Isolation and Characterization of AaWRKY1, an Artemisia annua Transcription Factor that Regulates the Amorpha-4,11-diene Synthase Gene, a Key Gene of Artemisinin Biosynthesis

Abstract

Amorpha-4,11-diene synthase (ADS) of Artemisia annua catalyzes the conversion of farnesyl diphosphate into amorpha-4,11-diene, the first committed step in the biosynthesis of the antimalarial drug artemisinin. The promoters of ADS contain two reverse-oriented TTGACC W-box cis-acting elements, which are the proposed binding sites of WRKY transcription factors. A full-length cDNA (AaWRKY1) was isolated from a cDNA library of the glandular secretory trichomes (GSTs) in which artemisinin is synthesized and sequestered. AaWRKY1 encodes a 311 amino acid protein containing a single WRKY domain. AaWRKY1 and ADS genes were highly expressed in GSTs and both were strongly induced by methyl jasmonate and chitosan. Transient expression analysis of the AaWRKY1–GFP (green fluorescent protein) reporter revealed that AaWRKY1 was targeted to nuclei. Biochemical analysis demonstrated that the AaWRKY1 protein was capable of binding to the W-box cis-acting elements of the ADS promoters, and it demonstrated transactivation activity in yeast. Co-expression of the effector construct 35S::AaWRKY1 with a reporter construct ADSpro1::GUS greatly activated expression of the GUS (β-glucuronidase) gene in stably transformed tobacco. Furthermore, transient expression experiments in agroinfiltrated Nicotiana benthamiana and A. annua leaves showed that AaWRKY1 protein transactivated the ADSpro2 promoter activity by binding to the W-box of the promoter; disruption of the W-box abolished the activation. Transient expression of AaWRKY1 cDNA in A. annua leaves clearly activated the expression of the majority of artemisinin biosynthetic genes. These results strongly suggest the involvement of the AaWRKY1 transcription factor in the regulation of artemisinin biosynthesis, and indicate that ADS is a target gene of AaWRKY1 in A. annua.

Introduction

Artemisinin, isolated from traditional Chinese herb Artemisia annua L. by Chinese scientists in the 1970s, is a sesquiterpene lactone endoperoxide that provides the basis for effective treatments of malaria, especially for the cerebral and the chloroquine-resistant forms of this disease. Artemisinin-based combination therapies (ACTs) have now been recommended since 2001 by the World Health Organization (WHO) to reduce the odds of drug resistance. Currently, artemisinin derivatives are the only available drugs that are globally effective against the malaria parasite. Besides their antimalarial activity, artemisinins have also recently been reported to possess antiviral (Romero et al. 2006), anticancer (Efferth 2006) and antischistosomal activities (Utzinger et al. 2007). At present, the A. annua plant is still the only commercial source of artemisinins. The low artemisinin content in A. annua, ranging from 0.1 to 0.8% of the plant dry weight, makes artemisinin a relatively expensive drug, especially for economically disadvantaged people in developing countries where most malaria occurs. One of the most promising approaches to reduce the price of artemisinin-based antimalarial drugs is metabolic engineering of the plant in an attempt to obtain a higher artemisinin content. However, more knowledge about the biosynthetic pathway is required for this purpose.

The first committed step of artemisinin biosynthesis in A. annua is the cyclization of farnesyl diphosphate catalyzed by amorpha-4,11-diene synthase (ADS), which has been cloned independently by several groups (Chang et al. 2000, Mercke et al. 2000, Wallaart et al. 2001). Amorpha-4,11-diene is then consecutively transformed into artemisinic alcohol, artemisinic aldehyde, dihydroartemisinic aldehyde and dihydroartemisinic acid by a sequence of enzymatic reactions (Bertea et al. 2005). Genes encoding trichome-specific enzymes that are believed to be involved in this sequence of reactions have also been cloned recently (Teoh et al. 2006, Zhang et al. 2008, Teoh 2009). Based on these data, a hypothetical pathway of artemisinin biosynthesis has been summarized in Supplementary Fig. S1 (Pu et al. 2009). Thus far, there have been no reports on the regulation of artemisinin biosynthesis at the transcriptional level.

Coordinate transcriptional control of biosynthetic genes has emerged as a major mechanism dictating the final levels of secondary metabolites in plant cells. This regulation of biosynthetic pathways is achieved by specific transcription factors (Vom Endt et al. 2002). Transcription factors have been isolated and characterized for several plant secondary metabolic pathways, including those leading to biosynthesis of flavonoids, terpenoid indole alkaloids, benzylisoquinoline alkaloids and sesquiterpene phytoalexins. Transcription factors can often activate several genes of the same pathway, e.g. overexpression of the MYB transcription factor PAP1 from Arabidopsis resulted in strongly enhanced expression of phenylpropanoid biosynthesis genes as well as enhanced accumulation of lignin, hydroxycinnamic acid esters and flavonoids, including various anthocyanins that are responsible for the purple color (Borevitz et al. 2000). In the plant species Catharanthus roseus, a jasmonate-responsive transcriptional regulator of primary and secondary metabolism, ORCA3, has been shown to regulate the jasmonate-responsive activation of several terpenoid indole alkaloid biosynthesis genes (van der Fits and Memelink 2000). A WRKY-type transcription factor, GaWRKY1, from Gossypium arboreum has been suggested to participate in regulation of sesquiterpene biosynthesis in cotton (Xu et al. 2004). Recently, a Coptis japonica transcription factor CjWRKY1 has been isolated, and transient expression of cDNA in C. japonica protoplasts was shown to increase the level of transcripts of benzylisoquinoline alkaloid berberine biosynthetic genes (Kato et al. 2007). Therefore, overexpression of transcription factor genes has been proposed as a promising approach to induce entire secondary metabolic pathways (Verpoorte and Memelink 2002, Petersen 2007).

The WRKY proteins are one of the major groups of plant-specific transcriptional factors. They are encoded by a multigene family, with at least 72 and 100 members in Arabidopsis (Dong et al. 2003) and rice (Oryza sativa) (Wu et al. 2005), respectively. WRKY transcriptional factors share different numbers of WRKY domains, a 60 amino acid stretch containing a conserved amino acid sequence WRKYGQK together with a zinc finger-like motif (Eulgem et al. 2000). Numerous in vitro and in vivo experiments have demonstrated that WRKY proteins specifically bind to the W-box (T)TGAC(C/T), a cis-acting DNA element found frequently in the promoter of defense-related genes (Rushton et al. 1996, Yu et al. 2001, Zheng et al. 2006). The WRKY transcriptional factors can either up- or down-regulate the expression of a given gene (Oh et al. 2008). The WRKY proteins have been implicated in a variety of developmental and physiological roles in plants. Most WRKY proteins are involved in defense and disease responses (Eulgem et al. 2000, Turck et al. 2004, Park et al. 2006, Zheng et al. 2007). WRKY genes can also regulate embryogenesis (Lagace and Matton 2004), seed coat and trichome development (Johnson et al. 2002), senescence (Hinderhofer and Zentgraf 2001, Robatzek and Somssich 2001, Miao et al. 2004) and the biosynthesis of secondary metabolites (Xu et al. 2004).

