Abstract
The extracellular matrix (ECM) of cementum resembles other mineralized tissues in composition; however, its physiology is unique, and it contains molecules that have not been detected in other tissues. Cementum components influence the activities of periodontal cells, and they manifest selectivity toward some periodontal cell types over others. In light of emerging evidence that the ECM determines how cells respond to environmental stimuli, we hypothesize that the local environment of the cementum matrix plays a pivotal role in maintaining the homeostasis of cementum under healthy conditions. The structural integrity and biochemical composition of the cementum matrix are severely compromised in periodontal disease, and the provisional matrix generated during periodontal healing is different from that of cementum. We propose that, for new cementum and attachment formation during periodontal regeneration, the local environment must be conducive for the recruitment and function of cementum-forming cells, and that the wound matrix is favorable for repair rather than regeneration. How cementum components may regulate and participate in cementum regeneration, possible new regenerative therapies using these principles, and models of cementoblastic cells are discussed.
Introduction
Periodontium" refers to the tissues that collectively invest and support the teeth, and consists of the gingiva, periodontal ligament, alveolar bone, and cementum. The structure and composition of the periodontium are affected in many acquired and heritable diseases, and the most significant among these is periodontal disease. The hallmarks of periodontal disease are destruction of soft connective tissues, bone loss, and loss of connective tissue attachment to cementum; these alterations, if left untreated, lead to tooth loss. The aim of periodontal therapy is to regenerate and restore the various periodontal components affected by disease to their original form, function, and consistency ( Aukhil, 1991; Garrett, 1996). Regeneration requires restoration of alveolar bone height to the cemento-enamel junction, regeneration of gingival connective tissue destroyed by inflammation, formation of new acellular extrinsic fiber cementum on previously exposed root surfaces, synthesis of Sharpey’s fibers and their insertion into root surfaces, and re-establishing epithelial seal at the coronal portion. New therapeutic approaches available to achieve these objectives include use of barrier membranes for guided tissue regeneration, and applying growth factors and enamel matrix proteins to root surfaces; however, the effectiveness of these approaches is not predictable, especially on new cementum and attachment formation. These approaches and the principles behind them have been reviewed recently in several excellent publications ( Cochran and Wozney, 1999; MacNeil and Somerman, 1999; Wikesjö and Selvig, 1999; Bartold et al., 2000); therefore, they will not be discussed here. Cementum is the site where soft-tissue attachment has to be re-established, and cementum matrix is a rich source of many growth factors which influence the activities of various periodontal cell types ( Narayanan and Bartold, 1996; Saygin et al., 2000); therefore, in this review we will examine how the cementum can play a regulatory role in periodontal regeneration. In light of emerging evidence showing that local environment is a major regulator of how cells respond to environmental cues and signals ( Sastry and Horwitz, 1996; McKay, 1997; Michalopoulos and DeFrances, 1997), we believe that relatively little emphasis has been placed on the role of cementum in periodontal regeneration. Our goal is to highlight the manner by which cementum constituents can participate in and regulate periodontal regeneration, and to discuss how these principles can be extrapolated to regenerative therapy of the periodontium.
Wound-healing Events
To facilitate discussion on the possible role of cementum in periodontal regeneration, we will first briefly review the events and molecules involved in wound healing. Wound healing involves three overlapping phases that are interdependent ( Clark, 1996). The general principles and events have been characterized with the use of cutaneous wound models, and these apply to healing of surgical and diseased periodontal wounds as well. Traumatic injury damages blood vessels and causes hemorrhage and extravasation of blood, and a blood clot is formed. The blood coagulation process and activated complement pathway generate many polypeptide mediators, and the blood clot serves as a provisional matrix for the migration of inflammatory cells. Neutrophils and monocytes cleanse the wound of bacteria, foreign particles, and dead tissue, and macrophages and platelets secrete several polypeptide mediators that direct and regulate the activities of various cells participating in healing.
Re-epithelialization begins within hours of injury, when epithelial cells near the wound margin migrate to the wound site. The number of epithelial cells increases at the wound due to migration and proliferation, and eventually the breach in epithelium is sealed. During this time, angiogenesis and synthesis of collagens and other ECM components become active, and the clot is replaced by "granulation tissue". The wound is filled with activated fibroblasts, and a proportion of these cells is transformed into myofibroblasts. Appearance of the myofibroblasts corresponds to the beginning of connective tissue compaction, and the wound contracts. The wound contraction shrinks the wound size and brings the wound margins toward one another ( Singer and Clark, 1999). During the ensuing "tissue remodeling phase", the granulation tissue matrix is replaced with fresh connective tissue, and new blood vessels disappear through degradation and apoptosis. A fibrous scar replaces the wound when regeneration is not possible. During the following weeks and months, remodeling continues, and up to 70% of original tissue strength is restored ( Clark, 1996; Singer and Clark, 1999).
