Shape-dependent transmembrane transport of mesenchymal stem cell membrane-coated mesoporous silica nanoparticles enhances anti-glioma therapy

Abstract

The blood–brain barrier (BBB) remains a major obstacle to effective glioma chemotherapy, and although mesenchymal stem cell (MSC) membrane (MSCM) coated nanoparticles offer a biomimetic strategy for glioma-targeted delivery, the role of particle morphology in such systems remains unclear. Herein, doxorubicin-loaded mesoporous silica nanoparticles coated with MSC membranes (MSCM@MSN@DOX) were developed with three morphologies—spherical (AR∼1), short rod-shaped (AR∼1.5), and long rod-shaped (AR∼3) to study the shape effect of MSC biomimetic drug delivery system on glioma treatment. All formulations showed successful membrane coating, retention of MSC-associated proteins, and comparable drug loading, release, and stability, enabling assessment of shape-dependent delivery behavior. Compared with AR∼1 and AR∼1.5 counterparts, AR∼3 particles exhibited enhanced motility, higher uptake by bEnd.3 and GL261 cells, improved trans-BBB transport, and deeper tumor spheroid penetration. Mechanistic studies suggested that these effects were associated with stronger nanoparticle–cell membrane interactions and morphology-dependent endocytic behavior. In an intracranial GL261 glioma model, AR∼3 particles achieved greater brain/glioma-region accumulation, more pronounced tumor growth inhibition, and extended survival, with no overt systemic toxicity under the tested regimen. These findings identify particle morphology as an important optimization parameter within the investigated MSC membrane-coated MSN system and provide a basis for shape optimization of this biomimetic nanocarrier platform for brain-targeted drug delivery.

Graphical abstract

Highlights

• •MSC membrane-coated mesoporous silica nanoparticles with distinct aspect ratios were designed for glioma-targeting.
• •NTA and SPR analyses quantitatively linked nanoparticle shape to motility and cell-membrane interaction strength.
• •Bio-TEM and endocytosis studies revealed shape-dependent membrane wrapping and uptake mechanisms underlying BBB transport.
• •Long rod-shaped nanoparticles achieved superior BBB penetration and therapeutic efficacy with negligible systemic toxicity.

1. Introduction

Glioma is the most common and aggressive primary malignancy of the central nervous system (CNS), accounting for approximately 80% of all malignant brain tumors [1,2]. Despite multimodal treatment combining surgical resection, radiotherapy, and temozolomide-based chemotherapy, the median survival of patients with high-grade glioma remains below 15 months, with recurrence nearly universal [3]. A principal obstacle to effective pharmacological treatment is the blood–brain barrier (BBB), a highly selective vascular interface that severely restricts the CNS penetration of most therapeutic agents. Active efflux transporters expressed at the BBB further reduce the intratumoral accumulation of chemotherapeutics, leaving drug concentrations at the tumor site insufficient for sustained cytotoxicity [4,5]. The blood–tumor barrier (BTB) further compounds this challenge [6]. Together, these barriers limit the efficacy of systemically administered chemotherapy and underscore the need for delivery strategies specifically engineered to traverse the BBB and achieve meaningful glioma-site accumulation [7].

Nanoparticle-based drug delivery systems offer a rational strategy to address these limitations by improving drug stability, prolonging circulation half-life, and enabling passive or active accumulation at tumor sites [[8], [9], [10]]. However, conventional nanocarriers face persistent challenges: immune-mediated clearance [11], non-specific biodistribution [12], and insufficient trans-BBB transport efficiency [13] constrain their therapeutic utility in CNS applications. Cell membrane-coating strategies have emerged as a promising biomimetic approach to overcome several of these barriers simultaneously [[14], [15], [16]]. Among candidate membrane sources, mesenchymal stem cell (MSC) membranes are of particular interest because MSC-derived membranes have been reported to retain adhesion molecules and surface proteins associated with trans-endothelial migration and glioma tropism [14,17,18]. MSCs exhibit intrinsic tropism toward injured and tumor-associated tissues and can migrate across the BBB toward glioma lesions. Accordingly, MSC membrane coating may transfer membrane-associated features involved in endothelial interaction [19,20], immune tolerance [21], and tumor homing [22] to nanoparticles, thereby facilitating trans-BBB transport and glioma-site accumulation. Unlike MSC-based cell therapy, the biological functions of MSC membrane-coated nanoparticles mainly arise from retained membrane-associated proteins rather than paracrine secretion [23], although the specific components mediating BBB penetration and glioma homing remain to be clarified [24].

Despite growing interest in biomimetic nanocarriers, a critical design parameter has received comparatively little systematic attention: particle shape. The overwhelming majority of studies on brain-targeted nano delivery systems have focused on optimizing particle size [25], surface charge [26], ligand density [27], and pore architecture [28], while the influence of particle morphology on delivery performance remains poorly characterized [[29], [30], [31]]. Particle shape may represent an underexplored formulation parameter that modulates multiple sequential events in brain-targeted delivery [32]. Compared with spherical nanoparticles, non-spherical nanoparticles with comparable surface properties may exhibit altered diffusion behavior [33], membrane contact geometry [34], and cellular internalization processes [35], thereby potentially influencing endothelial interactions and trans-BBB transport efficiency [[36], [37], [38]]. These differences may be partly associated with shape-dependent membrane wrapping behavior during cell–particle interactions [39], although the underlying biophysical mechanism remains to be directly validated. For membrane-coated nanocarriers, particle shape may not only directly influence transport behavior but also affect membrane coating integrity, protein presentation, and the resulting nano–bio interface [40,41]. However, whether and how particle morphology modulates trans-BBB transport and glioma-targeted delivery in MSC membrane-coated systems remains unclear, representing a critical gap in the rational design of brain-targeted drug delivery systems.

Mesoporous silica nanoparticles (MSNs) provide a well-suited model platform for addressing this question. MSNs combine high specific surface area, tunable mesoporous channels, favorable drug-loading capacity, and surface-amenable functionalization [42,43] with a key practical advantage: their morphology can be controlled during synthesis to yield distinct particle shapes—most notably spherical and rod-shaped geometries—under otherwise comparable surface and compositional conditions [44]. This shape versatility, coupled with the widely explored biomedical applicability of silica-based nanomaterials, makes MSNs an experimentally tractable carrier for isolating the contribution of particle geometry to delivery outcomes, while minimizing confounding differences in surface chemistry and drug payload.

In this study, we engineered MSC membrane-coated MSNs with distinct morphologies to investigate how particle shape modulates BBB penetration and glioma-targeted delivery. By combining physicochemical characterization with analyses of particle diffusion, nanoparticle–cell membrane interactions, in vitro trans-BBB transport, cellular uptake, and in vivo glioma accumulation, we aimed to clarify the contribution of particle geometry to the delivery performance of MSC biomimetic nanocarriers. We hypothesized that nanoparticle morphology regulates trans-BBB delivery by altering diffusion behavior, membrane interaction, and endothelial transport processes (Fig. 1). Within the investigated MSC membrane-coated MSN system, this study identifies particle morphology as an important optimization parameter influencing brain-targeted delivery performance. These findings provide system-specific guidance for the shape optimization of MSC membrane-coated mesoporous silica nanocarriers, while further studies using other carrier materials and morphological ranges will be required to determine the broader applicability of these observations.

Fig. 1.

Schematic illustration of the assembly process and shape effect of MSCM@MSN@DOX with different morphologies.

