The cell surface of a restrictive fenestrated endothelium

Summary The location and chemical composition of anionic sites on the endothelium of the choriocapillaris was investigated with cationic ferritin and enzyme digestion techniques. Cationic ferritin administered intravenously initially labeled essentially all fenestral diaphragms. Within 30 min after injection, no diaphrams remained labeled, but they could be relabeled by a second cationic ferritin injection. Following perfusion of cationic ferritin, the entire luminal front of the endothelium was labeled: the plasmalemma and fenestral, vesicle, and channel diaphragms. Perfusion of neuraminidase or chondroitinase did not affect subsequent cationic ferritin binding. In contrast, heparitinase removed anionic sites on all structures except fenestral diaphragms. Cationic ferritin did not mark the endothelium following heparinase digestion. All sites were cleaved with pronase E. These results indicate that heparin is the anionic moiety on fenestral diaphragms while the glycocalcyces of the plasmalemma and vesicle and channel diaphragms are rich in a heparan sulfate proteoglycan. Furthermore, since the heparan sulfate localized to these structures was digested by both heparinase and heparitinase, it is in a form similar to heparin. These findings demonstrate that the endothelium of the choriocapillaris bears cell-surface anionic components that are different than those described for fenestrated endothelia lining other vascular beds.Supported by NIH EY 03776     

Redistribution of surface anionic sites on the luminal front of blood vessel endothelium after interaction with polycationic ligand

The ability of anionic groups on the luminal surface of blood vessels to redistribute by lateral migration under the influence of multivalent ligands was analyzed by electron microscopy, using cationized ferritin (CF). In vitro interaction of blood vessel segments with CF results in rapid aggregation of most anionic sites on the luminal fromt of the endothelium, followed by internalization or detachment of the CF patches, leaving most of the luminal surface devoid of anionic sites. Further incubation of such endothelial cells without CF results in regeneration of binding capacity for the polycationic label. Transport of CF, but not of native ferritin, across the endothelium by vesicle transport, followed by exocytosis of the interiorized CF clusters on the tissue front of the endothelium, was also observed. The possibility that such activities in the blood vessels in vivo may be associated with local changes in the normal distribution of the surface anionic sites as well as in accumulation of debris in the subendothelial layers of the vessels is suggested.     

Differentiated microdomains on the luminal surface of the capillary endothelium. I. Preferential distribution of anionic sites

Cationized ferritin (CF), introduced systemically in vivo or by perfusion in situ, binds preferentially to certain microdomains of the luminal plasmalemma of fenestrated capillaries (mouse pancreas and jejunum). The density and affinity of binding decrease in the following order: fenestral diaphragms greater than coated pits greater than plasmalemma proper. CF binds neither to the membrane of plasmalemmal vesicles and transendothelial channels nor to the corresponding stomatal diaphragms. The distribution pattern is the same when glutaraldehyde fixation precedes the administration of the tracer by perfusion, provided fixation is followed by quenching of residual free aldehyde groups. A much smaller cationic probe (alcian blue) perfused together with the fixative reveals a similar distribution pattern. The functional implications of the association of these microdomains with structures involved in capillary permeability are discussed.     

The endothelial pocket

Summary A new endothelial cell structure, named the endothelial pocket, has been found by combined transmission and scanning electron microscopic studies of renal peritubular capillaries. Transmission EM observations made on these and other fenestrated capillaries demonstrated that each pocket consists of an attenuated fold of fenestrated endothelium that projects 200 nm into the lumen above the rest of the endothelial surface. Beneath this luminal fold, there is a space and then another layer of fenestrated endothelium which abuts the basal lamina. The linear density of endothelial pockets was measured in the capillaries of the kidney cortex, intestinal mucosa and exocrine pancreas in mice and determined to be 0.067, 0.017 and 0.007 pockets·m^-1 respectively. Cationic ferritin decoration of the anionic sites on the luminal surface of the endothelium in these capillary beds revealed that both unlabelled and labelled diaphragms are clustered. In such specimens, the majority of the luminal diaphragms on endothelial pockets did not have cationic ferritin binding sites detectable by either scanning or transmission EM. On this account as well as on account of their general morphology, endothelial pockets appear to be multifold versions of the simple transendothelial channel.     

