ROS-Activatable “off-on” Fluorescent Prodrug Loaded Micelles for Targeted Therapy and in Situ Imaging of Breast Cancer Cells

Abstract

This study developed an intelligent drug-loaded micelle system based on folic acid-conjugated polyglycidol (FAPG), which incorporates a reactive oxygen species (ROS) responsive fluorophore-drug conjugate (BTBP-PTX, BP). The system integrates folate receptor-α (FR-α) targeting, esterase/ROS dual-responsive drug release, and aggregation-induced emission (AIE)-based fluorescence activation for tumor-targeted delivery and imaging. Under simulated tumor microenvironment conditions (esterase-rich, high ROS levels), the system displayed dual-responsive release behavior, achieving over 80% paclitaxel (PTX) release within 24 h through esterase-catalyzed ester bond cleavage and ROS-triggered oxalate bond hydrolysis. Cellular assays demonstrated that BP@FAPG enhanced cellular uptake by approximately 2.5-fold in MDA-MB-231 cells compared to non-targeted controls BP@GAPG, along with significant cytotoxicity (IC₅₀ = 60 nM) and deep penetration into three-dimensional tumor spheroids—effects primarily attributed to FR-α-mediated endocytosis. Furthermore, the AIE fluorophore was activated during drug release, allowing real-time fluorescence tracking of micelle localization and release dynamics.

Graphical abstract

Highlights

• •Dual-responsive BP@FAPG micelle enables PTX delivery and AIE imaging for breast cancer.
• •FAPG polymer shows low CMC, spontaneous self-assembly, and high interfacial efficiency.
• •Enhanced cellular uptake with potent cytotoxicity and deep spheroid penetration via FR-α endocytosis.
• •Real-time tracking of drug release via ROS-activated AIE fluorescence, linking therapy to imaging.

1. Introduction

Paclitaxel (PTX) is a cornerstone chemotherapeutic agent for treating breast, ovarian, and lung cancers. However, its clinical utility is severely compromised by poor aqueous solubility (<0.3 μg/mL), lack of tumor selectivity, and dose-limiting systemic toxicity (Han et al., 2016; Lu et al., 2023). The conventional formulation Taxol® relies on Cremophor EL to enhance solubility, yet this excipient triggers hypersensitivity reactions, neutropenia, and neurotoxicity in a significant proportion of patients (Zhang et al., 2024; Zhang et al., 2023a). These drawbacks emphasize the urgent need for advanced drug delivery systems capable of achieving controlled release, tumor-specific accumulation, and minimized off-target effects (Huang et al., 2018; Sofias et al., 2017).

Nanomedicine has emerged as a transformative strategy to address these challenges (Feng et al., 2019). Early nanoformulations primarily exploited passive targeting, leveraging physicochemical differences between tumor and normal tissues (Zhang et al., 2020a). The enhanced permeability and retention (EPR) effect remains the prototypical passive mechanism and is characterized by leaky vasculature and impaired lymphatic drainage (Liao et al., 2024; Adu-Frimpong et al., 2019; Yao et al., 2021). Additionally, tumor microenvironment (TME) distinctive features, such as acidic pH, elevated glutathione concentration, and overproduced reactive oxygen species (ROS), enable stimuli-responsive nanocarriers that disassemble selectively in the TME (Gu et al., 2022; Zhang et al., 2020b; Xiao et al., 2021; Zhang et al., 2023b; Zhang et al., 2023c). Despite these advances, passive strategies suffer from inter-patient and intra-tumoral heterogeneity, limited cellular uptake, and premature cargo leakage, resulting in suboptimal therapeutic indices (Hussain et al., 2024).

To achieve active targeting, ligands that bind overexpressed receptors on cancer cells are conjugated to nanocarriers, facilitating receptor-mediated endocytosis. Among these, folate receptors (FR-α) are particularly attractive (Xiao et al., 2025; Daniele et al., 2026). FR-α is overexpressed in >90% of epithelial ovarian cancers and substantial fractions of breast, lung, and colorectal carcinomas (Scaranti et al., 2020). Folate, a low-molecular-weight vitamin, exhibits nanomolar affinity for FR-α, excellent stability, and negligible immunogenicity (Li et al., 2025; Fernández et al., 2018). Conjugation of folate to nanocarriers via cleavable linkers (e.g., ester, disulfide) thus enables dual functionality: specific cellular recognition and internalization, followed by programmable linker cleavage by TME-enriched stimuli such as esterases (Hong et al., 2025; Liu et al., 2025a).

