Available online at www.sciencedirect.com

www.elsevier.com/locate/phytochem

Phytochemistry 69 (2008) 1641–1652

PHYTOCHEMISTRY

Isolation and functional analysis of two Cistus creticus cDNAsencoding geranylgeranyl diphosphate synthase

Irene Pateraki, Angelos K. Kanellis *

Group of Biotechnology of Pharmaceutical Plants, Laboratory of Pharmacognosy, Department of Pharmaceutical Sciences,

Aristotle University of Thessaloniki, 541 24 Thessaloniki, Greece

Received 18 December 2007; received in revised form 4 February 2008Available online 9 April 2008

Abstract

Cistus creticus ssp. creticus is an indigenous shrub of the Mediterranean area. The glandular trichomes covering its leaf surfacessecrete a resin called ‘‘ladanum”, which among others contains a number of specific labdane-type diterpenes that exhibit antibacterialand antifungal action as well as in vitro and in vivo cytotoxic and cytostatic activity against human cancer cell lines. In view of the prop-erties and possible future exploitation of these metabolites, it was deemed necessary to study the geranylgeranyl diphosphate synthaseenzyme (GGDPS, EC 2.5.1.30), a short chain prenyltransferase responsible for the synthesis of the precursor molecule of all diterpenes.In this work, we present the cloning, functional characterisation and expression profile at the gene and protein levels of two differentiallyexpressed C. creticus full-length cDNAs, CcGGDPS1 and CcGGDPS2. Heterologous yeast cell expression system showed that thesecDNAs exhibited GGDPS enzyme activity. Gene and protein expression analyses suggest that this enzyme is developmentally and tis-sue-regulated showing maximum expression in trichomes and smallest leaves (0.5–1.0 cm). This work is the first attempt to study theterpenoid biosynthesis at the molecular level in C. creticus ssp. creticus.� 2008 Elsevier Ltd. All rights reserved.

Keywords: Cistus creticus ssp. creticus; Secondary metabolism; Terpene biosynthesis; Labdane diterpenes; Geranylgeranyl diphosphate synthase (GG-DPS)

1. Introduction

Cistus creticus ssp. creticus is a perennial, Mediterraneanshrub, found mainly in degraded areas like maquis and gar-igues ecosystems. Its leaf surface is covered with nonglandular (stellate) and glandular, long, ball-headed,multi-cellular trichomes (Gulz et al., 1996). The resin or‘‘ladanum” excreted from these glands was known and usedsince ancient times for its officinal and aromatic properties(Demetzos et al., 1990; Gulz et al., 1996). These traditionalapplications were corroborated by recent findings showingthat resin’s labdane-type components such as ent-3b-acet-oxy-13-epi-manoyl oxide, ent-13-epi-manoyl oxide, (13E)-

0031-9422/$ - see front matter � 2008 Elsevier Ltd. All rights reserved.

doi:10.1016/j.phytochem.2008.02.005

* Corresponding author. Tel.: +30 2310 997656; fax: +30 2310 997662/997645.

E-mail address: kanellis@pharm.auth.gr (A.K. Kanellis).

labd-13-ene-8a,15-diol, (13E)-labd-7,13-diene-15-ol, (13E)-labd-7,13-diene-15-yl acetate, (13E)-labd-7,13-dienol,sclareol [(13R)-labd-14-ene-8,13-diol] or manoyl oxidederivatives, in free form or incorporated in liposomes,exhibited strong in vitro cytotoxic and cytostatic activityagainst human cancer lines (Demetzos et al., 1994; Dimaset al., 1998, 1999, 2001, 2006; Matsingou et al., 2005,2006). Further, sclareol in liposomal formulation reducedthe growth rate of human colon cancer tumors (HCT116)developed in SCID mice (xenografts) via apoptosis and cellcycle arrest without any significant side effects (Dimas et al.,2007; Hatziantoniou et al., 2006). The above findings pro-pose that sclareol or other labdane-type diterpenes incorpo-rated into liposomes may possibly evolve into novelchemotherapeutic agents for the treatment of several typesof human cancer. Within this context, an insight into themolecular mechanism governing the labdane diterpenes

1642 I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652

biosynthesis in C. creticus would be of great interest, inorder to make possible future applications and increasedproduction of these metabolites.

Diterpenes are produced from the condensation ofdimethylallyl diphosphate (DMADP) with three units ofisopentenyl diphosphate (IPP) (Koepp et al., 1995;Liang et al., 2002; Ogura and Koyama, 1998; Trapp andCroteau, 2001). The acyclic molecule (C20) produced,GGDP (geranylgeranyl diphosphate), undergoes a rangeof cyclisations to produce the parent skeletons of allditerpenoids. These reactions are catalyzed by diterpenesynthases (or cyclases) and are followed by a variety ofmodifications of the parent skeleton to produce the manythousands of different plant-origin diterpenes (Bohlmannet al., 1998; Pichersky and Gang, 2000; Trapp and Croteau,2001). Many metabolites, playing essential roles in plantphysiology such as carotenoids, chlorophyll and tocoph-erol side chains, phytol, the plant hormones gibberellinand abscisic acid are synthesized through the diterpenoidbiosynthetic pathway.

GGDPS (geranylgeranyl diphosphate synthase), is anhomodimeric, short chain trans-prenyltransferase responsi-ble for the synthesis of GGDP, and occurs nearly ubiqui-tously in plants, animals and bacteria (Ogura andKoyama, 1998). It is a branch point enzyme, which regu-lates in coordination with the other prenyltransferases(GDP and FDP synthase respectively) the precursor fluxtowards mono-, sesqui-, and diterpenoids production (Hef-ner et al., 1998; Laskaris et al., 2000). Consequently, thismay affect the ratio of the produced mono-/sesqui-/di-terp-enes. Thus, in view of the key role of this enzyme in diter-penes biosynthesis, the isolation and characterization of itscDNAs deserves further attention.

GGDPS cDNAs have been cloned and characterizedfrom several plant species [e.g. Arabidopsis thaliana (Okadaet al., 2000), Taxus canadensis (Hefner et al., 1998), Scopa-

ria dulcis, Croton sublyratus (Kojima et al., 2000), Hevea

brasiliensis (Takaya et al., 2003), Helianthus annuus (Ohet al., 2000) or Solanum lycopersicon (Ament et al.,2006)]. GGDPS seems to be encoded by a multigene familyin many plant species studied until today. This is obviousfrom the submissions in the nucleotide (GenBank,EMBL-Bank, RefSeq, PDB) and protein (SwissProt,PIR, PRF, PDB, UniProt, InterPro) sequence databases.In tomato, for example, three different cDNAs have beensequenced until today (accession numbers: DQ159949,Q1A7T0, Q1A7S9), in Adonis palaestina three (accessionnumbers: AY661706, AY661707 and AY661708), in Oryza

sativa three more (accession numbers: AK120768, Q6ET88and Q9XHX1), while in Daucus carota only two (accessionnumbers: AB027705 and DQ192185). In Arabidopsis, basedon the genome sequence, there are 12 putative GGDPSgenes (Lange and Ghassemian, 2003). Arabidopsis andtomato are the only plant species where this multigene fam-ily has been studied and the differential expression, as wellas the subcellular localization of specific members has beenreported (Ament et al., 2006; Okada et al., 2000).

