Multiple Forms of Endothelial Cell Growth Factor

THE .JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 260.No. 21. Issue ofSeptember 25, pp. 11389-11392, 1985 c 1985 by The American Society of Bioloacal Chemists, Inc. Printed in U . S . A .

Communication

between the carbohydrate and the anionic polypeptide may have biological significance (5). Additional studies have revealed that heparin: (i) increases the affinityof ECGF for its receptor, (ii) restores biological activity to denatured prepaRAPID ISOLATIONAND BIOLOGICALAND rations of purified ECGF, and (iii) potentiates the binding of CHEMICALCHARACTERIZATION* monoclonal antibodies to ECGF (5). Together these studies (2-5) suggest that heparin may act as a cofactor for ECGF (Received for publication, May 31, 1985) which may involve the stabilization of the secondary and/or Wilson H. Burgess, Tevie Mehlman, Robert Friesel, tertiary structure of the polypeptide (5). Warren V. Johnson, and Thomas MaciagS It is possible t o utilize heparin-Sepharose affinity chromaFrom the Department of Cell Biology, Revlon Biotechnology of existingpurification tography (2) withintheconstruct Research Center, Rockuille, Maryland 20850 procedures (6) to prepare relatively small quantities of the mitogen (3-5). ECGF partially purified by extraction and a Endothelial cell growth factor (ECGF)can be rapidly high to low molecular weight transition utilizing acid treatpurified from bovine brain to high specific activity ment, gel exclusion chromatography, and ammonium sulfate using heparin-Sepharose affinity chromatography. precipitation (6) can be used as starting materialfor heparinPurification of the mitogen by this method results in Sepharose affinity chromatography( 2 ) .ECGF purified in this relatively high yields of the polypeptide (10 to 100 pg/ manner is a single-chain polypeptide which possesses an kg of tissue) with biological activity on murine and human endothelial cells in the picogram range. The anionic isoelectric point and an apparentmolecular weight of product obtained is a mixture of two single-chainpoly- 20,000 ( 2 , 3, 5). Although thispurification procedure has peptides with apparent molecular weights of 17,000 resulted in the generation of sufficient quantities of ECGF (a-ECGF) and 20.000 (B-ECGF)bysodium dodecyl for receptor binding studies (3, 5) and antibody biosynthesis sulfate-polyacrylamide gel electrophoresis. The two (2), larger quantities are required for structural studies. We forms of ECGF can be separated by either NaCl gra- now report a rapid procedurefor the purification of relatively dient elution from heparin-Sepharose or reversed- large quantities of ECGF. Thisprocedure utilizes the affinity phase high pressure liquid chromatography. The two of ECGF for heparin and has resultedin the identification of polypeptides are related on the basis of similar: (i) multiple forms of the polypeptide growth factor. amino acid compositions, (ii) affinity for heparin-SephMATERIALS AND METHODS arose,(iii) cyanogen bromide and trypsin-derived Acrylamide, N,N’-methylenebisacrylamide, N’,N’,N’,N’-tetracleavage products, and (iv) biological activity. Furthermore, the cyanogen bromide fragments derived methylenediamine, ammonium persulfate, and SDSwere from BDH. and molecularweight standards were obtained from the two forms of ECGF also possess similar amino Heparin-Sepharose from Pharmacia. TheVydac C, column was fromRainin Instruments. acid compositions and mobilities on sodium dodecyl The Synchropak Cl8 column was from SynChrom. Aminoacid sulfate gels. These data suggest that there are at least standards, HPLC water, acetonitrile, CNBr, constant boiling HCI, two discrete molecular forms of ECGF in bovine brain and o-phthaldehyde reagentwere from Pierce Chemical Co. All other and that these two molecules are structurally related. chemicals were reagent-grade.

