Genome-edited allogeneic CAR-T cells: the next generation of cancer immunotherapies

Abstract

Chimeric Antigen Receptor T (CAR-T) cell therapy has revolutionized cancer immunotherapy, particularly in hematological malignancies. However, the clinical application of autologous CAR-T cells faces significant high cost and manufacturing challenges. Universal allogeneic CAR-T cells, derived from healthy donors, represent a promising solution to these obstacles. These “off-the-shelf” therapies aim to reduce the complexity and cost of CAR-T production. Despite exciting advancements in genome-editing technologies and promising clinical trial data, significant challenges remain, including graft-versus-host disease (GVHD), Host-versus-graft reaction (HVGR), off-target effects, genotoxicity, and manufacturing scalability. To address these concerns, genome-editing technologies such as ZFNs, TALENs, Meganucleases, CRISPR systems, base editing, and prime editing are being employed. This review summarizes the progress of universal allogeneic CAR-T cell therapies, addresses the critical challenges, and discusses the future directions for their clinical implementation.

Background

Chimeric Antigen Receptor T (CAR-T) cell therapy has emerged as a groundbreaking treatment in cancer immunotherapy, demonstrating remarkable clinical success, especially in hematologic malignancies such as acute lymphoblastic leukemia (ALL) and diffuse large B-cell lymphoma (DLBCL) [1]. This success has prompted its expansion into other cancers, with autologous CAR-T therapies in which patients’ own T cells are engineered to target and kill tumor cells showing durable remissions in various malignancies. Up to now, all CAR-T cell therapies approved by the Food and Drug Administration (FDA) and the National Medical Products Administration (NMPA) are directed to treat relapsed/refractory (R/R) hematological malignancies (Table 1). The commercialised CAR-T cell products are generally based on second-generation CARs, employing CD28 or 4-1BB as co-stimulatory domain fused to the CD3ζ signalling domain for T cell activation and facilitating the expansion and persistence of the genetically engineered cells in vivo [2]. Despite the potential of autologous CAR-T cell therapies, the process is marred by significant limitations. The need for individualized preparation, the extended timeframes for manufacturing, and the high costs involved make it a resource-intensive treatment [3–5]. Moreover, patients with insufficient T cell reserves, such as those who are heavily immunocompromised or who have T cell exhaustion, may be ineligible for autologous CAR-T therapy, thus limiting the overall accessibility of this innovative treatment [5].

Table 1.

Autologous CAR-T cell products on the market

Ratification	Trade name	Proper name	Target	Manufacturer	Indication	Date
FDA	Kymriah	Tisagenlecleucel	CD19	Novartis	R/R B-ALL	Aug. 2017
					R/R LBCL	May. 2018
					R/R FL	May. 2022
	Yescarta	Axicabtagene Ciloleucel	CD19	Kite Pharma	R/R LBCL	Oct. 2017
					R/R FL	Mar. 2021
	Tecartus	Brexucabtagene Autoleucel	CD19	Kite Pharma	R/R MCL	Jul. 2020
					R/R B-ALL	Oct. 2021
	Breyanzi	Lisocabtagene Marleucel	CD19	Juno Therapeutics	R/R LBCL	Feb. 2021
	Abecma	Idecabtagene Vicleucel	BCMA	Bristol Myers Squibb	R/R MM	Mar. 2021
	Carvykti	Ciltacabtagene Autoleucel	BCMA	Legend biotech	R/R MM	Feb. 2022
	Aucatzyl	obecabtagene autoleucel	CD19	Autolus Therapeutics	R/R B-ALL	Nov. 2024
NMPA	Yescarta	Axicabtagene Ciloleucel	CD19	Fosun Kite	R/R LBCL	Jun. 2021
	Carteyva	Relmacabtagene Autoleucel	CD19	JW Therapeutics	R/R LBCL	Sep. 2021
	Fucaso	Equecabtagene Autoleucel	BCMA	IASO Biotherapeutics	R/R MM	Jun. 2023
	Yuanruida	Inaticabtagene Autoleucel	CD19	Juventas	R/R B-ALL	Nov. 2023
	Saikaize	Zevorcabtagene Autoleucel	BCMA	CARsgen	R/R MM	Mar. 2024
	Carvykti	Ciltacabtagene Autoleucel	BCMA	Legend biotech	R/R MM	Aug. 2024

FDA, Food and Drug Administration; NMPA, National Medical Products Administration; R/R, relapsed/refractory; B-ALL, B-cell acute lymphoblastic leukemia; LBCL, large B-cell lymphoma; FL, follicular lymphoma; MCL, mantle cell lymphoma; MM, multiple myeloma

The development of universal allogeneic CAR-T cells, offering a promising solution to these challenges, involves a specific manufacturing process and mechanism (Fig. 1). By utilizing T cells from healthy donors, these “off-the-shelf” therapies can be produced in bulk, overcoming the need for patient-specific customization. By means of the establishment of “off-the-shelf” universal cell banks and the utilization of gene editing techniques to surmount the obstacles correlated with human leukocyte antigen (HLA)-mismatched cell therapy, the perils of manufacturing delays or failures, as well as the accidental transduction of blasts can be eliminated [6, 7]. This enables faster treatment timelines and offers a more scalable, cost-effective solution for patients who may otherwise be unable to receive autologous therapy. The potential to offer CAR-T therapy without the constraints of individualized preparation significantly broadens its availability and offers the possibility of treating a much larger pool of patients across various cancer types.

Fig. 1.

Manufacturing process and mechanism of universal allogeneic CAR-T cells. Allogeneic αβT cells are predominantly sourced from T lymphocytes obtained from third-party healthy peripheral blood mononuclear cells (PBMCs), umbilical cord blood (UCB), and induced pluripotent stem cells (iPSCs). Subsequent to the activation of αβT cells with anti-CD3/CD28 beads, techniques such as viral vector-mediated transgenesis or gene knock-in mediated by gene editing permit the permanent integration of recombinant DNA encoding a CAR and potentially extra genetic modifications. Subsequently, the T cells are amplified through the utilization of anti-CD3/CD28 beads and cytokines. Following the magnetic removal of residual TCRαβ⁺ cells by means of a sorting system like the cliniMACS device, the resultant product is cryopreserved for subsequent utilization. scFv, single-chain variable fragment. Created with BioRender.com

However, despite the significant promise of universal allogeneic CAR-T cell therapies, several bottlenecks in effectiveness and safety must be addressed before these therapies can become mainstream. The Graft-versus-Host Disease (GVHD) and host-versus-graft reaction (HVGR) remain major hurdles, as the use of donor T cells in a recipient with a different genetic makeup can trigger severe immune responses [3]. Additionally, while gene editing technologies offer solutions for these challenges, they also introduce potential concerns around off-target effects and genotoxicity, which must be carefully managed to ensure patient safety [5, 8]. Even with advances in gene editing, the effectiveness of universal CAR-T cells is still limited by factors such as tumor heterogeneity, antigen escape, T-cell exhaustion, and the suppressive nature of the tumor microenvironment (TME) [5, 9].

This review systematically delineates recent advances in universal allogeneic CAR-T therapy, with a focused analysis on gene-editing tools, manufacturing innovations, and clinical trial outcomes. We examine key bottlenecks centered on GVHD and HVGR, alongside challenges in solid tumors including TME suppression and trafficking barriers, while synthesizing current research strategies to overcome these hurdles. Furthermore, we summarize the advantages and limitations of mainstream gene-editing platforms and elucidate monitoring and mitigation strategies for safety concerns arising from off-target effects, genotoxicity, and long-term risks. Additionally, we present a comparative analysis of the relative merits of different donor-derived allogeneic T-cell sources and subpopulations employed in gene editing, as well as the evolution of gene delivery systems. We provide a detailed exposition of clinical trials involving gene-edited allogeneic CAR-T cells for hematological malignancies, summarizing lessons learned regarding clinical efficacy. Representative autologous and allogeneic CAR-T products are comparatively analyzed for their clinical efficacy in solid tumors. Finally, we explore the prospects of in vivo CAR-T engineering, which offers potential for off-the-shelf, low-cost, and patient-specific CAR therapies without ex vivo manipulation.

Critical challenges in allogeneic CAR-T therapy

A major clinical challenge of allogeneic CAR-T cell therapy is the inherent treatment-related morbidity and mortality, with GVHD being the main contributing factor. Numerous potential risk factors for these complications have been established in the context of allogeneic hematopoietic stem cell transplantation (HSCT), including HLA disparity [10], different stem cell sources [11], sex disparity [12], and more intensive GVHD prophylaxis regimens [13]. Allogeneic T cells recognize HLA molecules on the surface of antigen-presenting cells (APCs), and HLA mismatch between recipient and donor is one of the most serious risk factors for both GVHD and HVGR [14]. Therefore, one key strategy for utilizing CAR-T cells derived from healthy donors involves incorporating additional genetic modifications during the manufacturing process or screening specific T-cell sources/subpopulations, as will be described further in subsequent sections.

GVHD

Graft-versus-host disease (GVHD) remains the primary challenge in allogeneic CAR-T therapy. It occurs when T cells from the graft recognize the host as foreign and launch attacks to eliminate cells carrying foreign antigens. Acute GVHD can lead to systemic rashes, fulminant liver injury, and various gastrointestinal symptoms [15, 16], while chronic GVHD is often characterized by organ fibrosis and is accompanied by a high risk of infection-related mortality [17, 18].

The primary strategy for mitigating GVHD is based on targeted disruption or modification of the alloreactive mediator T-cell receptor of αβT cells (TCRαβ). Alloreactive αβT cells can recognize multiple different MHC molecules (humans HLA system), which may be associated with high frequency of allorecognition [15], therefore, disrupting the surface expression of TCRαβ to prevent them from interacting with MHC molecules in peptide complexes may alleviate GVHD alloreactivity. The TCRαβ protein complex, which consists of an α chain and a β chain, is associated with auxiliary molecules such as CD3 proteins. GVHD is mediated by antigen-specific TCRαβ, thus targeting genes related to the expression of the TCRαβ/CD3 complex can disrupt its cell surface expression at the genomic level [5]. The T-cell receptor β chain gene contains two possible constant regions (TRBC), while there is only one gene encoding the T-cell receptor α chain constant (TRAC). Therefore, one approach to efficiently prevent TCRαβ expression is to knock out the TRAC gene or replace some TCR subunits that hinder its antigen recognition function [19, 20]. Alternatively, knocking out either the TRBC1 or TRBC2 locus leads to the interruption of the β chain. In either case, the residual TCRαβ⁺ cells are strictly depleted by magnetic bead depletion systems, a process that redirects the specificity of the T cell products through the CAR, and greatly alleviates alloreactivity. This approach, disrupting donor-derived TCRαβ-related genes combined with removing residual TCRαβ⁺ cells, efficiently relieves GVHD. In addition, similar disruption of TCRαβ, which utilizes RNA interference (RNAi) to limit the assembly of the multimeric receptor complex at the protein level, has been used as an alternative method to genome editing [21, 22]. Notably, it should be cautioned that TCR knockout may compromise the in vivo persistence and functionality of T cells. For example, CRISPR-edited cells with complete TCR ablation exhibit diminished survival capacity compared to those with residual TCR expression, a difference that is likely linked to the disruption of IL-7/IL-15-dependent survival signaling resulting from TCR deficiency [23]. Furthermore, following repeated antigen stimulation, TRAC-knockout CAR-T cells display reduced secretion of IFN-γ and TNF-α, along with an increased proportion of cells arrested in the G0/G1 phase of the cell cycle [24, 25]. Such functional exhaustion may be attributed to the impairment of the synergistic interplay between CAR and TCR signaling [26].

Additional alternative cell therapies (non-αβ T cell-based) have demonstrated significant clinical potential. Liu E et al. employed cord blood-derived CD19 CAR-NK cells to treat 11 patients with R/R non-Hodgkin lymphoma (NHL) or chronic lymphocytic leukemia (CLL). By capitalizing on NK cells’ intrinsic MHC-independent cytotoxic mechanisms, this approach naturally circumvents GVHD while achieving an impressive 73% overall response rate (ORR). With a median follow-up of 13.8 months, 7 patients (64%) attained complete remission (CR) [27]. Concurrently, double-negative T cells (DNTs; CD3⁺ CD4⁻ CD8⁻) have exhibited dual functionality in allogeneic HSCT settings, maintaining potent graft-versus-leukemia (GVL) effects while suppressing GVHD. Clinical data demonstrate that third-party DNT infusion post allogeneic HSCT effectively prevents relapse in high-risk acute myeloid leukemia (AML) patients. Remarkably, none of 6 treated patients developed DNT-associated GVHD attributable to their immunoregulatory function, with 66.7% (4 patients) maintaining long-term relapse-free survival (RFS) [28]. Other innovative strategies, including induced pluripotent stem cells (iPSCs)-derived therapies and selective T cell subset engineering for universal CAR-T production, show sustained therapeutic efficacy and will be elaborated subsequently. Of particular interest, emerging research reveals that stem cell infusion timing during allogeneic HSCT significantly impacts acute GVHD incidence. The differential immune microenvironment in recipients across circadian phases suggests that chronotherapy approaches (e.g., morning infusions timed to circadian rhythm of immune cells) may substantially mitigate GVHD severity [29].

Novel pharmacological strategies are expanding therapeutic options for GVHD. For GVHD prevention after allogeneic HSCT, small-molecule calcineurin inhibitors (CNIs), cyclosporine and tacrolimus, remain fundamental prophylactic agents alongside cellular approaches [30]. Their efficacy against acute GVHD stems from inhibiting T-cell activation, specifically by blocking calcineurin-dependent NFAT signaling and preventing Lck-S59 dephosphorylation [31]. However, this broad immunosuppression non-selectively dampens immune responses, potentially compromising the efficacy of co-administered adoptive cell therapies (ACTs) like CAR-T cells. Furthermore, the immunomodulator antithymocyte globulin (ATG), when combined with the standard prophylaxis regimen (cyclosporine + methotrexate), demonstrated a reduction in the 24-month chronic GVHD incidence from 41.3% to 26.3%, while overall survival increased from 53.3% to 70.6% [32]. Emerging targeted agents, including post-transplant cyclophosphamide (PTCy) selectively eliminates alloreactive T cells via DNA damage-induced cytotoxicity while sparing regulatory T lymphocytes [33]. Additionally, targeting cytokine pathways also demonstrates promising clinical efficacy, such as JAK inhibitors (e.g., ruxolitinib) suppress pro-inflammatory cytokine cascades such as TNF-α, IFN-γ and IL-6, mitigating tissue injury in GVHD [30, 34].

HVGR

HVGR represents another major barrier in allogeneic CAR-T therapy. It originates from the host immune system recognizing infused CAR-T cells as foreign entities, thereby activating immune clearance mechanisms. The core pathological process involves host alloreactive T cells recognizing donor cell-surface HLA molecules via TCRs, while NK cells may also mediate HLA-independent rejection mechanisms [35]. Beyond HLA mismatching between donors and recipients, the CAR construct itself poses significant immunogenic risks as a foreign antigen [36, 37].

Suppressing host immune function constitutes the primary strategy for mitigating HVGR. Lymphodepleting regimens based on fludarabine-cyclophosphamide have demonstrated multiple therapeutic adaptations in clinical trials and are broadly integrated into HVGR prophylaxis protocols. Standardized conditioning in pivotal allogeneic CAR T-cell trials combines CD52-directed cytoreduction (e.g., alemtuzumab) and conventional chemolymphodepletion (fludarabine/cyclophosphamide) to achieve maximal lymphocyte eradication [5]. However, during alemtuzumab-mediated suppression of host T and NK cells, concomitant genome editing to knockout the CD52 gene in allogeneic CAR-T cells is essential to prevent therapeutic cells from inadvertent depletion [38]. Demonstrated in an in vitro study, dasatinib exhibits comparable efficacy in suppressing host alloreactive T or NK cells [39].

