Advancing Cancer Immunotherapy: Chimeric Antigen Receptor (CAR)-T Cell Engineering through Novel Screening Methods

Abstract

Cancer immunotherapy represents a paradigm-shifting achievement in oncology. Particularly, chimeric antigen receptor (CAR)-T cell therapy utilizing genetically engineered T cells has produced remarkable clinical responses in hematological malignancies. However, significant challenges still remain including limited efficacy in solid tumors and critical safety concerns. The functionality of CAR-T cells depends on their synthetic receptor, CAR, which redirects T cell specificity and enhances effector functions. Therefore, optimal CAR engineering is crucial for successful development of CAR-T cell therapy. In this review, we discuss the limitations of current CAR screening methods, which primarily assess antigen binding affinity in vitro and often fail to predict T cell function and in vivo therapeutic performance. Advanced cell-based screening platforms have been developed to overcome these limitations. We overview the principles of these CAR screening systems utilizing reporter cell lines. While most are based on the detection of antigen binding properties or CAR-T cell activation markers, we emphasize a FRET-based immunological synapse biosensor as a powerful system that directly assesses CAR activation upon antigen binding. This platform offers significant advantages in speed and scalability for predicting CAR-T cell functionality. We also discuss recent advances in CAR library screening directly in primary T cells, which provides more physiologically relevant data. Such advanced platforms are essential to accelerate the development of safe and effective CAR-T therapy for solid tumors, ultimately expanding the therapeutic potential of this transformative cancer treatment.

INTRODUCTION

Cancer treatment has historically relied on three modalities: surgery, radiation therapy, and chemotherapy. Surgery and radiation therapy achieve local tumor eradication. In contrast, chemotherapy provides systemic treatment through cytotoxic agents that target rapidly proliferating malignant cells (DeVita and Chu, 2008). The therapeutic potential of chemotherapy is constrained by limited specificity, as cytotoxic drugs cannot distinguish between malignant cells and healthy, rapidly dividing cells in bone marrow or the gastrointestinal mucosa (Yan et al., 2020). This results in dose-limiting toxicities and a narrow therapeutic window, driving the development of more selective therapeutic paradigms.

Cancer immunotherapy represents a transformative paradigm in oncology by harnessing the host immune system to target tumors. These include immune checkpoint inhibitors (ICIs), monoclonal antibodies that block inhibitory pathways exploited by tumors (Darvin et al., 2018); bispecific antibodies such as T cell engagers that physically link cancer cells to T cells (Wei et al., 2022); and cytokine therapy, which enhances the host immunity by administering immune-signaling proteins (Miller et al., 2009; Rosenberg et al., 1989). Adoptive cell therapy (ACT) represents a more active strategy of cancer immunotherapy, involving the ex vivo expansion and reinfusion of a patient’s own immune cells. The most prominent ACT is chimeric antigen receptor (CAR)-T cell therapy, in which T cells are genetically engineered to express synthetic receptors that recognize tumor-associated antigens on cancer cells, enabling potent and direct tumor cell lysis (Cappell and Kochenderfer, 2023; Hawkins et al., 2010). This approach has emerged as a breakthrough therapeutic modality, demonstrating remarkable clinical efficacy in patients with relapsed/refractory hematological malignancies.

Despite these successes, significant challenges limit the broader therapeutic application of CAR-T cell therapy. Major challenges include safety concerns such as cytokine release syndrome and immune effector cell-associated neurotoxicity syndrome, along with limited efficacy against solid tumors. Additional obstacles to the therapeutic outcomes of CAR-T cell therapy include tumor antigen escape and T cell exhaustion (Cappell and Kochenderfer, 2023; Sterner and Sterner, 2021; Zhang et al., 2023). Therefore, it is crucial to develop efficient and safe CAR-T cell therapy. Current CAR development relies on iterative optimization through in vitro assays and animal models, which often fail to predict clinical performance. This approach is time-intensive, costly, and provides limited insight into the complex cellular mechanisms underlying CAR-T cell function. Consequently, there is an urgent need for advanced CAR screening technologies.

This review examines the evolution of cancer immunotherapy with particular emphasis on CAR-T cell therapy, analyzing both its clinical impact and current limitations. We overview emerging cell-based screening platforms that enable systematic evaluation of CAR performance. These advances in screening methodology provide a framework for developing next-generation CAR-T therapies with improved clinical outcomes and reduced toxicity.

CANCER IMMUNOTHERAPY

Cancer immunotherapy is based on the principle that the immune system can recognize and eliminate malignant cells. This process, termed cancer immunoediting, involves a dynamic interplay between evolving tumors and the host immune system across three sequential phases: elimination, equilibrium, and escape (Mittal et al., 2014). During the elimination phase, innate and adaptive immune cells recognize and destroy nascent transformed cells. However, tumor variants that survive immune clearance enter an equilibrium phase, in which immune pressure suppresses but does not eradicate tumor growth. This equilibrium phase, which may persist for years or decades, exerts sustained pressure that drives the evolution of tumor cell variants with an enhanced capacity to evade immune recognition and destruction. Eventually, these immune-resistant clones expand unchecked during the escape phase, leading to clinically detectable cancer. Consequently, clinically diagnosed tumors represent populations that have successfully evaded immune surveillance through acquired mechanisms of immune escape.

Immune checkpoint inhibitor

Early immunotherapies attempted to alter the balance between tumor cells and the immune system through non-specific immune enhancement. For example, the treatment of cytokines such as IL-2 produced dramatic responses in some patients (Rosenberg et al., 1989). However, severe toxicity and low response rates restricted its therapeutic potential (Rosenberg, 2014). Modern immunotherapy adopted a more precise strategy. Instead of broadly boosting immune responses, researchers focused on identifying and blocking specific mechanisms that tumors use to evade immune detection. This targeted approach led to the development of immune checkpoint inhibitors, which disrupt the inhibitory pathways exploited by tumors to suppress anti-tumor activity of T cells. The durable clinical responses achieved with ICIs across various cancer types have established this strategy as a foundational component of cancer immunotherapy (Robert, 2020; Wei et al., 2024).

