Arabidopsis acyl-CoA-binding protein ACBP2 interacts with an ethylene-responsive element-binding protein, AtEBP, via its ankyrin repeats

Hong-Ye Li1,2 and Mee-Len Chye1,*1Department of Botany, University of Hong Kong, Pokfulam Road, Hong Kong, China (*author forcorrespondence; e-mail mlchye@hkucc.hku.hk); 2Present address: South China Institute of Botany, ChineseAcademy of Sciences, Guangzhou, 510650, China

Received 19 December 2003; accepted in revised form 11 February 2004

Key words: agroinfiltration, ankyrin repeats, autofluorescent protein fusions, protein–proteininteractions, yeast two-hybrid assay

Abstract

Cytosolic acyl-CoA-binding proteins (ACBP) bind long-chain acyl-CoAs and act as intracellular acyl-CoAtransporters and maintain acyl-CoA pools. Arabidopsis thaliana ACBP2 shows conservation at the acyl-CoA-binding domain to cytosolic ACBPs but is distinct by the presence of an N-terminal transmembranedomain and C-terminal ankyrin repeats. The function of the acyl-CoA-binding domain in ACBP2 has beenconfirmed by site-directed mutagenesis and four conserved residues crucial for palmitoyl-CoA binding havebeen identified. Results from ACBP2:GFP fusions transiently expressed in onion epidermal cells havedemonstrated that the transmembrane domain functions in plasma membrane targeting, suggesting thatACBP2 transfers acyl-CoA esters to this membrane. In this study, we investigated the significance of itsankyrin repeats in mediating protein-protein interactions by yeast two-hybrid analysis and in vitro protein-binding assays; we showed that ACBP2 interacts with the A. thaliana ethylene-responsive element-bindingprotein AtEBP via its ankyrin repeats. This interaction was lacking in yeast two-hybrid analysis uponremoval of the ankyrin repeats. When the subcellular localizations of ACBP2 and AtEBP were furtherinvestigated using autofluorescent protein fusions in transient expression by agroinfiltration of tobaccoleaves, the DsRed:ACBP2 fusion protein was localized to the plasma membrane while the GFP:AtEBPfusion protein was targeted to the nucleus and plasma membrane. Co-expression of DsRed:ACBP2 andGFP:AtEBP showed a common localization of both proteins at the plasma membrane, suggesting thatACBP2 likely interacts with AtEBP at the plasma membrane.

Introduction

Cytosolic acyl-CoA-binding proteins (ACBP) bindlong-chain acyl-CoAs and act as intracellular acyl-CoA transporters and maintain acyl-CoA pools(Kragelund et al., 1999). Recently, A. thalianacDNAs and their corresponding genes encodingnovel forms of membrane-associated ACBP, desi-gnated ACBP1 and ACBP2, have been characte-rized (Chye, 1998; Chye et al., 1999, 2000). BothACBP1 and ACBP2 show conservation at theacyl-CoA binding domain to cytosolic ACBPs but

are distinct from cytosolic ACBPs by the presenceof a hydrophobic domain at the N-terminus and ofankyrin repeats at the C-terminus.

We have previously demonstrated that ACBP1and ACBP2 are membrane-associated proteins.Western blot analysis using anti-ACBP1 and anti-ACBP2 antibodies on A. thaliana protein showedthat ACBP1 (Chye, 1998) and ACBP2 (Li andChye, 2003) are located in the microsome-con-taining membrane fraction and in the subcellularfraction containing large particles (mitochondria,chloroplasts and peroxisomes). Particle gene

Plant Molecular Biology 54: 233–243, 2004.� 2004 Kluwer Academic Publishers. Printed in the Netherlands.

233

bombardment of onion epidermal cells revealedACBP1:GFP and ACBP2:GFP fusion proteins aretransiently expressed at the endoplasmic reticulum(ER) and predominantly at the plasma membrane;subsequently we showed that the N-terminalmembrane domain of ACBP1 or ACBP2 per seis sufficient in targeting green fluorescent protein(GFP) to the plasma membrane (Li andChye, 2003). The localization of ACBP1:GFP isconsistent with our previous observations byimmunoelectron microscopy whereby ACBP1 waslocalized to the plasma membrane (Chye et al.,1999).

To further elucidate the function of ACBP2, itis essential to investigate the function and signifi-cance of its C-terminal ankyrin repeats. Sinceankyrin repeats are known to mediate protein-protein interactions (Michaely and Bennett, 1992;Bork, 1993), their presence at the C-terminus ofACBP2 suggests that they could potentiallymediate interaction of ACBP2 with other proteins.Hence, yeast two-hybrid analysis was employed inthe present study to investigate the significance ofthese repeats in mediating protein–protein inter-actions and to identify any interacting proteins. Abait-containing sequence encoding ACBP2 wasconstructed for yeast two-hybrid screens. The baitwas designed to screen a cDNA library derivedfrom A. thaliana (Kohalmi et al., 1998) to identifyproteins that interacts directly with ACBP2.In vitro binding assays were used to confirm theprotein–protein interactions. Subsequently thein vivo co-localization of ACBP2 and its interact-ing protein, AtEBP, was further investigated byexpression of GFP and DsRed fusion proteins(Goodin et al., 2002) in Nicotiana tabacum uponagroinfiltration.

