THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 264, No. 4, Issue of February 5 , pp. 2313-2323,1989 Printed in U.S.A.

The Binding of Heparin to Type IV Collagen: Domain Specificity with Identification of Peptide Sequences from the al(1V) and a2(IV) Which Preferentially Bind Heparin*

(Received for publication, September 29, 1988)

George G. Koliakos, Kokkona Kouzi-Koliakos, Leo T. FurchtS, Lorrel A. Reger, and Effie C. TsilibaryP From the Department of Laboratory Medicine and Pathology, University of Minnesota, Minneapolis, Minnesota 55455

Three distinctive heparin-binding sites were ob- served in type IV collagen by the use of rotary shad- owing: in the NC1 domain and at distances 100 and 300 nm from the NC1 domain. Scatchard analysis in- dicated different affinities for these sites. Electron microscopic analysis of heparin-type IV collagen inter- action with increasing salt concentrations showed the different affinities to be NC1 > 100 nm > 300 nm. The NC 1 domain bound specifically to chondroitinlderma- tan sulfate side chains as well. This binding was ob- served at the electron microscope and in solid-phase binding assays (where chondroitin sulfate could com- pete for the binding of [‘Hlheparin to NC1-coated sub- strata). The triple helix-rich, rod-like domain of type IV collagen did not bind to chondroitin/dermatan sul- fate side chains. In solid-phase binding assays only heparin could compete for the binding of [‘Hlheparin to this domain. In order to more precisely map potential heparin-binding sites in type IV collagen, we chemi- cally synthesized 17 arginine- and lysine-containing peptides from the al(1V) and a2(IV) chains. Three peptides from the known sequence of the al(1V) and a2(IV) chains were shown to specifically bind heparin: peptide Hep-I (TAGSCLRKFSTM), from the al(NC1) chain, peptide Hep-I1 (LAGSCLARFSTM), a peptide corresponding to the same sequence in peptide Hep-I from the a2 (NC1) chain, and peptide Hep-I11 (GEFYF- DLRLKGDK) which contained an interruption of the triple helical sequence of the al(1V) chain at about 300 nm from the NC1 domain, were demonstrated to bind heparin in solid-phase binding assays and compete for the binding of [‘Hlheparin to type IV collagen-coated substrata. Therefore, each of these peptides may rep- resent a potential heparin-binding site in type IV col- lagen. The mapping of the binding of heparin or related structures, such as heparan sulfate proteoglycan, to specific sequences of type IV collagen could help the understanding of several structural and functional properties of this basement membrane protein as well as interactions with other basement membrane and/or

*This study was supported by research Grant 39216 from the National Institutes of Diabetes, Digestive and Kidney Diseases/ National Institutes of Health (NIH) and grants from the American Diabetes Association, the Juvenile Diabetes Foundation Interna- tional, the Viking Children’s Fund, and the Minnesota Medical Foundation (to E. C. T.) and National Institutes of Health Grants AM-07651, CA-29995, CA-21463, and EI-39510 (to L. T. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Recipient of an Allen-Pardee Professorship. § To whom all correspondence should be addressed.

cell surface-associated macromolecules.

Type IV collagen, a basement membrane protein, has an unusual structure containing a large, noncollagenous NC1 domain at the carboxyl-terminal end, and numerous interrup- tions along the length of the Gly-X- Y triple-helical sequence. For instance, a total of 21 interruptions, 2 to 11 residues long, have been described in the total length of the deciphered sequence of the al(1V) chain (1-3). Matching interruptions plus an unmatched discontinuity of 20 amino acids have been reported in the known part of the sequence of the a2(IV) chain (4-6). Such interruptions have been hypothesized to contribute to increased flexibility of type IV collagen when compared to other interstitial collagens (7,8).

Several functional properties of type IV collagen have been elucidated thus far. For example, it is known that type IV collagen self-assembles to form an irregular polygonal struc- ture (9, 10); it binds to laminin (11-13), nidogen (14, 15), and heparan sulfate proteoglycan (12, 15, 16). In addition, we recently observed multiple interactions between type IV col- lagen and heparin, a glycosaminoglycan of which several disaccharide types may resemble the side chains of heparan sulfate proteoglycan (17). Rotary shadowing images examined by electron microscopy suggested that three different binding sites should exist in type IV collagen for heparin binding: in the NC1 domain and at distances of 100 and 300 nm from the NC1 domain, respectively. These heparin-binding sites in type IV collagen apparently have different affinities (17). This report extends our earlier observations and biochemically defines the major domains of type IV collagen which bind to heparin. The main noncollagenous NC1 domain contains the site that has the highest affinity for binding to heparin as determined by resistance to increasing salt concentrations. This domain also preferentially binds chondroitin/dermatan sulfate side chains. The other heparin-binding sites from the triple helix-rich region of type IV collagen fail to show appre- ciable binding to chondroitin/dermatan sulfate. Furthermore, three distinct peptides were identified from the published sequence of the al(1V) and a2(IV) chains, which specifically bound heparin and likely participate in the binding of type IV collagen to this specific glycosaminoglycan.

MATERIALS AND METHODS

Isolation and Purification of Type IV Collagen and Type IV Collagen Domains

Isolation of Type IV Collagen-Type IV collagen was isolated from the Engelbreth-Holm-Swarm tumor grown subcutaneously in mice which were rendered lathyritic by the addition of 0.1% P-aminopro-

2313

2314 Binding of Heparin to Domains of Type IV Collagen prionitrile fumarate (Sigma) in their drinking water, as described previously (18, 19).

DEAE-52 (Whatman) in 4 M urea (ultrapure grade, Schwarz/Mann Isolated type IV collagen was purified using batch incubation with

Biotech, ICN Biomedicals, Inc., Cleveland, OH), 0.05 M Tris-HC1, pH 8.6, containing 0.25 M NaCI, 1 mM EDTA, 50 pg/ml phenyl- methanesulfonic acid and 50 pg/ml chloromercuribenzoate, as re- ported previously (18,19). In several instances, ion exchange-purified type IV collagen was further purified by gel filtration through a Sephacryl S-400 (Pharmacia LKB Biotechnology Inc.) column (5 X 95 cm) equilibrated in 0.05 M Tris-HC1, pH 7.4, 2 mM dithiothreitol, 2 M urea, 1 mM glycine, 50 pg/ml phenylmethanesulfonic acid, and 50 pg/ml chloromercuribenzoate. The purity of type IV collagen was determined by SDS'-polyacrylamide gel electrophoresis (6% gel) and is reported elsewhere (19). The concentration of type IV collagen was determined by the method of Waddell (20, 21).

