- Email: [email protected]
Immunocytochemical localization of laminin, type IV collagen and fibronectin in rat retinal vessels
Eqn
Eye Res. (1988)
47, 317-327
Immunocytochemical Localization of Laminin, Type IV Collagen and Fibronectin in Rat Retinal Vessels EDWARD
ESSNER*
AND
~‘EN-LANG
LIN
Wayne State University, School of Medicine. Detroit, MI 48201, U.S.A.
Kresge Eye Institute,
(Received 23 January
1988 and accepted in revised form
541 E. Can$eld,
15 March
1988)
Laminin, type IV collagen and fibronectin have been identified as major components of the basement membrane (basal lamina) in various tissues. These antigens have also been identified in retinal vessels by light microscopic immunofluorescence but their precise location could not be determined at this level of resolution. In this study, we examined the localization of these constituents at the ultrastructural level using the protein A-immunoperoxidase technique. The basal lamina of all retinal capillaries, arterioles and venules was immunostained after exposure to antisera against laminin, type IV collagen and fibronectin. Staining was localized to the lamina densa, which appeared aa a single or double layer. Immunostaining for fibroneotin showed the weakest activity. The reaction was also seen in discrete patches between endothelial cells and pericytes. The inner limiting membrane of the retina was also reactive for laminin and type IV collagen but not for fibronectin. The results indicate that laminin, type IV collagen and fibronectin are components of the basal lamina in all types of retinal vessels. The presence of fibronectin at the endothelial-pericyte interface suggests that this protein may promote adhesion between cells and thus help to maintain the integrity of the vessel wall. Key umds: immunocytochemistry ; laminin ; type IV collagen : fibronectin; retinal vessels ; basal lamina.
1. Introduction
The basement membrane (basal lamina) is an extracellular structure that provides support to the tissue and promotes adhesion between cells and the extracellular matrix. The basal lamina of blood vesselsalso serves as a non-thrombogenic surface (Duance and Bailey, 1983) and as a filtration barrier (Farquhar, 1981). Several major constituents of the basement membrane have been identified biochemically or by immunocytochemistry. They include type IV collagen (Timpl, Martin, Bruchner, Wick and Wiedemann, 1978), which is the main structural component, laminin. a large glycoprotein that can bind to collagen (Charonis, Tsilibary, Yurchenco and Furthmayr, 1985) and heparan sulfate (BM-1) proteoglycan (Hassel et al., 1980), which may serve as a permeability barrier. Fibronectin, an adhesive glycoprotein, appears to be present in some basement membranes and is also found in the extracellular matrix (Hynes and Yamada, 1982). A basal lamina surrounds the endothelial cells and pericytes of retinal capillaries and the endothelial and smooth muscle cells of the larger vessels.The capillary basal lamina is of particular interest because it undergoes significant morphological and biochemical changes in diseases such as diabetes mellitus (Johnson, 1985). Identification of the main components of the basal lamina of retinal vesselsis still incomplete. However, analysis of basement membranesisolated from preparations of retinal microvesselsindicates the presenceof types I, IV and V collagen (Kennedy, Frank, Mancini and Lande, 1986). Laminin, collagen (types I, IV and V) and fibronectin have been localized by immunofluorescence microscopy in vessels of * To whom 00144835/88/080317 12
correspondence
should
+ 11 $03.00/0
be addressed 0 1988 Academic
Press Limited ERR 47
318
E. ESSSEK.
AND
WT.-L. LIN
human (Newsome and Hewitt, 1985; Jerdan and Glaser, 1986), bovine (Kohno, Sorgente, Patterson and Ryan, 1983) and rat (Gordon and Essner, 1986) retinas. However, the precise distribution of these antigens cannot be determined at this level of resolution. In the present work, we used the protein A-immunoperoxidase procedure to localize these proteins at the electron microscope level in retinal vessels of the rat.
2. Materials
and Methods
Preparation of tissue Male Sprague-Dawley rats (300-450 g) (Taconic Farms, Germantown, NY) were anesthetized with ether. The eyes were enucleated, slit at the limbus and placed in 4% paraformaldehyde-01 M phosphate buffer, pH 7.4. After several minutes, the cornea, lens and vitreous were removed and the posterior eyecup was fixed for a total of 15-2 h at 4’C. The localization of fibronectin was also examined after perfusion fixation. Animals were perfused retrogradely via the abdominal aorta with warm physiological saline using a Buchler monostaltic pump (Haake Buchler Instruments Inc., Saddlebrook, NJ) at a flow rate of 5 ml min-‘, followed by perfusion with warm 4 % paraformaldehyde (pH 7.4) until the eyes appeared white. The posterior eyecups were then removed and placed in fixative for 1 h. In preliminary experiments, retinal tissue was also fixed by immersion in one of the following fixatives at 4°C: (a) 4% paraformaldehyde-01 M phosphate buffer, pH 7.4, 605 M L-lysine, 601 M sodium periodate, 7.5% sucrose (2 h), (b) 4% paraformaldehyde-05 % glutaraldehyde-O.1 M phosphate buffer, pH 7.4 (2 h), (c) 1% glutaraldehyd&l M phosphate buffer, pH 7.4 (1 h) and (d) 1.25 % glutaraldehyde, 1.0 % paraformaldehydd 1 M phosphate buffer, pH 7.3 (1 h). Immunochemicals The anti-laminin antiserum was a purified IgG fraction (Collaborative Research, Lexington, MA). Anti-fibronectin antiserum was obtained from Cappel Laboratories, Inc. (Cochranville, PA) and from Transformation Research (Framingham, MA). No significant differences were observed between the two products. Anti-type IV collagen antiserum was the generous gift of Dr Hynda Kleinman, National Institute of Dental Research, Bethesda, MD. Non-immune rabbit serum and 3,3’-diaminobenzidine (DAB) were obtained from Sigma Chemical Co. (St Louis, MO). Protein A and Protein A-peroxidase were from E-Y Laboratories (San Mateo, CA). Immunocytochemical procedures Fixed tissues from all experiments were rinsed several times and stored overnight in 6 1 M phosphate buffer, pH 7.4, with 7.5% sucrose at 4°C. Non-frozen sections, 40-50 pm thick, were prepared from small blocks of tissue with a TC-2 tissue aectioner (Sorvall, Newton, CT) (Smith and Farquhar, 1965) and were stored in phosphate-buffered saline (PBS). The immunocytochemical procedure was essentially that of Novikoff et al. (1983). Briefly, tissue sections were exposed to each of the antisera for 2448 h at 4°C in covered, plastic, multiwell culture dishes. Each well contained 200 ,ul of antiserum. Antiserum dilutions in PBS ranged from 1: 10 to 1: 100. When several sections were included in one well, care was taken to ensure that they were fully immersed and did not overlap. After exposure to antiserum, the sections were rinsed four times (total of 40 min) in PBS and exposed to protein A-peroxidase (609 mg ml-l) for 1 h at 25°C in the dark. After a thorough rinse in PBS, the sections were incubated for peroxidase activity in medium containing 1 mg ml-’ DAB and 602% hydrogen peroxide in 61 M Tris-HCl buffer, pH 7.6, for 15 min in the dark at 25°C. Controls For the control experiments, each antiserum (diluted 1: 10 in PBS), was combined with its respective antigen (final concentration : 500 pg ml-‘) and left overnight at 4’C. The antigens were obtained from Collaborative Research Inc. (Bedford, MA). The resulting precipitate
LAMININ,
(‘OLLAGEN
AND
FIUKOSE(‘TIS
IS
VESSELS
310
was removed by centrifugation. Non-frozen sections of tissue fixed in 4% paraformaldehgde-01 M phosphate buffer, were incubated in the supernatants. followed by exposure to protein A-peroxidase and incubation in DAB medium as described above. In addition, the following controls were performed: (a) exposure to non-immune rabbit serum followed by protein A-peroxidase and incubation in DAB. (b) exposure to protein A-peroxidase and incubation in DAB without prior exposure t.o non-immune serum. (c) exposure to unlabeled protein ,4 and incubation in DAB, and (d) incubation in DAB wit,hout prior exposure to serum or to protein A. Tissues from all the experiments were dehydrated in alcohols and propylene oxide and embedded in Epon. The results of each experiment wert examined by light and elect’ron microscopy.
