272 Research Article

IntroductionActivating mutations in tyrosine kinase fibroblast growth factorreceptor 3 (FGFR3) result in several human skeletal dysplasiasincluding thanatophoric dysplasia (TD) and achondroplasia(ACH), which are the most common genetic form of lethal andnon-lethal skeletal dysplasias, respectively (Passos-Bueno at al.,1999). Probably the best documented effect of FGFR3 signaling incartilage is inhibition of chondrocyte proliferation, although themolecular mechanism of this phenotype is only beginning toemerge.

Several lines of evidence suggest that members of the signaltransducers and activators of transcription (STAT) family oftranscription factors may participate in the growth-inhibitory featureof FGFR3 signaling in chondrocytes. First, activation of FGFR3 bythe TD K650E mutation or FGF treatment, has been reported toresult in phosphorylation and nuclear localization of STAT1 inseveral non-chondrocyte cell models as well as in RCS chondrocytesor human primary chondrocytes, which is accompanied by inductionof the p21Waf1 (also known as CDKN1A) inhibitor of cell cycle andgrowth inhibition in 293T and RCS cells (Su et al., 1997; Lievensand Liboi, 2003; Nowroozi et al., 2005; Sahni et al., 1999; Legeai-Mallet et al., 1998). Second, in the cartilage of ACH- and TD-affected human fetuses as well as in mice carrying the ACH (G369C)or TD (K644E) mutations in FGFR3, STAT1 and STAT5 accumulate

and show nuclear localization, suggesting their activation (Legeai-Mallet et al., 1998; Legeai-Mallet et al., 2004; Chen et al., 1999; Liet al., 1999). Third, in two experimental studies, the loss of STAT1partially rescued the growth-inhibitory action of FGF signaling inboth in vitro and in vivo chondrocyte environments (Sahni et al.,1999; Sahni et al., 2001).

Since STAT1-mediated induction of p21Waf1 represents themechanism of cell growth inhibition by �-interferon (IFN�) (Chinet al., 1996; Bromberg et al., 1996), it is believed that thismechanism also occurs with FGFR3-mediated growth inhibition ofcartilage (Sahni et al., 1999; Legeai-Mallet et al., 2004). There are,however, several concerns regarding this model. First, since theloss of FGFR3 leads to significant skeletal overgrowth by increasedchondrocyte proliferation (Deng et al., 1996), the loss of Stat1should resemble the phenotype of Fgfr3-null mice, yet this is notthe case (Durbin et al., 1996). Second, in primary chondrocytesisolated from Stat1-null mice, RCS chondrocytes and PC12 cells,ERK and p38 MAP kinases appear to be candidates for FGF-mediated induction of p21Waf1 and growth inhibition (Murakami etal., 2004; Raucci et al., 2004; Krejci et al., 2004; Nowroozi et al.,2005). Third, crossing of Stat1-null mice with those carrying theACH mutation in FGFR3 did not rescue the ACH phenotypealthough increased chondrocyte proliferation was observed(Murakami et al., 2004).

Activating mutations in fibroblast growth factor receptor 3(FGFR3) cause several human skeletal dysplasias as a result ofattenuation of cartilage growth. It is believed that FGFR3inhibits chondrocyte proliferation via activation of signaltransducers and activators of transcription (STAT) proteins,although the exact mechanism of both STAT activation andSTAT-mediated inhibition of chondrocyte growth is unclear.We show that FGFR3 interacts with STAT1 in cells and iscapable of activating phosphorylation of STAT1 in a kinaseassay, thus potentially serving as a STAT1 kinase inchondrocytes. However, as demonstrated by western blottingwith phosphorylation-specific antibodies, imaging of STATnuclear translocation, STAT transcription factor assays andSTAT luciferase reporter assays, FGF does not activate STAT1

or STAT3 in RCS chondrocytes, which nevertheless respond toa FGF stimulus with potent growth arrest. Moreover, additionof active STAT1 and STAT3 to the FGF signal, by means ofcytokine treatment, SRC-mediated STAT activation orexpression of constitutively active STAT mutants does notsensitize RCS chondrocytes to FGF-mediated growth arrest.Since FGF-mediated growth arrest is rescued by siRNA-mediated downregulation of the MAP kinase ERK1/2 but notSTAT1 or STAT3, our data support a model whereby the ERKarm but not STAT arm of FGF signaling in chondrocytesaccounts for the growth arrest phenotype.

Key words: FGFR3, STAT, Cartilage, Chondrocyte, Fibroblast growthfactor, Growth arrest

Summary

STAT1 and STAT3 do not participate in FGF-mediatedgrowth arrest in chondrocytesPavel Krejci1,2,*, Lisa Salazar3, Helen S. Goodridge4, Tamara A. Kashiwada3, Matthew J. Schibler5,Petra Jelinkova1, Leslie Michels Thompson3 and William R. Wilcox6,7

1Institute of Experimental Biology, Masaryk University, 61137 Brno, Czech Republic2Department of Cytokinetics, Institute of Biophysics ASCR, 61265 Brno, Czech Republic3Department of Psychiatry and Human Behavior, University of California, Irvine, CA 92697, USA4Immunobiology Research Institute, Cedars-Sinai Medical Center, Los Angeles, CA 90048, USA5Brain Research Institute, David Geffen School of Medicine at UCLA, Los Angeles, CA 90095, USA6Medical Genetics Institute, Cedars-Sinai Medical Center, Los Angeles, CA 90048, USA7Department of Pediatrics, David Geffen School of Medicine at UCLA, Los Angeles, CA 90095, USA*Author for correspondence (e-mail: krejcip@sci.muni.cz)

Accepted 30 October 2007Journal of Cell Science 121, 272-281 Published by The Company of Biologists 2008doi:10.1242/jcs.017160

Jour

nal o

f Cel

l Sci

ence

273STATs in FGF signaling in cartilage

Regardless of the specific role of STATs, it is not clear howFGFR3 activates STATs. Activation of FGFR3 leads to tyrosinephosphorylation of STAT1, STAT3 and STAT5, depending on thecell system (Su et al., 1997; Sahni et al., 1999; Nowroozi et al.,2005), but the nature of the STAT-kinase involved is unknown.FGFR3 itself is a tyrosine kinase and can recruit SRC kinase (aknown STAT activator) (Klint et al., 1999; Cao et al., 1996), butthere is no evidence demonstrating direct FGFR3- or SRC-mediated tyrosine phosphorylation of STAT in chondrocytes. Thisstudy was undertaken to extend our knowledge regarding yetunclear aspects of STAT signaling in FGFR3 signaling in achondrocyte environment.

