Essential role for Abi1 in embryonic survival andWAVE2 complex integrityPatrycja M. Dubieleckaa,1, Kathrin I. Ladweinb,1, Xiaoling Xionga, Isabelle Migeottec, Anna Chorzalskaa,Kathryn V. Andersonc, Janet A. Sawickid, Klemens Rottnerb,e, Theresia E. Stradalb,f, and Leszek Kotulaa,2

aLaboratory of Cell Signaling, New York Blood Center, New York, NY 10065; bHelmholtz Centre for Infection Research, D-38124 Braunschweig, Germany;cDevelopmental Biology Program, Sloan-Kettering Institute, New York, NY 10065; dLankenau Institute for Medical Research, Wynnewood, PA 19096;eInstitute of Genetics, University of Bonn, 53117 Bonn, Germany; and fInstitute for Molecular Cell Biology, University of Münster, D-48149 Münster, Germany

Edited* by Hilary Koprowski, Thomas Jefferson University, Philadelphia, PA, and approved March 15, 2011 (received for review November 11, 2010)

Abl interactor 1 (Abi1) plays a critical function in actin cytoskeletondynamics through participation in the WAVE2 complex. To gaina better understanding of the specific role of Abi1, we generateda conditional Abi1-KOmouse model and MEFs lacking Abi1 expres-sion. Abi1-KO cells displayed defective regulation of the actin cyto-skeleton, and this dysregulation was ascribed to altered activity ofthe WAVE2 complex. Changes in motility of Abi1-KO cells weremanifested by a decreased migration rate and distance but in-creased directional persistence. Although these phenotypes didnot correlate with peripheral ruffling, which was unaffected,Abi1-KO cells exhibited decreased dorsal ruffling. Western blottinganalysis of Abi1-KO cell lysates indicated reduced levels of theWAVE complex components WAVE1 and WAVE2, Nap1, and Sra-1/PIR121. Although relative Abi2 levels were more than doubled inAbi1-KO cells, the absolute Abi2 expression in these cells amountedonly to a fifth of Abi1 levels in the control cell line. This findingsuggests that the presence of Abi1 is critical for the integrity andstability of WAVE complex and that Abi2 levels are not sufficientlyincreased to compensate fully for the loss of Abi1 in KO cells and torestore the integrity and function of theWAVE complex. The essen-tial function of Abi1 inWAVE complexes and their regulationmightexplain the observed embryonic lethality of Abi1-deficient em-bryos, which survived until approximately embryonic day 11.5and displayed malformations in the developing heart and brain.Cells lacking Abi1 and the conditional Abi1-KO mouse will serveas critical models for defining Abi1 function.

cell motility | lamellipodium | Rac | Arp2/3-complex | Hssh3bp1

Essential physiological processes such as cell migration, adhe-sion, and endocytosis are dependent on the coordinated re-

modeling of the actin cytoskeleton, but the complexity of themechanisms at play is only beginning to be appreciated (1, 2).Nucleation of actin filaments is catalyzed by different types ofprotein complexes, the most prominent of which include the Arp2/3 complex and members of the formin family of proteins (3). TheArp2/3 complex is regulated by additional proteins called “nucle-ation-promoting factors” (NPFs). The most prominent NPFs in-clude theWASP andWAVE subfamilies (4), not only because theyconstitute key signaling nodes that connect activation signals fromthe Rho family of small GTPases with the actin polymerizationmachineries but also because the N-WASP and Arp2/3 complexeshave emerged as frequent targets for viral or bacterial pathogensthat usurp the host actin cytoskeleton for their needs (5).Abl interactor 1 (Abi1) is one of the major regulators of actin

cytoskeleton reorganization through participation in several mul-tiprotein complexes that either associate with actin filamentsthrough spectrin (6) or Eps8 (7) or that regulate the dynamics ofactin polymerization more indirectly, for instance through regu-lation of small GTPase activities of the Rho family (4, 8). RhoGTPases are modulated by complexes of Abi1 with Eps8–Sos1 (9,10) and downstream actin polymerization e.g. by N-WASP (11)and WAVE complex (12–14). The ubiquitous WAVE complex,comprising the direct Rac interactor Sra-1/PIR121, its bindingpartner Nap1, plus Abi1, HSPC300/Brick1, and WAVE2, is con-sidered to be the essential functional output module linking Rac1

activation to actin polymerization that drives lamellipodia pro-trusion at the plasma membrane (13, 14). WAVE1 and WAVE3were proposed to have roles analogous toWAVE2 in the complex(8). Although the function of WAVE3 is less defined, the criticalrole of the other two WAVEs in dorsal ruffle formation wasdemonstrated (15, 16). Mechanistically, it is proposed that nativeWAVE2 complex is inactive (17–19) but is activated at themembrane by simultaneous interaction with prenylated Rac-GTPand acidic phospholipids as well as by serine/threonine phos-phorylation (20). Rac1 does not interact directly with WAVE butinstead binds to the Nap1-binding protein Sra-1 (12) and to Abi1(21). The Abi1-mediated tyrosine phosphorylation of WAVE2 byAbl regulates WAVE2 complex activity either by regulatingconformation of the complex, as suggested (22), or by regulatinginteractions among the complex components or with other cel-lular targets (10, 23, 24). The interaction of Abi1 with p85 (24)might modulate PI3 kinase activity (10); this activity would beconsistent with coincident modulation of WAVE complex activityby its interaction with PIP3, or PIP3-mediated membrane re-cruitment of WAVE2 (25).The Xenopus laevis homolog of Abi1, xlan4, was demonstrated