In A. annua, the glandular secretory trichomes (GSTs) are thought to be the site of biosynthesis and storage of artemisinin (Duke and Paul 1993, Duke et al. 1994, Li et al. 2006, Covello et al. 2007). In order to identify new genes of artemisinin biosynthesis, we generated expressed sequence tags (ESTs) using isolated GSTs of A. annua as starting material. In addition to ESTs representing the known genes involved in artemisinin biosynthesis, including FPS (farnesyl diphosphate synthase), ADS and CPR (cytochrome P450 reductase), we identified two ESTs among the 1,985 ESTs we sequenced that contained the consensus WRKY sequence. A full-length WRKY cDNA, designated AaWRKY1, was isolated and characterized.

In the present study, we report that two W-box elements are present in the ADS promoter. AaWRKY1 codes for a leucine zipper-containing WRKY protein, which interacted with the 3 × W-box elements derived from the ADS promoter in a yeast (Saccharomyces cerevisiae) one-hybrid system and in an electrophoretic mobility shift assay (EMSA). Overexpression of AaWRKY1 greatly activated the ADS promoters in stably transformed tobacco and in transiently transformed Nicotiana benthamiana and A. annua, and disruption of the W-box abolished this activation. Transient expression of AaWRKY1 cDNA in A. annua leaves clearly increased the level of transcripts of artemisinin biosynthetic genes. These data demonstrate that AaWRKY1 is a transcriptional activator of ADS, and participates in the regulation of artemisinin biosynthesis.

Results

Isolation of AaWRKY1 cDNA

To study the transcriptional regulation of ADS, a key enzyme in the biosynthesis pathway of artemisinin, the promoter of ADS (AY528931, cloned in our laboratory in 2004) was scanned with PLACE (Higo et al. 1999) (http://www.dna .affrc.go.jp/htdocs/PLACE/signalscan.html). The promoter was found to contain two reverse-oriented TTGACC W-box cis-acting elements (Fig. 1). Since, as mentioned above, W-box cis-acting elements are frequently found to be the binding sites of WRKY transcription factors (Eulgem et al. 2000), it is reasonable to assume that WRKY proteins are involved in the transcriptional regulation of the ADS gene.

Fig. 1

Analysis of the W-box elements of two ADS promoters. ADSpro1 and ADSpro2 represent the promoters derived from GenBank accession Nos. AY528931 and DQ448294, respectively.

The 10-celled biseriate GSTs were isolated by a method based on the mechanized abrasion of plant leaves, followed by filtration (Supplementary Fig. S2). A cDNA library of A. annua was then constructed using GSTs as the source of mRNA. In total, 1,985 randomly picked clones were sequenced. Two EST fragments that contained the consensus WRKY sequence and two EST fragments of the ADS gene were found in the cDNA library. The EST fragments that contained the consensus WRKY sequence were subjected to SMART-RACE-PCR. The full-length cDNA of AaWRKY1 (accession No. FJ390842) was found to be 1,478 bp long and contained a 936 bp open reading frame (ORF). The ORF encoded a protein of 311 amino acids (ACJ12926) with a calculated mol. wt. of 34.70 kDa and an isoelectric point of 5.31.

The AaWRKY1 protein contained one putative WRKY domain, together with one zinc finger-like motif in the C-terminal region (C-X7-CX23-H-X1-C) (Fig. 2). Based on the criteria described earlier (Eulgem et al. 2000), the AaWRKY1 protein appears to be a member of the group III WRKY family of plant transcription factors. Alignment of AaWRKY1 with some of the WRKY proteins that are involved in regulating plant secondary metabolism was performed (Fig. 2). AaWRKY1 exhibited 32.5% sequence identity with Arabidopsis AtWRKY53 protein, 13.2% sequence identity with C. japonica CjWRKY1 protein, and 15.9% identity with G. arboretum GaWRKY1 protein. However, other than for the conserved WRKY domain and the zinc finger-like motif, there was little sequence conservation among the WRKY proteins.

Fig. 2

Alignment of AaWRKY1 with some WRKY proteins. The deduced amino acid sequence of AaWRKY1 was aligned with Arabidopsis AtWRKY53 (NP_194112), Coptis japonica WRKY1 (accession No. BAF41990) and Gossypium arboretum WRKY1 (accession No. AAR98818) using Clustal W (Thompson et al. 1994) with default parameters through EMBnet (http://www.ch.embnet.org/software/ClustalW.html). Black and gray shadings, done with BOXSHADE 3.21 (http://www.ch.embnet.org/software/BOX_form.html), indicate conserved amino acid residues. Predicted domains are indicated above the sequences.

AaWRKY1 and ADS are highly expressed in GSTs and are induced by defense signals

To determine if AaWRKY1 expression coincides with ADS expression in A. annua, reverse transcription–PCR (RT–PCR) was carried out to analyze its expression patterns in roots, stems, leaves, flowers and GSTs. AaWRKY1 was highly expressed in GSTs and a moderate expression was observed in leaves and flowers. ADS showed exactly the same expression pattern (Fig. 3A). These results indicate that AaWRKY1 might target the ADS gene.

Fig. 3

The expression pattern of AaWRKY1 and ADS. (A) AaWRKY1 and ADS are highly expressed in glandular secretory trichomes (GSTs). Ethidium bromide-stained agarose gels showing RT–PCR results for roots (R), leaves (L), stems (S), flowers (F) and GSTs (T). (B) AaWRKY1 and ADS expression in response to MJ. (C) AaWRKY1 and ADS expression in response to chitosan. Leaves were sprayed with methyl jasmonate (MJ; 0.3 mM) or chitosan (150 mg l⁻¹). Leaves were harvested at the indicated times after treatment for total RNA preparation. Accumulation of AaWRKY1 and ADS transcripts was monitored by semi-quantitative RT–PCR. The Actin gene was used as an internal control.

Methyl jasmonate (MJ) and chitosan are proven effective elicitors of artemisinin biosynthesis in cell cultures and hairy root cultures of A. annua (Putalun et al. 2007, Baldi and Dixit 2008). Because ADS is a key enzyme of artemisinin biosynthesis, it is reasonable to assume that ADS transcription might be activated by these two elicitors. Also, if AaWRKY1 was a regulator of ADS, we would also expect an inducible expression of AaWRKY1. Thus, the expression of the ADS and the AaWRKY1 gene under MJ and chitosan treatments was investigated using semi-quantitative RT–PCR. Expression of AaWRKY1 was induced rapidly and robustly following treatment with 0.3 mM MJ or 150 mg l⁻¹ chitosan. In MJ-treated leaves, the transcript level of AaWRKY1 increased rapidly and peaked within 30 min; this activation persisted up to 1 h post-treatment and then declined to the original level by 2 h (Fig. 3B). ADS showed a slower rate of induction, with the transcript level increasing slightly from 0.5 to 2 h and reaching a maximum at 4 h. In chitosan-treated leaves, the transcript level of AaWRKY1 showed similar kinetics to those seen for MJ induction, whereas the level of ADS transcripts increased more rapidly compared with MJ induction, reaching its maximum at 1.5 and 2.0 h (Fig. 3C). Thus, the transcript level of AaWRKY1 increased rapidly and transiently upon elicitation, and preceded that of ADS.

