GASTROENTEROLOGY 1992;102:910-923

Barriers to Paracellular Esophageal Epithelium

Permeability

in Rabbit

ROY C. ORLANDO, ERIC R. LACY, NELIA A. TOBEY, and KATHRYN COWART Department of Medicine, University of North Carolina School of Medicine, Chapel Hill, North Carolina; and Department

of Anatomy and Cell Biology, Medical University

Morphological and electrophysiological techniques were used to define the location and nature of the barriers to diffusion across the intercellular space (paracellular pathway) of rabbit esophageal epithelium. Transmission electron microscopy and light microscopy coupled with histochemistry identified a series of tight junctions and an intercellular material staining positively for neutral and acidic glycoconjugates as likely barrier candidates. Additional studies with lanthanum and horseradish peroxidase showed that the barrier to diffusion of tracers was present throughout the stratum corneum and extended to the upper three to seven layers of stratum spinosum and that these findings were most compatible with the presence of the intercellular glycoconjugate material but not the tight junctions. Further positive staining for carbohydrate moieties at the electron microscopic level with periodic acid-thiocarbohydrazide-silver proteinate sug gested that the glycoconjugate material was synthesized in the cells of the barrier layers and packaged in intracellular membrane-bound vesicles before secretion into the intercellular space. Although tight junctions were present in series within stratum corneum and, less commonly, extended to two to three cell layers of upper stratum spinosum, analysis of tracer studies, freeze-fracture replicas, electrophysiological data, and mannitol fluxes, while not conclusive, provided little to support a major role for these junctions in barrier function in this tissue. n important function of all gastrointestinal epithelia is to act as barriers to the passive diffusion of ions and molecules from lumen to blood.’ To accomplish this task, epithelia come structurally equipped with hydrophobic cell membranes to seal off, to variable extents, the transcellular route and intracellular (junctional) complexes between cells to further seal off the paracellular pathway. The nature of the junctional complex governing permeability through the paracellular pathway in the simple columnar epithelium-lined portions of the gastrointestinal tract is the tight junction,ls’but its counterpart in

A

of South Carolina, Charleston,

South Carolina

the stratified squamous epithelium-lined esophagus has not been well defined. In the present investigation we used both morphological and electrophysiological techniques to establish the location and nature of the structural barriers governing diffusion through the paracellular pathway of rabbit esophageal epithelium. Rabbit esophageal epithelium was chosen because of its structural and functional similarities to human esophageal epithelium and because it has been used extensively for studying the pathogenesis of acid-induced (reflux) esophagitis.3-6 Portions of the present study have previously appeared in abstract form.7 Materials and Methods Animals All experiments were performed with New Zealand white rabbits weighing approximately 8-9 lb and having free access to food and water before use. Animal protocols were reviewed and approved by the institutional animal welfare committee.

Morphology Tissues for light and electron microscopy (EM) were fixed in one of the three following ways. 1. In situ. Rabbits were anesthetized by an IV injection of a 50:50 mixture of Valium (5 mg/mL; diazepam; Hoffman La Roche, Inc., Nutley, NJ) and pentobarbital (60 mg/ mL) before exposing the esophagus with a midline incision from mandible to midepigastrium. The exposed esophagus was ligated within the neck and just proximal to the esophagogastric junction. Fixatives appropriate for EM or light microscopy were injected by needle and syringe into the esophageal lumen until the organ was slightly distended. Fixative was also dripped on the exposed serosal surface, and the preparation was left in situ for 30-45 minutes. Then the animal was killed with an IV overdose of pentobarbital. The esophagus was removed and placed in the same fixative for 4-24 hours. 2. In vitro. Esophagi were obtained from rabbits killed by an IV overdose of pentobarbital. The organ was opened

0 1992 by the American

Gastroenterological 0016-5065/92/$3.00

Association

PERMEABILITY BARRIER IN ESOPHAGEAL EPITHELIUM 911

March 1992

longitudinally and then either cut into sections and placed in the appropriate fixative or stripped of its muscle layers, as described below, for mounting in the Ussing chamber, before being cut into sections and placed in fixative. To determine if regional differences existed in morphology, epithelium was divided into four segments of equal length for comparison of samples from different regions. 3. In vitro in Ussing chambers. Esophageal epithelium stripped of its muscle layers and mounted in Ussing chambers, as described below, had both mucosal and serosal bathing solutions drained and replaced with a fixative appropriate for either light microscopy or EM. Fixatives

