B cell engagement with HIV-1 founder virus envelope predicts development of broadly neutralizing antibodies

SUMMARY

Determining which immunological mechanisms contribute to the development of broad neutralizing antibodies (bNAbs) during HIV-1 infection is a major goal to inform vaccine design. Using samples from a longitudinal HIV-1 acute infection cohort, we found key B cell determinants within the first 14–43 days of viremia predict the development of bNAbs years later. Individuals who develop neutralization breadth had significantly higher B cell engagement with the autologous founder HIV Envelope (Env) within 1 month of initial viremia. A higher frequency of founder Env-specific naïve B cells was associated with increased B cell activation and differentiation, and predictive of bNAb development. These data demonstrate that the initial B cell interaction with the founder HIV Env is important for the development of broadly neutralizing antibodies, and provide evidence that events within HIV acute infection lead to downstream functional outcomes.

eTOC

Interrogating how HIV-infected patients develop broadly neutralizing antibodies can inform vaccine design. Townsley et al. show that elevated B cell interactions with the HIV founder envelope glycoprotein within the first month of infection lead to increased B cell activation and differentiation, which predict development of neutralization breadth years later.

Graphical Abstract

INTRODUCTION

Generating broadly neutralizing antibodies (bNAbs) that overcome the sequence diversity of the HIV-1 envelope (glyco)protein (Env) is thought to be a critical component toward the design of a protective vaccine, yet no HIV-1 vaccine candidate to date has successfully elicited bNAbs (Mascola and Montefiori, 2010; Saunders et al., 2012). However, a fraction of HIV-1 infected individuals (~20%) generate bNAbs years after initial infection (Burton et al., 2012; Haynes et al., 2012b; Kwong and Mascola, 2012; Mikell et al., 2011; Saunders et al., 2012). Passive administration of monoclonal bNAbs derived from infected individuals protect macaques from simian-human immunodeficiency virus (SHIV) infection (Gautam et al., 2018; Haigwood et al., 2004; Hessell et al., 2010; Julg et al., 2017; Mascola et al., 2000), underscoring the potential benefits of generating bNAbs in HIV-1 vaccine strategies. Therefore, understanding immunological and virologic factors that drive the generation of bNAbs in natural infection may provide insight into the design of a protective vaccine.

bNAbs have been shown to evolve from early autologous neutralizing antibody responses, ascribing importance to early B cell selection in the evolution of neutralization breadth years later (Doria-Rose et al., 2014; Liao et al., 2013). However, B cell dysfunction and hypergammaglobulinemia are hallmarks of HIV infection (Moir and Fauci, 2009, 2017), confounding B cell developmental pathways that lead to bNAbs. Previous studies have shown that HIV-1 infection alters B cell subset distributions, predominantly resulting in increased immature transitional, tissue-like memory (TLM), and plasmablast B cell populations, and decreased resting memory B cells (Cagigi et al., 2008; Doria-Rose et al., 2009; Mabuka et al., 2017; Moir and Fauci, 2009, 2017). B cell migration to germinal centers (GC) within lymph nodes and the gut associated lymphoid tissues (GALT) is thought to be a critical component of bNAb development, as these are the sites of antibody diversification and affinity maturation (Havenar-Daughton et al., 2016b; Mabuka et al., 2017; Mesin et al., 2016; Richardson et al., 2018; Tam et al., 2016). B cell GC activity, reflected by higher peripheral levels of chemokine CXCL13, has been shown to associate with broad neutralization (Havenar-Daughton et al., 2016b).

Chronic infection studies demonstrated neutralization breadth associates with increased time after seroconversion, higher plasma viral load (VL), greater viral diversity or superinfection, and lower CD4⁺ T (CD4⁺) cell counts (Cortez et al., 2012; Cortez et al., 2015; Gray et al., 2011; Landais et al., 2016; Mikell et al., 2011; Piantadosi et al., 2009; Rusert et al., 2016). More recent cross-sectional studies suggested race and infecting subtype associate with development of neutralization breadth (Havenar-Daughton et al., 2016b; Landais et al., 2016; Mabuka et al., 2017; Richardson et al., 2018; Rusert et al., 2016). However, the majority of these results relied on samples from primary or chronic infection studies that lacked pre-infection and very early acute infection time points (Euler et al., 2013; Gonzalez et al., 2018; Sanchez-Merino et al., 2016; van den Kerkhof et al., 2013). Additionally, the interaction of the autologous founder Env with B cells in acute infection has not been well studied. Whether factors during acute infection contribute to development of neutralization breadth remains unknown and could be key in informing effective vaccination strategies.

To determine if acute infection factors impact the generation of neutralization breadth, we assessed the longitudinal development of neutralization breadth in 73 individuals from the RV217 acute infection cohort (Robb et al., 2016) starting from pre-infection through 3–6 years of HIV-1 infection. We identified 16 individuals whose plasma were able to neutralize >70% of a panel of 34 viruses (broad neutralizers) and compared their acute infection parameters to 12 individuals whose plasma neutralized <35% of the same viral panel (non-broad neutralizers). Using these groups, we found that a reduction in naïve peripheral B cells 30–43 days following initial viremia strongly predicted the development of broadly neutralizing antibodies. Individuals with less than 160 B cells/mm³ were 42 times more likely to develop neutralization breadth. Increased founder Env-specific B cell populations at month 1 post-viremia were predictive of the development of broadly neutralizing antibodies, and founder Env-specific naïve B cells associated with frequencies of founder Env-specific activated memory, plasmablast, and integrin β7⁺ B cell subsets, suggesting early B cell engagement, differentiation, and migration to lymph nodes. These data show that early, heightened B cell responses to HIV infection were important in the development of broadly neutralizing antibodies, and that neutralization breadth is shaped by robust B cell responses within the first stages of acute infection.

RESULTS

Longitudinal development of neutralization breadth starting from acute HIV infection

RV217 was a prospective study that enrolled uninfected, high-risk individuals in Thailand and East Africa who were screened twice weekly for detectable HIV RNA, thereby tracking HIV-1 infection in individuals starting from the first days of initial viremia (Robb et al., 2016). Upon the first detectable viral RNA (viremia), intensive longitudinal sampling was performed in the acute and chronic phases of HIV infection to assess VL and cell counts of CD4⁺ T cells, CD8⁺ T cells, Natural Killer (NK) cells, and B cells. Plasma from the first 73 treatment-naïve individuals within the cohort were selected for longitudinal development of bNAbs. Within these individuals VL peaked at a median of 13–14 days post-viremia (days following the initial detection of HIV RNA (Robb et al., 2016)), before declining to a VL set point between days 30–43 (month 1) post-viremia (Figure 1A). Cell counts varied during acute infection, with the greatest changes coinciding with peak VL, then remaining largely unchanged after reaching early VL setpoint at days 30–43 post-viremia (Figure 1A). Time points starting from year 1 post-viremia were assessed for the presence of bNAbs against a panel of 34 viruses representing diverse subtypes, with only 3 viruses considered to be ‘Tier 1A’ (Table S1). Neutralization of the 34 Env-pseudovirus panel strongly correlated with neutralization of a previously published 208 Env-pseudovirus panel (Krebs et al., 2019). The panel included viruses capable of predicting neutralizing antibody epitope specificities via fingerprinting analysis, allowing for the simultaneous analysis of neutralization breadth, potency, and specificity (Doria-Rose et al., 2017; Georgiev et al., 2013; Raju et al., 2019). Neutralization breadth and potency were strongly associated (Figure 1B). Despite a higher set point VL in Thai individuals (Robb et al., 2016), there was no difference in neutralization breadth or potency between individuals from Thailand compared to East Africa (Figures 1C and S1A). Neutralization fingerprinting analysis revealed neutralizing antibodies potentially targeted multiple epitopes, with the majority predicted to target the CD4 binding site (bs) and the membrane proximal external region (MPER), independent of geographic origin (Figure 1D and Table S2). A few individuals were predicted to have multiple or evolving neutralization targets over time (Table S2). We were unable to ascribe any specificity with high confidence in 22 out of 73 individuals due to insufficient neutralization (<30% breadth). Since neutralization breadth, potency, and specificity did not differ in individuals from the two geographic regions, these neutralization data were grouped and analyzed together for the remainder of the study.

Figure 1. Longitudinal development of neutralization breadth starting from acute infection.

(A) Median VL (copies/mL, left axis) and cell counts (cells/mm³, right axis) of 73 ART-naïve participants. Days indicate the number of days since the first positive test for HIV-1 RNA. Day - 30 represents all pre-infection samples.

(B) Spearman correlation of each donor’s peak neutralization potency and breadth against the 34-virus panel, with rho and p-values indicated.

(C) Comparison of neutralization breadth of individuals from Thailand and East Africa at indicated time points. Peak neutralization represents each donor’s time point when the highest neutralization breadth was achieved.

(D) Neutralization fingerprinting was performed to delineate epitope specificity at the peak of neutralization breadth. Expanded slices contain multiple predicted epitopes.

(E) Spearman correlation of days post-viremia and neutralization breadth of each individual’s final available time point, with rho and p-values indicated.

(F) Individual plasma samples (single dots) ranked by increasing peak neutralization breadth against the 34-virus panel. Individuals who developed broad and non-broad neutralizing antibodies are represented by red and blue dots, respectively. Donors with 35–70% neutralization breadth or who had <35% breadth but were in the study for fewer than 3 years are shown in grey. 70% neutralization breadth (red) and 35% neutralization breadth (blue) are marked by dotted lines.

(G and H) Individual (thin lines) and median (thick line) longitudinal neutralization breadth of (G) broad and (H) non-broad neutralizers at specified days post-viremia. Vertical dotted black lines indicate the day of the first positive nucleic acid test. See also Figures S1–2 and Tables S1–3.

