Bone and ~~nerai, I8 (1992) 187-198 ~169-6009~92~$05.000 1992 Etsevier science Publishers B.V. All rights reserved.
K. Yonemuraa, A.S. Narayanana, Y. Mikib, R.C. Pagea and I-I. Okadab aDepartmentoJ Pathology. School of Medicine, University of Washington,Seatile, WA, USA ~~e~art#le~~t of ~nd~do~~tology and P~r~od~ntology~ Osaka ~~iv~r.~ity~ Osaka, Japan (Received 3 umber 1991) (Accepted 14 April 1992)
Summary Cementum is the mineralized interface through which collagen fibers of periodonta1 connective tissues are anchored onto the tooth surface. We have isolated and parti~ly characterized a mitogenic factor from human cementum which has properties different from other growth factors. Cementum was harvested from healthy human teeth, extracted in 1.0 M CH$OOH and mitogenic activities were fractionated by heparin-affinity chromatography. Proteins eluted by 0.4-0.6 M NaCI, which contained most of the cementurn mitogenic activity, were precipitated by trichloroacetic acid and resolved by HPLC through ion-exchange and reverse-phase columns. NaDodSOa-polyacrylamide 8el electrophoresis revealed that the purified preparation contained a M, 23 000 protein and this protein was associated with mitogenic activity. The purified cementum-derived growth factor (CGF) was bctive alone, but at suboptimal concentrations its activity was potentiated by small quantities of plasma-derived serum and epidermal growth fac,or (EGF). The activity was resistant to heat, but it was destroyed by trypsin digestion. Reduction and alkylation destroyed the mitogen~~acti~ty, however ele~trophoretic mobility was not affected. Binding of EGF to Bbroblast membranes was not affected by the CGF and assays to detect platelet.derived growth factor were negative. These characteristics indicated that CGF is a distinct molecular species. Our data show that cementum contains several mitogenic factors and that CGF is the major cementum mitogen.
Key words: Cementum; Growth factor: Fibroblasts
Cementum is a unique calcified structure through which collagen fibers from adjacent connective tissues are inserted onto the root surfaces of teeth. It consists Correspondence to: A.S. Narayanan, Department of Pathology SM 30, University of Washington, Seattle, WA 98195, USA.
188 of cellular and acellular portions and contains type I and III collagens, noncollagenous proteins and proteoglycans [l-6]. Cementum has a low metabolic turnover and, unlike most mineralized tissues, it does not have a blood supply .The structural integrity of eementum and its attachment to periodontal connective tissues are crucial for tooth structure, mobility and function. When pathogenic bacteria invade the teeth, connective tissues become inflamed and their matrix is destroyed and bacterial endotoxins are deposited on the ~ementum preventing the attachment of newconnective tissues. As a result, detachment of nonnative tissue from teeth and tooth loss occur. The basic cellular and molecular mechanisms required for regeneration of the damaged connective tissue and restoring its attachment to teeth are not clearly understood . Mineralized matrix of calcified structures is known to contain a variety of growth factors which include insulin-like growth factors (IGF), acidic and basic ~broblast growth factors (a- and bFGF), transforming growth factor-~ (TGF-/3), platelet-delved growth factor (PDGF) and bone morphogenetic proteins, and these are believed to play a role in tissue remodeling and wound repair [g-14]. It therefore appears possible that components released from cementum may influence the turnover and repair of surrounding connective tissues. This likelihood is supported by the observations that extracts of cementum promote the migration, attachment, growth, and protein and collagen synthesis of fibroblasts, and that these activities are either absent or present at low concentrations in adjacent tissues and in ~ementum affected by disease [l S-233.We have partially characterized the growth factors present in bovine cementum and shown that one of these growth factors is a M, 23000 protein. This protein appears to be a different molecular species from other growth factors although it resembles PDGF in some properties .In this paper we describe the isolation and partial characterization of this mitogen from human cementum.