and Tissue

Preparation

Electron Microscopy. The primary fixative consisted of 2.5% glutaraldehyde, 2% paraformaldehyde, and 0.1% picric acid in 0.1 mol/L cacodylate buffer.’ Tissues were cut into wedge-shaped pieces and remained in fixative for 3-48 hours. Tissues were rinsed in 0.1 mol/L cacodylate buffer and postfixed in either 1% 0~0, only at room temperature for 2 hours or 1% 0~0, plus 1% potassium ferrocyanide for 2 hours at 4°C in the dark.g Tissues postfixed in 0~0, only were washed in maleate buffer (pH 5.2) for 1 hour and treated with 1% uranyl acetate in maleate buffer (pH 5.2) for 1 hour followed by a l-hour maleate buffer wash. Tissues from both postfixation methods were dehydrated in a graded series of ethanols and embedded in Epon-araldite. Sections 0.25-0.5ym were stained with alkalinized toluidine blue for light microscopy. Thin sections were stained with a fresh mixture of saturated aqueous uranyl acetate followed by lead citrate. Light microscopy and transmission electron microscopy (TEM) were performed on an Olympus BH-2 (Lake Success, NY) and a JEOL 1200 EX, respectively. Freeze-fracture. Tissues immersed in the primary fixative as described above were rinsed in 0.1 mol/L cacodylate buffer and then soaked in a solution of 30% glycerol in Ringer’s solution for 1-2 hours. Specimens were rapidly frozen in either liquid freon 22 or propane, fractured in a Balzer’s freeze etch device (BAF-400T; Balzers; Hudson, NH), and shadowed with carbon-platinum. Before examination in the JEOL 1200 EX microscope, replicas of the fractured surface were freed from the adherent tissue by digestion in sodium hypochlorite, cleaned in chromic acid, and placed on formvar-coated grids (E. F. Fullam, Inc., Latham, NY). Histology. Tissues fixed in situ or in vitro (with or without mounting in Ussing chambers) were bathed in either Carnoy’s fixative, 1% calcium acetate formalin, or 1% calcium carbonate formalin. Carnoy’s tissue was fixed for 3 hours, transferred to 100% ethanol, and embedded as described below. Formalin tissue was fixed with 10% buffered formalin for 2-6 hours for immunohistochemistry studies and 18-24 hours for routine histochemistry. Tissues were transferred to 70% ethanol and processed for routine paraffin embedding. Sections were cut 6-8-pm thick, mounted on glass slides, and stained with Sudan black B and oil red 0 for lipids, Feulgen’s stain for DNA,” alcian blue at pH 2.5 for acidic (sulfated and nonsulfated)

glycoconjugates, alcian blue at pH 1.0 for sulfated glycoconjugates, and periodic acid-Schiff’s (PAS) for neutral glycoconjugates and glycogen. However, the latter was removed by combining PAS with diastase to digest the glycogen. Some tissue sections were incubated in staining solutions to visualize the glycoconjugates at the electron microscope level using the periodic acid-thiocarbohydrazide-silver proteinate (PATCH-SP) technique.” Extracellular tracer experiments. The EM tracers, horseradish peroxidase (HRP, Sigma Chemical Co., St. Louis, MO), 0.6-l%, or lanthanum chloride or lanthanum nitrate (Polysciences Inc., Warrington, PA), 1 mmol/L, were used to visualize permeation via the paracellular pathway. Tissues were exposed to tracer present in bathing solutions of normal or Ca’+-free Ringer’s solution or in EM fixative. Isolated tissue slices in vitro had both mucosal and serosal surfaces exposed to tracer simultaneously. In Ussing chamber-mounted tissue, only the mucosal or serosal surface was exposed to tracer present in Ringer’s solution or EM fixative from 30 minutes to 4 hours. Tissues from Ussing chambers were fixed for an additional 4-24 hours in tracer-free EM fixative. In some experiments, tissues mounted in Ussing chambers and paired by electrical resistance while in normal Ringer’s solution were exposed to tracer after thorough rinsing and replacement of both bathing solutions with Ca’+-free Ringer’s solution plus 5 mmol/L ethylene diamine tetraacetic acid (EDTA). In this instance, tracer was added to either the mucosal or serosal solution for l-4 hours before fixation for TEM in the chamber. Tissues exposed to HRP were either cut on a vibratome or cryosectioned and incubated in diaminobenzidine and 0~0, for visualization of tracer.” Electrophysiology Esophagi were obtained from rabbits killed by an IV overdose of pentobarbital. The organ was excised, opened, and pinned mucosal surface down in a paraffin tray containing ice-cold oxygenated normal Ringer’s solution. The submucosa was sharply dissected free of the underlying mucosa with a scalpel, a process that yielded a sheet of tissue consisting of stratified squamous epithelium and a small amount of underlying connective tissue. From this tissue, sections were cut and mounted as flat sheets between Lucite half-chambers with an aperture of 1.13 cm’ for measurements of potential difference (PD), short-circuit (Isc), and electrical resistance. In most experiments tissues were bathed with normal Ringer’s solution (with the following composition in mmol/L: NaC, 140; Cl-, 120; K+. 5; HCO,-, 25; Ca’+, 1.2; Mg’+, 1.2; HPO,‘-, 2.4; H,PO,-, 0.4), 297 mosmol/kg H,O, pH 7.5, when gassed with 95% 0,/5% CO, at 37’C. Luminal and serosal solutions were connected to calomel and Ag-AgCl electrodes with Ringer-agar bridges for measurements of PD and automatic short-circuiting of the tissue with a voltage clamp (World Precision Instruments, Inc., Sarasota, FL). Tissues were continuously short-circuited except for 5-lo-second periods when the open circuit PD was read. Resistance was calculated from the open-circuit PD and the Isc or from the current deflection to imposed voltage.

912 ORLANDO ET AL.

GASTROENTEROLOGY Vol. 102,No. 3

Figure 1. Low-power TEM showing the three epithelial strata of the rabbit esophagus. Note the distinct change in cell shape at the houndary between s. germinativum and spinosum and the loss of discrete nuclei in the s. corneum (original magnification X5250). In experiments to assess the contribution of tight junctions to epithelial permeability, electrical resistance was monitored in Ussing-chambered tissues before and during exposure to either (a) Ca’+-free Ringer’s solution (Ca’+ salts replaced by Mg2+ salts in normal Ringer’s) in the presence of 5 mmol/L EDTA, (b) hypertonic luminal solution of 1 mol/L mannitol, or (c) 5 umol/L cytochalasin B in both bathing solutions. Mannitol fluxes were performed in some experiments

to correlate the change in electrical resistance with a marker of paracellular permeability. This was performed by the addition of 10 mmol/L mannitol-Ringer’s solution and 10 pCi [‘4C]mannitol (ICN, Irvine, CA) to the luminal bath. After taking the “hot” side sample, 45 minutes were allowed for equilibration before sampling the serosal solution at two consecutive 45-minute intervals. Mannitol fluxes were determined using the counts obtained from a liquid scintillation counter. The mean value for the two

March 1992

PERMEABILITY

BARRIER

IN ESOPHAGEAL

EPITHELIUM

913

Figure 2. TEM showing the basal cells of s. germinativum. Arrows indicate the numerous desmosomes. Note the wide intercellular spaces (original magnification X17,850). Inset shows a freeze-fracture replica of a large gap junction (arrowheads) between the E and P (extracellular and protoplasm) fracture faces. D, desmosome (original magnification x54,000).

