Decellularized extracellular matrix hydrogel-mediated EVs therapy alleviates diabetic erectile dysfunction by targeting the miR-203a-3p/TMEM33 Axis

Abstract

Diabetic erectile dysfunction (DMED) is a common complication among male patients with diabetes. Therapies primarily based on phosphodiesterase type 5 inhibitors (PDE5Is) often yield suboptimal results. Extracellular vesicles (EVs) have emerged as a promising therapeutic strategy; however, their clinical translation is hindered by rapid in vivo clearance and limited inherent bioactivity. Therefore, we developed an innovative combined therapeutic approach. First, we fabricated an injectable, thermosensitive hydrogel (ECMTA-Hydrogel) by combining a decellularized porcine corpus cavernosum extracellular matrix (ECM) with tannic acid, which exhibits excellent biocompatibility and rapid gelation at body temperature (37 °C), enabling prolonged local retention of EVs within the corpus cavernosum. Second, we pretreated mesenchymal stem cells (MSCs) with pioglitazone to generate engineered EVs (PGZ-EVs) with enhanced bioactivity. Experiments confirmed that the ECMTA-Hydrogel facilitates sustained release of PGZ-EVs, extending their duration of action. In a DMED rat model, the combined ECMTA-PGZ-EVs therapy demonstrated superior efficacy, markedly improving erectile function and effectively reversing pathological phenotypic switching and apoptosis in corpus cavernosum smooth muscle cells (CCSMCs). Mechanistically, we identified that miR-203a-3p, highly enriched in PGZ-EVs, directly targets and suppresses the endoplasmic reticulum transmembrane protein TMEM33, thereby mitigating diabetes-induced endoplasmic reticulum stress (ERS). This pathway represents the key mechanism underlying the cytoprotective effects of PGZ-EVs. In summary, this study not only establishes an efficient natural matrix hydrogel-based delivery system but also augments the therapeutic potential of EVs via engineered modification. Furthermore, we elucidate a novel mechanism involving the miR-203a-3p/TMEM33/ERS axis, offering a promising therapeutic strategy for the clinical management of DMED.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04414-1.

Introduction

Erectile dysfunction (ED) is a frequent complication among male patients with diabetes mellitus (DM). Although oral phosphodiesterase type 5 inhibitors (PDE5Is) represent the first-line treatment for ED, their efficacy in diabetic patients remains often unsatisfactory. Consequently, there is an urgent need for novel therapeutic interventions for DMED [1, 2].

Recent evidence indicates that bone marrow-derived mesenchymal stem cell (MSC) transplantation can ameliorate DMED [3, 4]. However, clinical application is limited by potential risks such as tumorigenicity and pulmonary embolism [5]. Extracellular vesicles (EVs) secreted by MSCs present a promising alternative, circumventing these risks [6, 7]. Nonetheless, the burst release of EVs substantially undermines their therapeutic efficacy [8]. To address this limitation, we incorporated EVs into hydrogels to achieve controlled release. Hydrogels, recognized for their biocompatibility, tunable mechanical properties, and sustained-release capabilities, hold great promise in drug delivery and tissue engineering [9–11]. However, enhancing both the functionality and biosafety of hydrogels to meet specific therapeutic demands remains challenging. Here, we synthesized a hydrogel using a decellularized extracellular matrix (ECM) derived from the corpus cavernosum. This ECM hydrogel not only facilitates sustained EVs release but also exhibits mechanical properties congruent with native penile tissue [12, 13]. Moreover, the ECM’s rich composition of fibrous components can bind regenerative factors and cytokines, modulating cell division, migration, and recruitment during tissue regeneration, thereby promoting repair [14, 15]. We further incorporated tannic acid (TA) into the ECM hydrogel. Capitalizing on TA’s potent reactive oxygen species scavenging capacity and anti-inflammatory properties, we engineered an optimal microenvironment for EVs to exert tissue-reparative effects [16, 17]. This innovative design not only overcomes the bottleneck of local EV delivery but also provides a novel therapeutic strategy and theoretical foundation for DMED treatment.

MSCs-derived EVs carry diverse paracrine factors, including proteins and various RNA species, which enter target cells to exert significant biological effects [18, 19]. Among these, microRNAs (miRNAs) modulate gene expression at the transcriptional level, thereby regulating pathological and physiological processes in recipient cells [20]. Prolonged in vitro culture can lead to diminished biological and paracrine functions of MSCs. Numerous studies have demonstrated that preconditioning of MSCs with drugs, cytokines, or physical stimuli enhances the biological activity of the EVs they secrete [21, 22]. For instance, exosomes from hypoxic-preconditioned adipose-derived MSCs exhibit superior angiogenic potential and graft survival compared to those from untreated MSCs [23]. Huang et al. demonstrated that exosomes from atorvastatin-treated MSCs confer greater cardioprotection, potentially via enhanced endothelial cell function [24]. Pioglitazone, a highly selective and potent peroxisome proliferator-activated receptor gamma (PPARγ) activator, is a commonly prescribed antidiabetic agent [25, 26]. However, whether EVs derived from pioglitazone-pretreated MSCs exhibit enhanced therapeutic efficacy in DMED and the underlying mechanisms remain unexplored.

This study aimed to investigate the effects of BMSC-derived EVs from pioglitazone-pretreated cells (PGZ-EVs) delivered via ECMTA-Hydrogel on erectile function in a DMED rat model. By establishing a rat DMED model and administering intracavernosal injections of EVs from untreated or pioglitazone-pretreated BMSCs, we demonstrated that PGZ-EVs significantly enhanced the recovery of erectile function. To elucidate the underlying mechanisms, we performed high-throughput sequencing of EVs and PGZ-EVs to identify differentially expressed miRNAs. Subsequent target gene prediction and validation through in vivo and in vitro experiments revealed the involved pathways.

