Abstract
Odontoblasts produce most of the extracellular matrix (ECM) components found in dentin and implicated in dentin mineralization. Major differences in the pulp ECM explain why pulp is normally a non-mineralized tissue. In vitro or in vivo, some dentin ECM molecules act as crystal nucleators and contribute to crystal growth, whereas others are mineralization inhibitors. After treatment of caries lesions of moderate progression, odontoblasts and cells from the sub-odontoblastic Höhl’s layer are implicated in the formation of reactionary dentin. Healing of deeper lesions in contact with the pulp results in the formation of reparative dentin by pulp cells. The response to direct pulp-capping with materials such as calcium hydroxide is the formation of a dentinal bridge, resulting from the recruitment and proliferation of undifferentiated cells, which may be either stem cells or dedifferentiated and transdifferentiated mature cells. Once differentiated, the cells synthesize a matrix that undergoes mineralization. Animal models have been used to test the capacity of potentially bioactive molecules to promote pulp repair following their implantation into the pulp. ECM molecules induce either the formation of dentinal bridges or large areas of mineralization in the coronal pulp. They may also stimulate the total closure of the pulp in the root canal. In conclusion, some molecules found in dentin extracellular matrix may have potential in dental therapy as bioactive agents for pulp repair or tissue engineering.
Introduction
Dentinogenesis has been studied extensively as investigators attempt to better understand the formation and mineralization of this collagenous connective tissue, which resembles bone but has its own distinct features. A unique feature of dentin is that it is a mineralized tissue which surrounds the pulp, an unmineralized tissue, and gradually increases in thickness to the detriment of the space previously occupied by the pulp. The dental pulp not only functions to provide nutritional and sensory properties to dentin, but also has its own reparative capacity. This potential has important implications for dental therapy.
The areas of odontoblast biology and dentinogenesis, pulp biology, and the formation of reactionary and reparative dentin have been extensively reviewed during the last two decades (Lesot et al., 1993; Linde and Goldberg, 1993; Ruch et al., 1995; Smith and Lesot, 2001). Together, these and many other excellent articles constitute the milestones of our present knowledge. However, despite this apparent wealth of information, many assumptions and questionable hypotheses still prevail. Clarification of many of these points is still required to underpin future progress in our understanding of the properties of pulp and dentin. For example, considerable data have been obtained from studies on the early stages of dentin formation in embryonic tooth germs and simply extrapolated to the situation in mature teeth. While there may be common features, the tissue architecture and cellular environments in tooth germs and mature teeth are quite distinct. Following injury to the mature tooth, pulp progenitor cells may be recruited during the repair processes and differentiate into second-generation odontoblasts, neo-odontoblasts, or odontoblast-like cells. This plethora of names highlights our lack of understanding of the origin, nature, and fate of these replacement cells. Do they arise from undifferentiated/stem cells (Mann et al., 1996; Gronthos et al., 2000) or from some other derivation? Answers to these and many other questions may be key to our understanding of the mechanisms responsible for repair in the tooth.
In this review, we will consider separately the biology of dentin and pulp as a background to discussing the problems in pulp healing.
(I) Odontoblast Biology
(I-1) Derivation and differentiation of odontoblasts
Odontoblasts are derived from the dorsal cranial neural crest (CNC). Early-migrating cranial midbrain crest-derived cells populate the dental mesenchyme of mandibular molar teeth in the first branchial arch (Imai et al., 1996). Forebrain and midbrain CNC-derived cells migrate to the fronto-nasal mass (Osumi-Yamashita et al., 1994). Non-CNC-derived mesenchymal cells are also present in the arch, and consequently in the regions where teeth develop. At the late bud stage, the condensed dental mesenchyme consists of CNC-derived cells together with an increasing number of non-CNC-derived cells. CNC-derived pre-odontoblasts are aligned at the periphery of the dental papilla and adjacent to the inner enamel epithelium (Chai et al., 2000). Consequently, many of the pulp cells may be non-CNC-derived cells.
Characterization of the signaling processes responsible for the induction of odontoblast differentiation is fundamental to the understanding of both physiological dentinogenesis as well as tissue repair. The key roles played by growth factors in physiological odontoblast differentiation, and the recapitulation of such events during repair, have recently been reviewed (Smith and Lesot, 2001). Considerable focus has been placed on those growth factors, particularly of the Transforming Growth Factor-beta (TGF-β) family, which may be directly involved in signaling cytodifferentiation of odontoblasts and odontoblast-like cells. Expression of growth factors by the inner enamel epithelium leads to their sequestration within the dental basement membrane for presentation to the pre-odontoblasts (Ruch et al., 1995). While proof of which growth factors are responsible for signaling of odontoblast differentiation in vivo is lacking, Transforming Growth Factor β1 (TGF-β1), TGF-β3, Bone Morphogenetic Protein-2 (BMP-2), and Insulin-like Growth factor-1 (IGF-1) (Bègue-Kirn et al., 1992 , 1994) appear capable of signaling odontoblast differentiation in vitro.
Expression of growth factors by odontoblasts after differentiation may lead to their sequestration within the dentin matrix, where they may be released following injury to the tooth (reviewed by Tziafas et al., 2000; Smith and Lesot, 2001). Such a pool of growth factors may be able to signal reparative processes leading to odontoblast-like cell differentiation. Other growth factors may be expressed by odontoblasts, but not sequestrated within the dentin matrix. For instance, TGF-β3 is expressed more strongly than TGF-β1 (Sloan et al., 2000a) by human odontoblasts, but is not detected in the dentin matrix. TGF-β1, but not TGF-β3, has a high affinity for the proteoglycans decorin and biglycan within the dentin matrix, thereby facilitating its sequestration in this matrix. Thus, growth factors expressed by odontoblasts may play a variety of roles in tissue homeostasis, via either an autocrine or a paracrine action. Several angiogenic growth factors, presumably secreted by the odontoblasts (Roberts-Clark and Smith, 2000), may be important in the vasculature of the dentin-pulp complex, both physiologically and during tissue repair. Caution will be required in studies of the expression of growth factors in odontoblasts, since developmentally related temporal expression patterns may exist. Thus, the expression patterns in newly differentiated odontoblasts when primary dentinogenesis is very active may not necessarily reflect patterns seen later in the life of the tooth.
A striking feature of the odontoblast is its post-mitotic nature. During tooth formation in the mouse, cells of the odontoblastic lineage migrate from the central area of the dental papillae toward the periphery between days 14 and 18 of embryogenesis. Mouse pre-odontoblasts multiply through 14 to 15 cycles, with 10 hrs for each cell cycle (Ruch, 1987), although the number of cycles and their length are often ignored in humans. What is clear is that terminal differentiation requires a minimum number of cycles for cells to achieve competence to respond to inductive signaling (Ruch et al., 1976 , 1982).
Following the last mitosis and prior to odontoblast differentiation, the pre-odontoblasts align perpendicular to the basement membrane (BM), and only the daughter cell adjacent to the BM undergoes terminal differentiation into an odontoblast. This highlights the importance of the spatial configuration in signal presentation during odontoblast differentiation and the need for this feature to be considered during any mimetic approaches to pulp repair. It is assumed that the other daughter cell, located away from the BM, becomes incorporated within the Höhl layer (Höhl, 1896), although direct evidence is lacking (Fig. 1).
