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.     

A radioautographic study of the transport of 125I-labeled serum lipoproteins in rat aorta

Summary The transport of ¹²⁵I-labeled serum lipoproteins through the aortic endothelium was studied by radioautography. Rat aorta and heart was perfused in vitro with a medium containing human very low density (VLDL), low density (LDL), high density lipoprotein (HDL), delipidated HDL apolipoprotein or rat HDL. In all lipoproteins more than 95% of the radioactivity was TCA precipitable and lipid radioactivity was from 2–4% in HDL, 4–6% in LDL, 7–15% in VLDL. After 18–60 min of perfusion and wash with unlabeled medium, most of the aortic radioactivity was TCA precipitable and the percent of lipid counts was similar to that in the original lipoprotein. Following perfusion with VLDL, LDL, or HDL the radioautographic reaction was seen over the endothelium, the subendothelial space and the inner media, and was separated by an unlabeled zone from the reaction present over the adventitia. Uniform labeling of the entire wall was found after perfusion with HDL apolipoprotein. The presence of silver grains over endothelial cells in regions rich in plasmalemmal vesicles suggested that these organelles participate in the transport of the labeled lipoprotein, as was shown for lactoperoxidase (Stein and Stein, 1972). The present data indicate that cholesterol may enter the aortic wall as a constituent of lipoprotein particles. Since an HDL particle carries less than 1/20 of the cholesterol present in a LDL particle, it seems that the lower susceptibility of the female to atheromatosis might be related to the higher ratio of HDL to LDL particles in the female serum.The excellent technical help of Miss R. Ben-Moshe, Mrs. A. Mandeles, Mr. G. Hollander and Mrs. Y. Dabach is gratefully acknowledged. This study was supported in part by grants from National Institute of Health (No. 06-101-1), United States Public Health Service; Delegation Generale a la Recherche Scientifique et Technique of the French Government and from the Ministry of Health, the Government of Israel.     

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.     

Uptake of horseradish peroxidase from CSF into the choroid plexus of the rat,with special reference to transepithelial transport

Summary Protein uptake from cerebral ventricles into the epithelium of the choroid plexus, and transport across the epithelium were studied ultrastructurally in rats. Horseradish peroxidase (HRP, MW 40,000) was used as protein tracer. Steady-state ventriculo-cisternal perfusion with subatmospheric pressure (-10cm of water) in the ventricular system was applied. HRP dissolved in artificial CSF was perfused from the lateral ventricles to cisterna magna for various times, and ventriculo-cisternal perfusion, vascular perfusion or immersion fixation with a formaldehyde-glutaraldehyde solution was performed.Coated micropinocytic vesicles containing HRP were seen both connected with the apical, lateral and basal epithelial surface and within the cells. Heavily HRP-labeled vesicles were often fused with the lining membrane of slightly labeled or unlabeled intercellular spaces. Since the apical tight junctions of the epithelium never appeared open or never contained HRP in the spaces between the fusion points, and since the intercellular spaces between adjacent epithelial cells below the junctions only infrequently contained tracer after 5 min, by increasing amounts after 15–60 min of HRP perfusion, a vesicular transport of HRP from the apical epithelial surface to the intercellular spaces, bypassing the tight junctions, is suggested.In addition to the transepithelial transport, micropinocytic vesicles also transported HRP to the lysosomal apparatus of the epithelial cells. With increasing length of exposure to HRP, a sequence of HRP-labeled structures could be evaluated, from slightly labeled apical vacuoles and multivesicular bodies to very heavily labeled dense bodies.     