To investigate the early steps in the biosynthesis of diter-penes and their regulation in C. creticus ssp. creticus plant,two cDNAs encoding GGDPS (CcGGDPS1 andCcGGDPS2) were cloned and functionally characterized.Moreover, the levels of the transcripts and peptide accumu-lation were studied in several plant tissues. The resultsindicate that this enzyme is developmentally and tissue-reg-ulated showing maximum expression in trichomes and insmallest leaves (0.5–1.0 cm).

2. Results and discussion

2.1. Cloning of two full-length CcGGDPS cDNAs

Plant GGDPSs are encoded by multigene families whilethe genes that comprise these families seem to be differen-tially regulated and localized in different cellular compart-ments, like endoplasmic reticulum, plastids ormitochondria (Ament et al., 2006; Okada et al., 2000). Inspite of the extended studies of the GGDPS genes in manyplant species, only few of the isolated cDNAs have beenfunctionally characterized.

Two cDNAs corresponding to C. creticus GGDPSgenes, according to sequence comparison, CcGGDPS1

and CcGGDPS2, were isolated using 50 and 30 RACE tech-niques (accession numbers: AF492022 and AF492023,respectively). CcGGDPS1 full-length cDNA has a size of1456 bp and encodes a deduced protein of 363 amino acids.It contains a 50 untranslated region (UTR) of 167 bp and a30 UTR of 195 bp. The CcGGDPS2 full-length cDNA has asize of 1311 bp and encodes a deduced protein of 369amino acids. The 50 UTR is 115 bp while the 30 UTR hasa size of 86 bp. A putative polyadenylation signal wasfound only in the CcGGDPS2 (AATGAA), 23 nucleotidesupstream of the polyA, while no polyA signal was found inthe CcGGDPS1 cDNA, as is the case with many plant tran-scripts (Graber et al., 1999). Pairwise comparisons of theputative CcGGDPS peptides with other known GGDPSrevealed that they are highly similar with other plantGGDPS in a range of 73–53%, while the similarity withother non-plant eukaryotic and prokaryotic homologuesis considerably lower (Table 2). The similarity betweenthe C. creticus GGDPSs is 74% at the amino acid level.

In silico analysis, using the ChloroP v1.1 (Emanuelssonet al., 1999), TargetP 1.1 (Emanuelsson et al., 2000) and thePredotar V1.03 (Prediction of organelle targetingsequences, http://genoplante-info.infobiogen.fr/predotar/)programs showed that both CcGGDPS proteins containa putative transit peptide that directs the mature proteininto the plastids, with high probability (in a range of 57–97%). For the prediction of the cleavage site, the aboveprograms lack consistency. However, it has been provenexperimentally for the Taxus canadensis GGDPS that theGGDPS transit peptide extends till the region where thehigh conservation begins (Hefner et al., 1998). Likewise,the transit peptides for CcGGDPS1 and CcGGDPS2 pro-

I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652 1643

teins are presumed to extend from Met1 till Phe70 andPhe76 amino acids, respectively (Fig. 1). Consequently,the putative mature CcGGDPS1 and CcGGDPS2 proteinsconsist of 294 aa and have a size of 31.7 kDa. It is worth-while to notice that the putative mature CcGGDPS1 andCcGGDPS2 proteins have 85% identity, in pair wise com-parison using the ‘‘Blast 2 sequences” (http://www.ncbi.nlm.nih.gov/BLAST/bl2seq/wblast2.cgi) program.

Among the several conserved regions in all trans-pren-yltransferases there are two aspartate rich regions[DDxx(xx)D], the FARM (First Aspartate Rich Motif)and the SARM (Second Aspartate Rich Motif) located inthe enzymes’ large, central, hydrophobic cavity responsiblefor substrate binding and catalysis (Liang et al., 2002;Wang and Ohnuma, 1999). It has been shown, for the

Fig. 1. Alignment of CcGGDPS with other plant proteins. Conserved regionsthe hypothetical transit peptide is underlined with a bold dashed line. Conserveboxed, while similar amino acids are in gray boxes. Alignments were performshading was done with the BoxShade 3.21 program (http://www.ch.embnet.or

FDP synthase, that the FARM is bound to the allylic sub-strate – GDP – while the SARM is the binding site of theIPP (Koyama et al., 1994, 1995; Tarshis et al., 1996). Sub-stitution of any of the aspartate molecules in those motifs,using site-directed mutagenesis, with different amino acidsresulted in reduced enzymatic activity (for review, seeLiang et al., 2002). In addition to the above-mentionedmotifs, there is another characteristic region in theseenzymes, the Chain Length Determination region (CLD),which is responsible for the size of the resultant product.This region contains the FARM domain and five aminoacids upstream of it. According to Wang and Ohnuma(1999), the length of the final product is determined fromthe size of the fourth and fifth amino acids upstream theFARM, as well as from the type of the FARM motif

like FARM and SARM motifs are underlined with a bold solid line, whiled amino acid residues in all the sequences used in this alignment are blacked with the ClustalW program (http://www.ebi.ac.uk/clustalw/) while theg/software/BOX_form.html/).

Fig. 2. Phylogenetic relationships of short chain prenyltransferases basedon their amino acid sequences. The phylogenetic tree was constructed withthe Neighbor-Joining method. The numbers indicated are the bootstrapvalues. The species from which the enzymes were obtained are indicated.

1644 I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652

(DDxxxxD or DDxxD). FDP and GGDP synthases areclassified in several types according to this region (Wangand Ohnuma, 1999). In the case of plant and eubacterialGGDPS (type II GGDPS) the FARM has the DDxxxxxDmotif (while the GGDPS from all other organisms do nothave the two amino acid insertion). In these proteins thefourth and fifth position before the FARM are both devoidof aromatic amino acids (unlike the GDPS, the eukaryoticFDPS and the GGDPS from archaea) (Bouvier et al., 2005;Wang and Ohnuma, 1999). Both isolated cDNAs from theC. creticus contain the FARM (DDxxxxD) and SARMdomains and seem to be classified within the type II ofGGDPS, according to the CLD region sequence (Fig. 1).The fact that the plant and eubacterial GGDPS are classi-fied in the same GGDPS type and that the DDxxxxD motifis found in plant and bacterial GDPS and GGDPS as wellas in bacterial FDPS enzymes shows the common origin ofthese enzymes (Bouvier et al., 2005; Wang and Ohnuma,1999). In the phylogenetic tree shown in Fig. 2 it is also evi-dent that the deduced amino acid sequences of CcGGDPS1and CcGGDPS2 are clustered together with the plantGGDPS.

Chen et al. (1994) proposed that all GDPS, FDPS andGGDPS originate from a common ‘‘ancient” prenyl-transferase, which later evolved in all the short chainprenyltransferases. He proposed that this ancestral prenyl-transferase segregated in two clusters, regardless of thechain length of their final product, the first for the eubacte-rial and the second for the eukaryotic enzymes, while themodifications responsible for the product length evolvedindependently, after the eukaryote and bacterial enzymessegregation (Chen et al., 1994). From the phylogenetic treeshown in Fig. 2, it is obvious that all eukaryotic (plantsincluded) FDP synthases (type I) form a distinct cluster.The GGDPS from other eukaryotic organisms (animalsand fungi) are segregated separately, while the plants’GDPS and GGDPS, which are plastidial enzymes, aregrouped together with the bacterial prenyltransferases,supporting the notion of their common origin, throughendosymbiotic phenomena. It is interesting to point outthe fact that the plant FDPS, which is a cytoplasmicenzyme, segregates with the rest of the eukaryotic enzymes.Moreover, it is worthwhile to notice that several Arabidop-sis GGDPS enzymes which have been reported as cytoplas-mic enzymes (Okada et al., 2000) segregate along with theplastidial and bacterial GGDPS. Regarding the plantGGDPSs, it is apparent that peptides coming from thesame species are not clustered necessarily together, unlikethe CcGGDPSS1 and CcGGDPS2, but separately, proba-bly according to their different role in the plant cell. How-ever, as CcGGDPS1 and CcGGDPS2 are groupedtogether, this analysis does not provide any informationregarding their functional role in the C. creticus cells ortissues.

