Selection of cross-reactive T cells by commensal and food-derived yeasts drives cytotoxic TH1 cell responses in Crohn's disease

Abstract

Aberrant CD4⁺ T cell reactivity against intestinal microorganisms is considered to drive mucosal inflammation in inflammatory bowel diseases. The disease-relevant microbial species and the corresponding microorganism-specific, pathogenic T cell phenotypes remain largely unknown. In the present study, we identified common gut commensal and food-derived yeasts, as direct activators of altered CD4⁺ T cell reactions in patients with Crohn's disease (CD). Yeast-responsive CD4⁺ T cells in CD display a cytotoxic T helper cell (T_H1 cell) phenotype and show selective expansion of T cell clones that are highly cross-reactive to several commensal, as well as food-derived, fungal species. This indicates cross-reactive T cell selection by repeated encounter with conserved fungal antigens in the context of chronic intestinal disease. Our results highlighted a role of yeasts as drivers of aberrant CD4⁺ T cell reactivity in patients with CD and suggest that both gut-resident fungal commensals and daily dietary intake of yeasts might contribute to chronic activation of inflammatory CD4⁺ T cell responses in patients with CD.

Fig. 1. Altered CD4⁺ T cell reactivity against fungal microorganisms in CD patients.

a, Frequencies of reactive CD154⁺CD45RA⁻CD4⁺ T_mem cells against whole microbial lysates in healthy donors and patients with an IBD: healthy donors (HD, black dots), CD (red dots) and UC (gray dots). The letter n indicates the number of individual donors specified below each group. The P values are given at the top of each figure. b, Dot plot examples for the ex vivo detection of microorganism-reactive CD4⁺ T cells by ARTE. Absolute cell counts after magnetic CD154⁺ enrichment from 1× 10⁷ PBMCs are indicated. w/o, without. c, ITS-seq and analysis of stool samples from patients with CD (n = 38). The relative abundance of the top 12 fungal species identified by this analysis is shown. Less abundant or less represented fungal species are grouped under ‘other’. d, Serum anti-IgG and IgA ASCA concentrations in healthy donors and patients with IBD (HD, n = 46; CD, n = 88; UC, n = 73). Sera were considered positive if their activity was >20 relative units (RU) per ml. Both single-positive IgA or IgG samples and double-positive samples classified donors who are ASCA⁺. e, Frequencies of fungus-reactive CD4⁺ T cells in patients with CD, according to their ASCA status. f, Total CD4⁺ T_mem cells isolated from biopsies of inflamed and uninflamed intestinal tissue of patients with CD who are ASCA⁺ (n = 5). Cells were seeded at 200 cells per well in multiple wells, and expanded and restimulated with fungal lysates in the presence of autologous APCs. Mean frequencies of reactive CD154⁺ cells for all wells per patient are shown. g, Dot plot examples for restimulation of expanded CD4⁺ T_mem cells from an inflamed biopsy. The percentage of CD154⁺ TNF⁺ cells within CD4⁺ is indicated. Each symbol in a, d, e and f represents one individual donor and horizontal lines in a and e indicate the mean. Box-and-whisker plots are shown in c with center lines representing the median, box limits 25% (lower) and 75% (upper) quartiles and whiskers the minimum/maximum. Truncated violin plots with quartiles and range are shown in f. Statistical differences were obtained using the Kruskal–Wallis test with Dunn’s post hoc test in a with only significant differences shown; the two-tailed Mann–Whitney U-test was used in e.

Fig. 2. Enhanced T_H1 cell but not T_H17 cell responses against yeasts in patients with CD.

a,b, Ex vivo cytokine production of fungus-reactive T_mem cells following ARTE. a, IL-17A and IFN-γ staining of yeast-reactive CD154⁺ T cells. b, Statistical summary in healthy donors and patients with IBD. c, IL-17A and IFN-γ production of yeast-reactive CD4⁺ T cells in patients with CD according to their ASCA status. d, Cytokine production of S. cerevisiae-reactive T cells in healthy donors and patients with IBD. Each symbol in b, c and d represents one individual donor and horizontal lines indicate the mean. Truncated violin plots with quartiles and range are shown in c and d. Statistical differences (shown in figures) were obtained using the two-tailed Mann–Whitney U-test in c and the Kruskal–Wallis test with Dunn’s post hoc test in d; only significant differences are shown.

