Biomimetic lymph node-like scaffolds for optimized CAR-T cell expansion and potentiated antitumor efficacy

Abstract

Chimeric antigen receptor T cell (CAR-T) therapy has shown remarkable promise in treating hematological malignancies. However, the ex vivo expansion of CAR-T cells is time-consuming, potentially impairing CAR-T cell function. Physiologically, T cell activation and proliferation occur within the lymph node (LN) paracortex, a dynamic environment structured by a three-dimensional (3D) reticular network (RN) that promotes cell migration and mediator delivery. Mimicking this physiological niche offers a compelling strategy to improve CAR-T cell expansion. Inspired by the structure of the RN, we developed a biomimetic RN-like poriferous microsphere (PM) to establish a 3D culture platform optimized for both T cell and CAR-T cell proliferation. This engineered system not only significantly enhanced the proliferation rates of human T cells and CAR-T cells compared to conventional methods, but also preserved a higher proportion of central memory T cells (T_CM) and reduced the expression of exhaustion markers (PD-1, TIM-3, and LAG-3). Moreover, CAR-T cells expanded in PMs exhibited superior anti-tumor efficacy in both ex vivo and in vivo models, which correlated with the enrichment of pathways associated with robust T cell function at the RNA level. Overall, this biomimetic platform addresses critical limitations in human T/CAR-T cell expansion, preserving cell function and improving therapeutic outcomes.

---

1. Introduction

Chimeric antigen receptor T cell (CAR-T) technology has demonstrated substantial therapeutic potential against various tumors.^1,2 Encouragingly, the application of CAR-T leads to partial or complete remission in patients with hematological malignancies, including B-cell acute lymphoblastic leukemia³ and non-Hodgkin's lymphoma.⁴ The standard CAR-T treatment workflow involves the isolation of lymphocytes from patients’ peripheral blood, followed by genetic engineering and ex vivo expansion of CAR-T cells, and ultimately the reinfusion of these modified cells for the therapy.^5,6 However, the ex vivo expansion phase is the rate-limiting step of the overall therapeutic process.^7,8 This expansion can typically take 2–4 weeks to yield clinical-grade CAR-T cells for patients.⁹ Such a prolonged ex vivo culture duration may detrimentally alter or diminish the cytotoxic potency of CAR-T cells and reduce the presence of subpopulations of central memory T cells (T_CM) and stem cell-like memory T cells (T_SCM), which are vital for sustained therapeutic efficacy.^10,11 Therefore, there is an urgent need to develop an improved method to enhance the proliferation of T cells and CAR-T cells.

Tracing back to the physiological environment of T cells in vivo, T cells serve as vigilant sentinels of the immune system, distributed throughout various anatomical sites including the bone marrow, blood vessels, thymus, and lymph nodes (LNs).¹² T cell proliferation predominantly occurs within secondary lymphoid organs (SLOs), particularly the LNs.^13,14 Structurally, an LN comprises three main regions: the cortex, paracortex, and medulla.¹⁵ The paracortex compartment is permeated by a three-dimensional (3D) reticular network (RN) composed of fibroblastic reticular cells (FRCs) and reticular fibers, forming a fundamental scaffold for maintaining the LN structure.¹⁶ Upon antigen stimulation, T cells undergo rapid expansion within the paracortex, leading to an enlargement of the RN.¹⁷ The RN provides extensive surface area for T cell migration and enables the efficient delivery of soluble mediators in a relatively confined spatial environment, acting as “highways” for T cell trafficking and communication.^18,19 Based on this physiological insight, we hypothesize that mimicking the RN structure could offer a more optimal and biomimetic environment for enhancing T cell expansion.

Mimicking the RN structure that is stable ex vivo and allows for facile T cell infiltration and recovery is of great interest. Approaches involving hydrogels and extracellular matrix (ECM)-mimicking materials, such as Matrigel™ (Corning, USA), have been employed for culturing T cells in a 3D environment.^20,21 However, gel-like materials often suffer from poor structural stability, posing challenges for maintaining their integrity for a relatively long ex vivo culture duration.^22–24 Although a freeze-dried hydrogel may overcome the instability of a wet hydrogel, the recovery of viable cells from such unstable systems remains a significant challenge for ex vivo 3D culture.²⁵ Specifically, the mechanical stress²⁶ or enzymatic digestion²⁷ during the recovery procedure may potentially compromise T cell viability and function. The lyophilized LN also offers a highly biologically relevant scaffold for T cell culture,²⁸ however, its instability under prolonged ex vivo culture presents a significant practical challenge. Additionally, due to its biological origin, the possibility of large-scale production is limited.

In this study, we introduce hollow porous microspheres (PMs) as a biomimetic platform designed to replicate the structure of the RN found in LNs. We employed the biocompatible material polycaprolactone (PCL) to construct these microspheres and camphene to form porous network channels within the PMs. PCL-based PMs are stable under standard cell culture conditions,²⁹ ensuring structural integrity throughout the expansion process while also allowing for easy and gentle recovery of T cells. Furthermore, the use of PCL also makes this platform suitable for industrial-scale manufacturing. The resulting RN-like hollow PM system (Fig. 1) is specifically engineered to: (1) create an optimized spatial microenvironment that supports robust proliferation of T cells and CAR-T cells; (2) enhance the ex vivo expansion and cytotoxicity against targeted cells of CAR-T cells; and (3) promote the generation of CAR-T cells with durable functionality for in vivo therapeutic applications.

Fig. 1 Flowchart of the CAR-T cell cultivation process. This diagram illustrates the key steps in CAR-T cell culture. First, peripheral blood is collected and PBMCs are isolated. Next, PBMCs are activated with anti-CD3 and anti-CD28 antibodies to stimulate T cell activation and expansion. Subsequently, the T cells are genetically engineered to express CARs. These engineered CAR-T cells are then loaded into PMs and cultured for several days to facilitate their proliferation. Finally, the expanded CAR-T cells are harvested from the PMs for therapeutic applications.

