ERK signaling is a central regulator for BMP-4dependent capillary sprouting

Qian Zhou a, Jennifer Heinke a, Alberto Vargas a, Stephan Winnik a, Tobias Krauss a,Christoph Bode a, Cam Patterson b, Martin Moser a,⁎

a Department of Cardiology, University of Freiburg, Germanyb Division of Cardiology, Carolina Cardiovascular Biology Center, Chapel Hill, NC, USA

Received 2 November 2006; received in revised form 6 August 2007; accepted 6 August 2007

Time for primary review 21 daysAvailable online 11 August 2007

Abstract

Objective: Bone Morphogenetic Protein-4 (BMP-4) and Extracellular-Signal Regulated Kinases (ERK) play crucial roles in vascular diseases.Here, we demonstrate that BMP-4 not only signals through the classical Smad cascade but also activates ERK phosphorylation as an alternativepathway in human umbilical vein endothelial cells (HUVEC) and that Smad and ERK pathways communicate through signal crosstalk.Methods: HUVECs were treated with BMP-4 and/or MEK inhibitors. Smad 6 and constitutively active (ca) MEK1 were overexpressed. Lossof function of Smad 4 and Smad 6 was achieved by specific siRNA transfection. Cell lysates were analyzed by western blotting for Smad andERK phosphorylation. HUVEC spheroids were generated for angiogenesis quantification.Results: Treatment with BMP-4 results in a dose- and time-dependent activation of the MEK–ERK 1/2 pathway in addition to activation ofthe Smad pathway and is blocked by MEK inhibitors. Quantitative in-gel angiogenesis assays in the presence or absence of MEK inhibitorsdemonstrate that ERK signals are necessary for BMP-4 induced capillary sprouting. Furthermore sprouting is not blocked by inhibition of theSmad signaling pathway. Overexpression of the inhibitory Smad 6 inhibits ERK phosphorylation and ERK-induced capillary sprouting,whereas loss of function of Smad 4 has no effect.Conclusions: We demonstrate that ERK1/2 functions as an alternative pathway in BMP-4 signaling in HUVECs. Capillary sprouting induced byBMP-4 is dependent onERKphosphorylation. ERK is essential for efficient transduction ofBMPsignals and serves as a positive feedbackmechanism.On the other hand, stimulation of Smad 6 inhibits ERK activation and thus results in a negative feedback loop to fine-tune BMP signaling inHUVECs.© 2007 European Society of Cardiology. Published by Elsevier B.V. All rights reserved.

Keywords: BMP-4; ERK 1/2; In-gel angiogenesis; HUVEC; Signal transduction

This article is referred to in the Editorial by M.T. Rizzo(pages 375–376) in this issue.

Bone morphogenetic proteins (BMPs) are members of thetransforming growth factor-β (TGF-β) superfamily. Theyhave originally been identified by their ability to induceectopic bone and cartilage formation in rats [23]. Today morethan 20 BMP-related proteins have been identified, and it

becomes more and more apparent that BMPs are multifunc-tional proteins with a wide range of biological activities onvarious cell types. With respect to vascular development,BMP-2 and BMP-4 appear to be the most important familymembers [17].

BMPs signal through cell surface complexes of type I andtype II serine/threonine kinase receptors [21]. The type IIreceptor kinase is constitutively active and phosphorylatesthe type I receptor kinase after ligand binding. Onceactivated, the type I and type II receptor kinases formheterodimers and mediate intracellular signaling through thecentral signal messengers— the Smad (small mother againstdecapentaplegic) proteins [8].

Cardiovascular Research 76 (2007) 390–399www.elsevier.com/locate/cardiores

⁎ Corresponding author. Department of Cardiology, University ofFreiburg, Hugstetterstr. 55, 79106 Freiburg, Germany. Tel.: +49 761 2703401; fax: +49 761 270 3254.

E-mail address: moserm@medizin.ukl.uni-freiburg.de (M. Moser).

0008-6363/$ - see front matter © 2007 European Society of Cardiology. Published by Elsevier B.V. All rights reserved.doi:10.1016/j.cardiores.2007.08.003

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from

The Smad proteins form a group of transcription factorsmediating TGF-β andBMP signaling [27]. They are subdividedinto three classes, including receptor-regulated Smads (R-Smads), common-partner Smads (Co-Smads) and inhibitorySmads (I-Smads). To date, three R-Smads (Smad 1/5/8) areknown to mediate BMP signaling upon phosphorylation byBMP type I receptors. After activation, R-Smads release rapidlyfrom the receptor to form hetero-oligomeric complexes with theCo-Smad (Smad 4). The complex then translocates into thenucleus to regulate transcription of various target genes [3]. As anegative feedback loop, I-Smads (Smad 6 and 7) are activated.While Smad 6 inhibits BMP and TGF-β signaling with similarpotency, Smad 7 inhibits TGF-β signaling more efficiently [9].Smad 6 expression is induced by Smads 1 and 5 as well as otherstimuli, e.g. fluid shear stress. It inhibits complex formationbetween Smad 1 and Smad 4. When overexpressed at highconcentrations, Smad 6 also inhibits BMP signaling byinterfering with the R-Smads [6].

Findings over the past few years have suggested that BMPresponses result not only from activation of the Smad cascade,but from interactions among multiple signaling mechanisms[4]. Among these cascades, the MAPK (mitogen activatedprotein kinase) has been proposed to have important functionsunder physiological as well as pathological conditions [11].

