Immunohistochemical localization of periostin in tooth and its surrounding tissues in mouse mandibles during development

Authors

  • Hironobu Suzuki,

    Corresponding author
    1. Division of Oral Anatomy, Department of Oral Biological Science, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan
    2. Division of Periodontology, Department of Oral Biological Science, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan
    • Division of Oral Anatomy, Department of Oral Biological Science, Niigata University Graduate School of Medical and Dental Sciences, 2-5274 Gakkocho-dori, Niigata 951-8514, Japan
    Search for more papers by this author
    • Fax: 81-25-223-6499

  • Norio Amizuka,

    1. Division of Oral Anatomy, Department of Oral Biological Science, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan
    2. Center for Transdisciplinary Research, Niigata University, Niigata, Japan
    Search for more papers by this author
  • Isao Kii,

    1. Department of Life Science, Faculty of Biological and Biotechnology, Tokyo Institute of Technology, Yokohama, Japan
    Search for more papers by this author
  • Yoshiro Kawano,

    1. Division of Oral Anatomy, Department of Oral Biological Science, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan
    Search for more papers by this author
  • Kayoko Nozawa-Inoue,

    1. Division of Oral Anatomy, Department of Oral Biological Science, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan
    Search for more papers by this author
  • Akiko Suzuki,

    1. Division of Oral Anatomy, Department of Oral Biological Science, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan
    Search for more papers by this author
  • Hiromasa Yoshie,

    1. Division of Periodontology, Department of Oral Biological Science, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan
    Search for more papers by this author
  • Akira Kudo,

    1. Department of Life Science, Faculty of Biological and Biotechnology, Tokyo Institute of Technology, Yokohama, Japan
    Search for more papers by this author
  • Takeyasu Maeda

    1. Division of Oral Anatomy, Department of Oral Biological Science, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan
    2. Center for Transdisciplinary Research, Niigata University, Niigata, Japan
    Search for more papers by this author

Abstract

Previous reports have shown expression of immunoreactivity for periostin, originally identified as osteoblast-specific factor-2, in the periosteum and periodontal ligament. However, the developmental changes in its expression and the detailed immunolocalization have remained veiled. The present study was undertaken to examine the spatiotemporal expression of this protein in teeth and their associated tissues of mice during development at light and electron microscopic levels. In tooth germs at cap stage, periostin immunoreactivity was recognizable in the interface between inner enamel epithelium and preodontoblasts as well as in the mesenchymal tissues around cervical loop. Dental follicles around tooth germs at bell stage localized periostin immunopositivity in addition to the immunopositive areas observed in cap-staged tooth germs, although the functional significance of periostin has remained unclear in tooth development. Furthermore, periostin immunoreactivity was also found in the alveolar bone surface. In the incisors of both 7- and 21-day-old mice, immunoreaction for periostin was discernible in the lingual periodontal ligament and labial fibrous tissue adjacent to the papillary layer. After postnatal day 7, immunoreaction for periostin came to be restricted to the fibrous bundles in the periodontal ligament in accordance with the organization of the periodontal fibers, indicating its localization matched the morphogenesis of the periodontal ligament. Immunoelectron microscopic observation of the mature periodontal ligament verified the localization of periostin between the cytoplasmic processes of periodontal fibroblasts and cementoblasts and the adjacent collagen fibrils. Our findings suggest that periostin is involved at the sites of the cell-to-matrix interaction, serving as adhesive equipment for bearing mechanical forces, including occlusal force and tooth eruption. © 2004 Wiley-Liss, Inc.

