THE AMERICAN JOURNAL OF PATHOLOGY VOLUMEXXXVII JULY, I960 Nu3nr i REPAIR OF THE OLFACTORY MUCOSA WrrIH SPECIAL REFERENCE TO REGENERATION OF OLFACTORY CELTS (SENSORYNEuRONs) EDwiN W. SCHULTZ, MD. FromtheDeparmentofBacteriologyandExperimentalPathology, Stanford University School ofMedicine,Stanford, Calif. Although the normal structure of the epithelium of the olfactory mucosa has been known for many years, little has been reported on its repair. There are at least two reasons why the possible mechanism of repairofthis structureshouldbeofinterest: (i) Itisasurfacestructure in which nerve cell bodies ("olfactory cells") or a primitive type of neuron lies in a peripheral tissue. (2) The mucosa in which these cell bodies lieis not separated from the lamina propria by abasement mem- brane,butrestsdirectlyuponastructureofsomecomplexity, fromwhich different elements could participate in the repair. The present studies are an outgrowth of early work on chemical prophylaxis of experimental poliomyelitis in monkeys."2 They were prompted by observations showing that the resistance afforded by an intranasal irrigation with zinc sulfate solution a day or two prior to intranasalinoculationwitha given neuronotropic strain ofpoliomyelitis virus (MV strain) was usually followed in 3 or 4 months by resuscepti- bility to intranasal inoculation. The initial resistance induced could be explained by extensive destruction of the olfactory epithelium, thus interrupting the nervous pathway by which this strain could reach the central nervous system. It was not possible, however, for us to explain the ultimate return of susceptibility except on the possibility that some measure of restoration of the nervous pathway occurred in time. These observations, therefore,directedourattentiontothepossiblemechanism AidedinpartbyearlygrantsfromtheNationalFoundationforInfantileParalysis,Inc., NewYor. Receivedforpubliation, ber 28, 1959. 2 SCHUJLTZ Vol. 37, No. I underlying the return of susceptibility, and particularly to the pos- sibility that this might be related to a restoration of olfactory sensory neurons (olfactory cells). Apreliminarypaper on our observations was published in 1941,1 and enlarged on to some extent in 1942.2 Further observationshave beenmade from time to time since then, and it is the purpose of the present paper to summarize these. GROSS AND MICROSCOPIC STRUCTURE OF THE OLFACTORY REGION The gross anatomy of the nl passages in the rhesus monkey has been described by Geist.3 Certain qualifications of his description are required, however, onthe basis of our own observations. He stated that in the rhesus monkey, "there is nothing comparable to the human su- perior concha, this part of the wall being smooth and not appreciably elevated." He therefore recognized in the gross only the two lower conchas, whichhe designatedsuperior and inferior. While itis true that there is little if any bony ledge in evidence where the superior concha shouldbe,thereisaclearlydefined,tongue-likefoldofsofttissuedemon- strable onlowpowermicroscopic emtion of sections made through thecoronal plane. This tongue-like foldbegins 4 or 5 mm. belowthe top of the n vault and is i or 2 mm. in length The olfactory mucosa extends from the top of the nasal vault around the fold and down the lateralwall to alittle belowthelevel of the end of the fold. This is near theupper surfaceofwhatis designated by Geist as the superior concha, but which actually corresponds to the usual location of the middle concha,theupperportionofwhichinmanalsomarksthelowermostlimit of the olfactory mucosa. On the medial side, the olfactory mucosa ex- tends down to about the same level as on the lateral side. Most of the mucosalinesacleftwhichinthemonkeymeasuresabout i cm.indepth, I.5 cm. in length, and i mm. in width. The narrow width of the space covered bymostofthe mucosaisprobablya factor of importance in the effect produced by chemical agents, since, in addition to its inherent susceptibility to a given chemical agent, such a narrow cleftlike space would tendto retain the agent andprolongits action. Informationon themicroscopicstructure ofthe olfactorymucosa can be found in any histology textbookL The illustrations, however, are likelytobeintheformofschematicsketches,andnotphotomicrographs, aspresented inthis paper. Theliteratureprior toabout 1925 relating to the structure of the