The Mechanism of Cell Interaction and Response on Decellularized Human Amniotic Membrane: Implications in Wound Healing

The human amniotic membrane has been used as a biological dressing for skin burns,wounds, and chronic leg ulcers. Its therapeutic effects have been attributed to its ability to promote epithelialization, inhibit fibrosis, and act as an antimicrobial agent.¹ Human amniotic membrane products currently on the market, including Amniograft^® and Prokera^™ (Bio- Tissue Inc, Miami, Fla), are frozen with intact cellular structure, associated growth factors, and cytokines. ACELAGRAFT^™ (Celgene Cellular Therapeutics, Cedar Knolls, NJ) is distinct in that it is a decellularized and dehydrated human amniotic membrane (DDHAM) with all cells and associated growth factors removed. As a result, DDHAM can be shipped and stored at room temperature. DDHAM has demonstrated potential as a wound healing product in ongoing studies for the treatment of acute and chronic ulcers. Even with all cells and related factors removed, DDHAM still possesses the necessary biological properties to augment wound healing processes. In this work, a detailed mechanistic analysis of the mode of interaction of cells with DDHAM is presented. The response of human dermal fibroblasts to DDHAM was evaluated to complement the current clinical studies with DDHAM for the treatment of skin wounds. The biochemical, cell biology, and gene array studies together provide a mechanistic understanding of the wound healing process as it relates to DDHAM.

Methods

Preparation of membranes. DDHAM was prepared using proprietary methods as previously described.^2,3 In short, the amniotic membrane was excised from qualified term placentas and was washed and scraped to remove extraneous tissue and cells.This was followed by a decellularization of the tissue using deoxycholic acid and drying of the tissue using a gel dryer. An amniotic membrane treated as such is free of cells. The decellularized and dehydrated human amniotic membrane product is sold under the name ACELAGRAFT.

Biochemical analysis. The decellularized and dried membrane was completely solubilized using any of the 4 methods summarized in Scheme I. Tissue (1–3 mg) was solubilized in 1–3 mL of 10 mM HCl at 100^°C for 1 h. Tissue was also digested using either collagenase (ratio of 1:100 at 37^°C for 18 h) or pepsin (1 mg/mL) in 0.5 M acetic acid at 37^°C for 18 h. To identify growth factors (cytokines and/or hormones), the amniotic membrane tissue was treated by methods previously described.⁴ The tissue was treated with a buffer (5 mL) consisting of 2 M guanidium HC1 with 100 mM tris buffer, (pH 7.2, 5 mM EDTA, 1 mM DTT, 1 mM PMSF, and 1 mM β-mercaptoethanol) for 24 h at 4^°C.The resulting supernatant was dialyzed against water for 24 h at 4^°C and the sample was dried for biochemical analysis. Total collagen in DDHAM was quantitated using a Sircol assay kit (Accurate Chemical and Scientific Corp., Westbury, NY). Collagens I, III, and IV were quantitated using sandwich ELISAs with primary and secondary antibodies purchased from Rockland Immunochemicals, Inc. (Gilbertsville, Pa). Elastin and glycosaminoglycans (GAG) content were determined using the FASTIN and BLYSCAN dye based assay kits (Accurate Chemical and Scientific Corp., Westbury, NY). Fibronectin and laminin were quantitated using sandwich ELISA kits (Takara Bio Inc., Madison, Wis). Residual maternal hormones and growth factors were assessed using ELISAs from R&D Systems (Minneapolis, Minn).

Immunofluorescent imaging. To examine cell morphology in response to DDHAM, dermal fibroblasts (Cambrex Corp., Baltimore, Md) were seeded onto pieces of tissue previously secured onto tissue culture polystyrene (TCPS) 24 well plates using silicon O-rings and incubated for 3, 24, and 48 h. Cells were fixed with 10% formalin and stained with an antifibronectin antibody (Sigma-Aldrich, St. Louis, Mo) followed by an AlexaFluor 594 conjugated anti-rabbit antibody (Invitrogen Corp., Carlsbad, Calif); samples were also stained for F-actin with AlexaFluor 488 conjugated phalloidin.

