AMP-activated Protein Kinase Impairs EndothelialActin Cytoskeleton Assembly by PhosphorylatingVasodilator-stimulated Phosphoprotein*□S

Received for publication, September 14, 2006, and in revised form, November 1, 2006 Published, JBC Papers in Press, November 2, 2006, DOI 10.1074/jbc.M608866200

Constanze Blume‡, Peter M. Benz‡, Ulrich Walter‡, Joohun Ha§, Bruce E. Kemp¶1, and Thomas Renne‡2

From the ‡Institute for Clinical Biochemistry and Pathobiochemistry, Julius-Maximilians-University Wurzburg, Josef-SchneiderStrasse 2, D-97080 Wurzburg, Germany, §Department of Biochemistry and Molecular Biology, Kyung University College ofMedicine, 1 Hoegi-dong, Tongdaemun-gu, Seoul, Korea 130-701, and ¶St. Vincent’s Institute of Medical Research, Fitzroy,Victoria 3065 and Commonwealth Scientific and Industrial Research Organization Molecular and Health Technologies,Parkville, Victoria 3052, Australia

Vasodilator-stimulated phosphoprotein (VASP) is an actinregulatory protein that links signaling pathways to remodelingof the cytoskeleton. VASP functions are modulated by proteinkinases, which phosphorylate the sites Ser-157, Ser-239, andThr-278. The kinase responsible for Thr-278 phosphorylation,biological functions of the phosphorylation, and associationwith disease states have remained enigmatic. Using VASP phos-phorylation status-specific antibodies, we identified AMP-acti-vated protein kinase (AMPK), a serine-threonine kinase andfundamental sensor of energy homeostasis, in a screen forkinases that phosphorylate the Thr-278 site of VASP in endo-thelial cells. Pharmacological AMPK inhibitors and activatorsand AMPKmutants revealed that the kinase specifically targetsresidue Thr-278 but not Ser-157 or Ser-239. Quantitative fluo-rescence-activated cell sorter analysis and serum response fac-tor transcriptional reporter assays, which quantify the cellularF-/G-actin equilibrium, indicated that AMPK-mediated VASPphosphorylation impaired actin stress fiber formation andaltered cell morphology. In the Zucker Diabetic Fatty (ZDF) ratmodel for type II diabetes, AMPK activity and Thr-278 phos-phorylation were substantially reduced in arterial vessel walls.These findings suggest thatVASP is a newAMPKsubstrate, thatVASP Thr-278 phosphorylation translates metabolic signalsinto actin cytoskeleton rearrangements, and that this signalingsystem becomes down-regulated in diabetic vessels.

The regulation of actin dynamics is important for cytoskel-etal remodeling, cell morphology, and associated events such ascellmigration, adhesion, andmotility. In this complex scenario,Enabled/vasodilator-stimulated phosphoprotein (Ena/VASP)3

proteins coordinate different modes of actin organizationincluding actin filament assembly, cross-linking, and bundling.In mammals the Ena/VASP protein family comprises theVASP, Mena (the homolog of Drosophila Ena), and the Ena-VASP-like protein (Evl). These proteins participate in actin fil-ament networks at sites of cell-cell and cell-matrix interactionsandwithin lamellipodia and filopodia protrusions.Allmembersof the Ena/VASP family share a tripartite domain organizationconsisting of an N-terminal Ena/VASP homology (EVH)domain 1 (in VASP, residues 1–115), a central proline-richregion (PRR, 115–225), and a C-terminal EVH2 domain (225–380) (1, 2). VASP directly binds to globular actin (G-actin) andfilamentous actin (F-actin) via the EVH2 domain (2, 3). Resi-dues in the C terminus of the EVH2 domain mediate tetramer-ization of VASP and hetero-oligomerization with other Ena/VASP family members. These interactions are important forfilament elongation and branching and, consequently, for cellmigration and cell-cell or cell-matrix interactions (3, 4). Ena/VASP proteins promote actin polymerization and assembly,and several models of VASP-mediated F-actin control havebeen proposed, including the regulation of nucleation, bun-dling, branching, and capping (5, 6).VASP effects on actin turnover are regulated by phosphoryl-

ation, and in vitro studies have identified three serine/threo-nine phosphorylation sites (Ser-157, Ser-239, and Thr-278) inthe protein (7). Both in vitro phosphorylated VASP and VASPmutants that mimic defined phosphorylation states suggestthat phospho-VASP interferes with actin filament assembly,turnover, and branching. Mechanistically, VASP phosphoryla-tion seems to modulate F-actin binding, to reduce actin-poly-merization-promoting activity (8), and to interfere with anti-capping activity (6). Originally, VASP phosphorylation wasdescribed in platelets, where both cAMP- and cGMP-depend-ent protein kinases (PKA and PKG, respectively) were shown tophosphorylate residues Ser-157 and Ser-239 (9). Subsequently,the importance of phospho-Ser-157 (pSer-157) and phospho-

* This work was supported in parts by grants of the Deutsche Forschungsge-meinschaft Grants SFB 355 and SFB 688 (to U. W. and T. R.). The costs ofpublication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked “advertisement” inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

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

1 An Australian Research Council Federation Fellow.2 To whom correspondence should be addressed. Tel.: 49-931-201-36116;

Fax: 49-931-201-45137; E-mail: thomas@renne.net.3 The abbreviations used are: Ena/VASP, Enabled/vasodilator-stimulated

phosphoprotein; EVH, Ena/VASP homology; PK, protein kinase; ZDF,Zucker Diabetic Fatty; HUVEC, human umbilical vein endothelial cells;

FACS, fluorescence-activated cell sorter; TRITC, tetramethylrhodamine iso-thiocyanate; ACC, acetyl-CoA carboxylase; APMP, AMP-activated proteinkinase; G-actin, globular actin; F-actin, filamentous actin; SRE, serumresponse element; SRF, serum response factor; VASP, vasodilator-stimu-lated phosphoprotein; 8-Br, 8-bromo-; CA, constitutively active; DN, dom-inant negative; wt, wild type.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 282, NO. 7, pp. 4601–4612, February 16, 2007© 2007 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4601

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Ser-239 (pSer-239) in cyclic-nucleotide dependent-kinase sig-naling cascades was established in other cardiovascular cellsincluding aortic smooth muscle cells (10) and endothelial cells(11), in vessels of animalmodels, and in humans (for review, seeRef. 12).To quantify the magnitude of phosphorylation at these sites,

phosphorylation-specific antibodies anti-pSer-157 (5C6) andanti-pSer-239 (16C2) were developed (11). Analyses based onthese antibodies revealed that reduced Ser-157/Ser-239 phos-phorylation was associated with defective nitric oxide (NO)/cGMP signaling and endothelial dysfunction (13). AlthoughPKA and PKGmay phosphorylate purified VASP at Thr-278 toa small extent in vitro (7), a role for cyclic nucleotide-dependentprotein kinase phosphorylation of Thr-278 in vivo has neverbeen convincingly demonstrated. Indeed, neither the kinaseresponsible for Thr-278 phosphorylation nor the functionalconsequences of Thr-278 phosphorylation have been resolved.The AMP-activated protein kinase (AMPK) is a major regu-

lator ofmetabolism not only at the cellular but also at the wholeorganism level (14, 15). The evolutionarily conserved AMPK isa ubiquitously expressed heterotrimeric protein composed of acatalytic� subunit harboring a serine/threonine kinase domainand two regulatory � and � subunits (14, 16). AMP, generatedfrom metabolic stress-induced ATP depletion, activates theAMPK allosterically. Two signaling pathways involving the Ca2�/calmodulin-dependentprotein kinase kinase and the tumor sup-pressor LKB1 (also referred to as STK11) activate AMPK byphosphorylating its activation loop at Thr-172 (17, 18). Onceactivated, AMPK phosphorylates and inactivates several keyenzymes in energy-consuming biosynthetic pathways, therebyconserving ATP pools and preserving cellular energy homeo-stasis. Substrates of AMPK include acetyl-CoA carboxylase(ACC), 3-hydroxy-3-methylglutaryl-CoA reductase, and glyco-gen synthase, enzymes that control the synthesis of fatty acids,cholesterol, and glycogen, respectively (for review, see Refs.19–21). Disturbance of AMPK activity has been linked to thepathogenesis of type II diabetes mellitus and obesity (22, 23),and modulation of AMPK activity may offer novel therapeuticoptions for treatment of these diseases (24).We identifiedAMPK in a screen for kinases that phosphorylate

VASP residue Thr-278 in endothelial cells. Studies using pharma-cological agents, in vitro phosphorylations, and constitutivelyactive and dominant negative AMPK mutants confirmed thatVASP is a novel AMPK substrate. AMPK specifically targets theVASP Thr-278 site. Functionally, AMPK-mediated VASP Thr-278 phosphorylation reduced F-actin assembly, resulting indefective stress fiber formation and altered cell morphology. Inarterial vessel walls from Zucker Diabetic Fatty (ZDF) rats,which serve as a model for type II diabetes mellitus, AMPKactivity and VASP Thr-278 phosphorylation were severelyreduced. Taken together, our data show that metabolic sig-nals via AMPK-mediated VASP Thr-278 phosphorylationmay inhibit actin polymerization. In this way our studyestablishes a novel signaling pathway that directly linkscytoskeleton dynamics to energy metabolism that is impairedin diabetic vessels.

