Transport pathways for macromolecules in the aortic endotheliumI. Transendothelial channels revealed by three-dimensional reconstruction using serial sectionsкод для вставкиСкачать
THE ANATOMICAL RECORD 236:653-663 (1993) Transport Pathways for Macromolecules in the Aortic Endothelium: 1. Transendothelial Channels Revealed by Three-Dimensional Reconstruction Using Serial Sections KAZUSHIGE OGAWA, TOMOYOSHI WATABE, AND KAZUYUKI TANIGUCHI Department of Veterinary Anatomy, Faculty of Agriculture, Iwate University, Ueda, Morioka, Iwate (K.O., K.T.) and EM Application Laboratory, JEOL Ltd., Musashino, Akishima, Tokyo (T.W.), Japan ABSTRACT Three-dimensional organization of structures labeled by horseradish peroxidase as a tracer molecule in rat aortic endothelium was examined to elucidate the transport pathways for macromolecules. Aortic endothelium was divisible morphologically into four distinct parts, i.e., the parajunctional region (JR), peripheral region (PR), organelle region (OR), and nuclear region (NR). Almost all vesicles, intercellular clefts, and phagosomes were labeled by horseradish peroxidase. Peroxidase-positive vesicles tended to gather in the PR, occasional J R , and the upper part of the NR. Ultrathin serial micrographs revealed the transcellular channels composed of vesicles in the PR and J R , and the paracellular channels composed of the intercellular cleft without constrictions of tight and gap junctions. Transcellular channels were subdivided into three morphologically different types, where not only vesicles but invagination of abluminal membrane and intercellular cleft occasionally participated in their formation. Three-dimensional reconstructions from three series of consecutive electron micrographs revealed that almost all peroxidase-positive vesicles were connected directly or indirectly with the cell surface. These results indicate that the transcellular and paracellular channels are the transport pathways for macromolecules in the aortic endothelium and may suggest that the “shuttle” hypothesis is unsuitable to explain transport system for macromolecules because it postulates the existence of many free vesicles in the cytoplasm. o 1993 Wiley-Liss, Inc. Key words: Aortic endothelium, Permeability, Peroxidase perfusion, Rat, Three-dimensional reconstruction, Transendothelial channel, Transport pathway, Ultrathin serial section The modulation of vascular permeability is recognized as one of the most important endothelial functions and has been studied particularly in the capillary endothelium. Physiological measurements of the permeability properties revealed that the capillary wall serves as a highly selective barrier and is regarded as having the same function as an artificial membrane with “small pores and a few large pores,” 4-8 nm and 20-30 nm in radius, respectively (Grotte, 1956; Pappenheimer, 1953; Rippe and Haraldsson, 1986; Rippe et al., 1979; Taylor and Granger, 1984). The morphological equivalent of “small and large pores” has been searched for intensively and intercellular clefts and plasmalemmal vesicles have been considered as target structures (Bundgaard, 1984; Simionescu, 1983; Taylor and Granger, 1983; Wagner and Casley- Smith, 1981). Recently, a number of possible transport pathways in the endothelium have been illustrated in various kinds of capillaries (Clough, 1991; Simionescu, 1983; Taylor and Granger, 1983; Wagner and Casley-Smith, 1981). In arterial endothelium, permeability properties are 0 1993 WILEY-LISS,INC. not only important for transport of nutrients to the intimal and medial smooth muscle cells, but also in the close correlation between abnormal increases in endothelial permeability and atherogenesis (ROSS, 1986). According to the physiological data by O’Donnell and Vargas (19861, transport properties of arterial endothelium are similar to those of continuous capillary endothelium, although morphological differences such as endothelial thickness, intercellular junctional system, and vesicle density have been described between them (Londoiio and Bendayan, 1990; Simionescu and Simionescu, 1984). In addition, mechanical factors (i.e., blood flow, pressure oscillations, and sinusoidal stretch) have a large influence on the rate of water- Received October 20, 1992; accepted February 15, 1993. Address reprint requests to Kazushige Ogawa, PhD, Department of Veterinary Anatomy, Faculty of Agriculture, Iwate University, 3-18-8 Ueda, Morioka, Iwate 020,Japan. 