WHAT IS ALREADY KNOWN ON THIS TOPIC

Dual chimeric antigen receptor (CAR) expression has been achieved using two separate promoters, internal ribosomal entry sites or self-cleaving peptide sites.

These approaches dramatically increase the size of the genetic cassette and potentially lead to low transduction/transfection efficiency.

There is an urgent need to develop novel genetic constructs to increase antigen targeting range, while minimizing the cost, labor, and effectiveness of CAR T-cell therapy.

WHAT THIS STUDY ADDS

By incorporating a single compact bidirectional promoter into the Sleeping Beauty vector backbone, we obtained simultaneous expression of dual CARs, which ensures optimal functionality and offers a straightforward transfection process.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Bidirectional promoter-driven CARs offer a promising platform to improve the efficacy of antigen-targeting CAR T therapies and have further potential in combination with complementary therapies, for example, immune checkpoint inhibitor antibodies.

Background

Bidirectional promoters regulate 10% of human genes that are situated less than 1,000 base pairs apart from each other in head-to-head orientation.1 2 They are enriched with various transcription factor-binding sites, including GA-binding protein sites,3 which can enhance transcription activity in both directions, compared with unidirectional promoters. Studies have found that intergenic regions near CpG islands in the human genome function as bidirectional promoters4 and specifically favor bidirectional activity for genes arranged in a head-to-head orientation.5–8 Despite widespread interest in bidirectional promoters, only a small number of putative bidirectional promoters have been experimentally confirmed, and the functional mechanisms driving their activation are still poorly understood. Currently, bidirectional promoters in synthetic biology have not been widely adopted, especially in gene therapy such as chimeric antigen receptor (CAR)-T cells.

The US Food and Drug Federation has approved six CAR T-cell products for relapsed/refractory B-cell leukemia or lymphoma. Four of these targets the CD19 antigen on B cells. Single antigen-directed CAR T cells, such as CD19-targeted CAR T cells, have shown promising treatment outcomes.9 10 Unfortunately, the loss of target antigens in cancer cells is one of the primary reasons for disease relapse after CD19 CAR T-cell therapy.11 12 Targeting multiple molecules could overcome the limitation of antigen loss in CAR T-cell therapies. There are several combinations of target antigens for CAR T-cell hematologic malignancy, including CD19/CD20, CD19/CD22, and BCMA/other targets on B cells.13–17 These are accomplished by either pooling two single CAR T-cell products with different antigen-binding domains or a single CAR T-cell product capable of targeting two different antigens.11 18 19 Dual CAR T-cell strategies such as (1) sequential infusion of two separate CAR T-cell products, (2) cotransduction of T cells with two separate vectors encoding individual CAR structures, (3) bicistronic CARs that encode two CARs in one vector backbone, (4) tandem CARs linking two different scFvs in one CAR domain and (5) loop CARs linking the VL-VH of one scFv to the VL-VH of another scFv in one CAR domain have been growing and being used in clinical trials. However, these approaches have some limitations, as the sequential infusion is associated with a high manufacturing cost, as it requires rigorous quality control and can result in an uneven expansion of individual CAR cells. Similarly, cotransduction necessitates highly optimized protocols, resulting in high manufacturing costs for vectors and viruses. Last, bicistronic, tandem, and loop CARs employ large and complex genetic cassettes, which can potentially lead to low transduction efficiency.11 20–22

To minimize the cost, labor, and biases in CAR gene expression, there is a need to develop compact genetic constructs that enable multiantigen targeting. Gene therapy would benefit from the development of parsimonious genetic constructs able to simultaneously express two or more genes. Bidirectional promoters offer several advantages over existing platforms and have potential for not only expressing dual CARs but also immune checkpoint inhibitors, such as anti-PD-1, anti-CTLA-4, and anti-TIM-3 antibodies or immunostimulatory cytokines. However, the specific design and choice of the bidirectional promoter needs to be considered for each gene therapy application. In this study, we compared different bidirectional promoters in the Sleeping Beauty (SB) transposon system for the production of dual CD19 and CD20 CARs.

