Increased Motility and Suppression of Ex Vivo–Expanded Regulatory T Cells Designed for Adoptive Transfer Therapy in Ulcerative Colitis

Inflammatory bowel diseases (IBDs) such as Crohn’s disease and ulcerative colitis (UC) are medically intractable, and many patients do not benefit sufficiently from the available treatments.¹

A hallmark of IBD is the infiltration of T cells to the intestinal lamina propria. Although the potentially deleterious functions of proinflammatory T cells are balanced by anti-inflammatory regulatory T cells (Tregs) under homeostatic conditions, Tregs fail to control disease-driving T cells in IBD.² Thus, adoptive transfer of Tregs seems to be a promising approach for future therapy.³ While a previous study showed that ovalbumin-specific Tregs may counteract inflammation in Crohn’s disease,⁴ we recently developed a good manufacturing practice (GMP)-compliant protocol for the ex vivo expansion of autologous Tregs for adoptive transfer⁵ and a phase I dose-finding study in UC has now fully recruited patients.⁶

However, although increased expression of α4β7 integrin and L-selectin was described previously,⁵ it has not been explored to date whether these ex vivo–expanded autologous Tregs (hereafter called Investigational Medicinal Product [IMP]-Tregs) are functionally active for trafficking to the gut and how expansion affects their 3-dimensional (3D) motility and suppressive capacity.

Here, we present data suggesting that IMP-Tregs are equipped not only for successful gut homing in vivo, but also show enhanced motility in complex 3D matrices, predisposing them to increased effector T-cell suppression and substantiating them as a promising therapeutic candidate for UC.

To explore the gut-homing abilities of IMP-Tregs, we investigated adhesion to mucosal addressin cell adhesion molecule (MAdCAM)-1 and vascular cell adhesion molecule (VCAM)-1 as the intestinal endothelial ligands for the integrins α4β7 and α4β1.

Upon perfusion of nonexpanded and IMP-Tregs from the same donor through MAdCAM-1– and VCAM-1–coated capillaries,^7,8 we observed similar dynamic adhesion to both molecules, which was reduced substantially by treatment with the anti-α4β7 antibody vedolizumab or the anti-α4 antibody natalizumab (Supplementary Figure 1).

Supplementary Figure 1.

Dynamic adhesion assays of nonexpanded and IMP-Tregs in the presence or absence of (A) vedolizumab or (B) natalizumab and perfused through capillaries coated with coating buffer (uncoated), (A) MAdCAM-1 or (B) VCAM-1 as indicated. Representative images merged from 7 differentially colored high-power fields from a representative dynamic adhesion assay (only panel A) and pooled quantification of dynamic adhesion (n = 3). (C) Comparison of dynamic adhesion of nonexpanded and IMP-Tregs to MAdCAM-1 and VCAM-1 (n = 3). NTZ, natalizumab; VDZ, vedolizumab. ∗P < .05; ∗∗P < .01.

Adhesion to the endothelium is followed by transendothelial migration. To mimic this process, we coated porous membranes with MAdCAM-1 and VCAM-1 and assessed transmigration toward chemotactic chemokine (C-C motif) ligand 25 (CCL-25). We observed clearly increased migration of IMP-Tregs compared with nonexpanded Tregs over both MAdCAM-1– and VCAM-1–coated membranes (Figure 1A).

Figure 1.

Enhanced migratory features of IMP-Tregs. (A) Transmigration assays with nonexpanded or IMP-Tregs over membranes coated with MAdCAM-1 or VCAM-1 toward a chemokine (C-C motif) 25 (CCL-25) gradient. Left: Representative flow cytometry, absolute numbers of transmigrated cells are indicated. Right: Quantification of transmigrated cells; n = 3. ∗∗P < .01. (B) Labeled nonexpanded or IMP-Tregs were injected into the ileocolic artery of Rag1^-/- mice with dextran sodium sulfate colitis. Left: Representative intravital confocal microscopy. White arrows highlight extravasated Tregs. Right: Representative and quantitative flow cytometry of FarRed⁺ nonexpanded or IMP-Tregs among lamina propria mononuclear cells; n = 3. ∗P < .05. (C) Volcano plot of gene expression in nonexpanded and IMP-Tregs (n = 3 per group) as determined by bulk-RNA sequencing. Selected genes involved in cell migration are highlighted. The dashed horizontal line marks significance level (adjusted P < .05). FSC, forward scatter; SSC, sideward scatter.

Together, these data suggested that IMP-Tregs master key steps of gut homing, but the question of whether they successfully combine these aspects in vivo remained open. Thus, we adoptively transferred nonexpanded or IMP-Tregs in a humanized mouse model.⁹ Indeed, in vivo confocal microscopy showed extravasation of IMP-Tregs to the inflamed lamina propria. Subsequent analysis of lamina propria cells by flow cytometry showed increased recruitment of IMP-Tregs compared with nonexpanded Tregs to the inflamed gut (Figure 1B).

