Dental sensory receptor structure in human teeth

Pain, 13 (1982) 221-235 221 Elsevier Biomedical Press Research Reports Dental Sensory Receptor Structure in Human Teeth Margaret R. Byers ~, Scott J. Neuhaus and John D. Gehrig Center for Research in Oral Biololo~, School of Dentistry, Departments of Anesthesiology, Biological Structure, and Maxillofaciai Surgery, University of Washington, Seattle, Wash. 98195 ( U. S. A. ) (Received 31 July 1981. accepted 23 December 1981) Summa~ F;ectron microscopy was used to study normal human extracted teeth in order to define the junctions between sensory nerve endings and other cells in external pulp and inner dentin at the crown tip. Two sets of associated cells were found: (1) Connective tissue cells. The pulpal fibroblast network made occasional desmosome junctions with the odontoblast network, and the cells of each network formed many gap junctions and desmosomes with one another. (2) N,,rve endings. The terminal axons formed a succession of appositions with each other or with Schwann cells in the plexus of Raschkow and the cell-free zone, possibly with fibroblasts in the cell-free zone and odontoblast layer, and with odontoblasts in the odontoblast layer, predentin and dentin. The appositions between nerve endings and their companion cells at all levels usually maintained a regular intercellular spacing of at least 15 20 nm. In predentin and dentin, axons could be easily identified by their distinctive vesicles and mitochondria, and they often occurred within clusters of adjacent dentinal tubules; in the odontoblast layer axon identification was much more difficult. Axo-axonic appositions were found in the plexus of Raschkow, the cell-free zone, predentin and dentin; in many cases, bare axons were separated from each other only by a 5-10 nm ext-acellular space. Dental sensory mechanisms are discussed in relation to these observations. Introduction Sensory mechanisms in teeth are not yet fully understood, primarily because the interactions between sensory nerve endings and other cells are unclear. The anatomi- i Address correspondence to: Dr. M.R. Byers. Department of Anesthesiology, RN-10. University of Washington, Seattle, Wash. 98195, U.S.A. 0304-3959/82/0000-0000/$02.75 c,~ 1982 Elsevier Biomedical Press 222 cal relationships between sensory nerve endings and odontoblasts have been the most difficult to define mainly because of problems with fixation, sampling and cell identification [1,15]. Some electron microscopic studies have suggested that gap junctions occur between odontoblasts and sensory axons [20-21,26-30,381. How- ever, when axon identification was aided by axonal transport labeling, no axon- odontoblast gap junctions were found [9-10]; instead, there were associations between the two cell types, with parallel apposed membranes and a 20-40 nm intercellular cleft. This difference in identification of cells forming gap junctions may depend on the different species studied (human and cat teeth in the former studies, rat molars in ',he latter), or on difficulties with cell identification in the former studies. From what we know of gap junctions [6,36], if they join axons and odontoblasts, then the two cell types would be electrotonically coupled, and their functions would be interrelated. On the other hand, if sensory axons and odontoblasts are only apposed to one another, there either could be a special interaction at those sites, or the two cell types could merely be apposed because of the narrow space in which they carry out independent functions. Clearly, we need better information on axon-odontoblast junctions in order to understand the degree to which these cells interact. Another important association of sensory nerve endings in pulp and dentin occurs at the many sites at which nerve endings and/or axons are close to each other without an intervening Schwann cell or basal lamina, and with only a 10-20 nm extracellular cleft [9-10,3 !]