Human basic fibroblast growth factor gene encodes four

of 5 /5
Proc. Natl. Acad. Sci. USA 87 (1990) 2045 Biochemistry. In the article "Human basic fibroblast growth factor gene encodes four polypeptides: Three initiate trans- lation from non-AUG codons" by Robert Z. Florkiewicz and Andreas Sommer, which appeared in number 11, June 1989, ROBERT Z. FLORKIEWICZ*t AND ANDREAS SOMMER Synergen, Inc., 1885 33rd Street, Boulder, CO 80301 In addition, the footnotes in the lower right-hand corner of Proc. Natl. Acad, Sci. USA (86, 3978-3981), the authors request that the following corrections be noted: The author and affiliation lines should read as shown below. should read as shown below. *Present address: The Whittier Institute for Diabetes and Endocri- nology, 9894 Genesee Avenue, La Jolla, CA 92037. tTo whom reprint requests should be addressed. Correction

Embed Size (px)

Transcript of Human basic fibroblast growth factor gene encodes four

Proc. Natl. Acad. Sci. USA 87 (1990) 2045
Biochemistry. In the article "Human basic fibroblast growth factor gene encodes four polypeptides: Three initiate trans- lation from non-AUG codons" by Robert Z. Florkiewicz and Andreas Sommer, which appeared in number 11, June 1989,
ROBERT Z. FLORKIEWICZ*t AND ANDREAS SOMMER Synergen, Inc., 1885 33rd Street, Boulder, CO 80301
In addition, the footnotes in the lower right-hand corner
of Proc. Natl. Acad, Sci. USA (86, 3978-3981), the authors request that the following corrections be noted: The author and affiliation lines should read as shown below.
should read as shown below.
*Present address: The Whittier Institute for Diabetes and Endocri- nology, 9894 Genesee Avenue, La Jolla, CA 92037. tTo whom reprint requests should be addressed.
Correction
Proc. Natl. Acad. Sci. USA Vol. 86, pp. 3978-3981, June 1989 Biochemistry
Human basic fibroblast growth factor gene encodes four polypeptides: Three initiate translation from non-AUG codons
(gene expression/translational initiation/angiogenesis)
ROBERT Z. FLORKIEWICZ*t AND ANDREAS SOMMER* *The Whittier Institute for Diabetes and Endocrinology, 9894 Genesee Avenue, La Jolla, CA 92037; and tSynergen, Inc., 1885 33rd Street, Boulder, CO 80301
Communicated by William B. Wood, February 17, 1989
ABSTRACT Human basic fibroblast growth factor (bFGF) is an angiogenic polypeptide mitogen present in a wide variety of mesoderm- and neuroectoderm-derived tissues. bFGF cDNA and genomic clones predict a 17.8-kDa (155- amino acid) gene product based on the presence of a single putative translational initiator ATG codon. However, a bFGF protein isolated from human placenta contains two additional amino acids NH2-terminal to the predicted initiator methio- nine. We report here that the human cell line SK-HEP-1 coexpresses four molecular forms (17.8, 22.5, 23.1, and 24.2 kDa) of bFGF. The 17.8-kDa bFGF protein is translationally initiated at the previously predicted methionine (AUG) codon, whereas the 22.5-, 23.1-, and 24.2-kDa proteins initiate at unusual non-AUG codons. The higher molecular weight forms are colinear NH2-terminal extensions of the 18-kDa bFGF.
Human basic fibroblast growth factor (bFGF) is a heparin- binding polypeptide mitogen and chemoattractant for mes- enchyme-derived cells (1). It induces the synthesis of latent collagenase and plasminogen activator in capillary endothe- lial cells (2) and is angiogenic in vivo (3). Both cDNA and genomic clones for bFGF have been described (4, 5) and the single-copy gene for human bFGF (FGFB) has been localized on chromosome 4 (6).