Intracellular pathways of endocytosed transferrin and non-specific tracers in epithelial cells lining the rete testis of the rat

Summary The uptake and pathway of different markers and ligands for fluid-phase, adsorptive and receptor mediated endocytosis were analyzed in the epithelial cells lining the rete testis after their infusion into the lumen of these anastomotic channels. At 2 min after injection, diferric transferrin bound to colloidal gold was seen attached to the apical plasma membrane and to the membrane of endocytic coated and uncoated pits and vesicles. The injection of transferrin-gold in the presence of a 100-fold excess of unconjugated diferric transferrin revealed no binding or internalization of transferrin-gold. Similarly, apotransferrin-gold was neither bound to the apical plasma membrane nor internalized by these cells. These results thus indicate the presence of specific binding sites for diferric transferrin. At 5 min, internalized diferric transferrin-gold reached endosomes. At 15 and 30 min, the endosomes were still labeled but at these time intervals the transferrin-gold also appeared in tubular elements connected to or associated with these bodies or seen in close proximity to the apical plasma membrane. At 60 and 90 min, most of the transferrin-gold was no longer present in these organelles and was seen only exceptionally in secondary lysosomes. These results thus suggest that the tubular elements may be involved in the recycling of transferrin back to the lumen of the rete testis. The coinjection of transferrin-gold and the fluid-phase marker native ferritin revealed that both proteins were often internalized in the same endocytic pit and vesicle and shared the same endosome. However, unlike transferrin, native ferritin at the late time intervals appeared in dense multivesicular bodies and secondary lysosomes. When the adsorptive marker cationic ferritin and the fluid-phase marker albumin-gold were coinjected, again both proteins often shared the same endocytic pit and vesicle, endosome, pale and dense multivesicular body and secondary lysosomes. However, several endocytic vesicles labeled only with cationic ferritin appeared to bypass the endosomal and lysosomal compartments and to reach the lateral intercellular space and areas of the basement membrane. The rete epithelial cells, therefore, appear to be internalizing proteins and ligands by receptor-mediated and non-specific endocytosis which, after having shared the same endocytic vesicle and endosome, appear to be capable of being segregated and routed to different destinations.     

Molecular characteristics of pial microvessels of the rat optic nerve. Can pial microvessels be used as a model for the blood-brain barrier?

Pial microvessels have commonly been used in studies of the blood-brain barrier because of their relative accessibility. To determine the validity of using the pial microvessel as a model system for the blood-brain barrier, we have extended the comparison of pial and cerebral microvessels at the molecular level by a partial characterization of the glycocalyx of pial endothelial cells, in view of the functional importance of anionic sites within the glycocalyx. Rat optic nerves were fixed by vascular perfusion. Anionic sites on the endothelium were labelled with cationic colloidal gold by means of post- and pre-embedding techniques. The effects of digestion of ultrathin sections on subsequent gold labelling was quantified following their treatment with a battery of enzymes. Biotinylated lectins, viz. wheat germ agglutinin and concanavalin A with streptavidin gold, were employed to identify specific saccharide residues. The results demonstrate that the luminal glycocalyx of pial microvessels is rich in sialic-acid-containing glycoproteins. Neuraminidase, which is specific for N-acetylneuraminic (sialic) acid, and papain (a protease with a wide specificity) significantly reduce cationic colloidal gold binding to the luminal endothelial cell plasma membrane. Wheat germ agglutinin (with an affinity for sialic acid) binds more to the luminal than abluminal plasma membrane, whereas concanavalin A, which binds mannose, binds more to the abluminal surface. Similar results have been obtained for cerebral cortical endothelial cells. With respect to these molecular characteristics, therefore, the pial and cortical microvessels appear to be the same. However, since the two vessel types differ in other respects, caution is urged regarding the use of pial microvessels to investigate the blood-brain barrier. Received: 22 July 1996 / Accepted: 11 October 1996     

Distribution of sialoglycoconjugates on the luminal surface of the endothelial cell in the fenestrated capillaries of the pancreas

Sialic acid-bearing molecules on the luminal surface of the vascular endothelium in mouse and rat pancreatic capillaries were detected electron microscopically by using a procedure with ferritin hydrazide (FH), after preferential oxidation of sialyl residues with sodium periodate. The distribution of FH on the endothelial surface demonstrated the existence of microdomains with various densities of sialoglycoconjugates oxidizable by sodium periodate and accessible to the tracer. On the plasmalemma proper, FH binding sites were heterogeneously distributed. Their concentration on various microdomains decreased as follows: plasmalemma proper greater than coated pits greater than stomal diaphragms of plasmalemmal vesicles and transendothelial channels, and fenestral diaphragms. The membrane of plasmalemmal vesicles and transendothelial channels was not labeled by FH. Nonspecific binding of FH to the nonoxidized endothelial surface or that oxidized after neuraminidase treatment was relatively low.     