Isolation of the Main, Noncollagenous NCl Domain of Type I V Collagen-The globular domain of type IV collagen was prepared by digestion with bacterial collagenase (CLSPA Cooper Biomedical Inc., Malvern, PA) as previously described (19, 21). The isolated NC1 domain was further purified by gel filtration through a Sephacryl S- 300 (Pharmacia LKB Biotechnology Inc.) column (2.5 X 95 cm) equilibrated with 0.2 M NH4HC03, pH 8.5, as described elsewhere (21). Purity of the NC1 domain was tested by SDS-PAGE (10% gel) and rotary shadowing. Protein concentration was determined by the Lowry method (22).

Isolation of the Triple Helix-rich Domain of Type IV Collagen- Triple helical-domain fragments were prepared by a brief digestion of intact type IV collagen with pepsin (Cooper Biomedical Inc.) as described previously (9). Pepsin-treated type IV collagen was further purified by gel filtration through a Sephacryl S-400 column (5 X 95 cm) as described for intact type IV collagen (see above). The concen- tration of the triple helical fragments was determined by the method of Waddell (20).

Synthesis of Peptides Derived from the al(IV) and a2(IV) Chains Peptides derived from the sequence of either the NC1- or triple

helix-rich domains were synthesized according to the method of Barany and Merrifield (23) on a solid-phase support. Deprotection and release of peptides was achieved using HF containing 10% anisole for 1 h at 4 "C. Peptides were extracted with ether, dissolved in 10% acetic acid, filtered, and lyophilized. Peptides were checked for purity by HPLC, amino acid analysis, and eventually amino acid sequencing.

Peptides were HPLC-purified using a 21.1 X 250-mm C18 reverse- phase column and a gradient of acetonitrile containing 0.1% (v/v) of trifluoroacetic acid. To initially screen for biological activity, non- HPLC-purified peptides were used. Peptides which were positive for binding to heparin were further purified by HPLC and retested for activity. All three peptides which bound to heparin were active both before and after HPLC purification.

Solid-phase Binding Assays

Direct Binding of PHIHeparin in Solution to Substrata Coated with Protein, Fragments, or Peptides-The binding of [3H]heparin in solution to type IV collagen, pepsin-treated type IV collagen, isolated NC1 domain, and various peptides coated onto a solid support was tested by a modification of the method of Skubitz et al. (24). Type IV collagen and its major proteolytic fragments were coated onto plastic a t equimolar amounts. 50 p1 of ( a ) intact type IV collagen at 60 fig/ ml, ( b ) pepsin-treated type IV collagen at 52 pg/ml, (c) isolated NC1 domain at 20 pg/ml, and ( d ) each peptide at 50-500 pg/ml, all in PBS, were dried in 96-well plates (Immulon 1, Dynatech Laboratories, Inc., Alexandria, VA) by incubating overnight at 29 'C. As a control, 50 p1 of a solution of BSA at 8 pg/ml in PBS, an amount approxi- mately equimolar to that of type IV collagen and the major proteolytic domains, were coated in 96-well plates as described above. Before coating, intact type IV collagen was cleared of aggregates >200 S by centrifugation (15,000 rpm x 20 min, in a Ti70 rotor). The next day, 200 p1 of a solution containing 2 mg/ml BSA and 68 p M CaCL in 10 mM Tris-HC1-HEPES, pH 6.8 (buffer A), were added and incubated at 37 "C for 90 min, to "block residual sites on the plastic that would absorb protein. The solution was then aspirated and 50 p1 of [3H]

The abbreviations used are: SDS, sodium dodecyl sulfate; HPLC, high performance liquid chromatography; BSA, bovine serum albu- min; PBS, phosphate-buffered saline; HEPES, 4-(2-hydroxyethyl)-l- piperazineethanesulfonic acid.

heparin were added at increasing concentrations, in the same buffer. The wells which were coated with intact type IV collagen, or in several instances isolated NC1 domain, were treated concomitantly with 20 p1 of purified anti-NC1 IgG (see below) or normal rabbit IgG (which was used as a control), a t a concentration of 300 pg/ml. In the remaining wells, 20 pl of buffer A were added instead. The mixture was allowed to incubate for 3 h at 37 "C, and all wells were then washed three times with 200 p1 of buffer A also containing 0.1% Triton X-100. Bound [3H]heparin was solubilized by incubation at 60 "C for 30 min in the presence of 100 pl of a solution containing 0.5 N NaOH and 1% SDS. Radioactivity was quantitated with a Beckman LS-3801 scintillation counter.

Competition of the Binding of PHfHeparin in Solution to Solid- phase Type I V Collngen, Isolated NCl Domain, and Various Peptides by Sulfated Mucopolysaccharides-The isolated NC1 domain, intact type IV collagen, and various peptides were coated in Immulon 1,96- well plates, as described above. In type IV collagen-containing wells following incubation with buffer A, 20 p1 of anti-NC1 IgG at 300 pg/ ml were added. Subsequently, for all proteins, fragments, and peptides a constant amount of [3H]heparin (50,000-500,000 dpm, specific activity =lo6 dpm/pg) was added per well in buffer A, in the presence of increasing concentrations of ( a ) unlabeled heparin, ( b ) chondroi- tin/dermatan sulfate side chains, and (c) dextran sulfate. The mixture was allowed to incubate at 37 "C for 3 h, the wells were washed and processed, and solubilized bound [3H]heparin was quantitated as described above.

Competition of fH]heparin/Type IV Collagen Binding by Various Peptides-Type IV collagen was coated on the surface of 96-well plates as described above. A constant amount of heparin (500,000 dpm:500 ng) in buffer A was mixed with increasing concentrations of each separate peptide in the same buffer, and the mixtures were incubated for 2 h a t 37 "C in separate wells. At the end of the incubation period, 50 pl of the mixture containing the various con- centrations of each peptide incubated in the presence of [3H]heparin were transferred to type IV collagen-coated wells, following incuba- tion of the latter wells with buffer A. The wells were further incubated and washed, and bound [3H]heparin was quantitated as described above.

All the solid-phase binding experiments were done in triplicate, a minimum of five times. Both ion exchange- and gel filtration-purified type IV collagen were used with similar results. The amount of protein or peptide coated to plastic was quantitated by the use of '261-labeled proteins, domains, and peptides.

Purification of Anti-NC1 IgG Polyclonal antibodies directed against the purified NC1 domain

from murine type IV collagen were raised in rabbits as described previously (19). Briefly, the purified NC1 domain was injected sub- cutaneously in complete Freund's adjuvant in female New Zealand rabbits, in a total of three injections (100 pg/injection) a t 2-week intervals. Two weeks after the last injection, the animals were bled, and the antiserum was tested by enzyme-linked immunosorbent assay. This antiserum was found to be equally reactive with the NC1 domain and intact type IV collagen and did not show any significant reactivity with the triple helical domain of type IV collagen (19).

