Cytoplasmic desmosomes and intermediate filament disturbance following acrylamide treatment in cultured rat keratinocytes

TISSUE AND CELL, 1994 26 (1) 43-55 @ 1994 Longman Group UK Ltd.

A. H. M. SHABANA,

M. OBOEUF, and N. FOREST

CYTOPLASMIC DESMOSOMES AND INTERMEDIATE FILAMENT DISTURBANCE FOLLOWING ACRYLAMIDE TREATMENT IN CULTURED RAT KERATINOCYTES Keywords: Keratinocytes, acrylamide

desmosomes, cytokeratins,

intermediate filaments, internalization,

ABSTRACT. The present paper describes disturbances in the organization of tonofilaments and desmosomes of rat lingual and epidermal keratinocytes after treatment of the cells with acrylamide in culture. This treatment induced changes in cell shape, reduction of intercellular adhesion and a perinuclear accumulation of cytoplasmic organelles. Using specific antibodies for cytokeratins, the filaments were disorganized particularly in the perinuclear region. In untreated cells, keratin filament labelling was very weak or absent above and below the nucleus thus leaving a black nuclear space in fluorescine microscopy. Following acrylamide treatment, the keratin filament labelling covered the nuclear space which indicated the accumulation of these filaments all around the nucleus. Furthermore, the desmosomal junctions were often associated with thick keratin bundles. Antibodies for desmoplakins revealed a reduction in intercellular labelling and stronger cytoplasmic labelling. Ultrastructurally, well-developed long tonofilaments were found to associate with large desmosomal junctions. Furthermore. smallsized desmosomal structures were identified within the cytoplasm. Morphologically, these were identical to cell surface desmosomes and were almost always associated with well-developed tonofilaments. The effect of acrylamide on the protein kinase A activity might be implicated in the disturbances of the desmosome-intermediate filament complex and in the initiation of contractile forces necessary for pexinuclear accumulation of intermediate filaments and for the formation of intact cytoplasmic desmosomes. The acrylamide-induced intermediate filament and desmosomal changes may provide valuable information on the mechanism of intact cytoplasmic desmosome formation in several skin diseases and in squamous cell carcinoma.

Introduction

sizes and forms extending from tiny spots to extensive plaques that may be the result of lateral fusion of several desmosomes (Cowin et al., 1985a; 1985b). The two plasma membranes of the adhering cells are not in close contact but are separated by a 20-30 nm thick layer called desmoglea which often reveals a midline electron dense structure and two lateral zones of electron-dense anchoring filamentous material. In the cytoplasmic side the membranes are covered with symmetric electron dense plaques 10-40 nm thick. These cytoplasmic plaques are made of a proximal dense primary part and a distal loosely packed secondary part (Arnn and Staehelin, 1981). It was shown that cytokeratins bind certain components of the desmosomal molecules including desmocalmin

Within the epithelium, desmosomes play a major role in cell adhesion and are also the primary integrators of the cytokeratin containing intermediate filaments (Schwarz et al., 1990). They serve as specific attachment points for intermediate filaments (IF), creating a continuous network throughout the tissue by aligning the IF in all neighbouring cells. Typical desmosomes are disk-like structures varying in size between 200 and 600 nm. However, they can occur in a wide range of Laboratoire de Biologie-Odontologie, Institut Biomedical des Cordeliers, Escalier E, 2bme Etage. 15, rue de 1’Ecole de MCdecine, 75270 Paris Cedex 06, France. Correspondence to: Dr A. H. M. Shabana. Received 3 June 1993 Revised 31 August 1993 43

44

(Tsukita and Tsukita, 1985), band-6-polypeptide which binds only the acidic (type I) keratin family (Kapprell et al., 1988) and desmoplakins which may precisely interact with cytokeratins by their common periodicity of charged residues (Green et al., 1990). It was previously shown that acrylamide induces disturbances in the organization of IF and perinuclear accumulation of organelles in PtKl cells. This treatment does not alter protein synthesis or the integrity of the other cytoskeletal networks (Eckert, 1985). Also, it was shown that these effects are the result of inhibition of the CAMP-dependent protein kinase and keratin dephosphorylation (Eckert and Yeagle, 1988). The effects of acrylamide on keratinocytes IF has not been reported. In contrast to the PtKl cells, epidermal keratinocytes have more complex keratin filament networks and desmosomal junctions (Moll et al., 1982; Sawaf et al., 1990). Studies on the calcium-induced regulation of desmosomes have greatly helped us in understanding the mechanisms of desmosome assembly-disassembly and the interactions between the desmosomal molecules and tonofilaments, particularly the association between the desmoplakins and the cytokeratins (Mueller and Franke, 1982; Jones and Goldman, 1985; Green et al., 1990). Lowering calcium level in cell culture medium induces internalization of half desmosomes by pinocytosis (Duden and Franke, 1988), which represents a way for reduction of intercellular adhesion. These authors also observed occasional intact desmosomes in the cytoplasm and attributed their presence to the trypsinization of cells which may disturb the cell membrane. Decreased intercellular adhesion is a common landmark during development and in several pathological conditions characterized by hyperproliferation and increased cell movement, including wound healing as well as many malignant and premalignant lesions (Krawczyk and Wilgram, 1973; Ren et al., in intercellular 1990). The reduction adhesion is associated with a smaller number of desmosomal junctions particularly in invasive tumor cells. Electron microscopic investigations have shown that internalization of intact desmosomes is not uncommon in tumor cells and might represent an alternative way to reduce intercellular

