Seeds of A. annua were kindly provided by Dr. Xiao-Ya Chen at Shanghai Institutes for Biological Sciences. Seeds were surface-sterilized and germinated in flasks on 0.3% (w/v) Gelrite (Duchefa Biochemie B.V., Haarlem, The Netherlands) with 1/2 Murashige and Skoog (MS) medium (Murashige and Skoog, 1962). Seedlings that were approximately 10 cm in height were moved into the soil. Seeds of A. thaliana Col-0, Ler-0 and hy5-1 (Ler-0 background) lines from ABRC (Arabidopsis Biological Resource Center) were stratified at 4°C in dark for 2 days and then grown in pots containing a mixture of peat and vermiculite (3:1, v/v). Growth conditions for both A. annua and A. thaliana were 12h/12h light/dark photoperiod at 22°C, with a light intensity of 100 μmol m⁻² s⁻¹.

Genomic DNA was extracted from A. thaliana and A. annua following the standard CTAB protocol (Porebski et al., 1997). Total RNA was isolated from leaves using TRIzol reagent (Invitrogen) following the manufacturer's instructions. One microgram of total RNA was reverse-transcribed using the PrimeScript 1st Strand cDNA Synthesis Kit (TaKaRa) with oligo dT primer in a 20-μL system following the manufacturer's instructions. cDNA was stored at −80°C till further use.

The full-length cDNA sequence of QH6 (GenBank Accession No. AF276072.1) (Lu et al., 2002) was aligned with both genomic and cDNA sequences of a myrcene/(E)- β-ocimene synthase gene of A. thaliana (At4g16740) (Bohlmann et al., 2000) to estimate the positions of the exons and introns (Supplementary Figure S1). Because there was no information on the gene size of QH6, to increase the accuracy and efficiency of amplification, we designed primers according to the alignment to amplify QH6 genomic DNA in three overlapping fragments (primers and sequences are listed in Supplementary Table S1), each of which covers two predicted introns. Sequences of three amplified fragments were assembled. Approximately 0.1 μg genomic DNA of A. annua was amplified again with primers targeting two ends of the cDNA sequence (H-F and E-R) to confirm the assembly. Polymerase chain reaction (PCR) amplifications in this research were all conducted using LA Taq (TaKaRa) following the manufacturer's instructions, unless specifically noted. All amplicons were purified after agarose gel electrophoresis and subcloned into pGEM-T (Promega) to transform Escherichia coli DH5α and were sequenced for confirmation following standard molecular operation protocols (Sambrook et al., 1989). Sequences of all oligonucleotides used in this study are listed in Supplementary Table S1.

The upstream flanking sequence of QH6 was cloned using the genome walking strategy, following the instruction of a GenomeWalker Universal Kit (Clontech). Approximately 160 μg genomic DNA of A. annua was digested in 100 μL with 80 units of DraI, EcoRV, PvuII, and StuI (Promega) for 15 h and then ligated to adaptors at 16°C overnight. Three gene-specific reverse primers (GSP1-R, GSP2-R, and GSP3-R, Figure 1) were designed from the first extron of QH6 to amplify the upstream fragment with adaptor primers AP1 and AP2 by nested PCR.

Approximately 20 μg genomic DNA of A. annua was digested in a 50 μL system with ClaI, HindIII, and ApaLI overnight, respectively, resolved in a 1.2% agarose gel and blotted onto a Hybond N+ membrane (GE Healthcare). A 442-bp gene-specific fragment was amplified by primers Probe-F and Probe-R (Supplementary Table S1) and purified as a template for probe preparation. AlkPhos Direct Labeling Reagents and a CDP-Star™ Kit (GE Healthcare) were used for labeling and detection following the manufacturer's instructions. Membranes with probes were exposed to Kodak X-Omat Blue film (Perkin Elmer, Waltham, MA) for 48 h. The film was developed using standard methods.

We searched the region upstream from the QH6 translation initiation site for known cis-elements in plants using PLACE (A Database of Plant Cis-acting Regulatory DNA Elements) (Higo et al., 1999) and NSITE-PL (RegSite Plant DB, Softberry Inc., Mount Kisco, NY) databases.

