Similar viral load kinetics and shared Env gene mutation patterns in RMs infected with molecularly cloned SHIVs

(A) Plasma viral load kinetics measured by qPCR of viral RNA copies. Values below detection limits were set to 1 for visualization. (B) Mutation frequency at individual amino acid residues based on Env sequences recovered from each RM. Domain (top) and position (bottom) annotations follow HXB2 numbering. (C) Proportion of mutations unique to a single RM or shared by different numbers of RMs per group. Mutations at the same positions or identical mutations are counted separately. (D) Mutation frequencies of shared mutations. Each dot represents a mutation from one RM. Numbers above columns indicate the percentage of mutations with frequencies >0.5 (dashed line). (E) Examples of identical mutations shared among RMs. Titles indicate residues in the infecting T/F virus strain. Letters on individual bar segments indicate the one-letter codes of the mutated amino acid residues. +, insertion; –, deletion. Color-coded as in C, with grey denoting unmutated proportions.

Persistent and productive GC responses in RMs with chronic SHIV infection

(A) Flow cytometry gating strategies to identify B-cell populations in LN cells derived from SHIV-infected RMs. (B) Alluvial plots showing percentages of IgG+ GC, IgG+ Bmem and IgM+ Bmem populations among CD20+ LN cells during the infection time course in the three RMs in each infection group. (C-E) Alluvial plots showing percentages of SOSIP-binding (C), gp140-binding (D), and autologous neutralizing (E) IgG+ single B-cell cultures from different B-cell populations over the infection course. See also Figures S1 and S2, and Tables S1 and S2.

Distinct clonal dynamics of Env-reactive neutralizing and non-neutralizing IgG+ GC B-cell clones

(A) Expansion/contraction of Env-reactive neutralizing and non-neutralizing IgG+ GC B-cell clones in each infection group over time. ND, not detected. (B-C) Distributions of timepoint occurrence (B) and clone size (C) were compared between neutralizing (Neut) and non-neutralizing (NonN) clones. Two-sided Wilcoxon rank-sum tests at the clone level (pooled across RMs within each SHIV group) were used. (D) Clone counts of neutralizing and non-neutralizing IgG+ GC B-cell clones in individual RMs. (E) IGHV mutation frequency changes in Env-reactive clones spanning multiple timepoints. All nine clones (named “RM_Clone-ID”) that persisted across at least two timepoints and contained more than two clonal members at each timepoint are shown. Adjacent timepoints were compared using two-sided Wilcoxon rank-sum tests. Statistical annotations: *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001; ns, not significant. See also Table S3 and Document S1.

Limited clonal relationship between Env-reactive Bmem and IgG+ GC B cells

(A) Proportions of neutralizing and non-neutralizing Env-reactive B cells from different populations. Inserted numbers indicate cell counts. ND, not detected. (B) Clonal relationship between Env-reactive IgG+ GC and Bmem cells. Top sectors show B-cell populations (color-coded); lower sectors represent individual neutralizing or non-neutralizing clones, including lineages and singletons. Color gradients (red for neutralizing and blue for non-neutralizing) were used solely to visually distinguish adjacent individual clones. (C) Venn diagrams showing the distribution of B-cell clones across single or multiple populations. Blank regions indicate zero counts. (D-E) For lineages containing both IgG+ GC and Bmem members at the same timepoints: (D) VH mutation frequencies and neutralization activities for all seven RM_Clone-ID_Timepoint groups identified with >3 clonal members; (E) VH mutation frequency and neutralization differences between Bmem and IgG+ GC B cells for all 19 RM_Clone-ID_Timepoint groups identified. Each dot represents a Bmem–IgG+ GC B cell pair. See also Table S3.

Distinct functional types of Env-reactive B-cell clones defined by antigen binding and neutralization

(A) Log10-transformed fold-change of MFI ratios (SOSIP/gp140) for selected bnAbs and nnAbs, normalized to VRC01. MFI values at a same concentration (206 ng/ml, dashed lines in Figure S4A) were used; similar trends observed at other concentrations. bs, CD4bs-specific tier-1-neutralizing. (B) Composition of antigen specificities of Env-reactive antibodies identified from SHIV-infected RMs. (C) Venn diagrams categorizing Env-reactive antibodies by autologous neutralization and SOSIP/gp140 binding. Blank regions indicate zero counts. (D) Dot plots of SOSIP- vs. gp140-binding MFI values for Env-reactive antibodies in individual RMs, with a color gradient for autologous neutralization. (E) Similar plotting as (D), exemplifying four major types of Env-reactive B-cell clones (“RM_Clone-ID”). See also Figure S4, Table S3, and Document S2.

