NovelMechanismsofFibroblastGrowthFactorReceptor1 ... · 2009-10-10 · unknown. Here, we show that...

Novel Mechanisms of Fibroblast Growth Factor Receptor 1Regulation by Extracellular Matrix Protein Anosmin-1*□S

Received for publication, July 27, 2009 Published, JBC Papers in Press, August 20, 2009, DOI 10.1074/jbc.M109.049155

Youli Hu‡1, Scott E. Guimond§, Paul Travers¶, Steven Cadman‡, Erhard Hohenester�2, Jeremy E. Turnbull§,Soo-Hyun Kim‡3,4, and Pierre-Marc Bouloux‡4

From the ‡Centre for Neuroendocrinology, University College London Medical School, Royal Free Campus, London NW3 2PF, the§School of Biological Sciences, University of Liverpool, Liverpool L69 7ZB, the ¶Scottish Centre for Regenerative Medicine,University of Edinburgh, Edinburgh EH16 4SB, Scotland, and the �Department of Biological Sciences, Imperial College London,London SW7 2AZ, United Kingdom

Activation of fibroblast growth factor (FGF) signaling is initi-ated by a multiprotein complex formation between FGF, FGFreceptor (FGFR), and heparan sulfate proteoglycan on the cellmembrane. Cross-talk with other factors could affect this com-plex assembly and modulate the biological response of cells toFGF. We have previously demonstrated that anosmin-1, a gly-cosylated extracellularmatrix protein, interacts with the FGFR1signaling complex and enhances its activity in an IIIc isoform-specific and HS-dependent manner. The molecular mechanismof anosmin-1 action on FGFR1 signaling, however, remainsunknown. Here, we show that anosmin-1 directly binds toFGFR1 with high affinity. This interaction involves domains inthe N terminus of anosmin-1 (cysteine-rich region, whey acidicprotein-like domain and the first fibronectin type III domain)and the D2–D3 extracellular domains of FGFR1. In contrast,anosmin-1 binds to FGFR2IIIcwithmuch lower affinity and dis-plays negligible binding to FGFR3IIIc. We also show thatFGFR1-bound anosmin-1, although capable of binding to FGF2alone, cannot bind to a FGF2�heparin complex, thus preventingFGFR1�FGF2�heparin complex formation. By contrast, heparin-bound anosmin-1 binds to pre-formed FGF2�FGFR1 complex,generating an anosmin-1�FGFR1�FGF2�heparin complex. Fur-thermore, a functional interaction between anosmin-1 and theFGFR1 signaling complex is demonstrated by immunofluores-cence co-localization and Transwell migration assays whereanosmin-1was shown to induce opposing effects during chemo-taxis of humanneuronal cells.Our studyprovidesmolecular andcellular evidence for a modulatory action of anosmin-1 onFGFR1 signaling, whereby binding of anosmin-1 to FGFR1 andheparin can play a dual role in assembly and activity of the ter-nary FGFR1�FGF2�heparin complex.

FGF5 signaling plays an important role in a wide range offundamental biological responses (1–3). Both FGF and FGFRbind to heparan sulfate (HS) and heparin, a highly sulfated typeof HS produced in connective tissuemast cells. Heparan sulfateproteoglycans (HSPG) are the cell surface co-receptors essen-tial for the formation of functional FGF�FGFR signaling com-plex (4, 5). There are four structurally related FGFRs (FGFR1–4), which consist of an extracellular ligand-binding regioncontaining three immunoglobulin (Ig)-like domains (D1–D3), asingle transmembrane domain, and a cytoplasmic domain withprotein-tyrosine kinase catalytic activity. The 22 members ofthe FGF family bind to the interface formed by the D2/D3domains and the linker between these domains (6, 7), whereas aconserved positively charged region in D2 serves as the HSbinding site (8). An unusual stretch of seven to eight acidicresidues designated as the “acid box” is present in the linkerconnecting D1 and D2. Alternative splicing events occur togenerate various isoforms, including a truncated receptor lack-ing D1 and the D1–D2 linker or a full-length receptor that dif-fers in the secondhalf ofD3, designated as IIIb and IIIc isoforms(5). Two crystal structures have been proposed to demonstratehow the FGF�FGFR�heparin complex is assembled (9, 10).Recent evidence suggests that bothmay be biologically relevant(11, 12).The diversity of FGF signaling pathways and consequent bio-

logical functions require that activation of FGFR should betightly regulated. Such regulation can occur either at the level ofthe extracellular receptor-ligand complex assembly or via intra-cellular modulation of downstream effectors (13). Extracellularregulation mainly involves the interaction between each com-ponent of the FGF�FGFR�HS signaling complex. For example,FGF8 is shown to bind mostly to the FGFR IIIc isoforms,whereas FGF7 acts as the preferential ligand for the FGFR2 IIIbisoform (13, 14). Sequence specificity, length, and sulfation pat-

* This work was supported by Biotechnology and Biological SciencesResearch Council Grant BB/F007167/1.

□S The on-line version of this article (available at http://www.jbc.org) containssupplemental Figs. S1 and S2.

1 To whom correspondence may be addressed: Centre for Neuroendocrinol-ogy, UCL Medical School, Royal Free Campus, London NW3 2PF, UK. Tel.:44-20-7794-0500; Fax: 44-20-7317-7625; E-mail: [email protected].

2 Wellcome Trust Senior Research Fellow.3 To whom correspondence may be addressed: Division of Basic Medical Sci-

ences, St. George’s, University of London, London SW17 0RE, UK. Tel.:44-20-7794-0500; Fax: 44-20-8725-3326; E-mail: [email protected].

4 Both authors contributed equally to this work.

5 The abbreviations used are: FGF, fibroblast growth factor; FGFR, fibro-blast growth factor receptor; HSPG, heparan sulfate proteoglycan;D1–D3, immunoglobulin (Ig)-like domains 1–3; AB, acid box; NCAM,neuronal cell adhesion molecule; CR, cysteine-rich domain; WAP, wheyacidic protein-like domain; FnIII, fibronectin type III; KS, Kallmann syn-drome; GnRH, gonadotropin-releasing hormone; SPR, surface plasmonresonance; GPI, glycosylphosphatidylinositol; hTERT, human telomer-ase; BS3, bis-sulfosuccinimidyl suberate; SFM, serum-free medium;Sulfo-SANPAH, sulfosuccinimidyl-6-(4�-azido-2�-nitrophenylamino)-hexanoate.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 284, NO. 43, pp. 29905–29920, October 23, 2009© 2009 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

OCTOBER 23, 2009 • VOLUME 284 • NUMBER 43 JOURNAL OF BIOLOGICAL CHEMISTRY 29905

by guest on May 17, 2020

http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/cgi/content/full/M109.049155/DC1

http://www.jbc.org/

terns of HS are also important regulators of the FGF�FGFRinteraction (15, 16).Cell surface proteins other than FGFs and HSPGs partic-

ipate in FGFR signaling regulation. FLRT3 (a member ofthe fibronectin-leucine-rich transmembrane protein family)promotes FGF signaling and interacts with FGFR1 andFGFR4 via its extracellular fibronectin type III (FnIII) do-main (17). Sef (similar expression to fgf genes) functions asan antagonist of FGF signaling in zebrafish. The two FnIIIregions of Sef are essential for its function and interactionwith FGFR1 and FGFR2 (18). Neuronal cell adhesion mole-cule (NCAM), N-cadherin, and L1 have also been identifiedas functionally relevant in FGFR-mediated neurite out-growth (19–22). The FnIII domains of NCAMbind to the D2and D3 domains of FGFR1 (19) and FGFR2 (23) to induceligand-independent receptor phosphorylation.Anosmin-1, an extracellular matrix-associated glycosylated

protein, appears to be a novel member of the extracellularFGFR signaling modulators (24, 25). Loss-of-function muta-tions of anosmin-1 and FGFR1 are associated with Kallmannsyndrome (KS), underlyingX-linked, and autosomal dominant/recessive inheritance mode, respectively (26–28). KS is ahuman developmental genetic disorder characterized by loss ofsense of smell (anosmia) caused by abnormal olfactory bulbdevelopment and delayed, even arrested puberty caused by dis-rupted migration of the gonadotropin-releasing hormone(GnRH)-secreting neuron. We previously reported that anos-min-1 acts as an FGFR1IIIc isoform-specific co-ligand, whichenhances signaling activity. In human embryonic GnRH olfac-tory neuroblast FNC-B4 cells, anosmin-1 induced neuriteoutgrowth and cytoskeletal rearrangements through FGFR1-dependent mechanisms involving p42/44 and p38 mitogen-ac-tivated protein kinases andCdc42/Rac1 activation (25). A func-tional interaction is also demonstrable between anosmin-1 andFGFR1 in optic nerve oligodendrocyte precursor development(24). Structurally, anosmin-1 comprises an N-terminal cys-teine-rich domain (CR) and a whey acidic protein-like (WAP)domain, followed by four tandem FnIII repeats and a C-termi-nal histidine rich region (Fig. 1a). Current evidence suggeststhat anosmin-1 functions by affecting FGF2-induced activationof FGFR1 signaling rather than by directly stimulating thereceptor. However, the precise molecular mechanism of thisinteraction remains unclear.We now report for the first time that anosmin-1 directly

binds to FGFR1 using surface plasmon resonance (SPR), chem-ical cross-linking, and immunofluorescence co-localizationstudies in living cells. This interaction occurs between theN-terminal CR, WAP, and the first FnIII domain of anosmin-1and D2 and D3 ectodomains of FGFR1. Moreover, SPR studiesusing sequential injections and Transwell migration assays inimmortalized FNC-B4-hTERT cells suggest that anosmin-1can have opposing effects in the formation and activation of theFGF2�FGFR1�heparin complex depending on the order of theirbinding interactions with anosmin-1.