Materials and methods

Animals

All animal procedures were approved by the Institutional Animal Care and Use Committee of Shandong Provincial Hospital (No. 2023 − 175). Forty-eight 8-week-old male Sprague-Dawley rats were obtained from Jinan Pengyue Laboratory Animal Breeding Co., Ltd. and acclimatized for one week. To induce diabetes, 40 rats were fasted for 12 h and then received a single intraperitoneal injection of streptozotocin (STZ; 60 mg/kg in citrate buffer, Solarbio, China). The remaining eight rats (control group) received an i.p. injection of citrate buffer alone. Blood glucose levels were measured 72 h post-injection. Rats with blood glucose levels exceeding 16.7 mmol/L were considered successfully diabetic and included in the study. The rats were then randomly divided into six experimental groups (n = 8 per group):

Control group: Intracavernosal injection of 100 µL PBS.

Diabetes group: Intracavernosal injection of 100 µL PBS.

ECMTA group: Intracavernosal injection of 100 µL ECMTA hydrogel.

EVs group: Intracavernosal injection of 100 µL PBS containing 100 µg EV protein.

PGZ-EVs group: Intracavernosal injection of 100 µL PBS containing 100 µg PGZ-EV protein.

ECMTA-PGZ-EVs group: Intracavernosal injection of 100 µL ECMTA hydrogel containing 100 µg PGZ-EV protein. We conducted a subcutaneous apomorphine erectile test at the end of week 8. To evaluate erectile function, apomorphine (100 µg/kg, dissolved in ascorbic acid-supplemented normal saline) was administered subcutaneously into the dorsal neck of the rats. The animals were immediately observed for 30 min to record the frequency of penile erections. A valid erection was defined as penile enlargement, glans engorgement, or distal shaft exposure. Erectile dysfunction in the model group was confirmed if no erections were observed during the 30-minute period, or if the total erection frequency was significantly reduced compared to the normal controls. Diabetic rats exhibiting a negative erectile response to apomorphine were enrolled in the subsequent treatment group. Intracavernosal treatments were administered at the end of week 8, and terminal functional assessments including intracavernosal pressure measurements were conducted 4 weeks after treatment at week 12 (Fig. S4A). Fig. S4D shows the results of blood glucose measurements for each group of rats during our experiments.

Isolation and characterization of MSC-EVs

Rat bone marrow mesenchymal stem cells (BMSCs, OriCell) were cultured in medium supplemented with 10% fetal bovine serum (FBS) until 80% confluence. Cells were then incubated for 48 h in medium containing either 10% exosome-depleted FBS or 10% exosome-depleted FBS supplemented with 50 µM pioglitazone to generate PGZ-EVs. For each EV preparation, approximately 2 × 108 cells were used, and the conditioned medium (50 mL) typically yielded about 100 µg of EVs protein. The conditioned medium was collected, sequentially centrifuged, and filtered (0.22 μm). EVs were isolated via ultracentrifugation (100,000×g for 90 min), washed in PBS, and pelleted again (120,000×g, 90 min). The final pellet was resuspended in 200 µL PBS for further use. EV morphology was examined by transmission electron microscopy (TEM). Size distribution was determined by nanoparticle tracking analysis (NTA). Western blotting was performed to detect the EV surface markers CD81 and CD63 (27855-1-AP, 25682-1-AP, Proteintech, China).

Cell viability assay

The cytotoxicity of the ECMTA-Hydrogel was evaluated using a Cell Counting Kit-8 (CCK-8, C0037, Beyotime Biotechnology, China). Corpus cavernosum smooth muscle cells (CCSMCs) were seeded in 96-well plates (4,000 cells/well). After 24 h, cells were treated with various concentrations of ECMTA-Hydrogel (10–320 µg/mL). Following 72 h of incubation, optical density (OD) at 450 nm was measured using a microplate reader (Thermo Scientific, USA).

Erectile function assessment

Erectile function was evaluated by measuring the maximum intracavernous pressure (maxICP) and mean arterial pressure (MAP). Rats were anesthetized, and catheters were placed in the carotid artery (for MAP monitoring) and the corpus cavernosum (for ICP measurement). The cavernous nerve was exposed and electrically stimulated (5 V, 15 Hz, 5 ms pulse width, 60 s duration). ICP and MAP were recorded continuously using a BL-420 V pressure sensor system.

ELISAs for SOD, MDA and GSH

After euthanizing rats in each group, penile tissue was collected. An ELISA kit (DYC3419–2, KGE013, R&D Systems, USA; EEA020, Invitrogen, USA) was used to assess levels of malondialdehyde (MDA), superoxide dismutase (SOD), and glutathione (GSH) in the penile tissue. The unit for measuring SOD is U/mg protein, while the units for measuring MDA and GSH are nmol/mg protein. Furthermore, these measurements are calibrated based on the wet weight of sponge tissue.

In vivo tracking of EVs

EVs were labeled with a DiR fluorescent dye and injected into the rat corpus cavernosum, either in PBS or encapsulated within the ECMTA hydrogel. Fluorescence signals were monitored immediately after injection and at 1 and 2 weeks post-injection using an IVIS Spectrum In Vivo Imaging System (PerkinElmer).

Cell transfection

Third-passage CCSMCs were cultured in six-well plates. Lentiviral vectors for overexpressing miR-203a-3p (100314161, Shanghai Genechem Co., Ltd., China) and TMEM33 (59303, Genomeditech, China) were transduced into cells separately, following the manufacturer’s protocols.

Dual-Luciferase reporter assay

To verify whether TMEM33 is a direct target of miR-203a-3p, the full-length 3’-untranslated region (3’-UTR) of TMEM33 and a mutant version with an altered miR-203a-3p binding site were cloned into the pmirGLO vector (Promega). HEK 293T cells were seeded in 96-well plates and transfected with either miR-203a-3p lentivirus or a negative control. Subsequently, cells were co-transfected with 50 ng of pGL3 vector containing either the wild-type or mutant TMEM33 3’-UTR sequence, along with 10 ng of the pRL-TK vector (Promega), using Lipofectamine LTX (Invitrogen). After 24 h, firefly and Renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega).