PDGF-AB and goat anti-PDGF IgG were generously provided by Dr. Elaine Raines, Dept. of Pathology, University of Washington. Polyclonal rabbit antibody to bFGF was a generous gift from Dr. Andrew Baird, Whittier Institute for Diabetes and Endocrinology, San Diego, CA, and anti-aFGF antibody was pu~hased from R and D Systems, Minneapolis, MN. Human recombinant EGF was obtained from Sigma Chemical Company, St. Louis, MO. ~3H~-thymidine and [‘251]-EGFwere purchased from Amersham Life Sciences Products, Arlington Heights, FL. All chemicals used for our experiments were of ultrapure or comparable grade. Glassware and plasticware were siliconized using Prosil-28 (PCR, Gainesville, FL). Extraction of cememm
Pe~~ontally healthy human teeth were obtained at surgery during therapeutic extraction, adhering connective tissues removed and cementum scraped with
189 curettes under a microscope into 50 mM Tris-HCI buffer, pH 7.5, containing 2f rnM EDTA, 1 mM each of phenylmethanesulfonyl~uoride and ~-ethylmaleimide, and I flg/rnl each of pepstatin and leupeptin. It was dialyzed for 7 days at 4°C against 1.O M acetic acid containing proteinase inhibitors [18,24], and concentrated and dialyzed against 10 mA4 Na2HP04, 50 mA4 NaCl, pH 7.2 using Amicon Centriprep 10. HPLC of cementurn proteins
Proteins were loaded on a TSK Heparin-SPW HPLC column in 10 m&i Na2HP04, 50 mM NaCl, pH 7.2, and bound proteins were eluted using a O-2.0 M NaCl gradient at a flow rate of 0.5 ml/min in a Hewlett Packard HP 1090 apparatus. Fractions were collected and monitored for mitogenic activity. Indicated fractions were adjusted to 10% trichloroacetic acid (w/v) to precipitate the mitogen. The precipitate, which contained 96 f 5% (n = 3 experiments) of the total mitogenic activity, was reconstituted in 10 mM NazHPO4 buffer, pH 6.5, and loaded on a TSK CM-SPW column previously equilibrated with the same buffer. The column was eluted with a O-l .On/r NaCl gradient at a flow rate of 0.5 ml/min. Fractions with mitogenic activity were pooled, acidified with CFJCOOH and applied to a PBondpack C 1s column equilibrated with 0.1% aqueous CF&OOH. The column was eluted with a linear gradient of 0 - 100% @H&N in 0. t 1% CF$OOH at a flow rate of 0.5 ml~min. For mitogenic assays samples were concentrated and dialyzed in the presence of 2 mg/ml bovine serum albumin. Assay of DNA synthesis
Human gingival fibroblasts between fourth and twelfth transfer numbers were routinely used for the assays. Assays were performed in quadruplicate in 96-well dishes (Falcon) using 1 x IO4 cells/well. The cells were made quiescent by incubation with serum-free Dulbecco-Vogt (DV) medium for 48 h and then activated with the mitogen, After 22 h 10 @i/ml [3H]-thymidine in fresh serum-free medium was added. Six hours later the cells were washed with ice-cold phosphate-bu~ered saline (PBS) and 5% trichloroacetic acid, taken in 1.0 N NaOH, neutralized with WC1and counted in a liquid-scintillation counter . Measurement of fibroblast growth
Fibroblasts (lo4 cells) were plated in 48-well Falcon dishes. After 18 h (day 0) medium was removed and added with fresh DV-medium containing 1.0% plasma-derived serum (PDS), 1% PDS plus 10 ng/ml CGF or 10% fetal bovine serum (FBS). Every second day triplicate cultures were harvested by trypsin and cell number determined a hemocytometer. Tn the remaining cultures, media were replaced with respective fresh media and incubation continued. Chemical cleavage of disulfide bonds
Samples were reduced with 30 mM dithiothreitol in 500 mii4 Tris-HCI, pH 8.1 containing 6 M guanidine and 20 mM EDTA, at 37°C for 4 h in a nitrogen atmosphere. They were then alkylated with 70 mfMiodoacetic acid at 4°C for 1 h
uuder Nz 1251.