2. Materials and methods

2.1. Materials

Cetyltrimethylammonium bromide (CTAB) and 3-aminopropyltriethoxysilane (APTES) were obtained from Shanghai Aladdin Biochemical Technology Co., Ltd. (China). Tetraethyl orthosilicate (TEOS), aqueous ammonia, and anhydrous ethanol were purchased from Sinopharm Chemical Reagent Co., Ltd. (China). Dialysis bags (MWCO: 14,000 Da) were purchased from Greenbird Information Technology Co., Ltd. (China). Dulbecco's modified Eagle medium (DMEM), fetal bovine serum (FBS), penicillin–streptomycin solution, and 0.25% trypsin–EDTA solution were obtained from Gibco BRL (USA). The anti-VLA-4 antibody was purchased from Boster Biological Technology Co., Ltd. (China), and the anti-CXCR4 antibody was obtained from Nanjing Qingke Biotechnology Research Center Co., Ltd. (China). The anti-CD47 antibody and BCA protein assay kit were purchased from Beyotime Biotechnology (China). The DiR near-infrared fluorescent probe was purchased from Xi'an Ruixi Biological Technology Co., Ltd. (China), and D-luciferin potassium salt was obtained from Adamas Reagent Co., Ltd. (China). The Cell Counting Kit-8 (CCK-8), DAPI staining solution, 4% paraformaldehyde fixative, and TUNEL apoptosis detection kit were purchased from Beyotime Biotechnology (China). The Ki-67 immunohistochemistry kit was obtained from Shanghai Qiming Biotechnology Co., Ltd. (China). A 2.5% glutaraldehyde fixative solution was purchased from SPI Supplies (USA), and the liposome sensor chip (LIP chip) used for surface plasmon resonance analysis was obtained from Nicoya Lifesciences.

2.2. Preparation of MSCM@MSN@DOX with different morphologies

2.2.1. Synthesis of MSN with different morphologies

MSNs with different morphologies were fabricated using the template method. According to the aspect ratio of nanoparticles, the corresponding amount of CTAB was dissolved in 50 mL of purified water at 50 °C. Then, 0.57 mL of ammonia solution (28%–30%) was added, and the mixture was stirred for 1 h. Two-thirds of the corresponding amount of TEOS was then added. After vigorous stirring at room temperature for 1 h, the corresponding amount of APTES and the remaining one-third of TEOS were added, followed by 4 h of stirring. MSNs of different shapes were collected by centrifugation at 15,000 g for 20 min, washed and redispersed several times with deionized water and ethanol. The CTAB template was removed by stirring in 5% HCl/EtOH for 24 h. The MSNs were obtained by freeze drying after washing.

For MSNs of ARs∼1, the reactant concentrations are 0.05636 g CTAB, 0.0235 mL APTES, and 0.2271 mL TEOS. For MSNs of ARs∼1.5, the reactant concentrations are 0.1075 g CTAB, 0.0466 mL APTES, and 0.4565 mL TEOS. And for MSNs of ARs∼3, the reactant concentrations are 0.1567 g CTAB, 0.0689 mL APTES, and 0.6653 mL TEOS.

2.2.2. DOX loading

In order to unify the drug loading efficiency of MSNs with different shapes, different mass ratios of DOX/NH2-MSNs (1/4, 1/3, 1/2 and 1/1) were screened. In short, 1 mg NH2-MSNs with different shapes were dispersed in 1 mL PBS, and 500, 1000, and 2000 μg of DOX were added and stirred at 25 °C for 24 h. After centrifugation at 18,000 g for 10 min, the absorbance of supernatant was measured to calculate DOX content and the drug loading capacity of MSNs with different morphologies using the standard curve.

To obtain doxorubicin-loaded mesoporous silica nanoparticles (MSN@DOX), MSNs of different morphologies and equal mass of doxorubicin hydrochloride were dispersed in 5 mL water and stirred in the dark at 25 °C for 24 h. The particles were then collected by centrifugation at 15,000 g for 20 min to remove unloaded DOX.

2.2.3. Mesenchymal stem cell culture and membrane extraction

Healthy male SPF-grade C57BL/6J mice (4 weeks old) were purchased from the Production Department of Animals Core Facility, Nanjing Medical University. All experimental procedures complied with Nanjing Medical University's guidelines on laboratory animals' welfare. The work was approved by the Animal Ethics Committee of Nanjing Medical University (approval number: 2406047). To isolate MSCs from bone marrow, mice were sacrificed and immediately immersed in 75% ethanol for 10 min. Under aseptic conditions, rinse the tibiae and femurs three times with pre-cooled PBS buffer containing 1% penicillin–streptomycin, and then cut open both ends to flush the bone marrow cavity repeatedly with 1 mL of complete DMEM medium (with 10% fetal bovine serum and 1% antibiotics). The collected cells were filtered through a 70 μm cell strainer after resuspending, and then centrifuged at 300 g for 5 min. The precipitate was resuspended in fresh complete DMEM medium, and cultured in a 37 °C incubator with 5% CO₂ (Heracell 240i, Thermo Fisher Scientific).

To extract MSC membrane, MSCs were digested with trypsin, and pipetted to obtain a single-cell suspension. The suspension was then centrifuged at 1000 rpm for 3 min at room temperature. The precipitate was resuspended in PBS containing protease inhibitors and repeat a rapid freeze–thaw cycle for three times. Subsequently, the sample was sonicated three times using a probe-type sonicator under ice-bath conditions to ensure complete cell membrane lysis, and then centrifuged at 600 g for 10 min (3-30 KS, Sigma, USA). Carefully collect the supernatant and centrifuge at 70,000 g for 30 min at 4 °C (3-30 KS, Sigma, USA) to finally obtain MSCM.

2.2.4. Construction of MSCM@MSN@DOX with different morphologies

MSCM and MSN@DOX with different morphologies were mixed at a 1:10 mass ratio, followed by sonication five times using an ultrasonic disruptor under ice-water bath conditions. The sonicated mixture was then physically extruded through polycarbonate membranes with pore sizes of 800 nm and 400 nm sequentially using a liposome extruder, with 20 consecutive extrusions, to obtain MSCM@MSN@DOX with different morphologies.

2.3. Physicochemical characterization of MSCM@MSN@DOX with different morphologies

The particle size, PDI, and zeta potential were measured by dynamic light scattering (DLS) using Zetasizer Nano (Zetasizer Nano ZS90, Malvern, USA). The morphology of MSNs was determined with transmission electron microscopy (FEI Tecnai G2 Spirit BioTWIN, FEI, USA). Fourier-transform infrared (Nicolet iS 10, ThermoFisher, USA) spectroscopy was performed to confirm the complete removal of the CTAB template. Nitrogen adsorption analysis (TriStar II Plus, Micromeritics, USA) was used to measure the surface area and pore size of the MSNs. The BJH method was applied to obtain specific surface area, pore volume, and pore size distribution data from the isotherms.

2.4. Verification of MSCM membrane protein integrity and successful coating fusion

The morphology of successfully coated MSCM@MSN was determined with transmission electron microscopy (FEI Tecnai G2 Spirit BioTWIN, FEI, USA). To further verify MSCM membrane protein integrity, MSCM@MSN and pure MSCM were separately lysed with RIPA protein lysis buffer on ice for 30 min to extract proteins. After centrifugation at 12,000 g and 4 °C for 10 min (3-30 KS, Sigma, USA), the supernatant was collected to obtain MSCM@MSN and pure MSCM protein samples. Protein concentrations of all samples were determined using the BCA assay kit and subsequently adjusted to a uniform level. The sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) method was utilized to analyze the protein expression profiles of MSCM, and MSCM@MSN. Western blotting (WB) was employed to evaluate representative MSC-associated membrane proteins, including CD47, CXCR4, and VLA-4, with ATP1A1 used as a membrane-associated reference protein.

2.5. Particle motility

Nanoparticle tracking analysis (NanoSight NS300, Malvern, UK) was used to analyze particle motility by capturing motion videos of sufficient duration to acquire trajectory data. Trajectories with data exceeding 25 frames were selected, and the motility parameters of MSCM@MSN were analyzed using MATLAB.