SHABANA

ET AL.

adhesion (Klingmiiller et al., 1970; Franke et al., 1983). The mechanism by which intact desmosomes are internalized is not known. In the present paper we describe disturbances in the organization of IF and extensive accumulation of intact desmosomes in the cytoplasm of cultured rat keratinocytes following brief acrylamide treatment in a culture medium containing physiologic calcium level. We use a keratinocyte explant outgrowth technique in order to avoid internalization of desmosomes following cell trypsinization. The complete cytoplasmic desmosomes we observed might be newly synthetized structures in the wrong place or the result of internalization of pre-existing structures induced by physical forces exerted on the associated tonofilaments. Material and Methods Primary culture of rat keratinocytes

Sprague Dawley, 21-day-old fetal rat skin and tongue were cut into 1 mm3 dermal epidermal fragments using a punch instrument and collected in Eagle’s minimum essential medium with Earl’s salts (Boehringer, France), 2.5 pg/ml amphotericin, 100 II-J/ml penicillin, 100 fig/ml streptomycin (GIBCO) 10 ng/ml epidermal growth factor (GIBCO), and 10% filter-delipidized fetal calf serum. Primary cultures were initiated on 12mm diameter glass cover slides sterilized by autoclaving. The cover-slides were placed in 25well culture plates (Steriling). The cell culture medium was added as 1 ml/well, and the cultures were incubated in humid air containing 5% CO2 at 34°C (Jensen and Therkelsen, 1981). The tissue fragments were removed after the appearance of epithelial outgrowth around the tissue (48 hr). The outgrowth was composed almost entirely of epithelial cells (95%) which formed continuous sheets. Other cells in the cultures were predominantly fibroblasts. The medium was changed and the cells were further incubated for 48 hr. The acrylamide treatment of the rat keratinocytes was a modification of the method described by Eckert (1985). On the fourth day, cell culture medium deprived of the epidermal growth factor was used, and 24 hr later, 10 ~1 of 500 mM acrylamide (BioRad, France) in culture medium lacking EGF was added to the 1 ml culture medium per well (5 mM acrylamide). The cells were incu-

DESMOSOMES

AND INTERMEDIATE

FILAMENT

bated for 4 hr before examination. Equal volumes of the cell culture medium were added to the control cells and processed similarly.

DISTURBANCE

1s

thelial cytokeratin 7 which is not expressed by these cells. Fibroblasts provided a negative built-in control for the reactions of antidesmoplakin and anti-keratin antibodies.

Immunocytochemistry

Electron microscopy

The antibodies used for the localization of the desmosomal proteins and the cytoskeletal proteins were: mouse anti-desmoplakins 1 + II monoclonal antibody (Progen, Germany), rabbit anti-tubulin antibody (Bio makor, France) rabbit anti-actin antibody (Biomakor), rabbit anti-acactinin antibody (Biomakor) mouse anti-cytokeratin 14 monoclonal antibody (CKBl) (Sigma, France). The reaction sites were detected by using a goat anti-mouse IgG (H and L chains)-FITC or RITC labelled antibodies (Kirkegaard and Perry Inc, Maryland, USA) or goat anti-rabbit IgG (H and L chains)FITC labelled antibody (Biomakor). The specificities of these antibodies were confirmed in our laboratory by using frozen rat tissue sections known to contain (or not) the antigenic determinant searched for. The indirect immunocytochemical technique described earlier (Shabana et al., 1992) was used. The cells were washed three times using PBS (pH 7.4) before fixation and permeabilization. Pre-cooled solution of 50% methanol and 50% ethanol at -20°C was added (1 ml/well) in order to fix and permeabilize the cells. After 5 min, the wells were thoroughly washed and the excess of PBS was removed before the addition of the primary antibody. The pre-tested antibody dilutions were used to cover the epithelial sheets (about 100 pi/cover slide). The plates were incubated in a humid chamber at 37°C for 1 hr. After washing, the secondary labelled antibodies were incubated with the cells for 1 more hour. After thorough washing of the glass cover-slides, they were mounted and examined. A Leitz microscope equipped with epifluorescence filters for fluorescine and rhodamine was used to examine the cells, and photomicrographs were made using an oil immersion NPL Fluotar 25/0.75 and a 100/1~32-0~60 objective, and automatic photo-equipment. Haematoxylin and eosin staining was made on cells treated as above. Controls for the immunocytochemical technique were the use of the second antibody only and its use after a primary IgG monoclonal antibody specific for the simple epi-

