lable at ScienceDirect

Plant Physiology and Biochemistry 74 (2014) 304e314

Contents lists avai

Plant Physiology and Biochemistry

journal homepage: www.elsevier .com/locate/plaphy

Research article

Expression profiles of key phenylpropanoid genes during Vanillaplanifolia pod development reveal a positive correlation between PALgene expression and vanillin biosynthesis

Isabelle Fock-Bastide a,*, Tony Lionel Palama a, Séverine Bory a, Aurélie Lécolier b,Michel Noirot b, Thierry Joët c

aUniversité de la Réunion, UMR Peuplements Végétaux et Bioagresseurs en Milieu Tropical (PVBMT), Pôle de Protection des Plantes, 7, Chemin de l’IRAT,Ligne Paradis, 97410 Saint Pierre, Franceb IRD, UMR Peuplements Végétaux et Bioagresseurs en Milieu Tropical (PVBMT), Pôle de Protection des Plantes, 7, Chemin de l’IRAT, Ligne Paradis, 97410Saint Pierre, Francec IRD, UMR Diversité, Adaptation et Développement des Plantes (DIADE), 911 Avenue Agropolis, BP64501, 34394 Montpellier, France

a r t i c l e i n f o

Article history:Received 18 August 2013Accepted 25 November 2013Available online 4 December 2013

Keywords:Cinnamate 4-hydroxylaseFruit maturationPhenolic compoundsPhenylalanine ammonia-lyaseReal-time quantitative PCRReference genesVanilla planifolia

Abbreviations: C4H, cinnamate 4-hydroxylase; HHBAlc, p-hydroxybenzyl alcohol; HBAld, p-hydroxybafter pollination; OMT, O-methyltransferases; PAL, phPCAld, protocatechualdehyde; Van, vanillin; VanAc,alcohol; 4HBS, 4-hydroxybenzaldehyde synthase.* Corresponding author. Tel.: þ33 262 262 49 92 27

E-mail address: isabelle.fock@univ-reunion.fr (I. F

0981-9428/$ e see front matter � 2013 Elsevier Mashttp://dx.doi.org/10.1016/j.plaphy.2013.11.026

a b s t r a c t

In Vanilla planifolia pods, development of flavor precursors is dependent on the phenylpropanoidpathway. The distinctive vanilla aroma is produced by numerous phenolic compounds of which vanillinis the most important. Because of the economic importance of vanilla, vanillin biosynthetic pathwayshave been extensively studied but agreement has not yet been reached on the processes leading to itsaccumulation. In order to explore the transcriptional control exerted on these pathways, five key phe-nylpropanoid genes expressed during pod development were identified and their mRNA accumulationprofiles were evaluated during pod development and maturation using quantitative real-time PCR. As aprerequisite for expression analysis using qRT-PCR, five potential reference genes were tested, and twogenes encoding Actin and EF1 were shown to be the most stable reference genes for accurate normal-ization during pod development. For the first time, genes encoding a phenylalanine ammonia-lyase(VpPAL1) and a cinnamate 4-hydroxylase (VpC4H1) were identified in vanilla pods and studied duringmaturation. Among phenylpropanoid genes, differential regulation was observed from 3 to 8 monthsafter pollination. VpPAL1 was gradually up-regulated, reaching the maximum expression level atmaturity. In contrast, genes encoding 4HBS, C4H, OMT2 and OMT3 did not show significant increase inexpression levels after the fourth month post-pollination. Expression profiling of these key phenyl-propanoid genes is also discussed in light of accumulation patterns for key phenolic compounds.Interestingly, VpPAL1 gene expression was shown to be positively correlated to maturation and vanillinaccumulation.

� 2013 Elsevier Masson SAS. All rights reserved.

1. Introduction

Phenylpropanoid biosynthesis is among the most importantsecondary metabolism pathways in higher plants. It gives rise to alarge number of metabolites, including hydroxycinnamic acids,monolignols/lignin, coumarins, benzoic acids, stilbenes, anth

BAc, p-hydroxybenzoic acid;enzaldehyde; MAP, monthsenylalanine ammonia-lyase;

vanillic acid; VanAlc, vanillyl

; fax: þ33 262 262 49 92 93.ock-Bastide).

son SAS. All rights reserved.

ocyanins and flavonoids (Dixon et al., 2002; Vogt, 2010). The corereactions of phenylpropanoid pathway involve three enzymes,generally represented by several gene orthologs among genomes:phenylalanine ammonia-lyase (PAL; EC 4.3.1.4), cinnamate 4-hydroxylase (C4H; EC 1.14.13.11), and 4-coumarate: coenzyme Aligase (4CL; EC 6.2.1.12). PAL is the first enzyme in the pathway andcatalyzes the conversion of L-phenylalanine to trans-cinnamic acid.Afterward, C4H, which belongs to the cytochrome P450 super-family, hydroxylates t-cinnamic acid into para-coumaric acid,leading to the production of lignin and flavonoids (Dixon et al.,2002; Ehlting et al., 2006). Thus, PAL is at a metabolically impor-tant position, linking primary and secondary metabolism.

Phenolic compounds play a crucial role in fruit development,maturity and attractiveness. Fruit quality is indeed often associated

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314 305

with specific anthocyanins and characteristic benzenoid flavorcomponents, both metabolites deriving from L-phenylalanine pro-duced through the phenylpropanoid pathway (Griesser et al., 2008;Kumar and Ellis, 2001). Because of its wide use in food, beveragesand cosmetics, vanillin (4-hydroxy-3-methoxy-benzaldehyde) isone of the most popular flavors coming from the phenylpropanoidpathway. This secondary metabolite is synthesized in the “pods” ofaromatic Vanilla species, particularly in Vanilla planifolia. If vanillinis the most abundant compound, various phenolic derivatives andtheir glycosyl conjugates are also produced in vanilla pods. In fact,the distinctive vanilla flavor is made up of more than 250 com-pounds (Kanisawa et al., 1994; Walton et al., 2003). At harvest,green vanilla pods are flavorless. During pod development (from 7to 10 months depending on the vines), flavor precursors mainlyaccumulate in the form of non-aromatic glucosidic compounds.Their biosynthesis has been extensively studied in the plant and incell cultures (Tokoro et al., 1990). Nevertheless, the biosynthesispathway of vanillin remains a controversial issue (Dixon, 2011;Havkin-Frenkel et al., 1999; Walton et al., 2003). Conversely, theenzymatic hydrolysis of glucovanillin to vanillin (i.e. flavor emer-gence) has been greatly studied (Havkin-Frenkel et al., 1999;Kanisawa et al., 1994; Odoux et al., 2003).

Two main hypotheses have been advanced to explain glucova-nillin biosynthesis from p-coumaric acid. The first one proposes anoxidative route with the formation of a coenzyme A ester. In thisroute, the “ferulate pathway”, involves hydroxylation and methyl-ation steps before chain shortening. The second hypothesis pro-poses a non-oxidative route with chain shortening as the first stage,followed by hydroxylation and methylation of the aromatic ring(“benzoate pathway”). Until now, only a few enzyme steps in thephenylpropanoid pathway in vanilla tissue cultures have beenexplored (Pak et al., 2004; Podstolski et al., 2002). Among them, anunusual 4-hydroxybenzaldehyde synthase (4HBS), a cysteine pro-tease, was partially purified in vitro (Podstolski et al., 2002). Thisenzyme converts p-coumaric acid non-oxidatively to p-hydrox-ybenzaldehyde and may play a role in vanillin biosynthesis. Pro-teins, O-methyltransferases 2 and 3 (OMT2 and OMT3), were alsobiochemically and genetically characterized in V. planifolia. Theseenzymes seem to be close to caffeic acid O-methyltransferases,even if their potential in in vivo substrates is not yet clear (Li et al.,2006). According to the authors, these enzymes are likely to beinvolved in the biosynthesis of phenolic and flavonol components,which are part of the vanilla flavor.