EXPERIMENTAL PROCEDURES

Generation of Recombinant Protein Anosmin-1—We havepreviously described the generation of wild-type anosmin-1

(PIWF4) and its truncated forms comprising the CR andWAPdomain followed by one or two FnIII repeats (designated asPIWF1 or PIWF2 (Fig. 1a)) inDrosophila S2 cells (29). To gen-erate the PIF4 construct containing only the four FnIIIdomains, the corresponding cDNA coding sequence wasamplified by PCR using the 5�-CCCGGATCCACTCTGTA-CAAAGGTGTCCCCC-3� forward oligonucleotide primerwith a 5� BamHI restriction site and the 5�-CCCTCTAGATG-GAGAAGGCTTGTAATGATGT-3� reverse primer with a 3�XbaI site. The amplified cDNA fragment was subcloned into amodified Drosophila expression vector PMT/BiP/6His inwhich theV5 epitopewas removed from the original PMT/BiP/V5/6His vector (Invitrogen). Drosophila secretory signal ispresent at the 5�-end and the His6 epitope at the 3�-end. AQuikChange site-directed mutagenesis kit (Stratagene) wasused to introduce the single amino acid substitution of N267Kin the first FnIII domain and E514K and F517L substitution inthe third FnIII domain of PIWF4, creating the three mutantconstructs mPIWF4N267K, mPIWF4E514K, and mPIWF4F517L(Fig. 1a). All constructs were confirmed by DNA sequencingbefore being transfected into S2 cells, and recombinant pro-teins were generated according to previously described proto-cols (29). The schematic representation and purity of variousrecombinant anosmin-1 analogues are shown in Fig. 1.Generation of Recombinant FGFR1D1D3 and FGFR1D2D3

Proteins—pCEP-Pu FGFR1 constructs encoding the D1–D3 orthe D2–D3 ectodomains of human FGFR1 IIIc isoform(SwissProt entry P11362) have been described previously (30).For generation of FGFR1 ectodomain proteins (D1–D3 andD2–D3), 293-EBNA cells were cultured in Dulbecco’s modifiedEagle’smedium/F-12 (Invitrogen) supplementedwith 10% fetalcalf serum and 250 �g/ml G418. After transfection withpCEP-Pu FGFR1 expression constructs, which are maintainedepisomally in these cells, stable cell lines were obtained by se-lection in medium containing 1 �g/ml puromycin (Sigma).Secreted recombinant proteins in the conditioned mediumwere purified using TALON metal affinity resins (Clontech).Purity of the proteins was confirmed by SDS-PAGE and colloi-dal blue staining (Invitrogen), and quantification of eluted pro-teinswasmeasured byBradford assay (Bio-Rad). The schematicdomain structure and purity of recombinant FGFR1 D1D3 andD2D3 are shown in Fig. 1.SPR Analysis of Anosmin-1 and FGFR Interactions—Binding

analysis was performed using a BIAcore 3000 SPR-based bio-sensor (BIAcore AB) to quantify kinetic parameters for theinteraction between anosmin-1 and recombinant FGFR1D1D3and FGFR1D2D3 proteins. PIWF4 was immobilized onto aresearch grade CM4 chip (Biosensor AB, Uppsala, Sweden),whereas the truncated and mutated variants PIWF1, PIWF2,PIF4, mPIWF4N267K, mPIWF4E514K, and mPIWF4F517L werecoupled onto research grade CM5 chips (Biosensor AB, Upp-sala, Sweden). The CM4 chip is similar to sensor Chip CM5 butwith a lower degree of carboxymethylation, which can improvesensitivity for certain interactions. FGFR1D1D3, FGFR1D2D3,FGFR2c, and FGFR3c (the recombinant FGFR2c and FGFR3cectodomain proteins were generous gifts from Alan Brown(University of Cambridge)) were used as soluble analytes for thebinding assays. Immobilization of anosmin-1 to the chips was

Anosmin-1 on FGFR1 Signaling Complex Assembly

29906 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 284 • NUMBER 43 • OCTOBER 23, 2009


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

through its primary amino groups usingN-ethyl-N-(dimethya-minopropyl)carbodiimide/N-hydroxysuccinimide accordingto a standard amine-coupling protocol. Carboxymethyl groupson the chip surface were first activated using an injection pulseof 50 �l (flow rate, 5 �l/min) of an equimolar mix of N-ethyl-N-(dimethyaminopropyl)carbodiimide and N-hydroxysuccin-imide (final concentration, 0.05 M; mixed immediately prior toinjection). Following activationwithN-ethyl-N-(dimethyamin-opropyl)carbodiimide/N-hydroxysuccinimide, immobilizationof anosmin-1 was achieved by applying diluted anosmin-1(150–300 �g/ml) in HBS-EP buffer (0.01 M HEPES, 0.15 M

NaCl, 3.4mMEDTA, 0.005% polysorbate 20 (v/v)), pH 7.4) ontothe activated chip surface. Excess unreacted sites on the sensorsurface were deactivated with a 40-�l injection of 1 M ethanol-amine. Successful immobilization was confirmed by the obser-vation of a 3000–6000 response unit (RU) increase for mostanosmin-1 constructs and 600-RU increase for PIF4. Threeflow cells were coupled, and one flow cell contained a blanksensor chip serving as a reference surface.Different concentrations of analytes (FGFR1D1D3,

FGFR1D2D3, FGFR2c, and FGFR3c) in HBS-EP buffer wereinjected over the anosmin-1 sensor chip at a flow rate of 20�l/min. At the end of each sample injection (240 s), HBS-EPbuffer was passed over the sensor surface to monitor the disso-

ciation phase. Following 240 s ofdissociation, the sensor surface wasfully regenerated by injection of 50�l of 2 M NaCl in 100 mM sodiumacetate buffer (pH 4). The bulk shiftdue to changes in refractive indexwasmeasured using a reference sur-face and was subtracted from thebinding signal to correct for non-specific signals. Sensorgrams andkinetic parameters generated wereanalyzed using BIA Evaluation soft-ware version 3.0.SPR Analysis of Effects of Anos-

min-1 on FGF2�FGFR1�HeparinAssembly by Sequential Injection—To investigate the role of anos-min-1 on the formation of theFGF2�FGFR1�heparin complex, SPRanalysis was performed by sequen-tial injections of FGFR1, FGF2,and heparin over a PIWF4-coupledchip. All experiments were carriedout at 25 °C in HBS-EP buffer at aflow rate of 20 �l/min with the sam-ple injection of 240s. In one experi-ment, FGFR1 was first injected overPIWF4 to form a PIWF4�FGFR1complex on the sensor chip surface.FGFR1, either at a constant (250 nM)or at various concentrations, wasfirst injected over PIWF4 at con-stant HBS-EP buffer flow in the dis-sociation phase to remove unbound

FGFR1, leaving bound FGFR1 on the chip surface. This wasthen followed by the second injections of FGF2 at a constant (50nM) or at various concentrations, heparin alone or preincubatedFGF2�heparin at specified concentrations. To further analyzethe effect of heparin on PIWF4�FGF2�FGFR1 complex assem-bly, various concentrations (0, 0.1, 1, and 10 �g/ml) of heparinwere injected after the first injection of 250 nM FGFR1 and thesecond injection of 50 nM FGF2 over the PIWF4-coupled chip.The time lag within each injection was due to limitations in theBIAcore instrument. In another experiment, 10 �g/ml heparinwas first injected over PIWF4-coupled chip resulting inheparin�PIWF4 complex on the chip surface, followed by sub-sequent second injections of 50 nMFGFR1, 50 nMFGF2 alone orpreincubated FGFR1�FGF2 or FGFR1�FGF2�heparin complex.Cross-linking Studies of Interaction between Anosmin-1 and

FGFR1—Bis-sulfosuccinimidyl suberate (BS3) cross-linking ex-periments were carried out in a volume of 100 �l in phosphate-buffered saline. The mixture of PIWF4 and FGFR1D1D3 wasincubated for 1 h at room temperature, and cross-linking ofcomplex was initiated by adding BS3 (Pierce) to a final concen-tration of 0.01 mg/ml concentration, incubated for a further 15min, and then quenched by adding 20 mM Tris, pH 7.5.The photoactive cross-linking assay was carried out by using

PIWF1 labeled with heterobifunctional cross-linker Sulfo-

FIGURE 1. Generation of recombinant anosmin-1, anosmin-1 mutants, FGFR1D1D3, and FGFR1D2D3proteins. a, the schematic structures of recombinant proteins of anosmin-1 and FGFR1. Each domain in thewild type (PIWF4), point mutants (mPIWF4N267K, mPIWF4E514K, and mPIWF4F517L), and truncated (PIWF1, PIWF2,and PIF4) anosmin-1 protein analogues are represented by a shaded rectangle. V5 and 6His epitopes at the Cterminus are represented by a clear rectangle. Each immunoglobulin-like domain in the full ectodomain(FGFR1D1D3) and truncated form (FGFR1D2D3) of FGFR1 is represented by a half circle. The acid box (AB) isrepresented by a filled rectangle. H, histidine-rich region. b, 0.5–1 �g of purified recombinant proteins areloaded in each lane and visualized by colloidal blue staining. Molecular mass markers in kilodaltons are shownon the left.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

SANPAH containing an amine-reactive N-hydroxysuccinim-ide ester and a photoactivatable nitrophenyl azide. To labelPIWF1, 10-fold molar excess of Sulfo-SANPAH (Pierce) wasmixedwith 3.25�MPIWF1 and incubated at room temperaturefor 1 h. Non-reacted cross-linker was removed by dialysis over-night at 4 °C. Various amounts (0, 5, 10, and 15 �l) of Sulfo-SANPAH-labeled PIWF1wasmixedwith 233 nM FGFR1D1D3in a volume of 15 �l for 1 h at room temperature. The mixturewas thenplaced on ice, andphotolysiswas performedwith aUVlight source that irradiates at 365 nm at a distance of 5 cm for 15min. The photoactivated samples were heat-denatured at 95 °Cfor 5 min prior to SDS-PAGE electrophoresis.All cross-linked samples were then resolved by a 4–12%

