Proc. Nati. Acad. Sci. USAVol. 81, pp. 6963-6967, November 1984Biochemistry

Isolation of brain fibroblast growth factor by heparin-Sepharoseaffinity chromatography: Identity with pituitary fibroblastgrowth factor

(vascular endothelial cell proliferation/HPLC/amino acid composition/radioimmunoassay)

DENIS GOSPODAROWICZ*, JANNIE CHENG*, GE-MING LuI*, ANDREW BAIRDt, AND PETER BOHLENTt*Cancer Research Institute and the Departments of Medicine and Ophthalmology, University of California Medical Center, San Francisco, CA 94143; andtLaboratories for Neuroendocrinology, The Salk Institute for Biological Studies, La Jolla, CA 92037

Communicated by Harvey A. Itano, July 25, 1984

ABSTRACT Brain and pituitary fibroblast growth factors(FGF) have been purified to apparent homogeneity from crudetissue extracts by a three-step procedure, including salt precip-itation, ion-exchange chromatography, and heparin-Sepha-rose affinity chromatography. Brain and pituitary FGF havesimilar amino acid compositions and are indistinguishablewith respect to molecular weight (16,000 by polyacrylamidegel electrophoresis), retention behavior in reversed-phasehigh-performance liquid chromatography, and recognition byantibodies directed against the amino-terminal sequence of pi-tuitary FGF. Brain FGF preparations purified by heparin-Sepharose contain, in addition to the major FGF molecularspecies, at least two additional forms of the growth factor,which appear to be very similar by all the above criteria, ex-cept for retention in high-performance liquid chromatogra-phy.

Previous studies have shown that potent mitogens for meso-derm-derived cells, particularly for vascular endothelialcells, are present in both brain and pituitary tissue (1, 2).These factors, first detected on the basis of their ability tostimulate the proliferation of fibroblasts, have been namedpituitary and brain fibroblast growth factors (1). They arebasic mitogens (pl, 9.6) composed of single polypeptidechains with molecular weights of 14,000-16,000 (3, 4). Wehave recently isolated pituitary fibroblast growth factor(FGF) and determined its amino-terminal sequence to bePro-Ala-Leu-Pro-Glu-Asp-Gly-Gly-Ser-Gly-Ala-Phe-Pro-Pro-Gly (5). Brain FGF had previously been characterized astwo fragments of myelin basic protein (MBP) (4, 6), an iden-tification subsequently disputed (7, 8).We have developed a new procedure for the isolation of

FGF using heparin-Sepharose affinity chromatography(HSAC), as described by Shing et al. (9), for the purificationof a chondrosarcoma-derived basic growth factor. We reporthere the purification to apparent homogeneity of basic brainand pituitary FGFs, which appear to be identical. This find-ing resolves the previous controversy concerning the chemi-cal nature of brain FGF.

EXPERIMENTAL PROCEDURESMaterials. Frozen brain and pituitary tissues were ob-

tained from J. R. Scientific (Woodland, CA), kept in aRevco freezer (-800C), and used within a period of 2 weeks.All reagents were of analytical grade. Carboxymethyl-Seph-adex C-50 and heparin-Sepharose were from Pharmacia. TheVydac C4 reversed-phase HPLC column was from Separa-tions Group (Hesperia, CA). Crystalline bovine serum albu-min was from Schwarz/Mann. The Bio-Rad protein assay

kit, low molecular weight standards for NaDodSO4/polyac-rylamide gel electrophoresis, and the silver nitrate stain kitwere from Bio-Rad. Dulbecco's modified Eagle's medium(DME medium) H-16 was from GIBCO. Calf serum wasfrom Hyclone, Sterile Systems (Logan, UT).

