| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
INSERM U 397 Institut Louis Bugnard 31403 Toulouse Cedex 4, France
 |
ABSTRACT |
|---|
 TOP ABSTRACT RESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
These FGF-2s can exert their effects through different pathways. Extracellular FGF-2 binds to high-affinity transmembrane tyrosine kinase receptors (FGFR) and low-affinity receptors (heparan sulfate-containing proteoglycans) (11–13). Receptor activation stimulates intracellular mitogen-activated protein kinase (MAPK) and/or phospholipase C signaling pathways (14–16). But FGF-2 can also be internalized with both kinds of receptors into the cytoplasm (17–19) and translocate into the nucleus during the G1 phase of the cell cycle (20) by a mechanism distinct from that of nuclear endogenous FGF-2 (21). The nuclear HMM forms are involved in cell proliferation (22) and in oncogenesis (23, 24) while the AUG-initiated 18-kDa protein stimulates the migration (25), down-regulates its own receptor (22), and stimulates integrin synthesis (26). The inhibition of the expression of all FGF-2 isoforms leads to a loss of tumorigenicity in nude mice (27). A newly identified 34-kDa FGF-2 isoform has been recently characterized as a survival factor (8) while intravenous injection of 24 kDa-producing cells led to extensive lung metastases in nude mice (28). However, the details regarding the mechanisms by which the endogenous FGF-2s exert their intracellular effects remain to be elucidated.
We have previously shown that the three HMM (22, 22.5, and 24 kDa) and the 18 kDa (LMM) intracellular FGF-2s are found as components of large protein complexes of 320 and 130 kDa, respectively. Moreover, the coimmunoprecipitation of distinct proteins by anti-chloramphenicol acetyl transferase (CAT) antibodies in cells transfected with HMM-CAT and LMM-CAT fusion constructs, could reflect different activities of HMM and LMM FGF-2 (29). In an attempt to identify such associated proteins, we used the yeast two-hybrid system to screen a human lymphocyte cDNA library (30). We here report the characterization of a nuclear protein, FIF (FGF-2 interacting factor) which is able to bind HMM and LMM FGF-2 isoforms in the two-hybrid system and in in vitro binding assays but not some other members of the FGF family. Interestingly, coimmunoprecipitation experiments suggest that the FGF-2/FIF complex is nuclear in vivo. The FIF protein contains a leucine zipper, three hydrophobic heptad repeats, an acidic region, and a nuclear localization signal (NLS). Moreover, some truncated constructs revealed transactivating capacities.
|
RESULTS |
|---|
|
TOPABSTRACTRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Due to the lack of an initiation codon [Fig. 1
: the glutamic acid 72 is the first
amino acid of the isolated clone], a screening was performed on a
human hepatocyte cDNA library to isolate the full-length cDNA clone.
One clone, named FIF-504 (with a 504 aa open reading frame) was
isolated.3 This clone
contained 326 additional bp in the 5'-end compared with the FIF N1
clones. The downstream coding sequences were identical apart
from a divergence in the most 3'-region, comprising a 20-bp coding
sequence and a 189-bp noncoding sequence specific for FIF-504,
vs. 32 bp and 200 bp, respectively, which were specific for
FIF N1 (Fig. 1A). These two 3'-extremities probably originated
from an alternative splicing since the two first divergent
nucleotides GT in the clone 510 could correspond to the canonical
splice site that would be used to generate the FIF-504. Both
3'-sequences are well represented in the EST database, indicating that
both alternatively spliced forms of FIF are expressed in various
tissues. In FIF-504, the first methionine, surrounded by a good Kozak
consensus sequence for initiation of translation, is preceded 11 bp
upstream by an in-frame stop codon, thus avoiding any upstream
initiation. Based on the fact that there is no divergence among EST
sequences corresponding to the 5'-region of FIF, we presume that the
missing 5'-sequence in N1 is identical to that of FIF-504. The
5'-extended N1 clone was therefore denoted FIF-510 (Fig. 1A). Since no
difference in biochemical properties has yet been observed between both
FIF-504 and FIF-510, the two proteins will be considered as FIF. The
protein comprises two acidic domains (aa 164–190 and aa 323–335 in
Fig. 1B), a leucine-zipper motif (aa 370–391), and three overlapping
heptad repeats of hydrophobic amino acids (aa 259–314). One EST clone
(W40304) corresponds to an alternatively spliced form of FIF, in which
a 77-bp deletion introduces a stop codon 14 bp downstream from
the splice junction (Fig. 1A). We have recreated this COOH-truncated
FIF clone and named it FIF C1, which lacks the last 109 aa. During the
process of this investigation, two studies have reported the cloning of
a cDNA identical (API5-L1) (31) and highly similar
(AAC11long) (32) to that of FIF-504.
AAC11long and FIF-504 differ only by three amino
acids (residues 371–373, Fig. 1A); thus both proteins are likely to be
identical.
|
, FIF (FIF 504 and 510 gave the same
results) was only able to interact specifically with the two FGF-2
tested, i.e. the 18- and 24-kDa isoforms. FIF and FGF-2
interacted mutually in both bait and prey configuration. The
pAS-myc was used as a negative control. We did not detect
any homotypic interaction between FIF molecules indicating its
inability to form dimers at least in its fusion state in yeast.
