THE ANATOMICAL RECORD 220:258-266 (1988) The Anionic Matrix at the Rat Glomerular Endothelial Surface PRATAP S. AVASTHI AND VALSALA KOSHY Research Service, VA. Medical Center and Department of Medicine, University of New Mexico School of Medicine, Albuquerque, New Mexico 87108 ABSTRACT The anionic macromolecules at the glomerular endothelial cell surface are visualized only when stained with cationic stains. We investigated the arrangement and composition of this anionic matrix at the luminal surface. Rat kidneys were perfused with anionic ferritin (PI 4.5), ferritin (PI 7.41, or cationized ferritin (CF, PI 8.3). Anionic ferritin (PI4.5) did not bind to the capillary wall, ferritin (PI7.4) bound discontinuously only to the laminae rarae of the basement membrane, but cationized ferritin (CF, PI 8.3) bound as a thick continuous layer to the cell plasmalemma and bound to the anionic matrix in the fenestral spaces. These observations show that a n anionic matrix lines the entire capillary lumen surface, fills the fenestrae, and is interposed between the blood and the basement membrane a t the fenestrae. The anionic constituents at the capillary luminal surface were identified by in vivo digestion with specific enzymes. Absence of CF binding following digestion with specific enzymes was taken to indicate the presence of the particular glycoprotein known to be susceptible to the enzyme used. Neuraminidase digestion revealed that anionic sites over the surface plasmalemma are mainly from sialoproteins. In contrast, the matrix in fenestral channels contains heparan sulfate, hyaluronic acid, and sialoproteins. Papain digestion showed no glycolipids at the luminal surface. The functions of this continuous anionic layer located a t the luminal surface of glomerular capillaries have not yet been established. For ultrafiltration of water and macromolecules across the glomerular capillary wall, the major pathway appears to be through the endothelial fenestrae, across the three layers of glomerular basement membrane, and finally across the epithelial cell slit diaphragm (Shea and Morrison, 1975). The lamina densa of the glomerular basement membrane has been shown to function as the size-selective barrier for the neutral macromolecules traversing the glomerular capillary wall (Batsford et al., 1987; Caulfield and Farquhar, 1974). The glomerular capillary wall functions both as a charge-selective barrier and as a size-selective barrier (Bennett et al., 1976 Bohrer et al., 1978).Anionic macromolecules have been found to be particularly concentrated in both the laminae rarae of the basement membrane (Caulfield and Farquhar, 1976; Kanwar and Farquhar, 1979; Rennke et al., 1975). In addition, the epithelial cell glycocalyx and possibly the slit diaphragms also have anionic sialoproteins (Kerjaschki et al., 1984; Blau and Michael, 1971). The anionic barrier function of the glomerular capillary wall has been ascribed to all these structures but the precise location of the anionic barrier remains uncertain (Kanwar et al., 1980; Kanwar and Jakubowski, 1984; Vernier et al., 1983;Fishman and Karnovsky, 1985). Fenestrations in the glomerular endothelial cell cytoplasm in adult rats are mainly circular holes that traverse the entire cell cytoplasm and have a mean diameter of 75 nm (Avasthi et al., 1980).These fenestrae 0 1988 ALAN R. LISS, INC do not have a diaphragm and on routine electron microscopy appear to be mostly empty. In other blood vessels, endothelial cells have been shown to have a layer of anionic macromolecules that is attached to their plasmalemma and extends into the vascular lumen. Although invisible when unstained, a 10- to 20-nm-thick fuzzy layer of anionic macromolecules is visualized a t the vascular endothelial surfaces when stained with cationic electron-dense tracers (Luft, 1966; Copley, 1974).A similar layer of anionic macromolecular glycocalyx has been noted at the luminal surface of glomerular endothelial cells (Latta and Johnston, 1976). Owing to the fenestrations in the cytoplasm of this cell, the distribution of glycocalyx on the plasmalemma of this