DNA binding activity of AaWRKY1

As shown in Fig. 1, the promoter of the ADS gene cloned by our laboratory (AY528931, 2004) was designated ADSpro1, while the promoter of the ADS gene cloned by Kim et al. (2008) from a Korean A. annua (DQ448294) was designated ADSpro2. Both promoters contain two W-box elements, which are putative recognition sites for WRKY transcription factors. The position and sequence of Box W_A in the two promoters is well conserved, with the W-box cis-acting element (TTGACC) located in the region between −271 and −266 bp. Box W_B in the two promoters is also conserved, except that W_B in ADSpro1 is located in the region between −960 and −955 bp, while W_B in ADSpro2 is located in the region between −947 and −942 bp.

In order to test AaWRKY1 protein–W-box DNA interactions, an EMSA was performed. Production of recombinant AaWRKY1 protein in native conditions in a bacterial system (Escherichia coli) proved unsuccessful. Therefore, the protein was isolated in denatured form and dialyzed for refolding. The refolded protein band of AaWRKY1 was detected by SDS–PAGE (Supplementary Fig. S5). The DNA fragments containing the 3 × W-box and mutated 3 × W-box elements were triple tandem repeats of the W-box from the promoter of the ADS gene, and the complementary oligonucleotides could anneal to form double-stranded structures. Two shifted bands, W_A and W_B, were observed with the recombinant AaWRKY1 protein incubated with the 3 × W-box probe (Fig. 4A, C: lanes 2–6). No binding signal was detected with the mutated 3 × W-box elements as probes (Supplementary Fig. S6), and no binding signal was detected in the negative control, where the W_A-box × 3 was incubated with 0.5 μg of his-tagged AaIPPI (DQ666334) protein (Fig. 4A, lane 8) and the W_B-box × 3 was incubated with 2 μg of bovine serum albumin (BSA) (Fig. 4C, lane 8). Therefore, the observed probe retardation was assigned to the DNA binding activity of AaWRKY1. To verify AaWRKY1 binding capacity, the same amounts of W-box DNA were incubated with higher amounts of AaWRKY1: 1.5-, 2-, 2.5- and 3-fold that of lane 2 in lanes 3, 4, 5 and 6, respectively. The numbers of shifted protein–DNA complexes showed a tendency to increase gradually, and free DNA levels decreased accordingly. Thus, AaWRKY1 was able to recognize and interact with the W-box elements of the ADS promoter in vitro.

Fig. 4

The ability of the AaWRKY1 protein to bind to the W-box cis-acting elements, as analyzed by EMSA. (A) Titration of W_A-box × 3 DNA with AaWRKY1 protein stained with SYBR Green EMA for visualizing DNA. (B) The same gel as in A stained with SYPRO Ruby EMSA for visualizing protein. (C) Titration of W_B-box × 3 with AaWRKY1 protein stained with SYBR Green EMSA for visualizing DNA. (D) The same gel as in C stained with SYPRO Ruby EMSA for visualizing protein. Increasing amounts of AaWRKY1 protein were added to W_A-box × 3 DNA or W_B-box × 3 in a final 15 μl volume. M, DNA maker (DL2000). Lane 1, cis-acting element DNA only. Lanes 2–6, cis-acting element DNA with increasing amounts of AaWRKY1 protein (46, 69, 92, 115 and 138 ng). Lane 7, AaWRKY1 protein only (138 ng). Lane 8, cis-acting element DNA with 2 μg of BSA (A, B) and 0.5 μg of pET-AaIPPI (C, D), respectively. The arrows indicated the AaWRKY1–cis-acting DNA element complex.

In the EMSA, two shifted bands were observed with AaWRKY1 protein–3 × W-box DNA interactions (Fig. 4). It is possible that the large retardant band is an AaWRKY1 protein dimer, because AaWRKY1 in native-PAGE does not give a single band (Supplementary Fig. S5). Sequence analysis of the AaWRKY1 protein indicated that there was a leucine zipper-like motif (L_X6A_X6L_X6E_X6L_X6L) from amino acid position 16 to position 51 (Fig. 2); this motif may play a significant role in the formation of a homodimeric WRKY protein.

To determine if AaWRKY1 binds to the W-box of the ADS promoter in vivo, a yeast one-hybrid assay was performed. The W_A-box × 3 fragment was fused to a HIS reporter gene under the control of the GAL1 minimal promoter and introduced into yeast strain yWAM2. The ORF of AaWRKY1 was inserted into the pPC86 vector to generate pPC86-AaWRKY1. The pPC86-AaWRKY1 and the empty pPC86 vector were introduced into yeast strain yWAM2 containing a plasmid harboring a W_A-box × 3 fragment. On SD-Trp-Leu-His selective medium containing 15 mM 3-amino-1,2,4-triazole (3-AT), only the yeast clones harboring pPC86-AaWRKY1 could grow (Fig. 5). The result indicated that AaWRKY1 bound to the W_A-box and activated transcription in vivo.

Fig. 5

Yeast one-hybrid assay using the W_A-box × 3 as bait. Yeast cells carrying pPC86 or pPC86-AaWRKY1 were grown for 3 d at 30°C in SD/–his/–trp/–leu or SD/–his/–trp/–leu plus 15 mM 3-amino-1,2,4-triazole (3-AT), respectively.

AaWRKY1 activated ADSpro1 in transgenic tobacco

The transactivation of the ADS promoter by AaWRKY1 was confirmed in stably double-transformed tobacco plants harboring the reporter pCAMBIA-ADSpro1::GUS and the effector pBI-AaWRKY1 (Fig. 6A). Tobacco plants were first transformed with a binary vector pCAMBIA1301 harboring the ADSpro1::GUS reporter gene under the selective pressure of hygromycin, then the effector construct CaMV35S-AaWRKY1 was transformed into the primary transformant under the selective pressure of kanamycin and hygromycin. In the double transgenic tobacco plants, constitutive expression of AaWRKY1 activated the ADSpro1 promoter (Fig. 6B). Compared with the leaves of primary transformants, the GUS (β-glucuronidase) activities in leaves of secondary transformants were increased about 4- to 5-fold (Fig. 6C). These results demonstrated that AaWRKY1 strongly activated the ADSpro1 promoter in transgenic tobacco.