. Similar appositions occur between trigeminal terminals in corneal epithelium [34] along trigeminal nerves [441 and in other peripheral nerves [39]. Close axo-axonic appositions could allow for ephaptic interaction between sensory axons and may be the structural basis for coupling between axons in teeth [38]; alternatively, axon function may not require Schwann cell insulation at these sites. In this study of extracted human teeth, we have examined the cellular associations of nerve endings in 5 coronal regions: (1) plexus of Raschkow, (2) cell-free zone, (3) odontoblast layer, (4) predentin, (5) dentin. Good preservation of cell structure was achieved by using a rapid tooth cleavage and fixation technique modified from others [2,19,23], by using a formaldehyde-glutaraldehyde mixture for initial fixation, and by block staining with tannic acid and uranyl acetate. Selection of samples from the tip of the pulp horn and adjacent predentin and dentin ensured that the samples contained a high incidence of sensory nerve endings. Materials and Methods Non-carious teeth were obtained from donors, 16-26 years old, who required extractions for prophylactic periodontal or orthodontic purposes at the Oral Surgery Clinic at the University of Washington School of Dentistry. Samples included ! premolar, 2 second molars and 8 third molars. All were normally erupted teeth that had been functional for 1-6 years. Each was immediately immersed in fixative (2% 223 paraformaldehyde plus 1 • or 2~ glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4). The extracted teeth were carried quicHy to an adjacent laboratory where they were split into buccal and lingual halves as follows. A mesial-distal groove, 1-2 mm deep, was drilled on the mesial, occlusal and distal surfaces of the crown and root with a high speed carbide steel (no. 34) inverted cone drill set at 500,000 rev/min. The tooth-was held loosely on the buccal and lhagual sides with pliers to prevent flying fragments, and then side-cutting rongeur forceps were inserted into the notches and used to split the tooth mesio-distally. During the entire drilling and splitting procedures, the tooth was constantly irrigated with cold (4-10°C) fixative from a syringe. All teeth were split open and immediately immersed in fixative within 1-3 min from the time of the extraction. The buccai and lingual halves of each tooth were processed as follows: (A) The pulp horns were removed with fine tweezers from one of the half tooth crowns under a dissection microscope. In some cases, predentin or dentin fragments were left attached to the isolated pulp horn. Extirpated pulps were then cut int~ l-ram-thick transverse slices with razor blades, and fixation continued for 72 h in cold aldehyde fixative. Care was taken to identify slices by keeping them in individual vials during subsequent processing. (B) The remaining halves of the 11 teeth were each placed in a solution of 4~ glutaraldehyde, 10~ EDTA and 0.1 M sodium cacodylate buffer [42] for 6-12 weeks. The fully decalcified teeth were then trimmed to l-ram-thick transverse slices through the pulp horn and adjacent dentin, and were then rinsed in buffer for several days. Next, (!) half of the specimens from group A and group B were postosmicated for ! h (1~ osmium tetroxide, 0.1 M sodium cacodylate buffer, pH 7.4); this group was designated as regularly stained; (2) the remaining half of groups A and B were immersed for 90 rain in a solution of 4~ tannic acid, !% formaldehyde, I% glutaraldehyde, 0.1 M sodium cacodylate solution [41] and then postosmicated (1% OsO4) after reequilibration in 0.1 M buffer. These samples were then equilibrated in 0.2 M cacodylate buffer and block stained in 1~ aqueous urany! acetate solution for I h. All samples were then dehydrated in a graded series of ethanol, embedded in Epon, sectioned, stained with uranyl acetate and lead citrate, and examined in a Philips-300 transmission electron microscope. Results Tissue fixation and sample selection The quality of fixation was variable, but in most cases we could find areas near the tip of the pulp horn with adequate preservation of cell structure, a clear distinction between the 5 regions studied here (Fig. 1) and numerous axons in the nearby pulp (Fig. 2). Occasionally, there was regional retraction of odontoblasts into dentinal tubules (Fig. !). The electron microscope studies were restricted to the areas with the best cellular preservation and the most numerous axon bundles in the plexus and nearby pulp. 224 Fig. i. Light micrograph of external pulp plus attached dentin from tip of crown of undecalcified, regular-stained premolar. The 5 regions examined by electron microscopy i~., this study are shown here: plexus of Raschkow (PR), cell-free zone of Weil (CFZ), odontoblast layer (OD), predentin (PD) and dentin (D). On the left side of this field, some odontoblast nuclei have retracted into dentinai tubules (arrows). Magnification, 