Plant Science 157 (2000) 1–12

Lipid transfer proteins are encoded by a small multigene family inArabidopsis thaliana�

Vincent Arondel 1 2, Chantal Vergnolle2, Catherine Cantrel, Jean-Claude Kader *Laboratoire de Physiologie Cellulaire et Moleculaire, CNRS/Uni6ersite Pierre et Marie Curie UMR7632, Case 154, 4 Place Jussieu,

F-75252 Paris cedex 05, France

Received 26 October 1999; received in revised form 28 January 2000; accepted 21 February 2000

Abstract

Lipid transfer proteins (LTPs) are small, basic and abundant proteins in higher plants. They are capable of binding fatty acidsand of transferring phospholipids between membranes in vitro. LTPs from this family contain a signal peptide and are secretedin the cell wall. Their biological function is presently unknown. LTPs have been suggested to participate to cutin assembly andto the defense of the plants against pathogens. A genetic approach should prove useful to provide clues on their in vivo functions.Here, the characterization of the LTP gene family in Arabidopsis thaliana is described. At least 15 genes were identified, their mapposition determined and the expression pattern characterized for six of them. All the sequences exhibit the typical features of plantLTPs. The molecular weight is close to 9 kDa, the isoelectric point is near 9 (except for three acidic LTPs), and typical aminoacid residues such as cysteines are conserved. Genomic DNA blotting hybridization experiments performed using ltp1 to ltp6 asprobes indicate that ltps form distinct 1–3 gene subfamilies which do not cross hybridize. Expression studies indicate that all thegenes tested are expressed in flowers and siliques, but not in roots. Ltp1, ltp5 and ltp2 are expressed significantly in leaves, whileltp6 is detected only in 2–4-week-old leaves. In addition, ltp4 and ltp3 are strongly upregulated by abscisic acid (ABA). Tandemrepeats can be noted concerning ltp1 and ltp2 on chromosome 2, ltp3 and ltp4 on chromosome 5 and ltp5 and ltp12 onchromosome 3. While ltp7, ltp8 and ltp9 map at the same position on chromosome 2, the other genes are dispersed throughoutthe genome. The characterization of the Arabidopsis ltp gene family will permit to initiate a genetic approach for determining thein vivo function(s) of these proteins. © 2000 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Lipid transfer protein; Arabidopsis thaliana ; cDNA; Tandemly repeated genes; Abscisic acid

www.elsevier.com/locate/plantsci

1. Introduction

Proteins capable of transferring lipids betweenmembranes in vitro have been purified from awide range of living organisms [1]. Some of themhave been cloned and the amino acid sequencecomparisons revealed that these proteins fall intoseveral different classes that are unrelated basedon their primary structure. Although all theseproteins, called lipid transfer proteins (LTP), wereinitially supposed to participate to membrane bio-genesis, no clear evidence of such a role has beendemonstrated in vivo. Actually, the biologicalfunction of some LTPs [2] begins to be investi-gated. For example, the sec14 protein of yeast,which is a phosphatidylinositol transfer protein

Abbre6iations: ABA, abscisic acid; BAC, bacterial artificial chro-mosome; BLAST, basic local alignment search tool; EST, expressedsequence tag; LTP, lipid transfer protein; NMR, nucleic magneticresonance; PCR, polymerase chain reaction; PITP, phosphatidylinosi-tol transfer protein; RFLP, restriction fragment length polymor-phism; TAIR, The Arabidopsis Information Resource; T-DNA,transferred DNA; TIGR, The Institute for Genomic Research; YAC,yeast artificial chromosome.� 3 Accession numbers: ltp1: AF159798; ltp2: AF159799; ltp3:

AF159800; ltp4: AF159801; ltp5: AF159802; ltp6: AF159803.* Corresponding author.E-mail address: kader@ccr.jussieu.fr (J.-C. Kader).1 Present address: Laboratoire de Lipolyse Enzymatique, UPR

CNRS 9025, Universite de la Mediterranee, Marseille, France.2 These authors have contributed equally to this work

0168-9452/00/$ - see front matter © 2000 Elsevier Science Ireland Ltd. All rights reserved.

PII: S 0 1 6 8 -9452 (00 )00232 -6

V. Arondel et al. / Plant Science 157 (2000) 1–122

(PITP) was shown to be a sensor of the lipidcomposition of the Golgi membrane, and its ca-pacity to down regulate phosphatidylcholinebiosynthesis in this organelle was demonstrated. Inaddition, mammalian PITPs, which are struc-turally unrelated to sec14p, seem to play a key rolein phospholipase C mediated signaling throughtheir binding capacity to phosphoinositides.Therefore, it appears that these two different LTPsdo not possess identical physiological functionsand that neither of them transfer lipids betweenintracellular membranes.

In higher plants, LTPs form a very homoge-neous class of protein, if a sec14-like PITP isexcluded [3]. They are small (9 kDa), abundantand basic proteins that contain eight cysteineresidues [4,5]. They are capable of transferringseveral different phospholipids, and they can bindfatty acids [6] and acyl-CoA esters. Structural datahave been recently published, based on both X-raydiffraction [7] and nucleic magnetic resonance(NMR) [8] techniques. These results indicate thatLTPs contain a hydrophobic pocket capable toaccommodate a fatty acid or a lysophospholipidmolecule.

Numerous LTP cDNAs have been cloned fromdifferent plant species [4]. These data have indi-cated the existence of multiple isoforms, that aredifferently expressed and regulated [9–17]. How-ever, most of these genes are preferentially ex-pressed in epidermal cells of leaves and in flowers,and very rarely in roots.