As seen with previous studies (Gray et al., 2011; Landais et al., 2016; Mikell et al., 2011; Rusert et al., 2016), neutralization breadth and potency increased with time from infection (Figures 1E and S1B–C). A conservative threshold to define ‘broad neutralizers’ was set as having the ability to neutralize >70% of the viral panel (24 out of 34 viruses with an ID₅₀>40 and no MuLV neutralization detected). Using these criteria, we identified 16 broad neutralizers out of the 73 individuals tested (22%) (Figure 1F and Table S3). Three of the 16 broad neutralizers were excluded from further analysis due to the identification of superinfection during chronic infection (Table S3), an attribute previously associated with development of neutralization breadth and likely independent of acute infection determinants (Cortez et al., 2012; Cortez et al., 2015). Twelve individuals (16%) were termed ‘non-broad neutralizers’ using the criteria that they neutralized <35% of the viral panel and participated in the study for at least 3 years (>913 days post-viremia). Non-broad neutralizers were able to neutralize a few predominantly neutralization sensitive viral strains (Figure 2). The remaining 45 individuals (62%) developed >35% neutralization breadth but did not exceed 70% at any tested time point or dropped out of the study prior to year 3 (Figure 1F and Table S1).

Figure 2. Neutralization profile of broad and non-broad neutralizers against a 34-virus panel.

^aColumn subheadings specify the pseudoviruses within their respective subtypes. Values are ID50 plasma neutralization titers; the dilution that resulted in 50% reduction of infectivity as measured by RLU in a TZM-bl neutralization assay. Titer was color-coded as follows: (yellow) 1:20–1:99, (orange) 1:100–1:499, (dark orange) 1:500–1:999, (red) 1:1000–1:5000, (bright red) >1:5000

^bNeutralization breadth was defined as the percent of the 34-virus panel with a neutralization titer (ID50) ≥40 (twice the starting dilution).

^cPotency was defined as the geometric mean titer (GMT) of a single plasma sample calculated using the ID50 of all 34 viruses. See also Figure S2 and Tables S1–3.

While chronic infection samples have been extensively evaluated for neutralization breadth (Burton et al., 2012; Kwong and Mascola, 2012; Saunders et al., 2012), few studies have characterized the dynamics of neutralization breadth starting from early acute infection (Haynes et al., 2012b; Sanchez-Merino et al., 2016; van den Kerkhof et al., 2014). Therefore, longitudinal plasma was tested for bNAbs at month 1 (day 30–43), month 3 (day 68–133), month 6 (day 168–268), year 1 (day 376–541), year 2 (day 612–868), and year 3 (day 917–1262) (Figures 1G and H and Figure 3A). Because neutralization of heterologous viruses is rare during early acute infection, plasma from time points prior to year 1 (months 1, 3, & 6) were tested against a panel of 15 Env-pseudoviruses, a subset of the 34-virus panel (Table S4). Neutralization breadth and potency modestly increased starting at month 1 and fluctuated over time (Figure 1G and H and Figure S2A and B). Non-broad neutralizers showed little variation compared to the more dynamic neutralization profiles of broad neutralizers (Figure S2A and B). Five of the broad neutralizers showed an unexpected pattern: neutralization reached >70% breadth using the 34-virus panel and then decreased by >25% at a subsequent time point (Figure S2A, donors 20337, 30112, 30156, 30196, & 40123), signifying neutralization breadth and potency are dynamic responses and may not necessarily be maintained. Interestingly, when neutralization breadth narrowed (Figure S2A), the predicted neutralization specificity shifted in these donors (Table S2), potentially highlighting the interplay between viral escape and antibody evolution over time. Cross-neutralizing antibodies were detected as early as month 1, with broad neutralizers exhibiting statistically greater breadth and potency as early as month 3 compared to non-broad neutralizers (Figures 3A–C). Because differences in the development of neutralization breadth were seen as early as month 3, these data suggest that factors prior to month 3 may have influenced the development of broad neutralizing antibodies.

Figure 3. Early onset of neutralization in individuals who develop neutralization breadth.

(A) Heat map of neutralization between broad and non-broad neutralizers starting at month 1. Plasma was tested at longitudinal time points using the 15-virus panel at months 1–6, and against the 34-virus panel at years 1–3 post-viremia. Day ranges following first detectable HIV-1 RNA are shown in parentheses. Patients (sample ID) who were lost to follow up (LFU) or went on anti-retroviral treatment (On ART) are noted in addition to time points with no available sample (NS) or if background MuLV neutralization was detected (HB).

(B) Neutralization breadth and (C) potency of broad (red) and non-broad (blue) neutralizers. Lines indicating 70% and 35% neutralization breadth are shown by red and blue dotted lines, respectively. The limit of detection of the assay at a reciprocal titer of 20 (black dotted line) is shown in (C). The median, range, and upper and lower quartile of values within each group are indicated by the boxplots and p-values were calculated by Mann-Whitney. See also Table S4.

The number of peripheral B cells during acute HIV infection predict the development of broadly neutralizing antibodies

To determine if factors within acute infection associate with the generation of neutralization breadth, longitudinal VL and absolute cell counts of broad and non-broad neutralizers were analyzed. Absolute cell counts prior to HIV infection were not obtained; therefore, only post-infection cell counts were compared between these two groups. Evaluation of cell counts and VL as a continuum via Cox proportional hazard model analysis revealed that longitudinal peripheral reduction of B cells significantly associated with the likelihood of developing bNAbs (change per 200 cells/mm³ hazard ratio (HR)=0.37, 95% CI: 0.18–0.76, p=0.007) (Figure 4A and B), while a trend was observed for reduced CD4⁺ T cells and NK cells (p=0.076 and 0.052, respectively) (Figure S3A and B). Incorporating VL into a multivariate Cox proportional hazard model showed only B cells associated with neutralization breadth independent of VL (HR=0.28, 95% CI: 0.11–0.72, p=0.009).

Figure 4. The number of peripheral B cells during acute HIV infection predict the development of broadly neutralizing antibodies.

(A) Spline models of median (solid lines) absolute B cells counts (cells/mm³) are shown longitudinally for broad (red) and non-broad (blue) neutralizers. Shaded areas denote upper and lower 95% confidence intervals.

(B) Evaluation of B cell counts between broad (red) and non-broad (blue) neutralizers starting at day 2 through year 3 post-viremia and at the time point of peak neutralization breadth for each individual (chronic, range: day 376–2115). Dashed line indicates 160 B cells/mm³. The median, range, and upper and lower quartile of values within each group are indicated by the boxplots. Significant p-values by Mann-Whitney t-tests are shown.

(C) Forest plot indicating the number of B cells within the lower quantile is predictive of neutralization breadth at month1 post-viremia in broad neutralizers. Odds ratios (OR, solid symbols) and 95% confidence intervals (error bars) are indicated with the dashed line at 1 to indicate the cut-off for significance.

(D) Logistic regression of lower B cell quantiles with neutralization breadth. Odds ratios (OR), p-values, upper (U95), and lower (L95) 95% confidence intervals, and likelihood of neutralization breadth are indicated. Shaded boxes indicate p-values <0.05. See also Figure S3.

Further investigation using logistic regression at specific acute infection time points revealed that reduced peripheral B cell counts as early as month 1 were predictive of development of neutralization breadth (change per 200 cells/mm³ odds ratio (OR)=0.071, 95% CI: 0.004–0.478, p=0.02) and independent of VL when the covariate was incorporated into the model (OR=0.077, 95% CI: 0.005–0.534, p=0.03). Distinctions in B cell counts between groups were observed starting at month 1 using univariate Mann-Whitney t tests (Figure 4B). Sporadic differences between groups were observed for CD4⁺ and NK cell counts at month 6, but not consistently across all time points, as was observed for B cell counts (Figures S3B).

Within these individuals, VL measured at longitudinal time points was not significantly higher in broad neutralizers using univariate or multivariate approaches (Figure S3A and B). Neutralization breadth modestly directly correlated with peak VL (median 14 days) (Figure S3C). No significant correlation was observed with set point VL or VL at concurrent time points where neutralization breadth was observed (Figure S3C).

Since reduced peripheral B cell counts were observed in broad neutralizers at month 1 post-viremia, we next determined the peripheral B cell count that was predictive of neutralization breadth, as this information could be useful in predicting who would elicit broad neutralizing antibodies in future studies where ART is initiated early in infection. Forest plots generated from logistic regression of broad neutralizers, and all other participants who remained in the study for a minimum of three years, revealed participants with B cell counts within the lower 50% quantile were predicted to develop neutralization breadth (Figures 4C and D). Similar results were obtained when groups were divided into quartiles. The most significant result showed that individuals with fewer than 160 B cells/mm³ (lower 35% quantile) were 42 times more likely to develop neutralization breadth (Figure 4D). These data indicate a reduction in the peripheral B cell compartment starting at month 1 (30–43 days) post-viremia is predictive of the development of broadly neutralizing antibodies.

Greater changes in peripheral B cell subsets occur in broad neutralizers throughout acute infection

In an effort to understand the mechanism prompting B cell diminution in broad neutralizers as well as evaluating changes in the frequency of B cell subpopulations that may contribute to development of neutralization breadth, flow cytometry was performed on peripheral blood mononuclear cell (PBMC) samples at pre-infection, day 14 (peak VL), month 1, and chronic (peak neutralization breadth) time points. B cell subsets within the peripheral blood were defined by the expression of CD20, CD10, CD21, and CD27, on CD19⁺ B cells by increasing levels of differentiation: immature transitional (CD10⁺/CD27⁻), naïve (CD27⁻/CD21^hi), TLM (CD27⁻/CD21^lo), resting memory (CD27⁺/CD21^hi), activated memory (CD27⁺/CD21^lo), and plasmablasts, terminally differentiated B cells that secrete immunoglobulin, (CD27⁺⁺/CD20⁻/CD21^lo) as described previously (Moir et al., 2010) (Figure S4A–C). Integrin β7 was included to glean insight into B cell trafficking, as this has been shown to be a marker of homing to the GALT (Bargatze et al., 1995; Berlin et al., 1995). Thirty-seven founder envelope gene sequences were used to produce autologous founder gp140 Env proteins for each individual. As such, each flow panel was individualized for each participant with the inclusion of their respective founder gp140 Env glycoprotein(s) to capture the early autologous antigen-specific B cell responses.

