Chimeric antigen receptor (CAR) T cells are genetically engineered T cells expressing CAR, which is a non-major histocompatibility complex (MHC) restricted receptor containing a single-chain variable fragment (scFv) derived from an antibody, a transmembrane domain, and an intracellular signaling domain. Adoptive cell therapy using CAR T cells showed remarkable tumor specificity and robust anti-tumor immune responses, resulting in complete responses (1). To date, the Food and Drug Administration (FDA) has approved four autologous CAR T cells therapy products: tisagenlecleucel for acute lymphoblastic leukemia Novartis, 2017, axicabtagene ciloleucel for large B cell lymphoma Gilead, 2017, brexucabtagene autoleucel for mantle cell lymphoma in 2020 and relapsed or refractory B-cell precursor acute lymphoblastic leukemia (ALL) in 2021 (Gilead), and lisocabtagene maraleucel for relapsed or refractory large B-cell lymphoma (Bristol Myers Squibb, 2021) (2). Even though autologous CAR T cell therapies show remarkable therapeutic efficacy, they suffer several limitations. Autologous CAR T cell therapy requires T cell harvest from the patients, followed by genetic modification to express CAR and expansion, which could take at least 2 weeks. During this process, the progression of aggressive tumors could be lethal. Because the expansion of T cells depends on the quality attributes of the input cells, the inability to optimize the quality of a patient's T cells could lead to poor expansion yield. Expansion of tumor-infiltrating leukocytes provides an attractive strategy as the lymphocytes are primed against multiple tumor-associated antigens. However, their expansion ex-vivo is not effective, because of their exhausted phenotype and limited replicative potential (3). In addition to the above, the prohibitive cost of the procedure presents a critical challenge for autologous cell therapies (4). In contrast, allogenic cell therapy utilizes T cells from a donor. Because the donor does not undergo chemotherapy, the immune cells are more capable of eliminating target antigen harboring cells (5). Allogeneic cell therapy could be available as an ‘off-the-shelf' product, addressing the challenges of autologous cell therapy. Since matching of HLA-A, -B and -DR could potentially negate graft vs. host disease (GVHD), cell therapy products generated from selected individuals could apply to a wider population (6). Recently, Fate Therapeutics has reported the safety of administering allogeneic NK cell therapy products in doses up to 3 x 10⁸ cells (7). Hence, cell banks could be generated from optimal T cell subpopulations of healthy individuals, decreasing the production cost, while increasing the applicability and effectiveness. Therefore, allogeneic cell therapy has the potential to break the limitations of autologous cellular therapy (8). Owing to the promise of allogeneic cellular therapy, clinical trials evaluating the efficacy of allogeneic, ‘off-the-shelf', CAR T cells, targeting various tumor-associated antigens, are in progress (9, 10). CTX110 (CRISPR Therapeutics) (11), UCART19 in clinical trials NCT02808442 and NCT02746952 (Cellectis), and PBCAR0191 (Precision Biosciences) (12) are to name a few. Preliminary results show that UCART19 when administered to treat adult and pediatric patients with high-risk CD19⁺ relapsed/refractory B-cell acute lymphoblastic leukemia showed an efficacy of 88% of patients achieving complete responses (13). Furthermore, PBCAR0191 has shown similar efficacy rates compared to autologous CAR T cell therapy (14). Given the progress in cellular therapy, especially on the allogeneic front, the requirement for industrial-scale production of T cell products is inevitable. According to the Leukemia and Lymphoma Society, in the USA the expected number of new cases per year for leukemia is ~60,000, and for lymphoma is ~90,000 (https://www.lls.org/facts-and-statistics/facts-and-statistics-overview#Leukemia). While the number of CAR T cells per dose varies, estimates suggest the requirement of 3 trillion CAR T cells/year to treat hematological malignancies and 150 trillion CAR T cells for solid tumor malignancies. To meet the required batch sizes, scaling up a static flask-, in which the cells are grown on a flat surface (2D) will result in added labor, lab footprint, variability, risk of contamination, and poor process control. Stirred tank bioreactors (STRs) are available from small scale (1 L), bench scale (1–10 L) to pilot scales (10–1,000 L). Because the scale-up transfer kinetics are well characterized for STRs, they provide an excellent option, meeting the demand (15). Furthermore, as STRs are closed, automated, and GMP co-based culture compatible systems with an in-process control, T cell manufacturing in STRs significantly decreases the labor, batch to batch variation, and contamination risk. T cells are non-adherent, inherently cultured in suspension as single cells. This makes suspension-based bioreactors suitable for their expansion, without the artificial need to adapt them to suspension culture conditions and culture them in aggregates or attached to carriers. Similar to other cell types, expansion of T cells are sensitive to the buildup of cell metabolites, such as ammonia and lactate, in the culture media and hence requires media replenishment (16). Even though fresh nutrients can be supplied using the fed-batch culture method, the buildup of inhibitory agents in the media could potentially inhibit the cell growth, warranting continuous media perfusion. The attributes that make T cells suitable for expansion in 3D suspension cultures (such as stirred tank bioreactors where the cells are grown in 3-dimensional suspension culture) also present a challenge for automated, continuous, media change. Because the diameter of T cells ranges from 5 to 10 mm, perfusion of T cell cultures has been difficult, with major responsible factors being frequent filter fouling and cell escape. Hence, a scale-up technology equipped with continuous media perfusion enabling T cell expansion is urgently needed.