ResultsFGFR3 co-immunoprecipitates with STATs and phosphorylatesSTAT1 in a kinase assayAlthough it is well documented that FGFR3 activation leads toSTAT phosphorylation in cells, the nature of the tyrosine kinaseinvolved in this process is unknown. We tested whether FGFR3alone can act as a STAT kinase in chondrocytes. First, we askedwhether STAT1 and/or STAT3 interact with FGFR3 in cells. STATswere immunoprecipitated from HeLa cells or RCS chondrocytestransfected with C-terminally FLAG-tagged wild-type (wt)

FGFR3, a constitutively active FGFR3 mutant (K650E) or akinase-dead FGFR3 (K508M) mutant, and immunocomplexeswere probed for FGFR3 by FLAG (Fig. 1A) or FGFR3 (Fig. 1B)western immunoblotting (WB). Fig. 1 shows that FGFR3 co-immunoprecipitated with STAT1 and STAT3 in both cell types. Totest whether overexpressed FGFR3 activates STATs in RCSchondrocytes, we probed the RCS cell lysates for phosphorylatedSTAT1(Y701) and STAT3(Y705). Phosphorylation ofSTAT1(Y701) by FGFR3-K650E was detected but not of STAT3(Fig. 1B and data not shown).

Second, we used a cell-free kinase assay to test whether FGFR3can directly phosphorylate STAT1 and STAT3 on their activatorytyrosines 701 and 705, respectively. Wild-type or FGFR3-K650Ewere expressed in CHO cells, immunoprecipitated and subjectedto a kinase assay (Krejci et al., 2007), using recombinant STAT1or STAT3 as a substrate. Fig. 1C shows that FGFR3-wtphosphorylates STAT1 on tyrosine 701 weakly, in comparisonto FGFR3-K650E, which causes strong STAT1(Y701)phosphorylation; no phosphorylation of STAT3(Y705) was found(data not shown). When the recombinant, kinase-activeintracellular domain of FGFR3 was used as a kinase, STAT1 wasphosphorylated to a similar extent as with FGFR3-K650E,suggesting that the cell-borne FGFR3 does not phosphorylate

Fig. 1. FGFR3 interacts with STAT1 andSTAT3. FLAG-tagged wt, FGFR3-K650Eor FGFR3-K508M were expressed in HeLacells (A) or RCS chondrocytes (B) and thewhole cell lysates were subjected to WB forindicated molecules (left panel). Actinserves as a loading control. The samelysates were subjected to STAT1 (middlepanel) and STAT3 (right panel)immunoprecipitation (IP) followed by WB.Please note that the transfected FGFR3-FLAG was detected by FLAG antibody inA and by FGFR3 antibody in B. (C) FLAG-tagged wt or FGFR3-K650E,immunoprecipitated from CHO cells, orrecombinant tyrosine kinase (TK) domainof FGFR3 were subjected to an in vitrokinase assay with recombinant STAT1 as asubstrate. The level of STAT1phosphorylation was determined by WBwith antibody recognizing STAT1 onlywhen phosphorylated at Y701. Samplesincluding the FGFR inhibitor SU5402 (20�M) or those with ATP omitted serve asnegative controls for the kinase reaction.The levels of total STAT1 and FGFR3 serveas controls for the substrate and kinasequantity, respectively. Note that the FGFR3antibody was raised against theextracellular domain of FGFR3 and thuscannot recognize FGFR3-TK. (D) Thekinase assay was carried out as describedabove, using FGFR3-TK as a kinase, andrecombinant STAT1 and/or recombinantFRS2 as substrates. The level of STAT1phosphorylation was determined asdescribed above, whereas FRS2 tyrosinephosphorylation was determined by WBwith the 4G10 phosphotyrosine antibody.

Jour

nal o

f Cel

l Sci

ence

274

STAT1 via a co-immunoprecipitated intermediate. Next, we testedwhether FGFR3 can phosphorylate STAT1 in the presence of itsphysiological substrate fibroblast growth factor receptor substrate2 (FRS2). Fig. 1D shows that STAT1(Y701) phosphorylation doesoccur in the presence of recombinant FRS2, albeit to a lesser extent.

FGF signaling does not activate STATs in RCS chondrocytesFor the following experiments we used RCS chondrocytes, aFGFR3-expressing chondrocytic cell line that represents the bestcharacterized cell model for FGFR3-related skeletal dysplasias todate. RCS chondrocytes respond to FGFR3 activation with growtharrest similar to in vivo chondrocytes (Aikawa et al., 2001;Rozenblatt-Rosen et al., 2002; Dailey et al., 2003; Raucci et al.,2004; Krejci at al., 2005). First, we used WB with phosphorylation-specific antibodies to test whether FGFR3 activation, via FGF2treatment, leads to STAT phosphorylation. Fig. 2A,B shows thatRCS treatment with FGF2 did not lead to activatory tyrosine

Journal of Cell Science 121 (3)

phosphorylation of STAT1(Y701), STAT3(Y705) or STAT5(Y694)in contrast to IFN� or interleukin 6 (IL6) treatment that inducedsignificant tyrosine phosphorylation of all three STATs. Similarresults were found in cells treated with FGF2 together with heparinor with FGF1 (data not shown). Since FGF activates ERK1/2 inRCS cells (Fig. 2A), and MAP kinases are capable ofphosphorylating STATs at a conserved P(M)SP motif locatedbetween residues 720 and 730 (Decker and Kovarik, 2000), weprobed RCS cells for STAT1(S727) and STAT3(S727)phosphorylation. Fig. 2C shows that treatment with FGF2 leadsto phosphorylation of STAT3(S727); no phosphorylation ofSTAT1(S727) was found (data not shown).

Detection of activated STATs by WB can yield false-negativeresults due to a limited sensitivity of the method. We therefore usedmore sensitive ways to look for the STAT activation in FGF2-treated RCS chondrocytes. First, since the nuclear accumulation ofthe STATs represent a marker of their activation, we expressedSTAT1 or STAT3 tagged with green or yellow fluorescent protein(GFP or YFP, respectively) in RCS cells and examined theinfluence of FGF2 on subcellular localization of both fusionproteins by confocal microscopy. The bio-activity of both STATfusion proteins has been extensively examined and found not to bealtered by the tag (Köstner et al., 1999; Herrmann et al., 2003). Inuntreated cells, STAT1-GFP and STAT3-YFP were found equallydistributed between the cell nucleus and cytoplasm (Fig. 3C,D).Since RCS cells used for STAT imaging were grown in thepresence of 10% fetal bovine serum, the presence of nuclear STATsmay reflect basal activation by endogenous or serum-bornecytokines. Alternatively, the STATs may have a nuclear presencebecause of constant circulation of inactive STATs between thenucleus and cytoplasm, according to models proposed recently byReich and Liu (Reich and Liu, 2006). In RCS cells treated withpositive controls for STAT activation, i.e. with IFN� or IL6 for 30minutes, we observed rapid nuclear accumulation of the entirecytoplasmic complement of STAT1 or STAT3, respectively. Bycontrast, both short-term (10, 30 minutes, 1, 2 and 4 hours) andlong-term (8, 24 and 48 hours) FGF2 treatment did notsignificantly affect the subcellular distribution of either STAT (Fig.3C,D and data not shown). We also tested whether long-term FGF2treatment affects the ability of IFN� and IL6 to cause a nucleartranslocation of STAT1 or STAT3, respectively. Fig. 3D shows thatIL6-mediated nuclear translocation of STAT3 appears inhibited bya 48-hour pre-treatment of cells with FGF2. Nuclear translocationof STAT1 induced by 30-minute treatment with IFN� was notaffected by FGF2 (data not shown).

