Differential preferential binding of vedolizumab to T_Reg and T_Eff cells at different concentrations

To explore, whether non-linear exposure-efficacy correlations for vedolizumab might be due to α4β7-expressing immune cells not binding vedolizumab, we analysed the frequency of α4⁺β7⁺VDZ⁻ immune cells by flow cytometry using fluorescently labelled vedolizumab. We chose a concentration of 10 µg/mL, which was in the range of trough levels associated with optimal outcomes in a phase II trial.22 While the fraction of α4⁺β7⁺ cells was highest in CD4⁺ T cells and eosinophils, only a substantial portion of CD4⁺α4⁺β7⁺ T cells did not bind vedolizumab (online supplemental figure 1 and online supplemental table 12). Thus, we decided to further focus on subsets of CD4⁺ T cells.

To elucidate, whether vedolizumab binding to T_Reg and T_Eff cells differed at various concentrations, we used fluorescently labelled vedolizumab and performed flow cytometry analysis of PBMCs from patients with UC, CD and from healthy controls. We gated on CD4⁺CD127^lowCD25^highFoxp3⁺ T_Reg cells and CD4⁺CD127^highCD25^low T_Eff cells co-expressing integrin α4 and integrin β7 and quantified the fraction of these cells that bound fluorescently labelled vedolizumab (online supplemental figure 2). Here, we used concentrations of up to 50 µg/mL vedolizumab, a trough level that was associated with suboptimal outcomes in a phase II trial.22

The portion of α4⁺β7⁺ cells was significantly higher in T_Eff compared with T_Reg cells and comparable between UC, CD and healthy controls (online supplemental figure 3A and online supplemental table 13), while the expression of α4 and β7 per cell (as measured by mean fluorescence intensity (MFI)) was equal or higher on T_Reg compared with T_Eff cells (online supplemental figure 3B).

The portion of VDZ⁺ cells among α4⁺β7⁺ CD4⁺ T cells was similar between T_Reg and T_Eff cells from healthy donors and higher on T_Reg than on T_Eff cells from patients with UC and CD after exposure with 0.4 µg/mL vedolizumab. However, the fraction of VDZ⁺ T_Eff cells was significantly higher compared with T_Reg cells in healthy donors after exposure with 2 µg/mL vedolizumab and in all entities at 10 µg/mL vedolizumab with some individual differences. Following exposure with 50 µg/mL vedolizumab, virtually 100% of α4⁺β7⁺ T_Reg and T_Eff cells were positive for vedolizumab (figure 1A,B). In an additional series of experiments, we sought to confirm that these differences were not due to α4β7^low naïve T cells in either population. However, we could reproduce our findings, when additionally gating on CD45RO to exclusively select memory T cells (online supplemental figure 3C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Concentration-dependent binding profile of vedolizumab (VDZ) to T_Reg and T_Eff cells. Representative (A) and quantitative (B) flow cytometry of VDZ⁺ cells after gating on α4⁺β7⁺ T_Reg and T_Eff cells following incubation with the indicated concentrations of fluorescently labelled VDZ. Quantitative data are expressed relative to T_Eff cells. n=17–28 patients with IBD or healthy controls as indicated. (C) Representative (left) and quantitative (right) fluorescence microscopy of FACS-purified T_Reg and T_Eff cells stained with anti-β7 antibody (green) and different concentrations of fluorescently labelled VDZ (red) and counterstained with Hoechst (blue); scale bar 10 µm. Quantification of β7⁺ and β7⁺VDZ⁺ cells in eight high-power fields. n=5–6 (cells purified from leucocyte cones). Significant outliers were identified using Grubbs test and excluded from the analysis. Statistical comparisons were performed using two-way analysis of variance (ANOVA) with Sidak’s multiple comparison test (A, B) and Student’s t-test and mixed-effects analysis with Sidak’s multiple comparison test (C). Sample donor characteristics are listed in online supplemental table 2. CD, Crohn’s disease; FACS, fluorescence-activated cell sorting; T_Eff, effector T cell; T_Reg, regulatory T cell.

