Minicircle DNA-Engineered CAR T Cells Suppressed Tumor Growth in Mice

Recently, the chimeric antigen receptor-engineered T (CAR T; refs. 1, 2) cell therapy had successfully cured a few hematologic cancers. Now, hundreds of new CAR T therapies targeting blood and solid malignancies are being developed. However, current clinical-grade CAR T cells are manufactured mainly using retroviral or lentiviral vectors which have three problems: (i) potential insertional mutagenesis and oncogenesis (3); (ii) risks of generating replication-competent viruses (4); and (iii) high costs associated with technical challenges in scaling up the viral production (5).

On the contrary, nonviral methods, such as transposon and plasmid DNA, are not yet widely used in producing CAR T cells. Transposons carrying a CAR cassette (piggyBac or Sleeping Beauty) can integrate into the patients' genome, and therefore a long-lasting CAR expression can be achieved (6–8). However, the transposon-based CAR T production is complex and time-consuming, usually requiring purification and cotransfection of two plasmids, as well as a coculture with feeder artificial antigen-presenting cells or irradiated peripheral blood mononuclear cells (6–8). In addition, plasmid DNA (pDNA) is not suitable for producing therapeutic CAR T cells, because pDNA has a transient in vivo transgene expression due to transcriptional silencing by the bacterial DNA backbone (9).

To overcome this challenge, the minicircle DNA (mcDNA) was developed—two recombination sites, attB and attP, were added to flank the bacterial DNA backbone so that it can be removed through ΦC31 integrase-mediated recombination (9). The mcDNA vector, devoid of nearly all bacterial DNA, has significantly increased in vivo transgene expression levels (10–200 folds more) and in vivo expression duration (from 4 to 7 weeks; ref. 9). To date, mcDNA has been used to engineer human pluripotent stem cells (10).

In this study, we applied the mcDNA system to generate CAR T cells that specifically target the prostate stem cell antigen (PSCA) molecule (mcDNA-PSCA-CAR T cells). Our results demonstrated that mcDNA-PSCA-CAR can be easily transfected and highly expressed in human T cells. We then performed in vitro and in vivo assays and showed that mcDNA-PSCA-CAR T cells can specifically and effectively attack PSCA-positive cancer cells.

All animal experiments were conducted under the approval of Hebei Medical University Animal Care and Use Committee, Hebei, China. All NOD/SCID mice used in this study were healthy males (6–8 weeks of age, weighing 200–250 g), which were randomly assigned to experimental or control groups.

Human prostate cancer cell line RT4 and human bladder carcinoma cell line PC-3M were purchased from and authenticated by the ATCC. RT4 cells were cultured in McCoy 5A medium (ATCC) supplemented with 10% FBS (Gibco), whereas PC-3M cells were cultured in RPMI1640 medium (Gibco) supplemented with 10% FBS. These cells were used when they were approximately passages 18 to 30. Cells were maintained at 37°C in a humidified atmosphere containing 5% CO₂. RT4 and PC-3M cells were stained with a phycoerythrin (PE)-labeled mouse anti-PSCA IgG1 (Santa Cruz Biotechnology) and analyzed by FACS for PSCA expression on the cell surface. For Mycoplasma testing of these cells, both agar and broth methods were applied, and both cell lines were Mycoplasma-negative up until March 20, 2019.

The amino acid sequence of an anti-PSCA antibody's single-chain fragment variable (scFv) was previously published by Olafsen and colleagues (11), and its nucleotide sequence obtained through reverse genetics (see Results). The third-generation PSCA-CAR expression construct (containing ≥2 costimulatory domains) was generated by fusing the anti-PSCA scFv with the signaling domains of CD28, CD137, and CD3ζ. This PSCA-CAR construct (1,575 bp) was cloned into the pUC57 vector (4252 bp) and then cloned into the MC-Easy (SBI) parental minicircle plasmid MN511A-1, which contains a GFP cassette (7,064 bp) to form the parental minicircle PSCA-CAR plasmid (pMC-Easy-PSCA-CAR, 8651 bp).