Differentiated microdomains on the luminal surface of the capillary endothelium. II. Partial characterization of their anionic sites

To investigate the chemical nature of the cationic ferritin (CF)- binding sites of the differentiated microdomains of the capillary endothelium, the vasculature of the mouse pancreas and intestinal mucosa was perfused in situ with neuraminidase, hyaluronidase, chondroitinase ABC, heparinase, and three proteases: trypsin, papain, and pronase. Proteases of broad specificity removed all anionic sites, suggesting that the latter are contributed by acid glycoproteins or proteoglycans. Neuraminidase, hyaluronidase, and chondroitinase ABC reduced the density of CF-binding sites on the plasmalemma proper, but had no effect on either coated pits or fenestral diaphragms. Heparinase removed CF-binding sites from fenestral diaphragms and had no effect on coated pits. Taken together, these results indicate that the anionic sites of the fenestral diaphragms are contributed primarily by heparan sulfate and/or heparin, whereas those of the plasmalemma proper are of mixed chemical nature. The membranes and diaphragms of plasmalemmal vesicles and transendothelial channels do not bind CF in control specimens; this condition is not affected by the enzymic treatments mentioned above.     

The cell surface of a restrictive fenestrated endothelium

The choriocapillaris is one example of a capillary bed lined by a fenestrated endothelium that is restrictive to exogenous tracers and endogenous plasma proteins. In this study we have examined the distribution of cell-surface monosaccharides utilizing biotinylated lectin-avidin ferritin cytochemistry. Receptors for wheat germ agglutinin were localized to the plasmalemma and diaphragms of some fenestrae, vesicles, and channels at the luminal endothelial front in amounts greater than seen for the other lectins employed. The absence of labeling following inhibition with N-acetylglucosamine and after tissue digestion with N-acetylhexosaminidase, but not after neuraminidase indicated that this lectin marked N-acetylglucosamine residues and not sialic acid. Wheat germ agglutinin receptors were not affected by pronase E or trypsin digestion, but were partially removed by proteinase K. The latter also removed many fenestral diaphragms. Wheat germ agglutinin receptors were cleaved with endoglycosidase D. The combined results indicate that the wheat germ agglutinin receptor is of the low-mannose type and part of a protein with hydrophobic properties. Receptors for concanavalin A (mannose) and Ricinus communis agglutinin (galactose) were also localized to the plasmalemma and endothelial diaphragms. The examination of sections at different tilt angles revealed that these lectins bound to the endothelium in a non-random distribution, encircling diaphragms of fenestrae and channels. Soybean agglutinin (N-acetylgalactosamine) marked endothelial structures sparsely. Following digestion with pronase E or trypsin, receptor sugars for the latter three lectins were completely removed, indicating their presence on protease susceptible glycoproteins. These findings demonstrate that the endothelium of the choriocapillaris bears carbohydrate moieties that are different than those described for permeable fenestrated endothelia.     

Blood-brain barrier disruption in the hypothalamus of young adult spontaneously hypertensive rats

Vascular permeability and endothelial glycocalyx were examined in young adult spontaneously hypertensive rats (SHR), stroke-prone SHR (SHRSP), and Wistar Kyoto rats (WKY) as a control, in order to determine earlier changes in the blood-brain barrier (BBB) in the hypothalamus in chronic hypertension. These rats were injected with horseradish peroxidase (HRP) as an indicator of vascular permeability. Brain slices were developed with a chromogen and further examined with cationized ferritin, a marker for evaluating glycocalyx. Staining for HRP was seen around vessels in the hypothalamus of SHR and SHRSP, but was scarce in WKY. The reaction product of HRP appeared in the abluminal pits of endothelial cells and within the basal lamina of arterioles, showing increased vascular permeability in the hypothalamus of SHR and SHRSP, whereas there were no leaky vessels in the frontal cortex of SHR and SHRSP, or in both areas of WKY. The number of cationized ferritin particles binding to the capillary endothelial cells was decreased in the hypothalamus of SHR and SHRSP, while the number decreased in the frontal cortex of SHRSP, compared with those in WKY. Cationized ferritin binding was preserved in some leaky arterioles, while it was scarce or disappeared in other leaky vessels. These findings suggest that BBB disruption occurs in the hypothalamus of 3-month-old SHR and SHRSP, and that endothelial glycocalyx is markedly damaged there without a close relationship to the early changes in the BBB.     

