Traversing filaments in desmosomal and hemidesmosomal attachmentsFreeze-fracture approaches toward their characterization.код для вставкиСкачать
THE ANATOMICAL RECORD 1991-14 (1981) Traversing Filaments in Desmosomal and Hemidesmosomal Attach ments : Freeze-Fractu re Approaches Toward Thei r Characterization DOUGLAS E. KELLY AND AILEEN M. KUDA Department of Anatomy, Uniuersity of Southern California School of Medicine, Los Angeles, California90033 ABSTRACT Desmosomes and hemidesmosomes from larval newt epidermis were examined by freeze-fracture methods incorporating low osm olality fixation, short duration glycerination, complementary replica comparison, stereo imaging, and dark shadow printing. These procedures provide new evidence regarding the structure of "traversing" filaments as mediators of attachment between intermediate filaments and the cell membranes of desmosomes and hemidesmosomes. A detailed analysis of intramembranous particles and other structure in these attachments has also been possible. The relationship of this evidence to models of desmosomal structure suggested by other authors is discussed. Desmosomes are now generally regarded a s focal areas of firm intercellular membraneto-membrane adhesion a t which a complex network of cytoskeletal intermediate filaments (or tonofilaments) is anchored (reviewed by McNutt and Weinstein, '73; Staehelin, '74). Hemidesmosomes display outwardly similar structure but serve to attach cells to basal laminae and subjacent connective tissue strata (Kelly, '66; Shienvold and Kelly, '76). Little structural or biochemical information exists either a s to the nature of the intercellular adhesive vehicle or the mode of intermediate filament attachment to cell membranes. However, a number of possibilities have recently been proposed, largely as a result of studies using freeze-fracture. It is doubtful that intermediate filaments adhere directly to the cell membrane in desmosomal regions inasmuch as transmission electron micrographs of sectioned material frequently show them to converge toward, then arch away from, the dense proteinaceous cytoplasmic plaque that characterizes each side of a typical desmosome (Fawcett, '61;Kelly, '66).In some epithelia the intermediate filaments run a course parallel to the plaque, but at a distance of 20-40 nm from the cell membrane. In no instance have intermediate filaments been demonstrated to terminate a t a desmosome or insert directly into the desmosomal membrane. High resolution images often suggest a finely filamentous network surrounding the plaque and spanning the interval between desmo- 0003-276X/81/1991-0001$02.60 0 1981 ALAN R.LISS, INC. somal membrane on the one side and intermediate filaments on the other. On the basis of freeze-fracture images, McNutt and Weinstein ('73)first postulated that this network serves as the anchoring vehicle between intermediate filaments and the desmosomal membrane. We have introduced the term "traversing filaments" to denote fine filamentous profiles emanating from the intermediate filament region and extending onto the E-face of fractured desmosomal membranes (Kelly and Shienvold, '76). However, such images have been suspect for their similarity to nonspecific stress lines caused by plastic deformation. Freeze-fractured desmosomal membranes display aggregated intramembranous particles that are slightly larger and more irregular than the typical 10 nm intramembranous particles. These have been reported to cling predominantly to the E-fracture face in the desmosomes of unfixed non-keratinized epithelia (McNutt and Weinstein, '73; Kelly and Shienvold, '76), to remain with the P-fracture face in fixed non-keratinizing epithelia (Kelly and Shienvold, '76; Ishimura et al., '79)) or to be dispersed equally between the two faces (Breathnach et al., '72, '76; Staehelin, '74; Shimono and Clementi, '77). In keratinizing epithelial cells, the particles are found on both faces regardless of whether glutaraldehyde fixation has beenused (Breathnach et al., '73; '76). ~ Received June 13, 1980; accepted July 10, 1980. 2 DOUGLAS E. KELLY AND AILEEN M. KUDA Some workers have suggested that a relationship exists between the fine cytoplasmic traversing filaments and the intramembranous particles (McNutt and Weinstein, ’73; Kelly and Shienvold, ’76; Leloup et al., ’791,but Staehelin and Hull (‘78) diagrammed a scheme in which a network of small caliber filaments (termed “linkers”) extend from the desmosomal plaque of one cell, directly through its membrane, across the intercellular gap, through the membrane of a neighboring cell, and into its plaque. In this scheme, intermediate filaments are in turn anchored to the plaques by another set of filaments, the plaque being a region of interconnection between the two sets of small caliber filaments. In contrast, Leloup et al. (’79) have proposed that “protofilaments unravel from tonofilaments, course through the plaque, pierce their respective desmosomalmembranes and end by anchoring in the midline of the intercellular space, joining there with similar units from the neighboring cell.” Hence, while the various investigators are agreed on the existence of some form of traversing filaments, and that these likely connect directly or indirectly t o intermediate filaments, there is no consensus as to whether they terminate in the plaque, within the membrane, or within the intercellular