Masson’s trichrome staining

Masson’s trichrome staining was employed to evaluate smooth muscle and collagen fiber expression in the corpus cavernosum, assess the extent of tissue decellularization, and monitor the in vivo degradation of the hydrogel. Paraffin-embedded tissue sections were deparaffinized and hydrated, then stained following standard protocols, resulting in red-stained smooth muscle fibers and blue-stained collagen fibers. Image analysis was performed using Image-Pro Plus 5.0 software (Media Cybernetics, Inc., Bethesda, MD, USA).

Reverse transcription quantitative polymerase chain reaction (RT-qPCR)

Total RNA was extracted from tissue or cell samples using an RNA isolation kit (R0017M, Beyotime Biotechnology, China). cDNA was synthesized by reverse transcription. RT-qPCR was subsequently conducted on a LightCycler 480 II system (Rochester, USA) using SYBR Green PCR reagents (Toyobo, Osaka, Japan). The U6 small RNA and β-actin were used as internal controls for miRNA and mRNA normalization, respectively. Relative gene expression levels were calculated using the 2^(-ΔΔCT) method. All primer sequences are listed in Table 1.

Table 1.

Primer sequences used for RT–qPCR

Gene	Primer Sequence (5′ to 3′)
miRNA-203a-3p	Forward: CGCGGTGAAATGTTTAGGAC Reverse: AGTGCAGGGTCCGAGGTATT
TMEM33	Forward: CCTGGATGCAAAGGGTTCAAATA Reverse: AGCAAACTTCCTTGACCACTAAAG

Immunofluorescence staining

Penile tissues and CCSMCs were fixed with 4% paraformaldehyde. Tissues were cryoprotected by immersion in a 30% sucrose/PBS solution overnight at 4 °C, then sectioned into 5 μm slices. After blocking with 5% BSA for 1 h, sections or cells were incubated overnight at 4 °C with the following primary antibodies: GRP78 (1: 400, 11587-1-AP, Proteintech),α-SMA ༈1༚800, 67735-1-Ig, Proteintech༉, OPN ༈1༚800, 22952-1-AP, Proteintech༉, CD68 (1༚800, 25747-1-AP, Proteintech) and Desmin (1༚800, 16520-1-AP, Proteintech). After washing, the samples were incubated with CoraLite488- or CoraLite594-conjugated secondary antibodies ༈1༚500, Proteintech༉ for 1 h at room temperature. Nuclei were counterstained with DAPI ༈Beyotime Biotechnology, China༉, and images were captured using a fluorescence microscope ༈Nikon, Tokyo, Japan༉.

Western blot analysis

Tissues and cells were lysed using RIPA buffer supplemented with protease and phosphatase inhibitors. Protein concentration was determined via a BCA assay. Equal amounts of protein (20 µg) were separated by SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, USA). After blocking with 5% skim milk, the membranes were incubated overnight at 4 °C with primary antibodies against TMEM33(1:5000, 30639-1-AP, Proteintech), GRP78༈1:6000, 11587-1-AP, Proteintech༉, PERK༈1:3000, ༈1:6000, 24390-1-AP, Proteintech༉, pPERK༈1:6000, 82534-1-RR, Proteintech༉, ATF4༈1:800, 10835-1-AP, Proteintech༉, CHOP༈1:800, 15204-1-AP, Proteintech༉, β-actin༈1: 1000, 81115-1-RR, Proteintech༉, Bcl-2༈1༚1000, 26593-1-AP, Proteintech༉, Bax༈1༚1000, 50599-2-IG, Proteintech༉, Cleaved Caspase-3༈1༚2000, 25128-1-AP, Proteintech༉, OPN༈1༚1000, 22952-1-AP, Proteintech༉, and α-SMA ༈1༚800, 67735-1-Ig, Proteintech༉. Following incubation with HRP-conjugated secondary antibodies, protein bands were visualized using a Fujifilm LAS-4000 imaging system༈Tokyo, Japan༉ and analyzed with Image-Pro Plus 5.0 software.

High-throughput sequencing of miRNAs

Total RNA was extracted from three independent batches of EVs and PGZ-EVs using Trizol reagent. RNA quality and integrity were assessed using a Nanodrop 2000 (Thermo Fisher Scientific Inc., USA) and an Agilent 2100 Bioanalyzer (Agilent Technology, USA). miRNA microarray analysis was performed using the Affymetrix platform according to standard protocols (FlashTag Biotin HSR RNA Labeling Kit, hybridization, washing, and scanning). Differential expression analysis was conducted using the Limma package in R, with differentially expressed miRNAs (DEMs) defined as those with a fold change > 2 and an adjusted p-value (PADJ) < 0.05. Potential target genes of DEMs were predicted by integrating results from the TargetScan, miRWalk, GeneCards, and miRDB databases. Gene Ontology-Biological Process (GO-BP) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using Fisher’s exact test, with a significance threshold of p < 0.05.

Isolation and characterization of corpus cavernosum smooth muscle cells (CCSMCs)

CCSMCs were isolated from the penile tissues of 8-week-old rats. Briefly, Tissue were minced into small pieces measuring 1–3 mm³, then placed in an inverted culture flask. A small amount of Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% FBS, 1% penicillin, and 1% streptomycin should be added, and the flask should be placed in a 5% CO₂ incubator for 30 min [27]. The flask is then righted to fully submerge the tissue in the medium. Corporeal smooth muscle cells (CCSMCs) migrating from the tissue chunks are purified using differential adhesion velocity methods. CCSMCs were characterized by immunofluorescence staining for α-SMA and Desmin (Fig. S5A, B). To simulate a chronic diabetic environment in vitro, cells were treated with 30 mM glucose for 72 h. To induce endoplasmic reticulum stress (ERS) in vitro, CCSMCs were treated with 8 µg/mL tunicamycin (TM, dissolved in DMSO) for 12 h.

Flow cytometry analysis

Label free-floating EVs and EVs released from ECMTA hydrogels using the Exosome Labeling Kit (C3637S, Beyotime Biotechnology, China). After removing unbound dye by centrifugation-based purification, the labeled EVs were incubated with CCSMCs for 24 h at 37 °C. Following incubation, cells were washed, detached, and resuspended in PBS. The percentage of fluorescent cells, indicative of EV uptake, was quantified using a (ACEA NovoCyte, USA). Data were analyzed using FlowJo software. Untreated CCSMCs were used as a negative control to set the gating threshold.

Hydrogel preparation and characterization

Preparation and characterization of the decellularized extracellular matrix (ECM)

Fresh porcine penile tissue was decellularized to obtain the ECM. Tissues were cut into small pieces and soaked in sterile deionized water for 6 h, followed by agitation in 1% Triton X-100 solution (with antibiotics) for 48 h. The solution was replaced every 5 h. After thorough washing with PBS to remove residual detergents, the tissue was frozen at -80 °C for 96 h, lyophilized, and ground into a fine powder using a micro-grinder. The powder was sterilized by gamma irradiation and stored at -20 °C. For hydrogel preparation, 200 mg of ECM powder was digested with 40 mg pepsin in 20 mL of 0.1 M HCl for 48 h under sterile conditions. The pH was then neutralized to 7.0 using 1 M NaOH, and the ionic strength was adjusted with 10X PBS. This ECM solution forms a gel at 37 °C.

Preparation and Characterization of the ECMTA Hydrogel

First, 1 µM of TA was dispersed into 1 mL of ECM solution (1 µM mL − 1)to form the ECMTA solution. The mixturewas placed into a mold to synthesize the ECMTA at37 ℃. Scanning electron microscopy (SEM, FEI, USA) was used to investigate the micromorphology.

Fourier transform-infrared (FTIR) spectroscopy

FTIR analysis of the ECM and ECMTA hydrogels was performed using a Nicolet 6700 spectrometer (Thermo Scientific, USA) to characterize their functional groups and chemical bonds. Spectra were recorded in the range of 4000–400 cm⁻¹ at a resolution of 4 cm⁻¹.

Scanning electron microscopy (SEM)

The microstructure of the hydrogels was analyzed by SEM. Briefly, the formed hydrogels were frozen in liquid nitrogen, fractured, and the cross-sections were sputter-coated with gold. Images were acquired using a VEGA3 SEM (VEGA3, TESCAN, Czechia), and the average pore size was measured using ImageJ software.

In vitro EVs release profile

The release kinetics of EVs from the ECMTA hydrogel were evaluated using a bicinchoninic acid (BCA) protein assay. A 200 µL aliquot of ECMTA hydrogel containing 50 µg of EVs was added to the upper chamber of a Transwell insert (8 μm pore size, Corning) and incubated at 37 °C for 30 min to form a gel. PBS was added to the lower chamber. At predetermined time points, 20 µL of release medium was collected from the lower chamber and replaced with an equal volume of fresh PBS. The amount of EVs released was quantified, and the cumulative release curve was plotted.

Hematoxylin and Eosin (H&E) staining

Porcine penile tissues and rat organ specimens were fixed in 4% paraformaldehyde for at least 24 h. After dehydration through a graded ethanol series, tissues were embedded in paraffin, sectioned into 4–6 μm slices, and stained using a standard H&E staining kit (Servicebio, China).

Verhoeff-Van Gieson (EVG) staining

Paraffin-embedded sections of pig penile tissue (4–6 μm) were stained using a Verhoeff’s elastic fiber stain kit (Shyuanye, Shanghai, China) according to the manufacturer’s instructions to visualize elastic fibers.

Rheological analysis

The rheological properties of the ECMTA hydrogel were characterized using an Anton Paar MCR-301 rheometer with a 25 mm diameter parallel plate geometry. Hydrogel samples were molded into discs (25 mm diameter, 1 mm thickness). Frequency sweep tests were conducted from 0.1 to 100 Hz at a constant strain of 1%. Viscosity was measured by increasing the shear rate from 0.1 to 10 s⁻¹ at a constant frequency of 1 Hz.

Blood collection and biochemical analysis

After a 12-hour fast, blood samples were collected from anesthetized SD rats via cardiac puncture. Approximately 3–4 mL of whole blood was drawn into serum separation tubes (SST), allowed to clot at room temperature for 30 min, and then centrifuged at 3,000×g for 15 min at 4 °C. The resulting serum was aliquoted and analyzed for key biochemical parameters using a Hitachi 7020 automated clinical chemistry analyzer.

Statistical analysis

All experiments were performed with at least three independent replicates. Data are presented as mean ± standard deviation (SD). Statistical analyses were conducted using GraphPad Prism 8 software. Comparisons between two groups were analyzed using an unpaired, two-tailed Student’s t-test. For multiple group comparisons, one-way ANOVA was employed, followed by Tukey’s post hoc test for pairwise comparisons. A p-value of less than 0.05 was considered statistically significant.

Results

Preparation and characterization of ECMTA-Hydrogel

The extracellular matrix (ECM) hydrogel was fabricated from decellularized porcine corpus cavernosum tissue (Fig. S1A). Histological analyses, including hematoxylin and eosin (H&E), Masson’s trichrome, and Verhoeff–Van Gieson (EVG) staining, confirmed the effective removal of cellular components while preserving the native collagen and elastic fiber architecture in the decellularized tissues (Fig. 1A). Quantitative assessment further confirmed that the DNA content after decellularization treatment was 32.33 ± 4.04, while collagen was preserved (Fig. 1B). This measurement falls well below the rigorous 50 ng/mg threshold established by Crapo et al. [28]. This approach avoids potential local immunogenic reactions in the penis that could arise from hydrogels. Scanning electron microscopy (SEM) images of both native and decellularized tissues corroborated the maintenance of the ECM microstructure post-processing (Fig. 1C, D).

Fig. 1.

