Injection of Vocal Fold Lamina Propria-Derived Hydrogels Modulates Fibrosis in Injured Vocal Folds

Abstract

Vocal fold (VF) fibrosis, often resulting from phonosurgery, radiation, or trauma, causes irreversible voice dysfunction due to excessive ECM deposition and increased tissue stiffness. No FDA-approved treatments for VF fibrosis exist, highlighting the need for novel antifibrotic therapies. TGF-β1 contributes to fibroblast-to-myofibroblast activation, leading to increased ACTA2 expression and collagen production via SMAD3, YAP1, and integrin signaling pathways. Leveraging the principle that local cells respond to tissue-specific signals, our ECM hydrogel, derived from decellularized vocal fold lamina propria (VFLP-ECM), reduced ACTA2 expression in TGF-β1-stimulated VF fibroblasts, showcasing antifibrotic potential. This study evaluates the therapeutic potential of VFLP-ECM hydrogel in a rabbit VF injury model. VFLP-ECM hydrogel or bovine type I collagen injections were administered 7 days post-injury and evaluated on day 28. We compared two VFLP-ECM formulations: a manual process (VFLP (man)) and an accelerated automated method (VFLP (au)). VFLP (man) modulated fibrosis-associated gene expressions more effectively than VFLP (au) and collagen. Proteomics identified 229 proteins uniquely preserved in VFLP (man), including vitronectin, crucial in TGF-β1 signaling and ECM remodeling. Transcriptomic analysis suggests downregulation of fibrotic markers and inhibition of SMAD3, YAP1, and MRTFA, alongside upregulation of SMAD7, an inhibitor of TGF-β signaling. Notably, VFLP (man) treatment recovered stiffness comparable to uninjured controls (1.84 vs. 1.94 mN), whereas collagen-treated tissues remained stiff (2.7 mN), similar to the injury group (2.6 mN), indicating incomplete mechanical recovery. These in vivo data show that manually decellularized VFLP-ECM hydrogel attenuates fibrosis by disrupting key biochemical and mechanical cues driving myofibroblast activation.

Graphical abstract

1.0. Introduction

Vocal fold (VF) fibrosis is one of the most challenging clinical consequences of laryngeal injury, causing irreversible biomechanical disruption and voice dysfunction. It poses a significant barrier to effective rehabilitation for patients with voice disorders. Fibrosis is characterized by increased tissue stiffness, impaired mucosal wave propagation, and often irreversible dysphonia.^1–4 This condition accounts for a substantial proportion of chronic voice disorders and frequently occurs following phonosurgery, radiation therapy, intubation injury, or mechanical trauma.^5,6 Unlike benign VF lesions such as nodules or polyps, fibrotic remodeling often results in permanent damage due to a loss of tissue pliability and the failure to restore the native extracellular matrix (ECM).^1,7,8 Early intervention and prevention of fibrosis are critical, but current therapeutic options are limited. No FDA-approved treatments specifically target VF fibrosis, and clinical management is limited to voice therapy, surgical scar excision, or injection of bulking agents.⁹ Although these approaches may temporarily improve glottal closure or voice quality, none address the underlying pathophysiology of fibrosis. There is a pressing need for novel, mechanism-driven therapies to restore the integrity and function of the tissue.

Fibrosis is characterized by the disorganized accumulation of ECM components, particularly driven by the persistent fibroblast-to-myofibroblast activation.^1,10,11 Central to this process is transforming growth factor-beta 1 (TGF-β1) signaling, which activates SMAD3 and promotes the transcription of profibrotic genes, including COL1A1, ACTA2 (α-SMA), and FN1.¹² In parallel, non-canonical pathways—including YAP1, integrin-mediated mechanotransduction, and Rho GTPase signaling—further perpetuate myofibroblast persistence and matrix remodeling.¹³ These signaling pathways contribute to cytoskeletal reorganization, cell contraction, and overproduction of collagen, leading to matrix stiffening and impaired tissue compliance.¹⁴ Chronic activation of these pathways sustains the disorganized and stiff fibrotic tissue. It creates a feed-forward loop in which mechanical stiffening reinforces fibrotic signaling, further complicating efforts to restore normal function.^15,16 Therapies that specifically target these processes could provide relief to injured VFs.

Recent approaches to address VF fibrosis focus on harnessing local regenerative processes to promote constructive remodeling by providing signals to resident cells and, if needed, recruiting reparative cells to the injury site.^{1,4,5,12,17,18} One example is the use of injectable hydrogels made from ECM and derived components such as collagen, elastin, laminin, glycosaminoglycans, fibronectin, and matrix-bound vesicles (MBVs), which might provide structural support and biological signaling cues.^19,20 Decellularized ECM scaffolds support constructive tissue remodeling by minimizing fibrosis and foreign body responses. As they degrade, they release signals that modulate immune cells (e.g., macrophages, T cells, mast cells) and influence fibroblasts and epithelial cells.²¹ Injectable ECM hydrogels offer additional advantages by enabling minimally invasive delivery and the ability to conform to the site of delivery.^22,23 These bioactive materials can modulate inflammation and fibrosis by suppressing pro-inflammatory signals and promoting healing.

ECM biomaterials have been previously explored for VF regeneration with varying compositions, bioactivities, and therapeutic efficacies.^24–26 Here, we provide a brief overview of relevant efforts to develop adequate ECM-scaffolds for VF treatment. Porcine urinary bladder matrix (UBM-ECM) has been investigated in sheet form for VF repair but has failed to modulate scar formation effectively.^24,25 While UBM-ECM demonstrated biocompatibility and supported tissue regeneration in a canine hemilarynx reconstruction model, it resulted in scarring of the vocal fold mucosa.²⁴ Small intestinal submucosa (SIS), another xenogeneic ECM with broad availability and known regenerative properties, has been tested in injectable form.^6,26,27 Although SIS shares many core matrisome proteins with VF-derived ECM, SIS hydrogels alone did not promote scarless healing in acute VF injury models. However, when combined with mesenchymal stem cells (MSCs), SIS hydrogels improved glottal closure and enhanced neo-ECM deposition in scarred rabbit VFs.²⁸ Liver-derived ECM has also been proposed as a scaffold for VF repair and was shown to support regeneration in a canine model of acute injury.²⁹ Similarly, purified collagen, most often type I collagen, has also been used to treat VF injury. Type I collagen serves as a benchmark material due to its clinical availability, biocompatibility, mechanical relevance, defined signaling properties, and widespread use as a 3D injectable bulking scaffold. However, it has generally served as a control due to its limited ability to drive functional tissue regeneration.³⁰ Hyaluronic acid (HA), an endogenous ECM component, has been tested as an injectable VF treatment owing to its favorable viscoelastic properties and role in ECM homeostasis.^34,35 Bovine vocal fold lamina propria (Bov-VFLP-ECM) has also been evaluated as a potential biomaterial derived from a detergent-free, saline-based decellularization method, which supports cell infiltration and enables in vivo growth factor delivery.³¹ In a rabbit model of chronic VF scarring, Bov-VFLP-ECM gel decreased collagen content, increased lamina propria thickness, and triggered only a minimal humoral immune response. The choice of bovine tissue was guided by its anatomical similarity to the human VF and the high sequence homology of bovine type I collagen with its human counterpart.^32–34