2.6. Drug loading and in vitro release

To establish the calibration curve, a precise amount of doxorubicin hydrochloride was weighed and dissolved in a volumetric flask to prepare a 1 mg/mL stock solution. The stock solution was serially diluted to prepare solutions of 500.0, 200.0, 100.0, 50.0, 20.0, 10.0, 5.0, and 1.0 μg/mL. Absorbance was measured at 480 nm using a microplate reader to construct the DOX standard curve, with DOX mass concentration and absorbance plotted on the x- and y-axes respectively. The encapsulation rate (EE, %) and drug loading (DL, %) of MSCM@MSN@DOX were calculated according to Eqs. (2) and (3):

$$\text{EE}(\%)=\left(1-{\mathrm{W}}_{\mathrm{f}}\right)\times 100/{\mathrm{W}}_{\mathrm{t}}$$ (2)

$$\text{DL}(\%)=\left(1-{\mathrm{W}}_{\mathrm{f}}\right)\times 100/{\mathrm{W}}_{\mathrm{m}}$$ (3)

where W_f, W_t and W_m refer to the mass of the free DOX, the total DOX input and MSCM@MSN@DOX input, respectively.

The in vitro release behavior of MSCM@MSN@DOX with different morphologies was investigated using the dialysis method. MSCM@MSN@DOX was dispersed in triplicate in 2 mL of PBS buffer (pH 7.4), then placed into activated dialysis bags (MWCO: 14,000 Da, Greenbird Information Technology Co. Ltd., China). At predetermined time points, 200 μL of sample was taken, and fresh buffer was added at the same volume. The cumulative release of DOX was quantified by a full-band multifunctional microplate reader (Infinite 200, Tecan, Switzerland).

2.7. Storage stability

Freshly prepared MSCM@MSN@DOX of different morphologies were resuspended in PBS and stored at 4 °C. The hydrodynamic size and PDI were monitored for over a week using a particle size analyzer based on DLS under identical instrumental settings.

2.8. Cytotoxicity and in vitro anti-tumor efficacy evaluation

The cytotoxicity of MSCM@MSN@DOX on bEnd.3 and GL261 cells was investigated by CCK-8 assay in comparison with a control group treated with free DOX. Both cell suspension concentrations of bEnd.3 and GL261 cells were adjusted with sterile complete DMEM medium and seeded into 96-well plates at a density of 8 × 10³ cells per well, and cultured overnight (37 °C, 5% CO₂, humidified). The cells were incubated with DOX and MSCM@MSN@DOX with different morphologies at various DOX concentrations for 24 h, followed by addition of 200 μL CCK-8 reagent diluted with DMEM and incubation for 4 h at 37 °C. Absorbance was measured at 470 nm using a microplate reader (ELX800, Bio Tek, USA) at 0.5, 1, 2, and 3 h. Endothelial cell survival rate and tumor cell inhibition rate were calculated for each group.

2.9. In vitro cellular uptake

To investigate the uptake of the various formulation by cells, GL261 cells and bEnd.3 cells were seeded into 24-well plates at a density of 5 × 10⁴ cells per well and incubated for 24 h. Subsequently, MSCM@MSN@DOX nanoparticles of different morphologies were diluted to 75 μg/mL based on DOX concentration with DMEM medium, 500 μL of the diluted solution was added and incubated for 4 h. Followed by three PBS washes, nuclei were stained with DAPI, and the cellular uptake of the fluorescent nanoparticles was observed under a fluorescence microscope. For quantitative analysis of nanoparticle cellular uptake, 100 μL of RIPA lysis buffer was added to each well of the 24-well plate after removing drug-containing medium and washing three times with PBS. Cells were scraped from the plate bottom using a cell scraper, lysed at 4 °C with shaking for 30 min, and centrifuged at 12,000 g for 15 min (3-30 KS, Sigma, USA). Half of the supernatant was used for fluorescence measurement with a multifunction microplate reader, and the remaining half was used for protein concentration determination with a BCA assay kit.

2.10. In vitro BBB permeability

The bEnd.3 cells (8 × 10⁴ cells) were implanted into the 24-well transwell upper chamber to form an endothelial barrier. Transepithelial electrical resistance (TEER) was measured using a TEER meter (KingTech Technology Co., Ltd., China). When TEER values reached 200 Ω/cm², sufficient tight junctions were considered to have formed [45]. MSCM@MSN@DOX nanoparticles of different morphologies were diluted to 75 μg/mL based on DOX concentration with DMEM medium, then 200 μL of the diluted solution was added to the upper chamber of each well. At 2, 4, and 8 h time points, 100 μL of medium was withdrawn from the lower chamber, and fluorescence intensity was measured using a fluorescence microplate reader (Infinite 200, Tecan, Switzerland).

2.11. In vitro tumor spheroid penetration performance

In vitro tumor spheroid was established as follows: GL261 cells were seeded into a low-attachment 96-well plates (1 × 10³ cells per well). The plates were centrifuged at 251 g for 3 min (Xiangyi). The tumor spheroids were cultured till the diameter reached 300–500 μm [46].

MSCM@MSN@DOX were diluted to 75 μg/mL based on DOX concentration with DMEM medium, then added according to the wells and incubated for 8 h. The spheroids were gently washed twice with 200 μL of PBS after the incubation. Subsequently, 100 μL of PFA fixative was added to each well for fixation. Then, the spheroids were washed 2–3 times with PBS, and carefully transferred to confocal dishes to observe under a confocal laser scanning microscope (LSM800, Zeiss, Germany) using Z-stack scanning to evaluate the intratumoral distribution and penetration depth of different formulations within the tumor spheroids, followed by three-dimensional reconstruction based on the acquired image stacks.

2.12. Characterization of cellular endocytosis process

Bio-transmission electron microscopy (bio-TEM) was employed to observe the cellular endocytosis process of MSCM@MSN@DOX. The bEnd.3 cells were cultured into dishes and incubated with MSCM@MSN@DOX nanoparticles of different morphologies at 75 μg/mL based on DOX concentration with DMEM medium for 1 h. After removing drug-containing medium and washing three times with PBS, 2.5% glutaraldehyde fixative was added for pre-fixation for 30 min. Cells were gently scraped with a cell scraper to centrifuged into pellets, and fixed overnight at 4 °C in fresh 2.5% glutaraldehyde fixative. Cell pellets were embedded and made into ultrathin sections. Sections were stained with uranyl acetate and lead citrate, and then observed using a low-voltage transmission electron microscope (FEI Tecnai G2 120 KV, FEI, USA).

2.13. In vitro cellular endocytosis pathways verification

The bEnd.3 cells were seeded into 12-well plates at 1 × 10⁵ cells per well in DMEM medium and cultured for 24 h at 37 °C. Then, original medium was removed, and various endocytosis inhibitors diluted in DMEM medium (CPZ (10 μg/mL), MβCD (10 mmol/L), and EIPA (0.5 mmol/L)) were added to corresponding wells for 0.5 h pretreatment, with inhibitor-free groups as controls. After a 4 h incubation, cells were washed three times with PBS, stained with DAPI, and observed under an inverted fluorescence microscope. Meanwhile, a comparative experiment on energy suppression was conducted under the condition of 4 °C.

For quantitative analysis, 80 μL RIPA lysis buffer was added to each well of the 12-well plate after removal of drug-containing medium and PBS washing. Cells were scraped, lysed at 4 °C with shaking for 30 min, centrifuged at 12,000 g for 15 min. The fluorescence intensity was measured using a fluorescence microplate reader (Infinite 200, Tecan, Switzerland), and protein concentrations were determination with a BCA assay kit.

