Fibroblast growth factor-2 binds to the regulatory beta subunit of CK2 and directly stimulates CK2 activity toward nucleolin.

The presence of fibroblast growth factor-2 (FGF-2) in the nucleus has now been reported both in vitro and in vivo, but its nuclear functions are unknown. Here, we show that FGF-2 added to nuclear extract binds to protein kinase CK2 and nucleolin, a CK2 natural substrate. Added to baculovirus-infected cell extracts overexpressing CK2 or its isolated subunits, FGF-2 binds to the enzyme through its regulatory β subunit. Using purified proteins, FGF-2 is shown to directly interact with CK2 and to stimulate CK2 activity toward nucleolin. Furthermore, a mitogenic-deficient FGF-2 mutant protein has an impaired ability to interact with CK2 and to stimulate CK2 activity using nucleolin as substrate. We propose that in growing cells, one function of nuclear FGF-2 is to modulate CK2 activity through binding to its regulatory β subunit.

The fibroblast growth factors (FGFs) 1 family includes nine polypeptides of which FGF-1 and FGF-2 (acidic and basic FGF) are prototype members (reviewed in Refs. 1 and 2). FGF-2 exerts its pleiotropic effects in cell growth and differentiation through a dual receptor system consisting of four different high-affinity transmembrane receptor tyrosine kinases and low-affinity binding sites corresponding to heparan sulfate proteoglycans (reviewed in Refs. 2 and 3). In addition, FGF-1 and FGF-2 are translocated from outside the cell to the nucleus (reviewed in Ref. 2). Many other growth factors have been detected in the cell nucleus (reviewed in Refs. 4 and 5). In the case of Schwannomaderived growth factor and FGF-1, nuclear localization is necessary for mitogenic activity (6 -8). For FGF-2, nuclear translocation is correlated to cell proliferation, since it is no longer recovered in the nucleus of confluent cells (9,10). Therefore, these data strongly suggest that FGF-2, as well as other growth factors, plays specific, but still unknown, nuclear functions in addition to classical signaling through cell surface receptors.
We reported previously that in synchronized ABAE growing cells, a large increase of ribosomal genes transcription is tightly correlated to the nuclear translocation of FGF-2 and especially to an accumulation of the growth factor in the nucleolus. In contrast, in quiescent cells ribosomal genes transcription is 20-fold lower, and FGF-2 is exclusively found in the cytoplasm (9,10). Upon addition of FGF-2 to nuclei of serum-starved cells, a 6-fold increase of ribosomal genes transcription is observed together with an increase in phosphorylation of nuclear proteins, essentially of nucleolin (9,11); These data suggest that a nuclear function of FGF-2 could be to regulate ribosomal genes transcription by modulating the phosphorylation level of nuclear and nucleolar proteins as nucleolin. Nucleolin that is the major component of nucleolus is also a major substrate for protein kinase CK2 in rapidly proliferating tissues, and its phosphorylation level is correlated to the rate of ribosome biogenesis (reviewed in Ref. 12). CK2 is a serine/threonine protein kinase present in both the cytoplasm and the nucleus of cells and made of an heterotetramer composed of two catalytic ␣ and/or ␣Ј subunits and two regulatory ␤ subunits (reviewed in Refs. 13 and 14). Genetic studies showed that the kinase activity of CK2 is essential for viability of yeast (15). The involvement of CK2 in mitogenic signaling is strongly supported by approaches using microinjection of CK2 (16) or antibodies raised against CK2 (17,18). Moreover, several reports correlate an increase in CK2 activity with cell growth stimulation (reviewed in Ref. 13). Many CK2 substrates are nuclear proteins involved in the regulation of gene expression or cell cycle (reviewed in Refs. 13 and 14). Among them, only few were shown to physically associate with CK2. In the case of p53 and topoisomerase II, the interaction is mediated by the regulatory ␤ subunit of the potein kinase (19,20). Interestingly, the polyamine spermine that also interacts with CK2 ␤ subunit is a putative physiological regulator of the protein kinase (14,21,22). However, whether and how CK2 is involved in signal transduction remain unresolved. We further reported that while nucleolin is constantly present in the nucleolus, CK2 is detected in the nucleolus of growing but not confluent ABAE cells. Simultaneously, nucleolin is phosphorylated in growing cells and dephosphorylated in confluent cells (23). So, a putative target of FGF-2 in the nucleus could be the protein kinase CK2. Here, we report that in nuclear extract, FGF-2 associates with CK2 through its regulatory ␤ subunit. Furthermore, using purified proteins, we show that FGF-2 directly binds to CK2 and specifically modulates its activity using nucleolin as a substrate. Finally, a mutant nonmitogenic form of FGF-2 has an impaired ability to interact with CK2 and to stimulate CK2 activity toward nucleolin. for ABAE cells in (10). Sf9 cells were infected with EV55Dm␣ and or EV55Dm␤ viruses for the overexpression of CK2 subunits. At 3 days post-infection cells were washed in phosphate-buffered saline and lysed. Sf9 cells were broken in lysis buffer (20 mM Tris-HCl, pH 7.6, 12% saccharose, 2 mM EGTA, 2 mM phenylmethylsulfonyl fluoride, 5 mM dithiothreitol, 50 g/ml leupeptin, 50 g/ml aprotinin). Triton X-100 was added to 0.2%, and the homogenate was incubated for 3 min at 4°C, centrifuged 3 min at 1000 ϫ g yielding pellet that corresponds to nuclear fraction. Nuclear extracts were prepared from nuclear fractions of FM3A cells or insect cells overexpressing Drosophila melanogaster (DM) CK2 ␣ or ␤ or ␣ 2 ␤ 2 as described in Ref. 45.
