Proc. Natl. Acad. Sci. USAVol. 92, pp. 7956-7960, August 1995Cell Biology

Identification, purification, and characterization of a zyxin-related protein that binds the focal adhesion and microfilamentprotein VASP (vasodilator-stimulated phosphoprotein)

(cAMP/cGMP/protein kinase/actin polymerization/leading edge)

MATrHiAS REINHARD*, KARIN JOUVENALt, DOMINIQUE TRIPIERt, AND ULRICH WALTER*t*Medizinische Universitatsklinik, Klinische Biochemie und Pathobiochemie, Josef-Schneider-Strasse 2, D-97080 Wurzburg, Federal Republic of Germany; andtPharmaforschung H825, Hoechst AG, D-65926 Frankfurt/Main, Federal Republic of Germany

Communicated by David L. Garbers, University of Texas Southwestern Medical Center, Dallas, 7X, April 24, 1995 (received for reviewFebruary 15, 1995)

ABSTRACT VASP (vasodilator-stimulated phosphopro-tein), an established substrate of cAMP- and cGMP-dependentprotein kinases in vitro and in living cells, is associated withfocal adhesions, microfilaments, and membrane regions ofhigh dynamic activity. Here, the identification of an 83-kDaprotein (p83) that specifically binds VASP in blot overlays ofdifferent cell homogenates is reported. With VASP overlays asa detection tool, p83 was purified from porcine platelets andused to generate monospecific polyclonal antibodies. VASPbinding to purified p83 in solid-phase binding assays and theclosely matching subcellular localization in double-label im-munofluorescence analyses demonstrated that both proteinsalso directly interact as native proteins in vitro and possibly inliving cells. The subcellular distribution, the biochemicalproperties, as well as microsequencing data revealed thatporcine platelet p83 is related to chicken gizzard zyxin andmost likely represents the mammalian equivalent of thechicken protein. The VASP-p83 interaction may contribute tothe targeting ofVASP to focal adhesions, microfilaments, anddynamic membrane regions. Together with our recent identi-fication of VASP as a natural ligand of the profilin poly-(L-proline) binding site, our present results suggest that, bylinking profilin to zyxin/p83, VASP may participate in spa-tially confined profilin-regulated F-actin formation.

Focal adhesions are transmembrane junctions between termi-nal portions of microfilaments and the underlying extracellularmatrix with the heterodimeric integrins as the prevailingtransmembrane adhesive receptors (1-3). Understanding ofthe molecular basis of integrin-dependent cell adhesion andassociated signaling events (4, 5) requires elucidation of thefocal adhesion architecture. Based mostly on in vitro assays,multiple low-to-moderate affinity interactions have beenshown that allow the construction of several interdigitatingroutes that appear to link actin filaments to the transmembraneintegrins and provide docking points for different regulatoryproteins (for a review, see refs. 1-3). For instance, vinculininteracts with a-actinin and talin, which both bind to thecytoplasmic domains of integrin ,B subunits, and all three ofthem have actin-binding activity. Vinculin and a-actinin inaddition have been recognized as paxillin- and zyxin-bindingproteins, respectively. Although considerable insight into themolecular composition and structural organization of focaladhesions has been gained (1-3), current concepts concerningthe functional relationships between individual constituentsare still quite fragmentary.

Originating from the analyses of cyclic nucleotide-dependentplatelet inhibition, we characterized and purified the vasodi-

lator-stimulated phosphoprotein (VASP) as an in vitro and invivo substrate ofcAMP- and cGMP-dependent protein kinases(6-8). Cyclic nucleotide-dependent VASP phosphorylationlies at a point of convergence of two signaling pathways thatinhibit platelet activation and associated events like adhesion,shape change, and aggregation (9). Subsequently, VASP wasshown to localize to focal adhesions, stress fibers, highlydynamic membrane regions, and cell-cell contacts of certainepithelial cells (10, 11). Molecular cloning established VASPas a novel protein with a unique proline-rich central domain(11). Very recently, we identified VASP as a natural ligand forthe poly(L-proline)-binding site of profilin (12). Both VASPand the G-actin and phosphatidylinositol phosphates-bindingprotein profilin are important members of signal-transductionpathways (9, 13, 14) and may act in concert to relay signaltransduction to the actin cytoskeleton. Therefore, it is ofconsiderable interest to identify additional VASP-bindingproteins in order to define the structural and functionalintegration of VASP at its subcellular locations. Here wereport the identification and purification of a 83-kDa VASP-binding protein and its characterization as a mammalianprotein related to chicken zyxin (15-17).

