Mucoadhesive tumor-penetrating nanomedicine for intravesical chemo-immunotherapy against bladder cancer

Authors: Changhao Zhao, Dayong Hou, Kaiwei Wang, Jianjiang Chen, Yumin Wu, Nanhui Liu, Zhen Wang, Yanbin Liu, Yaowei Li, Zhishuai Zhang, Yi Liu, Zhuyu Hou, Xiaodan Yang, Xiao Liu, Wanhai Xu, Ziqi Wang, Han Zhang, Zhuang Liu

Abstract

Intravesical administrations of chemotherapeutics or immune-agonist Bacillus Calmette-Guérin (BCG) are first-line treatments for non–muscle-invasive bladder cancer (NMIBC). However, while urination prevents drug retention in the bladder, the bladder mucus, epithelial barrier, and dense tumor stroma form multiple physical barriers that restrict intratumoral drug penetration. Here, we developed thiol-functionalized bacterial membrane–coated nanoparticles loaded with a chemotherapeutic agent epirubicin (EPI) for highly effective intravesical chemo-immunotherapy against bladder tumors. The surface thiol groups enabled urine-resistant adhesion to the mucin-rich bladder mucosa through dynamic disulfide bonds, as demonstrated in both mouse and human bladders. Meanwhile, we unexpectedly found that bacterial membrane components up-regulated matrix metalloproteinases (MMPs), facilitating tight junction disruption and collagen degradation, thereby enhancing nanoparticle penetration into tumors. Intratumoral delivery of EPI by such nanomedicine would induce robust immunogenic cell death (ICD), which by synergizing with the immunoadjuvant properties of the bacterial membrane can elicit tumor-specific immune responses, resulting in potent antitumor efficacy in both NMIBC and muscle-invasive bladder cancer (MIBC) mouse models. Notably, combination with immune checkpoint blockade further amplified systemic antitumor immunity, leading to complete regression of orthotopic bladder tumors and marked inhibition of distant lesions. Our unique nanomedicine platform by addressing challenges in current intravesical therapies would be highly promising for potent intravesical chemo-immunotherapy of bladder malignancies.

INTRODUCTION

Bladder cancer is one of the most prevalent genitourinary malignancies, with more than 600,000 new cases diagnosed worldwide each year (1, 2). Based on the extent of tumor invasion into the bladder wall, bladder cancer is classified into non–muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC). In clinical practice, ~75% of bladder cancer cases are initially diagnosed as NMIBC (3, 4). Now, the standard treatment for NMIBC involves transurethral resection of the bladder tumor followed by regular intravesical treatment. First-line intravesical agents include chemotherapeutic drugs such as epirubicin (EPI) and gemcitabine and immune agonist Bacillus Calmette-Guérin (BCG). Despite these therapeutic interventions, 50 to 70% of patients experience tumor relapse within 5 years, highlighting the clinical challenge of tumor recurrence, especially for those with high-risk NMIBC (4–8). Unfortunately, a large proportion of patients with NMIBC would undergo pathological progression to MIBC later (9). For those patients with MIBC, radical cystectomy remains the definitive treatment, which, however, would irreversibly impair the quality of life (10). More unfortunately, many patients with MIBC would undergo further progression to systemic metastases to other organs, resulting in a remarked reduced 5-year survival rate to be as low as ~5% (9). Therefore, the best strategy is to fully control bladder cancer at the NMIBC stage. Developing more potent intravesical therapy is thus critical for bladder cancer treatment.

The limited therapeutic efficacy of conventional intravesical instillation treatments is attributed to the rapid drug clearance as well as the physiological barrier of the bladder mucosa. Conventional instillation agents are rapidly eliminated from the bladder during the first voiding of urine, typically resulting in the clearance of over 90% of the instilled drugs within 2 hours (11–15). Additionally, the bladder mucosa–covered urothelium and the dense tumor stroma constitute the barriers that impede drug penetration into the tumor (11, 12, 16, 17). Recent advances in nanocarrier engineering have attempted to enhance the intravesical drug delivery, primarily through mucoadhesive cationic nanoparticles or urease-powered nanobots/nanomotors. Cationic carriers, such as chitosan and cationic liposomes, can electrostatically interact with the negatively charged mucosal surface to enhance bladder retention (18–20). However, these noncovalent interactions are inherently weak and unstable, resulting in poor adhesion under urine flow and inefficient intratumoral drug delivery. Recently, urease-powered nanobots, which propel themselves by catalyzing urea into gas, offer improved bladder penetration (21–24). Nonetheless, those nanobots still can only stay in the bladder for 1 to 2 hours (during instillation) without the ability for long-term retention. Moreover, these strategies achieve tumor penetration depths of only a few hundred micrometers, which remain inadequate to meet clinical needs (21–23). Therefore, there is still a critical need for innovative strategies that can simultaneously overcome the clearance and penetration barriers of the bladder to optimize therapeutic outcomes in bladder cancer treatment.

On the other hand, while intravesical chemotherapy is normally used to treat low-risk and middle-risk NMIBC, intravesical BCG instillation is the first choice for high-risk NMIBC treatment (4, 5). Certain types of chemotherapeutic agents (e.g., anthracyclines such as EPI) can kill tumor cells by inducing immunogenic cell death (ICD), which by releasing damage-associated molecular patterns (DAMPs) and exposing tumor antigens could trigger tumor-specific immunity (25, 26). On the other side, for BCG immunotherapy, BCG bacterial components are recognized by antigen-presenting cells through pattern recognition receptors (PRRs), triggering the release of proinflammatory cytokines and chemokines and promoting the infiltration of immune effector cells (27, 28). Building on this understanding, the combination of ICD-inducing chemotherapeutics and immune agonists may generate synergistic effects, with chemotherapeutics to expose tumor antigens and immune agonists to amplify the subsequent immune response. Unfortunately, this combination therapeutics has not yet been realized in clinic for intravesical bladder cancer treatment.

Bacterial membrane vesicles and bacterial membrane–coated nanoparticles, enriched with pathogen-associated molecular patterns (PAMPs), have emerged as a promising type of biomaterial with inherent immunoadjuvant properties (29, 30). In this work, we developed thiol-functionalized Escherichia coli membrane–coated poly(lactic-co-glycolic acid) (PLGA) nanoparticles with loading of chemotherapeutic agent EPI (SH-EM-NP/EPI). It was revealed that thiol groups on the nanoparticle surface conferred urine-resistant mucoadhesion via forming dynamic disulfide bond with the mucin-rich bladder mucosa. Meanwhile, native bacterial membrane components promoted tumor penetration by up-regulating matrix metalloproteinases (MMPs), which degrade tight junctions and stromal collagen fibers within the tumor. As the results, such SH-EM-NP exhibited excellent mucoadhesive performance on both human and murine bladder mucosa and demonstrated markedly increased tumor accumulation with >2-mm penetration into bladder tumor tissues, a depth much larger than that achieved in previous reports using nanomotors (21–23). The delivered EPI would then induce robust immunogenic death of tumor cells, synergizing with the immunoadjuvant properties of the bacterial membrane to amplify antitumor immune responses. Compared with clinically used EPI or BCG, SH-EM-NP/EPI exhibited superior therapeutic efficacies, achieving tumor inhibition rates of 92.9 and 92.1% in NMIBC and MIBC mouse models, respectively. Excitingly, intravesically administrated SH-EM-NP/EPI by combining with anti–programmed death-ligand 1 (αPD-L1) immune checkpoint blockade (ICB) could elicit strong systemic antitumor immunity, resulting in complete regression of orthotopic (o.t.) bladder tumors and 93.4% inhibition of distant tumor growth. The bacterial membrane–based nanomedicine developed in this work is able to simultaneously address the two challenges in intravesical therapies, the limited intravesical drug retention and poor tumor penetration, while triggering both ICD and immune activation, particularly suitable for treatment of bladder cancer with substantial translational potential.

RESULTS

Preparation and characterization of SH-EM-NP/EPI

To prepare SH-EM-NP/EPI, EPI-loaded PLGA nanoparticles (PLGA-NP/EPI) were initially synthesized using a double-emulsion solvent evaporation method (31). Then, the extracted bacterial membranes derived from E. coli were coated on those nanoparticles (EM-NP/EPI) via coextrusion. Last, SH-modified nanoparticles (SH-EM-NP/EPI) were prepared by inserting 1,2-distearoyl-sn-glycero-3-phosphoethanolamine–polyethylene glycol 2000–thiol (DSPE-PEG2000-SH) into the membrane (Fig. 1A). Dynamic light scattering measurements demonstrated that the hydrodynamic diameter of EM-NP/EPI increased from 84 nm (for PLGA-NP/EPI) to 96 nm, accompanied by a shift in zeta potential from −17 to −24 mV, indicating successful coating of the PLGA-NP/EPI cores with bacterial membranes. Following modification with DSPE-PEG2000~-SH, the average diameter of SH-EM-NP/EPI increased to 108 nm, and their zeta potential rose to −18 mV (Fig. 1B and fig. S1). These changes can be attributed to the increased hydrodynamic nanoparticle size and the charge-shielding effect by DSPE-PEG2000-SH modification. Transmission electron microscopy (TEM) revealed a core-shell structure for those SH-EM-NP/EPI nanoparticles (Fig. 1C). SH-EM-NP/EPI nanoparticles exhibited good stability in mouse urine, showing no obvious change in their hydrodynamic diameters after 24 hours of incubation within mouse urine (fig. S2). Meanwhile, PLGA-NP, EM-NP, and SH-EM-NP exhibited comparable nanoparticle sizes and zeta potentials as their EPI-loaded counterparts (fig. S3).

