Anatomy, histochemistry and biochemistry of glucovanillin ...

Original article

Fruits, vol. 64 (4) 221

Anatomy, histochemistry and biochemistry of glucovanillin, oleoresin andmucilage accumulation sites in green mature vanilla pod (Vanilla planifolia;Orchidaceae): a comprehensive and critical re-examination.Abstract –– Introduction. Mature green vanilla pods accumulate 4-O-(3-methoxy-benzalde-hyde)-β-D-glucoside (glucovanillin), which, upon hydrolysis by an endogenous β-glucosidase,liberates vanillin, the major aroma component of vanilla. Sites of storage of glucovanillin in thepod have been controversially reported for decades; we aim, using precise and widely acceptedtechnical terminology, to clarify this controversy by providing an anatomical, histochemical andbiochemical evidence-based picture of glucovanillin accumulation sites. The pod also synthe-sizes an oleoresin and a mucilage of unknown constitutions; we report here their localizationand structures. Materials and methods. The pod anatomy was examined by light and epi-fluorescence microscopy. A protocol was established allowing fine hand-dissection of diverseanatomical parts of the pod (mesocarp, placentae, trichomes, intralocular interstitial cell-freeregion and seeds). Glucovanillin and γ-pyranones were extracted and analyzed by HPLC, whilethe structures of the mucilaginous polysaccharides were determined after permethylation.Results and discussion. Glucovanillin is essentially stored in the placentae (92%) and margi-nally in trichomes (7%); traces were measured in the mesocarp and intralocular interstitial cell-free medium. Trichomes store massive amounts of a fluorescing oleoresin (44%) rich in alke-nylmethyldihydro-γ-pyranones and synthesize a mucilage made of a glucomannan and a pecticpolysaccharide carrying monomeric arabinose and galactose side-chains. Conclusion. To date,the physiological roles of glucovanillin, long-chain pyranones, and mucilage remain unknown.

France / Vanilla planifolia / Orchidaceae / vanilla / trichomes / vanillin /oleoresins / fluorescence / mucilages / polysaccharides

Anatomie, histochimie, et biochimie des sites d’accumulation de laglucovanilline, d’une oléoresine, et d’un mucilage dans la gousse de vanilleverte mature (Vanilla planifolia ; Orchidaceae) : ré-examen critique d’ensemble.Résumé –– Introduction. Les gousses de vanille vertes matures accumulent du 4-O-(3-méthoxy-benzaldehyde)-β-D-glucoside (glucovanilline), qui, par hydrolyse à l’aide une β-glu-cosidase endogène, libère de la vanilline, le composé d’arôme majoritaire de la vanille. Les sitesde stockage de la glucovanilline dans la gousse ont été mentionnés depuis des décades de façoncontroversée; en utilisant une terminologie technique précise et largement acceptée, nous nousproposons de clarifier cette controverse en présentant un tableau, fondé sur des preuves ana-tomiques, histochimiques et biochimiques, des sites d’accumulation de la glucovanilline. Lagousse synthétise aussi une oléorésine et un mucilage de compositions inconnues ; nous don-nons ici leurs localisations et structures. Matériel et méthodes. L’anatomie de la gousse a étéexaminée en microscopies photonique et d’épifluorescence. Un protocole a été mis au pointqui permet une dissection manuelle fine des différentes parties anatomiques de la gousse (méso-carpe, placentae, trichomes, zone acellulaire interstitielle du locule et graines). La glucovanillineet les γ-pyranones ont été extraites et analysées par HPLC tandis que les structures des poly-saccharides du mucilage ont été déterminées par perméthylation. Resultats et discussion. Laglucovanilline est stockée essentiellement dans les placentae (92 %) et marginalement dans lestrichomes (7 %) ; des traces ont été détectées dans le mésocarpe et le milieu acellulaire interstitieldu locule. Les trichomes accumulent des quantités massives d’une oléorésine fluorescente(44 %) et riche en alkenylmethyldihydro-γ-pyranones et ils synthétisent un mucilage constituéd’un glucomannane et d’un polysaccharide pectique portant des chaînes latérales monoméri-ques d’arabinose et de galactose. Conclusion. Jusqu’à ce jour les rôles physiologiques de laglucovanilline, des pyranones à longues chaînes, et du mucilage demeurent inconnus.

France / Vanilla planifolia / Orchidaceae / vanille / trichome / vanilline /oléorésine / fluorescence / mucilage / polyoside

CIRAD, Persyst,UMR Qualisud, TA B-95 / 16, F-34398 Montpellier Cedex 5, France

[email protected]

Anatomy, histochemistry and biochemistry of glucovanillin, oleoresin and mucilage accumulation sites in green mature vanilla pod (Vanilla planifolia; Orchidaceae): a comprehensive and critical reexaminationEric ODOUX, Jean-Marc BRILLOUET*

* Correspondence and reprints

Received 24 February 2009Accepted 24 April 2009

Fruits, 2009, vol. 64, p. 221–241© 2009 Cirad/EDP SciencesAll rights reservedDOI: 10.1051/fruits/2009017www.fruits-journal.org

RESUMEN ESPAÑOL, p. 241

222 Fruits, vol. 64 (4)

E. Odoux, J.-M. Brillouet

1. Introduction

Amongst orchids, monocotyledonous plantsfrom the Orchidaceae family, a new sub-family, Vanilloideae, to which belongs thecommon vanilla (Vanilla planifolia Jacksonex Andrews, Vanilleae tribe), was recentlyrecognized through molecular taxonomy[1]. Within Vanilloideae (15 genera contain-ing 200 species), the genus Vanilla com-prises 110 species [2, 3] of which only threeare cultivated, namely, V. planifolia,V. tahitensis and V. pompona [4]. The term“vanilla” is used for both the plant and itsfruit.

Vanilla, a monopodial climbing vine,bears epigynous flowers. The syncarpousgynoecium constitutes a stigma, a style andan inferior tricarpellate unilocular ovarybearing placentae, from which rows ofnumerous anatropous ovules differentiateonly after pollination [5]. The fleshy fruits ofvanilla vine, improperly called vanillabeans, develop from the inferior ovary intoelongated pod-like berries with a trigonecross-section [6]; when fruits reach maturity,pods open longitudinally from the apicalend by two dehiscence splits, then progres-sively dry, in the end turning into a capsule[7]. Each carpel shows from the outside a

thin cutinized epicarp with thick-walledcells, a fleshy chlorophyllous mesocarpbearing itself two parietal placental bilobedlaminae (or ridges) running the fruit length,each lamina being subdivided into twolobes symmetrically bent towards the inter-lamina area from one side and one of thethree trigone angles to the other side [8](figures 1–3). Lobes bear myriads of verysmall crustose seeds [6]. In between the pla-cental ridges, differentiated from the inner-most endocarpic layer after fertilization hasoccurred, lie faint yellowish strips of muchelongated hair-like tubular cells, the tri-chomes (or papillae, or hairs) [5, 8–11]; thesecells synthesize and excrete in betweenthemselves and into the locular space a vis-cous mucilage embedding seeds, also calleda matrix [11], the nature of which isunknown, and expel oleoresin dropletswhen perturbed.

Although highly prized from very ancienttimes for its unique aroma, it is only recentlythat vanilla has attracted more attentionfrom plant physiologists, histologists andbiochemists [1, 12–15]. One of the major tar-gets in research on vanilla is obviously thebiosynthesis pathway of vanillin, its majoraroma compound, which implies that inter-est is focused on enzymes themselves [13,16, 17] and sites of synthesis and accumu-lation [10, 18]. Another research domainconcerns the vanilla lipids which may par-ticipate to some extent in its aroma [12].

As stated, most works have been focusedon the anatomical localization and enzy-matic mechanisms of vanillin synthesis andrelease from its non-aromatic precursor, 4-O-(3-methoxy-benzaldehyde)-β-D-glucoside(glucovanillin), all these efforts being madefor a better understanding of aroma genesisthrough diverse technological processesapplied to vanilla pods for production ofmarketable vanilla [5, 7, 9–11, 13, 18–22]; animproved knowledge of these complexmechanisms could ultimately lead to meta-bolic engineering of vanilla to, for instance,increase the levels of desirable metabolitessuch as vanillin [15, 23].

Three recent important breakthroughswere the anatomical localization of glucov-anillin in the inner region of green maturepods (Odoux et al., July 2003, [10]), very

Figure 1.Drawing of a longitudinally sectioned vanilla pod (partial view).

Glucovanillin, oleoresin and mucilage in green mature vanilla pod

Fruits, vol. 64 (4) 223

Figure 2. Split anatomy of vanilla pod (drawings from [9]); 2 placental laminae, photograph from Odoux (2008); micrograph from [10].

224 Fruits, vol. 64 (4)


soon confirmed by Joel et al. in September2003 [18]; the immunofluorescence localiza-tion of 4-hydroxy-benzaldehyde synthase(HBS) in the vanilla pod trichomes [18]; andthe purification and characterization of avanilla β-D-glucosidase responsible for therelease of aromatic vanillin from glucovanil-lin [14].

