGDM1400. Moreover, long-term virologic control was observed in a subset of participants following bNAb therapy. Importantly, no participants developed viral rebound from early in the study (week 10) until after completion of bNAb dosing and decline of bNAbs to subtherapeutic levels (week 28). These initial data suggest that the triple-bNAb cocktail could be explored further as an ART replacement strategy in PLWH. It is theoretically possible that the triple-bNAb cocktail will also prove highly effective for HIV-1 prevention. Long-acting versions of all three of these antibodies are currently being developed, which will probably facilitate the development of a triple-bNAb cocktail that can be administered once every 6 months for both HIV-1 therapy and prevention.

Methods

Study design

The study evaluated the safety, pharmacokinetics and antiviral activity of PGDM1400, PGT121 and VRC07-523LS bNAbs. Part 1 was a single-center, open-label study to evaluate a single IV dose of PGT121 and VRC07-523LS (group 1A: 30 mg kg⁻¹ per antibody, respectively; n = 3) or the combination of PGT121, VRC07-523LS and PGDM1400 (group 1B: 20 mg kg⁻¹ per antibody, respectively; n = 3) in adults without HIV at the Beth Israel Deaconess Medical Center (BIDMC), Boston, MA. Part 2 of the study was a multicenter, open-label trial of three monthly IV administrations of 20 mg kg⁻¹ each of PGT121, VRC07-523LS and PGDM1400 (group 2, n = 13) with the option of additional three (six in total) monthly IV infusions in PLWH on ART at three sites: BIDMC, Boston, MA; Orlando Immunology Center (OIC), Orlando, FL; and Houston AIDS Research Team (HART), McGovern Medical School at The University of Texas Health Science Center, TX. All participants in group 2 were asked to interrupt their ART 2 days after the initial IV administration of the three antibodies. Participants had HIV RNA monitoring every 2 weeks until the day 308 visit. If HIV RNA is confirmed detectable with >50 copies per ml in a repeat specimen, weekly quantitative HIV RNA testing is performed until confirmed undetectable (on 2 consecutive weekly assessments) or until criteria to reinitiate ART are met.

The ART restart criteria are as follows: (1) plasma HIV-1 RNA levels are ≥1,000 copies per ml confirmed on prompt repeat measurement, (2) two consecutive CD4⁺ T cell counts <300 cells per μl at least 2 weeks apart, (3) clinical progression to CDC stage 3 disease, (4) documented pregnancy or (5) request from the participant or HIV care provider. Volunteers who reinitiated ART were followed until their HIV-1 RNA levels were undetectable for two consecutive measurements. The protocol was approved by the BIDMC Institutional Review Board, the Advarra Institutional Review Board and the HART Committee for the Protection of Human Subjects. A safety monitoring committee, consisting of four external clinical experts on bNAbs and HIV clinical trials, provided an independent assessment of safety in this study. This study was registered at ClinicalTrials.gov, number NCT03721510.

Study participants

Participants were eligible for the study across groups if they did not have any clinically significant acute or chronic medical condition (besides HIV), such as chronic hepatitis B, active hepatitis C, significant psychiatric disorder, alcohol or substance use disorder, or chronic kidney or liver disease and if they had a body mass index >18 and <35. Sexually active participants had to be willing to use contraception for 3 months following investigational product (IP) administration and could not be pregnant or breastfeeding. Participants were eligible for group 1 if they were also 18–50 years of age, at low risk of HIV infection and willing to maintain low-risk behavior. PLWH (group 2) were eligible if they were 18–65 years of age, had CD4 ≥400 cells per μl, had no history of AIDS-defining illness within the previous 5 years, had no history of CD4 nadir <200 cells per μl and were on ART for a minimum of 24 months, with plasma HIV-1 RNA levels of <50 copies per ml for at least 12 months and not currently on a non-nucleoside reverse transcriptase inhibitor-based regimen. Before receipt of an investigational HIV vaccine candidate, monoclonal antibody or polyclonal immunoglobulin was an exclusion criterion. Volunteers were counseled during screening and throughout the trial regarding the potential risks of an analytical antiretroviral treatment interruption (ATI), monitoring procedures during the ATI and ART restart criteria. Willingness to undergo an ATI was confirmed. All participants gave written informed consent and successfully completed an assessment of understanding before the initiation of study procedures.

