- Email: [email protected]
Production of homogeneous basic fibroblast growth factor by specific enzymatic hydrolysis of larger microheterogeneous molecular forms
Journal of Biotechnology, 21 (1991) 83-92 © 1991 Elsevier Science Publishers B.V. All rights reserved 0168-1656/91/$03.50 ADONIS 016816569100145S
83
BIOTEC 00674
Production of homogeneous basic fibroblast growth factor by specific enzymatic hydrolysis of larger microheterogeneous molecular forms D. B e t b e d e r 2, p. Caccia ~, G. Nitti ~, F. Bertolero t, p. Sarmientos 1, F. Paul 2, p. Monsan 2, G. Cauet ~'* and G. Mazu6 t I Department ofBiotechnology, Farmitalia Carlo Erba, Milan, Italy and 2 BioEurope, Toulouse, France
(Received 15 March, 1991; revision accepted 20 May 1991)
Summary The 146-amino acid form of basic fibroblast growth factor (bFGF) was expressed in Escherichia coli and purified by a two step process including ion exchange and heparin-Sepharose chromatographies. However, the resulting protein consisted of a mixture of 146- and 145-amino acid forms, indicating that, besides the initial methionine, also the following residue (proline) was removed from the N-terminus. The same phenomenon was observed when the 155-amino acid form, which is biologically equivalent to the shorter one, was expressed in E. coli. Taking into account the previously known data concerning the possible mechanism of cleavage of the extended forms of bFGF in vivo, we developed an efficient enzymatic process that allows the production of an homogeneous 146amino acid form from recombinant NH2-end extended forms. This process takes advantage of the protecting effect that heparin exerts on bFGF. Accordingly, when bFGF, complexed to heparin, is treated with pepsin A, an aspartic protease with a broad specificity, only the Leu9-Prot0 peptide bond is cleaved generating the 146-amino acid form. Quantitative yields of this reaction are also achieved when bFGF is bound to a heparin-Sepharose column, allowing the integration of this enzymatic step directly during purification of the recombinant extended forms of bFGF. Correspondence to: P. Caccia, Department of Biotechnology, Farmitalia Carlo Erba, Viale Bezzi 24, 20146 Milan, Italy. * Present address." Transgene, 11, rue de Molsheim, 67082 Strasbourg Cedex, France.
84
Fibroblast growth factor; FGF; Heparin; Pepsin A; Affinity chromatography; Controlled proteolysis; Aspartic proteases
Introduction
Basic fibroblast growth factor (bFGF) is a heparin-binding non-glycosylated single chain protein with a basic isoelectric point. (Folkman and Klagsbrun, 1987; Gospodarowicz et al., 1987; Lobb, 1988). In vitro bFGF is a potent mitogen for cells derived from tissues of mesodermal and neuroectodermal origin. In vivo bFGF was shown to induce neoangiogenesis and to promote wound healing. It was originally isolated from bovine pituitaries (Gospodarowicz et al., 1984) as a 146-amino acid polypeptide (Esch et al., 1985). However, subsequent modifications of the purification procedures by the substitution of neutral for acidic extraction and the addition of protease inhibitors yielded N-terminal extended forms of the protein. The nucleotide sequences of cDNA clones predicted a primary translation polypeptide of 155 amino acids (Abraham et al., 1986a, b; Fig. 1) but longer forms were also detected (Ueno et al., 1986; Klagsbrun et al., 1987). Since the availability of cDNA clones (Abraham et al., 1986a, b) for bovine and human bFGF, several groups obtained recombinant active forms of this growth factor (Iwane et al., 1987; Squires et al., 1988; Fox et al., 1988; Barr et al., 1988). Expression of bFGF in Escherichia coli or in Saccharomyces cerevisiae led to mixtures of two or more polypeptides with N-terminal microheterogeneity (Fox et al., 1988; Barr et al., 1988). We previously developed efficient expression systems in E. coli for the 146and 155-amino acid forms (Bergonzoni et al., 1990; Isacchi et al., 1990) which yielded mixtures of 146/145 and of 154/153 respectively. Although it was reported that limited truncations of the amino terminal end of bFGF did not significantly Human 5 Met Ala Ala GI7 Set 25 Phe Pro Pro Gly His 45 Lea Arg Ile His Pro 65 Lys Leu Gln Leu Gln 85 Arg Tyr Leu Ala Met 105 Cys Phe Phe Phe Glu 125 Thr Ser Trp Tyr Val 145 Pro Gly Gln Lys Ala
basic F • F
I0 15 Ile Thr Thr Leu Pro Ala Leu Pro Glu Asp 30 35 Phe Lys Asp Pro Lys Arg Leu Tyr Cys Lys 50 55 Asp Gly Arg Val Asp Gly Val Arg Glu Lye 70 75 Ala Glu Glu Arg Gly Val Val Ser Ile Lys 90 95 Lys Glu Asp Gly Arg Leu Leu Ala Set Lys 1 10 115 Arg Leu G1u Ser Ann Asn Tyr Ann Thr Tyr 13 0 135 Ala Leu Lys Arg Thr Gly Gln Tyr Ly8 Leu 150 155 Ile Leu Phe Leu Pro Mot Ser Ala Ly6 Ser
20 Gly Gly Ser Gly Ala 40 Ann Gly Gly Phe Phe 60 Set Asp Pro His Ile 80 Gly Val Cys Ala Ann 100 Cys Val Thr Asp Glu 120 Arg Ser Arg Lys Tyr 140 Gly Ser Lys Thr Gly
Fig. 1. Primary sequence of 155-amino acid form of bFGF. Underlined amino acids were removed by controlled enzymatic cleavage with pepsin A.
85 affect its biological activity (Klagsbrun et al., 1987; Barr et al., 1988), we investigated several approaches to obtain homogeneous preparations of this polypeptide for a rigorous pharmaceutical development of the growth factor as wound healing agent. Here, we present a specific enzymatic hydrolysis process which resulted in the quantitative isolation of a single, homogeneous 146-amino acid form of bFGF starting from a mixture of the recombinant 154/153-amino acid forms.
