Study participants were recruited with approval of the Institutional Review Boards (Ethics committee of the Albert-Ludwigs-University, Freiburg, #407/16 & #14/17, University of Pennsylvania Institutional Review Board #814428). The study was performed in agreement with the principles expressed in the Declaration of Helsinki (2013). Written informed consent was received from participants prior to inclusion in the study.
Patient populations and specimen collection
CD patients were recruited at the IBD outpatient unit and endoscopy unit of the University Hospital of Freiburg. Additional CD samples were obtained from the biobank of the Immunology in IBD Initiative (I3) at the Hospital of the University of Pennsylvania. Following patients’ informed consent, peripheral blood, intestinal biopsies and clinical parameters including laboratory data were collected. 61 patients with Crohn’s disease and 25 healthy donors were included in the study (Table 1 and Supplementary Table 2–4). For analysis of peripheral blood samples patients were stratified into active and inactive disease based on stool calprotectin concentrations and Harvey Bradshaw Index (HBI)49. Intestinal biopsies were stratified into normal and inflamed mucosa based on macroscopic appearance during endoscopy. Due to limited availability of patient sample material and cell numbers per sample not all experiments were performed with all samples, the number of samples is indicated in the respective legend. Each experiment included samples from patients with active disease and in remission. Patient samples were chosen according to these clinical categories as measured by Calprotectin / HBI and based on endoscopic assessment of inflammation and not further selected based on additional parameters.
Isolation of mononuclear cells from the peripheral blood and the intestinal tissue was performed as previously described50,51. Briefly, isolation of peripheral blood mononuclear cells (PBMC) was done using a ficoll-histopaque density gradient centrifugation (Ficoll Paque PLUS, GE Healthcare Life Sciences). Blood drawn into EDTA tubes was diluted 1:1 with PBS, subsequently overlaid 2:1 on ficoll, and centrifuged for 20 min (800 g, acceleration/ brake 0, room temperature (RT)). The collected buffy coat was washed in phosphate buffered saline (PBS, Corning, #21-040-CV) and pelleted prior to further use (10 min, 500 g, acceleration/brake 9, RT).
Lymphocytes from intestinal biopsies isolated at the University Medical Center Freiburg used for flow cytometric analyses were mechanically homogenized through a 70 µm cell strainer (ThermoFisher, #08-771-2) before washing and biobanking (not subjected to an enzymatic digestion). Samples were obtained during ileocolonoscopy and origin grouped into ileum/colon/sigma. Analysis of biopsies that underwent flow cytometry revealed a mean of 65,449 recorded cells per sample with a minimum of 12,643 and a maximum of 198,257 cells per panel per analysis. For analysis of peripheral blood we plated 1 million PBMCs/well for staining. This amount of cells can in our experience not be reached with endoscopically obtained biopsies. 34.6% of biopsies were from the colon, 34.6% were taken from the ileum and 30.7% from the sigma. Samples were processed and frozen on the day of sample collection and thawed on the day of the respective experiment. For flow cytometry studies this approach was chosen to maximize cell yield and staining performance of chemokine receptors.
For mass cytometry experiments, lamina propria lymphocytes (LPL) and intraepithelial lymphocytes (IEL) were isolated using enzymatic digestion per I3 biobank SOPs. Freeze media (90% FBS (Gem Cell, 100–500), 10% DMSO (Sigma, D2650-5X10ML), DNase I (Sigma, #D5025-150KU, prepared stock of 4 mg/ml) and Collagenase/Dispase (Roche, 500 mg, #11097113001, prepared stock of 50 mg/ml)) were brought to RT. Biopsies were placed into a 50 ml Falcon tube containing 3 ml epithelial strip buffer (1X PBS, 5 mM EDTA (ThermoFisher, #15575020), 1 mM DTT (ThermoFisher, #R0861), 5% FBS, 1% penicillin/streptomycin (Gibco, #15140-122)) and incubated for 10 min in a 37 °C water bath. After vortexing, supernatant was transferred into an Eppendorf tube and spun down (17,000 × g, 10 min, 4 °C, accuSpin Micro 17 centrifuge). Supernatant was removed and remaining tissue was further digested in 5 ml of wash buffer (RPMI-1640 (Corning Life Sciences, #10-040-CV), 2% FBS, 1% L-Glutamine (Lonza, #17-605E), 1% pen/strep) supplemented with 50 µl of DNase I and 25 µl of Collagenase/Dispase stocks for 20 min at 37 °C. Following vortexing, the sample was gently strained through a 70 µm strainer and LPL were washed off with 20 ml of wash buffer. Cells were spun down (5 min, 800 × g, 4 °C), cryopreserved as described above and thawed on the day of the respective experiment. CyTOF acquisition was performed using LPL samples.
