Tissue substitutes with improved angiogenic capabilities: an in vitro investigation with endothelial cells and endothelial progenitor cells.

The use of implantable biomaterials, such as artificial skin substitutes used for dermal defects, remains limited by the low angiogenic potential of these products. The rapid in vivo degradation of growth factors contributes to the limiting of angiogenesis in biomaterials. Here, we report on collagen sponges in which vascular endothelial growth factor (VEGF) was immobilized through physical binding to heparin, covalently incorporated in the matrix via cross-linking with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) and N-hydroxysuccinimide. The in vitro release of VEGF over time and endothelial cell proliferation were investigated in matrices modified at varying heparin to EDC ratios either nonloaded or loaded with VEGF. ELISA demonstrated a significantly slower in vitro release of VEGF over a period of 5 days from heparinized matrices as compared to their unmodified and cross-linked counterparts. The effects of these modifications on the proliferation of endothelial cells and endothelial progenitor cells were evaluated after 1, 3 and 5 days either according to the bromodeoxyuridine assay or total cell counting with a Neubauer chamber. The endothelial and endothelial progenitor cells cultured in contact with heparinized matrices loaded with VEGF revealed both the highest rate of DNA synthesis and the highest total cell count. Furthermore, these results show that the cross-linking of collagen matrices - both in the presence and absence of heparin - leads to increases of the proliferative activities. We can assume that these changes lead to matrices with increased angiogenic capabilities.

Biomimetic modification of the TiO(2)/glass composite Ecopore with heparinized collagen and the osteoinductive factor BMP-2.

The porous TiO(2)/glass composite Ecopore has potential applications in hard tissue replacement. We describe the modification of Ecopore with the growth factor bone morphogenetic protein-2 (BMP-2) to add osteoinductive properties. Ecopore covalently coated with BMP-2 caused a weak induction of alkaline phosphatase in murine embryonal fibroblasts. In a rabbit bone defect model, BMP-2-coated Ecopore had moderately higher bone apposition rates and ingrown bone quantities at 6 weeks after implantation. To overcome loss of function due to chemical surface coupling, we filled the pore system of Ecopore with heparinized collagen sponge and loaded this secondary matrix with BMP-2. Heparinization of collagen filling increased the BMP-2 loading capacity of the matrix approximately 1.28-fold. Within 96 h, 17.0+/-0.1 and 10.1+/-0.2% of the used BMP-2 was released from non-modified and heparinized Ecopore/collagen, respectively, indicating that the heparin modification retarded BMP-2 release. Revealed by energy-dispersive X-ray spectroscopy analysis of implant cross-sectional areas, BMP-2-loaded Ecopore/collagen had significantly higher bony ingrowth quantities in rabbits, with the heparinized modification yielding the highest value (16.09+/-3.51%, p<0.005) compared with the non-heparinized matrix (10.72+/-4.07%, p<0.05) and the BMP-2-free controls (5.60+/-1.47%). This suggested a beneficial effect of the biomimetic modification of Ecopore with heparinized collagen for bone healing and integration.

Enhancing angiogenesis in collagen matrices by covalent incorporation of VEGF.