Fig. 3. Yeast-reactive CD4⁺ cells from patients with CD have a cytotoxic phenotype.

a, Overview of the experimental design. PBMCs of each of three healthy donors and patients with CD who were ASCA⁺ were stimulated with C. albicans, C. tropicalis and S. cerevisiae lysates. Reactive CD154⁺ T_mem cells were FACS purified (see Methods for sorting) and analyzed by scRNA- and TCR-seq. b, UMAP visualization showing the subset composition of yeast-reactive CD4⁺ T cells colored by different gene expression clusters. c, Dot plot visualization showing the expression of selected marker genes in each T cell cluster. Colors represent the normalized mean expression levels and size indicates the proportion of cells expressing the respective genes (T_cm cell, central memory cell; T_TM cell, transitional memory cell; T_EM cell, effector memory cell; T_FH, cell follicular helper cell; T_FR cell, follicular regulatory cell). d, Proportion of cells within each T cell cluster for individual donors (healthy, n = 3; CD, n = 3). e, Trajectory plot of cells ordered along pseudotime (left) and overlaid with their cluster identity (right). f, Heatmap depicting gene expression of cluster-defining genes along the trajectory branches identified in e. Each symbol in d represents one individual donor, and truncated violin plots with quartiles and range are shown.

Fig. 4. Yeast-responsive T_H1-CTLs from patients with CD have potent killing ability for IECs.

a, Ex vivo staining of cytotoxic markers (GZMB, PRF1, CD319 and CCL4) and IFN-γ of S. cerevisiae-reactive CD154⁺ T_mem cells. b, Statistical summary (HD, n = 16; CD, n = 22; UC, n = 15). c, Expression of cytotoxic markers in S. cerevisiae-reactive T cells of patients with CD according to their ASCA status (ASCA⁻, n = 8; ASCA⁺, n = 14). d, Real-time monitoring of cytotoxicity by electrical impedance measurement. Fungus-reactive CD154⁺ T_mem cells were incubated with primary IECs at the indicated T cell-to-target ratios (T:IECs) and IEC killing was monitored via changes in the cell index. Gray line: IECs without added T cells; blue line: IECs after adding the detergent Triton X-100 (positive control for maximal lysis); black lines: T cells from an HD; red lines: T cells from a patient with CD who was ASCA⁺. e, Summary of T cell-mediated killing after 35 h of coculture for M. restricta- and S. cerevisiae-reactive T_mem cells at different T cell-to-IEC ratios (healthy, n = 3; ASCA⁺ CD, n = 3). f, Statistical summary of T cell-mediated killing at a T cell-to-IEC ratio of 1:4 (HD S. cerevisiae, n = 5; M. restricta, n = 6; CD S. cerevisiae, n = 6, M. restricta n = 5). Each symbol in b, c, e and f represents one donor. Truncated violin plots with quartiles and range are shown in b and c. Error bars indicate mean and s.e.m. in e and f. Statistical differences were obtained with the two-tailed Mann–Whitney U-test in c and the Kruskal–Wallis test with Dunn’s post hoc test in f. Statistical differences to S. cerevisiae-reactive cells from patients with CD are indicated.

Fig. 5. Clonally expanded T_H1-CTLs are cross-reactive to multiple yeast species.

