Immunomodulatory effects of amniotic membrane matrix incorporated into collagen scaffolds

Abstract

Adult tendon wound repair is characterized by the formation of disorganized collagen matrix which leads to decreases in mechanical properties and scar formation. Studies have linked this scar formation to the inflammatory phase of wound healing. Instructive biomaterials designed for tendon regeneration are often designed to provide both structural and cellular support. In order to facilitate regeneration, success may be found by tempering the body’s inflammatory response. This work combines collagen-glycosaminoglycan scaffolds, previously developed for tissue regeneration, with matrix materials (hyaluronic acid and amniotic membrane) that have been shown to promote healing and decreased scar formation in skin studies. The results presented show that scaffolds containing amniotic membrane matrix have significantly increased mechanical properties and that tendon cells within these scaffolds have increased metabolic activity even when the media is supplemented with the pro-inflammatory cytokine interleukin-1 beta. Collagen scaffolds containing hyaluronic acid or amniotic membrane also temper the expression of genes associated with the inflammatory response in normal tendon healing (TNF-α, COLI, MMP-3). These results suggest that alterations to scaffold composition, to include matrix known to decrease scar formation in vivo, can modify the inflammatory response in tenocytes.

INTRODUCTION

Tendon healing following injury, even when surgically repaired, results in fibrocartilagenous scar formation which in turn leads to decreased ultimate tendon strength and disorganized collagen fibers. Adult wound repair occurs in the three overlapping phases: inflammation, proliferation, and remodeling.^1,2 The inflammatory phase in adult wound healing is characterized by the recruitment of cells to the site of injury.^1,3 Platelets release platelet-derived growth factor (PDGF) and transforming growth factor-beta (TGF-β) in order to initiate the chemotaxis of fibroblasts and inflammatory cells (neutrophils, macrophages) to the wound.³ Additionally, TGF-β stimulates macrophages to secrete the key inflammatory cytokines, tumor necrosis factor alpha (TNFα) and interleukin-1 (IL-1) during this phase.³ Scar formation is the ultimate end point of adult wound repair and studies have shown that the inflammatory response is an essential prerequisite for scarring.⁴ The limited healing of tendons calls for new techniques and materials that will promote tendon regeneration in vivo.

In contrast to adult wounds, healing in fetal cutaneous wounds proceeds in a regenerative fashion, with neither the typical inflammatory response nor the formation of scar tissue.^4–6 Extensive studies have shown that differences between adult and fetal wound healing are found at many levels of the wound healing cascade; cells, cytokines, growth factors, and the ECM.⁵ Fetal wounds exhibit lower levels of PDGF, TGF-β1 and TGF-β2⁷; leading to reduced numbers of immune cells in fetal skin and lower inflammatory reaction.^5,8,9 Pro-inflammatory cytokines (e.g., TGF-β1) are also less prevalent in fetal wounds.^10,11 Beredjiklian et al. examined differences between adult and fetal tendon wound healing via partial mid-substance tenotomies of the lateral extensor tendon in pregnant ewes and their fetal lambs.¹² Maternal and fetal tendons were harvested for histological and mechanical testing 7 days post-surgery. Testing showed that fetal tendons exhibited highly aligned and organized collagen fibers, no macroscopic abnormalities or adhesions, and no increases in TGF-β1 expression; all in direct contrast to the corresponding adult tendons.

We hypothesize the composition of biomaterials that mimic the extracellular matrix found in low inflammatory environments (such as the fetal wound environment) may be particularly relevant for modifying adult immune responses. The nonsulfated glycosaminoglycan hyaluronic acid (HA) is found in both adult and fetal wounds¹³ and assists in rapid cell proliferation and motility. HA is a significant component of the fetal, scarless wound healing cascade,¹⁴ and has been extensively studied in the context of burns and chronic wounds¹⁵ as well as reducing adhesion.¹⁶ Separately, the amniotic membrane (AM), known for its anti-inflammatory and antimicrobial properties as well as scarless wound healing capacity, is the innermost layer of the placenta. Clinicians and scientists have shown that the intact AM sheet is valuable in the treatment of corneal surfaces,^17–19 skin wounds,^20,21 oral cavity reconstruction²² and many other reconstruction applications.^23,24 In addition, the use of dry, micronized human amnion membrane has been shown to reduce cartilage degeneration in an osteoarthritic rat model.²⁵ While the AM has been implemented successfully in these wound healing applications and studied as an anti-inflammatory therapeutic for degenerative disease, its potential as a bioactive component in 3D biomaterials has not been extensively investigated.