225 ×. '~ale bar: 50 #m. Fig. 2. Small myelinated (M) and unmyelinated (A) axons in central pulp near the area shown in Fig. 1. Magn., 405×. Scale bar: 25 #m. Plexus of Raschkow Within the plexus we found fibroblasts, bundles of unmyelinated axons with associated Schwann cells and capillaries (Figs. I, 3-4). Myelinated axons were found close to the plexus but rarely within it. Most bundles of unmyelinated axons in the plexuswere completely surrounded by a Schwann cell process and basal lamina, and beyond that by numerous collagen fiber bundles (Figs. 3-4). Many of the axons were adjacent to one another, separated only by a 15-20 nm space (Figs. 3-4): a similar space separated the axons from Schwann cells. With tannic acid staining, fibrous material was revealed between apposed cells (Figs. 3-6). Cell-free .zone of Weil Withir the 'cell-free' zone there were fibroblast processes, capillaries and un- myelinated axons; most of these elements appeared to be crossing the zone en route to the odontoblast layer. The axon 'bundles' usually contained only 1-2 axons in this zone, and their ensheathing cells lacked a basal lamina; small bundles of collagen accompanied these axons (Figs. 5-6). In this zone, it became increasingly difficult to differentiate between fibroblasts, axons and Schwann cells. The ensheath- 225 ing cells around the axons did not have an associated basal lamina, and some of them might have been fibroblasts rather than Schwann cells. In Fig. 5, for example, two axons were identified because they were enclosed by an ensheathing cell, but two other nearby cell processes could be either free axons or fibroblasts. Sometimes there were apparent axo-axonic appositions in the cell-free zone (Fig. 6) but these were less common than in the plexus: Fibroblast processes could be identified by Figs. 3-4. Plexus of Raschkow. A bundle of unmyelinated axons is shown here from an undecalcified sample stained with tannic acid. The Schwann cell gSC) surrounds most axons but leaves some in contact with one another (A : A). Basal lamina (BL) surrounds the bundle. The outlined area in Fig. 3 is shown at higher magnification in Fig. 4. Scale bar: 0.5 #m. Magn., Fig. 3: 14,850x ; Fig. 4: 36,6C0×. Figs. 5-7. Cell-free zone. Unmyelinated axons (A) were identified in this zone when surrounded by a Schwann cell (SC), but no basal lamina accompanied the Schwann cells. Other cell processes (P) could be axons or fibroblasts. Gap junctions (arrowhead) were sometimes found between identi fled fibroblasts (F). Undecalcified samples stained with tannic acid. Scale bars: 0.5/tm. Magn., Fig. 5:17,100 × ; Figs. 6-7: 42,300 ×. 226 iheir occasional continuity with the fibroblast cell center, by their sheet-like exten- sions or by their cytoplasmic preservation quality. The processes of these identified fibroblasts occasionally formed gap junctions with one another (Fig. 7). Basal lamina was present around the capillaries ia the cell-free zone; so, its absence around Schwann cells or axons was not an artifact of preparation. Odontoblast laver The odontoblast cell bodies were joined to one another by numerous gap junctions and desmosomes, especially along the pulp-predentin border (Figs. 8- I0). In addition, this region contained unmyelinated terminal axons, some fibroblast processes and capillaries. The terminal axons only rarely had an ensheathing cell, so that they were difficult to distinguish from odontoblast or fibroblast processes. When a cell profile contained ribosomes, it was not axonal, since ribosomes are uncharacteristic of axons [39]. Cell profiles that contained numerous varied vesicles and mitochondria were probably axons (Figs. 9-10); their cross-sectioned mitochondria were usually rounder than those of the odontoblasts or fibroblasts. and they had more distinct mitochondrial membranes. Predentm Axon identification was much easier in the predentin than in the odontoblast layer. The odontoblast processes in predentin contained few organeiles except for microtubules and filaments, they were the largest processes in predentin, and they often had patches of dense material on the inner side of their cell membrane (Fig. 11). By contrast, the terminal axons were smaller, they usually contained clear or dense-cored vesicles and mitochondria, and their cytoplasm had a distinctive fixa- tion quality (Figs. 11-15). When tannic acid staining was used in decalcified samples, axonal cytoplasm appeared