Wound healing involves migration, adhesion, proliferation, and differentiation of several cell types. All these activities are triggered when polypeptide mediators bind to their cell-surface receptors and when integrins bind to ECM components. Different growth factors regulate different cell functions. For example, transforming growth factor (TGF)-α, a member of the epidermal growth factor (EGF) family, and heparin-binding EGF (HB-EGF), keratinocyte growth factor (KGF, also called FGF-7), and TGF-β and their receptors are involved in re-epithelialization. The expression of KGF is enhanced 100-fold 24 hours after injury ( Martin, 1997). FGF-1 and -2 (acidic and basic FGF), vascular endothelial cell growth factor (VEGF), TGF-β, angiogenin, angiotropin, angiopoietins, and hepatocyte growth factor are strongly angiogenic, whereas thrombospondins and angiostatin inhibit angiogenesis ( Conway et al., 2001). Platelet-derived growth factor (PDGF), which is mitogenic to fibroblasts and other mesenchymal cells, is an important growth factor in wounds, and gingival epithelium is a rich source of this growth factor in gingival wounds ( Green et al., 1997). Fibroblast collagen synthesis is activated by TGF-β and connective tissue growth factor (CTGF). The TGF-β also appears to induce the transformation of fibroblasts into myofibroblasts, and TGF-β and PDGF participate in wound contraction.
Cell migration and adhesion require active participation of integrins. For instance, migration of epithelial cells requires dissolution of α6 β 4 integrin, the receptor that mediates the attachment of hemidesmosomes and laminin. The receptors to fibronectin and tenascin, α5 β 1 and αv β 6 integrins, and vitronectin receptor αv β 5 are expressed, and relocalization of type I collagen receptor α2 β 1 occurs. The appearance of integrins that bind to fibrin and fibronectin seems to be rate-limiting for granulation tissue formation, and at the top of sprouting capillaries, endothelial cells actively express the integrin α5 β 3. Integrins are also required for wound contraction.
In addition to changes in integrin expression, proteolysis is necessary for cell migration through the fibrin clot and matrix and for tissue remodeling. Inflammatory cells and most other cells produce several different matrix metalloproteinases (MMPs) during wound healing. In cutaneous wounds, keratinocyte migration at the leading edge requires the fibrinolytic enzyme plasmin, and tissue-type plasminogen activator (tPA) and urokinase-type plasminogen activator (uPA) activate the plasmin. The expression of uPA, tPA, MMP-1 (also called interstitial collagenase), MMP-9 (gelatinase B), and MMP-10 (stromelysin-2) is up-regulated at the wound margin. Keratinocyte movement requires MMP-1 expression, and this enzyme also facilitates cell directionality; the MMP-1 expression is turned off once re-epithelialization is complete ( Parks, 1999). The MMPs also participate in remodeling of the granulation tissue matrix. Although there is considerable redundancy in the substrate specificity of MMPs, differences in their affinities to different substrates and their compartmentalization determine the course of their action and the composition of matrix laid down.
Several "master genes" participate in the wound-healing process. These genes code for proteins that control body plan formation during embryogenesis and development, and include homologues of TGF-β and BMPs and transcription factors encoded by "homeobox genes". The latter are characterized by the presence of a common 61-amino-acid DNA-binding motif called "homeodomain". These proteins regulate the expression of lineage-specific genes during normal embryonic development, and control pattern formation and cell fate and identity. In adult eukaryotic cells and during wound healing, the master genes participate in cell growth and differentiation, apoptosis, and cell-cell and cell-ECM interactions ( Srebrow et al., 1998; Cillo et al., 2001). Several homeobox genes are involved in vascular remodeling and angiogenesis ( Gorski and Walsh, 2000). The expression of master genes is regulated through interactions among cytokines, growth factors, ECM, and adhesion molecules, and growth factors that regulate their expression include TGF-β, BMPs, and FGFs. For example, the muscle segment homeobox gene MSX and "paired" PAX are regulated by FGFs and BMPs. Altered expression of these genes is associated with neoplastic and metabolic diseases ( Cillo et al., 2001). Programmed expression of master genes is likely to be essential for differentiation programs associated with wound healing, and aberrations in their regulation may lead to wound repair rather than regeneration.
Factors that Determine Whether Wounds Heal by Regeneration or Repair
Although the extent of injury and the amount of lost tissue that must be filled are important determinants, whether a damaged tissue heals by regeneration or is repaired by fibrous scar depends upon two crucial factors: (1) the availability of needed cell type(s), and (2) the presence or absence of the cues and signals necessary to recruit and stimulate these cells. These factors are not mutually exclusive. The signals are provided by diffusible factors such as growth factors and cytokines; however, the ECM regulates how cells respond to the signals.
In most vertebrates, the capacity to regenerate is limited to a few tissues, such as liver and bone. In these tissues, regeneration may mimic events of embryonic differentiation of multipotential stem cells. However, mammalian tissues constituted by "permanent cells" such as neurons and cardiac myocytes usually heal by repair. One reason for the failure of these tissues to regenerate was believed to be due to the absence of appropriate stem or progenitor cells. However, many adult tissues undergo self-renewal; therefore, they must presumably contain a life-long population of relatively slow-cycling (stem) cells. Indeed, recent evidence has shown that multipotent stem cells are present in non-regenerating adult tissues, including the brain ( McKay, 1997). The stem cells, which, in non-hematopoietic tissues, are referred to as mesenchymal stem cells or marrow stromal cells, can differentiate into many cell types. Embryonic stem cells are almost totipotent and malleable, and the diversification of the embryonic cells into cell types of their destined fate is largely complete shortly after development is complete. On the other hand, stem cells from adult tissues are relatively less malleable, and they may be pluripotent (hematopoietic and neuronal stem cells) or unipotent (epithelial stem cells are committed to differentiate into epithelial cells) ( Krebsbach et al., 1999; Fuchs and Segre, 2000). When a stem cell undergoes commitment to differentiate, it often proliferates rapidly first and then differentiates, and the differentiation to the final fate may occur in several stages (Fig. 1 ). These cells presumably respond to specific environmental signals to generate new stem cells or differentiated cells. Under normal homeostasis, differentiation of the stem cells is most likely triggered by instructive and stimulatory signals provided by the (local) environment, which consists of the ECM and growth factors sequestered in the matrix (Fig. 2 , see later). Differentiation and survival can also involve cross-talk with other cells of the same and different cell types ( Denker et al., 1999; Fuchs and Segre, 2000). Examples for the latter include mesenchymal cues for epithelial differentiation, and vice versa.