Conversely, both FGF-2 isoforms dimerize as already described (33). The
strength of the FIF/FGF-2 interaction is notable since several FIF
constructs, such as FIF C1 (FIF-510 lacking the last 109 aa),
interacted with FGF-2 with similar efficiency as the positive control
proposed by CLONTECH Laboratories, Inc. (Palo Alto, CA)
(SV-40 large-T antigen binding to p53) as measured by ß-galactosidase
activity in yeast extracts (Figs. 2B and 7A).
|
|
). However, coexpression of FGF
and FIF fusion proteins resulted in a high level of luciferase
activity. Similar results were obtained in HeLa cells (data not shown).
These results clearly demonstrate that the FIF and FGF-2 proteins
associate efficiently in mammalian cells.
|
, A and B). Under these conditions,
anti-FGF-2 antibodies brought down a 55-kDa protein that was
detected by the anti-FIF antibodies (Fig. 4A, lane 3). To verify that
the detected protein was not heavy chain immunoglobulins but FIF, the
blot was stripped and reincubated with rabbit antihuman FGF-2
antibodies (Fig. 4B). Interestingly, in the reciprocal
coimmunoprecipitation experiment, anti-FIF antibodies brought down only
the nuclear HMM FGF-2 isoforms (22, 22.5, and 24 kDa) and not the
cytosolic 18-kDa FGF-2 form (Fig. 4C, lane 2). This could suggest
either a nuclear localization of the FIF/FGF-2 complex and/or a higher
affinity for the HMM FGF-2 forms. Nevertheless, these results confirm
an in vivo association between FIF and FGF-2.
|
, interaction test). FGF-2 alone did
not bind to the resin (Fig. 5A, control). Furthermore, no binding of
any protein other than FGF-2 was observed when incubation was performed
with total Escherichia coli FGF-2 extract in place of
purified FGF-2 (data not shown). These results demonstrate the direct
and specific in vitro interaction between FIF and FGF-2.
|
). FIF did not associate with streptavidin beads alone (Fig. 5B, lane 10). However, FIF was found to interact with biotinylated
FGF-2 in a DTT-independent manner. The oligomeric nature of FGF-2 in
the nonreduced state disappeared with the addition of DTT,
indicating the involvement of disulfide bonds in the homotypic
interaction of FGF-2 (Fig. 5B, lanes 4 and 5). FIF C1, used to
distinguish FIF from FGF-2 tetramers, was still able to associate FGF-2
in a dose-dependent manner and with the same efficiency as the
wild-type molecule (Fig. 5B, lanes 6–9). The interaction is also
observed in the presence of 250 mM NaCl (data not shown).
Hence, these results confirm the data obtained with the immobilized
His-FIF protein (Fig. 5A) and point out the fact that the FIF/FGF-2
interaction occurs directly. It also indicates that the last 109
COOH-terminal residues of the FIF protein, lacking in FIF C1, are not
required for FGF-2 binding as already shown (Fig. 2).
Human FIF mRNA Expression Is Ubiquitous
To determine the size and tissue distribution of FIF
mRNA, a multiple tissue Northern blot with polyA mRNA extracted
from various human tissues (CLONTECH Laboratories, Inc.) was probed with a full-length FIF cDNA probe (Fig. 6). One major RNA transcript with an
apparent size of 3.6 kb was detected in all the examined tissues and
reflects the ubiquitous expression of this messenger. The FIF mRNA
appeared relatively more abundant in heart, pancreas, and placenta,
whereas the signal obtained from liver mRNA was comparatively weaker.
Tewari and colleagues (32) reported, apart from the dominant transcript
of approximately 4 kb, the presence of some less abundant transcripts
in murine tissues. However, we were unable to detect these transcripts
in human tissues.
|
). The 55-kDa protein is likely
corresponding to the characterized AUG-initiated full-length FIF since
it migrates as the FIF protein overexpressed in transfected COS-7 cells
(Fig. 7A) as well as the in vitro-translated protein (Fig. 7B). The similarity of the patterns of immunodetected FIF reflects the
high degree of conservation of this protein in various species [98%
identity between mouse and human protein (32)]. Nevertheless, a 53-kDa
protein, but not the 55 kDa isoform, was detected in normal human skin
fibroblast cells. This smaller form was also detected in several other
cell types such as MCF7, SK-Hep-1, HeLa human cells, and rat PC12 cells
(Fig. 7A). A 40-kDa protein, immunologically detected by two different
sera raised against FIF, was observed in some cell lines of different
species (PC12 and SMC as well as in mouse NIH-3T3). The nature of the
40- and 53-kDa forms are presently unknown.