Utilizing HLA-matched homozygous donors or reducing HLA expression represents a critical strategy to mitigate HVGR. However, identifying fully HLA-matched donors remains challenging. Alternative banks with a distinct HLA typing profile may avoid specific anti-HLA antibodies for specific cell batches [40, 41]. In the case of HLA mismatch, allogeneic T cells are susceptible to rejection after immune recognition. Knocking out the β2-microglobulin (B2M) subunit inhibits the expression of HLA class I (HLA-I) molecules, preventing the host immune cells from recognizing the therapeutic T cells as exogenous antigens via their TCRs, thus avoiding immune rejection [42, 43]. However, HLA-I negative T cells are vulnerable to lysis by recipient NK cells. In this regard, some studies have reported that a B2M-HLA-E-peptide or B2M-HLA-G-peptide fusion gene was specifically inserted into the B2M locus to prevent recipient NK cell-mediated lysis [44, 45]. Xu HG et al. showed that iPSCs with targeted knockout of HLA-A and HLA-B, but retaining endogenous HLA-C and non-classical HLA-E/F/G, exhibit enhanced protection against NK cell recognition. This protective effect surpassed that seen in HLA-A/B/C triple-knockout iPSCs [46]. Recently, Chen XF et al. confirmed that knocking out HLA-A/B (while retaining HLA-C/E) significantly reduces host T cell-mediated rejection [47].

Additionally, donor-derived T cells express HLA class II (HLA-II) molecules. Therefore, it is necessary to simultaneously prevent interactions between HLA-II and host CD4⁺ T cells to guard against lysis mediated by alloreactive CD4⁺ T cells [43, 48]. Disrupting the crucial transcriptional activator of HLA-II molecules (CIITA), as well as the genes encoding the α chain of HLA-II molecules (HLA-DRA, HLA-DQA, HLA-DPA) eliminates the expression of HLA-II [43, 48, 49]. Researchers utilized CRISPR-Cas9 to knock out RFX5, developing the CD7-targeted allogeneic CAR-T product RD13-01, which achieved the expected reduction in rejection [49], Strategies targeting other HLA-II transcription factors (e.g., RFXAP and RFXANK) also theoretically confer HVGR resistance [36, 50]. Furthermore, as a critical chaperone for HLA-II assembly and trafficking, CD74 deficiency indirectly suppresses HLA-II surface expression [51]. Employing CD74-targeted shRNA eliminates HLA-II and markedly reduces HVGR (NCT00840853).

Alternative approaches such as CAR construct optimization offer significant potential. Williams AM et al. engineered allogeneic CAR-T/NK cells expressing a cytotoxic alloimmune defense receptor (ADR) [52]. By hijacking 4-1BB (CD137), an activation-induced costimulatory molecule transiently expressed on alloreactive host lymphocytes, this receptor directs specific elimination of hostile immune cells. Besides, humanized tumor-reactive single-chain variable fragments (scFvs) have also demonstrated favorable safety profiles [53]. Alternatively, immunoglobulin heavy-chain-only recognition domains (e.g., VHH domains or nanobodies) can replace traditional scFvs to mitigate immunogenicity [54, 55].

Genome editing for allogeneic CAR-T cells

For more than a decade, scientists have been researching protein-based DNA-recognition domains fused with nucleases for the engineering of allogeneic CAR-T cells, including zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), homing endonucleases (Meganucleases), and the RNA-guided clustered regularly interspaced short palindromic repeats (CRISPR) systems. These mainstream gene editing tools specifically trigger double-strand breaks (DSBs) at target sites via chimeric nucleases, while minimizing off-target risks. Following DNA cleavage, cellular repair mechanisms mediate either gene inactivation (knock-out) via the error-prone non-homologous end joining (NHEJ) pathway, which frequently introduces disruptive insertions or deletions (indels) leading to functional gene loss [56], or gene insertion or correction (knock-in) via the homology-directed repair (HDR) pathway. HDR requires exogenous templates with homology arms for high-fidelity repair during S/G2 cell cycle phases [57, 58], but remains inefficient due to template dependency [5, 59, 60]. Genome-editing tools for T-cell modification are illustrated (Fig. 2), with their advantages and limitations summarized (Table 2).

Fig. 2.

Genome-editing tools for T-cell modification. (A) Zinc finger nucleases (ZFNs). (B) Transcription activator-like effector nucleases (TALENs). (C) Meganucleases megaTAL. MegaTAL consists of TALE arrays fused to a meganuclease that integrates the binding affinity of TALE arrays and the site-specific nuclease activity of meganuclease in a single structure. (D) Meganucleases ARCUS. A monomeric meganuclease known as I-Crel, is capable of carrying out both target site recognition and cleavage. (E) Clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 system. (F) CRISPR hybrid RNA-DNA (chRDNA)/Cas9 system. Using a hybrid RNA-DNA guide instead of sgRNA can enhance the specificity of the Cas9 nuclease. (G) Cas-CLOVER system. The Cas-CLOVER system’s dual guiding strategy boosts its fidelity compared to the single-guiding CRISPR system. This system consists of two dCas9, each fused to a Clo051 subunit. Clo051 activation demands its dimerization, which relies on each subunit being guided to the target site through a separate sgRNA. (H) Base editing system. Base editors contain a Nickase Cas9 (nCas9) variant, tethered to cytidine and adenosine deaminase for catalyzing highly targeted C→T or A→G base conversion, and introduces premature STOP codons or interferes with splicing sites to cause gene disruption without creating DSBs. (I) Prime editing system. nCas9 can also be fused to reverse transcriptase (RT) to form a complex with specialized guide RNA (pegRNA) for introducing larger sequence additions into a DNA region without generating DSBs. Created with BioRender.com

Table 2.

The advantages and limitations of different gene editing tools

Editing tools	Mechanism	Advantages	Limitations
ZFNs	Zinc finger protein recognition + FokI cleavage	·Mature early-stage technology ·Moderate specificity	·Limited target design (requires specific triplets) ·High off-target effects ·Low multiplex editing efficiency
TALENs	Protein-based DNA recognition + FokI nuclease cleavage	·Modular design flexibility ·High targeting specificity ·Lower off-target risk vs. CRISPR ·Edits GC-rich regions effectively	·Complex/time-consuming protein engineering ·Low delivery efficiency (requires mRNA/protein) ·low transduction efficiency for large plasmid size (~ 3 kb) ·Increased cytotoxicity with multiplex editing
Homing endonucleases	Large-scale nuclease recognizes long sequences (12–40 bp)	·Ultra-high specificity (long recognition) ·Lowest off-target risk ·Enables CAR transgene insertion via HDR	·Few natural target sites ·Difficult to engineer ·Delivery challenges ·Low editing efficiency (especially in primary cells)
CRISPR/Cas9	RNA-guided Cas9 nuclease cleaves dsDNA	·Simple design (only sgRNA needed) ·High-efficiency multiplex editing ·Broad cell type applicability ·Supports HDR for targeted insertion ·Broad clinical use	·PAM sequence dependency limits target sites ·High off-target risk ·DSBs may cause chromosomal abnormalities (e.g., translocations) ·Inefficient large-fragment integration ·Compromised in vivo persistence with multiplex editing
Base editing	Nickase Cas9 + deaminase (C→T or A→G conversion)	·No DSBs (improved safety) ·Avoids translocation risk ·Efficient stop codon introduction ·Precision point mutations ·Supports triple/quadruple editing	·Limited to specific base conversions ·Bystander editing effects ·Deaminase-induced RNA off-target effects ·Unknown long-term genetic stability ·Reduced T-cell expansion with multiplex editing
Prime editing	Cas9-reverse transcriptase fusion + pegRNA template	·Enables any DNA edit (insertion/deletion/replacement ·No donor DNA or DSBs required ·Theoretically lowest off-target risk	·Low editing efficiency (especially in primary cells) ·Complex delivery (long pegRNA) ·Unvalidated in vivo functionality ·Emerging tech and needs optimization

Genome editing tools

ZFNs

ZFNs are artificial chimeric nucleases based on Cys2-His2 zinc finger motifs, assembled into domain arrays to specifically bind target DNA and recruit FokI endonuclease domains for site-specific DSB induction [61, 62]. Usually, after 5 or 6 base pairs (bp) separate the two zinc finger binding domains, the dimerized FokI catalytic domain enters and cleaves double-stranded DNA [62, 63]. To reduce off-target cleavage, an obligate heterodimer FokI variant was generated to eliminate off-target events caused by homodimerization of the same ZFNs [64]. Though less widely used than CRISPR, ZFNs have clinical success, establishing reliability in genetic manipulation.

TALENs

Like ZFNs, TALENs need to be engineered in pairs to allow FokI to dimerize and cleave the required DNA target. TALENs consist of transcription activator-like effectors (TALEs) central DNA-binding domains, flanked by FokI catalytic domain (C-terminal) and nuclear localization signal domain (N-terminal) [64–66]. The key advantage of TALENs lies in their modular and predictable design: each typical tandem TALE repeat sequence (33–34 amino acids, highly conserved) has unique DNA binding specificity determined by divergent amino acids at positions 12 and 13—known as the repeat variable di-residue (RVD), specific amino acid pairs that confer DNA-binding specificity to individual nucleotide bases [65–67]. This straightforward RVD-code means the design and reprogramming of TALENs is more straightforward than ZFNs [68, 69], allowing them to recognize virtually any DNA sequence by combining fragments with appropriate RVDs, and providing flexibility for modifying multiple loci. TALENs also offer superior specificity over CRISPR in some contexts, making them ideal for high-precision edits and large-scale engineered T-cell manufacturing [65].

Homing endonucleases

Homing endonucleases (meganucleases) are naturally occurring restriction enzymes, distinct from modular ZFNs/TALENs/CRISPR systems—they act as single proteins to recognize and cleave target DNA [8]. Their small, monomeric structure facilitates vectorization and delivery. By recognizing long sequences (14–40 bp), they achieve highly specific cleavage with minimal off-target activity [70, 71]. Although this structure is difficult to re-purpose to target new sequences, and the cleavage activity may be affected by DNA methylation and chromatin structure at the target site [72, 73], megaTAL and ARCUS, two tools based on meganucleases, have been successfully applied to CAR-T cell therapy. MegaTALs are monomer hybrid nucleases, which have fusion of a TALE arrays with a meganuclease. This combination enables an increase in cleavage activity at highly specific target sequences, reduces the risk of mismatch, and leads to highly accurate editing of target genes with minimal off-target events [74, 75]. Moreover, ARCUS (from Precision Bioscience), derived from homodimeric I-Crel endonuclease, supports multiplex editing with extremely low off-target rates [76, 77].

CRISPR systems

CRISPR-Cas9 technology revolutionized gene editing with simplicity, flexibility, and efficiency. Unlike protein-DNA-dependent nucleases, CRISPR/Cas9 relies on RNA-DNA interactions for target recognition [68], derived from a microbial “adaptive immune response” to combat infections [78, 79]. It comprises non-specific Cas9 endonuclease (inducing DSBs) and a single guide RNA (sgRNA) recognizing a 20-bp target sequence. Downstream of the cleavage site, the protospacer adjacent motif (PAM), which is a short DNA sequence required for CRISPR target recognition, provides additional specificity through arbitrary variants of 5’-NGG [56]. HNH and RuvC domains within the Cas9 nuclease cleave the target and non-target strands of DNA to trigger DSBs [80, 81]. Precise control of DSBs timing and location reduces accidental indels, translocations, and p53-inactivated cell selection [82–84]. Cas9/sgRNA ribonucleoprotein (RNP) complexes activate Cas9 directly and degrade rapidly post-internalization, shortening off-target exposure while maintaining on-target efficiency [85–87]. Cas9 variants (nickase Cas9 [nCas9] and catalytically dead Cas9 [dCas9]) address DSB-related flaws. Paired nCas9/sgRNA complexes flanking the target site induce DSBs at specific locations, enabling relatively accurate control over DSB positioning and reducing off-target effects [88]. Meanwhile, the dCas9 variant can be easily combined with other functional enzymes to trigger various site-specific modifications independent of Cas9’s catalytic activity [89].

However, RNA-DNA recognition increases off-target risks. Thus, it is necessary to improve sgRNA by optimizing the design algorithms to decrease off-target activity and the risk of chromosome abnormalities [90]. In contrast to the traditional sgRNA, the off-target activity is obviously reduced using a hybrid DNA-RNA (chRDNA) guide to the target site. This DNA-DNA binding mode has limited affinity, which is convenient to prevent Cas9 activation in time in case of mismatch, while will not interfere with the on-target activity of Cas9 [91, 92]. Cas12a is an RNA-guided endonuclease, and its size is around 1/3 of the Cas9. Cas9 needs both CRISPR RNA (crRNA) and trans-acting CRISPR RNA (tracrRNA), while Cas12a only needs crRNA [93]. In addition, the sticky ends generated by Cas12a is superior to the blunt ends by Cas9, which is beneficial for site-specific insertion of the target gene or correction through HDR pathway [94]. The Cas-CLOVER system can also generate the sticky ends of DSBs, based on the fusion of naturally occurring Clostridium “Clo051” endonuclease with dCas9 [95]. This advanced strategy has higher fidelity and allows gene editing in inactive T cells, thus retaining the cells with a low differentiated phenotype [96].

Base editing

CRISPR-guided cytidine or adenosine deamination can mediate a highly precise C→T or A→G base conversion, respectively [97]. This base editing approach enables efficient disruption of splice sites or generation of premature stop codons without DSBs, thus avoiding triggering translocations and other chromosomal aberrations [98, 99]. It also alters amino acid codons, corrects single nucleotide variants, and creates point mutations [56]. However, base editing may also trigger off-target events, which may be related to the intrinsic affinity of deaminases for DNA, or the mismatches between sgRNA and undesired sequences [100]. In addition, base editing cannot be used to introduce reverse mutations (pyrimidine ↔ purine) and small DNA insertions, deletions or substitutions [101].

Prime editing

An alternative technique called prime editing has been developed to facilitate all possible base-to-base conversions and precisely targeted insertions or deletions without DSBs intermediate [102, 103]. Prime editing directly inserts the target gene into a specified DNA site using the prime editor (a nCas9 endonuclease fused to an engineered reverse transcriptase), guided by a prime editing guide RNA (pegRNA). This accurate genome editing method shows higher or similar efficiency and fewer byproducts than HDR, and has complementary advantages compared to base editing [101].

Off-target effects and genotoxicity

Off-target effects occur when gene-editing tools unintentionally modify genes other than the intended target, potentially causing unexpected side effects. CRISPR/Cas9, despite its high specificity, has been reported to cause unintended DNA breaks at sites with sequences similar to the target gene, increasing the risk of off-target mutations [104]. TALENs generally exhibit fewer off-target mutations but are not entirely risk-free, especially when editing large or complex genomic regions. These off-target effects pose significant challenges to the safety of allogeneic CAR-T therapies, as they may compromise immune function or lead to toxicities. Advanced delivery systems, such as ribonucleoprotein complexes, and modified guide RNAs (such as the high-fidelity Cas9 variants) have been developed to improve editing precision [105, 106].