2.14. In vitro cellular molecular interaction assessment

Surface plasmon resonance measurement was used to assess the cellular molecular interaction between bEnd.3 cells and MSCM@MSN@DOX. Cells were seeded into 6-well plates at 2 × 10⁵ cells per well in DMEM medium and cultured for 24 h, preparing for subsequent coupling with liposome chips. PBS buffer was run, and the chip CH1 and CH2 channel surfaces were activated for a contact time of 400 s at a flow rate of 20 μL/min. After chip activation, bEnd.3 cells were coupled to both CH1 and CH2 channels with PBS as the coupling buffer, contact time of 300 s, and flow rate of 20 μL/min. MSCM@MSN@DOX of different morphologies were sequentially injected at various concentrations, with a contact time of 270 s and dissociation time of 5730 s. Regeneration between concentrations was performed with 20 mM CHAPS solution for a contact time of 30 s.

2.15. In vivo distribution and targeting assay

The animal experiments were approved by the Institutional Animal Care and Use Committee of Nanjing Medical University (approval number: 2406047). An intracranial glioma murine model was established in male C57BL/6J mice aged five weeks. The GL261 model was chosen for its syngeneic and immunocompetent traits, which enable stable recapitulation of key features of glioma and the tumor microenvironment (TME). Briefly, GL261 cells were suspended at a density of 1 × 10⁵ cells/5 μL and injected into the mouse brains under anesthesia induced by pentobarbital sodium (1% w/v). The injection site was precisely located at a depth of 2 mm to the right side of the sagittal suture on top of the anterior fontanelle using stereotaxic apparatus.

During MSCM@MSN@DOX preparation, DiR dye was incorporated as a fluorescent tracer to label nanoparticles. After one week of the introduction of cells, modeled mice received a tail vein injection of dye-labeled nanoparticles. A small animal in vivo imaging system (Caliper, USA) was used to detect cerebral fluorescence signal intensity at 4, 6, 8, 12, and 24 h. At 24 h, mice were sacrificed, and major organs were harvested for ex vivo fluorescence distribution observation. Nanoparticle targeting to brain glioma tissue and biodistribution in other organs were quantified using Region-of-Interest (ROI) analysis performed by the Living Image software (Living Image 4.2, Caliper, USA).

2.16. In vivo therapeutic efficacy assessment

GL261-Luc glioma model mice were established following the protocol established previously. Modeled mice were randomly divided into five groups: saline group, free DOX group, MSCM@MSN@DOX (AR∼1) group, MSCM@MSN@DOX (AR∼1.5) group, and MSCM@MSN@DOX (AR∼3) group. After one week of cell injection, different formulations were administered via tail vein injection every 3 days for a total of five administrations at 200 μL (dosage equivalent to 2 mg/kg based on DOX weight and mouse body weight) each time. Chemiluminescent signals emitted by the glioma were measured every 7 days using an in vivo imaging system (Caliper, USA) to qualitatively evaluate tumor growth based on intensity levels observed. Anti-tumor efficacy of each formulation group was evaluated based on tumor growth inhibition rate and mouse survival.

TUNEL was employed for tumor cell apoptosis assessment. After five times of drug treatment, glioma model mice were sacrificed, and brain tumor tissues were carefully excised for frozen tissue sectioning. TUNEL probe was incubated with mouse brain tumor sections at 37 °C for 60 min, then washed three times with PBS, stained with DAPI for 5 min, and observed under a fluorescence microscope to assess tumor cell apoptosis.

Additionally, the brain tumor tissues were carefully excised and fixed with 4% paraformaldehyde to prepare paraffin tissue sections. A commercial Ki-67 protein immunohistochemistry kit was used for antibody incubation and staining of brain tumor sections. After incubation, observe the sections under an optical microscope to assess tumor cell proliferation. In order to display the pathological morphology of the tumor zone, the paraffin sections underwent hematoxylin and eosin (H&E) staining.

2.17. In vivo preliminary safety assessment

Healthy ICR male mice were randomly divided into saline group, MSCM@MSN@DOX (AR∼1) group, MSCM@MSN@DOX (AR∼1.5) group, and MSCM@MSN@DOX (AR∼3) group. Corresponding formulations were administered via tail vein injection at a volume of 200 μL (dosage equivalent to 2 mg/kg based on DOX). Injections were given every 3 days for a total of 5 doses, and body weight was continuously monitored. After 15 days, mice were sacrificed. Blood and major organ samples were collected. Blood samples were analyzed for routine biochemical indices. Organ samples were stained with H&E. Histopathological examination of stained tissue sections was performed under an optical microscope.

2.18. Sample size determination

Animal sample sizes were selected with reference to previously published studies employing comparable orthotopic glioma models and in vivo treatment designs [47], together with consideration of the objectives and expected variability of the respective experimental endpoints. For the exploratory biodistribution study, n = 3 mice per group was used for short-term assessment of in vivo nanoparticle distribution and brain/glioma accumulation. In the therapeutic study, longitudinal tumor bioluminescence was quantitatively analyzed using n = 3 mice per group, whereas n = 10 mice per group was used for survival analysis. The larger cohort for survival analysis was selected considering the potential attrition associated with intracranial tumor establishment and prolonged follow-up, as well as the greater inter-animal variability expected for long-term survival outcomes. For the preliminary safety study, n = 5 mice per group was used considering the repeated administration regimen and the approximately 15-day observation period. Overall, animal numbers were selected by balancing the scientific requirements of each experimental endpoint with animal-welfare considerations in accordance with the 3Rs principle.

2.19. Statistical analysis

Data were presented as the means ± standard deviation (SD). Statistical analysis of differences between the experimental groups was performed using multiple Student's t-tests or one-way ANOVA, as appropriate. Survival curves were generated using the Kaplan–Meier method and compared using the log-rank (Mantel–Cox) test. Two predefined families of pairwise comparisons were evaluated according to the original study objectives: (1) comparisons of each treatment group with the saline control, to assess treatment effects relative to untreated tumor progression; and (2) comparisons of the long rod-shaped MSCM@MSN@DOX formulation (AR∼3) with the other treatment groups, to evaluate its relative survival benefit. Within each comparison family, P values were adjusted for multiple comparisons using the Holm–Šídák method. Differences were considered statistically significant when P < 0.05. (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).

3. Results

3.1. Preparation and characterization of MSCM@MSN@DOX with different morphologies and transport-related properties

Here, we prepared MSNs with three different morphologies using the Janus method, including spheres (AR∼1), short rods (AR∼1.5), and long rods (AR∼3). These particles were synthesized under diluted TEOS and CTAB surfactant conditions with NH₃·H₂O as catalyst, and their size and morphologies were adjusted by changing the reaction conditions (Table S1). As CTAB concentration increased, the aspect ratios (ARs) of the particles also increased. TEM images (Fig. 2A) revealed that the synthesized MSNs were uniform in size with well-defined edges, presenting three distinct morphologies: spherical particles had diameters of approximately 60-80 nm; short rod particles had a major-axis diameter of 100-120 nm and minor-axis diameter of 50-60 nm; and long rods showed dimensions of about 180-220 nm in length and 50-60 nm in width. Rod-shaped MSNs showed approximately hexagonal cross-sections, and all three MSN formulations exhibited highly ordered porous channel structures. Dynamic light scattering further showed hydrodynamic sizes of 188.3, 192.4, and 254.1 nm for AR∼1, AR∼1.5, and AR∼3 particles, respectively, with low polydispersity indices (≤0.15) and negative zeta potentials (Fig. 2B, C and Table S2). The negative surface charge would drive the subsequent biomimetic cell membrane adsorption coating process, improve coating efficiency, and facilitate the formation of complete, dense biomimetic coatings with reduced exposed areas, thereby better achieving particle surface functionalization. The negative surface charge was favorable for the subsequent biomimetic membrane coating process. In addition, nitrogen adsorption-desorption isotherms confirmed the mesoporous characteristics of all MSN formulations (Fig. S1). The specific surface area and total pore volume followed the order of AR∼1.5 > AR∼3 > AR∼1, whereas the average pore size followed the order of AR∼1 > AR∼3 > AR∼1.5 (Table S3). FT-IR spectra showed the absence of characteristic CTAB signals, indicating complete removal of the template (Fig. S2). Together, these results confirmed the successful preparation of shape-defined MSNs with broadly comparable physicochemical characteristics, providing a suitable basis for subsequent evaluation of morphology-dependent biological performance.