For the ultrastructural examination the cells were washed in O-1 M sodium cacodylate buffer containing 0.1 M sucrose (pH 7.4). fixed for 30 min in 2% glutraldehyde in washing buffer, and post-fixed for 30 min in 0.5% osmium tetroxide in cacodylate buffer. After washing, the cells were stained with 2% uranyl acetate in 50% ethanol for 40 min during the conventional dehydration in increasing concentrations of ethanol solutions and finally in propylene oxide. Epon-araldite was used for embedding the cells on the glassslides. The blocks were heat-separated and thin sections were cut and stained with lead citrate solution (1.76 g lead citrate and 1.83 g lead nitrate in 50 ml H,O) for 30 min and examined with a Philips CM 12 electron microscope. Measurements and statistical analysis

All measurements were taken on electron photomicrographs of horizontally sectioned cells at the borders of the epithelial sheets. Cell surface desmosomes of both treated and untreated cells were measured and compared with desmosomes located within the cytoplasm of treated cells. 40 desmosomes were measured in each group. The desmosomal diameters expressed in nm units were compared and the difference between means was tested using the student t-test. A significant difference was accepted at values 295% confidence . Results Light microscopy

The epithelial outgrowth formed a continuous sheet of cells which surrounded the tissue fragment. The epithelial cells at the borders acquired morphological aspects typical of migrating cells as evidenced by the laterally expanding cytoplasmic lamellae with ruffled borders. A few cells were found to migrate short distances away from the epithelial sheet (1 mm was observed). Most of the epithelial cell divisions were localized in the periphery of the sheet, however these were also found within the sheet. Epithelial

DESMOSOMES AND INTERMEDIATE

FILAMENT DISTURBANCE

stratification occurred as early as the third day in culture. This stratification started from the tissue fragment side towards the borders of the outgrowth. After 5 days in culture there was still a border of epithelial cells in monolayer representing the expanding front within which some cells may temporarily overlap during cell migration or cell rearrangement. Haematoxylin and eosin stained cells were well-spread and had intimate intercellular contacts leaving very little or no intercellular spaces (Fig. la). The cytoskeletal network could be examined using the phase contrast accessories. This was spread in the cytoplasm around the nucleus and extended to the lateral membrane contacts. Treatment of the cells with acrylamide for 4 hr induced widening of intercellular spaces. Cell borders lacking intercellular contacts became concave, and those maintaining contacts became lightly stained and appeared devoid of organelles. The cytoplasmic density increased around the nucleus probably to the accumulation of organelles which were lacking in the cell borders (Fig. lb). In contrast to untreated cells, the cytoskeletal network was hardly visible in the perinuclear region. These modifications were best seen in the peripheral part of the monolayer, and accordingly we focused our results on the modifications in this part rather than on the stratified part of the culture.

47

actin, cu-actinin, tubulin and cytokeratin. Only the cytokeratin filaments were affected by acrylamide treatment. In both treated and untreated cells. the anti-actin labelled both F and G actin. The great majority of cells expressed only G actin. The formation of stress fibers was found in a few cells near the borders of the expanding epithelial sheets. The antibody for tubulins gave a diffuse cytoplasmic staining although some fine microtubules could be identified. Immunocytochemical examination of the intermediate filaments using antibody for cytokeratins showed marked differences in the organization of the keratin filaments. In the untreated cells, filaments of regular width and length were identified in the great majority of cells (Fig. lc). The fluorescine labelling was very weak or absent above and below the nucleus leadi,ng to a dark nuclear zone which corresponds to the nuclear size and form. The filaments extended from the peripheries of the nuclear zone to cell boundaries at areas where corresponding filaments of adjacent cells terminated. In well-spread cells, these filaments were highly organized and straight lines could be followed from the perinuclear zone to the cell borders (Fig. le). In less-spread cells, the filament width remained regular but the filaments may exhibit a complex network organization in the perinuclear zone and straight filaments in the cortical zone.

Immunocytochemistry

Keratin filaments in treated cells

The organization of the cytoskeletal network was examined by antibodies directed against

Acrylamide treatment of the epithelial cells induced modifications in the organization of

Fig. 1. The acrylamide-induced morphological (a, b) and immunocytochemical (c-h) modifications are presented; (a, c, e, g) untreated cells, and (b, d, f, h) treated cells. Without acrylamide (a), the keratinocytes are well-spread and in direct contact with each other. The cytoskeleton can be seen throughout the cytoplasm. The corresponding treated cells (b). have a densely stained perinuclear cytoplasm containing organelles and vesicles. The borders of the cells are free of these organelles and vesicles. The intercellular contacts are lost in certain areas leaving empty spaces (S). (Both a and b are haematoxylin and eosin stained). This change in cell morphology was associated with disorganization in the keratin intermediate filaments (c, d, e, f). The highly organized IF shown in (c) and (e) became extensively irregular after acrylamide treatment shown in (d) and (f), Note in (d) and (f) the masking of nuclear space (N) by the collapsing keratin filaments, and the variation in bundle diameter. Note also the thick filamentous labelling at the cortical cytoplasm particularly at sites of cell to cell contacts. The organization of desmoplakins was also altered following this treatment. Dotted lines at sites of intercellular contacts decorated the untreated cells in (g). These lines were absent or very limited after treatment in (h). The treated cells exhibited higher cytoplasmic labelling of both diffuse and dotted patterns in comparison with the untreated cells. Fig. la x 1100: Fig. lb x1100; Fig. lc x600; Fig. Id x1200; Fig. le x2100; Fig. lf x2100; Fig. lg x 2100: Fig. lh x2100.