a, TCR-β clonotype sizes were overlaid with UMAP visualization of the scRNA-seq dataset. b, Gini’s coefficient according to the different T cell clusters, depicting the distribution of TCR-β sequences (0 is total equality, that is, all clones have the same proportion; 1 is total inequality, that is, a population dominated by a single clone). To account for the uneven distribution of total sorted cells by samples, a down-sampled dataset was used taking 500 cells at random from each individual per fungal antigen (HD, n = 3; CD, n = 3). c, Number of T cells found in response to two or all three yeasts in healthy donors and patients with CD. The colors correspond to the different gene expression clusters from a. d, Clonal networks of antifungal TCR repertoires. Each dot represents a clonotype as defined by the TCR-β sequence. Connecting lines show TCR-β sequences that are shared between antifungal repertoires. The size of dots indicates clonal expansion. e, Percentage of identical TCR-β sequences found in samples stimulated with the different yeast species. f, Expression of selected marker genes according to TCR specificity. Colors represent the normalized mean expression levels and size indicates the proportion of cells expressing the respective genes. g, Cross-reactive TCRs from the single-cell dataset inserted into primary T cells using orthotopic TCR replacement. Dot plot examples show restimulation of expanded TCR-transgenic T cells for one TCR-α/β construct; numbers indicate percentage of reactive CD154⁺TNF⁺. h, Restimulation of TCR-transgenic T cells (n = 7) with different fungal species. i, Heatmap depicting the cross-reactivity of the individual TCR-transgenic T cells (n = 7) in relation to stimulation with S. cerevisiae. The initial observed cross-reactivity based on TCR-seq is indicated on the right. j, S. cerevisiae-reactive, cytotoxic, CD154⁺CD319⁺, single-cell clones (n = 10) sorted ex vivo, expanded and restimulated with different fungal species. The percentage of cross-reactivity is indicated in relation to stimulation with S. cerevisiae. Each symbol in b represents one individual donor, in h one TCR-transgenic T cell line and in j one T cell clone. Truncated violin plots with quartiles and range are shown in b and j.

Fig. 6. Food-derived yeasts can activate cross-reactive T_H1-CTLs in patients with CD.

a–c, PBMCs of healthy donors and patients with CD were stimulated with lysates of different cheese types and reactive CD4⁺ T cells analyzed by ARTE. a, Dot plot examples. Absolute cell counts after magnetic CD154⁺ enrichment from 1 × 10⁷ PBMCs are indicated. b, CD154⁺ T_mem cell frequencies and IFN-γ production in response to different cheese types (HD, n = 5; ASCA⁺ CD, n = 5). c, Ex vivo cytotoxic marker expression of reactive CD154⁺ T_mem cells in response to different cheese types (HD, n = 5; ASCA+ CD, n = 5). d, Reactive CD154⁺ T_mem cells from patients with CD who are ASCA⁺ were isolated after stimulation with gorgonzola lysate and expanded. Dot plots show reactivity on restimulation with the indicated antigens. e, Cross-reactivity of expanded gorgonzola-activated cells from patients with CD who are ASCA⁺ to different microbial and control antigens (Kluyveromyces lactis, Penicllium camemberti, P. roqueforti, Geotrichum candidum, n = 2; all other antigens, n = 6). f,g, TCR-transgenic T cells restimulated with lysates of gorgonzola or different fungal species commonly found in cheese. f, Dot plot examples with percentage of reactive CD154⁺TNF⁺ cells shown for one TCR-α/β construct. g, Cross-reactivity of the individual TCR-transgenic T cells (n = 7) in relation to stimulation with S. cerevisiae. Each symbol in b, c and e represents one individual donor and in g one TCR-transgenic T cell line. Truncated violin plots with quartiles and range are shown in b, c and e and mean values with s.e.m. in g. Statistical differences were obtained using the two-tailed Mann–Whitney U-test in b.

Extended Data Fig. 1. Antigen-reactive T cell enrichment (ARTE).

(a) Experimental set-up of ARTE in combination with multiplexing of differently stimulated samples. PBMCs were stimulated with different microbial lysates for 7 hours. Cells were labeled with a CD4-antibody-based fluorescent barcode using different anti-CD4 clones and mixed. Reactive CD154+ T cells were enriched by ARTE. Dot plot examples show C. albicans lysate-stimulated cells before enrichment and after magnetic CD154+ enrichment from 2x10e7 PBMCs. Percentage of CD154+ cells within CD4+ T cells and absolute cell counts are indicated. (b) Gating strategy following enrichment of CD154+ cells via ARTE.

Extended Data Fig. 2. Specificity of microbial lysate induced CD154 expression.

(a, b) Frequencies of microbial lysate stimulated CD154+ T cells stimulated in presence or absence of an anti-HLA-DR antibody of (a) healthy donors (n = 10) and (b) CD patients (n = 6). (c) Representative flow plots showing CD154 induction in presence or absence of an anti-HLA-DR antibody. Numbers of enriched CD154+ cells are indicated. (d) Following stimulation with microbial lysates, CD154+ cells were FACS-purified, expanded for several weeks and restimulated with the specific or unrelated microbial lysates in the presence of autologous antigen presenting cells. PMA-Ionomycin stimulation was used as positive control. Percentage of CD154+ TNFα+ cells within CD4+ is indicated (E. coli, F. prausnitzii, C. albicans, C. glabrata n = 5; K. pneumoniae n = 7; P. copri, B. fragilis, C. tropicalis n = 4; S. cerevisiae n = 8; M. restricta n = 2). (e) Total CD4+ memory T cells were isolated from biopsies of inflamed and uninflamed intestinal tissue of ASCA- CD patients (n = 5). Cells were seeded at 200 cells/well in multiple wells, expanded and re-stimulated with fungal lysates in presence of autologous APCs. Mean frequencies of reactive CD154+ cells for all wells per patient are shown. Each symbol in (a, b, d, e) represents one individual donor. Truncated violin plots with quartiles and range are shown in (e). Statistical differences: two-tailed paired t test in (a, b).

Extended Data Fig. 3. Correlation of T cell data with clinical parameters.

Reactive CD4+ T cell frequencies against C. albicans, C. tropicalis and S. cerevisiae were correlated to different clinical parameters in all CD patients (a-d) or ASCA+ CD patients (e-h). (a, e) Influence of age at diagnosis, disease localization or behavior, based on Montreal classification, on fungus-reactive T cell frequencies. n indicates the number of individual donors specified below each group. (b, f) Spearman correlation of yeast-reactive CD4+ T cell frequencies with disease duration or activity, as determined by Harvey-Bradshaw Index (HBI) or Crohn’s disease activity index (CDAI). (b, C. albicans n = 88, C. tropicalis n = 59, S. cerevisiae n = 44; f, C. albicans n = 50, C. tropicalis n = 38, S. cerevisiae n = 28). (c, g) Yeast-reactive T cell frequencies according to different treatments. IFX, Infliximab; VDZ, Vedolizumab; UST, Ustekinumab; ADA, Adalimumab; 5-ASA, 5-aminosalicylic acid. n indicates the number of individual donors specified below each group. (d, h) Yeast-reactive T cell frequencies according to sex. n indicates the number of individual donors specified below each group. Each symbol in a-h represents one individual donor. Horizontal lines indicate geometric mean values in (a, c, d, e, g, h).

Extended Data Fig. 4. Cytokine production of yeast-reactive CD4+ T cells.

(a) Proportion of IL-17A+IFN-γ+ cells reactive against the different fungal antigens. n indicates the number of individual donors specified below each group. (b) Proportion of IL-17A+IFN-γ+ cells in CD patients according to their ASCA status. n indicates the number of individual donors specified below each group. (c) Example plots for ex vivo cytokine production of S. cerevisiae-reactive Tmem following ARTE. Percentages of cytokine positive cells within CD154+ Tmem are indicated. (d) Cytokine production of C. tropicalis, C. albicans and M. restricta-reactive T cells in healthy donors and IBD patients. n indicates the number of individual donors specified below each group. Each symbol in (a, b, d) represents one individual donor, truncated violin plots with quartiles and range are shown in (b, d). Statistical differences: two-tailed Mann-Whitney test in (b); Kruskal-Wallis test with Dunn’s post hoc test in (d), only significant differences are shown.

Extended Data Fig. 5. Yeast-responsive Th1 cells express several cytotoxicity-associated markers.

(a) Dot plot visualization showing the expression of cytotoxicity-associated genes in each T cell cluster. Colors represent the normalized mean expression, and size indicates the proportion of cells expressing the respective genes. (b) Heatmap showing the expression of selected genes in each T cell cluster. Colors represent the Z-score-normalized expression amounts. (c) Dot plot visualization showing the expression of selected marker genes within the effector/Th17 cluster for healthy donors and CD patients according to antigen specificity. Colors represent the normalized mean expression levels, and size indicates the proportion of cells expressing the respective genes. (d) IL-17A producing cells were labeled with hashtag antibodies prior to scRNA sequencing (see Methods) and are plotted as overlay with the effector/Th17 cluster. These data show that only a subset of cells within this cluster has a Th17 signature. (e) Proportion of IFN-γ+ or IL-17A+ cells labeled with hashtag antibodies for individual donors and specificities (healthy n = 3, CD n = 3). (f, g) Expression of cytotoxic markers in (f) C. tropicalis and (g) C. albicans-reactive T cells of CD patients according to their ASCA status. (ASCA- n = 8; ASCA+ n = 14). Each symbol in (f, g) represents one individual donor. Truncated violin plots with quartiles and range are shown in (f, g). Statistical differences: Two-tailed Mann-Whitney test in (f, g).