Next, the degree of thiolation on SH-EM-NP/EPI was quantified using 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB; Ellman’s reagent). As presented in Fig. 1D, the free thiol content of SH-EM-NP/EPI was obviously higher than that of EM-NP/EPI, confirming successful thiol modification. As the bacterial membrane contains a small amount of endogenous cysteine, EM-NP/EPI also featured a low level of thiols. Subsequently, we evaluated the reactivity of SH groups on SH-EM-NP/EPI based on a click reaction between maleimide and SH (Fig. 1E) (32). As illustrated in Fig. 1F (left), EM-NP/EPI exhibited only a weak fluorescent signal after 2 hours of coincubation with Cyanine7–maleimide (Cy7-Mal), and this signal was attributable to the reaction between endogenous thiol groups present on the bacterial membranes and the maleimide moiety of the fluorescent dye (33). Conversely, SH-EM-NP/EPI displayed a markedly stronger fluorescence signal after reacting with Cy7-Mal (Fig. 1F, right), confirming the successful modification of thiol groups on SH-EM-NP/EPI and validated their chemical reactivity.

To evaluate the interbatch stability of membrane protein composition, quantitative proteomic profiling was performed using SH-EM-NP/EPI samples prepared from three independent batches (NP-1 to NP-3). The heatmap visualized the abundance profiles of the top 50 high-abundance membrane proteins across the three batches (Fig. 1G). Both the heatmap and Venn diagram revealed a high degree of consistency in membrane protein expression profiles, demonstrating the reproducibility and reliability of protein composition across different batches of samples (Fig. 1G and fig. S4).

Powdered formulations of biological products facilitate long-term storage and preserve activity (34). Next, the lyophilized powder of SH-EM-NP/EPI was prepared with cryoprotectant. The freshly prepared SH-EM-NP/EPI (EPI, 2 mg/ml) solution in 10% sucrose was lyophilized into dry powders and stored in nitrogen-filled glass vials to prevent oxidation. The formulation process was rather simple, which would permit easy scale-up. The resulting lyophilized formulations could be readily resuspended in sterile water (Fig. 1H). As presented in Fig. 1I and fig. S5, there were no obvious differences in size and zeta potential between the freshly prepared SH-EM-NP/EPI samples and resuspended ones, indicating that lyophilization had a negligible impact on the nanoparticle structure. Moreover, the reconstituted solution of SH-EM-NP/EPI displayed a characteristic core-shell morphology under TEM, consistent with that of the freshly prepared sample (fig. S6). In addition, the free thiol level quantified using DTNB was comparable between the freshly prepared sample and reconstituted solution, indicating that free thiol was protected against oxidation (Fig. 1J). These findings highlight the robustness of the lyophilized SH-EM-NP/EPI formulation and its potential for future clinical translation.

Bladder retention and mucoadhesive of SH-EM-NP

Intravesical instillation therapy is the standard of care for bladder tumor treatment. Nevertheless, the efficiency of intravesical drug delivery is substantially compromised by rapid urinary elimination. Therefore, engineering nanoplatforms to enhance bladder retention and improve drug exposure in bladder lesions are of great importance. To screen the optimal nanoparticles that can better enhance bladder adhesion, three formulations with distinct membrane protein/DSPE-PEG2000-SH mass ratios (1:1, 1, and 1) were synthesized for intravesical instillation. Nanoparticles prepared at the 1 ratio exhibited the strongest bladder adhesion, as evidenced by significantly higher fluorescence intensity compared with nanoparticles prepared at the other two ratios (fig. S7). Therefore, we selected the 1 ratio for subsequent experiments. Previous studies found that cationic nanoparticles could improve mucosal retention through electrostatic interactions with the negatively charged mucosa (19). Herein, we systematically compared the bladder retention performance of thiol- and amine-functionalized nanoparticles. We prepared 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine,4-chlorobenzenesulfonate salt (DiD)–labeled EM-NP, DSPE-PEG2000-NH2–modified nanoparticles (NH2-EM-NP), and DSPE-PEG2000-SH–modified nanoparticles (SH-EM-NP) and ensured equivalent fluorescence intensities across the three formulations. After intravesical instillation of the different formulations, time-dependent bladder retention was evaluated at 0.5, 3, 6, 12, and 24 hours by ex vivo fluorescence imaging (Fig. 2A). As shown in Fig. 2B (left), EM-NP exhibited rapid signal decay over time, indicating limited mucosal interaction and fast clearance by urinary flushing. In contrast, NH2-EM-NP displayed moderately improved bladder retention, while SH-EM-NP demonstrated the strongest and most sustained intravesical fluorescence across all time points. Quantitative analysis further confirmed this trend (Fig. 2B, right). At 24 hours postadministration, the fluorescence intensity in the SH-EM-NP group was 2.5-fold and 5-fold higher than that observed in the NH2~-EM-NP and EM-NP groups, respectively. Notably, the gradual decline in fluorescence intensity of SH-EM-NP over time revealed a clear time-dependent de-adhesion profile, consistent with dynamic thiol-disulfide exchange mediated covalent interactions between surface thiol groups and the mucin-rich bladder mucosa.

Almost all therapeutic agents administered via intravesical instillation are rapidly eliminated during the first voiding of urine because of their insufficient adhesion to the bladder wall (14). Thus, desirable bladder retention likely requires enhanced adhesion of the therapeutic agents to the bladder mucosa. To explore this, we performed confocal laser scanning microscopy (CLSM) to visualize the fluorescent signals of different DiD-labeled nanoparticles on the bladder wall following urine washing (Fig. 2C). As expected, while both NH2-EM-NP and SH-EM-NP showed stronger adhesion on the bladder walls than the nonmucoadhesive EM-NP, SH-EM-NP exhibited the strongest bladder wall adhesion ability among all nanoparticles (Fig. 2C and fig. S8). This finding indicates that nanoparticles with surface thiolation could effectively adhere to the bladder mucosa, helping to prevent their rapid clearance.

Considering the abundant disulfide bonds present in mucins, we hypothesized that the enhanced mucosa adhesion of SH-EM-NP primarily occurs through a dynamic thiol-disulfide exchange mechanism between thiol groups on SH-EM-NP and the disulfide bonds in mucins. To verify this hypothesis, Cy7-labeled mucin (Cy7_mucin) was coincubated with EM-NP and SH-EM-NP in solution. After 2 hours of reaction followed by centrifugation and washing, SH-EM-NP exhibited markedly higher fluorescence intensity than EM-NP, indicating the successful conjugation of Cy7_mucin onto SH-EM-NP (fig. S9). To confirm the involvement of thiol groups in bioconjugation, we used the click reaction between Cy7-Mal and thiol groups to monitor changes in the remaining free thiol groups’ content after the interaction between SH-EM-NP and mucin (32, 35). As expected, the thiol content (as measured by Cy7 fluorescence) decreased substantially following incubation with mucin, demonstrating thiol consumption during the reaction (fig. S10). To further validate the formation of disulfide bonds between SH-EM-NP and mucin, the disulfide bond reducer dithiothreitol (DTT) or l-ascorbic acid was added to the mixture before reaction as thiol-disulfide exchange could be dampened by reductive agents (36). Notably, in the presence of DTT, the fluorescence intensity of SH-EM-NP was strongly reduced upon reaction with Cy7_mucin (fig. S9). A similar decrease was observed when DTT was replaced with l-ascorbic acid (fig. S11). Namely, the conjugation of Cy7_mucin with SH-EM-NP was inhibited by both DTT and l-ascorbic acid, which can be attributed to blockade of the thiol-disulfide exchange reaction by reductants.

We further investigated the bladder adhesion behaviors of SH-EM-NP using human and mouse bladders. After immersing bladders within SH-EM-NP or EM-NP solutions for 2 hours followed by three washes with artificial urine (10 min each), notable fluorescence signals were observed from the mucosal layer of both human and mice bladder tissues treated with SH-EM-NP, whereas EM-NP–treated bladders exhibited rather weak fluorescence signals (Fig. 2, D and E). Meanwhile, in the presence of DTT, the fluorescence signal of SH-EM-NP on the bladder mucosal layer was significantly reduced, indicating that reductants could disrupt SH-EM-NP adhesion to the bladder mucosal surface (Fig. 2, D and E). These results indicate that SH-EM-NP adhesion to the bladder mucosa is mediated by dynamic thiol-disulfide exchange. The covalent interaction triggered by thiol-disulfide exchange enables the nanoparticles to effectively adhere to the bladder mucosal surface and protects them from rapid clearance by urine (Fig. 2F). Notably, this thiol-disulfide exchange–mediated adhesion is dynamic and reversible rather than permanent, thereby providing an initial anchoring effect for prolonged bladder retention. Furthermore, this mechanism demonstrates cross-species applicability, highlighting its promising potential for clinical translation.

Intratumoral penetration mechanistic study of SH-EM-NP

In addition to enhancing drug delivery through mucoadhesion, effective tumor penetration plays a pivotal role in intravesical bladder cancer treatment. To assess the tumor permeability of SH-EM-NP, we established an o.t. bladder tumor model in mice to conduct in vivo experiments (Fig. 3A). Cy5.5-labeled PLGA-NP, EM-NP, and SH-EM-NP were instilled into the bladders twice. At 24 hours postinstillation, bladder tumors were extracted and analyzed using confocal imaging (Fig. 3A). As presented in Fig. 3B, minimal fluorescent signals within tumor tissues were observed in the PLGA-NP group. In comparison, the EM-NP group displayed a more diffused fluorescence pattern, which penetrated slightly deeper into tumor tissues. Excitingly, SH-EM-NP were broadly distributed throughout tumor tissues, exhibiting the strongest intratumoral fluorescence signals penetrated into the tumor at a depth exceeding 2 mm. Notably, in clinical practice, a tumor penetration depth of 2 mm would be sufficient to infiltrate postoperative residual lesions, considering the thicknesses of human bladder mucosa and bladder wall are approximately at 100 μm and 2 mm, respectively (37, 38).