Anionic sites in the glomerular basement membrane. In vivo and in vitro localization to the laminae rarae by cationic probes

Cationized ferritin (CF) of narrow pI range (7.3-7.5) and the basic dye ruthenium red (RR) have been used as cationic probes to partially characterize anionic sites previously demonstrated in the glomerular basement membrane (GBM). When CF was given i.v. to normal rats and the left kidney was fixed by perfusion 15 min thereafter, clusters of CF molecules were found throughout the lamina rara interna (LRI), lamina rara externa (LRE), and mesangial matrix distributed at regular (approximately 60 nm) intervals. When kidneys were perfused with aldehyde fixative containing RR, small (20 nm) RR-stained particles were seen in the same locations distributed with the same 60 nm repeating pattern, forming a quasiregular, lattice-like arrangement. Fine (approximately 3 nm) filaments connected the sites and extended between them and the membranes of adjoining endothelial and epithelial cells. When CF was given i.v. followed by perfusion with RR in situ, both probes localized to the same sites. CF remained firmly bound after prolonged perfusion with 0.1-0.2 M KCl or NaCl. It was displaced by perfusion with buffers of high ionic strength (0.4-0.5 M KCl) or pH (less than 3.0 or greater than 10.0). CF also bound (clustered at approximately 60 nm intervals) to isolated GBM's, and binding was lost when such isolated GBM's were treated with buffers of high ionic strength or pH. These experiments demonstrate the existence of a quasi-regular, lattice-like network of anionic sites in the LRI and LRE and the mesangial matrix. The sites are demonstrable in vivo (by CF binding), in fixed kidneys (by RR staining), and in isolated GBM's (by CF binding). The results obtained with CF show that the binding of CF (and probably also RR) to the laminae rarae is electrostatic in nature since it is displaced by treatment with buffers of high ionic strength or pH. With RR the sites resemble in morphology and staining properties the proteoglycan particles found in connective tissue matrices and in association with basement membranes in several other locations.     

Radiation-induced changes of negative charge on the cell surface of primary human fibroblasts

By binding cationized ferritin (CF) to the plasma membrane of primary human fibroblasts, the amount and topology of negatively charged sites on cell surfaces were studied after X-irradiation. The CF binding was tested both on fixed and unfixed cells. Using various enzymes, the chemical nature of sites carrying the negative charges on cell surfaces was investigated. The results suggest that in unirradiated fibroblasts the CF binding occurred in a polarized manner, i.e. the particles were localized mainly on the apical surface of cells and formed clusters. The thin cytoplasmic protrusions and cell-to-cell contact sites bound CF to a greater extent than the bleb-like formations. Enzymatic digestion of surface polysaccharides showed that the main carriers of negatively charged sites are the glycosaminoglycans associated with the cell surface. The fixation of cells with glutaraldehyde did not influence the topology of CF binding either before or after the enzymatic treatment. After X-irradiation with 2.5 Gy the topology of CF binding did not change but the CF coverage of cells as well as the amount of ferritin particles per unit of surface area decreased within 10 min. The changes proved to be reversible as the values reached the pre-irradiation level by 1 h after irradiation.     