Materials and methods

Bacterial and yeast strains

Escherichia coli DH5a [supE44 DlacU169 (/80lacZDM15) hsdR17 recA1 endA1 gyrA96 thi-1relA1] was grown on LB (Luria-Bertani) medium(Sambrook et al., 1989). The two-hybrid libraryscreens were performed in the Saccharomyces ce-revisiae strain YPB2 [MATa ara3 his3 ade2 lys2trp1 leu2, 112 canr gal4 gal80 LYS2::GAL1-HIS3,URA3::(GAL1UAS17mers)-lacZ] (Kohalmi et al.,

1998). Co-transformants were plated on syntheticdextrose agar plates lacking leucine, tryptophanand histidine [SD–leu-trp-his] supplemented with15 mM 3-amino-1,2,4-triazole (3-AT) (Kohalmiet al., 1998).

Construction of a bait vector of GAL4(DB)-ACBP2 fusion

The bait plasmid pAT160 (Figure 1) was preparedby inserting the DNA encoding a truncated formof ACBP2 (amino acids 40–354) that lacks themembrane domain, into the centromere LEU2vector pBI-880 (a variant of pPC62 as described byChevray and Nathans (1992) and Kohalmi et al.(1998).

In the construction of pAT160, the 0.27 kbportion of ACBP2 cDNA encoding amino acids40–130 was first generated by PCR with templateplasmid pAT49, which contains the full-lengthACBP2 cDNA, and primers ML205 (5¢-CGTCACCCAGAGGAGTC-3¢) and ML266 (5¢-CATCACTCGGTACCTTCTG-3¢; KpnI siteunderlined). Forward primer ML205 correspondsto nucleotides þ118 to þ134 and reverse primerML266 corresponds to the complement of nucle-otides þ373 to þ391 on the ACBP2 cDNA(GenBank accession number AF178947). ThePCR product was cloned, with the KpnI internalsite adjacent to the T7 promoter, in pGEM-T Easyvector to generate plasmid pAT158. A 1.0 kbACBP2 cDNA fragment encoding amino acids131–354 of ACBP2 was cleaved off from plasmidpAT36 using KpnI and SacII and subsequentlyligated to similar sites in pAT158 to generatepAT159; plasmid pAT36 is an ACBP2 derivativein pBluescriptII KS(–) vector (pKS) containing thefull-length ACBP2 cDNA. Subsequently the1.3 kb SpeI fragment encoding amino acids40–354 of ACBP2 from plasmid pAT159 wassubcloned into the SpeI site of pBI-880 to obtainplasmid pAT160, with the 5¢ end of ACBP2 fusedto the 3¢ end of GAL4(DB).

In the construction of pAT230, the 0.74 kbportion of ACBP2 cDNA encoding amino acids215–354 of ACBP2 which contains the full-lengthankyrin repeats was cleaved off from plasmidpAT160 using NcoI and XbaI and subsequentlyreligated to generate pAT230. All constructs wereconfirmed by restriction digestion and nucleotidesequence analysis.

234

Yeast two-hybrid screening

Yeast strain YPB2 was transformed with this baitplasmid pAT160 and transformants were platedon [SD–leu] medium. Meanwhile an aliquot oftransformants was also tested on [SD–leu-his]medium supplemented with 15 mM 3-AT andabsence of growth on this medium would suggestthat the DB-‘bait’ fusion protein is unable to ini-tiate transcription of HIS3. Subsequently the bait-carrying strain was tested negative for b-galacto-sidase activity when using the X-Gal (5-bromo-4-chloro-3-indolyl-b-DD-galactopyranoside) colonyfilter assay. This further confirmed that the baitwas not able to activate transcription of the lacZreporter gene. Subsequently the prey vector pBI-771, a variant of pPC86 (Chevray and Nathans,1992; Kohalmi et al., 1998) was introduced intothis strain and its inability to grow on [SD–leu-trp-his] medium supplemented with 15 mM 3-AT andits lack of b-galactosidase activity were confirmed

before the bait was further used in cDNA libraryscreening.

In vitro protein binding assays

To further substantiate the observations in yeasttwo-hybrid screening, in vitro protein bindingexperiments were carried out according to Jarilloet al. (2001). All the constructs used in thesebinding studies were derivatives of vector pKS.The HindIII-SacI fragment from pBI-771 carryingGAL4(TA) (amino acids 768–881) was cloned intopKS at the corresponding restriction sites. TheGAL4(TA)-ACBP2 fusion construct was preparedby inserting a truncated form of ACBP2 encodingamino acids 2–346 from pAT49, on a SmaI-Bam-HI fragment, into the SmaI-BglII sites of pKS-TAwith the 5¢ end of TA-ACBP2 adjacent to the T3promoter.

The putative interacting candidates AtEBP andB114, as revealed from yeast two-hybrid screen-

Figure 1. Construction of ACBP2 and GAL4 DNA binding domain fusions. The numbers on the top lines indicate the positions of

amino acids in ACBP2. Nuclear localization signal consists of amino acids PKKKRKV. Plasmid pAT80 consists of the GAL4(DB)

fused to amino acids 215–354 of ACBP2. Plasmid pAT93 consists of amino acids 127–354 of ACBP2. Plasmid pAT89 consists of

amino acids 1–354 of ACBP2. Plasmid pAT134 consists of amino acids 1–354 of ACBP2. Plasmid pAT160 consists of amino acids 40–

354 of ACBP2. Plasmid pAT230 consists of amino acids 40–214 of ACBP2.