EXPERIMENTAL PROCEDURES

Animals—4–7-month-old male ZDF (fa/fa) rats (615.0 �24.9 g) with overt diabetes and their non-diabetic lean controls(�/�, 392.4 � 35.2 g) were from HarlanWinkelmann. Animalstudies were approved by the Regierung of Unterfranken.Cell Culture—Human umbilical vein endothelial cells

(HUVEC, Cambrex Bioproducts) were cultivated as described(25). ECV304, HEK293, and HeLa cells were grown in Dulbec-co’s modified Eagle’s medium containing 4.5 g/liter glucosesupplemented with 10% fetal bovine serum (26).Generation andCharacterization of Anti-pThr-278Antibody—

To generate antibodies that specifically recognize phospho-Thr-278 (pThr-278) VASP, rabbits were immunized with thefollowing phosphopeptide coupled to keyhole limpet hemocy-anin: 273CRRKApTQVGE282 (pT is phosphothreonine). Theantibody was immuno-selected using columns with immobi-lized phosphopeptide followed by a column with unphospho-rylated VASP peptide. Antibodies against the correspondingnon-phosphorylated VASP-peptide (anti-Thr-278) wereimmunopurified from the latter column and applied to the col-umn with immobilized phosphopeptide to remove pThr-278-cross-reacting antibodies. To test the specificity of anti-pThr-278 antibodies for 273CRRKApTQVGE282, titer plates(Maxisorb, Nunc) were coated by overnight incubation with100 �g/ml each of the phospho- and the unphosphorylated-VASP peptide. Anti-pThr-278 antibodies were applied in aserial 1:2 dilutions (starting concentration 2 �g/ml; 13 nM) andspecific bound antibody was detected with a secondary perox-idase-conjugated antibody to rabbit immunoglobulin (diluted1:4000) and a diammonium 2,29-azino-bis(3-ethyl-2,3-dihy-drobenzthiazoline-6-sulfonate) substrate solution for 30 min.The change of absorbance was read at 405 nm. All incubationswere done at 37 °C except for the coating step, which was doneat 4 °C.Cloning of VASP Mutants—VASP mutants were generated

by site-directed mutagenesis using the QuikChange Multi kit(Stratagene). Human VASP cDNA in a pcDNA3 vector(Invitrogen) and the following primers were used to exchangeSer-157, Ser-239, and Thr-278 with alanine or glutamic acid,respectively: 5�-CG GAG CAC ATA GAG CGC CGG GTCGCC AAT GCA GGA GGC CCA CC-3� (S157A); 5�-GGAGCC AAA CTC AGG AAA GTC GCC AAG CAG GAG GAGGCC TCAGGG-3� (S239A); 5�-G CTGGCC CGGAGA AGGAAAGCCGCGCAAGTTGGGGAGAAAACC-3� (T278A)and 5�-G CTG GCC CGG AGA AGG AAA GCC GAG CAAGTT GGG GAG AAA ACC-3� (T278E). Based on the aminoacids present at positions 157, 239, and 278, respectively, VASPmutants were designated AAA, AAE, and AAT. All constructswere confirmed by DNA sequence analyses.Screen for Kinases That Phosphorylate Thr-278 in Cells—To

screen for kinases that may phosphorylate the VASP Thr-278residue, confluent HUVEC, ECV304, or HEK293 cells, tran-siently transfected with cDNAs coding for human VASP in apcDNA3 vector, were treated with protein kinase activators orinhibitors. After incubation, the relative amount of pThr-278 ininhibitor- or activator-treated cells was quantified by Westernblot analysis with anti-pThr-278 antibody. PKA inhibitors used

AMPK Phosphorylates VASP

4602 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 7 • FEBRUARY 16, 2007

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

were H89 (0.5–5 �M for 10–60 min, Sigma) and Rp-8Br-cAMPS (50–500 �M for 10–60 min, BioLog), and PKA activa-tors used were Forskolin (1–10 �M for 0.5–60 min, Sigma) and8-Br-cAMP (50–200 �M for 0.5–60 min, BioLog). PKG effectson Thr-278 phosphorylation were analyzed with the inhibitorRp-8-Br-cGMPS (50–500�M for 10–60min, BioLog) andwiththe activators sodium nitroprusside (0.5–20 �M for 1–60 min,Sigma) and 8-Br-cGMP (50–200 �M, 0.5–60 min, BioLog).PKC was analyzed using the inhibitors Bis I (0.5–50 �M for0.5–8 h, Calbiochem) and Rottlerin (1–10 �M for 0.5–8 h,Sigma) and the activator phorbol 12-myristate 13-acetate(0.1–10�M for 2–30min, Sigma). To inhibit Ca2�/calmodulin-dependent protein kinase, we utilized KN93 (0.5–5 �M for0.5–8 h, Sigma), Ca2�/calmodulin-dependent protein kinase IIpeptide (0.5–5 �M for 0.5–8 h, Calbiochem), or Calcimycin(1–10�M for 10–30min, Sigma) for stimulation. PKB (Akt)wasinvestigated by Wortmannin (0.1–10 �M for 0.5–8 h, Calbio-chem) and fetal calf serum (0.5–20% for 0.5–3 h, Invitrogen).AMPK-driven Thr-278 was assessed using the AMPK activa-tors Metformin (1,1-dimethylbiguanide, 0.5–8 mM for 10–60min, Sigma), Phenformin (phenethylbiguanide, 0.5–8 mM for10–60 min, Sigma), and AICAR (5-aminoimidazole-4-carbox-amide-1�-riboside, 0.5–2.5 mM for 10–30 min, Calbiochem)and AMPK inhibitors compound C (2–20 �M for 0.5–8 h, Cal-biochem) and Indirubine (indirubine-3-oxime, 0.5–20 �M for0.5–8 h, Tocris).Transfection Experiments—ECV304 andHeLa cells were cul-

tivated in 6-well plates to 70% confluence and transiently trans-fected with 1.0 �g of cDNA coding for an Myc-tagged consti-tutively active AMPK �1 subunit mutant (CA�), a dominantnegative AMPK �1 form (DN�), or the wild-type �1 subunit(wt�) (29) using GenePorter 2 with Booster Reagent 2 (Peqlab).After 24 h of expression washed cells were lysed, and pThr-278levels were determined by Western blot analysis with anti-pThr-278 antibodies. Transfection efficiency was controlled bytransfecting with a pcDNA3GFP vector, which encodes greenfluorescent protein. Immunofluorescencemicroscopy revealedthat �35–40% HeLa and 20–25% ECV304 cells stained posi-tive for GFP after 24 h of expression.Analysis of VASP Phosphorylation—Cells or tissue were

immediately lysed in sample buffer containing 50 mM Tris, pH6.8, 2% SDS, 0.1% bromphenol blue, and 10% glycerol. Lysateswere separated by 8% SDS-PAGE under reducing conditionsand electro-transferred to a nitrocellulose membrane(Schleicher& Schull). Proteins on blots were probed using anti-bodies against AMPK�, phospho-AMPK�, phospho-ACC(Cell Signaling Technology), glyceraldehyde-3-phosphatedehydrogenase (Chemicon), Myc (Santa Cruz Biotechnology),VASP (M4, (9)), pSer-157-VASP (5C6, (11)), pSer-239-VASP(16C2, (27)), and pThr-278-VASP as well as horseradish peroxi-dase-conjugated secondary antibodies. Detection was performedusing a chemiluminescence technique (ECLPlus, AmershamBio-sciences.). The intensity of the individual bands on the x-ray films(XBA, Fotochemische Werke Berlin) was quantified by densito-metric scans (Scan Pack 3.0, Biometra).Immunochemistry—Formaline fixed and permeabilized