As it was mentioned previously, plant GGDPSs areencoded by multigene families, comprising of more thantwo members in many plant species. In order to isolate

other potential GGDPS cDNAs from C. creticus, 150cDNA fragments, amplified by RT-PCR and highly degen-erate primers using leaf trichome and leaf total RNA, were

I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652 1645

analyzed by sequencing and/or restriction patterns (datanot shown). All the above fragments corresponded to theCcGGDPS1 or CcGGDPS2 genes and it was not possibleto isolate a third GGDPS cDNA. From the Southern blotanalysis (Fig. 3) it is obvious that there are more than twohybridization bands in the lanes RV and RI, while there areonly two in the BG. The hybridization was performed witha radiolabeled conserved fragment of the CcGGDPS1cDNA (from 618 nt till 1094 nt), which contains the secondaspartate-rich motif. Therefore, it is possible that certainbands visualized in the blot (specially the more faint ones)correspond to genes encoding prenyltransferases otherthan GGDPS. Of course, one cannot exclude the possibilityof the occurrence of some more GGDPS genes in the Cistusgenome. From the results described above it can beassumed that C. creticus GGDPS is encoded by at leasttwo genes.

2.2. Production of polyclonal antibodies against CcGGDPS

protein

For analyses of the GGDPS enzyme at the protein level, apolyclonal antibody against the CcGGDPS was produced.For this purpose, a 192-amino acid peptide from theCcGGDPS1 (spanned from 171 aa till the end), 25 kDa insize, which had 84% identity with the CcGGDPS2 enzyme,was used as antigen for rabbit immunization. Western blotanalyses on total Escherichia coli protein extracts from cellsexpressing the recombinant CcGGDPS1 protein and ontotal protein extracts from C. creticus leaves revealed thatthe resultant rabbit antiserum was able to recognize the

12kb

2kb

1.6kb

0.5kb

RV BG RI

Fig. 3. Southern blot analysis. Genomic DNA (�30 lg) isolated fromCistus creticus leaves was digested with EcoRV (RV), BglII (BG) andEcoRI (RI), fractionated in 0.8% agarose-TAE gel, transferred to HybondN-membrane and hybridised with a (a32P)CTP-labeled fragment corre-sponding to a conserved region of the CcGGDPS1 cDNA. Molecularweights are indicated at the left.

recombinant, as well as the native CcGGDPS1 protein,while no cross-reaction with other E. coli or C. creticus pro-teins was observed (Fig. 4). In addition, the above antiserumwas able to recognize the recombinant CcGGDPS2 proteinexpressed in the ANY119 yeast cells (Fig. 5).

2.3. Functional characterization of CcGGDPS cDNAs

The ANY119 Saccharomyces cerevisiae mutant (MATa,bet2-1, ura3-52, his4-619), which is deficient in the b-sub-unit of type II geranylgeranyl transferase, BET2 gene(Jiang et al., 1995), was used to functional characterizethe CcGGDPS cDNAs. The bet2-1 gene is a recessive tem-perature-sensitive mutant allele that inhibits yeast cellgrowth at 37 �C (Jiang et al., 1995). This system has beenused before to demonstrate the function of putative GGDPsynthases from Taxus canadensis (Hefner et al., 1998).Although yeast GGDP does not exhibit any GTTase activ-ity, the increased GGDP cellular pool resulted fromGGDPS gene overexpression, compensates for the lowGGDP affinity of the mutated GTTase (Jiang et al.,1995). Based on these observations, several plasmid con-structs, containing CcGGDPS1 and CcGGDPS2 cDNAfragments, were expressed in the ANY119 cells. Theirgrowth was monitored under different temperatures (25and 37 �C) and growth media (Fig. 6), in order to examinethe CcGGDPS1 and CcGGDPS2 cDNA capability to com-plement the bet2-1 mutation.

Three different fragments for each cDNA were used forplasmid construction, as described in Section 3: the pFL1and pFL2 expressing the presumptive pre-mature or full-length proteins, the pM1 and pM2 coding for the hypothet-ical mature proteins and pT1 and pT2 expressing truncatedversion of the proteins. As mentioned above, bothCcGGDPS1 and CcGGDPS2 contain a transit peptide

1 2

C. creticus

75kD

50kD

37kD

25kD

E. coli

25 kD

3 4

25 kD

1 2

A

C. creticus

75kD

50kD

37kD

25kD

E. coli

B

25 kD

3 4

Fig. 4. Protein blot analysis of recombinant and native CcGGDPSproteins. (A) 12% SDS–PAGE stained with Coommassie Brilliant Blue R-250. Lane 1: total protein extract from E. coli cells expressing therecombinant CcGGDPS1 peptide; Lane 2: recombinant CcGGDPS1protein purified from Ni-NNF-Agarose column, used as antigen. (D)Immunodetection of GGDPS protein. Lane 3: total protein extracts fromE. coli cells expressing the recombinant CcGGDPS1 peptide; Lane 4: totalprotein extracts from Cistus creticus leaves (antiserum dilution 1:1000).Protein molecular weights are indicated in the right.

M FL1 M1 T1 M FL2 M2 T2

A

175kD

83kD

62kD

47,5kD

32,5kD

pY

ES2

AN

Y11

9

B

47,5kD

32,5kD

25kD

M2 T2

25kD

pYE

S2

AN

Y11

9

Fig. 5. Immunodetection of the recombinant CcGGDPS1 (panel A) and CcGGDPS2 (panel B) proteins expressed in the ANY119 cells. The constructshosted in ANY119 cells are indicated in each lane. ANY119 cells without any construct (ANY119) or with the empty vector (pYES2) were used as controls(antiserum dilution 1:1000). The molecular weights of the protein bands are indicated on the left side. Equal amounts of total proteins were loaded in eachlane (data not shown).

Selective-inducible (galactose-raffinose) medium

25 ºC 7 ºC

M1 FL1

pYES2 T1

M1 FL1

pYES 1

FL1

pYES2 T1

M1 FL1

pYES 1

Selective-repressive (glucose) medium

M2 FL2

pYES2 T2

M2 FL2

pYES2 T2

M2 FL2

pYES2 T2

pYES2 T2

CcGGDPS1

pYES2 T1

1

pYES2 T1

M1 FL1

1

FL1

pYES 1

M1 FL1

1

pYES2 T11

25 ºC 37 ºC

pYES2 T2

2

pYES 2

M2 FL2

2

2

pYES2 T2

M2 FL2

2

pYES 2

M2 FL2

CcGGDPS2

Fig. 6. Growth of ANY119 cells carrying the CcGGDPS1 and CcGGDPS2 constructs. The constructs hosted in each case are indicated above and beloweach Petri dish (pYES2, pFL1/2, pM1/2, pT1/2). The medium used is indicated on the right, while the incubation temperature is at the top.

