Human OX40L-CAR-Tregs target activated antigen-presenting cells and control T cell alloreactivity

Abstract

Regulatory T cells (T_regs) make major contributions to immune homeostasis. Because T_reg dysfunction can lead to both allo- and autoimmunity, there is interest in correcting these disorders through T_reg adoptive transfer. Two of the central challenges in clinically deploying T_reg cellular therapies are ensuring phenotypic stability and maximizing potency. Here, we describe an approach to address both issues through the creation of OX40 ligand (OX40L)-specific chimeric antigen receptor (CAR)-T_regs under the control of a synthetic forkhead box P3 (FOXP3) promoter. The creation of these CAR-T_regs enabled selective T_reg stimulation by engagement of OX40L, a key activation antigen in alloimmunity, including both graft-versus-host disease and solid organ transplant rejection, and autoimmunity, including rheumatoid arthritis, systemic sclerosis, and systemic lupus erythematosus. We demonstrated that OX40L-CAR-T_regs were robustly activated in the presence of OX40L-expressing cells, leading to up-regulation of T_reg suppressive proteins without induction of proinflammatory cytokine production. Compared with control T_regs, OX40L-CAR-T_regs more potently suppressed alloreactive T cell proliferation in vitro and were directly inhibitory toward activated monocyte-derived dendritic cells (DCs). We identified trogocytosis as one of the central mechanisms by which these CAR-T_regs effectively decrease extracellular display of OX40L, resulting in decreased DC stimulatory capacity. OX40L-CAR-T_regs demonstrated an enhanced ability to control xenogeneic graft-versus-host disease compared with control T_regs without abolishing the graft-versus-leukemia effect. These results suggest that OX40L-CAR-T_regs may have wide applicability as a potent cellular therapy to control both allo- and autoimmune diseases.

Fig. 1.. Generation of OX40L–CAR-T_regs with selective and sustained expression in FoxP3⁺ cells.

(A) Schema of OX40L-CAR and control constructs. L, leader, TM, transmembrane domain. (B) Transduction efficiencies measured by expression of NeonGreen fluorophore. Quantification of six independent experiments. (C) T_reg in vitro expansion over time in culture. A representative of three experiments from different donors is shown. (D and E) Surface expression of the OX40L-CAR on T_regs after incubation with biotinylated OX40L and staining with streptavidin-allophycocyanin. Cells were gated on the NeonGreen⁺ population. Representative flow cytometry plots (D) and quantification of five independent experiments (E) are shown. (F) Schema of OX40L–CAR-T_reg generation timeline. (G and H) Expression of FoxP3 at time of harvest (day 23) in different T cell subsets was measured. Representative flow cytometry plots (G) and quantitation of three independent experiments (H) are shown. Untrans., untransduced. (I) Expression of NeonGreen, as a surrogate for CAR or control vector expression, was measured during in vitro expansion. Data shown are a quantification of three independent experiments. Data are presented as the means ± SEM. Data were analyzed by an unpaired t test (B and D), one-way ANOVA with Tukey’s post hoc test (H) or two-way ANOVA with Šidák’s post hoc test (I). n.s., not significant; **P < 0.01 and ***P < 0.001.

Fig. 2.. OX40L–CAR-T_reg activation through OX40L leads to enhanced suppressive programs without enhancing proinflammatory cytokines.

(A) Experimental design of CAR-T_reg activation conditions with WT K562 (OX40L⁻), OX40L⁺ K562 (K562-OX40L⁺), or CD3/CD28 Dynabeads. Anti-OX40 antibody (OX40 Ab) added as indicated to prevent endogenous OX40L/OX40 interaction. (B to D) Expression of activation markers CD69 and CD71 was measured after CAR stimulation. Representative flow cytometry plots (B) and quantification of the frequency of CD69⁺ cells (C) and CD71⁺ cells (D) are shown. Data shown represent the summary of four independent experiments from four donors. (E) Expression of suppression markers CTLA-4, LAP, CD71, GARP, and LAG-3 and proinflammatory markers IFN-γ, TNF-α, IL-17A, and IL-2 after CAR stimulation. Markers were measured by flow cytometry, and the mean values are displayed on the radar plots. The left panel shows control T_regs, and the right panel shows CAR-T_regs. Additional quantitation is shown in fig. S2. Data are presented as the means ± SEM. Data were analyzed by two-way ANOVA with Tukey’s post hoc test. **P < 0.01 and ***P < 0.001.

Fig. 3.. CAR stimulation enhances suppressive function of T_regs.

(A) Experimental design of CAR-T_reg suppression of bead-stimulated T cells. CTV-labeled pan-T cells were activated with anti-CD3/CD28 beads. Separately, CAR or control T_regs were stimulated with WT K562 (OX40L⁻) or OX40L⁺ K562 cells. After 24 hours, beads were removed, and cells were cocultured at the indicated T_reg:pan-T cell ratio. (B to D) T cell proliferation was measured by CTV dilution in the coculture assay with the indicated T_regs. Representative flow cytometry plots (B) and percentage suppression of CD4 (C) and CD8 (D) T cell proliferation are shown. Shown is a representative experiment of three independent repeats. (E) Experimental design of CAR-T_reg suppression of unidirectional MLR. CTV-labeled responder PBMCs were cocultured with autologous CAR or control T_regs and irradiated allogeneic stimulator HLA-A*02⁺ PBMCs. (F to H) Representative CD4 T cell proliferation plots (F) and percentage suppression of CD4 (G) and CD8 (H) T cell proliferation are shown. Data in (G) and (H) are representative of three independent experiments. (I) Experimental design of CAR-T_reg–mediated suppression of costimulatory molecule expression on Mo-DCs. (J to L) Shown are the proportions of CD83⁺CD14⁺CD4⁻ Mo-DCs expressing costimulatory molecules CD80 (J), CD86 (K), and OX40L (L) after coculture with CAR-T_regs or control T_regs. Mean fluorescence intensity for corresponding markers is shown in fig. S3 [(A) to (C)]. [(J) to (L)] Shown is one representative experiment of six independent repeats. Data are presented as the means ± SEM. Data were analyzed by two-way ANOVA with Tukey’s [for (C) and (D)] or Šidák’s [for (G), (H), and (J) to (L)] post hoc test; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 4.. OX40L–CAR-T_regs remove OX40L from target cells through trogocytosis.

(A) Experimental design of OX40L transfer. WT K562 (OX40L⁻) or K562-OX40L⁺ target cells were cocultured for 3 hours with CAR-T_regs or control T_regs, followed by flow cytometric detection of OX40L on K562 cells or T_regs. (B to D) Representative flow cytometry plots (B) and quantification of OX40L on target CD4⁻ K562-OX40L⁺ cells (C) and on T_regs (D). Shown is a representative experiment of three independent experiments. (E to G) DiD lipid–stained WT K562 (OX40L⁻) or K562-OX40L⁺ target cells were cocultured with CAR-T_regs or control T_regs for 1.5 hours, followed by detection of DiD staining transfer by flow cytometry. Shown are the proportions of DiD⁺ CAR-T_regs or control T_regs (E), representative flow cytometry plots depicting counter-staining for DiD and OX40L (F), and proportions of OX40L⁺DiD⁺ and OX40L⁺DiD⁻ CAR-T_regs and control T_regs (G). Data shown are from a representative experiment of two independent repeats. Data are presented as the means ± SEM. Data were analyzed by two-way ANOVA with Tukey’s post hoc test. ***P < 0.001 and ****P < 0.0001.

Fig. 5.. OX40L-CAR mediates trogocytosis of OX40L from target cells.

(A) Representative images of T_reg (red) and K562 cell line (blue) cocultures stained with anti-OX40L (green). Mock-edited or OX40-KO CAR-T_reg and control T_regs were incubated with K562-OX40L⁺ cells for 15 min. Scale bars, 10 μm. (B) Quantification of OX40L fluorescence in 3D segmented, mock-edited, or OX40-KO CAR-T_reg and control T_regs cocultured with K562-OX40L⁺ cells. (C) Representative images of cocultures as in (A), except with WT (OX40L⁻) target cells. Scale bars, 10 μm. (D) Quantification as in (B), except with WT (OX40L⁻) target cells. Each dot in (B) and (D) represents a single cell. Data are presented as the median ± interquartile range (box) and total range (whiskers). Data were analyzed by a Kruskal-Wallis test with Dunn’s corrected post hoc test; **P < 0.01 and ****P < 0.0001.

Fig. 6.. OX40L–CAR-T_regs suppress APC-mediated T cell activation through OX40-OX40L.

(A) Schema of Mo-DC preconditioning by coculture with CAR-T_regs. (B and C) The ability of OX40L–CAR-T_reg–conditioned Mo-DCs and control T_reg-conditioned Mo-DCs to stimulate allo-CD4 (B) and allo-CD8 (C) T cell proliferation was quantified. The percentage suppression was determined by comparing the cell division stimulated with Mo-DCs without exposure to T_regs. Shown is a representative experiment of three independent repeats. (D and E) Shown are the percentages of CD4 (D) and CD8 (E) T cells expressing OX40 during allogeneic MLR after coculture with either control T_regs or CAR-T_regs. Data are presented as the means ± SEM. Data were analyzed by two-way ANOVA with Tukey’s post hoc test; *P < 0.05, **P < 0.01, and ****P < 0.0001.

Fig. 7.. OX40L–CAR-T_regs suppress GVHD in a xenogeneic transplant mouse model.

(A) Experimental design of xeno-GVHD mouse model. Human PBMCs (8 × 10⁶) were retro-orbitally injected into sublethally (250 cGy) irradiated NSG mice alone or together with either autologous control T_regs or OX40L–CAR-T_regs (8 × 10⁶ cells, 1:1 ratio with PBMCs). (B and C) Expression of OX40L on the surface of human leukocytes before injection into NSG mice or on day 7 after injection into NSG mice (n = 5). Shown are representative flow plots depicting OX40L expression on human CD3⁻CD20⁺ B cells, CD4⁺CD3⁺CD20⁻ and CD8⁺CD3⁺CD20⁻ T cells, and CD3⁻CD20⁻HLA-DR⁺ monocytes/DCs recovered from spleens on day 7 after injection (B) and quantification of samples from PBMCs before injection and the spleen after injection (C). (D to F) After induction of xeno-GVHD as shown in (A), mice were monitored twice weekly for survival end points (D), weight loss (E), and GVHD composite clinical score (F). See fig. S7A for GVHD scoring. Shown are representative results of two independent experiments. Data are presented as the means ± SEM. Data were analyzed using a log-rank (Mantel-Cox) survival test (D) and mixed effect two-way ANOVA with Šidák’s post hoc test [(C), (E), and (F)]. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.