Anti-NC1 IgG was prepared from immune serum using 50% am- monium sulfate precipitation and then pelleted by centrifugation (12,000 rpm X 30 min). The pellet was solubilized in H20 and was dialyzed first against 0.025 M Tris-HCI, pH 8.8, at 4 "C and then against the same buffer containing 0.035 M NaC1. It was then loaded onto a DEAE-Trisacryl M (IBF Biotechnics, Savage, MD) column (5.6 X 9 cm) which was equilibrated in the same buffer, as described (25). Unbound fractions of IgG were pooled and dialyzed against 10 mM Tris-HC1-HEPES, pH 6.8. After overnight dialysis, CaCb was added to the IgG solution to a final concentration of 68 pM, and the anti-NC1 IgG was loaded onto a heparin-Sepharose (Sigma) column (1.5 X 5 cm), equilibrated in the same buffer, Tris-HC1-HEPES, 68 p~ CaC12, pH 6.8. Fractions of nonbound IgG were pooled and reloaded for a second passage through the same column, which was previously washed with 2 M NaCl in 10 mM Tris-HC1-HEPES, pH 6.8. Unbound IgG was pooled, concentrated to 0.3 mg/ml, and kept at -70 "C in the same buffer containing 0.02% NaN3. Purified anti- NC1 IgG was tested for binding to NC1 by enzyme-linked immuno- sorbent assays.

Binding of Heparin to Domains of Type IV Collagen Rotary Shadowing

Type IV collagen was dialyzed against PBS or appropriate buffer (see below) overnight at 4 "C and was centrifuged to clear aggregates larger than 200 S, as mentioned above. The following experiments were performed. ( a ) 200 p1 of type IV collagen at 300 pg/ml in PBS, pH 7.4, were mixed with an equal amount of chondroitin/dermatan sulfate at 400 pg/ml in the same buffer. The mixture was allowed to incubate at 37°C for 1 h. ( b ) 200 p1 of type IV collagen at 300 pg/ml in 200, 240, and 260 mM KC1 (buffered with 10 mM phosphate, pH 7.4), respectively, were mixed with an equal amount of heparin at 400 pg/ml each in the corresponding buffer. Mixtures were allowed to incubate a t 37 "C for 1 h. (c) 200 pl of type IV collagen in PBS at 150 pg/ml were incubated in the presence of heparin (final concentration 200 pg/ml) and anti-NC1 IgG (total IgG added 600 ng) a t 37 "C for 1 h.

In all instances, 100 pl of each mixture were then added to 900 p1 of a solution containing 20% glycerol and 0.15 M NHaHC03. Each sample was then gently mixed, sprayed on freshly cleaved mica sheets, and processed for rotary shadowing as mentioned in previous reports (11, 19, 21). Replicas were floated in HzO, collected on 300-mesh copper grids, and then examined with a Philips-300 transmission electron microscope operating at 60 kV.

Histograms of the distribution of chondroitin/dermatan sulfate or heparin along the length of type IV collagen were constructed in each case with the use of a semi-automated Optomax system from prints

al

w >

0)

U c C

30 .- ._ m c

20 0

0 A

c al 3 u l o E

8 LL

0 0 0.2 0.4 0.6 0.8 1 .o

Length of Type IV Collagen

C

2315

at a final magnification of X 500,000 as described elsewhere (19, 21). Binding events were selected in each case based on the following criteria: 1) type IV collagen molecules should be clearly visualized throughout their whole length. 2) The length of type IV collagen molecules should be 350-450 nm. 3) Each binding event should be clearly visualized.

[3H]Heparin (0.4 mCi/mg) was purchased from Du Pont-New England Nuclear. Heparin (porcine intestinal mucosa, grade I, 15 kDa), dextran sulfate (8 kDa), and chondroitin/dermatan sulfate (porcine skin, type B, 20 kDa) were all purchased from Sigma. [3H] Heparin of the same source as unlabeled heparin was previously shown to behave similarly in binding studies to unlabeled heparin (24).

Iodination '''I labeling of type IV collagen, pepsin-treated type IV collagen,

and isolated NC1 domain was performed by the use of Enzymobeads (Bio-Rad) as reported elsewhere (21). Peptides were iodinated by the chloramine-T method (26).

RESULTS

Rotary Shadowing-Heparin was previously shown to bind in three distinct sites along the length of type IV collagen: the NC1 domain, and in the triple helix-rich domain, at

50

40

30

20

10

0

Length of Type IV Collagen

0) C U c ._ .- m - 0

0 0.2 0.4 0.6 0.8 1 .o Length of Type IV Collagen

FIG. 1. Histogram of the distribution of heparin along the length of type IV collagen constructed from examination of rotary shadowing images. The total length of type IV collagen was divided into 20 equal segments, each 20 nm long. The main noncollagenous NC1 domain is represented in the first 10 nm (first of 20 segments). Type IV collagen (150 pglml) was incubated with heparin (200 pg/ml) at 37 "C for 1 h in phosphate buffer containing 0.2 M KC1 ( A ) , 0.24 M KC1 ( B ) , or 0.26 M KC1 (C). The number of binding events in each case was 70.

2316 Binding of Heparin to Domains of Type IV Collagen

FIG. 2. A, rotary shadowing images from an experiment in which chondroi- tin sulfate (200 pg/ml) was co-incubated with type IV collagen (250 pg/ml) a t 37 "C for 1 h in PBS. Arrows indicate chondroitin/dermatan sulfate side chains. Bar equals 100 nm. B, histogram of the distribution of chondroitin/der- matan sulfate along the length of type IV chondroitin constructed from rotary shadowing images. The total length of type IV collagen was divided into 20 segments, each 20 nm long. The first segment (0-10 nm) represents the NC1 domain. The number of binding events used to construct this histogram was 70.

A

60

40

20

0 0

distances of 100 and 300 nm from the NC1 domain (17). By affinity chromatography, [3H]heparin bound to type IV col- lagen eluted with a linear KC1 gradient at the following ionic strengths: 0.18, 0.22, and 0.24 M KCl, respectively. In order to determine the exact site where each of these three different peaks of [3H]heparin bound to type IV collagen, we coincu- bated type IV collagen and heparin at the following ionic strengths: 0.20, 0.24, and 0.26 M KC1 and then examined interactions by rotary shadowing. At 0.2 M KC1, the histogram showed two distinctive peaks for heparin binding to type IV collagen: the first peak was observed in the NC1 domain (0- 20 nm from the carboxyl-terminal end of type IV collagen) and represented 31% of binding events. The second peak was observed at a distance of 80-120 nm from the NC1 domain and also represented 31% of binding events (15% of binding events occurred at 80-100 nm and 16% occurred at 100-120 nm) (Fig. LA).

