Characterization and Molecular Cloning of a

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 266,No. 25,Issue of September 5,pp.

Characterization and Molecular Cloning a ofPutative Binding Protein for Heparin-binding Growth Factors* (Received for publication, May 10, 1991)

Dianqing Wu, Mikio Kan, GordonH. Sato, Tetsuji OkamotoS, andJ. Denry Satoi From the W. Alton Jones Cell Science Center, Inc., Lake Placid, New York 12946

A novel M. 17,000 heparin-binding protein was purified from culture medium conditioned by A431 human epidermoid carcinoma cells. This protein, designated HBpl7, was found to bind the heparin-binding peptide growth factorsHBGF-1 and HBGF-2 in a noncovalent, reversible manner. In addition HBpl7 was found to inhibit the biological activities of both HBGF1 and HBGF-2. Both the binding and inactivation of HBGF-1 and HBGF-2 by HBpl7 were abolished by heparin. Full-length 1163-base pair HBpl7 cDNA was cloned and sequenced by using the polymerase chain reaction technique. The deduced primary structure of HBpl7 consisted of 234 amino acids including each of five partial peptide sequences obtained from proteolytic fragments of purified HBpl7. The encoded protein included a 33-residue N-terminal signal sequence for secretion and a single potential N-linked glycosylation site. No homology with any known protein was found for the deduced primary structure of HBpl7. The expression of HBpl7 mRNA was found to occur preferentially in normal human keratinocytes and in squamous cell carcinomas. This pattern of HBpl7 gene expression suggests that this binding protein for HBGFs 1 and 2 has a physiological role in squamous epithelia.

Heparin-binding growth factors (HBGF)’ are a family of seven polypeptide growth factors that include HBGF-1 (acidic FGF), HBGF-2 (basicFGF), K-FGF, Int-2, FGF-5, KGF, and FGF-7 (1-3). Of these HBGF-1and HBGF-2 have been implicated in a number of biological processes in vitro and in vivo such as growth stimulation of mesodermal- and neuroectodermal-derived cells, angiogenesis, wound healing, and tis-

* This research was supported by Grants CA40294 and CA37589 from the National Cancer Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solelyto indicate this fact. The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank withaccessionnumber(s) M60047. $Present address: Dept. of Maxillofacial Surgery I, Hiroshima University Dental School, Hiroshima 734, Japan. $ To whom correspondence should be addressed W. Alton Jones Cell Science Center, Inc., 10 Old Barn Rd., Lake Placid, NY 12946. The abbreviations used are: HBGF, heparin-binding growth factor; FGF, fibroblast growth factor; EGF, epidermal growth factor; PDGF, platelet-derived growth factor; HPLC, high performance liquid chromatography; RP-HPLC, reverse-phase high performance liquid chromatography; ACN, acetonitrile; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; DSS, disuccinimidyl suberate; PCR, polymerase chain reaction; bp, base pair(s); kb, kilobase pair(s); RACE, rapid amplification of cDNA end; DME medium, Dulbecco’s modifiedEagle’s medium; FBS, fetal bovine serum.

sue regeneration (1, 3, 4). Both HBGF-1 and HBGF-2 were initially isolated from neural tissueas M , 16,000-18,000 polypeptides (5,6). Theyare 55% identical in aminoacid sequence (7), they have similar biological activities (1,3), andthey may act through the same cell surface receptor molecules (8). However, HBGF-1 and HBGF-2 have several functionally defined differences that are based on their interactions with heparin. HBGF-2 is biologically more potent than HBGF-1 in the absence of heparin (9); theactivity of HBGF-1 but not HBGF-2 is potentiated by heparin (10,ll); andHBGF-2 has a greater affinityfor immobilized heparin than HBGF-1(12). Tumor angiogenesis is apotentialrate-limiting step in tumor development and metastasis (13-15). Both HBGF-1 and HBGF-2 have been shown to have strong angiogenic activity (16-18). Since many tumors have been found to produce HBGF-1 or HBGF-2 (1, 3), they maybe directly involved in the vascularization of tumors(14). However, neither HBGF-1 nor HBGF-2 has conventional signal sequences for secretion (19, 20), and many cells that produce HBGF-1 or HBGF-2 release them at very low levels (21-23). In addition HBGFs have been found to associate with extracellular matrix components (23-25), but the mechanisms by which HBGFs 1 and 2 are translocatedto extracellular spaces remain unknown. Binding proteins for growth factors including epidermal growth factor (EGF) (26, 27), nerve growth factor (NGF) (28, 29), insulin-like growth factors (IGFs) (30-33), insulin (34), platelet-derived growth factor (PDGF) (35), and transforming growth factor-/3 (TGF-/3) (36-38) have been identified, and binding proteins have been suspected to be involved in making HBGF-1 and HBGF-2 available to cell surface receptors (1). Dennis et al. (39) have found that a2-macroglobulin is a binding protein for HBGF-2, but because it appears to bind HBGF-2 irreversibly, a2-macroglobulin is not likely to function asa physiological carrier of HBGF-2. Herewe describe a heparin-binding protein purified from culture medium conditioned by A431 human epidermoid carcinoma cells (40,41), which was observed associated with an HBGF-2 activity produced by these cells. This protein, designated HBpl7, is shown to bind to both HBGF-1 andHBGF-2 in areversible manner and thus may act as a carrier for these HBGFs. The effects of HBpl7 on the biological activities of HBGF-1 andHBGF2 are examined, and the sequence of cloned HBpl7 cDNA is reported. MATERIALSANDMETHODS

Cell Culture and Cell Proliferation Assay-The epidermoid carcinoma cells A431-AJC, a clone of A431 cells (40, 41), and Kane’ were cultured in DME/F-12 nutrient medium. A431-4 (421, a nontumorigenic clonal variant of A431 cells, was obtained from Dr. T. Kawamot0 (Biochemistry Dept., Okayama University Dental School, Okayama, Japan) andwas cultured in DME/F-12 medium supplemented

16778

* T. Okamoto, unpublished results.