In part 1, a sentinel volunteer from subgroup 1A and a sentinel volunteer from subgroup 1B received infusions at least 24 h apart to allow observation of IP-related adverse events. If no reactogenicity or adverse events that met the pause criteria occurred within 24 h after IV infusion for the sentinel volunteer in subgroup 1A, then the remaining volunteers in subgroup 1A were infused. The same applied to the sentinel volunteer in subgroup 1B.

Endpoints

The primary endpoints were safety and tolerability as well as pharmacokinetics. Safety and tolerability endpoints included (1) proportion of participants with moderate or greater reactogenicity (for example, solicited AEs) for 3 days following each IV infusion of PGT121, PGDM1400 and VRC07-523LS; (2) proportion of participants with moderate or greater PGDM1400 and/or PGT121 and/or VRC07-523LS bNAb-related unsolicited AEs, including laboratory safety parameters (biochemical, hematological) following IV infusion of PGDM1400 and/or PGT121 and/or VRC07-523LS for the first 56 days post administration of IPs; and (3) proportion of participants with PGDM1400 and/or PGT121 and/or VRC07-523LS-related SAEs throughout the study period. Pharmacokinetic endpoints included elimination half-life (t_1/2), clearance (CL/F), volume of distribution (Vz/F) and area under the concentration decay curve. The secondary endpoints focused specifically on group 2 PLWH: (1) proportion of volunteers who meet ART reinitiation criteria (defined as plasma HIV-1 RNA ≥1,000 copies per ml and/or CD4 count <300 cells per μl) on two consecutive measurements following ART interruption and time to meeting ART reinitiation criteria; (2) bNAb serum levels at the time of viral rebound; (3) change in CD4⁺ T cell count and frequency compared with the baseline as measured by single platform flow cytometry; (4) development of HIV-1 sequence variations in epitopes known to result in reduced PGDM1400 and/or PGT121 and/or VRC07-523LS neutralization susceptibility; and (5) development of anti-PGT121, anti-VRC07-523LS and anti-PGDM1400 antibodies.

IPs

PGDM1400 is a recombinant, fully human monoclonal antibody (mAb) of the IgG1 isotype that binds to the HIV envelope. PGDM1400 was formulated in a 20 mM acetate, 9% sucrose, 0.008% polysorbate 80, pH 5.2 formulation buffer at a concentration of 50 mg ml⁻¹. Each 10 ml vial contained 6 ml of PGDM1400. PGT121 is a recombinant, fully human mAb of the IgG1 isotype that binds to the HIV envelope. PGT121 was formulated in a 20 mM acetate, 9% sucrose, 0.008% polysorbate 80, pH 5.2 formulation buffer at a concentration of 50 mg ml⁻¹. Each 10 ml vial contained 6 ml of PGT121. VRC07-523LS is a recombinant, fully human mAb of the IgG1 isotype that binds to the HIV envelope. VRC07-523LS was formulated at a concentration of 100 ± 10 mg ml⁻¹ in a buffer composed of 50 mM histidine, 50 mM sodium chloride, 5% sucrose and 2.5% sorbitol at pH 6.8. Each standard 10 ml or 3 ml glass vial contained 6.25 ± 0.1 ml or 2.25 ± 0.1 ml, respectively, of VRC07-523LS. The M428L/N434S is known to increase affinity for the neonatal Fc receptor (FcRn) and extends antibody half-life in vivo²⁴.

Safety assessments

Local and systemic reactogenicity safety data were collected for 3 days after IP administration (see protocol and schedule of procedures in Supplementary Information). Data on unsolicited AEs were collected for 56 days following IP administration. Potential immune-mediated diseases (pIMDs) were considered AEs of special interest as they could potentially be caused by immune responses to the IP; pIMDs included autoimmune diseases and other inflammatory and/or neurologic disorders that may or may not have an autoimmune etiology. Data on pIMDs and SAEs were collected through study day 308. Blood samples for serum chemistry and hematology were collected throughout the study, while urine samples for pregnancy testing and urinalysis were collected at screening, before each IP administration and throughout the study thereafter. Blood samples were collected from PLWH to determine HIV-1 RNA levels and CD4⁺ T cell counts throughout the study. Medical monitoring was provided by a protocol safety review team and an independent safety monitoring committee. Local and systemic AEs were graded by the NIAID Division of AIDS (DAIDS) Table for Grading the Severity of Adult and Pediatric Adverse Events Version 2.1, July 2017. For the first 24 h after IP infusion or injection, any infusion-related reactions, including cytokine release syndrome, were graded by the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) v5.0, November 27, 2017. Peripheral blood was collected to determine PGDM1400, PGT121 and VRC07-523LS serum levels, HIV sequencing and immunogenicity, among other research assessments outlined in the study protocol (Supplementary Information).

