PROTEIN

EXPRESSION

AND

PURIFICATION

2,304-312

(19%)

Isolation and Characterization of Dog Hearts’

of Insoluble Collagen

Shizuko Takahashi,2 Mengjia Zhao, and Calvin Eng Department of Medicine, Division of Cardiology, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, New York, New York 10461 Received August 2, 1991, and in revised form September

13, 1991

A procedure for isolating insoluble heart collagen has been developed. The method involves the use of defined optimal conditions of sonication that yield no thermal denaturation of the triple-helical structure nor disruption of the primary structure of the collagen molecules; this is followed by extraction of isolates with nondenaturing agents. The amino acid residues of the isolates are then reacted with dansyl chloride to allow determination of amino-terminal residues and quantification of the collagen. The method has several advantages over existing procedures: (i) There is no other method available for isolation of undenatured insoluble heart collagen in almost pure form (consists of 9 6 % of type I collagen) and in a good yield. Sonication of tissue at or below 4°C for a total of 120 s (15 8 sonication repeated 8 times at 120-s intervals) yielded insoluble collagen fibers with 9 0 % yield and a 20-fold purification as determined by the increase in Hyp content of the isolates. Extraction of these isolates with 0.6 M KC1 and 1 M NaCl at 4°C resulted in a 22-fold purification with 7 0 % yield, while the classical extraction method with nondenaturing reagents yielded only S-fold purification. (ii) There has been little study of the derivatization of an insoluble protein (collagen) with dansyl chloride. The Lys residues of collagen could be recovered as c-Dns-Lys in 6 4 % yield from a reverse-phase C-18 column by high-performance liquid chromatography. This assay allows measurement of 0.1-100 nmol t-Dns-Lys. (iii) The method generates direct information concerning the quantity of collagen and its nature with respect to amino groups. 0 1991 Academic Press, Inc.

The collagens represent major structural proteins of all animal tissues. In tissues of developing animals, the ’ This work was supported by National Institutes of Health Research Grants IPOI AG 05554, HL 37412, and HL 27219, a Grant-inAid from the American Heart Association-Winthrop Pharmaceuticals 901347, a Grant-in-Aid and an Established Fellowship from the New York Heart Association, and a grant from the Squibb Research Laboratories. * TO whom correspondence should be addressed. 304

collagen molecules are assembled into fibrils and may undergo intra- and intermolecular crosslinking to form highly insoluble structures assembled into orderly aggregates in the extracellular space. Ultimately, that process gives rise to fibers. The length and thickness of the assembly of collagen molecules are determined by the type of collagen involved, by interactions with other components of the extracellular matrix, and by the cells in the tissue. Thus, the kinds of fibers formed are related to their function. Large cable-like bundles are formed by type I collagen, while fine fibrillar-like structures are formed by type III and type V collagens, which usually occur associated with type I collagen (1). Thus, tendon collagens consist of type I collagen, while the heart contains type I (dominant type) (2-4), III (3,5), V (3), IV (3), and VI (6) collagens. In many cases, the fibers in these structures are insoluble; that is, they are difficult to extract or purify without the use of degradative methods. At present, no method for isolating undegraded or undenatured insoluble collagen of the heart in pure form has been reported. The present study was undertaken to develop a reproducible combination of procedures for isolating undenatured insoluble collagen3 from the left ventricle of dog 3 Terms used: insoluble collagen refers to the collagens that are not extractable in either nondenaturing agents (0.5 M acetic acid, 0.1-l M NaCl, at 4°C) or denaturing agents (8 M urea or 6 M guanidine-HCl, at 4”C, or 1% SDS, at 25°C). Most insoluble collagens are solubilized by gelatinization (denaturation) by autoclaving. This insolubility of the collagen in fibers is due to the occurrence of both intra- and intermolecular crosslinks between the collagen chains (1). Soluble collagen refers to the collagens that are extractable in acetic acid (5-500 mM, at 4°C) and remain soluble in neutral buffers in the presence of 0.2 M NaCl at 4°C. This soluhility is due to the occurrence of fewer crosslinks between collagen chains than occur in insoluble collagens. Undenatured collagen refers to the collagen that exhibits the properties of native collagen, including occurrence of the polyproline-like helices of a chains and the triple-helical coiled-coil structure of collagen molecules. The triple-helical structure restricts proteolytic attack by general proteinases. Both N- and C-termini of collagen molecules lack the triple-helical structures and therefore are susceptible to attack by general proteinases. Denatured collagen refers to the collagen that has lost the properties of native collagen, such as loss of polypro1046-5928/91$3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.