Immune checkpoints comprise a network of co-stimulatory and co-inhibitory receptors on T cells that regulate the duration and amplitude of immune responses. While these pathways are essential for maintaining self-tolerance and preventing autoimmunity, tumors exploit these natural regulatory mechanisms to evade immune destruction. The most well-characterized checkpoint pathways involve cytotoxic T lymphocyte-associated protein-4 (CTLA-4) (Rowshanravan et al., 2018; Schildberg et al., 2016) and programmed cell death protein-1 (PD-1) (Freeman et al., 2000; Salmaninejad et al., 2019). CTLA-4 primarily functions in lymphoid organs during T cell priming, competing with the co-stimulatory receptor CD28 for ligand binding and thereby attenuating T cell activation (Collins et al., 2002; Schildberg et al., 2016). In contrast, PD-1 operates predominantly in peripheral tissues and tumor microenvironment (TME). When PD-1 on activated T cells engages its ligands PD-L1 or PD-L2, which are frequently upregulated on tumor cells and other components within the TME, it delivers inhibitory signals that promote T cell exhaustion and functional impairment (Freeman et al., 2000; Latchman et al., 2001; Salmaninejad et al., 2019).

Immune checkpoint inhibitors are monoclonal antibodies that block these inhibitory interactions. Anti-CTLA-4 antibodies (e.g., ipilimumab) and anti-PD-1/PD-L1 antibodies (e.g., nivolumab, pembrolizumab, atezolizumab) physically block receptor-ligand binding, effectively releasing the brakes on T cells and reactivating their suppressed anti-tumor immune responses (Fig. 1A). These agents have demonstrated unprecedented and durable clinical efficacy across an expanding spectrum of malignancies, including melanoma, non-small cell lung cancer, and renal cell carcinoma (Marei et al., 2023). The success of these first-generation ICIs has led to intense investigation into additional checkpoint targets, including lymphocyte-activation gene 3 (LAG-3), T cell immunoglobulin and mucin-domain containing-3 (TIM-3), and T cell Immunoreceptor with Ig and ITIM domains (TIGIT), aiming to further enhance anti-tumor immunity and overcome therapeutic resistance (Marei et al., 2023).

Fig. 1.

Cancer immunotherapy. (A) Immune checkpoint inhibitors are monoclonal antibodies targeting co-inhibitory receptors such as PD-1 on T cells or PD-L1 on cancer cells. By blocking these inhibitory interactions, ICIs can reactivate the suppressed anti-tumor immune responses of T cells. (B) CAR-T cell therapy, a prominent form of adoptive cell therapy, utilizes patient-derived T cells that are genetically engineered to express CAR constructs. CAR comprises an scFv, hinge, transmembrane, co-stimulatory and signaling domains. CAR-T cells recognize and bind to tumor-associated antigens, which trigger the CAR signaling pathways. Activated CAR-T cells express activation markers such as CD69, and exhibit multiple effector functions including transcription factor activation, cytokine production, and granzyme release for tumor cell killing.

Adoptive cell therapy

Adoptive cell therapy represents the most active therapeutic strategy in cancer immunotherapy. Rather than modulating the immune response of patients, ACT involves ex vivo manipulation and expansion of a patient’s immune cells, typically T cells. These cells are then reinfused into the patient, serving as a powerful, living therapeutic agent that actively target and destroy malignant cells (Du et al., 2023; Pinto et al., 2025; Rohaan et al., 2019; Waldman et al., 2020). This approach effectively overcomes a limitation of ICIs, which are often ineffective in patients lacking sufficient pre-existing endogenous anti-tumor immune responses. By providing a substantial quantity of activated, tumor-specific T cells, ACT directly addresses this immunological deficit, thereby executing a robust and targeted cytotoxic attack against cancer cells.

Several ACT modalities have achieved clinical success through distinct mechanisms and applications. The tumor-infiltrating lymphocyte (TIL) approach utilizes the patient’s native anti-tumor immune responses. Patient T cells are directly harvested from the TME, where they have already demonstrated the capacity for infiltration and survival in this challenging environment. These TILs are then expanded ex vivo to clinically relevant numbers, often in the billions, using high-dose IL-2 before reinfusion. The clinical success of this modality in solid tumors, particularly advanced melanoma, highlights its ability to harness the existing anti-tumor immune repertoire (Andersen et al., 2016; Ben-Avi et al., 2018; Lee et al., 2017; Spiess et al., 1987).

CAR-T cell therapy is a clinically successful ACT modality. This approach involves the genetic engineering of T cells to express chimeric antigen receptors (Asmamaw Dejenie et al., 2022; Pinto et al., 2025). The CAR construct is typically composed of a single-chain variable fragment (scFv) for tumor-associated antigen recognition and intracellular T cell signaling domains (Asmamaw Dejenie et al., 2022; Labanieh and Mackall, 2023). This design enables CAR-T cells to specifically recognize and bind to the antigens on the surface of cancer cells in an MHC-independent manner, bypassing a key mechanism of tumor immune evasion (Gross et al., 1989; Kuwana et al., 1987) (Fig. 1B). For example, CD19-targeting CAR-T cell therapies, such as tisagenlecleucel and axicabtagene ciloleucel, have demonstrated exceptional remission rates in patients with relapsed/refractory B-cell malignancies (Cerrano et al., 2020; Ghafouri et al., 2021).

In contrast to CAR-T cells, engineered T cell receptor (TCR) therapy modifies T cells to express a new high-affinity TCR that recognizes tumor-specific antigens presented by MHC molecules. This approach leverages the natural mechanism of T cell recognition, as the engineered TCR recognizes the intracellular antigens that are processed and presented on the cell surface by MHC molecules. This capability significantly broadens the scope of potential targets to include a vast range of intracellular proteins, making it particularly promising for solid tumors (Ghafouri et al., 2021; Ping et al., 2018; Shafer et al., 2022).

Despite their therapeutic promise, the widespread adoption of ACTs is hindered by significant translational challenges. These include high manufacturing complexity and cost, as well as the potential for severe toxicities such as cytokine release syndrome and immune effector cell-associated neurotoxicity syndrome. To address these challenges, current research efforts target the development of off-the-shelf allogeneic products, safety enhancements through suicide genes or on/off switches (Dagar et al., 2023).

CHIMERIC ANTIGEN RECEPTOR (CAR)-T CELL THERAPY

CAR-T cell therapy represents a landmark achievement in cell therapy. This technology, which engineers a patient’s T cells into living therapeutic agents that actively target and destroy malignant cells, has transformed treatment outcomes for patients with advanced hematological malignancies who have exhausted conventional therapeutic options. However, this remarkable success is accompanied by significant challenges, including limited efficacy in solid tumors and critical safety concerns. A comprehensive understanding of these achievements and obstacles is essential to guide the scientific and clinical priorities for developing next-generation CAR-T cell therapies (Dagar et al., 2023).

