PERMEABILITY OF INTESTINAL CAPILLARIES : Pathway followed by Dextrans and Glycogens

The pathway followed by macromolecules across the wall of visceral capillaries has been studied by using a set of tracers of graded sizes, ranging in diameter from 100 A (ferritin) to 300 A (glycogen). Polysaccharide particles, i.e. dextran 75 (mol wt ~75,000; diam ~125 A), dextran 250 (mol wt 250,000; diam ~225 A), shellfish glycogen (diam ~200 A) and rabbit liver glycogen (diam ~300 A), are well tolerated by Wistar-Furth rats and give no vascular reactions ascribable to histamine release. Good definition and high contrast of the tracer particles were obtained in a one-step fixation—in block staining of the tissues by a mixture containing aldehydes, OsO₄ and lead citrate in phosphate or arsenate buffer, pH 7.4, followed by lead staining of sections. The glycogens and dextrans used move out of the plasma through the fenestrae and channels of the endothelium relatively fast (3–7 min) and create in the pericapillary spaces transient (2–5 min) concentration gradients centered on the fenestrated sectors of the capillary walls. The tracers also gained access to the plasmalemmal vesicles, first on the blood front and subsequently on the tissue front of the endothelium. The particles are temporarily retained by the basement membrane. No probe moved through the intercellular junctions. It is concluded that, in visceral capillaries, the fenestrae, channels, and plasmalemmal vesicles, viewed as related parts in a system of dynamic structures, are the structural equivalent of the large pore system.     

INTESTINAL CAPILLARIES : I. Permeability to Peroxidase and Ferritin

Horseradish peroxidase (mol. diam. 50 A) and ferritin (mol. diam. 110 A) were used as probe molecules for the small and large pore system, respectively, in blood capillaries of the intestinal mucosa of the mouse. Peroxidase distribution was followed in time, after intravenous injection, by applying the Graham-Karnovsky histochemical procedure to aldehyde-fixed specimens. The tracer was found to leave the plasma rapidly and to reach the pericapillary spaces 1 min post injection. Between 1 min and 1 min 30 sec, gradients of peroxidase reaction product could be demonstrated regularly around the capillaries; their highs were located opposite the fenestrated parts of the endothelium. These gradients were replaced by even distribution past 1 min 30 sec. Ferritin, followed directly by electron microscopy, appeared in the pericapillary spaces 3–4 min after i.v. injection. Like peroxidase, it initially produced transient gradients with highs opposite the fenestrated parts of the endothelium. For both tracers, there was no evidence of movement through intercellular junctions, and transport by plasmalemmal vesicles appeared less efficient than outflow through fenestrae. It is concluded that, in the blood capillaries of the inintestinal mucosa, the diaphragms of the endothelial fenestrae contain the structural equivalents of the small pore system. The large pore system seems to be restricted to a fraction of the fenestral population which presumably consists of diaphragm-free or diaphragm-deficient units.     

Permeability of muscle capillaries to microperoxidase

In this study we attempted to identify a morphologic counterpart of the small pore of muscle capillaries. The existence of such a pore has been postulated by physiologists to explain the permeability of muscle capillaries to small macromolecules. We injected mice intravenously with microperoxidase (MP) and fixed specimens of diaphragm at intervals of 0-250 s after the injection to localize the tracer by electron microscopy. The small size of MP (1,900 mol wt and 20 A molecular diameter [MD]) ensures its ready passage through the small pore since the latter is thought to be either a cylindrical channel 90 A in diameter or a slit 55 A wide. MP appears in the pericapillary interstitium within 30 s of initiation of its intravenous injection. The patterns of localization of MP observed within clefts between adjacent capillary endothelial cells indicate that some endothelial junctions are permeable to this tracer. Although small vesicles transfer MP across the endothelium, we do not believe that the vesicles transfer substantial amounts of MP into the pericapillary interstitium. We did not obtain evidence that MP crosses the endothelium of capillaries through channels formed either by a single vesicle or by a series of linked vesicles opening simultaneously at both surfaces of the endothelial cell. From our observations we conclude that some endothelial junctions of capillaries are permeable to MP, and that these permeable junctions are a plausible morphologic counterpart of the small pore.     

