THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 267, No. 29, Issue of October 15, pp. 21211-21219,1992 Printed in U. S. A .

Identification and Distribution of a Novel, Collagen-binding Protein in the Developing Subepicardium and Endomysium*

(Received for publication, December 9, 1991)

James G. Tidball From the Department of Ph.ysiologica1 Science and Jerry Lewis Neuromuscular Research Center, University of California, Los Angeles, California 90024-1527

A protein doublet (M, = 68,000) that copurifies with chicken cardiac collagen types I and I11 is purified and characterized in the present study. Peptide mapping and amino terminus sequencing for both 68-kDa poly- peptides show they have similar structures. This is supported by amino terminus sequencing of a 39-kDa proteolytic fragment of each polypeptide. The 68-kDa polypeptides appear at PI 6.7-6.8 in two-dimensional gels. Under nonreducing, electrophoretic conditions, the doublet appears as a large multimer or aggregate. Amino acid sequencing of the protein shows that its amino terminus contains a heptapeptide (VCJXXGK) that appears in the heparinlfibrin-binding domain of fibronectin and the collagen-binding domain of lami- nin. Cardiac myocytes synthesize and secrete the pro- tein in vitro onto cell surfaces and onto the substratum. Indirect immunofluorescence shows the protein first appears in the chicken subepicardium at -10 days following fertilization. As collagen accumulates in the subepicardium and the volume of the subepicardial space increases, the 68-kDa protein is found predomi- nantly at the interface between myocardial cells and the connective tissue and between epicardial cells and the connective tissue. In adult hearts, the protein is also present at lower concentrations in endomysia1 connective tissue. The 68-kDa protein is also present in the skeletal muscle endomysium of embryonic chick- ens. Electron microscopic immunocytochemistry shows the 68-kDa protein is located at the surface of subepi- cardial collagen fibers. In addition, a direct interaction between the 68-kDa protein and collagen are indicated by : 1) equilibrium gel filtration of the 68-kDa protein in the presence of gelatin, 2) gelatin affinity chroma- tography of the 68-kDa protein, and 3) comigration of type I collagen and the 68-kDa protein during gel filtration under reducing conditions. The 68-kDa pro- tein exhibits no collagenase activity under native con- ditions or in zymograms. Together, the data indicate that the 68-kDa protein is a novel collagen-associated protein appearing in late epicardial development.

The mechanical performance of the heart is determined in part by the organization and composition of its extracellular

* This work was supported by National Institutes of Health Grant HL-42227. Protein sequencing performed at UCLA Protein Micro- sequencing Facility was aided by a BRS Shared Instrumentation Grant 1 SlORR05554-01 from the National Institutes of Health. 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.

matrix (ECM).’ Pathological and experimental conditions that lead to changes in cardiac ECM composition can result in profound changes in cardiac function (discussed by Factor, 1990). For example, aortic constriction leads to an increase in concentration of ECM molecules in the myocardium (Bar- tosova et d., 1969; Weber et al., 1987; Skosey et d., 1972), that is associated with an increase in blood pressure (Lund et al., 1979). An increase in papillary muscle resting tension occurring during aging (Capasso et al., 1986) can also be attributed to increased ECM concentration (Bing et al., 1978; Eghbali et al., 1989).

Most investigations of the ECM composition of the heart have focused on the identity and distribution of connective tissue proteins in the myocardium and papillary muscles. However, a layer of connective tissue lying at the surface of the heart between myocardium and epicardium is the cardiac site containing the highest concentrations of ECM compo- nents and the most rapid rate of accumulation of collagen types I and I11 from the latter half of embryonic development through adulthood (Tidball, 1992). This site, called the sub- epicardium, is also greatly enriched in fibronectin. These observations suggest that the subepicardium may play an important, yet unexplored role in cardiac mechanics.

The goal of this study was to investigate the identity and distribution of other, putative structural proteins in the sub- epicardium and to provide a better characterization of the biochemical and structural changes that occur during devel- opment in this scantly studied cardiac site. In this study, it is shown that a 68-kDa protein that copurifies with cardiac collagens types I and I11 is a collagen-binding protein secreted by myocytes and is located on the surface of collagen type I fibers in the embryonic subepicardium. This novel protein is most enriched on subepicardial collagen fibers lying near the superficial surface of embryonic myocardial cells or deep surface of epicardial cells, and may provide an indirect link between myocardial cells and the subepicardium.

MATERIALS AND METHODS

Protein Purification-Initial stages of protein purification are based on established purification techniques for collagen and procol- lagen (Miller and Rhodes, 1982; Sage and Bornstein, 1982). Hearts (-10 g) were dissected from adult chickens (White Leghorn) and homogenized in 10 volumes of 50 mM Tris-HC1, pH 7.5, containing 4.5 M NaCl, 20 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 2 mM N-ethylmaleimide, and 1 pg/ml pepstatin (buffer 1). The homog- enate was stirred at 7 “C for 3 h and then centrifuged at 50,000 X g at 4 “C for 1 h. The supernatant was discarded, and the pellet was again homogenized in buffer 1 and then centrifuged for an additional 1 h at 50,000 X g at 4 “C. The supernatant was discarded, and the

The abbreviations used are: ECM, extracellular matrix; SDS- PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; BSA, bovine serum albumin; DPBS, Dulbecco’s phosphate-buffered saline.

21211

21212 68-kDa Collagen-binding Protein in the Subepicardium pellet was resuspended in 20 volumes of 0.5 M acetic acid at pH 2.5 and stirred overnight at 7 'C.

The suspension was then centrifuged at 1,200 X g for 15 min and the pellet discarded. The supernatant was centrifuged at 50,000 X g for 2 h at 4 "C and the pellet discarded. NaCl was then added slowly to the supernatant at 4 'C with stirring to make 2 M NaC1. The solution was allowed to stir for 2 h, and the precipitate was allowed to settle at 7 "C without stirring. The solution was centrifuged at 35,000 X g at 4 "C for 1 h. The supernatant was collected, and additional NaCl was added slowly to make 4.5 M NaCl. That solution was then centrifuged at 35,000 X g for 1 h at 4 "C.

