Immunogold quantitation of laminin, type IV collagen, and heparan sulfate proteoglycan in a variety of basement membranes

A series of basement membranes was immunolabeled for laminin, type IV collagen, and heparan sulfate proteoglycan in the hope of comparing the content of these substances. The basement membranes, including thin ones (less than 0.3 micron) from kidney, colon, enamel organ, and vas deferens, and thick ones (greater than 2 micron), i.e., Reichert's membrane, Descemet's membrane, and EHS tumor matrix, were fixed in formaldehyde, embedded in Lowicryl, and treated with specific antisera or antibodies followed by anti-rabbit immunoglobulin bound to gold. The density of gold particles, expressed per micron2, was negligible in controls (less than or equal to 1.1), but averaged 307, 146, and 23, respectively, for laminin, collagen IV, and proteoglycan over the thick basement membranes (except for Descemet's membranes, over which the density was 16, 5, and 34, respectively) and 117, 72, and 64, respectively, over the lamina densa of the thin basement membranes. Lower but significant reactions were observed over the lamina lucida. Interpretation of the gold particle densities was based on (a) the similarity between the ultrastructure of most thick basement membranes and of the lamina densa of most thin basement membranes, and (b) the biochemical content of the three substances under study in the EHS tumor matrix (Eur J Biochem 143:145, 1984). It was proposed that thick basement membranes (except Descemet's) contained more laminin and collagen IV but less heparan sulfate proteoglycan than the lamina densa of thin basement membranes. In the latter, there was a fair variation from tissue to tissue, but a tendency towards a similar molar content of the three substances.     

Structure, composition, and assembly of basement membrane

Basement membranes are thin layers of matrix separating parenchymal cells from connective tissue. Their ultrastructure consists of a three-dimensional network of irregular, fuzzy strands referred to as "cords"; the cord thickness averages 3-4 nm. Immunostaining reveals that the cords are composed of at least five substances: collagen IV, laminin, heparan sulfate proteoglycan, entactin, and fibronectin. Collagen IV has been identified as a filament of variable thickness persisting after the other components have been removed by plasmin digestion or salt extraction. Heparan sulfate proteoglycan appears as sets of two parallel lines, referred to as "double tracks," which run at the surface of the cords. Laminin is detected in the cords as diffuse material within which thin wavy lines may be distinguished. The entactin and fibronectin present within the cords have not been identified as visible structures. The ability of laminin, heparan sulfate proteoglycan, fibronectin, and entactin to bind to collagen IV has been demonstrated by visualization with rotary shadowing and/or biochemical studies. Incubation of three of these substances-collagen IV, laminin (with small entactin contamination), and proteoglycan-at 35 degrees C for 1 hr resulted in a precipitate that was sectioned for electron microscopic examination and processed for gold immunolabeling for each of the three incubated substances. Three structures are present in the precipitate: 1) a lacework, exclusively composed of heparan sulfate proteoglycan in the form of two parallel lines, similar to double tracks; 2) semi-solid, irregular accumulations, composed of the three initial substances distributed on a cord network; and 3) convoluted sheets, which are also composed of the three initial substances distributed on a cord network but which, in addition, have the uniform appearance and thickness of the lamina densa of basement membrane. Hence these sheets are closely similar to the main component of authentic basement membranes.     

Lack of heparan sulfate proteoglycan in a discontinuous and irregular placental basement membrane

A discontinuous basement membrane of variable width that surrounds spongiotrophoblast cells of rat placenta was examined for the presence of type IV collagen, laminin, a heparan sulfate proteoglycan, entactin, and fibronectin using monospecific antibodies or antisera and the indirect peroxidase technique. At the level of the light microscope, the basement membrane was immunostained for type IV collagen, laminin, entactin, and fibronectin. Heparan sulfate proteoglycan immunostaining, however, was virtually absent even after pretreatment of sections with 0.1 N acetic acid, pepsin (0.1 microgram/ml) or 0.13 M sodium borohydride. Examination in the electron microscope confirmed the lack of immunostaining for heparan sulfate proteoglycan, whereas the other substances were mainly localized to the lamina densa part of the basement membrane. The absence of heparan sulfate proteoglycan in this discontinuous and irregular basement membrane even though type IV collagen, laminin, entactin, and fibronectin are present, suggests that heparan sulfate proteoglycan may have a structural role in the formation of basement membrane.     