CAR-T cell therapy in hematological malignancies

CAR-T cell therapy has demonstrated transformative clinical efficacy in patients with relapsed/refractory B-cell malignancies. These therapies primarily target CD19, a transmembrane glycoprotein expressed on most B cell leukemias and lymphomas. Pivotal clinical trials have established clinically meaningful response rates and durability in historically treatment-refractory patient populations. In studies of pediatric and young adult patients with relapsed/refractory B cell acute lymphoblastic leukemia (ALL), anti-CD19 CAR-T therapy achieved complete remission rates exceeding 80% (Testa et al., 2024). In adults with relapsed/refractory diffuse large B-cell lymphoma (DLBCL), objective response rates ranged from 50-80%, with substantial proportions of patients maintaining complete responses for extended periods (Lyu et al., 2021). These clinical outcomes resulted in regulatory approval of the first CAR-T cell therapeutics. Tisagenlecleucel (Kymriah^®) and axicabtagene ciloleucel (Yescarta^®) have been incorporated into standard treatment algorithms for multiple B-cell malignancies (Cerrano et al., 2020; Ghafouri et al., 2021). The therapeutic platform has expanded to additional hematological malignancies, including the development of B-cell maturation antigen (BCMA)-targeting CAR-T therapies for multiple myeloma (Shah et al., 2020). These clinical successes have validated the CAR-T platform and catalyzed research into applications across diverse cancer types.

CAR-T cell therapy targeting solid tumors

CAR-T cell therapy has shown clinical success in hematologic malignancies, however, its effectiveness against solid tumors remains limited. This difference arises from fundamental biological differences between blood and solid cancers. The challenges in CAR-T therapy for solid tumors include target antigen identification, tumor trafficking and infiltration, and suppression by the TME (Fig. 2A). The selection of appropriate target antigens for CAR-T therapy presents a greater challenge for solid tumors compared to hematologic malignancies. Ideal targets should be abundantly and uniformly expressed on tumor cells, with minimal presence on normal tissues (Chen et al., 2024; Dagar et al., 2023; Marofi et al., 2021). While CD19 and BCMA serve as effective targets in B-cell and plasma cell malignancies, tumor-specific antigens are rarely found in solid tumors. Most tumor-associated antigens are overexpressed on cancer cells but also present on normal tissues (Friedman et al., 2018). Furthermore, solid tumors frequently exhibit significant antigen heterogeneity as target antigen expression varies widely among different cancer cell subpopulations. Consequently, CAR-T therapy directed at a single antigen may eradicate antigen-positive cells while leaving behind antigen-negative or low-expressing variants, leading to immune escape and disease relapse (Mishra et al., 2024; Wagner et al., 2020).

Fig. 2.

Current challenges in CAR-T cell therapy. (A) CAR-T cell therapy targeting solid tumors faces significant challenges, including tumor-specific antigen selection, physical barriers during trafficking and infiltration into tumor sites, and the immunosuppressive tumor microenvironment (TME). ECM, extracellular matrix; CAF, cancer-associated fibroblast; TAM, tumor-associated macrophages; TAN, tumor-associated neutrophils; MDSC, myeloid-derived suppressor cells; Treg, regulatory T cells. (B) CAR-T cell therapy is associated with severe safety concerns, including on-target off-tumor toxicity (OTOT), cytokine release syndrome (CRS), and immune effector cell-associated neurotoxicity syndrome (ICANS).

In addition, trafficking to and infiltrating solid tumors represent substantial barriers. Unlike hematologic cancers, where malignant cells circulate in accessible compartments, CAR-T cells should navigate from the systemic circulation to discrete tumor sites, extravasate through tumor vasculature, and penetrate dense stromal barriers comprising fibroblasts, extracellular matrix, and other structural components to reach target cells (Kalli et al., 2023; Kuczek et al., 2019; Sterner and Sterner, 2021). This migration process results in insufficient CAR-T cell accumulation within the tumor, thereby limiting their therapeutic efficacy. Even after successfully infiltrating the tumor, CAR-T cells encounter immunosuppressive tumor microenvironments. Solid TMEs employ multiple mechanisms to attenuate T cell function, including elevated expression of inhibitory checkpoint ligands such as PD-L1 on tumor and stromal cells, enrichment of immunosuppressive cell populations including regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs), and secretion of inhibitory mediators such as TGF-β and adenosine (Quail and Joyce, 2013; Rodriguez-Garcia et al., 2020). These immunosuppressive factors are compounded by metabolic constraints including hypoxia and nutrient depletion (Alvarado-Ortiz and Sarabia-Sa, 2025; Jain et al., 2014) that further compromise T cell proliferation, survival, and cytotoxic capacity.

Safety of CAR-T cell therapy

The potency of CAR-T cells drives their therapeutic efficacy, however, remarkable effectiveness is frequently offset by severe toxicity (Fig. 2B). A major safety concern is on-target, off-tumor (OTOT) toxicity, which arises when the target antigen is also expressed on healthy, non-malignant cells (Morgan et al., 2010; Richman et al., 2018; Thistlethwaite et al., 2017). This results in the unintended destruction of normal tissues alongside malignant cells. A well-documented example is B cell aplasia following anti-CD19 CAR-T therapy (Kalos et al., 2011). Since CD19 is expressed on healthy B cells, the treatment results in depletion of the entire B cell lineage, a condition manageable with lifelong immunoglobulin replacement therapy. However, such toxicity poses a significant barrier in solid tumors, where tumor-associated antigens often have detectable expression in essential organs, such as the lung, liver, and heart, raising the potential for fatal collateral damage.

Cytokine release syndrome (CRS) represents the most common and extensively characterized acute toxicity associated with CAR-T cell therapy. CRS is triggered by the rapid and massive release of pro-inflammatory cytokines by activated CAR-T cells and other recruited immune cells such as macrophages. Interaction with a large tumor burden generates a robust activation signal leading to exponential increases in cytokines such as IFN-γ, TNF-α, and notably IL-6 (Brentjens et al., 2010; Morgan et al., 2010; Sheth and Gauthier, 2021; Singh et al., 2017). Clinical manifestations of CRS range from mild flu-like symptoms to severe, life-threatening conditions characterized by hypotension, capillary leak, hypoxia, and multi-organ dysfunction. The severity of CRS strongly correlates with tumor burden at infusion, as larger tumor loads produce greater CAR-T cell activation and cytokine release. Management of severe CRS typically involves supportive administration of tocilizumab, an IL-6 receptor antagonist that can rapidly reduce symptoms (Kotch et al., 2019).