uniformly dense, the cell membranes were clearly defined, the spacing between the cell processes was narrow, and fibrous material connected them (Fig. 12). With regular staining, the intercellular cleft was usually 15-20 nm wide, except at occasional narrow spots (Fig. 13) or at wider separations (Fig. 14). In most cases, part of the axon-odontoblast apposition curved away from the electron beam axis, so that the cells overlapped obscuring part of the cleft (Figs, 11, 13, 15). Frequently, two or more axons were clustered together (Fig. 15). Dentin In dentin, we found small regions in which several (3-10) adjacent dentinai tubules contained nerve-like profiles (Fig. 16) and then larger areas without many (or any) innervated dentinal tubules. Usually, the innervated tubules contained one large (2-3.5/tm diameter) pale odontoblast process and a much smaller axon-like profile (0.l- l .0 /~m diameter) containing varied vesicles and mitochondria (Figs. !6-17). When cell membranes were clearly visible, the axons and odontoblasts were usually separated by a 15-20 nm spacing (Figs. 17-20), although at a few spots the cleft was only about lO nm wide. Often there were two adjacent axon-like cells in dentinal tubules (Figs. 19-20); the extracellular space separating adjacent bare 227 Figs. 8-10. Odontoblast layer. These micrographs show odontoblasts connected to one another by gap junctions (arrowhead) or desmosomes (d). Several odonloblastic collaterals would have been indis- tinguishable from free axons in other planes of section (.-,). However, in these sections, the conn¢ctior, with odontoblastic cytoplasm containing ribosomes (R) is clear. Two axon-like (A) cells are shown: they contain vesicles, round mitochondria, and no ribosomes. Undecalcified samples, block slained with uranyl acetate (Fig. 8) or regular staining (Fig. 9); decalcified sample (Fig. 10). Scale bars: 0.5 ~tm. Magn., Fig. 8: 30,600X ; Fig. 9" 20,700× ; Fig. i0: 44,100x. axons was typically very narrow (5-10 nm) in '.he dentinal tubules. Rarely, an enlarged axon was found filled with mitochondria and glycogen-like particles (Fig. 19), perhaps indicating the extreme tip of the axon. 228 Figs. II 15. Predentin. Typical odontoblast processes and associated axons (A) are shown here. t)cca~ionally the odontoblast processes were small (Fig. I I; *), but usually they were much larger than the axons and contained paler cyt~plasm composed mostly of microtubules and filaments. Ofter~, patches of dense material were found along the in,ler side of the odontoblast membrane (Fig. I !, arrows). Tannic a~:id staining (Fig. ]2) e,fllanced membrane structure and showed fine fibers connecting adjacent cell processes. Axo,ls in predentin varied m size (Figs. 13-15), they usually contained a variety of vesicles and ,nitochondria and they sometiraes oc~::~rred in clusters (Fig. 15). Decalcified samples; tannic acid stain (Fig. 12) or regular stain (Figs. I I, 13-15). Scale bars: 0.5 ~m. Magn., Fig. I I: 22,950×', Figs. 12-15: 42,300 45,000 -~:. Fig. 16. A cluster of adjacent dentinal tubules at the predentin-dentin border; each tubule contains an axon (A) and an odontoblast process (*). Decalcified sample; regular staining. Magn.. 10,350 ~". Figs. 17-20. Dentin. Typical examples of dentinal axon morphology arc shown here, A 15 20 iml tlef! separated the apposed cells except at occasional narrower spots ~ ~-). In Fi B. 19, an unusually large a~on is filled with mitochondria (m) and glycogen-like particles (g); the small process (p) is pr~bahlv an ax,~n because its fixation quality resembles the adjacent axon. Often, two ax~ms were adjacent (l:ig. 20) separated by a narrow cleft (5-10 rim). Undecaicified samples ~Fig. 17L dt:~alcified samples (Figs 18-20}; all regularly stained. Scale bars: 0.5 pro. Magn., Fi~s, 17 20: 42,3~H~ 45,t)0!~, - 230 Discussion Cell identification The definition of cell junctions involved in dental sensory mechanisms has been difficult, primarily because of problems with cell identification in the odontoblast layer and deiatin. Early studies by Frank [19-21] and Arwill [2] clearly showed that the terminal axons in these areas have no ensheathing Schwann cell or basal lamina separating them from other cells, and that the morphology of the odontoblast collaterals can be indistinguishable from that of axons. In addition, pulpal fibro- blasts send processes across the cell-free zone into the odontoblast layer, which also can resemble terminal axons. In the present study we found that identification of sensory nerve endings in the odontoblast layer was always uncertain, but could have greater reliability in the following circumstances: (!) Terminal sensory axons are beaded unmyelinated structures with numerous agranular vesicles, occasional dense-core vesicles, and mitochondria occupying the enlarged 'bead' regions [2,9- 10,13.