To further investigate the underlying mechanisms driving the enhanced tumor penetration of SH-EM-NP, we performed transcriptome sequencing [RNA sequencing (RNA-seq)] and quantitative proteomic analysis of bladder tumor tissues with or without SH-EM-NP treatment. Using the volcano plot, 65 up-regulated and 67 down-regulated genes were identified between the two groups (fig. S12). To investigate the functional roles of these regulated genes, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and Gene Ontology (GO) enrichment analyses of regulated genes were conducted. As illustrated in Fig. 3 (C and D), we selected 24 pathway entries presented in the plot, which could be divided into four main categories, namely, immune signaling pathways, inflammatory responses, immune responses, and extracellular matrix (ECM). With treatment of SH-EM-NP, crucial inflammatory pathways were enriched, such as the nuclear factor κB (NF-κB) signaling pathway, inflammatory responses, the tumor necrosis factor (TNF) signaling pathway, and interleukin-6 (IL-6) and cytokine-mediated signaling pathways. In GO enrichment analysis, several ECM-related biological processes, including ECM organization and collagen catabolic processes, were also markedly enriched. Moreover, several crucial immune pathways were significantly up-regulated, including antigen processing and presentation, PRR signaling [such as Toll-like receptors (TLRs) and nucleotide-binding oligomerization domain–like receptors (NOD-like receptors)], natural killer cell–mediated cytotoxicity, and chemokine/cytokine signaling involved in the immune activation.

Next, we conducted quantitative proteomic profiling to achieve a more comprehensive characterization of protein expression changes. The quantitative proteomic evaluations identified 56 up-regulated and 139 down-regulated proteins (fig. S13). As demonstrated in Fig. 3 (E and F), we selected 28 pathways from the KEGG and GO enrichment analyses. These pathways could be categorized into three major functional ECM remodeling, inflammatory responses, and immune responses. With treatment of SH-EM-NP, the NF-κB signaling pathway and TLR signaling pathway were significantly enriched, aligning well with the results of RNA-seq. Meanwhile, pathways related to tight junction, cell adhesion, ECM organization, collagen catabolic processes, and tissue remodeling were also significantly enriched (Fig. 3, E and F). Specifically, SH-EM-NP–treated tumors exhibited significant down-regulation of ECM-related proteins, including collagens (e.g., CO4A1 and CO4A2), laminins (e.g., LAMC1 and LAMB1), and proteoglycans (e.g., PGS1 and PGS2), as well as proteins involved in cell junctions and adhesion, such as tight junction–associated proteins (e.g., CADH2 and CAD13) and integrins (e.g., ITA5 and ITA1). Notably, matrix-degrading enzymes including MMPs (MMP-2 and MMP-9) and a disintegrin and metalloproteinase (ADAM9 and ADA17) were significantly up-regulated (Fig. 3G). Based on RNA-seq and proteomic analyses, we propose that the enhanced intratumoral penetration of SH-EM-NP is primarily driven by the bacterial membrane–induced activation of inflammatory signaling, which would then lead to up-regulated matrix-degrading enzymes such as MMP to degrade ECM and tight junction protein, consequently creating a more permeable matrix to facilitate deeper nanoparticle infiltration.

To further understand the unique role of bacterial membranes in enhancing tumor penetration of those SH-EM-NP nanoparticles, we used thiol-modified red blood cell (RBC) membrane–coated nanoparticles (SH-CM-NP) as another control for comparative analysis. RBC membranes were selected as a biologically derived but immunologically inert membrane control as they lack PAMPs and do not actively induce inflammatory responses. We first established a multicellular spheroid model to mimic the architecture of tumor tissues in vitro (fig. S14, left). Then, different Cy5.5-labeled nanoparticles were separately coincubated with MB49 tumor spheroids for 2 hours, followed by incubation in fresh serum-free medium for an additional 24 hours. As presented in fig. S14, the fluorescence signals of PLGA-NP and CM-NP were predominantly confined to the periphery of the tumor spheroids. In comparison, EM-NP and SH-CM-NP exhibited a more diffused fluorescence pattern, with fluorescence penetrating slightly deeper into the spheroids. The signals from SH-EM-NP achieved the greatest penetration into the interior of tumor spheroids. Based on these findings, we inferred that the coated bacterial membrane plays a critical role in enhancing tumor penetration, while surface thiolation would further endow nanoparticles with prolonged adhesive properties, thereby facilitating even deeper tumor penetration.

It is well-known that the epithelial mucosal barrier is principally maintained by tight junctions (13, 20). Thus, to validate the proposed possible mechanism of the enhanced tumor penetration of SH-EM-NP, we examined whether these nanoparticles could disrupt tight junctions and modulate the integrity of the bladder epithelial barrier. Initially, immunofluorescence staining was conducted to examine tight junction–associated proteins such as E-cadherin and occludin (OCC) on SV-HUC-1 cell monolayers. After treatment with CM-NP, SH-CM-NP, EM-NP, and SH-EM-NP for 8 hours, CLSM revealed that the bacterial membrane–based EM-NP and SH-EM-NP induced marked reductions and discontinuity in E-cadherin and OCC staining, indicating disruption of tight junction integrity (fig. S15). The tight junctions in cells exposed to CM-NP and SH-CM-NP, which lacked bacterial membrane components, remained intact, similar to the findings in the control group (fig. S15).

Subsequently, we performed a Transwell monolayer model assay using SV-HUC-1 and MB49 cells to measure transepithelial electrical resistance (TEER), a well-established indicator of the fluctuation of tight junctions (20, 39). After exposing the apical side of the cell monolayer to different nanoparticles for 8 hours, TEER was measured using a volt-ohm meter (Fig. 3H, left). Both EM-NP and SH-EM-NP significantly reduced TEER in SV-HUC-1 and MB49 cell monolayer models, indicating disruption of tight junctions induced by those bacterial membrane–coated nanoparticles. In contrast, CM-NP and SH-CM-NP had no significant effect on TEER values (Fig. 3H, right, and fig. S16). After replacing the nanoparticle-loaded medium with fresh medium at 8 hours and continuing incubation, the TEER values in EM-NP and SH-EM-NP groups returned to baseline levels at 24 hours, suggesting gradual recovery of cell tight junctions in the absence of those nanoparticles. Meanwhile, we observed significantly elevated levels of TNF-α, IL-6, MMP-2, and MMP-9 in the supernatants of SV-HUC-1 and MB49 cells treated with EM-NP and SH-EM-NP for 8 hours, and those levels returned to the normal range after removal of nanoparticles and incubation in fresh cell medium (24 hours), indicating that the bacterial membrane–containing nanoparticles could activate inflammatory responses and up-regulate MMPs (Fig. 3I and fig. S17), consistent with the above RNA-seq and proteomic data.

To further verify a causal role of MMP activity in mediating enhanced tumor penetration, we next performed an in vivo MMP inhibition study using the broad-spectrum MMP inhibitor GM6001. Cy5.5-labeled SH-EM-NP were intravesically administered to tumor-bearing mice in the presence or absence of GM6001, followed by tumor sectioning and fluorescence imaging to assess intratumoral nanoparticle distribution. As shown in Fig. 3J, inhibition of MMPs markedly diminished nanoparticle penetration into the tumor tissue. Compared with the SH-EM-NP group, in which nanoparticles distributed deeply throughout the tumor parenchyma, GM6001-treated tumors exhibited markedly reduced fluorescence intensity inside the tumor, with nanoparticles mainly localized at the superficial tumor regions. These results demonstrate that bacterial membrane–induced MMP up-regulation plays an important role in facilitating intratumoral nanoparticle penetration.

Building on these findings, a potential mechanism underlying the enhanced intratumoral penetration of SH-EM-NP is proposed in Fig. 3K. Bladder tumor tissues, characterized by densely packed collagen fibers and matrix proteins, along with an epithelial barrier formed by tight junctions of the bladder epithelium, collectively constitute physical barriers that limit nanoparticle penetration (11, 20, 40). The thiolated surface of SH-EM-NP enables their dynamic and reversible adhesion to the bladder mucosal surface, providing both temporal and spatial advantages for efficient tissue penetration while allowing subsequent nanoparticle detachment and penetration into underlying tumor tissue upon tumor microenvironmental (TME) remodeling. This prolonged retention allows the PAMPs present in the bacterial membranes of SH-EM-NP to be recognized by TLRs on both the bladder epithelium and cells within the TME, subsequently leading to the activation of classical inflammatory pathways, such as TLR/NF-κB signaling, which then induce robust secretion of proinflammatory cytokines (e.g., TNF-α and IL-6) and matrix-degrading enzymes (e.g., MMP-2 and MMP-9). These inflammatory mediators by promoting the degradation of stromal collagen fibers and tight junction proteins would ultimately create a more permeable TME and enhance intratumoral penetration of nanoparticles.

