FGF-2/fibroblast growth factor receptor/heparin-like glycosaminoglycan interactions: a compensation model for FGF-2 signaling. ROBERT PADERA,*,â GANESH ...
FGF-2/fibroblast growth factor receptor/heparin-like glycosaminoglycan interactions: a compensation model for FGF-2 signaling ROBERT PADERA,*,† GANESH VENKATARAMAN,† DAVID BERRY,‡ RANGA GODAVARTI,‡2 AND RAM SASISEKHARAN‡,1 *Harvard Medical School, Boston, Massachusetts 02115, USA; †Harvard-MIT Division of Health Sciences and Technology, ‡Division of Bioengineering and Environmental Health, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA ABSTRACT Heparin-like glycosaminoglycans (HLGAGs) play a central role in the biological activity and signaling behavior of basic fibroblast growth factor (FGF-2). Recent studies, however, indicate that FGF-2 may be able to signal in the absence of HLGAG, raising the question of the nature of the role of HLGAG in FGF-2 signaling. In this study, we present a conceptual framework for FGF-2 signaling and derive a simple model from it that describes signaling via both HLGAG-independent and HLGAG-dependent pathways. The model is validated with F32 cell proliferation data using wild-type FGF-2, heparin binding mutants (K26A, K119A/R120A, K125A), and receptor binding mutants (Y103A, Y111A/W114A). In addition, this model can predict the cellular response of FGF-2 and its mutants as a function of FGF-2 and HLGAG concentration based on experimentally determined thermodynamic parameters. We show that FGF-2-mediated cellular response is a function of both FGF-2 and HLGAG concentrations and that a reduction of one of the components can be compensated for by an increase in the other to achieve the same measure of cellular response. Analysis of the mutant FGF-2 molecules show that reduction in heparin binding interactions and primary receptor site binding interactions can also be compensated for in the same manner. These results suggest a molecular mechanism that could be used by cells in physiological systems to modulate the FGF-2mediated cellular response by controlling HLGAG expression.—Padera, R., Venkataraman, G., Berry, D., Godavarti, R., Sasisekharan, R. FGF-2/fibroblast growth factor receptor/heparin-like glycosaminoglycan interactions: a compensation model for FGF-2 signaling. FASEB J. 13, 1677–1687 (1999)
Key Words: FGF receptor z thermodynamic parameters z extracellular matrix
The fibroblast growth factor (FGF)2 family plays a key role in development and morphogenesis, as well as in the pathogenesis of several disease states 0892-6638/99/0013-1677/$02.25 © FASEB
such as tumor growth and aberrant neovascularization (1–3). FGF family members bind to two types of receptors: a high-affinity signaling receptor present on the cell surface and a low-affinity binding receptor present both on the cell surface and in the extracellular matrix (ECM) (2, 4 – 6). The highaffinity signaling receptors are the fibroblast growth factor receptors (FGFR), of which there are four (FGFR1–FGFR4) (5, 7). The low-affinity receptor is the heparin-like glycosaminoglycan (HLGAG) component of heparin sulfate proteoglycans (HSPG) (5, 8 –12). Understanding the nature of the interactions of FGF-2 (basic FGF) with FGFR and the role of HLGAGs in this process is a fundamental issue that is still poorly understood (8). This is complicated by the fact that HLGAGs are complex polysaccharides present not only on the cell surface, but also in the ECM. As a result, they interact with FGF-2 both on the cell surface and in the ECM compartments (8, 13, 14). Thus, HLGAGs can play many roles along the trajectory of FGF-2 signaling events. HLGAGs provide a locale for a stable reservoir of FGF-2 in the ECM that can be released, upon appropriate stimuli, for cell signaling (15). FGF-2 bound by HLGAG in the ECM is protected from heat, acid, and protease degradation (16). HLGAGs have been implicated in the induction of FGF-2 dimerization or in the stabilization of self-associated FGF-2 molecules (11, 17– 19). This activity is thought to be important as a prerequisite for FGFR clustering. Finally, HLGAGs are thought to participate in interactions between 1
Correspondence: MIT, 77 Massachusetts Ave., Building 16 –561, Cambridge, MA 02139, USA. E-mail: ramnat@ mit.edu 2 Present address: Genetics Institute, One Burtt Rd., Andover, MA 01810 3 Abbreviations: ECM, extracellular matrix; FGF, fibroblast growth factor; FGF-2, basic FGF; FGFR, FGF receptor; HLGAG, heparin-like glycosaminoglycan; HRP, horseradish peroxidase; HSPG, heparin sulfate proteoglycans; IL, interleukin; PCR, polymerase chain reaction. 1677
FGF-2 and FGFR in a ternary signaling complex on the cell surface that could also involve FGFR:FGFR interactions directly induced or stabilized by HLGAG (20, 21). In physiological systems, HLGAGs have been shown to be important in various normal and pathological processes (3, 22). However, an ongoing debate in the literature centers around the nature of the role of HLGAGs in FGF-2 signaling and whether their presence is an absolute requirement for the activity (23–26). In this study, we propose a framework to study signaling by FGF-2 through FGFR1, in both the presence and absence of HLGAGs. Results presented herein indicate, as has been suggested by others (23, 24), that HLGAGs are not absolutely required for FGF-2 signaling but that HLGAGs facilitate signaling at a much lower FGF-2 concentration than is possible in its absence. More important, we demonstrate that the absence of FGF-HLGAG interactions can be compensated by increasing the FGF-2 concentration to achieve signaling in the absence of HLGAG. Similarly, HLGAG can be used to compensate for the reduction in the interactions between FGF-2 and FGFR. Through site-directed mutagenesis, we have explored the contribution of the HLGAG and FGFR binding residues of FGF-2 in the signaling process and how changes in the binding energy resulting from a mutation can be compensated. The experimental data provide a basis for the development of a thermodynamic formalism, enabling the determination of parameters for FGF-2 signaling as a function of both FGF-2 and HLGAG concentrations. We present a framework for systematically analyzing these compensatory effects and discuss how nature may take advantage of these principles in physiological and pathological processes.