Light microscopic immunolocalization of type IV collagen, laminin, heparan sulfate proteoglycan, and fibronectin in the basement membranes of a variety of rat organs

Immunohistochemical methods were used to determine whether type IV collagen, laminin, fibronectin, and heparan sulfate proteoglycan were present in diverse basement membranes. Antisera or antibodies against each substance were prepared, tested by enzyme-linked immunosorbent assay, and exposed to frozen sections of duodenum, trachea, kidney, spinal cord, cerebrum, and incisor tooth from rats aged 20 days to 34 months. Bound antibodies were then localized by indirect or direct peroxidase methods for examination in the light microscope. Immunostaining for type IV collagen, laminin, fibronectin, and heparan sulfate proteoglycan was observed in all of the basement membranes encountered. Fibronectin was also found in connective tissue. In general, the intensity of immunostaining was strong for type IV collagen and laminin, moderate for heparan sulfate proteoglycan, and weak for fibronectin. The pattern was similar in the age groups under study. Very recently the sulfated glycoprotein, entactin, was also detected in the basement membranes of the listed tissues in 20-day-old rats. It is accordingly proposed that, at least in the organs examined, type IV collagen, laminin, fibronectin, heparan sulfate proteoglycan, and entactin are present together in basement membranes.     

Immunocytochemistry of extracellular matrix in the lamina propria of the rat testis: electron microscopic localization

The distribution of laminin, type IV collagen, heparan sulfate proteoglycan, and fibronectin was investigated in the rat testicular lamina propria by electron microscopic immunocytochemistry. Distinct patterns were observed for each antigen within the extracellular matrix (ECM) layers of the lamina propria. Laminin, type IV collagen, and heparan sulfate proteoglycan all localized to the seminiferous tubule basement membrane. Type IV collagen and heparan sulfate proteoglycan, but not laminin, localized to the seminiferous tubule side of the peritubular myoid cells. All four of the antigens were localized between the peritubular and lymphatic endothelial cells. Failure to localize fibronectin in the ECM layer between the Sertoli and peritubular myoid cells tends to support the concept that adult Sertoli cells do not produce this protein in vivo. Intracellular immunostaining was insufficient to allow unambiguous identification of the cellular source of any of the ECM molecules.     

Fine structure of the glomerular basement membrane and immunolocalization of five basement membrane components to the lamina densa (basal lamina) and its extensions in both glomeruli and tubules of the rat kidney

Electron microscopic immunostaining was used to examine the localization of type IV collagen, laminin, entactin , heparan sulfate proteoglycan, and fibronectin within the basement membranes of the rat kidney. In preliminary experiments, various methods of processing formaldehyde-fixed kidney were compared using antilaminin antiserum and the indirect immunoperoxidase method. Little or no laminin immunostaining of the glomerular basement membrane was present in sections unless they had been frozen-thawed; and even in this case, the immunostaining was light in comparison to that of basement membranes in adjacent tubules. However, when frozen-thawed sections were treated with 0.5% sodium borohydride, immunostaining was then as strong in glomerular as in tubular basement membranes. Accordingly, this treatment was applied to frozen-thawed sections before immunostaining for any of the substances under study. Immunostaining of the glomerular basement membrane for each of the five substances was fairly uniform throughout the lamina densa (also called basal lamina), but uneven in the lamina lucida interna and externa (also called lamina rara interna and externa) in which stained bands extended from the lamina densa. Similarly in the basement membranes of tubules, immunostaining for the five substances was localized to the lamina densa and bands extending into the lamina lucida. When the ultrastructure of the glomerular basement membrane was examined, three structures were found: (1) a network of 4-nm-thick "cords," which seems to be the main component; the cords are closely packed in the lamina densa and more loosely arranged in the lamina lucida interna and externa; (2) straight, hollow 7-10-nm-thick structures referred to as " basotubules "; and (3) 3.5-nm elements composed of minute paired rods, referred to as "double pegs." The distribution of the cords, but not that of the other two structures, was related to the immunostaining pattern. It is concluded that (1) to fully reveal the antigenicity of the glomerular basement membrane, frozen-thawed sections must be treated with sodium borohydride prior to immunostaining, possibly because this basement membrane is more compact than the others; and (2) in both glomerular and tubular basement membranes, type IV collagen, laminin, entactin , heparan sulfate proteoglycan and fibronectin are colocalized in the lamina densa and its extensions to the laminae lucidae . Since the distribution of the cords corresponds to that of immunostaining, it is likely that the five substances are present within the cords.     