Fig. 2.

Preparation and characterization of MSCM@MSN@DOX with different morphologies. TEM images of MSN (scale bar = 100 nm) and MSCM@MSN (scale bar = 200 nm) (A, D). Particle size distribution and zeta potential of MSN (B, C). Particle size distribution and zeta potential of MSCM@MSN. (E, F) Coomassie Brilliant Blue staining of total proteins (G). Western blot analysis of representative membrane proteins (CD47, CXCR4, and VLA-4) in MSCM and lysed MSCM@MSN, with ATP1A1 used as a membrane-associated reference protein (H). DOX release profiles of MSCM@MSN@DOX at pH 7.4 (I). Representative motion trajectories of MSCM@MSN@DOX (J). Cumulative mean square displacement (MSD) curves calculated using MATLAB (K). Data in (B–E) and (H) are presented as mean ± SD (n = 3).

Doxorubicin hydrochloride was loaded into MSNs through electrostatic adsorption to produce MSN@DOX with different morphologies. To enable subsequent biological comparison among the three formulations, the DOX-to-MSN feeding ratio was first optimized (Fig. S3). Drug loading capacity and encapsulation efficiency were evaluated at four DOX/MSN mass ratios (1:1, 1:2, 1:3, and 1:4). Among these conditions, a DOX/MSN ratio of 1:1 gave comparable drug loading capacity and encapsulation efficiency among the three morphologies, with a loading level of approximately 43%, and was therefore selected for subsequent studies.

We then prepared MSCM@MSN@DOX with different morphologies using the conventional cell membrane camouflage method. TEM images showed MSCM@MSN with the three corresponding morphologies (AR∼1, AR∼1.5, and AR∼3), while the dense lipid layer visible at the arrow-marked positions on the nanoparticle surface supported successful coating of the MSCM shell (Fig. 2D). After biomimetic coating, the hydrodynamic diameters of all particles increased to varying degrees (Fig. 2E), and the zeta potentials became more negative compared with those of the uncoated particles (Fig. 2F), further supporting successful MSC membrane coating. As the functions of MSC membranes mainly depend on their surface proteins, we further evaluated whether these proteins were retained on MSCM@MSN with different morphologies. SDS-PAGE analysis showed that the overall protein profiles of native MSC membranes were largely retained after coating onto MSNs of different morphologies (Fig. 2G). Western blot further confirmed that representative MSC-associated membrane proteins, including CD47, CXCR4, and VLA-4, were preserved on the surface of MSCM@MSN (Fig. 2H). ATP1A1 was included as a membrane-associated reference protein, and comparable ATP1A1 signals were observed among the analyzed MSCM@MSN samples. These results collectively indicated that MSC membranes were successfully coated onto MSNs of different morphologies while retaining key membrane-associated features, thereby providing a relatively comparable biomimetic platform for subsequent evaluation of morphology-dependent delivery behavior.

The in vitro release of DOX from MSCM@MSN@DOX (AR∼1, ∼1.5, ∼3) was compared in PBS (pH 7.4) at 37 °C (Fig. 2I). The three membrane-coated formulations exhibited comparable short-term DOX release profiles under the tested conditions, with no significant inter-group differences within 12 h and cumulative release reaching approximately 26%. No obvious burst release was observed during the tested period. In addition, MSCM@MSN@DOX (AR∼1, ∼1.5, ∼3) were stable in PBS after one week of storage at 4 °C, with no significant changes in particle size or PDI (Fig. S4). Together, these results indicated that differences observed in subsequent biological studies were unlikely to arise from major disparities in release characteristics or storage stability among the three formulations.

Multiple-particle tracking provides a nondestructive and highly sensitive method for evaluating particle motion behavior in complex fluid environments. Here, we used this approach to investigate the motility of MSCM@MSN@DOX in fresh PBS as a simplified fluid environment. The trajectories of the particles were captured, and representative trajectories of MSCM@MSN@DOX (AR∼1, ∼1.5, ∼3) were mapped in Fig. 2J. Spherical particles moved within a relatively limited area, whereas both short-rod and long-rod particles displayed a broader movement range, suggesting improved diffusion behavior. The accumulative mean square displacement (MSD), average velocity and diffusion coefficient were calculated on a time scale of 1 s (Fig. 2K and S5). As shown in Fig. 2K, rod-shaped particles exhibited higher MSD, average velocity, and diffusion coefficient than spherical particles, with the AR∼3 group showing the greatest motility. In summary, rod-shaped particles exhibited higher velocity and diffusion-associated behavior than spherical particles. The enhanced motility of long rod-shaped particles may be related to translational-rotational coupling and hydrodynamic effects associated with shape anisotropy. These results suggested that particle motility could be modulated by morphology, thereby partially influencing diffusion behavior. The enhanced motility also increases the probability of contact between drug-loaded systems and target sites after transport via the bloodstream, thereby facilitating subsequent cellular interaction and transmembrane transport.

3.2. BBB-associated interaction and tumor uptake evaluation of MSCM@MSN@DOX with different morphologies

Since particle morphology may influence the initial contact of nanoparticles with cells, we next investigated whether the morphology-dependent differences were also reflected in cellular uptake. Based on the concentration-dependent cell viability results shown in Fig. 3A and B, 75 μg/mL was selected as an intermediate working concentration between the tested concentrations of 50 and 100 μg/mL. This concentration was chosen to provide sufficient nanoparticle/drug exposure for subsequent uptake and transport measurements while maintaining acceptable viability of bEnd.3 cells. We then evaluated the cellular uptake of the particles in both bEnd.3 and GL261 cells, representing the endothelial and tumor sides of the delivery process, respectively. Fluorescence microscopy (Fig. 3C and D) showed obvious DOX red fluorescence signals in both cell types, confirming efficient cellular association and internalization of the MSCM@MSN@DOX formulations under the tested conditions. From the qualitative results (Fig. 3E and F), in both bEnd.3 cells and GL261 cells, fluorescence signals intensified with increasing nanoparticle aspect ratio, with the long rod-shaped particles (AR∼3) showing significantly higher uptake than the other two groups. These results indicated that particle morphology significantly affected cellular interaction and uptake efficiency at both the endothelial and tumor-cell levels, providing a basis for the subsequent differences in BBB transport and tumor penetration.

Fig. 3.

BBB-associated interaction and tumor uptake evaluation of MSCM@MSN@DOX with different morphologies. Viability of bEnd.3 and GL261 cells after treatment with different formulations at various concentrations (A, B). Representative qualitative visualization of the cellular uptake by bEnd.3 and GL261 cells, respectively (scale bar = 50 μm) (C, D). Quantitative analysis of cellular uptake in bEnd.3 and GL261 cells, respectively (E, F). Schematic illustration of the in vitro monolayer BBB model (G). Fluorescence intensity of MSCM@MSN@DOX detected in the lower chamber of Transwell inserts over time (H). Data are presented as mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001; ns, no significance.