The contraction and expansion of B cell subpopulations were found throughout HIV infection, as seen previously, but these changes in B cell frequencies differed between broad and non-broad neutralizers (Moir et al., 2010; Moir and Fauci, 2009, 2017) (Figure 5A and B). Compared to pre-infection, HIV infection in broad neutralizers triggered significant alterations of all B cell subpopulations throughout acute infection (Figure 5A and C), whereas fewer significant changes were detected in non-broad neutralizers (Figure 5B and D). In both groups, there was a contraction of resting memory (CD27⁺/CD21^hi) B cells, while activated memory (CD27⁺/CD21^lo), TLM (CD27⁻/CD21^lo), and plasmablast (CD27⁺⁺/CD20⁻/CD21^lo) populations increased by day 14 (Figure 5A–D). B cell frequencies at month 1 post-viremia within the broad neutralizers continued to contract or expand relative to pre-infection levels, while few changes were occurring in the non-broad neutralizers at the same time point (Figure 5A–D). Frequencies of naïve B cells (CD27⁻/CD21^hi) and integrin β7⁺ (CD19⁺/integrin β7⁺) B cells, a gut homing marker, were significantly reduced and amplified, respectively, in broad neutralizers, but the abundance of these subpopulations in non-broad neutralizers remained unaltered after infection (Figure 5A–D). Peripheral levels of CXCL13, previously shown to be a marker of GC activity (Havenar-Daughton et al., 2016b), did not significantly differ between broad and non-broad neutralizers until chronic infection when neutralization breadth was observed (Figure S5A). Broad neutralizers also sustained greater TLM (CD27⁻/CD21^lo) B cells at day 14 and month 1 and plasmablast (CD27⁺⁺/CD20⁻/CD21^lo) frequencies at month 1 compared to non-broad neutralizers (Figure 5E). Because plasmablast populations can also be defined using CD38 (Doria-Rose et al., 2009; Yu et al., 2016), we also analyzed the plasmablast population defined as CD19⁺CD27⁺CD38⁺, which yielded similar population frequencies (Figure S5B). By chronic infection, the B cell subpopulation frequencies were similar between broad and non-broad neutralizers (Figure 5E).

Figure 5. Greater changes in peripheral B cell subsets occur in broad neutralizers throughout acute infection.

(A and B) Contraction and expansion of B cell subsets represented by radar plots of median normalized scores of B cell subpopulation frequencies of (A) broad and (B) non-broad neutralizers. Colored lines represent the indicated time points pre- and post-infection, with the subpopulation indicated outside of the radar plot. Scores were calculated by standardizing each subpopulation observation across all time points.

(C and D) Heat map of p-values by Wilcoxon signed-rank test analyzing B cell subset frequency changes at post-infection time points compared to pre-infection in (C) broad and (D) non-broad neutralizers. Significant p-values were color-coded as follows: (yellow) p<0.05, p>0.01, and (orange) p<0.01.

(E) Comparison of B cell subset frequencies at specified time points between broad (red) and non-broad (blue) neutralizers. The median, range, and upper and lower quartile of values within each group are indicated by the boxplots. Reported p-values were generated by Mann-Whitney t tests. See also Figure S4.

Longitudinal total IgG was assessed to determine if extensive B cell dysfunction resulting in hypergammaglobulinemia was associated with neutralization breadth. Higher concentrations of total IgG were found in broad neutralizers during chronic infection with trends at pre-HIV and day 14 (Figure S5C). However, when IgG concentrations were normalized to individual pre-infection levels, no significant difference was found between broad and non-broad neutralizers (Figure S5D), suggesting the extent of B cell dysfunction in acute HIV infection, as measured by IgG concentration, was not associated with subsequent neutralization breadth.

Reduction in peripheral naïve B cells associates with B cell migration, activation and differentiation

We next wanted to ascertain whether certain B cell subpopulations during acute infection associated with the development of neutralization breadth, similar to the absolute peripheral B cell counts (Figure 4). To understand the cause of the reduced peripheral absolute B cell counts, subpopulation frequencies were applied to available B cell counts (Figure 6A and B). As seen previously (Moir et al., 2010; Moir and Fauci, 2009), the majority of CD19⁺ B cells consisted of naïve (CD27⁻/CD21^hi) and resting memory (CD27⁺/CD21^hi) subpopulations (Figure 6A and B). Similar to absolute B cell counts (Figure 4A and B), significantly reduced naïve (CD27⁻/CD21^hi) B cells were observed within the broad neutralizers at all time points tested (Figure 6C). The resting memory B cell subpopulation (CD27⁺/CD21^hi) was also significantly lower in broad neutralizers during chronic infection with trends at day 14 and month 1 (Figure S5E). Logistic regression revealed the development of neutralization breadth significantly associated with a reduction in peripheral naïve B cells (CD27⁻/CD21^hi) as early as day 14, month 1, and chronic time points (OR=0.63, 95% CI: 0.37–0.92, p=0.036; OR=0.64, 95% CI: 0.36–0.89, p=0.037; OR=0.53, 95% CI: 0.28–0.81, p=0.016, respectively). Only naïve (CD27⁻/CD21^hi) populations directly correlated with absolute B cell counts starting at day 14 (Figure 6D), indicating a reduction of peripheral naïve B cells (CD27⁻/CD21^hi) was the main driver responsible for the reduction of total peripheral B cell counts in the broad neutralizers.

Figure 6. Reduction in peripheral naïve B cells associates with B cell migration, activation, and differentiation.

(A and B) Pie charts of absolute B cell subsets at day 14 in (A) broad and (B) non-broad neutralizers.

(C) Comparison of peripheral absolute naïve B cells in broad and non-broad neutralizers at indicated longitudinal time points following initial viremia. Chronic represents each donor’s time point when the peak neutralization breadth was achieved. Median B cell counts (cells/mm³) with min to max intervals are shown with p-values calculated by Mann-Whitney t tests.

(D) Spearman correlation comparing frequencies of day 14 peripheral naïve B cells to day 14 peripheral absolute B cell counts.

(E) Spearman correlations comparing frequencies of naïve B cells at month 1 to frequencies of month 1 activated memory, resting memory, tissue-like memory (TLM), plasmablasts, integrin β7⁺, and founder gp140⁺ B cells of broad (red) and non-broad (blue) neutralizers. Rho and p-values are shown for each correlation. See also Figure S5.

By month 1, naïve B cell (CD27⁻/CD21^hi) frequencies significantly inversely correlated with frequencies of other B cell subpopulations including activated memory (CD27⁺/CD21^lo), resting memory (CD27⁺/CD21^hi), TLM (CD27⁻/CD21^lo), integrin β7⁺ (CD19⁺/integrin β7⁺), plasmablasts (CD27⁺⁺/CD20⁻/CD21^lo), and frequencies of founder-Env-specific B cells (CD19⁺/gp140⁺) (Figure 6E and Figure S5F), indicating that naïve B cells were engaging founder envelope glycoproteins and undergoing activation, migration and differentiation.

The initial interaction between founder Env and naïve B cells is important for the development of neutralization breadth

To further understand whether initial autologous founder Env interactions with B cells are important for development of subsequent neutralization breadth, we categorized the frequencies of founder Env-specific B cells (CD19⁺/gp140⁺) into tertiles (low, medium, and high) using cumulative incidence to determine the likelihood of developing neutralization breadth. Individuals with medium or high frequencies of autologous founder Env-specific B cells (CD19⁺/gp140⁺) were significantly more likely to develop neutralization breadth compared to individuals with low frequencies (Figure 7A). In addition, the frequencies of month 1 autologous founder Env-specific (CD19⁺/gp140⁺) B cells positively associated with peak neutralization breadth (Figure S5G), and were predictive of the development of bNAbs in logistic regression analysis (OR=1.13, 95% CI: 1.02–1.28, p=0.035). These data demonstrate that early B cell engagement of the autologous founder gp140 Env within 1 month of infection was predictive of the development of broadly neutralizing antibodies years later.

Figure 7. The initial interaction between founder Env and naive B cells in acute infection is important for the development of neutralization breadth.

(A) Cumulative incidence curves indicating the probability of developing neutralization breadth using high, medium, or low frequencies of founder Env-specific B cells 1 month (day 30–43) following initial viremia. Reported p-value by LogRank test.

(B) Comparison of frequencies of founder Env-specific B cells between broad (red) and non-broad (blue) neutralizers at month 1. The median, range, and upper and lower quartile of values within each group are indicated by the boxplots and p-values by Mann-Whitney t tests.

(C) Biplot of unsupervised Principal Component Analysis (PCA) of founder Env⁺ B cell phenotypes at month 1. Each dot represents a broad (red) or non-broad (blue) neutralizer plotted in 2 dimensions using their projections on the first 2 principal components (PC). Arrows point in the direction of B cell subsets accounting for the variability observed between groups as projected onto the 2 dimensions of the biplot.

(D) Cumulative incidence curves indicating the probability of developing neutralization breadth using high, medium, or low frequencies of founder Env-specific naïve B cells 1 month following initial viremia. Reported p-value by LogRank test.

(E) ROC curve using the B cells frequencies of founder Env-specific B cell subsets at month 1 for the prediction of neutralization breadth. Area under the Curve (AUC) for founder Env-specific B cells and founder-Env B cell subsets is shown for those reaching a predictive power over 0.70.

