Differential specificity of endocrine FGF19 and FGF21 to FGFR1 and FGFR4 in complex with KLB

Background

Recent studies suggest that betaKlotho (KLB) and endocrine FGF19 and FGF21 redirect FGFR signaling to regulation of metabolic homeostasis and suppression of obesity and diabetes. However, the identity of the predominant metabolic tissue in which a major FGFR-KLB resides that critically mediates the differential actions and metabolism effects of FGF19 and FGF21 remain unclear.

Methodology/Principal Findings

We determined the receptor and tissue specificity of FGF21 in comparison to FGF19 by using direct, sensitive and quantitative binding kinetics, and downstream signal transduction and expression of early response gene upon administration of FGF19 and FGF21 in mice. We found that FGF21 binds FGFR1 with much higher affinity than FGFR4 in presence of KLB; while FGF19 binds both FGFR1 and FGFR4 in presence of KLB with comparable affinity. The interaction of FGF21 with FGFR4-KLB is very weak even at high concentration and could be negligible at physiological concentration. Both FGF19 and FGF21 but not FGF1 exhibit binding affinity to KLB. The binding of FGF1 is dependent on where FGFRs are present. Both FGF19 and FGF21 are unable to displace the FGF1 binding, and conversely FGF1 cannot displace FGF19 and FGF21 binding. These results indicate that KLB is an indispensable mediator for the binding of FGF19 and FGF21 to FGFRs that is not required for FGF1. Although FGF19 can predominantly activate the responses of the liver and to a less extent the adipose tissue, FGF21 can do so significantly only in the adipose tissue and adipocytes. Among several metabolic and endocrine tissues, the response of adipose tissue to FGF21 is predominant, and can be blunted by the ablation of KLB or FGFR1.

Conclusions

Our results indicate that unlike FGF19, FGF21 is unable to bind FGFR4-KLB complex with affinity comparable to FGFR1-KLB, and therefore, at physiological concentration less likely to directly and significantly target the liver where FGFR4-KLB predominantly resides. However, both FGF21 and FGF19 have the potential to activate responses of primarily the adipose tissue where FGFR1-KLB resides.     

The F-BAR protein NOSTRIN participates in FGF signal transduction and vascular development

F-BAR proteins are multivalent adaptors that link plasma membrane and cytoskeleton and coordinate cellular processes such as membrane protrusion and migration. Yet, little is known about the function of F-BAR proteins in vivo. Here we report, that the F-BAR protein NOSTRIN is necessary for proper vascular development in zebrafish and postnatal retinal angiogenesis in mice. The loss of NOSTRIN impacts on the migration of endothelial tip cells and leads to a reduction of tip cell filopodia number and length. NOSTRIN forms a complex with the GTPase Rac1 and its exchange factor Sos1 and overexpression of NOSTRIN in cells induces Rac1 activation. Furthermore, NOSTRIN is required for fibroblast growth factor 2 dependent activation of Rac1 in primary endothelial cells and the angiogenic response to fibroblast growth factor 2 in the in vivo matrigel plug assay. We propose a novel regulatory circuit, in which NOSTRIN assembles a signalling complex containing FGFR1, Rac1 and Sos1 thereby facilitating the activation of Rac1 in endothelial cells during developmental angiogenesis.     

Testing effects of signal transduction pathways on cadherin junctional complex assembly using quantitative image analysis

Cadherin adhesion molecules function in numerous cell biological processes that influence embryo development, normal cell physiology, and pathophysiology of many disease processes. Cadherins nucleate the assembly of the adherens junction, a cell-to-cell adhesion plaque that is prominent in simple epithelial cells and found in many cell types. Numerous cell biological approaches have been used to study this interesting class of molecules. Here, we outline methodology used in our studies of junctional complexes to examine effects of signaling molecules on assembly mechanisms. This is a quantitative method that allows the investigator to test the combined effect of two different signaling processes to determine whether these two signals act in concert within the same pathway. We discuss how this method could be generalized to other studies to examine consequences of various experimental manipulations on the assembly of cellular structures.     

