Our cohort included n=75 BD patients with mucocutaneous or ocular clinical manifestations attending the Behçet’s Centre of Excellence at the Royal London NHS Trust. BD diagnosis met International Criteria 2014 for BD (27) (Table 1). Age and gender-matched adult healthy controls (HC, n=41) were recruited alongside, in good general health and had no symptoms of infection for at least 3 weeks prior to blood collection. The study was conducted in compliance with the Helsinki Declaration and under Ethical Approval P/03/122 granted by The Queen Mary Research Ethics Committee and City Research Ethics Committee. All patients were stratified according to their disease activity state (active/inactive) and clinical phenotypes (mucocutaneous/ocular), based on their BD current activity form (BDCAF) score and clinical evaluation at the time of their visit to the clinic and blood sample collection. All participating patients and healthy volunteers gave their written informed consent prior to taking part in the study.

Peripheral blood samples (25-30mL) were collected from BD patients and HC into EDTA-vacutainers (Becton Dickinson). Peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation over Ficoll-Paque™Plus (GE Healthcare) and suspended in foetal bovine serum (FBS, Thermo Fisher Scientific) containing 10% dimethyl sulfoxide (Sigma), 100 IU/mL penicillin, and 100 mg/mL streptomycin (Gibco) for cryopreservation and storage in liquid nitrogen until subsequent use. For serum samples, 5mL of peripheral blood was collected in ‘clot activator’ vacutainers. Tubes were centrifuged for 5 min at 1500xg before serum was collected and stored at -80 °C until used.

Phenotyping of B and T cell subpopulations was carried out using PBMCs directly ex vivo or after in vitro cell culture. For membrane markers, cells were stained with a Live/Dead discriminant dye (BioLegend), followed by blocking with Fc receptor binding inhibitor Human TruStain FcX™, (BioLegend). The cells were then incubated with different cocktails of fluorochrome-conjugated antibodies (Supplementary Tables 1-3), washed in MACS buffer, and fixed with 2% paraformaldehyde (PFA). For analysis of intracellular cytokines or intranuclear markers, after membrane staining, cells were fixed/permeabilised with Cyto-Fast™ Fix/Perm Buffer Set (BioLegend) or True-Nuclear™ Transcription factor buffer set (BioLegend) respectively and incubated with specific antibodies as listed in Supplementary Table 4-6. Data were acquired with FACSDiva or SpectroFlow^® software on LSRII or Aurora (Cytek) (Beckton Dickinson) cytometers respectively. Live single cells were gated on FSC-A versus SSC-A and SSC-A versus Live/Dead dye FACS plots. Fluorescence-minus-one control samples were used for adjustments of gates. Data were analysed using FlowJo v10.10 (TreeStar).

Frozen PBMC from BD patients and HC were thawed and treated with 100mg/mL DNase I (Merck) for 10 minutes, washed in RPMI medium, and adjusted to a concentration of 1x10⁶cell/mL in serum-free RPMI medium. For proliferation analyses, PBMC were stained with 3µM CMFDA cell tracker™ green dye. Then cells were then washed, resuspended in complete RPMI medium [containing 20% FBS (Thermo Fisher Scientific), 100IU/mL penicillin, 100mg/mL streptomycin (Gibco); 5mM L-glutamine (Gibco) and 0.1mM non-essential amino acids (Gibco)] and plated in 96-well round bottom plates at a density of 2x10⁵cells/well. For ex vivo intracellular cytokine evaluation: cells received 50ng/mL of phorbol 12-myristate 13-acetate (PMA) (Adipogen), and 250ng/mL of ionomycin (Iono, eBioscience), in addition to of 2µM BD Golgi Stop (Monensin) (BioLegend), and 1µL/million cells of BD Golgi Plug (Brefeldin) (BD Bioscience), and incubated for 4 hours. Vδ2T-cells were specifically stimulated within the total PBMCs (28–30), using 100ng/mL (E)-4-Hydroxy-3methyl-but-2-enyl pyrophosphate (HMB-PP) (Sigma Aldrich) and 20ng/mL IL-15 (Peprotech). Non-stimulated controls were included for each sample for comparative analysis. A positive control culture for plasma cell induction was set-up using PBMC stimulated with 5µg/mL anti-human IgG/A/M (Abcam), 2.5µg/mL CpG (Alpha Diagnostics Int), 1µg/mL soluble CD40L (Enzo Life Sciences), and 50ng/mL IL-21 (Peprotech) (31). For intracellular cytokine analysis, cells were harvested at 6, 12, and 24 hours. Cultures were maintained for 5 days at 37 °C in a humidified atmosphere with 5%CO2 and supplemented every day with 50µM β-mercaptoethanol (Sigma). Cells were collected on day 5 for staining with the γδT and B cell antibody panels (Supplementary Tables 3, 4). For analysis of autoantibody levels in supernatant, cultures were maintained for 10 days (32). On day 4, cultures (both stimulated and non-stimulated) were supplemented with human recombinant (hr) IL-2 (20U/mL), IL-6 (50ng/mL), IL-10 (50ng/mL), IL-15 (10 ng/mL) (Peprotech) and mouse anti–CD40L IgG (1µg/mL; clone TRAP1, Biolegend) to sustain B cell proliferation and differentiation. On day 7, cultures were supplemented with hr IL-6 (50ng/mL), IL-15 (10ng/mL) (Peprotech), hr hepatocyte growth factor (20ng/mL) (Cell Sciences, Canton, MA), and hyaluronic acid (100µg/mL, Sigma) to promote plasma cell differentiation and viability (32). After 10 days, supernatants were collected and frozen at -80 °C for subsequent analysis. Cell viability of cultured cells was assessed by flow cytometry with Zombie Ultra-Violet Live Dead viability dye.