Moreover, microscopic analysis of FACS-isolated T_Reg and T_Eff cells stained with a non-competing anti-β7 antibody and incubated with different concentrations of vedolizumab confirmed that less T_Reg than T_Eff cells bound vedolizumab at a concentration of 10 µg/mL, which was not the case in cells exposed to 50 µg/mL vedolizumab (figure 1C, online supplemental figure 3D).

We further explored, whether the activation status of the cells and associated differences in α4β7 integrin conformation might be relevant for the differential binding pattern. However, we were also able to reproduce a right-shifted binding profile of T_Reg cells following stimulation with MnCl₂, phorbol-12-myristat-13-acetat/ionomycine and anti-CD3/CD28 (online supplemental figure 4, online supplemental table 14), suggesting that this is not the case.

In synopsis, our data showed that blocking α4β7 integrin with vedolizumab on T_Reg cells requires higher concentrations of the antibody compared with T_Eff cells. This implied that clinical efficacy of vedolizumab might at least partly result from residual T_Reg cell homing at concentrations already completely blocking T_Eff cell homing.

Differential vedolizumab binding to T_Reg and T_Eff cells leads to differential dose-dependent adhesion and transmigration profiles

To explore the functional relevance of our findings, we analysed FACS-purified CD4⁺CD127^lowCD25^high T_Reg and CD4⁺CD127^highCD25^low T_Eff cells in functional assays in vitro. Sorting achieved a purity of >99% for both cell types, cells were viable and T_Reg cells exhibited marked suppressive abilities, when co-cultured with T_Eff cells (online supplemental figure 5).

We analysed the impact of in vitro treatment with vedolizumab on the dynamic adhesion of T_Reg and T_Eff cells to MAdCAM-1 (figure 2A). We focused on 10 µg/mL and 50 µg/mL as the most clinically relevant concentrations.25 Consistent with α4β7 integrin expression, adhesion was significantly higher for untreated T_Eff compared with untreated T_Reg cells (figure 2B,C). Either treatment led to a reduction of the dynamic adhesion of both cell types. However, inhibition of adhesion of T_Reg cells was substantially lower compared with T_Eff cells after treatment with 10 µg/mL vedolizumab, while almost complete inhibition of adhesion and no difference between T_Reg and T_Eff cells could be observed after treatment with 50 µg/mL vedolizumab (figure 2D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Concentration-dependent adhesion and transmigration of T_Reg and T_Eff cells in functional assays in vitro. (A–D) Dynamic adhesion of T_Reg and T_Eff cells treated with different concentrations of vedolizumab to MAdCAM-1. (A) Schematic representation of the experimental setup; fluorescently labelled T_Reg and T_Eff cells were incubated with different concentrations of VDZ, perfused through MAdCAM-1-coated capillaries and adhering cells were quantified using confocal microscopy. (B) Representative microscopic images of adhered cells (overlay of counted high-power fields) and (C) quantification of the background-corrected number of T_Reg and T_Eff cells incubated with or without 10 or 50 µg/mL VDZ adhering to MAdCAM-1. (D) Relative inhibition of adhesion of T_Reg and T_Eff cells to MAdCAM-1 after treatment with 10 or 50 µg/mL VDZ. n=22 (cells purified from leucocyte cones). (E–G) Transmigration assays with CD4⁺ T cells. The fraction of T_Reg and T_Eff cells in the transmigrating cells was quantified by flow cytometry. Representative (E) and quantitative flow cytometry (F) of transmigrating CD4⁺ cells after treatment with different concentrations of VDZ. **p<0.01 compared with all treatment groups. (G) T_Reg/T_Eff ratio of transmigrated cells after treatment with 10 versus 50 µg/mL VDZ (G). n=8–17 patients with IBD or healthy controls as indicated. Statistical comparisons were performed using paired t-test (C, D) and mixed-effects analysis with Tukey’s multiple comparisons test and paired t-test (F, G). Sample donor characteristics are listed in online supplemental table 3. CD, Crohn’s disease; MAdCAM-1, mucosal addressin cell adhesion molecule-1; T_Eff, effector T cell; T_Reg, regulatory T cell; VDZ, vedolizumab.