Materials and methodsPrimary cells and cell lines

Raji (human Burkitt lymphoma cell line) and K562 (human immortalized myelogenous leukemia cell line) cells were maintained in RPMI 1640 medium supplemented with GlutaMAX (Thermo Fisher Scientific, Waltham, Massachusetts, USA), penicillin (100 U/mL)—streptomycin (100 µg/mL) and 10% heat-inactivated fetal bovine serum (FBS) (Thermo Fisher Scientific, Waltham, Massachusetts, USA) at 37°C with 5% CO2. To generate CD19 and CD20 antigen-bearing K562 cell lines, SB transposons (pSBbi-RP; kind gift from Eric Kowarz)23 bearing CD19 and CD20 antigens (pSBbi-RP-CD19 and pSBbi-RP-D20) and transposase (pCMV(CAT)T7-SB100; kind gift from Zsuzsanna Izsvak)24 were transfected using the MaxCyte OC-25 electroporation protocol. Electroporation was performed with 2×106 K562 cells in 25 µL with 1 µg of pCMV-CAT and 5 µg of pSBbi-RP-CD19 or pSBbi-RP-CD20. The pSBbi-RP transposon vector carries a red fluorescent marker (dTomato) and a puromycin-resistance gene. Cells were selected with puromycin antibiotic after 24 hours of electroporation. Successful transfection was confirmed by analyzing the cells using flow cytometry and fluorescence microscopy. PBMCs were obtained from healthy blood donors with informed consent.

CAR construct generation in the Sleeping Beauty backbone

To generate bidirectional, dual CD19/CD20 CAR constructs, we PCR amplified the eukaryotic promoters EF-1α (1.2 kb), hPGK (527 bp), RPBSA (673 bp) and CMV (675 bp) from existing vectors. The EF-1α and RPBSA promoters from pSBbi-RP plasmid,23 hPGK promoter from pCCLsin plasmid (kind gift from Professor Dr Naldini)25 and CMV promoter from pCDNA3.1(-) plasmid were amplified and cloned into an SB backbone (pSBbi-RP). CD19 CAR (1582 bp) FMC63 anti CD19-scFv, the human CD8 hinge domain, the human CD8 transmembrane (TM), 4-1BB intracellular co-stimulatory domain and the CD3ζ activation domain were cloned in the forward orientation from each promoter. Next, green fluorescent protein (GFP) (717 bp) and CD20 CAR (1675 bp) 1F5 anti-CD20-scFv, with the human CD8 hinge domain, the human CD28 TM, CD28 intracellular co-stimulatory domain and the CD3ζ activation domain, were cloned in the reverse orientation (figure 1A). CD19 CAR expression was assessed using a biotinylated CAR19 detection peptide (CD19 CAR detection reagent, human, catalog number: 130-129-550, Miltenyi Biotec) and anti-biotin antibody, while the expression of CD20 CAR was evaluated based on the level of GFP expression throughflow cytometry. For the unidirectional dual CD19/CD20 CAR construct, GFP, CD19 CAR, and CD20 CAR were amplified from the bidirectional dual CAR construct and positioned after the EF-1α promoter with the 2A self-cleaving peptide (P2A) inserted between the GFP-CD19 and CD20 CAR in a unidirectional fashion. To construct the single EF-1α-driven CD19 CAR in the SB backbone, EF-1α and CD19 CARs were amplified from the bidirectional dual plasmid (CAR20.GFP.EF1.CAR19) and cloned and inserted into the pSBbi-RP backbone. Sequences were confirmed by Sanger sequencing

Figure 1

Different bidirectional promoter-driven dual CD19/20 CAR T-cell generation. (A) Schematic illustration of the Sleeping Beauty backbone bearing four different eukaryotic promoters (EF-1α, hPGK, RPBSA and CMV). (B) Schematic design of CAR T-cell generation. (C) Flow cytometry plot representing forward CAR19 expression on day 14 after transfection. (D) Flow cytometry plot representing GFP expression on day 14 after transfection. (E) Flow cytometry plot representing double CAR19+ and GFP+ cell per cent expression on day 14 after transfection. (F) Transposition efficiency of CAR19, GFP and CAR19+GFP+ on day 14 after transfection. (G) Mean fluorescent intensity of CAR19 and GFP on day 14 after transfection. All results are summarized as the mean±SEM from at least three independent healthy blood donors. Significant differences were determined by two-way analysis of variance, and asterisks indicate significant p values as follows: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. CAR, chimeric antigen receptor; CD, cluster of differentiation; FBS, fetal bovine serum; GFP, green fluorescent protein; IL, interleukin; P2A, 2A self-cleaving peptide; PBMCs, peripheral blood mononuclear cells; TM, transmembrane.