MATERIALS AND METHODSPurification of VASP. VASP was purified from human

platelets essentially as described (6) with the modificationsindicated (10).

Porcine platelet VASP was purified as a byproduct of p83.Pooled hydroxylapatite fractions (40-100 mM phosphate)obtained during the purification of p83 from the particulateplatelet fraction (see below) were subjected to sequentialorange A and Mono-S HR chromatography (6) followed by afinal hydroxylapatite step.

[32P]VASP Overlay. For [32P]VASP overlays, proteins wereseparated by SDS/PAGE and blotted onto nitrocellulose.Nitrocellulose sheets were blocked as described (10) and wereincubated for at least 30 min at room temperature with[32P]VASP at 0.1 ,ug/ml in blocking medium (10) supple-mented with 0.5 mM dithiothreitol. Nitrocellulose sheets werewashed three times for 5 min with phosphate-buffered saline(PBS) containing 0.3% Triton X-100, 0.05% Tween 20, and 0.5mM dithiothreitol. Dried sheets were exposed to an autora-diographic film.For radiolabeling of VASP, porcine platelet VASP was

phosphorylated by cGMP-dependent protein kinase (6) at30°C in the presence of 2.5 ,uM [y-32P]ATP (5.9 TBq/mmol)for 1 h followed by an additional 30 min with 200 ,uM [t-32p]

Abbreviations: VASP, vasodilator-stimulated phosphoprotein; 2-D,two dimensional.ITo whom reprint requests should be addressed.

7956

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

1, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995) 7957

ATP (64 GBq/mmol). Free ATP was removed by centrifugationthrough a MicroSpin S-200 HR column (Pharmacia).For the overlay of chicken blood cell proteins separated by

two-dimensional (2-D) gel electrophoresis, human [32P]VASPat 0.8 jig/ml, phosphorylated with [y-32P]ATP of 3-fold higherspecific activity, was used.

Purification of p83. The soluble fraction (45,900 x g super-natant) of a hypotonic platelet lysate, obtained from 8 x 1012porcine platelets essentially as described (6), was subjected toone additional freeze/thaw cycle, adjusted to pH 7.0, clearedby centrifugation, and applied to an S-Sepharose FF (Phar-macia) column (4 x 33 cm). This and all subsequent steps wereperformed at 4°C. Alternatively, a 250mM NaCl extract of theparticulate platelet fraction dialyzed against buffer A (10 mMNaPi, pH 7.0/4 mM EDTA/50 mM 2-mercaptoethanol) wasused as starting material for S-Sepharose FF-chromatography.

Proteins were eluted from the S-Sepharose FF column witha 1000-ml gradient from 0 to 500 mM NaCl prepared in bufferA. Fractions that were positive for p83 in the [32P]VASPoverlay procedure were pooled, and p83 was precipitated bythe addition of solid ammonium sulfate (final concentration,30% saturation). The precipitate was dissolved in about 6 mlof buffer A containing 75 mM NaCl, dialyzed against the samebuffer, and loaded (0.5 ml/min) onto a column of 5-6 ml ofhydroxylapatite (Bio-Gel HT, Bio-Rad). The column waseluted (1 ml/min) with a 35-ml linear gradient from 10 mMKH2PO4 adjusted with NaOH to pH 7.5 containing 50 mM2-mercaptoethanol and 75 mM NaCl to 100 mM KH2PO4/NaOH (pH 7.5) containing only 50 mM 2-mercaptoethanol.P83-containing fractions were pooled, dialyzed against 10 mMKH2PO4/NaOH, pH 7.5/150 mM NaCl and applied (0.25ml/min) to a 1-ml HiTrap chelat column (Pharmacia), pre-charged with 500 ,1 of 100mM ZnCl2. The column was washedwith buffer B (25 mM sodium borate, pH 7.5/1 M NaCl).Contaminating proteins together with some p83 were eluted(0.5 ml/min) by 15 ml of 0.25 mM ZnCl2 in buffer B.Subsequent elution with 1 mM ZnCl2 in buffer B yieldedapparently homogeneous p83 as judged from Coomassie blue-stained gels. After removal of the ZnCl2, p83 was dialyzedagainst 25 mM Tris-HCl, pH 8.0/0.1 mM EGTA/150 mMNaCl/0.015% Nonidet P-40 (Boehringer Mannheim)/20%(vol/vol) glycerol/0.5 mM dithiothreitol for storage at -80°C.Alternatively, protamine agarose chromatography was used asthe final purification step. The sample eluted from the hy-droxylapatite column was applied (0.25 ml/min) to 2.5 ml ofprotamine agarose (Sigma) in 10 mM KH2PO4/NaOH, pH7.5/150 mM NaCl/4 mM EDTA/50 mM 2-mercaptoethanol/20% glycerol. The column was washed with the same buffercontaining 2 M NaCl and eluted (0.5 ml/min) with 300 mMarginine hydrochloride (pH 9.0)/2 M NaCl/4 mM EDTA/50mM 2-mercaptoethanol, yielding essentially pure p83. Con-centrations of purified p83 were estimated from Coomassieblue-stained gels using bovine serum albumin as a standard.Immunological Methods. A New Zealand White female

