Heparin type IV collagen interactions: equilibrium binding and inhibition of type IV collagen self-assembly.

Interactions between type IV collagen and heparin were examined under equilibrium conditions with rotary shadowing, solid-phase binding assays, and affinity chromatography. With the technique of rotary shadowing and electron microscopy, heparin appeared as thin, short strands and bound to the following three sites: the NC1 domain, and in the helix, at 100 and 300 nm from the NC1 domain. By solid-phase binding assays the binding of [3H]heparin in solution to type IV collagen immobilized on a solid surface was found to be specific, since it was saturable and could be displaced by an excess of unlabeled heparin. Scatchard analysis indicated three classes of binding sites for heparin-type IV collagen interactions with dissociation constants of 3, 30, and 100 nM, respectively. Furthermore, by the solid-phase binding assays, the binding of tritiated heparin could be competed almost to the same extent by unlabeled heparin and chondroitin sulfate side chains. This finding indicates that chondroitin sulfate should also bind to type IV collagen. By affinity chromatography, [3H]heparin bound to a type IV collagen affinity column and was eluted with a linear salt gradient, with a profile exhibiting three distinct peaks at 0.18, 0.22, and 0.24 M KCl, respectively. This suggested that heparin-type IV collagen binding was of an electrostatic nature. Finally, the effect of the binding of heparin to type IV collagen on the process of self-assembly of this basement membrane glycoprotein was studied by turbidimetry and rotary shadowing. In turbidity experiments, the presence of heparin, even in small concentrations, drastically reduced maximal aggregation of type IV collagen which was prewarmed to 37 degrees C. By using the morphological approach of rotary shadowing, lateral associations and network formation by prewarmed type IV collagen were inhibited in the presence of heparin. Thus, the binding of heparin resulted in hindrance of assembly of type IV collagen, a process previously described for interactions between various glycosaminoglycans and interstitial collagens. Such regulation may influence the assembly of basement membranes and possibly modify functions. Furthermore, qualitative and quantitative changes of proteoglycans which occur in certain pathological conditions, such as diabetes mellitus, may alter molecular assembly and possibly permeability functions of several basement membranes.

Interactions between type IV collagen and heparin were examined under equilibrium conditions with rotary shadowing, solid-phase binding assays, and affinity chromatography.
With the technique of rotary shadowing and electron microscopy, heparin appeared as thin, short strands and bound to the following three sites: the NC1 domain, and in the helix, at 100 and 300 nm from the NC1 domain. By solid-phase binding assays the binding of ['Hlheparin in solution to type IV collagen immobilized on a solid surface was found to be specific, since it was saturable and could be displaced by an excess of unlabeled heparin. Scatchard analysis indicated three classes of binding sites for heparin-type IV collagen interactions with dissociation constants of 3, 30, and 100 nM, respectively. Furthermore, by the solid-phase binding assays, the binding of tritiated heparin could be competed almost to the same extent by unlabeled heparin and chondroitin sulfate side chains. This finding indicates that chondroitin sulfate should also bind to type IV collagen. By affinity chromatography, ['Hlheparin bound to a type IV collagen affinity column and was eluted with a linear salt gradient, with a profile exhibiting three distinct peaks at 0.18, 0.22, and 0.24 M KC1, respectively. This suggested that heparin-type IV collagen binding was of an electrostatic nature. Finally, the effect of the binding of heparin to type IV collagen on the process of selfassembly of this basement membrane glycoprotein was studied by turbidimetry and rotary shadowing. In turbidity experiments, the presence of heparin, even in small concentrations, drastically reduced maximal aggregation of type IV collagen which was prewarmed to 37 "C. By using the morphological approach of rotary shadowing, lateral associations and network formation by prewarmed type IV collagen were inhibited in the presence of heparin. Thus, the binding of heparin resulted in hindrance of assembly of type IV collagen, a process previously described for interactions between various glycosaminoglycans and interstitial collagens. Such regulation may influence the assembly of base- Pardee Professor of Pathology. ment membranes and possibly modify functions. Furthermore, qualitative and quantitative changes of proteoglycans which occur in certain pathological conditions, such as diabetes mellitus, may alter molecular assembly and possibly permeability functions of several basement membranes.