Frozen serum samples from HC and BD patients were thawed, diluted 1:250 in the sample diluent buffer while cell culture supernatants were used undiluted. All samples were tested in duplicate to determine levels of anti-HSP60 IgG/A/M antibodies using commercial ELISA kits (Enzo Life Sciences) following manufacturer’s instructions. Plates were read using a Clariostar microplate reader (BMG Labtech) and absorbance recorded at 450nm. Antibody quantitation was performed by interpolation from a standard curve.

The supernatants of cultured cells (for 1 day and 5 days) culture supernatants were used to evaluate cytokines concentration using the Luminex R&D Systems Discovery Assay (Biotechne). This assay was a custom-made, premixed, multi-analyte ELISA incorporating reagents for quantitative detection of IL-4, IL-10, IL-17, IL-21, CXCL13, IFN-γ, and TNF-α. Assay samples were prepared and read in MagPix Luminex instrument following the manufacturer’s instructions. Cytokine concentrations were evaluated using Luminex software against standard curves generated with calibration reagents.

Data are presented as mean values with standard error of the mean (SEM). Normality of data sets was evaluated by Shapiro Wilk test. If normally distributed, two-group comparisons were performed using two-tailed unpaired t-test. Multi-group comparison was performed using one-way ANOVA. For skewed data sets, non-parametric Mann–Whitney and Kruskal–Wallis tests (followed by Dunns multiple-comparison test) were used. Grouped data sets were analysed using 2-way ANOVA. In multiple tests, Tukey’s, uncorrected Fisher and Dunn’s tests were used to correct P-values. All analyses were performed, and graphs created with Prism version 10.0.3 (GraphPad). P-values < 0.05 were considered statistically significant.

Human γδT-cell acquisition of a ‘follicular helper-like’ state has previously been suggested to promote B-cell differentiation (5), but it remains unclear whether this process could contribute to autoantibody generation in BD. To assess this possibility, we first used flow cytometry to profile γδT-cell and B-cell compartments in PBMC from BD patients and HC donors. The total proportion of CD19⁺B cells was significantly reduced in active BD patients (5.5 ± 0.73%) and inactive BD patients (6.05 ± 0.76%) compared to HC (10.11 ± 1.4%; P = 0.0061 and 0.01 respectively) (Figure 1A). This reduction was more evident in the subgroup of patients with mucocutaneous BD (Supplementary Figure 1A).

Next, we investigated γδT-cell profiles within the CD3⁺ pool. The total percentage of CD3⁺T-cells was significantly higher in active BD patients (74.65 ± 2.67%) and inactive BD patients (76.4 ± 2.16%) compared with HC (63.5 ± 1.96%; P = 0.0027 and <0.001 respectively) (Figure 1B). The frequencies of Vδ1 and Vδ2 T-cells were comparable between healthy and patient groups (Figure 1C). Additionally, the total percentage of CD3⁺ and Vδ2T-cells were significantly higher in the subgroup of patients with ocular BD (Supplementary Figures 1B, C).

