An ultrastructural and immunohistological study of the rat olfactory epithelium: Unique properties of olfactory sensory cells

~

Differentiation

Differentiation (1 985) 29 :243-253

Springer-Verlag1985

An ultrastructural and immunohistological study of the rat olfactory epithelium: Unique properties of olfactory sensory cells Michael Vollrath'*, Michael Altmannsberger', Klaus Weber and Mary Osborn

' Department of Otorhinolaryngology and

Department of Pathology, University of Goettingen, D-3400 Goettingen, Federal Republic of Germany Max Planck Institute for Biophysical Chemistry, D-3400 Goettingen, Federal Republic of Germany

Abstract. The olfactory epithelium contains three cell types : basal cells, supporting cells and sensory neurons. Electron microscopy as well as immunofluorescencemicroscopy with intermediate-filament antibodies were used to study the rat olfactory epithelium in order to obtain more information about these different cell types and to try to investigate their histogenetic origins. We found mitoses in the basal-cell layer, as well as multiple centrioles and tonofilaments in some basal cells. As revealed by electron microscopy, the supporting cells contained tonofilaments and reacted strongly with antibodies to keratin, in line with their known epithelial nature. When antibodies to other intermediatefilament types were used, i.e. glial fibrillary acidic protein, vimentin, desmin and neurofilaments, no reaction was seen in the cells of the olfactory epithelium, with the exception of occasional staining of a few axons in the subepithelial layer by neurofilament antibodies. In particular, the cell bodies, dendrites and most axons of the sensory neurons were negative for a variety of antibodies against neurofilaments. Olfactory sensory neurons therefore belong to the very few cells in adult animals which seem to lack intermediate filaments. We discuss whether this finding is related to the fact that these cells are also unique among neurons in that they are not permanent cells but constantly turn over.

microvilli of supporting cells. This structure is ciliated, and the cilia are believed to be the loci of odour perception. An axon extends from the proximal part of each olfactory sensory cell. Axons from neighbouring receptor cells combine and penetrate the basal membrane to form specialized, non-myelinated nerve bundles (fila olfactoria) which eventually connect with the brain. Olfactory axons, which have a diameter of approximately 0.2 pm, have the highest length diameter ratio of sensory cells in the body [15, 241 and hence have the slowest conducting velocity of all neurons [23]. Because of these properties, olfactory sensory cells are generally accepted as being neurons [23, 24, 32, 401. The perikarya of olfactory sensory cells are located between the supporting cells and the basal cells. In contrast to almost all other neuronal cells in the body that are formed during embryogenesis and then persist as post-mitotic cells, olfactory neurons are constantly renewed during the life of the animal [30-35, 441. Thus, olfactory neurons are of wide interest not only with respect to their function within the olfactory epithelium, but also as a model for neuronal plasticity as well as for studies involving neurogenesis in adult animals and in man. The derivation of each of the three cell types of the olfactory epithelium is still a matter of discussion. Andres

Introduction

The ultrastructure of the olfactory epithelium has been elucidated by a number of light- and electron-microscopic studies ([2, 3, 21, 22, 26, 27, 47, 53, 541; for reviews, see [28, 521). Three types of cells can be distinguished in the mammalian olfactory epithelium (Fig. 1): olfactory sensory cells, supporting cells and basal cells. The basal cells, which are located towards the basal side of the olfactory epithelium, usually extend through the lower third of the epithelium. The cell bodies of the supporting cells form the upper layer of the olfactory epithelium and are connected to the basal membrane by thin extensions of their somata and by hemidesmosomes [52]. Olfactory sensory cells have a peripheral dendritic process which extends to the surface of the epithelium, where it forms the olfactory vesicle embedded in the

*

Present address; Department of Otorhinolaryngology, University Medical School, D-3000 Hannover, Federal Republic of Germany

Fig. 1. Schematic representation of the olfactory epithelium showing the arrangement of the supporting cells (1) the olfactory sensory cells (2), the light basal cells (3) and the dark basal cells (4). m u . . microvilli; t.p., terminal plate; b m . , basal membrane; A, cells of the interfascicular glia surrounding the axonal processes

244

[2, 41 has shown that the basal cells consist of two distinct cell populations : light, undifferentiated basal cells and dark basal cells located next to the basal membrane ((2, 41; Fig. 1). He believed the light basal cells - the globosal basal cells of Graziadei and Monti-Graziadei [31] - to be primitive blastema or stem cells which could differentiate into mature sensory cells. This interpretation, which was originally based on the ultrastructural similarity between primitive neuroblasts and light basal cells, has been proved to be correct both by de- and regeneration experiments [31-351 as well as by autoradiographic investigations of undisturbed, resting olfactory epithelium [32,44,45]. The derivation of the dark, true basal cells is still a matter of controversy: whereas some authors believe that they are self renewing [2, 4, 32, 331, others [52, 541 believe that they are derived from the light basal cells, since mitoses are rarely found in the basal part of the olfactory epithelium [52]. The origin of the supporting cells is also controversial. While it was originally suggested that they might be derived from the pool of light basal cells [30, 661, the more widely held view is that the supporting cells are a separate, selfrenewing entity in the mature olfactory epithelium [ 4 6 , 31, 32, 391. Recently, intermediate-filament typing has proved to be a valuable tool in the classification of different cells in normal and pathological material, especially with respect to their histogenic origin (for a review, see [48]). Intermediate filaments can be distinguished by their diameter (7-11 nm) from both microfilaments (6 nm) and microtubules (20-25 nm). Intermediate-sized filaments (IFs) consist of five different classes which can be distinguished immunologically. Each IF type is characteristic of one or more cell types (for reviews, see [20, 36, 48, 491): 1 . Epithelia and cells of epithelial origin contain keratin filaments (17, 18, 20, 63, 641. 2. Cells of mesenchymal origin contain vimentin [9, 191. 3. Glia filaments are specific for astrocytes and Bergmann glia [lo, 11, 13, 50, 611. 4. Desmin is typical of most muscle cells [25, 621. 5. Most, but not all, neurons contain neurofilaments [42, 51, 58, 60, 61, 671. In the course of analysing chemically induced tumours of the olfactory epithelium in rats, with special emphasis on their histogenetic origin (our, unpublished results), investigation of normal olfactory epithelium became desirable. Although the ultrastructure of normal olfactory epithelium is already well established, some new details were demonstrated inside the basal-cell layer. This report also presents, for the first time, a detailed immunohistological study of the olfactory epithelium using antibodies specific for the different intermediate-filament types. Methods Electron microscopy