Intravesical instillation of SH-EM-NP/EPI to treat bladder cancer

Encouraged by the sustained bladder retention and deep tumor penetration of SH-EM-NP, we further validated their therapeutic potential for bladder cancer treatment. NMIBC represents the most prevalent subtype of bladder cancer, accounting for ~75% of all clinical cases (3). We first evaluated the therapeutic efficacy of SH-EM-NP/EPI using an NMIBC model, which was established by intravesically engrafting aggressive MB49-luciferase (MB49-Luc) bladder cancer cells into bladders of C57BL/6J mice (41). Five days after implantation, multiple tumors were observed by histological analysis in the urothelial layer with no evidence of muscle layer invasion (fig. S18). Then, mice carrying NMIBC tumors were randomly divided into six control (saline), E. coli–derived cytoplasmic membranes [(EM) protein concentration, 2 mg/ml], EPI (1 mg/ml), BCG (1 mg/ml), EM-NP/EPI (1 mg/ml EPI), and SH-EM-NP/EPI (1 mg/ml EPI). Each group received intravesical instillation treatment (50 μl per mouse) every 4 days for four times. Tumor growth progression was tracked by measuring the bioluminescence signals of MB49-Luc cells on days 0, 4, 8, 12, and 14 postimplantation (Fig. 4A). As illustrated in Fig. 4 (B to D), negligible bioluminescence was observed on day 14 in mice treated with SH-EM-NP/EPI, whereas mice in the other groups exhibited prominent bioluminescence signals. Although BCG and EPI, the first-line intravesical therapeutics, delayed bladder tumor progression to some extent, their efficacy was significantly inferior to that of SH-EM-NP/EPI. Based on the tumor average bioluminescence intensities, SH-EM-NP/EPI exhibited the most significant therapeutic efficacy, achieving a tumor growth inhibition (TGI) rate of 93% (Fig. 4D). Notably, SH-EM-NP/EPI treatment significantly improved survival, extending the median survival time of mice from 18 days (control group) to 54 days (Fig. 4E). In addition, no significant weight loss was observed in SH-EM-NP/EPI–treated mice compared with other groups, which may be attributed to the minimal tumor burden and reduced treatment-associated toxicity in this group of mice (Fig. 4F).

Compared with NMIBC, MIBC exhibits markedly higher malignant potential, with deeper tissue infiltration and increased metastatic propensity (42). Encouraged by the promising therapeutic efficacy of SH-EM-NP/EPI in NMIBC, we further validated their potential for treating MIBC. To establish a murine model of MIBC, MB49-Luc cells were injected directly into the bladder wall of C57BL/6J mice. Three days postimplantation, histological analysis confirmed marked tumor infiltration into the bladder muscle layer (fig. S19). The MIBC-bearing mice were then randomly assigned to five control (saline), EM (protein concentration, 2 mg/ml), EPI (1 mg/ml), EM-NP/EPI (1 mg/ml EPI), and SH-EM-NP/EPI (1 mg/ml EPI). The treatment protocol is presented in Fig. 4G. SH-EM-NP/EPI significantly inhibited MIBC tumor growth, with this group exhibiting weaker bioluminescence signals on day 14 than the other groups (Fig. 4, H to J). After completing the treatment regimen, bladder tumor tissues were resected and weighed, revealing the lowest tumor burden in the SH-EM-NP/EPI group (Fig. 4, K and L). Specifically, the TGI rates were 92.1, 66.4, 43.7, and 12% in the SH-EM-NP/EPI, EM-NP/EPI, EPI, and EM groups, respectively. Together, SH-EM-NP/EPI exhibited robust antitumor efficacy in both NMIBC and MIBC models.

Next, we evaluated the in vivo biosafety of SH-EM-NP/EPI. In previous two types of tumor model experiments, no significant body weight loss was observed in mice treated with SH-EM-NP/EPI (Fig. 4F and fig. S20). The levels of key serum biomarkers, including alanine transaminase, aspartate transaminase, alkaline phosphatase, serum creatinine, and blood urea nitrogen, remained within normal ranges (fig. S21). Additionally, hematoxylin and eosin (H&E) staining of major organs, including the heart, liver, spleen, lungs, and kidneys, uncovered no observable structural abnormalities or tissue damage (fig. S22). Collectively, these results confirmed the favorable in vivo biosafety profile of SH-EM-NP/EPI. To further assess systemic safety and biodistribution following intravesical instillation, we performed ex vivo fluorescence imaging of major organs at 24 hours post–intravesical instillation. As shown in fig. S23, strong fluorescence signals were detected in the bladder, whereas no obvious nanoparticle accumulation was observed in the heart, liver, spleen, lungs, or kidneys. This bladder-restricted distribution pattern indicates that SH-EM-NP/EPI remain largely confined to the intravesical space with minimal systemic exposure.

Given that the proposed mechanism of SH-EM-NP/EPI involves inflammatory responses and modulation of epithelial barrier, we next systematically evaluated local bladder safety at tissue levels. First, to directly assess whether the inflammatory activation and tight junction modulation induced by SH-EM-NP/EPI would compromise the urothelial integrity, we performed a focused comparison between phosphate-buffered saline (PBS) and SH-EM-NP/EPI. Each group received intravesical instillation treatment every 4 days for four times. As shown in fig. S24, SH-EM-NP/EPI administration resulted in a moderate increase in bladder-associated leukocyte counts, as determined by flow cytometry, as well as elevated TNF-α and IL-6 levels, as measured by enzyme-linked immunosorbent assay (ELISA), on day 1 following intravesical instillation. These changes could be attributable to the activation of local innate immune responses. Notably, these inflammatory indicators gradually declined over time and returned to baseline levels by days 7 and 14, indicating that the inflammatory responses triggered by SH-EM-NP/EPI were transient and self-limiting. To further examine the integrity and recovery of the urothelial barrier, immunohistochemical staining of the tight junction proteins OCC and zonula occludens-1 (ZO-1) was performed on 1, 7, and 14 days postinstillation (fig. S25). In the SH-EM-NP/EPI group, OCC and ZO-1, although showing a transient reduction and partial disorganization on day 1, exhibited well-preserved and continuous staining patterns comparable to that in the PBS group on days 7 and 14. These results indicate that SH-EM-NP/EPI induces a reversible modulation of the urothelial barrier rather than persistent epithelial damage.

Subsequently, to further assess the local safety of SH-EM-NP/EPI, we expanded the analysis to include first-line chemotherapeutic drugs EPI at the same dose as a clinically relevant comparator. Each group received intravesical instillation treatment every 4 days for four times. Bladder weight, bladder wall thickness, urinary GP-51 levels (a marker of urothelial injury), and histopathological changes were examined at 1, 7, and 14 days after treatment (figs. S26 and S27). Free EPI resulted in obvious bladder edema, increased bladder wall thickness, elevated urinary GP-51 levels, and evident epithelial injury at early time points. In contrast, SH-EM-NP/EPI caused markedly milder and more transient changes, which gradually resolved over time and approached levels comparable to that in the PBS group by days 7 and 14. The reduced local toxicity of SH-EM-NP/EPI is likely attributable to the nanoparticle-mediated controlled release of EPI, which limits acute epithelial exposure to high drug concentrations. Therefore, although SH-EM-NP/EPI transiently activates local inflammation and modulates epithelial tight junctions to facilitate therapeutic penetration, it does not induce persistent urothelial damage or long-term bladder toxicity.

Antitumor immune responses induced by SH-EM-NP/EPI

Encouraged by the favorable therapeutic outcomes of intravesically administrated SH-EM-NP/EPI, we next investigated its antitumor immune responses (Fig. 5A). We first investigated the maturation of dendritic cells (DCs), which play a crucial role in connecting the innate and adaptive immune systems (43). As illustrated in Fig. 5B, all treatment groups exhibited increased proportions of mature DCs (CD11c^+^CD86^+^CD80^+^) in inguinal lymph nodes compared with the control group, with SH-EM-NP/EPI eliciting the highest level of DC maturation (Fig. 5B). It is well-known that bacterial membranes could activate the DCs through TLR signaling (29, 30). In Fig. 5B, we observed that EM induced slightly DC maturation. We speculate that this limited immunostimulatory effect was attributable to its poor bladder mucosal adhesion capability. Moreover, in EPI-containing formulations treated groups, such as EPI, EM-NP/EPI, and SH-EM-NP/EPI, elevated expression levels of endogenous danger signals calreticulin (CRT) and high mobility group box 1 (HMGB1) were observed in tumor tissues. This observation demonstrated that EPI induced ICD in tumor cells, which could enhance tumor immunogenicity and further promote DC maturation (Fig. 5, B and C). SH-EM-NP/EPI triggered the most significant ICD and DC maturation through their bladder adhesion and intratumoral penetration capacity. The tumor cell ICD synergizing with the immunoadjuvant effects of bacterial membranes would ultimately contribute to potent immune activation.