To examine whether the observed morphology-dependent uptake was also present in the absence of membrane coating, we additionally evaluated cellular uptake of morphology-matched uncoated MSN@DOX. The uncoated formulations also exhibited significant morphology-dependent uptake in both bEnd.3 and GL261 cells (Fig. S6 and S7), with the AR∼3 formulation showing significantly higher uptake than the AR∼1 formulation. These findings are consistent with previous studies showing that the geometry of inorganic nanoparticles can influence cellular internalization through differences in particle orientation, membrane-contact geometry, and membrane wrapping [48,49]. Importantly, the present results indicate that morphology-dependent uptake is not unique to the MSC membrane-coated system and that particle geometry itself makes an important contribution to cellular internalization. In the biomimetic formulations, this geometric contribution is retained, while the outer interface additionally contains membrane lipids and retained membrane-associated proteins. Therefore, cellular uptake of MSCM@MSN@DOX may reflect the combined influence of particle geometry and the membrane-derived biological interface rather than a purely membrane-mediated effect.

To further determine whether the morphology-dependent differences in cellular interaction were reflected in trans-BBB transport, we conducted a Transwell assay using an in vitro BBB model (Fig. 3G). The accumulated amount of all three formulations transported across the endothelial barrier into the lower chamber increased over time (Fig. 3H), indicating that MSCM@MSN@DOX with different morphologies could all cross the endothelial barrier to a certain extent. This result also suggests that the retained MSC membrane-associated features may contribute to trans-BBB interaction and transport, possibly through retained adhesion molecules and chemokine receptor-related membrane functions. Among them, the long rod-shaped formulation (AR∼3) exhibited the highest trans-BBB efficiency at all tested time points (2, 4, and 8 h), showing clear morphology-dependent differences in transport efficiency. Combined with its higher uptake in both bEnd.3 and GL261 cells, the AR∼3 formulation displayed a dual advantage at the endothelial and tumor-cell levels, together with the highest trans-BBB transport efficiency, which may be beneficial for brain-targeted delivery against glioma.

To further evaluate whether the morphology-dependent advantages observed above could be maintained in a three-dimensional tumor-like barrier, we established a GL261 multicellular spheroid model and examined the spatial distribution of different formulations by CLSM Z-stack imaging (Fig. 4A). Clear morphology-dependent differences in spheroid penetration were observed (Fig. 4B). In the spherical formulation group (AR∼1), red fluorescence was mainly confined to the peripheral region of the spheroids, and little fluorescence was observed in the inner region as the scanning depth increased (beyond 40 μm), indicating limited penetration into the tumor-like structure. As the aspect ratio increased, the short-rod group (AR∼1.5) showed moderately improved penetration compared with the spherical group, but fluorescence in the deeper layers (40–50 μm) remained relatively weak. In contrast, the long-rod formulation (AR∼3) exhibited the strongest fluorescence intensity across the scanned sections, with pronounced signal still detectable in deeper regions of the spheroids, and the fluorescence was notably stronger than that of the other two groups. The 3D reconstructed images (Fig. 4C) further supported this trend. Compared with the other two groups, the tumor spheroids in the long-rod formulation group (AR∼3) showed stronger red fluorescence intensity and deeper fluorescence penetration, whereas the spherical formulation group (AR∼1) exhibited much weaker fluorescence that was largely confined to the superficial regions of the spheroids. These results indicated that particle morphology played an important role in penetration through dense tumor-like tissue. Notably, the superior penetration of AR∼3 particles was consistent with their enhanced motility observed in multiple-particle tracking, and may also be related to their stronger cellular interaction and trans-BBB transport behavior described above.

Fig. 4.

GL261 multicellular spheroid penetration of MSCM@MSN@DOX with different morphologies. Scheme figure of the in vitro GL261 multicellular spheroid model and CLSM Z-stack imaging (A). Penetration performance of MSCM@MSN@DOX with different morphologies into GL261 multicellular spheroids (Scanning depth: 5–50 μm, scale bar = 250 μm; red fluorescence represents doxorubicin hydrochloride) (B). Corresponding 3D reconstructed images generated from the Z-stack scans of the spheroids (C).

3.3. Shape-dependent nanoparticle–cell interaction and endocytic mechanism

To investigate how particle morphology influences transmembrane uptake during BBB-related transport, we first visualized the internalization process of MSCM@MSN@DOX with different morphologies by bio-TEM. Distinct morphology-dependent differences were observed during the early stages of cell-particle interaction and membrane wrapping (Fig. 5A). Spherical particles were typically associated with relatively localized membrane contact and were often internalized through flask-shaped invaginations enclosing single particles. In contrast, rod-shaped particles more frequently showed extended contact along the cell membrane and more pronounced membrane wrapping or invagination structures during internalization. Depending on the initial contact orientation of the rod-shaped particles, the extent of membrane deformation appeared to vary, suggesting that the early contact mode may influence the subsequent uptake process. These observations suggested that morphology-dependent differences might already emerge at the initial membrane-contact and wrapping stages, potentially contributing to the uptake and transport differences described above.

Fig. 5.

Shape-dependent cell interaction and endocytic mechanism of MSCM@MSN@DOX with different morphologies. Bio-TEM images of nanoparticle uptake by bEnd.3 cells. (scale bars = 500 nm and 1 μm, as indicated). Lower panels show enlarged views of the dashed-box regions (A). Representative SPR binding kinetic curves between MSCM@MSN@DOX and bEnd.3 cells for groups AR∼1 (B), AR∼1.5 (C), and AR∼3 (D), respectively. Quantitative cellular uptake by bEnd.3 cells following treatment with different endocytosis inhibitors for groups AR∼1 (E), AR∼1.5 (F), and AR∼3 (G), respectively. Data are presented as mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, no significance.

Given the morphology-dependent differences in membrane contact and wrapping observed by bio-TEM, we next used surface plasmon resonance (SPR) to further quantify the interaction between MSCM@MSN@DOX and the bEnd.3 cell surface. All three formulations generated measurable binding responses, indicating effective interaction between the MSCM@MSN@DOX formulations and the bEnd.3 cell surface (Fig. 5B, C and D). Notably, the response intensity increased with particle aspect ratio, and the AR∼3 formulation showed the strongest binding signals among the three groups. Kinetic analysis (Table S4) further revealed morphology-dependent differences in binding behavior. Long rod-shaped particles (AR∼3) exhibited the highest association rate constant (K_a) and the lowest dissociation rate constant (K_d), resulting in the lowest equilibrium dissociation constant (K_D), which indicated faster particle-cell binding, greater binding stability, and stronger overall binding affinity. In contrast, the spherical formulation (AR∼1) showed a markedly weaker interaction with the bEnd.3 cell surface, with a K_D of 1.07 × 10⁻⁴ M. These results were consistent with the TEM observations that rod-shaped particles formed more extended membrane contact during internalization, and suggested that particle morphology modulated the initial nanoparticle-cell interface, thereby contributing to the subsequent differences in uptake and trans-BBB transport observed above. Together with the morphology-dependent uptake observed in the uncoated MSN controls, these findings suggest that particle geometry provides an underlying physical contribution to cellular internalization, while the biomimetic membrane interface may further modulate the subsequent nanoparticle–cell interaction and endocytic process.