However, several limitations remain in developing ECM-based therapies for VF repair, primarily due to the poor modulation of VF fibrosis. Decellularization methods carry some challenges: physical techniques like freeze-drying and sonication may damage the scaffold structure, and there is a lack of validation for immunogenicity and recellularization.³⁵ At the same time, some detergent-based protocols fail to remove DNA, posing risks for clinical use. Overall, the lack of optimized decellularization protocols for VF-specific ECM remains a barrier towards clinical translation and remains an active research area.^36,37 To address these concerns, our group developed a porcine VFLP-ECM with a unique composition, including MBVs and structural macromolecules. We previously reported that VFLP-ECM hydrogels suppressed myofibroblast activation and reduced ACTA2 expression in both human vocal fold fibroblasts (HVOX) and neonatal dermal fibroblasts (HDFn), suggesting potential antifibrotic properties.^{3,19,25,38,39} VFLP-ECM hydrogels also inhibited the activation of “M1-like” macrophages in vitro when stimulated by LPS. This anti-fibrotic and anti-inflammatory potential makes VFLP-ECM hydrogels a compelling therapeutic target for treating VF injuries.⁴⁰ Despite these promising in vitro findings, the in vivo therapeutic effects of VFLP-ECM remain largely unexplored. In this study, we evaluated the efficacy of VFLP-ECM hydrogel in a rabbit model of VF injury by transcriptional and translational shifts as well as tissue biomechanics 28 days after injection.

We also compared two formulations of VFLP-ECM: one generated using a traditional manual decellularization protocol and another produced through a streamlined, automated method that can yield raw material for the hydrogel in just 3 hours, compared to 1-2 days for the manual process. Although both formulations preserved core ECM components, only the manually prepared VFLP-ECM (VFLP (man)) triggered an antifibrotic response 28 days post-delivery. This formulation inhibited key profibrotic pathways, including TGF-β1/SMAD3 signaling and the activation of YAP1 and MRTFA, and reduced stiffness changes after injury, resulting in tissue that was comparable to uninjured tissue. Conversely, the automated ECM (VFLP (au)) produced transcriptomic and mechanical responses akin to those of injected bovine type I collagen. Still, VFLP (au) and Col1 materials resulted in improved outcomes compared to injury alone.

These findings provide evidence that VFLP-ECM, primarily produced through manual decellularization, is a biologically active biomaterial capable of modulating fibrosis-related pathways and promoting functional tissue recovery. By integrating transcriptomic, proteomic, and mechanical data, this study offers mechanistic insight into ECM-mediated antifibrotic therapy and establishes a foundation for future translational efforts in regenerative laryngology.

2.0. Methods

2.1. Materials and Reagents.

Laryngeal samples were obtained from a local market in Nahunta, Raleigh, NC. Reagents, including sodium deoxycholate (SDC), DNase Type I, ethanol, peracetic acid, and pepsin, were purchased from Millipore Sigma-Aldrich. Pluristrainer filters with 5-μm and 15-μm pore sizes were used to prepare injectable hydrogels. Type I collagen (FibriCol, Col1, Advanced Biomatrix) was used as a hydrogel control.

2.2. Vocal fold Lamina propria Extracellular Matrix (VFLP-ECM) Derivation

2.2.1. VF Decellularization

VFLP from the porcine vocal folds was dissected from the underlying muscle. The porcine VFLP-derived ECM was then processed using a manual (classical) or an automated protocol.

Manual Method:

Isolated porcine VFLPs were washed, pooled, and stored with minimal alterations to the intrinsic structure before decellularization.

Automated Method:

As previously described, VFLPs were snap-frozen in liquid nitrogen and minced into approximately 3 × 3 mm pieces using a commercially available garlic mincer (Amazon).^25,41,42 The minced tissues were stored at −20°C until further processing.

As detailed in Table 1, for both the manual and automated methods, tissues were centrifuged to remove excess water at 3000 RCF for 30 seconds and standardized by 1g of tissue mass per batch before undergoing sequential decell treatment. Although the same series of reagents was used in both methods, the increased surface area created by mincing in the automated method significantly reduced decellularization time. Notably, the automated approach required only one-third of the time needed for the manual protocol, thanks to a pump-controlled system that automated the reagent flow. After the decell protocol, the resulting dECM was collected, freeze-dried, and subjected to quality control assessments to evaluate decell efficacy.

Table 1.

Summary of decellularization protocols for generating vocal fold lamina propria extracellular matrix (VFLP-ECM). Each step includes the reagent used, exposure time, and order of application. The manual and automated protocols differ in duration, reagent exposure, and sequence, which may influence ECM composition and bioactivity.

Scale: 1 gram of native tissue		Time (minutes)
Step	Reagent	Manual	Automate
Wash 1	DPBS 1X	15	5
	DI_Water	15	5
	DPBS 1X	15	5
Detergent	SDC (4% W/V)	120	30
Wash 2	DPBS 1X	15	5
	DI_Water	15	5
	DPBS 1X	15	5
Enzymatic	DNAsel	120	30
Wash 3	DPBS 1X	15	5
	DI_Water	15	5
	DPBS 1X	15	5
Sterilization	Et-OH (4% V/V) & Perac. Acid	30	30
	0.1% V/V
Wash 3	DPBS 1X	15	5
	DI_Water	15	5
	DPBS 1X	15	5
Total time		450	150

2.2.2. Double-Stranded DNA (dsDNA) Quantification.

Three (3) mg of dried dECM was enzymatically digested using Proteinase K solution (Qiagen) and 180 μL of ATL digestion buffer (Qiagen). The specimens were placed in a heat block at 55°C overnight until full digestion occurred, ensuring no visible particles remained in suspension. The dsDNA in the samples was quantified using the QuantiFluor^® dsDNA System (Promega). Dilutions were performed to reduce the concentration of the ATL buffer below 400-fold to prevent interference with the readings. The protocol for measuring dsDNA was followed according to the manufacturer’s conditions using a fluorescent well plate reader.

