Primary structure of the heparin-binding site of type V collagen

139

Biochimica et Biophysica Acta, 1035 (1990) 139-145

Elsevier BBAGEN 23346

Primary structure of the heparin-binding site of type V collagen Yoshihito Yaoi 1, K a h o k o Hashimoto 1, Hiroyuki Koitabashi 1,2, Kazuhiko T a k a h a r a 3, M a n a b u Ito 3 and Ikunoshin Kato 3 1 Biology Division, National Cancer Center Research Institute, 2 Department of Chemistry, College of Arts and Sciences, Tokyo University, Tokyo and 3 Biotechnology Research Laboratories, Shiga (Japan)

(Received 2 January 1990)

Key words: Primary structure; Heparin binding site; Collagen V

The abilities of collagens, type I, II, III, IV, and V, to bind heparin were examined by heparin-affinity chromatography and binding studies with [3sS]heparin. At a physiological pH and ionic strength, only type V collagen bound to heparin. Collagens type I and II showed higher affinities than types III and IV for heparin, but did not bind to a heparin column at a physiological ionic strength. The heparin binding site of type V collagen was located in a 30 kDa CNBr fragment of the at(V) chain, and the amino acid sequence of this fragment was determined. The 30 kDa fragment contained a cluster of basic amino acid residues, and enzymatic cleavage within this basic domain greatly reduced the heparin-binding activities of the resulting peptides. Thus this basic region is probably the heparin-binding site of type V collagen.

Introduction Interactions between proteins and proteoglycans are essential for the structure of the extracellular matrix. As in vitro models of these interactions, the binding of heparin to matrix proteins such as fibronectin, laminin or collagen have been studied extensively [1-6]. However, discrepant results have been obtained for the interactions between collagen and heparin. Smith and Brandt [7] reported that under physiological conditions type XI collagen strongly bound to a heparin agarose column, whereas types I, II and V bound only weakly and could be eluted at a physiological ionic strength (0.15 M NaC1). On the other hand, LeBaron et al. [8] found that heparin or heparan sulfate bound to microtiter wells coated with type V collagen under physiological conditions. They also showed that in this assay system type V collagen had higher affinity than fibronectin, laminin, or collagen types I, II, III, IV or VI for heparin and heparan sulfate. Keller et al. [9] reported that type I collagen bound to heparin-Sepharose at 0.15 M NaC1, and was eluted with 2 M NaCl/6 M urea, and that the triple-helical conformation of collagen was essential for this binding. Koliakos et al. [10] showed that type IV collagen has at least two

Correspondence: Y. Yaoi, Biology Division, National Cancer Center Research Institute, Tsukiji 5-1-1, Chuo-ku, Tokyo 104, Japan.

heparin-binding sites, but that the binding activity was destroyed by its partial digestion with pepsin. In the present study, we investigated the interaction between type V collagen and heparin by affinity chromatography and by binding studies with radioactive heparin. We found that type V collagen bound strongly to heparin at a physiological pH and ionic strength, whereas types I, II, III and IV did not. The heparinbinding site in the type V collagen molecule was located in a 30 kDa CNBr fragment of the al(V ) chain, and we determined the amino acid sequence of this site. Materials and Methods

Materials The following materials were used: Bovine pepsinized type I, II, III, IV and V collagens (Koken), human pepsinized type V collagen, trypsin (EC 3.4.21.4) (TPCK-treated), and cyanogen bromide (Sigma), Staphylococcus aureus V8 proteinase (Behringer) and lysyl endopeptidase (Wako Pure Chemicals). Affinity chromatography Heparin-affinity chromatography was performed on a colunm of TSK-gel heparin 5PW (0.75 × 7.5 cm, Toso). Native collagens were dissolved in PBS/2 M urea, and chromatography was carried out at room temperature i n t h e presence of 2 M urea. Peptide fragments were fractionated in PBS without urea. HeparinSepharose 6LB (Pharmacia) was used for large scale chromatography.