&-minute flux periods was reported as the value for the tissue. Statistics The data for the electrophysiology experiments were analyzed using the Student’s t test for paired samples. Results TEM and Freeze-Fracture The rabbit esophageal epithelium is lined by a partially keratinized moist stratified squamous epithelium. The epithelium has approximately 25-35 cell layers; they can be separated structurally into the following three groups: those comprising the basal layers or stratum germinativum (s. germinativum), those comprising the middle layers or stratum spinosum (s. spinosum), and those comprising the most luminal layers or stratum corneum (s. corneum) (Figure 1).

The s. germinativum consists of two to three rows of irregular cuboidal-shaped cells. These cells have large centrally positioned nuclei and a cytoplasm packed with free ribosomes and a moderate number of mitochondria. Adjacent cells interdigitate through fingerlike processes that are connected by desmosomes across a relatively wide intercellular space. Hemidesmosomes attach the basal aspect of the cells to the underlying basal lamina (Figure 2). On freeze-fracture replicas the plasma membrane was noted to contain many large gap junctions as well as desmosomes, but tight (occluding) junctions of either the zonulae or fasciae type were not observed (Figure 2). The s. spinosum consists of 15-20 layers of extremely long disc-shaped (squamous) cells. These cells, 90-IlO-pm long and 2-3-pm thick in the upper layers and 30-60-ym long and about h-pm thick in the lower layers, had a nucleus that was centrally located with a notable margination of the chromatin and intermediate-sized filaments throughout the cy-

914

ORLANDO ET AL.

Figure 3. TEM of the middle region of s. spinosum. (A) Cell at the fop of the figure is layer 12 from the lumen. Arrows indicate degenerating mitochondria (original magnification X11,000). Inset shows a small gap junction between P fracture face (right side) and E fracture face (lefi side) (original magnification X156,000). (B) High-power EM of layers 18 and 19 from the lumen showing the numerous desmosomes along adjacent microvilli. Arrows indicate membrane-bound vesicles with dense masses as well as dendritic filamentous material inside (original magnification X49,000).

toplasm (Figure 3). The cells also contained Golgi complexes, mitochondria, endoplasmic reticulum, and ribosomes in various stages of degeneration as the cells proceeded from the lower to upper boundary of this stratum (Figure 3A). Within the upper three to seven layers of the s. spinosum the cells containedabundant 120-150-nm membrane-boundvesicles (Figure 3B). These vesicles, which were bounded by a symmetrical unit membrane, con-

GASTROENTEROLOGY Vol. 102, No. 3

tained both amorphous dark-staining masses and a fine dendritic network of strands similar to that found in the intercellular space adjacent to these cells. These same structures were present between the cells of the s. corneum, but there the findings were much more prominent. Typical membranecoating granules (Odland bodies), as observed in keratinized epithelia, were not identified either intracellularly or extracellularly. The plasma membrane of the cells in the s. spinosum was elaborated into short stubby microvilli and microplicae, about 30% of which were attached to a microvillus of an adjacent cell via a desmosomal connection (Figure 3B). In addition, in the upper two to three layers, tight junctions were identified by TEM at the circumferential edge of these elongated cells. The tight junctions were short, consisting of one to two points of fusion between the outer leaflets of adjacent cell membranes (data not shown). Because freeze-fracture replicas did not expose these junctions, further characterization as either zonulae, fasciae, or maculae occludentes could not be performed. The freezefracture replicas of this region, however, did show small gap junctions (Figure 3A). The s. corneum consists of 7-12 layers of squamous cells, the transition of which from those of s. spinosum was abrupt, being demarcated by several significant alterations (Figure 4). First, the nuclear envelope was usually absent, although in the lower layers of s. corneum, the nuclear contents remained within the cell in large clumps, often giving the impression at the light microscopic level of an intact nucleus, By the upper one to two layers of this zone, the nuclear material appeared to have dissolved in most cells, because Fuelgen’s stain was negative for both intracellular or extracellular DNA. Second, the cells of this layer acquired an “asymmetric unit plasma membrane,” with the inner leaflet appearing three to four times thicker than the outer leaflet. This appearance was not caused by actual differences in the membrane but by the adherence of a 7-9-nm-thick layer of dark-staining material to the cytoplasmic surface of the inner leaflet. Third, between adjacent cells of all layers of the s. corneum were tight junctions (Figure 4). These tight junctions were shown on freeze-fracture to have one to two, rarely three, strands on the P (or protoplasm) face of the plasma membrane. The strands were generally arranged in either a parallel array or as an interlacing meshwork (Figure 5A and B). Although only short segments of junctions were visible in replicas, strands always appeared continuous; this appearance suggested that they extended around the periphery of the cell as occluding-type junctions. In contrast to desmosomes, which were evenly spaced over the cell surface at the ends of microvillar tips,