cell is expected to be quite different. The purpose of this study was to map the distribution of glycocalyx on the plasmalemma of the fenestrated glomerular endothelial cell and to ascertain whether the fenestral spaces are occupied by a n anionic fiber matrix that may be invisible by the routine electron microscopy. Furthermore, some of the constituents of this glycocalyx were histochemically identified. Received May 27, 1987; accepted July 23, 1987 GLOMERULAR ENDOTHELIAL GLYCOCALYX MATERIALS AND METHODS Animals Female Sprague-Dawley rats (150-200 gm body weight) were utilized in these experiments. To minimize variability, if any, due to sex and age, only the female rats in a narrow range of body weight were utilized. Materials Native ferritin (horse spleen, isoelectric point, PI 4.55.5) from Sigma Chemical Company, St. Louis, and cationized ferritin (PI 8.2-8.4, CF) from Miles Laboratory, Elkhart, IN, were purchased. Neutral ferritin (PI 7.4) was prepared by the method of Danon et al. (1972). All batches of ferritin were checked for the stated PI by isoelectric focusing (Righetti and Drysdale, 1974), and monomolecular dispersion was checked by electron microscopy. All ferritins were dialyzed against 0.15 M NaCl at 4°C for 24 hr. Dulbecco's phosphate-buffered saline and minimum essential medium were purchased from Gibco Laboratories (Grand Island Biological Company, Grand Island, NY). The final perfusion solution (DPBS) contained Dulbecco's phosphate-buffered saline (pH 7.27.4) with 5% minimum essential medium and 14 mM glucose. Mannitol was obtained from Invenox, Chagrin Falls, OH, and lissamine green from Serva Entwicklunglabor, Heidelberg. Experimental Protocol Rats were anesthetized by intramuscular injection of ketamine HCl(87 m g k g body weight) and xylazine (13 m g k g body weight). Laparatomy was performed; the abdominal aorta was isolated, and a polyethylene catheter (18-to 22-gauge) was positioned so that the catheter tip lay adjacent to the left renal artery. Two ligatures, one above and the other below the left renal artery, were tied on the aorta, so as to perfuse only the left kidney. Open-circuit perfusion was carried out by making a n opening in the left renal vein. With a n occlusive roller pump molter pump, model No. RL 175 110, International Medical Corporation, Englewood, CO), the renal vasculature was cleared of blood by a 2- to 7-min perfusion of 10-15 ml of DPBS at 37"C, gassed with 0 2 . Through a side-arm, perfusion pressures were continuously monitored with a Statham pressure gauge and a calibrated electronic pressure monitor (Electronics for Medicine, White Plains, NY). Following the washing out of blood from the vasculature, ferritin was perfused (7 mg/100 gm body weight in 2 ml of 0.15 M saline) through the renal vasculature. Unbound ferritin was removed by one more minute of DPBS infusion followed by perfusion of fixative. Details of the sequence of perfusions for all the experiments are given in Table 1. Except for the period of fixative perfusion, throughout the entire experiment, perfusion pressures were generally 80-150 mmHg. Tissue Fixation and Processing for Electron Microscopy Tissues were fixed by intrarenal arterial perfusion of 30 ml of Karnovsky's fixative in 10 min. Kidneys were removed, and 1-mm cubes of renal cortex were cut and further fixed in Karnovsky's fixative overnight (12-18 hr), and were stored in 0.1 M sodium cacodylate buffer (pH 7.4) till further processed. The tissues were postfixed in osmium tetroxide, dehydrated in acetone, and embed- 259 ded in LX 112 resin (Ladd, Burlington, VA). Thin sections ( 60 nm) were stained with lead citrate. Viewing and photography were done on a Hitachi HU 11-B EM (75 kV). From each rat, a minimum of three glomeruli were studied. The results were qualitatively assessed by comparing the number of ferritin particles on the experimental rat capillary luminal surface to the appropriate control. To resolve the issue whether CF was completely filling the fenestrae or was attached only to its walls, first, three serial thin sections (gray) of one glomerulus were mounted