DESMOSOMES

AND INTERMEDIATE

FILAMENT DISTURBANCE

the IF (Fig. Id). In contrast to the untreated cells, the treated cells exhibited accumulation of keratin filaments all around the nucleus which masked the nuclear zone. Keratin bundles varied markedly in width, thus increasing the effects of IF disorganization. Inter-filament contacts at sites of cell-cell adhesion were clearly labelled. This was the result of reduction of IF-related intercellular contact and the thickening of the cortical bundles (Fig. If). Organization of desmosomal proteins

In untreated cells, desmoplakins I and II were detected as dotted lines at the lateral cell-to-cell contacts. Dots at the dorsoventral aspects or in the cytoplasm were also found but represent a small percentage of the lateral dotted lines. The background labelling which represents the soluble pool of desmoplakins was low (Fig. lg). The acrylamide-treated cells showed marked reduction in lateral membrane labelling. Dotted labelling at dorsoventral membranes or in the cytoplasm was relatively very high. The antibody strongly labelled the cytoplasm indicating a high pool of soluble desmoplakins (Fig. lh). Ultrastructure

The ultrastructural examination of the untreated cells showed the presence of numerous desmosomal junctions of different sizes. The width of desmoglea was thinner than that of the cytoplasmic plaque. These desmosomes were associated with tonofilaments of moderate density. The filaments often disappeared in the cortical matrix rich in microfilaments. The intercellular contact was also assured by adherens junction associ-

4Y

ated with microfilaments (Fig. 2a). The intermediate filaments formed thicker bundles deeper in the cytoplasm and in the perinuclear zone. The acrylamide-treated cells exhibited irregular intercellular contacts and the intercellular communication was maintained by cytoplasmic projections and intercellular junctions. Desmosomes were of different sizes, however, the width of desmoglea was increased in relation with that of the cytoplasmic plaque. This was particularly true where well-developed thick and long tonofilaments were inserted in the cytoplasmic plaque (Fig. 2b). The cortical matrix was not rich in microfilaments although adherens junctions were also present and associated with microfilaments. The appearance of long and wide tonofilament bundles was more frequently observed suggesting lateral intermediate filament cross-linking. The most striking change was the presence of typical desmosomal structures within the cytoplasm. These were often associated wtth thick tonofilament bundles. Cytoplasmic desmosomes were classically found in parts of the cytoplasm where welldeveloped tonofilaments extended to surface desmosomes and never in cell areas featuring concave borders. They may also be found in two adhering cells (Fig. 2b) or more often in only one of the two cells (Fig. 3a). The fine structure of cytoplasmic desmosomes was strikingly similar to that of the surface desmosomes (Fig. 3b). They took the form of two symmetrical electron-dense plaques with a central area of moderate electron density corresponding to the desmoglea. In contrast to surface desmosomes which exhibited variation in the width of desmo-

Fig. 2. Comparing the tonofilament - desmosome complex in untreated cells (u), and treated cells (b), revealed important ultrastructural modifications induced by acrylamide. In untreated cells (a) the intercellular contacts were associated with intermediate filaments of moderate density (T) ending in desmosomal plaques at the cell membrane (D), and by actin rich cytomatrix containing microfilaments (M) ending in adherens junctions (A). These two cytoskeletal systems were co-localized in the cortical matrix. The acrylamide treated cells (b) had remarkably dense tonofilaments running a straight course at the cortical cytomatrix and ending in desmosomal plaques (D). The actin rich cortical cytomatrix was less evident. however, adherens junctions (A) and the associated microfilaments (M) could be identified. Numerous small size desmosomcs (arrows) were found in the cytoplasm of the two adhering cells shown in the Figure. The surface desmosomes (D) were larger in diameter and their desmoglea were wider particularly in those associated with thick tonofilaments. A microtubule is indicated by an arrow head. Fig. 2u ~2600; Fig. 26 ~26,000.

DESMOSOMES AND INTERMEDIATE

FILAMENT DISTURBANCE

glea, the central areas of cytoplasmic desmosomes were of uniform diameter. The electron-dense plaques were often associated with tonofilaments. The orientation of the tonofilaments was either perpendicular, oblique or parallel to the long axis of the desmosomal plaques. Another configuration was the encircling of a desmosome with tonofilament bundles, i.e. beginning from one electron-dense plaque and ending in the other, which indicated the absence of membrane around the desmosome. Comparing the size of the desmosomes at the cell surface and within the cytoplasm, the internalized desmosomes had electron-dense plaques varying in length between 56 and 303 nm with a mean value of 116 nm. The corresponding values for the surface membrane desmosomes in treated cells were 114 to 445 nm with a mean of 204 nm. The difference between the means was statistically significant (P < 0.001). In untreated cells, desmosomal size varied between 70 and 346nm with a mean value of 199nm. This was also significantly larger than the size of cytoplasmic desmosomes (P < O.OOl), but not different from surface desmosomes of treated cells. The intracytoplasmic desmosomes were almost always associated with well-developed tonofilament bundles. When the cells in contact were examined, cytoplasmic desmosomes were identified in cells having tonofilament accumulations in one pole of the cell. The cells with wide-spread tonofilament distribution did not contain desmosomes. Discussion In the present study, we used antibody which