1646 I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652

presumably targeting them to the plastids. The length ofthis peptide was not possible to be accurately determinedin silico. However, based on the notion suggested byHefner et al. (1998) that a transit peptide can be consideredthe part of the protein expanded at the N-terminal non-conserved region, the transit peptides for CcGGDPS1and CcGGDPS2 were regarded the regions of 69 and75 aa in length, respectively, starting from the 50 end(Fig. 1). The hypothetical transit peptides were removedfrom the inserts of the pM1/2 constructs to check theirimpact on the catalytic activity of the correspondingenzymes. The pT1/2 peptides expressed a truncated versionof the CcGGDPS proteins, shorter than the hypotheticalmature ones. It should be noted that the primers used forcloning the fragments of interest were designed in a way

that a ‘‘Kozak motif” (ANNATGG) was formed aroundthe start-codon (ATG) of each recombinant transcript(Table 1), for effective translation (Kozak, 1987, 1991).

To ensure that the recombinant GGDPS proteins wereefficiently expressed in the ANY119 cells hosting the aboveconstructs, western blot analysis was performed in totalyeast protein extracts. Fig. 5 demonstrates that the GGDPSantibody recognized all GGDPS proteins of the expectedsize (FL1: 39.2 kDa; M1: 31.7 kDa; T1: 30.9 kDa; FL2:39.7 kDa; M2: 31.7 kDa; T2: 30.9 kDa). No bands weredetected in lanes containing ANY119 cells with the emptyvector, or ANY119 cells with no vector, suggesting thatthe CcGGDPS antibody recognized specifically the recom-binant CcGGDPS proteins and not other yeast pren-yltransferases. Moreover, all constructs were capable of

Table 1Primers used for the amplification of the cDNA fragments used for the constructs inserted in ANY119 yeast cells

Primers used for CcGGDPS1 constructs

GGDPS1FL.FOR1 50 CC AAG CTT ATA ATG GGT ACT GTT AGC ATC AGT T CG 30

GGDPS1TRM.FOR1 50 CC AAG CTT ATA ATG GGT TTC GAT TTC AAG TCC TAC ATG 30

GGDPS2TR.FOR1 50 CC AAG CTT AAC ATG GTT CAG AAG GCG AAT TCC 30

GGDPS1.REV1 50 GC TCT AGA CTA ATT CTG CCT ATA AG C AAT G 30

Primers used for CcGGDPS2 constructs

GGDPS2FL.FOR1 50 CC AAG CTT AAA ATG GGT ACT GTG AAT GTG AAT CTG G 30

GGDPS2TRM.FOR1 50 CC AAG CTT AAA ATG GGT TTT GAT TTC CAG TCT TAT ATG 30

GGDPS2TR.FOR1 50 CC AAG CTT AAG ATG GCG GAT TCC GTT AAT CAA GCG 30

GGDPS2.REV1 50 GC TCT AGA CTA GTT TTG CCT ATA AGC AAT G 30

With bold are indicated the START (ATG) and the STOP codon, while the HindIII (AAGCTT) and XbaI (TCTAGA) restriction sites are underlined.

Table 2Identity percentage of the Cistus creticus GGDP synthases with otherknown homologous enzymes

GGDP synthases(accession number)

Identitypercentage(%) with theCcGGDPS1

Identitypercentage(%) with theCcGGDPS2

Croton sublyratus (BAA86284) 73 71Hevea brasiliensis (BAB60678) 73 70Medicago truncatula (ABE80654) 72 70Tagetes erecta (AAG10424) 71 67Sinapis alba (CAA67330) 70 70Scoparia dulcis (BAA86285) 70 66Arabidopsis thaliana (Q0WUL9) 69 70Catharanthus roseus (CAA63486) 67 67Abies grandis (Q8W1R9) 57 59Taxus � media (Q6Q291) 56 58Oryza sativa (Q9XHX1) 53 54Streptomyces coelicolor

(NP_624521)26 26

Mycobacterium tuberculosis

(CAA17477)24 24

Homo sapiens (NP_001032354) 22 21Mus musculus (BAA76512) 21 20Saccharomyces cerevisiae

(AAB68296)19 19

The peptide sequences were compared using the SMS2 program (Bioin-formatics organisation, http://bioinformatics.org/sms2/ident_sim.html).

I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652 1647

expressing all versions of the CcGGDPS proteins, but withdifferent expression efficiencies. The pM1 and pM2 con-structs resulted in higher protein expression levels com-pared with the rest of the constructs (Fig. 5A and B).

The growth of the ANY119 cells was monitored in twodifferent synthetic culture media. The first one containedglucose which suppresses (minimize the activity) theGAL1 promoter that controls the expression of the recom-binant GGDPS proteins, while the second one containedgalactose and raffinose, which induce the GAL1. Thegrowth of the ANY119 cells hosting the CcGGDPS con-structs was tested at standard temperatures (25 �C), whichpermit normal growth of the bet2-1 mutants and at hightemperatures (37 �C), which restrict normal growth of thebet2-1 mutants. ANY119 cells hosting the empty pYES2vector were served as controls.

All ANY119 cells grew efficiently at 25 �C, irrespectiveof the construct or the medium used (Fig. 6). This was

expected, since this temperature did not restrict the growthof these cells, even in the absence of BET2 protein. In con-trast, only the cells expressing the FL1/2 and M1/2 pep-tides were able to grow at 37 �C in galactose-raffinosecontaining media, conditions that activated the GAL1 pro-moter and induced the recombinant CcGGDPS expression.No growth was observed in glucose, which suppressed theGAL1 activity. It is concluded, therefore, that only FL1/2and M1/2 peptides coded for active GGDPS enzymes,since they were able to reverse the ANY119 high-tempera-ture sensitivity phenotype. T1/2 truncated peptides did notcomplement GGDPS activity under these conditions. Cellsexpressing the mature GGDPS1 protein (M1) grew nor-mally at 37 �C in media that suppressed the GAL1 pro-moter (Fig. 6). This was probably due to a minimumGAL1 activity under these conditions (and consequentlya minimum M1 peptide synthesis), which apparently wassufficient for bet2-1 complementation. This was consistentwith the highest expression of M1/2 constructs as evi-denced by the higher amounts of the corresponding poly-peptides in western blots (Fig. 5). It seems that the first69 and 75 aa of the CcGGDPS1 and 2 proteins, respec-tively, were not necessary for the enzymes’ function, a factthat strengthens the hypothesis that these regions repre-sented actual transit peptides. It is interesting to mentionthat the size of hypothetical CcGGDPS mature proteins(32 kDa) coincided with the size of the protein detectedin the western blot of total leaf extracts from C. creticus

leaves (Fig. 4B). The removal of 6 or 9 aa from the matureCcGGDPS1 and CcGGDPS2 peptides, respectively (con-structs pT1 and pT2), was sufficient to abolish the GGDPSactivity of the above proteins.