space. An accurate interpretation of the nature of particles and filaments of such minute dimensions is obscured by imaging phase granularity, replica granularity, replica plastic deformity, and intentional or unintentional sublimation of the specimen prior to replication. In our prior investigations (Kelly and Shienvold, ’76), we were disturbed that images of traversing filaments seen extending between intermediate filaments and exposed desmosoma1 membrane E-face were far more obvious and numerous in unfixed tissue than when glutaraldehyde fixation was employed. We have therefore sought technical modifications that might clarify the reality and extent of these fine filamentous components. This report summarizes results obtained by the use of abbreviated glycerination in conjunction with glutaraldehyde fixation a t an osmolality lower than has been used with prior methods. Stereo imaging in which tilt axis and replica shadow azimuth are coordinated, complementary replica comparison, and dark shadow photographic printing have also been used to advantage. We believe that these procedures allow added assurance as to the reality of traversing filaments and provide additional information about their behavior during the freeze-fracture replication process. The resultant observations offer new insight on the mode of termination of these filaments within desmosomal membranes, and also clarify the architecture of hemidesmosomes. MATERIALS AND METHODS Larval west coast newts, Taricha torosa, were maintained in the laboratory after hatching. As noted in our prior studies, the epidermis of these animals provides highly ordered and distinct desmosomes and hemidesmosomes. Animals of 23-50 mm (approximate lengths) were anesthetized with tricaine (ethyl maminobenzoate methanesulfonate) and tail pieces were excised. The tissues used for the images in this report were fixed a t room temperature for 30minutes in 1.Wo glutaraldehyde with 0.1 M cacodylate buffer (pH 7.4) and 5mM CaC1, (310mOsm), embedded in 3.0%agar, and sliced on a Sorvall TC-2 tissue chopper.The 100 pm slices were freed of agar and glycerinated a t room temperature for limited periods ranging from 1 to 2.5 hr (total time) in 10% and 20% glycerol in cacodylate butYer. After sandwiching between two gold discs the tissues were quickly frozen in liquid Freon 22 cooled to liquid nitrogen temperature, then fractured a t -110°C in a Balzers BAF 301 instrument equipped with a double replica device. Platinum-carbon replication of the fractured surfaces was performed with minimal etching and a t a 45“ angle. Additional carbon reinforcement was applied from 90”in two layers as the tissue was allowed to warm to -30°C. Replicas were processed through methanol, household bleach, distilled water, 1.0% acetic acid, and several water rinses before being picked up for study on 75 or 100 mesh parlodion-coated grids. Replicas were examined with a JEOL-1OOC microscope equipped with a goniometer stage; stereo pairs were obtained after adjustment of the goniometer stage to achieve parallel alignment of tilt axis and the shadow azimuth of the specimen. This maximizes the stereoscopic effect of the 12-14” angle of tilt (see Observations-Technical). Final dark shadow photographic prints were made with an intermediate positive transparency, which was a direct contact exposure of the original EM negative. NOTE: Stereo-pair micrographs are mounted for viewing with a standard cartographer‘s stereo-viewer or using a wall-eyed viewing technique. Cross-eyed viewing will provide a confused “false-stereo” image. DESMOSOMAL TRAVERSING FILAMENTS 3 Fig. 1. Two freeze-fracture stereo-electron micrograph pairs of a desmosome depicting conventional light shadow imaging (top pair) and dark shadow imaging (bottom pair). Note that finer details of the desmosome such as 1&15 nm tonofilaments (TO),6-8 nm traversing filaments VR), and intramembranous particles of the P-face (P) are revealed with more distinct elevation when the dark shadow imaging is employed. The rest of the replica-stereo pairs in this report have been reprodud by the dark shadow imaging technique. Note example of a traversing filament, whi+ extends from the cytoplasm onto the desmosomal E-face (arrow). (El, E-face; (IS),intercellular space. x 58,600. The accompanying micrographof a sectioned desmosomedisplays major components and the approximate fracture (dotted line) achieved in the replica. 4 DOUGLAS E. KELLY AND AILEEN M. KUDA OBSERVATIONS Technical considerations The use of relatively low OSmOlalitY fixation and shortened duration of glycerol treatment has resulted in better quality freeze-fracture replicas of lama1 amphibian desmosomes and hem~desmosomesthan we have achieved with Previous procedures. In particular, fine filaments, which heretofore were only suggested in unfixed preparations and not seen with claritY in fixed material, are now revealed distinctly and with consistency. Analysis of fine structural features in replicas of desmosomes and hemidesmosomes has been aided by the coordinated use of several refined techniques. First, stereo-imaging is helpful in establishing a n accurate appreciation of the full three-dimensional relief contained in freeze-fracture replicas. Analysis Of single images is reasonably accurate with regard to the two lateral dimensions but Often provides a n underestimation of depth. Therefore, the illustrations