654 OGAWA ET AL. soluble molecule transport in the arterial endothelium (Chien, 1978) and may cause some differences in transport properties between arterial and capillary endothelium. Morphological studies on the transport pathways for macromolecules in the arterial endothelium have been largely restricted to two-dimensional analyses based on random thin sections. Bundgaard (1983) and Frgkjaer-Jensen (1980) demonstrated that they provided less information compared to three-dimensional analyses based on serial sections. This suggests that insufficient morphological analyses have so far been done in the arterial endothelium to understand its transport properties for macromolecules. In the present study, therefore, we wanted to elucidate transport pathways in the aortic endothelium by three-dimensional reconstruction using ultrathin serial sections and horseradish peroxidase (HRP) as a tracer molecule: HRP labels both “small and large pores” owing to its small radius of 2.98 nm. Since we have already demonstrated the existence of channels composed of fused vesicles in the aortic endothelium (Ogawa et al., 1986), our particular focus here is on the detailed nature of channels of fused vesicles and intercellular clefts. MATERIALS AND METHODS Tissue Preparation Young adult Sprague-Dawley rats of both sexes weighing about 130 gm were used in this study. Under sodium pentobarbital anesthesia (Somnopentyl, Pitman-Moore, Inc., Washington Crossing; 0.1 ml per 100 gm body weight, by intraperitoneal injection), HRP (type 11,Sigma Chemical Co., St. Louis) dissolved in 0.4 ml of physiological saline was injected into the saphenous vein with a concentration of 15-20 mg per 100 gm body weight, and allowed to circulate for 2-5 min before sacrifice. The abdominal aorta was removed between celiac and renal arteries, and immersed for 120 min at 4°C in a mixture of 2% paraformaldehyde and 2.5% glutaraldehyde buffered with 0.1M cacodylate to pH 7.4. During fixation, the aorta was cut into segments less than 2 mm in length. After fixation, the tissues were rinsed for 3 hr a t 4°C in the same buffer containing 8%sucrose. Non-frozen cross sections of 80100 pm in thickness were cut with a Microslicer (Dosaka EM Co. Ltd., Kyoto) and incubated in the medium of Graham and Karnovsky (1966) for 30-60 min at room temperature to visualize HRP. After rinsing, sections were postfixed with 2% OsO, in the same buffer for 120 min at 4”C,dehydrated through a graded ethanol series followed by propylene oxide, and embedded in Spurr’s resin. Some sections were prestained en bloc with 1%uranyl acetate for 60 min before dehydration. Control rats received physiological saline (0.4ml per 100 gm body weight) instead of HRP to evaluate the specificity of reaction products. coated slightly with carbon and viewed with a JEM1200EX electron microscope at an acceleration voltage of 60 KV especially focused on the plasmalemmal vesicles and intercellular clefts labeled with HRP. In some grids, staining and/or carbon-coating were omitted. In several portions photographed serially, plasma membrane and cytoplasmic organelles labeled with HRP were traced on a transparent sheet from each print and stored through a digitizing tablet in the computer. Three-dimensional reconstruction was calculated using a JIM-5000 (JEOL Ltd., Akishima) high-speed image processing system with a three-dimensional reconstruction program. We reconstructed three endothelial segments from three series of consecutive electron micrographs through the computer. The magnification on the electron microscope was calibrated with a carbon replica of an optical grating (2,160 lines per mm). The thickness of serial sections ranged between 20-25 nm according to the method of Frgkjaer-Jensen (1980) which compares the number of serial sections containing a single vesicle with the maximum diameter of the same vesicle. RESULTS Localization of Horseradish Peroxidase and Its Regional Difference Aortic endothelium was divided into four distinct regions according to Simionescu et al. (1974) in continuous capillary endothelium, i.e., parajunctional region (JR) located around the intercellular junction and characterized by low frequency or absence of vesicles; peripheral region (PR) located in the peripheral part of the cell and characterized by large vesicle population; organelle region (OR) located adjacent