Mechanical and Structural Responses of Silk Fibroin Scaffolds to MechanoCulture T6^® Bioreactor Loading and Enzymatic Degradation

Abstract

Silk fibroin is a semicrystalline biopolymer derived from Bombyx mori cocoons that can be fabricated into various biomaterials including aligned porous scaffolds. Silk fibroin is an ideal biomaterial polymer due to its tunable pore sizes and mechanical properties, which match those of aligned soft tissues, as well as its biocompatibility, non-toxic effects, and tunable degradation. This in vitro platform was developed from ice-templated anisotropic silk fibroin scaffolds toward modeling aligned soft tissues using mechanical loading via a bioreactor. Different scaffold fabrication post-lyophilization parameters gave varied self-assembly of the amino acid building blocks of silk fibroin, as explored through crystalline structures and in vitro degradation. X-ray scattering revealed a longer-range order of the crystalline domains when scaffolds were subjected to slower post-lyophilization processing. This structural difference also manifested in different rates of enzymatic degradation, where protease XIV was able to cleave the amorphous regions between smaller crystalline domains more rapidly. MechanoCulture T6 bioreactor stimulation occurred for 5, 10, or 25 days at 1 Hz, 10% strain for 30 minutes with 11.5 hours of rest periods to mimic skeletal muscle stimulation for hypertrophy. Hydrated uniaxial rheology was used to assess Young’s modulus (E), ultimate tensile stress (UTS), and strain at break. The mechanical properties and internal porosity were found to be independent of storage, loading, and time. Scanning electron microscopy (SEM) and nano-computed tomography (nano-CT) showed minimal differences in scaffold structural properties after bioreactor loading. Ice-templated silk fibroin scaffolds were shown to be suitable for new approach methods requiring mechanical stimulation.

Graphical Abstract

1. Introduction

Silk fibroin, obtained from Bombyx mori silkworm cocoons, is a natural biopolymer used in the formation of biomaterials due to its biocompatibility, controllable degradation rate, and tunable physical properties.^1-3 Silk fibroin polymer solutions can be formed into various biomaterial formats including films, nanoparticles, scaffolds, and hydrogels.^{1, 3} Molecular weight, concentration, and crystallinity are all tunable parameters that affect the mechanical and physical properties of the resulting silk fibroin-based structures.^{1, 2, 4} Silk fibroin is known to form physical crosslinks via secondary structure formation within the protein, resulting in crystalline beta-sheet regions.^5-7 The organization of beta-sheet structures through hydrogen bonding yields crystalline and amorphous polymer regions, which decrease optical transparency and increase mechanical strength over time.^{7, 8} Crystalline beta-sheet structures can be intentionally induced by applying shear forces, adding salt, increasing the temperature, or through water vapor annealing. However, they will also form unintentionally as the biopolymers interact, aiming to minimize free energy.^{1, 5, 7, 9, 10} The quantity and organization of beta-sheet structures are tunable, resulting in controllable degradation rates, mechanical properties, and thermal properties as the ratio of amorphous and crystalline domains shifts.

Porous three-dimensional silk fibroin scaffolds can be formed through salt-leaching^{1, 11-20} or ice templating techniques.^21-30 Ice-templating is the process of freezing an aqueous silk solution, then lyophilizing to sublimate the ice, leaving a silk fibroin scaffold.²⁴ Pore shapes within these scaffolds are manipulated through the freezing step, resulting in isotropic pores or anisotropic channels.^{23, 24, 26, 31} To achieve various pore architectures, the freezing step is performed in a mold to produce anisotropic scaffolds^{23, 27, 31} or through freezing in a −20°C or −80°C freezer to produce an isotropic scaffold.^{4, 25, 26} Previous work has explored the parameter space of controlling the freezing rate, which is dictated by the freezing temperature, to assess anisotropic pore sizes.^{23, 31} Previous work with isotropic silk fibroin scaffolds has assessed tunable degradation rates^{4, 32, 33} and compressive mechanics for various fabrication parameters.^{26, 27} Initial work has begun to explore pore size tunability^{23, 31} and mechanical properties of aligned silk fibroin scaffolds,³¹ but degradation information has yet to be completed for anisotropic scaffolds. Previous efforts have aimed to understand the dynamics of these silk fibroin scaffolds prior to in vivo use, enabling an improved understanding of bioactive molecule delivery,^{27, 34} strategies to promote cellular infiltration,^{23, 25, 27} and timelines for tunable degradation^{2, 4, 33} as a function of fabrication parameters.

Silk fibroin-based biomaterials have been successful in applications involving cellular maturation, differentiation, and remodeling responses. Cellular responses can be tuned through modulation of mechanical properties, structural organization, and bioactive molecules. Several examples have shown differentiation of stem cells due to structural properties,^{16, 28} as well as migration and phenotype shifts due to biological cues.^{7, 27, 34} Specifically, initial in vitro work with anisotropic ice-templated silk fibroin scaffolds has examined the extent of the inclusion of cell adhesion cues from extracellular matrix components²⁷ and tunable mechanics^{27, 31} relevant to cell culture applications. Recent work has begun to address the scalability of aligned scaffold volumes and fundamental control of physical properties through secondary structure crystallite size.³¹ Fundamental understanding of scalability and dynamic temporal changes to these scaffolds will provide avenues for in vitro assessments, including the development of three-dimensional disease models (Duchenne’s muscular dystrophy, rare diseases)^{35, 36} and pharmaceutical development (drug testing)^37-39 prior to in vivo studies. For example, silk fibroin porous scaffolds provide the necessary structural cues for key long-term in vitro applications, including mimicry and support of muscle growth, enabling cardiac²⁷ and skeletal muscle^{31, 40, 41} tissue studies.

Skeletal muscle is a dynamic organ that is prone to atrophy or hypertrophy depending on total signaling and loading to the system.^42-44 Mechanical loading and motion are essential to skeletal muscle growth to stimulate signal cascades in the mechanotransduction pathways.^{42, 45-47} While short-term studies in materials such as collagen type I hydrogels can provide insight into cell behavior over short times, the development of new approach methodologies (NAMs) for functional long-term studies is needed.⁴⁸ Long-term mechanotransduction assessments require mechanical stimulation in a mechanical bioreactor, such as the CellScale MechanoCulture^™ T6 Bioreactor, Flexcell^® Tension Systems, or custom-built systems.^{47, 49} Development of long-term culture platforms necessitates the ability to mimic physiologically relevant mechanical motion⁴⁷ and understand the impacts of this motion on a relevant biomaterial prior to investigating clinically relevant hypotheses in these systems. Notably, previous work by Somers et al., determined relevant time scales over which the onset, magnitude, and repetition of mechanical motions impacted proliferation outcomes of three-dimensional in vitro skeletal muscle growth.⁵⁰ Moreover, other skeletal muscle cellularized scaffolds resulted in contractile, mature muscle tissue mimics after 6 to 32 days in culture, highlighting the need for long-term culture platforms for striated muscle studies.⁴⁵

In this work, fundamental connections between structure-function relationships of beta-sheet organization and tunable mechanics and degradation rates were evaluated. These efforts are important for the development of silk fibroin scaffolds in in vitro new approach methodologies, which can provide insight into disease pathologies as an alternative to animal models. Preclinical studies utilizing in vitro NAMs would alleviate the rising cost of animal experiments and regulatory changes associated with animal models. While existing skeletal muscle NAMs have been utilized for various applications, the time scale of these assessments has been limited.^{48, 51} For example, wound healing or disease modeling can take weeks to months to recapitulate applicable phenotypes.^{48, 51} The findings reported here will inform the selection of a long-term material formulation for specific applications prior to preclinical testing.

2. Methods

2.1. Decellularized Extracellular Matrix

Similar protocols as reported by Stoppel et al. were used to decellularize adult skeletal muscle tissue.^{27, 52} Adult porcine skeletal muscle (antibiotic free pork chops) was obtained from Whole Foods Market. The tissue was cut into small pieces (10 mm × 10 mm × 10 mm) and washed with 1X phosphate buffered saline (PBS) (Fisher Bioreagents, Pittsburgh, PA, USA) containing 1% penicillin and streptomycin (pen-strep) for 30 minutes, then decellularized with 1% (weight/volume) sodium dodecyl sulfate (SDS) (Boston Bioproducts, Milford, MA, USA). The SDS solution was changed every 2 hours on the first day and 3 times per day until the tissue appeared white and was completely decellularized (approximately 6 days). To ensure full decellularization, wax embedding, sectioning, and hematoxylin and eosin (H&E) staining were performed to ensure all the cells have been removed (Supplemental Figure 1). The fully decellularized skeletal muscle pieces were washed in ultrapure water for 48 hours to remove any SDS. The pieces underwent acetone precipitation to remove any unwanted small molecules or residual SDS, followed by 3 rinses in ultrapure water for 20 minutes each. Excess liquid was aspirated from samples, and the decellularized tissue was then frozen at −80°C for 24 hours and lyophilized (Labconco, Kansas City, MO, USA) for 48 hours. The pieces were milled through a 40 mm mesh on a single speed mini cutting mill (Thomas Scientific, Thorofare, NJ, USA) and then solubilized through pepsin digestion in 0.1 N hydrochloric acid (LabChem, Zelienople, PA, USA). After digestion, pepsin was neutralized with 1 M sodium hydroxide (Fisher Chemical, Pittsburgh, PA, USA). The decellularized extracellular matrix (dECM) solution was then frozen and lyophilized until use. Reconstituted dECM was diluted with ultrapure water to a stock concentration of 20 mg dECM per mL water. The dECM solution was further diluted during scaffold fabrication to a final concentration of 0.2 mg dECM per mL solution.