Since the survival of ingrowing cells in biomaterials for regenerative processes largely depends on the supply of nutrients and oxygen, angiogenesis plays an important role in the development of new materials for tissue engineering. In this study we investigated the possibility of enhancing the angiogenic properties of collagen matrices by covalent incorporation of the vascular endothelial growth factor (VEGF). In a previous paper we already reported the use of homo- and heterobifunctional cross-linking agents for modifying collagen matrices [1]. In the present work the angiogenic growth factor was linked to the collagen with the homobifunctional cross-linker disuccinimidyldisuccinatepolyethyleneglycol (SS-PEG-SS) in a two step procedure. The efficiency of the first reaction step-the reaction of SS-PEG-SS with VEGF--was evaluated by western blot analysis. After 10 minutes virtually all of the dimeric molecules VEGF were on average modified by conjugation with 1 cross-linking molecule. The biological activity of the conjugate was investigated by exposing endothelial cells to non-modified VEGF and to VEGF conjugated to the cross-linker. The conjugation only had a limited effect on the mitogenic activity of VEGF. We therefore applied the cross-linking reaction to the VEGF-collagen system. In a first approach the changes were evaluated by the in vitro exposure of HUVECs to non-modified matrices, to matrices in which the VEGF was simply admixed and to matrices in which the VEGF was covalently incorporated. The angiogenic properties were evaluated in vivo with the chorioallantois membrane model. In this assay the chorioallantois membrane of the chicken embryo was exposed to the same set of matrices. The covalent incorporation of VEGF has a small but significant effect both on the formation of microvessels in the chorioallantois membrane and the tissue ingrowth into the implant. The covalent incorporation of angiogenic growth factors may thus be considered as a promising approach for enhancing the angiogenic capabilities of collagen matrices. Also the cross-linking with the homobifunctional cross-linking agent has a positive effect on the angiogenic potential of the collagen matrices.

The impact of vacuum freeze-drying on collagen sponges after gas plasma sterilization.

The sterilization of porous collagen sponges remains a challenging procedure. Gamma irradiation denatures collagen, resulting in dramatic changes to its structure. Ethylene oxide leaves toxic residues requiring weeks to evaporate. This study investigated the impact on cell behavior of gas plasma treatment when combined with vacuum freeze-drying. The goal of this procedure is to eliminate the molecules of hydrogen peroxide remaining after the sterilization process, together with their decomposition products, from the scaffolds. These molecules hinder the immediate use of the porous designs. Collagen and EDC/NHS-heparinized collagen scaffolds were sterilized with gas plasma. H2O2 released by the collagen specimens was measured by peroxidase test both immediately and also 1 week after the plasma treatment. Further measurements were done 24, 36, 48 and 72 h after vacuum freeze-drying. The activity of these scaffolds was further evaluated in relation to the proliferation, migration and differentiation of human umbilical vein endothelial cells (HUVECs). Both immediately after exposure to gas plasma and also 1 week later, the collagen designs contained significantly higher concentrations of H2O2 than scaffolds having also undergone vacuum freeze-drying. This procedure achieved faster decontamination of the remaining H2O2. Following vacuum freeze-drying, sponges already allowed HUVEC proliferation after 48 h, but in non-lyophilized specimens after gas plasma treatment alone, cell death occurred as early as only 1 week later. These data highlight the advantages of carrying out vacuum freeze-drying following gas plasma sterilization. The results show the substantial impact of sterilization of porous materials made for tissue engineering.

Enhanced dermal regeneration using modified collagen scaffolds: experimental porcine study.

Ongoing research has achieved much progress towards the development of new artificial skin substitute products. However, effective implant material for correcting full-thickness defects (such as those arising from extensive burns, tumor resection, hereditary or congenital defects and chronic wounds) has not yet become available. Following split-thickness skin grafting, contraction and scar formation are unavoidable. These phenomena are believed to be due to poor dermis regeneration. Collagen dermis substitute has been developed for the treatment of deep, poorly vascularized tissue defects. However, their application is problematic, because scaffolds of this kind fail to adequately induce the neoangiogenesis needed for regeneration. To overcome these shortcomings a number of matrices have been chemically modified. Furthermore, these matrices first require implantation and follow-up by skin grafting 3 to 5 weeks later. In this article we describe new materials made of modified collagen which enhance dermal regeneration and neoangiogenesis. The procedure was applied in successfully dealing with full-thickness defects in pigs, with subsequent split-thickness skin grafting being performed immediately afterwards. Histological findings revealed that the neodermis obtained resembles a normal dermis structure. These scaffolds have the potential of serving as an off-the-shelf skin replacement in the reconstruction of deep wounds, thus supporting wound-healing processes.

Effects of modified collagen matrices on human umbilical vein endothelial cells.