Immunogold localisation of endogenous immunoglobulin-G in ultrathin frozen sections of the human placenta

Summary Endogenous immunoglobulin-G was localised in ultrathin frozen sections of human term placenta by use of an indirect immuno electron-histochemical methodology. Immunoreactivity of endogenous IgG to rabbit anti-human immunoglobulin-G antibody was visualised by use of protein-A — colloidal gold complex. Gold marked the syncytiotrophoblast in both coated and uncoated regions of the apical plasmalemma, in vesicles and multivesicular bodies, and in vesicles near the basal plasmalemma. Immunoreactivity was also seen in the interstitial space between the trophoblast and the fetal endothelial layer as well as in various types of vesicles within the endothelial cells. No immunoreactivity was seen in the intercellular clefts of the endothelium. The pattern of localisation observed is consistent with receptor-mediated uptake of immunoglobulin-G into the syncytiotrophoblast of the human placenta followed by release into the interstitial space and then vesciular transport through the endothelium.     

Horseradish peroxidase as a tracer for capillary permeability studies

The permeability of capillaries was investigated utilizing an in vivo injection of horseradish peroxidase (HRP) and an in situ perfusion of a balanced salt solution containing HRP and lanthanum chloride. In the continuous capillaries of heart and muscle, HRP diffuses mainly through intercellular junctions, while in testicular capillaries, the transport is via micropinocytotic vesicles. The diffusion and micropinocytotic transport of HRP was demonstrated in both directions, i.e. from the capillary lumen to the interstitium and vice versa. Lanthanum can be used as a bidirectional inhibitor of micropinocytosis. The transport of HRP is then almost completely hindered in testicular capillaries. In heart muscle, the effect on HRP transport is not significant, due to second transport pathway, i.e. intercellular cleft passage.     

ULTRASTRUCTURAL STUDIES ON THE PERMEABILITY OF THE MESOTHELIUM TO HORSERADISH PEROXIDASE

Peritoneal mesothelium was exposed for 2–60 min to solutions of horseradish peroxidase by incubation in vitro, or after intraperitoneal injection in vivo. Peroxidase was localized, with the electron microscope in the intercellular clefts of the mesothelium, often along their entire lengths, in vesicles adjoining or contiguous with the clefts, and along the peritoneal and basal surfaces of the cell, and also in intracytoplasmic vacuoles. The intercellular junctions of peroxidase-treated mesothelium did not differ from those of controls: open and closed junctions were present in both groups. Intercellular localization was also obtained when the mesothelium was exposed to peroxidase during or after fixation. Although intracellular absorption of peroxidase and its incorporation into larger vacuoles were observed, there was no clearcut evidence of vesicular transport across the mesothelium in these experiments. These findings are consistent with physiologic data which postulate that mesothelial transport can be accounted for, at least in part, by passive diffusion through a system of pores, and they suggest that these pores are located in the intercellular clefts.     

PERMEABILITY OF MUSCLE CAPILLARIES TO EXOGENOUS MYOGLOBIN

Whale skeletal muscle myoglobin (mol wt 17,800; molecular dimensions 25 x 34 x 42 Å) was used as a probe molecule for the pore systems of muscle capillaries. Diaphragms of Wistar-Furth rats were fixed in situ at intervals up to 4 h after the intravenous injection of the tracer, and myoglobin was localized in the tissue by a peroxidase reaction. Gel filtration of plasma samples proved that myoglobin molecules remained in circulation in native monomeric form. At 30–35 s postinjection, the tracer marked ~75% of the plasmalemmal vesicles on the blood front of the endothelium, 15% of those located inside and none of those on the tissue front. At 45 s, the labeling of vesicles in the inner group reached 60% but remained nil for those on the tissue front. Marked vesicles appeared on the latter past 45 s and their frequency increased to ~80% by 60–75 s, concomitantly with the appearance of myoglobin in the pericapillary spaces. Significant regional heterogeneity in initial labeling was found in the different segments of the endothelium (i.e., perinuclear cytoplasm, organelle region, cell periphery, and parajunctional zone). Up to 60 s, the intercellular junctions and spaces of the endothelium were free of myoglobin reaction product; thereafter, the latter was detected in the distal part of the intercellular spaces in concentration generally equal to or lower than that prevailing in the adjacent pericapillary space. The findings indicate that myoglobin molecules cross the endothelium of muscle capillaries primarily via plasmalemmal vesicles. Since a molecule of this size is supposed to exit through both pore systems, our results confirm the earlier conclusion that the plasmalemmal vesicles represent the large pore system; in addition, they suggest that the same structures are, at least in part, the structural equivalent of the small pore system of this type of capillaries.     