to be critical for CNS development, and Abi2 might be similarlyrequired in mice (26). In mice Abi2 has regulatory functions incell–cell adhesion, cell migration, and tissue morphogenesis (27).Abi2-KO mice exhibit a severe memory loss thought to be causedby deficiencies in dendritic spine and adherens junction formationand abnormal cell–cell communication (27). A role for Abi inregulation of the cytoskeleton and in cytokinesis has been con-firmed in Dictyostelium (28). Dictyostelium has only one Abi gene;thus it offers no specific conclusion for roles of mammalian Abi1or Abi2. Moreover, numerous studies have concluded that Abiproteins have overlapping functions in actin cytoskeleton regu-lation (29, 30). Studies using conventional Abi1 (31) and WAVE2(16) mutant alleles suggested a critical role of Abi1 in embryonicdevelopment, thus suggesting that a conditional allele wouldprovide more definitive information about Abi1, including itsfunction in adult organs.Here we present the initial characterization of MEFs isolated

from a conditional Abi1-KO mouse. Biochemical analyses ofAbi1-KO cells indicated that the presence of Abi1 is critical forWAVE2 complex integrity. As a physiological readout of Abi1gene disruption, we observed impairment of local actin poly-merization processes leading to decreased dorsal ruffle formationand cell migration. We also found that Abi1 is essential for em-bryonic development, because Abi1-deficient embryos survivedonly until embryonic day 11.5 (E11.5) and displayed malforma-tions in the developing heart and brain.

Author contributions: P.M.D., K.V.A., J.A.S., K.R., T.E.S., and L.K. designed research; P.M.D.,K.I.L., X.X., I.M., A.C., and L.K. performed research; P.M.D., K.R., T.E.S., and L.K. contributednew reagents/analytic tools; P.M.D., K.I.L., X.X., I.M., K.V.A., J.A.S., K.R., T.E.S., and L.K.analyzed data; and P.M.D., K.R., T.E.S., and L.K. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1P.M.D. and K.I.L. contributed equally to this work.2To whom correspondence should be addressed. E-mail: lkotula@gmail.com.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1016811108/-/DCSupplemental.

7022–7027 | PNAS | April 26, 2011 | vol. 108 | no. 17 www.pnas.org/cgi/doi/10.1073/pnas.1016811108

Dow

nloa

ded

by g

uest

on

Dec

embe

r 7,

202

0

ResultsGeneration of Abi1 Floxed Strain. To generate the Abi1 conditionalknockout mouse, we prepared a targeting vector that would re-sult in conditional deletion of Abi1 exon 1, which is consistentwith the alternatively spliced Abi1 mRNA (6) (Fig. 1A). Designof the targeting vector and subsequent steps of cloning of thetransgenic animal, including F1 transmission of the transgeneand production of the homozygous Abi1 floxed (fl/fl) strain (Fig.1 B and C), are described in SI Materials and Methods.

Absence of Abi1 Causes Embryonic Lethality at Day E11.5 ofDevelopment. It has been shown that Nap1 is essential for a se-ries of actin-mediated cell-migration events during early mouseembryogenesis (32). These events include migration of the ante-rior visceral endoderm, required for specification of the anterior–posterior body axis of the animal, and migration of the mesodermand endoderm germ layers (32). As aWAVE complex component,Abi1 is involved in the regulation of cell migration; therefore it wasexpected that Abi1 also might regulate morphogenetic eventsduring early mouse embryogenesis. By excision of the floxed exon 1from the germ line, a null allele of the Abi1 gene was obtained.Using animals carrying this null Abi1 allele (SI Materials andMethods), timed matings were carried out, and the phenotypes ofthe homozygous null embryos from day E8.5 to E11.5 were ana-lyzed.UnlikeNap1mutations, which cause developmental arrest atday E9.5, Abi1-null embryos displayed lethality between E10.5 andE11.5 (Fig. 1D). Gross examination of E8.0–E11.0 embryosrevealed marked developmental delay with prominent neural tubeand cardiac abnormalities similar to those observed for N-Wasp–null embryos (33). As early as E8.5, Abi-null embryos exhibiteddevelopmental delay (smaller size, some unturned embryos). Un-dulation of the neural tube was observed, and in some embryos theneural tube was open. Cardiac tissue was clearly apparent in Abi1-null embryos, but at late stages the heart was markedly dilated(cardiac edema). Abi1-null embryos developed somites and limbbuds; however, severe abnormalities were observed in the forma-tion of branchial arches. Although the specific developmentalabnormalities leading to death of the Abi1-null embryos remainuncertain, defective cardiac function and/or abnormalities in bran-chial arch formation might be the causes. Nonetheless, some Abi1-null embryos were observed to have a beating heart up to E10.5. NoAbi1-null embryos survived beyond E11.5, although lack of Abi1seemed not to affect gastrulation. The most marked defect was thedevelopmental delay of Abi1-null embryos and the pronouncedneural and cardiac defects.

