Divergent cytokine and transcriptional signatures control functional T follicular helper cell heterogeneity

Abstract

CD4⁺ T follicular helper (T_FH) cells support tailored B cell responses against multiple classes of pathogens. To reveal how diverse T_FH phenotypes are established, we profiled mouse T_FH cells in response to viral, helminth and bacterial infection. We identified a core T_FH signature that is distinct from CD4⁺ T follicular regulatory and effector cells and identified pathogen-specific transcriptional modules that shape T_FH function. Cytokine-transcriptional T_FH programming demonstrated that type I interferon and TGFβ signaling direct individual T_FH phenotypes to instruct B cell output. Cytokine-directed T_FH transcriptional phenotypes are shared within human germinal centers, but distinct T_FH phenotypes dominate between donors and following immune challenge or in antibody-mediated disease. Finally, we identified new cell surface markers that align with distinct T_FH phenotypes. Thus, we provide a comprehensive resource of T_FH diversity in humans and mice to enable immune monitoring during infection and disease and to inform the development of context-specific vaccines.

Fig. 1. Functionally heterogeneous T_FH phenotypes are induced by diverse pathogen infections.

a–i, Analysis of draining lymph node cells from wild-type (a–c,h,i) and ZsGreen_T-bet reporter (d–g) mice infected with the indicated pathogens at the early peak GC response (LCMV day 12, influenza day 10, T. muris day 21, H. polygyrus day 12 and C. rodentium day 12). Representative plots of CD4⁺CD44⁺Ly6C⁺CD162⁺ T_eff cells and CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells with histograms displaying Bcl-6 expression (a). Frequencies of T_FH cells in CD4⁺CD44⁺ gate (b) and the ratio of T_FH:T_eff cells (LCMV n = 8, influenza n = 8, T. muris n = 6, H. polygyrus n = 9, C. rodentium n = 10 mice per group) (c). Representative plots of ZsGreen_T-bet reporter expression (T_FH cells are blue, T_eff cells are red) (d). ZsGreen_T-bet reporter⁺ frequency and gMFI of CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells (LCMV n = 8, influenza n = 8, T. muris n = 6, H. polygyrus n = 10, C. rodentium n = 7 mice per group) (e). Immunofluorescence staining of draining lymph node GCs. Red arrows indicate ZsGreen_T-bet reporter⁺ T_FH cells. Yellow, CD4; blue, IgD; magenta, GL7; green, ZsGreen_T-bet reporter. Scale bar, 200 μm (f). Correlation of ZsGreen_T-bet reporter expression with the T_eff:T_FH cell ratio across infections (g). Frequency of T_FH cells that produced IFNγ, IL-4 or IL-17A (h). IFNγ⁺ T_FH1, IL-4⁺ T_FH2, and IL-17⁺ T_FH17 cells are included in the entire T_FH population (i). The inner slice displays cytokine coexpression. The outer ring (green) indicates the proportion of ZsGreen_T-bet⁺ T_FH cells expressed. The data are from 6–10 mice per group and are presented as the mean ± s.e.m. Statistical tests included one-way analysis of variance (ANOVA) for multiple comparisons and Pearson correlation with two-tailed P values. ****P < 0.0001 or otherwise indicated. Source Data

Fig. 2. Diverse pathogen infection tailors B cell responses.

a–d, Analysis of draining lymph node cells (a–c) and serum (d) from wild-type (a,b) and ZsGreen_T-bet (c) reporter mice infected with the indicated pathogens at the early peak GC response. Analysis of B220⁺IgD^loCD95⁺CD38⁻ GC B cells (a) and B220⁺IgD^loCD95⁻CD38⁺ MBCs (LCMV n = 8; influenza n = 8; T. muris n = 6; H. polygyrus n = 9; and C. rodentium n = 8 mice per group) (b). Correlation of the frequency of MBC with the ratio of T_eff:T_FH cells and ZsGreen_T-bet reporter expression gMFI across infections (c). Serum IgG isotype concentration (d). The data are from 6–10 mice per group and are presented as the mean ± s.e.m. Statistical tests included one-way ANOVA for multiple comparisons and Pearson correlation with two-tailed P values. ****P < 0.0001 or otherwise indicated. Source Data

