In vivo CAR-T cell therapy: New breakthroughs for cell-based tumor immunotherapy

ABSTRACT

Chimeric antigen receptor (CAR)-T cell immunotherapy represents an evolutionary advance in the treatment of cancer, yet it faces challenges such as manufacturing complexity, high cost, and time-consuming process. In recent years, the strategy of in vivo CAR-T cell therapy is emerging as a promising approach to improve anti-tumor effectiveness and safety. Briefly, T cells are genetically modified to express CAR protein directly in the body by delivery of vectors. With the continuous optimization of gene delivery systems, gene editing technologies and CAR structures, advancements in in vivo CAR-T therapies have notably enhanced safety, effectiveness, and application in clinical settings. Here, we review the key platforms of in vivo gene delivery and the progress of in vivo CAR-T cell therapy for cancers. We discuss the challenges of in vivo CAR-T cell therapy, such as safety issues of gene delivery, the persistence and function of CAR-T cell, and the immunosuppressive microenvironment in solid tumors.

Introduction

Despite extensive research, cancer treatment remains a critical global issue.¹ Advances have been achieved in standard cancer therapies such as surgery, radiotherapy, and chemotherapy, but their therapeutic efficacy is still restricted, especially in late-stage or refractory cancers.² In contrast, immunotherapy employs stimulated immune cells that specifically target and eliminate cancer cells, thereby reducing adverse effects and offering significant advantages in antitumor therapy.³ However, immunotherapies also face several challenges limiting their clinical success. Conventional T-cell receptor (TCR) therapies rely on antigen recognition presented by major histocompatibility complex (MHC) molecules and require sufficient co-stimulatory signals to activate T cells, but tumor cells frequently escape immune surveillance by reducing MHC class I molecules. To overcome this challenge, T cells engineered to express chimeric antigen receptors (CARs) have emerged as a promising approach.⁴ These CAR-modified T lymphocytes can recognize and respond to cancer cells independent of MHC restriction.

CAR-T cell therapy is an innovative form of cancer therapy based on effector T lymphocytes. Using genetic engineering techniques, T cells can be endowed with a receptor system known as CAR. This allows the modified T cells to accurately identify and kill tumor cells from the body effectively.⁵ The CAR is a composite protein featuring an antigen-binding domain, an intracellular signaling unit, a transmembrane region, and a hinge region, each playing a distinct role.⁶ The antigen-binding domain is usually derived from the variable region of a monoclonal antibody, most commonly a single-chain variable fragment (scFv) including variable heavy (VH) and variable light (VL) domains for surface antigen recognition. The hinge region is a connecting portion that links antigen-binding and transmembrane domains, providing stability and flexibility for optimal design of CARs.⁷ The CAR is anchored to the T cell membrane via its transmembrane region. Finally, the intracellular signaling domain contains different numbers of intraceller signaling molecules for T cell activation, including intracellular CD3ζ, CD28 and 4-1BB co-stimulatory domains.⁸ To date, CARs structures have evolved through five generations (Figure 1). The changes in different generations of CARs have mainly occurred in their intracellular region.⁹ The first-generation CARs contained a CD3ζ domain alone, with a weak ability to activate T cells.¹⁰ The second-generation contains CARs introduced a costimulatory domain (e.g., CD28, 4-1BB or OX-40) to enhance T cell proliferation. The third-generation CARs incorporated multiple costimulatory signaling domains but failed to improve the efficacy of CAR-T cells. Based on second-generation CARs, the fourth-generation CARs were further engineered to express a transgenic protein, such as cytokines, enabling its release at the tumor site.^11–13 Currently, the most recent fifth-generation CARs have been designed to express a cytokine receptor domain (e.g., IL-2 receptor) to further promote CAR-T cell activation, or a drug-dependent switch receptor to control CAR-T cell function.¹⁰

Figure 1.

Evolution of the CAR structure. The evolution of CAR constructs from the first to the fifth generation. All five generations of CAR constructs share four common domains: the extracellular antigen-binding structural domain, a hinge region, a transmembrane region, and the intracellular signaling structural domain. The major difference among generations arises from modifications within the intracellular signaling region. The first-generation CAR contains only one CD3ζ signaling molecule. In second-generation CAR, the CD3ζ molecule integrates the co-stimulatory domains CD28 or 4-1BB. In terms of structure, the third-generation of CAR has two different co-stimulatory domains (CD28 4-1BB/ICOS-4-1BB) compared to the second-generation CARs. Fourth-generation CARs incorporate gene sequences encoding cytokines and other molecules. Once activated, CAR-T cells release immunostimulatory cytokines, including IL-12 and IL-15. This design is conducive to increasing T cell infiltration in tumor tissue. Fifth-generation CARs are universal CARs that contain three key domains: TCR activation signals, co-stimulation signals, and cytokine signals. They can achieve more physiological cytokine signal transduction and comprehensively enhance T cell function.