Plasmalemmal vesicles are involved in transendothelial transport of albumin, lysosomal enzymes and mannose 6-phosphate receptor fragments in capillary endothelium

With the use of immunoelectron microscopy we have demonstrated the presence of lysosomal enzymes (acid alpha-glucosidase and glucocerebrosidase) and fragments of the 270 kDa receptor for mannose 6-phosphate and insulin-like growth factor II in blood plasma, plasmalemmal vesicles of endothelial cells and pericapillary spaces in human skeletal muscle tissue. At these locations, the three proteins colocalized with albumin known to be transported from the capillaries into the pericapillary spaces. Immunoblot analysis of plasma revealed the presence of relatively high molecular weight polypeptides in this material. These observations strongly suggest that high molecular weight species of lysosomal enzymes can pass the endothelial barrier in skeletal muscle tissue.     

Surface charges associated with fenestrated brain capillaries. II. In vivo studies on the role of molecular charge in endothelial permeability

We report on the effect of the net charge of a tracer (ferritin) on its permeability in fenestrated capillaries of the brain. Our experiments show that the charge of this tracer actually influences its interaction with the endothelium. Three phases of tracer-endothelial interaction could be discriminated. Anionic and slightly cationic derivatives (pH 4.5-7.8) do not show any affinity to the luminal endothelial membrane. Ferritin derivatives with a pI value between 7.8 and 9.3 result in the labeling of the fenestrae without coating additional luminal plasmalemmal structures (i.e., coated pits and plasmalemmal vesicles). Tracers with a high positive net charge (pI greater than 9.3) led to their endocytotic uptake and extravasation by some transcytotic mechanism. Extravasated cationic ferritin accumulates in the endothelial basement membrane and binds to striated collagen fibrils. It is suggested that the pericapillary collagen fibrils of fenestrated brain capillaries act as a charge filter with respect to macromolecules.     

Permeability of muscle capillaries to small heme-peptides. Evidence for the existence of patent transendothelial channels

Two heme-peptides (HP) of about 20-A diameter (heme-undecapeptide [H11P], mol wt approximately 1900 and heme-octapeptide [H8P], mol wt approximately 1550), obtained by enzymic hydrolysis of cytochrome c, were sued as probe molecules in muscle capillaries (rat diaphragm). They were localized in situ by a perixidase reaction, enhanced by the addition of imidazole to the incubation medium. Chromatography of plasma samples showed that HPs circulate predominantly as monomers for the duration of the experiments and are bound by aldehyde fixatives to plasma proteins to the extent of approximately 50% (H8P) to approximately 95% (H11P). Both tracers cross the endothelium primarily via plasmalemmal vesicles which become progressively labeled (by reaction product) from the blood front to the tissue front of the endothelium, in three successive resolvable phases. By the end of each phase the extent of labeling reaches greater than 90% of the corresponding vesicle population. Labeled vesicles appear as either isolated units or chains which form patent channels across the endothelium. The patency of these channels was checked by specimen tilting and graphic analysis of their images. No evidence was found for early or preferential marking of the intercellular junctions and spaces by reaction product. It is concluded that the channels are the most likely candidate for structural equivalents of the small pores of the capillary wall since they are continuous, water-filled passages, and are provided with one or more strictures of less than 100 A. Their frequency remains to be established by future work.     

Quantitative immunocytochemical studies of endogenous albumin in rat aortic endothelial and mesothelial cells

Endogenous albumin was revealed over cellular structures of rat ascendent aorta endothelia and mesothelium, with high resolution and specificity, by applying the protein A-gold immunocytochemical approach. This approach allows albumin distribution to be studied under steady-state conditions. The cellular layers evaluated were the aortic endothelium, the capillary endothelium (vasa vasorum), and the mesothelium externally lining the aorta at this level. Gold particles, revealing albumin antigenic sites, were preferentially located over plasmalemmal vesicles and intercellular clefts of endothelial and mesothelial cells, though with different labeling intensities. The interstitial space was also labeled. Morphometrical evaluation of plasmalemmal vesicles demonstrated a higher surface density for these structures in capillary endothelial cells (12%) compared with those in aortic endothelial (5%) and mesothelial cells (2%). Quantitation of gold labeling intensities over these structures revealed a higher labeling over plasmalemmal vesicles of capillary endothelium than over those of aortic endothelium and mesothelium. This result, together with the higher surface density of plasmalemmal vesicles found in capillary endothelium, suggest an important role of these structures in the transendothelial passage of endogenous albumin, particularly for capillary endothelium. On the other hand, labeling densities over mesothelial clefts were found to be higher than those of capillary and aortic endothelia. Results from this study concur with the proposal of a differential passage of albumin according to the cell lining considered, and suggest to a role for mesothelial intercellular clefts in contributing to the presence of albumin in interstitial spaces.     