to the nucleus and characterized by a high concentration of organelles; nuclear region (NR) located in the central part and occupied chiefly by the nucleus itself. The last three regions in the aortic endothelium were similar to those in the capillary endothelium in their morphological features. Morphological characteristics of the aortic endothelium were recognized in the JR and summarized as follows. First, the JR was not always situated between PRs as in the capillary endothelium. It held one of six positions according to the combination of adjoining regions and was frequently observed in three of them as follows: JR lying between PRs; JR lying between PR and OR; PR lying between ORs. Second, JRs were subdivided into 2 groups (JR1 and JR2) characterized respectively by high- and low-vesicle population. There was every indication that the latter (JR2) corresponded to the JR lying between PRs. Four distinct regions in the aortic endothelium are schematically summarized in Figure 1. Reaction products of peroxidase were observed on the external side of both luminal and abluminal membranes, in the intercellular clefts and the subendothe- Serial Sectioning Thin cross-sections of aortic endothelia were cut serially with a diamond knife mounted on an LKB U1trotome NOVA and collected on Formvar-covered grids with a single-slot of 1 x 2 mm. The serial sections as thin and uniform as possible were selected by judging the interference color. The obtained serial sections were stained with uranyl acetate and lead citrate, G IC IEL L P S Abbreviations Gap junction Intercellular cleft Internal elastic lamina Lumen Phagosome Subendothelial space TRANSENDOTHELIAL CHANNELS IN RAT AORTA 655 OR Fig. 1 . A diagram of four morphologically distinct regions in rat aortic endothelium. JR1: parajunctional region lying between organelle regions characterizedby a high concentrationof vesicles. JR2: parajunctional region lying between peripheral regions charac- terized by a low concentrationof vesicles. NR: nuclear region occupied by nucleus. O R organelle region characterized by a high concentration of organelles. PR parajunctional region characterizedby a high concentration of vesicles. lial space, and within almost all vesicles and most phagosomes, whereas they were not found in the aortic endothelium of the control animals. The reaction products revealed deep invaginations of the abluminal membrane (Fig. 2). Peroxidase-positive vesicles tended to gather in the PR, JR1, (Fig. 3) and the luminal part of NR. Tight junctions and/or gap junctions were usually observed in the intercellular membranes and formed a constriction less than 2 nm in the intercellular clefts unlikely to be permeable to HRP. Therefore, the intercellular cleft was divided into luminal and abluminal parts, although the latter was often filled with reaction products of peroxidase. of intercellular cleft was recognized as a third type. The transcellular channels connecting the luminal parts of intercellular clefts with the abluminal space, however, were not observed in the present study. In the highly interdigitating portion, a chain of fused vesicles (or a vesicle) was frequently (not rarely) found even in a plane of a single section. This connected an abluminal part of the intercellular cleft with an abluminal space or another part of the abluminal intercellular cleft and formed a shortcut pathway for macromolecules (Fig. 6a). Intercellular clefts without the constriction were sometimes found. They were about 20 nm in width along most of its length (Figs. 6a, 7) and were regarded as one of paracellular channels. Serial micrographs revealed that intercellular clefts varied conspicuously from simple profiles to complicated profiles in intricacy of interdigitation (Figs. 6, 7) and from permeable profiles to impermeable ones that are present at the intercellular junctions (Fig. 6). Transendothelial Channels Channels formed by vesicles through endothelial cell cytoplasm (transcellular channels) inevitably contained the reaction products of peroxidase, but were rarely found in a plane of a single section. They were usually recognized only by three-dimensional organization of vesicles from serial electron micrographs. All channels were recognized in the PR and JR of the endothelium. The present three-dimensional observation revealed the existence of a t least three morphologically different types of transcellular