Next, we used an ELISA-based EMSA assay to probe whetherFGF2 activates STATs in RCS chondrocytes. Cells grown in thepresence of 10% fetal bovine serum were treated with FGF2 forup to 72 hours and analyzed for active nuclear STAT1 or STAT3(Fig. 4A). Again, no significant increase of active, nuclear STAT1or STAT3 was found upon FGF2 treatment. Surprisingly, there wasa progressive reduction of basal active nuclear STAT withprolonged FGF2 treatment.

Finally, we used a luciferase-based reporter assay to test forFGF2-mediated STAT activation in RCS chondrocytes. Cells weretransfected with three different reporters expressing fireflyluciferase controlled by promoters containing STAT responsiveenhancer motifs (GAS, Ly6e and M67) (Khan et al., 1993; Besseret al., 1999), together with an internal control reporter encodingRenilla reniformis luciferase driven by the thymidine-kinasepromoter. Fig. 4B shows that FGF2 did not activate any of the three

Fig. 2. FGF2 treatment leads to serine but not tyrosine phosphorylation ofSTAT in RCS chondrocytes. (A,B) RCS chondrocytes were serum-starved for12 hours, treated with FGF2, IL6 or IFN� for the indicated times, andanalyzed for ERK1/2, STAT1, STAT3 and STAT5 by WB with antibodyrecognizing the given molecule only when phosphorylated at a specific site.The arrow indicates Y694-phosphorylated STAT5. The same membrane wasre-probed with antibody recognizing the given molecule regardless of itsphosphorylation. (C) RCS chondrocytes were serum-starved for 12 hours,treated with FGF2 (100 ng/ml) in the presence of heparin (1 �g/ml) for theindicated times, and analyzed for STAT3 by WB with antibody recognizingSTAT3 only when phosphorylated at S727. The appearance of S727-phosphorylated STAT3, triggered by FGF2, is indicated by an arrow. Thelower band appears to be a crossreactivity of the antibody as indicated by theposition of the total STAT3 detected on the re-probed membrane.

Jour

nal o

f Cel

l Sci

ence

275STATs in FGF signaling in cartilage

Fig. 3. FGF2 treatment does not trigger nuclear accumulation of STAT1-GFP or STAT3-YFP in RCS chondrocytes. (A,B) RCS chondrocytes were transfected withSTAT1-GFP or STAT3-YFP vectors. Forty-eight hours after transfection, the cells were harvested and analyzed for STAT1 and STAT3 by WB. Note the amount ofSTAT fusion proteins in transfected cells (lane 2) in comparison to untransfected controls (lane 1). (C,D) RCS chondrocytes were transfected with STAT1-GFP (C) orSTAT3-YFP (D), grown for 48 hours and treated with FGF2, IL6 and IFN� for the indicated times, fixed and analyzed for STAT subcellular distribution by confocalmicroscopy. The DAPI staining indicates the cell nucleus. Note that FGF2 did not influence the distribution of either protein in contrast to positive controls (IFN�- orIL6-treated cells) that show nuclear accumulation of all the STAT1-GFP or STAT3-YFP. Also note the incomplete IL6-mediated nuclear translocation of STAT3-YFPin cells pre-treated with FGF2 for 48 hours, suggesting an interference with canonical STAT3 signaling by the FGF2 stimulus. Scale bar: 10 �m.

Jour

nal o

f Cel

l Sci

ence

276

STAT reporters used. Interestingly, we observed a slight decreasein the basal luciferase activity of two out of three reporters in cellstreated with FGF2 for 72 hours, again suggesting an inhibition ofendogenous STAT activity by chronic FGF2 treatment.

Addition of active STATs to FGF signal does not enhanceFGF2-mediated growth arrestAlthough the FGF-FGFR3 signaling does not appear to activateSTAT1 or STAT3 in RCS chondrocytes (Figs 2-4), both STATs are

Journal of Cell Science 121 (3)

present and thus may contribute to FGF2-mediated growth arrestwhen activated. We tested this hypothesis by determining whetheraddition of active STAT1 or STAT3 sensitizes RCS cells to FGF-mediated growth arrest. First, we added active STATs to the FGFsignaling via co-stimulation of RCS cells with IFN� or IL6. Cellswere grown in presence of FGF2, IFN� and/or IL6 for 72 hoursand counted to test if IFN� and/or IL6 contribute to the FGF2-mediated growth arrest. Despite clear STAT activation in cellstreated with IFN� and/or IL6, there was no contribution from

either cytokine to FGF2-mediated growth arrest (Fig.5A,B).

Next, we transfected RCS cells with a constitutivelyactive SRC kinase mutant (Y529F; ca-SRC), that is knownto activate STAT3 via phosphorylation at Y705 (Shi andKehrl, 2004). As evidenced by a STAT luciferase reporterassay, ca-SRC-expressing cells show significant amountsof STAT activation, although they respond to FGF2treatment identically to untransfected cells (Fig. 5C,D,E).Finally, we expressed a constitutively active STAT3 mutant(A662C/N664C; ca-STAT3) in RCS chondrocytes anddetermined its contribution to FGF2-mediated growtharrest. Fig. 5G shows significant activation of the STATluciferase reporter in cells expressing ca-STAT3. However,this activation does not sensitize RCS chondrocytes toFGF2-mediated growth arrest (Fig. 5G,H). Similarexperiments could not be carried out with wt and ca-STAT1since we were unable to express STAT1 in sufficientquantities in RCS chondrocytes (data not shown).

siRNA-mediated downregulation of STAT1 or STAT3does not rescue FGF-mediated growth arrestTaken together, activation of STAT1 or STAT3 does notcontribute to FGF2-mediated growth arrest in RCSchondrocytes (Fig. 5). Since it has been suggested thatSTATs function in FGF-mediated chondrocyte growtharrest independent of their activation (Raucci et al., 2004),we asked whether downregulation of STAT1 or STAT3, viaRNA interference, rescues the FGF2-mediated growtharrest of RCS chondrocytes. Cells were transfected withsiRNAs directed against STAT1, STAT3 or ERK1/2 andanalyzed for target downregulation as well as for siRNAspecificity 24 hours later. At an 80 nM siRNA scale, asignificant silencing of all three targets was observed (Fig.6A), although the transfection with siRNA targetingERK1/2 also led to a slight STAT1 downregulation. Next,cells were transfected with siRNAs targeting STAT1,STAT3 or ERK1/2, grown for 24 hours and probed for theirresponse to FGF2 in a 72-hour-long growth experiment.Fig. 6B shows that untransfected RCS cells responded tovarious levels of FGF2 with the usual amount of growtharrest. This phenotype did not appear to be significantlyrescued by STAT1 or STAT3 downregulation (Fig. 6B). Bycontrast, downregulation of ERK1/2 led to at least partialreversal of the growth arrest with all FGF2 concentrationsused.

