FGFR3 intracellular mutations induce tyrosine phosphorylation in the Golgi and defective...

1773 (2007) 502–512www.elsevier.com/locate/bbamcr

Biochimica et Biophysica Acta

FGFR3 intracellular mutations induce tyrosine phosphorylation in the Golgiand defective glycosylation

Linda Gibbs ⁎, Laurence Legeai-Mallet

INSERM U781, Hôpital des Enfants Malades, 149 rue de Sèvres-75015 Paris, France

Received 22 August 2006; received in revised form 8 December 2006; accepted 20 December 2006Available online 20 January 2007

Abstract

Mutations of the Fibroblast Growth Factor Receptor 3 (FGFR3) gene have been implicated in a series of skeletal dysplasias includinghypochondroplasia, achondroplasia and thanatophoric dysplasia. The severity of these diseases ranges from mild dwarfism to severe dwarfism andto perinatal lethality, respectively. Although it is considered that the mutations give rise to constitutively active receptors, it remains unclear howthe different mutations are functionally linked to the severity of the different pathologies. By examining various FGFR3 mutations in a HEK cellculture model, including the uncharacterized X807R mutation, it was found that only the mutations affecting the intracellular domain, inducedpremature receptor phosphorylation and inhibited receptor glycosylation, suggesting that premature receptor tyrosine phosphorylation of the nativereceptor inhibits its glycosylation. Moreover, these mutations appeared to be associated with elevated receptor signaling in the Golgi apparatus. Inconclusion, although pathological severity could not be correlated with a single factor arising from FGFR3 mutations, these results suggest thatintracellular domain mutations define a distinct means by which mutated FGFR3 could disrupt bone development.© 2007 Elsevier B.V. All rights reserved.

Keywords: Chondrodysplasia; Receptor; Kinase; Signaling; Trafficking

1. Introduction

Fibroblast Growth Factor Receptor 3 (FGFR3) belongs to afamily of four genes (FGFR1–4) coding for receptor tyrosinekinases [1]. These proteins constitute a family of structurallyrelated proteins exhibiting an extracellular domain composed ofthree immunoglobulin-like domains, an acid box, a single trans-membrane domain and a split tyrosine kinase (TK) domain [1].

FGF receptor dimerization and phosphorylation is inducedthrough the binding of its ligand (FGF) in the presence ofsoluble or cell-surface heparan sulfate proteoglycans [2–6]. Thedimer state is stabilized by a conformational change in theextracellular domain [7]. As a result the two monomers comeinto contact. This stimulates catalytic activity resulting intransphosphorylation of key tyrosine residues in the cytoplas-

Abbreviations: ACH, achondroplasia; BFA, Brefeldin A; ER, endoplasmicreticulum; FGFR, Fibroblast Growth Factor Receptor; HCH, hypochondropla-sia; TDI and TDII, thanatophoric dysplasia types I and II; TK, tyrosine kinase;WT, wild type⁎ Corresponding author.E-mail address: [email protected] (L. Gibbs).

0167-4889/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.bbamcr.2006.12.010

mic domain of the receptor that serve as docking sites for theadaptor proteins and effectors that propagate FGFR3 signals viadifferent signaling cascades [8,9].

FGF signaling through FGFR3 is implicated in the regulationof many cellular processes including proliferation, differentia-tion, migration, wound healing, and angiogenesis [10,11].Mutations in FGFR3 have been shown to be responsible for avariety of skeletal dysplasias [12] of increasing severity,including achondroplasia (ACH) which is the most commonform of human dwarfism [13,14], the milder disorder hypochon-droplasia (HCH) [15,16] and the more severe thanatophoricdysplasia types I and II (TD I and TDII) [13,17,18]. Thethanatophoric dysplasia subdivision arises from the presentationof curved femurs in TDI but not in TDII. TDIs caused by theK650M mutation can also be associated with Severe Achon-droplasia with Developmental Delay and Acanthosis Nigricans(SADDAN) [19]. These FGFR3 mutations appear to arise fromspontaneous missense neo-mutations [20].

The molecular basis underlying the various chondrodyspla-sias is thought to be a consequence of the constitutive activationof mutant receptors [21–23]. These dominant active mutations

mailto:[email protected]

http://dx.doi.org/10.1016/j.bbamcr.2006.12.010

503L. Gibbs, L. Legeai-Mallet / Biochimica et Biophysica Acta 1773 (2007) 502–512

in FGFR3 inhibit endochondral bone growth and cause ACH or,TD in murine transgenic models [24–26].

A number of mechanisms have been proposed to explain howthe mutations affect FGFR3 activation. These include increasedligand-binding affinity or decreased ligand-mediated receptordown-regulation in the case of G380R [27] and stabilized ligand-induced dimers in the case of R248C (TDI) [22,28]. Activatingmutations in the TK domain, such as is the case of the K650N,K650M and K650E mutations (corresponding to HCH, TDI andTDII, respectively), lead to ligand-independent receptor activa-tion and constitutive receptor tyrosine phosphorylation [22].More recently, two mutations in the murine equivalent ofK650E have been found to inhibit FGFR3 trafficking to theplasma membrane from the endoplasmic reticulum [29,30].However it remains unclear how these mechanistic aberrationsrelate to disease severity.

