Fig 1.
TLMs are significantly less diverse and mutated than other MBCs in PLWH.
A: Flow-cytometric quantification of frequencies of IgM and IgG MBC subsets within IgD- B cells, n = 10 PLWH on ART (triangles), n = 12 for PLWH with elite control (circles) and n = 10 for HIV-negative controls (squares). Gating strategy shown in S1 Fig. MBC subsets are colour-coded as follows: activated (orange), resting (teal) and TLM (purple). Donors that underwent bulk BCR sequencing are highlighted with black symbols. No statistical significance was found in the Benjamini-Hochberg multiple testing adjusted p-values between donor groups for each MBC subset in turn using Mann-Whitney U tests with the compare_means function from the ggpubr package. B: Clonal diversity indices of MBC subsets stratified by IgG versus IgM expression for PLWH (n = 5). 100 BCR sequences were randomly sampled from each donor/subset for comparison. C: VH mutational burden (as % of nucleotides) of BCR sequences analysed in (B). D: Calculation of mutations between B cell subsets for PLWH (n = 5). Lineages containing multiple MBC subsets were selected, and BCR sequences underwent pairwise comparison to identify unique AA mutations (FWR1 - FWR3). The median was taken per comparison for each lineage and then averaged to give a value per donor. The x axis shows the comparison group, and the y axis the average proportion of unique AA mutations per comparison. E-F: BCR phylogenetic network analysis to ascertain MBC subset evolution within PLWH (n = 5). Networks were created for each lineage, with each node representing a BCR (self-linkages excluded). Edges between nodes were drawn if the tip-to-tip distance was below a set threshold, defined as the position of first apex in the rate of change curve for the tip-to-tip distances. The network was directed by only allowing edges from closer nodes to germline to further nodes from germline. Once these networks were created, the hub score (E) and out-degree to in-degree ratio (F) was calculated (hub score representation, panel (E), left hand side), to provide global and local measures of node connectivity, respectively. Throughout, statistical significance was assessed using Mann-Whitney U tests with the compare_means function from the ggpubr package (p > 0.05 not marked, p ≤ 0.05 *, p < 0.01, **, p < 0.001 ***, p < 0.0001 ****).
Fig 2.
Minor transient spike in HIV viremia triggers antibody response.
A: Dynamics of viral load (V/L, copies/mL) and CD4 + T cell count (per mm3) from an elite controller (index participant). Grey bars indicate timepoints of stored blood (PBMC/plasma) samples. B: Maximum Likelihood Phylogenetic analysis of env nucleotide sequences (amplicons spanning AA35-683, HXB2 numbering) isolated from the index participant. Sequences are coloured according to the sampling timepoint. Red nodes represent bootstrap support > 70%. Clade B sequences (NL4-3 and Ba-L) were included as outgroups for phylogeny rooting. Root branches have been shortened to prevent large branches caused by differences between participant sequences and outgroups. C-D: Maximum Likelihood Phylogenetic analysis of env nucleotide sequences (amplicons spanning AA35-683, HXB2 numbering) isolated from an individual on ART at one timepoint (C) or the index participant at the viral blip timepoint (D). Red nodes represent bootstrap support > 70%. E: Anti-Env IgG titre in the plasma from the index and a HIV-negative participant. Plasma titre was assessed via semi-quantitative ELISA against the 1A3, 1D5 and 2E4 Env proteins (encoded by env sequences isolated from the index participant). F: Pseudovirus neutralization 50% inhibitory dilution (ID50) titres of plasma from the index and HIV-negative participant, respectively. Pseudovirus was expressed using autologous env sequences cloned from the index participant including 6 pseudoviruses from the pre blip timepoint (month 0), 8 pseudoviruses from the blip timepoint (month 41) and 8 pseudoviruses from the post blip timepoint (month 65). The dotted line represents the detection limit of the assay (ID50 = 1:100). If no neutralization was detected, samples were assigned an ID50 of 0. If only a single data point of the plasma dilution series was above 1:100, no accurate ID50 estimation could be determined and thus samples were assigned an ID50 of 1:100. The box plot lines show the median and 10-90th percentile.
Fig 3.
Transient viral blip induces expansion of Env-reactive IgG+ memory B cells.
