- Email: [email protected]
Replacement of receptor neurones after section of the vomeronasal nerves in the adult mouse
Brain Research, 147 (1978) 297-313 © Elsevier/North-Holland Biomedical Press
297
REPLACEMENT OF RECEPTOR NEURONES AFTER SECTION OF THE VOMERONASAL NERVES IN THE A D U L T MOUSE
P. C.
BARBER and G. RAISMAN
Laboratory of Neurobiology, National Institute for Medical Research, Mill Hill, London NW7 1.4A (England)
(Accepted September 19th, 1977)
SUMMARY Eight days after vomeronasal nerve section or removal of the accessory olfactory bulb, the majority of receptor cells of the vomeronasal neuroepithelium degenerate and disappear, leaving a regular framework consisting of supporting cells and their radial processes. The cell clusters at the boundaries of the epithelial sheet (which have been shown to be actively dividing in the normal, unoperated adult mouse) are also spared. The epithelium is subsequently repopulated by receptor cells appearing first in the basal part of the receptor cell layer and later occupying the full width of the receptor layer. These cells are anatomically fully differentiated receptor cells with normal sensory dendrites. Their axons form conspicuous intraepithelial neuromatous masses. Administration of [3H]thymidine on days 10-20 postoperatively labels some clusters of supporting cells and virtually all of the receptor cells, indicating that the repopulation of the epithelium is due to new formation of receptor cells.
INTRODUCTION It is well established that the olfactory chemoreceptor neurones undergo a reaction of retrograde degeneration when subjected to axotomy by severing the olfactory nerves or removing the olfactory bulb. They can also be destroyed by direct application of toxic chemicals or by mechanical interference. There is now evidence that after such loss there may be replacement of these sensory neurones from a stem cell with restoration of normal structure of the olfactory epithelium and possible return of olfactory function, even in adult animals (see ref. 35 for a review). Nagahara 25 in the mouse, saw extensive degeneration of olfactory receptor cells 3 days after olfactory nerve section or bulbectomy, followed by restoration of normal mucosal cellularity in 90 days. He proposed that the basal part of the mucosa contains 'resting cells', presumably without axons, which persist after the original
298 neurosensory cells have degenerated, and which give rise to new neurosensory cells by mitosis. After the destruction of the olfactory mucosa by intranasal irrigation with 1 ~ zinc sulphate, various authors have reported regeneration of a normally arranged neuroepithelium in the frog 33, rabbit 24, mouse 20 and monkey30, 31, although an earlier study 82 failed to show regeneration in the rat. In monkeys, Schultz3O, 31 found that regeneration of the olfactory mucosa after zinc sulphate took 3-4 months, and corresponded in time to the return of susceptibility to infection by intranasal inoculation of a neuronotropic strain of poliovirus to which zinc sulphate treatment had earlier conferred resistance. This suggested that mucosal regeneration was accompanied by axonal regrowth to the olfactory bulb, thus providing a pathway for the virus infection. Recent studies using systemically injected tritiated thymidine have demonstrated in normal adult animals a continuous formation and presumably therefore a turnover of olfactory receptors from a basally located stem cell (see ref. 2 for refs). It has also been shown in the frog that after receptor degeneration produced by olfactory nerve section, mitosis and thymidine labelling of the basal cells is greatly stimulated, and that labelled cells repopulate the epithelium, coming to resemble olfactory receptors, both light and electron microscopically7. Axons, presumably arising from the new receptors, have been observed to grow back and to reestablish synaptic contacts in the glomeruli of the olfactory bulb s. In the pigeon after olfactory nerve section, Oley et al. 26 demonstrated restoration of normal morphology of the olfactory mucosa, regeneration of the olfactory nerves and their synapses in the bulb, and recovery of the ability to perform olfactory discrimination tasks. A series of biochemical experiments by Margolis and his collaborators11,16-1s also provide strong evidence for the regenerative replacement of olfactory receptor cells in the adult mouse. The receptor cells contain a unique 'olfactory marker' protein 16 whose synthesis occurs only in the olfactory epithelium t8. This protein, together with an unusual dipeptide, carnosine (fl-alanyl-L-histidine)17 and the enzyme carnosine synthetase, are present in the receptor cells in the olfactory epithelium and in their axons and axon terminals in the olfactory bulb. Within the first few days after unilateral olfactory nerve section 11, all three biochemical markers disappear from the olfactory epithelium, during which time the histology shows that the receptor cells degenerate. Over the period from 2 to 4 weeks after operation, the biochemical parameters return towards normal levels and there is an increased mitotic activity in the basal cell layer of the epithelium, and repopulation of the receptor cell layer. Observations on the olfactory bulbs over the same time periods indicated that there is a similar loss followed by reappearance of the biochemical markers, and electron microscopy showed that during the first week there is a degeneration of olfactory nerve axons and terminals whereas over the period when the biochemical markers are returning there is a reappearance of olfactory nerve axons and terminals making synaptic contacts. With the wisdom of hindsight it is interesting to examine the reasons why Le Gros Clark la considered that there was no regeneration of olfactory receptors after lesions of the olfactory bulb. Using protargol silver staining of dendrites ('olfactory
299 rods') to identify receptor cells, he reported what he interpreted (with some puzzlement) as a selective survival of about half of the receptors from the olfactory epithelium of the rabbit at as long as 6 months after olfactory bulb lesions. ('The explanation of these residual receptors still eludes us'). It now seems likely (see also 36) that what he was seeing at this survival time was an incomplete stage in the repopulation of the epithelium following an earlier total receptor cell loss, a loss which he had himself reported in his own material at 2-3 days survival 15, but which he later came to regard as an artefact caused by a supposed inability of the protargol silver stain to show dendrites in the remaining cells seen in the epithelium at this time. ('The immediate results of the removal of the olfactory bulb may be accompanied by a defective silver impregnation . . . which at one time led me to suppose that all the receptors are affected by such a lesion'). It now seems more likely that the absence of dendrites was real, and that the cells remaining at 2-3 days survival were not receptor cells. The vomeronasal epithelium has been said to differ from the olfactory epithelium in being incapable of regeneration after nerve section 25. This has been ascribed to the absence of either Bowman's glands or of basal cells4,1°,z2,27, both of which are thought to act as sources of new cells in the regenerating olfactory epithelium. The vomeronasal epithelium has also been reported not to exhibit continuous turnover 22. In the previous paper 2 we described in the vomeronasal sensory epithelium of the adult mouse a region of dividing cells which may represent a low level of turnover, or a process of continued postnatal growth. In the present study we provide light and electron microscopical evidence for regeneration of vomeronasal receptors after degeneration caused by section of the vomeronasal nerves or removal of the accessory olfactory bulb. MATERIALS AND METHODS 20 adult (about 30 g b.w.) male albino mice of the Parkes strain were subjected to unilateral (left-sided) olfactory bulbectomy. The operation was performed under tribromoethanol anaesthesia ('Avertin', Winthrop, Surrey, 0.3 mg/g b.w., i.p. in alcoholic saline) with the head immobilised in a stereotaxic holder. Using an 18-gauge needle attached to a suction pump, the bulb was removed by aspiration through a dorsal craniotomy. After various survival times, the animals were anaesthetised and killed by perfusion through the left ventricle with a mixture of 1 ~ glutaraldehyde and 1 ~ formaldehyde in 0.1 M phosphate buffer at pH 7.4. The vomeronasal organs were dissected out and decalcified in 2.75 ~ EDTA at pH 7.4 for 3 days. Tissues were then processed for embedding in paraffin wax or postfixed in 2 ~ buffered osmium tetroxide and embedded in Epon (TAAB, Reading). Mice were killed at survival times of 4 days (n ----2 mice), 8 (n = 6), 9 (n -- 2), 10 (n ~-- 2), 16 (n ---- 1), 20 (n = 2), 32 (n ~ 4), and 50 days (n = 1). Two animals killed on day 20, and 2 killed on day 32 received 10 daily injections of [3H]thymidine (6-[aH]thymidine, sp. act. 26 Ci/mmole), The Radiochemical Centre, Amersham) from days 10 to 20 postoperatively at a dosage of 1/zCi/g b.w./day i.p. in 0.2 ml sterile saline.
300 A u t o r a d i o g r a p h y was carried out on 7 # m thick sections from wax e m b e d d e d blocks; details of the autoradiographic procedure were as described in the previous publication 2. After development the sections were stained through the emulsion with haematoxylin a n d eosin. Sections for electron microscopy were cut on an L K B U l t r o t o m e III, m o u n t e d on u n c o a t e d copper grids a n d stained with alcoholic u r a n y l acetate a n d lead citrate. I n a further 9 mice the v o m e r o n a s a l nerves were cut bilaterally. U n d e r Avertin anaesthesia, with the head immobilised in a stereotaxic holder, a m i c r o m a n i p u l a t o r was used to insert a knife constructed from a fragment of razor blade vertically t h r o u g h a dorsal c r a n i o t o m y at a p o i n t 1.0 m m rostral to the frontal pole of the cerebral cortex. The knife was lowered into one bulb until the tip touched the floor of the cranial cavity. It was then raised 0.5 m m a n d d r a w n coronally across the midline, so that it completely severed the v o m e r o n a s a l nerves bilaterally, where they lie close to the midline on the medial surfaces of the m a i n bulbs, rostral to the accessory bulbs. After survival times of 8 days (n = 2 mice), 16 (n ---- 2), 32 (n = 2), 44 (n = 1), 68 (n ---- 1), a n d 112 days (n = 1), the animals were perfused with 4 ~ phosphate buffered formaldehyde, a n d processed as above for paraffin wax embedding. All surgical lesions were verified by light microscopy of paraffin e m b e d d e d sections taken coronally through the olfactory bulbs. The criterion for b u l b e c t o m y was complete absence of accessory bulb tissue, a n d the criterion for v o m e r o n a s a l nerve section was complete absence of accessory bulb glomeruli.