Collagen-glycosaminoglycan (CG) scaffolds have been used in a wide variety of applications for skin, peripheral nerve, and cartilage tissue engineering as well as 3D environments for in vitro studies of cell behavior.^26–32 It has been shown that in severe skin wounds the application of these scaffolds decreases the population of myofibroblasts at the wound site.³³ By delaying these contraction processes, collagen-based scaffolds have induced regeneration and decreased scar formation in skin and nerve defects.³⁴ Our lab has recently developed scaffold variants to mimic the anisotropic microstructure of native tendons.³⁵ While typically fabricated with chondroitin sulfate, modifications to the scaffolds’ glycosaminoglycan (GAG) content were subsequently used to modulate growth factor sequestration, much like the native tendon ECM.³⁶ The capacity to alter the biomolecular environment and cellular response via scaffold composition may be a way to address needs for immunomodulatory biomaterials for tissue regeneration. The work described here combines CG scaffolds with fetal wound healing-inspired matrix (HA, AM) in order to study their ability to temper pro-inflammatory conditions in vitro.

MATERIALS AND METHODS

Analysis of amniotic membrane

Amnion membrane isolation

In collaboration with Carle Foundation Hospital Tissue Repository (Urbana, IL), human placentas were obtained following uncomplicated vaginal births. The amniotic membrane (AM) matrix components have been isolated from these placentas as described.³⁷ Briefly, excess blood was washed from the fetal membrane and the AM was mechanically separated from the placenta.³⁷ The AM was cut into sections, washed, and then decellularized via incubation in thermolysin (125 μg/mL).³⁸ Decellularized AM were then rinsed with shaking in PBS to remove cellular debris and stored in PBS at 4°C. Following 24–48 h of storage, in which the spongy layer of the AM was allowed to swell, decellularized AM pieces were scraped (spatula) in order to fully separate the AM matrix of the compact layer from the underlying spongy layer. The remaining matrix (basement membrane, compact layer) was lyophilized and stored in a dessicator until further use.

Amnion membrane characterization

Biopsies of placental tissue were fixed in 10% formalin and embedded in paraffin in a cross sectional orientation. Serial slices of 5 μm thickness were cut and mounted on slides. Slides were deparaffinized and stained with either H&E (H&E) or Masson’s Trichrome stain. All slides were mounted with Permount and a coverslip prior to imaging on an optical microscope (Leica Microsystems, Germany). The isolated decellularized and dried amniotic membrane was analyzed for its collagen and GAG content via a hydroxyproline³⁹ and a 1,9-dimethylmethylene blue (DMMB)⁴⁰ assay, respectively.

Fabrication of CG scaffolds

Preparation of collagen suspension

A suspension of collagen and one of three components was made by homogenizing type I collagen from microfibrillar collagen (Collagen Matrix, Oakland, NJ) in 0.5M acetic acid with either (1) chondroitin sulfate from shark cartilage (Sigma Aldrich, St Louis, MO), (2) hyaluronic acid from Streptococcus equi (Sigma-Aldrich, St. Louis, MO), or (3) ground (mortar and pestle) and, further, homogenized amniotic membrane as collected above.²⁷ Scaffolds containing CS and HA components were made with a 11:1 collagen:component ratio while C:AM were made at a 5:1 w:w ratio due to the high AM collagen content. All scaffolds were made with a target total density of 0.5% w/v.⁴¹ The suspensions were stored at 4°C and degassed prior to use.⁴¹

Fabrication of collagen-based scaffolds via freeze drying

Isotropic CG scaffolds were fabricated as previously described.^35,42 The collagen suspension was transferred to an aluminum tray mold and placed into the freeze-dryer with a shelf temperature of 4°C. The suspension was frozen to a final freezing temperature of −40°C. To form a porous, sponge-like scaffold, ice crystals were sublimated under vacuum (200 mTorr) at 0°C.

Crosslinking of CG scaffold

In order to sterilize and dehydrothermally crosslink the scaffolds, the lyophilized sheets were placed in a vacuum oven (Welch, Niles, IL) at 105°C under vacuum for 24 h.²⁷ A biopsy punch was used to cut 6 mm diameter cylinders from the 4mm thick scaffold sheet for use in all experiments. Scaffolds were hydrated by soaking in 100% ethanol overnight and washing in PBS for 24 h. Scaffolds were then crosslinked via carbodiimide chemistry by immersing in 1-ethyl-3-[3-dimethylaminopropyl] carbodiimide hydrochloride (EDC) and N-hydroxysulfosuccinimide (NHS) at a molar ratio of 5:2:1 EDC:NHS:COOH for 1.5 h under shaking at room temperature.^43,44 Prior to use, scaffolds were washed with and stored in PBS at 4°C.