Isolation of Brain and Pituitary FGF. Bovine brains (4 kg)or pituitaries (1.8 kg) were extracted with 0.15 M ammoniumsulfate (pH 4.5) as described (1, 3, 4). Partially purified FGFwas prepared by ammonium sulfate precipitation and batchadsorption/elution, using carboxymethyl Sephadex C-50, asdescribed (3, 4). FGF-containing fractions eluting from theion-exchange column with 0.6 M NaCl/0.1 M sodium phos-phate, pH 6.0, were pumped (35 ml/hr) through a heparin-Sepharose column (1.6 x 5 cm; bed vol, 10 ml) that had beenequilibrated at room temperature with 10 mM Tris HCl, pH7.0/0.6 M NaCI. The column was washed (flow rate 35 ml/hr) with 10 mM Tris HCl, pH 7.0/0.6 M NaCl, and then with10mM Tris HCl, pH 7.0/1.1 M NaCl, until the absorbance ofthe eluate at 280 nm became negligible. Mitogenic activitywas then eluted with a linear 2-hr salt gradient of 1.1 M to 2M NaCl in 10 mM Tris HCl (pH 7.0) at 35 ml/hr. Fractionswith biological activity were pooled and kept frozen at-80'C. Unless otherwise stated, total protein was deter-mined by the dye fixation assay (10), using bovine serumalbumin as a standard, and/or by amino acid analysis (seebelow). Aliquots of HSAC-purified FGF were analyzed byreserved-phase HPLC on a Vydac C4 column [25 x 0.46 cm;particle size, S ,tm; pore size, 300 A; using a gradient of ace-tonitrile in 0.1% (vol/vol) trifluoroacetic acid]. Further de-tails are contained in the figure legends.Amino Acid Analysis. Amino acid analysis was performed

on a Liquimat III analyzer (Kontron, Zurich, Switzerland)equipped with an o-phthalaldehyde fluorescence detectionsystem and a proline conversion accessory according to pre-viously described micromethodology (11).NaDodSO4/PAGE. Aliquots (0.5 ,ug of protein) from the

bioactive HSAC fractions were added to a sample buffercomposed of 15% (vol/vol) glycerol/0.1 M dithiothreitol/2%(wt/vol) NaDodSO4/75 mM Tris'HCl, pH 6.8/2 mM phenyl-methylsulfonyl fluoride/2 mM EDTA/1 mM N-ethylmaleim-ide/1 mM iodoacetic acid. Samples were boiled for 3 minand then applied to an exponential gradient (10%-18%) poly-acrylamide slab gel with a 3% stacking gel (12, 13). Electro-phoresis' was for 4 hr at 20 mA. Gels were stained using theBio-Rad silver nitrate stain kit as described by the manufac-turer (14).

Bioassay. The mitogenic activity of column fractions wasdetermined using bovine vascular endothelial cells derivedfrom adult aortic arch as described (7, 15). Briefly, cells

Abbreviations: ABAE cells, adult bovine aortic endothelial cells;CGF, chondrosarcoma growth factor; FGF, fibroblast growth fac-tor; FPLC, fast-protein liquid chromatography; MBP, myelin basicprotein; HSAC, heparin-Sepharose affinity chromatography.

6963

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

6964 Biochemistry: Gospodarowicz et aL

were seeded at an initial density of 2 x 104 cells per 35-mmdish containing 2 ml of DME medium H-16 supplementedwith 10% calf serum and antibiotics (7, 15). Six hours later, aset of triplicate plates was trypsinized and cells were count-ed to determine the plating efficiency. Ten-microliter ali-quots of the appropriate dilution of each fraction (with DMEmedium/0.5% bovine serum albumin) were then added tothe dishes every other day. After 4 days in culture, triplicateplates were trypsinized, and final cell densities were deter-mined by counting cells in a Coulter counter.RIA. Amino-terminally directed antibodies against pitu-

itary FGF were obtained by immunizing 3-month-old male

1.

Ec0cam

zm0

U)m

A

Proc. Natl. Acad Sci. USA 81 (1984)

and female New Zealand White rabbits against the bovineserum albumin-conjugated synthetic decapeptide Pro-Ala-Leu-Pro-Glu-Asp-Gly-Gly-Ser-Tyr [Tyr10-FGF(1-9)I, whichrepresents the amino-terminal sequence of pituitary FGF (5).These antibodies recognize both synthetic antigen and nativepituitary FGF on an equimolar basis and are capable of in-hibiting the FGF-induced proliferation of vascular endotheli-al cells in vitro (5). An RIA was established using the radio-iodinated antigen as a tracer and antiserum (716 B4 and B8)at a final dilution of 1:5000 (5).