2.4. CcGGDPS gene transcripts and proteins tissuedistribution

To study the regulation of CcGGDPS at the transcrip-tional and protein levels, RT-PCR was applied to detectspecifically the CcGGDPS1 and CcGGDPS2 transcript lev-els and western blotting to analyse total CcGGDPS proteinaccumulation (Fig. 7). Roots and seeds exhibited very lowexpression of both genes. The mRNA steady state levels ofCcGGDPS1 and 2 genes were higher in leaf trichomes thanin all other tissues studied (Fig. 7A). This corroborates

CcGGDPS peptides

01020304050607080

Root Stems FlBud Leaf Trich Seeds

Tissues

Rel

ativ

e pe

ptid

e le

vels

CcGGDPS transcripts

0

1

2

3

4

5

6

Root Stems FlBud Leaf Trich Seeds

Tissues

CcGGDPS transcripts

0.00

0.50

1.00

1.50

2.00

2.50

XS S M L

Leaf developmental stage

Rel

ativ

e tr

ansc

ript

le

vels

Rel

ativ

e tr

ansc

ript

leve

ls

CcGGDPS peptides

02468

10121416

XS S M L

Leaf developmental stage

Rel

ativ

e pe

ptid

e le

vels

A

B

CcGGDPS2

CcGGDPS1

CcGGDPS2

CcGGDPS1

Fig. 7. Accumulation of GGDPS polypeptides and transcripts in severalCistus tissues (A) and in different leaf developmental stages (B) usingprotein blot and RT-PCR analyses, respectively. FlBuds: flower buds,Trich: trichomes. Leaf developmental stages: XS: leaves 61 cm; S: leaves1–2 cm; M: leaves 2–3 cm; L: leaves P3 cm. Ten micrograms of totalprotein were loaded in each lane.

1648 I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652

with findings showing that glandular trichomes is the site ofterpenoid biosynthesis (Gang et al., 2001; Gershenzonet al., 1989, 1992). The results show also that the accumu-lation of the two CcGGDPS gene products followed differ-ent tissue pattern. This is an implication of the differentroles of the two GGDPS genes could have in Cistus tissues.After trichomes, accumulation of CcGGDPS1 was higherin leaves followed by the flower buds and stems, whileCcGGDPS2 was higher in flower buds and next in leaves.In terms of total CcGGDPS transcript steady state levels,leaves, flower buds and stems followed trichomes. The highlevels of the CcGGDPS transcripts in the flower buds could

be explained from the fact that increased needs for isopre-noids maybe needed for the protection of this sensitive tis-sue as well as the for the formation of the flower pigmentsand volatiles (Pichersky and Gang, 2000; Tholl et al.,2005). Differential expression of the GGDPS gene familymembers has also been reported for several genes inArabidopsis (Okada et al., 2000) and for two in tomato(Ament et al., 2006).

In view of the evidence that GGDPS genes were differen-tially expressed in tissues studied, we next focused ourstudy on the accumulation of CcGGDPS protein in thesame tissues. Indeed, the levels of CcGGDPS immunoreac-tive polypeptides paralleled the expression of total GGDPStranscripts in the above tissues, with trichomes being therichest tissue in CcGGDPS protein. These findings suggestthat the control of CcGGDPS synthesis is mainly at thetranscription level.

2.5. Developmental regulation of CcGGDPS protein and

transcripts

Initial studies indicated that C. creticus trichomes fromvery young (0.5–1.0 cm) and young leaves (1–2 cm) con-tained the highest amount of labdane-type diterpenes(Falara and Kanellis, unpublished results). Similarly, otherinvestigators showed that the production of secondarymetabolites in leaves and trichomes is under developmentalcontrol (Gershenzon et al., 2000; McConkey et al., 2000;Valkama et al., 2004). Therefore, it was deemed necessaryto study the CcGGDPS protein and mRNA profile duringCistus leaf development. Leaves of different size rangingfrom extra small to large were harvested and total proteinextracts analysed in terms of their immunoreactiveCcGGDPS protein content (Fig. 7B). It is evident thatCcGGDPS polypeptides’ accumulation was inverselyrelated to the size of the leaves, that is the smallest or youn-gest leaves expressed the highest amount of protein(Fig. 7B).

In order to examine if these differences were due to thevariations in transcript levels and also to distinguish whichmember of the CcGGDPS gene family was responsible forthis pattern, we monitored the level of the correspondingmRNAs in the same leaves by RT-PCR analysis(Fig. 7B). Total transcript accumulation of the twoCcGGDPS genes paralleled the corresponding proteinaccumulation during leaf development, being the maxi-mum expression of both genes in the extra small leaves.Thereafter, CcGGDPS1 and 2 transcript abundance exhib-ited a progressive decline. However, it seems that bothgenes were expressed in relatively equal amounts duringthe whole period, suggesting a similar developmental andtranscriptional control during leaf development. ThisCcGGDPS1 and 2 expression pattern at the mRNA andprotein levels resembled the labdane-type diterpenes’ accu-mulation during C. creticus leaf development (Falara andKanellis, unpublished data). Monoterpene accumulationand expression of their biosynthetic genes, for example in

I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652 1649

peppermint (Mentha � piperita L.), although are underdevelopmental control as in C. creticus, however, they fol-low a different pattern with maximum accumulation ofcompounds and the corresponding mRNA in medium sizeleaves aged close to 12-day of growth (McConkey et al.,2000). This disparity can be attributed either to differentphysiological functions of diterpenes and monoterpenesduring leaf development or due to different type of tri-chome in these two species and the way that these com-pounds are stored and excreted. In peppermint, thepeltate trichomes contain a storage cavity where thesemetabolites are stored (Turner et al., 2000), whereas theC. creticus trichomes do not store the produced metabo-lites, but they excrete them to the environment (Gulzet al., 1996). Similarly, Kuntz et al. (1992) showed thatGGDPS gene is developmentally regulated in Capsicum

annum leaves where the levels of the correspondingmRNAs are reduced with leaf age. Finally, the observeddecline in CcGGDPS1 and 2 transcript levels in older leaves(Fig. 7B) and the corresponding lower production of lab-dane-type diterpenes (Falara and Kanellis, unpublisheddata) might be explained by the substitution of terpenoidsby a thicker wax cuticle which possibly makes the leavesless sensitive to pathogens, herbivores and other biotic orabiotic stresses (Viougeas et al., 1995).

In conclusion, it is shown that the functional character-ised CcGGDPS1 and 2 genes and the corresponding pro-teins are tissue and developmentally regulated and thiscontrol is exerted at the transcript level.

3. Experimental

3.1. Plant material

C. creticus ssp. creticus plants were grown outdoors, inthe area of Thermi, Thessaloniki, Greece, at the premisesof the National Agricultural Research Foundation. Planttissue was collected, immediately frozen in liquid nitrogen,and stored at �80 �C.

3.2. Total RNA, genomic DNA purification and Southern

blot analysis

Total RNA and genomic DNA from C. creticus tissueswere extracted following the protocols described by Pate-raki and Kanellis (2004). Genomic DNA (30 lg) wasdigested with EcoRV, BglII and EcoRI endonucleases,fractionated in 0.8% (w/v) agarose gel, transferred on tonylon membrane (Nytran� 0.45, Schleicher & Schuell)and hybridised with a radiolabeled CcGGDPS1 fragment,corresponding to a highly conserved region of the gene.Probe labelling was carried out using the RadPrimeDNA Labeling System (Invitrogen, Life Technologies,Madison, WI, USA). Blot hybridization, membrane wash-ing and autoradiography were performed as described byChurch and Gilbert (1984).