51

recognizes cytokeratin 14 which is expressed by all stratified epithelia in man and many experimental animals (Roop et al., 1987; Fuchs and Marchuk, 1983; Shabana et al., 1989). In the present paper, we report the cross reactivity of the antibody with rat’s oral and skin stratified epithelia. Nelson and Sun (1983) showed that cytokeratin 14 is synthesized by basal cells, assembled with other cytokeratins into filaments, and maintained during epithelial differentiation. Thus it represents a reliable marker for the study of keratin filament organization at different stages of keratinocyte differentiation. Following acrylamide treatment, a marked disorganization of keratin intermediate filaments was found immunocytochemically which is related but not identical to that induced in the PtKl cells (Eckert, 1985). Work from several laboratories has shown that phosphorylation may play an important role in the control of keratin filament assembly-disassembly in the epithelial cells (Gilmartin et al., 1980; Yeagle et al., 1990; Eriksson et al., 1992). Several kinases are implicated in the phosphorylation of keratin intermediate filaments including protein kinase C, calmodulin-dependent protein kinase and CAMP-dependent protein kinase (Eckert, 1985; Yeagle et al., 1990; Yano et al., 1991). Treatment of the PtKl epithelial cells with acrylamide was previously shown to induce aggregation of the keratin and vimentin rich IF (Eckert, 1985). These filaments collapsed around the nucleus leaving the cell lamellae free from IF. It was shown that acrylamide treatment inhibits the phosphorylation of the IF by affecting a CAMPdependent protein kinase (Eckert and Yeagle, 1988). Yeagle et al., (1990) showed

Fig. 3. Cytoplasmic desmosomes may be found in only one of the two adhering cells as shown in (n). Their fine structure is presented in (b). The upper cell does not possess cytoplasmic desmosomes. In contrast, the lower cell contains a large number of cytoplasmic desmosomes (arrows). The later cell shows morphological evidence of contraction including increased intercellular spaces (S) and concave outlines at free borders. The cell. however, maintains contact with the cells on the left and upper side through numerous cytoplasmic processes, adherens junctions and a zone of grouped surface desmosomes (D). The tonofilaments (T) arc dense and grouped at the region of desmosomal junctions and extend to juxtanuclear cytoplasm. In association with these bundles, numerous cytoplasmic desmosomes are found. The upper cell exhibits wider cytoplasmic distribution of tonofilaments. The ultrastructure of the cytoplasmic desmosomes (b), was identical to that of surface desmosome; the symmetric electron-dense plaques, the lameller pattern of the middle part, the moderate electron-dense tilamentous part corresponding to the desmoglea, and the bilateral. tonotilament insertions in the desmosomal plaques. Fig. 3a x11,000; Fig. 3b x70,OMl.

52

conformational changes in keratin molecule upon changes in phosphorylation state. This is manifested by altered mobility or flexibility of the terminal regions of the keratin molecule that may interfere with interactions of the keratin filaments with other cytoplasmic components and ending by formation of aggregates. These authors also found changes in the N terminal following CAMPdependent phosphorylation that may alter interactions of IF with other components of the cell surface and result in perinuclear accumulation as shown in the PtKl cells. In the present study, the keratin-based intermediate filaments did not completely collapse in a perinuclear zone. The filaments accumulated at high density in the perinuclear region, however, thick bundles connected these to the cell membrane where desmosomes were found. This observation might suggest that keratin filaments not connected to desmosomes collapsed in the perinuclear zone. Alternatively, the thick bundles might be more resistant to the effects of acrylamide through extensive cross-linking or other mechanisms. The changes in cell morphology following acrylamide treatment of rat keratinocytes is comparable to the effect of CAMP-dependent protein kinase inhibitor on mammalian cells (Fernandez and Lamb, 1991). Cells injected with a specific kinase A inhibitor immediately acquire a prophase morphology which is characterized by contraction and perinuclear accumulation of the cytoplasm. The above authors also observed reorganization of microtubules following injection of the kinase inhibitor which was not observed after acrylamide treatment of rat keratinocytes and PtKl cells. This difference might be explained by a lighter inhibitory effect of acrylamide on the kinase in comparison with the effect of specific kinase inhibitor. We have previously described the organization of keratin filaments in migrating keratinocytes (Shabana et al., 1992). In the leading edge or lamellipod, the IF form interconnected network that fades gradually towards the protrusion region as defined by Dunn and Brown (1987). In contrast, they form straight and thick bundles in the uropod extending from the nuclear region to the distal cell junction. These different organizations suggest not only that keratin filaments are dynamic but also that they are

SHABANA ET AL.

associated with stable cell compartments. Since cell locomotion requires extension of lamellipod followed by nuclear displacement, the length of uropod increases before its retraction (DiMilla et al., 1991). It is tempting to speculate that tension is created on the filaments connecting the nucleus to the distal cell junction. We related the thick and long bundles extending from the desmosomes to the nuclear zone after acrylamide treatment to those found in the uropod, the tension being created by the cytoplasmic accumulation in the juxtanuclear area. There are several reasons to propose that these bundles are under physical tension; (1) their organization in straight lines, (2) their parallel organization in the form of thick tonofilament bundles as shown in electron microscopy, (3) the widening of the desmoglea, and (4) their localization in the cell uropods or equivalent cell processes subjected to physical tension. It is important to note that microfilaments and intermediate filaments are ultrastructurally associated and co-localized before and during desmosome formation as shown in primary mouse keratinocytes (Green ef al., 1987). On the molecular level, cross-linking IF with microfilaments can be established by plectin through its ability to bind IF and the microfilament associated molecules such as fodrin (reviewed by Wiche, 1989; Foisner and Wiche, 1991). The complete cytoplasmic desmosomes described in this paper should be distinguished from cytoplasmic half desmosomes. Cytoplasmic half desmosomes were observed following calcium deprivation in the cell culture medium. At low calcium, the cells lose their surface desmosomal junctions by affecting the calcium dependent proteins; calmodulin, desmocalmin (Tsukita and Tsukita, 1985), desmogleins and desmocollins (Matty et al., 1987; Steinberg et al., 1987). The cells exhibit cytoplasmic aggregates of electron-dense plaques and half desmosomes are found to associate with cytoplasmic vesicles. These desmosomal structures contain the desmosomal markers desmoplakins (Jones and Goldman, 1985; Duden and Franke, 1988), desmoglein and desmocollins (Mattey and Garrod, 1986a,b). It was suggested that internalization of desmosomal plaques takes place during the trypsinization of cultured epithelial cells and accordingly

DESMOSOMES

AND INTERMEDIATE

FILAMENT

half desmosomes may be present in the membrane of pinocytotic vesicles (Duden and Franke, 1988). Half desmosomes were also internalized by pinocytosis during the transformation of bladder epithelial cells into fibroblastic cells by the transforming growth factor alpha (Gavrilovic et al., 1990). It is tempting to speculate that the capacity to internalize and reform desmosomes is critical for developmental events such as cell separation and rearrangement and for the pathogenesis of epithelial disorders including tumors. Cytoplasmic desmosomes described here are identical morphologically to those observed in vivo in squamous cell carcinoma more than 20 years ago (Klingmiiller et al., 1970). We show in the present paper that intracytoplasmic demosomes preserve their morphological characteristics, particularly the pair of dense plaques, the lamellated appearance of the central discs and the association with tonofilaments. In vivo, intracytoplasmic desmosomes are not restricted to squamous cell carcinoma. It was also reported in epithelioid sarcoma where the IF may be accumulated in the juxtanuclear region forming whorled arrangements (Ishida et al., 1992). Several epidermal and oral mucosal hyperproliferative diseases may similarly show this phenomenon including keratoacanthoma and Bowen’s disease (Sato and Seiji, 1973). Bowen’s disease is characterized by perinuclear aggregation of tonofilaments in the dyskeratotic cells and desmosomal structures associated with these aggregates. There are three possible ways by which desmosomes may be found in the cytoplasm. It can be assumed that the occurrence of invaginations of the plasma membrane is a normal event in keratinocytes and that the invaginated plasma membrane can contain desmosomal structures. These structures are resistant to enzymatic destruction and thus can be found in the cytoplasm (Schenk, 1975). However, the cytoplasmic desmosmes we observed may be located within the juxtanuclear cytoplasm, away from membrane interdigitations. Another possibility is that cytoplasmic desmosomes may be the result of new formation in the wrong place. This hypothesis is supported by; the absence of membrane vesicles around at least one side of the desmosomal plaques, the regularity in

DISTURBANCE

53

size of cytoplasmic desmosomes in contrast to the wide variation in surface desmosome diameter, the regularity in structure which argues against expected degradation of internalized surface desmosomes, and the bilateral association with tonofilaments which is more difficult to consider in a hypothesis based on internalization of surface desmosomes. A mechanism for desmosome new formation in the cytoplasm may be based on disturbances in desmosomal protein and glycoprotein transport to the cell borders. We have observed accumulation of cytoplasmic organelles and vesicles in the juxtanuclear region. These vesicles may contain desmocollins and desmogleins. In the presence of cytoplasmic pools of other desmosomal plaque proteins, all desmosomal components will be available for assembly in the cytoplasm. It would be interesting to test the effect of acrylamide on kinase C which was shown to inhibit desmoplakin transport to the cell membrane (Sheu et al., 1989). The third possibility is the internalization of complete desmosomes from the cell surface. In calcium switch experiments, it was shown that extended incubation of the cells in high calcium medium results in stable desmosomes which cannot be split when treated with low calcium medium. Instead, these are endocytosed intact by one of the partner cells after ripping out of the membrane (Mattey and Garrod, 1986b; Buxton and Magee, 1992). Under such conditions, the presence or absence of membrane around the internalized desmosomal plaque may depend on the area of the cut membrane and on cell proteolytic activity. 4 hr incubation with acrylamide may be sufficient for the proteolysis of the internalized membrane but not the more resistant desmosomal components. Entire desmosome internalization might not occur before stabilization of the desmosomes and in the absence of intermediate filament bundles bearing tension created during cell contraction. Sato and Seiji (1973) suggested that desmosomes were drawn in the cytoplasm by the tonofilaments. The morphological configurations of tonofilaments provide evidence that the desmosomes are under tension. The qualitative and quantitative differences between the cytoplasmic and the surface membrane desmosomes suggest that they are distinct structures. One might think that small-sized