Substances present in the circulation as well as those in the local environment provide stimulus and directional signals for cells to migrate. These substances, which include growth factors, other soluble mediators, and ECM components, also promote cell division and differentiation, and more than 2000 such molecules have been purified or cloned so far ( Fuchs and Segre, 2000). During the course of wound healing, the spectrum and concentration of molecules change continuously. More than one substance is present in the local microenvironment, and the combined effect of these molecules may be additive, synergistic, or antagonistic.
Regulation by the Local Environment
The ECM is a multicomponent three-dimensional structure composed of collagens, fibronectin, elastin, other non-collagenous proteins, and proteoglycans. It serves as substratum for cell adhesion and promotes cell spreading and cytoskeletal organization ( Folkman and Moscona, 1978; Böhmer et al., 1996; Assoian and Zhu, 1997). The ECM determines the three-dimensional cell architecture and transmits and translates external mechanical and tensional forces to appropriate response signals ( Huang and Ingber, 2000). Adhesion to ECM is essential for cell cycle transit by anchorage-dependent cells. The ECM also regulates gene expression of growth factors, growth factor receptors, and other proteins, and determines the outcome of a cell’s response to growth factors. For example, in the presence of ECM, cell division is suppressed, and cells differentiate ( Adams and Watt, 1993; Juliano and Haskill, 1993; Gumbiner, 1996; Sastry and Horwitz, 1996).
Cells bind to the ECM through integrins, and the binding initiates a cascade of signaling reactions. Signaling reactions activated include tyrosine phosphorylation of focal adhesion kinase (FAK) and other signaling proteins, activation of mitogen-activated protein kinase (MAP kinase) cascade, expression of c-fos, and elevation of certain cyclin levels ( Juliano and Haskill, 1993; Yamada and Miyamoto, 1995; Assoian, 1997; Assoian and Zhu, 1997; Lewis et al., 1998). Signaling pathways induced by ECM components cooperate with those activated by growth factors in mediating their biological functions, and both integrin- and growth-factor-induced signals are necessary for expression of G1 cyclins and cell cycle progression from G0/G1 to S-phase. During G1, the expression of D1 and A cyclins and activation of cyclin-dependent kinase (cdk)-2 require adhesion-mediated integrin signals ( Guadagno et al., 1993; Fang et al., 1996; Zhu et al., 1996). Integrin- and growth-factor-receptor-induced signaling pathways converge at the MAP kinases; however, while growth factors cause a transient activation of the extracellular-signal-regulated kinases 1 and 2 (ERK1/2), the ECM induces a sustained increase of their activity. The latter is associated with increased cyclin D1 expression ( Zhu and Assoian, 1995; Böhmer et al., 1996; Weber et al., 1997). The ECM also up-regulates cdk inhibitors (CKI) p27kip1 (p27) and p21cip1/wafl (p21) levels, and this is a mechanism contributing to ECM-induced cell-cycle arrest ( Koyama et al., 1996; Raines et al., 2000).
The composition of the ECM can be responsible for the recruitment of specific cell types during wound healing. For example, fibronectin and type I collagen are chemotactic to and promote adhesion of most cell types, whereas laminin and type IV collagen are selective for certain cell types. Tenascin, on the other hand, is anti-adhesive. Thus, the composition of ECM can determine which cells are enlisted during wound healing and whether healing occurs by regeneration or repair.
Many growth factors are sequestered in the ECM, and ECM interactions can modify the binding of growth factors to their cell-surface receptors ( Schonherr and Hausser, 2000). ECM components can bind to these molecules at multiple sites, and the proteoglycans can interact through protein core as well as glycosaminoglycan chains. During inflammation, soluble mediators such as growth factors, cytokines, and chemokines are secreted by inflammatory cells, and some mediators are also released from damaged tissue by degradation. The activities of these molecules are directed toward cells participating in wound healing. On the other hand, the activity of growth factors sequestered in the ECM may be limited to the cell types needed for the maintenance of normal tissue homeostasis. Some of the sequestered molecules and ECM components may be tissue-specific in their distribution. The ECM composition and growth factors available together are likely to regulate which receptors are expressed and which biochemical signaling events are induced ( Saito and Narayanan, 1999; Cuklerman et al., 2001; Heldin, 2001). In addition, they will also determine which pathway is chosen among a network of several parallel pathways and how cells respond functionally; these in turn will determine the course of wound-healing events.
Other factors that determine cellular response include cell density and interaction with other cells. The importance of intercellular interactions is illustrated by the stromal response to epithelial injury and vice versa, and by the differentiation of embryonic cells into teratocarcinomas when injected into nude mice ( Bradley, 1990; Denker et al., 1999; Zieske, 2001).