Interactions of basement membrane components

The binding of laminin, type IV collagen, and heparan sulfate proteoglycan to each other was assessed. Laminin binds preferentially to native type IV (basement membrane) collagen over other collagens. A fragment of laminin (Mr 600 000) containing the three short chains (Mr 200 000) but lacking the long chain (Mr 400 000) showed the same affinity for type IV collagen as the intact protein. The heparan sulfate proteoglycan binds well to laminin and to type IV collagen. These studies show that laminin, type IV collagen and heparan sulfate proteoglycan interact with each other. Such interactions in situ may determine the structure of basement membranes.     

Localization of binding sites for laminin, heparan sulfate proteoglycan and fibronectin on basement membrane (type IV) collagen

Rotary shadowing electron microscopy was used to examine complexes formed by incubating combinations of the basement membrane components: type IV collagen, laminin, large heparan sulfate proteoglycan and fibronectin. Complexes were analyzed by length measurement from the globular (COOH) domain of type IV collagen, and by examination of the four arms of laminin and the two arms of fibronectin. Type IV collagen was found to contain binding sites for laminin, heparan sulfate proteoglycan and fibronectin. With laminin the most frequent site was centered approximately 81 nm from the carboxy end of type IV collagen. Less frequent sites appeared to be present at approximately 216 nm and approximately 291 nm, although this was not apparent when the sites were expressed as a fraction of the length of type IV collagen to which they were bound. For heparan sulfate proteoglycan the most frequent site occurred at approximately 206 nm with a less frequent site at approximately 82 nm. For fibronectin, a single site was present at approximately 205 nm. Laminin bound to type IV collagen through its short arms, particularly through the end of the lateral short arms and to heparan sulfate proteoglycan mainly through the end of its long arm. Fibronectin bound to type IV collagen through the free end region of its arms. Using a computer graphics program, the primary laminin binding sites of two adjacent type IV collagen molecules were found to align in the "polygonal" model of type IV collagen, whereas with the "open network" model, a wide meshed matrix is predicted. It is proposed that basement membrane may consist of a lattice of type IV collagen coated with laminin, heparan sulfate proteoglycan and fibronectin.     

Basement membrane complexes with biological activity

We have studied the reconstitution of basement membrane molecules from extracts prepared from the basement membrane of the EHS tumor. Under physiological conditions and in the presence of added type IV collagen and heparan sulfate proteoglycan, gellike structures form whose ultrastructure appears as interconnected thin sheets resembling the lamina dense zone of basement membrane. The major components of the reconstituted structures include laminin, type IV collagen, heparan sulfate proteoglycan, entactin, and nidogen. These components polymerize in constant proportions on reconstitution, suggesting that they interact in defined proportions. Molecular sieve studies on the soluble extract demonstrate that laminin, entactin, and nidogen are associated in large but dissociable complexes which may be a necessary intermediate in the deposition of basement membrane. The reconstituted matrix was biologically active and stimulated the growth and differentiation of certain cells.     

Three-dimensional network of cords: the main component of basement membranes

Basement membranes were divided into two types: 1) thin basement membranes, such as those of the epidermis, trachea, jejunum, seminiferous tubule, and vas deferens of the rat, the ciliary process of the mouse, and the seminiferous tubule of the monkey, and 2) thick basement membranes, such as the lens capsule of the mouse and Reichert's membrane of the rat. High-magnification electron microscopy was used to examine both types after fixation either in glutaraldehyde followed by postosmication or in potassium permanganate. The basic structure of thin and thick basement membranes was found to be a three-dimensional network of irregular, fuzzy strands referred to as "cords"; the diameter of these cords was variable, but averaged 4 nm in all cases examined. The spaces separating the cords differed, however. In the lamina densa of thin basement membranes, the diameter of these spaces averaged about 14 nm in every case, whereas in the lamina lucida it ranged up to more than 40 nm. Intermediate values were recorded in thick basement membranes. Finally, the third, inconstant layer of thin basement membranes, pars fibroreticularis, was composed of discontinuous elements bound to the lamina densa: i.e., anchoring fibrils, microfibrils, or collagen fibrils. In particular, collagen fibrils were often surrounded by processes continuous with the lamina densa and likewise composed of a typical cord network. Finally, two features were encountered in every basement membrane: 1) a few cords were in continuity with a 1.4- to 3.2-nm thick filament or showed such a filament within them; the filaments became numerous after treatment of the seminiferous tubule basement membrane with the proteolytic enzyme, plasmin, since cords decreased in thickness and could be reduced to a filament, and 2) at the cord surface, it was occasionally possible to see 4.5-nm-wide sets of two parallel lines, referred to as "double tracks." On the basis of evidence that the filaments are type IV collagen molecules and the double tracks are polymerized heparan sulfate proteoglycan, it is proposed that cords are composed of an axial filament of type IV collagen to which are associated glycoprotein components (laminin, entactin, fibronectin) and the double tracks of the proteoglycan.     