Different truncations of the QH6 promoter region were amplified from the cloned upstream fragment using a common reverse primer, QH6-R, of which two nucleotides were mutated to generate an NcoI site for subsequent gene expression, and different forward primers with a PstI restriction site. All amplified promoter fragments were purified after gel electrophoresis, subcloned into the pGEM-T vector, and confirmed by sequencing.

For expressing GUS in A. thaliana, two different truncations immediately before and after G-box of QH6 promoter were amplified using forward primers QH6+GB-F and QH6-GB-F, respectively, with QH6-R. Each of the forward primers contain a PstI site at the 5′ end for further cloning work. Both of the promoter truncations were digested by NcoI and PstI and ligated to the same digested pCAMBIA1391Z vector (CAMBIA) to drive GUS expression. These constructs were named GB+ (with promoter beginning at the predicted G-box), and GB− (with promoter beginning immediately after the predicted G-box) (Figure 2).

For expressing luciferase gene (LUC), the 1400 bp fragment upstream translation initiation site was amplified using QH6-1400-F, which also contains a PstI site at the 5′-end, and QH6-R. To mutate its G-box core from CACGTG to ACGATC, a megaprimer strategy was adopted (Ke and Madison, 1997). In brief, the forward primer QH6-1400-MF was used with QH6-R to amplify the first short fragment and to introduce the mutation. This fragment was purified and used as a megaprimer with 1400-F to generate the 1400 bp promoter fragment with the G-box changed. Full open reading frame of luciferase gene (LUC) was digested from pSPLuc+NF (Promega) and then cloned between NcoI and HpaI sites in pCAMBIA1390 (CAMBIA) to give pCAMBIA1390-LUC. 1400 (1400 bp full length promoter of QH6) and 1400M (mutate G-box of 1400) were fused between PstI and NcoI in pCAMBIA1390-LUC to give 1400::LUC and 1400M::LUC, respectively.

Constructs were transformed into Agrobacterium tumefaciens GV3130 by electroporation with an Eppendorf 2510 electroporator (Yi and Kao, 2008). A positive colony for each construct was confirmed by PCR and inoculated for A. thaliana transformation by floral dipping according to Clough and Bent (Clough and Bent, 1998).

T0 seeds were germinated on 0.3% Gelrite plates with 1/2 MS medium (Murashige and Skoog, 1962) and 15 μg/mL hygromycin for screening. Seedlings growing well were moved into pots containing a mixture of peat and vermiculite (3:1, v/v) after 2 weeks. Genomic DNA was extracted from mature leaves to study the copy number and T-DNA insertion site by genome walking. Ten independent homozygote T2 lines were screened by PCR according to the T-DNA Primer Design Tool (http://signal.salk.edu/tdnaprimers.2.html) and the previous genome walking result (data now shown).

For A. thaliana plants transformed with different promoter truncations, mature leaves from homozygous transgenic plants with approximately 12 rosette leaves were sampled every 2 h for two continuous days and were frozen in liquid nitrogen immediately. RNA was isolated from a bulk of leaves from at least four plants to minimize individual variation, and 1 μg was reverse-transcribed as described above. To quantify gene expression, a 25-μL reaction system contained 12.5 μL SYBR Premix Ex Taq II (TaKaRa), 1 μL of each forward and reverse primers, 0.4 μL cDNA template and 9.1 μL RNase-free H₂O. A Thermal Cycler Dice Real Time System TP800 (TaKaRa) was used and the reaction program was 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s.

The second exon of GUS was amplified by primers GUS-Q-F and GUS-Q-R for real-time quantification. Actin8 (Hewezi et al., 2010) was amplified by primers ACT8-Q-F and ACT8-Q-R as a reference for normalizing GUS expression.

For each sample, PCR assays were carried out in triplicate. Melting curve analysis was used to confirm the specificity of the method. The results were analyzed using a TP800 Thermal Cycler Dice Real Time System (Takara). The threshold cycle (CT) values were automatically determined by the TP800 software (Takara). Expression values were calculated according to the 2^−ΔCT method (for each gene, ΔC_T = C_T,Gene − C_T,actin) (Livak and Schmittgen, 2001).

For luciferase activity assay, the luminescent signal was detected and 4 min exposure images were acquired by IVIS Lumina XR Imaging system (Perkin Elmer). 2.5 mmol/L luciferin (with 0.01% Triton X-100) was sprayed 1 h before detection (Millar et al., 1992). The radiance was calculated by Living Image Software and ROI (Region of Interest) tool.