The most commonly used biomaterials fail to ensure sufficient angiogenesis for fast in vivo incorporation. This results in central necrosis and consequent infection. One way of obtaining a high angiogenic response is the application of vascular endothelial growth factor (VEGF). To obtain a sustained release of these cytokines, heparin was incorporated into collagen matrices using 1-ethyl-3(3-dimethyl-aminopropyl) carbodiimide (EDC) and N-hydroxysuccinmide (NHS). The functionality of the heparinized collagen matrices was then enhanced by immobilization of VEGF via its heparin-binding domain. This procedure changed in vitro degradation behavior and water-binding capacity. Accelerated endothelial cell proliferation was also achieved. A range of different heparin and EDC/NHS concentrations in combination with VEGF induced variation in endothelial cell growth and tubulogenic formation. Polymerized collagen scaffolds presented biointeractive systems with integrated angiogenic activity. This may become a useful tool in the clinical therapy of disorders connected with wound healing.

Modulation of angiogenic potential of collagen matrices by covalent incorporation of heparin and loading with vascular endothelial growth factor.

One of the prominent shortcomings of matrices for tissue engineering is their poor ability to support angiogenesis. We report here on experiments to enhance the angiogenic properties of collagen matrices. Our aim is to achieve this goal by covalently incorporating heparin into collagen matrices and by physically immobilizing angiogenic vascular endothelial growth factor (VEGF) to the heparin. The immobilization of heparin was performed with 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC) and N-hydroxysuccinimide (NHS). Carboxyl groups on the heparin are activated to succinimidyl esters, which react with amino functions on the collagen to zero length cross-links. This modification leads--in addition to the incorporation of heparin--to gross changes in in vitro degradation behavior and water-binding capacity. As a first approach to testing angiogenic capabilities, endothelial cells were exposed to nonmodified and heparinized collagen matrices. This exposure leads to an increase in endothelial cell proliferation. The increase can be further enhanced by loading the (heparinized) collagen matrices with VEGF. Evaluation of the angiogenic potential of heparinized matrices was further investigated by exposing them to the chorioallantoic membrane of chicken embryos and to the subcutaneous tissue of rats. Both approaches show that heparinized matrices have substantially increased angiogenic potential. In particular, the loading of heparinized matrices with VEGF invokes a further increase in angiogenic potential. It is apparent that the physical binding of VEGF to heparin allows for a release that is beneficial to angiogenesis. By varying the heparin and EDC/NHS concentrations during the modification process and by varying the loading with VEGF, the angiogenic potential as well as the degradation behavior can be adapted to obtain matrices that fulfill specific angiogenic requirements in the field of tissue engineering.

High density binding of proteins and peptides to poly(D,L-lactide) grafted with polyacrylic acid.

The use of graft polymers for the functionalisation of biomaterial surfaces is already widespread. We investigated the adsorptive and covalent binding of a variety of proteins and peptides to poly(D,L-lactide) grafted with polyacrylic acid. Covalent attachment was achieved through coupling of amino groups of the protein/peptide to the carboxyl groups of the graft polymer by using a water-soluble carbodiimide and N-hydroxysuccinimide. Binding densities were determined by automated amino acid analysis after acid hydrolysis of both the poly(D,L-lactide) and the adsorbed and covalently bound proteins. Experiments in the absence and presence of the coupling reagents allow to discriminate between adsorptive and covalent binding. Although the adsorptivc binding is quite substantial in absolute terms, the amount of adsorbed protein is relatively low as compared to the total amount of bound protein. Total binding densities of 20-30 microg/cm2 can easily be achieved. Depending on the concentration and on the properties of the proteins and peptides, between 5% and 80% of the totally bound protein may be physically adsorbed. Densities expressed in molecules/10 nm2 vary from 0.5 molecule fibronectin to 2,000 laminin-peptide molecules: their binding densities clearly correlate with their respective molecular masses. Obviously, the binding densities are governed by their individual three-dimensional space requirements rather than the density of the available carboxyl groups. From the number of carboxyl groups/10 nm2 (18,000-30,000 COOH/10 nm2) the average length of the acrylic acid graft polymer molecules was estimated. Based on the assumption that about 10 copolymer chains can be accommodated on 10 nm2, the average length of the polymer chains, which corresponds to the thickness of the graft phase, is estimated to be 0.5-1 microm. The organisation of the proteins and peptides within the polyacrylic acid phase was further investigated by experiments in which a protein (BSA) and a peptide (Val-Lys) were allowed to react in either a singular, a consecutive or a simultaneous way. Together with XPS and IR-ATR surface characterisation experiments a three-dimensional picture of the arrangement of the immobilised proteins and peptides within the graft polymer phase emerges.