rabbit was immunized with electrophoretically purified andelectroeluted p83 starting from partially purified preparations.The p83 rabbit antiserum AS83-1 and a polyclonal mouseanti-VASP serum were affinity-purified essentially as de-scribed (18) by using the respective purified proteins blottedonto nitrocellulose. Immunoblotting (Western blotting) andimmunofluorescence analysis were done essentially as de-scribed (10).

Solid-Phase Binding Assay. Solid-phase binding assays weredone essentially as described (12) with 10-80 ng of p83 per mlfor coating and 2.2 ,ug of VASP per ml as the soluble ligand.Two-Dimensional Gel Electrophoresis of Chicken Blood

Cell Proteins. Blood cells were pelleted for 20 min at 940 x gfrom the 60 X g (30 min) supernatant of anti-coagulatedchicken blood and were washed twice. Proteins prepared in 9M urea/2% (wt/vol) Nonidet P-40 (Boehringer Mannheim)/

2% (vol/vol) 2-mercaptoethanol/0.8% (wt/vol) Ampholines,pH 3-10 (Serva), were separated by isoelectric focusing at 15°Cfor 24,700 V-hr on Immobiline dry strips (pH 3-10.5, Phar-macia) in the presence of the internal standards carbonicanhydrase II from bovine erythrocytes (pl 5.4, Sigma) andmyoglobin from horse heart (pl 6.8 and 7.2, Sigma). Afterfocusing, the strips were prepared for SDS/PAGE and appliedto an SDS slab gel.

Microsequencing and Sequence Analysis. Partially purifiedp83 (hydroxylapatite fractions) was CCl3COOH-precipitatedand separated by SDS/PAGE. The Coomassie blue-stained gelpiece containing p83 was excised, dehydrated by washing with1:1 (vol/vol) 0.2 M N-methylmorpholine acetate, pH 8/ace-tonitrile, and reswollen with 0.2 M N-methylmorpholine ace-tate, pH 8/0.02% Tween-20 containing endoproteinase Lys-C(sequencing grade, Boehringer Mannheim). After a 4-hr in-cubation at 30°C, the resulting peptides were expelled from thegel matrix by a shrinking step with 60% (vol/vol) acetonitrile/0.1% CF3COOH. Peptides were separated by HPLC on a C8column (1 x 250 mm; Applied Biosystems), and selectedpeptides were subjected to microsequencing with a model477A protein sequencer (Applied Biosystems).

Protein data base searches and sequence alignments weremade by using the FASTA and BESTFIT programs [GeneticsComputer Group (Madison, WI) (GCG) program package;ref. 19], respectively. For statistical evaluation, z values (20)were calculated for the BESTFIT similarity scores from themean and standard deviation of scores of the randomlyshuffled sequences.

RESULTSIdentification of a 83-kDa VASP-Binding Protein (p83) in

Blot Overlays. When total platelet proteins, separated bySDS/PAGE and transferred to nitrocellulose, were probedwith [32P]VASP, a single protein of -83 kDa (p83) wasintensely labeled (Fig. 1). Occasionally, additional smallerbands of various sizes were observed, which most likelyrepresented degradation products (e.g., see Fig. 3, lanes 1' and

Ox lOx 50xOngin -

97.4 -

66.357.6

42.839.2

25.8

Dye front

FIG. 1. Competition by unlabeled VASP of [32P]VASP binding top83 present in total platelet homogenates (autoradiogram). Proteins ofa human platelet homogenate (25 ,ug per lane) were separated bySDS/PAGE and blotted onto nitrocellulose. Individual lanes wereincubated with 0.1 ,ug of [32P]VASP per ml in the absence (lane OX)or presence of a 10-fold (lane lOX) or 50-fold (lane 5OX) excess ofunlabeled VASP.

Cell Biology: Reinhard et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

1, 2

020

7958 Cell Biology: Reinhard et al.

2'). Competition with a 10- to 50-fold excess to unlabeledVASP quenched [32P]VASP binding to p83, demonstratingthat this interaction is specific and saturable (Fig. 1).

Purification of p83 from Porcine Platelets. P83 was purifiedfrom porcine platelets as outlined in Fig. 2A by using[32P]VASP overlays as a detection tool. Detection of theVASP-binding protein p83 was strongly dependent op thepresence of a specific fibrinogen receptor (integrin a1nb33)antagonist or on high concentrations of a Ca2+ chelator duringplatelet preparation or platelet lysis, respectively (not shown),suggesting that p83 is susceptible to platelet aggregation-dependent proteolysis by calpain (21, 22). In a first chromato-graphic step VASP-binding activity was recovered at neutralpH in the 70-220 mM NaCl eluate of a cation exchanger(S-Sepharose FF) and was nearly quantitatively precipitated bythe addition of ammonium sulfate (30% saturation). In asubsequent hydroxylapatite step, about half of the p83 loadedonto the column was recovered with an equivalent of 25-40mM phosphate-i.e., before the bulk of other proteins waseluted. Therefore, hydroxylapatite chromatography is an effi-cient purification step. Fractions eluted with 40-60 mMphosphate, though containing most of the remaining p83, werediscarded because coeluting contaminants were difficult toremove. Finally, proteins from pooled fractions were separatedon a chelating column that had been precharged with ZnCl2.Elution with 250 ALM ZnCl2 yielded both p83 and several lowermolecular weight proteins (Fig. 2B, lane 1), whereas subse-quent elution with 1 mM ZnCl2 resulted in apparently homo-geneous p83 (Fig. 2B, lane 2). Alternatively, chromatographyon protamine agarose and arginine (pH 9.0) elution was usedas a final purification step with similar efficiency. From a totalof 8 x 1012 platelets, obtained from 60 liters of porcine blood,-10-20 ,ug of p83 could be purified.

Aplatelet lysate

45,900 x g

supematant pellet250 mM NaCIextract

S-Sepharose FF chromatography

ammonium sulfate precipitation(0-30% saturation)

hydroxylapatite chromatography

Zn2t-chelate protamine agarosechromatography chromatography

1 mM ZnCelution

B0.c

X0lC* OC

purified p83

-0

I~

0 10 15 20 25Elution Volume, ml

30

,/ arginine elution(pH 9)

1.00

'.75 '

).50 E

-25 cN

FIG. 2. Puriflcation of p83 from porcine platelets. (A) Flow-chartfor purification of p83. (B Left) Elution profile of the final purificationstep, Zn2+-chelat chromatography. (B Right) Analysis of aliquots ofthe 0.25 mM ZnCl2 eluate (lane 1) and the subsequent 1 mM ZnCl2eluate (lane 2) by SDS/PAGE and Coomassie blue-staining.