2. Experimental section

2.1 Materials

Poly (ε-caprolactone) (PCL; Macklin, ∼80 000), polyvinyl alcohol 1788 (PVA; Macklin), chloroform (Chengdu Kelong Chemical Co., Ltd), camphene (CAS No. 79-92-5, Aladdin Industrial Corporation, Shanghai, China), 75% ethyl alcohol (Chengdu Kelong Chemical Co., Ltd), human Interleukin-2 (IL-2, C013; Novoprotein), the anti-human CD3 monoclonal antibody (Clone: OKT3, BioLegend), the anti-human CD28 monoclonal antibody (Clone: CD28.2, BioLegend), X-vivo™ 15 cell culture medium (04-418Q; Lonza Biosciences), 0.25% Trypsin-EDTA (25200056; Gibco), polybrene (H8761; Solarbio), Dulbecco's modified Eagle's medium (DMEM, 11995065; Gibco), RPMI-1640 (22400121; Gibco), Dulbecco's phosphate buffered saline (DPBS, 14190144; Gibco), fetal bovine serum (FBS, Z7185FBS-500; ZETA Life), HER2-His protein (Novoprotein), penicillin/streptomycin (P/S, SV30010; Hyclone), carboxyfluorescein diacetate succinimidyl ester (CFSE, BioLegend, USA). LDH assay kit (Beyotime Biotech Inc), D-luciferin (40902ES03; YEASEN), ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China), isoflurane (R510-22; RWD Life Science), and the Trizol reagent (15596-018; Invitrogen) were used. The anti-human CD3 FITC, CD4 APC, CD8 PE-Cy7, CD45RO PE-Cy7, CD62L PE, CD95 APC, PD-1 PE-Cy7, TIM-3 PE, LAG-3 Qdot 655, and anti-His APC antibodies and their isotype controls used for flow cytometry were purchased from BioLegend (USA). Cytometric bead array (CBA) flex set reagents for IL-2, TNF-α, IFN-γ, GM-CSF, and IL-6 were purchased from BD Biosciences (San Diego, CA, USA). Female NTG mice (6–8 weeks old) were purchased from SPF (BEIJING) Biotechnology Co., Ltd and housed under specific pathogen-free (SPF) conditions. Peripheral blood mononuclear cells (PBMCs) were purchased from Milecell Biotechnologies (Shanghai, China).

2.2 Cell lines and culture

Human ovarian cancer cell line SKOV3 (SKOV3), acute T cell leukemia cell line Jurkat (Jurkat), and human embryonic kidney 293T (HEK293T) cells were obtained from ATCC. Jurkat cells stably expressing enhanced green fluorescent protein (Jurkat-EGFP), SKOV3 cells stably expressing green fluorescent protein (SKOV3-GFP), and SKOV3-luciferase cell lines (SKOV3-luc, SKOV3 cells stably expressing firefly luciferase) were generated via lentiviral transduction. SKOV3, SKOV3-GFP, SKOV3-luc, and HEK293T cells were cultured in DMEM containing 10% FBS and 1% P/S. Jurkat-EGFP cells were maintained in RPMI-1640 supplemented with 10% FBS and 1% P/S (1640++).

2.3 Synthesis of different sized PMs

The porous PCL microspheres were synthesized by the oil-in-water emulsion method. Firstly, 6 g of camphene and 2 g of PCL were dissolved in 20 mL chloroform, and the mixture was gently stirred for 24 h to form the homogeneous oil phase. Secondly, the oil phase was added dropwise into 1% polyvinyl alcohol (PVA) aqueous solution. The mixture was then stirred at 200 rpm for 3 h, followed by stirring at 230 rpm for 20 h. The formed microspheres were sorted through a screen cloth. Subsequently, the PMs were freeze-dried for 72 h to ensure complete sublimation of the camphene from the microspheres. Finally, the PMs were disinfected with 75% ethanol and washed twice with phosphate buffered saline (PBS) before seeding of T cells.

2.4 Scanning electron microscopy (SEM) of human LN and PMs

Human LNs were washed with Dulbecco's phosphate buffered saline (DPBS) three times, followed by decellularization with 2% SDS at 37 °C and slightly oscillated for 2 days. Then, the decellularized human LNs were sliced into 4 mm sections and freeze-dried for 72 h. Subsequently, both human LN slices and PMs were gently adhered to the conducting resin, coated with ion (Carbon coater, Hitachi, Japan) and observed using a scanning electron microscope (SEM, Inspect, FEI). The pore size of human LNs and PMs, and the diameter of PMs were measured using ImageJ (version 1.8.0).

2.5 Brightfield and fluorescence microscopy imaging

A ZOE fluorescent cell imager (Bio-Rad) was used for brightfield and fluorescence imaging. Jurkat-EGFP cell samples were placed in the brightfield mode first to locate the cells, then switched to green fluorescence mode. After focusing on the cells, the required high-power objective lens was switched and the fine-tuning knob was used for precise focusing. Then, the brightfield and green fluorescent images were captured and saved.

2.6 Real-time monitoring of Jurkat-EGFP cells in PMs

1 × 10⁶ Jurkat-EGFP cells were dispersed in 100 μL 1640++, and seeded into 20 mg PMs for 15 min of incubation. Subsequently, 1 mL of the 1640++ was added to each group. Expansion and morphology of Jurkat-EGFP cells within PMs or in control plates were monitored for 50 h using a live-cell imaging system (Incucyte S3; Sartorius).

2.7 Confocal microscopy

1 × 10⁶ CFSE-labeled human T cells were seeded in PMs, incubated overnight at 37 °C and imaged using a Zeiss LSM 880 confocal microscope with Zeiss ZEN Black software (version 2.3).

2.8 T cell isolation and expansion

The isolated PBMCs were seeded into culture flasks pre-coated with anti-CD3 and anti-CD28 antibodies at 4 °C overnight. T cells were cultured in X-vivo™ 15 Cell Culture Medium supplemented with 400 IU mL⁻¹ human IL-2 (X-vivo™ 15 + 400 IU IL2) at 37 °C with 5% CO₂ for 48 h.

2.9 T cells cultured in in PMs and in a control plate

1 × 10⁶ human T cells were dispersed in 100 μL X-vivo™ 15 + 400 IU IL2 and then incubated into 20 mg PMs or added into a 6-well plate (control plate) for 15 min. Subsequently, an additional 3 mL of X-vivo™ 15 + 400 IU IL2 was added to each group. The cells were then cultured at 37 °C with 5% CO₂. The cells were collected by gently pipetting the PMs using a micropipette.

2.10 Cell counting and viability statistics

A 20 μL homogeneous cell suspension was pipetted into the sample chamber, and the cell number was determined using a Countstar Automated Cell Counter (Shanghai Ruiyu Biotechnology Co., Ltd). An equal volume of 0.4% trypan blue solution was added to the cell suspension, and viable (unstained) and dead (blue-stained) cells were counted. The cell viability percentage was calculated as follows:

Cell viability percentage = (total number of cells − number of dead cells)/total number of cells × 100%.

2.11 Construction of Her2-CAR

The structure of CAR consisted of anti-HER2 scFv, CD8 hinge, CD8 TM, 4-1BB costimulatory domain and CD3ζ. The HER2-CAR gene fragment was amplified by PCR with primers containing homologous arms for homologous recombination. The amplified fragment was then cloned into the linearized lentiviral vector pWPXLd using the ClonExpress II One Step Cloning Kit.

2.12 Lentivirus production

Lentiviral particles were generated by co-transfecting HEK-293T cells with the HER2-CAR expression plasmid, packaging plasmid psPAX2, and envelope plasmid pMD2.G using the calcium phosphate transfection system. Virus-containing supernatants were collected at 48 h and 72 h post-transfection, sequentially filtered through a 0.45 μm membrane, and concentrated by ultracentrifugation. The resulting viral pellet was resuspended in DPBS and stored at −80 °C until use.

2.13 Preparation of Her2-CAR-T cells

Isolated T cells (3 × 10⁵) were transduced with lentivirus at 30 multiplicity of infection (MOI) in 24-well plates containing 8 μg mL⁻¹ polybrene. Then the plates were centrifuged at room temperature and 1000 g for 1 h, and cells were incubated overnight. Fresh X-vivo™ 15 + 400 IU IL2 was added to replace the virus-containing medium, and cells were incubated for an additional 48 h.