HeLa cells in 2-well Chamberslides (Nunc), and aortic rat tis-sues cryosections were incubated overnight with primary anti-

bodies against VASP (polyclonal M4 and monoclonal IE273(28)) and phospho-VASP (see above) followed by secondaryanti-mouse and anti-rabbit antibodies conjugated to Alexa-fluor 488 or 594 (Molecular Probes) for 1 h. All incubationswere performed in 5% goat serum, phosphate-buffered saline atroom temperature. Stained sections were investigated using aNikon Eclipse E600 microscope equipped with a C1 confocalscanning head and a 100� oil immersion objective. Imageswere acquired using the EZ-C1 2.10 software from Nikon.Quantification of Cellular F-actin Content by FACS—HeLa

cells were co-transfected (GenePorter 2 and Booster Reagent 2,Peqlab) with pcDNA3 vectors coding for VASPmutants (AAA,AAE, AAT; 2.5 �g each) or Myc-tagged AMPK �1 subunitscorresponding to wild-type (wt�), a constitutively activemutant (CA�), or a dominant negative (DN�) mutant (5 �geach) (29). After 24 h of expression in Dulbecco’s modifiedEagle’s medium with all supplements, cells were trypsinized,fixed, permeabilized (see above), and co-stained for F-actinusing Alexa-fluor 647 phalloidin and for Myc-tagged AMPK�1subunit using anti-Myc antibody followed by fluorescein iso-thiocyanate-conjugated anti-mouse antibody. The mean cellu-lar F-actin content determined by phalloidin staining ofoverexpressing cells was quantified using FACScan (BDBiosciences) as described (3).Reporter Gene Assays—We utilized a serum response factor

(SRF)-based transactivation assay to quantify changes in cellu-lar actin dynamics as described (30, 31). Briefly, HeLa cellsgrown on 6-well plates were transiently transfected usingGenePorter2 and Booster2. Each transfection mix includescDNAs coding for VASPmutants (0.1 �g), AMPKmutants (0.5�g), serum response element (SRE) reporter vector (0.5 �g,pSRE, Stratagene), Renilla luciferase control vector (0.25 �g,phRL-TK, Promega), and pcDNA3 vector without insert (0.55�g, Invitrogen). After 24 h cells were lysed, and luciferase activ-ities were measured in the supernatants using the Dual Lucif-erase Reporter Assay System (Promega) with a luminometer(Lumat, Berthold). Firefly luciferase activity was normalized toRenilla luciferase activity and expressed relative toMOCK- (setto 0%) and VASP AAA-transfected cells (set to 100%).In Vitro VASP Phosphorylation—AMPK phosphorylation

assays were performed at 30 °C in 40 �l reaction mixtures con-taining 80 ng of purified recombinant human VASP and 100milliunits of purified AMPK (Upstate Biotechnology Inc.) inreaction buffer (5 mM Hepes, pH 7.5, 0.1 mM dithiothreitol,0.25%Nonidet P-40, 7.5mMMgCl2, 50�MATP, 300�MAMP).Reactions were stopped at different time points by the additionof SDS sample buffer and heating at 90 °C for 10 min.Statistics—Results were expressed as themeans� S.D. Com-

parisons between two groups were analyzed using Student’s ttests. Significant difference was defined as p values �0.05.

RESULTS

AMP-activated Protein Kinase Phosphorylates VASP at Posi-tion Thr-278—To analyze VASP Thr-278 phosphorylation andto identify the kinase that phosphorylates residue Thr-278, wegenerated a phosphorylation status-specific anti-peptide anti-body (anti-pThr-278) using a phosphopeptide spanning theThr-278 phosphorylation site (273CRRKApTQVGE282). Immu-

AMPK Phosphorylates VASP

FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4603

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

nopurified antibody specifically bound to the phosphopeptidebut not to the unphosphorylated control (Fig. 1A). Further-more, anti-pThr-278 did not cross-react with recombinantunphosphorylated VASP inWestern blotting (see Fig. 3, upperlane, t � 0 min). To confirm that anti-pThr-278 detects thephosphorylation status of cellular VASP at the third site, weoverexpressed VASP and protein phosphatase 2B, which hasbeen shown to dephosphorylate pThr-278-VASP (28). Resultsfrom Western blots incubated with anti-pThr-278 indicatedthat protein phosphatase 2B dose-dependently decreasedpThr-278 inHEK293 cells (Fig. 1B). Together, the data confirmthe specificity of anti-pThr-278 antibody for pThr-278 in vivo.Using anti-pThr-278, we systematically screened for serine-

threonine kinases that may phosphorylate VASP in intact cells.First, we selected kinases whose consensus sequences matchedthe Thr-278 flanking residues. Subsequently, endothelial celllines (ECV304 and HUVEC), which express a high amount ofendogenous VASP (11), as well as HEK293 cells, which tran-

siently overexpress VASP, were in-cubated with specific activators orinhibitors of the selected kinasesand analyzed for pThr-278 and totalVASP by Western blotting (de-scribed under “Experimental Proce-dures” in detail). Pretreatment withthe PKA activator Forskolin clearlyaugmented pSer-157 in ECV304cells as indicated by the phospho-specific antibody (anti-pSer-157),Fig. 1C, lower panel). Phosphoryla-tion of Ser-157 of VASP led to ashift in apparent molecular mass inSDS-PAGE from 46 to 50 kDa, andanti-VASP antibodies detected thecharacteristic doublet of Ser-157-non-phosphorylated and -phospho-rylated protein after stimulation(anti-VASP, Fig. 1C, middle panel,the upper band was detected byanti-pSer-157). Notably, PKA acti-vation did not increase pThr-278over basal levels. SummarizedpThr-278 signal intensities of thedoublet did not significantly differfrom the pThr-278 level of VASPfrom unstimulated cells (100 versus97 � 8%, Fig. 1C, upper panel andcolumn bar). Furthermore, relativesignal intensities of the 46- and 50-kDa band from anti-pThr-278 andanti-VASP blots were not obviouslydifferent (Fig. 1C, compare upperand middle panels), suggesting thatphosphorylation atThr-278 occurredindependent of the Ser-157 phospho-rylation status. Similarly, PKG activa-tors such as 8-Br-cGMP failed tochange Thr-278 phosphorylation in

cells (100 versus 105 � 9%, Fig. 1C, right, upper panel andcolumn bar). Under these conditions the activator clearlyincreased phosphorylation of the PKG-preferred VASPphosphorylation site, Ser-239 (Fig. 1C, lower panel, anti-pSer-239). Note, in contrast to phosphorylation at the firstsite, phosphorylation at the second site does not induce ashift in molecular weight (middle panel, anti-VASP). Fur-thermore, pThr-278 was independent of PKC, PKB, orCa2�/Calmodulin-dependent protein kinase activity (notshown). Next, we wondered whether the AMPK may phos-phorylate VASP at Thr-278. AMPK activity inhibitorsIndirubine and compound C, concentration (Fig. 2A)- andtime-dependent (Fig. 2B), significantly decreased pThr-278 byup to 45% relative to basal pThr-278 levels in untreatedHUVEC. Consistently, AMPK activating agents such as Met-formin, Phenformin, or 5-aminoimidazole-4-carboxamide-1�-riboside increased pThr-278 in a concentration (Fig. 2C)- andtime-dependent (Fig. 2D) manner up to maximally 56, 72, and

FIGURE 1. Analysis of Thr-278 phosphorylation status with phospho-specific antibodies. A, to test thephospho-specificity of anti-pThr-278, microtiter plates were coated with 100 �g/ml each of a phosphopeptidespanning the Thr-278 phosphorylation site (Œ) or the corresponding unphosphorylated peptide (f) followedby incubation with anti-pThr-278 antibodies in serial 1:2 dilutions starting at 2 �g/ml. Specifically boundantibody was detected with a horseradish peroxidase-conjugated secondary antibody followed by the chro-mogenic substrate diammonium 2,29-azino-bis(3-ethyl-2,3-dihydrobenzthiazoline-6-sulfonate), and absorb-ance was measured at 405 nm. The figure is representative for a series of five independent experiments. B, thespecificity of anti-pThr-278 antibodies for pThr-278-VASP was analyzed using protein phosphatase 2B (PP2B)that dephosphorylates pThr-278. HEK293 cells were co-transfected with 1.0 �g of human VASP in pcDNA3vector and 0.25 or 1 �g of cDNA coding for protein phosphatase 2B or vector without insert (MOCK). After 24 hcells were lysed and analyzed for pThr-278 (anti-pThr-278 antibody) and total VASP (anti-VASP antibody, M4)by Western blot. C, ECV304 cells were treated with PKA or PKG activators Forskolin (5 �M, 10 min) or 8-Br-cGMP(200 �M, 15 min), respectively. Phosphorylations at the sites Thr-278, Ser-157, and Ser-239 were analyzed andcompared with untreated cells by Western blotting using phospho-specific antibodies. The intensity of pThr-278 band was quantified from Western blots by densitometric scans; pThr-278 levels in untreated cells were setto 100%. The columns give means � S.D. (n � 4; not significant (n.s.), treated versus untreated).