All markers included in the CyTOF panel were previously tested for compatibility with the dissociation protocol and optimized. In order to exclude significant influence of different lymphocyte isolation protocols on the obtained results we validated our findings from the CyTOF experiments in a flow cytometry cohort with a different isolation protocol.
Cytokine production and multiparametric flow cytometry
For intracellular cytokine staining cells were stimulated with ionomycin (Iono, Sigma, Germany; final concentration 1 µg/ml) and phorbol 12-myristate-13-acetate (PMA; Sigma, Germany; final concentration 50 ng/ml) for 5 h in the presence of brefeldin A (GolgiPlug; BD Biosciences, Germany; 0.5 μL/ml) and monensin (GolgiStop; BD Biosciences, Germany; 0.325 μL/ml) for 4 h at 37 °C. To allow for intracellular staining, cells were treated with the FoxP3 Kit (Thermo Fisher, Germany). 2% paraformaldehyde was used to fix the cells after staining. Flowcytometric analysis was performed with a CytoFLEX (Beckman Coulter, Germany). The following antibodies were used for flow cytometry experiments: CD8 BV650 (RPA-T8, BioLegend), PD-1 BV786 (EH12.1, BD Biosciences), PD-1 BV421 (EH12.2H7, BioLegend), PD-1 PerCP-eFluor710 (eBioJ105, eBioscience), CD26 FITC (2A6, eBioscience), CD6 PE (BL-CD6, BioLegend), CD27 PE-Dazzle594 (M-T271, BioLegend), CD39 PerCP-eFluor710 (eBioA1, eBioscience), CD69 PE-Cy7 (FN50, eBioscience), CD161 APC (191B8, Miltenyi), IFN-γ APC-eFluor780 (4 S.B3, eBioscience), Fixable viability dye BV510 (eBioscience), Fixable viability dye APC-eFluor780 (eBioscience), IL-17 BV605 (BL168, BioLegend), IL-17F BV786 (O33-782, BD Biosciences), RORγt PE (# 600380, R&D Systems), CD56 BV650 (5.1H11, BioLegend), TCRαβ AlexaFluor488 (IP26, BioLegend), TCRαβ BV421 (IP26, BioLegend), TCRγδ PE (5A6.E9, Life Technologies), CD8 PE-Dazzle594 (RPA-T8, BioLegend), TCR Vα7.2 PE-Cy7 (3C10, BioLegend), TCR Vα7.2 FITC (3C10, BioLegend), TCR Vα24Jα18 PE-Cy7 (6B11, Invitrogen), CD3 AlexaFluor700 (SK7, BioLegend), IL-17 PE (eBio64DEC17, eBioscience), TNF PE-Cy7 (MAb11, BioLegend), pSTAT3 Alexa 647 (4/P-STAT3, BD Biosciences), CD3 PerCP (SK7, BD Biosciences), CD8 Krome Orange (B9.11, Beckman Coulter), CD45RA PeCy7 (HI100, BD Biosciences), Isotype IgG2a κ Alexa 647 (MOPC-173, BD Biosciences), CD4 BV786 (L200, BD Biosciences), CD4 BV421 (RPA-T4, BioLegend).