Major barriers to ex vivo CAR T cell therapy

Traditional CAR-T therapy employs an ex vivo engineering strategy as follows: T cells are isolated from the peripheral blood of patients, and then the CAR gene is introduced into the T cells via viral vectors or non-viral delivery systems. Following ex vivo expansion, CAR-T cells are reinfused into the patients.^14,15 This technology has demonstrated significant efficacy in hematologic malignancies (e.g., B-cell leukemia and lymphoma, multiple myeloma) and certain solid tumors.^16–18 Several CAR-T cell products targeting antigens such as CD19 and BCMA have been approved for clinical application.^19,20 However, traditional ex vivo CAR-T cell therapies are limited by complex manufacturing processes, high production costs, long production times, and potential immunotoxicity, thus severely restricting widespread accessibility to this therapy.^21–23

The quality of CAR-T cells produced in vitro is highly influenced by the manufacturing process and production environment.²⁴ Because of the potential risks associated with genetically modified T-cell therapies, CAR-T cells are subject to the regulatory strategies of the Food and Drug Administration (FDA). The manufacturing process typically begins in a translational research laboratory and then moves to a good manufacturing product (GMP) facility. Clinical-grade reagents and materials are used for the production of CAR-T cells and validated to ensure product quality and safety.²⁵

The production of CAR-T cells usually takes several weeks for isolation, modification, and expansion of T cells following the conventional ex vivo procedure.²⁶ The long manufacturing time limits access to ex vivo CAR-T cell therapy. During this period, patients undergo bridging therapy and lymphodepletion to control disease progression and optimize conditions for CAR-T cell expansion and implantation.²⁷ However, the use of bridging therapies, including chemotherapy, immunotherapy, targeted therapy or radiotherapy, may cause additional risks, and inappropriate timing of treatment may directly affect CAR-T cell efficacy. In addition, lymphodepletion regimens are strongly associated with adverse events, including cytokine release syndrome (CRS) and neurotoxicity after CAR-T therapy, further increasing the complexity of treatment decisions.^28,29

Antigen loss or escape is one of the barriers limiting the long-term efficacy of ex vivo CAR-T cell therapy. Tumor cells evade immune recognition by down-regulating or completely losing the expression of target antigens. As a result, CAR-T cells fail to efficiently recognize and eliminate tumor cells, which becomes an important factor for treatment failure.³⁰ The clinical effectiveness of CAR-T cell-based therapy largely relies on the in vivo persistence and function of infused CAR-T cells.⁸ Universal CAR-T cells for off-the-shelf applications have been developed through transduction of allogeneic T cells. However, allogeneic responses may induce rejection of universal CAR-T cells, thereby significantly impairing their antitumor activity.³¹ Ex vivo CAR-T cell therapy that uses allogeneic T cells may trigger host immune rejection or exacerbate graft-versus-host disease.^32,33

Collectively, several factors, including intricate and costly production procedures, patient selection, and product heterogeneity, have limited the widespread use of current ex vivo CAR-T cell products.³⁴

Generation of in vivo CAR-T cells

To break the limitations of ex vivo CAR-T cells, a novel approach has been established to generate CAR-T cells directly in vivo (Figure 2). Unlike the traditional ex vivo approach, this novel technology selectively transfers CAR genes into the patient’s T cells in the body through targeted delivery systems such as engineered viral vectors, lipid nanoparticles, and polymer nanocarriers.^23,35 The in vivo method bypasses the intricate manufacturing procedures, prolonged treatment cycle, high production costs, and potentially T cell exhaustion and heterogeneity linked to ex vivo CAR-T cell manipulation.³⁶ Notably, the in vivo strategy can achieve precise regulation of CAR-T cell function by targeting and transferring the CAR-encoding gene construct into T cells.^37,38 In vivo CAR-T cell therapy has demonstrated promising antitumor efficacy and safety in animal models. For example, in a mouse model of acute lymphoblastic leukemia, CAR-T cells were successfully generated by tail vein injection of an Adeno-associated virus (AAV) vector carrying the CAR gene and effectively cleared leukemia cells without causing significant systemic toxicity.³⁹ These results laid a foundation toward clinical development of in vivo CAR-T cell therapy to improve therapeutic response, especially in treating neoplastic diseases. The in vivo gene delivery system for T cell engineering needs to meet specific requirements such as accurate T-cell targeting, effective gene editing, and minimal toxicity (Table 1). A variety of targeted delivery platforms have been established for in situ CAR-T cell generation, mainly through viral vectors or nanocarriers (Figure 3).

Figure 2.

Ex vivo CAR-T cell therapy and in vivo CAR-T cell therapy. The ex vivo CAR-T cell therapy involves isolating T cells from the patient’s blood, introducing the CAR gene to the T cells using a delivery vector, expanding the engineered CAR-T cells outside the body, and then reinfusing them to the patient. In contrast，in vivo CAR-T cell therapy bypasses the process of ex-vivo gene modification and cell expansion. By intravenously injecting carrier particles, the CAR gene is directly delivered to the patient’s T cells, generating CAR-T cells in situ.

Table 1.

Classification of in vivo CAR-T delivery vectors.