Immune effector cell-associated neurotoxicity syndrome (ICANS) is a serious neurological toxicity associated with CRS. The symptoms of ICANS vary widely, including confusion, aphasia, tremors, seizures, and in severe cases, fatal cerebral edema. Although the pathophysiology remains under investigation, current evidence suggests that inflammatory cytokines disrupt the blood-brain barrier, resulting in endothelial activation and infiltration of immune cells into the central nervous system, contributing to neurotoxicity (Hunter and Jacobson, 2019; Sheth and Gauthier, 2021; Velasco et al., 2023).

CAR MOLECULAR ARCHITECTURE AND ENGINEERING APPROACHES

Effective CAR-T cells are required to exhibit high potency and persistence to overcome the immunosuppressive barriers within the TME, while high potency substantially elevates the risk of CRS, ICANS as well as OTOT toxicity (Hunter and Jacobson, 2019; Sheth and Gauthier, 2021). Thus, it is challenging to develop effective and safe CAR-T therapies, particularly for solid tumors. The functionality of CAR-T cells depends on their synthetic receptor, CAR, which is engineered to redirect T cell specificity and enhance effector functions. Therefore, optimal CAR engineering is crucial for successful CAR-T cell therapy development. This section provides an overview of CAR molecular architecture and conventional approaches for the CAR constructs.

Molecular architecture of CAR

The basic structure of CAR is composed of an scFv domain that specifically binds to tumor-associated antigens and signaling domains for T cell activation (Fig. 1B). The extracellular scFv domain is connected to intracellular signaling domains through hinge and transmembrane domains (Rafiq et al., 2020). Upon encountering tumor cells, the scFv domain binds target antigens and triggers TCR signaling through the CAR signaling domains. This activation prompts CAR-T cells to release cytotoxic granules containing perforin and granzymes, which permeabilize target cell membranes and induce apoptosis. They also secrete pro-inflammatory cytokines including IFN-γ and TNF-α that amplify anti-tumor responses and recruit additional immune cells (Hu et al., 2024).

scFv domain: The extracellular scFv domain has been the primary focus of CAR development due to its fundamental role in determining tumor specificity and therapeutic potency. A single-chain variable fragment is an engineered fusion protein composed of antibody variable heavy and light chains connected by a short flexible peptide linker. scFv maintains antigen-binding specificity despite its compact size relative to a whole antibody. The scFv sequence determines the specificity for target antigen (Chang and Chen, 2017; Dotti et al., 2014). In addition, the scFv affinity is a critical determinant of CAR-T cell potency. High-affinity interactions generally enhance CAR activation and its functionality. However, excessively high antigen-binding affinity of scFv can mitigate the serial-killing capacity of CAR-T cells (Mao et al., 2022). It may also increase OTOT toxicity due to the recognition of low-level antigens expressed in normal tissues (Duan et al., 2021; Richman et al., 2018). Thus, it is essential to fine-tune the affinity of scFvs for CAR-T cell functionality. Additionally, the epitopes targeted by scFvs can be crucial for CAR-T cell function. For example, several scFvs target mesothelin (MSLN), a well-known solid tumor antigen, but each binds to a different epitope (Tomar et al., 2022; Zhang et al., 2015). Because the extracellular region of MSLN is known to undergo shedding within the tumor microenvironment (Liu et al., 2020), it is important to select the scFv with an epitope in the membrane-proximal region of MSLN. Therefore, both intrinsic antigen characteristics and the targeted epitope are critical considerations for identifying the optimal scFv sequence for successful CAR-T therapy.

Hinge domain: The hinge or spacer region serves as a flexible linker that determines the distance between CAR-T and target cells, directly affecting the formation of the immunological synapse (IS) critical for the CAR function and therapeutic efficacy. Optimal hinge length and structure are essential for effective scFv binding to target epitopes (Guest et al., 2005; Schafer et al., 2020). Inadequate length impairs antigen binding, whereas excessive length disrupts efficient IS formation. Current FDA-approved CAR-T therapies employ the hinge domains derived from CD8α, CD28, or IgG4, yet this limited repertoire contrasts with the diverse target antigens (Cappell and Kochenderfer, 2023; Guedan et al., 2019). Antigen-specific hinge optimization therefore represents a promising strategy for improved CAR design.

Transmembrane domain: The transmembrane (TM) domain serves to anchor CARs in the T cell membrane. Early CAR constructs employed the CD3ζ-derived TM domain, which was subsequently discovered to undergo unintended heterodimerization with endogenous TCR components. This prompted a shift toward CD8α- and CD28-TM sequences (Jayaraman et al., 2020; Sterner and Sterner, 2021). However, the CD28-TM containing CARs can heterodimerize with endogenous CD28 or form homodimers, eliciting potent T cell activation but potentially accelerating T cell exhaustion (Leddon et al., 2020; Muller et al., 2021). Notably, TM domains function beyond simple membrane anchoring, modulating CAR surface expression, spatial organization, and interactions with endogenous signaling machinery, features that position them as pivotal determinants of CAR performance.

Intracellular signaling domain: The intracellular signaling domain serves as the engine of CAR-T cells, translating antigen binding into T cell activation. The primary signaling component is the CD3ζ domain derived from the TCR, which contains 3 immunoreceptor tyrosine-based activation motifs (ITAMs). When the scFv domain of a CAR recognizes the target antigen on tumor cells, Lck phosphorylates the CD3ζ-ITAMs and the subsequent binding of ZAP70 facilitates the activation of T cells (Larson and Maus, 2021). Alternative signaling domain strategies involve incorporating downstream CD3ζ effectors, for example LAT and SLP76 as primary signaling components (Tousley et al., 2023).