DiscussionAlthough the FGF system represents one of the majorregulators of cartilage, many aspects of FGF signaling inthe chondrocyte environment remain unclear, including therole of STAT transcription factors in FGF-regulated events

FGF2 _

10’ 30’ 1h 2h 4h 8h 12h 24h 36h 48h 72h _

IFNγ _ _ _ _ _ _ _ _ _ _ _ _

+

STA

T1

activ

atio

n (O

D45

0-65

0nm

)

0.25

0.2

0.15

0.05

0.025

0

control

FGF2 _

10’ 30’ 1h 2h 4h 8h 12h 24h 36h 48h 72h _

IL6 _ _ _ _ _ _ _ _ _ _ __ __

+

0

0.8

0.6

0.4

0.2

control

STA

T3

activ

atio

n (O

D45

0-65

0nm

)

A

B

0.24

0.18

0.12

0.6

0

**

**F-L

uc/R

-Luc

(x1

0-3) 90

60

30

0

**

**

4

3

2

1

0

**

reporter: pTal-Luc-4xGas pZLuc-TK-3xLy6e pTATA-TK-Luc-4xM67

FGF2 _

+ _ _

+ _ _

+ _

IFNγ _ _

+ _ _ _ _ _

+IL6

_ _ _

_

_

+ ̀ _ _ _

Fig. 4. FGF2 does not trigger STAT transcriptional activity. (A) Growing RCSchondrocytes were treated with FGF2 for the indicated times and cell nuclei wereisolated and analyzed for active STAT1 (upper graph) or STAT3 (lower graph) bySTAT ELISA-based EMSA assay as described in the Materials and Methods. STATactivation was quantified by spectrophotometry. Note that FGF2 did not significantlyelevate active nuclear STAT1 or STAT3. Rather, a diminution of basal active nuclearSTATs appeared with prolonged FGF2 treatment. The data represent an average fromtwo samples with the indicated range. The dashed line indicates basal STAT nuclearactivity. Cells treated for 30 minutes with IFN� or IL6 serve as positive controls forSTAT activation. (B) RCS chondrocytes were transfected with three different STATfirefly luciferase (F-Luc) reporter vectors and a control Renilla luciferase (R-Luc)vector, stimulated with FGF2, IFN� or IL6 for 72 hours and analyzed for luciferaseactivity as described in the Material and Methods. Data represent an average from fourwells with indicated standard deviation. Statistically significant differences relative tocontrol are indicated (Student’s t-test; **P<0.01).

Jour

nal o

f Cel

l Sci

ence

277STATs in FGF signaling in cartilage

in cartilage. Despite genetic evidence suggesting a role for STAT1in FGF-mediated inhibition of cartilage growth (Su et al., 1997;Sahni et al., 1999; Legeai-Mallet et al., 1998; Legeai-Mallet et al.,2004; Chen et al., 1999; Li et al., 1999), the molecular mechanismof this role remains obscure. In this study, we attempted tocharacterize some of the unclear aspects of STAT participation inFGF signaling, namely the mechanism of FGF-mediated STAT

activation and the contribution of STATs to FGF-mediatedchondrocyte growth inhibition.

STAT1 activation by FGFR3Activation of FGFR3 has been reported to lead to tyrosinephosphorylation of STAT1, STAT3 or STAT5, depending on thecell system (Su et al., 1997; Sahni et al., 1999; Nowroozi et al.,

Fig. 5. Addition of active STATs does not sensitize RCS chondrocytes to FGF2-mediated growth arrest. (A) RCS chondrocytes were treated with FGF2, IL6 andIFN� for 30 minutes and analyzed for activatory phosphorylation of STAT1, STAT3 and ERK1/2 by WB. The levels of total STAT1, STAT3 and ERK1/2 serve asloading controls. (B) RCS chondrocytes were treated with FGF2, IL6 and IFN� for 72 hours and counted. The data represent an average from at least three wellswith the indicated standard deviation. The arrow indicates maximal growth arrest in cells treated with 20 ng/ml of FGF2. Note that all FGF2 treatments led tostatistically significant growth inhibition (Student’s t-test; P<0.01; results not shown on the graph) relative to untreated control, and that IL6 and/or IFN� co-treatment with FGF2 led to statistically significant differences relative to FGF2 alone (Student’s t-test; *P<0.05, **P<0.01). (C) RCS chondrocytes weretransfected with vector carrying constitutively activated SRC kinase (ca-SRC), grown for 24 hours and analyzed for the indicated molecules by WB. (D) RCSchondrocytes were transfected with empty vector or vector carrying ca-SRC together with STAT firefly luciferase (F-Luc) reporter vector (pTATA-TK-Luc-4xM67)and a control Renilla luciferase (R-Luc) vector (pRL-TK), grown for 96 hours and analyzed for luciferase activity as described in the Material and Methods. Thedata represent an average of four wells with indicated standard deviation. Statistically significant differences relative to control are indicated (Student’s t-test;*P<0.05, **P<0.01). (E) RCS chondrocytes were transfected with vector carrying ca-SRC, treated with FGF2 for 72 hours and counted. Note that all FGF2treatments led to statistically significant growth inhibition (Student’s t-test; P<0.01; not shown on the graph) relative to untreated controls, but overexpression ofca-SRC did not alter the FGF2-mediated growth arrest. The data represent an average from three wells with the indicated standard deviation. (F,G,H) RCSchondrocytes were transfected, analyzed for transgene expression, luciferase activity and FGF2-mediated growth arrest as described above except that vectorsexpressing wt or constitutively active STAT3 (ca-STAT3) were used instead of ca-SRC. Note that all FGF2 treatments led to statistically significant growthinhibition (Student’s t-test; P<0.01; not shown on the graph) relative to untreated controls, but overexpression of wt or ca-STAT3 did not alter the FGF2-mediatedgrowth arrest.

Jour

nal o

f Cel

l Sci

ence

278

2005), but the nature of a STAT-kinase involved in this process isunknown. Since FGFR3 itself is a tyrosine kinase, we askedwhether it can directly phosphorylate STAT1 or STAT3. We foundthat FGFR3 is capable of phosphorylating STAT1(Y701) in a cell-free kinase assay (Fig. 1). Although one cannot rule out that thisphosphorylation is an artifact of the method where large quantitiesof both kinase and substrate are present, several findings suggestthat direct FGFR3-mediated phosphorylation of STAT1(Y701) isphysiologically relevant. First, FGFR3 did not phosphorylateSTAT3(Y705) despite the similarity in the motif sequences(GYIKT for STAT1 versus PYLKT for STAT3) that are bothrecognized by known STAT kinases (Kisseleva et al., 2002).Second, the difference in magnitude of STAT1 phosphorylationobserved here between wt and FGFR3-K650E corresponds to thedifference in 293T cells, where FGFR3-K650E caused 20 timesstronger STAT1 phosphorylation than wt FGFR3 (Su et al., 1997).Third, FGFR3 phosphorylated STAT1 in the presence of itsphysiological substrate FRS2 (Fig. 1).

We conclude that FGFR3 can potentially function as a STAT1kinase in chondrocytes, directly activating STAT1 through itsphosphorylation at Y701. In a kinase assay, FGFR3 equallyphosphorylated both STAT1 and FRS2 (Fig. 1D). In contrast tonormal cells, TD chondrocytes may have different substrateselectivities since FGFR3-K650E is active in the endoplasmicreticulum unlike wt FGFR3, which is active at the cell membrane(Lievens and Liboi, 2003; Raffioni et al., 1998). It is thus likelythat cytoplasmic STAT1 is phosphorylated predominantly byFGFR3-K650E whereas membrane-anchored FRS2 is a preferredsubstrate for wt FGFR3. In RCS chondrocytes, the immatureFGFR3 forms immunoprecipitated preferentially with STAT1,supporting this hypothesis (Fig. 1B). In contrast to STAT1, wefound that FGFR3 does not phosphorylate STAT3 in a kinase assay.This implies that STAT3 activation by FGFR3 signaling, when itoccurs, involves intermediates, such as JAK2 and SRC kinases,recently shown to activate STAT3 in response to FGF in endothelialcells (Deo et al., 2002).