The pellet was collected and resolubilized in 0.1 M sodium acetate a t pH 5.0 (buffer 2) at 50 pg/ml. An aliquot containing 250 pg was applied to a cation-exchange column (Mono S, Pharmacia LKB Biotechnology Inc.) and then eluted with a linear gradient of buffer 2 against increasing concentrations of buffer 2 containing 0.1 M NaCl. Eluted samples were analyzed by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDS-PAGE, Laemmli, 1970). The aliquot eluting in buffer 2 containing 14 mM NaCl was applied to a gel filtration column (Superose 12, Pharmacia) and the peak fraction collected and analyzed by SDS-PAGE. This sample appeared as a 68-kDa band under reducing conditions and 235-kDa under nonreducing conditions. Light loading of reducing, 8% acryl- amide gels show the protein as a doublet at 68 kDa. However, because subsequent findings show the 68-kDa polypeptides to be closely related, they will be referred to collectively as P68 in this study.

Antibody Froduction-Samples of purified P68 protein were emul- sified in Freund's complete adjuvant (50 pg/ml) and injected subcu- taneously in a New Zealand White rabbit following collection of preimmune serum. The rabbit was then boosted with 50 pg/ml of P68 in Freund's incomplete adjuvant 4 weeks following the initial injection and then boosted once during each of the following 3 weeks. Serum was then collected from an ear vein.

Antibody Purification-Antisera were purified according to the technique of Cox et al. (1983). Samples of -200 mg of P68 in SDS- PAGE sample buffer were loaded on a SDS-PAGE gel and electro- phoresed. A band corresponding to -60-75-kDa was cut from the gel and transferred electrophoretically to nitrocellulose, The nitrocellu- lose was washed in buffer 3 containing 0.2% gelatin, 0.05% Tween- 20, and 3% bovine serum albumin (BSA) for 2 h at 4 "C and then cut into 4-mm2 pieces and incubated for 14 h in anti-P68 diluted 1:5 in buffer 3. The nitrocellulose was rinsed in buffer 3. The nitrocellulose was rinsed in buffer 3 several times before placing for 2 min in 4 ml of 0.2 M glycine-HC1 at pH 2.8. The glycine-HC1 was then neutralized with 50 mM Tris, pH 9.0.

Immunoblot Analysis-Samples of chicken heart or chicken hind- limb muscle were homogenized in -10 volumes of 80 mM Tris buffer at pH 6.8 containing 0.1 mM dithiothreitol and 70 mM SDS. Samples were then boiled and centrifuged to remove insoluble material. The samples were analyzed by SDS-PAGE using gels containing 10% acrylamide and 0.13% bisacryamide or 8% acrylamide and 0.13% bisacrylamide. Following electrophoresis, some gels were stained with Coomassie Blue, and others were electrophoretically transferred to nitrocellulose (Burnette, 1981) for 3 h at 1.0 amps. The nitrocellulose sheets were blocked overnight in 50 mM Tris, pH 7.6, containing 150 mM NaCl, 0.1% NaN3, (buffer 3) to which 0.05% Tween-20, 0.2% gelatin, and 3% non-fat dry milk were added. The nitrocellulose sheets (immunoblots) were overlaid for 90 min with anti-P68 serum diluted in buffer 3 containing 0.2% gelatin, 0.5% Tween-20, and 5% inactivated horse serum. The immunoblots were washed in several changes of buffer 3 containing 0.2% gelatin and 0.5% Tween-20 and overlaid with affinity purified, '9-goat anti-rabbit IgG. This second antibody was iodinated by the chloramine-T technique and then diluted to lo4 Bq/ml in buffer 3 containing 0.2% gelatin, 0.05% Tween-20, and 2% horse hemoglobin. The immunoblots were washed overnight in several changes of buffer 3, air dried, and autoradi- ographed.

Immunofluorescence and Immunoelectron Microscopy-Hearts and hindlimb muscles were dissected from embryonic or adult White Leghorn chickens and processed for either immunofluorescence ob- servation or immunoelectron microscopy. Tissues for immunoflu- orescent labeling were frozen in 10.2% polyvinyl alcohol, 4.3% poly- ethylene glycol and sectioned at -20 "C at 12-pm thickness. Sections were placed on microscope slides coated with 0.4% gelatin and 0.04% chromium potassium sulfate and stored at -20 "C. Sections were then labeled for 90 min at room temperature with anti-P68 diluted 1:50 in buffer 3 or with goat anti-collagen type I diluted 1:50 (Southern Biotech, Birmingham, AL) or with preimmune serum diluted 150.

Sections were rinsed with buffer 3 and incubated for 1 h in buffer 3 containing 0.2% gelatin, 1.5% non-fat dry milk, and 0.5% Tween-20. Sections were subsequently covered for 45 min with fluorescein isothiocyanate-conjugated goat anti-rabbit IgG or rabbit anti-goat IgG, as appropriate for primary antibody, before washing with buffer 3 and water. Stained sections mounted in water-miscible media were viewed by epifluorescence.

Hearts to be viewed by electron microscopy were dissected and the epicardium removed with fine forceps to expose the underlying sub- epicardium. Hearts were then immersed in 2% paraformaldehyde in Dulbecco's phosphate-buffered saline (DPBS) for 30 min at 4 "C before rinsing in several changes of DPBS. Hearts were then incu- bated in anti-P68 or preimmune serum diluted 1:lOO in buffer 3 at 6 "C for 12 h and then rinsed in several changes of buffer 3. Hearts were immersed in buffer 3 containing 0.5% Tween-20, and 1.5% BSA for 1 h and then immersed in ferritin-conjugated, goat anti-rabbit IgG diluted 1:200 in buffer 3. After several washes in buffer 3 con- taining 0.5% Tween-20, hearts were fixed in 1.4% glutaraldehyde in 0.2 M sodium cacodylate buffer at pH 7.2. Regions of ventricular myocardium were then cut from the hearts, rinsed in cacodylate buffer, and fixed in 1% osmium tetroxide for 30 min at room temper- ature. The samples were dehydrated in a graded series of ethanols and embedded in epoxy resin for thin sectioning. Sections were stained with uranyl acetate and lead citrate before observation by electron microscopy.