Expression, Purification, and Biotinylation of Wild-type or Mutant FGF-2-⌬1 mutant form of FGF-2 was generated as described in Ref. 30. Expression and purification of recombinant wild-type or ⌬1 FGF-2 were performed as described in Ref. 10. Recombinant human FGF-2 was biotinylated on cysteine residues, modification that did not affect the biological activity of the growth factor (46). Iodoacetyl-LC-biotin was purchased from Pierce.
Loading of Biotinylated FGF-2 on Streptavidin Beads-Beads (Dynabeads M-280 streptavidin beads, Dynal) were first extensively washed with binding buffer (15 mM Tris, pH 7.5, 5 mM MgCl 2 , 0.1% Tween 20, 150 mM NaCl), washed twice with binding buffer containing 0.1% bovine serum albumin and once more with binding buffer. Beads were mixed with biotinylated FGF-2 (biotFGF-2) at a ratio of 1 g of biot-FGF-2 for 10-l beads and incubated 30 min in 300 l of binding buffer at room temperature under constant rotation. Beads were then extensively washed for a complete removal of free FGF-2.
Biotinylated FGF-2 Interaction with Nuclear Extract Prepared from FM3A Cells-Loaded beads (50 or 10 l) were incubated 1 h at 4°C in 400 l of binding buffer with proteins (40 or 4 g) from FM3A cells nuclear extract. After extensive washing, bound proteins were either directly resuspended in sample buffer by boiling beads 4 min in 1 ϫ Laemmli sample buffer or eluted at 4°C with 100 l of elution buffer (15 mM Tris, pH 7.5, 5 mM MgCl 2 , 0.1% Tween 20, 400 mM NaCl). Eluates were either trichloroacetic acid-precipitated and resuspended in 1 ϫ Laemmli sample buffer or tested for CK2 activity (see below). For each experiment, the same procedure was performed with unloaded beads as control.
Biotinylated FGF-2 Interactions with Recombinant CK2-Reactions were performed as described above with 10 l of loaded or unloaded beads and 20 ng of purified recombinant Drosophila CK2 or 1-2 g of proteins from nuclear extract prepared from insect cells overexpressing Dm CK2 ␣, ␤, or ␣ 2 ␤ 2 . Quantities of the different nuclear extracts included were adjusted to yield equal amounts of each subunit per assay. Bound proteins were resuspended by boiling beads 4 min in 1 ϫ Laemmli sample buffer.
Test of the Eluates for CK2 Activity-Aliquots of the eluates (20 l) prepared as described above were incubated 2 min at 30°C with 100 M [␥-32 P]ATP (1500 cpm/pmol), 1 mg/ml of dephosphorylated casein (Sigma) and 0, 5, or 20 g/ml heparin in a final volume of 66 l at final concentrations of 15 mM Tris, pH 7.5, 5 mM MgCl 2 , 0.1% Tween 20, 150 mM NaCl. Reactions were stopped by adding 16 l of 5 ϫ Laemmli sample buffer. One-third of each reaction was analyzed by SDS-PAGE.