0304-4165/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)

140

SDS-PA GE and blot analysis SDS-PAGE was carried out in Laemmli's buffer system [11] under reducing conditions. The proteins were transferred to nitrocellulose membranes (Schleicher & Schuell, BA85) by the method of Towbin et al. [12]. The membranes were washed with PBS which contained 0.05% Tween 20 (T-PBS), overlaid with 2 /~Ci/ml of [35S]hepadn (17.35 mCi/g, Amersham) in T-PBS, and allowed to stand at 4 ° C overnight. The membranes were then washed with T-PBS, and bound [35S]heparin was detected by autoradiography with Fuji C R M F X-ray film.

Fragmentation and fractionation of peptides Type V collagen (10 mg) was dissolved in 2 ml of 70% formic acid and trated with 40 mg of CNBr at room temperature for 20 h. Digestions with proteinases (10 /~g/ml) were carried out in 50 mM ammonium bicarbonate (pH 7.8) at 3 7 ° C for 4 h (trypsin) or 15 h (V8 proteinase and lysyl endopeptidase). Reverse-phase HPLC was performed on a column of RPC-SC18 (Jasco). Peptides were eluted with a linear gradient from 0.01% trifluoroacetic acid to 60% acetonitrile at a flow rate of 1 ml/min.

Amino acid analysis and protein sequencing Amino acid sequencing was carried out with a Protein Sequencer 477A (Applied Biosystems). Amino acid compositions were determined after hydrolysis of samples with 6 M HC1 that contained 0.1% phenol at l l 0 ° C for 24 h. Results

Binding of collagens to heparin The affinities of bovine collagens types I, II, III, IV and V for heparin was examined by heparin-HPLC

B 0.5M NaCI

0.5M NaCI

IV

<

.

lt0

20

30

40

.""

'0.3

50

T i m e (min)

Fig. 2. Heparin-HPLC of a mixture of collagens. Bovine collagens were dissolved in 10 m M phosphate b u f f e r / 2 M urea (pH 7.4) at 1.5 mg of each collagen per ml, and fractionated by heparin-HPLC with a linear gradient of NaC1. R o m a n numerals indicate the elution positions of standard collagens.

(Fig. 1). Collagen types I, II, III and IV did not bind to the column at 0.15 M NaC1 (pH 7.4). As an example, the elution profile of type I collagen was shown in Fig. 1A. However, under the same conditions, type V collagen bound t o the column and was eluted from the column at 0.5 M NaC1 (Fig. 1B). Fig. 2 shows the gradient elution profile of a mixture of collagens. Types III and IV did not bind to the column in 10 mM phosphate buffer (pH 7.4). Types I and II bound to the column but began to be eluted at an NaC1 concentration of less than 0.15 M, while type V was eluted at about 0.3 M NaC1. Type V collagen was heat-denatured at 50 ° C for 15 rain, and then fractionated by heparin-HPLC into bound and unbound fractions (Fig. 3A). SDS-PAGE of these fractions revealed that the a 1 chain bound to the heparin column, whereas the a 2 chain was eluted in the flow through fractions (Fig. 3B). Next, the various types of collagen were separated by SDS-PAGE, and then transferred to a nitrocellulose membrane to examine their bindings of [3sS]hepadn. As shown in the autoradiogram in Fig. 4B, [35S]hepadn bound selectively to the cq chain of type V collagen. These results indicate that only type V collagen has the ability to bind heparin under physiological conditions, and that the binding site is in its otI chain.

E c

o

CNBr fragmentation of type V collagen

(N ,<

10

20

30 '

1'0

20

30

T i m e (rain)

Fig. 1. Heparin-affinity H P L C of collagens. Bovine pepsinized collagens type I (A) and type V (B) were dissolved in P B S / 2 M urea at 1.5 m g / m l , and aliquots of 0.15 ml were loaded on a column of TSK-gel heparin 5PW. The column was washed with P B S / 2 M urea, and then bound material was eluted by adding 0.5 M NaCI to the starting buffer.

To locate the heparin-binding site in Otl(V ) m o r e precisely, we fragmented type V collagen with CNBr and isolated the heparin-binding peptides by heparinHPLC (Fig. 5A). The unadsorbed fraction, and that bound to the column and eluted with 0.5 M NaC1, were examined by SDS-PAGE. As shown in Fig. 5B, the major heparin-binding peptide in the adsorbed fraction was a 30 kDa fragment. Binding of [35S]heparin on a nitrocellulose membrane confirmed that this 30 kDa fragment was the major heparin-binding peptide (Fig. 5B, lane d).