CAR T-cell generation and expansion

Fresh PBMCs were isolated from healthy donor blood samples using Ficol-Paque Premium (GE HealthCare, Bio-Sciences-AB, Sweden) and density gradient centrifugation. Plasmids were isolated from DH5-α using endotoxin-free Qiagen plasmid maxi kits. PBMC (4×106) were suspended in P3 primary cell 4D-Nucleofector X kit reagent (Lonza, Basel, Switzerland) with CAR constructs (SB transposon; 2 µg) and pCMV-CAT (SB transposase; 0.4 µg) at a 5:1 ratio in a total volume of 20 µL.

Electroporation was performed using 4D-Nucleofector followed by incubation in TexMACs medium supplemented with 5% FBS, 10 ng/mL interleukin (IL)-7, and 5 ng/mL IL-15 at 37°C with 5% CO2 for 10 min. In parallel, autologous PBMC feeder cells (10–15×106) were irradiated at 25 gray and mixed with TexMACs medium supplemented with 5% FBS, 10 ng/mL IL-7, and 5 ng/mL IL-15. The feeder cells and electroporated cells were combined in 48-well plates and further incubated for 14 days, with media changes on day 4 and day 7 using fresh TexMACs medium supplemented with 5% FBS and 10 ng/mL IL-7 and 5 ng/mL IL-15. On day 14, the efficiency of CD19 and CD20 CAR was analyzed by incubating the cells with specific ligands, and flow cytometry was used to detect the expression levels (figure 1B).

In vitro cytotoxic assay

Cytotoxicity was assessed on day 14 post-electroporation. CAR T cells were co-cultured with Nalm-6 (CD19+), Raji (CD19+CD20+), CD19-K562, and CD20-K562 cancer cells at different effector to target (E:T) ratios (0.5:1, 1:1, 2:1 and 5:1) in RPMI media with 10% FBS in V bottom 96-well plates without cytokines for 24 and 48 hours. Flow cytometry analysis was performed to analyze the per cent killing. The cell cultures from different E:T ratios were stained with antibodies specific to CD3-Alexa Fluor 700, CD19-BV510 and Zombie NIR dye (#B331983). To analyze the per cent killing, we first gated Zombie NIR negative population and this population was gated with CD3+ (Y axis) and CD19+ (X axis) to separate viable CAR T cells and residual target cells. Whereas, for CD19-K562 and CD20-K562 we gated CD3+ (Y axis) and dTomato (X axis) in viable cells population. The per cent killing was then analyzed by comparing the total cells in the target alone.

To mimic the prolonged and repetitive cytotoxic activity of single anti-CD19 and dual anti-CD19/20 CAR T cells required for in vivo activity against B-cell lymphomas, a longer time frame cytotoxicity assay was set-up using 106 CD19-K562 and 106 CD20-K562 with 106 CAR T cells were co-cultured in G-Rex24-well plates (Wilson Wolf, USA) with RPMI media with 10% FBS and without cytokines. The results were analyzed on day 4 and day 8. Total cells were harvested, and cytotoxicity was analyzed by staining with 7-aminoactinomycin-D (7AAD) or 4’,6-diamidino-2-phenylindole to distinguish between dead and live cells. The remaining live cells were stained with target-specific and T cell (CD3)-specific antibodies. The absolute number of effector and target cells was determined using flow cytometry, and per cent killing was calculated using the following formula:

Cytokine release assay

A total of 105 CAR T cells were co-cultured with 105 Raji cells at a 1:1 ratio for 24 hours in a 96-well plate. The levels of IL-2, interferon-gamma (IFN-γ), tumor necrosis factor-alpha (TNF-α), and cytokines were analyzed by flow cytometry using the BDTM cytokine bead array Human Th1/Th2 Cytokine Kit II (BD Bioscience, USA). The data were analyzed by FCAP Array V.4 software (BD Bioscience, USA).