0.2 0.4 0.6 0.8 1 .o Length of Type IV Collagen

When a concentration of 0.24 M KC1 was used, heparin was observed to bind only in the NC1 domain (0-20 nm from the COOH terminus of type IV collagen). A total number of 51% of the heparin type IV collagen binding events were present at this site, whereas the rest of the binding events were distributed randomly in all other segments of type IV collagen (Fig. 1B).

Finally, at 0.26 M KC1, no distinctive peak for heparin binding was observed throughout the length of type IV colla- gen (Fig. 1C).

These rotary shadowing data indicate that the binding interaction which disappears at 0.2 M KC1 (the lowest salt concentration used) corresponded to the binding site which is located at 300 nm from the NC1 domain (17). At 0.24 M KC1, the peak which corresponded to the binding interaction lo- cated at 100 nm from the NC1 domain was eliminated,

Binding of Heparin to Domains of Type IV Collagen 2317

FIG. 3. Rotary shadowing images of type IV collagen incubated in the presence of both heparin and anti- NC1 IgG. Anti-NC1 IgG is shown to bind to the NC1 domain (A-D) , thus making the size of the globule larger than usual (arrowheads). Several heparin molecules (arrows) are shown to bind at 100 (C) and 300 (D) nm from the NC1 domain. Bar equals 100 nm.

TABLE I Incidence of binding events of heparin along the length of type IV

collagen in rotary shadowing images (expressed as percentage of the total number of bound heparin molecules examined)

The number of binding events in all other sites along the length of type IV collagen was <4.5%.

Distance from the NC1 domain Total number of binding

0-20 nm 80-120 nm 280-320 nm events

Type IV collagen plus 28% 25% 18% 147 heparin (from Fig. 2 in Ref. 17)

heparin plus a-NC1 IgG

Type IV collagen plus 3% 40% 41% 90

whereas the heparin binding interaction with the NC1 peak persisted. The latter peak or binding interaction was elimi- nated only at the highest salt concentration used (0.26 M KCI).

When chondroitin/dermatan sulfate was tested for binding to type IV collagen by rotary shadowing, it bound to the NC1 domain of type IV collagen (Fig. 2 A ) . Only one peak for chondroitin/dermatan sulfate binding was present in the con- structed histogram, which contained 69% of binding events in this major noncollagenous domain of type IV collagen. The rest of the binding events were distributed randomly along the triple helix-rich domain of type IV collagen (Fig. 2B) .

When type IV collagen was incubated in the presence of heparin and anti-NC1 IgG concomitantly, the following were observed by rotary shadowing: (a) anti-NC1 IgG was bound to the NC1 domain of most type IV collagen molecules ex- amined, resulting in a substantially larger size of the NC1 globules (Fig. 3, A and B ) , and ( b ) in such molecules, heparin did not bind to the NC1 domain, although heparin binding at the sites at 100 and 300 nm from the NC1 domain was maintained (Fig. 3, C and D and Table I).

Solid-phase Binding Assays-Previous work with this ap- proach showed that [3H]heparin bound to intact type IV collagen in a saturable manner (17). This report describes

that the isolated NC1 domain also binds [3H]heparin (Fig. 4A). When the isolated NC1 domain was coated onto plastic surfaces at 20 pg/ml, about 6% of the added [3H]heparin was observed to bind at the lowest concentrations added; at the highest concentrations, about 0.5-0.6% of added [3H]heparin was bound (Fig. 4A). This binding of heparin to the NC1 domain was also saturable and appeared to be specific, since it could be displaced by an excess of unlabeled heparin (Fig. 4B) . Perhaps paradoxically, chondroitin/dermatan sulfate was observed to compete better for the binding of [3H]heparin to the NC1 domain than unlabeled heparin. Fifty percent of bound [3H]heparin (which represented 25 ng) could be com- peted by chondroitin/dermatan sulfate at a concentration of 0.02 p ~ , whereas -1 p~ (or 50 times more) heparin and 10 p~ (or 500 times more) dextran sulfate were required to achieve 50% displacement of the NC1-bound [3H]heparin. Therefore, chondroitin sulfate should bind to the isolated NC1 domain with higher affinity than heparin and at near or similar sites.

The binding of heparin to the triple helix-rich domain of type IV collagen was also examined. Originally, pepsin-treated type IV collagen, which lacks the NC1 domain, was used, but this fragment failed to bind heparin in solid-phase binding assays (data not shown). Failure of heparin binding was not due to the inability of this fragment to adhere to plastic surfaces. By the use of lZ5I-labeled type IV collagen, NC1 domain, and triple helical domain, we observed that 40-50% of each protein or fragment bound to Immulon 1 96-well plates, when dried at concentrations of 20-80 pg/ml (data not shown). We then examined the possibility that pepsin affected the integrity of the triple helix-rich domain by attacking, at least partially, pepsin-sensitive, noncollagenous interruptions of the triple helix. Such interruptions could be involved in the binding of heparin to the rod-like part of type IV collagen. TO test this possibility, plastic surfaces were coated with intact type IV collagen and incubated with increasing concentrations of [3H]heparin, in the presence of 6 pg of anti-NC1 IgG. This purified IgG was first shown to block the binding of [3H] heparin in solution to NC1-coated substrata. Normal rabbit

Binding of Heparin to Domains

A of

1 1.2 1

0.4 o'6 b! 0.2 f 0.0 7 I I I

Type IV Collagen

B 1200

1000

800

600

400

200

0 0 100 200 300 ,001 .01 .1 1 i o 1 0 0 i o 0 0

Concentration of inhibitor (pM) [HI Heparin added (ng)

FIG. 4. A, binding of [3H]heparin in solution to domain NC1 coated onto plastic. [3H]Heparin was allowed to interact with domain NC1 for a period of 2 h at 37 "C before washing and quantitation. Bars represent standard deviation. B, competition of the binding of [3H]heparin in solution to domain NC1-coated substrata by chondroitin/ dermatan sulfate (A-A), unlabeled heparin (M), and dextran sulfate (A-A). 500 ng of [3H]heparin were used per well inthis experiment (specific activity -lo6 dpm/Fg). Standard deviation was less than 3%.

2000

1000

0 BSA NC1 NCl+NRIgG NCl+AntiNCl

BSA c 4 C4+ANC1

FIG. 5. A, binding of [3H]heparin in solution to domain NC1- coated substrata, in the absence of IgG (NCl), in the presence of normal rabbit IgG(NC1 + NRZgG), or specific anti-NC1 IgG (NCI + AntiNCl). 500 ng of [3H]heparin were added per well. BSA represents nonspecific binding of [3H]heparin to BSA-coated substrata. Stand- ard deviation was less then 5%. B, binding of [3H]heparin in solution to type IV collagen-coated substrata without IgG (C4) or in the presence of specific anti-NC1 IgG (C4 + ANCl). 500 ng of [3H] heparin were added per well. BSA represents nonspecific binding of [3H]heparin to BSA-coated substrata. Standard deviation was less than 5%.