Periostin consisting of 811 amino acids, initially identified as osteoblast-specific factor-2, is a disulfate-linked 90 kDa protein secreted by osteoblasts (Takeshita et al., 1993). This protein, however, was renamed periostin, both to avoid its confusion with a transcription factor that also used this acronym, known as Cbfa1 (Ducy et al., 1997), as well as on account of predominant localization in the periosteum (Horiuchi et al., 1999). Since one of the biological functions in periosteum is the bearing of mechanical force, periostin was predicted to act at the site of cell-to-cell and/or cell-to-matrix interactions (Horiuchi et al., 1999). An analysis of the amino acid sequence has revealed a putative complex protein structure with four repeats of a characteristic domain (Horiuchi et al., 1999). The amino acid sequence of periostin has a high homology with the Fasciclin I that is expressed in the central nervous system of insects, serving in neural cell-to-cell contact (Bastiani et al., 1987; Zinn et al., 1988). Therefore, periostin has been regarded as a member of the Fasciclin I family, which includes βig-h3 (Skonier et al., 1992), Algal-CAM (Huber and Sumper, 1994), and midline fasciclin (Hu et al., 1998), mimicking a function of Fasciclin I.

The periodontal ligament anchoring the tooth to alveolar bone is a dense collagenous tissue, which is usually exposed to various mechanical stress, including occlusal force, during mastication, resulting in an active tissue remodel. This tissue consists of heterogeneous cell populations, including fibroblasts, osteoblasts, cementoblasts, macrophages, and vascular endothelial cells as well as various extracellular matrices. The extracellular matrix plays a pivotal role in maintaining local histological architectures for a variety of tissues. Many extracellular matrix-associated proteins, including connective tissue growth factor (Moussad and Brigstock, 2000), nephroblastoma overexpressed (NOV) (Perbal, 2001), vitronectin, and fibronectin, may regulate cellular activities such as cell adhesion, migration, mitogenesis, differentiation, survival, angiogenesis, wound healing, and tumorgenesis (Moussad and Brigstock, 2000). In addition to periosteum, the exclusive localization of periostin has been formed in the periodontal ligament of adult mice by in situ hybridization and immunohistochemistry (Horiuchi et al., 1999, Wilde et al., 2003), indicating a possibility that periostin interacts with these extracellular matrices in the periodontal ligament. Furthermore, periostin has been regarded as a fetal protein that is involved in the morphogenesis and subsequent development of various tissues including bone and periodontal tissues (Kruzynska-Frejtag et al., 2001, 2004; Ito et al., 2002). In contrast to the periodontal ligament, little information is available regarding periostin expression in tooth both at the mature and the developmental stage. However, to our knowledge, detailed localization of this protein remains unclear in the developing teeth.

The present study was thus undertaken to examine the spatiotemporal localization of periostin in developing mouse teeth and their surrounding tissues by immunocytochemical techniques to provide evidence for verifying the biological functions of this protein. Furthermore, immunoelectron microscopy was employed to demonstrate the localization of periostin immunoreactivity in the periodontal ligament of mature mice.

MATERIALS AND METHODS

The care and use of animals followed the Guiding Principles for the Care and Use of Animals, approved by Niigata University in accordance with the principles of the Helsinki Declaration.

Tissue Preparation

Under anesthesia with an intraperitoneal injection of chloral hydrate (400 mg/100 g body weight), 7-, 10-, 14-, and 21-day-old SPF/VAF mice and unborn mice at 17 days of gestation were perfused through the left ventricle either with 4% paraformaldehyde in a 0.1 M phosphate buffer (PB; pH 7.4), or with a mixture of 4% paraformaldehyde and 0.0125% glutaraldehyde in 0.067 M PB for light microscopy or immunoelectron microscopy, respectively. Mandibles were immediately removed and immersed in the same fixatives for an additional 8 hr at 4°C. The specimens were decalcified with 10% EDTA-2Na (ethylene diamine tetraacetic acid disodium salt) solution at 4°C and equilibrated with a 30% sucrose solution overnight for cryoprotection. Cryostat sections were prepared and mounted onto glass slides precoated with 3-aminopropyltriethoxy silane.

