A model for de novo synthesis and assembly of tight intercellular junctions. Ultrastructural correlates and experimental verification of the model revealed by freeze-fracture

The structure and function of intercellular tight (occluding) junctions, which constitute the anatomical basis for highly regulated interfaces between tissue compartments such as the blood-testis and blood-brain barriers, are well known. Details of the synthesis and assembly of tight junctions, however, have been difficult to determine primarily because no model for study of these processes has been recognized. Primary cultures of brain capillary endothelial cells are proposed as a model in which events of the synthesis and assembly of tight junctions can be examined by monitoring morphological features of each step in freeze-fracture replicas of the endothelial cell plasma membrane. Examination of replicas of non-confluent monolayers of endothelial cells reveals the following intramembrane structures proposed as 'markers' for the sequential events of synthesis and assembly of zonulae occludentes: development of surface contours consisting of elongate terraces and furrows (valleys) orientated parallel to the axis of cytoplasmic extensions of spreading endothelial cells, appearance of small circular PF face depressions (or volcano-like protrusions on the EF face) that represent cytoplasmic vesicle-plasma membrane fusion sites, which are positioned in linear arrays along the contour furrows, appearance of 13-15 nm intramembrane particles at the perimeter of the vesicle fusion sites, and alignment of these intramembrane particles into the long, parallel, anastomosed strands characteristic of mature tight junctions. These structural features of brain endothelial cells in monolayer culture constitute the morphological expression of: reshaping the cell surface to align future junction-containing regions with those of adjacent cells, delivery and insertion of newly synthesized junctional intramembrane particles into regions of the plasma membrane where tight junctions will form, and aggregation and alignment of tight junction intramembrane particles into the complex interconnected strands of mature zonulae occludentes. The distribution of filipin-sterol complex-free regions on the PF intramembrane fracture face of junction-forming endothelial plasmalemmae corresponds precisely to the furrows, aligned vesicle fusion sites and anastomosed strands of tight junctional elements.(ABSTRACT TRUNCATED AT 400 WORDS)     

Morphogenesis of tight junctions in the peritoneal mesothelium of the mouse embryo

The peritoneal mesothelium of mouse embryos (12 to 18 day of gestation) was studied by freeze-fracture and in sections in order to reveal the initial formation of the tight junctions. Freeze-fracture observations showed three types of tight junctions. Type I consists of belt-like meshworks of elevations on the P face and of shallow grooves on the E face. No tight junctional particle can be seen either on the elevations or in the grooves. Type II shows rows of discontinuous particles on the elevations on the P face. Type III consists of strands forming ridges on the P face. On the E face, the grooves of Type II and III appear to be narrower and sharper than those of Type I. Quantitatively, Type I junctions are most numerous during the early stages (day 12-13) of embryonic development, while Type III junctions become more common in the later stages, and are the only type seen by day 18. Observations on sections, however, fail to distinguish between the three types. The results suggest that an initial sign of tight junction formation is close apposition of the two cell membranes in the junctional domain, without tight junctional particles. Later, the particles appear to be incorporated in the tight junctions and the strands form by fusion of the particles.     

Junctional structures in digestive epithelia of a cephalopod

Intercellular junctions have been studied in the epithelia of digestive organs of Sepia officinalis (digestive gland, digestive duct appendages and caecum) by conventional staining, lanthanum tracer and freeze-fracturing techniques. In the three organs studied the same junctional complex occurs, consisting of a belt desmosome, a septate junction and gap junctions. The septate junction is of pleated-sheet type and the gap junction has its particles on the P face of the fracture. Circular structures have been found in the digestive gland septate junctions. Neither continuous nor tight junctions have been found. These results show that Cephalopods have junctional structures very close to those of other Molluscs and of Annelids. Some small differences between the septate junctions of the three organs could be related to their different physiology.     

The occluding junctions of mouse duodenal enterocytes during development

Summary The architecture of occluding junctions during the differentiation of the mouse duodenum was studied in freeze-fractured material. Irregular zonulae occludentes (ZO) (Type I) are numerous during fetal life, and are characterized by their irregular width, and by the presence of basal open-ended extensions fused with the discontinuous basal strand of the ZO. Regular ZOs (Type II), typical of the adult villous epithelium, appear after Type I junctions by day 16 of gestation. Two patterns are distinguishable: in the first, parallel strands of ridges and furrows are found without crossing branches; in the second pattern, the junction zone is organized like a network of short branches forming various types of polygons. In fetal and adult mice fasciae occludentes (FO) (Type III) are present on the lateral cell membranes; in unfixed specimens particles are found in the furrows of the E-face and pits on the ridges of the P-face. In fixed tissues, the particles are aligned on the ridges of the P-face. These results indicate that fixation with glutaraldehyde modifies considerably the affinity of junctional particles toward the P-face during the fracture process. Moreover, the presence of numerous large FOs on the lateral cell membranes of enterocytes during late fetal life and in the adult, is possibly related to cell movement along the intestinal villi.     