In all treatments, the time when lights were switched on (dawn) was defined as Zeitgeber time 0 (ZT0). Prism (GraphPad Software) was used to perform the statistic analysis. Results were presented as mean ± SD. Transcript abundances were expressed as relative values to corresponding ZT0 levels of each line. An unpaired Student's t-test was conducted to examine the differences between the groups.

The fragment from −258 to −425 bp upstream of the QH6 translation initiation site was amplified by PCR with primers +GB-F and GB-R and was subcloned into the EcoRI/SpeI sites of pHIS2.1 (Clontech) as +GB. G-box -deleted (-GB) and -mutated (GBM) fragments were amplified using reverse primer GBR and forward primer -GB-F or GBM-F, respectively, and were fused to pHIS2.1. A fragment containing four tandem repeats of G-box (GACACGTGGC) was synthesized by annealing 4GB-F and 4GB-R primers (Supplementary Table S1) and cloned to pHIS2.1 as 4GB. The HY5-AD fusion vector was constructed by amplifying the HY5 (At5g11260) open reading frame from Arabidopsis cDNA with primers HY5-F and HY5-R and introducing the purified PCR product to the SmaI site of pGADT7-Rec2 (Clontech). The plasmids were transformed into Saccharomyces cerevisiae Y187 and HY5/+GB, HY5/-GB, HY5/GBM and HY5/4GB double transformants were selected on -Trp/-Leu double drop-out plates. A binding ability assay was performed by dotting 5 μL liquid culture on -Trp/-Leu/-His triple drop-out plates. The liquid culture was series-diluted while the initial concentration of OD₆₀₀ = 0.1 was taken as 0.1.

For EMSA, the HY5 open reading frame was amplified with primers HY5EX-F and HY5EX-R from A. thaliana cDNA and cloned into pGEX-4T-1 (GE Healthcare) between EcoRI/XhoI sites. The expression construct was transformed into E. coli BL21(DE3)pLysS (Invitrogen) for prokaryotic expression. The GST-HY5 fusion protein was purified using a MagneGST protein purification system (Promega). The biotin labeled 52-bp probes of the QH6 promoter containing G-box (391-442) and the mutant (M391-442) were synthesized by GenScript (Nanjing). An unlabeled probe was used as a competitor. The GST-only protein was purified and was used as a negative control. The interaction of the HY5 protein and DNA fragment was determined using a LightShift Chemiluninescent EMSA kit (Pierce) following the manufacturer's instruction. The 20-μL binding system was resolved on an 8% polyacrylamide gel in 1 × TBE and was semi-dry blotted onto Hybond-N+ membrane (GE Healthcare). After a three-hour cross-linking at 60°C, the signal was detected according to the manufacturer's instruction with Kodak X-Omat Blue film (Perkin Elmer, Waltham, MA).

According to the alignment of the A. annua QH6 cDNA sequence (Lu et al., 2002) with A. thaliana ocimene/myrcene synthase (At4g16740) cDNA and genomic DNA sequences (Chen et al., 2003), approximate positions for introns and exons of QH6 were predicted (Supplementary Figure S1). Three primer pairs were designed to amplify QH6 genomic DNA into three overlapping fragments, each of which covered two introns according to the prediction. The amplified fragments were assembled and re-confirmed by amplifying A. annua genomic DNA with primers H-F and E-R from two ends of the QH6 coding region. The QH6 gene has seven exons and six introns with the highly conserved tandem arginine domain (RR) in the first exon and the aspartate-rich divalent binding domain (DDXXD) in the fourth exon, as a typical Class III terpene synthase (Trapp and Croteau, 2001; Aubourg et al., 2002; Lange and Ghassemian, 2003). There was no hit with obvious similarity when we searched GenBank with sequence of any of introns (data not shown). The gene structure and positions of conserved domains and designed primers are shown in Figure 1.

The sequence upstream of QH6 was cloned by genome walking, and the first 2-kb fragment flanking the QH6 translation initiation site was confirmed by re-amplification from A. annua genomic DNA directly. By searching databases of PLACE (A Database of Plant Cis-acting Regulatory DNA Elements) (Higo et al., 1999) and NSITE-PL (RegSite Plant DB, