It is well documented that genes involved in primary metabolic processes such as photosynthesis are circadianly regulated (Sugiyama et al., 2001). For example, in crassulacean acid metabolism plants, the gene for phosphoenolpyruvate carboxylase kinase is regulated by a circadian oscillator to enhance enzyme activity in the dark (Bakrim et al., 2001). In other branches of metabolism, attention has been paid to the circadian pattern of plant hormones such as abscisic acid (Thompson et al., 2000), or photoprotective pigments, including phenylpropanoids (Borevitz et al., 2000). However, reports on the biosynthetic rhythm of terpenes, the biggest group of plant secondary metabolites, has been limited to in planta chemical analysis and emission profiles of flowers, grasses, shrubs, and trees (Bertin et al., 1997; Hansen et al., 1997; Helsper et al., 1998; He et al., 2000a, 2000b).

In nature, plants are the main source of volatile organic compounds (VOCs) emitted to the atmosphere, among which some terpenoids and benzenoic compounds are released in diurnal or nocturnal patterns (Loreto et al., 1996; Staudt et al., 1997; Kolosova et al., 2001). β-Pinene is one of the most widely detected VOCs (Geron et al., 2000), and in some plants, α-pinene, β-pinene, and sabinene may account for more than 80% of the monoterpenes emitted (Bertin et al., 1997). In Quercus ilex grown under natural conditions, terpene emission was found to peak at noon, with the quantity shifting with light intensity and temperature (Bertin et al., 1997; Staudt and Bertin, 1998). The emission of α- and β-pinene in laboratory tests peaked in as short as 30 min after the light was switched on (Loreto et al., 1996). Because this plant has no special storage tissue (such as the oil glands of peppermint, Mentha piperita) for terpenoids produced, the emission pattern implies fluctuations in the activity of specific enzymes, which could be controlled at any step from transcription to posttranslational modification, or in the availability of substrate.

The first committed reactions in the formation of the various classes of terpenes are catalyzed by terpene synthases, which convert acyclic diphosphate substrates (geranyl diphosphate [GPP], farnesyl diphosphate [FPP], geranylgeranyl diphosphate [GGPP], etc.) into acyclic or cyclic terpenes (compare with Greenhagen and Chappell, 2001). These terpene products may be further modified by other enzymes, such as P450 monooxygenases, leading to various terpenoid derivatives (Schuler, 1996). Monoterpenes, which are formed from GPP by monoterpene synthases, are widely present in plants (Croteau, 1987), and some of them account for the flower scent of plants (e.g. Borg-Karlson et al., 1994). In different developmental stages of peppermint, coincident temporal change in enzyme activities, enzyme protein levels, and steady-state transcript abundances of monoterpene synthases was demonstrated, indicating that most of the monoterpene synthases in this plant are regulated at the level of gene expression (Gershenzon et al., 2000; McConkey et al., 2000).

Artemisia annua is an annual herb widely distributed in Asia, Europe, and North America. Different from most of the other plants in the Asteraceae family, the pollination of A. annua is mediated by wind or insects (McVaugh, 1984). The plant is rich in terpenoids (van Geldre et al., 1997; Tan et al., 1998). Among these are the sesquiterpene lactone artemisinin, which is one of the most efficient drugs against Plasmodium species involved in malaria (van Agtmael et al., 1999), and monoterpenes camphor, 1, 8-cineole, α-pinene, and β-pinene, etc. (Ahmad and Misra, 1994), which give the plant a sweet scent. A key enzyme for artemisinin synthesis, amorpha-4, 11-diene synthase, has recently been purified, and its cDNA was cloned (Bouwmeester et al., 1999; Mercke et al., 2000). At least two monoterpene synthases (QH1 and QH5) were proved to be inducible at the transcriptional level by mechanical wounding. In vitro assay showed that both of them converted GPP into (3R)-linalool, which, however, could not be detected from the plant extract (Jia et al., 1999). Here, we report a new A. annua monoterpene synthase cDNA, QH6, which encodes a (−)-β-pinene synthase. This is the first reported terpene synthase of which transcripts and enzymatic products were shown to fluctuate in a circadian pattern.

Figure 1 Nucleotide and deduced amino acid sequences of QH6. The conserved motifs are in boldface. Positions and directions of primers used are underlined and labeled. The HindIII site, before which the fragment is used as template for probe labeling in northern (more ...)

Figure 2 Phylogenetic analysis of monoterpene synthases of which the enzymatic functions reported. The neighbor-joining method was used for the protein sequences. Bootstrap values were labeled besides the branches, and the scale bar indicates 20% sequence (more ...)

The recombinant plasmid pET32b/QH6, in which the transit peptide sequence had been truncated, resulted in successful expression of QH6 fusion protein in Escherichia coli after isopropyl β-d-thiogalactoside induction. With GPP as substrate, the fusion protein catalyzed the formation of two products in the ratio of 94:6, as revealed by two peaks in gas chromatography (GC). Mass spectra (MS) showed these to be β-pinene (the major product) and α-pinene (the minor product; Fig. 3). By coinjection with authentic standards for GC, both products were found to be (−)-enantiomers (Fig. 4). These results were supported by ¹H nuclear magnetic resonance (NMR) spectra of the products. The signals from the major product were 4.625 (br s, 1H), 4.556 (br s, 1H), 2.539 (m, 1H), 2.455 (t, 5.4 Hz, 1H), 2.316 (dtd, 9.9 Hz, 5.8 Hz, 1.2 Hz, 1H), 2.248 (m, 1H), 1.975 (m, 1H), 1.84 (m, 1H), 1.82 (m, 1H), 1.420 (d, 9.8 Hz, 1H), 1.236 (s, 3H), and 0.720 (s, 3H), and those from the minor product were 5.185 (t of sextet, 3 Hz, 1.5 Hz, 1H), 2.162 (d of sextet, 17.5 Hz, 2.5 Hz, 1H), 1.931 (td, 5.6 Hz, 1.5 Hz, 1H), 1.658 (br q, approximately 2 Hz, 3H), 1.151 (d, 8.5 Hz, 1H), and 0.835 (s, 3H). These signals are within experimental error of literature values (Badjah-Hadj-Ahmed et al., 1992). These in vitro results indicate that QH6 encodes a β-pinene synthase, which produces α-pinene as a minor product. The 94:6 ratio of β-:α-pinene in in vitro assay was similar to that measured from A. annua leaf extract (92:8), but higher than that from stems (89:11) or inflorescences (79:21).

Figure 3 Identification of the major enzymatic product as β-pinene, and the minor as α-pinene. MS are of the major product (A), authentic β-pinene (B), minor product (C), and authentic α-pinene (D).

Figure 4 Enantiomer-selective GC identification of α- and β-pinene produced by the recombinant enzyme are (−)-enantiomers. Left, Identification of the major product; right, identification of the minor product. A, A mixture of (+)- (more ...)

Figure 5 Transcript abundance of QH6 in different organs of A. annua A, Northern blot; B, ethidium bromide-stained gel for monitoring equal amount of total RNA loaded per lane. Each of 20 μg of total RNA was isolated from roots (lane 1), stems (lane 2), (more ...)

Figure 6 Effects of mechanical wounding (approximately 40% damage of upper leaf surface; A) and Verticillium dahliae elicitor treatment (1 μg of Suc equivalent mL−1; B) on the steady-state level of QH6 transcripts. In each panel, an ethidium (more ...)

Figure 7 Circadian pattern of transcriptional regulation of QH6. A, Fluctuation of steady-state level of QH6 transcripts with day-night shifting (d 1–3); B, when plants were entrained by LL photoperiod, the oscillation pattern was in advanced cycles (d (more ...)

We then investigated if other environmental factors could affect QH6 expression. Examination of the diurnal pattern of QH6 expression by northern-blot analysis showed that the steady-state mRNA level fluctuated with the day-night rhythm. With a 12 h/12 h (light/dark, LD) photoperiod, the QH6 mRNA level was generally higher in the day (L) than in the night (D). However, the mRNA level started to increase around 9th h in the dark phase (D9), approximately 3 h prior to the light phase. This accumulation reached its peak level in 12 h, i.e. at the 9th h in light (L9; Fig. 7A).

A key issue is whether plants of A. annua could still maintain a circadian pattern of QH6 transcript accumulation when entrained in constant light (LL) or constant dark (DD; Fig. 7). After the plants were moved to a LL regimen for 3 d (d 6 of Fig. 7), the fluctuation was still clearly detectable, with the peak level appearing at D9, about 12 h ahead that of the original cycle at L9. The rhythm was further accelerated, as in d 8, two peaks appeared at around D9 and L12, respectively (Fig. 7B). When plants were entrained in DD, the transcripts accumulated much slower, and the level did not reach the peak until phase D3 in d 4. At the 3rd d under DD (d 6), the peak was delayed by about 15 h, and it was delayed by 12 h on d 8. In addition, the oscillation magnitude was reduced after 5 d of DD treatment (Fig. 7C). Therefore, the fluctuation cycles were advanced under LL and were delayed under DD conditions. When plants were returned from LL to LD, the peak of QH6 transcript abundance stayed at phase L12 in the 1st d, and then went back to its original phase L9 on the day after (Fig. 7B). For plants from DD, the fluctuation also returned to its original phase on the 2nd d under LD (Fig. 7C).