The aim of this study was to compare two attributes of mutantreceptor biochemistry (tyrosine phosphorylation and receptorglycosylation), and the pathology with which the mutation hasbeen associated. The study analyzed wild type and sevenmutated FGFR3s including the uncharacterized X807R muta-tion [31], in a HEK cell culture model. Even though all themutated receptors were found to have an elevated level ofphosphorylated tyrosine, these levels did not appear to correlatewith disease severity. Tyrosine phosphorylation was identifiedon native and partially glycosylated isoforms for the K650N,K650M, K650E and X807R mutations and appeared to disruptglycosylation. However, the degree of defective glycosylationdid not correlate with disease severity. The three K650mutationsand the X807R mutation elicited a high degree of intracellulartyrosine phosphorylation primarily associated with Rab6-positive structures suggesting the existence of ectopicallyactivated signaling cascades associated with the Golgi appara-tus. The study therefore finds that disease severity ofchondrodysplasias cannot solely be attributed to receptoractivation or receptor activation at the plasma membrane.

2. Materials and methods

2.1. Generation of human FGFR3 constructs

Full length human wild type FGFR3 cDNAwas kindly provided by Dr. M.Hayman (New York). Single point mutations were generated by site-directedmutagenesis (Quikchange® Site-Directed Mutagenesis, Stratagene). Enzymeswere purchased from New England Biolabs, Roche Diagnostics and Invitrogen.Reverse transcriptase-PCR (First Strand Synthesis, Roche Diagnostics) wasperformed on patient RNA using primer sets R248C/Y373C-P1/R248C/Y373C-P2 and G380R-P1/G380R-P2. Site-directed mutagenesis for K650N, K650Mand K650E and X807Rwas performed on wild type FGFR3. All constructs werecloned into pCDNA3.1 (Invitrogen, Europe). PCR and site-directed mutagenesisprimers are listed in Supplementary Table 1. The presence of mutations wasconfirmed by sequencing.

2.2. Western blotting (WB) and immunoprecipitation (IP)

Transfected cells were lysed in RIPA Buffer (50 mM Tris–HCl pH 7.6,150mMNaCl, 1%Nonidet-P40, 0.5% sodium deoxycholate, supplemented withprotease and phosphatase inhibitors). ForWestern blotting, 20 μg of cell lysate inRIPA buffer were used. Immunoprecipitation was performed by incubating 1 μgrabbit anti-FGFR3 (polyclonal against the cytoplasmic domain)/100 μg protein

with protein G agarose (Roche). Immunoblotting: anti-FGFR3 (Sigma) or mouseanti-phosphotyrosine (Cell Signaling) were used at a concentration of 1:2000;secondary antibodies anti-rabbit or anti-mouse HRP (Amersham) were used at aconcentration of 1:10000 and were detected by chemiluminescence (ECL,Amersham Pharmacia Biotech). Where necessary, PVDF membranes werestripped in 2% SDS, 100 mM β-mercaptoethanol, 50 mM Tris, pH 6.8.

2.3. Glycosidase treatments

Anti-FGFR3 immunoprecipitates were resuspended in 50mM sodium citrate(pH 5.5), 1% SDS, 1% β-mercaptoethanol. The sample was dividedproportionally for untreated or EndoH (endoglycosidaseH, Roche) and PNgaseF(peptide: N-glycosidase F, Roche) treatments, and used according to the man-ufacturer's instructions.

2.4. Cell culture, transfections and treatments

Human Embryonic Kidney cells (HEK) stably expressing the vitronectinreceptor (293 VnR) were provided by Dr. R. Baron (Yale University). Cells wereincubated in DMEM supplemented with 10% FCS (Invitrogen), and antibiotics.Transfection of constructs into cells at 50–80% confluency using Fugene6(Roche)was performed according to themanufacturer's instructions. BrefeldinA(BFA) treatment: 5 μg/ml for 2 h at 37 °C (Epicentre Technologies; TEBUFrance). Nocodazole treatment: 10 μg/ml 2 h at 37 °C (BIOMOL; TEBU,France). The tyrosine kinase inhibitor Sugen5402 (gift from G. Mc Mahon,SUGEN) was dissolved in DMSO and used at 25 μM overnight. Control cellswere incubated with DMSO alone. Tunicamycin treatment (5μg/ml; SigmaAldrich) was overnight in DMEM supplemented with 10% FCS.

2.5. Immunofluorescence

Transfected cells were fixed in 4% PFA for 20 min, permeabilized in 0.1%TritonX-100/PBS and blocked in 10% normal sheep serum. Incubation withprimary antibodies; anti-FGFR3 at 1:400; anti-phosphotyrosine at 1:200; rabbitanti-calnexin at 1:100, (Sigma); mouse anti-GM130 at 1:100, (BD TransductionLaboratories); rabbit anti-rab6, at 1:100 (Tebu, France); rabbit anti-Sec31, at1:100 (a gift from Prof. W. Hong), was for 1 h at room temperature. Incubationwith secondary antibodies Alexa488 and Alexa568 at 1:400 (Molecular Probes)was for 1 h at room temperature. For anti-Rab6; after fixing in PFA, cells werepre-incubated in 50 mM NH4Cl/PBS for 10 min, with antibody incubation in0.05% saponin and 0.2% normal sheep serum in PBS.

3. Image analysis and statistics

Experiments were performed on three or more independentoccasions. The quantification ofWestern blotting was performedusing NIH Image. Statistical differences were identified byapplication of the Student t test. The variations from the meanvalues shown are equal to the Standard Deviations from theserespective means. Immunofluorescence images were selected asgood representatives of the experimental condition.