Protein

HBGF-binding

16779

with 10% fetal bovine serum (FBS, HyClone Laboratories, Logan, and a model 470 gas phase sequenator (Applied Biosystems, Santa UT) and10 nM EGF (Upstate Biotechnology, Inc., Lake Placid, NY). Clara, CA) as described by Crabb et al. (53). The N-terminal seME-180 (43), A253 (38), SCC-25 (44) and FaDu squamous cell car- quences determined for HBpl7 and peptides P1 to P4 were: KVVSE cinomas, SK-MEL-1 melanoma (45), SK-LMS-1 sarcoma (45), HeLa QKDTLGNTQIKQKSRPGNKGKFVTKDQAN; QKDTLGNTQIK cervical adenocarcinoma (46), MCF-7(47) and BT-20 (48) mammary QKSRPGNK(G)XFV; FSCVFAGNPTSCLXLXDC(E); RVYXKQV adenocarcinomas, DU145 prostate adenocarcinoma (49), and SKCO- A(R); and SSLKLVSSTLFGN, respectively. With 0-lactoglobulin A 1 (45) and HT-29 (45) colon adenocarcinomas were purchased from as a test standard the sequenator gave initial yields of 40% and the American TypeCulture Collection (Rockville, MD) and were repetitive yields in excess of 90%. cultured in DME/F-12 medium with 2% FBS. Normal human keraThe amino acid sequences EQKDTLGN and GNTQIKQK from tinocytes were purchased from Clonetics Corp. (San Diego, CA) and the N terminus of HBpl7 and CVFAGNPT from peptide P2 were cultured in KGM medium from the same supplier. Swiss 3T3 and used to synthesize 23-residue degenerate oligonucleotideprimers with human foreskin fibroblasts were cultured in DME/F-12 medium with a model 381A DNA synthesizer (Applied Biosystems). The primer 10% FBS. pools were:A, 5'-GAXCAXAAXGAYACNYTNGGNAA-3'; B, 5'Cell proliferation was assessed by the incorporation of [3H]thymi- GGNAAYACNCAXATZAAXCAXAA-3'; and 3"ACXCAC, dine into Swiss 3T3 cells. 1 X lo4 cells/well were grown to confluence NAAXCGNCCNTTXGGNTG-5'. X was A or G, Y was T or C, X at 37 "C in 24-well plates in DME/F-12 medium with 10% FBS. The was A, T, or C, and N was any nucleotide. cells were serum starved in nutrient medium for 24 h, and growth Iodination of HBGFs, Receptor Binding Assay, and Affinity Crossfactors were added for an additional 22h. The cells were then linking to Receptor-Bovine HBGF-1 (Upstate Biotechnology, Inc., incubated with 0.5 pCi/ml [3H]thymidine (20 mCi/mmol; Du Pont- Lake Placid, NY) and HBGF-2 (R andD Systems)were labeled with New England Nuclear) for 2 h, washed three times with phosphate- NaIz5I (Amersham Corp.) to specific activities of 5.9 X lo5 and 2 X buffered saline, fixed with 5% trichloroacetic acid, solubilized with IO5 cpm/ng, respectively. The iodination of the HBGFs and the 0.1 N NaOH, and counted by scintillation spectrometry in a model binding of 9 - H B G F - 1 t o HepG2 human hepatoma cells (54) were LS7500 scintillation counter (Beckman Instruments). done as described by Kan et al. (55). Heparin Affinity Chromatography-20 liters of DME/F-12 medium Radiolabeled HBGF-1 and HBGF-2 were covalently cross-linked conditioned by A431-AJC cells were concentrated 10-fold with a to HBpl7with disuccinimidyl suberate (DSS) (Pierce Chemical Co.). hollow fiber concentrator (Amicon, Danvers, MA) and passed over a 4 pg of HBpl7 were incubated with 1 X 10' cpm Iz5I-HBGFin a total 5-ml heparin-Sepharose column (PharmaciaLKB Biotechnology volume of 20 pl at 22 "C for 2 h. DSS was added to a final concentraInc.) a t 4 "C. The column was washed with phosphate-buffered saline tion of 1 mM, and the reaction was incubated at 4 "C for 20 min. and either it was developed with a linear 0.65-2.15 M NaCl gradient in phosphate-buffered saline or it was washed with 0.65 M NaCI, and SDS-PAGE sample buffer was added, the samples were heated to bound material was eluted with 0.95 M NaC1. The eluate was dialyzed 100 "C for 2 min, and thecross-linked complexes were electrophoresed in a 15% polyacrylamide gel by the method of Laemmli (50). against Milli-Q water (Millipore, Bedford, MA) and lyophilized. DNA Amplification by Polymerase Chain Reaction (PCR)-cDNA Gel Filtration and Reverse-phose HPLC-The lyophilized heparin column eluate was dissolved in 0.5mlof 35% acetonitrile (ACN sequences were amplified by PCR as described by Saiki et al. (56). Burdick and Jackson, Muskegon, MI) in 0.1% trifluoroacetic acid First strand cDNA was synthesized from 1 pgof poly(A+) mRNA (Pierce Chemical Co.) and separated on a TSK G4000 SW gel with a Red Module kit (Invitrogen, San Diego, CA). PCR was done filtration column (0.75 X 60 cm) (Phenomenex, Palos Verde, CA) at with a GeneAmp kit (Perkin Elmer-Cetus) using the synthetic olia flow rate of 1ml/min in the same solvent. Material eluting between gonucleotide primers described above. 40 cycles of amplification were M,10,000and 35,000 wasconcentrated by evaporation under vacuum carried out with the following denaturation, annealing, and extension (Speed Vac; Savant Instruments, Farmingdale, NY) and was injected steps: 94 "C X 1 min; 40 "C X 2 min, and 72 "C X 3 min for reactions onto an analytical C-4 RP-HPLC column (0.46 X 25 cm) (Vydac; with degenerate primers or 94 "C X 40 s; 55 "C X 1 min, and 72 "C X The Separations Group, Hesperia, CA). Bound proteins were eluted 2 min for RACE reactions. The final amplification cycle was followed with a linear gradient of 23-33% ACN in 0.1% trifluoroacetic acid in by a 10-minextension reaction at 72 "C. Rapid amplification of cDNA 30 min at a flow rate of 1 ml/min with a model 344 gradient HPLC end (RACE) reactions were done as described by Frohman et al. (57). DNACloningana'Sequencing-PCR products were purified by system (Beckman, San Carlos, CA). In the final purification step HBpl7 was rechromatographed on an analytical C-4 RP-HPLC col- electrophoresis in agarose gels and were blunt end-ligated into the umn (Vydac) that was developed over 30 min with a linear gradient SK Bluescript vector (Stratagene, La Jolla, CA). Recombinant SK of 25-30% ACN in 0.1% trifluoroacetic acid at a flow rate of 1 ml/ plasmids were isolated using a Midi plasmid preparation kit (Qiagen, Studio City, CA), and inserts were sequenced by the dideoxy chain min. Affinity Chromatography of A431 HBGF-2 on Immobilized termination method (58) with a Sequenase version 2.0 DNA sequencHBpl7"Purified HBpl7 was covalently coupled to Affi-Gel 10 (Bio- ing kit (U.S. Biochemical Corp.) Complete HBpl7 cDNA and amino Rad) a t 1mg of protein per ml of gel according to themanufacturer's acid sequences were used with the FASTA search program (59) to instructions. HBGF-2 partially purified from A431-conditioned me- find homologous sequences in the GenBank (release 67; Intellidium by heparin affinity chromatography was bound to the immobi- Genetics, Inc., Mountain View,CA) and PIR (release 26; National lized HBpl7. HBGF-2, detected by its mitogenic activity, was eluted Biomedical Research Foundation (NBRF), Washington, D. C.) data from the HBpl7 column by 1mg/ml heparin. bases. Identification of A431 HBGF-2 by Western Blot-A431-conditioned Southern and Northern Hybridizations-For Southern hybridizamedium was fractionated on a heparin-Sepharose column (Pharma- tions RACE products were separated in 1.2% agarose gels and tanscia) at 4 "C. Protein that eluted in 0.65-2.15 M NaCl was further ferred to nitrocellulose (Micron Separation, Inc., Westboro, MA). fractionated by RP-HPLC on an analytical C-4 column (Vydac) that The membrane was hybridized with a 32P-labeled192-bpPCR product was developed with a linear ACN gradient in 0.1% aqueous trifluo- in 5 X SSPE (1X = 0.18 M NaC1, 10 mM sodium phosphate, pH 7.4, roacetic acid. Material that eluted in 31% ACN and was mitogenic 1 mM EDTA), 5 X Denhardt's solution (lx = 0.02% polyvinylpyrrolfor Swiss 3T3 cells was electrophoresed with purified bovine brain idone, 0.02% Ficoll, 0.02% bovine serum albumin), 50% formamide, HBGF-2 (R and D Systems, Minneapolis, MN) in a 15% polyacrylamide-SDS gel (50) and transferred to nitrocellulose (Schleicher & 0.05% SDS, and 10 pg/ml salmon sperm DNA for 10 h at 42 "C. The Schuell). Membrane-bound HBGF-2 was detected with rabbit anti- membrane was washed three times at 42'C with 1 X SSC (0.15 M bodies to bovine HBGF-2(R and DSystems) using biotinylated NaC1,0.013 M sodium citrate, pH 7.0) containing 0.1% SDS. The airsecond antibodies and streptavidin-alkaline phosphatase (Amersham dried membrane was exposed to Kodak XAR5 film. Poly(A+) RNA extracted from cells with RNAzol (Cinna/Biotex, Corp.) with the chromogenic substrates BCIP and NBT (Bio-Rad). Fragmentation of HBpl7 andPeptide Purification-10 pg of puri- Friendswood, TX) and isolated by chromatography on oligo(dT)tied HBpl7 was digested with Staphylococcus aureus V8 protease (51) cellulose (Collaborative Research, Bedford, MA) was used in Northas described by Wu et al. (52). Four of the resulting peptides were ern hybridizations. 5 pg of each mRNA were electrophoresed in a 1% purified by RP-HPLC on a C-4 column (0.46 X 25 cm; Vydac) that formaldehyde-agarose gel (60) and transferred to a nylon membrane was developed with a linear gradient of 10-30% ACN in 0.1% triflu- (Zeta-Probe; Bio-Rad). Hybridization and washing conditions were the same as those for Southern hybridizations. Poly(A') RNAs from oroacetic acid. HepG2 (551, Hep3B (551, SK-HEP-1 (45), and normal fetal liver cells Amino Acid Analysis and N-terminal Amino Acid SequencingAmino acid analysis and microsequence analysis of HBpl7 and four were kindly provided by Dr. W. L. McKeehan (W. Alton Jones Cell of its peptides were done using a model 420A PTC amino acid analyzer Science Center, Lake Placid, NY).