Genome editing tends to generate some unexpected genetic risks, including insertion/deletion mutations, large deletions, loss-of-heterozygosity, chromosomal translocations and even loss [107]. DSBs-dependent editing poses a risk of genomic instability. Genotoxicity may arise, although in the case of on-target cleavage, when the generated DSBs interacts with other nuclease-induced DSBs or spontaneous DSBs within the genome. Especially after using multiplex gene editing strategies, the risk of translocation becomes particularly obvious [38]. During the UCAR-T19 gene editing process, simultaneous targeting of the TRAC gene (chromosome 14) and the CD52 gene (chromosome 1) unexpectedly led to the detection of the t(1;14) chromosomal translocation in all batches of UCART19 cells [108]. These risks arise from the integration of transgenes into the host genome, which can disrupt normal cellular processes or activate oncogenes, potentially leading to secondary malignancies. Viral vectors, commonly used for gene delivery, are particularly prone to random integration, which may inadvertently activate proto-oncogenes or disrupt tumor suppressor genes. Although CRISPR/Cas9 and TALENs offer more precise editing, they are not immune to off-target effects, which can result in unintended genetic alterations and increase the risk of cancer formation [109]. Currently, base editing and prime editing are theoretically considered safer as they do not require DSBs. However, these technologies face practical challenges due to their limited editing scope and suboptimal efficiency. Furthermore, like other editing technologies, complex multiplex editing impairs T cell viability and expansion capacity [98, 110, 111]. Additionally, deaminases in base editing may induce RNA off-target effects, and the long-term genetic stability remains unknown [98, 111].

To mitigate these risks, researchers have developed multiple strategies. These include gRNA optimization approaches such as adding cytosine (C) extensions to the 5’ end of conventional gRNAs to precisely regulate Cas9 activity. Experimental validation demonstrates that this strategy reduces p53 activation and cytotoxicity while enhancing homology-directed repair efficiency by over twofold, with compatibility extending to Cas12a systems [112]. Alternatively, introduction of the ‘tevopreQ1’ motif enhances stability, achieving >95% precise editing rates in MMR-deficient cells such as MLH1-knockout K562 cells, significantly outperforming traditional pegRNA [113]. Additional approaches encompass innovative precision editing technologies, safety switches like the RQR8 suicide gene [114], and transposon-based systems [115], all designed to minimize harmful genomic integrations. Nevertheless, persistent monitoring remains indispensable for ensuring therapeutic safety [116, 117].

Risks monitoring

The long-term risks associated with universal allogeneic CAR-T therapy include delayed toxicities, secondary malignancies, and immune-related complications. Despite genetic modifications aimed at reducing T-cell alloreactivity, GVHD remains a potential concern. Prolonged immunosuppression may increase susceptibility to opportunistic infections or trigger autoimmune disorders. Furthermore, off-target editing-induced genomic instability could lead to uncontrolled cellular proliferation, elevating the risk of malignant transformation [118]. Flow cytometry serves as the primary method for assessing the immunophenotype and composition of circulating CAR-T cells [119, 120]. However, its inherent limitations, including low throughput and a hypothesis-driven approach, hinder unbiased screening and molecular profiling of these cells. A holistic strategy integrating multi-dimensional technological platforms and end-to-end management protocols is essential to monitor off-target effects, genotoxicity, and long-term risks throughout the allogeneic CAR-T editing process.

The advancement of highly sensitive off-target detection methodologies is paramount. Zhu M et al. developed a novel technique that captures single-stranded DNA bound by replication protein A (RPA), combined with strand-specific nuclease digestion of double-stranded DNA to enhance signal-to-noise ratio. This approach enables the detection of off-target effects for diverse editing tools, including Cas9, cytosine base editors (CBEs), adenine base editors, and prime editors. It exhibits high sensitivity, broad applicability, and reveals heterogeneity. Notably, it identified significant differences in off-target sites for the same guide RNA (gRNA) across different editors (e.g., Cas9 vs. CBE), with off-target events frequently occurring in open chromatin regions [121].

Off-target prevention and genomic risk control can be enhanced by optimizing editing tools, controlling editing duration, and strengthening in vitro and in vivo validation. Beyond employing high-fidelity Cas9 variants or base editors to minimize non-specific cleavage, designing highly specific gRNAs and pre-screening potential off-target sites using algorithms (e.g., COSMID) are critical. DISCOVER-seq leverages chromatin immunoprecipitation of the endogenous DNA repair protein MRE11 to directly capture CRISPR-Cas cutting sites in living systems, enabling label-free off-target detection [122]. This technique facilitates in situ identification of off-target events (sensitivity ~ 0.3%) in animal models and patient-derived iPSCs, precisely delineating Cas enzyme molecular activity and serving as a safety assessment tool for clinical gene editing. Additionally, Yu Z et al. designed PEAC-seq, which modifies the Prime Editor system to insert a fixed sequence tag, enabling simultaneous detection of off-target edits and chromosomal translocations (e.g., large deletions, translocations), providing a solution for identifying editing-induced chromosomal rearrangements [123].

In vitro, whole-genome sequencing (WGS) and whole-exome sequencing (WES) are employed to identify large-scale genomic alterations and point mutations, respectively, effectively detecting insertions/deletions (indels) and structural variants [8, 124]. In vivo, integrating single-cell multi-omics (scRNA-seq + scATAC-seq) elucidates CAR-T cell gene expression profiles, clonal evolution, and immunophenotypes, enabling early detection of off-target mutations or aberrant cellular proliferation for effective risk management [125]. These tools collectively contribute to safeguarding the safety and efficacy of CAR-T therapy throughout its lifecycle. However, persistent challenges include insufficient sequencing coverage/depth, sequencing biases, and high costs, underscoring the ongoing need for diagnostic innovation [125, 126].

The full spectrum of adverse editing consequences may only manifest as unexpected side effects during allogeneic CAR-T therapy. For instance, Shen J et al., utilizing the FDA Adverse Event Reporting System (FAERS) and VigiBase databases, identified types of second primary malignancies (SPMs) with high incidence post-CAR-T therapy (e.g., T-cell lymphoma, acute myeloid leukemia), observing a significantly earlier onset time compared to non-CAR-T populations [127]. Therefore, rigorous long-term monitoring, patient follow-up, and biospecimen archiving are crucial, particularly in the early stages of therapeutic deployment [5].

Screening the sources/subpopulations of allogeneic CAR-T cells

At present, most allogenic CAR-T cells are mainly manufactured from T cells in peripheral blood mononuclear cells (PBMCs) and, rarely, from umbilical cord blood (UCB) and iPSCs. Allogeneic CAR-T cells derived from healthy donor PBMCs are related to the ability to make multiple vials from a single apheresis product, and therefore quickly obtain from previously manufactured products, which are not impacted by immune inhibitors or exposure to chemotherapeutic agents, compared with autologous T cells from patients. Moreover, cell banks expressing different subtypes of HLA complex are able to be generated to potentially match the HLA types of patients. Allogeneic CAR-T cells can be derived from the HSCT donor or HLA-matched sibling donor, but only for patients who relapsed after allogeneic HSCT. It has been reported that twenty patients with B-cell malignancies received CD19-targeted CAR-T cells manufactured from each recipient’s allogeneic HSCT donor, and they did not receive chemotherapy before infusion. GVHD was not reported, and among these patients, 6 achieved CR and 2 achieved partial response (PR) [128]. However, selecting T cells from donors according to immune characteristics may have unique advantages of reducing heterogeneity, as discussed below (Table 3).

Table 3.

Comparison of several types of allogeneic CAR-T cell therapy

Cell sources	Advantages	Limitations
PBMCs	·Rich T cell pool ·High proliferative potential ·Established isolation protocols (standardized clinical-grade methods)	·Significant donor variability ·Costly donor screening ·Requires rigorous HLA matching (alloreactivity risk)
UCB	·More naive phenotype ·High proliferative capacity and persistence ·Low immunogenicity (low risk of GVHD)	·Limited quantity of cells ·Difficult donor tracing (no additional cell infusion)
iPSCs	·Unlimited source ·Homogeneous (defined TCR specificity and HLA haplotype)	·Tumorigenic potential ·Additional genetic modifications &. quality controls ·Complicated procedure ·High cost

Cell types	Advantages	Limitations
αβ T cells	·Designed to avoid alloreactivity ·High availability	·Complicated genetic modification ·Risk of off-target mutations ·Residual TCRαβ + cells lead to GVHD ·Slow proliferation in vitro ·High cost
γδ T cells	·Multiple killing modes ·Cytotoxic phenotype ·HLA-independent antigen recognition (low risk of GVHD) ·Effective infiltration and resistance to hypoxia ·Efficacy in solid tumours ·TME amelioration	·Limited quantity of cells ·Short persistence in vivo ·Slow proliferation in vitro ·Low availability
NK T cells	·Advantages of dual NK cell and T cell properties ·Multiple killing modes ·Low risk of GVHD ·TME amelioration	·Limited quantity of cells ·Short persistence in vivo
CIK cells	·Cytotoxic phenotype ·Low immunogenicity	·High heterogeneity ·Limited quantity of cells ·Short persistence in vivo
VST cells	·Virus-specific TCR (non-alloreactive TCR) ·Prevent virus-induced cancers or infections ·Boosting in vivo via TCR stimulation	·Short persistence in vivo without TCR stimulation with viral antigens

UCB, umbilical cord blood; iPSCs, induced pluripotent stem cells; CIK cells, Cytokine-induced killer cells; VST cells, virus-specific memory T cells; GVHD, Graft-versus-host disease; TME, Tumor microenvironment; TCR, T-cell receptor

UCB T cells

UCB is an enriched source of hematopoietic stem cells (HSCs). HSCs are able to self-renew, expand in vitro and to differentiate into T cells, although the total of HSCs is limited in each UCB [129]. Such CAR-T cells have higher proliferative capacity and persistence, moreover, the unique naive antigen phenotype contributes to the decreased alloreactivity of UCB grafts [130]. The characteristics of UCB T cells also include impaired NFAT signalling and lower immunogenicity, which may be associated with the decreased risk of GVHD [131], and thus allows less strict restrictions on HLA disparities [132]. In addition, UCB T cells usually express lower levels of T cell inhibitory receptors compared to those of PBMCs [133], and CCR7-enriched UCB CD8⁺ T cells reinforced tumor-homing ability to mediate a more effective antitumor effect by rapidly infiltrating tumors and inducing cytotoxic CD8⁺ and CD4⁺ Th1 T cells in the TME. Placenta-derived stem cells have a unique HLA expression pattern, in which extravillous cytotrophoblast cells only express HLA-C, HLA-E and HLA-G, while syncytiotrophoblast cells are HLA-negative [134]. It has been reported that the expression of HLA-E and HLA-G in universal CAR-T-19 cells prohibit host NK cell-mediated lysis and were not recognizable by allogeneic T cells [44, 45]. Although the research on HLA specificities of placental-derived CAR-T cells has not yet been clinically assessed, it indicated the potential in cancer cellular immunotherapy.

iPSCs

Another source of allogeneic CAR-T cells is iPSCs, which theoretically have unlimited self-renewal and proliferation capacity, while maintaining the pluripotency and lineage differentiation potential, including T cells and NK cells with prominent antitumor activity [135, 136]. A bank of iPSCs lines with HLA haplotypes could be generated to reduce the risk of alloreactivity of iPSCs-derived CAR-T cells by HLA matching between host and graft. In the case of HLA mismatch, TCR and HLA of allogeneic CAR-T cells are eliminated by gene editing to reduce immune rejection. A recent study indicated that iPSCs-derived cytotoxic T lymphocytes (iPS-rCTLs) are able to survive persistently and maintain the cytotoxicity in vivo, iPS-rCTLs transduced with anti-CD38 CAR can inhibit tumor growth, and there is no risk of CD38-mediated fratricide [137]. In addition, these CAR-T cells are generated from a single clonal engineered pluripotent cell line with the capacity for clonal expansion, and are therefore homogeneous [135]. However, the tumorigenic potential of undifferentiated iPSCs has not been ruled out [138], such as teratomas and other adverse reactions [139].

γδ T cells

γδ-T cells are unconventional T cells lacking MHC restriction, thus reducing the risk of GVHD without gene editing. And γδ-T cells are critical components of the innate immune system, constituting only 1–5% of circulating lymphocytes, while they are modulators of adaptive immunity to indirectly target tumor cells [140], are mainly divided into 3 subtypes based on delta chain rearrangements: Vδ1, Vδ2, and Vδ3 [141]. Vδ2 T cells are enriched in peripheral blood, while Vδ1 and Vδ3 T cells are predominantly tissue-resident such as intestinal mucosa, skin and liver [142, 143]. Especially the Vδ1 subtype, has a homing advantage over αβT cells, and thus can better infiltrate tumors, particularly those with a hypoxia TME [144, 145]. γδT cells can be regarded as APCs to present tumor antigens to αβT cells, leading to the expansion of αβT cells, and mobilizing other immune cells to attack tumor cells [145, 146]. Capsomidis A et al. demonstrated that transforming γδT cells with CAR are able to increase the cytotoxicity, while retaining the ability to infiltrated tumor cells and cross-present antigens [146]. In addition, γδT cells can recognize tumor cells by endogenous receptors (Vδ1TCR, CD16, NKG2D and NKp30) to kill tumors: pathways like CD16-mediated antibody-dependent cellular cytotoxicity, perforin/granzyme-triggered cytotoxicity and Tumor Necrosis Factor-Related Apoptosis-Inducing Ligand (TRAIL)/FAS Ligand (FASL)-mediated apoptosis [147].

A recent study indicated that γδ T cells can also be derived from iPSCs. Wallet et al. manufactured iPSC-derived γδ CAR-T cells which showed continuous tumor cell killing with the presence of IL-15, and produced significantly less IFN-γ and other inflammatory cytokines to ameliorate the humoral immune response compared to αβ CAR-T cells [148]. Historically, the clinical research on γδ T cells is limited because of the lower number in peripheral blood. Nevertheless, Neelapu S et al. generated the first-in-class allogeneic CD20-targeted γδ CAR-T cell therapy (ADI-001) for patients with R/R B-cell lymphoma (BCL), which demonstrated that it had a comparable safety profile to autologous CAR-T cells, and 4 (67%) of 6 patients achieved CR at day 28 [149]. Recently, ADI-001 exhibited dose-dependent expansion and persistence, clinical responses were associated with ADI-001 cellular kinetics, regardless of extensive degree of HLA mismatch [150]. Further studies explored strategies to optimize γδ CAR-T efficacy. Lee D et al. constructed allogeneic γδ CAR-T cells by selecting Vδ2 T cells with naturally high CD16 (FcγRIII) expression and integrating CD19-CAR with an IL-15 module [151]. This therapy demonstrated dual antitumor mechanisms in ovarian cancer models, retaining CD16-mediated antibody-dependent cellular cytotoxicity (ADCC) while enhancing CAR targeting and IL15-driven expansion, prolonging survival without inducing GVHD.

NK T cells

Another candidate is NKT cells, which are a subset of innate-like T lymphocytes that express NK cell surface markers and recognize specific glycolipids presented by a non-polymorphic HLA class I-like molecule CD1d [152, 153]. NKT cells acquire the characteristics of NK cells after developing in the thymus, forming an important bridge between innate and adaptive immunity, and triggering quantities of polyfunctional cytokines within hours after activation that protect the host from cancers, infections, and autoimmune diseases [154, 155]. In addition, NKT cells indirectly participate in the antitumor immune response through the interactions with dendritic cells (DCs), tumor-associated macrophages (TAMs) and the activation of T cells and NK cells [156–158]. Considering that the majority of cancers are CD1d-negative, NKT cells can not exert the direct killing function of NK cells. However, the CAR engineering is able to compensate for this defect. Besides the CAR-mediated targeted killing ability, CAR-NKT cells are able to eliminate tumor cells by using perforin/granzyme, TNF-α, TRAIL and FASL in a non-targeted way, which makes them more suitable for eliminating extremely heterogeneous tumors [141]. According to the studies, CAR-NKT cells can exert dual specific killing ability through CAR-/CD1d- mediated mechanisms, generating a stronger anti-tumor effect, and effectively eliminate relapsed lymphoma without causing excessive immune response compared to conventional CAR-T cell therapy [159, 160].

