Sequential-Delivery Gel Microspheres Integrating HAase-Mediated ECM Remodeling and GA-Active Targeting for Deep Tumor Penetration in TACE-Treated HCC

Abstract

Objective

Transcatheter arterial chemoembolization (TACE) for hepatocellular carcinoma (HCC) is limited by inadequate intratumoral drug penetration. We developed a sequential-delivery composite gel microsphere system integrating hyaluronidase (HAase)-mediated microenvironment remodeling, glycyrrhetinic acid (GA)-mediated active targeting, and pH-responsive drug release as a potential strategy to address this barrier.

Methods

Bufalin (BUF)-loaded GA-PEG-HYD-PLGA nanoparticles and HAase were co-encapsulated within carboxymethyl chitosan/sodium alginate gel microspheres containing BaSO₄ for radiopacity (HAase/BUF NPs-GM). The system was characterized for particle size, drug loading, pH-responsive release, and hemocompatibility. Antitumor efficacy, deep tumor penetration, and mechanisms were evaluated in Huh-7 cells and a rabbit VX2 orthotopic liver tumor model via TACE-guided arterial delivery.

Results

HAase/BUF NPs-GM exhibited a favorable size (270 ± 61.89 μm), pH-dependent BUF release, and good biocompatibility, with an IC₅₀ of 55.07 nM against Huh-7 cells. The system reduced hyaluronic acid content across all tumor regions, consistent with enhanced penetration of the therapeutic payload into the tumor core. Compared with controls, HAase/BUF NPs-GM combined with TACE suppressed tumor growth and angiogenesis, induced apoptosis, inhibited proliferation, and was associated with increased intratumoral CD3⁺ T cell infiltration and reduced expression of EMT-related markers. Mechanistically, the system induced ROS overproduction, mitochondrial membrane potential decrease, and G2/M cell cycle arrest.

Conclusion

This strategy, which combines microenvironment remodeling, active targeting, and responsive release, provides a promising preclinical proof-of-concept for addressing the penetration limitations of conventional TACE. Further translational investigation, including validation in orthotopic or patient-derived xenograft models and long-term toxicity assessment, is warranted.

Graphical Abstract

Introduction

Transcatheter arterial chemoembolization (TACE) is a cornerstone treatment for unresectable hepatocellular carcinoma (HCC), delivering locoregional therapy while preserving normal liver parenchyma.^1–3 Clinically, TACE includes conventional TACE (c-TACE) using iodized oil emulsions and drug-eluting bead TACE (DEB-TACE) using preloaded microspheres. Despite widespread use, the objective response rate of DEB-TACE remains suboptimal, ranging from 39.02% to 58% for primary liver cancer and 19.5% to 33.33% for secondary liver cancer.^4,5 A critical factor underlying these limited outcomes is inadequate intratumoral drug penetration.^6,7 c-TACE formulations form mucus-like aggregates that are readily cleared from tumor vasculature, whereas DEB-TACE lacks real-time visualization to ensure optimal microsphere distribution.⁸ In both approaches, drug delivery is largely confined to perivascular regions, leaving tumor cells in deeper, poorly vascularized areas unaffected.^9,10 This penetration deficit contributes to the high recurrence rates after TACE and underscores the need for strategies that enable deep drug transport into tumor parenchyma.¹¹

Beyond inadequate penetration, TACE faces additional limitations, including hypoxia-induced angiogenesis, incomplete embolization due to collateral circulation, and post-embolization syndrome. Among these, inadequate intratumoral drug penetration represents a fundamental physical barrier that limits the efficacy of any TACE-delivered agent, regardless of drug potency. Elevated interstitial fluid pressure (IFP), driven largely by hyaluronic acid (HA) accumulation in the extracellular matrix (ECM), collapses tumor blood vessels and impedes convective transport from the vasculature into the tumor interstitium.^6,7,11–13 According to Starling’s law, once IFP reaches a critical threshold, interstitial fluid flow ceases and drug transport becomes diffusion-limited.^12,13 Degrading HA to normalize IFP is therefore a rational strategy to enhance drug penetration, and we leverage this concept as the first step of our sequential delivery design.¹⁴

To achieve deep tumor penetration after TACE, a delivery system must sequentially overcome multiple biological barriers. Integrating combination therapy with smart embolic materials has been proposed to overcome the therapeutic ceiling of TACE.¹⁵ We therefore designed a composite gel microsphere integrating three synergistic mechanisms in a sequential manner. First, hyaluronidase (HAase) is co-encapsulated to degrade HA, reduce IFP, and facilitate subsequent nanoparticle extravasation and interstitial transport. Second, glycyrrhetinic acid (GA) is conjugated to enable active targeting to HCC cells, which overexpress GA receptors with minimal expression in normal hepatocytes.^16–18 Chen et al showed that GA-modified liposomes exhibited excellent targeting to Hepa1-6 cells with minimal uptake in normal L02 cells,¹⁹ while Wang et al reported GA-based polymeric nanodrugs with 3.0-fold higher uptake in HepG2 cells than in normal hepatocytes.²⁰ These findings, together with studies using self-assembled micelles and glutathione-responsive systems,^21,22 validate GA as an effective ligand for HCC-specific delivery. Third, a pH-responsive hydrazone (HYD) bond is incorporated to trigger drug release within the acidic tumor microenvironment (TME), addressing the suboptimal intracellular release observed in our previous GA-PEG-PLGA nanoparticle. This sequential strategy-HAase-mediated HA degradation, GA-directed active targeting, and HYD-triggered release-is designed to overcome the barriers limiting TACE-based delivery.

As the therapeutic payload, we selected bufalin (BUF), a bioactive compound derived from the traditional Chinese medicine venenum bufonis. BUF exhibits potent antitumor activity across multiple solid tumors, with well-documented efficacy against HCC.^23–26 It is the principal active component of “Huachansu”, a clinically used agent for HCC in China.²⁷ However, its clinical translation is hindered by poor water solubility, short half-life, narrow therapeutic window, and dose-limiting cardiotoxicity.^28–30 Encapsulation within GA-polyethylene glycol (PEG)-HYD-poly(lactic-co-glycolic acid) (PLGA) nanoparticles offers a strategy to overcome these pharmacokinetic limitations while enhancing tumor-specific delivery. BUF was selected as a model payload to validate the proof-of-concept for our sequential-delivery platform, rather than as a replacement for clinically established TACE agents such as doxorubicin or cisplatin. Unlike doxorubicin, which is limited by cumulative cardiotoxicity and multidrug resistance, or irinotecan, which shows significant interpatient pharmacokinetic variability, BUF has a well-defined pharmacological profile with documented anti-HCC activity and the ability to overcome multidrug resistance. Its challenging physicochemical properties make it an ideal stress test for our delivery platform. Three considerations guided this selection. First, BUF’s poor solubility, short half-life, and cardiotoxicity provide a stringent test of the platform’s delivery capability. Second, BUF’s anti-HCC activity and ability to overcome multidrug resistance make it relevant for patients with prior chemotherapy exposure. Third, the platform is designed to be payload-flexible; successful proof-of-concept with BUF supports future application with clinically established TACE agents.

In this study, we engineered a composite gel microsphere encapsulating BaSO₄, HAase, and BUF-loaded GA-PEG-HYD-PLGA nanoparticles (BUF NPs) within carboxymethyl chitosan/sodium alginate (CMCS/ALG) gel microspheres (BaSO₄-CMCS/ALG@GA-PEG-HYD-PLGA-BUF/HAase, HAase/BUF NPs-GM or GA dual load). While some commercial DEB-TACE microspheres incorporate radiopaque elements, the CMCS/ALG matrix used here is inherently non-radio-opaque. BaSO₄ incorporation enables real-time DSA-guided visualization of microsphere distribution during TACE, addressing the lack of intraprocedural imaging feedback in conventional DEB-TACE. By combining this system with TACE, we aim to achieve: (1) HAase-mediated HA degradation to facilitate deep penetration, (2) GA-mediated active targeting to HCC cells, (3) HYD-based pH-responsive drug release within the TME, and (4) enhanced antitumor efficacy with reduced systemic toxicity. We evaluated this strategy in a rabbit VX2 hepatic tumor model, a clinically relevant orthotopic model that recapitulates human vascular anatomy and TACE procedural conditions. While the present study is preclinical, the proposed platform is conceptually positioned as a potential next-generation DEB-TACE adjunct. Rather than replacing existing embolic agents or chemotherapeutics, it could complement current TACE regimens by adding microenvironment-modulating and active-targeting capabilities to existing embolic microsphere technologies.