Equilibrium Gel Filtration-A gel filtration column (Superose 12) was equilibrated with 0.1 M sodium phosphate buffer, pH 7.6, con- taining 150 mM NaCl (PBS) to which 0.1% 0-mercaptoethanol and 1 mg/ml gelatin (porcine skin, Sigma) were added. A sample of 100 pg of P68 was then added to 200 pl of column buffer and applied to the column using automated liquid chromatography (fast protein liquid chromatography, Pharmacia). Purified, lyophilized P68 requires boil- ing in reducing buffers to be solubilized. The elution time of the sample was then determined by monitoring the absorbance of the eluate at 280 nm. In alternative experiments, the column was equili- brated in the same column buffer excluding gelatin and the sample solubilized in buffer without gelatin before performing identical gel filtration of the sample. Samples of P68 were also subjected to gel filtration in the absence of @-mercaptoethanol. Control experiments in which BSA replaced P68 for equilibrium gel filtration in the presence or absence of gelatin were also performed.

Gel Filtration of P68 and Type I Collagen-A gel filtration column (Superose 12) was equilibrated with 50 mM sodium phosphate buffer, pH 7.6, containing 150 mM NaC1, 0.1% @-mercaptoethanol, and 0.1% Triton X-100. Samples of P68 (80 pg) or soluble type I collagen (80 pg, rat tail, UBI, Inc.) were applied to the column separately or mixed together and then applied to the column for separation by liquid chromatography. Protein concentration of the eluate was measured continuously at 280 nm, and 1-ml fractions collected. Proteins present in peak fractions were assayed by SDS-PAGE. Elution times for P68 alone, type I collagen alone, and P68 in the presence of collagen type I were determined.

Gelatin Affinity Chromatography-A 0.3-mg sample of P68 in DPBS was applied to a gelatin-Sepharose column equilibrated in DPBS. The column was washed to base line using DPBS and then eluted with DPBS containing 1 M NaCl. Peak fractions were meas- ured for protein concentration using Bradford assay and were ana- lyzed by SDS-PAGE and immunoblotting with anti-P68.

Two-dimensional Gel Electrophoresis-Two-dimensional electro- phoretic separations of P68 were performed by the procedure of O'Farrell(l975). Samples of partially purified P68 obtained following 4.5 M NaCl precipitation were solubilized in sample buffer by repeated freeze-thawing in liquid nitrogen before applying to the first dimen- sion for separation by charge. Partially purified samples were used for two-dimensional analysis because purified samples showed great variability in charge that appeared to be artifacts of preparation.

Amino Acid Analysis and Amino-terminal Sequencing-Samples of P68 or proteolytic fragments of P68 were electrophoresed by SDS- PAGE and then transferred electrophoretically to polyvinylidine difluoride membranes (Immobilon-P, Bedford, MA). Immobilon transfers were stained in Coomassie Blue for 10 min and destained in 10% methanol, 5% acetic acid in water. Protein bands correspond- ing to each P68 polypeptide or their proteolytic fragments were separately cut from the immobilon sheet, air dried, and cysteine residues derivatized by carboxymethylation (Hirs, 1967) or pyridiny- lation (Friedman et al., 1970). Samples were either hydrolyzed for 6 h in 1 N HCl under nitrogen, if used for amino acid analysis (Applied Biosystems, model 420 analyzer), or extracted with 0.1% trifluoroa-

68-kDa Collagen-binding Protein in the Subepicardium 21213

cetic acid if used for amino-terminal sequencing (Applied Biosystems model 470A sequencer).

Peptide Mapping-Samples of P68 to be analyzed by V8 protease peptide fragment mapping were separated by SDS-PAGE so that both P68 polypeptide components could be identified separately. Gels were stained for 5 min in Coomassie Blue, destained 5 min, and then each 68-kDa band cut separately from the gel and inserted in a lane of a second SDS-PAGE gel to which 30 pl of 125 mM Tris, pH 6.8.1 mM EDTA, 0.1% SDS, 20% glycerol, and 25 ng/ml V8 protease were added. The samples and protease were then run into the stacking gel, the current turned off for 1 h, and then electrophoretic separation continued. Some gels were then stained with Coomassie Blue while other, identically prepared samples were prepared for amino-terminal sequencing.

Zymogram Assay-The possibility that P68 is a collagenase was investigated by zymogram assay. A sample of -50 pg of P68 was applied to an SDS-PAGE gel containing 1 mg/ml of gelatin. Another lane of the gel was loaded with 20 pg of bacterial collagenase as a positive control. Following electrophoresis, gels were rinsed with agitation in distilled water containing 2% Triton X-100 for 30 min. They were then incubated overnight a t 37 "C in 50 mM Tris, pH 7.5, containing 5 mM CaC12. Gels were stained with Coomassie Blue for 1 h and then destained.

Collagenase Assay under Native Conditions-The possibility that P68 is a collagenase that is active under native conditions was examined by testing for proteolysis of type I collagen (rat tail tendon, acid soluble, UBI), using bacterial collagenase as a positive control. Aliquots of 0.5 mg/ml of type I collagen in 50 mM Tris, pH 7.5, containing 10 mM CaClZ and 0.1% NaNR were prepared. Either 100 pg of P68, 100 pg of collagenase, or no enzyme was added to each aliquot. An identical group was prepared in which 10 mM EDTA replaced CaC12 in the buffer. Samples were incubated in closed containers for 19 h a t 37 "C. At the end of incubation, samples were centrifuged for 10 min a t 10,000 X g and the supernatants collected. The supernatants were then analyzed by liquid chromatography (Superose 12) with protein concentration in the eluate monitored continuously a t 280 nm. A sample of freshly solubilized type I collagen was also analyzed identically by liquid chromatography for compari- son to experimental samples.

Cell Culture-Hearts were dissected from 14-day-old embryonic, White Leghorn chickens and minced into l-mm3 fragments under sterile conditions. Cells were dissociated by trituration in 0.5% trypsin with EDTA and centrifuged for 5 min in a clinical centrifuge. Pelleted cells were resuspended in 5% embryo extract and 1% L-glutamine in Eagle's minimal essential medium.