Purification of CK2 and Nucleolin-Expression and purification of recombinant Drosoplila CK2 and isolated ␣ subunits were performed as described previously (19,26). CK2 and nucleolin was purified from CHO cells (Mons University, Hainaut, Belgium). Nuclei were isolated and lysed in a low ionic strength buffer and the resulting S2 extract was applied to an heparin-Sepharose column (Pharmacia Biotech Inc.) (31). Fractions containing nucleolin and CK2 were then subjected to anionexchange chromatography (Mono Q, Pharmacia). CK2 and nucleolin were eluted with a 0.2-1 M KCl gradient produced with 50 mM Tris-HCl, pH 7.9, 30% glycerol. Fractions were analyzed on a 12% SDS-PAGE and tested for CK2 activity.
Phosphorylation Assays-To achieve maximal labeling, nucleolin purified from CHO cells was dephosphorylated prior to performing phosphorylation assays. Alkaline phosphatase-agarose (Sigma) was first extensively washed with 50 mM Tris, pH 8, 1 mM MgCl 2 , 0.1 mM ZnCl 2 and washed twice with same buffer including 0.1% bovine serum albumin for saturation. 10 g of nucleolin purified from CHO cells was incubated 1 h at 37°C with 10 units of saturated alkaline phosphataseagarose in a final volume of 100 l with final concentrations of 50 mM Tris, pH 8, 2 mM MgCl 2 , 0.1 mM ZnCl 2 , 0.5 mg/ml bovine serum albumin, 10 g/ml leupeptin, and 0.1 mM phenylmethylsulfonyl fluoride. After a quick centrifugation, phosphatase-free dephosphorylated nucleolin was recovered in the supernatant. Phosphorylation reactions (25 l) contained 25 ng of dephosphorylated nucleolin, 100 M [␥-32 P]ATP (1000 cpm/pmol), 10 mM Tris, pH 7.5, 150 mM NaCl, 1 mM MgCl 2 when recombinant CK2 was used or 4 mM MgCl 2 when CHO purified CK2 was used, 0.1% Tween 20, 10 g/ml leupeptin, 0.1 mM phenylmethylsulfonyl fluoride and increasing amounts of the indicated growth factor or compound. All incubations were performed for 2 min at 30°C. Phosphorylation reactions were initiated by adding 5 ng of recombinant Drosophila CK2 ␣ or ␣ 2 ␤ 2 or 0.1 unit of CK2 purified from CHO cells (1 unit of activity is the amount of enzyme that catalyzes the incorporation of 1 pmol of phosphate into dephosphorylated casein at 5 mg/ml in 1 min at 30°C). All reactions were stopped by adding 6 l of 5 ϫ Laemmli sample buffer. [␥-32 P]ATP (3000 Ci/mmol) was supplied by Amersham Corp., cytochrome c and EGF by Sigma, and insulin by Novo-Nordisk.
SDS-PAGE, 32 P Analysis, Silver Staining, and Immunoblotting-For silver staining, boiled samples were resolved by 10 or 12% SDS-PAGE (47) and then the gel was incubated for 15 min in acetic acid 10%, methanol 30%. After washing with H 2 OmQ for 15 min, the gel was incubated for 20 min in glutaraldehyde 10%. After washing for 1 h, it is incubated in a silver solution extemporaneously prepared as follows: 4 ml of AgNO 3 20% is added drop by drop in 74 ml of H 2 OmQ, 21 ml of NaOH 0.36%, 1 ml of NH 4 OH. After washing with H 2 OmQ for 2 min, it is incubated in a solution containing 100 ml of H 2 OmQ, 50 l of citric acid, 10%, and 50 l of formaldehyde, 37.5%, until silver staining is judged fine. Then the gel is transferred in H 2 OmQ where it can be kept. For 32 P analysis and immunoblotting, boiled samples were resolved by 10 or 12% SDS-PAGE and electrotransferred to nitrocellulose membranes (48). For 32 P analysis, Western blots were autoradiographed to visualize phosphorylated proteins and scanned using a FujiX phosphorimager to quantitate the extent of phosphorylations. For immunoblotting, nitrocellulose membranes were blocked for 1 h with TBS-T (15 mM Tris, pH 8, 150 mM NaCl, 2 mM MgCl 2 , 0.1% Tween 20), 3% milk powder. Then filters were incubated 1-2 h with specific antibodies as indicated. After washing, the blots were incubated with an anti-rabbit horseradish-peroxydase conjugated secondary antibody (Promega). Protein-antibody complexes were visualized by an enhanced chemiluminescense (ECL) Western blotting detection system according to the manufacturer's specifications (Amersham). To detect the ␣ and the ␤ subunit of CK2, we used an anti-CK2 ␣ subunit polyclonal antibody (UBI) or a polyclonal antibody directed against a synthetic ␤ CK2 peptide. This peptide, which corresponds to the carboxyl-terminal amino acid sequence (205-215) of the ␤ subunit of Drosophila CK2, was synthesized by Neosystem Laboratories (Strasbourg, France) and conjugated to ovalbumin for immunization to yield the corresponding antibodies pAB␤CD (Eurogentec, Seraing, Belgium). Serum was used at the 1/500 dilution.
