Parvalbumin-, substance P- and calcitonin gene-related peptide-immunopositive axons in the human dental pulp differ in their distribution of varicosities

Information on the frequency and spatial distribution of axonal varicosities associated with release of neurotransmitters in the dental pulp is important to help elucidate the peripheral mechanisms of dental pain, mediated by myelinated versus unmyelinated fibers. For this, we investigated the distribution of axonal varicosities in the human dental pulp using light- and electron-microscopic immunohistochemistry for the vesicular glutamate transporter 2 (VGLUT2), which is involved in the glutamatergic transmission, and syntaxin-1 and synaptosomal nerve-associated protein 25 (SNAP-25), combined with parvalbumin (PV), which is expressed mostly in myelinated axons, and substance P (SP) and calcitonin gene-related peptide (CGRP), which are expressed mostly in unmyelinated axons. We found that the varicosities of the SP- and CGRP-immunopositive (+) axons were uniformly distributed throughout the dental pulp, whereas those of PV+ axons were only dense in the peripheral pulp, and that the expression of PV, VGLUT2, syntaxin-1, SNAP-25, SP and CGRP was significantly higher in the varicosities than in the axonal segments between them. These findings are consistent with the release of glutamate and neuropeptides by axonal varicosities of SP+ and CGRP+ unmyelinated fibers, involved in pulpal pain throughout the human dental pulp, and by varicosities of PV+ fibers, arising from parent myelinated fibers, and involved in dentin sensitivity primarily in the peripheral pulp.

For this, we here investigated the distribution of varicosities in the human dental pulp from human premolar teeth using light microscopic immunohistochemistry for VGLUT, combined with parvalbumin (PV), which is expressed mostly in parent myelinated axons, and substance P (SP) and calcitonin gene-related peptide (CGRP), which are expressed mostly in unmyelinated axons, in the sensory root of the trigeminal ganglion 17,18 . We also used electron microscopic immunohistochemistry to examine the varicosities that expressed PV, SP, CGRP, VGLUT2, and the SNARE proteins syntaxin-1 and synaptosomal nerve-associated protein 25 (SNAP-25).

Results
Distribution of axonal varicosities within the dental pulp. Most axons in the dental pulp that were immunostained for PV, SP or CGRP were within axonal bundles in the radicular pulp and occasionally in the core of the coronal pulp. The PV-immunopositive (PV +) axons branched extensively out of these bundles and towards the dentin, whereas the SP+ and CGRP+ axons branched frequently in the peripheral pulp, distal to the parietal nerve plexus (Figs. 1, 2).
While only few PV+ axons in the radicular pulp and the core of the coronal pulp had any varicosities, virtually every PV+ axon in the peripheral pulp displayed multiple varicosities. The SP+ and CGRP+ axons, on the other hand, did not show such regional differences and had equally dense varicosities throughout the dental pulp (Figs. 1,2). Quantitative analysis revealed that the number of varicosities per unit length of the PV+ axons was significantly higher in the peripheral pulp than in the core of the coronal and the radicular pulp. Conversely, for the SP+ and CGRP+ axons, it was similar among the three pulpal regions. The number of varicosities per unit length for PV+, SP+ and CGRP+ axons was similar in the peripheral pulp, whereas it was significantly higher for the SP+ and the CGRP+ axons than for the PV+ axons in the core of the coronal and radicular pulp ( Table 1). The VGLUT2+, syntaxin-1+ and SNAP-25+ axons had numerous branches and varicosities in the peripheral pulp (Fig. 3). In the core of the coronal and the radicular pulp, many VGLUT2+ axons but few syntaxin-1+ or SNAP-25+ axons had varicosities. In addition, double immunofluorescent staining revealed that the majority of SP+ and CGRP+ axons and few of the PV+ axons expressed VGLUT2 (Fig. 4). The fraction of VGLUT2-expressing SP+ (32.6 ± 7.8%) and CGRP+ (39.2 ± 5.9%) axons was significantly higher than that of VGLUT2-expressing PV+ axons (6.2 ± 3.6%). PV+ axons did not co-stain for SP or CGRP, suggesting that in the dental pulp, PV is expressed in a different axon type than CGRP or SP (Fig. 5).
Ultrastructure of the axonal varicosities in the dental pulp. At electron microscopy, axonal varicosities were observed on unmyelinated fibers and on unmyelinated segments of myelinated fibers: Myelinated fibers entered the dental pulp as myelinated and then lost their myelin. The varicosities but not the axonal segments between them contained numerous small vesicles and mitochondria; the varicosities of SP+ and CGRP+ axons also contained large vesicles with dense cores. Varicosities were not covered by Schwann cells, which can be favorable structure for the release of neurotransmitters. The density of immunostaining for PV, SP, CGRP, VGLUT2, syntaxin-1 and SNAP-25 was significantly higher in the varicosities than in the interim segments of the axon, pointing to varicosities as a likely site of accumulation and release of glutamate and neuropeptides in the dental pulp (Fig. 6).

Discussion
The main findings of the present study are that while the varicosities of the SP+ and CGRP+ axons were uniformly distributed throughout the human dental pulp, those of PV+ axons were only dense in the peripheral pulp, and that the expression of SP, CGRP, VGLUT2, syntaxin-1 and SNAP-25 was significantly higher in the varicosities than in the axonal segments between them. These observations support the hypothesis that axonal varicosities are sites of release of glutamate and neuropeptides by SP+ and CGRP+ unmyelinated fibers, involved in pulpal pain throughout the human dental pulp, and PV+ fibers, arising from parent myelinated fibers and involved in dentin sensitivity in the peripheral pulp only. The axonal varicosities in the dental pulp contained vesicles and mitochondria and densely expressed VGLUT2, syntaxin-1 and SNAP-25, which are involved in vesicular loading and exocytosis of neurotransmitters www.nature.com/scientificreports/ and the neuropeptides SP and CGRP, which in turn, have been shown to mediate nociception and neurogenic inflammation [2][3][4]19 . This suggests that the axonal varicosities represent the site of release of glutamate and neuropeptides in the dental pulp, analogous to the varicosities of autonomic nerves, which also contain numerous vesicles and mitochondria, express SNARE proteins, and do not establish synaptic contact with adjacent effector cells 20,21 . Several lines of evidence suggest that the glutamate signaling by peripheral sensory axons plays a crucial role in acute nociception and pathological pain [2][3][4]22 . Thus, direct application of glutamate to peripheral tissues induces excitation of nociceptive Aδ and C fibers and their subsequent sensitization 3 , and injection of glutamate into deep facial tissues evokes acute pain and induces mechanical allodynia and heat hyperalgesia 4,23,24 . In addition, it has been shown that the pulpal axons express the metabotropic glutamate receptor mGluR5 and that its expression is increased following pulpal inflammation 8,25 . By analogy, acute and pathologic dental pain may be mediated by glutamate released from the varicosities of the pulpal axons. www.nature.com/scientificreports/ That the majority of SP+ and CGRP+ axons, but few of the PV+ axons coexpressed VGLUT2, which is analogous to the respective neurons in the TG 26 , suggests that the VGLUT2-dependent glutamate release in the dental pulp is mediated primarily by SP+ and CGRP+ unmyelinated axons, and to a much lesser degree by PV+ axons, arising from parent myelinated axons. Considering that the VGLUT1+ primary sensory neurons express mostly PV, NF200, and SSEA4, which are expressed in neurons with myelinated axons, and rarely CGRP and IB4, which are expressed in neurons with unmyelinated axons [26][27][28] , the pupal PV+ axons arising from parent myelinated axons are likely to express VGLUT1 and be involved in the VGLUT1-dependent glutamate release.
Considering that in the sensory root of TG 98.8% of the PV+ parent axons are myelinated, and 99.5% of the SP+ and 96.5% of the CGRP+ parent axons are unmyelinated 17,18 , PV+ fibers may arise from parent myelinated A fibers, and SP+ and CGRP+ fibers may arise from unmyelinated C fibers. Since it is typical for myelinated fibers to lose their myelin sheath and become thinner during their course to the peripheral target 29 , PV+ fibers lose their myelin during their course to the dental pulp. Thus, ~ 66%, 79% and 99% of the PV+ parent fibers, which are myelinated in the sensory root of the TG, become unmyelinated at the root pulp, coronal pulp and peripheral pulp, respectively 30 . Varicosities were observed within the unmyelinated segment of the PV+ fibers, but not within the their myelinated segment.
PV is expressed by a small subset of proprioceptive neurons in the dorsal root ganglion (DRG) 18,31,32 , but in a large number of low-threshold mechanoreceptive and nociceptive neurons in the TG 18,33 . Also, the human dental pulp, which has very few, if any, proprioceptors [34][35][36] , is densely innervated by a large number of PV+ axons 30 . Considering these findings, PV may label large number of pulpal axons arising from parent myelinated axons. SP and CGRP have been routinely used as markers for primary sensory neurons with unmyelinated fibers 37 even though they have been shown to also label some DRG neurons with myelinated fibers 38,39 . That PV+ pulpal axons did not co-stain for SP or CGRP, which is consistent with findings in the rat dental pulp 40 , suggests that PV and SP/CGRP are expressed by distinct axon types.
In agreement with our previous observations, the PV+ axons branched extensively in the peripheral pulp distal to the parietal plexus of nerves, and many of them extended towards the dentinal tubules 30 . That almost all PV+ axons possessed numerous varicosities in the peripheral pulp but few had any varicosities in the radicular pulp and the core of the coronal pulp suggests that glutamate released from numerous varicosities of PV+ axons, following excitation in the dentin and peripheral pulp, can induce dental hypersensitivity by activating adjacent and nearby axons. This possibility is supported by the observation that numerous axonal varicosities in the peripheral pulp also express mGluR5 8 . On the other hand, that there are few PV+ axonal varicosities in the radicular pulp and the core of the coronal pulp is consistent with the observation that a considerable fraction of PV+ axons in these regions of the dental pulp are myelinated and thus do not express functional glutamate receptors and are unlikely target for either autocrine or paracrine activation and/or sensitization by glutamate 30 .
The peptidergic (SP+ or CGRP +) neurons in the TG coexpress the markers for glutamatergic transmission VGLUT 26,28,41,42 and glutaminase 43 . We also showed previously that pulpal SP+ or CGRP+ axons frequently coexpress VGLUT2 9 . Taken together, these findings suggest that pulpal peptidergic C fibers may release glutamate as well as neuropeptides. Also, since at variance with the PV+ varicosities, the SP+ and CGRP+ varicosities were uniformly distributed throughout the dental pulp, the co-release of glutamate and neuropeptides and the subsequent autologous activation of these fibers or of adjacent fibers during inflammation, is more likely to occur uniformly throughout the dental pulp and not be confined to the peripheral pulp.  Tissue preparation. We used maxillary premolars from nine human subjects, 15-28 years of age; the teeth were healthy and were extracted for orthodontic indication at the Department of Oral Surgery, Kyungpook National University Hospital. The pulps were extracted in toto by cutting the teeth along their longitudinal axes. The tissues were fixed for 5 h in a solution of 4% paraformaldehyde in phosphate buffer (PB, 0.1 M, pH 7.4) for light microscopy (6 pulps) or in a mixture of 4% paraformaldehyde and 0.01% glutaraldehyde for electron microscopy (3 pulps). The tissues were then immersed in a solution of 30% sucrose in PB at 4 °C. On the next day, 20-30 µm-thick sections for light microscopy were cut on a cryotome, and 50-60 µm-thick sections for electron microscopy were cut on a Vibratome (Leica Biosystems, Wetzlar, Germany).
Light microscopic immunohistochemistry. Two protocols were used to prepare sections for light microscopy, immunoperoxidase for single staining and immunofluorescence for double staining. Quantitative analysis. The number of varicosities per unit axonal length was determined using sections stained with immunoperoxidase. Images were obtained at 40 × with an Exi digital camera (Q-imaging Inc., Surrey, CA) attached to a Zeiss Axioplan 2 microscope (Carl Zeiss, Göttingen, Germany), and saved as TIFF files. The number of varicosities and the length of the PV+, SP+ and CGRP+ axons in each peripheral, coronal, and radicular pulp were measured from a total of 12-16 images from each pulpal region in 3-4 sections of each of three human dental pulps. Continuous strings of axonal beads with an apparent linear arrangement were considered a single axon.