(F) IgM binding antibodies against respective autologous founder Env gp140 at indicated longitudinal time points.

(G and H) Antibody binding to (G) BG505 or (H) ZM106.9 SOSIP constructs at longitudinal time points.

(I) B cell receptor (BCR) IgM and IgG isotype usage of founder-Env specific B cells at month 1 per individual.

(J) IgM and IgG heavy chain germline usage of founder Env-specific B cells at month 1. Heavy chains that constitute 50% of heavy chain usage for each isotype are shown. Heavy chains that account for fewer than 2% of the germline usage are grouped together as “other”. See also Figures S6–S7 and Table S5.

Similar to the total peripheral B cell subsets as described above, B cell subpopulations of founder-Env specific CD19⁺ B cells were analyzed to determine if there were phenotypic differences in early antigen-specific B cells between groups. Autologous founder-Env specific B cell subpopulations were analyzed as described above, and the frequencies of antigen-specific IgM⁺ and IgG⁺ B cells were included to determine the impact of these early isotypes in bNAb development. Expansion of founder-Env specific B cell subpopulations was evident in both broad and non-broad neutralizers (Figure S6A–D). Minimal differences in Env-specific B cells were found at day 14, but by month 1, HIV infection of broad neutralizers triggered significant expansion of founder-Env specific B cells with higher frequencies of naïve (gp140⁺/CD27⁻/CD21^hi), activated memory (gp140⁺/CD27⁺/CD21^lo), TLM (gp140⁺/CD27⁻/CD21^l°), and IgM⁺(CD19⁺/gp140⁺/IgM⁺) B cells compared to non-broad neutralizers (Figure 7B and Figure S6E–F). An unsupervised Principal Component Analysis (PCA) was performed to determine which founder Env-specific B cell subsets at month 1 best explain the variance between broad and non-broad neutralizers. The distribution of individual donors showed that most of the variance between groups was due to high founder Env-specific naïve (gp140⁺/CD27⁻/CD21^hi) and IgM⁺(CD19⁺/gp140⁺/IgM⁺) B cell frequencies within the broad neutralizers (Figure 7C). When categorized into tertiles using cumulative incidence, individuals with medium or high frequencies of founder Env-specific naïve B cells were significantly more likely to develop breadth compared to individuals with low founder-specific naïve B cell frequencies (Figure 7D). In addition, logistic regression analysis revealed that higher frequencies of founder Env-specific naïve B cells were significantly associated with and predictive of the development of breadth (OR=4.14, 95% CI: 1.56–16.68, p=0.017), with the frequency of founder-Env⁺ IgM⁺ (IgM⁺/gp140⁺/CD19⁺) and founder-Env⁺ TLM (gp140⁺/CD27⁻/CD21^l°) B cells trending (OR=1.28, 95% CI: 1.02–1.76, p=0.06 and OR=6.48, 95% CI: 1.63–72.25, p=0.05, respectively). Receiver operating characteristic (ROC) plots showed that month 1 founder Env-specific naïve B cell frequencies were effective at distinguishing broad and non-broad neutralizers, achieving a positive predictive value (PPV) of 0.85 and an area under the curve (AUC) of 0.78 (Figure 7E). Founder Env-specific B cells (CD19⁺/gp140⁺), founder-Env TLM (gp140⁺/CD27⁻/CD21^lo), IgM⁺ (IgM⁺/gp140⁺/CD19⁺), and activated memory (gp140⁺/CD27⁺/CD21^lo) subpopulations were also able to discriminate between these groups AUC of >0.74 (Figure 7E). Combined, this data provides evidence that the initial interaction between founder Env and naïve B cells is important for the development of neutralization breadth.

In an effort to understand if B cell engagement of founder Env translated into secreted antibody responses, longitudinal plasma samples were measured for IgM, IgA, and IgG binding antibodies to each individual’s unique autologous founder gp140 envelope glycoprotein. Broad neutralizers expressed elevated IgM responses to autologous founder Env by day 14 with a trend at month 1 compared to non-broad neutralizers (Figure 7F). Evaluation using a Cox proportional hazard model via a continuum revealed that heightened longitudinal autologous Env IgM responses significantly associated with the development of neutralization breadth (change per 1 standard deviation of signal:noise HR=1.52, 95% CI: 1.174–1.958, p=0.001). Conversely, IgA and IgG responses to autologous founder gp140 Env were similar between the groups (Figure S6G and H). Longitudinal plasma were also tested for neutralization against pseudoviruses expressing autologous founder envelopes. Autologous neutralization to the founder Env was detected by month 3, with no observed difference in autologous IgM, IgG, or IgA binding or neutralization between groups at that time point (Figure S6I and Table S5).

Cross-reactive binding antibody responses were also analyzed by measuring plasma IgM, IgA, and IgG responses to consensus gp140s spanning 7 HIV subtypes as used previously (Haynes et al., 2012a; Liu et al., 2013). A trend was observed for higher heterologous IgM binding responses at day 14, but overall IgM, IgG, and IgA cross-binding responses to heterologous gp140 Envs were similar between groups (Figure S6J). Because HIV envelopes expressed as gp140 proteins do not adequately reflect binding antibody responses to native-like trimers or bind to trimeric-specific broadly neutralizing monoclonal antibodies (Sanders et al., 2013; Yasmeen et al., 2014), plasma antibody binding to BG505 and ZM106.9 Env SOSIPs were also evaluated to assess differences in binding antibodies to trimeric epitopes. Broad neutralizers had a significantly higher magnitude of antibodies binding to BG505 and ZM106.9 SOSIP by month 3 (Figure 7G–H), suggesting the potential of early specificity to trimeric-dependent epitopes.

Finally, since the initial interaction between the founder Env and naïve B cells was found to be important for the development of neutralization breadth, we sequenced the heavy chain of the founder antigen-specific B cells at month 1 to determine if there was a bias in gene usage of the early founder-Env specific B cell receptors. No significant difference was observed between the length of CDRH3 or somatic hypermutation (SHM) between broad and non-broad neutralizers (Figure S7A and B) at month 1 post-viremia. In agreement with higher frequency of IgM⁺ B cells (Figure 7B), a significantly higher number of IgM heavy chains were obtained from broad neutralizers compared to non-broad neutralizers (Figure 7I). Heavy chain gene usage of IGHV 4–34, 4–39, and 3–23 dominated both founder-specific IgM⁺ and IgG⁺ B cells (Figure 7J), with no significant difference in heavy chain gene usage or BCR diversity of antigen-specific cells between broad and non-broad neutralizers (Figure S7 C–E). At month 1 post-infection, there was also no evidence of clonal expansion in either group. Taken together, this data indicates that the initial engagement of the HIV Env with B cells is not biased towards specific heavy chain gene usage in broad neutralizers compared to non-broad neutralizers.

DISCUSSION

The generation of bNAbs is a pivotal, but as of yet, unachievable component of HIV-1 vaccine design. Gathering information from HIV-1 infected individuals that produce bNAbs can provide valuable insight into mechanistic aspects of the immune system that initiate or guide bNAb development (Burton et al., 2012; Haynes et al., 2012b; Kwong and Mascola, 2012; Mikell et al., 2011; Saunders et al., 2012). In the present study, we investigated acute infection factors that were found to be important in the development of neutralization breadth in the prospective RV217 cohort that included samples prior to, and within the first days, months, and years following HIV-1 infection.

Using these longitudinal samples, we stratified groups into ‘broad’ and ‘non-broad’ neutralizers and used these groups to identify factors within acute infection that associated with development of bNAbs. By first assessing clinical information, longitudinal analyses revealed a reduction in peripheral B cell counts significantly associated with development of bNAbs. Peripheral B cell depletion was previously shown using the highly efficacious 17-D yellow fever vaccine that transiently reduced peripheral naïve and memory B cells, providing evidence that peripheral B cell decrement may be associated with desired functional antibody responses (Kohler et al., 2012). Reduced peripheral absolute B cell counts were driven by a reduction of peripheral naïve B cells, a population which is significantly inversely associated with and predictive of the development of bNAbs as early as day 14. An influx of B cells into germinal centers within the lymph nodes may have caused the reduction of B cell counts in the periphery, a prospect which may be beneficial to the development of neutralization breadth. As seen previously in vaccinated macaques who develop neutralizing antibodies compared to macaques that do not develop neutralizing antibodies, higher frequencies of GC B cells were observed in lymph nodes (Havenar-Daughton et al., 2016a). Because our analysis is based on peripheral cell counts and lymph node samples were not obtained within this cohort, we are unable to determine if lymph node or GALT B cell counts differ between broad and non-broad neutralizers. Individuals with low B cells counts (fewer than 160 B cells/mm³) at month 1 had the highest probability of achieving neutralization breadth (42 times more likely), and this parameter could be used to identify potential broad neutralizers in other acute infection cohorts where lymph node biopsies or other tissue samples may be available (Ananworanich et al., 2012; Byrne et al., 2016).

Despite the overall reduction of absolute B cells, broad neutralizers had higher frequencies of autologous founder Env-specific B cells, which were predictive of the development of bNAbs. The association of the naïve B cell subpopulation with B cell memory and plasmablast subpopulations suggests naïve B cells engaged the founder HIV Env and differentiated into memory and antibody-secreting cells (Moir and Fauci, 2009, 2017). The significantly higher frequencies of gp140-specific naïve, activated memory, and TLM populations, as well as the dynamic changes within the overall B cell subpopulations during the first month of infection provides evidence that B cells of broad neutralizers were more responsive to founder engagement, activation, and terminal differentiation relative to the non-broad neutralizers. Higher engagement of founder Env with naïve B cells led to heightened secretion of IgM binging antibodies by day 14 post-viremia. Taken together, this study suggests early, improved engagement of the HIV-1 Env led to a robust B cell responses favoring the development of neutralization breadth.