To understand the molecular alterations driving such superior homing capacity of IMP-Tregs, we explored the transcriptome of nonexpanded and IMP-Tregs. Indeed, IMP-Tregs displayed a specific profile of differentially expressed genes associated with cell migration (Figure 1C), suggesting that the expansion process imprints an enhanced migratory phenotype on IMP-Tregs.

After reaching the target tissue, Treg-mediated suppression of inflammation requires their migration through the extracellular matrix of the intestine to establish sufficient proximity with proinflammatory cells. To model this process, we performed 3D motility assays, whereby Tregs are placed in a complex collagen network to monitor spontaneous migration.¹⁰ IMP-Tregs showed substantially increased migration speed compared with nonexpanded Tregs. Moreover, the motile fraction and the directional persistence of these cells was clearly higher in IMP-Tregs (Figure 2A–C), and the expression of regulatory genes in IMP-Tregs was increased substantially (Figure 2D).

Figure 2.

Enhanced 3D motility and T-cell suppression of IMP-Tregs. (A) Representative maximum intensity projection along the z-axis of nonexpanded and IMP-Tregs moving through 3D collagen gels. Cell trajectories over 5 minutes are indicated in orange (scale bar: 25 μm). (B) Cell speed and directional persistence of nonexpanded and IMP-Tregs from 1 representative of 3 donors. (C) Quantification of speed, motile fraction, and directional persistence of nonexpanded and IMP-Tregs from 1 representative of 3 donors over 5 minutes. ∗P < .05; ∗∗∗P < .001. (D) Heat-map depicting expression of genes associated with suppressive function between nonexpanded and IMP-Tregs as determined by bulk-RNA sequencing (reads are normalized to the mean expression on day 0). (E) Gating strategy defining T-cell proliferation in suppression assays. (F) Representative histograms showing T-cell proliferation in the presence of nonexpanded and IMP-Tregs at a Treg–to–T-cell ratio of 1:1, 1:5, and 1:10. (G) Percentage suppression in the first generation of divided cells in the presence of nonexpanded and IMP-Tregs at the indicated Treg–to–T-cell ratios (n = 3). CFSE, carboxyfluorescein succinimidyl ester.

Thus, we hypothesized that enhanced motility together with higher suppressive capacity increases the immunosuppressive function by increasing the frequency and efficacy of close encounters with effector T cells. To test this, we performed carboxyfluorescein succinimidyl ester dilution assays and measured the inhibition of T-cell proliferation at various Treg–to–T-cell ratios. IMP-Tregs more efficiently suppressed autologous T cells at all ratios tested compared with nonexpanded Tregs (Figure 2E-G). In addition, the observed differences between nonexpanded and IMP-Tregs were more pronounced for smaller Treg–to–T-cell ratios, in which fewer Tregs must migrate over longer distances to retain the overall suppression of co-cultures that contain more Treg cells.

Taken together, these results underscore the functional fitness of our IMP-Tregs to reach their target tissue as well as to establish contact with and to suppress their proinflammatory counterparts. Our data suggest that enhanced transendothelial migration and suppressive function are a direct consequence of increased motility in 3D environments after expansion because Tregs must cover the distance between individual cells to exert T-cell suppression and active cellular movement is also a prerequisite to leave the peripheral circulation and transmigrate.

Thus, beyond substantiating the promising potential of our IMP-Tregs as a conceptually novel cell-based approach specifically supporting anti-inflammatory T-cell functions for the therapy of IBD, these data also have important implications for future basic and translational studies on immune cell interactions in intestinal inflammation because they underscore that it seems important to take tissue-migratory properties into account and to model them appropriately.

Supplementary Methods

Donors and Samples

This study was approved by the ethics committee of Friedrich-Alexander-Universität Erlangen-Nürnberg (151_12B, 417_19) and all donors provided their written informed consent.

Ex Vivo Treg Expansion

The Tregs used in this study were produced according to our previously published GMP-compliant Treg expansion protocol and were subjected to the implemented lot-release criteria to finally fulfill the criteria of the IMP.¹ Nonexpanded Tregs were directly cryopreserved after CD25-enrichment and stored in liquid nitrogen identical to the IMP-Tregs after ex vivo expansion until further use.