3. Results In preliminary experiments, the best results in terms of uniformit,y and consistency of immunostaining were obtained from Gssues fixed in 4”/n paraformaldehyde or in the lysine-periodate-paraformaldehyde fixative. The addition of glutaraldehyde. even at low concentrations, resulted in greater variability of staining in the sect,ion and within individual vessels while higher concentrations were inhibitory. Semi-thin sections were examined by light microscopy to determine the intensity and distribution of peroxidase immunostaining. In sections from tissue exposed t,o anti-laminin antiserum, the reaction, which appeared reddish brown in color, was seen in the walls of the retinal vessels and in the inner limiting membrane (Fig. 1). Background staining was negligible, although it was somewhat, more apparent after exposure to higher concentrations of antiserum. Sone of the control preparations showed peroxidase activity in the retinal vessels or elsewhere in the section (Fig. 2). At the electron-microscope level, anti-laminin immunostaining was localized to the basal lamina of capillaries (Fig. 3), arterioles (Fig. 4) and venules (Fig. 5). In general. the outermost basal lamina adjacent to surrounding Miller cells and glial cells was stained more strongly than the inner one between the endothelial cell and pericytes (Figs 3, 5). In some vessels, the inner basal ia.mina was virtually unreact,ive (Fig. 4). The plasma membranes that delimit the basal lamina, especially those of Miiller and glial cells, were also stained (Fig. 5). Within the basal lamina, peroxidase immunostaining was generally confined to the dense, central portion (lamina densa). In some regions, however, the lamina densa appeared in the form of two distinct layers, both of which were immunostained (Fig. 5). As was noted by light microscopy. no peroxidase immunostaining was observed in blood-vessels from the various control preparations. Light microscopic examinat.ion of tissues exposed t,o ant.i-type IV collagen antiserum, showed relatively strong immunostaining in all retinal vessels and in thp inner limiting membrane (Fig. 6). At the ultrastructural level. peroxidase reaction product was localized in t,he basal lamina of capillaries (Fig. 7). arterioles (Fig. 8) and venules (Fig. 9). As in the case of laminin, the outermost portions of the basal lamina were generally stained more strongly. JVit,hin the basal la,mina immunostaining was localized to the lamina densa whether in single or in b-layered form (Fig. 8): The @ma membranes of glia.1 cells delimiting the basal lamina wvre also stained (Fig. 8). Sane of t,he controls showed a rea&on in the \,cAssels by eit,her light or electron microscopy. After exposure to anti-fibronectin antiserum, peroxidase immunostaining was localized in the basal lamina of capillaries (Figs 10, 11) artc>rioles and renules. Xo differences were seen between perfused and non-perfused tissues except that in the I”.:!
FIGS. l-5.
See facing
page for legend
LAMININ,
COLLAGEN
AND
FIBRONECTIN
IN
VESSELS
321
latter, light staining was also sometimes present on the luminal surface. In contrast to the other two antigens, fibronectin immunostaining was much weaker and more variable and some vessels showed little or no staining. The plasma membranes that delimited the basal lamina were stained, as were the invaginations that arise from these membranes (Fig. 10). In addition to the basal lamina, peroxidase reaction product was concentrated in discrete patches along the interface between the endothelial cells and pericytes (Fig. 11). As was the case for the other antigens, the controls showed no staining of the blood-vessels.
4. Discussion In this study, an immunoperoxidase procedure was used to determine the distribution of laminin, type IV collagen and fibronectin in retinal vessels of the rat. Light microscopy of immunoreacted tissues showed staining in the walls of capillaries, arterioles and venules. No reaction was seen in control tissues. These results are in agreement with earlier immunofluorescence studies of retinal vessels in several species (see Introduction). However, in our study we found no fibronectin staining by light microscopy in the inner limiting membrane as has been reported in the bovine eye (Kohno, Sorgente, Patterson and Ryan, 1983). The very low level of reaction product seen at the ultrastructural level could well be due to adsorption artefact. Whether these differences are due to species variations or to other factors is not known. At the electron microscope level, it was evident that the three antigens were localized in the basal lamina of the retinal vessels. The fact that, in some instances, the basal lamina between the endothelial cell and pericyte was less reactive than the outermost basal lamina is probably due to failure of the antibodies to penetrate the blood-vessel wall. However, differences in concentration of antigen between the inner and outer basal lamina cannot be excluded. Staining of the plasma membranes delimiting the basal lamina is interpreted as an artefact since it was also seen in control tissues. Artefactual staining of these membranes has also been reported by Laurie. Leblond and Martin (1982). The basal lamina of most cells consists of two main regions : a relatively wide, dense.
FIG. 1. Light micrograph of semi-thin section of tissue incubated in anti-laminin antiserum. The wall of the retinal blood vessels and the inner limiting membrane (arrow), are immunostained. x 450. FIG. 2. Light followed Toluidine x 450.
micrograph of semi-thin section from control tissue incubated in non-immune serum, by exposure to protein A-peroxidase and incubation in DAB medium. Section was stained with Blue to reveal the outer nuclear layer. None of the blood-vessels are immunoreactive.
FIG. 3. Electron capillary nucleus
micrograph of anti-laminin-treated tissue. Peroxidase reaction product is localized to basal lamina. Note weaker staining of inner portions (arrows). P. Pericyte; L. lumen; N, of outer nuclear layer. x 4600.
FIG. 4. Electron micrograph of portion of retinal arteriole after exposure to anti-laminin antiserum. Peroxidase immunostaining is localized to the basal lamina, which shows faint, double-layered appearance (arrows). Note staining of delimiting outer membrane (M). The basal lamina between smooth muscle cell (S) and endothelium (E) is virtually unreactive. L. Lumen. x 30700.