HBGF-binding Protein

16780 A

FIG. 1. Purification of HBpl7 by HPLC. A , following heparin affinity chromatography and gel filtration HPLC thepooled protein from A431-conditioned medium was fractionated by RP-HPLC on an analytical C-4 column that was developed with a linear gradient of 23-33% ACN in 0.1% trifluoroacetic acid in 30 min a t a flow rate of 1 ml/min. 1-ml fractions were collected and assayed for mitogenic activity on 3T3 cells as indicated by the hatched bars. B, the major absorbance peak, which eluted from the RP-HPLC column in 27% ACN, was rechromatographed on a C-4 column; the columnwas developed with a linear gradient of 25-30% ACN in 30 min a t a flow rate of 1 ml/min. The purified material was resolved into a major silver stained band of M,17,000 by SDS-PAGE in a 15% polyacryl-

RESULTS

Serum-free culture medium conditioned by A431 cells was subjected to heparin affinitychromatography. Heparin-bound proteins that eluted in 0.65-0.95 M NaCl were associated with a considerable level of mitogenic activity when assayed by ["]thymidine incorporation into serum-starved Swiss 3T3 cells (results notshown). This mitogenic activity was further purified by gel filtration HPLC andC-4 reverse-phase HPLC (Fig. lA). In the latter step the mitogenic activity co-eluted with a minor absorbance peak in 31% ACN and was separated from the major absorbance peak, which eluted in 27% ACN. After rechromatography on C-4 the major absorbance peak was resolved as a M , 17,000 band by SDS-PAGE (Fig. 1B); this protein was designated HBpl7. The mitogenic activity peak was assayed for the presence of HBGF-2 since an HBGF2-like activity had been observed in A431-conditioned medium (61) and in A431 cell extracts (62a). An immunoreactive band of approximately M , 18,000, which co-migrated with bovine HBGF-2, was detected by HBGF-2 antibodies in a Western blot (Fig. IC). This material was also mitogenic for human umbilical vein endothelial cells, and it partially inhibited the binding of lZ5I-HBGF-1to HepG2 hepatoma cells (data not shown).Togethertheseresults confirmed the presence of bioactive HBGF-2 in A431-conditioned medium. Since a portionof the HBGF-2 activity in A431-conditioned medium routinely eluted from immobilized heparin in 0.95 M NaCl while bovine brain HBGF-2 characteristically required 2 M NaCl to be released from heparin (12), we considered the possibility that A431 HBGF-2 was associated with a binding protein that was itself a heparin-binding protein. As HBpl7 co-purified with A431 HBGF-2 activity untilthe reversephase HPLC steps, HBpl7 was examined for HBGF binding activity. Immobilized purified HBpl7 was found to retain heparin affinity-purified HBGF-2 activity from medium conditioned by A431 cells (Fig. 2 A ) , and the mitogenic activity could be eluted from the HBpl7column by 1 mg/ml heparin. In addition purified HBpl7 protected the activity ofA431 HBGF-2, which like bovine brain HBGF-2 is acid-labile (l), in a solution of 0.1% trifluoroacetic acid (Fig. 2B). Heparin has similarly been found to protect HBGF-2 as well as HBGF1 from inactivation by acid (62). Cross-linking experiments using the bifunctional reagent DSS were done to investigate further thephysical interactions between HBGFs andHBpl7.HBpl7 could be covalently cross-linked to '"I-HBGF-1 forming a complex of approximately M , 34,000 that could be detected by SDS-PAGE under reducing conditions (Fig. 3). This molecular species did not result from the dimerization of HBGF-1, as it was not detected in the absence of HBpl7 or in the presence of excess unlabeled HBGF-1 (Fig. 3A). In addition the covalent complex was not formed in the absence of DSS under either reducing or nonreducing conditions (Fig. 3B). No DSS-induced covalent complexes between HBpl7 andbovine "TI-HBGF-2 or mouse '*'IEGF orBSA weredetected under conditions inwhich HBGF1 was cross-linked to HBpl7 (datanot shown). Since HBGF2 was retained by an HBpl7 affinity column (Fig. 2), the inability of DSS to cross-link HBGF-2 and HBpl7most likely reflected a lack of sufficiently closely spaced primary amino groups in the noncovalent complex (63). amide gel run under reducing conditions (inset, panel B ) . Edman degradation of purified HBpl7 yielded an N-terminal sequence of KVVSEQKDTLGNTQIKQKSRPGNKGKFVTKDQAN. C , A431 HBGF-2 (lane A ) and bovine brain HBGF-2 (lane B ) were electrophoresed in a 15% polyacrylamide-SDS gel and blotted to nitrocellulose. Membrane-bound HBGF-2 was detected by indirect enzymelinked immunoassay.