While HAase-mediated ECM remodeling, GA-based active targeting, and pH-responsive drug release have each been explored separately, their integration within a single TACE-deliverable microsphere platform has not been previously reported. This combinatorial approach is designed to address the sequential physical barriers that limit conventional TACE in a coordinated manner, representing the core innovation of this study. Recent advances in TME-responsive nanomedicine have demonstrated strategies such as enzyme-triggered transcytosis, size-switchable nanoparticles, hypoxia-responsive systems, stepwise-responsive Au-based delivery systems, sequential-mediated hepato-escape lipid nanoparticles, and TME-responsive polymer nano-immunomodulators. However, most of these approaches focus on single mechanisms or are designed for systemic administration, which is limited by low tumor accumulation. Our work distinguishes itself by integrating three complementary mechanisms within a single TACE-deliverable microsphere platform, enabling sequential overcoming of the physical barriers that limit conventional TACE.

Materials and Methods

Materials

BUF was purchased from Chengdu Herbpurify Co., Ltd (Chengdu, China). ALG and CMCS were obtained from Shanghai Maclin Biochemical Technology Co., Ltd (Shanghai, China). Huh-7 hepatocellular carcinoma cells (BNCC337690) and normal THLE-2 liver cells (CRL-2638™) were purchased from BeNa Culture Collection and ATCC, respectively. Huh-7 cells were authenticated by STR profiling upon receipt and confirmed mycoplasma-negative by PCR. All experiments used cells between passages 5 and 10. New Zealand White rabbits (2.0–2.5 kg) were supplied by Jiashan Shengwang Ecological Farm.

Synthesis and Characterization of GA-PEG-HYD-PLGA

GA-PEG-HYD-PLGA was synthesized via three sequential steps (Figure 1A). First, NH₂-PEG-NH₂ (1.0 g), GA, 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC, 2.0 eq), and 4-Dimethylaminopyridine (DMAP, 0.1 eq) were reacted in chloroform at room temperature for 12 h. The resulting GA-PEG-NH₂ was isolated by rotary evaporation, ether precipitation, filtration, and drying (yield: 82.3%). Second, OH-HYD-COOH (72 mg), lactide (1.58 g), glycolide (0.42 g), and stannous octoate were heated in anhydrous toluene at 130 °C for 12 h under N₂ to obtain PLGA-HYD-COOH (yield: 86.5%). Finally, PLGA-HYD-COOH (1.0 g) was reacted with GA-PEG-NH₂ (1.1 eq), EDC (2.0 eq), and DMAP (0.1 eq) in chloroform at room temperature for 24 h. After solvent removal, the residue was dialyzed (MWCO 3500 Da) for 48 h and lyophilized to yield GA-PEG-HYD-PLGA (yield: 58.2%). Purity of GA-PEG-NH₂, PLGA-HYD-COOH, and the final conjugate was analyzed by HPLC (Shimadzu LC-20AD) with an ELSD detector (Shimadzu ELSD-LT II) using a C18 column (4.6 × 250 mm, 5 μm) and gradient elution of water/acetonitrile at 1.0 mL/min; purity was >95% by area normalization. Structure was confirmed by ¹H NMR (Bruker Avance-400) in CDCl₃, FT-IR (Thermo Scientific Nicolet 6700), and GPC (Prominence Prep LC-20A). The final product was stored as a solid powder at −20 °C under dry, dark conditions.

Figure 1.

Synthesis and characterization of PLGA-HYD-PEG-GA. (A) Synthetic route. (B) ¹H NMR spectrum of PLGA-HYD-PEG-GA in CDCl₃. a: PLA methine peak (–CH–); b: PGA methylene peak (–CH₂–); c: PEG methylene peak (–CH₂CH₂O–); d: PEG methoxy peak (–OCH₃); e: PLA methyl peak (–CH₃); f: characteristic peak of glycyrrhetinic acid. (C) FT-IR spectra of PLGA-HYD-COOH and PLGA-HYD-PEG-GA. (D) GPC chromatograms showing the molecular weight shift after GA conjugation.

Preparation of BUF NPs and HAase/BUF NPs-GM

BUF NPs were prepared by oil-in-water emulsion solvent evaporation and stored at 4 °C. HAase/BUF NPs-GM were fabricated using a high-voltage pulse microencapsulation apparatus. Briefly, CMCS and ALG were dissolved in deionized water at an optimized mass ratio, followed by addition of Na₂SO₄ and BUF NPs to form the dispersed phase. BaCl₂ aqueous solution was used as the receiving phase. The dispersed phase was electrosprayed under controlled flow rate and current intensity to obtain BaSO₄-CMCS/ALG@BUF NPs, which were washed three times with ultrapure water and dried at 50 °C for 12 h under vacuum. The resulting microspheres were then incubated with HAase (1:2 mass ratio, BUF NPs: HAase) at 4 °C for 12 h under gentle agitation (300 rpm). The conjugate was dialyzed (MWCO 14 kDa, 24 h) and lyophilized to obtain HAase/BUF NPs-GM.

Characterization of BUF NPs and HAase/BUF NPs-GM

BUF NPs were characterized for particle size, polydispersity index (PDI), and zeta potential (Malvern Nano-ZS90); morphology by transmission electron microscopy (TEM, JEM 2100F); encapsulation efficiency (EE) and drug loading content (LC) of BUF and HAase; and in vitro drug release at pH 4.5, 6.5, and 7.4.

HAase/BUF NPs-GM were characterized for particle size distribution (Nikon TI-s inverted microscope) and morphology by SEM (Hitachi SU8010). EE and LC were quantified for BUF and HAase, and retained HAase activity was measured. BaSO₄ crystal structure was analyzed by XRD (Rigaku Ultima IV). Mechanical properties (hardness, resilience, springiness) were tested using an FTC TMS-PRO texture analyzer. Real-time visualization was confirmed by digital subtraction angiography (DSA, Siemens Artis Q). In vitro release profiles at different pH values and hemocompatibility were also evaluated.

In vitro release studies were conducted using a dialysis bag method (MWCO 14 kDa) at 37 ± 0.5 °C in PBS at pH 7.4, 6.5, and 4.5. Buffer solutions were prepared according to the Chinese Pharmacopoeia: pH 7.4 and 6.5 buffers mainly contain potassium dihydrogen phosphate and sodium hydroxide, while pH 4.5 buffer mainly contains sodium acetate and glacial acetic acid. Briefly, 2 mL of BUF NPs or HAase/BUF NPs-GM dispersion was placed into a dialysis bag and immersed in 100 mL of release medium under continuous stirring at 100 rpm. At predetermined time points, 1 mL of release medium was withdrawn and replaced with an equal volume of fresh pre-warmed PBS to maintain constant volume and sink conditions. BUF concentration was quantified by HPLC (Agilent 1200) using a C18 column (Agilent Zorbax Extend-C18, 4.6×250 mm, 5 μm) with methanol/water (67:33, v/v) at 1.0 mL/min. Detection wavelength was 296 nm, and column temperature was 30 °C. The method was validated for linearity (R² > 0.999), precision (RSD < 2%), and accuracy (recovery: 98–102%).

For hemocompatibility, each formulation (free BUF, BUF NPs, blank gel microspheres, GA single load, and GA dual load) was mixed with RBC (2%, v/v) suspension and incubated at 37 °C for 2 h. 0.9% sodium chloride solution served as the negative control (0% hemolysis) and deionized water as the positive control (100% hemolysis). After incubation, samples were centrifuged at 1006g for 5 min, and the absorbance of the supernatant was measured at 590 nm using a microplate reader. Hemolysis (%) was calculated as: [(Asample − Anegative) / (Apositive − Anegative)] × 100%.