NKT cells only account for 0.1-1% of circulating T lymphocytes, resulting in the scarcity of CAR NKT cells sources. One strategy to collect quantities of NKT cells is to use NKT cell-derived iPSCs. NKT redifferentiated from NKT-iPSCs, and retained the function of secreting IFN-γ, exhibiting strong cytotoxicity in vitro and have similar antitumor activity similar to that of parental NKT cells in human xenograft mice models [161]. Another problem is the poor persistence of NKT cells in vivo. Tian G et al. indicated that CD62L⁺ NKT cells have prolonged persistence and antitumor activity in vivo [162]. In order to investigate strategies of enriching CD62L⁺ NKT cells, the introduction of certain cytokines would generate the expected results. In trials of anti-GD2 CAR-NKT cells, the researchers found that the co-expression of IL-15 increased the proportion of CD62L⁺ cells and improved the persistence of CAR-NKT cells in vivo [163, 164].

Notably, Yang Lili’s team pioneered an alternative approach to resolve NKT scarcity and persistence simultaneously—allogeneic HSC-engineered invariant NKT (HSC-iNKT) cells [51]. By differentiating gene-edited CD34⁺ HSCs in artificial thymic organoids, this method generates >10¹¹ clinical-grade iNKT cells per donor. The engineered HSC-iNKT products can be engineered to co-express IL-15 or CARs while retaining intrinsic tumor-homing receptors (e.g., CCR5/CXCR3) and exhibiting low HLA-I/II expression to evade host rejection. Critically, HSC-iNKT cells demonstrate enhanced persistence, leverage dual (NK/TCR) or triple (NK/TCR/CAR) antitumor mechanisms, and incorporate safety switches (e.g., sr39TK) for off-the-shelf applications. Further refining this platform, their latest study establishes a xenogeneic-feeder-free, GMP-compatible 5-stage differentiation system that achieves >10⁶-fold expansion of UCB-derived allogeneic CAR-NKT cells [165]. These cells exhibit stable hypoimmunogenicity via epigenetic silencing of HLA/NK-ligand genes, enable triple-targeting efficacy against BCMA-loss multiple myeloma, and eliminate immunosuppressive myeloid cells in the TME without inducing GVHD or cytokine release syndrome (CRS) in vivo. Building upon this platform, their recently study demonstrates CD33 CAR-NKT cells uniquely overcome conventional CAR-T limitations in myeloid malignancies [166]. This CD33-targeted variant achieve >95% eradication of bone marrow-resident leukemic stem/progenitor cells, a key obstacle in AML/Myelodysplastic Syndromes (MDS) treatment. The GMP-compatible manufacturing process and absence of GVHD/CRS risks further validate their potential as a scalable “off-the-shelf” therapeutic, advancing clinical translation for intractable hematologic cancers.

Cytokine-induced killer (CIK) cells

Cytokine-induced killer (CIK) cells constitute a heterogeneous cell groups including NK, NKT and T cells [167], and previous study indicated that CIK cells are terminally differentiated activated CD8 cytotoxic T-EMRA lymphocytes [168]. They express T-cell markers and contain a high percentage of NK-like T cells, which are characterized by the enrichment of CD3CD56 cytotoxic cells, and generated by PBMCs in the presence of IL-2, anti-CD3 antibody and IFN-γ [169, 170]. Although this procedure brings heterogeneity, the advantages of both MHC-dependent and MHC-independent cytotoxicity, as well as minimal alloreactivity make it applicable in clinical trails. The clinical data of patients with R/R ALL confirmed that donor-derived CIK-cell therapy had anti-leukemia activity without occurrence of GVHD, and 28 days after infusion, 8 (61.5%) of 13 patients achieved CR/incomplete hematologic recovery (CRi) [171]. At present, the small number of treated patients makes it difficult to evaluate the variability of CAR CIK production associated with the clinical results. In addition, researchers claim that CIK cells can also be derived from UCB, which mostly benefits patients who fail to collect or manufacture autologous CAR-T cells [172].

Virus-specific memory T cells

Virus-specific T (VST) cells have a TCR that can specifically recognises viral antigens, and have an effective treatment for infections with Epstein-Barr virus (EBV), cytomegalovirus (CMV), Varicella Zoster virus (VZV) or Adenovirus (ADV) [9, 173]. Purified memory virus-specific T cells can provide protection against viral diseases without GVHD [174]. One possibility is the limited repertoire of these memory T cells, because the quantity of T cells and TCR diversity are directly proportional to the risk of alloreactivity.

Actually, the insertion of a specific CAR avoids resistance to EBV-specific T cells or immune escape caused by EBV antigen loss, thus enhancing the antitumor effect [175, 176]. Nearly 40% of patients with Hodgkin lymphoma express EBV-related antigens in tumor cells and VST cell therapies showed good tolerance and remission rates [9]. Most viruses effectively activate innate and Th1 polarized immunity, therefore, virus vaccines or oncolytic viruses are able to strengthen the activity of CAR-transduced VST cells by their native TCR [173]. Stimulating TCR leads to the proliferation of VST cells, thus eliminating EBV-infected memory B cells [176]. Similarly, the duration of response (DOR) and persistence of CAR-T cells were increased via CMV vaccination, while CAR molecules were engrafted onto CMV-specific T cells to mediate the lasting remission of B cell malignancies [177]. In addition, Caruana I et al. reported that CMV-specific VST was implanted with a specific CAR of disialoganglioside GD2, which can be enhanced with a CMV vaccine in neuroblastoma mice model, thus to inhibit tumor growth [178].

Efficiency-driven gene delivery

Although advanced gene editing tools can significantly reduce off-target events, their editing efficiency is largely comparable to that of traditional tools. Moreover, innovative manufacturing technologies are essential for producing CAR-T cells at the scale required for widespread clinical use. Complications and toxicity associated with engineered T cells, potentially caused by the antigen-binding properties of the CAR, include the potential for off-target and on-target toxicity, as well as CRS and neurotoxicity after infusion. Moreover, regardless of the tool used, up to 50% of T cells may remain unedited. The commonly used method to remove unedited cells in clinical trials involves magnetic bead-mediated depletion systems, such as the CliniMACS system. However, some advanced gene delivery strategies, such as γ-retrovirus, lentiviral vectors, transposons, mRNA-based and DNA-based platforms, are improving the efficiency, scalability, and safety of CAR-T cell production.

γ-Retrovirus and lentiviral vectors

γ-Retrovirus was the first vector used in the manufacture of CAR-T cells [179], and currently approved products Yescarta and Tecartus both use γ-retroviral vectors for gene delivery. However, γ-retroviral integration sites are often close to promoters, and integration into proto-oncogenes or tumor suppressor genes may pose a potential cancer risk [180]. In addition, γ-retrovirus cannot infect non-dividing cells and significantly impacts the host transcriptome [181]. Therefore, lentiviral vectors have gradually replaced them in clinical trials, offering a broader range of target cells, higher transduction efficiency, greater transgene capacity, and lower immunogenicity. Because lentiviral vectors integrate randomly, the risk of insertional mutagenesis persists. Additionally, the multi-plasmid transfection system used in their packaging process makes lentivirus production challenging, and batch quality is prone to variation [182]. Nevertheless, long-term experience (over 2 years) with γ-retroviral and lentiviral transduction indicates that the risk of transformation in T-cell therapy is low [183, 184].

Transposons systems

Transposons are natural and mobile DNA fragments, consisting of a transposase and target genes with corresponding binding sites, such as the Sleeping Beauty (SB) and PiggyBac (PB) systems. Singh et al. used the SB11 transposon to generate CD19-targeting CAR-T cells achieving a CAR positivity rate over 90% [115]. Meanwhile, Jin et al. found that SB100X was about 3.6 times more effective than SB11 in generating CAR-T cells [185]. Kubo et al. constructed EPHB4-targeting CAR-T cells using PB, achieving a CAR positivity rate of 78.5% [186]. However, when the transgene exceeds 5 kb, transposition efficiency drops significantly. In addition, since the SB transposon tends to integrate into transcriptional units and the PB transposon is prone to integrate into transcription start sites and CpG islands, both carry the risk of insertional mutagenesis or oncogenesis [187].

mRNA-based platforms

mRNA delivery technology represents a cutting-edge, non-integrative approach to CAR-T cell manufacturing, typically involving mRNA electroporation, lipid nanoparticles, or exosome-mediated mRNA transduction. While mRNA is inherently unstable, electroporated mRNA functions immediately in the cytoplasm, which not only enhances expression efficiency but also avoids the possibility of insertional mutagenesis. Beatty et al. successfully generated CAR-T cells with a positivity rate of over 98% by electroporating mesothelin-specific CAR mRNA into activated T cells [188]. However, CAR expression via this method is limited by the inherent instability of mRNA molecules, resulting in short expression duration and frequently necessitating repeated infusions, leading to higher treatment costs. In addition, lipid nanoparticle-mediated mRNA delivery exhibits lower cytotoxicity while achieving similar transduction efficiency [189]. However, it also faces challenges related to poor stability, susceptibility to degradation, and material compatibility [190]. Another approach is exosome-mediated mRNA delivery, which offers structural stability and low immunogenicity, although the cytotoxicity of exosome-programmed CAR-T cells appears relatively weak [190].

DNA-based platforms

DNA delivery is conceptually similar to mRNA delivery, encompassing DNA nanovector electroporation and nanomaterial-mediated targeted DNA delivery. Bozza M et al. devised a DNA nanovector called nS/MARt through engineering modification of the original plasmid pEPI, and electroporated the CAR-encoding gene into cells, enabling permanent expression of the transgene [191]. Moreover, the nS/MARt-transduced CAR-T cells exhibited stronger tumor infiltration and lysis ability in vivo, without genomic integration or insertional mutagenesis risk. Another approach to shorten the production cycle and reduce cost is to directly generate engineered T cells in vivo. For example, Smith TT et al. used synthetic DNA nanocarriers to directly program leukemia-specific CAR-T cells in vivo, achieving the rapid reduction of tumor burden in B-ALL mice [192].

Clinical trials of genome-edited CAR-T cells

Gene-edited allogeneic CAR-T therapy has achieved landmark breakthroughs in clinical studies for hematologic malignancies (Table 4). However, its efficacy against solid tumors remains suboptimal. Current clinical trials focus primarily on genetic engineering approaches targeting tumor-specific antigens (Table 5). Within the solid TME, multiple factors critically impair the function and persistence of allogeneic CAR-T cells, particularly the immunosuppressive TME and trafficking barriers. This review delineates the key challenges and mechanisms underlying therapeutic failure in solid tumors and further highlights promising engineering strategies (e.g., chemokine receptor expression, TGF-β resistance edits). Through comparative analysis of response rates, persistence, and toxicity profiles across different edited products, we elucidate pivotal factors influencing success, including lymphodepletion regimens, editing efficiency, and cell source. This analysis aims to provide insights for the design of future clinical investigations.

Table 4.

Currently available clinical trials of genome-edited allogeneic CAR-T cells therapy in hematologic malignancies