Figure 8 Diurnal fluctuation of β-pinene content. A, β-Pinene in total VOCs collected in each 3-h time phase (nanograms per plant); B, β-pinene in volatiles extracted from juvenile leaves (nanograms per gram of fresh weight). Time phases (more ...)

It is interesting to note that the QH6 mRNA level and in planta (−)-β-pinene content displayed a circadian pattern of variation, and their peak levels even appeared in the same time phase (L9). This suggests that the biosynthesis of (−)-β-pinene in A. annua is largely controlled at the transcriptional level. Such a regulatory pattern is similar to that of monoterpene biosynthesis in peppermint glandular trichomes (McConkey et al., 2000), which showed coincidental temporal changes in steady-state transcript abundance and enzyme activities. Furthermore, the similar fluctuation of (−)-β-pinene emission suggests that this monoterpene is probably not stored in glandular trichomes for long periods of time, but is released directly after synthesis from glandular or nonglandular tissues. Although α- and β-pinene were reported to take part in direct plant defense against herbivores and pathogens, allelopathy, and plant pollination (Langenheim, 1994), little is known about the biological or ecological function of (−)-β-pinene in A. annua. In contrast to QH1 and QH5, which could be induced by wounding, QH6 transcription was suppressed by mechanical wounding or fungal elicitation. It seems unlikely that (−)-β-pinene synthase is involved in the chemical defense of A. annua. One possible explanation for this suppression is that those up-regulated “defensive enzymes” such as QH1 and QH5 would have more substrate (GPP) available, if other, possibly nondefensive, monoterpene synthases were down-regulated. A similar result was obtained with ponderosa pine (Pinus ponderosa) in which it was shown that an increase in limonene content was accompanied by the decreased levels of α-pinene, β-pinene, 3-carene, and myrcene (Sturgeon, 1979).

It should be mentioned that (−)-β-pinene is not the only example of a plant volatile that can be synthesized and emitted in a diurnal mode. Floral volatiles from rose (Rosa hybrida L. cv Honesty; Helsper et al., 1998) and snapdragon (Antirrhinum majus; Kolosova et al., 2001), as well as damage-induced volatiles of cotton (Gossypium hirsutum; Loughrin et al., 1994), are emitted in a clear circadian rhythm. Delfine et al. (2000) and Sharkey (1996) pointed that monoterpene might improve thermotolerance of photosynthesis in leaves and all isoprenoids are thought to protect membrane from denaturation. It is possible that a peak amount of pinene, as found in this study for A. annua, is synthesized and released for protective purpose when light intensity and temperature are reaching their maximal values under nature conditions.

Isopentenyl diphosphate (IPP) is the biosynthetic precursor of all the terpenoids. In Arabidopsis, transcripts of DXS, the first enzyme in the pathway to IPP biosynthesis in plastids (Lichtenthaler, 1999), vary in a diurnal pattern that peaks at 4 h in the light phase (Estévez et al., 2000; Harmer et al., 2000). DXR generates IPP precursors and was thought to catalyze the critical step in the entire pathway (Mahmoud and Croteau, 2001). With A. annua, we found that both QH6, one of the monoterpene synthases, and DXR, a key enzyme of the plastid isoprenoid pathway, are up-regulated in the later light phase (S. Lu and X.-Y. Chen, unpublished data) when the biological clock in plant is set to allocating assimilated carbohydrates to different pathways (Harmer et al., 2000). At this point, it is still unclear whether the QH6 transcription is directly regulated by an intrinsic circadian clock or by some intermediates in the pathway downstream of DXR, e.g. IPP or GPP. It remains to be learned what signal regulates DXR and QH6 transcription.

For circadian analysis, a group of cultivated plants were entrained under LD. One group was kept under LD as control, and the rest were divided for two treatments. One was removed to LL, and the other to DD. After 5 d, all of these treated plants were returned to the original LD cycles. Juvenile leaves were collected from 3 d prior to different photoperiod treatments to 1 d after returning to LD, at 3-h intervals.

In each of the triplicate experiments, the plants were destroyed once a leaf was sampled.

The transformed E. coli cells were cultured overnight at 37°C in Luria-Bertani (LB) medium supplemented with 100 μg mL⁻¹ ampicillin. A 500-μL aliquot of the saturated culture was used to inoculate 50 mL of fresh LB medium containing 100 μg mL⁻¹ ampicillin. After adding 1 mm isopropyl β-d-thiogalactoside, the induced culture was allowed to grow at 22°C overnight. The cells were pelted and resuspended in 10 mL of cold assay buffer (Jia et al., 1999) and were disrupted by sonication. After centrifugation, the supernatant was then transferred to a glass tube and the insoluble fraction was resuspended in 10 mL of the same assay buffer.