Preparation and characterization of ECMTA-hydrogel. (A, B) HE staining, Masson staining, EVG staining, and quantitative analysis before and after decellularization. (C, D) SEM images of penile tissue before and after decellularization. (E) Real-world states of the hydrogel at different temperatures. (F) Infrared spectra of ECM and ECMTA. (G) SEM image of ECMTA-hydrogel. (H) Flow properties of ECMTA-hydrogel as a function of temperature. (I) CCK8 assay showing cell viability after treatment of CCSMCs with different concentrations of ECMTA-hydrogel (n = 3). (*P < 0.05, **P < 0.01, ***P < 0.001; ns: no significant difference; bar graphs represent the mean ± SD)

To enhance the functionality of the ECM-based hydrogel, tannic acid (TA)—a natural polyphenol known for its anti-inflammatory, antioxidant, and antibacterial properties—was incorporated to form the ECMTA-Hydrogel [29, 30]. This modification aimed to mitigate the high oxidative stress microenvironment characteristic of diabetic penile tissue. ELISA assays measured MDA, GSH, and SOD levels to evaluate penile oxidative stress in rats. Results showed significantly elevated penile oxidative stress in the DMED group, while ECMTA treatment improved this condition (Fig. S1B). Fourier transform-infrared (FTIR) spectroscopy confirmed the successful integration of TA, as evidenced by characteristic absorption peaks in the ECMTA spectrum compared to ECM alone (Fig. 1F). Rheological characterization demonstrated the thermosensitive behavior of the ECMTA-Hydrogel, which transitions from a liquid state at ambient temperature to a stable gel at 37 °C, enabling facile injection and localized delivery into the corpus cavernosum (Fig. 1H). Visual inspection at different temperatures validated this sol–gel transition (Fig. 1E).

SEM imaging of the lyophilized ECMTA-Hydrogel revealed a homogeneous porous morphology with well-interconnected networks, indicative of uniform cross-linking and favorable microstructure for bioactive molecule encapsulation (Fig. 1G). Biocompatibility and safety were systematically evaluated both in vitro and in vivo. CCK-8 assays performed on corpus cavernosum smooth muscle cells (CCSMCs) exposed to various concentrations of ECMTA-Hydrogel showed no significant cytotoxicity, confirming its high biocompatibility (Fig. 1I). Moreover, after four weeks of in situ implantation in the penile tissue of healthy SD rats, histological evaluation of major organs and analysis of blood biochemical parameters revealed no signs of tissue damage or systemic toxicity (Fig. S2A, D). HE staining and immunofluorescence analysis of the penis revealed no abnormal macrophage infiltration at the hydrogel injection site, thereby avoiding potential risks of immune - fibrosis - functional impairment in the penile region (Fig. S2B, C), further attesting to the biosafety of the hydrogel.

ECMTA-Hydrogel prolongs the retention and release of EVs

TEM analysis confirmed the characteristic cup-shaped morphology of both BMSC-EVs and PGZ-EVs (Fig. 2A). NTA revealed a particle size distribution ranging from 50 to 120 nm (Fig. 2C). Western blot analysis confirmed the presence of EV-specific markers CD81 and CD63 (Fig. 2B). Immunofluorescence imaging showed that EVs were uniformly distributed within the ECMTA-Hydrogel matrix (Fig. 2D). A sustained-release profile, quantified by BCA assay, demonstrated that the ECMTA-Hydrogel enabled a gradual release of over 90% of the encapsulated EVs over 20 days (Fig. 2E). To verify that EVs released from hydrogels retain intact vesicular structure and activity, we analyzed EVs released from hydrogels at different time points using TEM, NTA, and WB. Results demonstrated consistent vesicular structure and activity across all time points (Fig. S3A-C). Furthermore, by integrating our NTA and BCA measurements. Statistical analysis confirmed no significant difference in this ratio between the EVs group and PGZ-EVs group (Fig. S3D). After injecting EVs and ECMTA-EVs into the rat corpus cavernosum (Fig. 2F, Fig. S4B), In vivo imaging revealed strong initial fluorescence in both EV and ECMTA-EV groups post-injection. However, while the signal in the EV group diminished rapidly and was nearly undetectable after two weeks, the ECMTA-EV group maintained a significant fluorescence signal, indicating prolonged retention (Fig. 2G, Fig. S4C). Histological examination (Masson’s trichrome stain) confirmed that the ECMTA-Hydrogel did not disrupt the normal architecture of the corpus cavernosum, degraded progressively over four weeks, and was fully biocompatible. This feature prevents the potential for penile cavernous obstruction, and even penile necrosis, that could result from injecting hydrogel into the penis (Fig. 2H). We treated CCSMCs with EVs released from ECMTA hydrogels on days 11 to 15 and with free EVs for 24 h. Immunofluorescence colocalization and flow cytometry results demonstrated comparable uptake efficiencies between the two EV groups (Fig. S5C-F). Furthermore, we collected EVs released from ECMTA between days 11–15 and treated CCSMCs under high-glucose conditions. Western blot analysis demonstrated that these EVs effectively suppressed pathological phenotypic switching (Fig. S5G–I). These findings indicate that the ECMTA hydrogel can sustainably release EVs while effectively preserving their structural integrity and biological activity.

Fig. 2.

Characterization of EVs and PGZ-EVs and in vivo/in vitro release of EVs from ECMTA-hydrogel. (A, C) TEM images and size distribution of EVs and PGZ-EVs. (B) Western blotting analysis of CD81 and CD63 expression in EVs, PGZ-EVs, and MSCs. (D) Immunofluorescence observation of EVs distribution within ECMTA-hydrogel. (E) Quantitative measurement of ECMTA-hydrogel-released EVs via the in vitro BCA assay. (F) Schematic diagram of injecting EVs/ECMTA-EVs into the rat corpus cavernosum. (G) IVIS imaging of DiR-labeled EVs retention in rat penile tissue. (H) Masson’s trichrome stain of penile tissue after ECMTA-EVs injection

ECMTA-PGZ-EVs ameliorate erectile dysfunction in DMED rats

After 4 weeks of treatment in each group, Erectile function, assessed by the maxICP/MAP ratio, was significantly impaired in DMED rats compared to the control group. Treatment with ECMTA-Hydrogel and EVs improved this ratio, with PGZ-EVs exhibiting superior efficacy relative to plain EVs. The most pronounced therapeutic effect was observed in the ECMTA-PGZ-EVs group. This result indicates that our treatment maintains improvements in erectile function and cavernous body structure for at least 4 weeks. underscoring the dual benefit of EVs engineering and sustained hydrogel delivery (Fig. 3A, C).