in this article are portrayed in stereo-pairs. They are arranged for viewing with a StereO-VieWer; crosseyed Viewing will require reversal of the images presented, left for right, in order to avoid confusing false stereo-imaging. Wall-eyed viewing can be used in lieu of a stereo-viewer and without reversing the images. Second, we have found that fine details are more distinctly presented when a "dark shadow" print is used for final analysis. Two identical stereo-images (made from the same negatives) are shown in Figure 1. The stereo-pair has been prepared by the conventional light shadow technique, and the lower pair illustrates dark shadow processing through the use of an intermediate tramparencY. Both Pairs Provide adequate stereo-imaging if oriented as presented here, SO that the dark sides of fine raised details face downward in each illustration. However, close comparison of the two images will show a n enhanced distinctness and appreciation for elevation of finer details (e.g. filaments and particles) in the lower, dark shadow stereo-pair. Because of this advantage, all subsequent figures in this report are prints processed by the dark shadow technique. In such images, light areas in the replica represent regions of heavy platinum deposition, and dark areas are regions protected from such deposition (shadowed in a freer sense of the word). With the exception of the top pair in Figure 1, all figures are presented with the direction (azimuth) of metal deposition from the top. This is opposite to the conventional orientation used for light shadow images. Third, a n aspect commonly overlooked in the use of stereo-pairs with freeze-fracture replicas involves coordination of metal deposition direction and tilt =is. Much of the threedimensional quality of a single conventional micrograph of a freeze-fracture replica derives from shadow direction, with the attendant necessity of orienting the micrograph SO that the direction of metal deposition is from the bottom of a figure. Inverting this relationship reverses perceived depth and confuses true relief, unless a reversed, or dark, shadow image isutilized. In any stereo-pair, three-dimensional information is provided by the angular differencein the optic with each member ofthe pair has been obtained and viewed. This is achieved tiit alongan axisthat necesthrough specimen sarily lies vertical in the &sewer's field of vision. Rotation of that axis gradually reduces, and at 90" cancels, the three-dimensional effect. Metal deposition direction and tilt axis azimuth can be potentially reinforcing or antagonistic in presentation of depth. If maximal reinforcement is toresult, then the metal &Position azimuth of areplica must be arranged to coincide with the tilt axis used in stereo electron microscopy. All of the illustrations in this report,as well as many others we have studied, have been coordinated carefully to provide this reinforcement. Fourth, we report here the first published complementary replica images of desmosomes of comand hemidesmosomes. comparison plementary replicas provides some additional insight into the arrangement of finer components of these adhesive structures. When complementary replicas have been achieved and the same d e ~ ~ s o m e&midesmosame 10cated on each replica, we have taken stereo pairs of both members using coordinated tilt and &adow in order to provide the advantages of all the above refinementsin a complementary pair comparison. Several examples are included in this report ( ~ i2 ,~3 and ~ .8). Biological findings In previous studies (e.g. Breathnach et al., '72, '73, '76; McNutt and Weinstein, '73; Kelly and Shienvold, '76; Caputo and Peluchetti, '77; Staehelin and Hull, '78; Brown and Ilic, '79) it has been noted that the intramembranous particles characteristic of desmosomes are about 10-15 nm in diameter, rather closely packed, but of irregular shape and orientation. When DESMOSOMAL TRAVERSING FILAMENTS 5 Fig. 2. Stereo-pairsof complementaryreplicas (topand bottom) in which split membranes of three desmosomal territories are visible (encircledby dotted lines). Note the difference in the caliber of tonofilaments (TO)and traversing filaments (TR), both of which appear as elevations on both sides of the replica. Typical desmosomal intramembranous particles are Seen on P-faces (P)of each desmosomal territory, but the corresponding desmosomal E-face is relatively featureless in these replicas. One desmosomal territory discloses traversingfilamentsextendingfrom the cytoplasm onto the surfaceof the E-face (arrows). x 64,000 fixation is avoided, the particles of l q a l newt epidermal desmosomes appear to remain adherent to the E-fracture face, but infixedpreparations they are found predominately on the Pfracture face (Kelly and Shienvold, '76). These relations and sizes are confirmed in the present observations of differently fixed epidermis (Figs. 1, 2, 3 and 51, but in addition 8-10 nm thread- or ridge-like profiles are also a part of the P-face desmosomal territory, wedged among the aggregated particles. In most fixed preparations, the corresponding E-face appears smooth and devoid of particulation. However, in some of the present replicas, we have noted 6 DOUGLAS E. KELLY AND AILEEN M. KUDA Fig. 3. Stereo-pairsof complementaryreplicas (topand bottom) of the split membranes of a desmosomedepicted at higher magnification.The desmosomal