FIG. 5. Portion of pericytic venule after exposure to anti-laminin antiserum. Immunostaining is localized to both outer and inner basal laminae although the latter is slightly less reactive. The two-track form of the lamina densa is evident (arrows). The delimiting membrane (M) is also stained. L. Lumen ; E, rndothelium; P. pericyte. x 32500.
I“L’
1: p:sss l-:1.::\ Sl) \\‘ I, I.1 s
LAMININ,
COLLAGEN
AND
FIBRONECTIN
IN
VESSELS
323
central region (lamina densa) separated from the plasma membrane by a less dense zone (lamina rara). When the lamina densa is bordered on the exterior by cells rather than extraceIlular matrix, as in the glomerular basement membrane (Farquhar, 1981), a second lamina rara (externa) is also present. The blood-vessels of the retina are closely surrounded by Miiller cells and other glial cells ; their basal lamina. together with that of pericytes, abuts the basal lamina of the endothelial cells. As a result, the lamina rarae externa are often obliterated and the lamina densa of the adjacent cells is fused into a single layer. The original lamina rarae remain as narrow spaces between the margins of the lamina densa and the plasma membranes of opposing cells. It is interesting to note that the lamina densa of the retinal vessels in our study often appeared in the form of two distinct layers. We have noted this type of configuration in previous studies (Gordon and Essner, 1985). It has also been described in the ‘endoepithelial ’ capillaries of the urinary bladder (Wolff, 1977) and in capillaries of the brain (Schmidley and Wissig, 1986), where it has been interpreted as being due to incomplete fusion of the lamina densa. The main finding in the present work is that peroxidase immunostaining for all three antigens is always associated with the lamina densa, whether in single or bi-layered form, and is absent from the lamina rarae. The peroxidase procedure is widely used for the immunocytochemical localization of antigens at the light and electron microscope levels. It is worth noting, however, that under certain conditions, the resolution of the method may be limited by the propensity of oxidized DAB to diffuse from one site to another. Such diffusion is especially prevalent under conditions of poor fixation or excessive accumulation of reaction product (Novikoff, Novikoff, Quintana and Davis, 1972; Courtoy, Picton and Farquhar, 1983). The problem is particularly important when determining the precise location of antigens within structures of relatively narrow dimensions such as the basal lamina. Recently, Abrahamson (1986) showed that in the glomerular basement membrane, the localization of laminin obtained with a post-embedding technique differed from that seen with the immunoperoxidase procedure. In the present work, care was taken to limit immunostaining to low but detectable levels in order to minimize diffusion artefact. This was possible because of the high concentrations of reactive antigens in the basal lamina. The localization of minor antigenic components may require the use of additional methods in which diffusion is not a factor. There is general agreement that laminin and type IV collagen are true components FIG. 6. Light micrograph antiserum. Immunostaining membrane (arrow). Staining x 760.
of semi-thin section from tissue incubated in anti-type IV collagen is localized in the wall of the retinal vessels and in the inner limiting of nuclei (N) in outer nuclear layer is due to Toluidine Blue stain.
Fra. 7. Electron micrograph of capillary after exposure to anti-type IV collagen antiserum. Immunostaining is localized to inner and outer basal lamina and to the inner limiting membrane (ILM). Arrow points to double-layered portion of lamina densa. P, Pericyte. x 5900. FIG. 8. Retinal arteriole from tissue exposed to anti-type IV collagen antiserum. Immunostaining is localized to the inner (top arrow) and outer (bottom arrow) basal lamina. Double-layered lamina densa is shown at bottom arrow. M, Delimiting membrane of basal lamina: 6, smooth muscle cell: E. endothelium ; L. lumen. x 22500. FIG. 9. Electron micrograph of venule from tissue exposed to anti-type IV collagen lmmunostaining is seen throughout the basal lamina and in the inner limiting membrane Periryte. x 1300.
antiserum. (ILM). P.
324
E. ESSNEK
AND
tV.-L.
LIS
Fra. 10. Electron micrograph of tissue exposed to anti-fibronertin antiserum. Capillary shows light, patchy immunostaining in basal lamina and associated vesicles (arrow). Inner limiting membrane (ILM) is virtually unreactive. E. Endothelium : P, pericyte. x 12500. FIG. 11. Capillary between endothelisl
from anti-fibronectin-treated cell (E) and pericyte (P).
tissue, x 26300.
shows
immunostaining
in patches
(arrows)
LAMININ.
COLLAGEN
AND
FIBRONECTIS
IX
VESSELS
325
of the basal lamina (Johnson, 1985). However, there is some controversy as to whether fibronectin is also a constituent (Madri, Roll, Furthmayr and Foidart, 1980). Although fibronectin has been demonstrated in the basal lamina of several tissues by immunocytochemical means (Stenman and Vaheri, 1978; Hynes, 1981; Laurie, Leblond and Martin, 1982), the localization is thought by some investigators to reflect the presence of plasma (cold, insoluble globulin) rather than tissue fibronectin (Martinez-Hernandez, Marsh, Clark, Macarak and Brownell, 1981; Amenta, Clark and Martinez-Hernandez, 1983). The two forms are completely cross-reactive (in the same species) and therefore immunologically indistinguishable from each other when the usual type of antibodies is employed (Ruoslahti and Vaheri, 1975). However, fibronectin immunostaining can be demonstrated in tissue that has been exhaust,ively perfused, a procedure that elutes albumin and presumably other plasma proteins, from the basal lamina (Courtoy, Timpl and Farquhar, 1982; Courtoy and Boyles, 1983). This situation may not apply to retinal vessels since they are characterized by a blood-retinal barrier which prevents transendothelial transport of proteins from blood to tissue front. In addition, studies in our laboratory and elsewhere (Shiose, 1970; Smith and Rudt, 1975) have shown that peroxidase and certain other hemeproteins are not cytochemically demonstrable in the basal lamina following iv injection. Thus, it appears unlikely that the reaction we observed, albeit weak and variable, was due to plasma rather than to tissue fibronectin. In retinal capillaries, fibronectin immunostaining was also observed in the form of discrete patches located between the endot’helial cells and pericytes. Such deposits were not seen in control tissue or in tissue reacted for laminin or type IV collagen and are thus unlikely to represent non-specific deposits of reaction product. The precise location of the patches has not been determined. They could correspond to gap junctions such as those described in cerebral capillaries (Cuevas et al., 1984) or to other regions which we have observed in retinal capillaries, where the plasma membranes of endothelial cells and pericytes are closely opposed and the basal lamina is absent. Alternatively, the reactive patches may be associated with focal regions similar to those described by Courtoy and Boyles (1983) that’ are characterized by concentrations of fibrils and adhesion plaques in the pericytes. The presence of fibronectin in the endothelial-pericyte interface might be expected to increase cohesiveness between cells and thus contribute to maintaining the integrity of the vessel wall.
ACKNOWLEDGMENTS This work was supported by NIH Grant EY 04831 and by Research to Prevent Blindness, Inc. We thank Penny Wang for her technical assistance.