Impermeability of hagfish cerebral capillaries to horseradish peroxidase

Summary Brain capillaries and their permeability to intravenously injected horseradish peroxidase, HRP, (MW: 40,000) were examined electron-microscopically in an attempt to find a structural explanation for the poorly developed blood-brain barrier in the hagfish, Myxine glutinosa. In particular, it was the aim of this study to examine the role of the numerous endothelial vesicles and tubules in the transport of this tracer between blood and brain. Many of the vesicles and tubules were found to be in continuity with the luminal or abluminal surfaces, but tubules generating channels through the endothelial cells were never observed. The cleft between adjacent endothelial cells was obliterated by punctate junctions. HRP, which was allowed to circulate for up to 35 min, was not found in the basal lamina or in the surrounding brain parenchyma. Few of the luminal vesicles and tubules were marked by the tracer. In the intercellular cleft HRP was stopped by the junctions. It is concluded that the hagfish like other vertebrates has a blood-brain barrier to HRP, and the numerous vesicles and tubules occurring in hagfish brain endothelium are not involved in the transendothelial transport of this macromolecule.     

The ultrastructural basis of alveolar-capillary membrane permeability to peroxidase used as a tracer

The permeability of the alveolar-capillary membrane to a small molecular weight protein, horseradish peroxidase (HRP), was investigated by means of ultrastructural cytochemistry. Mice were injected intravenously with HRP and sacrificed at varying intervals. Experiments with intranasally instilled HRP were also carried out. The tissue was fixed in formaldehyde-glutaraldehyde fixative. Frozen sections were cut, incubated in Graham and Karnovsky's medium for demonstrating HRP activity, postfixed in OsO4, and processed for electron microscopy. 90 sec after injection, HRP had passed through endothelial junctions into underlying basement membranes, but was stopped from entering the alveolar space by zonulae occludentes between epithelial cells. HRP was demonstrated in pinocytotic vesicles of both endothelial and epithelial cells, but the role of these vesicles in net protein transport appeared to be minimal. Intranasally instilled HRP was similarly prevented from permeating the underlying basement membrane by epithelial zonulae occludentes. Pulmonary endothelial intercellular clefts stained with uranyl acetate appeared to contain maculae occludentes rather than zonulae occludentes. HRP did not alter the ultrastructure of these junctions.     

Endocytotic pathways across the human gall-bladder mucosa: permeability studies using horseradish peroxidase

Summary Human gall-bladder epithelium obtained straight from the operating theatre was incubated in an Ussing chamber with the fluid phase marker, horseradish peroxidase (HRP), for up to 60 min. When the marker was presented on the apical surface, within 30 min it had moved readily across the apical cytoplasm in transport vesicles to receptosomes and into the lateral intercellular space, extending across the basement membrane into the lamina propria. When HRP was presented at the basal aspect, within 30 min it had moved through the lamina propria, across the basement membrane and into the lateral intercellular space. By 60 min, only small amounts had been taken up by the epithelial cells and transported to receptosomes. These data indicate a rapid transmucosal endocytotic pathway for blood-or bile-borne macromolecules.     

Endocardial endothelium in the rat: junctional organization and permeability

Selective permeability of endocardial endothelium has been suggested as a mechanism underlying the modulation of the performance of subjacent myocardium. In this study, we characterized the organization and permeability of junctional complexes in ventricular endocardial endothelium in rat heart. The length of intercellular clefts viewed en face per unit endothelial cell surface area was lower, and intercellular clefts were deeper in endocardial endothelium than in myocardial vascular endothelium, whereas tight junctions had a similar structure in both endothelia. On this basis, endocardia endothelium. might be less permeable than capillary endothelium. However, confocal scanning laser microscopy showed that intravenously injected dextran 10000 coupled to Lucifer Yellow penetrated first the endocardial endothelium and later the myocardial capillary endothelium. Penetration of dextran 10000 in myocardium occurred earlier through subepicardial capillary endothelium than through subendocardial capillary endothelium. Penetration of tracer might thus be influenced by hydrostatic pressure. Dextran of MW 40000 did not diffuse through either endocardial endothelium or capilary endothelium. The ultrastructure of endocardial endothelium may constitute an adaptation to limit diffusion driven by high hydrostatic pressure in the heart. Differences in paracellular diffusion of dextran 10000 between endocardial endothelium and myocardial vessels, may result from differing permeability properties of the endocardium and underlying myocardium.     