For the ultrastructural investigation of the olfactory epithelium, adult white Wistar rats (strain Bar WISW SPF; Winkelman Versuchstierzucht, Borchen, FRG) with an average weight of 280 g were used. The olfactory epithelium is easily damaged during fixation and was therefore fixed in situ by intracardial perfusion [2, 521. After an overdose of sodium pentobarbitdl (Nembutal), the thoracic cavity was opened, the descending aorta was ligated, and a perfusion

syringe was inserted into the left ventricle. After wide dissection of the right atrium, a high-speed perfusion of the brain and upper portion of the body was performed using solution I [40 mg Onkovertin (low-molecular-weight dextran), 9.0 g NaCl, 0.3 g KCl, 0.2 g CaClz and 0.05 g MgClz dissolved in 1,000 mlO.1 A4 phosphate buffer, pH 7.2) warmed to 37" C until the reflux was clear (isometric pressure conditions for approximately 2 min). This was followed by the application of the fixative (6 ml 25% glutaraldehyde, 6 ml 25% paraformaldehyde, 45 ml 0.2 M cacodylate buffer and 100 ml distilled water, pH 7.35). For the first 60 s the fixative was used at 37°C and, for the rest of the perfusion time (20 min), it was cooled down to 3"-6" C. The perfusion was completed by the administration of solution 2 (6.2 g saccharose in 100 ml 0.1 M phosphate buffer, pH 7 . 4 7 . 8 ) for 10 rnin at 3"-6" C. Correctly performed perfusions showed a fibrillation of muscles immediately after the start of fixation, followed by a general muscle stiffness involving only the upper part of the body. After perfusion, the nose was opened, and the olfactory region (ethmoturbinalia IIIV) was removed and immersed in fixative (2% Os04 and 5% saccharose in 0.1 M phosphate buffer, pH 7 . 4 7 . 8 ) for 1 h at 4" C. The olfactory epithelium was prepared using a binocular microscope. Tissue samples were dehydrated in an alcohol series and treated with saturated uranyl acetate in 70% alcohol for 12 h. After complete dehydration, the tissue was put through two changes of propylene oxide (5 rnin each) and then embedded in Araldite. Semi-thin sections were cut on an ultramicrotome (Ultracut OmU 3; Reichert and Jung) using glass knives and stained with toluidine blue. Areas of interest were marked. Ultrathin sections (thickness, 50-100 pm) were cut, mounted on copper grids and stained with lead citrate. lmmunofluorescence microscopy

Immunohistological investigations, although possible with alcohol-fixed materail [l], yielded the most reliable results when performed on frozen tissue samples. After an overdose of Nembutal, the nose was opened, and the entire olfactory region (ethmoturbinalia 11-IV) was removed, cut along the axis of the nasal septum into two equal parts and fixed on wooden blocks (2 x 1 x 1 cm)with the cut side next to the block by covering it with Tissue Tek I1 (Miles, Naperville, Ill). The whole block was frozen in isopentane cooled by liquid nitrogen. The frozen material was wrapped with Parafilm to prevent desiccation and was stored at - 70" C. Tissue was cut on a cryostat at approximately -20" C dried on the slide at room temperature, extracted for 1 rnin with Triton X-100 in phosphate-buffered saline (PBS) and then fixed in acetone at - 10"C. After the application of the first antibody for 3 W 5 min and a wash in PBS, the slide was incubated with an appropriate fluorescein-isothiocyanate (F1TC)-coupled second antibody at 37" C for 30 min. After washing in PBS, the slides were mounted in Mowiol4-88. An tibodies Keratin antibodies. Both conventional and monoclonal antibodies against keratin were used. Keratin was isolated from cow snout and injected into guinea pigs. The antibody was affinity purified on the same antigen bound to Sepharose 4B [17, 491. The monoclonal keratin antibody lu-5, which

245

has a broad specificity ([65]; our unpublished results), recognizes squamous and simple epithelia, and was a kind gift from Dr. T. Staehelin. Vitamin antibody. Vimentin was isolated from rabbit chondrocytes and injected into sheep. The antibody was affinity purified on vimentin from rabbit chondrocytes bound to Sepharose 4B [49]. Desmin antibody. This was raised in rabbits against desmin purified from pig stomach [49]. Glialfibrillary acidic protein (GFAP) antibody. GFAP antibody was raised in rabbits against GFAP isolated from pig spinal cord [57].The monoclonal GFAP antibody has been described elsewhere [14]. Neurofilament antibodies. Both polyclonal and monoclonal antibodies specific for neurofilaments were available. These included rabbit polyclonal antibodies specific for the 68%lodalton (kd) polypeptide (NF-L), the 160-kd polypeptide NF-M and the 200-kd polypeptide (NF-H) as well as antibodies containing determinants against all three neurofilament polypeptides (for characterization of these antibodies,

see [56, 58, 60, 611; also included were mouse monoclonal antibodies NR4 directed against NF-L and antibody NE14 directed against NF-H (for characterization see [14]). Appropriate FITC- or rhodamine- labelled second antibodies were purchased from Cappell Laboratories (Cochranville, PA) or from Miles-Yeda (Israel). Second antibodies were applied at a concentration of 0.2-0.5 mg/ml. Results Electron-microscopic findings

Inside the olfactory epithelium, there is a stratification of nuclei belonging to the different cell types (Fig. 1). The nuclei of the supporting cells are closest to the lumen, and then come the nuclei of the mature sensory neurons. The lower part of the epithelium is occupied by undifferentiated light basal cells, with the dark basal cells located immediately on and above the basal membrane. Figure 2a shows the two types of basal cells: the undifferentiated light basal cells and the dark basal cells are seen to be separated by empty spaces of different sizes. This loosening of cellular contact within the dark basal cells, resulting in the enlargement of the intracellular spaces, is such a frequent finding

Fig. 2a. Electron micrograph of light basal cells (left) and dark basal cells (right) separated by the empty spaces referred to by some authors (e.g., 1511) as the basal labyrinth (15).Note the tonofilaments (arrows). b Light basal cell. Note the large nucleus and scanty cytoplasm; inside the cytoplasm, abundant polyribosomes are visible. The tonofilaments on the lower /eft (arrow) belong to the process of a supporting cell. a x 4,100; b x 16,200

246

241

that Seifert (521 wondered whether it reflected the normal in situ situation and referred to it as the bas1 labyrinth [52]. The undifferentiated, primitive ' neuroblastic' nature of light basal cells is evident in Fig. 2b: a large nucleus containing finely dispersed chromatin is surrounded by a small rim of cytoplasm, in which abundant polyribosomes, but few special cell organelles, are visible. A single mitochondrion is present in the lower part of the cytoplasm. The tonofilaments visible in the left comer of the Fig. 2b belong to a basal process of a supporting cell inserting on the basal membrane. Figure 3a shows the basal part of the olfactory epithelium that includes the basal membrane and the most caudal portions of the dark basal cells, two of which are undergoing mitosis. Note especially the tonofilaments in the enlargement of one of these cells shown in Fig. 3b, and the accumulations of centrioles in some mitotic cells (Fig. 3ce). These centrioles are similar to those seen in the olfactory vesicle, in which they form the basal bodies (blepharoplasts) of typical olfactory cilia (Fig. 3f). The striking similarity between the centrioles inside basal cells and the basal bodies of olfactory cilia is especially well demonstrated in the perifocal haze surrounding the centrioles (Fig. 3d) and in equivalents of rootlet fibres and rudimentary outgrowing cilia (Fig. 3e). The possible implications of these findings with respect to embryohistogenesis are discussed below. Immunohistolog ical findings

Immunohistochemical investigations of frozen sections of the olfactory epithelium yielded a strong immunofluorescence with antibodies specific for keratin (Figs. 4a, c and 5). This staining was strongest just above the basal membrane and in the upper third of the epithelium. In the lower two-thirds of the epithelium, the reaction was thin and thread like (e.g. Fig. 5b, c, e). The supporting cells were strongly stained. Thus, the oval nuclei of the supporting cells present in the upper third of the epithelium were surrounded by a filamentous basket positive for keratin (Figs. 4a, 5c); strong staining of tonofilaments of the terminal plate at the upper surface as well as in the basal processes connecting to hemidesmosomes at the basal membrane was also observed (Fig. 4a). Light basal cells lying above the basal membrane as well as sensory cells did not appear to be stained by the keratin antibody (cf. Fig. 5c, d). Increased staining with the keratin antibody was seen along the basal membrane in a position corresponding to that occupied by the dark basal cells; however, because of the proximity of the supporting cells and dark basal cells, we cannot say whether all of the dark basal cells were positive with the keratin antibody, although our immunofluorescence pictures are consistent with such an interpretation. Similar staining patterns were obtained with the keratin guinea-pig antibody (Fig. 4a, c) and with the keratin monoclonal antibody lu-5 (Fig. 5). The latter is a broadspecificity keratin antibody which reacts with a wide variety of epithelia, including rat liver (Fig. 5g). Staining of the olfactory epithelium with a monoclonal antibody which re-

Table 1. Staining properties of the olfactory epithelium with different intermediate-filamentantibodies Antibody against

Olfactory epithelium Supporting Sensory Light Dark cell cell basal cell basal cell