The fractions of PV+, SP+ and CGRP+ axons that coexpress VGLUT2 were analyzed using sections stained with double immunofluorescence. Images were taken with a confocal microscope (LSM 510 Meta; Carl Zeiss) at 40 × at a same optical slice thickness for all channels, and saved in TIFF format. The PV+, SP+ and CGRP+ axons that coexpress VGLUT2 in the peripheral and coronal pulp were counted from a total of 12-16 images from 3-4 sections of each of three human dental pulps. All axons longer than 1 cm in the images (> 45 µm in tissue) were counted. Continuous strings of axonal beads with an apparent linear arrangement were considered to represent single axons. The differences among the three pulpal regions, and among the PV+, SP+ and CGRP+ axons was examined by one-way analysis of variance (ANOVA) and Scheffe's F-test (significance was set at p < 0.05).

Electron microscopic immunohistochemistry.
To improve penetration of antibodies for electron microscopy, sections were cryoprotected in 30% sucrose in PBS overnight, frozen on dry ice for 30 min, and thawed in PBS. They were then pretreated in PBS-buffered solutions of 1% sodium borohydride for 10 min, 3% H 2 O 2 for 10 min, and 10% NDS for 30 min. After that, the sections were transferred to a solution of the primary antibody, mouse anti-PV at 1:1,000, rat anti-SP at 1:500, mouse anti-CGRP at 1:1,000, guinea pig anti-VGLUT2 at 1:500, mouse anti-syntaxin-I at 1:2,000 or rabbit anti-SNAP-25 at 1:3,000 for an overnight incubation at room temperature. After several rinses in PBS, sections were transferred to 2% NDS for 30 min and then to an appropriate biotinylated secondary antibody, donkey anti-mouse, donkey anti-rat, donkey anti-guinea pig or donkey anti-rabbit at 1:200 for 2 h. Avidin-biotin-peroxidase binding was with ExtrAvidin peroxidase (Sigma-Aldrich) at 1:5,000 for 1 h, and peroxidase was revealed according to a standard diaminobenzidine (DAB) protocol. After additional staining in 1% osmium tetroxide in PB for 1 h, the immunostained sections were dehydrated with a series of ethanol dilutions, and embedded in Durcupan ACM (Fluka, Buchs, Switzerland). The resin was cured Scientific RepoRtS | (2020) 10:10672 | https://doi.org/10.1038/s41598-020-67804-x www.nature.com/scientificreports/ in a 60 °C oven for 48 h. One square mm chips of the embedded tissue were cut out and glued onto support resin blocks with cyanoacrylate. Thin sections, cut with a diamond ultramicrotome knife (Dupont), were mounted in series on formvar-coated Ni grids and the grids were counterstained with uranyl acetate and lead citrate and examined on a Hitachi H-7500 electron microscope (Hitachi, Tokyo, Japan) at 80 kV. Photomicrographs were taken with a SC1000 CCD camera (Gatan, Pleasanton, CA) and saved as TIFF files using DigitalMicrograph software.
Immunohistochemical controls. All antibodies used in this study have been used in our laboratory and their specificity has been confirmed in previous experiments 17,18,26,30 . Nevertheless, we routinely processed sections where we omitted either the primary or the secondary antibodies. Sections incubated without the primary or secondary antibodies completely lacked specific staining. We also immunostained sections after preadsorption with blocking peptides, according to the respective supplier's protocol: for example, sections incubated with 18 µg/mL PV blocking peptide lacked specific immunostaining for PV. With electron microscopy, we studied the immunostaining in adjacent serial thin sections: consistent staining in serial sections of the same tissue elements confirmed specificity.

Data availability
The datasets generated and/or analyzed in the course the current study are available from the corresponding author on request.