In further support of physiological interaction between STATsand FGFR3, we found that FGFR3 co-immunoprecipitates withboth STAT1 and STAT3 from HeLa cells and RCS chondrocytes

Journal of Cell Science 121 (3)

(Fig. 1A,B). The strength of the STAT-FGFR3 interactioncorrelated positively with the amounts of FGFR3 activation.Surprisingly, however, we found that a significant amount of bothSTATs associate with the FGFR3-K508M mutant suggesting thatFGFR3 activation (i.e. autophosphorylation) is not a prerequisitefor STAT binding (Fig. 1), but may facilitate binding. Although wecannot rule out the possibility that FGFR3-K508M isheterophosphorylated by endogenous FGFRs, our data suggest thatFGFR3 pre-assembles with STAT in FGF-naive cells.

FGF treatment of RCS chondrocytes leads to potent activationof ERK1/2 (Krejci et al., 2004; Raucci et al., 2004) (Fig. 2A),which can phosphorylate STAT at a conserved P(M)SP motiflocated between residues 720 and 730 (Decker and Kovarik,2000). We found that such phosphorylation of STAT3(S727)occurred in FGF2-treated RCS cells (Fig. 2C). Although we didnot detect similar phosphorylation of STAT1, probably because ofthe low level of STAT1 in RCS cells, it is probable thatSTAT1(S727) is also phosphorylated by ERK1/2 in RCS cells(Decker and Kovarik, 2000). Serine 727 phosphorylation ofSTAT3 is necessary for its full transcription factor activity,depending on the promoter and/or cell type (Wen et al., 1995; Kimand Baumann, 1997). The functional significance of STAT3(S727)phosphorylation in FGF signaling in RCS chondrocytes is,however, unclear since it occurs without concomitant activatoryphosphorylation at Y705.

Contribution of STATs to FGF-mediated growth arrest in theRCS chondrocytesAlthough murine models have played an important role indelineating the features of FGFR3-related skeletal dysplasias,unraveling the precise molecular mechanisms of aberrant FGFsignaling in the chondrocyte environment requires an in vitroexperimental model. Although human primary chondrocytes caneasily be obtained from the long bone epiphyses of TD fetuses,such cells rapidly lose their differentiated, cartilage-like phenotypeupon in vitro cultivation, including their growth inhibitory responseto a FGF stimulus. As early as 72 hours in culture, human primarychondrocytes begin responding to FGF treatment by proliferating,which is typical for undifferentiated mesenchymal cells but

Fig. 6. RNAi-mediated downregulation ofSTAT1 and STAT3 does not rescue the FGF-mediated RCS growth arrest. (A,B) RCSchondrocytes were subjected to ERK1/2, STAT1and STAT3 RNAi as described in the Materialsand Methods. 24 hours later, the cells treatedwith specific siRNA (arrow) were eitheranalyzed for ERK1/2, STAT1, STAT3 and actinlevels by WB (A), or treated with FGF2 for 72hours and counted (B). The data represent anaverage from three wells with the indicatedstandard deviation. Note that only ERK1/2downregulation lead to statistically significantrescue (Kruskal-Wallis test, *P<0.05) of theFGF2-mediated growth arrest.

Jour

nal o

f Cel

l Sci

ence

279STATs in FGF signaling in cartilage

opposite to the outcome of FGF signaling in cartilage (P.K. andW.R.W., unpublished) (Legeai-Mallet et al., 1998). This rapid lossof cartilage-like properties of FGF signaling renders culturedhuman chondrocytes impractical for experimental studiesaddressing the mechanisms of chondrocyte-specific FGF signaling.Although several other cell models have been used to explore themechanism of FGF signaling in a chondrocyte environment(Henderson et al., 2000; Yamanaka et al., 2003), RCS chondrocytesrepresent the best studied chondrocyte cell model to date. UsingRCS chondrocytes, several essential features of FGF signaling inchondrocytes have been elucidated, including the mechanisms ofFGF-mediated chondrocyte growth arrest, cytoskeletal alterations,loss of chondrocyte extracellular matrix, mechanisms of FGF andC-natriuretic peptide signaling crosstalk and others (Aikawa et al.,2001; Ben-Zvi et al., 2006; Rosenblatt-Rosen et al., 2002; Daileyet al., 2003; Raucci et al., 2004; Krejci et al., 2004; Krejci et al.,2005; Krejci et al., 2007; Priore et al., 2006). Since they respondto FGF treatment with potent growth arrest that is accompanied bythe activation of STAT1 (Sahni et al., 1999), we used RCSchondrocytes in this study.

In contrast to previously published data (Sahni et al., 1999; Ben-Zvi et al., 2006), we were unable to detect STAT1 activationfollowing FGF treatment of RCS chondrocytes using four differentexperimental approaches, i.e. WB with the STAT phosphorylation-dependent antibodies, imaging of nuclear translocation of activeSTAT, STAT ELISA-based EMSA assay and STAT luciferasereporter assay (Figs 2-4). We presently do not know the reason forthis difference.

In this study, we systematically evaluated the requirement ofboth STAT1 and STAT3 for the FGF-FGFR3-mediated growthinhibition of RCS chondrocytes. As discussed above, we wereunable to detect STAT1 and STAT3 activation by FGF despitecombining several different approaches (Figs 2-4). Next, wedetermined whether addition of active STAT1 or STAT3 to the FGFsignal, by means of cytokine treatment, SRC-mediated STATactivation, or expression of a constitutively active STAT mutantcould sensitize RCS chondrocytes to the FGF-mediated growtharrest and found that none of the experimental interventions had aneffect (Fig. 5). Finally, we tested whether STAT1 and STAT3contribute to the FGF-mediated growth arrest independently oftheir activation, as proposed earlier (Raucci et al., 2004), usingsiRNA-mediated acute knockdown. Although the informationvalue of the RNAi experiment is diminished because of the slightdownregulation of STAT1 by siRNA targeting ERK1/2, we foundno contribution of STAT1 or STAT3 downregulation to the FGF-mediated growth arrest in RCS chondrocytes (Fig. 6). Altogether,we did not find a direct contribution of either STAT1 or STAT3 tothe growth inhibitory signaling of FGF-FGFR3 in RCSchondrocytes. Our data thus confirm that the ERK1/2 arm, but notthe STAT arm of FGF signaling is directly responsible for thegrowth arrest phenotype in chondrocytes, as demonstrated earlier(Raucci et al., 2004; Krejci et al., 2004). Our findings cannotentirely reflect what is occurring in the cartilage growth plate,where several signaling systems influence chondrocyteproliferation in a complex spatiotemporal relationship. Also, sinceRCS chondrocytes express wt FGFR3, our findings might notreflect the signaling of mutated FGFR3, particularly that ofFGFR3-K650E, which is a strong STAT1 activator (Su et al., 1997)(Fig. 1). Lastly, STATs may contribute to the FGF signaling incartilage by acting in FGF-regulated processes not directly relatedto proliferation.