Gene expression studies could also be a further approach toexplore variations in the phenylpropanoid pathway in V. planifoliapods. Gene expression analysis helped to understand the signalingand metabolic pathways underlying cellular and developmentalprocesses. Of all the methods for quantifying gene expression,reverse transcription quantitative real-time polymerase chain re-action (qRT-PCR) is considered the most reliable, thanks to its su-perior sensitivity and specificity, evenwith limited amounts of RNA(Van Guilder et al., 2008). Nevertheless, in order to accurately andreliably quantify gene expression, endogenous reference genesmust be selected. Genes encoding transcripts involved in basiccellular metabolism are frequently used for this purpose. The sixmost frequently used reference genes are Actin, b-tubulin, Elonga-tion Factor 1 a (EF1), 18S rRNA, Glyceraldehyde-3-Phosphate Dehy-drogenase (GAPDH) and Ubiquitin (Kumar et al., 2011).

To the best of our knowledge, no research has been carried outto evaluate gene expression over time in V. planifolia pods and,specifically, within the phenylpropanoid pathway. Consequently,the purpose of the present work is to identify PAL and C4H geneorthologs that are actively transcribed in developing pods and thento quantify their expression levels during pod maturation, usingqRT-PCR, as well as for the previously characterized Vp4HBS,

VpOMT2 and VpOMT3 genes. Nevertheless, in order to explorechanges in gene expression, a necessary prerequisite was to iden-tify reliable reference genes as internal normalization control. Inaddition, expression profiling of these key phenylpropanoid genesis discussed in light of accumulation patterns for metabolitesinvolved in flavor precursors, i.e. p-hydroxybenzoic acid, p-hydroxybenzyl alcohol, p-hydroxybenzaldehyde, proto-catechualdehyde, vanillic acid, vanillyl alcohol, and vanillin.

2. Methods

2.1. Plant material

V. planifolia vines were grown in a shade house of the “Coop-érative Provanille” (Saint-André, La Réunion). When flowersappeared, they were hand-pollinated and labeled. Harvest dateswere defined in accordance with the beginning of vanillin and itsglycosyl conjugate (glucovanillin) accumulation in pods (Kanisawaet al., 1994; Havkin-Frenkel et al., 1999). Thus, four pods werecollected every month, from 3 to 8 months after pollination (MAP).They were randomly sampled in duplicate from different plants,immediately frozen in liquid nitrogen and stored at �80 �C untilutilization. These podswere used for metabolomic analysis (Palamaet al., 2009). For the present study, the same pods were used forRNA extraction and gene expression studies.

2.2. RNA extraction and cDNA synthesis

For each stage, frozen pods were ground together to a finepowder in liquid nitrogen with a mortar and a pestle. The poolsobtained were then analyzed in duplicate for RNA extraction.

Total RNA was extracted from 100 mg homogenized plant ma-terial using an RNeasy Plant Mini Kit (Qiagen, Hilden, Germany).Putative genomic DNA contaminationwas eliminated by treatmentwith recombinant DNase I (RNase-Free DNase Set, Qiagen). In orderto increase its quality, total RNA was finally purified and concen-trated using RNeasy MinElute Cleanup columns (Qiagen) accordingto the manufacturer’s instructions. RNA concentration and puritywere estimated by spectrophotometry at 260 and 280 nm. OnlyRNA samples with 260/280 ratios upper 1.9 were used for cDNAsynthesis. RNA integrity was also evaluated by agarose gelelectrophoresis.

cDNA was synthesized from 1 mg of total RNA for each sampleusing the ImProm-II Reverse Transcription System Kit (Promega,Madison, USA) in a final volume of 20 mL, as per the manufacturer’sinstructions. Two independent reverse transcription reactions wereperformed for each sample. Reactions without the reverse tran-scriptase were also performed to check the absence of genomicDNA (No-RT controls). The final cDNA samples were then diluted40-fold prior to use in qRT-PCR.

2.3. Partial cDNA sequences

At the beginning of this work, only a few sequences forV. planifolia and orchids were available in public databases,including VpOMT2 (GenBank ID: DQ400399), VpOMT3 (GenBank ID:DQ400400). Vp4HBS was also available (Havkin-Frenkel et al.,2003). But for classical housekeeping genes such as b-tubulin,Actin, EF1, and other candidate genes like VpPAL1 and VpC4H1, se-quences were not available in libraries. For those genes, degenerateprimers were used to obtain specific PCR fragments.

Primers were designed using the CODEHOP (COnsensus-DEgenerate Hybrid Oligonucleotide Primers) strategy (http://www.bioinformatics.weizmann.ac.il/blocks/codehop.html).

Table 1Comparison of gene sequences from V. planifolia with known related nucleotide sequences.

Gene name Primer sequence (50e30) forwardand reverse

cDNA fragmentsize (bp)

E value Nucleotideidentity (%)

Blast hit (GenBank ID)

b-tubulin CCA GAT CGG CGC Caa rtt ytg ggaGGG CGA AGC Cca cca tra ara a

721 0 81 Cymbidium faberi (JN177714)

Actin GCA TCG TGT CCA Act ggg ayg aya tGGA GGG GCC ACC ACC ttd aty ttc at

784 0 86 Vanda hybrid cultivar Princess Mikasa ’Blue’ (HQ596370)

EF1 GAA GGA GGC CGC Cga rat gaa yaaCCT GCA GGG CCT CGt grt gca tyt c

732 0 86 Mimosa pudica (AB468825)

PAL TGC AAA AAG AAC TGA TCC Gat tyy tna ayg cCGG AAA ATT GGG CAA ACA Tta ryt tnc cda t

866 0 82 Bromheadia finlaysoniana (X99997)

C4H TGT TCA CCT ACA AGG GCc arr aya tgg tCGC CGA AGG GGA TGT ACy bra art crt t

1008 0 81 Sorghum bicolor (XM_002458638)

Table 2Specific primers used for qRT-PCR.

Gene Primer sequence (50e30) forward and reverse Amplicon size (bp) Efficiency (E)a r2

b-tubulin F: TGGGCGAAAGGACACTACACR: CACCCAAAGAATGGCACACT

121 1.99 0.99

Actin F: GGGTTACTCCTTTACGACCACAR: GCTGCTCTTCGCTGTCTCAA

112 1.98 0.97

EF1 F: TTGCTTGCCTTCACTCTTGGR: TTCATCGTACCTTGCCTTGG

96 2.01 0.99

GAPDH F: GCTCCGATGTTTGTTGTTGGR: CTTCTGTGTTGCTGTGGTTGC

177 1.88 0.97

18S rRNA F: GCCTGGAATCTTTCGGTTTGR: GCTCGTCGCACAACAACATC

178 1.96 0.99

VpHBS F: GTGCGTCCAGTTAGCGTTGR: CTGCGTGGTTCACATCCATT

112 1.95 0.98

VpOMT2 F: GGTGGTGATGTGTTCGCTTCR: GCAGTCCTCGTCGTTCCAAT

84 1.86 0.99

VpOMT3 F: GGGAACCACCACTCGCTACR: TGGAGAACCGCATCCTTTAC

147 2.00 0.98

VpC4H1 F: CCTTTCTCAGGGGCTACCTCR: TATCCATCGGCTTCGTGTTC

117 1.97 0.99

VpPAL1 F: GGCGAAGAAACTCCACGAR: AATCTGAGGACCGAGCCATT

94 1.94 0.98

a Serial dilutions of themixed cDNAs from all tested samples were used to calculate the gene-specific PCR amplification efficiency (E) and correlation coefficient (r2) for eachgene. E ¼ 10(�1/slope), according to Bustin et al. (2009).