4. Results

4.1. Glycosylation of FGFR is affected by the K650 mutations

The FGFR3mutations associated with bone growth disordersare found in different domains and depicted in Fig. 1A. Theseinclude the intracellular domain (K650N/HCH, K650M/TDI,K650E/TDII, X807R/TDI), the transmembrane domain(G380R/ACH) and the extracellular domain (R248C/TDI,Y373C/TDI). The mutation arising from the stop codon readthrough (X807R/TDI) is predicted to generate a protein the size

Fig. 1. A biochemical comparison of the glycosylation and tyrosine phosphorylation of the WT and mutant FGFR3s. (A) Schematic depiction of FGFR3 proteinincluding the Ig-like domains (Ig1–3) of the extracellular domain, the transmembrane domain (TM) and the split tyrosine kinase domains (TK1 and TK2) of theintracellular domain. Also shown is the position of the mutations examined in this study. (B) FGFR3s for WT, G380R, R248C, Y373C, K650N, K650M and K650E allappeared as three discrete bands of differing mobilities (105, 115, 130 kDa) by Western blot, as a consequence of glycosylation. EndoH treatment removed mannose-rich moieties, and PNGase treatment removed all glycosylated moieties from FGFR3. (C) Tunicamycin treatment inhibited glycosylation of native WT and K650MFGFR3 proteins. (D) The levels of the native and differently glycosylated WT, G380R, R248C, Y373C, K650N, K650M and K650E FGFR3 isoforms were examinedat 24 and 48 h by Western blot (upper panels). The relative levels of the native and differently glycosylated isoforms from at least five experiments were quantified at24 h (lower graphics). The asterisks mark the significantly lower relative levels of the fully glycosylated isoforms compared with that of the WT (*p<0.01;**p<0.001). (E) The detection of tyrosine phosphorylation on the FGFR3s. WT, G380R, R248C, Y373C, K650N, K650M and K650E FGFR3s wereimmunoprecipitated by anti-FGFR3 and examined by Western blot with anti-FGFR3 or anti-phosphotyrosine. (F) WT and K650M FGFR3s were treated withSugen5402 (tyrosine kinase inhibitor) and examined by Western blotting of anti-FGFR3 immunoprecipitations. Sugen5402 treatment reduced tyrosinephosphorylation and increased the relative level of the fully-glycosylated isoform of the K650M mutant.

504 L. Gibbs, L. Legeai-Mallet / Biochimica et Biophysica Acta 1773 (2007) 502–512

of FGFR3 plus 141 amino acid residues. The consequences ofthese mutations were analyzed in HEK cells by transient trans-fection of FGFR3 expression plasmids.

Twenty-four hours post-transfection, FGFR3 protein withthree mobilities was detected byWestern blot analysis of FGFR3immunoprecipitates from transfected cell extracts (Fig. 1B).Glycosidase treatments of the protein extracts revealed the wildtype and each of the mutant FGFR3 proteins (R248C, G380R,Y373C K650N, K650M, K650E) exist as an unglycosylatedform (highest mobility, 105 kDa), an EndoH glycosidase-sensitive form (intermediate mobility; 115 kDa) and an EndoHinsensitive but PNGase-sensitive form (lowest mobility,130 kDa). This suggested that FGFR3was subject to progressiveglycosylation from native to fully mature receptor. This wasconfirmed by treating the WT and K650M FGFR3 transfectedcells with the N-glycosylation inhibitor tunicamycin. Theinhibition of glycosylation reduced the level of the 115 kDaand 130 kDa isoforms compared with the untreated controls,24 h post-transfection (Fig. 1C).

By quantifying the levels of the different FGFR3 glycosy-lated isoforms at 24 and 48 h, it was found that the relative levelsof the non-glycosylated receptors had declined and the fullyglycosylated receptors had increased at 48 h compared with 24 hpost transfection (p<0.05; Fig. 1D). The proportional level of

fully glycosylated receptors found with the WT receptor at 24 h(38±8%, n=15) was similar to that found with R248C (35±16%, n=8), G380R (36±12%, n=5), Y373C (37±11%, n=6)mutations. However the proportional level of fully glycosylatedreceptors found with K650N (24±9%, n=5), K650M (7±4%,n=10) and K650E (4±4%, n=10) was significantly less thanthat found with the WT, whereas the proportional level of non-glycosylated K650M (31±8%) and K650E (33±15%) recep-tors, was significantly greater than that found with the WT (12±7%). This suggested that mutations associated with amino acidK650 of the intracellular domain inhibited glycosylation and themutations that occur in the transmembrane or extracellular do-mains do not inhibit glycosylation.

4.2. Receptor phosphorylation is greater for FGFR3 withintracellular mutations

To examine the relationship between glycosylation andreceptor tyrosine phosphorylation, anti-FGFR3 immunopreci-pitates of transfected cells were probed with a phosphotyrosineantibody on Western blots (Fig. 1E). A low level of phos-photyrosine was detected on the wild type receptor. Amarginallygreater level of tyrosine phosphorylated FGFR3 was found withthe G380R, R248C and Y373C mutants. The level of tyrosine


phosphorylated FGFR3 was greater again with the K650 mu-tants, where K650M>K650E>K650N. Moreover phosphotyr-osine was detected primarily on the fully glycosylated isoformsof the WT, G380R, R248C, Y373C FGFRs whereas phos-photyrosine was also detected on the non- and partially glyco-sylated K650 mutant isoforms. These results suggested thattyrosine phosphorylation of the K650 mutants in their non-or partially-glycosylated forms, inhibits their further glyco-sylation. This is especially so for K650M and K650E, andless marked for K650N. This suggestion was supported by ana-lyzing cells transfected with either the WT or K650M mutantFGFR3s following treatment with the tyrosine kinase inhibi-tor Sugen5402 (Fig. 1F). Following Western blot of FGFR3immunoprecipitates, it was found that the level of tyro-sine phosphorylated FGFR3 was strongly reduced, althoughsome tyrosine phosphorylated FGFR3 was detected with theK650M mutant (Fig. 1F). The reduction of tyrosine phos-phorylation of the K650Mmutant FGFR3 was associated with alarge relative increase in the level of fully glycosylated iso-form. Such a relative increase in the level of fully glycosylatedisoform was not detected in cells transfected with the wild typereceptor.