Keratin Vimentin Neurofilaments GFAP Desmin a

+ -

-

-

-

-a -

-

+a

-

Because of the proximity of supporting cells and basal cells, we cannot be sure whether all dark basal cells were positive and whether all light basal cells were negative with the keratin antibody. In the subepithelial layer, vimentin stained connectivetissue elements and blood vessels, while desmin stained some blood vessels. Positive staining of a few axons was detected in the subepithelial layer with different neurofilament antibodies, while GFAP stained interfascicularglia

cognizes desmosomes [48a] showed desmosomes along the processes of the supporting cells (Fig. 4 0 . The most interesting immunohistological finding was that none of the cells inside the olfactory epithelium showed a positive immunofluorescence with a variety of polyclonal and monoclonal antibodies specific for neurofilaments. In each case, the position of the epithelium was identified either by observing the section by phase microscopy (e.g. Fig. 4e, f) or by using a double label with the keratin antibody (e.g. Fig. 4c, d). Thus, antibodies directed against all three neurofilament triplet proteins, as well as antibodies specific for NF-L, NF-M or NF-H, stained neither the sensory neurons nor other cell types. In addition, only a few of the axons present in the subepithelial layer showed some weak staining with neurofilament antibodies. This point is most clearly illustrated in Fig. 4c and d, which show the same section stained with antibodies to keratin (Fig. 4c) and to neurofilaments (Fig. 4d). We are unable to say from sections such as that shown in Fig. 4d whether the positive staining with neurofilament antibodies is of the axons of the fila olfactoria or of other nerves present in this layer, such as the nervus terminals or branches of the trigeminal nerve (for corresponding ultrastructural studies which only occasionally showed IFs in such axons, see [3, 5, 411). The neurofilament antibodies used in the present study stain neuronal cell types for instance, in rat brain and rat optic nerve, very strongly (data not shown, but see [14, 561). When antibodies against GFAP were used, a positive reaction of interfascicular glia was seen (Fig. 4g, h). When antibodies specific for the remaining intermediate-filament groups were used, no staining of the olfactory epithelium was seen. Thus, desmin antibodies only stained some blood vessels in the connective tissue (Fig. 49, while vimentin antibodies stained vascular smooth-muscle cells and cells in the connective tissue (Fig. 4j). The immunohistological results are summarized in Table 1.

Fig. 321-f. Electron micrographs of a basal-cell layer just above the basal membrane (bm).Inside two mitotic cells (arrowheads) and one tonofilament-containing cell (arrow), there are accumulations of centrioles. b-d Details of tonofilaments (b) and centrioles (M) inside the cells shown in a. Note the perifocal haze in d (arrow). e Centrioles inside basal cells, resembling basal bodies. Note rootlet equivalents (urrows) and the initiation of a cilium (arrowhead). f Olfactory vesicle with basal bodies and cilia. Compare these mature basal bodies with the centrioles in b-d. a x 5,000; b x 19,000; c-f x 30,000

248

Fig. 4a-j. Immunocytochemistry of the olfactory epithelium using antibodies against intermediate filaments and immunofluorescence microscopy. a, b Staining of the olfactory epithelium with guinea-pig antibodies to keratin (a) and with rabbit antibodies to the neurofilament triplet (b), c, d Staining of the same section using double-label fluorescence with guinea-pig antibodies to keratin (c) and with the monoclonal neurofilament antibody NR4 against NF-L (a). e, f Phase (e) and fluorescence (f) micrographs of a section stained with the monoclonal neurofilament antibody NR4. g, h Phase (g) and fluorescence (h) micrographs of a section stained with the monoclonal GFAP antibody G-A-5. i Rabbit antibody against desmin, j Sheep antibody against vimentin. The position of the olfactory epithelium when present is indicated by an avterirk according to the corresponding phase micrograph, and the limits of the olfactory epithelia are marked by black dotted lines on phase micrographs (i.e. e, g) or white dotted lines on fluorescence micrographs (b, i, j). L, lumen. Note the strong staining of the supporting cells and some basal cells by keratin antibodies in the olfactory epithelium cut either longitudinally (a, c, arrow) or transversely (c, arrowhead). Note that none of the cell types in the olfactory epithelium is stained with either rabbit neurofilament antibodies (b) or monoclonal antibodies to neurofilaments (d, f). but that positive staining of a few axons is visible in the subepithelial layer (b, d, f). Antibodies to GFAP (h) stain interfascicular glia cells in the subepithelial layer. Antibodies to desmin do not stain the olfactory epithelium but stain some vascular smooth-muscle cells in the subepithelial layer (i).Antibodies to vimentin do not stain the olfactory epithelium but stain vascular smooth-muscle cells and conncctive-tissue cells in the subepithelial layer (i).a, b x 700;c, d x 200; e-j x 170

249

Discussion The olfactory epithelium is a primitive sensory organ, the histology of which is remarkably constant throughout evolution. Thus, at the light-microscopic level, the olfactory epithelia of fish, amphibia, reptilia, birds and mammals show the same stratified texture of supporting cells, sensory cells and basal cells [40,46]. The stimulus-transforming cells within this epithelium are primary sensory cells similar to those found in the periphery of invertebrates [12]. Kolmer, who gave an extensive description of the olfactory epithelium at the light-microscopic level, understood the bipolar sensory cell, with its peripheral dendritic process to the surface of the epithelium and its central process (axon, film olfactorium) to the olfactory bulb, to be the “sensory gan-

glion cell analogue to the spinal ganglion cells” [40]. Only recently, Kauer [37, 381 has demonstrated the cellular continuity of the entire sensory cell - from the olfactory vesicle to the olfactory bulb - by the retrograde transport of horseradish peroxidase. Ultrastructural investigations [4, 5, 22, 27, 29, 32, 331 have confirmed Kolmer’s interpretation of olfactory sensory cells being true neurons (for comprehensive ultrastructural descriptions of the entire olfactory mucosa, see [28, 31, 521). Thus, it is striking that olfactory sensory cells are - in contrast to all other neurons - not terminal cells, but are constantly renewed out of the pool of light basal cells (for references, see Introduction). This renewal takes place continuously, even in undistorted epithelium, so that the turnover rate of normal olfactory sensory cells is about 30 days [32, 44, 451. Furthermore, the em-