Fig. 6

Activation of the ADSpro1 promoter by co-expression of the effector construct CaMV35S-AaWRKY1 with the reporter construct ADSpro1::GUS in transgenic tobacco plants. (A) Schematic diagram of the reporter and effector constructs used in the co-expression experiment. (B) RT–RCR analysis of the GUS and AaWRKY1 transcripts in transgenic tobacco. Actin amplification was used as the constitutive control. R (CK) represents the transgenic tobacco line harboring the reporter construct as the control. The plasmid harboring the effector construct (E) was transformed into the transgenic tobaccos harboring the reporter construct. E-1:R, E-2:R, E-3:R, E-4:R and E-10:R represent five transgenic tobacco lines carrying the effector construct (E) and the reporter construct (R) at the same time. (C) Measurement of GUS activities in the different transgenic tobacco lines. GUS activity was normalized in every double transformant relative to the transgenic tobacco line harboring the reporter construct R (CK). Data were statistically analyzed by SPSS, and asterisks indicate values that are significantly different (**P < 0.01) from the reporter construct R (CK).

AaWRKY1 activated ADSpro2 in a transient expression system

Kim et al. (2008) reported another ADS promoter, ADSpro2 (DQ448294) from a Korean A. annua. In order to confirm that AaWRKY1 could also activate ADSpro2, we amplified the ADSpro2 promoter from our A. annua strain and generated three mutants: ADSpro2: mW_AADSpro2, mW_BADSpro2 and mW_ABADSpro2 (see Materials and Methods). We then tested the interactions of AaWRKY1 with the wild-type and mutant ADS promoters in an Agrobacterium-mediated transient expression system in N. benthamiana and A. annua leaves. When only the reporter plasmid was introduced into N. benthamiana, the GUS expression directed by the wild type and mutant ADSpro2 was similar (Fig. 7a, c, e, g). When both reporter and effector plasmids were introduced into N. benthamiana, the ADSpro2-directed GUS expression increased approximately 4-fold (Fig. 7b). Compared with the wild-type ADSpro2, mW_AADSpro2-. mW_BADSpro2- and mW_ABADSpro2-directed GUS expression decreased approximately 40–50, 60–70 and 100% (Fig. 7d, f, h), respectively. Similar results have also obtained in the transient expression system in A. annua (Fig. 8). These results indicated that AaWRKY1 protein transactivated the ADS promoter in a W-box-dependent way.

Fig. 7

Transactivation of ADSpro2::GUS gene expression by AaWRKY1 in agroinfiltrated Nicotiana benthamiana leaves. A transient assay of GUS activities was carried out in N. benthamiana leaves 2 d after infiltration with Agrobacterium tumefaciens cells harboring (a) ADSpro2::GUS; (b) ADSpro2::GUS/35S::AaWRKY1; (c) mW_B ADSpro2::GUS; (d) mW_B ADSpro2::GUS/35S:: AaWRKY1; (e) mW_AADSpro2::GUS; (f) mW_AADSpro2/35S:: AaWRKY1; (g) mW_ABADSpro2::GUS; and (h) mW_ABADSpro2::GUS/35S:: AaWRKY1. GUS activity was normalized in treatments (b–h) relative to that of N. benthamiana leaves agroinfiltrated with ADSpro2::GUS (a). Duncan multiple comparison tests were done, A, B, C and D indicate values that are significantly different (**P < 0.01) from each other.

Fig. 8

Transactivation of ADSpro2::GUS gene expression by AaWRKY1 in Artemesia annua leaves. A transient assay of GUS activities was carried out in A. annua leaves 3 d after co-cultivation with Agrobacterium tumefaciens cells harboring ADSpro2::GUS, mW_B ADSpro2::GUS, mW_AADSpro2::GUS or mW_ABADSpro2::GUS. GUS activity from mW_ABADSpro2::GUS was used for normalization. Duncan multiple comparison tests were done, A, B, C and D indicate values that are significantly different (**P < 0.01) from each other.

AaWRKY1 regulates the expression of the tobacco sesquiterpene synthase gene

When AaWRKY1 was ectopically expressed in tobacco, the transcript level of the 5-epi-aristolochene synthase gene (EAS4) was up-regulated, and the transcript level of EAS4 declined when anti-AaWRKY1 was expressed. The transcript level of the squalene synthase gene (SQS) was almost invariable (Fig. 9). Since the promoter region of EAS4 reportedly contains a W-box (Yin et al. 1997), AaWRKY1 might regulate EAS4 expression through interaction with the W-box in the gene promoter. The fact that transcription factors often have at least partially conserved functions between different species has been reported (Lloyd et al. 1992, Borevitz et al. 2000).

Fig. 9

AaWRKY1 regulates the expression of the tobacco 5-epi-aristolochene synthase gene (EAS4). (A and B) Expression levels of the transgene in the AaWRKY1-overexpressing tobacco line-4 (OE-4) and antisense AaWRKY1 line-1 (AS-1) were estimated by RT–PCR. Actin amplification was used as constitutive control. WT indicates the wild-type tobacco line. (C) RT–RCR analysis of the EAS4 and SQS transcripts in AaWRKY1-overexpressing line 4 (OE-4), antisense-1 line (AS-1) and the WT tobacco line. SQS, squalene synthase.

AaWRKY1 activated genes of the artemisinin biosynthetic pathway in A. annua

Since overexpression of AaWRKY1 greatly activated the ADS promoters in stably transformed Nicotiana tabacum and in transiently transformed N. benthamiana and A. annua, it was of interest to know whether AaWRKY1 would also activate other genes of the artemisinin biosynthetic pathway. Therefore, the pBI-AaWRKY1 construct and the empty vector pBI121 were introduced into leaves of A. annua by vacuum-assisted agroinfiltration (see Materials and Methods). The success of the transformation was confirmed by GUS staining (Supplementary Fig. S7). When the PBI-AaWRKY1 construct was introduced into A. annua leaves, transcript levels for AaWRKY1 were 5-fold higher than those of the control after 72 h of co-cultivation (Fig. 10A, B). Semi-quantitative RT–PCR results showed that the expression levels of known genes of the artemisinin biosynthesis pathway were significantly increased with the overexpression of AaWRKY1, except for that of FPS (Fig. 10A, B). The transcripts of HMGR, ADS, CYP71AV1 and DBR2 increased 5.18 ± 0.75-, 6.56 ± 0.65-, 3.01 ± 0.57- and 3.84 ± 0.80-fold, respectively (Fig. 10B). Thus, overexpression of AaWRKY1 could strongly activate the expression of the majority of artemisinin biosynthetic genes.

Fig. 10

AaWRKY1 activated the known genes involved in the artemisinin biosynthesis pathway. (A) RT–PCR analysis of gene expression levels in empty vector PBI121 (control) and PBI-AaWRKY1 transiently transformed A. annua. (B) Comparison of gene transcript levels in PBI121 (control) and PBI-AaWRKY1 transiently transformed Artemesia annua. The gene transcript levels of PBI121 transiently transformed A. annua were set to 1 and compared with the transcript levels in PBI-AaWRKY1 transiently transformed A. annua. Signals obtained from Actin were used for normalizations. Data were statistically analyzed by SPSS, and asterisks indicate values that are significantly different (**P < 0.01).

Discussion

Because of the tremendous medicinal significance of artemisinin, its biosynthesis and metabolic engineering are presently very attractive research topics worldwide. To date, several genes encoding enzymes of the artemisinin biosynthetic pathway have been cloned and characterized (Bouwmeester et al. 1999, Chang et al. 2000, Wallaart et al. 2001, Teoh et al. 2006, Zhang et al. 2008, Teoh 2009). However, the transcriptional regulation of this important pathway is still not yet understood. In the present study, we isolated and characterized the first transcription factor, AaWRKY1, from A. annua. The other EST containing the consensus WRKY sequence was also identified from the cDNA library and isolated by RACE (rapid amplification of cDNA ends), and was designated AaWRKY2. The AaWRKY2 protein contained one putative WRKY domain, together with one zinc finger-like motif in the C-terminal region (C-X5-C-X23-H-X1-H), which was clustered with the group II WRKY protein. AaWRKY1 was highly expressed in the GSTs on the leaves and flowers of A. annua (Fig. 3A). As has been reported for other genes related to artemisinin biosynthesis, such as ADS, CYP71AV1 and DBR2 (Teoh et al. 2006, Zhang et al. 2008), the expression of AaWRKY1 was fast, transient and preceded that of ADS upon induction by either MJ (Fig. 3B) or chitosan (Fig. 3C). AaWRKY1 showed a strong induction by MJ and chitosan; however, AaWRKY2 showed insignificant variation in response to the same treatments (Supplementary Fig. S4). Because AaWRKY1 and ADS in GSTs were found to have a similar expression pattern, this enabled us to suggest that AaWRKY1 is a transcription factor regulating the ADS promoter.

AaWRKY1 is a group III WRKY transcription factor, based on its sequence analysis. It contains a putative nuclear localization signal (NLS) sequence (₁₀₅KKRK₁₀₈) (Fig. 2) and, indeed, the targeting experiment showed that AaWRKY1 was localized to the nucleus (Supplementary Fig. S3). The results indicate that AaWRKY1 is a nuclear protein, which is consistent with its function as a transcriptional regulator. AaWRKY1 contains one WRKY domain with a WRKYGQK core sequence that is the major WRKY core sequence (Eulgem et al. 2000). The EMSA and yeast one-hybrid results indicated that AaWRKY1 interacts with the cis-acting elements (TTGACC) of the W-box in the promoter region of the ADS gene (Figs. 4, 5). Furthermore, the in vivo transactivation by AaWRKY1 of the ADS promoter was confirmed with stably transformed tobacco plants (Fig. 6). The activation of the ADS promoter by AaWRKY1 required functional W-box elements in the ADS promoter, as was also confirmed experimentally (Figs. 7, 8). These results strongly suggest that ADS is a target gene of AaWRKY1.

Transcription factors can often activate several genes of the same pathway. The results presented here demonstrated that AaWRKY1 has a similar propensity. Transient expression of AaWRKY1 activated the expression of HMGR, ADS, CYP71AV1 and DBR2 of the artemisinin biosynthesis pathway (Fig. 10, Supplementary Fig. S1). It is possible that the W-box also existed in the promoters of CYP71AV1, HMGR and DBR2 in A. annua. Indeed, the W-box has been found in the promoters of cytochrome P450 genes of many other plants such as Arabidopsis (Narusaka et al. 2004), C. japonica (Kato et al. 2007) and cotton (Xu et al. 2004), while two W-box elements were also found in the HMGR1 promoter of Camptotheca acuminate (Burnett et al. 1993). Isolation and characterization of the promoters of HMGR, CYP71AV1, DBR2 and other genes in the artemisinin biosynthetic pathway would be beneficial for further evaluation of the function of WRKY transcription factors in artemisinin metabolism. Generation of transgenic A. annua plants that overexpress or silence the AaWRKY1 gene is currently under way in our laboratory. Detailed analysis of these transgenic plants will provide a more complete picture of how AaWRKY1 regulates artemisinin biosynthesis, and will also address whether overexpression of the AaWRKY1 gene could be a new strategy for increasing artemisinin production in transgenic A. annua.

FPS is considered to be a regulatory enzyme in plant isoprenoid biosynthesis (Dudley et al. 1986, Hugueney et al. 1996, Ramos-Valdivia et al. 1997, Chen et al. 2000). In the present study, overexpression of AaWRKY1 did not increase FPS expression. In a salicylic acid-induced A. annua plant, FPS expression was also not induced, although artemisinin content increased (Pu et al. 2009). Overexpression of the FPS gene in A. annua has also been shown not to increase the content of artemisinin significantly (Han et al. 2006). These results suggest that FPS might not be a rate-limiting factor in artemisinin biosynthesis. This would be similar to the observation that overexpression of Arabidopsis FPS1S did not significantly increase the content of sterols (Cunillera et al. 2000, Masferrer et al. 2002).

ADS showed a slower rate of induction in MJ-treated leaves than in chitosan-treated leaves following AaWRKY1 expression (Fig. 3B, C). It is possible that the induction of the transcript of AaWRKY1 was stronger and longer (0.5–1.5 h) in chitosan-treated leaves than in MJ-treated leaves (0.5–1.0 h). It might also be that other transcription factors are involved in the regulation of ADS expression, because other cis-acting elements involved in salicylic acid responsiveness, defense and stress responsiveness, ethylene responsiveness, auxin responsiveness and gibberellin responsiveness are also present in the ADS promoter (Supplementary Table S1). More than one transcription factor can coordinately regulate the expression of pathway genes; for example, in maize, the majority of the structural genes encoding enzymes committed to anthocyanin biosynthesis are coordinately regulated by the bHLH protein-encoding gene R and the MYB gene C1 in the aleurone layer of maize kernels (Mol et al. 1998). The bHLH transcription factor (three ESTs), bZIP protein (two ESTs), zinc finger protein (two ESTs including one C₂H₂ type and one CCCH type), AP2 domain transcription factor (one EST), trihelix transcription factor (one EST) and MYC transcription factor (one EST) were all found in the cDNA library (data not shown). Characterization of these transcription factors will aid in finding activator proteins that regulate the artemisinin biosynthesis pathway. In turn, further elucidation of the regulatory machinery controlling artemisinin biosynthetic genes will allow greater advancement of artemisinin biotechnology.

Materials and Methods

Plant materials

The high artemisinin-yielding strain 001 of A. annua L. from Sichuan Province, China (Chen et al. 2000, Zhang et al., 2006, Wang et al. 2007) was used in this study. Artemisia annua seedlings were micropropagated as described by Zhang et al. (2006). The rooted in vitro plantlets were transported into pots and grown in a greenhouse with 16 h/25°C day and 8 h/20°C night. Nicotiana tabacum cv. Wisconsin 38 (W38) was used for transformation, and N. benthamiana was used for agroinfiltration.