Immunocytochemical and labelled tracer approaches to uptake and intracellular routing of Immunoglobulin-G (IgG) in the human placenta

Summary Isolated lobules of freshly delivered human term placenta were (a) subjected to an indirect immunoelectron ultracryo method in which the immunoreactivity of endogenous Immunoglobulin-G (IgG) to rabbit anti-human IgG antibody was localized with protein-A-colloidal gold and (b) extracorporeally perfused and human IgG molecules complexed to horseradish peroxidase (HRP) added to the maternal perfusate and the uptake of IgG-HRP over different perfusion durations visualized ultrastructurally by using diaminobenzidine cytochemistry. Immunoreactivity to anti-human IgG antibody was localized all along the apical plasmalemma, in apical coated and uncoated vesicles, in apical and juxtanuclear multivesicular bodies, and in basal vesicles of the syncytiotrophoblast layer of the placenta. The stroma separating the syncytiotrophoblast from the foetal endothelium as well as vesicles within the endothelium were immunoreactive. No immunoreactivity was localized in paracellular clefts of endothelia. A similar distribution of exogenous IgG-HRP was observed for the perfused placentae. When bovine IgG-HRP or HRP alone were used as control tracers no uptake was seen for the former whilst the latter was observed only in early endosomal vesicles of the syncytiotrophoblast. The pattern of localization visualized in both studies is consistent with receptor-mediated uptake of IgG by the syncytiotrophoblast and a vesicular transport of IgG across the foetal endothelium.     

Endocytosis in the uterine luminal and glandular epithelial cells of mice during early pregnancy

We have localized horseradish peroxidase (HRP) in the mouse uterus after intravenous administration on days 1 and 5 of pregnancy in an effort to understand how serum proteins reach the uterine lumen. Direct movement of HRP into uterine and glandular lumina was blocked by the epithelial tight junctions on both days. In luminal and glandular epithelial cells at both times, HRP was localized in endocytic vesicles along the basolateral membranes, multivesicular bodies (mvb), elongated dense bodies below the nucleus (bdb), and many small vesicles near the apical surface of the cells. The uptake of HRP was most extensive in the luminal epithelium on day 1: the number of tracer-containing apical vesicles and bdb was largest, and there were also clusters of vesicles containing the tracer above the nucleus. Acid phosphatase was localized on day 1 in mvb and bdb in both cell types, indicating that these structures are lysosomes. It appeared that HRP followed two pathways after basolateral endocytosis by the epithelial cells: it was transported to the apical region of the cells, where it was present in small vesicles that may release their contents into the uterine or glandular lumina, or it was transported to lysosomes. To investigate whether macromolecules may be transported from the uterine lumen to the stroma, we also studied endocytosis at the apical pole of luminal epithelial cells after intraluminal injection of HRP. There was no detectable uptake of HRP from the lumen on day 1, and no tracer was detected in the intercellular spaces or basement membrane region. On day 5, a large amount of HRP was taken up from the lumen into apical endocytic vesicles, mvb, and dense bodies, but tracer was not present in the Golgi apparatus, lateral intercellular spaces, or the basement membrane region at the times studied. These observations indicate that there was no transport of luminal macromolecules to the uterine stroma on day 1, while the possibility of transport on day 5 requires further study.     