2.14 Her2-CAR-T cells’ cytotoxicity assay

HER2-CAR-T cells were co-cultured with target cells at different E : T ratios (1 : 1, 2.5 : 1 and 5 : 1) in 96-well plates. Initially, 1 × 10⁴ tumor cells were seeded and co-cultured with 1 × 10⁴ (1 : 1), 2.5 × 10⁴ (2.5 : 1), and 5 × 10⁴ (5 : 1) HER2-CAR-T cells in triplicate. Released lactate dehydrogenase (LDH) in supernatants was quantified using a LDH (lactate dehydrogenase) Kit according to the manufacturer's instructions. The specific cytotoxicity was determined using the following equation:

[(Experimental release − effector spontaneous release − target spontaneous release)/(target maximal release − target spontaneous release)] × 100%.

The real-time killing of tumor cells by HER2-CAR-T cells expanded in PMs or in control plates was monitored using a live-cell imaging system.

2.15 Flow cytometry

0.5–1 × 10⁶ isolated cells were collected and centrifuged at 350 g for 3 min, then washed with DPBS. The cells were resuspended in 50 μL DPBS and stained with fluorescent-labeled antibodies, then incubated in the dark at 4 °C for 30 min. For HER2-CAR-T detection, cells were first stained with the HER2-His primary antibody, followed by an APC anti-His Tag secondary antibody to determine HER2-CAR expression. After staining, cells were washed twice and resuspended in 400 μL buffer in flow cytometry tubes for analysis. Acquired data were analyzed using NovoExpress (version 1.4.0) and Flowjo (version 10.8.1) software to gate populations and assess marker expression.

2.16 Cytokine production measurement

T cells were cultured in 96-well plates or within PMs for 24 h, and then the supernatants were collected. Separately, CAR-T cells were co-cultured with SKOV3 tumor cells at a 2.5 : 1 E : T ratio for 24 h, and then the culture supernatants was collected. Concentrations of IFN-γ, TNF-α, GM-CSF, IL-2 and IL-6 were measured using the CBA Human Cytokine Kit and analyzed by FCAP software.

2.17 RNA-Seq analysis

Total RNA was extracted from HER2-CAR-T and PM-HER2-CAR-T using the Trizol reagent. Subsequently, RNA quality and quantity were assessed, and libraries were prepared. Sequencing was performed on a NovaSeq 6000 platform. Finally, quantification and analysis of gene or transcript expression were performed.

2.18 Ovarian tumor model and animal experiments

Female NTG mice were intraperitoneally (i.p.) injected with 3 × 10⁵ SKOV3-luc cells. After 3 days, mice were randomized into four treatment groups: PBS, control non-transduced T cells (MOCK), HER2-CAR-T, and PM-HER2-CAR-T for different treatment. Then, 5 × 10⁶ MOCK T cells, 5 × 10⁶ HER2-CAR-T cells, or 5 × 10⁶ PM-HER2-CAR-T cells were injected intraperitoneally per mouse. An equal volume of PBS was injected as the control group. Weekly bioluminescence imaging (BLI) was performed to monitor tumor progression. For BLI, mice were anesthetized with 2% isoflurane and subsequently received an intraperitoneal injection of D-luciferin (150 mg kg⁻¹ in 200 μL PBS). After 10 min of D-luciferin administration, BLI signals were acquired using the AniView100 in vivo Imaging System (BioLight Biotechnology, Guangzhou, China). Regions of interest (ROIs) were manually defined over the abdominal cavity, and total photon flux (photons/second) was quantified using AniView software (version 1.00.0046). Data were analyzed using GraphPad Prism software (version 8.0), with significance set at p < 0.05. Results are presented as mean ± standard deviation (SD).

2.19 Statistical analysis

Statistical significance of measured differences in this study was assessed using Student's t-test for a comparison between two independent groups. For comparisons involving three or more groups, one-way analysis of variance (ANOVA) was applied, followed by appropriate post-hoc tests for multiple comparisons where applicable. All statistical analyses were performed using GraphPad Prism software (version 8.0). Results were presented as the mean ± SD from biological replicates. Each experiment was performed with a minimum of three independent replicates, as indicated in the figure legends. In all figures, statistical significance is indicated by the following symbols: *p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001.

2.20 Ethics committee approval

This study has adhered to the ethical principles of the World Medical Association Declaration of Helsinki and has been approved by the Biomedical Ethics Committee of West China Hospital, Sichuan University (Ethical Review Approval No. 2024 (2302)). All animal experiments were conducted in strict accordance with the protocols approved by the Biomedical Ethics Committee of West China Hospital, Sichuan University (Approval No. 20230504007).

3. Results

3.1 Structure of the human LN paracortex

LNs are the SLOs where most T cells proliferate following antigenic stimulation, especially in their paracortex area.³⁰ The structural framework of the LN paracortex is RN, formed by FRCs. The interstices of RN are occupied by lymphocytes, macrophages and antigen-presenting cells (APCs) like dendritic cells (DCs, Fig. 2A).³¹ The SEM imaging of decellularized LN paracortex further reveals the structure of the RN (Fig. 2B). Measurements indicate that the pore sizes of the RN range from 10 μm to 40 μm (Fig. 2C), which are suitably dimensioned to accommodate approximately 1 to 5 naive lymphocytes.¹⁵

Fig. 2 Construction and characterization of human LN-like PMs. (A) Schematic diagram of the paracortical structure of a human LN. The blue and red arrows indicate the inlet and outlet of the lymphatic circulation in a human LN, respectively. The area within the red dashed box shows the FRC structures formed within the paracortical region of the LN, as well as the high endothelial venules (HEVs), T cells, and DCs within the paracortical region. (B) SEM image of the paracortical region of a human LN. Bars represent 30 μm. (C) Pore size distribution of the paracortical region of a human LN, measured using ImageJ (version 1.8.0). (D) Process for constructing PMs using the oil-in-water method. (E) Representative SEM images of PMs with three different nominal diameters (approximately 200 μm, 400 μm, and 600 μm). Scale bars: 200 μm. (F) Diameters of PMs of different sizes, measured using ImageJ (version 1.8.0). (G) Pore sizes of PMs of different sizes, measured using ImageJ (version 1.8.0). (H) Representative images showing the initial loading of Jurkat-EGFP cells into PMs of three nominal diameters (200 μm, 400 μm, and 600 μm) after 15 min of incubation. All bars represent 100 μm. (I) Jurkat-EGFP cells were added to PMs of three sizes and allowed to proliferate for 50 h. All bars represent 500 μm. (J) The proportion of Jurkat-EGFP cells within the three-sizes of PMs was calculated after 50 h of incubation. The data are presented as mean ± SD (n ≥ 3 independent experiments). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3.2 Formation and characterization of LN-like PMs

Inspired by the structure of the RN in the LN paracortex and considering the appropriate size of the carrier, we synthesized PMs using PCL via an oil-in-water method (Fig. 2D and Fig. S1). To facilitate effective T cell interactions, it is important to form interconnected and open network structures within these PMs. Camphene was employed as a pore-forming material; it sublimes in a dendritic manner to create interconnected network channels.³² Through this method, we successfully fabricated PMs in three distinct size categories: small (S) (S, 100–200 μm), medium (M) (M, 400–500 μm), and large (L) (L, >600 μm), as shown in Fig. 2E and F. SEM analysis confirmed that the pores within the PMs were inter-connected (Fig. 2E). The average pore size of S and M-PMs was approximately 20 μm, while that of the L-PMs was around 40 μm (Fig. 2G), all of which were within the range of pore sizes observed in the human RN (Fig. 2C).