March

1992

PERMEABILITY

BARRIER

IN ESOPHAGEAL

EPITHELIUM

915

Figure 4. TEM showing s. corneum above and s. spinosum below dashed line. Tight junctions (circled) are shown to be present at multiple levels within s. corneum. An increase in width of the intercellular space between layers I and z indicates that the upper cell is desquamating. The arrow shows a degenerating nucleus (N) in upper s. corneum (original magnification X12,600). Upper inset shows a tight junction in s. corneum consisting of two short fusion points (arrowheads) (original magnification X100,000). Lower inset shows the abrupt acquisition of an “asymmetrical unit membrane” (arrowheads) in the s. corneum cells compared with the typical tripartite plasma membrane of s. spinosum (arrows) below (original magnification X200,000).

tight junctions were only found at the peripheral margins of the cell. Within the luminal cell layer, the tight junctions between adjacent cells were in many areas pulled apart so that the cells appeared to be desquamating “intact” (i.e., shedding into the lumen without obvious breaks in their plasma membranes) (Figure 6). However, lysosomes were rarely observed in the surface cells, and it is of interest that there was no identifiable mucous layer on the epithelial surface either by TEM or in light microscopic sections stained with PAS. In addition to tight junctions and desmosomes, the intercellular space of the s. corneum and the uppers. spinosum contained three other structures. First structure was abundant irregularly shaped masses comprised of densely packed fine filaments (0.4-0.6nm apart) and small globular particles (Figure 7). These masses, which never occluded the intercellular space, were always adherent to one of the plasma membranes and usually in contact with a desmosoma1 plaque (Figure 8A). Second structure was abundant fine filamentous strands emanating from the outer leaflet of the plasma membrane (Figures 4, 7, and 8A). These strands, which branched and anastomosed to form a loose dendritic network bridging the intercellular space, were similar to the glycocalyx identified on the surface of the cells facing the esophageal lumen. Third structure was sparse, small,

smooth-surfaced lipidlike bodies. These were always found adjacent to a plasma membrane in narrowed spaces such as those created by desmosomes (Figure 8B). The lipid bodies were of two morphological types: those with an internal lamellarlike arrangement that could only be identified on freeze-fracture (suggesting their destruction by tissue preparations used for TEM) and those with a smooth outer surface and a homogenous interior (Figure 8B and C). Additional changes noted in cells from the s. spinosum-corneum boundary to the esophageal lumen included a progressive decrease in the number of both intracellular membrane-bound vesicles and desmosomal connections (remnants of desmosome dissolution were evident by the presence of isolated plaques within the cell or intercellular space) and an increase in the amount of dark-staining intercellular masses (Figure 7). In many cases it appeared that the masses may have been derived from the degradation of the intercellular desmosomal plaques; however, the observation that both components of the masses, i.e., parallel fibrils and amorphous mats, were found in membrane-bound intracellular vesicles argues against this possibility. Histochemistry Histochemical stains were performed for the purpose of identifying the nature of the intercellular

916

ORLANDO

ET AL.

GASTROENTEROLOGY

Vol. 102, No. 3

Figure 6. EM of cells adjacent to lumen showing dissolution of tight junctions (arrowheads) as cells pull apart during desquamation (original magnification X66,150).

ther staining with PATCH-SP to localize carbohydrate moieties by TEM not only showed positive staining of the glycocalyx and intercellular masses but stained positively the material present within the intracellular membrane-bound vesicles in cells of the upper s. spinosum and s. corneum (Figure 10). Barrier Experiments

Figure 5. Freeze-fracture replicas showing one to two strands (arrowheads) of occluding junctions between adjacent cells at layers 6-7 (A) and l-2 (B) (original magnification X62500).

materials (Figure 9A-D). Sections stained with Sudan black B and oil red 0 revealed a moderate number of scattered lipid droplets predominantly within the cells of the s. corneum (Figure 9D). The small lipid bodies observed by TEM in the intercellular space were not observed in these histological sections, presumably because they were too small to be resolved on light microscopy. Notably densely packed PAS-positive material was localized throughout the intercellular space of all layers of the s. corneum, and this material extended into the intercellular spaces of the upper 5-10 layers of s. spinosum (Figure 9A). Acidic glycoconjugates stained with alcian blue, pH 2.5, were also localized to the intercellular spaces but were confined to the upper layers of the s. corneum (Figures 9B and C). The intercellular material in the upper two to three layers of the s. corneum also stained with alcian blue, pH 1.0, but this reaction was very weak (data not shown). Fur-

After identifying the presence of tight junctions and an intercellular glycoconjugate material as potential barriers to diffusion across the intercellular

Figure 7. TEM showing the acquisition of dense intercellular masses of material (arrows) within the intercellular space of the s. corneum. Numbers on right side indicate cell layer from lumen. Progressive dissolution of desmosomes in the upper layer (small arrowheads) is also apparent when compared with the intact desmosomes found in the lower layer (large arrowheads) (original magnification X30,000).

March 1992

PERMEABILITY

BARRIER IN ESOPHAGEAL

EPITHELIUM

917

space of rabbit esophageal epithelium, tracer and electrophysiologic experiments were conducted to clarify their respective roles. Tracer exposures were performed on tissues mounted in Ussing chambers for periods ranging up to 6 hours to ensure adequate time for maximum penetration of tracer to occur. Ussing chambers were useful not only for permitting access to either mucosal or serosal sides for tracer exposure but for measurements of PD, Isc, and electrical resistance, parameters valuable in- ensuring the integrity of the individual tissues being studied as representative of the species.3*5 Tracer Studies

Figure 8. EMS of intercellular spaces of s. corneum. (A) Thin section showing intercellular mass (arrowhead) associated with desmosome (D) (original magnification ~108,750). Inset shows two morphologies of these masses, the amorphous mat (asterisk) and the fibrillar part where there is a parallel arrangement of darkly staining fibrils (arrows) (original magnification X240,000). (E) Thin section shows two lipid bodies (arrowheads) adjacent to the desmosome (D) (original magnification ~105,000). (C) Freeze-fracture replicas showing lamellar lipid body (arrow) and smooth lipid body (arrowhead), both adherent to the plasma membrane. MV, microvillus (original magnification X100,000).