All characterized cDNA clones contain one putative initi- ator methionine codon (ATG) from which synthesis of a 155-amino acid (17.8 kDa) bFGF protein species is thought to start (4). However, we have shown that bFGF purified from human placenta has a 2-amino acid extension NH2-terminal to the putative initiating methionine (5). Recently, a 25-kDa bFGF protein has been identified from guinea pig brain extracts (7). In this report, we demonstrate that multiple molecular forms ofbFGF (17.8, 22.5, 23.1, and 24.2 kDa) are coexpressed in the human hepatoma cell line SK-HEP-1. In addition, we have used a SK-HEP-1 cDNA to show that both in vitro transcription-translation and in vivo COS-1 cell expression experiments result in the synthesis of multiple bFGF protein species. Selective mutagenesis of this cDNA indicates that the 17.8-kDa protein is translationally initiated at the previously predicted AUG codon (4), while the 22.5-, 23.1-, and 24.2-kDa proteins initiate translation at non-AUG codons.
MATERIALS AND METHODS Preparation of Cell Lysates and Immunoblot (Western)
Analysis. SK-HEP-1 cells (1 x 109) were lysed in 10.0 ml of lysis buffer containing 50 mM Tris HCl buffer (pH 7.5), 400 mM NaCl, 1 mM MgCl2, 1% Nonidet P-40, and 1 ,uM phenylmethylsulfonyl fluoride. Nuclei and cell debris were removed by centrifugation at 16,000 x g for 10 min at 4°C.
The cell lysate was then purified further by heparin- Sepharose (HS) chromatography as described (8). Fractions from the HS chromatography were analyzed for the presence of bFGF by 12% sodium dodecyl sulfate/polyacrylamide gel electrophoresis (SDS/PAGE) (9) and Western blotting to nitrocellulose (10). The Western blots were probed with affinity-purified rabbit anti-bFGF antibodies raised against human bFGF (5) or against synthetic bFGF peptides. The second antibody used for color development was alkaline phosphatase-conjugated goat anti-rabbit antiserum as de- scribed by the manufacturer (ProtoBlot system, Promega Biotec). In Vitro Transcription-Translation. The entire 1108-
base-pair (bp) human bFGF cDNA (5) was first subcloned into the plasmid vector pGEM-4Z at the unique EcoRI site. The orientation of the insert was determined by standard procedures. mRNA was transcribed according to the manu- facturer's recommendations from Xba I-linearized template by using the phage SP6 promoter and the pGEM in vitro transcription system (Promega Biotec). mRNA transcribed in vitro was confirmed to be of one size class and to be of the correct sense by RNA (Northern) blot-hybridization analysis with 5'- and 3'-specific oligonucleotides as hybridization probes (data not shown). In vitro translation of the mRNA was performed in wheat germ extracts (Bethesda Research Laboratories) in the presence of [35S]methionine as suggested by the manufacturer. We found it was unnecessary to cap the 5' end of mRNA transcribed in vitro prior to in vitro trans- lation in wheat germ extracts. Immunoprecipitations of the reaction mixtures with anti-bFGF antibodies were performed as described (10). Briefly, 1.0 ml ofthe lysis buffer was added to the in vitro translation reaction mixture; then 5 ,ul of rabbit anti-bFGF antiserum was added, and the mixture was incu- bated for 2 hr at 4°C. Protein A-Sepharose was then added, and the mixture was incubated for an additional 30 min at 4°C. The pellet was washed six times with the same lysis buffer (ice cold), and the immunoprecipitates were eluted from protein A-Sepharose with SDS/PAGE sample buffer (9), fractionated by 12% SDS/PAGE, and visualized by fluorog- raphy.
Site-Directed Mutagenesis. Site-directed oligonucleotide mutations at nucleotide positions 199-201 (CTG to CTT) and 364-366 (ATG to GCT) (see Fig. 3) were introduced by using a Bio-Rad mutagen kit and following procedures precisely as described by the manufacturer. The mutation at nucleotide positions 241-243 (CTG to CTT) was constructed by resyn- thesizing a fragment of DNA (by using four overlapping synthetic oligonucleotides) between the Xho I site at nucle- otide 192 and the Apa I site at nucleotide 352. The frame-shift mutation introduced at the unique Apa I site (see inverted triangle in Fig. 3) was constructed by linker insertion with complementary synthetic oligonucleotides so that the nucle-
Abbreviations: bFGF, basic fibroblast growth factor; HS, heparin- Sepharose. tTo whom reprint requests should be addressed.
3978
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Proc. Natl. Acad. Sci. USA 86 (1989) 3979
otide sequence at this site, after insertion, would be 5'- GGCCTCTAGAGCCGGCC-3'. This mutation keeps the up- stream putative CUG start codons unchanged, but as trans- lation proceeds past the oligonucleotide insertion, the reading frame changes by plus one. Since the antibodies used in all of the experiments described were directed against the 17.8-kDa protein, which initiates translation at the AUG codon down- stream from the insert, no immunoreactive signal would be detected for the higher molecular weight proteins (22.5, 23.1, or 24.2 kDa) if their translation were initiated upstream from the linker insertion. If the source of the multiple higher molecular weight proteins was 3' to the inserted oligonucle- otide, then the immunoreactive signal would remain the same. COS Cell Expression. The bFGF "wild-type" and muta-
genized cDNAs were subcloned into the hybrid expression vector pJC119, and COS-1 cells were transfected as de- scribed (11, 12). Briefly, the expression vector pJC119 uti- lizes the simian virus 40 late promoter and polyadenylylation signal. Plates (60 mm) of COS-1 cells were transfected with 10 ,ug of DNA by using 500 tzg of DEAE-dextran per ml for 30 min, followed by a 2-hr incubation in medium supple- mented with 100 ,uM chloroquine; 48 hr after transfection, cell lysates were prepared as described above. COS-1 cell lysates were then incubated with HS for 2 hr at 40C. After centrifugation, the HS pellets were washed three times with buffer containing 20 mM Tris HCl (pH 7.5), 0.5 M NaCl, and 1 ,uM phenylmethylsulfonyl fluoride and then three times with the same buffer containing also 1 M NaCl. Finally, protein was released from the HS resin with SDS/PAGE sample buffer, fractionated by 12% SDS/PAGE, and ana- lyzed for the presence of bFGF proteins by Western blotting as described above.
Bioassay. COS-1 cell lysates were prepared as described above except that heparin-bound protein was eluted from the HS resin in buffer containing 3M NaCl instead ofSDS/PAGE sample buffer. Aliquots from these 3 M NaCl eluates were then diluted into phosphate-buffered saline containing 0.01% gelatin and assayed for stimulation of 3T3 cell DNA synthe- sis. This 3T3 cell mitogenicity assay, as described (7), mea- sures the incorporation of [3H]thymidine into CCl3COOH- precipitable radioactive material.