IgG did not interfere with the binding of [3H]heparin to domain NC1 (Fig. 5A). These data provided evidence that in the presence of anti-NC1 IgG, the NC1 domain of type IV collagen does not bind [3H]heparin in solid-phase binding assays. In similar experiments performed in the presence of anti-NC1 IgG, the binding of [3H]heparin to intact type IV collagen-coated substrata was decreased by -60% (Fig. 5B). The residual binding of [3H]heparin to type IV collagen was presumed to be contributed by sites in the triple helix-rich domain. These findings were confirmed by rotary shadowing examination of samples of type IV collagen which were incu- bated with heparin and anti-NC1 IgG concomitantly (Fig. 3 and Table I; see above). This heparin binding was saturable (Fig. 6A) and could be competed by an excess of unlabeled heparin. When type IV collagen was coated onto plastic surfaces at 60 wg/ml and a large excess of [3H]heparin was added, approximately 3% of added [3H]heparin bound at the lowest concentrations, whereas -0.6-0.7% was observed to bind at the highest concentrations (Fig. 6B). Unlabeled hep- arin at 4 PM was required to compete for 50% of the amount of bound [3H]heparin. Interestingly, neither chondroitin/der- matan sulfate nor dextran sulfate could compete well for the binding of [3H]heparin to the triple helix-rich domain of type IV collagen. More than 1 mM of each of these two mucopoly- saccharides was apparently required for 50% displacement of bound [3H]heparin, a concentration in excess of the range used in the solid-phase binding experiments (Fig. 6B). These data indicate that the triple helix-rich domain of type IV collagen should bind heparin but not chondroitin/dermatan sulfate. This finding was confirmed by examination of rotary shadowing images of type IV collagen which was incubated in the presence of chondroitin/dermatan sulfate (Fig. 2, A and B; see above).

Identification of Peptide Sequences from alfIV) and a2fIV) Which Bind Heparin-In order to more precisely map the heparin-binding domains in type IV collagen, a variety of regions were selected from the published sequence of the a1 and a2 chains of type IV collagen based on several criteria: first, the presence of lysine and/or arginine residues was favored because these amino acids have a relatively high number of positive charges. This was considered to be a valid criterion, since electrostatic forces may play an important role

Binding of Heparin to Domains of Type IV Collagen

A 3.0" B

3000 - 2.5 - E,

E 2.0 - 2000 -

3 0 n C .- L

E 1000 - I - 3 m 0 . . . ...., . .... '..I - - ""1 - -*""I ' ' . '"T ' .-

1 1 0 100 11 0 200 400 600 800 ,001 .01 . I

[HI Heparin added (ng) Concentration of inhibitor (pM)

FIG. 6. A, Binding of [3H]heparin to type IV collagen coated substrata, in the presence of 6 pg of specific anti- NC1 IgG. 500 ng of [3H]heparin was added and was allowed to interact with the triple helix-rich domain of type IV collagen for 2 h a t 37 "C before washing and quantitation. Bars represent standard deviation. B, competition of the binding of [3H]heparin to type IV collagen-coated substrata in the presence of 6 pg of anti-NC1 IgG by unlabeled heparin (M), chondroitin/dermatan sulfate (A-A), and dextran sulfate (A-A). Concentration of added [3H]heparin per well was 500 ng (specific activity: lo6 dpm/pg). Standard deviation was less than 5%.

2319

TABLE I1 Peptides tested for binding to heparin

From the NC1 domain"

Peptide 1 NPLCPPGTKIL L Y ~ NC1; 17-27 Peptide 2 (Hep-I) TAGSCLRKFSTM 011 NC1; 49-60 Peptide 3 APFIECHGRGTCN ( ~ 1 NC1; 171-185 Peptide 4 MFKKPTPSTLKAGELR 0(1 NC1; 201-216 Peptide 5 (Hep-11) LAGSCLARFSTM (YZ NC1; 49-60 Peptide 6 TPFICRNGGRGTTCH 012 NC1; 168-184

From the triple helix-rich domain of the a,(IV) chain*

Peptide 7 Peptide 8 Peptide 9 Peptide 10 (Hep-111) Peptide 11 Peptide 12 Peptide 13 Peptide 14 Peptide 15 Peptide 16 Peptide 17

GLSFQGPK

GMPGJGEK GEFYFDLRLKGDK GPKGEPGKIVPLPG GLPGKPGSMDKVDMGSMKG GVPGKDGQAGQPGQP GEKGDKGLPGLD GHATEGPK

GPYDIIKGQP -

GQA~VQEKGDFATKGEK

-

GVKGDKGNPGWPGAP

221-228 244-260 270-277 531-543 633-645 930-945 975-989

1115-1126 1228-1235 1263-1277 1350-1359

Peptides are numbered from the carboxyl-terminal end of type IV collagen. * Discontinuities of the Gls-X-Y sequence, when Dresent, are underlined. Peptides are numbered from the

amino-terminal end of type IV collagen.-

in the binding of heparin to type IV collagen (17). Second, sequences from the NC1 domain were selected which have a relatively low (or negative) value of hydropathy index. Such sequences in vitronectin, laminin, and fibronectin were found to bind heparin (27-30). To apply this criterion, the hydrop- athy index of the sequence of the a1 and a 2 NC1 chains (31- 34) was constructed, using the Intelligenetics (Mountain View, CA) computer program with a span setting of six amino acids (data not shown). Third, for the triple helix-rich domain, sequences mainly were selected containing discontinuities of the triple helix from the published sequence of the complete length of the al(1V) chain (12). This possibility was suggested by the lack of binding of pepsin-treated type IV collagen to heparin in solid-phase binding assays (see above). We rea- soned that pepsin partially cleaved several of the discontinu- ities and could thus abolish binding to heparin. In addition to discontinuities, several triple helical sequences were selected based on their lysine or arginine content (27) to test for

binding to heparin, as a potential control. A total of 17 peptides were synthesized and tested.

The amino acid sequences of the peptides which were chemically synthesized and used in this study and their local- ization in the al(1V) and a2(IV) chains are indicated in Table 11.