3.3. Cloning and sequence analysis of C. creticus GGDPS

full-length cDNAs

An initial cDNA fragment, 470 bp, corresponding to a C.

creticus GGDPS was amplified by PCR with degenerateprimers [GGDP.SEN2: 50 CA(CT) GA(CT) GA(CT)(CT)T(AGCT) CC(AGCT) TG(CT) ATGG 30 andGGDP.AS1: 50 (AG)TC (AGCT)TT (AGCT)CC(AGCT)GC (AGCT)GT (AGCT)TT (AGCT)CC 30)] from C. creti-

cus genomic DNA. For the cloning of the 50 of the GGDPScDNAs, 50 RACE was performed using the ‘‘50 RACE Sys-tem for Rapid Amplification of cDNA Ends” (Invitrogen,Life Technologies, Carlsbad, CA) and the gene specific pri-mer GGDP.AS3 (50 GCC AAC TGT GGA CCC CGTCAT GTG CTC 30). From this procedure two PCR frag-ments were produced, corresponding to two differentCcGGDPS cDNAs. The 30 of these cDNAs were obtainedusing 30 RACE techniques with the primers: GGDP.SEN2,specGGDPS2.FOR (50 ATT GCC GCG TGT GAG CTCGTC GGC GG 30), and oligodT (50 ACT AGT CTC GAGTTT TTT TTT TTT TT TTT TT 30). As template for the50 and 30 RACE reactions was used single strand cDNA syn-thesized using the oligodT primer and reverse transcriptaseSuperScript II (Invitrogen, Life Technologies, Carlsbad,CA) from DNaseI-treated (Promega, GmbH, Mannheim,Germany) total RNA isolated from young leaves. PCRproducts were cloned into pGEM-T Easy (Promega GmbH,Mannheim, Germany) vector and sequenced using the LI-COR Long Read 4200 automated sequencer and the ‘‘Sequi-therm EXCELII” kit (Epicentre, Madison, WI, USA).Phylogenetic analyses were performed using neighbor join-ing analysis by MEGA 3.1 program. Bootstrap values werecalculated by distance analysis for 1000 replicates.

3.4. Antibody production against CcGGDPS1

For the antibody production, the CcGGDPS1 fragmentfrom the Phe172 until the Asn363 amino acids was used ascorresponded to a highly conserved and hydrophilic region.The GGDPa/b.FOR (50 ATA CTC GAG TTC GGC GAGGAT ATC GCG GTG 30) and GGDPa/b.REV (50 ATCGGA TCC TCA TTC TGC CTA TAA GCA ATG 30)primers, containing the XhoI and BamHI restriction sites,respectively, were used for the amplification of the corre-sponding cDNA fragment. The fragment was cloned inthe E. coli expression vector pET-16b (Novagen, Madison,WI, USA) and expressed in E .coli BL21(DE3) cells. Therecombinant protein was 6XHIS-tagged and was purifiedfrom BL21(DE3) cells’ protein extract, under denaturatedconditions with 8 M Urea, using a Ni-NNF-Agarose(Novagen, Madison, WI, USA) column, according to themanufacturer’s instructions. The purified recombinant pro-tein (serving as antigen) was injected in rabbits for anti-body production, in the facilities of the Institute ofMolecular Biology and Biotechnology, Foundation ofResearch and Technology, Heraklion, Greece. Antiserumof the rabbit was collected and stored at �80 �C.

Table 3Primers used for the CcGGDPS1 and CcGGDPS2 expression studies

Gene Primers Sequence

CcEF1a CcEF1a.For1 50 GGT CCT ACT GGT TTA ACCACT G 30

CcEF1a.Rev1 50 CTC GGA GAA GGT CTC CACAAC C 30

CcGGDPS1 CcGGDPS1.For1 50 AGT TGG CGT ATC CCC CGCCCG 30

CcGGDPS1.Rev1 50 TAC ATC TGT CTC TAA GCAGTC GC 30

CcGGDPS2 CcGGDPS2.For1 50 GAT GTG ACG AAA TCT TCCGTG 30

CcGGDPS2.Rev1 50 CTT TTA TGC TTC TTT CATTCA TAG 30

1650 I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652

3.5. Generation of GGDP synthase yeast expression

constructs and complementation assays

PCR primers containing a HindIII (AAGCTT) or aXbaI (TCTAGA) site were designed based on the 50 and30 termini of the CcGGDPS1 and CcGGDPS2 cDNAs(Table 1), and used for the amplification and cloning ofthe cDNA fragments in the pYES2 yeast expression vector(Invitrogen, Life Technologies Carlsbad, CA). Three frag-ments were amplified for each cDNA. The first one wasdesigned to express the hypothetical full-length CcGGDPSproteins (pFL1 and pFL2 constructs), the second toexpress the hypothetical mature proteins (pM1 – fromPhe70 till STOP codon-, and pM2 – from Phe76 till STOPcodon-constructs) while the third one to express a trun-cated version of each protein (pT1 – from Met76 till STOPcodon-, and pT2 – from Met85 till STOP codon-con-structs). The fragments amplification was performed usingthe Platinum Pfx DNA polymerase, an enzyme with proof-reading activity, (Invitrogen, Life Technologies, Carlsbad,CA) in order to minimize the possibility of sequence dis-crepancies. The constructs were verified with sequencingand inserted in ANY119 yeast cells, with the LiAc method(Becker and Lundblad, 1998). The CcGGDPS recombi-nant proteins were expressed in yeast cells under the con-trol of GAL1 promoter of the pYES2 vector. S.

cerevisiae strain ANY119 was a gift from Dr. S. Ferro-Novick (Howard Hughes Medical Institute, Yale Univer-sity) (Jiang et al., 1995). Isolated colonies, correspondingto each construct, were streaked on plates with syntheticmedia containing glucose (2%, v/v) (suppress the GAL1promoter) or galactose (2%, v/v) together with raffinose(1%, v/v) (induce the GAL1 promoter), as carbon source(lacking uracil) and grown for 3 days at 25 or 37 �C. Forthe GGDPS protein expression in liquid media, isolatedcolonies were used to inoculate media containing glucose(2%, v/v), cells were grown for 16 h and then collected,rinsed with sterile water and grown for four more hoursin media containing galactose/raffinose (2% and 1%, v/v,respectively).

3.6. Protein extraction and western blot analysis

Total protein extraction from C. creticus: 250 mg of sev-eral plant tissues collected from plants growing outdoorswere grinded and homogenized in 0.5 ml of extraction buf-fer (2% SDS, 0.2 M Tris–HCl, pH 8.0, 5 mM EDTA, 5 mMMgCl2, 10% (v/v) glycerol, 2% (v/v) 2-mercaptoethanol,2 mg/ml PMSF, 3% (w/v) PVPP). The homogenate wasincubated for 10 min at 94 �C and centrifuged for 10 minat 15,000g. The supernatant was collected and stored at�20 �C.

Total protein extraction from yeast: cells collected from5 ml liquid culture were resuspended in 0.5 ml of yeast pro-tein extraction buffer (1% NP40, 120 mM NaCl, 1 mlPMSF). Equal volumes of glass beats were added and vor-texed for 20 min at 4 �C. The mixture was centrifuged for

5 min at 13,000 rpm and the supernatant was collectedand stored at �20 �C.

Total protein extraction from E. coli cells and westernblot analyses were performed according to Sambrooket al. (1989). Ten micrograms of total crude proteinextracts were loaded in each lane shown in Fig. 7. The pro-teins of interest were detected using as primary antibodythe anti-GGDPS, in dilution 1:1000, and as secondary ahorseradish peroxidise conjugated goat anti-rabbit anti-body (dilution 1:2000) and 3,30 diaminobenzidine (DAB)/hydrogen peroxide.

