Identification of a Novel Domain of Fibroblast Growth Factor 2 Controlling Its Angiogenic Properties*

Fibroblast growth factor 2 (FGF-2) is a potent factor modulating the activity of many cell types. Its dimerization and binding to high affinity receptors are considered to be necessary steps to induce FGF receptor phosphorylation and signaling activation. A structural analysis was carried out and a region encompassing residues 48–58 of human FGF-2 was identified, as potentially involved in FGF-2 dimerization. A peptide (FREG-48–58) derived from this region strongly and specifically inhibited FGF-2 induced proliferation and migration of primary bovine aorta endothelial cells (BAEC) in vitro , and markedly reduced FGF-2-dependent angiogenesis in two distinct in vivo assays. To further investigate the role of region 48–58, a polyclonal antibody raised against FREG-(48–58) was tested and was found to block FGF-2 action in vitro . Human FGF-2 has three histidine residues, one falling within the region 48–58. Chemical modification of histidine residues blocked FGF-2 assays in modified Boyden chambers (31). BAEC on gelatin-coated polycarbonate filters (12- (cid:1) m pores) (Costar) were to FGF-2 (Invitrogen), vascular endothelial growth factor (VEGF) (RnD Systems), epidermal growth factor (EGF) (Invitrogen), or fibronectin (Invitrogen) human recombinant factors, dissolved at the reported concentration in Dulbecco’s modified Eagle’s medium, 0.1% BSA. FREG-(48–58) or SCR were supplemented to the growth factor solution at the reported final concentration. Assays were carried out at 37 °C in 5% CO 2 , for 6 h. Migrated cells were counted at (cid:2) 400 magnification in 15 fields for each filter, and the average num- ber (cid:3) S.D. of cells/field was reported. All experiments were performed least three times in AbFREG48–58 interaction with immobilized FGF-2. Increasing quantities of FGF-2 were spotted onto nitrocellulose membrane (0, 6, 12.5, 25, 50, and 100 ng/spot, respectively); membrane was then incubated with AbFREG48–58 (1:200 in PBS), and bound antibody was revealed with anti-rabbit peroxidase-conjugated IgG and chemiluminescence. AbFREG48–58 recognized the immobilized FGF-2. This experiment was performed three times with similar results. B , FREG-(48–58) competition. Increasing quantities of FREG-(48–58) (up to 100 ng/ml) dose-dependently competed for the interaction between FGF-2 and AbFREG48–58, carried out as reported in panel A . C , FGF-2 induced BAEC proliferation at 48 h, in the presence of AbFREG48–58 (1:2000; 1:4000) or in the presence of preimmune rabbit serum (same dilutions). AbFREG48–58 (1:4000) significantly inhibited FGF-2-induced proliferation, while the control antibody was ineffective. The two antibodies did not modify cell number under basal conditions.

Fibroblast growth factor 2 (FGF-2) is a potent factor modulating the activity of many cell types. Its dimerization and binding to high affinity receptors are considered to be necessary steps to induce FGF receptor phosphorylation and signaling activation. A structural analysis was carried out and a region encompassing residues 48 -58 of human FGF-2 was identified, as potentially involved in FGF-2 dimerization. A peptide (FREG-