The bacterial E. coli strain ZYCY10P3S2T (SBI) was used to generate the minicircle PSCA-CAR (mcDNA-PSCA-CAR). Transformed bacteria carrying pMC-Easy-PSCA-CAR was cultured with LB medium containing 50 μg/mL kanamycin, with a rotation speed of 250 rpm at 30°C overnight. To assist the generation of mcDNA-PSCA-CAR, the expression of ΦC31 integrase (recombinase) and SceI endonuclease (to degrade the bacterial backbone) was induced by adding l-arabinose at an OD₆₀₀ from 4 to 6. Minicircles (mcDNA-PSCA-CAR) were produced from the parental minicircle plasmid (pMC-Easy-PSCA-CAR) through ΦC31-mediated recombination between the ΦC321 attB and attP sites on the parental plasmid (pMC-Easy-PSCA-CAR), resulting in separation of the parental minicircle plasmid into the minicircle DNA (mcDNA-PSCA-CAR) and the parental bacterial backbone. In addition, l-arabinose also induced SceI endonuclease that degraded the bacterial backbone. After an additional 5 hours, bacteria cells were harvested and mcDNA-PSCA-CAR purified using QIAGEN's EndoFree Plasmid Maxi Kit. For every 400 mL overnight culture, one 2500 column and 100 mL of buffers P1, P2, and P3 were used to ensure complete lysis and high yield of mcDNA-PSCA-CAR. The Maxi-prepared mcDNA-PSCA-CAR was then confirmed with restriction analysis.

Peripheral blood samples were obtained from healthy donors at Hebei Blood Center, Hebei, China. Written informed consent was obtained from all donors; the studies were conducted in accordance with the World Medical Association Declaration of Helsinki and approved by the Medical Research Ethics Committee of Hebei Medical University, Hebei, China. Peripheral blood mononuclear cells (PBMC) were isolated from whole blood samples by Ficoll-Paque density gradient centrifugation (GE Healthcare). PBMCs from the interphase were collected, washed with complete medium for 3 times, and then cultured in RPMI-1640 complete medium supplemented with 1,000 U/L IFNγ (PeproTech) at 37°C with 5% CO₂ overnight. The PBMCs were incubated with 1 μg/mL CD3/OKT-3 (Biogems) and 1 μg/mL CD28 (Biogems) for 1 day to activate the T cells. The activated T cells were then cultured with 500 U/mL recombinant human IL2 (PeproTech) and 10 U/mL recombinant human IL15 (PeproTech) to expand for 3 to 7 days.

Prepared T cells were propagated in RPMI1640 medium supplemented with 10% FBS, glutamine, 100 U/mL penicillin–streptomycin, and 500 U/mL recombinant human IL2. The purity and density of the T-cell population were determined by flow cytometry using conjugated mAbs specifically targeting CD3, CD4, and CD8. Generation of PSCA-CAR T cells was achieved by electroporation using a 4D-Nucleofector system (Lonza): 5 × 10⁶ of activated T cells were electroporated using program EO115 with 2.5 μg of pmaxGFP electroporation control (Lonza) or equimolar amount of mock mcDNA control plasmid MN601A-1 (SBI), or mcDNA-PSCA-CAR in 100 μL of P3 Primary Cell Buffer according to the manufacturer (Lonza). After electroporation, cells were resuspended in 2 mL of prewarmed RPMI1640 medium containing 10% FBS and incubated in anti-CD3/CD28-mAb–coated culture plates at 37°C in 5% CO₂ to reactivate cells. Culture medium was half replaced by fresh complete medium containing IL2 (500 U/mL) every 2 to 3 days. Twenty-four hours after electroporation, the transfection efficiency was evaluated by the fluorescence intensity and the viability of transfected T cells was verified using FACS according to the light scattering properties. The generated CAR T cells were then phenotypically analyzed by flow cytometry for T-cell markers CD3, CD4, and CD8, and cell surface PSCA-CAR expression was monitored using protein L-labeling.