Basement membranes are composed of a number of distinctive macromolecules including type IV collagen, laminin, heparan sulfate and chondroitin sulfate proteoglycans, entactin/nidogen, etc. (1). Most of these macromolecules, i.e. type IV collagen, laminin, and heparan sulfate-proteoglycan are known to self-assemble (2-5) and can also bind to each other as well as to other components of basement membranes (1). For example, EHS'-derived basement membrane-like heparan sulfate proteoglycan as well as heparin have been reported to bind to type IV collagen via the main noncollagenous, NC1 domain (6, 7). A morphological study indicated that heparan sulfate proteoglycan bound along the length of type IV collagen (8). However, resolution of this heparan sulfate proteoglycan was not adequate and in view of this drawback, the use of only a morphological approach did not allow for a better characterization of the interaction. In a different study, by affinity chromatography heparan sulfate proteoglycan was shown to bind to pepsin-extracted type IV collagen which lacks the noncollagenous NC1 domain (6). Further studies are required to better characterize the binding events between intact type IV collagen and the above mentioned glycosaminoglycans.
Previous studies revealed that other interstitial collagens, such as collagen types I, 11, and 111, bind to heparin and a number of proteoglycans via their protein core and side chains (9). In most instances this binding resulted in regulation of assembly of the above-mentioned collagens into fibrils (9)(10)(11)(12)(13)(14)(15). We therefore undertook the current study in order to examine and characterize ( a ) interactions between intact, EHS-derived type IV collagen and heparin, several disaccharide units of which resemble those of heparan sulfate proteoglycan although the latter has a lower degree of sulfation, and ( b ) the possible effects of heparin binding on the process of type IV collagen assembly.
Isolation of Type IV Collagen-Type IV collagen was isolated from The abbreviations used are: EHS, Engelbreth-Holm-Swarm; PBS, phosphate-buffered saline; BSA, bovine serum albumin; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid. the EHS tumor grown subcutaneously in lathyritic mice, as previously described (16,17). Isolated type IV collagen was used following purification by bath incubation with DEAE-52 (Whatman) in 4 M urea, (ultra-pure grade, Schwarz/Mann Biotech, ICN Biomedicals), 0.25 M NaCl, 0.05 Tris-HC1, pH 8.6, containing 1 mM EDTA, 50 pg/ ml phenylmethanesulfonyl fluoride and 50 pg/ml chloromercuribenzoate. At the end of the incubation, the suspension was centrifuged to pellet the DEAE beads, and the supernatant was dialyzed against 0.05 M Tris-HC1, pH 7.4, containing 2 M guanidine HC1 (ultra-pure grade, Sigma), 2 mM dithiothreitol, and the above-mentioned protease inhibitors (buffer A). Following dialysis, type IV collagen was centrifuged to remove aggregates larger than 50 S, at 40,000 rpm (rotor Ti-70) in a Beckman L8-M ultracentrifuge for 90 min in aliquots of 30 ml, and the supernatant was kept on ice until further use. In several instances, following incubation with DEAE-52, type IV collagen was further purified by gel filtration through a Sephacryl S-400 (Pharmacia LKB Biotechnology Inc.) column (5 X 95 cm). The column was equilibrated in 0.05 M Tris-HC1, pH 7.4, containing 2 M urea, 2 mM dithiothreitol, 1 mM glycine, and the above-mentioned protease inhibitors (buffer B). The purity of ion-exchange and gel filtrationpurified type IV collagen was determined by sodium dodecyl sulfatepolyacrylamide gel electrophoresis and as described elsewhere (2, 16,17). Protein concentration was determined by the method of Waddell (17, 18).