2.2.3. Hematoxylin and Eosin (H&E)

Tissue samples were fixed in 10% neutral buffered formalin for 24 hours. Subsequently, the fixed tissues were dehydrated through a graded ethanol series (70%, 80%, 95%, 100%) for 1h at each concentration. The tissue was then cleared in xylene or a xylene substitute to remove ethanol and embedded in paraffin for sectioning and subsequent H&E staining.

2.2.4. Proteomic Analysis

A discovery proteomic workflow was employed to characterize and compare the overall protein composition of VFLP-Manual vs VFLP-Auto. VFLP-ECMs from three independent batches (each decell batch was pooled from at least 7 larynges) were used for each scaffold type, resulting in three independent replicates. Each sample was then suspended in 1mL of 50mM ammonium bicarbonate (pH 8.0) containing 2% (w/v) sodium deoxycholate (Sigma-Aldrich) to facilitate tissue digestion and homogenization. Protein concentration in each sample was determined using a bicinchoninic acid (BCA) assay. For subsequent mass spectrometry (MS) analysis, samples were prepared following a filter-aided sample preparation (FASP) protocol.

All samples were analyzed using a discovery proteomics workflow LC-MS/MS (MS). The prepared peptides were injected into an EASY-nLC 1200 system (Thermo Scientific, San Jose, CA) interfaced with an Orbitrap Exploris 480 (Thermo Scientific, Bremen, DE) mass spectrometer equipped with an EASY-Spray ion source. Peptides were first trapped on a Thermo Fisher Scientific Acclaim PepMap 100 column (C18, 3 μm particle size, 75μm inner diameter [ID], 20 mm length; catalog no. 164946) before separation on an EASY-Spray analytical column (C18, 2μm particle size, 75 μm ID, 250 mm length; catalog no. ES802A) maintained at 45°C.

Raw MS data files were processed using Proteome Discoverer version 2.5.0.400 (Thermo Fisher Scientific) in a label-free quantitation (LFQ) workflow. Peptides were searched against combined Sus Scrofa databases (SwissProt and TrEMBL; taxonomy ID = 9823). The resulting protein identifications and quantitative data were used to compare the protein composition of the acellular manual vs automated VFLP-ECM scaffolds.

2.2.5. VFLP Hydrogel Preparation

Digestion Process:

Independent of dECM derivation, the material was ground and digested using the protocol developed by Freytes et al. This protocol shows that a concentration of 10 mg/mL allows for complete digestion while preserving the hydrogel-forming properties of the ECM.⁴³ Briefly, 100 mg of dECM was digested in 10 mL of a pepsin solution in 0.1M HCl. The digestion was performed for 48 hours at low agitation until all the major ECM particles were digested. Aliquots of the pre-hydrogel were then stored at −20°C until use.

Preparation of Injectable Hydrogels:

To ensure the resultant hydrogels could be readily injected through a 27-gauge needle in the animal model, the pre-hydrogel decellularized ECM (dECM) solution was first passed through 200-μm pluriStrainer filters to remove any large particulates. For the preparation of injectable ECM hydrogels (either dECM-based or Type I collagen-based), 600 μL of a 10 mg/mL ECM solution was combined with 100 μL of 10× Dulbecco’s Phosphate-Buffered Saline (DPBS) and 50–90 μL of 0.1 M NaOH, added dropwise to adjust the pH to 7.4. The mixture was maintained on ice to prevent premature gelation, and cold deionized (DI) water was used to bring the total volume to 1mL. This approach ensures the hydrogels remain at a physiologically relevant pH, achieve the desired protein concentration, and possess sufficient fluidity for injection before their thermal gelation at 37°C.

2.3. Rabbit Injury Model

All procedures were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the NYU Grossman School of Medicine (approval numbers IA18-01360 and IA19-00011). These approvals ensured that the study design and animal handling adhered to institutional and national guidelines for the ethical and humane use of laboratory animals. A total of 30 New Zealand white rabbits (Oryctolagus cuniculus) supplied by Charles River Laboratories were used for this study. Rabbits were randomly assigned to five experimental groups of 6 animals each. Four groups of animals underwent VF injury with the following treatments detailed below: 1) collagen type 1 (‘Col1’); 2) VFLP (man), 3) VFLP (au), and 4) no treatments (‘Injury only’). An additional group was used as a “no injury, no injection” control. Three larynges from each condition were used for transcriptome analysis, and the other 3 were used for mechanical analysis.

2.3.1. Vocal Fold Injury

Rabbits were sedated with Ketamine (35 mg/kg) and dexmedetomidine (0.15 mg/kg) administered intramuscularly. If additional sedation was needed, mask isoflurane (2-3%) was used. While dorsal, a Storz 8574B (11cm) pediatric laryngoscope was inserted orally, and the vocal folds were visualized using a 3mm 85-degree flexible Ambu^® aScope^™ 4 RhinoLaryngo Slim endoscope connected to an aView^™ 2 Advance monitor. Topical lidocaine was applied to the vocal folds. Bilateral elevation in the membranous mucosal cover and epithelia was achieved using a custom microsurgical needle. Once mucosal elevation was completed, 2 mm curved micro cup cutting forceps (Karl Storz) were used to strip the mucosa, creating bilateral superficial injuries. Lidocaine was applied to the injury sites, and atipamezole (1mg/kg) was administered intramuscularly to reverse anesthesia.

2.3.2. Hydrogel Injections

Seven days post-injury, rabbits were sedated as described above. A Snowden-Pencer^™ Killian speculum was used to open the oral cavity. Under visualization, a 27-gauge, 5-inch Kel-F Hub Needle and Hamilton syringe were used to inject 20μl of collagen type 1, VFLP (man), and VFLP (au) (n=6 per condition: both vocal cords from each of 3 animals were analyzed separately) bilaterally for Transcriptome Analysis. For Mechanical Testing, only one vocal fold was treated and analyzed per animal, as only one VF can achieve full anterior commissure preservation. This paradigm resulted in three samples per condition for Mechanical Testing. Twenty-eight days after treatment, rabbits were sedated and euthanized via intravenous injection of sodium pentobarbital (390mg/kg) through the marginal ear vein. The larynx was excised, and the lumen was exposed. The entire larynx was pinned onto a Sylgard^® dissection dish. Each vocal fold underwent mucosal removal via dissection. Under stereoscopic magnification, an 18-gauge needle bevel was used to make incisions along the inferior and superior margins, extending from just medial to the arytenoid cartilage to just lateral to the anterior commissure. Fine-tipped forceps grasped the mucosa at the medial incision point, and the needle was used to dissect the mucosa from the muscle further. Once removed, the mucosa from both folds were placed in separate conical tubes and immediately flash frozen and placed in −80°C until further processing.