Protein A

HBGF-binding

16781

-

0 -

36 29

E,

r L-

10

20

30

40

Volume (ut)

B

-

0

FIG.3. Covalent cross-linking of HBGF-1 to HBpl7. A, '"Ialone HBGF-1 exposed to DSS (lane 1 ) and '*'I-HBGF-l cross-linked with DSS to HBpl7 in the presence ( l a n e 2) or absence ( l a n e 3) of excess unlabeled HBGF-1 were electrophoresed in a 12% polyacrylamide-SDS gel and were detected by autoradiography. B, '*'I-HBGF1 incubated with HBpl7 in the presence (lane 1 ) or absence (lanes 2 and 3 ) of DSS was subjected to SDS-PAGE under reducing (lanes 1 and 2) or nonreducing ( l a n e 3 ) conditions and was detected by autoradiography.

-

0

E

E

r

c 0

I

0

10

20 30 Volume (ul)

40

FIG. 2. Interactions between A431 HBGF-2 and HBpl7. A , heparinaffinity-purified HBGF-2 from A431-conditioned medium was passed over a column of immobilized HBpl7, and the column was eluted with 1 mg/ml heparin. Aliquots of the flow-through (O), eluate (O), andnonfractionated sample (0)were normalized and assayed for the stimulation of [3H]thymidine incorporation by Swiss 3T3 cells. B, A431 HBGF-2 was left untreated (O), or itwas incubated with (0)or without (A) 0.5 pg/ml HBpl7 in 0.1% trifluoroacetic acid (pH 2.0) for 4 h a t room temperature. Aliquots were then assayed for mitogenic activity on Swiss 3T3 cells.

In a functional assay purified HBpl7 inhibited the abilities of both HBGF-1 and HBGF-2 to stimulate theincorporation of [3H]thymidine into 3T3 cells (Fig. 4,A and B ) . 1.5 pg/ml HBpl7 half-maximally inhibited the mitogenic activity of 10 ng/ml HBGF-1, while 12 pg/ml HBpl7 half-maximally inhibited the activity of 0.5 ng/ml HBGF-2. Thus, this effect of HBpl7 appeared more potent for HBGF-1 thanHBGF-2. By contrast, HBpl7did not inhibit PDGF-stimulated DNA synthesis in 3T3 cells (results not shown). The inhibitory effect of HBpl7 on the activities of HBGF-1 and HBGF-2 was not observed in the presence of 10 pg/ml heparin. To determine whether the inhibition of HBGF activity by HBpl7 occurred at the level of ligand-receptor binding lZ5I-HBGF-1was incubated with HepG2 human hepatoma cells in the presence or absence of HBpl7 under conditions inwhich HBGF-1 was not internalized (55). HBpl7 inhibited the binding of radio-

100

rn

z

C

50-

P

B

"

1..'. 0

'

0.1

1

10

1w

Proleln $0 I ml)

FIG. 4. Effects of HBpl7 on the biological activities of HBGFs 1 and 2. Increasing concentrations of HBpl7 were added to quiescent cultures of Swiss 3T3 cells with 10 ng/ml HBGF-1 ( A ) or 0.5 ng/ml HBGF-2 ( B ) in the presence (A) or absence (0)of 10 pg/ ml heparin. After 22 h a t 37 "C, ['Hlthymidine incorporation was assayed during an additional 2-h period. C, the binding of '2sII-HBGF1 to HepG2 cells in the presence of increasing concentrations of HBpl7 was measured a t 4 "C as described by Kan et al. (55).

labeled HBGF-1 to intact HepG2 cells in a concentrationdependent manner indicating that the inhibition of HBGF-1 activity by HBpl7 resulted from decreased ligand binding (Fig. 4C).In thepresence of heparin no inhibition of binding could be detected (data not shown). To determine the entire primary structure of HBpl7 the

16782


PCR method (56) was usedto amplify and clone HBpl7 cDNA from A431-derived poly(A’) mRNA. Four V8 protease-generated fragments of purified HBpl7 were purifiedby RPHPLC (Fig.5A) and sequenced inorderto design PCR primers. None of the N-terminal peptide sequences determined were homologous to GenBankDNA sequences (release 67) or NBRF-PIR protein sequences (release 26). Three pools (A-C) of 23 base degenerate primers were designed based on the N-terminal sequences of HBpl7 and V8 protease-generated fragment P2 (Fig. 5A). The oligonucleotide primer sequences used are described under “Materials and Methods.” Polymerase chain reactions withC primers paired with either A or B primers yielded cDNA products thatdiffered in length by 18 base pairs(Fig. 5B). This size difference was consistent with the known positions in the N terminus of HBpl7 of the peptides corresponding to primerpools A and B. The larger PCR product wasclonedin the SK Bluescript vector and sequenced; the insert consisted of 192 bp and contained sequences corresponding toall three primer pools. The 5‘ and 3’ regions flanking this192-bp cDNA sequencewere obtained by using the RACE technique (57). The 3‘-RACE and 5‘RACE reactions yielded 0.95- and 0.35-kb fragments, respectively, which hybridized to theoriginal 192-bp HBpl7 cDNA (Fig. 6). These fragments were also cloned in the SK Bluescript vector and sequenced. The full-lengthsequences of HBpl7 cDNA and itsdeduced protein product are depicted Fig. in 7. HBpl7 cDNA consisted of 1146 bp excluding the poly(A) tail, and its longest open reading frame encoded 234 amino acids, which included the five amino acid sequences obtained from the N termini of HBpl7 and V8 protease-generated peptides P1 to P4. The first methionine codon of this open reading frame starteda t nucleotide 98 andwas immediately preceeded bythe sequence

3

p 1353. 1078 . 872 603.