Instability of the WAVE2 Complex in Abi1-KO MEFs. After isolationand PCR screening of MEF cell clones, Southern blotting con-firmed Cre-mediated deletion of exon 1, which resulted in the lackof Abi1 expression (Fig. 2 and Figs. S2 and S3). Using Abi1-KOcells, we investigated how loss of Abi1 affects the stability of otherWAVE2 complex subunits. Western blot analysis of total cellextracts demonstrated that WAVE2, Sra-1, and Nap1 proteinlevels were reduced significantly, although not abolished, in Abi1-KO cells as compared with the Abi1-expressing parental cell line(Fig. 3). A twofold increase in Abi2 protein levels in Abi1-KO vs.control cells also was observed. An anti-WAVE1 antibody mir-rored the significant down-regulation observed for WAVE2, butthe opposite was observed for WAVE3 (Fig. 3). However, itshould be noted that a pan-WAVE antibody capable of recog-nizing allWAVE proteins still showed reducedWAVE1/2/3 levelsin Abi1-KO clones compared with control (Fig. 3 Right and Fig.S4A). This finding is consistent with the observed phenotypes andmay indicate that increased WAVE3 levels cannot compensatefor the reduction of WAVE1 and WAVE2, probably because of

Fig. 1. Generation of conditional Abi1-KO mouse. (A) Design of the tar-geting vector and subsequent steps of recombinant Abi1 allele modification.Abi1 allele, wild-type Abi1 allele; targeted allele, the targeting vector con-taining exon 1 with the short homology arm (SA) extending 3′ and the longhomology arm (LA) on the 5′ side of exon 1. Exon 1 is flanked by two loxPsites, and the neomycin gene cassette is inserted 3′ to exon 1 and internal tothe distal loxP site. The neomycin gene cassette is bound by two frt sites andincludes an additional loxP site as indicated (SI Materials and Methods). Thetarget region is 0.9 kb and includes exon 1. Floxed allele, the recombinantAbi1 allele following frt-mediated removal of the neomycin cassette; de-leted allele, the recombinant Abi1 allele following Cre-mediated removal ofexon 1. (B) PCR analysis of genotypes of parental and F1 mice. HeterozygousAbi1 mice carrying the targeted allele were bred with the frt deleter strain(SI Materials and Methods). Animals were genotyped with primers LAN1 andA2 for the neomycin gene cassette (Neo+/frt+/loxP+), with primers mAbi1-loxP35′ and Flankneo13′ (SI Materials and Methods and Fig. S1) for the wild-type allele (Upper panel, lower band) or with the floxed Abi1 allele, whichlacks the neomycin gene and the 3′ frt site (Neo−/frt−/loxP+) (Upper panel,upper band). Under the PCR conditions used, no amplification of the neo-mycin-positive allele (Neo+/frt+/loxP+) was observed with primers mAbi1-loxP35′ and Flankneo13′. Animals 1 and 3 are heterozygous mice positive forthe floxed allele lacking neomycin and the wild-type allele (Upper); and theylack the neomycin gene (Lower). Animal 2 contains only wild-type alleles(Lower). (C) Southern blot analysis of Abi1 transgenic mice. Southern blot-ting was performed as described (SI Materials and Methods). Mouse genomicDNA was digested with AflII enzyme and hybridized with a 543-bp PB1/2probe. Expected sizes were 4.15 kb for the wild-type (wt); 5.49 kb for thetargeted allele (loxP+/neo+), (indicated by stars); and 3.65 kb for the floxed

allele with the neomycin cassette deleted (loxP+/neo−). (D) Phenotype ofAbi1-null embryos. Phenotypic analyses show exencephaly and pericardialedema in Abi-null embryos. The table summarizes observations of embryosat days E9.5, E10.5, and E11. Embryo death occurred between E10.5–E11.5.

Dubielecka et al. PNAS | April 26, 2011 | vol. 108 | no. 17 | 7023

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 7,

202

0

the relatively low abundance of the predominantly neuronal-specific WAVE3 (34) in these cells.

Increased Levels of Abi2 Expression in an Abi1-KO Cell Line ReachedOnly 22% of the Endogenous Abi1 Level in the Control Floxed CellLine. To determine the absolute concentration of Abi1 proteins incell lysates, predetermined amounts of purified recombinant Abi1or Abi2 protein standards were added into MEF cell lysate sam-

ples from Abi1-KO and Abi1 floxed (fl/fl) control cell lines. Thecomparative evaluation of the control recombinant Abi1 andAbi2(Fig. S5) amounts vs. their endogenous levels, as detected byrespective antibodies, allowed determination of the absoluteamount of endogenous Abi1 or Abi2 proteins. Standard curveswere generated by plotting the band intensities against the con-centrations of control recombinant Abi GST-tagged proteins(∼80 kDa) (Fig. S6). Based on the standard curves, we calculatedthat, although the relative intensity of Abi2 was about two timeshigher in Abi1-KO cells than in control cells, the difference in theabsolute amount of endogenous Abi2 vs. Abi1 in control cells wasless than 10%: 0.914 ± 0.096 ng Abi2 vs. 11.88 ± 0.55 ng Abi1 permicrogram of total protein in the sample. This evaluation in-dicated that despite enhancement of Abi2 expression in Abi1-KOcells, its level did not exceed 22% of the endogenous Abi1 incontrol cells. Thus, it seems likely that the failure to up-regulateAbi2 sufficiently is one of the reasons Abi2 does not reconstitutethe WAVE complex in Abi1-KO cell lines.