Interactions of basement membrane components

The binding of laminin, type IV collagen, and heparan sulfate proteoglycan to each other was assessed. Laminin binds preferentially to native type IV (basement membrane) collagen over other collagens. A fragment of laminin (M_r 600 000) containing the three short chains (M_r 200 000) but lacking the long chain M_r 400 000) showed the same affinity for type IV collagen as the intact protein. The heparan sulfate proteoglycan binds well to laminin and to type IV collagen. These studies show that laminin, type IV collagen and heparan sulfate proteoglycan interact with each other. Such interactions in situ may determine the structure of basement membranes.     

Formation of a supramolecular complex is involved in the reconstitution of basement membrane components

Basement membrane macromolecules, including type IV collagen, laminin, and heparan sulfate proteoglycan, do not aggregate when incubated alone. Rather, precipitation occurs in the presence of equimolar amounts of laminin and type IV collagen but variable amounts of heparan sulfate proteoglycan. This interaction requires native laminin and type IV collagen. Heparan sulfate proteoglycan increases the precipitation of laminin particularly in the presence of type IV collagen. Fibronectin does not cause type IV collagen to precipitate. These studies show that the components of basement membrane interact in a highly specific manner and suggest that such interactions may be involved in the deposition of basement membrane in situ.     

Localization of type IV collagen, laminin, heparan sulfate proteoglycan, and fibronectin to the basal lamina of basement membranes

Electron microscopic immunostaining of rat duodenum and incisor tooth was used to examine the location of four known components of the basement-membrane region: type IV collagen, laminin, heparan sulfate proteoglycan, and fibronectin. Antibodies or antisera against these substances were localized by direct or indirect peroxidase methods on 60-microns thick slices of formaldehyde-fixed tissues. In the basement- membrane region of the duodenal epithelium, enamel-organ epithelium, and blood-vessel endothelium, immunostaining for all four components was observed in the basal lamina (also called lamina densa). The bulk of the lamina lucida (rara) was unstained, but it was traversed by narrow projections of the basal lamina that were immunostained for all four components. In the subbasement-membrane fibrous elements or reticular lamina, immunostaining was confined to occasional "bridges" extending from the epithelial basal-lamina to that of adjacent capillaries. The joint presence of type IV collagen, laminin, heparan sulfate proteoglycan, and fibronectin in the basal lamina indicates that these substances do not occur in separate layers but are integrated into a common structure.     

Hypoplastic basement membrane of the lens anlage in the inheritable lens aplastic mouse (lap mouse)

Adult homozygous lap mice show various eye abnormalities such as aphakia, retinal disorganization, and dysplasia of the cornea and anterior chamber. In the fetal eye of a homozygous lap mouse, the lens placode appears to develop normally. However, the lens vesicle develops abnormally to form a mass of cells without a cavity, and the mass vanishes soon afterward. Apoptotic cell death is associated with the disappearance of the lens anlage. We examined the basement membranes of the lens anlage of this mutant by immunohistochemical methods under light microscopy using antibodies against basement membrane components of the lens anlage, type IV collagen, fibronectin, laminin, heparan sulfate proteoglycan, and entactin and by transmission electron microscopy. Immunohistochemistry showed the distribution and intensity of antibody binding to the lens anlage to be almost the same for each these antibodies regardless of the stage of gestation or whether the anlagen were from normal BALB/c or lap mice. Thus, positive continuous reactions were observed around the exterior region of the lens anlage from day 10 of gestation for type IV collagen, fibronectin, laminin, heparan sulfate proteoglycan antibodies, and at least from day 11of gestation for entactin antibody. The basement membrane lamina densa of both normal and lap mice was shown by electron microscopy to be discontinuous at days 10 and 10.5 of gestation. However, by day 11 the lamina densa was continuous in the lens anlagen of normal mice but still discontinuous in the lap mice. By day 12 of gestation, the lamina densa had thickened markedly in normal mice, whereas in lap mice it remained discontinuous and its thinness indicated hypoplasia. These results indicate that, while all basement components examined are produced and deposited in the normal region of the lens anlage in the lap mouse, the basement membrane is, for some reason, imperfectly formed. The time at which hypoplasia of the basement membrane was observed in this mutant coincided with the stage during which apoptosis in the lens anlage occurred. This result may indicate a possibility of the relationship between the basement membrane and apoptosis in this mutant.     