Compared to polyethylene glycol (PEG), the gold standard hydrophilic polymer, linear polyglycerol (LPG) offers a hyperbranched yet linear topology with abundant pendant hydroxyl groups, enabling high-density ligand grafting without compromising solubility (Wang et al., 2024; Cai et al., 2020). LPG also confers prolonged blood circulation and superior tumor accumulation versus PEG analogues, attributable to reduced protein corona formation and enhanced conformational flexibility (Wang et al., 2016; Bochenek et al., 2020). Herein, folate is tethered to LPG via an ester linkage to yield FAPG, a folate-decorated amphiphilic polymer (Liu et al., 2025b). In aqueous media, FAPG self-assembles into micelles with folate-enriched coronas, priming them for FR-mediated endocytosis. Upon internalization, esterases in tumor cells hydrolyze the folate-LPG ester, destabilizing the micelle and liberating the encapsulated cargo.

A parallel challenge in nanomedicine is real-time visualization of drug release to confirm spatiotemporal accuracy. Conventional approaches append always-on fluorophores (e.g., Cy5.5, FITC) to nanocarriers; however, persistent background signals confound the interpretation of release events (Ding et al., 2025; Mahmood et al., 2026). Stimuli-activatable probes, such as GSH-cleavable disulfide-linked rhodamine or pH-sensitive squaraine, partially mitigate this issue but often suffer from incomplete activation or off-target responses due to TME heterogeneity. Aggregation-induced emission (AIE) fluorogens offer a superior alternative (Wang et al., 2023; Tang et al., 2026; Wang et al., 2025). The fluorophore 2-(benzo[d]thiazol-2-yl)-4-bromophenol (BTBP) is a well-established excited-state intramolecular proton transfer (ESIPT) -active AIEgen (Song et al., 2025; Wang et al., 2026; Yang et al., 2023; Wei et al., 2020; Yang et al., 2016). In its free form, BTBP exhibits strong green emission due to an efficient ESIPT process, which relies on an intact intramolecular hydrogen bond between the phenolic -OH and the benzothiazole nitrogen atom (Huang et al., 2023; Ma et al., 2025). Capitalizing on this principle, we engineered a ROS-responsive PTX prodrug (BTBP-PTX, BP) by conjugating PTX to the phenolic -OH of BTBP through an oxalate ester linker. In the intact BTBP-PTX prodrug, the conjugation disrupts the intramolecular hydrogen bond and blocks the ESIPT process, resulting in a non-emissive “off” state. Upon ROS-triggered cleavage of the oxalate linker in the tumor microenvironment, free BTBP is released. The restored phenolic -OH re-establishes the intramolecular hydrogen bond, reactivating ESIPT and producing strong aggregation-induced emission fluorescence (“on” state) (Kaur et al., 2023; Ma et al., 2022). This ESIPT-blockage/restoration mechanism ensures that fluorescence activation is strictly dependent on ROS-mediated prodrug cleavage, enabling real-time, background-free tracking of drug release. The hydrophobic prodrug is encapsulated within the FAPG micelle core, forming the integrated theranostic nanoparticle BP@FAPG. Following FR-mediated endocytosis, esterase-triggered micelle disassembly releases the prodrug, which is subsequently cleaved by ROS to concomitantly activate PTX cytotoxicity and AIE fluorescence. This dual enzyme/ROS responsiveness circumvents reliance on any single TME cue, enhancing reliability across heterogeneous tumors, while the fluorescence “off-to-on” signal serves as a direct, real-time readout of PTX release.

In this study, the actively targeted BP@FAPG micelles exhibited significantly improved performance over their non-targeted counterpart, BP@GAPG, attributable to the folate-mediated targeting moiety (Scheme 1). Subsequent assays confirmed the superior cellular imaging capability and anticancer activity of the BP@FAPG micelles. This integrated targeting approach represents a significant advancement over conventional drug delivery systems, providing a promising strategy for the precision delivery of paclitaxel within the complex tumor microenvironment. Moreover, it establishes a valuable platform to create integrated theranostic nanocarriers for other chemotherapeutic agents.

Scheme 1.

Schematic illustration of tumor cell killing and imaging function of BP@FAPG micelles.

2. Experimental section

2.1. Synthesis

2.1.1. Synthesis of 2-(benzo[d]thiazol-2-yl)-4-bromophenol (BTBP)

A mixture of 5-bromo-2-hydroxybenzaldehyde (2 g, 9.94 mmol) and 2-aminobenzenethiol (1.17 mL, 10.9 mmol) was dissolved in a 1:1 (v/v) mixture of THF and MeOH in a round-bottom flask. The reaction mixture was heated under reflux for 5 h. After completion, the mixture was cooled to room temperature, and the precipitate was filtered to afford compound BTBP as a white solid (2.284 g, 75% yield). ¹H NMR (400 MHz, Chloroform‑d) δ 8.03–7.97 (m, 1H), 7.93 (ddd, J = 0.6, 1.4, 7.8 Hz, 1H), 7.79 (d, J = 2.4 Hz, 1H), 7.53 (ddd, J = 1.2, 7.4, 8.2 Hz, 1H), 7.48–7.41 (m, 2H), 7.00 (d, J = 8.8 Hz, 1H).