Fig. 3. Identification of conserved T_FH and T_FR signatures.

a–j, Bulk RNA-seq of sorted CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁻IL-21–GFP⁺ T_FH, CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁻IL-21–GFP⁻ T_FH, CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁺IL-21–GFP⁺ T_FR, and CD4⁺CD44⁺PD-1⁻CXCR5⁻FoxP3-RFP⁻IL-21–GFP⁻ T_eff cells from the draining lymph nodes of the mice infected with pathogens described in Fig. 1. MA plot visualizing the log fold change (FC) in the mean expression of genes expressed differentially between the T_FH and T_eff populations for the five infections (a). UpSet plot showing intersections of genes differentially expressed between the T_FH and T_eff populations for each infection (b). Common DE genes across three or more infections were used to derive the core T_FH signature. False discovery rate (FDR) < 0.05. Heatmap of core T_FH signature genes (row-based z scores of normalized log₂ counts per million) for cytokine and cell surface receptor genes (c) and transcriptional regulator genes (d). GSEA of differentially expressed genes in T_FH (T_FH versus T_eff comparison), T_FR (T_FR versus T_eff), and T_eff cells (T_eff versus T_FH) for all infections (e,f). NES of T_FH, non-T_FH, GC T_FH, GC B, human T_FH and cancer-associated T_FH cell programs (e) and precursors of exhausted (T_PEX), exhausted progenitor (T_PROG) and long-term hematopoietic stem cell (HSC-LT) gene programs in the T_FH, T_FR and T_eff populations (f). The NES score represents the enrichment of genes (sets) relative to each comparison, correcting for multiple testing. UpSet plot of the T_FR signature showing intersections of upregulated genes expressed differentially between the T_FR versus T_FH cell contrast population and the T_FR versus T_eff cell contrast population for five infections (g). FDR < 0.05. Heatmap of T_FR signature genes shared across contrasts (T_FR versus T_FH and T_FR versus T_eff) (row-based z score of normalized log₂ counts per million) for cytokine and cell surface receptor genes (h) and transcriptional regulator genes (i). Bubble plot of Bcl-6 network transcriptional regulator genes in the T_FH core showing log₂FC differences in the T_FH (T_FH versus T_eff contrast) and T_FR (T_FR versus T_eff contrast) populations (j). The size of the bubble represents the −log₁₀(FDR), and the color indicates the log₂FC compared to T_eff cells. The colored bar indicates genes present in the T_FH core and Bcl-6 gene sets from Extended Data Fig. 5g. Novel genes proposed to be independent of the Bcl-6 network are indicated with gray bars. The data represent independent samples of 2–3 per cell type per infection. Source Data

Fig. 4. Identification of pathogen-specific T_FH phenotypes.

a–g, Bulk RNA-seq of sorted CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁻IL-21–GFP⁺ T_FH, CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁻IL-21–GFP⁻ T_FH cells from the draining lymph nodes of mice infected with the indicated pathogens (as in Fig. 1). MDS plot of CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁻IL-21–GFP⁺ T_FH transcriptomes showing the separation of samples by infection type along dims 1 and 2 (a). Genes expressed differentially for T_FH (single infection vs. all other infections) and T_eff cells (single infection versus all other infections) (FDR < 0.05) (b). Orange, common pathogen-specific DE genes in both T_FH and T_eff cells; blue, pathogen-specific signatures unique to T_FH cells; red, pathogen-specific signatures unique to T_eff cells. Heatmaps of pathogen-specific T_FH signature genes (average row-based z score of normalized log₂ counts per million) for selected cytokine and cell surface receptor genes (c) and transcriptional regulator genes (d). Cytokine pathway genes from Mouse Genome Informatic (MGI) database gene sets indicated in colored circles. IFN (red), TNF (purple), IL-1 (pink), TGFβ (blue), IL-4 (green), IL-6 (yellow) and IL-17 (orange). Heatmap of pathogen-specific T_FH signature genes (average row-based z score of normalized log₂ counts per million) for selected signature-defining cytokines and chemokines for each infection (e) and lineage-defining transcription factors (f). GSEA of pathogen-specific signatures via MGI MSigDB ‘Hallmark’ gene sets for cytokine signaling and response (g). Bubble size represents the -log₁₀(FDR), and color indicates the NES. The data represent independent samples of 2–3 per cell type per infection. h, Analysis of draining lymph node CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells from wild-type mice infected with indicated pathogens (as in Fig. 1) displaying gMFI of pathogen-specific T_FH signature marker (LCMV n = 7; T. muris n = 5; and C. rodentium n = 7 mice per group). Data show experiments of 5–7 mice per group and mean ± s.e.m. Statistical tests were one-way ANOVA of multiple comparisons and Pearson correlation with two-tailed P values. ****P < 0.0001 or otherwise indicated. Source Data

