CAR T Cell Protocol: Step-by-Step Laboratory Techniques

Manufacturing chimeric antigen receptor T cells involves a chain of interconnected laboratory steps, each building on the last: collecting a patient’s white blood cells, isolating T cells, activating them, inserting a new gene that encodes the CAR, expanding the modified cells to therapeutic numbers, and then rigorously testing the final product before it reaches the patient. The entire process typically spans one to two weeks in a controlled cleanroom environment, and the choices made at each stage shape the potency, safety, and persistence of the cells that ultimately get infused.

Harvesting and Isolating T Cells

The process begins with leukapheresis, a blood-draw procedure that selectively collects white blood cells while returning red cells and plasma to the patient. The resulting product is a mixed bag of immune cells, including T cells, monocytes, B cells, and natural killer cells. Because CAR therapy needs T cells specifically, the next task is pulling them out of the crowd.

Most protocols enrich for CD3-positive T cells, since CD3 is a surface marker found on virtually all T cells but not on other white blood cells. Magnetic bead-based selection is the workhorse here. In one common approach, peripheral blood mononuclear cells are thawed and rested overnight before CD3-positive T cells are isolated using magnetic beads conjugated with antibodies against CD3 and CD28.¹ The overnight rest step helps improve viability after the stress of freezing and thawing. Patients with high monocyte counts can pose a challenge at this stage, since monocytes tend to stick to beads and clog the selection process, sometimes requiring extra purification steps.

Activating T Cells

Freshly isolated T cells are in a quiescent state and need a strong wake-up signal before they will accept a new gene and start dividing. In the body, antigen-presenting cells deliver this signal through two simultaneous cues: one through the T cell receptor and another through costimulatory molecules on the cell surface. Laboratory activation mimics this by providing both signals artificially.

The most established method uses beads coated with antibodies against CD3 and CD28, which partially recreate the way antigen-presenting cells stimulate T cells.² These magnetic beads, commonly sold under the Dynabeads brand, are mixed with T cells at a one-to-one ratio in culture medium supplemented with interleukin-2 at around 300 international units per milliliter, then incubated at 37°C with 5% carbon dioxide for 48 hours.³ An alternative is a soluble nanomatrix reagent called TransAct, which delivers the same CD3/CD28 signals but without physical beads, making downstream removal simpler. TransAct is typically diluted 1:100 in culture medium before being added to the cells.³

The choice between beads and soluble reagents affects more than convenience. Bead-based activation gives a potent, sustained signal, but the beads must be magnetically removed before infusion. Soluble reagents wash out more easily during media changes. Both approaches, however, share the same goal: pushing T cells into rapid division so they become receptive to gene delivery.

Getting the CAR Gene into T Cells

Once T cells are activated and dividing, the next step is inserting the DNA sequence encoding the chimeric antigen receptor. This is where the science gets most technically demanding, and there are two broad families of approach: viral vectors and non-viral delivery.

Viral Vector Transduction

Lentiviral and gamma-retroviral vectors are the most widely used tools for delivering the CAR gene. Both work by packaging the CAR DNA inside a disabled virus particle that can integrate into the T cell’s genome, giving stable, long-term expression. Lentiviral vectors are produced by transfecting a packaging cell line (usually HEK293T cells) with plasmids encoding the viral components and the CAR transgene. Optimized calcium phosphate-based transfection protocols help achieve high viral titers.⁴

In some protocols, T cells are activated and transduced simultaneously. One recent method adds lentiviral particles along with TransAct and the transduction-enhancing agent polybrene (at 15 micrograms per milliliter) plus interleukin-2, all in a single step, then incubates for 48 hours.⁵ Combining activation and transduction saves time but demands careful optimization of reagent concentrations.

Transduction efficiency varies widely depending on the method. A head-to-head comparison of different enhancement techniques for retroviral transduction found that combining spinoculation (centrifuging cells with the virus to force physical contact) and retronectin (a fibronectin fragment that co-localizes virus and cells) produced the highest efficiency at roughly 63%, compared with about 34% for spinoculation plus polybrene and only about 10% for retronectin or polybrene alone.⁶ These numbers matter: low transduction efficiency means fewer CAR-positive cells in the final product, which can mean lower potency or the need for longer expansion times to hit dose targets.

Non-Viral Gene Delivery

Viral vectors work well but come with significant costs and long lead times for manufacturing the virus itself. Non-viral alternatives are gaining ground, especially transposon systems and CRISPR-based approaches. Transposon systems use a two-component setup: a plasmid containing the CAR gene flanked by specific DNA sequences (inverted terminal repeats) and a transposase enzyme that cuts the gene out of the plasmid and pastes it into the cell’s genome.⁷ The Tc Buster system, for example, has been used to deliver a multi-gene cassette encoding a CD19-targeting CAR along with a selectable marker and a reporter gene into primary human T cells.⁸

The appeal is straightforward: DNA plasmids are far cheaper and faster to produce than viral vectors, and the process sidesteps the complex biosafety requirements of working with live virus. The trade-off is that transposon integration is somewhat less controlled than lentiviral insertion, and achieving consistently high efficiency in primary T cells remains an active area of development.

