Introduction.
Versican is a large interstitial chondroitin sulphate proteoglycan initially isolated from human fibroblast cultures (Zimmermann & Ruoslahti 1989) and similar proteoglycans have been found in various connective tissues (Shinomura et al. 1990, Zimmermann et al. 1994, Landolt et al. 1995). Versican interferes in vitro with the attachment of cells to various extracellular matrix components (Yamagata et al. 1989). In vivo , versican seems to act as a barrier to migratory neural crest cells and outgrowing axons (Landolt et al. 1995), and also may participate in the control of keratinocyte and dermal fibroblast proliferation (Zimmermann et al. 1994). Versican has a peptide domain homologous to hyaluronan (HA)- binding domain of aggrecan and is thought to be able to bind HA via this domain to form large aggregate structure (LeBaron et al. 1992). HA is a large glycosaminoglycan, without core protein and is believed to be involved in tissue formation and cell differentiation (Toole 1991).
Versican (Bartold et al. 1995, Yamauchi et al. 1997) and HA (Linde 1973, Mangkornkarn & Steiner 1992, Nieminen et al. 1993) have been also identified in the dental pulp and recently a proteoglycan aggregate consisting of versican, HA and link protein was isolated from the rat incisor dental pulp (Shibata et al. 2000). A previous histochemical study showed that both HA and versican were localized in the radicular dental pulp of the rat molar, but that there was considerable regional variation in the staining intensities for both molecules (Shibata et al. 1999). These results lead to the hypothesis that the distribution of both molecules can vary with physiological conditions in the dental pulp. As a first trial to confirm this, we determined to investigate the developmental changes in histochemical localization of both molecules, since rat molars undertake a series of developmental changes including completion of crown formation, root formation and eruption during the postnatal period.
A number of structural differences have been described between the coronal and radicular parts of the dental pulp, including those in the vascular system (Nakamura 1986), the nervous system (Maeda et al. 1987) and the number of immunocompetent cells (Okiji et al. 1992). These studies indicate that there are different physiological properties such as in defence mechanisms between the two regions of dental pulps. Therefore, in line with the above hypothesis, we also sought to assess regional differences in the distribution of HA and versican.

Materials and methods.
All animals were housed in facilities approved by the Tokyo Medical and Dental University. The animal-use protocol form conforming to NIH guidelines as stated in the ‘Principles of Laboratory Animal Care’ (NIH Guidelines 1985) was reviewed and approved by the Screening Committee for Animal Research of the Tokyo Medical and Dental University prior to the study.
Two pregnant Sprague-Dawley rats were purchased from Sankyo laboratories (Tokyo, Japan) and 30 newborn rats were raised during the experimental period. Postnatal rats, 1, 7, 14, 21, 28, 35, 42 and 49 days old, were anaesthetized with ether and killed by cervical dislocation. Three to four rats in each age group were used. The mandibles were dissected and immersed in 4% paraformaldehyde solution (0.1 mol L 1 phosphate buffer, pH 7.4, at room temperature) for 1 day, decalcified in 10% EDTA for 2 weeks at room temperature and embedded in paraffin. Five to six teeth samples of the mandibular first molars in each age group were investigated. Sagittal and horizontal sections (4 m) were mounted on glass slides coated with poly-L-lysine.
The immunohistochemistry for versican and histochemical staining for HA has been described in detail (Shibata et al. 1999). Briefly, sections digested with trypsin (Wako Chemicals, Tokyo, Japan) were treated with monoclonal antibodies 12C5 (antiversican HAbinding region, the Developmental Studies Hybridoma Bank, Iowa City, IA, USA) and CS-56 (antichondroitin sulphate chain, Seikagaku Corporation, Tokyo, Japan), or with biotin-labelled HA-binding protein (Seikagaku Corporation, Tokyo, Japan). These probes have been well characterized (Asher et al. 1991, Asari et al. 1992, Shibata et al. 2000). We earlier confirmed that 12C5 reacts with the chondroitinase ABC digested core protein of the large proteoglycan from the rat dental pulp (Shibata et al. 2000). Its molecular weight (400–500 Kda) is equivalent to the versican core protein. Furthermore, antichondroitin sulphate monoclonal antibodies, including CS-56, are known to recognize versican, but not decorin (Sorrell et al. 1999). The streptoavidin-biotin method was then applied to the sections using a HISTOFINE SAB kit (Nichirei, Tokyo, Japan). Biotin-labelled antimouse IgG + IgM + IgA in this kit was used as the secondary antibody for immunostaining. Finally, sections were treated with AEC (3-amino-9-ethylcarbazole) (Nichirei, Tokyo, Japan) to reveal any reaction. Negative control sections for versican immunohistochemistry were incubated with normal mouse IgG instead of the primary antibodies. Negative controls for HA staining were prepared by treating sections with Streptomyces hyalurolyticus hyaluronidase (Seikagaku Corporation, Tokyo, Japan) after trypsin digestion. All sections were examined after counterstaining with haematoxylin.

