Neuroscience Vol. 84, No. 2, pp. 559–567, 1998 Copyright  1998 IBRO. Published by Elsevier Science Ltd Printed in Great Britain. All rights reserved Pergamon PII: S0306-4522(97)00503-4 0306–4522/98 $19.00+0.00

DIRECT EVIDENCE OF TRIGEMINAL INNERVATION OF THE COCHLEAR BLOOD VESSELS

Z. VASS,*† S. E. SHORE,†‡ A. L. NUTTALL†§¶ and J. M. MILLER† *Department of Otorhinolaryngology, Albert Szent-Gyo¨rgyi Medical University, Szeged, Hungary, H-6725 †Kresge Research Institute, Department of Otolaryngology, University of Michigan, Ann Arbor, MI 48109-0506, U.S.A. ‡Department of Otolaryngology, Medical College of Ohio, Toledo, OH 43699, U.S.A. §Oregon Hearing Research Center, NRC04, Oregon Health Sciences University, Portland, OR 97201-3098, U.S.A.

Abstract––This paper provides the first detailed description of the trigeminal innervation of the inner vasculature. This system provides a newly discovered neural substrate for rapid vasodilatatory responses of the to high levels of activity and sensory input. Moreover, this discovery may provide an alternative mechanism for a set of clinical disturbances (imbalance, hearing loss, tinnitus and headache) for which a central neural basis has been speculated. Iontophoretic injections of biocytin were made via a glass microelectrode into the trigeminal ganglion in guinea-pigs. Tissue for histological sections was obtained 24 h later. Labeled fibers from the injection site were observed as bundles around the ipsilateral spiral modiolar blood vessels, as individual labeled fibers in the interscala septae, and in the ipsilateral stria vascularis. The dark cell region of the cristae ampullaris in the vestibular labyrinth was also intensively labeled. No labeled fibers were observed in the neuroepithelium of the cristae ampullaris or the . These results confirm and localize an earlier indirect observation of the trigeminal ganglion projection to the . This innervation may play a role in normal vascular tone and in some inner ear disturbances, e.g., sudden hearing loss may reflect an abnormal activity of trigeminal ganglion projections to the cochlear blood vessels.  1998 IBRO. Published by Elsevier Science Ltd.

Key words: vestibular labyrinth, headache, herpes zoster, sensory neuron, hearing loss, guinea-pig.

Trigeminal neurons may play a role in maintaining blood flow that is responsible for generating vascular normal vascular tone and causing pathophysiological headaches.17 vascular responses in the cochlea.18 Indirect evidence Sympathetic adrenergic innervation of the cochlea for this includes observations that: (i) auditory and extends from the cerebral arterial system to the vestibular symptoms often occur with migraine head- modiolar vessels,28 and physiological studies have ache,21,35 which may be related to trigeminal vascular demonstrated that the sympathetic nervous system disturbances; and (ii) some neurotrophic viruses such plays a significant role in the control of cochlear as herpes zoster, which are localized in the trigeminal microcirculation.22 ganglion, may cause sensorineural hearing loss.7 In contrast to the innervation of cerebral blood The cochlea receives its blood supply from the vessels, parasympathetic innervation of cochlear anterior inferior cerebellar artery, which is a branch blood vessels has not been found. However, the of the basilar artery,4 and cochlear vasoregulation presence of substance P, calcitonin gene-related shows similarities to regulation of cerebral blood peptide and neuropeptide Y in the guinea-pig spiral flow. Cerebral arteries are innervated by several modiolar artery (as in cerebral blood vessels36) systems that contribute to the control of brain blood suggests that other vasodilative mechanisms may be flow.5 Parasympathetic cholinergic fibers increase present.8,30 Additional support for this view is the cerebral blood flow by releasing dilating transmit- demonstration by Vass et al.31 that capsaicin has a ters.25 Extraparenchymal cerebral arteries are also direct effect on cochlear blood flow, suggesting a role innervated by fibers originating mainly from the of primary sensory neurons in cochlear blood flow trigeminal ganglion. These fibers also release dilating regulation. The first evidence that the trigeminal transmitters,1 contributing to an increased cerebral ganglion innervates the cochlea was demonstrated by infusion of a retrograde tracer into the cochlea.34 ¶To whom correspondence should be addressed. However, the terminal endings, pathways and target Abbreviations: DAB, 3,3*-diaminobenzidine tetrahydro- vessels of these putative sensory neurons in the chloride; EDTA, ethylenediaminetetra-acetate. cochlea are unknown. 559 560 Z. Vass et al.