2.2. Silk Fibroin Extraction

Silk fibroin solution was prepared through previous methods.¹ Bombyx mori silk cocoons were cut to remove the insect and debris. 5 grams of the cocoons were boiled in 2 liters of boiling sodium carbonate solution (0.02 M) for 30 minutes (Sigma-Aldrich, St. Louis, MO, USA) to remove sericin proteins coating the fibroin fibers. After extraction, the degummed fibers were rinsed in ultrapure water for 20-minute intervals 3 times, then allowed to dry in a fume hood for 48 hours. To solubilize the degummed fibers, 9.3 M of aqueous lithium bromide (Sigma-Aldrich, St. Louis, MO, USA) was added at a 1:4 silk fibroin mass to salt solution ratio. The silk fibroin solution was dialyzed with 3.5 kDa MW cut off dialysis tubing (3,500 MWCO, ThermoScientific, Rockford, IL, USA) against ultrapure water. After dialysis, silk was centrifuged three times at 3,000 RPM for 20 min at 4 °C to remove insoluble particulates. The solutions were stored at 4 °C. To find the resulting concentration of silk fibroin solution, the weight by volume percentage was found by drying the solution overnight and comparing the dry and wet masses. The silk solution was diluted with ultrapure water to 5% (weight/volume).

2.3. Anisotropic Silk Fibroin Scaffolds

Silk fibroin solution at the desired 5% (weight/volume) concentration was gently mixed with skeletal muscle dECM to achieve a final dECM concentration of 0.2 mg dECM per mL (Figure 1A). To mimic skeletal muscle tissue alignment, the scaffolds were frozen in an anisotropic direction, as previously described by Stoppel et al.²⁷ The freezing arrangement was achieved by placing crushed dry ice pellets and 100% ethanol, mixed to form a slurry, on one side of a metal divider and the silk fibroin solution on the other. 184 Sylgard elastomer (Dow Corning, Midland, MI, USA) was used as the base to help maintain the metal mold. The silk fibroin solution froze across the container, starting at the metal plate and spreading across the solution until it was fully frozen (Figure 1B). The resulting frozen silk and water block was put on a lyophilizer (Labconco, Kansas City, MO, USA) to sublimate the ice and remove it from the sample. This lyophilization (−80 °C and 0.275 mbar) produced a water-soluble, aligned, silk fibroin scaffold, due to the pores that were created when the ice crystals were sublimated. To render the scaffold water insoluble a post-lyophilization step of water annealing (23°C and 0.0728 MPa, 6 L desiccator with 500 mL H₂O) or autoclaving (121°C and 0.01 MPa) was performed (Figure 1C). It should be noted that differences in pressure, temperature, and humidity during these processes may alter rates of formation and internal organization. Care was taken to ensure the same conditions were used throughout the formulation processes. From the large scaffold sample (73 mm × 65 mm × 8 mm), smaller strips (20 mm × 10 mm × 3 mm) were cut out for experimental use.

Figure 1.

(A) Scaffolds made with silk fibroin or silk fibroin with decellularized extracellular matrix (dECM) (B) Scaffolds were formed isotropically (iso) or anisotropically (aSF) dependent on the method of freezing (C) After lyophilization, scaffolds underwent post-lyophilization steps to induce crystallinity. Created in BioRender.

2.4. Isotropic Silk Fibroin Scaffolds

Silk fibroin solution at the desired 5% (weight/volume) concentration was poured into wells of a 6-well plate. Isotropic scaffolds were not formed with dECM since they were only used to assess crystallinity (Figure 1A). The plate was left overnight in a −80°C freezer (Figure 1B). The resulting frozen silk and water block was put on a lyophilizer (Labconco, Kansas City, MO, USA) to sublimate the ice and remove it from the sample. This lyophilization (−80 °C and 0.275 mbar) produced a water soluble, silk fibroin scaffold, due to the pores that were created when the ice crystals were sublimated. To render the scaffold water insoluble a post-lyophilization step of water annealing (23°C and 0.0728 MPa, 6 L desiccator with 500 mL H₂O) or autoclaving (121°C and 0.01 MPa) was performed, with a control of no post-lyophilization step (Figure 1C).

2.5. Fourier-Transform Infrared Spectroscopy

Quantification of secondary protein structures was performed using Fourier-transform infrared spectroscopy (FTIR) analysis. Prior to FTIR analysis, scaffolds were dried in a fume hood for 48 hours. Spectra were collected with a Nicolet iS50 FTIR Spectrometer (Thermo Fisher Scientific, Waltham, MA) at the UF Nanoscale Research Facility, equipped with an attenuated total reflections (ATR) zinc selenium (ZnSe) crystal. Each measurement consisted of 128 scans with a resolution of 4 cm⁻¹ over a range of 4,000-650 cm⁻¹. Background spectra were collected using the same conditions and subtracted from each sample spectrum in addition to subtracting spectra to remove background water and carbon dioxide contributions. Relative secondary structure content was determined through deconvolution (Supplemental Figure 7) as described by Hu et al., where the amide I region (1,590–1,710 cm⁻¹) can be split into regions based on the protein secondary structure: 1,605–1,615 cm⁻¹ as side chain/aggregated strands, 1,616–1,637 cm⁻¹ and 1,697–1,703 cm⁻¹ as beta-sheet structure, 1,638–1,655 cm⁻¹ as random coils, 1,656–1,662 cm⁻¹ as α-helical bands, and 1,663–1,696 cm⁻¹ as turns.⁵

2.6. X-Ray Diffraction

Silk fibroin scaffolds (no dECM) were cut (32 mm diameter × 2 mm thickness) and placed in a shallow well sample holder. X-ray diffraction data was collected on a Malvern Panalytical Empyrean (Malvern, Worcestershire, UK) diffractometer between 1-50° at a rate of 2° per minute with a copper long fine focus high resolution source tube (Empyrean CU LFF HR) set to 45 kV and 40 mA with a wavelength of 1.5418 Å. Intensity versus 2Θ was plotted and degree of crystallinity was obtained by deconvolution of peaks similar to previous silk literature.^{53, 54} Bragg’s Law was used to convert 2Θ to real-space and reciprocal-space values. Intensity values were normalized to the maximum intensity per sample. Supplemental Figure 2 shows replicate peaks for deconvolution that were averaged to obtain the crystalline content.

2.7. In vitro Degradation

Samples were cut (small cubes: 5 mm × 5 mm × 5 mm) from the bulk scaffold. Discrete sampling was used in which one scaffold piece (n=3) was assessed for a singular time point and discarded after measurement to limit repeated handling and drying of the samples. Samples were dried to gain the initial starting mass in a 1.5 mL centrifuge tube. One milliliter of Protease XIV (Sigma Aldrich, catalog number P4147) solution in 1X PBS at an activity concentration of 1 U/mL was added to the centrifuge tube containing the sample. The centrifuge lids were closed, and samples were stored in a 37°C incubator on a rocking plate. Fresh enzyme solution was replaced every 24 hours throughout the time course. When a sample reached its designated time point, the sample was rinsed with ultrapure water before aspirating all liquid. Samples were left to dry for 72 hours in a fume hood before measuring the final mass. Portions of the degraded scaffolds were taken for SEM, FTIR analysis, or nano-CT imaging. Time points for analysis were 24, 48, 72, 96, 120, and 144 hours. Mass loss was calculated as a percentage of the remaining mass compared to the initial mass prior to degradation. Best fit lines from the mass loss data, as well as FTIR deconvolution, can be found in Supplemental Figure 3.

2.8. Bioreactor Setup for Mechanical Stimulation

After being cut to size (20 mm × 10 mm × 3 mm), the anisotropic scaffolds (dECM, WA) were autoclaved (122 C, 18.5 PSIG, 15 min) in 0.5X phosphate buffered saline (PBS) containing dECM on a short liquid cycle for sterility (dECM, WA postA, Figure 1A, C). All subsequent steps were performed in a biosafety cabinet to ensure sterility. To mimic the experience of skeletal muscle growth from literature, scaffolds were placed in well plates on an orbital shaker for 5 days (Day 5) with media changes every 36 hours. After 5 days of growth,⁵⁰ scaffolds were transferred to a sterile MechanoCulture T6 Bioreactor (CellScale, Ontario, Canada). Up to 6 scaffolds were loaded into the bioreactor at a resting length of 10 mm with clamps on each end of the scaffold. After loading, the bioreactor was filled with DMEM media containing 1% pen-strep and 10% fetal bovine serum (FBS) or ultrapure, sterile water. The bioreactor was programmed to provide uniaxial mechanical stimulation, in the direction of alignment, with a 10% strain, 1 Hz frequency, 30 min of active stimulation, and 11.5 hours of rest. The bioreactor was housed at 37 °C in a 5% CO₂ humidified atmosphere and the media or water was changed every 36 hours. Scaffolds were removed from the bioreactor after 5 (Day 10) or 10 days (Day 15) for assessment with a control of 0 days on the mechanical bioreactor. Further assessment was added for 25 days (Day 30) in the bioreactor to demonstrate longer term use.