Fig. 3.

Assessment of erectile function and penile Masson’s trichrome staining in rats post-treatment. (A) ICP curves for each group, with red lines indicating 60-second electrical stimulation periods. (B) Masson’s trichrome-stained penile tissue from different groups post-treatment. Smooth muscle appears red, while collagen appears blue. (C) Ratios of max ICP to MAP (n = 3). (D) Data represent the ratio of smooth muscle to collagen (n = 3). (*P < 0.05, **P < 0.01, ***P < 0.001, #P < 0.05, ##P < 0.01, ###P < 0.001; bar graphs represent the mean ± SD)

ECMTA-PGZ-EVs reverse pathological phenotypic switching and apoptosis in CCSMCS

In the pathophysiology of DMED, the switch from contractile to synthetic smooth muscle cells represents a key pathological mechanism. The primary biomarker for the contractile phenotype is α-SMA, while the synthetic phenotype typically involves osteopontin (OPN). Masson’s trichrome staining showed a significantly decreased smooth muscle-to-collagen ratio in the DMED group, which was ameliorated by ECMTA-Hydrogel and EVs treatments, most effectively by PGZ-EVs and ECMTA-PGZ-EVs (Fig. 3B, D). Immunofluorescence and Western blot analyses corroborated these findings, showing that DMED increased the expression of the synthetic phenotype marker osteopontin (OPN) and the pro-apoptotic proteins Bax and Cleaved Caspase-3, while decreasing the contractile phenotype marker α-SMA and the anti-apoptotic protein Bcl-2. ECMTA-PGZ-EVs treatment most potently reversed these pathological changes (Fig. 4A-D, Fig. S8A, B).

Fig. 4.

Phenotypic switching and apoptosis in CCSMCs of rats across treatment groups. (A) Representative immunofluorescence images showing α-SMA expression in penile tissues of rats across groups. (B) Representative immunofluorescence images showing α-SMA and OPN expression in rat penises, indicating phenotypic switching. (C) Western blotting analysis of BAX, cleaved-caspase3, BCL-2, α-SMA, and OPN protein expression in rat penises. (D) Data represent relative expression levels normalized to β-actin (n = 3). (*P < 0.05, **P < 0.01, ***P < 0.001, # P < 0.05, ##P < 0.01, ### P < 0.001; bar graphs represent the mean ± SD)

PGZ-EVs-derived miR-203a-3p directly targets TMEM33

High-throughput miRNA sequencing of EVs and PGZ-EVs identified 13 downregulated and 5 upregulated miRNAs in PGZ-EVs (Fig. 5A, C, Fig. S6A). Gene Ontology (GO) enrichment analysis and KEGG Pathway Enrichment Analysis indicated a significant association of the target genes with the endoplasmic reticulum stress (ERS) pathway (Fig. 5B). ERS refers to the accumulation of misfolded or unfolded proteins within the ER lumen exceeding a threshold level, leading to the activation of the unfolded protein response (UPR) signaling cascade from the ER to the nucleus [31, 32]. Extensive research indicates that ERS induces phenotypic switching and apoptosis in smooth muscle cells [33, 34]. Our previous study has confirmed that ERS plays a crucial role in the pathogenesis of DMED [35]. To further validate our findings, we simulated an in vitro diabetic environment by stimulating CCSMCs with high glucose, while excluding the effects of hyperosmolarity using mannitol. Western blot analysis revealed that high glucose stimulation significantly elevated CCSMCs ERS levels and increased pathological phenotypic conversion and apoptosis-related protein expression. Following administration of the ERS inducer TM, the aforementioned protein expression levels further increased. Conversely, treatment with the ERS inhibitor 4-PBA significantly reduced ERS levels, diminished pathological phenotypic switching, and decreased apoptosis-related protein expression (Fig. S7A, B). Additionally, another study indicates that upregulating miR-203a-3p exert therapeutic effects in diabetic retinopathy [36].Among the upregulated miRNAs, miR-203a-3p was selected for further investigation due to its reported role in diabetic complications. RT-qPCR confirmed that miR-203a-3p was downregulated in the Corpus cavernosum of DMED rats and in CCSMCs under high-glucose conditions (Fig. 5D, Fig. S6B). By integrating target gene prediction results from Targetscan, miRDB, and mirwalk, and cross-referencing with ERS-related genes in the GeneCards database, we identified transmembrane protein 33 (TMEM33) as a potential target gene of miR-203a-3p (Fig. 5E). This was validated by Western blot, which showed that miR-203a-3p overexpression in CCSMCs suppressed TMEM33 expression (Fig. 5F, G). A dual-luciferase reporter assay confirmed a direct interaction, as miR-203a-3p overexpression reduced luciferase activity in cells transfected with the wild-type, but not a mutant, TMEM33 3’UTR (Fig. 5H).

Fig. 5.

miR-203a-3p enriched in PGZ-EVs targets and downregulates TMEM33. (A) Heatmap showing differentially expressed miRNAs in EVs and PGZ-EVs. (B) GO-BP enrichment analysis and KEGG pathway enrichment analysis of target genes for differentially expressed miRNAs in EVs and PGZ-EVs. (C) Bar chart showing miR-203a-3p expression in EVs and PGZ-EVs (n = 3). (D) RT-qPCR demonstrating miR-203a-3p expression levels in penises of normal rats versus DMED rats (n = 3). (E) Venn diagram predicting miR-203a-3p target genes associated with ERS using Targetscan, micwalk, miRDB, and GeneCards databases. (F) Western blotting showing TMEM33 expression in CCSMCs between the NC group and miR-203a-3p overexpression group. (G) Bar graph showing relative expression of TMEM33 normalized to β-actin (n = 3). (H) Dual luciferase assay demonstrating miR-203a-3p inhibition of TMEM33 activity (n = 3). (*P < 0.01, **P < 0.001, ***P < 0.001; bar graphs represent mean ± SD)

PGZ-EVs attenuate endoplasmic reticulum stress in DMED rats

Western blot analysis of ERS pathway proteins revealed significant upregulation of PPERK, PERK, GRP78, ATF4, CHOP, and TMEM33 in the DMED group (Fig. 6A, B). These changes were mitigated by EV treatments, with the most substantial reduction observed in the ECMTA-PGZ-EVs group. Immunofluorescence staining for GRP78 provided consistent results(Fig. 6C, Fig. S8C).