SEM analysis

Scanning electron microscopy (SEM) was used to visualize the cross sectional interior microstructure of dry, uncrosslinked scaffolds. Samples were imaged with a JEOL JSM-6060LV scanning electron microscope using secondary electron and backscattered electron detectors under low vacuum.

Mechanical testing

Compression testing was carried out on crosslinked, hydrated samples. Scaffolds from each group (CS, HA, AM) were cut from the scaffold sheet (~4 mm thick) with 12 mm biopsy punches before being hydrated and cross-linked as described above. Using a TA.XTplus Texture Analyzer (StableMicro Systems Ltd., Surrey, UK) equipped with a 1 kg load cell, the stress-strain behavior of each scaffold variant was measured. Samples were loaded in unconfined compression to approximately 75% strain at a rate of 0.01 mm/min. The elastic moduli were determined from the linear elastic region (~1–10% strain) of each stress–strain plot.⁴⁴

Pull down sequestration assay

The degree of cytokine sequestration in each CG scaffold variant was determined via a pull down assay.³⁶ Three hydrated, crosslinked scaffolds were placed in a single well of an ultra-low attachment 24-well plate (Fisher, Waltham, MA) with 1 mL of a pH 7.4 PBS solution supplemented with 1 ng/mL interleukin-1 beta (IL-1β) (ProSpec, Israel) and 1% bovine serum albumin (BSA). Control wells containing the IL-1β solution and no scaffolds were used. The scaffolds and controls were incubated under gentle shaking at 37°C for 1 h. Immediately following incubation, an ELISA kit (R&D Systems, Minneapolis, MN) was used to measure the amount of IL-1β remaining in solution. The difference between the concentration of IL-1β remaining in the solutions containing scaffolds and the solutions that did not have scaffolds (controls) was regarded to be the amount of cytokine in the scaffolds. IL-1β pull down for each CG variant was reported as a percentage of the total IL-1β concentration in the loading solution.

Cell culture

Tenocyte isolation and culture

Tenocytes (tendon cells) were isolated from 2- to 3-year-old horses and expanded in growth media: high glucose Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA), 1% Antibiotic-Antimycotic (Invitrogen, Carlsbad, CA), 1% l-glutamine (Invitrogen, Carlsbad, CA), and 50 μg/mL ascorbic acid (Wako, Richmond, VA) in standard culture flasks.⁴⁵ The cells were cultured at 37°C and 5% CO₂ and the media was changed every 3 days until the tenocytes reached confluence. Passage 5 cells were used for all experiments.

Scaffold seeding and culture conditions

Hydrated, cross-linked scaffolds were soaked in growth media for 1 h at 37°C before being placed in ultra-low attachment 6-well plates (Fisher, Waltham, MA). Confluent tenocytes were trypsinized and resuspended in growth media at a concentration of 2.5 × 10⁵ tenocytes per 20 μL. Scaffold variants were seeded with tenocytes using a previously established static seeding method.³⁵ The scaffolds were incubated for 15 min at 37°C following the addition of 10 μL of the cell suspension to one side of the scaffold. The scaffolds were then flipped over and another 10 μL of cell suspension was added for a total of 2.5 × 10⁵ tenocytes seeded on each scaffold. To allow for initial cell attachment, scaffolds were placed in the incubator for 2 h. After this period, additional media was added and scaffolds were incubated at 37°C and 5% CO₂ for 24 h before the media was changed for one of three inflammatory media variants. These variants included: (1) growth media (control), (2) media supplemented with 0.1 ng/mL of the pro-inflammatory factor interleukin-1 beta (IL-1β) (inflammatory), and (3) media supplemented with 1 ng/mL of IL-1β (high inflammatory). Scaffolds were cultured in culture media as described above, with or without the addition of serum, for the duration of the experiment. Media changes occurred every 3 days.

Histological analysis of cell-seeded scaffolds

Following 7 days of culture, tenocyte-seeded scaffolds were fixed in 10% formalin overnight and then washed and stored in PBS. Fixed scaffolds were blocked with PBS containing 2% bovine serum albumin and 0.1% Tween 20 for 1 h under shaking at room temperature. Then scaffolds were incubated overnight in vimentin antibody (Cell Signaling Technology, 1:800 dilution), washed and stored in 0.1% Tween 20 PBS. Prior to embedding at −80°C, scaffolds were soaked in 20% sucrose for 1 h followed by Optimal cutting temperature compound (OCT) for 1 h. Sections of 25 micron thickness were obtained using a Leica CM3,050 S cryostat and imaged via fluorescence microscopy (Leica DMI4,000B fluorescence microscope, Qimaging camera). Fluorescent and brightfield channels were merged using ImageJ.