All non-specific plant LTPs characterized so farcontain a signal peptide, and immunolocalizationdata indicate that they locate to the cell wall [18].These proteins have also been shown to besecreted by cell cultures [15,19]. This localizationtherefore preclude a priori an intracellular role forthese proteins. Possible biological functions havebeen suggested. LTP might play a role in cutinand wax assembly [20,21]. Another possible role isbased on the antifungal properties displayed bysome LTP [22]. These proteins might play a role inthe defense of the plant against pathogen attack[23–25]. Indeed, it has been shown that increasingthe level of an LTP in transgenic tobacco enhancesthe resistance of the plant towards a pathogen[26]. A possible way to find a role for theseproteins would consists in obtaining mutants ortransgenic plants that express antisense RNA. Thephenotypic characterization of these plants would

provide clues with regards to the in vivo functionof these proteins. Arabidopsis thaliana seems to bethe most appropriate plant material for a geneticapproach, since it is very easy to transform [27],that numerous tools are available that allows re-verse genetics (transferred DNA, (T-DNA), [28] ortransposons tagged lines) and that the genomeprograms have yielded a considerable amount ofgenomic and cDNA sequences [29,30]. Here, thecharacterization of the Arabidopsis ltp gene familyis described.

2. Material and methods

2.1. Plant and DNA materials

A. thaliana (ecotype Columbia:2) plants weregrown at 25°C with a 16 h-photoperiod (150 mEs−1 m−2) as described [3]. Plant material wasrapidly collected and immediately frozen in liquidnitrogen and stored at −80°C prior to nucleicacid isolation. Abscisic acid (ABA) treatmentswere performed on plants at the rosette stage. Theplants were transferred to a nylon mesh floatingon a liquid nutrient solution for 4 days. ABA(10−4 M) was then added and the plants werecollected 24 or 48 h afterwards.

cDNA clones were obtained from the Arabidop-sis Biological Resource Center at Ohio State Uni-versity and the recombinant inbred lines from theNothingham Arabidopsis Stock Center. The CICyeast artificial chromosome (YAC) library [31] wasobtained from Dr D. Bouchez (INRA Versailles,France). Rab 18 cDNA and the ribosomal DNAprobes were obtained from Dr M. Delseny(CNRS-Perpignan, France).

2.2. Nucleic acids purification

The plant material was ground to a fine powderin liquid nitrogen. RNAs and genomic DNA wereextracted as previously described [3].

2.3. Northern and Southern blot hybridizationanalyses

RNA was fractionated on 1.5% formaldehydeagarose gels and transferred onto Hybond Nmembrane (Amersham, UK). Genomic DNA (1mg per lane) was restricted, fractionated on 0.8%

V. Arondel et al. / Plant Science 157 (2000) 1–12 3

agarose gel and transferred onto Positive™ mem-branes (Appligene, France) according to manufac-turer’s instructions.

Hybridizations were carried out as describedpreviously [3] at 65°C with randomly primedcDNA probes or at 50°C with oligonucleotideprobes.

2.4. Gene mapping and sequencing

YAC library screening was performed and yeastDNA was prepared according to Ref. [32]. Se-quencing was performed either according to Ref.[33] using the sequenase version 2.0 kit (USB,USA), or by automated sequencing (CompanyESGS, France).

Mapping data were processed by D. Bouchez(INRA Versailles, France) with respect to physicalmapping and Sean May (Nothingham) with re-spect to restriction fragment length polymorphism(RFLP) mapping. Other routine DNA manipula-tions were as in Ref. [34].

2.5. Bioinformatics

Identification and search for LTP through data-bases (GenBank, The Arabidopsis InformationResource, TAIR) was performed using both key-word searching and the basic local alignmentsearch tool (BLAST) (BLASTX, BLASTN andTBLASTN) softwares [35]. Sequence comparisonsand phylogenic analyses were performed using theClustalW software [36] and Phylip package [37]which is based on the neighbor joining methodvalidated by bootstrap statistical analysis. Matureamino acid sequences were aligned by ClustalWusing a PAM matrix. The results from the align-ment were used for constructing the tree using theneighbor joining method. Those programs wereaccessed from the Infobiogen web server (http://www.infobiogen.fr), using the default options un-less otherwise indicated.

3. Results

3.1. Identification of ltp-related cDNAs

Expressed sequence tag (EST) database (Univer-sity of Minnesota) was searched for files contain-ing the words lipid and transfer. More than 200

entries were found. The cDNAs were classed intofamilies based on The Institute for Genomic Re-search (TIGR) tentative consensus. The remainingsequences were compared to these consensus andthe ESTs that presented more than 93% identityover a 100 nucleotide stretch were considered asbeing encoded by the same gene. Only ‘typical’LTPs were retained, that is, proteins that con-tained around 120 amino acid, with the typicalLTP amino acid pattern [4]. Larger proteins (150aa and more), and 7 kDa LTPs [38] were excluded.Six different classes were defined and the largestrepresentative from each class was sequenced andcharacterized. These cDNAs were designated ltp1to 6; ltp1 and ltp2 corresponding to the ltp1 andltp2 genes already described by Ref. [18] and ltp3is likely to be identical to Clark and Bohnert ltp3gene [39]. In addition, ltp4 is identical to ltp-a2which was purified from a crude cell wall prepara-tion [25]. BLAST alignments were carried outusing these six genes and gene products as ‘probes’and nine additional ltp genes were identified.These genes were designated ltp7 to ltp15 (Table1).

3.2. Sequence analysis

All of the deduced proteins share a similarhydrophobic profile, and contain a typical signalpeptide [40]. The characteristics of the proteinscoded for by these cDNAs are summarized inTable 1. The number of amino acid ranges from112 to 123 or from 89 to 98 if the signal peptide isexcluded. The isoelectric point of the matureproteins is usually close to 9, with the exceptionsof ltp5 (11.4), ltp6 and ltp12 (7.7), ltp15 (7.5).Interestingly, ltp8, ltp9 and ltp14 are acidic, withpI of 4.9, 5.2 and 4.2, respectively. These are thefirst acidic plant ltps ever reported. These proteinsare rich in alanine, glycine and cysteine residues,and devoid of glutamic acid (except for ltp11 toltp14), tryptophane (except for ltp14) and histidine(except for ltp13), except for ltp9 and ltp15 whichcontain all 20 amino acids. The ArabidopsisATA7 protein [41], which appears to be related toltps, has been added to Table 1. It is slightly largerthan ltps (114 amino acids plus 26 residues for thesignal peptide).