This study was undertaken to provide a direct and vestibular labyrinth. The sections were mounted on measure of any trigeminal ganglion projections to the to chrome–alum–gelatin-coated slides for subsequent peroxidase reaction. cochlea, through the use of anterograde transport of biocytin injected into the trigeminal ganglion in the guinea-pig. Localization of biocytin

After several washes in phosphate-buffered saline, all the EXPERIMENTAL PROCEDURES sections were incubated for 2 h at room temperature in avidin-D–horseradish peroxidase (Vector Laboratories, Pigmented guinea-pigs (250–400 g; NIH Outbred strain, Inc., No. A-2004) dissolved in 1% Triton X-100 (Sigma Murphy Breeding Laboratory, Plainfield, IN, U.S.A.) were 9002-93-1) in 0.1 M phosphate buffer (pH 7.4). The sections anesthetized with ketamine hydrochloride (Ketaset; 80 mg/ were rinsed in three changes of 0.1 M phosphate buffer kg) and xylazine (Rompun; 4 mg/kg) administered intra- (pH 7.4). For enzymatic reaction to reveal peroxidase muscularly. Periodic drug supplements were used to label, the sections were incubated for 7 min in 0.075% maintain anesthetic levels throughout the procedure. The 3,3*-diaminobenzidine tetrahydrochloride (DAB; Sigma; n guinea-pigs ( =9) were placed in a stereotaxic frame (David D-5905) and 0.005% H2O2 (Sigma; H-1009). Following the Kopf) and a longitudinal incision was made in the scalp. reaction, the sections were washed three times for 5 min in Using stereotaxic coordinates (0.37 cm caudal to bregma, 0.1 M phosphate buffer. The slides were counterstained for 0.45 cm lateral from the midline and 1.3 cm ventral to 3.5 min in 1% aqueous Neutral Red (I.T. Baker, Inc.; R bregma), a small hole was drilled in the skull without 747-03) buffered at pH 4.8 with acetate buffer. The slides disturbing the meninges, and a glass micropipette (tip were washed in distilled water, dehydrated in graded diameter 25–30 µm; World Precision Instruments Co.) was ethanol, cleared in xylene and coverslipped using Permount. lowered by a micromanipulator into the left trigeminal The reacted sections were studied using a Leitz Dialux ganglion. Iontophoretic injections of biocytin were delivered microscope equipped with a drawing tube. Anterogradely via a glass microelectrode using a constant-current genera- labeled neurons and neuron terminals in the trigeminal tor (Stoelting, 1 µA a.c. for 20 min). Biocytin (Sigma ganglia, cochlea, vestibular labyrinth, and Chemical Co., St Louis, MO, U.S.A.; B-4261) was prepared were mapped using a camera lucida. as a 5% solution in 0.05 M Tris buffer (pH 7.6). The solution could be stored at 4)C for up to seven days with no 14 noticeable loss of activity. Advantages of the use of Controls biocytin over intracellular markers (i.e. horseradish peroxi- dase and Lucifer Yellow) are its small size and its high Vehicle control. In two animals, a glass microelectrode affinity for avidin. The small size of biocytin allows the filled with 0.05 M Tris buffer (pH 7.6) was positioned in a experimenter to use unbeveled electrodes (i.e. glass micro- stereotaxic frame and iontophoretic injections were made pipettes filled with an electrolyte) for injection and also into the left trigeminal ganglia. After 24 h, the ipsilateral permits a rapid diffusion within the cell far from the and contralateral trigeminal ganglia and cochlea were injection sites. Dental cement was used to seal the opening, prepared for histological evaluation. and the animal was sutured and allowed to recover. After 24 h, the animal was deeply anesthetized with pentobarbital Endogenous peroxidase control. In order to test for the and perfused transcardially with 200 ml of 0.05% sodium presence of endogenous peroxidase activity in trigeminal nitrite in normal saline, followed by 750 ml of mixed primary afferent neurons, the trigeminal ganglia and coch- aldehyde fixative (1.25% glutaraldehyde and 1.0% parafor- leae of two control animals, not exposed to avidin-D– maldehyde in phosphate buffer; pH 7.4). The flow of fixative horseradish peroxidase, were fixed by vascular perfusion, to the cochleae, vestibular labyrinth and trigeminal ganglia sectioned serially at 30 µm and processed according to by way of the common carotid arteries was maximized the DAB technique. Sections were then examined under by occlusion of the brachial arteries and the descending bright- and dark-field illumination. aorta. Following perfusion–fixation, the trigeminal ganglia were quickly dissected and placed in fresh fixative (2% paraformaldehyde–2.5% glutaraldehyde in 0.1 M phosphate Control for non-specific labeling. To control for the possi- buffer, pH 7.2–7.4) for 12 h at 4)C. The tissue was then bility that some of the visualized reaction product may have immersed overnight in 30% sucrose in 0.12 M phosphate resulted from the non-specific deposition of incubation buffer (pH 7.4). The ipsilateral and contralateral tympanic medium components, several sections from each experimen- bullae, cochlea and vestibular labyrinth were also removed tal animal were reacted in an incubation medium lacking the after perfusion–fixation. The apical end of the cochlea was chromogen DAB. opened and the same cold fixative was gently perfused through the opening in the apex and the . Chemical control. Finally, to investigate the possibility The cochlea and the vestibular labyrinth were then that the visualized reaction product resulted from the non- postfixed overnight and decalcified for four weeks in 3% enzymatic hydrolysis of incubation medium components, EDTA. Each specimen was frozen with dry ice in 100% several sections from each experiment were heated for ethanol and serially sectioned at 30-µm increments at the 15–20 min at 90–100)C to inactivate the enzymes prior to trigeminal ganglion and at 15 µm through the cochlea incubation in the complete medium.