VASP and p83 Interact as Native Proteins. Blot overlaysshowed that VASP binds to denatured p83. Therefore, it wasof interest to examine whether VASP and p83 also interact asnative proteins. This question was addressed by using solid-phase binding assays in which purified p83 was coated to thesurface of microtiter wells and VASP was applied as a solubleligand. VASP strongly bound to surface-coated p83, demon-strating that the native proteins interact directly (data notshown).VASP and p83 Show a Closely Matching Subcellular Dis-

tribution. A rabbit antiserum AS83-1 raised against porcineplatelet p83 was characterized in Western blots of porcine andhuman platelets, chicken blood cells, and human skin fibro-blasts. In all but the chicken cells, p83 was specifically stained.In the human platelet preparation, minor putative degradationproducts were also labeled by AS83-1 antiserum (Fig. 3, lane2"). [32P]VASP overlays of proteins from chicken blood cellsrevealed VASP binding to a protein of slightly smaller size(Fig. 3, lane 3'), which did not react with the antiserum (Fig.3, lane 3"). Some species selectivity of AS83-1 antiserum wasalso evident when labeling of porcine and human p83 werecompared (Fig. 3; compare lanes 1' and 2' to lanes 1" and2"; note that the protein applied in lane 1" is 20% of that inother lanes).

If VASP and p83 interact directly in living cells, they areexpected to display an overlapping subcellular distribution.This was investigated by double-label immunofluorescenceusing the monospecific rabbit antiserum (AS83-1) raisedagainst porcine platelet p83 and a polyclonal mouse anti-VASPserum S2. Fig. 4 presents a comparison of VASP and p83localization in human dermal fibroblasts. The subcellularlocalization of both proteins was indistinguishable. In partic-ular, focal adhesions, microfilaments of stress fibers (Fig. 4 a,a', b, and b'), and arcs (not shown), and the periphery ofprotruding lamellae (Fig. 4 c and c') were labeled both byaffinity-purified antibodies and the original antisera againstVASP (Fig. 4 a '-c'; ref. 10) and p83 (Fig. 4 a-c). Both proteinswere located at stress fibers in a characteristic dotted pattern.A more detailed inspection revealed that the stress fiber

s°s% AS~~~~~C

1 2 3 4 1' 2' 3' 4' 1" 2" 311 4"

kDa

97.4-66.3-57.6-42.8_39.225.8

CoomassieBlue

Overlay Anti-p83

FIG. 3. Analysis of p83 expression by Western blotting and[32P]VASP binding. Proteins (25 ,ug per lane, except for lane 1" with5 ,ug) from porcine platelets (lanes 1-1"), human platelets (lanes 2-2"),chicken blood cells (lanes 3-3"), and human skin fibroblasts (lanes4-4") were separated by SDS/PAGE and analyzed by Coomassie bluestaining (lanes 1-4), [32P]VASP overlay (lanes 1'-4'), and Westernblotting using the AS83-1 anti-p83 serum at a 1:1000 dilution (lanes1"-4"). Note that the chicken VASP-binding protein (lane 3') migratedslightly faster than the equivalent mammalian proteins (lanes 1', 2',and 4').

Proc. Natl. Acad Sci. USA 92 (1995)

.1

1 2.... OHgin

-97.4-66.3-57.6.42.8'39.2

-25.8- Dye front

Dow

nloa

ded

by g

uest

on

Mar

ch 2

1, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995) 7959

FIG. 4. Comparison of the subcellular distribution of p83 (a-c) and VASP (a'-c') by double-label immunofluorescence analysis of human skinfibroblasts. Localization of p83 and VASP was revealed by polyclonal rabbit anti-p83 (AS83-1) and mouse anti-VASP (S2) antisera followed byrhodamine-conjugated donkey anti-rabbit and FITC-conjugated sheep anti-mouse secondary antibodies, respectively. (a and a') Survey ofsubcellular p83 and VASP distribution. (b and b') p83 and VASP association with focal adhesions and stress fibers, which show a dotted stainingpattern. (c and c') Colocalization of p83 and VASP at the periphery of a protruding lamella (arrow heads). [Bar in c' (valid for all panels) = 20gm.]decoration with p83 showed the same periodicity and spacingand coincided with that of VASP (data not shown).P83 Is a Zyxin-Related Protein. Partially purified p83