At day 1, HA staining (Fig. 1a) and immunostaining for 12C5 (Fig. 1b) were found throughout in the interior parts of the mandibular first molar dental pulp.
At day 7, HA staining was found throughout the dental pulp, but a strong staining area appeared in the subodontoblastic layer (Fig. 2a), whilst immunostaining for 12C5 was still restricted to the interior dental pulp (Fig. 2b).
At day 14, the mandibular first molar had already finished its crown formation. HA staining (Fig. 3a) and immunostaining for 12C5 (Fig. 3b) were found throughout the dental pulp, but strong reactions for both molecules were seen in the subodontoblastic layer.
At day 21, the molar had erupted and its root formation had started. Figure 4a,c,e shows HA staining and Fig. 4b,d,f shows immunostaining for 12C5 cut longitudinally (Fig. 4a,b), and transversely through the coronal pulp (Fig. 4c,d) and the radicular pulp (Fig. 4e,f ). Although strong reactions for both HA and 12C5 were seen in the subodontoblastic layer of the coronal pulp (Fig. 4a–d), a strong reaction for HA and a weak reaction for 12C5 were seen in the subodontoblastic layer of the radicular pulp (Fig. 4a,b,e,f ). A strong reaction for 12C5 gradually shifted centripetally with increasing distance from the coronal pulp toward the root (see Fig. 4b). Immunostaining for CS-56 showed an identical staining pattern to 12C5 at this stage (Fig. 4g) and at all other ages examined (data not shown). These results confirmed versican localization in the corresponding area.

Figures 1, 2, 3. Serial sections showing HA staining (a) and immunostaining for 12C5 antibodies (b). Bars = 100 m. At day 1 postnatal, uniform staining patterns for both molecules are seen in the interior parts of the dental pulp (arrow in 1a,b). At day 7, a strong reaction for HA is seen in the subodontoblastic layer (arrow in 2a), whereas the reactivity for versican is still restricted to the interior pulp (arrow in 2b). At day 14, the tooth crown formation has almost finished and strong reactions for both molecules are seen in the subodontoblastic layer (arrows in 3a,b).

At days 28 and 35, the staining pattern for both molecules was similar to that at day 21 (data not shown).
At day 42, although the subodontoblastic layer in the coronal pulp still showed strong reactions for both molecules (Fig. 5a–d), a 12C5-negative area first appeared in the central pulp region (Fig. 5a,c). In the corresponding area, HA staining was detectable but weak (Fig. 5b,d). Perfusion-fixation sections showed similar results (data not shown), indicating these findings were not fixation artifacts. Meanwhile, the staining pattern in the radicular pulp for either molecule did not change (Fig. 5a,b).
At day 49, the 12C5-negative, low-HA area in the central coronal pulp expanded (Fig. 6a,b).
No immunostaining was seen in the negative control section at day 21 (Fig. 7a). Similarly, positive reactions for HA completely disappeared after digestion with the Streptomyces hyalurolyticus hyaluronidase in the negative control section (Fig. 7b). Similarly, none of the negative control sections showed positive reactions for immunostaining or HA staining at any stage (data not shown).

Discussion.
HA and versican were histochemically localized in all dental pulps studied during the postnatal period until 49 days of age, but the staining patterns for both molecules varied with age. Notably, strong reactions for both molecules appeared in the subodontoblastic layer of the coronal pulp by the completion of crown formation. However, the pattern in the radicular pulp was different; a strong reaction for HA and a weak reaction for versican were seen in the subodontoblastic layer of the radicular pulp. This was also the most common pattern observed in the radicular pulp in our previous study (Shibata et al. 1999).
During cavity preparation, odontoblasts are separated from the predentine, but not from the subodontoblastic layer, and shift inward together with it (Sato 1989, Ohshima 1990). Therefore, HA and versican in this layer may play a role in supporting and protecting odontoblasts. Yamauchi et al. (1997) made a similar conclusion based on the localization of versican and link protein in the subodontoblastic layer of the bovine dental pulp. Moreover, the subodontoblastic layer contains a nerve plexus (Maeda et al. 1987) and a capillary network (Nakamura 1986, Ohshima 1990), and hence HA and versican also seem to be involved in maintaining these structures.