Flow cytometric analysis

CAR T cells (2×105 cells per test) were labeled with antibodies at 4°C for 15 min. Flow cytometry was performed using a MACSQuant Analyser 10 Flow Cytometer (Miltenyi Biotec, Bergisch Gladbach, Germany), and data were analyzed by FlowJo V.10.7.1 software (FlowJo). Cells were subjected to SSc-FSc and gated as the CD3+ T-cell population. For transfection efficiency: CD19 CAR detection reagent, anti-biotin PE, and CD3-APC (Miltenyi Biotec, Bergisch Gladbach, Germany); T-cell phenotype: CD3-PerCP (UCHT1), CD8-PE (SK1), (BD Bioscience, New Jersey, USA), CD4-APC (BioLegend, San Diego, California, USA); memory phenotype: CD3-PerCP (UCHT1), CD45RO-VioGreen (UCHL1), and CD62L-VioBlue (DREG-56) (BD Bioscience, New Jersey, USA); exhaustion: CD3-PerCP, PD1(CD279)-APC, TIM-3-PE, LAG3-PE, and TIGIT-APC (BioLegend, San Diego, California, USA); activation: CD3-PerCP, CD25-PE, and CD69-APC (BioLegend, San Diego, California, USA); and for cytolytic activity: CD3-PE (OKT3), CD19-APC (HIB19) (BioLegend, San Diego, California, USA), and 7AAD-PerCP (BD Bioscience, New Jersey, USA).

Fluorescence microscopy

Bright field and dTomato-expressing cell images were captured using an Olympus Model IX-81 inverted fluorescence microscope.

In vivo antitumor effect of CAR T cells

NOD-scid IL2Rgammanull (NSG) mice were obtained from Jackson Laboratory (Bar Harbor, Maine, USA) and bred under specific pathogen-free conditions at the University of Otago Animal Research Center. The Animal Ethics Committee from the University of Otago Animal Research Center approved the animal studies under the Animal Use protocol AUP-23–11. A small sample size was chosen to evaluate the efficacy of bidirectional CAR designs in an in vivo setting for the first time. Initial intentions were to gather early evidence of bidirectional functionality prior to potential clinical applications.

Age-matched female NSG mice (n=6 per treatment) were subcutaneously injected in the ventral lateral flank with 1×106 Raji-luciferase cells following a standard operating protocol (SOP NO: AWO 002) from the University of Otago Animal Welfare Office. Six days following tumor administration, the mice were randomly assigned to treatment groups consisting of phosphate-buffered saline (untreated) and 5×106 EF1.CAR19 CAR T cells or CAR20.GFP.EF1.CAR19 CAR T cells using healthy donor PBMCs. Treatments were administered by intravenous injection of the lateral tail vein (SOP NO: AWO 004).

Tumor burden was measured using digital calipers every second day (volume calculated using the formula V=0.5×length×width2). Tumors were additionally measured by live imaging using the PerkinElmer IVIS X5 system, and mice were administered 150 mg/kg D-luciferin potassium salt (LUCK-1GR) by intraperitoneal injection (SOP NO: AWO 003). Mice were anesthetized by isoflurane induction until effect, followed by a 12 min incubation, after which five images were taken with 1 min delays. This procedure was performed on day 5 post tumor and repeated every week until the end of the study. Images were analyzed using Living Image, V.4.1, software (PerkinElmer), and the bioluminescent signal flux for each mouse was expressed as a total flux of photons/second.

Data analysis and statistics

DNA sequences were analyzed by Geneious Prime and SnapGene software. Schematic illustrations were made by using BioRender. All the in vitro experiments were performed with at least three independent healthy donors (N=3), and the data are presented as the means±SEMs or means±SDs. Representative flow cytometry plots are from one donor. Student’s t-test, one-way and two-way analysis of variance (ANOVA) were used to determine statistical significance.

The data from in vivo experiments were expressed as the mean±SD of the total flux of 3–6 mice (N=6) following the loss of >3 mice data excluded from analysis to maintain valid statistical comparison. Statistical differences were determined by two-way ANOVA and by multiple t-tests. Mice were removed from the study when tumors reached over 1000 mm³ or exceeded 15 mm in any dimension. The Kaplan-Meier survival curve was plotted at the end of the study. Statistical analysis was performed using GraphPad Prism V.8 (GraphPad Software). P values<0.05 were considered statistically significant and are represented as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

ResultsExpression of dual CD19/CD20 CAR T cells using different bidirectional eukaryotic promoters

Eukaryotic promoters EF-1α, hPGK, RPBSA and CMV, previously identified as having bidirectional activity in driving reporter genes,26 27 were tested for their ability to drive the expression of two CAR cassettes.