Intracellular Uptake of HAase/BUF NPs-GM

To evaluate HAase-mediated enhancement of cellular uptake and the hepatoma-targeting ability of GA-decorated nanoparticles, Nile Red (NR) was used as a fluorescent tracer in place of BUF. Huh-7 cells were treated with free NR, NR NPs, BaSO₄-CMCS/ALG@GA-PEG-HYD-PLGA-NR (GA single load (NR)), BaSO₄-CMCS/ALG@GA-PEG-HYD-PLGA-NR/HAase (GA dual load (NR)), BaSO₄-CMCS/ALG@PEG-HYD-PLGA-NR/HAase (No GA dual load (NR)), or BaSO₄-CMCS/ALG@PEG-PLGA-NR/HAase (No GA and HYD dual load (NR)) at an NR concentration of 150 μM. After 2–72 h of incubation, cellular fluorescence intensity was quantified by flow cytometry (FCM, Beckman Coulter DxFLEX).

In vitro Antitumor Activity and Toxicity Evaluation

Huh-7 cells were treated with free BUF, BUF NPs, GA single load, GA dual load, No GA dual load, or No GA and HYD dual load at BUF concentrations of 0, 2.5, 25, 50, 100, 150, and 200 nM for 72 h. Four hours before the end of incubation, 20 μL of MTT (5 mg/mL) was added to each well. After incubation, the supernatant was aspirated, and 150 μL of DMSO was added to dissolve formazan crystals under shaking for 10 min. Absorbance at 570 nm was measured using an ELISA reader (LB941, Berthold) to determine IC₅₀ values. For toxicity evaluation in normal cells, THLE-2 cells were co-incubated with control, BUF NPs, GA single load, or GA dual load at the IC₅₀ concentration of BUF for 72 h, after which cell viability was assessed.

In vivo Antitumor Effect and Toxicity

Thirty New Zealand White rabbits were fasted for 6 h (water ad libitum) before surgery and anesthetized with intravenous sodium pentobarbital (30 mg/kg, 3% solution) via the marginal ear vein. Anesthesia depth was monitored by loss of pedal withdrawal reflex and respiratory rate. Supplemental doses (one-third of the initial dose) were given intravenously as needed to maintain surgical anesthesia. The subxiphoid abdomen was shaved, disinfected with povidone-iodine, and draped sterilely. A 3-cm midline laparotomy was performed to implant a 0.2 cm³ VX2 tumor fragment into the left hepatic lobe via trocar guidance. The abdominal wall was closed in layers, and prophylactic intramuscular streptomycin (2 × 10⁵ IU) was given. Tumor engraftment was confirmed by ultrasonography on postoperative day 10.

The rabbit VX2 hepatic tumor model was selected because it is the most established large animal platform for TACE research, providing sufficient vascular caliber for human-compatible interventional catheters and DSA-guided superselective arterial delivery. While VX2 is not of hepatocellular origin, it shares key physical tumor microenvironment features-including elevated IFP, HA accumulation, hypoxia, and hypervascularity-that are relevant to the drug delivery barriers investigated here.

TACE was performed under DSA guidance. After standard preparation, the right femoral artery was exposed via layered dissection under general anesthesia. Arterial access was achieved with an 18G needle, followed by insertion of a 4-F vascular sheath. A 2.5-F coaxial microcatheter was advanced via a microguidewire to superselectively catheterize the hepatic artery. Angiography was correlated with preoperative CT to localize tumor lesions and vasculature, enabling superselective cannulation of tumor-feeding arteries.

Prior to treatment allocation, tumor volumes were measured by ultrasonography on day 10 post-implantation. Rabbits were randomly assigned using stratified randomization based on baseline tumor size. No significant differences in baseline tumor volumes were observed among groups (p > 0.05). Rabbits were assigned to five groups (n = 6 per group): normal saline (NS), free BUF, GA single load, GA dual load, and No GA dual load. The BUF dose was 2 mg/kg (except for the NS group). These five groups permitted two independent single-variable comparisons: (i) GA dual load versus GA single load, differing only in co-encapsulated HAase, allowing assessment of HAase-mediated ECM remodeling; and (ii) GA dual load versus No GA dual load, differing only in GA decoration, allowing assessment of GA-mediated active targeting. The free BUF and NS groups served as additional reference controls. The free BUF group received blank gel microspheres after free BUF administration. Blank microspheres were prepared identically to drug-loaded microspheres, with the same BaSO₄ content, particle size, and mechanical properties, ensuring comparable embolization across all treatment groups. All non-NS groups received the same embolization procedure via superselective hepatic artery catheterization. The NS group received normal saline alone without embolization. Embolization efficacy, radiopacity, and tumor volume were monitored by DSA and ultrasonography. Body weight and tumor volume were recorded periodically. Tumor volume measurements and histopathological evaluations were performed by investigators blinded to treatment allocation.

Sample size (n = 6 per group) was determined based on preliminary experiments and published VX2 TACE studies, which typically employ 5–8 animals per group to achieve sufficient statistical power (≥80%) for detecting biologically meaningful differences in tumor growth inhibition. The sample size was also constrained by ethical considerations and animal availability. Individual animal tumor volume data were collected for all rabbits and used to calculate group means and standard deviations.

On day 28 post-treatment, all rabbits were euthanized by intravenous injection of an overdose of sodium pentobarbital (150 mg/kg) via the marginal ear vein, followed by exsanguination. Death was confirmed by cardiac arrest, cessation of respiration, and bilateral mydriasis. Tumors were immediately harvested. Blood samples were collected via cardiac puncture prior to euthanasia. Necrosis was assessed by H&E staining, and apoptosis by TUNEL assay. Heart, liver, spleen, lung, and kidney tissues were collected and analyzed for in vivo toxicity using H&E staining. The ARRIVE 2.0 Essential 10 checklist is provided as Supplementary Material.

Deep Penetration of HAase/BUF NPs-GM

After pharmacological experiments, liver and tumor tissues were collected and split into two portions: Tissue I (stored at −80 °C for HA quantification) and Tissue II (frozen sections for penetration assessment). For Tissue I, samples from three regions (tumor margin, intratumoral, tumor core) were homogenized in lysis buffer, centrifuged (2500 rpm, 5 min), and supernatants used to quantify HA levels. For Tissue II, frozen sections were incubated with rabbit anti-CD31 primary antibody, followed by CoraLite488-labeled goat anti-rabbit IgG secondary antibody, counterstained with DAPI, and imaged by confocal laser scanning microscopy (CLSM, Leica STELLARIS 8 DIVE) to evaluate penetration depth, spatial distribution, and correlation with vascular architecture. Additional sections were stained with DAPI (room temperature, 10 min), washed with PBS, and examined by fluorescence microscopy. Fluorescence intensity was quantified to assess drug delivery efficiency in the three tumor regions.

Internalization Pathways

Huh-7 cells were pre-incubated with specific endocytosis inhibitors (free GA, 10 μM; 2-deoxyglucose, 20 mM; chlorpromazine, 10 μM; hexadimethrine, 20 μM; genistein, 50 μM; methyl-β-cyclodextrin, 5 mM) for 1 h, followed by the addition of HAase/BUF NPs-GM for 72 h. Cellular uptake of GA dual load was then analyzed by FCM.

Intracellular Localization

Cells were seeded in confocal dishes and treated with free NR, NR NPs, GA single load (NR), GA dual load (NR), No GA dual load (NR), or No GA and HYD dual load (NR). After 72 h incubation, intracellular drug distribution was analyzed by CLSM following staining with DIO, Lyso-Tracker, and DAPI.

Mechanism of Antitumor Effect in vitro

Huh-7 cells were treated for 72 h with control, free HAase + BUF NPs, GA dual load, No GA dual load, or No GA and HYD dual load. After incubation, cells were stained with 10 mM DCFH-DA for 20 min and analyzed by CLSM to assess intracellular reactive oxygen species (ROS). Mitochondrial membrane potential was measured by JC-1 staining and FCM. Apoptosis and cell cycle distribution were evaluated by Annexin V-FITC/PI dual staining.