The subcellular localization of the FIF protein was studied both
by subcellular fractionation (endogenous FIF) and by immunofluorescence
labeling of FIF-transfected COS-7 cells. The Western blot presented in
Fig. 7C shows that the endogenous 55-kDa protein could be detected both
in the nuclear and cytoplasmic compartments (COS-7, NIH-3T3) while the
40-kDa form protein in NIH-3T3 appears to be exclusively nuclear.
However, in SK-Hep-1 cells the 55-kDa protein was found exclusively in
the nuclear fraction. The analysis of the FIF amino acid sequence
revealed the presence of a potential bipartite NLS between the lysine
residues 454 and 475 (Fig. 1B). This sequence is similar to the NLS of
the p53 tumor suppressor protein (Fig. 7D). The C1 protein was found in
the cytoplasm in transfected COS-7 cells in contrast to the full-length
FIF that was essentially nuclear (Fig. 7E). In some cells, however, the
protein was detected in the cytoplasm, confirming the results obtained
in subcellular fractionation of FIF-transfected COS-7 cells.
Overall, these results suggest that the full-length FIF protein is
essentially nuclear and that the sequence comprising residues 454–475
does play a role of NLS in the targeting of the wild-type protein to
the nucleus. The nuclear localization of the 40-kDa protein (Fig. 7, A
and C) is likely due to the presence of the NLS motif in the
COOH-terminal part, implying that this would correspond to a
NH2-truncated form of FIF.
The Interaction with FGF-2 Involves Two Distinct Regions in FIF
To define the FGF-binding region in the FIF sequence, progressive
deletion constructs were generated. As shown in Fig. 8A, two
NH2-terminal deletion constructs, lacking the
first 71 residues (N1) and 95 residues (N2), retained the capacity to
bind FGF-2. In contrast, all the N3 constructs, lacking the first 107
residues, were unable to interact with FGF-2. These results suggest the
presence of a first FGF-2-binding region (FBR-1) spanning proline 96 to
glutamine 107 in the FIF protein. Since the COOH deletions C3 (1–378),
C4 (1–345), and C5 (1–311) retained the ability to bind FGF-2, we can
conclude that the carboxy-terminal part of the protein beyond
phenylalanine 311 is not involved in the FGF-2 binding. Moreover, it
should be noted that in these deletions, as well as in several other
constructs, such as C1 (1–426), N1 (72), and N1C2 (72–402), the
binding capacity to FGF-2 appeared to be enhanced. This could be due to
differences in the structural conformation offering a better
accessibility to the FGF-2 binding sites through the lack of the
carboxy-terminal extremity.
|
)
has been shown to mediate binding between transcriptional coactivators
and nuclear receptors (35). This motif is present in FBR-1
(L102–L106). The replacement of the conserved leucine residues 102 and
105 by alanine (mutations generally done to disable this functional
binding motif) did not diminish the FGF binding, suggesting that this
short sequence is not involved in the FIF/FGF-2 interaction (data not
shown).
With respect to the deletions of the C terminus, a large difference in
FGF-2-binding capacity was observed between the C5 (1–311) and the C6
(1–273) constructs, indicating that the sequence including threonine
274 to phenylalanine 311 represents a second important region for the
FIF/FGF-2 interaction (Fig. 8B, FBR-2). Furthermore, the data suggest
that the two FBRs are necessary for FGF-2 binding since C6, lacking the
FBR-2, vs. N3C3, N4C3, N5C3, N3C5, and N6, only lacking the
FBR-1, did not exhibit any FGF-2 binding activity.
Taken together, these results indicate the presence of two mutually dependent regions in the FIF protein (spanning residues 96–107 and 274–311, respectively) which are both required for the interaction with FGF-2.
Truncation-Dependent Transactivation Capacity in the FIF
Protein
To determine whether the FIF protein could act as a
transactivator, each deleted-FIF cDNA was fused to a GAL4-DNA binding
sequence in the pAS2 plasmid. Yeast cells were transformed with each
construct, and the ß-galactosidase activity was measured from liquid
cultures.
As shown in Fig. 8, the results obtained with the N6C4 (231–345), C5
(1–311), and C6 (1–273) fusion proteins indicate that a potential
transactivation region is comprised between residues 231 and 273.
Notably, all constructs with a COOH terminus beyond glutamine 378
failed to transactivate the reporter gene. Moreover, in constructs with
only a COOH-terminal truncation, the region between phenylalanine 311
and leucine 346 seems to play an important role in the regulation of
the transactivating activity of FIF [compare C5 (1–311) with C4
(1–345)]. A region rich in acidic amino acids between the glutamate
residues 323 and 335 (pI = 2.8) (present in the C4 but not in the
C5 construct) could play a role in the inhibition of the
transactivation. However, this repressing activity appears to be
dependent on the nature of the N-terminal part of the protein (compare
C4 and N3C4 with N6C4). Conceivably, some intramolecular interactions
between the region from residues 108 to 231, and the acidic region
between residues 311 and 345, could contribute to this inhibition
(compare N4C3 or N3C4 with N6C4). This is further supported by the fact
that all constructs comprising both regions, except N3C3, did not show
any transactivating activity.
