Nanoparticle-based strategy in CAR-T cell immunotherapy: challenges, implications, and perspectives

Abstract

Chimeric antigen receptor (CAR)-T cell therapy has achieved remarkable progress in treating hematologic malignancies, yet its broader application faces challenges such as manufacturing complexity, solid tumor microenvironment barriers, and immune toxicity. Nanoparticles (NPs), leveraging their precise delivery, immunomodulation, and multifunctional integration capabilities, offer innovative strategies to optimize CAR-T cell therapy. This review provides a comprehensive elucidation of the fundamental framework of CAR-T cell therapy and the challenges in oncological applications. Subsequently, we systematically summarized the synergistic mechanisms between NPs and CAR-T cell therapy, including optimization of genetic modification, enhancement of tumor site infiltration, modulation of immunosuppressive tumor microenvironments, mitigation of tumor antigen heterogeneity, real-time monitoring, and dynamic control of cellular activity. Ultimately, it highlights the emerging paradigm of artificial intelligence integration within this domain while discussing the associated technical obstacles and future prospects of this combined therapeutic approach.

Introduction

Chimeric antigen receptor (CAR)-T cell therapy, an advanced form of adoptive cell immunotherapy, enhances the immune system’s ability to combat tumors by genetically modifying donor or patient T cells to express CARs [1, 2]. The therapy generally consists of five critical steps: T cell isolation, activation, genetic engineering for CAR expression, CAR-T cell expansion, and reinfusion into patients to target cancer cells [3] (Fig. 1). CARs, expressed on the engineered CAR-T cells and designed to recognize specific cell surface structures, are mainly composed of four functional domains: the extracellular domain (ECD) with a single-chain variable fragment (scFv) for recognizing antigens and bypassing the limitations of the major histocompatibility complex (MHC), a transmembrane domain (TMD) for structural anchoring, a hinge domain (HD) for connecting ECD and TMD, and an intracellular domain for signal transduction [4, 5]. After recognizing tumor antigens, these engineered CAR-T cells eliminate tumor cells through three mechanisms: (1) Through the perforin and granzyme pathway, perforin released by CD8⁺ T cells forms pores on the tumor cell membrane, allowing granzymes to enter and induce cell death [6]; (2) Through cytokine release, they regulate the tumor microenvironment (TME) and promote the polarization of macrophages to the M1 phenotype [7]; (3) Through the Fas/Fas ligand (FasL) pathway, the binding of Fas and FasL activates the adaptor protein Fas-associated death domain (FADD), triggering apoptosis [8] (Fig. 1). Notably, during CAR-T cell-mediated attack on target cells, granzyme B enters the tumor cell cytoplasm through perforin-formed pores in the plasma membrane, directly activating caspase-3. The activated caspase-3 subsequently cleaves gasdermin E (GSDME), thereby triggering pyroptosis, a proinflammatory form of programmed cell death, that ultimately leads to tumor cell membrane perforation, cellular swelling, and rupture [9, 10]. This pyroptotic process promotes the massive release of damage-associated molecular patterns (DAMPs) from tumor cells [11]. These DAMPs specifically activate antigen-presenting cells (APCs, e.g., macrophages) infiltrating the TME, enhancing their antigen uptake and presentation capabilities. Consequently, this mechanism further amplifies the antitumor immune response mediated by CAR-T cells [12].

Fig. 1.

The conventional CAR-T cell manufacturing process and the tumor-killing mechanisms. Blood is collected by venipuncture or apheresis, and T cells are isolated from either the patient or a healthy allogeneic donor. Subsequently, T cell activation is achieved using aAPCs or beads CD3/CD28. Genetic modification to generate CAR-T cells is then performed via either viral or electroporation gene delivery methods. During the patient’s lymphodepleting pre-treatment, the engineered CAR-T cells undergo ex vivo expansion in culture systems. Finally, the amplified CAR-T cell products are infused back into the patient, enabling specific recognition and elimination of tumor cells

Over the past 30 years, CAR-T cell technology has evolved to the fifth generation. The first-generation CAR-T cells achieve basic T cell activation by combining scFv with the CD3ζ signaling domain, but their efficacy and durability are limited [13]. The second generation introduces costimulatory domains (CSDs) such as CD28 and CD147 [14]. Incorporating an additional costimulatory domain into the third-generation framework significantly enhances the activity, proliferation, and durability of T cells [15]. The fourth generation, termed TRUCK (T-cells redirected for universal cytokine-mediated killing), integrates cytokine or immunomodulator genes to regulate the TME while simultaneously recruiting immune cells [15]. The latest fifth-generation CAR-T cells can respond to signals in the TME and provide a more controllable regulation method [16]. In particular, the emergence of Boolean logic-gated CAR-T cells has achieved precise control of activity through logical gates such as AND, OR, and NOT, improving specificity and efficacy while reducing side effects [17, 18]. Currently, seven CAR-T cell therapies, including Kymriah™, Yescarta™, Tecartus™, Breyanzi^®, Abecma^®, Carvykti™, and Aucatzyl^®, have received the Food and Drug Administration (FDA) approval for hematologic malignancies [19, 20]. Recent studies also explore CAR-T cell applications in non-cancer fields such as autoimmune diseases, infections, and transplantation, highlighting their transformative potential [21–24] (Table 1).

Table 1.

Summary of clinical trials using CAR-T treated pathologies beyond tumor

Disease type	Condition	Target	Study Type	Interventions	Phase	Status	NCT
Autoimmune diseases	ITP	BCMA	Interventional	Biological: autologous anti-BCMA chimeric antigen receptor T cells	II	Unknown	NCT05315778
	SLE	CD19	Interventional	Other: CAR T-cell therapy	I	Recruiting	NCT06150651
	SLE/SjS/SSc/DM/AAV	CD19	Interventional	Biological: CD19 targeted CAR-T cells	I	Recruiting	NCT06056921
	Scleroderma	CD19/BCMA	Interventional	Biological: Assigned Interventions CD19/BCMA CAR T-cells	Early I	Recruiting	NCT05085444
	SjS	CD19/BCMA	Interventional	Biological: Assigned Interventions CD19/BCMA CAR T-cells	Early I	Recruiting	NCT05085431
	Lupus nephritis	CD19/BCMA	Interventional	Biological: Assigned Interventions CD19/BCMA CAR T-cells	Early I	Recruiting	NCT05085418
	SLE/SjS/SSc/IM/AAV	CD19	Interventional	Biological: Anti-CD19-CD3E-CAR-T cells	NA	Recruiting	NCT06373081
	Crohn’s disease/Ulcerative colitis/DM/Still diseases	CD7	Interventional	Biological: CD7 CAR T-cells	Early I	Recruiting	NCT05239702
	MS	CD19	Interventional	Biological: KYV-101 anti-CD19 CAR-T cell therapy Drug: Standard lymphodepletion regimen	I	Recruiting	NCT06138132
	AAV/IIM/Lupus nephritis/SjS/SLE/SSc	CD19/BCMA	Interventional	Biological: CD19-BCMA CAR-T cells	I/II	Not yet recruiting	NCT06350110
	AAV/IM/SjS/SLE/SSc/APS	CD19	Interventional	Biological: BRL-301	NA	Recruiting	NCT05859997
	MG	CD19	Interventional	Drug: CD19 CAR-T cells injection	I	Recruiting	NCT05828225
	Neuromyelitis optica	CD19	Interventional	Drug: CD19 CAR-T cells injection	I	Recruiting	NCT05828212
	AIHA	CD19	Interventional	Biological: ThisCART19A	I	Recruiting	NCT06212154
	IIM/MS/MG/NMOSD/MOGAD	BCMA	Interventional	Biological: CT103A cells Drug: Cyclophosphamide and fludarabine	Early I	Recruiting	NCT04561557
	IIM/ITP/RA/SjS/SLE/ SSc	CD19	Interventional	Biological: T cell injection targeting CD19 chimeric antigen receptor	Early I	Not yet recruiting	NCT06417398
Infectious diseases	HIV	bNAbs	Interventional	Biological: CAR-T cells	I	Unknown	NCT04863066
	HIV	gp120	Interventional	Biological: CAR-T cells	I	Recruiting	NCT03240328
	HIV	gp120	Interventional	Biological: CMV/HIV-CAR T Cells	Early I	Recruiting	NCT06252402
	HIV	gp120	Interventional	Drug: Cyclophosphamide Biological: LVgp120duoCAR-T cells, low dose | high dose 1 more	I/II	Recruiting	NCT04648046
	EBV	LMP1	Interventional	Biological: Cohort A: 3.0 × 10^6 CAR-T cells/kg Cohort B: 9.0 × 10^6 CAR-T cells/kg Cohort C:1.5 × 10^7 CAR-T cells/kg	I	Completed	NCT05864924
Transplantation	Kidney transplantation	CD19/BCMA	Interventional	Drug: Cyclophosphamide Biological: CART-BCMA huCART19	I	Recruiting	NCT06056102
	Haematopoietic cell transplantation	CD19/CD20	Interventional	Drug: Donor-derived CD19 CAR Therapy Bridged Allo-HSCT and Sequential Donor-derived CD22 CAR Therapy	I	Not yet recruiting	NCT06326008

ITP immune thrombocytopenia, SLE systemic lupus erythematosus, SjS Sjogren’s syndrome, SSc systemic sclerosis, DM dermatomyositis, AAV anti-neutrophil cytoplasmic antibody‐associated vasculitis, IM inflammatory myopathy, MS multiple sclerosis, IIM idiopathic inflammatory myopathy, APS antiphospholipid syndrome, MG myasthenia gravis, AIHA autoimmune haemolytic anaemia, NMOSD neuromyelitis optica spectrum disorder, MOGAD myelin oligodendrocyte glycoprotein antibody‐associated disease, RA rheumatoid arthritis, BCMA B-cell maturation antigen

Despite its promise, challenges remain, such as high manufacturing costs, limited efficacy in solid tumors due to microenvironmental suppression, and severe adverse effects like cytokine release syndrome (CRS) [25–28]. In recent years, nanoparticles (NPs) have unlocked unprecedented opportunities in cancer immunotherapy due to their unique size effect, high specific surface area, precise targeting, and drug co-delivery capabilities, which have been successfully applied in tumor imaging, immunosuppressive microenvironment reprogramming, and vaccine delivery [29–32]. Through material selection and surface modification, the physical and chemical properties of NPs can be precisely regulated to optimize biocompatibility and improve efficiency, providing an innovative solution to break through the bottleneck of CAR-T cell therapy: NPs enhance CAR-T cell production by reducing genotoxicity and improving gene delivery efficiency, mimic artificial antigen-presenting cells (aAPCs) for T cell expansion, and enable in situ T cell engineering [33–35]. For solid tumors, NPs deliver chemokines, remodel the TME, enable multi-antigen targeting to improve CAR-T cell infiltration and persistence [36]. Additionally, NPs facilitate real-time monitoring and spatiotemporal control of CAR-T cell activity, mitigating CRS and neurotoxicity risks [37, 38].

Diverse categories of NPs have been developed for cancer immunotherapy, including polymer-based NPs, lipid-based NPs, and inorganic NPs, each exhibiting unique advantages and distinct characteristics (Table 2). Herein, this review outlines the current obstacles faced by CAR-T cell therapy and systematically summarizes the three core application advancements of NPs in CAR-T cell therapy: optimizing manufacturing process, enhancing anti-tumor efficacy, enabling real-time monitoring and toxicity control. We also deeply discuss the technical challenges and future development directions of this combined therapy, exploring how artificial intelligence (AI) can aid the development of personalized immunotherapy to provide a theoretical basis for promoting clinical transformation.

Table 2.

A summary of the pros and cons of representative nanoparticles

Types of nanoparticles	Examples	Pros	Cons	References
Polymer-based nanoparticles	Polymer nanoparticles Dendrimers Polymeric micelles	Excellent delivery carrier Tailored size and surface properties Sustained-release capability for controlled drug delivery Good stability	Risk of particle aggregation Immunogenicity Potential hepatosplenic toxicity Limited clinical application	[296]
Lipid-based nanoparticles	Liposomes Lipid nanoparticles	Good biocompatibility and high bioavailability Extensive clinical adoption Payload flexibility Low immunogenicity Ease of surface modification	Potential for drug leakage Immune perturbations in vivo Risk of hepatosplenic accumulation	[297, 298]
Protein-based nanoparticles	Ferritin, Albumin, Virus like particles	Good biocompatibility and biodegradability Bioactivity modulation Low toxicity	Potential immunogenicity Stability concerns Limited drug loading capacity	[299, 300]
Inorganic nanoparticles	Gold nanoparticles Iron oxide nanoparticles Silica nanoparticles	Unique biological, chemical, electrical, magnetic properties Suitable for diagnostics, imaging, and photothermal therapy High stability Tunable size and shape	Potential toxicity Limited solubility Clearance issues Regulatory hurdles	[301, 302]
Peptide nanoparticles	Stimuli-responsive peptide nanoparticles	Versatile functionality Biodegradability Targeted delivery potential	Potential immunogenicity Complex synthesis and stabilization requirements	[303]
Hybrid nanoparticles	Peptide-polymer hybrid nanoparticles Polymer-lipid hybrid nanoparticles	Large payload capacity High biocompatibility Good stability Low toxicity	Technical challenges in design and manufacturing Potential immunogenicity	[304]
Biomimetic nanoparticles	Membrane-coated nanoparticles Extracellular vesicles	High biocompatibility Prolonged blood circulation time Low immunogenicity Targeting capability	Complex fabrication Issues with non-specific accumulation Potential immunogenicity concerns	[305, 306]

Challenges in CAR-T cell therapy

CAR-T cell therapy encounters numerous hurdles, including inefficient manufacturing processes, suboptimal efficacy on solid tumors, poor CAR-T cell persistence, substantial antigen escape events, and serious side effects [39–41] (Fig. 2).

Fig. 2.

Challenges in CAR-T Cell Therapy. (A) In conventional CAR-T manufacturing, viral vectors present challenges including immunogenicity, potential carcinogenicity, and limited cargo capacity, while electroporation may lead to cell membrane damage and cytotoxicity. (B) The intrinsic characteristics of solid tumors restrict CAR-T efficacy and persistence in multiple dimensions: physical barriers such as dense ECM and aberrant vasculature impede T-cell infiltration into the TME; metabolic features including hypoxia, acidic conditions, and nutrient depletion, combined with immunosuppressive components, accelerate CAR-T functional exhaustion; additionally, tumor antigen heterogeneity further compromises therapeutic efficacy. (C) Moreover, CAR-T cell therapy faces safety challenges such as CRS, on-target/off-target toxicity, and insufficient real-time monitoring capabilities

Manufacturing process

The clinical manufacturing procedure of CAR-T cells encompasses multiple in vitro steps, including T-cell isolation, CAR expression on T cells, and their amplification, demanding sophisticated clinical management, substantial infrastructure resources, and significant financial investment. Among these processes, the key step lies in the genetic modification of T cells for CAR expression via either viral or non-viral transduction [42, 43]. To date, viral vectors, especially gamma-retrovirus and lentivirus, have been extensively used because of their high gene transduction efficiency and ability to integrate genes stably [42]. However, due to multiple problems of viral vectors, including immunogenicity, potential carcinogenic risks, complex production, high costs, and limited loading capacity, research has shifted to non-viral gene delivery technologies for alternative solutions [40, 44–46].