2.1.2. Synthesis of BTBP-PTX

BTBP (500 mg, 1.63 mmol) was dissolved in dry dichloromethane (8 mL) under a nitrogen atmosphere. An excess of oxalyl chloride (276 μL, 3.26 mmol) was added dropwise to the solution at 0 °C. After stirring at room temperature for 1 h, the solvent and excess oxalyl chloride were removed under reduced pressure. The flask was purged with nitrogen, and a solution of paclitaxel (2.09 g, 2.45 mmol) in dry dichloromethane and triethylamine (340 μL, 2.45 mmol) were added dropwise at 0 °C. Upon completion of the reaction, the mixture was concentrated under reduced pressure. The crude product was purified by column chromatography (DCM: MeOH = 5:1) to give the target compound as a white solid (633 mg, 32% yield). ¹H NMR (400 MHz, DMSO‑d₆) δ 9.41 (d, J = 8.6 Hz, 1H), 8.43 (d, J = 2.4 Hz, 1H), 8.02 (d, J = 8.2 Hz, 1H), 7.99–7.94 (m, 2H), 7.92–7.87 (m, 2H), 7.79–7.59 (m, 6H), 7.56–7.38 (m, 9H), 7.28–7.22 (m, 1H), 6.20 (s, 1H), 5.87–5.80 (m, 1H), 5.73–5.67 (s, 1H), 5.53 (d, J = 9.0 Hz, 1H), 5.38 (d, J = 7.2 Hz, 1H), 4.85 (d, J = 2.2 Hz, 2H), 4.65 (s, 1H), 3.98 (d, J = 4.6 Hz, 1H), 3.52 (d, J = 7.2 Hz, 1H), 2.22–2.18 (s, 3H), 2.12 (s, 3H), 1.81–1.74 (m, 1H), 1.65–1.55 (m, 2H), 1.53 (d, J = 1.4 Hz, 3H), 1.46 (s, 3H), 1.00 (d, J = 10.4 Hz, 6H). HRMS (ESI) m/z calcd for C₆₂H₅₇BrN₂O₁₇S [M + H]⁺1213.2561, found 1213.2540. ¹³C NMR (101 MHz, DMSO) δ 202.43, 172.75, 169.89, 168.81, 166.31, 165.24, 162.50, 160.98, 155.30, 151.34, 139.32, 139.27, 135.05, 134.53, 134.49, 133.43, 131.36, 130.21, 130.17, 130.03, 129.60, 128.71, 128.30, 127.49, 127.43, 126.50, 125.21, 122.43, 122.01, 120.97, 119.19, 110.86, 83.64, 80.32, 76.83, 75.41, 74.83, 74.56, 74.30, 73.67, 70.52, 69.67, 57.48, 56.36, 46.16, 43.02, 40.15, 39.94, 39.73, 39.52, 39.31, 39.10, 38.89, 36.59, 34.77, 26.36, 22.52, 21.35, 20.70, 13.86, 9.82.

2.1.3. Synthesis of BTBP-BTBP

BTBP (500 mg, 1.63 mmol) was dissolved in dry dichloromethane (5 mL) in a 50 mL round-bottom flask under a nitrogen atmosphere. Oxalyl chloride (69 μL, 0.82 mmol) and triethylamine (153 μL, 1.63 mmol) were added dropwise to the solution at 0 °C. The reaction mixture was then allowed to warm to room temperature and stirred for 5 h. After completion, the crude product was purified by column chromatography (Hexane) to afford the target compound as a white solid (348 mg, 32% yield). ¹H NMR (400 MHz, Chloroform‑d) δ 8.55 (d, J = 2.5 Hz, 2H), 8.12 (d, J = 8.2 Hz, 2H), 7.95 (dd, J = 1.2, 8.0 Hz, 2H), 7.62 (dd, J = 2.4, 8.6 Hz, 2H), 7.53 (ddd, J = 1.2, 7.0, 8.2 Hz, 2H), 7.43 (ddd, J = 1.2, 7.2, 8.0 Hz, 2H), 7.23 (d, J = 8.6 Hz, 2H).