(I-2) Ultrastructure and functional activities of odontoblasts and Höhl cells
(I-2-1) Polarizing and young secretory odontoblasts
Young polarizing odontoblasts are small, ovoid cells with a high nucleus/cytoplasmic ratio, rudimentary rough endoplasmic reticulum (RER), and a poorly developed Golgi apparatus. Essentially, they are cells undergoing re-organization as they differentiate in preparation for their secretory function. With increasing differentiation, gap-type junctions increase in number and size between odontoblast cell bodies at locations where the distal junctional complex is actively developing. The distal junctions form a physical barrier between the predentin compartment and the odontoblast cell body.
Secretory odontoblasts are aligned along the periphery of the pulp on the formative surface of dentin, and both the precise chronology of tooth formation and the ultrastructural observations of the odontoblast through its life-cycle suggest some degree of pre-programming in the activity of these cells. Decreasing volume of the pulp chamber during secondary dentinogenesis has been reported to be associated with the programmed cell death of odontoblasts. The nature of such programming of odontoblast activity or survival is unclear; however, its elucidation could provide a powerful tool for the regulation of dentinogenesis during repair. During tooth injury leading to odontoblast death, the cells of the Höhl layer can re-express some transcription factors such as c-fos and differentiate into new odontoblasts for reparative dentinogenesis (Kitamura et al., 1999; Mitsiadis et al., 1999).
Functionally, the secretory odontoblast can be considered to consist of two distinct parts: the cell body involved in the synthesis and control of cellular and extracellular proteins, and the process whereby secretion and limited re-internalization occur. The odontoblast process consists of one main trunk with numerous lateral branches along its length. The process is limited by a plasma membrane and contains predominantly cytoskeletal components. Actin microfilaments interconnect microtubules and are denser in the sub-plasmalemmal undercoat. Intermediary filaments of the vimentin type are also present. Nestin, another intermediary filament family member, is mainly found in neural-crest-derived cells, including odontoblasts. In the main trunk, immunolabeling was positive for nestin, whereas small lateral branches were unstained. Both coated and uncoated small vesicles can be seen in the processes, and clathrin at the surface of membrane-coated vesicles has been assumed to play a role in the internalization and backward transport of degraded or fragmented molecules from the ECM. Transport from the plasma membrane to the Golgi would allow for turnover of ECM molecules and provide some retro-control through stimulation or inhibition of ECM synthesis.
Lateral branching may contribute to cell-cell communication and cell-matrix communication and may possibly be involved in peritubular dentin formation. Odontoblasts, like many other cells, may use intermediary filaments to provide spatial information. In mature dentin, there are minute openings of canaliculi into the lumen of the tubules. These canaliculi contain the lateral branches and constitute a tiny network that interconnects the tubules.
Compositional differences between predentin and the mineralized dentin matrix led to the hypothesis that there may be two levels of secretion from the odontoblast (Linde, 1984). Secretion of collagen and proteoglycans at the proximal level would give rise to predentin, while secretion of other non-collagenous components, including dentin phosphoprotein, at a distal level has been suggested to occur close to the mineralization front.
(I-2-2) Aged odontoblasts and apoptosis
Little is known about the ultrastructure of old odontoblasts, and much of our knowledge is based on the rat incisor, which represents a rather special model for aging in view of its continuous growth (Takuma and Nagai, 1971). While the comparison among polarizing, pre-secretory, young, and older secretory odontoblasts is of course easier in a single tooth, it is unclear whether such observations are relevant to teeth of limited growth, as in humans. A reduction in length and cytoplasmic organelles with aging would be in accord with a decreased synthetic and secretory capacity of these cells. An increase in number and size of lysosomes and phagosomes would be expected to be associated with gradual cellular degeneration and self-destruction. Whether this degenerative pattern is common to all teeth or is merely a reflection of the continuous growth of the rat incisor is unclear.
What
The apparent longevity of odontoblasts and their capacity for fluctuations in secretory activity, however, raise an interesting question. Are all odontoblasts the same, or are there differences in the programmed life of the cells? Odontoblasts are post-mitotic cells, but their numbers decline with age, and the reported apoptosis (Franquin et al., 1998) implies differences in the life span of these cells. It is unclear if this is a programmed event or whether it is the result of environmental influences, such as crowding during continued secondary dentin deposition. Feedback from lysosomal accumulation of fragments of degraded molecules and/or specific gene expression might also possibly contribute to odontoblast survival. In view of the developmental events leading to odontoblast differentiation, it is hard to envisage programmed differences in cell life span. Despite differences in survival time between odontoblasts, their longevity compared with that of many cells of the body is remarkable and suggests a strong influence of expression of cell survival factors.
(1–3) The extracellular matrix of predentin and dentin
Varieties of molecules have been identified in predentin/dentin, and more will almost certainly be reported. They have been extensively reviewed by many authors (for review, see Butler et al., 2003), and we shall focus here only on the comparison between dentin and pulp matrix components, in relation to their properties to promote mineralization and repair of the tissues. The Table summarizes our present knowledge.
The non-collagenous proteins may be subdivided into molecules predominantly found in mineralized tissues (bone and dentin), and more ubiquitous molecules. For many years, some proteins, such as dentin phosphoprotein (DPP) and dentin sialoprotein (DSP), were believed to be present exclusively in dentin (MacDougall et al., 1997) It is now established that bone cells also synthesize DSP, and probably DPP, but in a ratio that is estimated to be about 1:400 of that in dentin (Qin et al., 2002). Dentin matrix protein-1 (DMP-1) is expressed by both bone and dentin (George et al., 1993). The gene encoding this protein is found near the locus of the DSPP gene. This is also the locus where the dentin genetic defect dentinogenesis imperfecta type II has been identified. In vitro studies have shown that these phosphorylated proteins can be nucleators, and they have been strongly implicated in crystal growth during dentinogenesis, although the nucleating characteristics of these molecules in vivo remain to be established. In addition to a role in dentin mineralization, these molecules exhibit some signaling function, adhesive properties through their RGD sequence, and may possibly be protective against complement-mediated cell toxicity. Recently, 4 molecules have been grouped as a family of
DPP and DSP have long been considered as unique phenotypic markers of dentin and secretory odontoblasts. Since bone cells also expresses DSPP, DSP, and DPP, these can hardly be considered as dentin-specific molecules, as is also the case for DMP-1. Pre-secretory ameloblasts also express these molecules at some very early stages of odontogenesis. Thus, the list of the various molecules presented in the Table and the lack of absolute specificity for any of them suggest that, for the identification of cells presumed to be of the odontoblastic lineage, a panel of various markers must be considered. The combination of expression of DPP, DSP, DMP-1, and nestin may be valuable. However, we still have very limited understanding of how the expression of these various components is modulated under different conditions.
Some molecules that are not phosphorylated have been identified in dentin. The γ-carboxyglutamic-containing molecule osteocalcin, present in odontoblasts, was long believed to be associated with mineralization. However, increased mineralization was observed in bones of mice showing deletion of this gene (Dulcy et al., 1996), thus requiring revision of the role of this protein.
The presence of amelogenin in dentin was first attributed to a degradation process. It was hypothesized, and partially demonstrated immunohistochemically, that amelogenin migrates throughout dentin and is further incorporated (internalization) into odontoblast processes and pulp cells (Nakamura et al., 1994). Enamelysin (MMP20), which acts on amelogenin as substrate, has been detected within odontoblasts and lends support to the hypothesis of active degradation of amelogenin following internalization. However, it was recently demonstrated that alternatively spliced amelogenins, expressed either by all exons (A+4) or by all exons except exon 4 (A-4), are also synthesized by odontoblasts and may be isolated from bovine and rat dentin. In preliminary studies, A+4 and -4 were identified as a differentiation factor and a chondrocyte-inducing agent (CIA). They interact with bone and cartilage transcription factors (Nebgen et al., 1999; Veis et al., 2000).