The MAPK pathway refers to a cascade of proteinkinases that are highly conserved in evolution. There aredifferent isoforms, among which the Extracellular-SignalRegulated Kinase (ERK) is of great importance in regulatingproliferation and differentiation [2]. ERK is activated by avariety of extracellular agents, including growth factors,hormones and neurotransmitters. Upon stimulation, adaptorproteins are recruited to the plasma membrane, where theyinduce the activation of small GTP-binding proteins (e.g.Ras). Signals are then transmitted to downstream proteins ofthe cascade, including Raf and MEK. The activated MEKsare dual specificity kinases with a unique selectivity towardsERK [1]. Depending on cell type- and target gene-spe-cificity, activated ERK pathways have been reported toenhance [7] or inhibit [15] Smad activity.

Although both BMP-4 and ERK have been proposed toplay crucial roles in the pathophysiology of vasculardiseases [5], nothing is yet known about interactions ofthese pathways in endothelial cell lines. Here, we de-monstrate that BMP-4 activates the ERK 1/2 cascade inHUVECs in parallel with the classical Smad pathway andthat both signaling pathways communicate through signalcrosstalk.

1. Materials and methods

1.1. Materials

BMP-4 was purchased from Chemicon International. Allantibodies and the MEK inhibitors PD98059 and U0126were purchased from Cell Signaling Technology. Reagentswere dissolved as recommended by the manufacturer. Cell

culture media and reagents were from Cambrex BioScience,Walkersville and PromCell, Heidelberg.

1.2. Cell culture

The investigation conforms with the principles outlined inthe Declaration of Helsinki for the use of human tissues.HUVECs were kindly provided by H. Augustin, Freiburg,Germany. Cells were seeded in 75 cm2 plastic culture disheswith 10 ml EBM containing EGF Single Quots. The mediumwas supplemented with hEGF, hydrocortisone, GA 1000,BBE and 10% fetal calf serum. Cells were incubated at 37 °Cin an atmosphere containing 5% CO2 and used for exper-iments before they reached passage 6. All experiments wereperformed in triplicate using two different isolates ofHUVECs.

1.3. Transient transfection of cells

Smad 6 expression construct pcDNA3-Flag(N)-mSmad 6was generously provided by T. Imamura, Tokyo, Japan [9].For transient transfections, DNA plasmids were introducedinto HUVECs using FuGene6 transfection reagent accordingto the manufacturer's instructions (Roche Applied Science).Transfection was confirmed using primary antibody againstthe Flag tag (Stratagene).

Adenovirus coding for flag-tagged Smad 6 (AdSmad 6)was kindly provided by M.-J. Goumans, Amsterdam,Netherlands. Infection with Smad 6 or recombinant adeno-virus encoding green fluorescent protein (AdGFP) as controlwas performed at 1!109 total virus particles/ml in 2% FCS-containing medium. Infected cells were then cultured for anadditional 24 h before in-vitro angiogenesis assay wasperformed. Infection efficacy was confirmed by fluorescentmicroscopy for GFP and by using primary antibody againstthe Flag tag (Stratagene).

Adenovirus coding for constitutively active MEK1(caMEK1) and GFP under the control of a bicistronicpromoter was kindly provided by K. Takimoto, USA [12].Infection with caMEK1 or control virus was performed at1!109 total virus particles/ml in serum-free medium for12 h. Infected cells were then cultured for 48 h in 2% FCS-containing medium. Successful infection of more than85% of cells was confirmed by fluorescent microscopy forGFP.

1.4. Stimulation and inhibition experiments

Cells grown to 80% confluence in 35 mm dishes wereserum starved for 2 h in OptiMEM. BMP-4 was dissolved asindicated by the manufacturer and was directly added to thecells. After gentle swirling, cells were incubated for 20 min,unless otherwise indicated. Controls were treated with bufferonly. For inhibition experiments, appropriate concentrationsof the MEK inhibitors were directly added to cells 30 minprior to the addition of BMP-4.

391Q. Zhou et al. / Cardiovascular Research 76 (2007) 390–399

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from

1.5. Western blotting

After treatment, cells were washed twice with ice-coldphosphate buffered saline (PBS), lysed on ice in RIPA bufferand then centrifuged at 1600 g for 1 min at 4 °C to removeinsoluble material. The supernatant was collected and totalcellular protein was quantified using Bradford protein assay.Equal amounts of protein (30–50 μg) were then loaded andseparated by a 15% SDS-PAGE. Electroblotting was per-formed to transfer proteins to nitrocellulose. After blockingwith TBST supplemented with non-fat dried milk, themembrane was incubated overnight at 4 °C with theappropriate primary antibody. Specific proteins weredetected using horseradish peroxidase-conjugated goatanti-rabbit IgG (Dako) and visualized by an ECL system(Amersham Bioscience).