As mentioned previously, the N3C3 construct differed from the other similar truncated forms like N2C3 and N4C3 as it displayed transactivation capacities. This could imply a potentially inhibitory role of FBR-1 (aa 96–107) on transactivation. Likewise, one could assume that the sequence comprising residues 108–149, without any further NH2 extension, could adopt a specific conformation and/or allow the association of a specific binding factor leading to transactivation.
In conclusion, a region delimited from residues 231–273 can confer a transactivating potential of the FIF protein that is probably masked in the full-length protein.
|
DISCUSSION |
|---|
|
TOPABSTRACTRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
The interaction, initially observed in yeast, was reproduced in
mammalian cells (Fig. 3) and appears specific, since other FGF members
tested, such as FGF-1, FGF-3, and FGF-6, fail to interact with FIF
protein despite the high degree of sequence homology between these
different FGF members (Fig. 2). FGF-2 and FIF also coimmunoprecipitated
reciprocally (Fig. 4). The direct association between FIF and FGF-2 was
demonstrated by two reciprocal in vitro-binding assays (Fig. 5). Therefore, it appears that FIF constitutes a specific intracellular
partner of FGF-2.
We show that FIF is widely expressed in multiple human tissues (Fig. 6)
as well as in several other species such as monkey, rat, mouse, and
hamster (Fig. 7A). All the tested cells, namely fibroblasts, neural
(PC12), epithelial (RPE), smooth muscle (SMC), carcinoma (HeLa), and
hepatoma cells (SK Hep-1) express FIF. We can also note a parallel
ubiquitous expression of the FGF-2 in these cells (data not shown).
Two FIF cDNAs were cloned which differ in their 3'-extremities, generating FIF-504 and FIF-510 (504 and 510 residues, respectively) with an apparent molecular mass of 55 kDa. This corresponds well with the theoretical weight of 56,734 and 57,725 Daltons, respectively. These proteins, however, did not exhibit any detectable difference with respect to their affinity for FGF-2. Two other proteins, immunorelated to the full-length 55 kDa, with apparent molecular masses of 40 and 53 kDa were detected in some cell types. The protein AAC-11long, cloned by Tewari et al. (32), is identical to FIF apart from three residues that could come from an inversion between mouse and human sequences as confirmed in a recent publication concerning the localization of the FIF gene on the X chromosome (31). Furthermore, no human EST clones containing the sequence reported by Tewari et al. have been reported. AAC-11 was described as two isoforms with apparent molecular masses of 55 and 25 kDa. However, a 25-kDa form was never recognized by our antibodies.
FIF does not present any homology with two recently cloned FGF-binding proteins such as the FGF-1 intracellular binding protein (FIBP) (36), or the ribosomal L6/TAX-responsive element-binding 107 protein, which is able to associate with different FGF-2 isoforms (37).
FIF appears mainly located in the nucleoplasm (SK-Hep-1; Fig. 7, E and
C) even if fractionation assays in COS-7 and NIH-3T3 cells, and
in situ detection assays in COS-7, revealed its presence in
the cytoplasm. The nuclear localization is due to the presence of a
bipartite NLS in the C terminus, which is homologous to the one
described for the p53 protein (38). The functionality of this NLS was
demonstrated by in situ immunocytochemistry experiments
(Fig. 7E). The 40-kDa protein, which could either correspond to a
maturation product of the 55-kDa protein or a protein originating from
a downstream alternative initiation codon, is entirely nuclear and is
hence likely to contain the C-terminal portion of the full-length
FIF.
Interestingly, in the coimmunoprecipitation experiments, the anti-FIF
antibodies only brought down the nuclear HMM FGF-2 isoforms (Fig. 4C).
This indicate that the FIF/FGF-2 complex is nuclear in vivo
and/or that FIF exhibits a higher affinity for the HMM FGF-2 isoforms.
Nevertheless, we did not observe any significant differences in the
intensity of the interaction between FIF and the 18- or 24-kDa FGF-2
forms in two-hybrid experiments (data not shown). It is possible that
posttranslational events participate in the stabilization of the
FIF/FGF-2 complex in the nucleus. Several putative predicted
phosphorylation sites [protein kinase A (PKA), protein kinase C (PKC),
MAPK, casein kinase II (CK2), and ABL tyrosine kinase sites] are
located in or close to the NLS in FIF. Thus, phosphorylation might be
involved in the regulation of the compartmentalization of FIF within
the cell, as already shown in other nuclear proteins (39, 40, 41, 42).
Preliminary pull-down assays using antiphosphotyrosine antibodies on
total COS-7 cell extract immunoprecipitated the 55-kDa protein,
assuming a potential role of phosphates on tyrosine residues (data not
shown). Moreover, FGF-2 has been found to bind to the ß-subunit of
the CK2 (43). FIF contains six putative CK2 phosphorylation sites that
could be involved in the regulation of the localization and/or activity
of FIF.