Isolation of glandular secretory trichomes and cDNA library construction

Leaves from 4-month-old (pre-flowering) A. annua were used for the isolation of GSTs. The collected leaves were chilled in ice-cold water for 10 min, then abraded using a cell disrupter (Bead Beater, Biospec Products, Bartlesville, OK, USA). The extraction mixture contained 20–30 g of leaves, 80–100 g of glass beads (0.5 mm diameter), 20–30 g of XAD-4 resin and 100 ml of isolation buffer [25 mM MOPSO, pH 6.6, 200 mM sorbitol, 10 mM sucrose, 5 mM thiourea, 2 mM dithiothreitol (DTT), 5 mM MgCl₂, 0.5 mM sodium phosphate, 0.6% (w/v) methylcellulose and 1% (w/v) polyvinylpyrrolidone (M_r 40,000)]. GSTs were isolated as described by Teoh et al. (2006). Total RNA from GSTs was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany). The cDNA library was constructed with the Creator SMART cDNA Library Construction Kit (Clontech, Mountain View, CA, USA) following the protocols suggested by the manufacturer.

AaWRKY1 cDNA isolation

Randomly picked clones were sequenced with an ABI 3700 DNA sequencer using a BigDye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA). Sequences were assembled and edited using PHRED/PHRAP (Ewing and Green 1998), with a minimatch score of 35, and all contigs and singlets were imported into a gap4 database for viewing (Bonfield et al. 1995). The non-redundant unigenes (contigs and singlets) were subjected to similarity searches against the GenBank non-redundant (nr) protein databases using the BLASTX algorithm with default parameters. The EST fragment which contains the consensus WRKY sequence was identified and subjected to SMART-RACE-PCR (nested PCR) with first-strand cDNA as temple using a primer binding to the pDNR-LIB vector arm (5′-AAG CAG TGG TAT CAA CGC AGA GT-3′) (forward) and two gene-specific primers (reverse). The gene-specific primers for AaWRKY1 are GSP1 (5′-CGC TAC CCT CTA ACC CAT CAT CAG T-3′) and GSP2 (5′-ACT CAG ATC GTA CGC TTT GTG GGC T-3′).

Subcellular localization of AaWRKY1

The ORF of AaWRKY1 was amplified by a sticky-end PCR method (Zeng 1998) to introduce an Xbal sticky end (CTAG) upstream of the start codon and a SmaI blunt end (CCC) downstream of the ORF. The sticky-end PCR product (XbaI upstream and SmaI downstream) was ligated into pBI221 digested with XbaI and SmaI to generate the CaMV35S::AaWRKY1-GFP construct, which was verified by sequencing. The CaMV35S::GFP construct was used as a control. Onion inner epidermal cells were stretched out inside-up on MS medium, and bombarded by using the PDS-1000/He system (Bio-Rad, Hercules, CA, USA) at 1100 p.s.i. with DNA-coated gold particles. The transformed cells were cultured on MS medium at 25°C for 24 h, and then observed under a confocal microscope (Leica TCS SP2, Germany). Green fluorescent protein (GFP) was excited at 488 nm with an argon laser, and fluorescence was detected at 500–540 nm and imaged with a 40.0 × /1.25 HCX PL APO objective lens.

Heterologous expression and purification of AaWRKY1

The ORF of AaWRKY1 was amplified with forward primer 5′-GAT GGA TCC ATG GAA AGT GTT TGT GTT TAT-3′ (with the BamHI site underlined) and reverse primer 5′-GTA CTC GAG AAA TTT GAA ATC AAG GTC TAA-3′ (with the XhoI site underlined), and then fused to the BamHI–XhoI sites in the pET-30a (+) vector (Novagen, Madison, WI, USA). After sequence confirmation, the pET-AaWRKY1 plasmid was transformed into E. coli strain BL21 (DE3) competent cells for protein expression. Escherichia coli cells containing pET-AaWRKY1 were cultured in LB medium supplied with 50 mg l⁻¹ ampicillin at 37°C. When the OD₆₀₀ of the cultures reached 0.8, 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added and the cultures were incubated for another 4 h at 25°C. The cells were harvested by centrifugation and resuspended in 10 ml of 0.05 M sodium phosphate buffer (pH 7.5) containing 20 mM imidazole, and sonified on ice for 5 min at 50% pulses using a Ultrasonic Crasher (Ningbo Scientz Biotechnology Co. Ltd., China). The homogenate was centrifuged for 10 min at 10,000 × g and 4°C. Since most of the expressed AaWRKY1 proteins existed in the inclusion body, the pellet was resolved in 8 mol l⁻¹ urea, then the proteins were isolated as denatured forms through a pre-packed 5 ml Hi-trap Ni column (Pharmacia) following the manufacturer’s protocol. For refolding, the purified proteins were dialyzed in turn against 50 mmol/l phosphate-buffered saline buffers containing 6, 4, 3, 2, 1 and 0 mol l⁻¹ urea (pH 7.4) at 4°C, for at least 4 h for each urea concentration. After removing the precipitates by centrifugation, the dialyzed protein solution was lyophilized, and stored at −80°C before use.

Electrophoretic mobility shift assay

The DNA binding ability of AaWRKY1 was analyzed by EMSA. Two oligonucleotides were synthesized, including the sense strand sequence of W_A-box × 3 (5′-AAT GTT GAC CCA TTC ATC ACA ATG TTG ACC CAT TCA TCA CAA TGT TGA CCC ATT CAT CAC-3′) and the corresponding antisense strand sequence, the two oligonucleotides were incubated for 5 min at 95°C, and thereafter 5 min at 65°C to renature the products. The oligonucleotides of W_B-box × 3 (5′-AAT GGG TCA ATT GGG GAG TGA ATG GGT CAA TTG GGG AGT GAA TGG GTC AAT TGG GGA GTG-3′) and the corresponding antisense strand sequence were synthesized and annealed to form double-stranded DNA using the same method. The core sequence of the W_A-box TTGACC was mutated into TTGAAA and the core sequence of the W_B-box GGTCAA was mutated into AATCAA. The mutated W_A-box and W_B-box probes were formed by the same way.

The EMSA was performed with the Electrophoretic Mobility Shift Assay kit (Invitrogen, USA) following the manufacturer’s instruction. The DNA–protein binding reaction was performed by incubating double-stranded oligonucleotides with purified protein at room temperature for 30 min in a total volume of 15 μl. The reaction system contained 20 mmol Tris–HCl (pH 7.6), 30 mmol KCl, 0.2% (w/v) Tween-20, 1 mmol DTT and 10 mmol (NH₄)₂SO₄. The binding mixture was resolved on a 6% non-denaturing polyacrylamide gel in 0.5 × TBE buffer. The gel was first stained with SYBR Green EMSA stain for visualizing DNA, and photographed; the same gel was stained with SYPRO Ruby EMSA stain for monitoring protein, and photographed.

