Engineering cryoelectrospun elastin-alginate scaffolds to serve as stromal extracellular matrices

Abstract

Scaffold-based regenerative strategies that emulate physical, biochemical, and mechanical properties of the native extracellular matrix (ECM) of the region of interest can influence cell growth and function. Existing ECM-mimicking scaffolds, including nanofiber mats, sponges, hydrogels, and nanofiber-hydrogel composites are unable to simultaneously mimic typical composition, topography, pore size, porosity, and viscoelastic properties of healthy soft-tissue ECM. In this work, we used cryoelectrospinning to fabricate 3D porous scaffolds with minimal fibrous backbone, pore size and mechanical properties similar to soft-tissue connective tissue ECM. We used salivary glands as our soft tissue model and found the decellularized salivary gland (DSG) matrix to have a fibrous backbone, 10–30 μm pores, 120 Pa indentation modulus, and ~200 s relaxation half time. We used elastin and alginate as natural, compliant biomaterials and water as the solvent for cryoelectrospinning scaffolds to mimic the structure and viscoelasticity of the connective tissue ECM of the DSG. Process parameters were optimized to produce scaffolds with desirable topography and compliance similar to DSG, with a high yield of >100 scaffolds/run. Using water as solvent, rather than organic solvents, was critical to generate biocompatible scaffolds with desirable topography; further, it permitted a green chemistry fabrication process. Here, we demonstrate that cryoelectrospun scaffolds support penetration of NIH 3T3 fibroblasts 250 to 450 μm into the scaffold, cell survival, and maintenance of a stromal cell phenotype. Thus, we demonstrate that elastin-alginate cryoelectrospun scaffolds mimic many structural and functional properties of ECM and have potential for future use in regenerative medicine applications.

Graphical Abstract

1. Introduction

The extracellular matrix (ECM) comprises the proteinaceous scaffold within the connective tissue that plays a crucial role in regulating the function of parenchymal tissue in organs. Many pathologies in the human body are promoted by a diseased stroma^1–3 and are accompanied and/or caused by ECM stiffening, including fibrotic diseases and cancer^4,5. In advanced stages of many fibrotic diseases, which contribute to up to 45% of deaths worldwide⁶, a fibrotic stroma ultimately leads to loss of organ function⁷. The conversion of tissue-resident stromal cells into myofibroblasts is thought to drive fibrosis in many organs⁸. The myofibroblasts produce excess levels of ECM proteins, creating a matrix that is stiffer and denser than a homeostatic matrix, which contributes to disease progression^6,9. Preventing or reversing the conversion of tissue-resident stromal cells into myofibroblasts is one possible strategy for therapeutic remediation of fibrotic diseases. The delivery of mesenchymal stromal cells (MSCs) into diseased organs in mice has shown promise in preventing myofibroblast conversion, remediating disease, and improving organ function⁸. MSCs can be delivered through an intravenous injection but show poor engraftment and transient therapeutic effects¹⁰. Scaffolds that mimic soft-tissue stromal ECM can help in both localizing the MSCs at the organ of interest and preventing their conversion into myofibroblasts by providing key mechanical and biochemical cues. Hence, compliant scaffolds that can maintain the phenotype of MSC-like cells in vitro may have future applications in regenerative medicine to slow or reverse fibrotic disease.

Soft-tissue organs have unique ECM compositions that enable tightly regulated biochemical and mechanical properties for specific lineage commitment and differentiation during organ development and maintenance of cellular function in adult organs. Healthy soft-tissue stromal ECM is composed of an insoluble backbone of ECM proteins and soluble hydrogel-forming glycosaminoglycans (GAGs) of varying concentrations, depending on the organ. The mechanical and biochemical cues from the ECM (e.g., composition, topography, pore size, porosity, viscoelasticity) modulate cell viability, growth, homeostasis, migration, and differentiation. For example, the topography, pore size, and porosity modulate the amount of ECM backbone material interacting with the cell and thereby affect cell viability and growth^11,12. The matrix viscoelastic properties and the extent of cell attachment to the matrix regulate the cell-generated traction forces and the substrate resistance that the cells experience in response, which impact cell phenotype and differentiation^13–18. As matrix mechanics drive cell health and disease progression, artificial matrices for in vivo implantation and in vitro culture must closely emulate the native, healthy ECM to maintain cellular function as expected in a healthy organ.

Synthetic scaffolds of various types, including nanofiber mats, sponges, hydrogels, and nanofiber-hydrogel composites, have been engineered to simulate various aspects of ECM. Nanofiber mats have fibrous topography, impenetrable pores, and high stiffness, typically in the MPa range or higher^19,20, which make them good candidates for basement membrane mimetics for monolayer epithelial or endothelial growth^21–24. However, they fail to mimic both the 3D topography and the stiffness of soft-tissue stromal ECM within the connective tissue compartment. Sponges fabricated by freeze-drying, particulate- or salt-leaching, gas foaming, or phase separation have excessive ECM backbone and excessively large pore sizes (> 100 μm) when stiffness is in the sub-kPa range; hence, cells attach flush against the backbone, similar to 2D culture¹³. Hydrogels allow tunable stiffness in the kPa range for soft tissue scaffolds. However, bulk hydrogels, e.g., alginate, poly(ethylene glycol) (PEG), lack an insoluble fibrous backbone that mechanically supports cells. Their extremely small pore sizes, in the submicron range, support molecular movement but, without cell attachment sites, impede cellular movement crucial for cell migration and organization. While existing hybrid nanofiber-hydrogel scaffolds reinforce hydrogels with a fibrous backbone, they may have a non-homogenous distribution of cell anchorage points²⁵. Hence, a new fabrication strategy is necessary to produce scaffolds that concurrently mimic these essential properties of soft-tissue stromal ECM, including porous topography with 10–50 μm pores^26–29, minimal insoluble fibrous backbone, soluble hydrogel cushion, and sub-kPa range stiffness (~100–600 Pa)^30–34.