AMPK Phosphorylates VASP

4604 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 7 • FEBRUARY 16, 2007

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

73%, respectively, over basal levels. All tested AMPK inhibitorsand activators similarly affected pThr-278 levels in HEK293and ECV304 cells (data not shown), suggesting that VASPmight be a novel AMPK substrate in vivo. Notably, in non-stimulated HUVEC about 40% of total VASP was phosphoryl-ated at Thr-278 as indicated by a comparison of signals frompThr-278 and anti-Thr-278 antibodies. Anti-Thr-278 antibodydetects the corresponding peptide, which was used for genera-tion of anti-pThr-278, in the non-phosphorylated form. Under

basal conditions the ration of Thr-278-phosphorylated to totalVASP was even higher in HEK293 and ECV304 (about 1:2). Toconfirm that the employed conditions of AMPK inhibitors andactivators efficiently targeted enzyme activity in the treatedcells, we analyzed phosphorylation of AMPK itself at residueThr-172, which correlates with AMPK activity (32). AMPKThr-172 phosphorylation paralleled pThr-278 levels (Fig. 2,A–D), supporting the hypothesis that VASP is a substrate ofactivated AMPK in vivo.

FIGURE 2. AMPK inhibitors and activators modulate Thr-278 phosphorylation in living cells. To test the AMPK activity on Thr-278 phosphorylation,confluent HUVEC were incubated with AMPK inhibitors or activators of varying concentrations (A and C) and incubation times (B and D). Washed cells werelysed and analyzed by Western blotting using antibodies against pThr-278, total VASP, and activated AMPK (anti-pAMPK). Changes in pThr-278 were quantifiedfrom Western blots and blotted relative to basal pThr-278 levels in untreated cells (0% value). The columns give the means � S.D. (n � 4; *, p � 0.05, treatedversus untreated). A, pThr-278 was dose-dependently reduced after 4 h of incubation of cells with the AMPK inhibitors Indirubine (5 and 20 �M) or CompoundC (2 and 20 �M). B, time course of AMPK activation and pThr-278 in cells incubated with Indirubine (10 �M) or Compound C (10 �M) C, pThr-278 and AMPKactivity were stimulated with Metformin (0.5 and 2 mM) or Phenformin (0.5, 2 mM) for 1 h or with 5-aminoimidazole-4-carboxamide-1�-riboside (AICAR; 0.5, 2.5mM) for 30 min. D, time dependence of pThr-278 and AMPK activation after stimulation with Metformin (2 mM), Phenformin (2 mM), or 5-aminoimidazole-4-carboxamide-1�-riboside (2.5 mM).

AMPK Phosphorylates VASP

FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4605

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

AMPK Specifically Targets the VASP Residue Thr-278 inVitro—To examine the specificity of AMPK for any of the threeVASP phosphorylation sites, recombinant purified VASP wasincubated with active AMPK (Fig. 3). At 0.5, 2, 8, and 30 min,aliquots from the reactionmixture were collected and analyzedfor total VASP (anti-VASP) and its phosphorylations at resi-dues Ser-157, Ser-239, and Thr-278 using the phospho-specific

anti-VASP antibodies 5C6 (detects pSer-157), 16C2 (detectspSer-239), and anti-pThr-278. Western blots revealed thatVASP Thr-278 phosphorylation occurred rapidly within 0.5min and increased for 30 min. Under these conditions, Com-pound C inhibited Thr-278 phosphorylation, confirming theselectivity of active AMPK-mediated VASP phosphorylation(not shown). In contrast, AMPKdid not phosphorylate residuesSer-157 or Ser-239 within the incubation time. Taken togetherthese data indicate that AMPK specifically utilized the Thr-278phosphorylation site in vitro.AMPK Activity Mutants Modulate VASP Thr-278 Phos-

phorylation—To confirm the effect of AMPK inhibitors andactivators on pThr-278 levels by an independent approach, weemployed AMPK mutants affecting enzyme activity. ECV304and HeLa cells were transfected with cDNAs coding for a con-stitutively active mutant of the AMPK �1 subunit (CA�), adominant negative variant (DN�), the wild-type �1 subunit(wt�), or empty vector (MOCK). CA� is a truncated form ofwild type �1 subunit (amino acids 1–312) containing a muta-tion that alters Thr-172 to an aspartic acid (T172D), whichmimics phosphorylation. Phosphorylation of Thr-172 of the�1subunit is essential for the enzyme activity. In the DN� AMPKform, aspartic acid 157 of the �1 subunit is substituted by analanine residue (29). After 24 h, lysed cells were analyzed forpThr-278 levels, total VASP, and expression of the constructs.Additionally, phosphorylation of the established AMPK sub-strate ACC (33) was analyzed as an internal measure of AMPK

signaling. Compared with MOCK-transfected cells (set to 100%), CA�increased pThr-278 levels inECV304 and HeLa cells to 123 and129%, respectively (Fig. 4, upperlanes, A and B). Consistently, inhi-bition of AMPK activity using DN�reduced pThr-278 down to 60 and88% in ECV304 or HeLa cells,respectively, whereas wt� did notsignificantly affect pThr-278 levels(104% in both cell lines; Fig. 4, col-umn bars in A and B). The relativedegrees of phosphorylation of Thr-278 by the various AMPK formswere similar to those observed forthe AMPK substrate ACC. Theseresults are in accordance with ourfindings using pharmacologicalAMPK inhibitors and activators(Fig. 2) and support the conclusionthat AMPK phosphorylates VASPat the Thr-278 site in vivo.AMPK-mediated VASP Thr-278

Phosphorylation Decreases CellularF-actin Content—VASP phospho-rylation modulates F-actin forma-tion in vitro (7, 8). However, in cellsthe precise consequences of differ-ential VASP phosphorylation forthe F-/G-actin ratio and cytoskele-

FIGURE 3. AMPK specifically targets VASP residue Thr-278 in vitro. Puri-fied recombinant human VASP (80 ng) was incubated with active AMPK (100milliunits) at 30 °C. At different time points (0.5, 2, 8, 30 min), aliquots werecollected, separated by SDS-PAGE, and analyzed for phosphorylation byimmunoblotting using phospho-specific antibodies directed against theVASP phosphorylation sites Thr-278 (anti-pThr-278), Ser-157 (anti-pSer-157),or Ser-239 (anti-pSer-239). Polyclonal anti-VASP antibodies quantified totalVASP in the samples. The blot is representative of four independentexperiments.

FIGURE 4. AMPK mutants modulate VASP Thr-278 phosphorylation in cells. ECV304 (A) or HeLa (B) cellswere transiently transfected with cDNAs encoding Myc-tagged wild-type (wt�), dominant-negative (DN�),and constitutively active (CA�) AMPK�1 subunits or with empty vector (MOCK). After 24 h cells were lysed,separated by SDS-PAGE, and analyzed by immunoblotting using antibodies against pThr-278 (anti-pThr-278),total VASP (anti-VASP), Myc-tag (anti-Myc), and the AMPK substrate acetyl-CoA carboxylase phosphorylated atSer-79 (anti-pACC). The blot is representative of four independent experiments. VASP phosphorylation atThr-278 was quantified using densitometric analysis (Scan pack 3.0) and normalized for total VASP protein (n �4; *, p � 0.05, AMPK activity mutants versus MOCK).