A quantitative study of intramembrane changes during cell junctional breakdown in the dystrophic rat retinal pigment epithelium

Previous electron microscope freeze-fracture and tracer studies have revealed that intercellular junctions in the retinal pigment epithelium (RPE) of Royal College of Surgeons (RCS) rats with inherited retinal dystrophy [5] break down between three and six postnatal weeks [6, 7]. In this study quantitative computer techniques were used to analyze the freeze-fracture changes in the dystrophic RPE. The following parameters were measured: length of tight junctional strands/micron2; number of tight junctional strand anastomoses/micron2; number of gap junctional aggregates/micron2; area of gap junctional aggregates/micron2; and density of background intramembrane particles/micron2. At three postnatal weeks, the dystrophic junctional complex membrane is similar to normal, but at 10 weeks and later there are dramatic decreases in tight junctional strand length/micron2 and number of anastomoses/micron2, as well as in the number/micron2 and area of gap junctions/micron2, while the density of background particles/micron2 is dramatically increased. Correlational analysis revealed that changes in gap and tight junctions were significantly related to each other and to the increase in background particle density. The diameter of background particles within the normal and post-breakdown dystrophic junctions was measured in order to see whether the dispersal of gap and tight junctional particles (8-10 nm) into the surrounding membrane contributes to the increased particle density. These measures showed that background particles in all size ranges were more numerous in the dystrophic RPE, but that the largest increase was in the smallest diameter particles (6-7 nm). Thus, while gap and tight junctional sized particles contribute to the increase, particles from other sources may also be involved. Particle density of apical and basal membranes in the normal and in the 10 week and older dystrophic RPE was analyzed to study the effects of tight junctional breakdown on the distribution of intramembrane particles. These measures showed that particle density was greater basally than apically in the normal RPE and that particle density in both membranes decreased slightly in the dystrophic RPE, but that their ratio remained unchanged. It has been shown previously that even a single intact tight junctional strand is sufficient to maintain differences in particle density between apical and basal surfaces [14, 15] and in the majority of abnormal dystrophic junctional complexes at least one tight junctional strand remains intact.(ABSTRACT TRUNCATED AT 400 WORDS)     

Relationship of sertoli-sertoli tight junctions to ectoplasmic specialization in conventional and en face views

Ectoplasmic specializations are actin filament-endoplasmic reticulum complexes that occur in Sertoli cells at sites of intercellular attachment. At sites between inter-Sertoli cell attachments, near the base of the cells, the sites are also related to tight junctions. We studied the characteristics of ectoplasmic specializations from six species using conventional views in which thin sections were perpendicular to the plane of the membranes, we used rare views in which the sections were in the plane of the membrane (en face views), and we also used the freeze-fracture technique. Tissues postfixed by osmium ferrocyanide showed junctional strands (fusion points between membranes) and actin bundles, actin sheets, or both, which could be visualized simultaneously. En face views demonstrated that the majority of tight junctional strands ran parallel to actin filament bundles. Usually, two tight junctional strands were associated with each actin filament bundle. Parallel tight junctions were occasionally extremely close together ( approximately 12 nm apart). Tight junctional strands were sometimes present without an apparent association with organized actin bundles or they were tangential to actin bundles. En face views showed that gap junctions were commonly observed intercalated with tight junction strands. The results taken together suggest a relationship of organized actin with tight junction complexes. However, the occasional examples of tight junction complexes being not perfectly aligned with actin filament bundles suggest that a precise and rigidly organized actin-tight junction relationship described above is not absolutely mandatory for the presence or maintenance of tight junctions. Species variations in tight junction organization are also presented.     