Fig. 5. Pathogen-specific T_FH phenotypes are directed by distinct cytokine signaling pathways.

a–d, Analysis of draining lymph nodes from Tgfbr2-Lck^Cre and Lck^Cre control mice infected with influenza A virus in intact mice (a,c,d) and 50:50 bone marrow chimera mice (b). CXCR3⁺ T_FH1, CXCR3⁻CCR6⁻ T_FH2, and CCR6⁺ T_FH17 cells within CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells are displayed as parts of the whole population (a,b). Frequency within T_FH population, and representative histograms of CXCR3 and CCR6 expression on T_FH cells (n = 9–10 mice per group) (a). The inner slice of the pie chart displays CXCR3⁺CCR6⁺ coexpression. Wild-type and Tgfbr2-Lck^Cre cells were identified by congenic marker expression (b). Frequency of CXCR3⁺ T_FH1, CXCR3⁻CCR6⁻ T_FH2, and CCR6⁺ T_FH17 cells and frequency of IFNγ⁺ T_FH1 cells within CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells (n = 10 mice per group). The inner slice of the pie chart displays CXCR3⁺CCR6⁺ coexpression. IgG1 or IgG2c class switched B220⁺IgD^loCD95⁺ GC B cells and CD86⁻CXCR4⁺ dark zone (DZ) or CD86⁺CXCR4⁻ light zone (LZ) B cells on B220⁺IgD^loCD95⁺ GC B cells (n = 7–10 mice per group) (c). Serum IgG isotype concentration (d). e–h Draining lymph node analysis of Ifnar^−/− and Ifnar ^+/+ control mice infected with LCMV in intact mice (e,g,h) and 50:50 bone marrow chimera mice (f). CXCR3⁺ T_FH1, CXCR3⁻CCR6⁻ T_FH2, and CCR6⁺ T_FH17 cells within CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells are displayed as parts of the whole population (e,f). Frequency within T_FH population, and representative histograms of CXCR3 and CCR6 expression on T_FH cells (n = 7 mice per group) (e). The inner slice of the pie chart displays CXCR3⁺CCR6⁺ coexpression. Wild-type and Ifnar^−/− cells were identified by congenic marker expression (f). Frequency of CXCR3⁺ T_FH1, CXCR3⁻CCR6⁻ T_FH2, and CCR6⁺ T_FH17 cells and frequency of IFNγ⁺ T_FH1 cells within CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells (n = 8 mice per group). The inner slice of the pie chart displays CXCR3⁺CCR6⁺ coexpression. IgG1 or IgG2c class switched B220⁺IgD^loCD95⁺ GC B cells and CD86⁻CXCR4⁺ DZ or CD86⁺CXCR4⁻ LZ B cells from B220⁺IgD^loCD95⁺ GC B cells (n = 7 mice per group) (g). Serum IgG isotype concentration (h). The data are from 7–10 mice per group and are presented as the mean ± s.e.m. Statistical tests were a two-tailed unpaired Student’s t-test for the intact system and a two-tailed paired Student’s t-test for the bone marrow chimera model. ****P < 0.0001 or otherwise indicated. Source Data

Fig. 6. Cytokine-guided T_FH phenotypes are evident in human lymphoid tissue.