Structural specificity in a FGF7-affinity purified heparin octasaccharide required for formation of a complex with FGF7 and FGFR2IIIb

Variations in sulfation of heparan sulfate (HS) affect interaction with FGF, FGFR, and FGF-HS-FGFR signaling complexes. Whether structurally distinct HS motifs are at play is unclear. Here we used stabilized recombinant FGF7 as a bioaffinity matrix to purify size-defined heparin oligosaccharides. We show that only 0.2%-4% of 6 to 14 unit oligosaccharides, respectively, have high affinity for FGF7 based on resistance to salt above 0.6M NaCl. The high affinity fractions exhibit highest specific activity for interaction with FGFR2IIIb and formation of complexes of FGF7-HS-FGFR2IIIb. The majority fractions with moderate (0.30-0.6M NaCl), low (0.14-0.30M NaCl) or no affinity at 0.14M NaCl for FGF7 supported no complex formation. The high affinity octasaccharide mixture exhibited predominantly 7- and 8-sulfated components (7,8-S-OctaF7) and formed FGF7-HS-FGFR2IIIb complexes with highest specific activity. Deduced disaccharide analysis indicated that 7,8-S-OctaF7 comprised of DeltaHexA2SGlcN6S in a 2:1 ratio to a trisulfated and a variable unsulfated or monosulfated disaccharide. The inactive octasaccharides with moderate affinity for FGF7 were much more heterogenous and highly sulfated with major components containing 11 or 12 sulfates comprised of predominantly trisulfated disaccharides. This suggests that a rare undersulfated motif in which sulfate groups are specifically distributed has highest affinity for FGF7. The same motif also exhibits structural requirements for high affinity binding to dimers of FGFR2IIIb prior to binding FGF7 to form FGF7-HS-FGFR2IIIb complexes. In contrast, the majority of more highly sulfated HS motifs likely play FGFR-independent roles in stability and control of access of FGF7 to FGFR2IIIb in the tissue matrix.     

Multivalent protein-carbohydrate interactions. A new paradigm for supermolecular assembly and signal transduction

Many biological recognition processes involve the binding and clustering of ligand-receptor complexes and concomitant signal transduction events. Such interactions have recently been observed in human T cells in which binding and cross-linking of specific glycoprotein counter-receptors on the surface of the cells by an endogenous bivalent carbohydrate binding protein (galectin-1) leads to apoptosis [Pace, K. E., et al. (1999) J. Immunol. 163, 3801-3811]. Importantly, different counter-receptors associated with specific phosphatase or kinase activities were shown to form separate clusters on the surface of the cells as a result of galectin-1 binding to the carbohydrate moieties of the respective glycoproteins. This suggests that the unique separation and organization of signaling molecules that results from galectin-1 binding is involved in delivering the signal to die. The ability of galectin-1 to induce the separation of specific glycoprotein receptors was modeled on the basis of molecular and structural studies of the binding of multivalent carbohydrates to lectins that result in the formation of specific two- and three-dimensional cross-linked lattices. These latter studies have been recently highlighted by X-ray crystallographic results showing that a single tetravalent lectin forms distinct cross-linked complexes with four different bivalent oligosaccharides [Olsen, L. R., et al. (1997) Biochemistry 36, 15073-15080]. In this report, binding and cross-linking of multivalent carbohydrates with multivalent lectins is shown to be a new paradigm for supermolecular assembly and signal transduction in biological systems.     

FGF and FGFR signaling in chondrodysplasias and craniosynostosis

The first experimental mouse model for FGF2 in bone dysplasia was made serendipitously by overexpression of FGF from a constitutive promoter. The results were not widely accepted, rightfully drew skepticism, and were difficult to publish; because of over 2,000 studies published on FGF‐2 at the time (1993), only a few reported a role of FGF‐2 in bone growth and differentiation. However, mapping of human dwarfisms to mutations of the FGFRs shortly, thereafter, made the case that bone growth and remodeling was a major physiological function for FGF. Subsequent production of numerous transgenic and targeted null mice for several genes in the bone growth and remodeling pathways have marvelously elucidated the role of FGFs and their interactions with other genes. Indeed, studies of the FGF pathway present one of the best success stories for use of experimental genetics in functionally parsing morphogenetic regulatory pathways. What remains largely unresolved is the pleiotropic nature of FGF‐2. How does it accelerate growth in one cell then stimulate apoptosis or retard growth for another cell in the same type of tissue? Some of the answers may come through distinguishing the FGF‐2 protein isoforms, made from alternative translation start sites, these appear to have substantially different functions. Although we have made substantial progress, there is still much to be learned regarding FGF‐2 as a most complex, enigmatic protein. Studies of genetic models in mice and human FGFR mutations have provided strong evidence that FGFRs are important modulators of osteoblast function during membranous bone formation. However, there is some controversy regarding the effects of FGFR signaling in human and murine genetic models. Although significant progress has been made in our understanding of FGFR signaling, several questions remain concerning the signaling pathways involved in osteoblast regulation by activated FGFR. Additionally, little is known about the specific role of FGFR target genes involved in cranial bone formation. These issues need to be addressed in future in in vitro and in vivo approaches to better understand the molecular mechanisms of action of FGFR signaling in osteoblasts that result in anabolic effects in bone formation. J. Cell. Biochem. © 2005 Wiley‐Liss, Inc.     