Defect in Formation of PDGF-Induced Dorsal Ruffles in Abi1-KO MEFCells. Abi1, as the critical component of WAVE2 complex, isknown to be involved in actin reorganization resulting inlamellipodia and in actin-rich peripheral and dorsal ruffle for-mation upon growth factor stimulation (9, 30, 35). Therefore, weasked whether any defects in these structures were observed incells lacking expression of a functional Abi1 gene. We found thatupon PDGF treatment, both Abi1 control and Abi1-deficientcells displayed peripheral as well as dorsal ruffle formation, asassessed by Alexa Fluor 594–phalloidin staining, and WAVE2was localized to these structures, as expected, although it was lessprominent on average in Abi1-deficient clones (Fig. 4A). Abi1was highly enriched in dorsal ruffles and localized at the tips ofruffles and in the cell periphery in Abi1 control cells but couldnot be detected in the Abi1-KO cell lines (Fig. 4A). Evaluationof the number and topology of dorsal ruffles revealed an ∼50%decrease in the formation (Fig. 4 B and C) and lower apparentprominence of these structures in Abi1-KO cells (Fig. 4D). Thisobservation indicates an important role of Abi1 in the formationof dorsal ruffles. However, despite the reported critical role ofthe WAVE2 complex in peripheral ruffle formation (15, 36)and the observed decrease in WAVE2 complex levels, PDGF-induced peripheral ruffling was not affected in Abi1-KO cells(Fig. 4 B and C). Moreover, neither the expression levels of Eps8and Sos1 nor total Rac activation levels in response to PDGFcorrelated with the loss of Abi1, indicating that defective Racactivation downstream of the Eps8/Abi1/Sos1 complex does notcause the observed dorsal ruffling phenotype (Fig. S4 B and C).

Cells Lacking Abi1 Exhibit Cell-Motility Defects. Abi1 has been im-plicated as a regulator of actin cytoskeleton-dependent cell mo-tility, mainly as an integral component of the WAVE2 complex.Therefore, we examined Abi1-KO MEF cell lines in a series ofmotility assays (Fig. 5). Random cell motility was recorded byvideomicroscopy and was analyzed using the cell-tracking tool (SIMaterials and Methods). Control cells migrated an average of390 μm, whereas KO cells migrated an average of 260 μm. The

Fig. 3. Expression levels of WAVE proteins, Sra-1, and Nap1 in Abi1-KO cells. Western blotanalysis of MEF cell lysates of Abi1-KO control (fl/fl) and Abi1-KO cell lines. WAVE1, -2, -3, Nap1,Sra-1, and Abi2, protein levels were evaluatedwith specific antibodies in total cellular lysatesobtained from two clones [c1 (#3-6) and c2 (#3-11)] lacking Abi1 expression and from the pa-rental line MEF #3. A significant decrease inprotein levels of the WAVE-complex compo-nents Sra-1 and Nap1,WAVE1, and totalWAVE1/2/3 and a significant (twofold) increase inWAVE3 and Abi2 levels were observed in KOcells vs. control. Levels of protein expression ± SEM were evaluated based on band intensities from several independent experiments (Center). Anti-Rac1antibody (Rac1) was used as loading control. Abi1 was detected with monoclonal antibody 1B9 (SI Materials and Methods).

Fig. 2. Characterization of Abi1-KO MEF cell lines. (A) Cre-recombinase–mediated exon 1 deletion of Abi1 in MEF cell lines. DNA from MEF cells wasgenotyped with primers DL75′ and Neogene13′ (SI Materials and Methodsand Fig. S1). Genotyping of MEF cell lines based on the PCR result is indicatedbelow the panel. Floxed, Abi1-floxed allele; deleted, exon 1-deleted allele;Abi1 (fl/fl), genotype homozygous for the floxed allele; Abi1 (del/del),genotype homozygous for the deleted allele; Abi1 (fl/del), heterozygousgenotype. Note that in cell line #8 clone d3 has both alleles, indicating thatin this cell line Cre-recombinase failed to recombine on one allele; hence theheterozygous genotype, Abi1 (fl/del), is observed. Primer sequences are listedin SI Materials and Methods. (B) Southern blot analysis of MEF cell genomicDNA. MEF cell genomic DNA was digested with AflII enzyme and hybridizedwith the 543-bp PB1/2 probe. Expected sizes were 4.15 kb for wild type; 3.65kb for the floxed allele (neomycin cassette deleted; loxP+/neo−); and 4.84 kbfor the deleted allele [exon 1 and neomycin cassette deleted; loxP−/neo−

(Ex1del)]. (C) Cre-recombinase–mediated loss of Abi1 expression in MEF#3 cell line subclones. Western blot analysis of Abi1 expression in parentalMEF #3 Abi1 (fl/fl) and in exon 1-deleted Abi1 (del/del) MEF cell lines. Celllysates of the indicated cell lines were blotted with antibody 7B6 (23).Clones #3-1 through #3-11 are subclones of the parental MEF#3 obtainedfollowing transient Cre-recombinase expression. Abi1(+/+) MEF WT, mouseembryonic cells expressing the wild-type Abi1 gene. Protein-stained gel ofthe samples used for Western blot was used as loading control.