Using the founder gp140 Envs tailored for each individual allowed us to identify donor-specific founder-positive B cells, potentially increasing the sensitivity of these initial interactions. However, even though these founder gp140 Envs are a mix of monomers, dimers, and trimers (Wieczorek et al., 2015), they may not have captured the full spectrum of antigen-positive B cells as they were not conformationally constrained. Analysis of SOSIP Envs provided evidence that broad neutralizers were able to elicit elevated binding antibodies to SOSIP proteins, which may indicate that the Env of founder viruses of broad neutralizers have unique features that drive the development of neutralization breadth as seen with other studies (Kouyos et al., 2018). We recently reported in detail the development of MPER-directed bNAbs from one individual within the current study, and found that it was the founder Env displaying MPER that initiated the B cell lineage imparting neutralization breadth (Krebs et al., 2019). Analysis of additional founder Env sequences and their biophysical properties are needed to determine whether the founder Envs of the broad neutralizers possess attributes favoring the initiation of B cell lineages that lead to bNAbs.

In contrast viral features, broad neutralizers may possess B cell features that favor better engagement of the founder Envs. Recent studies showed that maintaining some auto- or poly-reactive memory B cell clones may be beneficial (Burnett et al., 2018), as memory B cells with autoreactive B cell receptors continue to circulate and undergo somatic hypermutation, which may abrogate their autoreactivity and allow the B cell to increase affinity and subsequent neutralization of its target (Burnett et al., 2018). Higher frequencies of autoreactivity have been shown to be present in HIV-1 infected individuals with bNAbs (Moody et al., 2016). Sequencing the heavy chains of founder Env-specific B cells at month 1 did not reveal differences between groups in frequency or diversity of gene usage, although combined, HIV Envs from both groups favored binding to 4–34, 4–39, and 3–23. Further exploring BCR diversity overtime is warranted in follow-up studies to determine their effect on the development of broadly neutralizing antibodies.

Since viral load has previously been shown to be associated with the development of neutralization breadth in other studies (Cortez et al., 2012; Cortez et al., 2015; Gray et al., 2011; Landais et al., 2016; Mikell et al., 2011; Piantadosi et al., 2009; Rusert et al., 2016), the lack of association of viral load with neutralization breadth in this cohort was somewhat surprising. The significant reduction of B cell counts with neutralization breadth was independent of viral load, which may point towards a greater complexity than only antigenic load stimulation resulting in neutralization breadth, and instead may reflect viral diversity or B cell engagement of conserved epitopes across strains. Further exploring the phenotypes of NK cells and CD4 T cells, such as follicular helper (TFH)-like CD4T cells, is warranted in follow-up studies to determine their effect on the development of broadly neutralizing antibodies.

These data demonstrate development of neutralization breadth is influenced by factors within the first month of acute infection. Early, heightened engagement of the founder Env favored the development of broadly neutralizing antibodies, providing evidence that neutralization breadth is shaped by robust B cell responses within the first stages of infection. The limitations of this study include a relatively small number of individuals within each group, and a lack of lymph node samples at early time points to glean further insight into mechanisms within germinal centers. The benefit of this prospective cohort allowed the opportunity to evaluate responses prior to and early following initial viremia. In conclusion, the results outlined within this study offer insights into the early acute infection events that lead to downstream functional outcomes, and underline the importance of the initial interaction of the viral Envelope glycoprotein with naïve B cells.

STAR Methods

RESOURCE AVAILABILITY

Lead Contact

Further information and reasonable requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Shelly J. Krebs (skrebs@hivresearch.org).

Materials Availability

All unique reagents generated in this study are available from the Lead Contact with a completed Material Transfer Agreement.

Data and Code Availability

Founder envelope sequences generated in this study have been deposited into GenBank under accession numbers MW443137-MW443225.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Human subjects

RV217 was an East Africa and Thailand cohort that prospectively enrolled 2276 high-risk men and women prior to HIV infection (Robb et al., 2016). The protocol was approved by appropriate local review boards and the Walter Reed Army Institute of Research, and written informed consent was obtained from all participants in their written languages. As previously described (Robb et al., 2016), men and women, 18 to 50 years of age at high-risk for HIV infection were recruited from bars, clubs, and other locations associated with transactional sex. To be eligible for study entry, participants had to meet at least one of the following four criteria within the previous 3 months: had exchanged goods for sex, had unprotected sex with a known HIV-positive partner, had unprotected sex with three or more partners, and had symptoms of a sexually transmitted infection. In the first part of the study, which involved surveillance of participants who were not infected, volunteers who had at least one of these high risk criteria underwent small-volume blood collections by fingerstick measurement twice weekly for the presence of HIV RNA and large-volume blood collections every 6 months during the surveillance phase. Small-volume blood samples were tested for HIV-1 RNA within 24 to 48 hours after collection. Following the first positive HIV-1 RNA detection (viremia), large-volume blood samples were obtained twice weekly for the first 4 weeks, and every 3 months thereafter for long-term follow-up. We followed local HIV care and treatment guidelines which did not incorporate routine treatment of acute infection until the START trial results were announced. From that point onward everyone was treated, regardless of national treatment initiation criteria. Prior to that time, we treated anyone with severe acute retroviral syndrome and did not prevent anyone from starting treatment if they and their physician chose to initiate treatment. All pregnant women started ART immediately. The first 73 treatement-naïve infected individuals enrolled in the initial RV217 study (Robb et al., 2016) who remained in the study for at least one year without treatment were tested for bNAbs at longitudinal time points throughout infection. Day −30 represented the data ≥30 days prior to detectable viremia (range 199–653 days prior to viremia, median −350). Day 2 was within the first days a blood sample was reactive for HIV-1 RNA and prior to peak VL (range 1–5 days post-initial viremia, median 2 days). Day 14 coincided with peak VL and incorporated cell data from time points within 4 days of the peak VL time point when no cell counts were available at peak VL (range 4–29 days, median 14). All other time points were as follows: month 1 (range 30–43 days, median 35), month 3 (range 68–133 days, median 90), month 6 (range 168–268 days, median 253), year 1 (range 376–541 days, median 442), year 2 (range 612–868, median 726), year 3 (range 917–1262, median 1084). The chronic time point corresponded to the day of the highest neutralization breadth of the individual against the 34-virus panel (range 170–2115, median 764). All samples were ART-naïve. Set point VL was defined as the average VL of all samples collected between days 42 and 365, as described previously (Robb et al., 2016).

Cell lines

Human embryonic kidney (HEK)-derived 293T cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA), and HeLa-derived TZM-bl reporter cells were acquired through the NIH AIDS Reagent Program (ARP, Bethesda, MD). HEK293T and TZM-bl cells were maintained in complete Dulbecco’s Modified Eagle Medium (herein referred to as DMEM) containing high glucose Dulbecco’s Modified Eagle Medium (DMEM, Thermo Fisher, Waltham, MA), 1X Penicillin-Streptomycin (Pen Strep, Thermo Fisher, Waltham, MA) and 10% fetal bovine serum (FBS, Gemini Bio Products, West Sacramento, CA) at 37°C/5% CO₂. Expi293F cells, obtained from Thermo Fisher, were maintained in Expi293 Expression Medium (Thermo Fisher, Waltham, MA), at 37°C/10% CO₂ with shaking at 120 RPM.

METHOD DETAILS

Neutralization

The presence of neutralizing antibodies was assessed in longitudinal samples using the well-described TZM-bl neutralization assay that included a 29-virus reference panel of full-length env of viruses previously selected to map epitope specificity of serum antibodies (Doria-Rose et al., 2017; Georgiev et al., 2013). This original reference panel was dominated by pseudoviruses encoding subtype B envelopes and was thus expanded to a 34-virus panel to increase subtype diversity. HIV env DNA constructs were obtained from the NIH ARP, James Binley (Torrey Pines Institute), David Montefiori (Duke University Medical Center), Dana Gabuzda (Dana Farber Research Center), and Sodsai Tovanabutra (MHRP). Similarly, 37 pseudovirues encoding autologous envelopes were produced to measure autologous neutralization longitudinally in plasma from respective donors. Env-pseudoviruses capable of a single round of infection were produced by co-transfection of HEK-293T cells with plasmids expressing env in the pSG3ΔEnv backbone. Pseudovirus stocks were titrated by luciferase reporter gene expression in TZM-bl cells as previously described (Li et al., 2005) by targeting 100,000 relative light units (RLU) for each pseudovirus after a 48-hour incubation period. 178 samples from 73 HIV-1-infected individuals were evaluated for bNAbs starting 30 days post-viremia and up to 6 years. Using a robotic microneutralization assay (Krebs et al., 2019), samples were analyzed using 4-fold dilution of plasma starting at 1:20. Neutralization was measured by reduction of luciferase gene expression as previously described (Montefiori, 2005). Briefly, indicator virus was incubated with serial 4-fold dilutions of plasma samples in duplicate before being added to TZM-bl cells. After a 48-hour incubation, luciferase activity was measured using Britelite Plus Reporter Gene Assay System (Perkin Elmer, Coraopolis, PA) substrate solution. Neutralization activity is represented as the reciprocal plasma dilution that resulted in 50% reduction (ID₅₀ or IC₅₀) of RLU. Positive neutralization was defined as 50% inhibition of infection of an HIV strain at ≥1:40 plasma dilution, 2-fold higher than the limit of the assay, with no murine leukemia virus (MuLV) neutralization detected at 1:20 plasma dilution. Breadth was calculated as the percentage of the 15 or 34 virus panel neutralized by ≥1:40 plasma dilution. A more stringent ID50 cut-off of 100 was also used instead of 1:40, and 15 out of th e16 individuals would still be classified as being ‘broad neutralizers’, with no change in the results. Potency was calculated as the geometric mean titer (GMT) of plasma against the entire 15- or 34-virus panel.