acid amino-terminal leader sequence, which constitute a high-affinity site for Ni21 that can be cleaved with thrombin. The mutations were introduced into FGF-2 by the overlap extension polymerase chain reaction (PCR) methodology developed by Higuchi et al. (27). Briefly, two primary PCR reactions (12 cycles) were set up using the original FGFpET14b construct as template, followed by secondary PCR reactions using the primary PCR products as the template. The amplified secondary PCR product was cloned into pET14b expression vector (Novagen) using NdeI and SpeI restriction enzymes. After amplification and cloning in pET-14b vector, the sequence of the mutated genes was verified using Sequanase (U.S. Biochemicals Inc., Cleveland, Ohio). The plasmid containing the gene in pET-14b was isolated, purified, and used to transform the host BL21(DE3) cells. Expression, isolation, and purification of mutant FGF-2 in E. coli
MATERIALS AND METHODS
Overnight cultures (100 ml) of BL21(DE3) (Novagen) containing the FGF-2 genes in pET-14b were grown to an OD600 of 0.5, induced with 1 mM iso-phenyl-thio-galactoside (IPTG) for 2 h and harvested. The cell pellet was resuspended in 1/20th volume binding buffer (20 mM Tris, 500 mM NaCl, 5 mM imidazole). The resuspended culture was placed in an ice bath, sonicated for 2 min using a Branson sonicator (model no. 450, power 3, 50% pulse; CT) and centrifuged at 4°C and 15,000 3 g for 30 min. The supernatant was purified by Ni21 affinity chromatography using Sepharose 6B Fast Flow resin covalently linked to nitrilotriacetic acid (Novagen). Briefly, the resin was charged with five column volumes of 200 mM NiSO4 and equilibrated with five column volumes of binding buffer. A 6 –10 ml sample was then applied, followed by 12 ml binding buffer, 9 ml 15% elution buffer (20 mM Tris, 500 mM NaCl, 200 mM imidazole), and 10 ml 100% elution buffer. Typically, 1 mg of purified FGF was obtained per liter of culture. Purity of the eluted protein was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis with a Mini Protean II electrophoresis apparatus (Bio-Rad, Hercules, Calif.). Visualization of proteins in gels (12% gels) was accomplished with 0.1% Coomassie blue stain or silver stain (Bio-Rad). Heparin-POROS chromatography was used to confirm the loss of heparin binding for the mutants K26A, K119A/R120A, and K125A. Circular dichroism was performed on each of the mutant proteins and shown to be identical with wild-type FGF-2 protein (data not shown).
Proteins and reagents
FGF-2 assay (ELISA)
Recombinant human FGF-2 and monoclonal FGF-2 antibody 11.1 were generously provided by Scios, Inc. (Mountainview, Calif.). Extreme care was taken to ensure the absence of any HLGAGs during the storage, handling, expression, isolation, and purification of FGF-2 and the mutants described below. Goat anti-mouse horseradish peroxidase- (HRP) conjugated antibodies and SuperSignal ULTRA chemiluminescent HRP substrates were obtained from Pierce (Rockford, Ill.). Heparin (HLGAG) from porcine mucosa was obtained from Celsus Laboratories (Columbus, Ohio) and the average molecular mass was assumed to be ;17,000 Da.
Samples of standard (Scios, FGF-2) and test (recombinant and mutant) proteins were prepared by serial dilution in phosphate-buffered saline in the concentration range of 1–200 ng/ml. They were added to a 96-well microtiter plate and incubated at 37°C for 2 h to coat the wells. The wells were washed with water and incubated with blocking buffer (borate-buffered saline containing 0.05% Tween 20, 1 mM EDTA, 0.25% BSA, 0.05% NaN3). Fifty microliters of murine mAb 1 (mAb 11.1, which binds to linear sequence residues 21–27, concentration 1000 ng/ml) was added to the FGF-2coated wells and incubated at 37°C for 2 h. After washing and blocking, 50 ml of HRP-conjugated anti-mouse immunoglobulin G 1 antibody (Pierce Chemicals) was added at a concentration of 1000 ng/ml and incubated at 37°C for 2 h. After washing and blocking the wells, they were incubated with 150 ml of a substrate for HRP (3, 39, 5, 59 tetramethylbenzidine; Pierce) for 30 min at room temperature. One hundred microliters of 1 M H2SO4 was added to stop the reaction. The plate was read in a microtiter plate reader at 450 nm and the
Site-directed mutagenesis FGF-2 was produced as a soluble protein in BL21(DE3) Escherichia coli host, using the pET14b system (Novagen, Madison, Wis.). This construct has a histidine tag (six consecutive histidines) and a thrombin cleavage site in a 21 amino 1678
Vol. 13
October 1999
The FASEB Journal
PADERA ET AL.