Myocytes and fibroblasts were separated by first culturing cells on tissue-grade Petri dishes a t 37 "C, 5% COz for 90 min to allow fibroblast attachment to the dish (Polinger, 1970). Nonattached cells were removed and transferred to a Petri dish coated with 2% gelatin and then incubated overnight a t 37 "C (Puri and Turner, 1978). The myocytes did not attach to the substratum in either of these two plating techniques. The myocytes were then transferred to a Petri dish coated with 2% gelatin and 10% horse serum in Eagle's minimal essential medium. Myocytes attached to this serum-coated substra- tum.

Fibroblasts, attached to tissue grade Petri dishes, and myocytes, attached to serum-coated dishes, were cultured for 7 days a t 37 "C. The supernatants from both cultures were collected and protein precipitated from the supernatants with 80% ammonium sulfate. P68 will precipitate a t ammonium sulfate concentrations less than 80%. The precipitated protein was then boiled in SDS-PAGE sample buffer and used for immunoblots. Following supernatant collection, attached cells were rinsed with Eagle's minimal essential medium and then scraped from their dishes. The cells were boiled in SDS-PAGE sample buffer and used for immunoblots. In other culture dishes, fibroblasts grown on 2% gelatin-coated coverslips or myocytes grown on serum- coated coverslips were prepared for immunofluorescence with anti- P68 as described above.

RESULTS

Established techniques for purifying soluble collagen by acetic acid extraction and NaCl precipitations (Miller and Rhodes, 1982; Sage and Bornstein, 1982) were found in the present study to yield collagen type I, collagen type I11 and a significant proportion of 68-kDa protein (Fig. 1). Further purification of P68 using ion-exchange and gel filtration (Fig.

20c

11t

9t

6E

4:

FIG. 1. Reducing SDS-PAGE separation of cardiac collagen preparation. A, molecular mass standards, indicated in kilodaltons, 200 (myosin heavy chain), 116 @-galactosidase), 95 (glycogen phos- phorylase), 68 (BSA), 43 (ovalbumin). B, 2.0 M NaCl precipitate of chicken heart extract stained with Coomassie Blue. C, 4.5 M NaCl precipitate of chicken heart extract; 68-kDa protein subject to further purification is indicated by arrow. D, autoradiograph of immunoblot of 4.5 M NaCl precipitate incubated with anti-collagen type 111 and radioiodinated second antibody. E, autoradiograph of immunoblot of 4.5 M NaCl precipitate incubated with anti-collagen type I and radioiodinated second antibody.

20(

11 9

6

4

A B C U t F G H I J K L M

FIG. 2. Reducing SDS-PAGE of the following samples sil- ver stained following electrophoresis. A , molecular mass stand- ards, indicated in kilodaltons. B-L, sequential fractions from ion- exchange column. M, peak fraction of P68 from gel filtration of samples obtained from lanes F and G in ion-exchange chromatogra- phy. This sample was used for antibody production.

2) shows that the protein could be resolved into two polypep- tides a t -68-kDa by SDS-PAGE (Fig. 3). These procedures yielded 200-300 pg of purified 68-kDa proteinlg of starting material. The 68-kDa polypeptides display isoelectric points a t PI 6.7 and 6.8 (Fig. 4). Amino acid analysis of this protein shows an absence of hydroxyproline and relatively low glycine levels, indicating that P68 is not a collagen (Table I).

Peptide maps of the P68 polypeptides show identical pat- terns for the two polypeptides following V8 protease digestion (data not shown). A major proteolytic fragment was found at 39 kDa in digests of both polypeptides. Amino-terminal se- quencing shows a high level of similarity between the amino termini of the intact polypeptides (Table 11) as well as the 39- kDa proteolytic fragment (Table 111). Furthermore, nonre-

21214 68-kDa Collagen-binding Protein in the Subepicardium

200

116

95

68

m w

43

A B C D FIG. 3. SDS-PAGE separations on 8% acrylamide gels

stained with Coomassie Blue. A , molecular weight standards. R, appearance of P68 a t -235 kDa in nonreducing SDS-PAGE. C, molecular weight standards. D, resolution of P68 into two polypep- tides under reducing conditions. Upper band is identified as P68 (heavy) and lower band as P68 (light).

"

"

' h

14 FIG. 4. Two-dimensional separation of P68 enriched frac-

tion, showing that the 68-kDa polypeptides are primarily focused at PI = 6.7 (single arrowhead) and PI = 6.8 (double arrowhead), although less prominent, more acidic forms are also present. Silver-stained.

duced samples separated by SDS-PAGE show anti-P68 binds to a broad band found at -235 kDa (Fig. 3). This indicates that P68 polypeptides are bound to one another by disulfides.

Gel filtration of reduced P68 shows that the protein elutes at 16 ml under the conditions used in these experiments, while soluble type I collagen elutes a t 17 ml under identical chro- matographic conditions (Fig. 5). However, chromatography of P68 in the presence of collagen type I shows that an early peak eluting at 12.7 ml contains both P68 and collagen I (Fig. 5). P68 and type I collagen are therefore capable of association under reducing conditions.

Equilibrium gel filtration of P68 in the presence or absence of gelatin under reducing conditions indicates that P68 can also interact with thermally denatured collagen type I (Fig. 6). Control experiments in which BSA replaced P68 in gel

TABLE I P68 amino acid analysis

% molar Residues/molecule

cys 2.8 ASP 8.7 Glu 9.4 Ser 11.5 G ~ Y 12.2

His 1.5 '4% 5.6 Thr 7.2 Ala 7.1 Pro 6.6

TY r 2.0 Val 7.7 Met 1.0 Ile 3.6 Leu 6.8

Phe 2.3 LY s 4.0 HvD 0

17 53 57 70 74

9 34 44 43 40

12 47

6 22 41

14 24 0

TABLE I1 NH,-terminal sequencing of P68 polypeptides and similar domains of

other ECM molecules Eu& BdB?.' P h S i m i l d E I w m

EbsUmhllr€sUlKsM

p68 (ligh!?

fih& 4

fihd" 3

f i b m d n 6 R b m d " 7

' Boxed residues include those that are identical or products of similar codons as P68 (light). Those residues with overlying dots are products of similar, non-identical codons.

P68 (heavy) and P68 (light) refer to relative positions of P68 polypeptides in doublet appearing in SDS-PAGE.