Anionic surface properties of aortic and mitral valve endothelium from New Zealand white rabbits

The luminal surfaces of the endothelium lining the two surfaces of the aortic arterial (AAR) and ventricular (AVT), and mitral ventricular (MVT) and atrial (MAT), valve cusps were studied with cationic ferritin (CF) and ferritin (Fer)-conjugated lectins (WGA, RCA, SBA). The arterial (AAR) and ventricular (MVT) surfaces of the aortic and mitral cusps, which are exposed to more turbulent fluid mechanical forces and lower wall shear stresses, had the greatest density of CF labeling. The endothelia of the four surfaces displayed a gradient of decreasing density from the nuclear region to the periphery. Neuraminidase, chondroitinase ABC and AC, heparinase, heparitinase, hyaluronidase (testicular), and pronase E digestions suggested that a significant number of the anionic sites labeled by CF are associated with sialoglycoproteins and glycosaminoglycans such as chondroitin 4/6 sulfates, dermatan sulfates, and heparan sulfates. The localization of WGA receptors on the endothelium of AAR and MVT demonstrated a greater density of sialyl moieties than on the AVT and MAT. There was no binding of Fer-RCA with specificity for D-galactopyranosides or Fer-SBA with affinity for N-acetylglucosamine and D-galactose to the endothelium unless it was first treated with neuraminidase. Hence, sialic acids are shown to be among the more superficial components of this glycocalyx and to be largely responsible for the greater densities over the endothelium of AAR and MVT.     

Evidence for the sorting of endocytic vesicle contents during the receptor-mediated transport of IgG across the newborn rat intestine

Fc receptors on the luminal membranes of intestinal epithelial cells in the neonatal rat mediate the vesicular transfer of functionally intact IgG from the intestinal lumen to the circulation. In addition, there is a low level of nonselective protein uptake, but in this case transfer does not occur. To determine whether a specialized class of endocytic vesicles could account for the selective transfer of IgG, mixtures of IgG conjugated to ferritin (IgG-Ft) and unconjugated horseradish peroxidase (HRP) were injected together into the proximal intestine of 10-d-old rats, and the cellular distribution of these two different tracers was determined by electron microscopy. Virtually all apical endocytic vesicles contained both tracers, indicating simultaneous uptake of both proteins within the same vesicle. However, only IgG-Ft bound to the apical plasma membrane, appeared within coated vesicles at the lateral cell surface, and was released from cells. HRP did not bind to the luminal membrane and was not transferred across cells but was confined to apical lysosomes as identified by acid phosphatase and aryl sulfatase activities. To test the possibility that the binding of IgG to its receptor stimulated endocytosis, HRP was used as a fluid volume tracer, and the amount of HRP taken up by cells in the presence and absence of IgG was measured morphologically and biochemically. The results demonstrate that endocytosis in these cells is constitutive and occurs at the same level in the absence of IgG. The evidence presented indicates that the principal selective mechanism for IgG transfer is the binding of IgG to its receptor during endocytosis. Continued binding to vesicle membranes appears to be required for successful transfer because unbound proteins are removed from the transport pathway before exocytosis. These results favor the proposal that IgG is transferred across cells as an IgG-receptor complex.     

Ultrastructure of cell surface coat (glycocalyx) in rat urinary bladder epithelium

Earlier statements to the contrary, the present study demonstrates the presence of a cell surface coat (glycocalyx) on the luminal plasma membrane of the superficial transitional epithelial cells lining the urinary bladder of male Buffalo rats. This coat was demonstrated with ruthenium red, an electron dense stain, which revealed a surface layer, 60-80 A thick, separated from the outer leaflet of the plasma membrane by an electron lucent layer, approximately 30 A thick. The structure of the glycocalyx was not affected by 12 weeks of treatment with dibutylnitrosamine, a known bladder carcinogen.     

The glycocalyx of the mouse uterine luminal epithelium during estrus, early pregnancy, the peri-implantation period, and delayed implantation. I. Acquisition of Ricinus communis I binding sites during pregnancy

Mouse uteri were examined during estrus, early pregnancy, the peri-implantation period, and delayed implantation to determine whether changes in the surface coat of the luminal epithelium could be associated with receptivity of the uterus to the presence of blastocyst-stage embryos or blastocyst adhesion. By using alkaline bismuth subnitrate to label periodate-oxidized glycols within the glycocalyx we were able to measure the thickness and examine the morphology of the glycocalyx by electron microscopy. Ferritin-conjugated Ricinus communis agglutinin (RCA-I) demonstrated the presence of D-galactose at terminal, nonreducing positions within the glycocalyx. A relatively thick (0.06-0.1-micron) surface coat was present during estrus, but contained almost no RCA-I binding sites. During Day 3 of pregnancy the surface coat remained up to 0.1 micron thick and RCA-I binding sites were present. At Day 4 and during delay the glycocalyx had a fibrillar appearance, contained RCA-I binding sites, and was reduced to 0.06-0.08 micron in thickness. During Day 5 of pregnancy the thickness of the surface coat was greatly reduced, but there remained uniform lectin binding adjacent to the plasma membrane both at sites of blastocyst attachment and between implantation sites. The results indicate that the luminal epithelium of the mouse uterus acquired RCA-I binding sites during pregnancy and that the thickness of the surface coat was greatly reduced at the time of implantation.     