a–e, scRNA-seq of sorted CD3⁺CD4⁺CD45RA⁻CD45RO⁺CXCR5⁺CD27⁺ human tonsillar T_FH cells from three healthy adult donors. UMAP dimensional reduction of data depicting 11 clusters on the basis of Louvain clustering via the Jaccard similarity index (k = 9) (a). Overlay of the mean ranked scores of each pathogen-specific signature onto UMAP clusters (b). Mean of the ‘TotalScore’ from the singscore: simpleScore function collated for each cluster. Pathogen-specific upregulated signature genes used. Rank scores of pathogen-specific signatures for each pseudosample on the basis of cluster and individual donors (c). Influenza A versus LCMV (left); LCMV versus T. muris (middle); influenza A versus H. polygyrus (right). Heatmap of the mean rank scores of pathogen-specific signatures for each cluster (row-based z score of the mean rank score) (d). GSEA-vissE analysis for the comparison of C7 versus C8 (e). Top gene sets of selected clusters as bar plots of DE gene counts with corresponding gene statistics (FDR as color shade) from the DE analysis. GSEA was performed via the Hallmarks c2 (‘CP:REACTOME’, ‘CP:PID’, ‘CP:BIOCARTA’,‘CP:KEGG’) and c5 (‘GO:BP’,‘GO:MF’) collections from the MSigDB using a two-tailed approach correcting for multiple testing with the FDR adjusted to P ≤ 0.05. f, Ranked score of pathogen-specific signatures in human dataset T_FH cells from donors in SARS-CoV-2 vaccine (n = 5), influenza vaccine (n = 1), malaria infection (adult n = 3, child n = 3), peanut allergy (n = 3), asthma (n = 4) and autoimmune contexts (healthy donor n = 6, systemic lupus erythematosus n = 8). Ranked scores normalized to the first time point for signatures. Data are presented as mean ± s.e.m. g, Visium spatial tonsil data displaying a ranked score of pathogen-specific signatures onto GCs highlighted by the core T_FH cell signature. h, Xenium spatial human lymph node data displaying a ranked score of pathogen-specific signatures in T_FH cells and GCs highlighted by the core T_FH cell signature. Source Data

Fig. 7. Cell surface proteins identify distinct human tonsillar T_FH and cT_FH phenotypes.

a–e, Flow cytometry analysis of CD3⁺CD4⁺CD45RA⁻CD45RO⁺CXCR5⁺ T_FH cells from five PBMC and six tonsil healthy adult donors. t-distributed stochastic neighbor embedding (t-SNE) and heatmap expression of T_FH cell markers of PD-1, CXCR3, CCR6 and CCR4 (a). T_FH cell populations identified by PD-1 and CD57 with expression of the T_FH cell markers CXCR5, ICOS, OX40 and CCR4 on each population (b). Histograms of PD-1, CD57, CD69 and CD82 expression on T_FH cells (c). Histograms of CD127, CD99, CD71 and CD151 expression on T_FH cells (d). T_FH populations separated into quadrants (Q1–Q4) based on expression of CD99 and CD151 markers (e). f, scCITEseq surface protein expression (log counts) of sorted CD3⁺CD4⁺CD45RA⁻CD45RO⁺CXCR5⁺CD27⁺ human tonsillar T_FH cells from three healthy adult donors (as in a–e) of cluster-identifying markers overlaid onto the scRNA-seq UMAP. g, T_FH cell surface and activation markers in CD3⁺CD4⁺CD45RA⁻CD45RO⁺CXCR5⁺ human T_FH cells from matched tonsil, adenoid tissue and PBMC from five healthy juvenile donors. h,i, SARS-CoV-2 Spike tetramer⁺ CD3⁺CD4⁺CD45RA⁻CXCR5⁺ cT_FH cells from SARS-CoV-2 infected donors (h) at time of infection (11 of 11 donors; n = 11) and 6 months convalescence (10 of 11 donors; n = 10) and nine COVID-19 mRNA vaccinated donors (n = 9) (i) at 7 days post-vaccine and 3 months post-vaccine. Frequency of Spike tetramer⁺PD-1⁺ T_FH cells, gMFI expression for CXCR3 and CCR6, and frequencies and dot plots of CD151/CD99 quadrant-separated T_FH cell populations. Data show experiments of 8–11 samples per group. Statistical test was a two-tailed paired Student’s t-test. P values are indicated. Source Data