When the nucleotide sequences are compared,the identity between two sequences is always lowerthan 60%, except for the couples ltp1/ltp2 and

V. Arondel et al. / Plant Science 157 (2000) 1–124

ltp4/ltp3 which exhibit 79 and 77% homology,respectively. Because the sequence identity is re-stricted to a smaller area, only ltp1 and ltp2 werefound to cross hybridize when washed at moder-ately stringent conditions (65°C, 0.3 M NaCl).

All genes for which both the gene and thecDNA sequences are available were found to con-tain an intron 9–12 nucleotides upstream from thestop codon. Aside from ltp8 and ltp12 (intron sizeclose to 450 nt), the size of this intron is alwaysclose to 100 nucleotides.

When compared to each other, the amino acidsequences deduced from the ltp genes exhibit from22 to 79% identity and 55 to 90% similarity whenconservative replacements are taken into account(Table 2). ATA7 shares 19–30% identity and 47–56% similarity to Arabidopsis LTPs. Cysteineresidues are conserved in all sequences (Fig. 1).The central region of the protein is also strongly

conserved (residues 42–57) and the residues val 7,leu 11, tyr 17, gly 21, gly 34, leu 38, asp 48, arg 49,leu 77, pro 78 can be found in more than 80% ofthe sequences.

The comparison with other plant mature LTPproteins can be done through a phylogenetic tree(Fig. 2). A large homogenous group (group 1) thatcontains only dicot genes comprises 32 membersfrom Nic.ta1 to Pha.vu. Within group 1, severalclades can be distinguished that follows botanicalclassification. This is obvious for Cruciferae andSolanaceae. A second group (Group 2) appearsmore heterogeneous: it comprises six ArabidopsisLTP proteins (plus ATA7 [41]), two castor bean,two tobacco, two rice and one pine tree proteins.Monocotyledons can be considered to belong to athird group (Group 3), which includes a barleyclade which seems to represent a transition ofsome sort with group 1.

Table 1Characteristics of mature Arabidopsis ltps deduced from cDNA sequencesa

MW (kDa)Amino acid numberGroup (TIGR)Intron size pIBACESTGene

9.3120 (25+) 93AC005499118F16T7ltp1 9.3TC8103TC9063107AC005499124J8T7 9,4ltp2 (25+) 93 9,4TC14044TC14108

ltp3 (23+) 9286C11T7 9,2 9,1AB016890 95 TC8698ltp4 9,18,8(23+) 89TC9488108AB01689090B4T7

TC14119111 TC10173 (25+) 93 9.9 11,4ltp5 124E14T7 AL133452

TC13850ltp6 7.79,9(19+) 94TC1100288AC012562174N16T7

10.5(25+) 98No 9.7No intron?AC006439No ESTltp7ltp8 AI997024 AC005957 439 No (25+) 91 9 4.9

(28+) 90 9.8ltp9 5.2No EST AC007267 No intron? NoTC53883No BAC (22+) 94 9.9ltp10 9.4?AI998609

246G16T7275H8T7

108AL035678 No 8.5ATTS5582 9.3(28+) 89ltp117.7No 9.7(25+) 93464AL133452K2B3T7ltp12

(27+)92ltp13 No?AB011475 8.510.6No EST?AB016880No EST Noltp14 (25+) 91 9.6 4.2

11No EST 7.5AF076275 ? No (22+) 96ltp15(26+) 114 12.7 6.5ATA7 AF037589 AC006439 2 introns

a For each gene, the expressed sequence tag (EST) sequenced is mentioned and the accession number of the bacterial artificialchromosome (BAC) indicated. The size of the intron typical of lipid transfer proteins (LTPs) — which is localized at the 3% endof the open reading frame — is indicated in base pairs; question marks indicate the lack of either gene or cDNA sequenceinformation which cannot allow formal identification of introns. The tentative consensus (TC) number(s) given by The Institutefor Genomic Research, TIGR [44]) is indicated. The number of amino acids corresponds to the mature protein after signalsequence (number of amino acids in parentheses) removal. The most probable cleavage site was determined according to [40].Molecular weight (MW) and isoelectric point (pI) determinations were performed based on the mature protein sequences.Sequence data relative to ltp7 to ltp15 and ATA7 were retrieved from GenBank.

V.

Arondel

etal./

Plant

Science

157(2000)

1–

125

Table 2Comparison of amino acid sequences of Arabidopsis ltpsa

Ltp5ltp1 ltp6 ltp7 ltp8 ltp9 ltp10 ltp11 ltp12 ltp13 ltp14 ltp15ltp2 ltp3 ltp4

25.2 (51.3) 26.3 (47.4) 27.2 (50.8) 29.8 (53.5) 20 (42.6) 23.1 (51.3) 19 (47.4) 22.6 (52.2) 20.2 (49.1) 28.9 (50) 22.4 (47.4) 20.2 (48.2)22.6 (44.3) 26.7 (56)ATA7 23.9 (52.9)39 (66) 55.6 (76.8) 33.3 (69.9) 26.3 (58.9) 42.6 (79.8) 37.6 (63.4) 44.8 (76)64 (81) 25 (55.9)ltp1 24 (59.1) 19.4 (54.2)54 (78)54 (79)47 (75)

45.4 (72.2) 37.2 (66) 48 (66) 33 (67) 22.3 (55.3) 28.9 (66) 35.2 (57.1) 44.3 (70.1) 21.9 (55.9) 19.8 (61.3) 16.8 (46.9)47.3(73.9)ltp2 49.5(74.7)46.3 (78.7) 48.5 (78.8) 41.3 (79.3) 28.7 (60.6) 46.9 (77.1) 42.4 (67.4) 52.6 (84.2)54.6(79.4) 24.5 (54.8)79.3(90.2) 25.3 (62) 22 (53.1)ltp345.7 (75.3) 51.5 (76.8) 40.2 (78.3) 30.8 (60.6) 41.4 (70.7) 41.3 (65.2) 53.7 (76.8)ltp4 27.1 (57.6) 24.5 (57.1) 22.4 (49)49.5(76.3)36.7 (67) 48.5 (69.7) 34.4 (64.5) 28.1 (56.2) 41 (71.6) 36.6 (60.2) 43.8 (70.8) 24.2 (56.4)ltp5 25 (60.2) 19.4 (53.1)