REFERENCES Abrahamson, D. R. (1986). Post-embedding colloidal gold immunolocalization of laminin to the lamina rara interna, lamina densa and lamina rara externa of renal glomerular basement membranes. J. H&o&m. Cytochem. 34, 84-53. Amenta, P. S., Clark, C. C. and Martinez-Hernandez, A. (1983). Deposition of fibronectin and laminin in the basement membrane of the rat parietal yolk sac: immunochemical and biosynthetic studies. J. Cell Biol. 96, 104-11. Charonis. A. S., Tsilibary, E. C.. Yurchenco, P. D. and Furthmayr, H. (1985). Binding of laminin to type IV collagen: a morphological study. J. CeEZBiol. 100. 1848-53.
326
E. ESSNER
AND
W.-L.
LIN
Courtoy, P. J. and Boyles, J. (1983). Fibronectin in the microvasculature : localization in the pericyte-endothelial interstitium. J. Ultrastruct. Res. 83, 258-73. Courtoy, P. J., Picton, D. H. and Farquhar, M. G. (1983). Resolution and limitations of the immunoperoxidase procedure in the localization of extracellular matrix antigens, J. Histochem. Cytochem. 31, 945-51, Courtoy, P. J., Timpl, R. and Farquhar, M. G. (1982). Comparative distribution of laminin, type IV collagen, and fibronectin in the rat glomerulus. J. Histochem. Cytochem. 30, 874-86. Cuevas, P., Gutierrez-Diaz, J. A., Reimers, D., Dujovny, M., Diaz, F. G. and Ausman, J. I. (1984). Pericyte endothelial gap junctions in human cerebral capillaries. Anat. Embryol. 170, 1559. Duance, V. C. and Bailey, A. J. (1983). The nature, structure and function of the vascular basement membrane. In Biochemical Interactions at the Endothelium. (Ed. Cryer, A.). Pp. 41-69. Elsevier: New York. Farquhar, M. G. (1981). The glomerular basement membrane. A selective macromolecular filter. In Cell Biology of the Extracellular Matrix. (Ed. Hay, E. D.). Pp.33578. Plenum : New York. Gordon, S. R. and Essner, E. (1985). Plasma membrane-associated vesicles in retinal capillaries of the rat. Am. J. Anat. 174, 161-72. Gordon, S. R. and Essner, E. (1986). Actin, myosin and laminin localization in retinal vessels of the rat. Cell Tissue Res. 244, 583-9. Hassel, J. R., Robey, P. G., Barrach, H. J., Wilczek, J., Rennard, S. I. and Martin, G. R. (1980). Isolation of a heparan-sulfate-containing proteoglycan from basement membrane. Proc. Nat. Acad. Sci. U.S.A. 77, 4494-9. Hynes, R. 0. (1981). Fibronectin and its relation to cellular structure and behavior. In Cell Biology of the ExtraceElular Matrix. (Ed. Hay, E. D.). Pp. 295-327. Plenum: New York. Hynes, R. 0. and Yamada, K. M. (1982). Fibronectin : multifunctional modular glycoprotein. J. Cell Biol. 95, 369-77. Jerdan, J. A. and Glaser, B. M. (1986). Retinal microvessel extracellular matrix: an immunofluorescent study. Invest. Ophthalmol. Vis. Sci. 27, 194-203. Johnson, L. D. (1985). Basement membrane structure, function and alteration in disease. In Structure and Properties of Cell Membranes, Vol. 1. (Ed. Benga, G.). Pp. 189217. CRC Press: Boca Raton, FL. Kennedy, A., Frank, R. N. and Mancini, M. A. (1986). In vitro production of glycosaminoglycans by retinal microvessel cells and lens epithelium. Znvest. Ophthalmol. 27, 74654. Kennedy, A., Frank, R. N., Mancini, M. A. and Lande, M. (1986). Collagens of the retinal microvascular cells in vitro. Exp. Eye Res. 42, 177-99. Kohno, T., Sorgente. N., Patterson, R. and Ryan, S. J. (1983). Fibronectin and laminin distribution in bovine eye. Jpn. J. Ophthalmol. 27, 496505. Laurie, G. W.. Leblond, C. P. and Martin, G. R. (1982). Localization of type IV collagen, laminin, heparan sulfate proteoglycan, and fibronectin to the basal lamina of basement membranes. J. Cell Biol. 95, 3494. Madri, J. A., Roll, F. J., Furthmayr, H. and Foidart, J. M. (1980). Ultrastructural localization of fibronectin and laminin in the basement membranes of the murine kidney. J. Cell Biol. 86, 682-7. Martinez-Hernandez, A., Marsh, C. A., Clark, C. C., Macarak, E. J. and Brownell, A. G. (1981). Fibronectin : its relationship to basement membranes. II. Ultrastructural studies in rat kidney. Collagen Rel. Res. 1, 40518. Newsome. D. A. and Hewitt, A. T. (1985). Laminin detection in normal and retinitis pigmentosa human retina. Curr. Eye Res. 4, 92532. Novikoff, A. B., Novikoff, P. M., Quintana, N. and Davis, C. (1972). Diffusion artifacts in 3,3’-diaminobenzidine cytochemistry. J. Histochem. Cytochem. 20, 745-9. Novikoff, P. M., LaRusso, N. F.. Novikoff, A. B., Stockert, R. J., Yam, A. and LeSage, G. D. (1983). Immunocytochemical localization of lysosomal B galactosidase in rat liver. J. Cell Biol. 97, 1559-65.
L.-\MINIS.
C:OLL,iGES
ANI)
FI BKOSEC’TIS
IS
VESSELS
:?27
Ruoslahti, E. and Vaheri, E. (1975). Interaction of soluble fibroblast surface antigen wit,h fibronertin and fibrin. Identity with cold-insaluble globulin of human plasma. ,I. E’rp, Med. 141, 497-501. Schmidley. J. W. and Wissig, S. L. (1986). Basement membrane of central nervous system capillaries lacks Ruthenium Red-staining sites. Microvnsc. Res. 32, 300-14. Shiose. Y. (1970). Electron mirroxropic studies on blood-retinal and bloodpaqueous harriers. .Jpn.
J.
Ophthabmol.
14. 7% XT.
Smith. R. E. and Farquhar. M. C:. (1965). Preparation of non-frozen sections for electroIl microscope cytochemistry. K.(‘.S. Sci. I~tr. Sr~s 10. l:S-18. Smith. It. S. and Rudt, L. A. (1975). Ocular. rascaular and epithelial barriers to microperoxidase. In.vest. Ophthalmol. 14, 55f%60. St’enman. S. and Vaheri. A. (1978). Distribution of a major connective tissue prot,ein. fibronectin. in normal human tissues. J. Ex~J. Lll~d. 147. lWS64. Timpl. R.. Martin, G. R.. Bruchner. D., \Vick. U. and Wirdemann. H. (1978). Xature of the vollagenous protein in a tumor basement membrane. Eur. .J. Biochem. 84. 43-52. \Volff. ,J. k. (1977). I:ltrast,ructure of the terminal vascular bed as related to function. In dlicrocirculntion. Vol. 1. (Eds Kalry. (:. and Altura. H. M.). Pp. 95- 130. University Park Press : Baltimore.