Other growth factors have been identified in dentin, including TGF-β, IGF-1 and -2, FGF-2 (Finkelman et al., 1990; Cassidy et al., 1997; Zhao et al., 2000), and various angiogenic growth factors (Roberts-Clark and Smith, 2000). The state of activation of some of these growth factors is unclear, although much of the TGF-β appears to be in active form and to be associated with various ECM components of dentin (Smith et al., 1998). This is corroborated by the biological effects of dentin matrix preparations containing TGF-β1 on odontoblast differentiation and the inhibition of these effects by neutralizing antibodies (Bègue-Kirn et al., 1992). However, in mice overexpressing TGF-β1, reduced expression of DSPP was detected and led to defective hypomineralized dentin formation (Thyagarajan et al., 2001). Interaction with ECM components of the dentin matrix may confer protective effects on growth factor activity, since activated TGF-β1 at least has a relatively short half-life. The presence of these growth factors in dentin emphasizes the biologically active nature of the matrix and the need to move away from regarding it as a relatively inert material.
The diverse range of molecules found in dentin has challenged researchers for several decades and still poses numerous questions in terms of their functional relationships. Our understanding of their precise tissue locations is still limited, and yet this is fundamental to an appreciation of their functions. The complex interactions between and among many of these molecules may be critical to their function in tissue development and homeostasis, and our appreciation of such interactions is only starting to unfold. Nevertheless, it is clear that several of these molecules may have potent biological effects in the reparative situation, and that the matrix interactions occurring may be important regulators of reparative processes.
(I-4) Dentin
Dentin is a complex heterogeneous structure, a manifestation of the cellular processes that take place during its formation. Thus, mantle dentin reflects secretory events that occur as the odontoblasts complete their differentiation. The main bulk of circumpulpal dentin is comprised of inter- and peri-tubular dentin matrices permeated by dentinal tubules containing the odontoblast processes as these cells retreat in a pulpal direction. Estimates of tubular density have varied from 18,000 to 83,000 per square millimeter, but a mean of 21,000 in deeper areas and 18,800 in the middle areas of dentin appears more realistic (Schilke et al., 2000).
During phylogenetic development, osteodentin, which is found in ancestral species, disappears, and orthodentin becomes predominant. Matrix of osteodentin morphology is still detected sometimes in pathological dentin, e.g., during genetic diseases, drug-induced diseases, or in response to caries. The persistent presence of odontoblasts on the formative surface of orthodentin has implications for their capacity to be up-regulated and participate in repair by secreting tertiary dentin (Smith et al., 1995), thereby allowing for the deposition of dentin at the pulp/dentin interface. In contrast, osteocyte-like cells become entrapped within the osteodentin matrix, thus limiting their capacity to contribute to reparative processes.
(II) Pulp Biology
(II-1) The pulp during crown and root formation
Most tooth germs develop within an environment derived from the first branchial arch with the exception of the maxillary incisors, which originate from the fronto-nasal bud. The future pulp contains a mixture of resident mesenchymal cells, already present at the site where the dental lamina and tooth buds develop, together with para-axial mesenchymal-derived cells and cells migrating from the cranial neural crest (CNC).
It is generally assumed that the derivation of the different pulp cell populations in the crown and root of the tooth is similar, but evidence to support this view is lacking. The epithelial-mesenchymal interactions responsible for tooth development in the crown and root and culminating in odontoblast differentiation show many parallels, although differences do exist in the structure of the epithelial tissues. For example, ameloblastin (also named amelin or sheathlin) and enamelin, but not amelogenin, are expressed in the Hertwig root sheath (Bosshardt and Nanci, 1997). This is in contrast to the situation in the enamel organ during crown development, where amelogenin contributes 90% of the secreted matrix.
The morphogenesis of the vasculature is quite different in the pulp of the crown and root. In the crown, blood vessels that penetrate the apical area of the pulp infiltrate and branch around the odontoblast layer. As seen in corrosion resin casts of dog pulp, the sub- and odontoblastic layers are fed by capillaries forming successive small glomerular well-individualized structures that feed limited areas about 100–150 μm wide. In contrast, in the root, resin casts indicate a somewhat continuous network of sub-odontoblastic capillaries (Takahashi, 1985). In the root, arteries and veins are also located centrally, although the vessels in the two parts of the pulp appear quite distinct as a result of their formation at different stages of tooth development.
Crown and root differences in nerve development also exist, although information on events in the root is limited. The number of nerve fibers does not appear to be stable in the tooth, perhaps reflecting the changing environment to which it is subjected. This implies a capacity for neural regeneration within the pulp and offers exciting opportunities for exploitation in any approaches to the regeneration of pulp tissue.
Differences in development of the crown and root of the tooth have received only limited attention and yet may hold valuable clues to differences in tissue structure and behavior in these two regions of the tooth. In the absence of more definitive information, caution is required in assumptions of similarities between the crown and root of the tooth.
(II-2) The pulp in the mature adult tooth
(II-2-1) Pulp cells
Once tooth formation is complete, the pulp is totally surrounded by a mineralized environment, and because of the continuous slow secretion of physiological secondary dentin, the space occupied by the pulp is gradually reduced. There is still, however, communication via the apical foramen between the pulp and the periapical tissues.
The adult pulp contains cells that are responsible for the formation and turnover of a complex non-mineralized ECM. Most of the cells are fibroblast-like cells (Fig. 1). Macrophages, nerves, and capillaries may also be observed. Division of these cells within the crown of the adult pulp appears to be limited, although cell renewal following apoptosis probably occurs. While, superficially, all the fibroblasts appear morphologically similar, variations in their proliferative activity suggest that they represent a heterogeneous cell population (Moule et al., 1995). Such heterogeneity is further suggested from studies in human deciduous teeth where 183 fibroblast strains were isolated, with only 6 of these strains being capable of forming mineralization nodules (Tsukamoto et al., 1992). These observations indicate that isolation of primary cultures of pulp cells may give rise to considerable selection of cell phenotype, which in turn could influence the interpretation of data from these cells.
Pulp fibroblasts are elongated, with a large nucleus and well-developed RER. The Golgi apparatus located near the nucleus and the presence of secretory vesicles reflect the synthetic capacity of these cells. Lysosomes are also present, although it is not known if the same fibroblasts are involved simultaneously in the synthesis and catabolism of ECM components. Nevertheless, there appears to be considerable turnover of ECM components, particularly in the younger pulp. Pulp fibroblasts do not exist in isolation, but are connected by desmosome-like and gap junctions, which facilitate intercellular communication. This also implies that if the cells move or translocate, this communication may be broken. This motility is perhaps facilitated by the presence of α-smooth-muscle actin, which contributes to contraction of the cells and/or the extracellular matrix (Alliot-Licht et al., 2001).
Defense cells have also been identified in the healthy pulp, including mostly dendritic cells and histiocytes/macrophages. A few lymphocytes and mast cells are also present. Most, if not all, are Class II molecule-expressing cells, implicated in immune responses. Some years ago, Jontell et al.(1987) reported the presence of immunocompetent cells in normal pulp. Cells expressing the major histocompatibility complex (MHC) class II molecules and with characteristic dendritic morphology have been identified in the pulp (Jontell et al., 1998). B-lymphocytes are rarely encountered in the normal dental pulp, but macrophages are present, probably for phagocytic activity. These various cells are especially numerous in the peripheral areas of the pulp, where they may participate in the immunosurveillance of this tissue and contribute to the pulp responses to dentinal caries (Jontell et al., 1998; Okiji et al., 2002). Cooperation between and among the various cell populations in the pulp may be an essential feature of normal tissue homeostasis, including pulp cell dynamics, and such interactions must be elucidated if we are to understand the physiological behavior of pulp and its response to injury. While the mediation of such interactions is poorly understood at present, interesting insights have arisen from null mutation studies for the growth factor TGF-β1 (D’Souza et al., 1998). This growth factor, which is physiologically expressed by pulp cells together with its other isoforms (Sloan et al., 2000a,b), is a potent immunosuppressant and may play a key role in the regulation of inflammatory cell activity in the pulp.
(II-2-2) Pulp extracellular matrix
Pulp fibroblasts produce a complex ECM, which is substantially different from that of dentin and other soft connective tissues (see Table). In the pulp, large intercellular spaces contain type I and type III collagen fibrils (56 and 41%, respectively), in contrast to the predominantly type I collagen of dentin. Types V (2%) and VI collagen (0.5%) are also found but in lesser amounts. Microfibrils 10–14 nm in diameter containing fibrillin are associated with microfibrils of type VI collagen. Differences in the non-collagenous ECM components also exist between pulp and dentin. Fibronectin, of both pulp and serum origin, is prevalent in pulp. Chondroitin 4- and 6-sulfate (60%), dermatan sulfate (34%), and keratan sulfate (2%) glycosaminoglycans (GAGs) are associated with protein cores and, consequently, are found as proteoglycans (PGs) in the pulp. Among PGs, decorin, biglycan, and versican are present. This pattern of PGs in pulp differs substantially from that in dentin, where CS-4 accounts for 81%, CS-6 for 14%, and CS/DS for 2%. Together with hyaluronic acid, which accounts for about 2% of the GAGs found in the pulp, pulp PGs contribute significantly to the viscosity of the intercellular matrix of the pulp. Major differences also exist in the lipids found in pulp and dentin (Goldberg and Septier, 2002).
While the derivation of the pulp cells may be key to our understanding of their secretory activity, it is probable that control of their behavior by growth factors and cytokines will also be important. Bone Morphogenetic Proteins (BMPs), their transcripts, and the types IA and II receptors are expressed by pulp cells (Takeda et al., 1994a,b; Gu et al., 1996). Moreover, pulp cells express many members of the TGF-β superfamily, types I and II receptors for TGF-βs, activins, and BMPs (Toyono et al., 1997; Sloan et al., 1999 , 2000a). The effects of the TGF-β family members on pulp cell behavior are poorly understood, but could prove a fruitful area of study, especially with respect to pulp mineralization.
The lack of mineralization in pulp might be explained by the absence of specific molecules which have been identified in dentin. Immunostaining and in situ hybridization suggest that DSP, DPP, DMP-1, and osteocalcin are predominantly expressed by odontoblasts. Osteonectin, found widely in the dental papilla of tooth germs, is found in the odontoblasts alone, and not in the pulp in adult teeth. Osteopontin and BSP are present in both dentin and pulp (Yokota et al., 1992). These substantial differences in composition between dentin and pulp could partly explain why pulp is unable to mineralize under physiological conditions. The presence of mineralization inhibitors (e.g., the high concentration of PGs) in pulp may also contribute to the absence of mineralization. Thus, the regulation rather than the initiation of mineralization may be the more crucial question to be addressed, i.e., what prevents mineralization in pulp and other soft connective tissues rather than what initiates it in dentin.
(II-3) Renewal of cell populations in normal and experimental conditions
Little is known of the life span of pulp fibroblasts. A decline in the total number of cells has been reported in both the human and rat pulp with aging (Lavelle, 1968; Bernick, 1987). Only limited inferences can be drawn from these studies, since no attempt was made to distinguish among different pulp cell populations. Significant decreases in numbers of sub-odontoblast cells with aging have been observed in the human and rat pulp (Murray et al., 2002a,b). Interestingly, with increasing age, while the fibroblast density decreased in the human pulp, small but significant increases in fibroblast density were seen in the rat. While these observations relate to numbers rather than turnover of pulp cells, analysis of the data suggests that perhaps fibroblast density is less-well-regulated than that of other cells of the pulp in the rat, and also than fibroblasts of the human pulp.
Significant increases in cell numbers were seen in the central pulp core and within the peripheral sub-odontoblast layer after administration of a fatty-acid-deficient diet (Vermelin et al., 1995). These workers hypothesized that killer cells are normally responsible for destruction of excess pulp cells to maintain a balance in cell numbers, and that their action was impaired by this diet, thereby disrupting the normal dynamics of pulp cell renewal. Such increases in cell density under experimental conditions are in contrast to the low rate of cell division normally seen in the coronal pulp. This may indicate that the rate of cell division can fluctuate with the experimental conditions or that existing approaches to the study of pulp cell division have not been capable of identifying all dividing cells. [3H] thymidine studies of pulp cell division beneath exposed cavities restored with calcium hydroxide to investigate odontoblast-like cell differentiation suggested that dividing cells move from the central to peripheral sites of the pulp (Fitzgerald et al., 1990). The existence of a true self-renewing pulp stem cell population (Gronthos et al., 2000 , 2002; Young et al., 2002) has important implications for odontoblast-like cell differentiation during repair and is discussed further below. Our knowledge of pulp cell dynamics is still limited, and yet the renewal and potentiality of different cell populations in the pulp are key to understanding and exploiting the behavior of this tissue.
Cell death among the different cell populations in the pulp is still poorly understood. Using the TUNEL procedure to detect apoptosis in rat molars, investigators have detected a few positive cells in the outer areas of pulp (Vermelin et al., 1996). Because of the false-positive results that may be seen with the TUNEL method, other approaches have also been adopted. Immunostaining of healthy human pre-molars for transglutaminase revealed restriction of positive cells, again, to the outer areas of the pulp, and, ultrastructurally, apoptotic cells in young rat molar pulps show characteristic morphological changes (Vermelin et al., 1996). In this context, it has been suggested that apoptotic bodies may contribute to the initiation of biological mineralization (Kardos and Hubbard, 1982). Thus, the role of apoptosis may go beyond the simple regulation of cell density and renewal of cell populations.
(III) Pulp Repair
(III-1) Reaction of dentin and pulp to mild injury —healing, regeneration and repair
(III-1-1) Reactionary dentin
The ability of the odontoblast to respond to injury (e.g., caries, cavity preparation) and up-regulate its secretory activity leading to deposition of reactionary dentin is well-established (Smith et al., 1994). The important feature of this response is that there is no cell renewal and the odontoblasts have to survive the injury. This is in contrast to reparative dentinogenesis, where the intensity of the injury is of a magnitude that results in odontoblast death and cell renewal by a new generation of odontoblast-like cells (Lesot et al., 1993; Smith et al., 1994). The process of reactionary dentinogenesis involves up-regulation of odontoblast activity, often in quiescent cells at the stage of physiological secondary dentinogenesis, in response to the injury stimulus. The nature of the signaling process from this stimulus may be rather variable and has been hypothesized to result from the release of growth factors and other bio-active molecules from the dentin matrix during injury (Smith et al., 1995). Consequently, the up-regulatory signaling may be rather non-physiological and lead to compositional differences in matrix secretion during primary dentinogenesis. Behind a calcio-traumatic line (Fig. 2), a more or less tubular dentin is secreted, bearing most of the characteristics of orthodentin, although not always positively stained by the “Stains-all” method. This suggests that reactionary dentin does not always contain phosphorylated proteins (Takagi and Sasaki, 1986), or perhaps that the proteins present at that location are not phosphorylated. By definition, reactionary dentin is secreted by surviving odontoblasts, and thus other pulp cells are not involved in its synthesis. A variety of bioactive molecules may participate in the signaling of reactionary dentinogenesis, although relatively few have been characterized. Members of the TGF-β family, including TGF-β1, TGF-β3, and BMP-7, are capable of up-regulation of secretion (Sloan and Smith, 1999; Sloan et al., 2000b), although undoubtedly there may be other molecules that are capable of participation in these signaling processes.