Fig. 1. Coronal section through the vomeronasal organ (VNO) in a normal adult mouse. E, neurosensory epithelium; L, lumen of the VNO; S, cartilaginous nasal septum. The nasal cavity was decalcified, embedded in wax, and sections cut at 7 ym were stained with haematoxylin and eosin. Scale bar, 250 ym. Fig. 2. VNO at 4 days after bilateral section of the vomeronasal nerves to show one of the earliest signs of the retrograde reaction - - i.e. generalised thinning of the packing density of cells in the neurosensory epithelia (E) on both sides (cf. Fig. 1). Scale bar, 250 ym. Fig. 3. VNO at 8 days after bilateral section of the vomeronasal nerves to show loss of receptor cells (r) and persistence of the supporting cells (s). In the regions (arrows) of the neuroepithelium adjacent to the dorsal and ventral boundaries with the respiratory epithelium, the full thickness of the neuroepithelium remains populated by cells. Scale bar, 250 ym. Fig. 4. VNO at 8 days after unilateral olfactory bulbectomy to show a complete loss of receptor cells on the operated side (O, as in Fig. 3), and a normal appearance of the neurosensory epithelium on the unoperated side (U, cf. Fig. 1). Scale bar, 250 ym. Fig. 5. Higher magnification of the VNO at 8 days after vomeronasal nerve sectior~ to show the extent of cell loss in the receptor cell layer (r), persistence of supporting cells (s), and persistence of cells throughout the full thickness of the neuroepithelium in the regions adjacent to the dorsal and ventral boundaries (arrows) with the respiratory epithelium (c). B, blood vessel; L, lumen of the VNO. Scale bar, 250 ym. Fig. 6. Enlargement of part of the neuroepithelium from Fig. 5 to show that the supporting cells (s, which remain after retrograde loss of the receptor cells) have radial processes (arrow heads) extending from the lumen of the VNO (L) through the depopulated receptor cell layer (r) to the basal lamina (arrows). The few remaining nuclei in the receptor cell layer represent cells still in the process of degeneration, or a small number of surviving receptors. In addition, other cell types (e.g. capillary endothelium) are present. Scale bar, 20 ym.
302 RESULTS Adult mice were subjected to unilateral olfactory bulbectomy (OBX) or to bilateral vomeronasal nerve section (VNNX), and killed at various times thereafter. Four animals received 10 daily injections of [3H]thymidine from days 10 to 20 postoperatively. The bulbectomies completely removed the accessory olfactory bulb (AOB) on the operated side in all animals, and additionally the contralateral accessory bulb or vomeronasal nerve (VNN) was damaged to a variable extent in about half of the cases. Damage to the main bulb was extensive on the operated side but slight on the control side. The vomeronasal nerve sections completely severed all the vomeronasal nerve fascicles bilaterally, as judged by complete absence of glomeruli from the accessory bulbs. In all cases there was widespread damage to the main olfactory bulbs, and in many cases to the anterior poles of the accessory bulbs. In the vomeronasal organs (VNO) the sequelae of olfactory bulbectomy and vomeronasal nerve section were similar and will be considered together. 4 days survival (n = 2, OBX) Degenerative changes were seen amongst the neurosensory cells of the VNO. The cells appeared darker, shrunken, and fewer in number than in normal animals. Other components of the VNO showed no difference from normal (Figs. 1 and 2). 8 days survival (n = 6 0 B X , 2 V N N X ) There was a marked reduction in the number of cells in the sensory epithelium of the VNO on the operated side. The supporting cells remained normal in number and location, while virtually all of the receptor cells deep to them had disappeared (Figs. 3, 4). It is likely that this selective loss of the receptor cells represents a retrograde reaction to axotomy rather than a local effect of the lesion on the VNO (such as ischaemia): no other components of the VNO were affected, and since the blood supply of the VNO is from branches of the sphenopalatine artery, it is unlikely to be damaged by intracranial surgery 14. In contrast to the main part of the epithelial sheet, there was no cell loss in the regions of the neuroepithelium adjacent to the boundaries with the ciliated respiratory epithelium (Figs. 3, 4 and 5). In the previous paper 2 we showed that in normal, unoperated mice, these are regions of active cell division as shown by the presence of labelled nuclei at short survival times after [3H]thymidine administration. One possible explanation is that the cells here are immature and without axons, and thus would not have suffered direct injury as a result of cutting the vomeronasal nerve or bulbectomy. In 6 out of 8 mice, the area left free of receptor cells was occupied by supporting cell processes which formed a regular pattern of radial strands extending from the luminal surface to the basal margin of the epithelium (our Fig. 6; a similar arrangement of supporting cell processes in the normal olfactory epithelium is illustrated by Cajal, Fig. 409 in a Vol. 2). In the remaining 2 mice at this survival time, the supporting cell bodies had become directly apposed to the submucosa, thus completely eradicating the space formerly occupied by the receptor cells.
303
16 day survival (n = 1 0 B X , 2 VNNX) and 20 day survival (n = 2 0 B X ) At 16 days the space deep to the supporting cell layer in the epithelium was no longer vacant. Ceils appeared in groups, and the groups tended to coalesce to form a basally situated layer separated from the superficial supporting cell layer by a cell free zone (Fig. 7). At 20 days more cells were seen in the basal part of the receptor layer but the superficial part of the receptor layer was still relatively poorly repopulated with cells (Fig. 8). 32 days survival (n = 4 0 B X , 2 VNNX) The sensory epithelium had variable cellularity on the lesioned side: most regions appeared virtually repopulated throughout the full depth of the epithelium from basal to superficial layers, but some still had a lower than normal cell density in the region immediately deep to the supporting cells. Numerous circumscribed eosinophilic areas, typically oval or circular in profile, and containing few nuclei were seen extending into the receptor layer from the basal surface of the epithelium (arrows in Fig. 9). The arrangement of the supporting cell layer was frequently distorted over these areas. Electron microscopy (see below) shows these areas to be vomeronasal nerve neuromata. 44 to 112 days survival (n -----1 0 B X , 3 VNNX) At times longer than 32 days, the appearances were basically similar, with repopulation of the sensory epithelium well advanced. As at 32 days, the extent of recovery was still variable in different regions of a single VNO, but most of the epithelium appeared normal. Protargol - silver staining of the epithelium at 112 days survival showed sensory dendrites extending to the luminal surface, indicating the return of differentiated neurosensory cells (Fig. 10). Numerous circumscribed acellular or poorly cellular areas were also seen at these longer survival times, and in silver stained sections showed a fibrous structure (Fig. 10). Thymidine labelling In the above series, all of 8 animals showed virtually complete loss of receptor cells, but persistence of supporting cells in the VNO 8 days after OBX or VNNX. In all 15 animals which survived for 16-112 days, the epithelium showed a recovery of cell numbers which was greater at longer survival times. [aH]thymidine labelling could be used to demonstrate that this observed sequence is not due to the persistence of preexisting cells (e.g., as a result of incompleteness of the lesion). For this purpose 4 mice were given 10 daily injections of [3H]thymidine i.p. from days 10 to 20 after bulbectomy - - i.e. the injections were started at a time after degeneration of all axotomised receptor cells was known to have been complete. In 2 animals killed on day 20, it was seen that virtually all nuclei in the deeper layers of the epithelium were labelled - - i.e. they represent newly proliferated cells (Figs. 11 and 12). Some groups of supporting cell nuclei were also labelled (Figs. 13, 14 and 15). In 2 mice killed on day 32, there was a similar pattern of labelling, except
t.O
2
305 that most nuclei were lightly labelled (cf. Figs. 15 and 16), indicating that further cell division had taken place after the period of thymidine administration.