ART counseling

Participants who had not initiated or made plans to initiate ART by the final study visit received ART counseling again at their final study visit.

HIV-1 RNA levels and CD4⁺ T cell measurement

Plasma HIV-1 RNA levels were measured at BIDMC using the Roche COBAS AmpliPrep/COBAS TaqMan HIV-1 Test, version 2.0 (lower limit of quantification (LLoQ) = 23 RNA copies per ml), until it was replaced with the Hologic Aptima HIV-1 Quant Assay (LLoQ = 32 RNA copies per ml). HIV-1 viral load was measured at OIC and HART using the Abbott Real-Time HIV-1 assay (LLoQ = 40 RNA copies per ml). Viral load assays were performed at LabCorp or at BIDMC. CD4⁺ T cell counts were measured using a clinical flow cytometry assay performed at LabCorp or at BIDMC.

Determination of bNAb serum levels

PGT121 and PGDM1400

PGT121 and PGDM1400 levels were determined by quantitative enzyme-linked immunosorbent assay that measures the ability of each mAb to bind to their specific biotinylated anti-idiotype antibody captured on 96-well streptavidin plates (Thermo Fisher). Standard curves consisted of either PGT121 or PGDM1400 IgG mAb. Samples with bNAb concentrations below the LLoQ were designated as the LLoQ value for plotting purposes. Samples with bNAb concentrations above the upper limit of quantification were further tested at various dilution factors and resulting concentrations falling within the linear range of the standard curves, corrected for the dilution factor and averaged.

VRC07-523LS²⁵

VRC07-523LS anti-ID antibody (5C9) was coated on bare Meso Scale Diagnostics (MSD) 96-well plates overnight at 4 °C. The plates were washed and blocked for 1 h with shaking. The blocked plates were washed, and diluted test samples as well as VRC07-523LS standards and controls were added to the assay plates. Plates were incubated with shaking for 2 h at room temperature. Plates were washed, and Sulfo-Tag-labeled mouse anti-human IgG detection antibody in assay diluent was applied to the plates and allowed to associate with complexed anti-ID–VRC07-523LS within the assay wells for 1 h, with shaking. The plates were washed to remove unbound detection antibody, and MSD Read Buffer containing Sulfo-Tag substrate was immediately added to the wells. Plates were read using the MSD Sector Imager S600. As current is applied to the plate, areas of well surface that form a full anti-ID–VRC07-523LS–Sulfo-Tag anti-human IgG complex emit light in an electrochemiluminescent reaction. The amount of VRC07-523LS sandwiched by the anti-ID and anti-human IgG antibodies is directly proportional to the concentration of reactive VRC07-523LS. Serial dilutions of sample within the dynamic range of the standard curve were interpolated to assign a sample concentration. Analysis was performed using MSD Discovery Workbench Software and Microsoft Excel.

Limits of quantitation for analytes

Analyte	LLoQ (µg ml−1)
PGT121	0.485
PGDM1400	0.630
VRC07-523LS	0.8

Half-life calculation

To estimate the half-life of each treatment, the concentration–time relationship was analyzed. The natural logarithm of the concentration was regressed against time using a linear regression model implemented through the R lm function. The resulting model provided coefficients representing the intercept (log_conc_0) and the slope (ke), which signified the initial logarithmic concentration and the decay rate constant, respectively. The half-life of the treatment was computed using the equation half-life = −log(0.5)/ke, where ke represents the decay rate constant obtained from the linear regression model.