For flow cytometry analysis, BD FACS Aria II (BD Biosciences) was used. T cells were harvested 24 hours after transfection and stained with different mouse anti-human antibodies labeled with various fluorescent tags for 30 minutes at 4°C in the dark. The following conjugated antibodies were used: CD3-PE or CD3-FITC (BD Biosciences), CD4-PE (BD Biosciences), and CD8-FITC (BD Biosciences). Biotinylated protein L (GenScript) was used to detect cell surface PSCA-CAR expression. For FACS staining, 1 × 10⁶ T cells were harvested and placed in a 5 mL tube and washed 3 times with 3 mL of ice-cold PBS containing 4% BSA as the Wash Buffer. After washing, cells were resuspended in 200 μL of ice-cold Wash Buffer and incubated with 1 μg of protein L at 4°C for 45 minutes. Cells were washed with 3 mL of the ice-cold Wash Buffer 3 times, and then incubated in the dark with 10 μL of PE-conjugated streptavidin (SA-PE, BD Biosciences) at 4°C for 30 minutes. After washing for additional two times, cells were examined for cell surface PSCA-CAR expression by FACS. All FACS data were analyzed using FlowJo 8.1.1 software (FlowJo, LLC).

Normal T, mock T, or PSCA-CAR T cells (10⁶ cells) were cultured for 10 days at 37°C in 5% CO₂. Cells were cultured with recombinant human IL2 (500 U/mL) to proliferate and recombinant human IL15 (10 U/mL) to promote survival. The number of cells was calculated every other day, and the proliferation curve was generated.

Normal T, mock T, or PSCA-CAR T cells were cocultured with PC-3M or RT4 cells in increasing effector-to-target cell ratios (2:1, 5:1, 10:1, 20:1) in flat-bottomed 96-well plates. Cocultures were harvested 24 hours later and analyzed using Cell Counting Kit-8 according to the manufacturer (DOJINDO).

Normal T, mock T, or PSCA-CAR T cells (2 × 10⁶ cells) were cocultured with PC-3M or RT4 cells in 96-well plates (effector-to-target cell ratio, 10:1). Supernatants were collected after coculturing these cells for 24 hours. ELISA (eBioscience) was applied to analyze IFNγ and IL2 secretion.

To reconstruct the humanized immune system in the NOD/SCID mice, mice were injected with 4 × 10⁷ PBMCs for 4 weeks. Four weeks later, NOD/SCID mice with the humanized immune system were subcutaneously inoculated with 3 × 10⁶ PC-3M cells to develop the tumor xenografts. Approximately 15 days later, when the tumor volume reached 250 mm³, mice received either 1 × 10⁷ normal T cells or PSCA-CAR T cells through tail vein on days 1 and 8 (8 mice per group). For the NS control group, mice were treated with 0.5 mL saline. The tumor size was measured using a Vernier caliper every other day, and the tumor volume was calculated as: volume = (length × width²)/2. Mice were sacrificed 4 weeks after normal T or PSCA-CAR T-cell treatment. The PSCA CAR T cells in mouse peripheral blood were detected by FACS as described above. ELISA (eBioscience) was used to detect IFNγ and IL2 secretion in mouse blood serum. To examine the pathology and T-cell infiltration, tumor samples were fixed and stained with an anti-PSCA antibody (Abcam) and an anti-CD3 antibody (Abcam) using IHC. These experiments were independently repeated three times.

All results are expressed as mean ± SEM. Statistical significance was assessed using ANOVA. A value of P < 0.05 was considered statistically significant.

To generate PSCA-CAR T cells using mcDNA, we designed and generated the minicircle construct, mcDNA-PSCA-CAR. We first used reverse genetics to generate the nucleotide sequence of the anti-PSCA scFv (Fig. 1A) from the published antibody amino acid sequence (11). The immunoglobulin kappa light chain (IgG kappa) as the leader sequence was added to the N-terminus of the PSCA scFv domain for efficient cell surface expression of the chimeric receptor (Fig. 1A). Intracellular domains of CD28, CD137, and CD3ζ were included to enhance costimulatory signaling, and the CD8 Hinge and transmembrane domains (CD8Hinge, CD8TM) were used to link the extracellular and intracellular regions (Fig. 1A). To insert the PSCA-CAR construct into the parental plasmid, we added EcoRI and BamHI restriction sites to its 5′- and 3′-ends, respectively (Fig. 1A).