olfactory mucosa has been reviewed by Schaeffer,4 Parker5 and Hopkins.6 Little relating to its normal structure has been addedsince. Anextensivebibliographyon theanatomy, physiology, and pathology of the olfactory system in vertebrates and invertebrates has recently been published by Airkem, Inc.7 July, I96 REPAIR OF OLFACTORY MUCOSA 3 MATERIALAND METHODS AUlofthe observations weremadeonmonkeys. Onepercentsolutions of zincsulfate (C.P. or U.S.P.) in distilledwater were employed for the intranasal treatments. These were applied by irrigation of the nasal passage while the animals were under ether anesthesia and suspended in a fullyinverted position in a specially designed rack. The irrigations were carried out with a 50 cc. Luer syringe to which a short (12 cm.) rubber tubewasattached. The latter carried asmall bulb at its terminal end which was fitted snugly into the external nares. The fluid was put throughthenlpassagegentlywhiletheanimal'smouthwasheldopen and the tongue pulled forward with forceps to avoid inhalation of the solution. After one nl passage had been irrigated, the other was similarly treated. The animals were kept in the inverted position until they began to rouse from the anesthetic. Theprocedureemployedinharvestingthe tissueswas as follows: The animals were exsanguinated by cutting the neck vessels with a sharp razor, immediately after which the tissues were fixed in situ by gentle perfusion with the fixative. About 20 CC. of the fixative was first intro- duced via one carotid, and then a similar amount via the other, a 20 cc. Luer syringe with a 22 gauge needle. Of a number of fixatives initially tried, the one whichproved most satisfactory with the staining procedures employed was Bouin's solution. It was found that the per- fusion must be carried out gently enough not to separate the olfactory mucosa artificially from the lamina propria or to otherwise disrupt the tissue. Preliminary perfusion with sline solution was not found to im- prove the fixation or the staining results, and, if anything, impaired the results. Itsuse was therefore abandoned early in the work. After theperfusion, the softtissues frombothsides of the nasal vault were removed as one connected mass, usually with the olfactory nerve fila and olfactory bulbs included. To accomplish this satisfactorily re- quiredpractice.Thefirststepconsistedofcarefullybreakingawaysmall portions of the frontal and nasal bones with a bone forceps of suitable size. After the area had been adequately exposed, it was a relatively simple matter to detach the lamina propria from its bony attachments and to extract the mucosa in toto from both sides, along with the olfac- tory nerve fila and olfactory bulbs. Toaccomplish removal of the latter two as part of the interconnected mass, it was necessary to remove the bony spicules carefully within the olfactory foramen, the homologue of the cribiform plate in man. After its removal, the tissue mass was reoriented on a glass slide to bringthepartsbackasnearlyaspossibleintotheir normal relationships. 4 SCHULTZ Vol.37, No. z Fixation in Bouin's solutionwascontinued for one ormore days, follow- ingwhichthetissuewasputin35percentalcohol,andfromthere carried through graded alcohols. The tissues were finally blocked in paraffin. Sections were cut as thinlyas possible (5 p or less) through the coronal plane of the normally oriented tissue, and mounts were made from dif- ferent levels through all or most of each mucosa. This orientation made TABLE I sPIs OF NASAL MCOSA Mucosal Mucosal Mucosal Stage specimens Stage specimens Stage specimens of examined of examined of examined repair (no.) repair (no.) repair (no.) Controls* 48 I4days 3 29days 3 i dayt 13 I5days 4 30days 7 2days II i6days 3 3I-40days 10 3days II 17days 4 4I-;0days 7 4days 9 Igdays 2 5i-6odays I0 5days 6 I9days 2 6I-70days 5 6days 3 20days 4 7i--odays 7 7days 4 2Idays 7 8I-90days 4 8days 5 22days 3 3-4months 25 9days 6 24days 2 4-5months 27 iodays 9 25days 5 5-6months 4 xI cdays 4 26days 5 6-7months II I2days I 27days I 7-8months II 13days I0 28days 5 12months I *Controlanimal wereuntreated. t"Days" indicateinterval following intranasal irrigation with zinc sulfate solution. possible observations on all 4 surfaces of the mucosa, at different levels, with portions of the respiratory mucosa often included. Generally in- cluded also in the sections were