Unraveling the Relevance of Tissue-Specific Decellularized Extracellular Matrix Hydrogels for Vocal Fold Regenerative Biomaterials: A Comprehensive Proteomic and In Vitro Study

Abstract

Decellularized extracellular matrix (dECM) is a promising material for tissue engineering applications. Tissue-specific dECM is often seen as a favorable material that recapitulates a native-like microenvironment for cellular remodeling. However, the minute quantity of dECM derivable from small organs like the vocal fold (VF) hampers manufacturing scalability. Small intestinal submucosa (SIS), a commercial product with proven regenerative capacity, may be a viable option for VF applications. This study aims to compare dECM hydrogels derived from SIS or VF tissue with respect to protein content and functionality using mass spectrometry-based proteomics and in vitro studies.

Proteomic analysis reveals that VF and SIS dECM share 75% of core matrisome proteins. Although VF dECM proteins have greater overlap with native VF, SIS dECM shows less cross-sample variability. Following decellularization, significant reductions of soluble collagen (61%), elastin (81%), and hyaluronan (44%) are noted in VF dECM. SIS dECM contains comparable elastin and hyaluronan but 67% greater soluble collagen than VF dECM. Cells deposit more neo-collagen on SIS than VF-dECM hydrogels, whereas neo-elastin (~50 μg/scaffold) and neo-hyaluronan (~ 6 μg/scaffold) are comparable between the two hydrogels.

Overall, SIS dECM possesses reasonably similar proteomic profile and regenerative capacity to VF dECM. SIS dECM is considered a promising alternative for dECM-derived biomaterials for VF regeneration.

Graphical Abstract

Decellularized Extracellular Matrix (dECM) derived from the same source as the target tissue is thought to augment regeneration, but limited dECM is obtained from small organs like vocal folds (VF). VF dECM and a commercial alternative, Small Intestinal Submucosa (SIS) possess similar proteomic content. Additionally, hydrogels from SIS and VF dECM stimulate deposition of the ECM components comparable levels.

1. Introduction

Decellularized extracellular matrix (dECM) has become a major naturally derived biomaterial in tissue engineering and regeneration. Depending on the application, dECM-based biomaterials are manufactured in forms of injectables, bioinks, and electrospun scaffolds.^[1–3] With dECM materials, cells and immunogenic molecules are largely removed that minimize adverse host response. Meanwhile, structural, core-matrisome proteins [e.g., collagen, proteoglycans, glycoproteins], macromolecules [e.g., glycosaminoglycans (GAGs) etc.] and matrisome-associated proteins [e.g., ECM-affiliated proteins, ECM-Regulators] are mostly preserved in the dECM materials.^[4,5] These preserved ECM structural and signaling molecules are essential to provide a native-like, tissue-specific microenvironment for local cell functions in tissue repair.^[6,7]

Mammalian vocal folds (VF) have unique ECM-structural and mucosa-immunological profiles that likely benefits from a tissue-specific regenerative approach like dECM.^[8–10] Mass spectrometry has aided in profiling the proteome of organs of interest through measuring the mass to charge ratio (m/z) of fragmented proteins in the sample. For complex ECM profiles like VF, proteomic analyses are imperative to generate a deep understanding of the interaction between biological systems and biomaterials for bioengineering applications.^[11,12]

In VF tissue engineering, mass spectrometry was used to identify and compare proteins in native and decellularized human VF mucosa to determine which extent dECM preserved a niche for cellular attachment and infiltration.^[13] The 219 proteins that remained in dECM included those involved in wound healing activities such as ECM production and organization. Although traditional histology did not identify antigenic cellular proteins in the VF dECM, mass spectrometry found residual antigenic proteins including Prxs1, 2 and 4, CuZnSOD and GPx3. These results indicated a possible source of varied immune responses to dECM biomaterials in vivo.

Proteomic analysis was also applied to evaluate the variability of VF dECM samples during manufacturing. Conventional decellularization protocols at the bench scale produce highly varied batches of dECM due to manual reagent changes and biological variation between animals.^[14] Customized bioreactors have been designed to reduce variability between dECM hydrogels through automation and aid in scale-up manufacturing. Mass spectrometry was implemented to compare conventionally produced VF dECM with dECM produced using the bioreactor.^[14] In this study, 2430 unique proteins were identified, including 84 core matrisome proteins. Bioreactor-VF dECM contained a greater overall number of protein types by 10-fold compared to the conventional dECM protocol.

While the general biomimetic benefit of dECM is well-acknowledged, manufacturing VF-specific dECM biomaterials at the commercial scale is almost impossible because only very minute amounts of matrix material can be derived from the small human VF with sizes ranging between 240 and 2400 mm³.^[15] To address this scalability challenge, large organs such as Small Intestinal Submucosa (SIS) or Urinary Bladder Matrix (UBM) have been used for extraction of dECM in developing VF regenerative biomaterials.^[8,16] The larger size of porcine SIS can generate a larger volume of material from a single batch for testing and gel production. Several dECM-derived scaffolds are already clinically approved including Cook Biotech’s Biodesign^® from Small Intestinal Submucosa (SIS) and ACELL’s Matristem UBM^®.^[17,18]

To date, only limited studies provided data regarding the importance of tissue-specific dECM for VF tissue regenerative biomaterials. In one in vitro study, mass spectrometry was used to compare the proteomes of dECM extracted from porcine VF and UBM tissues.^[19] Compared to that of UBM, dECM from VF tissue contained a higher concentration of the TGF-β1 sequestering protein, LTBP4. The presence of this molecule may help reduce TGF-β1-mediated fibrotic activity. When Human Vocal Fold Fibroblasts (HVFF) were cultured on VF dECM, UBM, or collagen hydrogels, fibrosis-associated genes Col1A1 and ACTA2 were found downregulated on VF dECM hydrogels. However, neo-ECM production by HVFF as functions of dECM tissue sources were not evaluated and therefore the overall regenerative efficacy could not be confirmed. In an animal study, MSCs encapsulated in SIS dECM hydrogels and MSCs alone were injected to scarred rabbit VF.^[20] Compared to MSC-only controls, more neo-hyaluronan and neo-collagen was found in scarred VF at Week 8 post-injection. Improved glottal function was also identified in the SIS-MSC group, indicating the SIS provided a niche for MSCs to regenerate functional tissue. As a VF dECM group was not included in this study, whether SIS dECM possessed comparable regenerative capacity to VF dECM remained to be confirmed.

The primary goal of this study was to compare the protein components, biocompatibility and functional features between VF and SIS-dECM biomaterials using a combination of mass spectrometry-based proteomics and in vitro studies. Molecular components and pathways related to ECM, angiogenesis and immune systems were the focus of these analyses. VF dECM was hypothesized to contain target tissue-specific bioactive molecules, while SIS dECM was hypothesized to fulfill the fabrication challenge of supply and accessibility.

2. Results

2.1. Efficacy of Decellularization Protocols for Porcine Vocal Folds

We first tested seven dECM protocols of their efficacy in removing cells and their residual DNA from porcine VF. (Figure 1A). In selecting a protocol, we used a literature value of < 50 ng DNA/mg tissue.^[21] While higher levels of DNA have been suggested to be acceptable, standard levels for DNA removal in dECM products have not yet been designated.^[1] 50 ng is a widely adopted and stringent target for decellularization protocols.^[18,21–25] We sought to achieve a comparable level of decellularization with reduced heat exposure time and consequent protein degradation by minimizing nuclease incubation time. We also sought to increase efficacy of gelation by using isopropanol to delipidize the tissue.

Figure 1.

A. Decellularization Process of Porcine Vocal Folds. VF were dissected from a porcine larynx, then minced to increase surface area. Cell lysis was performed for 24 h in 3M sodium chloride at 4°C. Incubation was performed with nucleases at 37 °C for 6 hours breaks down DNA, Delipidization in Isopropanol for 24 h at 4 °C, followed by additional nuclease incubation cycles to maximize DNA removal. A 6-hour incubation cycle is repeated twice in the selected protocol. Homogenization by cryomilling produces a white powder. B. Nanograms Residual DNA for Each Decellularization Protocol. Incubation cycles lasted either 6 or 24 h. **** p < 0.0001 C. Formation of VF and SIS dECM Hydrogels. 20 mg/ML dECM Powder is suspended in 2 mg/mL pepsin in 0.05 M HCl for 48 h while stirring at a 30 rpm for 48 h. The solution is neutralized with 1 M NaOH and 10X PBS, diluted to 16 mg/mL, and incubated at 37 °C to induce gelation. D. Storage Modulus of 16 mg/mL SIS dECM and VF dECM hydrogels, and 2 mg/mL Collagen I hydrogel controls over 2 h.