Analysis of mutations in fibroblast growth factor (FGF) and a pathogenic mutation in FGF receptor (FGFR) provides direct evidence for the symmetric two-end model for FGFR dimerization

Two competing models for fibroblast growth factor (FGF) receptor (FGFR) dimerization have recently emerged based on ternary FGF-FGFR-heparin crystal structures. In the symmetric two-end model, heparin promotes dimerization of two FGF-FGFR complexes by stabilizing bivalent interactions of the ligand and receptor through primary and secondary sites and by stabilizing direct receptor-receptor contacts. In the asymmetric model, there are no protein-protein contacts between the two FGF-FGFR complexes, which are bridged solely by heparin. To identify the correct mode of FGFR dimerization, we abolished interactions at the secondary ligand-receptor interaction site, which are observed only in the symmetric two-end model, using site-directed mutagenesis. Cellular studies and real-time binding assays, as well as matrix-assisted laser desorption ionization-time of flight analysis, demonstrate that loss of secondary ligand-receptor interactions results in diminished FGFR activation due to decreased dimerization without affecting FGF-FGFR binding. Additionally, structural and biochemical analysis of an activating FGFR2 mutation resulting in Pfeiffer syndrome confirms the physiological significance of receptor-receptor contacts in the symmetric two-end model and provides a novel mechanism for FGFR gain of function in human skeletal disorders. Taken together, the data validate the symmetric two-end model of FGFR dimerization and argue against the asymmetric model of FGFR dimerization.     

Real-time analysis of ternary complex on particles: direct evidence for partial agonism at the agonist-receptor-G protein complex assembly step of signal transduction

We developed a novel and generalized approach to investigate G protein-coupled receptor molecular assemblies. We solubilized a fusion protein consisting of the beta(2)-adrenergic receptor and green fluorescent protein (GFP) for bead-based flow cytometric analysis. beta(2)-Adrenergic receptor GFP bound to dihydroalprenolol-conjugated beads, providing a K(d) for the fusion protein and, in competition with beta(2)-adrenergic receptor ligands, K(d) values for agonists and antagonists. Beads displaying chelated nickel bound purified hexahistidine-tagged G protein heterotrimers and, subsequently, the binary complex of agonist with beta(2)-adrenergic receptor GFP. The dose-response curves of ternary complex formation revealed maximal assembly for ligands previously classified as full agonists and reduced assembly for ligands previously classified as partial agonists. Guanosine 5'-3-O-(thio)triphosphate-induced dissociation rates of the ternary complex were the same for full and partial agonists. Soluble G protein, competing with ternary complexes on beads provided an affinity estimate of agonist-receptor complexes to G protein. When performed simultaneously, the two assemblies discriminated between agonist, antagonist or inactive molecule in a manner appropriate for high throughput, small volume drug discovery. The assemblies can be further generalized to other G protein coupled receptor protein-protein interactions.     

betaKlotho is required for fibroblast growth factor (FGF) 21 signaling through FGF receptor (FGFR) 1c and FGFR3c

Fibroblast growth factor (FGF) 21, a structural relative of FGF23 that regulates phosphate homeostasis, is a regulator of insulin-independent glucose transport in adipocytes and plays a role in the regulation of body weight. It also regulates ketogenesis and adaptive responses to starvation. We report that in a reconstituted receptor activation assay system using BaF3 cells, which do not endogenously express any type of FGF receptor (FGFR) or heparan sulfate proteoglycan, FGF21 alone does not activate FGFRs and that betaKlotho is required for FGF21 to activate two specific FGFR subtypes: FGFR1c and FGFR3c. Coexpression of betaKlotho and FGFR1c on BaF3 cells enabled FGF21, but not FGF23, to activate receptor signaling. Conversely, coexpression of FGFR1c and Klotho, a protein related to betaKlotho, enabled FGF23 but not FGF21 to activate receptor signaling, indicating that expression of betaKlotho/Klotho confers target cell specificity on FGF21/FGF23. In all of these cases, heparin enhanced the activation but was not essential. In 3T3-L1 adipocytes, up-regulation of glucose transporter (GLUT) expression by FGF21 was associated with expression of betaKlotho, which was absent in undifferentiated 3T3-L1 fibroblasts. It is thus suggested that betaKlotho expression is a crucial determinant of the FGF21 specificity of the target cells upon which it acts in an endocrine fashion.     