Arana stated that glucovanillin is mainlylocated (~70%) in the photosynthetic mes-ocarp of the green mature pod, the rest ofit being found in the inner placental portion,and the enzyme responsible for its hydrol-ysis (a β-glucosidase) being exclusivelypresent in the chlorophyllous mesocarp[19];as a consequence, glucovanillin present inthe heart of the pod would have to migratetowards the mesocarpic region to be hydro-

lyzed and release vanillin. Since then, anextremely detrimental confusion has devel-oped and still exists concerning the in vivosites of vanillin synthesis and accumulationof glucovanillin and of its enzymatic hydrol-ysis. Odoux et al. [10] demonstrated that glu-covanillin accumulates mainly in the pla-centae and to a lower extent in thetrichomes, while Joel et al. [18] asserted soonafterwards that vanillin, as measured afterhydrolysis of its glucoconjugate, is synthe-sized in a unique hairy tissue, the trichomes.Former authors showed that the enzymehad exactly the same anatomical localiza-tion as its substrate and that it is cytoplasmicand/or apoplasmic (not vacuolar); libera-tion of vanillin from its precursor would nottherefore be conditioned by a hypotheticaloutward diffusion of glucovanillin, as men-tioned by Arana [19]. Another matter of con-fusion concerns the final fate of vanillin: asa confirmation of hearsay reported inSwamy’s writings who literally and onlywrote “unicellular hairs [i.e., the trichomes]are said to secrete the vanillin.” [5], Joel et al.[18], also quoted by Havkin-Frenkel andBelanger [15], asserted that vanillin issecreted by trichomes as part of a “denselypacked secreted matrix that accumulates inthe fruit cavity”. However, no quantitativeproof of that had ever been provided.

The purpose of our contribution is thus,through reconsideration of our formerlypublished data [10] and of the previous andrecent literature, and by providing new,detailed quantitative data, to unveil andunderline indisputable facts with regards tothe fate of vanillin in the vanilla pod. Thefive main compartments of the pod [(meso-carp + epicarp), placentae, trichomes,seeds, and interstitial intralocular cell-freeregion] were meticulously separated andaccurately analyzed for their basic constitu-ents (sugars, organic acids, proteins, cellwalls and oleoresin) and phenolics includ-ing their glucosylated forms, and their tissuemass distribution is presented. Finally,through a cross-examination of already pub-lished results and anatomical, histochemicaland biochemical data given in the presentwork, we propose an evidence-based pic-ture of the sites of accumulation of glucov-anillin in the green mature pod.

Figure 3. Mode of dissection of trichome strips and placental lobes of vanilla pods (not to scale).


Fruits, vol. 64 (4) 225

Oleoresin, firstly defined as a volatile oilcontaining resin [24], is nowadays widelyaccepted as an “oil containing non-resinousmaterials” obtained from certain plants (e.g.,Capsicum spp., Cisteacae). However, in theperfume industry, in the case of vanilla, themeaning of “oleoresin” was extended to ahydroalcoholic extract of the pods [25], thuscontaining in addition to lipophilic materialsphenolics such as vanillin and its glucocon-jugate, glucovanillin; this “oleoresin” can befurther fractionated, giving a liposolublefraction. Vanilla pods are oleoresin-rich (3%of fresh weight) [25]. Long-chain γ-pyranoneswere isolated from a pentane extract (theliposoluble “oleoresin” fraction) of wholenon-dissected crushed vanilla pods [12] andwere said to arise from epicuticular wax.We now report on the exact tissue localiza-tion of oleoresin long-chain γ-pyranonesin the vanilla pod and their fluorescenceproperties.

Moreover, a thorough, extensive charac-terization of the unknown mucilaginoussubstance secreted into the locule is pre-sented for the first time. With respect to sus-pected functions of pistil mucilaginouspolysaccharides with regards to pollentubes [26], its possible role in the vanilla podis discussed.

2. Materials and methods

2.1. Plant materials

Fifty sound mature green vanilla pods(~8 months after pollination) were usedfrom Kerala State (India). At this maturitystage, i.e., green mature non-dehiscent, theywill be designated as vanilla pods through-out our paper. A sample of ten pods was ran-domly chosen amongst this population.Their average weight was: (22.38 ± 1.83) g(n = 10). It had been formerly checked [27]on five rings (2-mm width) regularly distrib-uted along the central pseudoprismatic zoneof fresh pods (figure 1, 1/3 of the podlength) (n = 5) that, within a single pod, con-centrations of glucovanillin and otherrelated phenolics were constant. Othervanilla pods were from Gutiérrez Zamora(Veracruz, Mexico).

2.2. Light microscopy

Transverse slices (1-mm thickness) weredipped in 0.1 M phosphate buffer (pH 7.0)containing 1% glutaraldehyde, 2% parafor-maldehyde and 1% caffeine. The mediumwas placed under vacuum for 15 min, andsections left to incubate for 24 h at 4 °C.Samples were then dehydrated by succes-sively dipping in 70% ethanol for 30 min,95% ethanol for 2 × 30 min, and ethanol for8 h. They were then impregnated in amedium containing Technovit 7100 resin(100 mL; Kulzer Werheim, Germany) towhich had been added Technovit 7100accelerator (1 g), Technovit polyethyleneglycol PEG 400 (2 mL), and triethylene gly-col dimethacrylate (0.4 mL) diluted twicewith ethanol for 2 h at 4 °C. Samples wereimpregnated for 24 h at 4 °C in the undi-luted impregnation medium before theywere finally embedded at 37 °C in theimpregnation medium (15 mL) with 1 mL ofTechnovit 7100 hardener. Sections (5 µm)were obtained with a Historange LKB micro-tome. They were then oxidized for 5 min in1% periodic acid, washed with distilledwater and stained in the dark for 10 minwith Schiff reactant (PAS) prepared as fol-lows: dispersion of basic fuschin (CI 42500;1 g) in boiling water (200 mL), filtering thesolution once it was cooled to 50 °C, andadding sodium metabisulfite (2 g) and1 M HCl (20 mL); then, after 24 h in thedark, activated charcoal (0.5 g) was added,and the medium filtered. Sections werewashed with distilled water until the wash-ing liquid was colorless and then stainedwith Naphtol Blue Black [(CI 20470); 1 g in7% (v/v)] for 5 min at 60 °C. Sections werequickly washed with distilled water, thentreated with 7% acetic acid, and finally driedfor 15 min at 60 °C. Sections also underwenta Sudan Red 7B histochemical test for totallipids [28]. Standard control procedureswere conducted at the same time. Sectionswere examined under a DMRXA Leicamicroscope (Leica Microsystems, Wetzlar,Germany).

2.3. Fluorescence microscopy

Fresh cross-sections (100 µm) wereobtained from pods using a Microm

226 Fruits, vol. 64 (4)


vibratome and observed with a LeicaDM6000 epifluorescence microscope (filtercube A, excitation BP 340–380 nm, emissionLP 425 nm) (Leica Microsystems, Rueil-Mal-maison, France). Images were processedthrough Volocity 4.0.1 (Improvision, Lexing-ton, MA, USA).

2.4. Dissection of vanilla pods

Estimation of the fresh mass distribution ofthe five compartments [(mesocarp + epi-carp), placentae, trichomes, seeds, and loc-ule seed-free acellular space] forming avanilla pod was conducted as follows: thecentral pseudoprismatic zone of a fresh pod(figure 1) was transversally divided with arazor blade into several cylinders (~2-cmheight); they were then frozen (–20 °C). Acylinder was then longitudinally cut with ascalpel at the angle opposite to the twoangles where the dehiscence splits are(figures 2, 3), and penetrating into the loc-ule not further than the mesocarp width toavoid damaging inner placentae and seeds;it was then unrolled from the two endsshowing the inner faces of the pseudoprismconstituting three paired placental laminaecarrying seeds and three intermediate well-visible seed-free faint yellowish stripslocated at the locule corners and separatedfrom placental laminae by frost-generatedlongitudinal cracks, the trichomes. The sur-face of placental laminae was gentlyscraped with a spatula to remove most of theseeds, remnant ones being eliminated bygently brushing placentae with a smoothpaint brush. Trichome faint yellowish stripswere separated from the placental laminaeunder the stereo microscope with a scalpelrunning along their long sides and followingcracks; strips (width 2 mm) were then cutinto squares (length 2 mm), pinned on theedge, and meticulously separated from theunderlying green mesocarp with a razorblade. The same was done to recover theplacental laminae.

Placental laminae were submitted to asecondary dissection: under the stereomicroscope, laminae were pinned on theedge and meticulously separated with arazor blade into their upper portion (1: thelobes themselves) and their feet (2: the baseof laminae) (figure 3).

The intralocular interstitial cell-freemedium surrounding the seeds was obtainedas follows: portions of fresh pods were lon-gitudinally cut through the V-shaped tri-chome corner and dipped in pentane. Spon-taneously detached seeds were recoveredand, after elimination of remnants of funi-cles with tweezers under the stereo micro-scope, dried at ambient temperature for 4 h;at that stage, the mucilaginous gangueadheres upon drying to the seeds and wasthen analyzed for phenolics. A separatepreparation of seeds was rapidly washedwith distilled water (2 min), and seeds wereanalyzed for phenolics after drying.

New scalpels and razor blades were usedfor each step to avoid cross-contaminationof the different tissues. Histological purity ofthe four tissues was checked under thestereo microscope.