readings were plotted vs. the standard FGF concentration to determine the concentrations of the test samples. Cell culture BaF3 and F32 cells were a generous gift of Dr. David Ornitz (Washington University, St. Louis, Mo.). These cells were maintained in suspension culture in RPMI 1640 with 10% calf serum, 10% WEHI-3-conditioned media, 2 mM L-glutamine, 100 U/ml penicillin, and 100 mg/ml streptomycin. Cultures were grown in 25 cm2 flasks in a 37°C/5% CO2 incubator and were passaged 1:10 by dilution thrice weekly. WEHI-3 cells (ATCC, TIB 68) were grown in suspension culture in RPMI 1640 medium supplemented with 10% calf serum and 0.05 mM 2-mercaptoethanol in a humidified 37°C/5% CO2 incubator. Supernatants from the WEHI-3 cells [which produce interleukin 3 (IL-3), an antiapoptotic cytokine for F32 cells] were collected by incubating 3 3 106 cells/ml in their media for 3 days. Supernatants were recovered by centrifugation at 800 3 g, filtered through an 0.45 mM filter and stored at 270°C. RPMI 1640, calf serum, 1003 penicillin-streptomycinglutamine, and 2-mercaptoethanol were from Sigma Chemical (St. Louis, Mo.). Proliferation studies F32 cells were collected from the propagating culture, washed three times with IL-3-deficient media (i.e., without WEHI-3conditioned media), and counted with an electronic cell counter (Coulter, Inc., Hialeah, Fla.). The cells were resuspended in IL-3-deficient media to a density of 1 3 105 cells/ml. This cell suspension (1 ml) was added to each well of a 24-well tissue culture plate along with the appropriate concentration of FGF-2 and heparin. The negative control wells contained the cell suspension without growth factor or heparin; positive control wells contained the cell suspension with 2.9 nM (50 ng/ml) recombinant FGF-2 and 29.4 nM (500 ng/ml) heparin. The cells were incubated for 72 h in the 37°C/5% CO2 humidified incubator and counted with a Coulter counter at the experimental end point. These experiments benefit from the use of the F32 cell line, which lacks cell surface HLGAGs and has been genetically engineered to express FGFR1 (9). The concentration of HLGAG is dependent only on the amount added in the media; no pretreatment with chlorate or heparinase was needed to rid the cells of cell surface HLGAGs, allowing an accurate measurement of concentration effects. Data analysis For simplicity, the data have been normalized as follows. The proliferative index was defined as the increase in cell number for the experimental case divided by the increase in cell number for the positive control. The positive control were F32 cells incubated with 50 ng/ml wild-type FGF-2 and 500 ng/ml heparin. The positive control conditions were found to yield the maximal proliferative response for FGF-2, using the F32 cells. The proliferative index was independently determined for each experiment in order to eliminate potentially confounding factors, such as receptor number per cell, [R] in the equations that follow. The model described below predicts that at very high FGF-2 concentrations in the absence of HLGAG (Ho50), the signaling tends toward k3/k4. In the experiments shown (as well as in data that are not shown), the maximal signaling for FGF-2 in the absence of HLGAG (HLGAG-independent signaling in Fig. 2) is ;65% of the maximal signaling for FGF-2 in the presence of HLGAG (HLGAG-dependent signaling in HLGAG MODULATION OF FGF-2 SIGNALING
Figure 1. F32 cell proliferation as a function of wild-type FGF-2 concentration in the absence of heparin. The proliferation index was calculated as described in Materials and Methods and plotted against the total FGF-2 concentration. The filled circles represent the experimental data (averaged over four experiments) along with the standard deviations. The solid line represents the expected response calculated from Equation 3, with k1 5 1.08 nM, k3 5 0.65, k4 5 1 and setting Ho 5 0.
Fig. 2). Therefore, we have set k3 5 0.65 and k4 5 1 to account for this information.
RESULTS Framework and rationale Figure 1 shows a FGF-2 cell-mediated response in a concentration-dependent fashion in the absence of HLGAGs, as described in earlier reports (23, 24). These experiments were carried out in the F32 cell line, which is devoid of endogenous HLGAGs. Increasing the FGF-2 concentration in the absence of any HLGAGs increases the proliferation index (cellular response or signal) of the F32 cells. The data suggest that FGF-2 can signal even in the absence of HLGAGs. We propose a simple and unified framework for addressing HLGAG-dependent and HLGAG-independent signaling by FGF-2, shown in Fig. 2A. In this framework, efficient and optimal signaling can be achieved by a ternary interaction between FGF-2: FGFR:HLGAG. The signaling in the absence of HLGAGs need not be as efficient, but can nevertheless be assumed to be a parallel signaling mechanism. Such a framework can account for the observed HLGAG-mediated FGF-2 dimerization or oligomerization and consequent receptor dimerization and clustering observed (3). Furthermore, no assump1679
Figure 2. Schematic model of FGF-2 signaling. A) FGF-2 and FGFR come together to form a signaling complex on the cell surface in the absence (left) or presence (right) of HLGAG. Signaling in the absence of HLGAG may be less efficient than in the presence of HLGAG, indicated by a thinner arrow. HLGAG is known to induce FGF-2 dimerization leading to FGFR clustering and enhanced signaling. In the absence of HLGAG, FGF-2 self-association may play the same role at high concentrations. B) Five distinct yet related interactions can therefore be identified: 1) Interaction of FGF-2 with FGFR, 2) interaction of FGF–2 with HLGAG leading to FGF-2 dimerization, 3) interaction of FGF–2 with itself to form FGF-2 dimers or oligomers, 4) interaction of HLGAG with FGFR, and 5) interaction of FGFR with itself to form FGFR dimers or oligomers. The model makes no assumptions regarding the stoichiometry of the interactions between the components. Panel C summarizes the equations of panel B as Equations 1 and 2 (see text).