Figures 4. Serial sections, respectively, and 4g is an adjacent section to a at day 21. HA staining (a,c,e) and immunostaining for 12C5 (b,d,f ) and for CS-56 (g) cut longitudinally (a,b,g), and transversely through the coronal pulp (c,d) and the radicular pulp (e,f ). In the coronal pulp, strong reactions for both HA and 12C5 are seen in the subodontoblastic layer (arrows in a-d). A strong reaction for 12C5 gradually shifted centripetally with increasing distance from the coronal pulp toward the root (arrowheads in b and f ), although a strong staining reaction for HA is still seen in the subodontoblastic region in the radicular pulp (arrowheads in a and e). Note a weak reaction for 12C5 is seen in the subodontoblastic layer of the radicular pulp (*in b and f ). Immunostaining for CS-56 shows an identical staining pattern to 12C5 (g). A weak reaction for CS-56 is also noted in the subodontoblastic layer of the radicular pulp (*in g). OB, odontoblasts. Bars = 100 m.

Compared to the radicular pulp, the coronal pulp has stronger defence mechanisms, including a higher number of immunocompetent cells (Okiji et al. 1992), and a more extensive subodontoblastic nerve plexus and capillary network (Nakamura 1986, Maeda et al. 1987, Sato 1989, Ohshima 1990). This is probably because the coronal pulp is more susceptible to exogenous stimuli than the radicular pulp. We suggest that the co-localisation of HA and versican seen in the coronal subodontoblastic layer can offer stronger support to the odontoblasts and can maintain more extensive nervous and/or vascular systems than HA alone seen in the radicular pulp.

Figure 5. Serial sections, respectively, at day 42. HA staining (a,c) and immunostaining for 12C5 (b,d) cut longitudinally (a,b), and transversely through the coronal pulp (c,d). Although the subodontoblastic layer in the coronal pulp shows strong reactions for both molecules (arrows in a-d), a 12C5-negative area is seen in the interior (arrowheads in b and d). In the corresponding area, HA staining was detectable but weak (arrowheads in a and c). The staining pattern in the root pulp for both molecules is identical to that at day 21. Bars = 100 m.


Figure 6. Serial sections showing HA staining (a) and immunostaining for 12C5 (b) at day 49. A versican-deficient, low-HA area in the central coronal pulp (arrows) expands at this age. Bars = 100 m.


Figure 7. Negative controls for immunostaining (a) and HA staining (b) at day 21 never show positive reactions. Bars = 100 m.

Versican is known to be involved in tissue formation by capturing a temporal space for succeeding tissues/cells in embryonic tissues (Shinomura et al. 1990, Landolt et al. 1995). Since the versican expression becomes extensively restricted in postnatal tissues (Bignami et al. 1993), its disappearance can be regarded as a sign of maturation. We think that versican in the dental pulp also has this function and that its disappearance in the interior of the coronal pulp shows the loss of this function with age, whereas its existence in the subodontoblastic layer contributes to supportive functions described above. Furthermore, since the coronal pulp is formed earlier, this change seems to occur earlier than in the radicular pulp.

References.

Asari A, Miyauchi S, Miyazaki Ket al. (1992) Intra- and extracellular localization of hyaluronic acid and proteoglycan constituents (chondroitin sulfate, keratan sulfate, and protein core) in articular cartilage of rabbit tibia. Journal of Histochemistry and Cytochemistry 40, 1693-704.
Asher RA, Perides EG, Vanderhaeghen J-J, Bignami A (1991) Extracellular matrix of central nervous system white matter: demonstration of an hyaluronate-protein complex. Journal of Neuroscience Research 28, 410-21.
Bartold PM, Moule AJ, Li H, Rigby P (1995) Isolation and characterization of the proteoglycans synthesized by adult human pulp fibroblasts in vitro. International Endodontic Journal 28, 163-71.
Bignami A, Perides G, Rahemtulla F (1993) Versican, a hyaluronate-binding proteoglycan of embryonal precartilaginous mesenchyma, is mainly expressed postnatally in rat brain. Journal of Neuroscience Research 34, 97-106.
Landolt RM, Vaughan L, Winterhalter KH, Zimmermann DR (1995) Versican is selectively expressed in embryonic tissues that act as barriers to neural crest cell migration and axon outgrowth. Development 121, 2303-12.
LeBaron RG, Zimmermann DR, Ruoslahti E (1992) Hyaluronate binding properties of versican. Journal of Biological Chemistry 267, 10003-10.
Linde A (1973) A study of the dental pulp glycosaminoglycans from permanent human teeth and rat and rabbit incisors. Archives of Oral Biology 18, 49-59.
Maeda T, Iwanaga T, Fujita T, Takahashi Y, Kobayashi S (1987) Distribution of nerve fibers immunoreactive to neurofilament protein in rat molars and periodontium. Cell and Tissue Research 249, 13-23.
Mangkornkarn C, Steiner JC (1992) In vivo and in vitro glycosaminoglycans from human dental pulp. Journal of Endodontics 18, 327-31.
Nakamura TK (1986) Scanning electron microscopic observations of the vascular system of rat molar pulps. Okajima Folia Anatomica Japan 62, 331-64.
Nieminen P, Vanio S, Jernvall J, Lukinmaa P-L, Lehtonen E, Thesleff I (1993) A chondroitin sulfate epitope in mammalian dental pulp and its developmental expression in mouse dental papilla. Journal of Dental Research 72, 1460-72.
NIH Guidelines (1985) Principles of Laboratory Animal Care. NIH publication nos 86-23 (revised 1985).
Ohshima H (1990) Ultrastructural changes in odontoblasts and pulp capillaries following cavity preparation in rat molars. Archives of Histology and Cytology 53, 423-38.
Okiji T, Kawashima N, Kosaka T, Matsumoto A, Kobayashi C, Suda H (1992) An immunohistochemical study of the distribution of immunocompetent cells, especially macrophages and Ia antigen-expressing cells of heterogeneous populations, in normal rat molar pulp. Journal of Dental Research 71, 1196-202.
Sato O (1989) Responses of pulpal nerves to cavity preparation in rat molars: an immunohistochemical study using neurofilament protein (NFP) antiserum. Archives of Histology and Cytology 52, 433-46.
Shibata S, Kaneko S, Yanagishita M, Yamashita Y (1999) Histochemical localization of hyaluronan and versican in the rat molar dental pulp. Archives of Oral Biology 44, 373-6.
Shibata S, Yoneda S, Yanagishita M, Yamashita Y (2000) Isolation of proteoglycan (versican) aggregate from rat dental pulp. Archives of Oral Biology 45, 563-8.
Shinomura T, Jensen KL, Yamagata M, Kimata K, Solursh M (1990) The distribution of mesenchyme proteoglycan (PG-M) during wing bud outgrowth. Anatomy and Embryology 181, 227-33.
Sorrell JN, Carrino DA, Baber MA, Caplan AI (1999) Versican in human fetal skin development. Anatomy and Embryology 199, 45-56.
Toole BP (1991) Proteoglycans and hyaluronan in morphogenesis and differentiation. In: Hay ED, ed. Cell Biology of Extracellular Matrix , 2nd edn. New York, USA: Plenum Press, 305-41.
Yamagata M, Suzuki S, Akiyama SK, Yamada KM, Kimata K (1989) Regulation of cell-substrate adhesion by proteoglycans immobilized on extracellular substrates. Journal of Biological Chemistry 264, 8012-8.
Yamauchi S, Cheng H, Neame P, Caterson B, Yamauchi M (1997) Identification, partial characterization, and distribution of versican and link protein in bovine dental pulp. Journal of Dental Research 76, 1730-6.
Zimmermann DR, Dours-Zimmermann MT, Schubert M, Bruckner-Tuderman L (1994) Versican is expressed in the proliferating zone in the epidermis and in association with the elastic network of the dermis. Journal of Cell Biology 124, 817-25.
Zimmermann DR, Ruoslahti E (1989) Multiple domains of the large proteoglycan, versican. EMBO Journal 8, 2975-81.