141

B (1

b

c kD

0. SM

NoCI

--

¢,¢,2 - -

200

- - 116

--

97

--86

E

c 0

I 10

I 20 Time

30

(min)

Fig. 3. Heparin-HPLC of heat-denatured type V collagen. (A) Bovine type V collagen was dissolved in PBS/2 M urea (1.5 m g / m l ) and denatured at 5 0 ° C for 15 min. Heparin-HPLC was carried out as described for Fig. 1. (B) SDS-PAGE of type V collagen (a), and the unadsorbed (b) and adsorbed (c) fractions from the heparin column. 4-20% gradient gels.

For determination of the primary structure of the heparin-binding site, human type V collagen was treated with CNBr, and the 30 kDa heparin-binding fragment was partially purified by affinity chromatography on a column of heparin-Sepharose CL-6B, followed by digestion with V8 proteinase. Fig. 6A shows the elution

A~ m

n

B "

profile of the total digest on reverse-phase HPLC. The digest was dissolved in 10 mM phosphate buffer that contained 0.05 M NaC1 (pH 7.4), and loaded on a heparin-HPLC column. The column was washed with the same buffer, and adsorbed material was then eluted by the addition of 0.5 M NaC1 to the buffer. Two peptides, termed H-1 and H-2 (Fig. 6C), were recovered in the heparin-bound fraction. These two peptides were no longer capable of binding to the heparin column at 0.15 M NaC1, indicating that their affinities for heparin were lower than that of the original 30 kDa fragment. Peptide H-2 was digested with lysyl endopeptidase, and separated into three fractions, L-l, L-2 and L-3, by reverse-phase HPLC (Fig. 7). Peptide L-2 was further split with trypsin into two fragments, T-1 and T-2 (Fig.

8).

! fl

III IV V 1 fl fll I V V

TYPE

Fig. 4. Binding of [35S]heparin to collagens on a nitrocellulose membrane. A, Coornassie blue staining and B, autoradiogram after binding of [35S]heparin. 5% gels.

The results of protein sequencing of these peptides are shown in Fig. 9. Several residues could not be identified by the usual sequencing procedures, and are designated as X in Fig. 9. For their identification, peptides L-3 (containing two Xs), T-1 (one X) and T-2 (one X) were subjected to amino acid analysis and found to contain 1.81, 0.93 and 1.03 mol of hydroxy-

142 A

kD

0.5M

NaCl

97 -

6643-

31-

E c

30 K

22-

% N

6

14-

Fig. 5. Heparin-HPLC of CNBr fragments of type V collagen. (A) The CNBr fragments were dissolved in PBS and fractionated by heparin-HPLC without urea. (B) SDS-PAGE of CNBr fragments (a), unadsorbed (b) and adsorbed (c) fractions from the heparin column and (d) the binding of [ 35S]heparin to total CNBr fragments. 4-204 gradient gels.

C

0

H-l

H-2

I-

0

10

Retention

20

time

30

40

0

IO

20

30

40

(mln)

Fig. 6. Reverse-phase HPLC of the V8 proteinase digest of the 30 kDa CNBr fragment of human type V collagen. A, total digest; B, unadsorbed; and C, adsorbed fractions from heparin.

143 30K: GIXGDRGEIGPP*GPRGEDGPEGP H-1 : RGPRGITGKP*GPXGNSGGDGPAGPP*GERGP

L-3

H-2: IGPP*GPRGEDGPEG I_-1: P°GPRGQRGPTGPRGE L-2: IGPP*GPRGEDGPEGP L-3: LGVP*GLP*GYP*GRQGPXGSIGFP*GFP*GADGEXGGRGTP*GK T-1 : GEDGPEGPXGR L-2

T-2: IGPP*GPRGPLGPP'GEXGK

E

Fig. 9. N-terminal amino acid sequences of the peptides derived from the 30 kDa CNBr fragment (30K) of human type V collagen. Unidentiffed amino acids and hydroxyproline are shown by X and P*, respectively.