tions regarding the stoichiometry of the interactions were made, enabling us to develop a set of thermodynamic parameters for the different interactions based on experimental data. Under the experimental conditions, different groups of interactions can be unified into single thermodynamic parameters by simplifications of the equations in Fig. 2B. In Fig. 2C, the thermodynamic parameters k1 and k3 describe FGF-2 signaling in the absence of heparin. Within k1 are embedded FGF-2:FGFR interactions, FGFR:FGFR interactions, and FGF-2:FGF-2 interactions. Similarly, the parameters k2 and k4 relate to the possible interactions of HLGAG, FGF-2, and FGFR and the signaling resulting from them. The parameters k3 and k4 are also a measure of the efficiency of signaling of the FR and FHR complexes, respectively. Therefore, the following two equations will be used in conjunction with experimental data to determine these parameters. Assuming that the observed saturation results from limiting receptor concentrations, we can derive from first principles the relationship between normalized signaling (y) and the initial concentrations of FGF 1680
Vol. 13
October 1999
(Fo) and HLGAGs (Ho).3 Assuming these two parallel reactions are additive, the signaling for this unified model can be written as:
3
Consider the reactions, k1 k3 ¡ FR O ¡ SIGNAL F1RO ¢ k2 k4 ¡ FHR O ¡ SIGNAL F1H1R O ¢
Where F 5 FGF, R 5 Receptor, H 5 Heparin, and S 5 Signal S 5 k3(FR) 1 k4(FHR), Smax 5 k4Ro, Ro 5 (FR) 1 (FHR) 1 R k1 5
~F!~R! ~FR!
y5
~1!
k2 5
~F!~R!~H! ~FHR!
S k3 ~FR! 1 k4 ~FHR! 5 Smax k4Ro
~2!
~3!
Substitute for (FR), (FHR), and Ro from (1), and (2) into (3),
The FASEB Journal
PADERA ET AL.
k3 k4 1 Ho k1 k2 y5 k4 k4 k4 1 Ho 1 k1 k2 Fo
(Eq. 3; see footnote 3)
This equation represents a 2-dimensional surface in Ho and Fo that relates to the total signal S. Isosignaling contours on this surface suggest that the same amount of signaling can be achieved by several combinations of the FGF-2 and HLGAG concentrations. This treatment of the signaling events as thermodynamic processes further suggests that FGF2:FGFR complexes (and hence signaling) can be modulated by the concentrations of the components of the complex. For instance, a reduction in the FGF-2 concentration can be compensated by an increase in the heparin concentration (and vice versa) to achieve the same measure of signaling over an appropriate concentration range. Similarly, if a mutation in FGF-2 were to change k2, one could compensate by increasing FGF-2 concentration, heparin concentration, or both to achieve the same number of FR or FHR complexes (and therefore signaling), as with wild-type FGF-2. The above framework allows us to systematically study the contributions of these components—FGF-2 and HLGAG—in signaling. Using wild-type FGF-2, we studied the effects of FGF-2 and HLGAG concentrations on cellular response in F32 cells (see Materials and Methods), as well as the effects of the mutations on the thermodynamic parameters and cellular response. Wild-type FGF-2 The thermodynamic parameters that govern the interaction between FGF-2 and FGFR in the absence of HLGAG can be determined from the data presented in Fig. 1. Setting Ho 5 0 in Equation 3 and using the values of k3 and k4 described in Materials and Methods, the value of k1 calculated from the data is 1.08 nM. The calculated trend for the signaling using this k1 value in Equation 3 is also shown in
k3 k4 ~F!~R! 1 ~F!~R!~H! k1 k2 y5 ~F!~R! ~F!~H!~R! k4 1 1R k1 k2
S
D
As a first approximation, at equilibrium, F ' Fo, and H ' Ho (see text), and simplifying k3 k4 1 Ho k1 k2 y5 k4 k4Ho k4 1 1 k1 k2 Fo
S
D
HLGAG MODULATION OF FGF-2 SIGNALING
Figure 3. F32 cell proliferation as a function of wild-type FGF-2 concentration in the presence of 30 nM heparin. Proliferation index was calculated as described in Materials and Methods and plotted against the total FGF concentration added. The filled circles represent the experimental data (averaged over four experiments) along with the standard deviations. The solid line represents the signaling calculated from Equation 3 with k1 5 1.08 nM, k3 5 0.65 and k4 5 1.
Fig. 1. It can be seen that this model fits the data very well. Figure 3 shows FGF-2 signaling as a function of FGF-2 concentration in the presence of 500 ng/ml heparin. Maximal proliferation occurs at an FGF-2 concentration of 0.6 nM (10 ng/ml), corresponding to a 503 reduction in the amount of FGF-2 needed to reach plateau than when the experiment is run in the absence of heparin (Fig. 1). Furthermore, increasing either the FGF-2 concentration or the heparin concentration beyond these points does not alter signaling, confirming that the receptor concentration on the cell surface is the rate-limiting reactant (22). Using the above determined value for k1, the value of k2 can be determined from Equation 3. The value of k2 for wild-type FGF-2 is calculated to be 4.0 nM. The signaling calculated from Equation 3 is plotted on the same graph, and again shows that the fit between the experimental data and the model is quite good. Using these parameters for wild-type FGF-2, signaling can be written as: y5
0.60 1 0.25Ho 1 0.93 1 0.25Ho 1 Fo
(Eq. 4)
This equation predicts that setting the FGF-2 concentration at 2.9 nM and varying the heparin concentration will result in an increase in signaling in a hyperbolic fashion. At low heparin concentrations, the HLGAG-independent signaling mechanism is dominant and the signaling corresponds to that of 1681
that in this plot, two sets of data (varying FGF concentrations at Ho50 and Ho529 nM) are used to regress the model parameters, and this is used to predict the third data set (FGF constant, varying Ho). For this data set, the predicted data are compared to the experimental data for varying Ho, keeping Fo constant. FGF-2 mutants
Figure 4. Model prediction of signaling as a function of heparin concentration. Over the concentration range of 0.001 nM to 100 nM, the prediction from Equation 4 is drawn on semi-log axes. Also shown in the graph is the experimentally measured proliferation index for the case of an FGF-2 concentration of 2.9 nM (50 ng/ml), with heparin concentration varying over the range of 3 pM to 58 nM (1000 ng/ml). The filled circles represent the experimental data (averaged over four experiments) along with the standard deviations.