Exposure of esophageal epithelium mounted in Ussing chambers to HRP or lanthanum from the mucosal surface failed to shown any penetration of tracer into or between the cells of any complete layer. However, in selected tissues where the outer layer was in the process of desquamating, sections occasionally revealed tracer within the first intercellular space but not beyond (data not shown). This same observation was made in freshly excised tissues that had been simultaneously exposed to tracer from both the mucosal and serosal surfaces, indicating that the findings observed in the Ussing-chambered tissues were not artifacts of mounting. Exposure of esophageal epithelium mounted in Ussing chambers to either lanthanum or HRP from the serosal surface showed that neither tracer entered the cells. Also both tracers penetrated the intercellular spaces of all layers of the s. germinativum and most layers of the s. spinosum (Figure 11).Both lanthanum and HRP penetration were observed to end within the intercellular space of the s. spinosum three to seven cell layers below its boundary with the s. corneum (Figure 11). A similar pattern of tracer penetration into freshly excised tissues was also observed when incubations were performed with exposure to tracers simultaneously from both sides. Although extracellular tracer from the serosal side generally permeated to several layers below the cellular level at which tight junctions were observed, rare areas were found where tight junctions and tracer (HRP) were coincident. However, TEM of these areas showed that HRP was present on both sides of the junction (Figure 11). Because the pattern of restricted tracer penetration mirrored the distribution of glycoconjugate material rather than tight junctions, another tracer study was performed to further corroborate a relationship between glycoconjugate material and barrier function. The intercellular spaces of all layers was exposed to tracer via the lateral cut edges

918

ORLANDO

ET AL.

GASTROENTEROLOGY

Vol. 102,No. 3

Figure 9. Light micrographs of histochemical stains for glycoconjugates (A-C) and lipids (0). Dashed line separates s. corneum above from s. spinosum below. There is dense intercellular staining with PAS (A) for neutral glycoconjugates and alcian blue, pH 2.5 (II), for acidic glycoconjugates. Note the distribution of PAS throughout the s. corneum and into the upper layers of the s. spinosum (A), whereas alcian blue predominantly stains the intercellular material within the s. corneum (II). The purple color with a combined alcian blue-PAS stain reveals that the regions have both acidic and neutral glycoconjugates; the pink color identifies those areas of s. spinosum containing predominantly neutral glycoconjugates (Cl. Scattered intracellular lipid droplets are identified in the s. corneum but not in the s. spinosum after staining with oil red 0 (0) (original magnification x975).

of the tissue, effectively bypassing any barrier function contributed by tight junctions. This was performed by first excising the tissue and fixing it without exposure to tracer. After recutting to expose the intercellular spaces through the lateral cut edge, tissues were then exposed for 24 hours to fixative containing 2% lanthanum nitrate as tracer.13 As a result of this process, lanthanum was observed to penetrate the intercellular spaces of the s. germinativum and those of the lower two thirds of s. spinosum, but tracer remained unable to penetrate the intercellular spaces of the s. corneum or upper s. spinosum (Figure 12). Notably, this distribution of lanthanum within the intercellular space was similar to that observed with exposures to HRP (Figure 11). Electrophysiology

Experiments

Additional studies to assess the relationship between tight junctions and paracellular permeability in rabbit esophageal epithelium were performed by monitoring resistance in Ussing-chambered tissues before and during exposure to either a Ca’+free, EDTA-containing (5 mmol/L) solution, a luminal solution made hypertonic by the addition of 1

mol/L mannitol or the addition to both bathing solutions of cytochalasin B, 5 j.tmol/L. Tissues exposed to each of these three environments have been shown to show an increase in paracellular permeability as a result of the opening of tight juncfions.‘4-‘8 The first maneuver, exposure to Ca2+-free bathing solution in the presence of the calcium chelator, EDTA, resulted in a small decline in resistance of 8% + 4%, whereas controls exposed to normal Ringer’s increased 4% f 2% (Table 1). Notably, despite the reduction in resistance in Ca’+-free EDTA solution, there were no identifiable changes in either tight junctional morphology or degree of penetration of HRP into the tissue from either mucosal or serosal side (data not shown); also, an additional study of [14C]mannitol fluxes performed in tissues exposed to Ca2+-free EDTA documented that the decrease in resistance was accompanied by a small, but statistically insignificant, increase in mannitol permeability (Table 2). The second and third maneuvers, exposure to either a luminal bathing solution made hypertonic with 1 mol/L mannitol or to luminal and serosal cytochalasin B, failed to show any decline in resistance; instead, resistance increased (significantly for hypertonic mannitol) compared with resistance observed

PERMEABILITY

March 1992

Figure 10. TEM of sections reacted for PATCH-SP for carbohydrates. Small darkly staining globules indicate a positive reaction in the dendritic glycocalyxlike material of the intercellular space and in the masses found in membrane-bound vesicles (arrowI~eod& D, desmosome (original magnification X120,000).

in simultaneously (Table 1).