2.1.4. Synthesis of FAPG and GAPG

Folic acid (293 mg, 0.66 mmol) was dissolved in dry DMF (5 mL) in a 50 mL round-bottom flask. To this solution, DCC (151 mg, 0.73 mmol) and DMAP (8 mg, 0.06 mmol) were added, and the mixture was stirred at room temperature for 3 h. Subsequently, LPG (500 mg, 0.11 mmol) was added, and the reaction was allowed to proceed for 24 h. Upon completion, the crude mixture was transferred into a dialysis bag (MWCO 3500 Da) and dialyzed against distilled water for 24 h. The dialysate was then filtered and lyophilized to obtain the final product FAPG as a yellow solid (625 mg, 80% yield). ¹H NMR (400 MHz, DMSO‑d₆) δ 8.65 (d, J = 9.2 Hz, 6H), 7.91–7.53 (m, 12H), 7.37–6.94 (m, 12H), 6.92–6.49 (m, 12H), 4.91–4.42 (m, 54H), 4.28–3.78 (m, 24H), 2.39–1.63(m, 24H), 1.10 (s, 9H).

GAPG was synthesized following the same procedure as described for FAPG, yielding a white solid (492 mg, 73% yield). ¹H NMR (400 MHz, DMSO‑d₆) δ 7.85 (d, J = 9.8 Hz, 12H), 7.41 (d, J = 8.4 Hz, 12H), 4.61 (s, 54H), 4.11–3.86 (m, 18H), 2.07–1.71(m, 12H),1.36 (s, 12H), 1.13 (s, 9H).

2.2. Preparation of drug-loaded micelles

The drug-loaded micelles were prepared using the film-dispersion method. Briefly, FAPG, BTBP-BTBP, and BTBP-PTX at different mass ratios (10:1:0.5,10:2:1 and 10:3:1) were dissolved in methanol. The solvent was then evaporated by rotary evaporation to form a thin BP@FAPG film. This film was subsequently re-dispersed in ultrapure water, and the mixture was filtered (0.22 μm) to afford the final micelle solution. Notably, BTBP-BTBP was co-encapsulated as a built-in fluorescence amplifier. Similar to the ROS-activatable BTBP-PTX, BTBP-BTBP is non-fluorescent under normal physiological conditions but releases fluorescence upon cleavage of its oxalate bond in the ROS-overexpressed tumor microenvironment, thereby enhancing the real-time imaging sensitivity of the drug-loaded micelles. The BP@GAPG micelles were prepared using the same method. The drug loading content of BTBP-PTX was quantitatively determined via high-performance liquid chromatography (HPLC). Briefly, the drug-loaded micelles were redissolved in methanol and ultrasonicated for 15 min to ensure complete dispersion. The resulting solution was then injected (20 μL injection volume) into an Agilent InfinityLab Poroshell 120 C18 column (4 μm, 250 mm × 4.0 mm) for separation under the following conditions: mobile phase, H₂O/acetonitrile = 15/85 (v/v); flow rate, 0.5 mL/min; detection wavelength, 254 nm.

2.3. In vitro cytotoxicity

Cytotoxicity was evaluated in vitro using the Cell Counting Kit-8 (CCK-8) assay on two representative cell lines: L929 murine fibroblasts, serving as a normal cell model, and MDA-MB-231 human triple-negative breast cancer cells as a tumor model. After seeding cells in 96-well plates at a density of 5 × 10³ cells per well and allowing them to adhere overnight, the culture medium was replaced with fresh medium containing serially diluted samples. L929 cells were treated with BTBP-PTX, BP@FAPG, and BP@GAPG micelles, while MDA-MB-231 cells were exposed to free PTX, BTBP-PTX, BP@FAPG, and BP@GAPG micelles. Following 48 h of incubation at 37 °C under 5% CO₂, 10 μL of CCK-8 reagent was added to each well, and the plates were incubated for an additional 2 h. Absorbance was measured at 450 nm using a microplate reader, and cell viability was expressed as a percentage relative to untreated controls.

2.4. Cellular uptake

The cellular uptake behavior of micelles was qualitatively and visually investigated using confocal laser scanning microscopy (CLSM). MDA-MB-231 cells were seeded in glass-bottom confocal dishes at a density of 5 × 10⁴ cells per dish and cultured for 24 h at 37 °C under 5% CO₂ to ensure complete adhesion. Subsequently, the cells were treated with free BTBP, free BTBP-PTX, BP@GAPG micelles, and BP@FAPG micelles (each at a PTX-equivalent concentration of 0.1 μM) and incubated for a specified duration. Following incubation, the cells were washed three times with phosphate-buffered saline (PBS), stained with Hoechst 33342 for 15 min, and washed again with PBS. Cellular uptake was observed by detecting fluorescence signals at the appropriate emission wavelengths for BTBP (green) and Hoechst 33342 (blue).