On day 14 post electroporation, the EF-1α promoter showed stronger expression of CD19 and CD20 CAR in both directions, 78.0%±5.4 CD19-CAR expression from the forward direction and 71.8%±8.4 CD20-CAR expression from the reverse direction, among all bidirectional promoters. Surprisingly, the hPGK and RPBSA promoters exhibited stronger CAR expression from the reverse direction (CD20 CAR: 84.1%±5.5 and 65.4%±11.5) but minimal CAR expression from the forward direction (CD19-CAR: 29.9%±5.8 and 21.2%±8.5). In addition, the CMV promoter showed weak expression of CARs in either direction. The EF-1α promoter also exhibited a stronger double CAR-positive (CD19+CD20+) cell population than hPGK, RPBSA, and CMV (59.9%±11.7 vs 18.8%±8.8, 18.4%±9.4 and 0.13%±0.11, respectively) (figure 1C–F).

As shown in figure 1G, the EF-1α promoter displayed the highest CAR19 mean fluorescent intensity (MFI) compared with hPGK, RPBSA, and CMV (7476.3±1030.5 vs 606.7±158.7, 435.3±169.8, and 120.3±26.8, respectively). However, all three promoters except CMV were able to express variable levels of CAR20 MFI on CD3+ T cells (EF-1α: 3943±1471.5; hPGK: 9552.0±1981.1; RPBSA: 2960±771.5 and CMV: 313.7±18.8).

Bidirectional promoter-driven expansion and phenotypic differentiation of CD19 and CD20 CAR T cells

We started with 4×106 PBMCs on day 0 for SB transposition. On day 14, four CAR constructs showed a significant level of cell expansion (EF-1α: 9.9±1.3×106, hPGK: 10.1±0.88×106, RPBSA: 5.9±0.5×106 and CMV: 7.2±1.1×106) (figure 2A). There was no statistically significant difference in fold expansion between each promoter used. Next, we evaluated the T-cell subset phenotype, and no statistically significant differences were found in the CD4+ to CD8+ T-cell ratio for either of the four CAR constructs (figure 2B). The memory phenotype was determined by the expression of CD45RO and CD62L, as represented by figure 2C. Compared with untransduced constructs, the CAR constructs differed in their degree of apparent central memory T-cell phenotype differentiation (EF-1α>hPGK>RPBSA>CMV), but no significant differences were observed between naïve, effector memory and terminal effector T-cell phenotypes (figure 2D).

Figure 2

Bidirectional CD19/20 CAR T-cell expansion and phenotype. (A) Expansion of total cell numbers. Cell numbers were determined by trypan blue exclusion assay on day 0, day 7 and day 14 after cell transfection. (B) Mean percentages of helper CD3+/CD4+ T cells and cytotoxic CD3+/CD8+ T cells on day 14 after transfection. (C) Flow cytometric representation of CD3+ T-cell memory phenotypes. (D) Memory phenotype of CD3+T cells on day 14 after transfection showing the percentage of naïve (TN), central memory (TCM), effector memory (TEM), and terminal effector (TE) T cells. All results are summarized as the mean±SEM from at least three independent healthy blood donors. Significant differences were determined by Student’s t-tests and two-way analysis of variance. Asterisks indicate significant p values as follows: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

The EF-1α bidirectional CAR construct demonstrates enhanced cytotoxicity

To determine the cytotoxic potential of different CAR constructs featuring bidirectional promoters, we employed Nalm-6, Raji, CD19-K562, and CD20-K562 cell lines. Prior to conducting the cytotoxic assay, we confirmed the expression of CD19 and CD20 markers in the cell lines (online supplemental figure 1A–C). The CAR construct using a CMV promoter was excluded from the study due to its failure to express CAR in any orientation. As shown in figure 3A–D, EF-1α bearing CAR T cells, which expressed the highest levels of both CD19 and CD20 CAR, also demonstrated enhanced cytotoxicity against the Nalm-6, Raji, CD19-K562, and CD20-K562 cell lines compared with the CAR constructs driven by other promoters, even at a low E:T ratio (online supplemental figure 2A–D). In addition, the expression of CD19/CD20 B-cell marker on the surface of leukemia (Nalm-6), lymphoma (Raji) and overexpressed (CD19-K562 and CD20-K562) cells were analyzed by flow cytometry after overnight (24 hours) co-culture (online supplemental figure 3A). The CD19/CD20 markers were rapidly downmodulated from the target cells that were co-cultured with EF-1α bearing CAR T cells even at a low E:T ratio however hPGK and RPBSA promoter bearing CAR T cells also significantly downmodulate CD19/CD20 in high E:T ratios (online supplemental figure 3B–E).