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314306

The conserved regions of target proteins were determined byaligning their amino acid sequences obtained from the NCBI data-base using ClustalW2 (version 2.1) (http://www.ebi.ac.uk/Tools/services/web/toolform.ebi?tool¼clustalw2). Protein blocks werecreated from the conserved region of proteins using the BlocksMultiple Alignment Processor program (http://bioinformatics.weizmann.ac.il/blocks/process_blocks.html). Retained primershave a melting temperature (Tm) in the range of 60e62 �C, arandom nucleotide composition and are 18e30 bases long. Ex-pected amplicons are 732e1009 bases long (Table 1). Reactionconditions involved 40 cycles of denaturing at 95 �C for 45 s (firstcycle for 5 min), annealing at specific Tm for 1 min and extending at72 �C for 30 s. A mix of cDNA from all pod samples (500 ng) wasused as template. Specific amplification products were thensequenced in double stands. Results were confirmed by BLASTNand BLASTX against the nr database at NCBI to get at least 80%identities to the published counterparts in other plants (Table 1).The resulting sequences were deposited in GenBank (http://www.ncbi.nlm.nih.gov) with the accession numbers KF513172,KF513173,KF513174,KF513175,KF513176 for VpPAL1, VpC4H1, b-tubulin,Actin, EF1, respectively.

2.4. Primer design for quantitative real-time PCR

The partial sequences obtained were used to design primers forquantitative real-time PCR. Primer pairs for qRT-PCR amplificationwere designed using previous sequences obtained from degen-erated primers or from published sequences (GAPDH, 18S rRNA,

Vp4HBS, VpOMT2, VpOMT3) using Primer 3 (v.0.4.0) (http://frodo.wi.mit.edu/primer3/). Retained parameters were a Tm between 60and 62 �C, a primer size between 18 and 22 bp and about 50% GCcontent. Amplicon lengths were optimized to 90e150 bp to ensureoptimal polymerization efficiency and minimize the impact of RNAintegrity on relative quantification of gene expression.

The probability of formation of hairpin structures and primerdimerization was also checked using the Oligo Calculator (version3.26) algorithm (http://www.basic.northwestern.edu/biotools/OligoCalc.html). Homologies of the cDNA sequences with data-base sequences were determined using BLASTX and BLASTN ho-mology searches. The size of PCR products was further checked on2.5% agarose gels (Table 2). In addition, target amplicons weresequenced. Specific primer sequences are given in Table 2.

2.5. PAL and C4H phylogenetic analysis

Homologous sequences of VpPAL1 and VpC4H1 were identifiedusing BLAST (http://www.ncbi.nlm.nil.gov/BLAST). Sequences werealigned using ClustalW2 and phylogenetic trees were built usingthe neighbor-joining method in MEGA (v. 5) software (Tamuraet al., 2011). The replication of bootstrap test was 1000.

2.6. Quantitative real-time PCR

The cDNA samples obtained were analyzed in duplicate usingreal-time PCR. Finally, four biological replicates per stage of poddevelopment were evaluated for each gene tested.

Trifolium pratense (PAL3) (DQ073808)Trifolium pratense (PAL2) (DQ073810)Trifolium pratense (PAL4) (DQ073811)Pisum sativum (PAL2) (D10003)Pisum sativum (PAL1) (D10002)Phaseolus vulgaris L. (M11939)Glycine max (S46988)

Fabaceae

100

100

87

100

54

95

59 Glycine max (S46988)Trifolium pratense (PAL1) (DQ073809)Populus trichocarpa (PAL3) (EU603318)Populus trichocarpa (PAL1) (EU603319)Populus tremuloides (PAL1) (AF480619)

Salicaceae

Prunus avium (PAL1) (AF036948)Rubus idaeus (PAL2) (AF237955) Rosaceae

Populus trichocarpa i(PAL5) (EU603320)100

100100

57

100

59

DICOTS

p p ( ) ( )Populus trichocarpa (PAL4) (EU603322)Populus tremuloides (PAL2) (AF480620)

Salicaceae

FabaceaeMedicago truncatula (XM 003591829)RutaceaeCitrus limon (PAL6) (U43338)RosaceaeRubus idaeus (PAL1) (AF237954)

Petunia x hybrida (PAL1) (AY705976)Petunia exserta (PAL1) (JF793917)100

52 61

55

93

Petunia axillaris (PAL1) (JF793918)SolanaceaeNicotiana attenuata (PAL1) (DQ768746)

Nicotiana attenuata (PAL2) (DQ768747)Capsicum annuum (EU616575)Ageratina adenophora (PAL2) (GQ259805)Ageratina adenophora (PAL1) (GQ259804) Asteraceae

Arabidopsis thaliana (PAL3) (AY528562)A bid i l t b l t (PAL3) (XM 002871035)

100

76

100

99100

Arabidopsis lyrata subsp. lyrata (PAL3) (XM 002871035)Arabidopsis thaliana (PAL4) (AY303130)

BrassicaceaeArabidopsis thaliana (PAL2) (NM 115186)Arabidopsis lyrata subsp. lyrata (PAL2) (XM 002877863)Arabidopsis thaliana (PAL1) (NM 129260)Brassica napus (PAL1-2) (DQ341309)Narcissus tazetta var. chinensis (PAL1) (GU574806)Lycoris radiata (FJ603650) Amaryllidaceae

100

100

10091

Lycoris radiata (FJ603650) y

Phalaenopsis x Doritaenopsis hybrid cultivar (AY281156)Bromheadia finlaysoniana (X99997)Vanilla planifolia VpPAL1

Orchidaceae

Musa acuminata AAA (EU856393)Musa acuminata cultivar Calcutta (EU856394)Triticum aestivum (X99705)Bambusa oldhamii (PAL3) (GU338002)

100

81

10072

93

83

59

MONOCOTS

Bambusa oldhamii (PAL3) (GU338002)Hordeum vulgare subsp. vulgare (AK356545)Phyllostachys dulcis (PAL1) (FJ594738)Phyllostachys parvifolia (PAL1) (FJ594737)Phyllostachys angusta (PAL1) (FJ596643)Bambusa oldhamii (PAL1) (AY450643)Bambusa oldhamii (PAL2) (FJ715635)Oryza sativa Japonica (NM 001062143)

Poaceae

52100

100

97

91100

78

y p ( )Zea mays (BT067796)Zea mays (PAL3) (NM 001111864)Saccharum officinarum (EF189195)Phyllostachys edulis (FJ195650)Bambusa oldhamii (PAL4) (GU592807)Pinus pinaster(PAL2) (AY321089)Pinus pinaster (PAL1) (AY321088) Pinaceae (Outgroup)100

100

99100

90

Fig. 1. Phylogenetic tree depicting the relationship between various PALs. The analysis involved 57 nucleotide sequences aligned using ClustalW and the phylogeny of the PALsequences was constructed using the neighbor-joining method with MEGA 5 (Tamura et al., 2011). The percentage of replicate trees in which the associated taxa clustered togetherin the bootstrap test (1000 replicates) is shown next to the branches. Bootstrap percentages of >50 are given. The Pinus pinaster PAL1 and PAL2 sequences were used to root the tree.GenBank ID was put in brackets.

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314 307

Quantitative RT-PCR was performed with an ABI PRISM 7000Sequence Detection System (Applied Biosystems, Foster City, USA)using SYBR Green detection chemistry. The reactions were pre-pared in a total volume of 13 mL containing 1 mL of diluted cDNA(corresponding to 25 ng of initial total RNA), 0.5 mL of each gene-specific primer (200 nM) and 6.5 mL of 2� SYBR Green Master Mixreagents (Applied Biosystems). The following standard thermalprofile was used for all PCR reactions: polymerase activation(95 �C for 10 min) then amplification and quantification cyclesrepeated 45 times (95 �C for 15 s and 60 �C for 1 min). No-template controls (NTCs) were included for each primer pair.Dissociation curves for each amplicon were analyzed to verify thespecificity of each amplification reaction. A melting curve wasgenerated for each sample to ensure the purity of the amplifiedproducts.