1.6. In vitro angiogenesis assay

HUVEC spheroids were generated as published [13].Briefly, HUVEC were grown as hanging drops of 500 cellseach for 24 h in a cell culture incubator. Spheroids wereplaced into a collagen gel generated by the addition of pH 7.4adjusted collagen extract containing carboxymethylcelluloseat a ratio of 1:1 to prevent sedimentation of spheroids prior topolymerization of the collagen gel. After adding spheroids,the gel was rapidly aliquotted into wells of a 24 well plate andincubated at 37 °C for 24 h to polymerize. For stimulation orinhibition experiments BMP-4 protein or MEK inhibitorswere added to the gel before polymerization started. Toquantify in-gel angiogenesis the cumulative length of allcapillary-like sprouts originating from the core plain of anindividual spheroid was measured at 10!magnification using

Fig. 1. Dose and time dependence of intracellular signaling induced by BMP-4. (A): BMP-4 results in dose-dependent ERK and MEK phosphorylation, while totalERK and total Smad 5 remain constant. (B) Time course of BMP-4 activated signaling pathways. HUVECswere treatedwith BMP-4 (50 ng/ml) for the indicated times.(C andD)Densitometric analysis of n=3western blots. Smad phosphorylation and ERKphosphorylation reach their maximum at 20min as shown in the densitometricanalysis of the time course, whereas only Smad phosphorylation is significantly increased at 10 min already. ⁎ indicates pb0.05.

392 Q. Zhou et al. / Cardiovascular Research 76 (2007) 390–399

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from

a digitized imaging system. At least 10 spheroids percondition were analyzed.

1.7. Small interfering RNA (siRNA) transfection

Smad 4-siRNAs (siSmad4) Smad 6-siRNA (siSmad6)and their complementary RNA strands were purchased fromInvitrogen. The sequences were siSmad4: (forward: 5′-GGAAUUGAU CUC UCAGGAUTTand reverse 5′-AUC CUGAGA GAU CAA UUC CTT), siSmad6: (forward: 5′-CCACAU UGU CUU ACA CUG AAA CGG A and reverse: 5′-UCC GUU UCA GUG UAA GAC AAU GUG G). siRNAtransfection was performed according to the manufacturer'sprotocol (Invitrogen). Transfection efficiency was confirmedby quantitative real-time PCR.

1.8. Real-time PCR

Quantitative real-time PRC analysis was performed usingthe real-time PCR detection system (Bio-Rad) withsequence-specific primer pairs for Smad 4 (forward primer:5′-CAC AGG ACA GAA GCC ATT GAGAGA and reverseprimer: 5′-CGG TGA CAC TGA CGC AAA TCA AAG);Smad 6 (forward primer: 5′-CCC CCGGCTACT CCATCAAGG TGT and reverse primer: 5′-GTC CGT GGG GGCTGT GTC TCT GG). Quantification was performed usinglightcycler software. Total RNAwas extracted from HUVECusing the RNA Mini Kit (Bio-Rad). Reverse transcriptionwas performed with iScript cDNA-Kit (Bio-Rad). Knock-down efficiency was calculated using theΔΔCt method. Thehousekeeping gene human RNA-Polymerase was used forinternal normalization (Ambion).

1.9. Statistical analysis

All experiments were performed in triplicate. Western blotsshown are representative for at least three independentexperiments. Densitometric analysis was performed usingQuantity One® 1-D Analysis Software Version 4.4, Bio-Rad.Relative densitometry results for phosphorylated MAP kinaseand Smad signals were correlated for total MAP kinase orSmad signals (fold increase). Quantification of in-gelangiogenesis assay was performed using ImageJ 1.37vprovided by the NIH. Data were expressed as mean±SEMand comparisons were calculated by Student's t-test orMann–Whitney U-Test respectively. Statistical analysis was per-formed using GraphPad Prism 4.0, USA. A p value of b0.05indicated statistical significance.

2. Results

2.1. BMP-4 induces activation of the MEK–ERK pathway

In a search for non-Smad-dependent signaling pathwaysinduced by BMP-4 that are required for maximal activation, weinvestigated the MAPK pathway in HUVECs. In cells

incubated with increasing concentrations of BMP-4 for20 min, we observed a dose-dependent phosphorylation ofERK 1/2 and its activator MEK (Fig. 1A). The amount ofphospho ERK increased 5.6 fold after stimulation with 50 ng/ml BMP-4 compared to untreated cells. Concentrations of totalERK were not affected by BMP-4 stimulation. To investigatethe time course of signal transduction after BMP-4 stimulation,we incubated the cells for different periods of time. As shown inFig. 1B–D, both Smad 1/5 and ERK 1/2 phosphorylationreached a maximum at 20 min. At 10 min Smad phosphory-lation attained already almost maximal concentrations, whileERK phosphorylation remained still relatively low. These dataindicate that BMP-4 induces ERK phosphorylation inHUVECs in a dose- and time-dependent manner, with Smadphosphorylation occurring faster than ERK phosphorylation.

To specify the upstream mediator of BMP-4-inducedERK activation, HUVECs were pre-incubated with the MEKinhibitors PD98059 and U0126. Pre-treatment with either ofthe MEK inhibitors for 60 min blocked BMP-4 inducedphosphorylation of ERK 1/2. In comparison to phosphoERK, the phosphorylation of Smad 1/5 was not influencedsignificantly by the MEK inhibitors (Fig. 2). These resultsindicate that BMP-4 induced phosphorylation of ERK 1/2 inHUVECs is MEK dependent, but MEK activation is notnecessarily required to activate the Smad 1/5 pathway.