AMPK Phosphorylates VASP

4606 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 7 • FEBRUARY 16, 2007

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

FIGURE 5. AMPK-mediated VASP Thr-278 phosphorylation interferes with F-actin assembly. HeLa cells were transiently co-transfected with cDNAs codingfor wild-type (wt�), dominant negative (DN�), or constitutively active (CA�) forms of the AMPK�1 subunit and VASP-mutants. A, schematic representation ofthe VASP mutants AAA, AAE, or AAT in the VASP domain structure. VASP consists of a central proline-rich region (PRR) that is flanked by two Ena/VASPhomology domains (EVH1 and EVH2). The first phosphorylation site Ser-157 is located within the polyproline region, whereas the second and third phospho-rylation sites (Ser-239, Thr-278) are within the EVH2 domain. To block VASP phosphorylation at positions Ser-157, Ser-239, and Thr-278, we exchanged theseresidues to alanines. To imitate phosphorylation of Thr-278, the residue was mutated to glutamic acid. In another case it was left unchanged to allow AMPKphosphorylation specifically at this residue. In the scheme the dark, gray, or open boxes indicate an alanine, glutamic acid, or threonine residue at the indicatedpositions, respectively. B, after 24 h, overexpressing cells were lysed and probed for pThr-278-VASP (anti-pThr-278), total VASP (anti-VASP), or AMPK mutantsusing the Myc tag (anti-Myc) by Western blotting. Representative blots from seven experiments are shown. C, after 24 h, AMPK and VASP mutants overex-pressing cells were fixed, and F-actin was stained with Cy5-phalloidin and quantified using FACS-analysis. F-actin content of VASP AAA-overexpressing cellsand MOCK-transfected cells was set to 100 and 0%, respectively. Column bars represent means � S.D. (n � 7; *, p � 0.007 for the comparisons shown). D, HeLacells in six-well plates were transfected with the reporter pSRE, coding for SRE-driven firefly luciferase, the control plasmid phRL-TK, which codes Renillaluciferase, and expression vectors for AMPK and VASP mutants in the indicated combinations. Cells were harvested after 24 h of culture, and luciferase activitieswere measured in cell lysates as described under “Experimental Procedures,” with firefly luciferase activity normalized to Renilla luciferase activity. Normalizedluciferase activities in VASP-AAA, and MOCK-transfected cells were assigned values of 100 and 0%, respectively. Columns present means � S.D. (n � 4; *, p �0.01). E, immunocytochemistry of VASP and AMPK mutants. Transfected cells were fixed, permeabilized, and stained using antibodies against the Myc tag(anti-Myc) of the AMPK mutants and TRITC-conjugated phalloidin to stain actin fibers. The white arrowheads indicate stress fibers.

AMPK Phosphorylates VASP

FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4607

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

ton dynamics have remained obscure. To analyze potentialeffects of pThr-278 on actin turnover independent of VASPphosphorylation at the Ser-157 and Ser-239 residues, we gen-erated three VASPmutants. Phosphorylation sites Ser-157 andSer-239 were changed into alanine residues. Additionally, Thr-278 was either mutated into a glutamic acid residue, whichmimics a constitutively phosphorylated threonine residue(mutant AAE: S157A, S239A, T278E) or exchanged by an ala-nine (AAA),which cannot be phosphorylated or left unchangedto allow AMPK-mediated phosphorylation (AAT, Fig. 5A). Ini-tially, we confirmed that transiently expressed VASP-AAT is asubstrate for our AMPK mutants (CA�, DN�, wt�). Westernblot analysis with the anti-pThr-278 antibody indicated thatDN�-AMPK decreased, whereas CA�-AMPK increased Thr-278 phosphorylation of VASP-AAT as comparedwith that pro-duced by wt�-AMPK overexpressed in HeLa cells (Fig. 5B). Toevaluate the effect on AMPK-mediated VASP phosphorylationon F-actin assembly, we used a quantitative FACS assay (3) inwhich the mean F-actin content of cells expressing AMPKmutants and VASP-AAT was measured relative to that ofMOCK-transfected cells (0% value). Signals of the F-actin spe-cific dye (Cy5-conjugated phalloidin) were normalized to cellsthat express phosphorylation-resistant VASP-AAA (100%).Overexpression ofVASP-AAE,whichmimicsAMPKphospho-rylated VASP-AAT, reduced cellular F-actin amount to 57% inthe absence of AMPK mutants and served as a positive control(Fig. 5C). Consistently, co-transfection of VASP-AAT withwt�-AMPK or CA�-AMPK diminished F-actin content to 64or 46%, respectively, of the VASP-AAA value. In contrast, inDN�-AMPK- and VASP-AAT-overexpressing cells F-actinamount was only slightly reduced (91%), indicating thatAMPK-driven VASP phosphorylation at the third site nega-tively interferes with F-actin assembly. To confirm that AMPK-

meditated Thr-278 phosphoryla-tion impairs actin polymerizationby an independent approach, weutilized a SRF transcriptionalreporter assay in intact adherentcells (Fig. 5D). The assay quantifiescellular G-actin pool depletion dueto a rise of SRE binding SRF expres-sion (30, 31). The conversion of G-to F-actin both by inducing F-actinassembly and inhibiting F-actin dis-assembly stimulates SRF activity,which increases SRE-dependentexpression of the luciferase reportergene (3). Transient VASP overex-pression induces F-actin assembly(34, 35) and SRE-driven luciferaseactivity (3). Consistently, normal-ized luciferase activity of HeLa cellstransfected with VASPmutants wasmarkedly elevated as comparedwith MOCK-transfected controlcells (set to 0%). Luciferase activitywas maximal in VASP-AAA-trans-fected cells (set to 100%). A single

negative charge at the third site in VASP-AAE reduced the sig-nal to 30%, whereas the activity of VASP-AAT-expressing cellswas in between VASP-AAA and -AAE (66%, not shown). Insupport of our hypothesis that phosphorylation at Thr-278impairs F-actin assembly, co-expression of VASP-AAT withwt�-AMPK and, more pronounced, with CA�-AMPK reducedthe activity (70 and 33%, respectively). In contrast, luciferaseactivity in DN�-AMPK- and VASP-AAT-expressing cells washigh (95%) and close to the VASP-AAA level (100%). Together,these results strongly suggest that AMPK-mediated Thr-278phosphorylation interferes with VASP-induced actin filamentassembly in cells. Additionally, we used immunofluorescenceto visualize the consequences of VASP-Thr-278 phosphoryla-tion on the cytoskeleton in transfected HeLa cells (Fig. 5E).Consistent with the localization of endogenous AMPK� sub-units (36), overexpressedMyc-tagged wt� were found to be dif-fusely distributed in the cytoplasm. Corresponding to the FACSdata and luciferase assay results, these cells presented few actinstress fibers, which were visualized by fluorescence-labeled phal-loidin. Furthermore, stress fibers were amplified in VASP-AAAas comparedwithVASP-AAE transfected cells, which served asinternal controls (supplemental Fig. S1). Coexpression of theVASP-AAT mutant with wt�-AMPK strongly increased actinfilament formation as compared with MOCK-transfected cells.Importantly, actin fiber formation was further enhanced incells expressing DN�-AMPK. In line with the hypothesisthat AMPK activity negatively regulates F-actin content andstress fiber formation, actin filaments were reduced in cells thatco-expressed the CA�-AMPK and VASP-AAT. These datademonstrate that AMPK activity impairs F-actin assembly byVASP Thr-278 phosphorylation, resulting in defective stressfiber formation.

FIGURE 6. AMPK-mediated VASP phosphorylation is decreased in aortic tissue in a rat model of type IIdiabetes mellitus. Aortas were isolated from ZDF and lean rats and rinsed, and tissue lysates were preparedimmediately. Proteins were separated by SDS-PAGE and analyzed by Western blotting. A, aortic tissues wereprobed using antibodies against VASP phosphorylated at Thr-275 (anti-pThr-278) or polyclonal anti-VASPantibodies, which quantify total VASP. B, pThr-275 was quantified from Western blot signals using densitomet-ric analysis and was normalized to total VASP. Columns give means � S.D. (n � 4; *, p � 0.05, lean versus ZDF).C–F, aortas from ZDF versus lean controls were probed for AMPK (anti-AMPK) (C), active AMPK phosphorylatedat Thr-172 (anti-pAMPK) (D), the AMPK substrate acetyl-CoA carboxylase phosphorylated at Ser-79 (anti-pACC)(E), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (anti-GAPDH) (F). The blots show representa-tive results from n � 6 per group.