A standard curve was obtained using five serial dilutions ofpooled cDNAs from all tested samples as the template andwas usedto calculate the correlation coefficient (r2) for each gene. Gene-specific PCR amplification efficiency (E) for each gene was calcu-lated as follows: E ¼ 10(�1/slope) (Bustin et al., 2009). The primer

sequences, expected amplicon lengths, r2 and PCR efficiencies of allPCR products were provided in Table 2.

Data were analyzed using the SDS 2.1 software (Applied Bio-systems) which performed the baseline correction and automati-cally calculated the threshold for determining cycle quantification(Cq) values for each reaction, i.e. the fractional cycle number wherefluorescence increases above the threshold.

2.7. Determination of expression stability of reference genes

Gene expression levels were calculated for all reference genesbased on the number of Cq. To analyze expression stability and toidentify the most suitable reference genes, the two freely availablestatistical algorithms, geNorm (Vandesompele et al., 2002) andNormFinder (Andersen et al., 2004) were used in accordance withthe authors’ recommendations.

Relative expression levels were analyzed using the REST(Relative Expression Software Tool) 2009 software v. 2.0.13 (Pfafflet al., 2002). Briefly, relative expression was calculated usingqPCR efficiency (E) and the cycle quantification value difference

Brassicaceae

Brassica napus (C4H1) (DQ485130)

Brassica napus (C4H1) (DQ485129)

Brassica rapa (AB300317)

Brassica napus (C4H2) (DQ485131)

Isatis tinctoria (GU014562)

100100

53

54

100

Arabidopsis lyrata subsp. lyrata (XM 002881068)

Arabidopsis thaliana (NM 128601)

Arabidopsis thaliana (U71080)

Arabidopsis thaliana (U37235)

Medicago truncatula (DQ335792)

Medicago sativa (L11046)100

99100

100

52Fabaceae

Amaryllidaceae

A t

Medicago sativa (L11046)

Trifolium pratense (EU574001)

Cicer arietinum (AJ007449)

Allium cepa (AY541032)

Allium sativum (GU456382)

Zinnia elegans (U19922)

100

52

100

63

Asteraceae

Rutaceae

DICOTSHelianthus tuberosus (Z17369)

Echinacea angustifolia (EU676019)

Citrus sinensis (C4H2) (AF255014)

Citrus x paradisi (AF378333)

Ruta graveolens (AF548370)

Populus tremuloides (C4H1 1) (DQ522292)

100

100

100

98

Salicaceae

Populus tremuloides (C4H1-1) (DQ522292)

Populus trichocarpa (C4H1) (XM 002325602)

Populus tremuloides (U47293)

Populus tremuloides (C4H2-2) (DQ522295)

Populus tremuloides (C4H2-1) (DQ522294)

Prunus avium (C4H1) (GU990522)100

100

100

100

100

7158

Rosaceae

Fabaceae

Prunus armeniaca (HM204477)

Malus x domestica (C4H1) (DQ075002)

Rubus occidentalis (FJ554629)

Rubus coreanus (EU123531)

Glycine max (FJ968526)

Leucaena leucocephala (GU183363)

10088

100

99

Orchidaceae

Poaceae MONOCOTS

Leucaena leucocephala (GU183363)

Acacia auriculiformis x A. mangium (EU275980)

Vanilla planifolia VpC4H1

Phyllostachys edulis (EU780142)

Zea mays (NM 001155686)

Sorghum bicolor (XM 002458638)

100

100

100

79

Poaceae

Pinaceae (outgroup)

Bambusa oldhamii (GU188741)

Hordeum vulgare subsp. vulgare (AK250541)

Triticum aestivum (AK332115)

Pinus taeda (AF096998)

Pinus pinaster (C4H2) (JN013973)100

10097

100

Fig. 2. Phylogenetic tree depicting the relationship between various C4Hs. The analysis involved 43 nucleotide sequences aligned using ClustalW and the phylogeny of the C4Hsequences was constructed using the neighbor-joining method with MEGA 5 (Tamura et al., 2011). The percentage of replicate trees in which the associated taxa clustered togetherin the bootstrap test (1000 replicates) is shown next to the branches; those below 50 are omitted. The C4H sequences of two Pinaceae were used to root the tree. GenBank ID wasput in brackets.

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314308

(DCq) of sample vs. control for both target and reference genes.The geometric mean of Cq (CqRef) for the identified referencegenes, i.e. Actin and EF1, was used to calculate relative geneexpression ratios (Pfaffl et al., 2002). The formula proposed bythese authors is as follows: Ratio ¼ (Etarget)DCqtarget

(mean control�mean

sample)/(ERef)DCqRef(mean control�mean sample).

2.8. Statistical analysis

All results were analyzed using the Statistica software package(v. 6.1, 1997 for Microsoft Windows). Four statistical methods werecarried out: one-way ANOVA, principal component analysis (PCA),linear regression, and non-linear regression.

One-way ANOVA allowed means to be compared. To avoid falsepositives, the significance p value was arbitrarily low (p < 0.0001).When the F test was significant, a Bonferroni comparison of meanswas carried out.

Principal component analysis (PCA; normalized Varimax) wasapplied to the gene expression and metabolite accumulationdataset recorded over time to highlight the reduced number of

factors involved in the generalized variance. Expressed mathe-matically, principal component analysis consisted of a translation(date are centered, and often reduced), followed by an axis rotation.The main objective of the method is to obtain K new independent,i.e. orthogonal, variables from N initial ones (K< N). The choice of Kis determined using Kayser’s criteria (all components with eigen-value <1 represent noise; all components with eigenvalue >1 canbe interpreted). Each component is defined by its eigenvector andtheir correlation with initial variables allows their interpretation.Moreover, each component explains a part of the variance for eachvariable (i.e. relative contribution, r2). Finally, the Varimax optionallows non-normal variables to be used, leading to robustcomponents.

The relationship between VpPAL1 expression and pod develop-ment (expressed in months after flowering) was tested using linearregression with classical least square criteria.

Finally, the relationship between vanillin content and VpPAL1expressionwas analyzed using the non-linear least square method.Fitting was carried out using logistic model y ¼ M/(1 þ ea(x�b)) andthe LevenbergeMarquardt algorithm.

0.0000.0200.0400.0600.0800.1000.1200.140

Stab

ility

valu

e

Decreasing expression stability --->

0.0990.111

0.167

0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

0.160

0.180

V2/3 V3/4 V4/5Pairwise Variations (V)

B

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

b-tubulin 18S rRNA GAPDH ActinEF1

Averag

e e

xp

ressio

n s

tab

ility (

M)

<::::: Least stable genes Most stable genes ::::>

A

Fig. 3. Reference genes analyses. (A) GeNorm analysis for the reference genes. Gene expression stability and ranking of candidate reference genes in accordance with geNormanalysis. A lower average expression stability value (M) indicates more stable expression. Two technical duplicates of four biological replicates of V. planifolia pods across 3, 4, 5, 6, 7and 8 months after hand-pollination of flowers were used. The optimal number of reference genes for normalization by pairwise variation (V) was determined by geNorm analysis.A variation of <0.15 indicates no significant contribution of an additional reference gene to the normalization factor. In our experimental set, the use of the two most stablereference genes should allow a reliable normalization. (B) NormFinder analysis for the reference genes. A lower stability value indicates the most stable expression in all samples.

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314 309

3. Results

3.1. PAL and C4H phylogenetic analysis

Each degenerated primer pair gave a single amplicon and partialsequences for the genes were obtained by sequencing. All resultsobtained were confirmed by BLASTN or BLASTX against nr databaseat NCBI to get at least 80% identities to the published data in otherplants (Table 1). Consequently, the initial sequences for PAL(VpPAL1) and C4H (VpC4H1) are now available for V. planifolia.