Binding and transcytosis of glycoalbumin by the microvascular endothelium of the murine myocardium: evidence that glycoalbumin behaves as a bifunctional ligand

The binding and transport of glycoalbumin (gA) by the endothelium of murine myocardial microvessels were studied by perfusing in situ 125I-gA or gA-gold complexes (gA-Au) and examining the specimens by radioassays and EM, respectively. After a 3-min perfusion, the uptake of radioiodinated gA is 2.2-fold higher than that of native albumin; it is partially (approximately 55%) competed by either albumin or D-glucose, and almost completely abolished by the concomitant administration of both competitors or by gA. D-mannose and D-galactose are not effective competitors. Unlike albumin-gold complexes that bind restrictively to plasmalemmal vesicles, gA-Au labels the plasma-lemma proper, plasmalemmal vesicles open on the lumen, and most coated pits. Competing albumin prevents gA-Au binding to the membrane of plasmalemmal vesicles, while glucose significantly reduces the ligand binding to plasmalemma proper. Competition with albumin and glucose gives additive effects. Transcytosis of gA-Au, already detected at 3 min, becomes substantial by 30 min. No tracer exit via intercellular junctions was detected. gA-Au progressively accumulates in multivesicular bodies. The results of the binding and competition experiments indicate that the gA behaves as a bifunctional ligand which is recognized by two distinct binding sites: one, located on the plasma membrane, binds as a lectin the glucose residues of gA; whereas the other, confined to plasmalemmal vesicles, recognizes presumably specific domains of the albumin molecule.     

Transtubular transport of proteins in rabbit proximal tubules

The purpose of the present experiments was to study possible different pathways of intracellular transport of proteins after luminal and basolateral uptake in isolated rabbit proximal tubules. Tubules were exposed to cationized ferritin (CF) in the perfusion fluid and horseradish peroxidase (HRP) in the bath simultaneously or to HRP in the bath alone for 30 min. The peritubular fluid (bath) and perfusion fluid were then exchanged and the tubules either fixed immediately or allowed to function during chase-periods for 10, 20, 30, or 60 min before fixation to follow the migration of the proteins through the cells. The proteins were to a large extent found separated in different vacuoles and lysosomes at all time periods studied, indicating separate pathways after uptake via the luminal and basolateral membranes respectively. About 0.5% of the CF taken up by the cells was transported through the cells and became located in the intercellular spaces. HRP was transported from the peritubular fluid to the apical cytoplasm of the tubules indicated by a gradual accumulation of small HRP-containing vesicles, first in the basal part of the cells and then in the apical cytoplasm. In tubules perfused with both CF and HRP in the perfusate, the CF and HRP were found together in apical vacuoles and lysosomes. After perfusion with HRP alone, this tracer was found in similar large vacuoles and lysosomes in the apical cytoplasm, in contrast to the small HRP-filled vacuoles seen after uptake from the bath.     

Fatty acids binding to albumin increases its uptake and transcytosis by the lung capillary endothelium

To determine whether uptake and transcytosis of albumin (A) in continuous capillary endothelia are modified when this protein carries fatty acids, the transport of albumin-oleic acid and albumin-palmitic acid complexes was compared with that of defatted albumin. The probes, either radioiodinated or tagged with 5-nm gold particles (Au), or both, were perfused in situ or injected in vivo; after 3 or 30 min lung fragments were radioassayed or examined by electron microscopy. Both in situ and in vivo, the uptake of fatty acid-carrying albumin (A-FA) was consistently 2 to 3 times higher than that of defatted A. Electron microscopy revealed that A-FA complexes tagged with gold were taken up and transported across the endothelium by plasmalemmal vesicles. Morphometric analysis showed that as compared with A-Au, at 3 min the density of (A-FA)Au bound to plasmalemmal vesicles was 2 to 3 times higher, and the extent of transcytosis was increased. Uptake of the iodinated albumin was more effectively competed by A-FA complexes than by defatted A, suggesting a higher affinity of the former for the albumin binding sites of the endothelium. The results indicate that when carrying fatty acids, albumin is taken up specifically and with high affinity by the capillary endothelium, a process that may play a role in the transport of fatty acids from the plasma to the cells where they are metabolized.     