37 (72) 42.4 (77.2) 27.8 (55.6) 55.3 (81.9) 38.9 (64.2) 41.2 (69.1) 22.9 (48.9)ltp6 26.8 (62.8) 20.4 (53.1)32.7 (68.4) 26.5 (57.1) 38.4 (72.7) 36 (60) 42.6 (68.3) 24.2 (51) 24.2 (53.1) 18.4 (48)ltp7

25 (55.4) 35.7 (71.4) 41.3 (70.6) 33.7 (66.3) 24.2 (54.3) 22.6 (60.4) 17.2 (44.8)ltp825 (54.2) 31.5 (57.6) 25.8 (54.6) 26.9 (56.5) 18.5 (58.2) 22.5 (51)ltp9

33 (57.4) 35.1 (70.1) 19.8 (48.9)ltp10 27.8 (60.6) 16.5 (51)ltp11 38.9 (61) 26.1 (59.8) 23.7 (57.1) 22.3 (55.2)

22.7 (57) 23.5 (62.8) 19.6 (58.3)ltp1227.7 (56.5)ltp13 20 (43.8)

25.2 (58.3)ltp14

a The percentage of identical residues is indicated. The number of similar residues (i.e. when conservative replacements are taken into account) is indicated between parentheses. Ltp1 is from Thoma et al. [18] and ATA7 from Rubinelli et al. [41].

V. Arondel et al. / Plant Science 157 (2000) 1–126

Fig. 1. Sequence alignment of ltp-deduced proteins of Arabidopsis. The probable cleavage site of the signal peptide wasdetermined according to Ref. [40], and only the mature deduced amino acid sequences were aligned using the ClustalW software[36], followed by processing with the EDITALN software. Amino acids that are strictly conserved are indicated by thecorresponding symbol. *, amino acids conserved in 80% of the sequences; + , in 60%; :, in 40%; and ., in 20%.

3.3. Southern blot hybridization analyses

When hybridized and washed at high stringency(68°C, 15 mM NaCl), each probe reveals only oneband on DNA restricted with most enzyme chosen(data not shown). At lower stringency (hybridiza-tion at 50°C, 1.2 M NaCl, washes at 50°C, 0.3 MNaCl), several additional bands can be detected,depending on the probe and the restriction enzymeused. Ltp1 and ltp2 exhibit a similar pattern ofhybridization. This is due to the strong nucleotidesimilarities between the two genes combined withthe fact that both genes are tandemly repeated onchromosome 2 (see below). The same holds true inthe case of ltp3 and ltp4 which locate to chromo-some 5. The ltp1/2 family comprises two strongbands plus up to five additional weak bands de-pending on the enzyme and the ltp3/4 family onestrong band plus up to four additional weakbands. Ltp5 exhibits usually one to two bandswhile ltp6 detects one to two strong bands and upto four additional weak ones. This suggests thatltp1 and ltp2 represent a small subfamily of two tothree genes, ltp3 and ltp4 a subfamily of two

genes, ltp5 and ltp6 two subfamilies of one or twogenes each. The weakly hybridizing bands (10–12)might correspond to other ltps or to unrelatedgenes (Fig. 3).

3.4. Gene expression studies

Northern analyses where carried out on totalRNA isolated from different tissues. All geneswere found to be highly expressed in flowers, andthe transcript can always be detected in siliques.Virtually no RNA can be detected in roots, whilethe expression of the genes varies in leaves. Onlyltp1, ltp2 and ltp5 are significantly expressed inleaves. While the level of ltp5 mRNA seems toremain constant during the development of theleaf up to its senescence, the expression of ltp1 ismaximum in young leaves and decreases with timeafter bolting. Ltp1 mRNA is barely detected atday 45, when the leaves start senescing. Ltp 6 isslightly expressed during the first 24 days aftergermination (Fig. 4).

ABA was found to induce strongly the expres-sion of ltp4 and ltp3 in fully expanded leaves,

V. Arondel et al. / Plant Science 157 (2000) 1–12 7

while it does not alter significantly the expressionof the other ltp genes.

Because there might be some cross reactionsbetween the entire cDNA probes, the same experi-ments were carried out using specific oligonucle-otides, and identical results were obtained (datanot shown).

3.5. Mapping experiments

Ltp4 and ltp2 genes were found to detect anRFLP between Columbia and Landsberg usingEcoRV as restriction enzyme. They were mappedusing 90 recombinant imbred lines, and found tolocate on chromosome 5, cosegregating withmarker m211A (ltp4) and on chromosome 2, be-tween markers m323 (2.2 cM) and m529 (3.4 cM)(ltp2). Ltp1, ltp5 and ltp6 were found to hybridizeto CIC YACs 3G1 and 2G9 for ltp1, 8E1, 7A4,10B4, 9C9, 9D9, 10A11, 11G7 and 6F4 for ltp5,1C12, 10B10 and 11D1 for ltp6. These data indi-cate that ltp1 maps to chromosome 2 (close toposition 71cM), ltp5 to chromosome 3 (close to

Fig. 2.