Co-stimulatory domain: First-generation CAR contained only CD3ζ-ITAMs in its intracellular signaling domain but exhibited poor proliferation and short survival in vivo, resulting in clinical failure (Asmamaw Dejenie et al., 2022; Lindner et al., 2020). The breakthrough came with second-generation CARs, which incorporated co-stimulatory domains, such as CD28 or CD137 (4-1BB), alongside the intracellular signaling domain CD3ζ-ITAMs. This additional signal dramatically enhances T cell activation, proliferation, cytokine production, and long-term persistence of CAR-T cells, enabling the development of FDA-approved CAR-T therapeutics. Notably, co-stimulatory domains differentially influence signaling pathways: CD28-based CAR-T cells proliferate rapidly and produce abundant Th1 cytokines, whereas 4-1BB-based CAR-T cells exhibit delayed responses but demonstrate reduced T cell exhaustion (Long et al., 2015; Salter et al., 2018). Current investigations evaluate alternative costimulatory molecules such as ICOS, OX40, and further optimization of signaling domain combinations will improve the therapeutic efficacy of CAR-T cell therapy (Guedan et al., 2014, 2018; Tan et al., 2022).

Conventional CAR engineering approaches

The extracellular scFv domain has been the primary focus of CAR development due to its fundamental role in determining tumor specificity and therapeutic potency. Conventional scFv screening approaches include bacterial or yeast display techniques to isolate antibody clones that bind to target antigens with high affinity (Jaroszewicz et al., 2022). Bacterial or yeast display techniques present scFv fragments on the bacterial or yeast surface where they can be screened against target antigens through iterative rounds of binding selection and amplification. This approach allows for the generation of large diverse scFv libraries and enables the identification of high-affinity binders through techniques such as fluorescence-activated cell sorting (FACS) or magnetic bead selection.

Following initial screening, the selected scFvs are evaluated using surface plasmon resonance (SPR) or biolayer interferometry (BLI) (Duan et al., 2021; Sarker et al., 2022). SPR is a label-free optical technique that measures real-time binding kinetics by detecting changes in the refractive index near a sensor surface when biomolecules bind. This method provides quantitative data on binding affinity (K_D), association rates (k_on), and dissociation rates (k_off), thus it is an important method to characterize the scFv-antigen interactions. BLI is another label-free optical biosensing technique that measures biolayer thickness changes on fiber optic biosensors when target molecules bind to immobilized ligands. BLI offers advantages including faster analysis times, reduced sample consumption, and the ability to measure crude samples without extensive purification.

These in vitro screening methods efficiently identify scFvs with high binding affinity to target proteins. However, the purified proteins may not maintain their native conformation as found on cell membranes (Le et al., 2024; Mahboob et al., 2025; Wessel et al., 2020). The epitopes identified during in vitro screening can be occluded or altered at the cell membrane interface, possibly due to membrane constraints, glycosylation patterns, or protein-protein interactions. Additionally, in vitro experimental conditions differ substantially from complex cellular environments, including the presence of competing membrane proteins, glycocalyx components, and dynamic membrane topology that can influence antigen accessibility and presentation (Watanabe et al., 2019). Furthermore, high binding affinity of scFvs measured in vitro does not always correlate with optimal CAR-T cell function. Excessively high-affinity interactions may impair CAR-T cell serial killing capacity, reduce trafficking efficiency, or promote activation-induced cell death (Duan et al., 2021; Mao et al., 2022; Richman et al., 2018). Consequently, conventional screening approaches often fail to predict CAR-T cell therapeutic outcomes, highlighting the critical need for innovative cell-based screening platforms that better recapitulate the native tumor microenvironment and can assess functional CAR-T cell responses rather than binding affinity alone.

ADVANCED PLATFORMS FOR CAR ENGINEERING AND OPTIMIZATION

As discussed in the previous section, CAR engineering has been primarily focused on identifying high-affinity scFvs through in vitro screening methods. While these traditional approaches enable high-throughput scFv screening, incorporating the complex cellular environment into the screening process is necessary to more effectively identify functionally relevant scFvs for CAR-T cell therapy. Additionally, CARs contain multiple modules beyond the scFv domain, and combinations of the modules can critically influence the specificity, sensitivity, and therapeutic efficacy of CAR-T cells. However, most CARs currently in development utilize a restricted set of molecules for these modules, primarily CD8α, CD28, 4-1BB, and CD3ζ. This limited molecular repertoire reflects the low-throughput nature of traditional CAR development, leaving an extensive design space largely unexplored. These limitations have driven the development of novel cell-based screening strategies that enable the discovery of more potent, persistent, and safer CAR architectures. This section reviews recent advances in cell-based platforms for comprehensive CAR screening and optimization (Table 1).

Table 1.

Characteristics of cell-based CAR screening methods

	Screening method	Principle	Readout	Test system	Advantages	Limitations	References
Antigen Binding Properties	Malibu-Glo assay	scFv binding affinity	Luminescence (Fig. 3A, upper panel)	scFv-luciferase fusion protein	Cell-based evaluation of scFv binding affinity	Difficult to predict the CAR-T functions; Protein purification required	Natarajan et al., 2020
	SynNotch system	scFv binding avidity	Reporter gene expression (Fig. 3A, lower panel)	HEK293F cell	Cell-based evaluation of scFv binding avidity	Difficult to predict CAR-T functions	Ma et al., 2021
T Cell Activation Reporter Cell Lines	CAR-J assay IL-2 promotor system NF-κB, NFAT reporter cell line Triple parameter reporter (TPR) cell line	T cell activation markers	CD69 level (Fig. 3B, upper panel) IL-2, Reporter gene expression (Fig. 3B, middle panel) Reporter gene expression (Fig. 3B, lower panel)	Human T lymphocyte Jurkat cell Murine T hybridoma B3Z cell Human T lymphocyte Jurkat cell	Rapid Measurement of T cell activation markers	Detection of end-point T cell activation markers susceptible to diverse factors beyond CAR activation; Lack of full physiological context of primary T cells	Bloemberg et al., 2020 Di Roberto et al., 2020 Rydzek et al., 2019 Jutz et al., 20166
Antigen-induced CAR Activation	FRET-based CAR-activation sensor	Antigen binding induced CAR activation	FRET/CFP emission ratio (Fig. 3C)	HEK293A, Human T lymphocyte Jurkat cell	Most direct measurement of antigen binding-induced CAR activation; Real-time measurement of Immunological synapse formation	Lack of full physiological context of primary T cells	Lee et al., 2025
CAR Library Screening in Primary T cells	Pooled screening of primary CAR-T cells	T cell activation markers	T cell activation markers (e.g. CD69) (Fig. 4)	Human primary T cell	Most physiologically relevant system	Detection of end-point T cell activation markers susceptible to diverse factors beyond CAR activation	Castellanos-Rueda et al., 2022; Goodman et al., 2022; Gordon et al., 2025; Rios et al., 2023