Immunohistochemistry for Periostin

Cryostat sections were processed for the avidin-biotin complex method according to Hsu et al. (1981). After the inhibition of endogenous peroxidase activity with absolute methanol containing 0.3% hydrogen peroxide for 30 min, they were primarily incubated with a rabbit polyclonal antiserum against periostin (Horiuchi et al., 1999) at a dilution of 1:100 for 2–3 hr at room temperature. Sections were then incubated with a biotinylated goat anti-rabbit IgG (1:1,000; Vector Laboratories, Burlingame, CA) and subsequently with an avidin-peroxidase complex (ABC kit; Vector) for 60 min each at room temperature. The antigen-antibody reaction sites were made visible by incubation with 0.04% 3-3′-diaminobenzidine and 0.03% hydrogen peroxide. After counterstaining with 0.03% methyl green, the sections were dehydrated through an ascending series of ethanol and coverslipped with Permount (Fisher Scientific, Springfield, NJ).

For immunoelectron microscopy, three additional adult mice (6- and 7-week-old) were perfused in the same fixative mentioned above. The decalcified cryostat sections were immunostained and postfixed with 1% osmium tetraoxide reduced with 1.5% potassium ferrocyanide in a 0.1 M cacodylate buffer for 3 hr at 4°C and were dehydrated in ascending acetones prior to embedding in epoxy resin (Epon 812, Taab, Berkshire, U.K.). Ultrathin sections were prepared with an ultramicrotome and examined with a Hitachi H-7000 transmission electron microscope (TEM; Hitachi, Tokyo, Japan) following brief staining with uranyl acetate and lead citrate.

Immunocytochemical Control

Immunocytochemical controls were performed by replacing the primary antiserum with normal nonimmune rabbit serum or 0.01 M phosphate-buffered saline and omitting biotinylated anti-rabbit IgG or avidin-peroxidase complex. These sections did not show any immunoreactivity for periostin. The detailed characterization of this antiserum has been previously described (Horiuchi et al., 1999).

RESULTS

Localization of Periostin in Mandibles of 17-Day-Old Fetuses

The periosteum associated with the developing mandible strongly immunoreacted with the antiserum against periostin. At this stage, the tooth germs of first and second molars were at bell and cap stages, respectively. Intense periostin immunoreactivity was discernible in the tooth germ at cap stage (Fig. 1A); the interface between inner enamel epithelium and preodontoblasts, where neither enamel nor dentin had formed, showed periostin immunoreaction (Fig. 1B). Furthermore, the mesenchymal tissues around the cervical loop also exhibited periostin immunoreactivity (Fig. 1A and C).

Figure 1.

The localization of periostin in the mandible of 17-day-old fetuses. A: The second molar of the mandible of 17-day-old mouse fetus. The tooth germ is at cap stage. B: A closer view of the boxed area in A. The periostin immunoreaction is recognizable at the interface between inner enamel epithelium (iEP) and preodontoblasts (PO). C: A higher magnification of the boxed area in A. The mesenchymal tissues around the cervical loop also show periostin immunoreactivity. Note absence of immunoreaction in the cellular elements. Original magnification: A, 180×; B, 1,200×; C, 1,200×. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com].

In the tooth germ at bell stage, the dental follicles surrounding the tooth germs also displayed periostin immunoreaction (Fig. 2B). The immunoreactivity for periostin at bell stage demonstrated the same distribution pattern as to that observed in the cap-staged tooth germ; the interface between the more differentiated inner enamel epithelium and preodontoblasts exhibited periostin immunoreaction (Fig. 2C). Moreover, stratum intermedium showed a very faint immunoreaction for periostin. However, cellular elements in the tooth germ including inner and outer enamel epithelial cells as well as preodontoblasts lacked periostin immunoreaction (Fig. 2C). The surface of the surrounding alveolar bone also exhibited immunopositivity for periostin (Fig. 2D). The mesenchymal tissues around the apical region of the epithelial root sheath, that is, the epithelial diaphragm, localized a weak but specific immunoreactivity of periostin (Fig. 2D). In spite of our careful observation, we failed periostin immunoreactivity in the cellular elements in the mesencymal tissues (Fig. 2E). However, the connective tissue interconnecting between the epithelial root sheath and opposing alveolar bone was devoid of any periostin immunoreaction (Fig. 2D).