Shared functional characteristics of Env-reactive recombinant antibodies derived from different RMs

(A) Env-binding MFI values of rAbs to autologous and heterologous SOSIP/gp140. Functional types of source B-cell clones are color-coded (legend in B). Dashed lines indicate binding cutoffs. (B) Heatmaps of rAb (rows) neutralization against autologous and heterologous pseudo-viruses. Autologous SHIV neutralization using culture supernatants (Sup_SHIV) and SOSIP/gp140 binding (panel A) are included for comparison. (C) Maximum heterologous neutralization vs. gp140 binding of rAbs (panel B), with dashed cutoffs for neutralization and binding. (D-E) Heatmaps (D) showing percentages of inhibition of analyte rAbs (rows) binding to autologous SOSIP or gp140 antigens in the presence of bnAbs or nnAbs competitors. UMAP coordinates are displayed in parallel within (D) and as 2D plots in (E), labeled with clusters identified using the DBSCAN algorithm. RMs and Types in panel D use the same color-coding as in B. Color-coding in panel E shows types as in B, and shape-coding shows RMs as in A. See also Table S3, S4, and S5, Figures S3, and S4, and Document S3.

Heterogeneous BCR genetics underlying similar functional profiles in genetically diverse RMs

(A-B) Venn diagrams showing the numbers of shared and distinct IGHV, IGKV, and IGLV alleles in the MF BCR repertoire (A) and Env-reactive BCRs identified (B) across three RMs per infection group. (C-D) Bubble plots of IGV allele usage in the MF BCR repertoire (C) and Env-reactive BCRs (D), mapped to UMAP coordinates based on sequence similarity. Bubbles are color-coded by IGV family (C) or functional type (D) and scaled by usage percentages among unique CDR3s (uCDR3, C) or Env-reactive BCRs identified (xBCR, D). IGKV and IGLV are combined as light chain repertoire in percentage calculation. Numbers indicate counts of unique CDR3s from Rep-Seq (C) or Env-reactive BCRs identified (D). (E–F) Distribution of MADM of IGV allele usage frequencies across RMs, calculated from original percentages (E) or percentages smoothed within IGV genes (F). The distribution for uCDR3 represents the combined MADM values from 1000 multinomial down-sampled simulations (see Methods). Empirical p values were calculated between the median MADM of xBCR and 1000 median values of randomly-sampled and size-matched uCDR3 subsets (see Methods). ***, empirical p < 0.001. Alleles with zero frequencies across all three RMs were excluded to prevent distortion by an excess of MADM = 0 values. (G) Cosine similarity scores of IGV allele usage between RMs. For the uCDR3 source, dots indicate medians from 1000 multinomial down-sampling simulations (see Methods), with error bars showing 25th – 75th percentiles. Lines connect the same RM pair comparisons from uCDR3 or xBCR sources, with solid lines for original percentages and dashed lines for percentages smoothed within IGV genes. Line colors indicate the significance of empirical p values: grey (***) for p < 0.001, blue (**) for p < 0.01, green (*) for p < 0.05, and orange (ns) for not significant. See also Table S6 and Figure S3.

Feeder cell lines for single B-cell cultures, and autologous neutralization activities from the culture supernatants, related to Figure 2

(A) Kinetics of IgG production in culture supernatants from RM IgM+ and IgG+ single B cells cultured with the MS40L feeder cell line. Each dot represents an individual IgG+ culture well. Dashed lines indicate detection limits, and error bars (in red) represent the median values. (B-C) Comparison of the original MS40L and engineered MEC-147 feeder cell lines in supporting single B-cell cultures from different B-cell populations. Panel (B) shows the culture of B cells isolated from the lymph nodes of an RM after DTaP immunization, while panel (C) shows the culture of B cells isolated from PBMCs of a healthy human donor. The top panels display the percentage of IgG+ culture wells among all single-cell sorted wells. Error bars indicate the mean + SD across three 96-well culture plates. The bottom panels show OD450 values from IgG ELISA assays, with each dot representing an individual IgG+ culture well. Dashed lines indicate detection limits, and error bars (in red) represent the median values. (D) Single nucleated cells were isolated from LN biopsies of infected RMs at various time points following SHIV infection. B cells of different phenotypes, including mature follicular (MF), IgM+ Bmem, IgG+ GC, and IgG+ Bmem cells, were identified from isolated LN cells by flow cytometry and sorted into single B-cell cultures with MEC- 147 feeder cells. Supernatants were harvested after 18 days of culture, and IgG-producing cultures were identified using standard ELISA assays. IgG-containing culture supernatants were tested for neutralization activity against autologous SHIV pseudo-viruses using standard TZM-bl assays. Each dot represents an individual single B-cell culture supernatant containing clonal IgG. Error bars in red indicate median values, while blue dashed lines indicate cutoff thresholds for defining neutralizing versus non-neutralizing cultures. Inserted red numbers represent the percentage of neutralizing cultures among IgG+ cultures for the corresponding B-cell populations (see Table S2 for details).