2.5. Determination of glucovanillin and related phenolics

Portions of mesocarp, placentae, trichomes,seeds embedded in their watery and oilygangue, and washed seeds (~50 mg each,dry weight) were extracted by 2 mL of[methanol:water] (50:50, v/v) for 2 h atambient temperature with intermittent son-ication, then centrifuged (14 000 g, 5 min);the supernatant was then submitted toquantitative analysis. HPLC separation ofphenolics and their glycoconjugates wasperformed using an Agilent 1100 separationsystem (Agilent Technologies, Waldbronn,Germany) including a quaternary pumpcoupled to a diode array detector and con-trolled by Chemstation A.10.02 software.Separations were achieved using a [(250 ×4.6)-mm internal diameter] Modulocart QS-Lichrospher 5-μm ODS2 column (Interchim,Montluçon, France) with a guard column,operated at 30 °C. The mobile phase con-sisted of [water:formic acid] (98:2, v/v) (elu-ant A) and [water:acetonitrile:formic acid](18:78:2, v/v/v) (eluant B). The flow ratewas 0.5 mL·min–1. The elution program wasas follows: 8–13% eluant B (0–10 min); 13–20% eluant B (10–30 min); 20–8% eluant B(30–35 min). Triplicate samples wereinjected at a level of 10 μL. The columneffluent was monitored at 280 nm. The


Fruits, vol. 64 (4) 227

column eluate was then split and0.25 mL·min–1 was directed to a LCQ iontrap spectrometer fitted with an electrosprayinterface (Thermo Finnigan, San Jose, USA).Experiments were conducted in both nega-tive and positive modes. Scan range was100–2000 atomic mass unit (a.m.u.) andscan rate 1 scan·sec–1. The desolvation tem-peratures were (250 and 300) °C in the pos-itive and negative ion modes, respectively.High spray voltage was set at 4000 V (pos-itive) and 3500 V (negative) ion modes.Nitrogen was used as the dry gas at a flowof 5 for the auxiliary gas and 55 for thesheath gas. Identifications were achieved onthe basis of the ion molecular masses, UV-visible spectra and injection (+co-injection)of standards when available.

Quantification was achieved by injectionof solutions of known concentrations of glu-covanillin, ρ-hydroxybenzaldehyde, vanillicacid and vanillin; glucosides were quantita-tively hydrolyzed by sweet almond β-glu-cosidase and released aglycons were con-verted into their glucosides [29]. Data wereobtained from ten separate dissections andexpressed as mg·100 g–1 fresh weight.These aglycons and their glucosides will bethereafter designated as “phenolics”; theglucosides themselves will be named“aroma precursors”.

2.6. Extraction of the crude mucilage

Fresh trichomes (~1 g) were disintegrated inpentane and, after filter paper filtration,dipped in distilled water (50 mL); the slurrywas left under magnetic stirring for 2 h. Aftercentrifugation (14 000 g, 5 min), the crudemucilage was precipitated from the waterphase by adding ethanol to 80% final con-centration. After standing overnight at ambi-ent temperature, the fresh mucilage wasrecovered by centrifugation, freeze-driedand weighed.

2.7. Determination of water and oleoresin

Viscous material contaminating bits of pla-cental laminae and trichome strips, andseeds, was very quickly blotted with a filter

paper (5 s) without exerting pressure andthe four fresh compartments (including themesocarp) from a single pod were weighed,dipped in pentane and left under magneticstirring overnight; the mesocarp, placentaeand trichomes were primarily torn to pieceswith tweezers while seeds were crushed ina mortar. They were then withdrawn fromthe solvent, which rapidly evaporated(30 s); then the fresh materials wereweighed, left overnight in a vacuum ovenat 80 °C, and weighed again. They werethen pulverized in liquid nitrogen into a finepowder with a Dangoumeau ball mill (topimpact frequency). Data were obtainedfrom ten separate dissections.

A bulk amount of oleoresin was preparedas follows: vanilla pods (~60 g) were frozenand the outer part was eliminated with ahand vegetable peeler (no green color left);trichomes were then hand-dissected andseeds were eliminated by flotation in coldwater (4 °C). The seed-free trichomes weredisintegrated in hexane ([solvent:wet mat-ter] ratio 10) with an ultra-Turrax. After fil-tration, the organic phase was recoveredand excess distilled water added, and themedium was sonicated for 1 h; after centrif-ugation (14 000 g, 5 min), the organic phasewas again water-washed three times; a cleardeep yellow oleoresin was finally obtainedby vacuum evaporation of the solvent (yield0.6 g).

2.8. Determination of sugars and organic acids

The dry defatted mesocarp, placentae andtrichomes (~5 mg each, dry solids) wereextracted by 500 µL of acidified distilledwater (pH 2.3) for 2 h at ambient tempera-ture then centrifuged (14 000 g, 5 min). HPLCanalyses were performed using a Spectra-SERIES separation system P100 (ThermoSeparation Products, USA) including a qua-ternary pump coupled to a diode array detec-tor and controlled by Chemstation A.10.02software. Separations were achieved usinga [(250 × 4) mm i.d.) Lichrospher 100 RP18e(5 µm) column (Interchim, Montluçon,France) with a guard column, operated at40 °C. The mobile phase consisted of[methanol:0.01 M phosphoric acid] (30:70,

228 Fruits, vol. 64 (4)


v/v). The flow rate was 0.7 mL·min–1. Dupli-cate samples were injected at a level of20 µL. The column effluent was monitoredat 210 nm with a Shimadzu differentialrefractometer. Quantification was achievedby injection of solutions of known concen-trations of sucrose, glucose, fructose, andoxalic and malic acids; data were expressedas mg·100 g–1 fresh weight.

2.9. General

Determination of nitrogen was performedas follows: after pyrolysis under oxygen at1050 °C, nitrogen oxides were reduced intoN2 which was measured with a catharome-ter. Cell walls were measured according toHuber [30]. Polysaccharides constitutingneutral sugars were determined, afterhydrolysis with 2 M trifluoroacetic acid for1 h 30 min at 120 °C [31], by GLC of theiralditol acetate derivatives [32] at 210 °C ona fused-silica DB-225 capillary column[(30 × 0.32) mm i.d.], 0.25-µm film; J&W Sci-entific] with H2 as the carrier gas; inositolwas the internal standard. Uronic acids weremeasured by the m-phenylphenol method[33] using galacturonic acid as a standard.Uronic acids were also identified from theacid hydrolyzate by separation through highpH anion-exchange chromatography(HPAEC) as described by Brillouet et al. [34].Methylation structural analysis of thepolysaccharides was performed withsodium methyl sulfinyl carbanion andmethyl iodide in dimethyl sulfoxide [35];partially methylated alditol acetates wereanalyzed as described by Pellerin et al. [36].

The fluorescence spectrum of oleoresinwas measured with a F-4500 Hitachi spec-tro-fluorimeter. Oleoresin constituents wereseparated by HPLC using the same systemand column used for analysis of phenolics.The mobile phase consisted of acetonitrile(eluant A) and dichloromethane (eluant B).The flow rate was 0.8 mL·min–1. The elutionprogram was as follows: eluant A 100% (0–4 min); 100–45% (4–5 min); 45–20% (5–20 min); 20–100% (20–25 min). Oleoresinwas diluted in [hexane:dichloromethane](2:1, v/v), then duplicate samples wereinjected at a level of 10 µL. The columneffluent was monitored with a 1100 diode

array detector at 260 nm (Agilent Technol-ogies) coupled to a RF-10AXL Shimadzu flu-orescence detector.

3. Results

3.1. Anatomy of green mature vanilla pods

Observing anatomical and histological ele-ments from the outside towards the centrallocule of the fruit (figures 1, 2) makes it pos-sible to find: a thin epicarp (~20 µm thick-ness) covered with a cuticle; a fleshy pho-tosynthetic greenish mesocarp (~3-mmthickness) made of parenchymatous cellsembedding several vascular bundles; threeyellowish placentae bearing six longitudinalbi-lobed placental paired laminae ontowhich seeds are attached through funicles;three vitreous hairy areas (trichomes)located at the three corners of the pod; anda two cell-layered endocarp. French notedthat the trichomes, although adjacent to theplacentae, are not attached to them [11], i.e.,they have no physical contact with placentalcells, being simply bound through their dis-tal ends to endocarpic cells. Upon section,the vanilla pod exhibits two easily visibledehiscence lines.

The whole set of data given in this paperwill refer to the five main compartmentsconstituting a vanilla pod: (1) the mesocarp(+ epicarp), (2) placentae, (3) trichomes,(4) seeds and the intralocular interstitialcell-free region defined as the watery andoily space surrounding seeds in thelocule (5). Due to minute expected masses,no efforts were made to dissect the epicarpseparately; similarly, funicles were takenwith the placentae. Furthermore, only thepseudoprismatic zone (figure 1) will beconsidered since distributions might berather different at the apical and distal endsdue to higher weight proportions of vascularbundles.

It is also very important to recall thealready mentioned fact that, after freezing apod, a well demarcated faint yellowish trig-one becomes well visible (figure 2 and [10]);it corresponds to the placentae and tri-chomes (see below).