Expanding Cells to Therapeutic Numbers

A patient typically needs tens of millions to billions of CAR T cells for treatment, so the modified cells must be expanded dramatically after gene delivery. The expansion phase usually lasts about seven to ten days and requires careful control of temperature, gas exchange, and nutrient delivery.

Bioreactor Platforms

Static culture in bags or flasks works for small-scale or early-phase production, but automated bioreactors offer better reproducibility and scalability. The Quantum Flex system, a hollow-fiber bioreactor, has demonstrated 150- to 200-fold expansion in just seven days, and this held true across a range of starting cell amounts from one million to fifteen million cells.⁹ The hollow-fiber design mimics the capillary structure of blood vessels, letting nutrients and waste products exchange efficiently while keeping cells in a protected environment.

More recent work has pushed toward integrating multiple manufacturing steps into a single device. One group demonstrated that activation, viral transduction, and expansion could all be performed inside the same hollow-fiber system without ever opening the closed circuit, reducing the risk of contamination and hands-on labor.¹⁰ Automating these steps is a big deal for the field because manual handling is one of the main sources of batch-to-batch variability and a driver of the enormous cost of current commercial products.

Cytokine Choice Shapes the Final Product

The cytokines added to culture medium during expansion are not just growth fuel; they steer the T cells toward particular functional states. Interleukin-2 is the traditional workhorse, but it tends to push cells toward a more differentiated effector phenotype. CAR T cells expanded with interleukin-7 and interleukin-15 instead showed an increase in less-differentiated memory-type cells, including naive and central memory subsets, compared to cells grown in interleukin-2, which favored effector memory cells.¹¹ Less-differentiated T cells tend to engraft better and persist longer after infusion, which can translate into more durable responses in patients. This finding has made the IL-7/IL-15 combination increasingly popular in newer manufacturing protocols.

Quality Control and Release Testing

Before a CAR T cell product can be infused, it goes through a battery of quality checks that together determine whether the batch meets release criteria. These tests assess identity, purity, potency, sterility, and safety.

Confirming CAR Expression by Flow Cytometry

The most basic check is confirming that the T cells actually express the CAR on their surface. Flow cytometry using antibodies or protein ligands that bind the CAR construct is the standard method. Increasingly, high-dimensional panels are being deployed. One group developed a 36-marker spectral flow cytometry panel that simultaneously captures CAR expression, T cell subset identity, exhaustion markers, and in vitro cytotoxicity in a single assay, providing a detailed fingerprint of the product at various stages of manufacturing.¹² This kind of deep profiling is moving from research into routine manufacturing because it catches problems, like excessive exhaustion or skewed subset ratios, that simpler two- or three-color panels miss.

Measuring Vector Copy Number

Regulators require manufacturers to measure how many copies of the CAR transgene have integrated into each cell’s genome. Too few copies means weak expression; too many raises theoretical safety concerns about insertional mutagenesis. Droplet digital PCR has emerged as the preferred method over traditional real-time PCR because it does not require a standard curve and shows better repeatability and a lower detection limit.¹³ The technique works by partitioning a DNA sample into thousands of tiny droplets, each of which undergoes an independent PCR reaction. Counting positive versus negative droplets gives an absolute measure of copy number without the calibration headaches of older methods. For products made with two different vectors, multiplex droplet digital PCR can quantify each vector’s copies independently by targeting unique sequences in each construct.¹⁴

Potency and Cytotoxicity Assays

Proving that CAR T cells can actually kill target cells is the core potency test. A validated approach co-cultures CAR T cells with target tumor cells at a one-to-one ratio for 24 hours, then uses flow cytometry to measure dead target cells, subtracting the background death rate from a control well containing non-CAR T cells and the same target cells.¹⁵ Several other readout methods exist for quantifying the same phenomenon: lactate dehydrogenase release from dying cells, chromium-51 release, live-cell imaging platforms like IncuCyte, and impedance-based real-time systems like xCELLigence.¹⁶ Each has its strengths: real-time systems let you watch killing kinetics unfold over hours rather than getting a single snapshot, while flow-based methods give you the added ability to phenotype the surviving populations.