Non-viral techniques, like electroporation, offer a safer and more adaptable means of gene transfer, free from the size limitations of viral vectors. Electroporation, which utilizes pulsed voltage to create transient pores in the cell membrane, can facilitate the entry of cargo into the cell and enable the expression of the transgene, possessing advantages like large cargo capacity and the ability to be used for mRNA modification [47, 48]. Nevertheless, this technique, contingent upon electrical field strength and pulse duration, induces membrane disruption, cytoplasmic loss, and cytotoxicity, potentially altering gene expression patterns [49, 50]. (Fig. 2A). Therefore, it is imperative to explore novel cargo delivery systems to address CAR-T cell manufacturing problems and drive its development.

Efficacy and persistence

Unlike hematologic malignancies, the treatment of solid tumors remains a Gordian knot. Although recent clinical trials have demonstrated that CAR-T cell therapy shows some efficacy in certain solid tumor types, such as sarcomas and neuroblastomas, the overall therapeutic effectiveness remains limited [51, 52]. The inherent nature of solid tumors contributes to the constrained efficacy and persistence of CAR-T cells (Fig. 2B). As proof: (1) Upon injection into the bloodstream, CAR-T cells migrate to the tumor site via chemokine signals. Nevertheless, solid tumors suppress chemokine release, and the absence of matching receptors prevents their effective localization [53]. (2) After reaching the tumor site, physical barriers such as dense stroma and abnormal vasculature significantly hinder the migration, penetration, and extravasation of CAR-T cells, thereby limiting the attack on tumor cells [54]. (3) TME in solid tumors is typically characterized by hypoxia, acidity, oxidative stress, and nutrient depletion, along with immunosuppressive components such as soluble factors, cytokines like transforming growth factor-β (TGF-β), and immunosuppressive cells such as regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs) [54, 55]. This hostile environment impairs CAR-T cell proliferation and accelerates functional decline, leading to non-responsiveness and tumor relapse. (4) Tumor antigen heterogeneity, another characteristic of solid tumors, also impacts CAR-T cell efficacy [56]. The absence or alteration of tumor cell antigens prevents single-target CAR-T cells from recognizing them. Although multi-target CARs can be designed to deal with this, it may lead to the risk of on-target/off-tumor toxicity [57].

Safety

Although CAR-T cell therapy exhibits a potent killing effect, the shortfall in control over efficacy, persistence, and localization after adoptive cell transfer results in an adverse immune response harmful to healthy tissues, such as CRS, immune effector cell associated neurotoxicity syndrome (ICANS), and immune effector cell-associated hemophagocytic lymphohistiocytosis-like syndrome (IEC-HS) [58]. CRS is correlated with a substantial elevation in serum concentrations of pro-inflammatory cytokines induced by activated CAR-T cells, among which interleukin (IL)−6 functions as the key mediator [19]. Notably, this cytokine surge cascade arises not only from direct secretion by activated CAR-T cells but also from another key driver: pyroptosis, a gasdermin-mediated cell death pathway during CAR-T-mediated killing, that releases large amounts of DAMPs [12]. However, excessive DAMP release and subsequent hyperactivation of APCs further amplify proinflammatory cytokines (e.g., IL-1β, IL-6), ultimately triggering CRS [59]. Importantly, the level of GSDME correlates with CRS severity, underscoring the pivotal role of pyroptosis in CRS pathogenesis [9]. Clinically, CRS manifests as persistent fever, hypotension, hypoxia, or even multiple organ failure [19, 60]. Fortunately, CRS responds to therapies directed at IL-6 and utilizing corticosteroids [61]. Recent studies show that pretreatment with the CD19-targeting monoclonal antibody tafasitamab reduces CRS severity in CD19-CAR-T therapy through CD19 antigen occupancy while enhancing antitumor efficacy and overall survival, offering a promising strategy to advance the achievement of a rational balance between CAR-T potency and CRS risk [62]. ICANS correlates with on-target/off-tumor recognition of cerebral antigens by CAR-T cells targeting CD19 and CD22, resulting in neurological symptoms including encephalopathy, tremor, and cerebral edema [63, 64]. Other than supportive care and corticosteroids, IL-18 blockade and in situ conjugation of polyethylene glycol (PEG) to the surface of CAR-T cells (‘PEGylation’) have demonstrated potential in treating ICANS [65–67]. Like ICANS, IEC-HS frequently manifests in patients with serious CRS and is characterized by macrophage activation, leading to uncontrolled cytokine-driven inflammation. Studies have shown that corticosteroids +/- anakinra can be used for treating IEC-HS [68, 69]. In addition, CAR-T cell therapy presents other safety concerns, such as graft-versus-host disease, off-target toxicity, anaphylactic reactions, potential viral insertional oncogenesis, and inadequate monitoring [70] (Fig. 2C). Thus, further research is required to address the aforementioned issues.

Integration of nanoparticles with CAR-T cell therapy

NPs, composed of different biomaterials such as lipids, polymers, and inorganic materials, exhibit beneficial properties, including reduced toxicity, low cost, and easy customization [8]. Over the past decade, NPs have presented feasible strategies in cancer immunotherapy by means of vaccine development and immunomodulatory drug delivery, indicating their potential to overcome the aforementioned challenges of CAR-T cell therapy.

Engineer CAR-T cell therapy

Due to their varied materials, decent stability, and expanded cargo-loading options, NPs are used to engineer CAR-T cells both in vitro and in vivo to improve availability, thereby becoming desirable alternatives to conventional gene delivery techniques and boosting production efficiency [38, 71] (Fig. 3 and Table 3). Table 4 compares different gene delivery approaches in CAR-T cell therapy, highlighting key pros and cons.

Fig. 3.

Use of nanoparticles for CAR-T cell generation and schematic diagram of the CAR-T-MNPs serving as the biomimetic drug delivery. (A) For in vitro production, T cells are first isolated from patient blood, activated, and then genetically modified via NP-mediated delivery of CAR-encoding genetic material. The engineered CAR-T cells undergo ex vivo expansion before reinfusion into the patient to exert antitumor effects. Functionalized NPs can serve as aAPC platforms to activate and proliferate CAR-T cells, significantly shortening preparation. Additionally, NPs act as cell-sorting tools to enhance CAR-T cell purity. (B) For in vivo generation, functionalized NPs carrying CAR-encoding genetic payloads are directly injected into the patient’s bloodstream. These NPs selectively transduce T cells, enabling in vivo expansion of CAR-T cells for targeted tumor eradication. This process can be optimized through antibody or SORT ligand modification strategies to achieve high-efficiency in vivo transfection. (C) The CAR-T-MNPs combine the target-specificity of CAR-T cells with the drug-loading capacity of NPs, thereby prolonging circulation time and enhancing tumor-targeting efficacy. SORT: selective organ targeting; CAR-T-MNPs: CAR-T-cell membrane-coated NPs

Table 3.

Examples of nanoparticle-mediated transfection of CAR-T cells

	Carrier	Cargo	Target cells	Transfection efficiency	Disease	References
In vitro	PBAE-based nanoparticles	Leukemia-targeted 194-1BBz CAR plasmid DNA	Primary human T cells	3.8%	B-cell acute lymphoblastic leukemia.	[90]
	Ionizable lipid Nanoparticle	CD19-targeted CAR mRNA	Jurkat cells Primary human T cells	Equivalent to electroporation	Acute lymphoblastic leukemia	[33]
	Lipid Nanoparticle	CD19-targeted CAR mRNA	Human T cells	22.6%	Leukemia	[80]
	Ionizable lipid nanoparticle	CD19-targeted CAR mRNA	Jurkat cells Primary human T cells	B10 LNPs are comparable to electroporation	Acute lymphoblastic leukemia	[79]
	Gold nanoparticles	eGFP-mRNA	Jurkat cells	45%	NA	[76]
In vivo	PBAE-based nanoparticles	Leukemia-targeted 194-1BBz CAR plasmid DNA	Primary human T cells	7.1% among CD3+ T cells on day 24	B-cell acute lymphoblastic leukemia	[90]
	Cationic polymer nanoparticles	Plasmid CAR with a T cell-specific CD2 promoter	T cells in humanized NSG mice	14.3%	Ovarian cancer	[91]
	PBAE polymer nanoparticles	Disease-specific CAR mRNA	Circulating T cells	75 ± 11% of T cells on day 2	Leukemia Prostate cancer Hepatocellular carcinoma	[307]
	Imidazole-based synthetic lipidoids	Cre mRNA	Primary mouse T cells	8.2%	NA	[308]
	SORT lipid nanoparticles	CD19-targeted CAR mRNA Cre recombinase mRNA	Splenic T cells	7% of total splenic T cells	Lymphoreplete B cell lymphoma	[98]

Table 4.

Comparison between viral and non-viral vectors in CAR-T cell therapy [49, 309–311]

Types of vectors	Viral vector	Non-viral vector
		Electroporation	Nanoparticle
Advantages	Easy to use Wide range of transfection mechanisms	Large cargo capacity Low manufacturing costs	Targeting potential Highly customizable Low cytotoxicity Suitable for in vivo and in vitro applications
Disadvantages	Size-limited cargo Immunogenicity Risk of genotoxicity High variability in small-scale production High production costs	Limited large-scale manufacturing Risk of cytotoxicity Loss of cytoplasmic content Limited in vivo application	Classically low transfection capacity Toxicity related to materials Potential hepatosplenic toxicity
Efficiency	Various	Good	Various
Viability	Good	Poor	Good

For in vitro transfection, Bozza and colleagues devised a non-integrating DNA nanovector that is free of viral components and capable of extrachromosomal replication, successfully generating CAR-T cells with enhanced anti-tumor efficacy in vitro. In addition, this nanovector is characterized by persistent transgene expression, non-immunogenicity, large cargo capacity, ease of manufacture, and affordability, underscoring its multifaceted advantages. Its in vitro validation, nevertheless, relies on healthy donor T cells rather than patient-derived ones and lacks TME simulation, meaning further research is still needed for its clinical translation [72]. mRNA features easy production, low cost, no concerns regarding long-term side effects, and flexible protein-encoding ability, thus serving as a potent tool for T cell gene engineering in vitro [73]. Moreover, NPs can encapsulate mRNA to protect it and facilitate T-cell uptake, thereby reprogramming tumor-associated genes in T cells via transient expression [33]. Researchers have employed miscellaneous types of NPs to boost mRNA delivery efficiency. For example, it was demonstrated that the structure of cationic polymers, the key components of polymer-based NPs, is pivotal in governing both gene transfection efficiency and cytotoxicity. In contrast to linear poly (2-dimethyl amino) ethyl methacrylate) (pDMAEMA), the comb- and Sunflower-shaped pDMAEMAs exhibited higher transfection efficiency and comparatively lower cytotoxicity, furnishing a valuable avenue for designing efficacious vectors to ensure successful in vitro gene delivery. However, their transfection efficiency remains lower than that of viral vectors, with transfection efficiency in primary Human T cells achieving only 25% (for mRNA) and 18% (for plasmid DNA), alongside limited in vivo validation and cell type specificity, hindering clinical translation due to unmet efficacy and generality demands [74, 75]. Gold NP-mediated vapor nanobubble photoporation, which attained transfection rates of up to 45% and increased the number of transfected viable cells five times compared to electroporation in Jurkat T cells, is also a promising in vitro mRNA delivery technique. Nonetheless, it lacks validation in primary human T cells, has unclear long-term effects on cell functionality, and faces scalability challenges [76]. Recently, ionizable lipid nanoparticles (LNPs) have been applied for in vitro mRNA delivery to human T cells [77, 78]. Through excipient optimization to enhance delivery efficiency and mitigate cytotoxicity, the B10 formulation emerged as the top performer, tripling the functional delivery compared to the standard one. Moreover, its generated CAR-T cells exhibited equivalent cancer cell-killing efficacy to those induced by lentivirus and electroporation. Even so, the mechanisms by which B10’s excipient composition enhances delivery remain unclear [79]. Furthermore, LNPs extend their applicability to other cell types, such as macrophages and natural killer (NK) cells. A study showed that by optimizing the LNP formulation and mRNA modification, in vitro mRNA transfection was achieved and anti-CD19 CAR macrophages were successfully produced, showing significant cytotoxic effects on B lymphoma [80]. Overall, among current mRNA nanocarriers for in vitro transfection, LNPs demonstrate the highest clinical potential, although each platform necessitates tailored optimization to overcome its particular limitations toward translational application.

The activation and expansion of CAR-T cells engineered in vitro is labour and time intensive, underlining the need for an efficient way to streamline production. T cell activation demands three signals: T cell receptor (TCR) signaling, costimulation, and cytokine signaling. In vivo, APCs deliver these signals, while in vitro, aAPCs are deployed to drive T cell activation [81]. Despite the capacity of prevalent commercial microbead aAPC systems like Dynabeads to restore T cell traits, issues including suboptimal T cell expansion rates, misregulated or limited functions of the generated T cells, and the necessity for extra steps to remove microbeads render the in vitro T cell activation and proliferation process cumbersome [82–84]. Fortunately, NPs can be modified and serve as aAPC platforms, enabling in vitro activation and proliferation of CAR-T cells and shortening the preparation time (Fig. 3A). Metzloff et al.. achieved one-step efficient production of mRNA CAR-T cells by directly conjugating fragments of CD3 and CD28 antibodies onto the surface of LNPs mimicking APCs, which subsequently alleviated tumor burden in murine xenograft models [85]. Notably, activating lipid nanoparticles (aLNPs) slashed the cell culture time for mRNA CAR-T cells by half, from 48 to 24 h. Simultaneously, dispensing with the need for activation magnetic beads substantially reduced both the preparation time and manufacturing costs, while the study is limited to preclinical murine models, long-term safety and scalability for clinical use remain unproven [85].

NPs could also serve as a pivotal tool in cell sorting, playing a significant role in CAR-T cell preparation (Fig. 3A) [86]. NPs significantly enhanced magnetic labeling efficiency by utilizing magnetic amplification by sequentially connected antibodies and nanoparticles (MagSCAN) technology. This method increased the purity of positive fractions from 35 to 50% with conventional methods to 80%, thereby enabling the efficient isolation of high-purity CAR-T cells while maintaining cell viability above 90%. The sorted CAR-T cells demonstrated superior performance in anti-tumor therapy: when co-cultured with tumor cells, their killing efficiency was markedly enhanced, evidenced by increased tumor cell mortality and elevated release of pro-inflammatory molecules, such as tumor necrosis factors (TNFs) and interferon-gamma (IFN-γ). Additionally, the expression levels of cytokines, including IL-6 and IL-10, were significantly upregulated, indicating enhanced anti-tumor activity. Thus, the NP-based MagSCAN technology effectively improved the therapeutic efficacy of CAR-T cell therapy by enhancing both cell purity and functional activity. Notably, the linkers and MNPs involved may trigger immune responses, making direct clinical application unadvisable. Moreover, its effectiveness in whole blood without removing residual antibodies remains to be verified [87].

To circumvent the constraints of complexity, heavy cost, and high personalization associated with large-scale implementation of ex vivo-produced CAR-T cell products, in vivo T cell engineering holds promise (Fig. 3B) [88]. Direct reprogramming of patient-derived T cells into CAR-T cells enables the generation of patient-specific cellular therapeutics. Polymer nanoparticles (PNPs) and LNPs are the most common and promising nanovectors employed in CAR-T cell engineering [89]. Utilizing PNPs as delivery vectors for in vivo generation of CAR-T cells constitutes a viable strategy. DNA-loaded PNPs delivered the transgene encoding leukemia-specific 194−1BBz CAR into circulating T-cell nuclei, leading to the regression of leukemia in mice, with an efficacy comparable to that of conventionally manufactured CAR-T cells. Yet, these studies are confined to murine models, and off-target transgene expression, though minimal, remains a concern [90]. Notably, in a recent study, Zhu and colleagues reported that they successfully generated CAR-T cells in vivo by injecting a supramolecular hydrogel system based on cationic PNPs and demonstrated its potent anti-tumor efficacy in NSG humanized xenograft mouse models. This method boasted strong local retention and low systemic toxicity, yet it relied on local tumor injection and had limited efficacy against metastases [91]. In comparison, PNPs enabled systemic administration, offering greater convenience in hematological tumors, but suffering from insufficient penetration in solid tumors. Both strategies avoided the risks associated with viral vectors, providing complementary optimization approaches for CAR-T therapy [90, 91]. LNPs, another ideal carrier for promoting CAR-T cell therapy, enable in vivo generation of CAR-T cells by transporting therapeutic mRNA to lymphocytes [92]. Nonetheless, the current in vivo T cell production efficiency is suboptimal, and direct transfection of mRNA into in vivo T cells remains a technical hurdle [93]. The design and modification of NP surfaces may facilitate overcoming these hurdles by enhancing the in vivo-specific transfection of T cells (Fig. 3B). Zhou et al. constructed a CD3 antibody-modified LNP system loaded with a plasmid encoding IL-6 short hairpin RNA (shRNA) and CD19-CAR, which targeted and transfected T cells to generate IL-6 knockout CAR-T cells [92]. These engineered cells could kill leukemia tumor cells with high CD19 expression, reduce IL-6-induced CRS, and exhibit comparable anti-tumor effects to traditional CAR-T cells prepared ex vivo, greatly prolonging the survival of leukemia model mice. However, the transfection efficiency of this system remains lower than that of lentiviral methods, and off-target effects may still pose potential risks [92]. Furthermore, studies have demonstrated that anti-T cell marker selection (e.g., CD3/CD5/CD7) and antibody-conjugated lipid nanoparticles (Ab-LNPs) dosage critically regulate CAR expression levels in T cells. In vivo investigations revealed that CD3- and CD7-LNPs effectively delivered CAR mRNA, inducing robust CAR expression in circulating T cells, whereas CD5-LNPs failed to achieve efficient transfection or expression even at maximal doses. Remarkably, increasing the dosage of CD3- and CD7-LNPs further enhanced CAR expression [94].

Notably, targeted ligand modification isn’t essential for LNP-mediated in vivo T cell transfection. Certain LNPs with optimized formulations can circumvent the liver tropism of non-targeted carriers and accumulate in extrahepatic tissues [95]. Wang and co-authors developed a novel technique termed selective organ targeting (SORT), allowing for the controlled delivery of nucleic acids to target tissues without modifying the targeting ligands of NPs [96]. By incorporating a fifth component, the SORT molecule, into the traditional four-component LNP, they achieved precise regulation of tissue-specific delivery to the livers, lungs, and spleens of mice after intravenous injection [96]. Some researchers have identified the relevant mechanism: serum proteins can recognize exposed SORT molecules, adsorb to LNPs, and interact with target-organ cell receptors for targeted delivery [97]. Álvarez-Benedicto et al. have demonstrated that they achieved in situ production of CAR-T cells via intravenous injection of spleen-SORT LNPs unmodified by targeting ligands, reducing tumor metastasis, and improving the overall survival rate of B-cell lymphoma mice [98]. Notably, SORT LNPs have also been engineered for targeted delivery to other organs, including the lung, brain, pancreas, placenta, bone marrow, and lymph nodes [99–104]. To advance clinical translation, large animal model studies should be conducted to validate the safety and efficacy of SORT LNPs in a physiological environment more analogous to humans. Furthermore, optimizing LNP formulations and CAR designs can enhance targeting efficiency and therapeutic efficacy against highly aggressive tumors, thereby mitigating tumor escape.

Compared with conventional PNP and LNP-based gene delivery approaches, a recent study designed a novel in vivo CAR-T cell generation strategy through membrane fusion-mediated CAR protein delivery [105]. The researchers developed virus-mimetic fusogenic nanovesicles (FuNVs) as CAR protein carriers, which efficiently fused with T-cell membranes in vivo to deliver anti-CD19 CAR proteins directly to T cells, thereby generating functional anti-CD19 CAR-T cells. Although the strategy streamlined CAR-T production, avoided CRS, and showed potent anti-tumor effects, it might not have suited patients with impaired T-cell function, and fusogens with higher specificity were needed to reduce non-specific binding [105]. Interestingly, CAR-T cells can also serve as starting materials for the fabrication of NPs (Fig. 3C). Illustratively, the mesoporous silica NPs loaded with near-infrared (NIR) dye IR780 were coated with CAR-T cell membranes that specifically recognize Glypican-3⁺ hepatocellular carcinoma (HCC) cells, while poly (D, L-lactide-co-glycolic acid) NPs encapsulating the anticancer drug cisplatin were functionalized with anti-human epidermal growth factor receptor 2 (HER2)-specific CAR-T cell membranes. What’s more, 3-bromopyruvate-mixed Au NR@Cu₂₋ₓSe NPs were coated with CAR-T cell membranes capable of specifically recognizing CD19 antigens on the surface of NALM6 tumor cells, demonstrating potent anti-tumor efficacy both in vitro and in vivo [106–108]. Therefore, CAR-T cell membrane-coated NPs (CAR-T-MNPs) also represent a promising biomimetic delivery platform that combines CAR-T cell targeting specificity with NP drug-loading capacity, potentially augmenting the efficacy of various cancer treatments. For advancing the development of this field in the future, it is necessary to further improve the stability of loaded substances and verify the long-term in vivo metabolic safety of nanomaterials.

Enhance the anti-tumor efficacy of CAR-T cell therapy

Promote aggregation at tumor sites

Insufficient infiltration of T cells into tumor tissues impedes the efficacy of CAR-T cell therapy in treating solid tumors. To address this, NPs have been utilized to promote the aggregation of CAR-T cells to tumor sites (Fig. 4). IL-12 nano-stimulators (INSs), designed as redox-responsive human serum albumin NPs with IL-12, exemplify this approach [109] (Fig. 4A). Through bioorthogonal conjugation, INSs were conjugated with CAR-T cells to form INS-CAR-T biohybrids. Upon tumor antigen stimulation, the surface reducibility of CAR-T cells increased, prompting INS to release IL-12 in a responsive manner. This, in turn, drove the secretion of chemokine C-C motif ligand (CCL)−2, CCL5, and CXC chemokine ligand (CXCL)−10, selectively recruiting and expanding CD8⁺ CAR-T cells within the tumor and bolstering the anti-neoplastic potency. Although this design enables precise antigen-responsive activation, its effectiveness depends on specific antigen expression and may be limited in tumors with low antigen levels [109]. Beyond enhancing chemokine secretion, NPs can reshape the TME by removing physical barriers via photothermal therapy, thereby facilitating CAR-T cell therapy [110] (Fig. 4B). Tang et al. developed a hydroxyethyl starch-polycaprolactone (HES-PCL) nanodelivery system to co-load the TGF-β receptor inhibitor LY2157299 (LY) and photosensitizer indocyanine green (ICG), generating LY/ICG@HES-PCL NPs. These NPs were targeted to the lymphoma site via the enhanced permeability and retention (EPR) effect. The photothermal effect of ICG could destroy the extracellular matrix (ECM), dilate blood vessels, and loosen compact tissue, accelerating LY release. This release process could upregulate chemokines CXCL9/10/11 and the CAR-T cell CXC chemokine receptor (CXCR)−3 at the tumor site, facilitating CAR-T cell infiltration and enabling long-term anti-tumor effects. Despite its controllability, the utility of this system is limited by the shallow penetration depth of light [110, 111]. Recent research has shown that chromium NP-based photoimmunotherapy harbors potential for enhancing CAR-T cell infiltration in solid tumors. The trivalent Cr3⁺ ions generated upon degradation of the nanocomposites could upregulate the expression of chemokines CCL3 and CXCL13, facilitating CAR-T cell infiltration. Even so, the long-term metabolic safety of Cr-based materials requires further validation [112]. Collectively, these technologies provide complementary strategies to promote the aggregation of CAR-T cells to tumor sites. The integration of precise targeting and microenvironment remodeling may offer promising directions for future research.

Fig. 4.

Strategies for nanoparticle-enhanced CAR-T Cell tumor infiltration. (A) Nano-stimulators enable redox-responsive IL-12 delivery, promoting secretion of chemokines to enhance CAR-T cell tumor recruitment and expansion. (B) NPs can be combined with PTT to co-deliver TGF-β inhibitor LY and photosensitizer ICG. This approach utilizes photothermal effects to disrupt ECM, dilate blood vessels, promote CAR-T cells infiltration, and upregulate chemokine expression. (C) Magneto-acoustic manipulation technology improves the transport efficiency of CAR-T cells conjugated with immunomagnetic NPs to tumor sites, enabling precise tumor localization and effective infiltration. PTT: photothermal therapy

Researchers could also employ magnetic-acoustic manipulation technology to enhance the transport efficiency of CAR-T cells to tumor sites (Fig. 4C). Immunomagnetic particles carrying anti-CD3/CD28 antibodies were conjugated to CAR-T cells via click chemistry, generating CAR-T-cell-based live microrobots (M-CAR-Ts) [113]. Under magnetic fields, M-CAR-Ts moved against blood flow, bypassed obstacles, and precisely targeted tumor periphery regions, while under acoustic fields, M-CAR-Ts further penetrated tumor tissues, promoting their infiltration into tumor sites. Additionally, the anti-CD3/CD28 antibodies also stimulated CAR-T cell proliferation and activation, significantly improving their tumor-suppressive effectiveness [113]. Nonetheless, it should be noted that this study was conducted solely in mouse models, which exhibit significant physiological differences from humans. The precision and stability of the magnetic-acoustic driving system may be compromised in deep tissues due to potential interference, potentially reducing its efficacy in larger animals or humans. Moreover, the immunomagnetic beads could elicit immune responses, and their long-term safety remains uncertain.

Modulate hostile TME

Upon infiltrating tumor sites, CAR-T cells encounter the hostile TME, which features hypoxia, acidity, and nutrient deprivation, along with various inhibitory cells and cytokines. These factors cause T cell dysfunction and induce treatment resistance [114–116]. NPs can reverse the immunosuppressive TME through multiple modalities, thereby enhancing the therapeutic efficacy of CAR-T cells in solid tumors (Fig. 5). Dong et al. developed catalase-encapsulated CaCO₃ nanoparticle-constructed colloidosomal microreactors (CCaP CSs), aiming to regulate tumor hypoxia and acidity within the TME [117]. By leveraging CaCO₃’s proton-scavenging capacity to neutralize acidic conditions and catalase’s ability to decompose H₂O₂, this dual-modality approach effectively alleviated hypoxia while reducing lactate accumulation. Notably, this TME remodeling strategy reversed immunosuppression and significantly potentiated the tumor suppression efficacy of epidermal growth factor receptor (EGFR)-expressing CAR-T cells in modeled human triple-negative breast cancer tumor xenografts. However, CCaP CSs rely on intratumoral injection, limiting their applicability to inaccessible tumors, and their oxygen generation depends on endogenous H₂O₂ levels, which may restrict efficacy in severely hypoxic environments [117]. Apart from directly modifying the hypoxic and acidic states of the TME, NPs can also indirectly remodel the immuno-hostile TME by regulating tumor cell metabolism [114]. The developed APHA@CM nanocatalyst, prepared by encapsulating horseradish peroxidase (HRP)-loaded Au/polydopamine nanoparticles (Au/PDA NPs) and Ag₂S quantum dots, targeted tumor cells by leveraging CAR-T cell membranes. This membrane-camouflaged design enhances tumor targeting and reduces immune clearance, a distinct advantage over CCaP CSs’ passive accumulation. Additionally, APHA@CM integrates multimodal imaging for precise therapeutic guidance, a feature absent in CCaP CSs. Au NPs mediated synergistic sonodynamic therapy (SDT) to induce immunogenic cell death (ICD) and reprogram the immunosuppressive TME. The PDA layer and HRP enhanced the efficiency of SDT, and HRP catalyzed the production of oxygen to alleviate hypoxia. Meanwhile, Au NPs exhibited glucose oxidase-like enzymatic activity, which inhibited glycolysis, reduced lactate efflux, improved the metabolic microenvironment, and promoted the activation and survival of CAR-T cells. Despite these advantages, APHA@CM’s reliance on external stimulation, such as ultrasound/laser for SDT/PTT, complicates clinical operation, and the long-term biocompatibility of Au-based materials remains to be verified [118] (Fig. 5A).

Fig. 5.

