(Received for publication, January 2, 1996, and in revised form, October 3, 1996)
From the Central Research Laboratory, Nippon Suisan Kaisha Limited, 559-6 Kitanomachi, Hachioji, Tokyo 192, Japan
Here, we describe the isolation of a rainbow
trout cDNA clone that contains the entire fibroblast growth
factor-2 (FGF-2; basic FGF) coding region. Interestingly, the rainbow
trout cDNA contains a translation stop codon just upstream of the
primary initiating methionine codon and so cannot give rise to the
longer forms of FGF-2 that are produced in mammals by alternative
translation initiation at leucines farther upstream. Transfection of
human FGF-2 cDNA into fish cells shows that fish cells can initiate protein synthesis at an upstream leucine CUG codon; surprisingly, however, synthesis is initiated only at the most 5 CUG and not at the
two subsequent CUG codons or the methionine AUG codon also used in
mammalian cells. Like other FGF-2 proteins, bacterially produced
rainbow trout FGF-2 protein binds tightly to heparin-Sepharose and also
promotes the proliferation of fibroblast cells. However, the protein
differs from all others previously identified at amino acids 121-123,
which are part of the proposed highly conserved receptor-binding
domain. Comparisons of the efficacies of recombinant wild-type and
mutant rainbow trout FGF-2 proteins demonstrate that these three amino
acids are critical to the activity of FGF-2.
Fibroblast growth factor-2 (FGF-2 1; also known as basic fibroblast growth factor) is a member of a family that now comprises nine related proteins sharing 30-55% sequence homology (1) and heparin binding affinity. Acidic FGF (FGF-1) and basic FGF (FGF-2) are the prototypes. Other members of the FGF family include the oncogene products FGF-3 (Int-2), FGF-4 (Hst/K-FGF), FGF-5, FGF-6, and FGF-7 (keratinocyte growth factor) and the more recently identified FGF-8 (androgen-induced growth factor) (2) and FGF-9 (glial activating factor) (3). FGF-2 has been cloned from human, bovine, mouse, rat, and Xenopus (4-9). The growth factor promotes the proliferation and differentiation of a wide range of mesoderm- and neuroectoderm-derived cells and, in Xenopus, is a potent inducer of mesoderm formation in developing embryos (9, 10). The strong mitogenic effect of FGF-2 on many cell types has suggested its use as a therapeutic drug in several clinical situations associated with tissue regeneration or repair (11, 12).
Two classes of FGF-2 cell-surface receptors have been described. The "low affinity" class has been identified as a heparan sulfate proteoglycan (13-15), whereas the "high affinity" class consists of at least four members, designated FGF receptors 1-4 (16, 17). Each member of this latter class has an immunoglobin-like structure and an intracellular tryosine kinase domain. Identifying the structure of the FGF-2 proteins that determines their interaction with these receptors has been of great interest. Two regions, corresponding to amino acids 33-77 and 115-124 in human FGF-2, have been proposed to be involved in the receptor binding and mitogenic activity of this growth factor (18). Three-dimensional structural studies, as well as other data, have indicated that the region corresponding to amino acids 115-124 in human FGF-2 is the core sequence required for the binding of FGF-2 to its receptors (19-21). In support of this assignment, this region is highly conserved among all the FGF-2 sequences obtained to date (4-9).
It has been known that several forms of FGF-2 are synthesized as a result of alternative translation initiation at an AUG start codon or two or more CUG codons located farther upstream (7, 22, 23). Subcellular localization studies have demonstrated that the AUG-initiated form is cytoplasmic, whereas the CUG-initiated forms contain a nuclear localization sequence and are mostly recovered in the nucleus (22, 24). Both forms are biologically active (25, 26), but differences in their functions have not yet been discerned.
We have undertaken the isolation of the rainbow trout homologue of FGF-2 to study its potential role in fish embryonic development, to further explore structure-function relationships through sequence comparisons, and to exploit its activities for fish cultivation. The cDNA we isolated encodes a 17.3-kDa protein that is highly homologous to mammalian FGF-2, but lacks the alternative translation start sites that give rise to the nuclear targeted forms of mammalian FGF-2 and also differs at amino acids 121-123 in the putative receptor-binding domain. To test the significance of this latter difference, we have used site-directed mutagenesis to change these amino acids in rainbow trout FGF-2 to those of mammals or to a completely different sequence and have tested their biological activities with fish and mammalian cell lines.