Quiescent human gingival ~broblasts were washed with cold PRS containing 0.1% bovine serum albumin and incubated with gentle mixing for 2 h at 4°C in binding medium (DV medium containing 25 mM HEPES and 2% plasmaderived serum), After incubation with growth factors as specified, [‘251]-EGF was added to give a final concentration of 0.25 ng/ml and incubation continued for 1 h. The cells were washed three times with cold PBS containing 0.1% bovine serum albumin, cells solubi~iz~ in 1% Triton X-100 containing 0.1% bovine serum albumin and counted in a gamma counter 1261. N;af)[email protected]
gel ~~~c~~~~~~~e~~~ was performed
in 15% separating gel slabs and proteins were visualized by staining with CoomassieL g . For determining mitogenic activity brilhant blue or by silver-staip.1 proteins were eluted from gel slices by extraction with 0.5 ml of 0.5% SDS in 1.O M acetic acid for 24 h at 4°C ; the extracts were dialyzed and concentrated using Centricon 10. ~~o~e~~co~ce~~~a~~~nswere determined by the Bradford assay  using a BioRad (Richmond, CA) protein assay kit, Proteins in CM-SPW and Crs fractions were measured by comparing peak areas to those obtained for known concentrations of bovine serum albumin at 280 nm. Results Previously we observed that acid extracts of human cementum contain most of the cementum mitogenic activity . In order to separate the activity based on differences in affinity to heparin, cementum was extracted in 1.O M CH3COOH and the extract was subj~ted to HPLC through a column of heparin. Approxi-
RetenttonTlme (min) Fii. 1. Heparinafilnity HPLC of human cementurn extract. The extract was bade an a column of TSKHeparin-SPW and eluted with a linear gradient of O-2.0M NaCl at a flow rate of 0.5ml/min. I&ml fractions were collected and monitored for mitogenic activity. Mean 1_ SD of quadruplicate assays are shown.
Fig, 2. HPLC of Helen-fraction B from Fig. 1. on TSK CM-SPW column. Proteins were precipitated by 10% trichloroa~tic acid and separated as described in the Methods. Fractions were processed as described in Fig. 1.
mately 10% of the loaded activity did not bind to the column (peak A, Fig. 1) and this fraction, which is likely to contain insulin-like growth factors and EGF [9) was not characterized further. Two minor peaks of activity bound strongly to the column and these were eluted by 0.9-I .2 and 1.4-l .8 M NaCl (peaks C and D resp~tively). These fractions, which accounts for < 10% of the total activity, were previously shown to represent a- and b-FGF 1241,therefore they were not further processed. The remaining activity, which accounted for the 52-70% of loaded activity in several experiments, eluted at the trailing end of the protein peak at 0.4-0.6 AdNaCl (peak B, Fig. 1). Proteins in the peak B fractions were precipitated by 10% trichloroacetic acid and separated by HPLC in TSK CM-SPW column as described in Fig. 2. Approximately 40% of loaded proteins and m 10% of mitogenic activity did not bind to the column rlpcak A, Fig. 2). The remaining mitogenic activity was eluted by 0.55-0.75 M NaCI (peak B, Fig. 2); fractions under this peak were pooled and separated by reverse-phase HPLC through a PBondpack Cls column (Fig, 3). A single peak of mitogenic activity was obtained between 3540% CH$N. The Cts fraction represented a 6000-fold purification of mitogenic
Fig. 3. C,s-reverse-phase chromatography of peak El proteins from Fig. 2. The fractions were pooled, separated on 0.11% trifluoroacetic acid using a 0 - 100% CHsCN gradient and assayed for mitogenic activity as described in the methods.
192 Table 1
Purification summary for human cementem derived mitogenic factor from a representative preparation Fraction
Acid extract Heparin-SPW TCAb CM*SPW CM
28 200 753 349 8.1 0.4
15 630 1 365 10 785 88 875
Fold purification 1 42 91 719 5925
Recovery of activity 100 67 59 21 12
u Wr:its/mgprotein; one unit of activity is equal to the [SH]-thymidine incorporation induced by 10% fetal bovine serum, Assays were pcrfo~ed in a total volume of 100id serum-free DV medium containing 0.5% fetal bovine serum plus CGF fractions. b Trichloroacetic acid precipitate.
activity and 98%) was lost upon reduction (not shown). PDGF has similar properties and Table 3
Action of anti-P~GF and anti-CGF antibodies on Mitogenic Stimulation by CGF and PBGF Experiment
cpm x IO-”
None x-PDGF None x-PDGF None x-PDGF
I.5 + 1.6 f 5.3 + 5.6 It 10.4 f 1.5 f
0.1 0.2 0.6 0.7 0.9 0.2
100 107 100 106 100 19
1.2 f 1.1 f 5.1 f 3s f 8.9 f 8.1 f
0.1 0.2 0.6 0.6 1.0 1.1
100 92 100 69 100 97
None CGF PDGF-AB
None x-CGF None x-CGF None x-CGF
a 10 n&ml Cts fraction and 2 ng/ml human platelet PDGF-AB. b Mitogens ‘mm incubated with 50 +tg/mlgoat anti-PDGF IgG or 1:100 dilution of anti-CGF antibody 2SN232for 2-h at 37°C and centrifuged. ’ [3Hl-thymidine uptake, mean f SD of quadruplicated.