Material and Methods
Proteolytic treatment of soluble complexes bFGF-heparin and bFGF-heparan sulfate in solution 1.6 mg of 154/153-amino acid form of bFGF were complexed to 1.6 mg of heparin or heparan sulfate in 10 mM citrate-phosphate buffer pH 4.0, 10°C. After 10 min incubation time 160/xg of pepsin A (Sigma P-6887, 3200-4500 units per mg protein) were added to the solution. Samples of the reaction medium were taken at different times and directly loaded onto a heparin-Sepharose column preequilibrated with 10 mM phosphate buffer pH 8.0/0.5 M NaC1 at 4°C. The column was then washed extensively with the same buffer and the bFGF was eluted with 3 M NaC1 in 10 mM phosphate buffer pH 8.0. The pooled fractions were desalted on a Sephadex G25 column previously equilibrated in 10 mM phosphate buffer pH 6.0 and subjected to the SDS-PAGE analysis. Using heparan sulfate as protective agent, 146-bFGF was obtained in quantitative yield after 1 h. Pepsin treatment of bFGF-heparin complex led to the same quantitative yields just in 6.5 h incubation time.
Pepsin A treatment of bFGF bound to a heparin-Sepharose column 50 mg of bFGF (154/153) were loaded at a flow rate of 2.5 ml min -I on a heparin-Sepharose column (Pharmacia) (2.6 x 22.5 cm) previously equilibrated in 25 mM citrate buffer pH 4.0/0.5 M NaC1 at 4°C. A solution of 680 U ml-~ of pepsin A (Sigma) in 10 mM citrate-phosphate buffer was continuously recycled through the column at 2.5 ml min-~ for 3 h at 4°C. The column was then washed extensively with a 25 mM phosphate buffer pH 8.0/0.5 M NaCI to inactivate and to eliminate the enzyme. The bFGF was step-eluted with 3 M NaCI in 25 mM phosphate buffer pH 8.0. The bFGF-containing fractions were pooled, concentrated by ultrafiltration on Amicon PM 10 membrane and desalted on a Sephadex G25 column (Pharmacia) previously equilibrated in 10 mM phosphate buffer pH 6.0.
N-terminal sequence analysis Automated N-terminal sequence analysis was performed on a Model 477A Pulsed Liquid Phase Sequencer (Applied Biosystems, CA, U.S.A.) with on-line
86 Model 120A PTH-Analyzer. Normal-1 program with little modifications was used. All sequencing materials and reagents were purchased from Applied Biosystems.
C-terminal sequence analysis Time-course study of carboxypeptidase P digestion of bFGF was performed at room temperature in 10 mM sodium acetate buffer pH 3.8/0.05% Brij-35, using an enzyme to substrate ratio of approximately 1:100 (w/w) (Lu et al., 1988). Digestions of 0.5-1 nmol of protein were carried out for 2 h with 0.1-0.2 /~g of CpP (Boehringer); N-leucine was added as an internal standard. At different times 10/100 p,l aliquots were withdrawn and subjected to amino acid analysis after automated derivatization with PITC on a Model 420A Derivatizer (Applied Biosystems) and subsequently injected into a RP-HPLC with on-line Model 130A analyzer. Separation of the derivatized PTC-AAs was achieved on a PTC-C18 column ( P / N 0711-0204, 220 x 2.1 mm, 5/~, Applied Biosystems).
Bioassays An endothelial cell-strain derived from bovine aorta (BAEC) was used to study the proliferative response induced by bFGF. Cells were plated at 2500 cells per well in 96-well microtitre plates in complete medium consisting of Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 13% fetal bovine serum (FBS) (Gibco U.K.). After attachment, complete medium was replaced with experimental media consisting of DMEM supplemented with 0.5% FBS, 0.1% bovine serum albumin (BSA) (Sigma U.S.A.) and the desired concentrations of bFGF (Erbamont). The cultures were incubated for 3 d at which time they were fixed with formalin and stained with 0.5% crystal violet. After staining wells were thoroughly washed to remove unincorporated dye. Methanol (95%; 0.1 ml per well) was added to each well to extract the dye to an extent proportional to the amount of cells grown per well. Plates were transferred for automatic reading to a spectrophotometric microplate reader equipped with a 540 nm filter. For the synthesis of plasminogen activator, BAEC (3 × 104 cells in 0.2 ml per well) were seeded in 96-well microtitre plates in complete growth medium that was replaced after attachment with DMEM supplemented with 0.5% FBS, 0.1% BSA and the test concentrations of bFGF. After incubating for 28 h, cultures were washed and cells were lysed with a solution containing 0.5% Triton X-100. Aliquots of the cell-lysates were assayed for plasminogen activator activity using a chromogenic substrate (Spectrozyme PL) and plasminogen (both reagents from American Diagnostica Inc.) for the amidolytic assay. Results
Expression of recombinant bFGF The natural methionine residue present at the NH2-end of the molecule (Fig. 1) allowed the direct expression of the nucleotide sequence of the 155-amino acid
87
form in E. coli. The expression of the 146-amino acid form starting with a proline required the insertion of an additional methionine residue just upstream to the Pro residue. In both cases, N-terminal sequence analysis of the purified recombinant bFGF revealed the presence of double sequences. In the case of the 155-amino acid form, the product consisted of a constant equal mixture of the 154- and the 153-amino acid forms starting respectively with NH2-Ala-Ala-Gly-Ser-etc. and NH2-Ala-Gly-Ser-etc. Thus, the initial methionine was efficiently removed and also about 50% of the isolated bFGF lost the subsequent alanine residue. A similar result was obtained when the 146-amino acid form was expressed in E. coli obtaining a mixture of the two 146- and 145-amino acid forms. Even in this case, the initial methionine was efficiently removed but also the following residue was partially lost. Interestingly, by varying the conditions of the bioprocess (time and temperature), it was possible to modify the proportion of these forms indicating that this degradation occurred within the bacterial cell and it was probably due to some bacterial aminopeptidase (unpublished results).
Controlled enzymatic processing of bFGF The purified recombinant 154/153 mixture was incubated with two different aspartic proteases: pepsin A and cathepsin D. Aliquots of the reaction mixture were taken at various time intervals and submitted to SDS-PAGE analysis showing that in the absence of any protecting agent, bFGF was quickly digested into small peptides. On the contrary, the treatment of bFGF (154/153) with pepsin A (10 : 1, w/w; pH 4.0; 10°C) in the presence of heparin or heparan sulfate (1:1, w/w), resulted in the progressive and complete conversion of the 154/153-amino acid forms to a lower molecular weight form which comigrated with our 146/145-amino acid form standard (Fig. 2). After electroblotting on a PDVF membrane the protein was submitted to automated N-terminal sequence analysis on a pulsed liquid phase sequencer. The first three cycles resulted in a single homogeneous sequence: Pro-Ala-Leu that corresponds to the intact N-terminal end of the known 146-amino acid form. No other sequence was detected showing that, despite the
q"~
~U~' ~ B l ~
~
~
. 1
2
3
4
. 5
. 6
~ .