RESULTSIsolation of Brain and Pituitary FGF. The heparin-Sepha-

rose affinity chromatography profile of partially purifiedbrain FGF is shown in Fig. 1. Most of the protein (>99%)was not retained by the column (Fig. lA), and the unad-sorbed material had little biological activity (<0.3% that ofthe input) (Fig. 1B). Elution of the column with 1.1 M NaClresulted in the elution of another protein fraction with little

2.5 - Ala

14iin_ 4 lo

x3I

- 2_ I en

-Jwu

!E00CM

z4co0(nmo

FRACTION NUMBER

I0I-

Cl)SC,)-J-Jwu

In0.2 - 4 lo

II~~- -

~~~~~I0~~~~~~I U0.1~~~~~~~~.oZ_ %0

50 6027080LI 11.0 i u50 607080

FRACTION NUMBER

PROTEIN (ng/ml)

FIG. 1. Purification by HSAC of a partially purified preparationof brain FGF. (A) The FGF preparation (0.6 M NaCi eluate fromcarboxymethyl Sephadex column, 98 ml, 7.3 mg/ml), containing>90% of the biological activity present in crude extract (pH 4.5) waschromatographed on a heparin-Sepharose column. Fractions of 10ml and 3 ml were collected during sample loading and column wash-ing; fractions of 1.4 ml were collected during gradient elution. Pro-tein concentration in fractions 5-14 was determined by weighing a 2-ml aliquot after dialysis and lyophilization. For bioassay, aliquots ofthe pooled fractions 5-14, 28-30, and 31-33 were diluted with DMEmedium/0.5% bovine serum albumin, and 10-pil aliquots containing2 ,.g, 16 ng, and 4 ng, respectively, were added to low densityABAE cell cultures. Bioassay results for these pooled fractions areshown in histogram form. Aliquots of fractions 50-67 of the NaClgradient were diluted 1:500 in DME medium/0.5% bovine serumalbumin, and 10-jl aliquots were added to low-density ABAE cellcultures. The final cell densities of the cultures after 4 days areshown for each fraction (A). Densities of control cultures after 4days were 6 x 104 cells per 35-mm dish. (B) Mitogenic activities offractions from various purification steps on low density ABAE cells.NaCl (0.6 M) carboxymethyl Sephadex fraction (a), NaCl (0.6 M)HSAC fraction (tubes 5-14, *), NaCl (1.1 M) HSAC fraction (tubes28-30, A; tubes 31-33, A), NaCl (1.5-1.6 M) HSAC gradient elutionfractions (tubes 55-62, a).

'2

0xx

C,)

C,)-J-JwUL

I 10

PROTEIN (ng/ml)i0

FIG. 2. Purification by heparin-Sepharose affinity chromatogra-phy of a partially purified preparation of pituitary FGF. (A) The pi-tuitary FGF preparation (0.6 M NaCl carboxymethyl Sephadex elu-ate, 124 ml, 7 mg/ml) was chromatographed on a heparin-Sepharosecolumn, as described in Fig. 1A. Protein concentrations in fractions6-15, 33-34, and 54-64 were determined as described in Fig. 1A.Ten microliters of the pooled fractions 6-15 and 33-34, containing 2and 10 ng of protein, respectively, and fractions 54-64 of the NaClgradient were assayed as outlined in Fig. 1A. (B) Mitogenic activitiesof fractions at various purification steps on low density ABAE cellcultures. Conditions were as shown in Fig. 1B. NaCl (0.6 M) car-boxymethyl Sephadex fraction (a), NaCl (0.6 M) heparin-Sepharosefraction (tubes 6-15, *), NaCl (1.1 M) heparin-Sepharose fraction(tubes 33-34, A), NaCl (1.5-1.6 M) gradient elution fractions (tubes56-65, a).

Proc. Nart. Acad. Sci. USA 81 (1984) 6965

Table 1. Purification of brain and pituitary FGF

Protein Maximal Recovery ofrecovered, mitogenic effecti ED50,* Total activity,t biological Purification

Purification step mg ng/ml ng/ml units x 105 activity, % factor

BrainCrude extract 15,000 150 x io3 14 x 103 10.7 100 1Carboxymethyl Sephadex C-50 180 2 x 103 210 8.6 80 66Heparin-Sepharose 0.033 0.4 0.04 8.3 77 350,000

PituitaryCrude extract 26,000 75 x 103 7 x 103 37.1 100 1Carboxymethyl Sephadex C-50 450 1.5 x 103 120 37.5 101 58Heparin-Sepharose 0.150 0.4 0.05 30.0 81 140,000

