Glial cell line-derived neurotrophic factor (GDNF), a distant member of the transforming growth factor-β (TGF-β) family, was originally isolated and purified from a conditioned medium of the B49 rat glial cell line (Lin et al., 1993, 1994). The GDNF is a relatively new family of neurotrophins which mediates trophic effects on neuronal survival, growth, and target innervation. GDNF promotes the survival and the differentiation of many peripheral neurons, such as sympathetic, parasympathetic, and sensory neurons (for reviews, Lin et al., 1993; Airaksinen and Saarma, 2002; Enomoto, 2005). GDNF sends signals through a multicomponent receptor complex comprising a glycosylphosphatidylinositol-anchored cell surface molecule—the GDNF family receptor (GFRα)—and RET (for “rearrangement during transfection”) tyrosine kinase, triggering the activation of multiple signaling pathways in responsive cells (for a review, Airaksinen and Saarma, 2002). However, GDNF preferentially binds with GFRα 1, but not directly with RET. The complex of GDNF-GFRα 1 is required for the subsequent activation of the tyrosine kinase RET receptor (Treanor et al., 1996; Enomoto, 2005). Damage to motor axons leads to an increase in GDNF expression in the Schwann cells after elevation of the RET expression (Naveilhan et al., 1997; Höke et al., 2002).
A series of immunohistochemical and electron microscopic studies have revealed that Ruffini endings, corresponding to mechanoreceptors of slowly adapting type II (Chambers et al., 1972; Biemesderfer et al., 1978), are the primary mechanoreceptor in the periodontal ligament (for review, Maeda et al., 1999). The periodontal Ruffini endings are characterized by elaborate ramifications of the expanded axon terminals and by an association with a peculiar Schwann cell called the terminal Schwann cell (Byers, 1985; Maeda et al., 1989; Kannari et al., 1991). These terminal Schwann cells, analogues to the lamellar or laminar cells of the cutaneous mechanoreceptors, play crucial roles in the development and regeneration of the periodontal Ruffini endings (cf., Maeda et al., 1989). The lingual periodontal ligament of the rodent incisor has been reported to show a characteristic distribution of the Ruffini endings; they are restricted to the alveolar half of the ligament called the alveolus-related part (ARP), whereas never appearing in the other tooth half called the tooth-related part (TRP) (Sato et al., 1988). Thus, the lingual ligament of the rodents represents a favorite area for investigating the morphological properties of the periodontal Ruffini endings.
Immunocytochemical and experimental studies have shown that the periodontal Ruffini endings have a high potential for neuroplasticity (cf., Maeda et al., 1999; Wakisaka et al., 2000), as supported by findings that their morphological regeneration is completed around postoperative 4 weeks following transection of the inferior alveolar nerve in rats, and quantitative analyses showing that the neural density of the periodontal nerves decreases to a minimum around postoperative 3 days but recovers to normal levels by postoperative 4 weeks (Atsumi et al., 1999b, 2000). Because the periodontal Ruffini endings—including the terminal Schwann cells—have been reported to express trkB, a high affinity neurotrophin receptor, that binds brain-derived neurotrophic factor (BDNF)-neurotrophin-4/5 (NT-4/5) (Ochi et al., 1997; Atsumi et al., 1999a), the involvement of a trkB/BDNF-NT-4/5 signaling pathway had been considered for their regeneration process, as supported by observations of mice lacking bdnf and nt-4/5 (Harada et al., 2003; Jabbar et al., 2007). However, an analysis of the reduction rate of neural density in the Ruffini endings implies a possibility for the involvement of these two neurotrophins in the regeneration of the periodontal Ruffini endings in a stage-specific manner.
Following nerve injury, previous reports have demonstrated that GDNF reduced the degeneration of motor neurons after spinal root avulsion and distal nerve axotomy (Yuan et al., 2000), promoted axonal regeneration (Dolbeare and Houle, 2003), and enhanced the regeneration of dorsal roots into the adult rat spinal cord (Iwakawa et al., 2001). Recently, we have revealed a close relationship between the expression of GDNF immunoreaction and expansion of the axonal profiles in the periodontal Ruffini endings (Aita et al., 2006; Igarashi et al., 2007), suggesting that GDNF is a key molecule for the maturation and maintenance of the periodontal Ruffini endings. However, no information is available regarding the involvement of GDNF in the regeneration of the periodontal Ruffini endings. Thus, this study examined changes in the expression of GDNF immunoreactions in the periodontal ligament of rat incisors receiving a transection of the inferior alveolar nerve. Because our previous reports have demonstrated the exhibition of GDNF immunoreaction in the terminal Schwann cells but not in the axon terminals (Aita et al., 2006; Igarashi et al., 2007), this study used immunohistochemistry for S-100 protein, a reliable marker for these cells, to demonstrate the regeneration process of the periodontal Ruffini endings. Furthermore, changes in GDNF were assessed in the trigeminal ganglion at protein and mRNA levels by immunohistochemistry and a real time reverse transcriptase polymerase chain reaction (RT-PCR) method.