~
7
~~ D - / bFGF 146 • ~ --Lysozyme 8
9
Fig. 2. SDS-PAGE analysis on PhastGel High Density (Pharmacia) using the Pharmacia Phast System stained with Coomassie brillant blue R-250. The 154-153 bFGF/hcparin complex 1: 1 (w/w) digested with pepsin A was analysed after 1 rain (lane 1), 2 h (lane 2), 4 h (lane 3), 6 h (lane 4). The 154-153 bFGF/heparan sulfate complex 1:1 (w/w) was treated with pepsin A. Aliquots of the digestion mixture w e r e analysed after 1 rain (lane 5), 20 rain (lane 6), 40 rain (lane 7), 60 rain (lane 8). Molecular weight standards (lane 9): ]54/153-aa bFGF (1"1kDa), 146-aabFGF (16.2 kDa), lysozymc (14.4 kDa).
88 presence of three Leu-Pro sites on the b F G F molecule, when the elongated forms of b F G F were complexed to heparin or to heparan sulfate, the digestion with pepsin A cleaved specifically the Leu9-Prol0 peptide bond. Other glycosaminoglycans (chondroitin and dermatan sulfate) did not protect efficiently the b F G F molecule and many fragments were generated by incubation with pepsin A, cathepsin D or chymotrypsin (data not shown). A controlled proteolytic cleavage of the "heparin-protected" NH2-extended b F G F molecule, was also achieved after digestion with cathepsin D, although a longer incubation time was required for this proteolytic reaction, possibly due to the lower activity of the enzyme preparation used (data not shown). When chymotrypsin was added under similar conditions but at pH 7.5, a single polypeptide of about 14,000 Da was detected after gel electrophoresis of the incubation mixture showing no evidence of the presence of a 146-amino acids form.
Large scale enzymatic processing of bFGF on heparin-Sepharose column The results obtained in solution and in small batch reactions were verified on a larger scale process with the 154/153 mixture of elongated b F G F bound to a heparin-Sepharose column. Accordingly, 50 mg of b F G F (154/153) were loaded on a heparin-Sepharose cohl.,r..r, r,,~,,iously equilibrated in 10 mM citrate-phosphate buffer pH 4.0. ,-k solution of pepsin A was recycled continuously through the column for 3 h at 4°C. Thereafter the column was washed with a phosphate buffer
kDa m4 5 I ~2/o ~18 ~16
3
! 2 3 4 Fig. 3. SDS-PAGE analysis on 15% polyacrylamidegel stained with Coomassie Blue R-250. The purified homogeneous 146-aa form of recombinant bFGF after pepsin A treatment on heparin-Sepharose column (lane 3), the 146/145-amino acid form of bFGF (lane 2) and the 154/153-amino acid form of bFGF (lane 1) were loaded on gel with low molecularweight standards (lane 4).
89
04
¸
37~I
0.3
ff
J
II
U
o
Cq e,I O2
3211,
< 0.1
InJe~:tlon
I I I
"5
I I I
I0 I
I
I
)
15 I
i
i
i
i
20 i
i
i
)
i
25 i
i
i
i
i
30 i
i
i
)
RT (mini Fig. 4. Reverse-phase HPLC analysis of the homogeneous 146-aa form of bFGF. The protein eluted from heparin-Sepharose column after controlled proteolytic cleavage with pepsin A was subjected to an HPLC column (Vydac CTR 214 C4; 0.46x25 cm) and eluted with a linear gradient of 32-37% acetonitrile in 0.1% TFA at a flow rate of 1 ml min -I.
at alkaline pH both to inactivate and to eliminate the enzyme. The resulting polypeptide was eluted from the column with 3 M NaCI in 25 mM phosphate buffer at pH 8.0. The bFGF-containing fractions were collected and desalted on a Sephadex G25 column. The pooled fractions analyzed on SDS-PAGE showed a single band with a molecular weight corresponding to that of the standard 146/145 bFGF form (Fig. 3). Reverse-phase HPLC analysis resulted also in a single peak (Fig. 4). N-terminal sequence analysis yielded a single sequence corresponding to the correct, intact N-terminal end of the 146-amino acid form, i.e. Pro-Ala-Leu-. C-terminal sequence analysis showed the expected carboxyterminal end of the bFGF molecule, i.e. -Ala-Lys-Ser. Thus, also when bFGF was bound to the heparin-Sepharose resin, pepsin A was capable of cleaving specifically the molecule at the Leug-Pro]0 bond generating a homogeneous 146-amino acid form.
In vitro bioassays of the bFGF forms The biological activity of the mixture of the 154/153-amino acid form was compared to that of the homogeneous 146-amino acid form obtained by the proteolytic process described. Two activities, the induction of a proliferative response and the synthesis of plasminogen activator, were studied in bovine aortic endothelial ceils (BAEC). Both assays confirmed the in vitro biological equivalence
90
®
100 90,
*
*s
.....
"'0
50, 40,
.d
30,
[..
'°O'I
....... -
,01
I---O-- bFGF146aa
,I
1
10
100
1000
I0000
ng ml" l
"
®
_ta=.,.,---n
x
_•
O' _'
bFGF154/153
aa
o t
I0
I O0
1000
10000
ng ml" I
Fig. 5. In vitro biological activity of bFGF. A. Mitogenic effect of the 154/ 153-aa form • - • and of the homogeneous 146-aa form o . . . . o on BAEC cells. The concentration of bFGF was plotted against the percent of maximal stimulation obtained with the 154/153-aa form. Values represent the mean of 4 determinations; standard deviation was always less than 20%. B. Induction of the tissue plasminogen activator synthesis (PA) in BAEC. Cells were incubated for 28 h with the different forms of bFGF (154/153-aa • - - • ; 146-aa o . . . . o). Cell-lysates were assayed for PA activity using plasminogen and a chromogenic substrate for the amidolytic assay. Data are the mean of four replicates; standard deviations were always less than 20%.
o f t h e 1 5 4 / 1 5 3 - as c o m p a r e d e n z y m a t i c p r o c e s s (Fig. 5 A , B).