Strategies for nanoparticles reverse the immunosuppressive TME. (A) NPs can alleviate TME hypoxia and acidity: CaCO₃ NPs alleviate acidity through proton scavenging while catalase-mediated decomposition of H₂O₂ generates oxygen, reducing lactate accumulation. The APHA@CM nanocatalyst system employs HRP for oxygen production and utilizes the glucose oxidase-like activity of gold NPs to inhibit glycolysis and suppress lactate efflux, thereby improving metabolic conditions. (B) NPs deliver immunomodulators such as TLR7/8 agonists, PI3Kδ inhibitors, STING agonists, or anti-tumor cytokines in the TME, remodeling the immunosuppressive TME and augmenting the anti-tumor potency of CAR-T cells. (C) NPs can be co-encapsulated with hydrogels, enabling localized delivery of CAR-T cells, IL-15 cytokines, and anti-PD-L1 antibodies to effectively overcome immunosuppression in solid tumor microenvironments. HRP: horseradish peroxidase

NPs can also deliver diverse immunomodulatory drugs to suppress immunosuppressive cells and inhibitory molecules within the TME, thereby blocking immunosuppressive pathways (Fig. 5B). To illustrate, polymeric micellular nanoparticles (PMNPs) were engineered to co-deliver Toll-like receptor 7 and 8 (TLR7/8) agonists and phosphatidylinositol-3 kinase p110δ (PI3Kδ) inhibitors. This combination therapy promoted type 1 macrophage polarization while reducing MDSC accumulation and selectively depleting tissue-resident Tregs, thereby mitigating the immunosuppressive TME and paving the way for enhancing CAR-T cell therapy outcomes [119]. In another study, Zhu and colleagues developed nanovesicles expressing anti-programmed cell death Ligand 1 (PD-L1) and loaded with the interferon gene stimulator (STING) agonist cGAMP [120]. Upon inhalation, these nanovesicles swiftly accumulated in lung tissue and ferried cGAMP to PD-L1 highly-expressing cells in tumor tissue via the programmed cell death protein 1 (PD-1)/PD-L1 interaction. The internalized cGAMP triggered the release of pro-inflammatory cytokines, spurring CAR-T cell infiltration and depleting immunosuppressive cell populations such as Tregs and MDSCs. Meanwhile, by blocking PD-L1, the nanovesicles prevented CAR-T cell exhaustion, remodeling the immunosuppressive TME and augmenting the proliferation and anti-tumor potency of CAR-T cells in solid tumors [120]. Both strategies target immunosuppressive cells and amplify pro-inflammatory signals, yet differ in their delivery routes and primary mechanisms. Systemically administered PMNPs modulated myeloid cell polarization and Treg function, while inhalable lung-targeted nanovesicles integrated STING-mediated immune activation with checkpoint blockade to directly enhance CAR-T cell persistence. Together, these approaches demonstrate the versatility of NPs in remodeling the TME to improve immunotherapy, with their efficacy in solid tumors being influenced by route- and target-specific characteristics. Other than the aforementioned studies, immunomodulatory drugs like TGF-β inhibitors, A2a adenosine receptor antagonists, and α-GalCer agonists have also been documented [36, 121–123].

Numerous studies have validated the role of pro-inflammatory cytokines, another type of immunomodulatory agent, in potentiating anti-tumor immunity [124, 125]. Nevertheless, systemic administration of cytokines may lead to severe systemic toxicity [126]. Consequently, a NP-based “cellular backpack” strategy has been devised to accurately deliver cytokines at the opportune time and location, thereby synergistically enhancing T cell cytotoxicity and maximizing therapeutic outcomes while minimizing toxicity [127]. Tang et al. developed nanogels (NGs) loaded with IL-15 superagonist complex (IL-15Sa) [128]. Upon CAR-T cells recognizing tumor antigens, the NGs responded to the elevated reduction potential on the T cell surface, conjugated the IL-15Sa-loaded NPs with the CAR-T cell plasma membrane, and released IL-15Sa into the TME. This promoted the proliferation of intratumoral T cells while averting impacts on peripheral blood T cells. Compared with systemic administration of IL-15Sa, the NGs system boosted intratumoral T cell proliferation by a factor of 16, allowed for the safe use of up to 8 times the dose of IL-15Sa without eliciting toxic reactions, and markedly improved the tumor clearance rate [128]. Nevertheless, the “cellular backpack” strategy suffers from drawbacks such as complex manufacturing processes, lengthy production times, and potential impairment of T cell physiological functions. To address these and enhance CAR-T cell anti-tumor efficacy, Liu et al. labeled T cells by adding NPs containing unnatural azido sugars to the culture medium during cell expansion and employed click chemistry to directly attach anti-tumor cytokines to the cell surface. The approach augmented the in vivo effector function of T cells after adoptive transfer, activated the endogenous immune system to promote antigen spreading, and helped identify more tumor-specific antigens. It should be noted that metabolically labeled cytokines lack dynamic release regulation, potentially resulting in sustained systemic immune activation [129]. Besides the aforementioned strategies, CAR-T cells and auxiliary NPs can be co-encapsulated into hydrogels for local therapy (Fig. 5C). A biodegradable hydrogel reservoir for the localized delivery of CAR-T cells, IL-15 cytokine, and anti-PD-L1 antibody post-tumor resection was developed to prevent tumor recurrence, effectively overcoming immunosuppression in the solid TME and CAR-T cell depletion. Despite this promising approach, the necessity for surgical implantation of hydrogels may increase procedural complexity [130]. Overall, these findings indicate that localized delivery and stimulus-responsive release represent key strategies for enhancing T cell therapies. Technical pathways should be selected based on the characteristics of the TME: for instance, immunogenic tumors may benefit from nanogel-based systems, metabolic labeling offers advantages in manufacturing simplicity, while hydrogels are particularly suitable for surgically accessible tumors.

Restrict tumor antigen heterogeneity

Tumor antigen heterogeneity, characterized by antigenic differences and high mutation diversity among tumor cells, poses a major challenge to CAR-T cell therapy [131]. Researchers have carried out extensive investigations and devised multiple strategies to address tumor antigen heterogeneity, including multi-target CAR-T cells against diverse tumor-associated antigens (TAAs), BiTE (bi-specific T cell engager)-secreting CAR-T cells that enhance tumor recognition or universal CAR-T cells targeting switchable adaptors for flexible antigen engagement [132–134] (Fig. 6A).

Fig. 6.

Strategies to restrict tumor antigen heterogeneity. (A) Non-NP-based strategies to restrict tumor antigen heterogeneity. (B) The multifunctional NP (NbFGFR4-GrB-h-HFn) co-delivers nanobodies NbFGFR4 targeting tumor-specific antigen FGFR4 and broad-spectrum CD71 ligands (h-HFn). This design enables robust binding and GrB delivery across gastric cancer cells with variable FGFR4 expression, significantly enhancing the recognition of antigenically diverse tumors. (C) F-AgNPs utilize membrane fusion to modify exogenous antigens onto tumor cell surfaces. These in situ-modified antigens provide abundant recognition sites for CAR-T cells, enabling them to overcome limitations imposed by intrinsic tumor antigen heterogeneity and exert potent antitumor effects. (D) NP-vaccine combination strategy: Administered subcutaneously or intravenously, vaccine NPs are either internalized by peripheral DCs or directly transported to lymph nodes for DC interaction. Activated DCs migrate to lymph nodes to enhance CAR-T cells, thereby directing their tumor-specific killing. Subsequent tumor lysis releases diverse antigens (including non-vaccine-targeted neoantigens), which are captured by DCs and presented in draining lymph nodes. This process activates T-cell responses, broadening the immune system’s antigen recognition spectrum through antigen spreading. F-AgNPs: fusogenic antigen-loaded nanoparticle

NP-based CAR-T cell therapy also bears great prospects in restricting tumor antigen heterogeneity. The multitargeting NPs capable of binding both tumor-specific antigens (TSAs) and widely expressed antigens were designed to enhance the recognition of heterogeneous tumor cells [135]. Specifically, they fused the nanobodies (Nbs) targeting the TSA FGFR4 of gastric cancer with granzyme B (GrB) and conjugated the complex onto human ferritin (h-HFn), generating NbFGFR4-GrB-h-HFn NPs. These NPs utilize NbFGFR4 to target FGFR4 while leveraging h-HFn’s binding to CD71 (a receptor highly expressed on tumor cells) for dual targeting, enabling robust binding to gastric cancer cells across varying FGFR4 expression levels and delivering GrB. Notably, even in FGFR4-low-expressing MGC803 cells, CD71 ensures effective tumor engagement, thus overcoming antigen heterogeneity [135] (Fig. 6B). Similar strategies have been adopted in HCC characterized by high FGFR4 expression. FGFR4-ferritin NPs were administered to alpacas to construct a Nb phage library, from which FGFR4-targeting Nbs with high specificity and affinity were screened. The Nb-derived CAR-T cells, developed accordingly, overcame the antigen downregulation issues due to the inhibitory microenvironment and tumor heterogeneity, demonstrating remarkable anti-tumor efficacy both in vitro and in vivo. While Nb-derived CAR-T cells provide the benefit of sustained tumor surveillance via T cell expansion, their efficacy is limited by the dependence on FGFR4 expression, restricting their use in tumors with low or absent antigen levels [136]. In contrast, NbFGFR4-GrB-h-HFn NPs demonstrate broader applicability across a range of FGFR4 expression levels. However, NPs’ delivery efficiency is compromised by the structural barriers of solid tumors, which needs further study.

While these pioneering efforts have demonstrated the potential to enhance the efficacy of CAR-T cells in treating solid tumors, they remain confined to tumors possessing suitable intrinsic antigens, which underpin the functionality of CAR-T cells. In response, some researchers proposed leveraging exogenous antigens to perform in situ modification on solid tumors, thereby providing ample targets and achieving the redirection of immune recognition [137] (Fig. 6C). The developed fusogenic antigen-loaded nanoparticles (F-AgNPs) exploit the properties of tumor cell membranes after malignant transformation, namely negative charge and enhanced fluidity, facilitating their efficient fusion with F-AgNPs containing the cationic lipid DOTAP. Additionally, it employed tumor-derived extracellular vesicles (EVs) to propel antigen diffusion within tumor tissues. Through in situ antigen modification on the tumor cell membrane, this platform redirected CAR-T cells, regulated the expression level of exogenous antigens via dose adjustment, and averted target antigen loss through repeated administration. Consequently, a tumor-inherent antigen-independent target-redirected universal CAR-T (TRUE CAR-T) cell therapy was established, effectively overcoming tumor antigen heterogeneity. Nevertheless, the tumor selectivity of F-AgNPs relies on the membrane properties of tumor cells; off-target risks remain due to potential uptake by normal cells. Moreover, their tendency to accumulate in the liver and spleen necessitates further investigation into long-term safety [137].

Besides, coupling NPs with vaccine therapy to target diverse tumor antigenic epitopes and engage endogenous immunity also shines as a beacon of hope for overcoming tumor heterogeneity in CAR-T cell therapy [138–140] (Fig. 6D). The amph-vax vaccine developed by Ma et al. consists of amphiphilic ligands and adjuvants. Upon administration, it selectively migrates to lymph nodes and modifies the surface of APCs, significantly enhancing antigen presentation efficiency. This vaccine activates and expands potent tumor-killing CAR-T cells while overcoming tumor antigen heterogeneity through endogenous immune system activation and antigen spreading. Mechanistically, IFN-γ secreted by CAR-T cells not only sustains their cytotoxicity and effector factor expression but also recruits dendritic cells (DCs) and promotes antigen uptake. Under IFN-γ stimulation, DCs secrete IL-12, creating a positive feedback loop that amplifies CAR-T effector responses [138, 141]. Furthermore, coupling this vaccine with NPs may enable more efficient lymph node-targeted delivery and APC anchoring, thereby potentiating CAR-T cell therapeutic efficacy. In another study, Zhang et al. fabricated a nanovaccine that encapsulated the TLR-9 agonist CpG ODN and was combined with tumor cell lysates (TCLs) and neoantigens [142]. Experimental results indicated that the NPs could efficiently transport CpG ODN and diverse tumor antigens to the draining lymph nodes. Therein, the CpG ODN spurred DC maturation and drove the secretion of proinflammatory cytokines like IL-6 and IFN-γ. Meanwhile, mature DCs manifested improved antigen-presenting capabilities, activating both T cells and B cells to elicit cellular and humoral immune responses. Furthermore, the vaccine-induced antibody generation synergized with CD16 CAR-T cells through antibody-dependent cellular cytotoxicity (ADCC) to impede tumor growth. Notably, the molecular mechanisms of ADCC and therapeutic synergy remain poorly defined. The use of TCLs and neoantigens in the nanovaccine introduces compositional heterogeneity, potentially resulting in batch variability and complicating standardized production. Future work should focus on validating findings in broader tumor models and species, clarifying the mode of action, and improving manufacturing consistency and safety evaluation [142].

In addition to the aforementioned aspects, NPs can augment the anti-tumor efficacy of CAR-T cell therapy via multiple mechanisms. Primarily, NPs mimic APCs to directly activate and stimulate T cells, exhibiting remarkable therapeutic potential in animal models [141]. Recent research has shown that NPs integrated CAR ligands into the immunological synapse (IS), actively recruiting APCs rather than merely mimicking them. This strategy leverages natural immune cell interactions to enhance targeting specificity, thereby optimizing CAR-T cell activation and bolstering anti-tumor efficacy. However, its effectiveness depends on CD169 expression by APCs, which may limit applicability in tumors with low CD169 levels [143]. Secondly, NPs can be combined with other immunotherapies to further amplify the anti-neoplastic activity of CAR-T cells. For instance, free delivery of immune checkpoint inhibitors such as blockers of PD-1, lymphocyte-activation gene 3 (LAG-3), T cell immunoreceptor with Ig and ITIM domains (TIGIT), T cell immunoglobulin and mucin-domain containing 3 (TIM-3), or cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) bears great prospects in relieving T cell immunosuppression [144–147]. Alarmingly, extensive inhibition by free blockades may incite autoimmune reactions. To address this issue, researchers utilized LNPs to co-deliver CAR mRNA and small interfering RNA (siRNA) targeting PD-1 to T cells, temporarily disrupting endogenous PD-1 expression and reducing systemic immunosuppression risks. Despite these benefits, the efficacy of this approach relies on RNA stability and may require repeated administrations to sustain therapeutic effects [148]. Moreover, anchoring anti-PD-1 nanogels to CAR-T cell surfaces enhanced treatment efficacy through spatial-temporal co-localization and controlled drug release. Even so, the anchoring efficiency depends on cell surface modification steps, which may increase clinical complexity [149]. The combination of NPs-based CAR-T cell therapy with traditional therapies such as targeted therapy and radiotherapy unveils potential. For example, PMNPs concurrently delivered PI3Kδ inhibitors and TLR7/8 agonists, enhancing T cell adaptive immune responses and augmenting the anti-tumor effect of local radiotherapy [119, 150]. NPs could also modify tumor cells by delivering the Fas gene to heighten the killing power of CAR-T cells against tumor cells. Yet, this strategy relies on the Fas-FasL pathway, which may limit its efficacy in tumors with low FasL expression [151]. In summary, each of the aforementioned technologies presents distinct advantages. Optimal strategy selection should be guided by specific tumor characteristics, target antigen expression profiles, and therapeutic context.