Neutralization Fingerprinting

Epitope delineation by fingerprinting analysis was performed as previously described (Doria-Rose et al., 2017; Georgiev et al., 2013; Raju et al., 2019). Neutralization breadth below 30% against the panels were omitted from analysis. A cutoff of ≥0.25 predicted relative prevalence (coefficient) was used as a threshold to determine epitope specificity and lower residual score (<0.1) and median of scores (<0.06) values associated with a higher confidence for the given prediction. Predictions were confirmed via the use of HIV-2 MPER chimera and peptide competition mapping experiments as described previously (Krebs et al., 2019). Briefly, plasma neutralization was tested using HIV-2/HIV-1 chimeras with modified MPER regions and used in neutralization assays as described above.

Envelope glycoprotein production

Uncleaved founder and consensus gp140 proteins were produced from founder env sequences identified by SGA (Kijak et al., 2017). These gp140’s were a mixture of monomers, dimers and trimers (Wieczorek et al., 2015), and incorporated the native leader peptide, an R to S mutation in the gp120/gp41 cleavage site, and full MPER sequences followed by a short GGGS linker sequence and a C-terminal AviTag. Sequences, codon-optimized for expression in human cells, were synthesized (Genscript, Piscataway, NJ) and cloned into a custom pcDNA3.4 expression vector (Thermo Fisher, Waltham, MA) using sequence- and ligation-independent cloning (SLIC) methods. Proteins were expressed by transient transfection in Expi293 cells (Thermo Fisher, Waltham, MA) per manufacturer’s instructions. Gp140 proteins were purified from clarified cell culture supernatants 4 days post-transfection using Galanthus nivalis lectin (GNL) resin (Vector Labs, Burlingame, CA) affinity chromatography. When needed, proteins were further passed through a Q-sepharose (GE Healthcare, Chicago, IL) column to remove host cell protein contaminants. All gp140s were purified to 90% purity or higher, as assessed by SDS-PAGE and Coomassie staining in reducing and non-reducing conditions. An additional post-purification control step included binding to CD4-Ig, as measured by bio-layer interferometry, to insure proper gp140 folding.

Flow cytometry

Cryopreserved peripheral blood mononuclear cells (PBMCs) from blood collected from volunteers were stained with the following anti-human antibodies: CD3 (BV510), CD16 (BV510), CD4 (BV510), IgM (Cy5PE), IgG (BV421), and CD21 (phycoerythrin-Cy7) (BD Biosciences, San Jose, CA); CD19 (ECD) (Beckman Coulter, Atlanta, GA); integrin β7 (fluorescein isothiocyanate), CD14 (BV510), CD56 (BV510), CD8 (BV510), CD20 (BV570), CD27 (BV605), and CD10 (BV711) (BioLegend, San Diego, CA). Aqua live/dead stain (Thermo Fisher, Waltham, MA) was added for viability discrimination. Antigen-specific B cells were sequentially stained using biotinylated founder gp140 probes that were added to the PBMC/antibody mixture before excess streptavidin labeled with phycoerythrin was added. Some individuals were infected with multiple transmitted viruses compared to others. Since we did not want to bias our approach, we produced and included all gp140 founder envelope glycoproteins from individuals infected with multiple founders, thus recreating the initial probable interactions of the B cells with the HIV envelope glycoproteins within these donors. The total amount of envelope glycoprotein included within the flow cytometry phenotyping experiment was equivalent across all donors, whether they were infected with multiple or single viruses. Data was collected on a FACSAria II (BD Biosciences, San Jose, CA) and analyzed using FlowJo v.9.9.6 (TreeStar). Pre-infection time points and uninfected donors were used as baseline comparators to infected samples. The gating strategy is shown in Figure S4. Founder gp140⁺-specific cells at month 1 were sorted into TCL buffer (Qiagen) with 5% BME prior to BCR sequencing.

BCR sequencing

Month 1 antigen-specific B cells from 9 broad neutralizers and 7 non-broad neutralizers were sorted and lysed. The B cell lysates from each donor were pooled after adding 2.5X volumes of NucleoZOL and processed using the NucleoSpin RNA Set for NucleoZOL kit (TaKaRa Bio, Inc; Mountain View, CA) per manufacturer’s instructions. Purified RNA was concentrated using NucleoSpin RNA Clean-up XS columns, eluted in RNase-free water, then reverse-transcribed to full-length cDNA and amplified with the SMART-Seq v4 Ultra Low Input RNA kit (TaKaRa) per manufacturer’s instructions.

Antibody V(D)J regions were amplified from cDNA using a combination of previously published VH (pool)+IgG-or-IgM and SMART+IgG-or-IgM primer sets. Primer sequences are composites of Illumina Adaptors (see STAR methods, underlined) and portions complementary to the VH chain (Briney et al., 2016) or 5’ PCR Primer II A (SMART primer) sequence. Briefly 10 ul reaction volumes consisted of 2 ul 5X PrimeSTAR GXL buffer, 1.6 ul of 1.25 mM dNTPs, 0.2 ul of each 10 μM primer or primer pool, 0.2 ul PrimeSTAR GXL DNA polymerase (TaKaRa), and 2 ng of cDNA. cDNA was amplified as follows: initial denaturation at 94°C, 2 min; then 30 cycles of 98°C, 10 sec / 58°C, 15 sec / 68°C 1 min 30 sec; then 68°C, 5 min; and a 4°C hold. One ul of amplified product was screened on a 1% agarose gel. PCR product was purified using AMPure XP beads (Beckman Coulter, Atlanta, GA) and subjected to a second round of nested amplification using the high fidelity 2X KAPA HiFi Hotstart enzyme (Kapa Biosystems) and index primers from the Nextera XT Index Kit v2 (Illumina) per respective manufacturer’s instructions but with the following PCR parameters: initial denaturation at 95°C, 3 min; then 8 cycles of 95°C, 30 sec / 55°C, 30 sec / 72°C, 30 sec: then 72°C, 5 min; and a 4°C hold. Purified indexed libraries were diluted, pooled and sequenced on the MiSeq instrument using the paired end 600 bp MiSeq Reagent v3 (Illumina) per manufacturer’s instructions.

Overlapping paired-end reads were merged and quality trimmed using pRESTO (Vander Heiden et al., 2014). Sequences with only one representative read were discarded. Filtered reads were error corrected and repertoires constructed from the reads using IgReC (Shlemov et al., 2017). VDJ assignment and heavy chain mutation frequency were performed using IgBLAST (Ye et al., 2013) and additional analyses were performed within R studio. Total counts of each sequence were calculated taking into account the counts of the deduplicated reads. The IgReC counts and sequences with a total count <20 reads or that were nonfunctional were discarded. Pairwise sequence Levenshtein distances within IgG or IgM of each donor were calculated using stringdist (Loo, 2014). Sequences were hierarchically clustered based on the distances with a cutoff of 0.05 and the top sequence by count was selected as the representative for the cluster. The frequency of heavy chain usage in IgM and IgG isotypes was calulcated for all donors analyzed or separated broad and non-broad neutralizer groups.

Determination of plasma CXCL13 levels

CXCL13 levels of diluted plasma were determined by Luminex (Millipore Sigma, Burlington, MA) using the manufacturer’s protocol. Briefly, diluted plasma was incubated with anti-CXCL13 antibody conjugated beads before being washed and read on a FlexMap3D (Luminex, Austin, TX). The standard curve was used to determine relative concentrations of plasma CXCL13.

Total IgG quantification

FortéBio Octet biolayer interferometry was performed using protein G biosensors per manufacturer’s protocol (FortéBio, Fremont, CA). Briefly, protein G biosensors (FortéBio, Fremont, CA) were pre-wet in hydrating solution for 10 minutes immediately prior to use. Tips were then transferred into wells with sample or a serially diluted IgG control. Total IgG concentrations were determined using data analysis software 9.0 (FortéBio, Fremont, CA) based on binding rates (nm) compared to serial dilution IgG controls after reference depleted IgG serum subtraction.

Antibody Binding

Longitudinal binding antibodies to biotinylated SOSIPs, autologous gp140, or consensus gp140s ConAl, ConAE01, ConB, ConC, ConD, ConG, and ConM were evaluated using a multiplex Luminex assay. Antigens were covalently coupled to uniquely coded carboxylated magnetic microspheres (Luminex Corp., Austin TX) per manufacturer’s protocol. Briefly, microspheres are activated by incubation in buffer containing 1-Ethyl-3[3-dimethylaminopropyl]carbodiimide hydrochloride and N-hydroxysulfosuccinimide for 20 min. Following activation, beads were incubated with antigen for 2 hr to allow coupling via the primary amine. Biotinylated peptide antigens were bound to SA-coated microspheres for 2 hr followed by addition of free biotin to quench the reaction. Following coupling, coated microspheres were washed and stored at −80°C in PBS containing 0.1% BSA, 0.05% sodium azide and 0.02% Tween-20. Coupling efficiency and specificity were confirmed by testing pooled plasma from HIV-infected and uninfected healthy controls (SeraCare, Milford MA), HIV-IG (NIH AIDS Reagent Program) and purified normal human IgG (Sigma-Aldrich, St. Louis MO). Ig isotype specificity was determined by detection with R-phycoerthrin (PE)-conjugated mouse anti-human IgG (0. 5μg/mL), IgA (5μg/mL), or IgM (5pg/mL) (Southern Biotech, Birmingham AL). Fluorescent signals were measured with FlexMap3D with xPONENT v4.2 software and the median fluorescence intensity was obtained and compared to a standard curve. Pooled plasma from HIV-infected and -uninfected healthy controls, HIV-IG, normal human IgG, and no-sample control wells were included on each assay plate. Mean flourscent intensity (MFI) from samples (signal) was compared to the MFI of healthy control plasma (noise), and a ratio of signal:noise was graphed at each time point.