Fig. 2. Phylogenetic tree for plant ltps. The phylogenetic treewas built using the protein sequences indexed in GenBank,after removal of the signal peptide [40]. The neighbor joiningmethod/UPGMA version 3.573c from the PHYLIP package[37] was used, as indicated in Section 2. Sequences are men-tioned by six first letters of the Latin name of the plant fromwhich they were obtained followed by a number when therewere several lipid transfer protein (LTP) in a same species.Their accession numbers in GenBank are: Bras.na: Brassicanapus 1: X60318 2: U22175 3:U22105 4:U22174-Bras.ra:Brassica rapa L31938-Bras.ole: Brassica oleracea 1: L33904 2:L33905 3: L33906 4: L33907-Hor.vu: Hordeum 6ulgare 1:U18127 2: Z37114 3: X68656 4: X68654 5: X96979 6: Z66529,Z66528, U63993 7: U88090 8: X60292, X59253 9 Z3715-Tri.du: Triticum durum X63669-Ory.sa: Oryza sati6a 1:U16721 2: U29176 3:D15364 4:D22795 5: D16036 6:U291767: U29176 8: X83433 9: X83434 10: Z23271 11: X83435 12:D15678- Zea: Zea mays 1: U66105 2: J04176-Sor.bi: Sorghumbicolor 1: X71667 2: X71668 3: X71669-Nic.ta: Nicotianatabacum 1: D13952, 2: U14167, 3:U14168, 4: X62395-Ric. co:Ricinus communis 1: M86353 2: D11077-All.ce: Allium cepaS79815-Pha.vu: Phaseolus 6ulgaris : U72765-Pin.ta: Pinustaeda U10432-Pru.du: Prunus dulcis 1: X96714 2:X96716-Ger.hy: Gerbera hybrida Z31588-Hel. an: Helianthus annuusX92648-Dau.ca: Daucus carota M64746-Gos.hi: Gossypiumhirsutum 1: U64874 2:U15153-Lyc.pe: Lycopersicon pennellii1: U66466 2: U66465-Lyc.es: Lycopersicon esculentumU81996-Spi.ol: Spinacia oleracea M58635. Arabidopsisthaliana clones are ATA7 (AF037589), ltp1, ltp4, ltp3, ltp4,ltp2, ltp6. Ltp 10 aminoacid sequence is deduced from ex-pressed sequence tags (ESTs) and ltp7, ltp8, ltp9, ltp11, ltp12,ltp13, ltp14 and ltp15 are deduced from whole BAC se-quences.

V. Arondel et al. / Plant Science 157 (2000) 1–128

position 72) and ltp6 to chromosome 3. Ltp 1 and2 appear to be close together on chromosome 2,ltp4 and ltp3 on chromosome 5. Based on genomicsequencing and mapping data available from theTAIR database, ltp7, ltp8 and ltp9 map to thesame position (29 cM) on chromosome 2, whileltp12 is clustered with ltp5. Ltp11 and ltp15 lo-cates to chromosome 4 (89 and 31 cM, respec-tively), ltp13 and ltp14 to chromosome 5 (94 and117 cM, respectively) and the map position ofltp10 remains unknown (Fig. 5).

4. Discussion

Lipid transfer proteins are an ubiquitous proteinfamily in higher plants, whose biological functionremains unknown. One of the problems encoun-tered in studying LTPs is the number of isoformesthat can be detected. For instance, more than tengenes have been described in rice [17]. Arabidopsisis the most suitable organism for obtaining anexhaustive collection of ltp isoformes, because thesmall size of its genome suggests that genesfamilies are likely to contain few members. Themain reason for using Arabidopsis for such astudy is that an important part of its genome hasbeen already sequenced, and that many ESTs areavailable. ESTs are particularly well suited forlooking for LTPs since these proteins are shortand almost half of the amino acid sequence isincluded in an average EST. The other reason isthat LTPs are frequently expressed at high levels,and their mRNA represents usually a few percentof a plant total mRNAs. They are therefore wellrepresented in cDNA libraries. This was found tobe the case for ltp1 to ltp6, while ltp8, ltp9 to ltp12are represented by one to three ESTs. No ESTcould be found for the other genes.

A total of 14 ltp genes could be evidenced inArabidopsis based on 80% of the genome sequenceand ten through EST analysis, for a total of 15different genes. It is therefore likely that the Ara-bidopsis 9 kDa ltp gene family contains about15–20 genes. However, sequencing errors mighthamper the discovery of additional genes. Forinstance, a couple of nucleotide modificationsshould be enough to detect another gene on bacte-rial artificial chromosome (BAC) T1N24(AF149413). The Southern blot hybridization datasuggest that ltp genes consist in small subfamiliesof one to three genes which may, or may not,weakly cross hybridize at low stringency. Clarkand Bohnert [38] have recently characterized threecDNAs (ltp1 to ltp3) In A. thaliana (Wassilewskijaecotype) and the corresponding ltp1 and ltp2genes [16,39]. These three members of the LTPfamily are similar to those presented in the studyalthough performed in a different ecotype. Theseauthors have also noted that there was very littlecrosshybridization, even between ltp1 and ltp2.Therefore, Southern genomic DNA analyses arelikely to underestimate the number of ltp genes.

Fig. 3. Southern blot hybridization analysis of ltp gene familyin Arabidopsis. Columbia genomic DNA (1 mg per lane) wasrestricted during 4 h with 5 U of the restriction endonucleasesBamHI (1), BglII (2), ClaI (3), EcoRI (4), EcoRV (5),HindIII (6), XbaI (7) and XhoI (10). The DNA was fraction-ated on agarose gel, transferred to nylon membrane, andprobed with 32P-labeled cDNA probes. The hybridizationswere carried out at 50°C in 0.6 M NaCl, and the washes wereperformed in 2×SSC, 0.1% SDS at 50°C. Membranes whereexposed to X-ray films between intensifying screens. Themolecular weight markers were Lambda DNA restricted byHindIII; they are indicated by stars (from top to bottom, inkilobase pairs: 23.1; 9.4; 6.5; 4.4; 2.3; 2.0; 0.56) on the left sideof each membrane.

V. Arondel et al. / Plant Science 157 (2000) 1–12 9

Fig. 4. Northern blot hybridization analysis of ltp gene expression. Total RNA (10 mg) were fractionated on 1.5% formaldehyde-agarose gels, transferred to a nylon membrane and probed with cDNA inserts labeled by random priming using [32P]dCTP.Hybridizations were carried out at 45°C in 50% formamide. Washes were performed at 62°C in 2×SSC 0.1% SDS. A Rab18probe was used as a positive control for abscisic acid (ABA) induction, and a ribosomal DNA probe (rib) to assess for evenloading of the RNA. RNAs were from 40-day-old plants for organ specific expression studies (A), from 28-day-old leaves forABA induction (C), and from leaves extracted at different stages of development (B). The induction by ABA was carried out in10−4 M for 24 or 48 h. L, leaves; S, siliques; F, flowers; R, roots.