HIV-1 env gene sequencing and production of pseudoviruses

Single genome amplification assays were performed by isolating HIV-1 RNA and reverse transcribing to viral cDNA or isolating DNA from CD4⁺ T cells isolated with a negative selection kit (Stem Cell) from frozen PBMCs²⁶. First-round and nested PCRs were carried out with Platinum SuperFi II Polymerase (Invitrogen) together with HIV B primers as previously described⁵ (Supplementary Table 7). Amplicons from cDNA and DNA dilutions resulting in ≤33% positive wells were considered to result from amplification of a single cDNA or DNA molecule according to the Poisson distribution (P < 0.05) and were then sequenced. For each participant who rebounded, 22–52 rebound sequences were analyzed, and selected viral sequences were used to generate cytomegalovirus-promoter-based pseudoviruses²⁷.

TZM-bl neutralization assay

The neutralization activity of the bNAbs against participant-derived Env pseudoviruses was measured as a function of reduction in Tat-regulated luciferase (Luc) reporter gene expression in TZM-bl cells^27,28,29,30. Pseudoviruses were tested against individual bnAb clinical drug products using a primary concentration of 50 μg ml⁻¹ with fivefold dilution series in duplicate wells. IC₅₀ and IC₈₀ titers were calculated as bNAb concentrations that inhibited 50% or 80% of viral infection, respectively. All assays were performed in a laboratory meeting the standards of Good Clinical Laboratory Practice.

Calculation of ID80 titers

For each individual bNAb, ID80 titers at the time of viral rebound were calculated as (serum concentration of bNAb)/IC₈₀. The bNAb serum concentrations closest to the timepoint of the first HIV RNA level >200 copies per ml were used.

Intact proviral DNA assay

CD4⁺ T cells from frozen PBMCs were isolated using the EasySep Human CD4⁺ T Cell Isolation Kit (Stem Cell), and DNA was subsequently extracted with the QIAmp DNA Mini Kit (Qiagen). The IPDA was then performed on the DNA samples as described previously³¹.

Sequence analyses

Env gene sequences were extracted manually and codon aligned using the webtool Gene Cutter on the Los Alamos HIV database (https://www.hiv.lanl.gov/content/sequence/GENE_CUTTER/cutter.html). Alignments were further refined manually. All participant sequences were found to be subtype B using the recombinant identification program (RIP) program using the consensus sequences. Five sequences were found to be highly enriched for apolipoprotein B mRNA editing catalytic polypeptide-like (APOBEC) motif G-to-A substitutions and stop codons using the HYPERMUT tool (https://www.hiv.lanl.gov/content/sequence/HYPERMUT/hypermut.html), so were excluded from further analysis: two among the 98063-baseline sequences, one in the 98059-rebound sequences, one in the 65021-baseline sequences and one in the 65010-baseline-V2-B11a sequence. Five sequences from 65010 were found to have been contaminated with the pseudovirus sequence SG3 (B.UNITED_STATES.1987.SG3.L02317). Six additional from this subject had short stretches that were SG3 like, as ascertained using RIP (www.hiv.lanl.gov/content/sequence/RIP/RIP.html)³², and these were also removed from the analysis to exclude the possibility of in vitro recombination. The other 62 sequences from 65010 did not resemble SG3 and were consistent with being a quasispecies from a single host. Additional sequences were generated to confirm this finding. The bNAb-resistance mutations were identified using defined signature sites³³ by comparing baseline and rebound Env sequences at these sites together with matched bNAb neutralization data. For some Envs, canonical resistance signatures were not found despite bNAb resistance, and putative mutations underlying such resistance were identified by manual inspection of resistant and sensitive sequences together with information on proximity to bNAb epitopes. The PHYML³⁴ maximum likelihood tree, using a general time reversible model, empirical equilibrium frequencies and eight rate categories, estimated the proportion of invariable sites and gamma shape parameter from the data, starting with an NJ tree, and was run using the PHYML at the Los Alamos National Laboratory HIV database (https://www.hiv.lanl.gov/content/sequence/PHYML/interface.html). The HXB2 reference strain (K03455) was used as an outgroup to identify mutational positions in Env. The full tree with all participant data is shown in Fig. 6, while participant-specific sublineage trees were excised from this tree for the subsequent extended data figures. Sequence logos were obtained from the web tool AnalyzeAlign on the Los Alamos HIV database (https://www.hiv.lanl.gov/content/sequence/ANALYZEALIGN/analyze_align.html).