Fig.7. Radioreceptor assay for [“‘I]-EGF binding. Confluent quiescent Bbroblast were incubated with growth factors incubated for 2 h at 4°C in binding medium, and then iodinated EGF was added to give a final concentration of 0.25 ng/ml. After I h, the cells were washed and counted. Mean + SD of quadruplicate values are shown. , EGF; PDGF-AB; o-0, CGF.
it also has moderate affinity to heparin [9,28], therefore experiments were done to examine whether the CGF and PDGF are one and the same. In the first experiment the effect of an anti-PDGF antibody on mitogenic activity of CGF was examined. This antibody, which recognizes both A and B chains of PDGF , inhibited PDGF action, but it did not affect CGF (Table 3). On the other hand, a monoclonal antibody raised against bovine CGF  inhibited the CGF action, but not PDGF (Table 3). Anti-aFGF and anti-bFGF antibodies also did not affect the mitogenic activity of CGF (data not shown). In the second experiment the effect of CGF on the bmding of [‘251]-EGF to tibroblast membranes was examined; however, while PDGF decreased the [‘*‘I]-EGF binding, CGF had no effect (Fig. 7). Discussion Our data show that human cementum contains mitogenic factors which can be fractionated based on differences in affinity to heparin. These include those which do not bind or bind only weakly to heparin, FGFs which bind strongly to heparin, CGF which has moderate aflinity, and other unidentified factors extractable by guanidine [l&24]. The cementum resembles other mineralized structures in this respect; however, while IGF-like molecules which do not bind to heparin and FGFs account for the major portion of mitogenic activity in alveolar bone and other bones, they are only minor components in cementum [9,24]. The mitogenic factor designated as CGF is the major component in bovine  and human cementum (Fig. 1), and it accounts for -70% of mitogenic activity extracted from this tissue by CH$OOH. We have modified the previously described procedure  to purify human
196 CGF. This procedure, which involves precipitation by trichloroacetic acid and HpLC through CM-SPW and Cl8 columns, achieved -6OOO-fold purification and yielded an almost homogeneous preparation containing the M, 23 000 protein as the chief component. The preparation contained additional staining material at M, - 15 000, however the staining was absent after reduction and it had no detectable mitogenic activity. A M, 19 000 contaminant, which was a major component in previous preparations ,was removed by using the current procedure (Fig, 4). Mitogenic activity is associated with the M, 23 000 band; however if the activity is due to this protein or others comigrating with it could not be determined by 2D gels due to low recovery of activity from gels. The human CGF obtains here and bovine CGF reported previously  have almost identical properties. Both proteins were mitogenic to skin and periodontal ligament f‘lbroblasts and towards bovine aortic smooth muscle cells (data not shown, ),They were heat stable, but susceptible to reduction. They were active alone, however manifested synergism with other factors such as EGF. In this property the CGF resembles PDGF, however, while PDGF inhibits EGF binding, the CGF does not. A compa~son of the properties of CGF from both human and bovine ~mentum reported here and previously  with those of other growth factors indicates that the former is a different molecular species. For example, even though the CGF resembles PDGF in affinity to heparin, synergism with ‘progression factors’, heat stability and susceptibility to reduction, it is not inhibited by anti-PDGF antibodies. While reduction converts PDGF into subunits of M, 14 400,16 000 and I7 000 ,no subunits are formed from the CGF. We were also unable to detect PDGF in crude cementum extracts as well as Cis fractions, either by radioreceptor assays, or by ELISA using antibodies which will recognize both A and B chains of PDGF. PDGF was not detected even after CMcellulose chromatography to remove binding proteins. Heat resistance, chromatographic properties and lack of inhibition by anti-FGF antibodies indicate that the CGF is not FGF, and its molecular