235

ings, were digested from their corresponding preyfusions in pBI-771 as SalI-NotI fragments andthese fragments were subsequently cloned intotheir corresponding restriction sites in pKS. ThecDNA encoding the Brassica napus lysophospha-tidic acid acyltransferase (BnLPAAT, gift from DrM. Frentzen; GenBank accession number Z95637)from plasmid pQEfam1 was digested with EcoRIand BglII and cloned into the EcoRI and BamHIsites of pKS.

Subsequently, in vitro transcription/translationwas used to generate polypeptides for binding as-says (Jarillo et al., 2001). The polypeptides wereanalyzed by SDS-PAGE and autoradiography(Jarillo et al., 2001). The binding of in vitro tran-scription/translation products to the GAL4(TA)-ACBP2 fusion protein, immobilized to protein A/agarose beads, using monoclonal antibody againstGAL4(TA) was investigated according to Jarilloet al. (2001).

Construction of plasmids used in co-localization

All binary vectors used in this study were deriva-tives of plasmids pGDG and pGDR which containgenes encoding the autofluorescent proteins GFPand DsRed, respectively (Goodin et al., 2002). AXhoI-HindIII fragment encoding ACBP2 (aminoacids 2–354) from plasmid pAT94 was cloned intothe corresponding sites on pGDR to create plas-mid pAT226 in which ACBP2 is fused downstreamof DsRed; plasmid pAT94 is an ACBP2 derivativethat carries the ACBP2 on an EcoRI-EcoRIrestriction fragment. The 1.0 kb XhoI-BamHIfragment containing the full-length AtEBP cDNAfrom plasmid pAT175 was ligated to the corre-sponding sites of pGDG to produce plasmidpAT225 in which AtEBP is fused downstream ofGFP; plasmid pAT175 contains the full-lengthAtEBP cDNA cloned into the XhoI and NotI sitesof pYES2 vector (Invitrogen, USA). The cloningjunctions in all constructs were confirmed bynucleotide sequence analysis.

Plant material for agroinfiltration

Tobacco (N. tabacum cv. Xanthi) plants weregrown in a greenhouse at 22 �C for 6 weeks. Twodays before agroinfiltration they were maintainedin a growth chamber at 22 �C under 16 h light/8 hdark as specified by Goodin et al. (2002).

Preparation of Agrobacterium suspension foragroinfiltration

Derivatives of Agrobacterium tumefaciens strainLBA4404 containing plasmid pGDG, pGDR,pAT225 or pAT226 were maintained on LuriaBertani (LB) solid medium supplemented withkanamycin (50 lg/ml) and streptomycin (25 lg/ml) and grown at 28 �C for 2 days. For agroinfil-tration, agrobacteria were grown at 28 �Covernight, in LB medium supplemented withkanamycin (50 lg/ml) and streptomycin (25 lg/ml). Preparation of Agrobacterium suspension andagroinfiltration of tobacco leaves in planta werecarried out following the procedures of Yang et al.(2000).

Laser-scanning confocal microscopy

A Zeiss LSM 510 inverted confocal laser-scanningmicroscope equipped with helium/neon lasers andmultitracking was used for the analysis of GFPand DsRed localization following the settings de-scribed by Goodin et al. (2002) with minor modi-fications. GFP fluorescence was excited at 488 nm,filtered through a primary dichroic (UV/488/543),a secondary dichroic of 545 nm and subsequentlythrough BP505–530 nm emission filters to thephotomultiplier tube (PMT) detector. DsRedfluorescence was excited at 543 nm, the emissionwas passed through similar primary and secondarydichroic mirrors and finally through a BP560–615 nm emission filter to the PMT detector. Theimages were processed using the LSM 510 soft-ware (Zeiss).

Results

Yeast two-hybrid screening

After attempts with several constructs (Figure 1c–f) failed in cDNA library screens, either due tohigh background in X-Gal filter assays (plasmidspAT80 and pAT93) or the lack of positive clonesin library screens (plasmids pAT89 and pAT134),GAL4(DB)-ACBP2 bait construct pAT160 (Fig-ure 1b), which consists of amino acids 40–354 ofACBP2, was eventually used in yeast two-hybridanalysis. The yeast strain YPB2 transformedwith this bait could not grow on [SD–leu-his]

236

supplemented with 15 mM 3-AT medium (Fig-ure 2a) and was tested negative in X-Gal colonyfilter assays (Figure 2b), while YPB2 transformedwith pAT80 grew on [SD–leu-his] supplementedwith 15 mM 3-AT medium (Figure 2a) and wastested positive in X-Gal colony filter assays (Fig-ure 2b), suggesting that the pAT160 bait alonecould not activate the transcription of reportergenes HIS3 and lacZ and was suitable for twohybrid screens. A GAL4(TA) activation tagged A.thaliana cDNA library was introduced into theyeast strain YPB2 harboring plasmid pAT160. Thenumber of independent transformants was deter-mined to be 2 · 106 after transformation withplasmid pAT160 and plating of an aliquot of theyeast transformation mixture on [SD–leu-trp]. Atotal of 600 putative positives were selected on[SD–leu-trp-his] supplemented with 15 mM 3-ATmedium. Initially 5 mM 3-AT was added to theselective medium according Kohalmi et al. (1998);however, to reduce background, we subsequentlyincreased the concentration of 3-AT to 15 mM.When transformants were screened for b-galacto-sidase activity with a X-Gal colony filter assay,twenty yeast clones that turned blue at differentintensities were identified as clones encodingputative interactors.