2.9. Dynamic Mechanical Analysis

The width and thickness of the scaffold samples were measured with digital calipers (20 mm × 10 mm × 3 mm). Dynamic Mechanical Analysis (DMA) occurred on an Anton Paar MCR 702e Rheometer (Anton Paar, Graz, Austria). Each sample was placed into tensile clamps, and the clamps were tightened to 10 centinewton per meter using a torsion screwdriver. The SRF upper and lower clamps of the MCR 702e were used (U-SRF5 and the L-SRF5/LD, respectively). The gap width was recorded as the height of the sample between the clamps (10 mm). A custom immersion cup system (Anton Paar) was attached to the instrument to keep the samples hydrated in DMEM supplemented with pen-strep and FBS during testing or ultrapure water. Anton Paar RheoCompass software was utilized to perform static and tensile testing. Each tensile test included a pre-stretch at 0.1 N to ensure the sample did not sag.

Hydrated static testing was used to extend the samples from rest (original length at 0.1 N) to break strain. Static tests were run at 1 mm per minute (equivalent to 10% per minute) strain amplitudes. Three samples for each scaffold formulation were used (n=3). From each static test, the Young’s modulus, strain at break, and ultimate tensile strength were calculated from the stress-strain curves. Young’s modulus was calculated as the slope of the stress-strain curve from 0 to 3 mm/mm. Break strain was calculated as the strain at the maximum stress before breakage. Ultimate tensile strength is reported as the maximum stress achieved by each sample. Representative stress-strain curves with the calculated properties can be found in Supplemental Figure 4, 5, & 6. The mean ± standard deviation for the Young’s modulus (E), strain at break, and ultimate tensile strength (UTS) values can be found in Supplemental Table 1.

2.10. Scanning Electron Microscopy

Cross sectional pieces of the scaffolds were cut to examine the x, y, and z planes of the scaffolds. Samples were attached to conductive tape mounted on ZEISS/LEO SEM Pin Stub Mount, Ø12.7mm × 9mm pin height (Catalog No. 16202, Ted Pella, Inc., Redding, CA) and left to dry in a fume hood for 24 hours. A Phenom Pure benchtop SEM (ThermoFisher Scientific, Waltham, MA, USA) was used to image samples in a charge reduction sample holder at 5kV and 500X magnification to confirm alignment and architecture structure. ImageJ (US NIH, Bethesda, MA, USA) was utilized for analysis of the SEM images with pore diameters and alignment measurements given in Supplemental Figure 8.

2.11. Nano-Computed Tomography

Assessment of the internal structure and porosity was conducted using nano-computed tomography (Nano-CT) at the University of Florida Nanoscale Research Facility. A small section of scaffold was stained with Lugol’s iodine solution (Sigma-Aldrich, St. Louis, MO, USA) for 24 hours. The stained scaffold pieces were rinsed with ultrapure water for 12 hours on a tube invertor. The scaffolds were then scanned via an Xradia 620 Versa (Carl Zeiss X-Ray Microscopy, Oberkochen, Baden-Württemberg, Germany) at 70kV and 8W with 0.4X magnification and analyzed using VGStudio Max (Hexagon AB, Stockholm, Sweden).

The 3D reconstruction from nano-CT was converted to a mesh using VGStudio Max (Volume Graphics GmbH, Heidelberg, Germany). The meshes were imported into Blender (Blender, Amsterdam, Netherlands) using the Bullet Physics engine add on and set to passive rigid body mesh collision settings. To assess pore interconnectivity, 50 sphere particles of a set diameter (5, 10, or 50 μm), were dropped in the direction of alignment into the top of the mesh (Supplemental Video 1-3, 0.5 μm spheres in Supplemental Figure 9). The retention of the spheres was calculated from snapshots of the spheres.

2.12. Statistical Analysis

Experimental data are mainly expressed as mean ± standard deviation (SD) with n=3, unless otherwise stated. GraphPad Prism (La Jolla, CA) was utilized to analyze these data. Analysis is completed with appropriate-sized analysis of variance (ANOVA). If significance was found, Tukey post hoc testing was used for pairwise comparisons. Statistical significance is reported as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

3. Results and Discussion

Silk fibroin scaffolds formed by anisotropic ice-templating result in channeled materials with different structures on each face of the material.^{21, 23, 31} Fabrication methods and crystalline induction strategies can influence mechanical properties as well as overall pore architecture.^{4, 7, 21, 23, 26, 55} To further expand the utility of these materials as new approach methodologies (NAMs) to provide long-term, in vitro, preclinical data for tissue engineering, regeneration, or disease pathology studies, trends in rates of degradation under mechanical loading conditions for various potential formulations should be understood.⁵⁶ An understanding of the interplay between formulation conditions, mechanical loading, and rates of degradation will enable intentional scaffold design based on biological application. A major contributor to scaffold performance is the total content and organization of crystal structures within these scaffolds.

The amino acid composition of proteins dictates the mechanical, thermal, and physical properties of native silk fibers or resulting silk fibroin-based biomaterials.^57-59 A variety of repetitive amino acid domains interact through inter- and intra-molecular forces, resulting in crystalline beta-sheet organization which increases mechanical and thermal properties.^{57, 60} Silk fibroin is known to form physical crosslinks via secondary structure formation within the protein, resulting in ordered beta-sheet crystalline regions.^{5, 10} The hierarchical organization of these beta-sheet crystalline regions, via hydrogen bonding, yields an interspersed array of crystalline regions and amorphous polymer domains.^{7, 31, 57} The rate of spontaneous formation and the resulting hierarchical organization lead to shifts in mechanical properties between conditions with similar compositions, altering mechanics, but also altering propensity for material degradation. It has been demonstrated in silk fibroin hydrogels that the duration allotted for the protein to rearrange into its equilibrium structure affects the mechanical properties and beta-sheet content.⁷ Silk fibroin hydrogels that were allowed to spontaneously gel over 60+ days resulted in significantly higher beta-sheet content than those that had energy quickly introduced to the system (sonicated for 15 seconds).⁷ Crystalline beta-sheet structures can be intentionally induced via external forces, such as the application of shear forces, by increasing the temperature, changing the pH, or through water vapor annealing, but will also form even without intentionality as the biopolymers interact, aiming to minimize free energy. ^{1, 5, 7, 10, 61}

In ice-templated silk fibroin scaffolds, this crystallinity is intentionally induced via water vapor annealing or through autoclaving, and these methods for inducing crystallinity impact initial static and dynamic mechanical properties.³¹ Scaffolds were formed by freezing in a −80 °C freezer (iso) or anisotropically (aSF) with a slurry of dry ice and ethanol. Scaffolds were then lyophilized before undergoing specific post-lyophilization beta-sheet induction. No post-lyophilization (NA), water annealing (WA), autoclaving (A), or water annealing with an additional autoclaving step for sterilization (WA postA) were used as the crystalline induction comparisons (Figure 1), with an emphasis on the time and temperature driving formation. In this work, prior literature is expanded upon to quantify how crystalline region organization contributes to the long-term performance of anisotropic silk scaffolds intended for use as in vitro NAMs that enable the study of a wide range of biological phenomena, such as rare disease progression, bioactive molecule delivery, and the influence of mechanical perturbation.

3.1. Crystalline domain determination in silk fibroin scaffolds

Efforts have been made in the silk field to characterize fibers through X-ray techniques,^62-67 which can yield information about crystalline size and spacing. X-ray scattering has been used to assess silk dope and self-assembly within silk glands providing information on the crystallite organization of precursor Bombyx mori silk solutions.^68-70 While the native crystallinity of silk dope and silk spinning in the gland is helpful from an evolutionary standpoint, understanding self-assembly of regenerated Bombyx mori silk fibroin is more relevant to the development of rationally designed silk fibroin-based materials. X-ray diffraction (XRD) has previously described isotropic scaffolds with different post-lyophilization techniques.^{53, 71} It was shown that increasing time of annealing caused shifts in peak location toward crystalline structures.^{53, 71} Most commonly, beta-sheet content in silk fibroin materials is assessed through Fourier-transform infrared (FTIR) spectroscopy, with crystalline content ranging from 40-60% based on post-lyophilization steps and material format.^{4, 26, 55} Here, the differences in crystalline content are shown for different post-lyophilization processing methods of anisotropic silk fibroin scaffolds. Therefore, the differences between no post-lyophilization step, water annealing, and autoclaving are compared for isotropic and anisotropic silk fibroin scaffolds, with a focus on crystalline organization (Figure 2) and the resulting degradation kinetics (Figure 3) and dynamic mechanical responses (Figure 5).

Figure 2.