Protocol 3, which used 18 total hours of nuclease incubation in 6 h cycles, was sufficient to reduce the residual DNA concentration to <50 ng DNA/mg tissue (Figure 1B). The residual DNA in Protocol 3 was significantly less than Protocol 2 where two 6 h cycles were used, and Protocol 5 where a single 24 h cycle was used (43.4 ± 5.9 ng DNA/mg tissue v/s 127 ± 35 ng DNA/mg tissue and 118 ± 35 ng DNA/mg tissue respectively, p < 0.0001). While protocols 4, 6, and 7 also reduced DNA content below the 50-ng/mg target, Protocol 3 was selected for subsequent experiments as the dECM protocol that met the 50 ng/mg DNA target with the shortest incubation time.

The commercially produced dECM from Cook Biotech that we used for comparison to our VF dECM throughout our research is produced from porcine SIS. The DNA content of their powdered SIS dECM is not publicly accessible. However, Cook Biotech’s Oasis^®, which is produced using the same peracetic acid and ethanol-based methods, has been reported to contain 0.42 ± 0.01 ng DNA/mg dry tissue.^[26]

The peracetic acid and ethanol method alone, while highly effective for decellularization of SIS, does not achieve sufficient penetration for effective decellularization of VF. To decellularize VF, peracetic acid has been used in combination with sodium deoxycholate and nucleases.^[27] However, peracetic acid can damage collagen structure with exposure of longer than 1–2 hours, while sodium deoxycholate also damages ECM components.^[4] In our pilot work, a protocol involving exposure to sodium deoxycholate for 2 h, nucleases for 48 h, and peracetic acid for 1 h was required to achieve comparable levels of decellularization to our Protocol 3.^[28] Hydrogel gelation was noted to be inconsistent, which hampered the reproducibility of experiments and potential for scaleup in manufacturing.

2.2. Hydrogel Formation and Characterization

To form hydrogels, 20 mg/mL VF dECM produced using Protocol 3 was solubilized in pepsin in hydrochloric acid. Once neutralized, diluted to a concentration of 16 mg/mL, and incubated at 37 °C, hydrogels formed within 30 minutes. Both SIS dECM and VF dECM hydrogels could be manipulated with a spatula without losing integrity. Time sweeps were performed to evaluate gelation kinetics and viscoelasticity. 2 mg/mL collagen I hydrogels were used as controls. All hydrogels possessed storage moduli (G’) greater than loss moduli (G”) throughout the experiment, and reached a plateau in under 30 minutes. This confirmed our observations of the time to gelation. The plateau, or final G’ for SIS dECM (178 ± 47 Pa) and VF dECM (144 ± 59 Pa) hydrogels were comparable and not significantly greater than that of the collagen hydrogels (25.53 ± 13.75 Pa). No significant difference was identified between groups (p = 0.21) (Figure 1D).

2.3. Protein Identification, Gene Ontology Enrichment, and Analysis

We then performed mass spectrometry-based proteomics on the three sample types: native VF tissue, VF dECM produced using Protocol 3, and Cook Biotech’s SIS dECM. Using Venn Diagrams, we determined the number of proteins in common and unique to each tissue type (Figure 2A). A total of 620 proteins were identified from the Native VF samples. Of these proteins, 63% (391/620) were shared between VF dECM and SIS dECM. A further 29% (182/620) were shared with only VF dECM, and 6% (36/620) with only SIS dECM. 2% of proteins (11/620) were unique to Native VF, 9% of proteins (36/411) to SIS dECM, and 3% of proteins (17/614) to VF dECM. A total of 24 proteins were shared by the two dECM types (6% of SIS dECM v/s 4% of VF dECM), but not identified in native VF. The decellularization process eliminates abundant proteins (i.e., albumin) that in native tissues serve to “block” the signal of less abundant proteins. This dynamic range issue results in hidden proteins in the native samples.

Figure 2.

Comparison of identified proteins in Native VF, VF dECM, and SIS dECM A. Total identified proteins produced using Functional Enrichment Analysis Tool (FunRICH). B. Principal component analysis (PCA) of the proteomics data for each tissue sample (n = 6 per tissue type). The PCA graph depicts in group variation for Principal component 1 (PC1) and Principal component 2 (PC2). PC1 accounts for 40.59% of variation between the tissue samples and PC2 accounts for 24.15%. C. Protein loadings plot. The protein loadings plot describes the degree to which specific proteins contribute to PC1 and PC2. Circles indicate proteins with relatively high loading scores discussed in text.

Principle component analysis enables a more detailed look at variability between samples, as well as sources of variability. From principal component analysis (Figure 2B), 40.59% and 24.15% of variability were attributed to PC1 and PC2 respectively. To gain further insights into how content of specific proteins impacts observed variability between samples, a protein loadings plot was generated. The proteins that contributed most strongly to PC1 and PC2 was based on the magnitude of loading scores (−1 to 1) from the origin. For instance, ALB and MYH2 had strong contributions and positive correlation with PC1 and PC2, respectively. While ALB had a log10 abundance greater than 2 in all sample types, it was most abundant in native VF, leading to the high loading score. MYH2 had the highest log10 abundance of all proteins in native VF (3.37) and VF dECM (3.22), but was much lower in SIS dECM (0.81). This difference caused its high PC2 loading score. Likewise, the loading score of BPIFB1, which also had a high log10 abundance in native VF and VF dECM but was absent in SIS dECM, showed a strong negative correlation with PC1. In contrast, COL6A3 was negatively correlated with both PC1 and PC2, and had high but differing log10 abundance in native VF (2.71), VF dECM (2.97), and SIS dECM (2.9). The protein loadings plot allows us to tie our variability data more directly from PCA to notable differences between sample types

Samples of native VF and VF dECM demonstrated a high degree of similarity based on the first principal component, PC1, with the majority of variability introduced between the VF tissues and SIS. However, PC2 indicated variation between native VF and VF dECM. Based on sample distribution, the highest degree of similarity was noted in the SIS dECM samples, which originated from two bulk lots produced using standardization procedures. In contrast, high in-group sample variability was observed within Native VF and VF dECM. The greater in-group similarity identified in SIS dECM samples is beneficial for product consistency in scale-up for manufacturing. Overall, these analyses showed that while our VF dECM is more similar to Native VF, the majority of proteins are also shared with SIS dECM decellularized with peracetic acid and ethanol.

2.3.1. Comparison of Matrisome and Matrisome-Associated Proteins by Tissue Type

Core matrisome proteins represent the main structural and mechanical components of the ECM and include collagens, glycoproteins, and proteoglycans.^[5] Matrisome-associated proteins include ECM-affiliated proteins, ECM Regulators, and secreted factors, which have roles in directing cell behavior toward wound healing. Venn Diagrams were used to identify which specific core matrisome and matrisome-associated proteins overlapped or were distinctively present in native VF, VF dECM from Protocol 3 and Cook Biotech’s SIS dECM (Figure 3A). Respective abundance curves were also created to rank the top 100 collagens and VEGF-signaling associated proteins and provide a representation of the shifts in protein abundance between sample types (Figure 4). The purpose of these analyses was to provide a deeper comparison of core matrisome and matrisome-associated proteins between sample types and to evaluate their roles in regeneration.

Figure 3.

Venn diagram of identified A. core matrisome and B. matrisome-associated proteins. Selected pathways (e.g., elastic fiber formation, angiogenesis, etc.) in which these proteins play a role are highlighted.

Figure 4.

Rank abundance curve highlighting identified collagens and proteins involved in Reactome’s VEGF signaling pathway. The green-shaded area represents the top 100 proteins. Only collagens and VEGF-related proteins are highlighted in this figure, not all proteins in the samples.