Figure 2.

The localization of periostin in the mandible of 17-day-old fetuses. A: Lower magnification of the mandible of 17-day-old mouse fetus. The periostin immunoreaction, colored brown, is localized in tissues surrounding the tooth germ at bell stage. B: A closer view of the boxed area in A. The immunoreactive dental follicles are seen to surround a tooth germ of molar. C: Higher-magnified image of the cuspal region and enamel organ. A dental follicle (DF) shows strong staining (arrows) indicative of periostin, while there is a weak reaction in the external soft tissues. Note immunopositivity neither in the outer (oEP) nor inner (iEP) enamel epithelium. Arrowheads indicate immunoreaction between iEP and preodontoblasts. Moreover, the stratum intermedium (SI) shows a very faint immunoreaction for periostin. D: Immunoreaction for periostin is observed in the cell layer (arrows) adjacent to the surface of the alveolar bone. The region of the epithelial root sheath (arrowheads) is immunostained with periostin. No apparent positivity is recognizable in the region intervening between the root sheath and the alveolar bone. E: Higher-magnified image of the area as indicated by arrowheads in D. The extracellular matrix shows the periostin immunoreactivity at the apical region of the epithelial root sheath. Original magnification: A, 30×; B, 70×; C, 200×; D, 180×; E, 360×. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com].

Periostin in Mandibles of 7-Day-Old Mice

The mesenchymal tissue adjacent to the labial and lingual cervical loops in the incisal apices of mandibles showed a weak immunoreaction for periostin (Fig. 3A–D). Its immunoreaction was not localized in the cellular elements, but in the extracellular matrix (Fig. 3C). The periostin peptide was uniformly localized in the lingual periodontal ligament, including both the tooth half and alveolar half of the ligament, which were termed the tooth- and alveolus-related parts, respectively (Fig. 3E) (Beertsen et al., 1974). These immunoreactions exclusively appeared as the filamentous elements in the periodontal ligament. Although the osteoblasts on the alveolar bone expressed intense periostin immunoreaction, neither the cementum nor dentin had its immunoreaction at light microscopic level (Fig. 3E). However, periostin immunoreaction could depict the outlines of cementoblasts appeared positive (Fig. 3E). In the labial connective tissue of incisors, the fibrous tissue adjacent to the papillary layer displayed periostin immunopositivity (Fig. 3F). However, only a weak immunoreaction was recognizable in the spatially matched region adjacent to the stratum intermedium of ameloblasts (Fig. 3G). In contrast, the connective tissue facing the alveolar bone localized an intense periostin immunoreaction (Fig. 3F and G). All stages of ameloblasts and odontoblasts failed to show any specific immunoreaction for periostin (Fig. 3A and E–G).

Figure 3.

Periostin in the mandibles of 7-day-old mice. A: Low-powered magnification of a part of the mandible of a 7-day-old mouse. Periostin, indicated by a brown color, surrounds a partial incisor. B: Highly magnified image of the labial cervical loop. Faint immunoreactivity (an arrow) is found in the tissue surrounding the cervical loop. C: A closer view of the boxed area in B. The periostin immunoreaction is recognizable on the border between preameloblasts and mesenchymal tissue. D: The lingual cervical loop is covered with tissues (arrows) weakly positive for periostin. E: The localization of periostin (brown color) in the region of the lingual periodontal ligament (PL) of incisor. Periostin immunoreactivity is exclusively localized in the periodontal ligaments, especially in the filamentous elements, whereas no periostin is observable in cementum (Ce), dentine (D), or odontoblast layers (Od). However, periostin immunoreaction is observed around the cementoblasts (arrows). F: Higher magnification of the labial connective tissues adjacent to the papillary layer and ameloblasts. Periostin immunopositivity is localized in fibrillar structures (asterisk) in parallel to the incisal axis, the neighboring papillary layer, and in the soft tissues (an arrow) close to the alveolar bone. G: Higher magnification of labial connective tissues adjacent to the stratum intermedium and ameloblasts. Unlike E, connective tissues (asterisk) adjacent to the stratum intermedium show the poor immunoreactivity for periostin despite its presence (an arrow) close to the alveolar bone. H: Lower magnification of the first molar. Note the immunopositive dental follicle encapsulates the molar. I: At the region of the epithelial sheath of Hertwig, periostin is discernible in the cell layer covering the alveolar bone and dental follicle. J: At the coronal region, the first and the second molars facing intermediating the immunopositive dental follicle (arrows). K: When observed in the cervical region of the tooth germ, the immunopositive cell layer adjacent to the alveolar bone and dental follicle appears integrated. Original magnification: A, 13×; B, 90×; C, 540×; D, 110×; E, 280×; F, 400×; G, 400×; H, 45×; I, 230×; J, 160×; K, 400×. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com].