AMPK Phosphorylates VASP

4608 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 7 • FEBRUARY 16, 2007

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

AMPK-mediated VASP Phosphorylation Is Reduced in theAorta of ZDF Rats—It has been proposed that AMPK signalingcascades may be involved in many of the metabolic perturba-tions observed in type II diabetes mellitus (for review, see Ref.37). To analyzeAMPK-mediatedVASPphosphorylation in dis-eased vessels, we utilized ZDF rats that are leptin-receptor defi-cient and serve as a rodent model of type II diabetes with endo-thelial dysfunction (38) and decreased AMPK activity (39). Weinvestigated the degree of VASP phosphorylation in any of thethree sites and localization of these differentially phosphoryla-

ted proteins in the aorta of diabeticanimals. In rats, VASP residues Ser-154, Ser-236, and Thr-275 corre-spond to the human phosphoryla-tion sites Ser-157, Ser-239, andThr-278, respectively. Notably, ourhuman VASP phospho-specificantibodies cross-react with the rathomologs (not shown). Total VASPprotein in ZDF rat aortas was indis-tinguishable from controls (Fig. 6A,lower panel). pThr-275 signal inten-sities determined from Westernblots were reduced to 30% as com-pared with control animals (lean,100%; Fig. 6, A and B). In contrast,the fraction of Ser-154-phosphoryl-ated VASP (indicated by the shiftedupper band) appeared to be inde-pendent of Thr-275 phosphoryla-tion status (Fig. 6A). In vessels ofcontrol animals, anti-pThr-278 andanti-Thr-278 determined the ratioof Thr-278 phosphorylated to totalVASP as 1:4, consistent with lowpThr-278 levels in non-stimulatedhumanplatelets (7). Consistentwithearlier reports (40, 41), expressionof AMPK protein was not altered indiabetic animals as compared withlean controls (Fig. 6C). The amountof active, Thr-172-phosphorylatedAMPK coincided with pThr-275levels and was largely reduced invessels of diabetic animals (Fig. 6D).Another sign of defective AMPKsignaling was diminished AMPKphosphorylation of ACC (pACClevels shown) in vessels from ZDFrats (Fig. 6E), whereas glyceralde-hyde-3-phosphate dehydrogenaselevels were unaffected (Fig. 6F).To further dissect the phos-

phorylation status of the three phos-phorylation sites, we analyzedaortic ring sections for pSer-154,pSer-236, and pThr-278 using ourphospho-specific anti-VASP anti-

bodies in immunohistochemistry (Fig. 7). In accordance withthe Western blot results, pThr-275 was severely diminished invessels of ZDF rats as compared with lean controls, whereastotal VASP signals were indistinguishable in both groups. Thereduced pThr-275 levels were most conspicuous in the endo-thelial and subendothelial smooth muscle cell layers (Fig. 7,A–F). In contrast, pSer-154 did not differ in aortas of ZDF andcontrol rats as judged from fluorescence intensities in immu-nohistochemical sections, confirming the Western blot analy-ses. Ser-154-phosphorylatedVASP almost completely localized

FIGURE 7. Immunolocalization of differentially phosphorylated VASP in aortic tissue of type II diabeticrats. Aortic tissues were isolated from male ZDF and lean rats and embedded, and then tissue cryosectionswere made and stained with antibodies against total VASP (anti-VASP) followed by Alexa488-conjugatedsecondary antibody (A, D, G, K, N, Q) or with phospho-specific antibodies, which detect rat VASP phosphoryl-ated at Ser-154 (anti-pSer-157) (H and L), at Ser-236 (anti-pSer-239) (O and R), or at Thr-275 (anti-pThr-278) (Band E) followed by Alexa594-coupled secondary antibodies. As controls, sections were stained using antibod-ies against total AMPK (anti-AMPK) (T and W) against active AMPK phosphorylated at Thr-172 (anti-pAMPK) (Uand X) or against the AMPK substrate acetyl-CoA carboxylase, which is AMPK-phosphorylated at Ser-79 (anti-pACC) (V and Y) followed by Alexa488-conjugated secondary antibody. Pictures were taken using a NIKONconfocal scanning microscope equipped with a 100� objective. The pictures show representative results fromn � 5 animals per group.

AMPK Phosphorylates VASP

FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4609

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

at the luminal plasmamembrane of the endothelial layer (Fig. 7,G–M). Consistent with data from models of defective vascularNO/cGMP signaling in rabbits, rats, and humans (12, 13),VASP phosphorylation at the PKG site Ser-236 was largelyreduced both in endothelial and smooth muscle cells of theaortas of ZDF rats (Fig. 7, N–S). Furthermore, in accordancewith theWestern blot signals, immunohistochemistry revealedreduced phosphorylation levels of ACC and AMPK, whereasAMPK antigen was unchanged (Fig. 7, T–Y). These resultsdemonstrate thatAMPK-mediatedThr-275 phosphorylation islargely decreased in the aortas of diabetic rats and suggestVASP pThr-275 as a potential vascular marker for pathologicalglucose metabolism.

DISCUSSION

The present study identifies VASP as a novel AMPK sub-strate in vitro in living cells and in a rat model for diabetes. Wedemonstrate that AMPK-mediated phosphorylation of the res-idue Thr-278 reduces F-actin accumulation activity of VASPand impairs stress fiber formation.Defective cytoskeleton orga-nization has important implications for a variety of actin-basedprocesses such as changes of cellular morphology and cell-cellor cell-matrix adhesion and is suspected to be important forendothelial dysfunction in diabetic vessel disorders (42, 43).Weidentify AMPK as the first kinase that phosphorylates the Thr-278 site of VASP in vivo. AMPK-mediated VASP phosphoryla-tion could provide a directly link from metabolic energy regu-lation to cytoskeletal control.Originally, AMPKwas discovered to inhibit anabolic and ini-

tiate catabolic pathways by rapid, direct phosphorylation oftheir rate-limiting enzymes. Indeed, AMPK activity preservesglucose homeostasis by phosphorylating glycogen synthase and6-phosphofructo-2-kinase (44) and switches off ATP-consum-ing enzymes such as acetyl-CoA carboxylase (33) or hydroxy-methylglutaryl-CoA reductase in anabolic fatmetabolismpath-ways (for a complete list of AMPK substrates see Kahn et al.(15)). Recently, the importance of AMPK-driven signaling cas-cades has evolved. Using transgenic mouse models, AMPK sig-naling was demonstrated to be involved in the regulation offood intake by integrating hormonal and nutrient signals in thehypothalamus (45). Furthermore, AMPK activity is importantin the cardiovascular system since mutations in regulatory �2subunit are associated with hereditary heart disease in humans,i.e. the Wolff-Parkinson-White syndrome (46). Our work fur-ther expands the role of AMPK. We demonstrated that AMPKphosphorylates VASP, a protein critically involved in actinpolymerization, bundling, and nucleation (1, 47). AMPK spe-cifically targets the VASP Thr-278 site, but not Ser-157 andSer-239 (Fig. 3), and the phosphorylation status of Thr-278seems to be independent of PKA- and PKG-mediated VASPphosphorylation at Ser-157 and Ser-239 (Fig. 1C). Indeed, theamino acid sequence flanking Thr-278 (T) in human VASP(GLMEEMNAMLARRRKATQVGEKTP) is a variant of theproposed consensus AMPK recognition motif, harboring thecore motif Hyd-(basic, X)-X-X-(S/T)-X-X-X-Hyd (X is anyamino acid;Hyd is a hydrophobic residue) (16). ThisAMPK siteis located at the C-terminal end of the B-block (positions 259–276) within the VASP EVH2 domain. The EVH2 domain and

especially residues within the B-block are known to promoteF-actin binding to VASP and actin polymerization in vitro.Recently, analyses of VASP-driven F-actin accumulation con-firmed the importance of residues within the B-block for actinpolymerization in living cells (3, 10). Both studies utilized thatthe transcription factor SRF is regulated by changes of the cel-lular G-/F-actin equilibrium. VASP-stimulated SRE-driventranscription is dependent on the F-actin binding B-block adja-cent to the Thr-278 phosphorylation site and, indeed, F-actinassembly driven by VASPmutants lacking this critical segmentis defective (3). Consistent with these studies, our data indicatea crucial role of Thr-278 phosphorylation for VASP-mediatedactin filament formation in cells (Fig. 5). AMPK-mediatedVASP Thr-278 phosphorylation may modulate the electro-static interactions between the nearby arginine residues and thenegatively charged actin filament. Indeed, PKG- and PKA-me-diated phosphorylation of Ser-239 within the A-block, anothermotive important for VASP-mediated F-actin formation, sup-port the concept that phosphorylation of sites within EVH2inhibit actin polymerization in vitro (8) and in cells (10).Although the detailed molecular mechanisms of phospho-