Immunogenicity assessments

HIV-specific T cell responses were measured by IFNγ ELISpot assay using potential T cell epitope (PTE) Env, Pol and Gag peptide libraries^35,36. ELISPot assay was performed as follows. White membrane plates (Millipore) were coated at 4 °C overnight with 10 µg ml^–1 anti-human IFNγ (Mabtech). Rested PBMCs were plated at 2 × 10⁶ with PTE Env, Pol or Gag pools at 2 µg ml^–1 for 18 h at 37 °C. Development was achieved by the addition of biotin (Mabtech), antibiotin (VectorLabs) and chromagen (Pierce). Background subtraction was noted by a matched DMSO peptide concentration of 0.4%. Multiparameter intracellular cytokine staining assays were performed essentially as described³⁷. Intracellular cytokine staining assays were performed with 10⁶ PBMCs incubated for 6 h at 37 °C with medium, 10 pg ml^–1 phorbol myristate acetate and 1 µg ml^–1 ionomycin (Sigma-Aldrich), or 1 µg ml^–1 HIV-1 Env, Gag or Pol peptide pools. Cultures contained monensin (GolgiStop, BD Biosciences), brefeldin A (GolgiPlug, BD Biosciences) and 1 µg ml^–1 mAb against human CD49d (clone 9F10). Cells were then stained with predetermined titers of mAbs against CD3 (clone SP34.2; Alexa700, BD, 0.25 µl per test), CD4 (clone L200, BV711, BD, 0.75 µl per test), CD8 (clone SK1, BUV805, BD, 1 µl per test), CD45RA (clone 5H9, APC-H7, BD, 2.5 µl per test) and CCR7 (clone EH12.1, BB700, BD, 2.5 µl per test) and stained intracellularly with IFNγ (clone B27, BUV395, BD, 2.5 µl per test). IFNγ+ responses were counted after subtraction of 2 times the values in the PBS controls (gating strategy in Supplementary Fig. 1).

Assessment of ART drug levels in plasma

Plasma samples from participants without viral rebound (98055, 98056, 98057, 98063 and 98064) from the last study timepoint were quantitatively analyzed for emtricitabine and tenofovir at the University of Nebraska Medical Center. These two antiretroviral drugs were chosen for testing because they are present in commonly used combination antiretroviral regimens.

Sample size and statistical analysis

The sample size for the safety and tolerability analysis was 18 participants according to the study design used to characterize the safety profile of 1, 3 or 6 IV infusions of PGT121, VRC07-523LS and PGDM1400. For life-threatening adverse events related to the active product, if none of the 18 participants in either group 1 or group 2 is receiving the active product experience such reactions, then the exact 95% upper confidence bound for the rate of these adverse events in the population is 16.1%. This was an exploratory proof-of-concept trial and the analysis was descriptive, and no formal null hypothesis was tested. The frequency of moderate or greater reactogenicity events was determined and compared between groups. The frequency of SAEs judged possibly related, probably related or related to the IP was determined. All AEs were analyzed and grouped by seriousness, severity and relationship to the IP (as judged by the investigators). An interim safety analysis of group data was carried out after each regimen according to the study schema without unblinding the study to investigators or participants. At the end of the study, a full analysis was prepared. Missing data were excluded from the statistical analysis. The sample size for pharmacokinetic analysis was three per dose for subgroups 1A and 1B and 12 for group 2, sufficient for the planned analyses based on previous experience with PGT121, PGDM1400 and VRC07-523LS pharmacokinetics^1,24. The sample size to evaluate the antiviral activity of 3 or 6 IV infusions of PGT121, VRC07-523LS and PGDM1400 in PLWH following ART interruption was approximately 12. As this was an exploratory proof-of-concept trial and analysis was descriptive, no formal null hypothesis was tested.