ABAE Cells Culture-Primary cultures of ABAE cells were established as described previously (49). Cells were seeded in 35-mm dishes at 2000 cells/dish and routinely grown at 37°C with 10% CO 2 in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% heat-inactivated calf serum. Purified wild-type FGF-2 (1 ng/ml of culture medium) or ⌬1 FGF-2 (1 ng/ml, 10 ng/ml or 50 ng/ml of culture medium) was added every 2 days. Each day, cells were trypsinized and counted.

FGF-2 Interacts Specifically with CK2 in a Nuclear
Extract-To address our hypothesis that FGF-2 acts as a regulator of protein kinase CK2, we first investigated whether FGF-2 can interact with CK2. Biotinylated FGF-2 immobilized on streptavidin-beads (biotFGF-2) was incubated with proteins of nuclear extract prepared from FM3A cells. In order to visualize pattern and amount of proteins from nuclear extract that interact with FGF-2, proteins bound to biotFGF-2 were separated on SDS-PAGE and silver stained. As shown in Fig. 1, no protein were recovered with unloaded beads indicating that proteins bound to loaded beads really interact with biotFGF-2. Then, we observed that a limited number of proteins from the whole nuclear extract bound biotFGF-2 and that the amount of bound proteins reached saturation, whereas the amount of input proteins still increased. These points attest that in our test the interaction observed with biotFGF-2 is specific.
Then, in order to test for CK2 activity, bound proteins were eluted and incubated with [␥-32 P]ATP in a kinase assay buffer with dephosphorylated casein as exogenous substrate. As shown in Fig. 2A, two bands corresponding to phosphorylated casein (32 and 26 kDa) and an additional endogenous phosphoprotein (100 kDa) were observed. The phosphorylation of casein was markedly inhibited by heparin, a potent and specific inhibitor of CK2: 75% inhibition with 5 g/ml heparin to 90% inhibition with 20 g/ml heparin. Similarly, the phosphorylation of the 100-kDa protein was inhibited by 70 and 85% with 5 and 20 g/ml heparin, respectively. In the absence of exogenously added casein, the 100-kDa protein was the only phosphorylated protein (data not shown). We subsequently probed the filter with a specific serum raised against nucleolin. As shown in Fig. 2B, the phosphorylated 100-kDa band was recognized by anti-nucleolin antibodies, suggesting that the phosphoprotein was nucleolin. Taken together these data show that FGF-2 retain a protein kinase activity inhibited by heparin that phosphorylates casein and a 100-kDa protein that could be nucleolin.
Further characterization of the protein kinase activity bound to biotFGF-2 was carried out by immunoblotting experiments. Equal amounts of nuclear extract proteins were incubated with increasing amounts of biotFGF-2, and bound proteins were analyzed by immunoblotting with a polyclonal antibody raised against the ␣ subunit of CK2. As shown in Fig. 3A, CK2 ␣ was detected among the proteins bound to biotFGF-2. The signal of bound CK2 ␣ increased with the amount of biotFGF-2 until CK2 ␣ was exhausted in input proteins. Then, increasing amounts of nuclear extract proteins were probed with equal amounts of biotFGF-2. As shown in Fig. 3B, bound CK2 ␣ increased and reached saturation, albeit input CK2 ␣ still increased. As a control for binding specificity, similar immunoblotting experiments were performed using specific antibodies raised against two other protein kinases Cdk1 (or Cdc2) and Cdk2. These two protein kinases were selected, because Cdk1 has been shown to phosphorylate casein and nucleolin (24), and Cdk2 has been shown to control the transition between the G 1 and S phases of the cell cycle (reviewed in Ref. 25), the transition when the nuclear translocation of FGF-2 occurs (10). As shown in Fig. 3C, neither Cdk1 nor Cdk2 were detected among proteins bound to biotFGF-2. The same observation was made for the mitogen-activated protein kinase or MAP kinase (data not shown).