Fig. 6.

PGZ-EV treatment downregulates ERS levels in the penises of DMED rats. (A) Western blotting shows the expression of TMEM33, GRP78, PERK, PPERK, ATF4, and CHOP proteins in rat penises across groups. (B) Data represent relative expression levels normalized to β-actin (n = 3). (C) Representative immunofluorescence images showing GRP78 expression in penises of rats across groups. (*P < 0.05, **P < 0.01, ***P < 0.001, #P < 0.05, ##P < 0.01, ###P < 0.001; ns: no significant difference; bar graphs represent mean ± SD)

The miR-203a-3p/TMEM33 axis mediates the protective effects of PGZ-EVs in vitro

To further validate our findings, We simulated a diabetic environment in vitro using high-glucose stimulation. Western blot and immunofluorescence staining revealed significantly elevated ERS levels in CCSMCs under high-glucose conditions. Treatment with PGZ-EVs effectively reduced ERS levels (Fig. 7A-D), inhibited pathological phenotypic switching, and decreased apoptosis. However, overexpression of TMEM33 markedly attenuated these protective effects (Fig. 7E-H). To validate the therapeutic mechanism of PGZ-EVs from multiple angles, we induced a high ERS environment in CCSMCs using the endoplasmic reticulum stress inducer TM. Western blot and immunofluorescence staining results showed trends consistent with those observed under high glucose conditions (Fig. S9A-H). This confirmed that PGZ-EVs exert their therapeutic effects through the miR-203a-3p/TMEM33 pathway.

Fig. 7.

Effects of PGZ-EVs on endoplasmic reticulum stress (ERS) levels, pathological phenotypic switching, and apoptosis in CCSMCs. (A) Western blotting showing protein expression levels of TMEM33, GRP78, PERK, PPERK, ATF4 and CHOP in CCSMCs across different groups. (B) Data represent relative expression levels normalized to β-actin (n = 3). (C) Representative immunofluorescence images showing GRP78 expression in CCSMCs from different groups. (D) Quantify the expression levels of GRP78 in CCSMCs across each group using ImageJ software (n = 3). (E) Western blotting showing protein expression levels of BAX, Cleaved Caspase-3 and Bcl-2 in CCSMCs across different groups. (F) ImageJ software was used to quantitatively analyze the expression levels of apoptosis-related proteins in CCSMCs across different groups. Data are presented as relative expression levels normalized against β-actin (n = 3). (G) Western blotting showing protein expression levels of α-SMA and OPN in CCSMCs across different groups. (H) ImageJ software was used to quantitatively analyze the expression levels of Phenotypic Switching-related proteins in CCSMCs across different groups. Data are presented as relative expression levels normalized against β-actin (n = 3). (*P < 0.05, **P < 0.01, ***P < 0.001; ns: no significant difference; bar graphs represent the mean ± SD)

Discussion

Diabetic erectile dysfunction represents a significant clinical challenge, with current first-line therapies often yielding suboptimal results. This study introduces a novel combinatorial strategy that synergizes an advanced biomaterial delivery system with engineered extracellular vesicles to address the key pathophysiological processes of DMED. Our findings demonstrate that: (1) The ECMTA-Hydrogel, derived from decellularized porcine corpus cavernosum and functionalized with tannic acid, serves as an exceptional thermosensitive platform for the sustained local delivery of EVs; (2) Pretreatment of MSCs with pioglitazone yields PGZ-EVs with enhanced therapeutic potency; and (3) The superior efficacy of PGZ-EVs is mechanistically linked to the delivery of miR-203a-3p, which directly targets TMEM33 to ameliorate endoplasmic reticulum stress, thereby mitigating pathological phenotypic switching and apoptosis in CCSMCs.

The complex anatomy and hemodynamics of the corpus cavernosum pose a major hurdle for achieving effective local drug concentration. While hydrogels have emerged as promising EV delivery vehicles [37], many synthetic variants raise biosafety concerns due to acidic degradation byproducts [38, 39]. Our ECMTA-Hydrogel overcomes these limitations by leveraging the innate biocompatibility of a native decellularized matrix [40, 41]. This hydrogel undergoes a favorable sol-gel transition at body temperature, enabling minimally invasive injection and precise localization. The incorporation of the polyphenol tannic acid further endows the system with antioxidant and anti-inflammatory properties, crucial for counteracting the diabetic oxidative microenvironment [42, 43]. While the ECMTA hydrogel alone exhibited measurable improvements in the local microenvironment by scavenging ROS, the profound structural tissue repair and cellular phenotypic reversal were primarily driven by the biologically active PGZ-EVs. Our release kinetics and in vivo tracking data conclusively show that this system effectively prevents the burst release of EVs, ensuring prolonged local bioavailability and significantly extending the therapeutic window without causing tissue damage or adverse inflammatory responses. This prolonged physical retention does not necessarily alter the intrinsic cellular uptake mechanism, but it provides a significantly expanded time window for CCSMCs to efficiently internalize the EVs.