(A-D) Fourier transform infrared (FTIR) spectroscopy of 30-min degummed, 5% (w/v), ice-templated silk fibroin scaffolds (A) Representative normalized spectra for isotropic (iso) scaffolds with no post-lyophilization (NA), water annealing (WA), or autoclaving (A) steps (B) Beta-sheet content determined by deconvolution of isotropic FTIR spectra (n=3) (C) Representative normalized spectra for anisotropic scaffolds with water annealing, or autoclaving steps (D) Beta-sheet content determined by deconvolution of anisotropic (aSF) FTIR spectra (n=3) (E-J) X-ray diffraction of 30-min degummed, 5% (w/v), ice-templated silk fibroin scaffolds (E) Representative normalized intensity of isotropic scaffolds with different post-lyophilization processing conditions (F) Total crystalline content determined by deconvolution of isotropic scaffold XRD spectra (n=3) (G) Representative normalized intensity of anisotropic scaffolds with different post-lyophilization processing conditions (H) total beta-sheet crystalline content determined by deconvolution of anisotropic scaffold XRD spectra (n=3). (I-J) X-ray diffraction of 30-min degummed, 5% (w/v), ice-templated silk fibroin scaffolds comparing isotropic (iso) and anisotropic (aSF) scaffolds with water annealing (WA), autoclaving (A), or a combination of both (WA postA). Data are expressed as mean ± standard deviation. Analyzed with 1-way ANOVA with Tukey’s post-hoc analysis. Statistical significance is reported as *p<0.05, ***p<0.001, and ****p<0.0001.

Figure 3:

In vitro degradation of anisotropic silk fibroin scaffolds via protease XIV (A) Silk fibroin amino acid composition and secondary structure with protease XIV cleavage sites Y, F, H, W (highlighted in blue). Adapted with permission from Pacheco et al.⁷ Copyright 2025 American Chemical Society. Close up view of lamellar beta-sheets where protease XIV cannot penetrate the crystalline domains. (B) Mass loss over time of scaffolds containing no ECM (open symbols) or with ECM (closed symbols) and post-lyophilization methods of water vapor annealing (WA, blue) and autoclaving (A, pink). (C) Changes in beta-sheet content via FTIR deconvolution over time of degraded scaffolds containing ECM. (D) Representative SEM images of a scaffolds containing ECM with water annealing post lyophilization step after one day of degradation in vitro with protease XIV. (E) Analysis of changes in pore size over time of degraded scaffolds containing ECM with water annealing post lyophilization step. (F) Pore size evaluation of SEM images from day 1 and day 2 scaffolds after degradation. Data are expressed as mean ± standard deviation. Analyzed with 2-way ANOVA with Tukey’s post-hoc analysis. Statistical significance is reported as *p<0.05 and ****p<0.0001.

Figure 5:

(A-B) Representative stress-strain curves of hydrated extensional test at a rate of 10 mm/min of (A) Water annealed scaffolds pre- and post-autoclave sterilization at Day 0 (B) Day 10 scaffolds stored in water (H₂O, black) or media (DMEM, blue). (C-F) Hydrated storage in water (H₂O, open symbol) or media (DMEM, closed symbol) at Day 5, Day 10, or Day 15 (shades of blue) (C) Total relative beta-sheet content (n=8). (D) Extensional Young’s modulus values (n=3). (E) Strain at break (n=3). (F) Ultimate tensile strength (n=3). Data are expressed as mean ± standard deviation. Analyzed with 2-way ANOVA with Tukey’s post-hoc analysis. Statistical significance is reported as *p<0.05 and ***p<0.001. Statistical significance is non-significant for all mechanical testing.

Aikman et al. was the first to assess anisotropic silk fibroin scaffolds via FTIR.³¹ Notably changes in beta-sheet content were compared for water annealing and autoclaving post-lyophilization steps, where lower silk fibroin concentration (3%) produced no significant differences, but higher silk fibroin concentrations (5%) were significantly different.³¹ These increases in beta-sheet content were linked to increases in plastic deformation through tensile testing. Specifically, water annealing (23°C and 0.0728 MPa for 6 hours) produced less total beta-sheet content with a more elastic response, while autoclaving (121°C and 0.01 MPa for 1 hour) resulted in higher beta-sheet content, but more plastic deformation.³¹ Molecular Dynamics (MD) simulations predicted that larger crystalline structure sizes in the silk fibroin polymers lead to more plastic deformation, while more closely packed, smaller crystalline domains lead to less plastic deformation.³¹ However, this has not yet been confirmed experimentally through the size or location of crystalline domains. Moreover, the MD simulations focused on ratios of structural components in a single molecule and did not predict the interactions of full-length polymer systems or inter-molecular interactions. Tools like AlphaFold struggle to accurately determine structures of large proteins like silk fibroin (around 390 kDa), thus there are limited resources to predict the self-assembly of crystalline domains within silk fibroin, with most computational efforts focusing on the structure of the beta-sheet region.⁶⁹

When comparing post-lyophilization steps for silk fibroin ice-templated scaffolds, it is expected that the scaffolds with no post-lyophilization step will have the lowest crystalline content. These scaffolds are still water-soluble, and the protein backbone still contains a substantial amount of bound water. Adding post-lyophilization steps excludes water from the backbone of the silk fibroin protein, inducing beta-sheet formation and rendering the scaffolds water-insoluble.²⁶ The extent of water exclusion and beta-sheet formation is thermodynamically driven and is highly dependent on time and temperature.⁵ Water vapor annealing has been shown to increase beta-sheet content as the time of annealing increases.^{4, 26} Autoclaving scaffolds induces beta-sheet content faster due to the addition of heat. However, longer times for beta-sheet induction are hypothesized to cause different organizations of beta-sheet structures due to relaxation time of the silk fibroin polymer.

First, FTIR was used to assess total beta-sheet crystalline content, providing semi-quantitative results based on relative total secondary structure content. FTIR is useful for identifying changes in beta-sheet content through the different post-lyophilization fabrication methods. Infrared absorbance spectra were deconvoluted based on known wavenumber peaks contributing to different secondary structures previously shown by Hu et al.⁵ Isotropic scaffolds (Figure 2A, B) showed significant increases (p<0.0001) in beta-sheet content between the three post-lyophilization methods. Similarly, the anisotropic scaffolds (Figure 2C, D) show significant increases (p<0.0001) in beta-sheet content between the water annealing (aSF WA) and autoclaving conditions (aSF A). For the isotropic scaffolds with no post-lyophilization steps, total beta-sheet content was around 36%. For both the isotropic and anisotropic scaffolds, the beta-sheet content is around 50% for the water annealed scaffolds, while the autoclaved scaffolds have around 60% beta-sheet content (p<0.0001). This demonstrates that the direction of ice-templating has a limited impact on overall beta-sheet content (p=0.1154). The main driving factor for beta-sheet content is the temperature and the length of time of beta-sheet induction, where the higher temperature during autoclaving was hypothesized to induce beta-sheets more quickly.

XRD was used to assess crystalline structure of isotropic and anisotropic silk fibroin scaffolds with water annealing, autoclaving, and no post-lyophilization fabrication steps, as well as autoclave sterilization post water annealing. The comparison between water annealing and autoclaving was chosen to compare with previous MD simulation results³¹ with a no post-lyophilization step as the control. Isotropic and anisotropic scaffolds were generated to minimize the effects of symmetry and crystallite anisotropy on the nanometer level, as opposed to the macroscopic scale observed at the sponge surfaces. Isotropic scaffolds were shown to have no crystalline anisotropy, as demonstrated by replicate scans at multiple sample orientations (rotational direction of the sample in comparison to the incident X-ray source), with no changes in peak location and size (Figure S2A, B). In comparison, multiple sample loading angles showed changes in higher-order peaks for the anisotropic scaffolds, demonstrating anisotropy at both the crystallite (nanometer) and scaffold sheet (macroscopic) levels (Figure S2C).

XRD data for isotropic silk fibroin scaffolds show multiple changes in relative peak intensity and peak location (Figure 2E). The scaffolds with no post-lyophilization step (iso NA) appear as a singular broad peak. The water annealed (WA) scaffolds have larger higher-order scattering features in the small Q-space region (0.5-1.0 Å⁻¹). The autoclaved (A) scaffolds have two more defined peaks in the middle (1.0-2.0 Å⁻¹) and less intensity in the higher-ordered peaks. Of interest is the difference in higher-ordered spacing, where the dominant peaks for the autoclaved scaffolds are located in the smaller Q-space region. Additionally, the smaller-order spacing, at higher Q-space, shows a step-like decrease in intensity from 2.0-3.0 Å⁻¹. These equally spaced regions are representative of longer-range periodicity, with lowering intensity hypothesized to result from smaller equally spaced crystalline domains. Inversely, the autoclaved scaffolds, which have sharper, non-equally spaced peaks in higher-ordered regions, are hypothesized to be representative of larger, non-ordered crystalline domains. This matches well with the MD simulation predictions of crystalline organization previously hypothesized.³¹ The water annealed scaffolds have smaller but more equally spaced crystalline domains, while the autoclaved scaffolds have larger and less frequent crystalline domains due to the application of heat during autoclaving. Previous literature has identified crystalline domains, where the peak deconvolution (between 1.0-2.0 Å⁻¹) is representative of beta-sheet domains.^{54, 72} Similarly to the FTIR, the total crystalline content from XRD (Figure 2F) matches the hypothesis that increasing post-lyophilization step temperature increases overall crystalline content. The scaffolds with no post-lyophilization step show the lowest crystalline content, around 42%. The water annealed scaffolds are significantly higher (56%, p=0.0270), with the autoclaved scaffolds even significantly higher (60%, p=0.0008). Here, the differences between the isotropic water annealed and isotropic autoclaved scaffolds are not significant (p=0.3219), highlighting the influence of changes in shape and organization of the crystallites rather than total overall content.