to
the
146-amino
acid
form
obtained
by the
Discussion
b F G F w a s o r i g i n a l l y i s o l a t e d f r o m b o v i n e b r a i n as a p o l y p e p t i d e o f 146 a m i n o a c i d s ( G o s p o d a r o w i c z e t al., 1984). L a t e r , it b e c a m e c l e a r t h a t t h i s f o r m w a s
91 generated by proteolytic processing of N-terminal extended forms. Indeed, the addition of protease inhibitors such as pepstatin and variations of the pH of the extraction procedure led to the isolation of the 154-amino acid form (Ueno et al., 1986; Klagsbrun et al., 1987). In addition, the nucleotide sequence of human and bovine cDNA clones would predict the expression of the 155-amino acid form as shown in Fig. 1 (Abraham et al., 1986a, b). Based on these observations, it was thus suggested that acidic extraction of bFGF led to the activation of aspartic protease(s) acting at the N-terminal end of the protein generating a truncated form of 146 amino acids (Klagsbrun et al., 1987). In this article we describe an enzymatic process able to generate a homogeneous 146-amino acid form of bFGF which may be considered as a reproduction of the proteolytic conversion observed during extraction and purification from tissues. The presence of specific glycosaminoglycans, as potential protecting agents, was found necessary to avoid further cleavages of the polypeptidic chain of bFGF. Previous findings already demonstrated that heparin and heparan sulfate were able to efficiently protect bFGF from trypsin and partially from chymotrypsin (Sommer and Rifkin, 1989; Rosengart et al., 1988). Three Leu-Pro sites are present in the 154/153 bFGF forms. However, it is quite interesting to note that only the site close to the NH2-end domain was sensitive to proteolytic action in the conditions used in our experiments. This differential sensitivity suggests that the region of the protein close to the other two Leu-Pro sites should be somehow important f o r the interactions with such glycosaminoglycans. Specificity of the protective action of heparin and heparan sulfate is another important point. Indeed, other glycosaminoglycans such as chondroitin and dermatan sulfate were unable to protect efficiently bFGF from proteolytic cleavages, possibly due to a lower affinity for bFGF. Finally, since our controlled proteolytic cleavage works efficiently even when bFGF is immobilized on a heparin-Sepharose column, it could be easily integrated in the purification process of a recombinant NH2-extended form and allow the large-scale production of a homogeneous and fully active 146-amino acid form.
Acknowledgements We would like to acknowledge the excellent technical expertise of O. Cletini and B. Valsasina for their help in the analytical part of the work and F. Roletto and C. Cristiani for the biological assays.
References Abraham, J., Mergia, A., Whang,J., Tumolo,A., Friedman,J., Hjerrild, K., Gospodarowicz,D. and Fiddes, J. (1986a) Nucleotide sequence of a bovine clone encoding the angiogenicprotein, basic fibroblast growthfactor. Science 233, 545-548.
92 Abraham, J., Whang, J., Tumolo, A., Friedman, J., Gospodarowicz, D. and Fiddes, J. (1986b) Human basic fibrohlast growth factor: nucleotide sequence and genomic organization. Eur. Mol. Biol. Org. J. 5, 2523-2528. Barr, P., Cousens, L., Lee-Ng, C., Medina-Selby, A., Masiarz, F., Hallewell, R., Chamberlain, S., Bradley, J., Lee, D., Steimer, K., Poulter, L., Burlingames, A., Esch, F. and Baird, A. (1988) Expression and processing of biologically active fibroblast growth factors in the yeast Saccharomyces cerevisiae. J. Biol. Chem. 263, 16471-16478. Bcrgonzoni, L., Isacchi, A., Sarmientos, P. and Cauet, G. (1990) New derivatives of human/bovine basic fibroblast growth factor. European Patent Application No. 0363675. Esch, F., Baird, A., Ling, N., Ueno, N., Hill, F., Denoroy, L, Klepper, R., Gospodarowicz, D., Bohlen, P. and Guillemin, R. (1985) Primary structure of bovine pituitary basic fibroblast growth factor (FGF) and comparison with the amino-terminal sequence of bovine brain acidic FGF. Proc. Natl. Acad. Sci. U.S.A. 82, 6507-6511. Folkman, J. and Klagsbrun, M. (1987) Angiogenic factors. Science 235, 442-447. Fox, G., Schiffer, S., Rohde, M., Tsai, L., Banks, A. and Arakawa, T. (1988) Production, biological activity and structure of recombinant basic fibroblast growth factor and an analog with cysteine replaced by serine. J. Biol. Chem. 263, 18452-18458. Gospodarowicz, D., Cheng, J., Lui, G., Baird, A. and Bohlen, P. (1984) Isolation of brain fibroblast growth factor by hcparin-Sepharose affinity chromatography: identity with pituitary fibroblast growth factor. Proc. Natl. Acad. Sci. U.S.A. 81, 6963-6967. Gospodarowicz, D., Ferrara, N., Schweigerer, L. and Neufeld, G. (1987) Structural characterization and biological functions of flbroblast growth factor. Endocr. Rev. 8, 95-114. Isacchi, A., Caccia, P., Cauet, G., Bertolero, F., Bergonzoni, L., Roletto, F., Ziliotto, R., Garofano, L., Jacob, C., Sarmientos, P., Mazud, G., Carminati, P. and Roncucci, R. (1990) Human basic fibroblast growth factor. Production of a clinical grade recombinant protein from E. coli. Chimicaoggi 6, 72-74. Iwane, M., Kurokawa, T., Sasada, R., Seno, M., Nakagawa, S. and Igarashi, K. (1987) Expression of cDNA encoding human basic fibroblast growth factor in E. coli. Biochem. Biophys. Res. Commun. 146, 470-477. Klagsbrun, M., Smith, S., Sullivan, R., Shing, Y., Davidson, S., Smith, J. and Sasse, J. (1987) Multiple forms of basic fibroblast growth factor: Amino-terminal cleavages by tumor cell- and brain cell-derived acid proteinases. Proc. Natl. Acad. Sci. U.S.A. 84, 1839-1843. Lobb, R. (1988) Clinical applications of heparin-binding growth factors. Eur. J. Clin. Invest. 18, 321-326. Lu, S., Klein, M. and Lai, P. (1988) Narrow-bore High Performance Liquid Chromatography of Phenylthiocarbamyl amino acids and Carboxypeptidase P digestion for protein C-terminal sequence analysis. J. Chromatogr. 447, 351-364. Rosengart, T., Johnson, W., Friesel, R., Clark, R. and Maciag, T. (1988) Heparin protects HeparinBinding Growth Factor-I from proteolytic inactivation in vitro. Biochem. Biophys. Res. Commun. 152, 432-440. Sommer, A. and Rifkin, D. (1989) Interaction of heparin with human basic fibroblast growth factor: protection of the angiogenic protein from proteolytic degradation by a glycosaminoglycan. J. Cell. Physiol. 138, 215-220. Squires, C., Childs, J., Eisenberg, S., Polverini, P. and Sommer, A. (1988) Production and characterization of human basic fibroblast growth factor from Escherichia coli. J. Biol. Chem. 263, 16297-16302. Ueno, N., Baird, A., Esch, F., Ling, N. and Guillemin, R. (1986) Isolation of an amino-terminal extended form of basic fibroblast growth factor. Biochem. Biophys. Res. Commun, 138, 580-588.