Abbreviations used in the figures

AICA anterior inferior cerebellar artery ME middle ear BA basilar artery Oph ophthalmic branch of the trigeminal ganglia EE external ear SM scala media ET Eustachian tube SMA spiral modiolar artery ISS interscala septae ST scala tympani Mand mandibular branch of the trigeminal ganglia SV scala vestibuli Max maxillar branch of the trigeminal ganglia SV stria vascularis Trigeminal ganglia and cochlear blood vessels 561

Fig. 1. Schematic of the anatomical relationship of the trigeminal ganglia, cochlea, vestibular labyrinth and basilar artery. The cochlea and vestibular labyrinth receive innervation from the ophthalmic branch of the trigeminal ganglia through the basilar artery and anterior inferior cerebellar artery. In contrast, the middle ear and Eustachian tube receive their innervation from the mandibular branch of the trigeminal ganglia. Also shown are the anatomical sites of Figs 2–6.

RESULTS labeled cells or axons in the contralateral trigeminal ganglion. Localization of labeled cells Trigeminal ganglion. The light microscopic resolu- Cochlear . After 24 h survival time follow- tion of the neurons and axons labeled by the uptake ing injections of 5% biocytin into the trigeminal of biocytin permit detailed examination of the ganglion, mid-modiolar serial sections revealed projection pathways and terminal fields (Fig. 1). labeled fibers emerging from the injection site and The biocytin injection sites were characterized by a collecting into bundles around the ipsilateral spiral central core of labeled neurons that appeared to be modiolar blood vessels. The terminal labeling was completely filled by biocytin, giving a Golgi-like adjacent to blood vessels, whose lumens are apparent appearance. Most cells around the injection site in many sections. Labeled bundles were observed exhibited a very dense labeling, with the reaction from the basal to the apical turns of the spiral products appearing as stippled cytoplasmatic gran- modiolar blood vessel (Figs 1, 3). ules (Fig. 2a). The granules were small, of fairly Labeling was found to alternate from one side of uniform size, roughly spherical in shape and evenly the modiolus to the other in a stair-step fashion, distributed throughout the cytoplasm. The nucleus consistent with the spiral orientation of the modiolar was clearly visible in the majority of labeled neurons. vessels observed in the mid-modiolar cross-sections. Labeled axons were seen emerging from the central In the basal turn, where there are more vessels, the core of the injection site (Fig. 2b). There were no labeled fibers gave rise to many collaterals that 562 Z. Vass et al.