The injury to the odontoblast that produces a reactionary dentinogenic response may well be responsible for the calcio-traumatic line that delineates the dentin matrix secreted pre- and post-injury. This line is indicative of an aberration in matrix secretion and mineralization and implies that the injury resulted in abnormal odontoblast behavior at this point. The relationship between the degree of injury that an odontoblast can withstand and still survive is unclear. Correlation of caries lesion progression and reactionary/reparative dentinogenic events is hampered by lack of chronological information on tissue changes that would distinguish odontoblast survival and renewal. Morphological changes in odontoblasts beneath caries lesions have been reported (Bjørndahl et al., 1998), and in very active lesions, tertiary dentinogenic processes may be absent altogether (Bjørndahl and Darvann, 1999). However, these data do not allow for discrimination of reactionary from reparative dentinogenesis. Similar problems can exist in the study of pulpal responses beneath cavity preparations where surgical procedures can cause odontoblast death (Kitamura et al., 2001). Nevertheless, examination of the relationship among depth of cavity preparation, odontoblast numbers, and the tertiary dentinogenic response beneath the cavity indicates that if the cavity is prepared carefully enough, extensive odontoblast loss is seen only when pulpal exposure is approached (About et al., 2001; Murray et al., 2001). This raises an important question as to the ability of the odontoblast process to self-repair after cavity preparation. The latter studies imply that it is only when the process is cut near the cell body that the odontoblast is no longer able to self-repair such injury, since it is unlikely that any cell could withstand an on-going loss of integrity in its membrane. The nature of the repair events that might result in restitution of the integrity of the membrane of the cut odontoblast process, however, remains to be elucidated and will provide a fruitful area of study.
(III-1-2) Reparative dentin and odontoblast-like cells—factors influencing the recruitment and differentiation of secondary odontoblasts
With injury to the tooth of greater or more sustained intensity, localized odontoblast death will probably result. If suitable tissue conditions prevail, a new generation of odontoblast-like cells or “secondary odontoblasts” may differentiate from progenitor cells within the pulp and secrete a reparative dentin matrix. In the case of the injury leading to pulpal exposure, this reparative dentinogenesis may give rise to dentin bridge formation. The different possible derivations for the progenitor cells that have been suggested to give rise to this new generation of odontoblast-like cells have been alluded to above (Section II-3). They include the sub-odontoblast cells in the layer of Höhl, fibroblasts, undifferentiated mesenchymal cells from the pulp core, and vascular-derived pericytes. It is unclear, however, whether any of the adult-derived resident progenitor cell populations represents a defined stem cell population (Gronthos et al., 2000 , 2002; Young et al., 2002). Dedifferentiation and transdifferentiation of certain mature cells lead to re-evaluation of the concept of stem cells in repair processes.
It is possible that all of these derivations for the progenitor cells are valid and that the term “odontoblast-like cell” has been used rather loosely to describe any cell in the pulp capable of depositing a mineralized matrix after injury to the tooth. The phenotype of the primary odontoblast may be defined by the morphology of the cell and the matrix it secretes as well as by its pattern of gene expression, leading to the synthesis and secretion of characteristic dentin matrix proteins. Few odontoblast-like cells would fulfill all of these criteria, although it must be recognized that they are cells capable of repair and therefore may display non-physiological traits. Nevertheless, it seems likely that, at times, the term “odontoblast-like cell” may have been used inappropriately. Attempts should be made to relate the cell originating from the progenitor cell to the phenotype of the cell responsible for repair in any pulpal injury situation.
The spectrum of matrix appearance observed during pulp repair can range from an atubular fibrodentin to a relatively regular tubular reparative dentin. A fibrodentin response may often precede secretion of a more tubular dentin on its surface, which raises questions as to the dentinogenic specificity of the former. Interpretation of many of these tissue changes is often hampered by the absence of data either on gene expression by the cells responsible for the matrix secretion or on the nature of the matrix components secreted. While attempts have been made in some studies to characterize the cell phenotype during in vivo experimental reparative dentinogenesis (D’Souza et al., 1995), in vitro studies may provide a simpler model for interpretive purposes (About et al., 2000).
A striking feature of the comparison between primary odontoblast and odontoblast-like cell differentiation is the absence of involvement of dental epithelium in the latter situation. The inductive signal for odontoblast differentiation during tooth development has been proposed to be growth factor in nature, probably of the TGF-β family (Smith and Lesot, 2001). TGF-β isoform expression by odontoblasts (Sloan et al., 2000b) leads to their sequestration within the dentin matrix (Cassidy et al., 1997), from where they may be released (Smith et al., 2000) or exposed (Zhao et al., 2000) by demineralizing agents. Dentin matrix extracts are capable of inducing odontoblast-like cell differentiation when implanted in vivo within exposed cavity preparations (Smith et al., 1990). Inclusion of TGF-β neutralizing antibodies blocked odontoblast-like cell differentiation (Bègue-Kirn et al., 1992). While TGF-β1, TGF-β3, follistatin, BMP-2 or -4, and IGF-1 all showed some effects on odontoblast-like cell differentiation in cultured dental papillae, only TGF-βs stimulated gradients of odontoblast-like functional differentiation over large areas (Bègue-Kirn et al., 1994). In vivo implantation of TGF-β1 or BMP-2, -4, or -7 in pulp-capping situations (Rutherford et al., 1993 , 1994; Nakashima, 1994a,b; Nakashima et al., 1994; Six et al., 2002b) and implantation at central pulpal sites (Tziafas et al., 1998) gave rise to variable results. The complex injury situation modeled in these studies may have been influenced by the mode of presentation of the growth factor for signaling cell differentiation and could have contributed to heterogeneous data. However, in a simpler model of injury involving a needle-punch injury to the odontoblast layer in cultured tooth slices, TGF-β3, but not TGF-β1, was capable of inducing odontoblast-like cell differentiation in association with the agarose beads used for growth factor delivery (Sloan and Smith, 1999). De novo reparative dentinogenesis involving odontoblast-like cell differentiation could be observed when TGF-β1, in an alginate hydrogel, was applied to the cut pulpal surfaces of cultured human tooth slices (Dobie et al., 2002). Clearly, anomalies exist in the reported effects of the different members of the TGF-β family, and especially TGF-β1 and -β3, on odontoblast-like cell differentiation. Differences in effects may simply reflect immobilization of the growth factor at its site of action by interaction with the extracellular matrix components (Smith et al., 1998), or there may be a requirement for presentation of the growth factor in a particular conformational or complex arrangement. Detailed studies of signal transduction events during odontoblast-like cell differentiation are required to correlate the nature of signaling processes with specific morphological and functional changes in the cell.
(III-1-3) Odontoblast-like cells: gene regulation and transcription factors
Little is known of gene regulation and transcription factors that are involved in the recruitment and differentiation of odontoblast-like cells. However, those approaches provide powerful tools to aid our understanding of the cellular events taking place.