Quantification of TGF-β1 protein release

Tenocyte-seeded scaffolds were cultured in serum-free media variants for 24 h. Following this culture period, the amount of TGF-β1released into the media was measured via an ELISA kit (R&D Systems, Minneapolis, MN). Concentrations were determined using a standard curve and a four parameter logistic (4-PL) curve-fit. Results were normalized to acellular scaffold variants cultured in control media.

Quantification of cell metabolic activity

Metabolic activity of the tenocytes within the collagen scaffolds was measured using a nondestructive alamarBlue^® assay (Invitrogen, Carlsbad, CA).^35,41 Briefly, scaffolds were removed from culture, rinsed in PBS, and incubated under gentle shaking at 37°C for 2 h in a 1× alamarBlue^® solution containing the media variant of the primary culture. Using a fluorescent spectrophotometer, resorufin fluorescence was measured (excitation: 540 nm, emission: 590 nm) and compared with a standard curve created from a known number of cells from the start of the experiment. The standard curve was run on well-plated cells, with known numbers of cells in each well, ranging from 25% to 300% of the number of cells seeded per scaffold. Metabolic activity was tracked over a 7 day period and, at each time point, was interpolated to the standard curve and reported as a percentage of the total number of seeded cells.

Gene expression analysis through RNA isolation and real-time PCR

An RNeasy Plant Mini kit (Qiagen, Valencia, CA) was used to isolate RNA from cell-seeded scaffolds. The scaffolds were rinsed in PBS, cut in half with a razor, and immersed in the kit’s lysis buffer for 5 min on ice.⁴⁶ Following the kit’s instructions, RNA was isolated and total RNA was quantified via spectrophotometry. The QuantiTect Reverse Transcription kit (Qiagen, Valencia, CA) was used to reverse transcribe the isolated RNA in a Bio-Rad S1000 thermal cycler. Real-time PCR reactions were performed in an Applied Biosystems 7900HT Fast Real-Time PCR System (Carlsbad, CA) to measure gene expression levels for collagen I (COL1A2), matrix metalloproteinase-3 (MMP-3), scleraxis (SCXB), and tumor necrosis factor-alpha (TNF-α). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a housekeeping gene (differences in GAPDH expression between groups were compared and found to be statistically insignificant: p > 0.05). Sequence Detection Systems software v2.4 (Applied Biosystems, Carlsbad, CA) was used to complete analysis. All results were expressed as fold changes relative to expression levels of cells in scaffolds of the same type and at the same time point but cultured in the control media.

Statistical analysis

One-way analysis of variance (ANOVA) followed by Tukey-HSD post-hoc test was performed on all data sets except for metabolic activity. The metabolic activity was analyzed using a two-way, repeated measures ANOVA followed by Tukey-HSD post-hoc test. A p values < 0.05 was used for significance. All analyses were based on a minimum of n = 4 scaffolds. Error is reported as the standard error of the mean in the figures.

RESULTS

Placental structure and amnion composition

Histological staining of placental biopsies shows the structure of the amniotic membrane in relation to the rest of the placenta [Fig. 1(A,B)]. H&E staining [Fig. 1(A)] shows a single layer of epithelial cells on the surface of the amniotic membrane. Masson’s Trichrome staining [Fig. 1(B)] indicates that the amniotic membrane is rich in collagen. Quantitative analysis determined AM collagen content (40.83 ±0.04 wt %) and sulfated GAG content (0.22 ±0.11 wt %).

FIGURE 1.

Amniotic membrane histological analysis, scaffold mechanical and microstructural analysis. A: Placental section stained with H&E. Arrows indicate amniotic membrane. Scale bar: 200 μm. B: Masson’s Trichrome staining of placental section. Arrows indicate amniotic membrane. Scale bar: 200 μm. C: Elastic modulus of scaffold variants under compression. (*) significance (p <0.05) between scaffold groups. D: SEM images of scaffold variants; collagen:chondroitin sulfate (C:CS), collagen:hyaluronic acid (C:HA), collagen: amniotic membrane (C:AM) (left to right). Scale bar: 100 μm. E: Histological staining of vimentin in cell-seeded scaffolds with varying composition at day 7. Left to right: collagen:-chondroitin sulfate (C:CS), collagen:hyaluronic acid (C:HA), collagen: amniotic membrane (C:AM). Scale bar: 200 μm.