250

251

bryological development of the olfactory epithelium from the olfactory placode - a specialized ectodermal formation distinct from the neural plate - emphasizes its peculiarity in comparison to regular neurons derived from the neural epithelium of the neural tube or the neural crest. Whereas in normal neurons after axon injury, there is Wallerian degeneration of the distal part and axonal regeneration beginning from the proximal stump, in olfactory sensory cells, the entire cell including both processes perishes and is replaced by cells differentiating out of the pool of basal blastema cells [33]. Sensory cells

Our most interesting immunohistological finding was that sensory cells were negative when tested with a variety of polyclonal and monoclonal antibodies specific for neurofilaments. This negative reaction was seen not only in blastema cells but also in the perikarya, dendrites and axons of mature olfactory sensory cells (although note that some axons in the subepithelial layer were positive, e.g. Fig. 4b). This negative reaction with neurofdament antibodies underlines the exceptional nature of olfactory sensory cells. Although, if these cells contained only very few neurofilaments, it might be difficult to detect them by fluorescence, the antibodies used here reveal neurofilaments in other rat tissues, including the optic nerve and brain. The further argument that putative neurofilaments might be modified in a special manner and therefore not be identified by immunofluorescence is unlikely because of the number of different neurofilament antibodies used. As shown in the figures and in Table 1, sensory cells were likewise negative when tested with antibodies against desmin, vimentin and GFAP. A comparison of phase and fluorescence micrographs of olfactory epithelium stained with broad-specificity keratin antibodies (e.g. Fig. 5c, d) suggests that the olfactory neurons were not stained. Thus, olfactory neurons appear to lack intermediate filaments. In accordance with our immunohistological findings, our ultrastructural studies did not show 10-nm filaments inside the olfactory epithelium, which is in good agreement with other electron-microscopic investigations of the olfactory epithelium [52, 54, 661. It could, however, be argued that the cytoplasm is too dense to detect 10-nm filaments, and further studies of this epithelium, e.g. after detergent extraction, would add to the strength of this argument. Olfactory sensory cells thus seem to belong to the few neuronal types which show only a faint reaction or no reaction at all with antibodies to neurofilaments [60].This group includes certain dorsal-root ganglion cells that also show no reaction with antibodies to neurofilaments [60], a finding which is in accordance with parallel ultrastructural investigations, since in both electron microscopic [16] and immunoelectron-microsopic investigations [56], the small

dark cells of spinal ganglia have been shown to lack neurofilaments. Interestingly, the analogous cells inside the retina, i.e. bipolar cells, also appear negative when tested with antibodies against intermediate filaments and, in particular, with those against neurofilaments [59]. Rods and cones of the retina as well as one type of horizontal cell in certain species are also reported to be negative for IFs (for references, see [SS]). Too little is currently known to judge whether the apparent lack of IFs in these different neuronal types can be correlated with a functional advantage for these particular cell types. The parallel with other cells of neuronal origin which lack neurofilaments suggests that olfactory sensory cells may be of a rather 'primitive' neuronal type. The data from both our ultrastructural and fluorescence studies of olfactory sensory cells are consistent with their proven derivation from light basal cells, since these cells also appear to lack intermediate filaments. Supporting cells

The supporting cells exhibited features, both ultrastructurally and by immunofluorescence, which justify their designation as epithelial cells. Thus, electron micrographs revealed tonofilaments in supporting cells, and a strong reaction with antibodies to keratin both around their nuclei and in their extensions was seen using immunofluorescence microscopy. Whether the supporting cells are likewise derived from basal cells as previously suggested [30, 661, or whether they represent a self-regenerating entity [4-6, 321 cannot be decided from the present results. Since at least some tonofilament- and hence keratin-containing dark basal cells were observed, and since these are most probably derived from light basal cells, a similar origin for supporting cells cannot be excluded. Basal cells

Our ultrastructural investigation of the most caudal layer of dark basal cells yielded some hitherto unknown findings that may be relevant to the cytogenetic derivation of these cells. In the cytoplasm of a very small number of basal cells, accumulations of centrioles reminiscent of the basal bodies (blepharoplasts) of olfactory cilia were found. These exhibited rootlet equivalents and a typical perifocal haze. Their ultrastructure alone does not provide sumcient evidence to decide whether these accumulations of centrioles are always rudimentary blepharoplasts or whether they resemble the multiple microtubule-organizing centres (MTOCs) demonstrated in neuroblastoma cells in culture [43, 551. Thus, these cells may be derived from light basal cells which have originally initiated these elements typical of olfactory sensory cells, and have then moved position to the basal membrane and become dark basal cells. In dark basal cells, tonofilaments (Fig. 3 b) could be demon-