In a second approach, we investigated the impact of different concentrations of vedolizumab on MAdCAM-1-dependent transmigration of T_Reg and T_Eff cells in vitro. T cells were left to transmigrate towards CCL25 over MAdCAM-1-coated transwell plates with 3 µm pores in the presence of different vedolizumab concentrations. Treatment with all concentrations of vedolizumab led to a significant and similar reduction of transmigration of cells from patients with UC and CD as well as healthy controls. However, the ratio of transmigrated T_Reg to T_Eff cells was substantially higher after treatment with 10 µg/mL vedolizumab compared with 50 µg/mL (figure 2E–G).

Taken together, these data supported the notion that differential binding of vedolizumab to T_Reg and T_Eff cells has functional implications for T cell adhesion and transmigration.

Differential vedolizumab binding to T_Reg and T_Eff cells leads to differential homing to the inflamed gut in vivo

We next aimed to address, whether we could detect similar effects in vivo. To this end, we made use of a previously described humanised mouse model of T cell homing to the inflamed gut (figure 3A), in which we had earlier shown a comparable or even higher reduction of T_Reg cell homing to the gut after treatment with high vedolizumab concentrations.10 Based on our above findings, we now investigated the effect of treatment with 10 µg/mL vedolizumab. Intravital confocal microscopy demonstrated active trafficking of the transferred cells (figure 3B). As expected, more untreated T_Eff cells homed to the gut compared with untreated T_Reg cells. Interestingly, treatment with 10 µg/mL vedolizumab led to substantially reduced homing of T_Eff cells, while T_Reg cell homing was not significantly affected as assessed by flow cytometry and lightsheet fluorescence microscopy (figure 3C,D). These observations further supported our concept of differential responses of T cell subsets to vedolizumab.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

In vivo homing of T_Reg and T_Eff cells in a humanised mouse model. (A) Schematic representation of in vivo homing assays. Fluorescently labelled T_Reg or T_Eff cells±10 µg/mL VDZ were injected into the ileocolic artery of anesthetised mice for subsequent quantification of homed cells by LPMC isolation and flow cytometry or lightsheet microscopy. (B) Visualisation of homed cells (green) using intravital confocal microscopy. Red: blood vessels stained with anti-CD31; blue: nuclear counter-stain with Hoechst. Arrows indicate adhering human cells. Scale bar 100 µm. (C) Representative (left) and quantitative (right) flow cytometry of FarRed⁺ human T_Reg and T_Eff cells accumulating in the lamina propria of Rag1^−/− mice after treatment with either isotype control or with 10 µg/mL VDZ. n=6 per group (cells purified from leucocyte cones). (D) Representative (left) and quantitative (right) lightsheet fluorescence microscopy. Arrows indicate adherent cells still inside the vasculature (yellow) or extravasated into the tissue (green). Quantification of homed cells in 15 representative 3D cubes from three individual experiments (relative to the number of injected cells). Scale bar 100 µm. Statistical comparisons were performed using one-sample t-test and Student’s t-test. LPMC, lamina propria mononuclear cells; T_Eff, effector T cell; T_Reg, regulatory T cell; VDZ, vedolizumab.