Fruits, vol. 64 (4) 229

Since Joel et al. [18] asserted that phenolicsubstances are secreted into the apoplasm(i.e., into the cell-free space surroundingseeds), specific attention was paid to theintralocular interstitial cell-free volume. Itwas measured or calculated by three meansand the result checked as follows: relativesurfaces of the mesocarp (S1), placentae(S2), trichomes (S3), and locule (S4) wereobtained by cutting them out from a hardpaper photograph of a pod transverse sec-tion (figure 4); after weighing, throughpaper density, the percent surfaces were cal-culated; given that the total surface of thetransverse section was 51.75 mm2, the totallocular surface (S4) was 10.9% of the whole,i.e., 5.67 mm2. Although seeds are bulbousand pseudospherical [6, 9], for a simpler cal-culation, we assimilated a seed to a sphereof 230 µm diameter; we covered the avail-able locule surface with a maximum numberof seed disks (surface of a seed sectionS5 = 4.15 × 104 µm2; volume of a seed =6.37 × 10–3 mm3). Going to a ring of5.67 mm2 and 230 µm thickness (= a seeddiameter), it came to a maximum of117 seeds packed in an intralocular seedmonolayer. Knowing that the volume of thisring is 1.30 mm3 and that the total volumeof seeds is 0.75 mm3, then the empty vol-ume was 0.55 mm3, i.e., ~ 43% of the loculartotal volume. Finally, we filled a graduatedtube with glass beads (diameter = 0.8 cm)up to 60 mL (bead diameter and total vol-ume proportional to actual seed diameterand total locular volume) and added waterup to the surface of the beads: interstitialvolume was found to be equal to 26 mL(~ 43% of total volume) and the number ofbeads to 115. By counting seeds on a freshsection under the stereo microscope, it cameto ~120. Thus, on average, 117 seeds arecontained in a tranverse section of 230 μmthickness.

The mesocarp (S1), placentae (S2) andtrichome (S3) surfaces were (33.5, 10.5 and2.1) mm2, respectively. Finally, the percentdistribution of volumes in the pseudopris-matic portion of a green mature vanilla podwas: [mesocarp:placenta:trichomes:seeds:intralocular interstitial cell-free medium] =[64.7:20.3:4.1:6.2:4.7]. The inter-trichomevolume was neglected in this estimationsince French stated that these hairs become

cemented upon aging [11]. Finally, the den-sities of the respective compartments were:[mesocarp:placentae:trichomes:seeds] = [1.00:1.07:0.88:0.85]; they may be comparedwith their respective oleoresin and watercontents (see below).

3.2. Distributions of fresh masses, dry solids and organic materials in mature green vanilla pods

The five compartments from ten greenmature frozen vanilla pods were obtainedby hand-dissection, and more specificallyfor trichomes and placental lobes (figure 3).Due to their minuteness, hand-dissection oftrichomes required a specific techniquegreatly facilitated by frost-generated crackswhich formed through the mesocarp at theboundary between placentae and hairs.Fresh materials were weighed separately foreach pod. Most of the fresh pod matter ismade of the fleshy mesocarp followed bythe placentae; trichomes, seeds and theintralocular interstitial cell-free mediumeach represented ~4–5% of the pods(table I); until now, the mass proportion oftrichomes had not been determined.

Figure 4. Mode of estimation of the intralocular interstitial cell-free volume of vanilla pods (seed diameter: 230 µm).

230 Fruits, vol. 64 (4)


Dry solids were defined as the residue leftafter extraction by pentane and vacuumheat-dehydration. The materials soluble inthis solvent (oleoresin) were liquid and oilyat ambient temperature and were consid-ered separately. Dry solids were unequallydistributed between compartments from thevery rich seeds to the very poor trichomes(table I); the dry solid contents of the mes-ocarp and placentae differed by ~6% infavor of the latter, which could be relatedto placental glucovanillin content. The tri-chome water content was ~50% only (freshweight basis), which explains their very lowdry solids content.

Proteins (N × 5.7), as percent of freshmatter, were more represented in the pla-centae than in the mesocarp and trichomes(table II). If one expresses protein contentof the water phase only of each tissue(excluding oleoresin), then it comes to: mes-ocarp 0.55%, placentae 1.35% and trichomes0.58%; again, placentae appear ~2.5 times

richer than the two other compartments;indeed, placental laminae were morestrongly bluish-stained than the mesocarpby the Naphtol Blue Black (protein mate-rial), indicating a high metabolic activity[10]. Knowing that an unknown part of thecellular water is contained in the apoplasm,the symplasmic protein concentrationswould be even higher.

Soluble sugar contents in the mesocarpand placentae were similar at ~3% (table II);in the case of trichomes, it was ~1.5% of totalfresh matter, but also ~3% of the hydratedmatter. Sucrose, glucose and fructose werethe constituting sugars.

3.3. Distribution of glucovanillin and related phenolics within mature green vanilla pods

On a fresh matter basis, placentae werefound to be richer in glucovanillin (2.2 times)

Table I.Distribution of fresh matter and dry solids contents from vanilla pods (% of fresh matter, n = 10).

Compartments of vanilla pods

Fresh matter distribution Water Oleoresin1 Dry solids2

Mesocarp 64.65 ± 2.35 87.66 ± 2.05 0.82 ± 0.09 11.52 ± 0.27

Placentae 21.79 ± 1.10 80.75 ± 3.69 1.49 ± 0.25 17.76 ± 0.82

Trichomes 3.60 ± 0.22 49.65 ± 5.12 43.71 ± 6.02 6.64 ± 0.80

Seeds 5.25 ± 0.15 13.33 ± 1.15 32.31 ± 2.15 54.36 ± 2.35

Intralocular cell-free medium 4.71 (calculated) Not determined Not determined Not determined

1 Oleoresin = materials soluble in pentane.2 Dry solids = [(fresh matter) – (water + oil)].

Table II.Organic composition of vanilla pods (% of fresh matter, n = 10).

Compartmentsof vanilla pods

Proteins(N × 5.7)

Sugars1 Organic acids2 Cell walls

Mesocarp 0.48 ± 0.05 2.94 ± 0.15 0.53 ± 0.04 5.08

Placentae 1.09 ± 0.05 3.02 ± 0.22 1.12 ± 0.03 3.24

Trichomes 0.29 ± 0.09 1.51 ± 0.09 0.03 ± 0.01 Not determined

Seeds 4.27 ± 0.11 – – –

1 Sugars = glucose + fructose + sucrose.2 Organic acids = oxalic + malic acids.


Fruits, vol. 64 (4) 231

than trichomes (table III), a ratio slightlylower than our previously reported one (3.3times) [10] for green mature pods of the samephysiological age. It is worth noting thatconcentrations of glucovanillin in the pla-centae and trichomes, when expressed as apercent of their respective water phases, areclose to each other, as are the levels of sugars(table II). Contrary to our previous study [10],we found, although at trace level (table III),some glucovanillin in the mesocarp, as didJoel et al. [18] (0.56%, fresh matter basis), andseeds were glucovanillin-free. The glucov-anillin mass distribution within a pod was:placentae 92.2%, trichomes 7.0% and mes-ocarp 0.8%. On a dry solids basis, it repre-sents ~48% of the placentae and ~59% of thetrichomes.

In addition to that, the placental laminaewere found to be heterogenous for theircontents of glucovanillin, p-hydroxybenzal-dehyde β-D-glucoside and their aglycons:indeed, a gradient of concentrations wasfound from their base to the lobes of thelaminae (figure 3, table III), lobes beingtwice as poor as the bases of the laminae.

A different distribution was found forp-hydroxybenzaldehyde β-D-glucoside. Al-though placentae were the richest compart-ments (fresh weight basis), the trichomesand mesocarp showed significant levels andits mass distribution was somewhat differentfrom that of glucovanillin: placentae 81.5%,trichomes 17.2% and mesocarp 1.4%. Theaverage (glucovanillin:p-hydroxybenzalde-hyde β-D-glucoside) concentration relative

ratios were also different: ~[85:1] in placen-tae, and ~[31:1] in trichomes. We also carriedout a less complete study on green maturepods from Mexico; they were richer inglucovanillin and p-hydroxybenzaldehydeβ-D-glucoside was relatively more abundantcompared with glucovanillin (respectiveconcentrations: in placentae 11.0% and0.5%; in trichomes 1.3% and 0.2%); wefound that the above ratio was again differ-ent between placentae and trichomes:~[22:1], and ~[6:1], respectively. Thus,differently from trichomes, the accumula-tion of glucovanillin is more largely favored(three to four times) in placentae thanp-hydroxybenzaldehyde β-D-glucoside; al-though the meaning of this differenceremains obscure, this suggests a tissue met-abolic segregation with regards to the bio-synthesis (and/or its regulation) of these twoaroma precursors. It must be noted thattraces of 4-O-(3-methoxy-benzoic acid)-β-D-glucoside (vanillic acid glucoside) and4-O-(benzoic acid)-β-D-glucoside were alsofound in both compartments.

The concentrations of glucovanillin andrelated phenolics in the intralocular intersti-tial cell-free medium embedding seeds werealso measured. They were: glucovanillin27 µg·mL–1 and vanillin 5 µg·mL–1, for atotal level of ~0.003% in interstitial liquid,i.e., ~0.006% of the total phenolics con-tained in the vanilla pod.

Finally, aglycons were also present.Whether they were not yet glucosylated orhad been deglucosylated from their parent

Table III.Distribution of phenolic glucosides and their aglycons in green mature vanilla pods (% of fresh weight, n = 10).