7024 | www.pnas.org/cgi/doi/10.1073/pnas.1016811108 Dubielecka et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 7,

202

0

migration rate of control cells was 0.54 μm/min, compared with0.36 μm/min for KO cells. However, Abi1-KO cells exhibited in-creased directional persistence. The net distance vs. true distanceratio was 0.25 for control cells vs. 0.37 for KO cells (arithmeticmeans, P < 0.0001 for all comparisons). Thus, migration rate anddistance were reduced, but directional persistence was increasedin Abi1-KOMEF cells compared with control cells. In the wound-healing assay, a statistically significant impairment of woundclosure by cells lacking Abi1 was observed: 3 μm2/min by controlcells vs. 2.39 μm2/min by KO cells (Fig. 6 and Movies S1, S2, andS3). Taken together, these results indicate that Abi1 removalcaused reduced wound-closure rates, probably because of theimpairment in cell migration.

DiscussionHere we report the production and initial phenotypic character-ization of MEF cells lacking a functional Abi1 gene. Among otherstrategies used to generate Abi1-KO animals and cell lines, tar-

geting of exon 1 has proven successful. Following Cre-mediatedrecombination, MEF cells lacking Abi1 mRNA and protein ex-pression were obtained. In this study, we focused on the charac-terization ofWAVE2 complex-dependent processes, knowing thatparticipation in the complex is a critical aspect of Abi1 function.Abi1-KO MEF cells are viable and exhibit no apparent loss

of ability to form lamellipodia, stress fibers, or peripheral ruf-fles. However, Abi1-KO cells displayed decreased efficiency ofPDGF-induced dorsal ruffling formation. In addition, dorsalruffles formed in Abi1-KO cells were much less prominent anddisplayed less intense F-actin staining compared with control celllines. Because the levels of the Eps8/Sos1 complex did not cor-relate with defective dorsal ruffling, we assume that this pheno-type is a consequence of the significant instability of WAVE2complex in the absence of Abi1. The levels of WAVE2 complexcomponents WAVE2, Sra-1/PIR121, and Nap1 were significantlylower in cells lacking Abi1. These data are consistent with pre-vious siRNA knockdown experiments showing that upon down-

Fig. 4. Cell morphologies of Abi1-KO cell lines. (A and B) Localization of Abi1 and WAVE2 in control and Abi1-deficient MEF cells after PDGF stimulation.Control Abi1 (fl/fl) and Abi1 (del/del) cell lines were grown on glass coverslips, serum-starved, and treated with PDGF. Cells then were immunostained withantibodies to Abi1 [anti-Abi1 (4E2)], WAVE2 (#1735), and phalloidin as described in SI Materials and Methods. Abi1 andWAVE2 localized to circular dorsal andperipheral ruffles of precursor cells butwere absent (Abi1) orweaker (WAVE2) in these structures in KO cells. (C) Quantification of dorsal and peripheral ruffling.(Left) MEF control (fl/fl), and Abi1-KO cells (del/del) were plated on glass coverslips and serum-starved overnight. Cells were stimulated with PDGF, fixed, andstained with Alexa Fluor 594-conjugated phalloidin to detect F-actin. (Right) The percentages of cells with different cell morphologies in response to PDGFtreatmentwere quantified in the indicatedMEF cell lines. Note the reduction of cells displaying dorsal (Upper) but not peripheral (Lower) ruffles in independentAbi1-null clones (c1, c2, d1, d2) compared with their respective control Abi1 (fl/fl) cells (#3 or #8), as indicated. Categories of cell morphologies for quantificationwere as follows: with ruffles, without ruffles, or with ambiguous morphology. At least 100 cells were analyzed and categorized for each condition. Columnsshow the percentage of cells of the respective morphology displayed as arithmetic means ± SEM of at least three independent experiments. (D) F-actin stainingin PDGF-induced dorsal ruffles in control Abi1 (fl/fl) cells and in Abi1-KO MEF cells [Abi1 (del/del)]. F-actin staining of ruffles revealed different topology andapparent reduction of F-actin in dorsal ruffles in Abi1-KO cells vs. the control cell line. Images were taken from two focal planes, A (Left) and B (Right).