310. 271 234.

!!

.

4

,

1353. 1078 . I.’ 872 . 603

.

j

310 271 234

FIG. 6. 5’and 3‘ rapid amplification of cDNA ends (RACE). A 192-bp HBpl7 cDNA fragment was extended in the 5’ and 3’ directions using the RACE protocols of Frohman et a/. (57). The 3’RACE products (lanes I and 2 ) and the 5’-RACE products (lanes 3 and 4 ) were separated ona 1.5% agarose gel and stainedwith ethidium bromide (lanes I and 3) or blotted onto nitrocellulose and hybridized with a 192-bp [”P]HBp17 cDNA fragment.

5’-GCTGCAGCC-3’, which is similar to the consensus translationstart sequence5’-GCCGCCPuCC-3’ of vertebrate mRNAs (64). Thus, this methionine codon most likely initiated the translation of HBpl7. In addition there was an inframe stop codon 90 nucleotides upstream of this methionine codon. The aminoacid sequence from residue to 6 21 following the initiation methionine was very hydrophobic and could form the core of a signal sequence for secretion. Compared with othersignal peptide sequences (65,66), the most probable cleavage site for the HBpl7signal sequence was Ser-33. This prediction was in agreement with the observed N-terminal residue of purified HBpl7 being Lys-34. There were two consecutive stop codonsat HBpl7cDNA nucleotides 800 and 803, which were followed by at least 341 bp of 3”noncoding region. Asingle poly(A)addition signalsequence 5’-AATAAA-3‘ (67) waslocated a t nucleotide 1125. Fromthis analysis we deduced that mature HBpl7was a secreted protein, which was derived from the proteolytic processing of a 234-residue precursor molecule. The HBpl7 precursorhad a single potential N-linked glycosylation site a t Asp-99. Homology searches of the NBRF-PIR and GenBank data bases revealed no sequences homologous toHBpl7orHBpl7 cDNA, respectively. Northernhybridizations with the 192-bp HBpl7 cDNA Fraction fragment as a radiolabeledprobe were used to examine a number of normal and tumor-derived human cells for the presence of HBpl7 mRNA (Fig. 8). A single 1.2-kb mRNA species was detected in A431 poly(A+) RNA, which was consistent with the deduced size of full-length HBpl7cDNA (Fig. 7). This 1.2-kb HBpl7 mRNA was also expressed in normal keratinocytesandinthe five squamous cell carcinomas ME180, FaDu, A253, Kane, and SCC-25 in addition to A431 cells (Fig. 8). HBpl7 mRNA was notdetected in normal human foreskin fibroblasts, fetal liver cells, HepG2 and SKHEP-1 hepatomacells, or MCF-7 mammaryadenocarcinoma FIG. 5. HBpl7 peptides generated with V8 protease and cells (Fig. 8). Also negative were SK-LMS-1 sarcoma cells, synthesis of HBpl7 cDNA by polymerase chain reaction. A, peptides released from reduced and carboxymethylated HRpl7by S. SK-MEL-1 melanoma cells, Hep3B hepatoma cells, and the aureus V8 protease were separated by RP-HPLC on a C-4 column. following adenocarcinoma cell lines: HeLa, DU 145, BT-20, The N-terminal amino acid sequences determined by Edman degra- and HT-29 (data not shown). SKCO-1 colon carcinoma was dation of peaks Pl-P4 are provided under “Materials and Methods.” the only adenocarcinoma testedin which HBpl7 mRNA could B, PCR products synthesized on mRNA templates with primer pools be detected. Interestingly, the A431-4 nontumorigenic variant A and C (lane 1 ) or B and C (lane 2) were separated on a 1.5% of A431 cells (42) did not express detectable levels of HBpl7 agarose gel and stained with ethidiumbromide. Size markers in lane mRNA (Fig. 8). 3 were: 603,310,271,234, 194, and 118 bp.

C


16783

28s.

18s.

OOACCATGM D - HF ES

nnccrorn

TOAQAQAQTC E R V

TATTQQMAC V W K

8

~ c m o c ~ o oCUTCCUOC

CF VAN 0 PS TC

L L KK

D

l

8 -

TCAW~TAA

410

AOCTCMGGA

0

PI

uo

M Q r m O C M OUTCTOCOC TCACAOMAR O V A NR L SR O DK I C R 1

480 ACATCTQTAQ

2

8

P3

ATATTCCAAO ACAQCTQTQA Y S K T A V K T TAAGCTAOTC K L V S S

R

510 MACCAOAQT OTOCAQAMQ V C R K D F P E

AGCTCCACTC T L F Q N T

F4

5m

TAmcoOAA K P R K E (Do

GTCCCCCAGO S P E R H

OAOCACATW ACOOCMAGA I Q K K TE T S P S L

QACCATOOCC T Y A

ACCMAOCTC T K A P

CCOAQTOTQT E C V

OACTGCCCTQ T A L

OAQTTCTQTQ E F C Q E

QAOAQAClTO T W S S C L

ACOTCATOCT

810 AATQAOQTCA

V A

OATITCCAQ

S

CACMAOCCC AQQAAGOAQA K T E N 1 6 8

OICCAOCOCC TCTAQCCTAQ T

O

l

8

'

L

148

600

MACAOAGAT

560 CAQTGACCCA

8

QOAOOACCCA CATATQGCAA E D P D Y A N

720 ACCAOAGGAA a R

~

OAQCTCTCTC T F F

780 TCCTCAOCAT L S I

228

(iD0

7m

FIG. 8. Northern analysis of HBpl7 mRNA expression.

w

AATCCAQTCT

TCWCATTCT

RNAs from the following sources were examined for the expression of HBpl7 mRNA: Kane, A431-4, ME180, A253, FaDu, A431 and SCC-25 squamous cellcarcinomas; HepG2 and SK-HEP-1 hepatoma celllines;MCF-7 mammary adenocarcinoma; SK-CO-1 colon adenocarcinoma; human foreskin fibroblasts ( H F F ) ;and normal human keratinocytes ( H K C ) . 5 pg of poly(A+) RNA from each source were electrophoresed in a 1%agarose gel and blottedto a nylon membrane. The membrane was probed with a 192-bp ["'P]HBpl7 cDNA fragm ment.