Differential vedolizumab binding to T_Reg and T_Eff cells correlates with the availability of α4β7 integrin in vivo

We reasoned that for in vivo action of vedolizumab in patients with IBD, the remaining availability of free α4β7 molecules at a certain exposure is crucial. Thus, to understand, how different vedolizumab concentrations affect available α4β7 integrin, we exposed PBMCs to ascending doses of vedolizumab in vitro and subsequently labelled free binding sites. While we observed no significant difference in the ratio of T_Reg and T_Eff cells with free vedolizumab binding sites at a concentration of 2 µg/mL, a significantly higher portion of T_Reg compared with T_Eff cells had free α4β7 molecules available on their surface after incubation with 10 and 50 µg/mL vedolizumab. At a concentration of 110 µg/mL vedolizumab (in the range of the highest serum levels observed in patients25), the abundance of cells with free binding sites was similar again (figure 4A,B, online supplemental figure 6A,B), further supporting the concept of a right-shifted T_Reg cell response to vedolizumab.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Quantification of T_Reg and T_Eff cells with free vedolizumab binding sites in vitro and in vivo. (A) Representative and quantitative flow cytometry of free VDZ binding sites on α4⁺β7⁺ T_Reg and T_Eff cells after incubation with different concentrations of unlabelled VDZ in vitro and consecutive staining with saturating concentrations of fluorescently labelled VDZ. Quantitative data are expressed relative to T_Eff cells. n=5–6 patients with IBD and healthy controls. (B) T_Reg/T_Eff ratio of cells with free vedolizumab binding sites (free VDZ BS) after treatment with different concentrations of vedolizumab. n=5–6 patients with IBD and healthy controls. (C) T_Reg/T_Eff ratio of cells with free vedolizumab binding sites in patients undergoing vedolizumab therapy stratified according to VDZ trough levels. Staining was performed at week 2 and/or 6 of treatment, trough levels were determined using vedolizumab drug level ELISA. n=8–58 samples from patients with IBD per serum group, some patients provided blood at week 2 and week 6. Statistical comparisons were performed using one-sample t-test (A) and Student’s t-test (B, C). Sample donor characteristics are listed in online supplemental tables 4 and 5. T_Eff, effector T cell; T_Reg, regulatory T cell.

To explore, whether this holds also true in vivo, we determined the serum trough levels in patients with IBD receiving vedolizumab therapy at week 2 and 6 and simultaneously determined free binding sites. In an exploratory analysis, we observed an optimum in the ratio of T_Reg and T_Eff cells with free α4β7 molecules available in the range of 40 to 55 µg/mL vedolizumab trough level and a significantly reduced ratio at even higher serum levels (figure 4C, online supplemental figure 6C).

In conclusion, these data suggested that certain vedolizumab exposure levels go along with higher residual availability of functional α4β7 integrin on T_Reg compared with T_Eff cells in patients with IBD in vivo.

Single-cell RNA sequencing identifies an ITGB1⁺PI16⁺ T_Reg cell subset ‘resistant’ to vedolizumab

To further dissect the mechanisms underlying our observations, we decided to use single-cell RNA sequencing. To this end, we FACS-purified CD4⁺CD45RO⁺α4⁺β7⁺ cells binding fluorescently labelled vedolizumab (VDZ⁺) or not (VDZ⁻) at a concentration of 10 µg/mL. Re-analysis of sorted cells confirmed that all selected cells expressed the integrins α4 and β7 (online supplemental figure 7A). Moreover, we observed that the vast majority of α4⁺β7⁺VDZ⁻ cells also stained positive for fluorescently labelled MAdCAM-1 (online supplemental figure 7B) and observed dynamic adhesion of CD4⁺CD45RO⁺α4⁺β7⁺VDZ⁻ cells to MAdCAM-1, corroborating that α4β7 integrin expressed on cells not binding vedolizumab is functional (online supplemental figure 7C).