Mechanism of Antitumor Effect in vivo

Frozen sections of Tissue II were immunohistochemically stained for proliferating cell nuclear antigen (PCNA), hypoxia-inducible factor 1-alpha (HIF-1α), Vimentin, N-cadherin, and CD3⁺ T cells. Positive signals were quantified using image analysis software to assess tumor proliferation, metastatic potential, and immune cell infiltration.

Statistical Analysis

Data are presented as mean ± SD. All statistical analyses were performed using SPSS software (version 19, IBM Inc., USA). Normality was assessed using the Shapiro–Wilk test, and homogeneity of variance using Levene’s test. For multiple-group comparisons, one-way ANOVA followed by the LSD post-hoc test was used. For two-group comparisons, Student’s t-test was used. When data did not meet assumptions of normality or equal variance, non-parametric tests were used instead. Comparisons were pre-planned: each treatment group was compared with the NS group, and the two mechanistic comparisons were specified a priori. A sensitivity analysis using Bonferroni correction for these pre-planned comparisons yielded conclusions consistent with the primary analysis. Statistical significance was defined as p < 0.05, and p < 0.01 was considered highly significant.

Sample size (n = 6 per group) was determined based on preliminary experiments and published VX2 TACE studies, which typically employ 5–8 animals per group to achieve sufficient statistical power (≥80%) for detecting biologically meaningful differences in tumor growth inhibition. The sample size was also constrained by ethical considerations and animal availability. Individual animal tumor volume data were collected for all rabbits and used to calculate group means and standard deviations.

Predefined exclusion criteria were established prior to the study. Animals were excluded if they met any of the following criteria: (1) failure of tumor engraftment confirmed by ultrasonography on day 10 post-implantation; (2) death during the TACE procedure or within 48 h post-procedure due to procedural complications unrelated to tumor burden; (3) severe postoperative complications requiring euthanasia before the scheduled endpoint; (4) technical failure of superselective hepatic artery catheterization preventing complete agent delivery. For in vitro experiments, data points were excluded if they deviated more than two standard deviations from the mean of replicate measurements, or if technical failure of the assay was documented. All exclusions were pre-specified and applied consistently across groups.

Results

Characterization of GA-PEG-HYD-PLGA

GA-PEG-HYD-PLGA was successfully synthesized and characterized. ¹H NMR (Figure 1B) confirmed GA conjugation to PEG, with characteristic peaks of PEG/PLGA and no impurities: δ 7.90/7.71 (semicarbazide bond), δ 5.22/1.57 (PLA-CH-/CH₃), δ 4.83 (PGA-CH₂-), δ 3.64 (PEG-CH₂-), and δ 0.81–1.37 (GA angular-CH₃-/CH₂-). FT-IR spectra (Figure 1C) of GA-PEG-HYD-PLGA versus PLGA-HYD-COOH showed amide I (1653.00 cm⁻¹) and II (1552.70 cm⁻¹) bands, confirming conjugation of PLGA-HYD-COOH to the terminal amine of GA-PEG. GPC further corroborated formation, with the molecular weight of PLGA-HYD-PEG increasing by approximately one GA unit (Figure 1D). These data confirm the successful synthesis of GA-PEG-HYD-PLGA.

Characterization of BUF NPs and HAase/BUF NPs-GM

BUF NPs were spherical with a mean diameter of 145 ± 3.35 nm, a PDI of 0.16 ± 0.03, and a zeta potential of −15.4 ± 2.1 mV (Figure 2A). EE and LC of BUF in the NPs were 82.86 ± 3.54% and 55.28 ± 2.35%, respectively. In vitro release showed pH-dependent behavior: slow release at pH 7.4, with accelerated release at pH 6.5 and 4.5 due to HYD bond cleavage (Figure 2B), consistent with previous reports.^31,32

Figure 2.

Characterization of BUF NPs and HAase/BUF NPs-GM. (A) TEM images of BUF NPs. (B) In vitro BUF release from BUF NPs at pH 4.5, 6.5, and 7.4. (C) TEM and (D) SEM images of HAase/BUF NPs-GM. (E) XRD pattern of HAase/BUF NPs-GM. The upper red curve is the experimental pattern; the middle black vertical lines and lower green curve are the standard reference pattern of orthorhombic BaSO₄ (Barite, ICDD/JCPDS No. 00–024-1035). Major experimental peaks are labeled with their d values and match the reference Bragg positions; no additional crystalline phases were detected. (F) DSA images of HAase/BUF NPs-GM. (G) In vitro BUF release from HAase/BUF NPs-GM at pH 4.5, 6.5, and 7.4. (H) Hemocompatibility of HAase/BUF NPs-GM.

HAase/BUF NPs-GM were spherical with a diameter of 270 ± 61.89 μm (Figure 2C) and a rough, granular surface with variable pores (Figure 2D). Similar alginate/chitosan-based microspheres have been reported for local drug delivery in solid tumors.³³ LC and EE were 24.87 ± 2.62% and 38.93 ± 4.10% for BUF, and 55.86 ± 3.24% and 37.24 ± 4.35% for HAase, respectively. Retained HAase activity was 6540.77 ± 45.52 IU/L. The swelling ratio in PBS (pH 7.4) at 37 °C was 40.64 ± 2.16% after 6 h, indicating moderate swelling suitable for TACE embolization. XRD confirmed the presence of orthorhombic BaSO₄ (Figure 2E), with major diffraction peaks matching the standard reference pattern (Barite, ICDD/JCPDS No. 00–024-1035) and no additional crystalline phases. Texture analysis showed hardness 7.23 ± 0.11 N, resilience 0.41 ± 0.04, and springiness 0.24 ± 0.02 mm, meeting the mechanical requirements for TACE. DSA confirmed real-time embolization monitoring (Figure 2F). The microspheres showed pH-dependent BUF release, with accelerated release under acidic conditions (Figure 2G). Hemolysis rates for free BUF, BUF NPs, blank microspheres, GA single load, and GA dual load were 1.36 ± 0.13%, 0.27 ± 0.04%, 0.54 ± 0.05%, 0.81 ± 0.18%, and 0.41 ± 0.39%, respectively, all below the 5% safety threshold (Figure 2H). These results indicate that HAase/BUF NPs-GM possess the key attributes required for TACE delivery: suitable size and mechanical properties, radiopacity, pH-responsive release, and favorable biocompatibility.

Intracellular Uptake

Cells treated with GA dual load or No GA dual load showed significantly stronger fluorescence than free NR-treated cells (p < 0.01; Figure 3A and B). Fluorescence intensity was higher in HAase-loaded and GA-decorated groups than in non-loaded or non-decorated controls at all time points. At 72 h, the GA dual-loaded system showed approximately one order of magnitude (roughly 16-fold or greater) higher mean fluorescence intensity than the GA single-load system (Figure 3B), indicating a pronounced non-additive enhancement rather than simple additive targeting. These findings indicate that HAase-mediated HA degradation enhances nanoparticle accessibility to tumor cells, while GA modification potentiates target specificity and cellular uptake in HCC cells, consistent with previous reports.^33–35

Figure 3.

In vitro assessment of HAase/BUF NPs-GM. (A) Flow cytometry analysis of Huh-7 cell uptake after incubation with the indicated formulations for 2, 6, 12, 24, 48, and 72 h. (B) Quantification of mean fluorescence intensity from (A). (C) Cell viability of Huh-7 cells treated with free BUF, BUF NPs, GA single load, GA dual load, No GA dual load, or No GA and HYD dual load at BUF concentrations of 0–200 nM for 72 h. (D) IC₅₀ values derived by nonlinear fitting. (E) Viability of normal THLE-2 cells treated with BUF NPs, GA single load, or GA dual load at 104.97 nM BUF for 72 h. Data are mean ± SD from three independent experiments (n = 3). **p < 0.01, *p < 0.05 vs the indicated group (one-way ANOVA with LSD post-hoc test, SPSS version 19).