To better understand the phylogenetic relationship betweenVpPAL1 or VpC4H1 and related genes from other plant species,alignments with sequences with a higher score of similarity wereperformed. Specifically, 57 PAL sequences and 43 C4H sequenceswere included to generate a maximum likelihood phylogenic treewith MEGA 5 program (Figs. 1 and 2).

3.1.1. VpPAL1The gymnosperm PAL1 and PAL2 sequences from Pinus pinaster

were selected as outgroup. Indeed, gymnosperms are considered tobe ancestral to angiosperms, on the basis of morphological char-acters and 18S rRNA sequences (Chaw et al., 1997). So, a single tree,considered as the most parsimonious tree, was obtained (Fig. 1). Asexpected, VpPAL1 clustered with orchids Phalaenop-sis�Doritaenopsis hybrid cultivar and Bromheadia finlaysoniana.Within the Orchidaceae family, sequence identity was evaluated atbetween 77 and 78%. VpPAL1 was also phylogenetically closelyrelated to the Poaceae family (Fig. 1).

3.1.2. VpC4H1The pine (Pinus taeda) C4H sequence was used as outgroup. The

phylogenetic tree showed two separate groups with high boot-strap values (99%). Dicot and monocot C4H genes were clearlyseparated in monophyletic groups (Fig. 2). VpC4H1 gene fromV. planifolia was included, as expected, within the group ofmonocots. Within this clade, C4H sequences identity with vanillacDNA ranged between 76% (for Phyllostachys edulis) and 81% (forBambusa oldhamii).

3.2. Expression stability analysis

The geNorm and NormFinder algorithms were used to evaluatethe stability of the five potential reference genes.

3.2.1. geNorm analysisQuantities of the five potential reference genes, calculated for

each biological sample were used in geNorm to determineexpression stability values (M). The pairwise variation (Vn/Vnþ1) ofeach gene with the others is also generated (Vandesompele et al.,2002). In the end, the highest stability value (M ¼ 0.217) wasfound for EF1 and actin whereas the lowest stability value(M ¼ 0.573) was observed for b-tubulin. The V2/3 was lower thanthe cut-off level (0.15) (Vandesompele et al., 2002), belowwhich noadditional reference gene is required (Fig. 3A). Therefore, the bestcombination according to geNorm is actin and EF1. Inclusion ofadditional reference gene (GADPH) increases the pairwise variation

Fig. 4. Expression ratios of VpPAL1, Vp4HBS, VpC4H1, VpOMT2 and VpOMT3 during V. planifolia pod maturation. Expression levels were measured by qRT-PCR using the primers listedin Table 2. Sample duplicates were used in all qPCR runs (four biological replicates). Ratios refer to up- or down-regulation of gene expression for each month compared to the thirdmonth after hand-pollination (control). Actin and EG1 were used as reference genes for normalization. P-values were estimated based on the Bonferroni’s test. The * symbol at agiven time indicates the ratio of gene transcript to control was significant at P < 0.01.

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314310

(V2/3 ¼ 0.099 and V3/4 ¼ 0.111). We therefore deemed two refer-ence genes were optimal for our experiments.

3.2.2. NormFinder analysisTo confirm geNorm results, expression datawere evaluatedwith

the NormFinder algorithm. As shown in Fig. 3B, the lowest stabilityvalue, obtained for Actin, indicates the most stably expressed gene(stability value ¼ 0.052; Fig. 3B), followed by EF1 (0.063), 18S rRNA(0.093), GAPDH (0.101) and b-tubulin (0.120). Therefore, the moststable genes according to NormFinder were actin and EF1.

In conclusion, both geNorm and NormFinder identified the sametwo stably expressed genes, even if the ranking of other genes wasslightly different (Fig. 3).

3.3. Expression of phenylpropanoid genes during pod maturationand relationship with phenolic compounds accumulation

Transcripts of the genes studied were detected in all samples atvarious amounts using qRT-PCR (Fig. 4).

One-way ANOVA showed no significant variation in expressionover time for VpOMT2 and VpOMT3. Using Bonferroni’s test(p< 0.01) as criteria, gene expression of Vp4HBS increased between3 and 4 months after pollination (MAP), but was constant from 4 to8 MAP (Fig. 4). In contrast, the expression level of VpC4H1 washighest at earlier stages (3 MAP), while its expression was ratherdown-regulated during pod maturation. In fact, only VpPAL1expression exhibited large variations over time. A linear increasewas observed from 3 to 7 MAP: y ¼ 1.234��2.72 (r ¼ 0.994;p ¼ 0.0006). Expression then seemed to decrease slightly after 7MAP. In addition, VpPAL1 expression was more than 6-fold higherin 7-month-old pods than in 3 month-old fruits.

To analyze the effect of maturation on gene expression andaccumulation of related phenolic compounds, a PCA analysis wascarried out including time, gene expression andmetabolite content.The metabolites, which amounts were previously determined onthe same biological samples using LCeMS (Palama et al., 2009),were p-hydroxybenzoic acid (HBAc), p-hydroxybenzyl alcohol(HBAlc), p-hydroxybenzaldehyde (HBAld), protocatechualdehyde(PCAld), vanillin (Van), vanillic acid (VanAc) and vanillyl alcohol

Table 3Principal components of the diversity of gene expression and metabolite accumu-lation profiles in five different stages of vanilla pod development. Numbers withoutbrackets are correlation coefficients. Coefficients highly correlated to the time arewritten in bold. Numbers in brackets are relative contributions expressed as a per-centage (10� r2) computed only for significant correlations.

Time PC 1 PC 2 PC 3

0.925 0.177 0.246

VpPAL1 0.955 (90.3) 0.215 0.124Vp4HBS 0.491 0.559 0.422VpC4H1 �0.036 �0.095 �0.980 (95.0)VpOMT2 �0.046 0.957 (91.6) 0.153VpOMT3 0.024 0.981 (96.4) �0.076HBAc 0.973 (94.7) 0.116 0.124HBAlc �0.682 �0.601 �0.409HBAld 0.958 (91.8) 0.007 0.277PCAld 0.483 0.516 0.424Van 0.983 (96.6) 0.047 0.090VanAc 0.965 (93.1) 0.181 0.044VanAlc 0.966 (93.3) �0.050 �0.158

Eigen value 7.407 (57.0) 2.956 (22.7) 1.712 (13.2)

Metabolites abbreviations. HBAc: p-hydroxybenzoic acid; HBAlc: p-hydroxybenzylalcohol; HBAld: p-hydroxybenzaldehyde; PCAld: protocatechualdehyde; Van:vanillin; VanAc: vanillic acid; and VanAlc: vanillyl alcohol.

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314 311

(VanAlc). Three components accounted for 92.8% of the generalizedvariance present in the dataset. The first was highly correlated totime and VpPAL1 expression (Table 3). It accounted for 91.2% ofVpPAL1 expression variation over time. Strikingly, this componentwas also closely correlated with HBAc, HBAld, VanAc, VanAlc andVan content (Table 3, Fig. 5). This component could therefore beinterpreted as the consequence of maturation on VpPAL1 expres-sion and vanillin accumulation, and this finding strongly suggests aclose relationship between VpPAL1 expression and secondarymetabolite production. The first component also accounted for 24%of Vp4HBS expression (not statistically significant), but less than 1%of VpC4H1, VpOMT2 and VpOMT3 transcript level variations(Table 3). The second component was closely correlated toexpression of VpOMT2 and VpOMT3 and could be interpreted asvariations between pods, independent of time. This componentcontributed for 31.2% of variations in Vp4HBS expression. Thiscomponent was also correlated to PCAld and HBAlc. In particular,PCAld content increased when HBAlc content decreased. The thirdcomponent was merely correlated with variations in VpC4H1

Fig. 5. Principal component analysis (PCA) of the distribution of gene expression andmetabolite accumulation profiles at five development stages of vanilla pod. The plotrepresents principal component 1 (PC1) versus principal component 2 (PC2).

expression, independent of the two first factors. Consequently, PCAled to synthetic results, coherent with ANOVA results. Significantly,the role of VpPAL1 in maturation was confirmed. In contrast, othergenes should have little (Vp4HBS) or no role (VpC4H1, VpOMT2 andVpOMT3) in flavor development over time.