While attempting to detect FGF-mediated activation of STATs,we surprisingly found that prolonged FGF treatment inhibitsSTAT1 and STAT3 activation by IFN� or IL6, respectively. Thisinhibition was evident using several approaches to detect STATactivation (Figs 3, 4), suggesting that chronic FGF stimulusinterferes with cytokine-induced STAT activation in RCSchondrocytes. Some cytokine-STAT signaling systems, such as theIL6 family of cytokines that signals via the gp130 receptor,represent important positive regulators of endochondral growth(Sims et al., 2004). Our data opens the possibility that FGFsignaling exerts some of its effects on cartilage through interferingwith canonical STAT pathways. This hypothesis is currently underinvestigation.

Materials and MethodsCell culture and western immunoblottingRCS chondrocytes, HeLa cells and CHO cells were propagated in Dulbecco’smodified Eagle’s medium (DMEM) or Opti-MEM (Gibco-BRL, Gaithersburg, MD),containing 10% fetal bovine serum (Atlanta Biological, Nordcross, GA) andantibiotics. All the data are representative of at least three independent experiments.For western immunoblotting, cells were lysed in immunoprecipitation buffer (50 mMTris-HCl pH 7.4, 150 mM NaCl, 0.5% NP-40, 1 mM EDTA, 25 mM NaF, 0.1 mMdithiothreitol (DTT), 1 �g/ml leupeptin, 10 �g/ml soybean trypsin inhibitor, 1 mMPMSF, 8 mM �-glycerophosphate, 10 mM Na3VO4, 1 �g/ml aprotinin). Lysates wereresolved by SDS-PAGE, transferred onto a PVDF membrane and visualized byluminescence (Amersham, Piscataway, NJ). Antibodies against the followingproteins were used: actin, FGFR3 and FRS2 (Santa Cruz Biotechnology, Santa Cruz,CA); ERK1/2, P-ERK1/2T202/Y204, STAT1, STAT3, STAT5, P-STAT1Y701, P-STAT3Y705, P-STAT5Y694, P-STAT1S727, P-STAT3S727 (Cell Signaling, Beverly,MA); FLAG (Sigma, St Louis, MO); SRC (Biosource, Camarillo, CA); and 4G10(Upstate Biotechnology, Lake Placid, NY).

Signal transduction studies, FGFR3 kinase assays and STATELISA-based EMSA assayCells were serum starved for 12 hours before treatment with 10 ng/ml FGF2, 20ng/ml IL6 (R&D Systems, Minneapolis, MN) and 40 ng/ml IFN� (Calbiochem, SanDiego, CA). When heparin (Gibco-BRL) was used, the concentration was 1 �g/ml.The FGFR3 kinase assays were carried out as described before (Krejci et al., 2007).Briefly, the kinase reactions were performed in 50 �l of kinase buffer (60 mM Hepes-NaOH pH 7.5, 3 mM MgCl2, 3 mM MnCl2, 3 �M Na3VO4, 1.2 mM DTT)supplemented with 2.5 �g PEG, 10 �M ATP and recombinant human STAT1 (500ng), STAT3 (250 ng; Active Motif, Carlsbad, CA), or FRS2 (500 ng; Abnova, TaipeiCity, Taiwan) as a substrate. The recombinant FGFR3 intracellular domain (E322-T806; Cell Signaling) was used at 300 ng per reaction. To obtain C-terminally FLAG-tagged human wt FGFR3 or FGFR3-K650E, vectors expressing both kinases weretransfected into CHO cells and the kinases were purified by immunoprecipitationwith FLAG antibody as described previously (Krejci et al., 2007).Immunocomplexes were washed with kinase buffer and subjected to the kinase assay.SU5402 was obtained from Calbiochem. The solutions used for STAT and FGFR3co-immunoprecipitation experiments have been described elsewhere (Bonofiglio etal., 2005). STATs were immunoprecipitated from 500 �g of total HeLa lysate using2 �g of STAT1 or STAT3 antibody (Santa Cruz). For STAT ELISA-based EMSAassay, growing RCS cultures were treated as desired and cell nuclei were isolatedusing a Nuclear Extract Kit (Active Motif) as directed by the manufacturer. TheELISA-based EMSA assay (TransAM Stat Assay kit; Active Motif) was used toquantify the amount of active STAT1 and STAT3 in nuclear extracts. Briefly, theactive STATs were purified from a nuclear lysate upon binding an immobilizedoligonucleotide containing a 5�-TTCCCGGAA-3� motif, and detected by ELISA.

Vectors, cell transfection and RNA interferenceVectors carrying C-terminally FLAG-tagged human wt or FGFR3-K650E have beendescribed elsewhere (Krejci et al., 2007). pRK7 vector containing C-terminallyFLAG-tagged kinase-dead FGFR3 was created by subcloning an FGFR3 segmentcontaining a kinase-inactivating K508M substitution from PFR3K508M (Raffioni etal., 1998) into the C-terminally FLAG-tagged human wt FGFR3. The vectorscontaining STAT1 or STAT3 C-terminally fused to GFP or YFP (Köstner and Hauser,1999; Herrmann et al., 2003) were obtained from H. Hauser and G. Müller-Newen,respectively. Vectors expressing wt and constitutively active (ca-)STAT1(A656C/N658C) were obtained from D. Frank and T. Ouchi, respectively (Liddle etal., 2006; Sironi and Ouchi, 2004). Vectors containing C-terminally FLAG-taggedwt STAT3 or ca-STAT3 mutant (A662C/N664C) were obtained from Addgene(Cambridge, MA). Vector containing ca-SRC kinase mutant (pUSEamp-SRC-Y529F) was obtained from Upstate Biotechnology. The following vectors were usedfor STAT luciferase reporter assays: pZLuc-TK-3xLy6e, pTATA-TK-Luc-4xM67

Jour

nal o

f Cel

l Sci

ence

280

(Addgene), pTAL-Luc-4xGas and pRL-TK (Promega, Madison, WI). Beforetransfection, the extracellular matrix was degraded by treatment with 0.3% bacterialcollagenase (type II; Invitrogen, Carlsbad, CA). Cells were transfected withFuGENE6 (Roche Diagnostics, Penzberg, Germany) or Lipofectamine 2000(Invitrogen) transfection reagent according to the manufacturer’s protocol.

RNA interferenceCells were transfected with siRNA using X-tremeGENE according to manufacturer’sprotocol (Roche). The siRNA transfection was carried out in 2 ml of transfectionmedium (Opti-MEM; Invitrogen) in six-well plates (Costar, Cambridge, MA). ThesiRNA concentration was 80 nM, the RNA/X-tremeGENE ratio (�g/�l) was 1:5 andthe amount of transfected cells was 1�105. The medium was changed 4 hours aftertransfection and the cells were cultured for 24 hours and either harvested for WB ortreated with FGF2 for 72 hours and counted. The siRNAs against ERK1/2, STAT1and STAT3 were purchased from Dharmacon (Lafayette, CO).