The degree of glycosylation and of tyrosine phosphorylationof the FGFR3 mutants were compared with pathology severity(Table 1). Pathology severity was ranked in accordance to whathas been described for the long bone dysplasia, placing the TDIand TDII with equal but most severe rank [32,33]. It was foundthat there was no simple or significant correlation between thesefactors and the severity of the disease pathology associated withthe different mutations. However the analysis suggested that theK650 mutations function differently to those mutationsaffecting the transmembrane or extracellular domains wherethe pathology severity is positively correlated with the degree ofdisrupted receptor glycosylation (r2=0.97).

4.3. The X807R mutation affects FGFR3 glycosylation

The FGFR3 stop codon read-through mutation X807R, wasidentified in the TDI pathology but its functional relationship tothe pathology is not known. Three FGFR3 isoforms of differingmobilities were detected in cells transfected with the X807R

Table 1No statistically significant correlation between bone pathologies from which theFGFR3 mutations arise, and glycosylation processing defects or the level oftyrosine phosphorylation

Mutant Pathology Pathology rank pTyrosineFGFR rank

Glycosylationdefect rank

WT none 1 1 1K650N HCH 2 2 2G380R ACH 3 1 1R248C TDI 4 1 1Y373C TDI 4 1 1K650M TDI 4 4 3K650E TDII 4 3 3Correlation coefficient with pathology, r2 0.13 0.11

The table lists the ranking scores for each parameter used, prior to determiningthe correlation coefficient r2.

mutant (Fig. 2B). The X807R mutant was found to exist asunglycosylated and glycosylated forms. The intermediatemobility isoform was both PNGase and EndoH sensitive andthe lowest mobility isoform was PNGase sensitive but endoHinsensitive. The size of the X807R isoforms suggested that theyall included the predicted 141 amino residue read-through. Byquantifying the levels of the different isoforms 24 h post-transfection, it was found that the relative level of fullyglycosylated isoform (12±9%, n=8) was similar to what wasfound with the K650M and K650E mutants and significantlylower thanwhat was foundwith theWTreceptor (Fig. 2C and seeFig. 1D). Also consistent with an inhibition of glycosylation, therelative level of non-glycosylated X807R mutant receptor (49±15%) was significantly greater than that found with the WT. TheX807R mutant was also found to be tyrosine phosphorylated onthe non- and partially-glycosylated isoforms (Fig. 2D). Thereforethis suggested that the X807R confers a constitutively activestatus to FGFR3. Moreover X807R was found have similarcharacteristics to the other intracellular domain mutations(K650N, K605E and K650M) suggesting that the intracellularmutations enable tyrosine phosphorylation prior to glycosylation.

4.4. Greater levels of phosphotyrosine-positive puncta occurwith the intracellular FGFR3 mutations

The process of protein glycosylation is related to traffickingthrough the endoplasmic reticulum and Golgi apparatus.Typically, the glycosylation that is EndoH sensitive takes placein the endoplasmic reticulum whereas glycosylation insensitiveto EndoH takes place in the Golgi apparatus. Using immuno-fluorescence to examine cells 24 h post-transfection, it was foundthat the wild type and the mutant FGFR3s were associated withintracellular structures throughout the cytoplasm. In every casethere was some co-localization of these structures and punctatestructures detected by the Golgi marker antibody, GM130 (Fig.2E). This suggested that the resistance of the K650N, K650E,K650M and X807R mutants to complete glycosylation was notbecause they were prevented from reaching the Golgi apparatus.

Given that receptor tyrosine phosphorylation was also foundto inhibit receptor glycosylation, a phosphotyrosine antibodywas used for immunofluorescence (Fig. 3A). High phosphotyr-osine levels were essentially associated with FGFR3 transfectedcells (with a low level occasionally detectable in untransfectedcells). The punctate structures detected by the phosphotyrosineantibody also co-localized with some of the intracellularstructures detected by the FGFR3 antibody suggesting thatthese punctate structures include tyrosine phosphorylatedFGFR3. The intensity and the extent of phosphotyrosine-positive puncta appeared greater in the K650N, K650E,K650M and the X807R mutants compared with the othermutants and the wild type FGFRs (Fig. 3B).