Similarly to the peak intensity and location in the isotropic scaffolds, the anisotropic scaffolds show differences between the larger order domains and smaller order periodicity. The same shifts in the higher order Q-space region show a shift from a sharper peak in the water annealed scaffold to a broader peak shifted toward higher order in the autoclaved scaffold (Figure 2G). Additionally, the water annealed scaffolds still show smaller, more frequent, but lowering in intensity, peaks in the larger Q-space region, corresponding to the smaller, but more frequent crystalline domains. Total crystalline content determined by deconvolution (Figure 2H) shows statistical similarities between water annealing and autoclaving scaffolds (on the order of 60%). This aligns more closely with the results of Aikman et al., where anisotropic silk fibroin scaffolds showed no significant differences and little change in beta-sheet content by FTIR between water annealing and autoclaving post-lyophilization steps.³¹

Interestingly, the difference between isotropic and anisotropic scaffolds are very minimal when compared through FTIR and XRD. The direction of ice-templating (macroscopic anisotropy) does not have significant impact on overall beta-sheet content (p>0.05), instead the temperature and length of time of beta-sheet induction, where the higher temperature during autoclaving produces the largest differences. This can best be seen through comparison of water annealed, autoclaved, and water annealed plus autoclaved scaffold comparisons (Figure 2I and Figure 2J). The water annealed plus autoclaved scaffolds (WA postA) were added to match the structure of the sterilized scaffolds in the mechanical property assessment. The scaffolds added to the bioreactor were sterilized to confirm their future use in cell culture conditions. The shift at larger order Q-space regions can be observed between the water annealed (WA) to water annealed with the addition of autoclaving (WA postA) scaffolds. The small equal, decreasing peaks, at longer range spacing also disappear through the addition of autoclaving, matching the shape of the autoclaved scaffolds instead of the water annealed scaffolds, highlighting the potential for continuing shifts in intra- and inter-molecular interactions within these scaffold formulations.

Overall, the organization of crystalline domains in silk fibroin scaffolds was shown to match that of the previously predicted MD simulations.³¹ Here water annealed scaffolds, formed at lower temperatures, were found to have higher order periodicity with smaller crystallite sizes. Isotropic water annealed (iso WA) scaffolds were found to have a crystallite size of 1.45 nm at a Q-space of 1.5 Å⁻¹, determined by the Scherrer equation. Similarly, anisotropic water annealed (aSF WA) scaffolds maintained a similar crystallite size of 1.01 nm. The autoclaved scaffolds showed less frequency, but larger, non-ordered crystalline domains. Isotropic and anisotropic autoclaved scaffolds were found to have crystallite sizes of 4.56 nm and 2.83 nm, respectively, determined by the Scherrer equation at a Q-space of 1.5 Å⁻¹. These differences in peak intensity and location demonstrate water annealed scaffolds contain smaller, more frequent crystallites, while autoclaved scaffolds contain larger, more dispersed crystallites. Moreover, the addition of an autoclaving step following water annealing preparation shows further shifts in crystallite ordering and size, demonstrating the impact of heat on thermodynamic stability and polymer reorganization. The impacts of the changes in crystalline size and spacing will be explored through degradation and mechanical assessments.

3.2. Crystalline content decreases degradation kinetics

Capturing degradation kinetics of biomaterials is important for considerations toward in vitro or in vivo applications.⁷³ The degradation of materials is driven by external factors (enzymatic, hydrolysis, oxidation, mechanical, etc.),² and the influence of each of these can vary by application. For in vivo work, typically in biomaterial implantation, the rate of degradation should closely follow the rate of new tissue formation.^{2, 73} This ensures that the appropriate structure is maintained through the initial growth stages, but that the material is fully removed by the time the new tissue is fully formed. For silk fibroin materials, in vitro degradation experiments have been conducted on isotropic scaffolds,⁴ micro- and nano-particles,⁷⁴ as well as hybrid materials like particle-laden scaffolds.⁷⁵ Mathematical determination has revealed specific kinetic rate parameters for in vitro degradation for isotropic silk fibroin scaffolds.⁴ The same type of scaffold has also been shown to be degraded by immune cells when implanted in rodent models, with time dependence examined over an 8 week period.³⁴ Particle-laden silk fibroin scaffolds have been developed for multi-phase degradation of two different time scales within the same material, with the outer porous scaffold degrading faster than then entrapped microparticles.⁷⁵ Previous in vitro work has utilized several model enzymes for degradation experiments, with protease XIV being a common choice.^{2, 4, 26, 32, 55, 75, 76} However, degradation experiments have not been performed for the ice-templated aligned silk fibroin scaffolds described here.

The formation of beta-sheet structures leads to robust mechanical properties.⁵⁷ These beta-sheet structures can be formed through multiple methods previously discussed, such as temperature, shear forces, pH, salt, and temperature changes which modify the size and organization of secondary structures. ^{1, 5, 7, 10, 61} In silk fibroin scaffolds, crystallinity is set by exposing the scaffolds to vacuum conditions in the presence of water or heat, where the autoclaved scaffolds have larger crystalline regions with more spread-out and irregular amorphous regions whereas the water annealed scaffolds have smaller crystalline regions with more frequency in amorphous regions disruption.³¹ The crystallinity affects the degradation based on the availability of where the enzyme can penetrate based on steric hinderance of various secondary structures and amino acids within the crystalline regions. Protease XIV is known to cleave at tyrosine (Y), phenylalanine (F), tryptophan (W), and histidine (H).⁴ Pacheco et al. showed through HPLC that the degumming of silk fibroin reduces molecular weight as well as alters amino acid composition for the remaining biopolymer.⁷ In the 30-minute degummed silk fibroin, used for these scaffolds, previous HPLC data results hypothesize the loss of the N- and C-termini of the unmodified, full-length silk fibroin heavy chain during the degumming step due to thermal degradation.⁷ Further breakdown of the silk fibroin protein is hypothesized to occur in the amorphous regions of the protein. Unmodified, full-length heavy chain, of Bombyx mori (UniProtKB/Swiss-Prot: P05790.4), contains 277 tyrosine, 29 phenylalanine, 11 tryptophan, and 5 histidine, with 3.6%, 20.7%, 0%, and 40% present in the N- and C-termini, respectively. The remaining phenylalanine, tryptophan, and histidine residues are located within the amorphous regions, often not participating in the formation of beta-sheet regions.⁷ However, tyrosine is located in both the amorphous and crystalline domains with more tyrosine located in the crystalline region (Figure 3A). Of the residues cleavable by protease XIV remaining in the 30-minute degummed silk fibroin, it is expected that 85% are within crystalline domains and may not be accessible at the start of degradation.

Given the more open crystalline structure and potential availability of the amorphous regions, it was hypothesized that the water annealed scaffolds would degrade faster than the autoclaved scaffolds due to the previous MD simulations, confirmed by XRD. Here, 5 mm cubes of ice-templated scaffolds with post-lyophilization steps of water annealing (WA) or autoclaving (A) were exposed to 1 U/mL protease XIV solution in PBS. The activity concentration was chosen to allow for comparison to previous literature results for silk fibroin based-biomaterial degradation^{4, 34} as well as its representation toward physiologically relevant conditions for other enzyme concentrations.⁷⁵ Degradation was measured by mass loss over time and endpoint scaffolds were saved for FTIR analysis or SEM imaging. The mass loss over time data (Figure 3B) shows that the water annealed scaffolds degrade faster than the autoclaved scaffolds, as hypothesized. The addition of decellularized ECM peptides does not greatly change the degradation rate of either post-lyophilization fabrication method (Figure S3A). These results align with previous degradation rates for isotropic silk fibroin scaffolds³⁴ that fully degrade around 7 to 9 days. Additionally, to confirm the hypothesis that cleavage at amorphous sites over crystalline regions drives the faster degradation of the water annealed scaffolds, the secondary structure was assessed through FTIR (Figure S3B & D). The beta sheet content (Figure 3C) significantly increased during the first day of degradation for the water annealed scaffolds, while the autoclaved scaffolds showed a slight increase over the course of the degradation period. The same trend holds for scaffolds containing no ECM (Figure S3C). This further confirms the hypothesis as enrichment of beta-sheet regions occurs through the cleavage and breakdown of amorphous regions.