2.2. BMP-4 induces sprouting of capillaries

In order to investigate the effect of BMP-4 on capillarysprouting, HUVEC spheroids were seeded in collagen gelsprepared with VEGF, BMP-4, or control. After 24 h,capillary sprouts grew radially in all three dimensions.Collagen gel-embedded HUVEC spheroids under controlconditions showed a low level of spontaneous sprouting withan average sprout length of 70 μm/spheroid (Fig. 3A; F).HUVEC spheroids were readily responsive to stimulation byincreasing doses of BMP-4 (Fig. 3B; F). Sprout lengthincreased from 70 μm/spheroid under control conditions to157 μm/spheroid after stimulation with 100 ng/ml BMP-4

Fig. 2. Inhibition of BMP-4 induced ERK 1/2 activation by MEK inhibitors.Cells were pre-treated with theMEK inhibitors for 60min prior to the addition ofBMP-4 (50 ng/ml). BMP-4 induces strong ERK phosphorylation. This effect islost in the presence ofMEK inhibitors. Representative blot for three experiments.

393Q. Zhou et al. / Cardiovascular Research 76 (2007) 390–399

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from

(pb0.05) and 262 μm/spheroid with 200 ng/ml BMP-4(pb0.05). This effect was comparable to stimulation byVEGF (Fig. 3C; F), which was used as a positive control, and

induced outgrowth to 280 μm/spheroid and 327 μm/spheroidat concentrations of 25 ng/ml and 50 ng/ml, respectively.These data indicate that BMP-4 exerts stimulative effects on

Fig. 3. Effect of BMP-4 and ERK 1/2 on capillary sprouting. HUVEC spheroids were seeded in collagen gels preparedwith control (A), BMP-4 200 ng/ml (B) andVEGF 50 ng/ml (C). (F) HUVEC spheroids were readily responsive to increasing doses of BMP-4 and VEGF. (D) The BMP-4 effect was strongly reduced afterpreincubation with the MEK inhibitor U0126 (10 μM). (G) Quantitative analysis of sprouts after stimulation with BMP-4 and with or without U0126. ⁎ Indicatespb0.05 compared to control or BMP-4+U0126. (E, H) BMP-4 (50 ng/ml) induced capillary sprouting was not affected by siRNA knock down of Smad 4[siSmad4]. ⁎ Indicates pb0.05 compared to transfected spheroids without BMP-4. [siNeg=non-specific control siRNA].

394 Q. Zhou et al. / Cardiovascular Research 76 (2007) 390–399

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from

capillary sprouting in HUVEC that are similar to the effectinduced by the angiogenic growth factor VEGF providingfurther support for the pro-angiogenic effect of BMP-4 inHUVECs.

2.3. ERK 1/2 is essential for capillary sprouting

Recent experiments have revealed the importance of ERK-MAPK signaling in angiogenesis. Given the observations thatBMP-4 activates ERK 1/2 in HUVECs and is also capable toinduce capillary sprouting, we sought to investigate the role ofBMP-4 induced ERK 1/2 phosphorylation in mediatingcapillary sprouting. HUVEC spheroids were embedded incollagen gels with or without the MEK inhibitor U0126 andBMP-4. As shown in Fig. 3D and G, these spheroids show asignificant reduction of total sprout length by 73.7% (pb0.05)in comparison to spheroids incubated with BMP-4 alone. Thus

endothelial cell sprouting is dependent on ERK signaling uponstimulation with BMP-4.

2.4. Smad 4 is not necessary for capillary sprouting

To investigate if endothelial cell sprouting is dependent onthe Smad pathway, the expression of the Co-Smad Smad 4wassilenced in HUVECs employed for the sprouting assay. Smad4 forms complexes with the R-Smads, and is necessary fornuclear translocation of the Smad complex, and thus for targetgene transcription [6]. Consequently, loss of function of Smad4 results in impairment of the Smad pathway. We achieved aknock-down efficacy of 82.7% as confirmed by quantitativereal-time PCR. As shown in Fig. 3E and H, total sprout lengthof BMP-4 stimulated spheroids was not affected by specificsilencing of Smad 4 compared to HUVEC treated with controlsiRNA. Thus, Smad 4 and downstream Smad signaling is not

Fig. 4. Smad 6 blocks ERK 1/2 phosphorylation and inhibits endothelial sprouting. (A) HUVECs were transfected with FLAG-Smad 6 expression vector andtreated with BMP-4 (50 ng/ml). Overexpression of Smad 6 does not result in non-specific Smad 1/5 phosphorylation. BMP-4 induced ERK phosphorylation iscompletely blocked when Smad 6 is overexpressed. (B) Representative HUVEC spheroid transduced with Smad 6 overexpressing adenovirus. (C) RepresentativeHUVEC spheroid that was transduced to overexpress Smad 6 and stimulated with BMP-4. BMP-4 induced sprouting was blocked by Smad 6. (D) Quantitativeanalysis of the experiments performed in (B) and (C). (E) siRNA knock down of Smad 6 results in increased sprouting compared to control transfected spheroids.⁎ Indicates pb0.05 compared to control.

395Q. Zhou et al. / Cardiovascular Research 76 (2007) 390–399

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from

necessary for endothelial cell sprouting. These data supportour findings that ERK signaling is the central mediator inBMP-4 induced HUVEC sprouting.

