ARI0003: Co-transduced CD19/BCMA dual-targeting CAR-T cells for the treatment of non-Hodgkin lymphoma

Abstract

CD19 CAR-T therapy has achieved remarkable responses in relapsed/refractory non-Hodgkin lymphoma (NHL). However, challenges persist, with refractory responses or relapses after CAR-T administration linked to CD19 loss or downregulation. Given the co-expression of CD19 and BCMA in NHL, we hypothesized that dual targeting could enhance long-term efficacy. We optimized different dual-targeting approaches, including co-transduction of two lentiviral vectors, bicistronic, tandem, and loop and pool strategies, based on our academic anti-CD19 (ARI0001) and anti-BCMA (ARI0002h) CAR-T cells. Comparison with anti-CD19/CD20 or anti-CD19/CD22 dual targeting was also performed. We demonstrate that anti-CD19/BCMA CAR-T cells can be effectively generated through the co-transduction of two lentiviral vectors after optimization to minimize competition for cellular resources. Co-transduced T cells, called ARI0003, effectively targeted NHL tumor cells with high avidity, outperforming anti-CD19 CAR-T cells and other dual-targeting approaches both in vitro and in vivo, particularly in low CD19 antigen density models. ARI0003 maintained effectiveness post-CD19 CAR-T treatment in xenograft models and in spheroids from relapsed CART-treated patients. ARI0003 CAR-T cells were effectively manufactured under Good Manufacturing Practice conditions, with a reduced risk of genotoxicity compared to other dual-targeting approaches. A first-in-human phase 1 clinical trial (CARTD-BG-01; this study was registered at ClinicalTrials.gov [NCT06097455]) has been initiated to evaluate the safety and efficacy of ARI0003 in NHL.

Keywords: CAR-T cells; co-transduction; dual targeting; lymphoma.

Graphical abstract

Figure 1

Generation of dual-targeting CAR-T cells by co-transduction of lentiviral vectors (A) Schematic representation of CAR constructs. (B) Representative flow cytometry plots of CAR staining on the T cell membrane in single transduction for ARI0001 (top), for ARI0002h (center), and in co-transduction (bottom) at the indicated MOIs. (C) Quantification of CAR expression shown in (B). Bar graphs depict mean percentage of transduction in single or co-transduction ±SD (n = 6 healthy donors). Two-way ANOVA tests. ∗p < 0.05; ∗∗p < 0.005; ∗∗∗p < 0.0005; ∗∗∗∗p < 0.0001. (D–F) T cells after single or co-transduction with high MOIs (MOI = 10) were studied at protein (D), DNA (E), and RNA (F) levels. Bar graphs depict mean ± SD. Each dot represents a healthy donor (n = 5). (D) Percentage of CAR expression on the T cell membrane analyzed by flow cytometry. (E) Integrated DNA LV copies per genome determined by qPCR (n = 6). (F) CAR mRNA levels in co-transduced and single-transduced CAR-T cells determined by real-time PCR (n = 6). (G) Quantification of CAR expression for the ARI0001 CAR codon optimized version 1 (top) and for the ARI0002h CAR (bottom) after single- and co-transduction (n = 3). Two-way ANOVA tests. ∗p < 0.05; ∗∗p < 0.005; ∗∗∗p < 0.0005; ∗∗∗∗p < 0.0001. (H) Representative flow cytometry plots of CAR staining on the T cell. An MOI of 10 was used for single transductions, while three combinations of MOIs were used for CAR co-transduction. Representative of two experiments using different healthy donors. (I) Percentage of CAR⁺-T cells after transduction with a single LV or after co-transduction (ARI0003). Each dot represents a donor (n = 6 for healthy donors and n = 4 for NHL patients). (J) CAR distribution within the CAR⁺ population of ARI0003. Mean ± SD from n = 6 healthy donors and n = 4 NHL patients.