'' Laminin B1 chain from Drosophila mehogas ter (30). Fihronectin from bovine plasma (44). ' Fihronectin from human serum (45).

Fragment of rat fibronectin (46). Fragment of chicken fibronectin (47).

Parenthetical percentage includes residues that are products of * Precursor of human von Willebrand factor (48).

codons with conservative substitutions.

TABLE 111 NH,-terminal sequencing of 39-kDa proteolytic fragments of P68

oolvaeotides

' Boxed residues include those that are identical. 39-kDa (heavy) and 39-kDa (light) refer to the 39-kDa fragments

of the heavy and light chains of the P68 doublet appearing in SDS- PAGE. ' % similarity calculated only for those residues for which definitive

calls were made in analysis. Those residues separated by a stroke (X/ X) represent two amino acids that may be found in that position. Those replaced by a hyphen could not be called with certainty.

filtration experiments showed no change in BSA elution times in the presence or absence of gelatin. Gelatin affinity chro- matography of P68 also indicates that denatured collagen type I and P68 can interact directly, in that following appli- cation of 0.3 mg of P68 to a gelatin affinity column and washing to base line, 0.1 mg of P68 was retained on the

68-kDa Collagen-binding Protein in the Subepicardium 21215

r'

I I

. _ 'i ' f l ?:' . .

column buffer p

' 0 0

. . 4 ~ . 1 , I \

.,i. :' is , ,

. . \ \ ~ ,~. J

~~ . ,-

I. _ - ~ ~i .I " . ~ . ~ L i" -~7- ' 4 :c, ' 3 .'

. . f , u w \ : YO! "'2: 1,- 5 )

FIG. 5. Gel filtration of P68 and type I collagen under re- ducing conditions. Upper panel, P68 alone. Middle panel, acid- soluble collaeen alone. Lower oanel. P68 and tvDe I collaeen mixed in . , ". Y

rior to application to gel filtration column.

I O i ? ! A i 6 18 20 72

t!.UTlON VOLUMt (n ls)

FIG. 6. Elution profiles of P68 applied to gel filtration col- umn under reducing conditions in the absence (solid line) or presence (dotted line) of 1 mg/ml gelatin. P68 consistently elutes in earlier fractions if gelatin were present in the column buffer.

column until eluted with 1 M NaCl (data not shown). P68 was not retained on control columns consisting of agarose only when samples were applied in identical procedures as used for gelatin-agarose affinity chromatography. Furthermore, elec- tron microscopic studies of F68 distribution shows the protein distributed along the surface of collagen type I fibers in the epicardium (Fig. 7). Binding of the ferritin-conjugated second antibody in these preparations appears a t an approximate periodicity of 70 nm, suggesting that P68 binds to a specific site on the collagen molecule. Control preparations showed no collagen labeling (data not shown).

Antibody labeling of embryonic and adult chicken hearts with anti-P68 and anti-collagen type I shows both proteins enriched in the subepicardial connective tissue (Fig. 8). How- ever, collagen type I is found in embryonic myocardia while P68 is restricted to the subepicardium. Within the subepicar-

FIG. 7. Electron micrograph of subepicardial collagen fi- bers labeled with anti-P68 and ferritin-conjugated, second antibody. Some of the ferritin grains, indicating the location of P68, are marked with arrowheads.

dium, P68 is most concentrated at the interface between the myocardium and subepicardium (Fig. 9). In adult chicken hearts, P68 also appears in endomysia1 connective tissue, although the subepicardium is most enriched in P68 (Fig. 9). No labeling was shown on control sections (data not shown). P68 is also present in immunoblots of skeletal muscle (Fig. 10) and appears in antibody-labeled sections concentrated in skeletal muscle perimysium and endomysium, sites also en- riched in type I collagen (Fig. 9).

Cardiac myocytes and fibroblasts in culture synthesize and secrete P68, although the protein does not accumulate in the culture media (data not shown). Cardiac cells retain P68 at their surfaces, as shown by indirect immunofluorescence (Fig. 11). P68 distribution at the cell surface is finely punctate. The protein also appears concentrated at the ends of long processes of cells cultured on gelatin and serum-coated sub- strata, which may reflect a role for P68 in contributing to adhesion to substrata.

Although P68 distribution and binding affinity for collagen resembles other connective tissue proteins, P68 is a unique molecule. Immunoblots show no cross-reactivity between pol- yclonal anti-fibronectin and P68 (Fig. lo), even though com- parisons of amino-terminal sequence data of the P68 proteins with known sequences shows similarity between P68 and part of fibronectin and laminin B1 chain and, to a lesser extent, von Willebrand factor (Table 11). However, an internal se- quence bears no resemblance to any known, sequenced ECM proteins (Table 111).

Collagenases are also known, collagen-binding proteins with masses of -68 kDa. Zymograms used to test the possi- bility that P68 is a gelatinase revealed no proteolysis in the gels although positive control lanes loaded with bacterial collagenases showed collagenolytic activity (data not shown). P68 is therefore not the -68-kDa, collagen-binding collagen- ase previously identified in connective tissues (e.g. Vartio and Baumann, 1989).

Incubation of P68 with type I collagen under native condi- tions also showed no collagenolysis, although bacterial colla- genase produced collagen degradation under identical condi- t,ions. Tiquid chromatography of acid-soluble, type I collagen showed the protein to elute a t 19.8 and 20.5 ml under nonre- ducing conditions used here (Fig. 12). Incubation of type I collagen with bacterial collagenase in the presence of CaZ+ followed by gel filtration showed that in addition to a major peak a t -20 ml, additional major peaks appeared a t 24 ml and 30-34 ml (Fig. 13). Collagen samples treated identically with P68 showed only minute peaks a t -20 ml (Fig. 14), indicating that P68 was bound to collagen that was subsequently centri-

21216 68-kDa Collagen-binding Protein in the Subepicardium

FIG. 8. Surface of 12-day-old chicken embryonic heart. Ar- rowheads indicate subepicardial connective tissue layer. Above arrow- heads is epicardium, below lies myocardium. A and R, Nomarski optics and indirect immunofluorescence images of section labeled with anti-P68. C and D, Nomarski optics and indirect immunofluo- rescence images of section labeled with anti-collagen type I. Arrows indicate myocardial collagen fibers. Bars = 30 pm.

fuged out of the preparation at the end of incubation. Dupli- cate samples in which CaC12 was replaced by EDTA showed major peaks only at -20 ml for both collagenase and P68- treated collagen (Fig. 14). Thus, the association of P68 with collagen appears to be calcium-dependent. Small peaks rep- resentingpolypeptides less than the mass of collagen appeared occasionally in calcium-free preparations of collagen incu- bated in the presence of collagenase, P68, or buffer only. These polypeptides are interpreted to be products of limited proteolysis resulting from processes not mediated by collagen- ase.