To isolate the rainbow trout FGF-2 cDNA, a probe was
first generated by PCR. Oligonucleotide primers were designed based on the amino acid sequence of the highly conserved region of FGF-2 and
were then used to amplify rainbow trout testis cDNA as described previously (27). Products of the expected size were subcloned into the
pSK vector (Stratagene) and sequenced. The primers that yielded a
357-bp fragment highly homologous to human (68.4%) and Xenopus (68.3%) FGF-2 proteins were as follows: primer 1, TA(T/C)TG(T/C)AA(A/G)AA(T/C)GG(A/G/C/T)GG(A/G/C/T)TT(T/C)TT; and
primer 2, CAT(A/G/C/T)GG(A/G/C/T)A(A/G)(A/G)AA(A/G/C/T)A(A/G)(A/G/T)AT(A/G/C/T)GC(T/C)TT(T/C)TG. A rainbow trout testis cDNA library in gt10 (random
hexanucleotide-primed) was screened by hybridization with a probe made
from the cloned PCR fragment.
Reverse transcription-PCR was conducted as described previously (28) with some modifications. Briefly, poly(A)-selected RNA was prepared from various rainbow trout tissues and from a cell line (RTG-2) derived from rainbow trout ovary (29). The cDNAs were synthesized using a first strand synthesis kit (Stratagene). An aliquot of the first strand reaction was then subjected to PCR analysis using rainbow trout FGF-2-specific primers F1 and F2, which generated a 233-bp product: primer F1, AGACGGACAACAGATGAATG (positions 706-725); and primer F2, CAGCCCAGGCTGAGTTTTAA (positions 938 to 919).
Confirmation of the Upstream Stop CodonTo confirm the presence of the upstream stop codon, an aliquot of the first strand reaction (from brain, ovary, and RTG-2) was subjected to PCR using primers F3 and F4, which generated a 663-bp product spanning the first methionine codon: primer F3, TTCGTGATAACATTTCAGCG (positions 43-62); and primer F4, CGGTCCAAACAGTCTTCCAT (positions 705 to 684). The resulting PCR fragments were subcloned into the pSK vector and then sequenced.
Western Blotting for Determination of Rainbow Trout FGF-2Tissues were homogenized in lysis buffer (10 mM Tris-HCl, pH 7.5, 0.1% Triton X-100) in the presence of protease inhibitors as instructed by the supplier (Boehringer Mannheim). After clarification by centrifugation, the extracts were applied to heparin-Sepharose CL-6B (Pharmacia Biotech Inc.). The column was washed with lysis buffer containing 0.5 M NaCl and eluted with the buffer containing 2 M NaCl. The eluates were dialyzed against distilled water at 4 °C and lyophilized. Dissolved samples or cultured cells solubilized in the same buffer were electrophoresed on a 16% SDS-polyacrylamide gel and transferred to a polyvinylidene difluoride membrane (Millipore Corp.). After blocking, the membranes were immersed in a solution containing antiserum from a rabbit immunized with recombinant rainbow trout FGF-2 that had been purified from Escherichia coli as described below. The membranes were then reacted with alkaline phosphatase-conjugated anti-rabbit IgG (Sigma), and the reacted bands were visualized by nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate.
Transient FGF-2 Expression in Cultured CellsA rainbow
trout FGF-2 expression plasmid (pCMV-rtFGF-2) was constructed by
inserting the 475-bp EcoRI fragment described below into the
EcoRI site of pcDNA3 (Invitrogen). A human FGF-2
expression plasmid (pCMV-hmFGF-2) was constructed by inserting the DNA
fragment extending from the first methionine codon to the stop codon of human FGF-2 cDNA (purchased from R&D Systems, Oxon, United Kingdom) into pcDNA3. A human FGF-2 expression plasmid (pCMV-hfFGF-2)
carrying full-length human FGF-2 cDNA (from plasmid pcDNA3-14
(23)) was a kind gift of Dr. R. Z. Florkiewicz. The schematic
structures of the expression plasmids are shown in Fig. 4A.
5 × 105 RBCF-1 cells, derived from goldfish fin (30),
or COS-7 cells (31) were inoculated into 60-mm culture dishes. After
16 h at 37 °C, the cells were transfected with 2 µg of
expression plasmids as described previously (27). After 2 days, the
cells were analyzed by Western blotting.
