Tendon-Derived Matrix Crosslinking Techniques for Electrospun Multi-Layered Scaffolds

Abstract

Tendon tears are common and healing often occurs incompletely and by fibrosis. Tissue engineering seeks to improve repair, and one approach under investigation uses cell-seeded scaffolds containing biomimetic factors. Retention of biomimetic factors on the scaffolds is likely critical to maximize their benefit, while minimizing the risk of adverse effects, and without losing the beneficial effects of the biomimetic factors. The aim of the current study was to evaluate cross-linking methods to enhance the retention of tendon-derived matrix (TDM) on electrospun poly(ε-caprolactone) (PCL) scaffolds. We tested the effects of ultraviolet (UV) or carbodiimide (EDC:NHS:COOH) crosslinking methods to better retain TDM to the scaffolds and stimulate tendon-like matrix synthesis. Initially, we tested various crosslinking configurations of carbodiimide (2.5:1:1, 5:2:1, and 10:4:1 EDC:NHS:COOH ratios) and UV (30s 1 J/cm², 60s 1 J/cm^2, and 60s 4 J/cm²) on PCL films compared to un-crosslinked TDM. We found that no crosslinking tested retained more TDM than coating alone (Kruskal-Wallis: p>0.05), but that human adipose stem cells (hASCs) spread most on the 60s 1 J/cm² UV- and 2.5:1:1 EDC-crosslinked films (Kruskal-Wallis: p<0.05). Next, we compared the effects of 60s 1 J/cm² UV- and 2.5:1:1 EDC-crosslinked to TDM-coated and untreated PCL scaffolds on hASC-induced tenogenesis. UV- crosslinked scaffolds had greater modulus and stiffness than PCL or TDM scaffolds, and hASCs spread more on UV-crosslinked scaffolds (ANOVA: p<0.05). FTIR spectra revealed that UV- or EDC-crosslinking TDM did not affect the peaks at wavenumbers characteristic of tendon. Crosslinking TDM to electrospun scaffolds improves tendon-like matrix synthesis, providing a viable strategy for improving retention of TDM on electrospun PCL scaffolds.

1. Introduction

Rotator cuff (RC) tears are a leading cause of function-limiting pain and shoulder dysfunction. Over 300,000 RC repairs are performed annually in the USA,¹ but re-tear rates range from 20% to 94%, depending on the size of the tear, and a variety of other factors.^2–5 Furthermore, 30% of RC tears are irreparable,⁶ and the need for more effective surgical options is increasing as the population ages and maintains high functional demands. Even successful surgical treatment results in fibrotic scar tissue formation, not tendon regeneration, which can result in on-going pain and loss of motion, and increased risk for re-tear.^{2, 7–9} Extracellular matrix patches, whether derived from xenografts or allografts, are used to augment repairs but cell infiltration and remodeling are slow, and consistently result in fibrosis.^{6, 10, 11} The optimal properties of extracellular matrix grafts for rotator cuff repairs are not yet fully defined.¹² Tendon tissue engineering seeks to address this gap.

Tendon tissue engineering often incorporates synthetic polymer scaffolds as part of a strategy to induce a more regenerative tendon-like phenotype. Non-woven methods of fabricating these scaffolds are common, and include electrospinning approaches.^13–19 Multilayered electrospun scaffolds facilitate rapid cell infiltration and deposition of a tendon-like matrix by human adipose-derived stem cells (hASCs).^{9, 13, 20} While synthetic polymer scaffolds offer the ability to control microarchitectural properties, they do not provide biological signals to direct cell behavior.²¹ Such signals can be incorporated via biomimetic factors, including various growth factors or biologically-sourced materials such as tendon-derived matrix (TDM). TDM, developed by powdering/pulverizing or dissolving native tendon tissue, stimulates tenogenesis,^{13, 22–24} but is not fully retained on TDM-coated, poly(ɛ-caprolactone) (PCL) scaffolds during in vitro culture.¹³ The loss of TDM or other biomimetic materials from a biomaterial leads to concerns relating to potential for aberrant differentiation or immune responses if this approach were used in vivo. Extracellular matrices have been used as both scaffolds and incorporated into synthetic scaffolds in tissue engineering, but most methods to produce ECM causes a loss of ECM integrity and changes its mechanical and degradation properties. Crosslinking ECM proteins becomes a vital tool when designing scaffolds to control for mechanical properties, degradation rates, and immunogeneity.^{25, 26} There are physical (e.g., ultraviolet radiation or dehydrothermal) and chemical (e.g., carbodiimide or glutaraldehyde) crosslinking methods to stabilize ECM proteins, but some crosslinking methods have negative effects on differentiation and cell attachment.^27–29 Carbodiimide chemical crosslinking targets the carboxyl groups in amino acid side chains like the aspartic acid (D) in the RGD motif commonly found in the ECM protein fibronectin³⁰ or the glutamate (E) in found in integrin binding motifs found in collagens.³¹ Carboxyl group (COOH) crosslinking is popular since it is not toxic and the byproducts can be easily removed by washing.³² In this reaction, the COOH conjugates with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) to form an active ester. Then, it reacts with a primary amine from adjacent amino acids in collagen forming the crosslink with an amide bond and producing an isourea byproduct.³³ The isourea byproduct is unstable, therefore, N-hydroxysuccinimide (NHS) is used to form a more stable ester, while also improving crosslinking efficiency.³³ Nevertheless, the use of carboxylate groups and primary amines as cross-linking substrates can reduce bioactivity.³⁴ Ultraviolet (UV)-light produces radiation that cross-links aromatic side chains, such as phenylalanine and tyrosine,^{31, 34} and is frequently used for stabilization of biological matrices. These crosslinking methods could bind the TDM to itself, attaching it around the multilayered electrospun fibers to improve retention on the scaffold. The objectives of this study were to screen the effect of these crosslinking techniques on 1) retention of TDM, and 2) biomimetic effects of TDM on PCL films and multilayered electrospun PCL scaffolds.