2.4. Transcriptome Analysis

Frozen vocal fold samples (0.5-3mg) were stored at −80°C in RNALater-ICE (Invitrogen) until RNA isolation. The tissue was homogenized in 1ml of buffer RLT+βME (Qiagen) using a Bead Mill 4 (Fisherbrand) for three cycles of 30s at a speed of 4m/s with 2min ice rests interspersed between cycles. Total RNA was isolated using the RNeasy Plus (Qiagen) extraction kit with the protocol adjusted proportionally for an input of 825μL of homogenized solution. RNA quality and quantity were assessed on the Bioanalyzer 2100 instrument using the RNA 6000 Pico Kit (Agilent). mRNA-seq libraries were prepared using a target input of 100ng of intact (RIN>7) RNA with the KAPA mRNA HyperPrep kit, implementing a fragmentation step of 94 °C for 6min and 15 cycles of final amplification (Roche). Library conversion and deep sequencing were performed using the App-A Universal Library Conversion kit and App-D Primer kit on a DNBSEQ-G400C sequencer with the FCL PE100 sequencing set (Complete Genomics).

RNA-seq data in the FASTQ file format was generated and demultiplexed on the DNBSEQ-G400C sequencer. Raw RNA-seq data were quality-controlled during and after sequencing to identify potential technical issues. Raw RNA-seq sequencing reads were mapped to the rabbit reference genome (assembly UM_NZW_1.0,⁴⁴ RefSeq annotation GCF_009806435.1-RS_2023_02) using STAR v2.7.10b ⁴⁵ to generate read counts for each of the annotated genes. RefSeq transcript annotations were supplied to facilitate mapping reads spanning known splicing junctions. The filterByExpr function from the edgeR package⁴⁶ was used to filter out lowly expressed genes for differential expression analyses. Raw gene read count data were normalized using the trimmed mean of M-values from edgeR. Differential expression analysis was performed using the linear model approach provided by the limma package.⁴⁷ To account for potential correlations between two vocal fold samples from the same animals, the animal identifier was included as a blocking variable in fitting the linear model, and the within-animal correlation was estimated using the ‘duplicateCorrelation’ function. For comparisons of each treatment against the uninjured control, the p-values for the coefficient/contrast of interest were adjusted for multiple testing of the Benjamini and Hochberg method ⁴⁸ to control for the expected false discovery rate (FDR). For comparisons between treatment groups and the ‘Injury only’ group, raw p-values less than 0.05 were used as the significance threshold for differential expression. Functional enrichment analysis of differentially expressed genes, including predicted enriched regulators and molecular pathways, was performed using QIAGEN’s Ingenuity Pathway Analysis (Redwood City, CA).⁴⁹ The raw RNA-seq data have been deposited in the NCBI GEO repository.

2.5. Mechanical Analysis

2.5.1. Tissue Preparation and Experimental Setup

Harvested larynges were bisected in the sagittal plane while preserving the anterior commissure, initially preserved at −80°C, and later thawed at 4°C before testing. Each sample was securely mounted onto custom-made molds made of Ortho-Jet resin (Lang Dental, Illinois, USA) to ensure stability during testing and mounted onto a MA646 specimen holder (Biomomentum Inc., Laval, QC, Canada). To maintain moisture, samples were covered in phosphate-buffered saline (PBS) at room temperature, and the tissue was allowed to equilibrate before testing.

2.5.2. Automated Indentation Mapping

Automated indentation mapping was performed using Biomomentum Mach-1 v500css (Laval, Quebec, Canada) with a calibrated uniaxial load cell (±1.5 N, Force Resolution 75 μN) according to established protocols.⁵⁰ An indentation mapping grid was implemented across the glottic and subglottic surface, comprising a minimum of 40 testing points per sample. Testing points were randomly distributed, with a higher spatial density concentrated in the mid-membranous region, which was defined as the area of interest. Indentation measurements were conducted using a 1.0 mm diameter spherical indenter. The indentation was performed at a constant velocity of 0.45 mm/sec and an indentation depth of 0.7mm, followed by a 5-second relaxation phase.

2.5.3. Analysis

Data processing was conducted using Mach-1 Analysis Software Version 4.1.0.19 (Biomomentum, Montreal, QC, Canada), and results are presented as mean values with corresponding standard deviations (SD) for each sample. Normal force values (mN) were extracted from the load vs. displacement curves at each measured position at 0.3 mm indentation depth. Structural stiffness (mN/mm) was calculated as the difference in measured load (at 0.3 and 0.0 mm) divided by the indentation depth (0.3 mm). The normal tissue displacement (mm) under a 1.96 mN load was also derived from the normal force versus position curves. Heat maps were created for each test parameter and superimposed onto sample images to visualize vocal fold mechanical changes under varying testing conditions. Outliers ± 5 SD from the mean were excluded from the analysis. A standardized color gradient was applied, with lower values depicted in blue and higher values in red.

2.5.4. Histological Processing and Analysis

Following biomechanical testing, vocal fold specimens were sectioned at the midportion along the anterior-posterior axis using a precision-engineered 3D-printed blade holder with dual razor blades spaced 3mm apart. Samples were fixed in 10% NBF, dehydrated, paraffin-embedded, sectioned at 5μm, and stained with H&E following standardized protocols. Heat maps were aligned with indentation maps and histological sections to reflect the biomechanical properties at the corresponding region of the H&E slide.

2.6. Statistical Analysis

Statistical analyses were performed using GraphPad Prism software (version 10.5). For mechanical characterization data, when indicated, comparisons between all treatment groups and the Injury-only group were conducted using Fisher’s Least Significant Difference (LSD) post hoc test following analysis of variance (ANOVA). Gene expression differences between groups were assessed using Student’s t-test. Statistical significance was defined at a threshold of p < 0.05.

3.0. Results

This project focused on developing an injectable vocal fold lamina propria extracellular matrix (VFLP-ECM) hydrogel for the treatment of vocal fold injuries. Prior in vitro data showed that VFLP-ECM effectively modulated alpha-smooth muscle actin (ACTA2), a robust marker of myofibroblast activation, at transcriptomic and protein levels.^8,12,19,51 These findings underscore the potential of tissue-specific scaffolds like VFLP-ECM in managing fibrosis. Figure 1 shows a visual summary of the experimental design and treatment groups.

Figure 1. In vivo Study overview for vocal fold injury treated with vocal fold lamina propria extracellular matrix (VFLP-ECM) hydrogels.