SDS-PAGE and analyzed by polyclonal anti-FGFR1 (H-76,Santa Cruz Biotechnology), monoclonal anti-V5 (Invitrogen),and anti-anosmin-1 polyclonal antibody raised against recom-binant protein PIWF1 as previously described (29).Generation of COS7 Cells Co-expressing FGFR1 and

Anosmin-1-GFP—To generate the full-length FGFR1 IIIcexpression construct with a C-terminal 3xMyc tag, the codingsequence was amplified by PCR primers introducing BamHI(5�-end) and SalI (3�-end) restriction sites and subcloned intopCMV-3Tag-9 vector (Stratagene). Anosmin-1-GFP was gen-erated by PCR amplification of the full-length human KAL1gene coding sequences, including the signal peptide, by usingprimers that introduced SalI (5�-end) and BamHI (3�-end)restriction sites. The digested PCR product was ligated topEGFP-N1 vector to create a C-terminal green fluorescenceprotein (enhanced GFP). COS7 cells were cultured in Dulbec-co’s modified Eagle’s medium supplemented with 10% fetalbovine serum, 50 units/ml penicillin, and 50 �g/ml streptomy-cin in a 5% CO2 at 37 °C. FUGENE reagent (Roche AppliedScience) was used for all transfections according to the manu-facturer’s protocol. To establish stably transfected anosmin-1-GFP linage, the transfected COS7 cells were selected in G418(600 �g/ml)-containing medium for 2 weeks before being ana-lyzed by immunofluorescence. To confirm whether the exoge-nous constructs were expressing the respective proteins, totalcell lysates of the transfected COS7 cells were prepared in lysisbuffer (1% Triton X-100, 50 mM Tris-HCl at pH 8.0, 150 mM

NaCl) containing 1% aprotinin, 100 �g/ml phenylmethylsulfo-nyl fluoride, 10 mM sodium fluoride, 1 mM sodium orthovana-date, and 1 mM dithiothreitol, separated on SDS-PAGE, trans-ferred onto nitrocellulose membrane, and probed withmonoclonal antibodies against Myc-epitope (9E10) and greenfluorescence protein (3E1), both from Cancer Research UK.Secondary antibody was peroxidase-labeled horse anti-mouseIgG (Vector Laboratories).Immunofluorescence—The COS7 cells were plated on a cov-

erslip in a 24-well culture plate 24–30 h after transfection.Once the cells were attached, depending on the experiment, theculture medium was changed to serum-free medium for 12 hbefore stimulation with 25 ng/ml FGF2 for 20 min. Cells werefixed in chilled 4% formaldehyde on ice, washed in phosphate-buffered saline and 10mM ammonium chloride, and permeabi-lized with 0.05% Triton X-100. For staining of cell surface pro-teins, the permeabilization step was omitted. After blockingwith 3% bovine serum albumin in phosphate-buffered saline,

cells were incubated with rabbit polyclonal FGFR1 antibodies(C-15 or H-76, both from Santa Cruz Biotechnology), whichwere subsequently detected by goat anti-rabbit IgG conjugatedwith Alexa Fluor 555 (Invitrogen Molecular Probes). Hoechst(Invitrogen) was used for nuclear staining. Coverslips weremounted on slides using ProLong Gold Antifade (Invitrogen).Each experiment was repeated a minimum of three times, andthe most representative images from random fields wereshown. Confocal microscopy was performed using an Axiovert200M/LSM 510 Meta laser-scanning confocal microscope(Zeiss, UK). All imageswere taken under aZeiss Plan-Apochro-mat 40 � 1.3 numerical aperture oil-immersion objective andpinhole setting at 1.00 Airey unit. Twelve-bit single directionalimages were collected to peak by separate excitation ofenhanced GFP, Alexa Fluor 555, and Hoechst using the 488,543, and 405 nm lasers, respectively. All images shown are sin-gle confocal sections.Telomerase-mediated Immortalization of FNC-B4 Cells—A

primary neuroblast culture, FNC-B4, originated from humanfetal olfactory epithelium, has been previously described (31).To establish an immortal derivative of FNC-B4 (designated asFNC-B4-hTERT), cells grown in F-12 Coon’s modificationmedium supplemented with antibiotics and 10% fetal bovineserum were sequentially transduced with two different replica-tion-defective retroviral vectors. First, pWXL-Neo-Eco, anamphotropic retroviral construct containing the mouse basicamino acid transporter (ecotropic receptor) and the neomycinresistance marker, was transfected into AM12 amphotropicretrovirus packaging cell line. After 48 h, the live virus stockwasharvested to infect the FNC-B4 cells in the presence of Poly-brene (4 �g/ml). After selection with G418 (400 �g/ml), theecotropic receptor expressing FNC-B4 cells were furtherinfected with pBabe-puro-hTERT, an ecotrophic retroviralvector containing the catalytic subunit of human telomeraseand puromycin resistance, which had been produced inBOSC293packaging cell line. After selectionwith puromycin (1�g/ml), resistant colonies were pooled and regularly passageduntil the cells overcame the cellular senescence and continuedto grow, as compared with the pBabe-puro empty vector-in-fected cells, which ceased to proliferate (supplemental Fig. S1).CellMigrationAssay—Migration assayswere performedon a

24-well Transwell with 8.0-�m polycarbonate membrane(Corning Costar), which was coated with 0.2 mg/ml gelatin inphosphate-buffered saline. The lower compartmentwas loadedwith either SFM, FGF2 (1 nM), PIWF4 alone (1, 10, or 50 nM), or10 nM PIWF4 in the presence of 20 �M SU5402. In the topcompartment, serum-starved FNC-B4-hTERT cells (5 � 104cells in 200�l of SFM)were plated and incubated at 37 °C in 5%CO2 for 4 h before quantification of the migrated cells. To gen-erate HS-deficient FNC-B4-hTERT cells, 30 mM sodium chlo-rate was added in the culturemedium throughout all processes.For the PIWF4�heparin functional interaction study (Fig. 9c),the lower compartmentwas loadedwith SFM, 1�g/ml heparin,10 nM PIWF4 alone, or 10 nM PIWF4 plus varying concentra-tions of heparin, whereas the sodium chlorate-treated cellswere loaded onto the Transwell insert. For the PIWF4/FGFR1functional interaction study (Fig. 9d), sodium chlorate-treatedcells were preincubated with SFM, 10 nM PIWF4, or 10 nM




http://ww

w.jbc.org/

Dow

nloaded from


http://www.jbc.org/

PIWF4 plus 10 nM FGFR1D1D3. After 1-h incubation, cellswere gently spun down, resuspended in SFM and loaded ontothe top chamber, while the mixture of 10 nM PIWF4 and 200ng/ml heparinwas loaded into the lower compartment. For thisexperiment, cells were incubated for 20 h to allow significantmigration.Migrated cells on the bottom of themembrane werefixed in 96% methanol, stained with hematoxylin solution(Sigma), and mounted onto the glass slides. Chemotaxis wasquantified by counting five random fields per membrane usinga 20� objective. The average values for each membrane wereexpressed as number of cells per high power field. All experi-ments were performed at least twice in duplicates.

RESULTS

Determination of Direct Binding of Anosmin-1 to FGFR1—Direct interaction of anosmin-1 and FGFR1 was first detectedusing SPR analysis. In this study, the recombinant wild-typefull-length anosmin-1 and its various mutant analogues wereimmobilized individually on the sensor chip surface, throughtheir native amine groups generating comparable couplingresponse unit (RU) increases. Soluble FGFR1 D1D3 and D2D3proteins were injected at serial dilutions. Representative SPRsensorgrams for each experiment are shown (Figs. 2 and 3). Theresults demonstrate that full-length anosmin-1 (PIWF4) candirectly bind to FGFR1 with a relatively high association rateand very low dissociation rate generating a nanomolar range(�10nM) dissociation constant (Kd), which indicates high bind-ing affinity between the two proteins (Fig. 2a and Table 1).Similar binding affinities were observed not only with theFGFR1D1D3 containing the three extracellular Ig domains, butalso with FGFR1D2D3 with only two Ig domains, missing thefirst Ig domain (D1) and the acid box located in the D1–D2linker region (Fig. 3a and Table 1). This indicates that FGFR1binding to anosmin-1 is mainly dependent on its second (D2)and third (D3) Ig domains. Interestingly, however, an approxi-mate 3-fold increase of the Kd values was obtained in the bind-ing of PIWF4with FGFR1D2D3 (�2 nM) as comparedwith thatwith FGFR1D1D3 (�7 nM). One interpretation of these datacould be that the acid box together with the D1 domain ofFGFR1 may play a regulatory role in the interaction with anos-min-1 by partially blocking the anosmin-1 binding site withinthe D2–D3 domains, a scenario reminiscent of the intramolec-ular autoinhibition by these domains.Identification of Domain-specific Interaction of Anosmin-1

with FGFR1—To further determine the specific domain(s) ofanosmin-1 binding to FGFR1, we generated C-terminal trun-cated mutants PIWF1 and PIWF2, encompassing the N-termi-nal CR andWAP followed by one or two FnIII domains, respec-tively (Fig. 1b). SPR analyses using immobilized PIWF1 orPIWF2 clearly demonstrate that both are still capable of bind-ing to FGFR1D1D3 and FGFR1D2D3 (b and c of Figs. 2 and 3).As seen with PIWF4, they both bind to FGFR1D2D3 withslightly higher affinity than to FGFR1D1D3. The only differ-ence was that PIWF2 generated similar Kd values to the full-length counterpart, whereas PIWF1 showed an approximate5-fold lower binding affinity (Table 1). These observations sug-gested that the N-terminal CR, WAP, and the first FnIIIdomains of anosmin-1 (i.e. PIWF1) are sufficient for FGFR1

binding and the additional FnIII domains are dispensable in thisinteraction.Because there is growing evidence showing that FnIII

domains in NCAM and L1 are involved in direct FGFR1 bind-ing (19, 20), we sought to determine whether the N-terminalregion (CR and WAP) of anosmin-1 was required for theFGFR1 binding by testing the PIF4 construct, which containsonly the four FnIII domains. In addition, we introduced threepoint mutations (N267K, E514K, and F517L) into the full-length PIWF4 to investigate the effects of KS-related missensemutations in the FnIII domains. When SPR assays were con-ducted using these mutant anosmin-1 proteins, PIF4 showedno apparent bindingwith the two forms of FGFR1 ectodomains(Figs. 2d and 3d). N267K substitution in the first FnIII domainalso leads to a complete loss of FGFR1 binding. By contrast,E514K and F517L substitutions in the third FnIII domain stillretained nanomolar range binding affinity, albeit 2- to 6-foldreduced, to both FGFR1D1D3 and D2D3 (Table 1 and e–g ofFigs. 2 and 3). Taken together, these data suggest that the first,but not the third, FnIII domain is critical for FGFR1binding andthe N-terminal CR and WAP domains are also required forFGFR1 interaction.Determination of Anosmin-1 Binding to FGFR2c and FGFR3c—

