Overexpression of T-bet in HIV infection is associated with accumulation of B cells outside germinal centers and poor affinity maturation

Abstract

Nearly all chronic human infections are associated with alterations in the memory B cell (MBC) compartment, including a large expansion of CD19^hiT-bet^hi MBC in the peripheral blood of HIV-infected individuals with chronic viremia. Despite their prevalence, it is unclear how these B cells arise and whether they contribute to the inefficiency of antibody-mediated immunity in chronic infectious diseases. We addressed these questions by characterizing T-bet-expressing B cells in lymph nodes (LN) and identifying a strong T-bet signature among HIV-specific MBC associated with poor immunologic outcome. Confocal microscopy and quantitative imaging revealed that T-bet^hi B cells in LN of HIV-infected chronically viremic individuals distinctly accumulated outside germinal centers (GC), which are critical for optimal antibody responses. In single-cell analyses, LN T-bet^hi B cells of HIV-infected individuals were almost exclusively found among CD19^hi MBC and expressed reduced GC-homing receptors. Furthermore, HIV-specific B cells of infected individuals were enriched among LN CD19^hiT-bet^hi MBC and displayed a distinct transcriptome, with features similar to CD19^hiT-bet^hi MBC in blood and LN GC B cells (GCBC). LN CD19^hiT-bet^hi MBC were also related to GCBC by B cell receptor (BCR)-based phylogenetic linkage but had lower BCR mutation frequencies and reduced HIV-neutralizing capacity, consistent with diminished participation in GC-mediated affinity selection. Thus, in the setting of chronic immune activation associated with HIV viremia, failure of HIV-specific B cells to enter or remain in GC may help explain the rarity of high-affinity protective antibodies.

Fig. 1.. Accumulation of T-bet⁺ B cells in non-GC areas of LN in HIV infection.

(A and B) Images of cells expressing T-bet alone or in combination with staining for CD20⁺ B cells, CD4⁺, or CD8⁺ T cells in LN sections isolated from HIV-uninfected (A and B, right) and HIV-infected (B, left) individuals. Red asterisks indicate enlarged images from the red boxed inset illustrating differences in T-bet expression patterns between cell populations. (C) Frequencies of T-bet⁺ cells among B and T cells were quantified by histocytometry in LN sections from HIV-infected (n = 12) and HIV-uninfected (n = 8) individuals. (D) Histocytometry plots and gating strategy for defining areas within LN sections. Ki-67 expression was used to define GC (Ki-67⁺IgD⁻) and non-GC (Ki-67⁻) zones within CD20⁺ follicles, and the extrafollicular (EF) zone was defined by Boolean gate subtraction. (E) Frequencies of CD20⁺ cells expressing T-bet (left graph) and numbers of CD20⁺T-bet⁺ (right graph) cells in each zone (n = 11 HIV-infected) were quantified by histocytometry. (F) Distance of CD20⁺T-bet⁺ cells to GC and non-GC zones (n = 11 HIV-infected). In (C), (E), and (F), each individual is color-coded per Table 1, and shaded bars represent medians. *P < 0.05, **P < 0.01, ***P < 0.001 by Wilcoxon matched-pairs signed rank test after obtaining significance by Friedman analysis of variance (ANOVA) test on full set; ns, not significant.

Fig. 2.. T-bet expression and signature profile among CD19^hi LN MBC of HIV-infected individuals.

(A) Heatmap depicting expression of several markers based on T-bet expression in defined populations from histocytometric analyses. (B) Frequencies of CD19^hi (left) and T-bet intensities (right) among defined populations measured by conventional flow cytometry on LN mononuclear cells isolated from HIV-infected individuals (n = 15 and 13, respectively). (C) Quantitative analyses of mean fluorescence intensity (MFI) for each signature-defining marker by B cell population or CD3⁺ cells performed by conventional flow cytometry on LN mononuclear cells (n = 8 to 10). In (B) and (C), each individual is color-coded per Table 1, and black horizontal bars represent medians. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by Wilcoxon matched-pairs signed rank test after obtaining significance by Friedman ANOVA test on B cell subpopulations.

Fig. 3.. Unique transcriptional profile for HIV-specific MBC in LN of infected individuals.

(A) Frequencies of HIV gp140⁺ B cells among CD19^hi and CD19^lo MBC and GCBC measured by conventional flow cytometry on LN mononuclear cells isolated from HIV-infected individuals (n = 13). Each individual is color-coded per Table 1, and black horizontal bars represent medians. **P < 0.01, ***P < 0.001 by Wilcoxon matched-pairs signed rank test after obtaining significance by Friedman ANOVA test on full set. (B) Sorting strategy used to isolate multiple LN B cell populations used for RNA-seq (each population is indicated by color-coded boxes). The numbers refer to the percentage of cells within the gated population relative to the total number of cells shown in the dot plot. (C) Unsupervised principal components analysis of sorted populations using all expressed genes, performed on HIV-infected participants HIV1, HIV3, HIV5, and HIV7 (Table 1) and HIV-uninfected participants (n = 5). (D) Heatmap depicting scaled, normalized log₂ expression values of selected differentially expressed genes. (E) Gene set enrichment analysis of GCBC (top), memory precursor (middle), and atypical B cell signature genes (bottom) ranked by differential expression between HIV gp140⁺ and total MBC in HIV-infected individuals; NES, normalized enrichment score.