In the molars at postnatal 7 days, the immunopositive dental follicle and the external soft tissues encapsulated the tooth germ of the first molar (Fig. 3H). The periostin-immunoreactive cell layer, including osteoblasts, covered the surface of the peripheral alveolar bone (Fig. 3I). Dental follicles running in narrow spaces between the coronal regions of the first and the second molars displayed periostin immunopositivity (Fig. 3J), indicating continuity with the immunoreactive cell layer covering the alveolar bone. Furthermore, dental follicles close to the epithelial sheath of Hertwig also exhibited a periostin immunoreaction (Fig. 3K).

Periostin in Mandibles of 10-Day-Old Mice

By 10 days after birth, root formation had actively proceeded in the first molar, though it was not exposed to the oral cavity (Fig. 4A). The dental follicles and external soft tissues with strong periostin immunoreaction encompassed the tooth crown (Fig. 4A–C). Observation at higher magnification showed periostin immunoreaction exclusively in the extracellular matrix, not in the cellular elements, in the dental follicle (Fig. 4B and C). Both the connective tissue facing the alveolar bone beneath the bifurcation area of tooth roots, corresponding to a future interradicular septum, and the alveolar bone demonstrated a weak immunopositivity for periostin (Fig. 4A). Consistent with the findings in the 7-day-old mandible, the cell layer surrounding interalveolar septum was intensely positive in periostin immunoreaction (Fig. 4B). Furthermore, a faint immunopositive line separated condensed cells of the dental pulp and cells of the surrounding tissue (Fig. 4D). In the interfacing region between the tooth germ and surrounding alveolar bone, a bundle of the immunopositive fibrous tissues ran obliquely from the cementum toward the opposing alveolar bone, presumably identical to future periodontal ligament fibers (Fig. 4E).

Figure 4.

Periostin in the mandibles of 10-day-old mice. A: An image of a whole tooth germ stained with antiserum against periostin at 10 days after birth. An intensely immunoreactive tissue including dental follicles encapsulates the tooth germ. B: At the cervical region of the first and the second molars, periostin immunoreactive fibrillar tissues are observed on the alveolar bone and appear to come together and connect the immunopositive dental follicles. C: A closer view of the boxed area in B. The extracellular matrix is not cellular elements immunopositive for periostin. D: Periostin is found in the cell layer adjacent to both the alveolar bone and the root sheath. Note that a faint immunopositive line (arrows) separates condensed cells of the dental pulp and cells of the surrounding tissue. E: Fibrillar connective tissues positive for periostin are observed to run obliquely from the region of the cementum toward the opposing alveolar bone. Original magnification: A, 50×; B, 140×; C, 700×; D, 200×; E, 140×. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com].