Proteomics

Plasma was analyzed using SomaScan Assay v4.1 for human plasma (catalog number 900-00021), measuring the expression of 6,596 unique human protein targets using 7,596 SOMAmer reagents, single-stranded DNA aptamers, according to the manufacturer’s standard protocol (SomaLogic) (Supplementary Information). The assay used standard controls, including 12 hybridization normalization control sequences used to control for variability in the Agilent microarray readout process, as well as 5 human calibrator control pooled plasma replicates and 3 quality control (QC) pooled replicates used to mitigate batch effects and verify the quality of the assay run using standard acceptance criteria. The readout was performed using Agilent microarray hybridization, scan and feature extraction technology. Twelve hybridization control SOMAmers are added alongside SOMAmers to be measured from the serum samples and controls of each well during the SOMAmer elution step to control for readout variability. Plate calibration was performed by calculating the ratio of the calibrator reference relative fluorescence unit (RFU) value to the plate-specific calibrator replicate median RFU value for each SOMAmer. The resulting ratio distribution was decomposed into a plate scale factor defined by the median of the distribution and a vector of SOMAmer-specific calibration scale factors. Normalization of QC replicates and samples was performed using adaptive normalization by maximum likelihood with point and variance estimates from a normal US population. Post-calibration accuracy was estimated using the ratio of the QC reference RFU value to the plate-specific QC replicate median RFU value for each SOMAmer. The resulting QC ratio distribution provided a robust estimate of accuracy for each SOMAmer on every plate. Plate-specific acceptance criteria are as follows: plate scale factor between 0.4 and 2.5 and 85% of QC ratios between 0.8 and 1.2. For platform standardization, all RFU values across all samples were converted to SomaScan v4.0 using Somascan internal scaling factors. For SomaO reference serum and plasma datasets, EDTA plasma and serum baseline samples were collected from individuals without any underlying conditions and represent a healthy adult proteome from a mix of male and female individuals ranging from 18 to 90 years of age. We used the Linear Models for Microarray Data (limma) R package to identify differentially expressed proteins. The method involves fitting a linear model to the data and then performing a t-test to identify proteins that are differentially expressed between two or more groups. P values were corrected for multiple testing using the Benjamini–Hochberg method. R packages ggplot2 and ComplexHeatmap were used to generated the figures. Differential expression analysis was conducted with an n of 5 for each group (pre- and post-bNAbs) using the limma package in R. Pathway enrichment analysis was performed using Gene Set Enrichment Analysis against the C2 database from the Molecular Signatures Database (https://www.gsea-msigdb.org/gsea/msigdb/collections.jsp) to elucidate the biological pathways and processes modulated following bNAb administration.

HLA typing

HLA class I genotyping of the group 2 participants was performed using a targeted next-generation sequencing method as described previously with minor updates³⁸. Locus-specific primers were used to amplify polymorphic exons of the HLA-A and HLA-B (exons 1–4) and HLA-C (exons 1–5) genes with the Fluidigm Access Array System and Juno LP 48.48 IFCs (Fluidigm). The Fluidigm PCR amplicons were pooled and subjected to sequencing on an Illumina MiSeq sequencer (Illumina). HLA alleles and genotypes were called using the Omixon HLA Explore (version 2.0.0) software (Omixon).

Statistics

Group-based comparisons were analyzed statistically using a Wilcoxon rank sum test. Depending on the null hypothesis, one-sided or two-sided P values were obtained and are mentioned in the figure legends. The recombination detection program RIP uses a z-test on hamming distances, as previously described³².

Changes to methods after trial commencement

Protocol version 4.0 modified the inclusion and exclusion criteria: exclude volunteers who have clear documentation of multi-antiretroviral resistance, as per the request of the FDA, to allow enrollment of individuals into group 2 that had initiated ART during any phase of infection. Protocol version 5.0 added the two clinical study sites OIC and HART. These changes were made to facilitate recruitment and had the approval of the safety monitoring committee, protocol safety review team and institutional review board.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Extended data

Extended Data Fig. 1 CD4 + T-cell counts overtime and Group 1 serum bNAb pharmacokinetics.

CD4 + T-cell Counts over Time in participants that had received 3 (a) or 6 bNAb infusions (b). Pharmacokinetics of PGT121, VRC07-523LS and PGDM1400 in Group 1 A and 1B. Levels of VRC07-523LS (green), PGDM1400 (blue) and PGT121 (red) in serum in participants in group 1 A after a single infusion of PGT121 and VRC07-523LS at 30 mg/kg each (c) or in group 1B after infusion of PGT121, VRC07-523LS and PGDM1400 at 20 mg/kg each (d), as determined by anti-idiotype specific Binding Antibody Multiplex Assay. Data are mean ± s.d. Mean half-life (t1/2) of each bNAb is indicated in days.

Extended Data Fig. 2 Participant 65021: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. Data demonstrates that pseudoviruses were completely sensitive to VRC07-523LS and resistant to PGDM1400 at baseline and rebound. PGDM1400 resistance might be explained by the presence of the resistance signatures N-130 and I-165 in all sequences. For PGT121, two baseline sequences were sensitive while the rebound sequences all likely arose from a resistant baseline variant carrying the resistance conferring mutation Y-330, consistent with PGT121 selection.