Binding of PHIHeparin to Synthetic Peptides-The binding of increasing concentrations of [3H]heparin to peptides dried on plastic surfaces was first examined. Three peptides, peptide Hep-I, a dodecapeptide from the a1 (NC1) chain, peptide Hep-11, a homologous dodecapeptide from the a 2 (NC1) chain, and peptide Hep-111, from the triple helix-rich domain of al(IV), which also contained a discontinuity 7 residues long (total length being thirteen), all bound heparin in a concen- tration-dependent and saturable manner (Fig. 7, A-C). The rest of the peptides tested were negative for binding to [3H] heparin. When peptide solutions of Hep-I, Hep-11, and Hep- I11 at 500 pg/ml each were plated, peptides Hep-I1 and Hep-

Binding of Heparin to Domains of Type IV Collagen

0 100 200 300 400 500

3[H] Heparin added (ng)

300

200

100

0 0 200 400 600 800 1000 1200

3[H] Heparin added (ng)

200

150

100

50

0

beled heparin (Fig. 8, A-C). In all cases, unlabeled heparin at approximately 0.2-0.3 p~ was able to displace 50% of the bound, [3H]heparin. Dextran sulfate could also compete for the binding of [3H]heparin to the peptides a t concentrations of 0.5-1 pM, whereas -500 times more chondroitin sulfate was required to achieve a similar competition.

Next, the specificity of the binding of peptides Hep-I, Hep- 11, and Hep-I11 to heparin was determined based on the ability of the peptides to compete for the binding of [3H]heparin in solution, to intact type IV collagen-coated substrata. These solid-phase competition binding assays indicated that Hep-I, Hep-11, and Hep-111, but none of the other peptides are able to compete (ie. peptide 3; Fig. 9A; peptides 8 and 9, Fig. 9B). About 10 pg of each of the three peptides, when incubated with [3H]heparin, could reduce the amount of [3H]heparin binding to type IV collagen by -50% (Fig. 9, A and E). Therefore, peptides Hep-I, Hep-11, and Hep-111 apparently bound heparin specifically when coated to plastic or when present in solution.

DISCUSSION

We have previously demonstrated that heparin binds to type IV collagen with multiple interactions (17). By the method of rotary shadowing, heparin appeared to bind to three distinctive sites in type IV collagen, which were the NC1 domain, and sites located at distances of 100 and 300 nm from the NC1 domain. These multiple interactions had three different K d values and were shown to be electrostatic in nature (17). In this report, we extend the original obser- vations and examine the specificity for binding to heparin of each of the two major domains of type IV collagen: the main noncollagenous, NC1- and the triple helix-rich domains.

The technique of rotary shadowing suggested that the strongest heparin-binding site is localized in the NC1 domain of type IV collagen as determined by resistance to increased salt concentrations. The site of higher affinity for binding heparin in type IV collagen, following the NC1 domain-site, should correspond to that located at 100 nm from the NC1 globule. The site which is located at 300 nm from the NC1 globule should then represent the site of lowest affinity for binding to heparin, on type IV collagen. The salt concentra- tions used in these rotary shadowing experiments were se- lected based on our previous data in which heparin bound to a type IV collagen affinity column and eluted with a linear KC1 gradient at three different ionic strengths of 0.18, 0.22, and 0.24 M KCl, respectively, all above physiologic ionic strength (17).

In solid-phase binding assays, both the main noncollage- nous and triple helix-rich domains of type IV collagen could bind [3H] heparin in a concentration-dependent and saturable manner. This binding was apparently specific, since it could be displaced by an excess of unlabeled heparin. The data indicate that integrity of the triple helix-rich domain, includ- ing interruptions of the Gly-X- Y sequence, should be impor- tant for binding of this domain to heparin. Pepsin-treated type IV collagen, which may have some of the discontinuities at least partially cleaved by pepsin, failed to bind heparin in solid-phase binding experiments. The lack of binding of hep- arin to pepsin-extracted, placental type IV collagen has been previously observed by the use of a type IV collagen affinity column, although it was not attributed to the treatment of type IV collagen with pepsin (35). In order to circumvent this difficulty, we used intact type IV collagen and sterically blocked the NC1 domain by the use of specific anti-NC1 IgG, which was present in heparin solutions throughout the incu-

0 200 400 600 800 1000 12001400

3[H] Heparin added (ng)

FIG. 7. Binding of ['Hlheparin in solution to peptide Hep-I ( A ) , peptide Hep-I1 ( B ) , and peptide Hep-I11 (C)-coated sub- strata (peptides Hep-I, Hep-11, and Hep-I11 were coated at a concentration of SO0 sg/ml). [3H]Heparin was allowed to interact with each peptide for 2 h at 37 "C before washing and quantitation. Bars represent standard deviation.

I11 were observed to bind 5-15 times more 13H]heparin than peptide Hep-I. In order to determine whether peptides differ- entially bound to plastic surfaces, three peptides were ran- domly selected, they were synthesized with an extra tyrosine residue at their amino end, and were subsequently labeled with 1251. All three peptides were observed to bind to plastic surfaces to a similar extent, when plated at concentrations 5, 50, or 500 pg/ml. Approximately 1-2% of each added peptide bound to plastic at the lower concentrations and 0.1-0.2% bound at the highest concentrations used (data not shown).

The binding of [3H]heparin in solution to peptides Hep-I, Hep-11, and Hep-I11 could be competed by an excess of unla- bation periods. We have-demonstrated that under our exper-

Binding of Heparin to Domains of

A 3000 t 1

Type IV Collagen 2321

B 50000 I 1

nnr n l 1 1 1 0 100 1000 ,001 .01 .1 1 1 0 100 1000 " . Con'kntration of inhibitor (pM) concentration of inhibitor (pM)

C 20000 ,

,001 .01 .1 1 1 0 100 1000 Concentration of inhibitor (pM)

FIG. 8. Competition of the binding of [3H]heparin in solution to peptide Hep-I ( A ) , peptide Hep-I1 ( B ) , and peptide Hep-I11 (C)-coated substrata, by unlabeled heparin (U), dextran sulfate (A-A), and chondroitin sulfate (A-A). Added [3H]heparin was 50 ng. Standard deviation was less than 5%.

8ooo v A B '1 12000

h

E 10000

p 8000

n 6000 1

m 0 1

gg of peptide added pg of peptide added

FIG. 9. A, competition of the binding of [3H]heparin in solution to type IV collagen-coated substrata, by increasing concentrations of peptide Hep-I, (U-U), peptide Hep-I1 (+-e), and peptide 3 (W). 500 ng of [3H] heparin were added per well. Standard deviation was less than 4%. B, competition of the binding of [3H]heparin in solution to type IV collagen-coated substrata, by increasing concentrations of peptide Hep-I11 (a-O), peptide 8 (+-e), and peptide 9 (W). Added t3H]heparin was 500 ng. Standard deviation was less than 5%.

imental conditions, the presence of anti-NC1 IgG totally abolished the binding of heparin to the NC1 domain, both in solid-phase binding assays and rotary shadowing experiments. In contrast, intact type IV collagen incubated with heparin in the presence of anti-NC1 IgG bound a substantial amount of heparin in solid-phase binding assays (approximately 40% of the amount which was bound in the absence of anti-NCl IgG). This amount of bound [3H]heparin should then repre-

sent the binding to the triple helix-rich domain, exclusive of the NC1 domain. Also, in rotary shadowing images, binding of heparin in both sites of the rod-like part of type IV collagen was maintained in the presence of anti-NC1 IgG.