Sponsors	Product	Tumor (R/R)	Target	Modifications	Edit platforms	Phase	Study start	Patients (received ≥ one infusion)	Toxicity	Outcomes	Trial number	References
Institut De Recherches Internationales Servier	UCART19	B-ALL	CD19	TRAC & CD52 KO	TALENs	I	Jun. 3, 2016	21	CRS 91% and Grade 3/4 CRS 14%; Grade 1/2 ICANS 38%; Grade 4 cytopenia 32%; Grade 1 skin GVHD 10%	2 treatment-related deaths; CR/CRi 67% (28 days); PFS 27% (6 months), and OS 55%	NCT02808442	[114]
	UCART19	B-ALL	CD19	TRAC & CD52 KO	TALENs	I	Aug. 1, 2016	25	DLTs 12%; Grade 3 + CRS 24%; Grade 3 + ICANS 4%; Grade 4 cytopenia 16%; Grade 3 + infections 28%; Grade 1 skin GVHD 8%	4 treatment-related deaths; ORR 48% (median follow-up time at 12.8 months); DOR and median RFS 7.4 months; PFS 2.1 months; OS 13.4 months	NCT02746952	[108]
Cellectis	UCART123	AML	CD123	TRAC & CD52 KO	TALENs	I	Jun. 19, 2017	-	-	-	NCT03190278	[193]
	UCART22	B-ALL	CD22	TRAC & CD52 KO	TALENs	I	Oct. 14, 2019	5	Grade 1/2 CRS 60%; no DLTs, GVHD or ICANS	CRi 40%	NCT04150497	[194]
	UCARTCS1A	MM	BCMA	TRAC & CS1 KO	TALENs	I	Nov. 21, 2019	-	-	-	NCT04142619	[195]
	UCART20 × 22	B-NHL	CD20/22	TRAC & CD52 KO	TALENs	I/II	Nov. 1, 2022	3	Grade 3- CRS 100%; no DLTs, GVHD or ICANS	CR 66.6%, PR 33.3%	NCT05607420	[196]
Allogene Therapeutics	ALLO-501	LBCL/FL	CD19	TRAC & CD52 KO	TALENs	I	May. 1, 2019	46	No DLTs, GVHD or Grade 2 + ICANS; Grade 1/2 CRS 21.7% and Grade 3 + CRS 2.2%; cytopenia 82.6%; Grade 3 + infections 23.9%	1 treatment-related deaths; LBCL CR 36.4% (6 months); In consolidation cohort, FL (n = 4) ORR 100% and CR 75%	NCT03939026	[197]
	ALLO-715	MM	BCMA	TRAC & CD52 KO	TALENs	I	Sep. 23, 2019	43	Grade 3 + adverse events 88%; CRS 55.8% and Grade 3 + CRS 2.3%; ICANS 14%; infections 53.3% and Grade 3 + infections 28%	CR 25%; PR 45.8%; The median DOR 8.3 months	NCT04093596	[198]
	ALLO-501 A	LBCL	CD19	TRAC & CD52 KO	TALENs	I/II	May. 21, 2020	-	-	-	NCT04416984	[199]
	ALLO-605	MM	BCMA	TRAC & CD52 KO	TALENs	I	Aug. 11, 2021	-	-	-	NCT05000450	-
	ALLO-501 A	LBCL	CD19	TRAC & CD52 KO	TALENs	I	Jun. 18, 2024	-	-	-	NCT06500273	-
Chinese PLA General Hospital	UCART019	DLBCL	CD19	TRAC & B2M KO	CRISPR-Cas9	I/II	Jun. 1, 2017	2	Patient 1 Grade 4 CRS and cytopenia; patient 2 Grade 1 CRS; no GVHD or other toxicity	Patient 1 did not improve despite intensive medical intervention; patient 2 recovered without any treatment after cell infusion	NCT03166878	[200]
	UCART19 × 20/22	B-HM	CD19/20/22	TRAC KO	CRISPR-Cas9	I/II	Jan. 2, 2018	-	-	-	NCT03398967	-
	GC502	B-ALL	CD19/7	TRAC & CD7 KO	CRISPR-Cas9	I	Sep. 29, 2021	4	Grade 3 CRS 50%; Grade 2 CRS 50%; Grade 3 febrile neutropenia 100%, Grade 4 thrombocytopenia 25%, Grade 3 anemia 75%; no ICANS or GVHD	CR/CRi 75% (28 days); PR with extramedullary involvement 25% (1 month)	NCT05105867	[201]
	ET-901	B-NHL	CD19	TRAC & SPPL3 KO	CRISPR-Cas9	I/II	Sep. 6, 2023	9	Grade 3 + CRS 33.3%; Grade 1/2 ICANS 33.3%; no DLT or GVHD	ORR 100%; CR 66.6%; PR 33.3%	NCT06014073	[202]
Caribou Biosciences	CB-010	B-NHL	CD19	TRAC & PD-1 KO	ChRDNA-Cas9	I	May. 26, 2021	-	-	-	NCT04637763	[203]
	CB-011	MM	BCMA	TRAC & B2M KO	ChRDNA-Cas12a	I	Feb. 6, 2023	-	-	-	NCT05722418	[45]
	CB-012	AML	CCL-1	TRAC, B2M & PD-1 KO	ChRDNA-Cas12a	I	Dec. 5, 2023	-	-	-	NCT06128044	[204]
CRISPR Therapeutics AG	CTX110	B-HM	CD19	TRAC & B2M KO	CRISPR-Cas9	I/II	Jul. 22, 2019	32	CRS 56%; no CRS 3 + or GVHD; ICANS 9.4%, Grade 3 + ICANS 6.3%; Grade 3 + infections 12.5%; Grade 1 hypogammaglobulinemia 3.1%	Best ORR 67%; CR 41%; CR 19% (6 months); CR 7.4% (24 months)	NCT04035434	[205]
	CTX120	MM	BCMA	TRAC & B2M KO	CRISPR-Cas9	I	Jan. 22, 2020	-	-	-	NCT04244656	-
	CTX130	TCL	CD70	TRAC, B2M & CD70 KO	CRISPR-Cas9	I	Jul. 31, 2020	39	CRS 67%, Grade 1/2 CRS 59%, Grade 3 + CRS 7.7%; Grade 1/2 ICANS 10%	4 treatment-related deaths; ORR 46.2% (7.4 months)	NCT04502446	[206]
	CTX112	B-HM	CD70	TRAC, B2M, CD70, TGFBR2 & Regnase-1 KO	CRISPR-Cas9	I/II	Mar. 10, 2023	-	-	-	NCT05643742	[207]
Xinqiao Hospital of Chongqing	GC027	T-ALL	CD7	TRAC & CD7 KO	CRISPR-Cas9	I/II	Sep. 2, 2019	-	-	-	ChiCTR1900025311	[208]
	GC027	T-ALL	CD7	TRAC & CD7 KO	CRISPR-Cas9	I	Mar. 1, 2021	-	-	-	NCT04264078	-
Nanjing Medical University	CTA101	DLBL	CD19	TRAC & CD52 KO	CRISPR-Cas9	I	Dec. 1, 2019	-	-	-	NCT04026100	-
Xuzhou Medical University	CTA101	B-ALL	CD19/22	TRAC KO	CRISPR-Cas9	I	Dec. 10, 2019	-	-	-	NCT04154709	-
Zhejiang University	CTA101	B-ALL/NHL	CD19/22	TRAC & CD52 KO	CRISPR-Cas9	I	Jan. 8, 2020	6	CRS 100%, Grade 1/2 CRS 83.3%, Grade 3 CRS 16.7%; Grade 3 + infections 50%; no DLTs, GVHD or ICANS	1 treatment-related deaths; CR 83.3% (28 days); CR/CRi 62.5%	NCT04227015	[209]
	RD13-01	T-ALL/TCL/AML	CD7	TRAC, CD7 & HLA II KO	CRISPR-Cas9	I	Sep. 10, 2020	12	Grade 1/2 CRS 83%; no DLTs, GVHD, ICANS or Grade 3 + CRS	1 treatment-related deaths; ORR 81.8% (28 days); CR 63.6%;	NCT04538599	[49]
Fate Therapeutics	FT819	B-HM	CD19	TRAC KO	CRISPR-Cas9	I	Jul. 6, 2021	-	-	-	NCT04629729	[210]
Wugen, Inc.	WU-CART-007	T-ALL/LBL	CD7	TRAC & CD7 KO	CRISPR-Cas9	I/II	Jan. 14, 2022	18	Grade 3 + adverse events 44%; CRS 78%, Grade 1/2 CRS 72%, Grade 3 CRS 5.6%; Grade 1 ICANS 5.6%; no GVHD or cytopenia	CR/CRi 58%	NCT04984356	[211]
AvenCell Europe GmbH	Allo-RevCAR01-T-CD123	AML	CD123	TRAC and B2M KO	CRISPR-Cas9	I	Aug. 1, 2023	-	-	-	NCT05949125	[212]
Great Ormond Street Hospital For Children NHS Foundation Trust	TT52CAR19	B-ALL	CD19	TRAC & CD52 KO	CRISPR-Cas9	I	Aug. 12, 2020	6	Grade 2 CRS 33.3%; Grade 4 ICANS 16.7%; Grade 1 skin GVHD 16.7%	4 achieved molecular remission	NCT04557436	[25]
	BECAR-7	T-ALL	CD7	TRBC, CD52 & CD7 KO	Base editing	I	Jan. 4, 2022	3	Grade 1/2 CRS 100%, ICANS, cytopenia and GVHD	1 treatment-related deaths; ORR 66.7%	ISRCTN15323014	[98]
	BECAR-7	T-ALL	CD7	TRBC, CD52 & CD7 KO	Base editing	I	Apr. 19, 2022	-	-	-	NCT05397184	-
	BECAR-33	AML	CD33	—	Base editing	I	Jan. 11, 2023	-	-	-	NCT05942599	-
Beam Therapeutics	BEAM-201	T-ALL/LL	CD7	TRAC, CD52, PD-1 & CD7 KO	Base editing	I/II	May. 25, 2023	-	-	-	NCT05885464	-
Precision Biosciences	PBCAR0191	B-ALL/NHL	CD19	TRAC KO	ARCUS	I/II	Mar. 11, 2019	-	-	-	NCT03666000	[213]
	PBCAR20A	NHL/CLL/SLL	CD20	TRAC KO	ARCUS	I/II	Mar. 24, 2020	-	-	-	NCT04030195	-
	PBCAR269A	MM	BCMA	TRAC KO	ARCUS	I/II	Apr. 30, 2020	-	-	-	NCT04171843	-
	PBCAR19B	B-NHL	CD19	TRAC KO	ARCUS	I	Jun. 16, 2021	-	-	-	NCT04649112	-
Poseida Therapeutics	P-BCMA-ALLO1	MM	BCMA	TRBC & B2M KO	Cas-CLOVER	I/II	May. 5, 2022	21	CRS 43%; Grade 1 CRS 33%; Grade 2 CRS 10%; no CRS 3+; Grade 1 ICANS 14%; no DLTs or GVHD	ORR 90%; CR 28.6%	NCT04960579	[214]

R/R, relapsed/refractory; HM, hematopoietic malignancies; AML, acute myeloid leukemia; ALL, acute lymphoblastic leukemia; B-ALL, B-cell acute lymphoblastic leukemia; T-ALL, T-cell acute lymphoblastic leukemia; BCL, B Cell Lymphoma; TCL, T-Cell Lymphoma; LBCL, large B Cell Lymphoma; DLBCL, diffuse large B cell lymphoma; NHL, non-Hodgkin lymphoma; B-NHL, B Cell non-Hodgkin lymphoma; FL, follicular lymphoma; CLL, chronic lymphocytic leukemia; SLL, small lymphocytic lymphoma; MM, multiple myeloma; BCMA, B-cell Maturation Antigen; TRAC, T-cell Receptor Alpha Constant; TRBC, T-cell Receptor Beta Constant; B2M, β2-microglobulin; PD-1, programmed death-1; CLL-1, C-type lectin-like molecule-1; TGFBR2, transforming growth factor β receptor 2; GVHD, graft-versus-host disease; DLTs, dose-limiting toxicities; CRS, cytokine release syndrome; ICANS, immune cell associated neurotoxicity syndrome; CR, complete response; CRi, complete response with incomplete hematologic recovery; OS, overall survival; ORR, overall response rate; PR, partial response; DCR, disease control rate; DOR, duration of response; PFS, progression-free survival; RFS, relapse-free survival

Table 5.

Currently available clinical trials of genome-edited CAR-T cells therapy in solid tumors

Sponsors	Product	Tumor (R/R)	Target	Edit loci	Edit platforms	Phase	Study start	Patients (received ≥ one infusion)	Toxicity	Outcomes	Trial number	Cells source	References
Poseida Therapeutics	P-MUC1C-ALLO1	ST	MUC1	TRAC & B2M KO	Cas-CLOVER	I/II	Feb. 15, 2022	3	No adverse effects	Tumor elimination 100% (2 weeks); PR 33.3%	NCT05239143	Allogeneic	[215]
Guangdong Pharmaceutical University	Anti-MUC1 CAR-T cells	NSCLC	MUC1	PD-1 KO	CRISPR-Cas9	I/II	Feb. 1, 2018	6	No CRS or other adverse events	33.3% patients lung tumor size shrunk significantly (4 weeks)	NCT03525782	Autologous	[216]
Sun Yat-Sen University	MUC1-targeted CAR-T cells	BC	MUC1	PD-1 KO	CRISPR-Cas9	I/II	May. 17, 2019	12	Anemia and lymphopenia 33.3%; acute fever 50%; skin rash 41.7%; diarrhea, vomiting and urinary tract infection 16.7%; no DLTs, ICANS and Grade 3 + CRS	SD 41.7%	NCT05812326	Autologous	[217]
Chinese PLA General Hospital	MPTK-CAR-T	ST	mesothelin	TRAC & PD-1 KO	CRISPR-Cas9	I	Mar. 19, 2018	15	No CRS or ICANS; no observed on-target/off-tumor toxicity, autoimmune activity; effusion increase 26.7%	SD 46.7% (4 weeks); SD 13.3% (12 weeks); median PFS 7.1 weeks	NCT03545815	Autologous	[218]
	GC008t	ST	mesothelin	PD-1 KO	CRISPR-Cas9	I	Nov. 30, 2018	9	Grade 1 CRS 22.2%; no observed on-target/off-tumor toxicity, autoimmune activity.	2 patients repeat infusion failure; SD 57.1%, PFS 80 days; PR 28.6%, PFS 160 days	NCT03747965	Autologous	[219]
	TGFβR-KO CAR-EGFR T Cells	ST	EGFR	TGFBR KO	CRISPR-Cas9	I	Mar. 15, 2022	—	—	—	NCT04976218	Autologous	—
CRISPR Therapeutics AG	CTX130	RCC	CD70	TRAC, B2M & CD70 KO	CRISPR-Cas9	I	16-Jun-20	16	Grade 1/2 CRS 50%; infection 25%; no DLTs or Grade 3+; No hemophagocytic lymphohistiocytosis, GVHD or ICANS	CR 6.3% (36 months); SD 75%; DCR 81.3%	NCT04438083	Allogeneic	[220]
	CTX131	ST	CD70	TRAC, B2M, CD70, TGFBR2 & Regnase-1 KO	CRISPR-Cas9	I/II	Apr. 1, 2023	—	—	—	NCT05795595	Allogeneic	[221]
Zhejiang University	PD1-PSMA-CART cells	CRPC	PSMA	PD-1 KO	CRISPR-Cas9	I	Dec. 23, 2021	—	—	—	NCT04768608	Autologous	—
Allogene Therapeutics	ALLO-316	RCC	CD70	TRAC & CD52 KO	TALENs	I/II	Feb. 24, 2021	17	CRS 65%; Grade 3 CRS 6%; no GVHD or ICANS	Best PR 17.6%; ORR 12%; DCR 71%; CD70 + ORR 22% and DCR 100%	NCT04696731	Allogeneic	[222]

ST, solid tumors; NSCLC, Non-small Cell Lung Cancer; BC, Breast Cancer; RCC, Renal Cell Carcinoma; CRPC, Castration-Resistant Prostate Cancer; MUC1, Glycoprotein mucin 1; EGFR, epidermal growth factor receptor; PSMA, prostate-specific-membrane-antigen; TRAC, T-cell Receptor Alpha Constant; B2M, β2-microglobulin; PD-1, programmed death-1; TGFBR2, transforming growth factor β receptor 2; GVHD, graft-versus-host disease; DLTs, dose-limiting toxicities; CRS, cytokine release syndrome; ICANS, immune cell associated neurotoxicity syndrome; SD, stable disease; CR, complete response; PR, partial response; DCR, disease control rate; PFS, progression-free survival

Haematological malignancies

B-Cell malignancies: CD19 and CD22 targeting

As the pioneering target in CAR-T therapy, CD19 remains a cornerstone for treating R/R B-cell malignancies due to its near-universal expression (>95%) on malignant B cells. Benjamin R et al. recruited 7 children and 14 adults with R/R B-ALL across two multicenter phase 1 studies to evaluate the safety and antileukemic activity of UCART19 (NCT02808442) [114]. The researchers manufactured UCART19 by lentivirally transducing healthy donor T cells with a CAR construct encoding the 4G7 anti-CD19 scFv linked to 4-1BB costimulatory and CD3ζ activation domains. Additionally, the CAR was co-expressed with RQR8, a CD20 mimotope enabling rituximab-mediated elimination of UCART19 in cases of excessive toxicity. They electroporated TALENs mRNA targeting TRAC and CD52 genes into the cells to disrupt TCR expression (reducing GVHD risk) and confer resistance to alemtuzumab (enhancing UCART19 persistence during lymphodepletion). At 28 days post-infusion, 14 of 21 patients (67%) achieved CR/CRi; median DOR was 4.1 months, 6-month progression-free survival (PFS) was 27%, and overall survival (OS) was 55%. Subsequent phase 1 dose-escalation results (median follow-up: 12.8 months) showed an ORR of 48%, median DOR and relapse-free survival (RFS) of 7.4 months, PFS of 2.1 months, and OS of 13.4 months (NCT02746952) [108]. Jain et al. transduced donor-derived T cells with a lentiviral vector expressing an anti-CD22 CAR (anti-CD22 scFv-41BB-CD3ζ), similarly modified using Cellectis’ TALENs technology to disrupt TRAC and CD52 genes (NCT04150497) [194]. Their study found UCART22 therapy to have no unexpected toxicities: manageable grade 1/2 CRS occurred in 3/5 patients (60%), with no dose-limiting toxicities (DLTs), GVHD, or immune effector cell-associated neurotoxicity syndrome (ICANS); 2/5 patients (40%) achieved CRi. Neelapu SS et al. reported no DLTs or GVHD, limited grade 2 + ICANS, and low-grade CRS in trials for R/R large B-cell lymphoma (LBCL) or follicular lymphoma (FL) patients (NCT03939026) [197]. The median time from enrollment to therapy was 5 days, demonstrating rapid delivery of ALLO-501. The 6-month CR rate in LBCL patients (36.4%) and infection rates were comparable to autologous CAR-T trials. Analysis revealed that consolidation dosing had similar safety and efficacy to single dosing, with a 100% ORR and 75% CR rate in evaluable patients post-consolidation. To elucidate cell-extrinsic mechanisms of allogeneic CAR-T rejection, Jallouk AP et al. compared expansion kinetics across 11 recipients of allogeneic CAR-T therapy (NCT04416984) [199]. Their findings indicate that eliminating alloreactivity and enhancing adaptability improves expansion, persistence, and efficacy of allogeneic ALLO-501 A therapy.

ALLO-501 is primarily indicated for R/R LBCL and FL, exhibiting superior efficacy: in LBCL, the ORR is 61.5% with a CR rate of 46.2%; in FL, the ORR reaches 82.6% with a CR rate of 52.2%. The median duration of response is 23.1 months, and CAR-T cells persist for approximately 4 months. It achieves low toxicity profiles through a conditioning FCA regimen (fludarabine + cyclophosphamide + anti-CD52 antibody ALLO-647) to suppress host immunity, combined with TRAC and CD52 gene knockout technologies [197]. The incidence of grade ≥ 3 CRS is only 5%, with no GVHD observed, making it suitable for high-risk or elderly patients. In contrast, UCART19 is targeted at R/R B-ALL, with both ORR and CR/CRi rates of 48%; 75% of patients achieve minimal residual disease (MRD)-negative, but the median duration of response is only 7.4 months [108]. It employs a conditioning regimen of fludarabine + cyclophosphamide (with alemtuzumab added in some cases) and TALEN-mediated TRAC gene knockout, yet remains associated with significant safety risks: 80% of patients develop CRS (24% grade ≥ 3), requiring intensive supportive care. Clinical selection should consider disease type and risk stratification. ALLO-501 holds advantages in efficacy and safety, while UCART19 necessitates combination therapies to consolidate efficacy. Future efforts should focus on optimizing gene editing and combination strategies to improve long-term outcomes.