Compartmentsof vanilla pods

4-O-(3-methoxy-benzaldehyde)-β-D-glucoside

4-O-(benzaldehyde)-β-D-glucoside

3-methoxy-4-hydroxybenzaldehyde

4-hydroxy-benzaldehyde

λmax (nm) 272.0, 307.6 268.2 279.6, 309.2 284.2

(M+HCOO)– adduct with formiate ion

359 329 – –

Mesocarp 0.026 ± 0.019 0.0006 ± 0.0004 0.0020 ± 0.0010 0.0006 ± 0.0005

Placentae 8.589 ± 1.628 0.0995 ± 0.0286 0.2010 ± 0.0760 0.0121 ± 0.0051

Bases 9.634 ± 1.991 0.1221 ± 0.0410 0.2420 ± 0.0610 0.0155 ± 0.0063

Lobes 4.982 ± 1.273 0.0555 ± 0.0212 0.1950 ± 0.0450 0.0079 ± 0.0028

Trichomes 3.932 ± 1.277 0.1290 ± 0.0530 0.1470 ± 0.0430 0.0140 ± 0.0050

232 Fruits, vol. 64 (4)


glucoconjugates by vanilla β-glucosidase isnot known [14]. Their levels were, as a per-cent of total aglycon, i.e., (free + glucosidi-cally-linked), placentae; vanillin 4.5%,p-hydroxybenzaldehyde 21.8%; trichomes:vanillin 7.1%, p-hydroxybenzaldehyde 20.3%.If the free aglycons arose from the enzy-matic deglucosylation of their precursors,one must note a different behavior betweenthese two phenolics which might reflect adifferent maximum velocity (Vm) of theenzyme on these two substrates which wasnot known until now.

3.4. Distribution of oleoresin within mature green vanilla pods

We determined the contents of pentane-sol-uble materials (oleoresin) in the four com-partments [8, 9] (table I). While the meso-carp and placentae showed low levels(<1.5%, fresh weight basis), the trichomeswere found to be extremely rich, with ~45%of an oily material (fresh weight basis). Sucha result was not unexpected since lightmicrographs of unstained thick fresh sec-tions (figure 5) showed a high density ofoily brown-yellow material in their distaland apical portions and droplets of the samecolor; the trichomes were also intensely red-dish-stained by Sudan Red (figure 5), andexhibited a very intense white-bluish fluo-rescence (figure 5). Placentae showed aweak yellowish fluorescence only. ByHPLC, we checked that trichome oleoresinwas devoid of glucovanillin and related phe-nolics, and of fluorescent chlorophyll. Itsabsorbance and emission spectra showedmaxima at λmax = 260 nm, consistent withan enone group [12], and λem = 458 nm inhexane (λexc = 325 nm), respectively, cor-responding to its fluorescence in the bluerange of the visible spectrum. Separation ofoleoresin constituents by HPLC afforded ahomologous series (compounds 1 to 6; fig-ure 6) showing a very strong UV absorbanceand fluorescence. Peak 3 gave a pseudomo-lecular ion at (M+H)+ at m/z = 433. We didnot detect these compounds in the sepa-rately dissected pod epicarp. We also havequalitative evidence (HPLC analysis) thatthe deep yellow color of the oleoresin is dueto the presence of β-carotene.

3.5. Distribution and characterization of the mucilage within mature green vanilla pods

The mesocarp was directly homogenized indistilled water while viscous materialimpregnating trichome strips and, to a lowerextent, placentae was rapidly drained witha filter paper, then was treated like the mes-ocarp. A mucilaginous material was recov-ered by precipitation of the extract in 80%ethanol. The mesocarp and placentae werefound to be mucilage-free, while the tri-chome dry solids consisted of ~32% of thismaterial which was shown to be essentiallypolysaccharidic in nature. The feeling ofhigh viscosity obtained when manipulatingthe seeds with one's fingers is due to themucilage itself since it was the same with theoleoresin-free one redissolved in water. Thishuge concentration is reflected by theintense fuchsia periodic acid-Schiff (PAS)staining of the intracellular portion of thetubular cells (figure 5), whereas, in other tis-sues, only cell wall polysaccharides (cellu-lose, hemicelluloses and pectic substances)stained fuchsia. The intralocular interstitialcell-free medium was not stained since theexcreted water-soluble mucilage was elim-inated into the glutaraldehyde fixative solu-tion. Bluish-stained protein material wasalso visible in the cell-free space. The com-position of this polysaccharide mixtureshows that uronic acids, arabinose andgalactose are the major sugar constituents,followed by mannose and glucose(table IV). Xylose, rhamnose and fucose areminor constituents. After saponification [38],no ester-linked ferulic acid was observed.Structural analysis of this polysaccharidemixture (table V) reveals a high density ofterminal units (arabinose and galactose) anda glucomannan.

4. Discussion

To our knowledge, the first mention of anexclusive localization of an unknown vanil-lin-releasing system in the inner portion ofthe vanilla pod was given by De Lannessan,who wrote: “The flavouring matter is notcontained in the outer fleshy part of the fruit,


Fruits, vol. 64 (4) 233

since, when cutting it into thin lamellaewhen fresh, and letting them separately dry,one can observe that only lamellae from theinner region are odorant.” [8]. The sameyear, Mangin wrote: “These fruits are muchelongated siliques or capsules with fleshyand thick walls containing an aromaticpulp… That pulp itself is the vanilla; the restof the fruit is only its envelope.” [39]. Thus,

anatomically speaking, the vanilla pod, bythe end of the nineteenth century (and verylikely much earlier), had been experimen-tally demonstrated as heterogeneous withregards to aroma genesis, i.e., the enzymatichydrolysis of glucovanillin into vanillin.

We produced recent quantitative evi-dence of a preferential sequestration of glu-covanillin and β-glucosidase, the two

Figure 5.Vanilla pod trichome sections (J-shaped corner, see figure 2).

234 Fruits, vol. 64 (4)


components of the vanillin-releasing sys-tem, in the inner pod region (i.e., placentaeand trichomes) [10], in agreement with theabove quoted authors but in contradictionwith previous data by Arana [19]. In thepresent work, we unambiguously and moredefinitely proved, through fine dissections,that the glucosylated non-aromatic precur-sor of vanillin, i.e., glucovanillin, is essen-tially stored in the placental laminae of thegreen mature pods (~92% of total glucovan-illin), the trichomes containing only about7% of it. Regarding the intralocular intersti-tial cell-free medium, we showed that itsglucovanillin content (if any) was infinites-imal. Whether the fleshy photosyntheticmesocarp at the green mature stage couldbe considered as glucovanillin-free [10] or ifit contains a minute proportion of this mol-ecule (this work and [18]) is still a matter ofconjecture: indeed, although its hand-dis-section was performed under the stereomicroscope along the sharp boundary withthe whitish placental zone, a slight contam-ination cannot be excluded, and, for exam-ple, a 1% placental contamination wouldincrease by one-third the mesocarpic gluco-vanillin content. A recent laser capturemicrodissection technique coupled withmicroanalysis [40] could be an efficientapproach to tackle this point.

In the aqueous phase of the placentae,glucovanillin showed an estimated concen-tration of 338 mM, a value similar to the onepreviously reported (300 mM [10]), butlikely to be higher since an unknown partof water is apoplasmic (hydration of cellwalls). In the trichome water phase, itreached 252 mM. It is worth noting that con-centrations of glucovanillin, when expressedas a percent of the water phases of the pla-centae and the trichomes, respectively, arevery similar and so are the levels of sugars.Thus, although these two cell types areontogenetically, morphologically and func-tionally (mucilage and oleoresin synthesesin trichomes only) very different, their aque-ous phases are very much comparable fortheir phenolic and sugar contents. Since themaximum solubility of glucovanillin in purewater is lower than 100 mM at ambient tem-perature (Odoux, unpubl. res.), the actualphysico-chemical status of glucovanillin inplacentae is not known: the vacuolarmedium (> 93% of total cell volume [10]) inwhich glucovanillin is present might beoversaturated due to the presence of otherelements (sugars, proteins, etc.). Glucovan-illin might also be under crystalline or amor-phous insoluble forms although the placen-tal cells appear translucent under the stereomicroscope (figure 1). Appearance of awhitish zone, i.e., the placentae, upon freez-ing (figure 2) leads to favoring the firsthypothesis.

The presence of glucovanillin in tri-chomes was formerly illustrated by theirintense epifluorescence under unfiltered UVexcitation [11, 18]. While we observed a blu-ish-white intense fluorescence of the tri-chomes (figure 5) and of expelled droplets,we found that placental laminae, althoughtwice as rich in glucovanillin as trichomes,exhibited a faint greenish-yellow fluores-cence. More precisely, it is mainly the lam-inae lobes (i.e., the region where the

Figure 6. UV- and fluorescence-HPLC chromatograms of trichome oleoresin (λ = 260 nm) from vanilla pods.

Table IV.Composition of the trichome mucilage (anhydrosugar mole %) from vanilla pods.