Through investigation into the effects of crystallinity on degradation, structural effects were explored at endpoints through analysis of SEM images. SEM images were taken of the porosity on all three faces: in the direction of alignment, on the side of the scaffold, and from the top. During degradation, it was visually clear that the scaffolds were shrinking in overall bulk size over time. This can be measured through changes in porosity over time. Figure 3D and Figure 3E show changes to pore sizes from degradation in protease XIV after one day. Pore diameters for non-degraded scaffolds had an average pore diameter of 37 μm, while scaffolds degraded after the course of 2 days for water annealing and autoclaving had pore sizes with an average size of 41 μm and 90 μm, respectively. Collecting and handling the scaffolds for the correct orientation for imaging over the entire time course was not achievable due to the extent of degradation beyond two days, thus pore diameter measurements are available for only days 1 and 2 (Figure 3F). There are significant differences between pore sizes (p<0.0001) for each time point of degradation of the autoclaved scaffolds. Additionally, there are significant differences (<0.0001) between the water annealing and autoclave post-lyophilization steps after two days of degradation. This is notable since the pore sizes of the autoclaved scaffolds are significantly larger than the water annealed scaffolds, which was not expected as the water annealed scaffolds degrade quicker. Interestingly, there does not seem to be a strong correlation between the pore size and degradation rate, allowing for independent tuning of porosity and degradation rates in silk fibroin scaffolds through freezing rate and post-lyophilization method, respectively.

Given that the pore size is large enough to promote transport and diffusion of the enzyme through the scaffold as long as the pores are interconnected, shifts in degradation rate are hypothesized to be due in part to the availability of cleavage domains. The ability of the enzyme to access cleavage sites is hindered by sterics within the secondary structure of the silk fibroin, with larger crystallite sizes preventing access to amino acids deep within the crystallite regions.⁷⁷ Hydrogen bonding between the beta-sheets to form large crystals does not allow for equal access to cleavable sequences. As the crystallites grow, it is also expected that access to inter-spaced amorphous regions decreases, limiting the binding ability of the enzyme. Specifically, supramolecular interactions, comprised from many silk fibroin chains aggregating, contribute to the lamellar crystalline structures, further limiting the accessibility of cleavage sites. The larger crystallite size in the autoclaved scaffolds increases the activation energy and reduces thermodynamic accessibility, slowing enzyme kinetics and requiring disruption of the crystalline structure. As the amorphous regions are more flexible and accessible, these regions are targeted for degradation first. Biodegradable polyesters show similar limited accessibility for depolymerization through increased crystallinity.⁷⁸ Hydrolytic breakdown of the polymer chains by enzymes are hindered by crystalline domains in more hydrophobic and crystalline high density polyethylene-like materials.^{78, 79} Recently, Schwab et al. tuned polyester crystallinity through the introduction of branched monomers which increased hydrolysis rates as crystallinity decreased.⁸⁰ Overall, the accessibility of cleavage sites in the amorphous region of the secondary structure allows for the water annealed scaffolds to degrade faster than the autoclaved scaffolds, where the larger crystallites, induced through different post-lyophilization steps, inhibit the enzyme from cleaving protected amino acids.

3.3. Developing anisotropic silk scaffolds as NAMs for Muscle Applications

To develop long-term contractile muscle tissues, it is critical to understand the influence of the scaffold during long-term cell culture parameters. Other 3D biomaterial platforms have been utilized for long-term culture, but face some challenges, mainly in terms of continuous mechanical motion and contraction of the scaffold material. For example, hydrogel and sponge systems are popular for 3D culture, with collagen and Matrigel systems being used for muscle applications.^{48, 81, 82} However, hydrogels systems are typically used in conjunction with passive static stretch to align cells, rather than oscillating movement, for functionalization.⁴⁸ While this can still stimulate mechanotransduction pathways, it limits the future use and tunability of the platform toward dynamic applications. Additionally, hydrogel systems have the potential for changes in scaffold volume over time making the separation of passive stretch versus shrinkage effects difficult in 3D evaluation of tissue.⁸¹ Previous work with anisotropic silk fibroin scaffolds hypothesized their ability to maintain their structure and function over long time periods of mechanical stimulation. Aikman et al. demonstrated that the scaffolds could withstand 100 hours of continuous mechanical motions from strains within the linear region.³¹ While this is only an estimate of potential long-term use, this manuscript will expand upon the potential for long-term applications (>2 weeks). Previous work has also investigated engineering parameters to form ice-templated anisotropic silk fibroin scaffolds with pore sizes maintaining ample room for cell infiltration.^{23, 27, 31} The extensional mechanical properties were also found to mimic native skeletal muscle with 1000 to 1600 kPa Young’s modulus.³¹

As the magnitude of strain and frequency dependence has already been determined for the aligned silk fibroin scaffolds and is known to match native skeletal muscle tissue function,³¹ further work in this study will look at the effects of repeated mechanical stimulation over the course of 5 or 10 days. These time frames are motivated by the ideal 5-days of delayed onset of mechanical stimulation and an average of one hour of motion per day commonly shown toward the development of murine and human skeletal muscle tissues in vitro.^{45, 47, 50} Somers et al. has also highlighted the limited data available to confirm appropriate length of stimulation regimens, but the silk materials highlighted here are able to meet and exceed lengths of time commonly found in literature.⁴⁷ Figure 4 shows the timeline of storage and mechanical motion experienced by the scaffolds in cell culture media containing 1% pen-strep and 10% fetal bovine serum (DMEM) or water (H₂O) conditions. First, scaffolds were stored in their respective solution in a well plate for 5 days (5 Day) on a rocking plate in a cell culture incubator (37°C) to mimic the conditions of initial skeletal muscle cell growth and infiltration. Next, the scaffolds were transferred into the MechanoCulture T6 bioreactor with a gap length of 10 mm. Mechanical stimulation was applied for 30 minutes every 11.5 hours at 10% strain and 1 Hz frequency, to mimic native skeletal muscle motion and align with prior in vitro culture studies.^{47, 50} The mechanical motion cycle was repeated for an additional 5 days (10 Day) or an additional 10 days (15 Day).

Figure 4:

Bioreactor and timeline justification. (A) Timeline for scaffolds includes storage in water or media on a shaker plate in the incubator for 5 days. Then scaffolds are added to the MechanoCulture T6 bioreactor (in water or media respectively) for mechanical stimulation at 10% strain and 1 Hz frequency for 30 minutes every 11.5 hours. Created in BioRender. (B) Images of the bioreactor loaded with scaffolds containing water (H₂O) or media (DMEM).

Silk fibroin is a bioinert protein that contains no native cell adhesion sites.^{8, 27, 83, 84} To accommodate the future use of cellular components in these scaffolds toward in vitro NAMs or in vivo implantation, characterization was completed with the addition of decellularized extracellular matrix (dECM) components, aligning with prior formulations in the literature.^{27, 31} Skeletal muscle extracellular matrix components are vital for skeletal muscle growth, contraction, and normal physiological function.⁸⁵ For these dynamic systems, where mechanical recapitulation is vital for mimicking skeletal muscle, the mechanical forces and stimulation regimens are mediated through cell-material interactions and mechanotransduction pathways.⁸⁶ Thus, integrin binding sites and other adhesive cues in the dECM are necessary to promote adhesion in the biomaterial scaffold.⁸⁷ The extracellular matrix in this work is derived from decellularized porcine skeletal muscle, but other growth factors could be used such as those shown in Jameson et al. where prefabrication additions allow the factor to be entangled in the polymer matrix or post-fabrication methods where the scaffold can be soaked in a solution containing the factor.³⁴ Here the dECM is incorporated into the silk fibroin polymer solution prior to scaffold formation. More specific factors could be incorporated depending on the final application or intended use of the scaffold.

Previous tensile assessments of anisotropic ice-templated silk fibroin scaffolds were performed hydrated in water or only 24 hours of storage in different solutions (PBS, DMEM, FBS) with prior results predicting potential for long-term use.³¹ Since the use of these scaffolds as cellularized materials would require use in physiologically relevant conditions, tensile testing was expanded here to determine the influence of mechanical loading and storage in water or cell culture media conditions over longer time frames. The scaffolds were made at the same fabrication parameters of 30-minute degummed silk fibroin at 5% concentration with 0.2 mg/mL dECM content with water annealing post-lyophilization steps used in literature.³¹ In addition to previous literature, here the silk fibroin scaffolds were sterilized through liquid autoclaving to mimic the conditions necessary for the scaffold prior to any cell culture. Thus, all the scaffolds used in the remainder of this work, for the bioreactor experiments and assessment, match the crystalline structure of the water annealing scaffolds with the addition of autoclaving sterilization (dECM, WA postA). The additional autoclaving step produces some changes in crystallinity due to the manipulation of water with the silk fibroin backbone. Other sterilization methods may be used, such as those previously seen with isotropic silk scaffolds, gamma irradiation, ethylene oxide, and hydrogen peroxide gas plasma.⁵⁵ However, access to these other methods were limited for this work. While the addition of this sterilization step makes it difficult to compare absolute values of mechanical property data to previous literature, the assessment included within are more relevant to the use of these aligned scaffolds as NAMs. These sterilized anisotropic dECM silk fibroin scaffolds are used for the remaining assessments within this manuscript.