In competition assays, we observed that the binding of [3H] heparin to the main noncollagenous, NC1 domain could be more easily competed by chondroitin/dermatan sulfate than by heparin. Dextran sulfate was a poor competitor. The bind-

2322 Binding of Heparin to Domains of Type IV Collagen

ing of [3H]heparin to the triple helix-rich domain could be competed by unlabeled heparin, whereas neither dextran sul- fate nor chondroitin/dermatan sulfate were efficient compet- itors. These data provide evidence that the main, noncollag- enous domain of type IV collagen should bind both chondroi- tin/dermatan sulfate and heparin, albeit with different affinities. In contrast, the triple helix-rich domain should preferentially bind heparin. We confirmed this specificity by rotary shadowing experiments, in which chondroitin/derma- tan sulfate was demonstrated to specifically bind to the NC1 domain but not to the rod-like part of type IV collagen.

We then asked which contiguous amino acid sequences in type IV collagen were involved in the binding to heparin. To address this question at least in part, peptides were chemically synthesized from the published sequence of the al(1V) and a2(IV) chains. The criteria for selection were based on avail- able information about possible participation of lysinelargi- nine residues in binding to heparin (27, 29, 30), as well as values of hydropathy index which would be compatible with such binding events (27-29).

Another major criterion for selecting peptides from the triple helix-rich domain of al(IV) was the presence of discon- tinuities in the Gly-X-Y sequence, which we reasoned should participate in heparin binding (see above). The synthesized discontinuities represented the vast majority of the interrup- tions which occur in the sequence of the al(IV) chain. From a total of 17 chemically synthesized peptides, peptides Hep-I ( d - N C l ) , Hep-I1 (a2-NC1), and Hep-111 (containing a seven- amino acid-long discontinuity of the triple helix of the al(1V)) were shown to specifically bind heparin and also compete for the binding of [3H]heparin to intact type IV collagen in solid- phase binding experiments. All three peptide sequences con- tained lysine and/or arginine residues. Interestingly, when the end lysine residue was omitted from the synthesis of peptide Hep-111, its ability to bind heparin was maintained, although at somewhat lower levels (data not shown). The presence of positively charged residues in heparin-binding peptides is compatible with the ionic nature of interactions with heparin, as described previously (17, 27-29). Peptide Hep-I1 represents the sequence in the a 2 NC1 chain which is homologous to the sequence of Hep-I in the a1 NC1 chain. It differs from Hep-I by 3 amino acid residues, but only one substitution is conservative. It is possible that the sequences of these two peptides in the intact type IV collagen may function in concert insofar as binding to heparin or related proteoglycans is concerned.

None of the peptides derived from the triple helical se- quences had any heparin-binding activity, although some peptides contained 2 or more lysine residues. It is worth noting that the sequence of the interruption contained in peptide Hep-111, which binds heparin, represents a major site for cleavage by pepsin (1). If this peptide sequence in type IV collagen were involved in heparin binding events, then pepsin treatment of type IV collagen could result in impairment of such binding. This was indicated by experiments in which pepsin-treated type IV collagen failed to bind heparin. This observation that a discontinuity of the triple helix of the al(IV) chain can apparently bind heparin endows this partic- ular sequence (YFDLRLK), with an additional function other than the possible contribution to the flexibility of type IV collagen (36). Importantly, may of these interruptions appear to be conserved in different species ( X ) , pointing to preser- vation of important functions, one of which could be the binding to heparin or heparin-related structures. The site of the heparin-binding interruption is localized at amino acid residues 534-540 from the amino-terminal end of al(1V)

chain of type IV collagen. Provided that the average length of type IV collagen in rotary shadowing images is approximately 400 nm (19) and the total length of the rod-like segment of al(1V) contains 1413 residues (37) (ie. 3.5 amino acids/nm), the interruption in peptide Hep-111 should correspond to a distance -150 nm from the amino terminus. Since we do not know the contribution of other interruptions in locally alter- ing the apparent length of the rod, this site could roughly correspond to the site which is located at 100 nm from the NH2 terminus of type IV collagen, previously identified as a heparin binding site by rotary shadowing. It is possible, how- ever, that other sequences from a2(IV) are also involved in heparin binding at this site, and this possibility needs to be explored when the complete sequence of aZ(IV) becomes available. Furthermore, it is likely that one or more different peptide sequences which we have not examined in the a 1 or a2 (NC1) chains preferentially bind chondroitin/dermatan sulfate. However, it is also possible that post-translational modification(s) or a specific conformation of either peptide Hep-I or peptide Hep-11, which is missing from the synthetic peptides (but should exist in the corresponding sequence in the intact molecule), could be required for the latter binding.

In conclusion, our data indicate that the NC1 domain binds both chondroitin/dermatan sulfate and heparin. The triple helix-rich of type IV collagen preferentially binds heparin. The heparin binding with the highest affinity is localized to the NC1 domain. Discontinuities of the triple helix-rich do- main of at least the al(1V) chain are presumably a major determinant for heparin binding in this domain. Furthermore, we have identified three peptide sequences from the al(IV) and a2(IV) chains which apparently bind heparin in a specific manner. Peptides Hep-I and Hep-I1 should contribute, per- haps in a complementary way, to the binding of at least heparin to the NC1 domain. Peptide Hep-111 could be poten- tially a component of the sequence which binds heparin at a distance of 300 nm from the carboxyl-terminal domain of type IV collagen.

It would be interesting to know whether type IV collagen and the sequences contained in peptides Hep-I, Hep-11, and Hep-111 also bind to chondroitin or heparan sulfate proteogly- can from cell surfaces or basement membranes. Such binding events could potentially bear an effect on the structure or function of basement membranes depending upon the avail- ability of various components. Furthermore, since several lysine residues of basement membrane proteins, including type IV collagen and its NC1 domain, are chemically modified by an excess of glucose (21, 38-41), it would be exciting to know whether the binding of type IV collagen to proteoglycans (12, 15, 16) is altered in diabetes mellitus. This information might help explain, at least in part, the aberrant structure and function(s) of several basement membranes in late stages of the disease.