P-face(P) displaysclosely aggregated, irregularly shaped, 10- 15nm particles, many of which are attached or related to enlongate ridge-like structuresof approximately the same or smaller caliber (arrows).In the upper replica, the desmosomal E-face (Elshows a slight indication of a finer particulate “pebbling.” (IS), intercellular space. x 117,000 Fig. 4. Stereo-pair showing the E-face aspect of a desmosomal territory (E).A distinct pebbling is evident, the individual particles of which are smaller in diameter and ofmore irregular shape than either the general or desmosomal intramembranous particles found adherent to the P-face (P) of the split membrane of the adjacent cell. The intercellular space of the desmosome displays linear profiles (arrows), some of which seem to enter the two desmosomal membranes. x 72,700 Fig. 5. Stereo-imagesof a replica showing the apical junctional region between two surface epidermal cells. The apical epithelial surface is at the bottom of the picture, and next to it a tight junction (TJ)is visible in the replica. Two desmosomal territories (outlined) are identifable by the typical particulate aggregates along the P-face (P) of the cell on the right. to the line where fractured cytoplasm intersects the Traversing filaments (TR)are visible coursing from tonofilaments (TO) cytoplasmic leaflet of the desmosomal membrane. They appear to penetrate that leaflet and relate to the P-face desmosomal ridges or particles. Linear profiles in the intercellular space (arrows) may also relate to the particles after penetrating the external desmosomal membrane leaflet. (E), E-face. x 72,700 DESMOSOMAL TRAVERSING FILAMENTS 7 8 DOUGLAS E. KELLY AND AILEEN M. KUDA E-face desmosomal territories in which a distinct finer “pebbling” is apparent (compare Figs. 2, 3, 4 and 6). The size of these E-face particles is quite variable, but nearly always smaller than 8 nm. The typical P-face particulation is regularly seen in all desmosomal territories whether or not exposed E-faces of the same desmosome appear smooth or pebbled. Therefore, E-face pebbling does not seem to be the result of transfer of entire particles from the P-face. The reason for appearance or nonappearance of E-face pebbling is unclear; it is possible that the structure underlying this particulation is often obscured by the thickness of the replica or by the angle of metal deposition. Thus far no close correlation has been possible between P-face particles and the E-face pebbling. It is difficult to determine if E-face pebbling is due only to elevated particles or is in part representative of minute indentations. Cytoplasmic filaments are replicated distinctly and with less distortion in fixed preparations (see also McNutt, ’76). This is true of both intermediate filaments of 10-12 nm diameter and finer microfilaments (less than 8 nm), which occupy desmosomal as well as other peripheral regions of the epithelial cells studied (Figs. 1, 5 and 6). Interestingly, both categories of filaments appear as elevations in the replicas and never as pits, a behavior in the freeze-fracture process that is not unlike that shown by myosin filaments when skeletal muscle is fractured (Bullivant et al., ’72). Apparently, during fracturing such filaments are plucked slightly out of the frozen cytoplasm on both faces of the fracture. Moreover, the resulting images do not show precise alignment of filaments on the opposite fracture faces, indicating that distortion produced by fracturing is random. These are particularly evident when complementary replicas are studied (Fig. 2). Groups of intermediate filaments (commonly identified as tonofilaments in epidermis) make looping excursions past the cytoplasmic plaques of desmosomes or hemidesmosomes. It does not appear that these tonofilaments terminate a t the membranes or plaques of desmosomes or hemidesmosomes. Rather, a second population of filament appears to intervene between tonofilament loops and desmosomal or hemidesmosomal membranes. These “traversing” filaments are clearly visible in the present replicas of fixed desmosomes as narrow aggregated elevations, each somewhat less than 8 nm in diameter (Figs. 2, 5 and 6). Like tonofilaments, each traversing filament appears as an elevation, even in complementary replicas, and corresponding depressions are not observed. Profiles of traversing filaments are distinct and regular in these preparations. Though elevated from the fracture surface, they do not display the excessive plastic deformity that has been encountered in the same areas of unfixed material (Kelly and Shienvold, ’76). As in those previous studies, traversing filaments are occasionally seen to extend from the cytoplasm in the desmosomal plaque region onto the E-face of a split membrane (see Figs. 1 and 2). Such images may represent exposed, but not displaced, insertions of the traversing filaments into the membrane interior. Alternately, they may be severed and extruded filament tips that fall back on the replica surface after the fracture process. More important, the present studies disclose images in which a given traversing filament profile can be seen to extend through the cross-fractured cytoplasmic leaflet to end in relationship to a P-face ridge or particle (Figs. 5, 6 and 7). Images observed to date do not disclose with certainty whether traversing filaments extend beyond the particles to penetrate completely