A: Gating strategy for FACS-sorting of 1A3 Env-PE and 1D5 Env-BV786 specific memory B cells (Aqua live/Dead-, CD4-, CD19 + , IgM- IgG+) from index participant at the blip (month 41) and post-blip (month 65) timepoints, or a HIV negative participant. B: Percentage of Env-reactive B cells within IgG + MBCs stratified by participant. C: Proportion of MBC subsets (resting, activated, TLM, intermediate) within total IgG + MBCs in PLWH with chronic viremia (n = 11) and HIV-negative (n = 15) controls are shown on the left panel. The analysis was based on flow-cytometric phenotyping of the expression of CD21 and CD27. Statistical significance was assessed using Mann-Whitney U tests for each subset in turn between the chronic viremia and HIV-negative donors (p > 0.05 not marked, p ≤ 0.05 *, p < 0.01, **, p < 0.001 ***, p < 0.0001 ****). The right panel shows data on these MBC subsets for the elite controller during the viral blip (divided into Env+ and Env- cells) compared to the post-blip timepoint. D: UMAP visualization of single cell transcriptomes from TLM, activated and resting MBCs (n = 301) sorted from the elite controller post blip and Env+ memory B cells from the blip timepoint. Single cell libraries were prepared via the Smart-seq 2 pipeline [32], a total of 88 resting, 88 activated, 88 TLM and 161 Env + MBC are profiled. Cells are annotated by their sorting identity. E: Expression of selected genes associated with memory B cell subsets profiled in (D). The fraction of cells is shown by the dot size, and the mean expression level is reflected by the colour. F: Similarity of single-cell transcriptomes of Env + IgG+ memory B cells with the resting, activated and TLM IgG + B cell subsets sequenced in (D). Similarity is calculated as probability using the Celltypist annotation tool trained on the sorted post-blip MBC subsets. G-H: IgG subclass usage (G) and nucleotide somatic hypermutation levels (V + J gene), (H)of sequenced memory B cell populations shown in (D). One way-ANOVA with Tukey’s multiple comparison used for statistical testing. *p < 0.032, **p < 0.021, ***p < 0.0002 and ****p < 0.0001.
Fig 4.
Transient viral blip induces a complex response across both B- and non-B cell subsets.
A: UMAP visualization of annotated single-cell transcriptomes (48,602 cells) recovered from circulating MACS-selected B cells spiked with PBMC in 3:1 ratio of the index participant (elite controller) across all timepoints. B: Bar plot showing proportion (% of non-B cells) of non-B cell subsets across all three timepoints. C-D: Dot plot showing expression of selected marker genes (C) and CITE-seq protein markers (D) across annotated cell clusters from (A). The fraction of cells expressing a particular marker is shown by dot size and the marker mean expression level is reflected by the colour. E: UMAP visualization of finely annotated B cell transcriptomes (42,149 cells) extracted from the dataset shown in (A). F: Bar plot showing proportion (% of B cells) of annotated B cell subsets across all three timepoints. G-H: Dot plot showing expression of selected marker genes (G) and CITE-seq protein markers (H) across annotated B cell clusters from (E). The fraction of cells expressing a particular marker is shown by dot size and the marker mean expression level is reflected by the color. I: Box plots showing number of inferred cell-cell interactions (left) and interaction strength (right) across timepoints using CellChat package. Both B- and non-B cells included. J: Isotype usage analysis depicted as % of B cells within each subset and timepoint. K: Violin plot showing VDJ mutation frequency stratified by subset and timepoint. L: GSEA of selected Hallmark and KEGG gene sets based on pre-ranked DEG comparing all B cells from blip (red dots) and post-blip timepoint (blue dots) to the reference (pre-blip B cells). Dot size reflects normalized enrichment score (NES). Vertical black lines indicate the threshold for statistical significance. M: Heatmap showing scanpy expression score of selected HALLMARK, KEGG and GOBP gene sets in each B cell subset and timepoints. Scaled by row.
Fig 5.
TLM subsets are heterogenous and display innate-like B cell characteristics.
A: PCA based on VJ gene usage by elite controller B cells coloured by B cell subset and timepoint. B: Bar chart showing percentage of B cells using IGHV4-34 segment within each subset and timepoint. C: Bar chart showing clonality (clone size Gini index) of each B cell subset stratified by timepoint. D: Circos-style plots depicting clonal overlap between each B cell subset pair stratified within each timepoint. The thickness of the grey connecting line between circles reflects the number of overlapping BCR clonotypes. E: Single-cell BCR network plots of all pre-blip clones consisting of at least three cells, coloured by B cell subset annotation (left), heavy chain isotype class use (middle) and mutation count (right). Each circle/node represents a single B cell with a corresponding set of BCR(s). Cells belonging to the same clone are connected with a black line. F: UMAP visualization of all B cells described in Fig 4, coloured by GEX-based pseudotime and overlayed by inferred developmental lineages (black arrows), generated by slingshot package. G: UMAP generated by the BCR-seq-based pseudobulk VDJ trajectory analysis (dandelion package) of B cell dataset described in Fig 4, coloured by calculated pseudotime and B cell subset annotation.
Table 1.
Smart Seq 2 and Env sequencing primer sequences.
Table 2.
B cell phenotyping flow cytometry and CITE-Seq panels. Where suppliers do not provide antibody concentrations the volume of antibody used (μL) is provided.
Table 3.
Bulk BCR sequencing primers.