2.5. In vitro dead/live imaging

The viability of MDA-MB-231 cells after treatment with various formulations was evaluated using live/dead cell fluorescence staining. Specifically, MDA-MB-231 cells were plated into 24-well plates at 5 × 10⁴ cells per well in complete medium and allowed to adhere for 24 h. The cells were then treated with culture medium containing free BTBP (100 nM), free BTBP-PTX, BP@FAPG micelles, and BP@GAPG micelles (all at a PTX-equivalent concentration of 60 nM) for 24 h. A separate control was established using L929 cells treated with BP@FAPG alone. After the treatment period, all samples were rinsed three times with PBS, stained with Calcein-AM and PI, and examined under a fluorescence microscope.

2.6. 3D Tumor Spheroid Anticancer activity Assessment

MDA-MB-231 cells were formed into tumor spheroids via the hanging drop method and cultured for 48 h until mature, structurally intact spheroids were obtained. Subsequently, free BTBP-PTX and the prodrug-loaded micelles were added, both administered at a PTX-equivalent concentration (60 nM). The spheroids were stained using Calcein-AM and Propidium Iodide, and overall fluorescence distribution as well as morphological changes were analyzed by imaging under a Nikon Ti inverted microscope.

2.7. Statistical Analysis

All statistical analyses were conducted using GraphPad Prism 8 (GraphPad Software, USA). Comparisons between two groups were performed with the two-tailed Student's t-test. Data are expressed as mean ± standard deviation (SD). Differences were considered statistically significant at *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

3. Results and discussion

3.1. Synthesis and characterization

Scheme 2 illustrates the synthetic routes for the FAPG and GAPG polymers. Their chemical structures, molecular weights, and molecular weight distributions were systematically characterized using proton nuclear magnetic resonance (¹H NMR) spectroscopy and gel permeation chromatography (GPC). The successful synthesis of FAPG and GAPG was confirmed by ¹H NMR spectroscopy, which displayed the characteristic resonance peaks of the LPG backbone along with signals corresponding to the conjugated folic acid and gambogic acid ligands (Fig. S6 and Fig. S7). Their number-average molecular weights (Mn) were determined by GPC to be 8739 g/mol and 7432 g/mol, respectively (Fig. 2a). The polydispersity indices (PDI) were determined to be 1.03 for GAPG and 1.05 for FAPG. These low PDI values indicate a high degree of molecular homogeneity within each polymer batch. Such uniformity is advantageous for the formation of consistent and stable micellar carriers and reflects the good controllability and reproducibility of the synthetic methodology.

Scheme 2.

Synthesis of BTBP-PTX, FAPG, and GAPG.

Fig. 2.

(a) GPC of FAPG and GAPG; (b) Surface tensions of FAPG and GAPG in aqueous solution; (c) Size (diameter) distributions of FAPG, GAPG, BP@FAPG, and BP@GAPG micelles (mean ± SD, n = 3); TEM images of BP@FAPG and BP@GAPG micelles; (d) Size (diameter) of BP@FAPG and BP@GAPG micelles in FBS buffer by DLS (48 h).

3.2. Fluorescence analysis of BTBP/BTBP-PTX

The AIE property of the BTBP fluorophore is the core mechanism enabling the “off-on” fluorescence response in our system. To verify that BTBP is an AIEgen, we investigated its fluorescence behavior in DMSO/H₂O mixtures. As shown in Fig. 1a, BTBP exhibits very weak fluorescence in pure DMSO (molecularly dissolved state, “off”). When the water fraction exceeds 70% ($f_w\ge 70\%$), BTBP molecules aggregate, and a dramatic fluorescence enhancement is observed (Fig. 1b), reaching approximately 8-fold higher intensity at $f_w=90\%$. This water-fraction-dependent fluorescence enhancement is the hallmark of aggregation-induced emission, confirming that BTBP is an AIEgen. This behavior is consistent with recent reports on benzothiazole-based AIE fluorophores. It is worth noting that BTBP exhibits both ESIPT (characterized by the dual enol/keto emission and large Stokes shift) and AIE (characterized by the water-fraction-dependent fluorescence enhancement). These two properties are not mutually exclusive but coexist in this benzothiazole-based fluorophore, as previously reported for similar derivatives (Song et al., 2025; Wei et al., 2020; Ahmad et al., 2021). Leveraging this AIE characteristic, we constructed the ROS-responsive probe BTBP-PTX and systematically evaluated its fluorescence response. Concentration-dependent experiments (Fig. 1c) confirmed the high stability of the probe, as evidenced by minimal background fluorescence in the absence of H₂O₂. A strong linear correlation between fluorescence intensity and H₂O₂ concentration was observed in the range of 0–10 μM. Beyond 10 μM, the fluorescence signal plateaued, indicating complete activation of the prodrug. The time-dependent response curve (Fig. 1d) further demonstrates that upon exposure to a fixed concentration of H₂O₂, the oxalate ester linker is specifically cleaved, releasing the hydrophobic BTBP fluorophore. This release immediately induces BTBP aggregation, leading to a rapid and significant increase in fluorescence intensity within a short timeframe. This indicates fast response kinetics, efficiently converting the chemical recognition of H₂O₂ into an optical “off-on” signal suitable for real-time dynamic monitoring. These results collectively demonstrate that BTBP-PTX possesses high sensitivity and quantitative detection capability for H₂O₂ at pathophysiological concentration levels.