2. Materials and Methods

2.1. Tendon-Derived Matrix

Flexor and extensor tendons were harvested from adult female porcine hind limbs obtained from a local slaughterhouse. The tissue was minced, lyophilized, and pulverized (6750 Spex SamplePrep Freezer Mill; Spex CertiPrep) before being placed in a sieve to pass through a 75-µm wire mesh. The sieved TDM powder was stored at −80°C until use. We made a TDM-slurry to coat the films and scaffolds by mixing 0.1 g TDM into 10 mL PBS (1% w/v), and stirring it for 5 days at 4°C.

2.2. Poly(ɛ-caprolactone) Films

Poly(ɛ-caprolactone) films (PF) were created using 5% w/v PCL in 7:3 dichloromethane:ethanol. PCL solution was deposited into custom polydimethylsiloxane molds attached to glass slides and allowed to dry for 24 hours. Dried PCL films were then coated with 1% TDM in PBS and allowed to dry for 48 hours or coated with PBS and allowed to dry for control groups. TDM coated films were left un-crosslinked (TF) or crosslinked using UV light (UF) or chemically using a carbodiimide solution (EF).

2.3. Electrospun Multilayer Scaffolds

Multilayered electrospun PCL scaffolds were prepared as described previously.¹³ Briefly, poly(ɛ-caprolactone) (440744, Millipore Sigma) was dissolved in a solution of 70/30 dichloromethane/ethanol at 10% w/v for 24 hours. PCL solution was electrospun into a 1.25 g/l NaCl saline bath through a 25-gauge needle at 17 kV. Layers were collected from the surface of the bath at 2-minute intervals for a total of 35 layers. Scaffolds were then cut into 0.5×4 cm test strips and rehydrated in a series of ethanol baths then stored in PBS until TDM coating or cell seeding. We coated TDM onto the scaffolds at 2 mg TDM/cm² by dipping them into 1% TDM solution, and then dried them for 48 hours before use.

2.4. Crosslinking

2.4.1. PCL Films

After coating with TDM, chemical crosslinking was achieved using three different carbodiimide solutions (14mM 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) and 5.5 mM N-hydroxysuccinimide (NHS); Sigma). Crosslinking solutions were made with three different ratios of EDC:NHS:COOH: 2.5:1:1 (EF2.5), 5:2:1 (EF5), 10:4:1 (EF10). For all ratios, TDM films were incubated in crosslinking solutions for 4 hours, then incubated in 0.1M Sodium Phosphate solution for 2 hours to hydrolyze remaining COOH,³⁵ then washed with PBS four times for an hour each time. In films purposed for biochemical content assays, washes were collected.

UV crosslinking was achieved using a Spectrolinker XL-1000 (Spectroline) crosslinker with a 254 nm source in three configurations: 1 J/cm² energy concentration for 30s (UF30) or 60s (UF60) at 18 cm distance from source, or for 60s at half distance (9 cm) (UFHD) from source (4 J/cm²). These cross-linking times were used to accommodate the possibility of future cross-linking each individual layer of multi-layered scaffolds during fabrication (2-minute intervals between collection of each layer).

For cell attachment film assays, PCL films (PF), TDM-coated, un-crosslinked films (TF), uncoated glass slides (negative control), fibronectin coated glass (positive control), and crosslinked films of each type (EF2.5, EF5, EF10, UF30, UF60, UFHD) were blocked with 7.5% Bovine Serum Albumin. Both sides of each film were sterilized under UV light for 10 minutes, then pre-wetted with PBS prior to seeding.