Fig. 2. (a) Bright-field micrograph of neurons and axons labeled by the uptake of biocytin permits detailed examination of the projection pathways and terminal fields in the ipsilateral trigeminal ganlgion. The injection site is characterized by a central core of labeled neurons that appears to be completely filled by biocytin, giving a Golgi-like appearance. Most cells around the injection site exhibit a dense labeling, with the reaction products appearing as stippled cytoplasmatic granules. Scale bar=25 µm. (b) Bright-field micrograph of the ipsilateral trigeminal ganglion 24 h following injection of 5% biocytin into the trigeminal ganglion. Labeled axons are seen emerging from the central core of the injection site. The axons show parallel arborization. Scale bar=200 µm. formed a dense arborization of highly varicose fibers. convey the radiating arterioles and radial collecting Figure 4a and b shows a modiolar section of the venules supplying and draining the stria vascularis. cochlea and clearly demonstrates a labeled spiral The neurons that had taken up the biocytin were modiolar artery. often extensively labeled, producing a Golgi-like The spiral modiolar artery is situated in the inter- appearance. They then continued radially along the space between the acoustic nerve centrally and the external wall of the membranous part of the scala spiral ganglion peripherally. In mid-modiolar sec- vestibuli and terminated in the stria vascularis. tions, we were not able to determine whether the labeling was associated with the spiral modiolar Stria vascularis. In all cases, a large number of artery only or both the artery and the vein. anterogradely labeled fibers and terminal boutons was There were also some labeled fibers in the spiral seen in the ipsilateral stria vascularis (Figs 1, 5a). ganglion. These bundles did not show any significant These fiber bundles are typically localized only in the localization and there was no relationship to blood stria vascularis, which is well separated from the spi- vessels. The specificity of the labeling procedure is ral ligament. The labeled structures were brown and confirmed by the total absence of peroxidase reaction amorphous in appearance and were homogeneously products in the surrounding area. stained. The labeled fiber bundles ran together with On the contralateral side there was only light the vessels and followed their arborization (Fig. 5b). labeling around the spiral modiolar artery. Labeled neurons could also be seen in the cross- section of the vessel of the spiral prominence. There Interscala septae. The course of fibers antero- were some labeled nerve fibers on the contralateral gradely labeled with biocytin and their terminal fields side around the stria vascularis, but the density of could be followed in the interscala septae, which these granules was much less than on the ipsilateral Trigeminal ganglia and cochlear blood vessels 563

Middle ear. Following biocytin injections into the trigeminal ganglion, labeled fibers were seen emerg- ing into the ipsilateral middle ear mucosa (Fig. 1). Labeled fibers were also localized in the Eustachian tube (Fig. 7). In addition, there were some labeled terminals under the mucosa of . On the contralateral side, there were also a few labeled boutons but the density was much less than on the ipsilateral side.

Controls. The specificity of biocytin neurochemis- try used in the current study was confirmed by the negative results obtained for all sham operations and for all controls. The sham-operated vehicle control in normal animals demonstrated no peroxidase labeling in either the ipsi- or contralateral trigeminal ganglia, in the cochlea, the labyrinth or the middle ear sections, indicating the specificity of biocytin.