-58) derived from this region strongly and specifically inhibited FGF-2 induced proliferation and migration of primary bovine aorta endothelial cells (BAEC) in vitro, and markedly reduced FGF-2-dependent angiogenesis in two distinct in vivo assays. To further investigate the role of region -58, a polyclonal antibody raised against FREG-(-58) was tested and was found to block FGF-2 action in vitro.
Human FGF-2 has three histidine residues, one falling within the region 48 -58. Chemical modification of histidine residues blocked FGF-2 activity and FREG-(48 -58) inhibitory effect in vitro, indicating that histidine residues, in particular the one within FREG-(48 -58) region, play a crucial role in the observed activity. Additional experiments showed that FREG-(48 -58) specifically interacted with FGF-2, impaired FGF-2-interaction with itself, with heparin and with FGF receptor 1, and inhibited FGF-2-induced receptor phosphorylation and FGF-2 internalization. These data indicate for the first time that region 48 -58 of FGF-2 is a functional domain controlling FGF-2 activity.
Fibroblast growth factor 2 (FGF-2) 1 belongs to the fibroblast growth factor family and is one of the most potent growth factors regulating cellular functions including proliferation and chemotaxis, tissue and organ development, and tissue repair (1)(2)(3). FGF-2 has been suggested to play a key role in the etiology and progression of several pathological conditions including Parkinson's and Alzheimer's diseases (4 -5), tumor growth and progression (6 -9), atherosclerosis, and restenosis after angioplasty (10). A specific role has been indicated for FGF-2 in tumor angiogenesis (11)(12) and in neuronal differentiation (13)(14).
In light of the key role FGF-2 plays in both physiologic and pathologic conditions, structural-functional relationships studies on this factor are of large interest. FGF-2 dimerization and its interaction with heparan sulfates are necessary steps required for binding to high affinity receptors and for the following signaling activation. Several reports investigated functionally relevant regions of FGF-2 by analyzing crystallographic structures or by mutagenesis studies. FGF-2 dimer has not been crystallized, and different studies based on molecular simulations indicated different regions possibly lying at the FGF-2 dimer interface (15)(16)(17), whereas crystallization (18 -20) and other studies (21)(22)(23)(24)(25)(26) identified residues crucial for FGF-2 activity and involved in the interaction with heparin and high affinity receptors. Finally, specific residues were found to control the FGF-2 ability to induce production of urokinase-type plasminogen activator, likely in the FGF-2 nuclear localization site (27)(28).
Despite all ongoing investigations, a specific site regulating the whole FGF-2 activity by controlling FGF-2 interaction with itself, has not yet been identified. We recently showed that FGF-2 directly interacts with PDGF-BB (29 -30), leading to a marked inhibition of its angiogenic properties in vitro and in vivo (31). In the present study we report the novel identification of an FGF-2 functional domain controlling in vitro and in vivo FGF-2 properties, likely by controlling dimer formation.
Regions at the interface of the predicted FGF-2 dimer were identified by solvent accessibility analysis as described above. According to these analyses, the FGF-2 region including the sequence DPHIKLQLQAE (here referred to as FREG-(48 -58)) was identified. FREG-(48 -58) was then custom synthesized by Research Genetics Inc. As a sequencespecificity control, two different scrambled versions of FREG-(48 -58) (KHIAQLDEPLQ and KLQLDIEAHPQ), were synthesized by Neosystem Laboratoire. They are here referred to as SCR. Peptide purity, assessed by reverse-phase HPLC and mass spectrometry, was higher than 95% in all cases. Different batches of FREG- (48 -58) and SCR gave similar results in all biological assays.

Cell Isolation and Cell Culture
BAEC were derived from adult bovine aorta and cultured as previously reported (40). Cell purity was evaluated by DiI-Ac-LDL uptake and was consistently Ͼ97%. BAEC at passage 3-7 and at 80% confluence were used in all assays.

Proliferation and Migration Assays
Proliferation assays were carried out as previously reported (31). After 24 h of starvation, BAEC were incubated with Dulbecco's modified Eagle's medium-BSA (0.1%) alone or containing human recombinant growth factors (10 ng/ml) as reported, in the absence or in the presence of FREG- (48 -58) or SCR, the antibody raised against FREG-(48 -58) (AbFREG-(48 -58)), or preimmune control serum (AbP-I). BSA fraction V (Sigma) was used throughout the study. Time course experiments were carried out at 1-, 2-, and 3-day time points at 10 ng/ml FREG- (48 -58) concentration. Cells were then harvested and counted with a hemacytometer. The FREG-(48 -58) dose used in most experiments (i.e. 10 ng/ml) was identified in dose dependence tests carried out in cell proliferation and receptor phosphorylation experiments.
Migration assays were carried out in modified Boyden chambers (Costar Scientific Corporation) as previously reported (31). Briefly, BAEC seeded on gelatin-coated polycarbonate filters (12-m pores) (Costar) were exposed to FGF-2 (Invitrogen), vascular endothelial growth factor (VEGF) (RnD Systems), epidermal growth factor (EGF) (Invitrogen), or fibronectin (Invitrogen) human recombinant factors, dissolved at the reported concentration in Dulbecco's modified Eagle's medium, 0.1% BSA. FREG- (48 -58) or SCR were supplemented to the growth factor solution at the reported final concentration. Assays were carried out at 37°C in 5% CO 2 , for 6 h. Migrated cells were counted at ϫ400 magnification in 15 fields for each filter, and the average number Ϯ S.D. of cells/field was reported. All experiments were performed at least three times in duplicate.