With respect to the core matrisome proteins, a total of 13 collagen types were common in all sample types. Seven of these 13 collagens were in the top 100 protein abundance in native VF, 8 in VF dECM, and 10 in SIS dECM. Specifically, COL1A1, COL1A2, COL5A1 and COLA2 are key to the formation of neo-collagen fibers.^[5] The COL6 family and COL14A1 are involved in fibrillogenesis, or self-assembly of the collagen network.^[29,30] COL3A1, a collagen essential to the elasticity of healthy VF tissue that is replaced by Collagen 1 in scar tissue, was also identified in all sample types.^[31,32] A total of 18 glycoproteins were shared across the three sample types. Several of these proteins (EMILIN1, FBLN1, FBLN5, FBN1) contribute to elastic fiber formation, an essential process repaired VF to regain elasticity.^[33–36] Laminins, involved in basement membrane formation, integrin binding, and cell signaling were identified in all sample types (LAMA4, LAMB1, LAMB2, LAMC1), native VF only (LAMA3), or VF dECM only (LAMB3, LAMC2).^[5,36] Seven proteoglycans were identified across all sample types. Among them, ASPN and LUM are involved in collagen binding,^[37,38] whereas BGN, DCN, and HSPG2 are involved in angiogenesis and ECM binding.^[39–42] HAPLN1 and VCAN were found only in native VF and VF dECM, which are involved in the regulation of cellular adhesion and hyaluronan binding.^[43,44] A sample-by-sample comparison of core matrisome protein abundnance using hierarchical clustering analysis is available in Figure S1.

Among the matrisome-associated proteins, a total of 9 ECM-affiliated proteins were noted across all sample types (Figure 3B). Annexins (ANXA1, ANXA2, ANXA5, ANXA6, ANXA8, ANXA11) are involved in angiogenesis, adoption of cellular morphology, and immune modulation.^[45–47] LMAN1, which regulates blood clot formation, was shared only by native VF and VF dECM. LGALS1, a protein related to cellular adhesion and signaling, is unique to VF dECM.^[48,49] In addition, five proteins were unique to SIS dECM that are involved in angiogenesis (GREM1) or cellular adhesion, migration, and signaling (C1QA, GPC6, LGALS2, LGALS4).^[50–53] Ten ECM-regulator proteins were shared by all three sample types that have roles in regulating angiogenesis and blood clots such as fibrin clot formation, heparin binding, and blood coagulation.^[54,55] Other roles include growth factor binding (A2M, F2).^[56,57] Of the ECM-regulators unique to native VF and SIS dECM, AGT, plays a role in angiogenesis.^[58] Other secreted growth factors, cytokines, and chemokines are usually difficult to detect by mass spectrometry.^[59] As expected, only one secreted factor was detected in all tissue types (S100A1), and one by native VF and VF dECM (S100A14).

Proteins involved in angiogenesis, an important process in VF repair, were identified using a rank abundance curve for proteins in the VEGF-signaling pathway. The three most abundant VEGF signaling-associated proteins in all three sample types were HSP90AA1, HSPB1, and CRK. These proteins regulate VEGF binding, endothelial cell outgrowth, and endothelial cell migration, respectively.^[60–62] HSP90AA1 and CRK were ranked in the top 100 in SIS dECM. On the other hand, HSPB1 (76^th) and HSP90AA1 (100^th) were in the top 100 of VF dECM. These proteins were all ranked >100^th in native VF.

2.3.2. Pathway and Network Enrichment Analyses

ECM organization and Immune pathways were highlighted in Voronoi Diagrams generated using Reactome. In ECM organization, the sub-pathway laminin interactions and sub-sub pathway anchoring of fibers in collagen formation were overrepresented for VF dECM from Protocol 3 in comparison to native VF and Cook Biotech’s SIS dECM (Figure 4A–C). Voronoi diagrams visually showed the comparative overrepresentation of ECM organization pathways in the dECMs compared to native VF, i.e. laminin interactions in VF dECM and fibronectin matrix formation in SIS dECM. Signaling by receptor tyrosine kinases includes pathways associated with angiogenesis including signaling by VEGF and PDGF. In our analyses, VEGFR-mediated vascular permeability, a process involved in inducing angiogenesis, was represented at comparable levels between sample types.^[63] Signaling by PDGF was more highly represented in VF and SIS dECM compared to native VF. This pathway involves the recruitment of smooth muscle cells or pericytes and is therefore instrumental in neo-blood vessel stabilization.^[64] MET signaling, while not as commonly associated with angiogenesis, also plays a role by stimulating VEGF release and endothelial cell motility, as well as regulating blood vessel permeability.^[65] MET signaling was overrepresented in VF dECM compared to our other samples, notably with respect to the MET promotes cell motility pathway.

In contrast, immune system pathways such as neutrophil degranulation and signalling by interleukins were overrepresented in native VF compared to either SIS or VF dECM source. This visualization provides an overview of information upon which we based more in-depth analysis of immune system pathways in our tissue sources.

In our further analyses on statistically significant proteins involved in the innate immune system, 17 abundant proteins were allocated to the neutrophil degranulation pathway, 14 of which (AHSG, PMG1, CCT2, CBR1, CCT8, PGAM1, SERPINB1, AP2A2, GPI, HBB, SERPIN1, STOM) were exclusively found in this pathway (Table 2). Neutrophil degranulation, a pathway involved in inflammatory disease in the VF, was also found to be overrepresented in the native VF.^[66] Of proteins involved in neutrophil degranulation, CRISP3 was found exclusively in native VF, MME in both native VF and VF dECM, and RAC1 exclusively in SIS dECM. Among statistically significant proteins in the adaptive immune system, four proteins were allocated to the MHC II adaptive immune response, of which one, CAPZ8 was not found in VF dECM (Table 3). The interleukins pathway in cytokine signaling contained 13 proteins, two of which (SOD1 and CA1) were identified exclusively in native VF, while CA2 was found in native VF and SIS dECM.

Table 2.

Alternate IDs of proteins with statistical significance (p <0.01583) assigned to the “Innate Immune System” sub-pathway via Reactome. Color intensity correlates to the log10 spectra of protein abundance.

Table 3.

Alternate IDs of proteins with statistical significance (p <0.01583) assigned to the “Adaptive Immune System” and “Cytokine Signaling in Immune System” sub-pathways via Reactome. Color intensity correlates to the log10 spectra of protein abundance.

Gene Ontology (GO) term enrichment analysis of overrepresented proteins in Cytoscape helps to produce results that are more consistent and reproducible. The average number of proteins involved in ECM-related (Figure 6A), angiogenesis-related (Figure 6B), and immune related (Figure 6C) GO terms were identified in the biological process category.

Figure 6.

Gene ontology of biological process terms using BiNGO Cytoscape plug-in. A. ECM-related terms. B. Angiogenesis related terms C. Immune-related terms. The number of proteins involved in each term identified in SIS dECM, VF dECM, and Native VF samples (n = 6/group) was determined. Statistical analysis was performed by ANOVA with post-hoc Tukey Tests. *p< 0.05, ** p< 0.01, *** p < 0.001, **** p < 0.0001. These terms are represented in an interaction diagram in Figure S4.

ECM-related GO terms (Figure 6A) showed that SIS dECM contained comparable numbers of proteins with VF dECM, but significantly greater than native VF in the “extracellular matrix organization” (17.2 ± 0.75 v/s 17.0 ± 0.63 v/s 15.7 ± 1.0 proteins) and “collagen fibril organization” (6.8 ± 0.41 v/s 6.2 ± 0.41 v/s 6.0 ± 0.0 proteins) terms. Significantly greater protein number involved in cell-matrix adhesion were identified in VF dECM (10.0 ± 0.0 proteins) compared to SIS dECM (5.8 ± 2.9 proteins), with no significant difference with native VF (8.5 ± 1.4 proteins). Proteins involved in two processes were only identified in SIS dECM: collagen biosynthetic process (2.5 ± 1.2 proteins) and collagen metabolic process (3.3 ± 1.6 proteins). Cell adhesion was only identified in native VF and VF dECM. Comparable protein numbers were found for all sample types for integrin activation, integrin-mediated signaling pathway, and calcium-independent cell-matrix adhesion. No significant difference in protein number was identified between native VF and VF dECM ECM-related terms.

For angiogenesis-related biological process GO terms (Figure 6B), while comparable protein numbers were identified for “negative regulation of angiogenesis” in native VF, VF dECM, and SIS dECM (v/s 2.3 ± 3.6 v/s 2.0 ± 3.1 proteins), these terms were predominantly identified in SIS dECM. Proteins were only found in SIS dECM for the term “response to hypoxia” (6.7 ± 3.3). This biological processes was key to initiation of angiogenesis.