To address the bottlenecks of CAR-T cell therapy, such as immunosuppressive TME, cell exhaustion, and antigen heterogeneity, recent studies have focused on designing novel CAR variants incorporating innate immune domains. By activating innate immune signals and synergizing with adaptive immune effects, these variants significantly enhance the function and efficacy of CAR-T cells [152–156]. As core pattern recognition receptors (PRRs) of the innate immune system, the TLR family participates in the costimulation of CAR-T cells by regulating T cell activation, differentiation, and cytokine secretion [157]. Incorporating the Toll/interleukin-1 receptor (TIR) domain of TLR2 into CAR constructs targeting CD19 or mesothelin upregulated the secretion of proinflammatory cytokines (e.g., IL-2 and IFN-γ) and the expression of genes related to cell adhesion and synaptic transmission, significantly enhancing the in vitro and in vivo cytotoxic activity of CAR-T cells against leukemia and solid tumors [153]. In a clinical trial (NCT02822326), 1928zT2 CAR-T cells harboring the TLR2 domain achieved complete remission in patients with relapse and refractory acute lymphoblastic leukemia (ALL), and mild CRS occurring in some treated patients was effectively managed with tocilizumab treatment [158]. The integration of the TLR4 domain also shows promise. Studies by Mikolič et al. demonstrated that second-generation anti-CD19 CAR-T cells containing the TLR4 domain exhibited stronger activation in both hematologic malignancies and solid tumor models. Although the response to solid tumors was slower, all tumor-bearing mice ultimately achieved complete remission [155].

The signaling module, composed of MyD88 (Myeloid Differentiation Primary Response gene) and CD40, represents another key optimization strategy. As a critical adaptor protein in the TLR pathway, MyD88 triggers the NF-κB (nuclear factor kappa B) and MAPK (mitogen-activated protein kinase) pathways by activating IRAK (IL-1 receptor-associated kinase) and TRAF6 (TNF receptor-associated factor 6) [159]. CD40, a costimulatory receptor, further amplifies these signals via TRAF proteins. Their synergy significantly enhances T cell proliferation, survival, and cytokine production [159]. Studies by Prinzing et al. confirmed that MyD88/CD40-containing CAR-T cells maintained a less differentiated phenotype (downregulated TBET and Blimp-1 expression) and upregulated the expression of cell cycle regulatory genes (MYB and FOXM1) after repeated stimulation, directly supporting their anti-exhaustion potential [160].

As a C-type lectin receptor (CLR), NKG2D (natural killer group 2-member D) recognizes stress-induced ligands (e.g., MICA, MICB, and ULBPs) that are upregulated on tumor cells due to DNA damage or oncogenic stress. Via adaptor proteins DAP10 (activating the PI3K-Akt pathway) or DAP12 (activating the Syk/ZAP70 pathway), NKG2D provides activation and signal integration for CAR-T cells without additional costimulatory domains, effectively addressing tumor antigen heterogeneity [161, 162]. Currently, NKG2D-CAR-T products such as CYAD-01 have demonstrated significant tumor clearance potential in clinical studies for acute myeloid leukemia (AML), myelodysplastic syndromes (MDS), and multiple myeloma (MM) [163].

These findings confirm that the integration of innate immune signals with CAR-T cells synergizes innate and adaptive immune mechanisms, laying a foundation for next-generation therapies. Notably, NPs, with their excellent carrier properties, have enabled the precise delivery of innate immune ligands such as TLR agonists, effectively enhancing the targeted activation efficiency of immune cells and providing critical technical support for regulating the functions of innate immune components [119, 142, 164]. In the future, integrating NP delivery technology and AI-based design optimization is expected to further overcome the technical bottlenecks of CAR-T therapy, promote its application in solid tumors and long-term efficacy, and provide a more robust and durable solution for cancer immunotherapy.

Strengthen the monitoring and control of CAR-T cell therapy

In CAR-T cell therapy, severe adverse reactions like CRS partly result from inadequate monitoring of the location and duration of CAR-T cell-induced tumor cytotoxicity [165]. Consequently, it is imperative to develop CAR-T cell tracking methods. This will provide a comprehensive understanding of their in vivo biodistribution and behavioral patterns in patients, enabling timely troubleshooting to ensure safety and permitting prompt evaluation of treatment efficacy for therapy optimization. NPs present a significant opportunity for leveraging clinical imaging techniques to monitor the location, distribution, and activity of CAR-T cells. For instance, Kiru et al. devised a microfluidics-based iron oxide labeling system for CAR-T cells to enhance monitoring capabilities [166]. Through convective transfer and mechanical cell volume exchange, iron oxide NPs were incorporated into anti-B7-H3 CAR-T cells, obviating the need for transfection agents or genetic manipulation. This innovative approach allows for non-invasive detection of the labeled T cells using photoacoustic imaging (PAT), magnetic resonance imaging (MRI), and magnetic particle imaging (MPI), thereby providing a novel means to monitor CAR-T cells in solid tumors [166]. In comparison, one study employing iron oxide NPs achieved over 99% labeling efficiency by moderately activating T cells with IL-2 and CD3, without resorting to chemical or mechanical intervention [167]. By utilizing dextran-coated iron oxide NPs as an MRI contrast agent and the T2* relaxation measurement technique, researchers tracked CAR-T cells and monitored their distribution in intracranial gliomas. Experimental results demonstrated that the NPs could efficiently penetrate the subcellular cytoplasmic vesicles of T cells for labeling, without affecting the viability and effector functions of the T cells. Moreover, the distribution of T cells in the TME was found to be predictive of the anti-tumor treatment outcome [167]. However, the sole iron oxide NPs currently available for clinical application pose a risk of severe side effects in some patients, such as hypotension, hypersensitivity reactions, or allergic reactions [168]. Encouragingly, recent research has suggested that MegaPro-NPs may offer enhanced safety [169]. By labeling CAR-T cells with MegaPro-NPs and conducting in vitro cell iron uptake and activity function tests, along with in vivo MRI, MPI, and bioluminescence imaging (BLI) to monitor tumor growth, it has been confirmed that MegaPro-NPs can be effectively utilized for in vivo tracking of CAR-T cells in preclinical glioblastoma models [169]. Given that MegaPro-NPs have completed a Phase II clinical trial, the application in clinical cancer patients is expected shortly (NCT03407495). In another study, an in vitro cell-labeling method based on heparin and protamine was employed to attach ⁸⁹Zr-labeled dual-modal positron emission tomography (PET)/near-infrared fluorescent (NIRF) silica NPs to CAR-T cells [170]. This approach enabled non-invasive tracking of infused CAR-T cells via whole-body PET and NIRF imaging for over one week in tumor-bearing mice, providing valuable insights into their spatial distribution and fate. Nevertheless, a key limitation of this technique lies in the gradual dissociation of NPs from CAR-T cells, followed by their uptake by tumor cells, which may compromise the accuracy of cell tracking and fail to reflect the true distribution of viable CAR-T cells. Furthermore, the radiological safety profile of ⁸⁹Zr requires further evaluation before clinical translation [170].

Conventional CAR-T cells, lacking artificial control in vivo, can elicit rapid and durable clinical responses that may lead to severe, even fatal adverse events, particularly on-target/off-tumor toxicity effects and CRS. On-target/off-tumor toxicity effects stem from the expression of CAR-T cell targets in normal tissues, whereas CRS results from the intense interplay between CAR-T cells, tumor cells, and host immune cells, which triggers the swift activation and expansion of CAR-T cells, followed by massive cytokine release [39]. Consequently, beyond using NPs to monitor CAR-T cell distribution and activity, it is essential to employ them for regulating CAR-T cell function and quantity (Fig. 7). This will help avert excessive activation and proliferation, thereby curtailing side effects and safeguarding treatment safety. In response, researchers have developed “switch” mechanisms, such as a gelatinase Matrix metalloproteinase 2/9-based nano-modified switch, to modulate CAR-T cell activity, leveraging the high expression of gelatinase in the TME and its minimal presence in normal tissues [171]. By embedding a gelatinase-responsive peptide within methoxy poly(ethylene glycol)-polycaprolactone (PEG-PCL) copolymers, researchers obtained a PEG-PCL conjugate that was cleaved upon secretion of gelatinase into the TME. This cleavage caused NPs to deform and release drugs into the tumor tissue. Thus, NPs with switch molecules allowed precise control and continuous, stable activation of CAR-T cells, eliminating adverse reactions from sudden cytokine surges [171]. Similarly, by combining light-switchable CAR-T cells (LiCAR-Ts) with an upconversion nanoplate (UCNP) system, researchers utilized UCNPs to convert NIR light into blue light to activate LiCAR-T cells. The anti-tumor immune response was precisely triggered only when tumor antigens and light coexist, realizing a spatio-temporally regulated immune response [172]. This combined system has been validated in melanoma and lymphoma mouse tumor models, demonstrating enhanced CAR-T cell therapy safety and marked suppression of tumor growth, thereby providing a promising strategy for precision immunotherapy [172, 173]. Employing an analogous remote-control mechanism, another team utilized gold nanorods to convert NIR light into heat, enabling remote control of CAR-T cell activity through photothermal signals. Coupled with thermal-sensitive genetic switches like 7 H-YB, this system induced localized expression of IL-15 Superagonists at 40–42 °C, improving the precision, safety, and antitumor potency of the treatment [174] (Fig. 7A). Despite their therapeutic benefits, these technologies present distinct limitations. Nano-modified switch relies on endogenous matrix metalloproteinase (MMP) expression, which restricts its applicability in tumors with low MMP levels. LiCAR-T depends on external light activation, and its efficacy diminishes in deep tumors due to significant attenuation of light intensity. Although the photothermal approach, mediated by gold nanorods, can trigger transgene expression via heat shock response elements, its precision is comparatively lower due to thermal diffusion, which may cause off-target activation in surrounding normal tissues.

Fig. 7.

Nanoparticles designed to modulate the function and quantity of CAR-T cells. (A) UCNPs and gold nanorods can convert NIR light into blue light and heat, respectively, to activate CAR-T cells, enabling antitumor responses triggered only in the coexistence of tumor antigens and light, thus enhancing treatment safety. (B) NP-based cytokine regulation strategy. (C) NPs switch tumor cell death phenotype. UCNPs: Upconversion nanoplatelets; NIR light: near-infrared light

Besides the “switch” mechanism, NPs can improve CAR-T cell therapy safety by suppressing cytokine release. Zhou et al. used LNPs, which outperform cationic polymers, to encapsulate IL-6 shRNA and CAR gene plasmids, then modified them with CD3 antibody [92]. They created IL-6 knockdown CAR-T cells in vivo via intravenous injection, maintaining anti-tumor effects while reducing IL-6 release and CRS incidence, bypassing the complexity and cost of traditional in vitro methods. However, this strategy offers limited control over the broader cytokine network, as secretion of other proinflammatory cytokines such as IL-2 and TNF-α persists. Moreover, its clinical applicability may be constrained by variable transfection efficiency and potential off-target effects, warranting further investigation into the consistency and safety of cytokine modulation [92] (Fig. 7B). Another study functionalized melanoma-specific CAR-T cells with superparamagnetic iron oxide nanoparticles (SPIONs) [175]. SPIONs serve as contrast agents for X-ray tomography, MRI, micro-computed tomography, and magnetic guidance carriers [176]. Following SPION internalization, T cells acquired magnetic controllability, enabling magnetic field-guided precise tumor targeting while significantly reducing off-target distribution and associated damage. Importantly, targeting pyroptosis or DAMPs release is a well-recognized strategy to mitigate CAR-T-associated inflammatory complications, and the SPIONs-based modification strategy aligns with this approach: it preserves overall CAR-T cytotoxicity while suppressing secretion of cytokines (e.g., TNFα, IFNγ), but also shifts target cell death from inflammatory pyroptosis to non-inflammatory apoptosis. This occurs via a mechanism wherein reduced cytokine and granzyme B secretion inhibits the expression and cleavage of gasdermin proteins, thereby markedly lowering systemic inflammatory risks and ultimately enhancing both safety and targeting precision in melanoma treatment. Nevertheless, these findings are limited to in vitro models and require further validation [175] (Fig. 7C).

Recently, CAR⁺ EVs derived from CAR-T cells have emerged as a novel tactic to mitigate immunotherapy side effects, attributed to their natural nanostructure, low immunogenicity, and high biocompatibility. The biomolecular constituents and functions within them closely correlate with the source cells and can mediate intercellular crosstalk, participating in biological processes such as immunity and inflammation [177]. A previous report has shown that CAR-T cells could release EVs, primarily in the form of exosomes bearing CAR constructs, which expressed antibodies against EGFR and HER2, exhibiting potent anti-tumor effects with low toxicity in breast cancer [178]. Building on this, Haque and colleagues innovatively produced “off-the-shelf” targeted CAR exosomes using in vitro parental cells and CAR plasmids, simplifying the cumbersome traditional CAR-T cell preparation [179]. They demonstrated that CD19 CAR exosomes, instead of whole CD19 CAR-T cells, could induce cytotoxicity in CD19-positive leukemia B cells while sparing negative cells and curtail the CRS risk, thereby charting a novel course for cancer nano-immunotherapy [179]. Apart from their robust anti-tumor potential, CAR⁺ EVs serve as both CAR-T cell activation markers and predictors of therapy-related adverse reactions. In vitro assays revealed that following co-culture of CD19 CAR-T cells with CD19⁺ target cells, the release of CAR⁺ EVs spiked within 1 to 4 h, signifying a close link between CAR⁺ EV release and target cell binding, thereby representing an immediate response to CD19 CAR-T cell activation. In vivo, CAR⁺ EVs could be detected in plasma just 1 h after patients received CD19 CAR-T cells, and their plasma presence paralleled the disappearance of circulating CAR-T cells, further attesting to their capacity to reflect CAR-T cell activation in vivo [180]. Furthermore, the clinical trial highlighted plasma CAR⁺ EVs as early predictors of ICANS. The data revealed that ICANS patients exhibited significantly higher plasma CAR⁺ EV levels in the early phase (e.g., 1 h and day 1 post-infusion) compared to non-ICANS patients. Specifically, a plasma CAR⁺ EV concentration above 132.8 CAR⁺ EVs/µL at 1 h post-infusion predicted ICANS with 89.29% sensitivity and 74.58% specificity, while a level above 224.5 CAR⁺ EVs/µL on day 1 achieved 96.55% sensitivity and 80.36% specificity. These results suggested that CAR⁺ EVs could predict ICANS up to 4 days before symptom onset (median onset: day 5) and correlate with ICANS severity [180]. Taken together, the applications of NPs offer novel pathways to improve CAR-T cell therapy safety and propel the advancement of cancer immunotherapy.