lz5I-Labeling of Type IV Collagen-Type IV collagen was labeled with lZ5I-NaI (Amersham Corp.) by a modification of the lactoperoxidase method (2,4). Briefly, type IV collagen was dialyzed in buffer B (without glycine) overnight at 4 "C. The next day, type IV collagen was centrifuged at 40,000 rpm for 20 min to remove large aggregates (see above). Approximately 500 pg of protein in 1-2 ml of buffer were mixed with 50 p1 of rehydrated lactoperoxidase-glucose oxidase Enzymobeads in Hz0 (Bio-Rad) in the presence of 100 pl of 0.2 M sodium phosphate buffer, pH 7.0. In sequence, 5 mCi '"I-NaI (Amersham Corp., IMS 300) and 25 pl of a solution containing 1% ~-D-glucose were added and the reaction was allowed to proceed for 30-40 min on ice. Following the end of the incubation, 100 p1 of 0.1 M dithiothreitol in H 2 0 were added, and labeled type IV collagen was separated from free "'I-NaI by gel filtration through a Sephadex G-25 (Sigma) column (1 x 25 cm) equilibrated in buffer B. Aliquots of 5 pl from eluted fractions (-500 pl/fraction) were then quantitated in a gammacounter (gamma Trac 1193, TM-analytics, Elk Grove Village, IL).
The radioactive peak which contained type IV collagen was then collected, dialyzed against buffer A, and stored on ice until further use. Incorporated "' I in type IV collagen was tested by trichloroacetic acid precipitation and was found to be 90-95% precipitable. Unlabeled type IV collagen was mixed with T -t y p e IV collagen and the mixture (at a final specific activity of 20,000 cpm/pg) was used to determine the amount of protein coated in 96-well polystyrene plates which were used for solid-phase binding assays (see below).
Solid-phase Binding Assays-A modification of the method of Skubitz et al. (19) was used. Type IV collagen, in PBS containing 0.02% sodium azide, and precleared of aggregates by centrifugation, was coated in 96-well polystyrene Immulon 1 plates (Dynatech). Aliquots, 50 pl/well, of a solution containing 60 pg/ml type IV collagen were dried overnight by incubation at 29 "C. Under these conditions 1.3 pg of type IV collagen was adsorbed/well, as determined by the use of lZ5I-type IV collagen. The concentration of 60 pg/ml was found to be optimal, since it was the minimal concentration required to bind approximately 100% or the plateau value of [3H]heparin binding (6,000 dpm). As a control, 50 pl of bovine serum albumin (BSA) (fatty-acid free, fraction V, ICN Immunobiologicals) was coated/well at 8 pg/ml, a protein concentration equimolar to that of type IV collagen. The next day, 200 pl of a solution containing 2 mg/ml BSA, 68 p~ CaC12 in 0.01 M Tris-HC1 buffered at pH 6.8 with 0.01 M HEPES (buffer C) were added to each well and incubated for 120 min at 37 "C. This step was used to "block" uncoated sites in the plastic which would adsorb additional protein. After removal of buffer, (a) 50 pl of [3H]heparin at various concentrations was added in buffer C (containing 2 mg/ml BSA) in the absence of unlabeled heparin or, ( b ) a constant amount of [3H]heparin (50,000 dpm) was incubated in the presence of either unlabeled heparin or other sulfate polysaccharides (chondroitin/dermatan sulfate and dextran sulfate). Increasing concentrations of each of these three polysaccharides were used concomitantly with 50 pl of [3H]heparin in buffer C. In all instances incubation time was 2 h, at 37 "C, at the end of which solutions were removed from the wells. Unbound [3H]heparin was removed by washing the wells three times with 200 pl of buffer C containing 0.05% Triton X-100. Bound [3H]heparin was then solubilized by incubation with 100 p1 of 0.05 N NaOH and 1% sodium dodecyl sulfate for 30 min at 60 "C and quantitated in a Beckman LS-3801 scintillation counter. Specific [3H]heparin binding was determined as total binding of [3H]heparin to type IV collagen minus binding of [3H]heparin to BSA-coated wells. A! ! experiments were performed at least three to five times in triplicate.
[3H]Heparin, of the same source as unlabeled heparin, was previously shown to behave similarly in binding studies to unlabeled heparin (19).