83
BIOTEC 00674
Production of homogeneous basic fibroblast growth factor by specific enzymatic hydrolysis of larger microheterogeneous molecular forms D. B e t b e d e r 2, p. Caccia ~, G. Nitti ~, F. Bertolero t, p. Sarmientos 1, F. Paul 2, p. Monsan 2, G. Cauet ~'* and G. Mazu6 t I Department ofBiotechnology, Farmitalia Carlo Erba, Milan, Italy and 2 BioEurope, Toulouse, France
(Received 15 March, 1991; revision accepted 20 May 1991)
Summary The 146-amino acid form of basic fibroblast growth factor (bFGF) was expressed in Escherichia coli and purified by a two step process including ion exchange and heparin-Sepharose chromatographies. However, the resulting protein consisted of a mixture of 146- and 145-amino acid forms, indicating that, besides the initial methionine, also the following residue (proline) was removed from the N-terminus. The same phenomenon was observed when the 155-amino acid form, which is biologically equivalent to the shorter one, was expressed in E. coli. Taking into account the previously known data concerning the possible mechanism of cleavage of the extended forms of bFGF in vivo, we developed an efficient enzymatic process that allows the production of an homogeneous 146amino acid form from recombinant NH2-end extended forms. This process takes advantage of the protecting effect that heparin exerts on bFGF. Accordingly, when bFGF, complexed to heparin, is treated with pepsin A, an aspartic protease with a broad specificity, only the Leu9-Prot0 peptide bond is cleaved generating the 146-amino acid form. Quantitative yields of this reaction are also achieved when bFGF is bound to a heparin-Sepharose column, allowing the integration of this enzymatic step directly during purification of the recombinant extended forms of bFGF. Correspondence to: P. Caccia, Department of Biotechnology, Farmitalia Carlo Erba, Viale Bezzi 24, 20146 Milan, Italy. * Present address." Transgene, 11, rue de Molsheim, 67082 Strasbourg Cedex, France.
84
Fibroblast growth factor; FGF; Heparin; Pepsin A; Affinity chromatography; Controlled proteolysis; Aspartic proteases
Introduction
Basic fibroblast growth factor (bFGF) is a heparin-binding non-glycosylated single chain protein with a basic isoelectric point. (Folkman and Klagsbrun, 1987; Gospodarowicz et al., 1987; Lobb, 1988). In vitro bFGF is a potent mitogen for cells derived from tissues of mesodermal and neuroectodermal origin. In vivo bFGF was shown to induce neoangiogenesis and to promote wound healing. It was originally isolated from bovine pituitaries (Gospodarowicz et al., 1984) as a 146-amino acid polypeptide (Esch et al., 1985). However, subsequent modifications of the purification procedures by the substitution of neutral for acidic extraction and the addition of protease inhibitors yielded N-terminal extended forms of the protein. The nucleotide sequences of cDNA clones predicted a primary translation polypeptide of 155 amino acids (Abraham et al., 1986a, b; Fig. 1) but longer forms were also detected (Ueno et al., 1986; Klagsbrun et al., 1987). Since the availability of cDNA clones (Abraham et al., 1986a, b) for bovine and human bFGF, several groups obtained recombinant active forms of this growth factor (Iwane et al., 1987; Squires et al., 1988; Fox et al., 1988; Barr et al., 1988). Expression of bFGF in Escherichia coli or in Saccharomyces cerevisiae led to mixtures of two or more polypeptides with N-terminal microheterogeneity (Fox et al., 1988; Barr et al., 1988). We previously developed efficient expression systems in E. coli for the 146and 155-amino acid forms (Bergonzoni et al., 1990; Isacchi et al., 1990) which yielded mixtures of 146/145 and of 154/153 respectively. Although it was reported that limited truncations of the amino terminal end of bFGF did not significantly Human 5 Met Ala Ala GI7 Set 25 Phe Pro Pro Gly His 45 Lea Arg Ile His Pro 65 Lys Leu Gln Leu Gln 85 Arg Tyr Leu Ala Met 105 Cys Phe Phe Phe Glu 125 Thr Ser Trp Tyr Val 145 Pro Gly Gln Lys Ala
basic F • F
I0 15 Ile Thr Thr Leu Pro Ala Leu Pro Glu Asp 30 35 Phe Lys Asp Pro Lys Arg Leu Tyr Cys Lys 50 55 Asp Gly Arg Val Asp Gly Val Arg Glu Lye 70 75 Ala Glu Glu Arg Gly Val Val Ser Ile Lys 90 95 Lys Glu Asp Gly Arg Leu Leu Ala Set Lys 1 10 115 Arg Leu G1u Ser Ann Asn Tyr Ann Thr Tyr 13 0 135 Ala Leu Lys Arg Thr Gly Gln Tyr Ly8 Leu 150 155 Ile Leu Phe Leu Pro Mot Ser Ala Ly6 Ser
20 Gly Gly Ser Gly Ala 40 Ann Gly Gly Phe Phe 60 Set Asp Pro His Ile 80 Gly Val Cys Ala Ann 100 Cys Val Thr Asp Glu 120 Arg Ser Arg Lys Tyr 140 Gly Ser Lys Thr Gly
Fig. 1. Primary sequence of 155-amino acid form of bFGF. Underlined amino acids were removed by controlled enzymatic cleavage with pepsin A.
85 affect its biological activity (Klagsbrun et al., 1987; Barr et al., 1988), we investigated several approaches to obtain homogeneous preparations of this polypeptide for a rigorous pharmaceutical development of the growth factor as wound healing agent. Here, we present a specific enzymatic hydrolysis process which resulted in the quantitative isolation of a single, homogeneous 146-amino acid form of bFGF starting from a mixture of the recombinant 154/153-amino acid forms.
Material and Methods
Proteolytic treatment of soluble complexes bFGF-heparin and bFGF-heparan sulfate in solution 1.6 mg of 154/153-amino acid form of bFGF were complexed to 1.6 mg of heparin or heparan sulfate in 10 mM citrate-phosphate buffer pH 4.0, 10°C. After 10 min incubation time 160/xg of pepsin A (Sigma P-6887, 3200-4500 units per mg protein) were added to the solution. Samples of the reaction medium were taken at different times and directly loaded onto a heparin-Sepharose column preequilibrated with 10 mM phosphate buffer pH 8.0/0.5 M NaC1 at 4°C. The column was then washed extensively with the same buffer and the bFGF was eluted with 3 M NaC1 in 10 mM phosphate buffer pH 8.0. The pooled fractions were desalted on a Sephadex G25 column previously equilibrated in 10 mM phosphate buffer pH 6.0 and subjected to the SDS-PAGE analysis. Using heparan sulfate as protective agent, 146-bFGF was obtained in quantitative yield after 1 h. Pepsin treatment of bFGF-heparin complex led to the same quantitative yields just in 6.5 h incubation time.