We next evaluated the cell loading efficiency and subsequent proliferation dynamics of Jurkat-EGFP cells within the S, M, and L-PMs. Jurkat-EGFP cells were co-incubated with S, M and L-PMs for 15 min. After incubation, Jurkat-EGFP cells could not enter S-PMs (Fig. 2H). For M-PMs, only a small portion of cells were loaded, and most Jurkat-EGFP cells were observed to be scattered around M-PMs. Most Jurkat-EGFP cells were observed to enter L-PMs. After 50 h of incubation, Jurkat-EGFP cells proliferated exclusively within L-PMs (Fig. 2I and SI Videos S1–S3). Although Jurkat-EGFP cells could partly enter M-PMs, the number of cells within M-PMs remained unchanged after 50 h of observation (Fig. 2J).

As human T cells were unable to be loaded into S-PMs, we focused on using M-PMs and L-PMs to culture CD3/CD28-activated human T cells. A total of 1 × 10⁶ CD3/CD28-activated human T cells were seeded into M-PMs and L-PMs and allowed to proliferate for 10 days. Light microscopy of human T cells cultured in M-PMs and L-PMs showed the gradual formation of high-density clusters between the cells and the PMs, suggesting T cell proliferation within both PM types (Fig. 3A). However, quantitative analysis demonstrated that the number of proliferating human T cells in the L-PMs was significantly greater than in both M-PMs and the control plate (standard cell culture plate, Fig. 3B). Furthermore, confocal microscopy of CFSE-labeled T cells confirmed that L-PMs could accommodate a greater number of human T cells than M-PMs (Fig. 3C and D). After 10 days, the proportion of T_CM on the control plate, in L-PMs and in M-PMs, was 43.47%, 69.87% and 62.25%, respectively (Fig. 3E and F). The proportion of effector memory T cells (T_EM) was notably reduced in the L-PMs (27.06%) and M-PMs (34.42%) compared to the control plate (51.85%), indicating that L-PMs and M-PMs promoted a significantly higher T_CM and lower T_EM phenotype.

Fig. 3 Proliferation and phenotypic characteristics of human T cells in M-PM and L-PM. (A) Representative light microscopy images of human T cells proliferating in the control plate, M-PM, and L-PM after 10 days of culture. All bars represent 100 μm. (B) Quantification of the number of proliferating human T cells in M-PM and L-PM plates after 10 days compared to control plates. Data are presented as mean ± SD (n ≥ 3 independent experiments). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (C) and (D) Representative confocal microscopy images of CFSE-stained human T cells in L-PM (C) and M-PM (D). (E) Flow cytometric analysis of CD62L and CD45RO expression in human T cells cultured in PMs for 10 days. (F) Percentages of T cell subsets (T_CM, T_EM, effector T cells (T_EF) and naive T cells (T_N)) calculated from the flow cytometry analysis of CD62L and CD45RO expression on human T cells in PMs (n = 3). Data are presented as mean ± SD (n ≥ 3 independent experiments). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3.3 Co-culture and expansion of human T cells in PMs

Given the superior performance of L-PMs in supporting T cell expansion, they were selected for subsequent human T cell culture. Human T cells cultured within L-PMs exhibited a significantly larger average cell diameter (Fig. 4A) and higher viability (Fig. 4B) compared to those in control plates at different initial densities. After 10 days of culture, the ratio of CD4⁺ to CD8⁺ T cells remained comparable between the L-PMs and control groups (Fig. 4C), indicating no significant shift in the T cell subset distribution. Notably, after 4 days of culture, the proportion of T_CM cells in L-PMs reached 88.84%, significantly exceeding the 80.16% observed in the control plate (Fig. 4D and E). However, after 10 days of culture, the T_CM proportion in the control plate fell to 19.76%, whereas it remained at 55.99% in L-PMs. The maintenance of T_CM cells in L-PMs suggests a potential for enhanced cellular cytotoxicity. The proportions of both PD-1⁺ and TIM-3⁺ positive human T cells, and LAG-3⁺ human T cells in L-PMs were also significantly lower than those in control plates (Fig. 4F and G). The spontaneous secretion of inflammatory cytokines by human T cells in L-PMs was also increased compared to that in control plates after 10 days of culture. The secretion of secreted interferon-γ (IFN-γ) and granulocyte-macrophage colony-stimulating factor (GM-CSF) by T cells in L-PMs was significantly higher (Fig. 4H), while interleukin-6 (IL-6) levels remained similar across both groups (Fig. 4I), suggesting enhanced cytotoxic potential with a potentially lower risk of cytokine release syndrome (CRS).

Fig. 4 Proliferation, phenotype, and cytokine secretion of human T cells cultured in L-PM. (A) Diameter of human T cells after 10 days of culture in control plates or L-PM with an initial cell count of 3 × 10⁴, 5 × 10⁴, and 1 × 10⁵ (n = 3). (B) Viability of human T cells after 10 days of culture in control plates and 600 μm L-PM (with an initial cell count of 3 × 10⁴, 5 × 10⁴, and 1 × 10⁵, respectively) (n = 3). (C) Flow cytometric analysis of CD4⁺ and CD8⁺ T cell subsets after 10 days of culture in control plates and L-PM (n = 3). Data represent percentages of CD4⁺ and CD8⁺ cells (n = 3). (D) Flow cytometric analysis of CD62L and CD45RO expression in human T cells after 10 days of culture in control plates and L-PM. (E) Percentages of T cell subsets (T_N, T_CM, T_EM, and T_EF) calculated from the flow cytometric data in Fig. 4D. (F) Expression levels of PD-1 and TIM-3 depletion markers on T cells cultured in control plates and L-PM for 10 days, and calculation of the positive rate (n = 3). (G) Expression levels of LAG-3 depletion markers on T cells cultured in control plates and L-PM for 10 days, and calculation of the positive rate of LAG-3⁺ (n = 3). (H) Secreted concentrations of IFN-γ and GM-CSF in the culture supernatant after 10 days of culture (n = 3). (I) Secreted concentration of IL-6 in the culture supernatant after 10 days of culture (n = 3). Data in Fig. 4 are expressed as mean ± SD (n ≥ 3 independent experiments). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3.4 Co-culture and expansion of CAR-T cells in PMs