Allogeneic CAR T cell generation requires genetic engineering, which involves knocking out T cell receptor (TCR), selective knockout of MHC I/II allele, on top of the genetic manipulation needed for the expression of the CAR molecule. Even though cutting edge gene-editing technologies, such as Zinc- finger nucleases (ZFNs) and MegaTALs, clustered regularly interspaced short palindromic repeats (CRISPR/Cas9), and transcription activator-like effector nuclease (TALEN), provide tools for successful deletion and insertion of desired genes, although 100% efficacy cannot be guaranteed, thus, resulting in the presence of unedited cells in the culture (17). The presence of TCR-positive cells in the final allogeneic cell therapy product could result in graft vs. host disease, which could be life-threatening. Hence, depletion of undesired cells is mandatory and requires a separate unit operation, which increases the chances of cell loss and contamination. To achieve the clinical demand of high cell numbers, scaling up the depletion step could potentially involve unreasonable processing time and/or non-viable scale-out.

Here, we describe a closed and scalable platform for T cell manufacturing. This platform offers in-process control and includes T cell expansion in a stirred tank bioreactor with continuous media perfusion. Following T cell expansion, undesired cell depletion is performed in-vessel, avoiding a unit operation, while reducing labor and contamination risks. Furthermore, cell processing and concentration are performed in a closed manner, without compromising cell yield and the characteristics of the cells.

Here, we describe the methods utilized to isolate T cells from peripheral blood mononuclear cells (PBMCs), activate and expand in different culture conditions, process parameters used for T cell expansion, cell purification, concentration, and characterization.

Allogeneic CAR T cells can be used as ‘off-the-shelf' cell therapy and hold the promise to increase the applicability of the cell therapy product to the population on a wider scale. Current estimates suggest the need for batch sizes of 2,000 L to meet the demand for CAR T cells used for hematological and solid tumor malignancies (18). Given the large footprint, risk of contamination, variability, and poor process control in static 2D cultures (the cells are grown on flat surfaces such as culture flasks), stirred tank bioreactors to provide an excellent platform for cell expansion of T cells, offering well-characterized scale-up kinetics, in-process control and low risk of contamination. Translating 2D expansion processes into 3D expansion, in stirred tank bioreactors, was successfully shown for adherent cell types (19). However, the absence of inherent perfusion capability of STRs and the small size of T cells (5-10 mM in diameter) proved to be formidable obstacles to achieving high yields of T cells in STRs. Here, we describe a GMP-compatible, closed, and scalable platform for T cell expansion in perfusion-enabled STR. In addition, our platform offers in-unit and potentially scalable cell depletion magnetic technology, avoiding a unit operation, and lowering the risk of contamination and labor (Figure 6).