We directionally cloned the construct into pUC57 and then into the MC-Easy MN511A-1 plasmid which contains a GFP cassette (SBI) to generate the parental minicircle plasmid pMC-Easy-PSCA-CAR (Fig. 1B, from left to middle). Eight pMC-Easy-PSCA-CAR bacterial clones (lanes 1–8) were selected for restriction and sequencing analysis, and nearly all clones contained the PSCA-CAR construct (1,575 bp) except clone 2 (lane 2, Fig. 1C). Next, we cultured E. coli (strain ZYCY10P3S2T) transformed with pMC-Easy-PSCA-CAR and added L-arabinose to induce the expression of ΦC31 integrase and SceI endonuclease. The ΦC31 integrase mediated a recombination event between attB and attP sites (Fig. 1B, middle) and separated the pMC-Easy-PSCA-CAR into mcDNA-PSCA-CAR and the bacterial backbone which was degraded by SceI endonuclease (Fig. 1B, from middle to right). As shown in Fig. 1D, after L-arabinose induction, the minicircle (mcDNA-PSCA-CAR) was successfully generated without the presence of the parental minicircle plasmid (Fig. 1D, both plasmid and mcDNA are in the undigested, closed circular form).

After large-scale preparation and purification, we transfected mcDNA-PSCA-CAR and the control plasmids into human T cells via electroporation (normal T: control cells without electroporation; pmaxGFP transfected T: electroporation control cells, transfected with the control plasmid pmaxGFP; mock T: mcDNA control cells, electroporated with the mcDNA control plasmid MN601A-1, which contains a GFP cassette; PSCA-CAR T: cells electroporated with mcDNA-PSCA-CAR, which contains a GFP cassette) using a 4D-Nucleofector system (Lonza). Twenty-four to 48 hours after transfection, obvious green fluorescence can be observed in transfected T cells (Fig. 2A).

Twenty-four hours after electroporation, we performed FACS analysis for the four different treatments (Fig. 2B). All the transfection experiments were performed with cells obtained from three donors, and representative FACS result from one donor is shown (Fig. 2B). Our FACS results showed that the transfection efficiency of mcDNA-PSCA-CAR (PSCA-CAR T cells, 58.66%) was similar to the minicircle control (mock T cells, 56.72%) but was slightly higher than cells electroporated with plasmid DNA (pmaxGFP transfected T cells, 48.24%). As a control, cells without electroporation had no GFP signal (normal T cells, 0.44%). Statistical analysis of transfected cells from three different donors is shown in Fig. 2B (right). In addition, the electroporation inflicted moderate damage to transfected cells (Fig. 2C) as the viability of the transfected T cells (PSCA-CAR T cells, 60.2%) was lower than that of nontransfected cells (normal T cells, 96.8%), and the statistics is shown in Fig. 2C (right). However, electroporation had no effect on the proliferation of the transfected T cells as described below (Fig. 2E).

Protein L, a bacterial surface protein that can specifically bind to the IgG kappa and PSCA scFv domains (Fig. 1A; refs. 12, 13), was used to detect the expression of PSCA-CAR on the cell surface; the PSCA-CAR–bound protein L was then stained with a secondary PE-SA antibody and analyzed by FACS. A majority (88.96%) of PSCA-CAR T cells expressed PSCA-CAR on the cell surface, whereas normal T and mock T cells had nearly no expression (0.10% and 0.11%, respectively; Fig. 2D) with the statistical analysis from three donors shown in Fig. 2D. To examine whether electroporation was harmful to the proliferation of transfected cells, we performed cell proliferation assays 3, 5, 7, and 10 days after electroporation into cells from three different donors. As is shown in Fig. 2E, mock T and PSCA-CAR T cells proliferated normally as compared with nonelectroporated normal T cells. Finally, electroporation and the expression of PSCA-CAR did not affect the phenotypes of T cells as the expression patterns of CD3, CD4, and CD8 remained unchanged (Fig. 2F), and the statistical analysis of transfected cells from three donors is shown on the right.

We asked whether PSCA-CAR T cells can specifically target and damage PSCA-expressing cells in vitro. First, we selected two cell lines based on their PSCA expression profiles: the human bladder carcinoma cell line PC-3M is known to be PSCA-negative in cultures, whereas the human prostate cancer cell line RT4 is reported to overexpress PSCA under cultured conditions (14–16). Our FACS data confirmed that a majority of RT4 cells (81.6%), but not PC-3M cells (0.01%), had PSCA expression (Fig. 3A), which was consistent with the literature (16).