2.4.2. Multilayered Electrospun Scaffolds

TDM group scaffolds were allowed to dry after TDM coating, while crosslinked scaffolds were prepared in the same way as crosslinked films. Crosslinking methods were selected based on the most promising results obtained from PCL films (Section 3.2). UV crosslinked scaffolds were exposed to an energy concentration of 1 J/cm², with an 18 cm distance from the source, for 60 seconds on each side of the scaffolds. Chemical crosslinking was done at 2.5:1:1 EDC:NHS:COOH concentration. All scaffolds were rinsed in PBS after coating or cross-linking, then allowed to dry and were stored at 4^oC until use. Prior to cell seeding, scaffolds were surface sterilized with UV light for 10 minutes on each side and prewet prior to cell seeding or characterization through spectrometry or biochemistry.

2.5. Cell Culture

Human Adipose Stem Cells (hASCs) previously isolated from lipoaspirate waste from three de-identified female donors^{13, 22} (IRB Exempt) were combined to form a superlot and used at passage 4. On films, hASCs were seeded at 20,000 cells per film. On scaffolds, hASCs were seeded at 0.5×10⁶ cells/cm² on each side, with 15 minutes incubation between seeding each side to allow for cell attachment to the scaffold. Scaffolds were incubated in 6 well plates coated in 2% agarose and suspended between two Teflon rings. Cell culture media (Advanced DMEM (Life Technologies) with 1% penicillin-streptomycin-fungizone (Life Technologies), 4mM L-Glutamine (Life Technologies), and 15 mM l-ascorbic acid-2-phosphate (Sigma)) included no exogenous growth factors. Scaffolds designated for cell attachment assays were cultured with no FBS, all others contained 10% FBS (Hyclone). Scaffolds were cultured for up to 28 days, with media changes every third day. Media was collected at each change for biochemical analysis. Scaffolds were harvested for cell attachment at 0, 4, and 72 hours, fixed, stained, and imaged for histology. Additional scaffolds were incubated for 28 days, harvested, and digested to determine dsDNA, sulfated glycosaminoglycan (s-GAG), and hydroxyproline content after culture.

2.6. Biochemical Assays

To appraise crosslinking efficiency, PCL films were incubated in PBS at 37°C for 72 hours, then the PBS solutions and films were harvested, films were lyophilized, dry weight obtained, then solutions and films were digested for 1 week in papain (125 µg/mL) at 60°C. For assessment of elution of extracellular matrix from the scaffolds, the scaffolds were lyophilized, dry weight obtained, and then the scaffolds were placed in PBS at 37°C for 7 days. The PBS was mixed 1:1 with papain and digested for 18 hours at 60°C. dsDNA content was quantified using the Picogreen Assay (Life Technologies), s-GAG content was quantified using 1,9-dimethylmethylene blue,³⁶ and hydroxyproline content was quantified as previously described using a 7.46:1 ratio to calculate the amount of collagen.^{13, 37}

2.7. Cell Attachment and Spreading

Films were harvested at 24 hours after seeding. Multilayered electrospun scaffolds were harvested at 0 hours, 4 hours, and 72 hours. All films/scaffolds were fixed in 4% paraformaldehyde for ten minutes and stored at 4°C until staining and imaging. Fixed films were stained with Hoechst 33342 (Fisher) and Acti-stain 488 Phalloidin (Cytoskeleton Inc.), then the center of each film was removed using an 8mm biopsy punch and the punched section was immobilized to a glass slide for imaging. Fixed scaffolds were immobilized to glass slides for imaging. Samples were stained for nuclei and actin and three images per sample were taken on a Zeiss Axiovert S100 microscope (Carl-Zeiss). Images were imported into FIJI (2.3.0/1.53t) and the Bio-Formats plugin was used to calculate cell number and surface area. Mean cell count and average single cell area (total actin area/cell number) are reported for each treatment.

2.8. Scanning Electron Microscopy

Four 2 cm x 5 cm strips were cut from PCL, TDM-coated (TDM), carbodiimide crosslinked (EDC), and UV crosslinked (UV) electrospun scaffolds and dried. Strips were sputter coated in gold and imaged using a Hitachi S-4800 Field Emission SEM. Six images were taken randomly throughout each sample.