To map the regions in the FIF protein involved in the interaction with
FGF-2, a set of truncated fusion constructs were cotransformed with the
FGF-2 into the yeast (Fig. 8). It is inferred from these data that aa
96–107 are required for FGF binding (FBR-1). A second centrally
located region in the full-length FIF (aa 274–311) was also shown to
interact with FGF (FBR-2). Both regions are required for the
interaction. Thus, through these two binding domains, FGF-2 could
induce a structural refolding of FIF, allowing either a restrained or
activated state. We can notice that the FBR-II comprises at least one
heptad repeat but not the canonical leucine zipper (Figs. 1 and 8). It
is possible that this hydrophobic structural feature may constitute a
spatial motif that is recognized by FGF-2. Nevertheless, the leucine
zipper motif in the L6/TAX-REB107 protein is not essential for the
binding to FGF-2 (37).
We can indeed observe that the transactivating activity of the FIF occurs essentially when a COOH-terminal part of the tested constructs is truncated. Intramolecular interactions between the COOH-terminal region beyond residue 311 and a region NH2-terminal to residue 231 could explain the absence of transactivating activity of the full-length protein. The region comprising residues 231–273 is present in all constructs with transactivating capacity, thus delineating a putative transactivating domain. It is possible that posttranslational modifications and/or alterations in the protein conformation could unmask and activate the transactivating domain. Such conformational changes could potentially be induced after phosphorylation of consensus sites. Following the example of the heat shock factors (HSF) model, inactive monomeric HSF1 is bound to hsp70 or hsp90 in nonstressed cells (44, 45). Following stress, however, the HSFs are released and subsequently acquire a trimeric active conformation with transactivating properties. Similarly, one could speculate that the FIF-FGF pair proceeds in a similar manner toward the FIF activation in response to specific signals. Preliminary results revealing homodimerization capacity for some truncated forms of FIF strengthen this eventuality (data not shown).
Numerous studies have shown that FGF-2 is able to protect cells from apoptosis (46, 47, 48, 49). However, FGF-2 has also been reported to promote apoptosis in some cell types, which suggests that the role of FGF-2 in programmed cell death could be cell type specific or dependent on the physiological status of the cell (50, 51, 52). The role of endogenously produced FGF-2 on apoptosis has been addressed by antisense approaches (46, 49) and by overexpression in vitro (47) and in vivo (52). It is not clear to which extent the HMM FGF-2 isoforms interfere in the regulation of apoptosis. It has been demonstrated, however, that cell surface receptor-mediated activation of MAPKs and a subsequent up-regulation of Bcl-2 and Bcl-xL levels are crucial for the antiapoptotic effect of FGF-2, which at least in RPE cells is mediated by a FGF2-induced production of FGF-1 (49, 53, 54).
Several studies suggest the necessity of a complementation between
FGFR-mediated signaling and the direct interaction between
intracellular proteins and internalized FGF-2 (20, 55) as well as
internalized FGF-1 (36, 56, 57) for normal proliferative FGF-mediated
activity. In addition the HMM FGF-2 forms have been attributed specific
functions, which often are correlated to malignant or
pathophysiological states (recently reviewed in Ref. 58). The
expression of HMM FGF-2 is enhanced in transformed and stressed cells
(24) and is negatively correlated to the cell density in normal cells
(59). Selective overexpression of HMM FGF-2 in NIH-3T3 fibroblasts
allowed growth and survival in low serum (8, 22, 60). Similarly,
expression of HMMW FGF-2 in pancreatic acinar AR4–2J cells rendered
these cells serum independent (61). The 24-kDa isoform was found to
render HeLa cells resistant toward 
-radiation (62), and elevated
levels of this isoform were also in correlation in enhanced mouse
intestinal epithelial cells survival after -irradiation (63).
Moreover, selective expression of 24-kDa FGF-2 in NBT-II cells was
shown to induce lung metastases after subcutaneous injection in nude
mice, in contrast to 18-kDa FGF2-producing cells (28).
Very little is known about the molecular mechanisms underlying HMM FGF-2-specific signaling. FIF is, to our knowledge, the first nuclear protein to interact directly with HMM FGF-2 isoforms and constitutes a potential mediator of these signals.
The antiapoptotic property of the shorter AAC-11 form of FIF (32) could be linked to the putative HSF-like behavior of FIF. The biophysical status of FIF (phosphorylation and/or maturation with a parallel change in the conformation) could be an important switch regulating FGF-2-binding and transcriptional activity. Moreover, expression of the HMM FGF-2 isoforms after serum stimulation of quiescent cells (59) and heat shock-stressed cells (24) are enhanced. These FGF-2 nuclear isoforms could then act concomitantly with FIF to stimulate or repress specific genes in response to stress.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Isolation of the Complete FIF cDNA
The FIF cDNA isolated in yeast was used as a probe to screen a
human hepatocyte-cDNA library inserted in the EcoRI site of
the 
ZAP II vector following the protocol provided by the supplier
(Stratagene, La Jolla, CA). Automatic nucleotide
sequencing was performed using the AmpliTaq FS polymerase and a ABI
373A sequencer (Perkin Elmer Corp., Foster City, CA).