Scaffold microstructure, mechanics, and histological analysis

The elastic modulus of each of the CG scaffold variants varied between 0.1 and 1 kPa showing a significant effect of the included matrix [Fig. 1(C)]. Scaffolds containing chondroitin sulfate (CS) had an average elastic modulus of 0.169 ±0.010 kPa. Substituting hyaluronic acid (HA) for CS resulted in a significant increase in modulus to 0.511 ±0.052 kPa while adding amniotic membrane (AM) showed a further significant increase (1.065 ±0.083 kPa). The scaffold variants all showed an open porous network via SEM with pores on the order of 100 μm [Fig. 1(D)] and by day 7 of culture the tenocytes were well distributed throughout the scaffolds with no obvious differences in cell aggregation or distribution [Fig. 1(E)].

Tendon cell metabolic activity under pro-inflammatory media conditions

The metabolic activity of equine tenocytes cultured within the CG scaffold variants was evaluated prior to the pro-inflammatory challenge (day 0) and subsequently at days 1, 4, and 7 (Fig. 2). Each scaffold variant supported metabolic health over time. However, scaffolds containing HA or AM showed a greater increase in metabolic activity. Cells within C:AM scaffolds cultured in control media showed increased metabolic activity compared with C:CS (standard, control) scaffolds at all time points [Fig. 2(A)]. The same was seen in cells cultured in C:AM scaffolds with media supplemented with the pro-inflammatory cytokine IL-β at inflammatory [Fig. 2(B)] and high inflammatory levels [Fig. 2(C)]. Additionally, scaffolds containing hyaluronic acid also promoted significantly increased tenocyte metabolic activity over the C:CS scaffolds, especially in high inflammatory media, as seen at all time points [Fig. 2(C)].

FIGURE 2.

Tenocyte metabolic activity in scaffolds cultured in (A) control media, (B) inflammatory media containing 0.1 ng/mL IL-1β, and (C) high inflammatory media supplemented with 1 ng/mL IL-1β. (*) significance between groups indicated (p<0.05). The same data has been grouped by scaffold type and plotted in (D) collagen:chondroitin sulfate, (E) collagen:hyaluronic acid, and (F) collagen:amniotic membrane. (**) significance compared with all groups at given time point (p<0.05). (N.S.) no significant differences between groups at a given time point (p >0.05).

Reorganizing the data to display cell metabolic activity across all media conditions for each scaffold type [Fig. 2(D–F)], we find that, following the introduction of variant medias, there was no significant difference in cell metabolic activity between C:CS scaffolds across all media conditions [Fig. 2(D)]. Interestingly, while cell metabolic activity trended to increase throughout the 7 days experiment, cells in amniotic membrane matrix scaffolds showed significantly enhanced (p <0.05, day 4) cell metabolic activity in response to inflammatory or high inflammatory media formulations [Fig. 2(F)].

Gene expression of tenocytes cultured in CG scaffolds under IL-1β challenge

PCR was used to examine the relative expression of inflammatory and tenocyte phenotypic genes over the course of 7 days of culture in pro-inflammatory media conditions. The reported fold changes are compared with cells cultured in the same scaffold type but in control media. Tenocyte expression of tumor necrosis factor-alpha (TNF-α), a pro-inflammatory cytokine, is significantly downregulated (p <0.05) in AM containing scaffolds compared with C:HA scaffolds cultured in inflammatory media at day 7 [Fig. 3(A)]. Further, although not statistically significant, AM scaffolds also showed TNF-α downregulation compared with CS scaffolds (p =0.0814). When the cell-seeded scaffolds were cultured in high inflammatory media, both HA and AM scaffolds induced lower TNF-α expression compared with the standard C:CS scaffolds [Fig. 3(B)].

FIGURE 3.

Gene expression of tumor necrosis factor-alpha (TNF-α) in tenocytes cultured in three scaffold variants in (A) inflammatory (0.1 ng/mL IL-1β) and (B) high inflammatory (1 ng/mL IL-1β) media conditions. Gene expression was normalized to tenocytes cultured in respective scaffold in control (no inflammatory factors) media. (*) significance between groups indicated (p <0.05). (**) significance compared with all groups at given time point (p <0.05).