mass and mitogenic properties show that it is not EGF, IGF or one of the interleukins. Recently a new mitogen with same mol~ular mass as CGF has been described in the media of U-937 cells , but this growth factor belongs to the EGF family and it has greater affinity to heparin than CGF. The CGF does not appear to be TGF-j3 because the latter is not mitogenic to Bbroblasts and, while TGF-fl stimulates collagen synthesis, CGF does not [l&32]. Its heparin-elution properties and molecular size are different from that of IGF-I and IGF-II 19,141.The solubility properties of CGF are different from that of bone-morphogenetic proteins, and in bioassays done by Dr. AH. Reddi after subcutaneous implantation in rats we were unable to detect BMP activity in CGF preparations [18,24].While these differences indicate that the CGF has characteristic properties, further data on its primary structure are n=ssary to determine whether it belongs to the family of already described growth factors or it is a new species altogether. Interestingly, even though both PDGF and CGF act syner~sti~lly with EGF, only PDGF, not CGF, affects EGF-binding to cell membranes; this indicates that these growth factors synergize with EGF through different mechanisms. The CGF activity is susceptible to
reduction even though the electrophoretic migration does not change; this indicates that it requires intramolecular disulfide bonds for activity, The ~mentum contains not only growth factors, but also substances which promote fibroblast chemotaxis, attachment and collagen synthesis [15-201.The role of these substances and whether these activities are directed towards cells present in cementum or adjacent periodontal ligament, gingiva and dentin is not known. Interestingly, these activities are not detectable in other tissues which are in contact with cementum [15,16,18]. While these observations indicate that cementum has the potential to regulate the me~bolism and turnover of surrounding tissues, whether the active components are readily accessible is not clear as they are present immobilized in the mineralized matrix. It is likely however that the cementum, like other calcified tissues, serves as a storage site for these molecules . They could be released by degradation of the cementum matrix during in~ammation, thus the CGF could be a part of the response to injury and help to initiate the repair process. However its susceptibility to proteases may render it short-lived and this may be one reason why connective tissue attachment does not form effectively in chronic periodontitis. Nevertheless, the unique localization of CGF and other cementum components in the matrix indicates that they may play a key role in the formation of connective tissue on tooth root surfaces and sub~quent integration of collagen fibers into ~mentum.
This grant was funded by NIH grant DE-08229. We thank Dr. Elaine Raines for the gift of PDGF-AB and for IDGF assays, and Dr. A. Baird for anti-FGF antibody. We also thank Xidong Li and Linda Feldman for technical assistance. References Birkedai-Han~n H, Butler WT, Taylor RE. Proteins of the periodontium. characterization of the insoluble coiiagens of bovine dental ~mentum. Caicif Tissue Res ~977;72~3~~~. Christner P, Robinson P, Clark CC. A prciiminary characterization of human cementum collagen. Caicif Tissue Res 1977;23:147-150. Choveion A, Carmichael DJ, Pearson CH. The composition of the organic matrix of bovine cementum. Arch Oral Bioi 1975;20:537-541. Smith AJ, Leaver AG, Smith G. The amino acid composition of the non-coiiagenous organic matrix of human cementum. Arch Oral Bioi 1983:28:1047-1054. Bartoid PM, Miki Y, McAllister B, Namyanan AS, Page RC. Glycosaminogiy~ns of human cementum. J Periodont Res 1988;23:13-17. Bartold PM, Reinboth B, Nckae H, Narayanan AS, Page RC. Proteogiycans of bovine cementum: Isolation and characterization. Matrix 1990;10:10-19. Schiuger S, Yuodeiis R. Page RC, Johnson RH. In: Periodontal diseases: basic phenomena, clinical management, and occiusai and restorative interrelationships. Philadelphia, Lea & Pebiger; 1990;3842. Wirthiin MR, Hancock EB. Biologic preparation of diseased root surfaces. J Periodontal ~980;51:291297. Hauschka PV, Mavrakos AE, Iafrata MD, Doieman SE, Klagsbrun M. Growth factors in bone