ISOLATION

AND

CHARACTERIZATION

hearts. The methods include the following procedures: (i) Sonication under defined optimal conditions that causes little or no thermal denaturation of the triple-helical structure of collagen or shearing of the primary structure of collagen molecules was achieved by studying two collagens as models, acid-soluble guinea pig skin type I collagen and insoluble beef Achilles tendon collagen (AT).4 (ii) The derivatization of insoluble collagen with Dns-Cl4 was performed to modify the Lys and Hyl residues and amino-terminal residues, as Dns-Cl is a widely used fluorescent reagent for determination of amino acids and for amino-terminal residue analysis in soluble peptides and proteins (7-10). However, there has been little study of the derivatization of insoluble proteins with Dns-Cl. The insoluble collagens have been dansylated by us and the amino-terminal residues appearing at the cleaved sites of collagen molecules by sonication determined. We first established that the relevant dansylated amino acids are separated sufficiently from the products (Dns-OH and E-Dns-Lys) in HPLC and therefore can be analyzed in the acid hydrolysates. These treatments directly generate information concerning the quality and quantity of collagens and also identify the amino-terminal amino acid residues derived from the degradation or contamination of other molecules if these are present in the isolates. (iii) The “native” (or triple-helical) structure in the insoluble heart collagen isolated by sonication was studied by two procedures: One was by cross-reactivity between the antibodies directed against the epitopes in the triple-helical structure of collagen molecules, and epitopes in pepsin-solubilized sonicated collagens. The other was by determining the accessibility of the sonicated collagen to proteolysis by general proteinases. Because the native form of collagen is known to be more resistant than the denatured collagen to such proteolysis, these results could indicate whether native structures are preserved during the preparative treatment. (iv) Other biochemical characterization of insoluble heart collagen is achieved by determination of solubility in various reagents, stability to heat, amino acid composition, identification of types of collagens, and patterns of CNBr peptides on SDS-PAGE. MATERIALS

AND METHODS

Materids. Reagents were obtained as follows: Naacetyl-Lys (Na-AC-Lys) was from Sigma Chemical Co. (St. Louis, MO), and the octadecyl C-18 column (15.0 X line-like and triple-helical structures; this can be accomplished by application of heat or denaturing agents or the shearing of the polypeptide structure by chemical agents or proteolytic attack. ’ Abbreviations used: AT, Achilles tendon; C-collagenase, Chtridium histolyticum collagenase; T-collagenase, tissue collagenase; Dns or dansyl, 5-dimethylaminonaphthalene-1-sulfonyl; ELISA, indirect inhibition enzyme-linked immunosorbent assay; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis.

OF HEART

COLLAGEN

305

4.6 mm i.d.) from Supelco Inc. (Supelco Park, Bellefonte, PA). AT collagen was a gift from Dr. Sam Seifter, Department of Biochemistry, Albert Einstein College of Medicine. Purified C-collagenase (ll), T-collagenases (12,13), native acid-soluble guinea pig skin type I collagen (14), and rabbit anti-mouse skin type I collagen antiserum (4) were prepared in this laboratory. Other reagents were the same as those described in previous publications (4,11-14). Water was double-distilled and deionized, with or without Millipore filtering (0.45 pm membrane). Sonication. Sonication was performed with a Branson sonicator (Model W-220 F, Heat Systems-Ultrasonic Inc.) at output control 6. The microtip (a metal rod vibrator, 0.3-mm diameter) was centered in a tube (g-mm diameter) containing 1.5 ml of solution at 0°C and the tube was kept in ice. To maintain a constant temperature for sonication below 4°C (to minimize thermal denaturation), the sonicate temperature was measured immediately after sonication and at several subsequent intervals. Tris-Triton (10 mM Tris-HCl/ 2% Triton X-100, pH 7.5) buffer was used in three groups of experiments. In group A, a single sonication was performed for 15 or 30 s with a 45-s recovery. A single sonication of 15 s increased the temperature to 4°C from 0°C but the temperature returned to 3°C after 45 s. The sonicate temperature after 30 s of sonication was 5.5’C, and this returned to 5°C. In group B, a single sonication was performed for 15 s with a 120-s recovery. The temperature increased to 4°C and returned to 2°C after 120 s. In group C, repetitive sonication was performed for a total of 120 s (15 s, 8 times) in two subgroups of experiments: In subgroup a, a 45-s recovery was allowed and in subgroup b, a 120-s recovery. The temperature of subgroup a was increased to 5.5”C from 3°C after a 45-s recovery and that of subgroup b was increased to 4°C from 2’C after a 120-s recovery. On the basis of the preceding studies, the conditions for subgroup b were employed thereafter. Dansylution of AT collagen. Our initial studies (Table 1) indicated that the derivatization of AT collagen with Dns-Cl, determined by measurement of fluorescence, was maximal when dansylation was conducted as follows: Aliquots of AT collagen (1 mg) were placed in Eppendorf tubes and suspended in 0.8 ml of saturated NaHCO,; then 0.2 ml of 1% Dns-Cl in acetone was added to each tube at pH 8.5. The mixture was reacted for 24 h at 25°C in the dark. The dansylated collagen was collected by centrifugation at 4000g for 5 min at 25°C and transferred into a glass tube; NaHCO, and excess Dns-Cl were removed by washing five times successively with 5 ml each of water, acetonitrile, and acetone until the wash solution showed no detectable fluorescence. The material was then freeze-dried, hydrolyzed with 6 N HCl for 18 h at 105°C to allow measurement of fluorescence (15), and maintained for

306

TAKAHASHI,

ZHAO,

24 h at 110°C to allow measurement of Hyp (16). Undansylated collagen was used as a control. The hydrolysates were neutralized. Fluorescence was measured with excitation at 340 nm and emission at 530 nm and was expressed as excitation at 340 nm/pmol Hyp. The dansyl derivatives were analyzed by using HPLC on a reverse-phase octadecyl C-18 column as described (13,16), except for a minor modification. In brief, the column was eluted with a linear gradient of 67% solvent A (10 mM Tris-HChmethanol, 80:20, pH 7.75) and 33% solvent B (10 mM Tris-HCl:methanol, 10:90, pH 7.75), at 25°C for 45 min and then with 100% of solvent B. The amount of t-Dns-Lys was estimated from a standard curve constructed as noted below (13). To determine whether a linear relationship exists between the amount of e-Dns-Lys and the peak areas of the HPLC separation, varying amounts of standard tDns-Lys were chromatographed and analyzed. To determine the efficiency of dansylation and a linear relationship between the amounts of c-Dns-Lys of NWAcLys and the peak areas of HPLC separation, standards of Na-AC-Lys were derivatized with Dns-Cl in two groups: One was analyzed on HPLC and the measured peak area was compared with that of the standard EDns-Lys. The other group was analyzed on an amino acid analyzer to measure residual undansylated Na-AcLys. A large amount of NWAC-Lys (0.01-0.4 pmol) was used for the latter experiment. To determine the efficiency of dansylation and a linear relationship between the amount of c-Dns-Lys of AT collagen and the peak areas of the HPLC separation, varying amounts of AT collagen were derivatized with Dns-Cl in two groups: One group was subsequently hydrolyzed (6 N HCl, 105”C, 18 h) and analyzed by HPLC. The amounts of c-Dns-Lys were determined from the measured peak areas and referred to a standard curve; contents were expressed as nanomoles of t-Dns-Lys per micromole of Hyp or per milligram of collagen. The other group of samples was hydrolyzed for 24 h at 110°C and the undansylated Lys remaining after the dansylation procedure was determined either by HPLC or by amino acid analyzer. A large amount of collagen (10 mg) was used when measurement was made with the analyzer. Preparation and assay of sonicated acid-soluble collugen. The effect of sonication on the triple-helical structure of acid-soluble collagen was studied by measurements made using Ostwald viscometers with flow times for water at 25°C between 25 and 75 s, as described (12). Samples in two groups were sonicated: one was without heat treatment and the other was denatured by heating at 60°C for 20 min and chilling immediately in ice; controls were not sonicated. The viscosity of the sonicated or the heated sample was expressed as the percentage decrease in the initial specific viscosity from that of the untreated control. Aliquots (5-10 pg) of all samples were subjected to SDS-PAGE for examination of possible degradation of collagen.

AND

ENG

Preparation and assay of sonicated AT collagen. TO determine whether sonication causes a scission of peptide bonds in the insoluble collagen, we used the dansylation method of Gray (15) for amino-terminal amino acid residues. The AT collagen (2 mg/ml) was suspended in Tris-Triton buffer containing protease inhibitors (2 mM phenylmethanesulfonyl fluoride110 mM Nethylmaleimide/lO mM EDTA/l% aprotonin) and 0.02% NaN, (anti-bacterial agent) and sonicated. The fiber pellet adhering to the microtip was resuspended in fresh buffer for each sonication; the sonicates were saved for analysis. At the end of sonication, the fiber pellet was rinsed in water with an additional 2-s sonication, until the wash solution showed no absorbance at A 280nm; the material was then freeze-dried. Controls were without sonication. The collagens were then dansylated and analyzed as above. Preparation of dog heart tissue. The procedures used have been described (17). In brief, three normal male mongrel dogs (23 + 2 kg) were anesthetized. After cardiac arrest, the coronary arteries were perfused with phosphate-buffered saline and the heart was removed. The left ventricle was separated and the large vessels and epicardial and endocardial membranes were removed, the remaining tissue was then used for isolation of collagen. Isolation of collagen. The heart tissue was placed on ice and separated into thin pieces using forceps. The tissue was sonicated as above with modifications. The tissue (-0.1 g/3 ml) was sonicated first in a 14-mm-i.d. tube and then in a 9-mm-i.d. tube. The tissue was sonicated for a total of 5-150 s and the fiber pellet was collected. The fiber pellet remaining after 120 s of sonication was extracted for 24 h at 4’C successively with a lOOO-fold volume (w/v) of 0.6 M KC1 (3 times) and 1 M NaCl (3 times) containing 0.02% NaN,. The collagen fiber was collected by centrifugation at 27,000g for 20 min at 4”C, washed with water, freeze-dried, and stored at -70°C. For the preparation of control heart collagen, the tissue was extracted at 4°C consecutively with TrisTriton buffer, pH 7.5, containing the protease inhibitors and 0.02% NaN,, and then with 0.6 M KC1 and 1 M NaCl as above; the final residue was washed with water and freeze-dried. Another sample of the tissue was also extracted at 25°C for 24 h with 1% SDS in 10 mM TrisHCl buffer, pH 7.5, containing the protease inhibitors, and 2.5% 2-mercaptoethanol and 0.02% NaN,. The insoluble collagen was collected by centrifugation; SDS was removed by successive washings with water and freeze-dried. The respective preparations were designated sonicated collagen, control collagen, and SDScollagen. Morphological fiber. Scanning

examination

of

isolated

collagen

electron microscopy of the isolated collagen was performed as previously described (17). Extraction and solubilization of sonicated heart collugen. Extractions of the sonicated collagen with

