A. annua named as “Huhao 1,” originated from Chongqing, was developed in Shanghai after selection for several years. Plants were grown in the greenhouse with a 16/8 h light/dark photoperiod at 25°C.

T-shape trichomes were collected from the capitulum of A. annua with laser capture microdissection. The RNA from T-shape trichomes was extracted and sequenced (Soetaert et al., 2013). Arabidopsis ABC protein sequences were obtained from the Arabidopsis Information Resource (TAIR) database. A. annua putative ABC transporters were searched performing a BLASTP analysis against the transcriptome database using Arabidopsis ABC transporter protein sequences as queries with an “E” value over e⁻¹²⁰. Then the sequences of polypeptides corresponding to A. annua ABC transporters were analyzed in the Conserved Domain Database (CDD) at NCBI (Cakir and Kilickaya, 2013). The ABC transporters protein sequences from A. annua and PDR protein sequences from Arabidopsis were aligned with ClustalX. The phylogenetic tree was constructed by MEGA software (Tamura et al., 2011). Based on the RNAseq databases, we predicted the full-length AaPDR3 sequence. To obtain the open reading frame (ORF) of AaPDR3, the cDNA was synthesized with 0.5 μg total RNA isolated from leaves of A. annua, and the ORF was amplified using the gene-specific primers (Table S1). The phylogenetic tree analysis was performed with MEGA software version 5 via the neighbor-joining method based on amino acid sequence alignment, and the bootstrap analysis was performed using 1,000 replicates. Roots, stems, young leaves (the two youngest leaves), old leaves (from the 15th to the 16th leaf), buds and flowers of the A. annua plants were collected for RNA extraction using plant RNA isolation reagent (Tiangen, Beijing, China) following the manufacturer's instructions. The leaves from the Leaf0 (meristem), Leaf1, Leaf2, Leaf3, Leaf4, Leaf5, and Leaf6 counted from the apical top of the main stem were collected from 5-month-old A. annua. The total RNA was used to synthesize the first-strand cDNA. All the tissues and leaves collected from three plants were separately pooled for each determination. For hormone treatment, 2-month-old A. annua plants were treated with 100 μM MeJA (Sigma-Aldrich, USA), and then sampled at 0, 0.5, 1.5, 3, 6, 9, 12, 24 h, water with 1% concentration of DMSO as a mock treatment. The fifth leaves collected from three plants were separately pooled for each determination for RNA isolation. Real-time qPCR was carried out using the SuperReal PreMix Plus (SYBR Green) kit (Tiangen, Beijing, China) on lightcycle®96 (Roche, Mannheim, Germany). Three biological repeats were measured for each sample.

The full-length ORF of AaPDR3 was cloned into BamHI and XbaI sites of pHB-GFP vector. The recombinant plasmid was introduced into Agrobacterium tumefaciens strain GV3101 for A. tumefasciens-based Nicotiana benthamiana leaves transient expression (Voinnet et al., 2003). To confirm the localization of AaPDR3, we co-expressed the fusion protein GFP-AaPDR3 and the plasma membrane protein PIP1-mCherry in tobacco leaf epidermal cells. The GFP signal was observed after 2–3 days by Leica TCS SP5-II confocal laser microscopy (Leica, Wetzlar, Germany).

Genomic DNA was extracted from fresh young leaves of A. annua using the CTAB method. The upstream region 2,059 bp of AaPDR3 was obtained from the genome database of A. annua, amplified from genomic DNA with primers containing PstI and BamHI restriction sites and inserted into pCAMBIA1391Z vector. The resulting construct was transformed into A. annua plants, as described previously (Zhang et al., 2009).

The 346 bp fragment of AaPDR3 was amplified, cloned into gateway cloning vector pENTR vector using pENTR™/SD/D-TOPO® Cloning Kit (Invitrogen, Carlsbad, CA, USA), and then transferred to the destination vector pHELLSGATE12 via the LR recombination reaction (Invitrogen). The recombination plasmids (pHB-GFP-AaPDR3 and pHELLSGATE12-iAaDPR3) were introduced into A. tumefaciens strain EHA105 and transformed into A. annua plants, as described previously (Zhang et al., 2009).