The structural integrity and unique biochemical composition of the ECM are likely to be essential for tissue homeostasis under healthy conditions and for regeneration after injury. The importance of local environment in influencing cell behavior is illustrated by the expression of adult progenitor phenotype cells by fetal hematopoietic stem cells implanted into the adult spleen ( Geiger et al., 1998). The composition of the ECM changes with the progression of wound healing. For instance, during the early stages, when granulation tissue is actively being constituted, types III, IV, V, and VI collagen, fibronectin, and vitronectin are actively expressed, and these are replaced by type I collagen at later stages ( Eckes et al., 2000). The changing composition and quality of the ECM do not mimic the local environment of the healthy tissue. The type and concentration of polypeptide mediators available change continuously during the course of wound healing. Thus, destruction of the local environment and failure to reconstitute it are likely to be significant reasons for the failure to regenerate after injury. In this case, the provisional matrix offered by the granulation tissue may be conducive for repair rather than regeneration.
It is now recognized that cells needed for regeneration are present in most tissues. These cells may be "stem cells" removed from terminal differentiation by several stages, or precursor or progenitor cells immediately before full differentiation (Fig. 1 ). An interesting possibility is that, whereas pre-existing ECM may provide signals for precursor cells to differentiate in healthy adult tissues, the ECM synthesized by a "stem" cell and differentiating stem cell may contribute to the local environment for the differentiation of the next stage cell (Fig. 2 ).
Local Factors in Periodontal Healing
Many of the events, growth factors, and soluble mediators described above also participate in the healing of periodontal tissues after injury by trauma and other causes, including infection. The periodontium has been shown to contain biologically active mediators, and their role has been recently reviewed ( Cochran and Wozney, 1999; MacNeil and Somerman, 1999; Bartold et al., 2000; Saygin et al., 2000). The concentrations of these molecules vary in different periodontal components, and appear to be relatively higher in alveolar bone and cementum. The inability of these molecules to be detected in soft tissues, especially in the gingiva, is likely to be due to their low concentration rather than to their absence ( Miki et al., 1987; Nishimura et al., 1989; Somerman et al., 1989; Nakae et al., 1991). Although substances present in one periodontal component can influence other periodontal structures, under healthy conditions these factors are likely to affect mostly the adjacent cells. Under pathological conditions, the role of these components is likely to be minor, due to destruction of and biochemical alterations in the local environment and because of the presence of serum and inflammatory-cell products in relatively large amounts.
Biology of Cementum
Although cementum is anatomically an integral part of the tooth, functionally it is a component of the periodontium, and its major role is to serve as the site of attachment for principal collagen fibers (Sharpey’s fibers). The biology of cementum and molecular and cellular aspects of cementogenesis have been discussed elsewhere ( Schroeder, 1992; Bosshardt and Selvig, 1997; Saygin et al., 2000), and here we will review only those aspects relevant to periodontal healing. Three cementum types differing in the presence of cells and collagen fibers are distinguished in humans. The acellular afibrillar cementum covers teeth at and along the cemento-enamel junction; it consists of mineralized matrix, but lacks collagen fibrils and embedded cells. It is deposited in isolated patches over enamel and dentin, and its deposition begins at the end of enamel maturation and continues for an indefinite time ( Bosshardt and Selvig, 1997). The cells responsible for depositing this cementum have not been identified, and it appears that connective tissue cells lay the matrix when they come into contact with the enamel surface. Acellular extrinsic fiber cementum (AEFC), which serves the primary attachment function, covers cervical and middle portions of the roots, and it is usually confined to the coronal half of the root. It consists of a dense fringe of collagenous fibers implanted into dentinal matrix and, most importantly, perpendicular to the root surface. The production of AEFC commences shortly after crown formation and continues to grow as long as the adjacent periodontal ligament remains undisturbed. It is produced by cementoblasts that differentiate closest to the advancing root edge. Cellular intrinsic fiber cementum (CIFC) is found on dentin where no acellular extrinsic fiber cementum has been laid down, and its formation commences closest to the advancing root edge on the forming root. This cementum contains cementocytes entrapped in the mineralized matrix, and collagen fibers present in this cementum are parallel to the root surface. The CIFC has mainly a repair function. The CIFC may overgrow acellular extrinsic fiber cementum and vice versa, and this is called mixed stratified cementum. This cementum is confined to apical root portions and furcations, and it appears to participate mainly in the repair of previously resorbed roots ( Schroeder, 1992; Bosshardt and Selvig, 1997).
In rodents, cementogenesis begins with the deposition of a matrix on the dentin surface by Hertwig’s epithelial root sheath (HERS), disruption of the HERS, migration and organization by ectomesenchymal cells from the dental papilla, and their subsequent differentiation into cementoblasts ( Cho and Garant, 1989; Slavkin et al., 1989; Thomas and Kollar, 1989). The matrix they produce surrounds cells producing cementum, the cementoblasts, and the cementoblasts are generated by differentiation of progenitor cells, which in mice are believed to arise from the dental follicle. Epithelial cells of the Hertwig’s root sheath are believed to undergo epithelial mesenchymal transformation ( Pitaru et al., 1994; Bosshardt and Selvig, 1997).