the desmosomal membrane. Nor is their nature within the plaque region, or the architecture of their attachment to tonofilament loops, yet discernible. Fig. 6. Higher magnification stereo-pair depicting freeze-fractured desmosomal membrane faces (P and E). The P-face particulation is typical, and a faint indication of E-face pebbling is apparent in this replica. Traversingfilaments (TR)are visible near and within the cross-fractured desmosomal cytoplasmic membrane leaflet. Stereo-imaging reveals the close relationship or continuity that exists between the P-face desmosomal ridges and particles and the traversing filaments. Linear images within the intercellular space appear to reach the P-face (arrows).Such images are suggestive of filamentous components extending into or across the intercellularspace, but in our studies these have been seen with less frequency than images of traversing filaments penetrating from the cytoplasm to join the P-face particles. X 72,700 Fig. 7. High magnification stereo-pair showing a free%-fractured desmosome in which both cytoplasmic and intramembranous components are visible in each of the participant cells. Typical desmosomal P-face particulation is apparent within the membrane of the cell on the left. Traversing filaments (TR)extend from the region of tonofilaments(TO) and in some instances (arrows) can be traced across the fractured cytoplasmic leaflet where they display continuity with P-face P-face; (E), E-face. x 122,000 particles. (P), DESMOSOMAL TRAVERSING FILAMENTS 9 10 DOUGLAS E. KELLY AND AILEEN M. KUDA Some of the replicas provide images suggesting a filamentous structure within the intercellular desmosomal gap (see Figs. 4,5 and 6). Occasionally these extracellular linear profiles appear to penetrate the extracellular leaflet and end on the P-face. The images examined to date, however, do not reveal a special relationship of these filaments with the P-face particles. The basalmost cells of larval newt epidermis display prominent hemidesmosomes. These can often be studied in the same replicas in which desmosomes appear. As in the case of desmosomes, the clarity of hemidesmosomal components has been improved with the newer techniques. In our prior studies (Shienvold and Kelly, '76) some larval newt epidermal hemidesmosomal intramembranous particles were noted to be much larger than the population in desmosomes. These 20-30 nm particles were observed to cling with equal frequency to P- and E-facesin fixed preparations, while adhering to the P-face almost exclusively in unfixed material (a pattern unlike that of desmosomes).The present observations of fixed preparations confirm the size difference for most hemidesmosoma1 particles, but comparisons of complementary replicas (Fig. 8) suggest a slightly different interpretation regarding distribution. The largest (20-30 nm) hemidesmosomal particles are quite irregular in shape, are somewhat widely separated and tend to cling almost exclusively to the P-face in these fixed preparations. The E-face hemidesmosomal territory does disclose particles, but for the most part these are smaller, even more widely separated, and quite similar to the 10-15nm desmosomal particles. Moreover, close inspection reveals that the smaller particles are also found on the P-face territory scattered among the large particles. In addition, the hemidesmosomal P-face displays numerous longer, 10 nm caliber, ridgeor thread-like profiles that also resemble those of the desmosomal P-face. These are rare on the E-face. The hemidesmosomal E-face also displays a somewhat roughened or pitted background, the depressions of which do not precisely correlate withparticles, large or small, of the complementary P-face. Like the situation in desmosomes, traversing filaments are a n integral component of hemidesmosomes, spanning the region between tonofilament bundles and the hemidesmosomal membrane. However, a relationship of traversing filaments to the special large particles of the hemidesmosomes is not indicated clearly in the images obtained thus far. Nor have we yet been able to demonstrate a connectionbetween traversing filaments and the smaller particles seen widely scattered on both P- and E-face hemidesmosomal territories. DISCUSSION Desmosomes and hemidesmosomes a r e highly complex structures in their intramembranous and cytoplasmic organization. While much remains to be worked out concerning the details of this complexity, some new insight is provided by the techniques and images described here. First of all, the intramembranous organization within desmosomal territories is structurally more complicated than generally perceived. Our findings confirm previous observations concerning the characteristic 10-15 nm Fig. 8. Complementaryreplicas (top and bottom) shown in stereo-pairs and depicting hemidesmosomal territories along the basalmost cell membrane of larval newt epidermis. Abundant tonofilaments (TO)are seen in the cleaved cytoplasm to the right. Characteristic large (20-30 nm) irregularly shaped particles are loosely clustered in each of many hemidesmosomal domains distributed over the P-face (P). Smaller (10-15 nm) particles are seen amongthe larger ones and also on the complementary E-face hemidesmosomalterritories. Elongate linear profiles are also found on the P-face (arrows), in some instances in close proximity to the large P-face particles. The