(A) Bilateral injury in the leporine model, followed by a waiting period for re-epithelialization to create a small pocket for hydrogel retention. (Bi) ECM was manufactured for injection to treat vocal fold injury in the rabbit model. Three types of ECM were tested: VFLP processed using the classical/manual method (VFLP-man), VFLP processed using an automated method (VFLP-au), and type I collagen (Col1) as a hydrogel control. Rabbit vocal folds were collected 28 days post-treatment. (Bii) Schematic representation of the expected effect of VFLP hydrogels in modulating the cell fate response of critical cell lines in the vocal fold. Fibrosis after injury will be assessed by comparing untreated injury against treatment groups, using transcriptomic analysis and mechanical property evaluation.

3.1. Hydrogel-ECM derivation.

We developed two versions of the VFLP-ECM scaffold: (1) using classical methods previously reported by Wrona et al. ²⁰ and (2) through a new workflow that enhances ECM production yield.^4,25,52 The latter involves identical reagents with modified exposure times as developed by Mora et al. Detailed timelines and protocols for these methods are presented in Table 1 of the Methods section. We procured larynges to assess the performance of both biomaterials as a treatment in an in vivo injury model. We dissected the VFLP tissues following the procedures described in the Methods. Figure 2A displays a macroscopic view of the porcine larynx, showcasing the enclosed VFLP tissues. Tissues were decellularized based on centrifugation weight conditions, where they were spun down in a centrifuge for 30 seconds at 3000 RCF to remove excess water and standardize the initial batch weight. dECM was produced and lyophilized for grinding before digestion using the Freytes method based on the pepsin digestion of the ECM.⁴³ The dsDNA content was quantified as shown in Fig. 2B. Native tissue had an average dsDNA content of 7.5 μg/mg dry tissue. After decellularization, total dsDNA content was significantly reduced. Still, the final content in the VFLP (man) version was 30 times higher than that in the automated VFLP. Similar amounts of dsDNA have been previously reported in ECM scaffolds without impacting their effectiveness in the host. ^53,54

Figure 2.

Injectable VFLP-ECM hydrogel derivation. A) Representative images of the porcine larynx. Following the methods described in Table 1, decellularized forms of the tissues are presented for manual processing - large pieces - and automated processing - cumulated small pieces. B) Quality control: double-stranded DNA content in non-decellularized (native) tissue versus decellularized VFLP (manual) and VFLP (automated). C) Representative images of H&E staining of the native VFLP and the decellularized ECMs (automated and manual). D) An injectable version of the VFLP-prehydrogel is available before storage for use in in vivo experiments.

H&E histology was performed on the material to evaluate the efficacy of the decellularization protocol, as depicted in Fig 2C. The absence of intact nuclei in the images confirmed the effective removal of nuclei, a result of the decellularization process (both manual and automated). As previously reported, the dECM (au and man) was digested using pepsin and filtered through a pluriStrainer to achieve a hydrogel with a particle size suspension suitable for administration through a 25-gauge needle. After filtration, the final, sterile product is shown in Fig 2D.

3.2. Discovery Proteomics Workflow (VFLP (man) vs. VFLP (au))

We analyzed the protein profiles of both ECM types with three replicates per ECM to assess their compositional differences (Figure 3). Prior to analysis, the dataset underwent preprocessing to include only high-confidence proteins and sub-units while excluding flagged proteins.

Figure 3. Comparative discovery proteomics of VFLP-ECM: manual vs. automated.

A) Venn diagram detailing proteins identified in manual and automated VFLP-ECM processes. Gene names highlighted within squares indicate further analysis via GeneMANIA, showing potential proteins in the material that physically interact with TGF-β1 (green font), ACTA2 (magenta), or both (orange). B) Volcano plot illustrating more proteins and subunits significantly identified in the manual VFLP-ECM. Triple asterisks (***) denote findings of particular interest due to their potential functional interaction inhibiting myofibroblast formation.

This processing identified 1,793 proteins and sub-units for VFLP (au), 2,179 for VFLP (man), and 1,748 proteins common to both methods. A Venn diagram was generated to illustrate the overlap and unique identifications between the two methods (Figure 3A). Although most proteins (1,748) appeared in both datasets, indicating a substantial overlap, 45 proteins were exclusively detected in VFLP (au) and 431 were solely detected in VFLP (man). To investigate the functional significance of these unique proteins, we employed GeneMania to analyze their physical interactions and pathway linkages, particularly with key fibrosis markers such as TGF-β1 and ACTA2. Key interacting proteins are detailed in Figure 3A. We used color-coded fonts to list between proteins that interact with TGF-β1 (in green) and those that interact or are pathway-related with ACTA2 (in magenta). Notably, the analysis of unique proteins in the VFLP (man) identified a critical matrix component, Vitronectin (VTN), abundant in this type of scaffold.

For proteins common to both methods, we conducted a differential expression analysis to assess changes in relative abundance between VFLP (au) and VFLP (man). Mean abundance was calculated for each protein across the three biological replicates using scaled abundance values generated by Proteome Discover. The mean abundances were then used to calculate a log2 fold change (log2FC) between the two methods, quantifying expression level differences for each protein or subunit detected. Statistical significance was evaluated using a paired t-test, and p-values were adjusted using the Benjamini-Hochberg method to control the false discovery rate. The results were visualized in a Volcano Plot (Figure 3B), highlighting proteins significantly upregulated (in red) and downregulated (in blue) in VFLP (man) compared to VFLP (au). This analysis identified 105 upregulated proteins in VFLP (man). However, as previously reported for other tissues, no significant differences in matrisome-associated proteins were observed across biomaterials.²⁵ The ten most statistically significant upregulated and downregulated proteins were labeled.

3.3. Transcriptome Analysis

We profiled the transcriptomic changes in the vocal fold (VF) tissues isolated from each animal 28 days post-injection to evaluate the molecular effects of different treatment conditions following injury. To provide biological context, Figure 4 summarizes the complexity of fibrotic signaling pathways and key transcriptional targets relevant to ECM-mediated modulation of fibrosis.

Figure 4. Pathways involved in fibrosis, highlighting both canonical and non-canonical processes.

This includes the canonical TGF-β/SMAD signaling pathway and non-canonical pathways such as PI3K/AKT, MAPK, and Wnt/β-catenin signaling. The figure illustrates how integrin conformational changes induced by mechanical stress in the ECM facilitate fibroblast mechanotransduction and stress fiber formation. The complex signaling network subsequently influences gene transcription by regulating mechanosensitive co-transcription factors, such as myocardin-related transcription factor A (MRTFA), further driving pro-fibrotic genes.