studied,

paired

Ringer controls

Discussion The nature and distribution of the barriers to diffusion across the paracellular pathway of stratified squamous epithelia appears to be highly variable and dependent on the organ under study. Thus, for example, based on tracer studies, there are at least three different structures configured in five different ways acting as barriers to diffusion across the intercellular space.‘s~*9-z4 The barrier structures include (a) tight junctions (zonula occludentes), e.g., frog skinz4; (b) an intercellular lamellar-shaped lipid, e.g., mouse esophagus and epidermis”; and/or (c) an intercellular glycoprotein, e.g., rabbit oral epithelium.23 These structures, which may act alone as barriers, are also reported to act in combination as barriers, and further such combinations, e.g., tight junctions and intercellular material, may vary to the extent that the two structural barriers either reside within the same topographical area, e.g., oral epithelium of rabbit or rat1g*23or within two distinct topographical areas, e.g., skin of the grass snake.” In the present investigation the structural barriers to permeation through the paracellular pathway

BARRIER

IN ESOPHAGEAL

EPITHELIUM

919

were sought in rabbit esophageal epithelium. By TEM two structures were identified that appeared capable of contributing to the paracellular barrier function in this tissue, i.e., tight junctions and/or an intercellular material staining positive for neutral and acidic glycoconjugates. Tight junctions were identified throughout the s. corneum, were sparse in the upper two to three layers of s. spinosum, and were absent in all deeper layers. The intercellular material was identified within the intercellular spaces of all layers of s. corneum and extended into the intercellular spaces for 5-10 layers of upper s. spinosum. Notably, the electron-dense tracers, HRP or lanthanum, failed to penetrate the intercellular space when applied exclusively to the mucosal surface unless disruption of the first cell layer was noted but could, when applied exclusively to the serosal surface, penetrate the intercellular spaces of the s. germinativum and lower two thirds of the s. spinosum. Neither HRP nor lanthanum were able to penetrate into the intercellular spaces of the upper s. spinosum. Based on this distribution of tracer, which localized the barrier predominantly to s. corneum and upper layers of s. spinosum, the structural barrier to molecules/ions the size of HRP (4-5 nm) and lanthanum (2 nm) in rabbit esophageal epithelium was more compatible with the presence of the intercellular glycoconjugate material than the tight junctions. This concept was further strengthened by the studies with laterally diffusing lanthanum, because it would have been impossible to cut through tight junctional regions at the margins of each of the cell layers of s. corneum and uppers. spinosum such that lanthanum was uniformly impeded from entering the intercellular spaces from either edge for the entire region. The filling of the intercellular spaces by glycoconjugate material, however, would be consistent with this observation. Although the tracer studies discussed above support the intercellular glycoconjugate material as the major barrier in rabbit esophageal epithelium, a role for the tight junctions in barrier function, perhaps to smaller ions, could not be excluded. This was particularly true in the present studies, because the intercellular glycoconjugate material overlapped and extended, in all but rare instances, beyond the entire distribution of the tight junctions. In an attempt to address this issue, it was noted that rabbit esophageal epithelium is an electrically “tight” tissue as reflected by an electrical resistance of > 1000 Q/cm2 (Table 1).Although the correlation is imperfect,” the studies of Claudez6 and Madara and Dharmsathaphornz7 indicate that tight junctional strand number (assessed on freeze-fracture replicas) generally corre-

920

ORLANDO ET AL.

GASTROENTEROLOGY

Vol. 102, No. 3

lar to that in rat oral epithelia,lg suggests that the individual tight junction does not account for the low permeability across esophageal tissue. Nonetheless, tight junctions were located in all layers of s. corneum and occasionally within the upper two to three layers of s. spinosum, and as such could still account for low tissue permeability by acting as resistors in series; thus, a tissue with an s. corneum of, for example, 10 cell layers with individual tight junctions averaging one to two strands could behave as if it had a single junction with a IO-ZO-strand num-

lumen

Figure 11. Low-power EM showing intercellular spaces filled with horseradish peroxidase (arrowheads) permeating from the serosal side to layer 17-18(dotted line) from the lumen. Dashed line marks the boundary between s. spinosum and s. corneum (original magnification X4950). Inset shows high-power EM from an unusual area where horseradish peroxidase penetrated to the level of the tight junction. Tracer (arrows) is present on both sides of the junction (arrowheads) (original magnification x149,990).

lates with electrical resistance. The present analysis of rabbit esophageal epithelium, however, showed that individual tight junctions, which appeared as only brief “kisses” by TEM, had only one or two strands, and rarely three. This finding, which is simi-

Figure 12. Light micrograph (epoxy resin section) of s. spinosum and s. corneum showing the cut edge of the tissue (lefl side) exposed to lanthanum. Black arrowheads indicate penetration of tracer into the intercellular spaces of mid to lower layers of s. spinosum. Open arrowheads indicate intercellular spaces of upper s. spinosum and s. corneum that are free of lanthanum. Dashed line indicates approximate margin between s. spinosum and s. corneum (original magnification ~2000).

PERMEABILITY BARRIER IN ESOPHAGEAL EPITHELIUM

March 1992

Table 1. The Electrical Resistance of Rabbit Esophageal Epithelia Before and After Exposure to Three Treatments Known to Increase Permeability Across EpitheIiaJ Tight Junctions Resistance Agent

n

Postexposure

Initial

Ca’+-free Ringer’s + EDTA 9 1849+ 242 (5 mmol/L] 9 1851+277 Ringer’s control Mannitol 5 2095 +309 (1 mol/L) 5 209Ok205 Ringer’s control Cytochalasin B 4 2497f289 (5wol/L) 4 2529 + 265 Ringer’s control

(2)

1739f273 19162288

-8 f4'= 4+2

2615 k464 2268f249

23f8' 8+4

2897f395 2941 + 281

15 f 2 17 +1

NOTE. Electrical resistance is expressed in Q X cm’. R (%) indicates percentage change in resistance from initial resistance, after exposure to Ca-free Ringer’s plus EDTA for 45 minutes and mannito1 or cytochalasin B for 1 hour. Values are means rt SEM. A negative sign indicates a decline in electrical resistance (i.e., increase in permeability); a positive sign indicates an increase in electrical resistance (i.e., decrease in permeability). “P < 6.95 compared with simultaneously studied, paired control.