These data show that in a nuclear extract, biotFGF-2 associates with nucleolin and CK2 whether the enzyme was de-tected as a kinase activity or as an immunodetected protein. As previous experiments using purified proteins indicated to us that FGF-2 does not directly bind nucleolin (data not shown), and as nucleolin is a natural CK2 substrate, we especially investigated the interaction between FGF-2 and CK2.
Interaction between FGF-2 and CK2 Is Mediated by the CK2 ␤ Subunit-In order to determine which CK2 subunit may interact with FGF-2, we expressed isolated ␣ or ␤ subunits or CK2 holoenzyme in a baculovirus-directed insect cell expression system as described previously (26). BiotFGF-2 was directly added to nuclear extracts prepared from these insect cells. Bound proteins were analyzed by immunoblotting experiments with antibodies raised against CK2 ␣ or ␤ subunit. As shown in Fig. 4A, biotFGF-2 interacts with the holoenzyme ␣ 2 ␤ 2 as well as with CK2 ␤ subunit alone but not with CK2 ␣ subunit alone. Then, in order to know whether FGF-2 directly binds to CK2, biotFGF-2 was incubated with recombinantpurified CK2. As shown in Fig. 4B, biotFGF-2 binds to purified CK2. Therefore, these experiments show that the interaction between FGF-2 and CK2 is direct and is mediated by CK2 ␤ subunit. So, to further demonstrate that FGF-2 binds CK2 in a nuclear extract of FM3A cells, we analyzed nuclear extract proteins bound to biotFGF-2 by immunoblotting with a polyclonal antibody raised against CK2 ␤ subunit. Indeed, CK2 ␤ subunit was detected among proteins bound to biotFGF-2 (data not shown).
FGF-2 Directly Stimulates Phosphorylation of Nucleolin by CK2-To investigate whether the direct interaction between FGF-2 and CK2 could modulate the kinase activity, we per-   1. FGF-2 retain just few proteins from a whole nuclear extract. Nuclear extract (NE) prepared from FM3A cells was incubated with biotinylated FGF-2 (biot-FGF-2) loaded on streptavidin beads (ϩ) or with unloaded beads (Ϫ). To visualize bound proteins compared with the whole nuclear extract, input and bound proteins were resolved by SDS-PAGE and analyzed by silver staining. Increasing amounts of input proteins (1, 2, 5, or 10 g) were incubated with biotFGF-2 (ϩ). As a control, 10 g of input proteins were incubated with unloaded beads (Ϫ). For comparison, 1 g of input proteins (NE input) and biotFGF-2 is shown on first and last lane, respectively. Migration of molecular mass standards (kDa) is indicated to the left.
formed phosphorylation assays using purified proteins. CHOpurified nucleolin was used as natural CK2 substrate. CK2 was purified from CHO cells, or CK2 holoenzyme and CK2 ␣ were purified from insect cells infected with recombinant baculovirus. As shown in Fig. 5A, phosphorylation of nucleolin by CK2 was clearly stimulated in the presence of increasing amounts of FGF-2 to reach a maximum at 10 Ϫ7 M FGF-2 and then de-creased back to the basal level at 10 Ϫ6 M FGF-2. In contrast, phosphorylation of nucleolin by the catalytic ␣ subunit alone did not increase upon addition of FGF-2. Quantitation of 32 P incorporation revealed that phosphorylation of nucleolin by recombinant or CHO CK2 was stimulated 3-or 5.5-fold, respectively, by 10 Ϫ7 M FGF-2 ( Fig. 5B and 6, respectively). As a control, we examined the effects of cytochrome c, a molecule of similar molecular weight and isoelectric point to FGF-2, and of heat-denatured FGF-2. None of them had any effect on CK2 activity in this assay, indicating that the basicity of the molecule is not sufficient to explain the effect of FGF-2 (Fig. 5B). Then, in comparison with FGF-2 effect, we examined the effect of two agents known for inducing CK2 activity: spermine, a commonly used CK2 activator but of very low molecular weight, and protamine, a mix of basic polypeptides with an average molecular mass of 4 -4.5 kDa. An efficient stimulation of CK2 activity using nucleolin as substrate was observed for spermine concentrations in the millimolar range and for protamine concentrations in the micromolar range (Fig. 5B). This underlined that FGF-2 stimulated CK2 activity at a much lower concentration than spermine and protamine, two specific activators of CK2. Phosphorylation assays performed with dephosphorylated casein as substrate indicated an FGF-2-induced stimulation of CK2 activity that reached a maximum at 10 Ϫ6 M FGF-2 (data not shown). Therefore, taken together these data strongly support the hypothesis that FGF-2 directly stimulates CK2 activity through its interaction with the ␤ subunit of the protein kinase.