Global gene expression changes relative to the ‘no injury’ experimental group are shown in Figure 5A. Hierarchical clustering of these global expression profiles revealed that all treatment groups diverged from the injury-only condition, with the type I collagen (Col1) and porcine-derived VFLP (VFLP (au)) treatments clustering closely together. Meanwhile, the manually decellularized VFLP (VFLP (man)) formed a distinct cluster. Pairwise correlation analysis of their global expression changes between treatment groups is also visualized in Figure 5B, illustrating broader patterns of transcriptional modulation. Compared to the Col1 control, gene expression demonstrated a proportional correlation along the trend line. This response was consistent across all treatment groups, VFLP (au) and VFLP (man). A less steep trend line was observed when comparing the injury group with VFLP (au) treatment. To further evaluate how these groups cluster with each other, a Principal Component Analysis (PCA) was performed. The PCA (Figure 5C) further supports these findings, with sample centroids and individual replicates reflecting similar relative positioning observed in the heatmap (Figure 5A).

Figure 5. Transcriptomic characterization of leporin-treated vocal folds relative to healthy tissue.

(A) Heatmap of average log fold change (logFC) values across all sequenced genes, with hierarchical clustering performed on group-averaged expression profiles. (B) Pairwise correlation of global gene expression between groups, shown as scatter plots with marginal distributions and corresponding R² values. (C) Principal component analysis (PCA) of the full transcriptomic dataset: small markers represent individual vocal fold samples, and large markers indicate group centroids.

Despite intra-group variability, Col1 and VFLP (au) samples clustered proximally in multidimensional space. Meanwhile, VFLP (man) samples separated more distinctly between Col1/VFLP (au) and Injury-only groups. These global transcriptional profile differences suggest that the products derived from altered modes of ECM processing may influence the healing of vocal cords in vivo and may impact the therapeutic efficacy of ECM.

To further investigate the effect of our ECM on the fibrotic response, we compared the expression profiles of the samples from injury and treated animal groups against samples without treatment (‘Injury only’) and performed targeted gene analysis by filtering our transcriptomic dataset using a list of 2490 fibrosis-associated genes curated by the knowledgebase of the Ingenuity Pathway Analysis (IPA) (Supplementary Table S1.). Given the limited understanding of the key molecular drivers of vocal fold fibrosis, we retained the complete list of fibrosis-annotated genes for analysis. Although TGF-β1 transcript levels were not significantly affected, its receptor signaling activity was evaluated by analyzing both canonical (e.g., SMAD3-related) and non-canonical (e.g., MAPK, PI3K/AKT) downstream targets, which showed differential expression consistent with TGF-β1-mediated responses. This is consistent with the role of TGF-β1 as part of a latent protein complex, where post-transcriptional activation governs its biological effects.^10,55,56 Volcano plots were generated for each treatment comparison (Figure 6A), displaying the global transcriptomic changes and highlighting fibrosis-associated genes in green. The position of ACTA2, a marker of myofibroblast activation, is indicated for reference.

Figure 6.

Transcriptional changes of fibrosis-related genes induced by hydrogel treatments in the rabbit vocal cord injury model. A) Volcano plot displaying the log2 fold change (x-axis) vs. log10 scale raw p-values (y-axis) of each gene in each treatment versus injury comparisons. Green dots indicate fibrosis-related genes curated by the IPA knowledgebase. Differentially expressed genes were defined as fold change > 2 and raw p-value < 0.05. B) Venn diagram of differentially expressed fibrosis-related genes identified from each treatment. C) Expression changes of the ACTA2 gene, a canonical biomarker of myofibroblast activation. D) Heat map illustrating the overall expression changes of all detected fibrosis-related genes (2,490 genes in total, ordered by their overall log2 fold changes).

Overall, more genes were identified as downregulated compared to the injured group before VFLP (man) treatment, as depicted in Panel A of Figure 6. Notably, among differentially expressed genes (Log2 fold change > 1 & raw p-value < 0.05), the VFLP (man) group exhibited a more significant number of downregulated fibrosis-related genes (129 down vs. 7 up-regulated genes) compared to the other two treatment groups (48 down vs. 10 up-regulated genes for Col1, and 56 down vs. 14 up-regulated genes for VFLP (au)). A Venn diagram (Figure 6B) showing the overlap of differentially regulated fibrosis-associated genes further illustrates the unique transcriptional profile of VFLP (man) in modulating fibrosis-related genes. Consistent with our previously published in vitro data, Figure 6C shows ACTA2, a widely recognized marker of myofibroblasts. We focused on this marker not because it is unique—there are others—but because it was the targeted reporter used in prior in vitro studies. This marker exhibits downregulated expression levels in the VFLP treatment groups, supporting the antifibrotic potential of VFLP-ECM biomaterials. This trend was further supported by the hierarchical clustering of the expression changes of all fibrosis-annotated genes (Figure 6D), where VFLP (man) samples clustered distinctly from all other groups. These findings suggest VFLP (man) exerts a more pronounced inhibitory effect on the fibrotic response during VF healing.

3.4. Mechanical Analysis of the Folds

Mechanical testing revealed significant differences in structural properties across experimental conditions, as summarized in Table 2.

Table 2.

Descriptive statistics of normal force, structural stiffness, and displacement by treatment groups

	Normal Force at 0.3 mm (mN)		Structural Stiffness at 0.3mm (mN/mm)		Displacement at 1.96N (mm)
Groups	Mean	SD	Mean	SD	Mean	SD
No Injury	1.84	0.47	5.34	1.47	0.30	0.07
Injury alone	2.60	0.32	8.39	0.95	0.23	0.03
Col1	2.77	1.11	6.59	2.50	0.27	0.06
VFLP (au)	2.09	0.66	6.33	2.05	0.30	0.09
VFLP (man)	1.94	0.42	5.77	1.22	0.29	0.08

To compare groups, a one-way ANOVA followed by LSD posthoc tests was carried out. Figures 7 A and B depict the distinct differences in structural properties across the experimental conditions.

Figure 7. Biomechanical properties of vocal folds following hydrogel treatment.

(A) Structural stiffness measurements derived from force-displacement testing. (B) Normal force measured at a 0.3 mm displacement. (C) Correlation between ACTA2 expression (logFC) and structural stiffness across samples. Asterisks indicate statistical significance: *p < 0.05, **p < 0.01. (D) Representative heatmaps illustrating tissue stiffness projected onto the tested vocal fold (VF) area for all experimental groups. Warmer colors indicate regions of higher stiffness, whereas colder colors correspond to lower stiffness.

At a displacement of 0.3 mm, all injured and treatment groups displayed elevated normal force measurements relative to the uninjured control. The injury-alone group registered a normal force of 2.60 ± 0.32 mN, significantly higher than the no-injury group (1.84 ± 0.47 mN, p = .008) and the VFLP (man) group (1.94 ± 0.42 mN, p = .0201). This increase in normal force in the injury-alone group suggests a stiffened tissue response post-injury. Although still higher than baseline, the VFLP (man) group exhibited a significant reduction in normal force compared to the injury alone, nearing the values observed in the no-injury group.

Similarly, the injury-alone group exhibited the highest stiffness, measuring 8.39 ± 0.95 mN/mm, which was significantly greater than that in the no-injury group (5.34 ± 1.47 mN/mm, p = .001), the VFLP (man) group (5.77 ± 1.22 mN/mm, p = 0.004), and the VFLP (au) group (6.33 ± 2.05 mN/mm, p = 0.022). Heatmap visualization of stiffness (Figure 7D) further corroborated these findings. The injury-alone group displayed significantly increased stiffness, indicative of fibrotic changes and decreased tissue compliance post-injury. Conversely, the VFLP (man) group showed a more uniform stiffness distribution, with reduced stiffness (blue) areas, indicating restoring tissue properties closer to baseline. While the VFLP (au) group exhibited lower stiffness than the injury-alone group, it presented a more varied profile, with some areas still demonstrating elevated stiffness (red).

Displacement at 1.96 N was lowest in the injury alone group (0.23 ± 0.03 mm), compared to no injury (0.30 ± 0.07 mm), indicating increased tissue stiffness following injury. Col1 treatment (0.27 ± 0.06 mm), VFLP (au) (0.30 ± 0.09 mm), and VFLP (man) (0.29 ± 0.08 mm) treatments improved displacement compared to the injury alone group. VFLP (au) and VFLP (man) values were closer to the baseline observed in the no-injury group (0.30 ± 0.07 mm).

Treatment with Col1, VFLP (au), and VFLP (man) helped restore tissue mechanical properties, with VFLP (man) showing the most promising recovery. While all treatments reduced stiffness compared to the injury-alone group, VFLP (man) showed the most significant improvement, particularly in normal force and stiffness, while also showing higher compliance.

4.0. Discussion

Vocal fold (VF) injury represents a significant clinical challenge due to its detrimental impact on vocal function and the persistent difficulty in achieving effective, long-term treatment. The lamina propria of the vocal folds is composed of a specialized extracellular matrix (ECM) that imparts the viscoelastic properties necessary for phonation. Following injury, fibrotic remodeling of the ECM and scar tissue formation disrupt the layered architecture and biomechanical function of the superficial lamina propria (SLP), leading to increased stiffness, reduced pliability, and altered mucosal wave propagation. These changes manifest clinically as dysphonia and may result in permanent voice impairment. Current treatment options for VF scarring and related voice disorders are limited in efficacy, with most strategies focusing on symptom management rather than restoration of native tissue architecture or biomechanical function.

Injectable biomaterials are widely used for vocal fold (VF) repair due to their compatibility with in-office, image-guided delivery and ability to conform to laryngeal anatomy. Current clinical injectables vary in composition and duration, ranging from temporary bulking agents, such as gelatin- and collagen-based products, to longer-lasting materials, including hyaluronic acid (HA), autologous fat, and synthetic fillers like calcium hydroxylapatite and polydimethylsiloxane. These materials are injected into the vocal fold muscle to increase the overall mass of the vocal folds or adjasent tissues. These approaches do not attempt to replace or alter the lamina propria. None of these agents address the underlying fibrotic microenvironment or promote true tissue regeneration within the lamina propria. This lack of current treatment options highlights the need for next-generation biomaterials to provide structural support, actively modulate fibrosis, and facilitate functional VF healing.

Preparing dECM biomaterials hinges on effective decellularization protocols that remove cellular remnants while preserving the three-dimensional integrity of the ECM. Approaches range from simpler techniques (e.g., freeze-drying with sonication) to more robust combinations of detergents, enzymes, and chemicals. Sonication can substantially improve cell-removal efficiency alongside freeze-drying, but it risks weakening the scaffold structure. Detergent-enzymatic treatments (commonly employing Triton X-100 or TrypLE Express) often achieve near-complete decellularization; however, detergent residues may introduce cytotoxicity or disrupt collagen fibers. Our protocol using sodium deoxycholate plus DNase has demonstrated effective removal of dsDNA within hours, though peracetic acid or prolonged enzyme exposures can degrade key ECM proteins. An optimal strategy must strike a balance between thorough decellularization, minimal ECM damage, reproducibility, and scalability. Notably, while SIS and UBM can be decellularized with arguably consistent results using peracetic acid, this method struggles to adequately penetrate VF tissues, reflecting how tissue architecture can influence protocol success. The automated decell version sought to meet these criteria. Once decellularized, scaffolds must be evaluated for their final composition and bioactivity.

Two proteomic studies illustrate these distinctions in the final acellular ECM. Mika Brown et al. compared porcine VF dECM, obtained via an optimized protocol (Protocol 3), to native VF tissue and SIS.^6,57 They found that VF dECM shared about 63% of its identified proteins with SIS and retained many unique matrisome components linked to cell–matrix adhesion. However, SIS was rich in proteins associated with wound healing and innate immunity, implying it could be beneficial in promoting regenerative responses. Notably, Brown et al. reported that VF dECM had the highest total collagen content by dry mass, although soluble collagen was reduced compared to native VF. The authors concluded that while VF-specific ECM more closely mirrored the native tissue, SIS emerged as a scalable alternative because it effectively supported cell viability, stimulated elastin, and hyaluronan production, and showed lower cross-sample variability. Alternatively, our previous study detailed proteomic characterization of decellularized porcine VF lamina propria (VFLP-ECM) compared to UBM.⁵² Our analysis identified nearly 2000 proteins in VFLP-ECM, many of which were shared with UBM, but some were present at significantly different abundances. Notably, proteins potentially modulating TGF-β1 signaling, such as latent TGF-β1 binding protein (LTBP4), EMILINs, decorin, and dermatopontin, were abundant in VFLP-ECM. These molecules are crucial for regulating fibrotic processes, suggesting that VF-specific scaffolds may inhibit myofibroblast activation and subsequent scar formation.

Although in our study, collagen injections showed modest improvements in antifibrotic gene expression; these effects were significantly less pronounced than those induced by VFLP (man). This suggests that the biological activity of VFLP (man) is driven by its unique composition, potentially enriched with signaling proteins not present in purified collagen. Proteomic analysis supported this, showing a substantially higher number of uniquely identified proteins in VFLP (man) (431) compared to VFLP (au) (45). These included vitronectin (VTN) and RHOA, proteins closely linked to TGF-β1 signaling and fibrotic remodeling (Figure 3 A). VTN peptides, which align with the pepsin digestion to produce the hydrogel, are reported to modulate in a competitive TGF-β1 stimulation by integrin interaction.⁵⁸ Their detection in VFLP (man) suggests that manual decellularization better preserves bioactive components that modulate the fibroblast phenotype and ECM remodeling. These findings are consistent with prior in vitro work by Mora-Navarro et al. (2020), which demonstrated that VFLP-derived ECM hydrogels reduced myofibroblast activation and fibrotic gene expression in TGF-β1-stimulated vocal fold fibroblasts.