In vitro Antitumor Activity and Toxicity Evaluation

Free BUF, BUF NPs, and four nanoparticle-loaded microsphere formulations all showed dose-dependent antitumor activity against Huh-7 cells. The GA dual load group showed lower cell viability and a reduced IC₅₀ (55.07 nM) compared with non-HAase-loaded controls (eg, GA single load, 79.68 nM; Figure 3C and D). At the IC₅₀ concentration of BUF (104.97 nM), none of the formulations significantly affected the viability of normal THLE-2 cells (Figure 3E). The GA dual load group exhibited the lowest IC₅₀ (55.07 nM), approximately 2-fold lower than free BUF (104.97 nM) and 1.45-fold lower than the GA single load group (79.68 nM). This reduction is consistent with the combined contributions of nanoencapsulation, GA-mediated active targeting, pH-responsive HYD cleavage, and HAase-mediated ECM remodeling, as supported by the enhanced cellular uptake (Figure 3A and B), pH-dependent release (Figure 2B and G).

In vivo Antitumor Effect and Toxicity

Tumor-bearing rabbits were confirmed for successful VX2 liver tumor establishment by ultrasound 10 days post-implantation. Baseline tumor volumes were 0.32 ± 0.08, 0.31 ± 0.09, 0.33 ± 0.07, 0.32 ± 0.08, and 0.30 ± 0.09 cm³ for the NS, free BUF, GA single load, GA dual load, and No GA dual load groups, respectively, with no significant differences among groups (p > 0.05). TACE was performed via femoral artery access under DSA guidance, followed by 18-day efficacy monitoring (Figure 4A–D). The 18-day follow-up (day 10 to day 28 post-implantation) was selected based on preliminary experiments and published VX2 TACE studies, which indicate that this duration is sufficient to observe significant tumor growth inhibition while minimizing animal suffering. Ultrasound-derived tumor volume changes (Figure 5A and B) showed progressive tumor growth in the NS group, whereas all treatment groups demonstrated antitumor activity. The GA dual load group showed significant reductions in tumor volume (p < 0.01) and tumor weight (p < 0.05; Figure 5C–E).

Figure 4.

Flowchart of the in vivo study and TACE procedure. (A) Tumor engraftment was confirmed by ultrasonography on day 10 post-implantation, followed by TACE treatment and an 18-day follow-up. (B) Schematic of the TACE procedure, including femoral artery catheterization and microcatheter delivery into the target vessel, followed by administration of each formulation via the left hepatic artery. (C) Femoral artery isolation and insertion of a 4-F catheter sheath. After TACE, embolization of the tumor-feeding artery was verified by DSA before administration of the delivery system. (D) DSA image; yellow arrow indicates the left hepatic artery, and blue arrow indicates tumor blood vessels.

Figure 5.

In vivo antitumor efficacy. (A) Representative ultrasound images of each group before and after TACE. (B) Representative images of VX2 tumors on day 18 post-TACE; each grid = 1 cm × 1 cm. (C–E) Tumor volume, tumor weight, and body weight after treatment. Data are mean ± SD (n = 6 per group). ^#p < 0.05, ^##p < 0.01 vs NS group (one-way ANOVA with LSD post-hoc test, SPSS version 19).

Liver function parameters (ALT, AST, ALP, TBil) revealed significant injury in the free BUF group compared with NS, whereas no differences were observed in the three delivery system groups (Figure 6A–D). Renal function (Cr, BUN) remained unchanged across all groups (Figure 6E and F). Cardiac function markers (CK, CK-MB, LDH) showed significant toxicity in the free BUF group but not in the delivery system groups (p < 0.01), indicating that encapsulation within the microsphere system mitigated BUF-induced liver and cardiac damage (Figure 6G–I).

Figure 6.

Serum biochemical markers. (A–I) Levels of ALT (A), AST (B), ALP (C), TBiL (D), Cr (E), BUN (F), CK (G), CK-MB (H), and LDH (I) in different groups. Data are mean ± SD (n = 6 per group).^##p < 0.01 vs NS group (one-way ANOVA with LSD post-hoc test, SPSS version 19).

Histopathological analysis (Figure 7A) revealed minimal tumor damage in the NS group, whereas the three delivery system groups exhibited larger necrotic areas than the free BUF group, with the GA dual load group showing the most pronounced necrosis. The free BUF group displayed severe myocardial damage, while delivery system groups showed reduced cardiotoxicity, with the GA dual load group exhibiting no significant myocardial damage. The GA dual load group also reduced lung metastasis. TUNEL staining (Figure 7B and C) demonstrated significantly higher apoptosis rates in the GA dual load and No GA dual load groups versus NS, with the GA dual load group showing the highest apoptosis rate. These findings indicate that the composite gel microsphere system enhances antitumor efficacy through sustained and targeted delivery while mitigating the off-target toxicities, particularly cardiotoxicity, that limit free BUF. Within the 18-day observation period, encapsulation reduced the acute and subacute cardiac and hepatic toxicity of BUF; however, this window is insufficient to assess delayed toxicity, cumulative toxicity upon repeated TACE, or late-onset organ injury.

Figure 7.

Histopathological analysis and TUNEL staining. (A) H&E staining of tumor and major organs in each group after TACE. (B) TUNEL staining and (C) quantification of TUNEL-positive rate. Data are mean ± SD (n = 6 per group). ^##p < 0.01 vs NS group (one-way ANOVA with LSD post-hoc test, SPSS version 19).

To dissect the individual contributions of HAase and GA, two single-variable comparisons were performed. The GA dual load group showed significantly smaller tumor volume and tumor weight than the GA single load group (Figure 5C and D), indicating a measurable contribution of HAase-mediated ECM remodeling. In contrast, the difference between the GA dual load and No GA dual load groups was modest and did not reach statistical significance at the whole-tumor level, suggesting that under TACE-based arterial delivery with co-encapsulated HAase, the contribution of GA-mediated active targeting to bulk tumor growth inhibition was less pronounced than that of HAase. The contribution of GA was more evident at the cellular level, where GA-decorated systems showed significantly higher uptake than non-GA-decorated controls (Figure 3A and B).

Deep Penetration of HAase/BUF NPs-GM

Figure 8A depicts HA content in tumor edge, middle, and core regions across groups. Compared with NS, no significant differences in HA levels were observed in the free BUF or GA single load groups. In contrast, the GA dual load group exhibited markedly reduced HA content across all tumor regions compared with the No GA dual load group (p < 0.01).

Figure 8.

Deep penetration of HAase/BUF NPs-GM. (A) HA content in tumor edge, middle, and core regions across groups. (B) Representative immunofluorescence images of VEGF and CD31 in tumor tissues (magnified view, 5×, of the boxed region). (C) DAPI-positive signals in tumor edge, middle, and core regions. Data are mean ± SD (n = 6 per group). ^#p < 0.05, ^##p < 0.01 vs NS group (one-way ANOVA with LSD post-hoc test, SPSS version 19).

Figure 8B shows vascular endothelial growth factor (VEGF) and cluster of differentiation 31 (CD31) immunofluorescence in tumor tissues. All three delivery system groups exhibited significantly reduced VEGF-positive fluorescence compared with NS (p < 0.01), while both the free BUF and delivery system groups demonstrated diminished CD31-positive fluorescence (p < 0.01). The GA dual load group exerted stronger inhibitory effects on VEGF and CD31 expression than the GA single load and No GA dual load groups.

To assess the spatial extent of drug efficacy, we examined DAPI positivity across tumor edge, middle, and core regions (Figure 8C). DAPI is a membrane-impermeant dye that stains only cells with compromised plasma membranes (ie, dead or dying cells); viable cells with intact membranes remain DAPI-negative. Therefore, DAPI-positive signals indicate regions of drug-mediated cell killing rather than serving as a direct tracer for nanoparticle or BUF distribution. Compared with NS (minimal DAPI positivity), the free BUF group showed significant DAPI-positive signals only at the tumor edge (p < 0.05), indicating that drug efficacy was confined to perivascular regions. The GA single load group showed DAPI positivity at the edge and middle (p < 0.01, p < 0.05), suggesting partial penetration. In contrast, both HAase-containing groups exhibited significant DAPI-positive signals across all three regions (p < 0.01), indicating that drug-mediated cell killing reached the tumor core. These findings are consistent with the conclusion that HAase-mediated ECM remodeling facilitated deep delivery of the therapeutic payload and its cytotoxic effect. Direct visualization of nanoparticle distribution by confocal z-stacking with depth profiling was not performed and is planned for future work.