Histidine Residue Chemical Modification
Chemical modification of histidine residues was achieved by diethylpyrocarbonate (DEPC) treatment, which specifically modifies the imidazolic ring of histidine residues (41). Briefly, an aliquot of a freshly prepared DEPC solution (Sigma) in anhydrous ethanol was added to FGF-2 in PBS, to reach 200 M DEPC and 55 M FGF-2 final concentration. After 15 min at 25°C the reaction was stopped with 100 mM Tris-HCl (pH 7.6). DEPC-induced modification of FREG-(48 -58) was carried out similarly. In control experiments DEPC was first inactivated by prior incubation with 100 mM Tris-HCl, and then added to the sample. Histidine-modified FREG-(48 -58) and FGF-2 were then tested in SPR (see below) and migration assays.

Fluorescence Analysis
Fluorescence spectroscopy experiments were carried out as previously reported with modifications (29, 42) on a Perkin Elmer LS55 fluorimeter at constant temperature (22°C). Briefly, 10 aliquots (0.5 l each) of FREG- (48 -58) or SCR (85 M, 0.1 mg/ml in PBS) were subsequently added to 90 l of FGF-2 in PBS (2 M, 36 g/ml). Under these conditions, the FGF-2 sample reached only 5% dilution. Fluorescence emission spectra were collected by excitation at 277 nm. Peak fluorescence was at 310 nm as previously reported for similar experimental conditions (43). Data were corrected for the buffer, FREG-(48 -58), and SCR contributions and for the dilution factor, and its value was plotted as a function of the amount of peptide supplemented. Data were analyzed according to the single site tight binding model (44).
Injection-FGF-2 (20 l, 1.25 g/ml) was injected for a 2-min association phase with or without increasing concentrations of FREG- (48 -58) or SCR, followed by HBS flow for a 30-s dissociation phase. Response, monitored at 25°C, was expressed in RU. Sensor chip regeneration was achieved by NaOH injections (50 mM, 10 l each). A 10 l/min flow rate was used throughout the experiments.

Overlay Experiments
Either FGF-2 or BSA (300 ng) were spotted onto a nitrocellulose membrane as previously reported (29), then blocked with 5% dry milk (Bio-Rad) in TPBS (0.1% Tween 20 in PBS). In addition to BSA, an additional control was introduced, corresponding to the mix of molecular weight standards (Amersham Biosciences) listed in the legend of Fig. 4. After the TPBS wash, the membrane was incubated with biotinylated FREG-(48 -58) (2 ml, 200 g/ml) for 4 h at room temperature and washed again. Interaction was detected by an avidin/biotinylated horseradish peroxidase kit (Vectstain ABC, from Vector) followed by chemiluminescent reaction and exposure to Kodak film (Eastman Kodak).