STUDIES ON BLOOD CAPILLARIES : II. Transport of Ferritin Molecules across the Wall of Muscle Capillaries

The pathway by which intravenously injected ferritin molecules move from the blood plasma across the capillary wall has been investigated in the muscle of the rat diaphragm. At 2 min after administration, the ferritin molecules are evenly distributed in high concentration in the blood plasma of capillaries and occur within vesicles along the blood front of the endothelium. At the 10-min time point, a small number of molecules appear in the adventitia, and by 60 min they are relatively numerous in the adventitia and in phagocytic vesicles and vacuoles of adventitial macrophages. Thereafter, the amount of ferritin in the adventitia and pericapillary regions gradually increases so that at 1 day the concentration in the extracellular spaces approaches that in the blood plasma. Macrophages and, to a lesser extent, fibroblasts contain large amounts of ferritin. 4 days after administration, ferritin appears to be cleared from the blood and from the capillary walls, but it still persists in the adventitial macrophages and fibroblasts. At all time points examined, ferritin molecules within the endothelial tunic were restricted to vesicles or to occasional multivesicular or dense bodies; they were not found in intercellular junctions or within the cytoplasmic matrix. Ferritin molecules did not accumulate within or against the basement membranes. Over the time period studied, the concentration of ferritin in the blood decreased, first rapidly, then slowly, in two apparently exponential phases. Liver and spleen removed large amounts of ferritin from the blood. Diaphragms fixed at time points from 10 min to 1 day, stained for iron by the Prussian Blue method, and prepared as cleared whole mounts, showed a progressive and even accumulation of ferritin in adventitial macrophages along the entire capillary network. These findings indicate: (1) that endothelial cell vesicles are the structural equivalent of the large pore system postulated in the pore theory of capillary permeability; (2) that the basement membrane is not a structural restraint in the movement of ferritin molecules across the capillary wall; (3) that transport of ferritin occurs uniformly along the entire length of the capillary; and (4) that the adventitial macrophages monitor the capillary filtrate and partially clear it of the tracer.     

Increased capillary permeability in muscularis layer of rat intestine caused by kidney extract.

Kidney extract and synthetic angiotensin II were injected into bilaterally nephrectomized rats in dosages capable of raising the mean arterial pressure by about 20 mmHg. Changes in ultrastructure and permeability for ferritin molecules were then examined in capillaries located in muscularis layer of the intestinal walls. Kidney extract with a high renin content was obtained from the renal cortex of rats by means of stepwise centrifugation methods. Animals injected with saline served as controls. In rats receiving kidney extract tissue edema was observed in the spaces around the blood and lymphatic capillaries. In these spaces ferritin molecules accumulated in high concentration indicating plasma protein leakage. Ferritin molecules within the endothelium were restricted within plasmalemmal vesicles, but were not found within interendothelial junctions or within the cytoplasmic matrix. Morphometric analysis of vesicular transport in the endothelial cells revealed a significant increase in labeling rate for the vesicles with ferritin molecules. These results suggest that the kidney extract contains substance(s) which increase capillary permeability for plasma proteins at least via increased vesicular transport, resulting in tissue edema.     

Specific binding sites for albumin restricted to plasmalemmal vesicles of continuous capillary endothelium: receptor-mediated transcytosis

The interaction of homologous and heterologous albumin-gold complex (Alb-Au) with capillary endothelium was investigated in the mouse lung, heart, and diaphragm. Perfusion of the tracer in situ for from 3 to 35 min was followed by washing with phosphate-buffered saline, fixation by perfusion, and processing for electron microscopy. From the earliest time examined, one and sometimes two rows of densely packed particles bound to some restricted plasma membrane microdomains that appeared as uncoated pits, and to plasmalemmal vesicles open on the luminal front. Morphometric analysis, using various albumin-gold concentrations, showed that the binding is saturable at a very low concentration of the ligand and short exposure. After 5 min, tracer-carrying vesicles appeared on the abluminal front, discharging their content into the subendothelial space. As a function of tracer concentration 1-10% of plasmalemmal vesicles contained Alb-Au particles in fluid phase; from 5 min on, multivesicular bodies were labeled by the tracer. Plasma membrane, coated pits, and coated vesicles were not significantly marked at any time interval. Heparin or high ionic strength did not displace the bound Alb-Au from vesicle membrane. No binding was obtained when Alb-Au was competed in situ with albumin or was injected in vivo. Gold complexes with fibrinogen, fibronectin, glucose oxidase, or polyethyleneglycol did not give a labeling comparable to that of albumin. These results suggest that on the capillary endothelia examined, the Alb-Au is adsorbed on specific binding sites restricted to uncoated pits and plasmalemmal vesicles. The tracer is transported in transcytotic vesicles across endothelium by receptor-mediated transcytosis, and to a lesser extent is taken up by pinocytotic vesicles. The existence of albumin receptors on these continuous capillary endothelia may provide a specific mechanism for the transport of albumin and other molecules carried by this protein.     

An electron microscopic study of the transport of peroxidases in the endothelium of mouse aorta

Summary Aortic endothelium presents a continuous barrier to diffusion of macromolecules. The cell margins overlap for long distances and there are multiple points of contact between the cell membranes at which the intercellular cleft is reduced to 30–40 Å or less, and free diffusion of lanthanum is impeded at some points of apposition. Macromolecular transport through the endothelium of mouse aorta was studied with the help of horseradish peroxidase (HRP) and bovine milk lactoperoxidase. Following injection of 0.25–0.5 mg of HRP no tracer was detected in the intercellular clefts even though it was seen in plasmalemmal vesicles and subendothelial space. However, when 5 mg of HRP was injected in either 0.05 or 0.5 ml of saline, transport of the enzyme occurred through both the intercellular clefts and via the plasmalemmal vesicles. On the other hand, lactoperoxidase of m.w. 82000 was transported through the plasmalemmal vesicles only. The findings were discussed with reference to the transport of serum lipoproteins and it was suggested that low and high density lipoproteins would be transported via the plasmalemmal vesicles.The excellent technical help of Miss R. Ben-Moshe and Mrs. A. Mandeles is gratefully acknowledged. This study was supported in part by a grant from the Myra Kurland Heart Fund, Chicago, Ill., and by a grant 06-101-1 of the National Institute of Health, United States Public Health Service.     

Distribution of endothelial vesicles in the microvasculature of skeletal muscle and brain cortex of the rat,as demonstrated by tannic acid tracer analysis

Summary Ultrathin serial sectioning and labeling with tannic acid have demonstrated that most plasmalemmal vesicles of rat vascular endothelial cells are not free, but rather are conjoined in three dimensions to form racemose invaginations from the cell surfaces. To elucidate the distribution of vesicles in these microvascular endothelial cells, we have examined terminal arterioles, capillaries and post-capillary venules of rat skeletal muscle and brain cortex, using tannic acid labeling and stereological methods, and have determined the proportions of free vesicles and the vesicles of luminal and abluminal invaginations, as well as the numerical density of vesicles. In the case of capillaries, regional differences in distribution have also been studied. The ratio of free vesicles is 6–7% and is constant throughout the muscle microvasculature. The distribution (proportions and numerical densities) of vesicles in the brain and muscle microvascular endothelial cells shows regionally distinctive patterns. In rapid-frozen, freeze-substituted endothelial cells, there are almost as many fused vesicles as seen in chemically fixed cells. Therefore, aldehydes do not seem to induce membrane fusion, and the distribution of vesicles seems to be preserved by chemical fixation. The structure and function of plasmalemmal vesicles are discussed.     

Transcytosis of albumin in capillary endothelium

We have used a variety of immunocytochemical procedures to localize albumin in transit through the capillary endothelium of the murine myocardium and thereby identify endothelial cell structures involved in albumin efflux. The most informative results were obtained with a protocol that included (a) removal of endogenous albumin by perfusion of the heart with PBS supplemented with 14 mM glucose, (b) perfusion of the heart vasculature with exogenous (bovine) albumin for various short time periods, (c) fixation of the vessels by formaldehyde- glutaraldehyde mixtures, (d) processing of fixed myocardium specimens through L. R. White embedding followed by sectioning, or direct thin frozen sectioning, and (e) reacting the surface of such specimens with antialbumin antibodies followed by gold-labeled secondary antibodies. The results indicate that (a) monomeric albumin binds (with low affinity) to the luminal surface of the capillary endothelium, (b) it is restricted in transit through the endothelium to plasmalemmal vesicles, and (c) it appears in the pericapillary spaces less than 15 s after the beginning of its perfusion. No albumin concentration gradients, centered with their maxima on the exits from intercellular spaces, were detected at any time points, including the shortest ones (15 and 30 s) investigated. Additional information comparing monomeric vs. polymeric albumin transcytosis was obtained using albumin-gold complexes. The results are discussed in terms of vesicular transport of albumin across the endothelium and the relations of this type of transport to the postulated pore systems of the physiological literature.     

Immunoelectron microscopic visualization of the transcytosis of low density lipoproteins in perfused rat arteries

A postembedding labeling technique was employed to visualize human native low density lipoproteins (LDL) during transcytosis in rat arterial endothelium. For this purpose human LDL was perfused through rat vasculature before fixation and processing for immunoelectron microscopy. The LDL particles were located on sections by anti-human apolipoprotein B-100 (LDL) antibodies and secondary antibodies or protein-A conjugated to 10-nm colloidal gold. LDL molecules were seen in plasmalemmal vesicles as well as in the subendothelial space. No colloidal gold was found in the intercellular junctions. Perfusion with reductively methylated LDL, which cannot bind to the LDL receptor, gave a similar labeling pattern, indicating that transcytosis of LDL via plasmalemmal vesicles is most likely receptor independent. Furthermore, the passage of LDL through intact vascular endothelium is a vesicular transport rather than an intercellular diffusion process.     

FINE STRUCTURAL LOCALIZATION OF A BLOOD-BRAIN BARRIER TO EXOGENOUS PEROXIDASE

Horseradish peroxidase was administered to mice by intravenous injection, and its distribution in cerebral cortex studied with a recently available technique for localizing peroxidase with the electron microscope. Brains were fixed by either immersion or vascular perfusion 10–60 min after administration of various doses of peroxidase. Exogenous peroxidase was localized in the lumina of blood vessels and in some micropinocytotic vesicles within endothelial cells; none was found beyond the vascular endothelium. Micropinocytotic vesicles were few in number and did not appear to transport peroxidase while tight junctions between endothelial cells were probably responsible for preventing its intercellular passage. Our findings therefore localize, at a fine structural level, a "barrier" to the passage of peroxidase at the endothelium of vessels in the cerebral cortex. The significance of these findings is discussed, particularly with reference to a recent study in which similar techniques were applied to capillaries in heart and skeletal muscle.     

Electron microscopic research on the capillary permeability of the primary portal plexus in rats in the prenatal period

Permeability of portal capillaries to intravascularly injected ionic lanthanum, ferritin and horse-radish peroxidase has been examined in rats on the 18th fetal day, and on days 1 and 9 of postnatal life. For several minutes, tracer molecules pass through the capillary wall and reach the median eminence. In the case of immature capillaries, the materials pass freely through the endothelial cells, and to a lesser extent are transferred via occasional plasmalemmal vesicles and fenestrae. As the maturation of capillaries proceeds their permeability via plasmalemmal vesicles and fenestrae increases considerably due to a gradual rise in the number of these structures. The plasmalemma of differentiated endothelial cells becomes impermeable to all the tracers. Only ionic lanthanum appears to penetrate through transendothelial channels and intercellular junctions between adjacent endothelial cells.     

STUDIES ON THE PERMEABILITY OF LYMPHATIC CAPILLARIES

The passageway for interstitial fluids and large molecules across the connective tissue lymph interface has been investigated in dermal lymphatic capillaries in the ears of guinea pigs. Numerous endothelial cells overlap extensively at their margins and lack adhesion devices at many points. The observations suggest that these sites are free to move as a result of slight pressure changes. Immediately following interstitial injections of tracer particles (ferritin, thorium, carbon, and latex spheres), many of the overlapped endothelial cells are separated and thus passageways are provided between the interstitium and lymphatic lumen. Tracer particles also occur in plasmalemmal invaginations along both connective tissue and luminal fronts. All of the tracer particles accumulate within large autophagic-like vacuoles. Very few particles of ferritin are observed in the endothelium after 24 hr; however, the vesicles containing the nonprotein tracer particles (carbon, thorium, and latex) increase in size and content and remain within the lymphatic endothelial cells up to 6 months. The role of vesicles in the transport of large molecules and particles is discussed in relation to the accretion of tracer particles within large vesicles and autophagic-like vacuoles in the endothelial cytoplasm.     