Eye Res. (1988)
47, 317-327
Immunocytochemical Localization of Laminin, Type IV Collagen and Fibronectin in Rat Retinal Vessels EDWARD
ESSNER*
AND
~‘EN-LANG
LIN
Wayne State University, School of Medicine. Detroit, MI 48201, U.S.A.
Kresge Eye Institute,
(Received 23 January
1988 and accepted in revised form
541 E. Can$eld,
15 March
1988)
Laminin, type IV collagen and fibronectin have been identified as major components of the basement membrane (basal lamina) in various tissues. These antigens have also been identified in retinal vessels by light microscopic immunofluorescence but their precise location could not be determined at this level of resolution. In this study, we examined the localization of these constituents at the ultrastructural level using the protein A-immunoperoxidase technique. The basal lamina of all retinal capillaries, arterioles and venules was immunostained after exposure to antisera against laminin, type IV collagen and fibronectin. Staining was localized to the lamina densa, which appeared aa a single or double layer. Immunostaining for fibroneotin showed the weakest activity. The reaction was also seen in discrete patches between endothelial cells and pericytes. The inner limiting membrane of the retina was also reactive for laminin and type IV collagen but not for fibronectin. The results indicate that laminin, type IV collagen and fibronectin are components of the basal lamina in all types of retinal vessels. The presence of fibronectin at the endothelial-pericyte interface suggests that this protein may promote adhesion between cells and thus help to maintain the integrity of the vessel wall. Key umds: immunocytochemistry ; laminin ; type IV collagen : fibronectin; retinal vessels ; basal lamina.
1. Introduction
The basement membrane (basal lamina) is an extracellular structure that provides support to the tissue and promotes adhesion between cells and the extracellular matrix. The basal lamina of blood vesselsalso serves as a non-thrombogenic surface (Duance and Bailey, 1983) and as a filtration barrier (Farquhar, 1981). Several major constituents of the basement membrane have been identified biochemically or by immunocytochemistry. They include type IV collagen (Timpl, Martin, Bruchner, Wick and Wiedemann, 1978), which is the main structural component, laminin. a large glycoprotein that can bind to collagen (Charonis, Tsilibary, Yurchenco and Furthmayr, 1985) and heparan sulfate (BM-1) proteoglycan (Hassel et al., 1980), which may serve as a permeability barrier. Fibronectin, an adhesive glycoprotein, appears to be present in some basement membranes and is also found in the extracellular matrix (Hynes and Yamada, 1982). A basal lamina surrounds the endothelial cells and pericytes of retinal capillaries and the endothelial and smooth muscle cells of the larger vessels.The capillary basal lamina is of particular interest because it undergoes significant morphological and biochemical changes in diseases such as diabetes mellitus (Johnson, 1985). Identification of the main components of the basal lamina of retinal vesselsis still incomplete. However, analysis of basement membranesisolated from preparations of retinal microvesselsindicates the presenceof types I, IV and V collagen (Kennedy, Frank, Mancini and Lande, 1986). Laminin, collagen (types I, IV and V) and fibronectin have been localized by immunofluorescence microscopy in vessels of * To whom 00144835/88/080317 12
correspondence
should
+ 11 $03.00/0
be addressed 0 1988 Academic
Press Limited ERR 47
318
E. ESSSEK.
AND
WT.-L. LIN
human (Newsome and Hewitt, 1985; Jerdan and Glaser, 1986), bovine (Kohno, Sorgente, Patterson and Ryan, 1983) and rat (Gordon and Essner, 1986) retinas. However, the precise distribution of these antigens cannot be determined at this level of resolution. In the present work, we used the protein A-immunoperoxidase procedure to localize these proteins at the electron microscope level in retinal vessels of the rat.
2. Materials
and Methods
Preparation of tissue Male Sprague-Dawley rats (300-450 g) (Taconic Farms, Germantown, NY) were anesthetized with ether. The eyes were enucleated, slit at the limbus and placed in 4% paraformaldehyde-01 M phosphate buffer, pH 7.4. After several minutes, the cornea, lens and vitreous were removed and the posterior eyecup was fixed for a total of 15-2 h at 4’C. The localization of fibronectin was also examined after perfusion fixation. Animals were perfused retrogradely via the abdominal aorta with warm physiological saline using a Buchler monostaltic pump (Haake Buchler Instruments Inc., Saddlebrook, NJ) at a flow rate of 5 ml min-‘, followed by perfusion with warm 4 % paraformaldehyde (pH 7.4) until the eyes appeared white. The posterior eyecups were then removed and placed in fixative for 1 h. In preliminary experiments, retinal tissue was also fixed by immersion in one of the following fixatives at 4°C: (a) 4% paraformaldehyde-01 M phosphate buffer, pH 7.4, 605 M L-lysine, 601 M sodium periodate, 7.5% sucrose (2 h), (b) 4% paraformaldehyde-05 % glutaraldehyde-O.1 M phosphate buffer, pH 7.4 (2 h), (c) 1% glutaraldehyd&l M phosphate buffer, pH 7.4 (1 h) and (d) 1.25 % glutaraldehyde, 1.0 % paraformaldehydd 1 M phosphate buffer, pH 7.3 (1 h). Immunochemicals The anti-laminin antiserum was a purified IgG fraction (Collaborative Research, Lexington, MA). Anti-fibronectin antiserum was obtained from Cappel Laboratories, Inc. (Cochranville, PA) and from Transformation Research (Framingham, MA). No significant differences were observed between the two products. Anti-type IV collagen antiserum was the generous gift of Dr Hynda Kleinman, National Institute of Dental Research, Bethesda, MD. Non-immune rabbit serum and 3,3’-diaminobenzidine (DAB) were obtained from Sigma Chemical Co. (St Louis, MO). Protein A and Protein A-peroxidase were from E-Y Laboratories (San Mateo, CA). Immunocytochemical procedures Fixed tissues from all experiments were rinsed several times and stored overnight in 6 1 M phosphate buffer, pH 7.4, with 7.5% sucrose at 4°C. Non-frozen sections, 40-50 pm thick, were prepared from small blocks of tissue with a TC-2 tissue aectioner (Sorvall, Newton, CT) (Smith and Farquhar, 1965) and were stored in phosphate-buffered saline (PBS). The immunocytochemical procedure was essentially that of Novikoff et al. (1983). Briefly, tissue sections were exposed to each of the antisera for 2448 h at 4°C in covered, plastic, multiwell culture dishes. Each well contained 200 ,ul of antiserum. Antiserum dilutions in PBS ranged from 1: 10 to 1: 100. When several sections were included in one well, care was taken to ensure that they were fully immersed and did not overlap. After exposure to antiserum, the sections were rinsed four times (total of 40 min) in PBS and exposed to protein A-peroxidase (609 mg ml-l) for 1 h at 25°C in the dark. After a thorough rinse in PBS, the sections were incubated for peroxidase activity in medium containing 1 mg ml-’ DAB and 602% hydrogen peroxide in 61 M Tris-HCl buffer, pH 7.6, for 15 min in the dark at 25°C. Controls For the control experiments, each antiserum (diluted 1: 10 in PBS), was combined with its respective antigen (final concentration : 500 pg ml-‘) and left overnight at 4’C. The antigens were obtained from Collaborative Research Inc. (Bedford, MA). The resulting precipitate
LAMININ,
(‘OLLAGEN
AND
FIUKOSE(‘TIS
IS
VESSELS
310
was removed by centrifugation. Non-frozen sections of tissue fixed in 4% paraformaldehgde-01 M phosphate buffer, were incubated in the supernatants. followed by exposure to protein A-peroxidase and incubation in DAB medium as described above. In addition, the following controls were performed: (a) exposure to non-immune rabbit serum followed by protein A-peroxidase and incubation in DAB. (b) exposure to protein A-peroxidase and incubation in DAB without prior exposure t.o non-immune serum. (c) exposure to unlabeled protein ,4 and incubation in DAB, and (d) incubation in DAB wit,hout prior exposure to serum or to protein A. Tissues from all the experiments were dehydrated in alcohols and propylene oxide and embedded in Epon. The results of each experiment wert examined by light and elect’ron microscopy.