Fig. 1.

(a) Fluorescence spectra of BTBP in H₂O/DMSO co-solvents with different volume fractions of water; (b) High-energy emission intensity of BTBP with different water fractions in H₂O/DMSO mixtures; (c) Fluorescence responses of BTBP-PTX toward different concentrations of H₂O₂; (d) Fluorescence responses of BTBP-PTX toward H₂O₂ for different time periods.

3.3. Surface-active properties of polymers

The critical micelle concentration (CMC) is a fundamental parameter for evaluating the self-assembly capability of amphiphilic molecules (Fig. 2b). As listed in Table 1, the CMC values for FAPG and GAPG are 11.15 μmol/L and 13.11 μmol/L, respectively. These low CMC values indicate that both polymers can spontaneously form micelles at low concentrations, a property that helps maintain micellar stability upon dilution in the bloodstream until they reach the target site. The minimum surface tension at the CMC (γ_CMC) was determined to be 53.02 mN/m for FAPG and 43.93 mN/m for GAPG. From these values, the corresponding surface pressure (Π_CMC) was calculated as 18.68 mN/m and 21.73 mN/m, respectively. Notably, both the γ_CMC and Π_CMC values of the synthesized polymers are superior to those of widely used surfactants such as Tween 80 and Triton X-100, highlighting their excellent efficiency in reducing interfacial energy (Tiwari et al., 2020; Acosta et al., 2005; Jaiswal et al., 2016). This indicates superior adsorption and spreading efficiency of their molecules at the air-water interface. This strong interfacial activity contributes to the formation of stable nanomicelles with a lower surface energy and a more compact shell, which helps minimize premature leakage of the encapsulated drug. To further analyze the interfacial behavior, the maximum surface excess concentration (Γ_max) and the minimum area per molecule (A_min) were calculated using the Gibbs adsorption isotherm. Both FAPG and GAPG exhibited relatively high Γ_max (1.51 × 10⁻⁶ mol/m² and 1.92 × 10⁻⁶ mol/m²) and reasonable A_min values (1.10 nm² and 0.86 nm²), suggesting that both polymers can achieve effective arrangements at the interface. Thermodynamic parameters provide quantitative insight into the self-assembly and adsorption processes. The free energy of micellization (ΔG0 mic) and the free energy of adsorption (ΔG0 ads) for both polymers were large negative values, confirming that both processes are thermodynamically highly spontaneous.

Table 1.

Surface Parameters of FAPG and GAPG.

Sample	CMC (μM)	γCMC (mN/m)	ΠCMC (mN/m)	Γmax × 10−6 (mol/m2)	Amin (nm2/molecule)	ΔG0 mic (kJ/mol)	ΔG0 ads (kJ/mol)
FAPG	11.15	53.02	18.68	1.51	1.10	−38.23	−39.46
GAPG	13.11	43.93	21.73	1.92	0.86	−37.83	−38.45

3.4. Characterization of micelles

Dynamic light scattering (DLS) analysis revealed that blank FAPG and GAPG micelles had hydrodynamic diameters of 51 nm and 68 nm (Fig. 2c), respectively, both with narrow size distributions, making them suitable for drug encapsulation and release. Prodrug-loaded micelles were successfully prepared using the film-hydration method. Transmission electron microscopy (TEM) images revealed that the micelles were relatively uniform spherical particles (Fig. 2c). Upon loading the hydrophobic prodrug BTBP-PTX, the micelle size increased: BP@FAPG diameter increased to 124 nm, and BP@GAPG to 106 nm, confirming successful encapsulation of the prodrug within the micellar core. The stability of the micelles was further evaluated over a period of 48 h in PBS containing 10% fetal bovine serum (FBS). Both the hydrodynamic diameter and polydispersity index (PDI) remained stable throughout this period, with no significant changes observed (Fig. 2d). The stability of BP@FAPG micelles under acidic conditions (pH 5.0) was further evaluated. As shown in Fig. S3, both the hydrodynamic diameter and fluorescence intensity remained essentially unchanged over 12 h, indicating that pH alone does not trigger micelle disintegration or premature fluorescence activation. This observation is consistent with previous reports on oxalate-based ROS-responsive probes, which likewise showed minimal pH-induced hydrolysis under acidic conditions (Liang et al., 2024). These results suggest that the micelles possess favorable stability characteristics, indicating their potential for sustained circulation in vivo.