DISCUSSION Although we know that adrenergic neurons inner- vate at least a portion of the cochlear vasculature,28 and sympathetic nerve fibers terminate on the spiral modiolar artery and extend to the radiating arteri- oles, radial collecting venules and spiral modiolar vein in the gerbil,6 an innervation of the richly vascular lateral wall is unknown. Many authors have suggested that the modiolar spiral artery is most important for adrenergic regulation of cochlear Fig. 3. Bright-field micrograph of the ipsilateral cochlea 27 24 h following injections of 5% biocytin into the trigeminal blood flow. In contrast, there is only indirect evi- ganglion. Labeled bundles from the basal to the apical turns dence supporting non-sympathetic control of vaso- of the spiral modiolar blood vessel are seen. Labeling was dilator tone.31,33 Moreover, the stria vascularis, the found to alternate from one side of the modiolus to the site of production, is perhaps the most other in a stair-step fashion, consistent with the spiral important vascular network in the cochlea, control- orientation of the modiolar vessels observed in the mid- modiolar cross-sections. In mid-modiolar sections, we were ling cochlear function and normal metabolic, ion and not able to determine whether the labeling was associated fluid balance.39 The intense oxidative enzyme activity with the spiral modiolar artery only or both the artery and in the stria vascularis is at a level almost equal to the the vein. Scale bar=200 µm. dark cells of the vestibular labyrinth.13 There is also a paucity of information about dark cell area vaso- side. The reaction product was easily distinguishable regulation mechanisms, particularly non-sympathetic from red blood cell peroxidase reaction product. regulation. In the current study, we found a significant number Vestibular labyrinth. There was a very intensely of labeled fibers related to the spiral modiolar artery labeled area in the ipsilateral dark cell region of the 24 h following injections of biocytin into the trigemi- . The area was sharply localized nal ganglion. These fibers probably play an import- close to the crista ampullaris and was observed in all ant role in maintaining vascular tone, in coordination the ampullae (Figs 1, 6). with the sympathetic innervation. Recent data No reaction product was observed around the revealed that labeled adrenergic fibers could be fol- vessels outside the dark cells. The dark cell region is lowed to the radiating arterioles and collecting a structure similar to the stria vascularis. The antero- venules.6 Similarly, our data suggest that the antero- gradely labeled fibers and terminal boutons were gradely labeled trigeminal fibers follow the interscala related to blood vessels. There were no labeled fibers septae, and are related to the radiating vessels and in the crista ampullaris or any other parts of the supply a very rich innervation to the stria vascularis. semicircular canal. The reaction product was distin- The localization and relationship to blood vessels of guishable from the subepithelial layers of melano- labeled fibers and boutons in the dark cells in the cytes, which are typical in the dark cell area. On the vestibular labyrinth were very similar to those in the contralateral side, there were a few labeled boutons stria vascularis, thus providing additional evidence but the density was much less than on the ipsilateral for a morphological and functional relationship side. between the stria vascularis and the dark cell region. 564 Z. Vass et al.

Fig. 4. (a) Bright-field micrograph shows a modiolar section of the cochlea and clearly demonstrates a labeled spiral modiolar artery. In the basal turn, where there are more vessels, the labeled fibers gave rise to many collaterals that formed a dense arborization of highly varicose fibers. Scale bar=50 µm. (b) Bright-field micrograph shows a modiolar section of the cochlea and clearly demonstrates a labeled spiral modiolar artery. The terminal labeling is adjacent to blood vessels, whose lumens were apparent in many sections. Scale bar=25 µm.

We postulate that the anterogradely labeled trigemi- These data indicate that the sensory innervations of nal neurons in the cochlea and vestibular labyrinth the middle ear and the inner ear are distinct. are the same as those retrogradely labeled by cochlear With regard to contralateral innervation of the infusion of wheatgerm agglutinin conjugated to horse- inner ear, this study also reveals minimal labeling of radish peroxidase described by Vass et al.34 The retro- the contralateral spiral modiolar artery, stria vascu- gradely labeled sensory nerves originated in the laris, vestibular labyrinth and middle ear. Edvinsson medial part of the trigeminal ganglion, mainly at the et al.10 have shown contralateral innervation of origin of the ophthalmic nerve. A similar ophthalmic the middle cerebral arteries of the rat using the origin exists for the middle cerebral artery, the dural fluorescent tracer True Blue. artery and the apical part of the basilar artery.40 The Given the observations of the current study, it is ophthalmic branch of the trigeminal nerve may par- interesting to speculate that this pathway may also be ticipate in the pathogenesis of migraine headaches, involved in tinnitus, vertigo and hearing loss in neurogenic inflammation and vascular permeability migraine and herpes zoster craniofacialis. There are due to the release of substance P and calcitonin many transient neurological disturbances associated gene-related peptide at the nerve terminal.2,29 with various types of migraine.26 Visual symptoms, There were also anterogradely labeled nerve fibers such as scintillating scotoma, are most common but in the middle ear, the mastoid cells and the region of somatosensory, motor, cranial nerve and brainstem the Eustachian tube, consistent with data provided by symptoms also occur.19 Among the brainstem symp- Oyagi et al.20 According to their study, the middle ear toms, vestibular manifestations are quite common mucosa is mainly innervated by the mandibular divi- and include non-specific dizziness, disequilibrium, sion of the trigeminal nerve. Vass et al.34 also demon- vertigo and motion intolerance.12,15 There are strik- strated retrograde labeling in the mandibular branch ing similarities between the occurrence of headache of the trigeminal ganglion after wheatgerm agglutinin– symptoms in cluster headaches and vertigo in benign horseradish peroxidase applications in the middle ear. positional vertigo or in Me´nie`re’s disease.11 Trigeminal ganglia and cochlear blood vessels 565