To further confirm specific interaction of thesecandidate proteins with ACBP2, library plasmids

of interest were rescued from these yeasttransformants for E. coli DH5a transformation.Subsequently, prey plasmids were extracted andreintroduced into the yeast intact strain YPB2 andused in co-transformation of a yeast strain thatcontained either the bait vector pBI-880 contain-ing GAL4(DB) only, pAT160 (GAL4(DB)-ACBP2) or an insert control bait of pAT157(GAL4(DB)-BnLPAAT) which carries the cDNAencoding lysophosphatidic acid acyltransferasefrom rapeseed (GenBank accession numberZ95637). The BnLPAAT clone was used because itencodes an enzyme involved in lipid metabolismthat we initially assumed could potentially interactwith ACBP2. When all of these transformedstrains were subjected to X-Gal filter assays, fourclones encoding interacting proteins were identi-fied. Their corresponding cDNA inserts in the preyvector pBI-771 were analyzed by restriction anal-ysis and subsequently by DNA sequence analysiswith the GAL(TA) specific forward primer BC304(5¢-CTATTCGATGATGAAGATACC-3¢) andthe ADH1-terminator reverse primer JN069 (5¢-TTGATTGGAGACTTGACC-3¢) (Kohalmiet al., 1998).

Results from analysis with the BLAST serverhttp://www.ncbi.nlm.nih.gov/cgi-bin/BLAST re-vealed that three interacting clones encoded a re-lated to AP2 protein, RAP2.3 (GenBank accession

Figure 2. Results of the HIS3 and lacZ reporter gene expression in the transformed yeast YPB2/pAT160 carrying GAL4(DB)-

ACBP2 bait. (a) HIS3 reporter: yeast transformants YPB2/pAT160 were plated on SD–leu medium (i), and SD–leu-his supplemented

with 15 mM 3-AT medium (ii); YPB2/pAT80 plated on SD–leu medium (iii), and SD–leu-his supplemented with 15 mM 3-AT medium

(iv). (b) lacZ reporter: yeast transformants were subjected to colony filter b-galactosidase assays. (i), YPB2/pAT160; (ii), YPB2/

pAT160-pBI-771 carrying both GAL4(DB)-ACBP2 bait and prey vector; (iii), YPB2/pAT80; (iv), YPB2/pAT80-pBI-771 carrying both

GAL4(DB)-ACBP2 (amino acids 215–354) bait and prey vector.

237

number AF003096; Okamuro et al., 1997).RAP2.3 is also known as ethylene-responsive ele-ment binding factor (ERF) protein AtEBP (Gen-Bank accession number Y09942; Buttner andSingh, 1997; Fujimoto et al., 2000). Of the threeclones encoding RAP2.3, two were identical andthey contained the full-length cDNA with 18 bp5¢-UTR and 239 bp 3¢-UTR. The third clone wastruncated at the 5¢ end and encodes only aminoacids E25 to E248 of RAP2.3, E248 being the lastamino acid in the open reading frame (ORF); italso contains 239 bp of 3¢-UTR. The amino acidsequence of AtEBP is shown in Figure 3a. AnAP2/EREBP (ethylene-responsive element-bind-ing protein) domain is present in AtEBP at aminoacids 76–143 (Okamuro et al., 1997).

The fourth positive clone, clone B114, encodesa truncated unknown protein corresponding toGenBank accession number AY133849. Thistruncated peptide includes amino acids G87 to theS161, S161 being the last amino acid in the predicted

ORF of AY133849. The deduced amino acid se-quence of B114 and the corresponding deducedfull-length peptide from AY133849 are shown inFigure 3b. This protein has no significanthomology to any known proteins in the databaseand conserved domains/motifs were lacking.

The results of X-Gal filter assays (Figure 4)revealed that the GAL4(DB)-ACBP2 fusioninteracted with the full-length AtEBP fused to theGAL4 DNA transcription activation domain, bythe production of significant levels of b-galactosi-dase activity (Figure 4a). AtEBP alone was notable to initiate transcription of lacZ for no bluecolor was detected in the X-Gal assay of YPB2harboring the AtEBP prey only (Figure 4b). Nointeractions were observed between ACBP2 andunrelated control proteins including GAL4(DB)(Figure 4c) and BnLPAAT (Figure 4d). Mean-while, the ACBP2 derivative lacking ankyrin re-peats (pAT230) did not show interaction withAtEBP for no blue signal could be detected in

Figure 3. Amino acid sequences of AtEBP and B114. (a). AtEBP containing 248 amino acids. The AP2/EREBP domain (K76–P143) is

in bold type. The location of the conserved YRG and RAYD elements are indicated by brackets. Putative nuclear localization signals

are underlined. (b). B114. The truncated region of B114 (G87–S161) is in bold type within the full-length deduced peptide sequence.