(hydroxylapatite eluate) was separated by SDS/PAGE, excisedfrom the gel, and subjected to proteolytic digestion. Microse-quencing of the resulting peptides yielded two peptide se-quences of 27 (peptide A) and 24 (peptide B) amino acids inlength. Both exhibited significant (average z values > 8) butlimited homology to chicken zyxin sequences (Fig. 5) withinthe proline-rich and third LIM domain, respectively (23).These data suggested that p83 is a zyxin-related protein,possibly the mammalian homologue of the chicken gizzardprotein. Zyxin, a low-abundance focal adhesion and stressfiber-associated a-actinin-binding protein (15, 16, 24), haspreviously been purified from chicken gizzard (16). In humanplatelets, a possible zyxin homologue has been observed, whichhas a somewhat higher apparent molecular mass (84 kDa) thanthe 82-kDa chicken protein (25). This is in agreement with theapparent molecular mass difference observed between mam-malian (Fig. 3, lanes 1', 2', and 4') and chicken (Fig. 3, lane 3')VASP-binding proteins. To investigate whether the chickenprotein that was recognized by VASP in blot overlays isidentical with zyxin, chicken blood cell proteins were separatedby 2-D gel electrophoresis followed by a [32P]VASP overlay. Asshown in Fig. 6, [32P]VASP labeled a protein extending over anapproximate pl range of 5.4 to 6.8, which is similar to thecalculated pl (23) and the experimentally determined pl

peptWd A 1 SGP VEAPSTGTGSAQPPSFTYAQQREK27chicen zyxmn 260 QFTAPSPSGPLSRPQPPNFTYAQQWER 286

T

pepte B 1 ,YAPRXSVXET EPTGTEEVT 241 1 1 1 P1 1 1 1

chicken zyxdn 467 KYAPRCSVCSEPIMPEPGKDETVR 490

FIG. 5. Alignment of the chicken zyxin sequence (23) to two p83peptide sequences obtained by endoproteinase Lys-C proteolysis ofporcine platelet p83 (single letter code for amino acids). X possiblyrepresents a cysteine (C) residue.

interval reported previously for chicken gizzard zyxin (16).Additional immunological data (not shown) also indicate thatthe chicken VASP-binding protein is indeed zyxin. The com-parison of the biochemical properties of chicken zyxin andmammalian p83 provides further evidence that they are re-lated. Both proteins behave similarly during ammonium sul-fate precipitation and bind to a hydroxylapatite matrix (16).Furthermore, zyxin is relatively insoluble at neutral pH (16),resembling partially purified p83, which (under these condi-tions) instantaneously precipitated when NaCl concentrationswere lowered by dialysis or dilution (not shown).

DISCUSSIONIn blots of total cellular proteins overlaid with 32p_phosphorylated VASP, a 83-kDa protein was labeled. Thisinteraction is specific and saturable, as it was quenched by anexcess of unlabeled VASP. [32P]VASP binding appears to bea very sensitive method to detect p83, since a rough estimate

.............................................. . ............kDa

...

66.3_57.6-42.8-39.2-

5.425.8 -

; ~~~~~~~~6.87,2

FIG. 6. [32P]VASP overlay of a chicken blood cell homogenateseparated by 2-D gel electrophoresis (autoradiogram): first dimension,isoelectric focusing on Immobiline dry strips (pH 3-10.5; from left toright); second dimension, SDS/PAGE. The positions of the isoelectricfocusing markers carbonic anhydrase II (pI 5.4) and myoglobin (pI 6.8and 7.2) are indicated.

Cell Biology: Reinhard et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

1, 2

020

7960 Cell Biology: Reinhard et al.

derived from dilution experiments with purified p83 demon-strated that about 30 ng are readily detectable in [32P]VASPoverlays (not shown).P83 purified here from porcine platelets was characterized

as a zyxin-related protein, based on biochemical and micro-sequencing data. In addition, the subcellular distribution ofp83, as revealed by indirect immunofluorescence, closelymatches that of chicken zyxin (15, 16, 24)-i.e., both proteinsare predominarktly associated with focal adhesions and stressfibers in a characteristic periodic pattern. Taken together,these data suggest that p83 is most likely the mammalianequivalent of chicken zyxin. However, the alignment of pep-tides A and B from p83 to the chicken zyxin sequence indicatesa lower degree of evolutionary conservation when comparedto other focal adhesion proteins, such as vinculin and non-muscle a-actinin from man and chicken (overall identities,95-96%). Both the AS83-1 antiserum raised against porcineplatelet p83 (our present data) and the original rabbit serumrecognizing chicken zyxin (17) showed limited interspeciescrossreactivity, pointing to obvious species differences be-tween both proteins.Two more zyxin-binding proteins have been detected pre-