Cytokine Release Profiling

Understanding which cytokines CAR T cells release, and in what quantities, provides insight into both potency and potential toxicity. Traditional single-cytokine ELISA remains the clinical standard for measuring key molecules like interferon-gamma and tumor necrosis factor-alpha in supernatants.¹⁷ Multiplex bead arrays allow simultaneous measurement of dozens of cytokines from the same sample. Newer single-cell platforms go a step further, isolating individual cells in microchambers to profile the cytokine secretion of each CAR T cell separately, revealing functional heterogeneity within a product that bulk assays would average out.¹⁷

Cytokine release syndrome is one of the most common and dangerous side effects of CAR T therapy, so some groups have developed in vitro models that attempt to predict it before infusion. One protocol co-cultures CAR T cells with tumor cells and autologous monocytes (since monocytes are key mediators of cytokine storm in patients), then measures the resulting cytokine output by ELISA or multiplex array.¹⁸ This three-way co-culture adds complexity but better reflects what happens in the body, where it is not just the T cells but also bystander immune cells amplifying the inflammatory cascade.

Formulation and Cryopreservation

After expansion and quality testing, the cells must be formulated for storage and transport. Most CAR T products are cryopreserved in liquid nitrogen so they can be shipped to clinical sites and thawed immediately before infusion. The standard cryoprotectant is DMSO, typically at 5-10% in a protein-containing base, cooled at about 1°C per minute in a controlled-rate freezer.

DMSO works well but is not harmless. It can cause side effects in patients at infusion, including nausea and cardiovascular reactions, and some patients are more sensitive than others. This has motivated work on DMSO-free alternatives. One approach used a combination of glycerol and betaine and achieved greater than 90% post-thaw recovery, but only when paired with a faster freezing rate of 3°C per minute rather than the traditional 1°C per minute.¹⁹ That finding is counterintuitive: for decades, the field has defaulted to slow cooling, so the idea that faster is better with a different cryoprotectant challenges established assumptions. These serum-free, DMSO-free formulations are still being validated for clinical use, but they represent a promising direction.

Built-In Safety Switches

Because CAR T cells can cause severe toxicity once inside the body, some constructs include a genetic safety switch, often called a suicide gene. One well-studied design fuses a caspase-9 domain to a dimerization domain. When a small-molecule drug called rimiducid is administered, it forces two caspase-9 molecules together, triggering the cell’s programmed death pathway. T cells carrying this construct have been shown to function normally against tumor targets in mice but can be rapidly eliminated when the dimerizing drug is given.²⁰ This kind of on-demand kill switch is especially important when the CAR targets an antigen found on both tumor cells and some healthy cells, since ongoing toxicity to normal tissue may need to be stopped quickly.

Allogeneic Manufacturing and Gene Editing

The protocol described so far is autologous: cells come from and go back to the same patient. An alternative that has entered clinical testing is allogeneic, or “off-the-shelf,” CAR T cells made from a healthy donor’s T cells. The manufacturing steps are largely the same, but with added gene-editing layers to prevent the donor cells from attacking the recipient’s tissues.

The primary edit disrupts the T cell receptor alpha chain, eliminating the donor T cell’s ability to recognize the recipient’s body as foreign and cause graft-versus-host disease. CRISPR/Cas9 and base editing have both been used for this purpose and reached clinical trials.²¹ More advanced approaches aim to make multiple edits simultaneously. One recent strategy combined two different CRISPR systems in the same manufacturing run: one to knock out genes that trigger immune rejection (like beta-2-microglobulin) without creating double-strand DNA breaks, and another to insert the CAR transgene directly into the T cell receptor locus, effectively disrupting the receptor and inserting the CAR in one move.²² Avoiding double-strand breaks where possible is desirable because those breaks can cause unintended rearrangements in the genome.

Allogeneic manufacturing has a compelling economic advantage: one donor’s cells can potentially produce hundreds of doses, compared to the single dose generated by autologous manufacturing. But the extra editing steps add manufacturing complexity, and the edited cells still face elimination by the recipient’s immune system over time, which can limit persistence.

Preclinical Testing in Animal Models

Before a CAR T product moves to human trials, preclinical data are required. The most common approach uses immunodeficient mice engrafted with human tumor cells, known as xenograft models. Patient-derived xenograft models, where the mice receive actual leukemic blast cells from patients rather than immortalized cell lines, offer a more clinically relevant test. These models have been shown to recapitulate both the therapeutic efficacy and the characteristic toxicities of CAR T therapy, including cytokine release syndrome and neurological inflammation, making them useful for evaluating safety as well as potency.²³

No single mouse model perfectly mirrors the human immune system, though. Xenograft models lack a functioning host immune system, so they cannot capture the interplay between CAR T cells and the recipient’s own immune cells, which plays a major role in both efficacy and toxicity in patients. Humanized mouse models, which are reconstituted with a human immune system, partially address this gap but are expensive and difficult to standardize. The field generally uses multiple model types in parallel, accepting that each one tells part of the story.