MATERIALS AND METHODS
All animal experiments were approved by the Niigata University Institutional Animal Use and Care Committee (approval number #161). The animals were housed in a temperature-controlled room under normal 12 hr light/12 hr dark laboratory conditions with free access to chow and water.
A total of 26 rats (8 weeks of age, weighing 200–250 g at surgery) were used in this experimental study. The transection of the inferior alveolar nerve was performed according to the methods described by Atsumi et al. (1999b, 2000) and Harada et al. (2003). Briefly, under anesthesia with by an intraperitoneal injection of 8% chloral hydrate (400 mg/kg), the inferior alveolar nerve exposed from the mandibular canal was transected at one side, and the cut ends of the nerve were returned into the canal, and then the wound was sutured (experimental group, n = 20). Six rats without any surgical treatment served as a control group. No postoperative treatment such as the administration of antibiotics was given to the operated rats.
Tissue Preparation for Immunohistochemical Staining
According to Youn et al. (1997), the regeneration process in this animal model can be divided into following the four stages: degeneration (postoperative (PO) 3 days), commencement of regeneration (PO 1 week), regeneration (PO 2 weeks), and completion of regeneration (PO 4–8 weeks). Thus, animals of the experimental group were allowed to survive for 3 days, 1, 2, 4, and 8 weeks after surgery (n = 4 at each stage). At appropriate survival periods, the animals were deeply anesthetized in the same way as described above, and perfused transcardially with a fixative containing 4% paraformaldehyde in a 0.1 M phosphate buffer (pH 7.4). The mandibles and the trigeminal ganglia were removed en bloc and immersed in the same fixative at 4°C for 12 hr. After decalcification of mandibles in a 10% ethylene diamine tetraacetic acid disodium (EDTA-2Na; Dojindo Laboratories, Kumamoto, Japan) solution for 3–4 weeks at 4°C with slight agitation, tissue blocks were equilibrated in a 30% sucrose solution overnight for cryoprotection. Frozen sections of mandibles including the incisors were serially cut at a thickness of 35 μm with a freezing microtome (Yamato Koki, Tokyo, Japan), collected in phosphate-buffered saline, pH 7.4, and treated as free-floating sections. Additionally, cryostat sections of the trigeminal ganglion were sectioned at a thickness of 10 μm in a cryostat (CM 3050S; Leica, Nussloch, Germany).
Immunohistochemistry for S-100 Protein and GDNF
Free-floating or cryostat sections were processed for immunohistochemistry according to our protocol as previously reported (Aita et al., 2006; Igarashi et al., 2007; Jabbar al., 2007). After an inhibition of endogenous peroxidase with 0.3% hydrogen peroxide in absolute methanol for 30 min at room temperature, frozen or cryostat sections were treated with 2.5% normal goat serum (Vector Lab., Burlingame, CA) for 60 min. The sections were primarily incubated with either a rabbit polyclonal antiserum against human GDNF (1:300; Santa Cruz Biotechnology, Santa Cruz, CA) or a rabbit polyclonal antiserum against bovine S-100 protein (1:500; Immunotech, Marseille Cedex, France) overnight at 4°C. The incubated sections were then reacted with a biotinylated goat anti-rabbit IgG (1:1000, Vector Lab.) and subsequently with a peroxidase-conjugated avidin (ABC Kit, Vector Lab.) for 90 min each at room temperature. The immunoreaction sites were visualized by an incubation in a 0.05 M Tris buffer (pH 7.6) containing 0.04% 3-3′-diaminobenzidine (DAB) and 0.03% hydrogen peroxide. The immunostained sections were counter-stained with 0.03% methylene blue.