The relationship between VpPAL1 expression and phenoliccompound accumulation was further investigated using non-linearregression between transcript andmetabolite profiles. Interestingly,more than 99% of vanillin content could be accounted for VpPAL1gene up-regulation (Fig. 6). Moreover, other metabolite variationsduring podmaturationmay rely on VpPAL1 gene expression. Indeed,highly positive correlations existed between VpPAL1 gene expres-sion and HBAc (r ¼ 0.921), HBAld (r ¼ 0.98), VanAc (r ¼ 0.884),VanAlc (r ¼ 0.792) and Van (r ¼ 0.96). Weak correlations wereobserved between VpPAL1 expression and other metabolites.

4. Discussion

4.1. Identification of reference genes for qRT-PCR in vanilla pods

Normalization with multiple reference genes is actually therule and guidelines have been provided to make qPCR experi-ments reliable (Bustin et al., 2009). In the last decade, manystudies on the validation of reference genes have been carried outon various species (Expósito-Rodríguez et al., 2008; Reid et al.,2006). Nevertheless, there is no information in the literatureregarding the selection of reference genes for gene expressionanalyses in vanilla.

In a recent study on the caffeoyl CoA O-methyltransferase-likegene family from V. planifolia, b-tubulin was used as an internalstandard, assuming its expression in all organs was uniform(Widiez et al., 2011). However, no previous testing of the expressionstability has been performed. The present work identified reliablereference genes with stable expression during pod maturation. Theexpression levels of five frequently used housekeeping genes (b-tubulin, actin, Elongation Factor 1 a (EF1), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and 18S rRNA) were assessedusing quantitative real-time RT-PCR. Based on statistical algo-rithms, the most stably expressed reference genes during devel-opmental stages of vanilla pods were actin and EF1. Additionally, wefound the least stable gene to be b-tubulin. This finding highlightedthe importance of the selection of reference genes in geneexpression analyses.

Fig. 6. Correlation between vanillin concentration and PAL expression level. Vanillinconcentration was evaluated by LCeMS analysis as described in Li et al. (2006).Numbers situated above the curve indicate months after pollination (3e7 MAP).

Fig. 7. Proposed pathways to vanillin taking together the expression of phenylpropanoid genes with the accumulation of phenolic compounds during pod maturation, adapted fromKanisawa et al. (1994) and Rasmussen and Dixon (1999). Vanillin would be arising from two separate routes from p-coumaric acid via non-oxidative or oxidative pathways(benzoate or ferulate pathways, respectively, to the right of the figure) or from t-cinnamic acid by non-oxidative chain shortening (Benzoic acids pathway, on the left of the figure).Genes with an observed significant up-regulation are in bold. OMT represents OMT2 and OMT3. Metabolite content during pod maturation is shown together with gene expressionprofile. Because of the toxicity of aglycone forms, this pathway is based on glucosides identified in extracts of green pods. The supposed biosynthetic origin of p-hydrox-ybenzaldehyde from t-cinnamic acid is indicated by dotted lines and a question mark, as described by Rasmussen and Dixon (1999). Abbreviations: PAL, phenylalanine ammonia-lyase; 4CL, 4-coumarate coenzyme A ligase; C4H, cinnamate 4-hydroxylase; OMT, O-methyltransferase; and HBS, 4-hydroxybenzaldehyde synthase.

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314312

4.2. VpPAL1 and VpC4H1: two new phenylpropanoid pathwaygenes expressed in the vanilla pod

This is the first report on a partial PAL sequence (VpPAL1) and apartial C4H sequence (VpC4H1) obtained from degenerated

primers in vanilla pods. The close relationship of the VpPAL1sequence with other PAL present in orchids was confirmed byphylogenetic analysis. Nevertheless, as in many plant species,multiple isoforms of PAL enzyme are expected to co-exist, withvarious implications in plant developmental programs and stress-

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314 313

responses (Dixon et al., 2002; Wanner et al., 1995). As an example,four PAL ortholog genes were described in the monocot B. oldhamii(Hsieh et al., 2010) and at least 4 non-redundant PAL transcriptsequences could be detected in the orchid Phalaenopsis aphroditetranscriptome (Su et al., 2013). If a major PAL gene has been pres-ently characterized in V. planifolia, the relative importance of otherputative PAL orthologs during pod maturation and vanillinbiosynthesis remains to determine.

cDNA sequences encoding C4H have been identified innumerous plants and were shown to be very conserved (Ehltinget al., 2006). Two classes of C4H co-exist even if the predominantdicots isoform is class I (only one copy) whereas monocots mainlymaintained class II (Batard et al., 2000). Both classes (class I andclass IIB) can be found in the same species, but, the two forms havebeen thus far only found in dicots (Nedelkina et al., 1999). As aconsequence, VpC4H1 is included in the typical clade IIA ofmonocots, and the exact C4H gene copy number remains to bedetermined for V. planifolia. Indeed, at least 2 non-redundant C4Htranscript sequences were described in Phalaenopsis sp. tran-scriptome (Su et al., 2013).

4.3. Changes over time in vanillin content are correlated totranscriptional variation of VpPAL1 gene

Temporal expression patterns of VpPAL1 and VpC4H1 wereinvestigated during pod development. VpPAL1 gene expressionwasgradually up-regulated during development (up to 6-fold higher in7 month-old pods than in 3 month-old pods). A rise in PAL geneexpression and PAL enzyme activity was frequently described inother developing fruits (Given and Venis, 1988; Griesser et al.,2008; Kumar and Ellis, 2001). Furthermore, the mRNA levelsencoding PAL genes were shown to increase in concurrently withphenolic compounds as flavonoids (Griesser et al., 2008). In fact,during fruit ripening, PAL genes were described to be key regulatorygenes for important biosynthetic pathways involved in the deter-mination of fruit quality (Griesser et al., 2008; Kumar and Ellis,2001).

In V. planifolia, the content of vanillin and related flavor pre-cursors (p-hydroxybenzoic acid, p-hydroxybenzaldehyde, vanillylalcohol and vanillic acid) also increased during pod maturation andthis is consistent with previous studies (Dignum et al., 2004;Kanisawa et al., 1994; Palama et al., 2009; Tokoro et al., 1990).Transcript accumulation of VpPAL1 during pod development thuscould result in the increase of related metabolites observed in themeantime, especially vanillin and glucovanillin. Indeed, the verystrong positive correlation, shown in Fig. 6, suggests a causal rela-tionship between the expression level of this PAL ortholog and thevanillin biosynthesized during pod maturation. Such results are inagreement with previous works demonstrating a positive rela-tionship between the accumulation of secondary metabolites andgenes activity (Singh et al., 2009). In conclusion, the VpPAL1 geneisoform identified by qRT-PCR using degenerated primers would bethe major gene ortholog among PAL gene family to be involved invanillin biosynthesis in vanilla pods.

In contrast to VpPAL1, there was a slight decrease in VpC4H1transcript abundance during pod maturation. In addition, thistemporal variation was independent of VpPAL1 expression(Table 3), and consequently, should play no role in vanillin accu-mulation. At this stage, two hypotheses can be proposed: (i) RT-PCRusing degenerated primers failed to catch the major C4H orthologplayer in vanillin biosynthesis during pod development, or (ii) thelevel of VpC4H1 transcript would be sufficient to allow the pathwayto operate without coordinated regulation (Blount et al., 2000).Transcriptome-wide analysis of vanilla pod maturation should helpclarifying this point. However, VpPAL1 could be the dominant

transcriptional control point for the vanillin biosynthetic pathwayas it has been previously demonstrated in other species for otherphenolics (Blount et al., 2000; Sircar and Mitra, 2008).