channels. The first type is present in the PR. It consists of a chain of fused vesicles and connects the luminal surface with the abluminal surface directly (Fig. 4). The second type is also recognized in the PR and consists of a chain of fused vesicles and invaginations of abluminal membrane forming a deep slit, not a tubule or a cone (Fig. 5). They fuse with each other and form the transcellular channel. The last type i F observed in the JR and consists of a chain of fused vesicles (or a vesicle) and an intercellular cleft. A chain of fused vesicles opens to the luminal or abluminal surface and fuses with the intercellular membrane. This channel serves as a passage route for macromolecules to bypass the intercellular junctions (tight junction and/or gap junction). In the serial micrographs of Figure 6, a vesicle opening simultaneously to the luminal space and the abluminal part Three-DimensionalReconstruction The three-dimensional reconstruction revealed that almost all peroxidase-positive vesicles were connected directly or indirectly (fused with other vesicles) with the cell surface. The reconstruction also indicated that vesicles had a tendency to gather in certain domains. In Figure 8 of a stereopair including the JR, vesicles were gathered in one endothelium between the luminal membrane and the abluminal part of the intercellular membrane. In this domain, vesicles opening to the lumen were close to the other vesicles opening to the abluminal part of intercellular cleft, although a transcellular channel was not observed. Transendothelial channels, i.e., transcellular and paracellular channels, are schematically summarized in Figure 9. DISCUSSION Methodological Validity According to the pore theory which describes the permeability properties of capillary walls, two systems of 656 OGAWA ET AL. Fig. 2. Deep invaginations with branches of abluminal membrane are filled with reaction products of peroxidase in the organelle region of the endothelium. Bar: 0.5 +m. Fig. 3. The first one of 11 serial electron micrographs used for three-dimensional reconstruction (Fig. 8).Peroxidase-positive vesicles are gathered in the parajunctional region lying between organelle regions. Bar: 0.5 km. TRANSENDOTHELIAL CHANNELS IN RAT AORTA 657 Fig. 4. Transcellular channel (arrow) composed of a chain of fused vesicles is recognized in the peripheral region of the endothelium. Bar: 0.2 pm. pores, small pores and a few large pores, perforate capillary endothelium. The former with an effective radius of 4-8 nm is assigned to a transport pathway for water and small solutes up to the size of serum albumin (7 nm in diameter). The latter with a radius of 20-30 nm transports macromolecules (Grotte, 1956; Pappenheimer, 1953; Rippe and Haraldsson, 1986; Rippe et al., 1979; Taylor and Granger, 1984). Besides cylindrical channels about 12 nm in diameter, small pores were also estimated to be slits about 8 nm in width. Previous authors used many kinds of tracer molecules to determine the morphological counterparts in capillary endothelium (Bundgaard et al., 1979; Clough and Michel, 1981; Karnovsky, 1967; Simionescu et al., 1975). In the present study, however, we adopted HRP as a tracer to evaluate the permeability properties and t o identify the transport pathways for macromolecules in the aortic endothelium because of the following reasons: (1) HRP is an intermediate-sized tracer, about 6 nm in diameter, and able to pass the endothelial layer through both small and large pores; (2) HRP serves as an enzymatic tracer and allows one to detect its precise localization with the electron microscope by large amounts of its reaction products. Actually, many authors have reported that HRP is observed in the intercellular clefts and vesicles of the arterial endothelium (Florey and Sheppard, 1970; Huttner et al., 1973; Okuda and Yamamoto, 1983; Schwartz and Benditt, 1972). Intercellular clefts and vesicles of the arterial endothelium are regarded as counterparts of the small and large pores of capillary endothelium (Bundgaard, 1984; Simionescu, 1983; Taylor and Granger, 1983; Wagner and Casley-Smith, 1981). Much larger tracers like ferritin (11 nm in diameter), however, can label vesicles only (Baldwin and Chien, 1985; Huttner et al., 1970,1973).In spite of these advantages, HRP shows a significant disadvantage. Cotran and Karnovsky (1967) reported that HRP induces histamine release and leads to an