	Delivery system	Representative Technology	Advantages	Disadvantages	Application Cases
Viral vectors	LV	HIV-based gene delivery system.	High transduction efficiency. Enables genomic integration and durable CAR expression. Engineerable targeted T cells (e.g., CD3-LV).	Higher HIV-associated immunogenicity. Risk of insertion mutagenesis. Limited carrier capacity (≤8 kb).	Generation of CD19-CAR-T cells using CD3-LV in a mouse model.40 VivoVec platform for B-cell lymphoma treatment.
	AAV	Envelope-less single-stranded DNA virus.	High transduction efficiency. High safety without genomic integration risk. Suitable for instantaneous expression.	Difficult to target T cells (requires antibody modification). Short expression time. Limited carrier capacity (≤4.7 kb)	Generation of CAR-T cells using AAV-CD4 CAR in mouse peritoneal cavity injection.41
Nonviral vectors	LNP	Antibody-coupled LNP (Ab-LNP). Carry CAR mRNA or DNA.	Highly biocompatible. Transient expression reduces side effects. Targetable modifications (e.g., CD3 antibodies). Simple and low cost production.	Low efficiency of T cell uptake Low endosomal escape efficiency. Transient mRNA expression (requires repeat administration).	Utilization of CD8-LNP for T cell reprogramming. 42,43 Construction of lipid nanocarriers targeting cd3 kills leukemia tumor cells with high CD19 expression and reduces IL-6-induced CRS.44
	Exosome	Engineered exosomes (e.g. tDC-Exo). Surface-modified anti-CD3/EGFR antibody.	Naturally low immunogenicity. Can carry multimodal molecules. (e.g., MHC, co-stimulatory molecules).	Difficult to isolate and purify. Unstable drug loading efficiency.	Modified exosomes for hematologic malignancies CAR-T therapy.45
	Polymer	Nanocarriers based on poly (beta-amino acid ester) (PBAE) polymers.	High transduction efficiency and specificity. High targeting and encapsulation capacity. Highly biocompatible and degradable.	Limited control of pharmacokinetics. High manufacturing costs. Low endosomal escape efficiency.	Polymeric nanocarriers enable in vivo treatment of leukemia, prostate cancer and HBV-induced hepatocellular carcinoma in mice by delivering tumor/virus-specific CAR/TCR mRNA.46

Figure 3.

Vector platforms for in vivo CAR delivery. Currently, two major types of delivery vectors have been engineered to enable in vivo CAR-T cell production: viral vectors and non-viral vectors. Among viral vectors, lentiviral and adeno-associated viral (AAV) systems are predominant, offering high transduction efficiency and stable gene expression. Non-viral vectors mainly include lipid nanoparticle vectors, polymer nanoparticle vectors, and exosomes. Compared to viral vectors, non-viral vectors provide enhanced flexibility, reduce the risk of insertion mutagenesis, and may lower manufacturing costs.

Viral vectors

Viral vectors can transfer target transgenes into mammalian cells and are widely used for the generation of CAR-T cells. There are two main classes of viral vectors for in vivo engineering of T cells: Lentiviruses and Adeno-associated viruses (AAVs).Viral particles are usually produced by HEK-293T cells or a packing cell line transiently transfected with recombinant plasmids.^47,48

Lentiviruses

As one of the predominant vectors employed in gene therapy, lentiviral vectors are HIV-1-based gene delivery vehicles that randomly integrate exogenous genes into the genome of host cells for long-term stable expression.⁴⁹ However, conventional lentiviral vectors are unable to selectively target T cells in vivo due to the lack of specific cell-targeting capabilities.⁵⁰

To achieve T-cell specificity, lentiviral vectors require envelope pseudotyping by replacing the native envelope glycoprotein (such as vesicular stomatitis virus G protein, VSV-G) with engineered ligands like CD3/CD4/CD8-targeting scFvs.⁵¹ An ideal targeted lentiviral vector for in vivo CAR-T cell therapy should possess the following key features: 1) Minimal off-target effects: To prevent the delivery of CAR gene to non-target cells, the viral vector needs to be engineered to ablate the binding of viral glycoproteins to natural surface receptors of non-target cells. 2) T cell specificity: With additional target-binding structural domains, the viral vector can bind T cells in a covalent or non-covalent manner to ensure the delivery of CAR gene to T cells. 3) Efficient gene delivery: The viral vector retains membrane fusion activity to allow the introduction of genetic material into target cells.⁵² To address the targeting issue, receptor-targeted lentiviruses have been developed by replacing the envelope proteins with proteins capable of recognizing specific surface receptors for targeted transduction. The transduction process does not require the activation of T cells, and the vector is capable of directly engaging T-cell surface receptors to achieve efficient gene delivery.³⁸ Modified lentiviral vector provides an efficient platform for in vivo receptor-targeted CAR-T cell delivery. This strategy enhances cellular transduction rates, minimizes off-target effects, and effectively transfers CAR genes to T cells (Figure 4). This approach results in direct in vivo production of tumor-targeting CAR-T cells, with reduced therapeutic costs.⁵³

Figure 4.

Viral delivery system for in vivo CAR-T cells. The viral vector surface carries an anti-CD3 single-chain variable fragment (scFv) fused to cocal glycoprotein, which specifically binds to CD3 on the surface of T cells. After binding, the membrane invaginates and forms an endosomal vesicle, releasing the viral core into the cytoplasm, where the viral RNA genome undergoes reverse transcription to form double-stranded DNA. The newly formed DNA is then imported into the nucleus and integrated into the T cell genome, thereby achieving stable expression of the CAR transgene.