Mapping data indicate that, although ltp genescan be found scattered through four chromosomesof Arabidopsis, four clusters exist that comprisenine genes. Tandem repeats can be noted based ongenomic sequences (AC005499, AB016890,AL133452) for ltp5-12, ltp4-3 and ltp1-2 [38]. Thetwo last pairs exhibit more than 75% identity atthe nucleic acid level. Taken together, these datastrongly suggest that these are duplicated genes.Other pairs of tandemly repeated LTP genes have

been characterized in various plants, such as ltp1and ltp2 from Sorghum 6ulgare [42], Wax 9D andWax 9C from Brassica oleracea [21] and LTP 4.2and LTP4.3 in barley [43]. Another characteristicof those genes is the existence of an intron locatedat the end of the open reading frame. This exis-tence is demonstrated for all the eight genes forwhich both cDNA and gene sequences are avail-able. This intron appears to be a general feature ofltp genes in plants.

V. Arondel et al. / Plant Science 157 (2000) 1–1210

The RNA hybridization analysis of all genestested indicate that transcripts are alwayspresent in flower and siliques. The main differ-ences between the genes is their pattern of expres-sion in leaves, which varies especially withregards to development. Clark and Bohnert [39]also noted that ltp1, ltp2 and ltp3 are highly

expressed in flowers and siliques. However, it hasbeen found that ltp2 is also expressed in leaves. Alltranscripts tested were undetectable in roots,which is also a general feature of plant ltps. It hasbeen observed that ltp4 and ltp3 from Arabidopsisare up-regulated by ABA. A similar result hasbeen obtained for LTP genes from other plantssuch as rice [17] and rapeseed [14]. Interestingly,the promoter region of the ltp4 gene from barley[43] contains elements which could be linked toABA responsiveness. In Arabidopsis [16] the pres-ence of several expression-controlling motifs inltp1 gene, also found in ltp2 gene by [39], similarto those previously reported in plant genes in-duced by various types of stress or by pathogenattack was observed.

The phylogenetic tree is very similar to thosealready published [17,39]. The group I defined byRef. [17] corresponds to the groups I and II, whilegroup III corresponds to group II by Vignols et al.[17]. As a control all rice sequences published bythese authors have been included and their differ-ent classes can be found at the same position inthis tree. Concerning dicots it seems plausible thattwo different ancestral genes exist: one corre-sponding to group I and the other one to groupIII. Interestingly, Arabidopsis genes arenot far from being equally represented in bothgroups (nine genes in group I versus six genes ingroup III). Monocots and Gymnosperms are alsorepresented in group III, so it is likely that theputative ancestral gene to this group has started todifferentiate early in higher plant evolution. Con-cerning group II, the barley subgroup appears tobe much more closely related to group I than togroup III. The situation appears to be different forthe other monocots of group II. This discrepancymight be due to the software used, and the factthat this tree is unrooted. In any case, the exis-tence of at least two different groups appearsclearly.

It has been suggested by Ref. [39] that theancestor to Brassicaceae possessed already severalcopies of ltp genes. The phyletic analysis suggestthat this ancestor might have possessed no lessthan five copies, A (precursor to ltp9, 13–15), B(precursor to ltp8, 11), C (precursor to ltp6, 10), D(precursor to ltp1, 2, 5, 7) and E (precursor toltp3, 4, 12). The closeness of genes such as ltp3and ltp4 suggests that this family of gene seems tocontinue to duplicate.

Fig. 5. Map position of ltp genes in Arabidopsis. Ltp1 andltp2 were found to hybridize to yeast artificial chromosomes(YACs) CIC 3G1 and 2G9, which map to chromosome 2(position 71 cM) close to restriction fragment length polymor-phism (RFLP) markers m429 and ve018. These two genes aretandemly repeated on bacterial artificial chromosome (BAC)T6A23 (AC005499). Ltp3 and ltp4 map to chromosome 5(position 111 cM) close to RFLP markers hst and agp50.They are tandemly repeated on bac MNC17 (AB016890).Ltp5 hybridizes to YACs CIC 8E1, 7A4, 10B4, 9C9, 9D9,10A11, 11G7 and 6F4. It locates to chromosome 3 (position71 cM), together with ltp12, close to markers AtEm1 andm457. ltp6 maps to chromosome 3 (position 29 cM), close tomarkers g4523 and mi357. Ltp7, ltp8 and ATA7 [41] arecontained in BACs T30D6 (AC006439), T15J14 (AC005957),ATF 20O9 (AL021749), respectively, which map to chromo-some 2 (position 29.2 cM), chromosome 2 (position 29.2 cM)and chromosome 4 (position 73.9), respectively. Ltp7 and ltp8do not appear to be tandemly repeated. Ltp 9 is at the locusAC007267 (chromosome 2 29.2 cM), ltp11 in BAC ATF17M5(AL035678, chromosome 4, 89cM), ltp13 in BAC K9L2(AB011475, chromosome 5, 94 cM), ltp14 in BAC MTG10(AB016880, chromosome 5, 117 cM) and ltp15 in BACT15F16 (AF076275, chromosome 4, 31 cM).

V. Arondel et al. / Plant Science 157 (2000) 1–12 11

5. Conclusion

Fifteen genes have been identified through45 000 ESTs and 102 megabases of genomicDNA. Although one cannot exclude that other ltpgenes exist in Arabidopsis, it is very likely that alarge majority of them have been identified [44].This will permit one to initiate a genetic approachfor determining the biological function of LTPs.The important number of genes makes an anti-sense approach difficult to carry out efficiently.Alternatively, it is possible to search for disruptedmutants. The availability of DNA sequence shouldallow a search by PCR for T-DNA tagged mu-tants. In addition, knowing the map position ofthese genes will permit to look for transposon-tagged mutants.

Acknowledgements

Part of the work presented in this article hasbeen funded by the GREG program 520 721. Wethank Dr D. Bouchez (INRA Versailles) forproviding us with the CIC YAC library, Dr M.Anderson (NASC, Nothingham) for Dr Dean’srecombinant inbred lines and the ABRC (OSU,USA) for the Arabidopsis EST clones. We aregrateful to Dr D. Bouchez for providing us withthe information concerning YAC anchoring andto Dr Sean May (Nothingham) for computingsegregation data. We thank Natalie Ferte for helpwith editing figures. We are grateful to Dr F.Grellet for in-depth critical reading of themanuscript and helpful advice concerning se-quence analysis. We are much indebted to Dr A.Zachowski for helpful suggestions concerning thepresentation of this manuscript.

References

[1] K.W.A. Wirtz, Phospholipid transfer, Annu. Rev.Biochem. 60 (1991) 73–99.

[2] K.W.A. Wirtz, Phospholipid transfer proteins revisited,Biochem. J. 324 (1997) 353–360.

[3] N. Jouannic, M. Lepetit, C. Vergnolle, C. Cantrel, A.M.Gardies, J.-C. Kader, V. Arondel, Isolation of a cDNAfrom Arabidopsis thaliana that complements the sec14mutant of yeast, Eur. J. Biochem. 258 (1998) 402–410.

[4] J.C. Kader, Lipid transfer proteins in plants, Ann. Rev.Plant Physiol. Plant Mol. Biol. 47 (1996) 627–654.

[5] J.-C. Kader, Lipid-transfer proteins: a puzzling family ofplant proteins, Trends Plant Sci. 2 (1997) 66–70.

[6] A. Zachowski, F. Guerbette, M. Grosbois, A. Jolliot-Croquin, J.C. Kader, Characterization of acyl-binding bya plant lipid-transfer protein, Eur. J. Biochem. 257(1998) 443–448.

[7] J.Y. Lee, K. Min, H. Cha, D.H. Shin, K.Y. Hwang,S.W. Suh, Rice non-specific lipid transfer protein: the 1.6A, crystal structure in the unliganded state reveals a smallhydrophobic cavity, J. Mol. Biol. 276 (1998) 437–448.

[8] J. Gomar, P. Sodano, D. Sy, D.H. Shin, J.Y. Lee, S.W.Suh, D. Marion, F. Vovelle, M. Ptak, Comparison ofsolution and crystal structures of maize nonspecific lipidtransfer protein: a model for a potential in vivo lipidcarrier protein, Proteins 31 (1998) 160–171.

[9] L. Sossountzov, L. Ruiz-Avila, F. Vignols, A. Jolliot, V.Arondel, F. Tchang, M. Grosbois, F. Guerbette, E.Miginiac, M. Delseny, et al., Spatial and temporal ex-pression of a maize lipid transfer protein gene, Plant Cell3 (1991) 923–933.

[10] A.J. Fleming, T. Mandel, S. Fofmann, P. Sterk, S.C. deVries, C. Kuhlemeier, Expression pattern of a tobaccolipid transfer protein gene within the shoot apex, Plant J.2 (1992) 855–862.

[11] K. Gausing, Lipid transfer protein genes specifically ex-pressed in barley leaves and coleoptiles, Planta 192(1994) 574–580.

[12] C.S. Kinlaw, S.M. Gerttulla, M.C. Carter, Lipid transferprotein genes of loblolly pine are members of a complexgene family, Plant Mol. Biol. 26 (1994) 1213–1216.

[13] Kotilainen, Y. Helariutta, P. Elomaa, L. Paulin, T.H.Teeri, A corolla- and carpel-abundant, non-specific lipidtransfer protein gene is expressed in the epidermis andparenchyma of Gerbera hybrida var Regina (Composi-tae). Plant Mol. Biol. 26 (1994) 971–978.

[14] I. Soufleri, C. Vergnolle, E. Miginiac, J.-C. Kader, Ger-mination-specific lipid transfer protein cDNAs in Bras-sica napus L., Planta 199 (1996) 229–237.

[15] P. Sterk, H. Booij, G.A. Scheleekens, A. Van Kammen,S.C. de Vries, Cell-specific expression of the carrot EP2lipid transfer protein gene, Plant Cell 3 (1991) 907–921.

[16] S. Thoma, U. Hecht, A. Kippers, J. Botella, S. De Vries,C.R. Somerville, Tissue-specific expression of a geneencoding a cell wall-localized lipid transfer protein fromArabidopsis, Plant Physiol. 105 (1994) 35–45.

[17] F. Vignols, M. Wigger, J.M. Garcia-Garrido, F. Grellet,J.C. Kader, M. Delseny, Rice lipid transfer protein(LTP) genes belong to a complex multigene family andare differentially regulated, Gene 195 (1997) 177–186.

[18] S. Thoma, Y. Kaneko, C. Somerville, A non-specificlipid transfer protein from Arabidopsis thaliana is a cellwall protein, Plant J. 3 (1993) 427–436.

[19] P. Coutos-Thevenot, T. Jouenne, O. Maes, F. Guerbette,M. Grosbois, et al., Four 9-kDa proteins excreted bysomatic embryos of grapevine are isoforms of lipid-trans-fer proteins, Eur. J. Biochem. 217 (1993) 885–889.

[20] D.P. Ma, H. Tan, Y. Si, R.G. Creech, J.N. Jenkins,Differential expression of a lipid transfer protein gene incotton fiber, Biochim. Biophys. Acta 1257 (1995) 81–84.

V. Arondel et al. / Plant Science 157 (2000) 1–1212

[21] J. Pyee, P.E. Kolattukudy, The gene for the majorcuticular wax-associated protein and three homologousgenes from broccoli (Brassica oleacera) and their expres-sion patterns, Plant J. 7 (1995) 49–59.

[22] B.P.A. Cammue, K. Thevissen, M. Hendriks, K. Egger-mont, I.J. Goderis, P. Proost, J. Van Damme, R.W.Osborn, F. Guerbette, J.C. Kader, W.F. Broekaert, Apotent antimicrobial protein from onion seeds showingsequence homology to lipid transfer proteins, Plant Phys-iol. 109 (1995) 445–455.