Cryoelectrospinning, also known as cryogenic electrospinning, low temperature electrospinning, and cold plate electrospinning, is an emerging technique to fabricate 3D nanofibrous scaffolds with high porosity and low bulk stiffness. The cryoelectrospinning process includes a cold collector plate, maintained at a temperature less than 0 °C^35–37 (figure S1), to collect deposited electrospun scaffolds, which are then lyophilized, a major difference from traditional electrospinning techniques. The cryogenically cooled collector plate promotes deposition of atmospheric water vapor as ice crystals on the collector plate and between deposited fibers. Upon contacting the collector plate at sub-zero temperature, the aqueous electrospun solution freezes to form ice crystals as well. When homogenous solutions with multiple components freeze, the solutes and solvent undergo phase separation, and the frozen water can subsequently be lyophilized to produce porous scaffolds. The nucleation of ice crystals allows scalable 3D growth, increased porosity, and consequentially, reduced scaffold density and kPa-range bulk stiffness³⁸. The remainder of the cryoelectrospinning process is similar to traditional electrospinning, in which a viscous, long-chain polymer solution becomes electrically charged when passed through a needle tip at high voltage (5–20 kV), and the electrostatic repulsion in the charged fluid exceeds the forces of surface tension, thereby pulling the charged droplet at the needle tip to the nearest electrical ground. The chain-chain interactions of the polymer molecules in the charged droplet prevent the droplet from falling to the electrically grounded collector plate as a drop and instead pull the droplet as a fiber with micro/nanometer diameter to the electrical ground. While the fibers deposit to form a 2.5D nanofiber mat in traditional electrospinning, when cryoelectrospun, they form a 3D fibrous scaffold following lyophilization. As shown in Table S1, cryoelectrospinning has primarily been explored with synthetic polymers (e.g., PLA^35,39–42, PLGA^37,43, PCL^38,41,44,45, PEU⁴³, PF⁴¹) dissolved in toxic organic solvents or glacial acetic acid, but natural polymers, such as silk fibronin^36,44,46, have also been used. Cryoelectrospun scaffolds have been shown to exhibit stiffnesses in the kPa or MPa range. While synthetic polymers are easier to manipulate, they do not provide biochemical cues that can be recognized by cells and non-biodegradable polymers require further chemical modification to be biodegradable^47,48. Additionally, organic solvents, even at low residual amounts, can have toxic effects on cells⁴⁹ and when used in combination with natural proteins, may alter the conformation of the proteins. We hypothesized that cryoelectrospinning of natural biomaterials, comprising ECM proteins and hydrogel materials with water as the solvent could yield biocompatible scaffolds that mimic the topography and viscoelasticity of soft-tissue stromal ECM.

In this study, we aimed to fabricate a scaffold that resembles the natural stromal ECM that supports a homeostatic mesenchymal phenotype. We used the salivary gland as our model soft-tissue organ and compared the topography and viscoelastic properties of the cryoelectrospun scaffold to a decellularized adult salivary gland matrix (DSG). We developed cryoelectrospinning methods to fabricate scaffolds that are similar to the natural stromal ECM found in adult submandibular salivary glands. We focused on fabricating a pliable environment for morphogenesis and hence, used a hydrogel material, alginate, and the insoluble ECM protein, elastin, as key components of the scaffold to mimic the elastic and gelling ECM composition of soft tissues. We explored the effects of solvent and process parameters on the topography and growth of cryoelectrospun scaffolds, developing a green chemistry approach to produce biocompatible scaffolds of unique porous topography with a fibrous backbone, resembling the structure of connective tissue ECM in soft tissues^26–28, which has not been previously reported with the cryoelectrospinning technique. We evaluated the ability of elastin-alginate cryoelectrospun scaffolds with the desirable topography (CES) to maintain stromal cell populations, by analyzing mesenchymal cell penetration and growth, phenotype, 3D organization, and function, focusing on the potential of CES to support healthy stromal growth and phenotype. ECM-mimicking CES scaffolds thus have potential for future in vivo delivery of stromal cells for tissue regeneration applications.

2. Materials and Methods

2.1. Materials

Scaffolds were fabricated using a soluble form of bovine neck elastin (ES12) from Elastin Products Company (Owensville, MI), alginate, and polyethylene oxide with a molecular weight of 400 kD (PEG-400 kD) from Sigma-Aldrich (St. Louis, MO), 85:15 poly(lactic-co-glycolic acid) (PLGA) (Cat. No. B6006–1) from DURECT Corporation (Cupertino, CA), and hexafluoroisopropanol (HFIP) from Sigma Aldrich. The reagents for crosslinking the scaffold were N-hydroxysuccinimide (NHS) from Thermo Fisher Scientific (Waltham, MA), ethyl dimethylaminopropyl carbodiimide (EDC), and calcium chloride dihydrate from Sigma-Aldrich. The reagents for cell culture were DMEM (high glucose), fetal bovine serum (heat-inactivated), Penicillin-Streptomycin (10,000 units/mL of penicillin and 10,000 μg/mL of streptomycin) from Thermo Fisher Scientific, or Antibiotic-Antimycotic Solution (10,000 units/mL penicillin, 10,000 μg/mL streptomycin and 25 μg/mL amphotericin B) from R&D Systems. Well-plates were coated with ultra-low adhesion polymer Lipidure from Amsbio (Cambridge, MA). Cell viability assays were performed with calcein-AM and ethidium homodimer from Sigma-Aldrich. Cell proliferation assays were performed using Cell Titer Glo-3D reagent from Promega (Madison, WI). Primary antibodies used for immunocytochemistry are detailed in Table S2, including antibodies against collagen I and collagen IV from MilliporeSigma (Burlington, MA), and perlecan from Santa Cruz Biotechnology (Dallas, TX) used for immunohistochemistry of decellularized salivary gland, and vimentin (clone LN-6) from Sigma-Aldrich and α-smooth muscle actin (α-SMA) from MilliporeSigma for cell culture samples. Secondary antibodies used were Alexa Fluor-488 AffiniPure F(ab’)₂ Fragment IgG, Cyanine Cy3 AffiniPure IgG, Alexa Fluor-647 AffiniPure F(ab’)2 Fragment IgG, and Alexa Fluor-488 AffiniPure F(ab’)₂ Fragment IgM from Jackson ImmunoResearch Laboratories (West Grove, PA). Other reagents used for immunocytochemistry include paraformaldehyde, Tween 20, bovine serum albumin, and phalloidin-rhodamine from Thermo Fisher Scientific, glutaraldehyde, Triton X-100, sodium chloride, and 4′,6-diamidino-2-phenylindole (DAPI) from Sigma-Aldrich, normal donkey serum (Cat. No. 017-000-121) from Jackson ImmunoResearch Laboratories, and Fluoro-Gel mounting medium from Electron Microscopy Sciences (Hatfield, PA). Triton X-100 from Sigma-Aldrich, NH₄OH and 1X phosphate buffered saline, pH 7.4 (PBS) from Thermo Fisher Scientific, and DNase I from StemCell Technologies (Vancouver, CA) were used in preparation of decellularized salivary glands for immunohistochemistry. Confocal imaging of decellularized salivary glands were performed using 50 mm glass bottom dishes (Cat. No. P50G-1.5–14F) from MatTek (Ashland, MA). Reagents used for the preparation of cell culture samples and decellularized salivary glands for scanning electron microscope (SEM) imaging include glutaraldehyde, sucrose, phosphate buffer, and hexamethyldisilazane (HMDS) from Sigma-Aldrich, and ethanol from Decon Labs (King of Prussia, PA).