Fig. 5a-g. lmmunocytochemistry of the olfactory epithelium with the monoclonal antibody lu-5 (a-c, e) or with a monoclonal antibody against desmosomes (0.Positive staining of rat liver by lu-5 is shown in g. The sections of the olfactory epithelium are arranged so that the lumen is at the top. In each instance, a section through the whole epithelium is shown, and the approximate positions of supporting cells ( I ) , olfactory neurons (2), light basal cells (3) and dark basal cells (4) are indicated (cf. Fig. 1). Note the strong staining of the basal cells above the basement membrane (a-c, e), and the strong staining of the supporting-cell nuclei (c) and of the processes of the supporting cells (b, c, e). d Phase micrographs of the section shown in e; a comparison of c and d suggests that the olfactory neurons and light basal cells are not stained by 111-5.Note in f that desmosomal staining is seen along the processes of the supporting cells, and in g, the strong staining of hepatocytes by lu-5 is visible. a-c, f, g Immunofluorescence; e peroxidase with counterstain to visualize the nuclei. a, b, f, g x 400;ce x 600

252

strated both by electron microscopy and by immunofluorescence microscopy using antibodies against keratin (Fig. 4a). Thus, perhaps a light basal cell can differentiate either into a sensory cell or into an epithelial cell. Like Seifert [52], who also suggested this mode of differentiation, we frequently found mitoses in the most basal cell layer. In some of these mitotic cells, we were also able to demonstrate similar accumulations of centrioles which were not incorporated into the mitotic apparatus (e.g. Fig. 3c-e). We interpret our data as indicating that light basal cells are capable of differentiating in two directions: towards true sensory cells (a route which is widely accepted) and likewise towards dark basal cells. The presence of multiple centrioles inside dark basal cells may indicate that the second route is possible even after the blastema cell has started to differentiate to form a sensory cell. Although findings on chemically induced tumours of the olfactory epithelium may be relevant to this question (our, unpublished results), final proof of whether the pool of light basal cells contains one or more types of blastema cell that have the capability of differentiating into different cell types (i.e. sensory cells, basal cells, and possibly even supporting cells [66]), or whether this pool contains different cell populations of different determination but similar ultrastructural appearance, will require cell-kinetic studies in which the temporal sequence of the different developmental steps of individual light basal cells can be determined. Our immunohistological investigation of the olfactory epithelium showed that in the subepithelial layer, cells surrounding the olfactory nerves (fila olfactoria) reacted positively when tested with antibodies against GFAP. Electronmicroscopic studies of interfascicular glia [5] support the idea that these are “glial cells, most probably astrocytes”, and such studies have distinguished them from Schwann cells [5, 411. Our results support such an interpretation, since these cells contain GFAP, the intermediate-filament type characteristic of astrocytes and Bergmann glia [JO, 11, 50, 611. Similar findings have recently been presented by others [7, 81. Since double labelling with GFAP and vimentin antibodies was not performed, we cannot, at present, exclude the possibility that these cells may also contain vimentin. Acknowledgements. We thank Susanne Isenberg and Elke Schmalstieg for skilled technical assistance, and Dr. Gerry Shaw for providing the rabbit neurofilament antibodies used in this study. This work was supported in part by grants from the Bultner Stiftung to M.V.and from the Deutschc Krcbshilfe to M.O.

References 1. Altmannsbergcr M, Osborn M, Schauer A, Weber K (1981)

Antibodies to different intermediate filament proteins. Cell typc-specific markers on paraffin-embedded human tissues. Lab Invest 45:427-434 2. Andres KH (1965) Zur Methodik der Perfusionsfixierung des Zentralnervensystems von SPugern. Gemeinsame Tagung der Nicdcrlandischen und Deutschen Gesellschaft fur Elektronenmikroskopie. Aachen. Cited in Seiferl K [52] 3. Andres KH (1965) Differenzierung und Regeneration von Sinneszellcn in der Regio olfactoria. Naturwissenschaften 52 :500 4. Andrcs K H (1966) Der Fcinbau der Regio olfactoria von Makrosrnatikcrn. Z Zellforsch 69: 14&154 5. Andrcs KH (1969) Der olfaktorische Saum der Katze. Z Zellforsch 96:25&274