The organic matrix of cementum is composed primarily of collagens. Type I collagen, which plays structural as well as morphogenic roles and provides scaffolding for mineral crystals, is the major species, and it accounts for 90% of all collagens. The type III collagen, which coats type I collagen fibrils, makes up ∼ 5%. Cementum contains two major non-collagenous proteins, bone sialoprotein (BSP) and osteopontin (OPN). These proteins, which are prominently expressed in acellular extrinsic fiber cementum and acellular afibrillar cementum, remain bound to the collagen matrix, and they possess cell attachment properties through the arg-gly-asp (RGD) sequence ( Ganss et al., 1999; Sodek et al., 2000). Both proteins are expressed during early tooth root development by cells along the root surface. Root-surface cells express the BSP, and it is also present in mature teeth. In contrast, OPN is present within the periodontal ligament region of mature teeth. These two proteins are believed to play a major role in the differentiation of the cementoblast progenitor cells to cementoblasts ( Saygin et al., 2000). The BSP is believed to serve adhesion function for root-surface cells and to participate in initiating mineralization. It is chemotactic to pre-cementoblasts and promotes their adhesion and differentiation ( Somerman et al., 1990; MacNeil et al., 1994). The OPN is expressed by many cells during periods of cementogenic activity. It regulates cell migration, differentiation, and survival via interactions with αv β 3 integrin, and it participates in inflammation by regulating monocyte-macrophage activation, phagocytosis, and nitric oxide production ( Giachelli and Steitz, 2000). In teeth and cementum, it may regulate biomineralization by at least two mechanisms: regulating bone cell differentiation, and matrix mineralization. Fibronectin, which is believed to bind cells to the ECM, and tenascin are present in the basement membrane of HERS during odontoblast differentiation, and later at the attachment site of periodontal ligament to the root surface. Other matrix components found in the cementum matrix include osteonectin, which is expressed by cementoblasts producing cellular extrinsic fiber cementum and cellular intrinsic fiber cementum, osteocalcin, and laminin. Several polypeptide growth factors with ability to promote the proliferation and differentiation of putative cementoblasts are sequestered in the cementum matrix. These include BMP-2, -3, and 4, PDGF, a- and bFGFs, TGF-β, and insulin-like growth factor-I (IGF-I) (Table ) ( Cochran and Wozney, 1999; MacNeil and Somerman, 1999; Saygin et al., 2000).
Many of these components are also present in bone; however, molecules unique to cementum have also been described. One of these is an IGF-I isoform referred to as cementum growth factor (CGF). This is a 14-kDa protein, which is larger than IGF-I in molecular size ( Ikezawa et al., 1997). The second molecule is a collagenous protein referred to as cementum attachment protein (CAP). Antibodies to CAP immunostain only cementum and not other periodontal components or other tissues ( Arzate et al., 1992). In bovine tooth germs, the CAP is expressed by cementoblasts, and in cementum its expression pattern is different from that of type I collagen ( Saito et al., 2001). The CAP promotes the adhesion and spreading of mesenchymal cells; however, it promotes the adhesion of mineralized-tissue-forming cells preferentially ( Olson et al., 1991; Pitaru et al., 1995).
The structure and biochemical composition of cementum are affected by several diseases ( Polson, 1986). In periodontitis, the chronic inflammatory process destroys collagen fibers of the gingiva, and this may extend to the root surface. In this disease, one significant pathological change is deposition of bacterial plaque substances, including the bacterial endotoxins. Destruction of connective tissue fibers leads to attachment loss, and damage to cementum becomes irreversible when the cementum surface is exposed to periodontal pockets. Alteration in the biochemical composition of cementum during periodontal disease results in loss of active substances and deposition of inhibitors such as endotoxins. Diseased cementum inhibits connective tissue cell attachment and growth and promotes epithelial attachment ( Terranova and Martin, 1982; Polson, 1986); this was the rationale for new therapeutic approaches in which diseased roots are conditioned to promote connective tissue attachment ( Bartold et al., 2000).
It is perhaps worthwhile to emphasize that certain aspects of cementogenesis and cementum biology (such as attachment mode of cementum to dentin, the rate of cementum apposition) seem to differ between species, most notably between rodents and large mammals, including humans ( Schroeder, 1992; Bosshardt and Schroeder, 1996). It still remains unclear whether these differences are due to spatio-temporal particularities (tooth size, the pace of development, etc.) or whether they mirror different molecular and/or cellular aspects of cementum formation and maintenance. Most of the information available is descriptive, and it is particularly scant with respect to humans. Resolving these issues is important, and, at this time, data derived from animal models must be used with great caution before conclusions for human applications are drawn. The in vitro/in vivo animal and human models developed recently (and described in more detail in the last sections of this review) should be excellent systems to clarify the interspecies differences at the cellular and molecular levels.
Possible Role of Cementum in Periodontal Regeneration
One major goal of regenerative periodontal therapy is new cementum formation and restoration of soft-tissue attachment to the cementum. Cementum regeneration requires cementoblasts, and the origin of cementoblasts and the molecular factors regulating their recruitment and differentiation are not fully understood. In vivo animal models to evaluate cementogenesis during tooth development, the expression pattern of specific matrix molecules, and in vitro experiments studying the effects of cementum components on periodontal cells have provided important clues on how cementum components can regulate cementum regeneration ( McCulloch, 1993; Thesleff and Nieminen, 1996; Ten Cate, 1997; MacNeil and Somerman, 1999; Saygin et al., 2000). During tooth development, dental follicle cells of ectomesenchymal origin appear to be responsible for cementogenesis. In adult mice, the cementoblasts are believed to be recruited as progenitor cells located paravascularly in the periodontal ligament or in endosteal spaces of the alveolar bone ( McCulloch, 1993; Pitaru et al., 1994). Although cementum formation in rodents differs from that in mammals ( Bosshardt and Schroeder, 1996), the observation of Liu et al. (1997) indicates that the periodontal ligament may be one source of cementoblast progenitors in adult humans. These investigators demonstrated that a small proportion of clones of cells cultured from human periodontal ligament form cementum-like mineralized nodules in culture and produce cementum-specific markers ( Bar-Kana et al., 1998). The cementoblasts may also be derived from stem cells present in the periodontal ligament, gingiva, or alveolar bone; this may be the case in healing periodontium, when the pool of available progenitors is likely to be reduced or absent. However, the molecules responsible for recruiting these cells and for their differentiation have not been identified. The mechanisms involved are likely to be more complex.