Subsequently, human epidermal growth factor receptor 2 (HER2)-targeted CAR-T cells (HER2-CAR-T) were co-cultured either within L-PMs or on conventional control plates for six days. The CAR expression levels in L-PM-cultured CAR-T cells (PM-HER2-CAR-T), at 57.10% (donor 1) and 37.1% (donor 2), were significantly higher than those in control plates (45.40% for donor 1 and 29.00% for donor 2) (Fig. 5A and B). The T_CM proportion of PM-HER2-CAR-T cells (87.56%, 44.67%) was significantly higher than that of HER2-CAR-T cells (70.86%, 28.85%) in donor 1 and donor 2 (Fig. 5C and D). The higher CAR positive rate and T_CM ratio indicate higher cytotoxicity ability of PM-HER2-CAR-T. In a cytotoxicity assay using SKOV3-GFP cells, which express high levels of HER2 (Fig. S2), PM-HER2-CAR-T cells demonstrated significantly stronger cytolytic activity across all tested effector-to-target (E : T) ratios (1 : 1, 2.5 : 1, and 5 : 1) (Fig. 5E and F and Fig. S3). The secretion of cytokines, including IL-2, IFN-γ, Tumor Necrosis Factor-α (TNF-α), and GM-CSF was significantly elevated in PM-HER2-CAR-T cells (Fig. 5G).

Fig. 5 HER2-CAR-T cells proliferated in L-PM. (A) and (B) HER2-CAR positivity rates after 6 days of culture of CAR-T cells from donors 1 and 2 in control plates or L-PM. (C) and (D) Flow cytometric analysis of CD45RO and CD62L expression in HER2-CAR-T cells from donors 1 and 2 after 6 days of culture in control plates or L-PM. The percentages of T cell subsets (T_N, T_CM, T_EM, and T_EF) were calculated based on flow cytometric data (n = 3). (E) Specific lysis of SKOV3 tuomor cells by HER2-CAR-T cells cultured in control plates or L-PM after 24 h of coculture at various effector cell to target cell (E : T) ratios was quantified by LDH release assay (n = 3). (F) Real-time monitoring of the killing effect of HER2-CAR-T cells on SKOV3 tumor cells at E : T ratios of 1 : 1, 2.5 : 1, and 5 : 1. Remaining tumor cell counts were monitored and analyzed using the Incucyte Intravital Imaging System (n = 3). (G) Secretion of IL-2, IFN-γ, GM-CSF, and TNF-α by CAR-T cells and SKOV3 target cells after 24 h of coculture (n = 3). Data in Fig. 5 are presented as mean ± SD (n ≥ 3 independent experiments). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3.5 Differences between Her2-CAR-T cells and PM-Her2-CAR-T cells at the RNA level

To explore the differences between PM-HER2-CAR-T cells and HER2-CAR-T cells, we conducted bulk RNA sequencing (RNA-Seq), which revealed a distinct set of differentially expressed genes (DEGs) (Fig. 6A and B). Gene ontology (GO) enrichment analysis of these DEGs demonstrated significant upregulation in biological processes such as the inflammatory response and Wnt signaling pathway; cellular components including cell–cell junction, receptor complex, membrane raft and plasma membranes; and molecular functions such as cytokine activity and GTPase activity (Fig. 6C). The upregulation of the inflammatory response is directly associated with enhanced T cell function. In the cell–cell junction pathway, enriched genes such as CDH5, which is related to cell adhesiveness,³³ indicate more intimate and efficient intercellular communication among PM-HER2-CAR-T cells.³⁴ The increased expression of genes associated with receptor complexes,³⁵ like interleukin-13 receptor alpha 1 (IL13RA1) that responds to IL-13 signaling,³⁶ indicates a heightened capacity for receptor–ligand interactions. The membrane raft³⁷ and the plasma membrane³⁸ related upregulated gene ATP-binding cassette transporter A1 (ABCA1) play a key role in preserving cholesterol balance of T cell membranes, thereby suppressing apoptosis and senescence.³⁹ Genes associated with cytokine activity in the GO analysis may enhance T cell secretion of functional cytokines, such as IL3 and IL17C, thereby positively correlating with T cell-mediated immune functions,⁴⁰ which is consistent with our Gene Set Enrichment Analysis results (Fig. 6D). Furthermore, the upregulation of the GTPase activity pathway suggests enhanced T cell capabilities in activation, signaling, migration and homing.⁴¹

Fig. 6 RNA-Seq analysis of HER2-CAR-T and PM-HER2-CAR-T cells. (A) Volcano plot of DEGs identified by RNA-Seq between HER2-CAR-T cells and PM-HER2-CAR-T cells. Genes showing significant differences (P < 0.05) and absolute log₂ fold change > 1 (|log₂FC| > 1) were selected for downstream analysis. (B) Heatmap of significantly differentially expressed genes (P < 0.05, |log₂FC| > 1) between HER2-CAR-T cells and PM-HER₂-CAR-T cells. (C) GO enrichment analysis of genes significantly upregulated in PM-HER2-CAR-T cells compared with HER2-CAR-T cells, with enrichment P value < 0.05. (D) GSEA analysis of cytokine activity of PM-HER2-CAR-T cells compared with HER2-CAR-T cells. (E) KEGG enrichment analysis of genes significantly upregulated in PM-HER2-CAR-T cells compared with HER2-CAR-T cells, with P < 0.05. (F) GSEA analysis of the JAK-STAT signaling pathway in PM-HER2-CAR-T cells compared with HER2-CAR-T cells. (G) GSEA analysis of cytokine-cytokine receptor interaction pathways compared with HER2-CAR-T cells.

KEGG pathway enrichment analysis further revealed significant upregulation of genes associated with the Wnt signaling pathway, the JAK-STAT signaling pathway, and cytokine–cytokine receptor interactions (Fig. 6E). The concurrent enrichment of the Wnt signaling pathway in both GO and KEGG analyses underscores its central role in regulating T cell development, proliferation, and differentiation, contributing to the enhanced functional state of PM-HER2-CAR-T cells.^42,43 The upregulation of the JAK-STAT signaling pathway represents a pivotal signaling cascade in T cells, involved in transmitting cytokine signals (Fig. 6F).⁴⁴ Genes related to the cytokine–cytokine receptor interaction pathway (e.g., IL3, IL17C, etc.) supports a positive correlation with T cell functions in immune regulation (Fig. 6G).^45,46

3.6 In vivo anti-tumor activity of CAR-T cells in PMs

Given the enhanced ex vivo tumor cytotoxicity of HER2-CAR-T cells expanded in L-PMs, we next assessed their therapeutic efficacy in vivo using a xenograft SKOV3 tumor model (Fig. 7A). Mice treated with PBS or MOCK T cells exhibited rapid tumor growth, as indicated by progressively increasing bioluminescence intensity over time (Fig. 7B). In contrast, tumor progression in mice receiving either PM-HER2-CAR-T cells or HER2-CAR-T cells was significantly suppressed within the first 10 days post-treatment. However, all mice in the HER2-CAR-T cell treated group relapsed at around day 17 post-treatment. Notably, only 1 of 3 mice in the PM-HER2-CAR-T cell treated group showed signs of relapse by day 31. The bioluminescence intensity of the PM-HER2-CAR-T cell treated group remained significantly lower than in the HER2-CAR-T cell treated group (Fig. 7C). Consistently, mice treated with PM-HER2-CAR-T cells exhibited a markedly prolonged survival period, exceeding 70 days, compared to other treatment groups (Fig. 7D). Throughout the treatment period, the body weight of mice across the MOCK, HER2-CAR-T, and PM-HER2-CAR-T groups remained stable, showing only a slight increase (Fig. 7E).

Fig. 7 In vivo antitumor efficacy of HER2-CAR-T cells in an intraperitoneal ovarian cancer model. (A) Schematic diagram of the experimental design for intraperitoneal human ovarian tumor treatment in NTG mice. The tumor model was established by intraperitoneal injection of 3 × 10⁵ luciferase-encoding SKOV3 cells (SKOV3-luc cells) per mouse. Three days later, each mouse received PBS (PBS), 5 × 10⁶ MOCK T cells, 5 × 10⁶ HER2-CAR-T cells, or 5 × 10⁶ PM-HER2-CAR-T cells (3 mice per group). (B) Representative in vivo IVIS images of NTG mice bearing ovarian tumors treated as indicated at representative time points after treatment. (C) Quantification of total bioluminescence intensity over time for each treatment group. Data are presented as mean ± SD (n ≥ 3 independent experiments). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (D) Kaplan–Meier survival curves of tumor-bearing NTG mice after different treatments. Statistical significance was assessed by the log-rank test. (E) Body weight changes of NTG mice during treatment. Animal experiments were approved by the Biomedical Ethics Committee of West China Hospital, Sichuan University (Approval No. 20230504007).

4. Discussion

The development of effective adoptive cell therapies, particularly the CAR-T cell therapy, critically depends on optimizing both the expansion and functional potency of T cells during ex vivo culture. Although T cells can form aggregates ranging from 10² μm to 10³ μm in conventional two-dimensional (2D) culture systems, these large clusters may suffer from hypoxia and nutrient deprivation, potentially leading to a necrotic core and impaired intercellular communication.^47,48 Standard 2D in vitro systems often fail to replicate the complex physiological microenvironment of lymphoid tissues, where T cells naturally reside within a dynamic 3D FRC network that facilitates constant interaction with various immune cells and APCs.^49,50 Studies have shown that after critical FRC network loss, immune cell recruitment, migration and dendritic cell-mediated activation of antiviral CD8⁺ T cells are seriously impaired.^50,51 3D environment is of great importance for T cell proliferation. Our study addresses this challenge by demonstrating the successful design and application of PMs to provide a more physiologically relevant 3D culture environment for T cells and CAR-T cells, leading to enhanced anti-tumor efficacy in vitro and in vivo.

Previous efforts to expand CAR-T cells ex vivo within 3D microenvironments have employed diverse strategies, including the use of synthetic artificial APCs like commercial microbeads (e.g., Dynabeads®, Thermo Fisher Scientific, USA) that utilize anti-CD3 and anti-CD28 antibodies to polyclonally activate T cells through engagement of the TCR complex and CD28 co-stimulatory receptor. Other approaches have introduced advanced biomaterials such as supported lipid bilayers (SLBs) formed on mesoporous silica micro-rods (MSRs)⁵² and CCL21-loaded 3D hydrogels,²⁴ designed to promote T cell activation and expansion by delivering specific biochemical signals. While these materials primarily focus on providing biochemical cues for T cell activation, our work highlights the critical role of the physical and structural properties of the culture substrate, which mimics the ECM of lymphoid organs, representing a new discovery.⁵³

Concurrently, significant advances have been made in the development of implantable T-cell scaffolds for in vivo applications, such as the subcutaneously injectable, biodegradable T-cell enhancing scaffold (TES), which reactivates pre-administered CAR-T cells through sustained co-stimulatory ligand presentation and controlled IL-2 release, thereby enhancing antitumor efficacy.⁵⁴ Another notable example involves porous poly(lactic-co-glycolic acid) (PLGA) scaffolds fabricated via microfluidic technology and functionalized with T-cell activating signals.⁵⁵ The stability of T cell scaffolds for in vitro use is crucial for maintaining a consistent structural framework during long-term in vitro T cell culture, and is helpful for T cell separation and collection from the scaffolds.

To address these limitations, we developed biocompatible PMs using PCL as the primary material. PCL's inherent strength ensures structural robustness of the 3D scaffold, maintaining its integrity at physiological temperatures (37 °C),⁵⁶ which is essential for long-term T cell culture. The physical dimensions of the carrier are critical; nanospheres are too small to support effective T cell entry and proliferation, while excessively large carriers (1–3 cm) may hinder T cell–T cell interactions and complicate cell retrieval. Our findings emphasize the importance of both surface chemical properties and microsphere morphology in facilitating efficient T cell culture and loading. Based on Live cell imaging results, we observed that T cells could not effectively enter 200 μm and 400 μm PMs. This finding underscores the significance of appropriately sized channels within the PMs. To address this, we synthesized channeled PCL microspheres with pore sizes exceeding 600 μm, which enabled efficient T cell loading and enhanced functionality. These results indicate that macroscopic features, such as the shape and overall size of these LN–mimicking microspheres, significantly influence T cell behavior and functional outcomes.

Our PM design incorporates interconnected channels formed through the sublimation of solid camphene, in contrast to conventional materials that typically generate isolated, independent pores. This interconnected architecture enhances internal communication among T cells within the PMs and effectively addresses a common limitation in porous materials, buoyancy. Initially, PMs tended to float on the culture medium surface due to air trapped within their pores. However, treatment with 75% ethanol effectively reduced the surface tension, allowing the microspheres to completely sink into the culture medium and fully integrate with the cell culture conditions (Fig. S4). This surface modification likely increased wettability as well, promoting greater cell entry and interaction with the porous framework. Additionally, the high surface area provided by the porous PCL microspheres offers a more efficient platform for T cell loading and culture compared to non-porous, dense structures.

Beyond in vitro expansion, the ultimate aim of CAR-T cell therapy is to achieve sustained anti-tumor efficacy in vivo. While CAR-T cells are infused back into patients once sufficient numbers are achieved, a recognized limitation of currently used CAR-T cells is their potential for a gradual decline in efficiency and lack of enduring vitality in vivo. The proportion of T_CM cells is crucial for long-term expansion and persistence of CAR-T cells.⁵⁷ Our findings suggest that culturing T cells within PMs may promote the maintenance of stem cell-like phenotypes, potentially influenced by the specific mechanical cues provided by the PM structure. This preservation of less differentiated subsets may underlie the superior persistence and anti-tumor efficacy of PM-expanded CAR-T cells in vivo. The elevated production of cytokines observed in PM-expanded CAR-T cells further suggests a more potent cytotoxic profile against cancer cells, both in vitro and in vivo.