olfactory nerve fila, olfactory bulbs, adjacent meninges, and a small portion of the frontal lobe of the brain. The staining procedure found most satisfactory and used almost ex- clusively was that described by Bodian8I9 for staining nerve fibers. Al- though various modifications were tried during the earlyperiod of these studies, we did not findany of these superior. An essential ingredient in the reagents employed was "Protargol," suplied at that time by the July,zI960 REPAIR OF OLFACTORY MUCOSA 5 WinthropChemicalCompany for"tissuestainingonly," aproductwhich we understand is no longer available. We have since tried protargols provided by the same company for medicinal use without obtaining successful results. With careful observation of the details of the pro- cedure, the Bodian method with the protargol above referred to yielded excellent results on Bouin's fixed material. However, inadvertent over- fixation with Bouin's solution at one time caused the loss of a large amount of valuable material. The total number of specimens of olfactory mucosa from normal and treatedmonkeys thatwere successfullyprocessed and studied cannot be stated definitely, but can be said to exceed 300 in number, which, with at least io sections per specimen, would mean at least 3,000 sections. Actually, the number of slides examined, usually with 4 or 5 sections per slide, greatly exceeded this figure, since an average of considerably more than io slides with "repeats" for better staining results, was the rule. A partial list of the total number of specimens examined and their distribution from the standpoint of the stages of repair represented is giveninTable I.Asindicated, 48 ofthe specimens came fromuntreated, normal monkeys; 68 fromanimals treatedwith zinc sulfate solutionless than io days previously; 84 from animals treated IO to 30 days pre- viously; 27 treated from3' to 6odayspreviously; i6 treated from 6i to godayspreviously, and 79treated from3 months tooneyearpreviously. The greatest attention was given to mucosa treated 3 months or more previously, afterwhichtimeanimalshadshownthemselves resusceptible to theneuronotropic virus employed. OBSERVATIONS Since the structure of the normal olfactory mucosa is not the primary theme of this report, this will be dealt with only insofar as certain facts relating to it have a bearing on this study. The epithelium is a fairly thickstructurecomposedofolfactorycells (sensoryneurons), ofcolum- nar shaped sustentacular or supporting cells, and of basal cells, usually rather squat, the whole presenting the appearance of a pseudostratified epithelium (Figs. I to 4). This structure, it is important to note, rests directly on the lamina propria, and is not separated from it by a base- mentmembrane. Inthelaminapropriaare found theglands of Bowman, ducts from which open at the surface of the mucosa, providing a serous secretion.Also foundarebundlesofaxons fromtheolfactorycellswhich, soon after emergence from the epithelium, become ensheathed by Schwann cells; these constitute unmyelinated nerve (Fig. i). Normally, the bodies of the olfactory cells lie mostly in the midzone of the epithelium, but may at times be seen either near its base or near 6 SCHULTZ Vol. 37, No. z its surface. The axons of the olfactory cells occur as fine, threadlike processes of uniform diameter which course between the cells of the mucosa in a winding or undulating manner to reach the lamina propria. Heretheybecome grouped intobundles of increasing size whichbecome ensheathed by Schwann cells soon after reaching the lamina propria. The dendritic process is usually thicker and more variable in form than the axon. It may be bayonet-like, its point surmounted by a short, deli- cate hair (olfactory cilium); it may be in the form of a thick process, solidlystainedorlacy, ineither casewithbulbous or fingerlike terminals surmountedbyoneormoreolfactoryhairs, ormayshowothervariations in form (Figs. 2 to 4). It has been postulated that differences in odor perception might be related to differences in function of the individual sensory elements. Should this be true, and differences in function be associated withdifferences instructure aswell, these could be accounted for, to some extent, by the differences in form we have observed. The ends of the dendrites come close to the surface of the mucosa, while the olfactory cilia extend