4.4. HBAld accumulation in vanilla pod is not related to Vp4HBSgene expression

4HBS has been much studied for its role in benzoic acid syn-thesis in species such as Daucus carota (Sircar and Mitra, 2008).4HBS was specific in converting p-coumaric acid to p-hydrox-ybenzaldehyde (HBAld) (Podstolski et al., 2002; Sircar and Mitra,2008).

The Vp4HBS gene was identified (Havkin-Frenkel et al., 2003).Various 4HBS enzymatic activities were observed in organs of va-nilla (Podstolski et al., 2002). In particular, the time of highest ac-tivity in pods was observed at 8 MAP. Surprisingly, HBAld contentreached its maximum level at the 11th MAP.

In our work, Vp4HBS gene expression was observed to increase(about 1.4-fold) between 3 and 4 MAP and this subsequentlyremained stable up to the last developmental stage studied (8MAP). Thus, Vp4HBS transcript levels did not show, after 3 MAP, anincrease similar to the HBAld accumulation in pods (Palama et al.,2009). These observations are very far from the relationship be-tween VpPAL1 expression and vanillin level. In fact, HBAld accu-mulation was independent of Vp4HBS expression (Table 3). Thisdiscrepancy is similar to the observations of Podstolski et al. (2002)on 4HBS activity and HBAld content. In fact, Vp4HBS would beinvolved in another pathway than the one of vanillin biosynthesis.As described in other organs (Havkin-Frenkel et al., 2003), Vp4HBSexpression should be associated with production of other benzoateacid derivatives. The existence of multiple isoforms for the Vp4HBSgene in vanilla pods could explain the absence of correlation be-tween HBAld and Vp4HBS transcript levels. Thus, another isoformthan the one studied here may be implicated in vanillin biosyn-thesis. This conclusion is supported by the higher level of activityobserved for this enzyme in roots and embryos rather than in pods(Podstolski et al., 2002).

Another possible explanation could involve the production of p-hydroxybenzaldehyde in vanilla pod via another pathway. Thiscompound could also come from t-cinnamic acid through a 4-coumarate: coenzyme A ligase (4CL) enzyme (Fig. 7), as describedfor other species (Rasmussen and Dixon, 1999; Wildermuth, 2006).This supports the multiple pathways to vanillin (Dixon, 2011;Kanisawa et al., 1994). This biosynthesis route could also beinvolved in producing phenolic compounds (flavor precursors) invanilla pods. Further molecular analyses will be required to confirmour findings. HBAld glucoside could also originate from p-hydrox-ybenzyl alcohol glucoside. Indeed, the decrease of p-hydroxybenzylalcohol glucoside during pod maturation has been detected by LCeMS analysis (Palama et al., 2009), whereas the level of p-hydrox-ybenzaldehyde increased (Fig. 7). This conclusion goes along withthe report of previous authors (Kanisawa et al.,1994). It is noted thatmany previous authors also reported the release of p-hydrox-ybenzaldehyde from related b-glucosides (Walton et al., 2003).

4.5. OMT2 and OMT3 expression levels do not change during vanillapods maturation

The third group included genes encoding for OMT2 and OMT3described by Li et al. (2006). Our results showed that VpOMT2 andVpOMT3 are really expressed in vanilla pods. Moreover, theexpression pattern of VpOMT2 was closely correlated with theexpression pattern of VpOMT3. This expression did not fluctuateduring pod maturation, suggesting another metabolic pathway.This could be related to the low level of these enzymes observed in

I. Fock-Bastide et al. / Plant Physiology and Biochemistry 74 (2014) 304e314314

pod tissue. A higher level of enzymatic activity was detected inroots and flowers (Li et al., 2006). Consequently, VpOMT2 andVpOMT3 are probably not implicated in vanillin biosynthesis. Thesefindings are supported by the absence of caffeic and ferulic acids,and their glucosides, in pods using LCeMS analysis (Palama et al.,2009).

In conclusion, our study gives the first, through partial, insightinto phenylpropanoid metabolism in vanilla during pod develop-ment using analysis of gene expression. We highlighted the centralrole of a PAL gene ortholog in vanillin biosynthesis, called VpPAL1.Concerning C4H, there is an obligate role for this enzyme throughboth ferulate and benzoate pathways. However, the C4H orthologdirectly involved needs to be identified. In case VpC4H1 is involved,the low transcriptional control evidenced in this report suggestspost-transcriptional regulations. Finally, taking together the effectof maturation on gene expression and accumulation of relatedphenolic compounds, some genes, namely VpOMT2, VpOMT3 andVp4HBS, appeared not implicated in vanillin synthesis pathway.

Acknowledgments

The authors thank Bertrand Côme and the Coopérative Prova-nille (Saint-André, Reunion Island) for providing the plantmaterials.

Author contribution: I.F.B designed the study, performed exper-iments, analyzed data and wrote the manuscript; T.L.P., S.B., A.L.performed experiments, discussed the results and comment on themanuscript; M.N. analyzed data, performed statistical analysis andwrote the manuscript; T.J. analyzed data and wrote the manuscript.All authors read and approved the manuscript.

Appendix A. Supplementary data

Supplementary data related to this article can be found at doi:10.1016/j.plaphy.2013.11.026.

References

Andersen, C.L., Jensen, J.L., Ørntoft, T.F., 2004. Normalization of real-time quanti-tative reverse transcription-PCR data: a model-based variance estimationapproach to identify genes suited for normalization, applied to bladder andcolon cancer data sets. Cancer Res. 64, 5245e5250.

Batard, Y., Hehn, A., Nedelkina, S., Schalk, M., Pallett, K., Schaller, H., Werck-Reichhart, D., 2000. Increasing expression of P450 and P450-reductase proteinsfrom monocots in heterologous systems. Arch. Biochem. Biophys 379, 161e169.

Blount, J.W., Korth, K.L., Masoud, S.A., Rasmussen, S., Lamb, C., Dixon, R.A., 2000.Altering expression of cinnamic acid 4-hydroxylase in transgenic plants pro-vides evidence for a feedback loop at the entry point into the phenylpropanoidpathway. Plant Physiol. 122, 107e116.

Bustin, S.A., Benes, V., Garson, J.A., Hellemans, J., Huggett, J., Kubista, M., Mueller, R.,Nolan, T., Pfaffl, M.W., Shipley, G.L., Vandesompele, J., Wittwer, C.T., 2009. TheMIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin. Chem. 55, 611e622.

Chaw, S.M., Zhaekikh, A., Sung, H.M., Lau, T.C., Li, W.H., 1997. Molecular phylogeny ofextant gymnosperms and seed plant evolution: analysis of nuclear 18S rRNAsequences. Mol. Biol. Evol. 14, 56e68.

Dignum, M.J.W., van der Heijden, R., Kerler, J., Winkel, C., Verpoorte, R., 2004.Identification of glucosides in green beans of Vanilla planifolia Andrews andkinetics of vanilla b-glucosidase. Food Chem. 85, 199e205.

Dixon, R.A., 2011. Vanillin biosynthesis e not as simple as it seems? In: Havkin-Frenkel, D., Belanger, F.C. (Eds.), Handbook of Vanilla Science and Technology,first ed. Blackwell Publishing Ltd, New York, pp. 292e298.

Dixon, R.A., Achnine, L., Kota, P., Liu, C.J., Reddy, M.S.S., Wang, L., 2002. The phe-nylpropanoid pathway and plant defense e a genomics perspective. Mol. PlantPathol. 3, 371e390.