5. Conclusions

In conclusion, our study presents a novel strategy for the ex vivo expansion of HER2-CAR-T cells using PCL-based PMs, offering a 3D biomimetic culture system that enhances T cell loading efficiency, functional performance, and in vivo anti-tumor activity and persistence. Moving forward, future investigations should aim to unravel the underlying mechanisms through which the PM architecture modulates T cell differentiation and longevity, while also refining the microsphere design for clinical translation.

Author contributions

Huajin Zhang: conceptualization, data curation, writing – original draft, visualization, and investigation; Fujun Liu: software, validation, and writing – review and editing; Junyilang Zhao: methodology; Yong Wang: investigation; Yuge Shen: visualization and data curation; Qiqi Li, Hui Luo, Rong Li, Fan Zhu, and Shuo Xie: investigation and data curation; Yinhao Wei, Xupeng Gou and Danling Hu: investigation and validation. Yu Chen: formal analysis; Zhengji Li: formal analysis and software; and Hanshuo Yang: conceptualization, writing – review and editing, funding, acquisition, and supervision. All authors have made significant contributions to the research and have approved the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data supporting this paper, including the IR characterization of PCL and PM, Her2 expression of SKOV3 cells, images of HER2-CAR-T and PM-HER2-CAR-T cytotoxicity against SKOV3-GFP cells, pretreatment of PM, and the movies of real-time monitor of Jurkat-EGFP cells incubated in S, M, and L-PMs have been incorporated in SI. See DOI: https://doi.org/10.1039/d5tb01594d

Acknowledgements

This work was supported by the National Natural Science Foundation of China (82203030) and Key Program of Hainan Province (ZDYF2025LCLH007). Special thanks to Tianmu Wu from Chengdu Wuhou Baichuan Stomatological Clinic Co., Ltd, Chengdu, China, for his significant contributions to in vivo experiments and analyzing part of the data in this study. We also thank Shuang Li from Minshan Cell Engineering Technology Research Institute (Celenov), China (Sichuan) Free Trade Pilot Zone, Chengdu High-tech Zone, Chengdu, Sichuan, China, for her contribution to statistical analysis in this study.