Cell-based CAR screening systems measuring antigen binding properties

To overcome the limitations of in vitro screening methods, several cell-based scFv screening systems have been developed (Fig. 3A). One method named Malibu-Glo involves fusing an scFv with luciferase, treating cells with this fusion protein, and assessing scFv binding affinity through luminescence intensity (Natarajan et al., 2020). This approach provides a rapid, simple, and sensitive method for measuring the antigen binding affinity of scFvs in a cellular context. Another strategy employs co-culture of CAR-expressing cells with antigen-expressing cells, followed by flow cytometry analysis to determine the proportion of CAR-positive doublets, thereby evaluating binding capacity to target cells (Shepherd et al., 2024).

Fig. 3.

Principles of cell-based platforms for CAR engineering and optimization. (A) Cell-based CAR screening systems measuring antigen binding properties. Upper panel: Luciferase-fused scFvs are co-cultured with cancer cells and the measured luminescence intensity correlates with the scFv affinity. Lower panel: scFv-containing SynNotch system is used to assess the scFv avidity. Upon engagement with cancer cells, mechanical tension induces proteolytic cleavage of the SynNotch receptor, releasing an intracellular transcriptional activator that drives the expression of a fluorescent protein. Thus, fluorescence intensity correlates with the scFv avidity. (B) Cell-based CAR screening systems based on T cell activation reporter cell lines. Upper panel: Upon co-culture with cancer cells, the CD69 expression level in CAR-expressed Jurkat cells is measured by flow cytometry. Middle panel: For cytokine-based readouts, a self-cleaving P2A peptide and a GFP open reading frame were inserted downstream of the final exon of the endogenous IL-2 gene in B3Z cells, enabling GFP expression following CAR activation-induced IL-2 production. Lower panel: Jurkat cell line expressing response elements for T cell transcription factors (NF-κB, NFAT, AP-1), a minimal promoter, and fluorescent proteins, is used to evaluate the CAR-induced transcription factor activation by fluorescent intensity. (C) CAR optimization system evaluating antigen-induced CAR activation. A FRET-based immunological synapse biosensor system consists of CAR libraries with various scFvs, CFP and YFP separated by an optimized linker, and ZAP70-SH2 domains. Upon binding to target antigens, the CAR-ITAMs in the biosensor are phosphorylated and subsequently recruit ZAP70-SH2 domains, increasing the FRET signal between CFP and YFP. Thus, live-cell FRET measurement enables the detection of antigen-induced CAR activation at the immunological synapse.

Additionally, researchers have incorporated a synthetic Notch (SynNotch) system to measure the antigen-binding property of scFvs (Ma et al., 2021; Morsut et al., 2016). Notch is a transmembrane protein, which is activated by mechanical forces generated by cell-cell interactions (Gordon et al., 2007; Mumm et al., 2000). Its engineered variant, the SynNotch receptor, is composed of an extracellular recognition domain, e.g. scFv, a Notch core domain including a protease cleavage site, and an intracellular transcriptional activator domain. Upon binding target antigen via an scFv, mechanical tension induces proteolytic cleavage of the SynNotch receptor, liberating the intracellular transcriptional activator that drives expression of desired genes, for example, a fluorescent protein in the responding cell. The subsequent FACS analysis thereby allows identifying scFvs that selectively bind to the target cells with high avidity.

While these methods serve as useful tools for assessing the binding capacity of scFvs to target antigens, the selected scFvs do not necessarily correlate with optimal CAR-T cell functionality when incorporated into cell-based therapeutics. Indeed, the functional behavior of CAR-T cells depends on multiple factors beyond scFv affinity, including other biophysical properties of scFv as well as intracellular CAR domains, making subsequent validation in primary T cells indispensable (Ma et al., 2021; Natarajan et al., 2020; Shepherd et al., 2024). Furthermore, multiple studies have reported that high-affinity scFvs can increase OTOT toxicity, rendering them both ineffective and potentially dangerous (Caruso et al., 2015; Drent et al., 2017; Liu et al., 2015).

CAR screening platforms based on T cell activation reporter cell lines

As scFv antigen binding properties cannot accurately predict optimal CAR-T cell functions, cell-based screening platforms have been developed to directly assess CAR-induced T cell responses and applied to screen effective CAR candidates (Fig. 3B). These platforms employ engineered reporter cell lines to provide fast, reproducible, and quantitative readouts of CAR activation. Among these, a human T lymphocyte Jurkat cell line has become particularly prominent due to its conserved T cell signaling machinery, such as CD69 (Cibrian and Sanchez-Madrid, 2017) and IL-2 (Fernandez-Riejos et al., 2008; Schneider et al., 1977; Schwenk and Schneider, 1975), making it an effective surrogate for primary T cell activation. Thus, the Jurkat cell has become the most employed cell line in CAR-T therapy research (Bloemberg et al., 2020; Duong et al., 2013; Gordon et al., 2022).

CAR-J assay is a Jurkat-based system utilizing flow-cytometric detection of the surface activation marker CD69 to assess CAR-induced T cell activation (Bloemberg et al., 2020). This assay is particularly useful for rapid screening and short-listing novel scFv candidates and for identifying CARs with deleterious auto-activation or tonic signaling. A study demonstrated the predictive power of the CAR-J assay by identifying anti-EGFRvIII CARs that showed tonic activation in Jurkat cells. The same CARs, when expressed in primary T cells, later caused target-independent proliferation, an effector phenotype, and progressive loss of target-specific response, demonstrating the clinical relevance of the assay for initial quality control.

Another CAR screening system is based on the detection of IL-2 production, a reliable marker of T cell activation (Di Roberto et al., 2020). In this system, a self-cleaving P2A peptide and a GFP open reading frame (ORF) were integrated immediately downstream of the final exon of the endogenous IL-2 gene in the murine T hybridoma cell line B3Z. The P2A peptide ensures that GFP remains cytoplasmic rather than being fused to and secreted with IL-2. Consequently, when CAR-expressing B3Z reporter cells are activated by binding to target cells, they produce IL-2 and simultaneously express cytoplasmic GFP. This enables high-throughput FACS-based screening, as the fraction of GFP-positive cells directly correlates with IL-2 secretion levels upon stimulation.