Cytokine secretions. Media samples were removed at 1, 24, and 48 h to test for fibronectin, laminin, GAGs, and collagens using assays as previously described. Cytokine secretions from the same supernatants were quantitated using multiplexed antibody arrays (Biosource, Camarillo, Calif). The group of cytokines tested included several interleukins (1–17), macrophage inducible proteins (MIPs), interferons (IFNs), tumor necrosis factor-alpha (TNF-α), eotaxin, and granulocytemacrophage colony-stimulating factor (GM-CSF). The biochemistry of the tissue-cell interaction was investigated using competitive binding assays. Fibroblasts were incubated with TCPS-immobilized DDHAM for 2 h in the presence of fibronectin, laminin, collagen, and antibodies to several integrins (β1/CD29, α6 subunit/CD49f, β2, α4/CD49D, and α5). Cells were hematoxylin and eosin (H&E) stained to visualize cell binding to DDHAM.

Gene expression profiling. Dermal fibroblasts (DF) purchased from Cambrex (Baltimore, Md) were cultured in Fibroblast Growth Medium (Cambrex, Baltimore, Md). Three 6-cm x 8-cm sheets of DDHAM were hydrated with culture medium prior to seeding with 1 x 106 adult dermal fibroblasts. After 5 days of incubation, approximately 2 x 106 cells were recovered. Control cultures of fibroblasts were grown in triplicate on TCPS Petri dishes. Two micro arrays were processed from each culture to produce a total of 6 data sets for the growth on DDHAM and TCPS control. Cells from both conditions were recovered in trypsin replacement TrypLE Express prior to lysis. Total RNA was isolated with the RNeasy Mini Kit (Qiagen, Valencia, Calif). RNA concentrations were determined with a NanoDrop (ND-1000) spectrophotometer and integrity was determined with an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, Calif). Ten micrograms of total RNA from each culture were run on the Affymetrix GeneChip platform (Santa Clara, CA). Total RNAs were converted to labeled cRNAs and hybridized to oligonucleotide Human Genome U133A 2.0 arrays according to the manufacturer’s methods (Affymetrix, Santa Clara, Calif). Image files were processed with the Affymetrix MAS 5.0 software to generate .chp text files. Data from the text files were normalized and analyzed with Agilent GeneSpring 7.3 software.

Results

Biochemical analysis. Results indicated that DDHAM is essentially an ECM-like material, composed of collagens (56%), elastin (18%), and other ECM components, such as GAGs (0.5%), fibronectin (0.3%), and laminin (Table 1)—these ECM molecules along with 10%–14% water account for about 90% of the total dry weight of the tissue. The remaining 10% are components yet to be identified. Further, as summarized in Table 2 and Table 3, DDHAM does not contain any detectable growth factors, cytokines, or hormones.

Cell morphology and fibronectin matrix assembly. The morphology of fibroblasts bound to DDHAMwas examined using actin and fibronectin immunofluorescent staining techniques. As shown in Figure 1, fibroblasts on fibronectin were well spread, displaying actin stress fibers characteristic of an adherent cellular phenotype. However, fibroblasts on collagen were not as well spread. They showed numerous filopodial projections emanating from the central cell body. By 24 h and 48 h of culture, the fibroblasts proliferated and densely populated the surface of the tissue, with differences in cellular morphology between the various substrates being less apparent, as fibroblasts elaborate their own matrix proteins. This demonstrates that the fibroblasts were able to modify their extracellular environment and therefore, no longer responded to the original experimental extracellular substrate.

As shown in Figure 2, fibronectin staining of fibroblasts cultured on DDHAM at the 1-h timepoint showed intense intracellular fibronectin staining and no extracellular staining. In comparison, fibroblasts on fibronectin-coated surfaces showed no evidence of intracellular fibronectin. Instead, a diffuse extracellular staining representing the fibronectin preadsorbed to the surface was observed.

By 48 h of culture, for both the fibronectin-coated surfaces and the DDHAM samples, cells had spread extensively, as shown by the network of actin filaments. Fibroblasts on both surfaces had also elaborated a dense fibronectin matrix. Qualitatively, the fibronectin fibrils on the DDHAM sample appeared to be more intensely fluorescent, suggesting enhanced fibronectin matrix deposition on DDHAM as compared to fibronectin coated-surfaces. These data show that DDHAM supports, and perhaps even enhances, one of the most important functions of fibroblast activity the ability to elaborate and remodel the extracellular matrix.

Competitive cell binding assays. To further examine the possibility that fibroblasts interact with DDHAM via well known integrin-fibronectin interactions,⁵ competitive cell binding assays were performed. In these studies, fibroblasts were seeded on tissue culture trays with immobilized DDHAM. Fibroblasts were allowed to bind to DDHAM for 2 h in the presence of medium. In a separate experiment, fibroblasts were seeded on the immobilized tissue culture trays in the presence of fibronectin, laminin, collagen, or the integrin antibodies (as described in Methods). If the fibroblasts bound to DDHAM by recognizing fibronectin in the tissue, then addition of fibronectin should compete with that interaction and prevent the fibroblasts from binding to DDHAM.