54

SHABANA

desmosomes are easier to pull within the cytoplasm than larger ones. However, a contradictory argument may be that stabilized desmosomes (subject for intact internalization) should be larger than non-stabilized newly-formed ones. Extensive studies are required in order to understand the mechanism of cytoplasmic desmosome formation by acrylamide treatment and under physiological calcium level. The present model may provide valuable information on this mechanism of formation

ET AL.

and the functional significance of cytoplasmic complete desmosomes. Acknowledgements

We are thankful for the valuable discussions with Prof. Goldberg, Faculty of Dentistry, Montrouge, and for the technical assistance of Mme N. Martin in cell culture. The secretarial work of Mr E. Marie-Rose and Mme K. Marlin is much appreciated.

References Arnn J., Staechelin L. A 1981. The structure and the function of spot desmosomes. Dermatology 20, 33&339. Buxton R. S., Magee A. I. 1992. Structure and interactions of desmosomal and other cadherins. Semin. Cell Biol., 3, 157-167. Cowin P., Franke W. W., Grund C., Kapprell H.-P., Kartenbeck J. 1985a. The desmosome-intermediate filament complex. In The cell in contact (eds. G. Edelman and J. P. Thiery), pp. 426-460. John Wiley and Sons, New York. Cowin P., Kapprell H.-P., Franke W. W. 1985b. The complement of desmosomal plaque proteins in different cell types. J. Cell Biol., 101, 1442-1454. DiMilla P. A., Barbee K., Lauffenburger D. A. 1991. Mathematical model for the effects of adhesion and mechanics on cell migration speed. Biophys. J., 60, 15-37. Duden R., Franke W. W. 1988. Organization of desmosomal plaque proteins in cells growing at low calcium concentrations. J. Cell Biol., 107, 1049-1063. Dunn G. A., Brown A. F. 1987. A unifying approach to analysing cell motility. J. Cell Sci. (Sup. 8) 81-102. Eckert B. S. 1985. Alteration of intermediate filament distribution in PtKl Cells by acrylamide. Eur. J. Cell Biol., 37, 169174. Eckert B. S., Yeagle P. L. 1988. Acrylamide treatment of PtKl cells causes dephosphorylation of keratin polypeptides. Cell Motil. Cytoskeleton, 11, 24-30. Eriksson J. E., Opal P., Goldman R. D. 1992. Intermediate filament dynamics. Curr. Op. in Cell Biol., 4, 99104. Fernandez A., Lamb N. 1991. Mitotic control in mammalian cells, Positive and negative regulation by protein phosphorylation. In Cellular regulation byproteinphosphorylation (ed. L. M. G. Heilmeyer), pp. 397-409. SpringVerlag Berlin, Heidelberg. Foisner R., Wiche G. (1991). Intermediate filament-associated proteins. Curr. Opin. Cell Biol., 3, 75-81. Franke W. W., Moll R., Mueller H., Schmid E., Kuhn C., Krepler R., Antlieb U., Denk N. 1983. Immunocytochemical identification of epithelium-derived human tumors with antibodies to desmosomal plaque proteins. Proc. Nod. Acad. Sci. USA, 80, 54?-547.

Fuchs E., Marchuk D. 1983. Type I and type II keratins have evolved from lower eukaryotes to form the epidermal intermediate filaments in mammalian skin. Proc. Natl. Acad. Sci. USA, 80, 5857-5861. Gavrilovic J., Moens G., Thiery J. P., Jouanneau J. 1990. Expression of transfected transforming growth factor alpha induces a motile tibroblast-like phenotype with extracellular matric degradation potential in a rat bladder carcinoma cell line. Cell Regul., 1, 1003-1014. Gilmartin M. E., Culbertson V. B., Freedberg I. M. 1980. Phosphorylation of epidermal keratins. J. fnoest. Dermatol., 75, 211-216. Green K. J., Geiger B., Jones J. C. R., Talian J. C., Goldman R. D. 1987. The relationship between intermediate filaments and microfilaments before and during the formation of desmosomes and adherens type junctions in mouse epidermal keratinocytes. J. Cell. Biol., 104, 138%1402. Green K. J., Parry D. A. D., Steinart P. M., Virata M. L. A., Wagner R. M., Angst B. D., Nilles L. A. 1990. Structure of the human desmoplakins: Implication for function in the desmosomal plaque. J. Biol. Chem., 26.5, 2603-2612.

Ishida T., Oka T., Matsushita H., Machinami R. 1992. Epithelioid sarcoma: An electron microscopic, immunohistochemical and DNA flew cytometric analysis. Virchows Arch. A., Pathol. Amt. Histopathol., 421, 401-408. Jensen P. K. A., Therkelsen A. J. 1981. Cultivation at low temperature as a measure to prevent contamination with tibroblasts in epithelial cultures from human skin. J. Inuest. Dermatol., 77, 210-212.