FGF-2 in the absence of heparin. Increasing the HLGAG concentration increases the overall proliferation rapidly, resulting in saturation at heparin concentrations greater than 5.8 nM. Figure 4 shows the predicted signaling at this FGF-2 concentration as a function of heparin concentration. The experimental data corresponding to an FGF-2 concentration of 50 ng/ml with varying heparin concentration are also plotted in Fig. 4. At low heparin concentrations (,0.06 nM), the experimental data indeed show that the signaling is equivalent to that of signaling in the absence of heparin at the same FGF-2 concentration, as predicted by the model. The proliferation index reaches a plateau at a heparin concentration of 5.8 nM, as predicted by the model. As mentioned above, Equation 4 predicts a 2-dimensional surface for the dependence of signaling on FGF-2 and heparin concentration. Figure 5 shows the dependence of signaling (z axis) on the concentration of FGF-2 and heparin (x and y axis, respectively). This surface shown in Fig. 5 allows us to predict the proliferation response as a function of heparin concentration at a given FGF-2 concentration, and vice versa. Furthermore, the experimental data shown in Figs. 1, 3, and 4 can be superimposed on this 3-dimensional plot. Such a figure combines the results of the modeling and the experimental data into one plot; this format will be used to present data on the FGF-2 mutants. It is important to note 1682
Vol. 13
October 1999
Site-directed mutagenesis studies have identified specific sites on FGF-2 as being involved in HLGAG binding (28, 29) or FGFR binding (30). Specific mutants that alter the binding interaction of FGF-2 to HLGAG or FGFR can therefore be used to further investigate and corroborate the unified model presented above. The FGF-2 heparin binding mutants K26A, K119A/R120A, and K125A and the FGF-2 receptor binding mutants (Y103A and Y111A/ W114A) were chosen for study; the results are described below. K26A K26 is situated in the primary heparin binding site and is thought to form salt bridges with the anionic sulfates of HLGAG (28, 30, 31). A mutation of K26 to Ala significantly reduces FGF-2 binding to HLGAGs
Figure 5. 3-Dimensional plot of signaling by wt-FGF-2. The concentration axes (Fo and Ho) are log scales and the proliferation index is plotted on a linear scale. In the absence of heparin (Ho50, the front end of the surface), the proliferation index increases with increasing FGF-2 concentration and saturates at ;0.65. This surface enables easy visualization of the dependence of heparin concentration as well as the FGF concentration on the proliferation index. The experimental data from Figs. 1 and 3 used to calculate the thermodynamic parameters are also shown (the two curves parallel to the Fo axis); data used to compare model predictions (Fig. 4) are shown parallel to the Ho axis. For the sake of clarity, the surface at these locations is cut to highlight the location of the experimental points.
The FASEB Journal
PADERA ET AL.
bridge interactions with HLGAG (28, 29). A double mutation of K119 and R120 to Ala significantly reduces FGF-2 binding to HLGAGs (28, 29). Experiments with varying K119A/R120A concentrations in the absence of HLGAG resulted in a k1 of 5.63 nM; in the presence of HLGAG, k2 was found to be 7.29 nM (2). Using these values, signaling as a function of K119A/R120A and heparin concentration can be described by Equation 6. y5
Figure 6. 3-Dimensional plot of signaling by K26A. The axes are as described in Fig. 5. Two sets of experimental data (parallel to the Fo axis) are used to calculate the parameters for Equation 5. The predicted surface fits very well with data on varying heparin concentration (analogous to Fig. 4 for wt FGF). The data are plotted as described in the legend to Fig. 5.
(28, 30). The values for k1 and k2 were calculated to be 2.53 and 18.03 nM for K26A in the same way as described for wild-type FGF-2. Using these values, signaling as a function of K26A and heparin concentration can be described by Equation 5. y5
0.26 1 0.055Ho 1 0.40 1 0.055Ho 1 Fo
(Eq. 5)
Figure 6 shows the 3-dimensional graph of the signaling for this mutant as predicted by Equation 5. Note that although a significant amount of binding energy has been disrupted by this mutation, complete activity can be recovered by increasing the K26A or the HLGAG concentration. The predicted profile for the dependence of signaling on the K26A and HLGAG concentration match almost exactly with the experimental data plotted on the same axes. Comparison of Fig. 6 with Fig. 5 shows that signaling in the absence of HLGAGs is almost identical, whereas in the presence of HLGAG, the signaling behavior of this mutant is shifted toward higher concentrations compared with wild-type FGF-2.
0.12 1 0.14Ho 1 0.18 1 0.14Ho 1 Fo
(Eq. 6)
Figure 7 shows the 3-dimensional graph of the signaling for K119A/R120A as predicted by the Equation 6. As with K26A, complete activity can be recovered by increasing the concentration of K119A/R120A or HLGAG. Figure 7 also shows that the experimental data fall exactly on the contours predicted by Equation 6. A comparison of Figs. 5 and 7 reveals that the surface profiles are almost identical between the wild-type FGF-2 and K119A/R120A except for a marginal shift in both the Fo and Ho concentrations. K125A A mutation of K125A also is reported to reduce interactions between FGF-2 and HLGAG (28, 29). However, the values for k1 and k2 from our experiments were found to be much higher than wild-type FGF-2 (157 nM and 84 nM (2) for K125A. These values for k1 and k2 indicate that the proliferative response will be less than that of wild-type FGF-2 at a given set of concentrations. Using these values, sig-
K119A/R120A The positively charged cluster of amino acids close to the carboxyl-terminal region of FGF-2 has also been implicated in HLGAG binding (28, 29). For instance, the amino acid residues K119, R120, and K125 are situated in the primary heparin binding site and are responsible for hydrogen bonding and salt HLGAG MODULATION OF FGF-2 SIGNALING
Figure 7. 3-Dimensional plot of signaling by K119A/K120A. The axes are as described in Fig. 5. The data are plotted as described in the legend to Fig. 5. 1683
(2), as compared to k2 of 1.08 nM (2) for wild-type FGF-2. Using these values, signaling as a function of FGF-2 and heparin concentration can be described by Equation 8. y5
Figure 8. 3-Dimensional plot of signaling by K125A. The axes are exactly as described in Fig. 5. The data are plotted as described in the legend to Fig. 5. Note that the surface is shifted toward higher Fo and Ho concentrations compared to Fig. 5.