increase in vascular permeability in some species including Sprague-Dawley rats. This may be caused by direct interaction of histamine with the endothelium to open its intercellular junctions through the endothelial contraction (Simionescu, 1983). However, this change in vascular permeability has occurred only in venular endothelium (Cotran and Karnovsky, 1967) and has not been observed in arterial endothelium (Florey and Sheppard, 1970; Huttner et al., 1973). These findings encouraged us to adopt HRP in the present study. In addition, reaction products were not observed in the aortic endothelium in our control experiments. This suggests that the reaction products originate from the exogenous peroxidase and may exactly label all kinds of transport pathways for macromolecules in the aortic endothelium. Almost all previous studies on the morphology of transport pathways for macromolecules were based on either tracer perfusion and random thin sectioning or ultrathin serial sectioning in the method. The former method is able to label vesicles taking part in the transport of tracer molecules. However, it cannot draw the organization of vesicles in the endothelium, because the diameter of vesicles is equal to the thickness of conventional thin sections (Bundgaard, 1983). On the other hand, the latter method tends to overestimate three-dimensional organization of vesicles not always participating in the transport for macromolecules. Therefore, the combination of both methods as carried out in the present study may allow one to elucidate in 658 OGAWA ET AL. Fig. 5. Serial electron micrographs in the peripheral region. A vesicle (arrow in Fig. 5A) connectedwith invagination of abluminal membrane (arrow in Fig. 5B)fuses with another vesicle (*) opening to lumen and forms a transendothelial channel. Bar: 0.2 pm. TRANSENDOTHELIAL CHANNELS IN RAT AORTA Fig. 6. Five selected electron micrographs ( A 3rd; E D : 10th-12th; E: 18th) from 18 serial sections. A vesicle opening to abluminal part of the intercellular cleft (arrow in Figure 6b) fuses with luminal membrane (arrow in Figure 6d) and leads to a formation of a transcellular channel. Another vesicle opening simultaneously to abluminal part of 659 the intercellular cleft and ablumen makes a shortcut pathway (arrow in Figure 6a). Note that intricacy of interdigitation varies from a complicated profile to a simple profile, and tight and gap junctions recognized in Figures 6 b e are not observed in Figure 6a. Bar: 0.4 pm. 660 OGAWA ET AL. Fig. 7. Serial electron micrographs showing paracellular channel. Intercellular cleft without tight and gap junctions is about 20 nm in width along most of its length and about 15 nm in some most narrowing parts. Its intricacy of interdigitation varies from a mortise profile (Fig. 7A) to a simple profile. Bar: 0.2 Pm. Fig. 8. A stereopair of three-dimensional reconstruction in the parajunctional region. Note vesicles gathered between the luminal membrane and abluminal part of the intercellular membrane in the endothelium. Bar: 0.2 Fm. detail the precise transport pathways for macromole- clefts, and vesicles opening to the lumen (Florey and Sheppard, 1970; Huttner et al., 1973; Ogawa, unpubcules in the endothelium. Immersion fixation was preferred in this experi- lished data). This makes it difficult to grasp the precise ment because perfusion fixation tends to wash out transport pathways for macromolecules in these reHRP from the lumen, luminal region of intercellular gions. TRANSENDOTHELIAL CHANNELS IN RAT AORTA 66 1 Lumen Subendothelium Fig. 9. A diagram of three morphologically different types of transcellular channels and a paracellular channel in rat aortic endothelium. Type 1: transcellular channel formed by a chain of fused vesicles. Type 2: transcellular channel formed by a chain of fused vesicles and invagination of abluminal membrane. Type 3: transcellular chan- nel formed by intercellular cleft and a chain of fused vesicles (or a vesicle), which connects lumen with abluminal part of the intercellular cleft or luminal part of the intercellular cleft with ablumen. PCC: paracellular channel formed by intercellular cleft without tight and gap junctions. Transport Pathways for Macromolecules neously to lumen and ablumen forms the transcellular channel in the thin PR (Coomber and Stewart, 1986; Cornillie and Lauweryns, 1984; Palade and Bruns, 1968; Simionescu et al., 1975). In the