Our study so far has been focused on interactions with FGFR1,based on the fact that anosmin-1 showed FGFR1IIIc specificactivity in a BaF3 cell system (25) and a pointmutation presum-ably affecting the FGFR1IIIc-FGF8b interaction has been impli-catedwithKS cases (32). It is unknown, however, whether anos-min-1 has the capacity to interact with other FGF receptors.Thus, we further examined the binding affinity of anosmin-1with FGFR2IIIc and FGFR3IIIc by SPR (Fig. 4). SolubleFGFR2IIIc D1D3 protein injected over the PIWF4-coupledsensor chips generated amore than 7-fold higherKd value com-pared with that of FGFR1IIIc D1D3 (57.63 nM versus 7.64 nM),indicating that anosmin-1 preferentially binds to FGFR1 overFGFR2 (Tables 1 and 2). By contrast, anosmin-1 showed negli-gible binding to FGFR3c (Fig. 4d). Interestingly, PIF4, whichshowed no binding to FGFR1, demonstrated some weak bind-ing to FGFR2c, albeit with significant decrease of affinity with aKd increasing from57.63 nM to 380.15 nM,when comparedwithPIWF4 (Table 2). This again indicates a requirement of theN-terminal CR and WAP domains for optimal FGFR binding.Cross-linking Study on the Interaction of Anosmin-1 with

FGFR1—In the SPR assay, recombinant anosmin-1 analogueswere immobilized, and thus the interaction detected is in animmobilized phase. To investigate whether anosmin-1 andFGFR1 can interact in the solution state, we conducted twotypes of chemical cross-linking experiments, either by addingBS3 cross-linker directly to the reaction solution, or by couplinganosmin-1 with photoactive Sulfo-SANPAH cross-linker. IntheBS3 cross-linking studywhere 25 nMFGFR1D1D3was incu-bated with increasing amount of PIWF4, three bands weredetected by anti-FGFR1 (H-76) antibody. The bands corre-sponding to �60 and �120 kDa represent the monomeric andhomodimeric forms of FGFR1D1D3, respectively, as they arepresent even in the absence of PIWF4. A higher bandmigratingwith amolecular mass of�150 kDa, however, started to appearwith increasing intensity only when increased amount of




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

FIGURE 2. BIAcore analysis of anosmin-1 binding to FGFR1D1D3. Soluble FGFR1D1D3 at varying concentrations was injected over PIWF4 (a), PIWF2 (b),PIWF1 (c), PIF4 (d), mPIWF4N267K (e), mPIWF4E514K (f), and mPIWF4F517L (g) coupled sensor chips.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

FIGURE 3. BIAcore analysis of anosmin-1 binding to FGFR1D2D3. Soluble FGFR1D2D3 at varying concentrations was injected over PIWF4 (a), PIWF2 (b),PIWF1 (c), PIF4 (d), mPIWF4N267K (e), mPIWF4E514K (f), and mPIWF4F517L (g) coupled sensor chips.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

PIWF4 was added to the reaction. This band was thought torepresent the PIWF4�FGFR1D1D3 complex (�90 kDa ofPIWF4 and�60 kDa of FGFR1D1D3) (Fig. 5a). To confirm thisfinding, the same blot was reprobed with anti-V5 antibody,which detected two bands: one at �90 kDa, which correspondsto PIWF4, and another band at �150 kDa, which overlaps withthe band previously identified by the FGFR1 antibody, in-dicating that this band indeed contains the cross-linkedPIWF4�FGFR1D1D3 complex (Fig. 5b).We further used a poly-

clonal anti-anosmin-1 antibody to confirm complex formation.Again, incubation of an equal amount (50 nM) of PIWF4 andFGFR1D1D3 resulted in the appearance of a band of�150 kDa.The very faint backgroundband of the same size present even inthe absence of D1D3 indicates a self-dimerized PIWF4 (Fig. 5c).To eliminate the possibility that anosmin-1 protein in solutionmay stick to random proteins nonspecifically, the cross-linkingexperiments were repeated using an irrelevant recombinantprotein, Robo, which also contains two Ig domains. No appar-ent band corresponding to the PIWF4�Robo complex wasobserved (data not shown), confirming the specific interactionbetween PIWF4 and FGFR1D1D3.For the photoactive cross-linking assay, PIWF1 was first

labeled with Sulfo-SANPAH cross-linker, which is then acti-vated by photolysis after UV light exposure.Western blot anal-ysis of the cross-linked proteins with anti-FGFR1 antibodydetected an emerging band of �98 kDa, equivalent to the com-bined molecular mass of PIWF1 (�38 kDa) and FGFR1D1D3(�60 kDa), whose intensity increased in proportion to theamount of PIWF1 added (Fig. 5d). Again, when this blot wasreprobed with anti-V5 antibody, the presence of PIWF1 wasconfirmed in this band (Fig. 5e). Moreover, it was found thatPIWF1 could self-dimerize, ranging from monomeric to tri-meric forms in contrast to the full-length PIWF4, mostly pres-

FIGURE 4. BIAcore analysis of anosmin-1 binding to FGFR2c and FGFR3c. Soluble FGFR2IIIcD1D3 at varying concentrations was injected over PIWF4 (a),PIWF1 (b), and PIF4 (c), and soluble FGFR3IIIcD1D3 was injected over PIWF4 (d).

TABLE 1BIAcore analysis of anosmin-1 binding to FGFR1

Solubleanalyte

Coupledreagent kon koff Kd

M�1 s�1 s�1 nMFGFR1 PIWF4 1.01 � 0.18 � 104 7.68 � 0.98 � 10�5 7.64 � 0.68D1D3 PIWF2 8.82 � 3.71 � 103 1.01 � 0.59 � 10�4 11.93 � 8.05

PIWF1 4.07 � 1.32 � 103 1.50 � 0.87 � 10�4 35.10 � 8.57PIF4 NBa NB NB

mPIWF4N267K NB NB NBmPIWF4E514K 4.75 � 3.03 � 103 6.64 � 4.35 � 10�5 15.08 � 7.53mPIWF4F517L 5.28 � 2.48 � 103 2.36 � 1.20 � 10�4 48.63 � 19.86

FGFR1 PIWF4 3.92 � 0.07 � 104 1.04 � 0.37 � 10�4 2.66 � 0.96D2D3 PIWF2 2.24 � 0.09 � 104 2.30 � 2.27 � 10�4 3.79 � 2.69

PIWF1 6.84 � 0.60 � 103 1.10 � 0.37 � 10�4 16.20 � 6.18PIF4 NB NB NB

mPIWF4N267K NB NB NBmPIWF4E514K 8.25 � 3.61 � 103 1.99 � 0.66 � 10�4 26.70 � 10.62mPIWF4F517L 6.80 � 1.90 � 103 3.48 � 1.12 � 10�4 50.80 � 4.08

a NB, negligible binding.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

ent in monomeric form. Data from these cross-linking assaysalso indicate that anosmin-1�FGFR1 complex is formedwith 1:1stoichiometry.Co-recruitment of Anosmin-1 and FGFR1 to the Peripheral

Plasma Membrane—We have previously reported that anos-min-1 can be co-immunoprecipitated with FGFR1 (25). Ourcurrent studies showing direct binding between recombinantanosmin-1 analogues and FGFR1 ectodomains further supportthis observation. To demonstrate whether the physical interac-tions between these two proteins also occur in the presence ofintact extracellular matrix in living cells, we employed confocallaser scanning microscopy to visualize the subcellular localiza-tion of these proteins after immunofluorescence staining.When COS7 cells transfected with full-length FGFR1 expres-sion construct were immunostained with polyclonal anti-FGFR1 antibody (C-15), the signal was detected as a punctatepattern at the plasmamembrane and in cytoplasmic endosomalvesicles. FGFR1 staining was also detected in and around thecell nucleus (Fig. 6a), similar to previous reports (22, 33).When

cells transfected with anosmin-1-GFP construct were exam-ined, it was evident that GFP-tagged anosmin-1 protein wassecreted into surrounding medium, scattering to adjacent cells(Fig. 6a). Anosmin-1-GFP also showed a punctate pattern onthe cell periphery and within the cytoplasm, indicating accu-mulation along the plasma membrane as well as within theendosomal vesicles, as expected from an extracellular matrix-associated secretory protein, and also consistent with a previ-ous report (34).WhenFGFR1 and anosmin-1-GFP imageswereoverlaid, there was considerable overlap of the FGFR1 stain-ing with the GFP (marked with arrows in Fig. 6a), compatiblewith FGFR1�anosmin-1 co-localization in cells. No signalwas observed in untransfected cells (in the case of anosmin-1-GFP) or when primary antibodies were omitted (in thecase of FGFR1) (data not shown). Western blotting of lysatesfrom the transfected COS7 cells indicated that the exoge-nous constructs were expressing the respective proteins ofexpected size (Fig. 6c).It is possible that forced expression of exogenous genes at

high levels may lead to abnormal protein localization. Toaddress this possibility, we next examined the interaction of theendogenous FGFR1 protein in cells that are stably transfectedsolely with anosmin-1-GFP construct, because stable transfec-tion leads to a moderate, more physiologically relevant levelof protein expression. These modifications, however, alsoresulted in decreased immunofluorescence signal intensitydetected by the confocal microscopy (data not shown). Toenhance the visualization of FGFR1 and anosmin-1 interactionon the cell surface, we stained the cells without permeabiliza-tion, thereby helping to preserve plasma membrane integrity.In addition, we used an FGFR1 ectodomain-specific antibody(H-76) in the serum-starved cells after stimulation with FGF2,which allowed better detection of the extracellular domains ofFGFR1on the cell surface (Fig. 6b). In these cells, themajority ofthe endogenous FGFR1 detected on the plasma membrane co-localized with the anosmin-1-GFP. This result is consistentwith observations from the initial experiments using overex-pression constructs. Together, these data support the notionthat anosmin-1 directly interacts with FGFR1 in living cellsunder physiological conditions.Effect of Anosmin-1 on FGF2�FGFR1�HeparinAssemblyUsing