Yield(% fresh trichomes)

Rhamnose Fucose Arabinose Xylose Mannose Galactose Glucose Glucuronic acid

Galacturonic acid

2.1 0.83 0.29 24.49 2.86 11.69 18.87 8.66 25.01 7.30


Fruits, vol. 64 (4) 235

funicles carrying seeds are; figures 1–3)which show significant fluorescence,although far less intense than the trichomes’;one must recall that the placental lobes con-tain twice less glucovanillin than the bases.These observations are hardly reconcilablewith glucovanillin being the unique sourceof fluorescence. Indeed, trichome oleoresinitself exhibited a strong white-bluish fluo-rescence due to alkenylmethyl-dihydro-γ-pyranones (figure 6). Ramaroson-Raoni-zafinimanana et al., when analyzing theunsaponifiable portion of a vanilla integraloleoresin by gas-liquid chromatography(GLC) and NMR spectroscopy, characterizedthe first three homologs of this (1 to 3) series,namely, 2-nonadecen-10-yl, 2-heinecosen-12-yl and 2-tricosen-14-yl 2,3-dihydro 6-methyl 4H-pyran-4-ones [12] (figure 7). Dueto their high molecular weights, the upperhomologs (4 to 6) were not detected byGLC. These are: 2-pentacosen-16-yl, 2-hep-tacosen-18-yl and 2-nonacosen-20-yl 2,3-dihydro 6-methyl 4H-pyran-4-ones. Althoughwe did not measure them, the relative pro-portions of the 1→3 pyranone homologsreported by Maestro et al. [25] were strictlythe same as ours (1:2:3 = 0.10:0.41:1).

In conclusion, contrary to the aboveauthors who ascribed the UV fluorescenceof the hairy tissue (i.e., the trichomes) tophenolic substances that it secretes into theapoplasm [18], our results demonstrate thattheir very intense fluorescence is essentiallydue to alkenylmethyldihydro-pyranonescontained in their oleoresin. This is the firstmention of their presence in trichomes.Indeed, because they did not dissect theirvanilla pods to obtain the epicarp, andextracted neutral lipids from whole crushedbeans, Ramaroson-Raonizafinimanana et al.located long-chain γ-pyranones in epicutic-ular wax [12], which is devoid of these com-pounds. In fact, only trichomes seem to con-tain these pyranones. Although we did notmeasure their content, they represent,according to Maestro et al., and dependingon vanilla pod batches, from 3% to 25% ofvanilla oleoresin components detected byGLC, i.e., given the huge trichome oleoresincontent (~45% fresh weight), from ~1.4% to11% of the trichome fresh weight [25]. Glu-covanillin and vanillin weakly fluoresce tooat λem = 410 nm ([41]; G Conejero, “pers.

commun.”), i.e., in the purple range of thevisible spectrum, but, although representing~4% of trichome fresh weight and since

Table V.Structural analysis of the mucilaginous polysaccharides of vanillapods.

Sugar derivative Linkage indicated Mole %

2,3,5-Me3-arabinose1 terminal furanose 30.8

2,3,4-Me3-arabinose terminal pyranose 5.3

3,5-Me2-arabinose 2-linked 0.5



2,3,4,6-Me4-galactose terminal 19.5

2,3,4-Me3-galactose 6-linked 2.1

2,4,6-Me3-galactose 3-linked 2.2

2,4-Me2-galactose 3,6-linked 2.6

2-Me-galactose 3,4,6-linked 1.0

2,3,4,6-Me4-glucose terminal 1.1

2,3,6-Me3-glucose 4-linked 12.9

2,3-Me2-glucose 4,6-linked 1.2

2,3,4,6-Me4-mannose terminal 1.6

3,4,6-Me3-mannose 2-linked 0.1

2,3,6-Me3-mannose 4-linked 15.1

2,3-Me2-mannose 4,6-linked 1.7

1 1,4-di-O-acetyl-2,3,5-tri-O-methyl-arabinitol.

Figure 7.Structures of alkenylmethyldihydro-pyranones.

236 Fruits, vol. 64 (4)


placentae fluoresce faintly in the green-yel-low range, these phenolics cannot accountfor the observed trichome brilliant white-bluish fluorescence.

As stated above, within months followingour first article, Joel et al. confirmed afavored localization of vanillin (in fact glu-covanillin) in the central part of the pod [18]but, this time, attention was focused on thetrichome tissue. The authors’ assertion thata unique hairy secretory tissue, the trichomes,produces and accumulates (gluco)vanillin isnot easily reconcilable with our data. Theauthors drew a parallel between the elon-gation rate of tubular cells (trichomes)during maturation and the increase in con-centrations of glucosylated forms ofp-hydroxybenzaldehyde and vanillin in thewhitish inner part of the pod (i.e., placentaeand trichomes) during the same period.Such pseudoparallel evolution was taken asevidence for the trichomes being a uniquetissue synthesizing these phenolics. Later,Havkin-Frenkel and Belanger wrote, quot-ing the considered article, “the accumula-tion of these compounds was directlyrelated to the size of inner specialized cells”[15] (i.e., the trichomes). Actually, the pla-centae and trichomes were undeniably ana-lyzed in admixture, maybe because of thetediousness of separating them properly bydissection; indeed, the reported concentra-tions were those found in the placentae, notin the trichomes themselves (table III).Although a superposition of histograms canbe done for p-hydroxybenzaldehyde β-D-glucoside accumulation and trichomelength, this coincidental parallelism doesnot exist for glucovanillin, which almostreaches its maximum at 5–6 months (tri-chome length 10 µm) compared with 8–9 months (trichome length 300 µm) afterpollination (12.5% vs.14.0%, fresh weightbasis; note that the ordinate axis was scaledby mistake as % dry weight). Considering allour preceding comments, this ontogeneticalcharacter is definitely not valid evidence forthe trichomes being the unique site of syn-thesis and accumulation of glucovanillinand p-hydroxybenzaldehyde β-D-glucoside.

In the same article, Joel et al. [8] produceda confocal micrograph showing a positiveimmunostaining of trichome cytoplasm

with an antibody raised against vanillap-hydroxybenzaldehyde synthase (HBS)[13]; this was also questionably consideredas a second piece of evidence of vanillinsynthesis in the trichomes. Indeed, theenzyme produces, from p-coumaric acid,p-hydroxybenzaldehyde, not vanillin. Fur-thermore, since no experimental detailswere given for the preparation and charac-terization of this antibody, one may thinkthat, if it was raised against the enzymedescribed in the above reference, severalkinds of proteins (enzymes) could havebeen marked, mostly when the two enzymefractions obtained there were not pure,showing several bands in SDS-polyacryla-mide gel electrophoresis. Havkin-Frenkeland Belanger [15], and Havkin-Frenkel andPodstolski for vanilla cell cultures [37], andSicar and Mitra for carrot hairy roots [42],also mentioned the possibility that the HBSactivities which, in both cases, were notpurified to homogeneity, could be mediatedby several enzymes. In fact, an in situ directdisplay of the enzymatic activity would havebeen a more definitive proof.

Joel et al. [18], and Havkin-Frenkel andBelanger [15] displayed a micrograph of thecentral portion of the pod showing anintense catechin-HCl staining of a character-istic placental lamina improperly arrowedas the “secreted matrix” (compare theirshape in figure 2 and in the above refer-ences); this can be viewed as positive indi-rect evidence of their richness in phenolicsas stated in this paper. The authors men-tioned a descending staining gradient fromthe endocarp in the fruit cavity outwards incontradiction with our quantitative data,since we show here that there exists anincreasing gradient from the apical ends ofthe placental lobes towards the base of pla-cental laminae for their contents of glucov-anillin, p-hydroxybenzaldehyde β-D-gluco-side and their aglycons. We have nosuggestion to explain the existence of sucha gradient. Secondly, on the same micro-graph, the “vanillin secretory tissue”, i.e., thetrichomes, was not stained red. The authorsinterpreted this absence of staining of thespecial cells dedicated to vanillin biosynthe-sis as indicating that these cells would syn-thesize flavor compounds and secrete theseproducts into the vicinal tissue region


Fruits, vol. 64 (4) 237

and/or into the extra-cellular regions. How-ever, as mentioned by French, trichomes,although being adjacent to placental lami-nae, are not attached to them [11] (figures 1,5), but to the underlying endocarpic cellsfrom which they differentiated earlier intheir development. Thus, if vanillin weresynthesized in trichomes only, how and whywould it accumulate in such massiveamounts in placental laminae? Traffickingfrom the trichome symplasm to the laminaewould require crossing through several vac-uolar and cytoplasmic membranes, cellwalls, and the liquid locular gap in between,which seems neither physiologically or his-tochemically comprehensible, nor energet-ically feasible. More likely, the non-stainingmay have a more trivial cause: a perturba-tion of the catechin-HCl reaction caused bythe presence of ~45% oleoresin (freshweight basis; table I) in the trichomes hadoccurred.

Another matter of controversy is thenature of the secretions produced by the tri-chomes. Joel et al. [18] altered Swamy’s writ-ings [5], since this author never wrote thatvanillin is “secreted into tissues around theseeds”; starting from this questionableextrapolation, and referring to their histo-logical observations, they asserted thatvanillin (in fact glucovanillin) was secretedby trichomes “into the fruit cavity”; how-ever, no comments were given with regardsto the possible physiological function of thisso-called “vanillin secretion”. Concerningvanillin and its glucoside, we demonstratedhere that the quantitatively unsupportedsuggestion of Joel et al. [18] and the follow-ing references about their “secretion” by thetrichomes into the intralocular interstitialcell-free medium is almost unprovable inpractice; indeed, unpredictable (unavoida-ble?) minute contaminations by the V-shapedtrichome corner may still occur when lon-gitudinally cutting the pseudoprismatic por-tion, rendering it impossible to ascertainwhether the intralocular interstitial cell-freemedium is phenolic-free or not. If not, theamount of “secreted” phenolics appearsinfinitesimal compared with the placentaland, accessorily, trichome stocks.