Extended Data Fig. 1. Induction of T_FH response following diverse infection compared to steady state homeostasis.

a, Gating strategy to identify CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH and CD4⁺CD44⁺Ly6C⁺CD162⁺ T_EFF cells and Bcl-6 expression. b, Timecourse analysis of draining lymph node GC cells from wild-type mice infected with indicated pathogens displaying frequencies and counts of CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells (closed circle) and B220⁺IgD^loCD95⁺ GC B cells (open circle) (LCMV n = 3, influenza n = 4, H. polygyrus n = 4, C. rodentium n = 3 mice per group). T.muris data in. c, Steady state analysis of indicated lymph nodes from uninfected wild-type mice. CD4⁺CD44⁺ cell counts, frequency of CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells and frequency of T_FH cells produced IFNγ, IL-4, and IL-17 (n = 10 mice per group). d,e Analysis of draining lymph node cells from (d) wild-type and (e) ZsGreen_T-bet reporter mice infected with indicated pathogens at early peak GC response. (d) Total lymphocyte counts, CD4⁺CD44⁺ cell counts and frequency of CD4⁺CD44⁺ cells of live cells (LCMV n = 8, influenza n = 8, T. muris n = 6, H. polygyrus n = 9, C. rodentium n = 11 mice per group). (e) ZsGreen_T-bet reporter⁺ frequency and gMFI of CD4⁺CD44⁺Ly6C⁺CD162⁺ T_EFF cells (LCMV n = 8, influenza n = 8, T. muris n = 6, H. polygyrus n = 10, C. rodentium n = 7 mice per group). Data show experiments of 6–10 mice per group and mean ± SEM. Statistical tests: one-way ANOVA of multiple comparisons. **** P values are <0.0001 or otherwise indicated. Source Data

Extended Data Fig. 2. Conserved GC morphology in draining lymph nodes following infection with diverse pathogens.

Immunofluorescent staining of draining lymph node of infected ZsGreen_T-bet reporter mice. Boxes indicate zoom of GC shown in Fig. 1f. Data are representative of 2–3 mice per group. Yellow: CD4, blue: IgD, magenta. Scale bar consistent for all images, 200μm. Source Data

Extended Data Fig. 3. Heterogenous T_FH phenotypes display diverse cytokine profiles, similar to that of T_EFF and systemic cytokine milieu.

a–d Analysis of (a–c) draining lymph node cells and (d) serum from (a,b,d) wild-type and (c) IFNγ-GFPx4C13R reporter mice infected with indicated pathogens at early peak GC response. (a) Representative plots of CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cell IFNγ, IL-4, and IL-17 cytokine production. (b) Frequency of CD4⁺CD44⁺Ly6C⁺CD162⁺ T_EFF cell produced IFNg, IL-4, IL-17A. (c) IFNγ-GFPx4C13R reporter expression in CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells following T. muris infection. Representative plots, frequency of cytokine reporter with T_FH population and displayed as parts of whole T_FH population. Inner slice displays cytokine coexpression. (d) Serum cytokine concentration (pg/mL) in serum (n = 10 mice per group). Cytokines displayed as parts of total cytokines analyzed. Data show experiments of 7–10 mice per group and mean ± SEM. Statistical tests: one-way ANOVA of multiple comparisons. **** P values are <0.0001 or otherwise indicated. Source Data

Extended Data Fig. 4. Pathogen-induced T_FH phenotypes correlate tailored B cell responses.

a, Gating strategy to identify GC and MBC populations. b, Analysis of B220⁺IgD^loCD95⁺CD38⁻ GC B cells from ZsGreen_T-bet reporter mice infected with indicated pathogens at early peak GC response. (b) Correlation of frequency of GC B to the ratio of T_EFF/T_FH cells and to ZsGreen_T-bet reporter expression gMFI across infections. Data show experiments of 6–10 mice per group. Statistical tests: Pearson correlation of two-tailed P value. Source Data

Extended Data Fig. 5. Pathogen-specific transcriptional programs separate by cell type and infection and establishment of T_FH and T_FR core signatures.