Unstimulated PBMC samples were used to set the gates for flowcytometric analyses and each stimulation experiment included unstimulated PBMC samples. Due to limited cell numbers in biopsies, biopsies were not divided into a stimulated and an unstimulated sample.
Flow cytometric assessment of conventional and unconventional Tc17 cells
Percentages of γδ T cells, NKT cells and MAIT cells were assessed as fractions of live singlet CD3+ CD8+ CD4- T cells according to published gating strategies21,52. Specifically, unconventional T cell populations were identified by subgating for TCR γδ+ (γδ T cells), CD56+ TCR γδ- Vα7.2- (NKT cells), CD161 and Vα7.2+ (CD161hi Vα7.2+ and CD161mid Vα7.2+ MAIT populations) and conventional T cells (TCR γδ- CD56- Vα7.2-) of CD8+ T cells. For TCR γδ assessment, clone 5A6.E9 was used since it strongly stains both Vδ1 and Vδ2 subsets of γδ T cells. In selected experiments, this strategy was validated using a refined gating strategy assessing conventional T cell TCR αβ expression and expression of invariant NKT cell receptor on CD56+ cells using monoclonal 6B11 antibody34. Gating strategies are outlined in Supplementary Fig. 5.
Mass cytometry reagents were obtained from Fluidigm or generated by custom conjugation to isotope-loaded polymers using MAXPAR kit (Fluidigm). Mass cytometry antibodies used are shown in Supplementary Table 1. Prior to staining, cells were stimulated with PMA/Ionomycin for 5 h in the presence of golgi inhibitors as described above. Staining was performed as previously described53. Briefly, single-cell suspensions were pelleted, incubated with 20 μM Lanthanum-139 (Trace Sciences)-loaded maleimido-mono-amine-DOTA (Macrocyclics) in PBS for 10 min at RT for live/dead discrimination (LD). Cells were washed in staining buffer and resuspended in surface antibody cocktail, incubated for 30 min at RT, washed twice in staining buffer, pre-fixed with PFA 1.6%, washed, then fixed and permeabilized using FoxP3 staining buffer set (eBioscience), and stained intracellularly for 60 min at RT. Cells were further washed twice before fixation in 1.6% PFA (Electron Microscopy Sciences) solution containing 125 nM Iridium overnight at 4 °C. Prior to data acquisition on a CyTOF Helios (Fluidigm), cells were washed twice in PBS and once in dH2O. Mass cytometry data was acquired in one batch using bead-based normalization.
For analysis of mass cytometric data samples were first gated on Iridium intercalator positive, singlet LD negative CD45+ CD3+ CD8+ T cells using FlowJo (v10.6). Further analysis was performed using R (v3.6) (https://www.r-project.org). Analysis was performed using a modified version of the standard workflow of the CATALYST package (version 1.8.7) in Bioconductor54. Raw expression values were arcsinh transformed using a cofactor of 5. Clustering was performed using the cluster function from the CATALYST package applied to the following markers: IL-21, IFN-γ, TNF, IL-22, CCL3, IL-2, XCL1, GM-CSF, IL-13, IL-17A, IL-10. Briefly, a FlowSOM clustering was performed with a grid size of 10 × 10 for the self-organizing map followed by metaclustering with ConsensusClusterPlus with a predefined maximum number of 15 clusters. UMAP dimensional reduction was performed using the runDR function from the CATALYST package using the default settings and a random subset of 1000 cells. This package calls the runUMAP function from the scater package which uses the UMAP implementation from the uwot R package. The random seed for all analyses was set to 42. Hierarchical clustering was performed on unscaled data. R script and fcs files are available upon reasonable request.