slightly beyond the surface, where, as has been stated, theyarebathedbyseroussecretions from the glands of Bowman. In the absence of either a dendritic or axonal process, there is rela- tively little to enable one to differentiate olfactory cells definitely from sustentacular cells by the staining method we employed. While certain differences in the nuclei and cytoplasm, and in staining properties, may be observed, we have felt unwilling to identify cells as olfactory cells definitely unless either an axon or dendrite was seen to be associated with it, or toidentify agroup of cells as containing olfactory cells unless dendrites or axons were associated with the group. Undoubtedly, many incompletely differentiated olfactory cells were observed during the course of these studies, but we preferred to limit ourselves to criteria of which we could be certain. This report deals exclusively with the repair of the olfactory epi- theliuminmonkeys, followingitsdestructionbyzincsulfate.Thoroughly applied, a one per centsolution of this agent causes coagulation necrosis involvingthefulldepthoftheepitheliumdowntothelaminapropria,and usuallymost orallofitsentirearea. Smallareasmaybemissed attimes, butthesecan be easilyidentifiedbytheir normal structure. The respira- toryepithelium is little affected. The extent of the damage inflicted became evident on histologic ex- amination i, 2, or 3 days after application of the solution, during which timethenecrotic epitheliumwas foundintheprocess ofpeelingfrom the lamina propria. It usually came off en masse as a coherent membrane, often with many nuclei preserved by the fixative action of the solution (Figs. 5 to 8), the lamina propria being spared. Separation generally My,ZI6 REPAIR OF OLFACTORY MUCOSA 7 took place first at the arch of the nl vault, where it was most easily observed in sections cut in the coronal plane, but could be observed to take place also from other locations in suitable sections. It is important to add that its separation was usually so sharply defined and complete as to be comparable toa neatdissection. This has an important bearing in considering the repair that followed. Beng separation at the top of the nasal vault could be observed in about 24 hours and was usually completed everywhere in 3 or 4 days. However, delayed separation in some places was observed in exceptional cases as long as 8 days after treatment. As the separation proceeded from the vault downward, a single layer of new flat cells followed closely (Figs. 6 to 8). These cells increased rapidly in number so that by the third or fourth day the thickness of the new epithelium equaled or exceeded that of the normal. The cells making up this epithelium were derived largely if not entirely from the laminapropria,andwiththepossibleexceptionofthoseportionsborder- ingon the respiratory mucosa, did not arise from the latter, as reported by Smith10 in rats. In other words, the new epithelium, with minor ex- ceptions only, was not ciliated. In its early development, it resembled an atypical pseudostratified epithelium, with cells piled up in consider- able disarray (Fig. 9). The thick new epithelium was often associated with marked proliferative activity in the lamina propria as well, evi- denced by cords of cells extending from the new epithelium into the laminapropria (Fig. 9).Atleastpartofthenewcells werederived from ductcells of the glands of Bowman, somepossibly from the sheath cells of Schwann, and some possibly from other elements in the lamina pro- pria. The distinctive character of the new epithelium tended to be re- tained for some time, buttherewas also awell defined tendency for the cellstoorientthemselvesgraduallyasinthenormalolfactoryepithelium. In some areas the new epithelium might not attain more than I, 2, or 3 cellsindepthandwaseitherciliatedornoncillated. Suchareas,however, wereusuallycomparativelysmallandfewinnumber. Prior to the tenthdaytherewas relatively little to suggest that olfac- torycells wouldbe restored, althoughabnormallooking cells withproc- esses, along with what appeared to be degenerated nerve fibers, were found at times during this early period of repair. Such findings were usually interpreted as residua of damaged neurons carried upward by proliferatingcells. However, soonafter the tenth day, normal-appearing cells with dendritic processes could generally be