hemidesmosomal domains of the complementary E-face (E) display a dispersed, finer particulation as well as an array of pits in these fixed preparations. It has not been possible to relate the pits precisely to any particulate components of the complementary P-face. Traversing filaments (TR)are bunched in the cytoplasm, where they approach the hemidesmosomal membrane. The caliber of traversing filaments is characteristically about one-half that of tonofilaments. x 70,000. Hemidesmosomal components are illustrated in the above micrograph of sectioned basal epithelial cell surface and underlying connective tissue. The approximate fracture route observed in the replicas is indicated by the dotted line. DESMOSOMAL TRAVERSING FILAMENTS 11 12 DOUGLAS E. KELLY AND AILEEN M. KUDA irregular desmosomal intramembranous particles. They are closely packed and, in larval newt epidermis, cling almost exclusively to the P-face in fixed material and remain with the E-face when fixation is omitted (Kelly and Shienvold, '76). In addition, however, a bed of finer particulation can be found on the E-face of fixed desmosomes that should not be confused with cases of E-face particulation previously reported by other authors. Breathnach, et al. ('761, for example, noted that the typical 10-15 nm desmosomal particles are found scattered over both fracture faces in replicas of unfixed rat buccal epithelium, but their published images do not reveal the finer E-face pebbling noted here. Leloup et al. ('79), studying fixed calf muzzle stratum spinosum, described typical particles on desmosomal E-faces that were similar to, though less prominent than, those seen on the complementary P-faces (see also Ishimura et a]., '79; Pauli et al., '79). Thus far, the fine E-face particles seen in our study do not seem to display any complementary relationship to the larger desmosomal particles of the opposite face. They are seen to best advantage in thin replicas where platinum deposition is minimal and its angle is shallow, a relatively infrequent combination of circumstances. Thicker replication probably obscures the finer details so that the desmosomal E-face is similar to the surrounding territory. Caputo and Peluchetti ('77) reported a n elevated "plaque" over E-faces of fixed desmosomes of normal human epidermis, a feature limited to the stratum corneum. This plaque bears only superficial resemblance to the non-elevated E-face particulation reported here. Moreover, the replicas in the Caputo and Peluchetti study appear rather thick and hence are not apt to reveal the E-face particulation, if present, in their system. No functional significance for the finer E-face particulation can be adduced from our present evidence. However, one possibility is that they reflect sites of filament penetration to the extracellular space, a concept to be discussed further below. The present observations, combined with those ofMcNutt and Weinstein ('73) and Leloup et al. ('791, lead to a greater confidence in the reality of traversing filaments as an essential structural component of both desmosomes and hemidesmosomes. Traversing filament images were not obvious in our prior studies of fixed larval newt epidermis (Kelly and Shienvold, '76). With the fixation techniques now employed, the filaments are crisp and regular. Moreover, they are now revealed with greater precision than in our prior studies utilizing unfixed material, wherein small-caliber filament images were apparent, but equivocal, due to plastic distortion, which exaggerated their dimensions and obscured their directions. The traversing filaments regularly display a smaller caliber (5-8 nm) compared to nearby intermediate filaments (10-12 nm), and they occupy the region of the desmosomal plaque, although the plaque itself is not distinct in freeze-fracture preparations (see also Caputo and Peluchetti, '77). The traversing filaments appear to pierce the cytoplasmic leaflet of the cell membrane within the junctional territory; this is especially clear in desmosomal replicas where the cytoplasmic leaflet has been crossfractured. (Figs. 5, 6 and 7). More important, after penetration, the filaments appear to fuse with desmosomal particles (Fig. 7). It now seems likely that the irregular shapes of the desmosomal intramembranous particles are a reflection of their contiguity with inserted traversing filaments. Significantly, the smallest diameter of the particles (10 nm) is only slightly larger than the diameter of traversing filaments. Thus, the available evidence appears to favor the earlier suggestion regarding an essential role for the desmosomal particles in the anchorage of the traversing filaments into the desmosomal cell membrane (McNutt and Weinstein, '73; Skerrow and Matoltsy, '74; and Staehelin, '74). Alternatively, Leloup et al. ('79) have suggested that traversing filaments penetrate the cell membrane, but en route they attach to the cytoplasm-facingsurface of the cytoplasmic leaflet, forming a small bulge on the P-face, the desmosomal particle. This interpretation seems to point up the difficulty in discerning accurately whether it is the intramembranous particle or the traversing filament that pierces the cytoplasmic leaflet, or if the two components are in reality parts of the same entity. In our images, we interpret the continuity of the traversing filaments through the cytoplasmmembrane interface plus the