2.2. Animals

Mice used to source salivary glands were either CD-1 strain from Charles River Laboratories (Wilmington, MA) or C57B6 strain from Jackson Laboratories (Bar Harbor, ME). The care and handling of mice and tissue collection conformed to the requirements of and was approved by the Institutional Animal Care Use Committee (IACUC) of University at Albany, State University of New York.

2.3. Decellularization of salivary glands

Whole organs were resected from adult female CD-1 or C57Bl/6 mice. Decellularization of salivary glands was performed using a modification of a protocol developed for lung tissue⁵⁰. A pair of salivary glands were rotated via inversion at 4 °C in 40 mL sterile distilled water for 2 days in a 50 mL conical tube with water removal and replacement after one day. After water-induced lysis was complete, the water was replaced with 40 mL clearing solution (0.5% Triton X-100, 0.05% NH₄OH) and tumbled for one additional day at 4 °C. Decellularized samples were washed three times in PBS. DNA was removed from these samples using 0.5 mg/mL DNase I in PBS at room temperature for 30 minutes. These decellularized salivary glands were stored at 4 °C in preservation medium composed of DMEM/F12, 10% fetal bovine serum, and 1% penicillin-streptomycin (10,000 U/mL).

2.4. Immunofluorescent staining and imaging of decellularized glands

Whole decellularized salivary glands were immunostained for collagen I, collagen IV, and perlecan using 500 μL of diluted antibody solution/gland (Table S2). All primary antibodies were incubated overnight at 4 °C, followed by three washes using PBS. Secondary antibody incubations were at least 4 hours at room temperature. DAPI was used for nuclei staining. Immunostained glands were imaged in 50 mm glass-bottom dishes. Imaging was performed using an EVOS^® FL Cell Imaging System (Thermo Fisher Scientific) with the same exposures and/or laser configurations for all samples within an experiment.

2.5. Scanning electron microscopy (SEM)

Decellularized salivary glands and scaffolds seeded with cells were fixed with 4% paraformaldehyde-0.25% glutaraldehyde in 5% (w/v) sucrose and 0.6X PBS for 4 hours and 20 minutes, respectively, followed by 3% glutaraldehyde in 0.1 M sucrose-0.1M phosphate buffer (pH 7.4) for 2 hours. Samples were then washed in 0.1 M sucrose-0.1M phosphate buffer three times for 10 minutes each. We chose the lower cost, more accessible HMDS drying rather than the classical critical point drying method because HMDS drying preserves the ultrastructure as well as critical point drying⁵¹ while minimizing artifactual mesh holes associated with critical point drying⁵². Briefly, samples were dehydrated in graded ethanol series incubation of 25, 50, 70, 80, 95, 100, 100% for 15 minutes at each ethanol concentration. Samples were subsequently chemically dried at 3:1, 1:1, and 1:3 ethanol: HMDS for 15 min each and then in 100% HMDS thrice for 15 minutes each time. Samples were allowed to air dry overnight. Chemically dried biological samples and lyophilized scaffolds were sputter-coated with iridium-palladium for imaging. SEM imaging was performed using a Zeiss Leo 1550 field emission scanning electron microscope (Zeiss Leo Electron Microscopy Ltd., Cambridge, UK).

2.6. Scaffold fabrication and modification

Cryoelectrospinning was performed using protein-polymer solutions of 1% elastin, 1.5% alginate, and 3% PEG-400 kD in deionized water or 4% elastin and 4% PLGA in HFIP. The protein-polymer solution of choice was loaded into a 3-mL syringe. The syringe was connected to non-conductive perfluoroalkoxy tubing, which was, in turn, connected to a 25G needle. The collector plate was placed in a Styrofoam box and surrounded by adequate amounts of dry ice and ice to reach specific collector plate temperatures. The Styrofoam box was placed inside a repurposed cell culture incubator at 25 °C with a water pan to maintain humidity levels inside the chamber. The 25G needle was connected to a high voltage power source (Gamma High Voltage Research, Ormond Beach, FL) and the collector plate to the electrical ground. The fabrication was conducted at 17 kV needle voltage, 10 μL/min syringe flow rate, and 15 cm needle tip-to-collector spacing for 1 hour. The low flow rate of 10 μL/min and the small diameter of the 25G needle enabled electrospinning, rather than electrospraying, at high voltage. After 1 hour, the collector plate with the scaffold was immediately transferred to a lyophilizer (FreeZone freeze drier, Labconco, Kansas City, MI) and lyophilized for 2–3 hours.

The lyophilized elastin-alginate-PEG scaffolds were individually crosslinked in a 96-well plate with EDC and NHS crosslinking solution to stabilize the scaffold with preferential amide bond formation between elastin and alginate chains and possible ester bond formation. PEG-400 kD does not have pendant groups that can be crosslinked, and therefore, dissolves away in water. The crosslinking solution was prepared by dissolving 1.48 mg EDC and 1.78 mg NHS per 100 μL of 95% ethanol per scaffold. Scaffolds were rocked in crosslinking solution at 45 rpm for 2 hours, followed by a series of graded ethanol washes with 95, 70, 50, and 0% ethanol with 1.5% CaCl₂ for 15 min each to wash away residual EDC and NHS, and simultaneously ionically crosslink the alginate chains. The scaffolds were then frozen at −80 °C and lyophilized for 4 hours.

Elastin-alginate nanofiber (NF) mats were fabricated by traditional electrospinning using 1% elastin, 1.5% alginate, and 3% PEG-400 kD in deionized water in a process similar to the cryoelectrospinning process described above, except that the collector plate was maintained at room temperature, and the relative humidity levels were maintained below 35% using dehumidified air input.

Elastin-alginate freeze-dried scaffolds were fabricated using a 1% elastin-1.5% alginate solution in deionized water. 50 μL of the solution was added to each well of a 96-well plate and frozen at −80 °C overnight. The frozen scaffolds were lyophilized for 4 hours. The lyophilized scaffolds were crosslinked, following a similar procedure to that of cryoelectrospun scaffolds as described above.

All scaffolds were UV sterilized, soaked in 70% ethanol for 30 min, washed with 0.9% NaCl for 10 min and then hydrated in cell culture medium with 10% fetal bovine serum (FBS) and 5% Antibiotic-Antimycotic Solution (penicillin-streptomycin-amphotericin B) overnight before cell culture.

2.7. Cell culture

Mouse embryonic NIH 3T3 fibroblasts⁵³ of passage 12–17 were maintained in DMEM (High Glucose) medium containing 10% FBS and 1% penicillin-streptomycin. The NIH 3T3 fibroblasts were incubated in a 37 °C, 5% CO₂ humidified incubator and subcultured on day 3 or 4 when they were 70–80% confluent.

2.8. Cell culture on scaffolds

Mouse embryonic NIH 3T3 fibroblasts⁵³ were seeded at 75,000 cells/scaffold in 25 μL DMEM (high glucose) medium containing 10% FBS, 1% penicillin-streptomycin, and 25 mM CaCl₂ in ultra-low adhesion, polymer-coated, round-bottom, 96-well plates and incubated on a rotary shaker at 30 rpm for 2 hours to facilitate cell attachment to the cryoelectrospun scaffolds. Cell culture media were supplemented with 25 mM CaCl₂, a concentration at which cell culture was not negatively impacted, to prevent rapid disintegration of the scaffold. After two hours, each well was supplemented with 175 μL of fresh medium, and the well plate was incubated with rotary shaking for another 22 hours to increase the cell attachment efficiency. Well plates were transferred onto a static surface 24 hours after cell seeding. Cells were cultured on scaffolds for up to 7 days. Cell culture media were replenished every day by removing 150 μL of spent medium from each well and replenishing it with 200 μL of fresh media to avoid nutrient depletion and to retain certain amount of the conditioned medium.

2.9. Immunocytochemistry and confocal imaging of cell culture samples

Samples were fixed in 4% paraformaldehyde-0.25% glutaraldehyde in 5% (w/v) sucrose, 0.6X PBS for 15 minutes, permeabilized with 0.1% Triton X-100 in 1X PBS for 15 min and blocked with 20% donkey serum-3% bovine serum albumin in wash buffer (0.9% NaCl-50mM CaCl₂ in deionized water) for 2 hours at room temperature. Samples were then incubated with rhodamine phalloidin to reveal cytoskeletal F-actin and/or primary antibody against vimentin at 4 °C overnight, followed by incubation at room temperature for 2 hours with secondary antibody and DAPI to reveal the nuclei within the total cell population. Antibody combinations and concentrations used are detailed in Table S2. Samples were then mounted with Fluoro-Gel mounting medium for imaging. Confocal imaging was performed using a Leica SP5 confocal laser scanning microscope (Leica Microsystems, Mannheim, Germany).

2.10. Indentation testing

Mechanical properties including indentation modulus and sample viscoelastic relaxation time were determined using a micro-indentation tester (CellScale Biomaterials Testing, Ontario, Canada) as described previously⁵⁴. Briefly, samples were glued to a glass slide, immersed in a PBS bath, and then indented/loaded using a 3-mm spherical bead attached to a cantilever. Samples were deformed at a constant displacement rate of 4 μm/s. Upon reaching an indentation depth of 10% of the initial sample height, samples were held in their deformed state for up to 350 s (hold phase) and then allowed to relax by removal of the indentation force. Force (F) and displacement (δ) of the cantilever tip were measured as a function of time. The indentation modulus was determined as the elastic modulus (E_i) from the Hertz model by fitting the data from the loading region of the force-displacement curve measured by indentation to the Hertz contact equation for a spherical indenter $(F=\frac{4}{3}\frac{E_i{R}^{\frac{1}{2}}}{1-v^2}{\delta }^{\frac{3}{2}})$. Here, R is the radius of the spherical indenter (1.5 mm) and ν is the Poisson ratio of the sample (set at 0.49 to represent elastic, almost incompressible hydrogel materials⁵⁵). The matrix relaxation half time is a measure of the viscoelastic nature of a sample and is evaluated by its stress relaxation response. The stress relaxation response is observed during the hold phase of the force-time curve where the sample relaxes towards an equilibrium state, and the loading force required to maintain a constant strain reduces and reaches a steady-state value. The relaxation half time was computed from the stress relaxation response of the samples as the amount of time required for the stress/loading force to reach half of its peak value while maintaining a constant strain equal to 10% of the initial sample height.

2.11. LIVE/DEAD assay

Cell-scaffold constructs were incubated with 2 μM calcein-AM and 4 μM ethidium homodimer for 25 min at 37 °C and imaged using Leica SP5 confocal laser scanning microscope (Leica Microsystems, Mannheim, Germany) to reveal live cells in green and dead cells in red fluorescence. Quantitative analysis of live and dead cells was performed using ImageJ⁵⁶. The images were opened in ImageJ, and two separate images were obtained for live and dead cells by using the ‘Split Channels’ feature in Image>Color menu. The threshold of each image was adjusted. The numbers of live and dead cells were quantified by using the ‘analyze particles’ feature under the Analyze menu, setting the particle size range to 10–3000 μm², circularity to ‘0–1’, and excluding particles on edges.

2.12. Cell proliferation assay

Cell-Titer Glo^® 3D Viability Assay was performed to evaluate cell proliferation at 1, 4, and 7 days after cell seeding onto scaffolds per manufacturer’s instruction. Briefly, cell-scaffold constructs in the 96-well plate and Cell-Titer Glo^® 3D Viability Assay reagent were equilibrated to room temperature for 30 min. After samples were gently washed with 1X PBS, 75 μL of cell culture media and 75 μL of CellTiter-Glo 3D reagent was added. The contents of the well were vigorously pipetted up and down to disintegrate the scaffold and release the cells. The well plate was then shaken on a rotary shaker at 120 rpm, at 37 °C for 15 min, and then incubated at room temperature for 30 min to stabilize the reaction. 50 μL of the reaction mixture from each well was transferred into a 96-well white luminescence plate, diluted with 50 μL of cell culture media, and mixed well. Luminescence was determined using a Tecan Infinite 200 plate reader (Tecan US, Morrisville, NC).