[23] F. Garcia-Olmedo, A. Molina, A. Segura, M. Moreno,The defensive role of non-specific lipid-transfer proteinsin plants, Trends Microbiol. 3 (1995) 72–74.

[24] A. Molina, F. Garcia-Olmedo, Developmental andpathogen induced expression of three barley genes encod-ing lipid transfer proteins, Plant J. 4 (1993) 983–991.

[25] A. Segura, M. Moreno, F. Garcia-Olmedo, Purificationand antipathogenic activity of lipid transfer proteinsfrom the leaves of Arabidopsis and spinach, FEBS Lett.332 (1993) 243–246.

[26] A. Molina, F. Garcia-Olmedo, Enhanced tolerance tobacterial pathogens caused by the transgenic expressionof barley lipid transfer LTP4, Plant J. 12 (1997) 669–675.

[27] N. Bechtold, J. Ellis, G. Pelletier, In planta Agrobac-terium mediated gene transfer by infiltration of adultArabidopsis thaliana plants, CR Acad. Sci. Paris/Life Sci.316 (1993) 1194–1199.

[28] D. Bouchez, C. Camilleri, M. Caboche, A binary vectorbased on Basta resistance for in planta transformation ofArabidopsis thaliana, CR Acad. Sci. Paris/Life Sci. 316(1993) 1188–1193.

[29] M. Bevan, I. Bancroft, E. Bent, K. Love, H. Goodman,C. Dean, R. Bergkamp, W. Dirkse, M. Van Staveren, W.Stiekema, L. Drost, P. Ridley, S.A. Hudson, K. Patel, G.Murphy, P. Piffanelli, H. Wedler, E. Wedler, R.Wambutt, T. Weitzenegger, T.M. Pohl, N. Terryn, J.Gielen, R. Villarroel, N. Chalwatzis, et al., Analysis of a1.9 Mb of contiguous sequence from chromosome 4 ofArabidopsis thaliana, Nature 29 (391) (1998) 485–488.

[30] R. Cooke, M. Raynal, M. Laudie, F. Grellet, M.Delseny, P.C. Morris, D. Guerrier, J. Giraudat, F.Quigley, G. Clabault, Y.F. Li, R. Mache, M. Krivitzky,I.J. Gy, M. Kreis, A. Lecharny, Y. Parmentier, J. Mar-bach, J. Fleck, B. Clement, G. Philipps, C. Herve, C.Bardet, D. Tremousaygue, H. Hofte, et al., Furtherprogress towards a catalogue of all Arabidopsis genes:analysis of a set of 5000 non-redundant ESTs, Plant J. 9(1996) 101–124.

[31] C. Camilleri, J. Lafleuriel, C. Macadre, F. Varoquaux,Y. Parmentier, G. Picard, M. Caboche, D. Bouchez, A

YAC contig map of Arabidopsis thaliana chromosome 3,Plant J. 14 (1998) 633–642.

[32] S.I. Gibson, C.R. Somerville, in: C. Koncz, N.H. Chua,J. Schell (Eds.), Methods in Arabidopsis Research,World Scientific, Singapore, 1992, pp. 119–143.

[33] F. Sanger, S. Niecklen, A.R. Coulson, DNA sequencingwith chain terminating inhibitors, Proc. Natl. Acad. Sci.USA 74 (1977) 5463–5467.

[34] J. Sambrook, E.F. Fritsch, T. Maniatis, MolecularCloning. A Laboratory Manual, 2nd edition, ColdSpring Harbor Laboratory, Cold Spring Harbor, NY,1989.

[35] S.F. Altschul, W. Gish, W. Miller, E.W. Myers, D.J.Lipman, Basic local search alignment tool, J. Mol. Biol.215 (1990) 403–410.

[36] J.D. Thompson, D.G. Higgins, T.J. Gibson, CLUSTALW: improving the sensitivity of progressive multiple se-quence alignment through sequence weighting, positions-specific gap penalties and weight matrix choice, NucleicAcids Res. 22 (1994) 4673–4680.

[37] J. Felsenstein, PHYLIP — Phylogeny Inference Package(Version 3.2), Cladistics 5 (1989) 164–166.

[38] R. Kalla, K. Shimamoto, R. Potter, P.S. Nielsen, C.Linnestad, O.A. Olsen, The promoter of the barleyaleurone-specific gene encoding a putative 7 kDa lipidtransfer protein confers aleurone cell-specific expressionin transgenic rice, Plant J. 6 (1994) 849–860.

[39] A.M. Clark, H.J. Bohnert, Cell-specific expression ofgenes of the lipid transfer protein family from Arabidop-sis thaliana, Plant Cell Physiol. 40 (1999) 69–76.

[40] G. Von Heijne, Signal sequences: the limits of variation,J. Mol. Biol. 184 (1985) 99–105.

[41] P. Rubinelli, Y. Hu, H. Ma, Identification, sequenceanalysis and expression studies of novel anther-specificgenes of Arabidopsis thaliana, Plant Mol. Biol. 37 (1998)607–619.

[42] J. Pelese-Siebenbourg, C. Caellas, J.C. Kader, M.Delseny, P. Puigdomenech, A pair of genes coding forlipid-transfer proteins in Sorghum 6ulgare, Gene 148(1994) 305–308.

[43] A. Molina, I. Diaz, I.K. Vasil, P. Carbonero, F. Garcia-Olmedo, Two cold-inducible genes encoding lipid trans-fer protein LTP4 from barley show differential responsesto bacterial pathogens, Mol. Gen. Genet. 252 (1996)162–168.

[44] S.D. Rounsley, A. Glodek, G. Sutton, M.D. Adams,C.R. Somerville, J.C. Venter, A.R. Kerlavage, The con-struction of Arabidopsis expressed sequence tag assem-blies. A new resource to facilitate gene identification,Plant Physiol. 112 (1996) 1177–1183.

.