found in the new epi- thelium Thedendriticprocessseemedtobethe firstof thetwoprocesses toformand, fromthe first,wasdistinctiveenoughtobeeasilyidentified. The incidence of cells with processes increased sufficiently so that by 8 SCHULTZ VOI. -37,No. z the 20th day they were numerous in most areas. While such cells might remain relatively few in number in some portions, their number, after severalmonths,approachedthoseofanormalepitheliuminmostregions. In certain mucosal fragments, removed 6 months to a year after zinc sulfate treatment, the general structure of the epithellum in most por- tionsoftheolfactoryareawasessentiallynormal. Incertain of these late specimens, theepitheliumappearedthickerthannormalwithaseemingly higher incidence of olfactory cells per unit area. Indeed, were it not for theextent ofsuchportions, compared with the usual extent of the initial damage, it would have been easy to underestimate the probable extent of the initial damage or degree of restoration. As stated earlier, patches of epithelium sometimes escaped the action ofthezincsulfatesolution,presumablybecause thesolutionwasshunted aroundtheareasinquestion. Theseareas, however, weregenerallyeasily identified as "missed" and were usually not a problem in observations on the early stages of repair. In the later stages, they were less easily differentiated so that the average state of a given mucosal specimen had to serve as a measure of the probable extent of the repair. It should be addedhere thateven inthelate stage ofrepair, notallofagiven mucosa was likely to showa restoration approaching the normal. In some of the later specimens there were one or more areas, usually relatively small, in which the cells were only one or two deep, with few or no olfactory cells included, orwiththelaminapropriacovered byciliated respiratory epithelium. Explanation of these exceptions was not considered within the scope of an investigation limited to the question of whether or not olfactorycells arerestored. Because space would not allow this, the observations made on the progress of the repair cannot be presented on either a case-by-case or a fixed-time-interval basis. It must therefore suffice to summarize the kindsofobservationsmadewhichwereheldtoindicatethatintherepair, sensory elements (olfactory cells) were also restored. The evidence of this was based on a combination of the following kinds of observation: (a) thecharacterandextentoftheinitial damageinduced byzincsulfate solution-this constituting a base line for observations on subsequent events; (b) the mode of separation of the necrotic epithelium from the laminapropria, seemingly leaving the surface of the latter free of resid- ualolfactory cells (Figs. 5 to 8); (c) the character of the immediate re- epithelization of the denuded lamina propria with cells predominantly derived from it rather th from the adjacent respiratory epithelium (Fig. 9); (d) theappearance ofdistinctiveindividualcellsorofcompact nests of cells, associated or not with nerve fibers, in the new epithelium (Figs. IO to 22); (e) the increased incidence of cells with dendritic JuEly,I960 REPAIR OF OLFACTORY MUCOSA 9 processes after the third or fourth week of repair; (f) the exceptional locationofindividualolfactorycellsinnewly formedepithelium, suchas cells found close to the surface of a new epithelium (Figs. IO, II, 17); (g) abnormalities in the orientation, grouping and general distribution of olfactory cells in late epithelium retaining some of the earmarks of having regenerated; (h) the extent to which an essentially normal epithelial structure was found after several months (Figs. 23 and 24); and (i) the resusceptibility of monkeys to intranasal inoculation with a neuronotropicstrainofpoliomyelitis virus 3 or4months afterintranasal treatment with zinc sulfate solution. As reported by Schultz and Geb- hardt,2 this resusceptibility was associated with anatomic evidence of axonal regeneration and with the usual lymphocytic infiltration accom- panying neuronal transmission of the virus. Certain oddoccurrences ofolfactory cells were observed in specimens obtained from late stages of repair. Among these was the appearance of gland-like spaces lined in part by olfactory cells (Figs. 25 and 26), spaces seemingly formed by invaginations of portions of the olfactory mucosa (Fig. 27). In these the overlying surface epithelium