4.5. Phosphotyrosine-positive puncta are distinctly associatedwith the Golgi apparatus

In order to attempt to identify whether the phosphotyrosineactivity of FGFR3 is associated with subcellular sites of

Fig. 2. The intracellular mutation X807R displays similar biochemical properties to those of the K650 mutations. (A) Schematic representation of the X807R mutationwith the additional read-through domain of the FGFR3 protein that includes a hydrophobic rich region. (B) The X807R FGFR3 appeared as three discrete bands ofdiffering mobilities (approximately 128,139,153 kDa) by Western blot, as a consequence of glycosylation. EndoH treatment removed mannose-rich moieties, andPNGase treatment removed all glycosylated moieties from FGFR3. (C) The relative levels of the native and differently glycosylated X807R FGFR3s from eightexperiments were quantified at 24 h, where the relative level of the lowest mobility isoform (153 kDa and marked by asterisks) was significantly lower than that of theWT receptor (**p<0.001). (D) Tyrosine phosphorylation on the X807R FGFR3 mutation was assessed by immunoprecipitation by anti-FGFR3 and examined byWestern blot with anti-FGFR3 or anti-phosphotyrosine. (E) The X807R, K650M andWT receptors detected by immunofluorescence (green) partially colocalised withthe Golgi marker GM130 (red). Scale bars equal 2 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version ofthis article.)


glycosylation, the immunodetection by anti-phosphotyrosinewas compared with markers of the ER (calnexin), of ER exit(sec31) [34] and of the Golgi apparatus (rab6) [35] (Fig. 4).Calnexin was associated with structures throughout thecytoplasm as expected for a protein expressed in the ER (itretains unfolded or unassembled N-linked glycoproteins in theER). Few of the low level of phosphotyrosine-positive punctaco-localized with cells transfected with the wild type FGFR3.However there appeared to be proportionally more co-localiza-tion of phosphotyrosine-positive puncta with calnexin in theK650M and X807R mutants. Sec31 was detected as punctatestructures throughout the cytoplasm as expected for a proteininvolved in the transport of proteins from the ER to the Golgi

apparatus. There was little colocalization of phosphotyrosine-positive punctawith sec31 in the cells transfected with wild typeand X807R FGFR3s, but some co-localization was seen in thecells transfected with K650M. Rab6-positive puncta wereidentified as being restricted to a portion of the cytoplasmproximal to an aspect of the nucleus expected for a proteinassociated with different parts of the Golgi (medial, trans-Golginetwork and post-Golgi vesicles). In all FGFR3-transfectedcells a high proportion of phosphotyrosine-positive punctaappeared co-localized with rab6. Therefore, phosphotyrosineactivity arising from FGFR3 transfection appeared mainlyassociated with the Golgi apparatus in comparison to the otherintracellular structures examined.

Fig. 3. The expression of FGFR3 is associated with elevated levels of tyrosine phosphorylated proteins. (A) Immunofluorescence detection of FGFR3 (green) andtyrosine phosphorylated proteins (red) revealed that detectable levels of tyrosine phosphorylation are only associated with FGFR3-transfected cells exemplified hereby cells transiently transfected with the WT and K650M FGFR3 variants. The nuclei (blue) of non-transfected cells are marked by asterisks. (B) The levels ofdetectable tyrosine phosphorylated proteins varied depending upon the construct that was transfected (WT, G380R, Y373C, K650N, K650M, K650E and X807RFGFR3s). Scale bars equal 2 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)


Fig. 4. Greater association of FGFR3-positive puncta (green) in WT, K650M or X807R transfected cells, with rab6-positive puncta (red, lower nine panels) comparedwith calnexin (red, upper nine panels) or sec31-positive puncta (red, middle nine panels). Scale bars equal 2 μm. (For interpretation of the references to colour in thisfigure legend, the reader is referred to the web version of this article.)



4.6. Nocodazole treatment redistributes phosphotyrosine-positivepuncta

To further examine whether the phosphotyrosine-positivepuncta were mainly associated with the Golgi apparatus,transfected cells were treated with nocodazole. Nocodazoleaffects microtubule stability and therefore disrupts Golgiapparatus positioning in the cell without affecting its function.Treatment with nocadazole did not appear to affect the relativelevel of fully glycosylated receptor in cells transfected withwild type or K650M FGFR3s (Fig. 5A). Analysis byimmunofluorescence found that nocodazole treatment led tophosphotyrosine-positive puncta being more scatteredthroughout the cytoplasm in the cells transfected with wildtype and K650M and X807R FGFR3s (Fig. 5B). Therefore

Fig. 5. Treatments of FGFR3-transfected cells with nocodazole or BFA suggest thaapparatus. (A) Western analysis of transfected cells with either the WT or K650MImmunofluorescence detection of FGFR3 (green) and tyrosine phosphorylated proteinnocodazole. (C) Western analysis of transfected cells with either the WT, G380R, Y3glycosylation. (D) Immunofluorescence detection of FGFR3 (green) and tyrosine phoK650M or K650E isoforms and treated with BFA. Scale bars equal 2 μm. (For interpreweb version of this article.)

these results further support that the phosphotyrosine activityassociated with FGFR3 transfection is associated with theGolgi apparatus.

4.7. BrefeldinA inhibits glycosylation and the number ofphosphotyrosine-positive puncta

BFA reversibly disrupts Golgi apparatus function by inhibi-ting the anterograde transport from the endoplasmic reticulumto the Golgi apparatus. BFA treatment reduced the relative levelof the fully glycosylated FGFR3 especially notable in the cellstransfected with the wild type, G380R, Y373C and K650Nreceptors where the levels of the fully glycosylated isoformsin the untreated samples were relatively high (Fig. 5C). Asexpected, this suggested that the glycosylation processing was

t the tyrosine phosphorylated proteins are primarily associated with the Golgiisoforms and treated with nocodazole, shows no change in glycosylation. (B)s (red) of cells transfected with WT, K650M or X807R isoforms and treated with73C, K650N, K650M or K650E isoforms showing that BFA treatment inhibitssphorylated proteins (red) of cells transfected with WT, G380R, Y373C, K650N,tation of the references to colour in this figure legend, the reader is referred to the


unable to be completed because BFA was preventing FGFR3from leaving the ER on its route to the Golgi apparatus.