Electron microscopy At 50 days survival the regenerating sensory epithelium contained typical neurosensory cells lying in their normal position deep to the supporting cell layer. They had pale nuclei and the perikaryal cytoplasm contained a prominent mass made up of stacks of rough endoplasmic reticulum, and a characteristic and equally prominent, large organelle consisting of a paracrystalline array of regularly folded double smooth membranes of the type illustrated by Kolnberger 12 (see his Fig. 4b) and Bannister and Cuschieri 1 (see their Figs. 3 and 6). The sensory dendrites of the neurosensory cells extend through the supporting cell layer to reach the lumen, where they form a normal ending containing centrioles, surmounted by villi on the luminal surface and forming specialised junctional complexes with the adjacent supporting cellsS,6,1a, 29. In structure, therefore, the newly formed neurosensory cells resemble those in normal, unoperated mice. No systematic search was made to identify possible immature precursors of these neurosensory cells. Electron microscopy showed that the circumscribed acellular areas found in the light micrographs of the regenerating epithelium were in fact neuromata, consisting of masses of apposed unmyelinated axons whose structure resembled that of normal vomeronasal nerve axons. The arrangement of axons and ensheathing cell processes in the neuromata varied from an appearance (Fig. 18) close to that of normal vomeronasal nerves (Fig. 17) to a more disorganised structure. The areas resembling normal vomeronasal nerves contained masses of aligned parallel axons with a fairly uniform diameter of around 0.2-0.3 #m, and between 5 and 10 microtubules per profile. The only other notable axonal contents were elongated mitochondria and occasional dense cored vesicles of around 100 nm diameter. Groups of around 20 axons were segregated into irregularly shaped, interconnecting bundles by slender, finger-like processes of non-nervous cells whose cell bodies were closely applied to the nerve bundles. When compared with the normal vomeronasal nerves (cf. Figs. 17 and 18), however, even the most highly organized neuromatous masses did not achieve the same regularity
Fig. 7. VNO 16 days after olfactory bulbectomy to show the return of the newly formed receptor cells in patches (arrows) to the basal parts of the depopulated receptor cell layer. In the superficial part of the receptor cell layer there is a relatively cell free zone still remaining between the newly formed receptor cells and the supporting cell layer (s). Scale bar, 200 pm. Fig. 8. VNO 20 days after olfactory bulbectomy to show further re-population of the receptor cell layer (cf. Fig. 7) but still with some preferential localisation of the newly formed cells in the basal parts of the receptor cell layer. Scale bar, 200/~m. Fig. 9. VNO 32 days after olfactory bulbectomy to show nearly complete repopulation of the receptor cell layer but with localised acellular areas (arrows), which electron microscopy shows to be vomeronasal nerve neuromata. Scale bar, 200/~m. Fig. 10 VNO 112 days after olfactory bulbectomy: stained with protargol silver to show the sensory dendrites (arrows) of the newly formed receptor cells (r) extending through the supporting cell layer (s) to the lumen of the VNO (L). Two large neuromatous areas (N). Scale bar, 50/~m.
307 of axon arrangement, nor the same economy in the number of sheathing cell processes used to divide up the axon bundles: in the neuromata there are always fewer axons per bundle, and some degree of redundancy (i.e. reduplication) of sheathing cell processes. In the more disorganised areas, the axons are less regularly aligned, more varicose, and have large expansions packed with dense cored vesicles, mitochondria, large irregular dense bodies and autophagic vacuoles. Irregular fine filopodia extend from these expansions, and some profiles also show axoplasm with interlacing wispy material. These appearances are typical of those found at the growing ends of regenerating peripheral nerves in a variety of sites (e.g., ref. 19). An additional feature of such areas was the presence of clusters of vesicles in the axons. These vesicles were somewhat irregular in size and shape, but basically consisted of clear centred vesicles of around 50 nm diameter with occasional dense cored vesicles of 100 nm diameter. Occasionally the vesicles were clustered towards the outer surface of the axons at points of direct contact with non-neuronal sheathing processes ('synaptoid' contacts). In many areas (Fig. 20) the axons had a structure resembling the vomeronasal nerve terminals in the glomeruli of the normal accessory olfactory bulb (Fig. 19). The axons expanded into large irregular profiles which were aggregated in clusters. In such regions the axons had a dark background matrix and vesicle clusters of the type found in the glomerular terminals of normal vomeronasal nerve, the main difference being that the neuromatous masses had no indications of any kind of postsynaptic elements. DISCUSSION We find that after vomeronasal nerve section or removal of the accessory olfactory bulb, the majority of vomeronasal neurosensory cells have disappeared by 8 days. The time course of degeneration agrees with Nagahara z5 in being somewhat slower than that of the neurosensory cell s in the main olfactory mucosa. This reaction is selective for the neurosensory cells - - the supporting cells and the cells adjacent to the boundaries with the non-sensory epithelium are spared. In contrast to Nagahara 25 we find that at longer survival times the VNO progressively restores its cellularity by mitotic division of an unidentified stem cell type,
Figs. 11 and 12. Bright and dark field photographs of the same section through the VNO 20 days after unilateral olfactory bulbectomy and systemic administration of [aHlthymidine daily from days 10 to 20. Labelled nuclei are present throughout the regenerating neurosensory epithelium on the operated side (O) but are confined to clusters (arrows) in the regions of the neuroepithelium adjacent to the boundaries with the respiratory epithelium (especially the dorsal boundary) on the unoperated side (U). Seal e bar, 200 #m. Figs. 13 and 14. Bright and dark field photographs of the same section through the VNO 20 days after unilateral olfactory bulbectomy and systemic administration of [3H]thymidinedaily from days 10 to 20. Labelled nuclei are present in a high proportion of cells in the receptor cell layer and also in patches of supporting cell nuclei (arrows). Dashed lines drawn in on the dark field picture indicate the outline of the lumen (L) of the VNO. Scale bar, 100 urn.
308
Fig. 15. VNO 20 days after unilateral olfactory bulbectomy and systemic administration of [aH]thymidine daily from days 10-20. Many nuclei in the receptor cell layer (r) are heavily labelled and there are also some heavily labelled supporting cell nuclei (arrows). L, lumen; s, supporting cell layer. Scale bar, 50/~m. Fig. 16. VNO 32 days after unilateral bulbectomy and systemic administration of [aH]thymidine daily from days 10-20. There are still occasional heavily labelled cell nuclei, (white arrowhead) but compared with the previous stage (Fig. 15) the majority of nuclei in the receptor layer (r) are much more lightly labelled, implying that cell division has continued in the intervening period. Patches of supporting cell nuclei are also still heavily labelled (arrows). The apparent extracellular location of some silver grains is partly due to 'crossfire', and partly to grains lying over tangentially sectioned nuclei not in the plane of focus; both factors are accentuated by the thickness (7 tzm) of the section. Scale bar, 50/zm.
Fig. 17. A transverse section of part of a normal vomeronasal nerve fascicle taken at the level where the nerves lie on the medial surface of the olfactory bulb. The majority of axons have between 3 and 5 microtubules. N, part of the nucleus of a sheathing cell giving rise to processes (arrows). Scale bar,
1/~m. Fig. 18. Part of an intraepitbelial neuroma from the VNO of a mouse 50 days after OBX. The majority of axons are cut transversely. N, part of the nucleus of a sheathing cell; arrows, sheathing cell processes. (*) Basement membrane lined channels extending in from the external surface of the nerve. Scale bar, 1/~m. Fig. 19. A section across the edge of a glomerulus from the AOB of a normal mouse. Electron dense vomeronasal nerve terminals make synaptic contacts (arrows) with paler, mitral cell dendritic branches. g, capsule of glial, or possibly neuronal 29 sheet processes around the glomerulus. Scale bar, 1/~m. Fig. 20. Part of a neuromatous region from the vomeronasal neuroepithelium of a mouse 50 days after OBX, to show the resemblance to the normal glomerular terminals of the vomeronasal axons. Vesicle filled electron dense axonal profiles form an interwoven mass with no recognisable postsynaptic elements present, Scale bar, 1 t~m,
309
310 followed by migration and differentiation. The epithelium is repopulated by what are, at least structurally, normal receptor cells: protargol silver staining at 112 days survival showed the presence of sensory dendrites, and electron microscopy at 50 days survival confirmed this observation and showed that the epithelium contained fully differentiated receptor cells and typical vomeronasal axons. The majority of receptor nuclei in the regenerated epithelium at 20-32 days after operation can be labelled by prior administration of thymidine over the period from 10 to 20 days after operation. This indicates that these cells are not receptor cells which have somehow persisted unidentified from before the lesion (cf. ref. 14), since mature receptor cells have not been observed to undergo mitosis. One has to conclude that the labelled receptor cells have arisen by division and subsequent differentiation of a stem cell. The use of [aH]thymidine labelling also shows that, in contrast to Graziadei's findings in the frog olfactory epithelium 7, some new supporting cells are also formed during the process of regeneration of the mouse vomeronasal epithelium. Neither the location nor the identity of the vomeronasal stem cell is clear. In the olfactory epithelium the stem cell is basal (e.g. ref. 7, 9, 21). However, the VNO lacks an obvious basal layer of cells4,1°,22, and in our material from the vomeronasal epithelium after retrograde degeneration of the receptors we find no persisting basal layer of cells such as is seen in the main olfactory epithelium under these circumstances 7,e5. Nevertheless, repopulation of the regenerating vomeronasal epithelium seems to commence in the basal layers. It is possible that isolated basal cells are present in small numbers and have evaded discovery. However, it seems likely that at least part of the regeneration occurs by proliferation of cells at the boundaries of the sensory epithelium: these cells are actively dividing in the normal adult animal 2, and they persist in the acute phase of the retrograde reaction after vomeronasal nerve destruction. The present material does not enable us to reconstruct the precise migration path taken by the progeny of the stem cell division (see also ref. 23). During the stage of the most acute retrograde loss of neurosensory cells, the epithelium has a persisting regular skeleton (Fig. 6) made up of the radial processes of the supporting cells, which form a parallel array extending the full depth of the epithelium from the cell bodies at the luminal surface to the end feet on the basal lamina a. Possibly these processes offer some kind of guidance to the migrating neurosensory cells (e.g. analogous to the radial glia in the developing cerebellum2S). On the other hand, the supporting cell framework is not itself entirely composed of cells persisting from before the retrograde neurosensory cell loss, since we have observed patches of thymidine labelled supporting cells in the regenerated epithelium, implying that at least some supporting cells are formed as a result of the stem cell division stimulated by the retrograde loss of the neurosensory cells. The regenerating neurosensory epithelium is in general not completely repopulated, and contains neuromata consisting of unmyelinated axons closely resembling normal vomeronasal nerve axons. Since such neuromata were not seen in normal mice, they probably represent new axon outgrowth. In view of their common occur-
311 rence in the regenerating epithelium it is likely that they originate from the newly produced neurosensory cells, although we cannot completely rule out the unlikely possibility that a small number of neurosensory cells might somehow have persisted undetected after the original lesion and given rise to all the new axons in the neuromata. The axons form fairly normally arranged bundles and have typical growing tips. In some places there are signs of obstructed growth, with the formation of neuromatous tangles, and in other places the axons have a fully differentiated terminal structure and appearance, although they have not grown further than the epithelium and have not made contact with any identifiable postsynaptic elements. Studies on regeneration in the main olfactory system have mentioned neuroma formation at the site of surgical transection in regenerating pigeon olfactory nerve 26. Schultz 80 reported that axon bundles were unusually prominent in the lamina propria of the epithelium regenerating after zinc sulphate treatment in the monkey. Apart from this there have been no reports of intraepithelial neuromata such as we describe here in the regenerating VNO. The implication may be that regenerating olfactory receptors do not produce axons, or that they produce axons which distribute randomly, and are therefore not seen as obvious neuromata, or (which seems the most likely) that the regenerating olfactory axons return to the olfactory bulb in an orderly manner and with high efficiency in selecting their routes, so that axons do not become obstructed and diverted into neuromatous masses. This last suggestion is supported by reports of return of olfactory function after damage to the primary olfactory neurones. There are many clinical accounts of return of the sense of smell after its loss following head injury or upper respiratory tract surgery a4. After olfactory nerve section in pigeons ~6 and goldfish 37 the olfactory nerves have been seen to regenerate back to the bulbs and in both cases the histological evidence of regeneration was accompanied by recovery of the ability to perform olfactory discrimination tasks. Further supporting evidence for the regrowth of olfactory axons to the bulbs is provided by Schultz's finding a°,al of a return of susceptibility to infection by intranasal poliovirus after experimental damage to the olfactory mucosa in monkeys. During the course of the present work, the results of Harding et al. 11 became available; in this study the olfactory nerves were cut in mice, and it was shown that olfactory axons and terminals disappeared from the olfactory bulb glomeruli, together with three biochemical markers known to be specifically contained in primary olfactory neurones. The subsequent reappearance (with similar time course) of axons, terminals, and the biochemical markers provided good evidence that a high proportion of regenerating olfactory axons do return to the bulb (see also ref. 8). We now have some preliminary experimental results using an autoradiographic tracing method which provides direct evidence that after destruction of the vomeronasal nerves in the mouse at least some of the newly formed vomeronasal neurosensory cells have axons which reach as far as the olfactory bulb. Further work is in progress to investigate the extent to which these regenerating axons are capable of innervating bulb tissue.