Fig. 5. (a) Bright-field micrograph of the ipsilateral cochlea shows a large number of anterogradely labeled fibers and terminal boutons in the stria vascularis. These fiber bundles are typically localized only in the stria vascularis, which is well separated from the spiral ligament. The labeled structures were brown and amorphous in appearance and were homogeneously stained. Scale bar=25 µm. (b) Bright-field micrograph of the ipsilateral stria vascularis 24 h following injection of 5% biocytin into the trigeminal ganglion. The anterogradely labeled fiber bundles run together with the vessels and show its arborization. Labeled neurons could also be seen in the cross-section of the vessel of the spiral prominence. Scale bar=10 µm.

Simultaneous involvement of two cranial nerves by also evidence for a functional relationship between the herpes zoster virus has been recorded previously. capsaicin-sensitive sensory neurons of the anterior Herpes zoster oticus and facial paralysis are inferior cerebellar artery and cochlear blood vessels. frequently associated with tinnitus aurium and pro- The capsaicin-sensitive sensory neurons’ effect on gressive diminution of hearing, and recorded herpes cochlear blood flow is mediated by substance P, as zoster ophthalmicus is associated with ipsilateral demonstrated by a decrease in the capsaicin effect on deafness with and without vertigo.24 Hunt’s syn- cochlear blood flow after application of a substance P drome is recognized as a polycranial neuritis charac- blocker.33 Capsaicin-sensitive fibers have been shown terized by damage to sensory and motor nerves, to release substance P, and bilateral removal of including the auditory–vestibular apparatus.37 The the trigeminal ganglia decreases concentrations of frequent initial signs and symptoms, auricular substance P in all forebrain vessels, including the pain and herpetic rash, are correlated with initial rostral basilar artery.9,23 The substance P-containing hearing loss,38 but the anatomical pathways of capsaicin-sensitive sensory neurons responsible for migraine-caused vertigo and hearing disorders, as dilatation of cochlear vessels may be the antero- well as herpes zoster-caused vestibular and cochlear gradely labeled neurons observed in the cochlea and disorders are unknown. vestibular labyrinth in the present study.34 The This trigeminal sensory innervation of the cochlea reduced responsiveness to capsaicin in endolymphatic may have an important role in cochlear blood flow hydrops also suggests that the action of peptidergic regulation and thus may play a role in the symptom sensory fibers of the inner ear may be altered in this complex as described above, and in some vascular- condition.32 Vascular permeability and neurogenic related cochlear diseases. There is an increase in inflammation are also associated with trigeminal cochlear blood flow after perivascular, intra-arterial sensory innervation.3 A disturbance in the sensory and intracochlear capsaicin application,31 which may trigeminal system may thus be a major cause of inner be regulated by primary sensory neurons.32 There is ear pathology with reduced cochlear blood supply. 566 Z. Vass et al.

Fig. 6. Bright-field micrograph of the ipsilateral vestibular labyrinth 24 h following injection of 5% biocytin into the Fig. 7. Bright-field micrograph of the ipsilateral Eustachian trigeminal ganglion. There was an intensive labeled area in tube 24 h following injection of 5% biocytin into the the dark cell region of the crista ampullaris. The area was trigeminal ganglion. There were anterogradely labeled fibers very sharply localized close to the crista ampullaris and and terminal boutons under mucosa of the Eustachian tube. was observed in all of the ampullae. No reaction product Scale bar=50 µm. was observed around the vessels outside the dark cells. Scale bar=100 µm. increases observed after reduction in sympathetic tone.16 This system may therefore play a role in the The results we have described characterize the response of the system to metabolic stress (e.g., trigeminal ganglia innervation of the cochlear blood intense noise) and may provide a target for therapeu- vessels. The trigeminal sensory innervation may play tic manipulation. This neuronal system could be a role in the balance of normal vascular tone and in the basis for certain clinical observations linking some inner ear disturbances. We also propose that hearing changes with migraine headaches and other vertigo, tinnitus, and hearing deficit associated with disorders. migraine and herpes zoster arise by the excitatory ff e ect of the trigeminal ganglion. Acknowledgements—The present work was supported by NIH Grant R01 DC00105, Fogarty International Research Collaboration Award R03 TW00502 and by an Istva´n CONCLUSIONS Sze´chenyi Academy Fellowship, Hungary. We wish to thank Dr Richard Altschuler and his staff for their assist- These findings define a new control system for ance in the histological preparations. We acknowledge inner ear blood flow, capable of producing much Nadine Brown for her assistance in the preparation of this greater changes and with shorter latency than the manuscript.

