Odontoblast differentiation during tooth development requires epigenetic signals provided by the inner dental epithelium through basement membrane-mediated interactions. The basement membrane may act as a solid substratum, able to interact with plasma-membrane receptors of ectomesenchymal cells of the dental papilla, and/or as a potential reservoir of paracrine and autocrine morphogenetic factors regulating cell kinetics and providing the necessary inductive signals for primary dentinogenesis. Several growth factors have been associated with the onset of odontoblast differentiation: acidic and basic FGF, IGF-I and -II, neural growth factor, PDGF and molecules belonging to the TGF-? superfamily, which includes BMP 2, 4, 6 and 7. The requirement of basement membrane for odontoblast differentiation in isolated dental papillae grown on semisolid agar has recently been superseded by substitution with matrix molecules and growth factors: TGF-?₁ or BMP-2 combined with heparin or fibronectin and the EDTA-soluble fraction of dentine induced cytological and functional differentiation of odontoblasts, while IGF-I combined with heparin induced only cytological differentiation without matrix secretion.

Dentinogenesis can be initiated in the absence of dental epithelium and basement membrane as an intrinsic potential of mature dental pulp. This phenomenon, occurring stereotypically during wound healing and/or in response to external low-grade irritations in an appropriate bacteria-free pulp environment, is referred to as reparative dentinogenesis. Definitive data on the molecular mechanisms of reparative dentinogenesis and cytodifferentiation of odontoblast-like cells are not available. Several exogenous influences induce reparative dentinogenesis in vivo when applied as dressings at sites of exposed peripheral pulp or as implants placed at central pulp sites: the dentine matrix or dentine matrix components, purified of recombinant BMP-2, or BMP-4, or osteogenic protein-1 and Millipore filters soaked in a solution containing a high concentration of plasma fibronectin or an EDTA-soluble dentine fraction. Further, in primary cultures of dental pulp cells, growth factors (aFGF, bFGF, PDGF and TGF-?) exerted mitotic and matrix-formative effects. Exogenous growth factors such as aFGF, bFGF, PDGF, IGFs and TGF-?s, mediating the molecular mechanisms during wound healing, are now thought to play an important role in several tissues. Any influence of exogenous growth factors on dental pulp cells in vivo in regulating their proliferation, biosynthesis and differentiation has not been elucidated. Dentinogenic effects were reported recently after long-term intrapulpal implantation of TGF-?₁ purified from human platelets. We have now sought to investigate whether the short-term exposure of dental ectomesenchymal cells in vivo to filters containing recombinant bFGF, IGF-II or TGF-?₁ could induce the expression of an odontoblast-like phenotype and reparative dentine formation.

Small pieces not larger than approx. 1×2 mm, were cut from Millipore filters (150 ?m thick, 0.45 ?m pore size) and soaked in groups of eight for 24 h at room temperature in 2 ml of control or growth factor-containing solution (see below). The amounts of growth factor adsorbed to the filters were assessed quantitatively: adsorbed material was extracted from individual filters with 1 ml 0.68 M EDTA, pH 7.2, for 1 h and measured by enzyme-linked immunosorbent assay using antibodies against recombinant human bFGF (Amersham Life Sciences, USA), IGF-II (R&D Systems, Minneapolis, U.S.A.) or TGF-? (Amersham Life Sciences). High levels of binding of growth factor to the filters were confirmed in a second experiment in which both the total amount of growth factor adsorbed to the eight filters and that remaining in the solution were measured. Siliconized glass or plastic was used throughout the procedures.

The following implants were then prepared.

(i) Controls: 12 pieces of filter soaked in PBS with 0.1% dog serum albumin (Sigma, St Louis, U.S.A.).

(ii) 16 filters soaked in 100 (n=8) or 500 (n=8) ng human recombinant bFGF expressed in Escherichia coli (Sigma) per ml PBS/0.1% albumin control solution. Prior measurement of adsorption to the filters indicated that approx. 94% of the bFGF in the 100 ng/ml solution and 60% of the bFGF in the 500 ng/ml solutions had bound to the filters. The mean amounts of bFGF bound per filter (2 mm² surface area) were 24±5 (SD) ng and 75±8 ng for the 100 and 500 ng/ml solutions, respectively.

(iii) 16 filters soaked in 100 (n=8) or 500 (n=8) ng human recombinant IGF-II expressed in E. coli (Sigma) per ml PBS/0.1% albumin. Prior measurement of adsorption to the filters indicated that approx. 72% of the IGF-II in the 100 ng/ml solution and 50% of that in the 500 ng/ml solutions had bound to the filters. The mean amounts of IGF-II bound per filter (2 mm² surface area) were 18±3 ng and 64±6 ng for the 100 and 500 ng/ml solutions, respectively.

(iv) 16 filters soaked in 100 ng human recombinant TGF-?₁ expressed in E. coli (Life Technologies GmbH, Eggenstein, Germany) per ml PBS/0.1% albumin. Prior measurement of adsorption to the filters indicated that approx. 95% of the TGF-?₁ had bound to the filters. The mean amount of TGF-?₁ bound per filter (2 mm² surface area) was 24±4 ng.

Five healthy dogs 12–16 months of age were used. The animals were anaesthetized with pentothal (20 mg/kg body wt) and intubated with a cuffed endotracheal tube before the beginning of all experimental procedures.

The pulps of permanent upper and lower first molars and canines were exposed via buccal class V cavities cut with a No. 31 inverted-cone carbide bur in an air turbine with sterile saline spray. The cavities were washed with sterile saline and dried with cotton pellets. Implants were forced into central pulp sites of the exposed teeth. Two implants were placed in each molar by making two separate exposures. Light pressure was applied to the exposure site to control haemorrhage. A calcium hydroxide-containing cement (Life; Kerr Europe, Basel, Switzerland) was applied in contact with the amputated pulp and the cavities were further restored with zinc oxide–eugenol cement. Half of the specimens in each group were extracted after 1 week and processed for transmission electron microscopy. The remaining specimens were extracted after 3 weeks and processed for light microscopy.

These specimens were carefully extracted, cut labiolingually immediately after extraction and fixed for 10 min in 3% glutaraldehyde buffered with 0.1 M cacodylate, pH 7.3. The implant with the surrounding pulp tissue was then isolated and cut into six small samples, which were placed in the same fixative for 2.5 h, rinsed in 0.1 M cacodylate buffer, postfixed in 1% osmium tetroxide in the same buffer and embedded in Epon 812. At least 20 serial semithin sections were cut from each sample, stained with toluidine blue and examined by light microscopy for the cell response in close proximity to the filter. The occurrence of at least one group of cells containing more than five adjacent elongated cells with basally located nuclei in at least six serial sections was considered as cell polarization. Further ultrathin sections were cut and stained with uranyl acetate and lead citrate for examination in a Jeol 2000 FX II transmission electron microscope.

Teeth from the 3-week observation period were carefully extracted and fixed for 5 days in a buffered formalin solution, pH 7.2, after resection of their roots. They were then demineralized in 5% trichloroacetic acid for 8 weeks. Serial paraffin sections, 7 ?m thick, were cut and stained with Mayer’s haematoxylin–eosin, silver stain and Gomori trichrome. The occurrence of extracellular matrix in close proximity to the filter in at least six serial sections was considered as new matrix deposition. The pulpal responses at circumferential dentine and exposure sites were also evaluated.