3.7. RT-PCR reactions

RT-PCR techniques were used to monitor the expres-sion of the CcGGDPS1 and CcGGDPS2 genes in C. creti-

cus tissues. First strand cDNA synthesis was performedusing an oligodT primer and M-MLV reverse transcriptase(Invitrogen, Life Technologies, Carlsbad, CA) from DNa-seI-treated (Promega, GmbH, Mannheim, Germany) totalRNAs isolated from C. creticus tissues. The gene specificprimers used for the PCR reactions are shown in Table 3.The constitutively expressed C. creticus Eukaryotic Trans-lation Elongation Factor-1a cDNA (CcEF1a, accessionnumber EF062868) was used as internal control. Thecycling conditions were optimized for each pair of primersin a way that sharp bands corresponding to the expectedfragments were amplified with out any signs of non-specificproducts. The PCR amplified products for the CcGGDPS1cDNA was 612 bp, for the CcGGDPS2 was 261 bp, whilefor the CcEF1a was 411 bp. The fragments were clonedinto pGEM-T Easy (Promega GmbH, Mannheim, Ger-many) vector and sequenced using a LI-COR Long Read4200 automated sequencer and ‘‘Sequitherm EXCELII”

kit (Epicentre, Madison, WI, USA) to verify the amliconidentity. Relative transcript levels were visualized by settingthe cycle numbers so that the rate of PCR product ampli-fication was in the early exponential stage of the reaction.PCR products were resolved in agarose gel electrophoresis,photographed and quantified with ImageJ program athttp://rsb.info.nih.gov/ij/ (Abramoff et al., 2004).

I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652 1651

The experiments were performed in duplicates regardingbiological replications, where each individual sample wasa pool of approximately 10 plants. Two technical replica-tions were performed for each biological replicationmentioned.

Acknowledgements

This research was partially supported from a grant bythe Greek General Secretariat of Research and Technology(GSRT) awarded to A.K.K. (PENED 99ED 637).

References

Abramoff, M.D., Magelhaes, P.J., Ram, S.J., 2004. Image processing withImageJ. Biol. Int. 11, 36–42.

Ament, K., Van Schie, C.C., Bouwmeester, H.J., Haring, M.A., Schuur-ink, R.C., 2006. Induction of a leaf specific geranylgeranylpyrophosphate synthase and emission of -4,8,12-trimethyltrideca-1,3,7,11-tetraene in tomato are dependent on both jasmonic acid andsalicylic acid signaling pathways. Planta 224, 1197–1208.

Becker, D.M., Lundblad, V., 1998. Current Protocols in MolecularBiology. John Wiley & Sons, New York.

Bohlmann, J., Meyer-Gauen, G., Croteau, R., 1998. Plant terpenoidsynthases: molecular biology and phylogenetic analysis. Proc. Natl.Acad. Sci. USA 95, 4126–4133.

Bouvier, F., Rahier, A., Camara, B., 2005. Biogenesis, molecularregulation and function of plant isoprenoids. Progr. Lipid Res. 44,357–429.

Chen, A., Kroon, P.A., Poulter, C.D., 1994. Isoprenyl diphosphatesynthases: protein sequence comparisons, a phylogenetic tree, andpredictions of secondary structure. Protein Sci. 3, 600–607.

Church, G.M., Gilbert, W., 1984. Genomic sequencing. Proc. Natl. Acad.Sci. USA 81, 1991–1995.

Demetzos, C., Harvala, C., Philianos, S., 1990. A new labdane-typediterpene and other compounds from the leaves of Cistus incanus ssp.creticus. J. Nat. Prod. 53, 1365–1368.

Demetzos, C., Mitaku, S., Couladis, M., Harvala, C., Kokkinopoulos, D.,1994. Natural metabolites of ent-13-epi-manoyl oxide and othercytotoxic diterpenes from the resin ‘‘Ladano” of Cistus creticus.Planta Med. 60, 590–591.

Dimas, C., Demetzos, C., Marsellos, M., Sotiriadou, R., Malamas, M.,Kokkinopoulos, D., 1998. Cytotoxic activity of labdane type diterpenesagainst human leukemic cell lines in vitro. Planta Med. 64, 208–211.

Dimas, C., Demetzos, C., Mitaku, S., Vaos, B., Marselos, M., Tzavaras,T., Kokkinopoulos, D., 1999. Cytotoxic activity and antiproliferativeeffects of a new semi-synthetic derivative of ent-3beta-hydroxy-13-epi-manoyl oxide on human leukemic cell lines. Anticancer Res. 19, 4065–4072.

Dimas, C., Demetzos, C., Vaos, V., Ioannidis, P., Trangas, T., 2001.Labdane type diterpenes down-regulate the expression of c-mycprotein but not of Bcl-2 in human leukemia cell lines undergoingapoptosis. Leukemia Res. 25, 449–454.

Dimas, K., Hatziantoniou, S., Tseleni, S., Khan, H., Georgopoulos, A.,Alevizopoulos, K., Wyche, J.H., Pantazis, P., Demetzos, C., 2007.Sclareol induces apoptosis in human HCT116 colon cancer cellsin vitro and suppression of HCT116 tumor growth in immunodeficientmice. Apoptosis 12, 685–694.

Dimas, K., Papadaki, A., Tsimplouli, C., Hatziantoniou, S., Alevizopo-ulos, K., Pantazis, P., Demetzos, C., 2006. Labd-14-ene-8,13-diol(sclareol) induces cell cycle arrest and apoptosis in human breastcancer cells and enhances the activity of anticancer drugs. Biomed.Pharmacother. 60, 127–133.

Emanuelsson, O., Nielsen, H., Brunak, S., von Heijne, G., 2000.Predicting subcellular localization of proteins based on their N-terminal amino acid sequence. J. Mol. Biol. 300, 1005–1016.

Emanuelsson, O., Nielsen, H., von Heijne, G., 1999. ChloroP, a neuralnetwork-based method for predicting chloroplast transit peptides andtheir cleavage sites. Protein Sci. 8, 978–984.

Gang, D.R., Wang, J., Dudareva, N., Hee Nam, K., Simon, J.E.,Lewinsohn, E., Pichersky, E., 2001. An investigation of the storage andbiosynthesis of phenylpropenes in sweet basil. Plant Physiol. 125, 539–555.

Gershenzon, J., Maffei, M., Croteau, R., 1989. Biochemical and histo-chemical localization of monoterpene biosynthesis in the glandulartrichomes of spearmint (Mentha spicata). Plant Physiol. 89, 1351–1357.

Gershenzon, J., McCaskill, D., Rajaonarivony, J.I.M., Mihaliak, C.,Karp, F., Croteau, R., 1992. Isolation of secretory cells from plantglandular trichomes and their use in biosynthetic studies of monoter-penes and other gland products. Anal. Biochem. 200, 130–138.

Gershenzon, J., McConkey, M.E., Croteau, R., 2000. Regulation ofmonoterpene accumulation in leaves of peppermint. Plant Physiol.122, 205–213.

Graber, J.H., Cantor, C.R., Mohr, S.C., Smith, T.F., 1999. In silico

detection of control signals: mRNA 30-end-processing sequences indiverse species. Proc. Natl. Acad. Sci. USA 96, 14055–14060.

Gulz, P.G., Herrmann, T., Hangst, K., 1996. Leaf trichomes in the genusCistus. Flora 191, 82–104.