-14,21,34]; when a cell profile contained numerous vesicles and mitochondria, it was probably an axon. (2) The terminal axons are clustered in some areas [2,21,32], so that the odontoblast layer and pulp underlying a cluster of innervated dentinal tubules contained a higher incidence of axon-like profiles than did other odontoblastic regions. (3) The fixatives usually reacted differently with odontoblasts, fibroblasts and axons; so that each had a characteristic cytoplasmic density, mitochondrial structure and fixation quality [2,9-10,13-14,21-23]. (4) If ribosomes were found in a cell profile, then it was certainly not an axon. The earlier suggestion that gap junctions join axons to odontoblasts [26-30] seems to have misidentified the axons, since odontoblast collaterals can resemble axons (Figs. 8-10), and since some of the 'pale cell processes which may be axons' were continuous with cytoplasm containing ribosomes and rough endoplasmic reticulum [29]. A few of the axons in these samples may have been sympathetic rather than sensory, since some sympathetic axons end in peripheral pulp or near the basal side of the odontoblast layer [5,40], and would be indistinguishable from sensory axons [39]. However, the low incidence of sympathetic terminals compared with sensor)" endings in external pulp [5], and the apparent lack of sympathetic axons in dentin [3,5], suggests that all endings found in predentin and dentin in the present study were sensory, and that almost all axons in pulp or the odontoblast layer were sensory. In predentin and dentin, cell identification was much easier than in the odonto- blast layer, because of the distinctive morphology of the odontoblast process, the different cytoplasmic density and organelles of the axons and odontoblasts, and the different size of the two cell types. Junctions between connective tissue cells Both odontoblasts and dental fibroblasts derive from neural crest mesoderm [35] and are therefore connective tissue cells. We found numerous desmosomes and gap junctions between adjacent odontoblasts or between fibroblasts, as reported earlier by others [19-22,28]. We also found desmosome contacts between fibroblasts and 231 odontoblasts, and there may also have been gap junctions between fibroblasts and odontoblasts, as suggested by others [21,28]. Each of the types of desmosome nad a characteristic density, those between odontoblasts at the predentin border being most dense. The desmosomes between odontoblasts near preden',in were frequently adjacent to gap junctions, thereby forming complex junctions [19--21]; these were only found between odontoblasts. Junctions formed by dental sensory axons We did not find any evidence that sensory nerve endings form gap junctions with odontoblasts or with each other. This finding agrees with Holland's most recent study of cat canines [32] and with our autoradiographic studies of rat molars [9-10], but contradicts earlier studies by Holland [26,30]. It would, in fact, be extraordinary if gap junctions were present between the mesodermally derived odontoblasts [35] and the ectodermally derived sensory nerve endings, since, in vertebrates, gap junctions rarely occur between dissimilar cells [36]. When such 'heterologous' gap junctions are found, they link cells of similar embryonic origin [36]. The sensory axons in human teeth formed a succession of 'appositions' so that they were usually associated with a companion cell process as they passed from the pulpal plexus through the cell-free zone, the odontoblast layer, predentin, and into dentin, in the plexus and cell-free zone, the unmyelinated sensory axons formed appositions with one another as well as with Schwann cells. The axo-axonic appositions were numerous, they extended for long distances, and they were similar to those described in rat molars [9-10], rat incisors [7], cat teeth [31], or in peripheral nerves [39,44]. Similar axo-axonic appositions in human teeth are evident in the micrographs of others [2,4,14,19,25] but were not specifically mentioned