References

1. D. L. Wagner, E. Fritsche, M. A. Pulsipher, N. Ahmed, M. Hamieh, M. Hegde, M. Ruella, B. Savoldo, N. N. Shah, C. J. Turtle, A. S. Wayne and M. Abou-El-Enein, Nat. Rev. Clin. Oncol., 2021, 18, 379–393.
2. M. Passariello, L. Manna, R. Rapuano Lembo, A. Yoshioka, T. Inoue, K. Kajiwara, S. I. Hashimoto, K. Nakamura and C. De Lorenzo, Cell Death Discovery, 2025, 11, 58.
3. C. Roddie, J. Dias, M. A. O’Reilly, M. Abbasian, A. Cadinanos-Garai, K. Vispute, L. Bosshard-Carter, M. Mitsikakou, V. Mehra, H. Roddy, J. A. Hartley, V. Spanswick, H. Lowe, B. Popova, L. Clifton-Hadley, G. Wheeler, J. Olejnik, A. Bloor, D. Irvine, L. Wood, M. A. V. Marzolini, S. Domning, F. Farzaneh, M. W. Lowdell, D. C. Linch, M. A. Pule and K. S. Peggs, J. Clin. Oncol., 2021, 39, 3352–3363.
4. S. M. Larson, C. M. Walthers, B. Ji, S. N. Ghafouri, J. Naparstek, J. Trent, J. M. Chen, M. Roshandell, C. Harris, M. Khericha, T. Schweppe, B. Berent-Maoz, S. B. Gosliner, A. Almaktari, M. A. Ceja, M. S. Allen-Auerbach, J. Said, K. Nawaly, M. Mead, S. de Vos, P. A. Young, C. Oliai, G. J. Schiller, J. M. Timmerman, A. Ribas and Y. Y. Chen, Cancer Discovery, 2023, 13, 580–597.
5. L. Amini, S. K. Silbert, S. L. Maude, L. J. Nastoupil, C. A. Ramos, R. J. Brentjens, C. S. Sauter, N. N. Shah and M. Abou-El-Enein, Nat. Rev. Clin. Oncol., 2022, 19, 342–355.
6. L. Tang, Z. Huang, H. Mei and Y. Hu, Signal Transduction Targeted Ther., 2023, 8, 306.
7. I. Rahimmanesh, M. Tavangar, S. N. Zahedi, Y. Azizi and H. Khanahmad Shahreza, Adv. Biomed. Res., 2022, 11, 94.
8. H. Sudarsanam, R. Buhmann and R. Henschler, Front. Bioeng. Biotechnol., 2022, 10, 886637.
9. J. Jie, D. Mao, J. Cao, P. Feng and P. Yang, ACS Appl. Mater. Interfaces, 2022, 14, 37514–37527.
10. G. Gulden, B. Sert, T. Teymur, Y. Ay, N. N. Tiryaki, A. K. Mishra, E. Ovali, N. Tarhan and C. Tastan, Vaccines, 2023, 11, 313.
11. X. Wang, Y. Liao, D. Liu, J. Zheng and M. Shi, Mol. Ther., 2025, 33, 1380–1406.
12. S. Song, M. Senoussi, P. Escande and P. Villoutreix, PLoS Comput. Biol., 2023, 19, e1011168.
13. P. Bousso, Nat. Rev. Immunol., 2008, 8, 675–684.
14. M. Hommel and B. Kyewski, J. Exp. Med., 2003, 197, 269–280.
15. M. Bajenoff, J. G. Egen, L. Y. Koo, J. P. Laugier, F. Brau, N. Glaichenhaus and R. N. Germain, Immunity, 2006, 25, 989–1001.
16. A. Castenholz, Curr. Top. Pathol., 1990, 84(Pt 1), 1–32.
17. K. Jobin, D. Seetharama, L. Ruttger, C. Fenton, E. Kharybina, A. Wirsching, A. Huang, K. Knopper, T. Kaisho, D. H. Busch, M. Vaeth, A. E. Saliba, F. Graw, A. Pulfer, S. F. Gonzalez, D. Zehn, Y. Liang, M. Ugur, G. Gasteiger and W. Kastenmuller, Science, 2025, 388, eadq1405.
18. J. E. Gretz, E. P. Kaldjian, A. O. Anderson and S. Shaw, J. Immunol., 1996, 157, 495–499.
19. M. Sixt, N. Kanazawa, M. Selg, T. Samson, G. Roos, D. P. Reinhardt, R. Pabst, M. B. Lutz and L. Sorokin, Immunity, 2005, 22, 19–29.
20. E. Perez Del Rio, M. Martinez Miguel, J. Veciana, I. Ratera and J. Guasch, ACS Omega, 2018, 3, 5273–5280.
21. V. Leyva-Aranda, S. Singh, M. J. Telesforo, S. Young, C. Yee and J. D. Hartgerink, ACS Biomater. Sci. Eng., 2024, 10, 1448–1460.
22. G. Mattei, L. Cacopardo and A. A. Ahluwalia, Materials, 2020, 13, 438.
23. A. Y. Malkin, S. R. Derkach and V. G. Kulichikhin, Gels, 2023, 9, 715.
24. E. Perez Del Rio, F. Santos, X. Rodriguez Rodriguez, M. Martinez-Miguel, R. Roca-Pinilla, A. Aris, E. Garcia-Fruitos, J. Veciana, J. P. Spatz, I. Ratera and J. Guasch, Biomaterials, 2020, 259, 120313.
25. P. Agarwalla, E. A. Ogunnaike, S. Ahn, K. A. Froehlich, A. Jansson, F. S. Ligler, G. Dotti and Y. Brudno, Nat. Biotechnol., 2022, 40, 1250–1258.
26. J. J. Pandit, A. A. Yassin, C. U. Martin, G. L. Saux, A. Porgador and M. Schvartzman, Acta Biomater., 2025, 195, 83–93.
27. A. Autengruber, M. Gereke, G. Hansen, C. Hennig and D. Bruder, Eur. J. Microbiol. Immunol., 2012, 2, 112–120.
28. J. Shi, W. Wu, D. Chen, Z. Liao, T. Sheng, Y. Wang, Y. Yao, Q. Wu, F. Liu, R. Zhou, C. Zhu, X. Shen, Z. Mao, Y. Ding, W. Wang, G. Dotti, J. Sun, X. Liang, W. Fang, P. Zhao, H. Li and Z. Gu, Nat. Mater., 2024, 23, 844–853.
29. M. Mantecon-Oria, N. Diban, M. T. Berciano, M. J. Rivero, O. David, M. Lafarga, O. Tapia and A. Urtiaga, Membranes, 2020, 10, 161.
30. A. He, X. Li, Z. Dai, Q. Li, Y. Zhang, M. Ding, Z. F. Wen, Y. Mou and H. Dong, J. Nanobiotechnol., 2023, 21, 236.
31. Y. Shi, Y. Lu and J. You, Theranostics, 2022, 12, 5888–5913.
32. S. J. Hong, H. S. Yu and H. W. Kim, Macromol. Biosci., 2009, 9, 639–645.
33. W. Nan, Y. He, S. Wang and Y. Zhang, Front. Physiol., 2023, 14, 1234104.
34. L. F. K. Uhl and A. Gerard, Int. J. Mol. Sci., 2020, 21, 2674.
35. B. Hemmer, I. Stefanova, M. Vergelli, R. N. Germain and R. Martin, J. Immunol., 1998, 160, 5807–5814.
36. S. M. McCormick and N. M. Heller, Cytokine, 2015, 75, 38–50.
37. P. S. Kabouridis and E. C. Jury, FEBS Lett., 2008, 582, 3711–3718.
38. M. A. Al-Aghbar, A. K. Jainarayanan, M. L. Dustin and S. R. Roffler, Commun. Biol., 2022, 5, 40.
39. Y. Zhao, L. Zhang, L. Liu, X. Zhou, F. Ding, Y. Yang, S. Du, H. Wang, M. Van Eck and J. Wang, Arterioscler., Thromb., Vasc. Biol., 2022, 42, e311–e326.
40. N. Kim, S. Na, J. Pyo, J. Jang, S. M. Lee and K. Kim, Int. J. Mol. Sci., 2024, 25, 9341.
41. H. He, J. Huang, S. Wu, S. Jiang, L. Liang, Y. Liu, W. Liu, L. Xie, Y. Tao, Y. Jiang and L. Cong, J. Hematol. Oncol., 2021, 14, 171.
42. F. J. Staal, T. C. Luis and M. M. Tiemessen, Nat. Rev. Immunol., 2008, 8, 581–593.
43. L. Gattinoni, Y. Ji and N. P. Restifo, Clin. Cancer Res., 2010, 16, 4695–4701.
44. N. Fortelny, M. Farlik, V. Fife, A. D. Gorki, C. Lassnig, B. Maurer, K. Meissl, M. Dolezal, L. Boccuni, A. Ravi Sundar Jose Geetha, M. J. Akagha, A. Karjalainen, S. Shoebridge, A. Farhat, U. Mann, R. Jain, S. Tikoo, N. Zila, W. Esser-Skala, T. Krausgruber, K. Sitnik, T. Penz, A. Hladik, T. Suske, S. Zahalka, M. Senekowitsch, D. Barreca, F. Halbritter, S. Macho-Maschler, W. Weninger, H. A. Neubauer, R. Moriggl, S. Knapp, V. Sexl, B. Strobl, T. Decker, M. Muller and C. Bock, Nat. Immunol., 2024, 25, 847–859.
45. M. J. Podolska, R. Grutzmann, C. Pilarsky and A. Benard, Front. Immunol., 2024, 15, 1411047.
46. S. Yamaguchi, A. Nambu, T. Numata, T. Yoshizaki, S. Narushima, E. Shimura, Y. Hiraishi, K. Arae, H. Morita, K. Matsumoto, I. Hisatome, K. Sudo and S. Nakae, Sci. Rep., 2018, 8, 15750.
47. M. Li, L. Y. Chin, S. Shukor, A. G. Tamayo, M. V. Maus and B. Parekkadan, Autoimmunity, 2019, 52, 102–107.
48. I. L. Ch’en, J. S. Tsau, J. D. Molkentin, M. Komatsu and S. M. Hedrick, J. Exp. Med., 2011, 208, 633–641.
49. J. V. Stein and S. F. Gonzalez, J. Allergy Clin. Immunol., 2017, 139, 12–20.
50. M. Novkovic, L. Onder, J. Cupovic, J. Abe, D. Bomze, V. Cremasco, E. Scandella, J. V. Stein, G. Bocharov, S. J. Turley and B. Ludewig, PLoS Biol., 2016, 14, e1002515.
51. F. D. Brown and S. J. Turley, J. Immunol., 2015, 194, 1389–1394.
52. A. S. Cheung, D. K. Y. Zhang, S. T. Koshy and D. J. Mooney, Nat. Biotechnol., 2018, 36, 160–169.
53. J. Hyun, S. J. Kim, S. D. Cho and H. W. Kim, Biomaterials, 2023, 297, 122101.
54. D. K. Y. Zhang, J. M. Brockman, K. Adu-Berchie, Y. Liu, Y. Binenbaum, I. de Lazaro, M. C. Sobral, R. Tresa and D. J. Mooney, Nat. Biomed. Eng., 2025, 9, 268–278.
55. Z. Liao, J. Jiang, W. Wu, J. Shi, Y. Wang, Y. Yao, T. Sheng, F. Liu, W. Liu, P. Zhao, F. Lv, J. Sun, H. Li and Z. Gu, Natl. Sci. Rev., 2024, 11, nwae018.
56. D. D. de Souza Morais, C. B. B. Luna, E. B. Bezerra, D. C. de França, E. M. Araújo, E. P. do Nascimento, A. D. de Oliveira and T. J. A. de Mélo, Sustainability, 2022, 14, 9254.
57. D. Zhou, X. Zhu and Y. Xiao, Biomark. Res., 2024, 12, 132.

---