Sequential Injection Analysis—Having documented the directinteraction between anosmin-1 and FGFR1 using multiplecomplementary approaches, we sought to further determinewhether complex formation between anosmin-1 and FGFR1would affect the subsequent interaction with FGF2 and hepa-rin, and thus conducted a series of SPR experiments using asequential injection protocol. We first injected FGFR1D1D3over a PIWF4-coupled sensor surface, followed by a secondinjection with FGF2 at varying concentrations. Thus PIWF4-bound FGFR1 would be expected to remain on the sensor sur-face and unbound FGFR1 washed away by HBS-EP buffer pass-ing over the sensor surface. As shown in Fig. 7a, FGF2 did bindto PIWF4�FGFR1 complex in a dose-dependent manner. Thisinteraction was reproducibly observed in independent sequen-tial injection experiments where varying concentrations ofFGFR1 (first injection) and a constant concentration of FGF2(second injection) were used (Fig. 7b). However, when FGF2

FIGURE 5. Cross-linking study on the interaction of anosmin-1 withFGFR1D1D3. a and b, PIWF4 at the indicated concentrations was incubatedwith 25 nM FGFR1D1D3, and the complexes formed were cross-linked withBS3, resolved by 4 –12% SDS-PAGE, and identified with anti-FGFR1 polyclonalantibody (a) or anti-V5 monoclonal Ab (b). c, 50 nM PIWF4 with or without 50nM FGFR1D1D3 were cross-linked with BS3, and the cross-linked complex wasanalyzed by anti-anosmin-1 polyclonal Ab. d and e, PIWF1 labeled with pho-toactivatable heterobifunctional cross-linker Sulfo-SANPAH was incubatedwith 233 nM FGFR1D1D3. The cross-linked complex was detected by anti-FGFR1 pAb (d) or anti-V5 mAb (e). Š, monomer; ŠŠ, dimer; ŠŠŠ, trimer;4,cross-linked anosmin-1�FGFR1D1D3 complex.

TABLE 2BIAcore analysis of anosmin-1 binding to FGFR2c and FGFR3cAssociation rate constants (kon), dissociation rate constants (koff), and apparentdissociation constants (Kd) for soluble FGFR1D1D3, FGFR1D2D3, FGFR2c, andFGFR3c binding to immobilized anosmin-1. Values are shown as mean � S.D.derived from duplicate measurements in at least two separate experiments.

Solubleanalyte

Coupledreagent kon koff Kd

M�1 s�1 s�1 nMFGFR2c PIWF4 2.93 � 2.46 � 104 1.39 � 0.93 � 10�3 57.63 � 28.21

PIWF1 2.58 � 1.44 � 104 1.52 � 0.63 � 10�3 69.22 � 34.62PIF4 4.24 � 1.53 � 103 1.64 � 0.68 � 10�3 380.15 � 99.67

FGFR3c PIWF4 NBa NB NBa NB, negligible binding.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

was injected over PIWF4-coupled surface without the priorinjection of FGFR1, there was no change in RU, indicating thatFGF2 could not bind to PIWF4 alone (supplemental Fig. S2).These data suggested that binding of PIWF4 to FGFR1 did notimpair the ability of FGFR1 to bind to FGF2 and that, althoughPIWF4 does not bind to FGF2, PIWF4�FGF2�FGFR1 complexcould be formed on a pre-existing PIWF4�FGFR1 complex.Because a functional FGFR1 signaling complex would stillrequire HS, we then used a third injection with heparin atvarying concentrations to see if heparin could affect thisPIWF4�FGF2�FGFR1 complex. The results, however, showedthat heparin stripped FGF2 off from the PIWF4�FGF2�FGFR1complex in a dose-dependentmanner.At 10�g/ml heparin, theRU dropped from�260 to�120 RU, which is equivalent to thelevel of PIWF4�FGFR1 complex, rather than to the initialbaseline, indicating that stripping of FGF2 from PIWF4�FGF2�FGFR1 complex did not disrupt the PIWF4�FGFR1 inter-action (Fig. 7c). This observation was supported by the factthat when 10 �g/ml heparin was injected directly overPIWF4�FGFR1-bound surface, it did not change the RU levelsignificantly. Furthermore, when FGF2 was preincubated withheparin, it could not further bind to PIWF4�FGFR1 (Fig. 7d).Taken together, these data show that PIWF4-bound FGFR1still retains the ability to bind FGF2, but cannot lead to subse-quent FGF2�FGFR1�heparin complex formation.We have previously demonstrated that anosmin-1 binds to

heparin with a Kd around 2 nM (29), a binding affinity compa-rable to the anosmin-1�FGFR1 interaction in the present study.

We therefore asked whether hepa-rin-bound anosmin-1 could affectthe interaction of FGFR1 and FGF2�FGFR1�heparin complex formation.To address this, 10 �g/ml heparinwas first injected over PIWF4-coupled sensor chip allowingheparin�PIWF4 complex formationon the surface, followed by theinjection of either FGFR1, FGF2alone, or preincubated FGF2�FGFR1or FGF2�FGFR1�heparin. There wasno RU change when FGFR1 (50 nM)was injected (Fig. 8), even at higherconcentrations (data not shown),indicating that heparin-bound anos-min-1 lost the ability to bind FGFR1,such that the anosmin-1�FGFR1�heparin ternary complex couldnot beassembled. Notably, when FGF2 wasinjected, it reduced the RU almost tobasal level, equivalent to the pre-hep-arin injection state, suggesting thatFGF2 could strip off heparin fromanosmin-1, consistent with the ob-servation that heparin can competi-tively displace FGF2 from anosmin-1�FGFR1�FGF2 complex (Fig. 7c).However, a significant increase ofRU was observed when preincu-

bated FGFR1�FGF2 was injected (Fig. 8), suggesting that theanosmin-1�heparin interaction could favor subsequent bind-ing to pre-existing binary FGF2�FGFR1 complex, allowinganosmin-1�FGF2�FGFR1�heparin complex formation. Takentogether, these data demonstrate that the order of anos-min-1 interaction with FGFR1 and heparin can lead to dif-ferential effects on the ternary FGFR1�FGF2�heparin com-plex formation.Interactions of Anosmin-1with FGFR1 andHSPlayOpposing

Roles in the Migration of FNC-B4-hTERT Cells—Our findingsfrom the sequential injection experiments suggest thatheparin-bound anosmin-1 facilitates functional FGFR1 signal-ing complex formation, while anosmin-1-bound FGFR1 can-not. To investigate whether this dual role of anosmin-1 onFGF2�FGFR1�HS complex formation can influence the biolog-ical activity of FGFR1 in living cells, we have employed immor-talized human olfactory GnRH neuroblasts (FNC-B4-hTERT)to perform Transwell migration assays.The primary FNC-B4 cells, originally derived from human

fetal olfactory epithelium (31) have provided a useful modelsystem to study olfactory GnRH neurogenesis, and we havepreviously reported that both anosmin-1 and FGF2 caninduce neurite outgrowth in these cells (25). By using a ret-roviral vector expressing the catalytic subunit of humantelomerase (hTERT), we have established an immortal deriv-ative line of FNC-B4 cells (supplemental Fig. S1a). Our anal-yses indicate that the immortalized cells still maintain the

FIGURE 6. FGFR1 co-localizes with anosmin-1 in COS7 cells. a, COS7 cells were co-transfected transientlywith constructs expressing FGFR1–3xMyc and anosmin-1-GFP. The cells were fixed on the coverslips beforeincubation with polyclonal FGFR1 antibody (C-15) raised against C terminus of human FGFR1, which is subse-quently labeled by anti-rabbit secondary antibody conjugated with Alexa Fluor 555 (red). Anosmin-1 isdetected by GFP (green). Hoechst staining shows the nucleus (blue). Scale bars represent 10 �m. Co-localizationof FGFR1 and anosmin-1 is evidenced by overlapping fluorescence signals (yellow), as indicated with arrows inthe merged image. b, COS7 cells were stably transfected with anosmin-1-GFP by G418 selection. Selected cellswere plated on coverslips, serum-starved overnight, and stimulated with FGF2 (25 ng/ml) for 20 min to inducecell surface receptor recruitment. After fixing, unpermeabilized cells were analyzed for localization of endog-enous FGFR1 and anosmin-1 using polyclonal FGFR1 antibody (H-76) and GFP, respectively. FGFR1 H-76 anti-body, raised against the ectodomain of human FGFR1, was detected by anti-rabbit secondary antibody con-jugated with Alexa Fluor 555. The merged image indicates overlapping fluorescence signals (yellow) of theendogenous FGFR1 ectodomain (red) and stably transfected anosmin-1 (green) on the cell surface. Scale barsrepresent 10 �m. c, Western blotting analysis to confirm expression of respective tagged proteins in these cells.Exogenous FGFR1 was probed with anti-Myc antibody (9E10) to distinguish it from endogenous protein,although FGFR1 antibody was used for all immunofluorescence staining. Anosmin-1 expression was detectedwith anti-GFP antibody (3E1).