in those papers. With tannic acid stain, fine fibers were found connecting apposed cells together - - probably indicating membrane associated proteins or glycoproteins [42]. This fibrous material was found on all surfaces of the cells, not just at apposition sites; it also had approximately the same appearance at appositions between axoJ~s and Schwann cells as between axons and odontoblasts. In the odontoblast layer, axons identified autoradiographically were found to be apposed to each other by a narrow separatioa of 10-20 nm [9--10] and to the odontoblasts by a wider, more variable separation [2,9-10,21]. The association between ne'ves and odontoblasts can be remarkably regular with almost parallel apposed mcmbranes [2]. The axon-odontoblast associations appear to have adhesive properties, ;ince they were retained in areas of poor fixation or cell shrinkage, in samples in vhich pulp has separated from the odontoblast layer, or during cleavage of the crow l into small fragments [2,19-22]. However, Fearnhead [17] observed easy separation t,f nerves and odontoblasts during cell shrinkage. In prede ltin and dentin, the appositions formed by axons and odontoblasts were as parallel ~s those in the odontoblast layer. The observation that adjacent odonto- blasts also formed parallel appositions (Fig. 1 !) suggests that fixation may induce a parallel alignment of adjacent cells in this tissue. The axo-axonic appositions had an especially narrow extracellular cleft in preden- tin and dentin, perhaps to allow ephaptic communication betweea terminals for 232 synchronization of nerve signals, as suggested earlier [9-10,31]. Alternatively, fixa- tion may affect the axons in the dentinal tubules differently from those in pulp. Another possibility is that the close juxtaposition of neighboring terminal sensory axons can occur because sensory endings may not need Schwann cell insulation there to function properly. Similar appositions occur between corneal sensory endings [34], are common all along trigeminal nerve [31,44], and have been found in a wide variety of teeth [2,9-10,19-22]. Distribution of dentinal nerve endings Frank [21] reported that axon occurrence in dentinal tubules was highest in electron microscopic samples taken adjacent to the tip of the pulp horn, confirming Fearnhead's light microscopic data on the distribution of dentinal nerve endings [16]. In the present study, we found that the samples from the well-innervated coronal zone also contained many small 'patches' Of dentin in which the number of innervated dentinal tubules was high, often with 3-10 adjacent tubules innervated (Fig. 16); similar 'clusters' have been reported cartier [2,4,13,32]. In the present study we could not determine the distance that nerve endings or axons penetrated into dentin because fixation quality deteriorated beyond about 50 /~m into dentinal tubules. It is likdy, however, that human teeth are similar to monkey and cat teeth, for which recent autoradiographic studies show terminal axons penetrating as far as 150 #m into dentinal tubules [1 i,12]. Dental sensory mechanisms Several different cellular mechanisms have been suggested for the transduction of dental stimuli into nerve imptllses. (1) Odontoblastic transduction. It has been suggested for many years that the odontoblast is the primary receptor cell, and that it transduces extraceilular stimuli into signals that are then conveyed to the nearby sensory nerve endings [for review: 1,15,18]. In support of this theory is the observation that odontoblasts extend much closer to the sensitive dentino-enamel junction than do axons, in addition, there is some evidence for electrotonic coupling between axons in teeth and it has been suggested that this coupling is mediated by neuron-odontoblast gap junctions [26-30,38]. However, as mentioned earlier, evidence for such gap junctions is lacking in detailed EM studies [32] or autoradiograp~e studies [9-10]. In addition, Holland's studies of odontoblast processes show that they do not extend into outer dentin of mature teeth [27] and so they cannot reach the dentino-enamel junction either. (2) Independent neural transduction. It has been suggested that nerve endings in dentin, predentin and pulp may act independently to initiate nerve impulses [for review: 15]. This possibility is supported by the observation that electrical stimula- tion of dentin can activate a radial nerve inserted into a pulpless tooth [37]. Also, no changes in membrane potential were detectable by intracellular recording of cells in the odontoblastic layer during sensory transduction [43]. Finally, in rat incisors, sensory axons end near moribund odontoblasts [7]. (3~ Odontoblastic modut.ation of neural transduction. There are several ways in which odontoblasts might affect neural transduction depending upon the mechanism by which stimuli transfer through outer dentin to the underlying cells. If movement 233 of fluid in dentinai tubules is the key mechanism, as proposed by the hydrodynamic theory [8], the odontoblasts may provide a framework to hold sensory axons in the optimal position to detect fluid flow. At the very least, the odontoblasts must affect hydrodynamic events since they maintain dentinal matrix and tubules. If dentinal tubules become dried out, as in dentinal 'dead tracks," sensitivity of that region is reduced. It is interestillg that changes in pulpal fluid pressure can accompany increased dental sensitivity [24]. If the key extracellular event is the generation of electrical currents via piezoelec- tric, pyroelectric, or other mechanisms, odontoblasts could affect neural transduc- tion (a) by providing a framework to allow axons to position thems,~lves for optimal detection of such currents, or (b) by controlling specific extracellular electrolytes, thereby raising or lowering the excitability of the axons. Finally, transduction may involve chemical events [33] as well as electrical and/or mechanical ones. Odontoblastic modulation therefore could occur in more than one way. In conclusion, our studies of cell junctions in human teeth did not find gap junctions between axons and odontoblasts, but only between odontoblasts or fibroblasts. The use of tannic acid staining did not reveal unusual membrane-associ- ated material at the neuron-odontoblast or axo-axonic appositions. The similarity between the structure of sensory nerve endings in rodent, fish, cat and monkey teeth [9-13,22] and that found in human teeth suggests that all teeth have the same dental sensory mechanism(s). It is probable that the sensory nerve endings are the primary receptor; but their acti~,ity may be modulated by the odontoblasts. Acknowledgements We thank Dr. L.E. Westrum and Ms. Alison Ross for helpful advice, and P. O'Neill and J. Anderson for technical assistance. This work was supported by NtH Grants DE05159, DE00099, DE02600, and RR05346. References 1 Anderson, D.J., Hannam, A.G. and Matthews, B., Sensory mechanisms in mammalian teeth and their supporting structures, Physiol. Rev., 50 (1970) 171-195. 2 Arwill, T., Studies on the ultrastructure of dental tissue. I1. The predentine-palpal border zone. Odontol. Rev., 18 (1967) 191-208. 3 Arwill, T., Edwall, L., Lilja, J., Olgart, L. and Svensson, S.E., Uitrastructure of nerves in the dentinal-pulp border zone after sensory and autonomic nerve transection in the cat, Acta odontol. stand., 31 (1973) 273-281. 4 Avery, J.K., Structural elements of the young normal human pulp, Oral Surg., 32 (197 !) I 13-125. 5 Avery, J.K., Cox, C.F. and Chiego, D.J., Presence and location of adrenergic nerve endings in the dental pulps of mouse molars, Anat. Rec., 198 (1980) 59-71. 6 Bennett, M.V.L. and Goodenough, D.A., Gap junctions, electrotonic coupling and intercellular communication, Neurosci. Res. Progr., 16 (1978). 234 7 Bishop, M,A., A fine-structural survey of the pulpal innervation in the rat mandibular incisor, Amer. J. Anat., 160 (1981) 213-229. 8 Brannstrom, M,, Linden, L.A. and Astrom, A., The hydrodynamics o~the dental tubule and of pulp fluid, Caries Res., ! (1967) 310-317. 9 Byers, M.R., Fine structure of trigeminal nerve endings in rat molars. In: D.J. Anderson and B. Matthews (Eds.), Pain in The Trigeminal Region, Elsevier, Amsterdam, 1977, pp. 13-25. 10 Byers, M.R., Large and small trigeminal nerve endings and their associations with odontoblasts in tat molar dentin and pulp, Adv. Pain Res. Ther., 3 (1979) 265-270. ! I Byers, M.R. and Dong, W.K., Autoradiographic demonstration of sensory nerve endings in monkey teeth, Pain, Suppl. 1 (1981) 2018. 12 Byers, M.R. and Matthews, B., Autoradiographic demonstration of ipsilateral and contralaterai sensory nerve endings in cat dentin, pulp and periodontium, Anat. Rec., 201 (1981) 246-260. i 3 Corpron, RE. and Avery, J.K., The ultrastructnre of intradental nerves in developing mouse molars, Anat. Rec., 175 (1973) 585-606. 