Definitive Evidence for the existence of tight junctions in invertebrates

Extensive and unequivocal tight junctions are here reported between the lateral borders of the cellular layer that circumscribes the arachnid (spider) central nervous system. This account details the features of these structures, which form a beltlike reticulum that is more complex than the simple linear tight junctions hitherto found in invertebrate tissues and which bear many of the characteristics of vertebrate zonulae occludentes. We also provide evidence that these junctions form the basis of a permeability barrier to exogenous compounds. In thin sections, the tight junctions are identifiable as punctate points of membrane apposition; they are seen to exclude the stain and appear as election- lucent moniliform strands along the lines of membrane fusion in en face views of uranyl-calcium-treated tissues. In freeze-fracture replicas, the regions of close membrane apposition exhibit P-face (PF) ridges and complementary E-face (EF) furrows that are coincident across face transitions, although slightly offset with respect to one another. The free inward diffusion of both ionic and colloidal lanthanum is inhibited by these punctate tight junctions so that they appear to form the basis of a circumferential blood-brain barrier. These results support the contention that tight junctions exist in the tissues of the invertebrata in spite of earlier suggestions that (a) they are unique to vertebrates and (b) septate junctions are the equivalent invertebrate occluding structure. The component tight junctional 8- to 10-nm-particulate PF ridges are intimately intercalated with, but clearly distinct from, inverted gap junctions possessing the 13-nm EF particles typical of arthropods. Hence, no confusion can occur as to which particles belong to each of the two junctional types, as commonly happens with vertebrate tissues, especially in the analysis of developing junctions. Indeed, their coexistance in this way supports the idea, over which there has been some controversy, that the intramembrane particles making up these two junctional types must be quite distinct entities rather than products of a common precursor.     

A re-assessment of the tricellular region of epithelial cell tight junctions in trachea of guinea pig

The tricellular region of epithelial tight junctions was previously dismissed as a possible avenue of permeability. One reason was that the two parallel vertical fibers, which penetrate the depth of the tight junction, were apparently cross-linked. Another reason was that the tricellular region of the tight junction is deeper than the adjacent bicellular regions. In the course of several freeze-fracture studies of epithelial tight junctions we have made observations which led us to re-assess the tricellular region as an avenue of permeability. We believe that information from ectoplasmic fracture faces is less subject to artifacts and, in ectoplasmic fracture faces of tricellular regions, cross-linking of the vertical furrows has not been observed. In guinea pig tracheal epithelium the tricellular junction is only about 1 micron deep. Following exposure to cigarette smoke, lanthanum ion staining has been observed in some tricellular junctions. It seems that earlier reasons for dismissing tricellular regions of the tight junction as permeability sites may be insufficient and that there is some evidence to support a role in permeability.     

Carbon tetrachloride induced proliferation of tight junctions in the rat liver as revealed by freeze-fracturing

Application of carbon tetrachloride produced a progressive proliferation of tight junctions in the rat liver. This system proved to be rapid and highly reproducable and affords the opportunity for tracing the fate of tight junctions in freeze-fracture replicas, facilitating investigations on their formation and function. Beginning on day one carbon tetrachloride treatments resulted in the progressive loosening and fragmentation of the junctional meshwork. After three to four days the membrane outside the zonulae occludentes was extensively filled with proliferated discrete junctional elements often forming complex configurations. From the fifth day on the zonulae occludentes were restricted again predominantly around the bile canaliculus margins. But the junctional meshwork of the zonulae occludentes remained loosened in comparison to those in the control rats. It could be shown that tight junction proliferation on the lateral surface of the plasmalemma occurred both through de novo formation from discrete centers of growth by addition of intramembranous particles and through reorganization of preexistent junctional strands of the fragmented zonulae occludentes bodies. Whereas the large gap junctions close associated with the zonulae occludentes remained more or less unaffected during the experiments, small gap junctions increased in number after five days and were located at the margin or in the tight junction domain. It is assumed that the degeneration of the tight junctions served as a pool for intramembranous particles which form the gap junctions. The results of these observations are discussed in relation to those obtained in other systems.     