Indeed, as HBGF-2 binds to a form of 02-macroglobulin that is rapidly cleared from the circulation, it is likely that 02* 2-34 macroglobulin would mediate the inactivation and elimina170 wo tion of free HBGF-2 released into the plasma (39, 69). A TOTCOTMOT CCCTCTQTAT ACTITMAOC TCTCTACAQT CCCCCCMM T A T O A A C l l l similar role has been proposed for the binding of 02-macroglobulin to PDGF (35) and TGF-fi (36, 37). Inthisstudy we haveisolated from serum-freeculture medium conditioned by A431 human epidermoid carcinoma cells a M , 17,000 heparin-binding protein, designated HBpl7, 1050 that had the attributesof a binding protein for HBGFs 1 and lTCCATOOCC CACAWOCTA TOTQTITQAG CAQCOAAOAQ TCTKQAQCT QAATQADCCA 2. Purified HBpl7 bound HBGF-2(Fig. 2) and HBGF-1 (Fig. 3)in a noncovalentmanner.Theseinteractions could be reversed by either free HBpl7 or heparin, which suggested that the binding of HBpl7 and heparin to HBGFs 1 and 2 TlllTCAMA UJIUIUIAA M A FIG.7. HBpl7 cDNA and the deduced amino acid sequence were antagonistic events. In addition HBpl7 inhibited the of HBpl7. The nucleotide sequence of PCR-derived HBpl7 cDNA biological activities of HBGFs 1 and 2, as measured by the is presented with the deduced primary structure of HBpl7. Amino stimulation of ["Hlthymidine incorporationby Swiss 3T3 cells acid sequences obtained from the N terminus of HBpl7 and four of (Fig. 4), through the inhibition of receptor-mediated ligand its V8 protease-generated peptides are underlined.The sole potential N-linked glycosylation site a t residue 99 is also underlined. Nucleo- binding. The inhibition of HBGF activity by HBpl7 could tides found in the consensus sequence of vertebrate translation ini- also be blocked by heparin. These results might be expected tiation start sites (64) are marked by closed circles, and an in-frame if HBpl7 and heparinwere to bind tooverlapping regions of upstream stop codon is boxed. Asterisks mark two consecutive ter- HBGFs 1 and 2, and they suggest a mechanism mediated by mination codons at the end of the HBpl7 reading frame. A single heparin-like molecules by which HBGFs 1 and 2 could be polyadenylationsignalsequence (67) is ouerlined. This cDNA se- liberated from complexes with HBpl7 after being delivered quence has been assigned GenBank accession number M60047. to their sitesof action. The interactions between HBpl7 and HBGFs 1 and 2 were specific in that HBpl7 did not inhibit DISCUSSION PDGF-stimulated ["]thymidine incorporation by 3T3 cells, and HBpl7 could not becross-linked by DSS to EGF or Binding proteins distinct from cellular receptor molecules have beenidentified in tissues or plasma for peptide hormones bovine serum albumin (data not shown). HBpl7 cDNA was amplified by PCR from A431 mRNA, or growth factors including epidermal growth factor (26, 27), and its sequence was determined. The cloned HBpl7 cDNA nerve growth factor (28, 29), insulin-like growth factors I and I1 (30-33), insulin (34), platelet-derived growth factor (35), consisted of a total of 1146 base pairs followed by a poly(A) tract (Fig. 7). A 192-bp cDNA fragment hybridized to a single transforming growth factor-fi (36-38), and HBGF-2 (basic 1.2-kb speciesof A431 poly(A+) RNA(Fig. 8), which suggested FGF) (39). In general the plasma-derived binding proteins that the complete sequence of HBpl7 mRNA had been observed to inactivate their ligands and thus regulated ligand activity by reducingtheconcentration of freeligand. 02- tained. The cDNAsequence contained asingle long open Macroglobulin, a M , 720,000 serum protease inhibitor (68), reading frameof 234 codons, which began withan ATG codon has been found to act as a binding protein for PDGF (35), within a sequence homologous to the consensussequence for and each of five TGF-fi (36,37), and HBGF-2 (39). Since this serum glycopro- vertebrate translation start sites (64) encoded four tein binds HBGF-2 in an irreversible manner, it does not amino acid sequencesdetermined from intact HBpl7 and a carrier protein for HBGF-2 (39). peptides generatedby V8 protease (Fig. 7).This open reading appear to function as AGTQCAOOAC V O D

MAOAOMCO

GGTTCCTITA

e4a

AGAGATGTCA

16784

Protein

HBGF-binding

frame terminated with two consecutive stop codons. The deduced primary translation product of HBpl7 mRNA was a M, 26,247 precursor molecule with a 33-residue N-terminal signal peptide of Mr 3594 and a single potential N-linked glycosylation site at residue 99. The detection of Asn-99 during the sequencing of HBpl7 peptide P2 (Fig. 7) suggested that this residue was not glycosylated. The HBpl7precursor signal peptide predicted by comparison with known signal sequences (65, 66) coincided with that determined by Nterminal sequence analysis of the mature protein. Nucleotide or amino acid sequences homologous to HBpl7 mRNA and pre-HBpl7 could not be found in the GenBank (release 67) orNBRF-PIR (release 26) databases, respectively, which indicated that HBpl7was a novel protein. The molecular weight of mature HBpl7 estimated from its mobility in SDS-PAGEwas significantly smaller than theM, 22,653 determined from its deduced amino acid sequence. This discrepancy suggested either that HBpl7 assumed a highly compact conformation or that HBpl7was proteolytically processed at its C-terminal end. By amino acid analysis, purified HBpl7 contained no methionine residues (data not shown), which suggested that HBpl7 isolated from A431conditioned medium was truncated prior to Met-168 of the deduced HBpl7 precursor molecule (Fig. 7). Thus, theprecise C terminus of mature HBpl7 is notknown. As mature HBpl7 contained neither hydrophobic sequences of sufficient length to span the plasma membrane nor an endoplasmic reticulum retention signal (70, 71), we concluded that HBpl7 was a secreted protein. Poly(A+)RNA isolated from three normal human cell types and 18 tumor-derived cell lines were examined by Northern analysis for the expression of HBpl7 mRNA (Fig. 8, and results not shown). A single 1.2-kb mRNA species was detected in normal keratinocytes, in A431 cells, and in each of five additional squamous cell carcinomas, butit was not detected in foreskin fibroblasts, fetal liver cells, or in 10 of 11 non-squamous cancer cell lines isolated from diverse tissues. Thus, HBpl7 mRNA expression appeared to be restricted to squamous epithelial cells. The presence of HBpl7 mRNA in normal as well as malignant squamous epithelial cells suggested that HBpl7 was physiologically significant in normal epithelial tissue. The physiologicalroleof HBpl7 in uiuo has yet to be determined. However, as it has been reported that HBGF-2 acts as a mitogen for melanocytes (72, 73) and keratinocytes (74), a possible role for HBpl7 in epidermis is to transport HBGF-2 from keratinocytes, where both molecules are synthesized, to extracellular matrix binding sites or to HBGF receptors on target cells. HBpl7 may also act to transport HBGFs to distant binding sites from extracellular matrix binding sites proximal to the sources of HBGFs. In this role as anHBGF-binding protein HBpl7 may also mediate tumor formation by squamous cell carcinomas by promoting the neovascularization of developing tumor masses. In support of this hypothesis A431 cells transfected with HBpl7 sense cDNA were moretumorigenic in nude mice than control A431 cells, and nontumorigenic A431-4 cells (42), which did not express HBpl7 mRNA(Fig. 8), became tumorigenic when transfected with HBpl7 sense cDNA.~Finally, the identification of a secreted HBGF-binding protein produced by A431 cells,which also synthesize HBGF-2, suggests apotential method by which HBGF-2 could be transported from cells. Acknowledgments-We thank Louise Hartson for technical assistance, Christine Chapline, KarenWest, and Dr. John Crabb for D. Wu, G. H. Sato, and J. D. Sato, unpublished results.