As shown in Figure 3, cells showed rapid binding (within 2 h) to DDHAM in the presence of medium alone. Addition of fibronectin blocked that interaction as measured by no cell binding to DDHAM. A similar result was obtained with anti-integrin antibodies, which would functionally block the integrins on the fibroblasts. Laminin had no effect on cell binding, indicating that fibroblasts do not bind to DDHAM through laminin recognition. As with fibronectin,collagen was also able to compete with DDHAM for cell binding. Antibodies to integrins β1/CD29, β2, and α5 blocked initial fibroblast binding, antibodies to α6 subunit/CD49f,α4/CD49D and α5 were not able to block the binding, suggesting that the integrins β1/CD29, β2, and α5 participate in the initial binding of fibroblasts to DDHAM.

Cell secretion of ECM proteins on DDHAM. Fibroblasts were grown on DDHAM for 2–4 days untilthe cells were confluent.When cells bind to DDHAM in the short binding assays described here, they bind in “preferred areas” and in large clusters. However, when grown over a period of several days, they were able to completely cover the tissue. At the end of the 4-day growth period, culture medium was harvested and examined using a series of multiplexed antibody ELISAs to identify expressed cytokines and ECM proteins. The data obtained with cells on DDHAM was compared to fibroblasts grown on nontissue culture treated plates and tissue culture treated plates. Examination of medium for secretion of ECM proteins, such as fibronectin, laminin, GAGs, and collagens showed a 2-fold increase in the secretion of fibronectin when compared to cells on tissue culture/nontissue culture treated plates (Figure 4). No increase in secretion of collagen, laminin or GAGs was observed.

Cell secretion of cytokines on DDHAM. A 25-plex antibody array was used to examine the secretion of cytokines from DDHAM bound fibroblasts.These include several interleukins (1–17), MIPs, IFNs, TNF-α, eotaxin, and GM-CSF. Results show a modest increase (2–3 fold) in the expression of monocyte chemoattractant protein (MCP-1), IL-3, IL-12, and GM-CSF as a result of fibroblast culture on DDHAM compared to polystyrene plates (Figure 5). More dramatic increases were detected as a result of culture or growth on DDHAM including a 70- fold increase in IL-6 secretion, 140-fold increase in IL-8 secretion, and a 30-fold increase in eotaxin secretion.

Microarray analysis of dermal fibroblasts cultured on DDHAM. Genes up-regulated as a result of growth on DDHAM were identified from the microarray data sets by a 3-step filtering approach. Initially, a list of genes was compiled based on Affymetrix flag calls of “Present” in all data sets from fibroblasts cultured on DDHAM. This list was further refined to genes that had statistically significant differences (t-test) between the growth on control condition (TCPS) and the growth on DDHAM. Finally, genes with an expression level change of 3 or more fold were identified—this approach produced a list of 25 genes (Table 4). A comparison with the Gene Ontology database showed that the categories represented with the highest probabilities include response to chemical stimulus, response to stress, response to injury, immune defense, and inflammatory responses. Additionally, there are genes related to collagen metabolism, including MMP-3,⁶ MMP-1, super oxide dismutase, an oxygen radical scavenger shown to accelerate wound healing,⁷ and activator protein-1 (AP-1) complex components FOS and JUNB. Genes that were down regulated 3-fold or more by growth on DDHAM are shown in Table 5.

Discussion

Wound healing is a well-orchestrated dynamic process involving an intricate interaction between the ECM,cells,and a variety of growth factors and cytokines. During this process fibroblasts serve to assemble and remodel the extracellular matrix. The ECM serves as a physical conduit for cells to infiltrate into the wound space, and cells (and the ECM) provide the biochemical cues necessary to activate cellular responses, including growth factor secretion, that is required for wound repair.

Since the amniotic membrane is a complex, highly networked tissue, biochemical methods were developed to identify and quantitate its biochemical components including extracellular proteins, proteoglycans, and glycoproteins. The membrane was solubilized using a variety of specific methods summarized in Scheme I. For example, dissolution of the tissue in 10 mM HC1 allowed for the detection of elastin and GAGs, but not collagens. Collagens could only be quantitated after pepsin digestion of the tissue.The presence of residual of the more labile molecules, such as cytokines and hormones,required the use of collagenase digestion of the amniotic membrane, rather than a more general proteolytic enzyme like pepsin.

