The crystal structure of fibroblast growth factor 18 (FGF18)

Dear Editor, Fibroblast growth factors (FGFs) regulate a plethora of critical processes in development (Beenken and Mohammadi, 2009). These processes are mediated by signaling through four FGF receptors (FGFR1–4), which are high-affinity cell surface receptor tyrosine kinases. Receptors 1–3 undergo alternative splicing to generate band c-isoforms with altered ligand specificities and affinities. Ligand to receptor binding is not unique and one receptor can be activated by several FGFs. Heparan sulfate (HS) proteoglycan, a variably poly-sulfated glycosaminoglycan related to heparin, is an essential requirement for FGF signaling with FGFs varying in their specificities for different HS sulfation patterns (Galla-gher et al., 1992). The 18 mammalian FGFs that are capable of signaling through FGFRs share a conserved β-trefoil fold and are grouped into 6 subfamilies (Itoh and Ornitz, 2004). FGF18 belongs to the paracrine-acting FGF8 subfamily, which contains three members in humans: FGF8, FGF17, and FGF18. The biological activities of FGF8 and FGF17 are regulated by alternative splicing with four isoforms of FGF8 (a, b, e, and f) and two of FGF17. In contrast, FGF18 does not undergo alternative splicing. FGF18 has a number of functions in the developing and adult organism including a key role in skeletal development (Haque et al., 2007) and has undergone clinical trials for the treatment of osteoarthritis (Merck KGaA). Pathologically, FGF18 is implicated in colorectal (Shimokawa et al., 2003) and ovarian cancer (Wei et al., 2013) and has been proposed as an early marker and potential anticancer drug target. In vivo, FGF18 is expressed as a precursor polypeptide of 207 residues with the initial 27 hydrophobic residues forming a cleavable signal peptide. Residues 50–194 were expressed in Escherichia coli and purified to homogeneity. The interaction between FGF18 and heparin oligosaccharides was examined using isothermal titration calorimetry (ITC), Fig. S1A. FGF18 binds heparin oligosaccharides of 6 dp with an affinity of 2.2 ± 29.2 μmol/L. The interaction is enthalpy-driven (ΔH = −8.6 kcal·mol −1) with a negative entropic contribution (TΔS = −1.0 kcal·mol −1) giving an overall free energy (ΔG) of −7.6 kcal·mol −1. A heparin oligosaccharide of 8 dp was sufficient to dimerize FGF18 (Fig. S1B), as observed for FGF1 (Brown et al., 2013). FGF18 was incubated stoichiometrically with heparin (6 dp) and the complex isolated prior to crystallization in a mother liquor of 0.1 mol/L MES pH 6.5, 0.2 mol/L ammonium sulphate and 26% PEG 5000. FGF18 crystallized in the primitive monoclinic spacegroup P2 1 …


Isothermal Titration Calorimetry (ITC)
ITC was performed using a MicroCal VP-ITC machine (GE Healthcare). Titrations involved the addition of 10 µl aliquots of heparin oligosaccharides via a rotating stirrer-syringe to the calorimetric cell containing 1.4 ml of FGF18 at 4 min intervals for a total of 20 injections. A constant temperature of 25°C and stirring speed of 300 rpm was maintained throughout. Heats of dilution determined in the absence of FGF18 were subtracted from the titration data before curve fitting. Additionally a 4 µl injection was discarded from each dataset to remove the effect of titrant diffusion across the syringe tip during the equilibration process. Data were modelled by nonlinear least squares fitting and a full set of thermodynamic parameters derived using ΔG = ΔH -TΔS = -RTln(1/K D ), where ΔG, ΔH and ΔS are the Gibbs free energy, enthalpy and entropy of binding respectively. T is the absolute temperature, R = 1.98 cal mol -1 K -1 is the ideal gas law constant and K D is the dissociation constant.

Crystallization of FGF18
FGF18 was mixed in a 1:1 stoichiometry with heparin hexasaccharide and the complex purified using size-exclusion chromatography on a Superdex 200 10/30 column (GE Healthcare) equilibrated with 20 mM Tris pH 8.0, 150 mM NaCl.
The crystals were cryoprotected by the addition of 26 % (v/v) ethylene glycol (soaked for ~ 1 min) and vitrified by plunging in liquid nitrogen.

Structure determination and refinement
X-ray diffraction data were collected on a X8 PROTEUM (Bruker AXS) at a wavelength 1.54 Å. Data were indexed and scaled using PROTEUM 2 software (Bruker AXS) and structure factor amplitudes calculated from the merged intensities.
To provide phases, a molecular replacement search was performed using PHASER (McCoy et al., 2007) with the known crystal structure of FGF8b as a search model (chain M from PDB ID: 2FDB). Maximum likelihood-based restrained refinement was carried out using REFMAC (Murshudov et al., 2011) with manual rebuilding in Coot (Emsley and Cowtan, 2004). The refinement strategy was optimized using PDB_REDO (Joosten et al., 2011). The stereochemistry of the structure was assessed and validated with MolProbity (Chen et al., 2009). The final refinement statistics are shown in Table 1. The structure has been deposited in the PDB with PDB ID: 4JCM).

Structure comparison
The structures of other FGFs were obtained from the Protein Data Bank (Rose et al., 2011). FGF8b-FGFR2c interfaces were analyzed using PISA (Krissinel and Henrick, 2007) with interface 'hot spot' residues identified using HSPred (Lise et al., 2011).
Structural superpositions were performed using SuperPose (Maiti et al., 2004). The PDB2PQR server and the Adaptive Poisson-Boltzmann Solver (Unni et al., 2011) were used to calculate electrostatic potential. All images were prepared using PyMOL (DeLano, 2002) with secondary structure assigned with STRIDE (Heinig and Frishman, 2004).   (Sherry et al., 2001) and mutations identified in cancer samples from COSMIC (Forbes et al., 2011). FGF18 forms a 1:1 interaction with heparin oligosaccharides of 6 dp. (B) FGF18 dimerizes in the presence of heparin oligosaccharides of 8 dp.  structure alignment was performed using PROMALS3D (Pei et al., 2008) and JOY was used to annotate the 3D structural information (Mizuguchi et al., 1998) (Kyte and Doolittle, 1982). The hydrophobic β4-β5 loop of FGF18 (shown in stick representation) engages with the hydrophobic groove of the FGFR c-isoforms.
(D) In FGFR b-isoforms, the β4-β5 loop of FGF18 would engage a more hydrophilic region, explaining the preference of FGF18 for c-isoforms. In all panels FGF18 is shown in blue cartoon representation.