REFERENCES 1. Amann R., Schuligoi R., Lanz I. and Donnerer J. (1995) Histamine-induced edema in the rat paw—effect of capsaicin denervation and a CGRP receptor antagonist. Eur. J. Pharmac. 279, 227–231. 2. Arbab M. A. R., Wiklund L. and Swendgaard N. A. A. (1986) Origin and distribution of cerebral vascular innervation from superior cervical, trigeminal and spinal ganglia investigated with retrograde and anterograde WGA–HRP tracing in the rat. Neuroscience 19, 695–708. 3. Arvidsson B., Kristensson K. and Olsson Y. (1973) Vascular permeability to fluorescent protein tracer in trigeminal nerve and gasserian ganglion. Acta neuropath., Berlin 26, 199–205. Trigeminal ganglia and cochlear blood vessels 567

4. Axelsson A. (1968) The vascular anatomy of the cochlea in the guinea pig and man. Acta otolar., Stockh. 243, 1–134. 5. Bevan J. A. and Owman C. (1988) Cerebral circulation. In Noradrenergic Innervation of Blood Vessels (eds Burnstock G. and Griffith S. G.), pp. 37–62. CRC Press, Boca Raton, FL. 6. Brechtelsbauer P. B., Prazma J., Garrett C. G., Carrasco V. N. and Pillsbury H. C. (1990) Catecholaminergic innervation of the inner ear. Otolar. Head Neck Surg. 103, 566–574. 7. Byl F. M. and Adour K. K. (1977) Auditory symptoms associated with herpes zoster or idiopathic facial paralysis. Laryngoscope 87, 372–379. 8. Carlisle L., Aberdeen J., Forge A. and Burnstock G. (1990) Neural basis for regulation of cochlear blood flow: peptidergic and adrenergic innervation of the spiral modiolar artery of the guinea pig. Hearing Res. 43, 107–114. 9. Duckles S. P. and Buck S. H. (1982) Substance P in the cerebral vasculature: depletion by capsaicin suggests a sensory role. Brain Res. 245, 171–174. 10. Edvinsson L., Hara H. and Uddman R. (1989) Retrograde tracing of nerve fibers to the rat middle cerebral artery with True Blue: colocalisation with different peptides. J. cerebr. Blood Flow Metab. 9, 212–218. 11. Gilbert G. J. (1970) Cluster headache and cluster vertigo. Headache 9, 195–200. 12. Harker L. A. and Rassekh C. (1988) Migraine equivalent as a cause of episodic vertigo. Laryngoscope 98, 160–164. 13. Hiraide F. (1971) The histochemistry of dark cells in the vestibular labyrinth. Acta otolar., Stockh. 71, 40–48. 14. Izzo P. N. (1991) A note on the use of biocytin in anterograde tracing studies in the central nervous system: application at both light and electron microscopic level. J. Neurosci. Meth. 36, 155–166. 15. Kayan A. and Hood J. D. (1984) Neuro-otological manifestations of migraine. Brain 107, 1123–1142. 16. Miller J. M., Ren T. Y. and Nuttall A. L. (1995) Studies of inner ear blood flow in animals and human beings. Otolar. Head Neck Surg. 112, 101–113. 17. Moskowitz M. A., Reinhard J. F., Romero J., Melamed E. and Pettibone D. J. (1979) Neurotransmitters and the fifth cranial nerve: is there a relation to the headache phase of migraine? Lancet 2(8148), 883–885. 18. Moskowitz M. A. (1990) Basic mechanisms in vascular headache. Neurol. Clin. 8, 801–815. 19. Olesen J. (1990) The classification and diagnosis of headache disorders. Neurol. Clin. 8, 793–799. 20. Oyagi S., Ito J. and Honjo I. (1990) The trigeminal sensory innervation to the middle ear, Eustachian tube, pharynx: a study by horseradish peroxidase tracer method. Laryngoscope 100, 873–877. 