DESMOSOMES

AND INTERMEDIATE

Jones J. C. R., Goldman

FILAMENT

R. D. 1985. Intermediate

101,50&517. Klingmiiller G., Klchr H. U., Ishibashi

DISTURBANCE

filaments

and the initiation

Y. 1970. Desmosomen

im cytoplasma

55 of desmosomal

assembly. 1. Cd Biol I

entdifferenzierter

keratinocyten

dcs

platten epithelcarcinoma. Arch. Him Exp. Dermat~l., 238, 356-367. Krawczyk W. S., Wilgram G. F. 1973. Hemidesmosome and desmosome morphogenesis during epidermal wound healing. J. fJltrmfruc. Res., 45, 9>101. Mattey D. L., Gorrod D. R. 1986a. Calcium-induced desmosome formation in cultured kidney epithelial cells. J. Cell Sci., 85,95-111. Mattey D. L.. Garrod D. R. 1986b. Splitting and internalization of the desmosomes of cultured kidney epithelial cells by reduction in calcium concentration. J. Cell Sci., 85, 113-124. Mattey D. L., Suhrbier A., Parrish E., Garrod D. R. 1987. Recognition, calcium and the control of desmosome formation. In Junctional complexes in epithelid cells. pp. 4%65. CIBA foundation symposium 125. John Wiley and Sons, New York. Mall R., Franke W. W., Schiler D. L. 1982. The catalogue of human cytokeratins: patterns of expression in normal epithelial, tumor and cultured cells. Cell, 31, 11-24. Mueller M., Franke W. W. 1982. biochemical and immunological characterization of desmoplakins I and II. and major polypeptides of the desmosomal plaque. Differentiation, 23, 189-205. Nelson W.. Sun T.-T. 1983. The 50 and 58 k dalton keratin classes as molecular markers for stratified squamous epithelia: cell cultures studies. J. CeN Bid., 97, 244251. Rcn .I. Hamada J. I., Takeichi N., Fujikawa S., Kobayashi H. 1990. Ultrastructural differences in junctional intercellular communication between highly and weakly metastatic clones derived from rat mammary carcinoma. Cancer Res 50, 358-362. Roop D. R., Huitfeldt H., Kilkenny S. H. 1987. Regulated expression of differentiation-associated keratins m cultured cpidermal cells detected by monospecific antibodies to unique peptides of mouse epidermal kcratins. Differentration, 35, 143-150. Sato A., Seiji M. 1973. Electron microscopic observations of malignant dyskeratosis in leukoplakia and Bowen’s disease. Acta Dem. Venereal. (Stockh.), 53s, 101-110. Sawaf M. H.. Ouhayoun J. P. O., Shabana A. H. M., Forest N. 1990. Cytokeratin expression in human tongue epithelium. Am. J. Amt., 189, 155-166. Schenk P. 1975. Desmosomale strukturen im cytoplasma normaler und pathologischer keratinocyten. Arch. Dermutol. Res.. 253, 2>42. Schwarz M. A., Owaribe K., Kartenbeck J., Franke W. W. 1990. Desmosomes and hemidesmosomes: Constitutwe molecular components. Ann. Rev. Cell. Bioi., 6, 461-491. Shabana A. H. M.. Khattab R., Sawaf M. H., Forest N., Ouhayoun J. P. 1992. A switch in cytokeratin expression and intermediate filament organization associated with epithelial stratification. J. Eiol. Bum/e., 20, 33-34. Shabana A. H. M., Ouhayoun J. P., Sawaf M. H., Forest N. 1989. A comparative biochemical and immunological analysis of cytokeratin patterns in the oral epithelium of the miniature pig and man. Arch.7 Oral Bid.. 34, ?40259. Sheu H.-M., Kitajima Y., Yaoita H. 1989. Involvement of protein kinase C in translocation of desmoplakms from cytosol to plasma membrane during desmosome formation in human squamous cell carcinoma cells grown in low to normal calcium concentration. Exp. Cell Res., 185, 176190. Steinberg M. S.. Giudice G. J., Shida M., Pate1 N. H., Blaschuk 0. W. 1987. On the molecular organization, diversity and functions of desmosomal proteins. In Junctional complexes in epithelial cells. pp. 3-25. Ciba foundation symposium 125. John Wiley and Sons, New York. Tsukita S., Tsukita S. 1985. Desmocalmin: a calmodulin-binding high molecular weight protein isolated from drsmosomes. J. Cell Biol.. 10, 2070-2080. Wiche G. 1989. Plectin: General overview and appraisals of its potential role as a subunit protein of the cytomatrix. Crit. Rev. Biochem. Mol. Biol., 24, 41-67. Yeagle P. L., Frye J., Eckert B. S. 1990. Phosphorylation modulates keratin structure. Biochemistry, 29, 150X-1514. Yano T.. Tokui T.. Nichi Y., Nishizawa K., Shibata M., Kikuchi K., Tsuiki S., Yamaushi T., lnagaki M. 1991. Phosphorylation of keratin intermediate filaments by protein kinase C, by calmodulin-dependent protem kinasc and by CAMP-dependent protein kinase. Eur. J. Biochem., l!V, 231-290.