312 REFERENCES 1 Bannister, L. H. and Cuschieri, A., The fine structure and enzyme histochemistry of vomeronasal receptor cells. In D. Schneider (Ed.), Olfaction and Taste, Vol. IV, Wissenschaftliche MBH, Stuttgart, 1972, pp. 27-33. 2 Barber, P. C. and Raisman, G., Cell division in the vomeronasal organ of the adult mouse, Brain Research, in press. 3 Cajal, S., Ram6n Y., Histologie du systdme nerveux de l'homme et des vertdbr~s, Maloine, Paris, 1911, 993 pp. 4 Cuschieri, A. and Bannister, L. H., The development of the olfactory mucosa in the mouse: light microscopy, J. Anat. (Lond.), 119 (1975) 277-286. 5 Graziadei, P. P. C., Topological relation between olfactory neurons, Z. Zellforsch., 118 (1971) 449-466. 6 Graziadei, P. P. C., The olfactory mucosa of vertebrates. In L. M. Beidler (Ed.), Handbook o f Sensory Physiology, VoL IV, Springer, Berlin, 1971, pp. 27-58. 7 Graziadei, P. P. C., Cell dynamics in the olfactory mucosa, Tissue and Cell, 5 (1973) 113-131. 8 Graziadei, P. P. C. and Dehan, R. S., Neuronal regeneration in frog olfactory system, J. Cell Biol., 59 (1973) 525-530. 9 Graziadei, P. P. C. and Metcalf, J. F., Autoradiographic and ultrastructural observations on the frog's olfactory mucosa, Z. ZellJorsch., 116 (1971) 305-318. 10 Graziadei, P. P. C. and Tucker, D., Vomeronasal receptors in turtles, Z. Zellforsch., 105 (1970) 498-514. I 1 Harding, J., Graziadei, P. P. C., Graziadei, G. A. M. and Margolis, F. L., Denervation in the primary olfactory pathway of mice. IV. Biochemical and morphological evidence for neuronal replacement following nerve section, Brain Research, 132 (1977) 11-28. 12 Kolnberger, I., Vergleichende Untersuchungen am Riechepithel, insbesondere des Jacobsonschen Organs von Amphibien, Reptilien und S~ugetieren, Z. Zellforsch., 122 (1971) 53-67. 13 Kolnberger, I. and Altner, H., The occurrence of diffusion barriers and interreceptor contacts in the olfactory epithelium of the vomero-nasal organ. In D. Schneider (Ed.), Olfaction and Taste, Vol. IV, Wissenschaftliche, Stuttgart, 1972, pp. 34-39. 14 Le Gros Clark, W.E., Inquiries into the anatomical basis of olfactory discrimination, Proc. roy. Soc. B., 146 (1957) 299-319. 15 Le Gros Clark, W. E. and Warwick, R. T. T., The pattern of olfactory innervation, J. Neurol. Neurosurg. Psychiat., 9 (1946) 101-111. 16 Margolis, F. L., A brain protein unique to the olfactory bolb, Proc. nat. Acad. Sci. (Wash.), 69 (1972) 1221-1224. 17 Margolis, F. L., Carnosine in the primary olfactory pathway, Science, 184 (1974) 909-911. 18 Margolis, F. L. and Tarnoff, J. F., Site of biosynthesis of the mouse brain olfactory bulb protein, J. biol. Chem., 248 (1973) 451455. 19 Matthews, M. R., An ultrastructural study of axonal changes following construction of postganglionic branches of the superior cervical ganglion in the rat, Phil. Trans. B., 264 (1973) 479-508. 20 Matulionis, D. H., Ultrastructural study of mouse olfactory epithelium following destruction by ZnSO4 and its subsequent regeneration, Amer. J. Anat., 142 (1975) 67-90. 21 Moulton, D. G., Dynamics of cell populations in the olfactory epithelium, Ann. N. Y. Acad. Sci., 237 (1974) 52-61. 22 Moulton, D. G., C~elebi, G. and Fink, R. P., Olfaction in mammals - - two aspects: proliferation of cells in the olfactory epithelium and sensitivity to odours. In G. E. W. Wolstenholme and J. Knight (Eds.), Ciba Foundation Symposium on Taste and Smell in Vertebrates, Churchill, London, 1970, pp. 227-250. 23 Moulton, D. G. and Fink, R. P., Cell proliferation and migration in the olfactory epithelium. In D. Schneider (Ed.), Olfaction and Taste, VoL IV, Wissenschaftliche MBH, Stuttgart, 1972, pp. 20-26. 24 Mulvaney, B. D. and Heist, H. E., Regeneration of rabbit olfactory epithelium, Amer. J. Anat., 131 (1971) 241-251. 25 Nagahara, Y., Experimentelle Studien fiber die histologischenVerS.nderungen des Geruchsorgans nach der Olfactoriusdurchschneidung. Beitr~ige zur Kenntnis des feineren Baus des Geruchs. organs, Jap. J. reed. Sci., 5 (1940) 165-199. 26 Oley, N., Delian, R. S., Tucker, D., Smith, J. C. and Graziadei, P. P. C., Recovery of structure
313
27 28 29 30 31 32 33 34 35 36 37
and function following transection of the primary olfactory nerves in pigeons, J. comp. physiol. Psychol., 88 (1975) 477-495. Parsons, T. S., Evolution of the nasal structure in the lower tetrapods, Amer. Zool., 7 (1967) 397413. Rakic, P., Neuron-glia relationship during granule cell migration in developing cerebellar cortex. A Golgi and electronmicroscopic study in Macacus rhesus, J. comp. Neurol., 141 (1971) 283-312. Reese, T. S. and Brightman, M. W., Olfactory surface and central olfactory connexions in some vertebrates. In G. E. W. Wolstenholme and J. Knight (Eds.), Taste and Smell in Vertebrates, Churchill, London, 1970, pp. 115-150. Schultz, E. W., Regeneration of olfactory cells, Proc. Soc. exp. Biol., (N. Y.), 46 (1941) 41-43. Schultz, E. W., Repair of the olfactory mucosa. With special reference to regeneration of olfactory cells (sensory neurons), Amer. J. Path., 37 (1960) 1-19. Smith, C. G., Changes in the olfactory mucosa and the olfactory nerves following intranasal treatment with one per cent zinc sulphate, Canad. med. Ass. J., 39 (1938) 138-140. Smith, C. G., Regeneration of sensory olfactory epithelium and nerves in adult frogs, Anat. Rec., 109 (1951) 661-671. Sumner, D., Post-traumatic anosmia, Brain Research, 87 (1964) 107-120. Takagi, S. F., Degeneration and regeneration of the olfactory epithelium. In L. M. Beidler (Ed.), Handbook of Sensory Physiology, Vol. IV, Springer, Berlin, 1971, pp. 75-94. Thornhill, R. A., Cell division in the olfactory epithelium of the lamprey, Lampetra fluviatilis, Z. Zellforsch., 109 (1970) 147-157. Zippel, H. P., Baumgarten, R. yon and Westerman, R. A., Histologische, funktionelle und spezifische Regeneration nach Durchtrennung der ilia olfactoria beim Goldfisch, Z. vergl. PhysioL, 69 (1970) 79-98.