ber. However, when &sing-chambered esophageal epithelium was exposed to three maneuvers known to “open” tight junctions, i.e., luminal hypertonicity, cytochalasin B (a microfilament-active agent), or a Ca’+-free environment,‘4-‘a permeability increased little or none. Further tracer studies and mannitol fluxes performed on tissues in which resistance was reduced on exposure to a Ca’+-free environment continued to show a lack of penetration of HRP beyond that observed in tissues bathed in normal Ringer’s solution and only a small, but statistically insignificant, increase in mannitol permeability. Together, these morphological, flux, and electrophysiological data, while not conclusive, do not provide support for a role for the tight junctions in the barrier function of rabbit esophageal epithelium.

921

Our identification of an intercellular material as an important barrier to diffusion across the paracellular pathway in rabbit esophageal epithelium is consistent with the results reported for mouse and human esophagus by Elias et al.” However, the nature of the material in these latter tissues was reported to be a lamellar-lipid based on its appearance by TEM. In rabbit esophageal epithelium the intercellular material was, except in rare instances, devoid of lamination, and the application of histochemical techniques showed the material to stain densely positive with PAS-diastase and alcian blue at pH 2.5. Further, the intercellular material did not stain positive with either of the lipid stains (oil red 0, Sudan black B) used, although a number of small discrete “lipid bodies” were identified within cells on histochemical staining and within the intercellular space by TEM. These findings, therefore, suggest that neutral and acidic glycoconjugates, more compatible with glycoproteins than glycolipids, represent the barrier in rabbit esophageal epithelium. Of further interest, and in contrast to the report of Elias et a1.,2o this same type glycoconjugate material may also represent the barrier in human as well as other types of esophageal epithelia. This possibility is based on the results of histoc-hemical stains that show the presence of intercellular glycoconjugate materials in rat, monkey, and human esophageal epithelia28-30 similar to those noted in rabbit esophageal epithelia. The derivation of the intercellular material subserving barrier function in esophageal epithelium is also of interest. Al Yassin and Toner31 and Logan et a1.32 have previously shown the presence of PASand/or alcian blue-positive material, similar to that seen in the intercellular space, within the membrane-bound vesicles of s. spinosum or s. corneum in human esophageal epithelial cells. In contrast, in rabbit esophageal epithelium the material within the membrane-bound vesicles did not appear to stain positively with PAS or alcian blue; however, this

Table 2. Mannitol Flux and Electrical Resistance of Rabbit Esophageal to Ca’+-Free Ringer’s Plus EDTA for 135 Minutes

Epithelia

Before and After Exposure

Resistance Agent

n

Ca’+-free Ringer’s + EDTA (5 mmol/L) Ringer’s control

6 6

NOTE. Electrical resistance is expressed average value taken over the SO-minute averaging the fluxes measured over two tissues in each group. Values are means “P < 6.65 compared with simultaneously

Initial 2838 + 261 2851k 359

Postexposure

2727 + 228 3209k 362

$1 -3+6" 14 + 3

[14C]Mannitol flux (pmol . h-’ - cm-‘)

0.0017* 0.0005 0.0013+ 0.0006

in Q X cm’. R (%) indicates percentage change in resistance (with postexposure resistance as the period of the flux) from initial resistance. [‘4C]Mannitol flux for each tissue was determined by consecutive 45-minute periods; the mean flux presented in the table is the average flux for six + SEM. studied, paired control.

922

ORLANDO ET AL.

lack of positivity most likely reflected the presence of quantities too small to be detected by light microscopy, because the intracellular vesicles were observed by TEM to stain positively for the presence of carbohydrate moieties using the PATCH-SP technique.” This result suggests that the intercellular glycoconjugate material is probably synthesized and packaged into membrane-bound vesicles within cells of the barrier layers before secretion into the intercellular space. In summary, the present investigation shows that the major barrier to diffusion across the intercellular space of rabbit esophageal epithelium resides within the cell layers of s. corneum and upper s. spinosum. The nature of this barrier appears to be, to a significant degree, an intercellular glycoconjugate material, most likely glycoproteins. This material appears to be synthesized in the cells of the barrier strata and packaged, before secretion, as membrane-bound vesicles. After secretion, the tight junctions, for which little support as ion/molecular barriers could be defined in the current investigations, may by being the narrowest region between adjacent cells contribute to retention of the presumed larger-sized glycoconjugate material within the intercellular space. Additional studies are needed to better define the nature of the glycoconjugate material to understand how it exerts its barrier properties. References 1. Powell 2. 3.

4.

5.

6.

7.

8.

9.

DW. Barrier function of epithelia. Am J Physiol 1981;241:G275-G288, Fromter E, Diamond J. Route of passive ion permeation in epithelia. Nature 1972;235:9-13. Orlando RC, Powell DW, Carney CN. Pathophysiology of acute acid injury in rabbit esophageal epithelium. J Clin Invest 1981;68:286-293. Carney CN, Orlando RC, Powell DW, Dotson MM. Morphologic alterations in early acid-induced epithelial injury of the rabbit esophagus. Lab Invest 1981;45:198-208. Orlando RC, Bryson JC, Powell DW. Mechanism of H+ injury in rabbit esophageal epithelium. Am J Physiol 1984;246: G718-G724. Orlando RC, Powell DW. Studies of esophageal epithelial electrolyte transport and potential difference in man. In: Allen A, Flemstrom G, Garner A, Silen W, Turnberg LA, eds. Mechanisms of mucosal protection in the upper gastrointestinal tract. New York: Raven, 1984:75-79. Lacy ER, Tobey NA, Cowart K, Orlando RC. The esophageal mucosal barrier: structural correlates (abstr). Gastroenterology 1989;96:A281. Ito S, Karnovsky MJ. Formaldehyde-glutaraldehyde fixatives containing trinitro compounds (abstr]. J Cell Biol 1968; 39:168a. Karnovsky MJ. Use of ferrocyanide-reduced osmium tetroxide in electron microscopy. American Society of Cell Biology Proceedings, 11th Annual Meeting, New Orleans, 1971:146.