Bistability from double phosphorylation in signal transduction. Kinetic and structural requirements

Previous studies have suggested that positive feedback loops and ultrasensitivity are prerequisites for bistability in covalent modification cascades. However, it was recently shown that bistability and hysteresis can also arise solely from multisite phosphorylation. Here we analytically demonstrate that double phosphorylation of a protein (or other covalent modification) generates bistability only if: (a) the two phosphorylation (or the two dephosphorylation) reactions are catalyzed by the same enzyme; (b) the kinetics operate at least partly in the zero-order region; and (c) the ratio of the catalytic constants of the phosphorylation and dephosphorylation steps in the first modification cycle is less than this ratio in the second cycle. We also show that multisite phosphorylation enlarges the region of kinetic parameter values in which bistability appears, but does not generate multistability. In addition, we conclude that a cascade of phosphorylation/dephosphorylation cycles generates multiple steady states in the absence of feedback or feedforward loops. Our results show that bistable behavior in covalent modification cascades relies not only on the structure and regulatory pattern of feedback/feedforward loops, but also on the kinetic characteristics of their component proteins.     

Twist plays an essential role in FGF and SHH signal transduction during mouse limb development

Loss of Twist gene function arrests the growth of the limb bud shortly after its formation. In the Twist(-/-) forelimb bud, Fgf10 expression is reduced, Fgf4 is not expressed, and the domain of Fgf8 and Fgfr2 expression is altered. This is accompanied by disruption of the expression of genes (Shh, Gli1, Gli2, Gli3, and Ptch) associated with SHH signalling in the limb bud mesenchyme, the down-regulation of Bmp4 in the apical ectoderm, the absence of Alx3, Alx4, Pax1, and Pax3 activity in the mesenchyme, and a reduced potency of the limb bud tissues to differentiate into osteogenic and myogenic tissues. Development of the hindlimb buds in Twist(-/-) embryos is also retarded. The overall activity of genes involved in SHH signalling is reduced.Fgf4 and Fgf8 expression is lost or reduced in the apical ectoderm, but other genes (Fgf10, Fgfr2) involved with FGF signalling are expressed in normal patterns. Twist(+/-);Gli3(+/XtJ) mice display more severe polydactyly than that seen in either Twist(+/-) or Gli3(+/XtJ) mice, suggesting that there is genetic interaction between Twist and Gli3 activity. Twist activity is therefore essential for the growth and differentiation of the limb bud tissues as well as regulation of tissue patterning via the modulation of SHH and FGF signal transduction.     

A new model for chemotactic signal transduction in Dictyostelium discoideum

Dictyostelium discoideum amoebae were employed to study the refractoriness and adaptation of the rapid (5sec) accumulation of actin in their Triton-insoluble cytoskeletons following stimulation with specific chemoattractants. Amoebae became refractory within 10sec for this response but no adaptation occurred during this period. Amoebae desensitized for one attractant were not desensitized for another and responses to stimulation with a mixture of attractants were approximately additive. The characteristics of these processes are compared to published studies of adaptation in other chemoattractant-induced responses and a new model for the chemotactic signal transduction pathway is formulated. We conclude that intracellular cGMP accumulation may be on a separate branch of the pathway from the actin response.     

Cancer,oncogenes and signal transduction

A report on the European Molecular Biology Laboratory (EMBL) 'Oncogenes and Growth Control' meeting, Heidelberg, Germany, 17-20 April 2004.     