A variety of chemotactic factors, adhesion molecules, growth factors, and ECM constituents participate together in the recruitment of cementoblast progenitors, their expansion, and differentiation (Table , Fig. 3 ). Many of the same components may be available during periodontal healing; however, most of these molecules are pleiotropic and do not manifest cell specificity. Cell specificity can be achieved in several ways—for example, growth factors targeting specific cell types, unique ECM composition, and conditions permissive for needed cells but refractory to other cells. Evidence indicates that cementum components can regulate cellular activities by all of these mechanisms. For example, cementum contains molecules that promote chemotactic migration, adhesion, proliferation, and differentiation of some periodontal cell types better than others, and these molecules are not detectable in other periodontal structures ( Nishimura et al., 1989; Somerman et al., 1989; Pitaru et al., 1995; Arzate et al., 1996; Ikezawa et al., 1997; BarKana et al., 2000). Adhesion molecules that cause negative selection by excluding unwanted cells are also present in the cementum ( Olson et al., 1991). This cell specificity is not manifested by other ubiquitous matrix molecules, such as fibronectin. Most significantly, the cementum microenvironment contains all the components necessary for cell recruitment, proliferation, and differentiation, and molecules from the circulation are not necessary (Table , Fig. 3 ). For example, cells can escape cell cycle arrest and complete cell division and differentiation in the presence of cementum proteins alone ( Yokokoji and Narayanan, 2001). Needed molecules may also be present in enamel matrix extracts, although in this case the active components may or may not be the same as those of the cementum matrix.
The mechanism by which selection of cementoblast progenitors is achieved is unclear, and it most likely involves specific integrins and signaling events ( Liu et al., 1997; Ivanovski et al., 1999; Saito and Narayanan, 1999; Komaki et al., 2000). Once selected, the progenitor cells adhere to the root surface, and their expansion could be facilitated by growth factors present in the cementum matrix. Growth factors identified in cementum include IGF-I, FGFs, EGF, BMPs, and TGF-β ( MacNeil and Somerman, 1999). Although selection of cells can be achieved at the level of adhesion, it could also be by preferential proliferation. For example, the CGF is an isoform of IGF-I, and fibroblasts mitogenically respond to both growth factors similarly; however, osteoblasts respond better to the CGF ( Ikezawa et al., 1997).
These observations underscore the importance of restoring or providing the cementum microenvironment to initiate and promote new cementum formation. The integrity of cementum is chemically altered by disease due to deposition of bacterial endotoxins, and diseased cementum is removed during therapy. Dentin not covered by cementum undergoes resorption. Root conditioning is not likely to restore the original composition of the cementum local environment, and instead exposes molecules, especially type I collagen, which manifest poor cell specificity ( Polson, 1986). Applying some growth factors is not likely to provide the complete repertoire of the needed molecules, the concentration and type of which change continuously during the healing process. Similarly, while barrier membranes can facilitate population of the site by needed cells, they are not likely to provide the local environment necessary for their differentiation. Further, providing enamel proteins ( Lindhe, 1997; Hirooka, 1998), while likely to be conducive to early cementogenesis, may not provide appropriate environment for recruiting cementoblast progenitors in adults and for their differentiation. ECM components are indeed expressed during periodontal healing ( Lekic et al., 1996; Ivanovski et al., 2000); however, whether all molecules present in cementum matrix are expressed is not known. The temporal sequence of their expression is also important for initiating new cementum formation. These factors may explain why cementum regeneration is not always predictable for the available regenerative procedures.
During inflammation and wound-healing response following periodontal surgery, a battery of growth factors is available from both the circulatory and inflammatory cells. The provisional matrix formed by blood clot and granulation tissue has a composition different from that of the cementum environment under healthy conditions. This combination of growth factors and ECM is not likely to be conducive to the selection and function of cementoblast progenitors. If progenitors are not present and only stem cells are available, their differentiation to cementoblasts may require that events during early cementogenesis be recapitulated, and needed signaling molecules are not likely to be available in the wound-healing environment. Thus, it appears essential that the right combination of ECM components and growth factors is necessary to induce new cementum formation.
One factor that needs to be considered is that, in early and moderate periodontitis, acellular cementum (coronal half of the root) is affected, and the damage extends to cellular cementum in most advanced and furcally positioned lesions. These surfaces are almost always covered by cellular cementum during successful regeneration; whether this is adequate is unclear ( MacNeil and Somerman, 1999).
The growth factors and adhesion molecules present in cementum are also active toward cells of the gingiva, periodontal ligament, and alveolar bone ( Narayanan and Bartold, 1996; Bartold et al., 2000). Therefore, it is possible that cementum components have the potential to participate in the regulation of homeostasis and regeneration of these tissues. However, this is perhaps not significant under healthy conditions, because the growth factors present in cementum remain bound to the cementum matrix. Even if the inflammatory process releases them, their relative concentrations are likely to be less than those available from the blood and inflammatory cells; therefore, contributions by cementum molecules to the regeneration of other periodontal tissues are likely to be marginal or insignificant.