Limitations and solution strategies: no garden is without weeds

Over the past three decades, nanotechnology has witnessed rapid advancements, leading to the creation of numerous NPs, such as the most representative LNPs and PNPs, which are specifically designed for the targeted delivery of anti-tumor drugs. To date, at least 15 cancer nanomedicines have been approved worldwide, concurrently with over 80 novel drugs under clinical trial evaluation [181]. Currently, FDA-approved or clinically investigated nanomedicines are primarily based on organic nanomaterials, such as NP liposomes. Representative examples include Vyxeos^® (liposomal daunorubicin and cytarabine for injection) and Marqibo^® (liposomal vincristine sulfate injection), approved for treating leukemia after successful clinical trials [182] (Table 5). Extensive preclinical research has explored NP-based strategies, showing promising potential to boost tumor-antagonizing immunity (Table 6). However, NP-based CAR-T cell immunotherapy remains in its early stages, with only one ongoing clinical trial employing 64Cu SPIONs and PET-MRI to study cilta-cel’s in vivo transport in extramedullary myeloma, while recruitment for this trial continues (NCT05666700). Before the large-scale clinical translation of NP-based CAR-T cell therapy, improvements should be considered in the following aspects.

Table 5.

FDA-approved nanomedicines for anti-cancer therapy

Drug name	Company	Nanoparticle material	Active ingredients	Indications	Date of first approval
Oncaspar	Sigma TAU	Polymer protein conjugate	Pegaspargase	Acute lymphoblastic leukemia	1994
Doxil	Baxter Healthcare Corp	Liposome	Doxorubicin hydrochloride	Ovarian cancer AIDS-related Kaposi’s sarcoma Multiple myeloma	1995
DaunoXome	Galen (UK)	Liposome	Daunorubicin citrate	Kaposi’s sarcoma	1996
DepoCyt	IPacira Pharms Inc	Liposome	Cytarabine	Lymphomatous meningitis	1999
Abraxane	Bristol-Myers	Nanoparticle-bound albumin	Paclitaxel	Lung cancer Metastatic breast cancer Metastatic pancreatic cancer	2005
Marqibo	Acrotech	Liposome	Vincristine sulfate	Acute lymphoblastic leukemia	2012
Onivyde	Ipsen	Liposome	Irinotecan hydrochloride	Metastatic pancreatic cancer	2015
Vyxeos	Celator Pharms	Liposome	Cytarabine Daunorubicin	Acute myeloid leukemia	2017
Fyarro	Aadi Sub	Albumin-bound nanoparticles	Sirolimus	Malignant perivascular epithelioid cell tumor	2021

Table 6.

Representative nanoparticle-based strategies for enhancing the CAR-T cell therapy

Strategies	Nanoparticles	Cargos	Administration	Functions	References
CAR-T cell engineering	Spleen SORT lipid nanoparticles	CD19-targeted CAR mRNA	Intravenous injection	Produce CAR-T cells in situ	[98]
	Ionizable lipid nanoparticles	CAR mRNA and siRNA targeting PD-1	Co-incubation	Program CAR-T cells and downregulate PD-1 expression	[148]
	Activating lipid nanoparticles	CD19-targeted CAR mRNA	Co-incubation	Enable one-step activation and transfection of primary human T cells	[85]
	Magnetic nanoparticles	NA	Co-incubation	Enable high-performance magnetic cell sorting	[87]
Promote aggregation at tumor sites	Human serum albumin nanoparticles	IL-12	Intravenous injection	Release IL-12, recruit and expand CAR-T cells through chemokines at tumor sites	[109]
	LY/ICG@HES-PCL nanoparticles	LY ICG	Intravenous injection	Reshape the TME via photothermal therapy to facilitate CAR-T cell therapy	[111]
	Immunomagnetic particles	NA	Intravenous injection	Decorate CAR-T cells to enable magnetic-acoustic actuation for precise tumor targeting and in situ immune activation	[113]
Modulate hostile TME	CaCO3 nanoparticles	Catalase	Intratumoral injection	Neutralize tumor acidity, attenuate tumor hypoxia, and suppress lactate production	[117]
	Au/polydopamine nanoparticles	Horseradish peroxidase	Intravenous injection	Inhibit the glycolytic metabolism of tumor cells, reducing lactate efflux, reprogramming tumor immunosuppression	[118]
	Polymeric micellular nanoparticles	TLR 7/8 agonist and PI3k delta inhibitor	Intravenous injection Intraperitoneal injection	Diminish the immunosuppressive TME	[119]
	Protein nanogels	IL-15 Sa	Intravenous injection	Boost intratumoral T cell proliferation and limit toxic reactions	[128]
Restrict tumor antigen heterogeneity	Human ferritin nanoparticles	Granzyme B	Intravenous injection	Enable robust binding to gastric cancer cells and demonstrate superior cytotoxicity	[135]
	Fusogenic nanoparticles	Exogenous antigens	Intravenous injection Intraperitoneal injection	Leverage exogenous antigens to perform in situ modification on solid tumors	[137]
Strengthen the monitoring of CAR-T cells	Iron oxide nanoparticles	NA	Mechanoporation Intravenous injection	Achieve PAT, MRI and MPI multimodal imaging	[166]
	MegaPro nanoparticles	NA	Co-incubation Intracardiac injections	Realize MRI, MPI, and BLI multi-modality imaging	[169]
	89Zr-labeled dual-modal PET/NIRF silica NPs	NA	Intravenous injection Intraperitoneal injection	Accomplish PET/NIRF dual-modal imaging	[170]
Enhance the control of CAR-T cells	Gelatinase-responsive nanoparticles	Switch molecules	Intraperitoneal injection	Allow precise control and continuous, stable activation of CAR-T cells	[171]
	Upconversion nanoplates	NA	Intravenous injection Intratumoral injection	Convert NIR light into blue light to activate light-switchable CAR-T cells	[172]
	Lipid nanoparticles	Plasmid containing IL-6 shRNA and CD19-CAR	Intravenous injection	Engineer T cells to generate CAR-T cells with IL-6 gene suppression	[92]
	Superparamagnetic iron oxide nanoparticles	NA	Co-incubation	Enable magnetic control of CAR-T cells, suppress inflammatory cytokine release, and switch tumor cell death mechanism	[175]

Production of NPs

NP production faces challenges in sterility control, quality control, surface modification, and scalable manufacturing, all of which are critical for successful biomedical applications. Sterility and quality control ensure safety and batch consistency, while surface modification enhances stability, targeting, and therapeutic efficacy. Additionally, scalable production, a key bottleneck in transitioning NPs from the laboratory to clinical applications, requires solutions to environmental safety and regulatory issues [183].

Sterility and quality control

Like other medical devices or drugs, NPs for in vivo use must remain sterile to prevent treatment failure and toxic reactions [184]. Various sterilization techniques, including non-ionizing radiation, ionizing radiation, autoclaving, sterile filtration, and chemical agents such as formaldehyde and ethylene oxide, have been developed [185–187]. However, just as the saying “every coin has two sides” goes, each sterilization method possesses both merits and demerits, potentially adversely affecting the properties of NPs. Therefore, it is essential to consider factors such as NP formulation, batch size, and terminal sterilization limits to select the most suitable sterilization method for maximizing benefits [187]. It has been shown that purification strategies can remove residual production components, thereby avoiding toxicity [188]. Besides sterility, strict quality control of NP size, composition, and shape ensures batch consistency. During production, analytical methods such as dynamic light scattering, microscopy, and zeta potential measurements play crucial roles in characterizing NPs, maintaining quality, and achieving reproducible manufacturing [189, 190].

Surface modification and stability

Extensive research has been conducted on NP surface modification to prolong their circulation time, enable precise drug delivery, and provide targeted payload release [191]. As the most representative approach, PEGylation involves coating NPs with PEG to prevent aggregation, opsonization, and phagocytosis, thereby extending blood circulation duration and promoting accumulation at pathological sites via the enhanced EPR effect [192, 193]. However, the immunogenicity of PEGylated NPs can trigger anti-PEG antibody generation, thereby accelerating blood clearance and reducing therapeutic efficacy [194, 195]. To alleviate this concern, various strategies have been explored, including designing cleavable PEG structures, reducing PEG chain number and length to minimize antigenicity, and incorporating tumor-targeting ligands to enhance therapeutic efficacy and targeting [193, 196]. Alternative modification strategies, such as employing hyaluronic acid, polyglycerol, surfactants, or small targeting peptides (e.g., cyclic peptide Arg-Gly-Asp) and blood proteins like albumin, have also been investigated [197–200].

Scaling up production

Scalable production remains a critical bottleneck in clinical translation. Compared to small-scale laboratory preparation, the industrial-scale production of NPs requires stringent control of sterility and quality to ensure safety and product consistency, making the process more complex and costly [183]. Beyond selecting suitable sterilization methods, employing advanced software and analytical tools for process optimization, along with clarifying NP physicochemical properties, understanding key component interactions, and exploring characteristic-performance relationships, enhances nanomedicine manufacturing reproducibility [201].

Large-scale NP production poses significant environmental safety challenges. When handling dry nanomaterials, their small size allows them to disperse like dust in the air, potentially being inhaled and deposited in the lungs, causing tissue damage [202]. Even worse, some NPs can penetrate the skin barrier, making dermal exposure a potential risk. Therefore, personal protective measures must be strengthened during nanomedicine manufacturing to minimize the potential harm to personal health [203]. To predict immune interactions, extensive toxicity testing should be performed [201]. Pleasingly, manufacturing NPs entirely in liquid environments significantly reduces potential environmental impact, making closed production systems an effective strategy. Meanwhile, to guide clinical research and accelerate clinical translation of nanomedicines, the DELIVER framework has been introduced, encompassing critical elements including design, experimental, manufacturing, preclinical, clinical, regulatory, and business considerations [204].

Cargo delivery and expression of CAR

NPs face multiple barriers in mediating CAR cargo delivery, including target distribution barriers, extracellular barriers, and intracellular barriers. To enhance T cell transfection efficiency, researchers have optimized NP design through various strategies. This section will highlight the barriers and corresponding solutions encountered at each major step from NP administration to final CAR expression in chronological order (Fig. 8).

Fig. 8.

Barriers of nanoparticle as a carrier for CAR cargo. (A) Administration routes for NPs. (B) Intravenous injection of NPs into the bloodstream faces multiple challenges: First, plasma proteins adsorb onto the NP surface to form a protein corona, which alters their surface properties and affects in vivo distribution and stability. Second, blood shear stress may disrupt NP structure, interfering with their distribution and delivery functions. Additionally, endovascular nucleases, proteases, and other enzymes can degrade nucleic acids encapsulated within NPs, while interactions between NPs and blood cells further reduce their stability and delivery efficiency. (C) Throughout this process, NPs may also be cleared from the systemic circulation via the MPS, hepatobiliary elimination (by feces), or renal excretion (by urine). (D) The sequential steps and potential barriers in cargo delivery to target cells: After reaching target cells, NPs bind to the cell membrane and undergo cellular uptake through different endocytic routes. Upon uptake by T cells, NPs are typically encapsulated within endosomes, making effective endosomal escape critical. Following endosomal escape, NPs release their cargo to generate CAR-T cells, where DNA cargo must be transported to the nucleus for transcription, while mRNA cargo can directly bind to ribosomes in the cytoplasm to initiate translation. MPS: mononuclear phagocyte system

Reaching target cells

NPs face multiple barriers before reaching target cells, with these challenges varying based on the administration method. They can be administered via various routes, including intravenous injection, local injection (e.g., heart, brain, eye, tumor, or utero), intranasal administration, inhalation, and oral delivery [205] (Fig. 8A). Each administration route requires crossing specific barriers to enter the bloodstream. For example, oral delivery must overcome intestinal barriers, including mucus, immune components, and microbiota, while non-vascular methods like subcutaneous injections require overcoming barriers like the vascular endothelium [206]. Taking the commonly used intravenous injection in clinical practice as an example, NPs encounter multiple barriers upon entering the bloodstream (Fig. 8B). Following contact with blood, NPs rapidly adsorb plasma proteins and biomolecules, forming a protein corona, which often leads to opsonization, marking them for recognition and degradation by the mononuclear phagocyte system (MPS), including Kupffer cells, splenic macrophages, and bone marrow monocytes [196, 207]. NPs not phagocytosed by MPS may be eliminated via hepatobiliary or renal systems based on size [208, 209] (Fig. 8C). Meanwhile, the protein corona alters the surface properties of NPs, influencing their distribution and stability [210]. Furthermore, shear stress in the bloodstream may disrupt NP structures, further interfering with their distribution and delivery [211]. NPs may also interact with blood cells, reducing stability and delivery efficiency [212]. Worse still, intravascular enzymes such as nucleases and proteases degrade the nucleic acids encapsulated by NPs [213]. Due to these barriers, only a limited number of NPs reach their targets and effectively deliver the cargo [214].

To address these challenges, researchers have developed various targeting strategies. Passive targeting leverages NPs’ size and surface properties, whereas active targeting involves surface modifications with ligands like small molecules, aptamers, antibodies, peptides, or cells [214, 215]. Endogenous targeting exploits NPs’ composition to bind specific plasma proteins, while stimuli-responsive targeting utilizes materials that change under specific conditions [97, 99, 216]. In practical applications, multiple strategies are often combined to achieve the best results.

Binding and endocytic uptake

After reaching target cells, NPs need to bind to the cell membrane and cross this barrier to achieve intracellular delivery. The interaction between NPs and cell surfaces is influenced by multiple factors, including their charge, shape, and size [217]. For instance, anionic NPs may exhibit poor cellular uptake due to electrostatic repulsion with the negatively charged cell membrane, while excessive positive charge on cationic NPs can cause membrane damage [217]. Following the binding, NPs enter cells primarily through endocytosis or direct membrane fusion [218]. Nonetheless, T cells are notoriously difficult to transfect, significantly hindering NP-cell surface binding and subsequent cellular internalization [219]. Ligands such as anti-CD8 antibodies, transferrin, and integrin can be added to enhance binding and uptake [220, 221]. Particle size also critically influences endocytic pathways, including clathrin-mediated, caveolin-mediated, macropinocytotic, and clathrin/caveolin-independent mechanisms, making size optimization crucial for T cell uptake [218].