Figure 4. Colony filter b-galactosidase assays of candidate proteins AtEBP and B114 from yeast two-hybrid screens. (a), YPB2/

AtEBP+ACBP2; (b), PB2/AtEBP; (c), YPB2/AtEBP+GAL4 (DB); (d), YPB2/AtEBP+BnLPAAT; (e), PB2/AtEBP+pAT230; (f),

YPB2/B114+ACBP2; (g), YPB2/B114; (h), YPB2/B114+GAL4 (DB); (i), YPB2/B114+BnLPAAT; (j), YPB2/B114+pAT230.

238

YPB2/AtEBP+pAT230 (Figure 4e). Similar re-sults were obtained with B114, with a relativelyslight blue signal in YPB2/B114+ACBP2 (Fig-ure 4f) and the lack of blue signals in YPB2/B114(Figure 4g), YPB2/B114+GAL(DB) (Figure 4h),YPB2/B114+BnLPAAT (Figure 4i) and YPB2/B114+pAT230 (Figure 4j). Therefore, from yeasttwo-hybrid analysis, AtEBP and B114 were iden-tified as putative proteins that interact withACBP2 via the ankyrin repeats.

Confirmation of ACBP2-interacting proteins byin vitro protein-binding assays

All the constructs used in in vitro protein-bindingstudies were cloned into pBluescriptII KS(–) vec-tor (pKS). Subsequently polypeptides were gener-ated by in vitro transcription/translation and theresultant polypeptides were analyzed by SDS-PAGE. An autoradiograph of the gel showed thatthe estimated molecular masses of the four in vitrotranslation products, TA-ACBP2, BnLPAAT,AtEBP and B114, corresponded to 50, 43, 27 and8 kDa, respectively, in agreement with their cal-culated molecular masses (Figure 5).

Binding studies of in vitro transcription/trans-lation products to the GAL4(TA)-ACBP2 fusionprotein, immobilized to protein A/agarose beads,

using monoclonal antibody against GAL4(TA),showed that the GAL4(TA)-ACBP2 fusion proteinsignificantly binds AtEBP (Figure 5, lane 5).However, neither binding of GAL4(TA)-ACBP2to B114 (Figure 5, lane 6) nor binding of GAL4-(TA)-ACBP2 to the negative control BnLPAAT(Figure 5, lane 7) was observed. These resultssuggest that AtEBP, but not B114, binds ACBP2.

Co-localization of DsRed:ACBP2 and GFP:AtEBP

After agroinfiltration of tobacco leaves, observa-tions were carried out using a green filter toinvestigate the fluorescence pattern of GFP and ared filter to visualize the fluorescence of DsRed.Results revealed similar expression patterns ofnon-fused GFP protein from pGDG (Figure 6a)and non-fused DsRed protein from pGDR(Figure 6b); they were expressed throughout thewhole cell including the nuclei. In contrast, theDsRed:ACBP2 fusion protein was expressed inthe plasma membrane (Figure 6c). The GFP:AtEBPfusion protein was localized to the nucleus andplasma membrane (Figure 6d). When DsRe-d:ACBP2 and GFP:AtEBP were co-expressedfollowing co-infiltration of tobacco leaves, bothsignals, i.e. both fusion proteins, accumulated inthe plasma membrane with some GFP:AtEBP re-

Figure 5. ACBP2 and AtEBP interact in an in vitro protein binding assay. The left panel (Input) shows three input proteins produced

by in vitro transcription/translation. Of each sample 2 ll was loaded on 15% SDS-PAGE gel. The right panel (Binding) shows in vitro

protein binding between interacting proteins. Arrows show the location of the polypeptides.

239

tained in the nucleus (Figure 6e–g). There did notseem to be an observable decrease in AtEBP signalin the nucleus when ACBP2 was co-expressed inthe same cell.

Discussion

Several bait fusions between GAL4(DB) and dif-ferent portions of ACBP2 were designed in thisstudy because initial constructs were found notsuitable for yeast two-hybrid analysis. Figure 1shows the schematic construction of the baits usedincluding plasmid pAT80, constructed by fusingGAL4(DB) and a portion of ACBP2 containingthe ankyrin repeats only (Figure 1f), and plasmidpAT93, constructed by fusing GAL4(DB) and apartial ACBP2 containing the ankyrin repeats andthe acyl-CoA binding domain (Figure 1e). Whenthese two baits were tested for b-galactosidaseactivity using the X-Gal colony filter assays, bothexhibited high background levels of b-galactosi-dase activity. Thus, they could not be further usedin cDNA library screens. Subsequently, when thefull-length ACBP2 was fused with GAL4(DB) toconstruct a bait (plasmid pAT89 in Figure 1d), ittested negative in X-Gal colony filter assays andwas thus deemed suitable for use in screening thecDNA library. However no positives were ob-

tained on cDNA library screening and we deducedthat the short N-terminal hydrophobic signal-an-chor transmembrane domain of ACBP2 couldpossibly missort the ACBP2 fusion protein out ofthe nucleus. Indeed, subsequent results fromlocalization of GFP fusion proteins with ACBP2did confirm that the membrane domain in ACBP2per se targets the ACBP2:GFP fusion protein tosubcellular membranes (Li and Chye, 2003). It islikely that the GAL4(DB)-ACBP2 fusion proteinwas targeted to the subcellular membranes insteadof the nucleus, obliterating protein-protein inter-actions in the yeast cell. Hence, a nuclear locali-zation signal was added upstream of the full-lengthACBP2 (plasmid pAT134 in Figure 1c). Althoughthis bait tested negative in X-Gal colony filter as-says, no positives were obtained in further cDNAlibrary screening. Despite the problems encoun-tered with various constructs in initiating yeasttwo-hybrid screens, the outcome of this exerciserevealed that the stronger membrane localizationsignal in ACBP2 predominates over the intro-duced nuclear localization signal in theGAL4(DB)-NLS-ACBP2 fusion (Figure 1c).