Autologous Env antigen-binding activities of single B-cell culture supernatants, related to Figure 2

The same IgG-containing single B-cell culture supernatants prepared as described in Figure S1D were tested for binding activity against autologous SOSIP and gp140 antigens using standard Luminex assays, conducted in parallel and in a blinded manner to the autologous neutralization assays shown in Figure S1D. RMs were infected with either the BG505 (A) or CH505 (B) SHIV strains. B-cell populations are color-coded as indicated in the keys. Dashed lines indicate cutoff thresholds for defining antigen-binding versus non-binding cultures. Inserted numbers represent the percentage of antigen-binding cultures among IgG+ cultures for the corresponding B-cell populations. Additional antigens or capture antibodies, including CH505.gp120, MN.gp41, anti-RhIgG, and non-relevant controls (streptavidin, BSA, and ovalbumin), were included in the Luminex bead panel (see Table S2 for details).

Verification and supporting analyses, related to Figures 4, 6, and 7, Table S3, and Methods

(A) Related to Table S3. Proportions of identified autologous SOSIP- or gp140-binding B cells for which BCR sequences were determined. The blue horizontal line separates two categories: cultures with binding MFI > 300 (upper) and MFI ≤ 300 (lower). Stacked bars represent proportions normalized to the total counts of Env-binding B cells with MFI > 300, with pink indicating sequenced and grey indicating not sequenced. Numbers within bars denote the counts of Env-binding B cells in each category. (B) Related to Figure 4E. VH mutation frequency and neutralization differences between Bmem and IgG+ GC B cells for lineages containing both IgG+ GC and Bmem members at the same timepoints in individual RMs. Each dot represents a Bmem–IgG+ GC B cell pair. Only RM–Bmem population groups with more than two data points are shown. (C) Related to Figure 6A. Relationship between rAb binding MFI values to heterologous gp140 protein and the binding MFI values of corresponding single B-cell culture supernatants to CH505.gp120 and MN.gp41 proteins in the initial screening. Functional types of source B-cell clones are color-coded, and dashed lines indicate binding cutoffs. (D) Related to Methods. UMAP plots showing the same coordinates and clusters as in Figure 6E, with colors indicating individual assay batches. (E-F) Related to Figure 7. Normalized density plots display the distribution of distances calculated from the pairwise IGV gene distance matrix at two categorical levels: IGV gene (E) and family (F). Distances within the same category (“Within”) are shown in black, while distances between different categories (“Between”) are shown in grey. Dashed vertical lines in blue represent cutoff values, determined either at the intersection of within and between distance distributions or at the point of minimum difference when no intersection is observed. (G) Related to Figure 7. Distribution of MADM of IGV allele usage frequencies across RMs, calculated from percentages smoothed within IGV families. The distribution for uCDR3 represents the combined MADM values from 1000 multinomial down-sampled simulations (see Methods). Empirical p values were calculated between the median MADM of xBCR and 1000 median values of randomly-sampled and size-matched uCDR3 subsets (see Methods). ***, empirical p < 0.001. Alleles with zero frequencies across all three RMs were excluded to prevent distortion by an excess of MADM = 0 values. (H) Related to Figure 7. Cosine similarity scores of IGV allele usage between RMs. For the uCDR3 source, dots indicate medians from 1000 multinomial down-sampling simulations (see Methods), with error bars showing 25th – 75th percentiles. Lines connect the same RM pair comparisons from uCDR3 or xBCR sources, with solid lines for original percentages and dashed lines for percentages smoothed within IGV families. Line colors indicate the significance of empirical p values: grey (***) for p < 0.001, blue (**) for p < 0.01, green (*) for p < 0.05, and orange (ns) for not significant.

Env antigen-binding activities of rAbs, related to Figures 5 and 6

(A) Binding activities of reference bnAbs and nnAbs (human IgG1 format) for BG505 and CH505 SOSIP and gp140 antigens were tested in standard Luminex assays using serial dilutions. Vertical dashed lines indicate the concentration point (206 ng/ml) used for calculating MFI ratios in Figure 5A. (B) Binding activities of rAbs (rhesus IgG1 format) derived from representative B-cell clones of different functional types were similarly tested. Results for selected rAbs from BG505-infected (top row) and CH505-infected RMs (bottom row) are shown (named as “RM_Clone-ID (Type)”). Full data for all 146 rAbs are provided in Document S3. Antigens are color-coded as in panel (A). Dashed grey curves represent MFI values detected with Luminex beads conjugated to an anti-RhIgG capture antibody. Vertical dashed lines indicate the concentration points where anti-RhIgG MFI values align closely with those detected using culture supernatants during initial screening. (C) Comparison of MFI values for autologous SOSIP and gp140 antigens detected using culture supernatants (MFI (sup)) during initial screening and rAbs (MFI (rAb)) at selected concentration points (indicated by vertical dashed lines in panel (B) and Document S3). Linear regression models were fitted, and R2 values indicating the proportion of variance explained are shown in parentheses in the legends for corresponding antigens.