Quantitative Analysis of GDNF-Positive Neurons in Trigeminal Ganglion
The cross-sectional areas of GDNF-positive neurons in the trigeminal ganglion showing the nucleolus profiles were calculated from three sections per each animal using an NIH image (http://rsb.info.nih.gov/ij/download.html). The relative ratio of GDNF-positive neurons to total trigeminal neurons per section was also calculated in a double blind method. ANOVA analysis was used for the statistical comparison between control and postoperative periods. A P value of less than 0.05 was considered a significant difference.
Real Time PCR Procedure
Additional rats both with (3 days, 1, 2, 4, 8 weeks; n = 2 each) and without nerve injury (n = 2) were decapitated under deep anesthesia by diethyl ether. The trigeminal ganglia were removed, immediately frozen in dry ice, and stored at −80°C until use. Frozen tissue was homogenized in 1 mL TRIzol reagent (Gibco BRL, Gaithersburg, MD), and the total RNA was isolated according to the manufacturer's instructions. RNA was quantified with GeneQuant II (Amersham Pharmacia Biotech, Piscataway, NJ).
Samples from each experiment were run separately in duplicate. Reverse transcription was performed using a PrimeScript™ RT reagent Kit (TaKaRa, Ohtsu, Japan), and the reaction mix was subjected to quantitative real-time PCR to detect relative levels of the corresponding GDNF or GAPDH by using a Thermal Cycler Dice TP800 (TaKaRa) and SYBER® Premix Ex Taq™ II Kit (TaKaRa). The real-time PCR reaction was carried out in the presence of gene specific GDNF (forward 5′-CCG GAC GGG ACT CTA AGA TGA-3′ and reverse 5′-GTC AGG ATA ATC TTC GGG CAT ATT G-3′) and GAPDH (forward 5′-GGC ACA GTC AAG GCT GAG AAT G-3′ and reverse 5′-ATG GTG GTG AAG ACG CCA GTA-3′) primers. The ratio of GDNF to GAPDH mRNA copy numbers from the trigeminal ganglion was calculated by taking the average of mRNA copy numbers obtained from the two sets of experiments performed in duplicate. Based on this, the percent change in copy number of GDNF mRNA of the trigeminal ganglion was calculated. ANOVA analysis was used for the statistical comparison between postoperative periods as well as between control and experimental groups (Turkey-Kramer's HSD test). A P value of less than 0.05 was considered a significant difference.
Immunostaining with S-100 protein depicted the outlines of the periodontal Ruffini endings in the rat lower incisors (Fig. 1a). After thick S-100-immunopositive structures penetrated into the incisor periodontal ligament, they extensively branched out in a dendritic fashion (Fig. 1b). They were restricted to the alveolar half of the ligament referred to as the ARP but never appeared in the tooth half called the TRP. In addition, S-100 immunostaining demonstrated an association of the rounded cells with a kidney-shaped nucleus, which allows their easy identification as terminal Schwann cells, as reported previously (Byers, 1985; Sato et al., 1988; Maeda et al., 1989), at the ramified terminal portions of the periodontal Ruffini endings (Fig. 1b). The ordinary Schwann cells associated with the nerve fibers show S-100 immunoreactivity while nerve fibers did not.
GDNF-immunostaining also revealed many round cells and their thick processes of the Ruffini endings in the periodontal ligament of the rat incisor (Fig. 1c). The GDNF-immunoreaction in the terminal Schwann cells was localized in their cytoplasm but not in their nuclei. The distribution pattern of GDNF-positive structures was identical to that of S-100 positive ones in the ligament; these were restricted to the ARP, never appearing in the TRP.
PO Day 3
On Day 3, when previous studies reported the almost complete disappearance of the periodontal nerve fibers (cf., Atsumi et al., 2000; Harada et al., 2003), S-100 immunostaining showed the outlines of the Ruffini endings with the terminal Schwann cells in the ARP of the periodontal ligament. In addition to the rounded cells with S-100 immunoreaction, spindle-shaped cells appeared in the TRP where no positive elements existed under normal conditions (Fig. 2a). The long axis of these positive cells was in parallel to the tooth axis.
Transection of the inferior alveolar nerve induced the disappearance of GDNF positives structures from the lingual periodontal ligament except for a small number of dot-like positive elements in the ARP (Fig. 2b). The spindle-shaped cells in the TRP, as demonstrated with S-100 immunostaining, were devoid of GDNF immunoreaction.
PO Week 1
One week following the transection, the spindle-shaped cells with S-100 positivity increased in number in the TRP when compared with the previous stage (Fig. 2c). These cells frequently extended their short cytoplasmic processes. Furthermore, some S-100 positive cells were located near the tooth surface (Fig. 2c).