Tooth cavity preparation models of dental injury have been widely used to study odontoblast-like cell differentiation either with or without mechanical exposure of the pulp. Under carefully controlled conditions, cavity preparation can be undertaken without death of the underlying odontoblasts (Smith et al., 1994). However, odontoblast death can be readily induced during cavity preparation, even to the extent of aspiration of odontoblast nuclei within the dentinal tubules (Mjör, 1983). Animal models with smaller teeth may be more susceptible to injury during cavity preparation than larger teeth, where dissipation of the heat generated by cutting is easier. The cellular responses to these various degrees of injury are likely to be quite wide-ranging and should be considered in the comparison of data from different studies. Molecular characterization of the cellular responses to injury may be valuable in assessments of its impact on repair. Apoptosis was observed in odontoblasts 1 hr after cavity preparation in rat molars and was seen after 1 day in the underlying cells of the sub-odontoblastic region, with parallel expression of Bcl-2 (Kitamura et al., 2001).
A variety of molecules probably contributes to the signaling cascade, resulting in odontoblast-like cell differentiation. The nuclear proto-oncogenes, c-jun and jun-B, are known to control transcription via a factor termed “activator protein-1” (AP-1), which is stimulated by growth factors such as BMPs, Tumor Necrosis Factor, and Insulin-like Growth Factor. AP-1 stimulates the transcription of osteocalcin, alkaline phosphatase, and collagens. From 3 to 7 days after cavity preparation, c-jun and jun-B are expressed in the pulp cells underlying the cavity. After 14 days, they are expressed only in the formative cells lining the reparative dentin layer (Kitamura et al., 1999). Notch receptors and the Delta 1 ligand are expressed during tooth development, but not in adult teeth except after injury, when they appear to be up-regulated again. Notch 3 is associated with vascular structures, whereas Notch 1 is mostly found in a few pulp cells close to the injury site (Mitsiadis et al., 1999). Pulp cells involved in reparative dentinogenesis have been reported to express collagen type I, but not type III collagen and DSP (D’Souza et al., 1995). Our knowledge of gene, transcription factor, and growth factor expression during dental injury is still in its infancy but will contribute significantly to our understanding of the molecular control of odontoblast-like cell differentiation and repair.
(III-2) In vivo
experimental models of dental injury and repair
Preparation of class V cavities in sound human teeth extracted for orthodontic reasons has been widely used for the assessment of restorative procedures and materials. Despite the limitations of such a model, where the dentin thickness is limited and pathologic changes in the dentin resulting from the caries are absent, it probably represents the closest to the real clinical situation. Ethical considerations now prevent the use of such models, with the possible exception of re-analysis of archival histological material prepared some years ago.
In vivo animal models provide an acceptable alternative to human models, although species differences should be taken into consideration in the interpretation of tissue responses. The animals used include the monkey (Cox et al., 1992), the dog (Tziafas et al., 1998), the ferret (Smith et al., 1990), and the rat (Six et al., 2000). While such approaches are valuable for the examination of tissue responses in more physiological and clinical contexts, the complexity of the model and the control of possible experimental variables are important considerations. The surgical procedures giving rise to the tissue injury probably represent the most critical variable and have been alluded to in the preceding section of this review. Cavity preparation is well-recognized to be technique-sensitive, leading to a variety of tissue responses, and subsequent cavity treatment and restoration can also invoke pulpal responses (Mjör, 2002).
Both ethical and cost considerations have favored use of the rat model for in vivo studies by many researchers, despite the small size of the teeth. The special nature of this particular animal model warrants more detailed comments to put its use into perspective. The rat pulp demonstrates exceptional resilience and self-reparative capacity, which must be taken into account in the interpretation of experimental results. The continuously growing nature of the rat incisor limits its usefulness, and access to the mandibular first molar is difficult during surgery. The maxillary molar provides better surgical access, and Ohshima (1990) proposed cavity preparation on the mesial aspect of these teeth. Elimination of the gingival margin by electrosurgery allows cervical cavities to be prepared in a zone facing the middle of the coronal pulp in this tooth, thereby avoiding preparations near the tip of the pulp horn, where excessive reactionary dentinogenesis can be a problem (Goldberg et al., 1999). Cavities should be prepared with a tungsten carbide bur within 2 sec, followed by a short rinse with water, to avoid heat damage. The dorsal surgical presentation of the animal precludes use of water cooling during cutting; otherwise, water inhalation is a risk. The cavity is calibrated, and its size (width and depth) largely depends on the size of the bur.
The expertise of the investigator is crucial in these surgical procedures, and this reinforces the view that cavity preparation is a critical variable (Mjör, 2002). After cavity preparation, pulp exposure can be performed by mechanical perforation of the cavity floor with a steel probe. This approach avoids the extensive pulp damage caused by exposure during cutting with the bur, but does lead to dentin fragments being pushed into the pulp. While this tissue debris does not appear to invoke an inflammatory response (Decup et al., 2000), auto-induction of reparative dentinogenesis can be seen on the surfaces of the dentin particles (Tziafas et al., 1992). Even in the absence of any pulp-capping agent, spontaneous calcospherite formation can be observed in the mesial area of the coronal pulp chamber after exposed cavity preparation in the rat. This leads to deposition of an osteodentin-like matrix interspersed with unmineralized pulp remnants. Thus, any evaluation of the biological properties of implanted molecules must be compared with appropriate controls. Implanted molecules may be applied bound to a carrier (collagen) or adsorbed onto the surfaces of agarose beads or other substrates prior to the sealing of the cavity with a glass-ionomer cement to avoid bacterial infection. Generally, inflammatory processes will have subsided within 8 days, during which progenitor cell recruitment for reparative dentinogenesis will have occurred. Reparative dentinogenesis is initiated within 2 wks, and in the presence of calcium hydroxide, a thick dentin bridge will form within the subsequent 2 wks. While this rat model can be capricious in the hands of an inexperienced investigator, it is capable of providing valuable data in more experienced hands. Use of alternative, higher species of animals is not always viable in many countries because of ethical restrictions, or due to a high cost.
(III-3) In vitro
models of dental injury and repair
In vitro culture models have found widespread use because of their apparent simplicity. Cells emerging from cultured pulp explants give rise to a population of pulp-like cells with specific characteristics (Nakashima, 1991; Kasugai et al., 1993; Hanks et al., 1998; Ueno et al., 2001). After 15–20 days of culture, they are strongly positive for alkaline phosphatase and form mineralization nodules that stain positively with the von Kossa reaction. The nodules have been shown to be calcium- and phosphorus-rich by electron probe analyses (Stanislawski et al., 1997). Once differentiated, these pulp cells display an α-smooth muscle actin-rich phenotype and consequently may be derived from myofibroblasts or pericytes (Alliot-Licht et al., 2001). Pericytes have also been reported to be possible osteoblast progenitor cells (Brighton et al., 1992). What is unclear is the extent to which cell selection occurs during pulp explant culture. Both the culture conditions and outgrowth of cells from the explant will almost certainly favor selection of particular populations of migratory cells. Caution is therefore required in extrapolation of data from such cell cultures to the pulp in general.
Pulp cells in culture express collagen types I and III, osteonectin, osteopontin, fibronectin, and alkaline phosphatase, but not DPP or DSP. Selected clones of cells, however, express DSP and other molecules considered as odontoblast-specific markers, suggesting that spontaneous transformation of cell phenotype can occur during culture (Hanks et al., 1998). In some conditions, such as addition of β-glycerophosphate, cultured pulp cells will form mineralized nodules and express odontoblast markers (About et al., 2000).