Further gene expression analysis focused on markers of tendon phenotype (Fig. 4). Collagen I (COLI) is the primary ECM protein in tendon and is secreted by tenocytes. COLI gene expression was significantly downregulated in CS scaffolds following 1 day of inflammatory challenge by IL-1β when compared with C:HA scaffolds [Fig. 4(A)]. Following 4 days of culture in inflammatory media, tenocytes in C:CS scaffolds were expressing COLI at significantly lower levels than those in parallel HA or AM scaffolds and in the C:CS scaffolds in control media. Across all time points cells in both AM and HA scaffolds were maintaining COLI expression comparable to their respective controls. Scaffolds cultured in high inflammatory media showed no significant differences in COLI gene expression at day 1 [Fig. 4(B)]. Analysis of subsequent days shows significant downregulation of tenocyte COLI expression in all scaffolds (except HA at day 4) as compared with corresponding scaffolds in control media.

FIGURE 4.

Tenocyte gene expression levels of collagen I (COL1A2; A, B), matrix metalloproteinase-3 (MMP-3; C, D), and scleraxis (SCXB; E, F) as a function of scaffold content at days 1, 4, and 7 of culture in inflammatory (0.1 ng/mL IL-1β) (A, C, E) and high inflammatory (1 ng/mL IL-1β) (B, D, F) media. Gene expression was normalized to tenocytes cultured in respective scaffold in control (no inflammatory factors) media. (*) significance between groups indicated (p <0.05). (∧,∨) significant increase, decrease (respectively) compared with corresponding control (same scaffold type, same time point) (p <0.05).

Matrix metalloproteinase-3 (MMP-3) is involved in matrix breakdown in normal tendon remodeling and in early wound healing. Under the in vitro inflammatory conditions presented here, there was no significant difference in the MMP-3 gene expression between scaffolds or when compared with control media [Fig. 4(C)]. When considering the cells cultured in scaffolds exposed to high inflammatory media, MMP-3 expression is expectantly high in all groups at day 1 [Fig. 4(D)]. By day 4, tenocytes in AM scaffolds had significantly lower MMP-3 expression, no different from their respective control, as compared with cells cultured in CS or HA scaffolds.

The expression level of the pro-tenogenic transcription factor scleraxis (SCXB) was subsequently analyzed. While the scaffold variants used here did not contain aligned microstructural elements known to promote tenogenic expression profiles, results show little difference between groups due to scaffold type or media condition [Fig. 4(E,F)]. Cells in HA scaffolds show downregulation of SCXB expression compared with CS scaffolds (day1) and control media (day 7) in inflammatory media conditions [Fig. 4(E)]. SCXB expression is also downregulated in high inflammatory media compared with control media in HA scaffolds (day 1) and all scaffolds at day 7 [Fig. 4(F)].

Short-term protein release from pro-inflammatory tenocyte culture

In the absence of a pro-inflammatory signal (control media), cells seeded in CS scaffolds release significantly more TGF-β1 than those seeded in HA or AM scaffolds [Fig. 5(A)]. When the cell-seeded scaffolds are exposed to IL-1β in the inflammatory media, there was a significantly higher concentration of TGF-β1 in the media surrounding the collagen scaffolds containing CS as compared with the HA scaffolds [Fig. 5(B)]. In the high inflammatory media, there was no statistically significant difference between the amount of TGF-β1 released from each of the scaffold groups [Fig. 5(C)].

FIGURE 5.

Transforming growth factor-beta 1 (TGF-β1) release from tenocyte-seeded scaffolds of varying compositions. Measurements were made after 24 h of culture in serum-free medias: (A) inflammatory (0.1 ng/mL IL-1β) and (B) high inflammatory (1 ng/mL IL-1β). (*) significance between groups indicated (p <0.05).

Cytokine sequestration within scaffold variants

A pull down assay was employed to determine whether the three scaffold types were differentially sequestering or trapping the pro-inflammatory cytokine (IL-1β) and, thus, influencing the cellular response. While the results suggest increased pull down when moving from CS to HA to AM scaffolds, there is no statistically significant difference between the entrapment of IL-1β in the three scaffold variants (Fig. 6).

FIGURE 6.

Pull down of IL-1β by scaffolds with varying compositions. The degree of pull down was normalized to the concentration of IL-1β in a solution without scaffolds.

DISCUSSION

The inflammatory phase of adult wound healing is important for inducing the rapid closure of wounds but, as a result, leads to disorganized matrix and scar formation.⁴ Studies in the healing process following acute tendon injury, in particular, show upregulation in TGF-β1 mRNA and protein expression, disruption in collagen fiber continuity and organization, and decreases in ultimate stress and modulus when compared with uninjured tendons.^12,47 Increases in the gene expression of pro-inflammation (IL-1β, TNFα) and matrix degradation factors (MMP-3, MMP-13) as well as decreases in the expression of tendon ECM (COLI) and tendon-specific markers (SCX) have also been shown in a canine flexor tendon model.⁴⁸ In designing a biomaterial to drive tendon regeneration in vivo, the presence of an inflammatory environment should be considered.