Ehlting, J., Hamberger, B., Million-Rousseau, R., Werck-Reichhart, D., 2006. Cyto-chromes P450 in phenolic metabolism. Phytochem. Rev. 5, 239e270.

Expósito-Rodríguez, M., Borges, A.A., Borges-Pérez, A., Pérez, J.A., 2008. Selection ofinternal control genes for quantitative real-time RT-PCR studies during tomatodevelopment process. BMC Plant Biol. 8, 131e143.

Given, N.K., Venis, D.M.A., 1988. Grierson, purification and properties of phenylal-anine ammonia lyase from strawberry fruit and its synthesis during ripening.J. Plant Physiol. 133, 31e37.

Griesser, M., Hoffmann, T., Bellido, M.L., Rosati, C., Fink, B., Kurtzer, R., Aharoni, A.,Munoz-Blanco, J., Schwab, W., 2008. Redirection of flavonoid biosynthesisthrough the down-regulation of an anthocyanidin glucosyltransferase inripening strawberry fruit. Plant Physiol. 146, 1528e1539.

Havkin-Frenkel, D., Podstolski, A., Dixon, R.A., 2003. Vanillin biosynthetic pathwayenzyme from Vanilla Planifolia. U.S. Patent Appl. Publ. 49 pp., Cont. -in-part ofU.S. Ser. N� 462,576.

Havkin-Frenkel, D., Podstolski, A., Witkowska, E., Molecki, P., Mikolajczyk, M., 1999.Vanillin biosynthetic pathways: an overview. In: Fu, et al. (Eds.), Plant Cell andTissue Culture for the Production of Food Ingredients. Kluwer Academic/Plenum Publishers, New York, pp. 35e43.

Hsieh, L.H., Ma, G.J., Yang, C.C., Lee, P.D., 2010. Cloning, expression, site-directedmutagenesis and immunolocalization of phenylalanine ammonia-lyase inBambusa oldhamii. Phytochemistry 71, 1999e2009.

Kanisawa, T., Tokoro, K., Kawahara, S., 1994. Flavor development in the beans ofVanilla planifolia. In: Kurihara, K., Suzuki, N., Ogawa, H. (Eds.), Proceedings ofthe International Symposium. Springer, Tokyo, Japan, pp. 268e270.

Kumar, A., Ellis, B.E., 2001. The phenylpropanoid ammonia-lyase gene family inraspberry. Stucture, expression, and evolution. Plant Physiol. 127, 230e239.

Kumar, V., Sharma, R., Trivedi, P.C., Govind, K.V., Khandelwal, V., 2011. Traditionaland novel references towards systematic normalization of qRT-PCR data inplants. Austr. J. Crop Sci. 5, 1455e1468.

Li, H.M., Rotter, D., Hartman, T.G., Pak, F.E., Havkin-Frenkel, D., Belanger, F.C., 2006.Evolution of novel O-methyltransferases from the Vanilla planifolia caffeic acidO-methyltransferase. Plant Mol. Biol. 61, 537e552.

Nedelkina, S., Jupe, S., Blee, K., Schalk, M., Werck-Reichhart, D., Bolwell, G.P., 1999.Novel characteristics and regulation of a divergent cinnamate 4-hydroxylase(CYP73A15) from french bean: engineering expression in yeast. Plant Mol.Biol. 39, 1079e1090.

Odoux, E., Chauwin, A., Brillouet, J.M., 2003. Purification and characterization ofvanilla bean (Vanilla planifolia Andrews) b-D-glucosidase. J. Agric. Food Chem.51, 3168e3173.

Pak, F.E., Gropper, S., Dai, W.D., Havkin-Frenkel, D., Belanger, F.C., 2004. Character-ization of a multifunctional methyltransferase from the orchid Vanilla planifolia.Plant Cell Rep. 22, 959e966.

Palama, T.L., Khatib, A., Choi, Y.H., Payet, B., Fock, I., Verpoorte, R., Kodja, H., 2009.Metabolic changes in different developmental stages of Vanilla planifolia pods.J. Agric. Food Chem. 57, 7651e7658.

Pfaffl, M.W., Horgan, G.W., Dempfle, L., 2002. Relative expression software tool(REST�) for group-wise comparison and statistical analysis of relative expres-sion results in real-time PCR. Nucl. Acids Res. 30, e36.

Podstolski, A., Havkin-Frenkel, D., Malinowski, J., Blount, J.W., Kourteva, G.,Dixon, R.A., 2002. Unusual 4-hydroxybenzaldehyde synthase activity from tis-sue cultures of the vanilla orchid Vanilla planifolia. Phytochemistry 61, 611e620.

Rasmussen, S., Dixon, R.A., 1999. Transgene-mediated and elicitor-induced pertur-bation of metabolic channeling at the entry point into the phenylpropanoidpathway. Plant Cell 11, 1537e1551.

Reid, K.E., Olsson, N., Schlosser, J., Peng, F., Lund, S.T., 2006. An optimized grapevineRNA isolation procedure and statistical determination of reference genes forreal-time RT-PCR during berry development. BMC Plant Biol. 6, 27e38.

Singh, K., Kumar, S., Rani, A., Gulati, A., Ahuja, P.S., 2009. Phenylalanine ammonia-lyase (PAL) and cinnamate 4-hydroxylase (C4H) and catechins (flavan-3-ols)accumulation in tea. Funct. Integr. Genomic 9, 125e134.

Sircar, D., Mitra, A., 2008. Evidence for p-hydroxybenzoate formation involvingenzymatic phenylpropanoid side-chain cleavage in hairy roots of Daucus carota.J. Plant Physiol. 165, 407e414.

Su, C.L., Chao, Y.T., Yen, S.H., Chen, C.Y., Chen, W.C., Chang, A.Y.C., Shih, M.C., 2013.Orchidstra: an integrated orchid functional genomics database. Plant Cell.Physiol. 54, e11.

Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., Kumar, S., 2011. MEGA5:molecular evolutionary genetics analysis using maximum likelihood, evolu-tionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28, 2731e2739.

Tokoro, K., Kawahara, S., Amano, A., Kanisawa, T., Indo, M., 1990. Glucosides invanilla beans and changes of their contents during maturation. In: Bessiere, Y.,Thomas, A.F. (Eds.), Flavour Science and Technology. John Wiley and Sons, NewYork, pp. 73e76.

Van Guilder, H.D., Vrana, K.E., Freman, W.M., 2008. Twenty-five years of quantita-tive PCR for gene expression analysis. Biotechniques 44, 619e626.

Vandesompele, J., De Preter, K., Pattyn, F., Poppe, B., Van Roy, N., De Paepe, A.,Speleman, F., 2002. Accurate normalization of real-time quantitative RT-PCRdata by geometric averaging of multiple internal control genes. Genome Biol.3, 1e11.

Vogt, T., 2010. Phenylpropanoid biosynthesis. Mol. Plant 3, 2e20.Walton, N.J., Mayer, M.J., Narbad, A., 2003. Vanillin. Phytochemistry 63, 505e515.Wanner, L.A., Li, G., Ware, D., Somssich, I.E., Davis, K.R., 1995. The phenylalanine

ammonia-lyase gene family in Arabidopsis thaliana. Plant Mol. Biol. 27, 327e338.

Widiez, T., Hartman, T.G., Dudai, N., Yan, Q., Lawton, M., Havkin-Frenkel, D.,Belanger, F.C., 2011. Functional characterization of two new members of thecaffeoyl CoA O-methyltransferase-like gene family from Vanilla planifolia re-veals a new class of plastid-localized O-methyltransferases. Plant Mol. Biol. 76,475e488.

Wildermuth, M.C., 2006. Variations on a theme: synthesis and modification of plantbenzoic acids. Curr. Opin. Plant Biol. 9, 288e296.