The leaves were sampled from non-transgenic plants and transgenic plants for the histochemical GUS staining (Jefferson, 1987). The photographs were taken using an optical microscope (OLYMPUS, Japan). Two hundred milligrams of young leaves were powdered in liquid nitrogen, solubilized in the 2 volumes of buffer (100 mm Tris-HCl [pH 8], 50 mm KCl, 10 mm MgCl2, 20 mm DTT, and 2% Trixon-100) containing the protease inhibitors Cocktail and 1 mM phenylmethylsulfonylfluoride for 20 min on ice and centrifuged at 10,000 g for 10 min at 4°C twice. The supernatant was denatured by 2x sample buffer (125 mm TrisHCl [pH 6.8], 20% glycerol, 4% SDS, 200 mm DTT, and 0.05% bromophenol blue), incubated at 60°C for 15 min and clarified by centrifugation at 10,000 g for 1 min. The protein samples were separated on 8% SDS-PAGE gels and transferred onto nitrocellulose filters (0.45 μm pore size) (Millipore, USA). The membranes were blocked in 5% (w/v) non-fat milk powder for 2 h, and incubated with a 1:20,000 dilution of the primary antibody (Abmart, China) at 4°C overnight. The membranes were washed, incubated with a 1:10,000 dilution of goat anti-mouse alkaline phosphatase-conjugated secondary antibody (Sigma, USA), and detected using eECL Western Blot Kit (Kangwei Bio Inc., China).

The fresh samples were ground into fine powder in liquid nitrogen and freeze-dried for 72 h at −50°C. Fifty milligrams powder was suspended in 4 mL chromatographic-grade hexane in 10 mL glass tube with 100 μL trans-farnesol (77.6 μg/mL) as the internal standards, vigorously vortexed for 1 min and extracted for 40 min in an ultrasonic processor (JYD-650; Shanghai Zhisun Instrument Co. Ltd, China). The samples were centrifuged at 4,000 g for 10 min. The supernatants were filtered through 0.25-μm-pore-size filters, then concentrated and redissolved in 200 μL chloroform. GC-MS analysis was performed according to the methods described previously (Zhang et al., 2009). Three biological repeats were measured for each sample. Germacrene D was purchased from ChemFaces. B-caryophyllene and β-farnesene were purchased from Sigma-Aldrich.

The leaves of A. annua were collected, dried for 48 h at 50°C and ground into powder. One hundred milligrams of powder was extracted with 1 mL methanol for 30 min in an ultrasonic processor twice. The samples were centrifuged at 10,000 g for 10 min. The supernatants were filtered through 0.25-μm-pore-size filters and analyzed by the Waters Alliance 2695 HPLC system coupled with a Waters 2420 ELSD detector (Milford, USA) (Zhang et al., 2009). Three biological repeats were measured for each sample.

AaPDR3 was cloned into the SpelI and PstI sites of pDR196 by In-Fusion PCR cloning kits (Clontech, Palo Alto, CA, USA). The recombinant plasmid was transformed the strain AD1234567833 by the lithium acetate method. The yeast transformant was incubated in 50 mL SD medium (-uracil) at 29°C with shaking at 180 rpm, harvested at A600 = 1.0, and suspended by 50 mL half-strength SD medium (-uracil) containing β-caryophyllene, β-farnesene, and germacrene D, respectively. The cells were cultivated at 29°C with shaking at 180 rpm, harvested at the indicated times by centrifugation, washed twice with sterile water. The cells were disrupted with acid-washed glass beads in methanol for 15 min at 30 Hz (Yu and De Luca, 2013). Yeast cells were incubated in the culture media in the range of 0–1,200 μM β-caryophyllene for 1.5 h at pH 5.9. The cells harvested at the indicated times by centrifugation, washed twice with sterile water. Samples were centrifuged and filtered for GC-MS analysis.

