Two approaches to the reconstitution of tissues and organs are reviewed. The first consists of imitating the architecture of actual tissues and organs by combining cultured specialized cells with extracellular matrix components to produce a connective tissue substrate on or in which epithelial, mesothelial or endothelial cells can be plated or seeded and subsequently differentiate into mono or multilayered tissues and other structures. The second consists of providing an acellular framework of extracellular matrix constituents that can be occupied by adjacent host tissues after implantation in vivo and be remodeled by them to resemble the host tissues it is designed to replace. A paradigm for events in vivo, designed to study the process of remodeling of acellular matrices in vitro has been developed. The living skin equivalent (LSE), an example of a product fabricated using the first approach to tissue engineering, has been adapted to study events of extracellular matrix remodeling, relevent to the second approach to tissue engineering. After creating a disc shaped wound bed in an LSE, the wound is filled with a collagen matrix with or without added supplements and the process of epidermal wound closure and associated events in the dermis are followed. It is shown that fibroblast conditioned medium or a simple molecule such as ascorbic acid, added with no additional growth factors to the collagen matrix used to fill the wound bed, strongly stimulates the process of repair. Dermal fibroblasts from the adjacent tissue invade the collagen lattice that forms in the wound bed, and keratinocytes recruited from the wound edge overgrow the new dermal tissue. The applicability of the paradigm to the repair of vascular and other tissues will be discussed and approaches to optimizing the composition of acellular constructs considered.

Results of implantation studies in a variety of animal tissue models demonstrate that the rate of biogradation of a collagen scaffold should parallel the rate of wound healing observed in particular anatomic sites. This rapid degradation maximizes tissue regeneration and minimizes encapsulation of the implant. The following paper reviews the effects of crosslinking on the rate of tissue ingrowth and regeneration. In addition, preliminary mechanical data on newly developed soluble type I collagen fibers is presented as a possible advance in the production of high strength collagen based tissue scaffolds.

Using both in vivo and in vitro experiments we have demonstrated that: reconstituted collagen will undergo mineralization in a healing bony compartment; that this mineralization is the result of spontaneous precipitation of calcium salts due to the presence of alkaline phosphatase produced by the bone cells, and that once calcified, the collagen will undergo cellular resorption by tartrate-resistant multi-nucleate giant cells similar to osteoclasts. This sequence of events is quite different to that in the supra-bony soft-tissue compartment where no calcification of the collagen is apparent, the collagen matrix becomes infiltrated with fibroblast-like cells and little resorption of the matrix occurs during implantation.

We conclude that reconstituted collagen may be employed as both a tissue barrier, enhancing guided tissue regeneration, and a bone-substitute material, which becomes replaced by natural bone tissue.

We investigated the in vitro cytotoxicity of commercially available hexamethylenediisocyanate-crosslinked dermal sheep collagen (HDSC). HDSC was found to induce medium cytotoxic effects, as measured with methylcellulose cell culture. Apart from primary cytotoxicity, due to direct release of (extractable) cytotoxic products, HDSC was found to contain secondary cytotoxicity, possibly released by enzymatic interactions. In this study we found proof for this hypothesis, by exposing extracted HDSC to enzyme-containing medium. Furthermore we observed, that enzymatic pre-treatment can remove all secondary cytotoxic products, possibly due to detachment of pendants, which are still coupled to fragments of collagen molecules. The possibility of enzymatic pretreatment of HDSC, to obtain a non-cytotoxic/biocompatible material, may be important for in vivo applications.

A method is described for the production of collagen constructs formed from densely packed, native-banded collagen fibrils. This “Dense Fibrillar Collagen” is formed by concentrating collagen in situ prior to self assembly of the collagen into fibrils. The effect of altering the pH, ionic strength, and osmolarity of the concentrating solution was measured. Increasing the ionic strength and osmolarity of the concentrating solution increased the burst strength of the constructs; increasing the pH from 3.8 to 7.1 reduced the surface fibrillarity, degree of platelet uptake and “short term in vivo thrombogenicity. This construct is being considered as the basis of a small-caliber vascular prosthesis to support guided tissue regeneration.