Following single-cell sequencing, VDZ⁻ and VDZ⁺ samples were merged for comparative analysis. Clustering analysis using unique molecular identifiers at a resolution of 1 identified 11 distinct clusters (figure 5A). Using eight different marker genes (online supplemental figure 8A,B), clusters 9 and 10 were identified as T_Reg cell clusters (figure 5B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Single-cell RNA-sequencing of CD4⁺CD45RO⁺α4⁺β7⁺VDZ⁺ and CD4⁺CD45RO⁺α4⁺β7⁺VDZ⁻ cells. (A) UMAP plot showing clustering of 14 265 cells based on Leiden algorithm at resolution 1. (B) UMAP plots showing the distribution of cells from the VDZ⁺ and VDZ⁻ sample in all cells (left panel) and in the T_Reg cell clusters 9 and 10 (right panel). (C) Violin plots displaying the differential gene expression of selected genes in the T_Reg and T_Eff cell clusters from the VDZ⁺ and VDZ⁻ sample. (D) UMAP plots of the T_Reg cell subclusters showing cells expressing PI16, CCR10, ITGB1, CCR9 and CD38. T_Eff, effector T cell; T_Reg, regulatory T cell; UMAP, uniform manifold approximation and projection; VDZ, vedolizumab.

Our further analyses showed that—consistent with our previous data—the fraction of T_Reg cells was higher in the VDZ⁻ compared with the VDZ⁺ sample (figure 5B). Interestingly, the VDZ⁻ T_Reg cells also expressed T_Reg marker genes to a higher extent than VDZ⁺ T_Reg cells (online supplemental figure 8C). When comparing VDZ^- and VDZ⁺ T_Reg and T_Eff cells, we identified a specific signature of differentially expressed genes, many of which were associated with adhesion, extravasation and chemotaxis (figure 5C). As we aimed to characterise α4β7-expressing T_Reg cells not binding vedolizumab, we further focused on VDZ⁻ in comparison with VDZ⁺ T_Reg cells. Taking into account differential gene expression and the fraction of cells expressing the relevant genes, we identified a distinct T_Reg cell subpopulation expressing ITGB1, PI16 and CCR10, but not expressing CCR9 and CD38 that was predominant in the VDZ⁻ sample and almost completely absent in the VDZ⁺ sample (figure 5D, online supplemental table 16).

β1⁺PI16⁺ T_Reg cells show reduced vedolizumab binding in vitro and in vivo

To validate our findings, we stained PBMCs from healthy controls with antibodies against the different molecules identified above. We confirmed that a significantly higher portion of T_Reg cells not binding VDZ at 10 µg/mL expressed PI16 and β1 integrin compared with vedolizumab-binding T_Reg cells (figure 6A). Vice versa, the abundance of VDZ⁺ cells was lower in α4β7-expressing T_Reg cells positive for PI16 or β1 (figure 6B). Consistently, co-expression of β1 integrin and PI16 was observed in substantially more VDZ^- compared with VDZ⁺ cells and vedolizumab binding to β1⁺PI16⁺ cells was clearly lower than to β1⁻PI16⁻ cells (figure 6C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Flow cytometric validation of differentially expressed marker genes between VDZ⁺ and VDZ⁻ T_Reg cells in vitro and in vivo. (A) Representative (left) and quantitative (right) flow cytometry showing the fraction of α4⁺β7⁺ T_Reg cells binding (VDZ⁺) or not binding (VDZ⁻) vedolizumab at a concentration of 10 µg/mL and expressing PI16 or integrin β1. n=14 healthy donors. (B) Representative (left) and quantitative (right) flow cytometry showing vedolizumab binding to α4⁺β7⁺ T_Reg cells expressing PI16 or integrin β1. n=16 healthy donors.(C) Representative (left) and quantitative (right) flow cytometry showing the fraction of α4⁺β7⁺ T_Reg cells binding (VDZ⁺) or not binding (VDZ⁻) vedolizumab at a concentration of 10 µg/mL and co-expressing PI16 and integrin β1 or showing vedolizumab binding to α4⁺β7⁺ T_Reg cells co-expressing PI16 and integrin β1. n=16 healthy donors. (D) Representative (left) and quantitative (right) flow cytometry showing the fraction of α4⁺β7⁺ T_Reg cells with occupied or with free vedolizumab binding sites (VDZ BS) expressing integrin β1 and PI16 in patients treated with vedolizumab. n=57 samples from patients with IBD, some patients provided blood at week 2 and week 6. (E) Representative (left) and quantitative (right) flow cytometry showing free vedolizumab binding sites on β1⁺PI16⁺α4⁺β7⁺ T_Reg cells. n=57 samples from patients with IBD, some patients provided blood at week 2 and week 6. Statistical comparisons were performed using paired t-test. Sample donor characteristics are listed in online supplemental tables 6 and 7. T_Eff, effector T cell; T_Reg, regulatory T cell.