It is well-known that CD8^+^ T cells, the key cytotoxic effector cells, play a critical role in attacking tumor cells (44, 45). As illustrated in Fig. 5 (D and H), tumors treated with SH-EM-NP/EPI exhibited the highest CD8^+^ T cell counts. Notably, the percentage of granzyme B (GrzB)–positive T cells (CD3^+^CD8^+^GrzB^+^), key effectors responsible for tumor cell cytotoxicity, was also markedly increased following SH-EM-NP/EPI treatment (Fig. 5, E and I), indicating the activation of cytotoxic T cells. It is well-known that regulatory T cells (Treg cells) and M2-like macrophages that infiltrate the TME would contribute to immunosuppression. Reversing the immunosuppressive TME is crucial for effective tumor elimination (46–48). As presented in Fig. 5 (F and J), SH-EM-NP/EPI treatment resulted in obvious decreases in Treg cell (CD3^+^CD4^+^Foxp3^+^) counts in tumors. Meanwhile, in SH-EM-NP/EPI–treated tumors, we observed a significant down-regulation of M2-like macrophages (CD11b^+^F4/80^+^CD206^+^) and an up-regulation of M1-like macrophages (CD11b^+^F4/80^+^CD80^+^) (Fig. 5, G and K, and fig. S28). Notably, the levels of proinflammatory cytokines, including TNF-α, interferon-γ (IFN-γ), IL-6, and IL-12p70, were markedly elevated in the SH-EM-NP/EPI group (Fig. 5L). Previous studies have reported that bacterial-derived PAMPs can activate the TLR/NF-κB signaling pathway, stimulating DCs and macrophages to secrete proinflammatory cytokines (e.g., IL-6, IL-12, TNF-α, and IFN-γ). Among those cytokines, IL-12, TNF-α, and IFN-γ could induce macrophage polarization from the M2 to the M1 phenotype, while IL-6 and IFN-γ could block Foxp3 expression to inhibit Treg cells (49–51). Based on these observations, we propose that the bacterial membrane components of SH-EM-NP/EPI could contribute to reversing the immunosuppressive TME. Furthermore, immunofluorescence staining of bladder tumor sections revealed that SH-EM-NP/EPI treatment significantly increased the tumor infiltration of CD8^+^ T cells while reducing that of immunosuppressive M2-like macrophages in the TME (Fig. 5M), consistent with the flow cytometry results. These investigations demonstrated that SH-EM-NP/EPI could modulate the immunosuppressive TME and induce robust antitumor immune responses.

Based on these results, the antitumor mechanism of SH-EM-NP/EPI is proposed in Fig. 5N. Upon intravesical instillation, SH-EM-NP/EPI adhere to the bladder mucosa and penetrate deeply into tumor tissues. The sustained delivery of EPI exerts direct cytotoxic effects on tumor cells and simultaneously induces potent ICD, characterized by the release of DAMPs and enhanced tumor immunogenicity. In parallel, PAMPs from the bacterial membranes would serve as immunoadjuvants and synergize with ICD to enhance DC maturation. The dying tumor cells release abundant levels of tumor antigens, which are subsequently processed and presented by DCs, thereby initiating robust T cell–mediated adaptive antitumor immune responses. Additionally, PAMPs from SH-EM-NP/EPI could activate the TLR/NF-κB signaling pathway to up-regulate proinflammatory cytokines such as IL-6, IL-12, TNF-α, and IFN-γ, which could inhibit Treg cells and induce macrophage polarization from the M2 to the M1 phenotype. These factors synergized with EPI-induced ICD could contribute to reverse the immunosuppressive TME. Ultimately, SH-EM-NP/EPI chemo-immunotherapy elicits a robust antitumor immune response, offering a comprehensive and effective approach for bladder cancer treatment.

Intravesical SH-EM-NP/EPI treatment combined with ICB to achieve abscopal therapeutic effects

In the clinic, distant metastasis of bladder tumor often precludes the possibility of radical surgery, leading to a markedly diminished prognosis (52). Inspired by our previous findings that localized chemo-immunotherapy combined with ICB can elicit systemic antitumor immune responses (53), we hypothesized that the intravesical administration of SH-EM-NP/EPI combined with ICB could elicit a potent systemic immune response capable of attacking distant tumors, thereby addressing a critical clinical challenge in bladder cancer. An o.t./subcutaneous dual-tumor model was established in our study (54). First, an o.t. MIBC tumor was implanted into the bladder to serve as the primary tumor for intravesical treatment. Three days later, a second MB49 tumor was subcutaneously implanted into the flank of the same mouse to mimic a distant tumor that without direct drug administration (Fig. 6A). Notably, flow cytometry revealed significant up-regulation of PD-L1 in both o.t. bladder tumors and distal subcutaneous tumors following intravesical treatment with SH-EM-NP/EPI, suggesting the potential benefit of combining SH-EM-NP/EPI administration with αPD-L1 ICB therapy (Fig. 6B).

Inspired by these results, we next investigated the therapeutic efficacy and systemic antitumor immune responses of intravesical SH-EM-NP/EPI therapy in combination with intravenous αPD-L1 therapy using the o.t./subcutaneous dual-tumor bladder cancer model. Seven days after establishing the o.t. tumor model (4 days after implantation of subcutaneous tumors), the mice were randomly divided into six control (saline), αPD-L1 (1 mg/kg), BCG + αPD-L1 (BCG, 1 mg/ml; and αPD-L1, 1 mg/kg), EPI + αPD-L1 (EPI, 1 mg/ml; and αPD-L1, 1 mg/kg), SH-EM-NP/EPI (EPI, 1 mg/ml), and SH-EM-NP/EPI + αPD-L1 (EPI, 1 mg/ml; and αPD-L1, 1 mg/kg). The treatment schedule is shown in Fig. 6C. For o.t. bladder tumors, tumor sizes were monitored by serial bladder ultrasound, and bladder tumor tissues were resected and weighed at the end of the treatment course. As presented in Fig. 6 (D and E), BCG + αPD-L1 and EPI + αPD-L1 showed improved TGI than αPD-L1 monotherapy, but they were less effective than intravesical SH-EM-NP/EPI treatment. Notably, complete regression of o.t. bladder tumors was achieved in the SH-EM-NP/EPI + αPD-L1 group, highlighting the potent therapeutic benefit of αPD-L1 ICB in enhancing the efficacy of intravesical SH-EM-NP/EPI treatment.

For distal subcutaneous tumors, tumor sizes were continuously monitored, and tumor tissues were resected upon completion of the treatment regimen for further analysis. As presented in Fig. 6 (F to I), all treatment groups exhibited systemic therapeutic effects. Compared with the effects of αPD-L1 monotherapy (TGI of 43.6%), BCG + αPD-L1 (TGI of 45.9%) and EPI + αPD-L1 (TGI of 53.2%) did not achieve significantly better systemic therapeutic efficacy. Conversely, SH-EM-NP/EPI (TGI of 68.2%) induced stronger abscopal effects than BCG + αPD-L1 and EPI + αPD-L1. Notably, SH-EM-NP/EPI + αPD-L1 displayed the most potent systemic antitumor effect (TGI of 93.4%). Moreover, no significant changes in body weight were observed in mice following SH-EM-NP/EPI treatment. These results suggest that intravesical SH-EM-NP/EPI, through localized chemo-immunotherapy, can elicit a systemic antitumor effect, which can be further enhanced in combination with αPD-L1 therapy.

We further investigated the underlying mechanisms of the observed systemic antitumor effect achieved by intravesical SH-EM-NP/EPI administration. We first assessed DC maturation in the inguinal lymph nodes. Intravesical SH-EM-NP/EPI treatment significantly promoted DC maturation, which was further enhanced in combination with αPD-L1 (Fig. 6J). We next assessed the presence and functional status of CD8^+^ T cells in distant tumors. A significant increase in CD8^+^ T cell counts was observed following intravesical SH-EM-NP/EPI treatment (Fig. 6K). Their functional activity was also enhanced, as indicated by a marked increase in GrzB^+^ CD8^+^ T cell counts (Fig. 6L). Notably, this effect was further amplified when SH-EM-NP/EPI were combined with αPD-L1 therapy (Fig. 6, K and L). These investigations indicate that αPD-L1 effectively suppresses tumor-expressed PD-L1 and prevents immune evasion by the tumor, boosting the cytotoxic activity of tumor-specific T cells. Our findings thus demonstrate that local intravesical SH-EM-NP/EPI chemo-immunotherapy can induce ICD in o.t. bladder tumors, leading to the continuous release of tumor antigens, which together with the intrinsic adjuvant properties of the bacterial membrane would promote DC maturation and antigen presentation, thereby facilitating the priming and expansion of cytotoxic CD8^+^ T cells. These effector T cells would suppress tumor growth at the primary site and attack distant tumor lesions, resulting in a potent abscopal response. ICB with αPD-L1 to inhibit tumor immune evasion could further enhance systemic antitumor immunity.

DISCUSSION

In this study, we developed thiol-functionalized bacterial membrane–coated nanoparticles loaded with EPI for highly effective intravesical chemo-immunotherapy against bladder tumors. This nanomedicine platform has unique capabilities in enhancing bladder retention and improving intratumoral drug penetration. SH-EM-NP/EPI continuously induce ICD in tumor cells and, in synergy with the intrinsic immunoadjuvant properties of the E. coli membrane, amplify both innate and adaptive immune responses. SH-EM-NP/EPI exhibited markedly superior therapeutic efficacy compared with those clinically used EPI or BCG in both NMIBC and MIBC mouse models. Furthermore, when combined with PD-L1 blockade, SH-EM-NP/EPI achieved complete regression of o.t. MIBC bladder tumors and amplified systemic antitumor immune responses, resulting in robust shrinkage of distant tumors. This approach overcomes the limitations of conventional intravesical therapy and provides an innovative therapeutic strategy for bladder preservation in patients with advanced bladder cancer.

