The structural biology of the FGF19 subfamily

Abstract

The ability of the Fibroblast Growth Factor (FGF) 19 subfamily to signal in an endocrine fashion sets this subfamily apart from the remaining five FGF subfamilies known for their paracrine functions during embryonic development. Compared to the members of paracrine FGF subfamiles, the three members of the FGF19 subfamily, namely FGF19, FGF21 and FGF23, have poor affinity for heparan sulfate (HS) and therefore can diffuse freely in the HS-rich extracellular matrix to enter into the bloodstream. In further contrast to paracrine FGFs, FGF19 subfamily members have unusually poor affinity for their cognate FGF receptors (FGFRs) and therefore cannot bind and activate them in a solely HS-dependent fashion. As a result, the FGF19 subfamily requires α/βklotho coreceptor proteins in order to bind, dimerize and activate their cognate FGFRs. This klotho-dependency also determines the tissue specificity of endocrine FGFs. Recent structural and biochemical studies have begun to shed light onto the molecular basis for the klotho-dependent endocrine mode of action of the FGF19 subfamily. Crystal structures of FGF19 and FGF23 show that the topology of the HS binding site (HBS) of FGF19 subfamily members deviates drastically from the common topology adopted by the paracrine FGFs. The distinct topologies of the HBS of FGF19 and FGF23 prevent HS from direct hydrogen bonding with the backbone atoms of the HBS of these ligands and accordingly decrease the HS binding affinity of this subfamily. Recent biochemical data reveal that the ?klotho ectodomain binds avidly to the ectodomain of FGFR1c, the main cognate FGFR of FGF23, creating a de novo high affinity binding site for the C-terminal tail of FGF23. The isolated FGF23 C-terminus can be used to effectively inhibit the formation of the FGF23-FGFR1c-αklotho complex and alleviate hypophosphatemia in renal phosphate disorders due to elevated levels of FGF23.

Figure 1

An overview of FGF and FGFR structural biology. A, viewed on previous page) FGF1 is represented as a cartoon. N- and C-termini are labeled and the twelve β-sheets that compose the β-trefoil core are indicated. B, viewed on previous page) A schematic of FGFR shows its three Ig-like domains. D1 and the acid box (AB) are involved in autoinhibition, the heparan sulfate binding site (HBS) is located on D2 and alternative splicing takes place in the latter half of D3. D2 and D3 are necessary and sufficient for ligand binding. An intracellular kinase domain mediates downstream signaling. C) The formation of a 2:2:2 FGF:FGFR:HS dimer on the cell surface leads to intracellular transphosphorylation of the FGFR kinase domains and downstream signaling through PLCγ, FRS2α, and CRKL.

Figure 1

An overview of FGF and FGFR structural biology. A, viewed on previous page) FGF1 is represented as a cartoon. N- and C-termini are labeled and the twelve β-sheets that compose the β-trefoil core are indicated. B, viewed on previous page) A schematic of FGFR shows its three Ig-like domains. D1 and the acid box (AB) are involved in autoinhibition, the heparan sulfate binding site (HBS) is located on D2 and alternative splicing takes place in the latter half of D3. D2 and D3 are necessary and sufficient for ligand binding. An intracellular kinase domain mediates downstream signaling. C) The formation of a 2:2:2 FGF:FGFR:HS dimer on the cell surface leads to intracellular transphosphorylation of the FGFR kinase domains and downstream signaling through PLCγ, FRS2α, and CRKL.

Figure 2

HS-dependent dimerization of the FGF-FGFR complex A) A surface representation of the FGF2-FGFR1c-heparin ternary complex, PDB ID: 1FQ9. FGF2 is in dark grey, the D2 and D3 domains of FGFR1c are in light grey and heparin is represented as sticks in black. B) The complex has been rotated ninety degrees around an axis parallel to the plane of the page to reveal heparin binding the HBS in the membrane distal cleft of the complex. C) The boxed region from Figure 2A is expanded to show the FGF2 residues involved in hydrogen bonding to HS. There a total of 16 hydrogen bonds between FGF2 and HS. For the sake of clarity, only those three hydrogen bonds mediated by backbone atoms are shown.

Figure 3

Sequence alignments of FGFs. A, viewed on previous page. Sequences of human FGF19, FGF23, FGF21, FGF2, FGF4 and FGF10 are aligned. N- and C-terminal regions of some ligands are truncated for the sake of presentation. β-sheets are highlighted with grey in the alignment and the helical secondary structures in FGF19 and FGF23 are highlighted with dark grey. Numerous important residues are enclosed in boxes to emphasize their importance: the Cys-58 and Cys-70 residues that support FGF19’s unusual β10–β12 segment structure are boxed, along with other cysteines that form disulfide bridges; the GXXXXGXX(T/S) motif in the paracrine FGFs is indicated with boxes; the RXXR motif in FGF23 is also boxed. B) Orthologs of FGF19, FGF21, FGF23 and FGF2 are aligned and the β10–β12 segment is shown for the solved structures. FGF23 exhibits greater conservation in its HBS than do either FGF19 or FGF21. The location of the GXXXXGXX(T/S) motif is indicated with arrowheads above the alignment of FGF2 orthologs.