http://ww

w.jbc.org/

Dow

nloaded from



http://www.jbc.org/

molecular and cellular character-istics of the original GnRH neuro-blasts, show normal DNA damagecheckpoint control (data notshown), and, most importantly,express endogenous FGF2 andFGFR1 IIIc (supplemental Fig.S1b). These FNC-B4-hTERT cells,therefore, represent an ideal sys-tem to functionally confirm ourbiochemical data and have beenused in Transwell migration as-says as described under “Experi-mental Procedures.”The FNC-B4-hTERT cells plated

in the upper chamber of the Tran-swell showed minimal migrationtoward theSFM.WhenPIWF4(1and10 nM)was loaded in the lower cham-ber, however, cells exhibited a signifi-cant 3- to 6-fold increase inmigration

FIGURE 7. Effect of anosmin-1 binding to FGFR1 on the FGF2�FGFR1�heparin complex assembly by SPR using sequential injections. a, 250 nM

FGFR1was first injected over a PIWF4-coupled sensor chip, followed by the second injection of FGF2 at various concentrations. b, FGFR1 at variousconcentrations was first injected with the second injection of 50 nM FGF2. c, 250 nM FGFR1 was first injected, followed by the second injection of 50 nM

FGF2, and the third injection of heparin at various concentrations. d, 250 nM FGFR1 was first injected over a PIWF4-coupled sensor chip, followed by theinjection of 10 �g/ml heparin and the preincubated FGF2 with heparin, respectively. single solid arrow, first start of injection; single dashed arrow, firststop of injection; double solid arrows, second start of injection; double dashed arrows, second stop of injection; triple solid arrows, third start of injection;and triple dashed arrows, third stop of injection.

FIGURE 8. Effect of anosmin-1 binding to heparin on FGF2�FGFR1�heparin complex assembly by SPRusing sequential injections. 10 �g/ml heparin was first injected over a PIWF4-coupled sensor chip, followedby the second injection of 50 nM FGF2, 50 nM FGFR1D1D3, and preincubated FGF2�FGFR1D1D3,FGF2�FGFR1D1D3�heparin at the indicated concentrations, respectively. single solid arrow, first start of injec-tion; single dashed arrow, first stop of injection; double solid arrows, second start of injection; double dashedarrows, second stop of injection; triple solid arrows, third start of injection; and triple dashed arrows, third stop ofinjection.




http://ww

w.jbc.org/

Dow

nloaded from



http://www.jbc.org/

(p�0.01), comparedwithSFM.Thedose-dependentchemotacticmigration induced by PIWF4 most likely involves FGFR1�FGF2signaling activity, because FGFR1 antagonist SU5402 could abol-ish it and exogenous FGF2 can trigger similar migration of thesecells (Fig. 9, a and b). In contrast, when a higher concentration (50nM) of PIWF4 was loaded, the cell migration was decreased to abasal level (Fig. 9b), suggesting that PIWF4 canmediate opposingeffects on FGFR1 signaling activity depending on the level ofexpression.To establish the role of HS in this phenomenon, cells were

treated with sodium chlorate, which prevents endogenousHS sulfation during biosynthesis, without affecting the

expression of FGF2 and FGFR1. As shown in Fig. 9c, HS-deficient cells no longer exhibited chemotactic migrationtoward PIWF4, and heparin alone did not enhancemigrationof these cells, indicating that the presence of both PIWF4and HS are required to induce migration. As expected, che-motactic migration was restored when PIWF4 and heparinwere loaded together in the lower chamber, in a heparindose-dependent manner. These observations are consistentwith the notion that the anosmin-1�HS complex may facili-tate the formation of a functional FGFR1 signaling complexon the surface of these cells, inducing FGFR1-mediated che-motactic migration.

FIGURE 9. Opposing effects of anosmin-1 on FGFR1-mediated cell migration. FNC-B4-hTERT cells are plated in the top chamber of Transwells and allowedto migrate for 4 h toward the lower chamber containing different chemoattractants (SFM, FGF2, PIWF4, and PIWF4 plus SU5402) at varying concentrations. Arepresentative microscopic image of migrated cells after staining under each condition is shown in a and the quantification of the migrated cells is shown inb. **, p � 0.01 compared with SFM control. c, sodium chlorate (SC)-treated cells were plated on the top chamber and similarly allowed to migrate toward SFM,heparin, or PIWF4 plus heparin at the indicated concentrations. The quantification of migrated cells is shown. **, p � 0.01; *, p � 0.05, in comparison to 10 nM

PIWF4 treatment alone. d, SC-treated cells were preincubated with SFM, PIWF4, or PIWF4 plus FGFR1D1D3 for 1 h and then resuspended in SFM before beingloaded in the top chamber of Transwell. A mixture of 10 nM PIWF4 and 200 ng/ml heparin was introduced into the lower chamber. The numbers of migratedcells after 20 h were quantified. **, p � 0.01 compared with PIWF4 preincubation. All values are shown as mean � S.D. calculated from at least two separateexperiments performed in duplicates.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

Next, we preincubated the HS-deficient cells with 10 nMPIWF4 to allow the binding of anosmin-1 to the endogenousFGFR1 on the cell surface, which may then further recruitendogenous FGF2 to formaPIWF4�FGF2�FGFR1 complex (Fig.7, a and b). This progression of complex formation is expectedto occur preferentially in these cells, because FGF2 cannot bindto anosmin-1 directly without first interacting with FGFR1(supplemental Fig. S2b). Then, after 1-h preincubation, the cellswere washed and allowed tomigrate toward the lower chambercontaining PIWF4 and heparin. The results showed that prein-cubation with PIWF4, but not SFM, led to a 50% reduction incell migration, and this inhibitory activity of PIWF4 could bereversed by addition of FGFR1 D1D3 recombinant proteinin the preincubation mixture, which competitively inhibitsPIWF4 binding to the FGFR1 on the cell surface (Fig. 9d). Thesedata are consistent with the notion that PIWF4-bound FGFR1cannot further engage FGF2 to form a functional FGF2�FGFR1�HS complex (Fig. 7, c and d), thus anosmin-1 blocks theactivation of FGFR1 signaling complex.Taken together, these findings enable us to conclude that the

order of interactions of anosmin-1 with FGFR1 and HS canresult in opposing cellular responses and themultiprotein bind-ing interactions observed in our biochemical studies also occurin physiologically relevant cell systems.

DISCUSSION

We previously reported that anosmin-1 could modulateFGFR1 signaling involving p42/44 and p38 mitogen-activatedprotein kinases and Cdc42/Rac1 activation, in the process ofneurite outgrowth and cytoskeletal rearrangements. In heterol-ogous BaF3 lymphoblast cells, anosmin-1 enhanced FGF2 sig-naling specifically through FGFR1 IIIc isoform in aHS-depend-ent manner (25). It has been speculated that anosmin-1 affectsFGFR1 signaling by regulating FGF2�FGFR1�heparin signalingcomplex formation. The present study provides the first dem-onstration of high affinity direct binding between anosmin-1and FGFR1; we further propose a molecular mechanismwhereby binding of anosmin-1 to FGFR1 and heparin exhibitsdifferent effects on ternary FGFR1�FGF2�heparin complexformation.Anosmin-1 binding to FGFR1 is characterized by rapid asso-

ciation followed by a very slow dissociation pattern, giving adissociation constant Kd of 7 nM. The preferred FGF receptorfor anosmin-1 binding is FGFR1, because amuch lower bindingaffinity was observed for FGFR2c, with negligible bindingfor FGFR3c, which makes the preferential binding order:FGFR1�FGFR2��FGFR3. The difference in FGFR bindingprovides the explanation for the specificity of anosmin-1 inBaF3 cell proliferation, in which anosmin-1 functions only oncells expressing FGFR1c but not on FGFR2c and FGFR3c iso-forms. This finding also brings important new insights into themechanism of anosmin-1 action; high affinity binding of anos-min-1 to the respective receptor is both necessary and sufficientto ensure anosmin-1 activity on the receptor. It should be notedthat, although SPR studies have the limitation that they arecarried out in the absence of extracellular matrix, we have alsodemonstrated here equivalent interactions in functional cellassays (Figs. 6 and 9), thus supporting the biochemical data.

Fibronectin type III domains inNCAM, L1 andXFLRT3 playimportant roles in FGFR1 binding. In anosmin-1, at least one ofthe FGFR1 binding sites is likely located in the first FnIIIdomain because the N-terminal fraction of anosmin-1 with thesingle first FnIII domain retains FGFR1 binding capacity,whereas the N267K mutation in the first FnIII domain, but notthe other two substitutions in the third FnIII domain, leads toloss of FGFR1 binding. According to the x-ray scattering andconstrained modeling, the four FnIII domains of anosmin-1have an elongated and flexible inter-domain arrangement (35);hence the dimensional orientation of first FnIII domain wouldnot be affected by the other three FnIII domains. As observed inthe homology modeling (36), N267K was expected to causeeither disruption of protein folding or the loss of activity, byintroducing an additional basic residue near a large basic sur-face.Notably,N-terminal truncated anosmin-1 containing onlythe four FnIII domains completely loses the capacity to bind toFGFR1IIIc and shows a 7-fold decreased binding affinity toFGFR2IIIc. These results raise the possibility that the N-termi-nal CR and WAP domains of anosmin-1 may assist the firstFnIII domain to adopt an optimal orientation for FGFR1 bind-ing. This idea is supported by the fact that the surface of theWAP domain is acidic in charge and a large basic patch is pres-ent on the first FnIII domain surface (36), which makes it pos-sible for these neighboring domains to engage in electrostaticinteractions. The functional interdependence between theCR/WAP domain and the FnIII repeats has been previouslydocumented by in vivo studies; in both Caenorhabditis elegansand Drosophila, it has been reported that the removal of theFnIII domains or the equivalent substitution of N267K causedanosmin-1 protein to completely lose its biological activity andcell adhesion property (37, 38), and the additional phenotypeinduced by the N267K equivalent substitution in Drosophilawas suppressed by aWAPdomainmutation (37).Moreover, wehave previously reported that PIWF1 is sufficient to induceneurite outgrowth in human olfactory GnRH neuroblast cul-ture, but a point mutation in the WAP domain (C172R) abol-ishes this activity (25).In terms of the FGFR, the acid box has currently been iden-