on a single-slotted formvar-coated grid. Several sets of sequential sections were photographed to assess the pattern of CF attachment in the fenestrae. In addition, by means of a Hitachi 600 electron microscope with a tilt stage, paired photomicrographs of glomerular capillary walls were taken at a + l o " and a -10" tilt and were viewed through a stereoscopic viewer to obtain three-dimensional images. These procedures permitted resolution of the question whether ferritin was filling the entire fenestrae. Next, we considered the possibility that CF may have initially attached to the laminae rara interna, and then owing to the ongoing water filtration, additional CF may have accumulated in the fenestrae passively. To settle this question, we perfused CF in three rats with nonfiltering kidneys. Nonfiltering kidneys were produced by intrasaphenous vein injection of mannitol (0.1 gm per 100 gm rat weight) and by tying the left ureter (Holdaas et al., 1981). Lack of filtration by such maneuvers was ascertained by the injection of 1 ml of lissamine green dye (0.17 gm% in 0.3 M NaC1) in the tail vein of a similarly prepared rat. Dye in renal tubules was visualized through a dissecting scope. The dye was readily seen in the tubules on the unligated side but not on the side with the ureteric ligation, which confirmed the lack of filtration. In three rats with such nonfiltering kidneys, CF was injected and tissues were prepared and examined as outlined earlier. - Enzyme Digestion Experiments Enzymes were obtained from Sigma Chemical Company and Miles Laboratory. The purity, specificity, and enzyme activity of the enzymes used in this study are shown in Table 2. The enzymes used to remove the specific proteoglycans from the capillary lumen surface, their source, and the conditions utilized to optimize in situ enzyme action are shown in Table 3. For each enzyme experiment, enzyme solutions were prepared just before perfusion to preserve maximal activity. Optimal conditions for enzyme action were taken from the references shown in the extreme right-hand column of Table 3 (Pino et al., 1982; De Bruyn et al., 1978; Kanwar and Farquhar, 1980; Underhill and Toole, 1979; Gamse et al., 1978; Buonassisi and Root, 1975; Linker and Hovingh, 1972; Gill et al., 1981; Yamagata et al., 1968). Renal temperatures were maintained as required for the individual experiment (Table 3) by dripping saline (0.15 M) warmed to the required temperature over the kidney surface during the entire experiment. Also, all perfusion solutions were warmed to the required temperatures. Control rats for each enzyme were perfused with a similar volume of enzyme diluent (Table 3). Thus, each control rat underwent a perfusion sequence that was exactly similar to its corresponding enzyme experiment 260 P.S. AVASTHI AND V. KOSHY TABLE 1. Sequence of perfusion steps for native ferritin, for cationized ferritin, and for the enzymes used for in situ digestion of luminal anionic sites No. of rats (explcontrol)' Experiment Native ferritin (NF) Neutral ferritin Cationized ferritin Papain Hyaluronidase Neuramindase Heparitinase Chondroitinase ABC DPBS wash, min Sequence of perfusions Enzyme DPBS Tracer (PI) DPBS Fixative (4-5 ml), min wash, min (7 mg/100 g, bw) wash, min (Karnovsky's), ml 3 4 None No NF (4.5) 4 30 2 4 None No CF (7.4) 4 30 4 4 None No CF (8.3) 4 30 614 212 412 2/1 412 4-7 4 2-5 4-7 2 5-20 10 5-10 10 5-10 1 1 1 1 1 CF (8.3) CF (8.3) CF (8 * 3) CF (8.3) CF (8.3) 2 3 2 2-3 1-2 30 30 30 30 30 'Control rats were perfused, with diluent without the enzyme. All non-enzyme-controlshad identical CF binding; hence the number of controls for individual enzymes is small. TABLE 2. Purity, specificity and specific activity of enzymes Proteolytic activity Enzyme Hyaluronidase Not detected Neuraminidase < .002 U/mg protein Heparitinase Not detected Chondroitinase ABC Not detected Substrate specificity Specific activity Degrades only hyaluronic acid No 0-galactosidase activity < 1%a8ainst other GAGS Degrades hyaluronic acid and chondroitin 2,000 TRUImg protein' 160 U/mg protein 260 U/mg protein 37.5 pWmg protein 'TRU, Turbidity reducing units. 