Gene expression analysis of the VFs after treatment aligned with the proteomic data: VFLP (man) treatment resulted in 105 upregulated proteins and only one significant downregulated protein compared to VFLP (au), reflecting enhanced protein recovery rather than selective enrichment (Figure 3B volcano plot -proteomics-). Many of these upregulated proteins are involved in ECM remodeling and the regulation of fibrosis. While both ECM preparations shared 1,748 proteins, highlighting the preservation of the core ECM proteome, the distinct protein subset in VFLP (man) appears to confer enhanced biological activity. These compositional differences could be reflected functionally in vivo. Transcriptomic profiling showed that VFLP (man)-treated tissues exhibited a more substantial downregulation of fibrosis-associated genes than collagen and VFLP (au), indicating a more pronounced suppression of fibrotic signaling pathways.

Mechanical characterization further supported these findings. Vocal folds treated with VFLP (man) recovered viscoelastic properties similar to those of uninjured controls, whereas those treated with collagen or VFLP (au) remained stiffer—a hallmark of unresolved fibrosis. Since matrix stiffness can perpetuate fibrotic signaling through mechanotransduction pathways, this mechanical restoration likely contributes to the antifibrotic environment established by VFLP (man).

Recent therapeutic approaches have targeted these molecular mechanisms through cell- and biomaterial-based approaches. Autologous fibroblast injections have shown promise in clinical studies of VF scarring, resulting in partial restoration of vibratory function and improved phonation. Similarly, mesenchymal stem cells (MSCs) exhibit antifibrotic activity by inhibiting TGF-β1 signaling and downregulating the expression of myofibroblast markers. In addition, pharmacologic modulation of mechanosensitive pathways, such as the inhibition of YAP1⁶³ or the activation of NR4A1^64,65, has demonstrated potential in preclinical models of fibrosis. However, the clinical application of cell-based therapies remains limited by concerns over delivery, viability, and scalability.

To gain further mechanistic insight regarding the VFLP-ECM, we performed an Ingenuity Pathway Analysis (IPA) on the transcriptomic data. IPA shows predicted inhibition of key transcriptional regulators of fibrosis, including SMAD3, ERK1/2, YAP1, and myocardin-related transcription factor A (MRTFA)—all central mediators of TGF-β1 signaling and myofibroblast activation (See Figure S1).¹² The suppression of SMAD3 and MRTFA is particularly relevant, as both are directly involved in the expression of ACTA2 (α-SMA), COL1A1, and other pro-fibrotic genes. In parallel, IPA predicted activation of SMAD7, an inhibitory SMAD that blocks TGF-β signaling, suggesting a shift toward an antifibrotic regulatory state.

These findings are consistent with recent data on MRTF-A/SRF signaling in vocal fold fibrosis. A recent study demonstrated that MRTF-A and its co-factor SRF are upregulated in TGF-β1-stimulated fibroblasts and injured vocal folds in vivo. Pharmacologic inhibition of MRTF-A reduced α-SMA expression, collagen deposition, and epithelial thickening—hallmarks of fibrosis. Our observation that MRTFA is downregulated in VFLP (man)-treated tissues aligns with this work and provides further evidence that the antifibrotic effects of VFLP (man) may be mediated through suppression of mechanosensitive transcriptional pathways. Manual decellularization appears to preserve functionally important proteins that promote regulatory pathways—notably TGF-β1 and MRTFA-associated signaling—resulting in improved tissue repair and mechanical recovery. These findings underscore the importance of ECM processing methods in determining the biological activity of injectable biomaterials and highlight the therapeutic potential of VFLP-ECM in the treatment of vocal fold fibrosis.

Despite these promising results, several limitations should be acknowledged. This study assessed outcomes at a single time point (28 days post-injection), which may not fully capture the temporal dynamics of healing and fibrosis resolution. Additionally, only a single injection was administered. Multiple or higher-dose injections could yield more robust or sustained therapeutic effects. Future studies are needed to optimize dosing regimens, evaluate long-term outcomes, and further elucidate the mechanisms of action. Nevertheless, these findings represent an important first step in demonstrating the potential of tissue-specific ECM hydrogels as antifibrotic therapies in the vocal folds.

By integrating proteomic, transcriptomic, and mechanical analyses, this study provides a comprehensive evaluation of VFLP-ECM hydrogel therapy. It highlights the critical impact of the ECM preparation method on therapeutic efficacy. VFLP (man), in particular, emerges as a biologically active material capable of modulating fibrosis-associated pathways and promoting functional tissue repair, laying the groundwork for future translational research in regenerative laryngology.

5.0. Conclusion

This study demonstrates that VFLP-ECM hydrogels, particularly those produced through manual decellularization, exert potent antifibrotic effects in vivo by modulating gene expression, tissue mechanics, and fibrotic signaling pathways. VFLP (man) treatment was associated with transcriptomic profiles indicative of reduced myofibroblast activation, improved viscoelastic properties, and enrichment of bioactive proteins involved in ECM remodeling and TGF-β1 regulation. These outcomes were not replicated by automated ECM preparations or collagen, underscoring the impact of the decellularization strategy on functional efficacy. VFLP (man) appeared to attenuate fibrosis through coordinated suppression of SMAD3 and MRTFA—key transcriptional effectors of TGF-β1 and mechanotransductive signaling—while promoting antifibrotic mediators such as SMAD7. These findings provide mechanistic insight into how tissue-specific ECM can disrupt the feedback loops driving fibrosis in the VF, supporting its further development as a regenerative therapy for vocal fold scarring. Future studies should explore longitudinal outcomes, dosing strategies, and clinical translation to optimize the therapeutic application of VFLP-derived biomaterials.

HIGHLIGHTS:

Tissue-specific VFLP-ECM hydrogel restores vocal fold mechanics in vivo 28 days after injury.

VFLP- ECM hydrogel shows improved vocal fold repair compared to type I collagen.

Proteomic analysis indicates that matrisome proteins like vitronectin may drive the in vivo response.

Transcriptomic data reveal that ECM hydrogel inhibits fibrosis through non-canonical mechanisms.

Data Availability

Data will be made available on request.