Regarding immune-related biological processes (Figure 6C), more GO terms were identified in SIS dECM, with a particularly high number of proteins in “cellular response to stimulus”. The highest number of proteins identified in all three tissue types were related to “regulation of cell migration” (native VF: 15.5 ± 11.9; VF dECM: 23.2 ± 2.3; SIS dECM:23 ± 0.89 proteins). SIS dECM contained a significantly higher number of proteins than VF dECM in “innate immune response” (14.8 ± 0.9 v/s 2.8 ± 5.3 proteins) and “regulation of wound healing” (11.5 ± 0.4 v/s 9.0 ± 0.9 proteins). Compared to native VF, VF dECM had comparable proteins in “innate immune response” (11.8 ± 6.0 proteins) but significantly less proteins for “regulation of wound healing” (11.2 ± 0.4 proteins). Interestingly, only SIS dECM contained proteins involved in “regulation of acute inflammatory response” (3.3 ± 1.6 proteins), and “humoral immune response mediated by circulating immunoglobulin” (4.0 ± 0.0 proteins), while native VF and VF dECM had significantly more GO terms in “mucosal immune response”. In general, VF dECM, contained significantly less proteins associated with immune processes such as “regulation of wound healing”, and “regulation of cell migration”, and no proteins in “acute inflammatory response”. Although the presence of these proteins may create an adverse foreign body response, the complete absence of these immune-related proteins may also deter the initiation of effective wound healing.^[67]

To link Gene Ontology (GO) term enrichment to specific proteins, a table of the top 35 proteins identified in three categories of biological processes was generated for each sample type (Table 4). This analysis identified the comparative roles of top proteins in regeneration between tissue types. One biological process, muscle cell development, was exclusively linked to VF dECM in our analyses. Three biological processes, collagen biosynthetic process, cellular response to stimulus, and response to wounding, were identified only in the SIS dECM network. The process response to coagulation was linked to only native VF and SIS dECM, while the process extracellular structure organization was linked to only native VF and VF dECM. The remaining nine selected biological processes were common to all three tissue types.

Table 4.

Proteins associated with Biological Process Gene Ontology terms identified in Cytoscape.

2.4. Quantitative Comparison of Key Vocal Fold ECM Components by Tissue Source

To complement the mass spectrometry data, biochemical assays were performed to quantitatively compare the major ECM components collagen, elastin, and hyaluronan in native VF to our VF dECM and Cook Biotech’s SIS dECM. Total collagen by dry mass of samples (Figure 7A) was greatest in the VF dECM samples (237 ± 26 μg/mg), significantly greater than native VF (204 ± 40 μg/mg) and SIS dECM (167 ± 15 μg/mg). The quantity of total collagen by mass was greater in VF dECM than native VF because native VF contained mass from intra-cellular content while VF dECM contained ECM alone. Content of soluble collagen was significantly reduced in VF dECM (59 ± 3 μg/mg) compared to native VF (150 ± 19 μg/mg), with SIS dECM (98 ± 19 μg) between the two (Figure 7B).

Figure 7.

Quantitative Comparison of Key Vocal Fold ECM Components by Tissue Source A. Total Collagen Content (μg/mg tissue) B. Soluble Collagen Content (μg/mg tissue) C. Elastin Content (μg/mg tissue) D. Hyaluronan Content (μg/mg tissue). Statistical Analysis was performed by ANOVA with post-hoc Tukey Tests. *p< 0.05, ** p< 0.01, **** p < 0.0001

Following decellularization, elastin content was significantly reduced by 5-fold (native VF v/s VF dECM: 13 ± 1.7 μg/mg v/s 2.5 ± 0.54 μg/mg) (Figure 7C). Interestingly, VF dECM and SIS dECM (2.4 ± 0.54 μg/mg) contained comparable quantities of elastin.

Hyaluronan content was also significantly reduced by almost 2-fold after decellularization (native VF: 14 ± 2.6 μg/mg v/s 7.8 ± 0.95 μg/mg in VF dECM) (Figure 7D). Hyaluronan content of VF dECM and SIS dECM were not significantly different. The values of ECM component content are compiled in Table 5.

Table 5.

Summary of Quantitative Comparison of Key Vocal Fold ECM Components.

	Native VF	VF dECM	SIS dECM
Soluble Collagen (µg)	150 ± 19	59 ± 3	98 ± 19
Total Collagen (µg)	204 ± 40	237 ± 26	167 ± 15
Elastin (µg)	13 ± 1.7	2.5 ± 0.40	2.4 ± 0.54
Hyaluronan (µg)	14 ± 2.6	7.8 ± 0.95	8.9 ± 1.4

2.5. Viability and Cytotoxicity of Human Vocal Fold Fibroblasts

HVFF were cultured on VF and SIS dECM hydrogels as well as gelatin-coated glass and collagen I hydrogel controls to ensure the biocompatibility of our dECM hydrogels. HVFF exhibited greater than 90% viability on all substrates (Figure 8A). Between day 1 and 14, active proliferation of HVFF was observed as indicated by a significant increase in cellular density across substrates (Figure 8B). On day 1, cellular density was comparable on all substrates. On day 7, the HVFF density on gelatin-coated glass (175 ± 23 cells/mm²) was significantly greater than all other groups. By day 14, the cellular density on both collagen and glass groups were comparable (160 ± 26 cells/mm² v/s 185 ± 20 cells/mm²). Unexpectedly, proliferation was significantly lower on VF dECM (77 ± 4.7 cells/mm²) or SIS dECM (83 ± 5 cells/mm²). No significant difference was identified between cellular density of HVFF cultured on VF and SIS dECM across all time points.

Figure 8.

Viability and Cytotoxicity of Human Vocal Fold Fibroblasts A. Fluorescent imaging of HVFF on Glass, Collagen, SIS dECM, or VF dECM over 14 days. B. Count of HVFF (cells/mm²). C. Neo-Collagen Deposited per Scaffold D. Micrograms Neo-Elastin Deposited per Scaffold (in μg). E. Neo-Hyaluronan Deposited per Scaffold (in μg). Statistical Analysis was performed by Two-Way ANOVA with post-hoc Tukey Tests. *p< 0.05, ** p< 0.01

2.6. Deposition of Neo-ECM by HVFF on dECM Hydrogels

Neo-ECM assays were performed to evaluate the impact of dECM sources on HVFF functions. Soluble collagen, the form in which neo-collagens are deposited, was detected in initial cell-free “Day 0” samples at substantial quantities, but lost the majority of their soluble collagen by day 14. Soluble collagen content of cell-free SIS dECM, VF dECM and collagen hydrogels significantly decreased from 351 ± 22 μg/scaffold to 2.99 ± 0.35 μg/scaffold, from 171 ± 3.0 μg/scaffold to 6.63 ± 0.11 μg/scaffold from 224 ± 40 μg/scaffold to 9.16 ± 0.42 μg/scaffold, respectively.

The only significant increase in deposited collagen content identified within a sample type over time was for the collagen control between day 7 (168 ± 49 μg/scaffold) and day 14 (211 ± 39 μg/scaffold) (Figure 8C). A significant decrease in VF neo-collagen content was observed between day 7 (169 ± 42 μg/scaffold) and day 14 (125 ± 28 μg/scaffold). On the glass control, only small amounts of collagen were deposited after 14 days. The quantity of deposited soluble collagen after 14 days on collagen hydrogels, VF dECM hydrogels, and SIS dECM hydrogels (279 ± 50 μg/scaffold) were significantly greater than on glass. The quantity of deposited soluble collagen on SIS dECM hydrogels was significantly higher than VF dECM at all three time points, but not significantly different from collagen hydrogels on day 7 and day 14.

Negligible quantities of elastin and hyaluronan were identified in the initial “Day 0” hydrogels, due to the lower quantity of these components in the dECM, indicating that the presence of these ECM components at later time points was likely attributed to de novo production by HVFF. Neo-elastin was reliably detectable after 3 days on VF dECM, SIS dECM, and collagen samples and after 7 days on glass (Figure 8D). After 14 days, the quantity of neo-elastin deposited on VF dECM (46 ± 6.4 μg/scaffold) and SIS dECM (57 ± 10 μg/scaffold) was significantly greater than on collagen (33 ± 7.9 μg/scaffold). HVFF on glass produced the lowest quantity of neo-elastin (22 ± 4.2 μg/scaffold). No significant difference was identified between the two dECM types.

Neo-hyaluronan deposition was reliably detectable on VF dECM and SIS dECM samples after 3 days of HVFF culture, collagen samples after 7 days, and glass after 14 days (Figure 8E). Significantly greater quantities of neo-hyaluronan were produced by HVFF on VF dECM (5.6 ± 0.76 μg/scaffold) and SIS dECM (6.7 ± 1.1 μg/scaffold), compared to collagen (2.3 ± 0.76 μg/scaffold) and glass (1.5 ± 0.59 μg/scaffold). No significant difference in deposition of neo-hyaluronan was found between the two dECM sources, or between collagen and glass.