297
REPLACEMENT OF RECEPTOR NEURONES AFTER SECTION OF THE VOMERONASAL NERVES IN THE A D U L T MOUSE
P. C.
BARBER and G. RAISMAN
Laboratory of Neurobiology, National Institute for Medical Research, Mill Hill, London NW7 1.4A (England)
(Accepted September 19th, 1977)
SUMMARY Eight days after vomeronasal nerve section or removal of the accessory olfactory bulb, the majority of receptor cells of the vomeronasal neuroepithelium degenerate and disappear, leaving a regular framework consisting of supporting cells and their radial processes. The cell clusters at the boundaries of the epithelial sheet (which have been shown to be actively dividing in the normal, unoperated adult mouse) are also spared. The epithelium is subsequently repopulated by receptor cells appearing first in the basal part of the receptor cell layer and later occupying the full width of the receptor layer. These cells are anatomically fully differentiated receptor cells with normal sensory dendrites. Their axons form conspicuous intraepithelial neuromatous masses. Administration of [3H]thymidine on days 10-20 postoperatively labels some clusters of supporting cells and virtually all of the receptor cells, indicating that the repopulation of the epithelium is due to new formation of receptor cells.
INTRODUCTION It is well established that the olfactory chemoreceptor neurones undergo a reaction of retrograde degeneration when subjected to axotomy by severing the olfactory nerves or removing the olfactory bulb. They can also be destroyed by direct application of toxic chemicals or by mechanical interference. There is now evidence that after such loss there may be replacement of these sensory neurones from a stem cell with restoration of normal structure of the olfactory epithelium and possible return of olfactory function, even in adult animals (see ref. 35 for a review). Nagahara 25 in the mouse, saw extensive degeneration of olfactory receptor cells 3 days after olfactory nerve section or bulbectomy, followed by restoration of normal mucosal cellularity in 90 days. He proposed that the basal part of the mucosa contains 'resting cells', presumably without axons, which persist after the original
298 neurosensory cells have degenerated, and which give rise to new neurosensory cells by mitosis. After the destruction of the olfactory mucosa by intranasal irrigation with 1 ~ zinc sulphate, various authors have reported regeneration of a normally arranged neuroepithelium in the frog 33, rabbit 24, mouse 20 and monkey30, 31, although an earlier study 82 failed to show regeneration in the rat. In monkeys, Schultz3O, 31 found that regeneration of the olfactory mucosa after zinc sulphate took 3-4 months, and corresponded in time to the return of susceptibility to infection by intranasal inoculation of a neuronotropic strain of poliovirus to which zinc sulphate treatment had earlier conferred resistance. This suggested that mucosal regeneration was accompanied by axonal regrowth to the olfactory bulb, thus providing a pathway for the virus infection. Recent studies using systemically injected tritiated thymidine have demonstrated in normal adult animals a continuous formation and presumably therefore a turnover of olfactory receptors from a basally located stem cell (see ref. 2 for refs). It has also been shown in the frog that after receptor degeneration produced by olfactory nerve section, mitosis and thymidine labelling of the basal cells is greatly stimulated, and that labelled cells repopulate the epithelium, coming to resemble olfactory receptors, both light and electron microscopically7. Axons, presumably arising from the new receptors, have been observed to grow back and to reestablish synaptic contacts in the glomeruli of the olfactory bulb s. In the pigeon after olfactory nerve section, Oley et al. 26 demonstrated restoration of normal morphology of the olfactory mucosa, regeneration of the olfactory nerves and their synapses in the bulb, and recovery of the ability to perform olfactory discrimination tasks. A series of biochemical experiments by Margolis and his collaborators11,16-1s also provide strong evidence for the regenerative replacement of olfactory receptor cells in the adult mouse. The receptor cells contain a unique 'olfactory marker' protein 16 whose synthesis occurs only in the olfactory epithelium t8. This protein, together with an unusual dipeptide, carnosine (fl-alanyl-L-histidine)17 and the enzyme carnosine synthetase, are present in the receptor cells in the olfactory epithelium and in their axons and axon terminals in the olfactory bulb. Within the first few days after unilateral olfactory nerve section 11, all three biochemical markers disappear from the olfactory epithelium, during which time the histology shows that the receptor cells degenerate. Over the period from 2 to 4 weeks after operation, the biochemical parameters return towards normal levels and there is an increased mitotic activity in the basal cell layer of the epithelium, and repopulation of the receptor cell layer. Observations on the olfactory bulbs over the same time periods indicated that there is a similar loss followed by reappearance of the biochemical markers, and electron microscopy showed that during the first week there is a degeneration of olfactory nerve axons and terminals whereas over the period when the biochemical markers are returning there is a reappearance of olfactory nerve axons and terminals making synaptic contacts. With the wisdom of hindsight it is interesting to examine the reasons why Le Gros Clark la considered that there was no regeneration of olfactory receptors after lesions of the olfactory bulb. Using protargol silver staining of dendrites ('olfactory
299 rods') to identify receptor cells, he reported what he interpreted (with some puzzlement) as a selective survival of about half of the receptors from the olfactory epithelium of the rabbit at as long as 6 months after olfactory bulb lesions. ('The explanation of these residual receptors still eludes us'). It now seems likely (see also 36) that what he was seeing at this survival time was an incomplete stage in the repopulation of the epithelium following an earlier total receptor cell loss, a loss which he had himself reported in his own material at 2-3 days survival 15, but which he later came to regard as an artefact caused by a supposed inability of the protargol silver stain to show dendrites in the remaining cells seen in the epithelium at this time. ('The immediate results of the removal of the olfactory bulb may be accompanied by a defective silver impregnation . . . which at one time led me to suppose that all the receptors are affected by such a lesion'). It now seems more likely that the absence of dendrites was real, and that the cells remaining at 2-3 days survival were not receptor cells. The vomeronasal epithelium has been said to differ from the olfactory epithelium in being incapable of regeneration after nerve section 25. This has been ascribed to the absence of either Bowman's glands or of basal cells4,1°,z2,27, both of which are thought to act as sources of new cells in the regenerating olfactory epithelium. The vomeronasal epithelium has also been reported not to exhibit continuous turnover 22. In the previous paper 2 we described in the vomeronasal sensory epithelium of the adult mouse a region of dividing cells which may represent a low level of turnover, or a process of continued postnatal growth. In the present study we provide light and electron microscopical evidence for regeneration of vomeronasal receptors after degeneration caused by section of the vomeronasal nerves or removal of the accessory olfactory bulb. MATERIALS AND METHODS 20 adult (about 30 g b.w.) male albino mice of the Parkes strain were subjected to unilateral (left-sided) olfactory bulbectomy. The operation was performed under tribromoethanol anaesthesia ('Avertin', Winthrop, Surrey, 0.3 mg/g b.w., i.p. in alcoholic saline) with the head immobilised in a stereotaxic holder. Using an 18-gauge needle attached to a suction pump, the bulb was removed by aspiration through a dorsal craniotomy. After various survival times, the animals were anaesthetised and killed by perfusion through the left ventricle with a mixture of 1 ~ glutaraldehyde and 1 ~ formaldehyde in 0.1 M phosphate buffer at pH 7.4. The vomeronasal organs were dissected out and decalcified in 2.75 ~ EDTA at pH 7.4 for 3 days. Tissues were then processed for embedding in paraffin wax or postfixed in 2 ~ buffered osmium tetroxide and embedded in Epon (TAAB, Reading). Mice were killed at survival times of 4 days (n ----2 mice), 8 (n = 6), 9 (n -- 2), 10 (n ~-- 2), 16 (n ---- 1), 20 (n = 2), 32 (n ~ 4), and 50 days (n = 1). Two animals killed on day 20, and 2 killed on day 32 received 10 daily injections of [3H]thymidine (6-[aH]thymidine, sp. act. 26 Ci/mmole), The Radiochemical Centre, Amersham) from days 10 to 20 postoperatively at a dosage of 1/zCi/g b.w./day i.p. in 0.2 ml sterile saline.
300 A u t o r a d i o g r a p h y was carried out on 7 # m thick sections from wax e m b e d d e d blocks; details of the autoradiographic procedure were as described in the previous publication 2. After development the sections were stained through the emulsion with haematoxylin a n d eosin. Sections for electron microscopy were cut on an L K B U l t r o t o m e III, m o u n t e d on u n c o a t e d copper grids a n d stained with alcoholic u r a n y l acetate a n d lead citrate. I n a further 9 mice the v o m e r o n a s a l nerves were cut bilaterally. U n d e r Avertin anaesthesia, with the head immobilised in a stereotaxic holder, a m i c r o m a n i p u l a t o r was used to insert a knife constructed from a fragment of razor blade vertically t h r o u g h a dorsal c r a n i o t o m y at a p o i n t 1.0 m m rostral to the frontal pole of the cerebral cortex. The knife was lowered into one bulb until the tip touched the floor of the cranial cavity. It was then raised 0.5 m m a n d d r a w n coronally across the midline, so that it completely severed the v o m e r o n a s a l nerves bilaterally, where they lie close to the midline on the medial surfaces of the m a i n bulbs, rostral to the accessory bulbs. After survival times of 8 days (n = 2 mice), 16 (n ---- 2), 32 (n = 2), 44 (n = 1), 68 (n ---- 1), a n d 112 days (n = 1), the animals were perfused with 4 ~ phosphate buffered formaldehyde, a n d processed as above for paraffin wax embedding. All surgical lesions were verified by light microscopy of paraffin e m b e d d e d sections taken coronally through the olfactory bulbs. The criterion for b u l b e c t o m y was complete absence of accessory bulb tissue, a n d the criterion for v o m e r o n a s a l nerve section was complete absence of accessory bulb glomeruli.