To obtain sufficient product for NMR analysis, a large-scale preparation was conducted by inoculating 750 μL of an overnight culture into 750 mL of LB medium containing 100 μg mL⁻¹ ampicillin. The culture was prepared similarly to the small-scale assay, except that the supernatant of the sonicate was adjusted to a final concentration of 10 mm MgCl₂, 20 μm MnCl₂, and 200 μm GPP. Cyclopentane (30 mL) was used instead of pentane to minimize the solvent interference in subsequent NMR analysis. After incubation at 30°C overnight, the reaction mixture was extracted with cyclopentane (3 × 100 mL). The combined organic extracts were passed through a silica gel column (8 × 2 cm i.d.) overlaid with MgSO₄ (2 cm) to yield the terpene hydrocarbon fraction. A 2-μL aliquot was analyzed by GC and the total amount of product was estimated to be approximately 1 mg based on the peak area-to-weight ratio calculated from the (−)-β-pinene standard. The solvent was evaporated under a stream of nitrogen to near dryness for NMR analysis.

GC analyses were performed on a 6890 system (Hewlett-Packard, Palo Alto, CA) equipped with a Rtx-5 capillary column (30-m × 0.25-mm i.d., 0.10 μm df; Restek, Bellefonte, PA). Separation conditions were the following: injection port 200°C, flame ionization detector 250°C, a split ratio of 39:1 for plant extracts and enzymatic products, and splitless for volatile collections, and helium flow at 20 cm s⁻¹ (0.6 mL min⁻¹). The temperature program was 50°C for 5 min, an increase to 250°C (10°C min⁻¹), and 250°C for 5 min. GC-MS was performed on a 5890A instrument (Hewlett-Packard) equipped with a DB-5 ms column (60-m × 0.25-mm i.d., 0.10 μm df; J&W Scientific, Folsom, CA). Separation was achieved with splitless injection at 200°C, helium flow at 30 cm s⁻¹ (1 mL min⁻¹), and the identical temperature program as above. Mass spectra (m/z 35–500) were obtained on a ZAB-HF reverse-geometry double-focusing instrument at 70 eV with an electron-impact ion source (200°C). The accelerating voltage was 8 kV and the resolution was 1,000 (10% valley). Enantioselective GC was performed on a GC-9A instrument (Shimadzu, Tokyo) with a CycloSil-B capillary column containing a modified β-cyclodextrin stationary phase (30-m × 0.25-mm i.d., 0.25 μm df; J&W Scientific). The separation of chiral components was achieved with injection port and flame ionization detector at 200°C, helium flow at 29 cm s⁻¹ (0.8 mL min⁻¹), a 36:1 split ratio, and the following temperature program: 50°C for 1 min and a 5°C min⁻¹ ramp to 200°C, where the temperature was held for 10 min. Authentic (+)-α-pinene, (−)-α-pinene, (+)-β-pinene, and (−)-β-pinene from Sigma-Aldrich (St. Louis) were used as standards. The QH6 product was also coinjected with authentic samples to determine the pinene stereochemistries.

¹H NMR spectra were obtained on a AMX500 spectrometer (500.1 MHz for ¹H; Bruker, Billerica, MA) at 25°C in CDCl₃ solution (approximately 10 mm) and were referenced to internal tetramethylsilane.

For northern-blot analysis, a total of 20 μg of RNA was loaded per lane, and an ethidium bromide-stained gel or a duplicated blot probed for UEP transcripts was used to monitor the amount of RNA loaded. A fragment containing the first 566 bp of QH6 (released by EcoRI and HindIII digestion of pBK-CMV/QH6), which showed a relatively lower sequence identity with QH1 or QH5 (Jia et al., 1999), was ³²P labeled by the Prime-a-Gene System (Promega) to detect QH6 transcripts on the membrane. The blots were hybridized, washed as described (Ausubel et al., 1995), and exposed to x-ray film at −70°C. The images were collected with GeneGenius gel documentation system and were then digitalized with GeneTools software (Syngene, Cambridge, UK). Corresponding data from UEP transcripts were used for normalization.

We thank Prof. Rodney Croteau for help with the entire project. We are also grateful to Dr. William Wilson for his expert assistance in NMR, to Drs. Yong-Ping Huang and Hong-Chun Zhou for their assistance in chemical analysis, and to Elizabeth A. Hart, Hala G. Schepmann, and Dr. Xiang-Jun Zhou for helpful discussions.