Nonrandom distribution of sialic acid over the cell surface of bristle- coated endocytic vesicles of the sinusoidal endothelium cells

Previous studies with protein tracers have shown that the luminal surface of the vascular endothelium of the bone marrow is endocytic. The endocytosis occurs through the formation of large bristle-coated vesicles (LCV). The anionic charge distribution in this process was examined at the luminal surface of the endothelial cell, At pH 1.8, colloidal iron (CI), native ferritin, and polycationic ferritin (PCF) are bound by the luminal surface of the endothelial cell, but not at the sites of LCV formation. PCF used over a pH range of 1.8--7.2 (CI is unstable at higher pH levels) revealed LCV binding of this agent in increasing manner from pH 3.5 upwards. PCF binding at low pH (1.8) at the endothelial cell surface was markedly reduced by neuraminidase. Neuraminidase did not reduce PCF binding by the endothelial cell surface nor by the LCV at higher pH levels. It is concluded that the luminal surface of the endothelial cell has exposed sialic acid groups which are absent or significantly diminished at endocytic sites. The free surface of the endothelial cells as well as the sites of endocytosis have, in addition, anionic material with a pKa higher than that of sialic acid (pKa 2.6). These anionic materials may be different at the sites of endocytosis as compared to those present at the free cell surface.     

The charge distribution in the rough endoplasmic reticulum/golgi complex transitional area investigated by microinjection of charged tracers

The distribution of microinjected ferritin, ranging in charge from anionic to highly cationic, has been used to indicate differences in surface charge on the rough endoplasmic reticulum and the Golgi complex of intact cells. Highly cationic ferritins (HCF)(HCF1, pI 7.9-9.1; HCF2, pI 8.5-9.4; and HCF3.pI 9.5-10.1) were mostly bound and caused swelling of the rough endoplasmic reticulum. Cationic ferritin (CF) (pI 7.0-8.0) and anionic ferritin (AF) (pI 4.0-4.4) caused no changes in morphology. The distribution of these ferritins in the cytoplasmic space varied with their charge. Significantly more CF was bound to surfaces than was found in the free cytoplasmic space. Conversely, there was significantly more AF in the free cytoplasmic space than close to surfaces. Therefore, the intracellular surfaces are negatively charged. Comparison of the structures in the secretory pathway showed no differences in ferritin binding to transition vesicles, rough endoplasmic reticulum, Golgi saccules or secretory vesicles. The Golgi complex beads are not distinguished by their charge. It is therefore unlikely that charge differences play a role in regulating membrane-membrane interactions in this region of the secretory pathway.     

Differentiated microdomains of the luminal plasmalemma of murine muscle capillaries: segmental variations in young and old animals

We investigated the luminal surface of the continuous endothelium of the microvasculature of the murine heart and diaphragm to find out whether it has differentiated microdomains. The probes were ferritin molecules, cationized to pI's 6.8, 7.15, 7.6, 8.0 and 8.4, which were introduced by retrograde or anterograde perfusion through the aorta or vena cava after the blood was removed from the vasculature. The pattern of labeling was analyzed by electron microscopy and assessed quantitatively by morphometry in arterioles, capillaries, and venules identified in bipolar microvascular fields in the diaphragm. The results showed that the plasmalemma proper was heavily but discontinuously labeled by all cationized ferritins (CF) used, the labeling being less extensive on the venular endothelium. CF had access as individual molecules to a fraction of the vesicular population opened on the luminal front of the endothelium. Plasmalemmal vesicle labeling increased from approximately 10 to approximately 25% as the pI decreased from 8.4 to 6.8. Vesicle labeling also increased with CF concentration in the perfusate. All CF binding sites were removed by pronase and papain. Heparinase and heparitinase caused only a slight reduction in CF labeling. Neuraminidase decreased the extent and density of labeling, especially on the plasmalemma proper of the venular endothelium; this decrease was particularly pronounced in old animals.