Since data from mouse models has indicated that γδT-cells with a ‘Tfh-like’ profile can license immature B-cells to produce autoantibodies (5), we also analysed γδT-cell expression of key Tfh markers. Indeed, we detected a significantly increased percentage of CXCR5⁺PD-1⁺ double positive Vδ2T-cells in patients with active BD (0.42 ± 0.1%) compared with HC (0.13 ± 0.05%; P = 0.0045) (Figure 1D). Then we assessed mean fluorescence intensity (MFI) of Tfh markers. Both Vδ1 and Vδ2T cells from BD patients and HC donors showed comparable expression of CXCR5 and ICOS (Figure 1E). However, BD patients with active disease displayed significantly higher levels of CD40L expression on Vδ1T-cells (Active BD 204.6 ± 25.47 versus HC 100.6 ± 29.68; P = 0.0092) and higher levels of PD-1 expression on Vδ2T-cells (Active BD 148.49 ± 27.59 versus HC 50.33 ± 12.94; P = 0.0039) (Figure 1E). Altered expression of these markers in BD patients thus appeared to be more closely associated with the disease activity rather than clinical phenotype (Supplementary Figures 1D, E).

To further characterize the functional profile of γδT-cells in BD, we assessed ex vivo production of six cytokines associated with inflammatory responses and B cell-help (IL-4, IL-10, IL-17, IL-21, IFN-γ, and TNF-α). The levels of these cytokines were comparable between the BD and HC groups in both Vδ2⁺ (Figure 2) and Vδ1⁺T-cells (Supplementary Figure 1F). Overall, these data show that typical frequencies of B and γδT-cell frequencies are altered in BD, with increased expression of Tfh markers within the γδT-cell compartment in active disease.

To investigate the possible functional effects of Vδ2⁺ cells with a Tfh-like functional profile in BD, we next used microbial metabolite HMB-PP to selectively activate Vδ2⁺ T cells in total PBMC cultures and observed the impact on B-cell activation and differentiation over 5 days. As expected, total CD3⁺T cell numbers were not substantially altered after 5 days stimulation with HMB-PP (Figure 3A) whereas the Vδ2⁺ subset was selectively expanded in BD patients (Active BD; non-stimulated 2.53 ± 0.52% versus HMB-PP stimulated 11.4 ± 2.3%, P < 0.001) (Inactive BD; non-stimulated 3.32 ± 0.82% versus HMB-PP stimulated 14.64 ± 4%, P < 0.001), and HC donors (non-stimulated 2.6 ± 0.4% versus HMB-PP stimulated 11.2 ± 2.9%; P < 0.001) (Figures 3A, B). Additionally, CXCR5⁺PD-1⁺ Vδ2T-cell frequency fold change (calculated of non-stimulated and HMB-PP-stimulated cultures at each time point) increased over time in the BD group while decreasing in HC, with a significant difference observed by day 5 (BD: 5.9 ± 0.8% versus HC: 2.4 ± 1%, P = 0.04), (Figure 4A). This expansion was accompanied by dynamic changes in expression of selected Tfh markers, presented here as MFI fold change of increase between stimulated and non-stimulated conditions for each sample at each time point and presented as compared means. In the BD group, CD40L expression fold change decreased markedly by day 3 (P = 0.02) but recovered to levels comparable level with HC samples by day 5. In contrast, CXCR5 expression fold change peaked at day 3 in BD patients, which was significantly higher than in HC samples (P = 0.008). ICOS and PD-1 expression fold changes increased over time in both groups, reaching significantly higher values of ICOS in the BD group by day 5 (P = 0.04) (Figure 4B). These findings suggest altered dynamics of Tfh-like functional state acquisition by Vδ2T-cells from BD patients.

We next assessed the cytokine profiles of Vδ2T-cells after 6, 12, and 24 hours of HMB-PP stimulation compared with non-stimulated cells (Figure 5). Distinct expression patterns emerged over time. For instance, IL-4⁺ and IL-17⁺ cells tended to rise over time in both groups. In contrast, IFN-γ ⁺ and TNF-α ⁺ cells peaked at 6 hours in both groups and then decreased gradually, maintaining a significantly higher level of expression in BD patients (IFN-γ⁺ cells; BD: 10.1 ± 3% versus HC: 1.8 ± 0.3%, P = 0.03), (TNF-α⁺ cells; BD: 21 ± 1.9% versus HC: 14 ± 2.1%, P = 0.038). Interestingly, IL-10 and IL-21 also peaked at 6 hours but was maintained at significantly higher levels in the BD group by 24 hours (IL-10⁺ cells; BD: 3.4 ± 1.2% versus HC: 1.64 ± 0.38%, P = 0.03) (IL-21⁺ cells; BD: 2.54 ± 0.55% versus HC: 1.07 ± 0.3%, P = 0.04) (Figure 5).