Next, we obtained PBMCs from patients with IBD receiving clinical treatment with vedolizumab and assessed free vedolizumab binding sites (BS) on T_Reg cells together with the expression of the above markers. We observed that among T_Reg cells expressing α4β7 integrin with free binding sites for vedolizumab, cells expressing β1 and PI16 were significantly more abundant than among T_Reg cells already saturated with vedolizumab (figure 6D). Furthermore, among β1⁺PI16⁺ T_Reg cells, substantially more cells had free vedolizumab binding sites available than among β1⁻PI16⁻ T_Reg cells (figure 6E).

Together, these data corroborated our in silico findings and suggested that a β1⁺PI16⁺ T_Reg cell subset is the substrate of differential vedolizumab binding to T_Reg and T_Eff cells.

Vedolizumab-‘resistant’ β1⁺PI16⁺ T_Reg cells show a pronounced regulatory phenotype

Next, we aimed to further characterise the function of this T_Reg cell subset. Transcript levels in our single-cell dataset suggested that T_Reg cells not binding vedolizumab express a high level of regulatory markers and might therefore be a particularly suppressive cell population (online supplemental figure 8C).

Thus, we performed flow cytometry of CD4⁺CD25^highCD127^lowα4⁺β7⁺ T_Reg cells co-expressing integrin β1 and PI16 or not. We observed higher expression of CD25 per cell on β1⁺PI16⁺ than on β1⁻PI16⁻ gut-homing T_Reg cells (figure 7A). In addition, more β1⁺PI16⁺ T_Reg cells expressed Foxp3 and GITR and also to a higher extent (figure 7B,C). Functionally, after in vitro stimulation, a massively higher portion of β1⁺PI16⁺ T_Reg cells than β1⁻PI16⁻ T_Reg cells produced the suppressive cytokine IL-10 (figure 7D). These observations could also be reproduced using PBMCs from patients with IBD (online supplemental figure 9, online supplemental table 15). Finally, using in vitro co-culture suppression assays, VDZ^-, but not VDZ⁺ T_Reg cells clearly inhibited T_Eff cell proliferation (figure 7E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

Characterisation of vedolizumab-‘resistant’ β1⁺PI16⁺α4⁺β7⁺ T_Reg cells in the peripheral blood. (A) Quantitative flow cytometry showing mean fluorescence intensity (MFI) of CD25 on β1⁺PI16⁺α4⁺β7⁺ T_Reg compared with β1⁻PI16⁻α4⁺β7⁺ T_Reg cells. n=8 healthy donors. (B) Representative (left) and quantitative (right) flow cytometry showing the frequency of Foxp3-expressing β1⁺PI16⁺α4⁺β7⁺ T_Reg compared with β1⁻PI16⁻α4⁺β7⁺ T_Reg cells. n=8 healthy donors. (C) Representative (left) and quantitative (right) flow cytometry showing the frequency and mean fluorescence intensity (MFI) of GITR on β1⁺PI16⁺α4⁺β7⁺ T_Reg compared with β1⁻PI16⁻α4⁺β7⁺ T_Reg cells. n=6 healthy donors. (D) Representative (left) and quantitative (right) flow cytometry showing the frequency and the mean fluorescence intensity (MFI) of IL-10 on β1⁺PI16⁺α4⁺β7⁺ T_Reg compared with β1⁻PI16⁻α4⁺β7⁺ T_Reg cells after incubation with PMA, ionomycine and brefeldin A for 4 hours. n=8 healthy donors. (E) Representative flow cytometry of T_Eff cell proliferation as determined by dilution of CellTrace FarRed. Representative images from one out of five independent experiment (cells purified from leucocyte cones). Statistical significance was calculated using paired t-test. Sample donor characteristics are listed in online supplemental table 8. PMA, phorbol-12-myristat-13-acetat; T_Eff, effector T cell; T_Reg, regulatory T cell; VDZ, vedolizumab.