Transcriptomic signature analysis
Transcriptional profiling data published by Lee et al.35 was extracted from E-MTAB-331 (https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-331/), normalized using the vsnrma function from the vsn package, subsetted for common expression features and corrected for batch effects using Combat package in R. Subsequently, hierarchical clustering (hclust) was performed for CD8+ T cells from CD patients using the following markers: CD27, CD6, CD69, DPP4, ENTPD1, KLRB1, PDCD1; patients were stratified into two groups (cutree). Corresponding clinical data was gathered from35. Survival analysis was performed using the survminer R package. Patients that reached remission at minimum one timepoint in the follow up were included in the survival analysis (n = 31). The R script is available upon reasonable request.
ROC analysis: Time dependent ROC curve for prediction of flare free survival at 200 days of follow up was performed using a GSVA score based on phenotypic Tc17 signature markers CD6, CD69, CD27, CD39, PD1, CD26, CD161 and CD8+ T cell transcriptome data and corresponding clinical data from 35 untreated CD patients that were extracted from E-MTAB-33135.
T cell expansion under IL-17 polarizing conditions
For cultivation under IL-17 polarizing conditions cells were plated at 2 million PBMC/well in a sterile flat-bottom 96 well plate in complete medium (RPMI 1640 with 10% fetal calf serum, 1% penicillin/streptomycin solution and 1.5% 1 M HEPES; all Thermo Fisher, Germany) and stimulated with 25 µl/ml T cell activator (Stemcell Technologies, Canada). Cytokines were added to achieve the following final concentrations: 12.5 ng/ml for IL-1β, 5 ng/ml for TGFβ, 25 ng/ml for IL-6 (all Stemcell Technologies, Canada) and 25 ng/ml for IL-23 (Miltenyi Biotec, Germany). Cells in the control condition were stimulated with T cell activator only. Cells were cultured for 7 days at 37 °C, counted on days 3, 5, 7 and analyzed by flow cytometry on day 3 and 7 after restimulation with PMA/ionomycin.
STAT3 phosphorylation experiments
2 million PBMC/well were stimulated with IL-21 (10 ng/ml) (Miltenyi Biotec, Germany) in IMDM for 15 min at 37 °C with or without prior addition of Itolizumab (40 µg/ml) overnight and/or 15 min prior to stimulation. Cells were then treated with the BD Phosflow kit (BD Biosciences, Germany) according to manufacturer’s instructions and STAT3 phosphorylation was assessed via flow cytometry.
Itolizumab (Alzumab™, clone CD6D1, Biocon, India) was added to lymphocyte cell cultures at a final concentration of 40 µg/ml and anti-CD6 antibody, clone UMCD6 (Sigma–Aldrich, Germany) was added to a final concentration of 10 µg/ml for the respective experiments. Isotype antibodies were used to control for the addition of blocking antibodies. The following isotype control antibodies were used: For Itolizumab: purified human IgG1, κ isotype control recombinant antibody (clone QA16A12, BioLegend, USA) at a final concentration of 40 µg/ml, for UMCD6: purified mouse IgG1, κ isotype control antibody (clone MG1-45, BioLegend, USA) at a final concentration of 10 µg/ml.
FlowJo software v10.6 (FlowJo LLC, USA) was used to analyze flow cytometric and mass cytometric data. Statistical analysis was performed using GraphPad version 8 (Prism Software Inc., USA) and R version 3.6 (https://www.r-project.org) respectively. Essential R packages used are: SummarizedExperiment, CATALYST, flowCore, ComplexHeatmap, limma, lme4, edgeR, FlowSOM, tsne, nlme, MASS, Rtsne. Statistical tests used: Kruskal-Wallis-Test with Dunn’s multiple comparison’s test (Figs. 1A, C, D, F, 4), Mann–Whitney test (Figs. 1F, 5C), Wilcoxon test (Figs. 3, 6D–F), Log rank test (Fig. 5A). Marginal means in a linear model were used to assess statistical significance in Fig. 6C. A p value < 0.05 was considered significant. If not specified otherwise, ****indicates a p value <0.0001, *** <0.001, ** <0.01, * <0.05. Flow cytometric and mass cytometric original data plots are depicted using FlowJo as 2% contour plot with outlier setting or dot plots.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.