2.9. Mechanical Testing

Scaffolds were cut into dog-bone shapes along the expected direction of fiber alignment (n=5/group). Verhoeff stain lines were placed at the center and 5 mm on either side of the center using India ink to allow optical strain analysis.^{13, 14} Initial scaffold thickness was measured using a HR-2000M camera (Emergent Vision Technologies, Inc) and digital calipers in FIJI (2.3.0/1.53t). Prewetted scaffolds were sandwiched in 120-grit sandpaper, mounted, and clamped in the load frame (Acumen 3, MTS). Each sample was given a 0.5 g preload, followed by tensile strain to failure at a 1%/s strain rate on a 25 N load cell (MTS). Midsubstance stretch was calculated from digital images acquired at 50 Hz and interpolated to load frame data using custom MATLAB code and Microsoft Excel to calculate the linear region modulus, yield stretch, yield stress, and stiffness, as described previously.^{13, 14}

2.10. Fourier Transform Infrared Spectroscopy

Fourier transform infrared spectroscopy (FTIR) (Nicolet NEXUS 470, ThermoFisher) was performed for scaffolds across wavenumber from 800–4500 cm⁻¹ with a Diamond ATR crystal. Background spectra were collected before each session and scaffold data were corrected to background. Scaffold data were normalized to zero at 4500 cm⁻¹ wavenumber and the mean of three spectra for each group was plotted. We compared data for Amide I (~1630 cm⁻¹) and Amide II (~1560cm⁻¹), the most common absorption bands of collagen. For each amide, we analyzed the peak and the area under the curve (AUC). Additionally, we compared the 1655/1690 ratio, which correlates to collagen organization.³⁸

2.11. Statistical Analyses

We used JMP Pro 14 (SAS, Cary, NC) for all statistical analyses. We tested data for normality using Shapiro-Wilks test and equal variance using Brown-Forsythe test. We report all data as mean ± standard deviation. Parametric data were compared using analysis of variance (ANOVA) followed by Tukey post-hoc testing to determine differences between groups. We compared nonparametric data using the Kruskal-Wallis or the Friedman test followed by Wilcoxon Each Pair post-hoc testing. For all tests, we used α = 0.05 and had n > 3 scaffolds or films for all tests.

3. Results

3.1. PCL Film Biochemistry

We found no difference between TDM-coated and crosslinked films in the amount of dsDNA (p = 0.0948), s-GAG (p = 0.2007), or collagen (p = 0.2361) retained on the films after a 72-hour incubation (Figure 1). More dsDNA and s-GAG were washed off the scaffolds than retained across all scaffolds, while more collagen was retained on the on films than washed off (Figure 1C). Crosslinking had no effect on collagen retained (p = 0.1988) or collagen washed off (p = 0.6201).

Figure 1:

Box and whisker plots of dsDNA (A), sulfated glycosaminoglycans (s-GAG) (B), and collagen (C) content retained on films after a 72-hour incubation in PBS at 37°C or removed during washes while preparing TDM-coated (TF) or EDC- or UV-crosslinked PCL films (n=3). Groups with different letters above are significantly different from each other (Kruskal-Wallis, Wilcoxon Each Pair post-hoc test).

3.2. Films Attachment and Spreading

hASCs attached to all film substrates and spread over 3 days (Figure 2). More hASCs attached to films with TDM (coated or crosslinked) or fibronectin (positive control) than to PCL films or glass (negative control) (Figure 2B). There was no difference between crosslinking group on number of hASCs, and cell count was not different between positive control (fibronectin) and all TDM coated groups. hASCs spread to a greater extent on the EDC:NHS:COOH – 2.5:1:1 ratio – crosslinked films (EF2.5), the UV crosslinked scaffolds at 60s exposure at 1 J/cm⁻² (UF60), and the fibronectin coated films (positive control) compared to the other groups, including uncoated PCL and glass (negative control). hASC area did not differ between the un-crosslinked TDM-coated films and the remaining crosslinked films (EF5, EF10, UV30, UVHD). As crosslinking had no effect on TDM-retention in PCL films (Section 3.1), we compared the EDC 2.5:1:1 and UV 60s 1 J/cm² crosslinking methods to uncoated (PCL) and TDM-coated un-crosslinked multilayered electrospun scaffolds because they led to greater cell spreading than TDM alone and the increased exposure of the cells to the TDM could lead to better cell proliferation and biosynthesis.²⁷

Figure 2:

Representative images for hASCs cultured on their respective substrates after 72 hours (A). Green labels actin (Phalloidin-488) and blue labels nuclei (Hoechst 33342). Scale bar = 100 μm. Box and whisker plots of cell number per image (B) and average cell area (C) (total cell area divided by total number of cells/image; n=10–15 cells/image for glass and PCL films; n=56–152 cells/image for all other groups; 3–9 images per substrate). Groups with different letters above are significantly different from each other (Kruskal-Wallis, Wilcoxon Each Pair post-hoc test).