2 ~ glycosaminoglycans. ~ ~ ~ , TABLE 3. Conditions and enzymes used for removing the anionic sites from the luminal surface of glomerular capillaries Enzyme Papain Source Concentration' pH Temp., "C Time, min Reference 5-10 mglml 7.0 37 5-20 1-2U/ml 5.4 37 5-10 Pino et al. (1982) De Bruyn et al. (1978) Kanwar and Farquhar (1980) Underhill and Toole (1979) Buonassisi and Root (1975) Gamse et al. (1978) Gill et al. (1981) Linker and Hovingh (1972) Yamagata et al. (1968) Neuraminidase Papaya laytex (Sigma, Type IV) C1. perfringens (Sigma, Type X) Hyaluronidase Strep. hyaluronyticus (Miles) 50Ulml 5.6 37 10 Heparitinase F1. heparinum (Miles) 50Ulml 7.0 43 10 Chondroitinase Proteus vulgaris 2 Ulml 7.4 37 5-10 'Enzymes were diluted in 0.1 M sodium chloride-acetate to the concentrations specified for each enzyme. Conditions used were taken from the references given. GLOMERULAR ENDOTHELIAL GLYCOCALYX Fig. 1. Monomolecular dispersion of cationized ferritin (CF) in 0.15 M saline, PI 8.3. X225,OOO. Fig. 2. Capillary wall of a rat infused with neutral ferritin (PI 7.4). Ferritin is sparingly bound in clusters within the laminae rarae. CL, capillary lumen; BS, Bowman’s space. x49,OOO. 261 Fig. 3. Attachment of CF (PI 8.3) to the entire luminal front of a normal rat glomerular capillary wall. CF is attached to the endothelial glycocalyx, which fills the majority of the fenestrae. CL, capillary lumen; BS, Bowman’s space. ~58,000. Fig. 4. Composite of three sequential sections from a glomerular capillary loop. Many fenestrae have been sequentially cut but none show a central CF-free area. CL, capillary lumen. X35,OOO. 262 P.S. AVASTHI AND V. KOSHY Fig. 5. Paired photomicrographs of the same section taken at +lo" the entire fenestral spaces, including the center. CL, capillary lumen. and -10" stage tilt. Stereoscopic viewing reveals attachment of CF in X50,OOO. except that the enzymes were omitted from the diluent in controls. It should be noted that ferritins were not allowed to remain stationary or dwell in the renal vasculature, but were perfused through the vasculature.The number of rats, duration of perfusions, and sequence of perfusions for the experimental and control groups are given in Table 1.In enzyme perfusion experiments, the total enzyme quantity to be perfused was administered and each enzyme was retained in the renal vasculature for the indicated time (Table 1) by clamping the renal vessels. Tissue fixation and microscopy were done as outlined earlier. RESULTS Tracer Ferritins were dispersed as single particles for every batch of tracer utilized, as is shown in Figure 1. The PI of each batch of tracer was checked. For native ferritin the PI ranged between 4.4 and 5.5; neutral ferritins had a PI of 7.2-7.5 and cationized ferritin PI > 8, with major bands at 8.2 and 8.4. Binding of the Tracer Rat kidneys perfused with native ferritin (PI 4.5) showed no binding of the tracer to any component of the GLOMERULAR ENDOTHELIAL GLYCOCALYX 263 A clear zone between endothelial plasmalemma and basement membrane indicates the lack of binding sites parallel to the abluminal plasmalemma. Some binding of CF to fenestral floor, though decreased, is still seen (compare with Fig. 10)and is situated luminal to lamina rara interna (arrows). CL, capillary lumen; BS, Bowman’s Fig. 7. Detached endothelial and epithelial cells in a papain-treated space. ~79,000. kidney. No ferritin is bound at the luminal surface. CF is bound to Fig. 9. CF binding following hyaluronidase. CF binding in some of epithelial cell surface and a few ferritin particles are in the basement membrane. CL, capillary lumen; BS, Bowman’s space. x 58,000. the fenestrae is reduced (arrowheads), but only a minimal reduction is seen at plasmalemma glycocalyx. CL, capillary lumen; BS, Bowman’s Fig. 8. CF binding following 10-min neuraminidase treatment. space. ~ 4 0 , 0 0 0 . Marked reduction in CF binding to the endothelial glycocalyx is seen. Fig. 6. Glomerular capillary wall from a nonfiltering kidney. CF is attached to the entire luminal front, the ferritin in the fenestrae is