C

o

L-1

OJ

<

I

I

I

I

I

0

10

20

30

40

Retention

time

(rain)

Fig. 7. Reverse-phase HPLC of a lysyl endopeptidase digest of peptide H-2.

lysine residues per mol, respectively. These results indicate that the unidentified residues in these peptides were glycosylated hydroxylysines, whose PTH derivatives are not extracted by organic solvents in sequence analysis [13]. Additional evidence for the existence of

T-2

E C

o

P3 04

T-1

<

8

' 10

2 '0

Retention

3'0

time

4'0

(min)

Fig. 8. Reverse-phase HPLC of a tryptic digest of L-2.

hydroxylysine glycosides is that peptide bonds associated with these residues were not cleaved by lysyl endopeptidase and typsin. These peptides were arranged along the sequence of human al(XI ) reported [14], which is very similar to that of al(V) (Fig. 10). The residue numbers within the helical portion of al(V) were deduced from the previous reports [14,15]. The sequences of the homologous regions of human al(XI ) and 0t2(V) [16] were also compared. Discussion

The present results demonstrated that type V collagen bound to heparin at a physiological pH and ionic strength. Collagen types I and II showed higher affinity than types III and IV for heparin, but did not bind to a heparin column at a physiological ionic strength. Heparin-affinity chromatography is thus a convenient method for separation of different types of collagen. Keller et al. [9] reported that the binding of type I collagen to a heparin column in the absence of urea requires the triple-helical conformation of the collagen, and used 2 M NaC1/6 M urea to elute bound collagen. This binding probably involves different types of interactions between collagen fibers and heparin from those studied here. Koliakos et al. [10] found that type IV collagen also bound to heparin, but that its binding activity was destroyed by partial digestion with pepsin. Using heparin affinity chromatography and [35S]heparin-binding studies we found that the major binding site was present in a 30 kDa CNBr fragment of al(V). These results suggested that the primary structure of the binding site, rather than the quaternary structure of collagen, was important in the interaction with heparin. We therefore determined the partial amino acid sequence of the 30 kDa fragment, which is shown in Fig. 10. Since the cleavage of the Glu-365-Arg-366 bond by V8 proteinase greatly reduced the heparinbinding activities of the resulting two peptides, H-1 and H-2, the sequence around this cleavage site seems to be

144 T-1

---

'=

T-2

---

L-2

=

BrGN

V8

T

T

; al(V)

GIX

GDRGEIGPP*G

L

;

PRGEDGPEGP

1

K*GRIGPP*GPR

LK

cI2(V)

EP

-

V QI PH

-

IQ

L-3

GPLGPP*GEK*G

KLGVP*GLP*GY

300

280 cIl(Xl)

~=

H-2

=m

A P

I

E KR

,

R

----- ~

T

DP

S QA

DP TL

P

V ER

T AP

NR F

SD

L •

•

•

GFP*GFP*GADG EK*GGRGTP*GK 340 N PV

SS

PK

SQ

A

VA

DP

R

PK

AQ

ER

H-1

V8 •

•

•

P*GPRGQRGPT

1 •

•

GPRGERGPRG

•

ITGKP*GPXGN

360

LP

A

L

SGGDQPAGPP*

GERGP

38O

S E

L

, L-1

~

1

P*GRQGPK*GSI

320

NP VQ

A E

P KL PL

K T AP E

P D RP

P SI

IK

Q

Fig. 10. Primary structure of the heparin-binding site of human type V collagen. The vertical arrows indicate the positions of cleavage by CNBr, V8 proteinase (V8), lysyl endopeptidase (L) and trypsin (T). The 30 kDa CNBr fragment was digested with V8 proteinase and two heparin-binding peptides, H-1 and H-2, were isolated. Peptide H-1 was cleaved into L-l, L-2 and L-3 by lysyl endopeptidase. Peptide L-2 was further split into T-1 and T-2 by trypsin. Horizontal lines indicate the arrangement of all the peptides sequenced. The residue numbers of al(V) within the helical portion were given under the determined sequence. Basic amino acid residues in the putative heparin-binding region are indicated by triangles. For comparison, the homologous regions of human al(XI ) and a2(V) collagen chains are presented. P *, hydroxyproline; and K *, hydroxylysine.