CK2 Activity Is Not Directly Modulated by Other Growth Factors-To extend this analysis, the effects of several other growth factors (FGF-1, insulin, EGF, and NGF) were tested on CK2 activity. We selected FGF-1, since its nuclear localization has been related to its biological role (6,8) and EGF, NGF, and insulin since they have been proposed to be involved in CK2 regulation. For instance, experiments using antisense oligonu- FIG. 3. FGF-2 interacts with the protein kinase CK2 in a nuclear extract. Nuclear extract (NE) was incubated with biotFGF-2 (ϩ) or without (Ϫ) as described in the legend to Fig. 1. Input and bound proteins were resolved by SDS-PAGE, transferred to nitrocellulose, and analyzed by immunoblotting. A, 4 g of input proteins (NE input) were incubated with increasing amounts of biotFGF-2. Immunoblotting was performed with an anti-CK2 ␣ subunit polyclonal antibody. B, conversely, increasing amounts of input proteins (1, 2, 5, or 10 g) were incubated with biotFGF-2 (ϩ). As a control, 10 g of input proteins were incubated with unloaded beads (Ϫ). Immunoblotting was performed as described in A). C, 4 g of input proteins (NE) were incubated with biotFGF-2 (ϩ) and bound proteins (B) were immunoblotted with an anti-cdk1 or an anti-cdk2 antiserum. Western blots were overexposed as control.
cleotides involved CK2 in the mitogenic signaling elicited by EGF (27). Similar experiments strongly suggested an essential role for CK2 in controlling neurite growth (28). In addition, insulin has been reported to increase CK2 activity (reviewed in Ref. 29). Experiments were performed as described above with CK2 purified from CHO cells. As shown in Fig. 6, under these conditions, phosphorylation of nucleolin was stimulated 5.5fold by 10 Ϫ7 M FGF-2. FGF-1 was shown to stimulate CK2 activity by 2-fold at such a high concentration (10 Ϫ5 M) that the significance of this effect is unclear. None of the other growth factors tested so far appeared to directly modulate CK2 activity. Thus, in our conditions, FGF-2 seems to specifically activate the phosphorylation of nucleolin by CK2.
The ⌬1 Mutation of FGF-2 Affects Its Mitogenicity and Its Interaction with CK2-Taking into account both previously reported and present data, we hypothesized that the interaction between FGF-2 and CK2 might be involved in mitogenic pathway of the growth factor. Therefore, we studied a mutant form of FGF-2, noted ⌬1 FGF-2, that corresponds to a protein deleted of 6 residues (Lys 27 -Leu 32 ) located in the NH 2 -terminal part of the molecule (30). This mutant form was shown to interact with the high affinity receptor and with heparin as the wild-type form and thus the deletion did not affect significantly the physicochemical properties of the molecule. Nuclear extracts prepared from FM3A cells or from baculovirus-infected cells overexpressing CK2 were incubated with wild-type (WT) or ⌬1 biotFGF-2. Bound proteins were analyzed by immunoblotting experiments with an anti-CK2 ␣ subunit polyclonal antibody or/and an anti-Dm CK2 ␤ antiserum. As shown in Fig.  7A, binding of CK2 to ⌬1 FGF-2 was clearly reduced compared with WT FGF-2, especially the binding of CK2 from normal FM3A cells (left panel). Immunoblotting with antibodies raised against FGF-2 was performed to verify that equal amounts of WT and ⌬1 biotFGF-2 had been used (data not shown). Then, we tested ⌬1 FGF-2 for its ability to modulate phosphorylation of nucleolin by CK2. As shown in Fig. 7B, while a 3-fold stimulation of nucleolin phosphorylation was observed at 10 Ϫ7 M WT FGF-2, no stimulation was observed with the same concentration of ⌬1 FGF-2. A 2-fold stimulation was only detected with 10 Ϫ6 M ⌬1 FGF-2. Last, we tested ⌬1 FGF-2 for its mitogenic activity at 1 ng/ml, the lowest concentration of WT FGF-2 that promotes full proliferation of ABAE cells. As shown in Fig.  7C, in these conditions, ⌬1 FGF-2 failed to promote long term proliferation of ABAE cells compared with wild-type FGF-2 (1 ng/ml). Interestingly, ⌬1 FGF-2 at 10 ng/ml induced cell growth but not as efficiently as WT FGF-2 (about three times less). Therefore, ⌬1 FGF-2 failed to increase the phosphorylation of nucleolin by CK2 and the proliferation of ABAE cells in the range of concentrations determined to be optimal for WT FGF-2. Both stimulation of CK2 activity and mitogenic activity could be detected only if ⌬1 FGF-2 was used at a 10-fold greater concentration. These data demonstrate that the ⌬1 mutation of FGF-2 that affects its interaction with CK2 and its ability to directly stimulate CK2 phosphorylation of nucleolin concomitantly affects its mitogenic activity. DISCUSSION Although it is generally accepted