*Concentration of FGF preparation required to give a 50%o maximal response in the assay system.tOne unit of activity is defined as the quantity of FGF required to give half-maximal stimulation of cell proliferation in the assay systemdescribed.

biological activity (Fig. 1B). Gradient elution of the columnyielded a single protein peak eluting at 1.5-1.6 M NaCl (frac-tions 50-67). This fraction contained most of the biologicalactivity (83%; Table 1), with maximal stimulation of cell pro-liferation occurring at 0.4 ng/ml and half-maximal responseat 40 pg/ml (Fig. iB). The yield of brain FGF in five differentpreparations ranged from 30 to 55 /ig per kg of brain.A similar profile was observed when partially purified pi-

tuitary FGF was chromatographed on heparin-Sepharose(Fig. 2). The material in fractions 54-64 stimulated adult bo-vine aortic endothelial cell (ABAE cell) proliferation maxi-mally at a concentration of 0.4 ng/ml, while half-maximalresponse occurred with 50 pg/ml. Recovery of biological ac-tivity was 80% that of the input (Table 1). The yield of pitu-itary FGF from six different preparations varied from 150 to300 ,pg per kg of tissue.

Characterization of Brain and Pituitary FGF. The homoge-neity of brain and pituitary FGFs after HSAC was analyzedby NaDodSO4/polyacrylamide gel electrophoresis and re-versed-phase HPLC. As shown in Fig. 3, both brain and pi-tuitary FGF migrated as single bands with apparent molecu-lar weights of 16,000 (Fig. 3). The migratory behavior ofUSAC-purified brain and pituitary FGF was identical to thatof homogeneous pituitary FGF, which we had isolated usingdifferent procedures (5).Reversed-phase HPLC analyses of HSAC-purified brain

and pituitary FOF are shown in Fig. 4. Pituitary FGF is high-ly pure, as judged by the presence of only minor impuritieseluting before and after the biologically active UV absorbingpeak (Fig. 4A). The impurities amount to <10% of the mate-rial eluting in the main peak. In contrast, the elution profileof HSAC-purified brain FGF shows four distinct peaks (Fig.4B), the largest (fraction 19) having a retention time identicalto that of pituitary FGF. The three additional peaks (frac-tions 22, 25, and 30) do not represent unrelated impurities,because they possess biological activity. Furthermore, pitu-itary FGF and all forms of brain FGF are immunoreactivewhen assayed in an RIA using amino-terminally directedantibodies against pituitary FGF (Fig. 4 A and B). The aminoacid compositions of HSAC-purified brain and pituitaryFGFs are strikingly similar-(Table 2) and are very close tothat of pituitary FGF isolated by different procedures (5).This constitutes additional evidence for the purity of hepa-rin-purified FGF from both pituitary and brain. Interesting-ly, the amino acid compositions of the three forms of brainFGF as seen by HPLC are all similar (data not shown), sug-gesting that the various forms of brain FGF may be structur-ally related.

DISCUSSIONA three-step purification procedure has been developed forthe isolation of brain and pituitary FGF. The method in-cludes ammonium sulfate precipitation, adsorption to car-

boxymethyl Sephadex, and HSAC, and it yields substantialamounts of highly purified (>90%) brain or pituitary FGF in<48 hr. The heparin-Sepharose step alone is responsible for2400- and 5300-fold purification of pituitary and brain FGF,respectively, thus constituting an extremely powerful tech-nique. Brain and pituitary FGFs purified by HSAC possessthe full intrinsic activity and the same potency as fast-proteinliquid chromatography (FPLC)-purified pituitary FGF (5).Exposure of HSAC-purified FGF to acidic conditions of re-versed-phase HPLC (Fig. 4) results in drastic reduction(95%) of biological potency. This inactivation is similar tothat found previously for pituitary FGF purified by otherprocedures (5).Our combined data (molecular weight, HPLC retention