Binding assay in yeast one-hybrid system

All yeast transformation was carried out according to the transformation protocol of Gietz et al. (1995) using the yeast strain yWAM2 (MATαΔgal4Δgal80 URA3::GAL1-lacZlys2801^amberhis3-Δ200trp1-Δ63leu2ade2-101^ochreCYH2). The yeast reporter plasmid was constructed by inserting annealed complementary oligonucleotides containing the triple tandem copies of the W_A-box fragment from the ADS promoter (annealed to yield XbaI- and BamHI-compatible ends) into pRS315HIS. The sequence of the W_A-box × 3 fragment is 5′-CTT GAC CCA TTT TGA CCC ATT TTG ACC CAT TT-3′. Yeast cells transformed with pRS315HIS::W_A-box × 3 plasmid were grown on supplemented synthetic dextrose medium (SD) without leucine. The full-length AaWRKY1 fragment was reamplified from cDNA by PCR. The ORF was ligated into the pPC86 vector to generate pPC86-AaWRKY1 using the restriction sites EcoRI (forward) and SpeI (reverse). The pPC86-AaWRKY1 and the empty pPC86 vector were introduced into the yeast containing the pRS315HIS::W_A-box × 3 plasmid. The yeast strains transformed with the above constructs were grown on SD-Trp-Leu-His selective medium containing different concentrations of 3-AT at 30°C for 3 d.

Transactivation experiments in stably transformed tobacco and agroinfiltrated N. benthamiana

For the reporter construct, the ADSpro1 promoter (AY528931, cloned in our laboratory in 2004) was amplified from the A. annua genome by PCR with the primer pair ADSpro1F (5′-CGT ACT GCA GGC ATA AGA ACA TAC AAA GCA-3′) and ADSpro1R (5′-GCT ACC ATG GGA TTT TCA AAA CTT TGA ATA TAT G-3′). The 2,385 bp long ADSpro1 promoter was inserted into the pCAMBIA1301 vector, generating pCAMBIA-ADSpro1::GUS. For the effector construct, the coding region of AaWRKY1 was inserted into the reconstructed pBI121 vector with BamHI and XhoI restriction sites (PCR-based mutagenesis was applied to create an XhoI site immediately after the GUS TGA site) under the control of the cauliflower mosaic virus (CaMV) 35S promoter. The recombinant plasmid pCAMBIA-ADSpro1::GUS was introduced into the Agrobacterium tumefaciens EHA105 strain; this strain was then used to transform N. tabacum cv. Wisconsin 38 (W38). Transformants were identified through selection in MS medium containing 15 mg l⁻¹ hygromycin. The transgenic lines were confirmed by PCR using primers ADSpro1F and ADSpro1R with genomic DNA extracted from leaves of primary transformants as template. One positive primary transformant line was used as recipient for transformation of the effector gene. The effector construct pBI-AaWRKY1 was directly transformed into the transgenic GUS reporter plant, and double transformants were identified through screening for antibiotic resistance (kanamycin and hygromycin) and confirmed by PCR using AaWRKY1F and AaWRKY1R with genomic DNA extracted from leaves of secondary transformants as template.

For transient expression experiments, the 1,176 bp ADSpro2 promoter sequence (Kim et al. 2008) was isolated using primers ADSpro2F (5′-ACT GGT ACC CAT GGC TAA AAC CGA ATT CAA-3′) and ADSpro2R (5′-AGT CCA TGG CTT TTC ATA GAT GAG AAA CT-3′). The W_B-box (5′-AAT TT GGG GAG TG-3′) in ADSpro2 was mutated into 5′-AAT TT GGG GAG TG-3′ by the primer extension method with the primers mW_BboxF (5′-TTG GGA AAA ATT AAA TTA ATC AAT TGG GGA-3′) and mW_BboxR (5′-TTA ATT TAA TTT TTC CCA AGC CCA AGT CCA-3′); the resulting promoter was named mW_BADSpro1. The W_A-box (5′-AAT GCA TTC ATC AC-3′) in ADSpro2 was mutated into 5′-AAT GCA TTC ATC AC-3′ with the primer mW_AboxF (5′-TGT TTT TGA AAT GTT GAA ACA TTC ATC ACG-3′) and mW_AboxR (5′-TTT CAA CAT TTC AAA AAC ATA TTA GTC TTA-3′); the resulting promoter was named mW_AADSpro2. Double mutation mW_ABADSpro2 was carried out in the same way. ADSpro2, mW_BADSpro2, mW_AADSpro2 and mW_ABADSpro2 were inserted into the pCAMBIA 1301 vector, replacing the 35S promoter. The four above-mentioned ADSpro2::GUS, mW_BADSpro2::GUS, mW_AADSpro2::GUS, mW_ABADSpro2::GUS plasmids and 35S::AaWRKY1 were introduced into A. tumefaciens strain EHA105. In order to verify the effects of AaWRKY1 overexpression on W-boxes of the ADS promoter, a transient GUS activation was performed in tobacco leaves after infiltration with A. tumefaciens cells harboring (i) ADSpro2::GUS; (ii) ADSpro2::GUS/35S::AaWRKY1; (iii) mW_B ADSpro2::GUS; (iv) mW_B ADSpro2::GUS/35S::AaWRKY1; (v) mW_AADSpro2::GUS; (vi) mW_AADSpro2/35S::AaWRKY1; (vii) mW_ABADSpro2::GUS; and (viii) mW_ABADSpro2::GUS/ 35S::AaWRKY1. The ratio of 35S::AaWRKY1 to ADSpro2::GUS, mW_B ADSpro1::GUS, mW_A ADSpro2::GUS, mW_AB ADSpro2::GUS was adjusted to 1 : 1 for co-infiltration. Transient assay using N. benthamiana leaves was performed as described in the literature (Yang et al. 2000). Fully expanded leaves of 8-week-old N. benthamiana plants were used for agroinfiltration. After agroinfiltration, the N. benthamiana was covered with transparent plastic bags and maintained in a growth chamber at 22°C under 16 h light for 2 d. Experiments were done in triplicate; in each experiment six 12 mm discs from infiltrated areas of three independently infiltrated plants were transferred to Eppendorf tubes, frozen in liquid nitrogen and stored at −80°C.

Transient transformation of A. annua

Glycerol stocks of A. tumefaciens strain EHA105 containing 35S::GUS and A. tumefaciens strain EHA105 containing 35S::AaWRKY1 constructs were thawed and then streaked onto solid LB medium (pH 7.0, containing 100 mg l⁻¹ rifampicin and 100 mg l⁻¹ kanamycin), then individual single colonies were used to start 5 ml of liquid YEB medium (5 g l⁻¹ beef extract, 1 g l⁻¹ yeast extract, 5 g l⁻¹ peptone, 5 g l⁻¹ sucrose, 0.5 g l⁻¹ MgCl₂). These cultures were grown overnight with shaking at 28°C/200 r.p.m. The subsequent 100 ml cultures were started with 1 ml of the preceding 5 ml overnight culture and again incubated with shaking at 28°C/200 r.p.m. until OD₆₀₀ = 0.5. Bacterial cells were harvested by centrifugation at 6,000 × g for 5 min and washed once with 10 ml of washing solution containing 10 mM MgCl₂ and 100 μM acetosyringone. After centrifugation at 6,000 × g for another 5 min, the pellet of bacterial cells was resuspended in 50 ml of liquid MS medium.