Bacterial Expression and Purification of Rainbow Trout FGF-2
A set of primers was generated bearing the N- and
C-terminal ends of the rainbow trout FGF-2 coding sequence flanked by
an EcoRI + NcoI and EcoRI site,
respectively. These primers were used to generate a 474-bp PCR fragment
containing the entire open reading frame, which was then subcloned into
the pET-15e (32) NcoI and EcoRI sites. The
resulting plasmid, pET-rtFGF-2, was transformed into E. coli
DE21(DE3) cells harboring the bacteriophage T7 RNA polymerase gene,
which is under lac control. Cells were harvested by
centrifugation 3 h after 0.5 mM
isopropyl--D-thiogalactopyranoside induction and
suspended in 0.1 volume of buffer A (10 mM Tris-HCl, pH
7.5, 1 mM EDTA, 1 mM dithiothreitol). The
suspension was sonicated briefly; debris was removed by centrifugation;
and the supernatant was loaded onto a column of heparin-Sepharose
CL-6B. The column was then eluted by increasing the NaCl concentration
in buffer A from 0.2 to 2.0 M in 0.2 M
steps.
FGF-2 mutants were generated by deoxyoligonucleotide-directed mutagenesis as described previously (33, 34). The template DNA for mutagenesis was prepared from M13mp19 containing the cDNA encoding rainbow trout FGF-2, harboring the NcoI site at the translation initiation site. Mutants were confirmed by DNA sequencing. For bacterial expression, replicative form DNA was digested by NcoI and EcoRI and subcloned into NcoI/EcoRI-digested pET-15e. The purification of mutant FGF-2 proteins was performed as described above.
Mitogen Assay for Rainbow Trout FGF-2Both RBCF-1 cells and BALB/3T3 cells (35) were maintained at 37 °C in Dulbecco's modified Eagle's/Ham's F-12 medium (1:1) supplemented with 10% fetal calf serum. The mitogenic activity of FGF-2 was determined by the measurement of bromodeoxyuridine incorporation into the cultured cells. The incorporation assay was performed using serum-free Dulbecco's modified Eagle's/Ham's F-12 medium (1:1) supplemented with 1 mg/ml bovine serum albumin, 0.1 mg/ml aprotinin, 10 mg/ml transferrin, and 5 mg/ml fibronectin (36). Cells were harvested by trypsinization, rinsed once, resuspended with the serum-free medium, and inoculated into collagen-coated 96-well microplates at 4 × 103 cells/well with various concentrations of rainbow trout FGF-2 proteins or recombinant human FGF-2 (Boehringer Mannheim). After 2 days, cells were pulse-labeled with bromodeoxyuridine for 5 h, and bromodeoxyuridine incorporation was measured using an enzyme-linked immunosorbent assay kit as instructed by the supplier (Boehringer Mannheim).
As a first step in isolating the rainbow trout FGF-2 cDNA clone, we designed a set of degenerate oligonucleotide primers homologous to an amino acid sequence that is highly conserved among various species. These primers were then used to generate PCR fragments from rainbow trout testis cDNA to isolate a fragment that would serve as a screening probe. A PCR product of the length predicted from the human FGF-2 gene was isolated, cloned, and sequenced. The sequence encoded 115 amino acids that had 68.4% identity to the corresponding region of human FGF-2.
Using a probe generated from this fragment, ~5 × 105 plaques from a rainbow trout testis cDNA library were screened, and one strongly hybridizing clone was obtained. cDNA thus obtained was sequenced.
Structure of Rainbow Trout FGF-2The rainbow trout cDNA
contains an ATG initiation codon followed by an open reading frame that
encodes a 155-amino acid protein with a molecular mass of 17,317 Da.
The amino acid sequence has a >70% homology to all previously
isolated FGF-2 proteins and shows complete continuity with each, except
for the single amino acid missing from the mouse sequence (Fig.
1). As with other FGF-2 proteins, homology is higher in
the C-terminal region, which contains the putative receptor-binding
domains (18). Homology is also high between the rainbow trout and human
FGF-2 N-terminal domains, including a dearth of hydrophobic residues,
which indicates the absence of a classical signal peptide.