In addition to primary pulp cell cultures, attempts have been made to derive immortalized odontoblast-like cell lines from dental papilla cells. Transfection with simian virus 40 large T antigen gave rise to a cell line that was positive for DSPP, type I collagen, and alkaline phosphatase (MacDougall et al., 1995). Cell lines that expressed collagen type I, alkaline phosphatase, and osteocalcin have been reported after immortalization with human papillomavirus 18 (HPV 18 E6/E7) (Thonemann and Schmalz, 2000). With use of a gene for telomerase, immortalized cells possessing an odontoblast-like phenotype were also obtained. These cells express DMP-1 and DSPP and produce a mineralized dentin matrix both in vivo and in vitro (Hao et al., 2002). These immortalized odontoblast-like cell lines may be valuable tools for molecular investigations, but their ability to undergo cell division is at variance with the post-mitotic nature of odontoblasts under physiological conditions. The significance of these cell cycle differences remains to be established.
Tooth organ culture has been adopted for the study of dentinogenic events in rat incisor teeth, and this model afforded the important observation that separation of odontoblasts from contact with their dentin matrix leads to marked changes in their morphology (Heywood and Appleton, 1984). Although odontoblasts continue to show secretory activity when grown in association with the dentin matrix after pulp extirpation (Embery and Smalley, 1980), their functional behavior can probably be studied for time periods of only fewer than 24 hrs before cellular atrophy becomes significant. It is possible to grow these tissues for periods of up to 2 to 4 wks if thick slices of human or rat teeth are grown intact in culture without removal of the pulp tissues (Magloire et al., 1995; Sloan et al., 1998; Dobie et al., 2002). This approach also has the advantage that normal architectural relationships of the dentin-pulp complex are maintained during culture. While such models are valuable for studying events over periods of some days (Sloan and Smith, 1999), caution must be exercised over longer periods of weeks in the absence of information on whether or not physiological cell behavior is maintained. Nevertheless, such organ culture models provide valuable alternatives when in vivo experimentation is inappropriate and close control of experimental variables is required.
(III-4) Calcium hydroxide as a reference for induction of reparative dentinogenesis
Calcium hydroxide has been widely used as a pulp-capping agent for more than 60 years (Zander, 1939) and has been perceived by some as the gold standard against which other capping procedures should be evaluated. Its use on the exposed pulp can induce a reparative dentinogenic response leading to dentin bridge formation (Fig. 3). However, Ca(OH)2 does not have a specific target of action, and its behavior will be reviewed to assess the validity of use of this material as a reference.
As a pulp-capping agent, the strongly alkaline pH of Ca(OH)2 may contribute significantly to its action. Initially, this high pH appears to cause local necrosis of the pulp tissue around the injury site, and migration of pulp cells to this area is seen after approximately 2 wks (Schroder, 1985). While high pH environments are not generally conducive to biological activity, in this situation, Ca(OH)2 may prevent bacterial infection, thereby dampening inflammatory responses and providing a relatively sterile environment in which subsequent repair can occur. As cells migrate into the necrotic area, odontoblast-like cell differentiation commences and a mineralized reparative dentin bridge is formed (Schroder, 1985; Mjör et al., 1991). Two important questions arise from these observations: (1) What is the nature of the recruited progenitor cells for dentin bridge formation? and (2) What signaling processes are responsible for induction of differentiation of these cells to odontoblast-like cells?
DNA synthesis can be inhibited in the presence of Ca(OH)2 in vitro. Alkaline phosphatase activity and protein synthesis are diminished, although incorporation of the amino-acids leucine and proline, precursors of collagen and non-collagenous proteins, is increased (Alliot-Licht et al., 1994). Human dental pulp cells cultured on Ca(OH)2 differentiate into odontoblast-like cells after 4 wks, synthesize type I but not type III collagen, and display apical accumulation of actin and vimentin (Seux et al., 1991). Pulp cell activities observed after Ca(OH)2 application are dependent on the stage of examination and the tissue conditions prevailing at that time.
The signaling processes responsible for Ca(OH)2-induced odontoblast-like cell differentiation remain elusive. To suggest that differentiation represents spontaneous differentiation implies activation of an endogenous repair mechanism. Such a concept may prove founded if Ca(OH)2 is able to solubilize bio-active molecules with appropriate signaling functions from the dentin matrix (Tziafas et al., 2000). It has also been suggested that alterations in calcium levels in the cellular environment can invoke responses of apoptosis or differentiation of odontoblast-like cells (Kardos et al., 1998). However, there are probably aberrations in signaling, as suggested by the rather abnormal appearance of the initial matrix of the dentin bridge, which is usually of the osteodentin type rather than true tubular orthodentin. Discontinuities in dentin bridge formation give rise to so-called “tunnel defects” (Cox et al., 1996), the presence of which impairs the bridge’s properties as a permeability barrier. During later bridge formation, the matrix structure often becomes more regular, resembling true tubular orthodentin.
Thus, Ca(OH)2 appears to have relatively non-specific effects as an inducer of reparative dentinogenesis and considerable heterogeneity in its action. However, the broad experience gained with Ca(OH)2 in a variety of experimental and clinical situations makes it a useful treatment for comparative purposes for the evaluation of new pulp-capping agents.
(III-5) Signaling molecules for pulp repair
(III-5-1) Diversity of molecules and experimental approaches
The possibility for the use of a variety of biological macromolecules for signaling repair in pulp-capping and other treatment situations has offered exciting opportunities for the development of novel biomimetic approaches in restorative dentistry. A plethora of molecules—including growth factors, dentin matrix proteins, and extracellular matrix molecules—has been evaluated for their potential contributions to dental tissue repair. These studies were largely derived from the observations that dentin matrix is auto-inductive, obviating the requirement for dental epithelium to signal odontoblast-like cell differentiation. The histopathologic observations of dentinogenesis on the surfaces of dentin chips pushed into the pulp during cavity preparation led to studies which showed reparative dentinogenesis when demineralized dentin matrix (Tziafas and Kolokuris, 1990) or soluble dentin matrix extracts (Smith et al., 1990) were implanted into exposed pulps. These observations catalyzed more detailed studies of specific matrix molecules—such as fibronectin (Tziafas et al., 1992), bone sialoprotein (Decup et al., 2000), and the amelogenin polypeptides A+4 and A-4 (Six et al., 2002c)—for the evaluation of their dentinogenic effects during repair. The presence of growth factors in dentin matrix and their putative role in signaling odontoblast differentiation during embryogenesis have also led to examination of the effects of TGF-βs, BMPs, FGFs, and IGFs in reparative dentinogenesis (Rutherford et al., 1993 , 1994; Nakashima, 1994a,b; Nakashima et al., 1994; Tziafas et al., 1998; Sloan and Smith, 1999). In all of these various studies, reparative responses have been observed ranging from initial secretion of relatively non-specific atubular fibrodentin to apparently more specific tubular dentinogenic responses, often superimposed on atubular fibrodentin-like matrix. This raises many questions: What constitutes a specific dentinogenic response as opposed to secretion of a non-specific mineralized mass? To what degree are these molecules directly signaling odontoblast-like cell differentiation or indirectly influencing the tissue environment to facilitate signaling by other endogenous matrix molecules? Thus, several questions remain as to the signaling mechanisms in odontoblast-like cell differentiation during reparative dentinogenesis and which specific molecules are involved. From the viewpoint of potential clinical applications, such questions may be of lesser importance, since the final outcome of the treatment is the main goal. However, a mechanistic understanding of how that outcome is reached is important if it is to be achieved in a well-controlled manner.
(III-5-2) Strategies for pulp repair using specific classes of molecules
There is a plethora of biological molecules that could potentially influence the induction of reparative dentinogenesis and pulp repair. Strategies have been developed to investigate groups of molecules potentially available within the pulp after injury for natural mechanisms of repair and/or which may be able to recapitulate developmental events. A few of these strategies will be considered to provide guidance for future studies.