that actively proliferating cells contain higher level of CK2 activity than quiescent cells, conflicting data were reported concerning the regulation of CK2 activity in response to growth factors (29). Whether CK2 is involved in signal transduction remains unresolved and controversial (reviewed in Ref. 14). Here, we report that FGF-2 interacts with nucleolin and CK2 in a nuclear extract. FGF-2 directly binds to CK2 through the regulatory ␤ subunit of the protein kinase. Furthermore, in vitro, FGF-2 directly modulates CK2 activity toward nucleolin. Finally, a mutant nonmitogenic form of FGF-2 has an impaired ability to interact with CK2 and to stimulate CK2 activity toward nucleolin. These findings support the hypothesis that CK2 could be a target of FGF-2 involved in nuclear functions of the growth factor. We reported previously that in nuclei isolated from ABAE confluent cells, CK2 activity is inadequate to phosphorylate endogenous nucleolin, a major substrate for CK2 (31). The addition of exogenous CK2 to such nuclei induced not only protein phosphorylation but also ribosomal genes transcription (23). Similarly, addition of FGF-2 to nuclei isolated from ABAE serum-starved cells induced an increase of ribosomal genes transcription together with phosphorylation of nucleolin essentially (9,11). Moreover, the phosphorylation level of nucleolin has been correlated to the rate of ribosomal RNA transcription (reviewed in Ref. 12). These data strongly suggested that FGF-2 could control ribosomal genes transcription through modulation of CK2 activity toward specific substrates and especially nucleolin.
We report here in vitro experiments employing purified FGF-2, CK2, and nucleolin, in which FGF-2 induces once more an increase of nucleolin phosphorylation (Fig. 5). These latter data indicated that FGF-2 and CK2 could interact directly. Protein-protein interactions between FGF-2 and CK2 were first detected in experiments in which biotinylated FGF-2 was added to nuclear extracts (Fig. 3). Then, using purified components the possibility that a third component could mediate the binding of FGF-2 to CK2 was ruled out (Fig. 4B). Furthermore, protein-protein interaction experiments carried out with nuclear extracts prepared from baculovirus-infected cells overexpressing isolated CK2 subunits showed that FGF-2 interacts with CK2 ␤ subunit (Fig. 4A). Yet, we cannot exclude that the ␣ subunit participates to the interaction between FGF-2 and CK2. However, the interaction between FGF-2 and the ␤ subunit is necessary for the effect of FGF-2 on CK2 activity, since FGF-2 had no effect on CK2 ␣ activity (Fig. 5). Interestingly, p53 and topoisomerase II, two CK2 substrates, and the polyamine spermine, also bind to CK2 through its ␤ subunit. The presence of the ␤ subunit is required for the formation of a stable complex with each of the two substrates and for the activation of CK2 activity by the polyamine (19 -22). Unlike p53 and topoisomerase II, FGF-2 is not a substrate for CK2. Thus, FGF-2 is more closely related to spermine and also to protamine and so appears as another putative physiological regulator of CK2. These results further strengthen the idea that the ␤ subunit plays a key role in CK2 regulation and in its biological functions (reviewed in Ref. 14). The ␤ subunit is required for a fully active CK2 (32) and could play a role in substrate specificity (33,34). Recently, different polymeric forms of CK2 were correlated to different specific activities (35). Considering all these observations, it is conceivable that the binding of FGF-2 to CK2 ␤ subunit induces a conformational change, resulting in the modulation of CK2 activity, its substrate specificity, or both.
FGF-2 has now been reported in the cell nucleus in vitro in many systems and in vivo during chicken embryogenesis or Xenopus development (36,37). In cells producing FGF-2, four different isoforms have been described in different subcellular compartments (reviewed in Refs. 1 and 2). The subcellular distribution of these isoforms can be differentially regulated (38), and their dysregulated expression resulted in different phenotypes, suggesting that they act through distinct pathways (39). Therefore a pathway in which the growth factor is delivered to the nucleus may exist in addition to the classical tyrosine kinase transduction pathway. We reported previously that in growing cells FGF-2 is translocated to the nuclei during the G 1 /S transition of the cell cycle and accumulated to the (OⅪO), 1 ng/ml (OfO), or 10 ng/ml (---f---) ⌬1 FGF-2, or without growth factor (OEO). Duplicate dishes were trypsinized, and cells were counted at days of culture indicated.