Dubielecka et al. PNAS | April 26, 2011 | vol. 108 | no. 17 | 7025

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 7,

202

0

regulation of individual subunits, the remaining components ofthe WAVE2 complex are coordinately decreased (11, 37–39).Our results demonstrate that lack of Abi1 caused significant in-stability of the WAVE2 complex, thus confirming the critical roleof Abi1 in the complex function.In addition to WAVE2, we also observed a decrease in

WAVE1 levels, suggesting that Abi1 might be incorporated intocomplexes with WAVE1, as shown previously (20), and not only

with WAVE2. Although WAVE1 originally had been ascribeda role in dorsal but not peripheral ruffle formation (15), thisconclusion was contradicted more recently, because WAVE1 KOcells were observed to form multiple dorsal ruffles (36). Likewise,both WAVE1 and WAVE2 previously had been observed to ac-cumulate at the tips of lamellipodia and peripheral ruffles (14,40), thus indicating that potential functional differences betweenWAVE1 and WAVE2 are not caused by their differential func-tions in the formation of peripheral versus dorsal ruffles. BecauseAbi1-null fibroblasts have reduced expression of both WAVE2and WAVE1, we speculate that the observed defect in dorsal butnot peripheral ruffling simply reflects a higher sensitivity of dorsalruffles to abrogation of WAVE complex activity rather than dif-ferential subcellular functions of WAVE1 versus WAVE2. Thishypothesis is consistent with lower levels of total WAVE1/2/3proteins in Abi1-null fibroblast clones compared with controlcells. Consequently, lower WAVE levels cause lower output intoArp2/3 complex activity and F-actin polymerization, as indeedwas observed (Fig. 4D). Other consequences of reduced WAVEcomplex function and Abi1 deficiency are decreased random anddirected cellular motility.Although Abi2 expression was more than doubled in cells

lacking Abi1, this response did not compensate for the loss ofAbi1 in the apparent loss of integrity of WAVE2 complex, nordid it compensate for loss of Abi1 function pertaining to dorsalruffling and cellular motility, probably because the amount ofAbi2 is too low (demonstrated here to be less than 10% of Abi1in Abi1-expressing control cells). Nevertheless, the increasedamounts of Abi2 and concomitant increase of WAVE3 maywell contribute to the relative modesty of phenotypes observed inAbi1-deficient cells.Abi1 or Abi2 can be incorporated interchangeably into WAVE2

complex. For example, in leukocytes, Abi1 was found in theWAVE2 complex and Abi2 in WAVE1 complex, whereas bothAbi1 and Abi2 were present in WAVE2 complex in A431 cells(20). The basis for specificity of Abi protein incorporation intoWAVE1 or WAVE2 complex is not known, although, based onhigh sequence conservation of binding sites, the affinities of theAbi proteins for WAVE proteins are expected to be similar (20,22, 31). One explanation might be that Abi1 and Abi2 exhibitdifferent patterns of threonine and serine phosphorylation (20),which is required for full activation of WAVE2 complex by Racand acidic phospholipids (20). Nonetheless, concurrent down-regulation of WAVE1 and WAVE2 in Abi1-KO cells suggeststhat Abi1 incorporates into both WAVE2 and WAVE1 com-plexes or, alternatively, that reduced stability of subunits sharedby both WAVE1- and WAVE2-containing complexes, such asSra-1/PIR121 or Nap1, indirectly down-regulatesWAVE1 expres-sion.We cannot distinguish between these possibilities at present.Finally, the apparent dysregulation of WAVE3 expression alsowill have to be explored in the future.We propose that loss of WAVE2 complex function and dys-

regulation of WAVE1 and WAVE3 caused by the lack of Abi1might explain the early embryonic death of Abi1-null embryos.A nonoverlapping function of Abi proteins in embryonic de-velopment is indicated by the fact that Abi2 deficiency is not

Fig. 5. Impaired motility of MEFs lacking Abi1. The parameters of random cell motility, migration distance, rate, and directional persistence were evaluated inAbi1-null cell lines (c1 and c2) and Abi1 floxed (fl/fl) cells. Cells were seeded under the same conditions and monitored by time-lapse microscopy. Differentparameters of cell motility, as indicated above graphs (n = 40 movies per parameter/cell line) were evaluated using ImageJ. Directional persistence was de-termined as the ratio of total distance traveled over total length of the migration path. Data are presented as ± SEM, based on four independent experiments.

Fig. 6. Wound-healing migration assay. Cells were seeded under identicalconditions. After formation of a confluent monolayer, a wound was pro-duced by scratching with a fine pipette tip. Rate of wound closure was de-termined by measuring the area remaining uncovered by cells per unit oftime. Data represent means ± SEM of four independent assays of Abi1-KOcell lines (c1 and c2) and control (fl/fl) MEF cells.

7026 | www.pnas.org/cgi/doi/10.1073/pnas.1016811108 Dubielecka et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 7,

202

0

embryonic lethal (27), whereas, as we report here, death of Abi1-null embryos occurred around day E11.5.The absence of a gene required during developmental progres-

sion often is displayed phenotypically by histological abnormali-ties eventually leading to death at different stages of development.For example, inactivation of murine WAVE2 is lethal by E12.5(during mid-gestation) (16); inactivation of N-WASP is lethal atE11.0 (during organogenesis) (33, 41); inactivation of Cdc42 islethal at E3.5 (during implantation) (42); inactivation of Nap1is lethal at E9.0 (during gastrulation); and inactivation of Rac1 islethal at E7.5 (during gastrulation) (32, 43, 44). Abi2 mutantsare viable and without morphological defects (27). Inactivationmutants of two Sra-1 and HSPC300 genes have not yet been de-scribed. Abi1-null embryos developed normally through gastru-lation and survived to day E11.5 with marked neural and cardiacdefects. Thus, Abi1-signaling events are not required for gastru-lation but are critical during brain and heart development. Thecardiovascular phenotype is in agreement with observations froma conventional Abi1-KO model (45) that was reported while thismanuscript was under consideration.In summary, we show here that Abi1 loss causes impairments in