Acknowledgments-We are indebted to Drs. Jim McCarthy, Amy Skubitz, and Aristidis Charonis for many helpful discussions and suggestions; Ann Vogel and Marcy Krumwiede for skillful technical assistance; Dr. John Alegre for providing us with the hydropathy plots of the a1 (NCl) and or2 (NC1) chains; Dr. James White for providing facilities for the rotary shadowing technique; Drs. Bianca Conti-Tronconi and Bob Wohlhueter who synthesized and purified the peptides; and Carol El-Ghandour for typing the manuscript.

REFERENCES

1. Brazel, D., Oberbaumer, I., Dieringer, H., Babel, W., Glanville, R. W., Deutzmann, R., and Kuhn, K. (1987) Eur. J . Biochem.

2. Babel, W., and Glanville, R. W. (1984) Eur. J. Biochem. 143, 168,529-536

545-556

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20. 21.

22.

Binding of Heparin to Do

Killen, P. D., Burbelo, P., Sakurai, Y., and Yamada, Y. (1988) J. Biol. Chem. 263,8706-8709

Brazel, D., Pollner, R., Oberbaumer, I., and Kuhn, K. (1988) Eur. J. Biochem. 172,35-42

Sieblond, B., Qian, R., Glanville, R. W., Hofmann, H., Deutz- mann, R., and Kuhn, K. (1987) Eur. J. Biochem. 168,569-575

Schwarz, U., Schuppan, D., Oberbaumer, I., Glanville, R., Deutz- mann, R., Timpl, R., and Kuhn, K. (1986) Eur. J. Biochem.

Hofmann, H., Voss, T., Kuhn, K., and Engel, J. (1984) J. Mol.

Kilchherr, E., Hoffmann, H., Steigemann, W., and Engel, J.

Yurchenco, P. D., and Furthmayr, H. (1984) Biochemistry 23 ,

Yurchenco, P. D., and Ruben, G. C. (1987) J. Cell Biol. 105,

Charonis, A. S., Tsilibary, E. C., Yurchenco, P. D., and Furth- mayr, H. (1985) J. Cell Biol. 100, 1848-1853

Laurie, G. W., Bing, J. T., Kleinman, H. K., Hassell, J. R., Aumailley, M., Martin, G. R., and Feldmann, R. J. (1986) J.

Rao, C. N., Margulies, M. K., and Liotta, L. A. (1985) Biochem.

Dziadek, M., Paulsson, M., and Timpl, R. (1985) EMBO J. 4,

Fujiwara, S., Wiedemann, H., Timpl, R., Lustig, A., and Engel, J. (1984) Eur. J. Biochem. 143, 145-157

Timpl, R., and Dziadek, M. (1986) Znt. Rev. Exp. Pathol. 29, 1- 112

Tsilibary, E. C., Koliakos, G . G., Charonis, A. S., Vogel, A. M., Reger, L. A., and Furcht, L. T. (1988) J. Biol. Chem., in press

Kleinmann, H. K., McGarvey, L. M., Liotta, L. A., Gehron- Robey, P., Tryggvason, K., and Martin, G. R. (1982) Biochem-

Tsilibary, E. C., and Charonis, A. S. (1986) J. Cell Biol. 103,

Waddell, W. J. (1956) J. Lab. Znuest. 48 , 311-314 Tsilibary, E. C., Charonis, A. S., Reger, L. A., Wohlhueter, R.

M., and Furcht, L. T. (1988) J. Biol. Chem. 263 , 4302-4308 Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J.

157,49-56

Biol. 172,325-343

(1985) J. Mol. Biol. 186 , 403-415

1839-1850

2559-2568

Mol. Biol. 1 8 9 , 205-216

Biophys. Res. Commun. 128, 45-52

2513-2518

istry 21,6188-6193

2467-2473

#mains of Type IV Collagen 2323

(1951) J. Biol. Chem. 193 , 265-275 23. Barany, G., and Merrifield, R. B. (1980) in The Peptides (Gross,

E., and Meienhofer, J., eds) pp. 1-284, Academic Press, Or- lando, FL

24. Skubitz, A. P. N., McCarthy, J. B., Charonis, A. S., and Furcht, L. T. (1988) J. Biol. Chem. 263 , 4861-4868

25. Corthier, G., Boschetti, E., and Charley-Poulain, J. (1984) J. Zmmunal. Methods 66,75-79

26. Morrison, J. (1980) Methods Enzymol. 7 0 , 214-220 27. Suzuki, S., Oldberg, A., Hayman, E. G., Pierschbacher, M. D.,

28. Kyte, J., and Doolittle, R. F. (1982) J. Mol. Biot. 1 6 7 , 105-132 29. Charonis, A. S., Skubitz, A. P. N., Koliakos, G. G., Reger, L. A.,

Dege, J., Vogel, A. M., Wohlhueter, R., and Furcht, L. T. (1988) J. Cell Biol. 107 , 1253-1260

30. McCarthy, J. B., Chelberg, M. K., Mickelson, D. J., and Furcht, L. T. (1988) Biochemistry 27,1380-1388

31. Oberbaumer, I., Laurent, M., Schwarz, U., Sakurai, Y., Yamada, Y., Vogeli, G., Voss, T., Sieblond, B., Glanville, R. W., and Kuhn, K. (1985) Eur. J. Biochem. 147,217-224

32. Pihlajaniemi, T., Tryggvason, K., Myers, J. C., Kurkinen, M., Lebo, R., Cheung, M.-C., Prockop, D. J., and Boyd, C. D. (1985) J. Biol. Chem. 2 6 0 , 7681-7687

33. Schwarz-Magdolen, U., Oberbaumer, I., and Kuhn, K. (1986)

34. Hostikka, S., Kurkinen, M., and Tryggvason, K. (1987) FEBS Lett. 216 , 281-286

35. Keller, K. M., Keller, J. M., and Kuhn, K. (1986) Biochim. Biophys. Acta 882, 1-5

36. Glanville, R. W., Voss, T., and Kuehn, K. (1982) in New Trends in Basement Membrane Research (Kuehn, K., Schoene, H., and Timpl, R., eds) pp. 69-77, Raven Press, New York

37. Soininen, R., Haka-Risku, T., Prockop, D. J., and Tryggvason,

38. Cohen, M. P., Urdanivia, E., Surma, M., and Ciborowski, C. J.

39. Cohen, M. P., Urdanivia, E., Surma, M., and Wu, V.-Y. (1980)

40. Cohen, M. P., and Wu, V.-Y. (1981) Biochem. Biophys. Res.

41. Triieb, B., Fluchinger, B., and Winterhalter, K. H. (1984) Collagen

and Ruoslahti, E. (1985) EMBO J. 4,2519-2524

FEBS Lett. 208,203-207

K. (1987) FEBS Lett. 225 , 188-194

(1981) Diabetes 30, 367-371

Biochem. Biophys. Res. Commun. 95, 765-769

Commun. 100,1549-1554

Relat. Res. 4,239-251