RNA-seq of CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁻IL-21–GFP⁺ T_FH, CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁻IL-21–GFP⁻ T_FH, CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁺IL-21–GFP⁺ T_FR, and CD4⁺CD44⁺PD-1⁻CXCR5⁻FoxP3-RFP⁻IL-21–GFP⁻ T_EFF cells from draining lymph nodes of wild-type mice infected with indicated pathogens at early peak GC response. a, Gating strategy of sorted populations. b, Principal Component Analysis (PCA) plots of sorted IL-21⁺ T_FH, IL-21⁻ T_FH, T_FR, and T_EFF cell transcriptomes stratified by either (top) cell type or (bottom) infection type. Data show independent samples of 2–3 per cell type per infection. c, Analysis of draining lymph node cells from wild-type mice infected with indicated pathogens. gMFI of upregulated (CD200 and BTLA) and downregulated (CCR7) core T_FH signature markers for CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH cells. Data show experiments of 5–7 mice per group and mean ± SEM. d, UpSet plot of T_FR signature showing intersections of downregulated genes expressed differentially between T_FR vs. T_FH contrast and T_FR vs. T_EFF contrast for five infections. False discovery rate (FDR) < 0.05. e,f, MA plot visualizing log fold change against the mean expression of genes expressed differentially between (e) T_FR and T_FH cells and (f) T_FR and T_EFF cells for five infections. g, GSEA of differentially expressed genes in T_FH (T_FH vs. T_EFF contrast) and T_FR cells (T_FR vs. T_EFF contrast) combined for five infections. Barcode and ES plots displaying enrichment of genes (ES and NES) upregulated in core T_FH signature (T_FH Core Up), T_FH genes, genes indirectly regulated by Bcl-6 through repressor-of-repressor circuits (Indirect Bcl-6-rr), genes directly repressed by Bcl-6 (Direct Bcl-6-r), and Bcl-6 target genes (Bcl-6 targets) in T_FH cells (T_FH vs. T_EFF) contrast^,,,,. Statistical tests: one-way ANOVA of multiple comparisons and Pearson correlation with two-tailed P value. **** P values are <0.0001 or otherwise indicated. Source Data

Extended Data Fig. 6. IL-21⁺ T_FH and IL-21⁻ T_FH cells are transcriptionally similar across diverse infections.

a, UpSet plot displaying intersects of genes expressed differentially between IL-21⁺ T_FH and IL-21⁻ T_FH cells from separate infections in bulk RNA-seq analysis of CD4⁺CD44⁺PD-1⁺CXCR5⁺FoxP3-RFP⁻IL-21–GFP⁺ T_FH and CD4⁺CD44⁺PD-1⁺CXCR5⁺ FoxP3-RFP⁻IL-21–GFP⁻ T_FH cells as in Fig. 3. Three genes intersecting across 4 infections indicated with arrow. b-g, Analysis of CD4⁺CD44⁺PD-1⁺CXCR5⁺CD162⁻ T_FH cells and CD4⁺CD44⁺CD162⁺ T_EFF cells from draining lymph nodes from wild-type mice infected with indicated pathogens, LCMV or influenza A at early peak GC response. (b) Gating strategy to identify T_EFF, IL-21⁻ T_FH, and IL-21⁺ T_FH cells. (c) Representative plots and histograms of CD62L, CD127, and Ly6C staining on IL-21⁻ T_FH, IL-21⁺ T_FH cells and T_EFF cells in LCMV infection. (d-g) Frequency and gMFI of CD62L⁺, CD127⁺, and Ly6C⁺ on (d,f) IL-21⁻ T_FH, IL-21⁺ T_FH cells, frequency of total T_FH population (e,g) T_FH and T_EFF, frequency of CD4⁺CD44⁺ population. (d-e) in LCMV infection (n = 8 mice per group), (f-g) in influenza A infection (n = 6 mice per group). Data show experiments of 4–6 mice per group and mean ± SEM. Statistical tests: Two-tailed unpaired Student’s t test. **** P values are <0.0001 or otherwise indicated. Source Data