naling as a function of K125A and heparin concentration can be described by Equation 7. y5
0.0041 1 0.012Ho 1 0.0064 1 0.012Ho 1 Fo
(Eq. 7)
The 3-dimensional plot from the model and the experimental data for K125A are shown in Fig. 8. As with the previous mutants, the experimental data fit almost exactly with the model predictions. A comparison of Fig. 8 with Fig. 5 shows that the signaling behavior of this mutant is shifted toward significantly higher concentrations of K125A and heparin as compared to wild-type FGF-2, while the overall trends remain the same.
0.0059 1 0.062Ho 1 0.0091 1 0.062Ho 1 Fo
(Eq. 8)
The 3-dimensional plot from the model, along with the experimental data, is shown in Fig. 9. As with the previous mutants, the experimental data fit almost exactly with the model predictions. In contrast to the previous mutants, a comparison of Fig. 9 with Fig. 5 shows that the signaling behavior of this mutant is shifted from the wild-type FGF-2 in an asymmetric manner. The shift is much more dramatic along the Y103A axis and is shifted to a lesser extent along the heparin axis. Y111A/W114A Y111A/W114A is a mutation of FGF-2 in the secondary receptor binding site (30). Just as Y103A was an ideal mutant to investigate the primary receptor binding site, Y111A/W114A is used to determine whether changes in FGF-2 FGFR interaction at the secondary site of FGF-2 can be compensated for by an increase in FGF or heparin concentration. This mutant was inactive at all protein concentrations both in the presence and absence of heparin (see Discussion). DISCUSSION In this study, we have developed a simple, unified framework that suggests that FGF-2-mediated cellu-
Y103A Y103A is a well-studied mutant of the primary receptor binding site on FGF–2. The residue Y103 has been reported to be critical for the binding of FGF-2 to FGFR, and mutation of this amino acid leads to loss of activity (30). In contrast to native FGF–2, Y103A only minimally stimulates cell proliferation in the absence of heparin over the concentration range studied, giving a k1 value of 110 nM, a full two orders of magnitude higher than that of wild-type FGF-2. However, Y103A is able to signal to the same extent as wild-type FGF-2 in the presence of 29.4 nM heparin at 29.4 nM Y103A. In essence, the combination of 29.4 nM heparin and increased Y103A concentration is able to rescue the dramatic loss of activity that is seen in the absence of heparin for this mutant. This corresponds to a k2 value of 16.15 nM 1684
Vol. 13
October 1999
Figure 9. 3-Dimensional plot of signaling by Y103A. The axes and the plots are as described in Fig. 5. There is almost no cell response in the absence of HLGAG, but complete activity can be recovered at higher concentrations of FGF and HLGAG.
The FASEB Journal
PADERA ET AL.
lar responses in the absence of HLGAGs can proceed in an independent and parallel fashion to the responses with a ternary complex with HLGAGs. We have developed a thermodynamic framework for FGF-2-mediated cellular response in the presence or absence of HLGAGs, and the framework predicts that a cellular response can be achieved by appropriately adjusting the concentrations of FGF-2 and HLGAG. The model developed not only fits the experimental data extremely well, but can also predict trends of the cellular response at different HLGAG and FGF concentrations. The mathematical formulation for signaling by the FGF-FGFR-HLGAG systems developed in this study suggests that the same extent of cellular response can be achieved by several combinations of FGF-2 and HLGAG concentrations. Furthermore, as with other equilibrium reactions, this model also suggests that decreases in the binding interactions between any two components can be compensated for by increasing the effective concentration of either component. Compensation of HLGAG and FGF-2 components Mutations that affect the binding interactions between FGF-2 and HLGAG (K26A, K119A/K120A) are able to achieve the same magnitude of cellular response as wild-type FGF-2 in the presence of HLGAG, but at marginally higher FGF-2 or HLGAG concentrations. This is reflected in the much higher k2 values compared to wild-type, especially for K26A. On the other hand, the cellular response in the absence of HLGAG for these mutations is almost identical, as reflected in the similar k1 values for these mutants and wild-type FGF-2. Decoupling the HLGAG-independent and the HLGAG-dependent cellular responses in our framework has enabled us to identify the specific interactions that are affected by a mutation. The K125A mutant, which was expected to affect only the HLGAG binding, was found to influence the cellular response in both the presence and absence of HLGAG (k1 and k2 are over 50-fold higher compared to wild-type FGF-2). It is possible that this mutation affects additional interactions between FGF-2 and FGFR or influences other interactions, such as FGF–FGF dimerization. The interaction of FGF with FGFR has been mapped to two different sites, termed the primary and secondary receptor binding sites on FGF-2. Residues at the primary binding site (Y24, N101, Y103, L140, M142) contribute a significant amount of binding energy to the interaction with the receptor (24). Using Y103A as a model mutant, we show here that the contribution from this residue can be compensated for by increasing the protein concentrations. On the other hand, the secondary binding site residues of FGF-2 (amino acids 106 –115) (24) HLGAG MODULATION OF FGF-2 SIGNALING