The processing of the amniotic membrane includes mechanical scraping, deoxycholic acid decellularization, and heat drying. These steps may cause denaturation of polypeptides (eg, growth factors). To examine DDHAM for the presence of growth factors, guanidinium-HCl extraction methods⁴ were used. In control experiments, the tissue was processed with extraneously added growth factor. Results indicated that the guanidinium-HCl extraction method did not denature the growth factor or cytokine that was being detected. Further, amniotic membrane that had not been processed using the methods described above^2,3 showed the presence of growth factors (data not shown here). Thus, the extraction methods used in this study were valid and allowed detection of the presence of growth factors.

As shown throughout this work, DDHAM is an ECM-like tissue, which is devoid of cells. It likely mimics the scaffolding function of the ECM allowing fibroblasts to bind and contribute to the wound healing process at the wound site. DDHAM contains key cell adhesion protein molecules, such as fibronectin. Cells, including fibroblasts, recognize fibronectin in DDHAM via fibronectinintegrin interactions, and are stimulated to secrete fibronectin and assemble an extracellular matrix. The fibroblasts are also stimulated to secrete a variety of growth factors and cytokines that invigorate the wound healing process.

Fibroblasts cultured on DDHAM show similar morphologies to fibroblasts cultured on fibronectin—these results suggest that DDHAM provides a surface that is amenable to cellular adhesion, viability, and functionality. Effective wound repair requires remodeling of the extracellular matrix, particularly in the form of new matrix deposition. Therefore, the ability of fibroblasts cultured on DDHAM to deposit fibronectin, a component of the wound provisional matrix, into matrix fibrils was analyzed and compared to cells cultured on substrates precoated with fibronectin. It is well known from the literature that the deposition of fibronectin into the extracellular matrix (ECM) determines the matrix deposition and maintenance of other matrix proteins, such as collagen- 1, thrombospondin, tenascin-C, fibulin, and fibrinogen.^8–13 Fibroblasts cultured on DDHAM exhibited a propensity to deposit fibronectin onto matrix fibrils. In a wound milieu, DDHAM may serve not only as a physical conduit for infiltrating cells, but also as a provider of biochemical cues that induce signal transduction events resulting in the deposition of extracellular matrix in the wound space.

A more significant consequence of fibroblast binding to DDHAM is the secretion of cytokines, which are involved in recruiting other cells to the site of injury, initiating cell-cell communication and allowing for further progression of the wound healing process.¹⁴ The role played by these cytokines (ie, IL-6, IL-8, MCP-1, and eotaxin) in the wound healing process are multifunctional and intricate. Monocyte chemoattractant protein-1 (MCP-1), pro- and IL-8, belong to a family of chemotactic cytokines or chemokines that are involved in recruiting leukocytes to the wound site, promoting ECM production, and regulating repair processes.They are also involved in regulating wound healing processes, such as angiogenesis and hematopoiesis. In rat models, MCP-1 has been examined for its ability to accelerate reepithelization and collagen synthesis in wound healing.¹⁵

IL-8 stimulates early angiogenic activity in the wound site, a function VEGF takes over at later stages of the wound healing process. IL-8 can stimulate fibroblasts to form myofibroblasts,which facilitate wound closure.¹⁶ IL- 8 has been tested in a guinea pig model for deep, partialthickness skin burns with extraneous IL-8 therapy and has been shown to enhance the reepithelization process of wound healing.¹⁷

IL-6 is a proinflammatory cytokine that orchestrates early immune response by recruiting white blood cells to a wound site. White blood cells clean the wound of debris and bacteria. In IL-6 knockout mice, cutaneous wound healing is significantly delayed, arguing for an important role for this cytokine in accelerating wound healing.¹⁸ IL-6 has also been implicated in accelerated wound healing in a rat model.¹⁹

Conclusion

It is generally accepted that the presence of cells, growth factors, and cytokines are critical to the wound healing properties of the amniotic membrane. This work, however, describes an amniotic membrane product that contains sufficient biochemical entities to illicit biological responses despite being devoid of any cells and cellular components, growth factors, or cytokines. Dehydrated human amniotic membrane binds fibroblasts and induces them to release important ECM proteins and cytokines. In turn, these signals could have the ability to recruit new cells to the wound site and initiate the wound healing process. In a wound site, DDHAM may also manifest its wound healing properties through similar biological effects and responses as described in this study.

ACELAGRAFT^™ is a trademark of Celgene Corporation that refers to Celgene Corporation’s brand of dehydrated and decellularized human amniotic membrane.