Hatziantoniou, S., Dimas, K., Georgopoulos, A., Sotiriadou, N., Demet-zos, C., 2006. Cytotoxic and antitumor activity of liposome-incorpo-rated sclareol against cancer cell lines and human colon cancerxenografts. Pharmacol. Res. 53, 80–87.

Hefner, J., Ketchum, R.E.B., Croteau, R., 1998. Cloning and functionalexpression of a cDNA encoding geranylgeranyl diphosphate synthasefrom Taxus canadensis and assessment of the role of this prenyltrans-ferase in cells induced for taxol production. Arch. Biochem. Biophys.360, 62–74.

Jiang, Y., Proteau, P.J., Poulter, D., Ferro-Novick, S., 1995. BTS1

encodes a geranylgeranyl diphosphate synthase in Saccharomyces

cerevisiae. J. Biol. Chem. 270, 21793–21799.Koepp, A.E., Hezari, M., Zajicek, J., Stofer-Vogel, B., LaFever, R.E.,

Lewis, N.G., Croteau, R., 1995. Cyclization of geranylgeranyldiphosphate to taxa-4(5),11(12)-diene is the committed step of taxolbiosynthesis in Pacific yew. J. Biol. Chem. 270, 8686–8690.

Kojima, N., Sitthithaworn, W., Viroonchatapan, E., Suh, D., Iwanami,N., Hayashi, T., Sankaw, U., 2000. Geranylgeranyl diphosphatesynthases from Scoparia dulcis and Croton sublyratus cDNA cloning,functional expression, and conversion to a farnesyl diphosphatesynthase. Chem. Pharm. Bull. 48, 1101–1103.

Koyama, T., Obata, S., Saito, K., Takeshita-Koike, A., Ogura, K., 1994.Structural and functional roles of the cysteine residues of Bacillus

stearothermophilus farnesyl diphosphate synthase. Biochemistry 33,12644–12648.

Koyama, T., Tajima, M., Nishino, T., Ogura, K., 1995. Significance ofPhe-220 and Gln-221 in the catalytic mechanism of farnesyl diphos-phate synthase of Bacillus stearothermophilus. Biochem. Biophys. Res.Commun. 212, 681–686.

Kozak, M., 1987. An analysis of 50-noncoding sequences from 699vertebrate messenger RNAs. Nucleic Acid Res. 15, 8125–8148.

Kozak, M., 1991. An analysis of vertebrate mRNA sequences: intimationsof translational control. J. Cell Biol. 115, 887–903.

Kuntz, M., Romer, S., Suire, C., Hugueney, P., Weil, J.H., Schantz, R.,Camara, B., 1992. Identification of a cDNA for the plastid-locatedgeranylgeranyl pyrophosphate synthase from Capsicum annuum: cor-relative increase in enzyme activity and transcript level during fruitripening. Plant J. 2, 25–34.

Lange, B.M., Ghassemian, M., 2003. Genome organization in Arabidopsis

thaliana: a survey for genes involved in isoprenoid and chlorophyllmetabolism. Plant Mol. Biol. 51, 925–948.

Laskaris, G., van der Heijden, R., Verpoorte, R., 2000. Purification andpartial characterisation of geranylgeranyl diphosphate synthase from

1652 I. Pateraki, A.K. Kanellis / Phytochemistry 69 (2008) 1641–1652

Taxus baccata cell cultures. An enzyme that regulates taxane biosyn-thesis. Plant Sci. 153, 97–105.

Liang, P.H., Ko, T.P., Wang, A.H.J., 2002. Structure, mechanism andfunction of prenyltransferases. Eur. J. Biochem. 269, 3339–3354.

Matsingou, C., Dimas, K., Demetzos, C., 2006. Design and developmentof liposomes incorporating a bioactive labdane-type diterpene in vitrogrowth inhibiting and cytotoxic activity against human cancer celllines. Biomed. Pharmacother. 60, 191–199.

Matsingou, C., Hatziantoniou, S., Georgopoulos, A., Dimas, K., Terzis,A., Demetzos, C., 2005. Labdane-type diterpenes: thermal effects onphospholipid bilayers, incorporation into liposomes and biologicalactivity. Chem. Phys. Lipid 138, 1–11.

McConkey, M.E., Gershenzon, J., Croteau, R., 2000. Developmentalregulation of monoterpene biosynthesis in the glandular trichomes ofpeppermint. Plant Physiol. 122, 215–224.

Ogura, K., Koyama, T., 1998. Enzymatic aspects of isoprenoid chainelongation. Chem. Rev., 1263–1276.

Oh, S.K., Kim, I.J., Shin, D.H., Yang, J., Kang, H., Han, K.H., 2000.Cloning, characterization, and heterologous expression of a functionalgeranylgeranyl pyrophosphate synthase from sunflower (Helianthus

annuus L.). J. Plant Physiol. 157, 535–542.Okada, K., Saito, T., Nakagawa, T., Kawamukai, M., Kamiya, Y., 2000.

Five geranylgeranyl diphosphate synthases expressed in differentorgans are localized into three subcellular compartments in Arabidop-

sis. Plant Physiol. 122, 1045–1056.Pateraki, I., Kanellis, A.K., 2004. Isolation of high-quality nucleic acids

from Cistus creticus ssp. creticus and other medicinal plants.. Anal.Biochem. 328, 90–92.

Pichersky, E., Gang, D.R., 2000. Genetics and biochemistry of secondarymetabolites in plants: an evolutionary perspective. Trends Plant Sci. 5,439–445.

Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning: ALaboratory Manual, second ed. Cold Spring Harbor Laboratory, ColdSpring Harbor, NY.

Takaya, A., Zhang, Y.W., Asawatreratanakul, K., Wititsuwannakul, D.,Wititsuwannakul, R., Takahashi, S., Koyama, T., 2003. Cloning,expression and characterization of a functional cDNA clone encodinggeranylgeranyl diphosphate synthase of Hevea brasiliensis. Biochim.Biophys. Acta 1625, 214–220.

Tarshis, L.C., Proteau, P.J., Kellogg, B.A., Sacchettini, J.C., Poulter,C.D., 1996. Regulation of product chain length by isoprenyl diphos-phate synthases. Proc. Natl. Acad. Sci. USA 93, 15018–15023.

Tholl, D., Chen, F., Petri, J., Gershenzon, J., Pichersky, E., 2005. Twosesquiterpene synthases are responsible for the complex mixture ofsesquiterpenes emitted from Arabidopsis flowers. Plant J. 42, 757–771.

Trapp, S.C., Croteau, R.B., 2001. Genomic organization of plant terpenesynthases and molecular evolutionary implications. Genetics 158, 811–832.

Turner, G.W., Gershenzon, J., Croteau, R., 2000. Development of peltateglandular trichomes of peppermint. Plant Physiol. 124, 665–680.

Valkama, E., Salminen, J., Koricheva, J., Pihlaja, K., 2004. Changes inleaf trichomes and epicuticular flavonoids during leaf development inthree birch taxa. Ann. Bot. 94, 233–242.

Viougeas, M.A., Rohr, R., Chamel, A., 1995. Structural changes andpermeability of ivy (Hedera helix L.) leaf cuticles in relation to leafdevelopment and after selective chemical treatments. New Phytol. 130,337–348.

Wang, K., Ohnuma, S.I., 1999. Chain-length determination mechanism ofisoprenyl diphosphate synthases and implications for molecularevolution. Trend Biochem. Sci. 24, 445–451.