Pepsin A treatment of bFGF bound to a heparin-Sepharose column 50 mg of bFGF (154/153) were loaded at a flow rate of 2.5 ml min -I on a heparin-Sepharose column (Pharmacia) (2.6 x 22.5 cm) previously equilibrated in 25 mM citrate buffer pH 4.0/0.5 M NaC1 at 4°C. A solution of 680 U ml-~ of pepsin A (Sigma) in 10 mM citrate-phosphate buffer was continuously recycled through the column at 2.5 ml min-~ for 3 h at 4°C. The column was then washed extensively with a 25 mM phosphate buffer pH 8.0/0.5 M NaCI to inactivate and to eliminate the enzyme. The bFGF was step-eluted with 3 M NaCI in 25 mM phosphate buffer pH 8.0. The bFGF-containing fractions were pooled, concentrated by ultrafiltration on Amicon PM 10 membrane and desalted on a Sephadex G25 column (Pharmacia) previously equilibrated in 10 mM phosphate buffer pH 6.0.
N-terminal sequence analysis Automated N-terminal sequence analysis was performed on a Model 477A Pulsed Liquid Phase Sequencer (Applied Biosystems, CA, U.S.A.) with on-line
86 Model 120A PTH-Analyzer. Normal-1 program with little modifications was used. All sequencing materials and reagents were purchased from Applied Biosystems.
C-terminal sequence analysis Time-course study of carboxypeptidase P digestion of bFGF was performed at room temperature in 10 mM sodium acetate buffer pH 3.8/0.05% Brij-35, using an enzyme to substrate ratio of approximately 1:100 (w/w) (Lu et al., 1988). Digestions of 0.5-1 nmol of protein were carried out for 2 h with 0.1-0.2 /~g of CpP (Boehringer); N-leucine was added as an internal standard. At different times 10/100 p,l aliquots were withdrawn and subjected to amino acid analysis after automated derivatization with PITC on a Model 420A Derivatizer (Applied Biosystems) and subsequently injected into a RP-HPLC with on-line Model 130A analyzer. Separation of the derivatized PTC-AAs was achieved on a PTC-C18 column ( P / N 0711-0204, 220 x 2.1 mm, 5/~, Applied Biosystems).
Bioassays An endothelial cell-strain derived from bovine aorta (BAEC) was used to study the proliferative response induced by bFGF. Cells were plated at 2500 cells per well in 96-well microtitre plates in complete medium consisting of Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 13% fetal bovine serum (FBS) (Gibco U.K.). After attachment, complete medium was replaced with experimental media consisting of DMEM supplemented with 0.5% FBS, 0.1% bovine serum albumin (BSA) (Sigma U.S.A.) and the desired concentrations of bFGF (Erbamont). The cultures were incubated for 3 d at which time they were fixed with formalin and stained with 0.5% crystal violet. After staining wells were thoroughly washed to remove unincorporated dye. Methanol (95%; 0.1 ml per well) was added to each well to extract the dye to an extent proportional to the amount of cells grown per well. Plates were transferred for automatic reading to a spectrophotometric microplate reader equipped with a 540 nm filter. For the synthesis of plasminogen activator, BAEC (3 × 104 cells in 0.2 ml per well) were seeded in 96-well microtitre plates in complete growth medium that was replaced after attachment with DMEM supplemented with 0.5% FBS, 0.1% BSA and the test concentrations of bFGF. After incubating for 28 h, cultures were washed and cells were lysed with a solution containing 0.5% Triton X-100. Aliquots of the cell-lysates were assayed for plasminogen activator activity using a chromogenic substrate (Spectrozyme PL) and plasminogen (both reagents from American Diagnostica Inc.) for the amidolytic assay. Results
Expression of recombinant bFGF The natural methionine residue present at the NH2-end of the molecule (Fig. 1) allowed the direct expression of the nucleotide sequence of the 155-amino acid
87
form in E. coli. The expression of the 146-amino acid form starting with a proline required the insertion of an additional methionine residue just upstream to the Pro residue. In both cases, N-terminal sequence analysis of the purified recombinant bFGF revealed the presence of double sequences. In the case of the 155-amino acid form, the product consisted of a constant equal mixture of the 154- and the 153-amino acid forms starting respectively with NH2-Ala-Ala-Gly-Ser-etc. and NH2-Ala-Gly-Ser-etc. Thus, the initial methionine was efficiently removed and also about 50% of the isolated bFGF lost the subsequent alanine residue. A similar result was obtained when the 146-amino acid form was expressed in E. coli obtaining a mixture of the two 146- and 145-amino acid forms. Even in this case, the initial methionine was efficiently removed but also the following residue was partially lost. Interestingly, by varying the conditions of the bioprocess (time and temperature), it was possible to modify the proportion of these forms indicating that this degradation occurred within the bacterial cell and it was probably due to some bacterial aminopeptidase (unpublished results).
Controlled enzymatic processing of bFGF The purified recombinant 154/153 mixture was incubated with two different aspartic proteases: pepsin A and cathepsin D. Aliquots of the reaction mixture were taken at various time intervals and submitted to SDS-PAGE analysis showing that in the absence of any protecting agent, bFGF was quickly digested into small peptides. On the contrary, the treatment of bFGF (154/153) with pepsin A (10 : 1, w/w; pH 4.0; 10°C) in the presence of heparin or heparan sulfate (1:1, w/w), resulted in the progressive and complete conversion of the 154/153-amino acid forms to a lower molecular weight form which comigrated with our 146/145-amino acid form standard (Fig. 2). After electroblotting on a PDVF membrane the protein was submitted to automated N-terminal sequence analysis on a pulsed liquid phase sequencer. The first three cycles resulted in a single homogeneous sequence: Pro-Ala-Leu that corresponds to the intact N-terminal end of the known 146-amino acid form. No other sequence was detected showing that, despite the
q"~
~U~' ~ B l ~
~
~
. 1
2
3
4
. 5
. 6
~ .