DISCUSSION

The mechanical behavior of the heart is believed to be determined largely by the composition and distribution of ECM components. Collagens have been the most extensively studied of the cardiac ECM molecules and comprise 2-3% of

FIG. 9. Nomarski opt ics and indirect immunofluorescence of section labeled with anti-P68. A and B, adult chicken heart. Arrows indicate subepicardium containing most P68; scattered label- ing in myocardium corresponds to endomysium. Bar = 30 pm. C and D, thinner section of 19-day-old embryonic chicken subepicardium. Arrows indicate myocardium-subepicardium interface. Subepicar- dium lies above arrows. Most P68 lies at this interface, although some P68 lies at deep surface of epicardium. Bar = 5 pm. E and F, longitudinal section of 19-day-old embryonic chick skeletal muscle. Arrows indicate endomysium between individual fibers that labels most strongly with anti-P68. Bar = 25 pm.

the mass of the myocardium (Pearlman et al., 1982). This fraction consists of 85% collagen type I, 10% collagen type 111, and 5% collagen type V (Pearlman et al., 1982). All available evidence supports the hypothesis that cardiac stiff- ness, and therefore pressure-volume relationships, vary with the concentration of collagen in the heart. For example, after treatment of excised rabbit hearts with collagenase, trypsin, or elastase, only those hearts treated with collagenase dis- played a reduction in stiffness (O’Brien and Moore, 1966). Other observations supporting a relationship between cardiac stiffness and collagen concentration include those in which aortic constriction resulted in increased collagen concentra- tion and concomitant increase in stiffness (e.g. Skosey et al., 1972).

Although the collagens may therefore be important proteins in determining cardiac mechanics, the structure and compo- sition of the collagenous ECM of the heart has not been extensively investigated. In particular, little is known of the

68-kDa Collagen-binding Protein in the Subepicardium 21217

m

- W

A FIG. 10. Samples separated by reducing SDS-PAGE and

then immunoblotted. A, autoradiograph of immunoblot containing 4.5 M NaCl precipitate of cardiac collagen incubated with anti-P68 and radioiodinated second antibody. E?, immunoblot of adult chicken heart extract labeled with anti-fibronectin. C, immunoblot of adult chicken heart extract labeled with anti-P68. D, immunoblot of adult chicken skeletal muscle extract labeled with anti-P68.

FIG. 11. Embryonic chick myocytes after culture. A , phase microscopic appearance showing elongate shape of myocyte attached and spread on gelatin and serum-coated substratum. B, indirect immunofluorescent appearance of the same cell labeled with anti- P68 showing protein enriched a t process extending from end of cell projecting to the right in the micrograph. Elsewhere on cell, P68 has broadly distributed, punctate appearance. Bar = 8 qm.

". ~

NO COLLAGENASE 1 C O

FIG. 12. Gel filtration of acid-soluble, type I collagen elution profile by gel filtration under nonreducing conditions.

IO0 COLLAGENASE I

COLLAGEN + Co

=I 10 15 20 25 30 35 30

ELUTION VOLUME (mls)

P68 t COLLAGEN -I Ca I

0 4 - : -_I

10 15 20 25 30 35 4 0 ELUTION VOLUME (mls)

FIG. 13. Upper panel, elution profile of type I collagen following incubation with bacterial collagenase in presence of calcium. Gel filtration performed under nonreducing conditions. Lowerpanel, iden- tical concentration of P68 used in place of bacterial collagenase for incubation with type I collagen in the presence of calcium before gel filtration.

composition of the subepicardium, that layer of collagenous connective tissue surrounding the myocardium. The approach taken in the present study was to attempt to identify other major components of the cardiac ECM, especially those that interact with cardiac collagens, by investigating the identity of a 68-kDa protein that copurifies with collagens.

The association of P68 with collagens is supported by several observations. First, P68 copurifies with collagens type I and 111, a trait consistent with either structural similarities or an interaction between the proteins during purification. Second, equilibrium gel filtration of P68 in the presence or absence of gelatin shows that P68 elutes as a larger molecule in the presence of gelatin. Third, P68 is retained on gelatin- agarose columns until eluted with high salt buffers. P68 will also associate with acid-soluble type I collagen, as shown in gel filtration experiments. Finally, P68 is associated with the surface of some collagen type I fibers in the subepicardium, as shown by electron microscope immunolabeling. Thus, P68 has the distribution and behavior of a collagen-binding pro- tein and may contribute to the function of subepicardial collagen.

Indirect immunofluorescence shows that P68 does not in- teract with all embryonic cardiac collagen type I fibers, but

21218 68-kDa Collagen-binding Protein in the Subepicardium 100 - b COLLAGENASE t

COLL.ACEN + EDTA 0 8 0 ~ ~

O D 10 15 20 25 30 35 40

ELUTION VOLUME (mls)

IO0 h P68 + COLLAGEN t EDTA

0 0 10 15 2 0 25 30 35 40

ELUTION VOLUME (mls)

FIG. 14. Upper panel, elution profile of type I collagen following incubation with bacterial collagenase in presence of EDTA. Lower panel, identical concentration of P68 used in place of bacterial colla- genase for incubation with type I collagen in the presence of EDTA before gel filtration.

rather, is located in highest concentrations near the surfaces of cells lying on deep and superficial boundaries of the sub- epicardium. These observations, together with the finding that cardiac myocytes in vitro synthesize P68, indicates that cardiac myocytes near the subepicardium are engaged in syn- thetic and secretory activities not exhibited elsewhere in the embryonic myocardium. P68 is distributed more widely in the connective tissue of adult hearts, but the subepicardium re- mains the site of highest concentration in the adult heart. Although I have been unable to obtain data by in vitro assays to demonstrate that P68 mediates heart cell interactions with collagen type I, the molecules’ distribution supports that possibility. Furthermore, both sequencing data and the loca- tion of P68 at the surface of cardiac myocytes in vitro are also consistent with that interpretation. The possibility that P68 mediates cardiac myocyte-matrix adhesion is the subject of ongoing studies.