3.3.1. Mechanical loading and viscoelastic properties over time

As described in Figure 4, scaffolds were stored in water or cell culture media for 5 days before being transferred to the CellScale MechanoCulture T6 bioreactor filled with the same solution. The bioreactor provided mechanical stimulation over the course of 5 days or 10 days before scaffolds were assessed for mechanical properties and crystalline content. Hydrated extensional testing was completed at a rate of 1 mm/min with initial gap length at 10 mm, parallel to the direction of alignment. Representative hydrated stress-strain curves for anisotropic silk fibroin scaffolds pre- and post-autoclaving sterilization at day 0 (Figure 5A) are shown to demonstrate the differences in previous literature measurements³¹ versus the more relevant conditions for new approach methods development shown through sterilization. The pre-autoclaving sterilization (WA) mimics the data shown previously by Aikman et al. at the day 0 time point.³¹ Post-autoclave sterilization conditions at day 0 represent the rest of the scaffolds used throughout this manuscript (WA postA). Beta-sheet analysis from FTIR (Figure S4C) showed significant differences with the addition of autoclave sterilization, which would lead to changes in overall mechanical property values. Mechanical property values show that Young’s modulus, evaluated in the linear region, and ultimate tensile strength, evaluated outside of the linear region are significantly different for the pre- and post-sterilization conditions (Figure S4D, F). Break strain also shows significant differences between pre- and post-autoclave sterilization, however, due to maximum extensional displacement limits, the pre-sterilization conditions did not actually reach their break point and instead the maximum strain displacement is reported (Figure S4E). Anisotropic scaffolds pre-sterilization are shown to have minimal plastic deformation. However, a need to sterilize these scaffolds for cell culture applications through liquid-phase autoclaving alters these initial mechanical properties. The addition of the sterilization cycle increases beta-sheet content from around 50% to 60% (Figure S4C). As shown previously, higher beta-sheet content in porous silk fibroin scaffolds increases mechanical strength of the materials and produces a more brittle response. Day 0 and Day 5 scaffolds still underwent this sterilization step even though they were not attached to the bioreactor to maintain comparisons among all conditions in this manuscript. It is important to note that the resulting changes in plastic deformation and mechanical responses occur at strains above 0.3 mm/mm but are not critical in the analysis of these materials and their mechanical responses, where perturbations in linear region are maintained below 0.2 mm/mm strain.

Representative stress-strain curves demonstrate no observable trends between scaffolds stored in water (H₂O) or cell culture media (DMEM) solutions (Figure 5B). Similarities in stress-strain responses (Figure S5A-E) are seen in the evaluation of mechanical property values (Figure S6, Table S1). The stress-strain response of these materials exhibits a linear region at low strains with plastic deformation at high strains. Deconvolution of FTIR spectra (Figure S7) also shows minimal differences in secondary structure among scaffolds stored for 5 days or with the addition of mechanical motion for 5 or 10 days beyond the initial storage (Figure 5C). Beta-sheet content remains consistent, with an average between 58-64%. While there is some significance in the Day 10 beta-sheet content, the overall broadness of the FTIR peaks and the propagation of error though the deconvolution process makes statistical assessments difficult at this stage in addition to the deconvolution of secondary structures beyond beta-sheets (Figure S7). This significance is attributed to the extremely low standard deviation among data points (n=8) rather than the physical differences in the presence of beta-sheet structures since this effect is not observed through functional mechanical property assessments.

The extensional Young’s modulus values, taken as the slope of the linear region at low strain values (Figure S6), shows statistical similarities across all groups when stored in water or media over time (p=0.9845). Overall, there seems to be a decrease in modulus after introduction to the bioreactor for mechanical motion for all mechanical properties, yet minor difference between storage in water or media. The Young’s modulus (Figure 5D) for Day 10 and Day 15 conditions have similar magnitude to previous literature at short time scales, on the order of 2 MPa. Day 5 conditions have a fold change higher Young’s modulus, on the order of 3 MPa, but this can be attributed to the sterilization in the autoclave that increases beta-sheet content. The non-statistically significant trend toward a decrease in modulus after the introduction to the bioreactor is hypothesized to be due in part to some minor breakage that occurs upon loading. Similarly, the strain at break and ultimate tensile strength (UTS) are not statistically different across the storage solution and time on the bioreactor. The strain at break (Figure 5E) is defined as the strain at which an inflection point occurs on the stress-strain curves (Figure S6). While not statistically significant (p=0.7906), there appears to be a trend of decreasing strain at break with increasing time, as well as a decrease in magnitude when stored in media compared to water. The ultimate tensile strength is measured as the maximum stress achieved by the scaffolds (Figure S6). The ultimate tensile strength (Figure 5F) exhibits a similar responses to the strain at break, with a trend in magnitude that decreases over time and shows minimal differences between storage in water or media, although the differences are statistically insignificant (p=0.5871). Thus, future experiments with long-term culture exceeding 15 days may warrant additional testing. Extended storage with mechanical motion for 25 days (Day 30) (Figure S5E-F) shows that the same trends hold, with minimal differences in mechanical response over extended storage periods and stimulation in water or media (Table S1).

Silk fibroin materials are dynamically impacted when stored in PBS or DMEM, due to changes in salt concentration affecting silk fibroin self-assembly and amino acid interactions over time.⁸⁸ The addition of charges in solution alters hydration at the protein backbone, influencing stability of secondary structures such as beta-sheet structures.¹⁰ The biomaterial format impacts the extent of protein folding changes, with hydrated systems, such as hydrogels having the greatest impact.^{7, 10, 61} Nevertheless, this has not been thoroughly reported in ice-templated silk fibroin scaffolds. Surprisingly, the addition of salt to the storage solution does not significantly alter extensional viscoelastic properties in 30-min, 5% anisotropic scaffolds over time. It is hypothesized that the post-lyophilization processing steps (water annealing or autoclaving) and the secondary sterilization steps produce a maximum achievable crystallinity level, minimizing protein rearrangement. Together with the high polymer density found within these, it is hypothesized that minimal secondary structure rearrangement occurs over a 30-day time frame. This retainment of crystalline level and organization over time is unique for this type of silk fibroin material and likely indicative of the beta-sheet organization during post-lyophilization steps. Other types of silk fibroin biomaterials are highly dependent on the addition of salt and polymer reorganization over time due to chain mobility, within hydrogel and film materials.^{7, 27, 89} However, chain mobility is not present after the post-lyophilization steps in anisotropic silk fibroin scaffolds (no change in secondary structures, Figure S7A-G). Thus, future chain rearrangement, additional formation of beta-sheet structures, and changes to materials properties over time are not seen in ice templated silk fibroin scaffolds. The addition of loading via the bioreactor produces minimal changes to the material over time. Specifically, since the mechanical loading is set to a strain (10%) and frequency range (1 Hz) that is within the linear range of the material, no significant changes were expected, as this mechanical motion is performed in the elastic region. All mechanical properties were significantly similar values during the time scale and solutions experienced in the bioreactor. With limited chain rearrangement available, due to post-lyophilization processing, and recoverable deformation of the amorphous regions, the crystalline domains remain organized in their initial state and maintain long-term mechanical properties and structure over time. The extensional mechanical properties of the sterile aligned silk fibroin scaffolds, with the addition of physiologically relevant solution and mechanical stimulation, show agreement with the mechanical properties of native skeletal muscle and previous successful short-term in vitro systems,⁴⁷ indicating this combination of silk fibroin scaffolds and mechanical stimulation may be a suitable long-term mechanotransduction culture platform upon the addition of relevant cells.

3.3.2. Porosity and structural changes due to storage and loading

Given the nonsignificant differences observed through quantification of crystallinity and mechanical properties, the structural differences were investigated through scanning electron microscopy (SEM) and image analysis to confirm that the scaffolds remain intact on the micro (μm) and macro (mm) scale. It should be noted that the SEM (Figure 6) assessments must be performed on the dry, non-hydrated scaffolds, thus there are some limitations to using these scaffolds with SEM, as the scaffolds must be completely dried. Consequently, the pore size may not be representative of the scaffold in a hydrated environment due to swelling and shrinking in a dry or wet environment (Figure S8D).

Figure 6:

Representative scanning electron microscopy (SEM) images (into porosity, side profile, and top down) for aligned 30-minute, 5% (w/v), 6-hour water annealing (WA), silk fibroin scaffolds with dECM. Scale bars are 100 μm (A) freshly cut in water. (B) 5 days in media on a rocking plate. (C) 10 days in media (5 days rocking, 5 days in the bioreactor). (D) 15 days in media (5 days rocking, 10 days on bioreactor). (E) Pore diameter for each media condition as a function of time (n=40, 20 pores randomly assessed from 2 images). Data are expressed as mean ± standard deviation. Analyzed with 2-way ANOVA with Tukey’s post-hoc analysis. Statistical significance is reported as p>0.05. (F) Degree of alignment (n=20, 10 aligned pores randomly assessed from 2 images) as measured through ImageJ.

SEM micrographs were acquired for initial Day 0 scaffolds (Figure 6A) with no storage as well as post-storage and loading end points: Day 5 (Figure 6B), Day 10 (Figure 6C), and Day 15 (Figure 6D). Images were taken on each plane of the scaffold – into the porosity perpendicular to the alignment, along the side parallel to alignment, and from the top down, parallel to the alignment. Observationally, all scaffolds appear similar. Analysis of the pore diameters (Figure 6E) demonstrates no statistical differences in dry feature size. While there is some variability in the range of pore sizes from 0 to 100 μm, there are similarities among all scaffolds, regardless of time spent in storage or on the bioreactor. However, there are some minor, significant differences in pore diameter between the storage solutions (Figure S8A), which could be due to the difference in dehydration during preparation for SEM imaging. The orientation angle of alignment shows little differences between conditions stored in media for the duration of the bioreactor time (Figure 6F) or with the storage in water (Figure S8B). Overall, these results confirm the lack of influence of salt on the scaffold structure and properties within media, highlighting the benefits of these materials as long-term culture platforms. To confirm SEM results in a hydrated state, additional X-ray characterization was explored through nano-computed tomography (nano-CT) to view the three-dimensional structure of the scaffold in a hydrated environment.