Figure 2

ARI0003 cytokine production, cytotoxicity, and in vivo antitumor effects (A) Control T and CAR-T cells were cultured with the indicated target cells at an E:T ratio of 1:3. A Bioluminescence imaging (BLI)-based killing assay was used to evaluate the lysis of the indicated tumor cell lines over time. The mean ± SD of triplicates from a healthy donor is shown. Representative of two different experiments with two healthy donors (n = 2). (B and C) Control T and CAR-T cells were co-cultured with the indicated target cells at an E:T ratio of 1:1 in (B) and 3:1 in (C). Bar graphs depict mean ± SD. Each point represents a healthy donor (n = 3). (B) Specific tumor lysis was calculated relative to target cells without treatment using a 16-h luciferase-based killing assay. (C) Supernatants were collected 24 h after co-culture, and the release of TNF-α and IFN-γ was analyzed by ELISA. (D and E) The strength of interaction between single- and dual-targeting T cells and Ramos WT (D) or Ramos CD19 Low (E) target cells was measured. The percentage of total CAR-T cells remaining bound to target cells as the acoustic force ramp is applied from 0 to 1,000 pN is shown (right). The fold change of bound CAR-T cells at 1,000 pN as compared to UTD is shown (left). Data are presented as means ± SEMs from two (Ramos WT) or three (Ramos CD19 Low) healthy donors in technical triplicate (n = 2–3). p values were calculated using a one-way ANOVA test. ∗p < 0.05; ∗∗p < 0.01. (F) “Advanced” Burkitt lymphoma model. NSG mice were i.v. injected with Ramos-GFP-luciferase cells. After 14 days, mice were randomized and treated with 3 × 10⁶ T cells, including UTD control T cells (n = 4), ARI0001 (n = 6), ARI0002h (n = 6), or ARI0003 (n = 6). The Kaplan-Meier survival curves of treated mice are plotted, and mean survival (days) calculated. (G) “Advanced” Burkitt lymphoma model with low CD19 densities. NSG mice were i.v. injected with Ramos-GFP-luciferase-CD19 Low cells. After 14 days, mice were randomized and treated with 2 × 10⁶ control T cells (n = 7) or CAR⁺-T cells, including ARI0001 (n = 7), ARI0002h (n = 9), or ARI0003 (n = 8). The Kaplan-Meier survival curves of treated mice are plotted and mean survival (days) calculated. Log rank (Mantel-Cox) test. ∗p < 0.05; ∗∗p < 0.005; ∗∗∗p < 0.0005; ∗∗∗∗p < 0.0001.

Figure 3

Generation and expression characterization of different CD19-BCMA dual-targeting strategies (A) Schematic representation of CAR constructs. (B) Structural models of CAR constructs were generated using AlphaFold3, based on the input sequences of their extracellular domains (CD8α hinge and scFv). The BCMA-targeting regions are depicted in gray, the CD19-targeting regions in blue, the linkers connecting both regions are shown in purple, and the CD8 hinge in yellow. (C and D) CAR expression on the T cell membrane was analyzed by flow cytometry. (C) Levels of extracellular CAR and total CAR from single and dual-targeting strategies are shown. Bar graphs show the mean ± SD (n = 3 healthy donors). Total CAR was assessed using intracellular staining. Statistical analysis was performed using a two-way ANOVA. ∗∗p < 0.005; ∗∗∗∗p < 0.0001. (D) Flow cytometry plots from a representative healthy donor showing CD19-CAR and/or BCMA-CAR expression on the surface of T cells. (E and F) The CAR mean fluorescence intensity (E) and (F) the number of CAR molecules of dual-targeting strategies compared to ARI0001 (left) or ARI0002h (right) are shown (n = 3–6 healthy donors). The fold decrease in MFI and CAR molecule count relative to single-CAR constructs is calculated and analyzed using a one-sample t test. ∗∗p < 0.005; ∗∗∗p < 0.0005; ∗∗∗∗p < 0.0001. (G and H) Single and dual-targeting strategies were studied at both the DNA (G) and RNA (H) levels. (G) The number of lentiviral DNA copies integrated per genome for CD19 CARs (left) and BCMA CARs (right) compared to ARI0001 or ARI0002h was assessed using qPCR (n = 3 healthy donors) and analyzed using the one-sample t test. ANOVA test. ∗p < 0.05. (H) CAR RNA levels in the dual-targeting strategies compared to ARI0001 (left) or ARI0002h (right) were measured by real-time PCR (n = 3 healthy donors) and analyzed using a one-sample t test. ∗p < 0.05; ∗∗p < 0.005.