time, amino acid composition, immunoactivity, and bioac-tivity) suggest that pituitary FGF and the major form of brainFGF are structurally similar, possibly even identical. Thisconclusion is further supported by our preliminary finding(data not shown) that HSAC-purified brain and pituitaryFGFs are also indistinguishable with respect to their immu-noreactivity with murine 1gM monoclonal antibodies raisedagainst homogeneous pituitary FGF prepared by FPLC asdescribed (5).

Brain FGF consists of multiple molecular forms, whichappear to be structurally related based on their close similar-ities with respect to immuno- and bioreactivities and on theiramino acid compositions and molecular weights. Since all

92. X 105' A

92r.5 -- -t66 ft

B C: l) E F

op

31 -_-4

21 -- _

14.4 ----

FIG. 3. NaDodSO4/PAGE of heparin-purified brain and pitu-itary FGF. Samples (10 Ald, containing 0.5 yg of protein) were sub-jected to electrophoresis. Lanes B and D, HSAC-purified brainFGF; lane C, HSAC-purified pituitary FGF; lane E, pituitary FGFpurified by FPLC (5); lanes A and F, protein standard mixture in-cluding phosphorylase (Mr, 31,000), soybean trypsin inhibitor (Mr,21,500), and lysozyme (Mr, 14,400). Similar migration patterns wereobserved regardless of whether the samples were run under reduc-ing conditions.

Biochemistry: Gospodarowicz et aL

6966 Biochemistry: Gospodarowicz etaLP

A

.- E I L.LI-WI iWIWd5 10 15 20 25 30 35

FRACTION NUMBER

Table 2. Amino acid composition of brain and pituitary FGF

-6

-4

-3

Ino

U)

U)I-n

JLu

FIG. 4. Reversed-phase HPLC of heparin-purified pituitary andbrain FGF. (A) Pituitary FGF: 80 Al (20 ,ug of protein) of the pooledfractions 56-62 (Fig. 1A) was diluted with 930 A.l of 0.2 M acetic acidand 1 ml was injected into a C4 column equilibrated in 0.1% (vol/vol)trifluoroacetic acid. Protein was eluted with a linear 90-min acetoni-trile gradient in 0.1% trifluoroacetic acid [26%-36% (vol/vol) aceto-nitrile]. Flow rate was 0.6 ml/min, and fractions of 1.8 ml were col-lected. Aliquots of each fraction were bioassayed as described inFig. 1. Twenty-microliter aliquots were subjected to RIA. Absor-bance peaks corresponding to fractions 2-3 and fractions 34-35 didnot contain significant amounts of protein (by amino acid analysis).(B) Brain FGF: 1 ml (22 pug of protein) of the pooled fractions 55-62(Fig. 2A) was chromatographed as described for A. Bioassay was

carried out as described above, except that aliquots of each fractionwere diluted 1:10 with DME medium/0.5% bovine serum albuminfor assay. RIA was carried out as described above. Material in frac-tion 30, corresponding to the last UV-absorbing peak, was immuno-reactive when assayed at a higher dose. Absorbance peaks corre-

sponding to fractions 2-4 and 34-35 did not contain protein (by ami-no acid analysis).

forms are recognized by specific antisera raised against a

synthetic replicate of the amino-terminal sequence of pitu-itary FGF (5), it is likely that all forms of brain FGF sharethe same amino-terminal sequence already determined forpituitary FGF-Pro-Ala-Leu-Pro-Glu-Asp-Gly-Gly-Ser-ora sequence very similar. The nature of the structural differ-ences between the various forms of brain FGF is not yetclear, and it is likely that they are reflections of protein side-chain modifications or microheterogeneity of FGF.