GASTROENTEROLOGY Vol. 102, No. 3

10.

Preece A. A manual for histologic technicians. 3rd ed. Boston: Little, Brown, 1972:259-261:325-326, 375-376. 11. Thiery J-P. Mise en evidence des polysaccharides sur coupes fines en microscopic electronique. J Microscopic 1967;6:9871018. 12. Graham RC Jr, Karnovsky MJ. The early stages of absorption of injected horseradish peroxidase in the proximal tubules of mouse kidney: ultrastructural cytochemistry by a new technique. J Histochem Cytochem 1968;14:291-302. 13. Revel JP, Karnovsky MJ. Hexagonal array of subunits in intercellular junctions of the mouse heart and liver. J Cell Biol 1967;33:C7-C12. 14. Sedar AW, Forte JG. Effects of calcium depletion on the junctional complex between oxyntic cells of gastric glands. J Cell Biol 1964;22:173-188. 15. Palant CE, Duffey ME, Mookerjee BK, Ho S, Bentzel CJ. Ca’+ regulation of tight-junction permeability and structure in Necturus gallbladder. Am J Physiol 1983;245:C203-C212, 16. Landmann L, Stolinski C, Martin B. The permeability barrier in the epidermis of the grass snake during the resting stage of the sloughing cycle. Cell Tissue Res 1981;215:369-382. 17. Erlij D, Martinez-Palomo AM. Opening of tight junctions in frog skin by hypertonic urea solutions. J Membr Biol 1972;9:229-240.

18. Bentzel CJ, Hainau B, Ho S, Hui SW, Edelman A, Anagnostopoulos T, Benedetti EL. Cytoplasmic regulation of tight-junction permeability: effect of plant cytokinins. Am J Physiol 1980;239:C75-C89. 19. Shimono M, Clementi F. Intercellular junctions of oral epitheBum: studies with freeze-fracture and tracing methods of normal rat keratinized oral epithelium. J Ultrastruct Mol Struct Res 1976;56:121-136. 20. Elias PM, McNutt NS, Friend DS. Membrane alterations during cornification of mammalian squamous epithelia: a freezefracture, tracer, and thin-section study. Anat Ret 1977;189: 577-594.

21. Elias PM, Friend DS. The permeability barrier in mammalian epidermis. J Cell Biol 1975;65:180-191. 22. Henrikson RC, Stacy BD. The barrier to diffusion across ruminal epithelium: a study by electron microscopy using horseradish peroxidase, lanthanum, and ferritin. J Ultrastruct Res 1971;34:72-82.

23. Squier CA, Rooney L. The permeability of keratinized and nonkeratinized oral epithelium to lanthanum in vivo. J Ultrastruct Res 1976;54:286-295. 24. Martinez-Palomo A, Erlij D, Bracho H. Localization of permeability barriers in the frogepithelium. J Cell Biol1971;50:277287.

25. Mollgard K, Malinowska DH, Saunders NR. Lack of correlation between tight junction morphology and permeability properties in developing choroid plexus. Nature 1976;264: 293-294. 26. Claude P. Morphological factors influencing transepithelial permeability: a model for the resistance of the zonula occludens. J Membr Biol 1978;39:219-232. K. Occluding junction struc27. Madara JL, Dharmsathaphorn ture-function relationships in a cultured epithelial monolayer. J Cell Biol 1985;101:2124-2133. 28. Hopwood D, Logan KR, Coghill G, Bouchier IAD. Histochemical studies of mucosubstances and lipids in normal human oesophageal epithelium. Histochem J 1977;9:153-161. 29. Wislocki GB, Fawcett DW, Dempsey EW. Staining of stratified squamous epithelium of mucous membranes and skin of man

March 1992

and monkey by the periodic acid-Schiff method. Anat Ret 1951;110:359-376. 30. Rambourg A. Localization ultrastructurale et nature du materiel colore au niveau de la surface cellulaire par le melange chromique-phosphotunsstique. J Microscopic 1969;8:325-342. 31. Al Yassin TM, Toner PG. Fine structure of squamous epithelium and submucosal glands of human esophagus. J Anat 1977;123:705-721. 32. Logan KR, Hopwood D, Milne G. Ultrastructural demonstration of cell coat on the cell surfaces of normal human oesophageal epithelium. Histochem J 1977;9:495-504.

PERMEABILITY BARRIER IN ESOPHAGEAL EPITHELIUM

923

Received April 4.1991. Accepted July 30,1991. Address requests for reprints to: Roy C. Orlando, M.D., University of North Carolina School of Medicine, 326 Burnette-Womack Building, Campus Box 7080, Chapel Hill, North Carolina 27599. Supported by National Institutes of Health grant 5-ROlDK36013 and Marion Merrell Dow, Inc. (R. C. Orlando) and National Institutes of Health grant l-ROl-DK39074 (E. R. Lacy). The authors thank Lisa Von Hagen and Marion Hinson for their secretarial support in the preparation of this manuscript and Jan King for assistance with the histochemical staining.