viously. Zyxin binds to the N-terminal 27-kDa domain of thefocal adhesion- and microfilament-associated actin-bindingprotein a-actinin (24) and to the 23-kDa LIM-domain proteincCRP (chicken cysteine-rich protein) (23). Schmeichel andBeckerle (26) demonstrated that most of the CRP-bindingactivity resides in the first of three zyxin LIM-domains (amino&cids 349-406), whereas the a-actinin binding site is located inthe zyxin N-terminal proline-rich domain (amino acids 1-348).Recent data suggested that this proline-rich domain may alsobe involved in the zyxin-VASP interaction: thus, VASP hasbeen shown to bind to the Listeria monocytogenes surfaceprotein ActA (27). This interaction involves the 4-fold re-peated ActA proline-rich motifs (Asp or Glu)-Phe-Pro-Pro-(Pro or Ile)-Pro-Thr-(Asp or Glu)-(Glu or Asp)-Glu-Leu (28),which resemble several sequences located in the zyxin N-terminal domain between residues 81 and 134 (23). Moreover,microinjection of a corresponding ActA peptide caused adiffuse cytosolic VASP distribution and complete depletion ofVASP from lamellipodia and focal adhesions (28), where itnormally colocalizes with p83 or zyxin, respectively.VASP also interacts with the actin regulatory protein pro-

filin (12). Both VASP-binding proteins, listerial ActA (29, 30)and host-cell profilin (31), are involved in the actin-basedmotility of the intracellular bacterial parasite L. monocyto-genes. Considering the function of profilin in actin polymer-ization (32) and the direct interaction with VASP, VASP mightlink profilin to other proteins, like ActA and p83/zyxin andthus confine profilin-regulated F-actin formation to definedsubcellular sites. This may be especially relevant at the leadingedge of migrating cells, where VASP, p83, and profilin (33), aswell as highly dynamic actin polymerization are found (2,34-37). Similarly, VASP and p83/zyxin localization to focaladhesions and stress fibers may contribute to the actin turnoverobserved at these sites (37).

In the near future, further molecular characterization of p83and VASP should allow definition of the interaction betweenthese proteins in more detail and investigation of their possibleinvolvement in spatially confined actin polymerization.

The authors thank K. Abel (Wurzburg) for her help and adviceconcerning 2-D gel electrophoresis, Dr. T. Jarchau (Wurzburg) forhelp with the evaluation of results of data bank searches, Dr. B. M.Jockusch (Braunschweig, F.R.G.) for the collaboration in raisingpolyclonal mouse anti-VASP sera, and Dr. J. Wehland (Braunschweig)

as well as our colleagues of the Klinische Forschergruppe (Wurzburg)for a critical reading of the manuscript. M.R. is grateful to M. Kuhnfor excellent technical assistance. This work was supported by theDeutsche Forschungsgemeinschaft (SFB 176, A21).

1. Burridge, K., Fath, K., Kelly, T., Nuckols, G. & Turner, C. (1988)Annu. Rev. Cell Biol. 4, 487-525.

2. Luna, E. J. & Hitt, A. L. (1992) Science 258, 955-963.3. Gumbiner, B. M. (1993) Neuron 11, 551-564.4. Sastry, S. K. & Horwitz, A. F. (1993) Curr. Opin. Cell Biol. 5,

819-831.5. Shattil, S. J., Ginsberg, M. H. & Brugge, J. S. (1994) Curr. Opin.

Cell Biol. 6, 695-704.6. Halbrugge, M. & Walter, U. (1989) Eur. J. Biochem. 185, 41-50.7. Halbruigge, M., Friedrich, C., Eigenthaler, M., Schanzenbaicher,

P. & Walter, U. (1990) J. Biol. Chem. 265, 3088-3093.8. Butt, E., Abel, K., Krieger, M., Palm, D., Hoppe, V., Hoppe, J.

& Walter, U. (1994) J. Biol. Chem. 269, 14509-14517.9. Halbrugge, M. & Walter, U. (1993) in Protein Kinases in Blood

Cell Function, eds. Huang, C.-K. & Shaafi, R. I. (CRC, BocaRaton, FL), pp. 245-298.