Analysis by immunofluorescence suggested that the level ofintensity and number of phosphotyrosine-positive puncta weredramatically reduced following treatment with BFA in FGFR3-transfected cells (Fig. 5D). This again suggested that theinduction of phosphotyrosine levels primarily occurs post-ERand hence in the Golgi apparatus. However, despite thegenerally large reduction of phosphotyrosine immunoreactivity,more was found in those cells transfected with the three K650mutants and with X807R compared with the cells transfectedwith the wild type, G380R and Y373C. This suggested that thephosphotyrosine detected in the cells treated with BFA andtransfected with the K650 and X807R mutants may have arisenfrom those receptors that have yet to be glycosylated, asdetected by immunoprecipitation (Figs. 1E and 2C).

5. Discussion

Spontaneous missense neo-mutations in FGFR3 are respon-sible for chondrodysplasias such as HCH, ACH, TDI andTDII. Several mutations affecting particular amino acidresidues of the receptor have now been identified. Theanalysis of the X807R mutation further supports the idea thatthese mutations promote receptor tyrosine phosphorylation[32], and presumably promote concomitant tyrosine phos-phorylation of other protein substrates. The X807R representsa different type of mutation compared with the othercharacterized mutations, in that due to a read-through of thestop codon. The resultant predicted protein sequence has anelongated C-terminal domain. This additional C-terminaldomain may function by stabilizing tyrosine phosphorylation,perhaps by inducing some form of conformational change thatis similar to that described for the K650 mutations [21,26].The detection of tyrosine phosphorylation was found in asimilar fashion to that of the K650 mutated receptors, beingfound on the non- and partially-glycosylated X807R mutatedreceptor.

Further receptor tyrosine phosphorylation with any of themutated receptors, could not be provoked by the addition ofFGF ligand suggesting a limitation of this HEK model.However the fact that tyrosine phosphorylation was detectedwith all the receptors suggested that they have the capacity ofbeing phosphorylated in the absence of ligand. Anotherpotential problem with a transient transfection model is thatover-expressed proteins could accumulate in the ER. This mayhave occurred with this study as the immunodetection of theWT and mutated FGFR3s were found in structures throughoutthe cytoplasm and was similar to that of the ER marker calnexin(compare anti-FGFR3 in Figs. 2 and 3 with that of anti-calnexinin Fig. 4). Some of these structures though, were also marked bythe Golgi marker GM130, suggesting that all the receptors couldtraffic from the ER to the Golgi apparatus. This trafficking stepis indicative of correct receptor folding and N-glycosylation[36]. By quantifying the FGFR3 isoform levels of the WT andmutated receptors at 24 and 48 h post-transfection suggestedthat the correct progression of glycosylation could occur in this

model, with the relative level of the non-glycosylated FGFR3isoform decreasing in all the receptors when examined at 48 h.Moreover this analysis indicated that these glycosylation stepswere significantly disrupted only in the cells transfected withK650N, M and E, and X807R mutated FGFR3s.

The western analysis also found that those mutations locatedin the intracellular domain resulted in the tyrosine-phosphor-ylation of non- and partially-glycosylated receptor isoforms.This biochemical difference was corroborated by the immuno-fluorescence analysis, where it was found that these receptorsalso appeared associated with an enhanced level of tyrosine-phosphorylated proteins in intracellular structures that hadproperties consistent with being components of the Golgi ap-paratus. This suggested that those mutations located in theintracellular domain provoke premature phosphorylation of thede novo receptor.

This premature phosphorylation also appeared to inhibitreceptor glycosylation, concurring with previous analyses of themurine equivalents of the K650 mutations [29,30]. In contrast tothese previous studies where it has been suggested that theseconstitutively active mutated murine receptors accumulate inthe ER, the results here suggest a model where the constitutivelyactive human receptors, mutated in the intracellular domain,accumulate in the Golgi apparatus, where they promote furthertyrosine phosphorylation. In this study phosphotyrosine wasused as a readout of FGFR3 activity, whereas in previousstudies, particular tyrosine phosphorylated FGFR3 targets suchas Stat1, Stat3, Stat5, Jak1 and ERK1/2 were analysed [30,37].One of these studies suggested that murine equivalents of K650M and E phosphorylate these proteins in the ER because theirdetection was insensitive to BFA treatment. In this study, BFAdid not completely extinguish the immunodetection of phos-photyrosine and therefore the phosphotyrosine antibody mayhave detected these FGFR3 targets. One potential FGFR3signaling molecule post-ER in the case of K650N alone, couldbe Stat1 [30]. The murine equivalent of K650N FGFR3 wasfound to stimulate Stat1 phosphorylation with and without thepost-Golgi trafficking inhibitor monensin, but was found notstimulate Stat1 phosphorylation with BFA treatment (unlike themurine equivalents of K650 M and E) [30,37]. Therefore thissuggests that the phosphotyrosine detected in the Golgi byimmunofluorescence in this study, especially in the cases ofK650M, E and X807R, is either mainly derived from theFGFR3 itself or from other FGFR3 phosphotyrosine targets yetto be identified.

Hence, the receptors with intracellular domain mutationsdefine a distinct means by which mutated FGFR3 can disruptbone development, through enhanced and ectopic signalingemanating from the Golgi apparatus. This may explain whythe correlation between the degree disrupted receptor glyco-sylation and pathology severity may only hold for the K650mutated receptors. Moreover this study predicts that theX807R mutations gives rise to bone dysplasia by a similarmechanism to that found with the K650 mutations. It alsoseems clear that other factors in addition to tyrosinephosphorylation and disrupted receptor glycosylation areimplicated in the severity of the disease; and one additional


factor includes the intracellular localization of receptor activityto the Golgi apparatus.