are not conserved in the FGF-2 family (unpublished observation). These residues (e.g., Y111, W114), although not contributing significant binding energy, seem to be critical for FGF-2 activity. A Y111A/ W114A double mutant cannot be rescued by any of the changes in protein or HLGAG concentration in the range tested here (data not shown). It is possible that a mutation in the secondary receptor binding site affects k3 and k4, which govern the progression of the FR or FHR complex to the measured proliferative response. Physiological implications A common requirement shared by all tyrosine kinase receptors is the formation of a threshold number of appropriately phosphorylated cytoplasmic domains to stimulate the signaling cascade. This implies that a threshold number of active receptor-ligand complexes need to be present on the cell surface for an adequate period of time to result in effective signaling. Although a thermodynamic treatment of the signaling events is able to explain and predict several of the important trends in FGF-2 and FGF-2 mutant signaling, it is important to point out that the kinetic parameters of the interactions between FGF-2:FGFR: HLGAG are also very important. The off-rates of FGF-2 binding to FGFR have been shown to be dramatically reduced in the presence of HLGAGs (32), suggesting that the ternary complex is in the associated state for a much longer period than is the binary complex between FGF and FGFR. This ‘duration of complex’ might also be responsible for the differences in the efficiency of signaling in the absence of HLGAG (k3 vs. k4) and will be affected by the changes in concentrations. Recent studies have revealed that the intracellular transport of FGF-2 from the cell surface to the nucleus is dependent on HLGAGs (33), and this may contribute to differences observed in cell proliferation. In addition, the role of HLGAG binding to FGFR and how this affects FGF-2-mediated cellular response remains to be understood (20). Data presented here clearly show that though FGF-2 can signal in the absence of HLGAGs, the type and amount of HLGAG can significantly modulate FGF-2 signaling. HLGAG stabilizes the intermolecular interactions and provides a method for amplification of the system so that signaling can occur at lower FGF-2 concentrations than would be possible in the absence of HLGAG. This provides cells a manner whereby they can become either more sensitive or less sensitive to the presence of a given FGF-2 concentration in the extracellular environment. Physiologically, the cells can up- or downregulate the expression of HLGAGs on their surface, and thus modulate the nature and strength of signaling from the FGF-2 stored in the matrix (34). 1685
The data shown here indicate that HLGAG levels can modulate the activity of FGF-2. This raises an interesting issue in view of the physiological regulation of HLGAGs by cells themselves. Proliferation and binding data for FGF-2 have been obtained in a variety of different cell types under various conditions, sometimes yielding perplexing or contradictory results (35). Could some of these disparities be a result of differential expression of surface HLGAGs from cell type to cell type, or even from week to week, in a given culture? For example, could a primary capillary endothelial cell culture used at passage 9 have a different complement of cell surface HLGAG than one used at passage 13, yielding different results as a consequence? This underscores the utility of the F32 cells to investigate the role of HLGAG in FGF signaling, where the only HLGAG in the system is exogenously added. Although it is clear that HLGAGs can regulate FGF-2 signaling and biological activity, the mechanistic and molecular features of regulation of FGFR activity by HLGAGs are still under investigation. This study provides a framework for the design of experiments investigating the contributions of HLGAGs in the different steps of receptor complex formation leading to cellular activation. This work was supported in part from the Sloan-Cabot Foundation, Massachusetts Institute of Technology. We are grateful to Dr. Judy Abraham (Scios) for FGF-2 and to Dr. David Ornitz (Washington University) for the supply of F32, BaF3, and other reagents. We also thank Dr. Barbara Natke and others for critical comments.
11. 12.
13.
14.
15.
16.
17. 18.
19. 20.
21. 22. 23.
REFERENCES 24. 1. 2. 3. 4. 5. 6.
7. 8. 9.
10.
1686
Basilico, C., and Moscatelli, D. (1992) The FGF family of growth factors and oncogenes. Adv. Cancer Res. 59, 115–165 Sporn, M. B., and Roberts, A. B. (1990) Peptide Growth Factors and Their Receptors, Springer-Verlag, Berlin Klein, S., Roghani, M., and Rifkin, D. B. (1997). Fibroblast growth factors as angiogenesis factors: new insights into their mechanism of action. Exper. Suppl. (Basel) 79, 159 –192 Conrad, E., (1998) Heparin Binding Proteins, pp. 301–348, Academic Press, New York Rapraeger, A. C. (1993) The coordinated regulation of heparan sulfate, syndecans and cell behavior. Curr. Opin. Cell Biol. 5, 844 – 853 Ornitz, D. M., Xu, J., Colvin, J. S., McEwen, D. G., MacArthur, C. A., Coulier, F., Gao, G., and Goldfarb, M. (1996) Receptor specificity of the fibroblast growth factor family. J. Biol. Chem. 271, 15292–15297 Givol, D., and Yayon, A. (1992) Complexity of FGF receptors: genetic basis for structural diversity and functional specificity. FASEB J. 6, 3362–3369 Sasisekharan, R., Ernst, S., and Venkataraman, G. (1997) On the regulation of fibroblast growth factor activity by heparin-like glycosaminoglycans. Angiogenesis 1, 45–54 Ornitz, D. M., Yayon, A., Flanagan, J. G., Svahn, C. M., Levi, E., and Leder, P. (1992) Heparin is required for cell-free binding of basic fibroblast growth factor to a soluble receptor and for mitogenesis in whole cells. Mol. Cell. Biol. 12, 240 –247 Rapraeger, A. C., Krufka, A., and Olwin, B. B. (1991) Requirement of heparan sulfate for bFGF-mediated fibroblast growth and myoblast differentiation. Science 252, 1705–1708
Vol. 13
October 1999
25. 26. 27. 28.
29.
30.
31.