QUANTIFICATION AND STATISTICAL ANALYSIS

Prism v.8.0.2 (GraphPad Software, Inc., San Diego, CA) and R through R Studio (1.1.453, R Consortium, Boston, MA) were employed for all univariate and multivariate statistical analyses and graphical representations of data. Wilcoxon matched-pairs signed-rank test was used for intragroup comparisons, and Mann-Whitney nonparametric U tests were used for unpaired comparisons between groups. Correlations were determined by Spearman’s rank-correlation analysis. An additional Fisher’s exact analysis was employed to compare differences in neutralization epitopes, regional breadth and potency, and infecting subtypes. Cox proportional hazard (PH) models were used to interrogate associations between the time to reach peak neutralization breadth above 70% with individual clinical parameters and autologous IgM, IgG, and IgA binding antibodies. Data was analyzed as a continuum of broad neutralizers and individuals who remained in the study for at least 3 years by estimating the likelihood (hazard ratio) of reaching peak neutralization breadth above 70% in a bivariate analysis or multivariate analysis that controlled for VL. To have a better representation of the estimation generated from the modeling, the unit change for Cox PH model was defined as 200 cells/mm³ for cell counts, 200 copies/ml for VL, and 1 standard deviation unit change for autologous antibody isotype binding. Logistic regression was used to analyze the parameters that associate with the development of neutralization breadth. B cell frequencies were applied to the absolute B cell counts at the concurrent time points to obtain the absolute B cells for each subset. To have a better representation of the estimation generated from the modeling, the unit change was defined as 20 cells/mm³ for logistic regression analysis of the absolute peripheral B cells within the subpopulations. Logistic regression was also used for the B cell frequency of each subpopulation (% of CD19⁺ B cells), and the logistic regression for B cell frequency gp140⁺ B cells was performed using the change per 1% of the population. Forest plots were utilized to integrate and graphically display the estimated Odds Ratios (OR) along with 95% Confidence Intervals (CI) from logistic regression on each of the quantile (25%, 35%, 50%, 65%, 75%) of B cell counts. Locally weighted smoothing spline was applied to approximate and visualize the longitudinal dynamics of each cell count and VL against the days after 1^st positive HIV RNA (days post-viremia). Radar plots were generated using the median normalized scores of B cell subpopulation frequencies of broad and non-broad neutralizers. To avoid possible bias from different scales of subpopulations of B cell subpopulation frequencies, minimum-maximum normalization was applied to each respective subpopulation respectively to convert them into the same standardized scale across all time points. Cumulative incidence functions were used to estimate the proportion of the population with a higher probability of developing >70% neutralization breadth based on gp140⁺ (CD19⁺/gp140⁺) B cell frequencies and LogRank test for trend was used to verify the significant difference between the levels. To confirm the accuracy and prediction of the development of neutralization breadth with each gp140⁺ B cell subpopulation, receiver operating characteristic (ROC) curve analysis was performed for visualization and the area under the ROC curve (AUC) for each subpopulation was measured for quantitative analysis (Robin et al., 2011). Unsupervised principle component analysis (PCA) was performed to confirm the differences in gp140⁺ B cell subpopulation frequencies between broad and non-broad neutralizers through sample clustering. PCA-biplot based on the first two principle components was utilized to visualize the relative influence of each gp140⁺ B cell subpopulation. Heavy chain usage was adjusted for multiple comparisons using the false discovery rate (FDR). Significance is defined as a p<0.05, but trends are also included in the Figures when observed.

Supplementary Material

2

Table S1. Neutralization of a 34-virus panel by RV217 individuals. Related to Figures 1, and 2 and Table 1

3

Table S2. Delineation of antibody specificities in RV217 donor plasma. Related to Figure 1

4

Table S3. Demographics of broad and non-broad neutralizers. Related to Figure 1

5

Table S4. Early neutralization profile of broad and non-broad neutralizers against the 15-virus panel. Related to Figures 1 and 2

6

Table S5. Neutralization profile of broad and non-broad neutralizers against autologous pseudoviruses. Related to Figure 6 and Figure S6