Multiple Forms of Endothelial Cell Growth Factor

Endothelial cell growthfactor (ECGF’) istheprincipal polypeptide mitogen present in bovine neural tissue for human endothelial cells (1, 2) and is a member of a family of polypeptide mitogens which presently includes acidic fibroblast growth factor (FGF) and eye-derived growth factor-I1 (EDGF-11) (3). The biological activities of crude (4) and purified preparations of ECGF, acidic FGF, and EDGF-I1(3, 5 ) are potentiatedby the glycosaminoglycan heparin. Wehave previously reported that ECGF avidly binds to immobilized heparin (2) and have suggested that the structural interaction

Polyacrylamide gel electrophoresis in the presence of SDS was performed essentially as described by Laemmli (7) using a Mighty Small gel apparatus (Hoefer Scientific). Peptideswere isolated from SDS gels using the apparatus and methodology described by Hunkapiller et al. ( 8 ) .Amino acid analysis was performed on samples that were hydrolyzed in uacuo in 6 N HC1, 0.1% phenol for 24 h at 115 “C. Aminoacid compositions were determined basedonion-exchange separationandpostcoiumno-phthaldehydedetectionand by reversed-phase separation of phenylisothiocyanate-derivatized amino acids using the generalprocedures outlined by Waters Associates. Data collection and reduction was performed using a Waters Associates 840 system. RESULTS

The Rapid Purification ofa- and P-ECGF-AI1 procedures were performed at 4 “C. Bovine brain was homogenized in 1.3 * This work was supported by National Institutesof Health Grants volumes of 50 mM Tris-HC1, pH 7.4, containing 50 mM EDTA H L 310765 and AG 04807 (to T. M.). The costsof publication of this article were defrayed in part by the payment of page charges. This (Tris-EDTA) ina Waring blender for 3 min at 4 “C. The article must therefore he hereby marked “aduertisement” in accord- homogenate was centrifuged at 10,000 x g for 30 min and the pH of the supernatant was lowered to pH 4.5 with 1 M acetic ance with 18 U.S.C. Section 1734 solely to indicate this fact. j To whom correspondence should be addressed Department of acid. After 30 min at pH 4.5 the extract was centrifuged at Cell Biology, Revlon BRC 2 Research Court, Rockville, MD 20850. 10,000 X g for 30 min and the supernatant was quickly titrated The abbreviations used are: ECGF, endothelialcell growth factor; to pH 7.4 with 1 IU NaOH. The supernatant was subjected to FGF, fibroblast growth factor; EDGF-11, eye-derived growth factor11; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electro- saltfractionation with 50 and 95% saturatedammonium phoresis; HPLC, high pressure liquid chromatography; CNBr, cyan- sulfate and pellets were collected by centrifugation at 10,000 X g for 60 min.The 95% ammoniumsulfate pellet was ogen bromide; LE-11, murine lung capillary endothelial cells; HUVEC, human umbilical vein endothelial cells; EGF, epidermalgrowth factor. resuspended in Tris-EDTA ( ~ 1 ml/kg 0 tissue) and dialyzed

11389

Endothelial Cell Growth Factor

11390

TABLE I Purification of bovine brain a-ECGF and 0-ECGF Volume Procedure

Total units"

Total protein

Specific activity Purification Yield

unitslmg % -fold supernatant First 6.25 8,500 100 X lo7 432 144,750 1 Acid precipitation 7,800 (pH 4.5) 3.0 48 X 107 402 74,580 1 0-50% ammonium sulfate 8,950 2.64 X 107 1,640 16,165 42 4 830 21,580 2.2 X 107 1,019 35 2.35 50-95% ammonium sulfate NaCl gradientb 140 4.5 X 10'2.0 2.24 46 X 10' Pool A 1,157 Pool B 190 2.2 x lo6 4.37 0.5 X lo6 18 Pool c 190 6.7 X lo61.0 6.65 x lo6 2,315 Pool D 130 2.2 x lo6 2.21 1.0 x lo6 2,315 C4-HPLCc*d 4.0 X9,300 6 3.0 X lo6 lo6 5 0.75 a-ECGF 7.0 X lo6 0.8 16,300 3 5.2 X 105 0.075 8-ECGF One unit of activity is the amount of ECGF required to give half-maximum [3H]thymidineincorporation into quiescent murine endothelial cells (LE-11). Pools of the fractions derived from NaCl gradient elution off heparin-Sepharose are identified in Fig. 1. 'The protein concentration was determined by amino acid analysis. All other protein determinations were performed in duplicate by the method of Bradford (19). dPool C wasused as starting material for reversed-phase HPLC. The pool (190 ml) waslyophilized and reconstituted with Buffer A (8 ml) and 30% of this material was loaded into the C, column (see Fig. 2).