Synthesis of novel polyurethane copolymers containing covalently attached, well-oriented Arg-Gly-Asp (RGD) peptides was explored. A poly(tetramethylene oxide) based polyurethane was synthesized, and a bimolecular nucleophilic substitution reaction was then employed to incorporate ethyl carboxylate groups onto the polymer backbone. Cell-adhesive peptides Gly-Arg-Gly-Asp-Ser-Tyr (GRGDSY) and Gly-Arg-Gly-Asp-Val-Tyr (GRGDVY) were covalently bound to the polyurethane backbone via the formation of amide bonds. Nuclear magnetic resonance (NMR) and Fourier transform infrared (FTIR) spectroscopies were used to monitor the reactions. The amount of grafted peptide was determined by amino acid analysis. Electron spectroscopy for Chemical Analysis (ESCA) suggested the presence of the grafted peptide at the polymer-air interface in vacuo. Dynamic contact angle analysis showed that, in water, the peptide-grafted polyurethane surfaces were more polar than the underivatized polyurethane. The attachment and spreading of human umbilical vein endothelial cells (HUVECs) on the underivatized and peptide-grafted polyurethanes was investigated. The GRGDSY-and GRGDVY-grafted substrates supported cell adhesion and spreading even without serum in culture medium. The GRGDVYgrafted substrate supported a larger number of adherent cells and a higher extent of cell spreading than the GRGDSY-grafted substrate. These RGD-containing peptide grafted polyurethane copolymers may be useful in providing an easily prepared cell-adhesive substrate for various biomaterial applications.

Cell-biomaterial interactions for different material compositions were evaluated through quantitative examination of in vitro cell spreading behavior and shear-induced cell detachment. Biomaterial substrates consisted of identically prepared, well characterized bulk samples of implant grade cp Ti, Ti-6AI-4V, and Synamel. 3T3 fibroblasts were seeded on the substrate surfaces and incubated for 2 hours. Adherent cell morphology was then examined using scanning electron microscopy and digital image analysis. The cell detachment response was measured using a parallel plate flow chamber. When categorized according to observed cell morphology, projected cell areas were found to be lognormally distributed for all biomaterials and ln(cell area) was significantly greater for Synamel, while no difference existed between cp Ti and Ti-6AI-4V. In addition, the average post-shear fraction of adherent cells was significantly greater for Synamel substrates, as compared to cp Ti and Ti-6AI-4V substrates. The results from this study indicate that in vitro cell adhesion and detachment behavior are measurably influenced by biomaterial substrate composition.

To design novel biomaterials for hepatocyte transplantation it will be necessary to determine whether specific extracellular matrix (ECM) molecule(s) or the adhesive interactions between the surface and hepatocytes are responsible for regulation of hepatocyte function. Purified ECM molecules (laminin, fibronectin, types I and IV collagen) and a synthetic peptide containing the arginine-glycine-aspartate (RGD) cell-binding sequence were precoated at defined densities to non-adhesive polystyrene dishes. Hepatocytes cultured on dishes coated with a low density of ECM molecules (1 ng/cm2) maintained a round morphology, and high liver-specific protein secretion rates. In contrast, culturing hepatocytes on increasing ECM densities (50–1000 ng/cm2) resulted in extensive cell spreading, a loss of liver-specific protein secretion, and cell growth. Hepatocytes cultured on dishes coated with the RGD-containing peptide did not spread even on a high density of the peptide (10,000 ng/cm2), and albumin secretion remained high for hepatocytes cultured on all peptide densities (1–10,000 ng/cm2). These results suggest that a variety of ECM molecules and synthetic peptides are capable of inducing hepatocyte differentiation in vitro, and these effects depend on their ability to promote cell spreading.

The performance of biomaterials for in vivo and in vitro applications can depend critically on tissue cell adhesion and migration. We have been investigating the role that specific reversible interactions between cell adhesion receptors and complementary substratum-bound ligands play in the regulation of cell adhesion and migration. With an axisymmetric radial flow detachment assay (RFDA) [1] we measured cell-substratum adhesive strength for human smooth muscle cells (HSMCs) on surfaces coated with type IV collagen (CIV). We found that the critical shear stress for detachment increased linearly with increasing CIV coating concentration. Using time-lapse videomicroscopy and image analysis we tracked the movement of individual HSMCs over similar CIV-coated surfaces. Cell speed and persistence were determined for variations in CIV coating concentration by applying a persistent random walk model for individual cell movement. Cell speed reached a maximum at an intermediate concentration of CIV, supporting the hypothesis that an optimal cell-substratum adhesiveness exists for HSMC movement. This combination of techniques for measuring adhesion and motility provides a valuable tool to examine the role of cell-biomaterial interactions on cell behavior.

Results of preliminary investigations on the electrokinetic characterization and osteoblast morphology of ion implanted silicon wafers are presented. The goal of this study is to modify the surface of a material via ion implantation to control cellular interactions at the prosthesis-body interface.

Tissue response to any synthetic implant—whether vascular device, dental implant, orthopaedic device, or any other implant that is in contact with soft connective tissue or mineralized tissue—has been found to correlate with the composition, surface chemistry, and surface microgeometry of the implant material.

An enzyme immunoassay has been developed to semi-quantitate the degree of activation of platelets adhering to biomaterials. The method employs a monoclonal mouse anti-GMP-140 antibody. The glycoprotein GMP-140 is contained in alpha granules of resting platelets and is translocated to the outer membrane of platelets during activation via the thrombin-dependent pathway. A polyclonal anti-mouse IgG peroxidase conjugate is then added and platelet activation can be semi-quantitated by use of a suitable substrate, such as OPD.

The results obtained by this method compare well with scanning electron microscopy, and are shown to be less prone to the misinterpretation often associated with other conventional assay systems such as the determination of platelet adhesion by detection of ATP.

The adhesion and activation of platelets by biomaterials has been compared before and after coating these materials with phosphorylcholine (PC) derivatives. PC, in its form as a headgroup in membrane phosphoglyceride (lecithin) and ceramide (sphingomyelin) is a ubiquitous component of cell membranes, and displays extremely low interaction and binding with plasma proteins such as Factor XII [1,2], complement [2], fibrinogen [3,2], immunoglobulins [2] and albumin [2,4].

Using the immunoassay described here activation of platelets is seen to be dramatically reduced after coating surfaces, such as polyethylene, with PC compared with the untreated control. A PCcoated surface should therefore be an ideal starting point for developing tissue-inducing biomaterials. In addition endothelial cells contain GMP-140 in their secretory granules [5]. This would make this assay a potential tool in the investigation of endothelial cell - biomaterial interaction.

Glass surfaces with well defined surface morphologies have been prepared using photolithography to study the effect of surface morphology on cell adhesion and function. Using a transformed recombinant cell-line, AtT-20, as a model of shear sensitive cell, we have shown that cell-substratum adhesion strength is enhanced using a surface with uniform grooves without any loss in cellular function. Furthermore, using primary hepatocytes as a model for a cell whose function is sensitive to its shape, we have shown that surface morphology can modulate cell shape as well as its function.

The effects of nitrogen ion implantation of Ti-6AI-4V alloy on growth of Pseudomonas aeruginosa bacteria on surfaces of the alloy have been investigated. Results for ion implanted samples were compared with controls with similarly smoothly polished surfaces and with controls that had intentionally roughened surfaces. The test consisted of exposing sterile alloy samples to a microbiological broth, to which 24 hour-old cultures of Pseudomonas aeruginosa had been added. After bioassociation at normal temperature 37°C, bacteria adhering to the surface were fixed and treated with a new ruthenium tetroxide staining method, and quantified by use of scanning electron microscopy (SEM), back-scattered electron imaging and EDAX energy dispersive microanalysis. For smooth samples of the alloy, after a 12 hour growth period, the retained bacteria (revealed by the biologically incorporated ruthenium), decreased monotonically with nitrogen dose out to a total fluence of approximately 7 × 1017/cm2 in an affected depth of approximately 0.1500 μm. The SEM confirmed that the Pseudomonas aeruginosa adhered equally to control materials. The ruthenium studies revealed that the amount of bacterial adhesion is indirectly proportional to the nitrogen ion implantation of the titanium. The greater the percentage of nitrogen ion implantation in the titanium alloy, the less bacteria colonized the disk.