3. Discussion

The present study aimed to investigate the differences in proteomic composition between porcine VF dECM and SIS dECM, and the impact of tissue source on the production of neo-ECM in vitro to evaluate SIS dECM as a potential scalable alternative to VF dECM. In this study, a decellularization protocol for VF tissues was optimized to reduce nuclease incubation time and produce more stable, reliably gelling VF dECM hydrogels through delipidization.^[14,68]

3.1. Proteomic Insights into Tissue Specificity

Comparing the homogenized VF dECM and commercially produced SIS dECM powders with the composition of native VF, we found that VF dECM preserved many of the same processes based on GO term enrichment as native VF. The majority of these ECM-related processes were also present in SIS dECM. When abundance levels were taken into account, VF dECM was more similar to native VF than SIS dECM as expected, but SIS samples exhibited lower variability between samples.

Variability observed between VF dECM, and native VF could be partially accounted for by the removal of cellular proteins during decellularization. Conversely, the most probable reason that proteins were identified in VF dECM, but not native VF is that proteins present at low levels became detectable when more abundant cellular proteins were removed through decellularization.^[69] Differences in decellularization protocols could also cause differences in protein abundances between the two dECMs. While VF dECM is tissue-specific, the increased scalability and deceased variability of SIS dECM could provide practical advantages for application. In future work, additional proteomic analyses following cell culture would provide a comprehensive picture of variation in regenerative response due to tissue source, as Leng et al demonstrated in evaluating the response of epidermal cells to decellularized skin scaffolds.^[5]

Core matrisome proteins were largely preserved following our decellularization protocol. This includes collagens necessary to the durability of VF, glycoproteins involved in elastic fiber deposition and angiogenesis, and proteoglycans involved in angiogenesis and ECM binding. ECM regulators and soluble factors are largely lost during decellularization due to their solubility, but proteins involved in stimulating their secretion remain.^[5] These proteins are often expressed at low levels under healthy conditions, and secreted in response to injury.^[5]

However, pathway and GO term analyses indicated that decellularization caused an overall depletion of immune-related proteins in VF dECM prepared using Protocol 3 compared to native tissue. The overrepresentation of immune system pathways in native VF compared to VF and SIS dECMs likely represents the removal of inflammatory proteins from decellularization. That being said, certain proteins involved in inflammatory processes such as neutrophil degranulation were preserved in both dECMs. Neutrophil degranulation is an immune pathway that helps prevent infection, but overexpression is associated with inflammatory disease in the VF, such as laryngeal papilloma.^[8,56] Decreased expression of neutrophil degranulation and other inflammatory immune pathways in dECM biomaterials may contribute to improved regenerative responses. The lower abundance levels of proteins involved in the immune response in dECM compared to native VF could help prevent deleterious levels of inflammation.

Decellularization protocols do not necessarily remove all antigenic proteins.^[13] In a previous proteomic study on Bovine VF, Welham et al. listed proteins identified in their samples that are associated with processes that may cause deleterious responses to the xenogeneic dECM.^[13] These processes were identified from the biological process category of GO term enrichment and included: leukocyte-mediated immunity, oxidative stress response, cell killing, and biotic stimuli response.

In our GO term enrichment analyses, these processes were not identified as overexpressed. However, we did identify some of the proteins overlapped with those reported by Welham et al. in our samples (Table 6). The observed difference could be due to a combination of tissue source (porcine VF and SIS vs bovine VF), decellularization protocol, and differing extraction and analysis techniques during mass spectroscopy.

Table 6.

Comparison of Antigenic Proteins Reported from Previous VF dECM Literature^[13] in Native VF, VF dECM, and SIS dECM Samples. ╳ indicates the presence of the protein. Nil indicates the protein was not identified in samples.

Antigenic Proteins	Native VF	VF dECM	SIS dECM
HMGB1	╳	╳	╳
LMNB2	╳	╳	╳
PRDX1	╳	╳	╳
PRDX2	╳	╳	╳
PRDX4	╳	╳	╳
SOD1	╳	Nil	Nil
GPX3	╳	╳	╳
LYSC	Nil	Nil	Nil
DEF1	Nil	Nil	Nil
LPO	Nil	Nil	Nil
ALB	╳	╳	╳
C4BPA	Nil	Nil	Nil
HP	╳	╳	╳
ANXA1	╳	╳	╳
ANXA2	╳	╳	╳
ANXA5	╳	╳	╳
ANXA11	╳	╳	╳
ENO1	╳	╳	╳
LMNA	╳	╳	╳

While some researchers have suggested that all immunogenic proteins should be selectively removed from dECM biomaterials, this approach may prove a double-edged sword and undermine the immunomodulatory potential of dECM.^[13] For instance, we have identified proteins on this list such as annexins (ANXA1, 2, 5, and 11) as involved in immune mediation and angiogenesis. Additionally proteins that can cause severe, deleterious immune responses, such as those containing xenogeneic α-Gal epitopes, may be reduced to tolerable levels with existing decellularization techniques.^[71,72]

Several dECM sources have been reported to induce pro-regenerative immune responses, including SIS decellularized using a peracetic acid-ethanol method.^[73–76] The presence of the MHC II pathway may indicate that VF dECM is also capable of inducing a pro-regenerative response, though further validation is required. Proteomic analysis of macrophage response to VF dECM could be used to explore this possibility.^[73] Macrophages in the VF may polarize to inflammatory phenotypes that contribute to scar formation or anti-inflammatory phenotypes that contribute to tissue reconstruction. The induction of macrophage polarization towards the M2 phenotype would provide key evidence toward the applicability of VF dECM hydrogels toward VF repair.^[8,77] Rather than solely relying on proteomic analyses and the presence of remnant immunogenic proteins, these results should be used in combination with in vivo studies to predict the immune response.

With respect to angiogenesis, our abundance curve analysis identified HSP90AA1, HSPB1, and CRK as the most abundant proteins in the VEGF signaling pathways across all three sample types. Interestingly, our GO term analysis showed that SIS dECM contained the greatest number of proteins involved in response to hypoxia. VEGF signaling is involved in the sprouting of new blood vessels, and is an essential step in VF defect regeneration that has not been extensively explored for VF biomaterials.^[78] Our evidence suggests that dECM biomaterials possess intrinsic capacity to modulate VF endothelial cells for capillary regeneration, but functional assays are required for confirmation.

Taken together, the stimulation of regenerative processes including ECM production, immune modulation, and angiogenesis can lead to effective tissue regeneration in the VF. The similar proteomic profiles of SIS and VF dECM with respect to the aforesaid processes, indicate a potential for comparable VF-specific regenerative outcomes, which were partially verified with our subsequent functional analyses.

3.2. Validation of Proteomic Analyses through ECM Component Quantification

The use of quantitative and functional analyses is often needed to confirm the results of proteomic analysis in various aspects. For instance, proteomic analysis cannot quantify the ratio between total collagen versus the uncrosslinked, soluble collagens that contribute to hydrogel formation. The availability of soluble collagen is expected to affect the consistency of hydrogel formation and the production of healthy tissue when applied in vivo.^[29,79] Greater amounts of soluble, uncrosslinked collagen is associated with production of organized collagen and decreased scar formation, as in fetal skin, compared to lower amounts of soluble collagen.^[79] Our Sircol-based biochemical assay identified a reduction in soluble collagen content from native VF to VF dECM. The most likely cause was crosslinking during the isopropanol step.^[4,80] However, delipidization has been shown to generate more stable dECM hydrogels in a pancreas dECM hydrogel study, an important factor for scale-up to manufacture.^[80]

Based on biochemical analyses, elastin and hyaluronan levels were reduced following decellularization. However, the quantitative levels were comparable between VF dECM and SIS. The reduction from VF decellularization is expected from literature, as elastin may be lost during exposure to salts.^[81,82] As anticipated, hyaluronan quantities were significantly reduced in VF dECM samples. The primary reason is that hyaluronan, as a GAG bound to cell membranes, is often lost with cell removal, and GAG loss has been shown to increase with higher DNase concentrations.^[83,84]

Despite the reduction in elastin and hyaluronan quantities, core matrisome proteins involved in the elastin formation and hyaluronan binding processes were identified by our proteomic analyses. We therefore predicted that these proteins in the dECM scaffolds would stimulate production of elastin and hyaluronan by local cells, even after they were lost during decellularization.

3.3. In vitro Validation of Neo-ECM Production by dECM Hydrogels

The production of neo-ECM is an essential component in the repair of defects in soft tissues such as VF.^[85] Our proteomic analyses indicated overrepresentation of proteins and pathways involved in the production of neo-ECM components including collagen, elastin, and hyaluronan. In vitro neo-ECM deposition assays were used to determine whether these identified core matrisome proteins were bioactive.