Figure 4

ARI0003 shows enhanced functionality compared to other dual-targeting strategies (A) K652 cells were genetically modified to express truncated CD19, BCMA, or both antigens. CD19 and BCMA expressions were analyzed by flow cytometry. (B) The in vitro killing ability of the different strategies was analyzed after co-culture with K562 cells. Indicated target cells were co-cultured at an E:T ratio of 1:1 with CAR-T cells for 24 h. Specific lysis of cells expressing one or both antigens was measured in a luciferase-based killing assay. Bar graphs show the mean ± SD. Each point represents a healthy donor (n = 3). (C and D) The strength of interaction among all dual-targeting strategies with Ramos WT (C) or Ramos CD19 Low (D) target cells was measured. The percentage of total CAR-T cells remaining bound to target cells as the acoustic force ramp is applied from 0 to 1,000 pN is shown (left). The fold change of the percentage of bound CAR-T cells versus UTD T cells at 1,000 pN is shown (right). Data are presented as means ± SEMs from two donors in technical triplicate (n = 2). p values were calculated using a one-way ANOVA test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001. (E–H) Long-term cytotoxicity and proliferation of dual-targeting strategies. Control T and CAR-T cells were co-cultured with Ramos or Ramos CD19 Low for 7 days. (E) CD3⁺ (T lymphocytes) and GFP⁺ (tumor) populations after 7 days of co-culture in a representative healthy donor. (F) Quantification of the distribution of populations observed in (E). (G) T cell proliferation analyzed by flow cytometry. Each point represents a different experiment with a different healthy donor (n = 2). (H) Surface CAR expression levels across different CAR⁺ populations on ARI0003 and pooled T cells, both before and after co-culture with the Ramos WT cell line at an E:T ratio of 1:4. Results were obtained from three distinct healthy donors (n = 3) and are presented as the mean ± SD. Statistical significance was evaluated using a one-way t test. ∗p < 0.05.

Figure 5

Enhanced efficacy of ARI0003 in NSG mice and PDLS assays (A and B) ARI0003 in vivo antitumor activity in a “CAR-T limiting dose” Burkitt lymphoma xenograft model. NSG mice were i.v. inoculated with lymphoma cells expressing luciferase and treated 4 days later with 5 × 10⁵ control T cells (n = 7), ARI0001 (n = 9), pooled CAR-T (n = 9), ARI0003 (n = 10), or bicistronic CAR-T cells (n = 9) (60% CAR⁺-T cells for all groups). Disease progression was measured by bioluminescence photometry (IVIS). (A) Quantitative analysis of BLI as total flux (p/s) per individual animal in each group. (B) The Kaplan-Meier survival curves of treated mice are plotted and mean survival (days) calculated. Log rank (Mantel-Cox) test. (C–F) PDLS from a follicular lymphoma patient as a preclinical model to test ARI0003 CAR-T cells. (C) CD19 and BCMA antigen expression on PDLS before CAR-T cell co-culture analyzed by flow cytometry. (D) Images captured with Cytation 1 of representative PDLS that were formed for 4 days and then co-cultured with CAR-T cells for 2 more days. T cells are stained with CellTracker Violet (blue) and dead cells with propidium iodide (red). Loss of sphericity indicates CAR-T cell antitumor effect. (E) Specific killing of tumor spheroids by CAR-T cells overtime after co-culture at an E:T ratio of 1:2 as analyzed by flow cytometry. Data are presented as mean ± SD for triplicates. (F) Cytokine production by ELISA after 24-h co-culture at an E:T ratio of 1:2. Each dot in the ELISA represents a spheroid (n = 3). Two-way ANOVA test with Tukey’s correction. ∗∗p < 0.01; ∗∗∗p < 0.001.