arterial endothelium, Ogawa et al. (1986) reported the channels composed of a chain of fused vesicles for the first time, but there have been few confirmatory reports of this observation. Although Nag (1990) reported the channels in the hypertensive rat arteriolar endothelium using the same method as that of Ogawa et al. (1986), it is uncertain in our judgment whether the channel is precisely formed or not. As observed in the present study, transcellular channels show much more complicated profiles in naturally thick arterial endothelium. They may behave more transiently and dynamically than those in capillary endothelium because the arterial endothelium is influenced by greater mechanical factors in blood vessels such as fast blood flow, high pressure and flow oscillations, and intense sinusoidal stretch. Consequently, they may be hardly observed in a single conventional thin section unlike those in capillary endothelium, but they can be recognized by analyzing ultrathin serial sections. Most transcellular channels in capillary endothelium are usually present in the PR and composed only of vesicles, although Frgkjaer-Jensen (1980) reported a channel in the JR. In the aortic endothelium, transcellular channels are morphologically composed of invaginating abluminal membranes and intercellular clefts, in addition to vesicles. This implies that these structures are valid and suitable for the connection of lumen and ablumen across the long distance, the naturally thick arterial endothelium. We demonstrated that the discontinuity of intercellular junctions (tight junction and gap junction) occurs in the aortic endothelium. This indicates the existence of a paracellular channel, a transport pathway for macromolecules. In the arterial endothelium, it is widely recognized that tight junctions show discontinuous properties and disc-shaped gap junctions are intercalated or partly framed by tight junctional strands (Hutt- Although Chien (1978) divided aortic endothelium into four distinct parts according to Simionescu et al. (19741, he did not show anything about the properties of arterial endothelium. Therefore, we reported here the morphological characteristics of the aortic endothelium with special reference to the JR as compared to a continuous capillary endothelium. In the present study, almost all vesicles, intercellular clefts, and phagosomes were labeled with HRP in the aortic endothelium. This agrees well with previous studies (Florey and Sheppard, 1970; Huttner et al., 1973; Schwartz and Benditt, 1972). Since phagosomes have a close relation to the intracellular digestive system, we excluded them from the discussion on transendothelial transport system for macromolecules. Peroxidase-positive vesicles showed a tendency to assemble in the PR, in some JR where endothelium was usually rather thick, and in the luminal portion of NR. This may indicate that the transendothelial transport system is located in the PR and JR. Vesicles in the OR as well as NR cannot be expected to participate in this transport system because physical hindrances such as many organelles or nucleus and longer distances between lumen and ablumen exist in these regions. These vesicles may serve to increase the cell surface area, allowing changes in vascular diameter (Wolff, 1977) or possessing elastic properties of endothelium (FrgkjaerJensen, 1984). They may also be involved in regulation of cytosolic Ca concentration (Bundgaard, 1987). The functional aspects of these vesicles need to be studied further. Transendothelial channels can be regarded as one of the transport pathways for macromolecules. We demonstrated the existence of three morphologically different types of transcellular channels in the aortic endothelium by ultrathin serial sectioning and HRP labeling. In the capillary endothelium, many authors have reported the transendothelial channels usually found on a single thin section. In this case, a single vesicle or a chain of fused vesicles opening simulta- 662 OGAWA ET AL. Chien, S. 1978 Transport across arterial endothelium. Prog. Hemost. Thromb., 4:l-36. Clough, G. 1991 Relationship between microvascular permeability and ultrastructure. Prog. Biophys. Molec. Biol., 55:47-69. Clough, G., and C.C. Michel 1981 The role of vesicles in the transport of ferritin through frog endothelium. J. Physiol., 315:127-142. Coomber, B.L., and P.A. Stewart 1986 Three-dimensional reconstruction