ISOLATION

AND

CHARACTERIZATION

various reagents (0.5 M acetic acid, 8 M urea, 4°C and 1% SDS, 25°C) were performed as previously described (18). The extracts were collected by centrifugation, dialyzed, freeze-dried, and analyzed for contents of Hyp and proteoglycans. The collagen fiber (1% suspension in water) was solubilized by autoclaving at 121°C and 15 lb pressure for 20-120 min; it was then centrifuged at 27,000 g for 20 min. The Hyp contents of the supernatant and pellet were determined. Collagen fibers were solubilized also by limited pepsin digestion at 4°C for 24 to 96 h, and type I and type III collagens were separated by the method of Miller (19) using differential salt precipitation by NaCl. In this procedure, enzyme activity was terminated by addition of pepstatin (1 mg/ml) and the mixture centrifuged; the resulting supernatant was analyzed. Collagen fibers were also solubilized by digestion with CNBr by the method of Laurent et al. (20). After incubation at 25°C for 24 h, the reaction mixture was centrifuged, and the resulting supernatant was freeze-dried and analyzed. Degradation of sonicated heart collagen by proteinases. The susceptibility of sonicated heart collagen to proteolysis by general proteinases such as trypsin and chymotrypsin or by specific enzymes such as C-collagenase and T-collagenase was studied as previously described (4,11,12). In brief, the sonicated collagen (0.2% suspension in assay buffer) was used in two groups: one was unheated and other was heat-denatured (6O”C, 20 min). The samples were reacted with enzyme at an enzyme-to-substrate ratio of 1:50 (w/w) for 24 h at 37°C. At the end of the reaction, inhibitor (10 mM EDTA for T-collagenase, 10 mM o-phenanthroline for C-collagenase, 5 mM diisopropylphosphorofluoridate for serine proteinases, or 20 mM EDTA and 5 InM diisopropylphosphorofluoridate for Pronase E) was added and the samples were centrifuged, the resulting supernatants were freeze-dried and the Hyp contents analyzed. SDS-PAGE and immunoreactivity of sonicated heart collagen. Pepsin-solubilized or CNBr-digested collagens were subjected to SDS-PAGE (21), and protein and peptide bands were visualized by Coomassie brilliant blue staining and/or Western immunoblot staining (22). The latter was performed using an antiserum raised in a rabbit against undenatured acid-soluble mouse skin type I collagen. The antiserum used reacts with epitopes in antigen mouse type I collagens; about 70% of the reactivity is dependent on the conformational structure of undenatured collagen molecules and about 30% on epitopes in the primary structure of collagen (4). The presence of the native triple-helical structure of type I collagen in sonicated heart collagen was determined by cross-reactivity of the mouse antibodies as above to the sonicated collagen that had been pepsinsolubilized in two groups; one was unheated and the other was heat-denatured (6O”C, 20 min). The crossreactivity was determined by direct ELISA as previously described (4).

OF HEART

COLLAGEN

307

Other methods. The total collagen content of the sample was determined by measuring Hyp content, using both HPLC and an amino acid analyzer. The collagen content was expressed as mg of collagen/100 mg of sample protein, based on the assumption that the average Hyp content of collagens is 10% (1 pmol Hyp = 1 mg of collagen) and that for any protein, on the average 1 mg of protein yields 10 pmol of Leu equivalents in the ninhydrin reaction (4). The total amino acid contents of the sample were determined by using an automatic amino acid analyzer or by another ninhydrin reaction using Leu as standard (4). The proteoglycan content was determined by calorimetric methods using sodium glucuronate and proteoglycan as standards (23). The data were analyzed statistically, and Student’s t test was used to determine significance (P values of < 0.05). RESULTS AND DISCUSSION Dansylution of AT collagen. The dansyl derivatives eluting from the C-18 column at times close to that for t-Dns-Lys (46 min) were Dns-Val (40 min), Dns-NH, (42 min), Dns-Ile (49 min), and Dns-Leu (50 min). All of these were well-separated from t-Dns-Lys. The linearity of the standard curve for c-Dns-Lys in a working range was 0.1-10 nmol (Fig. la). The efficiency of dansylation of Na-AC-Lys and that of AT collagen were 87 f 5 and 84 f 5%, respectively, determined by the recovery of f-Dns-Lys and/or by the amounts of undansylated No-AC-Lys or Lys remaining after the dansylation procedure; in both cases 11 f 0.8% of Lys remained undansylated. The linearity of c-Dns-Lys ‘obtained from Ncu-AC-Lys and AT collagen isshown in Figs. lb and lc. The reliable lower limit of the method is 0.1 nmol (14.6 ng), which corresponds to about 385 ng of AT collagen. The recovery of t-Dns-Lys from the dansylated AT collagen was 0.20 f 0.02 pmol/mg collagen, which was about 83 + 5% of the Lys values found in undansylated controls (0.23 f 0.02 pmol/mg collagen); the recovery of internal standard in hydrolysates was 85 f 5%. Similar results were obtained with the dansylated soluble collagen used as a model control (data not shown). The lack of dansylated amino-terminal amino acid residues in the AT collagen suggests that the amino-terminal residues may be 2-pyrrolidone-5-carboxylic acid as reported for soluble type I collagens from rat tendon (24), chicken skin (25), and human skin (26). These results indicate that insoluble collagen was dansylated at its Lys residues to about the same degree as soluble collagen or other proteins (10). No e-Dns-Hyl was detected. Whether this is due to the interference of reaction between the E-amino group to Dns-Cl by the adjacent OH group of Hyl is unknown. Effect of sonication on acid-soluble collagen. Nishihara and Doty (27) reported that the sonication of acidsoluble calf skin collagen for 15 to 210 min below 8°C markedly decreased its viscosity, indicating that dena-

308

TAKAHASHI,

ZHAO.