protein microsequencing and amino acid analyses, Dr. Wallace L. McKeehan for supplying mRNA samples, Jeffrey Hall and Dr. Stephen Harris for oligonucleotide synthesis, and Polly Butler, Carol Baine, and Marina LaDuke for graphic art work. REFERENCES 1. Burgess, W. H., and Maciag, T. (1989) Annu. Rev. Biochem. 5 8 , 575-606 2. Finch, P. W., Rubin, J. S., Miki, T., Ron, D., and Aaronson, S. A. (1989) Science 245, 752-755 3. Gospodarowicz, D., Neufeld, G., and Schweigerer, L. (1986) Mol. Cell. Endocriml. 4 6 , 187-204 4. Kan, M., Huang, J., Mansson, P., Yasumitsu, H., Carr, B., and McKeehan, W.L. (1989) Proc. Natl. Acad.Sci. U. S. A. 8 6 , 7432-7436 5. Thomas, K. A,, Rios-Candelore, M., and Fitzpatrick S. (1984) Proc. Natl. Acad. Sci. U. S. A. 8 1 , 357-361 6. Bohlen, P., Baird, A., Esch, F., Ling, N., and Gospodarowicz, D. (1984) Proc. Natl. Acad. Sci. U. S. A. 8 1 , 5364-5368 7. Esch, F.,Ueno, N., Baird, A., Hill, F., Denoroy, L., Ling, N., Gospodarowicz, D., and Guillemin, R. (1985) Biochem. Biophys. Res. Commun. 133,554-562 8. Ruta, M., Burgess, W., Givol, D., Epstein, J., Neiger, N., Kaplow, J., Crumley, G., Dionne, C., Jaye, M., and Schlessinger, J. (1989) Proc. Natl. Acad. Sei. U. S. A. 86,8722-8726 9. Esch, F.,Baird, A., Ling, N., Ueno, N., Hill, F., Denoroy, L., Klepper, R., Gospodarowicz, D., Bohlen, P., and Guillemin, R. (1985) Proc. Natl. Acad. Sci. U. S. A. 82,6507-6511 10. Schreiber, A. B., Kenney, J., Kowselski, W. J., Fried, R., Mehlman, T., and Maciag, T. (1985) Proc. Natl. Acad. Sci. U. S. A. 82,6138-6142 11. Thomas, K. A., and Gimenez-Gallego,G. (1986) Trends Biochem. Sci. 11,81-84 12. Shing, Y., Folkman, J., Sullivan, R., Butterfield, C., Murray, J., and Klagsbrun, M. (1984) Science 2 2 3 , 1296-1299 13. Folkman, J., and Klagsbrun, M. (1987) Science 235,442-447 14. Folkman, J. (1990) J. Natl. Cancer Znst. 8 2 , 4-6 15. D’Amore, P. A. (1986) in Cancer Metastasis: Experimental and Clinical Strategies (Welch, D. R., Bhuyan, B. K., and Liotta, L. A., eds) pp. 269-283, Alan R. Liss, New York 16. Montesano, R., Vasselli, J. D., Baird, A., Guillemin, R., and Orei, L. (1986) Proc. Natl. Acad. Sci. U. S. A. 8 3 , 7297-7301 17. Shing, Y., Folkman, J., Handeschild, L., Lund, D., Crum, R., and Klagsbrun, M. (1985) J. Cell. Biochem. 2 9 , 275-287 18. Lobb, R. R.,Alderman, E. M., and Fett,J. W. (1985) Biochemistry 24.4969-4973 19. Abraham, J. A., Mergia, A., Whang, J. L., Tumolo, A., Friedman, J., Hjerrild, K. A., Gospodarowicz, D., and Fiddes, J. C. (1986) Science 233,545-548 20. Jaye, M., Howk, R., Burgess, W. H., Ricca, G.A., Chiu, I. M., Ravera, M.W., O’Brien, S. J., Modi, W. S., Maciag, T., and Drohan, W. (1986) Science 233,543-545 21. Klagsbrun, M., Sasse, J., Sullivan, R., and Smith, J. A. (1986) Proc. Natl. Acad. Sci. U. S. A. 8 3 , 2448-2452 22. Schweigerer, L., Neufeld, G., Friedman, J., Abraham, J. A., Fiddes, J. C., and Gospodarowicz, D. (1987) Nature 325,251259 23. Vlodavsky, I., Folkman, J., Sullivan, R., Fridman, R., IshaiMicbadi, R., Sasse, J., and Klagsbrun, M. (1987) Proc. Natl. Acad. Sci. U. S. A. 8 4 , 2292-2296 24. Dillario, J., Buffinger, N., Yamada, S., and Strohman, R.C. (1989) Science 244,688-690 25. Baird, A., and Ling, N. (1987) Biochem. Bwphys. Res. Commun. 142,428-435 26. Taylor, J. M., Cohen, S., and Mitchell, W. M. (1970) Proc. Natl. Acad. Sci. U. S. A. 67, 164-171 27. Taylor, J. M., Mitchell, W.M., and Cohen, S. (1974) J. Biol. Chem. 249,3198-3203 28. Varon, S., Nomura, J., and Shooter, E. M. (1968) Biochemistry 7,1296-1303 29. Greene, L. A., Shooter, E. M., and Varon, S. (1969) Biochemistry 8, 3735-3741 30, Zapf, J., Waldvogel, M., and Froesch, E. R. (1975) Arch. Biochem. Biophys. 168,638-645 31. Cohen, K. L., and Nissley, S. P. (1976) Acta Endocrinol. 83,243258