STAT luciferase reporter assayRCS chondrocytes were transfected with a plasmid containing ca-SRC or wt-STAT3or ca-STAT3, and a STAT firefly luciferase reporter plasmid and a control Renillareniformis luciferase plasmid in a 3:1.5:0.5 ratio using FuGENE6, according to theprotocol described above. Total amount of plasmid DNA was 5 �g used to transfect1�105 cells. When only the STAT reporter plasmid and the Renilla luciferase plasmidwere used, the total amount of transfected DNA was 4 �g and the plasmid ratio was3:1. Twenty-four hours after transfection, cells were treated with FGF2, IFN� or IL6for an additional 72 hours. The luciferase activity was determined using a Dual-Luciferase Reporter Assay kit (Promega).

Confocal microscopyCells were transfected with STAT1-GFP or STAT3-YFP as described above andgrown for 24 hours, treated as desired, fixed with 4% paraformaldehyde and mountedin medium (Vectashield; Vector Laboratories, Burlingame, CA) containing DAPI fornuclear staining. Confocal fluorescence and two photon laser scanned images weretaken on a Leica TCS-SP MP confocal and multiphoton inverted microscope(Heidelberg, Germany) equipped with an argon laser (488 nm blue excitation), adiode-pumped solid-state laser (561 nm yellow-green excitation), a helium-neonlaser (633 nm) and a two photon laser setup consisting of a Spectra-Physics MilleniaX 532 nm green diode pump laser and a Tsunami Ti-Sapphire picosecond pulsedinfrared laser tuned at 768 nm for UV excitation. Nomarski differential interferencecontrast (DIC) images were scanned using the argon laser for excitation.

We are grateful to H. Hauser, G. Müller-Newen, D. Frank and T.Ouchi for STAT vectors, and to P. Mekikian for an excellent technicalassistance. We also acknowledge the Carol Moss Spivak CellImaging Facility in the UCLA Brain Research Institute for the use oftheir confocal microscopes. This work was supported by NationalInstitutes of Health (5P01-HD22657), Yang Sheng Tang USAcompany, the Ministry of Education, Youth and Sports of the CzechRepublic (MSM0021622430), a Cedars-Sinai Medical GeneticsInstitute fellowship (P.K.), and a Winnick Family Scholars Award(W.R.W.).

ReferencesAikawa, T., Segre, G. V. and Lee, K. (2001). Fibroblast growth factor inhibits

chondrocytic growth through induction of p21 and subsequent inactivation of cyclin E-Cdk2. J. Biol. Chem. 276, 29347-29352.

Ben-Zvi, T., Yayon, A., Gertler, A. and Monsonego-Ornan, E. (2006). Suppressors ofcytokine signaling (SOCS) 1 and SOCS3 interact with and modulate fibroblast growthfactor receptor signaling. J. Cell Sci. 119, 380-387.

Besser, D., Bromberg, J. F., Darnell, J. E., Jr and Hanafusa, H. (1999). A single aminoacid substitution in the v-Eyk intracellular domain results in activation of Stat3 andenhances cellular transformation. Mol. Cell. Biol. 19, 1401-1409.

Bonofiglio, D., Gabriele, S., Aquila, S., Catalano, S., Gentile, M., Middea, E.,Giordano, F. and Ando, S. (2005). Estrogen receptor � binds to peroxisomeproliferator-activated receptor response element and negatively interferes withperoxisome proliferator-activated receptor � signaling in breast cancer. Clin. Cancer.Res. 11, 6139-6147.

Bromberg, J. F., Horvath, C. M., Wen, Z., Schreiber, R. D. and Darnell, J. E.(1996). Transcriptionally active Stat1 is required for the antiproliferative effects ofboth interferon alpha and interferon gamma. Proc. Natl. Acad. Sci. USA 93, 7673-7678.

Cao, X., Tay, A., Guy, G. R. and Tan, Y. H. (1996). Activation and association of Stat3with Src in v-Src-transformed cell lines. Mol. Cell. Biol. 16, 1595-1603.

Chen, L., Adar, R., Yang, X., Monsonego, E. O., Li, C., Hauschka, P. V., Yayon, A.and Deng, C. (1999). Gly369cys mutation to mouse FGFR3 causes achondroplasia byaffecting both chondrogenesis and osteogenesis. J. Clin. Invest. 104, 1517-1525.

Chin, Y. E., Kitagawa, M., Su, W.-C. S., You, Z.-H., Iwamoto, Y. and Fu, X.-Y. (1996).Cell growth arrest and induction of cyclin-dependent kinase inhibitor p21WAF1/Cip1

mediated by STAT1. Science 272, 719-722.

Journal of Cell Science 121 (3)

Dailey, L., Laplantine, E., Priore, R. and Basilico, C. (2003). A network oftranscriptional and signaling events is activated by FGF to induce chondrocyte growtharrest and differentiation. J. Cell Biol. 161, 1053-1066.

Decker, T. and Kovarik, P. (2000). Serine phosphorylation of Stats. Oncogene 19, 2628-2637.

Deng, C., Wynshaw-Boris, A., Zhou, F., Kuo, A. and Leder, P. (1996). Fibroblast growthfactor receptor 3 is a negative regulator of bone growth. Cell 84, 911-921.

Deo, D. D., Axelrad, T. W., Robert, E. G., Marcheselli, V., Bazan, N. G. and Hunt, J.D. (2002). Phosphorylation of Stat-3 in response to basic fibroblast growth factor occursthrough a mechanism involving platelet-activating factor, JAK-2, and Src in humanumbilical vein endothelial cells. Evidence for a dual kinase mechanism. J. Biol. Chem.277, 21237-21245.

Durbin, J. E., Hackenmiller, R., Simon, M. C. and Levy, D. E. (1996). Targeteddisruption of the mouse Stat1 gene results in compromised innate immunity to viraldisease. Cell 84, 443-450.

Henderson, J. E., Naski, M. C., Aarts, M. M., Wang, D., Cheng, L., Goltzman, D. andOrnitz, D. M. (2000). Expression of FGFR3 with the G380R achondroplasia mutationinhibits proliferation and maturation of CFK2 chondrocytic cells. J. Bone Miner. Res.15, 155-165.

Herrmann, A., Sommer, U., Pranada, A. L., Giese, B., Haan, S., Becker, W., Heinrich,P. C. and Müller-Newen, G. (2003). Stat3 is enriched in nuclear bodies. J. Cell Sci.116, 339-349.

Khan, K. D., Shuai, K., Lindwall, G., Maher, S. E., Darnell, J. E., Jr and Bothwell,A. L. (1993). Induction of the Ly-6A/E gene by interferon alpha/beta and gammarequires a DNA element to which a tyrosine-phosphorylated 91-kDa protein binds. Proc.Natl. Acad. Sci. USA 90, 6806-6810.

Kim, H. and Baumann, H. (1997). The carboxyl-terminal region of STAT3 controls geneinduction by the mouse haptoglobin promoter. J. Biol. Chem. 272, 14571-14579.

Kisseleva, T., Bhattacharya, S., Braunstein, J. and Schindler, C. W. (2002). Signalingthrough the JAK/STAT pathway, recent advances and future challenges. Gene 285, 1-24.