Rainbow Trout FGF-2 Lacks an Alternative N-terminal Variant
A
novel feature of the rainbow trout FGF-2 cDNA is the absence of an
upstream CUG codon, which could serve to initiate an alternative
N-terminal form of the protein like the nuclear targeted versions found
in mammalian cells. Instead, a TAG stop codon is present 15 bases
upstream of the initiating ATG codon. To test whether this stop codon
might be an artifact of cloning or particular to the tissue (testis)
used for the cDNA library, PCR fragments spanning this codon were
amplified and sequenced from rainbow trout brain, ovary, and RTG-2
cDNAs. The presence of the stop codon was confirmed in each case
(data not shown). To check further for the presence of an N-terminal
variant, Western blot analysis was performed using anti-rainbow trout
FGF-2 antibody. Western blot analysis of heparin-binding proteins from
ovary and testis detected a primary band with a relative molecular mass
of 18 kDa (Fig. 2, lanes 2 and 3).
This band comigrated with the recombinant rainbow trout FGF-2 used as a
standard (Fig. 2, lane 1). No bands were detected in the
22.5-24-kDa range, corresponding to the sizes of the mammalian FGF-2
variants. Thus, we conclude that rainbow trout FGF-2 lacks the
alternative N-terminal initiation variants found for mammalian
FGF-2.
Tissue Distribution of Rainbow Trout FGF-2 mRNA
We used
the highly sensitive reverse transcription-PCR technique to survey the
tissue distribution of FGF-2 mRNA in rainbow trout.
Expression could be detected in all tissues tested as well as the RTG-2
cell line (Fig. 3), with the highest levels observed in
brain, testis, and the RTG-2 cell line. No amplification was seen when
the avian myeloblastosis reverse transcriptase was omitted from the
cDNA synthesis reaction (data not shown).
Alternative Translation Initiation in Fish Cells
It is
possible that rainbow trout FGF-2 lacks upstream initiating CUG codons
because fish cells are unable to recognize them. To test this, a
plasmid carrying the full-length human FGF-2 cDNA under control of
the cytomegalovirus promoter (pCMV-hfFGF-2) was transfected into the
RBCF-1 cell line derived from goldfish and, as a positive control, into
COS-7 cells. As shown in Fig. 4B (lane 9), Western blot analysis of transfected cell lysate from COS-7 cells detected three bands with the expected apparent molecular masses
of 18, 22.5 and/or 23, and 24 kDa (7, 22, 23). Surprisingly, transfection of the RBCF-1 cells yielded only the largest species (24 kDa) (Fig. 4B, lane 6). As a further control,
both the RBCF-1 and COS-7 cells were transfected with a 5-truncated
form of the human cDNA (pCMV-hmFGF-2) that lacks the initiating
leucine codons, but retains the initiating methionine codon. Both cell
lines produced the expected 18-kDa protein when transfected with this
plasmid (Fig. 4B, lanes 5 and 8).
Although rainbow trout FGF-2 is highly homologous to
mammalian FGF-2 in the C-terminal region, it does contain
nonconservative substitutions at amino acids 121-123 within the
putative receptor-binding domain, which is highly conserved among all
other FGF-2 sequences known (Fig. 1). To evaluate the significance of
this difference, we compared the activities of the wild-type human and
rainbow trout FGF-2 proteins and also mutant rainbow trout FGF-2
proteins in which the human amino acids were inserted into positions
121-123 (FGF-2/TSW) or in which these three amino acids were replaced with alanines (FGF-2/AAA). Each of the recombinant proteins was produced in E. coli and purified by heparin-Sepharose CL-6B
affinity chromatography. As shown in Fig. 5, the elution
profiles of each of the FGF-2 proteins were similar, suggesting that
the mutations do not affect the heparin binding affinity. The mutations
also do not appear to affect the stability of the proteins, as both the
wild-type and FGF-2/AAA mutant rainbow trout proteins showed complete
stability when incubated for up to 48 h at 37 °C in the medium
(data not shown). In addition, the rates of degradation of the two
proteins by trypsin were indistinguishable, and heparin protected both
from trypsin digestion (data not shown). The biological activity of the
mutant and wild-type proteins was evaluated by testing for their
ability to induce proliferation of cultured fibroblast cell lines from
goldfish (RBCF-1) and mice (BALB/3T3). The human FGF-2 protein was
~4- and 6-fold more potent than the wild-type rainbow trout FGF-2
protein when tested with BALB/3T3 cells and RBCF-1 cells, respectively
(Fig. 6). Interestingly, when the human amino acids were
substituted for those of rainbow trout in positions 121-123 (mutant
FGF-2/TSW), activity was increased ~2-fold when tested with either
cell type (Fig. 6). These results suggest that amino acids 121-123
contribute to the activity of FGF-2. The importance of these three
amino acids is further underscored by the nearly complete absence of
activity of the FGF-2/AAA mutant with either cell line (Fig. 6).