2.5. Smad 6 inhibits phosphorylation of ERK 1/2 andendothelial cell sprouting

As we found induction of both the Smad as well as theMAPK cascade after stimulation of HUVECs with BMP-4,we aimed to investigate potential interactions between thesepathways. To inhibit the Smad pathway, we overexpressedSmad 6 — a well characterized inhibitory Smad. Recentresults indicate that Smad 6 elicits its effect by inhibiting thecomplex formation of Smad 1 with Smad 4, resulting inreduced expression of Smad target genes [10]. It was alsoreported that when overexpressed at higher concentrations,Smad 6 prevents the activation of Smad 1 and Smad 5 byinteracting with the type I BMP receptor [6,9]. To exclude thisrather non-specific effect of Smad 6 on the BMP receptor levelin our experiment, Smad 6 transfected cells were stimulatedwith BMP-4 and Smad 1/5 phosphorylation was detected byWestern blotting. As shown in Fig. 4A, the BMP-4 inducedSmad 1/5 phosphorylation in Smad 6 transfected cells wasequal compared to HUVECs transfected with control vector.Thus, overexpression of Smad 6 did not affect Smad 1/5phosphorylation, excluding an upstream effect of Smad 6 at thereceptor level in our experiments. To investigate crosstalkbetween the Smad and ERK pathway, we analyzed ERK 1/2phosphorylation in HUVECs upon stimulation with BMP-4,after blocking Smad signaling by overexpression of Smad 6(Fig. 4A).As expected, incubationwith BMP-4 increasedERK1/2 phosphorylation in HUVEC. In contrast, ERK phosphor-ylation upon stimulation with BMP-4was strongly inhibited inHUVECs overexpressing Smad 6. Even at baseline, withoutBMP-4 stimulation, ERK phosphorylation was lower in Smad6 overexpressing cells. These results suggest an interactionbetween the Smad and the ERK pathway and indicate thatSmad 6 not only inhibits the Smad pathway, but also elicitsinhibitory effects on ERK signaling. To functionally charac-terize the crosstalk between the Smad and the ERK pathway,the effect of Smad 6 on capillary sprouting was investigated.HUVECswere transducedwithAdSmad6 and then seeded intocollagen gels prepared with BMP-4 or control. As shown inFig. 4B–D, BMP-4 induced endothelial cell sprouting wasblocked by Smad 6 overexpression down to the level ofunstimulated spheroids. Along the same lines of evidence, lossof Smad 6 by siRNA knock down allows for enhancedendothelial cell sprouting (Fig. 4E). Together with the datapresented above, this observation supports the hypothesis, thatERK 1/2 signaling is essential for capillary sprouting and thatthis pathway is negatively regulated by Smad 6.

2.6. Smad 6 does not inhibit MEK phosphorylation

To investigate at which step in the cascade Smad 6interferes with the MEK/ERK cascade we analyzed the

phosphorylation of MEK and ERK individually. As shownin Figs. 1 and 4 BMP-4 induces phosphorylation of bothMEK and ERK 1/2. As MEK is the upstream mediator ofERK and ERK phosphorylation is inhibited by Smad 6 weasked if MEK phosphorylation was also affected by Smad 6.As shown in Fig. 4A Smad 6 does not interfere with MEKphosphorylation. These data indicate that Smad 6 differen-tially inhibits ERK but not MEK.

2.7. Smad 6 expression in HUVEC

We have used overexpression of Smad 6 as a tool todissect crosstalk between ERK and Smad signaling andfound that Smad 6 inhibits not only Smad but also ERKsignaling. To investigate if endogenous Smad 6 is capable toregulate these pathways under physiologic conditions weanalyzed the expression of Smad 6 in HUVEC cells. Asdetected by rt-PCR we found rather low levels of Smad 6RNA in unstimulated HUVECs but after stimulation withBMP-4 Smad 6 transcription was upregulated (Supplemen-tal Fig. 1). This indicates that sufficient amounts of Smad 6are present in endothelial cells after stimulation with BMP-4to regulate the crosstalk between the Smad and ERKpathways.

2.8. caMEK1 activates the phosphorylation of Smad 1/5

After determining that inhibition of Smad signaling hasinhibitory effects on ERK phosphorylation, we sought toinvestigate the role ofMAPK signaling on the Smad cascade inHUVEC. We overexpressed a constitutively active MEK1mutant (caMEK1) in HUVECs and analyzed the effect ofcaMEK1 on Smad 1/5 phosphorylation. Transfection efficacywas confirmed by visualizing bicistronic GFP-expression ofcaMEK1-virus by fluorescent microscopy. Furthermore ERK1/2 phosphorylation was detected as a positive control. Asshown in Fig. 5, Smad 1/5 was activated in HUVECsoverexpressing caMEK1, while no Smad 1/5 phosphorylation

Fig. 5. MEK1 stimulates Smad 1/5 phosphorylation. HUVECs weretransfected with caMEK1 adenovirus or control virus. Blots were probedwith phospho ERK 1/2 antibody to confirm functional transfection ofcaMEK. Smad 1/5 is strongly phosphorylated in caMEK overexpressing butnot in control cells. Total ERK serves as loading control.

396 Q. Zhou et al. / Cardiovascular Research 76 (2007) 390–399

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from

was detected in control cells. This indicates that caMEK1 iscapable to activate Smad 1/5.

