Fig 2: Identification of Tonsillar and Circulating Tfh Populations in Donor-Matched SamplesMatched PBMC and MNC-T from adult donors were stained ex vivo.(A) Dot plot illustrating the expression of CXCR5, PD-1, CXCR3, and ICOS within the total CD4+CD45RA− cells in a representative matched blood and tonsil sample pair.(B) Heatmaps showing the mean percentage of CXCR3+ and ICOS+ cells within populations expressing different combinations of CXCR5 and PD-1 in 12 donors. The white crossed boxes indicate the near-absence of the PD-1hiCXCR5hi population in peripheral blood.(C) Dot plot illustrating the gating strategy used to define populations of circulating CXCR5+PD-1+Tfh (cTfh CXCR3+, cTfh CXCR3−) and CXCR5− non-Tfh memory cells (cNon-Tfh) and tonsillar CXCR5hiPD-1hi GC Tfh (tTfh GC), CXCR5intPD-1int (tTfh CXCR3+, tTfh CXCR3−) Tfh and CXCR5−PD-1− non-Tfh memory cells (tNon-Tfh) within the total CD4+CD45RA− T cell pool in one representative donor.(D) The percentage of each Tfh/non-Tfh subset within the CD4+CD45RA− memory T cell population in blood and tonsils from the 12 donors.(E–G) Production of IL-21 (E), IL-4 (F), and IFN-γ (G) after PMA and ionomycin stimulation by Tfh and non-Tfh populations in 10 donors. Non-analogous subsets were compared within tonsil and peripheral blood using the Kruskal-Wallis test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Boxes show means and quartiles, and whiskers show minima and maxima.See also Figure S1.

Fig 3: CXCR5-PD-1+ICOS+ T Cells Contain HIV-Specific T Cells that Are Capable of Promoting Both HIV-Specific and Non-HIV-Specific Antibody Responses(A) Gag-specific T cells within CXCR5-PD-1+ICOS+ T cells or TFH cells were identified by OX40 and CD25 expression in LN cells stimulated with pooled Gag peptides for 18 h. Plots show representative data from one HIV LN.(B) Quantification of CD25hiOX40hi T cell frequency. Background level of CD25hiOX40hi T cells in vehicle-treated wells was subtracted from Gag-stimulated wells (n = 7).(C) Plasma cell phenotype of B cells cocultured with the indicated T cell subsets in the presence of Gag/Env peptides for 7 days. Representative data from one individual is shown.(D and E) Quantification of IgG concentration (D) or plasma cell number (E) in T:B cell cocultures after 7 days of Gag/Env peptide stimulation (n = 15).(F) Summary plot comparing total IgG antibody concentration with concentration of antibodies that recognize Gag/Env peptides or proteins. Cocultures contained either CXCR5-PD-1+ICOS+ T cells or TFH cells and were stimulated by Env/Gag peptides.(G) Bar graphs show the concentration of antibodies specific for Gag/Env protein asapercentage of total IgG produced by B cells in the presence of HIV peptide-stimulated CXCR5-PD-1+ICOS+ T cells or TFH cells. Each bar indicates one individual.Friedman test was performed and corrected using Dunn’s multiple comparisons test. Data are represented as mean ± SEM.Also see Figure S6.

Fig 4: CXCR5-PD-1+ICOS+ T Cells in the Blood Are Clonal Relatives of Cells in the LNs(A) Circos plots of TCR sequence overlap among different populations. Each thin slice of the arc represents a unique TCR sequence. Each plot represents data from one individual.(B) Bhattacharyya coefficient measurement for the TCR repertoire similarity between CXCR5-PD-1+ICOS+ T cells in the PBMC and the indicated T cell subsets from the LN. Similarity with naive cells was used as the baseline for comparison (n = 6).(C) Correlation between CXCR5-PD-1+ICOS+ T cells in the PBMC versus different LN subsets (n = 12 paired PBMC and LNs).(D) Correlation between CXCR5-PD-1+ICOS+ T cells in the PBMC and CD4:CD8 ratio. Data are from 12 HIV+ patients.(E) Sorted CD4+ T cell subsets from the PBMCs of HIV-infected individuals were cultured with B cells in the presence of SEB for 7 days. Plots show representative staining for plasma cells in cocultures.(F) Scatterplot quantifies IgG level and plasma cell number in cocultures with naive T cells (n = 11), CXCR5-PD-1-ICOS- T cells (n = 10), CXCR5-PD-1+ICOS+ T cells (n = 9), or cTFH cells (n = 7).For (B), Friedman test was performed and corrected using Dunn’s multiple comparisons test. For (C) and (D), association was measured by Pearson correlation.For (F), Kruskal-Wallis test was performed and corrected using Dunn’s multiple comparisons test. Data are represented as mean ± SEM.

Fig 5: CXCR5-PD-1+ICOS+ T Cells in HIV-Infected LNs Express TFH-Cell-Related Proteins and Function(A) Histograms showing fluorescent intensity of BCL6, Maf, and CD84 staining in naive (1), CXCR5-PD-1-ICOS- (2), CXCR5-PD-1+ICOS+ (3), and TFH cells (4). Plots summarize data from 8 HIV LNs.(B) Plasma cell phenotype of B cells cocultured with the indicated T cell subsets without or with SEB.(C-E) Quantification of total IgG (n = 13) (C), plasma cells (n = 15) (D), or CD4+ T cells (n = 15) (E) on the 7th day of each coculture condition.(F) IFN-? level in the supernatant of wells containing TFH cells or CXCR5-PD-1+ICOS+ T cells (n = 9).(G) Example gate for T-bet+ B cells. Naive IgD+CD27- B cells were excluded. Scatterplot quantifies T-bet+ B cells in 7 HC and 25 HIV-infected LNs.(H) Correlation between IFN-? expression in CXCR5-PD-1+ICOS+ T cells and T-bet+ B cell frequency in HIV-infected LN samples.For (A) and (C)–(E), Friedman test was performed and corrected using Dunn’s multiple comparisons test. For (F), Wilcoxon matched-pairs signed rank test was used. For (G), Mann-Whitney test was used. For (H), association was measured by Spearman correlation. Data are represented as mean ± SEM.Also see Figures S4D and S5A.