Members of the TGF-β family have been popular candidates for the induction of reparative dentinogenesis in view of their implication in developmental odontoblast differentiation (for review, see Smith and Lesot, 2001). BMP-7 or OP-1 was one of the first growth factors to be examined in pulp repair. In the monkey, BMP-7/OP1 induced widespread osteodentin formation throughout the pulp, including the total occlusion of the roots of treated teeth (Rutherford et al., 1993 , 1994). An irregular reparative osteodentin was deposited in the coronal pulp of rat molars 30 days after BMP-7 treatment, and extensive reparative dentin was observed in the roots of these teeth, beneath a calcio-traumatic line, with complete occlusion of the mesial root canal in many cases (Six et al., 2002a). These observations highlight an important aspect of reparative dentinogenesis, namely, that the coronal and root regions of the pulp may provide different environments for repair. The reasons for these differences are unclear, but their elucidation could prove fruitful for the development of novel endodontic therapies for clinical use. The above studies also indicate that consideration must be given to the regulation of reparative processes. Uncontrolled reparative dentinogenesis poses a problem in an endodontic situation, unless apical closure is adequate to prevent the spread of mineralization into periapical areas. Control might be exerted through the dose-dependency of the reparative events, but this requires further investigation. Such issues underline the need for a full understanding of the mechanisms of action of molecules such as BMP, before they can be effectively exploited for clinical use. BMP-7 can induce new bone formation at ectopic sites, and BMP receptors have been identified within dental pulp (Takeda et al., 1994a,b; Toyono et al., 1997). However, we have little understanding of the identities and phenotypes of the targeted cells or of the cellular processes affected during BMP-induced pulp repair.
In view of the auto-inductive effects of displaced dentin particles on reparative dentinogenesis in the pulp, the effects of some dentin matrix molecules have been examined. Phosphorylated proteins provide potential candidates in this context, and BSP has been selected for several reasons. The polyglutamic acid sequences in BSP provide hydroxyapatite (HAP)-binding sites and have the potential to mediate initial nucleation of HAP crystals. Fibrillogenesis of collagen is enhanced by BSP, which is also expressed by odontoblasts but not by pulp cells (Decup et al., 2000). Implantation of BSP into rat molar exposed pulps led to cellular changes in the pulp within 1 wk, and reparative dentinogenesis commenced after 2 wks, often in close association with the surface of dentin debris pushed into the pulp during perforation. After 1 mo, the mesial aspect of the coronal pulp was filled with a homogeneous mass of atubular dentin (Figs. 4a, 4b). In contrast to the studies with BMP-7, no effect of BSP was observed on the radicular pulp. Thus, these 2 bioactive molecules appear to target different areas of the pulp for repair. BSP might find clinical application in the induction of pulp mineralization in the coronal pulp, whereas BMP-7 could induce occlusion of both the radicular and coronal regions of the pulp (Six et al., 2002b). In the situation of coronal pulp repair, restoration of the physiological structural integrity of the tissue, i.e., a tubular orthodentin matrix, may be considered preferable, but in an endodontic application, apical sealing of the root canal is important, and a less permeable and homogeneous reparative tissue would be favored.
The effects of a parent molecule of the SIBLING family have also been investigated. MEPE (matrix extracellular phosphoglycoprotein), recently identified in odontoblasts (MacDougall et al., 2002), has been implanted via agarose beads into rat maxillary molars. After 1 mo, extensive reparative dentin formation was seen in the mesial third of the pulp, with occlusion of the root canal (Figs. 5, 6).
The role of epithelially derived molecular signaling in odontoblast differentiation during embryogenesis is well-established and has led to the examination of these molecules in pulp repair. A+4 and A-4 are found in dentin and represent a low-molecular-mass amelogenin polypeptide (Veis et al., 2000). Their cDNA has been identified in the odontoblast cDNA library and represents an alternatively spliced gene product. The first polypeptide is expressed by all the amelogenin gene exons (A+4), whereas A-4 is expressed by all the exons except the 4th (Veis et al., 2000). Implantation of A+4 and A-4 via agarose beads into rat molar pulps resulted in recruitment of a dense layer of cells on the bead surface after 1 wk, suggesting that these molecules have a chemotactic effect (Six et al., 2002c). In the root, instead of a single layer of odontoblasts at the border of the pulp, a double layer was seen, suggesting an unusual pattern of induction of odontoblast-like cell differentiation. After 15 days’ implantation of A+4, the reaction was restricted to an area near the exposure, where a thick dentin bridge developed after 30 days. In contrast, the initial reaction was more diffuse with A-4. Mineralization was observed both in the mesial coronal pulp and in the mesial root, and in many instances the latter was totally occluded with a dense matrix. Clearly, there is a need to screen a wide variety of bio-active molecules for potential use in pulp repair and to examine their mechanisms of action closely.
(IV) Conclusions
The last decade has proved to be an exciting time for pulp biology and has led to rapid advances in our knowledge of repair in this tissue. At the start of a new millennium, the use of biological molecules for the development of novel restorative treatment modalities in clinical dentistry is in sight. These approaches have potential applications in unexposed cavity preparations for protection of the pulp from harmful effects of dental materials by increasing the residual dentin thickness through reactionary dentinogenesis, as well as in exposed pulp situations for restoration of the structural integrity of the dentin wall by reparative dentinogenesis. In the severely compromised pulp, it may even be possible to use biological approaches in endodontic therapy to seal the root canal.
A thorough understanding of the various biological processes involved in repair will be pivotal to our future success in developing these approaches. Our review of this area has highlighted that, despite the considerable progress achieved, there are still numerous fundamental questions to be addressed. Emerging technologies in molecular and cell biology will be of crucial importance to our understanding of the cells responsible for repair and the molecular signals that direct and regulate these cells. Of fundamental importance will be an understanding of the nature and phenotype of the cells involved in repair, their molecular regulation, and the secretory behavior which gives rise to various mineralized matrices. Such knowledge will also underpin progress in tissue engineering in this area.
Rat molar. Polarized secretory odontoblasts (O) are seen in the upper right. The sub-odontoblastic layer of Höhl cells (H) is continuous with the outer part of the pulp (P). 650×.
Human tooth. Reactionary dentin is delineated from the dentin by a calcio-traumatic line (arrowheads) and from the pulp by predentin and odontoblast layers. 360×.
In the rat molar model, 8 days after implantation, Ca(OH)2 induces the formation of a reparative dentinal bridge. P, pulp; C, cavity. 120×.
Thirty days after implantation, BSP induces the formation of a homogeneous mineralized tissue
MEPE implantation. After 30 days, the mesial part of the pulp (P) and mesial root canal were filled with reparative dentin (arrowheads). C, cavity. G, gingiva. 10×.
Structure of the reparative dentin induced by MEPE implantation. Dentin debris and remnants of agarose beads, used as carriers (arrow), are embedded in the reparative dentin that occludes the perforation. 360×.
Footnotes
Acknowledgements
We acknowledge Dr. Ngampis Six’ contributions as a PhD student involved in the research on BSP, BMP-7, A+4 and A-4, and MEPE. We thank Dr. Erdjan Salih (Boston), Dr. Bruce Rutherford (Ann Arbor), Dr. Arthur Veis (Chicago), Dr. Pam DenBesten (San Francisco), and Acologix Inc. for their kind collaboration. We acknowledge the Université Paris V for an “Axe d’Excellence” grant, for the Research Grant CNRS-INSERM “Ingénierie tissulaire”, the Institut Français pour la Recherche Odontologique-IFRO, research grants from the Wellcome Trust and UK MRC, and EPSRC and the European Program COST Odontogenesis for their support.