198 matrix. Isolation of multiple types by affinity chormatography on Heparin-Sepharose. J Biol Chem 1986;261:12665-12674. 10 Baird A, B6hlen P. Fibroblast growth factors. In: Spom MB, Roberts AB, eds. Peptide growth factors and their receptors. I. Berlin: Springer-Verlag, 1990:369-418. 11 &y&in SM, Thompson AY, Bentx H, Rosen DM, MacPherson JM, Conti A, Siegel NR, Galluppi GR, P&- KA. Cartilage-inducing factor-A. Apparent identity to transforming growth factor beta. J Biol Chem 1986:261:5693-5695. 12 Kam RW, Reddi AH, Dissociative extraction and partial purification of OsteOgenin, a bone inductive protein from rat tooth matrix by heparin atTInity chromatography. Biochem Biophys Res Commun 1988;157:1253-1257. 13 Mayer H, Kukoschke K-G. Puriftcation and characterisation of a growth factor from porcine bone. Eur J Biochem 1989;181:409-415. 14 Rechler MM, Nissley SP. Insulin-like growth factors. In: Sporn MB, Roberts AB, eds. Peptide growth factors and their noeptors. I. Berlin: Springer-Verlag, 1990;263-367. 15 Nishimura K, Hayashi M, Matsuda K, Shigeyama Y, Yamasaki A, Yamaoka A. The chemoattractive potency of periodontal ligament, cementurn and dentin for human gingivval Bbroblasts. J Periodont Res 1989;24:146-148. I6 Somerman MJ, Foster RA, Imm GM, Sauk JJ, Archer SY. Periodontal ligament cells and gingival Bbroblasts respond differently to attachment factors in vitro. J Periodontol 1989;60:73-77. 17 McAllister B, Narayanan AS, Miki Y, Page RC. Isolation of a tibroblast attachment protein from cementum. J. Periodont Res 1990;25:99-105 18 Miki Y, Narayanan AS, Page RC. Mitogenic activity of cementum components to gingival flbroblasts. J Dent Res 1987;6631399-1403. 19 Somerman MJ, Archer SY, Shteyer A, Foster RA. Protein production by human gingival tibroblasts is enhanced by guanidine EDTA extracts of cementurn. J Periodont Res 1987;22:75-77. 20 Kawai T, Urist MR. Bovine tooth-derived bone morphogenetic protein. J Dent Res 1989;68:10691074. 21 Aleo JJ, De Renzis FA, Farber PA, Varboncoeur AP. The presence and biologic activity of cementumbound endotoxin. J Periodontol 1974;45:672-675. 22 Olson RH, Adams DF, Layman DL. Inhibitory effect of periodontally diseased root extracts on the growth of human gingival Rbroblasts. J Periodontol 1985;56:592-596, 23 Pitaru S, Melcher AH. Orientation of gingival tibroblasts and newly-synthesized collagen fibers in vitro. J Periodont Res 1983;18:483-500. 24 Nakae H, Narayanan AS, Raines E, Page R. Isolation and partial characterization of mitogenic factors from cementum. Biochemistry 1991;30:7047-7052. 25 Raines EW, Ross R. Platelet-derived growth factor 1. High yield purification and evidence for multiple forms. J Biol Chem 1982;257:5154-5160. 26 Bowen-Pope DF, Dicorleto PE, Ross R. Interactions between the receptors for platelet-derived growth factor and epidermal growth factor. J Cell Biol 1983;96:679-683. 27 Bradford MM. A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the priciple of protein-dye binding. Anal Biochem 1976;72:248-254. 28 Raines EW, Bowen-Pope DF, Ross R. Platelet-derived growth factor. In: Sporn MB, Roberts AB, eds. Peptide growth factors and their receptors. I. Berlin: Springer-Verlag. 1990;173-262. 29 Hart CE, Bailey M, Curtis DA, Osbom S, Raines E, Ross R, Forstrom JW. Purification of PDGF-AB and PDGF-BB from human platelet extracts and identificatian of all three PDGF dimers in human platelets. Biochemistry 1990;29:166-172. 30 Anate H, Olson SW, Page RC, Narayanan AS. Isolation of human tumor cells that produce cementum proteins in culture. Bone Miner 1992;18:15-30. 31 Higashiyama S, Abraham JA, Miller J, Fiddes JC, Klagsbrun M. A heparin-binding growth factor secreted by macrophage-like cells that is related to EGF, Science 1991;251:936-939. 32 Namyanan AS, Page RC, Swanson 3. Collagen synthesis by human Bbroblasts. Biochem J 1989;260:463+69. 33 Sommer A, Rifkin DB. Interaction of heparin with basic Bbroblast growth factor: protein of the ansiogenic protein from ‘proteolytic degradation by a glycosaminoglycan. J Cell Physiol 1989; 138:21%220.