Formation of tight and gap junctions in the inner enamel epithelium and preameloblasts in human fetal tooth germs

Human fetal primary tooth germs in the cap stage were fixed with a glutaraldehyde-formaldehyde mixture, and formative processes of tight and gap junctions of the inner enamel epithelium and preameloblasts were examined by means of freeze-fracture replication. Chains of small clusters of particles on the plasma membrane P-face of the inner enamel epithelium and preameloblasts were the initial sign of tight junction formation. After arranging themselves in discontinuous, linear arrays in association with preexisting or forming gap junctions, these particles later began revealing smooth, continuous tight junctional strands on the plasma membrane P-face and corresponding shallow grooves of a similar pattern on the E-face. Although they exhibited evident meshwork structures of various extents at both the proximal and distal ends of cell bodies, they formed no zonulae occludentes. Small assemblies of particles resembling gap junctions were noted at points of cross linkage of tight junctional strands; but large, mature gap junctions no longer continued into the tight junction meshwork structure. Gap junctions first appeared as very small particle clusters on the plasma membrane P-face of the inner enamel epithelium. Later two types of gap junctions were recognized: one consisted of quite densely aggregated particles with occasional particle-free areas, and the other consisted of relatively loosely aggregated particles with particle-free areas and aisles. Gap junction maturation seemed to consist in an increase of particle numbers. Fusion of gap junctions in the forming stage too was recognized. The results of this investigation suggest that, from an early stage in their development, human fetal ameloblasts possess highly differentiated cell-to-cell interrelations.     

Structural basis for some permeability properties of the air--blood barrier.

The structural basis for the permeability of the alveolar-capillary membrane to water-soluble solutes rests in part on the structure and function of its intercellular junctions and the pinocytotic vesicles within its cells. Intercellular junctions between endothelial cells of the pulmonary capillary bed differ both in permeability to enzyme tracers and in their structure. As determined by freeze fracture, the junctions in the arteriolar, capillary, and venular portion of the capillary network vary in complexity, and in the number of rows of particles constituting the junction. Because there are few particles associated with the junctions in the venular end of the capillary bed, these are considered to be the most permeable of the three types of vascular junctions. Epithelial junctions, in contrast, are impermeable to all enzyme tracers studied, and they are composed of a continuous, complex network of junctional fibrils. While intercellular junctions form seals of varying 'tightness,' pinocytotic vesicles provide a means for the transport of water-soluble macromolecules across the alveolar-capillary membrane.     

The structure and function of the smooth septate junction in a transporting epithelium: The malpighian tubules of the new zealand glow-worm Arachnocampa luminosa

Junctional complexes between the epithelial cells in the four distinct regions of the glow-worm Malpighian tubule were investigated by electron microscopy using thin sectioning, freeze-fracturing, osmotic disruption and tracer techniques. The lateral plasma membranes of all four cell types are joined by smooth septate junctions but the extent of the complex across the cell depth varies in the four different regions. The width of the septa, the interseptal spacing and the separation between the outer leaflets of the adjacent plasma membranes are different for each cell type. Gap junctions were identified only in the junctional complex between Type IV cells and were intercalated amongst large lateral sinuses. In oblique sections of lanthanum infiltrated tissue, the electron-lucent septa at the basal side of the junction are outlined by the tracer as it penetrates. In the Junctional complexes of all four regions the septa appear as short, distinct, linear bars. In tangential sections of gap junctions between Type IV cells, the junctions appear as a hexagonal array of intermembrane particles with a centre to centre spacing of 18 nm. Horseradish peroxidase did not penetrate the junctional complexes very far but readily passed through the basal lamina into the spaces between extracellular invaginations of the basement membrane of the cells. Junctional complexes in all four areas of the tubule have similar freeze-fracture faces. In freeze-fracture replicas of fixed tissue continuous ridges of fused particles are seen on the P face and complementary furrows are found on the E face. Junctional response to osmotically adjusted Ringer solutions was similar in all four cell types. Distortion or ‘blistering’ of the intercellular space between the septa of the junction occurred when the tissue was bathed in or injected with a hypertonic Ringer solution. The structure of these junctions, visualized by the different techniques, and the role of the septate junction in a transporting epithelium, are discussed.     

Structural domains of the tight junctional intramembrane fibrils.

Freeze-fracture reveals intramembrane fibrils lying along the intermembrane contacts that characterize tight junctions. Tight junctions from a variety of species are reexamined here by rapid freezing prior to freeze-fracture. The tight junction fibril is uprooted alternatively from either the cytoplasmic or the exoplasmic hemibilayer during freeze-cleavage, exposing two distinct but complementary views of its hybrid structure within the same replica. When the transmembrane fibril is uprooted from the exoplasmic hemibilayer it appears on the P-fracture face as a smooth-surfaced cylinder which is sometimes resolved into periodic globular structures. The lack of indication that the P-face cylinder has been pulled out through the opposite membrane half indicates that this domain of the fibril is, in large part, buried in the hydrophobic interior of the membrane. However, when the transmembrane fibril is uprooted from the cytosolic hemibilayer it appears on the E-fracture face as a row of irregular intramembrane particles. The irregular particles on the E-face aspect of the fibril are interpreted as corresponding to transmembrane protein segments that may very well make projections onto the cytosolic surface of the bilayer. En face views of the outermost junction strand between adjacent epithelial cells show periodic lines on the bilayer on each side of the junction which are interpreted as periodic transmembrane protein segments arising from the core structure of the tight junction fibril. If the backbone of the tight junction strand is an inverted cylindrical micelle, it must typically include proteins, which might anchor it to structures outside the membrane bilayer.     