Two-Hybrid System in Mammalian Cells
For interaction studies in mammalian cells, we constructed a
bigenic plasmid that allowed expression of the bait (FGF-2 fused to
VP16 transactivating domain) and the prey (GAL4 DNA-binding domain
fused to FIF 504 or 510) (see Plasmid Constructs
below). In the reporter plasmid, the firefly luciferase gene was
under the control of a minimal thymidine kinase promoter downstream
from the upstream activating sequences (UAS). Forty eight hours after
cotransfection in SK-Hep-1 cells, the luciferase activity was
determined by chemiluminescence assay as described by the manufacturer
(Promega Corp., Madison, WI).
Coimmunoprecipitations
Human stable transfected RPE-FGF cells (2 x
106) were rinsed in cold PBS, scraped,
pelleted, and incubated for 30 min at 4 C in 200 µl lysis buffer
(RIPA buffer: 50 mM Tris-HCl, pH 7.5, 150 mM
NaCl, 10 mM MgCl2, 1% Triton X-100,
0.1% SDS, 0.5% deoxycholic acid, 1 mM DTT) plus protease
and phosphatase inhibitors (10 mM
Na3VO4, 1 µg/ml
leupeptin, 1 µg/ml aprotinin, 0.1 mg/ml PMSF, 50 mM NaF,
50 mM ß-glycerophosphate, 100 µg/ml
N-tosyl-L-phenylalanine
chloromethyl ketone, 50 µg/ml
N-tosyl-L-lysine chloromethyl ketone,
100 µg/ml soybean trypsin inhibitor). The DTT was omitted in
experiments under nonreducing conditions. Cell debris was removed by
centrifugation at 10,000 x g for 10 min at 4 C.
Extracts were precleared with 50 µl of a mix of protein A and protein
G-Sepharose beads (Amersham Pharmacia Biotech, Uppsala,
Sweden) for 1 h at 4 C. Nonimmune serum (1:100) was added and the
mix remained 1 h at 4 C on a rotating wheel. After centrifugation
for 15 sec at 10,000 x g, the supernatant was
incubated for 2 h at 4 C with the primary antibody (rabbit
polyclonal antihuman FIF antibodies (1:200) or monoclonal antihuman
FGF-2 antibodies (Upstate Biotechnology, Inc., Lake
Placid, NY) (1:200) and 50 µl of protein A and protein G-Sepharose
beads. The beads were washed four times with RIPA buffer and twice with
PBS. The immobilized proteins were released by boiling in Laemmli
sample buffer and then analyzed by SDS-PAGE. Under conditions in which
the protein of interest comigrated with either the heavy or light chain
of immunoglobulins, proteins were heated at 75 C in Laemmli sample
buffer devoid of ß-mercaptoethanol and DTT. The rabbit polyclonal
antihuman FGF-2 antibodies used were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA).
Protein Production and Purification
The 18- and 24-kDa FGF-2 were produced and purified on a
heparin-Sepharose column by a FPLC (Amersham Pharmacia Biotech) as described previously (21). The His-FIF 504-, His-FIF
510-, and His-FIF C1-tagged proteins were produced using the inducible
pET expression system (pET-15b, Novagen, Madison, WI). in E.
coli BL21(DE3)pLysS and purified on a nickel-agarose column
(Ni-NTA, QIAGEN, Chatsworth, CA). For equilibration and
elution, 20 mM and 200 mM
imidazole, respectively, were used. An aliquot of each fraction was
analyzed in a SDS 10% polyacrylamide gel. The protein purity of the
fractions used was always higher than 95% as estimated by Coomassie
blue staining. Protein concentrations were determined by BCA assay
(Pierce Chemical Co., Rockford, IL).
In Vitro Interaction
The His-tagged recombinant FIF 504 and 510 purified on a nickel
agarose column (0.5 ml of FIF-NTA agarose = 100 µg of bound FIF)
were incubated with 20 µg of purified 18- or 24-kDa FGF-2 isoforms in
TKM50 (50 mM Tris-HCl, pH 7.4, 50 mM KCl, 5
mM MgCl2). After washing with 20
volumes of TKM100 (TKM50 but 100 mM KCl), the bound factors
were eluted with 0.5 ml of TKI (TKM100 + 1 M imidazole).
The eluted proteins were separated on SDS 12% acrylamide gel and
visualized by Coomassie blue staining. For FIF interaction with
biotinylated FGF-2, 200 ng of biotinylated FGF-2 were immobilized on 5
µl of streptavidin-coated-magnetic beads (Dynal, Oslo,
Norway) (33) and then incubated at 4 C in a Tween buffer (15
mM Tris-HCl, pH 7.5, 150 mM NaCl, 5
mM MgCl2, 0.1% Tween-20) with
50–500 ng of each purified full-length His-FIF protein or the
truncated form, His-FIF C1. After three washes with Tween buffer, the
bound proteins were eluted in SDS-sample buffer and analyzed by Western
blotting.
Northern Blot Analysis
A multiple tissue Northern blot, with polyadenylated mRNA
extracted from human cells of various tissues (MTN, CLONTECH Laboratories, Inc. Palo Alto, CA), was probed with the FIF 504
cDNA labeled with 32P-dATP using a
nick-translation kit (Promega Corp.).