To deliver genes to T cells, lentiviral vectors engineered to express agonistic CD3-specific scFv on the vector surface are capable of selectively targeting and activating T cells in vitro and in vivo. The targeted delivery system exhibits higher transduction efficiency of resting T cells compared to conventional lentiviral vectors, even in whole human blood.⁵⁴ When the engineered viral vector was injected into humanized mouse models, the CD19-CARs gene was selectively expressed in human CD8+ T cells. These modified T cells then specifically targeted and eliminated CD19-positive B cells through precise molecular recognition.⁵⁵ In a syngeneic tumor-bearing murine model, CD4-targeted lentiviral vectors led to accelerated tumor regression and enhanced therapeutic efficacy compared to CD8-targeted lentiviral vectors or a mixture of the two vectors, likely due to CD8+ T cell exhaustion.⁵⁶

AAV

In vivo CAR-T cell production has also been achieved using AAV vectors. AAV is a non-enveloped, single-stranded DNA virus with a genome length of 4.7 kb that can be modified to mediate DNA delivery into target cells.⁵⁷ The AAV vector genomes persist as episomes in the transduced target cells, driving stable expression of transgene.⁴⁹ Compared to adenoviruses, AAVs are less immunogenic and less likely to trigger an immune response.^58,59 In addition, recombinant AAV vectors lack viral coding sequences responsible for replication and pathogenicity, and do not require engineered lipids or chemical modifications for efficient transduction, which together contribute to their minimal toxicity and low immunogenicity.⁶⁰ A substantial amount of evidence from clinical studies utilizing AAV supports the safety and efficacy of gene therapy applications using recombinant AAVs.⁶¹ Higher transduction efficiency and more stable transgene expression make AAV ideal for in vivo CAR-T cell therapy.⁶² AAV tropism is governed by capsid modification. A recent study has engineered chimeric AAV-DJ vectors by fusing functional domains from eight wild-type AAV serotypes via DNA shuffling technology. This innovation significantly enhances transduction efficiency in T cells and enables in situ generation of CAR-T cells in vivo. In a humanized tumor mouse model, injection of AAVs carrying the CD4-CAR gene generated robust in vivo CAR-T cells and showed significant tumor-suppressive activity against human T cell leukemia.³⁹ These findings support the use of AAVs as a promising approach for the direct in vivo generation of CAR-T cells.

Non-viral vectors

Nanoparticles, generally measuring 10 to 1000 nm, are composed of lipids, polymers, inorganic substances, or biomaterials. They are widely employed as delivery platforms for transporting therapeutic agents, genetic material, or imaging probes to target cells and tissues.⁶³ Lipid nanoparticles (LNPs) are non-viral delivery systems capable of carrying nucleic acids and small molecule drugs into target cells. Due to the unique physicochemical features and biocompatibility, LNPs have become an essential approach for in vivo delivery of CAR constructs to T cells. LNPs are spherical vesicles composed of either a single (monolayer) or multiple (multilayer) phospholipid bilayers.⁶⁴ LNPs can encapsulate mRNA internally to avoid mRNA degradation in vivo and effectively preserve mRNA stability.^65–67 Techniques such as electroformation, hydration, and extrusion are commonly used for preparing liposomes. Among these, the thin-film hydration method – a classical approach – involves dissolving phospholipids and cholesterol in an organic solvent to form a thin film, which spontaneously self-assembles into bilayer vesicles upon aqueous hydration.⁶⁸ Furthermore, LNPs can enhance delivery efficiency by fusion with the cell membrane and facilitate the release of mRNA into the cytoplasm via pH-sensitive endosomal escape. By surface modification with adding target-specific ligands, LNPs can specifically interact with particular cell types.^69,70 The LNP-mediated mRNA delivery allows transient transgene expression in target cells in vivo.³⁸ Thus, in vivo delivery of CAR mRNA to T cells avoids permanent CAR expression to reduce the risk of severe side effects such as CRS.⁷¹

The LNP-mRNA delivery system for in vivo CAR-T generation was first documented in a preclinical study as a therapeutic strategy against cardiac fibrosis. Fibroblast activation protein (FAP) is an endogenous cardiac fibrosis antigen. The CD5-targeted LNPs encapsulating mRNA encoding FAP-specific CAR were injected into a mouse model of heart failure. This novel therapeutic strategy significantly reduced cardiac fibrosis and improved cardiac function.⁷² In the field of cancer diseases, an LNP system was developed to carry a plasmid containing IL-6 shRNA and CD19-CAR, with modification of anti-CD3 antibody on the surface.⁴⁴ This technology successfully produced CD19-specific CAR-T cells in vivo and reduced IL-6 expression, showing significant anti-tumor effects with less IL-6-induced toxicity.

LNPs-based delivery platforms for in vivo CAR-T cells overcome several problems associated with viral vectors, including limited vector capacity, insertional mutagenesis, and viral immunogenicity. The synthesis of mRNA is simple, avoiding the complex manufacturing process of viral vectors and reducing the production cost. Recently, an LNP-siRNA delivery system has been approved by FDA for the treatment of peripheral nerve disease caused by hereditary transthyretin amyloidosis.⁷³ In addition, transient mRNA expression can reduce side effects of long-term CAR expression, such as cytokine storms.⁷⁴ However, current LNP-based mRNA delivery methods face multiple challenges, such as poor T-cell uptake and low endosomal escape.^75,76 Based on the membrane fusion mechanism of HIV, a virus-like particle has been recently developed to display a mutant envelope glycoprotein gp160 for T cell recognition and deliver CAR mRNA via a capsid protein Peg10.⁴¹ This strategy improves the efficiency and safety of in vivo CAR-T cell therapy, while simultaneously providing a new approach for targeted delivery to other cell types via cell-specific membrane fusion.