ml

A

12345 678910

C

D

1 2 3 4 5 6 7 8 9 1 0

E

1 12 23 34 4 15 26 37 FIG. 1. Analysis of the purification and characterization of ECGF by SDS-polyacrylamide gel electrophoresis. The SDS gels shown in panels A-E are 15% (w/v) acrylamide. The standards used have molecular weightsof 94,000,67,000,43,000,30,000,20,100, and 14,400. Panel A , the samples applied to the lanes were: lane 1, molecular weight standards; lane 2, the heparin-Sepharose column load; lane 3, the column breakthrough; lane 4,0.5 M NaCl wash; lanes 5-10, selected fractions from early portions of the 0.5 to 1.5 M NaCl gradient (3000 ml, total volume). Panel B, lane 2, contains molecular weight standards; lanes 1 and 3-10 contain successive fractions (fractions 18-26, 50 ml each) from the NaCl gradient elution of the heparin-Sepharose column that contained mitogenic activity (6). The fractions in panel A , lanes 9 and 10, were combined as Pool A; those in panel B, lanes 1, 3, 4, and 5, were combined as Pool B; those in panel B, lanes 6-8, were combined as Pool C; and those in panel B, lanes 9 and 10, were combined as Pool D. Panel C, lane 1, contains molecular weight standards; lanes 2,3, and 4 contain samples representative of Pools B, C, and D, respectively. Panel D shows samples obtained from reversed-phase HPLC of biologically active fractions obtained from heparin-Sepharose (of Fig. 2.4). Lane 1 contains molecular weight standards; lanes 2 and 3 contain fractions from the ECGF-like protein region; lanes 4-6 contain fractions from the aECGF region; and lane 7 contains afraction from the 8-ECGF region of the C4 elution. Panel E shows the electrophoretic analysis of the cyanogen bromide digests of a- and 6-ECGF. The proteins (0.1-0.5 mg/ml) were dissolved in 70% (w/v) formic acid and a 500-fold molar excess of CNBr was added. The solution was purged with nitrogen and incubated at room temperature inthe dark for 20 h. The reaction

mg

overnight against 50 volumes of Tris-EDTA. Thesample was clarified by centrifugation at 10,000 x g for 60 min. Hydrated heparin-Sepharose (40 m1/2 kg of bovine brain) was added directly to the ECGF mixture. After 1 h at 4 "C with gentle stirring, a column (2.5-cm, inner diameter) was packed with the batch-adsorbed heparin-Sepharose mixture. The column was washed with 10 volumes of Tris-EDTA followed by 0.5 M NaCl in Tris-EDTA a t a flow rate of ~1 column volume/h. The biological activity was eluted witha linear gradientof 0.5 to 1.5 M NaCl inTris-EDTA(total volume 2 15 column volumes). The proteins eluted with the gradientwere examined for their biological activity and theirmobilities on SDS gels (Fig. 1,A-E). As shown in Fig. 1, B and C, two polypeptide species were observed and assigned names from their migratory position; (3-ECGF for the apparent M , z 20,000 species and a-ECGF for the apparent M , 17,000 species. The two polypeptides could be separated using heparin-Sepharose since (3-ECGFelutes before a-ECGF (Fig. 1,Band C). Greater than 90% of the biological activity was found in these pools (Table I, Pools B-D). The polypeptides were active biologically at 500 pg/ml on LE-I1 cells and human endothelial cells. Biologically active a- and (3-ECGF were also resolved (Figs. 24 and 1D) from one another andfrom a nonmitogenic, but immunoreactive, polypeptide which we have termed ECGFlikeprotein'usingreversed-phase HPLC. The biologically active (Fig. 2B) HPLC-purified material (Figs. 2A and ID) was used in the structural studies described below. The biological activity of these post HPLC fractions is stable for several weeks a t 4 "C and afterlyophilization. a-ECGF and(3-ECGFAre Related Polypeptides-The amino acid compositions of a- and (3-ECGF are similar (Table 11). Sincetheamino acidcompositions suggest relatively few methionines in either a- or (3-ECGF, digestion with CNBr was performed. The CNBr fragments derived from a-ECGF was terminated by adding 15 volumes of HPLC water and lyophilization. The samples were resuspended in HPLC water and analyzed by SDS-PAGE. Lane 1, contains molecular weight standards, lane 2 contains CNBr-digested a-ECGF, andlane 3 contains CNBr-digested 8-ECGF. Undigested a- and 8-ECGF are marked with arrows and stars. The CNBr fragments a-1, a-2, 8-1, and 8-2 are marked with arrows.