elongate nature of many of the desmosomal particles as favoring the incorporation of the filament tips into the membrane and a n intramembranous association with the particles. The degree to which desmosomal particles may reach the cytoplasmic or extracellular surfaces of the membrane is unclear (see Lempert and Elias, '79). The cytoplasmic extent reached by a given traversing filament also remains problematical. Staehelin and Hull ('78) have generalized that the intermediate filaments of DESMOSOMAL TRAVERSING FILAMENTS desmosomes loop through the cytoplasmic desmosomal plaque where they are in some manner associated with traversing filaments (which they term “linkers”).In newt epidermal desmosomes, tonofilament loops occur for the most part 20-40 nm away from the desmosomal plaque, a structure seen best in sectioned material (see micrograph accompanying Fig. 1; Kelly, ’66).In our freeze-fracture replicas, the margin of the zone rich in traversing filament profiles is found in association with the intermediate filament loops. Therefore, the traversing filaments must extend beyond the desmosomal plaque to reach the level of intermediate filaments; the substance of the plaque must be in part traversing filaments and in part another material that is packed around the filaments and not revealed in these preparations. The mode of attachment between intermediate filaments and traversing filaments is not clarified by the present evidence. In their study of calf muzzle epidermis, Leloup ekal. (‘79)have described profiles, roughly equivalent to our traversing filaments, from freeze-fracture images in which considerable etching is apparent. They related these images to prior biochemical studies of isolated desmosomal complexes (Drochmans et al., ’78), suggesting that intermediate filaments are composed of subfilaments surrounding a hollow central core. The tips of broken intermediate filaments in such preparations appear unraveled. Leloup et al. (‘79) have therefore interpreted the traversing filaments to be unraveled protofilaments from terminating intermediate filaments. This model is difficult to reconcile with the distinct non-terminating, non-raveling looping profiles of intermediate filaments that have been seen clearly in stereo-electron microscopic studies of sectioned newt epidermal desmosomes (Kelly, ’66).Moreover, definitive biochemical evidence as to the molecular structure of any of the several known types of intermediate filaments is yet lacking. In any case, it remains unclear how tensile forces might be transmitted from the cytoskeletal intermediate filaments to traversing filaments a t desmosomes. Similarly, the question of the mechanism by which adjacent membranes of the desmosomal system are interconnected across the intercellular space is unresolved. Staehelin and Hull (’78)envisioned individual traversing filaments (“linkers”) as extending from the cytoplasmic desmosomalplaque of one cell to that of the next, penetrating both leaflets of both cell membranes and incorporating a lattice-like array along the intercellular midline. In their 13 diagram a shorter form of traversing filament appears to attach intermediate filaments to the plaques. Leloup et al. (‘791, by contrast, diagrammed a hypothetical midline population of extracellular particles from which a second, extracellular type of thin filament radiates and extends from the extracellular space through the membrane leaflets, connecting en route with the desmosomal particles. Shimono and Clementi (‘77) have postulated that desmosomal external material is in the form of “side arms” that radiate from E-face particles within the membrane. Since this moiety is trypsin sensitive, unlike the mechanism that attaches tonofilaments to the membrane (see also Borysenko and Revel, ’73),the evidence favors the concept of different proteins/filaments on the two sides of each desmosomal membrane. Our images do not permit development of a more definitive model on this question, but they do provide evidence that cytoplasmic filaments penetrate the cytoplasmic leaflet and relate within the membrane to the desmosomal intramembranous particles. Some of the images of cross-fractured desmosomal external leaflets suggest also the possibility of filament penetration of that leaflet and a filamentous character for extracellular bridging material. These images are infrequent and less distinct compared to images of traversing filaments penetrating the cytoplasmic leaflet. In support of the idea of insertion of extracellular filaments, the fine “pebbling” noted on the desmosomal E-face and described above might be interpreted as a consequence of the external leaflet being penetrated. There is scant firm evidence published thus far to indicate that a single filament courses through both leaflets, much less that the same continuous filament is shared across the membranes of both adherent cells. Although this remains a possibility, we deem it at least equally feasible that different filamentous proteins are involved with intercellular linkage on the one hand and as an intermediate filamentto-cell membrane linkage on the other. This concept is supported by the observation that hemidesmosomes of the same epithelium display markedly different, larger intramembranous particles in addition to the 10-15 nm ones. If one assumes that the attachment of intermediate filaments to the cell membrane