3.5. In vitro drug release

To preliminarily verify the stimuli-responsive drug release behavior of the BP@FAPG system, we simulated drug release profiles under different physiological and pathological conditions in vitro. As shown in Fig. 3, in the PBS control group, cumulative drug release remained below 10%, confirming the excellent stability of the drug carrier and the absence of significant premature leakage under physiological conditions. In single-stimulus groups, drug release efficiency was relatively limited. The addition of esterase alone (25 U/mL) resulted in only 15% release, primarily attributed to esterase-triggered micelle disassembly without subsequent prodrug activation. In the presence of 10 μM H₂O₂ alone, the release increased to 30%. This release is mainly due to the cleavage of the oxalate ester bond in prodrug molecules that had weakly dissociated from the micelles. A pronounced synergistic effect was observed when both H₂O₂ and esterase were present. The release rate increased dramatically, reaching approximately 90% within 48 h. These results demonstrate that the PTX-loaded copolymer micelles exhibit esterase- and ROS-triggered drug release, maintaining stability under physiological conditions while enabling rapid and near-complete drug release in the tumor microenvironment. Compared to previously reported ROS-responsive prodrug systems that rely on a single stimulus (e.g., thioketal or boronate ester-based systems), our BP@FAPG micelles exhibit dual responsiveness to both ROS and esterase. This dual-responsive design offers a significant advantage by reducing reliance on any single tumor microenvironment cue, thereby enhancing the reliability of drug release across heterogeneous tumors with varying ROS levels or esterase expression.

Fig. 3.

In vitro release profiles of BP@FAPG micelles in different media.

3.6. Anticancer activity in vitro

The anticancer activity of the PTX prodrug and micelles was evaluated by comparing its cytotoxic effects on MDA-MB-231 cancer cells and L929 normal fibroblast cells. The experimental data indicated that all tested formulations exhibited excellent safety profiles in normal cells (Fig. 4a). Even at high concentrations (100 μM), the viability of cells treated with BTBP-PTX, BP@GAPG, or BP@FAPG remained above 95%, demonstrating good biocompatibility. In contrast, a clear concentration-dependent anticancer effect was observed in MDA-MB-231 cells, with cell viability decreasing significantly across the tested concentration range (Fig. 4b). Notably, BP@FAPG demonstrated the most potent cytotoxicity, outperforming both the free prodrug and the non-targeted counterpart BP@GAPG. This result highlights the superior therapeutic selectivity and efficacy of the actively targeted micellar system, which enhances cancer cell killing while reducing the off-target toxicity typically associated with free PTX.

Fig. 4.

(a) Cell viability of L929 cells after coincubation with BP@FAPG, and BP@GAPG; (b) Cell viability of MDA-MB-231 cells after coincubation with PTX, BTBP-PTX, BP@FAPG, and BP@GAPG. *p < 0.05, **p < 0.01.

3.7. Cellular uptake

To validate the cellular delivery efficiency of the BP@FAPG micellar system, we conducted qualitative and quantitative analyses of the uptake behavior of four different formulations in MDA-MB-231 cells (Fig. 5a). Confocal images revealed that free BTBP and free BTBP-PTX exhibited only faint green fluorescence with sparse intracellular distribution, indicating that the free form or prodrug was inefficient at entering cells and triggering aggregation. In contrast, the micellar formulations showed significantly enhanced intracellular fluorescence intensity, with a denser and widespread distribution. Notably, the folate-decorated targeting micelles displayed the brightest and most extensive green fluorescence coverage. Quantitative analysis confirmed that the fluorescence intensity of BP@FAPG was over 3-fold higher than that of BP@GAPG and significantly exceeded that of free BTBP (Fig. 5b). These results confirm that the micellar system dramatically improves intracellular delivery and AIEgen aggregation efficiency, and that folate modification further enhances cellular uptake and internalization via an active targeting mechanism. Compared to non-targeted micelles or free prodrugs, the FR-α-mediated endocytosis achieved by BP@FAPG represents a key advantage for tumor-specific drug delivery.

Fig. 5.

(a) CLSM images of MDA-MB-231 cells treated with free BTBP, BTBP-PTX, BP@GAPG, and BP@FAPG micelles; (b) Semiquantitative analysis of the fluorescence intensity by ImageJ software (n = 3). ***p < 0.001, ****p < 0.0001.