3. Discussion

Findings over the past few years show that BMP-4 hasimportant functions during early embryogenesis andvascular development [17]. BMP-2, 4 and 7 have beendemonstrated to inhibit aortic and pulmonary vascularsmooth muscle cell proliferation [16,18], but the effect ofBMP-4 on endothelial cells has not been well characterized,yet. BMP receptors are divided into type I and type IIreceptors, each comprising different receptor subtypes.Generally, the type I receptor is activated by the type IIreceptor after ligand binding and then initiates intracellularsignaling by phosphorylation of specific receptor-regulatedR-Smads, i.e. Smad 1/5 [21]. In the present study, wedemonstrate in HUVECs that BMP-4 not only activates theclassical Smad 1/5 pathway, but also activates ERK 1/2 in adose- and time-dependent fashion. Although activation ofSmad signaling occurs faster than activation of ERK, at20 minutes both pathways are fully stimulated (Fig. 1).Furthermore the direct upstreammediator of ERK activationMEK is phosphorylated in parallel to ERK. Accordingly,BMP-4 stimulated ERK activation can be blocked by theMEK inhibitors PD98059 and U0126, confirming that MEKis an upstream mediator of ERK phosphorylation afterstimulation with BMP-4 (Fig. 2).

BMP signaling has been implicated in important cellularfunctions such as endothelial cell sprouting [24]. Todetermine the functional impact of Smad and ERK mediatedBMP-4 signaling in HUVEC, we have established a capillarysprouting assay. Quantitative analyses show that BMP-4elicits pro-angiogenic effects, which are similar to the effectsinduced by VEGF. This effect was blocked when cells werepre-treated with the MEK inhibitor, indicating that ERK 1/2is essential for capillary sprouting in HUVECs (Fig. 3). Theimportance of ERK 1/2 itself on HUVEC differentiation hasbeen examined by Yang and colleagues. In a 3D gel matrixthey show that inhibition of ERK by the MEK inhibitorPD98059 does not interfere with HUVEC differentiation[25]. This finding indicates that the anti-angiogenic effect inour experiment is not likely to be the result of ERK inhibitionitself, but rather because ERK mediated activation of BMP-4is inhibited. In order to inhibit Smad signaling we over-expressed the endogenous Smad inhibitor Smad 6. Mostinterestingly, we found that Smad 6 not only is an inhibitor ofthe Smad pathway, but also elicits inhibitory effects on thephosphorylation of ERK 1/2 (Fig. 4A). It is noteworthy thatthe BMP-4 induced phosphorylation of the direct upstreammediator MEK is not affected by Smad 6 overexpressionindicating that Smad 6 interacts directly with ERK 1/2. Toconfirm that inhibition of ERK 1/2 by Smad 6 affectsendothelial cell sprouting in the same manner as inhibitionwith MEK inhibiting peptides we performed the HUVECspheroid sprouting assay using cells that overexpress Smad

6. As expected, in these experiments BMP-4 induced sprout-ing was effectively inhibited by Smad 6, indicating thatSmad 6 when overexpressed connects the Smad and ERKsignaling cascades (Fig. 3). To test whether Smad 6 may alsoserve as an endogenous regulator of Smad and ERKcrosstalk we examined levels of Smad 6 transcription inHUVECs and found that Smad 6 RNA is indeed highlyinducible upon stimulation with BMP-4 (SupplementalFig. 1). Interestingly, knock down of Smad 6 results inincreased spontaneous sprouting indicating that Smad 6serves as a negative regulator of endothelial cell sprouting(Fig. 4E). To differentiate between the Smad and the ERKinhibiting effect of Smad 6 we then decided to block Smadsignaling by loss of function of the ubiquitous Co-SmadSmad 4. Smad 4 has been shown earlier to be necessary foreffective Smad signaling [4]. Transfection of HUVECs withsiRNA specific for Smad 4 was highly effective in blockingSmad 4 transcription. In our experiments loss of function ofSmad 4 in HUVECs used for the sprouting assay does notaffect BMP-4 induced sprout length. Taken together this setof experiments indicates that ERK but not Smad signals arenecessary for endothelial cell sprouting and that Smad 6 is acentral regulator of Smad and ERK pathway crosstalk.

Recent data indicate that BMP signaling undergoesmodulation by crosstalk with other pathways [19]. However,the effect of Smad and ERK interactions has been the subjectof many controversies. In human mesangial cells a certainlevel of ERK activity is required for maximal induction ofSmad activity by TGF-β [7], whereas other data suggest aninhibitory effect of ERK on the Smad pathway [14,15]. In alandmark paper, Imamura et al. have demonstrated inhibitoryeffects of Smad 6 in TGF-β signaling, based on itsinterference with the phosphorylation dependent heterodi-merization of Smad 2 and Smad 4 [9]. Apart from the role ofSmad 6 in Smad–Smad complex formation, Smad 6 interactswith type I receptors at high concentrations, forming stablecomplexes, and thus inhibits signaling by various TGF-βsuperfamily members non-selectively [6]. In our experi-ments, even high levels of Smad 6 do not interfere withSmad 1/5 phosphorylation upon stimulation with BMP-4,making it unlikely that Smad 6 acts at the receptor level inthe experiments presented here (Fig. 4A). Instead, we showfor the first time that Smad 6 acts as an inhibitor of ERK 1/2phosphorylation in HUVECs. Thus, Smad 6 not only elicitsan inhibitory effect in the Smad cascade, but also blocksERK activity as part of a signaling pathway interaction. Thecentral role of Smad 6 in linking these pathways is under-lined by its upregulation upon pathway stimulation. Interest-ingly, this observation has also been confirmed in other cellsystems [26]. Smad 6 is inducible by BMP-4 in VSMC in avery similar fashion as described in our manuscript inendothelial cells. Furthermore, BMP-4 activates the Smad 1,p38 and ERK 1/2 pathways in pulmonary arterial smoothmuscle cells (PASMCs). However, while Smad signals areanti-proliferative in PASMCs, p38 and ERK result in cellproliferation and have anti-apoptotic activity in the same

397Q. Zhou et al. / Cardiovascular Research 76 (2007) 390–399

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from

cells, indicating that BMP-4 exerts complex effects in dif-ferent cell systems.