Fig. 1. Coronal section through the vomeronasal organ (VNO) in a normal adult mouse. E, neurosensory epithelium; L, lumen of the VNO; S, cartilaginous nasal septum. The nasal cavity was decalcified, embedded in wax, and sections cut at 7 ym were stained with haematoxylin and eosin. Scale bar, 250 ym. Fig. 2. VNO at 4 days after bilateral section of the vomeronasal nerves to show one of the earliest signs of the retrograde reaction - - i.e. generalised thinning of the packing density of cells in the neurosensory epithelia (E) on both sides (cf. Fig. 1). Scale bar, 250 ym. Fig. 3. VNO at 8 days after bilateral section of the vomeronasal nerves to show loss of receptor cells (r) and persistence of the supporting cells (s). In the regions (arrows) of the neuroepithelium adjacent to the dorsal and ventral boundaries with the respiratory epithelium, the full thickness of the neuroepithelium remains populated by cells. Scale bar, 250 ym. Fig. 4. VNO at 8 days after unilateral olfactory bulbectomy to show a complete loss of receptor cells on the operated side (O, as in Fig. 3), and a normal appearance of the neurosensory epithelium on the unoperated side (U, cf. Fig. 1). Scale bar, 250 ym. Fig. 5. Higher magnification of the VNO at 8 days after vomeronasal nerve sectior~ to show the extent of cell loss in the receptor cell layer (r), persistence of supporting cells (s), and persistence of cells throughout the full thickness of the neuroepithelium in the regions adjacent to the dorsal and ventral boundaries (arrows) with the respiratory epithelium (c). B, blood vessel; L, lumen of the VNO. Scale bar, 250 ym. Fig. 6. Enlargement of part of the neuroepithelium from Fig. 5 to show that the supporting cells (s, which remain after retrograde loss of the receptor cells) have radial processes (arrow heads) extending from the lumen of the VNO (L) through the depopulated receptor cell layer (r) to the basal lamina (arrows). The few remaining nuclei in the receptor cell layer represent cells still in the process of degeneration, or a small number of surviving receptors. In addition, other cell types (e.g. capillary endothelium) are present. Scale bar, 20 ym.
302 RESULTS Adult mice were subjected to unilateral olfactory bulbectomy (OBX) or to bilateral vomeronasal nerve section (VNNX), and killed at various times thereafter. Four animals received 10 daily injections of [3H]thymidine from days 10 to 20 postoperatively. The bulbectomies completely removed the accessory olfactory bulb (AOB) on the operated side in all animals, and additionally the contralateral accessory bulb or vomeronasal nerve (VNN) was damaged to a variable extent in about half of the cases. Damage to the main bulb was extensive on the operated side but slight on the control side. The vomeronasal nerve sections completely severed all the vomeronasal nerve fascicles bilaterally, as judged by complete absence of glomeruli from the accessory bulbs. In all cases there was widespread damage to the main olfactory bulbs, and in many cases to the anterior poles of the accessory bulbs. In the vomeronasal organs (VNO) the sequelae of olfactory bulbectomy and vomeronasal nerve section were similar and will be considered together. 4 days survival (n = 2, OBX) Degenerative changes were seen amongst the neurosensory cells of the VNO. The cells appeared darker, shrunken, and fewer in number than in normal animals. Other components of the VNO showed no difference from normal (Figs. 1 and 2). 8 days survival (n = 6 0 B X , 2 V N N X ) There was a marked reduction in the number of cells in the sensory epithelium of the VNO on the operated side. The supporting cells remained normal in number and location, while virtually all of the receptor cells deep to them had disappeared (Figs. 3, 4). It is likely that this selective loss of the receptor cells represents a retrograde reaction to axotomy rather than a local effect of the lesion on the VNO (such as ischaemia): no other components of the VNO were affected, and since the blood supply of the VNO is from branches of the sphenopalatine artery, it is unlikely to be damaged by intracranial surgery 14. In contrast to the main part of the epithelial sheet, there was no cell loss in the regions of the neuroepithelium adjacent to the boundaries with the ciliated respiratory epithelium (Figs. 3, 4 and 5). In the previous paper 2 we showed that in normal, unoperated mice, these are regions of active cell division as shown by the presence of labelled nuclei at short survival times after [3H]thymidine administration. One possible explanation is that the cells here are immature and without axons, and thus would not have suffered direct injury as a result of cutting the vomeronasal nerve or bulbectomy. In 6 out of 8 mice, the area left free of receptor cells was occupied by supporting cell processes which formed a regular pattern of radial strands extending from the luminal surface to the basal margin of the epithelium (our Fig. 6; a similar arrangement of supporting cell processes in the normal olfactory epithelium is illustrated by Cajal, Fig. 409 in a Vol. 2). In the remaining 2 mice at this survival time, the supporting cell bodies had become directly apposed to the submucosa, thus completely eradicating the space formerly occupied by the receptor cells.
303
16 day survival (n = 1 0 B X , 2 VNNX) and 20 day survival (n = 2 0 B X ) At 16 days the space deep to the supporting cell layer in the epithelium was no longer vacant. Ceils appeared in groups, and the groups tended to coalesce to form a basally situated layer separated from the superficial supporting cell layer by a cell free zone (Fig. 7). At 20 days more cells were seen in the basal part of the receptor layer but the superficial part of the receptor layer was still relatively poorly repopulated with cells (Fig. 8). 32 days survival (n = 4 0 B X , 2 VNNX) The sensory epithelium had variable cellularity on the lesioned side: most regions appeared virtually repopulated throughout the full depth of the epithelium from basal to superficial layers, but some still had a lower than normal cell density in the region immediately deep to the supporting cells. Numerous circumscribed eosinophilic areas, typically oval or circular in profile, and containing few nuclei were seen extending into the receptor layer from the basal surface of the epithelium (arrows in Fig. 9). The arrangement of the supporting cell layer was frequently distorted over these areas. Electron microscopy (see below) shows these areas to be vomeronasal nerve neuromata. 44 to 112 days survival (n -----1 0 B X , 3 VNNX) At times longer than 32 days, the appearances were basically similar, with repopulation of the sensory epithelium well advanced. As at 32 days, the extent of recovery was still variable in different regions of a single VNO, but most of the epithelium appeared normal. Protargol - silver staining of the epithelium at 112 days survival showed sensory dendrites extending to the luminal surface, indicating the return of differentiated neurosensory cells (Fig. 10). Numerous circumscribed acellular or poorly cellular areas were also seen at these longer survival times, and in silver stained sections showed a fibrous structure (Fig. 10). Thymidine labelling In the above series, all of 8 animals showed virtually complete loss of receptor cells, but persistence of supporting cells in the VNO 8 days after OBX or VNNX. In all 15 animals which survived for 16-112 days, the epithelium showed a recovery of cell numbers which was greater at longer survival times. [aH]thymidine labelling could be used to demonstrate that this observed sequence is not due to the persistence of preexisting cells (e.g., as a result of incompleteness of the lesion). For this purpose 4 mice were given 10 daily injections of [3H]thymidine i.p. from days 10 to 20 after bulbectomy - - i.e. the injections were started at a time after degeneration of all axotomised receptor cells was known to have been complete. In 2 animals killed on day 20, it was seen that virtually all nuclei in the deeper layers of the epithelium were labelled - - i.e. they represent newly proliferated cells (Figs. 11 and 12). Some groups of supporting cell nuclei were also labelled (Figs. 13, 14 and 15). In 2 mice killed on day 32, there was a similar pattern of labelling, except
t.O
2
305 that most nuclei were lightly labelled (cf. Figs. 15 and 16), indicating that further cell division had taken place after the period of thymidine administration.
Electron microscopy At 50 days survival the regenerating sensory epithelium contained typical neurosensory cells lying in their normal position deep to the supporting cell layer. They had pale nuclei and the perikaryal cytoplasm contained a prominent mass made up of stacks of rough endoplasmic reticulum, and a characteristic and equally prominent, large organelle consisting of a paracrystalline array of regularly folded double smooth membranes of the type illustrated by Kolnberger 12 (see his Fig. 4b) and Bannister and Cuschieri 1 (see their Figs. 3 and 6). The sensory dendrites of the neurosensory cells extend through the supporting cell layer to reach the lumen, where they form a normal ending containing centrioles, surmounted by villi on the luminal surface and forming specialised junctional complexes with the adjacent supporting cellsS,6,1a, 29. In structure, therefore, the newly formed neurosensory cells resemble those in normal, unoperated mice. No systematic search was made to identify possible immature precursors of these neurosensory cells. Electron microscopy showed that the circumscribed acellular areas found in the light micrographs of the regenerating epithelium were in fact neuromata, consisting of masses of apposed unmyelinated axons whose structure resembled that of normal vomeronasal nerve axons. The arrangement of axons and ensheathing cell processes in the neuromata varied from an appearance (Fig. 18) close to that of normal vomeronasal nerves (Fig. 17) to a more disorganised structure. The areas resembling normal vomeronasal nerves contained masses of aligned parallel axons with a fairly uniform diameter of around 0.2-0.3 #m, and between 5 and 10 microtubules per profile. The only other notable axonal contents were elongated mitochondria and occasional dense cored vesicles of around 100 nm diameter. Groups of around 20 axons were segregated into irregularly shaped, interconnecting bundles by slender, finger-like processes of non-nervous cells whose cell bodies were closely applied to the nerve bundles. When compared with the normal vomeronasal nerves (cf. Figs. 17 and 18), however, even the most highly organized neuromatous masses did not achieve the same regularity
Fig. 7. VNO 16 days after olfactory bulbectomy to show the return of the newly formed receptor cells in patches (arrows) to the basal parts of the depopulated receptor cell layer. In the superficial part of the receptor cell layer there is a relatively cell free zone still remaining between the newly formed receptor cells and the supporting cell layer (s). Scale bar, 200 pm. Fig. 8. VNO 20 days after olfactory bulbectomy to show further re-population of the receptor cell layer (cf. Fig. 7) but still with some preferential localisation of the newly formed cells in the basal parts of the receptor cell layer. Scale bar, 200/~m. Fig. 9. VNO 32 days after olfactory bulbectomy to show nearly complete repopulation of the receptor cell layer but with localised acellular areas (arrows), which electron microscopy shows to be vomeronasal nerve neuromata. Scale bar, 200/~m. Fig. 10 VNO 112 days after olfactory bulbectomy: stained with protargol silver to show the sensory dendrites (arrows) of the newly formed receptor cells (r) extending through the supporting cell layer (s) to the lumen of the VNO (L). Two large neuromatous areas (N). Scale bar, 50/~m.