Periostin in Mandibles of 14-Day-Old Mice

The first molar was not exposed to oral cavity at this stage and the apical foramen remained open widely at this stage (Fig. 5A). In accordance with the root formation, alveolar bone of the interradicular septum also had been formed by this stage (Fig. 5A). The periostin immunopositive tissue covered the interalveolar and interradicular septa, as did the ridge region of the interradicular alveolar bone, where the immunoreaction was weak except for the apical region of the tooth roots (Fig. 5A and B). Immunopositive fiber bundles obliquely extended from the mid region of tooth roots to the interradicular alveolar bone (Fig. 5B); the direction of the fibrous tissues appeared to be suitable for generating eruption force when these tissues would contract. The apical region displayed faint immunoreaction for periostin in the cells on the epithelial diaphragm as well as at the border between the dental pulp and periodontal ligament (Fig. 5C).

Figure 5.

Periostin in mandibles of 14-day-old mice. A: Whole image of the first molar. B: Parts of the periodontal ligaments interconnecting the interradicular septum localize periostin (brown color). In contrast, the entire periodontal ligaments associating with the interalveolar septum are immunoreactive for periostin. The immunopositive fibers run obliquely from the cementum of tooth roots to the middle height of the interradicular septum. Note the immunopositive fibrous structure identical to collagen bundles. C: The apical region of the tooth root shows very faint periostin in cells on the epithelial diaphragm and on the boundary between condensed cells in the dental pulp and the surrounding tissue. Original magnification: A, 40×; B, 135×; C, 135×. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com].

Periostin in Mandibles of 21-Day-Old Mice

In the incisal apex of mandibles, periostin immunoreactivity was recognizable at the developing sites of the labial and lingual cervical loops, i.e., developing sites, where the epithelium-mesenchymal interaction is known to take place (Fig. 6A–E), as observed at the previous stages. A closer view of the labial and lingual developing sites demonstrated the localization of periostin immunoreaction in the extracellular matrix, not in the cellular elements (Fig, 6B–E). The periodontal fibers running between the alveolar bone and incisor in the alveolus-related part showed immunopositivity for periostin, but the fibers running in parallel to the tooth axis lacked its immunoreaction (Fig. 6F). Periostin was broadly and intensely localized in the tooth-related part of the lingual periodontal ligament (Fig. 6F). In contrast, the labial connective tissue adjacent to the papillary layer localized periostin (Fig. 6G), but the matching region facing the stratum intermedium demonstrated poor immunoreaction (Fig. 6H). Throughout this investigation, odontoblasts and ameloblasts did not exhibit any specific immunoreaction for periostin (Fig. 6A, G, and H).

Figure 6.

Periostin in mandibles of 21-day-old mice. A: A low-powered image of the incisor of a 21-day-old mouse. BE: The boundary between the mesenchymal tissues and the labial (B and C) and lingual (D and E) cervical loops displays a weak immunoreaction for periostin. F: The fibers vertical to the incisal axis in the alveolus-related part of the lingual periodontal ligaments are positive for periostin. G: The labial connective tissue adjacent to the papillary layer shows the localization of periostin. H: The matching region facing the apical stratum intermedium shows the weak immunoreaction. I: A low-powered image of an entire first molar. J: Despite the broad distribution of periostin in the periodontal ligaments at this stage, the apical region lacks periostin immunoreactivity. Original magnification: A, 13×; B, 120×; C, 420×; D, 160×; E, 460×; F, 360×; G, 340×; H, 280×; I, 25×; J, 120×.

By day 21, the eruption and root formation of the first lower molar had almost finished, though the apical foraman remained unclosed (Fig. 6I). The periodontal ligament passing between the tooth root in either the interradicular or interalveolar septa revealed periostin immunoreactivity (Fig. 6I). Despite the broad distribution of periostin, however, the apical region lacked any periostin immunoreactivity except for weakly positive cells on the boundary between the dental pulp and periodontal ligament (Fig. 6J).

Immunoelectron Microscopic Observation on Periodontal Ligament of Mature Mice

Immunocytochemistry at the transmission electron microscopic level verified the localization of immunoreaction products for periostin on the cytoplasmic processes of fibroblasts in the periodontal ligament (Fig. 7A). The electron-dense deposits reactive to periostin were largely confined to their cell membranes tightly associated with the bundles of surrounding collagen fibers (Fig. 7C). A faint and dispersed immunoreactivity was discernible in the narrow area among collagen fibrils (Fig. 7B). Immunoreactive products for periostin were also localized on the cell membrane of the cytoplasmic processes of the cementoblasts, as identical to the periodontal fibroblasts (Fig. 7C). The immunolocalization of periostin observed in this study was summarized in Table 1.