Four polyurethanes that were synthesized with different hard segments as well as four commercial polyurethanes were investigated for in-vivo biostability. The four polyurethanes with the varying hard segments were based on a 3/2/1/ mole ratio of methylene diphenylene diisocyanate (MDI) or methylene dicyclohexylene diisocyanate (H12MDI); butanediol (BD) or ethylene diamine (ED); and polytetramethylene oxide (PTMO) (MW=I000). Two commercial polyurethanes were also used: Biomer (Ethicon, Inc.) and Pellethane (Dow).

Films of the polymers were implanted subcutaneously in rats for up to three months to assess their biostability. Polymer films were implanted with either a 100% strain applied or in the unstressed state. Measurement of tensile properties and surface properties before and after implantation was used to assess the stability of each of the polymers. Surface cracking was observed with scanning electron microscopy and the extent and depth of cracking was determined.

It was found that BD chain extended polymers maintained their tensile properties better than the ED chain extended polymers after implantation. H12MDI-based polyurethanes were more susceptible to surface cracking than MDI-based polyurethanes, possibly due to the lack of a crystallizable hard segment.

Human recombinant BMP-2 is known to induce cartilage and bone in a rat subcutaneous site. Here we demonstrate that BMP-2 can induce therapeutic bony union in segmental bone defects in three different animal models: internally fixed defects in sheep femur (2.5 cm) and dog mandible (3.0 cm). In each model, inactive bone matrix from the same species as the test animal was prepared by acid demineralization and extraction in guanidinium chloride. Human recombinant BMP-2 was reconstituted with the bone matrix and inserted in the defect site. Inactive matrix served as the negative control. Healing was monitored weekly by x-ray, At the final time points, specimens were evaluated for union by biomechanical testing and for bone formation by histology. In the first model, a 3 cm mandibular defect in dogs was made and stabilized with a stainless steel reconstruction plate. Animals were terminated at 3 and 6 months. In the second model, an osteoperiosteal defect was created by excising a 2.5 cm midshaft segment from the femur of a skeletally mature sheep. The gap was stabilized by an anterolateral fixation plate. Animals were sacrificed at 12 weeks postoperation. In both studies, BMP-2 mixed with inactive carrier induced bony union judged by radiography, biomechanical testing, and histology; by the same criteria, the defect implanted with the carrier matrix alone did not unite.

The ability of demineralized bone to induce cartilage and bone formation in ectopic sites was described by Urist [1]; this activity was named bone morphogenetic protein (BMP). The in vivo progression of demineralized bone to bone has been well-studied [1,2]. After initial proliferation of mesenchymal cells in the first five days after implantation, a cartilage model forms at days 5–10. The cartilage mineralizes and hypertrophies, and is gradually replaced by bone. The newly formed bone then continues to remodel, so that in rats a thin ossicle of bone surrounding bone marrow has formed by day 21. The therapeutic potential of a material which could induce bone formation and replace bone graft is very high, and demineralized bone or partially purified protein prepared from demineralized bone has been used to repair large bony defects in animal models and in human therapy.

More recently, some of the individual components of the bone morphogenetic activity have been isolated and characterized from bovine bone [3]. Extensive biochemical purification and molecular cloning have demonstrated that there are at least 7 distinct proteins present in bone. Molecular cloning of the bone morphogenetic proteins (BMPs) 2 through 7 demonstrated that they are related and all are members of the transforming growth factor/inhibin family of genes [4,5,6]. There may also be additional factors involved in cartilage and bone induction, as other investigators have described components different from these [7,8]. Expression of the recombinant human BMP-2, BMP-3, BMP-4, BMP-5, and BMP-7 in mammalian cells, followed by purification, allowed the demonstration that an individual BMP is sufficient to induce cartilage and bone in an ectopic site. A detailed study in rats of BMP-2-induced bone formation showed that the progression seen with the recombinant protein is the same as observed with demineralized bone or highly purified bone-derived, bone-inductive proteins [9].

With this background of rhBMP-2-induced ectopic bone formation, this study demonstrates the efficacy of BMP-2 in healing a bony defect. Using allogeneic bone as a matrix, rhBMP-2 shows efficacy in healing large defects in dog mandible and sheep femur.