FGF-R1 Phosphorylation
After 24 h of starvation, BAEC were incubated with FGF-2 (10 ng/ml) diluted in 0.1% BSA, in the absence or in the presence of FREG- (48 -58) or SCR (10 ng/ml) for 10 min. Cells were then washed in ice-cold PBS and scraped with lysis buffer (as above). Cell extracts were sonicated and centrifuged, and supernatants were collected. Immunoprecipitation was then carried out on samples (60 g of protein) by incubating with anti-FGF receptor antibody (mAb 125, Chemicon) (2 g) at 4°C overnight on a rocking bath, followed by incubation with 10 g of protein G-Sepharose 4 Fast Flow (Amersham Biosciences) for 2 h at 4°C. After washing with lysis buffer, complexes were dissolved in Laemmli buffer, separated by 8% SDS-PAGE, blotted onto nitrocellulose membrane, blocked as reported above, incubated with a monoclonal anti-phosphotyrosine antibody (clone PT66, Sigma) (1:2000 dilution) for 1 h at room temperature, revealed by chemiluminescence (Amersham Biosciences), and quantified by densitometry.
Production and Testing of the Rabbit Anti-FREG- (48 -58) Antibody (AbFREG48 -58) Antibody against FREG-(48 -58) (AbFREG48 -58) was developed according to standard protocols. Briefly, FREG-(48 -58) (1 mg) dissolved in 2 ml of PBS and supplemented with Complete Freund's Adjuvant, was administered by intramuscular injection in two 5-month-old New Zealand rabbits. After 22 and 44 days the antigen injection was repeated with un-complete Freund's Adjuvant. Fifteen days after the third injection, blood was collected, and the serum was separated and stored at Ϫ20°C in aliquots. Before some experiments, rabbit serum was collected after four antigen injections and showed activity at a higher dilution. The ability of this antibody to recognize the whole FGF-2 was tested by spotting increasing quantities of FGF-2 (up to 100 ng/spot) onto nitrocellulose membranes and then blocking with 5% dry milk. After TPBS washing, membranes were incubated with AbFREG48 -58 (1:200 in PBS) for 1 h at room temperature. Membranes were then washed and incubated with anti-rabbit peroxidase-conjugated IgG (Pierce) for 1 h at room temperature; detection was achieved by chemiluminescence (Amersham Biosciences). FREG-(48 -58) up to 100 ng/ml was used to compete for the observed interaction between FGF-2 and AbFREG48 -58. AbFREG48 -58 was then used in internalization and proliferation assays as reported.

In Vivo Angiogenesis Assays
Procedures involving animals and their care were conducted according to the institutional guidelines, in compliance with international laws and policies (Guide for the Care and Use of Laboratory Animals; United States National Research Council, 1996).

Angiogenesis in Chick Embryo Chorioallantoic Membrane (CAM Assay)
CAM assays were performed as previously reported (31). Fertilized White Leghorn chicken eggs (10/group) were incubated at 37°C at constant humidity. On incubation day 3, 2 ml of albumin were removed to allow detachment of the developing CAM. The window was sealed with a glass and the eggs returned to the incubator. On day 8, CAM was treated by applying the sponge as reported in the legend to Fig. 8. On day 12, blood vessels entering the sponge within the focal plane of the CAM and were counted at ϫ50 magnification in a double-blind fashion under a Zeiss SR stereomicroscope (Zeiss) and photographed in ovo with the MC63 Camera System (Zeiss). Embryos and their membranes were fixed in ovo in Bouin's fluid. Sponges, the underlying and immediately adjacent portions of CAM were removed, embedded in paraffin, sectioned at 8 m along a plane parallel to the CAM surface, and stained with a 0.5% aqueous solution of toluidine blue (Merck). The angiogenic response was assessed histologically by a planimetric method of "point counting" (48). The vascular density was indicated as the mean number of the occupied intersection points Ϯ S.D.

Angiogenesis in Matrigel Plugs
The angiogenesis assay in Matrigel plugs was carried out as reported (31,49). Briefly, Matrigel, a mixture of reconstituted basement membrane proteins (BD PharMingen, 600 l) supplemented with FGF-2 (150 ng/ml) alone or in the presence of FREG- (48 -58) or SCR (10 g/ml), was injected subcutaneously in CD1 mice (female, 20 g body weight). Matrigel plugs were removed after 8 days and processed for histology analysis. Histologic sections (7 m) were stained with Trichrome-Masson procedure (Bio-Optica). Vessels within the Matrigel were recognized by morphology and by the presence of red blood cells. Angiogenesis was evaluated blindly by two operators, by considering at least five different sections per Matrigel plug; each section was 100 m from the next. The total number of neo-vessels in the whole Matrigel area was measured with an Axioplan microscope (Zeiss) and was expressed as the number of vessels/mm 2 . When 2 vessels were cut longitudinally, as indicated by a long axis greater than 3-fold of the short axis, and were close to each other, they were counted as a single vessel. Ten animals were treated with FGF-2 alone, 14 with FGF-2 ϩ FREG-(48 -58), 10 with FGF-2 ϩ SCR.