AND

ENG

a

&-Dns-Lys (nmol)( A) 0

2

4

6

8

b

120

10

-0

-A %

-cY

F

.9

g 5

~1~~

60

I

I

t

0.0

0.2

0.4

0.6

0

0.8

1.0

E

(pmol)

5 =: 2.0

2 I

1P

2

4

10

A

20

1234

FIG. 2. Effects of sonication on viscosity and SDS-PAGE of acidsoluble collagen solution before and after denaturation by heating at 60°C for 20 min. (a) Percentage decrease of the initial specific viscosity (2.0) at 25°C of acid-soluble guinea pig skin type I collagen (0.05%, pH 7.5) is plotted against time. Open symbols, undenatured collagen; and closed symbols, heat-denatured collagen; squares, unsonicated controls; and triangles, sonicated collagens. Each data point represents the mean of two separate experiments with duplicate samples. (b) The SDS-PAGE (7.5% gel) of the samples used was that in (a) above. Protein bands were stained with Coomassie brilliant blue. Lanes 1,2 (0 h); 3,4 (16 h); 1,3 (control); 2,4 (sonicated). Assays are described in the text.

-

0

0

.

Time(h)

&-Dns-Lys (nmol)( A)

Na-AC-Lys

I

6

8

10

AT Collagen (mg) FIG. 1. (a) Linear relationship between the amount of t-Dns-Lys and the peak areas in the HPLC separation. Varying amounts of standard t-Dns-Lys are plotted against peak areas. Ordinate: unit of area. Abscissa: upper scale applied in a range of 1.0-10.0 nmol (open symbols), and lower scale applied in a range of 0.1-1.0 nmol (solid symbols); bars, SE. (b) Recovery of e-Dns-Lys after modification with Dns-Cl. Recovery of c-Dns-Lys after derivatization of varying amounts of standard NWAC-Lys with Dns-Cl was plotted against standard e-Dns-Lys. Unmodified Lys remaining in individual reaction mixtures was plotted against standard Lys. For this analysis an amino acid analyzer was used. Open symbols, e-Dns-Lys; and solid symbols, unmodified Lys. Data points represent the means of two separate experiments with duplicate samples; bars, SE. (c) Recovery of c-Dns-Lys after modification with Dns-Cl. Recovery of t-Dns-Lys after derivatization of varying amounts of AT collagen with Dns-Cl was plotted against standard t-Dns-Lys. Unmodified Lys remaining in individual hydrolysates is plotted against standard Lys. Open symbols, c-Dns-Lys; and closed symbols, unmodified Lys. Data points represent the means of two separate experiments with duplicate sam. . Ipies; bars, Sr;.

turation had occurred. However, sonication for only 2 to 3 min below 4°C was used in the present study. The viscosities of acid-soluble type I collagen of sonicated and controls were very similar (Fig. 2a), and decreased as expected on heat denaturation. The protein bands on SDS-PAGE of all collagens were also very similar and showed no cleaved fragments caused by sonication (Fig. 2b). These results indicate that the conditions of sonication outlined did not cause obvious denaturation or peptide cleavage of acid-soluble collagen molecules. Effect of sonication on AT collagen. One cleavage of the collagen chain would produce one Dns-amino acid residue over that of the control. Based on this assumption, results obtained from the AT collagen under various conditions were as follows. (i) The fluorescent intensity/pm01 Hyp values of the collagen sonicated for 15,120, and 150 s and dansylated at optimal conditions were respectively 1660 + 45, 1660 + 47, and 1660 f 59. These values were similar to those of the controls (Table 1). (ii) The chromatograms of hydrolysates of sonicated and control collagens were also similar and showed no Dns-amino acid or Dns-NH,, only the byproducts (Dns-OH and c-Dns-Lys) (Figs. 3a and 3b). The recoveries of c-Dns-Lys of sonicated and control collagens were similarly 0.2 * 0.02 pmol/mg collagen. Thus, there appear to be no detectable cleavages of collagen chains caused by sonication. Isolation of heart colhgen. Sonication of heart tissue for 15 s resulted in the formation of two fractions: one was a collagenous pellet which adhered to the microtip and contained about 95% of the collagen of the initial tissue, and the other was a homogenate containing over

ISOLATION TABLE

Optimization

of

Conditions

Sample Control

Time 04

25 25 25 37

12 24 48 24

for Dansylation

0.1% (10%) 661+40 N.D. N.D. N.D.

CHARACTERIZATION

1

Concentration Temp 03

AND

of AT

Collagen

of dansyl chloride

0.2% (20%) 1560 1662 1660 1672

+ + f +

37 53 47 46

0.6% (60%) 1660 k 53 1664 f 45 N.D. N.D.