It has been previously shown that fetal wounds have the capacity to facilitate scarless wound healing with a limited inflammatory response. Interestingly, this phenomenon is not solely defined by the sterile, intrauterine environment. Fetal marsupials such as the gray short-tailed opossum, which develop outside of the uterus, heal cutaneous wounds in a scarless fashion.⁴⁹ Research has also shown that scar formation occurs in adult sheep skin transplanted onto fetal lambs, but not in adjacent fetal skin wounds.⁵⁰ Due to the unique nature of these results, efforts have begun to focus on the ECM and the role it plays in wound healing.⁶ Hyaluronic acid (HA), shown to have a dominant presence in fetal wounds both in the wound ECM⁵¹ and surrounding fluid,¹³ has been used in three-dimensional skin grafts to reduce scar formation.⁵² Amniotic membrane has also been used in the treatment of skin wounds.^20,21 As a result, our effort here investigated whether CG scaffolds containing hyaluronic acid or amniotic membrane matrix could reduce the immune response associated with interleukin-1beta (IL-1β) in tenocytes.

Collagen-based scaffold content was altered by incorporating HA or AM in the collagen-glycosaminoglycan suspension instead of the standard chondroitin sulfate (CS). Prior to incorporation into the scaffold, the AM matrix was partially characterized for collagen and sulfated GAG content. The complete makeup of the amniotic membrane is not known; however, other studies have reported that in addition to these two matrix components, the AM also contains HA,²⁴ elastin and denatured collagen.⁵³ Following scaffold lyophilization, SEM analysis revealed that all scaffold variants had an open porous microstructure [Fig. 1(D)]. Mechanical testing on hydrated, crosslinked samples revealed significant increases in modulus when moving from CS to HA to AM scaffolds. Each scaffold contained the same overall material density and exhibited similar pore structure indicating that the matrix content affects scaffold modulus. While statistically significant, the relative differences in the moduli are still small compared with native tendon. However, these differences may alter local cellular behavior and warrant further investigation in the future. While the scaffolds tested here have insufficient mechanical strength for direct tendon application, separate efforts in our group have shown reinforcement strategies sufficient to boost the tensile modulus of the scaffold to greater than 1 MPa.^41,54 When examining the distribution of cells following the 7 days of culture, the representative images in Figure 1(E) do not show any evidence of cellular clumping in any of the collagen-based scaffolds, including C:AM.

Interleukin-1β is a pro-inflammatory cytokine secreted by macrophages during the inflammatory phase of wound healing and was used in this study as the pro-inflammatory challenge. We first examined metabolic activity of tenocytes seeded within these scaffolds over a 7-day period (Fig. 2). Not only do the amnion scaffolds support tenocyte metabolic health in control media, cells are significantly more active in AM scaffolds than CS scaffolds [Fig. 2(A)]. Further, tenocytes in AM scaffolds show increased metabolic activity in response to the pro-inflammatory challenge of IL-1β [Fig. 2(B,C)]. Since metabolic health is dependent on the number of cells present as well as the metabolic activity on a per cell basis, the increase of these in the AM scaffolds is a critical finding.

Genes associated with the early inflammatory response in tendon healing (TNF-α, COL1A2, MMP-3) were differentially regulated in tenocytes in each of the three different scaffold environments. Each of these has been studied within the context of the normal tendon healing cascade.⁴⁸ The pro-inflammatory factor TNF-α, which works alongside IL-1β, is normally upregulated in adult tendon healing.⁴⁸ The results seen here show an upregulation of TNF-α expression in CS scaffolds while there is downregulation of this pro-inflammatory marker in HA and AM scaffolds after 7 days of culture [Fig. 3(A)]. Especially in the high inflammatory media conditions [Fig. 3(B)], cells in HA and AM scaffolds show significant downregulation of TNF-α suggesting that these fetal ECM components may maintain their native anti-inflammatory properties within the CG scaffold platform.