cellular motility that depend on dynamic actin rearrangementsascribed to WAVE2 activity in actin polymerization. We provide

a biochemical explanation for the observed phenotypes, i.e., thatloss of Abi1 results in the loss of WAVE2 complex integrityand stability, and that this loss cannot be compensated by theincreased, but ultimately insufficient, levels of Abi2. These me-chanisms might underlie the critical role of Abi1 in embryonicdevelopment. Our conditional Abi1-KO model represents animportant tool for addressing the physiological functions of Abi1.

Materials and MethodsDetails of methods are presented in SI Materials and Methods. Transgenic EScells and conditional Abi1-KO mice were generated by conventional meth-ods. MEF lines obtained from conditional Abi-KOmice were used to generateAbi1-null MEF lines by transient transfection with Cre-recombinase. All stepswere confirmed by PCR and/or Southern blot. Cell morphologies and markerlocations were assessed by confocal and/or epifluorescence microscopy.

ACKNOWLEDGMENTS. We thank Sandrine Etienne-Manneville (InstitutPasteur, Paris, France) for helpful suggestions, Kai Städing for the help withantibody generation, and Brigitte Denker for technical help. This work wassupported in part by Grant R01 NS044968 from the National Institutes ofHealth, Grant W81XWH-08-1-0320 from the US Department of Defense, theFM Kirby Foundation, Inc. (L.K.), and the Deutsche Forschungsgemeinschaft(T.E.S. and K.R.).

1. Niedergang F, Chavrier P (2005) Regulation of phagocytosis by Rho GTPases. Curr TopMicrobiol Immunol 291:43–60.

2. Pollard TD, Borisy GG (2003) Cellular motility driven by assembly and disassembly ofactin filaments. Cell 112:453–465.

3. Faix J, Grosse R (2006) Staying in shape with formins. Dev Cell 10:693–706.4. Stradal TE, Scita G (2006) Protein complexes regulating Arp2/3-mediated actin

assembly. Curr Opin Cell Biol 18:4–10.5. Rottner K, Stradal TE, Wehland J (2005) Bacteria-host-cell interactions at the plasma

membrane: Stories on actin cytoskeleton subversion. Dev Cell 9:3–17.6. Ziemnicka-Kotula D, et al. (1998) Identification of a candidate human spectrin Src

homology 3 domain-binding protein suggests a general mechanism of association oftyrosine kinases with the spectrin-based membrane skeleton. J Biol Chem 273:13681–13692.

7. Biesova Z, Piccoli C, Wong WT (1997) Isolation and characterization of e3B1, an eps8binding protein that regulates cell growth. Oncogene 14:233–241.

8. Stovold CF, Millard TH, Machesky LM (2005) Inclusion of Scar/WAVE3 in a similarcomplex to Scar/WAVE1 and 2. BMC Cell Biol 6:11.

9. Scita G, et al. (1999) EPS8 and E3B1 transduce signals from Ras to Rac. Nature 401:290–293.

10. Innocenti M, et al. (2003) Phosphoinositide 3-kinase activates Rac by entering ina complex with Eps8, Abi1, and Sos-1. J Cell Biol 160:17–23.

11. Innocenti M, et al. (2005) Abi1 regulates the activity of N-WASP and WAVE in distinctactin-based processes. Nat Cell Biol 7:969–976.

12. Gautreau A, et al. (2004) Purification and architecture of the ubiquitous Wavecomplex. Proc Natl Acad Sci USA 101:4379–4383.

13. Innocenti M, et al. (2004) Abi1 is essential for the formation and activation ofa WAVE2 signalling complex. Nat Cell Biol 6:319–327.

14. Steffen A, et al. (2004) Sra-1 and Nap1 link Rac to actin assembly driving lamellipodiaformation. EMBO J 23:749–759.

15. Suetsugu S, Yamazaki D, Kurisu S, Takenawa T (2003) Differential roles of WAVE1 andWAVE2 in dorsal and peripheral ruffle formation for fibroblast cell migration. DevCell 5:595–609.

16. Yan C, et al. (2003) WAVE2 deficiency reveals distinct roles in embryogenesis and Rac-mediated actin-based motility. EMBO J 22:3602–3612.

17. Ismail AM, Padrick SB, Chen B, Umetani J, Rosen MK (2009) The WAVE regulatorycomplex is inhibited. Nat Struct Mol Biol 16:561–563.

18. Derivery E, Lombard B, Loew D, Gautreau A (2009) The Wave complex is intrinsicallyinactive. Cell Motil Cytoskeleton 66:777–790.

19. Chen Z, et al. (2010) Structure and control of the actin regulatory WAVE complex.Nature 468:533–538.

20. Lebensohn AM, Kirschner MW (2009) Activation of the WAVE complex by coincidentsignals controls actin assembly. Mol Cell 36:512–524.