The GDNF-positive elements showed almost the same distribution and morphology as those observed at the previous stage (Fig. 2d). In rare cases, however, a few rounded cells positive for GDNF occurred in the ARP, which contained the terminal Schwann cells associated with the periodontal Ruffini endings (Fig. 2d inset).
PO Week 2
This stage has been reported to be where the regenerating nerve fibers are increasing in number in the ARP (cf., Atsumi et al., 2000; Harada et al., 2003). In immunohistochemistry for S-100 and GDNF, the immuno-expression and distribution pattern of these positive elements were almost comparable to those observed at the previous stages; spindle-shaped cells with S-100 immunoreaction remained in the TRP (Fig. 2e) while no cells with GDNF-positivity existed there. In some cases, GDNF-positive structures were displayed in dendritic fashion in the ARP (Fig. 2f). The GDNF-positive structures appeared to consist of thin nerve fibers and probable Schwann cells because these structures were positive in S-100 immunoreaction (data not shown).
PO Week 4
At 4 weeks following transection, the distribution and terminal morphologies of the regenerated nerve fibers have been shown to return to normal levels (cf., Atsumi et al., 2000; Harada et al., 2003).
S-100 positive cells, spindle in shape, had completely disappeared from the TRP until this stage (Fig. 2g). S-100 reactive rounded cells and their cytoplasmic processes assembled in the ARP. These S-100 positive structures appeared in the periodontal Ruffini ending, with the terminal Schwann cells recognizable under normal conditions. However, their terminals appeared to be fewer with less frequent ramifications than the control animals.
In accordance with the regeneration of the Ruffini endings, intense GDNF reactive cells, round in profile, appeared in the ARP (Fig. 2h). Because of their profiles, location, and immunolocalization, these cells could be regarded as the terminal Schwann cells. The immunoreactive cells frequently extended their short cell processes. Similar to GDNF immunohistochemistry, however, their numbers appeared fewer than in the control.
PO Week 8
From PO week 4 to 8, the regeneration of the periodontal Ruffini endings had proceeded extensively. All immunostaining demonstrated a distribution and morphology of the periodontal Ruffini endings identical to those in the control animals (Fig. 3a,b).
Changes in GDNF-Immunoexpression in Trigeminal Ganglion
In the trigeminal ganglion of the control group (Fig. 4a) as well as the experimental group of PO day 3 (not shown) and week1 (Fig. 4b), a strong immunoreaction for GDNF appearing as a ring was found to surround immuno-negative trigeminal neurons, that is, satellite cells. In addition, a very few trigeminal ganglion neurons displayed GDNF immunoreaction at these stages (Fig. 4a,b). This immunolocalization pattern of GDNF reactive satellite cells remained unchanged, but a small population of trigeminal ganglion neurons came to show GDNF immunoreaction at postoperative weeks 2 (Fig. 4c) and 4 (Fig. 4d). A quantitative analysis of relative ratio of GDNF-positive neurons to total trigeminal neurons showed 6.0 ± 0.5% in control (1073 neurons), 6.9 ± 2.1% (855 neurons) at PO week 1, 9.0 ± 2.7% at PO week 2 (770 neurons), and 7.5 ± 1.2% at PO week 4 (1162 neurons). There were no significant difference except for between control and PO week 2 (P < 0.05). The cross-sectional areas of GDNF-positive and all trigeminal neurons exhibited 687.9 ± 180.5 μm2 in control, 765.1 ± 229.8 μm2 at PO week 1, 745.1 ± 234.8 μm2 at PO week 2, and 738.6 ± 312.1 μm2 at PO week 4. However, there was no significant difference.
The changes in immuno-expression of S-100 protein and GDNF during this experimental period are summarized in Table 1.
Table 1. Summary of changes in immuno-expression pattern during observation period
Real Time Quantitative RT-PCR Analysis in the Trigeminal Ganglion
Real-time RT-PCR analysis demonstrated a higher elevation of transcripted GDNF mRNA in the trigeminal ganglion between control (66.4 ± 11.9%) and experimental groups of PO week 2 (106.6 ± 15.8%), between PO day 3 (67.7 ± 18.0%) and week 2, and between weeks 1 (51.4 ± 20.2%) and 2 with significant difference (Fig. 5). The expression of GDNF mRNA at PO week 2 increased to 1.6 times more than at that of the control group. However, there was no significant difference in GDNF mRNA expression between PO weeks 2 and 4 (88.1 ± 27.2%), and between the control and other experimental groups, except for PO week 2.