Figure 3

Sequence alignments of FGFs. A, viewed on previous page. Sequences of human FGF19, FGF23, FGF21, FGF2, FGF4 and FGF10 are aligned. N- and C-terminal regions of some ligands are truncated for the sake of presentation. β-sheets are highlighted with grey in the alignment and the helical secondary structures in FGF19 and FGF23 are highlighted with dark grey. Numerous important residues are enclosed in boxes to emphasize their importance: the Cys-58 and Cys-70 residues that support FGF19’s unusual β10–β12 segment structure are boxed, along with other cysteines that form disulfide bridges; the GXXXXGXX(T/S) motif in the paracrine FGFs is indicated with boxes; the RXXR motif in FGF23 is also boxed. B) Orthologs of FGF19, FGF21, FGF23 and FGF2 are aligned and the β10–β12 segment is shown for the solved structures. FGF23 exhibits greater conservation in its HBS than do either FGF19 or FGF21. The location of the GXXXXGXX(T/S) motif is indicated with arrowheads above the alignment of FGF2 orthologs.

Figure 4

Paracrine FGFs mediate an epithelial-mesenchymal signaling loop. Ligands expressed in the epithelium signal through receptors expressed in the mesenchyme and vice versa.

Figure 5

The FGF19 structure A) The FGF19 structure, PDB ID: 2P23, is shown as a surface and as a cartoon. The β-sheets and α-helix are labeled along with the N- and C-termini of the protein. B) The HBS of FGF19 is superimposed onto that of FGF2. Both ligands are represented as ribbons, with FGF19 in dark grey and FGF2 in light grey. The atoms of the GXXXXGXX(T/S) motif are marked with dots in the FGF2 structure and identified. Arrows indicate the positions of the Leu-145 and Leu-162 where the FGF19 Cα trace diverges from the FGF2 Cα trace. C) The FGF19 HBS is shown and the C58-C70 disulfide bridge residues are represented as sticks along with the Leu-162 and Leu-145 residues that protect them from solvent. D) The FGF2 HBS is shown, the GXXXXGXX(T/S) motif is identified, and intramolecular contacts between the β1–β2 loop and β10–β12 segment are shown. Asn-36 of the β1–β2 loop interacts with the backbone atom of Arg-129 in the β10–β12 segment and Phe-39 of the β1–β2 region engages in Van der Waals interactions with the backbone of the β10–β12 region.

Figure 6

The FGF23 structure A) The FGF23 structure, PDB ID: 2P39, is shown as a surface and as a cartoon. The β-sheets and α-helix are labeled along with the N- and C-termini of the protein. The SOS molecule that is bound to FGF23 is represented as sticks in dark grey. B) An expansion of the box in Figure 6A showing a close-up of the SOS molecule and the residues in the HBS of FGF23 with which it binds. C) An overlay of the HBS of FGF23 with that of FGF2 showing their divergence. FGF2 is in light grey and FGF23 is in dark grey. The residues of the GXXXXGXX(T/S) motif are labeled and the Leu-138 and Pro-153 residues where the FGF23 Cα trace diverges from that of FGF2 are indicated with arrows. D) The FGF19 and FGF23 HBS regions are superimposed, with FGF19 shown in light grey and FGF23 in dark grey.

Figure 7

Superimposition of heparin onto the FGF19 and FGF23 structures. A) FGF19 is superimposed onto the FGF2-FGFR1c-heparin structure (PDB ID: 1FQ9) and the heparin from that structure is grafted onto FGF 19 to show the clashes that would occur were heparin to attempt to bind FGF 19 in the same fashion as it binds FGF2. B) Similarly for FGF23, heparin is superimposed to reveal the clashes that would occur were heparin to attempt to bind in a FGF2-like fashion.

Figure 8

A proposed model of the FGF23-FGFR1c-heparin-αklotho complex. A) FGF23 is superimposed onto FGF2 in the FGF2-FGFR1c-heparin complex (PDB ID:1FQ9) and is depicted in dark grey. The FGFRs from 1FQ9 are depicted in a medium tint of grey. Myrosinase (PDB ID: 1E6S), a glycosidase, is used as a model for the two KL domains of αklotho and is depicted in light grey. B) The FGF23-FGFR1c-heparin-αklotho complex has been rotated 90 degrees to show the HS binding cleft in the membrane-distal portion of the complex.