Biosynthesis and emission of terpenoid volatiles from Arabidopsis flowers

Abstract

Arabidopsis is believed to be mostly self-pollinated, although several lines of genetic and morphological evidence indicate that insect-mediated outcrossing occurs with at least a low frequency in wild populations. Here, we show that Arabidopsis flowers emit both monoterpenes and sesquiterpenes, potential olfactory cues for pollinating insects. Of the 32 terpene synthase genes in the Arabidopsis genome, 20 were found to be expressed in flowers, 6 of these exclusively or almost exclusively so. Two terpene synthase genes expressed exclusively in the flowers and one terpene synthase gene expressed almost exclusively in the flowers were characterized and found to encode proteins that catalyze the formation of major floral volatiles. A beta-glucuronidase fusion construct with a promoter of one of these genes demonstrated that gene expression was restricted to the sepals, stigmas, anther filaments, and receptacles, reaching a peak when the stigma was receptive to cross pollen. The observation that Arabidopsis flowers synthesize and emit volatiles raises intriguing questions about the reproductive behavior of Arabidopsis in the wild and allows detailed investigations of floral volatile biosynthesis and its regulation to be performed with this model plant system.

Figure 1.

Volatile Terpenes Emitted from Arabidopsis (Columbia Ecotype). (A) Release rates of major terpenes from 6-week-old flowering plants and parts of these plants determined by dynamic headspace sampling. Inflorescences were the source of emission for the monoterpenes and most of the sesquiterpenes. Data represent the mean of four replications ± se. The emission rate presented is that of a single plant per hour, although collections were made from five plants simultaneously over 8-h periods. (B) Gas chromatography of monoterpenes collected from 150 cut inflorescences during 12 h of closed-loop stripping (Donath and Boland, 1995). The peak of limonene coeluted with 2-ethyl-hexanol on the (5% phenyl)-polymethylsiloxane column, as shown here, but it was separable from this compound on a polyethylene glycol column. (C) A later portion of the same chromatogram as in (B), showing the sesquiterpene hydrocarbon region. The inset depicts a portion of the chromatogram of the same sample run on a polyethylene glycol (DB-WAX) column, which shows an improved separation of some components. Compounds marked with dots represent additional sesquiterpene hydrocarbons identified by GC-MS but not yet confirmed by comparison with authentic standards. IS indicates the internal standard, nonyl acetate. (D) Structures of the compounds identified by numbers in (A), (B), and (C). The number of each compound in (D) corresponds to the numbered columns in (A) and the numbered peaks in (B) and (C). Chirality was determined for all compounds except β-elemene and β-sesquiphellandrene.

Figure 2.

Organ-Specific Expression of Arabidopsis TPS Genes. Leaves, flowers, siliques, stems, and roots were collected from 6-week-old flowering plants. Total RNA was extracted and used for RT-PCR analysis. The expression of the 32 Arabidopsis TPS genes in each organ was examined in 30 RT-PCR procedures, of which the results from 28 are shown in this composite picture. As a result of the high sequence similarity between At4g13280 and At4g13300 and between At3g14520 and At3g14540, PCR products obtained from one or the other member of each pair of genes could not be distinguished. The results for At1g48800 and At1g48820, which showed no expression in any of the tested tissues, are not shown. M indicates DNA markers. The RT-PCR product for β-tubulin is shown at right. To prevent saturation conditions in the reactions with the β-tubulin primers, the cDNA amount was reduced twofold compared with that in the reactions with terpene synthase primers, and the number of cycles was reduced to 26, compared with 30 (see Methods for further details). The six TPS genes that are expressed exclusively or almost exclusively in the flowers are indicated with asterisks. The three TPS genes that were characterized further also are indicated (#).

Figure 3.

RNA Gel Blot Analysis of Selected TPS Genes. Gene-specific probes were amplified by PCR using internal primers as described for RT-PCR analysis, labeled with ³²P-dCTP, and hybridized to RNA gel blots. mRNA transcripts of At3g25810, At1g61680, and At5g23960 were detected in flower tissue only. R, root; Si, silique; St, stem; F, flower; L, leaf.

Figure 4.

Derived Amino Acid Sequences of Three Arabidopsis TPS Genes Involved in Floral Volatile Biosynthesis Aligned with Related Sequences. The sequences of the proteins encoded by At3g25810, At1g61680, and At5g23960 are aligned with the sequences of the monoterpene synthase 4S-limonene cyclase from spearmint (Ms4S-limo) (Colby et al., 1993) and the sesquiterpene synthase β-caryophyllene synthase from A. annua (Aaβ-caryo) (Cai et al., 2002). Amino acid residues conserved in three or more sequences are shaded. The R residue shown by a white letter on a black background in the At3g25810 sequence indicates the position of the first amino acid in the truncated At3g25810 construct described in the text (replaced by a Met). The M residue shown by a white letter on a black background in the At1g61680 sequence indicates the position of the first amino acid in the truncated At1g61680 construct described in the text.

Figure 5.

Gas Chromatographic Separation of Products Formed by Three Terpene Synthases Found in Arabidopsis Flowers. Extracts of E. coli expressing proteins derived from At3g25810, At1g61680, and At5g23960 were assayed in vitro with GPP or FPP, and products were separated on a (5%-phenyl)-methylpolysiloxane ([A] and [C]) or a β-cyclodextrin ([B]) column. Details of product separation, identification, and enantiomeric determination are described in Methods. (A) At3g25810 mature protein with GPP. Peak 1, α-pinene [98% (−)-1S]; peak 2, sabinene [75% (−)-1S, 25% (+)-1R]; peak 3, β-pinene [98% (−)-1S]; peak 4, β-myrcene; peak 5, limonene [92% (−)-4S, 8% (+)-4R]; peak 6, (Z)-β-ocimene; peak 7, (E)-β-ocimene; peak 8, α-terpinolene; peak 9, linalool (racemic, also produced in the assay of control E. coli cultures transformed with vector lacking the terpene synthase insert). (B) At1g61680 mature protein with GPP. Enantiomerically pure (+)-3S-linalool is the sole product. Unlabeled peaks also occurred in the control assay with an extract of E. coli transformed with the same expression vector lacking an insert. (C) At5g23960 with FPP. Peak 1, (−)-α-copaene; peak 2, β-elemene (chirality not determined); peak 3, (−)-(E)-β-caryophyllene; peak 4, α-humulene; peak 5, (E)-β-farnesene (also produced by the control). The asterisk indicates an unidentified product.

Figure 6.

Identification of (E)-β-Caryophyllene as the Major Enzyme Product of At5g23960 Catalysis. (A) Mass spectrum of (E)-β-caryophyllene produced by the incubation of FPP with a cell-free extract of E. coli BL21 Codon Plus expressing the At5g23960 protein. (B) Mass spectrum of authentic (E)-β-caryophyllene standard obtained under the same conditions as in (A). m/z, mass-to-charge ratio.

Figure 7.

Expression Patterns of the At3g25810 Promoter–GUS Fusion Gene during Flower Development. GUS staining was observed in sepals, stigmas, anther filaments, and receptacles of mature, but not immature, buds (A), newly opened flower (B), and older flower (C). In (D), petals and sepals were removed to show clearly the staining in the stigma of the older flower. The newly opened flower (B) is in the protogynous stage, in which the receptive stigma, protruding above the petals and the immature anthers, is accessible to cross-pollination. The older flower (C) is in the autogamous stage, in which the stamens have elongated to the level of the stigma and dehisced. The proximity of the stamens to the stigma at this stage facilitates self-pollination.