This study clearly demonstrated the chronological changes in the expression of GDNF in the lingual periodontal ligament of the rats following transection of the inferior alveolar nerve. Our current observations of its expression in terminal Schwann cells synchronize with the time course of the regeneration of the periodontal Ruffini endings as shown by previous studies (cf., Maeda et al., 1999; Wakisaka et al., 2000), suggesting that GDNF is one of the molecules responsible for regeneration of the periodontal Ruffini endings.
The terminal Schwann cells have been reported to play crucial roles in the development and regeneration of the periodontal Ruffini endings (cf., Maeda et al., 1999). Transection of the inferior alveolar nerve induced a migration of the spindle-shaped cells with S-100 immunoreaction into the TRP, which never contains the periodontal Ruffini endings under normal conditions as evidenced by the fact that they completely disappeared from there by postoperative week 4. This behavior of the spindle-shaped cells with the S-100 immunoreaction is consistent with previous reports (Atsumi et al., 1999b; 2000). Although the type of these S-100 positive cells remains unclear, they might be regarded as a phenotype of Schwann cells due to the expression of S-100 protein and the electron microscopic configurations (Atsumi et al., 1999b). In this experimental model, the periodontal nerve fibers completely disappeared within PO day 3, a few nerve fibers appeared around week 1, and the terminal formation of the periodontal Ruffini endings returned to normal by week 3 following injury (Youn et al., 1997). These findings suggest a close relationship between the behavior of the Schwann cells and regeneration of the periodontal axons. A developmental study also showed that the axons expanded after making contact with the terminal Schwann cells (Hayashi et al., 2000).
Neural recovery in the periodontal ligament seems to occur faster in the rat than in the cat. A previous report indicated incomplete recovery of nerve number and receptor structure in the cat periodontal ligament at 1 year after denervation (Holland and Robinson, 1987). This time lag may be explained by following three factors; species difference (rat versus cat), type of tooth (continuously growing tooth in rat versus root tooth in cat), and kinds of tissue. Indeed, a similar time lag of neural recovery has been reported in tooth replantation experiments (Holland and Robinson, 1987; Kamasaki et al., 1993; Yamada et al., 1999). Furthermore, the degeneration and regeneration processes of the periodontal Ruffini endings may occur more rapidly when compared with other tissues including the inner conical body of the rat mystical vibrissae (Renehan and Munger, 1986) and the rat elbow joint capsule (Sasamura, 1986). Thus, we may regard that a faster neural recovery as shown in this study is specific for the periodontal ligament of continuously growing tooth.
A notable finding in this study is that the migrated Schwann cells with spindle-shape lacked GDNF immunoreaction in spite of a positive reaction in the Schwann cells repositioned in the ARP. Our recent developmental study revealed that the GDNF-positive terminal Schwann cells occurred at postnatal 2 weeks when they were located near the expanded and dendritic axons of the periodontal Ruffini endings, suggesting that GDNF mediates the regeneration of the periodontal Ruffini endings (Igarashi et al., 2007). Although the functional role of S-100 proteins in the nervous system is not fully understood, it has been thought that the S-100 protein has possible intracellular functions including the regulation of cell growth/cell division and that of regulation of cell shape, energy metabolism, and signal transduction (for reviews, Zimmer et al., 1995; Schäfer and Heizmann, 1996). Furthermore, an experimental study on the denervated cutaneous mechanoreceptors suggests that trophic factors derived from specialized Schwann cells are essential for the induction of sensory receptors (Dellon and Munger, 1983). All this evidence suggests S-100 proteins may also perform neurotrophic functions in sensory neurons (Marshak, 1990; Van Eldik et al., 1991). Taken together with the expression pattern of GDNF, these findings indicate that the immunohistochemical property of the migrated Schwann cell into the TRP is that of an immature phenotype which lacks the ability to mediate neurotrophic effects.