While significant increases of elastin and hyaluronan were identified on scaffolds after 14 days, results of the neo-collagen assay were inconclusive. The hydrolysis-based control used in this study does not account for collagen degradation by MMPs and may not be sufficient to account for all collagen that is degraded over the culture period, skewing the amount of collagen deposited.^[5,86–89] The significant increase in quantity of elastin and hyaluronan partially validated the hypothesis that SIS and VF dECM stimulated de novo ECM production by HVFF.

However, further analyses such as histology and quantitative polymerase chain reaction (QPCR) are needed to support the regenerative potential of dECM hydrogels for VF tissue engineering. Virtual histology with the use of advanced imaging techniques such as nonlinear laser-scanning microscopy can visualize remodeled tissue and determine whether ECM organization is characteristic of healthy or scarred tissue.^[10,90–91] QPCR can be used to determine whether HVFF adopt an inflammatory or anti-inflammatory phenotype when cultured on SIS and VF dECM hydrogels by expression of markers such as ACTA2.^[19]

3.4. Future Directions

In our proteomic and biochemical analyses, some key ECM proteins were found to be significantly reduced in decellularized tissues compared to native VF. A strategic replenishment of depleted ECM components in dECM biomaterials may help boost its regenerative capacity for VF tissue engineering.

Native elastin is difficult to isolate and purify, limiting its utility in tissue engineering. In recent years elastin-like peptides (ELPs) have been synthesized from recombinant elastin genes in vitro with properties that may replicate the mechanical and biological functions of native elastin.^[92,93] For instance, ELP coatings on an electrospun TecoFlex^™ scaffolds upregulated production of neo-elastin and collagen III by HVFF after 7 days.^[92] If necessary, ELPs could be incorporated into the hydrogel formulation, through the incorporation of crosslinkable residues in the recombinant protein, though these methods remain challenging.^[93]

Replenishment of hyaluronan and other GAGs has also been suggested to improve the efficacy of dECM scaffolds.^[94,95] Hyaluronan is commonly used as a scaffold on its own and crosslinked with other materials. In its uncrosslinked form, hyaluronan degrades rapidly in vivo. Thiol-modified hyaluronan has been crosslinked with porcine UBM scaffolds, and induced increased contractility of smooth muscle cells and matrix remodeling over 28 days compared to UBM scaffolds without replenished hyaluronan.^[96] A hydrogel produced from methacrylated lung dECM and hyaluronan was found to increase cell proliferation and possess rheological mechanical properties comparable to native lung tissue.^[97] Hyaluronan-based biomaterials are already widely developed for VF biomaterials with encouraging outcomes.^[98] Therefore, the integration of hyaluronan to dECM scaffolds should be highly feasible.

Additionally, collagen loss to media from the control VF and SIS dECM hydrogels during the neo-ECM deposition experiments reflects a shortcoming common to existing VF injectables: rapid degradation. Degradation of VF biomaterials without compete tissue regeneration leads to a need for repeated reinjection.^[99] Crosslinking the dECM with itself or another naturally derived biomaterial could aid in biomaterial stability, resulting in a biomaterial that degrades as native tissue regenerates.^[100]

Beyond the ability to stimulate deposition of three key ECM components, further in vitro and in vivo assays would help to verify the bioactivity of the VF and SIS dECM hydrogels, and determine their applicability toward VF regeneration. The ability of VF and SIS dECM hydrogels to stimulate angiogenesis could be investigated through 3D tube formation assays and immunostaining for lumen development markers such as PECAM-1 and Actin. The immune response could be investigated through in vitro macrophage polarization experiments, and in animal models to evaluate the foreign body response. In both cases, QPCR and RNA-seq could be used to gain a more comprehensive outlook into the gene transcripts for a broad panel of angiogenic and pro-/anti-inflammatory markers.

4. Conclusions

Mass-spectrometry based proteomics have contributed to identifying the base proteomic composition of the dECM, as well as, examining the impact of composition on ECM regeneration. We demonstrated that VF dECM possesses greater similarity to native VF than SIS dECM, pointing toward tissue specific effects. However, SIS dECM stimulated the production of key VF neo-ECM components, elastin and hyaluronan, at comparable levels to VF dECM with less variation between samples. These results indicated that SIS dECM may prove a viable, more scalable alternative to tissue-specific VF dECM. As development of dECM hydrogels progresses toward applications in VF repair, we recommend further proteomic analyses with respect to functional response to the VF and SIS dECM hydrogels. This study demonstrated a direct link between proteomic content of dECM and the production of neo-ECM in the VF that will guide and help to optimize selection of dECM tissue source and evaluation methodology.

5. Experimental Section

5.1. Materials

Porcine larynges were sourced from a local farm and abattoir (Ferme Co-op Point du Jour, Bury, QC, Canada). Powder ECM from small intestinal submucosa was generously donated from Cook Biotech (West Lafayette, IN, USA). Ribonuclease A (RNase) was from Roche Diagnostics GmbH (Mannheim, Germany). Pepsin from porcine gastric mucosa (2500 U/mg protein), Deoxyribonuclease I (DNase) from Bovine Pancreas (400 KU/mg protein), Gelatin from bovine skin type B, Papain from papaya latex (10 U/mg protein), Fetal Bovine Serum, and Penicillin-Streptomycin (10,000 U/mL) were purchased from Sigma-Aldrich (St. Louis, MO, USA). 3 mg/mL Collagen I from Rat Tail, Gibco^® Cell Dissociation Buffer, and Phosphate Buffered Saline was bought from Gibco (Grand Island, NY, USA). Sodium Chloride and Thermo Scientific^™ Nunc^™ Lab-Tek^™ II Chamber Slide^™ System Slides were from Fisher Scientific (Saint-Laurent, QC, Canada). Phenylmethylsulfonyl fluoride (PMSF) and the Quant-IT^™ PicoGreen DNA assay kit (P7589) were from Thermo Fisher Scientific (Waltham, MA, USA). The Fastin^™ Elastin Assay (F2000), Sircol^™ Soluble Collagen Assay (S5000), Hyaluronan Purple-Jelley Assay (H1000), Denatured Collagen Standard 1mg/ml (S2010), and Fragmentation Reagent (SFRAG) were from Biocolor Life Science Assays (Carrickfergus, UK). Dulbecco’s modified Eagle medium was purchased from Life Technologies, Inc. (Burlington, ON). The Live/Dead Viability/Cytotoxicity Kit (L3224) was purchased from Invitrogen (Eugene, OR, USA).

Powder ECM from SIS was generously donated from Cook Biotech (West Lafayette, IN, USA). SIS dECM was produced according to Cook Biotech’s standard manufacturing protocols.^[102] The submucosa of the porcine small intestine was first mechanically separated from the other intestinal layers. Samples were washed in water and stored at −80 °C until additional processing. Defrosted SIS was then disinfected and decellularized in peracetic acid and ethanol. Rinses with high-purity water were used to remove the solution. The SIS dECM was then lyophilized into sheets and comminuted into powder. Terminal sterilization was performed with ethylene oxide gas under standard temperatures, pressures, and durations.

5.2. Decellularization of Porcine Vocal Folds

Porcine larynges were stored at −80 °C in PBS until use. A decellularization protocol was adapted from a previously developed protocol.^[68] Larynges were thawed in PBS overnight prior to dissection. VF were removed, washed to remove blood, and minced with a scalpel to increase surface area for decellularization. Minced VF were first agitated in 3 M sodium chloride on a shaker plate at 4 °C for 24 h to lyse cells.

The existing protocol used 24-hour enzymatic incubation steps. However, nucleases lose activity over time, and longer incubation times can damage the protein content of the ECM.^[101,103] Frequent enzyme changes, dubbed “cycles”, were predicted to decrease the total incubation time. Seven nuclease incubation protocols with varying incubation time, length of incubation cycle and number of incubation cycles were thus tested in this study. Nuclease Incubation was performed at 37 °C in 25 μg/mL DNase and 10 μg/mL RNase for 6 or 24 h under constant agitation to break down DNA in the tissue samples.

Delipidization was performed after the first incubation cycle by agitating minced VF in isopropanol for 24 h at 4 °C. The inclusion of a delipidization has been reported to enhance gelation kinetics and integrity.^[80] For Protocols 2, 3, 4, 6 and 7, additional nuclease incubation cycles followed delipidization. 1% Penicillin-Streptomycin and 1 U/mL PMSF were used in all steps to prevent bacterial growth and protease activity.