Extended Data Fig. 7. Pathogen-induced cytokine signaling influence T_FH and T_EFF bifurcation, T_FH subpopulations, and T_FH cells capacity to direct B cell and antibody output.

a, Gating strategy to identify CD4⁺CD44⁺CXCR5⁺PD-1⁺Ly6C⁻CD162⁻ T_FH1, T_FH2, T_FH17, and T_FH17.1 phenotypes based on expression of CXCR3 and CCR6 chemokine receptors. b-k, Analysis of draining lymph nodes from (b,c) wild-type, (d) Tgfbr2-Lck^Cre and Lck^Cre mice, (f) Ifnar^−/− and Ifnar^+/+ mice, (e,h,j,k) wild-type and Tgfbr2-Lck^Cre 50:50 bone marrow chimeras, or (g-i,k) wild-type and Ifnar^−/− 50:50 bone marrow chimera mice infected with (b,d,e,h,i,k) influenza or (c,f,g,h,j,k) LCMV. (b,c) gMFI expression of IFNg, IL-4, and IL-17 in T_FH1 (CXCR3⁺CCR6⁻), T_FH2 (CXCR3⁻CCR6⁻), T_FH17 (CXCR3⁻CCR6⁺) and T_FH17.1 (CXCR3⁺CCR6⁺) populations (influenza n = 10, LCMV n = 6 mice per group). (d,f) T_FH and T_EFF cell counts, T_FH and T_EFF cell frequencies, and T_FH/T_EFF cell ratio (influenza n = 9-10, LCMV n = 7 mice per group). (e,g) T_FH and T_EFF cell frequencies and T_FH/T_EFF cell ratio (influenza n = 10, LCMV n = 8 mice per group). (h) Comparison of effect of Tgfbr2-Lck^Cre and Ifnar ^−/− on T_FH frequency in chimeras. (i-j) Frequency of T_FH1, T_FH2, T_FH17 and T_FH17.1 cells within T_FH population in chimeric mice (influenza n = 8, LCMV n = 7 mice per group). (k) Comparison of effect of Tgfbr2-Lck^Cre and Ifnar ^−/− on T_FH1, T_FH2 and T_FH17 cell frequency in influenza and LCMV infection in chimeras (Tgfbr2-Lck^Cre influenza n = 10, Tgfbr2-Lck^Cre LCMV n = 7, Ifnar ^−/− influenza n = 8, Ifnar ^−/− LCMV n = 8 mice per group). l, Gating strategy to identify IgG1 and IgG2 class switched GC B cells, and dark zone (DZ) and light zone (LZ) GC B cells. Data show experiments of 6–10 mice per group and mean ± SEM. Statistical tests: Two-tailed unpaired Student’s t test for intact system; two-tailed paired Student’s t test for bone marrow chimera model. **** P values are <0.0001 or otherwise indicated. Source Data

Extended Data Fig. 8. Human lymphoid tissue T_FH phenotypes identified by core and pathogen-specific T_FH signatures.

a, Gating strategy to sort CD3⁺CD4⁺CD45RA⁻CD45RO⁺CXCR5⁺CD27⁺ human tonsillar T_FH cells from three healthy adult donors for paired scRNA-seq and CITE-seq. b, Overlay of mean ranked scores of core T_FH and T_EFF core signatures onto UMAP clusters. Mean of the ‘TotalScore’ from singscore::simpleScore function collated for each cluster. c, Distribution of T_FH cells across 11 clusters by donor. d, Proportion of pathogen-specific T_FH signatures overlapping with human cluster genes expressed differentially (one cluster vs. all other clusters). e, Overlay of mean ranked scores of each T_EFF cell pathogen-specific signature onto UMAP clusters. Mean of the ‘TotalScore’ from singscore::simpleScore function collated for each cluster. T_EFF cell pathogen-specific upregulated signature genes used. f, Gene set Enrichment and vissE analyses for the comparison Cluster 1 vs. Cluster 2. Top gene sets clusters of interest shown as barplot of DE gene counts with corresponding gene statistics (FDR as color shade) from the DE analysis. GSEA was performed using Hallmarks c2 (‘CP:REACTOME’, ‘CP:PID’, ‘CP:BIOCARTA’,‘CP:KEGG’) and c5 (‘GO:BP’,‘GO:MF’) collections from the MsigDB using a two-tailed approach correcting for multiple testing with false discovery rate (FDR) adjusted p ≤ 0.05. g, Ranked score of pathogen-specific signatures onto human dataset T_FH cells from donors in SARS-CoV-2 vaccine (n = 5), influenza vaccine (n = 1), malaria infection (adult n = 3, child n = 3), peanut allergy (n = 3), asthma (n = 4) and autoimmune contexts (healthy donor n = 6, systemic lupus erythematosus n = 8). Data are presented as mean values ± SEMs. h, Visium spatial tonsil data displaying ranked score of core T_FH signature. i, Xenium spatial human lymph node data displaying ranked score of core T_FH signature. j, Visium spatial tonsil data displaying ranked score of pathogen-specific signatures onto germinal centers highlighted by core T_FH signature. Source Data