The T cell expansion in agitated conditions in spinner flasks was found to be higher, compared to static 2D culture in T25 flask (Figures 1A,B). Agitation at a constant rate of 75 RPM enabled better T cell expansion compared to agitation at a lower speed of 50 RPM during activation, followed by agitation at an increased speed of 100 RPM. Costariol et al. have shown that the expansion of primary T cells is increased with agitation speed in an automated STR (20). In agreement, we have observed that agitation at lower speeds of 25 RPM and 35 RPM did not enable T cell expansion (data not shown). A study performed by Carswell et al. has shown that T cells in spinner flasks can be expanded without detrimental effect up to 120 RPM agitation speed spinner flask (21). Maintaining the tip speed, 120 RPM in a 125 mL spinner flask is equivalent to 95 RPM in a 1-L stirred tank bioreactor. Hence, we have evaluated the T cell expansion at a lower agitation rate of 65 RPM and a higher agitation rate of 88 RPM in a 1-L stirred tank bioreactor. In agreement with the data from Carswell et al., an increased VCD of T cells was observed at higher agitation rates in an STR (Figure 1C). Cell growth accompanied by the consumption of nutrients results in the accumulation of lactate in the cell culture, which hinders efficient cell growth and quality of the final cell therapy product (22, 23). Continuous media perfusion enables a fresh supply of nutrients and the removal of harmful metabolites while retaining the cells. However, owing to the 5–10-mm diameter of T cells, cell escape, and filter fouling have been major issues related to T cell culture media perfusion. A tangential flow filtration system provides an attractive solution as a cell retention system, where the movement of fluid enables media perfusion without fouling. Furthermore, ATF provides an additional advantage of self-cleaning induced by a backflush of alternating flow (24). Repligen provides ATF devices enabling perfusion from the scales of 0.5 L to 1,000 L, serving as an attractive option for continuous media perfusion in different scales of culture in STR. As the pore size of the hollow fibers in Repligen's ATF is 0.2 mm, we have employed ATF as a cell retention device for continuous media perfusion in STR. Initial testing of ATF for cell retention and media exchange resulted in successful perfusion of 1 VVD without filter fouling and cell loss (Table 1). Post inoculation of the STR with T cells at 0.5 × 10⁶ cells/mL, T cells were activated for 3 days with CD3+28 in the presence of recombinant human IL-2. After 3 days of activation, the cells were expanded in X-VIVO^TM 15 media containing recombinant human IL-2. Monitoring lactate levels, we have observed a 3-fold increase in lactate levels from day 7 to day 8 (Figure 2C). This is accompanied by an increase in viable cell density to 2.1 x 10⁶ cells/mL (Figure 2B) and a drop in glucose concentration to 5.4 mM on day 8 from 6.3 mM on day 7 (Figure 2D), warranting media perfusion. Enabling media perfusion on day 8 has resulted in steady-state levels of glucose and lactate (Figures 2C,D). Furthermore, continuous media perfusion has enabled exponential growth of T cells, yielding ~35 × 10⁶ cells/mL as final VCD. In comparison, T cells from the same donor were activated and expanded in 1L G-Rex®. T cell expansion in G-Rex® resulted in a VCD of 3.4 x 10⁶ cells/mL on day 14, yielding a 28-fold expansion of T cells. A 14-day culture in STR with ATF mediated perfusion resulted in a 167-fold expansion of T cells (Figure 2B). In addition, we have observed that agitation in STR in combination with ATF-mediated cell movement did not decrease cell viability or cell expansion (Figure 2E).

Turtle et al. reported in their clinical trials that CAR T cell therapy with a 1:1 CD4:CD8 T cell ratio resulted in 93% of the patients achieving bone marrow remissions (25). In Figure 3A, we show that the CD4:CD8 T cell ratio is maintained in the STR over 14 days of expansion, suggesting a 1:1 seeding ratio would be maintained by harvest. Growing evidence suggests that the presence of a high number of naïve, stem cell memory, and central memory T cells in the final CAR T cell therapy product results in relapse-free remissions, compared to lower efficacy observed in CAR T cell products containing a higher percentage of effector memory subset (26, 27). We show that T cell expansion in the tested conditions resulted in the final T cell subtype composition containing ~80% central memory subset, ~15% naïve and stem cell memory subset, <10% effector memory, and terminal differentiated subsets (Figure 3B). This phenotype can be explained by the higher replicative potential of naïve T cells (28) and stem cell memory T cells, which replicate and differentiate upon activation (29). The higher differentiation from naïve and T memory stem cells (Tscm) combined with the higher replicative potential of central memory T cells could have yielded a higher percentage of central memory T cells. Even though final T cell subtype concentrations depend on the initial T cell composition, T cell expansion in Quantum yielded similar phenotypic results (30). Presence of senescent and exhausted T cells in the final drug product results in lower efficacy, explained by lower replication potential and dysfunctional state of the T cells, respectively (31). We show that T cell expansion in STR did not result in >5% senescent (CD57⁺ KLRG1⁺) or exhausted T cells (CTLA4⁺/PD-1⁺) (Figure 3C). Polyfunctional strength index^TM (PSI) of T cells was predictive of clinical responses in Acute Myeloid Leukemia and could be used as a biomarker in immunotherapy (32). PSI of the immune cells is calculated by multiplying the number of cytokines produced by each cell with the amount of each cytokine. Evaluation of PSI of CD4⁺ and CD8⁺ T cells after stimulation for 5 hours suggested an increase in the number of effectors CD4⁺ and CD8⁺ T cells from day 0 to 14. In addition to an increase in the effector signature, we have observed an increase in stimulatory signature, specifically in CD8⁺ T cells (Figure 4D). Polyfunctional T cells are capable of producing 2+ cytokines upon stimulation with antigen and are considered a vital functional characteristic (33). Evaluation of the share of cells producing multiple cytokines suggested an increase in the percentage of CD4⁺ and CD8⁺ T cells producing multiple cytokines, specifically 5+ cytokines (Figure 4B). Furthermore, ~1% of CD4⁺ T cells have produced 11 cytokines, and ~1% of CD8⁺ T cells have produced 9 cytokines upon activation (Figure 4C). As assessed by the phenotypic characterization of T cells, an increase in the central memory phenotype could be translated into an increase in the\ number of cells producing multiple cytokines (34).