2.13. Scaffold pore size analysis using ImageJ

SEM images of cryoelectrospun scaffolds and fluorescence microscopy images of decellularized salivary glands were analyzed for pore size using ImageJ’s ‘analyze particles’ feature. The threshold for each image was adjusted (Image>Adjust>Threshold) to identify the pores from the scaffold background. Under the ‘Analyze’, ‘Set Measurements’ menu, the ‘fit ellipse’ feature was selected to identify the major and minor axis diameter of the pores. Pore size was analyzed by using the ‘analyze particles’ feature by setting the particle size range to 10–3000 μm², circularity to ‘0–1’, and excluding particles on edges.

2.14. Statistical Analysis

Data are presented as mean ± standard deviation. All in vitro cell culture and material characterization experiments were performed in triplicate, unless otherwise indicated. One-way ANOVA followed by Tukey’s post hoc test was performed using GraphPad Prism 9.2.0. p < 0.05 was considered significant.

3. Results

3.1. Decellularized salivary gland ECM exhibits porous topography with a fibrous backbone

To fabricate matrices with physiologically relevant ECM topography, we first examined the native ECM topography in decellularized female adult mouse submandibular salivary glands. We decellularized adult mouse submandibular salivary glands following a modified decellularization protocol developed for the lung, a similarly structured organ⁵⁰. An additional DNase treatment was necessary to remove lingering DNA adhering to the ECM (figure S2). SEM imaging of the cross-section of the decellularized salivary gland matrix (figure 1A) showed a fibrous basement membrane surrounding acinar epithelium (figure 1B) and as linear fibers and fiber bundles adjacent to ductal epithelium (figure 1C) and in interlobular regions (figure 1D). Stromal ECM is expected to surround the epithelial basement membrane; however, it was not abundantly visible, possibly due to removal of GAGs, the predominant soluble ECM components, during the decellularization process (figure 1E). Further, stroma content is higher during morphogenesis but is significantly reduced in adult tissue⁵⁷. While SEM reveals all structures in the ECM, it does not reveal the arrangement pattern of specific ECM proteins. To further illustrate the organization of specific ECM proteins, we immunostained the decellularized salivary glands for predominant ECM proteins collagen I, collagen IV, and heparan sulfate proteoglycan (perlecan) and viewed their arrangement by fluorescence microscopy. The ECM proteins were arranged in a porous topography with a fibrous backbone, forming 10–30 μm pores with delicate winding fibers (figure 1F). To evaluate existing scaffolds as suitable scaffolds to support mesenchymal or stromal cells, we compared the topography of the decellularized salivary gland ECM with current ECM mimics including nanofiber mats, sponges, hydrogels, or hydrogel-nanofiber composites. None of these currently available ECM-mimetic scaffolds truly recapitulate the topography, pore size, and stiffness of the native ECM of adult mouse salivary glands. Therefore, we set a benchmark to design and develop a scaffold with porous topography with a minimal fibrous backbone, pore size ~30 μm, and stiffness ~120 Pa to mimic salivary gland stromal ECM for stromal cell delivery.

Figure 1.

Salivary gland ECM topography analyses. (A-E) Scanning electron microscopy images of a cross-section of decellularized adult mouse salivary gland. Scale bar = 20 μm. A) Zoomed out image of overall cross-section depicting the acinar and ductal regions. B) Basement membrane of acinar epithelium. C) acinar clusters. D) Interlobular region. E) Stromal region. F) Immunocytochemistry of ECM protein expression in decellularized adult mouse salivary gland matrices showing microporous structural arrangement of ECM. Decellularized ECM immunostained with collagen I (COL-I, green), Collagen IV (COL-IV, red), and perlecan (grey) and merged (yellow). Scale bar = 20 μm.

3.2. Cryoelectrospinning of elastin-alginate with water as solvent produces 3D porous scaffolds

The choice of ECM proteins for scaffold fabrication is crucial since there are ~300 ECM core proteins⁵⁸, each providing unique biochemical and/or mechanical triggers to cells in the matrix. Of the 300 ECM proteins, collagen, elastin, and fibronectin have well-characterized functions in soft-tissue ECM⁵⁹. Morphogenetic environments require a pliable matrix for constant remodeling, and excess collagen increases the stiffness of the matrix, triggering a fibrotic phenotype in cells^60,61. Therefore, we chose not to include collagen, one of the primary components of the ECM of most tissues⁵⁹, but to use only elastin, a compliant protein, and alginate, a viscous hydrogel, as the biomaterials for fabrication, to emulate the viscoelastic nature of soft tissue ECM. We relied on serum in the cell culture medium for our initial fibronectin source. We also relied on stromal cells to secrete their own collagen and other ECM proteins onto the cryoelectrospun scaffolds.

To eliminate the use of toxic solvents for fabrication, we used water as the solvent, as both alginate and elastin can be dissolved in water. To delineate the effects of aqueous solvent on cryoelectrospinning, we compared cryoelectrospun scaffolds fabricated using an aqueous or an organic solvent. Our prior studies showed that traditional electrospinning of 4% elastin-4% PLGA in the organic solvent HFIP produced 2.5D mats with nanofibrous topography and improved elasticity over PLGA alone^62–64. Solutions for electrospinning require long-chain polymers to facilitate chain entanglement, which permits the deposition of fibers instead of microdroplets^65–67. While we used PLGA for this purpose in our previous work, PLGA also increases the stiffness of the scaffolds into the MPa range. Hence, we instead included PEG-400 kD to the 1% elastin-1.5% alginate solution at a concentration of 3% wt/v in water, to both facilitate chain entanglement essential for electrospinning and maintain the stiffness of the scaffolds low in the sub-kPa range. While we did not comprehensively explore material and solvent combinations for cryoelectrospinning, we compared cryoelectrospun scaffolds fabricated using 4% elastin-4% PLGA soluble in the organic solvent HFIP with ones fabricated using 1% elastin-1.5% alginate-3% PEG soluble in water, to analyze the effects of PLGA vs. PEG in the scaffold and the effects of an organic solvent and aqueous solvent. Our results showed that the elastin-PLGA cryoelectrospun scaffolds were very dense (figure 2A, B), and did not emulate the porous topography of salivary gland ECM (figure 1F). In contrast, we observed that cryoelectrospinning of elastin-alginate-PEG in water produced a taller, more porous, 3D scaffold when compared with elastin-PLGA electrospun in HFIP for the same fabrication duration (figure 2C vs. 2A). Further, we observed that at specific process conditions (detailed in section 3.3), the aqueous solvent facilitated the fabrication of CES that exhibited desirable porous topography with a minimal fibrous backbone and interconnected pores (figure 2D, 2E top left panel). We attempted to functionalize PEG-400 kD with carboxyl groups for crosslinking since PEG is a widely used hydrogel material; however, the length of the PEG polymer was a deterrent and prevented functionalization (data not shown). Therefore, PEG-400 kD was used solely for the purpose of increasing the electrospinnability of elastin-alginate solution. The PEG-400 kD dissolved away in water post EDC/NHS crosslinking due to the lack of cross-linkable pendant groups, yielding an elastin-alginate scaffold.