of vesicles in endothelium of blood-brain barrier versus highly permeable microvessels. Anat. Rec., 215:256-261. Cornillie, F.J., and J.M. Lauweryns 1984 Transendothelial channels in fenestrated endometrial blood capillaries of rats. Cell Tissue Res., 237:371-373. Cotran, R.S., and M.J. Karnovsky 1967 Vascular leakage induced by Exp. Biol. Med., 126: horseradish peroxidase in the rat. Proc. SOC. 557-561. Florey, Lord, and B.L. Sheppard 1970 The permeability of arterial endothelium to horseradish peroxidase. Proc. Roy. Soc. Lond. B., 174:435-443. FrZkjaerJensen, J. 1980 Three-dimensional organization of plasmalemmal vesicles in endothelial cells: An analysis by serial sectioning of frog mesenteric capillaries. J. Ultrastmet. Res., 73320. Fr@kjaerJensen,J . 1984 The plasmalemmal vesicular system in striated muscle capillaries and in pericytes. Tissue Cell, 16:31-42. Graham, R.C., and M.J. Karnovsky 1966 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., 14:291-302. Grotte, G. 1956 Passage of dextran molecules across the blood-lymph barrier. Acta Physiol. Scand., Suppl., 21 1 :1-84. Hiittner, I., M. Boutet, and R.H. More 1973 Studies on protein passage through arterial endothelium: 11. Regional differences in permeability to fine structural protein tracers in arterial endothelium of normotensive rat. Lab. Invest., 28:678-685. Hiittner, I., R.H. More, and G. Rona 1970 Fine structural evidence of specific mechanism for increased endothelial permeability in experimental hypertension. Am. J. Pathol., 61:395-412. Hiittner, I., C. Walker, and G. Gabbiani 1985 Aortic endothelial cell during regeneration. Remodeling of cell junctions, stress fibers, and stress fiber-membrane attachment domains. Lab. Invest., 53: 287-302. Karnovsky, M.J. 1967 The ultrastructural basis of capillary permeability studied with peroxidase as a tracer. J . Cell Biol., 35:213236. Londofio, I., and M. Bendayan 1990 Quantitative immunocytochemical studies of endogenous albumin in rat aortic endothelial and mesothelial cells. Biol. Cell, 69~161-169. Miihleisen, H., H. Wolburg, and E. Betz 1989 Freeze-fracture analysis of endothelial cell membranes in rabbit carotid arteries subjected to short-term atherogenic stimuIi. Virchows Archiv. B Cell Pathol., 56:413-417. Nag, S. 1990 Presence of transendothelial channels in cerebral endothelium in chronic hypertension. Acta Neurochirurgica, Suppl., 51:335 -337. ODonnell, M.P., and F.F. Vargas 1986 Electrical conductivity and its use in estimating a n equivalent pore size for arterial endothelium. Am. J. Physiol., 250:H16-H21. Ogawa, K.S., K. Fujimoto, and K. Ogawa 1986 Ultracytochemical studies of adenosine nucleotidases in aortic endothelial and smooth muscle cell. Ca+ +-ATPase and N a + , K+-ATPase. Acta Histochem. Cytochem., 19:601-620. Okuda, T., and T. Yamamoto 1983 The ultrastructural basis of the permeability of arterial endothelium to horseradish peroxidase. Freeze-fracture and tracer studies of rat thoracic aorta and basilar artery. Cell Tissue Res., 231:117-128. LITERATURE CITED Palade, G.E., and R.R. Bruns 1968 Structural modulations of plasmalemmal vesicles. J. Cell Biol., 37:633-649. Baldwin, A.L., and S. Chien 1985 Effect of plasma proteins on endothelial binding and vesicle loading of anionized ferritin in rabbit Pappenheimer, J.R. 1953 Passage of molecules through capillary walls. Physiol. Rev., 33:387-423. aorta. Arteriosclerosis, 5:451-458. Bundgaard, M. 1983 Vesicular transport in capillary endothelium: Rippe, B., and B. Haraldsson 1986 Capillary permeability in rat hindquarters as determined by estimations of capillary reflection coDoes it occur? Federation Proc., 42:2425-2430. efficients. Acta Physiol. Scand., 127:289-303. Bundgaard, M. 1984 The three-dimensional organization of tight junction in a capillary endothelium revealed by serial-section Rippe, B., A. Kamiya, and B. Folkow 1979 Transcapillary passage of albumin, effects of tissue cooling and of increases in filtration and electron microscopy. J. Ultrastruct. Res., 88:l-17. plasma colloid osmotic pressure. Acta Physiol. Scand., 105:171Bundgaard, M. 1987 Tubular invaginations in cerebral endothelium 187. and their relation to smooth-surfaced cisternae in hagfish (MyxRoss, R. 1986 The pathogenesis of atherosclerosis-an update. N. ine glutinosa). Cell Tissue Res., 249:359-365. Engl. J. Med., 314:488-500. Bundgaard, M., J . Fr@kjaerJensen,and C. Crone 1979 Endothelial plasmalemmal vesicles as elements in a system of branching in- Schwartz, S.M., and E.P. Benditt 1972 Studies on aortic intima: I. vaginations from the cell surface. Proc. Natl. Acad. Sci. USA, Structure and permeability of rat thoracic aortic intima. Am. J . 