Internalization Pathways

Figure 9A and B show that chlorpromazine treatment had no significant effect on fluorescence intensity compared with the GA dual load group. In contrast, pre-incubation with GA, 2-deoxyglucose, wortmannin, genistein, and methyl-β-cyclodextrin significantly reduced fluorescence intensity, with methyl-β-cyclodextrin showing the most pronounced decrease. The lack of effect with chlorpromazine indicates that clathrin-mediated endocytosis is not involved in the uptake of GA-modified NPs. The significant reduction following free GA pretreatment suggests that internalization occurs via specific interaction with GA receptors on HCC cells. The inhibitory effects of 2-deoxyglucose (energy depletion), wortmannin (macropinocytosis inhibitor), genistein (caveolae-mediated endocytosis inhibitor), and methyl-β-cyclodextrin (lipid raft disruptor) collectively suggest that internalization occurs via energy-dependent, caveolae-mediated endocytosis and macropinocytosis rather than clathrin-mediated pathways. This uptake mechanism may facilitate subsequent transcytosis and deep tumor penetration.

Figure 9.

Uptake and intracellular localization of GA dual load. (A) Flow cytometry analysis of Huh-7 cells pretreated with the indicated inhibitors for 1 h, followed by incubation with GA dual load for 72 h. (B) Quantification of fluorescence intensity from (A). (C) CLSM images of Huh-7 cells after 72 h treatment; cell membranes were labeled with Dio, lysosomes with Lyso-Tracker, and nuclei with DAPI (scale bar = 50 μm). (D) Quantification of mean fluorescence intensity. Data are mean ± SD from three independent experiments (n = 3). **p < 0.01 vs GA dual load group (one-way ANOVA with LSD post-hoc test, SPSS version 19).

Intracellular Localization

Subcellular localization of NR and NR-loaded delivery systems was assessed by CLSM. The GA dual load group exhibited significantly higher fluorescence intensity than other groups (Figure 9C). Predominant red NR fluorescence overlapped with blue cytoplasmic fluorescence, yielding yellow fluorescence, indicating that the delivery system primarily localizes to the cytoplasm after internalization. Flow cytometry (Figure 9D) confirmed that the GA dual load group had significantly higher signal intensity than other groups (p < 0.01).

Mechanism of Antitumor Effect in vitro

ROS analysis showed that all drug and delivery system groups exhibited elevated ROS levels compared with control (p < 0.01, p < 0.05), with the GA dual load group showing the highest levels (Figure 10A and B). Mitochondrial membrane potential reduction-reflected by increased JC-1 green fluorescence ratio-was significantly higher in all treatment groups versus control (p < 0.01), with the GA dual load group showing the greatest reduction (Figure 10C and D). FCM-based apoptosis analysis revealed significantly higher apoptotic percentages in all treatment groups versus control (p < 0.01), with the GA dual load group demonstrating the highest levels (Figure 11A and B). Cell cycle analysis indicated that all treatment groups altered Huh-7 cell cycle distribution compared with control (decreased G0/G1, increased G2/M and S phases). The GA dual load group exhibited the highest proportion of cells in G2/M and S phases and the lowest in G0/G1 (Figure 11C and D). These in vitro findings indicate that the complete system induces ROS overproduction, mitochondrial membrane potential reduction, apoptosis, and G2/M arrest in Huh-7 cells. These effects are consistent with the known mechanism of BUF and provide a plausible mechanistic basis for the in vivo antitumor activity; however, they do not by themselves establish that ROS/mitochondrial dysfunction is the causal mediator of tumor suppression in vivo, which would require dedicated rescue experiments.

Figure 10.

Mechanism of the in vitro antitumor effect. (A) Huh-7 cells were treated for 72 h with control, free HAase + BUF NPs, GA dual load, No GA dual load, or No GA and HYD dual load at 104.97 nM BUF, followed by DCFH-DA staining for ROS detection (scale bar = 50 μm). (B) Quantification of ROS mean fluorescence intensity. (C) JC-1 staining for mitochondrial membrane potential measured by flow cytometry. (D) Quantification of JC-1 green fluorescence ratio. Data are mean ± SD from three independent experiments (n = 3). **p < 0.01vs control group (one-way ANOVA with LSD post-hoc test, SPSS version 19).

Figure 11.

Apoptosis and cell cycle arrest induced by GA dual load. (A) Huh-7 cells were treated for 72 h with HAase + BUF NPs, No GA dual load, No GA and HYD dual load, or GA dual load at 104.97 nM BUF, and apoptosis was assessed by flow cytometry. Red characters indicated the percentage of total apoptosis cells. (B) Quantification of apoptosis. (C) Cell cycle analysis by flow cytometry. (D) Quantification of cell cycle distribution. Data are mean ± SD from three independent experiments (n = 3). **p < 0.01 vs the indicated group (one-way ANOVA with LSD post-hoc test, SPSS version 19).

Mechanism of Antitumor Effect in vivo

Immunofluorescence analysis (Figure 12A) revealed that compared with the free BUF group, all three delivery system groups exhibited significantly lower expression of PCNA, HIF-1α, vimentin, and N-cadherin (p < 0.01, p < 0.05), with the GA dual load group showing the most pronounced reductions (Figure 12B–E). Additionally, CD3⁺ T cell infiltration was significantly higher in all three delivery system groups compared with the free BUF group (p < 0.01, p < 0.05), particularly in the GA dual load group (Figure 12F). However, CD3⁺ is a pan-T cell marker and does not distinguish CD8⁺ cytotoxic T cells, CD4⁺ helper T cells, or regulatory T cells. In addition, effector molecules such as IFN-γ and Granzyme B, immunogenic cell death markers, and functional T cell-mediated killing were not assessed. This finding should therefore be interpreted as a correlative observation consistent with a possible immunomodulatory effect rather than as evidence of immune activation. The reduced expression of N-cadherin and vimentin in the delivery system groups indicates attenuation of EMT-related phenotypes; however, these markers are surrogate indicators and do not demonstrate a reduction in metastatic burden. The histological analysis in Figure 6A showed qualitatively fewer lung metastatic foci in the GA dual load group, but this observation was not systematically quantified. No metastasis model was established in this study, and metastatic foci were not counted. Therefore, no conclusion regarding anti-metastatic efficacy can be drawn from the present data.

Figure 12.

In vivo mechanism of the antitumor effect. (A) Immunohistochemical staining of tumor tissues from each group after TACE. (B–F) Positivity rates of PCNA (B), HIF-1α (C), vimentin (D), N-cadherin (E), and CD3⁺. (F) Data are mean ± SD (n = 6 per group). ^#p < 0.05, ^##p < 0.01 vs NS group (one-way ANOVA with LSD post-hoc test, SPSS version 19).

Discussion

TACE remains a cornerstone treatment for unresectable HCC, but its efficacy is limited by inadequate intratumoral drug penetration. Elevated IFP, driven largely by HA accumulation in the ECM, collapses tumor blood vessels and impedes convective transport, restricting therapeutics to perivascular regions.^6,7,11–13 According to Starling’s law, once IFP reaches a critical threshold, interstitial fluid flow ceases and drug transport becomes diffusion-limited.^36,37 Reversing elevated IFP and enhancing drug transport into the tumor core are therefore rational strategies to improve TACE efficacy. Current approaches include reducing interstitial fluid volume, normalizing tumor vasculature, degrading type I collagen, and degrading HA.³⁸ Among these, HA degradation is particularly attractive because HA binds approximately 15 water molecules per disaccharide unit, forming a gel-like matrix that impedes convective transport.³⁸ In this study, we developed a sequential-delivery gel microsphere system integrating HAase-mediated ECM remodeling, GA-mediated active targeting, and pH-responsive drug release within a single TACE-deliverable platform. The present findings provide preclinical proof-of-concept that this integrated strategy is feasible and is associated with enhanced penetration and antitumor efficacy in the rabbit VX2 model.