Plasma cell malignancies: BCMA targeting

Allogeneic CAR-T cells targeting B-cell Maturation Antigen (BCMA) are advancing rapidly. Koster CLBM et al. found that SLAMF7 (CS1) was overexpressed on MM cells, prompting them to disrupt this target in UCARTCS1 cells to diminish T-cell fratricide while simultaneously knocking out the TRAC locus to avoid GVHD (NCT04142619) [195]. Their preclinical studies showed that UCARTCS1 exerted potent antitumor activity against primary MM cells, MM cell lines, and in MM xenograft models. ALLO-715, an allogeneic anti-BCMA CAR-T cell therapy engineered to prevent GVHD and minimize rejection (NCT04093596) [198], was associated with grade ≥ 3 adverse events in 38 of 43 patients (88.0%). CRS occurred in 24 patients (55.8%), and infections in 23 patients (53.5%)—10 (23.3%) of grade ≥ 3 severity. Among evaluable patients, 11 (45.8%) achieved very good partial response (VGPR) and 6 (25%) achieved CR, with a median DOR of 8.3 months. Berdeja JG et al. developed CB-011 for MM patients, using ChRDNA-Cas12a editing to insert a BCMA-specific CAR into the TRAC locus (eliminating random integration) and integrating a B2M-HLA-E fusion gene into the B2M locus to evade host NK cell-mediated lysis (NCT05722418) [45]. P-BCMA-ALLO1, a non-viral transposon-generated allogeneic CAR-T cell, incorporates a human anti-BCMA CAR and an iCas9 safety switch [223]. In clinical testing, the Cas-CLOVER system was used to eliminate TRBC and B2M, demonstrating good tolerability with minimal CRS/ICANS risk in R/R MM patients. This trial reported PR as the best response in 3 of 5 patients (60%) [223]. Recent trials demonstrated that P-BCMA-ALLO1 exhibited a promising ORR (90%) and favorable safety profile when administered in combination with an optimized lymphodepletion regimen: CRS occurred in 43% of patients, with no cases of Grade 3 + CRS observed. Grade 1 ICANS was reported in 14% of patients, with no DLTs or GVHD documented (NCT04960579) [214].

ALLO-715 achieved an ORR of 70.8% and a CR of 25%, with intensive conditioning using the FCA regimen [198]. However, profound immunosuppression induced by this regimen significantly increased infection risk (53.3% overall infection rate, 28% grade ≥ 3), limiting its clinical utility. In contrast, P-BCMA-ALLO1 demonstrated a high ORR (90%) and a CR of 28.6% with markedly reduced toxicity through an optimized lymphodepletion regimen combined with gene-editing technologies, making it particularly suitable for patients with poor conditioning tolerance [214]. The core distinction lies in balancing immune clearance and cellular function: ALLO-715 trades intensive conditioning for deep remission while P-BCMA-ALLO1 achieves high efficacy with low toxicity via precise immune modulation of the FC regimen and preservation of TSCM populations. Future studies need to validate the long-term durability of P-BCMA-ALLO1 and explore combination strategies to improve CR rates. For R/R MM patients, individualized regimens should be selected based on performance status and prior treatment history—high-intensity FCA suits those tolerating profound immunosuppression, whereas optimized FC is preferable for balancing efficacy and safety.

Myeloid malignancies: CD123 targeting

CD123 (IL-3Rα), a cell surface target frequently expressed in hematologic and lymphoid malignancies, is present on leukemic blasts in approximately 80% of AML patients. Roboz GJ et al. used TALENs technology to disrupt TRAC and CD52 genes, employing an modified Toxicity Probability Interval (mTPI) design to evaluate the safety, tolerability, and preliminary antileukemic activity of UCART123 in R/R AML patients (NCT03190278) [193]. Prior in vitro studies demonstrated specific targeting of primary AML cells by UCART123 with minimal impact on normal hematopoietic progenitors.

T-cell malignancies: CD7 and CD70 targeting

CD7, a transmembrane glycoprotein expressed in over 95% of T-cell acute lymphoblastic leukemia (T-ALL) and T-cell lymphomas, represents an attractive therapeutic target for T-cell malignancies [224]. Unlike B-cell hematologic cancers, T-cell leukemias and lymphomas lack unique surface antigens. Allogeneic CAR-T cells targeting CD7 induce fratricide because CD7 is expressed on both healthy and malignant T cells. Multiplex base editing has entered cellular immunotherapy, enabling simultaneous editing of multiple genes. Promising initial clinical data have emerged from the first patients treated with this technology. To address fratricide, Great Ormond Street Hospital and Beam Therapeutics developed multiplex-edited allogeneic CAR-T cells from healthy donors. Chiesa R et al. electroporated activated T cells with coBE mRNA and sgRNAs targeting TRBC, CD52, and CD7 to generate allogeneic BECAR-7 T cells (ISRCTN15323014) [98]. Notable immunosuppression and cytopenia occurred in this trial. However, BE-CAR7 cells exhibited potent activity in all three patients, with antileukemic responses and deep remission observed in two patients within 28 days. Beam’s BEAM-201 additionally incorporates a mutant programmed death-1 (PD-1) immunosuppressive receptor designed to limit premature CAR-T exhaustion and enhance antitumor activity (NCT05885464). WU-CART-007 is a CRISPR/Cas9-edited allogeneic anti-CD7 CAR-T product with TRAC and CD7 deletion. It achieved a 58% composite complete response rate (CR/CRi) and demonstrated dose-dependent antitumor activity in R/R T-ALL patients (NCT04984356) [211]. Wang X et al. reported outcomes from a prospective GC027 study, the first-in-human universal CAR-T therapy for adult R/R T-ALL. After a single infusion without preconditioning, 80% of patients showed significant CAR-T expansion and sustained MRD-negative complete remission (ChiCTR1900025311) [208]. Hu Y et al. engineered allogeneic CD7-targeted CAR-T cells (RD13-01) using CRISPR to disrupt TRAC (preventing GVHD), CD7 (avoiding fratricide), and CIITA (resisting rejection). They also integrated an NK-inhibitory receptor-CD28 fusion to reduce NK-mediated lysis and a common γc cytokine receptor to enhance antitumor function (NCT04538599) [49]. At day 28 post-infusion, 9 of 11 patients (81.8%) achieved objective remission, including 7 patients (63.6%) with complete remission. After a median follow-up of 10.5 months, 4 patients maintained remission. Expansion of CD7-negative normal T cells was detected post-infusion.

CD70, a co-stimulatory protein in the TNF receptor family, shows limited expression on activated lymphocytes but high expression in many T-cell lymphomas (TCLs). Iyer SP et al. inhibited CD70 expression to mitigate fratricide and enhance cellular performance (NCT04502446) [206]. CTX130, as the first allogeneic anti-CD70 CAR-T product, was administered to 39 R/R TCL patients and observed a 46.2% ORR. In cohorts receiving dose level 3 or higher (n = 31), 51.6% (16/31) achieved overall response, including 19.4% (6/31) with CR and 32.3% (10/31) with PR.

Collectively, BE-CAR7 employed intensive preconditioning with alemtuzumab and achieved an ORR of 66.7%. Despite its precise base editing, it exhibited low efficiency [98]. CD52 knockout induces severe immunosuppression, restricting its application to bridging transplantation scenarios. WU-CART-007 utilized CRISPR for efficient editing of TRAC/CD7, combined with standard lymphodepletion, and achieved an ORR of 58% [211]. Its clear dose dependency and manageable toxicity render it more suitable for clinical translation. RD13-01 incorporated CRISPR-mediated multiplex editing (TRAC/CD7/CIITA) with integrated functional enhancement elements [49]. Even under mild preconditioning, it attained an 81.8% ORR. The allogeneic cells, modified for NK resistance, demonstrated prolonged persistence, yielding a 63.6% CR rate with balanced safety and durability. CTX130 targeted CD70 and featured simplified editing without target gene knockout [206]. It showed prominent safety under standard lymphodepletion but achieved a more limited ORR of 46.2%, which was attributed to target heterogeneity. The persistence of the allogeneic cells requires further validation. Future advancements may hinge on multiplex gene editing (e.g., incorporating PD-1 mutations and metabolic pathway enhancements) and functional optimization (e.g., γc receptor overexpression) to boost efficacy.

Dual-targeting strategies

Dual-target CAR-T cells may decrease the risk of relapse induced by single-target loss, similar to combination chemotherapy regimens. The aforementioned data indicate that CD7-targeting CAR-T therapy can create a therapeutic window by inhibiting the alloreactivity of patients’ own immunocytes to allogeneic CAR-T cells, achieving therapeutic effects in patients with T-ALL. Based on these studies, Ge J et al. developed GC502, a CD19/CD7 dual-targeted allogeneic CAR-T therapy for B-cell malignancies, disrupting the TRAC and CD7 loci to avoid GVHD and “cannibalism,” respectively [225]. To optimize anti-tumor efficacy, they incorporated a T-cell enhancer into the CAR construct in addition to disrupting TRAC and CD7. GC502 demonstrated robust anti-tumor efficacy and potential to suppress host anti-graft reactions in preclinical studies. Subsequently, Li S et al. reported early clinical results of GC502 in R/R B-ALL patients (NCT05105867) [201]. At day 28 post-infusion, 3 of 4 patients achieved CR/CRi, showing promising early outcomes with a manageable safety profile. Aranda-Orgilles B et al. transfected donor-derived T cells with a bicistronic lentiviral vector to express CARs targeting CD20 and CD22. Following TRAC and CD52 knockout via TALENs technology, UCART20 × 22 exhibited strong and sustained activity in vitro and in vivo [226]. Abramson JS et al. then conducted the first-in-human Phase 1/2 dose-finding and expansion study of UCART20 × 22 in R/R B-cell NHL (NCT05607420) [196]. Current data show UCART20 × 22 expansion in all patients, with no DLTs, GVHD, or ICANS observed; CRS was < Grade 3. The CR rate was 66.6% (2 of 3 patients), and the PR rate was 33.3%. Hu Y et al. developed the first CRISPR-edited universal CD19/CD22 dual-target CAR-T product (CTA101), disrupting TRAC and CD52 loci to achieve CR in B-ALL patients, with no observed gene editing-associated genotoxicity or chromosomal translocations (NCT04227015) [209]. At day 28 post-infusion, 5 of 6 patients (83.3%) achieved CR (median follow-up: 4.3 months), with 3 of 5 CR/CRi patients (60%) attaining MRD negativity.

GC502 avoids additional immunosuppression and relies on CD19/CD7 dual-targeting to reduce rejection, but this strategy may restrict cell expansion [201]. Its 75% CR rate is constrained by insufficient host immune clearance. UCART20 × 22 employs the intensive FCA lymphodepletion regimen combined with TALENs-mediated TRAC/CD52 editing, achieving a 66.7% CR rate [196]. However, it carries infection risks, and TALENs exhibit slightly inferior editing efficiency compared to CRISPR. CTA101 utilizes high-efficiency CRISPR editing of TRAC/CD52, combined with anti-CD52 mAb pretreatment to balance lymphodepletion intensity [209]. Coupled with allogeneic T cells from young donors, it achieves an 83.3% CR rate with more durable responses and no gene-editing associated toxicity. Among the three, CTA101 demonstrates superior overall performance due to precise editing, optimized lymphodepletion, and high-quality cell sources. UCART20 × 22 is more suitable for NHL patients who require intensive lymphodepletion, while GC502 provides a treatment option for cases where low toxicity is prioritized.

Enhanced function and novel platforms

Caribou Biosciences and CRISPR Therapeutics AG typically utilize CRISPR systems to disrupt both endogenous TRAC and B2M loci. O’Brien S et al. employed ChRDNA-Cas9 technology to disrupt TRAC and PD-1 genes for developing off-the-shelf anti-CD19 CAR-T therapy (CB-010). Compared to CAR-T cells without PD-1 knockout, CB-010 demonstrated statistically significant preclinical survival benefits in B Cell non-Hodgkin lymphoma (B-NHL) subtypes with prolonged therapeutic efficacy in vivo (NCT04637763) [203]. CB-012, a next-generation ChRDNA-Cas12a-edited allogeneic anti-C-type lectin-like molecule-1 (CLL-1) CAR-T therapy for R/R AML adults, was developed by Daver NG et al. through knocking out TRAC, PD-1, and B2M genes. This approach enabled site-specific insertion of a fully human anti-CLL-1 CAR into the TRAC locus and fusion of a B2M-HLA-E-peptide into the B2M locus (NCT06128044) [204]. For CTX110 (allogeneic CD19-directed CAR-T), McGuirk JP et al. used CRISPR/Cas9 editing to disrupt TRAC and B2M loci (NCT04035434) [205]. In heavily pretreated R/R LBCL patients, CTX110 achieved a best ORR of 67% (18/27) and best CR rate of 41% (11/27), with durable remissions and favorable safety during dose escalation. Nearly half of patients maintained CR for ≥ 6 months.

Terrett JA et al. further engineered CTX112 with dual genomic edits: Simultaneous disruption of TGFBR2 to block TGF-β-mediated immunosuppression in the TME and Regnase-1 to enhance functional persistence by increasing expansion capacity associated with central memory phenotypes (NCT05643742) [207]. Their preclinical tests revealed that CTX112 exhibited augmented expansion, enhanced effector cytokine secretion, and improved resistance to exhaustion.

The Chinese PLA General Hospital developed UCART019 using CRISPR/Cas9 to disrupt TRAC and B2M genes, reporting outcomes in two R/R DLBCL patients (NCT03166878) [200]. Notably, a patient with high tumor burden developed refractory grade 4 CRS despite intensive intervention. Conversely, a patient with lower tumor burden experienced self-resolving grade 1 CRS that resolved spontaneously over seven days post-infusion. These findings demonstrate a direct correlation between CRS severity and baseline tumor burden. Through genome-wide CRISPR screening, Wu Z et al. identified SPPL3, a key glycosylation-regulating gene that acts as a negative regulator of cell surface glycan synthesis. Knocking out SPPL3 resulted in the formation of a dense glycocalyx layer, which conferred protection against multiple immune recognition mechanisms. This included reduced surface accessibility of HLA-I molecules, thereby diminishing host T cell recognition and clearance via TCR-mediated mechanisms. It also decreased the exposure of NK cell-activating ligands, enhancing resistance to NK cell-mediated killing, and reduced Fas surface availability, inhibiting activation-induced cell death (AICD) through the Fas/FasL pathway (NCT06014073) [202]. The team developed TRAC/SPPL3 knockout anti-CD19 CAR-T cells (ET-901), which in a clinical trial involving 9 patients with B-NHL produced an ORR of 100% and a CR rate of 66.7%. Ottaviano G et al. engineered TT52CAR19 T cells via multiplex CRISPR editing of TRAC and CD52. TRAC disruption prevents GVHD, while CD52 elimination confers alemtuzumab-mediated survival advantage. In pediatric B-ALL patients (n = 6), 66.7% (4/6) achieved molecular remission, demonstrating potent anti-leukemic activity (NCT04557436) [25].