~
7
~~ D - / bFGF 146 • ~ --Lysozyme 8
9
Fig. 2. SDS-PAGE analysis on PhastGel High Density (Pharmacia) using the Pharmacia Phast System stained with Coomassie brillant blue R-250. The 154-153 bFGF/hcparin complex 1: 1 (w/w) digested with pepsin A was analysed after 1 rain (lane 1), 2 h (lane 2), 4 h (lane 3), 6 h (lane 4). The 154-153 bFGF/heparan sulfate complex 1:1 (w/w) was treated with pepsin A. Aliquots of the digestion mixture w e r e analysed after 1 rain (lane 5), 20 rain (lane 6), 40 rain (lane 7), 60 rain (lane 8). Molecular weight standards (lane 9): ]54/153-aa bFGF (1"1kDa), 146-aabFGF (16.2 kDa), lysozymc (14.4 kDa).
88 presence of three Leu-Pro sites on the b F G F molecule, when the elongated forms of b F G F were complexed to heparin or to heparan sulfate, the digestion with pepsin A cleaved specifically the Leu9-Prol0 peptide bond. Other glycosaminoglycans (chondroitin and dermatan sulfate) did not protect efficiently the b F G F molecule and many fragments were generated by incubation with pepsin A, cathepsin D or chymotrypsin (data not shown). A controlled proteolytic cleavage of the "heparin-protected" NH2-extended b F G F molecule, was also achieved after digestion with cathepsin D, although a longer incubation time was required for this proteolytic reaction, possibly due to the lower activity of the enzyme preparation used (data not shown). When chymotrypsin was added under similar conditions but at pH 7.5, a single polypeptide of about 14,000 Da was detected after gel electrophoresis of the incubation mixture showing no evidence of the presence of a 146-amino acids form.
Large scale enzymatic processing of bFGF on heparin-Sepharose column The results obtained in solution and in small batch reactions were verified on a larger scale process with the 154/153 mixture of elongated b F G F bound to a heparin-Sepharose column. Accordingly, 50 mg of b F G F (154/153) were loaded on a heparin-Sepharose cohl.,r..r, r,,~,,iously equilibrated in 10 mM citrate-phosphate buffer pH 4.0. ,-k solution of pepsin A was recycled continuously through the column for 3 h at 4°C. Thereafter the column was washed with a phosphate buffer
kDa m4 5 I ~2/o ~18 ~16
3
! 2 3 4 Fig. 3. SDS-PAGE analysis on 15% polyacrylamidegel stained with Coomassie Blue R-250. The purified homogeneous 146-aa form of recombinant bFGF after pepsin A treatment on heparin-Sepharose column (lane 3), the 146/145-amino acid form of bFGF (lane 2) and the 154/153-amino acid form of bFGF (lane 1) were loaded on gel with low molecularweight standards (lane 4).
89
04
¸
37~I
0.3
ff
J
II
U
o
Cq e,I O2
3211,
< 0.1
InJe~:tlon
I I I
"5
I I I
I0 I
I
I
)
15 I
i
i
i
i
20 i
i
i
)
i
25 i
i
i
i
i
30 i
i
i
)
RT (mini Fig. 4. Reverse-phase HPLC analysis of the homogeneous 146-aa form of bFGF. The protein eluted from heparin-Sepharose column after controlled proteolytic cleavage with pepsin A was subjected to an HPLC column (Vydac CTR 214 C4; 0.46x25 cm) and eluted with a linear gradient of 32-37% acetonitrile in 0.1% TFA at a flow rate of 1 ml min -I.
at alkaline pH both to inactivate and to eliminate the enzyme. The resulting polypeptide was eluted from the column with 3 M NaCI in 25 mM phosphate buffer at pH 8.0. The bFGF-containing fractions were collected and desalted on a Sephadex G25 column. The pooled fractions analyzed on SDS-PAGE showed a single band with a molecular weight corresponding to that of the standard 146/145 bFGF form (Fig. 3). Reverse-phase HPLC analysis resulted also in a single peak (Fig. 4). N-terminal sequence analysis yielded a single sequence corresponding to the correct, intact N-terminal end of the 146-amino acid form, i.e. Pro-Ala-Leu-. C-terminal sequence analysis showed the expected carboxyterminal end of the bFGF molecule, i.e. -Ala-Lys-Ser. Thus, also when bFGF was bound to the heparin-Sepharose resin, pepsin A was capable of cleaving specifically the molecule at the Leug-Pro]0 bond generating a homogeneous 146-amino acid form.
In vitro bioassays of the bFGF forms The biological activity of the mixture of the 154/153-amino acid form was compared to that of the homogeneous 146-amino acid form obtained by the proteolytic process described. Two activities, the induction of a proliferative response and the synthesis of plasminogen activator, were studied in bovine aortic endothelial ceils (BAEC). Both assays confirmed the in vitro biological equivalence
90
®
100 90,
*
*s
.....
"'0
50, 40,
.d
30,
[..
'°O'I
....... -
,01
I---O-- bFGF146aa
,I
1
10
100
1000
I0000
ng ml" l
"
®
_ta=.,.,---n
x
_•
O' _'
bFGF154/153
aa
o t
I0
I O0
1000
10000
ng ml" I
Fig. 5. In vitro biological activity of bFGF. A. Mitogenic effect of the 154/ 153-aa form • - • and of the homogeneous 146-aa form o . . . . o on BAEC cells. The concentration of bFGF was plotted against the percent of maximal stimulation obtained with the 154/153-aa form. Values represent the mean of 4 determinations; standard deviation was always less than 20%. B. Induction of the tissue plasminogen activator synthesis (PA) in BAEC. Cells were incubated for 28 h with the different forms of bFGF (154/153-aa • - - • ; 146-aa o . . . . o). Cell-lysates were assayed for PA activity using plasminogen and a chromogenic substrate for the amidolytic assay. Data are the mean of four replicates; standard deviations were always less than 20%.
o f t h e 1 5 4 / 1 5 3 - as c o m p a r e d e n z y m a t i c p r o c e s s (Fig. 5 A , B).