Vascular permeability (problem of the blood-brain barrier) in the pineal organ of the rainbow trout,Salmo gairdneri

Summary The problem of the blood-brain barrier in the pineal organ of the rainbow trout, Salmo gairdneri, was investigated following intraperitoneal or intracardial injections of several tracers and dyes with different molecular weights. As demonstrated at the light-microscopic level, repeated injections of trypan blue or horseradish peroxidase (HRP) resulted in an accumulation of these substances in the pineal epithelium (parenchyma). By use of the electron microscope, HRP was found in electron-dense bodies, probably lysosomes, in (i) the endothelial cells and perivascular macrophages 4 h after intraperitoneal injection, (ii) the supporting cells and intrapineal or luminal macrophages 8 h after injection, and (iii) the receptor cells 24 h after injection of the tracer. Ferritin particles penetrated the fenestrated endothelium of pineal capillaries. They were confined to vesicles, vacuoles and the smooth endoplasmic reticulum of the supporting cells as well as to the synaptic vesicles and the smooth endoplasmic reticulum of the pineal photoreceptors. The intercellular passage of tannic acid mixed with the fixative was blocked at the luminal junctional complex separating the pineal lumen from the basal portion of the pineal epithelium. The passive intercellular transport of substances with high molecular weight from the bloodstream to the cerebrospinal-fluid compartment is thus prevented. However, no blood-brain barrier exists for exogenously administered proteins, which are rapidly taken up by pineal cells and actively transported in a transcellular manner.The findings on the blood-brain barrier of the pineal organ of the rainbow trout are discussed with particular reference to the endocrine capacity of pineal sensory organs.Fellow of the Alexander von Humboldt Foundation, Federal Republic of Germany.     

Endocytosis in absorptive cells of cultured human small-intestinal tissue: Horseradish peroxidase,lactoperoxidase, and ferritin as markers

Summary The occurrence of endocytotic mechanisms in human small intestinal absorptive cells was investigated by culturing biopsy specimens in the presence of horseradish peroxidase (HRP), lactoperoxidase (LPO), and ferritin. The results indicate that both HRP and LPO entered the cells by apical endocytosis, after which they were transported via apical vesicles and tubules to the lysosome-like bodies. Ferritin, which showed a distinct affinity for the cell-coat glycoproteins, was not interiorized by the absorptive cells.These findings suggest that although human absorptive cells have an endocytotic mechanism, possibly fluid-phase endocytosis, cell-coat glycoproteins are not taken up by the cells, as indicated by the absence of ferritin in the apical vesicles and tubules, as well as the lysosome-like bodies. These findings provide indirect support for our hypothesis that the lysosome-like bodies have a function in the regulation of cell-coat glycoprotein transport via a crinophagic mechanism (fusion of apical vesicles and tubules with lysosome-like bodies) rather than via an exocytotic-endocytotic mechanism.     

Cytochemical observations on the transport of horseradish peroxidase in different segments of the nephron

Synopsis Kidney slices from rats injected with horseradish peroxidase (HRP) 5–10 min before sacrifice were fixed with formaldehyde vapour for 4 hr at 37°C and compared with tissue fixed by perfusion or by immersion. Much more of the injected protein was retained in extracellular and vascular spaces of the nephron in vapour-fixed than in perfusion-fixed or immersion-fixed tissue. The extracellular localization of HRP in the lateral intercellular spaces and in the infoldings of the basal cell membranes showed characteristic differences in different segments of the nephron. The high concentration of HRP in the lateral intercellular spaces of the collecting tubules, as well as the early location of small phagosomes containing HRP in the apical, lateral, and basal cell regions suggested that HRP was reabsorbed through the cytoplasm into the intercellular spaces or excreted in the opposite direction. The intercellular spaces in the terminal segments of the proximal tubules also showed high concentrations of HRP which suggests participation of these spaces in protein transport between the lumen and the peritubular capillaries. The extracellular concentration of HRP early after injection was found, by colorimetric assays of homogenates, to be several times higher in the papilla than in the cortex.