Having shown that Smad signaling interferes with the ERKpathway, we also asked if the MAPK cascade may converselyinteract with Smad signaling. Indeed MEK1 induces not onlyphosphorylation of ERK but also phosphorylation of Smad 1/5, which is the downstream effector of BMP signals (Fig. 5).Taking into consideration that Smad 1/5 was not suppressed inHUVECs pre-incubated with the MEK inhibitors our findingssuggest that MEK is sufficient, but not necessary, forphosphorylation of Smad 1/5. This result is consistent withfindings in osteoblastic cells, where collagen signals aretransduced via activation of the ERK pathway, which in turnenhances the transcriptional activity of Smad 1 in response toBMP [6,22].

Taken together, bidirectional mechanisms of crosstalkbetween the Smad and MAPK pathways exist in HUVECs(Fig. 6). Our observations suggest that the BMP-4 inducedactivation of ERK 1/2 occurs in conjunction with phosphor-ylation of Smad 1/5 and supports the hypothesis that ERK 1/2is part of a positive autoregulatory feedback loop in HUVECsto augment BMP-4 induced effects particularly necessary forendothelial cell sprouting. We found that not only Smad butalso ERK 1/2 activation is necessary to mediate full activityof BMP-4 signaling. This notion is supported by publisheddata indicating that ERK activity enhances TGF-β dependent

responses in human mesangial cells [7,20,21]. Thus,additional phosphorylation of Smads by MEK/ERK signalsis necessary for full activity of Smad signaling. In contrast, tocontrol overshooting Smad signaling, inhibitory Smads suchas Smad 6 are activated in parallel and interfere with Smadcomplex formation and signal transduction. In our case, weshow an inhibitory effect of Smad 6 on ERK1/2 phosphoryla-tion. This antagonistic effect of Smad 6 on ERK signalscontributes to a negative control feedback loop by inhibitingERK 1/2's activating role on Smad signals and thus helps toprevent excessive Smad signaling. We conclude that thebalance between the inputs of Smad 6 and MEK/ERK 1/2determine the level of transcriptional activity in the nucleusand therefore fine-tune BMP signaling in HUVECs. Smad 6and MEK/ERK are parts of a central regulatory system inBMP-4 mediated signaling in HUVECs. MEK/ERK acts asan enhancer of BMP signals and is necessary for cellsprouting. Both, BMP signals and ERK 1/2 are controlled bySmad 6, an endogenous inhibitor that is upregulated uponBMP stimulation.

Acknowledgements

We thank Bianca Engert and Fabian Klumpp for helpfulsupport.We thankH. Augustin, Germany, T. Imamura, Japan,K. Takimoto, USA and M.-J. Goumans, Amsterdam, The

Fig. 6. Schematic diagram of signaling crosstalk between ERK and Smad pathways in HUVEC. BMP-4 activates Smad 1/5 and the MEK–ERK pathway. MEKactivation results in additional phosphorylation of Smad 1/5, which is necessary to transfer the BMP-4 effect on capillary sprouting (positive feedback loop=P).To control overshooting BMP signaling we show that Smad 6 blocks BMP-4 induced ERK phosphorylation (negative feedback loop); —!=blocking effect;!=activating effect.

398 Q. Zhou et al. / Cardiovascular Research 76 (2007) 390–399

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from

Netherlands for generously sharing valuable reagents. Workin M.M's laboratory is supported by the Deutsche For-schungsgemeinschaft SFB TR 23 (A1) and by intramuralfunds of the University of Freiburg. C.P. is supported bygrants from the National Institutes of Health (HL61656,HL072347, AG024282, and HL65619) and is an EstablishedInvestigator of the American Heart Association and aBurroughs Wellcome Fund Clinical Scientist in TranslationalResearch.

All authors report no conflicts of interest.

Appendix A. Supplementary data

Supplementary data associatedwith this article can be found,in the online version, at doi:10.1016/j.cardiores.2007.08.003.

References

[1] Boulton TG, Nye SH, Robbins DJ, Ip NY, Radziejewska E,Morgenbesser SD, et al. ERKs: a family of protein-serine/threoninekinases that are activated and tyrosine phosphorylated in response toinsulin and NGF. Cell 1991;65:663–75.

[2] Chang L, Karin M. Mammalian MAP kinase signalling cascades.Nature 2001;410:37–40.

[3] Chen X, Weisberg E, Fridmacher V, Watanabe M, Naco G, WhitmanM. Smad4 and FAST-1 in the assembly of activin-responsive factor.Nature 1997;389:85–9.