Between all steps, the minced tissue was washed under running water in a Falcon^® 100 μm cell strainer with evenly spaced nylon pores to remove reagents. The lyophilized tissue was homogenized into powder using a SPEX SamplePrep 6775 Freezer Mill.

5.3. DNA Quantification

DNA quantification was performed according to established methods to identify a decellularization protocol that effectively reduced the DNA content of VF in the minimum total nuclease incubation time.^[4,104,105] Selection of a final protocol was based on reduction of DNA content to less than 50 ng per milligram of tissue.^[4] Decellularized VF powder from each decellularization protocol was digested in 250 μg/mL papain in 1X TE buffer (10 mM Tris-HCL, 1 mM EDTA, pH 7.5) at 60 °C for 16 h.^[27] Each dECM sample was produced from a mixture of VF (N=4), with three biological replicates and two technical replicates. After digestion 100 μL of PicoGreen^® Reagent was added to 100 μL of the digests and fluorescence measured at 520 nm in a SpectraMax i3x Multimode Plate Reader from Molecular Devices (San Jose, CA, USA).

5.4. Hydrogel Formation and Characterization

To prepare dECM pre-gels, 20 mg/mL SIS dECM or VF dECM powder was slowly stirred in a sterile filtered solution of 1 mg/mL pepsin and 0.05 M HCl. After 48 hours, the solution pH and salt concentration were neutralized with sterile filtered 1 M NaOH and 10X PBS respectively and diluted to 8 mg/mL with ddH₂O. Collagen I pre-gels were similarly prepared by pH and salt neutralization of the 3 mg/mL solution with 1 M NaOH and 10X PBS and diluted to 2 mg/mL with ddH₂O. After neutralization, the solutions were incubated at 37 °C for 30 min to form hydrogels. The gelation kinetics and viscoelastic properties of the hydrogels were analyzed using 2 h time sweeps on a TA Instruments Discovery HR-2 Rheometer.

5.5. Mass Spectrometry

Proteomics analysis was performed by the Clinical Proteomics Platform, Research Institute of the McGill University Health Centre, Montréal, Canada. Samples of native porcine VF, VF dECM, and SIS dECM powder (N = 6 for each group) were submitted for analysis. The native VF were homogenized using the same process described in Section 5.2, without undergoing decellularization and used as a control for the dECM.

At the Clinical Proteomics Platform, tissue lysates for each sample were loaded onto a single stacking gel band to remove lipids, detergents and salts. The single gel band containing all proteins was reduced with DTT, alkylated with iodoacetic acid and digested with trypsin. 2 ug of extracted peptides were re-solubilized in 0.1% aqueous formic acid and loaded onto a Thermo Acclaim Pepmap (Thermo, 75uM ID × 2cm C18 3uM beads) precolumn and then onto an Acclaim Pepmap Easyspray (Thermo, 75uM × 15cm with 2uM C18 beads) analytical column separation using a Dionex Ultimate 3000 uHPLC at 250 nL/min with a gradient of 2–35% organic (0.1% formic acid in acetonitrile) over 3 hours. Peptides were analyzed using a Thermo Orbitrap Fusion mass spectrometer operating at 120,000 resolution (FWHM in MS1) with HCD sequencing (15,000 resolution) at top speed for all peptides with a charge of 2+ or greater.

5.6. Bioinformatic Data Processing of Mass Spectrometry Data

Mascot database search, data validation and statistical testing in Scaffold Proteome Software, and data analysis were performed following data acquisition (Figure 9). The raw data were converted into *.mgf format (Mascot generic format) by Proteome Discoverer 2.1 for searching using the Mascot 2.6.2 search engine (Matrix Science, London, UK; version 2.6.2) against porcine protein sequences (Uniprot 2020; 49571 entries). Assuming the digestion enzyme trypsin, Mascot was searched with a fragment ion mass tolerance of 0.100 Da and a parent ion tolerance of 5.0 PPM. Carboxymethyl of cysteine was specified in Mascot as a fixed modification. Deamidation of asparagine and glutamine as well as oxidation of methionine and proline were specified in Mascot as variable modifications. A maximum of 1 missed cleavage was also used and a decoy search was performed. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [1] partner repository with the dataset identifier PXD033621.

Figure 9.

Illustrated workflow of the proteomics data analysis pipeline. The image of the LC-MS/MS equipment, a Thermo Scientific Ultimate 3000 HPLC and Orbitrap Fusion MS: Quadrupole-Orbitrap-Linear ion trap hybrid, was taken from the Thermo Fisher Scientific website

The database search results were loaded onto Scaffold Q+ Scaffold_4. (11.0, Proteome Software Inc., Portland, OR, USA) to validate MS/MS based peptide and protein identifications and to perform statistical tests.^[106] Peptide identifications were accepted if they could be established at greater than 95.0% probability by the Scaffold Local FDR algorithm. Protein identifications were accepted if they could be established at greater than 99.0% probability and contained at least 2 identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm.^[107] Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony. Proteins sharing significant peptide evidence were grouped into clusters. Decoy peptide false discovery rate (FDR) was determined to be 0.08% and decoy protein FDR was 0.8%. The statistical significance of the data were determined with ANOVA test, p-value < 0.05, and Benjamini-Hochberg correction, resulting in a corrected p-value < 0.02177. The data were also normalized.

All protein duplicates as well as proteins with incomplete identification information, including missing names and alternate IDs, were removed. The total spectra of the remaining 698 proteins were then log10 transformed.

The principal component analysis graph and corresponding scree plot were also created using the R programming language. A PCA loadings plot was generated in Markerview (SCIEX, v. 1.3.1) using Pareto scaling. Total area sums were used for data normalization with no weighting in order to visualize how each protein contributed to the corresponding principal components.

Pathway enrichment analysis was performed on the data using open-source Reactome pathway browser v3.7 and Reactome database v76.^[36] A list of protein alternate IDs and their corresponding log10 transformed total spectra was entered into the Reactome “Analysis Tools” feature for pathway identifier mapping, over-representation, and expression analysis. Pathways mapped to inputted proteins were highlighted in color. Yellow represents a relatively higher p-value of the overrepresentation statistical test, or weak overrepresentation of the pathway, and blue represents a relatively lower p-value of the overrepresentation statistical test, or strong overrepresentation of the pathway. These pathways were then displayed as Voronoi diagrams to present the p-value of the statistical test for over-represented molecular pathways, especially those related to the immune system and ECM organization.^[108]

Large-scale topological information molecular pathway network analysis was performed on the data using open-source bioinformatics software platform Cytoscape with plugin BiNGO 3.0.3 to calculate Gene Ontology (GO) term enrichment for the biological process, cellular component, and molecular function category.^[109–111] A hypergeometric test, Benjamini-Hochberg False Discovery Rate (FDR) correction, and a p-value threshold of 0.01 were implemented as statistical testing. Protein alternate IDs were used as input data and evaluated against a custom database of a porcine GO Annotation File (GAF) downloaded from Gene Ontology.^[112,113] The proteins were annotated to GO terms shown as nodes (circles) whose area was proportional to the number of genes in the data categorized to that node. Overrepresented GO terms were highlighted were highlighted in colour, with yellow representing a relatively higher adjusted p-value of the overrepresentation statistical test, or weak overrepresentation of the GO term, and orange representing a relatively lower adjusted p-value of the overrepresentation statistical test, or strong overrepresentation. White nodes represented those with no significant overrepresentation. For all 6 biological replicates of each tissue type (SIS dECM, VF dECM, and Native VF), the number of proteins annotated to each biological process GO term was determined for ECM, immune, and angiogenesis-related terms. The mean number of proteins and standard deviation were calculated for the GO terms of interest in each tissue type.

5.7. Quantitative Verification of dECM Components

ECM Quantification Assays were performed according to manufacturer instructions. For each assay, three replicates of 4–6 mg of dry, homogenized native VF, VF dECM, and SIS dECM powders were used. Native VF and VF dECM samples were produced from a mixture of VF (N=6) with samples selected from three decellularization batches.

Sircol^™ Soluble Collagen Assay.