FGF signaling through FGFR1 is required for olfactory bulb morphogenesis

During development, the embryonic telencephalon is patterned into different areas that give rise to distinct adult brain structures. Several secreted signaling molecules are expressed at putative signaling centers in the early telencephalon. In particular, Fgf8 is expressed at the anterior end of the telencephalon and is hypothesized to pattern it along the anteroposterior (AP) axis. Using a CRE/loxP genetic approach to disrupt genes in the telencephalon, we address the role of FGF signaling directly in vivo by abolishing expression of the FGF receptor Fgfr1. In the Fgfr1-deficient telencephalon, AP patterning is largely normal. However, morphological defects are observed at the anterior end of the telencephalon. Most notably, the olfactory bulbs do not form normally. Examination of the proliferation state of anterior telencephalic cells supports a model for olfactory bulb formation in which an FGF-dependent decrease in proliferation is required for initial bulb evagination. Together the results demonstrate an essential role for Fgfr1 in patterning and morphogenesis of the telencephalon.     

Differential signal transduction of alternatively spliced FGFR2 variants expressed in human mammary epithelial cells

Gene amplification and protein overexpression of fibroblast growth factor receptor 2 (FGFR2) characterize the SUM-52 breast cancer cell line developed in our laboratory. SUM-52 cells express nine distinct alternatively spliced isoforms of FGFR2. Among these isoforms are two otherwise identical FGFR2 variants that express either the C1 or C3 carboxyl terminus. FGFR2-C3 variants are not normally expressed by human mammary epithelial (HME) cells, and we have shown that overexpression of FGFR2-C3 in HME cells results in potent transformation. In particular, FGFR2-C3 expression leads to robust levels of constitutively tyrosine phosphorylated FRS2 in the absence of ligand stimulation. In contrast, overexpressed FGFR2-C1 requires constant stimulation with exogenous keratinocyte growth factor (KGF) to mimic the signaling capability of FGFR2-C3. However, activation of FRS2 that results from KGF-stimulated FGFR2-C1 signaling is transient and is associated with a mobility shift of FRS2 not observed when this signaling molecule is activated by the C3 isoform of FGFR2. Mutation of the only tyrosine phosphorylated site present in the C1 terminus and absent from C3, Tyr769, did not yield a receptor that rivaled the potent signaling of FGFR2-C3. We therefore conclude that aberrant expression of alternatively spliced isoforms of FGFR2 with the C3 carboxyl terminus in the SUM-52 breast cancer cells results in sustained activation of signal transduction leading to transformation.     

Kinetic model of BiP- and PDI-mediated protein folding and assembly

A mechanism for heavy chain binding protein (BiP)- and protein disulfide isomerase (PDI)- mediated protein folding and assembly has been proposed. It considers BiP chaperoning action and PDI catalytic activity. A kinetic model has been developed based on the proposed mechanism. The model was used for quantifying the influence of polypeptide concentration and ratio, and the effect of BiP and PDI concentration on the kinetics of folding and assembly. An optimum value for polypeptide concentration that minimizes assembly times was found, and different kinetic behaviors were identified for polypeptide concentrations higher or lower than the optimum. Pulse-chase experiments and the dependence of assembly time on unassembled polypeptides ratio predicted by the model are similar to those found during in vitro and in vivo folding and assembly of antibodies and human chorionic gonadotropin (hCG), as well as bovine pancreatic trypsin inhibitor (BPTI) folding. The model also explains the increase in folding and assembly rates during overexpression of BiP and PDI.     

Epistasis in a model of molecular signal transduction

Biological functions typically involve complex interacting molecular networks, with numerous feedback and regulation loops. How the properties of the system are affected when one, or several of its parts are modified is a question of fundamental interest, with numerous implications for the way we study and understand biological processes and treat diseases. This question can be rephrased in terms of relating genotypes to phenotypes: to what extent does the effect of a genetic variation at one locus depend on genetic variation at all other loci? Systematic quantitative measurements of epistasis – the deviation from additivity in the effect of alleles at different loci – on a given quantitative trait remain a major challenge. Here, we take a complementary approach of studying theoretically the effect of varying multiple parameters in a validated model of molecular signal transduction. To connect with the genotype/phenotype mapping we interpret parameters of the model as different loci with discrete choices of these parameters as alleles, which allows us to systematically examine the dependence of the signaling output – a quantitative trait – on the set of possible allelic combinations. We show quite generally that quantitative traits behave approximately additively (weak epistasis) when alleles correspond to small changes of parameters; epistasis appears as a result of large differences between alleles. When epistasis is relatively strong, it is concentrated in a sparse subset of loci and in low order (e.g. pair-wise) interactions. We find that focusing on interaction between loci that exhibit strong additive effects is an efficient way of identifying most of the epistasis. Our model study defines a theoretical framework for interpretation of experimental data and provides statistical predictions for the structure of genetic interaction expected for moderately complex biological circuits.