3.3. Scaffold TDM Retention

SEM images reveal qualitatively similar TDM in the TDM coated, EDC-, and UV-crosslinked samples; no TDM was visible on the PCL scaffolds (Figure 3). The remaining part of each scaffold used for SEM underwent PBS washes for 7 days to measure loss of collagen from the scaffolds.

Figure 3:

SEM images for PCL, TDM-coated, UV-crosslinked, and EDC-crosslinked multilayer electrospun scaffolds. Graphs show the collagen lost into PBS over 7 days. Groups with different letters above are significantly different from each other (Friedman Test, Wilcoxon Each Pair post-hoc test).

3.4. Mechanical Testing

Representative stress-stretch curves (Figure 4A), and Young’s modulus data (Figure 4B) show that incorporating TDM by crosslinking improved the linear-region modulus compared to PCL scaffolds (Figure 4B). UV crosslinking also improved the modulus compared to TDM coated scaffolds, while EDC crosslinking trended toward a greater modulus compared to TDM (p=0.0853). UV crosslinking improved stiffness compared to PCL and TDM scaffolds (Figure 4C), while the EDC crosslinking was intermediate – not significantly greater than PCL or TDM but not significantly different from the UV crosslinked scaffolds. Incorporating or crosslinking TDM had no effect on yield stretch (p=0.0773; Figure 4D) or yield stress (p=0.7543; Figure 4E).

Figure 4:

Representative stretch-strain curve (A) PCL, TDM-coated, UV-crosslinked, and EDC-crosslinked multilayer electrospun scaffolds. Mean ± SD linear region modulus (B), stiffness (C) yield stress (d), and yield stretch (e) in PCL, TDM, EDC, and UV electrospun scaffolds (n = 3–5). Groups with different letters above are significantly different from each other (ANOVA, Tukey’s post-hoc test).

3.5. Multilayered Electrospun Scaffold Biochemical Assays

On all seeded scaffolds, hASCs proliferated similarly over 28 days in culture based on the increase in dsDNA content by day 28 compared to day 0 (p=0.0011) or day 7 (p=0.0071) (Figure 5A), although there was no effect from treatment (p=0.1127). sGAGs were increased at all timepoints compared to day 0 (p<0.0001) though TDM coating or crosslinking did not improve sGAG compared to PCL within any timepoint (Figure 5B). Carbodiimide crosslinking the scaffolds (EDC) resulted in higher collagen at Day 28 (Figure 5C) compared to the PCL alone but not compared to TDM or UV, while neither the TDM or UV led to greater collagen than PCL at Day 28. dsDNA and s-GAG content was higher across all scaffolds at Day 28 compared to all other time points, and s-GAG and collagen was lower across all groups at Day 0 compared to all subsequent time points. In unseeded scaffolds, there were no changes over time for DNA (Figure 5D), s-GAG, (Figure 5E), or collagen (Figure 5F), although untreated PCL scaffolds had less collagen than the TDM-coated and UV- or EDC-crosslinked scaffolds (p = 0.0028). DNA, s-GAG, and collagen were significantly lower in unseeded scaffolds (Figure 5D-F) than in seeded scaffolds (Figure 5A-C). The UV scaffolds had significantly more collagen loss on Day 0 after 4 hours of incubation into the PBS wash, than PCL, TDM or EDC scaffolds, which lost no detectable collagen on Day 0 (Figure 3). Days 1 and 7 also saw no detectable collagen loss – from any treatment group. On Day 3, all groups lost collagen to the PBS, but there was no difference between treatments (p>0.05). Crosslinking the TDM to itself did not improve TDM loss. Finally, we measured ds-DNA, s-GAG and collagen within the media on Days 0, 3, 6, 9, 12, 15, and 27. There was no difference between treatment groups for the total dsDNA (Figure 5G), s-GAG (Figure 5H), or collagen (Figure 5I). Seeded scaffolds lost more s-GAG into the media beginning on Day 9 (p < 0.05) and more collagen over all time points (p < 0.05) than was lost from unseeded scaffolds.

Figure 5:

Box and whisker plots of dsDNA, sulfated glycosaminoglycans (s-GAG), and collagen content for seeded scaffolds (A-C), unseeded scaffolds (D-F), and culture media (G-I) over 28 days in culture (n=3). Groups with different letters above are significantly different from each other; # denotes seeded greater than unseeded (Friedman Test, Wilcoxon Each-Pair post-hoc test).