Klagsbrun, M., and Baird, A. (1991). A dual receptor system is required for basic fibroblast growth factor activity. Cell 67 229 –231 Yayon, A., Klagsbrun, M., Esko, J. D., Leder, P., and Ornitz, D. M. (1991) Cell surface, heparin-like molecules are required for binding of basic fibroblast growth factor to its high affinity receptor. Cell 64, 841– 848 Kato, M., Wang, H., Kainulainen, V., Fitzgerald, M. L., Ledbetter, S., Ornitz, D. M., and Bernfield, M. (1998) Physiological degradation converts the soluble syndecan-1 ectodomain from an inhibitor to a potent activator of FGF-2. Nat. Med. 4, 691– 697 Aviezer, D., Hecht, D., Safran, M., Eisinger, M., David, G., and Yayon, A. (1994) Perlecan, basal lamina proteoglycan, promotes basic fibroblast growth factor-receptor binding, mitogenesis, and angiogenesis. Cell79, 1005–1013 Vlodavsky, I., Miao, H.-Q., Medalion, B., Danagher, P., and Ron, D. (1996) Involvement of heparan sulfate and related molecules in sequestration and growth promoting activity of fibroblast growth factor. Cancer Metast. Rev. 15, 177–186 Folkman, J., Klagsbrun, M., Sasse, J., Wadzinski, M., Ingber, D., and Vlodavsky, I. (1988) A heparin-binding angiogenic protein— basic fibroblast growth factor—is stored within basement membrane. Am. J. Pathol. 130, 393– 400 Schlessinger, J., Lax, I., and Lemon, M. (1995). Regulation of growth factor activation by proteoglycans: what is the role of the low affinity receptors? Cell 83, 357–360 Venkataraman, G., Sasisekharan, V., Herr, A. B., Ornitz, D. M., Waksman, G., Cooney, C. L., Langer, R., and Sasisekharan, R. (1996) Preferential self-association of basic fibroblast growth factor is stabilized by heparin during receptor dimerization and activation. Proc. Natl. Acad. Sci., USA93, 845–580 Waksman, G., and Herr, A. B. (1998). New insights into heparin-induced FGF oligomerization. Nature Struct. Biol. 5, 527–530 Kan, M., Wang, F., Xu, J., Crabb, J. W., Hou, J., and McKeehan, W. L. (1993) An essential heparin-binding domain in the fibroblast growth factor receptor kinase. Science 259, 19918 – 19921 Bikfalvi, A., Klein, S., Pintucci, G., and Rifkin, D. B. (1997) Biological roles of fibroblast growth factor-2. Endocr. Rev. 18, 26 – 45 Basilico, C., and Moscatelli, D. (1992) The FGF family of growth factors and oncogenes. Adv. Cancer Res. 59, 115–165 Roghani, M., Mansukhani, A., Dell’Era, P., Bellosta, P., Basilico, C., Rifkin, D. B., and Moscatelli, D. (1994) Heparin increases the affinity of basic fibroblast growth factor for its receptor but is not required for binding. J. Biol. Chem. 269, 3976 –3984 Fannon, M., and Nugent, M. A. (1996) Basic fibroblast growth factor binds its receptors, is internalized, and stimulates DNA synthesis in Balb/c3T3 cells in the absence of heparan sulfate. J. Biol. Chem. 271, 17949 –17956 Filla, M. S., Dam, P., and Rapraeger, A. C. (1998) The cell surface proteoglycan syndecan-1 mediates fibroblast growth factor-2 binding and activity. J. Cell. Physiol. 174, 310 –321 Forsten, K. E., Courant, N. A., and Nugent, M. A. (1997) Endothelial proteoglycans inhibit bFGF binding and mitogenesis. J. Cell. Physiol. 172, 209 –220 Higuchi, R. (1990) in PCR Protocols: A guide to Methods and Applications (Innis, M. A., Gelfand, D. H., Sninsky, J. J., and White, T. J., eds) Academic Press, San Diego, California Thompson, L. D., Pantoliano, M. W., and Springer, B. A. (1994). Energetic characterization of the basic fibroblast growth factorheparin interaction: identification of the heparin binding domain. Biochemistry 33, 3831–3840 Li, L.-Y., Safran, M., Aveizer, D., Bo¨hlen, P., and Yayon, A. (1994). Diminished heparin binding of a basic fibroblast growth factor mutant is associated with reduced receptor binding, mitogenesis, plasminogen activator induction, and in vitro angiogenesis. Biochemistry 34, 731–736 Springer, B. A., Pantoliano, M. W., Barbera, F. A., Gunyuzlu, P. L., Thompson, L. D., Herblin, W. F., Rosenfeld, S. A., and Book, G. W. (1994). Identification and concerted function of two receptor binding surfaces on basic fibroblast growth factor required for mitogenesis. J. Biol. Chem. 269, 26879 –26884 Faham, S., Hileman, R. W., Fromm, J. R., Linhardt, R. J., and Rees, D. C. (1996). Heparin structure and interactions with basic fibroblast growth factor. Science 271, 1116 –1120
The FASEB Journal
PADERA ET AL.
32.
Nugent, M. A., and Edelman, E. R. (1992). Kinetics of basic fibroblast growth factor binding to its receptor and heparan sulfate proteoglycan: a mechanism for cooperativity. Biochemistry 31, 8876 – 8883 33. Sperinde, G. V., and Nugent, M. A. (1998). Heparan sulfate proteoglycans control intracellular processing of bFGF in vascular smooth muscle cells. Biochemistry 37, 13153–13164 34. Larrain J., Alvarez J., Hassell J. R., and Brandan E. (1997). Expression of perlecan, a proteoglycan that binds myogenic
HLGAG MODULATION OF FGF-2 SIGNALING
inhibitory basic fibroblast growth factor, is down regulated during skeletal muscle differentiation. Exp. Cell Res. 234, 405– 412 35. Bikfalvi, A., Savona, C., Perollet, C., and Javerzat, S. (1997). New insights in the biology of fibroblast growth factor-2. Angiogenesis 1, 155–173 Received for publication December 10, 1998. Accepted for publication May 14, 1999.
1687