The potent antitumor effects triggered by SH-EM-NP/EPI could be attributable to the synergistic interplay of multiple mechanisms. First, the surface-functionalized thiol groups on SH-EM-NP/EPI can undergo thiol-disulfide exchange reactions with disulfide bonds abundantly present in the bladder mucosal layer, resulting in covalent adhesion to the bladder wall. Building upon the robust bladder adhesion, the E. coli membrane component of the SH-EM-NP/EPI activates key signaling pathways, including TLRs, cytokine signaling, and NF-κB pathways, which up-regulate MMPs, thereby facilitating the degradation of tight junctions and collagen within the tumor stroma and markedly enhancing the trans-barrier intratumoral delivery of the intravesically administered therapeutic agent. In addition, SH-EM-NP/EPI, which penetrate into the tumor parenchyma, enable the sustained release of EPI, which continuously induces ICD in tumor cells and promotes the release of tumor antigens. Subsequently, robust tumor-specific adaptive immunity was elicited owing to the intrinsic immunoadjuvant properties of the E. coli membrane, thereby effectively suppressing both local and distant tumors. Meanwhile, the capability of SH-EM-NP/EPI to modulate the immunosuppressive TME would further favor antitumor immune responses. Thus, SH-EM-NP/EPI markedly enhanced the antitumor efficacy of intravesical therapy through a multistep cascade mechanism involving mucosal adhesion, tumor permeation, chemotherapeutic ICD action, and immune activation.

Our mucoadhesive tumor-penetrating nanomedicine holds great advantages for potential clinical translation in bladder cancer treatment. Different from live bacterial immunotherapies such as BCG, bacterial membrane–based SH-EM-NP/EPI nanomedicine may mitigate the risk of pathogen dissemination and the associated complications (28, 30). Compared with conventional chemotherapy, EPI in the SH-EM-NP/EPI formulation exhibits superior tumor penetration and retention. In clinical practice, postoperative residual bladder cancer lesions are typically smaller than 2 mm, the penetration depth achieved by our nanomedicine (>2 mm) is thus sufficient for treating human bladder cancer and thus holds strong clinical potential. Moreover, with the ability of simultaneous ICD-type tumor cell killing and immune activation, such SH-EM-NP/EPI may be a much more potent intravesical therapy compared with current first-line therapies such as BCG and chemotherapy for treatment of NMIBC. Furthermore, the combination of intravesical SH-EM-NP/EPI chemo-immunotherapy combined with systemic ICB therapy could trigger potent systemic tumor-specific immunity to effectively suppress both the primary bladder tumor and distant metastatic lesions, offering a safe and effective therapeutic strategy for patients with MIBC and advanced bladder cancers. Therefore, our mucoadhesive tumor-penetrating nanomedicine offers an innovative solution to multiple clinical challenges in bladder cancer treatment and represents a promising strategy for future clinical translation.

MATERIALS AND METHODS

Materials

Poly(lactide-co-glycolic acid) (acid terminated; Mw, 7000 to 17000; CAS: 26780-50-7), polyvinyl alcohol (PVA; Mw, 9000 to 10,000; 80% hydrolyzed; CAS: 9002-89-5), and mucin from porcine stomach (type III) were purchased from Sigma-Aldrich (Saint Louis, USA). EPI hydrochloride (CAS: 56390-09-1), sucrose (CAS: 57-50-1), and lysozyme (CAS: 12650-88-3) were purchased from Shanghai Macklin Biochemical Technology Co. Ltd. (Shanghai, China). DiD was purchased from Invitrogen. Cy7 maleimide (CAS: 2120392-49-4) and Cy7 NHS Ester (CAS: 1432019-64-1) were purchased from MedChemExpress LLC (New Jersey, USA). DSPE-PEG2000-SH (2 kDa) and DSPE-PEG-NH2 (2 kDa) were purchased from Ruixi biological Technology (Xi’an, China). Agarose (CAS: 9012-36-6), protamine sulfates (CAS: 9009-65-8), L-ascorbic acid (CAS: 50-81-7), dl-DTT (CAS: 3483-12-3), and 4′,6-diamidino-2-phenylindole (DAPI) (CAS: 28718-90-3) were purchased from Shanghai Aladdin Biochemical Technology Co. Ltd. (Shanghai, China). PBS, Dulbecco’s modified Eagle’s medium (DMEM), RPMI 1640 medium, and fetal bovine serum (FBS) were purchased from Gibco (USA). The bicinchoninic acid (BCA) protein assay kit was purchased from Thermo Fisher Scientific (Waltham, USA). d-Luciferin potassium was purchased from PerkinElmer Inc. (USA). ELISA kits were obtained from eBioscience (San Diego, USA). CD16/CD32 (catalog no. 101302), fluorescein isothiocyanate (FITC)–CD11c (catalog no. 117305), allophycocyanin (APC)-CD11c (catalog no. 117310), APC-CD80 (catalog no. 104714), phycoerythrin (PE)-CY5-CD80 (catalog no. 104711), PE-CD86 (catalog no. 105007), FITC-CD45 (catalog no. 103107), PE-CD45 (catalog no. 103105), peridinin–chlorophyll-protein complex (PerCP)-CY5.5-CD45 (catalog no. 147706), APC–PD-L1 (catalog no. 124311), PE-CD11b (catalog no. 101208), PerCP-F4/80 (catalog no. 123126), APC-CD3 (catalog no. 100236), FITC-CD3 (catalog no. 100203), PE-CD4 (catalog no. 100408), PE-CY5-CD4 (catalog no. 100514), PerCP-CD4 (catalog no. 100431), PE-Foxp3 (catalog no. 320008), PE-CD8 (catalog no. 100708), APC-CD206 (catalog no. 141707), PE-CY5-GrzB (catalog no. 372226), APC-GrzB(catalog no. 396407), and PD-L1 (catalog no. 124302) were purchased from BioLegend (San Diego, USA). MB49 cell lines, SV-HUC-1 cell lines, and MB49-Luc cell lines were purchased from Shanghai Fuheng Biotechnology Co. Ltd. (Shanghai, China). E. coli DH5α cells were purchased from Thermo Fisher Scientific (Waltham, USA). Transwell insert were purchased from Corning Company (USA). Female C57BL/6 mice (6 to 8 weeks) were purchased from GemPharmatech Co. Ltd. (Nanjing, China). Other chemicals were purchased from Sigma-Aldrich (Saint Louis, USA) unless otherwise noted.

Preparation of PLGA-NP/EPI nanoparticles

A double emulsion-solvent evaporation method was used to prepare PLGA-NP/EPI nanoparticles. A solution was prepared by dissolving 100 mg of PLGA-COOH in 2 ml of dichloromethane, followed by the addition of 0.2 ml of an EPI solution (50 mg/ml, with 10 mg of EPI in 200 μl of double-distilled water). The mixture underwent ultrasonication at 100 W in an ice water bath for 5 min, using a cycle of 3 s on and 5 s off. The above mixture was then added to 5 ml of a 5% PVA solution, and emulsification was repeated by ultrasonication (100 W) in an ice water bath for 10 min. The emulsion was gradually introduced into 20 ml of double-distilled water and continuously stirred at room temperature overnight. PLGA-NP/EPI nanoparticles were centrifuged at 21,000g for 30 min, washed thrice with PBS, and stored at 4°C until needed.

Drug loading and encapsulation efficiency were assessed using ultraviolet-visible spectroscopy at 485 nm, with calculations based on linear standard curves across the tested concentration range. The EPI solutions were diluted with water to achieve the desired concentration range for working solutions. The EPI recovered from washed supernatants during preparation (WR) was quantified.

The drug payload (WNP) is determined by subtracting WR from the total EPI amount (WT). The drug encapsulation and loading efficiencies were determined using the following equationsEncapsulation efficiency=WNP/WT×100%Loading efficiency=WNP/Wt×100%

WT is the total EPI weight used for NP preparation, WNP is the total EPI amount within the NPs, and Wt is the total NP weight.

For the synthesis of EPI-unloaded nanoparticles (PLGA-NP), 100 mg of PLGA-COOH was completely dissolved in 2 ml of dichloromethane. The above mixture was then added to 4 ml of a 5% PVA solution and emulsification was repeated by ultrasonication (100 W) in an ice water bath for 10 min. The emulsion was gradually added to 8 ml of double-distilled water and stirred at room temperature overnight. The PLGA-NP nanoparticles were obtained by centrifuging at 21,000g for 30 min, washed three times with PBS, and kept at 4°C until needed.

Isolation of EM

Cytoplasmic membranes from E. coli strain DH5α were isolated following a previously described method with modifications (30). E. coli DH5α cells were cultured overnight in liquid LB medium at 37°C with 200 rpm shaking. Bacteria were harvested at an optical density at 600 nm of 1.2 through centrifugation at 8000 rpm for 10 min at 4°C, followed by three washes with PBS. The cell pellet was resuspended in 10 ml of buffer A [1 M sucrose and 0.2 M tris-HCl (pH 8.0)] with lysozyme added to reach a concentration of 2 mg/ml. The cells were incubated for 2 hours at 37°C with shaking at 200 rpm. Spheroplasts were harvested by centrifuging at 8000 rpm for 10 min at 4°C, then resuspended in 10 ml of cold buffer B [20 mM tris-HCl (pH 7.2), 50 mM NaCl, and 5 mM EDTA] with 20% w/v sucrose for cell lysis via ultrasonic processing. The lysate was centrifuged at 10,000g for 20 min at 4°C to eliminate cell debris. After centrifuging the supernatant at 100,000g for 2 hours, the resulting pellet was resuspended in 2 ml of buffer B. This suspension was then layered on a discontinuous sucrose gradient (from bottom to 5 ml of 50%, 3 ml of 46%, 5 ml of 42%, 5 ml of 36%, 3 ml of 32%, and 5 ml of 27%). Ultracentrifugation was performed at 100,000g for 2 hours at 4°C using a Beckman Optima XPN-100 ultracentrifuge with a 32Ti rotor. A 5-ml sample of the cytoplasmic membrane fraction was extracted and diluted with cold buffer B to achieve a sucrose concentration of 10% (w/v). Membranes were collected after centrifuging at 100,000g for 1 hours at 4°C, then resuspended in PBS, and stored at −80°C for future experiments. The protein concentration of the final membrane preparation was quantified using a BCA protein assay kit.