Beyond delivery challenges, the inherent biological activity of naïve EVs can be insufficient. Preconditioning MSCs represents a potent strategy to enhance the bioactivity of their secreted EVs [44]. Pioglitazone, a PPARγ agonist, has been shown to augment the regenerative capacity of MSC-derived EVs in other diabetic contexts [45, 46]. Our study is the first to report its utility in generating EVs for DMED therapy. The corpus cavernosum of the penis is a network structure composed of numerous sinusoids, whose walls are primarily formed by CCSMCs [47].Our preliminary research confirms that pathological phenotypic switch and apoptosis of CCSMCs constitute one of the key pathological mechanisms in the pathogenesis of ED [48]. Hyperglycemia promotes SMCs phenotypic switch and apoptosis through multiple mechanisms, including endoplasmic reticulum stress, oxidative stress, and advanced glycation end products [33, 49, 50]. Contraction-type CCSMCs are crucial for maintaining cavernous structure and erectile function, whereas synthesis-type cells exhibit a dedifferentiated state, losing contractile capacity while gaining enhanced proliferation, migration, and ECM secretion abilities, leading to tissue fibrosis and vascular remodeling [51, 52]. Prolonged exposure to high stress further induces apoptosis in synthesis-type CCSMCs, resulting in reduced smooth muscle content, decreased cavernous compliance, and ultimately impaired erectile function [53, 54]. We found that PGZ-EVs were significantly more effective than their naïve counterparts in restoring erectile function and reversing the hallmark pathological features of DMED—namely, the phenotypic switch of CCSMCs from a contractile to a synthetic state and their increased susceptibility to apoptosis. Importantly, the functional and structural improvements observed 4 weeks after a single intracavernosal injection—such as the restoration of smooth muscle content and the reversal of fibrosis—highlight a durable structural regeneration rather than a mere transient pharmacological effect, underscoring the long-term clinical significance of this single-injection strategy.

To decipher the mechanism behind this enhancement, we focused on the miRNA cargo. Our high-throughput sequencing data showed that miR-203a-3p, along with miR-1b, miR-27a-5p, miR-3557-5p, and miR-9a-5p, was among the top five significantly upregulated miRNAs in PGZ-EVs compared to the control EVs. The miR-203a-3p is known to be dysregulated in diabetic complications and is implicated in protective pathways [36, 55]. Whereas its specific function in DMED remained unclear. In contrast, there is currently limited evidence supporting a significant protective effect of the other top-upregulated miRNAs (miR-27a-5p, miR-1b, miR-9a-5p, or miR-3557-5p) on diabetes and its complications. Therefore, we focused our mechanistic investigation on miR-203a-3p. We confirmed miR-203a-3p downregulation in DMED penile tissues and CCSMCs in a highly glucose-rich environment, identified TMEM33, an endoplasmic reticulum transmembrane protein involved in regulating the unfolded protein response, as its direct target. TMEM33 is a stress-induced endoplasmic reticulum transmembrane protein that participates in regulating the PERK and IRE1 signaling branches [56]. Furthermore, studies by Arhatte et al. demonstrated that inhibiting TMEM33 alleviates renal ERS, suggesting its potential as a therapeutic target [57]. The link between ERS and smooth muscle cell dysfunction is well-established, and our prior work has highlighted its pivotal role in DMED pathogenesis [35]. The findings that PGZ-EVs effectively suppressed the ERS pathway in vivo and in a chronic high-glucose (HG)-induced cellular model (further causally validated by the ER stress inhibitor 4-PBA), and that these benefits were substantially abolished upon TMEM33 overexpression, functionally validate the miR-203a-3p/TMEM33 axis as a crucial mechanism of action. Nevertheless, recognizing the multifactorial nature of DMED, we emphasize that ER stress represents one of several contributing mechanisms; other pathological pathways, such as oxidative stress and local inflammation, are also collectively targeted by the comprehensive cargo of PGZ-EVs and the hydrogel.

There are some limitations to this study. First, while we identified that miR-203a-3p is significantly enriched in the vesicles and plays a primary role in alleviating endoplasmic reticulum stress by targeting TMEM33, the sequencing data revealed other differentially expressed species. This suggests that the therapeutic effects may be supported by the synergistic actions of multiple biological components within the vesicles. Secondly, although our results demonstrate that this specific miRNA achieves a substantial recovery of the pathological phenotype, future research utilizing additional molecular tools could further characterize its individual contribution and its potential interactions with other molecules. This approach would allow for a more detailed mapping of the multifaceted therapeutic landscape. Finally, while no obvious immunogenicity or systemic toxicity was observed during the four weeks observation period, the long term safety and the complete degradation profile of the porcine derived hydrogel require further investigation over a longer duration to support future clinical translation.

Conclusion

In conclusion, we have developed and validated an innovative therapeutic paradigm for DMED that combines a sophisticated biomaterial strategy with engineered extracellular vesicles. The ECMTA-Hydrogel provides a biocompatible, injectable, and sustained-release platform that markedly improves the retention and duration of action of therapeutic EVs within the corpus cavernosum. Concurrently, the pioglitazone-induced engineering of MSC-EVs yields PGZ-EVs with superior bioactivity, largely attributable to the enrichment of miR-203a-3p. This miRNA acts by directly targeting TMEM33, thereby alleviating diabetes-induced endoplasmic reticulum stress, suppressing the pathological phenotypic switch of smooth muscle cells, and reducing apoptosis, ultimately leading to the recovery of erectile function. This work not only provides a promising combined therapeutic approach but also delineates a novel mechanistic axis, offering significant translational potential for the clinical management of DMED.

Author contributions

Hao Liu: Writing – review & editing, Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Zhenjie Zang: Writing – review & editing, Methodology, Data curation, Conceptualization. Danfeng Zhao: Supervision, Software, Resources, Conceptualization. Jing Zhang: Methodology , Conceptualization, Software. Zhenqing Wang: Writing – review & editing, Supervision, Methodology , Conceptualization. Qiang Fu: Investigation, Formal analysis, Data curation, Conceptualization. Keqin Zhang: Project administration, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Funding

This study was supported by the Shandong Provincial Natural Science Foundation (Grant number: ZR2021MH101, Grant number: ZR2023QH325), National Natural Science Foundation of China (Grant number: 82471653, Grant number: 82501950).

Data availability

Data will be made available on request.

Declarations

Competing interests

The authors declare no competing interests.