Next, to test whether PSCA-CAR T cells can damage target cells, we performed in vitro killing assays. Cells from three donors were nontransfected, mock transfected, or transfected with mcDNA-PSCA-CAR for 8 days, and their in vitro cytotoxic effects were statistically analyzed. First, we cocultured normal T, mock T, or PSCA-CAR T cells (effector cells) with PC-3M or RT4 cells (target cells) for 24 hours in various effector-to-target cell ratios (2:1, 5:1, 10:1, and 20:1) and evaluated the percentage of killing using a CCK-8 kit (Cell Counting Kit-8, DOJINDO). We found that PSCA-CAR T cells demonstrated strong cytotoxicity against PSCA-positive RT4 cells (Fig. 3B, left), but not PSCA-negative PC-3M cells (Fig. 3B, right). We also found that the killing efficiency of PSCA-CAR T cells was dose-dependent, and it increased with a higher effector-to-target ratio (Fig. 3B, left). Finally, as a negative control, normal T and mock T cells did not exhibit obvious cytotoxicity against PSCA-positive RT4 cells (Fig. 3B), indicating specific killing effects mediated only by PSCA-CAR T cells.

We then examined cytokine secretion of PSCA-CAR T cells produced from three donors. Our ELISA results showed that, upon coculture (effector-to-target cell ratio, 10:1), PSCA-CAR T cells secreted a large amount of IFNγ and IL2 in response to PSCA-positive RT4 cells, but not PSCA-negative PC-3M cells (Fig. 3C); while as a control, normal T and mock T cells exposed to RT4 or PC-3M cells released only background levels of IFNγ and IL2 (Fig. 3C). To sum up, PSCA-CAR T-cell–mediated cytotoxicity was dose-dependent, specific to the PSCA expression, and resulted in the secretion of IFNγ and IL2.

First, we examined PSCA expression levels in RT4- and PC-3M xenograft tumors. According to the literature, PC-3M had a strong PSCA expression in the xenograft tumor (15) whereas RT4′s, although it is known to highly express PSCA in culture (ref. 16; Fig. 3A), PSCA expression in vivo has not been tested. To examine their ability to express PSCA in vivo, we injected RT4 or PC-3M cells (3 × 10⁶ cells/mice, 15 days after injection, mice were sacrificed for analysis), respectively, into NOD/SCID immunodeficient mice and performed IHC using an anti-PSCA antibody (Fig. 4A). Our results showed that the PC-3M xenograft tumor expressed a large amount of PSCA, whereas the RT4 xenograft tumor expressed little PSCA (Fig. 4A). Therefore, we chose the PC-3M xenograft model for in vivo antitumor studies.

Next, we sought to examine the therapeutic effects of PSCA-CAR T cells in vivo. Cells were nontransfected, mock transfected, or transfected with mcDNA-PSCA-CAR and cultured for 8 days before harvesting. We applied a humanized immune system mouse model bearing tumor xenografts: we first transfused NOD/SCID immunodeficient mice with human peripheral blood for 28 days before subcutaneously inoculating 3 × 10⁶ PC-3M cells (PC-3M cells do not express PSCA in vitro, but express high levels of PSCA in vivo) to develop the xenograft tumors. Approximately 15 days after inoculation (day 1), we injected the mice (two injections in total, on day 1 and day 8, respectively) with xenograft tumors with saline (NS), 1 × 10⁷ normal T, or 1 × 10⁷ PSCA-CAR T cells and observed for 4 weeks. For each treatment group, 8 mice were used, and the experiment was repeated 3 times.

Our data suggested that the 28-day treatment with PSCA-CAR T cells resulted in significantly reduced tumor volumes (100.0 ± 6.16 mm³) as compared with those in the NS and normal T group (1834.7 ± 78.8 and 1435.4 ± 79.7 mm³, respectively, P < 0.0001; Fig. 4B and C), and throughout the course of this study, although xenograft tumors were growing rapidly in the NS and normal T groups, the tumor volumes of the PSCA-CAR T group were decreasing, suggesting a strong and long-term antitumor effect (Fig. 4C).