Eventually, when a large portion of ACBP2(amino acids 40–354) lacking only the membranedomain was fused to GAL4(DB) and used as bait(plasmid pAT160 in Figure 1b), yeast two-hybridscreens yielded three independent clones. Interest-

Figure 6. Localization of DsRed:ACBP2 and GFP:AtEBP. Representative tobacco leaf epidermal cells are shown by laser-scanning

confocal microscopy after agroinfiltration (a–d) or co-infiltration (e–g) of plasmid pGFP, pDsRed, pAT225 or pAT226. (a), GFP

expressed from pGDG; (b), DsRed expressed from pGDR; (c), DsRed:ACBP2 expressed from pAT226; (d), GFP:AtEBP expressed

from pAT225; (e), f and g, co-expression of DsRed:ACBP2 and GFP:AtEBP; e, red channel showing location of DsRed:ACBP2; (f),

green channel showing localization of GFP:AtEBP; (g), overlay of images e and f. Arrowheads indicate the position of nuclei. Bar,

20 lm.

240

ingly, all of the three clones encoded an AP2-re-lated protein RAP2.3 (GenBank accession numberAF003096; Okamuro et al., 1997) belonging to alarge family of recently discovered plant tran-scription factors with an AP2/EREBP domain,suggesting that ACBP2 interacts with RAP2.3.RAP2.3 is identical to AtEBP, an ERE-bindingprotein, which has been independently character-ized by Buttner and Singh (1997; GenBank acces-sion number Y09942). The interaction of AtEBP toACBP2 was further substantiated by in vitro pro-tein binding assays and in vivo studies using auto-fluorescent protein fusions in transient expressionby agroinfiltration of tobacco leaves. This interac-tion was obliterated when the ankyrin repeats wereremoved in plasmid pAT230, confirming that theyconstitute the interacting domain. Recently, Pet-rescu et al. (2003) have observed that recombinantmouse ACBP in rat hepatoma cells and transfectedCOS-7 cells interacts with the hepatocyte nuclearfactor-4a (HNF-4a), a nuclear binding protein thatregulates transcription of genes involved in bothlipid and glucose metabolism.

In comparison, the BnLPAAT cDNA encodingrapeseed lysophosphatidic acid acyltransferase,which we initially assumed could potentiallyinteract with ACBP2, was tested negative inACBP2 interaction in yeast two-hybrid analysisand in vitro protein binding assay. The peptideB114, previously identified in yeast two-hybridanalysis as a putative protein that could poten-tially interact with ACBP2 did not show binding toACBP2 on in vitro binding assays, suggesting thatit is an artifact from yeast two-hybrid analysis.The tendency of the yeast two-hybrid screen inyielding artifacts can be largely avoided by the useof appropriate plating conditions during thescreening process itself. However, if artifacts doescape selection under suboptimal conditions, invitro binding assays are always essential in con-firming results.

The AP2/EREBP proteins consist of a largefamily of recently discovered plant transcriptionfactors involved in plant growth and develop-mental regulation (Riechmann and Meyerowitz,1998). The AP2 DNA binding domain was initiallydiscovered in APETALA2 (AP2) in Arabidopsis, aprotein that plays a role in the ABC model offlower organogenesis (Jofuku et al., 1994). Subse-quently this DNA-binding domain has been foundto occur in many other plant transcription factors

and is unique to plants (Okamuro et al., 1997). Allof these proteins contain a conserved AP2 domain(Jofuku et al., 1994), which is involved in DNAbinding (Ohme-Tagaki and Shinshi, 1995) or inmediating protein–protein interactions (Okamuroet al., 1997). These proteins can be divided intotwo subfamilies based on the number of AP2DNA-binding domains present. The AP2 domainis present as two repeats in the AP2 subfamily,which includes AP2, AINTEGUMENTA (ANT)from Arabidopsis (Elliot et al., 1996) and Glossy15from maize (Moose and Sisco, 1996). ANT is re-quired for plant organogenesis and in regulatingapical meristem activity in Arabidopsis (Elliottet al., 1996). Glossy regulates leaf epidermal cellidentity during the vegetative phase change (Evanset al., 1994). In contrast, the AP2 domain ispresent as one repeat in the ERE-binding protein(EREBP) subfamily, which is named after the firstisolated genes, ethylene-response element-bindingproteins (Ohme-Tagaki and Shinshi, 1995). Pro-teins of the EREBP subfamily are mainly ex-pressed in response to biological or physical stress,such as pathogen attack, ethylene or ABA treat-ment, drought, cold or cadmium treatment. AtE-BP (RAP2.3) belongs to the EREBP subfamily bythe presence of one AP2/EREBP domain (Buttnerand Singh, 1997; Okamuro et al., 1997).