The use of bifunctional polyethyleneglycol derivatives for coupling of proteins to and cross-linking of collagen matrices.

The realization of three-dimensional (3D) degradable matrices which slowly release bio-active components represents a major challenge in the field of tissue engineering. In this paper we report on the usage of commercially available bifunctional agents for both the covalent coupling of proteins to and the cross-linking of collagen matrices. Proteins - horse radish peroxidase (HRP) was used as a model protein - were cross-linked with either a homobifunctional (disuccinimidyldisuccinatepolyethylene-glycol) or a heterobifunctional (N-hydroxysuccinimidylvinylsulfonepolyethyleneglycol) agent. In the case of the heterobifunctional cross-linking agent the collagen matrices were previously modified with succinimidylacetylthioacetate in order to introduce sulfhydryl groups. As compared with control experiments a 10-fold and 50-fold increase of immobilized proteins were achieved with the homobifunctional and heterobifunctional cross-linker resp. The HRP-PEG conjugates demonstrated a better long-term stability as compared to the non-treated HRP. The effects of the cross-linking agents and the thiolation reagent succinimidylacetylthio acetate on the in vitro degradation of the collagen matrices by collagenase were also investigated. In particular the reaction with succinimidylacetylthio acetate appears to offer interesting opportunities both for coupling active proteins and modulating the degradation times of collagen matrices.

Mutations in the Ca2+ binding site of the Paracoccus denitrificans cytochrome c oxidase.

Recent structure determinations suggested a new binding site for a non-redox active metal ion in subunit I of cytochrome c oxidase both of mitochondrial and of bacterial origin. We analyzed the relevant metal composition of the bovine and the Paracoccus denitrificans enzyme and of bacterial site-directed mutants in several residues presumably liganding this ion. Unlike the mitochondrial enzyme where a low, substoichiometric content of Ca2+ was found, the bacterial wild-type (WT) oxidase showed a stoichiometry of one Ca per enzyme monomer. Mutants in Asp-477 (in immediate vicinity of this site) were clearly diminished in their Ca content and the isolated mutant enzyme revealed a spectral shift in the heme a visible absorption upon Ca addition, which was reversed by Na ions. This spectral behavior, largely comparable to that of the mitochondrial enzyme, was not observed for the bacterial WT oxidase. Further structure refinement revealed a tightly bound water molecule as an additional Ca2+ ligand.

Perturbation of the CuA site in cytochrome-c oxidase of Paracoccus denitrificans by replacement of Met227 with isoleucine.

Subunit II of cytochrome-c oxidase contains a redox centre, CuA, with unusual spectroscopic properties; this site consists of two copper atoms and acts as the entry point for electrons from cytochrome c. We have constructed a site-directed mutant of cytochrome-c oxidase from Paracoccus denitrificans in which the CuA site has been disturbed by replacement of Met227 with isoleucine. The purified, fully assembled enzyme complex has been investigated with various techniques including metal analysis, EPR and visible spectroscopies, steady-state and fast kinetics. The stoichiometry of the metals in the enzyme remains unchanged but a clear perturbation of the CuA site can be observed in the EPR and near-infrared optical spectra. It is concluded that in the mutant CuA is still binuclear but that the two nuclei are no longer equivalent, converting the delocalized [Cu(1.5)....Cu(1.5)] centre of the wild type into a localized [Cu(I)....Cu(II)] system. Changes in the overall kinetics of the mutant are correlated with a diminished electron transfer rate between CuA and heme alpha.