We next investigated the ability of activated Vδ2T-cells to induce B cell proliferation and differentiation in vitro. First, we established a positive control culture system to confirm efficient B cell proliferation and differentiation under our culture conditions (Supplementary Figure 2). Next, we tested the effect of HMB-PP-activated Vδ2T-cells on B-cell proliferation and differentiation within total PBMC cell cultures. While the total proportions of CD19⁺B cells were similar in BD and HC groups following stimulation (Figure 6A), the frequency of proliferating B-cells detected was significantly higher in BD patients with active disease (4.6 ± 1.3% versus 32.1 ± 6.6%, P < 0.001), compared to both inactive BD (1.8 ± 0.3% versus 17.4 ± 2.5%, P = 0.06), and HC donors (7 ± 2% versus 21.5 ± 6.2%, P = 0.08), (Figure 6B).

Using the same culture system, we next investigated whether increased frequency of proliferating B-cells was linked with enhanced differentiation into plasma cells (CD19⁺CD27⁺CD38^hi). Following HMB-PP stimulation, the percentages of plasma cells were significantly higher in the Active BD group (10.85 ± 2.95%) compared to the inactive BD (4.2 ± 0.76%, P = 0.02) and HC groups (6.4 ± 1.45%, P < 0.001) (Figure 6C). Since expression of IRF-4 and PAX-5 in CD19⁺B-cells controls differentiation into plasma cells, we proceeded to test whether HMB-PP activation of Vδ2T cells could influence the expression of these transcription factors. Flow cytometry analyses showed that under HMB-PP-stimulated conditions, the frequency of IFR-4^hiPAX-5^lo cells was significantly higher in the Active BD group (25.13 ± 4.77%) compared to the inactive BD (15.14 ± 2.51%, P = 0.038) and HC (15.81 ± 3.18%, P = 0.19) (Figure 6D), indicating enrichment of a subset reported to display the highest potential for differentiation into antibody-producing plasma cells (31). Since changes in B-cell expression of IRF-4 and Pax-5 might be associated with the cytokine milieu generated by HMB-PP stimulation in total PBMC cell cultures, we next quantified supernatant levels of key cytokines and chemokines at early (Figure 7) and late (Supplementary Figure 3) timepoints of the cell cultures (on day 1 and 5 of culture). Using the Luminex assay to assess multiple cytokines and chemokines in cell culture supernatants (IL-4, IL-10, IL-17, IL-21, CXCL13, IFN-γ, TNF-α), we observed that all mediators apart from IL-10 were expressed at higher levels in the BD group, whereas IL-21 could not be detected in day 1 cultures (Figure 7). These early differences tended to resolve with increasing duration of culture, such that measured cytokine levels were comparable between BD and HC cultures by day 5 (Supplementary Figure 3; Supplementary Table 7).

Given that BD disease activity correlated with Vδ2 Tfh-like functional profile and B cell function across multiple assays, we next used clinical data to explore potential associations with serum levels of IgA, IgG, and IgM. Immunoglobulin concentrations in BD patients were similar to reported healthy adults mean levels and within normal range (Figure 8A). However, anti-HSP60 autoantibody levels were significantly higher in serum from patients with active BD (212.5 ± 36.4ng/mL) compared with inactive disease (57.65 ± 8.1 ng/mL; P = 0.0001) or HC donors (74.36 ± 10.9 ng/mL; P = 0.013) (Figure 8B). Independent from disease activity, supernatant levels of anti-HSP60 autoantibodies were significantly increased after HMB-PP stimulation of cell cultures of BD patients with ocular phenotype (3.25 ± 0.77 ng/mL) compared to mucocutaneous (0.3 ± 0.18 ng/mL; P < 0.0001) and HC donor cells (1.66 ± 0.33 ng/mL, P = 0.012) (Figure 8C). In the long-term cell cultures, cell viability remained high at day 10 when harvesting the supernatants for the functional autoantibody readout (Figure 8D).