the most important for heparin binding. This region has a high content of basic amino acids: the region from Hyl-342 to Lys-374 in Otl(W) contains nine basic amino acids as indicated by the triangles in Fig. 10, but Glu-365, whose carboxyl site was hydrolysed with V8 proteinase, is the only acidic amino acid. We propose that this basic domain is the most likely heparin-binding site of a l ( V ). Several cell adhesive glycoprtoeins, such as vitronectin and laminin, bind heparin. The heparin-binding site of vitronectin contains a cluster of basic amino acids, and constitutes the most hydrophylic region of the molecule [17]. Charonis et al. [5] identified one heparin-binding site in the laminin B t chain as a region rich in basic amino acids. They concluded that a high density of positive charges is not sufficient for the binding of heparin, but that the distance between charged groups is probably important. The basic domain of al(V) contains few hydrophobic amino acids, and so is consistent with their proposal. The spacial arrangement of basic residues in this domain should provide a clue to the structural requirements for protein-heparin interactions. In preliminary experiments we confirmed that type XI collagen obtained from cartilage also bound to a heparin column under physiological conditions, and was eluted with about 0.3 M NaC1. The arrangement of basic amino acids in the putative heparin-binding do-

main of a~(V) is well conserved in the corresponding region of al(XI ), but no such basic domain is present in the corresponding region of a2(V ) (Fig. 10). Since Glu355 in this region of al(V) is replaced by serine in at(XI), this domain of al(XI) must be even more basic than that of al(V). This basic domain of al(XI) might also be involved in the interaction of type XI collagen and heparin. Types V and XI are believed to be closely related not only structurally but also functionally [14]. The high affinities of these two collagens for heparin suggests their specific roles in the organization of the extracellular matrix.

Acknowledgements This work was supported by a Grant-in-Aid for Cancer Research (59010068) from the Ministry of Education, Science and Culture of Japan.

References 1 Stathankis, N.E. and Mosesson, M.W. (1977) J. Clin. Invest. 60, 855-865. 2 Ruoslahti, E. and EngvaU, E. (1980) Biochim. Biophys. Acta 631, 350-358. 3 Yamada, K.M., Kennedy, D.W., Kimata, K. and Pratt, R.M. (1980) J. Biol. Chem. 255, 6055-6063. 4 Sakashita, S., Engvall, E. and Ruoslahti, E. (1980) FEBS Lett. 116, 243-246.

145 5 Charonis, A.S., Skubitz, A.P.M., Koliakos, G.G., Reger, L.A., Dege, J., Vogel, A.M., Wohlhueter, R. and Furcht, LT. (1988) J. Cell Biol. 107, 1253-1260. 6 Lindahl, V. and Hook, M. (1978) Annu. Rev. Biochem. 47, 385417. 7 Smith, G.N., Jr. and Brandt, K.D. (1987) Collagen Rel. Res. 7, 315-321. 8 LeBaron, R.G., Hook, A., Esko, J.D., Gray, S. and Hook, M. (1989) J. Biol. Chem. 265, 7950-7956. 9 Keller, K.M., Keller, J.M. and Kuhn, K. (1986) Biochim. Biophys. Acta 882, 1-5. 10 Koliakos, G.G., Kouzi-Kallakes, K., Furcht, L.T., Reger, LA. and Tsilibary, E.C. (1989) J. Biol. Chem. 264, 2313-2323. 11 Laemmli, U.K. (1970) Nature 227, 680-685.

12 Towbin, H., Staelin, T. and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA 76, 4350-4354. 13 Butler, W.T., Finch, J.E. and Miller, E.J. (1977) Biochemistry 16, 4981-4990. 14 Bernard, M., Yoshioka, H., Rodriguez, E., Res, M., Kimura, T., Ninomiya, Y., Oleson, B.R. and Ramirez, F. (1988) J. Biol. Chem. 263, 17159-17166. 15 Seyer, J. and Kang, A.H. (1989) Arch. Biochem. Biophys. 271, 120-129. 16 Weil, D., Bernard, M., Gargano, S. and Ramirez, F. (1987) Nucleic Acid Res. 15, 181-198. 17 Suzuki, S., Oldberg, A., Hayman, E.G., Pierschbacher, M.D. and Ruoslahti, E. (1985) EMBO J. 4, 2519-2524.