Ehninger A et al. reported the preclinical development of Allo-RevCAR01-T-CD123, a switchable allogeneic CAR-T therapy for CD123-positive malignancies (NCT05949125) [212]. The RevCAR platform utilizes a dual-component reverse system comprising universal CAR-T cells expressing a short, non-immunogenic peptide binding domain that recognizes no human cell surface antigen, and soluble adaptors termed targeting modules (TMs) containing an scFv domain. Selection of R-TM123 with a short half-life enables CD123 engagement, while allowing rapid system shutdown via TM withdrawal to prevent persistent toxicity from sustained CAR-T activation. Their studies confirmed effective lysis of CD123⁺ AML cells by Allo-RevCAR-T cells redirected with R-TM123 in both in vitro and in vivo models. Additionally, FT819 represents the first clinically applied iPSC-derived off-the-shelf CAR-T therapy. This therapy targets B-cell malignancies including R/R BCL, CLL, and precursor B-ALL (NCT04629729) [210].

The unique ARCUS editing technology is authorized by Duke University. In preclinical disease models, PBCAR series therapies including CD19-targeted PBCAR0191 (NCT03666000) [213] and PBCAR19B (NCT04649112), CD20-targeted PBCAR20A (NCT04030195), BCMA-targeted PBCAR269A (NCT04171843), demonstrated efficacy in eliminating tumor cells and reducing tumor volume in vivo. Moreover, Precision Biosciences reports no evidence of GVHD in rigorous preclinical models.

CTX110, utilizing a standard FC lymphodepletion regimen combined with TRAC/B2M dual-gene knockout, achieved a 67% ORR in patients with LBCL [205]. This outcome confirms the synergistic effect between conventional preconditioning and gene editing. UCART019, derived from umbilical cord blood T cells, enables rapid manufacturing [200]. Although one patient achieved overall response (50%), another patient with high tumor burden developed grade 4 CRS without remission, suggesting that high tumor burden may offset the advantages of its cell source. For ET-901, the CRISPR screening led to the identification of the novel target SPPL3, and dual TRAC/SPPL3 knockout resulted in a 100% ORR in B-NHL [202]. This success highlights that editing efficiency (via precise targeting of immunoregulatory pathways) is more critical than the number of edited loci. TT52CAR19 T cells, engineered with TRAC/CD52 knockout and combined with alemtuzumab for host immune cell depletion, achieved a 66.7% ORR in pediatric B-ALL, demonstrating that multi-target editing combined with drug synergy can enhance CAR-T persistence [25]. In terms of cell sources, UCART019 cells face host rejection, while other products, though with unspecified sources, reduce GVHD risk via TRAC knockout. Overall, ET-901’s novel target discovery and TT52CAR19’s multi-mechanistic synergy show optimal potential, suggesting that integrating high-efficiency editing technologies (e.g., ARCUS or Cas12a), personalized lymphodepletion regimens (adjusted by tumor burden), and screening for immune evasion targets will be key to overcoming the bottlenecks in off-the-shelf CAR-T therapy.

Solid tumours

Overcome trafficking and TME barriers

The tumor microenvironment (TME) of solid tumors synergistically impairs CAR-T function through multiple mechanisms, including the infiltration of immunosuppressive cells and dysregulated metabolic stress [227]. Immunosuppressive cell populations constitute a central barrier within this milieu. Specifically, M2-polarized TAMs secrete TGF-β and IL-10, directly suppressing the cytotoxic activity of CAR-T cells [228]. Engineering CAR-T cells to secrete cytokines such as IL-12 [229], IL-15 [230], IL-21 [231], IL-2 and IFN-γ can enhance their anti-tumor activity [232].

Furthermore, myeloid-derived suppressor cells (MDSCs) deplete essential amino acids like arginine and cysteine, leading to CAR-T cell exhaustion due to metabolic deprivation [233], necessitating the blockade of key metabolic enzyme expression. Additionally, regulatory T cells (Tregs) competitively bind IL-2, thereby blocking proliferation signals for CAR-T cells [234]. To prevent TGF-β from activating Tregs and suppressing T-cell killing capacity, Chen et al. designed CAR-T cells capable of targeting and eliminating PD-1 and TGF-β to enhance anti-tumor potency [235]. Moreover, Tang N et al. utilized CRISPR/Cas9 technology to knock out the TGF-β receptor 2 (TGFBR2) gene in CAR-T cells, demonstrating in pancreatic cancer models that blocking the TGFβ signaling pathway significantly improves CAR-T cell ability to clear solid tumors [236]. Recent research by Zheng S et al. showed that engineering a TB15 chimeric receptor, comprising the extracellular domain of TGFBR2 fused to the intracellular signaling domain of IL-15, reverses TGF-β inhibitory signals into proliferative stimulatory signals. This strategy was validated in ovarian and liver cancer models to dually counteract immunosuppression and enhance CAR-T function, achieving a 95% tumor suppression rate and doubling survival time [237]. This TB15 inversion receptor represents a more advanced signal reprogramming concept, combining the dual advantages of “defense against suppression plus active activation,” positioning it as a key direction for next-generation intelligent CAR-T.

Concurrently, metabolic stress within the TME further exacerbates CAR-T cell dysfunction. Hypoxia stabilizes hypoxia-inducible factor-1α (HIF-1α), which not only promotes tumor angiogenesis but also induces CAR-T cell exhaustion by upregulating programmed death-ligand 1 (PD-L1) expression [238, 239]. Moreover, under hypoxic conditions, impaired adenosine kinase activity results in elevated extracellular adenosine concentrations; this adenosine binds to the A2A receptor on CAR-T cells, inhibiting TCR signaling [238, 240]. A2A receptor-deficient CAR-T cells can maintain activity in this context [241]. However, the substantial lactate produced by tumor cell aerobic glycolysis acidifies the microenvironment, directly inhibiting CAR-T cell migration and cytotoxic function. Latest research indicates that overexpressing the transcription factor Foxp3 in CAR-T cells, independent of its immunosuppressive function, drives mitochondrial dynamics remodeling by binding Drp1 protein. This redirects metabolic flux from glycolysis towards lipid metabolism, enhancing adaptability within the hypoxic microenvironment while simultaneously reducing HIF-1α accumulation and PD-L1-induced exhaustion [242].

The structural heterogeneity of solid tumors presents multiple barriers to CAR-T infiltration through aberrant vasculature, a dysregulated chemokine network, and extracellular matrix (ECM) obstacles, resulting in limited tumor infiltration and homing [243]. Intratumoral aberrant vascular branching and hemodynamic disturbances hinder CAR-T cell migration within tumor regions. Elevated interstitial fluid pressure further creates a physical barrier preventing CAR-T cell penetration into the tumor core [244]. Remodeling the tumor vasculature using short-term, low-dose anti-angiogenic agents such as anti-VEGF antibodies can improve blood perfusion and reduce interstitial fluid pressure [245], potentially facilitating CAR-T cell infiltration.

Dysregulation of the chemokine network compounds migration defects. Tumor cells frequently downregulate immune cell homing chemokines including CCL4, CXCL9, CXCL10, and CXCL11 (attract CD8⁺ T cells); CX3CL1 and CCL16 (attract CD8⁺ T cells and NK cells); CCL3 and CCL21 (attract T-cell and DC cells) [227]. Conversely, they overexpress pro-tumor chemokines that recruit immunosuppressive cells such as CCL5, CCL17, and CCL22 recruiting Tregs; CXCL1 and CXCL5 recruiting MDSCs; CXCL8 recruiting neutrophils; and CXCL12 recruiting TAMs. These chemokines variably promote immune escape [227]. This imbalance leads to mismatched migratory signals for CAR-T cells, preventing immune cell infiltration. One effective strategy involves genetically engineering CAR-T cells to express receptors matching chemokines secreted in the TME, actively guiding them towards the tumor site. For instance, overexpressing chemokine receptors like CXCR2, CCR4, or CCR5 in CAR-T cells enhances their migration towards tumors secreting the corresponding ligands [246–248].

The dense ECM, composed of collagen, fibronectin, and proteoglycans secreted by cancer-associated fibroblasts (CAFs), directly restricts CAR-T cell migration within solid tumors [249]. Strategies targeting key CAFs surface proteins such as fibroblast activation protein (FAP), for example using FAP inhibitors or FAP-CAR-T cells, may disrupt this physical barrier [250]. Furthermore, heparanase degrades heparan sulfate, a major ECM component, locally degrading the matrix to enhance penetration [251].

Clinical trials of genome-edited CAR-T in solid tumours

Current development of gene-edited CAR-T therapies for solid tumors focuses on engineering strategies targeting tumor-specific antigens to overcome immunosuppression. Key approaches include disrupting immune checkpoints (PD-1), TCR/MHC complexes, and inhibitory pathways (TGFBR2/Regnase-1) via CRISPR-Cas9/TALENs. These modifications enhance tumor specificity and persistence while mitigating allo-rejection, as evidenced in clinical trials targeting MUC1-C, mesothelin, and CD70. Such engineered cells represent a promising avenue to improve therapeutic efficacy.

Combination immunotherapy may be an effective approach to treating solid tumors. Glycoprotein mucin 1 (MUC1) is a transmembrane protein derived from the mucin family, which is a heterodimer of MUC1-N and the oncogenic subunit MUC1-C, related to tumor metastasis and progression, especially in gastric cancer [252]. MUC1-C is broadly present in tumors caused by the loss of cell polarity, exposure via hypoglycosylation and MUC1-N shedding, thus it may be more suitable as a tumor-selective target than MUC1-N. P-MUC1C-ALLO1 is an allogenic CAR-T targeting MUC1-C that uses the Cas-CLOVER system to disrupt the expression of TCR and MHC proteins, and showed robust infiltration and activity in triple negative breast cancer (TNBC) and ovarian cancer xenografts (NCT05239143) [215]. The early phase I experience indicated that P-MUC1C-ALLO1 is safe and tolerable with an early signal of efficacy at the low dose, and 1 of 3 patients (HR⁺, Her2^- Breast cancer) had a PR, and no toxicities related to P-MUC1C-ALLO1 were observed. In addition, Chen S et al. used SM3 scFv to construct MUC1-specific CAR, then knocked out PD-1 gene by CRISPR-Cas9 system in CAR-T cells and validated by sequencing (NCT03525782) [216]. The symptoms of all advanced Non-small Cell Lung Cancer (NSCLC) patients receiving were significantly improved in the first two weeks after Anti-MUC1 CAR-T cells infusion, and the size of lung tumor in 2 of 6 patients shrank remarkably within 4 weeks after infusion, which indicated that the curative effect of combined therapy was case specific. Despite half of the patients (3/6) having elevated cytokine levels, CRS was not observed, and no other adverse reactions occurred during the treatment of all patients. Similarly, Lin YG et al. constructed MUC1-specific CARs using 5E5 scFv, subsequently disrupted the PD-1 gene with the CRISPR-Cas9 system (NCT05812326) [217]. No serious CRS or predefined DLT was observed during the treatment of all advanced breast cancer (BC) patients, and 5 of 12 patients (41.7%) had stable disease (SD), which indicated that MUC1-targeted CAR-T cells therapy can control the progress of some advanced BC, and was safe and well-tolerated. In addition, pharmacokinetic studies showed that the improvement of clinical results was related to the high level of circulating CAR⁺ copy on the 7th day.

Mesothelin is a glycoprotein located on the cell membrane, which has generally limited distribution on normal tissues such as peritoneal, pleural, and pericardial mesothelial surfaces [253], while relatively specific expression at high levels in cancers including TNBC, pancreatic cancer, cholangio carcinoma, ovarian cancer and pleural mesothelioma, and is associated with higher tumor grade, later stage and poor prognosis, which makes mesothelin a potential important target for tumor treatment [254]. Wang Z et al. selected P4 scFv to construct the anti-mesothelin CAR to avoid potential CAR transgene immunogenicity, and generated mesothelin-directed 28ζ CAR-T (MPTK-CAR-T) cells with PD-1 and TCR disruption by CRISPR-Cas9 technology (NCT03545815) [218]. In a dose-escalation study, they showed that the CAR-T cells with PD-1 disruption would not cause unrestricted proliferation or persistence, nor would they elicit unexpected safety issues, and demonstrated that the natural TCR is conducive to the proliferation of CAR-T cells in solid tumors, while the crosstalk between endogenous TCR and CAR signaling needs to be further addressed. Based on these results, they subsequently developed GC008t for patients with mesothelin-positive advanced solid tumors in another pilot dose escalation study, which further demonstrated that the expansion and persistence of CAR-T cells with PD-1 disruption were not improved remarkably even in the case of natural TCR and lymphodepletion (NCT03747965) [219].

Clear cell renal cell carcinoma (ccRCC) is the most prevalent subtype of kidney cancer. Pal SK et al. reported the persistent response and stability observed in a Phase 1 dose-escalation study of CD70-targeting CTX130 (NCT04438083) [220], which previously achieved remarkable efficacy in R/R TCL patients. 1 (6.3%) of 16 patients achieved CR sustained over 36 months, and 12 (75%) patients achieved SD, reflecting a disease control rate (DCR) of 81.3%. The genetic modification of CTX131 is to carry out CRISPR editing on the basis of CTX130, and additionally disrupt TGFBR2 and Regnase-1, similar to the procedure of CTX112 (NCT05795595) [221]. Preclinical studies indicated that disrupting the synergistic effect of TGFBR2 and Regnase-1 can enhance the anti-tumor activity, inhibit tumor growth, prolong survival and increase the central memory T cell populations [207]. Another anti-CD70 allogenic CAR-T cell product for patients with ccRCC is ALLO-316, which utilizes TALENs gene editing to knock out TRAC and CD52 loci (NCT04696731) [222]. 3 of 17 patients (17.6%) achieved best PR at all time points. Among 9 patients diagnosed with CD70⁺ tumor, ORR was confirmed in 22%, and the DCR was 100%.

In terms of response rates, allogeneic products exhibit the advantage of multi-gene editing: although CTX130 (with TRAC, B2M & CD70 knockout) only achieved a 6.3% CR rate, one patient maintained CR for over 36 months, demonstrating its potential for deep and sustained remission [220]. ALLO-316 (with TRAC & CD52 knockout) showed a 17.6% PR rate due to its relatively limited editing scope, but lacked long-term remission data [222]. P-MUC1C-ALLO1 (with TCR & B2M knockout), leveraging the high specificity of MUC1-C, achieved a PR rate of 33.3%, the highest among allogeneic products [215]. Among autologous products, GC008t (with PD-1 knockout) yielded a 28.6% PR rate [219], outperforming the anti-MUC1 CAR-T (with PD-1 knockout), suggesting that TCR retention may enhance infiltration in solid tumors [216]. Regarding durability, CTX130, with its triple-editing strategy, stands out as the only product achieving >36 month CR. This confirms that the combination of “anti-rejection (TRAC + B2M) and anti-exhaustion (CD70)” is critical for long-term survival of CAR-T cells [220]. P-MUC1C-ALLO1, supported by continuous stimulation from MUC1-C, exhibits better durability than MPTK-CAR-T (mesothelin-targeted), though long-term clinical data remain limited [218]. Among autologous products, patients with PR in the GC008t group had a PFS of 160 days, longer than the SD duration observed in MUC1-targeted CAR-T [217, 219]. This indicates that moderate synergy between TCR and CAR signaling (rather than complete TCR knockout) may be more conducive to the long-term persistence of CAR-T cells in solid tumors. In terms of toxicity profiles, autologous products generally exhibited lower toxicity than allogeneic ones (except for grade 1 CRS in 22.2% of GC008t patients). However, their response rates and durability are overall inferior to those of allogeneic products with optimized editing (e.g., CTX130), which is partially attributed to the variable quality of patient-derived T cells.

Overall, the optimal strategy for solid tumor CAR-T therapy should satisfy three criteria: selection of targets with high tumor specificity (e.g., MUC1-C, CD70); editing strategies covering both “anti-rejection” and “anti-exhaustion” (e.g., CTX130’s TRAC, B2M & CD70 knockout); and synergy between lymphodepletion regimens and editing strategies to provide a proliferative niche for CAR-T cells. Future allogeneic products could further enhance solid tumor penetration by incorporating edits targeting genes resistant to the tumor microenvironment (e.g., TGFBR2). For autologous products, optimizing the signaling balance between TCR and CAR is essential to break through the current efficacy ceiling.