Klint, P., Kanda, S., Kloog, Y. and Claesson-Welsh, L. (1999). Contribution of Src andRas pathways in FGF-2 induced endothelial cell differentiation. Oncogene 18, 3354-3364.

Köstner, M. and Hauser, H. (1999). Dynamic redistribution of Stat1 protein in IFNsignaling visualized by GFP fusion proteins. Eur. J. Biochem. 260, 137-144.

Krejci, P., Bryja, V., Pachernik, J., Hampl, A., Pogue, R., Mekikian, P. and Wilcox,W. R. (2004). FGF2 inhibits proliferation and alters the cartilage-like phenotype of RCScells. Exp. Cell Res. 297, 152-164.

Krejci, P., Masri, B., Fontaine, V., Mekikian, P. B., Weis, M., Prats, H. and Wilcox,W. R. (2005). Interaction of fibroblast growth factor and C-natriuretic peptide signalingin regulation of chondrocyte proliferation and extracellular matrix homeostasis. J. CellSci. 118, 5089-5100.

Krejci, P., Masri, B., Salazar, L., Farrington-Rock, C., Prats, H., Michels Thompson,L. and Wilcox, W. R. (2007). Bisindolylmaleimide I suppresses fibroblast growthfactor-mediated activation of Erk MAP kinase in chondrocytes by preventing Shp2association with the Frs2 and Gab1 adaptor proteins. J. Biol. Chem. 282, 2929-2936.

Legeai-Mallet, L., Benoist-Lasselin, C., Delezoide, A., Munnich, A. and Bonaventure,J. (1998). Fibroblast growth factor receptor 3 mutations promote apoptosis but do notalter chondrocyte proliferation in thanatophoric dysplasia. J. Biol. Chem. 273, 13007-13014.

Legeai-Mallet, L., Benoist-Lasselin, C., Munnich, A. and Bonaventure, J. (2004).Overexpression of FGFR3, Stat1, Stat5 and p21Cip1 correlates with phenotypic severityand defective chondrocyte differentiation in FGFR3-related chondrodysplasias. Bone34, 26-36.

Li, C., Chen, L., Iwata, T., Kitagawa, M., Fu, X. and Deng, C. (1999). A lys644glusubstitution in fibroblast growth factor receptor 3 (FGFR3) causes dwarfism in mice byactivation of STATs and ink4 cell cycle inhibitors. Hum. Mol. Genet. 8, 35-41.

Liddle, F. J., Alvarez, J. V., Poli, V. and Frank, D. A. (2006). Tyrosine phosphorylationis required for functional activation of disulfide-containing constitutively active Statmutants. Biochemistry 45, 5599-5605.

Lievens, P. M.-J. and Liboi, E. (2003). The thanatophoric dysplasia type II mutationhampers complete maturation of fibroblast growth factor receptor 3 (FGFR3), whichactivates signal transducer and activator of transcription 1 (STAT1) from theendoplasmic reticulum. J. Biol. Chem. 278, 17344-17349.

Murakami, S., Balmes, G., McKinney, S., Zhang, Z., Givol, D. and De Crombrugghe,B. (2004). Constitutive activation of MEK1 in chondrocytes causes Stat1-independentachondroplasia-like dwarfism and rescues the Fgfr3-deficient mouse phenotype. GenesDev. 18, 290-305.

Nowroozi, N., Raffioni, S., Wang, T., Apostol, B. L., Bradshaw, R. A. and MichelsThompson, L. (2005). Sustained Erk1/2 but not STAT1 or 3 activation is required forthanatophoric dysplasia phenotypes in PC12 cells. Hum. Mol. Genet. 14, 1529-1538.

Passos-Bueno, M. R., Wilcox, W. R., Jabs, E. W., Sertie, A. L., Alonso, L. G. andKitoh, H. (1999). Clinical spectrum of fibroblast growth factor receptor mutations.Hum. Mutat. 14, 115-125.

Priore, R., Dailey, L. and Basilico, C. (2006). Downregulation of Akt activity contributesto the growth arrest induced by FGF in chondrocytes. J. Cell. Physiol. 207, 800-808.

Raffioni, S., Zhu, Y. Z., Bradshaw, R. A. and Thompson, L. M. (1998). Effect oftransmembrane and kinase domain mutations on fibroblast growth factor receptor 3chimera signaling in PC12 cells. A model for the control of receptor tyrosine kinaseactivation. J. Biol. Chem. 273, 35250-35259.

Raucci, A., Laplantine, E., Mansukhani, A. and Basilico, C. (2004). Activation of theErk1/2 and p38 mitogen-activated protein kinase pathways mediates fibroblast growthfactor-induced growth arrest of chondrocytes. J. Biol. Chem. 279, 1747-1756.

Jour

nal o

f Cel

l Sci

ence

281STATs in FGF signaling in cartilage

Reich, N. C. and Liu, L. (2006). Tracking STAT nuclear traffic. Nat. Rev. Immunol. 6,602-612.

Rozenblatt-Rosen, O., Mosonego-Ornan, E., Sadot, E., Madar-Shapiro, L., Sheinin,Y., Ginsberg, D. and Yayon, A. (2002). Induction of chondrocyte growth arrest byFGF: transcriptional and cytoskeletal alterations. J. Cell Sci. 115, 553-562.

Sahni, M., Ambrosetti, D., Mansukhani, A., Gertner, R., Levy, D. and Basilico, C.(1999). FGF signaling inhibits chondrocyte proliferation and regulates bonedevelopment through the STAT-1 pathway. Genes Dev. 13, 1361-1366.

Sahni, M., Raz, R., Coffin, J. D., Levy, D. and Basilico, C. (2001). Stat1 mediates theincreased apoptosis and reduced chondrocyte proliferation in mice overexpressingFGF2. Development 128, 2119-2129.

Shi, C.-S. and Kehrl, J. H. (2004). Pyk2 amplifies epidermal growth factor and c-Src-induced Stat3 activation. J. Biol. Chem. 279, 17224-17231.

Sims, N. A., Jenkins, B. J., Quinn, J. M., Nakamura, A., Glatt, M., Gillespie, M. T.,Ernst, M. and Martin, T. J. (2004). Glycoprotein 130 regulates bone turnover andbone size by distinct downstream signaling pathways. J. Clin. Invest. 113, 379-389.

Sironi, J. J. and Ouchi, T. (2004). Stat1-induced apoptosisis mediated by caspases 2, 3,and 7. J. Biol. Chem. 279, 4066-4074.

Su, S. W., Kitagawa, M., Xue, N., Xie, B., Garofalo, S., Cho, J., Deng, C., Horton, W.A. and Fu, X. (1997). Activation of Stat1 by mutant fibroblast growth-factor receptorin thanatophoric dysplasia type II dwarfism. Nature 386, 288-292.

Wen, Z., Zhong, Z. and Darnell, J. E., Jr (1995). Maximal activation of transcription byStat1 and Stat3 requires both tyrosine and serine phosphorylation. Cell 82, 241-250.

Yamanaka, Y., Tanaka, H., Koike, M., Nishimura, R. and Seino, Y. (2003). PTHrPrescues ATDC5 cells from apoptosis induced by FGF receptor 3 mutation. J. BoneMiner. Res. 18, 1395-1403.

Jour

nal o

f Cel

l Sci

ence