Figure 2.

Effects of solvent on topography and 3D growth of cryoelectrospun scaffolds. A,B) 2.5D Elastin-PLGA nanofibers fabricated with organic solvent HFIP. C,D) 3D Elastin-Alginate-PEG cryoelectrospun sponges fabricated with water as solvent. Water as solvent allowed 3D growth of elastin-alginate cryoelectrospun scaffolds with the desirable topography (CES) of high porosity, interconnected pores, and a minimal backbone. A,C) Bench top photos. Scale bar = 5 mm. B,D) SEM images. Scale bar = 10 μm. E) Effect of process parameters on scaffold topography. Scale bar =10 μm. Humidity > 35% and air temperature < 2 °C promote desirable topography (top left panel). Other combinations promote fibrous topography. Collector plate temperatures between −35 and −10 °C promote homogenous growth in X,Y, and Z dimensions.

3.3. High humidity and low air temperatures promote porous topography with minimal backbone in cryoelectrospun scaffolds

To delineate effects of collector plate temperature, chamber relative humidity, and air temperature on topography of cryoelectrospun scaffolds, we maintained two parameters constant, while modulating the third parameter. Atmospheric conditions had a substantial impact on the topography of the scaffolds by altering ice nucleation based on relative humidity levels in the electrospinning chamber. Collector plate temperature could also potentially impact scaffold growth and topography by affecting the rate of ice nucleation. We independently modulated the air temperature and humidity levels at collector plate temperatures above −35 °C. We observed that collector plate temperatures between −35 and −10 °C permitted homogenous scaffold growth (figure 2E), whereas air temperature and humidity affected the topography of these scaffolds. At collector plate temperatures ranging between −35 and −10 °C, when the air temperature was > 2 °C, irrespective of the relative humidity levels, cryoelectrospun scaffolds with fibrous topography (CES-F) resulted (figure 2E, right panels), whereas when the air temperature was between −10 and 2 °C and relative humidity levels > 35%, CES were generated, which showed desirable porous topography with a minimal fibrous backbone and interconnected pores (figure 2E, top left panel). Temperatures lower than −35 °C resulted in increased ice nucleation in the Z dimension instead of homogenous growth in the X, Y, and Z dimensions, frequently causing the scaffold to collapse onto itself (figure S3A) and showing heterogeneous scaffold growth (figure S3B). Therefore, we defined optimal conditions to fabricate CES as higher collector plate temperatures (between −35 and −10 °C), air temperatures between −10 and 2 °C, and a relative humidity > 35%.

3.4. Probe-array collector plate geometry promotes homogenous and distributed growth of cryoelectrospun scaffolds

Reproducibility and yield of a fabrication process are important factors affecting the scalability of scaffold production in future applications. To obtain homogenous, distributed scaffold growth, we explored different collector plate geometries. We hypothesized that homogenous distribution of the electric field over the collector plate could increase the yield of cryoelectrospun scaffolds. Hence, we designed a metallic probe-array collector plate, which increases the surface area in contact with the electrical ground, and compared scaffold growth on metallic probe-arrays with 3-mm (figure S4A,B) and 5-mm probe distancing (figure 3A) with that on a flat metal plate (figure 3B). Using COMSOL Multiphysics software, we simulated the electric field potential on the surface of the two types of collector plates used and observed the electric field potential to be focused on the probes (figure 3C) and to be approximately uniform over all the probes with a value of ~800 V (figure 3E), whereas the electric field potential was distributed over the flat plate (figure 3D) and varied across the working surface area from ~300 to ~800 V (figure 3F). This uniform potential over the probe-array collector plate allowed similar sized scaffolds (~3 mm in diameter and 2–3 mm in height) with the same topography to grow over most of the probes and hence, an overall homogenous scaffold growth over the probes (figure 3G) compared to that over the surface of the flat collector (figure 3H). Furthermore, the 5-mm probe-array (14×12 array, i.e., 168 probes) permitted a consistent yield of ~100 scaffolds/run in one hour, i.e., 60% of the total probes generated reproducible/usable scaffolds. The surface topography of these scaffolds remained relatively unaffected between the flat plate and 5-mm probe-array collector upon cryoelectrospinning at similar process parameters (figure 3I, J), demonstrating the capacity of the probe-array collector plate to homogenously generate 3D elastin-alginate scaffolds with a high yield. While the 3-mm probe-array collector allowed relatively homogenous growth of scaffolds in comparison with the flat collector, these scaffolds were too small to handle and difficult to remove from the collector. Furthermore, scaffolds fabricated on the 3-mm probe-array collector had mostly fibrous topography instead of the desired porous topography under similar process parameters (figure S4C, D), which might be caused by different electric field and ice nucleation profiles for the 3-mm probe-array compared with that of 5-mm probe-array. Hence, to homogenously fabricate CES, we chose the 5-mm probe-array collector.

Figure 3.

Effect of collector plate geometry on scaffold growth in X, Y, Z dimensions. A,B) Bench top photos of the collector plate. C,D) COMSOL simulation demonstrating the electric field line distribution over the collector plate. E,F) COMSOL simulation demonstrating the electric potential distribution over the collector plate. G,H) Bench top photos and I,J) SEM images of scaffolds grown on a metallic probe-array collector having 5-mm probe spacing (G,I) and flat collector (H,J), respectively. Scale bar = 20 μm. Metallic probe-array promoted distributed individual scaffold growth and increased growth in the Z dimension due to reduced electrical surface area in the X and Y dimensions. (A,C,E,G,I) Metallic probe-array collector plate. (B,D,F,H,J) Flat collector plate.