Extended Data Fig. 9. Novel markers identify distinct human T_FH subpopulations across tonsil and PBMC.

a–c, scCITEseq surface protein expression of sorted CD3⁺CD4⁺CD45RA⁻CD45RO⁺CXCR5⁺CD27⁺ human tonsillar T_FH cells from three healthy adult donors (as in Fig. 7). scCITEseq surface protein expression (log counts) of (a) T cell identity markers, (b) chemokine receptors, and (c) cluster-identifying markers, overlaid onto scRNA-seq UMAP. d-h, CD3⁺CD4⁺CD45RA⁻CD45RO⁺CXCR5⁺ human T_FH cells from five PBMC samples and six tonsil samples from healthy adult donors. (d), Gating strategy to identify T_FH cells in human tonsil, adenoid, and PBMCs with PD-1⁺ expression in tonsil and PBMCs displayed in dot plots. (e) T_FH surface marker frequency in tonsil CD57⁺ or CD57⁻ T_FH cells (n = 6 samples per group). (f) CD151/CD99 T_FH populations (Q1-Q4) in T_FH cells distinguished by CD57 and PD-1 from tonsil (n = 6 tonsils per group) and PBMCs (n = 5 PBMCs per group). (g) CXCR3, CCR6, and CCR4 expression in CD57/PD-1 and CD151/CD99 T_FH populations with key to identify T_FH populations distinguished by CD57 and PD-1, and T_FH populations (Q1-Q4) distinguished by CD151 and CD99. (h) gMFI expression of T_FH markers (PD-1, ICOS, OX40), chemokine receptor markers (CXCR3, CCR6, CCR4), and novel markers (CD127, CD99, CD71, CD151, CD43, CD69) with (+) and without (-) 4 hr PMA and ionomycin stimulation. Data show experiments of 5 PBMC samples per group and 6 tonsil samples per group and mean ± SEM. Statistical tests: one-way ANOVA of multiple comparisons. **** P values are <0.0001 or otherwise indicated. Source Data

Extended Data Fig. 10. Novel markers identify distinct human T_FH subpopulations in spike tetramer-specific PBMCs of SARS-CoV-2 infected and mRNA vaccinated donors.

a, CD3⁺CD4⁺CD45RA⁻CD45RO⁺CXCR5⁺ T_FH cells from six tonsil samples (n = 6) from healthy adult donors displaying T_FH surface marker frequencies in T_FH populations (Q1-Q4) distinguished by CD151 and CD99. b, Gating strategy to identify SARS-CoV-2 spike tetramer⁺ cT_FH cells. c,d Spike tetramer⁺ cells in (c) SARS-CoV-2 infection (11/11 donors; n = 11) or 6 months convalescence (10/11 donors; n = 10), or (d) nine COVID-19 mRNA vaccinated donors (n = 9) at 7 days post-vaccine and 3 months post-vaccine. Frequency Spike tetramer⁺ CD3⁺CD4⁺CD45RA⁻ T cells and CD3⁺CD4⁺CD45RA⁻CXCR5⁺ T_FH cells. Data show experiments of 8-11 samples per group. Statistical test: Two-tailed paired Student’s t test. **** P values are <0.0001 or otherwise indicated. Source Data