307 of axon arrangement, nor the same economy in the number of sheathing cell processes used to divide up the axon bundles: in the neuromata there are always fewer axons per bundle, and some degree of redundancy (i.e. reduplication) of sheathing cell processes. In the more disorganised areas, the axons are less regularly aligned, more varicose, and have large expansions packed with dense cored vesicles, mitochondria, large irregular dense bodies and autophagic vacuoles. Irregular fine filopodia extend from these expansions, and some profiles also show axoplasm with interlacing wispy material. These appearances are typical of those found at the growing ends of regenerating peripheral nerves in a variety of sites (e.g., ref. 19). An additional feature of such areas was the presence of clusters of vesicles in the axons. These vesicles were somewhat irregular in size and shape, but basically consisted of clear centred vesicles of around 50 nm diameter with occasional dense cored vesicles of 100 nm diameter. Occasionally the vesicles were clustered towards the outer surface of the axons at points of direct contact with non-neuronal sheathing processes ('synaptoid' contacts). In many areas (Fig. 20) the axons had a structure resembling the vomeronasal nerve terminals in the glomeruli of the normal accessory olfactory bulb (Fig. 19). The axons expanded into large irregular profiles which were aggregated in clusters. In such regions the axons had a dark background matrix and vesicle clusters of the type found in the glomerular terminals of normal vomeronasal nerve, the main difference being that the neuromatous masses had no indications of any kind of postsynaptic elements. DISCUSSION We find that after vomeronasal nerve section or removal of the accessory olfactory bulb, the majority of vomeronasal neurosensory cells have disappeared by 8 days. The time course of degeneration agrees with Nagahara z5 in being somewhat slower than that of the neurosensory cell s in the main olfactory mucosa. This reaction is selective for the neurosensory cells - - the supporting cells and the cells adjacent to the boundaries with the non-sensory epithelium are spared. In contrast to Nagahara 25 we find that at longer survival times the VNO progressively restores its cellularity by mitotic division of an unidentified stem cell type,
Figs. 11 and 12. Bright and dark field photographs of the same section through the VNO 20 days after unilateral olfactory bulbectomy and systemic administration of [aHlthymidine daily from days 10 to 20. Labelled nuclei are present throughout the regenerating neurosensory epithelium on the operated side (O) but are confined to clusters (arrows) in the regions of the neuroepithelium adjacent to the boundaries with the respiratory epithelium (especially the dorsal boundary) on the unoperated side (U). Seal e bar, 200 #m. Figs. 13 and 14. Bright and dark field photographs of the same section through the VNO 20 days after unilateral olfactory bulbectomy and systemic administration of [3H]thymidinedaily from days 10 to 20. Labelled nuclei are present in a high proportion of cells in the receptor cell layer and also in patches of supporting cell nuclei (arrows). Dashed lines drawn in on the dark field picture indicate the outline of the lumen (L) of the VNO. Scale bar, 100 urn.
308
Fig. 15. VNO 20 days after unilateral olfactory bulbectomy and systemic administration of [aH]thymidine daily from days 10-20. Many nuclei in the receptor cell layer (r) are heavily labelled and there are also some heavily labelled supporting cell nuclei (arrows). L, lumen; s, supporting cell layer. Scale bar, 50/~m. Fig. 16. VNO 32 days after unilateral bulbectomy and systemic administration of [aH]thymidine daily from days 10-20. There are still occasional heavily labelled cell nuclei, (white arrowhead) but compared with the previous stage (Fig. 15) the majority of nuclei in the receptor layer (r) are much more lightly labelled, implying that cell division has continued in the intervening period. Patches of supporting cell nuclei are also still heavily labelled (arrows). The apparent extracellular location of some silver grains is partly due to 'crossfire', and partly to grains lying over tangentially sectioned nuclei not in the plane of focus; both factors are accentuated by the thickness (7 tzm) of the section. Scale bar, 50/zm.
Fig. 17. A transverse section of part of a normal vomeronasal nerve fascicle taken at the level where the nerves lie on the medial surface of the olfactory bulb. The majority of axons have between 3 and 5 microtubules. N, part of the nucleus of a sheathing cell giving rise to processes (arrows). Scale bar,
1/~m. Fig. 18. Part of an intraepitbelial neuroma from the VNO of a mouse 50 days after OBX. The majority of axons are cut transversely. N, part of the nucleus of a sheathing cell; arrows, sheathing cell processes. (*) Basement membrane lined channels extending in from the external surface of the nerve. Scale bar, 1/~m. Fig. 19. A section across the edge of a glomerulus from the AOB of a normal mouse. Electron dense vomeronasal nerve terminals make synaptic contacts (arrows) with paler, mitral cell dendritic branches. g, capsule of glial, or possibly neuronal 29 sheet processes around the glomerulus. Scale bar, 1/~m. Fig. 20. Part of a neuromatous region from the vomeronasal neuroepithelium of a mouse 50 days after OBX, to show the resemblance to the normal glomerular terminals of the vomeronasal axons. Vesicle filled electron dense axonal profiles form an interwoven mass with no recognisable postsynaptic elements present, Scale bar, 1 t~m,
309
310 followed by migration and differentiation. The epithelium is repopulated by what are, at least structurally, normal receptor cells: protargol silver staining at 112 days survival showed the presence of sensory dendrites, and electron microscopy at 50 days survival confirmed this observation and showed that the epithelium contained fully differentiated receptor cells and typical vomeronasal axons. The majority of receptor nuclei in the regenerated epithelium at 20-32 days after operation can be labelled by prior administration of thymidine over the period from 10 to 20 days after operation. This indicates that these cells are not receptor cells which have somehow persisted unidentified from before the lesion (cf. ref. 14), since mature receptor cells have not been observed to undergo mitosis. One has to conclude that the labelled receptor cells have arisen by division and subsequent differentiation of a stem cell. The use of [aH]thymidine labelling also shows that, in contrast to Graziadei's findings in the frog olfactory epithelium 7, some new supporting cells are also formed during the process of regeneration of the mouse vomeronasal epithelium. Neither the location nor the identity of the vomeronasal stem cell is clear. In the olfactory epithelium the stem cell is basal (e.g. ref. 7, 9, 21). However, the VNO lacks an obvious basal layer of cells4,1°,22, and in our material from the vomeronasal epithelium after retrograde degeneration of the receptors we find no persisting basal layer of cells such as is seen in the main olfactory epithelium under these circumstances 7,e5. Nevertheless, repopulation of the regenerating vomeronasal epithelium seems to commence in the basal layers. It is possible that isolated basal cells are present in small numbers and have evaded discovery. However, it seems likely that at least part of the regeneration occurs by proliferation of cells at the boundaries of the sensory epithelium: these cells are actively dividing in the normal adult animal 2, and they persist in the acute phase of the retrograde reaction after vomeronasal nerve destruction. The present material does not enable us to reconstruct the precise migration path taken by the progeny of the stem cell division (see also ref. 23). During the stage of the most acute retrograde loss of neurosensory cells, the epithelium has a persisting regular skeleton (Fig. 6) made up of the radial processes of the supporting cells, which form a parallel array extending the full depth of the epithelium from the cell bodies at the luminal surface to the end feet on the basal lamina a. Possibly these processes offer some kind of guidance to the migrating neurosensory cells (e.g. analogous to the radial glia in the developing cerebellum2S). On the other hand, the supporting cell framework is not itself entirely composed of cells persisting from before the retrograde neurosensory cell loss, since we have observed patches of thymidine labelled supporting cells in the regenerated epithelium, implying that at least some supporting cells are formed as a result of the stem cell division stimulated by the retrograde loss of the neurosensory cells. The regenerating neurosensory epithelium is in general not completely repopulated, and contains neuromata consisting of unmyelinated axons closely resembling normal vomeronasal nerve axons. Since such neuromata were not seen in normal mice, they probably represent new axon outgrowth. In view of their common occur-
311 rence in the regenerating epithelium it is likely that they originate from the newly produced neurosensory cells, although we cannot completely rule out the unlikely possibility that a small number of neurosensory cells might somehow have persisted undetected after the original lesion and given rise to all the new axons in the neuromata. The axons form fairly normally arranged bundles and have typical growing tips. In some places there are signs of obstructed growth, with the formation of neuromatous tangles, and in other places the axons have a fully differentiated terminal structure and appearance, although they have not grown further than the epithelium and have not made contact with any identifiable postsynaptic elements. Studies on regeneration in the main olfactory system have mentioned neuroma formation at the site of surgical transection in regenerating pigeon olfactory nerve 26. Schultz 80 reported that axon bundles were unusually prominent in the lamina propria of the epithelium regenerating after zinc sulphate treatment in the monkey. Apart from this there have been no reports of intraepithelial neuromata such as we describe here in the regenerating VNO. The implication may be that regenerating olfactory receptors do not produce axons, or that they produce axons which distribute randomly, and are therefore not seen as obvious neuromata, or (which seems the most likely) that the regenerating olfactory axons return to the olfactory bulb in an orderly manner and with high efficiency in selecting their routes, so that axons do not become obstructed and diverted into neuromatous masses. This last suggestion is supported by reports of return of olfactory function after damage to the primary olfactory neurones. There are many clinical accounts of return of the sense of smell after its loss following head injury or upper respiratory tract surgery a4. After olfactory nerve section in pigeons ~6 and goldfish 37 the olfactory nerves have been seen to regenerate back to the bulbs and in both cases the histological evidence of regeneration was accompanied by recovery of the ability to perform olfactory discrimination tasks. Further supporting evidence for the regrowth of olfactory axons to the bulbs is provided by Schultz's finding a°,al of a return of susceptibility to infection by intranasal poliovirus after experimental damage to the olfactory mucosa in monkeys. During the course of the present work, the results of Harding et al. 11 became available; in this study the olfactory nerves were cut in mice, and it was shown that olfactory axons and terminals disappeared from the olfactory bulb glomeruli, together with three biochemical markers known to be specifically contained in primary olfactory neurones. The subsequent reappearance (with similar time course) of axons, terminals, and the biochemical markers provided good evidence that a high proportion of regenerating olfactory axons do return to the bulb (see also ref. 8). We now have some preliminary experimental results using an autoradiographic tracing method which provides direct evidence that after destruction of the vomeronasal nerves in the mouse at least some of the newly formed vomeronasal neurosensory cells have axons which reach as far as the olfactory bulb. Further work is in progress to investigate the extent to which these regenerating axons are capable of innervating bulb tissue.