CAR-induced T cell activation has also been assessed through transcription factors activation, such as NF-κB, NFAT, and AP-1 (Crabtree and Clipstone, 1994; Jain et al., 1992; Li and Verma, 2002; Macian, 2005; Shaulian and Karin, 2002). One such platform utilizes a Jurkat reporter cell line equipped with nuclear factor κB (NF-κB)-ECFP and nuclear factor of activated T cells (NFAT)-EGFP reporter genes (Rydzek et al., 2019), which are composed of response elements of NF-κB or NFAT, a minimal promoter, and ECFP or EGFP, respectively. This reporter system allows the detection of CAR-induced transcription factor activation via ECFP and EGFP fluorescent signals, providing a simple, quantitative, and reproducible readout of T cell activation. Its high-throughput potential was demonstrated in a large-scale screening of a ROR1-CAR library, where it successfully identified a functional CAR variant with a frequency of only 6 in 1.05×10⁶ variants. This discovery would be prohibitively time-consuming with traditional methods using primary T cells. Furthermore, the triple parameter reporter (TPR) cell line, which incorporates NF-κB-CFP, NFAT-EGFP, and AP-1-mCherry reporter genes into Jurkat cells (Jutz et al., 2016), enables simultaneous monitoring of three major transcription factors for effective CAR screening.

These cell-based platforms measuring T cell activation markers, such as CD69, IL-2, transcription factors, provide better prediction compared to accessing scFv affinity alone. However, these markers can be susceptible to diverse factors beyond CAR activation.

CAR optimization system evaluating antigen-induced CAR activation

The reporter cell lines detecting T cell activation markers are powerful CAR screening platforms, but they can be influenced by a wide range of factors besides antigen recognition. This can complicate the discrimination between activation resulting from specific CAR-antigen interactions versus non-specific responses. To address these limitations, a fluorescent resonance energy transfer (FRET)-based biosensor has been developed to provide more direct readouts of antigen-induced CAR activation (Lee et al., 2025) (Fig. 3C). This FRET-based immunological synapse biosensor incorporates CAR libraries with various scFvs, CFP and YFP separated by an optimized linker, and ZAP70-SH2 domains. Upon binding to target antigens on tumor cells, the ITAMs in the biosensor are phosphorylated and subsequently recruit ZAP70-SH2 domains, increasing the FRET signal between CFP and YFP. Live-cell FRET measurement therefore enables the real-time monitoring of antigen-induced local CAR activation at the IS.

This approach offers greater specificity than conventional methods because the FRET signal increases only when CAR-ITAMs are phosphorylated and recruit ZAP70-SH2 domains, events that occur specifically upon antigen binding and CAR clustering at the immunological synapse. This direct coupling between antigen engagement and signal output provides more accurate predictions of CAR therapeutic potency than methods measuring only downstream activation markers. This distinction proved critical in this study, demonstrating that a moderate-affinity scFv variant produced superior CAR-T cell responses despite the weaker antigen binding property compared to a high-affinity variant (Lee et al., 2025). The high-affinity CAR was prone to internalization and trogocytosis, a process in which T cells extract antigen from target cells, disrupting stable IS formation and impairing function. These findings demonstrate that lower binding affinity CARs can be highly effective if they promote stable and robust immunological synapses.

Consequently, screening systems that assess IS quality and CAR activation at the synapse, rather than relying solely on antigen binding affinity or general T cell activation markers, are essential for designing next-generation CARs that are both potent and safe. The study suggested a link between the scFv binding property, CAR activation level, immunological synapse quality, and the resulting CAR-T cell functionality. For high-throughput screening, the FRET-based assay was performed in conventional cell lines.

CAR library screening in primary T cells

CAR screening platforms utilizing reporter cell lines offer significant advantages in speed and scalability. However, their primary limitation is the lack of full physiological context. Cell lines do not possess the full cytokine repertoire, cytotoxic capacity, and complex differentiation pathways of T cells (Colin-York et al., 2019; Wilson et al., 2025). These limit their predictive ability to predict CAR-T cell function in vivo. To bridge the gap between in vitro discovery and clinical relevance, researchers have developed methodologies for screening CAR libraries directly in primary human T cells. This involves the simultaneous, pooled screening of CARs, followed by the identification of successful variants by next-generation sequencing (NGS) utilizing barcoded primers (Fig. 4). A significant technical advancement in this area is Golden Gate assembly using Type IIS restriction cloning (Castellanos-Rueda et al., 2022; Goodman et al., 2022; Gordon et al., 2025), which enforces ordered and directional assembly across diverse CAR domains, and blunt-ligation-based cloning strategy (Rios et al., 2023) for serial barcoded DNA assembly. These strategies facilitate the efficient generation of diverse CAR libraries through the fusion of domain libraries. Using such approaches, different research groups have constructed CAR libraries ranging from as few as 40 variants to, theoretically, up to 1.3×10⁶ (Castellanos-Rueda et al., 2022; Goodman et al., 2022; Gordon et al., 2025; Rios et al., 2023). Researchers have demonstrated the accuracy of CAR construct assembly and high reproducibility of CD69-based activation measurements (R²=0.96), thereby validating both the library construction and the screening methodology (Rios et al., 2023).

Fig. 4.

CAR library screening in primary T cells. To generate a pooled CAR-T cell population, a CAR library is introduced into human primary T cells via viral transduction. Activated CAR-T cells upregulating various T cell markers are screened and enriched by FACS-based sorting. The sorted CAR-T cells are lysed, and their CAR sequences are identified by next-generation sequencing using barcoded primers. This approach enables the identification of functionally superior CARs in physiologically relevant human primary T cells.

In this screening of pooled CAR-T cells, CD69 serves as a representative activation marker for evaluating CAR functionality. Additional assays are performed to assess various features of CAR-T cells, such as expansion and proliferation, cytokine production, cytotoxicity (Castellanos-Rueda et al., 2022; Goodman et al., 2022; Gordon et al., 2025; Rios et al., 2023). For example, CAR-T expansion and proliferation are typically determined by the dilution of CellTrace Violet (CTV). Cytokine secretion levels such as IL-2 and IFN-γ are quantified. Cytotoxicity was evaluated by measuring CD107a or granzyme expression using flow cytometry. As persistence is a major determinant of therapeutic efficacy of CAR-T cells, many studies also predict the persistence by measuring exhaustion markers and memory phenotype markers. Therefore, the pooled CAR-T cell libraries are screened for a range of clinically relevant T cell functions induced by CAR activation, moving beyond simple activation markers like CD69.