In a next step, we aimed to elucidate, whether the subset identified was also present in the gut of patients with IBD. Therefore, we performed in silico analyses with a publically available single-cell RNA sequencing dataset of CD45⁺ cells from the rectum of 11 patients with UC (GSE162335). We identified T_Reg cells and compared the expression of several key regulatory genes between β1⁺PI16⁺ and other T_Reg cells. In line with our single-cell data from the peripheral blood, we observed that many of these genes were expressed by a larger fraction of β1⁺PI16⁺ cells or at higher levels in these cells (figure 8A,B). To confirm these transcriptomic data, we isolated lamina propria mononuclear cells (LPMCs) from gut biopsies of patients with IBD and analysed CD4⁺CD25^highCD127^lowFoxp3⁺β7⁺ T_Reg cells co-expressing integrin β1 and PI16 or not using flow cytometry (online supplemental figure 10). β1⁺PI16⁺ T_Reg cells demonstrated a clearly higher expression of CD25 per cell (as determined by MFI) compared with β1⁻PI16⁻ T_Reg cells (figure 8C). Moreover, in vitro stimulation of LPMCs led to a significantly higher portion of β1⁺PI16⁺ than β1⁻PI16⁻ T_Reg cells producing IL-10 (figure 8D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8

Characterisation of β1⁺PI16⁺β7⁺ T_Reg cells in the lamina propria of patients with IBD. (A) UMAP plot showing clustering of T cells from a publicly available single-cell RNA sequencing dataset (GSE162335) of CD45⁺ LPMCs from the rectum of 11 patients with UC (left panel) and UMAP plots showing the distribution of T_Reg cells (right panel). (B) Heat map showing differential gene expression and the portion of cells expressing five prominent regulatory genes in β1⁺PI16⁺ compared to all other T_Reg cells from the dataset shown in (A). (C) Representative (left) and quantitative (right) flow cytometry showing mean fluorescence intensity (MFI) of CD25 on CD4⁺CD25^highCD127^lowFoxp3⁺β7⁺β1⁺PI16⁺ T_Reg compared with CD4⁺CD25^highCD127^lowFoxp3⁺β7⁺β1⁻PI16⁻ T_Reg cells. n=8 patients with IBD. (D) Representative (left) and quantitative (right) flow cytometry showing the frequency of IL-10 on CD4⁺Foxp3⁺β7⁺β1⁺PI16⁺ T_Reg compared with CD4⁺Foxp3⁺β7⁺β1⁻PI16⁻ T_Reg cells after incubation with PMA, ionomycine and brefeldin A for 4 hour. n=8 patients with IBD. Statistical significance was calculated using Student’s t-test. Sample donor characteristics are listed in online supplemental table 9. LPMC, lamina propria mononuclear cells; PMA, phorbol-12-myristat-13-acetat; T_Reg, regulatory T cell; UMAP, uniform manifold approximation and projection.

Collectively, these data strongly supported the notion that vedolizumab-‘resistant’ β1⁺PI16⁺ gut-homing T_Reg cells have a powerful regulatory function in the peripheral blood as well as in the intestine and might counteract inflammation in the gut.