Statistics
Data were expressed as mean Ϯ S.D. Student's two-tailed t test was performed, and p Յ 0.01 was considered statistically significant.

RESULTS
Analysis of FGF-2 Structure-Primary sequence analysis and three-dimensional structure evaluation were carried out on human FGF-2. A multiple sequence alignment was performed on the 23 members of the human fibroblast growth factor family (Fig. 1A). According to the consensus sequence at the 60% identity level, several residues are conserved within the family, while other regions show a high variability. In particular, the FGF-2 region 43-60 was found to be poorly conserved within the FGF family.
A docking simulation was carried out on two FGF-2 monomers (PDB code: 2BFH) leading to the prediction of a FGF-2 dimer (Fig. 1B); therefore regions at the dimer interface were predicted by solvent accessibility evaluation. Finally, on the high-resolution crystallographic model of the FGF-2/FGR1heparin complex (PDB file: 1FQ9), regions interacting with the FGF-R1 were identified by solvent accessibility analysis (reported in the legend of Fig. 1A). According to all these analyses, the FGF-2 region 43-60 shows one of the lowest intrafamily identities and, at the same time, many residues of this segment were predicted to be involved in FGF-2 dimerization and were found to be involved in receptor binding. All together these observations suggested that this portion of FGF-2 may play a specific role in FGF-2 dimerization and receptor binding. To investigate this hypothesis, the follow- FIG. 4. FREG-(48 -58) interacting properties. A, overlay assay of biotinylated FREG-(48 -58) on immobilized FGF-2. Biotinylated FREG-(48 -58) interacted with FGF-2 immobilized onto nitrocellulose (300 ng/spot), whereas no interaction was observed with immobilized BSA (300 ng/spot) or with immobilized molecular weight standards (Amersham Biosciences, 300 ng/spot) (Mix) consisting of a mixture of myosin, phosphorylase B, BSA, ovalbumin, carbonic anhydrase, trypsin inhibitor, and lysozyme. A representative experiment is shown. This experiment was performed three times with similar results. B, fluorescence spectra of FGF-2 (2 M) in the presence of increasing doses of FREG- (48 -58). Spectra were obtained by exciting at 277 nm and collecting spectra between 300 and 400 nm. Data were corrected for the dilution (which never exceeded 5%), for the buffer, FREG-(48 -58), and SCR contribution. Peak fluorescence at 310 nm, plotted as a function of FREG-(48 -58)/FGF-2 molar ratio, shows a dose-dependent and saturable quench. C and D, SPR analysis showed that FGF-2 interaction with immobilized FGF-2 (C) or with immobilized heparin (D) reduces as a function of FREG-(48 -58), while SCR shows no effect. E, FGF-2 interaction with FGF-R1 by SPR. Increasing concentrations of FREG-(48 -58) (1, 0 ng/ml; 2, 0.7 ng/ml; 3, 2.1 ng/ml; 4, 20 ng/ml; 5, 180 ng/ml; 6, 1.4 g/ml; 7, 14 g/ml) dose-dependently inhibited FGF-2 interaction with immobilized FGF-R1, up to 40%, reaching the plateau at 180 ng/ml. A representative experiment is shown. This experiment was repeated three times with similar results.
ing strategies were addressed: (i) one peptide derived from this region was designed, synthesized, and tested in vitro and in vivo, (ii) an antibody was developed against this peptide and tested in FGF-2-dependent in vitro assays, and (iii) histidine chemical modification was carried out because one of the three FGF-2 histidines falls in the peptide region (i.e. His in position 50).
The secondary structure of residues 43-60 is characterized by a disordered segment, a short helix, a ␤-strand, and an additional short disordered segment. We selected a peptide including the short helix and the short ␤-strand, i.e. region 48 -58 (DPHIKLQLQAE; FREG-(48 -58)), to avoid excessive flexibility. FREG- (48 -58) was therefore used in the following in vitro and in vivo analyses, and for rabbit immunization to raise a specific antibody.