3. Results In preliminary experiments, the best results in terms of uniformit,y and consistency of immunostaining were obtained from Gssues fixed in 4”/n paraformaldehyde or in the lysine-periodate-paraformaldehyde fixative. The addition of glutaraldehyde. even at low concentrations, resulted in greater variability of staining in the sect,ion and within individual vessels while higher concentrations were inhibitory. Semi-thin sections were examined by light microscopy to determine the intensity and distribution of peroxidase immunostaining. In sections from tissue exposed t,o anti-laminin antiserum, the reaction, which appeared reddish brown in color, was seen in the walls of the retinal vessels and in the inner limiting membrane (Fig. 1). Background staining was negligible, although it was somewhat, more apparent after exposure to higher concentrations of antiserum. Sone of the control preparations showed peroxidase activity in the retinal vessels or elsewhere in the section (Fig. 2). At the electron-microscope level, anti-laminin immunostaining was localized to the basal lamina of capillaries (Fig. 3), arterioles (Fig. 4) and venules (Fig. 5). In general. the outermost basal lamina adjacent to surrounding Miller cells and glial cells was stained more strongly than the inner one between the endothelial cell and pericytes (Figs 3, 5). In some vessels, the inner basal ia.mina was virtually unreact,ive (Fig. 4). The plasma membranes that delimit the basal lamina, especially those of Miiller and glial cells, were also stained (Fig. 5). Within the basal lamina, peroxidase immunostaining was generally confined to the dense, central portion (lamina densa). In some regions, however, the lamina densa appeared in the form of two distinct layers, both of which were immunostained (Fig. 5). As was noted by light microscopy. no peroxidase immunostaining was observed in blood-vessels from the various control preparations. Light microscopic examinat.ion of tissues exposed t,o ant.i-type IV collagen antiserum, showed relatively strong immunostaining in all retinal vessels and in thp inner limiting membrane (Fig. 6). At the ultrastructural level. peroxidase reaction product was localized in t,he basal lamina of capillaries (Fig. 7). arterioles (Fig. 8) and venules (Fig. 9). As in the case of laminin, the outermost portions of the basal lamina were generally stained more strongly. JVit,hin the basal la,mina immunostaining was localized to the lamina densa whether in single or in b-layered form (Fig. 8): The @ma membranes of glia.1 cells delimiting the basal lamina wvre also stained (Fig. 8). Sane of t,he controls showed a rea&on in the \,cAssels by eit,her light or electron microscopy. After exposure to anti-fibronectin antiserum, peroxidase immunostaining was localized in the basal lamina of capillaries (Figs 10, 11) artc>rioles and renules. Xo differences were seen between perfused and non-perfused tissues except that in the I”.:!
FIGS. l-5.
See facing
page for legend
LAMININ,
COLLAGEN
AND
FIBRONECTIN
IN
VESSELS
321
latter, light staining was also sometimes present on the luminal surface. In contrast to the other two antigens, fibronectin immunostaining was much weaker and more variable and some vessels showed little or no staining. The plasma membranes that delimited the basal lamina were stained, as were the invaginations that arise from these membranes (Fig. 10). In addition to the basal lamina, peroxidase reaction product was concentrated in discrete patches along the interface between the endothelial cells and pericytes (Fig. 11). As was the case for the other antigens, the controls showed no staining of the blood-vessels.
4. Discussion In this study, an immunoperoxidase procedure was used to determine the distribution of laminin, type IV collagen and fibronectin in retinal vessels of the rat. Light microscopy of immunoreacted tissues showed staining in the walls of capillaries, arterioles and venules. No reaction was seen in control tissues. These results are in agreement with earlier immunofluorescence studies of retinal vessels in several species (see Introduction). However, in our study we found no fibronectin staining by light microscopy in the inner limiting membrane as has been reported in the bovine eye (Kohno, Sorgente, Patterson and Ryan, 1983). The very low level of reaction product seen at the ultrastructural level could well be due to adsorption artefact. Whether these differences are due to species variations or to other factors is not known. At the electron microscope level, it was evident that the three antigens were localized in the basal lamina of the retinal vessels. The fact that, in some instances, the basal lamina between the endothelial cell and pericyte was less reactive than the outermost basal lamina is probably due to failure of the antibodies to penetrate the blood-vessel wall. However, differences in concentration of antigen between the inner and outer basal lamina cannot be excluded. Staining of the plasma membranes delimiting the basal lamina is interpreted as an artefact since it was also seen in control tissues. Artefactual staining of these membranes has also been reported by Laurie. Leblond and Martin (1982). The basal lamina of most cells consists of two main regions : a relatively wide, dense.
FIG. 1. Light micrograph of semi-thin section of tissue incubated in anti-laminin antiserum. The wall of the retinal blood vessels and the inner limiting membrane (arrow), are immunostained. x 450. FIG. 2. Light followed Toluidine x 450.
micrograph of semi-thin section from control tissue incubated in non-immune serum, by exposure to protein A-peroxidase and incubation in DAB medium. Section was stained with Blue to reveal the outer nuclear layer. None of the blood-vessels are immunoreactive.
FIG. 3. Electron capillary nucleus
micrograph of anti-laminin-treated tissue. Peroxidase reaction product is localized to basal lamina. Note weaker staining of inner portions (arrows). P. Pericyte; L. lumen; N, of outer nuclear layer. x 4600.
FIG. 4. Electron micrograph of portion of retinal arteriole after exposure to anti-laminin antiserum. Peroxidase immunostaining is localized to the basal lamina, which shows faint, double-layered appearance (arrows). Note staining of delimiting outer membrane (M). The basal lamina between smooth muscle cell (S) and endothelium (E) is virtually unreactive. L. Lumen. x 30700.