3.8. Live/dead cell imaging

The live/dead cell imaging results visually confirmed the anticancer efficacy and biocompatibility of BP@FAPG (Fig. 6). In MDA-MB-231 cells, treatment with BP@FAPG resulted in a markedly higher number of dead cells (red fluorescence from PI staining) compared to free drugs and non-targeted micelles, demonstrating its superior cytotoxic potency. Conversely, in normal L929 cells treated with BP@FAPG, both green fluorescence (from BTBP) and red fluorescence (from PI, indicating dead cells) were barely visible, with the vast majority of cells remaining alive. This observation further confirms the excellent biocompatibility and low off-target toxicity of the BP@FAPG system. The enhanced anticancer effect in MDA-MB-231 cells is directly attributable to the targeting-mediated efficient cellular uptake and subsequent drug release. Furthermore, when comparing the effects of BP@FAPG across different cell lines, the extent of cell death induced in MDA-MB-231 cells was significantly greater than that in L929 control cells. This distinct difference visually demonstrates its targeting specificity. Compared to conventional PTX formulations that lack such targeting capability, our BP@FAPG system offers a clear advantage in selectively killing cancer cells while sparing normal cells.

Fig. 6.

Fluorescence imaging of MDA-MB-231 cells incubated with different formulations. Nuclei were counterstained with Hoechst 33342 (λex = 405 nm, λem = 450 nm). The BTBP signal was acquired under 405 nm excitation with emission collected in the 480–550 nm range, while propidium iodide (PI) was excited at 488 nm with emission detected between 630 and 680 nm. Scale bar: (10×).

3.9. Evaluation of anticancer efficacy in a 3D tumor spheroid model

To assess therapeutic potential within a more complex, in vivo-like microenvironment, we employed a 3D tumor spheroid model for further validation. As shown in Fig. 7, treatment with free BTBP-PTX or non-targeted BP@GAPG micelles induced only limited fluorescence signals at the spheroid periphery and minor morphological changes. In striking contrast, treatment with BP@FAPG led to significant disruption of the spheroid's structural integrity and induced the most pronounced anticancer effect. Widespread red fluorescence (from PI staining) co-localized with intense and uniform green BTBP signal, confirming that FR-α-mediated active targeting improved micelle penetration and accumulation within the spheroid, and that efficient drug release subsequently induced large-scale apoptosis. These results provide a strong rationale for developing tumor-targeted delivery systems that integrate efficient therapy with real-time visualization capabilities.

Fig. 7.

Fluorescence images of 3D tumor spheroid cultured with BTBP, BTBP-PTX, BP@GAPG, and BP@FAPG stained with BTBP and PI dyes. The BTBP signal was acquired under 405 nm excitation with emission collected in the 480–550 nm range, and PI was excited at the wavelength of 488 nm, and emission wavelengths were at 630–680 nm. Scale bar: (10×).

4. Conclusion

In summary, we have successfully developed a novel micelle system, BP@FAPG, that integrates FR-α-mediated active targeting, esterase/ROS dual-responsive drug release, and AIE-based real-time release monitoring into a single platform. This system not only mitigates systemic toxicity associated with conventional paclitaxel formulations but also overcomes the limitations of single-stimulus responsiveness and non-visualizable drug release. Through comprehensive characterization and cellular evaluation, the carrier demonstrated superior surface activity, stimuli-responsive release, enhanced cellular uptake, potent cytotoxicity (IC₅₀ = 60 nM against MDA-MB-231 cells), and deep tumor spheroid penetration—attributes predominantly governed by FR-α-mediated endocytosis. Importantly, the ROS-triggered activation of BTBP provided real-time, in situ fluorescence tracking of drug release, establishing a self-reporting mechanism that bridges therapeutic action and imaging feedback. This integrated design represents a significant advancement over conventional drug delivery systems, offering a robust and programmable strategy for targeted paclitaxel delivery within the complex tumor microenvironment.

Future work will focus on in vivo validation of its targeting efficacy, biosafety, and theranostic performance, as well as exploring its combinatory potential with other treatment modalities such as immunotherapy or photodynamic therapy. Additionally, we plan to develop novel fluorophores based on the BTBP scaffold by replacing the bromine atom with other substituents to systematically investigate the structure-property relationship, which would enable deeper tissue penetration and improved imaging performance for in vivo applications. Such efforts will further enhance the versatility and translational potential of our micelle platform. Ultimately, this system provides a versatile blueprint for the next generation of intelligent nanocarriers that are capable of precise, trackable, and personalized cancer therapy.

CRediT authorship contribution statement

Jindong Li: Writing – original draft, Validation, Resources, Methodology, Investigation. Xinyu Zhang: Writing – review & editing, Validation, Resources, Methodology, Conceptualization. Ting Guo: Validation, Resources, Conceptualization. Ziliang Liu: Validation, Investigation, Conceptualization. Huiling Zhou: Validation, Investigation. Fang Zhang: Validation, Investigation. Yan Zhang: Writing – review & editing, Funding acquisition. Shengnan Yin: Supervision, Resources, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors 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

Supplementary material

Data availability

Data will be made available on request.