6, Bannister LH, Dodson HC (1975) Proliferative units in the olfactory epithelium of the mouse. J Anat 119:471498 7. Barber PC (1982) Regeneration of olfactory sensory axons into transplanted segments of peripheral nerve. Neuroscience 7:2677-2685 8. Barber PC, Lindsay RM (1982) Schwann cells of the olfiactory nerves contain glial fibrillary acidic protein and resemble astrocytes. Neuroscience 7: 3077-3090 9. Bennett GS, Fellini SA, Croop JM, Otto JJ, Bryan J, Holtzer H (1978) Differences among 100 8, filament subunits from different cell types. Proc Natl Acad Sci USA 75:43644368 10. Bignami A, Eng LF, Dahl D, Uyeda CT (1972) Localization of the glial fibrillary acidic protein in astrocytes by immunofluorescence. Brain Res 43:42%435 11. Bignarni A, Dahl D, Rueger DC (1980) Glial fibrillary acidic protein (GFA) in normal neural cell and in pathological conditions. In: Fedoroff F, Hertz L (eds) Advances in cellular neurobiology, vol 1. Academic Press, New York, pp 285-310 12. Bullock TH, Horridge GA (1965) Structure and function in the nervous system of invertebrates, vol 1. W.H. Freeman, San Francisco London 13. Dahl D. Bignami A (1973) Glial fibrillary acidic protein from normal human brain. Purification and properties. Brain Res 57 :343-360 14. Debus E, Weber K, Osborn M (1983) Monoclonal antibodies specific for glial fibrillary acidic protein and for each of the neurofilament triplet proteins. Differentiation 25: 193-203 25. de Lorenzo AJ (1957) Electron microscopic observations of the olfactory mucosa and olfactory nerve. J Biophys Biochem Cytol3 :839-867 16. Duce JR, Keen P (1977) An ultrastructural classification of the neuronal cell bodies of the rat dorsal root ganglion using zinc iodide-osmium impregnation. Cell Tissue Res 185:263277 17. Franke WW, Weber K, Osborn M, Schmid E. Freudenstcin C (1978) Antibody to prekeratin. Decoration of tonofilamentlike arrays in various cclls of epithelial character. Exp Cell Res 116:429-445 18. Franke WW, Applehans B, Schmid E, Freudenstein C, Osborn M, Weber K (1979) ldentification and characterization of epithelial cells in mammalian tissues by immunofluorescence microscopy using antibodies to prekeratin. Differentiation 15:7-25 19. Franke WW, Schmid E. Winter S, Osborn M. Weber K (1979) Widespread occurrence of intermediate-sized filaments of the vimentin type in cultured cells from diverse vertebrates. Exp Cell Res 123 :2 5 4 6 20. Franke WW, Schmid E, Schiller DL, Winter S, Jarasch ED, Moll R, Denk H, Jackson BW, Illmensee K (1982) Differentiation-related patterns of expression of proteins of intermediatesized filaments in tissues and cultured cells. Cold Spring Harbor Symp Quant Biol46:431-453 21. Frisch D (1964) Ultrastructural observations of the mousc nasal and olfactory mucosa. Anat Rec 148 :283 22. Frisch D (1967) Ultrastructure of mouse olhctory mucosa. Am J Anat 121:87-119 23. Gasser HS (1958) Comparison of the structure as revealed with the electron microscopc and the physiology of the unmedullatcd fibers in the skin nerves and the olfactory ncrvcs. Cxp Ccll Res (suppl) 5:3-17 24. Gasser HS, Palade GE (1956) Olfactory nerve fibers. J Gcn Physiol39 :473498 25. Granger BL, Lazarides E (1978) The existence of an insoluble Z-disc scaffold in chicken skeletal muscle. Cell 15: 12531268 26. Graziadei PPC (1965) Electron microscopic observations on the olfactory mucosa of the cat. Experientia 21 :274-275 27. Graziadei PPC (1966) Electron microscopic observations of the olfactory mucosa in vertebrates. In: Uyeda R (ed) Electron microscopy, vol 2. Maruzen. Tokyo, pp 487488 28. Graziadei PPC (1971) The olfactory mucosa of vertebrates. In: Beidler LM (ed) Handbook of sensory physiology, vol4. Chem-

253 ical senses. I. Olfaction. Springer, Berlin Heidelberg New York, pp 27-58 29. Graziadei PPC (1971) Topological relations between olfactory neurons. Z Zellforsch 1 18 :449-466 30. Graziadei PPC, Metcalf J F (1971) Autoradiographic and ultrastructural observations in the frog’s olfactory mucosa. Z a l l forsch 116:30%318 31. Graziadei PPC, Monti-Graziadei GA (1978) Continuous nerve cell renewal in the olfactory system. In: Jacobson M (ed) Handbook of sensory physiology, vol 9. Springer, Berlin Heidelbeg New York, pp 5 5 8 3 32. Graziadei PPC, Monti-Graziadei GA (1979) Neurogenesis and neuron regeneration in the olfactory system of mammals. 1. Morphological aspects of differentiation and struct t i l a l organization of the olfactory sensory organ. J Neurocytol 8: 1-18 33. Graziadei PPC, Monti-Grdziddei GA (1979) Neurogenesis and neuron regeneration in the olfactory system of mammals. 11. Degeneration and reconstitution of the olfactory sensory neurons after axotomy. J Neurocytol8: 197-213 34. Graziadei PPC, Okano M (1979) Neuronal degeneration and regeneration in the olfactory epithelium of pigeons following transection of the first crank1 nerve. Actd Anat (Basel) 104:22&236 35. Graziadei PPC, Karlan MS, Monti-Graziadei GA, Bernstein JJ (1980) Neurogenesis of sensory neurons in the primate olfactory system after section of the fila olfactoria. Brain Res 186 :289-300 36. Holtzer H, Bennett GS, Tapscott SJ, Croop JM, Toyama Y (1982) Intermediate-sized filaments: Changes in synthesis and distribution in cells of myogenic and non-myogenic lineages. Cold Spring Harbor Symp Quant Biol46:317-329 37. Kauer JS (1981) Olfactory receptor cell staining using horseradish peroxidase. Anat Rec 200: 331-336 38. Kauer JS (1983) Surface morphology of olfactory receptors. J Submicrosc Cytol 15: 167-171 39. Klein SL, Graziadei PPC (1983) The differentiation of the olfactory placode in Xenopus lueuis. J Comp Neurol217:17-30 40. Kolmer W (1927) Das Geruchsorgan. In: von Miillendorff W (ed) Handbuch der mikroskopischen Anatomie des Menschen, vol3. Springer, Berlin Heidelberg New York, pp 192-294 41. Kreutzberg GW, Gross G W (1977) General morphology and axonal ultrastructure of the olfactory nerve of the pike, Esox lucis. Cell Tissue Res 181 :443457 42. Lasek RJ, Hoffmann PN (1976) The neuronal cytoskeleton, axonal transport and axonal growth. In: Goldman R, Pollard T, Rosenbaum J (eds) Cell motility. Cold Spring Harbor, New York, pp 1021-1050 43. Marchisio PC, Weber K, Osborn M (1979) Identification of multiple microtubule initiating sites in mouse neuroblastoma cells. Eur J Cell Biol20:45-50 44. Moulton DG (1974) Dynamics of cell population in the olfactory epithelium. Ann NY Acad Sci 237:52-61 45. Moulton DG, Fink RP (1972) Cell proliferation and migration in the olfactory epithelium. In: Schneider D (ed) Olfaction and taste, vol4. Wissenschaftliche Verlagsgesellschaft, Stuttgart, pp 20-26 46. Negus SV (1958) The comparative anatomy and physiology of the nose and the paranasal sinuses. Livingstone, Edinburgh, p 34-48 47. Okano M, Weber AF, Frommes SP (1967) Electron microscopic studies of the distal border of the canine olfactory epithelium. J Ultrastruct Res 17:487-502 48. Osborn M, Weber K (1983) Tumor diagnosis by intermediate filament typing: a novel tool for surgical pathology. Lab Invest 48~372-394 48a. Osborn M, Weber K (1985) A monoclonal antibody recognizing desmosomes: use in human pathology. J Invest Dermatol (in press) 49.Osborn M, Geisler N, Shaw G, Sharp G, Weber K (1982)