However, the secretory role of trichomesin general has long been well documented

in the plant kingdom [43–46]. French, whenstudying ovular and placental trichomes ofAraceae, mentioned their mucilage-secret-ing activity [45], as did Fahn and Simony forFragonia species [46]. In the case of vanillapods, they are known to excrete a mucilageor sticky substance [11] and expel oleoresindroplets [8, 47]. We demonstrate here for thefirst time that trichomes actually producemassive intracellular amounts of mucilage,polysaccharidic in nature: it is made of amixture of a typical glucomannan (~35%)resembling the Amorphophallus konjac(monocotyledon) glucomannan [48] and ofabout 65% of a peculiar pectic polysaccha-ride. According to its structural analysis(table V), this latter polymer is a verydensely branched polysaccharide (~50% ter-minal arabinose and galactose) with only afew short chains of 5-linked arabinose, and3-, 6- and 3,6-linked galactose. Since we didnot analyze the structure of the acidic moi-ety, we are not able at the moment to iden-tify the points of branching, possibly galac-turonic acid. French showed that thismucilage is secreted into the intralocularand inter-trichome medium [11]; upon sec-tion, it shows in transmission electronmicroscopy a non-crystalline feather-likecriss-cross, suggesting an organized matrix.Similarly, trichomes from Araceae (mono-cotyledons) produce and secrete enormousquantities of it, which often fill the entire loc-ule [43].

These mucilaginous substances arepresent on all parts of the gynoecium ofmany plants [49] and their roles in the facil-itation of the pollen tube journey from thestigma to the ovule micropyle are not fullyelucidated [26]. Mollet et al. reported that,after extensive water-washing of style frag-ments from lily (a monocot), a crudepolysaccharide mixture was obtained bysoaking in 0.5 M imidazole-HCl buffer(pH 7.0) [50]; this and a stigma/stylarcystein-rich protein (adhesin) bind to eachother and promote adhesion of pollen tubesto the stigma epidermal extracellular matrix.Its chemical composition was very similar toours with respect to arabinose, mannose,galactose, glucose and glucuronic acidmolar ratios. The lily extract was, however,relatively richer in galacturonic acid. Never-theless, its [galacturonic acid/rhamnose]

238 Fruits, vol. 64 (4)


mole ratio (8.3) was similar to ours (8.8).Thus, it seems that, amongst vanilla com-plex mucilaginous polysaccharides, thepectic fraction would be a typical rham-nogalacturonan I with a high degree of inter-spersion of rhamnosyl units in the acidicbackbone [51]. However, in the case ofvanilla, since trichomes differentiate onlyafter fertilization [5], their mucilaginoussecretion does not play a part in pollen tubeguidance and nutrition; it is not known ifsimilar secretions originating from other celltypes exist at the pre-fecundation stage.Finally, the role of this polysaccharidicmucilage in the mature fruit, although hav-ing very strong similarities to lily pollen tubeadhesion pectin, remains unknown.

5. Conclusion

The green mature vanilla pod constitutesswollen placental laminae stuffed with glu-covanillin and related phenolics; secretorytubular trichomes filled with oleoresin richin γ-pyranone derivatives, glucovanillin anda mucilage, polysaccharidic in nature, alsopresent in the intralocular interstitial cell-free medium; and a fleshy mesocarp devoidof these components. Compared with theplacentae, trichomes store minor quantitiesof glucovanillin. Both placental laminae andtrichomes apparently harbor the enzymaticsystems for the synthesis of p-hydroxyben-zaldehyde, vanillin and related phenolicsand their further glucosylation. However,the implicated enzymes might be differentsince multiple routes are possible [52], andtheir biosynthesis and fate seem to be dif-ferently regulated in both compartments.Finally, the physiological roles, if any, ofglucovanillin and related phenolic gluco-sides, long-chain γ-pyranones, and muci-lage remain totally unknown.

Acknowledgements

We thank C. Dhuique-Mayer (UMR Qual-isud, CIRAD, Montpellier, France) for theseparation of carotenoids; P. Alter andG. Morel for their helpful technical assist-

ance; B. Guyot and N. Durand (UMR Qual-isud) for HPLC with fluorescence detection;and F. Lamotte and S. Catrix (UMR DAP,CIRAD-INRA, Montpellier, France) for theirhelp with spectrofluorimetry. We thankG. Conejero (Plant Cell Histocytology andImaging Platform, CIRAD-MIR, Montpellier,France) for epifluorescence microscopy. Wealso wish to thank M. Bourgois (Eurovanille,France) for the gift of Indian vanilla pods.

References

[1] Cameron K., Biogeography of Vanilloideae(Orchidaceae). 16th International BotanicalCongress (abstract), St. Louis, Missouri,USA, 1–7 August, 1999.

[2] Portères R., Le genre Vanilla et ses espèces,in: Bouriquet G. (Ed.), Le vanillier et la vanilledans le monde, Lechevalier, Paris, France,1954, pp. 94-290.

[3] Sotos-Arenas M.A., Vanilla, in: PridgeonA.M., Cribb P.J., Chase M.W., RamunsenF.N. (Eds.), Genera Orchidacearum, OxfordUniv. Press, UK, 2003, pp. 321–334.

[4] Stern W.L., Judd W.S., Comparative vegeta-tive anatomy and systematics of Vanilla(Orchidaceae), Bot. J. Linn. Soc. 131 (1999)353–382.

[5] Swamy B.G.L., On the life history of Vanillaplanifolia, Bot. Gaz. 108 (1947) 449–456.

[6] Cameron K.M., Recent advances in the sys-tematic biology of vanilla and related orchids(Orchidaceae: subfamily Vanilloideae), in:Proc. Vanilla, First Int. Cong., Princeton,USA, Carol Stream, Allured Publ. Corp.,USA, 2005, pp. 89–93.

[7] Havkin-Frenkel D., French J.C., Pak F.,Frenkel C., Inside vanilla: Vanilla planifolia’sbotany, curing options and future marketprospects, Perfum. Flavour. 30 (2005) 36–55.

[8] De Lanessan J.-L., Vanille, in: De LanessanJ-L. (Ed.), Les plantes utiles des coloniesfrançaises, Impr. Natl., Paris, France, 1886,pp. 62–74.

[9] Roux P., Études morphologiques et anato-miques dans le genre Vanilla, in: BouriquetG. (Ed.), Le vanillier et la vanille dans lemonde, Lechevalier, Paris, France, 1954,pp. 44–92.


Fruits, vol. 64 (4) 239

[10] Odoux E., Escoute J., Verdeil J.-L., BrillouetJ.-M., Localization of β-glucosidase activityand glucovanillin in vanilla bean (Vanilla plan-ifolia Andrews), Ann. Bot. 92 (2003) 437–444.

[11] French J.C., Development of vanilla-bearingplacental trichomes, in: Proc. Vanilla, FirstInt. Cong., Princeton, USA, Carol Stream,Allured Publ. Corp., USA, 2005, pp. 71–77.

[12] Ramaroson-Raonizafinimanana B., GaydouE.M., Bombarda I., Long-chain γ-pyrones inepicuticular wax of two vanilla bean species:V. fragrans and V. tahitensis, J. Agric. FoodChem. 47 (1999) 3202–3205.

[13] Podstolski A., Havkin-Frenkel D., MalinowskiJ., Blount J.W., Kourteva G., Dixon R.A.,Unusual 4-hydroxybenzaldehyde synthaseactivity from tissue cultures of the vanillaorchid Vanilla planifolia, Phytochem. 61(2002) 611–620.

[14] Odoux E., Chauwin A., Brillouet J.-M., Purifi-cation and characterization of vanilla bean(Vanilla planifolia Andrews) β-D-glucosidase,J. Agric. Food Chem. 51 (2003) 3168–3173.

[15] Havkin-Frenkel D., Belanger F.C., Applica-tion of metabolic engineering to vanillin bio-synthetic pathways in Vanilla planifolia, in:Verpoorte R. (Ed.), Applications of plant met-abolic engineering, Springer, Germany,2007, pp. 175–196.

[16] Pak F.E., Gropper S., Dai W.D., Havkin-Frenkel D., Belanger F.C., Characterizationof a multifunctional methyltransferase fromthe orchid Vanilla planifolia, Plant Cell Rep.22 (2004) 959–966.

[17] Li H.M., Rotter D., Hartmann T.G., Pak F.E.,Havkin-Frenkel D., Belanger F., Evolution ofnovel O-methyltransferases from the Vanillaplanifolia caffeic acid O-methyltransferase,Plant Mol. Biol. 61 (2006) 537–552.

[18] Joel D.M., French J.C., Graft N., KourtevaG., Dixon R.A., Havkin-Frenkel D., A hairytissue produces vanillin, Isr. J. Plant Sci. 51(2003) 157–159.

[19] Arana F.E., Action of a β-glucosidase in thecuring of vanilla, Food Res. 288 (1943) 343–351.

[20] Jones M.A., Vicente G.C., Criteria for testingvanilla in relation to killing and curing meth-ods, J. Agric. Res. 78 (1949) 425–434.

[21] Odoux E., Escoute J., Verdeil J.-L., The rela-tion between glucovanillin, β-glucosidaseactivity and cellular compartmentation dur-ing the senescence, freezing and traditional

curing of vanilla beans, Ann. Appl. Biol. 149(2006) 43–52.