Figure 7.

Immunoelectron microscopic image of periostin. A: Immunoreactive products to periostin, electron-dense deposits, are observed on the cytoplasmic processes (arrows) of periodontal fibroblastic cells (F). Note that the immunopositivity does not appear evenly on the entire surface of the fibroblastic cells. B: Higher-magnified image of the asterisk seen in A. The interspaces of collagen bundles close to the immunopositive cytoplasmic process are faintly positive for periostin. C: A cementoblast (Ce) shows the periostin immunopositivity on the cell membrane of the cytoplasmic processes (arrows) as well as in the interspaces of collagen bundles (arrowheads) close to the cytoplasmic processes, as was seen in the periodontal fibroblasts. Original magnification: A, 1,630×; B, 5,410×; C, 4,380×.

Table 1. Periostin expression of the mandibular molar during development
 CapBell7 days10 days14 days21 days
Outer enamel epithelium    
Inner enamel epithelium or ameloblasts  
Stellate reticulum    
Stratum intermedium+    
Interface between iEP and preodontoblasts++    
Preodontoblasts or odontoblasts
Dental papilla or pulp
The mesenchymal tissues around cervical loop or epithelial root sheath+++   
Periodontal ligament +++
Alveolar bone surface +++++
Dental follicle ++++ 

DISCUSSION

The present study confirmed the previous finding of intense and constant expression of immunoreaction for periostin in the periodontal ligament of 21-day-old mice (Horiuchi et al., 1999). It further succeeded in demonstrating the localization of periostin in developing teeth and their surrounding tissues, including the periodontal ligament, indicating that this protein has a biological function in the morphogenesis and subsequent development of various tissues. Interestingly, the cervical loop and dental follicles of the molar tooth germs in this study had periostin immunoreactivity, which then disappeared in advance of tooth development. This change in the expression pattern of periostin may be explained by a previous suggestion that periostin is produced as a kind of fetal protein (Kruzynska-Frejtag et al., 2001, 2004; Ito et al., 2002).

The present immunoelectron microscopy showed the immunolocalization of periostin in the cell membrane of the cytoplasmic extensions of periodontal fibroblasts, but neither in their cytoplasm nor the ground substance of the mature periodontal ligament. Periostin situates itself at sites of close association between cells, including the periodontal fibroblasts and cementoblasts and the extracellular components in the periodontal ligament, comparable with our previous report on the periosteum of adult mouse femurs (Hirose et al., 2003). This immunolocalization pattern may indicate that periostin modulates the remodeling and metabolism of extracellular matrices by mediating a cell-to-matrix interaction. Previous investigations on skeletal tissues have suggested the involvement of periostin in the cell-to-matrix interaction by evidence of the predominant localization of periostin in dense connective tissue such as periosteum, which is subjected to the physical force of exercise (Horiuchi et al., 1999; Hirose et al., 2003; Wilde et al., 2003). In addition, an analysis of the amino acid sequence of periostin predicted that periostin behaves as a kind of adhesion molecule (Takeshita et al., 1993). Thus, it is reasonable to consider that the attachments between these cells and collagen fibers via periostin might serve as an adhesive device for bearing mechanical stress in mature tissues, giving rise to the strength and rigidity of this tissue. The idea that periostin expression reflects a resistance against mechanical forces such as occlusal forces and/or tooth eruption loaded onto the periodontal ligament is supported by the following three findings obtained in this study. First, the labial connective tissue and lingual periodontal ligament showed different expression patterns of periostin immunoreactivity in mice incisors. Since the collagen fibers run in parallel to the tooth axis, referred to as vertical fibers in the labial connective tissue, it does not function as a tooth anchor. Second, only collagen fibers transversely running between the alveolar bone and incisor had periostin immunoreaction in the alveolus-related part, while the vertical fibers there lacked it. Third, periostin immunoreaction appeared in accordance with the organization of the collagen bundles in the periodontal ligament in molars.