seen up to the capillary luminal (CL) level, and the pattern of CF binding is the same as in Figure 3. ~ 4 3 , 0 0 0 . P.S. AVASTHI AND V. KOSHY 264 capillary wall. Neutral ferritin (PI7.4) was found in both laminae rarae of the glomerular basement membrane in discontinuous clusters (Fig. 2). This is in agreement with results reported by others (Rennke et al., 1975; Kanwar and Farquhar, 1979). Kidneys perfused with CF (PI 8.3) showed a layer of ferritin three to four particles thick attached to the entire luminal surface of the endothelium. This band of ferritin lined the luminal surface of endothelial cells, bound to plasmalemma at the sides of the fenestrae, filled the fenestral channels, and blended with ferritin bound to the lamina rarae interna (Fig. 3). CF filled the majority of the fenestrae up to the level of luminal plasmalemma. Thus, the endothelial glycocalyx forms a continuous layer at the luminal surface, covering the luminal plasmalemma and spanning the fenestrae. The finding of filling of fenestrae by CF could have been due to the CF attached to the rim of the fenestrae in the section even if CF did not bind to the center of the fenestrae. To examine this possibility, we obtained thin sequential sections. If CF was attached only to the walls, a CF-free area would be found in some of the sequentially cut fenestrae. Of the many sequentially cut fenestrae, none showed a central CF-free area (Fig. 4). Furthermore, the three-dimensional images obtained from the steroscopic paired micrographs of glomerular capillary wall sections (Fig. 5) demonstrate CF in the entire fenestral space. If CF were attached only to the fenestral walls, three-dimensional constructs would have revealed CF attached to a curved rim and a central area without CF. Instead, CF was found to be filling the entire fenestral spaces. These observations support our interpretation that CF attaches to the anionic fiber matrix that fills fenestrae completely up to the capillary lumen level. Next, we dealt with the possibility that following initial attachment of CF to the fenestral floor, additional CF may have accumulated passively due to ongoing water filtration. Nonfiltering kidneys were utilized to answer this question with the idea that if passive accumulation of CF in the fenestrae were occurring because of the water filtration, CF would bind only to the fenestral floor and would not be seen in the fenestral channel in the absence of filtration. Figure 6 shows that the pattern of CF binding in nonfiltering glomeruli was identical to that of the normal rat kidney. These observations are in agreement with the hypothesis that the anionic fiber matrix of endothelial glycocalyx in the fenestrae is made visible by the active binding of CF. Enzyme Digestion Experiments In these experiments, we tested the effects of a protease and a series of specific glycoprotein degrading enzymes on the anionic constituents at the luminal surface of the capillary wall. Papain digestion resulted in a complete loss of CF binding from the entire luminal front. Both the endothelial and the epithelial cells were detached from the basement membrane in places. CF was found in Bowman’s space and was bound to the surface of the epithelial cells only in the papain-digested kidneys (Fig. 7). Thus, papain increased the permeability of the barrier, which allowed CF to penetrate across the capillary wall. These results indicate that glycolipids are not a major constituent of the anionic sites at the luminal front of glomerular capillaries, since papain is expected to spare the glycolipids. Neuraminidase digestion erased the binding of CF from the plasmalemma glycocalyx at the luminal surface of endothelial cells (Fig. 8). It should be noted that CF failed to bind along the abluminal plasmalemma (Fig. 8). The extent of loss of anionic sites depended on the time allowed for the enzyme action. The loss was patchy following a 5min digestion; more extensive loss was observed after a long period of digestion. These results suggest that the anionic sites at the plasmalemma glycocalyx consist of sialoglycoproteins and