Intermediate filaments. Cold Spring Harbor Symp Quant Biol 46 :41 3-429 50. Schachner M, Hedley-Whyte ET, Hsu DW, Schoonmaker G, Bignami A (1977) Ultrastructural localization of glial fibrillary acidic proteins in mouse cerebellum by immunoperoxidase labeling. J Cell Biol75: 63-73 51. Schlaepfer WW (1977) Immunological and ultrastructural studies of neurofilaments isolated from rat peripheral nerve. J Cell Biol74 :226-240 52. Seifert K (1970) Die Ultrastruktur des Riechepithels beim Makrosmatiker. Eine elektronenmikroskopische Untersuchung. In : Bargmann W, Doerr W (eds) Normale und pathologische Anatomie, vol21. Georg Thieme, Stutgart, pp 1-99 53. Seifert K , Ule G (1965) Ober elektronenmikroskopische Untersuchungen an der Riechschleimhaut. Arch Ohren-Nasen-Kehlkopfheilkd 185: 767-771 54. Seifert K, Ule G (1967) Die Ultrastruktur der Riechschleimhaut der neugeborenen und jugendlichen weikn Maus. Z Zellforsch 76:147-169 55. Sharp GA, Osborn M, Weber K (1981) Ultrastructure of multiple microtubule initiation sites in mouse neuroblastoma cells. J Cell Sci 47: 1-24 56. Sharp GA, Shaw G, Weber K (1982) Immunoelectronmicroscopical localization of the three neurofilament triplet proteins along neurofilaments of cultured dorsal root ganglion neurons. Exp Cell Res 137 :403-41 3 57. Sharp GA, Osborn M, Weber K (1982) Occurrence of two different intermediate filament proteins in the same filament in situ within a human glioma cell line: An immunoelectron microscpical study. Exp Cell Res 141 :385-395 58. Shaw G, Weber K (1981) The distribution of the neurofilament triplet proteins within individual neurones. Exp Cell Res 136: 119-125 59. Shaw G, Weber K (1983) The structure and development of the rat retina: An immunofluorescence microscopical study using antibodies specific for intermediate filament proteins. Eur J Cell Biol30:219-232 60. Shaw G, Osborn M, Weber K (1981) Arrangement of neurofilaments, microtubules and microfilament-associated proteins in cultured dorsal root ganglion cells. Eur J Cell Biol24:20-27 61. Shaw G, Osborn M, Weber K (1981) An immunofluorescence microscopical study of the neurofilament triplet proteins, vimentin and glial fibrillary acidic proteins within the adult rat brain. Eur J Cell Biol 26:68-82 62. Small JV, Sobieszek A (1977) Studies on the function and composition of the 10 nm-filaments of vertebrate smooth muscle. J Cell Sci 23 :243-268 63. Sun TS, Green H (1978) Keratin filaments of cultured human epidermal cells. Formation of intermolecular disulfide bonds during terminal differentiation. J Biol Chem 253 :2053-2060 64. Sun TT, Shih C, Green H (1979) Keratin cytoskeleton in epithelial cells of internal organs. Proc Natl Acad Sci USA 76: 2813-2817 65. Von Overbeck J, Stlhli C, Gudat F, Carmann H, Lautenschlager C, Dorrmiiller U, Tackacs B, Miggiano V, Staehelin T, Heitz P (1985) Immunohistochemical characterization of an antiepithelial monoclonal antibody. Virchows Arch (in press) 66. Yamamoto M (1976) An electron microscopic study of the olfactory mucosa in the bat and rabbit. Arch Histol Jpn 38:359-412 67. Yen SH, Fields KL (1981) Antibodies to neurofilament, glial filament and to fibroblast intermediate filament proteins bind to different cell types of the nervous system. J Cell Biol 88: 115-126

Received February 1985 / Accepted in revised form June 1985