RESULTS
SK-HEP-1 Cells Synthesize Multiple Forms of bFGF. We examined SK-HEP-1 cell extracts for the presence of multi- ple forms ofbFGF by Western blot analysis (Fig. 1). The data show that SK-HEP-1 cells contain at least three immunore- active species of bFGF that bind to heparin. All three immunoreactive proteins were eluted from HS by >1 M NaCl, a characteristic of all bFGFs previously isolated (13) (Fig. LA, lanes 1-3). We then wanted to determine which amino acid sequences (domains) were conserved among the three immunoreactive bFGF proteins observed. To do this we carried out three competitive Western blot experiments: (0) with anti-bFGF antiserum alone or after preincubation with human recombinant bFGF (Fig. 1B, lanes 1 and 2, respectively); (ih) with anti-peptide antiserum specific for an NH2-terminal domain of bFGF alone or after preincubation with the corresponding synthetic peptide (Fig. 1C, lanes 1 and 2, respectively), and (iii) with anti-peptide antiserum specific for a COOH-terminal domain alone and after prein- cubation with the corresponding synthetic peptide (Fig. 1C, lanes 3 and 4, respectively). In each case, immunoreactivity was eliminated after preincubating the affinity-purified anti- sera with either recombinant bFGF or the corresponding synthetic peptide, respectively. The data show that all three
A
US
B C
1 2 3 4
FIG. 1. Western blot analysis of bFGF proteins from SK-HEP-1 cells. (A) SK-HEP-1 lysates probed with affinity-purified anti-bFGF antibodies. Lanes: 1, cell lysate before HS chromatography; 2, 1 M NaCI HS eluate; 3, 3 M NaCl HS eluate. (B) Identical SK-HEP-1 lysates probed with affinity-purified bFGF antibodies alone (lane 1) or after preincubation with 25 ,ug of recombinant bFGF (lane 2). (C) SK-HEP-1 lysates probed with affinity-purified anti-peptide-(40-63) antibodies alone (NH2-terminal) (lane 1) and after preincubation with 25 ,g of peptide-(40-63) (lane 2) or probed with affinity-purified bFGF anti-peptide-(147-153) antibodies alone (COOH-terminal) (lane 3) and after preincubation with 25 Ag of peptide-(147-153) (lane 4). Molecular masses are given in kDa.
immunoreactive proteins share similar, if not identical, NH2- terminal and COOH-terminal domains.
Multiple Species of-bFGF Are Synthesized in Vitro. We wanted to determine if the multiple bFGF species were encoded within a previously characterized bFGF cDNA that had been cloned from SK-HEP-1 poly(A)+ RNA (5). To do this, we subcloned that cDNA into a standard in vitro transcription system. The single class size mRNA tran- scribed in vitro was then used for translation by using a wheat germ in vitro translation system followed by immunoprecip- itaton and analysis by SDS/PAGE. The data in Fig. 2 show that the cDNA used to generate mRNA in vitro contains the coding information required for the synthesis of at least three immunoprecipitable species (17.8, 22.5, and 24.2 kDa) of bFGF. The immunoprecipitable band at 16 kDa is most likely a bFGF degradation product. DNA Sequence Predicts Translation Initiation at Upstream
CTG Codons. We examined the possibility that codons other than AUG were used to initiate translation of the higher molecular weight forms of bFGF because (0) sequence anal- ysis of the bFGF cDNA clone showed that it contained only one putative ATG translational initiator methionine codon (4, 5), (ii) alternative RNA splicing was not required to generate the multiple forms ofbFGF (data presented in Fig. 2), and (iii) pulse-chase experiments with SK-HEP-1 cells indicated that there is no precursor-product relationship between any ofthe multiple bFGF species observed (data not shown).
Fig. 3 shows the relevant nucleotide sequence of the bFGF
68 -
2
FIG. 2. In vitro translation of mRNA transcribed from the wild- type bFGF cDNA. RNA-depen- dent in vitro translations were per- formed in wheat germ extracts and immunoprecipitated as described. Immunoprecipitated samples were eluted from protein A-Sepharose, resolved on 12% SDS/PAGE, and visualized by fluorography. Lanes: 1, protein standards; 2, immuno- precipitated proteins from in vitro translation. Molecular masses are given in kDa.
Biochemistry: Florkiewicz and Sommer
3980 Biochemistry: Florkiewicz and Sommer
181 OTT] CGG CCG AGC GGC TCG AGG CTG GGG GAC CGO
Leu GGG CGC GGC CGC GCG CTG COG GGC GGG AGG [Leu] 250LCTTJ25 CTG GGGGGCCGG GGCCGG GGCCGTCCC CCG LOu 300 GAG CGG GTC GGA GGCOCGG GGCCGG GGCOCGG
GGG ACG GCG GCT CCC CGC GCG GCT CCA GCG , ~~~35C
GCT CGG GGA TCC CGG CCG GGC 0CC GCA GGG A/l1
.GCT] ACO ATGGCA bFGF TGA
Met Stop
FIG. 3. bFGF cDNA sequence analysis. The relevant sequence of the wild-type bFGF cDNA clone is shown with the position of site-directed mutations indicated in bold letters. The frame-shift mutation is indicated by the inverted triangle. The construction of all mutations is described in Materials and Methods.
cDNA used to generate mRNA for in vitro translation. When we considered the literature describing alternative transla- tional initiation codons (14-17), synthesis of a 24.2-kDa and 22.5-kDa bFGF protein could initiate translation at CTG codons in nucleotide positions 199-201 and 241-243, respec- tively, whereas synthesis of the 17.8-kDa bFGF would begin at the ATG codon in nucleotide positions 364-366. All three codons are contained within a nucleotide sequence described by Kozak (14) as most preferred for eukaryotic translational initiation, while another in-frame CTG codon at nucleotide position 228, which could initiate translation of a 23.1-kDa bFGF protein, is considered less preferred. Mutations were made as described in Materials and Methods.
Mutagenesis and in Vivo COS-1 Expression Show Multiple CTG Translation Initiations. To test the hypothesis that CTG codons could initiate translation of the 24.2-kDa and 22.5- kDa bFGF proteins, we conducted in vivo expression exper- iments using COS-1 cells transfected with the hybrid expres- sion vector pJC119 containing the following bFGF cDNAs: (i) a wild-type cDNA previously used for in vitro translation (see Fig. 2); (ii) a wild-type cDNA inserted in the incorrect orientation; (iii) a frame-shift mnutant with an altered (+1) reading frame beginning at nucleotide position 353 (between the CTG codons and the first ATG codon); (iv) a Guo-201 --
dThd mutation changing CTG codon to CTT; (v) a Guo-241 -+ dThd mutation changing CTG codon to CTT; and (vi) mutations at positions 364-366 changing ATG (methionine) to GCT (alanine).
Extracts from COS-1 cells transfected with these bFGF cDNA/pJC119 constructs were analyzed by Western blotting with anti-bFGF antibodies (Fig. 4A). A standard recombinant bFGF 17.8-kDa marker is presented in lane 1. COS-1 cells transfected with the wild-type cDNA (inserted into pJC119 in the correct orientation) synthesized at least three forms of bFGF (lane 5) with electrophoretic mobilities identical to the bFGF proteins expressed by the human SK-HEP-1 cells (see Fig. 1). This result supports the conclusion derived from the in vitro transcription-translation data, that the wild-type bFGF cDNA, although containing only one putative trans- lational initatior methionine codon, also contains the infor- mation required for the synthesis of at least two additional molecular forms of bFGF. COS-1 cells transfected with the wild-type cDNA inserted into the expression vector in the incorrect orientation (lane 7) or COS-1 cells mock- transfected (lane 8) do not synthesize bFGF. COS-1 cells transfected with the frame-shift mutation described in Fig. 3
1 2 3 4 5 6 7 8
0T-
bFGF, ng
FIG. 4. COS-1 cell expression (A) and bioactivity (B) of the wild-type and mutagenized bFGF cDNA. Lysates from the trans- fected cells were partially purified by HS chromatography and analyzed by SDS/PAGE and Western blotting with affinity-purified anti-bFGF antibodies (A). Lanes: 1, recombinant bFGF 17.8-kDa marker; 2, Guo-201 - dThd mutation (codon CTG to CTT at 199- 201); 3, Guo-243 -> dThd mutation (codon CTG to CTT at 241-243); 4, frame-shift mutation; 5, wild-type cDNA in the correct orientation; 6, ATG to GCT mutation at nucleotides 364-366; 7, wild-type cDNA in the incorrect orientation; and 8, mock-transfected COS-1 cells. Molecular masses are given in kDa. (B) HS-bound protein was eluted with 3 M NaCl and then assayed by the 3T3 cell mitogenicity assay. Quantitative Western blot analysis of these samples indicated a bFGF concentration of 1 ng/Al of 3 M NaCl eluate. *, Wild-type correct orientation; *, wild-type incorrect orientation; A, frame-shift mutation; a, ATG to GCT mutation; o, control COS-1 cells non- transfected.