are consistent with the findings reported previously (Latta et al., 1975). However, some residual binding of CF to the fenestral floor is still seen (Fig. 8). There was no detachment of endothelial or epithelial cells from the basement membrane in these neuraminidase-treated glomeruh, perhaps because of the short duration of enzyme action. These findings indicate that a majority of the anionic sites at the plasmalemmal glycocalyx are from neuraminic acid residues. Also, in the fenestral glycocalyx, a majority of anionic sites are removed with the removal of sialic acid. Digestion with hyaluronidase shows only a partial reduction in CF binding in some of the fenestrae (Fig. 9). Digestion with heparitinase decreased the binding of CF preferentially within the fenestrae, whereas the anionic sites in the endothelial glycocalyx at plasmaIemma were only minimally decreased (Fig. 10). These observations suggest that in the fenestrae the anionic sites of endothelial glycocalyx also contain hyaluronic acid and heparan sulfate glycosaminoglycans in addition to the sialoproteins (Figs. 9,lO). A 5-min digestion with chondroitinase ABC did not reduce CF binding to the luminal surface. At 10 min, some of the fenestrae had reduced CF binding, but others showed no change (Fig. 11).As chondroitinase ABC can degrade both chondroitin sulfate and hyaluronic acid (Yamagata et al., 1968), perhaps the relatively low activity of the enzyme against hyaluronic acid may have been responsible for these findings. Therefore, we are unable to decide from these observations whether chondroitin sulfate is also present in endothelial glycocalyx. DISCUSSION Although a glycocalyx has been previously noted at the glomerular endothelial surface, details of its distribution and composition have not been explored (Latta and Johnston, 1976; Rennke et al., 1975). The arrangement of macromolecules at the plasmalemmal surface is detailed here, and these observations show that the glomerular endothelial cell fenestrae are not empty holes as they often appear to be by routine electron microscopy. Instead of a fenestral diaphragm, these fenestrae are occupied by an anionic matrix that is visualized only following the binding of an electron-densetracer. In this respect, the ultrastructure of the matrix in the fenestrae is similar to the glycocalyx at the external surface of cells, which also remains invisible in unstained preparations. It appears that the glycocalyx attached to the endothelial plasmalemma at the fenestral rim extends into and fills the fenestral channels. We excluded the possibility of passive accumulation of ferritin at this location a) by the failure of the anionic or the neutral ferritin to similarly accumulate in the fenestrae and b) by the identical binding of CF in the nonfltering kidneys. Inclusion of a part of the fenestral rim within the GLOMERULAR ENDOTHELIAL GLYCOCALYX 265 Fig. 10. CF binding following heparitinase. Endothelial glycocalyx Fig. 11. CF binding following chondroitinase ABC. Slightly reduced over the plasmalemma shows a minimal reduction in attachment of binding in some fenestrae (arrowheads) is seen but the majority of CF. Majority of fenestrae do not bind CF. CL, capillary lumen; BS, fenestrae remain normal. CL, capillary lumen; BS, Bowman’s space. Bowman’s space. x 30,000. ~30,000. thickness of the tissue section could also have mimicked these findings. This possibility was excluded by examining sequential sections and by stereoscopic evaluation. By both these methods ferritin was seen in the fenestral center, which excluded the possibility that the ferritin was attached only to the rim of the fenestrae. Finally, an uncertainty remained that initially the cationized ferritin may have bound to the surface of the basement membrane and that the resultant clogging of the passage may have allowed passive accumulation of cationized ferritin in the fenestrae. This possibility was excluded by the finding of an identical binding pattern for cationized ferritin in the kidneys without water filtration. Thus, our findings demonstrate that the glomerular endothelial