OF HEART

309

COLLAGEN

Biochemical properties of sonicated heart collugen. The recovery of e-Dns-Lys from the sonicated heart collagen was 0.26 + 0.02 pmol/mg collagen (the recovery of internal standard was 90 + 5%), which was about 79 f 6% of the Lys values found in undansylated collagen (Table 3). The collagen also lacked new amino-terminal amino acid residues, suggesting that no detectable cleavage of collagen occurred during sonication. It also lacked Dns-

Note. The optimal conditions for dansylation of AT collagen were determined by measuring the fluorescence (excitation at 340 nm/ pmol Hyp) of dansylated collagen (1 mg) under various conditions; the concentration of acetone in the reaction mixture is shown in parentheses. N.D., not determined. Each data point represents the mean f SE of two separate preparations, with analysis in duplicates. The recovery of internal standard in each hydrolysate was 85 + 5%. Assay conditions are described in the text. The selected optimal conditions for dansylation of collagen therefore were 0.2% of dansyl chloride at 25°C for 24 h.

90% of the protein of the initial tissue (Table 2). Apparently, the sonication caused most of the noncollagenous proteins of the heart tissue to become liberated into a soluble and/or suspension phase, leaving most of the collagen as an insoluble network. Scanning electron microscopy of the collagen fibers showed the extensive collagen weave, free from myocytes or other cardiac cells (Fig. 4). Repetitive sonication (120 s) yielded about a 20-fold increase in the Hyp content of the fiber over that of the initial tissue with an 80% yield (Table 2). Consecutive extractions of the fiber with 0.6 M KC1 and 1 M NaCl resulted in a 22-fold increase (by removal of other substances) in Hyp content of the fiber with a 70% yield (Table 2); approximately 67% of that material was type I collagen (Fig. 5). The amino acid composition is similar to that of other collagens (Table 3), although Gly is somewhat low and there may be minor impurities. This preparation was used for the studies described below. When sonication was extended to 150 s, the Hyp content of the fiber was not increased, but some of the collagen was released into the suspension phase with a concomitant decrease in the yield of pellet (Table 2). This suggests that part of the collagen pellet may be formed by association of different sizes and types of collagens. The isolation of the collagen fiber was best when freshly prepared tissue rather than stored frozen tissue was used. The latter often does not form a collagen pellet upon sonication and centrifugation, but fragments into a suspension form which is difficult to handle and recover. The yield of control collagen was about 95%, but with only a 5-fold purification (Table 2). Removal of noncollagenous proteins by SDS extraction resulted in ll-fold purification with 97% yield (Table 2). The last method is simple and reliable, and the isolated (denatured) collagen can be used if the undenatured form is not required.

04

-

40 Time (min)

D

FIG. 3. Chromatograms of acid hydrolysates of dansylated insoluble collagens. The acid hydrolysates of dansylated collagen were loaded (10 ,ul) on a reverse-phase C-18 column and separated using HPLC, first with a linear gradient of solvents A and B, Tris-HClmethanol (pH 7.75), then with 100% solvent B. (a) Control AT collagen (1.0 mg/0.4 ml); (b) sonicated AT collagen (1.6 mg/0.4 ml); (c) sonicated heart collagen (1.1 mg10.4 ml). Lines at top of (a) indicate the elution times of corresponding standards: (1) Dns-OH; (2) DnsNH,; (3) c-Dns-Lys; and (4) di-Dns-Lys. Assays are described in the text.

310

TAKAHASHI, TABLE Isolation

of Insoluble

2 Heart

Collagen

HYP (wol)

Protein b%)

Starting tissue Sonicated for 15 s 60 s 80 s 100 s 120 s 150 s Sonicated for 120 s and extracted Control SDS

1.2

140.00

0.05

100

6.8 6.3 6.1 5.8 5.8 4.3

15.00 8.82 7.03 6.32 5.81 4.30

0.45 0.71 0.87 0.92 1.00 1.00

95 a1 85 80 80 60

9 14 17 18 20 20

5.0 6.8 7.0

4.89 30.00 13.00

1.12 0.23 0.54

70 95 97

22 5 11

WP (wol)l mg protein

Yield (a)

Purification (fold)

Treatment of tissue

Note. The yield of collagen prepared by sonication or by sonication and extraction from the left ventricle of normal dog heart (1 g) was determined by measuring the Hyp content of the preparation and expressed as the fold purification using the value of starting tissue as 1. Control and SDS collagens were prepared by extraction without sonication. Values shown are means for three separate preparations, with analyses in duplicate. Purification procedures and assay conditions are described in the text. The protein was determined by ninhydrin reaction and expressed as leucine equivalents; see text.

NH, and t-Dns-Hyl (Fig. 3~). The collagen fiber was virtually insoluble in several extracting agents: 0.5 M acetic acid, 8 M urea or 6 M guanidine-HCl, and 1% SDS. None of the extracts contained proteoglycans.