Future perspectives for solid tumors

Although allogeneic CAR-T therapy faces significant challenges in the treatment of solid tumors, numerous innovative strategies are outlining a promising blueprint for its future application. Future advances will extend beyond the optimization of CAR constructs alone, shifting toward multidimensional and systematic collaborative efforts.

The exploration of novel targets is fundamental to overcoming antigen heterogeneity. In the field of solid tumor CAR-T therapy, several emerging targets have demonstrated considerable clinical potential. For instance, the Claudin18.2-targeted therapy CT041 achieved an ORR of 37.8% and a DCR of 75.5% in 98 patients with advanced gastrointestinal cancers, with a median PFS of 4.4 months and median OS of 8.4 months. Among gastric cancer patients with measurable lesions, the ORR reached 57.4% and DCR was 83.0%, coupled with a favorable safety profile that included no severe neurotoxicity or treatment-related deaths [255]. Furthermore, IM96, which targets GUCY2C, was administered to 20 patients with metastatic colorectal cancer (55% of whom had liver metastases). Among the 19 evaluable patients, the DCR was 73.7% and ORR was 26.3% [256]. Only one case of ≥ grade 3 CRS was observed. Another promising target is GPC3. The GPC3-directed CAR-T product C-CAR031 was evaluated in 24 patients with advanced hepatocellular carcinoma, 83.3% of whom had extrahepatic metastases. Among 22 evaluable patients, 90.9% exhibited tumor shrinkage, with a DCR of 90.9% and an ORR of 50.0% [257]. Only one instance of grade 3 CRS was reported, indicating manageable safety. Additionally, AIC100, an ICAM-1-targeted therapy, was tested in 10 patients with thyroid cancer. Of the 9 evaluable subjects, the ORR was 22% (including one complete response) and DCR was 56%, with no serious adverse events reported [258]. These targets exhibit higher tumor specificity or are closely associated with tumor progression, collectively providing a new arsenal for the design of more precise and safer “off-the-shelf” therapeutic strategies.

A key to overcoming immunosuppressive solid tumor microenvironments and enhancing the efficacy of allogeneic CAR-T cells lies in developing innovative combination strategies. Among these, combination with local radiotherapy is highly promising. Radiotherapy not only directly kills tumor cells and exposes neoantigens but also remodels the local immune microenvironment, upregulating chemokines and tumor antigen expression. This may promote CAR-T cell recruitment and activation and induce an abscopal effect. Research by Sodji QH et al. demonstrated that low-dose radiopharmaceutical therapy (RPT) combined with CRISPR-edited GD2 TRAC-CAR T cells enhanced antitumor efficacy in solid tumors by improving CAR-T infiltration, cytokine secretion, and upregulation of Fas receptor on tumors. Pretreatment with ¹⁷⁷Lu-NM600 (a radiopharmaceutical agent) delivering 1.8 Gy RPT followed by CAR-T infusion resulted in complete regression in a neuroblastoma model, with a 100% survival rate at 125 days [259]. Additionally, combination with immunomodulatory agents, such as PD-1/L1 inhibitors, represents another major direction, aiming to reverse immune checkpoint-mediated exhaustion of CAR-T cells and achieve synergistic efficacy [260]. Meanwhile, oncolytic viruses (OVs) expressing immunostimulatory factors are also considered ideal partners. They not only selectively replicate within and lyse tumor cells but also convert “cold” tumors into “hot” ones, significantly improving CAR-T cell infiltration and functional persistence [261]. These multimodal combination strategies, by targeting different vulnerabilities within the tumor microenvironment, hold promise for synergistically enhancing the penetration and cytotoxic efficacy of genomically edited allogeneic CAR-T therapies in solid tumors, representing a critical direction toward clinical breakthroughs.

Furthermore, synthetic biology-driven “smart” CAR-T designs will endow cells with unprecedented sensing and response capabilities. Examples include engineering CAR-T cells to secrete specific cytokines (e.g., IL-12, IL-15) or ligands within the tumor niche, enabling self-sustained function and microenvironment remodeling [229–232]. Alternatively, regulatory circuits sensing the TME—such as CARs activated by specific TME factors (e.g., hypoxia, high ATP, or certain cytokines)—can achieve localized, tumor-specific activation [262].

The future of allogeneic CAR-T therapy in conquering solid tumors depends on the integration and iteration of new targets, combination strategies, intelligent cell designs, and in vivo reprogramming technologies (discussed in detail later). Through multidisciplinary and multi-technology collaborative innovation, we may ultimately break through the barriers posed by solid tumors and bring hope to more patients.

In vivo CAR-T engineering

Traditional CAR-T cell therapy relies on peripheral blood collection of patient T cells, ex vivo genetic engineering, and infusion of the expanded cells. This process is complex, time-consuming, and costly. In contrast, in vivo CAR-T technology employs a disruptive strategy. Targeted delivery systems directly introduce genetic material encoding the CAR into the patient’s T cells, enabling in situ T cell reprogramming. This approach bypasses ex vivo cell manipulation steps, reducing the treatment cycle from weeks to days while substantially lowering production costs.

The targeted delivery system represents the core breakthrough of in vivo CAR-T technology. This technology primarily utilizes two types of genetic payloads, mRNA vectors and viral vectors (LV/AAV). mRNA vectors enable direct expression of the CAR protein in the cytoplasm, providing transient expression, typically lasting several days, which reduces the risk of long-term toxicity. This feature makes mRNA particularly suitable for autoimmune disease treatment [263]. Viral vectors, conversely, mediate long-term stable CAR expression, rendering them suitable for malignant tumor therapy. Once the delivery system successfully introduces the genetic payload into T cells, the T cells utilize their own protein synthesis machinery to express CAR molecules on their surface, creating chimeric antigen receptors. These in situ generated CAR-T cells can immediately recognize and attack target cells. However, the success of in vivo CAR-T technology critically depends on the delivery efficiency and targeting specificity of the vector system. Optimizing vector design and surface functionalization strategies can significantly improve transfection efficiency. For instance, Wang Y et al. developed a T cell-specific membrane fusion-type pseudoviral particle system (T-FVLPmCAR) [264]. This system efficiently generated CAR-T cells in humanized mouse models using less than 1 µg of CAR mRNA, demonstrating significant dose-dependent anti-tumor efficacy against B-cell lymphoma.

Zhao G et al. developed a viral-mimetic fusogenic nanovesicle (FuNV) delivery system [265]. This system achieves efficient in vivo delivery of CAR genes by mimicking the viral membrane fusion mechanism. FuNVs incorporate engineered viral fusion proteins, such as those derived from measles or reovirus, on their surface. These fusion proteins target T cells via specific ligands like anti-CD3 single-chain antibodies. Upon contact with T cells, the fusion proteins mediate fusion of the nanovesicle with the cell membrane, directly delivering the encapsulated CAR protein or mRNA into the cytosol. This membrane fusion delivery mechanism bypasses degradation via the endosomal-lysosomal pathway, significantly enhancing delivery efficiency. In B-cell lymphoma models, a single injection of FuNVs loaded with anti-CD19 CAR generated functional CAR-T cells in vivo. Studies revealed that these in vivo generated CAR-T cells possessed anti-tumor activity comparable to conventional CAR-T cells. Furthermore, their transient nature significantly reduced risks like CRS. Additionally, combining them with an αOX40 agonistic antibody further enhanced anti-tumor efficacy without increasing toxicity. This finding provides a new strategy for developing safer CAR-T therapies.

Hunter TL et al. developed CD8-targeted lipid nanoparticles (CD8-L829-tLNP) based on a novel ionizable lipid, L829 [263]. This technology efficiently delivers mRNA encoding anti-CD19 or anti-CD20 CAR to CD8⁺ T cells in vivo, enabling transient expression of functional CAR receptors. It achieved precise reprogramming of CD8⁺ T cells in humanized mice, significantly depleting B cells and effectively controlling tumor growth. Similarly, Zhang Z et al. developed novel cardiolipin-mimetic lipid nanoparticles (CAMP LNPs) [266]. They designed novel CAMP lipids, such as PL40, which form polyhedral rigid nanoparticles by mimicking cardiolipin structure. This design significantly enhanced T cell targeting and mRNA transfection efficiency, achieving a 100-fold improvement over conventional carriers. It enabled antibody-free in vivo CAR-T cell therapy, circumventing the complexity of antibody modification and the risk of T cell exhaustion. This strategy offers a novel approach for treating inflammatory aging diseases.

Recently, Xu J et al. utilized nanobody-decorated lentiviral vectors (ESO-T01). They modified the vectors by mutating the viral envelope protein to narrow tropism, overexpressing CD47 on the membrane to evade immune clearance, integrating anti-TCR nanobodies for T cell targeting, and knocking down MHC-I to reduce immunogenicity. This system directly reprogrammed T cells in vivo to generate anti-BCMA CAR-T cells. In four patients with relapsed/refractory multiple myeloma, a single intravenous infusion without leukapheresis or lymphodepleting chemotherapy achieved a 100% objective response rate. Toxicity was manageable, including three cases of grade 3 CRS and one case of grade 1 ICANS [267]. This represents the world’s first clinical validation of a fully lymphodepletion-free in vivo CAR-T therapy, establishing a foundation for “off-the-shelf” therapeutics.

In vivo CAR-T, as a disruptive technology, is reshaping the landscape of cell therapy. It offers advantages including a shorter treatment cycle, lower costs, and broader applicability. However, challenges remain in vector construction, large-scale manufacturing, and quality control systems. Its regulatory pathway also differs from traditional cell therapy products, potentially necessitating new evaluation standards. Large-scale production of viral vectors faces issues like unstable titers and difficult purification. The manufacturing and quality control of nanoparticle formulations are similarly complex, requiring the development of new processes and analytical methods. With breakthroughs in targeted delivery systems and automated production processes, in vivo CAR therapy holds promise as a universal solution for cancer immunotherapy.

Conclusions

The “off-the-shelf” universal allogeneic CAR-T cell therapies are poised to transform the landscape of oncology, offering the potential for more efficient, accessible, and scalable cancer treatments. Despite significant challenges, including GVHD, HVGR, off-target effects, genotoxicity, and manufacturing scalability barriers, continued advances in gene-editing technologies, manufacturing innovations, and clinical research will pave the way for the widespread adoption of these therapies. In the near future, allogeneic CAR-T therapies could become a mainstream treatment option in oncology, offering new hope for patients with relapsed/refractory cancers and solid tumors. With ongoing research, collaboration, and technological progress, these therapies are positioned to become an essential part of the cancer treatment paradigm, leading to better outcomes and a higher quality of life for patients.

Abbreviations

ACTs

Adoptive cell therapies

ALL

Acute lymphoblastic leukemia

AML

Acute myeloid leukemia

APCs

Antigen-presenting cells

ATG

Antithymocyte globulin

ADR

Alloimmune defense receptor

ADCC

Antibody-dependent cellular cytotoxicity

ADV

Adenovirus

AICD

Activation induced cell death

B-ALL

B-cell acute lymphoblastic leukemia

B-NHL

B Cell non-Hodgkin lymphoma

BCL

B Cell Lymphoma

B2M

β2-microglobulin

Bp

Base pairs

BCMA

B-cell Maturation Antigen

BC

Breast cancer

CAR-T

Chimeric antigen receptor T-cell

CARs

Chimeric antigen receptors

CRISPR

Clustered regularly interspaced short palindromic repeats

CNIs

Calcineurin inhibitors

CBEs

Cytosine base editors

CrRNA

CRISPR RNA

CTLs

Cytotoxic T lymphocytes

CRS

Cytokine release syndrome

CR

Complete response

CRi

Complete response with incomplete hematologic recovery

CMV

Cytomegalovirus

CLL

Chronic lymphocytic leukemia

CLL-1

C-type lectin-like molecule-1

CAFs

Cancer-associated fibroblasts

CRPC

Castration-Resistant Prostate Cancer

ccRCC

Clear cell renal cell carcinoma

DSBs

Double-strand breaks

dCas9

Dead Cas9

DLTs

Dose-limiting toxicities

DCs

Dendritic cells

DOR

Duration of response

DCR

Disease control rate

DLBCL

Diffuse large B cell lymphoma

EBV

Epstein-Barr virus

ECM

Extracellular matrix

FDA

Food and Drug Administration

FAERS

FDA Adverse Event Reporting System

FL

Follicular lymphoma

FAP

Fibroblast activation protein

FASL

FAS Ligand

GVHD

Graft-versus-host disease

gRNA

Guide RNA

GVL

Graft-versus-leukemia

HVGR

Host-versus-graft reaction

HLA-I

HLA class I

HLA-II

HLA class II

HDR

Homology directed repair

HSCT

Hematopoietic stem cell transplantation

HSCs

Hematopoietic stem cells

HM

Hematopoietic Malignancies

HIF-1α 

Hypoxia-inducible factor-1α 

iPSCs

Induced pluripotent stem cells

iPS-rCTLs

iPSCs-derived cytotoxic T lymphocytes

ICANS

Immune cell associated neurotoxicity syndrome

LBCL

Large B Cell Lymphoma

MDS

Myelodysplastic Syndromes

MM

Multiple myeloma

mTPI

Modified Toxicity Probability Interval

MRD

Minimal Residual Disease

MDSCs

Myeloid-derived suppressor cells

MUC1

Glycoprotein mucin 1

NHL

Non-Hodgkin lymphoma

NHEJ

Non-homologous end joining

nCas9

Nickase Cas9

NFAT

Nuclear factor of activated T cells

NK

Natural killer

NKT

Natural killer T cells

NSCLC

Non-small Cell Lung Cancer

ORR

Overall response rate

OS

Overall survival

OVs

Oncolytic viruses

PTCy

Post-transplant cyclophosphamide

PBMCs

Peripheral blood mononuclear cells

PAM

Protospacer adjacent motif

PegRNA

Prime editing guide RNA

PD-1

Programmed death-1

PD-L1

Programmed death-ligand 1

PFS

Progression-free survival

PR

Partial response

PB

PiggyBac

R/R

Relapsed/refractory

RNAi

RNA interference

RVD

Repeat variable di-residue

RFS

Relapse-free survival

RPA

Replication protein A

RPD

Radiopharmaceutical therapy

scFv

Single-chain variable fragment

sgRNA

single guide RNA

SPMs

Second primary malignancies

SLL

Small lymphocytic lymphoma

SD

Stable disease

SB

Sleeping Beauty

TCR

T-cell receptor

TCRαβ

T-cell receptor of αβ T cells

TRAC

T-cell receptor α chain Constant

TRBC

T-cell receptor β chain Constant

TALENs

Transcription activator-like effector nucleases

TALE

Transcription activator-like effector

Th

t-helper

TracrRNA

Trans-acting CRISPR RNA

TAMs

Tumor-associated macrophages

TMs

Targeting modules

T-ALL

T-cell acute lymphoblastic leukemia

TCL

T-Cell Lymphoma

TNF

Tumor necrosis factor

TRAIL

Tumor Necrosis Factor-Related Apoptosis-Inducing Ligand

TGFBR2

Transforming growth factor β receptor 2

TNBC

Triple negative breast cancer

UCB

Umbilical cord blood

VZV

Varicella zoster virus

WGS

Whole-genome sequencing

WES

Whole-exome sequencing

ZFNs

Zinc finger nucleases

Author contributions

Jingchao Su collected the data and drafted the manuscript, Yuhua Li and Sanfang Tu designed the research and provided critical revisions, the other authors participated in the refinement of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (No. 82270233 and No. 82200246) and the Guangdong Basic and Applied Basic Research Foundation (2024A1515011659).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.