fenestrae are occupied by a fiber matrix of the macromolecules that are attached to the plasmalemma around the fenestrae and extend into the fenestral channels to fill the fenestral spaces. The functional significance of this anionic fiber matrix in the filtration pathway cannot be assessed from these mo~holo@calobservations’ Other observers have proposed that the fiber matrix in the endothelial transcellular channels in general may determine the molecular sieving across the capillary walls (Curry and Michel, 1980).Also, theoretical considerations favor the location of the anionic barrier at the very luminal surface rather than inside the layers Of the (Lassen, 1970). However, conflicting results have been obtained regarding the location of the charge-selective barrier in studies utilizing different tracers and Karnovsky, 1976; Farquhar et al., 1961; Masri et al., 1985). Therefore, despite the demonstration of a series of anionic layers in the filtration pathway, the precise location of myan the barrier in the glomerular capillary wall for serum anionic macromolecules (albumin) remains uncertain. Histochemical characterization of anionic sites in the endothelial glycocalyx reveals the following features: 1) A complete loss of CF binding following digestion with papain indicates that glycolipids are not likely contributors to the anionic sites. 2) The glycocalyx at the luminal surface derives a majority of its anionic groups from sialic acid residues, since digestion with neuraminidase preferentially erases the binding of CF from the glycocalyx covering the luminal and the abluminal plasmalemma. 3) Anionic sites in the glycocalyx within the fenestrae are from heparan sulfate proteoglycans, hyaluronic acid, and sialyl residues. The fenestral glycocalyx has a greater concentration of heparan sulfate proteoglycans. ACKNOWLEDGMENTS of we acknowledge the excellent technical Paul E. Clark and Judy Davis. The help of Drs. John Trotter, R. Chiovetti, and K.D. Gardner is acknowledged with Karin Woods provided excellent secretarial LITERATURE CITED Avasthi, P.S., A.P. Evan, and D. Hay (1980) Glomerular endothelial cells in uranyl-nitrate induced acute renal failure in rats. J. Clin. Invest., 65121-127. Batsford, S.R., R. Rohrbach, and A. Vogt (1987) Size restriction in the glomerular capillary wall. Importance of lamina densa. Kidney Int,,31,.710-717, Bennet, c.M., R.J. Glassock, R.L.S. Chang, W.M. Deen, C.R. Robertson, and B.M. Brenner (1976) Permselectivity of glomerular capil- 266 P.S. AVASTHI AND V. KOSHY lary wall: Studies of experimental glomerulonephritis in the rat using dextran sulfate. J. Clin. Invest., 57:1287-1294. Blau, E., and A.F. Michael (1971)Rat glomerular basement membrane composition and metabolism in aminonucleoside nephrosis. J. Lab. Clin. Med., 77:97-109. Bohrer, M.P., C. Baylis, H.D. Humes, R.J. Glassock, C.R. Robertson, and B.M. Brenner (1978) Permselectivity of glomerular capillary wall: Facilitated filtration of circulating polycations. J. Clin. Invest., 61:72-78. Buonassisi, V., and M. Root (1975) Enzymatic degradation of heparinrelated mucouolvsaccharides from the surface of endothelial cell .. . . ~ ~~ cultures. Biochim. Biophys. Acta, 385:l-10. Caulfield, J.P., and M.G. Farquhar (1974) The permeability of glomerular capillaries to graded dextrans: Identification of the basement membrane as the primary filtration barrier. J. Cell Biol., 63t883903. Caulfield, J.P., and M.G. Farquhar (1976)Distribution of anionic sites in glomerular basement membranes: Their possible role in filtration and attachment. Proc. Natl. Acad. Sci. USA, 73:1646-1650. Copley, A.L. (1974)Hemorheological aspects of the endothelium-plasma interface. Microvasc. Res., 8:192-212. Curry, F.E., and C.C. Michel (1980) A fiber matrix model of capillary permeability. Microvasc. Res., 20:96-99. Danon, D., T. Goldstein, Y. Marikovsky, and E. Skutelesky (1972) Use of cationized ferritin as a label of negative charges on cell surfaces. J. Ultrastruct. Res., 38.500-510. De Bruyn, P.P., S. Michelson, and R.P. Becker (1978) Nonrandom distribution of sialic acid over the cell