The periodontal ligament receives a dual innervation of sensory nerves both from the trigeminal ganglion and mesencephalic trigeminal nucleus (cf., Byers et al., 1986; Byers and Dong, 1989). However, an experimental study utilizing axonal transport has demonstrated that the periodontal Ruffini endings in the rat incisor originate only from the trigeminal ganglion (Byers and Dong, 1989). During the regeneration, we found a temporal increase in number of GDNF positive trigeminal ganglion neurons at PO week 2, when compared with the control group where the GDNF reaction was exclusively localized in the satellite cells. Current real time RT-PCR analysis indicated a significant elevation of GDNF mRNA at PO week 2. In a previous RT-PCR analysis of the contused rat spinal cord (Satake et al., 2000), transcripted GDNF mRNA began to increase within 30 min postinjury, peaked at 3 hr postinjury, and returned to the baseline level by 4 weeks postinjury. Because GDNF immunoreaction was recognizable in activated macrophages and microglia, the authors postulated that the up-regulation of GDNF may be a component of post-traumatic inflammation. However, this experimental model never induced inflammation in the periodontal ligament (Atsumi et al., 1999b; 2000), and the migrated S-100 positive cells were not macrophages, but immature Schwann cells (Atsumi et al., 2000; this study). Thus, we may consider that this temporal well-synchronized elevation at protein and mRNA levels is related to the appearance of trigeminal ganglion neurons with GDNF immunoreaction, as shown in this study. Indeed, a marked up-regulation of GDNF mRNA and protein has been found in dorsal root ganglion neurons following spinal cord injury (Sakurai et al., 1999; Widenfalk et al., 2001). However, we cannot exclude the possibility of the up-regulation of GDNF mRNA in the satellite cells with or without elevation in the neurons. Injury to trigeminal nerves caused alterations in the synthesis of various substances including neuropeptides, and the glia fibrillary acidic protein occurs in satellite cells following nerve injury (Stephenson and Byers, 1995; Hökfelt et al., 1996; Chudler et al., 1997). A recent work by Sun et al. (2008) also reported increased GDNF production in astrocytes in vivo and in vitro. Further investigation is needed to clarify whether a reaction of the trigeminal ganglion neurons or satellite cells induces a temporal elevation of GDNF mRNA following transection of the inferior alveolar nerve.
The present immunohistochemical observations support the involvement of GDNF in the regeneration of the periodontal Ruffini endings following transection of the inferior alveolar nerve. Because the heterogenous expression of trkB, a high affinity neurotrophin receptor for BDNF and NT-4/5 (Bothwell, 1991, 1995; Klein et al., 1991; Barbacid, 1994), has been found in the periodontal Ruffini endings (Ochi et al., 1997; Atsumi et al., 1999a), the NT-4/5 and/or BDNF/trkB signaling pathways have been considered necessary for the development and regeneration of the periodontal Ruffini endings. Indeed, mutant mice lacking either bdnf (Hoshino et al., 2003) or trkB (Matsuo et al., 2002) never developed periodontal Ruffini endings. Nt-4/5-depletion also induced a low density of the periodontal Ruffini endings from postnatal 1 to 8 weeks (Maruyama et al., 2005). In a regeneration study, furthermore, nt-4/5 depleted mice showed a delay in the regeneration of the periodontal Ruffini endings, but this delay was shortened by an exogenous administration of NT-4/5 (Jabbar et al., 2007). Bdnf heterozygous mice, which exhibit ∼50% of the BDNF protein levels of wild mice (Bianchi et al., 1996), also display a delayed regeneration of the periodontal Ruffini endings, with cessation of the regeneration at 60% of the normal level (Harada et al., 2003). All of these findings suggest the necessity of NT-4/5 and/or BDNF/trkB signaling pathways for the development and regeneration of the periodontal Ruffini endings. However, the chronological changes in reduction in neural density differed between bdnf heterozygous and nt-4/5 homozygous mice. In particular, BDNF and NT-4/5 are, respectively, regarded to function as triggers for neural sprouting and the maintenance of the periodontal Ruffini endings as well as to involve the regeneration at early stages. Because the expression of GDNF and the highest elevation of GDNF mRNA were recognizable at the maturation stage in this study, GDNF might be a key molecule for the maturation and maintenance of the periodontal Ruffini endings. Different kinds of neurotrophins including BDNF, NT-4/5, and GDNF might control the regeneration of the periodontal Ruffini endings in a stage-specific manner. With a consideration of our series of studies (Harada et al., 2003; Jabbar et al., 2007; this study), it is reasonable to conclude that the respective functions are for BDNF at the initial stage, NT-4/5 at the early stage, and GDNF at the maturation stage during the regeneration of the periodontal Ruffini endings.
The authors thank Messers M. Hoshino and K. Takeuchi, Division of Oral Anatomy, Niigata University Graduate School of Medical and Dental Sciences, for their technical assistance.