When considering genes associated with matrix deposition and remodeling, altering the CG scaffold composition also resulted in differences in expression (Fig. 4). Collagen I (COL1A2) is the primary ECM protein in tendon and is secreted by tenocytes but is downregulated in the first days following tendon injury compared with healthy tendons.⁴⁸ In this experiment, COL1A2 gene expression of scaffolds cultured in the inflammatory media is only significantly downregulated in the CS scaffolds when compared with the control media (day 4) [Fig. 4(A)]. HA and AM scaffolds show no significant differences in expression when compared with their respective controls. Our results show higher COL1A2 expression in the cells cultured in the inflammatory media in HA scaffolds compared with CS (day 1) and both scaffold types (day 4) [Fig. 4(A)]. Cell-constructs cultured in the high inflammatory media exhibited COL1A2 downregulation compared with the control in all groups (at days 4 and 7) except for HA at day 4 [Fig. 4(B)]. Typically upregulated in adult tendon injury,⁴⁸ MMP-3 expression was no different from the control media in the groups exposed to inflammatory media [Fig. 4(C)]. However, in high inflammatory media conditions, cells in AM scaffolds (day 4) showed significantly lower expression of MMP-3 that was no different from the control media [Fig. 4(D)]. These results together suggest that fetal wound inspired matrices alter the pro-inflammatory effects usually seen in tendon wound sites. Implementing scaffolds such as these in tendon regeneration applications may reduce inflammation-induced scar formation.

Tendon is a highly anisotropic tissue with tenocytes longitudinally aligned within its microstructure. Geometric cues such as this have been shown to play a role in the maintenance of the tenocyte phenotype in vitro.^35,55 However, due to ease of bulk fabrication, this study used isotropic CG scaffolds, eliminating any geometric cues that would maintain the tenocyte phenotype. Gene expression of the tendon-specific gene scler-axis (SCXB) was monitored over the 7 day culture period [(Fig. 4(E,F)]. Results give no indication that the addition of HA or AM into the scaffold either promotes or hinders expression of SCXB. Future work will employ anisotropic scaffolds that allow for tenocyte alignment and phenotypic preservation.⁵⁵

Important questions remain: why do scaffolds containing HA or AM alter the cell response to inflammatory signals? And what is the means of mechanism? Transforming growth factor-beta 1 (TGF-β1) is a protein that is highly expressed in adult wound healing but found at very low levels in fetal wounds. We hypothesized that AM-matrix decorated scaffolds may promote lower TGF-β1expression. An alternative hypothesis was that AM scaffolds may sequester the IL-1β from the media, preventing it from interacting with the cells within the scaffold.

With regards to the presence of TGF-β1, we report significantly less TGF-β1 released from the HA and AM containing scaffolds compared with the CS scaffolds [Fig. 5(A)]. Once IL-1β was introduced to the media, AM scaffolds were not significantly different from the other scaffold variants [(Fig. 5(B,C)]. However, scaffolds containing HA showed significantly lower (versus CS scaffolds) TGF-β1 release in the inflammatory media groups [Fig. 5(B)]. These results were normalized versus acellular scaffolds to reveal cell-specific secretions; given the potential for AM-matrix to be processed in a manner to facilitate release of large doses of growth factors,²¹ ongoing efforts are examining how these endogenous factors may also affect cell activity. To examine the interaction between IL-1β and the different scaffold variants, a pull down assay was employed. While no significant difference in sequestration of IL-1β was observed between scaffold groups, the overall trend suggests that AM scaffolds have the potential to be optimized to sequester more IL-1β than CS scaffolds (Fig. 6). Together these observations suggest that while material composition may be important in modifying the inflammatory response in vitro, the complete mechanism is not yet understood. Additional studies will seek to differentiate further, and over longer time periods, the impact of scaffold composition on growth factor production, sequestration, and release. Future studies will explore the mechanism behind the results seen here, including how cell-matrix interactions are directly influencing cellular response to a pro-inflammatory environment.

A limitation of this work is that supplementing media with a single pro-inflammatory factor is not adequate to replicate the complex inflammatory environment seen in vivo. Ongoing work is evaluating cell response to more complex inflammatory signals (e.g., TNF-α in combination with IL-1β) and in vitro macrophage phenotype in response to scaffold composition as well as inflammatory response and healing in an in vivo porcine defect model.

CONCLUSIONS

In an attempt to replicate the scarless, fetal wound healing properties in scaffolds for tendon regeneration, this study incorporated hyaluronic acid and amniotic membrane matrix into collagen-based scaffolds. In vitro experiments used interleukin-1 beta to create a pro-inflammatory environment. Bioactivity of the tenocytes cultured within these scaffold variants was assessed over a 7-day period. Following the pro-inflammatory challenge, equine tenocytes in amniotic membrane scaffold variants showed increased metabolic activity. These fetal wound inspired components also contributed to decreased tumor necrosis factor-alpha and matrix metalloproteinase-3 gene expression as well as increased expression of collagen I. Together, these results propose a biomaterial for modifying the inflammatory response associated with scar formation.