tified as a motif required for the interaction with N-cadherinand all the major NCAM isoforms (22). However, we haveshown in this study that FGFR1 lacking D1 and the acid boxbinds anosmin-1, with 3-fold higher affinity, indicating that theanosmin-1 binding motif is located within D2/D3 domains andthe linker between these domains. Even though these areas arealso required for FGF ligand binding, the binding motif forFGF2 and anosmin-1 does not seem to be completely overlap-ping because the SPR sequential injection demonstrated thatFGF2 is still capable of binding to the pre-formed anosmin-1�FGFR1 complex. It has been proposed that D1 and acid boxhave autoinhibition activity by potentially interfering with theligand�HS binding sites in D2 and D3. Deletion of D1 and acidbox is known to enhance the binding efficiency between FGFsand FGFR1 (39). This phenomenon was also observed withanosmin-1, suggesting that D1 and the acid box may exert amodulatory activity over anosmin-1 interactions with D2/D3.HSPG is essential for the stability of FGF�FGFR�HS ternary

complex (5). The initial binary complex formation is proposed




http://ww

w.jbc.org/

Dow

nloaded from


http://www.jbc.org/

to occur, either as FGF�FGFR or FGF�HS. In the first scenario,FGF�FGFR pairing is further stabilized by HS, as proposed inthe 2:2:2 FGF2�FGFR1c�heparin ternary complexmodel (40). Inthis case, the preformed FGF�FGFR complex seeks distinct sul-fation patterns ofHSof certain length,whichmay differ accord-ing to individual FGFs or FGFRs (41). In another scenario, HSfirst induces oligomerization of FGF molecules and subse-quently enables FGFR dimerization, as proposed in the 2:2:1FGF1�FGFR2c�heparin ternary complex model (9). These twobinary FGF�FGFR and FGF�HS complexes may co-exist as thedriving force for ternary signaling complex assembly, which canbe regulated by the different length and sulfation patterns ofheparin�HS saccharides (11). Herewe have shown that heparin-bound anosmin-1 facilitates FGF2�FGFR1 as the preferredfundamental binary complex, resulting in anosmin-1�FGF2�FGFR1�heparin complex formation.Wehave also observed thatshort heparin fragment-bound anosmin-1 loses binding capa-

bility with FGF2�FGFR1,6 suggest-ing that the modulatory effect ofanosmin-1 on ternary signalingcomplexformationisheparinlength-dependent. In contrast, anosmin-1-bound FGFR1 cannot interact witha FGF2�heparin complex. Althoughanosmin-1-bound FGFR1 can stillbind to FGF2, this is insufficientlystable for subsequent FGF2�FGFR1�heparin complex generation,because heparin rapidly dissociatesFGF2 from it. Therefore, we envis-age that anosmin-1 can play differ-ential roles during ternary FGF2�FGFR1�heparin complex assemblydepending onwhether it is bound toFGFR1 or heparin. These mecha-nisms, depicted in Fig. 10, are noveland distinct from the molecularmechanism used by other FGFR sig-nalingmodulators known so far. Forexample, it has been proposed thatcell adhesion molecules clusterupon homophilic binding, to induceco-clustering of the FGFR, leadingto receptor activation and biologicalactivity (42). Antiangiogenic formsof antithrombin block FGF2-in-duced angiogenesis by competingwith FGF2 for binding to heparin toprevent FGF2�FGFR1�heparin com-plex formation (43).Anosmin-1 and FGFR1 are

known to be involved in GnRHneu-ron migration and neurite out-growth (25, 34, 44, 45). We nowreport that anosmin-1 can functionas a chemoattractant inducingmigration of immortalized FNC-B4-hTERT cells via an FGFR1- and

HS-dependent mechanism (Fig. 9). HS-bound anosmin-1 willpreferentially associate with pre-existing FGF2�FGFR1 pairs tofacilitate FGF2�FGFR1�HS signaling complex formation on tar-get cells, enabling the FGFR1-mediated cellmigration as shownhere. In this situation, the role of anosmin-1 could be to presentthe appropriate HS to the complex, and if so, the endogenousHS context will be crucial in determining the anosmin-1-medi-ated responses. However, under the conditions where anos-min-1 level is high, HS-unbound anosmin-1 could diffuse freelyand bind to the FGF receptor on the host and neighboring cells,subsequently resulting in inhibitory effect on FGFR signalingcomplex formation. We have tested this scenario in our Tran-swell migration assays where endogenous HS synthesis is arti-ficially blocked to create an extracellular matrix environment

6 Y. Hu, S. E. Guimond, P. Travers, S. Cadman, E. Hohenester, J. E. Turnbull, S.-H.Kim, and P.-M. Bouloux, unpublished observations.

FIGURE 10. Putative model for the dual role of anosmin-1 on FGF2�FGFR1�HS complex formation. a, adiagram of model whereby anosmin-1 binding to FGFR1 inhibits FGF2�FGFR1�HS complex formation. FGF2 isable to bind to pre-formed anosmin-1�FGFR1 complex to form anosmin-1�FGF2�FGFR1 complex, but HS canstrip FGF2 off this ternary complex. Binary complex of FGF2�HS cannot bind to anosmin-1-bound FGFR1. Thus,FGF2�FGFR1�HS signaling complex would not form. b, a schematic representation of the manner in whichanosmin-1 binding to HS is capable of facilitating FGF2�FGFR1�HS complex formation. HS-bound anosmin-1preferentially binds to pre-formed FGF2�FGFR1 pair, resulting in anosmin-1�FGF2�FGFR1�HS complex forma-tion. The domain structure of anosmin-1 is based on the model proposed in Ref. 35. The cell membrane isshown as a double line. CR, cysteine-rich region; WAP, whey acidic protein-like domain; and I–IV, fibronectin-liketype III 1– 4.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

where anosmin-1 is mostly free (HS-unbound) and allowed tobind to FGFR1 on the cell surface. Our results confirmed thatanosmin-1 can inhibit FGFR1-mediated cell migration underthese conditions. Notably, anosmin-1 has been previouslyreported to inhibit the FGFR1�FGF2-induced cell migration inmouse oligodendrocyte precursor cells (24); however, themolecular mechanism for this phenomenon was hithertounknown.The diffusible characteristic of anosmin-1 and its dual activ-

ities on FGFR1 signaling complexmake it likely that anosmin-1will act as a guidance cue with combined chemoattractantand chemorepellant properties to direct cell migration andaxon targeting. This idea is supported by our current studieswhere anosmin-1 promotes GnRH neuroblast migration atlower concentration but inhibits at higher concentration.The binding capacity of HSPG and the expression levels ofanosmin-1 may determine these opposing activities by shift-ing anosmin-1 either into an FGFR1-bound or an HS-boundstate. This may provide one explanation as to why FGFR1mutations cause KS with a wide spectrum of reproductivephenotype (46). Confirming how individual FGFR1 muta-tions affect interaction with anosmin-1 may also offer fur-ther clues. Our current study showed that the two KS-relatedmutations of anosmin-1, E514K and F517L, resulted indecreased binding affinity to FGFR1, which in combinationwith the potentially altered protein stability/expression in vivomay contribute to the phenotype of KS. FGF8 has been cur-rently identified as one of six KS genes (47, 48). It is reasonableto assume that anosmin-1 can regulate FGF8�FGFR1�HS signal-ing complex formation in a manner similar to that proposedhere.

Acknowledgments—Weare grateful for RubyQuartey-Papafio for herkind help with SPR experiments. We are thankful for Dr. ShaneMinogue at the Centre for imaging, University College London Med-ical School, Royal Free Campus, for his expert advice in the use ofconfocal microscope and immunofluorescence. We are very gratefulfor the generous donation of the recombinant FGFR2 and FGFR3ectodomain proteins by Alan Brown (University of Cambridge). Wealso thank Dr. Federico Carafoli (Imperial College London) for theFGFR1 D1–D3 and D2–D3 expression constructs and technicalassistance in protein purification. The retroviral vectors used in cellimmortalization were generous gifts from Dr. Gordon Peters (CancerResearch UK, London Research Institute). Dr. David Ornitz (Wash-ington University, St. Louis, MO) has kindly provided the FGFR1 IIIccDNA construct.

REFERENCES1. Detillieux, K. A., Sheikh, F., Kardami, E., and Cattini, P. A. (2003) Cardio-

vasc. Res. 57, 8–192. Freeman, K. W., Gangula, R. D., Welm, B. E., Ozen, M., Foster, B. A.,

Rosen, J. M., Ittmann, M., Greenberg, N. M., and Spencer, D. M. (2003)Cancer Res. 63, 6237–6243

3. Ornitz, D. M. (2005) Cytokine Growth Factor Rev. 16, 205–2134. Eswarakumar, V. P., Lax, I., and Schlessinger, J. (2005) Cytokine Growth

Factor Rev. 16, 139–1495. Mohammadi,M.,Olsen, S. K., and Ibrahimi,O. A. (2005)CytokineGrowth

Factor Rev. 16, 107–1376. Plotnikov, A. N., Schlessinger, J., Hubbard, S. R., and Mohammadi, M.