Affinity Chromotography-A type IV collagen affinity column was prepared as follows: 10 ml of Affi-Gel-15 (Bio-Rad) was washed with 50 ml of H20, and the beads were resuspended and allowed to settle in 20 ml of 0.15 M sodium acetate buffer, pH 7.4. The beads were equilibrated in this buffer overnight at 4 "C (3 changes, 20 ml each). Type IV collagen was dialyzed overnight against the same buffer and was cleared of aggregates by centrifugation. Approximately 8.5 mg of type IV collagen was added to the beads and allowed to interact overnight with gentle rocking at 4 "C. The next day, 100 p1 of 1 M ethanolamine, pH 7.4, was added to inactivate any remaining reactive sites. Approximately 90% of the type IV collagen which was added coupled to the beads by this procedure. The beads were then packed in a 1.5 X 5-cm Econo-column (Bio-Rad). After packing, the column was washed with 150 ml of 1 mM Tris-HC1, pH 7.0. One mg of unlabeled heparin mixed with [3H]heparin (final specific activity, -8000 dpm/pg) in a volume of 1 ml was loaded onto the column and allowed to bind to type IV collagen by incubation overnight at 4 "C. The next day, the column was washed with 100 ml of 1 mM Tris-HC1 at pH 7.0 (until levels of [3H]heparin in the eluate were tested to be at background levels, below 100 dpm) and a gradient of 0-0.3 M KC1 in the same buffer was applied. Eluted fractions were collected (-1.2 ml/fraction) and 100 p1 of each fraction was quantitated in a Beckman LS-3801 scintillation counter.
Rotary Shadowing-Type IV collagen was dialyzed against PBS overnight at 4 "C and was centrifuged to clear aggregates larger than 20 S. 200 p1 of the supernatant was mixed with heparin in PBS (final concentrations of ligands were: 150 pg/ml type IV collagen and 50-100 pg/ml heparin). The following permutations were used as controls: (a) type IV collagen alone (150 pg/ml); (b) type IV collagen (150 pg/ml) mixed with BSA (100 pg/ml); and (c) type IV collagen (150 pg/ml) mixed with both BSA and heparin, each at 100 pg/ml. All samples were incubated for 1 h at 37 'C. 75 pl of each sample was then mixed with 20% glycerol in 0.15 M NH4HC03 pH 7.7 and sprayed on freshly cleaved mica sheets as previously reported (16,17,20). Rotary-shadowed replicas were collected on 300-mesh copper grids and examined with a Philips 300 transmission electron microscope operating at 60 kV. Two types of measurements were performed (a) statistical evaluation of the fields containing lateral associations or networks of type IV collagen was done as previously described (17); ( b ) a histogram of the distribution of heparin along the length of type IV collagen molecules was constructed from photographic images of complexes at a final magnification X 500,000 (17). In this instance, the distance of a binding event from the NC1 domain was divided by the total length of each molecule (which was traced from the NC1 domain). Three criteria were used for selecting binding events. 1) The whole length of type IV collagen molecules should be clearly visualized.
2) The length of type IV collagen should be 350-450 nm.
3) The binding of heparin to type IV collagen should also be clearly visualized. A similar approach was used previously to determine binding sites of intact laminin to type IV collagen (20). Statistical analysis was performed by testing the goodness of fit of the Poisson distribution (21).
Turbidity Measurements-Type IV collagen was dialyzed against PBS overnight and then cleared of aggregates that were >20 S by centrifugation. Aliquots of the protein (250 pg/ml) were then mixed with increasing concentrations of heparin in PBS (final heparin concentrations: 10, 100,200, and 400 pg/ml) at 0 "C. One-ml aliquots of each sample were then incubated for 60 to 90 min in quartz cuvettes at 35 or 37 "C. The following controls were used (a) type IV collagen (250 pg/ml) incubated in the presence of BSA (250 pg/ml); ( b ) heparin alone (400 pg/ml) incubated at the same temperature. Temperature was maintained by automatic control with a Peltier 111 Kinetics system (Beckman). The change of absorbance at 360 nm was followed over time with a Beckman DU-6 spectrophotometer. All types of experiments, with the exception of affinity chromatography were performed with both ion-exchange-and gel filtrationpurified type IV collagen (the affinity column was made only with ion-exchange-purified type IV collagen). Similar results were obtained in most of the experiments. In solid-phase binding assays, gel filtra-tion-purified type IV collagen was occasionally observed to bind slightly higher amounts of [3H]heparin.