express only the 17.8-kDa bFGF protein (Fig. 4A, lane 4). This result indicates that the 22.5-kDa and 24.2-kDa proteins are translationally initiated 5' to the ATG codon (nucleotide positions 364-366) and that the synthesis of the 17.8-kDa protein is, in fact, initiated at ATG at 364-366.
Direct mutagenesis of this ATG to GCT supports this interpretation, resulting in synthesis ofonly the 22.5-kDa and 24.2-kDa bFGF proteins (Fig. 4A, lane 6). The mutant cDNA containing a mutation at nucleotide position 201 giving codon CTT and eliminating the CTG expected to initiate synthesis of the 24.2-kDa protein expresses only the 22.5-kDa and 17.8-kDa proteins (Fig 4A, lane 2). This result implicates the CTG codon (positions 199-201) in the synthesis of the 24.2-kDa protein species of bFGF. Mutagenesis of codon CTG at nucleotide position 243, expected to eliminate syn- thesis of the 22.5-kDa protein, revealed a previously uniden-
A
24.2-m
."No: 4'm am ,W40 40 "m
.40""Am a* a*
Proc. Natl. Acad. Sci. USA 86 (1989) 3981
tified immunoreactive protein species-actually the top band of a 22.5/23.1-kDa doublet (Fig. 4A, lane 3). DNA sequence analysis of this mutant confirmed that the only nucleotide change was, as by design, at position 243, changing the codon at 241-243 from CTG to CTT. We suggest that, although the CTG at nucleotide position 226-228 (see Fig. 3) has a less preferred Kozak consensus sequence than that at 199-201 or at 241-243 or ATG at 364-366, it nonetheless appears to be recognized as a functional translational initiator codon (in vivo) for the synthesis of a 23.1-kDa bFGF protein species. During review of this manuscript we demonstrated that simultaneously mutagenizing both CTG codons at nucleotide positions 226-228 and 241-243 to CTJ followed by COS-1 cell expression results in the synthesis of only the 24.2-kDa and 17.8-kDa proteins as suggested above.
Bioactivity. The bFGF protein species synthesized by transfected COS-1 cells are bioactive in a 3T3 cell mitoge- nicity assay (Fig. 4B). All cell lysates tested were first purified by HS chromatography. The 3 M NaCl eluate of HS-purified lysates from control (nontransfected) COS-1 cells or from COS-1 cells transfected with the bFGF cDNA inserted in the incorrect orientation were not stimulatory in the mitogenicity assay. All other 3 M NaCl eluates from wild type as well as mutants synthesizing only the higher molec- ular weight proteins (24.2,23.1, and 22.5 kDa) or synthesizing only the 17.8-kDa protein stimulated nearly identical levels of [3H]thymidine incorporation. For all samples tested, mito- genic activity was neutralized by preincubation with anti- bFGF antibodies (data not shown).
DISCUSSION We have identified four molecular forms of human bFGF coexpressed by the human hepatoma cell line SK-HEP-1. Based on SDS/PAGE analysis and the predicted amino acid sequence, the four proteins have apparent molecular weights of 24.2, 23.1, 22.5, and 17.8 kDa. We have shown by in vivo expression of frame-shift and oligonucleotide-directed muta- tions that the synthesis of bFGF 24.2, 23.1, and 22.5-kDa proteins begins at non-AUG translational initiation codons. Thus, we demonstrate a normal animal cell gene encoding multiple colinear extended forms of the same protein in vivo, that these proteins are synthesized from one mRNA species, and that some initiate translation at non-AUG codons. Al- though the CTG codons described have reasonably good context in the +4 and -3 positions as described by the classic Kozak consensus sequence, it is not clear what nucleotide sequence around certain CTG codons affects the associated ribosome scanning model and translational initiation. We are beginning a more extensive mutagenesis program in an effort to understand this particular exception to the rules. In related experiments, primarily using in vitro translations, one form of
the MYC protooncogene has also been shown to utilize a non-AUG codon for translational initiation (16).