3.6. hASC Cell Attachment and Spreading

hASCs attached to the multilayer electrospun scaffolds and spread over three days in culture (Figure 6). There was no significant difference in cell number attached to the surface of scaffolds over time (p=0.57) or with treatment (p = 0.055). The hASCs spread more by 72 hours post-seeding than at earlier time points (p < 0.0001). hASCs spread to a greater extent on UV-crosslinked scaffolds at 72 hours compared to other treatment groups (p =0.003).

Figure 6:

Representative images for hASCs cultured on PCL, TDM-coated, UV-crosslinked, and EDC-crosslinked multilayer electrospun scaffolds after 72 hours (A). Green labels actin (Phalloidin-488) and blue labels nuclei (Hoechst 33342). Scale bar = 100 μm. Box and whisker plots of cell number (B) and average cell area (C) (total cell area divided by total number of cells/image; n=17–169 cells/image; 3 images per substrate). Groups with different letters above are significantly different from each other (Friedman Test, Wilcoxon Each-Pair post-hoc test).

3.7. Fourier Transform Infrared Spectroscopy

The spectra for the TDM, UV, and EDC scaffolds were more similar to each other than to the PCL scaffolds (Figure 7). All scaffolds presented peaks related to the ester group in PCL at 1721–1722 cm⁻¹ (Figure 7B). There was no difference in Amide I or Amide II peak wavenumber between TDM, UV, and EDC (Figure 7D; Table 1). The TDM, UV-, and EDC-crosslinked scaffolds exhibited Amide II peaks at 1548–1551 cm⁻¹, which were not present in the PCL samples. Additionally, the TDM, UV, and EDC scaffolds exhibited a peak at 3288–3294 cm⁻¹ for Amide A that was not present in the PCL samples (Figure 7B-C; Table 1). Area under the curve for Amide I and Amide II were greater in TDM, UV, and EDC samples than in PCL samples (p<0.0001), however, crosslinking with UV or EDC did not change the area under the curve compared to coating with TDM alone. There was no difference between TDM, UV, and EDC samples for the 1655/1690 ratio and all were greater than the 1655/1690 ratio on the PCL scaffolds.

Figure 7:

Fourier transform infrared (FTIR) spectra for PCL, TDM-coated, UV-crosslinked, and EDC-crosslinked multilayer electrospun scaffolds, individually (A) and grouped (B). Close up inspections of Amide A area (C) from wavenumber 2800–3700 cm⁻¹ and Amide I and II bands (D) from wavenumber 1510–1700 cm⁻¹.

Table 1:

Fourier transform infrared (FTIR) data.

	PCL	TDM	UV	EDC
Amide I Peak (cm−1)		1635 ± 7.09	1629 ± 2.11	1633 ± 3.61
Amide II Peak (cm−1)		1551 ± 1.93	1549 ± 1.93	1551 ± 1.61
Amide A Peak (cm−1)		3296 ± 9.25	3288 ± 9.39	3294 ± 4.52
Amide I AUC	5.84 ± 0.26	17.61 ± 3.43*	20.90 ± 2.98*	18.17 ± 2.38*
Amide II AUC	1.22 ± 0.05	10.17 ± 3.73*	14.28 ± 3.77*	10.29 ± 2.20*
Amide A AUC	12.10 ± 0.08	50.62 ± 11.80*	61.21 ± 11.31*	51.25 ± 7.48*
1655/1690 Ratio	0.40 ± 0.01	1.88 ± 0.52*	2.40 ± 0.46*	1.89 ± 0.16*

PCL = poly(ε-caprolactone); TDM = tendon-derived matrix; UV = ultraviolet crosslinked; EDC = carbodiimide crosslinked; AUC = area under curve.

^*denotes greater than PCL (ANOVA, Tukey’s post-hoc test).

4. Discussion

In this study, we demonstrated that carbodiimide crosslinking TDM to PCL electrospun scaffolds led to increased collagen content at Day 28 compared to PCL scaffolds, unlike coating with TDM alone, or UV crosslinking TDM to the PCL scaffolds. Neither coating with nor crosslinking TDM to the scaffolds stimulated matrix synthesis of glycosaminoglycans or promoted cell proliferation. Crosslinking with UV increased scaffold modulus and stiffness, while carbodiimide crosslinking trended toward improving scaffold modulus. Compared to TDM coating alone, the doses of carbodiimide or UV crosslinking tested did not enhance ECM retention on PCL films, nor did they reduce ECM lost from films or from electrospun scaffolds. Moreover, the crosslinking methods used did not impair the beneficial effects of TDM on numbers of cells attaching or on cell spreading, while 2.5:1:1 EDC and UV 60 demonstrated additional benefits compared to TDM for cell spreading on films, and of UV 60 on electrospun scaffolds. FTIR revealed similar spectra between the TDM, UV, and EDC scaffolds, with a distinct collagen signature that was not present on the PCL samples. The 1655/1690 cm⁻¹ ratio was similar in TDM, UV, and EDC scaffolds, suggesting that crosslinking at these does did not affect collagen alignment in the coated TDM. Thus, while the effects of crosslinking on TDM retention on the scaffolds were more modest than anticipated based on other studies,²⁷ these lower EDC doses were not detrimental to cell attachment, cell spreading, or to matrix synthesis. The collagen release for unseeded scaffolds in PBS was near zero for all groups at all timepoints except the UV scaffolds at Day 0 (Figure 3); this finding was confirmed in the unseeded scaffolds by measuring collagen lost to media. It is unclear why only the UV scaffolds released collagen on Day 0. The Day 0 release of collagen was not seen in unseeded scaffolds in media (Figure 5I) on a different batch of scaffolds, suggesting differences in handling for the upstream applications of these specific scaffold specimens. Furthermore, collagen release from unseeded scaffolds in the media was low, suggesting no evidence of matrix metalloproteinase activity in the media or from the TDM.