The blood-brain barrier in a reptile, Anolis carolinensis

An electron microscopic study was made of the ultrastructure and permeability of the capillaries in the cerebral hemispheres of the lizard, Anolis carolinensis. The brain of Anolis is vascularized by a loop-type pattern consisting exclusively of arteriovenous capillary loops. The ultrastructure of the endothelium and the arrangement of the various layers from the capillary lumen to the central nervous tissue is similar to that of mammals. The endothelial cells form a continuous layer around the lumen and are joined by tight interendothelial junctions. The basal lamina of the endothelium is also continuous and encloses pericyte processes. The cells of the nervous tissue rest directly on the basal lamina of the capillary and are separated from each other by a 200 Å space. Intravenously injected horseradish peroxidase (MW 40,000) and ferritin (MW 500,000) were used to study the permeability of the capillaries. The entry of horseradish peroxidase and ferritin into the intercellular spaces of the brain is restricted by the tightness of the interendothelial junctions. No vesicular transport of either tracer occurs; however, ferritin does enter the endothelial cells in vacuoles. No tracer molecules are present in the basal lamina, pericytes, or nervous tissue. The different responses of the endothelial cell to the tracers used in this study suggest that endocytotic activities of endothelial cells involve different processes. Vacuoles formed by marginal folds, vacuoles formed by endothelial surface projections or deep invaginations of the plasma membrane, 600–800 Å vesicles, and coated vesicles all seem to differ in the nature of the substances which they endocytose.     

Opening of blood-brain barrier in Triturus cristatus carnifex by hyperosmolar mannitol solutions

The permeability of the newt cerebral capillaries to lanthanum ion has been studied after perfusion with mannitol solutions of increasing molarity. In the control specimens lanthanum deposits were limited to the luminal side of the capillaries and tracer did not spread to the pericapillary spaces due to the tight junctions. Treatment with hypertonic solutions of mannitol (0.25M, 0.5M, 1M) caused opening of the blood brain barrier with a progressive increase in lanthanum between the endothelial cell edges, in the basal lamina and in the extracellular spaces of the nervous parenchyma in relation to the molarity of the mannitol solution. The spread of lanthanum is probably due to opening of the tight junctions between the endothelial cells, since pinocytotic vesicles labelled with tracer were not evident.     

Transendothelial transport of serum albumin: a quantitative immunocytochemical study

The steady-state distribution of endogenous albumin in mouse diaphragm was determined by quantitative postembedding protein A-gold immunocytochemistry using a specific anti-mouse albumin antibody. Labeling density was recorded over vascular lumen, endothelium, junctions, and subendothelial space. At equilibrium, the volume density of interstitial albumin was 18% of that in circulation. Despite this large difference in albumin concentration between capillary lumen and interstitium, plasmalemmal vesicles labeling was uniformly distributed across the endothelial profile. 68% of the junctions displayed labeling for albumin, which was however low and confined to the luminal and abluminal sides. The scarce labeling of the endothelial cell surface did not confirm the fiber matrix theory. The kinetics of albumin transcytosis was evaluated by injecting radioiodinated and DNP-tagged BSA. At 3, 10, 30, and 60 min, and 3, 5, and 24 h circulation time, blood radioactivity was measured and diaphragms were fixed and embedded. Anti-DNP antibodies were used to map the tracer in aforementioned compartments. A linear relationship between blood radioactivity and vascular labeling density was found, with a detection sensitivity approaching 1 gold particle per DNP-BSA molecule. Tracer presence over endothelial vesicles reached rapidly (10 min) a saturation value; initially localized near the luminal front, it evolved towards a uniform distribution across endothelium during the first hour. An hour was also needed to reach the saturation limit within the subendothelial space. Labeling of the junctions increased slowly, out of phase with the inferred transendothelial albumin fluxes. This suggests that they play little, if any, role in albumin transcytosis, which rather seems to proceed through the vesicular way.