HAase-mediated ECM remodeling was supported by the marked reduction of HA content across tumor edge, middle, and core regions in HAase-containing groups (Figure 8A). This extensive HA degradation would be expected to lower IFP based on established literature,^12,13,38 although IFP was not directly measured here. Consistent with this mechanism, HAase-containing groups showed significantly increased DAPI-positive signals across all three tumor regions, whereas non-HAase groups showed only perivascular DAPI positivity (Figure 8C). DAPI is a membrane-impermeant dye that stains only cells with compromised membranes; therefore, DAPI positivity reflects the spatial extent of drug-mediated cell killing rather than serving as a direct tracer of nanoparticle or BUF distribution. The presence of DAPI-positive signals in the tumor core of HAase-containing groups supports the conclusion that HAase-mediated ECM remodeling facilitated deep penetration of the therapeutic payload and its cytotoxic effect. These findings are consistent with previous reports showing that HAase-mediated HA degradation enhances nanoparticle penetration in solid tumors.^12,39

The non-additive enhancement of cellular uptake in the GA + HAase dual-loaded system compared with the GA single-loaded system warrants mechanistic interpretation. At 72 h, the dual-loaded system exhibited approximately 16-fold higher mean fluorescence intensity than the GA single-loaded system (Figure 3B), far exceeding simple additive targeting. Three complementary mechanisms may contribute. First, HAase-mediated HA degradation reduces the steric and viscous barrier of the ECM, increasing the effective surface area of tumor cells accessible to GA-decorated nanoparticles; in this sense, HAase functions as an “accessibility enhancer” for GA-mediated targeting rather than as an independent targeting ligand. Second, because HAase and BUF NPs are co-encapsulated within the same microsphere, HAase is released locally in the immediate vicinity of the nanoparticles, creating a spatiotemporally coordinated “pre-conditioned” microenvironment. Third, HA oligosaccharides generated by HAase activity may interact with CD44 and other HA-binding receptors, potentially modulating membrane lipid raft organization and endocytic activity. Because our internalization pathway analysis indicates caveolae-mediated endocytosis and macropinocytosis (Figure 9A and B), both sensitive to lipid raft dynamics, this represents a plausible additional synergistic mechanism. While the first two mechanisms are supported by our experimental data, the third remains a hypothesis and requires dedicated studies.

The pH-responsive HYD bond was designed to enable on-demand drug release within the acidic tumor microenvironment. In vitro, the complete system induced ROS overproduction, mitochondrial membrane potential reduction, apoptosis, and G2/M arrest in Huh-7 cells (Figures 10 and 11). These effects are consistent with the known mechanism of BUF and provide a plausible mechanistic basis for the in vivo antitumor activity.^23–26 However, these findings do not by themselves establish that ROS/mitochondrial dysfunction is the causal mediator of tumor suppression in vivo, and rescue experiments using an antioxidant such as N-acetylcysteine are required. The approximately 2-fold lower IC₅₀ of the complete system (55.07 nM) compared with free BUF (104.97 nM) reflects the combined contribution of nanoencapsulation, GA-mediated active targeting, HYD-based pH-responsive release, and HAase-mediated ECM remodeling. Notably, the magnitude of the IC₅₀ reduction is smaller than the approximately 16-fold increase in cellular uptake observed at 72 h, suggesting that downstream factors-including intracellular release kinetics, drug efflux, and intrinsic cellular sensitivity-also shape the net cytotoxic response. We therefore interpret the approximately 2-fold IC₅₀ reduction as evidence corroborating the mechanistic sequential rather than as a standalone demonstration of clinical superiority.

The contributions of HAase-mediated ECM remodeling and GA-mediated active targeting can be dissected through two single-variable comparisons built into the study design. The GA dual load versus GA single load comparison isolates HAase, since both groups contain GA, HYD, and identical nanoparticle and microsphere components; the significant reduction in tumor volume, tumor weight, HA content, and the deeper DAPI positivity observed in the GA dual load group therefore support a measurable contribution of HAase-mediated ECM remodeling. The GA dual load versus No GA dual load comparison isolates GA, since both groups contain HAase and HYD; the modest and non-significant difference in whole-tumor volume between these groups suggests that, under TACE-based superselective arterial delivery with co-encapsulated HAase, the marginal benefit of GA-mediated active targeting on bulk tumor growth is smaller than that of HAase. This does not imply that GA is ineffective; rather, its contribution is more clearly evident at the cellular level, where GA-decorated systems showed significantly higher uptake than non-GA-decorated controls and where free GA pretreatment reduced uptake of GA-decorated nanoparticles (Figure 3A and B; Figure 9A and B). These observations are consistent with previous studies demonstrating that GA receptors are overexpressed on HCC cells and that GA modification enhances HCC-specific delivery.^16–22 Nonetheless, the two factors are not strictly additive, and a full factorial design including a “No GA single load” group would be valuable in future studies to quantify interaction effects.

Compared with conventional DEB-TACE systems that rely solely on passive drug diffusion from embolic microspheres, our sequential-delivery platform offers three distinctive advantages. First, the co-encapsulation of HAase addresses the root physical barrier that limits drug extravasation, whereas conventional DEB-TACE microspheres lack any active TME-modulating capability. Second, while several HAase-conjugated nanoparticle systems have been reported for systemic cancer therapy,⁴⁰ their tumor accumulation remains limited by the hepatic first-pass effect and reticuloendothelial clearance; our TACE-based approach delivers the therapeutic payload directly into tumor-feeding arteries, maximizing local concentration while minimizing systemic exposure. Third, the GA-targeting ligand and HYD pH-responsive linker enable active cell recognition and on-demand intracellular drug release-features absent in most commercially available DEB-TACE products. Compared with our previous GA-PEG-PLGA nanomicelles, which showed suboptimal intracellular release, the incorporation of the HYD bond enables faster drug liberation upon lysosomal acidification. To our knowledge, this is the first study to integrate HAase-mediated HA degradation, GA-active targeting, and pH-responsive release within a single TACE-deliverable microsphere platform.

Beyond conventional DEB-TACE, several emerging strategies have been explored to enhance TACE-based therapy, including ultrasound-triggered microbubbles and photothermal therapy combined with TACE. Ultrasound-triggered microbubbles allow spatial and temporal control of drug release, but require an external ultrasound field and real-time imaging guidance, and their drug payload and sustained-release capacity are generally lower than those of gel microspheres. Photothermal therapy combined with TACE enables localized heating and may enhance drug penetration, but depends on photosensitizer accumulation and near-infrared laser irradiation, with limited tissue penetration depth and potentially heterogeneous heating in the deep tumor core. In contrast, our sequential-delivery platform does not require an external trigger device, provides sustained drug release over several days, achieves stable embolization, and enables deep delivery through HAase-mediated ECM remodeling; the incorporated BaSO₄ also allows real-time visualization under DSA. These features may be particularly advantageous for deep-seated or HA-rich hepatic tumors. We emphasize that these emerging strategies have their own applicable scenarios and that our platform is intended to complement, rather than replace, existing and emerging TACE technologies.

Beyond the proposed IFP-normalization mechanism, alternative pathways may contribute to the observed deep penetration. HAase-mediated HA fragmentation may trigger transcytosis in tumor endothelial cells,⁴⁰ and the small diameter of BUF NPs may enable passive diffusion through the remodeled ECM after HA degradation. Our internalization pathway analysis (Figure 9A and B) revealed that uptake occurs via energy-dependent, caveolae-mediated endocytosis and macropinocytosis rather than clathrin-mediated pathways-a mechanism that may facilitate subsequent transcytotic transport and deeper interstitial penetration. Although our experimental design does not fully distinguish these mechanisms, the collective evidence supports a multifactorial penetration process rather than a single dominant pathway.