[22] Márquez O., Waliszewski K.N., The effect ofthermal treatment on β-glucosidase inacti-vation in vanilla bean (Vanilla planifoliaAndrews), Int. J. Food Sci. Technol. 43(2008) 1993–1999.

[23] Siebert M., Sommer S., Li S.-M., Wang Z.-X.,Severin K., Heide L., Genetic engineering ofplant secondary metabolism. Accumulationof 4-hydroxybenzoate glucosides as a resultof the expression of the bacterial ubiC genein tobacco, Plant Physiol. 112 (1996) 811–819.

[24] Flammarion C., Oléo-résine, in: FlammarionE. (Ed.), Dictionnaire encyclopédique univer-sel, Lahure, Paris, France, 1902, pp. 49.

[25] Maestro Y., Bergia D., Lasserre C., Vanillaplanifolia extract, method for obtainingsame, and cosmetic or dermatological com-position containing same, World Pat. WO2007/ 034042 A2, 2007.

[26] Webb M.C., Williams E.G., The pollen tubepathway in the pistil of Lycopersicon peruvi-anum, Ann. Bot. 61 (1988) 414–423.

[27] Perez Silva A., Contribution à l’étude de lagenèse des composés d’arôme au cours duprocédé mexicain de transformation de lavanille (Vanilla planifolia Jackson), PhD The-sis, Univ. Montpellier II, France, 2006.

[28] Brundrett M.C., Kendrick B., Peterson C.A.,Efficient lipid staining in plant material withSudan Red 7B or Fluoral Yellow 088 in poly-ethylene glycol-glycerol, Biotech. Histo-chem. 66 (1991) 111–116.

[29] Odoux E., Changes in vanillin and glucova-nillin concentrations during the variousstages of the process traditionally used forcuring Vanilla fragrans in Réunion, Fruits 55(2000) 119–125.

[30] Huber D.J., Acidified phenol alters tomatocell wall pectin solubility and calcium con-tent, Phytochem. 30 (1991) 2523–2527.

[31] Albersheim P., Nevins D.J., English P.D., KarrA., A method for the analysis of sugars inplant cell-wall polysaccharides by gas-liquidchromatography, Carbohydr. Res. 5 (1967)340–345.

[32] Harris P.L., Henry R.L., Blakeney A.B., StoneB.A., An improved procedure for the methyl-ation analysis of oligosaccharides andpolysaccharides, Carbohydr. Res. 127(1984) 59–73.

240 Fruits, vol. 64 (4)


[33] Blumenkrantz N., Asboe-Hansen G., Newmethod for determination of uronic acids,Anal. Biochem. 54 (1973) 484–489.

[34] Brillouet J.-M., Williams P., Will F., Müller G.,Pellerin P., Structural characterization of anapple juice arabinogalactan-protein whichaggregates following enzymic dearabino-sylation, Carbohydr. Polym. 29 (1996) 271–275.

[35] Hakomori S.I., A rapid permethylation of gly-colipids and polysaccharides catalysed bymethyl sulfinyl carbanion in dimethyl sulfox-ide, J. Biochem. (Tokyo) 55 (1964) 205–208.

[36] Pellerin P., Doco T., Vidal S., Williams P.,Brillouet J.-M., O’Neill M.A., Structural char-acterization of red wine rhamnogalacturonanII, Carbohydr Res. 290 (1996) 183–197.

[37] Havkin-Frenkel D., Podstolski A., Vanillinproduction, US Patent US 2007/ 0044174A1, 2007.

[38] Michodjehoun-Mestres L., Amraoui W.,Brillouet J.-M., Isolation, characterization, anddetermination of 1-O-trans-cinnamoyl-β-D-glucopyranose in the epidermis and flesh ofdeveloping cashew apple (Anacardium occi-dentale L.) and four of its genotypes, J. Agric.Food Chem. 57 (2009) 1377–1382.

[39] Mangin A., Les vanilliers, in: Mame A. & sons(Eds.), Les plantes utiles, Tours, France,1886, pp. 134–141.

[40] Angeles G., Berrio-Sierra J., Joseleau J.-P.,Lorimier P., Lefèbvre A., Ruel K., Preparativelaser capture microdissection and single-potcell wall material preparation: a novelmethod for tissue-specific analysis, Planta224 (2006) 228–232.

[41] Ishikawa H., Kuwano A., Matsumoto K.,Complexation of vanillin and ethylvanillinwith α-, β-, and γ-cyclodextrin, J. Fac. Agric.Kyushu Univ. 52 (2007) 87–90.

[42] Sircar D., Mitra A., Evidence for p-hydroxy-benzoate formation involving enzymaticphenylpropanoid side-chain cleavage in

hairy roots of Daucus carota, J. Plant Phys-iol. 165 (2008) 407–414.

[43] Dalmer M., Ueber die Leitung der Pollen-schlauche bei den Angiosperm, JenaZeitsch. Naturwiss. 141 (1880) 530–566.

[44] Tilton V.R., Horner H.T., Stigma, style andobturator of Ornithogalum caudatum (Lil-iaceae) and their function in the reproductiveprocess, Am. J. Bot. 67 (1980) 1113–1131.

[45] French J.C., Structure of ovular and placen-tal trichomes of Araceae, Bot. Gaz. 148(1987) 198–208.

[46] Fahn A., Shimony C., Glandular trichomes ofFragonia L. (Zygophyllaceae) species: struc-ture, development and secreted materials,Ann. Bot. 77 (1996) 25–34.

[47] Simony R., Duquénoy P., Compléments à laconnaissance anatomique et histochimiquedes fruits de vanilliers cultivés, en particulierdans les départements et territoires françaisd’outre-mer, Etudes O.-M 36 (1953) 129–134.

[48] Katsuraya K., Okuyama K., Hatanaka K.,Oshima R., Sato T., Matsuzaki K., Constitu-tion of konjac glucomannan: chemical anal-ysis and 13C NMR spectroscopy, Carbohydr.Polym. 53 (2003) 183–189.

[49] Lord E.M., Adhesion and guidance incompatible pollination, J. Exp. Bot. 54 2003)47–54.

[50] Mollet J.-C., Park S.-Y., Nothnagel E.A., LordE.M., A lily stylar pectin is necessary for pol-len tube adhesion to an in vitro stylar matrix,Plant Cell 12 (2000) 1737–1749.

[51] O’Neill M., Albersheim P., Darvill A.G., Thepectic polysaccharides of primary cell walls,in: Dey P.M., Harborne J.B. (Eds.), Methodsin plant biochemistry, Acad. Press, Newyork,USA, 1990, pp. 415–441.

[52] Funk C., Brodelius P.E., Phenylpropanoidmetabolism in suspension cultures of Vanillaplanifolia Andr., Plant Physiol. 94 (1990)102–108.


Fruits, vol. 64 (4) 241

Anatomía, histoquímica, y bioquímica de los sitios de acumulación de laglucovainillina, de una oleorresina, y de un mucílago en la vaina de vainillaverde madura (Vanilla planifolia, Orchidaceae): re-examen crítico deconjunto.

Resumen – Introducción. Las vainas de vainilla verdes maduras acumulan 4-O-(3-metoxi-benzaldehído)-β-D-glucoside (glucovainillina), que, mediante hidrólisis con la ayuda de unaβ-glucosidasa endógena, libera vainillina, el mayor compuesto de aroma de la vainilla. Semencionaron desde hace décadas de modo controvertido los sitios de almacenamiento de laglucovainillina en la vaina; mediante el empleo de una terminología técnica precisa y amplia-mente aceptada, nos proponemos esclarecer dicha controversia presentando una tabla basadaen pruebas anatómicas, histoquímicas y bioquímicas, de los sitios de acumulación de la glu-covainillina. La vaina sintetiza asimismo una oleorresina y un mucílago de compuestosdesconocidos; facilitamos aquí sus respectivas localizaciones y estructuras. Material y méto-dos. La anatomía de la vaina se examinó en microscopía fotónica y de epifluorescencia. Sepuso en marcha un protocolo que permite una disección manual precisa de las diferentespartes anatómicas de la vaina (mesocarpio, placentae, tricomas, zona acelular intersticial dellóculo y semillas). Mientras que se extrajeron y analizaron la glucovainillina y los γ-pyranonesmediante HPLC, las estructuras de los polisacáridos del mucílago se determinaron mediantepermetilación. Resultados y discusión. La glucovainillina se almacena esencialmente en losplacentae (92 %) y marginalmente en los tricomas (7 %); se detectaron rastros en el mesocar-pio y el medio acelular intersticial del lóculo. Los tricomas acumulan cantidades masivas deuna oleorresina fluorescente (44 %) y rica en alkenylmethyldihydro-γ-pyranones y sintetizanun mucílago constituido por un glucomanano y un polisacárido héctico de cadenas lateralesmonoméricas de arabinosa y de galactosa. Conclusión. Hasta el día de hoy permanecendesconocidos los papeles fisiológicos que desempeñan la glucovainillina, los pyranones delargas cadenas y el mucílago.

Francia / Vanilla planifolia / Orchidaceae / vainilla / tricomas / vanillina /oleoresinas / fluorescencia / mucilagos / polisacáridos

Anatomy, histochemistry and biochemistry of glucovanillin ...

Documents

Transcript of Anatomy, histochemistry and biochemistry of glucovanillin ...