Tight junction formation is closely linked to the polar redistribution of intramembranous particles in aggregating MDCK epithelia

We have used freeze-fracture electron microscopy to investigate the relationship between the formation of the tight junction in the establishment of a differential distribution of intramembranous particles (IMPs) between the luminal and basolateral membranes of a canine kidney cell line (MDCK). This involves a characterization of the IMP distribution in these membranes in confluent monolayers of MDCK cells, in EGTA-dissociated cells, and in cells at various stages of reassociation. While normal confluent MDCK monolayer cultures exhibit tight junctions and an IMP differential distribution between the luminal and basolateral membranes, cultures dissociated with EGTA lose both formed tight junctional elements and the differential IMP distribution. We have also found that as tight junctions reform between reaggregating MDCK cells, intramembranous particles appear to rapidly redistribute with respect to them. An asymmetric distribution of these particles in the luminal and basolateral membranes is eventually achieved. Tight junction formation appears so closely linked to the genesis of IMP polarity that at early time points when only a string of tight junctional components spans the junctional zone, differential IMP distributions are seen. Thus, our dissociation studies suggest a close relationship between the integrity of the tight junction and the maintenance of IMP polarity between the luminal and basolateral membranes, while cell reassociation studies suggest that the tight junction may be functionally linked to the genesis of IMP polarity.     

Correlation of tight junction morphology with the expression of tight junction proteins in blood-brain barrier endothelial cells

Endothelial cells of the blood-brain barrier form complex tight junctions, which are more frequently associated with the protoplasmic (P-face) than with the exocytoplasmic (E-face) membrane leaflet. The association of tight junctional particles with either membrane leaflet is a result of the expression of various claudins, which are transmembrane constituents of tight junction strands. Mammalian brain endothelial tight junctions exhibit an almost balanced distribution of particles and lose this morphology and barrier function in vitro. Since it was shown that the brain endothelial tight junctions of submammalian species form P-face-associated tight junctions of the epithelial type, the question of which molecular composition underlies the morphological differences and how do these brain endothelial cells behave in vitro arose. Therefore, rat and chicken brain endothelial cells were investigated for the expression of junctional proteins in vivo and in vitro and for the morphology of the tight junctions. In order to visualize morphological differences, the complexity and the P-face association of tight junctions were quantified. Rat and chicken brain endothelial cells form tight junctions which are positive for claudin-1, claudin-5, occludin and ZO-1. In agreement with the higher P-face association of tight junctions in vivo, chicken brain endothelia exhibited a slightly stronger labeling for claudin-1 at membrane contacts. Brain endothelial cells of both species showed a significant alteration of tight junctions in vitro, indicating a loss of barrier function. Rat endothelial cells showed a characteristic switch of tight junction particles from the P-face to the E-face, accompanied by the loss of claudin-1 in immunofluorescence labeling. In contrast, chicken brain endothelial cells did not show such a switch of particles, although they also lost claudin-1 in culture. These results demonstrate that the maintenance of rat and chicken endothelial barrier function depends on the brain microenvironment. Interestingly, the alteration of tight junctions is different in rat and chicken. This implies that the rat and chicken brain endothelial tight junctions are regulated differently.     