The development of nanotechnology-based targeted drug delivery systems (NTDDS) has offered new prospects for cancer therapy.^41,42 NTDDS enables both passive targeting via the enhanced permeability and retention effect and active targeting through specific modifications.⁴³ However, systemic administration remains limited by low tumor accumulation (5–8%) due to the complex TME.⁴⁴ TACE overcomes this limitation by superselective intra-arterial delivery, concentrating the drug delivery system within tumor vasculature while embolizing its blood supply. By integrating TACE with NTDDS and addressing elevated IFP, we achieved enhanced deep penetration and targeted delivery efficiency. Specifically, HAase degraded HA across all tumor regions (Figure 8A), enabling nanoparticles to reach the tumor core (Figure 8C); GA modification enhanced cellular uptake (Figure 9A and B); and acidic lysosomal conditions triggered HYD cleavage, leading to mitochondrial dysfunction characterized by elevated ROS, decreased membrane potential, and G2/M arrest (Figure 9). This integrated sequential translated into superior antitumor efficacy, including suppressed tumor growth, reduced angiogenesis, and enhanced intratumoral immune infiltration (Figures 5–8 and 11). Notably, the enhanced CD3⁺ T cell infiltration observed in the fully functional GA/HAase group (Figure 12F) suggests that deep penetration of BUF may facilitate immunogenic cell death and subsequent antitumor immune responses. However, the observed reductions in HIF-1α, vimentin, and N-cadherin expression, together with increased CD3⁺ T cell infiltration, are correlative findings consistent with the proposed mechanisms rather than direct mechanistic validation. Dedicated studies-including functional assays for T cell activation, cytokine profiling, and direct hypoxia measurements-are warranted. While the present study focused on overcoming the physical barrier of drug penetration, we acknowledge that TACE treatment failure is multifactorial. Future studies combining our platform with anti-angiogenic agents or immunotherapies may address additional factors such as tumor hypoxia and immune evasion. Furthermore, while bufalin was used as the model payload in this proof-of-concept study, the sequential-delivery platform is designed to be payload-flexible. The physical delivery mechanisms, HAase-mediated ECM remodeling, GA-active targeting, and pH-responsive release, are independent of the specific drug payload and would be equally applicable to clinically established TACE agents such as doxorubicin or epirubicin.

Several limitations should be acknowledged. First, the rabbit VX2 model is not of hepatocellular origin and lacks a cirrhotic background; therefore, the results cannot be directly extrapolated to human HCC. Validation in orthotopic patient-derived xenograft or chemically induced HCC models is required. Second, the 18-day follow-up is insufficient to assess delayed or cumulative toxicity, and repeated TACE dosing was not evaluated. Future studies should extend the observation period, include repeated TACE administration, and perform serial cardiac and hepatic biomarker monitoring together with end-point organ histopathology. Third, IFP was not directly measured, and DAPI positivity is an indirect indicator of drug efficacy rather than a direct tracer of nanoparticle distribution. Future studies should employ direct IFP measurement and fluorescently labeled nanoparticles with confocal z-stacking to quantify penetration depth. Fourth, the in vitro ROS/mitochondrial findings are correlative with respect to the in vivo antitumor effect, and rescue experiments using N-acetylcysteine are required to establish causality. Fifth, the immune-related findings, including increased CD3⁺ T cell infiltration, are descriptive and correlative; CD4/CD8/Treg subsets, effector cytokines, immunogenic cell death markers, and functional killing assays were not evaluated. Sixth, no metastasis model was established, and the reduced expression of N-cadherin and vimentin should be regarded as hypothesis-generating rather than evidence of anti-metastatic efficacy. Seventh, microsphere stability, microcatheter passage, sterilization compatibility, and cGMP-scale manufacturing remain to be established. Finally, bufalin was used as a model payload; future studies should validate the platform with clinically established TACE agents such as doxorubicin or epirubicin.

In summary, while our experimental observations-including HA reduction, DAPI positivity, reduced HIF-1α and EMT marker expression, and increased CD3⁺ T cell infiltration-are consistent with the proposed sequential mechanisms, they represent indirect or endpoint evidence rather than direct mechanistic validation. We have therefore framed these findings as supportive of, rather than definitive proof of, the proposed mechanisms. Despite these limitations, the consistent efficacy data across multiple independent assays, combined with the favorable safety profile, support further translational development of this sequential-delivery system.

Conclusions

We developed a sequential-delivery composite gel microsphere system integrating HAase-mediated ECM remodeling, GA-mediated active targeting, and pH-responsive drug release within a single TACE-deliverable platform. In a rabbit VX2 orthotopic liver tumor model, TACE with HAase/BUF NPs-GM suppressed tumor growth, reduced angiogenesis, induced apoptosis, and was associated with increased intratumoral CD3⁺ T cell infiltration. The CD3⁺ T cell finding is correlative and warrants further investigation in combination immunotherapy. Within the 18-day observation period, encapsulation of BUF mitigated its cardiotoxicity and hepatotoxicity; however, long-term safety remains to be established.

This strategy provides a promising preclinical proof-of-concept for addressing the penetration limitations of conventional TACE. Its practical advantages include no requirement for an external trigger device, sustained drug release, stable embolization, and deep delivery via HAase-mediated ECM remodeling. The platform is payload-flexible and should be validated in future studies with clinically established TACE agents such as doxorubicin or epirubicin.

Before clinical translation, key manufacturing and regulatory challenges remain, including optimization of the high-voltage pulse microencapsulation process for cGMP-scale production, establishment of quality control parameters (sterility, endotoxins, and storage stability), and evaluation of sterilization compatibility. Further translational investigation-including validation in orthotopic or patient-derived xenograft models, long-term toxicity assessment, and direct mechanistic studies-is warranted.

Funding Statement

This work was supported by the National Natural Science Foundation of China (No.82104699), the Science Technology Department of Zhejiang Province (No. LGF22H290003), the Construction Fund of Key Medical Disciplines of Hangzhou (No.2025HZGF09), the Zhejiang Provincial Health Commission (No. 2025KY1134), the Major Project of Hangzhou Municipal Health Commission (No. Z20240019), the Science and Technology Development Support Program for Biomedicine and Health Industry of Hangzhou (No. 2022WJC230), and the “115” Plan Outstanding Talent Introduction Project of Hangzhou Science and Technology Bureau.

Abbreviations

TACE, transcatheter arterial chemoembolization; HCC, hepatocellular carcinoma; DEB, drug-eluting bead; TME, tumor microenvironment; IFP, interstitial fluid pressure; HA, hyaluronic acid; HAase, hyaluronidase; GA, glycyrrhetinic acid; HYD, hydrazone; BUF, bufalin; PEG, polyethylene glycol; PLGA, poly(lactic-co-glycolic acid); CMCS, carboxymethyl chitosan; ALG, sodium alginate; EE, encapsulation efficiency; LC, loading content; BUF NPs, BUF-loaded GA-PEG-HYD-PLGA nanoparticles; GA dual load or HAase/BUF NPs-GM, BaSO₄-CMCS/ALG@GA-PEG-HYD-PLGA-BUF/HAase; GA single load, BaSO₄-CMCS/ALG@GA-PEG-HYD-PLGA-BUF; No GA dual load, BaSO₄-CMCS/ALG@PEG-HYD-PLGA-BUF/HAase; No GA and HYD dual load, BaSO₄-CMCS/ALG@PEG-PLGA-BUF/HAase; NTDDS, nanotechnology-based targeted drug delivery systems; NR, Nile Red; FCM, flow cytometry; CLSM, confocal laser scanning microscopy; DMSO, dimethyl sulfoxide; NS, normal saline; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ALP, alkaline phosphatase; TBil, total bilirubin; Cr, creatinine; BUN, blood urea nitrogen; CK, creatine kinase; CK-MB, creatine kinase myocardial band; LDH, lactate dehydrogenase.

Data Sharing Statement

The data that support the findings of this study are available from the corresponding author Rangxiao Zhuang upon reasonable request.

Ethics Statement

This study has been approved with the following ethics approval: Institutional Animal Care and Use Committee (ACUG) at the ZJCLA -protocol #ZJCLA-IACUC-20010969. All animal procedures were performed in accordance with the Chinese Council on Animal Care Guidelines and the AVMA Guidelines for the Euthanasia of Animals (2020 Edition).

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that they have no competing interests.