Sequence comparisons between members of theAP2 family revealed a highly conserved motif re-ferred to as the RAYD element within the AP2/EREBP domain. The RAYD element contains aconserved core region that is predicted to form anamphipathic a-helix. This a-helical structure hasbeen implicated a role in DNA binding or inmediating protein–protein interactions importantfor RAP2.3 (AtEBP) function (Okamuro et al.,1997). This interaction may involve the formationof homo- or heterodimers, similar to that observedfor the MADS (MCM1, AG and ARG80, DEFand SRF) box family of plant regulatory proteins(Huang et al., 1996; Riechmann et al., 1996). Thusthe presence of a potential site for protein–proteininteractions in AtEBP suggests that ACBP2 andAtEBP may form heterodimers through theinteraction of the ankyrin repeats and the amphi-pathic a-helix, respectively. As a transporter andpool former of acyl-CoA esters, ACBP2 coulddonate acyl-CoA esters to regulatory factors as inthe case of the gene regulation of OLE1 inSaccharomyces cerevisiae in which saturated fatty

241

acids induce OLE1 transcription while unsatu-rated fatty acids repress its expression (Choi et al.,1996). In plants, fatty acid-derived signals havebeen implicated in the regulation of plant defenseand development (Farmer et al., 1998) while oleicacid has been demonstrated to activate a mem-brane-associated phospholipase D that belongs tothe phospholipase D family involved in hormonaland stress signaling (Wang and Wang, 2001).

Although AtEBP is predicted to be targeted tothe nucleus as determined with the PSORT WWWserver for the prediction of subcellular localizationof proteins (http://psort.nibb.ac.jp), GFP:AtEBPwas observed not only in the nucleus, also in theplasma membrane. Hence, interactions of ACBP2and AtEBP could occur subcellularly in the plas-ma membrane, a location common to both pro-teins from results using DsRed:ACBP2 andGFP:AtEBP fusions. Previous results fromACBP2:GFP transiently expressed in onion epi-dermal cells had shown that ACBP2 is targeted tothe plasma membrane (Li and Chye, 2003). Re-sults in this study, from agroinfiltration of DsRe-d:ACBP2 in tobacco leaves, reconfirm thesubcellular localization of ACBP2 to the plasmamembrane. Our earlier observation of the mem-brane-targeting domain in ACBP2 predominatingover an introduced nuclear localization signal(PKKKRKV) in plasmid pAT134 (Figure 1c)suggests that this strong membrane-targetingdomain predominates in maintaining theACBP2:AtEBP interacting proteins in the plasmamembrane. It has already been reported that sometranscription factors, such as the Arabidopsis floralidentity protein LEAFY (LFY), do move betweencells (Wu et al., 2003). Hence, a possible role forACBP2 could be the mediation of AtEBP move-ment between cells.

Acknowledgements

We thank W.L. Crosby (Plant BiotechnologyInstitute, NRC Canada) for provision of the yeasttwo-hybrid system, M.M. Goodin (University ofCalifornia, Berkeley) for provision of vectors inconstruction of autofluorescent protein fusionsand M. Frentzen (RWTH, Institut fur Biologie I,Aachen, Germany) for plasmid pQEfam1. Thiswork was supported by the Research GrantsCouncil of the Hong Kong Special Administrative

Region, China (Project HKU7232/00M). H.Y.L.was supported by a postgraduate studentship fromthe University of Hong Kong.

References

Bork, P. 1993. Hundreds of ankyrin-like repeats in functionally

diverse proteins: mobile modules that cross phyla horizon-

tally? Proteins Struct. Funct. Genet. 17: 363–374.

Buttner, M. and Singh, K.B. 1997. Arabidopsis thaliana

ethylene-responsive element binding protein (AtEBP), an

ethylene-inducible, GCC box DNA-binding protein interacts

with an ocs element binding protein. Proc. Natl. Acad. Sci.

USA 94: 5961–5966.

Chevray, P.M. and Nathans, D. 1992. Protein interaction

cloning in yeast: Identification of mammalian proteins that

react with the leucine zipper of Jun. Proc. Natl. Acad. Sci.

USA 89: 5789–5793.

Choi, J.Y., Stukey, J., Hwang, S.Y. and Martin, C.E. 1996.

Regulatory elements that control transcription activation and

unsaturated fatty acid-mediated repression of the Saccharo-

myces cerevisiae OLE1 gene. J. Biol. Chem. 271: 3581–3589.

Chye, M.-L. 1998. Arabidopsis cDNA encoding a membrane-

associated protein with an acyl-CoA binding domain. Plant

Mol. Biol. 38: 827–838.

Chye, M.-L., Huang, B.-Q. and Zee, S.Y. 1999. Isolation of a

gene encoding Arabidopsis membrane-associated acyl-CoA-

binding protein and immunolocalization of its gene product.

Plant J. 18: 205–214.

Chye, M.-L., Li, H.-Y. and Yung, M.-H. 2000. Single amino

acid substitutions at the acyl-CoA-binding domain interrupt14[C]palmitoyl-CoA binding of ACBP2, an Arabidopsis acyl-

CoA-binding protein with ankyrin repeats. Plant Mol. Biol.

44: 711–721.

Elliott, R.C., Betzner, A.S., Huttner, E., Oakes, M.P., Tucker,

W.Q.J., Gerentes, D., Perez, P. and Smyth, D.R. 1996.