14 Dahi, E. and Mjor, I.A., The structure and distribution of nerves in the pulp-dentin organ, Acta odontol, scand., 31 (1973) 349-356. 15 Dubner, R., Sessle, B.J. and Storey, A.T., Tooth, pulp and dentin. In: R. Dubner, B.J. Sessle and A. Storey (Eds.), The Neural Basis of Oral and Facial Function, Plenum Press, New York, 1978, pp. 109-146. 16 Fearnhead, R.W., Histological evidence for the innervation of human dentine, J. Anat. (Lond.), 91 (1957) 267-277. 17 Fearnhead, R.W., The neurohistology of human teeth, Proc. roy. Soc. Med., 54 (1961) 877-884. 18 Fearnhead, R.W., Innervation of dentinal tissues. In: A.E.W. Miles (Ed.), Structure and Chemical Organization of Teeth, Vol. !, Academic Press, New York, 1967, pp. 247-281. 19 Frank, R.M., Etude au microscope ~lectronique de rodontoblaste et du canalicule dentinaire humain, Arch. Oral Biol., ii (1966) 179-199. 20 Frank, R.M., Attachment sites between the odontoblast process and the intradentinal nerve fibres, Arch. Oral Biol., 13 (1968) 833-834. 21 Frank, R.M., Ultrastructural relationship between the odontoblast, its process and the nerve fibre. In: N.B.B. Symons (Ed.), Dentine and Pulp: their Structure and Functions, Livingstone, London, 1968, pp. 115-145. 22 Frank, R.M., Sauvage, C. and Frank, P., Morphological basis of dental sensitivity, Int. dent. J., 22 (1972) 1-19. 23 Furseth, R. and Mjor, I.A., Electron microscopy of human dentin, Acta odontoi, scand., 27 (1969) 577-593. 24 GreenhiU, J.D. and Pashley, D.H., Effects of desensitizing agents ~n the hydraulic conductance of human dentin in vitro, J. dent. Res., 60 (1981) 686-698. 25 Harris, R. and Griffin, C.J., Fine structure of nerve endings in the human dental pulp, Arch. Oral Biol., 13 (1968) 773-778. 26 Hollard, G.R., Membrane junctions on cat odontoblast, Arch. Oral Biol., 20 (1975) 55 !-552. 27 Holland, G.R., An ultrastructural survey of cat dentinal tubules, J. Anat. (Lond.), 122 (1976) i - ! 3. 28 Holland, R.G., Structural relationships in the odontobla~t layer. In: D.J. Anderson and B. Matthews (Eds.), Pain in the Trigeminal Region, Elsevier, Amstert~am, 1t~77, pp. 25-36. 29 Holland, G.R., Figures of junctions in the odontoblast layer. In: R. Dubner, B.J. Sessle and A. Storey (Eds.), The Neural Basis of Oral and Facial Function, Plenum, New York, 1978, pp. ! 19-120. 30 Holland, G.R., Microtubule and mierofilament populations of cell processes in the dental pulp, Anat. Rec., 198 (1980) 421-426. 31 Holland, G.R., The sub-odontoblas~ic ~-ellral plexus of the cat, J. Anat. (Lond.), 130 (1980) 457-467. 32 Holland, G.R., The incidence of dentinal tu~,des containing more than one process in the cuspai dentin of cat canine teeth, Anat. Rec., 200 (1981, 437-442. 33 Horiuchi, H. and Matthews, B., Response of intra~lental nerves to chemical and osmotic stimulation of dentine in the cat, Pain, 2 (1976) 49-59. 34 Hoyes, A.D. and Barber, P., Ultrastructure of cortical receptors. In: D.J. Anderson and B. Matthews (Eds.), Pain in the Trigeminal Region, Elsevier, Amsterdam, 1977, pp. 1-12. 235 35 Johnston, M.C., Bhakdinaronk, A. and Reid, Y.C., An expanded role of the neural crest in oral and pharyngeal development. In: J.F. Bosma (Ed.), Fourth Symposium on Oral Sensation and Perception, DHEW Pubfication No. N1H 75-546, Bethesda, Md, 1973, pp. 37-52. 36 Larsen, W.~., Structural diversity of gap junctions, a review, Tissue and Cell, 9 (1977) 373-394. 37 Matthews, B., Nerve impulses recorded from dentine in the cat, Arch. Oral Biol., 15 (1970) 523-530. 38 Matthews, B. and Holland, R.G., Coupling between nerves in teeth, Brain Res., 98 (1975) 354-363. 39 Ochoa, J., The unmyelinated nerve fibre. In: D.N. Landon (Ed.), The Peripheral Nerve, Wiley and Sons, New York, 1976, pp. 106-158. 40 Pohto, P., Sympathetic adrenergic innervation of permanent teeth in the monkey, Acta odontol. scand., 30 (1972) 117-126. 41 Thomas, H.F., The effects of various fixatives on the odontoblast process, J. dent. Res., 59A (1980) 403. 42 Van Deurs, B., Use of a tannic acid-glutaraldehyde fixative to visualize gap and tight junctions, J. ultrastruct. Res., 50 (1975) 185-192. 43 Winter, H.F., Bishop, J.G. and Dorman, H.L., Transmembrane potentials of odontoblasts, J. dent. Res., 42 (1963) 594-598. 44 Young, R.F., Fiber spectrum of the trigaminal sensory root of frog, cat and man determined by electronmicroscopy. In: D.J. Anderson and B. Matthews (Eds.), Pain in the Trigeminal Region. Elsevier, Amsterdam, 1977, pp. 137-147.