Basement membrane macromolecules: insights from atomic force microscopy

The major macromolecules of basement membranes-collagen IV, laminin-1, and heparan sulfate proteoglycan (HSPG)-have been analyzed by atomic force microscopy (AFM), both individually and in combination with each other. The positions of laminin binding to collagen IV were mapped and compared with the positions of imperfections in the amino acid sequence of collagen IV; the apparent molecular volumes of the HSPG proteoglycans were measured and used to estimate the corresponding molecular weights. Even the thin, thread-like strands of the polyanion heparan sulfate can be visualized with AFM without staining, coating, or fixation. These strands are single polysaccharide chains and are thus thinner than single-stranded DNA. The heparan sulfate strands in HSPG are necessary for protein filtration in kidney basement membranes. We propose that these thin strands filter proteins by functioning as an entropic brush-i.e., that they filter proteins by their constant thermally driven motion in the basement membrane. These AFM analyses in air are a step toward AFM analyses under fluid of basement membrane macromolecules interacting with each other.     

Macromolecular organization of basement membranes. Characterization and comparison of glomerular basement membrane and lens capsule components by immunochemical and lectin affinity procedures

The macromolecular components of bovine glomerular basement membrane (GBM) and lens capsules (anterior and posterior) solubilized by sequential extractions with denaturing agents were quantitated and characterized by polyacrylamide gel electrophoresis, CL-6B filtration, and DEAE-cellulose chromatography with the help of immunochemical techniques. Laminin, entactin, fibronectin, and heparan sulfate proteoglycan were primarily recovered (over 80%) from both basement membranes in a guanidine HCl extract which contained only a limited amount of the total protein (10-14%); most of the remainder of these noncollagenous components could be solubilized by the guanidine in the presence of reducing agent. Although a portion of the Type IV collagen could be obtained by these treatments, effective extraction of this protein depended on exposure to sodium dodecyl sulfate under reducing conditions. Immunoblot analysis revealed a remarkably similar pattern for GBM and lens capsule Type IV collagens with prominent bands of Mr = 390,000, 210,000, and 190,000 being evident. Fibronectin was present in much greater amounts in GBM than lens capsule while the reverse was true for entactin. In both GBM and lens capsules, the entactin (Mr = 150,000) exceeded laminin; the latter protein on immunoblotting was found to contain primarily the alpha-subunit (Mr = 200,000). The size of the heparan sulfate proteoglycan from anterior (Mr = 400,000) and posterior lens capsule (Mr greater than 500,000) was substantially larger than that from GBM (Mr = 200,000). During DEAE-cellulose chromatography under nonreducing conditions in a denaturing solvent, a portion of the Type IV collagen coeluted with the proteoglycan from these membranes. Considerable Bandeiraea simplicifolia I binding activity (alpha-D-galactose specific) was observed in GBM and lens capsule extracts and column fractions which could not be accounted for by laminin alone. Several components which reacted with this lectin were seen on transblots and among these Type IV collagen was identified. In contrast to the basement membranes from bovine tissues, the constituents from human GBM did not react with the B. simplicifolia I lectin.     

Heterogeneity of distribution pattern at the electron microscopic level of heparan sulfate in various basement membranes.