(1999) Cell 98, 641–650

7. Plotnikov, A. N., Hubbard, S. R., Schlessinger, J., and Mohammadi, M.(2000) Cell 101, 413–424

8. Kan, M., Wang, F., Xu, J., Crabb, J. W., Hou, J., and McKeehan, W. L.(1993) Science 259, 1918–1921

9. Pellegrini, L., Burke, D. F., von, Delft, F., Mulloy, B., and Blundell, T. L.(2000) Nature 407, 1029–1034

10. Schlessinger, J., Plotnikov, A. N., Ibrahimi, O. A., Eliseenkova, A. V., Yeh,B. K., Yayon, A., Linhardt, R. J., and Mohammadi, M. (2000) Mol. Cell 6,743–750

11. Goodger, S. J., Robinson, C. J., Murphy, K. J., Gasiunas, N., Harmer, N. J.,Blundell, T. L., Pye, D. A., and Gallagher, J. T. (2008) J. Biol. Chem. 283,13001–13008

12. Harmer, N. J., Ilag, L. L., Mulloy, B., Pellegrini, L., Robinson, C. V., andBlundell, T. L. (2004) J. Mol. Biol. 339, 821–834

13. Kim, S. H., Hu, Y., Cadman, S., and Bouloux, P. (2008) J. Neuroendocrinol.20, 141–163

14. Miki, T., Fleming, T. P., Bottaro, D. P., Rubin, J. S., Ron, D., and Aaronson,S. A. (1991) Science 251, 72–75

15. Guimond, S. E., and Turnbull, J. E. (1999) Curr. Biol. 9, 1343–134616. Ostrovsky, O., Berman, B., Gallagher, J., Mulloy, B., Fernig, D. G., Dele-

hedde, M., and Ron, D. (2002) J. Biol. Chem. 277, 2444–245317. Bottcher, R. T., Pollet, N., Delius, H., and Niehrs, C. (2004) Nat. Cell Biol.

6, 38–4418. Tsang, M., Friesel, R., Kudoh, T., and Dawid, I. B. (2002) Nat. Cell Biol. 4,

165–16919. Kiselyov, V. V., Skladchikova, G., Hinsby, A.M., Jensen, P. H., Kulahin, N.,

Soroka, V., Pedersen, N., Tsetlin, V., Poulsen, F. M., Berezin, V., and Bock,E. (2003) Structure 11, 691–701

20. Kulahin, N., Li, S., Hinsby, A., Kiselyov, V., Berezin, V., and Bock, E. (2008)Mol. Cell Neurosci. 37, 528–536

21. Saffell, J. L.,Williams, E. J., Mason, I. J.,Walsh, F. S., andDoherty, P. (1997)Neuron 18, 231–242

22. Sanchez-Heras, E., Howell, F. V., Williams, G., and Doherty, P. (2006)J. Biol. Chem. 281, 35208–35216

23. Christensen, C., Lauridsen, J. B., Berezin, V., Bock, E., and Kiselyov, V. V.(2006) FEBS Lett. 580, 3386–3390

24. Bribian, A., Barallobre, M. J., Soussi-Yanicostas, N., and de Castro, F.(2006)Mol. Cell Neurosci. 33, 2–14

25. Gonzalez-Martínez, D., Kim, S. H., Hu, Y., Guimond, S., Schofield, J.,Winyard, P., Vannelli, G. B., Turnbull, J., and Bouloux, P.M. (2004) J. Neu-rosci. 24, 10384–10392

26. Dode, C., Levilliers, J., Dupont, J. M., De, Paepe, A., Le, Du, N., Soussi-Yanicostas, N., Coimbra, R. S., Delmaghani, S., Compain-Nouaille, S., Ba-verel, F., Pecheux, C., Le, Tessier, D., Cruaud, C., Delpech, M., Speleman,F., Vermeulen, S., Amalfitano, A., Bachelot, Y., Bouchard, P., Cabrol, S.,Carel, J. C., Delemarre-van de,Waal, H., Goulet-Salmon, B., Kottler,M. L.,Richard, O., Sanchez-Franco, F., Saura, R., Young, J., Petit, C., and Hard-elin, J. P. (2003) Nat. Genet. 33, 463–465

27. Franco, B., Guioli, S., Pragliola, A., Incerti, B., Bardoni, B., Tonlorenzi, R.,Carrozzo, R., Maestrini, E., Pieretti, M., Taillon-Miller, P., Brown, C.,Wil-lard, H., Lawrence, C., Persico, G., Camerino, G., and Ballabio, A. (1991)Nature 353, 529–536

28. Legouis, R., Hardelin, J. P., Levilliers, J., Claverie, J. M., Compain, S.,Wunderle, V., Millasseau, P., Le Paslier, D., Cohen, D., Caterina, D., Bou-gueleret, L., Delemarre-Van deWaal, H., Lutfalla, G.,Weissenbach, J., andPetit, C. (1991) Cell 67, 423–435

29. Hu, Y., Gonzalez-Martínez, D., Kim, S. H., and Bouloux, P. M. (2004)Biochem. J. 384, 495–505

30. Carafoli, F., Saffell, J. L., and Hohenester, E. (2008) J. Mol. Biol. 377,524–534

31. Vannelli, G. B., Ensoli, F., Zonefrati, R., Kubota, Y., Arcangeli, A., Bec-chetti, A., Camici, G., Barni, T., Thiele, C. J., and Balboni, G. C. (1995)J. Neurosci. 15, 4382–4394

32. Pitteloud, N., Quinton, R., Pearce, S., Raivio, T., Acierno, J., Dwyer, A.,Plummer, L., Hughes, V., Seminara, S., Cheng, Y. Z., Li,W. P.,Maccoll, G.,




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

Eliseenkova, A. V., Olsen, S. K., Ibrahimi, O. A., Hayes, F. J., Boepple, P.,Hall, J. E., Bouloux, P., Mohammadi, M., and Crowley, W. (2007) J. Clin.Invest. 117, 457–463

33. Sandilands, E., Akbarzadeh, S., Vecchione, A., McEwan, D. G., Frame,M. C., and Heath, J. K. (2007) EMBO Rep. 8, 1162–1169

34. Cariboni, A., Pimpinelli, F., Colamarino, S., Zaninetti, R., Piccolella, M.,Rumio, C., Piva, F., Rugarli, E. I., and Maggi, R. (2004) Hum. Mol. Genet.13, 2781–2791

35. Hu, Y., Sun, Z., Eaton, J. T., Bouloux, P. M., and Perkins, S. J. (2005) J. Mol.Biol. 350, 553–570

36. Robertson, A., MacColl, G. S., Nash, J. A., Boehm,M. K., Perkins, S. J., andBouloux, P. M. (2001) Biochem. J. 357, 647–659

37. Andrenacci, D., Grimaldi, M. R., Panetta, V., Riano, E., Rugarli, E. I., andGraziani, F. (2006) BMC. Genet. 7, 47

38. Bulow, H. E., Berry, K. L., Topper, L. H., Peles, E., and Hobert, O. (2002)Proc. Natl. Acad. Sci. U.S.A. 99, 6346–6351

39. Olsen, S. K., Ibrahimi, O. A., Raucci, A., Zhang, F., Eliseenkova, A. V.,Yayon, A., Basilico, C., Linhardt, R. J., Schlessinger, J., and Mohammadi,M. (2004) Proc. Natl. Acad. Sci. U.S.A. 101, 935–940

40. Mohammadi, M., Olsen, S. K., and Goetz, R. (2005) Curr. Opin. Struct.Biol. 15, 506–516

41. Allen, B. L., and Rapraeger, A. C. (2003) J. Cell Biol. 163, 637–64842. Doherty, P.,Williams, G., andWilliams, E. J. (2000)Mol. Cell Neurosci. 16,

283–29543. Zhang, W., Swanson, R., Xiong, Y., Richard, B., and Olson, S. T. (2006)

J. Biol. Chem. 281, 37302–3731044. Gill, J. C., Moenter, S. M., and Tsai, P. S. (2004) Endocrinology 145,

3830–383945. Tsai, P. S., Moenter, S. M., Postigo, H. R., El, Majdoubi, M., Pak, T. R., Gill,

J. C., Paruthiyil, S., Werner, S., and Weiner, R. I. (2005) Mol. Endocrinol.19, 225–236

46. Pitteloud, N., Meysing, A., Quinton, R., Acierno, J. S., Jr., Dwyer, A. A.,Plummer, L., Fliers, E., Boepple, P., Hayes, F., Seminara, S., Hughes, V. A.,Ma, J., Bouloux, P., Mohammadi, M., and Crowley, W. F., Jr. (2006) Mol.Cell. Endocrinol. 254–255, 60–69

47. Chung, W. C., Moyle, S. S., and Tsai, P. S. (2008) Endocrinology 149,4997–5003

48. Hardelin, J. P., and Dode, C. (2008) Sex Dev. 2, 181–193




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

E. Turnbull, Soo-Hyun Kim and Pierre-Marc BoulouxYouli Hu, Scott E. Guimond, Paul Travers, Steven Cadman, Erhard Hohenester, Jeremy

Extracellular Matrix Protein Anosmin-1Novel Mechanisms of Fibroblast Growth Factor Receptor 1 Regulation by

doi: 10.1074/jbc.M109.049155 originally published online August 20, 20092009, 284:29905-29920.J. Biol. Chem.

10.1074/jbc.M109.049155Access the most updated version of this article at doi:

Alerts:

When a correction for this article is posted•

When this article is cited•

to choose from all of JBC's e-mail alertsClick here

Supplemental material:

http://www.jbc.org/content/suppl/2009/08/20/M109.049155.DC1

http://www.jbc.org/content/284/43/29905.full.html#ref-list-1

This article cites 48 references, 12 of which can be accessed free at


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/lookup/doi/10.1074/jbc.M109.049155

http://www.jbc.org/cgi/alerts?alertType=citedby&addAlert=cited_by&cited_by_criteria_resid=jbc;284/43/29905&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/284/43/29905

http://www.jbc.org/cgi/alerts?alertType=correction&addAlert=correction&correction_criteria_value=284/43/29905&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/284/43/29905

http://www.jbc.org/cgi/alerts/etoc

http://www.jbc.org/content/suppl/2009/08/20/M109.049155.DC1

http://www.jbc.org/content/284/43/29905.full.html#ref-list-1

http://www.jbc.org/

NovelMechanismsofFibroblastGrowthFactorReceptor1 ... · 2009-10-10 · unknown. Here, we show that...

Documents

Transcript of NovelMechanismsofFibroblastGrowthFactorReceptor1 ... · 2009-10-10 · unknown. Here, we show that...