[4] Derynck R, Zhang YE. Smad-dependent and Smad-independentpathways in TGF-beta family signalling. Nature 2003;425:577–84.

[5] Frank DB, Abtahi A, Yamaguchi DJ, Manning S, Shyr Y, Pozzi A, et al.Bone morphogenetic protein 4 promotes pulmonary vascular remodel-ing in hypoxic pulmonary hypertension. Circ Res 2005;97:496–504.

[6] Hata A, Lagna G, Massague J, Hemmati-Brivanlou A. Smad6 inhibitsBMP/Smad1 signaling by specifically competing with the Smad4tumor suppressor. Genes Dev 1998;12:186–97.

[7] Hayashida T, Decaestecker M, Schnaper HW. Cross-talk between ERKMAPkinase and Smad signaling pathways enhances TGF-beta-dependentresponses in human mesangial cells. FASEB J 2003;17:1576–8.

[8] Heldin CH, Miyazono K, ten Dijke P. TGF-beta signalling from cellmembrane to nucleus through SMAD proteins. Nature 1997;390:465–71.

[9] Imamura T, Takase M, Nishihara A, Oeda E, Hanai J, Kawabata M, et al.Smad6 inhibits signalling by the TGF-beta superfamily. Nature1997;389:622–6.

[10] Itoh FAH, Sugamura K, Heldin CH, ten Dijke P, Itoh S. Promotingbone morphogenetic protein signaling through negative regulation ofinhibitory Smads. EMBO J 2001;20:4132–42.

[11] Javelaud D, Mauviel A. Crosstalk mechanisms between the mitogen-activated protein kinase pathways and Smad signaling downstream ofTGF-beta: implications for carcinogenesis. Oncogene 2005;24:5742–50.

[12] Jia YTK. Mitogen-activated protein kinases control cardiac KChIP2gene expression. Circ Res 2006;98:386–93.

[13] Korff T, Augustin HG. Tensional forces in fibrillar extracellular matricescontrol directional capillary sprouting. J Cell Sci 1999;112(Pt 19):3249–58.

[14] KretzschmarMDJ, Timokhina I,Massague J. Amechanismof repressionof TGFbeta/ Smad signaling by oncogenic Ras. Genes Dev 1999;13:804–16.

[15] Kretzschmar M, Doody J, Massague J. Opposing BMP and EGFsignalling pathways converge on the TGF-beta family mediator Smad1.Nature 1997;389:618–22.

[16] Morrell NW, Yang X, Upton PD, Jourdan KB, Morgan N, Sheares KK,et al. Altered growth responses of pulmonary artery smooth musclecells from patients with primary pulmonary hypertension to transform-ing growth factor-beta(1) and bone morphogenetic proteins. Circula-tion 2001;104:790–5.

[17] Moser M, Patterson C. Bone morphogenetic proteins and vasculardifferentiation: BMPing up vasculogenesis. Thromb Haemost 2005;94:713–8.

[18] Nakaoka T, Gonda K, Ogita T, Otawara-Hamamoto Y, Okabe F, KiraY, et al. Inhibition of rat vascular smooth muscle proliferation in vitroand in vivo by bone morphogenetic protein-2. J Clin Invest 1997;100:2824–32.

[19] Nohe AKE, Knauss P, Petersen NO. Signal transduction of bonemorphogenetic protein receptors. Cell Signal 2004;16:291–9.

[20] Park S. Fine tuning and cross-talking of TGF-beta signal by inhibitorySmads. J Biochem Mol Biol 2005;38:9–16.

[21] Shi Y, Massague J. Mechanisms of TGF-beta signaling from cellmembrane to the nucleus. Cell 2003;113:685–700.

[22] Tamura Y, Takeuchi Y, Suzawa M, Fukumoto S, Kato M, Miyazono K,et al. Focal adhesion kinase activity is required for bone morphogeneticprotein — Smad1 signaling and osteoblastic differentiation in murineMC3T3-E1 cells. J Bone Miner Res 2001;16:1772–9.

[23] Urist MR. Bone: formation by autoinduction. 1965. Clin Orthop RelatRes 2002:4–10.

[24] Valdimarsdottir GGM, Rosendahl A, Brugman M, Itoh S, Lebrin F,Sideras P, ten Dijke P. Stimulation of Id1 expression by bonemorphogenetic protein is suffcient and necessary for bonemorphogeneticprotein-induced activation of endothelial cells. Circulation 2002;106:2263–70.

[25] Yang B, Cao DJ, Sainz I, Colman RW, Guo YL. Different roles of ERKand p38 MAP kinases during tube formation from endothelial cellscultured in 3-dimensional collagen matrices. J Cell Physiol 2004;200:360–9.

[26] Yang X, Long L, Southwood M, Rudarakanchana N, Upton P, JefferyT, et al. Dysfunctional Smad signaling contributes to abnormal smoothmuscle cell proliferation in familial pulmonary arterial hypertension.Circ Res 2005;27:1053–63.

[27] Zawel L, Dai JL, Buckhaults P, Zhou S, Kinzler KW,Vogelstein B, et al.Human Smad3 and Smad4 are sequence-specific transcriptionactivators. Mol Cell 1998;1:611–7.

399Q. Zhou et al. / Cardiovascular Research 76 (2007) 390–399

at Zentralbibliothek on January 16, 2013http://cardiovascres.oxfordjournals.org/

Dow

nloaded from