In addition to lipid-based nanosystems, polymer-based nanosystems are widely employed in in vivo CAR-T cell engineering.⁶⁴ Polymeric nanoparticles encapsulate nucleic acids to form versatile nanocarriers. Compared to liposomes, polymeric nanoparticles offer greater structural diversity and ease of modification, enabling precise control over particle size and morphology. These properties underpin their controllable cargo-loading capacity and enhanced delivery performance.⁷⁷ For gene delivery, polymeric nanoparticles typically incorporate positively charged groups (e.g., amino groups), which electrostatically bind negatively charged nucleic acids (e.g., DNA/mRNA). This binding protects nucleic acids from nuclease degradation while facilitating efficient intracellular delivery and release. Common polymers include polyethyleneimine (PEI), poly (2-dimethylaminoethyl methacrylate) (PDMAEMA), and poly (β-amino ester) (PβAE). While these nanoparticles exhibit high loading capacity, customizability, and low cytotoxicity, they face challenges such as low T-cell transfection efficiency and specific cellular barriers.⁷⁸ In a proof-of-concept study, PβAE nanoparticles encapsulating mRNA encoding a prostate tumor-specific CAR were systemically administered to mice, resulting in significantly prolonged survival compared to untreated controls.^79,80

Exosomes have also been used as nanocarriers for gene delivery. The nanosized vesicles are secreted by various cells, with a particle size range of about 40–160 nm.⁸¹ The natural properties of exosomes, such as higher stability and lower immunogenicity,⁸² make them an ideal vehicle for delivery of CAR mRNA in vivo. Recently, an engineered exosome platform was developed to express anti-CD3/CD28 scFvs on the surface for targeting T cells and encapsulate CAR mRNA by using the combination of the bacteriophage MS2 system with cargo-loading chaperone LAMP-2B. The modified exosomes can enable the in vivo production of CAR-T cells with anticancer capabilities, indicating the potential applications for producing CAR T cells in vivo.⁴⁵ In addition, a research team has engineered tumor antigen-stimulated dendritic cell-derived exosomes (tDC-Exo) to express anti-CD3 and anti-EGFR. The tDC-Exo not only activates endogenous T-cells but also facilitates T cell binding to cancer cells. This CAR-T cell therapy-mimicking platform further enhances the effectiveness of anti-tumor immunotherapy.⁸³

Lentiviral vectors can permanently integrate CAR genes into the DNA of target cells, thereby resulting in durable expression. This property renders them particularly suitable for hematologic malignancies that require prolonged immune surveillance.^40,84 AAV vectors show a strong tropism for the liver, making them particularly suitable for the treatment of hepatic diseases.³⁹ The transient expression conferred by LNP vectors renders them advantageous for therapeutic applications targeting acute diseases or aggressively progressing tumors.^42,72 Polymeric nanocarriers can achieve tissue-specific targeting through modular modifications. In addition, their strong barrier-penetrating capacity makes them more suitable for solid tumors.³⁶ Clinical translation of exosome carriers is still under investigation. Autologous exosomes offer the advantage of delivering personalized CAR constructs with reduced immunogenicity. However, challenges such as high manufacturing costs and stringent quality control requirements remain.^45,83

With the rapid development of multi-omics analysis technologies and network pharmacology platforms, cancer research has entered a new stage. By combining genomics, transcriptomics, proteomics, and metabolomics, multi-omics analyses facilitate the identification of tumor-specific antigens and provide valuable resources for the design of CAR constructs.^85–87 Omics analyses can be used to assess the impact of vectors on cellular metabolism, as well as their tissue distribution and immunogenicity.^72,88 Network pharmacology constructs multidimensional ‘disease–target–pathway’ networks to identify key regulatory nodes within the tumor microenvironment,⁸⁹ providing new strategies for optimizing in vivo CAR-T cell therapies. Network analyses enable precise screening of tumor targets. The identified targets can guide the engineering of vectors for specific delivery.^90–92 A research team constructed a machine learning model utilizing curated DNA LNP data from high-throughput screening studies to predict efficient LNP designs. Combining machine learning with LNP screening and optimization enables the customization of designs for specific therapeutic needs.⁹³ By combining multi-omics–based precision design with high-throughput screening platforms, CAR constructs and delivery strategies can be rapidly iterated. AI virtual clinical trial platforms, utilizing deep learning models, offer improved prediction of severe cytokine release syndrome.⁹⁴ These innovative approaches provide a theoretical foundation for the clinical development of in vivo CAR-T therapies.^90,94

Research progress of in vivo CAR-T cell therapy

For clinical applications, in vivo CAR-T cell therapy has received great attention from industry to improve therapeutic precision and reduce toxicity, through innovative delivery systems, optimization of gene editing and precise targeting strategies. Umoja Biopharma has developed a novel in vivo CAR-T cell engineering platform (VivoVec), a lentiviral vector expressing surface T-cell activation and costimulatory molecules for generation of CAR-T cells in vivo without lymphodepleting chemotherapy. This cocal-pseudotyped vector enhances the binding and transduction efficiency of viral particles to T cells by displaying anti-CD3 scFv on the viral surface and encoding a CAR transgene.^95,96 In humanized mouse models, the VivoVec system has been shown to successfully induce anti-CD19 CAR-T cells in vivo, resulting in the effective depletion of B cells.⁹⁷ In addition, animals tolerated a VivoVec clinical candidate UB-VV100 without significant toxicity, showing a favorable safety profile.^35,97 These results provide a promising foundation for the application of in vivo CAR-T therapy in clinical trials. In non-human primates, an intravenous injection of VivoVec particles encoding an anti-CD20 CAR transgene induced prolonged B cell depletion, further validating the safety and effectiveness of in vivo CAR-T cell therapy.³⁵