Proliferation and Migration Assays-FGF-2-induced BAEC proliferation was evaluated at 48 h in the absence and presence of increasing FREG-(48 -58) doses up to 1000 ng/ml ( Fig. 2A). FREG-(48 -58) showed a dose-dependent inhibitory activity and, at 10 ng/ml, lowered the FGF-2 mitogenic effect to the control level, while both scrambled peptides used as control (SCR) showed no activity. Therefore FREG- (48 -58) and SCR at 10 ng/ml were used for the following in vitro experiments.
Taken together these data indicate that a FREG-(48 -58)/ FGF-2 heterocomplex formation is observed in both solid phase and liquid phase conditions, with a K D (in fluorescence studies) falling in the nanomolar range.
These results indicate that, in the presence of FREG-(48 -58), FGF-2 internalization, which depends upon its binding to membrane-bound receptors, and FGF-R1 phosphorylation are markedly inhibited.
Histidine Residue Chemical Inactivation-Histidines are very reactive residues and often occupy functionally relevant sites in proteins. Human FGF-2 contains three histidines, His-16, His-35, and His-50, the latter falling within the FREG- (48 -58) region. In order to test whether histidine residues play a role in FGF-2 activity, their direct and specific chemical inactivation was achieved by DEPC treatment. Interestingly, DEPC-treated FGF-2 was unable to interact with itself ( Fig.  7A) and failed to induce BAEC chemotaxis (Fig. 7B). Further, inactivating FREG- (48 -58) histidine by DEPC incubation, abolished the FREG- (48 -58) ability to inhibit FGF-2-induced chemotaxis (Fig. 7B). Taken together these results show that histidines play a crucial functional role for FGF-2 activity in vitro, and indicate that the histidine present in FREG- (48 -58) is essential for its biological effects.
In Vivo Angiogenesis Assays-In order to evaluate whether FREG-(48 -58) modulates FGF-2 effects in vivo, two different assays were performed, i.e. the CAM assay and the Matrigel assay.
FREG- (48 -58) Effect in the CAM Assay-Ten CAM per group were treated with PBS, FGF-2 alone, FREG-(48 -58) alone, or FGF-2 ϩ FREG- (48 -58). Experiments were stopped at day 12 and examined macroscopically and histologically. FREG- (48 -58) abolished the new vessel formation induced by FGF-2, either inside the sponge trabeculae or at the boundary between the sponge and the CAM mesenchyme. Quantification of the effect achieved by histologic examination and planimetric vessel counting is reported in Table I, and a representative experiment is reported in Fig. 8A.
FREG- (48 -58) Effect in the Matrigel Assay-FGF-2-induced angiogenesis in Matrigel plugs injected subcutaneously in CD1 mice was also evaluated. Eight days after implant, Matrigel plugs were removed, histologically processed, and new blood vessel formation was measured. As shown in Fig. 8B, FREG-(48 -58) significantly inhibited (p Ͻ 0.01) FGF-2-induced vessel formation by 49 Ϯ 5%, while SCR was ineffective. Taken together, in vivo studies indicate that FREG-(48 -58) markedly inhibits FGF-2-induced angiogenesis in two different in vivo assays. DISCUSSION FGF-2 dimerization is reported to be a necessary event for receptor dimerization and signaling activation; therefore, region(s) directly involved in FGF-2 dimerization may represent crucial site(s) controlling the whole FGF-2 activity. The dimer in the absence of heparin has not been crystallized for FGF-2 although it is known for FGF-9 (50). Conversely, complexes in the presence of heparin (but lacking a direct FGF-FGF interaction) have been crystallized for FGF-1 (51) and FGF-2 (20). These studies revealed different dimerization/oligomerization features for different members of the FGF family, likely due to the differences in the ␤1, ␤2, ␤9, and ␤10 loops, outside the ␤ trefoil core (50). Different three-dimensional simulation-based studies predicted different organizations of the FGF-2 dimer in the absence of heparin (15)(16)(17) and indicated the theoretical possibility that FGF-2 monomers may interact and associate with different orientations. The dimer model predicted in this  For histologic assessment, every third