Evidence presented on the primary structure of P68 is consistent with both its role as a collagen-binding protein as well as a role in facilitating cell adhesion. Fibronectin and laminin are two well-characterized ECM proteins that support cell adhesion (Klebe, 1974; Graf et al., 1982) and are capable of binding to collagens (Engvall and Ruoslahti, 1977; Rao et al., 1982). The amino terminus of P68 contains a heptapeptide (VCLXXGK) that is similar to domains present in both fibronectin and laminin B1 chain. The similar heptapeptides are located in the heparin/fibrin-binding domain of fibronec- tin (- residue 140) (Hormann and Seidl, 1980; Garcia-Pardo et al., 1983) and the collagen-binding domain (domain IV) of laminin (residues 465-471) (Rao et al., 1982; Sasaki et al., 1982; Graf et al., 1987; Monte11 and Goodman, 1988). Whether this heptapeptide supports cell adhesion or P68 binding to other ECM molecules is under current investigation. A second region for which sequence data are available, the amino ter- minus of the 39-kDa fragment, displays no similarities to other ECM proteins. Thus, these data show that both P68 polypeptides are distinct proteins from other known ECM proteins to which they bear some structural and functional

similarities. Furthermore, the data indicate that there are large similarities in the primary structures of P68 heavy and light chains, although the nonidentical amino acid sequences show that they are distinct gene products.

The data presented show that myocytes accumulate P68 near their surfaces, at least in vitro. This differs from the production of most known ECM molecules in the heart. Early studies of cardiac hypertrophy in rats showed that during periods of increased collagen synthesis, there was a corre- sponding increase in cardiac fibroblasts proliferation (Grove et al., 1969), suggesting fibroblast rather than myocyte in- volvement in collagen production. More recently (Eghbali et al., 1989), in situ hybridization using cDNA for collagens type I and I11 showed the mRNA for these proteins located at interstitial sites where fibroblasts are typically located. Thus, fibroblasts are expected to be the primary, if not exclusive, site of collagen synthesis in the heart. The present study shows, however, that myocytes also contribute to the produc- tion of cardiac ECM. Coincidence of the time of first appear- ance of P68 and collagen type I in the subepicardium at embryonic day 10 (Tidball, 1992) may indicate interaction between myogenic and fibroblastic cells in subepicardial ECM production.

Several previously identified, -68-kDa polypeptides have been shown to bind to collagen, but each is distinct from P68 identified in the present investigation. For example, proteol- ysis of fibronectin with trypsin, cathepsin D, cathepsin G, or chymase yields 70-72-kDa fragments of fibronectin that bind to gelatin-Sepharose (Ruoslahti et al., 1979; Balian et al., 1979; Vartio et al., 1981, 1983; Hedin et al., 1988). P68 differs from fibronectin in most of its amino acid sequence as well as in its distribution. Fibronectin has been shown prominently and widely distributed in embryonic myocardia (Borg et al., 1982; Tidball, 1992), not restricted to the embryonic subepi- cardium as P68. Furthermore, polyclonal anti-sera to P68 do not recognize fibronectin nor does polyclonal anti-fibronectin bind P68. Other, less well-characterized, collagen-binding pro- teins, such as the 65-kDa platelet membrane receptor for collagen type I (Chiang and Kang, 1982) are distinct from P68 in their distribution. Similarly, P68 is distinct from the 67-kDa laminin/elastin receptor both in exhibiting no simi- larities in available sequence data and in distribution on the surface of collagen fibers rather than at the cell membrane (Mecham, 1991).

Collagenases are also -68-kDa, extracellular proteins ca- pable of binding to gelatin-agarose (e.g. Collier et al., 1988; Vater et al., 1978; Vartio and Baumann, 1989). Collagenases further resemble P68 in that they can appear as doublets by SDS-PAGE (Nagase et al., 1981; Stricklin and Hibbs, 1988). However, P68 has been shown to be both structurally and functionally distinct from collagenases in that no similarities in amino acid sequence were identified, no gelatinase activity appeared in zymograms of P68, and no collagenase activity was identified in assays conducted under native conditions. Furthermore, assays for collagenase activity of P68 provide further support for collagen binding activity of P68. Solutions of type I collagen incubated with P68 in the presence of Ca2+ and then centrifuged were depleted of collagen, while control solutions to which no P68 was added, showed substantial collagen remained in solution following centrifugation. How- ever, collagen incubated with P68 and EDTA followed by centrifugation showed that collagen remained in solution at concentrations that resembled control solutions containing collagen only. Thus, P68 may facilitate collagen precipitation through the calcium-dependent formation of aggregates that can be precipitated.

68-kDa Collagen-binding Protein in the Subepicardium 21219

Important questions remain concerning the function of P68. For example, if P68 is involved in indirectly mediating cell attachment to subepicardial collagen type I, what other pro- teins are involved in the association? Does the binding of P68 to collagen type I modify function or mechanical behavior of subepicardial collagen? For example, binding of the proteo- glycan decorin to the surface of collagen fibers may regulate increases in collagen fiber diameter (Vogel et al., 1984). Could P68 serve a role similar to decorin? These questions are under current investigation.

REFERENCES Balian, G., Click, E. M., Crouch, E., Davidson, J. M., and Bornstein, P. (1979)

Bartosova, D., Chvapil, M., Korecky, B., Poupa, O., Rakusan, K., Turek, Z.,

Bing, 0. H. L., Fanburg, B. L., Brooks, W. W., and Matsushita, S. (1978) Circ.