16785

32. Moses, A.C., Nissley, S. P., Cohen, K. L., and Rechler, M.M. J. W., and Sato, J. D. (1989) J. Biol. Chem. 264,17469-17475 (1976) Nature 2 6 3 , 137-140 53. Crabb, J. W., Johnson,M.,C. Carr, S. A., Armes, L. G., and 33. Hintz, R.L., and Liu, F. (1977) J. Clin. Endocrinol. & Metab. 4 6 , Saari, J. C. (1988) J. Bwl. Chem. 263,18678-18687 988-995 54. Aden, D. P., Fogel, A., Plotkin, S., Damjanov, I., and Knowles, B. B. (1979) Nature 282,615-616 34. Hoffman, R., Ristow, H.-J., Vesser, J., andFrank, W. (1973) Exp. Cell Res. 80,275-280 55. Kan, M., DiSorbo, D., Hou, J., Hoshi, H., Mansson, P.-E., and 35. Huang, J. S., Huang, S.S., and Deuel, T. (1989) Proc.Natl.Acad. McKeehan, W. L. (1988) J. Bwl. Chem. 2 6 3 , 11306-11313 Sci. U. S. A . 81,342-346 56. Saiki, R. K., Gelfand, D., Stoffel, S., Scharf, S. J., Higuchi, R., 36. O’Connor-McCourt, M.D., and Wakefield, L. M. (1987) J. Biol. Horn, G. T., Mullis, K.B., and Erlich, H. A. (1988) Science 39,487-491 Chem. 262,14090-14099 37. Huang, S. S., OGrady, P., and Huang, J. S. (1988) J.Biol. Chem. 57. Frohman, M.A., Dush, M. K., and Martin, G. R. (1988) Proc. 263,1535-1541 Sci. Acad. Natl. U. S. A . 86,8998-9002 38. Kanzaki, T., Olofsson, A., Morbn, A., Wernstedt, C., Hellman, 58. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. U., Miyazono, K., Claesson-Welsh, L., and Heldin, C.-H. (1990) Acad. Sci. U. S. A . 7 4 , 5463-5467 Cell 61,1051-1061 59. Pearson, W. R., and Lipman, D. J. (1988) Proc. Natl. Acad. Sci. U. S. A . 86,2442-2448 39. Dennis, P. A., Saksela, O., Haspel, P., and Rifkin, D. B. (1989) J. Biol. Chem. 264,7210-7216 60. Gerard, G. F., and Miller, K. (1986) Focus 8,3-5 40. Girard, D. J., Aaronson, S. A., Todaro, G. J., Arnstein,P., Kersey, 61. Okamoto, T.,Wu, D., Kan, M., and Sato, J. D. (1987) In Vitro J. H., Dosik, H., and Parks, W. P. (1973) J. Natl. Cancer Znst. Cell. & Deu. Biol. 2 3 , 41A 61,1417-1423 62. Masuda, Y., Yoshitake, Y., and Nishikawa, K. (1987) Cell Biol. 41. Fabricant, R. N., DeLarco, J. E., and Todaro, G. J. (1977) Proc. Znt. Rep. 11,359-365 Natl. Acad. Sci. U. S. A. 74,565-569 62a.Gospodarowicz, D., and Cheng, J. (1986) J. Cell. Physiol. 1 2 8 , 42. Buss, J. E., Kudlow, J. E., Lazar, C. S., and Gill, G. N. (1982) 475-484 Proc. Natl. Acad. Sci. U. S. A. 7 9 , 2574-2578 63. Lomant, A. J., and Fairbanks, G. (1976) J. Mol. Biol. 1 0 4 , 24343. Sykes, J. A., Whitescarver, J., Jernstrom, P., Nolan, J. F., and 261 Byatt, P. (1970) J. Natl. Cancer. Znst. 4 6 , 107-122 64. Kozak, M. (1987) Nucleic Acids Res. 16,8125-8148 44. Rheinwald, J. G., and Beckett, M.A. (1981) Cancer Res. 4 1 , 65. Watson, M. E. E. (1984) Nucleic Acids Res. 12,5145-5164 1657-1663 66. von Heijne, G. (1986) Nucleic Acids Res. 14,4683-4690 67. Proudfoot, N. J., and Bronwlee, G.G. (1976) Nature 2 6 3 , 21145. Fogh, J., and Trempe, G. (1972) in Human Tumor Cells in Vitro (J. Fogh, ed) pp. 115-159, Plenum Press, New York 214 68. Harpel, P. C. (1973) J. Exp. Med. 138,508-521 46. Jones, H. W., Jr., McKusick, V.A., Harper, P. S., and Wuu, K.69. Imber, M. J., and Pizzo, S. V. (1981) J. Biol. Chem. 2 6 6 , 8134D. (1971) Obstet. Gynecol. 38,945-949 8139 47. Soule, H. D.,Vazquez, J., Long, A., Alberts, S., and Brennan, M. (1973) J. Natl. Cancer Znst. 6 1 , 1407-1416 70. Munro, S., and Pelham, H. R. B. (1987) Cell 48,899-907 48. Stone, K. R., Mickey, D.D., Wunderli, H., Mickey, G. H., and 71. Pelham, H. R. B. (1989) Annu. Rev. Cell Bwl. 6,1-23 Paulson, D. F. (1978) Znt. J. Cancer 2 1 , 274-281 72. Halaban, R., Ghosh, S., and Baird, A. (1987) In Vitro Cell. & Deu. 49. Lasfargues, E.Y., and Ozzello,L. (1958) J.Natl. Cancer Znst. 21, Biol. 23,47-52 1131-1147 73. Halaban, R., Langdon, R., Birchall, N., Cuono, C., Baird, A., 50. Laemmli, U.K. (1970) Nature 2 2 7 , 680-685 Scott, G., Moellmann, G., and McGuire, J. (1988) J. Cell Bwl. 51. Houmard, J., and Drapeau, G. R. (1972) Proc. Natl. Acad. Sci. U. 107,1611-1619 S. A . 69,3506-3509 74. Shipley, D., G. Keeble, W.,W. Hendrickson, Coffey, J.R.E., J., 52. Wu, D., Wang, L., Sato, G. H., West, K. A., Harris, W. R., Crabb, Jr., and Pittelkow, M. R. (1988) J. Cell. Physwl. 138,511-518