Brain and pituitary FGF preparations, as reported earlier(16-18), are mitogenic for a wide variety of normal diploid

AsxThrSerGIxGlyAlaValMetIleLeuTyrPheHisTrpLysArgCysPro

HSAC-purifiedbrain FGF*

11.54.48.9

11.316.19.35.22.13.0

12.15.87.21.81.0

14.210.34.69.2

HSAC-purifiedPit. FGF*

11.24.48.812.915.39.05.62.33.3

12.15.96.52.31.0

14.010.04.09.3

Pit. FGFt104912169523

117731

131169

Values represent residues per molecule determined from 24-hr hy-drolysates (with 5.7 M HCI) of 10-30 pmol of protein and are notcorrected for hydrolysis losses. Procedures for obtaining values forall amino acids are described in ref. 11. Compositions are calculatedfor a 138 amino acid protein, which is in agreement with the ob-served molecular weight. Pit., pituitary.*Values are means of duplicate determinations.tAmino acid composition of pituitary FGF purified as described (5).

cultured cells derived from tissue originating from the pri-mary or secondary mesenchyme as well as from neuroecto-derm. These include rabbit chondrocytes, bovine granulosaand adrenal cortex cells, bovine corneal endothelial cells,and human umbilical endothelial cells. HSAC-purified brainand pituitary FGFs have identical spectra of activity (unpub-lished observations). Moreover, heparin-Sepharose affinitypurified brain and pituitary FGF are potent angiogenic fac-tors in vivo. Slow-release forms of brain or pituitary FGFcontaining 1 ug of the growth factor were capable of elicitinga maximal angiogenic response when implanted in either thehamster cheek pouch or in the chicken chorioallantoic mem-brane. Histological section revealed that neovascularizationtook place in the virtual absence of inflammatory cells (un-published observations).Although acidic and neutral forms of FGF have been re-

ported to be present in acidic (pH 4.5) pituitary and brainextracts (8, 19, 20), we find that >99% of the bioactivity sub-jected to heparin chromatography is strongly bound by hepa-rin-Sepharose. The high recovery of FGF (80%-83%) in the1.5-1.6 M NaCl fractions, and the presence of only smallamounts of bioactivity in side fractions from the ammoniumsulfate and carboxymethyl Sephadex steps (3, 4, 7), indicatethat FGF, as characterized in this report, is the major formof mitogen present in acidic extracts of brain and pituitarytissue. This is in contrast with earlier reports (8, 20) claimingthat significant amounts of acidic FGF can be present in par-tially purified preparations of brain or pituitary extracted atpH 4.5. It is also in contrast with a recent report (21) describ-ing the purification to homogeneity of an anionic form ofFGF from acidic (pH 4.5) brain extracts. Apparently multi-ple forms of FGF are present in brain that are biologically,but not necessarily structurally, related. The isolation of ba-sic or acidic FGF as the major mitogenic form may dependon minor differences in the isolation protocols or assaysused.The present study establishes the virtual identity of basic

brain and pituitary FGF. Since pituitary FGF has been par-tially characterized structurally (5) and has been shown to be

04o

Proc..Natl. Acad Sci. USA 81 (1984)

Proc. NatL. Acad Sci. USA 81 (1984) 6967

unlike any MBP fragment,. the previous identification ofbrain FGF as degradation products of MBP (6) must be con-sidered erroneous. It is likely that, as suggested by others(8), brain FGF copurified with MBP fragments when brainextracts were processed through the purification scheme de-scribed (4). Although both brain and pituitary FGF are dif-ferent from all other growth factors identified to date, theyshare common properties with a chondrosarcoma growthfactor (CGF), which is mitogenic for capillary endothelialcells (9). FGF and CGF are basic growth factors (pl, 9.6) thathave similar molecular weights (16,000-17,000 for brain andpituitary FGF, and 17,000-18,000 for CGF) and are mitogen-ic for vascular endothelial cells. They also share the propertyof being angiogenic in vivo (9, 22). The question of whetherFGF and CGF are related molecules can only be answeredwhen their complete structural characterization has been ac-complished.