Escaping from the endosomes

Upon uptake by T cells, NPs and their cargo are typically entrapped in endosomes, which gradually mature, acidify, and fuse with lysosomes, ultimately leading to cargo degradation [222]. Efficient endosomal escape is thus essential for successful delivery. Strategies include modifying NP composition (e.g., employing ionizable lipids, proton sponge molecules, or calcium ions) or designing specific structures like cuboplex nanostructures [223–226]. Further research into endosomal escape mechanisms and novel NP modifications is needed to advance clinical applications.

Expressing the cargo as CAR

Following endosomal escape, NPs need to release their cargo to generate CAR-T cells. For DNA cargo, NPs must deliver it to the nucleus for transcription, requiring T cells to be in an active mitotic state. Therefore, it is crucial to activate primary T cells, which can be achieved through aAPCs or CD3/CD28 magnetic beads [227]. In addition, incorporating nuclear localization signal (NLS) peptides or modifying DNA can enhance therapeutic transgene expression in T lymphocytes [90, 228]. For mRNA cargo, once released into the cytoplasm, it can directly bind to ribosomes to initiate translation, bypassing the need to enter the nucleus. However, mRNA offers short-term expression and requires frequent dosing to maintain efficacy, which may lead to cumulative toxicity and systemic inflammatory responses [140]. Potential solutions incorporate developing sustained-release delivery systems, optimizing mRNA sequences, investigating self-amplifying and trans-amplifying RNA systems, and exploring the application of circular RNA [229–233].

T cells have cell type-specific barriers and present additional gene delivery challenges. Autophagosome formation reduces polymer-mediated transfection efficiency, potentially addressed by manipulating autophagy via vacuolar protein sorting 34 and UNC-51-like kinase 1 complexes [234]. Slow endosomal acidification in T cells can be countered by pH-sensitive monomers with high pKa values to promote endosomal escape [235]. Overall, T cell transfection is a multi-step process, and the capability of NPs to overcome the aforementioned barriers is crucial for their successful delivery.

Efficacy and security in combined application

Combining NPs with CAR-T cell therapy confronts a gauntlet of obstacles that jeopardize anti-neoplastic effectiveness. Immunogenicity is a major concern; specifically, PEGylated LNPs can induce anti-PEG antibodies, leading to rapid clearance [236]. Moreover, immune cell membrane-coated NPs may also exhibit immunogenicity due to MHCs, affecting therapeutic outcomes [237]. Thus, a thorough investigation of the physicochemical properties of NPs and modification strategies, as well as their impact on pharmacokinetics and targeted delivery, is essential before NP clinical application. As previously mentioned, opsonization, MPS clearance, hepatobiliary, and renal system excretion will further reduce the concentration of NPs at target sites while increasing non-target tissue aggregation [207–209]. Some NPs could also suppress the functions of monocytes, macrophages, and NK cells, as well as pro-inflammatory cytokine production, leading to immune suppression [238]. Optimizing NP size, shape, charge, and surface modifications can enhance targeting, reduce immune suppression, and improve anti-tumor potency.

The potential hazards of NPs cannot be overlooked, as evidenced by the significant cytotoxicity demonstrated by magnetite and copper oxide NPs [239, 240]. LNPs may also induce adverse effects through multiple mechanisms, including IgM-mediated pseudoallergy, IgE-mediated allergic reactions, and autoimmune responses [241, 242]. In addition, the limited biodegradability of some NPs presents risks of long-term accumulation and chronic toxicity, but surface modifications and biodegradable materials can mitigate these risks [143, 243]. Moreover, many countries have enacted relevant laws and regulations to standardize the research and application of nanotechnology, thereby ensuring safety [244].

Regulatory frameworks

The clinical translation of NP-based strategy in CAR-T cell therapy also faces unique regulatory challenges: such combination products are subject to dual regulatory frameworks for both cell therapy products and NP-based therapy, but current guidelines lack integrated standards for such combined products. Understanding the existing regulatory landscape and addressing its inherent contradictions are crucial for advancing these therapies from laboratory research to clinical application.

CAR-T cell therapy

Globally, the United States, the European Union (EU), and Japan have established comprehensive regulatory frameworks for CAR-T cell therapy to facilitate innovation while ensuring patient safety. In the United States, the FDA classifies CAR-T products as human gene therapy products, overseen by the Center for Biologics Evaluation and Research (CBER) [245]. To accelerate approval, the FDA offers expedited pathways such as the Regenerative Medicine Advanced Therapy (RMAT) designation, which provides priority review and early regulatory communication for products addressing unmet medical needs without requiring demonstrated superiority over existing treatments [246]. The Breakthrough Therapy designation further shortens approval timelines for products showing substantial improvement in early clinical studies [247]. In the product development and manufacturing phases, manufacturers must submit detailed Investigational New Drug (IND) applications, including protocols for production and quality control. Manufacturing processes must strictly comply with current Good Manufacturing Practices (cGMP), with rigorous controls across the entire production process, from vector sourcing and T-cell handling to cryopreservation and transport, ensuring consistency and product stability [248].

In the EU, CAR-T cell therapy is classified as Advanced Therapy Medicinal Products (ATMPs), and their review system is jointly established by the Committee for Advanced Therapies (CAT) and the Committee for Medicinal Products for Human Use (CHMP) under the European Medicines Agency (EMA). The EU has established the Priority Medicine (PRIME) scheme, which targets innovative medicines for the treatment of serious diseases with unmet medical needs. Eligible CAR-T products can apply for this scheme and gain support, such as early communication with the EMA and accelerated evaluation [249]. Additionally, the conditional marketing authorization pathway allows for the approval of a product that can demonstrate significant benefits and controllable risks based on limited data, on the condition that further studies are conducted post-authorization to confirm its efficacy and safety [250]. Notably, ATMPs must submit marketing authorization applications through a centralized approval procedure, and the manufacturing process must comply with the GMP specifically developed by the EMA. Due to genetic modification, some member states also require environmental risk assessments to ensure public health and ecological safety [251].

Under the joint oversight of the Pharmaceuticals and Medical Devices Agency (PMDA) and the Ministry of Health, Labour and Welfare (MHLW), Japan classifies and regulates CAR-T products as regenerative medicine products. Japan has established the SAKIGAKE designation (Forerunner designation or pioneer drug), which is granted to innovative products first applied for in Japan and with potential to address unmet medical needs. Recipients enjoy preferences like priority review and shortened review cycles to accelerate market access [252]. Additionally, the Time-limited conditional approval pathway permits marketing approval based on non-confirmatory clinical trial data, though enterprises must supplement confirmatory data within a post-marketing specified period to retain marketing authorization [253]. During product development and clinical research, enterprises must comply with relevant regulations and guidelines. For example, the Act on the Safety of Regenerative Medicine implements risk-classified management for regenerative medicine technologies, with products of different risk levels subject to distinct review processes and requirements. Strict quality control standards also apply to manufacturing to ensure product quality and safety [251]. In summary, although the regulatory frameworks in the United States, EU, and Japan differ in classification and approval mechanisms, they share the core objective of ensuring safety, efficacy, and quality control. Through structured regulatory systems, expedited pathways, and lifecycle management, these regions support CAR-T innovation while safeguarding patient access to reliable therapies.

NP-based therapy

In the global regulatory landscape for NP-based therapy, the United States, Europe, and Japan have established tailored regulatory frameworks that align with their respective pharmaceutical oversight systems. These frameworks not only focus on the unique physicochemical properties and potential risks of nanomaterials but also provide guidance for their clinical translation [254]. In the United States, the FDA employs a multidimensional and adaptive regulatory strategy centered on risk assessment and product specificity. The agency has issued several key guidelines: the 2014 guidance Considering Whether an FDA-Regulated Product Involves the Application of Nanotechnology outlines criteria for nanomaterials, incorporating both size (1–100 nm) and size-dependent properties [255]; the 2018 guidance Liposome Drug Products specifies requirements for chemistry, manufacturing, controls (CMC), pharmacokinetics, and labeling [256]; and the 2022 guidance Drug Products, Including Biological Products, that Contain Nanomaterials refines the risk assessment framework, emphasizing evaluation of physicochemical properties, biological interactions, and long-term stability, along with the control of critical quality attributes (CQAs) [257]. The FDA also encourages early industry engagement and adopts “product-specific premarket reviews” to accommodate technological diversity [255].

In Europe, the EMA leads a regulatory approach focused on comparability studies and detailed control. Through scientific reflection papers, the EMA emphasizes the consistency and safety of nanomedicines and their reference products [258]. For intravenous liposomal formulations, the EMA requires demonstration of comparability in pharmaceutical quality, nonclinical data, and clinical performance [259]. For coated nanomedicines, it focuses on the stability of coatings, metabolic pathways, heterogeneity of surface coverage, process validation, and risk assessments of coating detachment [260]. For block copolymer micelles, early regulatory communication is advised to define CQAs, along with comprehensive documentation of synthesis, purification, and raw materials [261]. The European Nanomedicine Characterization Laboratory (EU-NCL) is advancing method standardization, though this effort remains under public consultation [262].

Japan’s regulatory system, overseen by the MHLW and the PMDA, follows a case-by-case methodology within existing pharmaceutical frameworks, supplemented by specialized provisions. Though no standalone nanomedicine regulation exists, the MHLW and PMDA issue joint reflection papers to elaborate on regulatory requirements: for example, block copolymer micelle products require data on synthesis, formation mechanisms, and stability [263]; a 2016 reflection paper on siRNA-loaded nanotherapeutics clearly outlined CMC standards, nonclinical toxicology, and clinical trial design [264]. Japan has established a Nanomedicine Initiative Working Group to discuss regulatory needs, but its regulatory progress lags behind that of the United States and Europe, with guidelines for emerging nanomedicine formulations still under development. Moreover, the approval process relies heavily on sponsors proactively submitting supporting data [254]. Overall, while the United States, European, and Japanese regulatory systems all prioritize safety and efficacy, they differ in emphasis: the United States system is risk-based and flexible, Europe emphasizes comparability and detailed control, and Japan adopts an incremental, case-based approach. These divergent frameworks necessitate compliance with multiple standards, underscoring the need for greater international harmonization to facilitate the global development of NP-based therapeutics.

NP-CAR-T combined therapy

When NPs are used in combination with CAR-T cell therapy, their integration of multiple technical dimensions, including nanomaterials, cell therapy, and gene editing, poses significant challenges to existing regulatory frameworks designed for single therapy. Due to the absence of specific guidelines for the approval and commercialization of such combined products, the regulatory landscape for NP-enhanced CAR-T cell therapy remains poorly defined. Multiple agencies may be involved in oversight. For example, in the United States, oversight may involve the FDA’s CBER, its Center for Drug Evaluation and Research (CDER), and the National Institutes of Health (NIH), which regulates gene therapies. However, the unclear demarcation of responsibilities among these agencies often leads to prolonged approval processes [265]. Unlike viral vectors with well-established regulatory pathways, non-viral vectors such as LNPs may reduce the risk of insertional mutagenesis but could necessitate the reconstruction of regulatory standards. For instance, NP-mediated CAR expression via mRNA is transient and may require repeated administrations, yet existing regulatory systems lack risk assessment indicators for such dosing strategies. Moreover, quality control indicators for viral vectors (e.g., viral titer, integration efficiency) are inapplicable to NPs, demanding the development of NP-specific parameters such as cargo encapsulation efficiency and transfection uniformity [266]. The synergistic interaction between NPs and CAR-T cells also introduces complex risk considerations: NPs carry the risk of accumulation, while CAR-T cells pose the risk of CRS; their combination may amplify these risks. However, current regulations do not mandate quantitative analysis of synergistic risks, making it difficult to balance therapeutic benefits against potential toxicities. Furthermore, as NP-based CAR-T cell therapy are still largely in the preclinical stage, the absence of long-term follow-up data fails to meet regulatory requirements for combination products [267].

To address these challenges, the following potential solutions can be adopted: First, promote interdisciplinary regulatory coordination by establishing a regulatory working group for NP-based cell therapy. This group should integrate expert resources from the fields of nanomaterials, cell therapy, and gene editing, and explore a “modular review” approach to evaluate the safety of NPs and CAR-T cells separately, thereby streamlining the approval process. Second, develop a dedicated regulatory framework. Drawing on the successful application of LNPs in severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccines, which accelerated the clinical translation of NP delivery systems, specialized guidelines for NP-based CAR-T cell therapy should be developed [268]. These guidelines should clarify key quality attributes of NPs (e.g., encapsulation efficiency, drug loading, stability) and safety evaluation indicators (e.g., maximum safe dose, organ accumulation threshold). Finally, optimize the risk-benefit assessment model by requiring the submission of quantitative data on the synergistic effects of NPs and CAR-T cells, for instance, by comparing the fold increase in efficacy with the incidence of additional toxicities. A dynamic risk monitoring system should also be established, including post-marketing continuous monitoring of NP biodistribution and long-term toxicity. Overcoming these regulatory hurdles requires collaboration among academia, industry, and agencies. Stakeholders should work within existing frameworks while advancing innovation through international standards, improved preclinical models, and transparent risk-benefit communication [269]. By establishing efficient and safety-conscious pathways, we can accelerate the clinical translation of these innovative combined therapies.

Artificial intelligence-powered nanoparticle-based CAR-T cell therapy

AI, a branch of computer science, aims to enable computers or computer-controlled machines to perform complex tasks requiring “human intelligence”. It is broadly categorized into three types: artificial general intelligence, artificial narrow intelligence, and artificial superintelligence. Key subfields include machine learning (ML), artificial neural networks, and deep learning [270]. In recent years, the rapid advancement of AI has ushered in a transformative era in medical and biological research [271–274]. The deep integration of AI and NPs is transforming CAR-T cell therapy from a standalone cellular treatment into a multi-technology-integrated precision medicine system. Through AI-driven target screening, product design optimization, delivery efficiency enhancement, efficacy monitoring, and toxicity prediction, this approach may establish a comprehensive “manufacturing optimization, efficacy maximization, and safety management” therapeutic framework.

Manufacturing optimization

To address the time-consuming and high complexity of CAR-T cell manufacturing, ML algorithms integrate multi-omics and functional data to accurately predict CQAs (e.g., CD4/CD8 ratios, memory cell subsets) for identifying optimal resource allocation and experimental conditions [275]. This approach also enables real-time quality prediction through early-phase parameter monitoring, facilitating scalable and standardized production of therapeutic cells [275]. Notably, large-scale collaborative initiatives are now underway to harness AI for advancing clinical applications of cell therapies. The European community-funded project on AI-driven, Decentralized Production for Advanced Therapies in the Hospital exemplifies such efforts [276].