Although preclinical studies have applied lentiviral vectors for in vivo CAR T-cell engineering, clinical evidence is limited.³⁸ For the first time, a clinical study has documented the application of in vivo CAR-T cell therapy in patients with relapsed or refractory multiple myeloma. This phase 1 trial evaluated ESO-T01, a lentiviral vector – mediated in vivo CAR T-cell therapy, in four patients with relapsed or refractory multiple myeloma. Two patients achieved stringent complete remission and two achieved partial remission. Cytokine release syndrome and transient hematologic toxicities were manageable. CAR T cells were detected in multiple tissues, confirming in vivo generation and expansion.⁹² Recent clinical trials demonstrate that in vivo CD19-targeted CAR-T cell therapy shows promising efficacy in relapsed/refractory B-cell malignancies. A multiply relapsed B-cell acute lymphoblastic leukemia patient achieved minimal residual disease-negative complete remission within one month without severe adverse events, while a high-tumor-burden diffuse large B-cell lymphoma case similarly attained complete remission. These findings suggest robust antitumor activity with favorable safety. Viral vectors can also be used for targeted therapy of solid tumors.⁹⁸

Despite great successes in hematological malignancies, CAR-T cell therapies still face numerous significant challenges in solid tumor therapy, including immunosuppressive tumor microenvironment, poor intra-tumoral trafficking, and the lack of tumor-specific antigens. Recently, delivery of CAR-T cells in an injectable lymph node-like scaffold increases CAR-T cell expansion and improves treatment of solid tumors.⁹⁹ This strategy promotes CAR-T cell expansion both ex vivo and in vivo. Furthermore, the scaffold could be loaded with other materials to enhance CAR-T cell therapy. Other innovations in localized delivery systems are also driving rapid progress of in vivo CAR-T cell therapy. An all-in-one implantable platform, Multifunctional Alginate Scaffold for T cell Engineering and Release (MASTER), has been developed to reduce the process of CAR-T cell manufacturing to one day. By loading MASTER with patient-derived T cells and retroviral CAR constructs, same-day implantation can generate in vivo CAR-T cells, effectively controlling tumor growth in a mouse lymphoma xenograft model.¹⁰⁰

Current lentiviral and AAV vectors can deliver transgenes with a maximal size of 8 kb and 4.5 kb, respectively. The helper-dependent adenoviral (HDAd) vectors, allowing loading up to 35 kb, have been used for in vivo prime editing of hematopoietic stem cells (HSCs). This novel technique could efficiently transduce HSCs and correct the mutation of sickle cell disease in a mouse model, highlighting the potential of precise editing for CAR-T cell therapy.⁸⁴ In addition, a novel VLP-based engineered system has been reported to generate multi-lineage CAR immune cells inside the body, showing robust and durable antitumor activity.¹⁰¹ Using VLP vectors to deliver CAR genes into hematopoietic stem cells allows their differentiation into long-lasting CAR-T cells, addressing the issue of insufficient CAR-T persistence in solid tumors. This technology provides a new therapeutic strategy for the durable treatment of solid tumors.

Currently, several clinical trials of in vivo CAR-T cell therapy for malignant tumors have been conducted worldwide. Tumor treatment will enter the era of in vivo precision immunotherapy in the future (Table 2).

Table 2.

Clinical trials targeting in vivo CAR-T therapy for malignant tumors.

NCT Number	Drug Name	Sponsor	Delivery vehicle	Target	Diseases	Estimated enrollement	Trial Phase	Reference
NCT06539338	INT2104	Interius BioTherapeutics	LV (CD7 scFv)	CD20	B cell cancers	30	Phase 1	143
NCT06528301	UB-VV111	Umoja Biopharma	LV (CD3 scFv, CD80 and CD58)	CD19	B cell cancers	106	Phase 1	144
NCT06743503	UB-VV400/410	Nanjing IASO Biotechnology Co., Ltd.	LV (CD3 scFv, CD80 and CD58)	CD22	B cell cancers	70	Phase 1	145
NCT06688435	KLN-1010	CSPC ZhongQi Pharmaceutical Technology Co., Ltd.	LV (CD3 antibody)	BCMA	Multiple myeloma	50	Phase 1	146
NCT05969041	MT-302	Myeloid Therapeutics	LNP	TROP	Epithelial tumours	48	Phase 1	147
NCT06478693	MT-303	Myeloid Therapeutics	LNP	GPC3	Liver cancer	48	Phase 1	148
NCT06890065	JY-231	Shenzhen Genocury Biotech Co., Ltd.	LV	CD19	B-Cell Lymphoma/Leukemia	20	Phase 1	149

The challenges of in vivo CAR-T cell therapy

The in vivo induced CAR-T cell therapy has become a hot research topic in the area of cancer immunotherapy, owing to remarkable advantages such as complete elimination of in vitro cell manipulation and expansion, as well as short treatment cycles. However, this technology still faces many challenges in terms of gene delivery safety, and innovative solutions are urgently needed for its clinical translation of the in vivo CAR-T cell therapy.

The safety of gene delivery systems is a major challenge in the development of in vivo CAR-T technologies. Certain viral vectors have the risk of oncogenic insertional mutagenesis during genomic integration, which may lead to abnormal cellular function or even tumorigenesis.^109,110 To reduce the risk associated with insertional mutagenesis, the currently employed strategies involve the improvement of CAR structures and the optimization of viral vector components, such as modified promoters and enhancers.^102,111 In addition, viral vectors and transgene products may trigger adverse immune responses in the body. The gene delivery vectors, including lentiviral vectors, retroviral vectors, AAV vectors, mRNA system, and plasmid system, should be carefully selected for a particular target.¹⁰³,¹¹² Different vectors have their own technical challenges. For instance, viral vectors commonly used for CAR gene delivery have high gene transduction efficiency, but limited cargo capacity and the risk of off-target effects.³⁶ Further modification of viral vectors will be necessary to enhance their targeting, improve delivery efficiency and reduce immune responses.