Bonthron. D.. Orr. E. C.. Mitsock. L. M.. Ginsbure. D.. Handin. R. I.. and

J. Biol. Chem. 254,1429-1432

and Vizek, M. (1969) J. Physiol. (Lond.) 200,285-295

Res. 43,632-637

Orkin, S . H: (1986) N k l e i c Acids'Res. 14,7125-7r27 Borg, T. K., Gay, R. E., and Johnson, L. D. (1982) Coll. Eel. Res. 2,211-218 Burnette. W. N. (1981) Anal. Biochem. 112. 195-203 Capasso,d. M., Malhotra, A,, Scheurer, J., and Sonnenblick, E. H. (1986) Circ.

Chiang, T. M., and Kang, A. H. (1982) J. Biol. Chem. 257,7581-7586 Collier, I. E., Wilhelm, S. M., Eisen, A. Z., Marmer, B. L., Grant, G. A., Seltzer,

J. L., Kronberger, A., He, C., Bauer, E. A., and Goldberg, G. I. (1988) J. Biol. Chem. 263,6579-6587

Res. 58,445-460

Cox, J. V., Schenk, E. A,, and Olmstead, J. A. (1983) Cell 35,331-339 Eghbali, M., Blumenfeld, 0. O., Seifter, S., Buttrick, P. M., Leinwand, L. A.,

Robinson, T. F., Zern, M. A., and Giamebrone, M. A. (1989) J. Mol. Cell Cardiol. 21,103-113

Engvall, E., and Ruoslahti, E. (1979) Int. J . Cancer 2 0 , l - 5 Factor, S. M. (1990) in Cardiac Myocyte-Connective Tissue Interactions in

Health and Disease (Robinson, T. F., and Kinne, R. K. H., eds) pp. 130-146, Karger, Base1

Friedman, M., K ~ l l , L. H., and Cavins, J. F. (1970) J. Biol. Chem. 245 , 3868- 3871

Garcia-Pardo. A,. Pearlstein. E.. and Franeione. B. (1983) J. Biol. Chem. 258, 12670-12674 '

A., and Yamada, Y. (1987) Cell 48,989-996

. ,

Graf, J., Iwamoto, Y., Sasaki, M., Martin, G. R., Kleinman, H. K., Robey, F.

Grove, D., Zak, R., Nair, K. G., and Aschenbrenner, V. (1969) Circ. Res. 2 5 , A T - A Q K

Hedin, U., Bottger, B. A., Forsberg, E., Johansson, S., and Thyberg, J. (1988)

Hirs, C. H. W. (1967) Methods Enzymol. 2 , 197-199 Hormann. H.. and Seidl. M. (1980) HorJoe-Sevkr's Z . Phvsiol. Chem. 361.

. ." ="

J. Cell Biol. 107 , 307-319

I . , I . I

1449-1452 '

Klebe, R. J. (1974) Nature 250,248-251 Kornblihtt, A. R., Umezawa, K., Vibe-Pedersen, K., and Baralle, F. E. (1985)

Laemmli, U. K. (1970) Nature 227,680-685 Lund, D. D., Twietmeyer T. A., Schmid, P. G., and Tomanek, R. J. (1979)

Mecham, R. P. (1991) Annu. Reu. Cell Biol. 7.71-91 Miller, E. J., and Rhodes, R. K. (1982) Methods. Enzymol. 82,33-64 Montell, D. J., and Goodman, C. S. (1988) Cell 53,463-473 Nagase, H., Jackson, R. C., Brinckerhoff, C. E., Vater, C. A,, and Harris, E. D.

Norton, P. A., and Hynes, R. 0. (1987) Mol. Cell. Biol. 7,4297-4307 OBrien, L. J., and Moore, C. M. (1966) Experientia 22,845-849 OFarrell, P. H. (1975) J. Biol. Chem. 250,4007-4021 Pearlman, E. S., Weber K. T., Pietra, G., Fisman, A. P., and Janicki, J. S.

Polin er, J S (1970) Ex Cell Res. 6 3 , 78-82 Puri,%. C.;and Turner,%. C. (1978) Exp. Cell Res. 1 1 5 , 159-173 Rao, C. N., Margulies, I. M. K., Tralka, T. S., Terranova, V. P., Madri, J. A,,

Ruoslahtl, E., Hayman, E. G., Kuusela, P., Shively, J. E., and Engvall, E. (1979)

Sage, H., and Bornstein, P. (1982) Methods Enzymol. 82,96-127 Sasaki, M., Kato, S., Kohno, K., Martin, G. M., and Yamada, Y. (1987) Proc.

Schwarzbauer, J. E., Tamkun, J. W., Lemischka, I. R., and Hynes, R. 0. (1983)

Skorstengaard, K., Jensen, M. S., Sahl, P., Petersen, T. E., and Magnusson, S.

Skosey, J. L., Zak, R., Martin, A. F., Aschenbrenner, V., and Rabinowitz, M.

Stricklin, G. P., and Hibbs, M. S. (1988) in Collugen (Nimni, M. E., ed) Vol. 1,

Tid! Vartio, T., and Baumann, M. (1989) FEES Lett. 255,285-289

all, J. G. (1992) Anat. Embryol. 185 , 155-162

Vartio, T., Sep a H and Vaheri, A. (1981) J. Biol. Chem. 266,471-477 Vartio, T., BarTati, g., DePetro, G., Miggiano, V., Stahli, C., Takacs, B., and

Vater, C. A., Mainardi, C. L., and Harris, E. D. (1978) Biochem. Biophys. Acta

Vogel, K. G., Paulson, M., and Heinegard, D. (1984) Biochem. J. 223,587-597 Weber, K. T., Janicki, J. S., Pick, R., Abraham, C., Schroff, S. G., Bashay, R.

EMBOJ. 4 , 1755-1759

Cardiovasc. Res. 13,39-44

(1981) J. Biol. Chem. 256,11951-11956

(1982) Lab. Inuest. 4 6 , 158-166

and Liotta, L. A. (1982) J. Biol. Chem. 257,9740-9744

J. Biol. Chem. 254,6054-6059

Natl. Acad. Sci. U. S. A. 84,935-939

Cell 35,421-431

(1986) Eur. J. Biochem. 161,441-453

(1972) Circ. Res. 31,145-157

p . 187-205, CRC Press, Boca Raton, FL

Vaheri, A. (1983) Eur. J. Biochem. 1 3 5 , 203-207

539,238-246

I., and Chen, R. M. (1987) Circ. 75 , (suppl. 1) 140-147