76:6439-6442. Pathol., 66:241-264. ner et al., 1973, 1985; Miihleisen et al., 1989; Okuda and Yamamoto, 1983). Our observation agrees well with these reports. We also demonstrated that the intercellular clefts below the intercellular junctions are usually filled with the reaction products of HRP. This may be caused by the discontinuous tight and gap junctions and the transcellular channels connecting lumen with intercellular cleft below the intercellular junctions. They may allow the intercellular cleft to be permeable to HRP due to the bypassing of the intercellular junctions. Schwartz and Benditt (1972) grouped the junctions between aortic endothelial cells into four categories, i.e., a single end-to-end or “butt” junction; a simple “overlap” junction; a “mortise” junction; more complicatedly folded junctions. Some investigators use these categories as parameters of permeability for macromolecules through intercellular clefts. However, our observation with serial electron micrographs such as in Figures 6 and 7 suggest that the Schwartz and Benditt groupings may not be valid because of the remarkable variety in intricacy of these interdigitations. It is widely accepted that “vesicular transport” is one of the ways in which molecules cross the capillary wall. Palade and Bruns (1968) and Simionescu (1983) proposed that plasmalemmal vesicles constitute a distinct population of membranes and shuttle between luminal and abluminal membranes to carry molecules in their contents. This “shuttle” hypothesis presumes many vesicles free in the cytoplasm. Recent studies by ultrathin serial sectioning have demonstrated that almost all plasmalemmal vesicles in the capillary endothelium open to the endothelial surface (Bundgaard et al., 1983; Coomber and Stewart, 1986; FrokjaerJensen, 1980,1984). These workers have thrown doubt upon the validity of this “shuttle” hypothesis. In the aortic endothelium, our previous study (Ogawa et al., 1986) demonstrated the different vesicle properties as shown by enzyme activities between luminal and abluminal sides. Our present observation with three-dimensional reconstructions revealed that almost all peroxidase-positive vesicles were connected with the cell surface (see Fig. 8). This suggests that the “shuttle” hypothesis does not apply to the aortic endothelium. At present, channels alone appear to be the means of transendothelial transport system via plasmalemmal vesicles in the aortic endothelium. However, since our three-dimensional reconstructions of vesicles were carried out on a few limited arterial endothelial segments, additional studies of other arterial endothelia should be carried out to extend our understanding of the transport behavior of these morphological entities. TRANSENDOTHELIAL CHANNELS IN RAT AORTA Simionescu, M., and N. Simionescu 1984 Ultrastructure of the microvascular wall: Functional correlations. In: The Cardiovascular System, Handbook of Physiology, vol. IV, sect. 2, Microcirculation. E.M. Renkin and C.C. Michel, eds. Williams and Wilkins Company, Baltimore, pp. 41-101. Simionescu, M., N. Simionescu, and G.E. Palade 1974 Morphometric data on the endothelium of blood capillaries. J. Cell Biol., 60: 128-152. Simionescu, N. 1983 Cellular aspects of transcapillary exchange. Physiol. Rev., 63:1536-1579. Simionescu, N., M. Simionescu, and G.E. Palade 1975 Permeability of muscle capillaries to small heme-peptides: Evidence for the existence of patent transendothelial channels. J. Cell Biol., 64.586607. 663 Taylor, A.E., and D.N. Granger 1983 Equivalent pore modeling: Vesicles and channels. Federation Proc., 42.2440-2445. Taylor, A.E., and D.N. Granger 1984 Exchange of macromolecules across the microcirculation. In: The Cardiovascular System, Handbook of Physiology, vol. IV,sect. 2, Microcirculation. E.M. Renkin and C.C. Michel, eds. Williams and Wilkins Company, Baltimore, pp. 467-520. Wagner, R., and J.R. Casley-Smith 1981 Review: Endothelial vesicles. Microvasc. Res., 21 :267-298. Wolff, J.R. 1977 Ultrastructure of the terminal bed as related to function. In: Microcirculation. G. Kaley and B.M. Altura, eds. University Park Press, Baltimore, vol. 1, pp. 95-130.