By using PCR to generate a homologous probe, we have isolated a rainbow trout FGF-2 cDNA clone that contains the entire coding sequence for the protein. An unexpected finding was that the cDNA contains an in-frame TGA stop codon just upstream of the initiating ATG codon, so rainbow trout cannot produce the alternative forms of FGF-2 found in mammals. The absence of alternative forms was confirmed by Western blotting of rainbow trout tissue extracts. A search of the EMBL data base revealed that an in-frame TGA stop codon is also present 15 bases upstream of the initiating ATG codon of Xenopus FGF-2, although its significance has not previously been remarked upon. Why rainbow trout lacks alternative translation initiation forms of FGF-2 awaits understanding of the roles of the different forms in mammals. Purified FGF-2 variants have the same mitogenic activity and seem to interact with the same cell-surface receptors. However, the alternative forms have been found to localize to different subcellular compartments, suggesting that they carry out distinct functions (22, 24, 37, 38): the human (22, 24) and rat (39) AUG-initiated forms are cytoplasmic, whereas the CUG-initiated forms are found primarily in the nucleus. Similar observations have been made for FGF-3 (40), suggesting that the alternative initiation mechanism might be a common feature among mammalian members of the FGF family.
Although the presence of an in-frame stop codon just 15 bp upstream of
the rainbow trout initiating methionine codon precludes alternative
initiation of FGF-2 from farther 5 CUG codons, we have shown that fish
cells are capable of initiating from such codons when present in the
proper context. Surprisingly, initiation occurred only from the most 5
CUG codon in the fish cells, whereas mammalian cells use, in addition,
two other CUG codons and the methionine codon. One possible explanation
of this result, based on the scanning model of Kozak (41), is that
initiation at the 5
CUG is actually more efficient in fish cells than
in mammalian cells, leaving too few scanning ribosomes to initiate at
the subsequent sites.
The encoded protein sequence of the rainbow trout cDNA is highly homologous to that of other species, especially in its C-terminal half. In this conserved region, human FGF-2 residues 115-124 (YRSRKYSSWY) have been shown to make a somewhat open loop on the surface of the molecule, which is presumed to bind to the FGF receptors (19, 20). It is therefore of particular interest that three nonconservative amino acid substitutions are present in rainbow trout FGF-2 residues 121-123 (PEM) as compared with human (TSW) and mouse, bovine, and Xenopus (SSW) FGF-2 proteins (Fig. 1).
To determine whether the substitutions of these residues could affect the FGF-2 biological activity or its specificity, two mutant FGF-2 proteins have been characterized in which FGF-2 residues 121-123 (PEM) have been substituted with TSW and AAA by site-directed mutagenesis. The FGF-2/AAA protein has similar heparin binding affinity compared with wild-type or FGF-2/TSW mutant protein (Fig. 5). Surprisingly, the mutant human (FGF-2/TSW) protein showed an increased activity relative to wild-type rainbow trout FGF-2 on the fish fibroblast cells as well as the mouse cells (Fig. 6). These results are consistent with previous studies that have suggested that residues 121-123 have an important role for both mitogenic function and heparin binding, but have presented data claiming to show little effect on heparin binding. We are beginning a more extensive mutagenesis program in an effort to further probe the functionality of this region of the protein.
In summary, we have isolated a cDNA that encodes the complete rainbow trout FGF-2. The cDNA reveals two striking differences from mammalian FGF-2: the absence of alternative N-terminal forms and the nonconservative substitution of amino acids in the region believed to be critical for receptor binding and mitogenic activity. The availability of recombinant rainbow trout FGF-2 should facilitate functional analysis of FGF-2 through comparative studies with the mammalian proteins and should also have practical applications to fish aquaculture.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) D85267[GenBank].
We are grateful to Dr. M. Brenner (NINDS, National Institutes of Health) for a critical reading of this manuscript, to Dr. R. Z. Florkiewicz for providing the human FGF-2 expression plasmid, and to the RIKEN cell bank for providing BALB/3T3 and COS-7 cells.