10. Reinhard, M., Halbrugge, M., Scheer, U., Wiegand, C., Jockusch,B. M. & Walter, U. (1992) EMBO J. 11, 2063-2070.

11. Haffner, C., Jarchau, T., Reinhard, M., Hoppe, J., Lohmann,S. M. & Walter, U. (1995) EMBO J. 14, 19-27.

12. Reinhard, M., Giehl, K., Abel, K., Haffner, C., Jarchau, T.,Hoppe, V., Jockusch, B. M. & Walter, U. (1995) EMBO J. 14,1583-1589.

13. Machesky, L. M. & Pollard, T. D. (1993) Trends Cell Biol. 3,381-385.

14. Sohn, R. H. & Goldschmidt-Clermont, P. J. (1994) BioEssays 16,465-472.

15. Beckerle, M. C. (1986) J. Cell Bio. 103, 1679-1687.16. Crawford, A. W. & Beckerle, M. C. (1991) J. Biol. Chem. 266,

5847-5853.17. Crawford, A. W. & Beckerle, M. C. (1993) in Guidebook to the

Cytoskeletal and Motor Proteins, eds. Kreis, T. & Vale, R. (OxfordUniv. Press, Oxford), pp. 247-248.

18. Smith, D. E. & Fisher, P. A. (1984) J. Cell Biol. 99, 20-28.19. Devereux, J., Haeberli, P. & Smithies, 0. (1984) Nucleic Acids

Res. 12, 387-395.20. Lipman, D. J. & Pearson, W. R. (1985) Science 227, 1435-1441.21. Wencel-Drake, J. D., Okita, J. R., Annis, D. S. & Kunicki, T. J.

(1991) Arterioscler. Thromb. 11, 882-891.22. Fox, J. E. B., Taylor, R. G., Taffarel, M., Boyles, J. K. & Goll,

D. E. (1993) J. Cell Biol. 120, 1501-1507.23. Sadler, I., Crawford, A. W., Michelsen, J. W. & Beckerle, M. C.

(1992) J. Cell Biol. 119, 1573-1587.24. Crawford, A. W., Michelsen, J. W. & Beckerle, M. C. (1992) J.

Cell Biol. 116, 1381-1393.25. Hensler, M. E. & Beckerle, M. C. (1992) Mol. Biol. Cell 3, 266a

(abstr.).26. Schmeichel, K. L. & Beckerle, M. C. (1994) Cell 79, 211-219.27. Chakraborty, T., Ebel, F., Domann, E., Niebuhr, K., Gerstel, B.,

Pistor, S., Temm-Grove, C. J., Jockusch, B. M., Reinhard, M.,Walter, U. & Wehland, J. (1995) EMBO J. 14, 1314-1321.

28. Pistor, S., Chakraborty, T., Walter, U. & Wehland, J. (1995) Curr.Biol. 5, 517-525.

29. Domann, E., Wehland, J., Rohde, M., Pistor, S., Hartl, M.,Goebel, W., Leimeister-Wachter, M., Wuenscher, M. &Chakraborty, T. (1992) EMBO J. 11, 1981-1990.

30. Kocks, C., Gouin, E., Tabouret, M., Berche, P., Ohayon, H. &Cossart, P. (1992) Cell 68, 521-531.

31. Theriot, J. A., Rosenblatt, J., Portnoy, D. A., Goldschmidt-Cler-mont, P. J. & Mitchison, T. J. (1994) Cell 76, 505-517.

32. Theriot, J. A. & Mitchison, T. J. (1993) Cell 75, 835-838.33. Buss, F., Temm-Grove, C., Henning, S. & Jockusch, B. M. (1992)

Cell Motil. Cytoskel. 22, 51-61.34. Stossel, T. P. (1993) Science 260, 1086-1094.35. Condeelis, J. (1993) Annu. Rev. Cell Biol. 9, 411-444.36. Cramer, L. P., Mitchison, T. J. & Theriot, J. A. (1994) Curr. Opin.

Cell Biol. 6, 82-86.37. Glacy, S. D. (1983) J. Cell Biol. 97, 1207-1213.

Proc. Natl. Acad. Sci. USA 92 (1995)

Dow

nloa

ded

by g

uest

on

Mar

ch 2

1, 2

020