* W. H. Burgess and T. Maciag, unpublished observations.

Endothelial Factor Cell Growth

11391

C

A ECGF-LP

I

Time (min.1

D

Time (min.1

FIG. 2. Separation and characterization of biologically active a- and fl-ECGF by reversed-phase HPLC. Samples from the heparin-Sepharose column were pooled,dialyzed against 10 mM ammonium bicarbonate, and lyophilized. The lyophilized protein was resuspended in solution A (0.1% trifluoroacetic acid) and applied (up to 300 pg) to a Vydac C, column. The column was developed with acetonitrile with trifluoroacetic acid added to balance AS,6nm(buffer B). The flow rate was 1.0 ml/min. After 5 min, the % buffer B was increased linearly in the following order: 10 min, 30%; 40 min, 40%; 45 min, 80%; 55 min, 80%;60 min, 5%. Panel A shows the trace of absorbance at 215 nm versus time. Panel E shows the ability of the reversed-phase HPLC samples (assayed at a dilution of 1:103) to stimulate [3H]thymidineincorporation in the LE-I1 cell assay (2). The inset topanel E shows the abilities of peak fractions of a-ECGF ( X - X ) and @-ECGF(W) to stimulate growth oflow cell seed density HUVEC (6). Panels C and D,reversed-phase HPLC analysis of peptides generated from digestion of aECGF (C) and @-ECGF(D)with trypsin. Trypsin digests were made in 0.1 M ammonium bicarbonate. Substrate concentrations were0.1-0.5 mg/ml and the final enzyme concentration was1:50 (w/w) of the substrate. The enzyme was added in two equal aliquots over a 16-h incubation period a t 37 “C. Reactions were terminated by lyophilization. The digests were resuspended in 0.1% trifluoroacetic acid and applied to a CIS column (4.1 X 250 mm). Thesolvent system was the same as thatin panel A . The column was equilibrated in buffer at a flow rate of 1 ml/min. After 5 min, the % buffer B was increased linearly to 40% over 50 min then increased to 80% over 5 min.

andp-ECGF were analyzed by SDS-PAGEand reversedphase HPLC. Digestion of both a - a n d P-ECGF with CNBr revealed two distinct fragments (Fig. 1E).These fragments could not be resolved by reversed-phase HPLC. Analysis of the CNBr-digested materialby SDS-PAGE in the absenceof reducing agents revealed prominent bands characteristic of the undigested material (data not shown). These data suggest the presence of a disulfide bond bridging the CNBr cleavage site. Individual CNBr fragments were obtained by electroelution from reduced SDS gels. The amino acid compositions of the individual CNBr fragments (a-1 versus p-1 and 01-2 versus p2) derived from either a - or P-ECGF are similar (Table 11). The electrophoretic mobilities of the a-1 and P-1 fragments are the same, whereas the mobilities of a-2 and P-2 differ

(Fig. 1E).These data indicate that digestion of either a- or P-ECGF with CNBr generates similar and distinct fragments. The similarities between a- and P-ECGF were further examined by reversed-phase HPLC analysisof trypsin-digested polypeptides (Fig. 2, C and D).The majority of the fragments identified for P-ECGF were also present in the a-ECGF digest. There were no major peaks of absorbance unique to the a ECGF digest. DISCUSSION

We havedevelopedaprocedurefor thepurification of multiple forms of ECGF from bovine brain. The success of the purification procedure is attributed to the transition of ECGF from a high to low molecular weight form (6) and the specificity of ECGF for heparin (2,5).Although both a-ECGF