surface of bristle-coated endocytic vesicles of the sinusoidal endothelial cells. J. Cell Biol., 78:379-389. Farquhar, M.G., S.L. Wissig, and G.E. Palade (1961) Glomerular permeability: I. Ferritin transfer across the normal glomerular capillary wall. J. Exp. Med., 113:47-66. Fishman, J.A., and M.J. Karnovsky (1985) Effect of the aminonucleoside of puromycin on glomerular epithelial cells in vitro. Am. J. Pathol., 118:398-407. Gamse, G., H.G. Fromme, and H. Kresse (1978)Metabolism of sulfated glycosaminoglycans in cultured endothelial cells and smooth muscle cells from bovine aorta. Biochim. Biophys. Acta, 544:514-528. Gill, J.P., J. Adler, C.K. Silbert, and J.E. Silbert (1981) Selective removal of glycosaminoglycans from cultured human skin fibroblast monolayers. Biochem. J., 194:299-307. Holdaas, H., G.F. Di Bona, and F. Kiil (1981) Effect of low-level renal nerve stimulation on renin release from non-filtering kidneys. Am. J. Physiol.,241:F 156- 161. Kanwar, Y.S., and M.G. Farquhar (1979)Anionic sites in the glomerular basement membrane: In vivo and in vitro localization to the lamina rarae by cationic probes. J. Cell Biol., 81:137-153. Kanwar, Y.S., and M.G. Farquhar (1980) Detachment of endothelium ~~ 1 --.I ~~ and epithelium from the glomerular basement membrane produced by kidney perfusion with neuraminidase. Lab. Invest., 42~375-384. Kanwar, Y.S., and M.L. Jakubowski (1984) Unaltered anionic sites of glomerular basement membrane in aminonucleoside nephrosis. Kidney Int., 25:613-618. Kanwar, Y.S., A. Linker, and M.G. Farquhar (1980) Increased permeability of the glomerular basement membrane to ferritin after removal of glycosaminoglycans (heparan sulfate) by enzyme digestion. J. Cell Biol., 86:688-693. Kerjaschki, D., D.J. Sharkey, and M.G. Farquhar (1984)Identification and characterization of podocalyxin-the major sialoprotein of the renal glomerular epithelial cell. J. Cell Biol., 98:1591-1596. Lassen, N.A. (1970)Large molecules. In: Capillary Permeability. C.Z. Crone and N.A. Lassen, eds. Academic, New York, pp. 549-559. Latta, H., and W.H. Johnston (1976) The glycoprotein inner layer of glomerular capillary basement membrane as a filtration barrier. J. Ultrastruct. Res., 57:65-67. Latta, H., W.H. Johnston, and T.M. Stanley (1975) Sialoglycoproteins and filtration barriers in the glomerular capillary wall. J. Ultrastruct. Res., 51:354-376. Linker, A,, and P. Hovingh (1972) Heparinase and heparitinase from flavobacteria. Methods Enzymol, 28:902-911. Luft, J.H. (1966) Fine structures of capillary and endocapillary layer as revealed by ruthenium red. Fed. Proc.,25:1773-1783. Masri, M.A., N.D. Boyd, and H. Mandin (1985) Role of fibrinogen in the glomerular permeability of the perfused isolated dog kidney. Kidney Int., 27t807-810. Pino, R.M., E. Essner, and L.C. Pino (1982) Location and chemical composition of anionic sites in Bruch's membrane of the rat. J. Histochem. Cytochem., 30:245-252. Rennke, H.G., R.S. Cotran, and M.A. Venkatachalam (1975) Role of molecular charge in glomerular permeability. Tracer studies with cationized ferritins. J. Cell Biol., 67:638-646. Righetti, P.G., and J.W. Drysdale (1974) Isoelectric focusing in gels. J. Chromatogr., 98:271-321. Ryan, G.B., and M.J. Karnovsky (1976) Distribution of endogenous albumin in the rat glomerulus; role of hernodynamic factors in glomerular barrier function. Kidney Int., 9:36-45. Shea, S.M., and A.B. Morrison (1975) A stereological study of the glomerular filter in the rat. Morphometry of the slit diaphragms and basement membrane. J. Cell Biol., 67:436-443. Underhill, C.B., and B.P. Toole (1979) Binding of hyaluronate to the surface of cultured cell. J. Cell Biol., 82:475-484. Vernier, R.L., D.J. Klein, S.P. Sisson, J.D. Mahan, T.R. Oegema, and D.M. Brown (1983)Heparan sulfate-rich anionic sites in the human glomerular basement membrane. Decreased concentration in congenital nephrotic syndrome. New Engl. J. Med., 309:lOOI-1009. Yamagata, T., H. Saito, 0. Habuchi, and S. Suzuki (1968) Purification and properties of bacterial chondroitinases and chondrosulfates. J. Biol. Chem.. 243:1523-1535.