The extracellular matrix plays a pivotal role in maintaining local histological architectures for a variety of tissues, including bone and periodontal ligament. Many extracellular matrix-associated proteins have been reported to regulate cellular activities such as cell adhesion, migration, mitogenesis, differentiation, survival, angiogenesis, wound healing, and tumorgenesis (Moussad and Brigstock, 2000). The turnover of extracellular matrices in the periodontal ligament appears higher than in other collagenous tissues due to mechanical stress, including occlusal forces and/or tooth eruption [cf. Schroeder (1986)]. These facts strongly suggest that periostin modulates the remodeling and metabolism of extracellular matrices by mediating a cell-to-matrix interaction. In addition to the well-organized periodontal ligament, we found periostin immunoreaction in the dental follicles of developing teeth as well as in the developing sites of incisors, signifying that this protein could be widely distributed in various fetal tissues in which the active remodeling of extracellular matrices takes place. Periostin has been reported to come to be highly restricted to the periosteum with aging, though it is earlier widespread in various fetal tissues (Kruzynska-Frejtag et al., 2001, 2004; Ito et al., 2002; Hirose et al., 2003). It is possible that periostin inhibits precocious cell differentiation to keep its proliferative potentiality in these cellular elements at the fetal stage.

In addition to the function of periostin as an adhesion molecule as suggested by the observations of periosteum (Horiuchi et al., 1999; Hirose et al., 2003) and the periodontal ligament (Horiuchi et al., 1999; Wilde et al., 2003; Kruzynska-Frejtag et al., 2004), periostin might further be involved in the homeostasis of the periodontal ligament. Periostin has been regarded as a member of the Fasciclin I family, including βig-h3 (Skonier et al., 1992), Algal-CAM (Huber and Sumper, 1994), and midline fasciclin (Hu et al., 1998), mimicking a function of Fasciclin I. Periodontal fibroblasts, major component cells in the periodontal ligament, are able to differentiate into multiple types of cells such as cementoblasts and osteoblasts and appear to play an essential role in response to mechanical forces of the tooth and in repair of a damaged matrix (Lekic and McCulloch, 1996). Although the periodontal fibroblasts have been well-known to have high alkaline phosphatase activity, the periodontal ligament is never calcified under normal conditions. Recently, Ohno et al. (2002) demonstrated that βig-h3 inhibits the mineralization of the periodontal ligament by the suppression of alkaline phosphatase activity. It would therefore be better to consider that periostin serves for the maintenance of homeostasis of the periodontal ligament by inhibiting mineralization.

It is noteworthy that a constant exhibition of periostin immunoreactivity was found around the developing sites—the labial and lingual cervical loops—of the incisor, as well as around the cervical loop of molar. Recently, Matias et al. (2003) have suggested the interaction between periostin and integrin family; the latter substances are capable of binding various ligands such as extracellular matrices and their associated proteins. The expression of β5 integrin (Yamada et al., 1994) and αv integrin subunit (Salmivirta et al., 1996) has been demonstrated during early tooth development by immunocytochemistry and in situ hybridization histochemistry, suggesting the involvement of the integrin family in the epithelial-mesenchymal interaction during tooth development. This similar expression pattern between periostin and the integrin family leads us to propose the possibility that periostin is involved in initial tooth development via cell-to-cell interaction, i.e., epithelial-mesenchymal interaction. Further investigation will have to clarify the functional significance of periostin during early tooth development.

Acknowledgements

The authors thank Mr. Masaaki Hoshino and Mr. Kiichi Takeuchi of the Divisions of Oral Anatomy, Niigata University Graduate School of Medical and Dental Sciences, for technical assistance.

Ancillary