Isolation of RBC plasma membranes

To isolate RBC membrane, whole blood was freshly collected from the orbital sinus of female C57BL/6. The blood underwent centrifugation at 3000g for 5 min and was washed twice with PBS to eliminate the serum. Washed RBCs were suspended in hypotonic (0.2×) PBS on ice to induce membrane rupture. The cell homogenate underwent centrifugation at 10,000g for 10 min at 4°C to eliminate hemoglobin. The two steps were repeated until the hemoglobin was adequately removed. The RBC membrane was preserved in PBS at −80°C for future use.

Generation and physicochemical characterization of SH-EM-NP/EPI

First, EM was shaken in a dry bath incubator at 37°C for 15 min before nanoparticle membrane coating. Then, PLGA-NP/EPI and EM were repeatedly coextruded using a 400-nm cutoff extruder with a polymer-to-membrane protein mass ratio of 1. SH-EM-NP/EPI was prepared by adding DSPE-PEG2000-SH to the suspension at a mass ratio of 1 (membrane protein to DSPE-PEG2000-SH) and stirring under nitrogen protection for 30 min at 200 rpm and 37°C. Subsequently, samples were centrifuged at 14,800 rpm for 30 min. Following supernatant removal, SH-EM-NP/EPI were collected and preserved in PBS for subsequent analysis.

To examine the morphology of the nanoparticles, 10 μl samples were deposited onto carbon-coated copper grids and incubated for 15 min. The residual fluid was absorbed using filter paper, and the samples on the grids were negatively stained with 1% (v/v) uranyl acetate for 8 min and air dried for analysis with a transmission electron microscope (HT7700, HITACHI, Tokyo, Japan). The hydrodynamic size distribution, ζ potential and polydispersity index were measured using a Zetasizer Nano ZS dynamic light scattering instrument (Malvern, UK).

Regarding the synthesis of EPI-unloaded EM-NP and SH-EM-NP, all the steps were identical except that PLGA-NP was used instead of PLGA-NP/EPI. Regarding the synthesis of NH2-EM-NP, all the steps were identical except that DSPE-PEG2000-NH2 was used instead of DSPE-PEG2000-SH. For the preparation of RBC plasma membranes coated nanoparticles (CM-NP and SH-CM-NP), all the steps were identical except that RBC plasma membranes were used instead of cytoplasmic membranes from E. coli. For the preparation of Cy5.5-labeled nanoparticles, Cy5.5-labeled PLGA at the same weight ratios were used to form different NPs.

Protein characterization of SH-EM-NP/EPI

Samples were initially centrifuged at 14,800 rpm for 30 min at 4°C. The precipitate was then lysed in radioimmunoprecipitation assay buffer on ice for 30 min and centrifuged again under the same conditions. To analyze the protein composition on SH-EM-NP/EPI, label-free quantitative proteomics analysis was performed (GENEWIZ Biotechnology Co. Ltd., Suzhou, China).

Reaction between SH-EM-NP and mucin in vitro

In vitro experiments confirmed the catalyst-free reaction between SH-EM-NP and mucin through dynamic thiol-disulfide exchange. First, mucin was thoroughly dissolved in an amine-free buffer to prepare a mucin solution (2 mg/ml). Cy7-NHS was added to mucin at a 1 molar ratio and gently agitated in light-protected conditions at room temperature for 2 hours. Postreaction, the solution underwent centrifugation at 21,000g for 1 hour to eliminate large mucin aggregates, and the supernatant was collected. The supernatant underwent ultrafiltration with a 100-kDa molecular weight cutoff filter to purify Cy7_mucin.

Nanoparticles of both modified and unmodified DSPE-PEG2000-SH were resuspended in PBS to a concentration of 0.3 mg/ml. Cy7_mucin, dissolved in PBS, was added to achieve a final concentration of 0.2 mg/ml. The mixture was then allowed to react at room temperature for 2 hours with shaking at 750 rpm. After the specified duration, the mixture was centrifuged at 21,000g for 30 min at 4°C and then washed three times with PBS to thoroughly eliminate any unattached Cy7_mucin. The precipitate was collected after centrifugation, resuspended in PBS, and imaged with the In Vivo Imaging System (IVIS). To confirm thiol involvement in conjugation, thiol-sensitive Cy7 maleimide was used and analyzed using the IVIS to observe changes in thiol numbers following reaction with mucin. To confirm disulfide bond formation between thiolated nanoparticles and mucin, l-ascorbic acid (1 mg/ml) and DTT (0.1 mg/ml) were introduced into the mixture before the reaction with Cy7_mucin as these reagents can inhibit thiol-disulfide exchange. The conjugation of Cy7_mucin on thiolated nanoparticles was observed and quantified by IVIS imaging system.

Animals and cells

Female C57BL/6 mice, aged 6 to 8 weeks, were acquired from GemPharmatech Co. Ltd. in Jiangsu, China. Mice were kept in groups of five per individually ventilated cage under a 12-hour light-dark cycle (08:00 to 00 light; 00 to 00 dark), with a stable room temperature of 21° ± 1°C and relative humidity between 40 and 70%. All mice had unrestricted access to food and water. The animal experiments were conducted in accordance with ethical guidelines and received approval from the Institutional Animal Care and Use Committee at Soochow University [approval number SYXK(Su)2021-0073].

MB49, MB49-Luc, SV-HUC-1 cell lines were purchased from Shanghai Fuheng Biotechnology Co, Ltd. (Shanghai, China). MB49 and MB49-Luc cells were maintained in DMEM supplemented with 10% FBS and 1% penicillin–streptomycin (PS) at 37°C in a 5% CO2 atmosphere. SV-HUC-1 cells were maintained at 37°C in 5% CO2 using Ham’s F-12 K medium (Gibco) supplemented with 15% FBS and 1% PS.

Human samples

Bladder tissue samples were collected from patients undergoing total cystectomy at Harbin Medical University Cancer Hospital. Studies were done under approval from the Ethical Committee of Harbin Medical University Cancer Hospital (KY-2024-24), and informed consent was obtained before enrollment for this study.

Bladder mucosa retention and adhesion assay

For the prolonged retention effect experiment, mice were intravesically instilled with DiD-labeled EM-NP, NH2-EM-NP, and SH-EM-NP (10 mg/ml, 50 μl) for 2 hours, followed by three times saline rinses. The mice were then given 24 hours of unrestricted access to food and water before undergoing bladder dissection. The adhesion effect was assessed using the IVIS imaging system and confocal microscopy. Bladder tissue samples were embedded in Tissue-Tek optimum cutting temperature compound and cryosectioned into 10-μm slices using a Leica Biosystems vibrating microtome. Cryosections were stained with ready-to-use DAPI for 10 min in the dark to visualize cell nuclei. Sections were washed thrice with 1× PBS containing Tween 20 (PBST), mounted using an antifluorescence quencher, and imaged with a Zeiss 710 confocal microscope.

To examine the adhesion effect of EM-NP, SH-EM-NP, and DTT-treated SH-EM-NP on the surface of bladder mucosa, we performed experiments using bladder tissues from mice and humans. Patient bladder tissues obtained from total cystectomy surgery were placed into saline. Bladder tissues were exposed to DiD-labeled formulations (10 mg/ml) for 2 hours, followed by triple washing with artificial urine. The samples were then prepared as previously outlined and imaged via confocal microscopy. In mice, DiD-labeled formulations (10 mg/ml, 50 μl) were instilled intravesically into the bladder for 2 hours. Following three washes with artificial urine to eliminate residual formulations, the samples were prepared as previously described and imaged via confocal microscopy.

Multicellular spheroids

Multicellular spheroids were generated from cultured MB49 cell lines. Standard 96-well plates with flat bottoms were precoated with 50 μl of a sterile 2.0% (w/v) agarose solution in DMEM to create a nonadherent surface. Subsequently, 1000 cells in 200 μl of culture medium were introduced into each agarose-coated well. After 7 days, cells formed spherical clusters averaging over 200 μm in diameter. Cy5.5-labeled nanoparticles (1 mg/ml) were incubated with multicellular spheroids for 2 hours, after which the medium was replaced and left for 24 hours. The spheroids were then washed three times with PBS and transferred to a confocal microscopy dish. Confocal microscopy with 25-μm Z-intervals from apex to equator was used to examine nanoparticle penetration into the spheroids. Zen software was used for quantitative analysis of nanoparticle penetration into multicellular spheroids.

TEER measurement assay

This study used Transwell plates to assess TEER. SV-HUC-1 human bladder epithelial cells and MB49 mouse bladder cancer cells were seeded at 1 × 10^4^ cells per well in 24-well Transwell plate insert chambers on day 0. Daily TEER measurements were conducted using a RE1600 epithelial volt-ohm meter. Once the TEER values reached a plateau, indicating the formation of fully established tight junctions, different formulations (with a membrane protein concentration of 20 μg/ml) diluted in culture medium were added to the inserts and coincubated for 8 hours. After removing the formulations, the medium was replaced with PBS, and TEER values were measured. The cells were incubated with fresh culture medium for 16 hours, after which the medium was replaced with PBS and TEER values were measured again.

Immunofluorescence of tight junction–associated proteins

SV-HUC-1 cells were exposed to various formulations, each containing a membrane protein concentration of 20 μg/ml, for 8 hours. Posttreatment, cells underwent washing with ice-cold PBS and fixation in 4% paraformaldehyde in PBS for 15 min. This was followed by three PBS washes and a 1 hour room temperature blocking step using 10% goat serum. Cells were incubated overnight at 4°C with primary antibodies (anti–E-cadherin and anti-OCC), washed with PBS, and then treated with goat anti-rabbit Alexa Fluor 488 at a 2000 dilution for 1 hour at room temperature. Nuclei were stained with DAPI. After washing three times with PBS, the samples were imaged using a confocal microscope.