312 REFERENCES 1 Bannister, L. H. and Cuschieri, A., The fine structure and enzyme histochemistry of vomeronasal receptor cells. In D. Schneider (Ed.), Olfaction and Taste, Vol. IV, Wissenschaftliche MBH, Stuttgart, 1972, pp. 27-33. 2 Barber, P. C. and Raisman, G., Cell division in the vomeronasal organ of the adult mouse, Brain Research, in press. 3 Cajal, S., Ram6n Y., Histologie du systdme nerveux de l'homme et des vertdbr~s, Maloine, Paris, 1911, 993 pp. 4 Cuschieri, A. and Bannister, L. H., The development of the olfactory mucosa in the mouse: light microscopy, J. Anat. (Lond.), 119 (1975) 277-286. 5 Graziadei, P. P. C., Topological relation between olfactory neurons, Z. Zellforsch., 118 (1971) 449-466. 6 Graziadei, P. P. C., The olfactory mucosa of vertebrates. In L. M. Beidler (Ed.), Handbook o f Sensory Physiology, VoL IV, Springer, Berlin, 1971, pp. 27-58. 7 Graziadei, P. P. C., Cell dynamics in the olfactory mucosa, Tissue and Cell, 5 (1973) 113-131. 8 Graziadei, P. P. C. and Dehan, R. S., Neuronal regeneration in frog olfactory system, J. Cell Biol., 59 (1973) 525-530. 9 Graziadei, P. P. C. and Metcalf, J. F., Autoradiographic and ultrastructural observations on the frog's olfactory mucosa, Z. ZellJorsch., 116 (1971) 305-318. 10 Graziadei, P. P. C. and Tucker, D., Vomeronasal receptors in turtles, Z. Zellforsch., 105 (1970) 498-514. I 1 Harding, J., Graziadei, P. P. C., Graziadei, G. A. M. and Margolis, F. L., Denervation in the primary olfactory pathway of mice. IV. Biochemical and morphological evidence for neuronal replacement following nerve section, Brain Research, 132 (1977) 11-28. 12 Kolnberger, I., Vergleichende Untersuchungen am Riechepithel, insbesondere des Jacobsonschen Organs von Amphibien, Reptilien und S~ugetieren, Z. Zellforsch., 122 (1971) 53-67. 13 Kolnberger, I. and Altner, H., The occurrence of diffusion barriers and interreceptor contacts in the olfactory epithelium of the vomero-nasal organ. In D. Schneider (Ed.), Olfaction and Taste, Vol. IV, Wissenschaftliche, Stuttgart, 1972, pp. 34-39. 14 Le Gros Clark, W.E., Inquiries into the anatomical basis of olfactory discrimination, Proc. roy. Soc. B., 146 (1957) 299-319. 15 Le Gros Clark, W. E. and Warwick, R. T. T., The pattern of olfactory innervation, J. Neurol. Neurosurg. Psychiat., 9 (1946) 101-111. 16 Margolis, F. L., A brain protein unique to the olfactory bolb, Proc. nat. Acad. Sci. (Wash.), 69 (1972) 1221-1224. 17 Margolis, F. L., Carnosine in the primary olfactory pathway, Science, 184 (1974) 909-911. 18 Margolis, F. L. and Tarnoff, J. F., Site of biosynthesis of the mouse brain olfactory bulb protein, J. biol. Chem., 248 (1973) 451455. 19 Matthews, M. R., An ultrastructural study of axonal changes following construction of postganglionic branches of the superior cervical ganglion in the rat, Phil. Trans. B., 264 (1973) 479-508. 20 Matulionis, D. H., Ultrastructural study of mouse olfactory epithelium following destruction by ZnSO4 and its subsequent regeneration, Amer. J. Anat., 142 (1975) 67-90. 21 Moulton, D. G., Dynamics of cell populations in the olfactory epithelium, Ann. N. Y. Acad. Sci., 237 (1974) 52-61. 22 Moulton, D. G., C~elebi, G. and Fink, R. P., Olfaction in mammals - - two aspects: proliferation of cells in the olfactory epithelium and sensitivity to odours. In G. E. W. Wolstenholme and J. Knight (Eds.), Ciba Foundation Symposium on Taste and Smell in Vertebrates, Churchill, London, 1970, pp. 227-250. 23 Moulton, D. G. and Fink, R. P., Cell proliferation and migration in the olfactory epithelium. In D. Schneider (Ed.), Olfaction and Taste, VoL IV, Wissenschaftliche MBH, Stuttgart, 1972, pp. 20-26. 24 Mulvaney, B. D. and Heist, H. E., Regeneration of rabbit olfactory epithelium, Amer. J. Anat., 131 (1971) 241-251. 25 Nagahara, Y., Experimentelle Studien fiber die histologischenVerS.nderungen des Geruchsorgans nach der Olfactoriusdurchschneidung. Beitr~ige zur Kenntnis des feineren Baus des Geruchs. organs, Jap. J. reed. Sci., 5 (1940) 165-199. 26 Oley, N., Delian, R. S., Tucker, D., Smith, J. C. and Graziadei, P. P. C., Recovery of structure
313
27 28 29 30 31 32 33 34 35 36 37
and function following transection of the primary olfactory nerves in pigeons, J. comp. physiol. Psychol., 88 (1975) 477-495. Parsons, T. S., Evolution of the nasal structure in the lower tetrapods, Amer. Zool., 7 (1967) 397413. Rakic, P., Neuron-glia relationship during granule cell migration in developing cerebellar cortex. A Golgi and electronmicroscopic study in Macacus rhesus, J. comp. Neurol., 141 (1971) 283-312. Reese, T. S. and Brightman, M. W., Olfactory surface and central olfactory connexions in some vertebrates. In G. E. W. Wolstenholme and J. Knight (Eds.), Taste and Smell in Vertebrates, Churchill, London, 1970, pp. 115-150. Schultz, E. W., Regeneration of olfactory cells, Proc. Soc. exp. Biol., (N. Y.), 46 (1941) 41-43. Schultz, E. W., Repair of the olfactory mucosa. With special reference to regeneration of olfactory cells (sensory neurons), Amer. J. Path., 37 (1960) 1-19. Smith, C. G., Changes in the olfactory mucosa and the olfactory nerves following intranasal treatment with one per cent zinc sulphate, Canad. med. Ass. J., 39 (1938) 138-140. Smith, C. G., Regeneration of sensory olfactory epithelium and nerves in adult frogs, Anat. Rec., 109 (1951) 661-671. Sumner, D., Post-traumatic anosmia, Brain Research, 87 (1964) 107-120. Takagi, S. F., Degeneration and regeneration of the olfactory epithelium. In L. M. Beidler (Ed.), Handbook of Sensory Physiology, Vol. IV, Springer, Berlin, 1971, pp. 75-94. Thornhill, R. A., Cell division in the olfactory epithelium of the lamprey, Lampetra fluviatilis, Z. Zellforsch., 109 (1970) 147-157. Zippel, H. P., Baumgarten, R. yon and Westerman, R. A., Histologische, funktionelle und spezifische Regeneration nach Durchtrennung der ilia olfactoria beim Goldfisch, Z. vergl. PhysioL, 69 (1970) 79-98.