3.5. Elastin-alginate cryoelectrospun scaffolds resemble the topography and viscoelasticity of decellularized salivary gland ECM

To evaluate the ability of CES to emulate salivary gland ECM, we compared their topography and viscoelastic properties with that of decellularized adult salivary gland matrix (DSG) by SEM and indentation testing, respectively. The CES exhibited porous topography with pores of 10–25 μm, which was similar to that of DSG (figure 4A–C). We further determined the indentation modulus and relaxation half times of CES and DSG by micro-indentation testing. CES and DSG had very low indentation moduli of ~ 120 Pa, whereas elastin-alginate freeze-dried sponges (FS), fabricated with the same material composition (1% elastin-1.5% alginate) and crosslinked with the same materials (EDC/NHS) as CES, exhibited a significantly higher indentation modulus of ~ 850 Pa (figure 4D). Further, we observed that both FS and CES demonstrated similar stress relaxation dynamics as DSG (figure 4E), despite the differences in observed stiffness/indentation modulus. While the stress relaxation dynamics of FS, CES and DSG were similar, higher variance was observed in CES and DSG, possibly due to differences in the nature of fiber bonds, pore sizes and spatial heterogeneity, which can impact the reorganization of fibers during relaxation.

Figure 4.

Elastin-alginate cryoelectrospun scaffolds (CES) mimicking decellularized adult salivary gland ECM in topographic and viscoelastic properties. A) SEM image of CES and B) fluorescence image of decellularized adult salivary gland matrix (DSG) showing topographical similarities. Scale bar = 20 μm. C) Pore size analysis using ImageJ. Comparison of D) compression modulus and E) relaxation half time of 1% elastin-1.5% alginate freeze-dried sponges (FS), 1% elastin-1.5% alginate-3% PEG CES and DSG showing that CES exhibited viscoelastic properties similar to DSG. ***, P<0.0001; ns, not significant.

3.6. CES promote 3D stromal growth and survival of NIH 3T3 fibroblasts

To investigate the ability of CES to support viable 3D stromal growth, NIH 3T3 fibroblasts, a well-established model mesenchymal cell line, were grown on CES for 24 hours followed by Live/Dead staining to determine the viability of attached cells. The majority of the cells (89% ± 2%) attached to CES were viable after 24 hours (figure 5A, B). The cells on CES formed 3D clusters (figure 5A) with penetration depths ranging between ~250 to 450 μm, with an average of 359 ± 96 μm (figure 5C). Furthermore, CellTiter Glo-3D viability assay was performed to evaluate cell growth on CES-H. The viable cell number on the scaffold increased slightly (but not significantly) from day 1 to 4 and then leveled off, confirming cell survival on the CES for 7 days (figure 5D).

Figure 5.

Elastin-alginate cryoelectrospun scaffolds with desirable topography (CES) promote viable 3D cell penetration and growth. A) Confocal images of Live/Dead stained NIH 3T3 fibroblasts on CES showing that majority of attached cells are viable and penetrate deep into the scaffold to form 3D clusters. B) Quantification of cell viability on CES after 24 hours using ImageJ. C) Quantification of cell penetration revealing that NIH 3T3 cells infiltrate into CES with depths ranging between 250 to 450 μm and have an average infiltration depth of 359 ± 96 μm. D) Time course of NIH 3T3 fibroblast growth on CES using Cell Titer-Glo^® 3D Cell Viability Assay.

3.7. CES promote typical adherent fibroblast morphology and stromal phenotype of NIH 3T3 fibroblasts

To evaluate the ability of CES to support typical adherent morphology, we cultured NIH 3T3 fibroblasts on traditionally electrospun elastin-alginate nanofiber (NF) mats, CES, and CES-F (figure 6). To analyze adherent cell morphology, NIH 3T3 cells were cultured on these scaffolds for 1 day followed by SEM imaging. NIH 3T3 fibroblasts remained rounded on NF mats (figure 6A, top panel); however, they demonstrated a spread-out, fibroblast morphology on both CES (figure 6A, middle panel) and CES-F (figure 6A, bottom panel). Maintenance of the characteristic adherent morphology for NIH 3T3 fibroblasts grown on CES (figure 6B, middle panel) and CES-F (figure 6B, bottom panel) was confirmed by visualizing the distribution of cytoskeletal F-actin (in red) and vimentin (in green). However, very few cells attached and grew on NF mats or exhibited elongated F-actin stress fibers (red) (figure 6B, top panel) because traditionally electrospun alginate nanofibers lose their fibrous structure after hydration with medium, forming a bulk hydrogel-like structure (figure S5A), and the alginate hydrogel does not have any adhesion sites for cell attachment and hence, showed low cell viability (figure S5B). NIH 3T3 fibroblasts formed 3D clusters with notable infiltration in CES (figure 6C, middle panel), but only a cell sheet on CES-F (figure 6C, bottom panel), demonstrating that the porous topography with minimal backbone and interconnected pores of CES is critical for infiltrated cell growth and 3D cell distribution. Additionally, maintenance of vimentin (in green) was observed in NIH 3T3 fibroblasts grown on CES over the course of 7 days (figure 6D), suggesting that CES support maintenance of stromal phenotype.

Figure 6.

Effects of scaffold properties on morphology, 3D organization, and marker expression of NIH 3T3 fibroblasts. A) SEM images showing the effect of scaffold topography on morphology of NIH 3T3 fibroblasts. NIH 3T3 fibroblasts maintain spread-out morphology on cryoelectrospun scaffolds with desirable topography (CES, middle) and cryoelectrospun scaffolds with fibrous topography (CES-F, bottom), but remaining isolated and rounded on traditionally electrospun nanofibers (NF, top) on day 1. Arrows denote individual cells. B,C) Confocal images of top view (B) and side view (C) of F-actin cytoskeleton organization (in red) and vimentin expression (in green) showing the effect of scaffold topography on attachment and infiltration of NIH 3T3 fibroblasts as well as mesenchymal marker expression on day 4. D) Confocal images showing maintenance of mesenchymal marker vimentin expression (green) over 7 days in NIH 3T3 fibroblasts grown on CES-H. Blue, DAPI-stained nuclei to show total cell population. Scale bar = 25 μm.