ECM provides 3D structure and support for cell and tissue biological, physical, and mechanical function. ECM allografts and xenografts have been used to augment rotator cuff repair, exhibiting lower retear rates and improved healing compared to rotator cuff repair alone, but need more clinical trials to understand their full impact, and cell infiltration and remodeling is slow.³⁹ ECM xenografts provide excellent biological support in vitro, but do not provide the needed level of mechanical support. Crosslinking ECM can improve the stiffness resulting in improved function.^40–42 However, in a rat model, a crosslinked small intestine submucosa ECM patch (CuffPatch, Arthrotek) was associated with an inflammatory response, while an uncrosslinked small intestinal submucosa ECM patch (Restore, DePuy Synthes) exhibited cell infiltration and tissue remodeling. In clinical trials, the Restore patch demonstrated limited functional improvement and proved unsuitable for augmenting larger rotator cuff tears.^{7, 12} Dermal allografts (like GraftJacket, Wright Medical) improve healing outcomes like pain and mobility, often for years after repair, but sometimes fail mechanically and re-tear.^{39, 43, 44} Degradable synthetic scaffolds offer an alternative solution for achieving mechanical support but lack the biomimetic cues of ECM.^{39, 45, 46}

Incorporating ECM proteins into synthetic scaffolds can add biological signaling cues while offering greater structural control. Coating poly(L-lactic acid) fibers with collagen and chondroitin sulfate led to increased gene expression of scleraxis, collagen type I, and tenomodulin in human mesenchymal stem cells.⁴⁷ Fibrin improved cell migration, matrix synthesis, and tendon gene expression compared to collagen in static culture, but tensile loading decreased tendon-like expression on fibrin while it increased on collagen.⁴⁸ Fibronectin, a component of the tendon ECM (~0.2% w/w^{49, 50}), is often used because cells can easily attach and migrate along fibronectin. While fibronectin incorporation does improve tendon gene expression and matrix synthesis over synthetic scaffolds alone,^{13, 51, 52} incorporating collagen leads to better cell infiltration on electrospun fibers,⁵¹ greater COL1A1, COL3A1, and DCN gene expression,⁵² and more collagen protein expression¹³ than fibronectin. Incorporating TDM, or ligament derived matrix, into electrospun scaffolds enhanced cell proliferation, tendon gene expression, and collagen protein synthesis^{13, 22, 24}. Using tendon ECM from ‘super-healer’ MRL/MpJ mice leads to better healing in vivo than tendon matrix from C57BL/6 mice, demonstrating that TDM proteins or molecules can greatly affect cellular responses.^{53, 54} The increased biological signaling cues from the combination of collagen I, along with other tendon ECM proteins such as collagen III, collagen VI, fibronectin, decorin, and biglycan could provide this benefit, but more work is needed to fully understand it.

EDC:NHS crosslinking disrupts the glutamate amino acids (E) in the Glycine (G) - variable (x) - Pyrrolysine (O) - Glycine (G) - Glutamate (E) - variable (x’) (GxOGEx’) integrin-binding sequence present in collagen; the more EDC:NHS used, the more integrin-binding sequences are disrupted.³⁴ Using the 5:2:1 EDC:NHS:COOH ratio reduced the availability of this sequence by ~80%, while crosslinking collagen scaffolds with UV at 0.96 J/cm² reduced free amine availability by <10%.³⁴ By these numbers, we estimate that our 2.5:1:1 EDC:NHS:COOH crosslinking ratio would reduce the GxOGEx’ sequence – as well as alternate motifs GLSGER, GQRGER, and GASGER⁵⁵ by only around 40%. Previously, a 5:2:1 EDC:NHS:COOH ratio reduced attachment of bone marrow-derived mesenchymal stem cells,²⁷ which did not occur here. It is unknown whether the reduced attachment was driven by the higher EDC:NHS:COOH ratio compared to the ratio used in this study, or how much the different scaffolds and cell type affected this attachment.