To investigate whether PSCA-CAR T cells continued to express PSCA-CAR in vivo, we performed FACS analysis for mouse peripheral blood and found that the PSCA-CAR T cells strongly expressed PSCA-CAR on the cell surface as long as 4 weeks after the treatment (Fig. 4D). In addition, our ELISA results showed that 28 days after the treatment, mouse serum IFNγ and IL2 levels in the PSCA-CAR T group were significantly higher than the other two groups (Fig. 4E). Taken together, our data suggest that PSCA-CAR T cells expressed PSCA-CAR in mouse peripheral blood and secreted significantly more cytokines for 28 days and played a persistent antitumor effect.

Finally, to study whether PSCA-CAR T cells effectively infiltrate into the tumor tissues in vivo, we performed IHC analysis. Although PC-3M cells do not express PSCA in vitro, they strongly express PSCA when forming xenograft tumors in vivo, and therefore these cells can be specifically targeted by PSCA-CAR T cells. Our results showed that PSCA was strongly expressed in the xenograft tumors in all three groups (Fig. 4F, middle row). In contrast, staining with the T-cell marker, CD3, revealed no infiltration of T cells in the NS group, few infiltrated T cells in the normal T-cell group, and notably more infiltrated T cells in the PSCA-CAR T-cell group (Fig. 4F, bottom row). This result suggests that PSCA-CAR T cells, but not normal T cells, specifically targeted and moved toward PSCA-positive cancer cells.

Clinical-grade CAR T cells are manufactured mostly using retroviral and lentiviral vectors (1, 2, 17, 18) and transposable elements (6, 7). A novel nonviral vector system, mcDNA, which is free of bacterial DNA backbone, has the advantages of higher transfection efficiency, stronger transgene expression, and longer expression duration in vivo (8, 9). We tested whether mcDNA can be used to produce CAR T cells, and our results showed that mcDNA-PSCA-CAR can be electroporated into human T cells with a high transfection efficiency (58.66%), a large portion of transfected T cells (60.2%) survived and proliferated, and 8 days after electroporation, strong expression of PSCA-CAR was detected in 88.96% cells. Moreover, as long as 4 weeks after injecting into mice, PSCA-CAR T cells were still present in mouse peripheral blood and played persistent roles in cytokine secretion, tumor infiltration, and antitumor effects (as discussed below). Our data demonstrated that mcDNA is feasible and effective in generating CAR T cells.

Currently, CAR T cells have demonstrated a remarkable therapeutic potential in blood tumors (17, 18) and solid tumors such as glioblastoma (19) and renal carcinoma (20). Although CAR T cells targeting against blood tumors demonstrated better potency, severe adverse effects occurred including patient deaths, brain edema, and cytokine release syndrome (17, 18). These adverse effects are caused by CAR T cells' “on-target, off-tumor” effects or relatively quick immune responses when attacking blood cancer cells (17, 18, 21). On the other hand, CAR T cells targeting against solid tumors had less side effects but limited antitumor efficacy (14, 19, 20, 22, 23), probably due to the immunosuppressive tumor microenvironment. However, most studies to date used viral-based methods to manufacture CAR T cells that have safety concerns including insertional mutagenesis, oncogenesis, and viral replication. Retroviral and lentiviral vectors preferentially integrate into oncogenic loci of the host genome and had caused blood cancers in clinical trials (3, 4). In addition, although advanced techniques of generating retroviral and lentiviral vectors can reduce the probability of creating pathogenic replication-competent viruses, these techniques only attenuate, but not eliminate, the chance of viral replication (4). In contrast, the mcDNA vector has no viral replication, merely rare random integration, and a much safer profile as compared with viral vectors (8, 9).

PSCA, a prostate tissue-restricted antigen, is expressed on the surface of primary and metastatic prostate cancers (24, 25). Given its very low expression in normal tissues, PSCA is an ideal target for CAR T immunotherapy against prostate cancers. Early reports showed that delayed tumor growth in mice had been achieved using second- and third-generation CAR T cells targeting PSCA (14, 22). In this study, we applied third-generation CAR T cells in a humanized immune system mouse model and observed significantly delayed tumor growth and decreased tumor burden (Fig. 4B and C). Our observation is similar to that of Abate-Daga and colleagues, in which the tumor burden was gradually reduced (22), but not to that of Hillerdal and colleagues, in which, despite the delayed tumor growth, the tumor burden was still increasing (14). One possible reason is that Abate-Daga and colleagues applied a humanized immune system mouse model (22) by injecting the mice with human PBMCs – such a model better resembles the immune system and tumor microenvironment in human patients; on the other hand, Hillerdal and colleagues used human xenograft nude mouse model without humanization (14) and therefore it could not accurately assess the effectiveness of CAR T cells in vivo.