requires a mechanism in hemidesmosomesthat is similar to that in desmosomes, it follows that the larger hemidesmosomal intramembranous components might be more related to the function of adhesion between cell membrane and 14 DOUGLAS E. KELLY AND AILEEN M. KUDA extracellular connective tissue components (basal lamina, collagen, etc.). Then it is plausible to relate the 10-15 nm particles of both desmosomes and hemidesmosomesto the insertion of cytoplasmic traversing filaments, and the 20-30 nm hemidesmosomal particles to insertion of extracellular attachment proteins peculiar to the hemidesmosomes. Two hemidesmosomal filament protein species should logically be involved; one (the traversing filaments) mediating attachment of intermediate filaments to the cell membrane; and the other (an extracellular vehicle) serving membraneto-connective tissue adhesion. Desmosomes, perhaps, incorporate a different extracellular vehicle that mediates membrane-to-membrane adhesion and does not involve a n intramembranous particle distinguishable from those connected to traversing filaments. This line of reasoning would be strengthened if a connection could be demonstrated between traversing filaments and small particles or ridges in hemidesmosomal membranes, a n aspect not accomplished in the current study or in prior work with unfixed hemidesmosomes (Shienvold and Kelly, '76). The above speculations and questions stress the desirability of efforts to define in biochemical terms the differences in the various filamentous and particulate components of desmosomes and hemidesmosomes now that some of their anatomic relationships are becoming more clearly understood. ACKNOWLEDGMENTS The authors are grateful to Dr. Richard L. Wood for critical review of the manuscript, and to Mrs. Delcina McMillan for expert secretarial assistance. This research was supported by a research grant (PCM 76-18755) from the National Science Foundation. LITERATURE CITED Borysenko, J.Z., and J.P. Revel (1973) Experimental manipulation of desmosome structure. Am. J . Anat., 137:403422. Breathnach, A.S., T. Goodman, C. Stolinski, and M. Gross (1973) Freeze-fracture replication of cells of stratum corneum of human epidermis. J. Anat. (Lond.) 114:65-81. Breathnach, A.S., C. Stolinski, and M. Gross (1972) Ultrastructure of foetal and post-natal human skin as revealed by the freeze-fracture replication technique. Micron, 3:287-304. Breathnach, A.S., M. Gross, B. Martin, and C. Stolinski (1976) A comparison of membrane fracture faces of fixed and unfixed glycerinated tissue. J . Cell Sci., 2I:437-448. Brown, D., and V. Ilic (1979) Freeze-fracture differences in plasma membranes of the stratum corneum and replacement layer cells of amphibian epidermis. J. Ultrastruct. Res., 67:5&64. Bullivant. S.. D.G. Rams, W.S. Bertaud, J.P. Chalcroft, and G.F. Grayston (1972) Freeze-fractured myosin filaments. J. Cell Biol., 55:520-524. Caputo, R., and D. Peluchetti (1977) The junctions of normal human epidermis. A freeze-fracture study. J. Ultrastruct. Res., 61:44-61. Drochmans, P., C. Freudemtein, J.C. Wanson, L. Laurent, T.W. Kennan, J. Stadler, R. Leloup, and W. Franke (1978) Structure and biochemical composition of desmosomesand tonofilaments isolated from calf muzzle epidermis.J. Cell Biol., 79:427-443. Fawcett, D.W. (1961)Intercellular bridges. Exp. Cell Res., 8:174-187. Ishimura, K., K. Takagaki, K. Egawa, and H. Fujita (1979) Characteristicfigures of intercellularjunctions of the hagfish intestinal epithelium. Freeze fracture observation. Arch. Histol. Jpn., 42:41%422. Kelly, D.E. (1966) Fine structure of desmosomes, hemidesmosomes and an adepidermal globular layer in developing newt epidermis. J. Cell Biol., 28:51-72. Kelly, D.E., and F.L. Shienvold (1976)The desmosome:Fine structural studies with freeze-fracture replication and tannic acid staining of sectioned epidermis. Cell Tiss.Res., 172:30%323. Leloup, R., L. Laurent, M.F. Ronveaux, P. Drochmans,and J.C. Wanson (1979)Desmosomes and desmogenesis in the epidermis of calf muzzle. Biol. Cell., 34:137-152. Lempert, T., and P.M. Elias (1979) Intraepidermal cell surface fine structure: Preservation and examination at high resolution. Anat. Rec., 193:927-937. McNutt, N.S. (1976) Plastic deformation of actin in the production of filaments on the E faces of freeze-fractured plasma membranes. J. Cell., Biol., 70:188a. McNutt, N.S., and R.S. Weinstein (1973) Membrane ultrastructure at mammalian intercellular junctions. Prog. Biophys. Mol. Biol., 26:45-101. Pauli, B.U., K.E. Kuettner, and R.S. Weinstein (1979)Intercellular junctions in FANFT-induced carcinomas of rat urinary bladder in tissue culture: in situ thin-section, freeze-fracture, and scanning electron microscopy studies. J. Microsc., 115.971-282. Shienvold, F.L., and D.E. Kelly (1976) The hemidemnosome: New fine structural features revealed by freeze-fracture techniques. Cell Tim. Res., 172:28%307. Shimono,M., and F. Clementi (1977) Intercellular junctions of oral epithelium. 11.Ultrastructural changes in rat buccal epithelium induced by trypsin digestion. J . Ultrastruct. Res., 59:lOl-112. Skerrow, C.J., and A.G. Matoltsy (1974) Isolation of epidermal desmosomes. J. Cell Biol., 63:515-523. Staehelin, L.A. (1974)Structure and functionof intercellular junctions. Int. Rev. Cytol., 39: 191-283. Staehelin, L.A., and B.E. Hull (1978) Junctions between living cells. Sci. Am., 238:140-152.