To further determine whether the morphology-dependent differences in membrane wrapping observed by bio-TEM, together with the differences in membrane contact and binding revealed above, were associated with distinct endocytic pathways, bEnd.3 cells were pretreated with inhibitors of major uptake routes before incubation with MSCM@MSN@DOX. As shown in Fig. 5E, F and G, uptake of all three formulations was markedly reduced under 4 °C treatment, indicating that internalization of all particle morphologies was largely energy-dependent. For the spherical formulation (AR∼1) (Fig. 5E), MβCD treatment caused the most evident reduction in uptake, whereas CPZ and EIPA showed relatively limited effects, suggesting that spherical particles were internalized mainly through lipid raft/caveolae-related pathways. For rod-shaped formulation (AR∼1.5 and AR∼3), the inhibitory effect of MβCD became more pronounced as the aspect ratio increased, while CPZ and low-temperature treatment also significantly reduced uptake, whereas the effect of EIPA remained relatively limited (Fig. 5F and G). These results suggested that uptake of rod-shaped particles still involved lipid raft/caveolin-related pathways, but with increasing aspect ratio, clathrin-mediated endocytosis appeared to contribute more prominently, particularly in the AR∼3 group. These findings were consistent with the morphology-dependent membrane wrapping and invagination patterns observed by bio-TEM. Together with the SPR results, these data further suggested that the initial binding strength between nanoparticles and the bEnd.3 cell surface was likely an important determinant of subsequent endocytic behavior. Particles with stronger binding affinity may form more stable interfacial contact with the cell membrane, thereby influencing membrane wrapping and the kinetics of endocytic entry. Accordingly, long rod-shaped particles, owing to their stronger membrane interaction and more extended contact mode, more readily engaged more complex endocytic pathways, providing a mechanistic basis for their superior uptake and BBB-transport performance.

3.4. In vivo distribution and glioma accumulation of MSCM@MSN@DOX with different morphologies

The in vivo distribution of MSCM@MSN@DOX with different morphologies was evaluated in an intracranial glioma-bearing mouse model to determine whether the morphology-dependent advantages observed in vitro could be reflected in brain/glioma-region accumulation in vivo. After two weeks of GL261 cells inoculation, DiR fluorescently-labeled MSCM@MSN@DOX with three different morphologies were intravenously administered to glioma-bearing mice. Simultaneously, mice without tumor modeling received an injection of DiR fluorescently-labeled MSCM@MSN@DOX (AR∼3) as Sham group.

Cerebral fluorescence intensity in all MSCM@MSN@DOX groups (Fig. 6A–D) showed varying degrees of increase relative to the Sham group, and fluorescence intensity maintained a continuous upward trend within 24 h. This prolonged retention may be related to the presence of the MSC membrane coating, which could help the nanoparticles reduce rapid clearance while preserving membrane-associated functions relevant to BBB crossing and glioma accumulation. In particular, retained membrane proteins such as CD47 may contribute to this behavior by supporting immune evasion-related properties of the biomimetic nanocarriers. The AR∼3 group showed the strongest cerebral fluorescence signal throughout the observation period, suggesting superior in vivo brain/glioma accumulation among the three formulations, which is likely due to stronger penetration resulting from shape effects. Ex vivo imaging at 24 h further supported the in vivo imaging results. Both qualitative and quantitative fluorescence analyses of major organs (Fig. 6B and E) showed that the overall distribution pattern was broadly similar among the three formulations. In contrast, fluorescence analysis of isolated brains indicated that all MSCM@MSN@DOX formulations were able to cross the BBB and accumulate in the glioma region, showing higher fluorescence intensity in brain tissue than that of the sham group (Fig. 6C and F). Among them, the AR∼3 group exhibited the highest brain-associated fluorescence signal, up to 4.014 × 10⁷, indicating superior brain/glioma accumulation. These findings showed that the MSC membrane-coated formulations exhibited favorable brain/glioma-region accumulation, while increasing particle aspect ratio further enhanced accumulation among the investigated biomimetic formulations.

Fig. 6.

In vivo distribution and glioma accumulation of MSCM@MSN@DOX with different morphologies.In vivo brain fluorescence imaging of orthotopic glioma-bearing mice at 4, 6, 8, 12, and 24 h after administration of DiR-labeled MSCM@MSN@DOX (A). Ex vivo fluorescence imaging of major organs and brains (B, C). Quantification of in vivo brain fluorescence intensity over time (D). Quantification of ex vivo organ and brain fluorescence intensity (E, F). Data are presented as mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, no significance versus G4 (long rod-shaped formulation).

3.5. Anti-glioma effect of MSCM@MSN@DOX with different morphologies

The in vivo anti-tumor efficacy of MSCM@MSN@DOX with different morphologies was further evaluated in an orthotopic GL261-Luc glioma model. As shown in the experimental timeline (Fig. 7A), treatment was initiated after intracranial tumor establishment and administered every 3 days for a total of five injections. Bioluminescence imaging performed 7 days after GL261-Luc inoculation confirmed successful tumor establishment in all groups (Fig. 7B). Subsequent imaging conducted on the 14th day revealed a progressive enhancement of glioma bioluminescence across all experimental groups, with average intensity reaching the 10⁷ order of magnitude, indicating continued tumor infiltration and expansion. After completion of the treatment regimen, clear differences in tumor progression emerged among the groups, and the MSCM@MSN@DOX (AR∼3) group exhibited the weakest intracranial bioluminescence signal. Monitoring was continued until the final imaging time point, and quantitative analysis showed that the tumor bioluminescence intensity in the MSCM@MSN@DOX (AR∼3) group decreased by 46.61% on day 28 compared with day 14 (Fig. 7C), indicating sustained inhibitory effects of the long rod-shaped formulation on glioma growth in vivo. Survival was simultaneously monitored to evaluate overall therapeutic benefit. The median survival times were 25.5, 26.5, 28.5, 28.5, and 37.0 days for the saline, free DOX, MSCM@MSN@DOX (AR∼1), MSCM@MSN@DOX (AR∼1.5), and MSCM@MSN@DOX (AR∼3) groups, respectively (Fig. 7D and Table S5). For the predefined comparisons with the saline control, the AR∼3 formulation showed a significant survival benefit after Holm–Šídák correction, whereas the differences for free DOX, AR∼1, and AR∼1.5 did not remain statistically significant after correction. Notably, both AR∼1 and AR∼1.5 showed numerical prolongation of median survival relative to saline (28.5 vs 25.5 days). In the corresponding unadjusted pairwise log-rank analyses, these differences reached statistical significance (AR∼1 vs saline, P = 0.038; AR∼1.5 vs saline, P = 0.019), but they no longer met the significance threshold after correction for the four predefined treatment-versus-saline comparisons. For the second predefined comparison family, the AR∼3 formulation showed significantly prolonged survival compared with every other group, including saline, free DOX, AR∼1, and AR∼1.5, after Holm–Šídák correction. Two mice in the AR∼3 group remained alive at the end of the observation period. Together with the longitudinal bioluminescence and histological findings, these results indicate that the AR∼3 formulation produced the most pronounced overall therapeutic benefit among the investigated formulations.

Fig. 7.

Anti-glioma effect of MSCM@MSN@DOX with different morphologies. Schematic illustration of the in vivo experimental design (A). In vivo bioluminescence imaging of orthotopic glioma-bearing mice at days 7, 14, 21, and 28 post-inoculations (B). Quantification of tumor bioluminescence intensity (n = 3) (C). Kaplan–Meier survival curves of glioma-bearing mice (n = 10) (D). Survival curves were compared using the log-rank (Mantel–Cox) test. Two predefined families of pairwise comparisons were evaluated: each treatment group versus saline, and the AR∼3 formulation versus the other treatment groups. P values were adjusted within each comparison family using the Holm–Šídák method. Representative TUNEL fluorescence, Ki67 immunohistochemistry, and H&E staining of tumor tissues after treatment (scale bar = 50 μm for TUNEL and H&E; scale bar = 20 μm for Ki67) (E). Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