Several studies have confirmed that many sesquiterpenes, with important biological functions, are produced in plant flower, leaf, secretory organ and root under constitutive, and stress conditions (Tholl, 2006). So we isolated T-shape trichomes from the capitulum of A. annua with laser capture microdissection and generated an RNA-Seq data based on RNA isolated from T-shape trichomes. Based on the T-shape trichomes transcriptome databases, we identified the 88 putative ABC transporters by performing a BLASTP analysis using Arabidopsis ABC transporter protein sequences as queries (Supplementary Information Data 1). We performed the phylogenetic analysis between PDR subfamily transporters found in Arabidopsis thaliana and the putative ABC transporters from A. annua. The result showed that four proteins were clustered with PDR transporters from Arabidopsis thaliana, and four PDR proteins (contig012562, contig001446, contig009129, and contig004541) were clustered with PDR transporters from Arabidopsis (Figure 1). Subsequently, we performed a phylogenetic tree analysis with the predicted amino acid sequences of four candidate PDR proteins and some PDR transporters containing Arabidopsis PDR transporters, NpPDR1, NtPDR1 and SpTUR2, showing that contig004541 protein sequence was similar to that of PDR proteins (AtPDR12, NpPDR1, NtPDR1, and SpTUR2) involved in terpene transport (Figure 2A). Therefore, this gene, named AaPDR3, was further examined as the candidate gene. AaPDR3, which is 4,278 bp in length, encodes a protein of 1,425 amino acids. This protein, belonging to the full-length size PDR subfamily, contains two nucleotide-binding domains (NBD) and two transmembrane domains (TMD; Figure 2B). Compare the conserved domain of known PDR transporters involved in terpene transport exhibited the high conservation in plants (Figure 2C). Besides, we analyzed the expression of AaPDR3 after the treatment with 100 μM MeJA, showing that MeJA induced the expression of AaPDR3 in A. annua (Figure S2).

Previous studies with CPS and BFS showed that the biosynthesis of related sesquiterpenes took place in roots, stems, leaves, and flower buds where they may play roles in defending the plant against fungal and worm attack (Lv et al., 2016). Consistent with these findings, investigation of AaPDR3 transcript level by RT-qPCR revealed that AaPDR3 expression level was the highest in T-shaped trichomes (Figure 3A). AaPDR3 is also determined in roots, stems, leaves, and flower buds (Figure 3A). Moreover, we analyzed the expression of AaPDR3 in leaves at different developmental stages. The expression level is the lowest in the youngest leaf (leaf0) and increased gradually with the leaves aging (Figure 3B).

Analysis of the encoded AaPDR3 protein by the subcellular prediction programs (Predotar: https://urgi.versailles.inra.fr/predotar/predotar.html; WoLF PSORT: http://www.genscript.com/psort/wolf_psort.html) predicted that this protein has no N-terminal signal peptide and is located to the plasma membrane. To examine the subcellular localization of AaPDR3 protein, the green fluorescent protein (GFP) fused to the N-terminal domain of AaPDR3 under CaMV35S promoter was transiently expressed in tobacco leaves. Results showed that GFP fluorescence of leaves expressing GFP-AaPDR3 was only observed in the plasma membrane (Figure 4A). The GFP fused to the N-terminal domain of AaPDR3 together with the established plasma membrane marker PIP1 (Siefritz et al., 2002) fused to mCherry were transiently co-expressed in tobacco leaves. The GFP-AaPDR3 green fluorescent signal was colocalized to the plasma membrane with PIP1-mCherry (Figure 4B). The results were consistent with those from prediction programs, indicating that AaPDR3 was localized in the plasma membrane and might function as a transporter.

To further investigate the tissue-specific expression pattern of AaPDR3 in A. annua, a 2,059-bp genomic fragment corresponding to the predicted AaPDR3 promoter sequence in our genome database was cloned from A. annua and then fused to GUS reporter gene. The recombinant plasmid was introduced into A. annua plants. GUS activity was analyzed in different tissues in A. annua. The result showed that GUS-staining was primarily restricted to T-shaped trichomes of old leaves in transgenic plants (Figures 5A–C). And GUS activity was also observed in roots in transgenic plants (Figure 5D), where a large number of sesquiterpenes are synthesized and stored.

To explore the function of AaPDR3 in A. annua, we generated 34 AaPDR3-RNAi transgenic plants using an RNAi strategy under the control of the CaMV35S promoter. In the RNAi transgenic plants, four independent lines with 14–34% observably downregulated AaPDR3 expression (Figure 6A) were selected for the detailed metabolic profiling analysis by gas chromatography-mass spectrometry (GC-MS) analysis (Figure S3). In contrast with the wild type, the suppression of AaPDR3 led to a 32–86% reduction of β-caryophyllene content (Figure 6B), while germacrene D and β-farnesene levels remained unchanged in AaPDR3-RNAi lines compared with wild type (Figure 6B). These data indicate that the repression of AaPDR3 markedly results in the suppression of sesquiterpene β-caryophyllene biosynthesis in A. annua.

