An anti-CD19/CTLA-4 switch improves efficacy and selectivity of CAR T cells targeting CD80/86-upregulated DLBCL

Abstract

Chimeric antigen receptor T cell (CAR T) therapy is a potent treatment for relapsed/refractory (r/r) B cell lymphomas but provides lasting remissions in only ∼40% of patients and is associated with serious adverse events. We identify an upregulation of CD80 and/or CD86 in tumor tissue of (r/r) diffuse large B cell lymphoma (DLBCL) patients treated with tisagenlecleucel. This finding leads to the development of the CAR/CCR (chimeric checkpoint receptor) design, which consists of a CD19-specific first-generation CAR co-expressed with a recombinant CTLA-4-linked receptor with a 4-1BB co-stimulatory domain. CAR/CCR T cells demonstrate superior efficacy in xenograft mouse models compared with CAR T cells, superior long-term activity, and superior selectivity in in vitro assays with non-malignant CD19⁺ cells. In addition, immunocompetent mice show an intact CD80^-CD19⁺ B cell population after CAR/CCR T cell treatment. The results reveal the CAR/CCR design as a promising strategy for further translational study.

Keywords: CAR T cells; CD19; CD80; CD86; DLBCL; FL; checkpoint ligand; chimeric checkpoint receptor; lymphoma; neurotoxicity.

Graphical abstract

Figure 1

Immune checkpoint ligands CD80 and CD86 are expressed in most tumor biopsies derived from DLBCL patients after receiving CAR (2^nd Gen) T cells (A) Evaluation of CD80/86 expression in lymphoma slides before and after treatment or at relapse as a mean H score, separated into tumor and tumor microenvironment cells where possible. See also Figures S1A‒S1C. Bar plots represent mean ± SEM of five high-power fields per slide. Statistical significance levels were determined by using a non-adjusted t test and reported according to p values (thresholds below). (B) Representative optical field section of DLBCL patient lymph node slides stained to show CD80/CD86 expression before and after treatment with tisagenlecleucel. Scale bars, 100 μm (original) and 20 μm (zoomed images). (C) Antigen expression of CD80 and CD86 in singularized DLBCL biopsy samples (CD19⁺/CD5^– gated) and peripheral blood samples of CLL (CD19⁺/CD5⁺ gated) and healthy donors (CD19⁺/CD20⁺ gated) measured by flow cytometry and plotted in comparison. Bar plots represent mean fluorescence intensity of each sample. (D) Comparison of CD80/CD86 transcript abundance across different cell types and lymphoma entities in the Brune et al. set. The horizontal line marks the median abundance of CD80/86 in healthy germinal center B cells. (E) CD86 transcriptional abundance across COO classification, upper plot from the Schmitz et al. set and lower plot from the Chapuy et al. set. (F and G) CD80/CD86 transcriptional abundance across genetic subtype clusters in the Chapuy et al. set and across revised LymphGen subtypes in the Schmitz et al. set. Horizontal line marks median abundance of all subtypes, which also serves as the reference group for individual group statistical testing. Statistical significance in transcriptomic data was evaluated using a one-sided unpaired non-adjusted Mann-Whitney U test and significance levels reported according to p values: ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001.

Figure 2

Design, expression, and activity of CD19/CD80/86-specific CAR T cells (A) Expression profiles of CD80 and CD86 in DLBCL cell lines used for in vitro experiments in this paper. See also Figure S2A. (B) Expression profiles of CD80 and CD86 in the Raji cell line used for in vitro and in vivo experiments. (C) Stylized representation of receptors used in this study. (D) Stylized representation of transduction vectors used in this study. (E) Representative FACS plots showing expression of anti-CD19 CAR and CTLA-4 receptors on transduced PBMCs. See also Figures S2B and S2C. (F) Expression of CD25 on CAR (+CCR) T cells and untransduced cells stimulated with CD19 and/or Ipilimumab 11 days after first activation. Bar plots represent mean fluorescence intensity of pooled (n = 2) samples. (G) Cytotoxicity of CAR/CCR and CAR T cells in co-culture with Burkitt and DLBCL lymphoma cell lines relative to controls without T cells. Significance levels derived from t tests comparing CAR/CCR with CAR (2^nd Gen), unmarked comparisons are not significant. Line plots represent mean ± SEM, n = 3 for constructs and n = 12 for no T cell control. (H) Depletion of patient-derived tumor cells in co-culture assay evaluated using FACS and comparing CAR/CCR T cells with CAR T cells, with both showing significant cell depletion compared with Mock. Bar plots represent mean ± SEM, n = 3. (I) Interferon-γ secretion of CAR/CCR and control CAR T cells in co-culture with tumor cell lines and patient tumor cells. Bar plots represent mean ± SEM, n = 3/4). Statistical significance was evaluated using a two-sided unpaired non-adjusted t test with significance levels reported according to p values: ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001.

Figure 3

CAR/CCR T cells improve response rates and survival time of mice in xenograft B cell lymphoma model (A) Timeline of injections and luminescence measurements in Raji lymphoma xenograft mouse trial in Rag2^tm1.1Flv Il2rg^tm1.1Flv (Rag2^– γc^–) mice evaluating first-line efficacy of CAR T constructs (n = 25). (B) Kaplan-Meier plot of mouse trial. Significance derived from pairwise log-rank test, p values reported. (C) Luminescence plot describing tumor burden. (D) Representative selection of mice pictures with luminescence overlay describing tumor burden. (E) Representative optical fields of spleen slides from Rag2^tm1.1Flv Il2rg^tm1.1Flv (Rag2- γc-) mice in a preliminary mouse trial. Scale bars, 100 μm.

Figure 4

CAR/CCR T cells show improved survival as a second line treatment after conventional CAR T (2^nd Gen) therapy in xenograft lymphoma model (A) Graphical representation of second-line xenograft mouse trial protocol. Rag2^tm1.1Flv Il2rg^tm1.1Flv (Rag2^– γc^–) received a Raji lymphoma intravenous xenograft (n = 22) and were treated with first-line (2^nd Gen) CAR T cells. Mice showing tumor relapse (n = 9) were divided into groups and treated with either CAR/CCR T cells (n = 5) or CAR (2^nd Gen) T cells (n = 4). (B) Kaplan-Meier plot of second-line mouse trial. Significance is evaluated via log-rank test and p value reported. (C) Luminescence plot describing tumor burden. (D) Representative selection of mice pictures with luminescence overlay describing tumor burden. (E) CD25 expression and cytokine secretion of stimulated CAR (+CCR) and untransduced T cells 24 days after first activation. Bar plots represent mean fluorescence intensity of pooled (n = 2) samples and mean of supernatant cytokines).

Figure 5

CAR/CCR T cells reveal reduced IL-2 and IL-6 secretion when co-cultured with tumor cells and reduced activation when co-cultured with healthy B cells or CD19⁺ MSC-derived cells (A) IL-2 secretion of CAR/CCR and control CAR T cells in co-culture with tumor cell lines and patient tumor cells. Bar plots represent mean ± SEM, n = 3/4. (B) Cytokine secretion across effector-target cell number ratios. Line plots represent mean ± SEM, n = 3; unmarked comparisons are not significant. (C) Flow cytometry plot of PMBCs expanded according to the transduction protocol stained to show CD16⁺ cells capable of secreting IL-6. See Figure S3E for gating strategy. (D) IL-6 secretion of PBMCs comparing samples containing CD16⁺ with samples without added CD16⁺ cells. Bar plots represent mean ± SEM, n = 3. (E) Graphical representation of healthy B cell depletion and cytokine secretion assay co-culturing transduced CAR T cells with heterogeneous PBMC samples from the same buffy coat containing healthy B cells. (F) Depletion of healthy B cells and associated cytokine secretion in co-culture. Bar plots represent mean ± SEM, n = 4/5 for flow cytometry, n = 3 for ELISA). (G) Graphical representation of VW-MSC differentiation and co-culture assay procedure. (H) FACS plots of post-differentiation VW-MSC-derived pericytes before and after MACS-based separation. Virtually all CD19⁺ cells also express the pericyte marker CD248 (see also Figure S4B). (I) IFN-γ secretion as a marker for CD19-directed CAR-induced activation comparing CAR/CCR, CAR (2^nd Gen), and mock T cells. Bar plots represent mean ± SEM, n = 4). Statistical significance was evaluated using a two-sided unpaired non-adjusted t test and significance levels reported according to p values: ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001.

Figure 6

CAR/CCR T cell model decreases CD80/86 positivity rate in an autochthonous lymphoma mouse model (A) Representative FACS plots showing expression of the mCAR and mCAR/mCCR. (B) Representative FACS plots showing expression of CD19, CD80, and CD86 on a stable mouse lymphoma cell line and non-malignant murine B cells used in this study. (C) Results of in vitro co-culture assays with mCAR (2^nd Gen) and mCAR/mCCR with the lymphoma cell line and healthy B cells. Bar plots represent mean ± SEM, n = 3. (D) Results of in vivo selectivity trial showing tumor size and CD80 positivity rate of B cells over time. For gating strategy see Figure S5C. Statistical significance was evaluated using a two-sided unpaired non-adjusted t test and significance levels reported according to p values: ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001.