Integrating ML and high-throughput NP synthesis, characterization, and functional testing enables data-driven candidate selection. Yamankurt et al. investigated Spherical nucleic acid candidates for cancer vaccines. They first identified 11 Spherical nucleic acid characteristics, generating 3,072 potential NP formulations, of which 960 were synthesized for high-throughput immunostimulation screening. Subsequently, they developed ML models using the experimental data to elucidate structure-immunoactivity relationships. This study demonstrates how combining ML with high-throughput NP screening can significantly reduce the number of designs requiring experimental testing, thereby shortening the research and development cycle and improving manufacturing efficiency [277]. While the work lays a methodological foundation for AI-aided NP design, it requires further improvement in model universality, multi-objective optimization, and production scale adaptation.

Efficacy maximization

A key objective in CAR-T cell research is identifying clinically safe and effective target antigens. Ideal targets are surface markers strongly and exclusively expressed on tumor cells but absent from vital non-malignant cells. AI tools facilitate this process by analyzing single-cell RNA and protein datasets to compare tumor and normal cell profiles.

In one study, an AI-powered deep learning model integrated multiple single-cell datasets and identified myeloid markers CD86 and colony-stimulating factor 1 receptor (CSF1R) as promising targets for AML [278]. Following tumor-target antigen identification, designing stable and high-affinity CAR proteins is crucial to trigger precise antitumor responses, with scFv structural optimization being particularly critical for enhancing therapeutic efficacy. Martarelli et al. employed an AI-powered computational platform to perform molecular docking and targeted molecular dynamics analysis on anti-CD30 monoclonal antibodies, enabling systematic screening of optimal scFv candidates for CAR construction. This study demonstrates the potential of computational modeling platforms to advance novel CAR engineering while reducing costs and resource consumption. However, the study relies on known antibody sequences, making it difficult to generate novel single-chain variable fragments [279]. Moreover, AI has been employed to predict optimal CAR signaling motifs by training on a library of 2,300 synthetic costimulatory domains, enabling optimized CAR-T cell phenotype, proliferation, and persistence. While the study features a rich library of motif combinations, the library construction depends on known signaling motifs and fails to cover motifs with unknown functions [280]. In the future, it is necessary to combine AI-generated design and multi-dimensional experimental validation to advance CAR design.

AI algorithms can also optimize NP delivery through multiple pathways, enhancing the efficacy of cancer therapy. Firstly, by employing ML and computational modeling, AI enables multidimensional optimization of NP characteristics, including size, surface charge, biointerface interactions, drug encapsulation efficiency, and controlled release kinetics [281]. Secondly, AI accelerates the development and optimization of novel NP formulations while enabling precise prediction of their in vivo behaviors. In a landmark study, drug-loaded ferritin NPs were combined with image-segmentation-based ML to analyze over 67,000 tumor vasculatures, successfully predicting NP permeability. This approach facilitated the design of genetically engineered protein NPs that significantly enhance transendothelial transport in low-permeability tumors, promoting NP accumulation within solid tumors [282]. Besides, AI algorithms facilitate the discovery of NPs targeting specific tissues and cells, thereby accelerating the realization of intelligent targeted delivery within the TME [283]. Some researchers developed a high-throughput DNA barcoding system, in which nucleic acid-barcoded LNPs were tracked via deep sequencing to quantify biodistribution. This approach precisely monitors LNP dynamics in both normal and tumor tissues, providing a powerful platform for screening highly tumor-specific nanocarriers [283]. By integrating nanosystems with sensors and imaging agents, combined with AI algorithms, this platform empowers real-time monitoring and feedback of tumor response, providing critical insights for treatment evaluation and adaptive strategy development. Such closed-loop feedback systems optimize drug delivery, dynamically adjust dosing, and even switch therapeutic modalities to maximize efficacy and overcome drug resistance [284].

Safety management

CRS and ICANS represent the primary side effects of CAR-T cell therapy. While studies on CRS and ICANS are limited by insufficient datasets and incomplete mechanistic understanding, AI has demonstrated promising potential in toxicity prediction. One group employed a ML approach with 16 models based on regression models and decision tree models to predict grade 4–5 CRS in anti-CD19 CAR-T-treated B-cell ALL patients. By measuring serum cytokine profiles, such as IFN-γ and sgp130 (soluble glycoprotein 130) before the onset of CRS, they achieved over 85% sensitivity and specificity [285]. Another study developed a ML-based logistic regression model to predict ICANS in 204 Axicabtagene Ciloleucel (axi-cel)-treated follicular lymphoma or diffuse large B-cell lymphoma patients, achieving 77% accuracy and 82% sensitivity [286]. Current predictive models, such as supervised multivariate models, generally fall short of the high-performance threshold (typically over 85%). However, expanding datasets and advanced AI models may enhance predictive accuracy, enabling proactive toxicity management and safer CAR-T cell therapy.

In addition to the aforementioned aspects, AI demonstrates significant potential to enhance NP-based CAR-T cell therapy through multifaceted approaches, particularly by leveraging large-scale data analytics to optimize patient selection, predict therapeutic responses, and improve early cancer screening and diagnosis [287–289]. It is noteworthy that current AI applications in cancer immunotherapy are still in their infancy, confronting substantial challenges including insufficient high-quality datasets, inherent complexities of biological systems and patient information, regulatory concerns, and the imperative for clinical validation through rigorous trials. Nevertheless, the development of integrated AI platforms incorporating multi-omics data such as genomics, proteomics, and imaging holds considerable promise to revolutionize the therapeutic workflow, from target antigen selection to clinical implementation, thereby ushering in a new era of precision-designed, patient-specific cellular immunotherapies with optimized efficacy and safety profiles [290, 291].

Conclusion and future perspectives

The combined application of CAR-T cell therapy and NPs offers an innovative solution for cancer treatment, with its core values manifested in the following three aspects: technological innovation, therapeutic enhancement, and safety control. Firstly, nanocarriers such as PNPs and LNPs enable efficient CAR delivery via mRNA transient expression or DNA site-specific integration, mitigating risks of insertional mutagenesis and cytotoxicity while reducing manufacturing timelines from weeks to days. Additionally, NPs enhance therapeutic efficacy by delivering chemokine modulators, degrading ECM barriers, and reprogramming immunosuppressive microenvironments, thereby improving CAR-T cell infiltration and functional persistence in solid tumors. Thirdly, safety control is ensured through iron oxide NPs enabling in vivo dynamic imaging of CAR-T cells, while “molecular switch” nanosystems precisely regulate CAR-T cell activity to minimize risks of CRS and neurotoxicity. Looking forward, the integration of AI is expected to substantially optimize this therapeutic strategy through AI-driven target screening, product design optimization, delivery efficiency enhancement, efficacy monitoring, and toxicity prediction, which will establish novel paradigms for developing more efficacious and safer immunotherapeutic approaches.

Despite significant strides in the application of NPs in CAR-T cell therapy over the past five years, its clinical translation still faces numerous challenges: uncertainty in NP mechanisms and physicochemical properties; scalability challenges in large-scale production; limitations in in vivo transduction targeting; concerns regarding long-term biocompatibility; potential immunogenic reactions; and translation challenges in experimental models. In this regard, future research should focus on:

1. Elucidating the pharmacokinetics, biodistribution, metabolism, clearance, and toxicity of NPs, along with their toxicity and immune system interaction mechanisms.

2. Simplifying the design of biomaterial systems, optimizing manufacturing processes, meeting regulatory requirements, and overcoming industrialization barriers.

3. Developing multifunctional integrated nanosystems by optimizing NP and nucleic acid designs or creating new materials to achieve efficient transfection, enhance therapeutic efficacy, and ensure safety.

4. Conducting in-depth studies in immunocompetent mice, immunodeficient and humanized mouse models, as well as non-human primates to promote clinical translation.

5. Integrating AI and high-throughput technologies to optimize nanomaterial design, or utilizing organoid and chip models to construct TME biomimetic systems for more accurate prediction of NPs’ behavior in humans [292, 293].

6. Further expanding research on targeting NK cells and other effector cell subsets, exploring their therapeutic potential beyond tumors, and advancing the field of adoptive cell therapy [294, 295].

With the deep integration of materials science, biomedicine, and clinical oncology, NP-based CAR-T cell therapy is poised to overcome existing hurdles in cancer therapy, ushering in a novel era of personalized and controllable “living drugs” in precision medicine. The academic community and the industry need to collaborate on innovation, improve international regulatory frameworks, accelerate the translation from laboratory to clinic, and provide safer and more effective treatment options for cancer patients.

Abbreviations

CAR

Chimeric antigen receptor

ECD

Extracellular domain

scFv

Single-chain variable fragment

MHC

Major histocompatibility complex

TMD

Transmembrane domain

HD

Hinge domain

TME

Tumor microenvironment

FasL

Fas ligand

FADD

Fas-associated death domain

GSDME

Gasdermin E

DAMPs

Damage-associated molecular patterns

CSD

Costimulatory domain

TRUCK

T-cells redirected for universal cytokine-mediated killing

FDA

Food and Drug Administration

CRS

Cytokine release syndrome

NP

Nanoparticle

aAPC

Artificial antigen-presenting cell

AI

Artificial intelligence

TGF-β

Transforming growth factor-β

Treg

Regulatory T cell

MDSC

Myeloid-derived suppressor cell

ICANS

Immune effector cell associated neurotoxicity syndrome

IEC-HS

Immune effector cell-associated hemophagocytic lymphohistiocytosis-like syndrome

IL

Interleukin

PEG

Polyethylene glycol

LNP

Lipid nanoparticle

NK

Natural killer

TCR

T cell receptor

aLNP

Activating lipid nanoparticle

MagSCAN

Magnetic amplification by sequentially connected antibodies and nanoparticle

TNF

Tumor necrosis factor

IFNγ

Interferon gamma

PNP

Polymer nanoparticle

shRNA

Short hairpin RNA

SORT

Selective organ targeting

FuNVs

Fusogenic nanovesicles

NIR

Near-infrared

HCC

Hepatocellular carcinoma

HER2

Human epidermal growth factor receptor 2

INS

IL-12 nano-stimulator

CCL

Chemokine C-C motif ligand

CXCL

CXC chemokine ligand

HES-PCL

Hydroxyethyl starch-polycaprolactone

LY

LY2157299

ICG

Indocyanine green

EPR

Enhanced permeability and retention

ECM

Extracellular matrix

CXCR

CXC chemokine receptor

M-CAR-T

CAR-T microrobot

EGFR

Epidermal growth factor receptor

CCaP CSs

Catalase-encapsulated CaCO3 nanoparticle-constructed colloidosomal microreactors

HRP

Horseradish peroxidase

PDA

Polydopamine

SDT

Synergistic sonodynamic therapy

ICD

Immunogenic cell death

PMNP

Polymeric micellular nanoparticle

TLR

Toll-like receptor

PD-L1

Programmed cell death ligand 1

STING

Interferon gene stimulator

PD-1

Programmed cell death protein 1

NG

Nanogel

IL-15Sa

IL-15 superagonist complex

TSA

Tumor-specific antigen

h-HFn

Human ferritin

Nb

Nanobodies

GrB

Granzyme B

F-AgNP

Fusogenic antigen loaded nanoparticle

EV

Extracellular vesicle

TRUE CAR-T

Target-redirected universal CAR-T

IS

Immunological synapse

LAG-3

Lymphocyte-activation gene 3

TIGIT

T cell immunoreceptor with Ig and ITIM domain

TIM-3

T cell immunoglobulin and mucin-domain containing 3

CTLA-4

Cytotoxic T-lymphocyte-associated protein 4

siRNA

Small interfering RNA

PRRs

Pattern recognition receptors

TIR

Toll/interleukin-1 receptor

ALL

Acute lymphoblastic leukemia

MyD88

Myeloid Differentiation Primary Response gene 88

NF-κB

Nuclear factor kappa B

MAPK

Mitogen-activated protein kinase

IRAK

IL-1 receptor-associated kinase

TRAF6

TNF receptor-associated factor 6

CLR

C-type lectin receptor

NKG2D

Natural killer group 2-member D

AML

Acute myeloid leukemia

MDS

Myelodysplastic syndromes

MM

Multiple myeloma

DC

Dendritic cell

ADCC

Antibody-dependent cell cytotoxicity

PI3Kδ

Phosphatidylinositol-3 kinase p110δ

PAT

Photoacoustic imaging

MRI

Magnetic resonance imaging

MPI

Magnetic particle imaging

BLI

Bioluminescence imaging

PET

Positron emission tomography

NIRF

Near infrared fluorescent

PCL

Polycaprolactone

LiCAR-T

Light-switchable CAR-T cell

UCNP

Upconversion nanoplate

SPION

Superparamagnetic iron oxide nanoparticle

MMP

Matrix metalloproteinase

MPS

Mononuclear phagocyte system

NLS

Nuclear localization signal

EU

European Union

CBER

Center for Biologics Evaluation and Research

RMAT

Regenerative Medicine Advanced Therapy

IND

Investigational New Drug

cGMP

Current Good Manufacturing Practices

ATMP

Advanced Therapy Medicinal Products

CAT

Committee for Advanced Therapies

CHMP

Committee for Medicinal Products for Human Use

EMA

European Medicines Agency

PRIME

Priority Medicine

PMDA

Pharmaceuticals and Medical Devices Agency

MHLW

Ministry of Health, Labour and Welfare

CMC

Chemistry, manufacturing, controls

CQAs

Critical quality attributes

EU-NCL

European Nanomedicine Characterization Laboratory

CDER

Center for Drug Evaluation and Research

NIH

National Institutes of Health

SARS-CoV-2

Severe acute respiratory syndrome coronavirus 2

ML

Machine learning

Authors’ contributions

LC, QL, and RX provided direction and guidance throughout the preparation of this manuscript. DS, ZKZ, RS, ZW, and PPZ wrote and edited the manuscript. ZKZ reviewed and made significant revisions to the manuscript. FP, HBL, GYC, HZQ, YDX, XZ, WJC, YJC, and SY collected and prepared the related papers. All authors read and approved the final manuscript.

Funding

This work was supported by the Natural Science Foundation of Hunan Province (2024JJ5494), Natural Science Foundation of Changsha City (Kq2403094), Hunan Provincial Development and Reform Commission of Innovative Research Program (2021-212-23), Hunan Innovative Province Construction Special Project (2021ZK4025), and the Hunan Provincial Department of Finance Gra (2024-31, 45, 2022 − 151, 2021 − 139 and 2020-83).

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.