section of 30 serial slides from each specimen was analyzed under a 144-point mesh inserted into the eyepiece of the photomicroscope. The total number of intersection points occupied by transversally cut vessels (3-10 m in diameter) were counted at ϫ250 magnification inside the sponge and at the boundary between the sponge and surrounding CAM mesenchyme in 6 randomly chosen microscopic fields per section and reported as mean Ϯ S.D. Asterisks indicate statistically significant difference from FGF-2 (p Ͻ 0.01). study shows a high content of hydrophobic residues at the dimer interface (i.e. 41%), which might contribute stability to the predicted dimer. FREG- (48 -58) was selected because its sequence is poorly conserved within the FGF family (Fig. 1A) (being specific for FGF-2, whereas other portions are conserved throughout the entire family), and because it falls at the predicted interface of the FGF-2 dimer. Further, it overlaps, or is close to, receptor-and heparin-binding residues. FREG- (48 -58) strongly and specifically inhibited FGF-2 activity in vitro and markedly reduced FGF-2-induced new vessel formation in vivo. An FGF-2/FREG-(48 -58) complex formation was observed and, likely as consequence of this, FGF-2 interaction with itself, with heparin, and with FGF-R1 was strongly reduced, as well as FGF-2-induced FGF-R1 phosphorylation and FGF-2 internalization. It is known that FGF-2 internalization follows its binding to either high affinity receptors or to heparan sulfate proteoglycans (51). Therefore, the observed FREG-(48 -58)-induced inhibition of FGF-2 internalization (Fig. 5, B and C) may be due either to reduced interaction with high affinity receptors (Fig. 4E), to the inhibition of FGF-2 selfinteraction (Fig. 4C), or to reduced interaction with heparan sulfate proteoglycans (Fig. 4D). In either case, the extensive inhibitory effect indicates that FREG-(48 -58) may interfere with an early event essential for the receptor activation. FGF-2 region 48 -58 may control an early event, likely dimerization, because the region 48 -58 stays at the predicted FGF-2 dimerization interface, essential to activate FGF-2 signaling. In this case, the observed reduced heparin-binding, receptor binding, and receptor activation may be consequences of dimer formation inhibition. However, a direct effect of FREG-(48 -58) on heparin binding and receptor binding cannot be ruled out at this stage. FREG-(48 -58) may affect FGF-2 dimer formation by directly interacting with FGF-2 and masking the dimerization site, although presently we do not have a definitive indication of the site of interaction with FGF-2. Alternative approaches confirmed that region 48 -58 may be an FGF-2 crucial functional domain. In fact, the antibody raised against FREG-(48 -58) recognized the whole FGF-2 and blocked FGF-2 internalization and its mitogenic effect, indicating that region 48 -58 is a relevant epitope and that it may have a crucial functional role. Further, histidine modification abolished FGF-2 and FREG- (48 -58) in vitro activities, indicating that histidines, and specifically His-50, play a crucial role for FGF-2 and FREG-(48 -58) activity, respectively. Site-directed mutagenesis studies will further characterize active residues in this region.
Several endogenous factors control angiogenesis (31, 53-54) and a number of molecules inhibiting FGF-2 are currently being investigated (47,(55)(56)(57)(58)(59)(60)(61)(62). Specifically inhibiting FGF-2dimerization with FREG-(48 -58) may represent a novel effective approach to modulate FGF-2 activity with potential therapeutic applications. To this respect, a biocomputing tool potentially helpful to investigate protein active sites has been recently made available (63) at crisceb.unina2.it/ASC/. In conclusion, results of the present study indicate a novel FGF-2 regulatory domain and identify a short peptide, which strongly inhibits FGF-2 activity both in vitro and in vivo. Further studies will be aimed at characterizing the potential therapeutic role of this peptide.