To correlate CAR-T cell functionality with CAR construct composition and sequence, the DNA barcode concept was introduced, a key technology for large-scale screening. The DNA barcodes are randomized nucleotide sequences approximately 6-18 base pairs in length, which can be incorporated during CAR library construction or tagged through primers during the amplification of CAR-T cell-derived genes. These platforms extend NGS-based pooled screening to include long-term phenotypes such as persistence and memory features, which are considered essential for a durable anti-tumor response in solid tumors (Goodman et al., 2022; Gordon et al., 2025). A screen of a 1.3×10⁶-member CAR library for glioblastoma (Gordon et al., 2025), for example, successfully identified a novel CAR that showed potent cytotoxicity, and importantly, enhanced persistence in a xenograft model, a key feature that a short-term activation screen would have missed. This demonstrates the value of screening for multiple, long-term phenotypes as a more reliable predictor of a CAR’s in vivo efficacy than short-term activation alone. Furthermore, such pooled library screening enables the assessment of the functional potential of scFv and diverse domain types that cannot be practically evaluated using traditional arrayed assays. Primary T-cell–based CAR library screening offers the greatest potential for clinically predictive outcomes, although the molecular mechanisms underlying the functional effects of the newly identified domains require subsequent investigation.

CONCLUSION AND PERSPECTIVES

Cancer immunotherapy has fundamentally shifted the treatment landscape in oncology. Among these approaches, CAR-T cell therapy stands out as a powerful method that engineers a patient’s own T cells into potent living therapeutics. This has resulted in remarkable clinical success for patients with relapsed/refractory hematological malignancies. However, several challenges limit its broader applications, including low efficacy in solid tumors and safety concerns. The functionality of CAR-T cell therapy depends on its synthetic receptor, CAR, thus advancing CAR design and engineering is crucial for successful therapeutic outcomes.

Traditional CAR screening methods, which focus on in vitro binding affinity, often fail to predict clinical outcomes. Cell-based scFv screening methods such as the SynNotch system were developed to address this limitation, however selected high-affinity scFvs still do not necessarily correlate with CAR-T functionality. Cell-based platforms assessing T cell activation markers (e.g. CD69, IL-2, transcription factors), including the CAR-J assay and TPR cell line, provide better prediction but remain susceptible to diverse factors beyond CAR activation. The FRET-based biosensor system was therefore developed to directly measure CAR activation upon antigen binding, providing the most direct readout currently available for predicting CAR functionality. The FRET-based IS biosensor also holds promise for fine-tuning CAR therapy as a personalized medicine approach. By testing a library of CARs in a patient’s own primary T cells, it could help identify the most effective CAR construct for the specific tumor before infusion. This would enable a personalized medicine approach, or “fine-tuning” of the CAR for each patient, maximizing the treatment’s effectiveness before infusion.

These CAR screening methods still rely on cell lines, which lack the full physiological context of primary T cells, such as cytokine repertoire and cytotoxic capacity. This limits their predictive power, therefore researchers have started screening CAR libraries directly in primary T cells by assessing the functionality of pooled CAR-T cells and identifying successful variants by next-generation sequencing. Nevertheless, there is a further need to advance the CAR screening principles that can be performed directly on primary T cells in a high-throughput manner to more accurately predict CAR-T cell function in a clinical setting. For example, the FRET-based assay currently performed in conventional cell lines for high-throughput screening, can be further applied in primary T cells to enhance clinical predictability. Looking ahead, we expect the development of diverse fluorescence-based biosensors, enabling direct monitoring of CAR activation at the molecular level in primary T cells. These technologies would allow investigators to visualize the activation of downstream signaling molecules, assess the directionality and magnitude of T cell signaling, and ultimately evaluate differentiation driven by CAR activation.

While most cell-based CAR screening methods identify optimal CAR constructs by measuring consequential or end-point functional outcomes, it is equally critical to investigate the molecular mechanisms triggered by individual CAR elements. Indeed, every component for CAR activation, including scFv affinity, epitope, hinge length, transmembrane domain, co-stimulatory domains, and signaling motifs, can differentially shape CAR-T cell signaling. Emerging high-resolution scFv screening technologies such as the SynNotch system, FRET-based visualization of initial CAR activation process, and integrative structure-function modeling can provide mechanistic insights into CAR functions. These insights would provide a strong rationale for refining CAR design and engineering strategies aimed at precisely tuning CAR-T cell signaling.

Current CAR screening methods primarily focus on basic metrics such as antigen binding properties and initial T cell activation. This simplistic approach fails to capture the nuanced and diverse functions of co-stimulatory domains, which are crucial for durable T cell persistence and long-term efficacy. To address these limitations, it would be valuable to develop innovative screening systems that move beyond fundamental CAR functionality and enable more comprehensive, clinically relevant evaluation. Another limitation is the lack of screening systems specifically designed to assess the ability of CAR-T cells to overcome the challenges induced by the immunosuppressive TME of solid tumors. Such systems would ideally assess T-cell performance under conditions of chronic antigen stimulation, hypoxia, low pH, or in the presence of suppressive cytokines such as TGF-β. Although efforts have been made to identify co-stimulatory molecules that prevent T cell exhaustion and promote memory-like differentiation under chronic antigen exposure, attempts to address the broader TME challenges remain limited. Future platforms must integrate components that recapitulate this hostile environment more comprehensively.

Therefore, it is necessary to further improve CAR screening platforms that can evaluate advanced engineering efforts, including the incorporation of logic gates to enhance tumor specificity, the constitutive or inducible expression of supplemental cytokines and chemokines to reshape the TME, and the integration of robust safety features to mitigate potential off-target toxicities. By adopting these refined screening principles, we can more effectively identify and advance next-generation CAR-T cell therapies with a higher probability of success against solid tumors. This continued innovation is essential for developing more effective and safer CAR-T cell therapies, potentially making CAR-T cell therapy a game-changing cancer immunotherapy for solid tumors as well.