β1⁺PI16⁺ T_Reg cells are ‘resistant’ to vedolizumab in vivo and enrich in the gut of patients with IBD responding to vedolizumab therapy

In a next step, we aimed to study, whether vedolizumab ‘resistance’ of α4β7-expressing β1⁺PI16⁺ T_Reg cells can also be observed in vivo. To this end, we quantified serum trough levels in patients with IBD receiving vedolizumab therapy and determined the availability of free vedolizumab binding sites on these cells. As expected, the portion of T_Eff cells with untargeted α4β7 integrin on their surface decreased in a dose-dependent fashion. However, this was not the case for α4β7-expressing β1⁺PI16⁺ T_Reg cells, while α4β7-expressing β1⁻PI16⁻ T_Reg cells exhibited a dose-dependent decrease similar to T_Eff cells (figure 9A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 9

Resistance of β1⁺PI16⁺α4⁺β7⁺ T_Reg cells to vedolizumab in patients with IBD in vivo and correlation with Gemini II and III trials. (A) Correlation of T_Eff cells (left), β1⁺PI16⁺α4⁺β7⁺ T_Reg cells (middle) and β1⁻PI16⁻α4⁺β7⁺ T_Reg cells with free vedolizumab binding sites (VDZ BS) as determined by flow cytometry with serum trough levels of vedolizumab as determined by ELISA in a cohort of patients with IBD treated with vedolizumab. Line showing simple linear regression. n=30 patients with IBD. (B, C) Representative (left) and quantitative (right) immunohistochemistry of human colon biopsies obtained from patients before or under treatment with VDZ. (B) CD4 (green), Foxp3 (red) and nuclei counterstain with Hoechst (blue), (C) Integrin β1 (green), Foxp3 (red) and nuclei counterstain with Hoechst (blue). Scale bar 100 µm (B), 50 µm (C). Quantification of eight high-power fields (HPF) per sample. n=6 (B), n=12–16 (C) patients with IBD. (D) Percentage of patients with CD from Gemini II and/or III trials achieving clinical remission at week 6 stratified according to VDZ though levels at week 6. 28–463 patients with CD per group. Boxes indicate remission rates with 95% Clopper-Pearson CI. (E) Comparison of the number of patients with or without clinical remission at week 6 in Gemini II and III with a trough level between 40–55 µg/mL and >55 µg/mL. Statistical comparisons were performed using Student’s t-test (B, C) and Fisher’s exact test (E). Sample donor characteristics are listed in online supplemental tables 10 and 11. T_Eff, effector T cell; T_Reg, regulatory T cell; VDZ, vedolizumab.

Since these data further suggested that residual T_Reg cell homing might crucially contribute to clinical efficacy of vedolizumab, we stained colon biopsies from responders to vedolizumab therapy obtained before the initiation of and under vedolizumab treatment for CD4 and Foxp3. While there was no quantitative difference in overall CD4⁺ T cells before and under therapy, the portion of Foxp3⁺CD4⁺ cells was significantly increased in patients with active therapy compared with before therapy (figure 9B). Interestingly, further stainings showed that more Foxp3⁺ cells present in the colon of patients treated with vedolizumab co-stained for β1 than before treatment (figure 9C). Again and on tissue level, this was consistent with the idea of residual gut homing of β1⁺ T_Reg cells under vedolizumab therapy.

Finally, we performed a post-hoc analysis of phase III data from the Gemini II and III trials of vedolizumab in patients with CD to correlate our observations to clinical outcomes. We determined the primary efficacy endpoint (remission rate at week 6) depending on the corresponding vedolizumab trough levels. Intriguingly, when stratifying for serum concentrations as in our cohort, remission rates in the range from 40 to 55 µg/mL vedolizumab were clearly higher than below and above. In a pooled analysis, the difference between the 40 to 55 µg/mL and the above 55 µg/mL group was significant (figure 9D,E). In conclusion, these observations were well reconcilable with non-linear dose–response characteristics due to residual homing of β1⁺PI16⁺ T_Reg cells.