To further assess the anti-glioma effect of MSCM@MSN@DOX, TUNEL staining was performed to evaluate tumor cell apoptosis in brain sections after treatment. As DOX exerts cytotoxicity mainly by inducing DNA damage, increased TUNEL-positive staining reflects enhanced apoptotic response in tumor tissue [50]. TUNEL staining results of mouse brain sections showed that, compared with the saline and free DOX groups, all three MSCM@MSN@DOX formulations increased the number of TUNEL-positive cells to varying degrees, with the MSCM@MSN@DOX (AR∼3) group showing the most pronounced apoptotic signal (Fig. 7E). These findings indicated that all three MSCM@MSN@DOX formulations exerted anti-glioma effects in vivo to varying degrees, while the long rod-shaped formulation showed the most pronounced induction of tumor cell apoptosis, consistent with its stronger inhibition of intracranial tumor progression observed by bioluminescence imaging. Moreover, immunohistochemistry (IHC) was applied for qualitative analysis of Ki67 expression, combined with H&E staining results of tumor regions, to jointly assess drug-mediated inhibition of tumor cell proliferation and assist in evaluating therapeutic efficacy. As Ki67 is a proliferation-associated nuclear marker, reduced Ki67-positive staining indicates suppression of tumor cell proliferation after treatment [51]. The results showed that after completion of the treatment regimen, Ki67-positive cells were reduced to varying degrees in the brain tumors of mice receiving MSCM@MSN@DOX, compared with saline and free DOX control groups. Among the three morphologies, the MSCM@MSN@DOX (AR∼3) group showed the greatest reduction in Ki67-positive staining (Fig. 7E). Together, these results indicated that all three MSCM@MSN@DOX formulations inhibited glioma cell proliferation in vivo to varying degrees, with the long rod-shaped formulation showing the strongest therapeutic effect. This result was consistent with its superior penetration and accumulation behavior observed in the previous experiments.

3.6. In vivo preliminary safety of MSCM@MSN@DOX with different morphologies

To further evaluate the in vivo preliminary safety of MSCM@MSN@DOX with different morphologies, healthy mice were intravenously administered the formulations for five consecutive doses under the same regimen used for efficacy evaluation. Body weight was monitored throughout the study, and blood samples and major organs were collected at the end of the treatment period for biochemical and histopathological analyses (Fig. 8A). During the observation period, the body weight in all three MSCM@MSN@DOX groups remained relatively stable, as well as in the control group (Fig. 8B). Hematoxylin and eosin (H&E) staining of major organs showed no obvious pathological abnormalities after repeated administration of any of the three formulations (Fig. 8G). Furthermore, blood biochemical analysis (Fig. 8C, D, E and F) showed that major hepatic and renal indicators (including alanine aminotransferase, aspartate aminotransferase, urea, and creatinine) were all within normal ranges and showed no significant differences compared with the control group, indicating no obvious adverse effects on hepatic or renal function. Together, these results suggest that MSCM@MSN@DOX with different morphologies did not induce obvious systemic toxicity under the tested dosing conditions.

Fig. 8.

In vivo preliminary safety of MSCM@MSN@DOX with different morphologies. Schematic illustration of the in vivo biosafety assessment (A). Body weight changes during the 15-day treatment period (B). Serum biochemical analysis of aspartate aminotransferase (AST), alanine aminotransferase (ALT), blood urea nitrogen (BUN), and creatinine (CRE) (C–F). H&E staining of major organs and brain (scale bar = 50 μm) (G). Data are presented as mean ± SD, n = 5; ns, not significant.

4. Conclusion

In this study, MSC membrane-coated mesoporous silica nanoparticles with three distinct morphologies (AR∼1, AR∼1.5, and AR∼3) were fabricated and evaluated as shape-defined biomimetic nanocarriers for glioma-targeted drug delivery. Physicochemical characterization confirmed successful membrane coating and retention of key MSC membrane proteins across all morphologies. Comparable drug loading, release profiles, and storage stability provided a relatively controlled basis for assessing shape-dependent delivery behavior.

Multiple-particle tracking analysis indicated that rod-shaped particles, particularly AR∼3, exhibited enhanced motility and diffusion-associated behavior. In vitro studies demonstrated that increasing aspect ratio was associated with improved cellular uptake in bEnd.3 and GL261 cells, higher trans-BBB transport efficiency, and deeper tumor spheroid penetration. Importantly, the additionally evaluated uncoated MSN@DOX series also exhibited significant morphology-dependent cellular uptake, with the AR∼3 formulation showing higher uptake than the AR∼1 formulation in both bEnd.3 and GL261 cells. These results indicate that particle geometry itself contributes importantly to morphology-dependent cellular internalization and that the uptake advantage observed in the biomimetic system cannot be attributed solely to MSC membrane coating. Endocytosis inhibition experiments and bio-TEM observations suggested that morphologically distinct particles may engage different endocytic pathways and membrane-wrapping configurations, contributing to shape-dependent internalization. SPR analysis further suggested stronger and more stable interactions between higher-aspect-ratio particles and bEnd.3 cell-associated surfaces, supporting the hypothesis that contact geometry influences subsequent endothelial interaction and uptake. In vivo experiments demonstrated that, under the conditions used in this study, AR∼3 particles achieved greater brain/glioma-region accumulation, more pronounced tumor growth inhibition, and significantly prolonged survival compared with the other investigated groups, with no overt systemic toxicity observed.

Taken together, these findings indicate that particle morphology represents an important optimization parameter within the investigated MSC membrane-coated MSN system. Nevertheless, several limitations of the present study should be acknowledged. First, the shape-dependent effects were evaluated using only three particle morphologies (AR∼1, AR∼1.5, and AR∼3) within a single mesoporous silica-based, MSC membrane-coated nanocarrier platform. Therefore, the observed relationships should be interpreted within the context of the present system rather than as a universally applicable shape rule for other nanomedicine platforms. Further studies using broader ranges of particle aspect ratios, alternative nanomaterial cores, and different biomimetic membrane systems will be required to determine the generalizability of these findings. Second, although the newly added uncoated controls demonstrate that morphology-dependent uptake can occur independently of MSC membrane coating, the complete coated and uncoated series were evaluated in independent experimental sets. Therefore, the present study does not quantitatively resolve the independent contribution of MSC membrane coating or its potential interaction with particle geometry. Future studies using fully parallel, morphology-matched coated and uncoated formulations will be needed to more precisely distinguish geometry-driven effects from those arising from the biomimetic membrane interface. Third, although the present results support associations among particle morphology, nanoparticle–cell membrane interaction, endocytic behavior, and transport performance, the underlying biophysical mechanisms have not yet been quantitatively resolved. Future work incorporating atomic force microscopy, direct membrane mechanics measurements, and three-dimensional computational simulations will be needed to quantify energy changes associated with membrane deformation and wrapping across particle geometries. Finally, the in vivo findings were obtained using a single orthotopic GL261 glioma model, and further validation in additional glioma models and more clinically relevant experimental settings will be necessary to evaluate the robustness and translational relevance of the observed shape-dependent delivery advantages. Overall, the present study provides system-specific formulation guidance for shape optimization of MSC membrane-coated mesoporous silica nanocarriers for brain-targeted controlled drug delivery.

CRediT authorship contribution statement

Yue Gong: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Hufeng Fang: Investigation, Formal analysis, Data curation. Xinrui Li: Investigation, Formal analysis, Data curation. Sijia He: Project administration, Methodology, Investigation, Formal analysis, Data curation. Zhenwei Ding: Methodology, Investigation, Formal analysis. Yingying Xu: Investigation. Tianyi Zhuang: Investigation. Jiaxin Li: Investigation. Wei Zhao: Investigation. Hongli Liu: Resources, Investigation. Xiang Cao: Investigation, Formal analysis. Kefan Song: Investigation, Formal analysis. Nan Shen: Investigation. Dan Su: Supervision, Resources, Project administration, Funding acquisition. Hongliang Xin: Writing – review & editing, Validation, Supervision, Resources, Project administration, Funding acquisition, Conceptualization. Wei Lv: Resources, Funding acquisition.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Hongliang Xin reports was provided by National Natural Science Foundation of China. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Data will be made available on request.