Rotary Shadowing
When examined with this morphological approach, heparin molecules (which are composed of side chains without a protein core) were visualized as thin, linear strands, approximately 2 to 3-nm thick and 15 to 20-nm long (Fig. 1). Heparin molecules were observed to bind to several discrete sites along the length of type IV collagen molecules (Fig. 1). A histogram of the distribution of heparin-type IV collagen-binding events, constructed as described under Materials and Methods, revealed three distinct binding sites. The highest percentage of binding events occurred in the NC1 domain. Two additional sites were evident in the triple helix-rich domain, one at a distance of approximately 100 nm from the NC1 domain and another at 300 nm from the NC1 domain (Fig. 2). These three binding sites were statistically significant when examined by the goodness of fit of the Poisson distribution (p < 0.001).
In a different experiment, type IV collagen was coincubated with heparin following passage of the latter through a type IV collagen affinity column. Heparin that bound to this column was eluted with a linear salt gradient, and three major peaks were obtained (see below). Subsequently, the ionic strength of each peak of heparin was adjusted to 0.15 M NaC1, and a sample of each was incubated with type IV collagen. Binding events were evaluated by rotary shadowing for each peak separately. All peaks of heparin had a similar distribution along the length of type IV collagen in that the majority of binding events were clustered in domain NC1 and at distances of 100 and 300 nm from the NC1 domain (Table I). These data provide evidence that charge heterogeneity of disaccharide units in commercially avzilable heparin prepa-   rations did not interfere with the binding of heparin along the length of type IV collagen.

Solid-phase Binding Assays
In solid-phase binding assays where type IV collagen was adsorbed onto plastic wells at a concentration of 1.3 pg/well, the binding of increasing concentrations of [3H]heparin in solution was found to be saturable (Fig. 3). Although some variability of the percentage of bound [3H]heparin was observed between different preparations of type IV collagen, on the average 25-30% of the added [3H]heparin bound at the lowest concentrations (which corresponded to 0.1-0.2 ng of [3H]heparin/well), and 1-2% of the added [3H]heparin bound at the highest concentrations (which corresponded to 5-7 ng of [3H]heparin/well). Fitting of the specific binding data from a representative experiment in the Scatchard equation indicated a complex curve fit (Fig. 4) with three apparent classes of binding sites. The three dissociation constants were cal-T n ._ 2 /"  Twenty different concentrations were used in triplicate wells to allow for more accurate quantitation of the binding. Type IV collagen used for this experiment was gel filtration-purified. culated to be, respectively: kdl = 3 nM ( r = 0.92, p < 0.001), kdz = 30 nM ( r = 0.97, p < 0.001) and kd3 = 100 nM ( r = 0.97, p < 0.001). Each of these three distinctive affinities may correspond to one of the three binding sites which were observed by rotary shadowing.

Competition of the Binding of PHIHeparin Binding to Type IV Collagen by Sulfated Polysaccharides
Specificity of the binding of [3H]heparin to type IV collagen with the solid-phase approach was tested by competing the binding of [3H]heparin with: ( a ) unlabeled heparin (15 kDa); ( b ) chondroitin/dermatan sulfate side chains (20 kDa); and (c) dextran sulfate (8 kDa). In this experiment, various concentrations of each of the above mentioned polysaccharides were added concomitantly in the presence of a constant amount of [3H]heparin to wells coated with 1.3 pg of type IV collagen. Both unlabeled heparin and chondroitin/dermatan sulfate were equally effective in their ability to compete with [3H]heparin for binding to type IV collagen (Fig. 5); 50% of These results suggest significant specificity in the type IV collagen-heparin interactions and also indicate that chondroitin/dermatan sulfate side chains should bind to type IV collagen as well.