Artemisia annua leaves from 4-week-old aseptic seedlings were cut and soaked with 50 ml of MS liquid medium containing A. tumefaciens cells supplemented with 0.005% (v/v; i.e. 50 μl l⁻¹) surfactant Silwet L-77 in 100 ml Erlenmenyer flasks, then the Erlenmenyer flasks were placed in the desiccator and a vacuum was applied at 30 inHg for 10 min under the control of a vacuum pump (Barnant Co., Barrington, IL, USA).The vacuum pump was then switched off and air was slowly let into the desiccator, so as to minimize the damage to the tissues. Liquid medium was removed and the explants were plated on MS solid medium; co-cultivation was in the dark at 25°C for 72 h.

The A. tumefaciens EHA105 strain containing either ADSpro2::GUS, mW_BADSpro2::GUS, mW_AADSpro2::GUS or mW_ABADSpro2::GUS was introduced into A. annua leaves by the above transient transformation method. After 72 h co-cultivation, the A. annua leaves were transferred to Eppendorf tubes, frozen in liquid nitrogen and stored at −80°C.

GUS activity assay

For stable transformation, three leaf discs of each transgenic plant were separately assayed for GUS activity, with each data point being the average of three transgenic plants. For transient GUS expression measurements in N. benthamiana, homogenates were made of six 12 mm discs from infiltrated areas of three independently infiltrated plants. For transient GUS expression measurements in A. annua, an equal amount of leaves was used. GUS activity was determined using an F-4500 Fluorescence spectrophotometer (Hitachi, Tokyo, Japan) as described in the literature (Jefferson et al., 1987) using 4-umbelliferyl-d-glucuronide as a substrate and 4-methylumbelliferone to calibrate the fluorometer. Protein content was determined according to the Bradford protein assay using BSA as a standard. GUS activity, normalized against protein concentration, was determined from the average of duplicate measurements per sample. Data were subjected to analysis of variance (ANOVA) and Duncan multiple comparison tests.

Gene expression analysis by semi-quantitative RT–PCR

For tissues expression analysis, total RNA was isolated from roots, leaves, stems, flowers and isolated trichomes of A. annua with an RNeasy Mini Kit (Qiagen). For elicitor treatments, 4-week-old plants were sprayed with 0.3 mM MJ or 150 mg l⁻¹ chitosan which was prepared according to a reported method (Young et al. 1982). Leaves from three plants were pooled for each time point, frozen in liquid nitrogen, and stored at −80°C before RNA extraction. Following co-cultivation, A. annua leaves were washed twice in aseptic distilled water and total RNA was isolated from explants with an RNeasy Mini Kit (Qiagen). First-strand cDNA synthesis was carried out with 1.0 μg of total RNA treated with DNase I (Roche) and Powerscript reverse transcriptase (Clontech). The resultant first-strand cDNA was used as a template for PCR. The conserved tobacoo Actin gene was used as an internal control.

The primers used were: actinF (5′-AAC TGG GAT GAC ATG GAG AAG ATA T-3′) and actinR (5′-TCA CAC TTC ATG ATG GAG TTG TAG G-3′) for Actin; AaWRKY1F (5′-ATG GAA AGT GTT TGT GTT TAT-3′) and AaWRKY1R (5′-AAA TTT GAA ATC AAG GTC TAA-3′) for AaWRKY1; ADSF (5′-ATG TCA CTT ACA GAA GAA AAA-3′) and ADSR (5′-TAT ACT CAT AGG ATA AAC GAG-3′) for ADS; HMGRF (5′-ATG ACT AGA GCG CCT GTT GTT-3′) and HMGRR (5′-GCA GAC ATC AAC GAC AAT TCA-3′) for HMGR; FPSF (5′-CTA TGT GGA TCT TGT GGA CCT-3′) and FPSR (5′-CCT GAA GAT TGA GAG TGT GGT-3′) for FPS; ADSF (5′-ATG TCA CTT ACA GAA GAA AAA-3′) and ADSR (5′-TAT ACT CAT AGG ATA AAC GAG-3′) for ADS; CYP71AV1F (5′-CGA AAG CAA TGG AGA AAG TAC-3′) and CYP71AV1R (5′-CTC AGT CTT TCT TTG CAT CGT-3′) for CYP71AV1; and DBR2F (5′-ATA AAT GAC AGA ACG GAC GAA-3′) and DBR2R (5′-GTC AAG AGA AGG GTA ATC CGT-3′) for DBR2. PCR products were electrophoresed on a 1% agarose gel and visualized after ethidium bromide staining. Pictures of gels were taken using a GelDoc EQ imager (Bio-Rad).

For transient gene expression analysis, band intensities were quantified using Quantity One software (Bio-Rad). The expression levels of genes involved in artemisinin biosynthesis were normalized using the Actin intensity as a loading control (set at 1.0). An average of three ethidium bromide-stained gels was determined and the standard deviation was calculated. Measurement data were expressed as mean ± SD. The analyses were performed using the statistical package for the Social Sciences (SPSS) statistical software for Windows, Version 9.0 (SPSS Inc., Chicago, IL, USA).

Funding

The National High-tech R&D Program (863) of the Ministry of Sciences and Technology, China (2007AA021501); the Knowledge Innovation Program of the Chinese Academy of Sciences (KSCX2-SW-329); the National Natural Science Foundation of China (30672623).

Acknowledgments

We thank Professor Xuejun Hua (Institute of Botany, the Chinese Academy of Sciences) for his help with the yeast one-hybrid assay, and Professor Rongxiang Fang (Institute of Microbiology, the Chinese Academy of Sciences) for his help with the subcellular localization assay.

Abbreviations

Abbreviations
• ACTs

  artemisinin-based combination therapies

• ADS

  amorpha-4,11-diene synthase

• 3-AT

  3-amino-1,2,4-triazole

• BSA

  bovine serum albumin

• CaMV

  cauliflower mosaic virus

• CPR

  cytochrome P450 reductase

• CYP71AV1

  cytochrome P450 enzyme

• DBR2

  artemisinic aldehyde Δ11(13) reductase

• DMAPP

  dimethylallyl diphosphate

• DTT

  dithiothreitol

• EMSA

  electrophoretic mobility shift assay

• EST

  expressed sequence tag

• FPP

  farnesyl diphosphate

• FPS

  farnesyl diphosphate synthase

• GFP

  green fluorescent protein

• GST

  glandular secretory trichome

• GUS

  β-glucuronidase

• HMGR

  3-hydroxy-3-methylglutaryl coenzyme A reductase

• IPP

  isopentenyl diphosphate

• MJ

  methyl jasmonate

• MOPSO

  3-(N-morpholino)-2-hydroxypropanesulfonic acid

• NLS

  nuclear localization signal

• ORF

  open reading frame

• RACE

  rapid amplification of cDNA ends

• RT–PCR

  reverse transcription–PCR.

The nucleotide sequence reported in this paper has been submitted to GenBank under the accession number FJ390842.