Neural regulation in tooth regeneration of Ambystoma mexicanum

The presence of nerves is an important factor in successful organ regeneration in amphibians. The Mexican salamander, Ambystoma mexicanum, is able to regenerate limbs, tail, and gills when nerves are present. However, the nerve-dependency of tooth regeneration has not been evaluated. Here, we reevaluated tooth regeneration processes in axolotls using a three-dimensional reconstitution method called CoMBI and found that tooth regeneration is nerve-dependent although the dentary bone is independent of nerve presence. The induction and invagination of the dental lamina were delayed by denervation. Exogenous Fgf2, Fgf8, and Bmp7 expression could induce tooth placodes even in the denervated mandible. Our results suggest that the role of nerves is conserved and that Fgf+Bmp signals play key roles in axolotl organ-level regeneration. The presence of nerves is an important factor in successful organ regeneration in amphibians. The Mexican salamander, Ambystoma mexicanum, is able to regenerate limbs, tail, and gills when nerves are present. However, the nervedependency of tooth regeneration has not been evaluated. Here, we reevaluated tooth regeneration processes in axolotls using a three-dimensional reconstitution method called CoMBI and found that tooth regeneration is nerve-dependent although the dentary bone is independent of nerve presence. The induction and invagination of the dental lamina were delayed by denervation. Exogenous Fgf2, Fgf8, and Bmp7 expression could induce tooth placodes even in the denervated mandible. Our results suggest that the role of nerves is conserved and that Fgf+Bmp signals play key roles in axolotl organ-level regeneration.


Results
Axolotl tooth structure and replacement. First, axolotl tooth histology and morphology were reexamined. The axolotl's teeth are found on four independent bones, the two dentaries and the two coronoids ( Fig. 1A,B). We further analyzed the mandibular skeleton using 3D imaging. Histologically stained samples were imaged using a CoMBI imaging system 16 . We could generate 3D structures of stained mandibular samples ( Fig. 1C-E). Meckel's cartilage is maintained even in adult axolotls ( Fig. 1C-E). Teeth are lined up on the dentaries but the tooth line(s) on the coronoids is disturbed (Fig. 1C-E, Movie 1). The teeth that are lined up on the mandibular bones usually have small tooth buds on the basal region ( Fig. 1C-D). It is particularly worth noting that some rootless teeth could be seen ( Fig. 1A-E, arrowheads). Rootless teeth were always small, and no attachment could be confirmed at the bottom of any rootless tooth. For this reason, the rootless teeth were easily lost. Histological observation on sections revealed that the Meckel's cartilage is connected with the muscles (Fig. 1F,G, asterisks). The teeth were attached to the dentaries and the tip of each tooth was exposed (Fig. 1F). At the coronoid level, the Meckel's cartilage runs between the dentary and the coronoid (Fig. 1G). The dentary holds the attached tooth and the tooth bud on the inner side (Fig. 1G). On the coronoid, however, the tooth alignment is disturbed. The dentary and the coronoid are thin structures without bone marrow at this point (Fig. 1G). Replacement tooth buds on the coronoid were identifiable in our sections (Fig. 1G,G').
The polyphyodonty of axolotls allows us to see various stages of tooth development within a single sample. Tooth development is initiated by invagination of the inner layer of the dental epithelium. Two thin layers invaginate into the oral mesenchyme ( Fig. 2A). At the end of these invaginated layers, a cup formation develops (Fig. 2B). The cells facing the mesenchymal cells differentiate into a dental papilla (Fig. 2C). The dental papilla cells generate odontoblasts, which deposit the predentine matrix (Fig. 2C). The predentine then mineralizes to become dentine (Fig. 2D). Cell proliferation can be observed in the basal portion of the odontoblast population by this stage (Fig. 2D'). Meanwhile, the tooth has elongated toward the surface of the oral epithelium ( Fig. 2E). At this stage, the pulp cavity becomes apparent. The attachment to the mandibular bone is strengthened by connective tissues (Fig. 2E). A functional tooth is shown in Fig. 2F. No cell proliferation was detected inside the tooth (Fig. 2F'). Each functional tooth is well innervated. Axons, visualized by Acetylated α-tubulin antibody, are readily apparent in the section (Fig. 2G,H,K). The developing tooth buds were also targets of innervation ( Fig. 2I,J,K), and nerves were present throughout this developmental process.
Dentectomy and tooth regeneration. To investigate tooth regeneration in an axolotl mandible, dentectomy was performed by removing the dorsal half of the dentary (Fig. 3, Movie 2). The tooth on the dentary is lined up and easy to target, and there are many detouched teeth and tooth buds (Fig. 3A,A'; Fig. 1). To ensure perfect removal of the tooth organs, including the oral epithelium, the dorsal side of the oral tissues was removed (Fig. 3B,B'). The dentectomized mandible was reconstituted into a 3D image (Fig. 3C-E). The absence of the dorsal half of the mandible is apparent in these images in contrast to those shown in Fig. 1C-E.
To observe the process of tooth regeneration in the axolotl mandible, we sectioned regenerating axolotl mandibles, and investigated their histology and gene expression patterns (Fig. 4). Histological observation revealed that oral epithelium covered the wound surface within 5 days after dentectomy (Fig. 4A,B). No invagination of the covering oral epithelium was observed at this time. Indications of the invagination of the oral epithelium were first found around 10 days after dentectomy (Fig. 4B, arrow). This invaginated epithelium then formed a tooth as in normal tooth development. Shh and Sp7 genes were investigated through in situ hybridization (Fig. 4C,D). Sp7, which has been reported as a late dental marker in vertebrates 17 , was found in the tooth buds in the intact oral tissues (Fig. 4C,the leftmost column). The expression pattern of Sp7 was consistent with Sp7's role in odontoblast differentiation (Fig. 4C). In tooth regeneration, Sp7 expression is observed in relatively later phases of tooth regeneration (Fig. 4C). It was first observed on day 10 around the mandibular bone (Fig. 4C, arrowheads). Later, Sp7 was found in the tooth bud mesenchyme where it contributes to odontoblast differentiation. Thus Sp7 is a mesenchymal marker gene. The other gene we examined, Shh, has been reported as a broad and reliable dental marker in tooth development 18,19 . Consistently, Shh expression was found as early as day 5 in the pre-migratory www.nature.com/scientificreports www.nature.com/scientificreports/ dental epithelium (Fig. 4D, arrowhead). Shh expression can be used to visualize the presumptive dental organs before epithelial invagination. By day 15, Shh expression was observed in the cup epithelium (Fig. 4D). Shh expression was maintained in the later phases of tooth regeneration. The epithelial expression pattern of Shh makes it a useful marker gene for visualizing the whole process of tooth regeneration in axolotls. Fgf and Bmp have been identified previously as nerve factor entities in limb and tail regeneration in axolotls 20,21 . Ectopic Fgf and Bmp application can substitute for the presence of nerve by fulfilling the role of nerves in denervated axolotl limbs 20 . An equivalent functionality of Fgf and Bmp has been reported in multiple organs and species 20,22,23 . Accordingly, we focused on the expressions of these genes in our examination of tooth regeneration in axolotls. Fgf2, Fgf8, Bmp2 and Bmp7 were investigated in the regenerating axolotl jaw (Fig. 4E-H). All genes were broadly expressed in the axolotl mandible. No obvious localizations of gene expression were observed with regard to Fgf2, Fgf8, Bmp2 and Bmp7 in tooth regeneration ( Fig. 4E-H). All four of these genes were also expressed in invaginating epithelial cells. Signals of Fgf2, Fgf8, Bmp2 and Bmp7 expression in the regenerating mandible were relatively ubiquitous, especially in comparison with the negative control (Fig. 4I). This gene expression pattern suggests the involvement of Fgf and Bmp signaling in tooth regeneration. nerve dependency of axolotl tooth regeneration. To examine the structures of nerves in detail, we used βIII-tubulin GFP transgenic axolotls, which are convenient for monitoring axon presence in tissues 24 . Axons projecting to the mandible could be observed in the smaller specimens (Fig. 5). Axons projecting from the trigeminal ganglia toward the mandible were also visible ( Fig. 5A-C). The nerves running into the ventral root exhibit a complex nerve projection pattern. In the proximal region of the dentary, the nerves branch apart into two major routes. In a βIII-tubulin GFP transgenic axolotl, GFP-positive fibers could be confirmed in the . This implies a positive relationship between tooth bud initiation and nerves. We next investigated the roles of nerves in tooth regeneration through denervation experiments. Our denervation procedure targeted the two major nerve routes in the mandibular region ( Fig. 5A-C). In the first denervation, the proximal region of each branch was dissected (Fig. 5A,B). Dentectomy was performed on the same day as the first denervation. Knowing that newly regenerating axons emerge from the dissected ends of nerves and that these newly forming axons are invisible because of their thinness, we also performed a second denervation (on day 10) on the more basal region (Fig. 5A). In the denervated mandible, GFP signals were almost absent initially (Fig. 5G,G'). By day 30, however, a few GFP-positive fibers could be seen (Fig. 5H,H'). On day 45, the axon presence remained much lower in denervated mandibles than in control mandibles (Fig. 5F,F' ,I,I'). These results indicate that our denervation procedure results in an aneurogenic state in the early phase but that innervation is somehow recovered in the later phases.
Regeneration was assessed in innervated and denervated mandibles. On day 25, regeneration was apparent in the control (innervated) mandible and absent in the denervated mandible (Fig. 6). Of note, the dentary bones were well restored in both samples ( Fig. 6A' ,B'), though there were holes in the regenerated dentaries, suggesting that bone regeneration was still ongoing (Fig. 6A' ,B'). The innervated dentary had regenerated teeth by day 25, though these regenerated teeth were immature compared to those in the intact region (Fig. 6A). In the denervated jaw, in contrast, the dentary was restored but the teeth were not regenerated (Fig. 6B). To investigate further, we made sections of the dentectomy samples and observed their histology ( Fig. 6C-F). On day 10, no apparent differences were observed between the innervated and denervated samples (Fig. 6C,E). In control (innervated) samples, the invagination of the dental lamina became confirmable on day 25 (Fig. 6D,D' , arrowheads). In the denervated samples, in contrast, no sign of an invagination of the oral epithelium was observed (Fig. 6F,F'). The extension of the dentary, however, was noticeably advanced on day 25 compared to day 10. This finding indicates that nerve presence is important for tooth regeneration in axolotls.
By day 42, tooth regeneration had somehow been restored in the denervated samples (Fig. 7, Movies 3 and 4). In the control (innervated jaw), tooth regeneration was fully completed by this time (Fig. 7A1-3,C,C'): not only functional teeth but also tooth buds for further replacement were observed (Fig. 7A3, insert). In the denervated mandibles, the dentary bone was regenerated and small teeth were observable in the regenerated region ( Fig. 7B1-3). The regenerated teeth were not attached to the dentary, however, suggesting that these teeth were www.nature.com/scientificreports www.nature.com/scientificreports/ still non-functional ( Fig. 7B1-3). Importantly, we observed differences between the denervated and intact mandibles pertaining to the tooth buds on the coronoids, where dentectomy had not been performed (Fig. 7C,C' ,D,D'). As in the intact mandibles, tooth buds were observed in the basal regions of the functional teeth ( Fig. 7C'), yet the number of tooth buds was apparently decreased in the denervated mandibles (Fig. 7D,D'). This suggests that nerve presence is important for tooth bud initiation not only in a bone regeneration context but also in tooth replacement. www.nature.com/scientificreports www.nature.com/scientificreports/ fgf and Bmp expression in denervated axolotl mandible. To investigate the effects of denervation on the expressions of these genes, we performed in situ hybridization. Dentectomy was performed with or without denervation, and samples were harvested at day 10. The in situ hybridization of Bmp2, Bmp7, Fgf2, and Fgf8 genes showed relatively broad expression patterns (Fig. 8A-J). No apparent differences in Bmp2 and Bmp7 expression were observed between the control and the denervated mandible (Fig. 8A-D). The signal intensity of Fgf2 and Fgf8 appeared to be slightly weakened in the denervated mandible (Fig. 8E-H). To confirm the decrease in Fgf2 and Fgf8 expression, quantitative RT-PCR was subsequently performed (Fig. 8K). Denervation decreased the expression levels of Fgf2 and Fgf8 but had no effects on Bmp2 or Bmp7 expression (Fig. 8K). Therefore, both results suggest that Fgf2 and Fgf8 gene expression were weakened by denervation, but that there were relatively few effects on Bmp gene expressions. www.nature.com/scientificreports www.nature.com/scientificreports/ tooth bud induction by exogenous fgf and Bmp expression in denervated mandible. To investigate the functions of Fgf and Bmp in tooth regeneration in the axolotl mandible, we electroporated Fgf and/or Bmp genes into the denervated axolotl mandible. To achieve this, we designed the following polycistronic expression vectors: FF-mCherry, which carries Fgf8, Fgf2, and mCherry, and BFF-mCherry, which carries Bmp7, Fgf8, Fgf2, and mCherry. Electroporation was performed on day 3 and day 10 after dentectomy and denervation. The genes introduced by electroporation were considered to be temporary. However, since axolotl tooth regeneration requires a relatively long time, we performed the electroporation twice in order to provide gene expression over a sufficient period. We electroporated Fgf and Bmp on the right side of each mandible, while on the left side we electroporated AcGFP vector as a control (Fig. 9A). Clear GFP signals could be seen in the oral mesenchyme. For technical reasons it was difficult to introduce the plasmid into the overlying wound (oral) epithelium. Shh expression was investigated on day 20. In the control side, denervated axolotl mandible failed to induce Shh expressing dental placodes over a large area although a few Shh expressing placode(s) were seen in the proximal region (Fig. 9D,G, Table 1). FF-mCherry and BFF-mCherry gave rise to a relatively faint but confirmable mCherry signal in the oral mesenchyme (Fig. 9B). This signal gradually faded but remained detectable for a month. On the FF-mCherry or BFF-mCherry electroporated side, Shh spots were both clearer and much more abundant (Fig. 9C,E,F,H, Table 1). Both the proportion of successfully inducted Shh spots and the number of Shh spots were increased in BFF-mCherry transfected mandibles www.nature.com/scientificreports www.nature.com/scientificreports/ than in FF-mCherry transfected mandibles (Table 1). Furthermore, Shh expression level was confirmed by quantitative RT-PCR (Fig. 9I). The control sample, which underwent denervation and AcGFP electroporation, showed very low Shh expression. The electroporation with FF-mCherry or BFF-mCherry increased the Shh expression level. Consistent with the results from in situ hybridization, BFF-mCherry electroporation induced higher Shh expression than FF-mCherry electroporation. These results suggest that 1) exogenous Fgfs can fulfill the roles of the primary nerves and can thereby substitute for them in the induction of dental placodes in a denervated mandible and that 2) Bmp signaling plays a supportive role. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Nerves are essential for organ-level regeneration in amphibians 23,[25][26][27] . Their limb, tail, and gill regeneration abilities are dependent on nerve presence. Based on those findings, we focused on the role of nerves in axolotl tooth regeneration. www.nature.com/scientificreports www.nature.com/scientificreports/ Hertwig published the first and fundamental study on tooth developmental events in salamanders 5 . Amphibian tooth morphology and structure have been well described previously, and the concept that amphibian tooth structures are generally similar to those of other vertebrates has been widely accepted 1 . The present study confirmed these insights. The axolotl mandible possesses haplodont dentition. As shown in Fig. 2, each axolotl tooth consists of a pulp cavity surrounded by a dentine cone. Enamel is deposited on the dentine by ameloblasts, which produce the enamel matrix 28 . As reported previously, such fundamental structures are conserved among species. Therefore, axolotl tooth development and regeneration may serve as a valuable model for tooth regeneration in mammals.
It is already widely known that axolotls are capable of tooth regeneration. A valuable description of their tooth regeneration has been published in the context of mandibular regeneration 11,12 , which has traditionally been studied through amputation of part of the upper/lower jaw. The amputated mandible grows a blastema from which the missing portion of the mandible regenerates 29 . The induced blastema then undergoes a mandibular regeneration process that includes tooth regeneration. It has remained unclear to date, however, whether jaw regeneration is an  www.nature.com/scientificreports www.nature.com/scientificreports/ essential condition for tooth regeneration, though it was reported that tooth regeneration started before jawbone reconstitution took place 30 . Consistently, the present study demonstrated that the process whereby new teeth are induced is independent of the dentary (Fig. 6A'). Given this, tooth regeneration is likely independent of jaw regeneration. On the other hand, functional teeth were only seen attached to the dentary, implying a relationship between the axolotl tooth and the dentary. This is as expected, although no report has been published on this relationship in urodele amphibians. Though tooth regeneration may be independent of jawbone regeneration, tooth regeneration may be dependent on the oral mesenchyme. Tooth development in mammals begins with the induction of a dental placode by the oral mesenchyme 31 . This dental placode then provides positive feedback to the mesenchyme, driving tooth development further. This mesenchymal-epithelium interaction has been considered essential for mammalian tooth development. In axolotl tooth regeneration, however, this interaction remains undescribed. What controls the inductive mesenchyme and how it is induced and maintained in tooth regeneration remains an open question.
Innervation in teeth has been conserved widely among species. Our study revealed the presence of axons in teeth in the axolotl mandible. Axons were observed in the pre-invagination oral epithelium (Fig. 5E,F), the developing tooth buds (Fig. 2I,J), and the mature tooth (Fig. 2K). The axons originate from the trigeminal ganglia ( Fig. 5A-C), then from many branches in the proximal region of the dentary (Fig. 5A-C). Our denervation procedure targeted major routes but was not expected to remove all of them. Insufficient removal of axons or axon regeneration may have caused the weak regeneration in the denervated mandible sample on the day 45 (Fig. 7B). The presence of nerves is necessary for successful tooth regeneration and replacement in axolotls as shown by the results depicted in Fig. 7B,D.
The present study also revealed the role of nerve presence in tooth regeneration (Fig. 6). The role of nerves in axolotl dental regeneration/development/replacement has not previously been clarified. The present results, specifically, the finding that, in the denervated mandible, invagination of the oral epithelium was delayed and Shh expression in the pre-invagination oral epithelium was suppressed (Figs. 6F and 9D,G), show that nerves play a role in the tooth bud induction process. Furthermore, axon presence in the pre-invagination oral epithelium follows a very suggestive pattern (Fig. 5E' ,F'). Axon fibers were more densely concentrated in the regions adjacent to the invaginating dental lamina, where the next dental lamina were about to emerge. This suggests that nerves play a role in dental lamina invagination. In keeping with hypothesis, the number of newly formed tooth buds was severely decreased in the denervated mandible. Although the coronoid was not damaged in our denervation procedure, the tooth buds on the coronoid were affected by the denervation. A similar observation has been reported in a bony fish 32 , where denervation prevented tooth turnover. However, innervation and tooth bud induction are not well understood, and further investigation will be necessary on this point.
The origin of the regenerating dental lamina has been the subject of disagreement. Graver claimed that the dental lamina did not arise de novo from the oral epithelium but rather were regenerated from the residual lamina in the stump 14 . Clemen demonstrated that the removal of the dental lamina from the vomer of the palatine resulted in toothless bones 13 . These findings suggest that the dental lamina come from the residual oral epithelium. The present study, however, demonstrates the importance of nerve presence for tooth regeneration in the axolotl mandible. Therefore, it is possible that the results of the previous studies arose in part from unintentional nerve damage caused by the procedures. As shown in Fig. 5A-C, the nerves in this region come from the caudal direction. In Graver's study 14 demonstrating that proximal removal of the dentary resulted in a toothless structure, the removal of the caudal (proximal) part of the dentine would have resulted in severe loss of nerves as well. In our study, nerve loss prevented tooth regeneration in the axolotl mandible (Fig. 6B). Thus similar damage to the nerves may have been the cause of the toothless structures observed in previous studies. Moreover, we removed the oral tissues as shown in Fig. 3. The shape of the portion removed was rectangular, and the stump was located on the short side of the rectangle. We have no reason to presume that there was less migration from the long side of the rectangle and more from the short side. Considering the present results, we think that the origin of regenerating dental lamina may need to be reinvestigated.
Fgf and Bmp signaling are core regulatory cascades in limb, tail, and gill regeneration in Ambystoma mexicanum 20,21,24,33 . The regeneration of each of these appendages depends on the presence of nerves 34 , which express the Fgf and Bmp genes and are assumed to secrete them from the ends of axons 35 . The present study confirms that one additional tissue utilizes this common regulatory system for regeneration: the teeth, which consist of multiple tissues and cell types.
We have shown that nerves are involved in the tooth structure ( Fig. 2I-K) and that the trigeminal ganglia express Fgf and Bmp genes 33 . Denervation results in the inhibition of tooth regeneration (Fig. 6). Ectopic expression of Fgf2 and Fgf8, however, rescues the otherwise nerve-dependent induction of dental placodes, which is enhanced by additional Bmp7 expression (Fig. 9E,H, Table 1). Bmp genes were not influenced by nerve presence (Fig. 8A,H,I), suggesting that Bmp expression in mandible tissues supports tooth regeneration in innervated and denervated axolotl mandibles. On the other hand, electroporation of additional Bmp7 had clear positive effects on dental placode formation in the denervated mandible (Table 1). This implies the presence of a threshold of Bmp signaling in dental placode formation and suggests that Bmps from nerves help to overcome this threshold. The presence of a threshold of Bmp-signaling is supported by similar observations in axolotl tail regeneration 21 . In tail regeneration induced on a lateral tail wound, additional Bmp7 input to Fgf2 and Fgf8 was necessary to induce regeneration even though tail tissues express Bmp genes. Given these findings, it is very likely that Fgfs and Bmp are correlated with nerve presence and play important roles in tooth regeneration.
Do nerves serve as Fgf and Bmp sources? As mentioned above, the trigeminal ganglia express Fgf and Bmp genes, suggesting that the nerves from the trigeminal ganglia are able to serve as the source of Fgfs and Bmps. However, another explanation is possible with regard to tooth regeneration. Nerves may control tooth regeneration by regulating Fgf and Bmp gene expression in the axolotl mandible. Actually, denervation influences Fgf gene expression in the axolotl mandible (Fig. 8). As nerves are known to express many secretory molecules, (2020) 10:9323 | https://doi.org/10.1038/s41598-020-66142-2 www.nature.com/scientificreports www.nature.com/scientificreports/ unidentified factors might be responsible for regulating Fgf and Bmp genes in the axolotl mandible. Further analysis is needed to clarify this point.
It is striking that the regenerative ability of axolotl tails, limbs, gills, and teeth is controlled by a common regulatory system. Furthermore, similar regulation by Fgf and Bmp genes has been reported in the limbs of Xenopus, Pleurodeles waltl, and Ambystoma mexicanum 20,23 . This supports the notion of a conserved regeneration system. Recently, it has been reported that ectopic Fgf signaling can induce intercalary regeneration in chicken limb buds 36 . Likewise, mouse digit tip regeneration can be enhanced by Bmp application 37,38 . These related discoveries imply the existence of a conserved regeneration system across species. If such a conserved regeneration system exists and can be identified, it is likely to lead to innovative medical solutions.
In conclusion, axolotl can regenerate their teeth as reported previously. We renewed and deepened the insights into axolotl tooth regeneration with modern experimental technologies because no updates in this area had been made for several decades. Moreover, we clarified that tooth regeneration is regulated in a nerve dependent manner, but the same is not true of jaw bone regeneration. The nerve roles in tooth bud regeneration can be replaced by Fgf and Bmp gene expression. Such conserved nerve roles and the substitute role of Fgf and Bmp for nerves in organ regeneration in axolotls imply that the conserved mechanisms serve to regulate the regeneration of various organs.

Materials and methods
Animals Axolotls (Ambystoma mexicanum) with nose-to-tail lengths of 8-12 cm were obtained from private breeders and housed in aerated water at 22 °C. Tubulin-GFP transgenic axolotls were kindly provided by the laboratory of Dr. E. Tanaka. In this study, the care and treatment of animals were carried out under protocols approved by the Animal Care and Use Committee of Okayama University. The animal experiments were performed following the guidelines of the animal care and use committee of the Okayama University. All efforts were made to minimize suffering according to the NIH Guide for the Care and Use of Laboratory Animals.
Surgery. Axolotls were anesthetized with MS-222 (Sigma-Aldrich, St. Louis, MO, USA) for about 10 min (depending on the animal size) and placed on ice for 1 h in order to slow their heartbeats. Icing the animals contributes greatly to good surgical recovery. Each animal was laid out with the upper-dorsal region facing up, and dentectomy was performed in the mandibular region. A part of each animal's mandible including the oral epithelium was dissected using forceps and scissors. About half of the dentary was dissected out, while the Meckel's cartilage was left intact. In cases of denervation, axon dissection was performed on the same day as dentectomy. The first denervation points are indicated in Fig. 5B. The second denervation was performed on day 10; the dissection points used in this denervation are shown in Fig. 5A. After each surgery, animals were kept on ice for 2 hours to allow their wounds to heal. All animals were subsequently kept in water.