VASP involvement in actin assembly remain to be elucidated,VASP anti-capping and filament bundling activities in vitro areclearly inhibited upon phosphorylation by PKA (6). Interest-ingly, phosphorylation of Ser-239 was primarily responsible forloss of anti-capping activity, whereas phosphorylation of Ena/VASP proteins by PKA at the phosphorylation site conserved inall vertebrate Ena/VASP proteins (Ser-157 in VASP) did notinterfere with anti-capping activity in vitro. Therefore, phos-phorylation at the conserved site (Ser-157) might serve otherfunctions than regulation of actin dynamics (6, 48). PreviousVASP studies investigating PKA- and PKC-mediated phospho-rylation at Ser-157 demonstrated an increased localization ofSer-157 phosphorylated VASP at the lamellipodium in endo-thelial cells (25). Although the function of VASP at this site isnot completely clarified, the protein seems to be important forlamellipodia dynamics and cell motility (49) most probably byits actin bundling activity (5). At the tip of lamellipodia theVASP concentration correlates with the velocity of protrusions(50), suggesting that VASP participates in actin turnover at thissite. At the leading edge, Ser-157 phosphorylation is dynami-cally regulated and increased during cell detachment and reat-tachment but reduced during cell spreading (51). As shown forthe VASP Ser-157 phosphorylation site, spatially regulatedphosphorylation of the Ser-239 and Thr-278 residues withinthe EVH2 domain by cyclic nucleotide- and AMPK-mediatedsignaling pathways could regulate VASP anti-capping and bun-dling activities in vivo. However, clarification of the precise spa-tial and temporal regulation of differential Ena/VASPphospho-rylation patterns is a challenging topic requiring futuresystematic investigation.One of the unexpected findings of this study is that the cyclic

nucleotide-dependent protein kinases PKA and PKG do notphosphorylate VASP at the Thr-278 residue in living cells. Pre-vious data, based primarily on in vitro experiments using puri-fied PKA catalytic subunit and PKG suggested that Thr-278might be targeted by these kinases (7). However, even afterlong-term incubation with PKA the magnitude of pThr-278

AMPK Phosphorylates VASP

4610 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 7 • FEBRUARY 16, 2007

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

was minor when compared with pSer-157 and pSer-239 phos-phorylation in vitro (7, 8, 10). The reasons for the disparitybetween results of the previously published in vitro studies andour investigations of VASP phosphorylation in cells is notentirely clear, but results obtained in intact cells provide morephysiologically relevant data. Indeed, in platelets specific acti-vation of PKA and PKG increased phosphorylations at the firstand second site but did not did not alter pThr-278 (7). In con-trast, AMPK specifically phosphorylates Thr-278 in vitro in pri-mary and stable cell lines and in animal models. The basalpThr-278 levels (Fig. 2) corresponds to the basal AMPK activityobserved in endothelial cells (52).The present study focuses on AMPK-mediated VASP phos-

phorylation specifically in endothelial cells and vascular walls,because AMPK signaling is impaired in diabetic vessels andmay contribute to diabetic retinopathy, nephropathy, ormicro-vascular leakage. An important finding was that decrease ofpThr-278was accompanied by reduced pSer-239, but not pSer-157, in vessels of diabetic animals (Fig. 7). The concomitantdecrease of pSer-239 and pThr-278 levels suggests the exist-ence of cross-talk between NO/cGMP signaling, which regu-lates Ser-239 phosphorylation in the endothelium (11), andAMPK-mediated pathways in diabetic conditions (Figs. 6and 7). Several mechanisms might explain co-regulation ofNO/cGMP-dependent phosphorylation of VASP-Ser-239 withAMPK-mediated phosphorylation of VASP-Thr-278. Recently,a direct link between AMPK and NO/cGMP signaling in endo-thelial cells was suggested by the fact that Metformin-stimu-lated AMPK can phosphorylate and activate endothelial nitricoxide synthase by an Hsp90-dependent mechanism (53). Acti-vated endothelial nitric oxide synthase stimulates PKG andincreases pSer-239. Conversely, pathological glucose metabo-lism, characterized by reduced AMPK activity, is associatedwith intracellular mediators like reactive oxygen species, whichcan uncouple endothelial nitric oxide synthase and, thus,decrease vascular bioavailability of NO (for a recent review, seeRef. 54) and in turn PKG activity and pSer-239. The coupling ofpSer-239 and pThr-278 seems to be of functional importancefor actin dynamics. In vitro studies demonstrate that phospho-rylation of both Ser-239 and Thr-278 inhibit actin polymeriza-tion (8). Consistently, pSer-239 and pThr-278 synergisticallyinterfere with F-actin formation in cells.4 Interestingly, actincytoskeleton is impaired in retinal endothelial cells of diabeticrats (55), and AMPK activation using 5-aminoimidazole-4-car-boxamide-1�-riboside or exercise, which activates AMPK inmany tissues, prevented the development of diabetes in ZDFrats (39, 56), supporting the contribution of AMPK signalingfor diabetic vessel disorders. It will be interesting to determinethe phosphorylation status of the three VASP phosphorylationsites in this model using our phospho-specific antibodies. Theset of phosphorylation specific antibodies offers the opportu-nity to quantify and to dissect PKA-, PKG-, and AMPK-medi-ated signaling via differential VASP phosphorylations in vivo.In summary, this study identifies VASP as new AMPK sub-

strate. AMPK is responsible for phosphorylation of the VASP

Thr-278 site in vitro, in living cells, and in vivo. AMPK-medi-ated VASP Thr-278 phosphorylation interferes with actinpolymerization and impairs formation of stress fibers, thus cre-ating a novel link between glucose metabolism and regulationof the endothelial actin cytoskeleton. AMPK activity andAMPK-mediated Thr-278 phosphorylation are reduced in thevascular wall of a rat model of type II diabetes mellitus, withpotential implications for diabetic actin-based vascular disor-ders such as microvascular leakage or diabetic retinopathy.

Acknowledgments—We are grateful to Dr. J. Bauersachs for donatingrat aortas, Dr. O. Ritter for providing the protein phosphatase 2BcDNA, Dr. Kai Schuh for helpful discussions, Dr. Stepan Gambaryanand Dr. S. M. Lohmann for critically reading the manuscript, andDr. Sabine Herterich for sequencing the VASP mutants (all at theUniversity of Wurzburg).

REFERENCES1. Krause, M., Dent, E. W., Bear, J. E., Loureiro, J. J., and Gertler, F. B. (2003)

Annu. Rev. Cell Dev. Biol. 19, 541–5642. Sechi, A. S., and Wehland, J. (2004) Front. Biosci. 9, 1294–13103. Grosse, R., Copeland, J. W., Newsome, T. P., Way, M., and Treisman, R.

(2003) EMBO J. 22, 3050–30614. Bear, J. E., Svitkina, T. M., Krause, M., Schafer, D. A., Loureiro, J. J.,

Strasser, G. A., Maly, I. V., Chaga, O. Y., Cooper, J. A., Borisy, G. G., andGertler, F. B. (2002) Cell 109, 509–521

5. Schirenbeck, A., Arasada, R., Bretschneider, T., Stradal, T. E., Schleicher,M., and Faix, J. (2006) Proc. Natl. Acad. Sci. U. S. A. 103, 7694–7699

6. Barzik, M., Kotova, T. I., Higgs, H. N., Hazelwood, L., Hanein, D., Gertler,F. B., and Schafer, D. A. (2005) J. Biol. Chem. 280, 28653–28662

7. Butt, E., Abel, K., Krieger, M., Palm, D., Hoppe, V., Hoppe, J., andWalter,U. (1994) J. Biol. Chem. 269, 14509–14517

8. Harbeck, B., Huttelmaier, S., Schluter, K., Jockusch, B.M., and Illenberger,S. (2000) J. Biol. Chem. 275, 30817–30825

9. Halbrugge, M., Friedrich, C., Eigenthaler, M., Schanzenbacher, P., andWalter, U. (1990) J. Biol. Chem. 265, 3088–3093

10. Zhuang, S., Nguyen, G. T., Chen, Y., Gudi, T., Eigenthaler, M., Jarchau, T.,Walter, U., Boss, G. R., and Pilz, R. B. (2004) J. Biol. Chem. 279,10397–10407

11. Smolenski, A., Poller, W., Walter, U., and Lohmann, S. M. (2000) J. Biol.Chem. 275, 25723–25732

12. Munzel, T., Feil, R.,Mulsch, A., Lohmann, S.M., Hofmann, F., andWalter,U. (2003) Circulation 108, 2172–2183

13. Schulz, E., Tsilimingas, N., Rinze, R., Reiter, B., Wendt, M., Oelze, M.,Woelken-Weckmuller, S., Walter, U., Reichenspurner, H., Meinertz, T.,and Munzel, T. (2002) Circulation 105, 1170–1175

14. Carling, D. (2004) Trends Biochem. Sci. 29, 18–2415. Kahn, B. B., Alquier, T., Carling, D., and Hardie, D. G. (2005) Cell Metab.