Figure 6

ARI0003 as therapeutic option for NHL following relapse to CD19 CAR-T cell treatment (A–D) ARI0003 antitumor efficacy in a mouse model of NHL relapse following ARI0001 treatment. (A) Schematic representation of the experiment design. NSG mice were inoculated with lymphoma cells expressing low CD19 densities and treated with 2 × 10⁶ ARI0001 (n = 28) or control T cells (n = 6). Two weeks later, animals treated with ARI0001 received a second treatment consisting of 3 × 10⁶ control T cells (n = 7), ARI0001 (n = 7), ARI0003 (n = 7), or pooled (n = 7) CAR-T cells (60% CAR⁺ expression). Disease progression was measured by bioluminescence photometry (IVIS). (B) Mice pictures from the day before treatment to 7 weeks after treatment. (C) Kaplan-Meier survival curves for treated mice and mean survival (days). ∗p < 0.05 and ∗∗p < 0.005 by log rank (Mantel-Cox) test. (D) Quantitative analysis of BLI as total flux (p/s) per individual animals in each group. (E–H) PDLS from patients who relapsed after CD19-CAR-T cell therapy served as an ex vivo preclinical model to challenge CAR-T cells. (E and G) Histograms of CD19 and BCMA antigen expression on PDLS before CAR-T cell treatment, for CD19⁺ and CD19⁻ PDLS, respectively. (F and H) Specific killing of tumor cells of each CAR-T cell effector group after co-culture with CAR-T cells at an E:T ratio of 1:2 was compared to control T cells and analyzed by flow cytometry, for CD19⁺ (F) and CD19⁻ (H) PDLS, respectively. Data presented are mean ± SD for triplicates. IFN-γ production after 24-h co-culture at an E:T ratio of 1:2 was analyzed by ELISA in PDLS supernatant. Each dot in the ELISA corresponds to a spheroid (n = 2–3).

Figure 7

CD19 and BCMA CAR co-transduction outperforms CD19 combined with CD20 or CD22 CARs (A) Schematic representation of CAR constructs. (B) The percentage of CAR-expressing T cells analyzed by flow cytometry is shown (n = 2 healthy donors). (C) Monospecific CAR-T cells and dual-CAR-T cells were cultured with Ramos CD19 Low cells at indicated E:T ratios. Tumor cell lysis was evaluated over time using a BLI-based killing assay. Data represent the mean ± SD of triplicates from two healthy donors. (D–F) Long-term cytotoxicity and proliferation of single and dual-targeting CAR-T cell strategies. CAR-T cells were co-cultured with Ramos CD19 Low cells for 7 days (n = 2 healthy donors). (D) Representative flow plots showing CD3⁺ (T cells) and GFP⁺ (tumor cells) populations. (E) Quantification of population distributions from (D). (F) T cell proliferation after 7 days of co-culture analyzed by flow cytometry. Bar graphs show the mean ± SD of the CD3⁺ T cell fold-increase. Each point represents an experiment with a different healthy donor (n = 2). p Values were calculated using a one-way ANOVA test. ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, ∗p ≤0.05

Figure 8

Characterization of ARI0003 batches produced under GMP conditions (A) Percentage of CAR⁺-T cells after transduction at clinical level analyzed by flow cytometry. Each dot represents an ARI0003 batch produced. (B) CAR distribution within the CAR⁺ population of ARI0003. (C) ARI0003 cells were co-cultured with Ramos WT target cells at indicated E:T ratios. Lysis of the tumor cells was evaluated 24 h after co-culture by flow cytometry. Results represent the mean ± SD of n = 5 clinical grade manufactured products. Statistics in (C) were performed with multiple comparisons using two-way ANOVA. ∗∗p < 0.005; ∗∗∗∗p < 0.0001.