Cell Types and Culture Conditions
Human retinal pigmentary epithelial cells (RPE) and human skin
fibroblasts were isolated and grown (
4 passages) as previously
described (24). Human RPE-FGF cells correspond to RPE cells stably
transfected with pREP-hFGF2 plasmid (see Plasmid Constructs)
and grow as RPE cells. Pheochromocytoma rat cells (PC12, kindly
provided by M. Weber) were grown in RPMI-1640 medium with 10% horse
serum and 5% FCS. The rat smooth muscle cells (SMC, kindly provided by
C. Vagner) were cultivated in DMEM with 10% calf serum. Cell lines
provided from ATCC, SK-Hep-1 (HTB 52), HeLa (CCL2), and
NIH-3T3 (HB-11601), were grown in DMEM with 10% FCS, whereas COS-7
cells (CRL 1654) and MCF-7 (HTB 22) were grown in DMEM with 5% FCS.
All cells were cultivated in 5% CO2 at 37 C.
Production and Immunopurification of FIF Antibodies
The sera of two rabbits were collected after three injections of
100 µg His-FIF 510 recombinant protein. To immunopurify antibodies, 5
mg of FIF 510 were linked to 500 µl of Affi gel-10 (Bio-Rad Laboratories, Inc.) in 0.1 M MOPS buffer, pH 7.5.
The unsaturated sites were blocked by ethanolamine before loading onto
a column. Ten milliliters of anti-His-FIF serum were added to the
column in a cyclic way and then washed with PBS/0.2% Triton, PBS/0.2%
Tween-20, and PBS. The immunoglobulins were eluted with 0.2
M glycine, pH 2.8/0.1% BSA and immediately neutralized
with 2 M Tris-HCl, pH 8. Protein concentrations were
determined by BCA assay (Pierce Chemical Co.).
DNA Transfection and Western Immunoblot Analysis
COS-7 monkey cells were transfected by the
diethylami-noethyl-dextran method as described previously (8). For
better separation of proteins in the size range of 40 to 60 kDa, we
used 10% SDS PAGE with Tris pH 8.3 in place of pH 8.8 (Fig. 7A).
SK-Hep-1 and HeLa cells were transfected by the use of Lipofectin
(Life Technologies, Inc.) as described by the
manufacturer. Cell extracts were treated for immunoblotting as
previously described (29). FGF-2 and FIF proteins were immunodetected
using rabbit polyclonal anti-human-FGF-2 antibodies (Santa Cruz Biotechnology, Inc., 1:1000 dilution) and rabbit polyclonal
antihuman-FIF antibodies (1:1000, see above), respectively. The blotted
proteins were detected by ECL (Amersham Pharmacia Biotech)
using either donkey peroxidase-linked antirabbit IgG or sheep
peroxidase-linked antimouse IgG (Amersham Pharmacia Biotech).
In Situ Immunocytochemistry
COS-7 cells grown on coverslips were transfected with pSG5 504
and pSG5 C1 (2 µg/ml) and prepared for immunofluorescence microscopy
as previously described (8). They were viewed under a fluorescence
microscope (Leica Corp., Deerfield, IL) and photographed
using 400 ASA HP5 film (Ilford).
In Vitro Translation
pSG5–504 and pSG5–510 plasmids, encoding FIF 504 and 510,
respectively, were digested by BglII and translated in the
presence of 35S methionine (Amersham Pharmacia Biotech) as previously described (8). Translation
products were analyzed by SDS-PAGE (10%) followed by autoradiography
of dried gels.
Plasmid Constructs
The details of each construct have been grouped in Table 1. The PCR primers used to generate some
constructs are shown in Table 2. Full
construction details are available upon request. pACT, pACT2, and pAS2
plasmids derive from MATCHMAKER (CLONTECH Laboratories, Inc.) two-hybrid vectors. The cloning of FGF-2 into pAS2 and
pACT is described elsewhere (33). The pET-15b vector (Novagen) was used
for protein production in E. coli. pET-FGF 155 and pET FGF
210 encoding the 18-kDa and 24-kDa isoforms of FGF-2, respectively,
were constructed as described previously (33). For production of the
different His-tagged FIF proteins, pET 15b constructs are described in
Table 1. The details regarding the construction of the pREP-hFGF2
plasmid, which expresses the CUG1–3- and AUG-initiated FGF-2 isoforms,
is described elsewhere (66). pSG5 FIF 504 and pSG5 FIF 510 were
constructed to analyze FIF protein in mammalian cells and for in
vitro translation experiments. pTHM3-FIF 510 and pTHM3-FIF 504
were designed to test the two-hybrid system in mammalian cells. The
corresponding reporter gene plasmid was derived from the SV40-pFlash
(SynapSys), which was modified through three successive insertions of a
double-stranded oligonucleotide containing a UAS recognition sequence
(sense oligo-sense: 5'-GCGGAGTACTGTCCTCCGGAGCTCG-3' with
SacI in bold; antisense oligo-antisense:
5'-AGCTGCGGAGTACTGTCCTCCGGAGC-TCGGATC-3'
(with a "sticky" degenerated SacI site, a wild-type
SacI, and a "sticky" BamHI site in
bold, respectively). After hybridization of the
complementary oligonucleotides, the double-stranded UAS fragment was
ligated into the SacI/BamHI sites of the
SV40-pFlash vector. The resulting construct was digested with
SacI/BamHI, and a second UAS fragment was
inserted. The procedure was repeated until three repeats had been
inserted. cDNA clones encoding mouse FGF-3 and human FGF-6 were kindly
provided by C. Dickson and F. Coulier, respectively.