Note Added in Proof. Recent studies using bovine adrenal cortexcapillary endothelial cells have shown that those cells are even moredependent on FGF in order to proliferate and preserve their pheno-type than ABAE cells (unpublished observations). The amino-termi-nal sequence of the various forms of brain FGF has been determinedand is identical in all cases.

We wish to thank K. von Dessonneck, R. Schroeder, and R.Klepper for excellent technical assistance and H. Scodel for manu-script preparation. Research in the laboratory of D.G. was support-ed by National Institutes of Health Grants (HL-20197 and EY-02186). Research at the Laboratories for Neuroendocrinology was

supported by National Institutes of Health Program Grants (HD-09690 and AM-18811) and by the Robert J. and Helen C. KlebergFoundation.

1. Gospodarowicz, D. (1974) Nature (London) 249, 123-127.2. Gospodarowicz, D., Greenburg, G., Bialecki, H. & Zetter, B.

(1978) In Vitro 14, 85-118.3. Gospodarowicz, D. (1975) J. Biol. Chem. 250, 2515-2520.4. Gospodarowicz, D., Bialecki, H. & Greenburg, G. (1978) J.

Biol. Chem. 253, 3736-3743.

5. Bohlen, P., Baird, A., Esch, F., Ling, N. & Gospodarowicz,D. (1984) Proc. Nati. Acad. Sci. USA 81, 5364-5368.

6. Westall, F. C., Lennon, V. A. & Gospodarowicz, D. (1978)Proc. Natl. Acad. Sci. USA 75, 4675-4678.

7. Gospodarowicz, D., Lui, G. M. & Cheng, J. (1982) J. Biol.Chem. 257, 12266-12278.

8. Thomas, K. A., Riley, M. C., Lemmon, S. K., Baglan, N. C.& Bradshaw, R. A. (1980) J. Biol. Chem. 255, 5517-5520.

9. Shing, Y., Folkman, F., Sullivan, R., Butterfield, C., Murray,J. & Klagsbrun, M. (1984) Science 223, 1296-1299.

10. Bradford, M. (1976) Anal. Biochem. 126, 144-150.11. Bohien, P. & Schroeder, R. (1982) Anal. Biochem. 126, 144-

152.12. Laemmli, U. K. (1970) Nature (London) 227, 680-685.13. Tauber, J. P., Cheng, J., Massoglia, S. & Gospodarowicz, D.

(1981) In Vitro 17, 519-530.14. Merril, C. R., Goldman, D., Sedman, S. A. & Ebert, M. H.

(1981) Science 211, 1437-1438.15. Gospodarowicz, D., Moran, J., Braun, D. & Birdwell, C. R.

(1976) Proc. Natd. Acad. Sci. USA 73, 4120-4124.16. Gospodarowicz, D., Mescher, A. L. & Birdwell, C. R. (1978)

in Gene Expression and Regulation in Cultured Cells: ThirdDecennial Review Conference, National Cancer InstituteMonograph no. 48 (Natl. Cancer Inst., Bethesda, MD), pp.109-130.

17. Gospodarowicz, D., Moran, J. & Mescher, A. L. (1978) in Mo-lecular Control of Proliferation and Cytodifferentiation, 35thSymposium of the Society for Developmental Biology, eds. Pa-pacanstantinou, J. & Rutter, W. J. (Academic, New York), pp.33-61.

18. Gospodarowicz, D., Vlodavsky, I., Fielding, P. & Birdwell,C. R. (1978) in Birth Defects, eds. Littlefield, J. W. & de-Grouchy, J. (Excerpta Medica, Amsterdam), pp. 233-271.

19. Lemmon, S. K., Riley, M. C., Thomas, K. A., Hoover,G. A., Maciag, T. & Bradshaw, R. A. (1982) J. Cell. Biochem.95, 162-169.

20. Gambarini, A. G. & Armelin, H. A. (1982) J. Biol. Chem. 257,9692-9697.

21. Thomas, K. A., Rios-Candelore, M. & Fitzpatrick, S. (1984)Proc. Natl. Acad. Sci. USA 81, 357-361.

22. Gospodarowicz, D., Bialecki, H. & Thakral, T. K. (1971) Exp.Eye Res. 28, 501-514.

Biochemist ': Gospodarowicz et aL