In all instances, nonspecific binding of [3H]heparin was less than 100 dpm.

Binding of PHIHeparin to Type IV Collagen by Affinity Chromatography
In this assay, 1 mg of heparin (also containing small amounts of [3H]heparin) was added in 1 ml of Tris-HC1, pH 7.0, to an affinity column containing type IV collagen. Heparin was allowed to interact with type IV collagen by incubation overnight at 4 "C. The next day, a salt gradient (0-0.3 M KC1) was applied to the column, and eluted fractions were tested for radioactivity. Three distinct major peaks of [3H]heparin were obtained at about 0.18, 0.22, and 0.24 M of KC1, all eluting at ionic strengths higher than physiologic (0.15 M). A fourth minor peak was present at a concentration of 0.26 M KC1 (Fig. 6). These data suggest that the interactions between heparin and type IV collagen may be ionic.
Because commercially available heparin preparations usually contain heterogeneous mixtures of disaccharide types, the possibility existed that the major peaks of heparin which were obtained by affinity chromatography represented different disaccharide types which preferentially bound to type IV collagen based on their ionic properties. In order to rule out this possibility, the peak eluting at an ionic strength of 0.18 M KC1 was randomly selected, adjusted t o 1 mM Tris-HC1, pH 7.0, and was tested by a second passage through the type IV collagen affinity column. Bound [3H]heparin was eluted with a linear salt gradient similar to the one mentioned above. In this instance, the profile of elution again revealed three major peaks. The ionic strength of each peak from the purified heparin coincided with that obtained from the commercially available mixture of heparin disaccharides (data not shown). Therefore, these data provide evidence for the existence of three major binding sites in type IV collagen for heparin and Total dpm of eluted, tritiated heparin, and ionic strength of the applied KC1 gradient (deduced from conductivity measurements) were plotted against collected fractions. Type IV collagen was used to bind to this affinity column following ion-exchange purification. thus corroborate the rotary shadowing data and the data obtained by Scatchard analysis of radiolabeled heparin-type IV collagen binding (see above).

Effect of Heparin Binding on the Process
of Type IV Collagen Polymerization Turbidity Measurements-Self-association of type IV collagen a t 250 pg/ml in PBS, was determined by the development of turbidity. As described previously (2, 16, 17), when prewarmed to 35 or 37 "C, type IV collagen self-associated and raised turbidity readily, without a lag phase, to a plateau value within 30-40 min (Fig. 7). In the presence of heparin, a dramatic decrease in the development of maximal turbidity occurred, indicating a decrease in type IV collagen self-assembly. At a concentration as low as 10 pg/ml, heparin coincubated with type IV collagen (250 pg/ml) at 37 "C decreased turbidity by 55%. When heparin was present in higher concentrations, it further suppressed the development of turbidity. For example, at 400 pg/ml, heparin caused -80% inhibi- tion of the maximal turbidity developed by type IV collagen. Some variability was observed between different preparations of type IV collagen in that occasionally, slightly higher concentrations of heparin were required to produce similar effects (data not shown). The presence of BSA at 100 pg/ml did not have any significant effect on maximal turbidity of type IV collagen. Heparin at 400 pg/ml in PBS, alone did not raise turbidity (data not shown). These observations suggest that the binding of heparin to type IV collagen has an inhibitory effect on the process of type IV collagen self-assembly into a network-like structure.
Rotary Shadowing-Type IV collagen in PBS (150 pg/ml) was incubated in the absence or presence of heparin (100 pg/ ml) a t 37 "C for 1 h and was then examined for the formation of lateral associations and networks by rotary shadowing and electron microscopy. When incubated alone, type IV collagen has assembled to laterally associated structures in ~7 3 % of the fields examined, as previously reported (17). The presence of heparin caused a dramatic decrease in network formation by type IV collagen, where networks were present in -27% of the fields (Fig. 8). This percentage represents approximately background levels (17). The presence of BSA in either the type IV collagen solutions or the mixtures of type IV collagen and heparin did not have any significant effect on numbers of lateral associations observed (Fig. 8). These observations corroborate the findings by turbidimetry in that they indicate that heparin inhibits the process of self-assembly of type IV collagen.