Generation of allogeneic CAR T cells requires deletion of TCR alpha coding TRAC locus. Different strategies are employed to excise TRAC locus and they widely differ in their efficiency. CRISPR/Cas9 shows efficiency of 70–80%, TALEN – 60 to 80%, Zinc finger nucleases – 20 to 40% and megaTALs show an efficiency of 75% (17). Presence of TCR-positive T cells in the allogeneic cell therapy product results in GVHD (35) which could be life-threatening if involves the clonal expansion of donor T cells (36). Hence, TCR-positive T cells have to be depleted before concentration and formulation. TCR-positive T cell depletion requires harvesting T cells from STR and process through an exclusive unit, which increases the chance of contamination and incubation of cells in unoptimized conditions. We have developed a proprietary magnetic technology, which facilitates cell depletion in STR while maintaining optimized conditions. In addition, cell depletion in STR avoids the need for an additional unit operation and decreases the chance of contamination. As a proof of principle, we have depleted CD4⁺ T cells after T cell expansion. As shown in Figure 5A and Table 2, 30 min of magnetic cell depletion effectively depletes CD4⁺ T cells at lower percentages (>99%) and higher percentages (97%). As per FDA guidelines, the residual bead percentage of <0.003% in the final cell therapy product ensures the minimal effect of beads in the system. Using magnetic depletion for 120 mins, we show a decrease in the residual bead percentage to 0.001% without cell loss and a significant drop in viability (Figure 5B and Table 3). Furthermore, our proprietary magnetic technology could potentially be scaled to yield similar results at different scales.

Post depletion of undesired T cells, concentration and formulation is performed to decrease the volume, amicable for therapeutic use. We have evaluated two automated, mutually exclusive, and closed systems for cell processing and concentration. The ekko^TM performs the concentration using acoustic technology and kSep concentrates the cells by fluidized bed formation using centrifugal force. As shown in Table 4, we have concentrated the cells at least 7 fold, recovering >70% cells without a significant drop in cell viability.

To the best of our knowledge, this is the first study to describe the successful application of ATF for cell retention and media perfusion in T cell manufacturing. Collectively, our data demonstrate the capabilities of our closed, GMP compatible, end-to-end platform in expanding T cells to clinically relevant doses, purifying cell cultures, in concentration, and formulation of cell products (Figure 6).

Here, we presented a closed, scalable, and automated, platform for T cell manufacturing. Continuous media perfusion leads to ~40 × 10⁶ cells/mL while maintaining nutrient and metabolite levels. Expanded cells show stemness over senescence markers. Post closed to harvest, cell concentration is enabled a closed, automated, equipment. For genetically engineered allogeneic CAR-T cells, selecting TCR-positive cells is needed. Our platform utilizes a proprietary in-vessel magnetic-based selection. This platform selectively and efficiently depletes undesired cell populations, avoiding the need for cell transfer into an additional unit of operation. This enables maintaining the cells under optimal culture conditions during the selection step and reduces contamination risk. Moreover, our proprietary in-vessel selection technology is highly scalable, and prepared to meet future clinical demand for a large number of allogeneic CAR-T cells.

Data may be available upon request at Lonza's discretion. Requests to access these datasets should be directed to Lonza, pharma@lonza.com.

HG, NU, and AS designed the various study parts, conducted experiments, and analyzed the data. AN contributed to the execution of the STR BRX runs. HG, NU, and AS have contributed to methodology development. YL, SR, and IF contributed to the conception and design of the study and gave intellectual input throughout the project and to the analysis of the data. HG wrote the manuscript. IF and AN provided critical revision to the manuscript. All authors approved the final manuscript.

All authors are current or previous employees of Lonza, a pharmaceutical company that develops and sells a wide range of products, including cell-based products for research and pharmaceutical use. Lonza may derive benefit from the sale of a product derived from this research. However, this does not alter our adherence to all the IJMS policies on sharing data and materials.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.