|
|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
This research was supported by CNRS and INSERM grants, and l’Association de la Recherche contre le Cancer and La Fondation de la Recherche Medicale.
1 These authors contributed equally to this work. 
2 A schematic representation of the different
truncated FIF constructs is shown in Fig. 8A.
3 The sequence data have been submitted to
the GenBank database under accession numbers AF229253 and
AF229254.
Received for publication February 7, 2000. Revision received July 31, 2000. Accepted for publication August 8, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  P. Krejci, K. Pejchalova, B. E. Rosenbloom, F. P. Rosenfelt, E. L. Tran, H. Laurell, and W. R. Wilcox The antiapoptotic protein Api5 and its partner, high molecular weight FGF2, are up-regulated in B cell chronic lymphoid leukemia J. Leukoc. Biol., December 1, 2007; 82(6): 1363 - 1364. [Full Text] [PDF]
|
|
|
|
 |
|
 P. Dvorak, D. Dvorakova, S. Koskova, M. Vodinska, M. Najvirtova, D. Krekac, and A. Hampl Expression and Potential Role of Fibroblast Growth Factor 2 and Its Receptors in Human Embryonic Stem Cells Stem Cells, September 1, 2005; 23(8): 1200 - 1211. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Bossard, L. V. d. Berghe, H. Laurell, C. Castano, M. Cerutti, A.-C. Prats, and H. Prats Antiangiogenic Properties of Fibstatin, an Extracellular FGF-2-Binding Polypeptide Cancer Res., October 15, 2004; 64(20): 7507 - 7512. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Garmy-Susini, E. Delmas, P. Gourdy, M. Zhou, C. Bossard, B. Bugler, F. Bayard, A. Krust, A.C. Prats, T. Doetschman, et al. Role of Fibroblast Growth Factor-2 Isoforms in the Effect of Estradiol on Endothelial Cell Migration and Proliferation Circ. Res., May 28, 2004; 94(10): 1301 - 1309. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Malecki, J. Wesche, C. S. Skjerpen, A. Wiedlocha, and S. Olsnes Translocation of FGF-1 and FGF-2 across Vesicular Membranes Occurs during G1-Phase by a Common Mechanism Mol. Biol. Cell, February 1, 2004; 15(2): 801 - 814. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Hu, S. Wang, Y. Zhang, C. A. Feghali, J. R. Dingman, and T. M. Wright A nuclear target for interleukin-1{alpha}: Interaction with the growth suppressor necdin modulates proliferation and collagen expression PNAS, August 19, 2003; 100(17): 10008 - 10013. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Claus, F. Doring, S. Gringel, F. Muller-Ostermeyer, J. Fuhlrott, T. Kraft, and C. Grothe Differential Intranuclear Localization of Fibroblast Growth Factor-2 Isoforms and Specific Interaction with the Survival of Motoneuron Protein J. Biol. Chem., January 3, 2003; 278(1): 479 - 485. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. S. Skjerpen, J. Wesche, and S. Olsnes Identification of Ribosome-binding Protein p34 as an Intracellular Protein That Binds Acidic Fibroblast Growth Factor J. Biol. Chem., June 21, 2002; 277(26): 23864 - 23871. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. T. A. Meij, F. Sheikh, S. K. Jimenez, P. W. Nickerson, E. Kardami, and P. A. Cattini Exacerbation of myocardial injury in transgenic mice overexpressing FGF-2 is T cell dependent Am J Physiol Heart Circ Physiol, February 1, 2002; 282(2): H547 - H555. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Goldfarb Signaling By Fibroblast Growth Factors: The Inside Story Sci. Signal., October 30, 2001; 2001(106): pe37 - pe37. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
, The start of the FIF N1 clone, isolated in the initial two-hybrid screening. ***
Indicates stop codons. The sequence in italics
corresponds to the FIF C1 construct and the alternatively spliced EST
clone W40304. The bracket indicates the putative splice
site which generates the two divergent 3'-extremities in the cDNA of
the FIF-504 and 510 sequences. B, Amino acid sequences of the human
FIF-510 and FIF-504. The circled glutamic acid residue
indicates the start of the N1 clone. The amino acids involved in the
canonical and the three overlapping heptad repeat motifs are
squared and linked. The LXXLL motif is
underlined and the nuclear localization signals (NLS)
are boxed. The small brackets delineate
the two acidic regions.