DISCUSSION
In this study we report the existence of multiple interactions between type IV collagen and heparin. Heparin was selected because it has a disaccharide unit structure which in some instances resembles that of heparan sulfate proteoglycan, a component of basement membranes and cell surfaces (22)(23)(24). Though heparin is known to be heterogeneous in charge and size, the appearance by rotary shadowing indicated that the variability in size was not large, as judged by the thickness and length of heparin molecules. T o our knowledge, this is the first report visualizing heparin with the technique of rotary shadowing and electron microscopy. Visualization of heparin was important in order to determine the localization of the heparin binding along the length of the rod-like, type IV collagen molecule. This morphological study revealed, following statistical evaluation, the existence of three distinct binding sites for heparin along the length of type IV collagen.
Heparin was previously shown to bind the main noncollagenous NC1 fragment of type IV collagen by affinity chromatography, but pepsin-extracted, placental type IV collagen failed to bind to a heparin-Sepharose affinity column (7). In contrast, we report here that in rotary-shadowed images, intact (not pepsin-extracted) EHS-derived, dimeric type IV collagen bound heparin at two distinct sites along the triplehelical domain of type IV collagen, located at 100 and 300 nm from the NC1 domain, respectively. Species differences and/ or different experimental conditions might account for the discrepancy of results. However, it is also possible that pepsin treatment cleaved, at least partially, interruptions of the Gly-X-Y sequence in the helix which are abundant in both the a l -and a2-chains of type IV collagen (the deciphered sequence of the al-chain contains a total of 21 interruptions (25,26), and several matching interruptions occur in the known sequence of part of the a2-chain (27)). If one or more of these interruptions were to be involved in the binding of type IV collagen to heparin, then treatment with pepsin could have at least partially impaired this binding. Indeed, ongoing experiments indicate that EHS-derived, pepsin-treated type IV collagen binds heparin only minimally.' Taken together, the above-mentioned observations indicate that heparin, several disaccharide units of which may resemble those of heparan sulfate proteoglycan, would perhaps require an intact triple helix-rich domain, including the interruptions, for binding. In addition, intact, EHS-derived, heparan sulfate proteoglycan was reported to bind to the triple helical domain (6,8) and the NC1 domain of type IV collagen as well (6). With the technique of rotary shadowing, Laurie et al. (8) observed that heparan sulfate proteoglycan bound to two distinctive sites along the rod-like part of intact type IV collagen. The distribution of the latter binding in the triple helix-rich domain of type IV collagen, although broader, was roughly similar to the distribution of heparin along the rodlike part of type IV collagen which we reported in this study ( Fig. 1 of this report and Fig. 6 (6) using affinity chromatography and zonal rate velocity sedimentation demonstrated both the binding of intact heparan sulfate proteoglycan and isolated protein core to the NC1-and triple-helical domains of type IV collagen.
In addition to rotary shadowing, we demonstrated specific binding of heparin to intact type IV collagen by solid-phase binding assays. Scatchard analysis of the binding data indicated three classes of binding sites, which corroborated the rotary shadowing observations. The three distinctive dissociation constants obtained of 3, 30, and 100 nM indicate that these binding sites are of relatively high affinity. Furthermore, specificity of this interaction was tested by competing the binding of [3H]heparin to type IV collagen with different sulfated polysaccharides. With this approach, we obtained evidence that heparin and probably chondroitin/dermatan sulfate side chains bound to intact type IV collagen. About the same concentration of each of these two competitors was required for 50% displacement of [3H]heparin bound to type IV collagen. This observation would suggest that type IV collagen may contain two different classes of binding sites, one of which should preferentially bind chondroitin sulfate.