AaPDR3 under the control of CaMV35S promoter was overexpressed in A. annua. We obtained 28 AaPDR3-overexpressing transgenic plants. Investigation of AaPDR3 transcript levels by qRT-PCR showed that the AaPDR3 expression was significantly increased in four AaPDR3-overexpression lines (Figure 6C). The four independent transgenic lines were identified by Western blot (Figure S4). Consistent with a role involved in sesquiterpenes biosynthesis transport in planta, the level of β-caryophyllene was increased to 0.48 mg/g FW in AaPDR3-overexpression lines compared to control (0.28 mg/g FW; Figure 6D). Little increases of β-farnesene and germacrene D were observed in AaPDR3-overexpression plants compared to wild type (Figure 6D). Taken together, AaPDR3 is involved in the sesquiterpene β-caryophyllene biosynthesis in A. annua. Moreover, the repression of AaPDR3 observably increased artemisinin contents in the RNAi plants (Figure S5).

A heterologous yeast expression system is an informative approach for elucidating the function of transporters (Morita et al., 2009; Shitan et al., 2013; Yu and De Luca, 2013). To investigate the function of AaPDR3 transporter, we expressed the AaPDR3 cDNA in the yeast strain AD12345678 lacking eight major ABC transporters and one transcription factor (Decottignies et al., 1998). Then we selected β-caryophyllene, β-farnesene, and germacrene D as the candidate substrates, respectively. The yeast cells of AaPDR3 transformant and the control (transformed with the empty vector PDR196) were incubated in half-strength Synthetic Dextrose (SD) medium contained 100 μM of each substrates, and the intracellular contents were quantitatively analyzed by LC-MS. Yeast cells expressing AaPDR3 accumulated more β-caryophyllene than the control along the same time course (Figure 7). The AaPDR3 transformants accumulated >44 nmol of β-caryophyllene per gram of cells compared with the control cells the contained almost 27 nmol at 9 h treated by 100 μM β-caryophyllene (Figure 7). The result demonstrated that expression of AaPDR3 increased β-caryophyllene influx. Both AaPDR3 transformants and the control were incubated in the culture media in the range of 0–1,200 μM β-caryophyllene. β-caryophyllene uptake by AaPDR3 followed Michaelis-Menten kinetics with K_m of 63.47 ± 8.81 pmol β-caryophyllene and a maximum transport rate V_max of 80.89 ± 2.46 pmol/g fresh yeast cells/min (Figure S6). No significant differences in the β-farnesene contents accumulated in AaPDR3 expressing yeast cells compared to that in the control group, as well as germacrene D (Figure S7). These results indicated that AaPDR3 was highly specific for the β-caryophyllene transport compared with β-farnesene and germacrene D in yeast.

The terpenoids are one of the largest groups of plant secondary metabolites (Croteau et al., 2000). Interestingly, the terpenoids are often transported from the cells where these compounds or metabolic intermediates are synthesized to neighboring cells, or even to other tissues or remote organs to be stored. Therefore, many transporter proteins participate in this biology process (Yazaki, 2005). Recently, the ATP-binding cassette (ABC) transporters have been reported to largely contribute to membrane transport of terpenoids in plants, especially for PDR subfamily (Jasiński et al., 2001; Van Den Brûle et al., 2002; Crouzet et al., 2013). Most of identified PDR transporters were expressed in specific tissues. For example, AtPDR8 is predominately expressed in roots and leaves. AtPDR2, AtPDR6, AtPDR9, and AtPDR13 are exclusively expressed in roots but not in shoots in Arabidopsis, while AtPDR14 is expressed only in shoots (Den Brule and Smart, 2002). The expression of AtPDR5 is mainly detected in roots and stems (Bienert et al., 2012). NtABCG5/PDR5, from N. tabacum, was highly expressed in the petals, stem and roots (Bienert et al., 2012). NtPDR1 was detected in stem and leaf tissues (Crouzet et al., 2013). These findings show that PDR genes are predominately expressed in roots and leaves. In plants, roots and leaves are the important tissues connected with environment. In this study, we characterized a PDR transporter AaPDR3 in A. annua. In our investigation, AaPDR3 was mainly active in old leaves, flowers, buds, and roots (Figure 3A). Notably, the GUS analysis exhibited that AaPDR3 was specifically expressed in T-shaped trichomes of old leaves and roots (Figure 5C). Likewise, AaPDR3 exhibited the tissues-specific expression pattern, suggesting that AaPDR3 plays an important role in the defensive compounds biosynthesis in T-shaped trichomes, flowers, buds, and roots. Moreover, the transcript level of CPS was also detected in leaves, flowers, buds, and roots (Lv et al., 2016), which is in accord with that of AaPDR3 in A. annua. Besides, the expression level of AaPDR3 was barely detected in the youngest leaf (leaf0), and increased gradually with the leaves aging (Figure 3B). Although the expression of CPS was highest in youngest leaf (leaf0), the CPS transcript level was also detected with the leaves aging (Lv et al., 2016). It means that β-caryophyllene is synthesized in young leaves and old leaves. From these results, we propose that AaPDR3 as a plasma membrane β-caryophyllene cellular uptake for gathering β-caryophyllene. Then the gathered β-caryophyllene is stored in the some cells of T-shape trichomes to reduce the cell damaged.