Models of Cementoblastic Cells and Their Potential Application for Developing New Strategies for Periodontal Regeneration
In this section, we will focus on an emerging approach to the development of new treatment modalities for restoration of attachment that involve possible application of cementoblastic cells.
Viable cementoblasts and/or periodontal ligament cells near the cementum appear to play a critical role in the regeneration of the tooth attachment apparatus. It is a well-known fact in the clinic that if an avulsed tooth is replanted into the tooth socket shortly after the avulsion (or the tooth is stored in the conditions that allow for cell survival), cementum-mediated attachment is efficiently re-established. In contrast, if the avulsed tooth is replanted without viable cells present, the healing process is frequently impaired, and severe complications (i.e., ankylosis, root resorption) are more likely to develop ( Boyd et al., 2000). This indicates that viable cementoblasts and/or intact molecules associated with them, in addition to cementum matrix (discussed above), are likely to be actively involved in recruiting cells that next differentiate into cementoblasts and form new cementum that is critical for re-establishing structurally and functionally sound attachment. Thus, the strategy of investigating cementoblastic cells as a possible source of factors that are specifically involved in cementum regeneration/healing has a clear basis in a well-known clinical observation.
In recent years, several laboratories have reported successful isolation and propagation of cells (from both animal and human sources) exhibiting an apparent cementoblastic phenotype in vitro ( D’Errico et al., 1995, 1997; Grzesik et al., 1998). These cells have been shown, reproducibly, to form histologically proven cementum-like tissue following in vivo transplantation into immunodeficient mice ( Grzesik et al., 1998, 2000; Handa et al., 2002). One of the most important benefits of the establishment of such combined in vitro/in vivo models is that they will allow for elucidation of the relationship between osteoblasts and cementoblasts. In contrast to bone, very little is known regarding the specific mechanisms involved in the maintenance of cementum structure and function in humans on the cellular and molecular level. Most of the information available is an extension of what is known for bone, and the general assumption is that cementoblasts’ physiology closely follows that of osteoblasts. This may well be true in a broad sense; however, given the clear differences in the structure and function between these two tissues, it is conceivable that different factor(s) are directing the distinct aspects of differentiation of osteoblasts and cementoblasts, respectively. This paradigm is of particular significance for the development of strategies for regenerating cementum with cell-based therapies (see below). Bone, in contrast to cementum, undergoes constant remodeling carried out by osteoclasts and is accompanied by bone marrow. The deposition of bone on the root surface by osteoblasts or osteoblastic precursors (either delivered or targeted to the tooth root surface) is therefore likely to result in the recruitment of osteoclasts, followed by the development of bone marrow; this, in turn, would significantly alter the local environment in periodontium and may result in root resorption, ankylosis, and subsequent loss of attachment. Thus, novel in vitro/in vivo models offer promise toward developing relatively simple and faithful means of recruiting cells with cementoblastic potential and their differentiation and excluding unwanted osteoblastic precursors. Last, the use of these systems allows for direct in vivo experimentation using human cells. This may be particularly important, because cementum has been reported to differ significantly between species in some aspects of its physiology ( Schroeder, 1992; Bosshardt and Schroeder, 1996).
The fact that cells with a cementoblastic phenotype can be cultured also offers the
possibility for development of a more radical treatment of cases where most of the
cementum in situ as well as adjacent soft tissues has been lost due to
pathology. In such a scenario, re-establishing the proper microenvironment (discussed
elsewhere) will not be sufficient to achieve regeneration, due to the lack (or low
abundance) of available precursors/progenitors of cementoblasts in the affected area. In
principle, cementogenic cells can be isolated from a relatively small specimen, expanded
in culture, and subsequently re-transplanted to the same patient. Primary human
cementoblasts can be efficiently grown from dissected fragments of healthy tooth root
cementum treated with collagenase. The culture conditions are fairly standard—it is
sufficient to maintain these cells in standard tissue culture medium (alpha minimal
essential medium [MEM], Dulbecco’s MEM/F12) supplemented with 10% fetal bovine serum
( Grzesik et al., 1998, 2000). In this manner, it is feasible to obtain large
numbers (in the range of 108-109) of committed cementogenic cells
from a single tooth. However, this system has a significant drawback because of the
requirement that healthy teeth be extracted for the cultures to be established. Also, it
is still not clearly established whether cementogenic cells can be reproducibly
obtained, by this approach, from teeth of aged and/or diseased patients. These
reservations somewhat deter the enthusiasm for the cell-transplantation approach. Such a
scenario would be undoubtedly more feasible if cells with cementogenic potential (or
cells that can be induced toward cementoblastic differentiation) could be obtained from
sources other than cementum, such as soft periodontal tissues (periodontal ligament,
gingiva) and/or even remote sites (bone marrow, fat). The use of sources other than
cementum for cell isolation, however, may require two critical conditions: Exclusion of unwanted cell lineages; this can be accomplished by the selection
of specific cells before culturing, selection during culture expansion, and/or
selection prior to delivery to the defect. Although molecules such as CAP show
some promise ( Liu et al.,
1997; BarKana et al.,
2000), currently, no defined molecules are known to discriminate
between cementogenic and non-cementogenic but osteogenic precursors. Inducing differentiation along the cementoblastic lineage—this can be
accomplished by pre-treatment of cells in vitro or in
vivo (i.e., the application of an inducing
stimulus concomitantly with cell delivery to the patient), or both, with a
specific inducing factor. At this time, however, no defined molecular factor
specifically inducing the cementoblastic phenotype is known.
Obviously, at this time, the application of cell therapy for treating cementum loss using cells isolated from soft tissues remains a highly hypothetical option, because methodology to satisfy these two conditions is not available. The first steps would be, then, to solve these two major problems relating to selecting cells with a proper differentiation potential and to test the general feasibility of this complex approach. Clearly, the recently developed in vitro/in vivo combined models of cementogenesis will be of great use for these purposes. In principle, the effect of any factor and its relation to the cementoblastic phenotype can be meaningfully studied in both the in vitro and in vivo conditions. It cannot be excluded that some of the already-known molecules, including pharmacological agents, are indeed capable of inducing cementoblastic differentiation—for example, a well-known immunosuppressant, cyclosporin A, has been shown to induce development of cementum in the gingival tissue of rats ( Ayanoglou, 1998). Similarly, the in vitro/in vivo system can also be used for further elucidation of the modes of action of some of the developed and currently tested compounds (such as Emdogain) with apparent cementum-/growth-promoting activities.
Once the molecules critical for cementum regeneration are identified, an efficient and relatively simple delivery system must be developed if they are to be used for periodontal regeneration. One of the major problems of applying bioactive molecules in in vivo settings is that they need to be present in a specific location and remain bio-available and bioactive for extended periods of time. For cementum regeneration, its anatomy—i.e., the fact that cementum is a mineralized "interface" tissue connecting mineralized dentin and non-mineralized fibrous periodontal ligament—implies that a successful delivery system should somehow reconcile these two distinct environments. An additional complication is that regenerated cementum must be strictly confined only to the tooth root surface, without compromising the integrity and structure of the periodontal ligament. One approach would be to target and bind bioactive molecules to the mineralized surface of the tooth root. Although some of the growth/differentiation factors, such as BMPs, have affinity to mineral of bones and teeth, many do not display this preference. The lack of efficient and stable binding of biologically active molecules to mineralized tooth root surfaces is likely to account for, at best, moderate and unpredictable therapeutic effects of many of those substances tested so far ( Caffesse et al., 1991; Choi et al., 1993).
In the last decade, several proteins have been found to bind mineral with high affinity, and this activity has been firmly linked to acidic motifs (poly-aspartate and poly-glutamate "stretches") present in these molecules. It is possible to utilize fusion proteins consisting of a bioactive part(s) (corresponding to factor[s] promoting cementoblastic differentiation, such as the CAP), and the mineral-binding domain derived from a different protein could allow for efficient targeting to the tooth root surface.
Integrin-matrix interactions are fundamental to most aspects of cellular metabolism and are especially important for the healing/regeneration and maintenance of the structure and proper functions of tissues and organs. Importantly, there is a clear positive correlation between integrin ligation and cell survival ( Meredith et al., 1996). For most integrin ligands, the receptor-binding domains have been mapped, and synthetic peptides, corresponding to these sequences, are biologically active. Several integrins as well as their natural ligands (e.g., collagens, BSP, OPN, CAP) are known to be expressed by cementoblastic cells ( Steffensen et al., 1992; Ivanovski et al., 2000), implying that they may be important regulators of cementogenesis. This also indicates that targeting integrin receptors may be beneficial for stimulating cementum regeneration. For example, in our laboratory, we have recently established that synthetic peptides containing the integrin-binding RGD sequence and the polyglutamate domain derived from BSP can be efficiently bound to synthetic hydroxyapatite/tricalcium phosphate ceramic. When such ceramics were next used as a vehicle for transplanting cementogenic cells into immunodeficient mice, cementum formed more efficiently and abundantly than when unaltered ceramics were used (Grzesik et al., unpublished data). This result indicates that integrins may be directly involved in some aspects of cementum formation and that increasing the numbers of active adhesive integrin ligands on mineralized surfaces enhances formation of cementum by committed cementogenic cells. Currently, we are evaluating whether this approach can be utilized for directing bioactive peptides to naturally occurring mineral matrices of bone and dentin in an experimental setting.
In summary, these are still early days of periodontal tissue regeneration via cell- and matrix-based therapies. Nevertheless, recent developments in basic science indicate that these approaches are undoubtedly feasible and, given their promise, worth exploring.
Stages in the differentiation of a stem cell. Embryonic stem cells are nearly totipotent, whereas a stem cell in an adult tissue may be pluripotent (for example, hematopoietic stem cell) or unipotent (epithelial cell, for example). An adult tissue may contain the stem cells and precursor cells separated from full differentiation by one or several steps. The periodontal ligament has been shown to contain precursor cells or progenitors to cementoblasts ( Pitaru et al., 1994). The differentiated cell may also consist of subpopulations of the same cell type ( McCulloch and Bordin, 1991; Fries et al., 1994).
Hypothetical model for maturation into fully differentiated cell. Under healthy conditions, pre-existing healthy matrix recruits precursor cells that are close to final differentiation and permits their differentiation to occur. During inflammation and wound repair, undifferentiated "stem" cells, presumably separated by several stages from full differentiation, interact with growth factors (GF) and available matrix, differentiate to the next stage, and produce matrix. This cell interacts with the matrix, which contains the ECM produced by the previous stage cell, and differentiates to the next stage. This process continues until the cell becomes fully differentiated.
Cell activities required for new cementum and attachment formation and cementum components that possibly regulate these processes. The list of molecules is not complete, and only some examples are given.
Footnotes
Acknowledgements
This work was supported by NIH grants DE-13061, DE-08229, DE-10491, DE-13475, and DE 14016.