Stoichiometry and redox behaviour of metals in cytochrome-c oxidase.

The early observation of extra copper in preparations of cytochrome-c oxidase has recently lead to a renewed interest in its stoichiometry and possible redox function. In various, pure preparations (heme A contents close to the theoretical value of 9.79 nmol/mg protein for the 13-subunit bovine enzyme) protein-related metal stoichiometries of 3 Cu, 2 Fe, 1 Zn, 1 Mg/monomer with M(r) 204266 were determined. Despite the presence of five potential redox metal ions, reductive and reoxidative titrations indicate the presence of only four one-electron-accepting/donating species in the ligand-free enzyme. Participation of two copper ions in a binuclear copper site acting as one-electron acceptor may explain both the observed copper stoichiometry and the redox behaviour. The homology of the C-terminal sequence of subunit II with one of the copper-binding sites in nitrous-oxide reductases provides possible ligands for complexing two copper ions in a binuclear center.

A comparative EPR investigation of the multicopper proteins nitrous-oxide reductase and cytochrome c oxidase.

The multicopper proteins, nitrous-oxide reductase (N2OR) and cytochrome c oxidase (COX), were investigated by EPR spectroscopy at microwave frequencies 2.4-35 GHz. Our results support a Cu-Cu interaction in COX and N2OR. At least 10 lines in the 2.7-GHz, 12 lines in the 4.6-GHz and 14 lines in the 9.2 GHz spectra were resolved for N2OR. Eight copper lines at 2.7 GHz, about nine lines at 4.6 GHz and about six lines at 9.2 GHz were resolved for COX. Simulations of the EPR spectra were consistent with most of the resonances of the multiline spectra, including regions in the center of the spectra where overlap of the three seven-line patterns is proposed. These simulations indicated that Cu-Cu interaction, in a mixed-valence [Cu(1.5) ... Cu(1.5)], S = 1/2 site is consistent with, if not proof of, the unusual spectral features observed for N2OR and COX.

Cytochrome c oxidase in Paracoccus denitrificans. Protein, chemical, structural, and evolutionary aspects.

Preparations and protein chemical characterizations performed with cytochrome c oxidase (E.C. 1.9.3.1) from the purple bacterium Paracoccus denitrificans are reviewed. The simplest catalytically competent complex of the enzyme consists of two subunits of 62012 and 2799 Da. The theoretical heme a/protein ratio of the purified enzyme is 22.0 nmol/mg. The amino acid sequences of both proteins are compared with examples of subunits I and II of mitochondrial terminal oxidases from the main kingdoms of eukaryotes. The significance of the emerging conserved features such as membrane penetration patterns, invariant residues, stoichiometry, and sites of prosthetic groups are discussed. The Paracoccus enzyme represents the only prokaryotic oxidase detailed so far, which is directly related to the mitochondrial oxidases by common ancestry in the growing O2 atmosphere.

Rapid purification of cytochrome c oxidase from Paracoccus denitrificans.

Two methods are described for the purification of cytochrome c oxidase from Triton X-100 extracts of the periplasma membrane of Paracoccus denitrificans. The first is a large-scale procedure for the preparation of 100-250 nmol of cytochrome c oxidase (10-20 mg) in 1 week. The second is a rapid procedure for isolating up to 25 nmol in 2-3 days. Owing to the high yields given by fast protein liquid chromatography (FPLC) on Mono Q columns, the overall yield is about 20%, whereas the yield in many other previously published procedures does not exceed 10%. The use of FPLC on Mono Q also offers a considerable saving of time.

Multifrequency EPR evidence for a bimetallic center at the CuA site in cytochrome c oxidase.