21. Parker W. (1991) Migraine and the in adults. Am. J. Otol. 12, 25–34. 22. Ren T., Laurikainen E., Quirk W. S., Miller J. M. and Nuttall A. L. (1993) Effects of stellate ganglion stimulation on bilateral cochlear blood flow. Ann. Otol. Rhinol. Lar. 102, 378–384. 23. Saito K. and Moskowitz M. A. (1989) Contributions from the upper cervical dorsal roots and trigeminal ganglia to the feline circle of Willis. Stroke 20, 524–526. 24. Scott M. J. Sr and Scott M. J. Jr (1983) Ipsilateral deafness and herpes zoster ophthalmicus. Archs Dermatol. 119, 235–236. 25. Segade L. A. G. and Quintanilla J. S. (1990) Distribution of postganglionic parasympathetic fibers originating in the pterygoplatine ganglion in the maxillary and ophthalmic nerve branches of the trigeminal nerve; HRP and WGA–HRP study in the guinea pig. Brain Res. 522, 327–332. 26. Sicuteri F. (1985) Symptomatologic interdependence or interindependence as a key alternative in migraine and cluster headache mechanisms. Cephalalgia 5, (Suppl. 2) 233–235. 27. Snow J. B. and Suga F. (1972) Control of cochlear blood flow. In Vascular Disorders and Hearing Defects (ed. Lorenzo A. J. D.), pp. 167–183. University Park Press, Baltimore, MD. 28. Spoendlin H. (1981) Autonomic innervation of the inner ear. Adv Otorhinolaryngol. 27, 1–13. 29. Sturzenegger M. H. and Meienberg O. (1985) Basilar artery migraine: a follow-up of 82 cases. Headache 25, 408–415. 30. Uddman R., Ninoyu O. and Sundler F. (1982) Adrenergic and peptidergic innervation of cochlear blood vessels. Archs Otorhinol. 236, 7. 31. Vass Z., Bari F. and Jancso´ G. (1994) Possible involvement of capsaicin-sensitive sensory nerves in the regulation of cochlear blood flow in the guinea pig. Acta otolar., Stockh. 114, 156–161. 32. Vass Z., Brechtelsbauer P. B., Nuttall A. L. and Miller J. M. (1995) Effect of endolymphatic hydrops on capsaicin evoked increase in cochlear blood flow. Acta otolar., Stockh. 115, 754–758. 33. Vass Z., Nuttall A. L., Coleman J. K. M. and Miller J. M. (1995) Capsaicin-induced release of substance P increases cochlear blood flow in the guinea pig. Hearing Res. 89, 86–92. 34. Vass Z., Shore S. E., Nuttall A. L., Jancso´ G., Brechtelsbauer P. B. and Miller J. M. (1996) Trigeminal ganglion innervation of the cochlea—retrograde transport study. Neuroscience 79, 605–615. 35. Vijayan N. (1990) Cluster headache and vertigo. Cephalalgia 10, 67–70. 36. Wahl M. and Schilling L. (1993) Regulation of cerebral blood flow—A brief review. Acta neurochir., Vienna, Suppl. 59, 3–10. 37. Wayman D. M., Pham H. N., Byl F. M. and Adour K. K. (1990) Audiological manifestations of Ramsay Hunt syndrome. J. Laryngol. Otol. 104, 104–108. 38. Wilson W. R. (1986) The relationship of the herpes virus family to sudden hearing loss: a prospective clinical study and literature review. Laryngoscope 96, 870–877. 39. Yamane H., Nakai Y., Konishi K., Sakamoto H., Iguchi H., Nakagawa T. and Morinaka K. (1991) Cochlear strial blood circulation. Acta otolar., Stockh. 481, 182–186. 40. Zhang Q. J., Hara H. and Kobayashi S. (1993) Distribution patterns of sensory innervation from the trigeminal ganglion to cerebral arteries in rabbits studied by wheat germ agglutinin-conjugated horseradish peroxidase anterograde tracing. Neurosurgery 32, 993–999.

(Accepted 30 September 1997)