ZHAO,

AND

ENG

Over 97% of this heart collagen was solubilized by autoclaving for 90 min; the noncollagen residue remaining after autoclaving lacked Hyp, indicating that the isolated preparation was free of elastin, since elastin is the only other protein that contains Hyp. Compared to heat-denatured collagen, the collagen fiber was significantly resistant to proteolysis by general proteinases (Table 4). These results indirectly support our other findings that the isolated heart collagen was of an undenatured form. Most of the sonicated heart collagen was solubilized by limited pepsin digestion (25,60,96% at 24, 48, 96 h, respectively) and by CNBr digestion (85%, 24 h). Typical protein bands and CNBr-peptide bands on SDS-PAGE were obtained (Figs. 5 and 6). Of the sonicated collagen, over 96% was type I collagen; this corresponded to approximately 67% of the collagen present in the left ventricle of dog hearts, which also contains some type III collagen. This was judged from the recovery of the Hyp and the protein bands on SDSPAGE (Fig. 5) of pepsin-solubilized collagen fibers. Some type V-like collagen was also present in the sample. The presence of type V-like collagen was inferred from (a) the sensitivity to C-collagenase, (b) the migration patterns on SDS-PAGE, and (c) the lack of reactivity against antibody to mouse type IV collagen. Our failure to separate type I and type III collagens by salt precipitation as reported by other investigators (28) may be due to the occurrence of intermolecular crosslinks between type I and III collagens. Such crosslinks have been shown to occur in human leiomyoma and calf aorta

FIG. 4. Scanning electron microscopic picture of heart collagen fibers. Collagen fibers were prepared from the tissue by a 15-s sonication. At the magnification shown (240X), large collagen cables of various sizes are apparent. Fine collagen fibers can be seen at higher magnification. Note that the fiber preparation is devoid of myocytes, indicating that the collagen fibers are separated from most of the cardiac cells by sonic&ion. Other conditions are described in the text.

ISOLATION

a

AND CHARACTERIZATION

OF HEART

311

COLLAGEN TABLE 4

b Action

of Selected Proteinases

on Sonicated

Heart

Collagen

% Heart collagen solubilized 8-

Undenatured

Enzyme used Control Clostridial collagenase Tissue collagenase Pepsin Trypsin Chymotrypsin Pronase E

(Y-

F1

2

3

4

5

6

FIG. 6. SDS-PAGE of pepsin-solubilized sonicated heart collagen. The pepsin-solubilized collagens were subjected to SDS-PAGE (7.5% gel) in two gels. Protein bands (a) stained with Coomassie brilliant blue and (b) visualized by radioautography of immunoblot staining using rabbit anti-(mouse skin) type I collagen antiserum at 1:lOO dilution. Lanes 1 and 4, acid-soluble mouse skin type I collagen (0.1 ag); lanes 2,3,5 and 6, pepsin-solubilized heart collagen (5 pg each), lanes 2 and 5, nonreduced and lanes 3 and 6 reduced with 0.1 M 2-mercaptoethanol for 2 h at 25°C. The j3 and a type I collagen chains are indicated. F, buffer front. Assays are described in the text.

collagens by Henkel and Granville (29), who suggested that this form of crosslink may be of special physiological importance. There were no significant differences between the protein bands on SDS-PAGE under reduc-

Heat-denatured

88.5 1.4 96.6 3.2 1.5 14.0

90.5 1.5 96.2 64.9 40.2 97.0

Note. Solubilization of sonicated heart collagen (3 mg) by proteinases (24 h, 37°C) was determined by measuring the Hyp content appearing in the soluble phase in two groups: one was undenatured and the other was heat-denatured (6O”C, 20 min); control was without enzyme. The extent of solubilization was expressed as the % of Hyp increase in the supematant over that of the control. Data points represent the means of two separate experiments, with analyses in duplicate. Assays are described in the text.

ing or nonreducing conditions (Fig. 5), and two of three bands corresponding to the /3 region of the collagen were stained with anti-mouse type I collagen antiserum (Fig. 5b), suggesting that these are structurally related. The pepsin-solubilized collagen cross-reacted with mouse antiserum to an extent of about 5% of the reactivity of

Mr (KDa)

4 -a!

TABLE 3 Amino

Acid Composition of the Collagen Preparation from Heart

Amino acid Hydroxyproline Aspartic acid Threonine Serine Glutamic acid Proline Glycine Alanine Valine Methionine Isoleucine Leucine Tyrosine Phenylalanine Hydroxylysine Hi&dine Lysine Arzinine

66-

Left ventricle of dog heart 80 66 22 34 90 118 260 97 35 9 20 36 11 23 7 10 33 53

Note. Average of duplicate analyses of samples hydrolyzed 24 h, 110°C. Values are expressed as residues per 1000 total residues.

ll-F

1

2

3

FIG. 6. SDS-PAGE of CNBr peptides of sonicated heart collagen. CNBr peptides were prepared from soluble type I collagen and from sonicated heart collagen and subjected to SDS-PAGE (12.5% gel). Samples were reduced as described in Fig. 5. Peptide bands were stained with Coomassie brilliant blue. Lane 1, acid-soluble mouse skin type I collagen (3 pg); lane 2, CNBr peptides from soluble type I collagen (5 pg); and lane 3, CNBr peptides from sonicated heart collagen (10 pg). The j3 and a type I collagen chains are indicated. Apparent M, was determined by coelectrophoresis with known protein standarda: bovine serum albumin (68 kDa), ovalbumin (43 kDa), and cytochrome c (11 kDa). F, buffer front. Assay conditions are described in the text.