Acknowledgements

We thank Prof.W. Hong, and DrsM.J. Morgan, O. Vielmeyerand J. Young for antibodies and advice. Part of this work wassupported by the European Skeletal Dysplasia Network(ESDN), EC contract QLG1-CT-2001-02188.

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.bbamcr.2006.12.010.

References

[1] D.E. Johnson, L.T. Williams, Structural and functional diversity in the FGFreceptor multigene family, Adv. Cancer Res. 60 (1993) 1–41.

[2] J. Huang, M. Mohammadi, G.A. Rodrigues, J. Schlessinger, Reducedactivation of RAF-1 andMAP kinase by a fibroblast growth factor receptormutant deficient in stimulation of phosphatidylinositol hydrolysis, J. Biol.Chem. 270 (1995) 5065–5072.

[3] M. Kanai, M. Goke, S. Tsunekawa, D.K. Podolsky, Signal transductionpathway of human fibroblast growth factor receptor 3. Identification of anovel 66-kDa phosphoprotein, J. Biol. Chem. 272 (1997) 6621–6628.

[4] M. Mohammadi, I. Dikic, A. Sorokin, W.H. Burgess, M. Jaye, J.Schlessinger, Identification of six novel autophosphorylation sites onfibroblast growth factor receptor 1 and elucidation of their importance inreceptor activation and signal transduction, Mol. Cell. Biol. 16 (1996)977–989.

[5] T. Spivak-Kroizman, M.A. Lemmon, I. Dikic, J.E. Ladbury, D. Pinchasi, J.Huang, M. Jaye, G. Crumley, J. Schlessinger, I. Lax, Heparin-inducedoligomerization of FGF molecules is responsible for FGF receptordimerization, activation, and cell proliferation, Cell 79 (1994) 1015–1024.

[6] A. Ullrich, J. Schlessinger, Signal transduction by receptors with tyrosinekinase activity, Cell 61 (1990) 203–212.

[7] J. Schlessinger, Ligand-induced, receptor-mediated dimerization andactivation of EGF receptor, Cell 110 (2002) 669–672.

[8] K.C. Hart, S.C. Robertson, D.J. Donoghue, Identification of tyrosineresidues in constitutively activated fibroblast growth factor receptor 3involved in mitogenesis, Stat activation, and phosphatidylinositol 3-kinaseactivation, Mol. Biol. Cell 12 (2001) 931–942.

[9] J. Schlessinger, Cell signaling by receptor tyrosine kinases, Cell 103 (2000)211–225.

[10] C. Basilico, D. Moscatelli, The FGF family of growth factors andoncogenes, Adv. Cancer Res. 59 (1992) 115–165.

[11] P. Blume-Jensen, T. Hunter, Oncogenic kinase signalling, Nature 411(2001) 355–365.

[12] A.O. Wilkie, G.M. Morriss-Kay, E.Y. Jones, J.K. Heath, Functions offibroblast growth factors and their receptors, Curr. Biol. 5 (1995) 500–507.

[13] F. Rousseau, J. Bonaventure, L. Legeai-Mallet, A. Pelet, J.M. Rozet, P.Maroteaux, M. Le Merrer, A. Munnich, Mutations in the gene encodingfibroblast growth factor receptor-3 in achondroplasia, Nature 371 (1994)252–254.

[14] R. Shiang, L.M. Thompson, Y.Z. Zhu, D.M. Church, T.J. Fielder, M.Bocian, S.T. Winokur, J.J. Wasmuth, Mutations in the transmembranedomain of FGFR3 cause the most common genetic form of dwarfism,achondroplasia, Cell 78 (1994) 335–342.

[15] G.A. Bellus, I. McIntosh, E.A. Smith, A.S. Aylsworth, I. Kaitila, W.A.Horton, G.A. Greenhaw, J.T. Hecht, C.A. Francomano, A recurrentmutation in the tyrosine kinase domain of fibroblast growth factor receptor3 causes hypochondroplasia, Nat. Genet. 10 (1995) 357–359.

[16] G.A. Bellus, I. McIntosh, J. Szabo, A. Aylsworth, I. Kaitila, C.A.

Francomano, Hypochondroplasia: molecular analysis of the fibroblastgrowth factor receptor 3 gene, Ann. N. Y. Acad. Sci. 785 (1996) 182–187.

[17] F. Rousseau, V. el Ghouzzi, A.L. Delezoide, L. Legeai-Mallet, M. LeMerrer, A. Munnich, J. Bonaventure, Missense FGFR3 mutations createcysteine residues in thanatophoric dwarfism type I (TD1), Hum. Mol.Genet. 5 (1996) 509–512.

[18] P.L. Tavormina, R. Shiang, L.M. Thompson, Y.Z. Zhu, D.J. Wilkin, R.S.Lachman, W.R. Wilcox, D.L. Rimoin, D.H. Cohn, J.J. Wasmuth,Thanatophoric dysplasia (types I and II) caused by distinct mutations infibroblast growth factor receptor 3, Nat. Genet. 9 (1995) 321–328.

[19] P.L. Tavormina, G.A. Bellus, M.K. Webster, M.J. Bamshad, A.E. Fraley, I.McIntosh, J. Szabo, W. Jiang, E.W. Jabs, W.R. Wilcox, J.J. Wasmuth, D.J.Donoghue, L.M. Thompson, C.A. Francomano, A novel skeletal dysplasiawith developmental delay and acanthosis nigricans is caused by aLys650Met mutation in the fibroblast growth factor receptor 3 gene,Am. J. Hum. Genet. 64 (1999) 722–731.