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
FITC anti-human Integrin β7	BioLegend	Cat# 321214 RRID:AB_830858
BV510 anti-human CD3	BD Biosciences	Cat# 740187 RRID:AB_2739940
BV510 anti-human CD14	BioLegend	Cat# 301842 RRID:AB_2561946
BV510 anti-human CD56	BioLegend	Cat# 318340 RRID:AB_2561944
BV510 anti-human CD16	BD Biosciences	Cat# 563830 RRID:AB_2744296
BV510 anti-human CD4	BD Biosciences	Cat# 562970 RRID:AB_2744424
BV510 anti-human CD8	BioLegend	Cat# 301048 RRID:AB_2561942
BV570 anti-human CD20	BioLegend	Cat# 302332 RRID:AB_2563805
BV605 anti-human CD27	BioLegend	Cat# 302830 RRID:AB_2561450
BV711 anti-human CD10	BioLegend	Cat# 312226 RRID:AB_2565876
PE-Cy7 anti-human CD21	BD Biosciences	Cat# 561374 RRID:AB_10681717
ECD anti-human CD19	Beckman Coulter	Cat# IM2708U RRID:AB_130854
Ax680 anti-human CD38	NIH	N/A
PE-Cy5 anti-human IgM	BD Biosciences	Cat# 551079 RRID:AB_394036
BV421 anti-human IgG	BD Biosciences	Cat# 562581 RRID:AB_2737665
Anti-IgG (Pan) PE	Southern Biotech	Cat# 9040-09 RRID:AB_2796601
Anti-IgA PE	Southern Biotech	Cat# 2050-09 RRID: AB_2795707
Anti-IgM (PE)	Southern Biotech	Cat# 9020-09 RRID: AB_2796577
Human IgG Isotype Control	ThermoFisher	Cat# 12000C RRID:AB_2532991
Bacterial and Virus Strains
E.coli DH5α	ThermoFisher Scientific	Cat# 18265017
Heterologous HIV-1 Env-pseudotyped viruses	John R. Mascola, NIH, Dana Gabuzda, James Binley, & David Montefiori (Kong R et al, 2016)	N/A
HIV-2 MPER chimeras	G. Shaw	(Davis et al., 2009)
pSV 10204v0A_04_A1_MT4	This study	N/A
pSV 102204_09_E7_GS	This study	N/A
pSV 10220v04_08_GS	This study	N/A
pSV 20337v02_03_B1	This study	N/A
pSV 20337v02_05_B8_MT3_MT2	This study	N/A
pSV 20337v02_10_F3_MT7_MT8	This study	N/A
pSV 20337v02_c02_16	This study	N/A
pSV 20337v02_c04_8	This study	N/A
pSV 20509v01_09_E10_MT1	This study	N/A
pSV 20511v01_06R_E12	This study	N/A
pSV 30112v02_04R_bB1_MT6	This study	N/A
pSV 30124v01_05_A19	This study	N/A
pSV 30124v01_07_F10_GS	This study	N/A
pSV 30156v0B_11_H13_MT6_MT1_MT7_GW	This study	N/A
pSV 30196v01_04_E7_MT5	This study	N/A
pSV 30486v01_02_bC3_MT3	This study	N/A
pSV 40363v03_04_K23_MT2	This study	N/A
pSV 40363v03_05_J12_GS	This study	N/A
pSV 40363v03_06_I16_GW	This study	N/A
pSV 40363v03_08_D6	This study	N/A
pSV 40363v03_c02env_79	This study	N/A
pSV 40436v09_09_D9_MT2_MT4_MT2	This study	N/A
pSV 40436v09_c06env_11	This study	N/A
pSV 40503v01_03_C10_MT17	This study	N/A
pSV 40503v01_08_H5_MT4_GW	This study	N/A
pSV 20178v01_02_GS	This study	N/A
pSV 20225v01_01_GS	This study	N/A
pSV 20233v01_10_GS	This study	N/A
pSV 20263v01_03_E5_GS	This study	N/A
pSV 20355v01_05_G2	This study	N/A
pSV 20631v03_11_E10_GS	This study	N/A
pSV 30507v01_13_A11	This study	N/A
pSV 40094v01_01R_M1	This study	N/A
pSV 40139v0J_03_F1	This study	N/A
pSV 40195v1a_01_C5_MT2	This study	N/A
pSV 40265v02_02_bG10_MT5_GW	This study	N/A
pSV 40511v01_08_D7	This study	N/A
Biological Samples
PBMCs from MHRP RV217 donors	M. Robb	(Robb et al., 2016)
Plasma from MHRP RV217 donors	M. Robb	(Robb et al., 2016)
Chemicals, Peptides, and Recombinant Proteins
Q5 High-Fidelity DNA Polymerase	New England Biolabs	Cat# M0491S
T4 DNA Polymerase	New England Biolabs	Cat# M0203S
ExpiFectamine 293 transfection kit	ThermoFisher Scientific	Cat# A14525
Polyethylenime	Polysciences Inc.	Cat# 23966
Expi293 Expression Medium	ThermoFisher Scientific	Cat# A14351
Opti-MEM	ThermoFisher Scientific	Cat# 31985
Galanthus nivalis lectin agarose	Vector Laboratories	Cat# AL-1243
Methyl alpha-D-mannopyranoside	Sigma	Cat# M6882
Q sepharose fast flow	GE Healthcare	Cat# 17051010
Dulbecco’s Modified Eagle Medium (DMEM)	Thermo Fisher	Cat# 11965126
Penicillin-Streptomycin	Thermo Fisher	Cat# 15140122
Fetal Bovine Serum (FBS)	Gemini Bio Products	Cat# 10438018
FuGene 6	Promega	Cat# E2692
LIVE/DEAD Fixable Aqua Dead Cell Stain	Thermo Fisher	Cat# L34957
DEAE-Dextran	Sigma	Cat# D9885-10G
Britelite PLUS	Perkin Elmer	Cat# 6066769
X-tremeGENE HP DNA Transfection Reagent	Sigma	Cat# 6366546001
Benzonase	Novagen	Cat# 70664-3
Streptavidin, R-Phycoerythrin Conjugate (SAPE)	ThermoFisher	Cat# S21388
anti-CXCL13 antibody conjugated beads	Millipore	Cat# HBCA1-MAG
Zeba Spin Desalting Columns 7K MWCO	ThermoFisher	Cat# 89890
FluoReporter Mini-Biotin-XX Protein Labeling Kit	Invitrogen	Cat# F6347
Microplate, 96 well PP, F-bottom	Greiner Bio-ome	Cat# 655209
Protein G Biosensors	FortéBio	Cat# 18-5083
Sample Diluent	FortéBio	Cat# 18-5028
Kinetics Buffer 10x	FortéBio	Cat# 18-1105
Human reference serum	Bethyl	Cat# RS10-110
NuPAGE 4–12% Bis-Tris Gel	Invitrogen	Cat# NP0321BOX
IgG Depleted Processed Serum	BBI	Cat# SF142-7
NuPAGE Sample Reducing Agent	ThermoFisher	Cat# NP0004
NuPAGE LDS Sample Buffer	Novex	Cat# 1606369
SimplyBlue SafeStain	ThermoFisher	Cat# LC6060
NuPAGE MOPS SDS Running Buffer	Invitrogen	Cat# NP0001-02
Bovine Serum Albumin	Sigma	Cat# A9576-50ML
DPBS without Ca & Mg	Quality Biological	Cat# 114-057-101
Tween 20	Sigma	Cat# P9416-100mL
10x PBS Solution, pH 7.4	TEKnova	Cat# P0196
2-mercaptoethanol	ThermoFisher	Cat# 21985023
Buffer TCL	Qiagen	Cat# 1031576
AMPure XP	Beckman Coulter	Cat# A63881
KAPA HiFi Hotstart enzyme	Kapa Biosystems	Cat# KK2602
RV217.10204v0A founder gp140	This study	N/A
RV217.10220v04.MAJ founder gp140	This study	N/A
RV217.10220v04.MIN founder gp140	This study	N/A
RV217.20178v01 founder gp140	This study	N/A
RV217.20225v01 founder gp140	This study	N/A
RV217.20233v01 founder gp140	This study	N/A
RV217.20263v01 founder gp140	This study	N/A
RV217.20337v02.a.4 founder gp140	This study	N/A
RV217.20337v02.b.2 founder gp140	This study	N/A
RV217.20337v02.c.2 founder gp140	This study	N/A
RV217.20337v02_02 founder gp140	This study	N/A
RV217.20337v02_10 founder gp140	This study	N/A
RV217.20355v01 founder gp140	This study	N/A
RV217.20509v01 founder gp140	This study	N/A
RV217.20511v01 founder gp140	This study	N/A
RV217.20631v03 founder gp140	This study	N/A
RV217.30112v02 founder gp140	This study	N/A
RV217.30124v01.MAJ founder gp140	This study	N/A
RV217.30124v01.MIN founder gp140	This study	N/A
RV217.30156v0B founder gp140	This study	N/A
RV217.30196v01 founder gp140	This study	N/A
RV217.30486v01 founder gp140	This study	N/A
RV217.30507v02 founder gp140	This study	N/A
RV217.40094v01 founder gp140	This study	N/A
RV217.40139v01 founder gp140	This study	N/A
RV217.40195v01 founder gp140	This study	N/A
RV217.40265v02 founder gp140	This study	N/A
RV217.40363v03.a.4 founder gp140	This study	N/A
RV217.40363v03.b.3 founder gp140	This study	N/A
RV217.40363v03_04 founder gp140	This study	N/A
RV217.40363v03_05 founder gp140	This study	N/A
RV217.40363v03_08 founder gp140	This study	N/A
RV217.40436v02 founder gp140	This study	N/A
RV217.40436v09_09 founder gp140	This study	N/A
RV217.40503v01.MAJ founder gp140	This study	N/A
RV217.40503v01.MIN founder gp140	This study	N/A
RV217.40511v01 founder gp140	This study	N/A
gp140 Consensus D	This study	N/A
BG505 SOSIP		(Doria-Rose et al., 2014)
ZM106.9 SOSIP	G.Joyce	(Joyce et al., 2017)
Consensus gp140 A, B, C, G, AE, and M	Duke PPF	(Haynes et al., 2012a)
Critical Commercial Assays
QIAquick Gel extraction kit	Qiagen	Cat# 28704
QIAprep Spin Miniprep Kit	Qiagen	Cat# 27106
Plasmid Purification Mega Kit	Qiagen	Cat# 12181
NucleoZOL	TaKaRa	Cat# 740404
NucleoSpin RNA Clean-up	TaKaRa	Cat# 740948
PrimeSTAR GXL DNA Polymerase	TaKaRa	Cat# R050A
SMART-Seq v4 Ultra Low Input RNA kit	TaKaRa	Cat# R400752
Nextera XT Index Kit v2	Illumina	Cat# TG-131-2001
MiSeq Reagent v3 Kit	Illumina	Cat# MS-102-3003
Deposited Data
RV217.10204v0A	This study	Genbank MW443137-MW443146
RV217.10220v04.MAJ	(Rolland et al., 2020)	N/A
RV217.10220v04.MIN	(Rolland et al., 2020)	N/A
RV217.20178v01	This study	Genbank MW443147-MW 443156
RV217.20225v01	(Rolland et al., 2020)
RV217.20233v01	This study	Genbank MW443157-MW443166
RV217.20263v01	(Rolland et al., 2020)	N/A
RV217.20337v02.a.4	(Rolland et al., 2020)	N/A
RV217.20337v02.b.2	(Rolland et al., 2020)	N/A
RV217.20337v02.c.2	(Rolland et al., 2020)	N/A
RV217.20337v02_02	(Rolland et al., 2020)	N/A
RV217.20337v02_10	(Rolland et al., 2020)	N/A
RV217.20355v01	(Rolland et al., 2020)	N/A
RV217.20509v01	(Rolland et al., 2020)	N/A
RV217.20511v01	(Rolland et al., 2020)	N/A
RV217.20631v03	(Rolland et al., 2020)	N/A
RV217.30112v02	(Rolland et al., 2020)	N/A
RV217.30124v01.	(Rolland et al., 2020)	N/A
RV217.30124v01.MIN	(Rolland et al., 2020)	N/A
RV217.30156v0B	This study	Genbank MW443167-MW443176
RV217.30196v01	This study	Genbank MW443177-MW443186
RV217.30486v01	This study	Genbank MW443187-MW443196
RV217.30507v02	(Rolland et al., 2020)	N/A
RV217.40094v01	(Rolland et al., 2020)	N/A
RV217.40139v01	This study	Genbank MW443197-MW443205
RV217.40195v01	This study	Genbank MW443206-MW443215
RV217.40265v02	(Rolland et al., 2020)	N/A
RV217.40363v03.a.4	(Rolland et al., 2020)	N/A
RV217.40363v03.b.3	(Rolland et al., 2020)	N/A
RV217.40363v03_04	(Rolland et al., 2020)	N/A
RV217.40363v03_05	(Rolland et al., 2020)	N/A
RV217.40363v03_08	(Rolland et al., 2020)	N/A
RV217.40436v02	(Rolland et al., 2020)	N/A
RV217.40436v09_09	(Rolland et al., 2020)	N/A
RV217.40503v01.MAJ	This study	Genbank MW443216-MW443225
RV217.40503v01.MIN Env sequence	This study	Genbank MW443216-MW443225
RV217.40511v01 Env sequence	(Rolland et al., 2020)	N/A
Experimental Models: Cell Lines
Human: HEK 293T ATCC	ATCC	Cat# CRL-3216; RRID: CVCL_0063
Human: HeLa-derived TZM-bl	NIH AIDS Reagent Program	Cat# 8129-442; RRID: CVCL_B478
Human: Expi293F	Thermo Fisher	Cat# A14527
Oligonucleotides
Forward HGS_VH1	This paper	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTAGCAGCCACAGGTGCCCACTCC
Forward HGS_VH2	This paper	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTAYCCCTTCMTGGGTCTTGTCC
Forward HGS_VH3	This paper	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTTMTTTTARRAGGTGTCCAGTGT
Forward HGS_VH4	This paper	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTGCTCCCAGATGGGTCCTGYCC
Forward HGS_VH5	This paper	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTGTTCTCCAAGGAGTCTGTYCC
Forward HGS_VH6	This paper	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTCTCCCATGGGGTGTCCTGTCA
Forward HGS_VH7	This paper	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTGCAGCAACAGGTGCCCACTCC
Forward 5’ SMART	(Su et al., 2017)	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGAAGCAGTGGTATCAACGCAGAGT
Reverse IgG	This paper	GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGSGATGGGCCCTTGGTGGARGC
Reverse IgM	This paper	GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGGTTGGGGCGGATGCACTCC
Software and Algorithms
xPONENT 4.2	Luminex	https://www.luminexcorp.com
FlowJo 9.9.6	FlowJo, LLC	https://www.flowjo.com
Data analysis software 9.0	FortéBio	https://www.fortebio.com
Prism v.8.0.2	GraphPad Software	https://www.graphpad.com
R studio	R Consortium	https://rstudio.com/
pRESTO	(Vander Heiden et al., 2014)	N/A
IgReC	(Shlemov et al., 2017)	N/A
IgBLAST	NCBI	https://ncbi.nlm.nih.gov
Stringdist	(Loo, 2014)	https://cran.r-project.org

HIGHLIGHTS.

• A reduction in total peripheral B cells associates with HIV neutralization breadth

• The initial interaction of founder Env with naïve B cells predicts bNAb development

• Increased B cell engagement with founder Env increases activation and differentiation