11392

Endothelial Factor Cell Growth

TABLEI1 Amino acid compositionsof a-ECGF, P-ECGF, and CNBr fragments

(ii) cross-reactivity with polyclonal and monoclonal ECGF antibodies, (iii) competition with ECGFfor binding to a high affinity endothelial cell-derived receptor, and (iv) potent biAmino acid u-ECGF p-ECGF CNBr CNBr CNBr CNBr a-lb b-lb a-2C 01-2= ological activity as endothelial cell mitogens which is potentiated by heparin (3). Aspartic acid 16.0 6.6 6.1 11.8 6.2 4.7 Threonine 8.1 9.6 3.0 4.5 5.1 3.1 It is of interestthata-ECGFissimilarto acidic FGF Serine 11.911.3 4.8 5.7 3.2 3.6 isolated from bovine neural tissue by Thomas et al. (10) with Glutamic acid 16.019.6 6.9 9.6 9.5 6.8 respect to its functional properties (3), apparent molecular Proline 5.95.7 NDd NDd NDd NDd weight on SDS gels, and amino acid composition. Similarly, 20.0 19.4 14.2 11.2 9.2 8.1 Glycine P-ECGF appears tobe identical to thesingle species of ECGF Alanine 6.1 6.9 4.5 3.0 2.9 2.2 originallypurified andcharacterized by Maciag et al. (6). 4.3 5.0 NDd NDd NDd NDd Cysteine‘ Valine 5.0 5.0 3.0 3.0 2.0 2.0 These data confirm and extend the antigenic and radiorecepMethionine’ 0.7 0.7 tor evidence for a unique family of endothelial cell polypeptide Isoleucine 2.7 6.0 6.4 2.9 3.0 3.3 mitogens (3) and argue thatcommon structures are sharedby Leucin@ 19.8 7.1 13.2 75 3.4 6.3 these polypeptides. Further structural studies promise to re7.6 6.5 3.2 2.4 3.9 3.5 Tyrosine veal the nature of the structure-function relationships be7.3 6.9 2.7 3.2 1.1 1.9 Phenylalanine tween theheparin- (2) andreceptor-binding ( 5 ) domains. Histidine 5.9 2.7 4.4 0.9 0.9 3.5 14.6 11.4 NDh NDh NDh NDh These studies are particularly important to the characterizaLysin@ 5.8 6.2 3.9 2.6 4.5 2.6 Arginine tion of an ECGF-like polypeptide which is not mitogenic for ND’ ND ND’ ND’ ND’ N D Tryptophan endothelial cells but shares certain functional (heparin-binda Normalized to 5 residues of valine. ing) and structural (monoclonal antibody-binding) properties * Normalized to 3 residues of valine. with a- and p-ECGF.’ The precise relationship of a- to pe Normalized to 2 residues of valine. ECGF and their relationships to other endothelial cell polyNot determined, analysisby postcolumn o-phthalaldehyde only. peptide mitogens (9-16) will depend upon the determination e Determined as cysteic acidafter performic acid oxidation (20). ’Determined as methionine sulfone after performic acid oxidation. of the primary structuresof these polypeptides and the spectrum of their biological attributes. 8 Acidhydrolysis of heparin (Upjohn) yieldsresidueswhichcomigrate with Leu and Lys. Not determined Lys co-migratingwith ammonia. Acknowledgments-We thank W. Terryand C. Smith for their Not determined. support and critical review of the manuscript, and L. Peterson for excellent secretarial expertise. (apparent M , = 17,000) and P-ECGF (apparentM , = 20,000) REFERENCES exhibit relatively high affinities for immobilized heparin, sub1. Maciag, T. (1984) Prog. Thromb. Hemostasis 7, 167-182 tle differences in their affinities can be detected using salt2. Maciag, T., Mehlman, T., Friesel, R., and Schreiber,A. B. (1984) gradient elution. The two forms of ECGF are indeed funcScience 225,932-934 tionally related as judged by their mitogenic activities for 3. Schreiber, A.B., Kenney, J., Kowalsky,W. J., Thomas, K. A,, Gimenez-Gallego, G., Rios-Canderlore, M., Di