Samples were prepared for the Sircol^™ Soluble Collagen Assay by overnight incubation in 1mL of 0.1 mg/mL pepsin in 0.5 M HCl at 4 °C. Digestion was halted with the addition of 100 μL acid neutralization reagent. To this solution, 200 μL Collagen Isolation and Concentration reagent was added and incubated again at 4 °C overnight. Concentrated collagen solutions were centrifuged at 12000 rpm and the supernatant aliquoted. Samples were prepared with 50 μL concentrated collagen solution and made up to 100 μL. Standards were prepared in the range of 0–45 μg. 1 mL Sircol Dye Reagent was added to each tube, and incubated at room temperature under gentle agitation for 30 min. The dye solution was centrifuged, and the supernatant removed, with tubes drained onto paper towels. Acid Salt Wash was added, and pellets centrifuged again to remove excess dye. After draining samples onto paper towels, 1000 μL Alkali Reagent was added and samples vortexed to release dye. Absorbance was measured at 555 nm.

Sircol^™ Insoluble Collagen Assay.

Insoluble collagen was assayed by incubating samples in 50 μL fragmentation reagent per 1 mg tissue for 2 hours at 65 °C. Fragmented collagen was then assayed as in the soluble collagen protocol, from the dye step forward. Denatured collagen standards of 0–60 ug were used. Absorbance was measured at 555 nm. In analysis, a 2.2X conversion factor was used to account for the 45% binding affinity of denatured collagen.

Fastin^™ Elastin Assay.

Elastin was extracted in 0.25 M Oxalic acid for 1 h at 100 °C to. After cooling to room temperature, samples were centrifuged at 10,000 rpm for 10 min. Supernatants were removed from samples, and the extraction process repeated on residual tissue. Samples were prepared from 100 μL with the first and second rounds assayed as separate samples. Standards were prepared containing 0–50 μg α-elastin. An equal volume of Elastin Precipitating Reagent was added to each sample and standard and vortexed. After 15 minutes, samples were centrifuged for 10 min, and supernatant discarded. 1 mL Fastin Dye Reagent was added to each tube. The precipitate was dispersed by vortexing, and samples were incubated at room temperature under gentle agitation for 90 min. Samples were centrifuged again and drained of dye before the addition of 250 μL Dye Dissociation Reagent. Samples were vortexed to release dye immediately and after a 10 min incubation period. Samples from round 1 of extraction were diluted at a 1:2 ratio, and samples from round 2 were not diluted. Absorbance was measured at 513 nm.

Purple-Jelley Hyaluronan Assay.

Tissue samples were digested in 400 μL Tris-HCl and 20 μL Proteinase at 65 °C for 3 h. After centrifuging at 12000 rpm for 10 min, supernatants were transferred to new microcentrifuge tubes mixed with 1 mL GAG precipitation reagent. After 15 min, samples were centrifuged, and supernatants discarded. Residues were extracted in 360 μL water for 15 min with intermittent vortexing before the addition of concentrated NaCl and Cetylpyridinium Chloride. Samples were left undisturbed for 30 min without mixing, then centrifuged, and residues discarded. The entire process was repeated from the GAG precipitation step once. GAG Precipitation Reagent was added a third time, but 500 μL 98% ethanol was added to residues after centrifugation rather than water and centrifuged again without mixing. Supernatant was removed and tubes dried by draining onto paper towels. The remaining pellet contained isolated HA, which was extracted with 100 uL water for 30 min using intermittent vortexing. Standards were prepared from 0–3 μg HA. 20 uL from each standard and sample was mixed with 200 μL Dye Reagent, and the absorbance measured at 655 nm.

5.8. Culture of Human Vocal Fold Fibroblasts

Human Vocal Fold Fibroblasts (HVFF) were generously supplied by Prof. Susan Thibeault (Department of Surgery, University of Wisconsin-Madison, USA). HVFF were cultured in T-75 flasks in a solution of Dulbecco’s modified Eagle medium (DMEM), 10%v/v fetal bovine serum, and 1%v/v penicillin/streptomycin under conditions of 5% CO₂ atmosphere and 37 °C. Media changes were performed every 3 days. Passages were performed when cells reached approximately 70% confluency using Gibco^® Cell Dissociation Buffer. Dissociated HVFF were centrifuged, suspended in fresh DMEM, and plated onto to fresh T-75 flasks. HVFF were used between passages 9 and 14 for this study.

5.9. Cellular Viability and Cytotoxicity

SIS dECM, VF dECM, and collagen I pre-gels were prepared as described in Section 5.4. On day 0 of cell experiments, 70 μL of SIS, VF dECM, and Collagen I pre-gels were pipetted into four wells for each time point. Pre-gels were incubated for at least 1 hr prior to cell seeding to form hydrogels with approximately 1 mm thickness.

Glass control wells were pre-coated with 0.1 mg/mL gelatin and incubated overnight prior to seeding. A total of sixteen 8-well chamber slides were used for each experiment, i.e. 4 replicates.

HVFF were seeded in each chamber well at a concentration of 10,000 cells per well. Additional media was then added for a total volume of 300 μL per well. HVFF were cultured under conditions of 5% CO₂ atmosphere and 37 °C over 14 days, with media changes performed every 2–3 days.

At each time point (1, 3, 7, or 14 days) media was removed from samples in randomly selected chamber slides and the HVFF and hydrogels were washed twice with PBS. 100 μL Calcein AM and Ethidium Homodimer solution was added to each well and incubated at room temperature while protected from light for 30 min. The dye solution was removed and each sample washed two additional times with PBS. A final 100 μL of PBS was added to keep samples hydrated, and all samples were imaged on a Zeiss LSM 800 Confocal Microscope the same day. Each image covered a region of 3 mm × 3 mm, and z-stacks were used to ensure imaging of all cells in the imaging region. Cell counting was performed using ImageJ.^[114]

5.10. Deposition of Neo-ECM Components

Chamber slides were prepared as described in section 5.8. However, two chamber slides were set aside without cell seeding for immediate “0 day” analysis. Cells were seeded on the remaining 14 chamber slides as described in section 5.9 and cultured for up to 14 days. Chamber slides were randomly selected for analysis on days 3, 7, and 14. This cell culture procedure was followed for both deposition of neo-Elastin and neo-Hyaluronan ECM Quantification assays. Three technical replicates were used for each assay.

Sircol^™ Soluble Collagen Assay:

The assay was performed as described in Section 5.6 with slight modifications in sample preparation. Volumes in the sample preparation step were reduced by half to ensure collagen would be detectable across all samples. Collagen content of control hydrogels incubated in DMEM without cells was measured at each time point to account for collagen degradation over time. For the assay, 100 μL samples were used.

Fastin^™ Elastin Assay:

To prepare samples for the Fastin^™ Elastin Assay, hydrogels were removed from wells and transferred to 1.5 mL microcentrifuge tubes. Cells were removed from glass controls using Gibco^® Cell Dissociation Buffer and centrifuged in PBS. The Fastin^™ Elastin was then performed as described in Section 5.6, with no dilution prior to the measurement of absorbance to account for the lower concentrations compared to tissue.

Purple-Jelley Hyaluronan Assay:

Samples were prepared for the Purple-Jelley^™ Hyaluronan Assay and the assay performed as described in Section 5.6.

5.12. Statistical Analysis

Statistical Analyses were performed using GraphPad Prism 9 (GraphPad Software, Inc., USA). Data was analyzed using one or two-way ANOVA followed by Tukey’s post-hoc test. Data values are presented as mean ± SD. Values of p < 0.05 were deemed statistically significant.

Figure 5.

Voronoi Diagrams of signaling by receptor tyrosine kinases, extracellular matrix organization, and immune system pathways generated using Reacfoam pathway analysis in Reactome of log10 total spectra for A. Native VF B. VF dECM and C. SIS dECM. Pathway enrichment is shown as a spectrum from dark blue (underrepresented) to yellow (overrepresented). Overview Voronoi Diagrams of all pathways identified in Reactome are available in Figures S1–S3.

Table 1.

Table of Decellularization Protocols and Residual DNA Averages for VF dECM and SIS dECM (if data available). Nucleases are not used in the decellularization of Cook Biotech’s SIS dECM, and complete details of the protocol are not available due to propriety.

Protocol Number	Total Nuclease Incubation Time (h)	Length of Each Incubation Cycle (h)	Number of Incubation Cycles	Residual DNA content (ng)
1	6	6	1	232 ± 12
2	12	6	2	127 ± 35
3	18	6	3	43.4 ± 5.9
4	24	6	4	28.4 ± 10
5	24	24	1	118 ± 35
6	48	24	2	32.3 ± 13
7	72	24	3	12.6 ± 6.2
SIS dECM	N/A	N/A	N/A	0.42 ± 0.01

Data Availability Statement

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [1] partner repository with the dataset identifier PXD033621. All other data that support the findings of this study are available from the corresponding author upon reasonable request.