In addition, both the abovementioned studies (14, 22) used viral vectors to produce CAR T cells. By applying mcDNA to generate PSCA-CAR T cells, we still detected PSCA-CAR–positive cells 28 days after injection, suggesting that PSCA-CAR was expressed persistently in vivo (Fig. 4D). In addition, PSCA-CAR T cells, but not normal T cells, secreted significantly higher levels of cytokines (Fig. 4E) and infiltrated tumor tissues by a greater extent and caused more tumor necrosis (Fig. 4F) as long as 4 weeks. Taken together, these results indicate that mcDNA-PSCA-CAR can be expressed for a long time in vivo, and PSCA-CAR T cells have a long-lasting antitumor effect, similar to CAR T cells transduced using lentivirus or retrovirus (14, 22).

Recently, there is one report (by Cheng and colleagues) applying mcDNA to generate CD19-CAR T cells (26). This is the first study to produce CAR T cells using mcDNA, and mcDNA-transduced CAR T cells had a strong and persistent CD19-CAR expression, showing the potential of using mcDNA as an alternative nonviral vector to produce clinical-grade CAR T cells (26). However, our studies are different in three ways. (i) The bacteria-free mcDNA system developed by Cheng and colleagues is based on multiplexed PCR using 96 pairs of primers, restriction digestion, and ligation-based cyclization which, although elegant, is labor-intensive, difficult to scale up, and therefore not suitable for translating from bench to bedside. In comparison, our system produced a single mcDNA in large-scale followed by endotoxin removal that is more applicable in the industry. (ii) Their CAR construct is a second-generation CAR targeting CD19, whereas ours is a third-generation humanized CAR targeting PSCA. (iii) After they electroporated their mcDNA prepared from a commercial kit (System Biosciences, same as ours), the cell viability was too low to continue functional analysis. However, after we electroporated our mcDNA prepared from the System Biosciences kit, T cells proliferated robustly within 10 days (Fig. 2E), and >88% of T cells expressed CAR 8 days after transfection (Fig. 2D). Such a difference might stem from different mcDNA constructs and/or downstream purification procedures. Therefore, our mcDNA production and purification protocols can be readily applied in the industry and for clinical purposes.

Taken together, our findings suggest that the mcDNA-CAR system is a simple yet efficient platform to generate clinical-grade antigen-specific CAR T cells in view of its clinical safety, low cost, and ease to produce. Our in vitro and in vivo results demonstrate that the mcDNA-PSCA-CAR T cells are specific, effective, and persistent against PSCA-positive tumor cells. This technique is safe as it does not possess risks such as viral replication, insertional mutagenesis, and oncogenesis. With standardized DNA purification procedures, mcDNA-CAR can be easily produced in a large scale, thus facilitating CAR T-cell manufacturing in the industry.

No potential conflicts of interest were disclosed.

Conception and design: J. Han, F. Gao, S. Geng, X. Ye, Z. Fu, L. Shi, J. Cai

Development of methodology: J. Han, F. Gao, S. Geng, X. Ye, T. Wang, P. Du, Z. Cai, Z. Fu, L. Shi, J. Cai

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Han, S. Geng, X. Ye, P. Du, Z. Cai, L. Shi, J. Cai

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Han, S. Geng, X. Ye, T. Wang, Z. Cai, L. Shi, J. Cai

Writing, review, and/or revision of the manuscript: J. Han, S. Geng, L. Shi, J. Cai

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Han, P. Du, L. Shi, J. Cai

Study supervision: S. Geng, T. Wang, Q. Li, J. Cai

This study is supported by research funds from Hebei Government Special Project of Giant Plan (2013–2018) and Hebei Government Special Project of Top Talents (2014–2019; to J. Cai).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.