Salvo, J., Barendothelial cells. In addition, they compete for endothelial ritault, D., Courty, J., Courtois, Y., Moenner, M., Burgess, W. cell receptor binding and are immunoreactive with anti-ECGF H., Mehlman, T., Friesel, T., Johnson, W., and Maciag, T. monoclonal a n t i b ~ d i e s . ~ (1985) J. Cell Biol., in press The results described here provide a structural basisfor the 4. Thornton, S. C., Mueller, S. N., and Levine, E. M. (1984) Science functionalsimilaritiesexhibited by a- and p-ECGF. The 222,623-625 similarities in the amino acid compositions of the two poly5. Schreiber, A. B., Kenney, J., Kowalski, W. J., Friesel, R., Mehlman, T., and Maciag, T. (1985) Proc. Natl. Acad. Sci. U. S. A , , peptides suggest that they are related. The relationship bein press tween the two polypeptides is further establishedby analysis 6. Maciag, T., Hoover,G. A,, andWeinstein, R. (1982) J. Biol. of their chemical and enzymatic cleavage products. Analysis Chem. 257, 5333-5336 of the CNBrcleavage products by SDS-PAGE in the presence 7. Laemmli, U. K. (1970) Nature (Lond.)227, 680-685 and absenceof reducing agents indicates thata- and p-ECGF 8. Hunkapiller, M. W., Lujan, E., Osterander, F., and Hood, L. E. bothhaveanintra-chain disulfide bond which spans the (1983) Methods Enzymol. 91, 227-236 9. Courty, J., Loret, C., Moenner, M., Chevallier, B., Lazente, O., cleavable methionine residue. Similarly, the HPLC profiles of Courtois, Y., and Barritault, D. (1985) Biochemie, in press trypsin-digested a- and p-ECGF alsosuggest a high degree of structural homology between the two polypeptides. These 10. Thomas, K., Rios-Candolore, M., andFitzpatrick, S. (1984) Proc. Natl. Acad. Sci. U. S. A. 81, 357-361 structural similarities do not preclude either minor alterations 11. Conn, G., and Hatcher, V. B. (1984) Biochem.Biophys.Res. in the primary structures of the polypeptides orintrinsic Commun. 124, 262-268 anomalies which affect the behavior of the polypeptides on 12. D’Amore, P. A,, and Klagsbrun,M. (1984) J.Cell Biol. 99,15451549 reversed-phase HPLC and their electrophoreticmobilities in 13. Gambarini, A. G., and Armelin, H. A. (1982) J . Biol. Chem. 257, SDS gels. This interpretation is consistent with thebehavior 9692-9697 of polypeptides such as calmodulin (17) in SDS gels and the 14. Kellet, J. G., Tanaka, T., Rowe, J. M., Shiu, R. P. C., and Friesen, des-asparaginylderivative of EGF (18) onreversed-phase H. G. (1981) J . Biol. Chem. 256,54-58 HPLC. 15. Klagsbrun, M., and Shing, Y . (1985) Proc. Natl. Acad. Sci. U. S. Together these data provide further evidence for the existA . 82,805-809 16. Lobb, R. R., and Fett, J. W. (1984) Biochemistry 23,6295-6299 ence of a distinct family of polypeptide growth factors and Kretsinger, R. H. (1980) are consistent with recent reports (3) that ECGF ( 2 ) , acidic 17. Burgess,W.H.,Jemiolo,D.K.,and Biochem. Biophys. Acta 623,257-270 FGF (lo), andEDGF-I1(9)shareconsiderablefunctional 18. Di Augustine, R. P., Walker, M. P., Klapper, D. G., Grove, R. I., homology. These similarities include: (i) affinity for heparin, Willis, W. D., Harvan, D. J., and Hernandez,0. (1985) J . Biol. A.B. Schreiber, W.H.Burgess, observations.

and T. Maciag,unpublished

Chem. 260, 2807-2811 19. Bradford, M. M. (1976) Anal. Biochem. 72, 248-254 20. Hirs, C. H. W. (1967) Methods Enzymol. 11, 197-199