Since cell extracts containing the higher molecular weight forms of bFGF appear to be as bioactive as extracts con- taining only the 17.8-kDa protein species, the physiological function of the multiple molecular weight bFGF proteins remains to be determined. Differences in their intracellular sorting, transport, or targeting may be one potential function. The multiple molecular forms ofbFGF could be differentially stored, each to be released from the cell in response to specific physiological signals; alternatively, the multiple bF- GFs may individually localize to different subcellular com- partments and modulate cell growth and differentiation. We have described a system that will allow the generation of mutants that produce only one specific molecular form of bFGF; such mutants can be used to determine the intracel- lular location and the physiological function of the individual molecular forms of bFGF.
We thank D. Abbott-Brown for establishing the bioassay used in our lab; T. Gleason and R. Weaver for oligonucleotide synthesis; B. Cooley for peptide synthesis and affinity purification of anti-bFGF antisera; R. Green for valuable discussions regarding in vitro trans- lations; T. Klein, C. Worland, and D. Higgins for preparation of the manuscript; and D. Hirsh, R. Thompson, R. Majack, and A. Flexer for critical review of this report.
1. Gospodarowicz, D. (1987) Nucl. Med. Biol. 14, 421-434. 2. Moscatelli, D., Presta, M. & Rifkin, D. B. (1986) Proc. NatI.
Acad. Sci. USA 83, 2091-2095. 3. Folkman, J. & Klagsbrun, F. M. (1987) Science 235, 442-447. 4. Abraham, J. A., Whang, J. L., Tumolo, A., Mergia, A., Fried-
man, J., Gospodarowicz, D. & Fiddes, J. (1986) EMBO J. 5, 2523-2528.
5. Sommer, A., Brewer, M. T., Thompson, R. C., Moscatelli, D., Presta, M. & Rifkin, D. B. (1987) Biochem. Biophys. Res. Commun. 144, 543-550.
6. Mergia, A., Eddy, R., Abraham, J., Fiddes, J. & Shows, T. B. (1986) Biochem. Biophys. Res. Commun. 138, 644-651.
7. Moscatelli, D., Joseph-Silverstein, J., Manejias, R. & Rifkin, D. B. (1987) Proc. Natl. Acad. Sci. USA 84, 5778-5782.
8. Squires, C. H., Childs, J., Eisenberg, S. P., Poverini, P. J. & Sommer, A. (1988) J. Biol. Chem. 263, 16297-16302.
9. Laemmli, U. K. (1970) Nature (London) 263, 789-798. 10. Florkiewicz, R. Z., Smith, A., Bergmann, J. E. & Rose, J. K.
(1983) J. Cell Biol. 97, 1381-1388. 11. Machamer, C. E., Florkiewicz, R. Z. & Rose, J. K. (1985)
Mol. Cell. Biol. 5, 3074-3983. 12. Rose, J. K., Adams, G. A. & Gallione, C. J. (1984) Proc. Natl.
Acad. Sci. USA 81, 2050-2054. 13. Lobb, R. R., Harper, J. W. & Fett, J. W. (1986) Anal. Bio-
chem. 154, 1-14. 14. Kozak, M. (1986) Cell 44, 283-292. 15. Peabody, D. S. (1987) J. Biol. Chem. 262, 11847-11851. 16. Hann, S. R., King, M. W., Bentley, D. L., Anderson, C. W. &
Eisenman, R. N. (1988) Cell 52, 185-195. 17. Curran, J. & Kolakofsky, P. (1988) EMBO J. 7, 245-251.
Biochemistry: Florkiewicz and Sommer