FIG. 5. Portion of pericytic venule after exposure to anti-laminin antiserum. Immunostaining is localized to both outer and inner basal laminae although the latter is slightly less reactive. The two-track form of the lamina densa is evident (arrows). The delimiting membrane (M) is also stained. L. Lumen ; E, rndothelium; P. pericyte. x 32500.
I“L’
1: p:sss l-:1.::\ Sl) \\‘ I, I.1 s
LAMININ,
COLLAGEN
AND
FIBRONECTIN
IN
VESSELS
323
central region (lamina densa) separated from the plasma membrane by a less dense zone (lamina rara). When the lamina densa is bordered on the exterior by cells rather than extraceIlular matrix, as in the glomerular basement membrane (Farquhar, 1981), a second lamina rara (externa) is also present. The blood-vessels of the retina are closely surrounded by Miiller cells and other glial cells ; their basal lamina. together with that of pericytes, abuts the basal lamina of the endothelial cells. As a result, the lamina rarae externa are often obliterated and the lamina densa of the adjacent cells is fused into a single layer. The original lamina rarae remain as narrow spaces between the margins of the lamina densa and the plasma membranes of opposing cells. It is interesting to note that the lamina densa of the retinal vessels in our study often appeared in the form of two distinct layers. We have noted this type of configuration in previous studies (Gordon and Essner, 1985). It has also been described in the ‘endoepithelial ’ capillaries of the urinary bladder (Wolff, 1977) and in capillaries of the brain (Schmidley and Wissig, 1986), where it has been interpreted as being due to incomplete fusion of the lamina densa. The main finding in the present work is that peroxidase immunostaining for all three antigens is always associated with the lamina densa, whether in single or bi-layered form, and is absent from the lamina rarae. The peroxidase procedure is widely used for the immunocytochemical localization of antigens at the light and electron microscope levels. It is worth noting, however, that under certain conditions, the resolution of the method may be limited by the propensity of oxidized DAB to diffuse from one site to another. Such diffusion is especially prevalent under conditions of poor fixation or excessive accumulation of reaction product (Novikoff, Novikoff, Quintana and Davis, 1972; Courtoy, Picton and Farquhar, 1983). The problem is particularly important when determining the precise location of antigens within structures of relatively narrow dimensions such as the basal lamina. Recently, Abrahamson (1986) showed that in the glomerular basement membrane, the localization of laminin obtained with a post-embedding technique differed from that seen with the immunoperoxidase procedure. In the present work, care was taken to limit immunostaining to low but detectable levels in order to minimize diffusion artefact. This was possible because of the high concentrations of reactive antigens in the basal lamina. The localization of minor antigenic components may require the use of additional methods in which diffusion is not a factor. There is general agreement that laminin and type IV collagen are true components FIG. 6. Light micrograph antiserum. Immunostaining membrane (arrow). Staining x 760.
of semi-thin section from tissue incubated in anti-type IV collagen is localized in the wall of the retinal vessels and in the inner limiting of nuclei (N) in outer nuclear layer is due to Toluidine Blue stain.
Fra. 7. Electron micrograph of capillary after exposure to anti-type IV collagen antiserum. Immunostaining is localized to inner and outer basal lamina and to the inner limiting membrane (ILM). Arrow points to double-layered portion of lamina densa. P, Pericyte. x 5900. FIG. 8. Retinal arteriole from tissue exposed to anti-type IV collagen antiserum. Immunostaining is localized to the inner (top arrow) and outer (bottom arrow) basal lamina. Double-layered lamina densa is shown at bottom arrow. M, Delimiting membrane of basal lamina: 6, smooth muscle cell: E. endothelium ; L. lumen. x 22500. FIG. 9. Electron micrograph of venule from tissue exposed to anti-type IV collagen lmmunostaining is seen throughout the basal lamina and in the inner limiting membrane Periryte. x 1300.
antiserum. (ILM). P.
324
E. ESSNEK
AND
tV.-L.
LIS
Fra. 10. Electron micrograph of tissue exposed to anti-fibronertin antiserum. Capillary shows light, patchy immunostaining in basal lamina and associated vesicles (arrow). Inner limiting membrane (ILM) is virtually unreactive. E. Endothelium : P, pericyte. x 12500. FIG. 11. Capillary between endothelisl
from anti-fibronectin-treated cell (E) and pericyte (P).
tissue, x 26300.
shows
immunostaining
in patches
(arrows)
LAMININ.
COLLAGEN
AND
FIBRONECTIS
IX
VESSELS
325
of the basal lamina (Johnson, 1985). However, there is some controversy as to whether fibronectin is also a constituent (Madri, Roll, Furthmayr and Foidart, 1980). Although fibronectin has been demonstrated in the basal lamina of several tissues by immunocytochemical means (Stenman and Vaheri, 1978; Hynes, 1981; Laurie, Leblond and Martin, 1982), the localization is thought by some investigators to reflect the presence of plasma (cold, insoluble globulin) rather than tissue fibronectin (Martinez-Hernandez, Marsh, Clark, Macarak and Brownell, 1981; Amenta, Clark and Martinez-Hernandez, 1983). The two forms are completely cross-reactive (in the same species) and therefore immunologically indistinguishable from each other when the usual type of antibodies is employed (Ruoslahti and Vaheri, 1975). However, fibronectin immunostaining can be demonstrated in tissue that has been exhaust,ively perfused, a procedure that elutes albumin and presumably other plasma proteins, from the basal lamina (Courtoy, Timpl and Farquhar, 1982; Courtoy and Boyles, 1983). This situation may not apply to retinal vessels since they are characterized by a blood-retinal barrier which prevents transendothelial transport of proteins from blood to tissue front. In addition, studies in our laboratory and elsewhere (Shiose, 1970; Smith and Rudt, 1975) have shown that peroxidase and certain other hemeproteins are not cytochemically demonstrable in the basal lamina following iv injection. Thus, it appears unlikely that the reaction we observed, albeit weak and variable, was due to plasma rather than to tissue fibronectin. In retinal capillaries, fibronectin immunostaining was also observed in the form of discrete patches located between the endot’helial cells and pericytes. Such deposits were not seen in control tissue or in tissue reacted for laminin or type IV collagen and are thus unlikely to represent non-specific deposits of reaction product. The precise location of the patches has not been determined. They could correspond to gap junctions such as those described in cerebral capillaries (Cuevas et al., 1984) or to other regions which we have observed in retinal capillaries, where the plasma membranes of endothelial cells and pericytes are closely opposed and the basal lamina is absent. Alternatively, the reactive patches may be associated with focal regions similar to those described by Courtoy and Boyles (1983) that’ are characterized by concentrations of fibrils and adhesion plaques in the pericytes. The presence of fibronectin in the endothelial-pericyte interface might be expected to increase cohesiveness between cells and thus contribute to maintaining the integrity of the vessel wall.
ACKNOWLEDGMENTS This work was supported by NIH Grant EY 04831 and by Research to Prevent Blindness, Inc. We thank Penny Wang for her technical assistance.