Fig. 4.. Profile of T-bet⁺ MBC in LN of HIV-infected individuals similar to GCBC but consistent with non-GC location.

(A) Expression of CD95 evaluated by conventional flow cytometry performed on LN mononuclear cells isolated from HIV-infected individuals (n = 12). (B) Apoptosis depicted and quantified by conventional flow cytometry performed on LN mononuclear cells isolated from HIV-infected individuals (n = 7), after incubation for 30 min with CD95L or CD40L (control). (C) Expression of CXCR5 and CXCR4 evaluated by conventional flow cytometry performed on LN mononuclear cells isolated from HIV-infected individuals (n = 9 to 12). (D) Images obtained by confocal microscopy depicting expression patterns of CD20, CXCL13, and T-bet in an LN section from an HIV-infected individual. Red asterisks are enlarged images from the red boxed inset illustrating distinct areas of expression for T-bet⁺ and CXCL13⁺ cells. (E) Imaged data from (D) were converted to flow data with gating scheme shown for delineation of CD20⁺ cells and depiction of CXCL13 by T-bet expression in CD20⁺ cells in dot plot and location. Quantification of frequencies of single or double CXCL13 and T-bet expression performed by histo-cytometry on LN sections from HIV-infected individuals (n = 9). In (A) to (C) and (E), each individual is color-coded per Table 1, and black horizontal bars represent medians. *P < 0.05, **P < 0.01, ***P < 0.001 by Wilcoxon matched-pairs signed rank test after obtaining significance by Friedman ANOVA test on B cell subpopulations.

Fig. 5.. T_H1 and T_FH cytokines modulate expression of T-bet and migration receptors.

(A) Conventional flow cytometry performed on LN GCBC and CD19^hi MBC depicting the expression of Bcl6, T-bet, and CXCR5 expression of an HIV-infected individual. Red pattern and numbers identify Bcl6 frequencies within each quadrant. (B) Conventional flow cytometry performed on tonsillar B cells isolated from HIV-negative donors depicting Bcl6 and T-bet expression before and after 48 hours of stimulation of IgD⁺ cells with anti-human BCR and CD40L with or without IFN-γ or IL-21. (C) Histograms and comparison (n = 10) of T-bet and CXCR5 protein expression under conditions in (B). (D) Comparison (n = 6) of gene expression for TBX21, S1PR2, and BACH2 under conditions in (B). *P < 0.05, **P < 0.01 by Wilcoxon matched-pairs signed ranked test.

Fig. 6.. BCR of CD19^hi MBC related to GCBC but with decreased mutation frequency and neutralization capacity.

(A) Determination of IGHV mutation frequencies (mut freq) among B cell populations sorted from LN of three HIV-infected individuals (described in Table 1). Data are presented as box-and-whiskers plots showing the two inner quartiles (boxes) with median (black lines), 1.5 times the inner quartile range (whiskers) and outliers (black circles). (B) HIV pseudotype neutralization by clonally related or epitope target-related mAbs reconstituted from HIV-specific MBC and GCBC of four HIV-infected individuals (related clones/targets are color-coded by individual per table S1). Each value is % neutralization of one clone or median neutralization of 2 to 11 related clones from each source. **P < 0.01 by Wilcoxon matched-pairs signed rank test. (C) Breadth of serologic HIV pseudotype neutralization activity correlated with frequency of CD19^hi LN MBC from HIV-infected individuals (n = 9, each individual is color-coded per Table 1). (D) Heatmaps indicating the numbers of shared clonal families (top triangle) and sequences (bottom triangle) from three HIV-infected individuals (identified in Table 1). Along the diagonal are total numbers of clones (top) or unique sequences (bottom). Percentages shown in parentheses were calculated by dividing the number of shared clones or unique sequences by the total in the smaller of the two samples. (E) Scatter-plots show the median IGHV mutation frequency of the two indicated populations for each shared clone from HIV-infected participant HIV1 (Table 1). Green equivalence and blue linear regression lines are depicted. The density (den) histograms show the mutation frequency of all sequences (bars) or sequences within shared clones between the two populations (dashed lines). P values by Wilcoxon signed rank test followed by Bonferroni correction. (F) Example of a multipopulation lineage tree with predicted internal node types. (G) Summary of predicted switches within lineage trees. Blue lines indicate more observed than expected switches; red lines indicate fewer observed than expected switches.