Vitamin-A-induced mucous metaplasia. An in vitro system for modulating tight and gap junction differentiation

Stratified squamous epithelia from 14-day chick embryo shank skin contain rare tight-junctional strands and only small gap junctions. Exposure of this tissue to retinoic acid (vitamin-A) (20 U/ml) in organ culture, however, induces mucous metaplasia, accompanied by tight-junction formation and gap-junction growth; untreated specimens continue to keratinize. To investigate sequential stages of junctional assembly and growth, we examined thin sections and freeze-fracture replicas at daily intervals for 3 days. During the metaplastic process, tight junctions assemble in midepidermal and upper regions, beginning on day 1 and becoming maximal on day 3. Two tight-junctional patterns could be tentatively identified as contributing to the emergence of fully formed zonulae occludentes: (a) the formation of individual ridges along the margins of gap junctions; (b) de novo generation of continuous ramifying strands by fusion of short strand segments and linear particulate aggregates near cellular apices. Gap junction enlargement, already maximal at day 1, occurs primarily three to four cell layers deep. Growth appears to occur by annexation of islands of 20-40 8.5-nm particles into larger lattices of islands separated by particle-free aisles. Eventually, a single gap junction may occupy much of the exposed membrane face in freeze-fractured tissue, but during apical migration of the cells such junctions disappear. The vitamin- A chick-skin system is presented as a responsive model for the controlled study of junction assembly.     

Freeze-fracture studies of frog atrial fibres.

The freeze-fracturing technique was used to characterize the junctional devices involved in the electrical coupling of frog atrial fibres. These fibres are connected by a type of junction which can be interpreted as a morphological variant of the "gap junction" or "nexus". The most characteristic features are rows of 9-nm junctional particles forming single or anastomosed circular profiles on the inner membrane face, and corresponding pits on the outer membrane face. Very seldom aggregates consisting of few geometrically disposed 9-nm particles are found. The significance of the junctional structures in the atrial fibres is discussed, with respect to present knowledge about junctional features of gap junctions in various tissues, including embryonic ones.     

The thin limbs of Henle's loop in the rabbit

Summary The thin limbs of short and long loops of Henle of the rabbit kidney were studied by freeze fracture techniques. According to TEM studies of thin sections four segments are discernible: descending thin limbs of short loops, descending thin limbs of long loops, subdivided into an upper and a lower part, and ascending thin limbs (Kaissling and Kriz 1979). This division is supported by findings obtained with the freeze fracture technique and based on differences in the organization of the junctional complexes as well as on differences in the internal morphology of the cell membranes. The descending thin limbs of short loops have junctional complexes established by several closely arranged junctional strands and numerous desmosomes. The upper parts of the long descending thin limbs have tight junctions consisting of a variable number of strands; their outstanding characteristic after freeze fracture is a high density of intramembrane particles in both luminal and baso-lateral membranes. The tight junctions of the lower part of the long descending thin limbs consist of several anastomosing junctional strands, which are, in contrast, loosely arranged; the cell membranes contain only a sparse population of intramembrane particles. The ascending thin limbs are characterized by shallow tight junctions (frequently consisting of only one single junctional strand). Moreover, the epithelial cells of this segment are heavily interdigitated; thereby the tight junctions are correspondingly lengthened.In addition, this study presents further evidence that remarkable species differences occur among thin limb epithelia. The junctional complexes of the long descending thin limbs of the rabbit are organized quite differently from those of small rodents (e.g., rat, Psammomys).The data of this study support the concept that the tight junctions are the main determinant of ionic conductances of the paracellular pathway. However, with reference to recent findings from microperfusion studies, it becomes obvious that a correlation of the junctional morphology with the transepithelial water permeability is lacking, at least for the thin limbs.This investigation was supported by the Deutsche Forschungsgemeinschaft; project Kr 546 Henlesche Schleife     

Structural integrity of hepatocyte tight junctions

The significance of discontinuities frequently found in freeze-fracture replicas of the tight junction was evaluated using complementary replicas of hepatocyte junctions from control and bile duct-ligated rats. An extensive analysis of complementary replicas using rotary platinum shadowing indicates that discontinuities in the protoplasmic (P) fracture face do not represent structural breaks in the tight- junctional network. In no case did P-face discontinuities correspond with interruptions in the groove network on the complementary extracellular (E) face. Quantitative analysis of replicas shows that P- face discontinuities result in part from "transfer" of material to the complementary E face (approximately 7% of the junctional length). However, many P-face discontinuities (7-30% of the junctional length) are matched only by a groove on the complementary E face. This finding demonstrates that a significant amount of material can be lost during freeze-fracture. An analysis of junctions from bile duct-ligated rats, which are known to have an increased paracellular permeability, shows comparable transfer and loss of material. However, the number of junctional elements and the tight-junction network density was significantly reduced by bile duct ligation. These observations indicate that discontinuities in tight-junctional elements result during the preparation of freeze-fracture replicas and are not physiologically important features of the junctional barrier. Variation in the number of elements provides the best explanation for observed differences in tight-junction permeability.