4. Discussion

Several pathologies arise in the human body due to an imbalance in the biochemical and mechanical cues delivered by the ECM to cells in organs^68–73. Hence, in addition to cell engineering strategies for regenerative medicine and in vitro culture purposes, scaffold engineering strategies are also critical. There is a pronounced need to engineer tissues that mimic not only the biochemical cues but also the physical and mechanical cues of healthy connective tissue ECM^13–17. In this work, we developed a novel fabrication strategy to produce 3D scaffolds (figure 2C) that closely mimic the topography, porosity, and pore size of endogenous salivary gland ECM (figure 4A–C) and potentially the stromal ECM of other soft tissues, e.g., lung, liver^26–28. Our cryoelectrospun scaffolds exhibited pore sizes of 10–25 μm with indentation modulus of ~120 Pa, which mirror the viscoelastic and topographic properties of salivary gland ECM (10–50 μm, ~120 Pa). Although porous scaffolds can be generated using porogen leaching, thermally induced phase separation or gas foaming techniques, these fabrication processes typically generate large pores, more than 100 μm in diameter with large heterogeneity in pore size and poor pore interconnectivity^74–76, resulting in poor cell ingrowth/cell penetration⁷⁷ and differentiation of stromal cells into chondrogenic or osteogenic lineages^78–80.

The use of water as the solvent for cryoelectrospinning is key to generating a desirable porous topography with a minimal backbone and interconnected pores. The aqueous solvent, in combination with the atmospheric water vapor condensing and depositing on the collector plate at sub-zero temperature, increased ice nucleation, which improved 3D growth and porosity of the scaffold (figure 2A, C). In this freezing process, phase separation of solutes and aqueous solvent occurs wherein the ice crystals nucleate together, and solutes separate out unidirectionally to form a backbone, which results in the unique topography (figure 2D), which has not been previously reported with cryoelectrospinning.

The fabrication strategy we developed incorporates ECM proteins and hydrogel materials in one step, in contrast to other fabrication strategies, where fibrous components are fabricated and subsequently embedded into hydrogel materials^38,55,81 before crosslinking. The technique reported here should allow for electrospinning of any water soluble, ECM protein-hydrogel material combination with a long-chain polymer, such as PEG-400 kD, to ensure the electrical conductivity, viscosity, and chain entanglement required for electrospinning. Our chosen biomaterials of 1% elastin-1.5% alginate composition mimicked the elastic modulus and relaxation properties of native salivary ECM (figure 4D, E) and can be supplemented with additional ECM components to engineer custom microenvironments. Embryonic microenvironments are pliable and have a low elastic modulus of 50–300 Pa⁸², to facilitate morphogenesis, cell expansion, and migration. Hence, the low elastic modulus of the CES at ~120 Pa makes them promising candidates for biomimetic matrices to regenerate or model a range of soft tissues.

The complex effects of process parameters on the cryoelectrospinning process generate a need for reproducible and homogenous scaffold growth. We improved the scaffold consistency, homogeneity, and yield by establishing thresholds for process parameters and evaluating alternative collector plate geometries. We successfully confirmed our hypothesis that electric field homogenization would improve scaffold homogeneity and yield through COMSOL simulation and fabrication optimization (figure 3). The homogeneous electric field and the geometry of the metallic probe-array regulated individual scaffold growth, scaffold distribution and size, and increased the scaffold yield of each cryoelectrospinning run to > 100 scaffolds in one hour. While probe-array collector plates have been used in traditional electrospinning to improve the scaffold porosity^83–85, they have not been previously used for cryoelectrospinning for distributed and homogenous growth. Further, delineation of the effects of solvent, air temperature, relative humidity and collector plate temperature on the cryoelectrospun scaffold topography (figure 2E) allowed us to determine the boundary conditions for these parameters to reproducibly produce scaffolds with desirable topography.

To demonstrate cell growth, viability, and phenotype maintenance on CES, we cultured stromal mesenchyme NIH 3T3 fibroblasts on CES, CES-F, and conventional electrospun NF mats. We demonstrated viable 3D cell growth on the CES (figure 5A, B) and determined that CES and CES-F topography allowed stromal fibroblasts to maintain their characteristic morphology, while NF mats did not (figure 6A), emphasizing the role of scaffold architecture on 3D cell growth. The high porosity of CES and CES-F may have maintained scaffold topography despite the alginate swelling, which could have, in turn, permitted cell attachment, growth, and cell-cell interaction, in contrast to NF mats (figure S5). Therefore, even though CES, CES-F, and NF mats were fabricated using the same material composition (1% elastin, 1.5% alginate, and 3% PEG-400 kD), cryoelectrospun scaffolds such as CES and CES-F offered a topographical advantage for cell attachment and maintenance of cell morphology over traditional electrospinning.

One of the key factors influencing cell phenotype in vivo is cell communication through cell-cell contacts in 3D. We demonstrated that CES, but not NF mats or CES-F, permitted 3D cell cluster growth (figure 6), reminiscent of cellular organization and interaction in vivo. Though CES-F favored cell attachment and maintenance of cell morphology, cells randomly attached to the scaffold and grew as mono- or bilayers (figure 6B, C, bottom panel). CES supported cell growth in all orientations, deep into the scaffold; however, CES-F favored the growth of cells only in the horizontal orientation and as cell sheets. Hence, the topography of CES not only mimics native ECM structure and organization but also facilitates cell-cell interactions similar to tissues in vivo by promoting cell growth in random orientations and grouping of cells into clusters.

We also demonstrated that CES maintained the adherent morphology of NIH 3T3 fibroblasts and mesenchymal marker, vimentin expression for 7 days, demonstrating retention of mesenchymal phenotype. The similar viscoelastic properties between CES and healthy, native ECM might be the primary contributing factor for healthy growth and phenotype retention of the stromal cells grown on CES. These results demonstrate the feasibility of using ECM-mimicking CES to support healthy stromal growth and function, laying the foundation for future in vitro soft-tissue modeling and stromal cell delivery in vivo for tissue regeneration applications.

5. Conclusion

Overall, we developed a cryoelectrospinning process to bioengineer a 3D porous matrix with a minimal fibrous backbone and interconnected pores. Our cryoelectrospinning process, in combination with the biomaterials chosen, yielded a scaffold with topographical and viscoelastic properties similar to native salivary gland ECM. NIH 3T3 fibroblasts attached to the scaffolds and maintained their mesenchymal phenotype. We laid the foundation to develop a matrix that can be used for recapitulating the stromal microenvironment, demonstrating potential for in vitro organ modeling and in vivo regenerative therapy applications.

Data Availability

The authors declare that all data supporting the findings of this study are available within the paper and its supplementary information.