Development of a mouse model for NMIBC

Cells were removed from tissue culture plastic by treating with 0.05% trypsin-EDTA for 5 min at 37°C. Cells were washed three times with PBS. Then, cells were resuspended in DMEM at 2 × 10^8^ cells/ml for MB49-Luc. Intravesical instillation in 8-week-old female mice was performed using 24.5-gauge plastic catheters. Before tumor cell infusion, 100 μl of protamine sulfate (10 mg/ml) was instilled into the bladder of anesthetized mice and maintained for 30 min. The bladder was emptied by applying digital pressure. A 100 μl tumor cell mixture was introduced into the bladder and left for 2 hours, with mice under inhalational isoflurane (1.5% for induction and 0.5% for maintenance) and on heating pads to sustain core body temperature. Postimplantation, mice were routinely observed for hematuria and tumor development.

Development of a mouse model for MIBC

Cells were removed from tissue culture plastic by treating with 0.05% trypsin-EDTA for 5 min at 37°C. Cells were washed three times with PBS and resuspended in PBS at a concentration of 1 × 10^7^ cells/ml for both MB49 and MB49-Luc. A bladder dissection was performed on 8-week-old female C57BL/6 mice, and 10 μl of a tumor cell mixture was injected into the bladder wall to create an o.t. bladder tumor. Subsequently, the abdominal cavity was sutured, and an intraperitoneal injection of penicillin-streptomycin was administered. Postimplantation, mice were routinely observed for hematuria and tumor development.

Establishment of o.t. and subcutaneous dual-tumor bladder cancer model

To establish o.t. and subcutaneous dual-tumor bladder cancer models, cells were detached from tissue culture plastic using 0.05% trypsin-EDTA for 5 min at 37°C. Following detachment, cells were washed three times with PBS, and viability was evaluated using the Countess II automated cell counter (Thermo Fisher Scientific). Subsequently, o.t. MB49 bladder tumor-bearing mice were first established following the protocol for the MIBC model. After three days, 1 × 10^6^ MB49 cells were injected subcutaneously into the mice’s right flanks. The animals were then regularly monitored for signs of hematuria and tumor growth.

Mouse in situ bladder tumor model intratumoral penetration experiments

Three days after establishing the o.t. MIBC bladder tumor model in female mice, Cy5.5-labeled PLGA-NP, EM-NP, CM-NP, SH-CM-NP, and SH-EM-NP (10 mg/ml, 50 μl) were intravesically instilled into the bladder twice, with a 4-day interval between treatments. Twenty-four hours after the second instillation, the tumor-bearing bladders were harvested and analyzed by confocal imaging. Tissue processing was performed as previously described.

In vivo MMP inhibition and intravesical nanoparticle penetration experiments

To evaluate the role of MMPs in mediating the tumor penetration of SH-EM-NP, an MMP inhibition study was performed using GM6001 (Ilomastat). An o.t. MIBC model was established in female mice as described above. Three days after tumor implantation, mice were randomly assigned to receive daily intraperitoneal injections of GM6001 at a dose of 50 mg/kg per day or vehicle control. GM6001 was dissolved in dimethyl sulfoxide (DMSO) and diluted with PBS to a final DMSO concentration below 2% before administration. GM6001 treatment was initiated on the day of the first intravesical instillation and continued daily until tissue collection. Cy5.5-labeled PLGA-NP, EM-NP, CM-NP, SH-CM-NP, or SH-EM-NP (10 mg/ml, 50 μl) were intravesically instilled into the bladder twice, with a 4-day interval between treatments. Twenty-four hours after the second instillation, tumor-bearing bladders were harvested for penetration analysis. Bladder tissues were processed as previously described, embedded, sectioned, and subjected to confocal imaging to assess the intratumoral distribution and penetration depth of the nanoparticles.

In vivo antitumor effect

In the NMIBC model, 5 days post–tumor establishment, mice were randomly assigned to six groups (n = 5) and received intravesical instillations of 50 μl of either PBS, EM, EPI, BCG, EM-NP/EPI, or SH-EM-NP/EPI every 4 days, totaling four treatments. The EPI and BCG concentrations were both 1 mg/ml, and the EM-containing groups received an equivalent membrane protein dose. Bioluminescence from MB49-Luc cells was assessed every 4 days using a PerkinElmer in vivo imaging system (IVIS). Body weight was recorded every other day for 14 days.

In the MIBC model, 3 days post–tumor establishment, mice were randomly assigned to five groups (n = 5) and received intravesical instillations of 50 μl of either PBS, EM, EPI, EM-NP/EPI, or SH-EM-NP/EPI every 4 days, totaling four treatments. The EPI concentration was 1 mg/ml, and the EM-containing groups received an equivalent membrane protein dose. Bioluminescence signals from MB49-Luc cells were measured on days 0, 4, 8, 12, and 14 using IVIS (PerkinElmer). Tumors were collected and weighed at the experiment’s conclusion. Body weight was recorded every other day for 14 days.

In the o.t. and subcutaneous dual-tumor bladder cancer model, mice were assigned to six groups 4 days post–tumor (i) control (saline, 50 μl intravesically, n = 5), (ii) αPD-L1 (1 mg/kg intravenously, n = 5), (iii) BCG + αPD-L1 (BCG at 1 mg/ml, 50 μl intravesically and αPD-L1 at 1 mg/kg intravenously, n = 5), (iv) EPI + αPD-L1 (EPI at 1 mg/ml, 50 μl intravesically and αPD-L1 at 1 mg/kg intravenously, n = 5), (v) SH-EM-NP/EPI (EPI at 1 mg/ml, 50 μl intravesically, n = 5), and (vi) SH-EM-NP/EPI + αPD-L1 (EPI at 1 mg/ml, 50 μl intravesically and αPD-L1 at 1 mg/kg intravenously, n = 5). Bladder tumor size was monitored by ultrasound imaging, and the bladder tumor tissues were excised and weighed at the end of the treatment. For the distal subcutaneous tumors, tumor size was measured every other day for a total of 14 days. At the end of the treatment, the tumors were harvested for further analysis. Body weight was also recorded every other day for 14 days.

In vivo ICD

Tumor tissues were collected posttreatment to assess in vivo CRT exposure and HMGB1 release. These tissues were fixed in 4% formalin, dehydrated, cleared, embedded in paraffin, and sectioned into 10-μm-thick serial slices. Following deparaffinization and antigen retrieval, sections were blocked using a 5% bovine serum albumin solution for 30 min at 37°C and then incubated overnight at 4°C with anti-CRT and anti-HMGB1 primary antibodies at a 300 dilution. The following day, sections were washed three times with 1× PBST and then incubated in the dark with Alexa Fluor 488– and Alexa Fluor 647–conjugated secondary antibodies (diluted 500) at 37°C for 30 min. To remove unbound antibodies, sections were washed three times with 1× PBST for 5 min each. Nuclei were counterstained with preprepared DAPI for 10 min in darkness. After another three washes with 1× PBST, the sections were mounted using an antifluorescence quencher and imaged using a confocal microscope (Zeiss).

Cytokine detection in tumor tissues

Tumor tissues were collected posttreatment and homogenized in cold PBS buffer with digestive enzymes to create single-cell suspensions for cytokine analysis. The postcentrifugation supernatant was collected for further analysis. Intratumor levels of IL-6, TNF-α, IFN-γ, and IL-12p70 cytokines were quantified using ELISA kits following the manufacturer’s guidelines. ELISA kits were sourced from Invitrogen.

In vivo flow cytometry

To analyze immune cells in tumors and lymph nodes, posttreatment tumor tissues and lymph nodes were collected and homogenized in cold PBS with digestive enzymes to create single-cell suspensions. Cells were initially incubated with anti-CD16/32, followed by staining with fluorophore-labeled antibodies targeting CD11c, CD80, CD86, CD45, PD-L1, CD11b, F4/80, CD3, CD4, CD8, CD206, GrzB, and Foxp3 for 30 min at room temperature, before flow cytometry analysis using BD Accurit C6 Plus and BD FACSAria III. For intracellular markers like GrzB, Foxp3, and CD206, cell membranes were fixed and permeabilized using Fixation Diluent and Permeabilization Diluent before intracellular staining.

H&E staining experiment

Sections were fixed in hematoxylin (Sigma-Aldrich, St. Louis, MO, USA) for 5 min at room temperature, followed by a 2 min rinse in running water for H&E staining. The sections were briefly immersed in acid alcohol, followed by sodium bicarbonate, and then dehydrated for 30 s. They were stained with eosin for 2 min and thoroughly washed using dehydrant and xylene.

Statistical analysis

Data are shown as means ± SD or range, as detailed in the figure legends. For comparisons involving more than two groups, one-way or two-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test was used, while Student’s t test was used for two-group comparisons. Survival curves were evaluated using the log-rank (Mantel-Cox) test. Statistical significance levels were defined as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, with “n.s.” indicating no significant difference. GraphPad Prism software version 8.0 was used to conduct the statistical analysis. No effect size was predetermined. All experiments were performed in three replicates or more. No animals and/or data were excluded. The dosing groups were filled by randomly selecting from the same pool of animals for in vivo experiments. All experimental procedures and quantification of results—including intravesical instillation, injection, isolation of the tumors or organs, tissue histological analysis, and flow cytometry—were done by two independent researchers.