21. Dubielecka PM, et al. (2010) Differential regulation of macropinocytosis by Abi1/Hssh3bp1 isoforms. PLoS ONE 5:e10430.

22. Leng Y, et al. (2005) Abelson-interactor-1 promotes WAVE2 membrane translocationand Abelson-mediated tyrosine phosphorylation required for WAVE2 activation. ProcNatl Acad Sci USA 102:1098–1103.

23. Xiong X, et al. (2008) Allosteric inhibition of the nonMyristoylated c-Abl tyrosinekinase by phosphopeptides derived from Abi1/Hssh3bp1. Biochim Biophys Acta 1783:737–747.

24. Dubielecka PM, et al. (2010) Abi1/Hssh3bp1 pY213 links Abl kinase signaling to p85regulatory subunit of PI-3 kinase in regulation of macropinocytosis in LNCaP cells.FEBS Lett 584:3279–3286.

25. Oikawa T, et al. (2004) PtdIns(3,4,5)P3 binding is necessary for WAVE2-inducedformation of lamellipodia. Nat Cell Biol 6:420–426.

26. Courtney KD, et al. (2000) Localization and phosphorylation of Abl-interactorproteins, Abi-1 and Abi-2, in the developing nervous system. Mol Cell Neurosci 16:244–257.

27. Grove M, et al. (2004) ABI2-deficient mice exhibit defective cell migration, aberrantdendritic spine morphogenesis, and deficits in learning and memory. Mol Cell Biol 24:10905–10922.

28. Pollitt AY, Insall RH (2008) Abi mutants in Dictyostelium reveal specific roles for theSCAR/WAVE complex in cytokinesis. Curr Biol 18:203–210.

29. Ryu JR, Echarri A, Li R, Pendergast AM (2009) Regulation of cell-cell adhesion by Abi/Diaphanous complexes. Mol Cell Biol 29:1735–1748.

30. Stradal T, et al. (2001) The Abl interactor proteins localize to sites of actinpolymerization at the tips of lamellipodia and filopodia. Curr Biol 11:891–895.

31. Echarri A, Lai MJ, Robinson MR, Pendergast AM (2004) Abl interactor 1 (Abi-1) wave-binding and SNARE domains regulate its nucleocytoplasmic shuttling, lamellipodiumlocalization, and wave-1 levels. Mol Cell Biol 24:4979–4993.

32. Rakeman AS, Anderson KV (2006) Axis specification and morphogenesis in the mouseembryo require Nap1, a regulator of WAVE-mediated actin branching. Development133:3075–3083.

33. Snapper SB, et al. (2001) N-WASP deficiency reveals distinct pathways for cell surfaceprojections and microbial actin-based motility. Nat Cell Biol 3:897–904.

34. Suetsugu S, Miki H, Takenawa T (1999) Identification of two human WAVE/SCARhomologues as general actin regulatory molecules which associate with the Arp2/3complex. Biochem Biophys Res Commun 260:296–302.

35. Ridley AJ, Paterson HF, Johnston CL, Diekmann D, Hall A (1992) The small GTP-bindingprotein rac regulates growth factor-induced membrane ruffling. Cell 70:401–410.

36. Legg JA, et al. (2007) N-WASP involvement in dorsal ruffle formation in mouseembryonic fibroblasts. Mol Biol Cell 18:678–687.

37. Steffen A, et al. (2006) Filopodia formation in the absence of functional WAVE- andArp2/3-complexes. Mol Biol Cell 17:2581–2591.

38. Kunda P, Craig G, Dominguez V, Baum B (2003) Abi, Sra1, and Kette control thestability and localization of SCAR/WAVE to regulate the formation of actin-basedprotrusions. Curr Biol 13:1867–1875.

39. Rogers SL, Wiedemann U, Stuurman N, Vale RD (2003) Molecular requirementsfor actin-based lamella formation in Drosophila S2 cells. J Cell Biol 162:1079–1088.

40. Hahne P, Sechi A, Benesch S, Small JV (2001) Scar/WAVE is localised at the tips ofprotruding lamellipodia in living cells. FEBS Lett 492:215–220.

41. Lommel S, et al. (2001) Actin pedestal formation by enteropathogenic Escherichia coliand intracellular motility of Shigella flexneri are abolished in N-WASP-defective cells.EMBO Rep 2:850–857.

42. Chen F, et al. (2000) Cdc42 is required for PIP(2)-induced actin polymerization andearly development but not for cell viability. Curr Biol 10:758–765.

43. Sugihara K, et al. (1998) Rac1 is required for the formation of three germ layersduring gastrulation. Oncogene 17:3427–3433.

44. Migeotte I, Omelchenko T, Hall A, Anderson KV (2010) Rac1-dependent collective cellmigration is required for specification of the anterior-posterior body axis of themouse. PLoS Biol 8:e1000442.

45. Ring C, Ginsberg MH, Haling J, Pendergast AM (2011) Abl-interactor-1 (Abi1) hasa role in cardiovascular and placental development and is a binding partner of thealpha4 integrin. Proc Natl Acad Sci USA 108:149–154.

Dubielecka et al. PNAS | April 26, 2011 | vol. 108 | no. 17 | 7027

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 7,

202

0