Figure 3

Bidirectional CD19/20 CAR T-cell cytotoxic function in vitro. Flow cytometry-based cytotoxicity assay were performed using (A) Nalm-6, (B) Raji, (C) CD19-K562 and (D) CD20-K562. CAR T cells and target cells were co-cultured overnight at different E:T ratios, X-axis. All results are summarized as the mean±SEM from at least three independent healthy blood donors. Significant differences were determined by two-way analysis of variance. Asterisks indicate significant p values as follows: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. CAR, chimeric antigen receptor; E:T, effector to target.

The EF-1α promoter drove stronger transcription of two transgenes using a bidirectional orientation, compared with a unidirectional orientation

Among the four promoters bearing dual CD19/CD20 CAR constructs, we chose the CAR construct with the EF-1α promoter (figure 4A) because of its strongest bidirectional activity. In our next experiment, we compared EF-1α promoter transcriptional activity between bidirectional and unidirectional orientations for dual CAR expression.

Figure 4

Comparisons of CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19 CAR T-cell efficiencies. (A) Schematic illustration of CAR20.GFP.EF1.CAR19 (B) Schematic illustration of EF1.GFP.CAR19.CAR20 (C) Schematic illustration of EF1.CAR19. (D) Flow cytometry plot representing CAR19 expression from CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19-transfected CD3+ T cells. (E) Flow cytometry plot representing GFP expression from CAR20.GFP.EF1.CAR19, and EF1.GFP.CAR19.CAR20-transfected CD3+ T cells. (F) Flow cytometry plot representing double CAR19+ and GFP+ cells. (G) Transfection efficiency of CAR19−, GFP− and CAR19+GFP+ cells on day 14 after transfection. (H) MFI of CAR19-positive and GFP-positive cells on day 14 after transfection. (I) Transfection efficiency of CAR19−, GFP− and CAR19+GFP+ cells on day 30 after transfection. (J) MFI of CAR19-positive and GFP-positive cells on day 30 after transfection. All results are summarized as the mean±SEM from at least three independent healthy blood donors. Significant differences were determined by Student’s t-test and two-way analysis of variance. Asterisks indicate significant p values as follows: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. CAR, chimeric antigen receptor; GFP, green fluorescent protein; MFI, mean fluorescent intensity; P2A, 2A self-cleaving peptide.

We next compared the bidirectional EF-1α CAR construct (GFP.CAR20.EF1.CAR19) with the EF-1α promoter driving the expression of both FMC63 CD19 CAR and 1F5 CD20 CAR in a sense direction, but with dual expression enabled by a P2A site (EF1.GFP.CAR19.CAR20) (figure 4B). As a control for this experiment, we generated an EF-1α-driven construct comprising only the FMC63 CD19 CAR (EF1.CAR19) (figure 4C).

The results showed that although GFP expression was similar from the forward or reverse position of EF-1α in both unidirectional and bidirectional CARs, surprisingly, CD19 CAR expression was much lower in the unidirectional CAR than in the bidirectional CD19 CAR. It was also noted that the expression level of CD19 CAR in the bidirectional construct was comparable to that of a single CD19 CAR. This indicates that higher expression of both transgenes occurs when placed in a bidirectional orientation of the EF-1α promoter (figure 4D–H). Remarkably, CAR19 and GFP expression was observed until day 30 post-transfection in CAR20.GFP.EF1.CAR19 CAR T cells (figure 4I,J).