Affinity chromatography indicated that interactions be- tween type IV collagen and heparin are electrostatic in nature, since they were abolished by increasing ionic strength. Similarly, interactions between type IV collagen and heparan sulfate, chondroitin and dermatan sulfate (both side chains and intact proteoglycans) were shown to be charge-mediated (10). Interestingly, three major distinct peaks of type IV collagen-bound tritiated heparin were apparent, each eluting at a different ionic strength. The possibility that different subpopulations of disaccharides in the commercially available heparin mixtures were preferentially eluted because of their charge properties was ruled out by a second passage of one major peak through the type IV collagen affinity column. This charge-selected subpopulation was shown to bind to the same affinity column. The elution profile following application of a linear KC1 gradient again contained three major peaks, and the ionic strength of each coincided with that of the original elution profile. Furthermore, each peak of heparin eluted from a type IV affinity column (at 0.18, 0.22, and 0.24 M KCl, respectively) was subsequently shown by rotary shadowing to bind in all three sites along the length of type IV collagen in the NC1 domain and at distances of 100 and 300 nm from the NC1 domain, respectively. Thus by three different approaches, rotary shadowing, solid-phase binding assays, and affinity chromatography, the existence of three distinct classes of binding sites for heparin in type IV collagen was indicated.
The effect of these multiple interactions on the process of self-assembly of type IV collagen was also examined. Previous studies have shown that assembly and fibril formation of at least several interstitial collagens is regulated by heparin, chondroitin sulfate side chains, and by a number of intact proteoglycans, either side chains or the protein core (11). For example, when added before nucleation in lathyritic tropocollagen preparations, chondroitin sulfate, dermatan sulfate, heparin and intact chondroitin sulfate or dermatan sulfate proteoglycan were shown to accelerate fibril formation (11). In contrast, when the same glycosaminoglycans were added to type I collagen after the nucleation phase, they delayed fibril formation (11). Oegema et al. (12) reported that proteoglycan derived from nasal cartilage delayed fibril formation by acid-extraced type I collagen. More recently Vogel et al. (14,15) have described a small, tendon-derived dermatan sulfate proteoglycan which specifically inhibits fibrillogenesis by type I and I1 collagen in vitro.
In this study, we report that type IV collagen assembly is also inhibited by one glycosaminoglycan, heparin. Both by turbidimetry and rotary shadowing, the polymerization process of this basement membrane collagen to a complex network was drastically reduced in the presence of heparin. It remains to be substantiated whether intact proteoglycans such as heparan sulfate have similar effects. If this were the case, then a more general picture emerges, that of an effect over self-association of several families of collagens by a variety of proteoglycans, via either their side chains or protein cores or both. The precise molecular mechanisms responsible for specific effects on the assembly of a certain type of collagen are not well understood. In the case of type IV collagen, the binding of heparin in rotary-shadowing images coincided with two of the sites where NC1 domain bound in the helix in order to initiate lateral assembly (at 100 and 300 nm from the NC1 domain, respectively; 16). This observation indicates that at least for steric reasons, if these sites were occupied by heparin (or potentially by related proteoglycans as well), assembly of type IV collagen should be hindered.
Both heparan sulfate and chondroitin sulfate proteoglycan occur in basement membranes (28,29) and cell surfaces as Heparin-type IV ( well (22-24, 30, 31). Also, at least one cell type, endothelial cells in tissue culture, have been reported to contain heparin sequences in the heparan sulfate chains of their proteoglycan (32). Cell surface-heparan sulfate proteoglycan in mammary epithelial cells was reported to bind to interstitial collagens (33). One could speculate that interactions between type IV collagen and proteoglycans of either cell surfaces or basement membranes might result in partial inhibition of molecular assembly, depending upon availability of varying components. In different basement membranes, even relatively small quantitative differences in the components could lead to different molecular assembly or function. This mechanism could partially explain basement membrane polymorphism in different tissues (34). Similarly, in certain pathological conditions, such as diabetes, the observed decrease of the amount of available proteoglycans (35-38) could result in a perturbed ultrastructural assembly of newly synthesized components which might be involved in the pathogenesis of abnormal function of diabetic basement membranes.