FIG. 7. The ⌬1 mutation of FGF-2 affects its mitogenicity, its interaction with CK2, and its ability to stimulate CK2 activity. A, interaction of ⌬1 FGF-2 with CK2. Biotinylated WT or ⌬1 FGF-2 loaded on streptavidin beads (ϩWT, ϩ⌬1) or as control unloaded beads (Ϫ) were incubated at 4°C with nuclear extract (4 g of protein) prepared from FM3A cells (left panel) or insect cells overexpressing D. melanogaster CK2 (right panel). Input nuclear extract (NE or ␣ 2 ␤ 2 ) and bound proteins (B) were resolved by SDS-PAGE, transferred to nitrocellulose, and probed with an anti-CK2 ␣ subunit polyclonal antibody or with an anti-D. melanogaster CK2 ␤ antiserum. B, effect of ⌬1 FGF-2 on phosphorylation of nucleolin by CK2. Purified and dephosphorylated nucleolin was incubated with 100 M [␥-32 P]ATP and recombinant purified CK2 in presence of increasing amounts of WT (OⅪO) or ⌬1 (OfO) FGF-2. Phosphorylation of nucleolin was analyzed as described in the legend to Fig. 4. A.U., arbitrary units. C, mitogenic activity of the ⌬1 FGF-2. Recombinant ⌬1 FGF-2 was tested for its capacity to stimulate long term proliferation of ABAE cells. Cells were plated in Dulbecco's modified Eagle's medium supplemented with 1 ng/ml WT FGF-2 nucleoli (9, 10). In confluent cells, FGF-2 is no longer present in both the nuclei and the nucleoli and simultaneously, CK2 is no longer present in the nucleoli. In contrast, nucleolin is constantly present in the nucleoli but highly phosphorylated in growing cells and dephosphorylated in confluent cells (23). ln line with these data, several authors described that CK2, or especially its ␤ subunit, was located mainly in the nucleus of growing cells but spread throughout cells that are serumstarved or confluent (16,17,40). Therefore, a coordinated spatiotemporal distribution of FGF-2 and CK2 in the nucleus could allow the regulation of specific nuclear protein phosphorylations, such as nucleolin phosphorylation during the G 1 /S transition. Support has been emerging for the concept that the differential subcellular distribution of kinases and phosphatases regulates the fine coordination of protein phosphorylations (41,42).
Here, we show that in vitro, the phosphorylation of nucleolin by CK2 is stimulated by increasing amounts of FGF-2 (Fig. 5). Surprisingly, this stimulation did not reach a plateau. This may be because in vitro, great amounts of FGF-2 finally induce aggregates or polymeric structures of CK2 that exhibit low specific activities. Indeed, in vitro FGF-2 is known to form multimers and CK2 exhibits different conformations correlated to different specific activities (35). But, such a modulation of CK2 activity has also been described previously, since basic polypeptides have a biphasic effect on phosphorylation of calmodulin (43). The maximal stimulation of CK2 activity is observed at 10 Ϫ7 M FGF-2. As translocated FGF-2 was estimated to be 240 pg/10 6 nuclei (10) and the average volume of the nucleus 0.2 ϫ 10 Ϫ12 liters (44), nuclear FGF-2 concentration should reach 6.5 ϫ 10 Ϫ8 M in ABAE cells. Thus, the concentration of 10 Ϫ7 M can be reached in specific locations of the nucleus, especially in the nucleolus where FGF-2 was shown to accumulate (9,10,38). This direct stimulation of CK2 activity was apparently not observed with FGF-1, EGF, NGF, and insulin (Fig. 6). Yet, further phosphorylation assays using different endogenous substrates have to be performed, because a given growth factor may modulate CK2 activity toward one or several but specific substrates.
The data presented in this report support the hypothesis that CK2 is a putative target of FGF-2 in the nucleus. The characterization of the interaction sites of the two partners and the isolation of FGF-2-CK2 complex in vivo are now under investigation. It is expected that our data will bring new insights both to the regulation of CK2 activity by growth factors and to the nuclear functions of FGF-2.