The non-viral vectors for gene delivery often face the hurdle of poor biodegradability and instability in the host body. The LNP-mRNA delivery system also suffers from low delivery efficiency, poor targeting capability and short-term expression.^{104,105,113,114} Their limited transfection efficiency is largely due to the insufficient escape from lysosomes, and the enzymatic degradation of mRNA.^106,115 To improve delivery efficiency, researchers have developed a variety of elegant and novel approaches.^107,116 Wyman et al. designed a KAALA peptide, a pH-sensitive peptide with a pH-dependent conformation, to promote the escape of DNA from the lysosome under acidic conditions.^108,117 Shaheen et al. updated the octaarginine-modified multilayered nanoparticles with the KALA peptide, significantly improving the expression of transgene.¹¹⁸ For in vivo T cell-targeted therapeutics, further modification of surfactant or lipid molecules of non-viral vectors can improve targeting, reduce drug toxicity to normal tissues, and enhance therapeutic efficacy.^119,120

Although in vivo CAR-T cell therapy has exhibited notable antitumor capabilities, the clinical effectiveness of CAR-T therapy is largely constrained by T cell exhaustion.¹²¹ To overcome this challenge, various strategies have been developed to enhance the persistence and antitumor functions of CAR-T cells in vivo. For example, the incorporation of co-stimulatory molecules (e.g., CD28 or 4-1BB) has been demonstrated to strengthen T cell activation and persistence by modifying intracellular signaling domains.^18,122 Additionally, the combination of immune checkpoint inhibitors and chemotherapy could block inhibitory signals and prevent T-cell exhaustion.^123,124 Depletion of T cell exhaustion-related genes or enhancement of CAR-T cell metabolic activity through genetic techniques such as CRISPR/Cas9 and TALEN may reduce CAR-T cell exhaustion within the hostile tumor microenvironment.¹²⁵ Moreover, inhibition of CD38, a potential exhaustion marker for CAR-T cells, was found to repress glycolysis, thereby improving the effectiveness and durability of CAR-T cell therapy.¹²⁶

Although viral vectors and nanomaterials hold broad application potential, the use of gene-editing technologies raises ethical concerns. Moreover, the gene-editing tools themselves may act as immunogenic antigens.¹²⁷ Immunogenicity remains a key challenge limiting the safety and durability of both ex vivo and in vivo gene therapies. In viral vectors, immunogenicity primarily arises from capsid or envelope proteins as well as the transgene products. These immune responses may lower the efficiency of transduction, restrict repeated dosing, and compromise the overall safety and persistence of gene therapy.¹²⁸ Research indicates that nanoparticle characteristics, including surface charge and size, markedly affect complement activation and the likelihood of cytokine release syndrome.¹²⁹ CAR-T–induced immunosuppression can increase the risk of infections and secondary malignancies. Since gene modification occurs in vivo, persistent expression may also pose risks of chronic toxicity or secondary primary malignancies.¹³⁰ Currently, there is no unified technical standard for CAR-T cell products. Regulatory frameworks primarily focus on ex vivo–modified cell therapies.¹³¹ In vivo CAR-T involves in situ modification of patient cells, making conventional testing methods difficult to directly apply for quality control of in vivo gene-edited products. Long-term follow-up systems are needed to dynamically monitor the genome and in vivo CAR-T expansion, in order to assess the durability of gene modifications and potential risks.

In vivo CAR-T cell therapy also face substantial challenges in treating solid tumors compared to hematologic malignancies.¹³² In addition to challenges with delivery safety and CAR-T cell persistence, the immunosuppressive tumor microenvironment (TME) represents a significant obstacle to the efficacy of in vivo CAR-T therapy in solid tumors and chronic inflammatory diseases. Within the TME, regulatory T cells (Tregs), myeloid-derived suppressor cells (MDSCs), and tumor-associated macrophages (TAMs) collaborate to form an immunosuppressive network. By secreting inhibitory cytokines such as TGF-β and IL-10, these cells suppress CAR-T cell activity and compromise therapeutic efficacy.¹³³ Such immunosuppressive niches are concomitantly established in chronic inflammatory milieus, exacerbating CAR-T cell dysfunction. Indeed, recent systematic work reveals how in-depth profiling of immune cell interactions and antitumor immunity within the TME provides critical mechanistic insights for CAR-T therapy. To counter these barriers, researchers are now developing multifaceted intervention approaches. Pharmacological interventions and genetic modifications can be employed to block the function of immunosuppressive cells, thereby converting the suppressive immune microenvironment.^134–136 In addition, antigen escape and tumor heterogeneity have led to limited efficacy of single-targeted CAR designs in solid tumors, thus multi-targeted or bispecific CAR constructs have been developed as promising strategies.^137,138 The immune microenvironment in chronic diseases exhibits immunosuppressive features mirroring those in solid tumors, characterized by inflammation-driven T cell exhaustion and immune dysregulation. Investigating these processes provides critical references for understanding CAR-T cell functionality within complex disease contexts.¹³⁹