AINTEGUMENTA, an APETALA2-like gene of Arabidopsis

with pleiotropic roles in ovule development and floral organ

growth. Plant Cell 8: 155–168.

Evans, M.M.S., Passas, H.J. and Poethig, R.S. 1994. Hetero-

chronic effects of glossy15 mutations on epidermal cell

identity in maize. Development 120: 1971–1981.

Farmer, E.E., Weber, H. and Vollenweider, S. 1998. Fatty acid

signaling in Arabidopsis. Planta 206: 167–174.

Fujimoto, S.Y., Ohta, M., Usui, A., Shinshi, H. and Ohme-

Takagi, M. 2000. Arabidopsis ethylene-responsive element

binding factors act as transcriptional activators or repressors

ofGCCbox-mediated gene expression. Plant Cell 12: 393–404.

Goodin, M.M., Dietzgen, R.G., Schichnes, D., Ruzin, S. and

Jackson, A.O. 2002. pGD vectors: versatile tools for the

expression of green and red fluorescent protein fusions in

agroinfiltrated plant leaves. Plant J. 31: 375–383.

Huang, H., Tudor, M., Su, T., Zhang, Y., Hu, Y. and Ma, H.

1996. DNA binding properties of two Arabidopsis MADS

domain proteins: binding consensus and dimer formation.

Plant Cell 8: 81–94.

Jarillo, J.A., Capel, J., Tang, R.H., Yang, H.Q., Alonso, J.M.,

Ecker, J.R. and Cashmore, A.R. 2001. An Arabidopsis

242

circadian clock component interacts with both CRY1 and

phyB. Nature 410: 487–490.

Jofuku, K.D., den Boer, B.G., Van Montagu, M. and

Okamuro, J.K. 1994. Control of Arabidopsis flower and seed

development by the homeotic gene APETALA2. Plant Cell 6:

1211–1225.

Kohalmi, S.E., Reader, L.J.V., Samach, A., Nowak, J.,

Haughn, G.W. and Crosby, W.L. 1998. Identification and

characterization of protein interactions using the yeast 2-

hybrid system. In: S.B. Gelvin and R.A. Schilperoort (Eds.),

Plant Molecular Biology Manual, Kluwer Academic Pub-

lishers, Dordrecht, Netherlands, pp. M1/1–30.

Kragelund, B.B., Knudsen, J. and Poulsen, F.M. 1999. Acyl-

coenzyme A binding protein (ACBP). Biochim. Biophys.

Acta 1441: 150–161.

Li, H.-Y. and Chye, M.-L. 2003. Membrane localization of

Arabidopsis acyl-CoA binding protein ACBP2. Plant Mol.

Biol. 51: 483–492.

Michaely, P. and Bennett, V. 1992. The ANK repeat: A

ubiquitous motif involved in macromolecular recognition.

Trends Cell Biol. 2: 127–129.

Moose, S.P. and Sisco, P.H. 1996. Glossy15, an APETALA2-

like gene from maize that regulates leaf epidermal cell

identity. Genes Dev. 10: 3018–3027.

Ohme-Tagaki, M. and Shinshi, H. 1995. Ethylene-inducible

DNA binding proteins that interact with an ethylene-

responsive element. Plant Cell 7: 173–182.

Okamuro, J.K., Caster, B., Villarroel, R., Van Montagu, M.

and Jofuku, K.D. 1997. The AP2 domain of APETALA2

defines a large new family of DNA binding proteins in

Arabidopsis. Proc. Natl. Acad. Sci. USA 94: 7076–7081.

Petrescu, A.D., Payne, H.R., Boedecker, A., Chao, H., Hertz,

R., Bar-Tana, J., Schroeder, F. and Kier, A.B. 2003. Physical

and functional interaction of acyl-CoA binding protein

(ACBP) with hepatocyte nuclear factor-4a (HNF-4a). J.

Biol. Chem. 278: 51813–51824.

Riechmann, J.L., Krizek, B.A. and Meyerowitz, E.M. 1996.

Dimerization specificity of Arabidopsis MADS domain

homeotic proteins APETALA1, APETALA3, PISTILLA-

TA, and AGAMOUS. Proc. Natl. Acad. Sci. USA 93: 4793–

4798.

Riechmann, J.L. and Meyerowitz, E.M. 1998. The AP2/

EREBP family of plant transcription factors. J. Biol. Chem.

379: 633–646.

Sambrook, J., Fritsch, E.F. and Maniatis, T. 1989. Molecular

Cloning: A Laboratory Manual, 2nd edn. Cold Spring

Harbor Laboratory Press, Plainview, NY.

Wang, C. and Wang, X. 2001. A novel phospholipase D of

Arabidopsis that is activated by oleic acid and associated

with the plasma membrane. Plant Physiol. 127: 1102–

1112.

Wu, X., Dinneny, J.R., Crawford, K.M., Rhee, Y., Citovsky,

V., Zambryski, P.C. and Weigel, D. 2003. Modes of

intercellular transcription factor movement in the Arabidopsis

apex. Development 130: 3735–3745.

Yang, Y., Li R. and Qi, M. 2000. In vivo analysis of plant

promoters and transcription factors by agroinfiltration of

tobacco leaves. Plant J. 22: 543–551.

243