Extended Data Fig. 6. Alterations of yeast-reactive CD4+ T cells are present in first-degree-relatives of IBD patients (IBD-FDR).

(a) Frequencies and (b) IL-17A production of yeast-reactive T cells in IBD-FDRs (n = 25) and non-FDR controls (n = 22). (c) IFN-γ staining of yeast-reactive T cells in IBD-FDRs (n = 25) and non-FDR healthy controls (n = 38). (d) Dot plot examples for IFN-γ staining of S. cerevisiae-reactive T cells. (e) Cytotoxic marker expression of C. tropicalis and S. cerevisiae-reactive CD154+ Tmem (non-FDR n = 16, IBD-FDR n = 14). (f) Heatmap depicting ex vivo cytotoxic markers production of yeast-reactive T cells within CD154+ Tmem measured by flow cytometry. (g) Serum anti-IgG and IgA ASCA antibody concentrations in non-FDRs (n = 46) and IBD-FDRs (n = 25). (h) Cytotoxic marker expression in IBD-FDRs in relation to elevated ASCA levels of the individual IBD-FDRs (ASCA < 10RU/ml n = 8, ASCA > 10 RU/ml n = 5). Individual donors are shown as dots. Each symbol in (a, b, c, e, g, h) represents one individual donor. Horizontal lines indicate mean values in (a). Truncated violin plots with quartiles and range are shown in (b, c, e). Box-and-whisker plots are shown in (h) with center lines represent the median, box limits represent 25% (lower) and 75% (upper) quartiles and whiskers represent minimum/maximum. Statistical differences: two-tailed Mann-Whitney test in (a, b, c, e, h).

Extended Data Fig. 7. Cross-reactivity of yeast-responsive T cells.

(a) α/β-TCR constructs of the most expanded cross-reactive TCRs identified from the single cell data set were inserted into primary human CD4+ T cells by orthotopic T cell receptor replacement using CRISPR/Cas. Dot plot examples showing CD4+ T cells with pulse only, knock-out of the endogenous TCR, combined knock-out of the endogenous TCR and knock-in of the transgenic TCR containing a murine β-chain constant region used as tracking marker. Transduced cells were further FACS purified and expanded to high purity. (b) TCR-α/β sequences selected from the scRNAseq data set for transgenic T cell generation. (c) S. cerevisiae-reactive Tmem from an ASCA+ CD patient were expanded and re-stimulated with various yeast species. Percentages of reactive CD154+ TNFα+ cells are indicated. (d) Cross-reactivity of expanded yeast-reactive T cells in ASCA- CD patients (C. albicans n = 4, S. cerevisiae, C. tropicalis n = 3) and ASCA+ CD patients (C. albicans, S. cerevisiae n = 4, C. tropicalis n = 3). Cross-reactivity is plotted as percentage in relation to total reactivity after stimulation with the initially used yeast species. (e) CMV serum antibody status of ASCA- CD patients (n = 39) and ASCA+ CD patients (n = 49) was determined by ELISA and percentage of CMV negative and CMV positive patients is indicated. (f) Yeast-reactive T cell frequencies of ASCA+ patients according to their CMV status (C. albicans CMVneg n = 32, CMVpos n = 17; C. tropicalis CMVneg n = 24, CMVpos n = 14; S. cerevisiae CMVneg n = 17, CMVpos n = 11). (g) TCR transgenic T cells (n = 7) were restimulated with common viral antigens. Cross-reactivity in relation to stimulation with S. cerevisiae is shown. Each symbol in (d, f) represents one individual donor and in (g) one TCR transgenic T cell line. Truncated violin plots with quartiles and range are shown in (d). Horizontal lines indicate geometric mean values in (f). Statistical differences: two-tailed Mann-Whitney test in (f).