In solid tumors, key elements of the TME comprise cancer-associated fibroblasts (CAFs) and the desmoplastic extracellular matrix they produce.¹⁴⁰ In liver cancer, the abundant presence of hyaluronic acid and collagen within the ECM contributes to a dense fibrotic barrier that hinders the effective infiltration of CAR-T cells.¹⁴¹ The development and progression of hepatocellular carcinoma (HCC) are closely associated with dysregulation of multiple key signaling pathways.¹⁴² HCC cells often downregulate target antigens such as GPC3 and AFP, resulting in antigen escape. These mechanisms promote tumor development.¹⁴³ The main immunotherapeutic strategies for HCC currently include immune checkpoint inhibitors and adoptive cell therapies. Studies have demonstrated that these immunotherapies can significantly prolong overall survival and reduce the risk of recurrence in HCC patients.¹⁴⁴ Post-liver transplantation HCC patients constitute an ideal population for in vivo CAR-T therapy, this approach bypasses the complex ex vivo manufacturing process. It enables rapid intervention during critical immunosuppressive windows. Liver-targeted vectors allow local delivery of CAR genes, reducing systemic toxicity.^145,146 Recent studies have shown that the gut microbiota is pivotal in regulating host immune responses and influencing the efficacy of cancer immunotherapy. The composition of the gut microbiota is intimately connected with both the therapeutic outcomes and toxicities of CAR-T cell therapy.^147,148 Modulating the gut microbiota can enhance the expansion and antitumor efficacy of in vivo CAR-T cell therapy. It may also decrease the risk of adverse effects such as cytokine release syndrome.¹⁴⁹ Personalized microbiota interventions in patients can precisely modulate and reshape the immune microenvironment. This approach offers new strategies for in vivo CAR-T cell therapy against solid tumors.

Conclusions and perspectives

As a revolutionary cancer treatment strategy, in vivo CAR-T cell therapy provides benefits compared with traditional ex vivo CAR-T cells. This novel strategy presents the potential to overcome many limitations of ex vivo CAR-T cell therapy, such as complex, costly and time-consuming manufacturing processes, highlighting an ideal therapeutic alternative to off-the-shelf CAR-T cells.

As a promising immunotherapy, in vivo CAR-T cell therapy is hindered by various limitations, including viral vector delivery efficiency, immune response, durability and solid tumor adaptability. However, these issues are expected to be resolved following continuous optimization of gene editing technologies, targeted delivery systems, and CAR structural designs. The in vivo CAR-T therapies are progressively moving from the laboratory to the clinic. In hematologic tumors, in vivo CAR-T cell therapy has demonstrated significant clinical potential. Moreover, this novel therapeutic is making breakthroughs in the treatment of solid tumors. The combination strategies with immune checkpoint inhibitors (e.g., PD-1 antibodies) or cytokines (e.g., IL-12-engineered exosomes) may further expand the application range into other indications.

Preclinical studies and early clinical trials have demonstrated the in vivo CAR-T cell therapy is safe and effective. Following remarkable advancements in innovative delivery technologies, expanded indications, and optimized safety, in vivo CAR-T cell therapy is expected to emerge as a key strategy in future tumor immunotherapy, bringing more efficient and safer therapeutic options to cancer patients.

Biography

Yi Zhang currently serves as the Director of the Biotherapy Center and Chief Scientist at the First Affiliated Hospital of Zhengzhou University, Academic Vice Dean of the Academy of Medical Sciences at Zhengzhou University, Executive Director of the Central Plains Cell and Immunotherapy Laboratory. He holds key positions in several national academic organizations, including Chair of the Tumor Immunotherapy and Biotherapy Committee of the Chinese Society of Immunology, and Vice Chair of the Tumor Biotherapy Professional Committee of the China Anti-Cancer Association. With 34 years of international research experience, Professor Yi Zhang is dedicated to overcoming critical bottlenecks in cell therapy and accelerating its clinical translation. His work has led to the discovery of novel regulatory mechanisms in immune cell function, the development of groundbreaking technologies, and their successful translation into clinical practice. Professor Zhang has directed more than 60 clinical trials in immune cell therapy, ranking first globally in the number of clinical trials targeting solid tumors with immune cell-based treatments.

Funding Statement

This work was supported by grants from National Key Research and Development Program Intergovernmental International Science and Technology Innovation Cooperation Project [2022YFE0141000]; the Key Program supported by the Joint Funds of the National Natural Science Foundation of China [U24A20734]; National Natural Science Foundation of China [82272873]; National Science Fund for Distinguished Young Scholars [82203548]; Henan Province Science and Technology Research Project [221100310100]; National Key R&D Program Frontier Biotechnology Key Special Project [2023YFC3403800] and Outstanding Youth Project of Henan Provincial Natural Science Foundation [252300421017]; National Key R&D Program Frontier Biotechnology Key Special Project [2023YFC3403800] and Outstanding Youth Project of Henan Provincial Natural Science Foundation [252300421017].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Ethical approval

Ethical approval was not required for this study, as it is a review article.

Abbreviations

CAR

Chimeric Antigen Receptor

CAR-T

Chimeric Antigen Receptor T

TCR

T-cell receptor

VH

Variable heavy

VL

Variable light

FDA

Food and Drug Administration

MHC

Major histocompatibility complex

CRS

Cytokine release syndrome

AAV

Adeno-associated virus

LV

Lentiviral vector

RT-LV

Receptor-targeted LV

LNP

Lipid nanoparticles

T-FVLP

T-cell-specific fusion virus-like particle

MHC

Major histocompatibility complex

EGFR

Epidermal growth factor receptor

VLP

Virus-like particle

MDSC

Myeloid-derived suppressor cells