In addition to being collected in a hydrated state, the 3D rendering obtained from nano-CT data can be used for multiple assessments through a variety of computational tools. Small representative scaffolds were cut from the long strips maintained in the bioreactor and soaked in iodine for contrast before being scanned. Nano-CT scans were collected in a hydrated state to prevent shrinkage from drying and were converted to 3D reference meshes. Figure 7 shows the 3D renderings of scaffolds for Day 0, Day 15, and degradation after 1 day. Here, pore interconnectivity is demonstrated with diffusion of objects through the pores in the direction of alignment. Following inspiration from Qazi et al.,⁹⁰ spheres of sizes 5 μm, 10 μm, and 50 μm were simulated and dropped through the pores. Building upon previous 3D scans of aligned silk scaffolds,³¹ only one post-lyophilization parameter (WA postA) was chosen for the longer mechanical stimulation time points to confirm no structural changes were found, as observed from the SEM imaging.

Figure 7:

Nano-CT scans of aligned silk fibroin scaffolds with water annealing and autoclave sterilization steps (A-C) 3D reconstruction (D-F) Blender simulation for pore size and interconnectivity assessment of 50 spheres (D) Scaffold at Day 0 (E) Scaffold stored for 5 days and mechanically stimulated for 10 days (Day 15) in water (F) Scaffold stored in 1 U/mL protease XIV solution for 1 day

Fifty spheres of each diameter were simulated and dropped in the direction of scaffold alignment. Figure 7 shows the frequency of total distance travelled by the sphere compared to the original length of the 3D rendering for the scaffolds at Day 0 (Figure 7D, Video S1), Day 15 (Figure 7E, Video S2), and following 1 day of degradation in protease XIV (Figure 7F, Video S3). Comparison between Day 0 and Day 15 shows minimal differences in the number of spheres able to traverse the scaffold. Most spheres were retained in the first 10% to 20% of the scaffold length, highlighting issues with directional pore interconnectivity. While the size of the spheres mimics the pore diameters estimated from the SEM images, the hydrated state was expected to be slightly larger due to swelling of the scaffolds (Figure S8D). However, similarities are seen between scaffolds with (Day 15) and without (Day 0) mechanical stimulation on the bioreactor, demonstrating that mechanical motion in the linear range does not impact pore size even when assessed in the hydrated state. Regardless, computational assessments of the nano-CT meshes show impeded transport of solid spheres through these materials, with increasing depths noted following degradation, highlighting that degradation opens up the internal pore structure. Greater difference was seen in the ability of the spheres to penetrate the scaffold after 1 day of degradation in protease XIV. The increase in pore size seen in Figure 3F, as well as the frequency of spheres that traverse the degraded scaffold, demonstrates bulk erosion of the scaffold from enzymatic degradation as the enzyme solution is able to diffuse through the porous scaffold more readily.⁷⁸ Additionally, a smaller sphere size of 0.5 μm was included (Figure S9) to demonstrate the ability of fluids and the enzyme solution to infiltrate the scaffold regardless of depth and interconnectivity, highlighting the limitations in enzyme accessibility are limited to the crystallite size (nanometer), rather than the diffusion of the fluid throughout the scaffold porosity.

While protease XIV is not produced by skeletal muscle cells, intended in the application of these scaffolds as NAMs, co-culture with other cell types or inclusion of enzymes to the system, may benefit the long-term pore infiltration of the system. The 1 U/mL concentration of protease XIV caused too much degradation of the bulk structure to perform mechanical testing after 1 day of degradation. However, lower activity concentrations or other enzyme types that target different amino acids may present an alternative method to tuning pore size and overall structural properties of the scaffolds after formation. During the scale up of scaffold dimensions to fit the materials in the bioreactor, interconnectivity limitations were found, which may lead to challenges with cellularization in the future. Smaller ice templated scaffolds previously used for cellularization did not experience limitations in cellular infiltration,^{23, 27} but this would need to be explored in future work toward the development of these aligned scaffolds as an in vitro platform. Additionally, other templating techniques exist that may allow for better pore interconnectivity such as magnetic templating,^{91, 92} hollow arrayed channels,²⁵ or even ice-templating at different temperatures.^{23, 31} Furthermore, future cellularization of the anisotropic scaffolds may also see changes in pore size as cells migrate into the scaffold. Previous literature has demonstrated changes in scaffold surface features from ECM deposition due to cellularization in anisotropic silk fibroin scaffolds²⁷ at much smaller dimensions than those generated here to fit in the bioreactor. Additionally, cellular remodeling may decrease porosity as cells lay new ECM during growth and maturation, resulting in changes to mechanical properties and mechanotransduction signaling,⁸⁶ which would warrant new investigations into scaffold structure and mechanical performance after cellularization.

Most importantly, one advantage that silk fibroin porous scaffolds provide to in vitro modeling with mechanical stimulation is the retention of the scaffold’s length and volume. The length of the scaffold does not shrink or elongate with mechanical stimulation (Figure S8C). Maintaining this stimulation within the linear region does not cause any permanent effects on the size and structure of the scaffold. Additionally, there is no compaction or shrinkage, which can be seen in passive hydrogel systems.⁸¹ Overall, the SEM imaging and 3D nano-CT scans show limited changes to the porosity and structure through storage and mechanical stimulation in the bioreactor. This would lead to the hypothesis that for the use of anisotropic silk fibroin scaffolds as NAMs, any environmental changes experienced by cells or tissue would be dictated solely by mechanical motion, rather than by changes resulting from scaffold breakdown, leading to a robust cellularized platform.

4. Conclusion

This work determined the tunability and impacts of crystallinity and long-term perturbations on anisotropic silk fibroin scaffolds. The tunability of silk fibroin scaffolds is driven by the post-lyophilization crystalline induction method. The induction methods remove bound water from the silk fibroin polymer backbone causing lamellar beta-sheet formation. Crystalline content and crystallite sizes were confirmed via XRD and showed shifts in intra- and inter-molecular interactions within these scaffold formulations. This thermodynamic phenomenon caused the formation of larger crystallite sizes with the inclusion of heat (A), whereas room temperature conditions for longer time periods produced the same crystalline content but smaller crystallite sizes (WA).

The differences in crystalline beta-sheet organization causes changes in degradation and mechanical properties. Larger crystallites achieve slower degradation kinetics and more plastic deformation whereas, the smaller more frequent crystallites have a faster degradation rate and more elastic mechanical response. The tunable degradation rates highlight the limited thermodynamic accessibility to protease XIV in large crystalline domains. The extensional mechanical properties of anisotropic silk fibroin scaffolds were shown to remain consistent through the addition of long-term mechanical perturbations. Mechanical stimulation was applied, in the linear region of the scaffolds, to mimic the extensional forces of native skeletal muscle movement. Static extensional properties and internal porosity were not affected by these mechanical perturbations over the course of 30 days. Additionally, storing the silk fibroin scaffolds in multiple solution types, to mimic physiologically relevant conditions, did not affect the long-term stability of the scaffolds over 30 days. Overall, structure-function relationships were found between differences in crystallite size and organization impacts on degradation, where crystallinity was able to maintain structural and mechanical performance within the linear, elastic, region of the materials. While long-term mechanical stimulation was performed only on 30-minute degummed silk fibroin formed into aligned water annealed scaffolds, the organization and crystallite size determined, connects polymer structure and time for beta-sheet organization through quantifiable results and predictability of these materials as long-term NAMs platforms, which may be expanded to other silk-fibroin scaffold formulations in the future.

Successful infiltration and cellularization of silk fibroin biomaterials in literature have demonstrated the intended future use of these scaffolds as cellularized in vitro NAM platforms. Cellular responses in silk fibroin biomaterials have previously shown how mechanical properties and biological component additions to the silk fibroin polymer systems support differentiation and maturation outcomes. While the scope of this work was focused on developing structure-function relationships for beta-sheet organization and resulting mechanical and structural properties, future work should include the addition of cellularization and assessment of the infiltration and mechanical outputs. Cells are hypothesized to infiltrate and adhere to the scaffolds based on relevant pore sizes and dECM inclusion as well as the successful recapitulation of native skeletal muscle mechanical properties and mechanical stimulation regimens. Due to the crystallite organization from post-lyophilization methods, mechanical properties and long-term mechanical response of anisotropic silk fibroin scaffolds are not expected to significantly change upon cellularization with skeletal muscle myoblasts.

Silk fibroin biomaterials have previously been utilized for muscular applications due to tunable pore sizes and mechanical properties, as well as their biocompatibility, non-toxic effects, and tunable degradation. Here the tunability was further explored through post-lyophilization methods, with crystalline organization and size determined for the first time in anisotropic silk fibroin scaffolds. This work lays the foundation for the development of anisotropic silk fibroin scaffolds as long-term in vitro new approach methods for soft tissue applications requiring mechanical motion.

Data Availability

Raw data used in this manuscript can be accessed here.