to
the
146-amino
acid
form
obtained
by the
Discussion
b F G F w a s o r i g i n a l l y i s o l a t e d f r o m b o v i n e b r a i n as a p o l y p e p t i d e o f 146 a m i n o a c i d s ( G o s p o d a r o w i c z e t al., 1984). L a t e r , it b e c a m e c l e a r t h a t t h i s f o r m w a s
91 generated by proteolytic processing of N-terminal extended forms. Indeed, the addition of protease inhibitors such as pepstatin and variations of the pH of the extraction procedure led to the isolation of the 154-amino acid form (Ueno et al., 1986; Klagsbrun et al., 1987). In addition, the nucleotide sequence of human and bovine cDNA clones would predict the expression of the 155-amino acid form as shown in Fig. 1 (Abraham et al., 1986a, b). Based on these observations, it was thus suggested that acidic extraction of bFGF led to the activation of aspartic protease(s) acting at the N-terminal end of the protein generating a truncated form of 146 amino acids (Klagsbrun et al., 1987). In this article we describe an enzymatic process able to generate a homogeneous 146-amino acid form of bFGF which may be considered as a reproduction of the proteolytic conversion observed during extraction and purification from tissues. The presence of specific glycosaminoglycans, as potential protecting agents, was found necessary to avoid further cleavages of the polypeptidic chain of bFGF. Previous findings already demonstrated that heparin and heparan sulfate were able to efficiently protect bFGF from trypsin and partially from chymotrypsin (Sommer and Rifkin, 1989; Rosengart et al., 1988). Three Leu-Pro sites are present in the 154/153 bFGF forms. However, it is quite interesting to note that only the site close to the NH2-end domain was sensitive to proteolytic action in the conditions used in our experiments. This differential sensitivity suggests that the region of the protein close to the other two Leu-Pro sites should be somehow important f o r the interactions with such glycosaminoglycans. Specificity of the protective action of heparin and heparan sulfate is another important point. Indeed, other glycosaminoglycans such as chondroitin and dermatan sulfate were unable to protect efficiently bFGF from proteolytic cleavages, possibly due to a lower affinity for bFGF. Finally, since our controlled proteolytic cleavage works efficiently even when bFGF is immobilized on a heparin-Sepharose column, it could be easily integrated in the purification process of a recombinant NH2-extended form and allow the large-scale production of a homogeneous and fully active 146-amino acid form.
Acknowledgements We would like to acknowledge the excellent technical expertise of O. Cletini and B. Valsasina for their help in the analytical part of the work and F. Roletto and C. Cristiani for the biological assays.
References Abraham, J., Mergia, A., Whang,J., Tumolo,A., Friedman,J., Hjerrild, K., Gospodarowicz,D. and Fiddes, J. (1986a) Nucleotide sequence of a bovine clone encoding the angiogenicprotein, basic fibroblast growthfactor. Science 233, 545-548.
92 Abraham, J., Whang, J., Tumolo, A., Friedman, J., Gospodarowicz, D. and Fiddes, J. (1986b) Human basic fibrohlast growth factor: nucleotide sequence and genomic organization. Eur. Mol. Biol. Org. J. 5, 2523-2528. Barr, P., Cousens, L., Lee-Ng, C., Medina-Selby, A., Masiarz, F., Hallewell, R., Chamberlain, S., Bradley, J., Lee, D., Steimer, K., Poulter, L., Burlingames, A., Esch, F. and Baird, A. (1988) Expression and processing of biologically active fibroblast growth factors in the yeast Saccharomyces cerevisiae. J. Biol. Chem. 263, 16471-16478. Bcrgonzoni, L., Isacchi, A., Sarmientos, P. and Cauet, G. (1990) New derivatives of human/bovine basic fibroblast growth factor. European Patent Application No. 0363675. Esch, F., Baird, A., Ling, N., Ueno, N., Hill, F., Denoroy, L, Klepper, R., Gospodarowicz, D., Bohlen, P. and Guillemin, R. (1985) Primary structure of bovine pituitary basic fibroblast growth factor (FGF) and comparison with the amino-terminal sequence of bovine brain acidic FGF. Proc. Natl. Acad. Sci. U.S.A. 82, 6507-6511. Folkman, J. and Klagsbrun, M. (1987) Angiogenic factors. Science 235, 442-447. Fox, G., Schiffer, S., Rohde, M., Tsai, L., Banks, A. and Arakawa, T. (1988) Production, biological activity and structure of recombinant basic fibroblast growth factor and an analog with cysteine replaced by serine. J. Biol. Chem. 263, 18452-18458. Gospodarowicz, D., Cheng, J., Lui, G., Baird, A. and Bohlen, P. (1984) Isolation of brain fibroblast growth factor by hcparin-Sepharose affinity chromatography: identity with pituitary fibroblast growth factor. Proc. Natl. Acad. Sci. U.S.A. 81, 6963-6967. Gospodarowicz, D., Ferrara, N., Schweigerer, L. and Neufeld, G. (1987) Structural characterization and biological functions of flbroblast growth factor. Endocr. Rev. 8, 95-114. Isacchi, A., Caccia, P., Cauet, G., Bertolero, F., Bergonzoni, L., Roletto, F., Ziliotto, R., Garofano, L., Jacob, C., Sarmientos, P., Mazud, G., Carminati, P. and Roncucci, R. (1990) Human basic fibroblast growth factor. Production of a clinical grade recombinant protein from E. coli. Chimicaoggi 6, 72-74. Iwane, M., Kurokawa, T., Sasada, R., Seno, M., Nakagawa, S. and Igarashi, K. (1987) Expression of cDNA encoding human basic fibroblast growth factor in E. coli. Biochem. Biophys. Res. Commun. 146, 470-477. Klagsbrun, M., Smith, S., Sullivan, R., Shing, Y., Davidson, S., Smith, J. and Sasse, J. (1987) Multiple forms of basic fibroblast growth factor: Amino-terminal cleavages by tumor cell- and brain cell-derived acid proteinases. Proc. Natl. Acad. Sci. U.S.A. 84, 1839-1843. Lobb, R. (1988) Clinical applications of heparin-binding growth factors. Eur. J. Clin. Invest. 18, 321-326. Lu, S., Klein, M. and Lai, P. (1988) Narrow-bore High Performance Liquid Chromatography of Phenylthiocarbamyl amino acids and Carboxypeptidase P digestion for protein C-terminal sequence analysis. J. Chromatogr. 447, 351-364. Rosengart, T., Johnson, W., Friesel, R., Clark, R. and Maciag, T. (1988) Heparin protects HeparinBinding Growth Factor-I from proteolytic inactivation in vitro. Biochem. Biophys. Res. Commun. 152, 432-440. Sommer, A. and Rifkin, D. (1989) Interaction of heparin with human basic fibroblast growth factor: protection of the angiogenic protein from proteolytic degradation by a glycosaminoglycan. J. Cell. Physiol. 138, 215-220. Squires, C., Childs, J., Eisenberg, S., Polverini, P. and Sommer, A. (1988) Production and characterization of human basic fibroblast growth factor from Escherichia coli. J. Biol. Chem. 263, 16297-16302. Ueno, N., Baird, A., Esch, F., Ling, N. and Guillemin, R. (1986) Isolation of an amino-terminal extended form of basic fibroblast growth factor. Biochem. Biophys. Res. Commun, 138, 580-588.