312

TAKAHASHI.

0.01

0.1

1

10

100

Collagen (us/ml)

FIG. 7. Examination of “native” structure in sonicated heart collagen with an antiserum to undenatured acid-soluble mouse type I collagen. An ELISA was performed in each case with the antiserum at a 1:lOO dilution and a O.l-ml aliquot (in well) of serial concentrations of antigen as shown on the abscissa. Open symbols, undenatured collagen; and closed symbols, heat-denatured collagen at 60°C for 20 min; squares, control mouse type I collagens; and triangles, pepsin-solubilized sonicated dog heart collagens. Data are means of two separate experiments in duplicate analysis. Assays are described in the text.

the acid-soluble mouse skin type I collagen, determined by ELISA. This reactivity was abolished when the same collagen was heat-denatured (Fig. 7), indicating that the isolated collagen probably retained the triple-helical structure of type I collagen (4). This allows the inference that the isolated insoluble heart collagen was of “native” form. The purification of insoluble heart collagen by sonication has definite advantages. At present, there is no other method by which to obtain such a high degree of purification (22-fold) of native insoluble heart collagen, free of proteoglycan and elastin, in such high yield. With classical extraction methods purification is only &fold. Similar methods may be applied to other species with a modification of sonication conditions according to the size and strength of the type I collagen fibers of the heart. ACKNOWLEDGMENTS We thank Dr. Sam Seifter for presubmission review of this paper and Dr. 0. 0. Blumenfeld for helpful discussions.

REFERENCES 1. Kuhn, K. (1987) in “Structure and Function of Collagen Types” (Mayne, R., and Burgeson, R. Eds.), pp. l-37, Academic Press, New York.

ZHAO.

AND

ENG

2. Borg, T. K., Buggy, J., Sullivan, T., Laks, J., and Terracio, L. (1986) J. Mol. Cell Cardiol. lS,(Suppl. l), Abstract 247. 3. Iimoto, D. S., Covell, J. W., and Harper, E. (1988) Circ. Res. 63, 399-408. 4. Takahashi, S., Barry, A. C., and Factor, S. M. (1990) Biochem. J. 266,233-241. 5. Weber, K. T., Janicki, J. S., Pick, R., Abrahams, C., Shroff, S. G., Bashey, R. I., and Chen, R. M. (1987) Circulation 75(Suppl. l), 40-47. 6. Adomian, G. E. (1985) J. Mol. Cell Cardiol. 17, (Suppl. XV), Abstract 1. 7. Weber, G. (1952) Biochm. J. 51, 155-167. 8. Gros, C., and Labouesse, B. (1969) Eur. J. Biochem. 7,463-470. 9. Gray, W. R. (1972) Enzyme structure, in “Methods in Enzymology” (Hirs, C. H. W., and Timasheff, S. N., Eds.), Vol. 25, pp. 121-138, Academic Press, New York. 10. Kaneda, N., Sato, M., and Yagi, K. (1982) Anal. Biochem. 127,

49-54. 11. Takahashi, S., and Seifter, S. (1972) J. Appl. Bacterial. 35,647657. 12. Takahashi, S., and Simpser, E. (1981) Hepatology, 1, 211-220. 13. Takahashi, S., and Biempica, L. (1985) Cancer Res. 45, 3311-. 3321. 14. Takahashi, S., Dunn, M. A., and Seifter, S. (1980) Gastroenterology 78, 1425-1431. 15. Gray, W. R. (1967) Enzyme structure, in “Methods in Enzymology” (Hirs, C. H. W., Ed.), Vol. 11, pp. 139-151, Academic Press, New York. 16. Takahashi, S., and Lee, M.-J. (1987) Anal. Biockm. 162, 553561. 17. Zhao, M., Zhang, H., Robinson, T. F., Factor, S. M., Sonnenblick, E. H., and Eng, C. (1987) J. Am. CoU. Cardiol. 10,1322-1334. 18. Takahashi, S., and Koda, K. (1984) Biochem. J. 220,157-164. 19. Miller, E. J. (1971) Biochemistry 10,1652-1658. 20. Laurent, G. J., Cockerill, P., McAnulty, R. J., and Hasting, J. R. B. (1981) Anal. Biochm. 113,301-312. 21. Laemmli, U. K. (1970) Nature (London) 227, 680-685. 22. Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad.

Sci. USA 76.4350-4354. 23. Davison, E. A. (1966) Complex carbohydrates, in “Methods in Enzymology” (Neufeld, E. F., and Ginsburg, V., Eds.), Vol. 8, pp. 52-59, Academic Press, New York. 24. Bornstein, P. (1969) Biochemistry 8,63-70. 25. Kang, A. H., and Gross, J. (1970) Biochemistry 9, 796-804. 26. Bornstein, P., and Click, E. M. (1970) Biochemistry 9,4699-4706. 27. Nishihara, T., and Doty, P. (1958) in “Recent Advances in Gelatin and Glue Research (Stainsby, G., Ed.), p. 262, Pergamon Press, Oxford. 28. Medugorac, I. (1982) Basic Res. Cardiol. 77, 589-598. 29. Henkel, W., and Granville, R. W. (1982) Eur. J. Biochem. 122, 205-213.