1, 15–2516. Hardie, D. G. (2003) Endocrinology 144, 5179–518317. Lizcano, J.M., Goransson,O., Toth, R., Deak,M.,Morrice,N.A., Boudeau,

J., Hawley, S. A., Udd, L., Makela, T. P., Hardie, D. G., and Alessi, D. R.(2004) EMBO J. 23, 833–843

18. Woods, A., Dickerson, K., Heath, R., Hong, S. P., Momcilovic, M., John-stone, S. R., Carlson, M., and Carling, D. (2005) Cell Metab. 2, 21–33

19. Kemp, B. E., Stapleton, D., Campbell, D. J., Chen, Z. P., Murthy, S.,Walter,M., Gupta, A., Adams, J. J., Katsis, F., vanDenderen, B., Jennings, I. G., Iseli,T., Michell, B. J., and Witters, L. A. (2003) Biochem. Soc. Trans. 31,162–168

20. Hardie, D. G. (2004) J. Cell Sci. 117, 5479–548721. Foretz, M., Ancellin, N., Andreelli, F., Saintillan, Y., Grondin, P., Kahn, A.,

Thorens, B., Vaulont, S., and Viollet, B. (2005) Diabetes 54, 1331–133922. Zhou, G., Myers, R., Li, Y., Chen, Y., Shen, X., Fenyk-Melody, J., Wu, M.,

Ventre, J., Doebber, T., Fujii, N., Musi, N., Hirshman, M. F., Goodyear,4 C. Blume, P. M. Benz, U. Walter, and T. Renne, unpublished results.

AMPK Phosphorylates VASP

FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4611

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

L. J., and Moller, D. E. (2001) J. Clin. Investig. 108, 1167–117423. Minokoshi, Y., Kim, Y. B., Peroni, O. D., Fryer, L. G., Muller, C., Carling,

D., and Kahn, B. B. (2002) Nature 415, 339–34324. Ahima, R. S. (2006) Nat. Med. 12, 511–51225. Profirovic, J., Gorovoy, M., Niu, J., Pavlovic, S., and Voyno-Yasenetskaya,

T. (2005) J. Biol. Chem. 280, 32866–3287626. Renne, T., Schuh, K., and Muller-Esterl, W. (2005) J. Immunol. 175,

3377–338527. Smolenski, A., Bachmann, C., Reinhard, K., Honig-Liedl, P., Jarchau, T.,

Hoschuetzky, H., and Walter, U. (1998) J. Biol. Chem. 273, 20029–2003528. Abel, K., Mieskes, G., and Walter, U. (1995) FEBS Lett. 370, 184–18829. Woods, A., Azzout-Marniche, D., Fouretz, M., Stein, S. C., LeMarchand,

P., Ferre, P., Foufelle, F., and Carling, D. (2000) Mol. Cell. Biol. 20,6704–6711

30. Sotiropoulos, A., Gineitis, D., Copeland, J., and Treisman, R. (1999) Cell98, 159–169

31. Geneste, O., Copeland, J. W., and Treisman, R. (2002) J. Cell Biol. 157,831–838

32. Stein, S. C.,Woods, A., Jones, N. A., Davison,M. D., andCarling, D. (2000)Biochem. J. 345, 437–443

33. Davies, S. P., Carling, D., and Hardie, D. G. (1989) Eur. J. Biochem. 186,123–128

34. Bachmann, C., Fischer, L., Walter, U., and Reinhard, M. (1999) J. Biol.Chem. 274, 23549–23557

35. Price, C. J., and Brindle, N. P. (2000) Arterioscler. Thromb. Vasc. Biol. 20,2051–2056

36. da Silva Xavier, G., Leclerc, I., Salt, I. P., Doiron, B., Hardie, D. G., Kahn, A.,and Rutter, G. A. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 4023–4028

37. Winder, W. W., and Hardie, D. G. (1999) Am. J. Physiol. 277, E1–E1038. Unger, R. H., and Orci, L. (2001) FASEB J. 15, 312–32139. Yu, X.,McCorkle, S.,Wang,M., Lee, Y., Li, J., Saha, A. K., Unger, R. H., and

Ruderman, N. B. (2004) Diabetologia 47, 2012–202140. Chen, M. B., McAinch, A. J., Macaulay, S. L., Castelli, L. A., O’Brien P, E.,

Dixon, J. B., Cameron-Smith, D., Kemp, B. E., and Steinberg, G. R. (2005)

J. Clin. Endocrinol. Metab. 90, 3665–367241. Wang, M. Y., and Unger, R. H. (2005) Am. J. Physiol. Endocrinol. Metab.

288, 216–22142. Brownlee, M. (2001) Nature 414, 813–82043. Frank, R. N. (2004) N. Engl. J. Med. 350, 48–5844. Marsin, A. S., Bouzin, C., Bertrand, L., and Hue, L. (2002) J. Biol. Chem.

277, 30778–3078345. Minokoshi, Y., Alquier, T., Furukawa, N., Kim, Y. B., Lee, A., Xue, B., Mu,

J., Foufelle, F., Ferre, P., Birnbaum,M. J., Stuck, B. J., and Kahn, B. B. (2004)Nature 428, 569–574

46. Arad, M., Benson, D. W., Perez-Atayde, A. R., McKenna, W. J., Sparks,E. A., Kanter, R. J., McGarry, K., Seidman, J. G., and Seidman, C. E. (2002)J. Clin. Investig. 109, 357–362

47. Reinhard, M., Jarchau, T., and Walter, U. (2001) Trends Biochem. Sci. 26,243–249

48. Loureiro, J. J., Rubinson, D. A., Bear, J. E., Baltus, G. A., Kwiatkowski, A. V.,and Gertler, F. B. (2002)Mol. Biol. Cell 13, 2533–2546

49. Bear, J. E., Loureiro, J. J., Libova, I., Fassler, R., Wehland, J., and Gertler,F. B. (2000) Cell 101, 717–728

50. Rottner, K., Behrendt, B., Small, J. V., andWehland, J. (1999)Nat. Cell Biol.1, 321–322

51. Howe, A. K., Hogan, B. P., and Juliano, R. L. (2002) J. Biol. Chem. 277,38121–38126

52. Drew, B. G., Fidge, N. H., Gallon-Beaumier, G., Kemp, B. E., and Kingwell,B. A. (2004) Proc. Natl. Acad. Sci. U. S. A. 101, 6999–7004

53. Davis, B. J., Xie, Z., Viollet, B., andZou,M.H. (2006)Diabetes 55, 496–50554. Munzel, T., Daiber, A., Ullrich, V., and Mulsch, A. (2005) Arterioscler.

Thromb. Vasc. Biol. 25, 1551–155755. Yu, P. K., Yu, D. Y., Cringle, S. J., and Su, E. N. (2005) Curr. Eye Res. 30,

279–29056. Pold, R., Jensen, L. S., Jessen,N., Buhl, E. S., Schmitz,O., Flyvbjerg, A., Fujii,

N., Goodyear, L. J., Gotfredsen, C. F., Brand, C. L., and Lund, S. (2005)Diabetes 54, 928–934

AMPK Phosphorylates VASP

4612 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 7 • FEBRUARY 16, 2007

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Thomas RennéConstanze Blume, Peter M. Benz, Ulrich Walter, Joohun Ha, Bruce E. Kemp and

by Phosphorylating Vasodilator-stimulated PhosphoproteinAMP-activated Protein Kinase Impairs Endothelial Actin Cytoskeleton Assembly

doi: 10.1074/jbc.M608866200 originally published online November 2, 20062007, 282:4601-4612.J. Biol. Chem. 

  10.1074/jbc.M608866200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

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

  http://www.jbc.org/content/282/7/4601.full.html#ref-list-1

This article cites 51 references, 26 of which can be accessed free at

by guest on March 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from