Basic fibroblast growth factor is a multi-potent cell regulatory factor that stimulates proliferation and angiogenesis. Controlled studies of basic fibroblast growth factor in animals have been hindered by the instability of this protein. In addition, many cells appear to require the continuous addition of basic fibroblast growth factor for optimal growth and function in culture. A system for the sustained delivery of active basic fibroblast growth factor might provide both a means to conduct log-term studies on activity and provide a practical alternative to multiple growth factor additions to cell cultures. Basic fibroblast growth factor was incorporated into standard polymer matrices, but the released growth factor had lost over 99% of its bioactivity. Loss of basic fibroblast growth factor activity was found to result from both physical inactivation and adsorption of the protein to surfaces. These problems were avoided by incorporating the growth factor into calcium cross-linked alginate microspheres containing heparinsepharose beads. Basic fibroblast growth factor was incorporated into these microspheres with 71% efficiency and active growth factor was released with predictable kinetics for up to 7 weeks. Release from these microspheres was controlled by the amount of heparin within the device and could be manipulated by simply altering the heparin content during fabrication. Alginate/heparin-sepharose microspheres were placed into growing cultures of bovine aortic endothelial cells and no cytotoxic effects were observed. Furthermore, microspheres containing growth factor provided long-term stimulation of cell proliferation and maintenance of endothelial cell morphology.

We have developed a lipid-based microcylinder for the controlled release of biological response modifiers. Lipid microcylinders are composed of 1,2-ditricosa-10,12-diynoyl-sn-glycerol-3-phosphocholine which form hollow cylinders 50–250 μm in length with a diameter approximately 0.5–1.0 μm. Amphiphilic molecules such as ganglioside (5–6 mol%) can be incorporated into the bilayers of the microcylinder, modulating surface characteristics of the structure.

In this study, we have examined the biocompatibility of lipid microcylinders and lipid microcylinders containing 6 mol% ganglioside. The interaction of microcylinders with peripheral blood monocytes and three cell lines: U937, a histocytic monocyte; K562, an erythroblast, and; a Jurkat derivative, a lymphoblast was assessed. Toxicity, as measured by proliferative status of the cell lines, was not evident at lipid concentrations up to 100 μg/ml lipid. Peripheral blood monocytes were observed to closely associate, but not engulf lipid microcylinders. However, when ganglioside was present in the lipid microcylinders this interaction was markedly decreased.

We have begun to measure the release rates of growth factors such as transforming growth factor-beta (TGF-β) from lipid microcylinders. Release of TGF-β from the microcylinders follows first-order kinetics. The rate of release can be modulated by increasing the temperature which results in a thermotrophic phase transition of the lipid at 43 °C.

Fibroblast growth factor (FGF) is a polypeptide found in two forms: basic and acidic. The basic form is produced by many more types of cells than the acidic form, although both bind to the same receptor. These proteins act on a variety of mesodermally and ectodermally derived cells, including chondrocytes, glial cells, myoblasts, endothelial cells, cornea and lens epithelia, adrenal cortical cells, ovarian granulosa cells, periosteal fibroblasts, and osteoblasts. Basic FGF was chosen for the present study for a variety of reasons. First, it has significant cross-species homology, with 98.7% correlation between human and both bovine and avian FGF. Less conservation has been observed in the acidic form. In addition, the basic form has been shown to be 30- to 100-fold more potent, depending on the target tissue.

Biocompatible electrodes were fabricated using ordinary semiconductor device processing techniques. For the purpose of identifying electrically stable electrode materials, which are biocompatible for long life, several kinds of contact metal were used. From a comparison of the metal elemental constants, it is suggested that the metal atom ionization energy, cohesive energy and the metal material conductivity may be the major factors to decide the material biocompatibility. Electrical testing shows Pd, with high atom ionization energy and low electrical conductivity, to have the best long life stability. In a test of 140 hours in saline solution with electrical power, the average maximum resistance deviation for a Pd electrode was 5%. The metal Ru, with high atom cohesive energy, has the best ability to withstand surface erosion for biocompatibility requirements. In a microscopic examination, the surface erosion for the Ru layer was invisible after 140 hours testing. The double-layer structures Pd/Ru and Mo/Ru were also tested. The maximum resistance deviations were found to be 7% and 9%, respectively.