Extended Data Fig. 3 Participant 98059: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. Participant 98059 was the only participant that demonstrated fully resistant virus at rebound, although one of two baseline viruses was sensitive to all three antibodies, while the other one showed resistance to PGDM1400. Rebound viruses carried the G459D signature, associated with VRC07-523LS resistance, and the G324R mutation in the key ³²⁴GDIR³²⁷ motif in the bNAb epitope known to confer PGT121 resistance. There were no clear PGDM1400 resistance signatures that differentiated between the sensitive baseline virus (V1-D6) and resistant baseline virus (V1-C4), however in V1-D6 there was an atypical additional pair of Cysteines in the hypervariable section of the V1 loop, which could have formed a disulfide bridge thus stabilizing the V1 loop structure, a phenomenon which has been associated with increasing sensitivity to V2-Apex antibodies. None of the rebound viruses carried the additional pair of Cysteines found in the sensitive baseline virus V1-D6.

Extended Data Fig. 4 Participant 98050: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. All rebound Env sequences were resistant to PGT121, attributable to the absence of N332 glycan in each of the viruses, and all rebound sequences showed PGDM1400 resistance, attributable to resistance signatures V-169 and N130. Baseline sequence data demonstrated that Envs had identical sequences to the rebounding Envs at PGDM1400 and VRC07-523LS contact sites, suggesting that PGDM1400 resistance existed at baseline and that VRC07-523LS did not exert any significant selection pressure.

Extended Data Fig. 5 Participant 98058 and 98060: viral characteristics.

Virological and pharmacokinetic dynamics (see Fig. 3 for details), phylogenetic tree and Env evolution. (a) 98058 and (b) 98060 had been infected with HIV-1 at the same time and from the same individual, explaining the substantial HIV-1 sequence similarities that were observed (Fig. 6). All rebound Env sequences were resistant to PGT121, attributable to the absence of N332 glycan in each of the viruses, and all except 1 rebound sequence in participant 98060 showed PGDM1400 resistance, attributable to resistance signatures N130.

Extended Data Fig. 6 Participant 65010: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. Rebounding virus was in the majority fully sensitive to all three bNAbs and demonstrated only single variants (that were resistant to PGDM1400 (2 of 7 tested pseudoviruses) or PGT121 (1 of 7 tested pseudoviruses). Virus was highly diverse at baseline, but only two variant sublineages were recovered in the rebound quasispecies.

Extended Data Fig. 7 Participant 65015: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. Rebounding virus was fully sensitive to all three bNAbs.

Extended Data Fig. 8 Participant 98056: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree and (c) Env evolution between baseline & PBMC proviral DNA at week 24 (approximately 16 weeks after the last bNAb administration) in the bNAb epitopes. The latter showed identical sequences to the baseline Envs at each bNAb contact site suggesting the absence of any selection pressure in this individual’s viral reservoir. Baseline resistance signatures in the core C-strand contact region of PGDM1400 (position 161, 165 and 169) was observed.

Extended Data Fig. 9 Participant 98005 and 98063: viral characteristics.

(a) Participant 98055 and (b) participant 98063. Virological and pharmacokinetic dynamics (see Fig. 3 for details), phylogenetic tree and Env evolution.

Extended Data Fig. 10 Neutralization profiles and complementarity of coverage for triple and dual bNAb combinations.

(a) IC80 titers for single and combination bNAbs against a heterologous panel of 374 pseudoviruses^5,7. Each row represents a pseudovirus, and the ordering of pseudoviruses for the left and right heatmaps is different. IC80 titers are colored per legend. For 2 and 3 bNAb combinations, IC80 titers were predicted using Bliss-Hill model on individual bNAb data using CombiNAber (https://www.hiv.lanl.gov/content/sequence/COMBINABER/combinaber.html). The left-most column indicates subtype of the virus colored per legend. (b) Percent coverage of sensitive viruses to only one bNAb in the combination (light green), and to two or more bNAbs is shown (dark green). Sensitivity threshold of IC80 < 1ug/ml is used. (c) Breadth-potency curves for triple and dual bNAb combinations showing overall coverage (light shades) or dual coverage (dark shades).