Multifrequency electron paramagnetic resonance (EPR) spectra of the Cu(II) site in bovine heart cytochrome c oxidase (COX) and nitrous oxide reductase (N2OR) from Pseudomonas stutzeri confirm the existence of Cu-Cu interaction in both enzymes. C-band (4.5 GHz) proves to be a particularly good frequency complementing the spectra of COX and N2OR recorded at 2.4 and 3.5 GHz. Both the high and low field region of the EPR spectra show the presence of a well-resolved 7-line pattern consistent with the idea of a binuclear Cu center in COX and N2OR. Based on this assumption consistent g-values are calculated for gz and gx at four frequencies. No consistent g-values are obtained with the assumption of a 4-line pattern indicative for a mononuclear Cu site.

Integral complexes of cytochrome c oxidase contain three coppers.

Metal contents have been determined in beef heart cytochrome c oxidase by inductively coupled plasma atomic emission spectroscopy. Integral complexes of this enzyme contain three copper and two iron atoms, as well as one zinc and one magnesium atom (Cu: 2, 91 +/- 0, 13; Fe: 2, 05 +/- 0, 17; Zn: 1, 03 +/- 0, 03; Mg: 1, 01 +/- 0, 10). The combination of these results with those reported for the c1 aa3-oxidase from Thermus thermophilus leads to the conclusion that subunit I is the universal two coppers and two hemes a binding catalytic unit common to all oxidases of the aa3-type. Subunit II, which binds the third copper ion, functions as an electron conducting unit, transferring electrons from cytochrome c to the four redox centers in subunit I. Preliminary titration experiments with NADH reveal, in agreement with this catalytic organisation, the presence of five redox centers.

Subunit II of cytochrome c oxidase from Paracoccus denitrificans. DNA sequence, gene expression and the protein.

Cytochrome c oxidase from the bacterium Paracoccus denitrificans, while being related to the mitochondrial enzyme in many ways, consists of only two to three different subunits. For the identification of its genes, a Paracoccus DNA library was constructed and screened with specific antibodies for expression of cloned inserts in E. coli. A positive clone expressing immunoreactive products in the molecular mass region of authentic subunit II revealed a high homology of its DNA-deduced amino acid sequence with subunit II sequences of the mitochondrial oxidases; several typical features, such as the transmembrane folding pattern and the presumed copper-binding site, are highly conserved between prokaryotic and mitochondrial polypeptides. A comparison with peptide sequencing data of the purified subunit established the presence of a characteristic N-terminal extension as well as a longer C terminus in the initial translation product of the Paracoccus subunit; by mass spectroscopy, the first N-terminally blocked residue of the mature polypeptide was identified as a pyroglutamate. No code abnormalities, but a highly specific codon usage were observed; no evidence for a localization of the subunit I gene directly adjacent to this gene has been obtained.

Cytochrome c oxidase is a three-copper, two-heme-A protein.

Metal contents of preparations of procaryotic (Paracoccus denitrificans) and eucaryotic (beef heart) cytochrome c oxidases have been determined by inductively coupled plasma atomic emission spectroscopy and shown to be stoichiometrically related to the protein contents. The results show that oxidases which possess subunits I and II have three copper atoms besides hemes a and a3 (Paracoccus denitrificans, Cu: 2.97 +/- 0.08 and Fe: 2.09 +/- 0.10; bovine heart, Cu: 2.83 +/- 0.07 and Fe: 1.94 +/- 0.12). Together with data reported for the c1 aa3 oxidase from Thermus thermophilus, the following conclusions can be drawn. Subunit I binds two copper atoms and both hemes a and a3 and thus is the universal terminal oxidase of this spectral type. Subunit II binds one copper and functions as an electron conductor. The mitochondrial respiratory complex IV binds, in addition to three copper and two hemes a, stoichiometric amounts of magnesium and zinc (bovine heart Mg: 0.98 +/- 0.05 and Zn: 1.01 +/- 0.04).