Expansion and phenotypic differentiation of EF-1α-driven dual CAR T cells

On day 14, all three EF-1α-driven dual CAR constructs showed significant levels of cell expansion compared with day 0 (online supplemental figure 4A). A significantly higher per cent of CD8+ T cells compared with CD4+ T cells was observed on day 14 in all three CAR constructs, as compared with untransduced (online supplemental figure 4B). In addition, CAR20.GFP.EF1.CAR19 CAR displayed a relatively higher central memory T-cell expansion as compared with the EF1.GFP.CAR19.CAR20 CAR (63.7%±9.1 vs 43.4%±6.5), while a 54.8%±8.3 central memory T-cell phenotype was observed in EF1.CAR19-transfected cells (online supplemental figure 4C, D).

CAR20.GFP.EF1.CAR19 CAR T cells have superior cytotoxicity to EF1. GFP CAR19 CAR20 CAR T cells in vitro

To evaluate antigen-specific cytotoxicity linked to CAR19 and CAR20 expression, K562 cells were engineered to overexpress CD19 and CD20 antigens (figure 5A). Parental K562 cells were used as a negative control, and CD19+/CD20+ Raji cells were used as a positive control.

Figure 5

Comparisons of the cytolytic activity of CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19 CAR T cells. (A) Fluorescence microscopy of K562 cells expressing dTomato. (B) Flow cytometric representation of CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19 CAR T cells co-cultured with Raji cells. (C) Per cent killing of CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19 CAR T cells to Raji cell line after 24 and 48 hours of co-culture. (D) Flow cytometric representation of CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19 CAR T cells co-cultured with CD19-K562 cells. (E) Flow cytometric representation of CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1α. CAR19 CAR T cells co-cultured with CD20-K562 cells. (F) Per cent killing of CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19 CAR T cells to CD19-K562, CD20-K562 and only K562 after 48 hours of co-culture. (G) Flow cytometric representation of CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19 CAR T cells co-cultured with mixed cultures of CD19-K562 and CD20-K562, CD19-K562 and CD20-K562 cells possessed dTomato; however, CD19-K562 cells were stained with anti-CD19 antibody, cells positive for anti-CD19 antibody and dTomato were CD19-K562, and only dTomato-positive cells were CD20-K562 (as represented in the template). All results are summarized as the mean±SE from at least three independent healthy blood donors. Significant differences were determined by one-way analysis of variance, two-way analysis of variance and Student’s t-test, and asterisks indicate significant p values as follows: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. CAR, chimeric antigen receptor; GFP, green fluorescent protein.

As shown in figure 5B,C, CAR20.GFP.EF1.CAR19 and EF1.CAR19 CAR T cells effectively killed Raji cells after 24 and 48 hours of co-culture compared with EF1.GFP.CAR19.CAR20 CAR T cells. Similarly, as demonstrated in figure 5D–F, after 48 hours of co-culture, CAR20.GFP.EF1.CAR19 and EF1.CAR19 CAR T cells killed 78.9%±5.5 and 76.38%±16.9 CD19-overexpressing K562 cells. In addition, only CAR20.GFP.EF1.CAR19 CAR T cells can kill 83.8%±2.1% of CD20-overexpressing K562 cells after 48 hours of co-culture. However, a very low baseline killing effect was observed in parental K562 cells (online supplemental figure 5A).

Furthermore, mixed co-culture results were analyzed on day 4 and day 8. As shown in figure 5G, CAR20.GFP.EF1.CAR19 CAR T cells were able to lyse both CD19-K562 and CD20-K562 cells, whereas EF1.CAR19 CAR T cells killed CD19-K562 cells but not CD20-K562 cells, as highlighted by the overgrowth of CD20-K562 cells. EF1.GFP.CAR19.CAR20 CAR T cells did not markedly lyse either CD19-K562 or CD20-K562 cells until day 8 of co-culture (figure 5G and online supplemental figure 5B, C). On day 8, a combination of CD19-K562 and CD20-K562 expanded effector cells from CAR20.GFP.EF1.CAR19 and EF1.CAR19 CAR T-cell group (online supplemental figure 5D). These data indicate the killing of CAR20.GFP.EF1.CAR19 and EF1.CAR19 CAR T cells were antigen-specific, as evidenced by the co-culture results from CD19-K562, CD20-K562, and mixed CD19/CD20-K562 cells.

The overall finding also represents CAR20.GFP.EF1.CAR19 CAR can express functional CARs in both directions when placed in head-to-head orientation from the EF-1α promoter. However, the function of the EF-1α promoter is limited when transcribing long RNA transcripts in a unidirectional orientation in the SB vector backbone.