Using the high-iron diamine thiocarbohydrazide silver proteinate (HID-TCH-SP) staining technique and enzymatic digestion, we investigated the ultrastructural distribution pattern of heparan sulfate side chains of heparan sulfate proteoglycan (HSPG) in various basement membranes (nerve, capillary, oral epithelial, muscle, and dental basement membranes). Four different distribution patterns of stain deposits were identified as heparan sulfate on the basis of enzymatic degradation by heparitinase. In some basement membranes associated with tooth germs and oral epithelium, HID-TCH-SP stain deposits were regularly located at both sides of the lamina densa, but few were observed in the lamina densa itself. In nerve, muscle, and capillary basement membranes, the stain deposits were localized at the external side of the lamina densa adjacent to the underlying connective tissue, but were not found in the laminae lucida and densa. In the internal basal lamina of junctional epithelium of gingiva, the stain deposits were detected mainly in the lamina lucida region. Finally, in some dental and oral epithelial basement membranes, the stain deposits were randomly distributed throughout both laminae lucida and densa. Thus, the present study demonstrated distinct differences in heparan sulfate distribution pattern among various basement membranes, suggesting their architectural heterogeneity.     

Heterogenous distribution of type IV collagen, entactin, heparan sulfate proteoglycan, and laminin among renal basement membranes as demonstrated by quantitative immunocytochemistry

Type IV collagen, entactin, heparan sulfate proteoglycan, and laminin antigenic sites were revealed on various rat renal basement membranes by use of protein A-gold immunocytochemistry. The basement membranes of the proximal and distal convoluted tubules, those of Bowman's capsule and glomerulus, and the mesangial matrix were labeled for all the antigens but to differing extents. Control experiments confirmed the specificity of these labelings. Quantitative evaluation revealed an important heterogeneity for each antigen among the various basement membranes. This heterogeneity suggests that the basement membrane components must arrange themselves in different ways, possibly to account for differences in functional properties of the various renal structures.     

High resolution immunoelectron microscopic localization of functional domains of laminin, nidogen, and heparan sulfate proteoglycan in epithelial basement membrane of mouse cornea reveals different topological orientations

Thin and ultrathin cryosections of mouse cornea were labeled with affinity-purified antibodies directed against either laminin, its central segments (domain 1), the end of its long arm (domain 3), the end of one of its short arms (domain 4), nidogen, or low density heparan sulfate proteoglycan. All basement membrane proteins are detected by indirect immunofluorescence exclusively in the epithelial basement membrane, in Descemet's membrane, and in small amorphous plaques located in the stroma. Immunoelectron microscopy using the protein A-gold technique demonstrated laminin domain 1 and nidogen in a narrow segment of the lamina densa at the junction to the lamina lucida within the epithelial basement membrane. Domain 3 shows three preferred locations at both the cellular and stromal boundaries of the epithelial basement membrane and in its center. Domain 4 is located predominantly in the lamina lucida and the adjacent half of the lamina densa. The low density heparan sulfate proteoglycan is found all across the basement membrane showing a similar uniform distribution as with antibodies against the whole laminin molecule. In Descemet's membrane an even distribution was found with all these antibodies. It is concluded that within the epithelial basement membrane the center of the laminin molecule is located near the lamina densa/lamina lucida junction and that its long arm favors three major orientations. One is close to the cell surface indicating binding to a cell receptor, while the other two are directed to internal matrix structures. The apparent codistribution of laminin domain 1 and nidogen agrees with biochemical evidence that nidogen binds to this domain.     

Basement membrane-like matrix of teratocarcinoma-derived endodermal cells: presence of laminin and heparan sulfate in the matrix at points of attachment to cells

Teratocarcinoma-derived endodermal PYS-2 cells are known to synthesize an extracellular matrix containing the basement membrane molecules laminin, type IV collagen, and heparan sulfate proteoglycan as major constituents (I. Leivo, K. Alitalo, L. Risteli, A. Vaheri, R. Timpl, J. Wartiovaara, Exp Cell Res 137:15-23, 1982). Immunoferritin techniques with specific antibodies were used in the present study to define the ultrastructural localization of the above constituents in the fibrillar network. Laminin was detected in matrix network adjacent to the basal cell membrane and in protruding matrix fibrils that connect the matrix to the cell membrane. Ruthenium red-stainable heparinase-sensitive 10- to 20-nm particles were often present at the junction of the attachment fibrils and the matrix network, or along the attachment fibrils. A corresponding distribution of ferritin label was observed for basement membrane heparan sulfate proteoglycan. Type IV collagen was found in the matrix network but not in the attachment fibrils. The results suggest that the PYS-2 cells are connected to their pericellular matrix by fibrils containing laminin associated with heparan sulfate-containing particles. These results may also have relevance for the attachment of epithelial cells to basement membranes.