consisted of either essentially normal olfactory epithelium, or of either ciliated or nonciliated cells, without the presence of olfactory cells. A similar oc- currence was the presence of islets of compact olfactory cells below a surface epithelium, thelatter being either of normal olfactory structure (Fig. 28) ornot. Howthesemighthavearisenwasnotclear. They could have arisen from either a burial of surface cells during the rapidly pro- liferating stage of the repair, or developed from potential olfactory cells in the depth of the tissue. Stillanother unexplained observation made on membranes in the late stages of repair was the occasional occurrence of short stretches of non- ciliated epithelium made up of somewhat cuboidal cells one or two cells in depth, at times with an occasional squat olfactory cell included. It would be of interest to know why such highly abortive restorations oc- curred in the midst of more or less fully repaired olfactory epithelium. Without knowledge of the usual extent of the initial damage, one might conclude that these represented the full extent of the original damage, aconclusion thatwouldbeatvariancewiththemanyobservations made onmembranes earlyin repair. DIscussIoN The observations reported here seem to justify the conclusion that in the repair of the olfactory mucosa following zinc sulfate-induced necro- sis, sensory neurons (olfactory cells) are largely restored. The question posed in these studies was not whether complete restoration occurred, IO SCHULTZ Vol. 37, No. z but whether or not restoration occurred at all. There was a basic reason for asking this question. In the olfactory cell we have a primitive type ofneuronrepresentedwhoseregenerative capacitymightdiffer fromthat ofother neurons in vertebrates. Prior to our first reports,'2 no evidence had been recorded indicating that the loss of olfactory cells could be followed byrestoration. The fact that areas were found in membranes late in repair, in which the restoration of olfactory cells was either slight or absent, constitutes, we believe, a separate problem, one relating to the factors that restrict the extent of restoration. Another problem considered to lie outside the scope of the present paper was the precise origin of the olfactory cells. Whilethisquestionwasofinterest,itwasheldforpossiblefuturestudies. In I938, Smith,10 on the basis of observations made on rats, following intranasal application of zinc sulfate solution, reported that the de- stroyed olfactory epithelium was replaced by ciliated respiratory type of epithelium only, and that specimens obtained up to two months after treatment failed to showa replacement of olfactory cells. In I95I, how- ever, Smith11 reported observations that led him to the conclusion that a regeneration of "sensory olfactory epithelium" did occur. The latter observations weremade onadult frogs in which the olfactory epithelium hadbeen either destroyed byzinc sulfate solution or removed by opera- tion. However, it should be noted that in neither of the two above in- vestigations did he use other than the hematoxylin and eosin stains in his sections. Success in demonstrating nerve fibers of olfactory cells by simple staining methods is at variance with our own early experience, and if this were possible, would have saved us an enormous amount of technical work. On the contrary, it has been our experience that the fibersofolfactorycellsaremorefastidious, ifanything, in theirreception ofstainingmethods thanarethenerve fibersin the olfactorybulbs, with whichstaining results could be comparedinthe samesections. Undoubtedly, manyolfactory orpotential olfactorycells, still without fibers,werenotidentifiedorclassifiedassuchinthespecimensexamined, since we wished to keep on secure ground and consider only cells with definitenerveprocesses. WNheredendriteswerenumerousinagivenarea, wecouldassumethatnerve cell bodies alsowere numerous. The reverse, however, wasnotnecessarily true, since someolfactorycellbodiesmight notyethave formed fibers. Just how early in a regenerating mucosa recognizable olfactory cells appear, we cannot state definitely, but these were observed soon after the tenth day. It is also not possible to state how long the restoration may continue, but the impression gained is that it may continue over a period of several months. Although mitotic figures were observed in the