CAR20.GFP.EF1.CAR19 CAR T cells exhibit a superior cytokine profile and lack inhibitory molecule marker expression in vitro

To determine the antigenic response of the dual CAR T cells, 105 CAR T cells were co-cultured with 105 Raji cells for 24 hours, and the cultured supernatants were analyzed for cytokine levels. As shown in figure 6A, CAR20.GFP.EF1.CAR19 and EF1.CAR19 CAR T cells cultured with Raji cells exhibited significant levels of IL-2 (2953.3±395.0 pg/mL and 2152.33±563.3 pg/mL), IFN-γ (3818.3±387.3 pg/mL and 3613.9±553.1 pg/mL) and TNF-α (339.46±59.3 pg/mL and 323.9±125.4 pg/mL). Additionally, very low cytokines were released in the culture conditions of untransduced or EF1. GFP.CAR19. CAR20 CAR T cells, which is below limit as shown in figure 6A.

Figure 6

Cytokine production, activation and exhaustion markers of CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19 CAR T cells in response to Raji cells. (A) Secretion levels of the cytokines IL-2, IFN-γ and TNF-α from CAR20.GFP.EF1.CAR19, EF1.GFP.CAR19.CAR20 and EF1.CAR19 CAR T cells. (B) Expression of the early activation marker CD69 in CD3+ T cells before and after co-culture. (C) Expression of the late activation marker CD25 in CD3+ T cells before and after co-culture. (D) Expression of TIM-3 in CD3+ T cells before 24 hours of co-culture. (E) LAG-3 expression in CD3+T cells before 24 hours of co-culture. (F) PD-1 expression in CD3+ T cells before 24 hours of co-culture. (G) TIGIT expression in CD3+ T cells before 24 hours of co-culture. All results are summarized as the mean±SEM from at least three independent healthy blood donors. Significant differences were determined by Student’s t-test or two-way analysis of variance, and asterisks indicate significant p values as follows: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. CAR, chimeric antigen receptor; GFP, green fluorescent protein; IFN, interferon; IL, interleukin; PD-1, Programmed cell death protein 1; LAG-3, Lymphocyte activation gene-3; TIGIT, T cell immunoreceptor with Ig and ITIM domains; TIM-3, T cell immunoglobulin and mucin domain-containing protein 3; TNF, tumor necrosis factor.

To further explore antigen-mediated activation of CAR T cells, we determined the early and late activation markers CD69 and CD25, respectively. CD69 is a very early activation marker induced following T-cell receptor engagement.28 As shown in figure 6B, CD69 was expressed in CAR20.GFP.EF1.CAR19 and EF1.CAR19 CAR T cells following Raji cell stimulation. As shown in figure 6C, Raji stimulation of CAR20.GFP.EF1.CAR19 and EF1.CAR19 increased CD25/IL-2Rα-chain expression. Neither CD69 nor CD25 markers were elevated in untransduced or EF1.GFP.CAR19.CAR20 CAR T cells after exposure to Raji cells (figure 6B,C). Moreover, exhaustion markers were not significantly expressed in any CAR-T cells when compared with untransduced T cells (figure 6D–G).

CAR20.GFP.EF1.CAR19 shows a significant tumor survival advantage compared with EF1.CAR19 in an advanced Raji-lymphoma model

Further comparison between the CAR20.GFP.EF1.CAR19 CAR design and traditional EF1.CAR19 CAR T cells were performed in vivo using a high-burden Raji NSG model. Mice were implanted with luciferase-expressing Raji cells and monitored over the course of 40 days. Bioluminescent imaging was performed on day 5 to assess the viable engraftment of Raji cells (figure 7A,B). CARs were transfected by nucleofection into healthy donor PBMCs in the same manner as in vitro functional testing. CAR T cells were passaged for 14 days and assessed for CAR surface expression and memory phenotypes 24 hours prior to intravenous administration (online supplemental figure 6A–C). Consistent with the in vitro results, the transfection efficiency of CAR20.GFP.EF1.CAR19 CARs were 88% and 81%, CAR19 (sense direction) and GFP.CD20 CAR (antisense direction), respectively. EF1.CAR19 CAR T cells have 90% CAR19 expression.

Figure 7

In vivo activity of CAR T constructs. (A) Schematic diagram of the in vivo experiment. (B) Biolumi