[20] Z. Vajo, C.A. Francomano, D.J. Wilkin, The molecular and genetic basis offibroblast growth factor receptor 3 disorders: the achondroplasia family ofskeletal dysplasias, Muenke craniosynostosis, and Crouzon syndrome withacanthosis nigricans, Endocr. Rev. 21 (2000) 23–39.

[21] M.C. Naski, Q. Wang, J. Xu, D.M. Ornitz, Graded activation of fibroblastgrowth factor receptor 3 by mutations causing achondroplasia andthanatophoric dysplasia, Nat. Genet. 13 (1996) 233–237.

[22] M.K. Webster, D.J. Donoghue, Constitutive activation of fibroblast growthfactor receptor 3 by the transmembrane domain point mutation found inachondroplasia, EMBO J. 15 (1996) 520–527.

[23] M.K. Webster, D.J. Donoghue, FGFR activation in skeletal disorders: toomuch of a good thing, Trends Genet. 13 (1997) 178–182.

[24] C. Li, L. Chen, T. Iwata, M. Kitagawa, X.Y. Fu, C.X. Deng, A Lys644Glusubstitution in fibroblast growth factor receptor 3 (FGFR3) causesdwarfism in mice by activation of STATs and ink4 cell cycle inhibitors,Hum. Mol. Genet. 8 (1999) 35–44.

[25] O. Segev, I. Chumakov, Z. Nevo, D. Givol, L. Madar-Shapiro, Y. Sheinin,M. Weinreb, A. Yayon, Restrained chondrocyte proliferation andmaturation with abnormal growth plate vascularization and ossificationin human FGFR-3(G380R) transgenic mice, Hum. Mol. Genet. 9 (2000)249–258.

[26] T. Iwata, C.L. Li, C.X. Deng, C.A. Francomano, Highly activated Fgfr3with the K644Mmutation causes prolonged survival in severe dwarf mice,Hum. Mol. Genet. 10 (2001) 1255–1264.

[27] E. Monsonego-Ornan, R. Adar, T. Feferman, O. Segev, A. Yayon, Thetransmembrane mutation G380R in fibroblast growth factor receptor 3uncouples ligand-mediated receptor activation from down-regulation, Mol.Cell. Biol. 20 (2000) 516–522.

[28] A. Weiss, J. Schlessinger, Switching signals on or off by receptor di-merization, Cell 94 (1998) 277–280.

[29] P.M. Lievens, E. Liboi, The thanatophoric dysplasia type II mutationhampers complete maturation of fibroblast growth factor receptor 3(FGFR3), which activates signal transducer and activator of transcription 1(STAT1) from the endoplasmic reticulum, J. Biol. Chem. 278 (2003)17344–17349.

[30] P.M. Lievens, C. Mutinelli, D. Baynes, E. Liboi, The kinase activity offibroblast growth factor receptor 3 with activation loop mutations affectsreceptor trafficking and signaling, J. Biol. Chem. 279 (2004) 43254–43260.

[31] F. Rousseau, P. Saugier, M. Le Merrer, A. Munnich, A.L. Delezoide, P.Maroteaux, J. Bonaventure, F. Narcy, M. Sanak, Stop codon FGFR3mutations in thanatophoric dwarfism type 1, Nat. Genet. 10 (1995) 11–12.

[32] G.A. Bellus, E.B. Spector, P.W. Speiser, C.A. Weaver, A.T. Garber, C.R.Bryke, J. Israel, S.S. Rosengren, M.K. Webster, D.J. Donoghue, C.A.Francomano, Distinct missense mutations of the FGFR3 lys650 codonmodulate receptor kinase activation and the severity of the skeletaldysplasia phenotype, Am. J. Hum. Genet. 67 (2000) 1411–1421.

[33] L. Legeai-Mallet, C. Benoist-Lasselin, A. Munnich, J. Bonaventure,Overexpression of FGFR3, Stat1, Stat5 and p21Cip1 correlates withphenotypic severity and defective chondrocyte differentiation in FGFR3-related chondrodysplasias, Bone 34 (2004) 26–36.

[34] B.L. Tang, T. Zhang, D.Y. Low, E.T. Wong, H. Horstmann, W. Hong,Mammalian homologues of yeast sec31p. An ubiquitously expressed form

http://dx.doi.org/doi:10.1016/j.bbamcr.2006.12.010


is localized to endoplasmic reticulum (ER) exit sites and is essential forER-Golgi transport, J. Biol. Chem. 275 (2000) 13597–13604.

[35] M. Roa, V. Cornet, C.Z. Yang, B. Goud, The small GTP-binding proteinrab6p is redistributed in the cytosol by brefeldin A, J. Cell Sci. 106 (Pt. 3)(1993) 789–802.

[36] L. Ellgaard, A. Helenius, Quality control in the endoplasmic reticulum,Nat. Rev., Mol. Cell Biol. 4 (2003) 181–191.

[37] P.M. Lievens, A. Roncador, E. Liboi, K644E/M FGFR3 mutants activateErk1/2 from the endoplasmic reticulum through FRS2alpha andPLCgamma-independent pathways, J. Mol. Biol. 357 (3) (2006) 783–792.

FGFR3 intracellular mutations induce tyrosine phosphorylation in the Golgi and defective...

Documents

Transcript of FGFR3 intracellular mutations induce tyrosine phosphorylation in the Golgi and defective...