In Vitro Dynamic Adhesion Assays

In vitro dynamic adhesion assays were performed as previously described.^2,3 In brief, miniature borosilicate capillaries (Vitrocom) were coated with fragment crystallizable (Fc) chimera of recombinant human vascular cell adhesion molecule 1 (rhVCAM-1) (BioLegend) or recombinant human mucosal addressin cell adhesion molecule 1 (rhMAdCAM-1) (R&D Systems) at a final concentration of 5 μg/mL in 150 mmol/L NaCl with 10 mmol/L HEPES for 1 hour at 37°C. Plastic tubing was connected to the capillaries and inserted in an adjustable flow rate peristaltic pump (Baoding Shenchen Precision Pump Company). Nonexpanded Tregs and IMP-Tregs were resuspended in adhesion buffer (pH 7.4; 150 mmol/L NaCl, 10 mmol/L HEPES, 1 mmol/L CaCl₂, 1 mmol/L MgCl₂, and 1 mmol/L MnCl₂) at a concentration of 1.5 million cells/mL, incubated with or without 30 μg/mL anti-α4β7 integrin antibody vedolizumab (Takeda), or the anti-α4β1 integrin antibody natalizumab (Biogen), and perfused through the capillaries using the peristaltic pump. After 3 minutes, capillaries were rinsed to remove all nonadhering cells and imaged using a SP8 confocal microscope (Leica). Cells of 8 randomly chosen fields of view were quantified for each capillary.

Transmigration Assay

Nonexpanded Tregs and IMP-Tregs were thawed and resuspended in serum-free X-Vivo 15 medium (Lonza) at a concentration of 2 million cells/mL. Membranes of Transwell inserts (3-μm pore size) were coated with rhVCAM-1 or rhMAdCAM-1 at a concentration of 5 μg/mL in 150 mmol/L NaCl with 10 mmol/L HEPES for 1 hour at 37°C. Subsequently, the Transwell inserts were placed into the wells of a 96-well Transwell plate (Corning) filled with 235 μL X-Vivo 15 medium supplemented with 0.5 μg/mL chemokine (C-C motif) ligand 25 (CCL-25). A total of 160,000 cells was transferred to the upper compartments of the Transwell inserts. After an incubation time of 4 hours at 37°C, the inserts were removed and transmigrated cells in the lower compartments were quantified by flow cytometry using a MacsQuant16 instrument. Data were analyzed with FlowJo v10.6 software (BD).

RNA Sequencing of Nonexpanded and IMP-Tregs

Messenger RNA of nonexpanded and IMP-Tregs was isolated with the NucleoSpin RNA Kit for RNA isolation (Machery Nagel). Bulk-RNA sequencing was performed with Novogene (Cambridge, UK). For differential expression analysis, conventional mapping was performed using STAR v2.6.1c (GRCh38 Ensembl 98). The reads per gene were counted using FeatureCount of the R v3.6.1 package Subread v2.0.1. Differential gene expression was conducted using DESeq v1.24.0.

3D Cell Motility Assays

3D cell motility assays were performed as described previously.⁴ A total of 300,000 nonexpanded Tregs or IMP-Tregs were mixed into 1.5 mL of a 1.2 mg/mL collagen solution in each well of a tissue culture–treated 6-well plate (Corning). After 1 hour of polymerization at 37°C, 5% CO₂, and 95% relative humidity (RH), 1 mL RPMI medium was added to each well. During the measurement, the 6-well plate was placed in an incubation chamber (37°C, 5% CO₂, 95% RH) mounted to the microscope stage. Ten fields of views were measured for each well. Migration trajectories were recorded using bright-field microscopy (10× objective) image stacks with a frame rate of 1 stack every 15 seconds, during which z-scans in 10-μm steps through a 500-μm gel were taken. Image stacks were compressed by computing the minimum intensity projection, maximum intensity projection, z-position of minimum intensities, and z-position of maximum intensities. Moving cells were detected automatically, and from each cell trajectory we computed the mean velocity and directional persistence. Cells were counted as motile if they covered a distance of more than 13 μm during 5 minutes.

Suppression Assays

Suppression assays were performed according to our GMP-compliant lot-release criteria as described previously.¹

Humanized In Vivo Gut Homing Model

As described previously,⁵ immunodeficient B6.129S7-Rag1<tm1Mom>/J (Rag1-/-) mice were treated with dextran sodium sulfate (1.5% in drinking water) for 1 week. On day 7, nonexpanded or IMP-Tregs were thawed, stained with CellTrace FarRed (Thermo Fisher), and 1.5 million cells were injected into the ileocolic artery of 3 mice. Gut homing was analyzed by intravital confocal microscopy (Leica SP8 microscope). One hour after injection, lamina propria mononuclear cells were isolated using a lamina propria dissociation kit (Miltenyi Biotec) followed by density gradient centrifugation with Percoll (GE Healthcare). Subsequently, the frequency of FarRed⁺ cells was determined by flow cytometry using a MacsQuant10 instrument. Approval for animal experimentation was obtained from the Government of Lower Franconia.

Statistical Analyses

Statistical analyses were performed using GraphPad Prism (GraphPad Software, Inc). All results are shown as means with SEM. The Shapiro–Wilk test was used to test for normal distribution. For data with normal distribution, statistical differences were tested using the paired or unpaired Student t test as applicable. A P value < .05 was defined as statistically significant.