In contrast, UV crosslinking removes the Glycine (G) - Phenylalanine (F) - Pyrrolysine (O) - Glycine (G) - Glutamate (E) - Arginine (R) (GFOGER) binding pattern by blocking the phenylalanine (F), which reduces α1 integrin binding and cell attachment, but UV crosslinking does not inhibit other GxOGEx’ binding patterns that both α1 and α2 integrins can bind to, such as Glycine (G)-Leucine (L)- Pyrrolysine (O)-Glycine (G)-Glutamate (E)-Arginine (R) (GLOGER).³¹ Therefore our finding that hASCs exhibited greater spreading on the UV-crosslinked scaffolds is concordant with greater loss of integrin-binding sequences present in collagen as a result of carbodiimide crosslinking: tendon cells respond to mechanical and biochemical signals via integrin binding.⁵⁶ While cell spreading was affected least with the 2.5:1:1 EDC:NHS:COOH ratio, further decrease in EDC crosslinking ratios could potentially both maintain crosslinking stability and allow integrin binding – a 0.7:0.28:1 ratio retained 80% of free amines.^{34, 57} However, further decreases could negatively impact other properties as mechanical properties, attachment of the TDM around the underlying polymer) and slow degradation rate. Despite this potential reduction of integrin signaling on EDC-crosslinked scaffolds,^{31, 34} we did not identify differences in collagen or s-GAG content between the UV-crosslinked or EDC-crosslinked scaffolds.

Crosslinking the scaffolds promoted greater matrix production in this study despite the reduced integrin binding, unlike the uncrosslinked TDM of untreated scaffolds. Similar results were seen with mesenchymal stem cells cultured on cartilage derived matrix, where crosslinked scaffolds promoted sGAG and collagen synthesis compared to uncrosslinked scaffolds.²⁷ We did not however observe any difference in collagen retention on films or scaffolds or collagen loss into washes or media between EDC- or UV-crosslinked scaffolds based on hydroxyproline assay. UV crosslinking increased scaffold stiffness in this study, while low levels of carbodiimide crosslinking (2.5:1:1 ratio) did not. In other work, EDC crosslinking at a higher ratio of 5:2:1 increased ECM stiffness,⁵⁷ which does increase collagen type I and type III expression.⁵⁸

The 1655/1690 ratio in FTIR spectra has had several interpretations in the literature where higher ratios have been attributed to characteristics such as increased alignment, insolubility, crosslinking, maturation, or ratio between trivalent and divalent crosslinks.^{38, 59–61} Most of these studies agree that higher ratios are consistent with increased crosslink and collagen retention. We found no difference between the 1655/1690 ratios in our UV or carbodiimide crosslinked samples compared to our uncrosslinked TDM samples, suggesting that crosslinking may not be the dominant factor behind the 1655/1690 ratio in our scaffolds. Instead, in our study, the consistent 1655/1690 ratio between crosslinked and uncrosslinked scaffolds could be due to the maturity or the alignment of the collagen in the underlying TDM preparation.

EDC-crosslinked, UV-crosslinked and TDM coated scaffolds expressed both Amide I and Amide II peaks on FTIR. Amide I represents mostly C═O stretching.⁶² For collagen, the C═O stretching for Amide I is evident around 1630 cm⁻¹ in our samples, which is in the range found for collagen across various species.⁵⁹ As expected, the PCL scaffolds did not exhibit a collagen Amide I peak. All scaffolds had a peak at 1721–1722 cm⁻¹, which was expected for PCL since the spectra obtained is similar to previous studies, and is attributable to C═O ester bonds in the ε-caprolactone monomer.⁶³

5. Conclusion

In conclusion, carbodiimide crosslinking methods improved collagen content at Day 28 compared to PCL scaffolds, unlike coating with or UV crosslinking TDM to the PCL scaffolds. In contrast, neither coating with nor crosslinking TDM to the scaffolds stimulated matrix synthesis of glycosaminoglycans or promoted cell proliferation. Crosslinking with UV increased scaffold modulus and stiffness, while carbodiimide crosslinking trended toward improving scaffold modulus. Nonetheless, both crosslinking techniques appeared to provide improvements in some features critical for tendon tissue engineering and will be evaluated further in future studies. Future studies will optimize crosslinking for ideal development of tendon-specific extracellular properties while maintaining regenerative potential and limiting host inflammatory response.