REFERENCES Abrahamson, D. R. (1986). Post-embedding colloidal gold immunolocalization of laminin to the lamina rara interna, lamina densa and lamina rara externa of renal glomerular basement membranes. J. H&o&m. Cytochem. 34, 84-53. Amenta, P. S., Clark, C. C. and Martinez-Hernandez, A. (1983). Deposition of fibronectin and laminin in the basement membrane of the rat parietal yolk sac: immunochemical and biosynthetic studies. J. Cell Biol. 96, 104-11. Charonis. A. S., Tsilibary, E. C.. Yurchenco, P. D. and Furthmayr, H. (1985). Binding of laminin to type IV collagen: a morphological study. J. CeEZBiol. 100. 1848-53.
326
E. ESSNER
AND
W.-L.
LIN
Courtoy, P. J. and Boyles, J. (1983). Fibronectin in the microvasculature : localization in the pericyte-endothelial interstitium. J. Ultrastruct. Res. 83, 258-73. Courtoy, P. J., Picton, D. H. and Farquhar, M. G. (1983). Resolution and limitations of the immunoperoxidase procedure in the localization of extracellular matrix antigens, J. Histochem. Cytochem. 31, 945-51, Courtoy, P. J., Timpl, R. and Farquhar, M. G. (1982). Comparative distribution of laminin, type IV collagen, and fibronectin in the rat glomerulus. J. Histochem. Cytochem. 30, 874-86. Cuevas, P., Gutierrez-Diaz, J. A., Reimers, D., Dujovny, M., Diaz, F. G. and Ausman, J. I. (1984). Pericyte endothelial gap junctions in human cerebral capillaries. Anat. Embryol. 170, 1559. Duance, V. C. and Bailey, A. J. (1983). The nature, structure and function of the vascular basement membrane. In Biochemical Interactions at the Endothelium. (Ed. Cryer, A.). Pp. 41-69. Elsevier: New York. Farquhar, M. G. (1981). The glomerular basement membrane. A selective macromolecular filter. In Cell Biology of the Extracellular Matrix. (Ed. Hay, E. D.). Pp.33578. Plenum : New York. Gordon, S. R. and Essner, E. (1985). Plasma membrane-associated vesicles in retinal capillaries of the rat. Am. J. Anat. 174, 161-72. Gordon, S. R. and Essner, E. (1986). Actin, myosin and laminin localization in retinal vessels of the rat. Cell Tissue Res. 244, 583-9. Hassel, J. R., Robey, P. G., Barrach, H. J., Wilczek, J., Rennard, S. I. and Martin, G. R. (1980). Isolation of a heparan-sulfate-containing proteoglycan from basement membrane. Proc. Nat. Acad. Sci. U.S.A. 77, 4494-9. Hynes, R. 0. (1981). Fibronectin and its relation to cellular structure and behavior. In Cell Biology of the ExtraceElular Matrix. (Ed. Hay, E. D.). Pp. 295-327. Plenum: New York. Hynes, R. 0. and Yamada, K. M. (1982). Fibronectin : multifunctional modular glycoprotein. J. Cell Biol. 95, 369-77. Jerdan, J. A. and Glaser, B. M. (1986). Retinal microvessel extracellular matrix: an immunofluorescent study. Invest. Ophthalmol. Vis. Sci. 27, 194-203. Johnson, L. D. (1985). Basement membrane structure, function and alteration in disease. In Structure and Properties of Cell Membranes, Vol. 1. (Ed. Benga, G.). Pp. 189217. CRC Press: Boca Raton, FL. Kennedy, A., Frank, R. N. and Mancini, M. A. (1986). In vitro production of glycosaminoglycans by retinal microvessel cells and lens epithelium. Znvest. Ophthalmol. 27, 74654. Kennedy, A., Frank, R. N., Mancini, M. A. and Lande, M. (1986). Collagens of the retinal microvascular cells in vitro. Exp. Eye Res. 42, 177-99. Kohno, T., Sorgente. N., Patterson, R. and Ryan, S. J. (1983). Fibronectin and laminin distribution in bovine eye. Jpn. J. Ophthalmol. 27, 496505. Laurie, G. W.. Leblond, C. P. and Martin, G. R. (1982). Localization of type IV collagen, laminin, heparan sulfate proteoglycan, and fibronectin to the basal lamina of basement membranes. J. Cell Biol. 95, 3494. Madri, J. A., Roll, F. J., Furthmayr, H. and Foidart, J. M. (1980). Ultrastructural localization of fibronectin and laminin in the basement membranes of the murine kidney. J. Cell Biol. 86, 682-7. Martinez-Hernandez, A., Marsh, C. A., Clark, C. C., Macarak, E. J. and Brownell, A. G. (1981). Fibronectin : its relationship to basement membranes. II. Ultrastructural studies in rat kidney. Collagen Rel. Res. 1, 40518. Newsome. D. A. and Hewitt, A. T. (1985). Laminin detection in normal and retinitis pigmentosa human retina. Curr. Eye Res. 4, 92532. Novikoff, A. B., Novikoff, P. M., Quintana, N. and Davis, C. (1972). Diffusion artifacts in 3,3’-diaminobenzidine cytochemistry. J. Histochem. Cytochem. 20, 745-9. Novikoff, P. M., LaRusso, N. F.. Novikoff, A. B., Stockert, R. J., Yam, A. and LeSage, G. D. (1983). Immunocytochemical localization of lysosomal B galactosidase in rat liver. J. Cell Biol. 97, 1559-65.
L.-\MINIS.
C:OLL,iGES
ANI)
FI BKOSEC’TIS
IS
VESSELS
:?27
Ruoslahti, E. and Vaheri, E. (1975). Interaction of soluble fibroblast surface antigen wit,h fibronertin and fibrin. Identity with cold-insaluble globulin of human plasma. ,I. E’rp, Med. 141, 497-501. Schmidley. J. W. and Wissig, S. L. (1986). Basement membrane of central nervous system capillaries lacks Ruthenium Red-staining sites. Microvnsc. Res. 32, 300-14. Shiose. Y. (1970). Electron mirroxropic studies on blood-retinal and bloodpaqueous harriers. .Jpn.
J.
Ophthabmol.
14. 7% XT.
Smith. R. E. and Farquhar. M. C:. (1965). Preparation of non-frozen sections for electroIl microscope cytochemistry. K.(‘.S. Sci. I~tr. Sr~s 10. l:S-18. Smith. It. S. and Rudt, L. A. (1975). Ocular. rascaular and epithelial barriers to microperoxidase. In.vest. Ophthalmol. 14, 55f%60. St’enman. S. and Vaheri. A. (1978). Distribution of a major connective tissue prot,ein. fibronectin. in normal human tissues. J. Ex~J. Lll~d. 147. lWS64. Timpl. R.. Martin, G. R.. Bruchner. D., \Vick. U. and Wirdemann. H. (1978). Xature of the vollagenous protein in a tumor basement membrane. Eur. .J. Biochem. 84. 43-52. \Volff. ,J. k. (1977). I:ltrast,ructure of the terminal vascular bed as related to function. In dlicrocirculntion. Vol. 1. (Eds Kalry. (:. and Altura. H. M.). Pp. 95- 130. University Park Press : Baltimore.