To identify the function of AaPDR3 in A. annua, we used RNAi to knock down the expression of AaPDR3. The repression of AaPDR3 resulted in an 86% reduction of β-caryophyllene content in AaPDR3-RNAi-20 transgenic A. annua line (Figure 6B), suggesting that AaPDR3 is essential for β-caryophyllene biosynthesis in A. annua. Transporters are the integral parts in metabolic networks, because they mediate multiple metabolic pathways. We speculated that the AaPDR3 repression would result in β-caryophyllene accumulated in the cells of T-shape trichomes in A. annua, which would prevent the β-caryophyllene biosynthesis. Our results, together with previous findings, indicated that AaPDR3 is involved in β-caryophyllene transport and plays an indispensable role in β-caryophyllene biosynthesis. AaPDR3 transporter reported here is the first transporter related to sesquiterpenes in A. annua, even in family Asteraceae.

Plant ABC transporters is a large and diverse group of proteins involved in the pathogen response, lipid deposition, and the transport of the phytohormones (Kretzschmar et al., 2011). Therefore, ABC transporters play an important part in plant growth, nutrition, development, and the interaction with the environment (Bird et al., 2007; Kuromori et al., 2010; Ding et al., 2011). Our results preferentially suggest that AaPDR3 is likely to be involved in the sesquiterpene β-caryophyllene transport based on four findings: (i) like other sesquiterpene transporters; the amino acid sequence of AaPDR3 is similar to that of PDR transporters involved in terpene transport (Figure 2A), (ii) the plasma membrane protein AaPDR3 is expressed in the tissues, including the T-shaped trichomes, buds, flowers, and roots, where the sesquiterpenes are synthesized (Figure 3A), (iii) increase and decrease in the AaPDR3 transcript level influence the sesquiterpene β-caryophyllene biosynthesis (Figure 6), and (iv) when AaPDR3 was expressed in yeast mutant AD1-8, yeast expressing AaPDR3 only took up β-caryophyllene faster than controls containing the empty vector (Figure 7). In fact, some ABC transporters are reported to have broad substrate specificity (Kolaczkowski et al., 1998). For example, PDR5 transporter from yeast was confirmed to export some compounds which had different structure and function (Wolfger et al., 2001; Lamping et al., 2010). In Arabidopsis, AtPDR12 is an ABA-uptake transporter in the guard cells and other cells (Kang et al., 2010). The plasma membrane transporter, AtPDR12, also contributes to the resistance of lead (Lee et al., 2005). When the yeast cells expressing AaPDR3 was incubated in SD medium contained 100 μM β-caryophyllene, β-farnesene, and germacrene D, respectively, our results showed that AaPDR3 exhibited narrow substrate specificity (Figure 7 and Figure S6).

Amazingly, we found that knockdown of AaPDR3 resulted in an increase in artemisinin content in AaPDR3-RNAi transgenic plants (Figure S5). AaPDR3 is a specific-expressed transporter gene in T-shape trichomes (Figure 4C), whereas artemisinin is specially synthesized in glandular trichomes. As we known, blocking the competitive pathways of artemisinin biosynthesis is very useful to improve the artemisinin content (Zhang et al., 2009; Lv et al., 2016). Both the artemisinin and dihydroartemisinic acid contents were increased, when CPS was suppressed by anti-sense in A. annua (Lv et al., 2016). The β-caryophyllene content in A. annua was up to 5–10% of the total essential oil (Brown, 2010). When AaPDR3 was down-regulated by RNAi in A. annua, the β-caryophyllene content was observably reduced in transgenic plants (Figure 6B), which might lead to the carbon altered through FPP to artemisinin biosynthetic pathway.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.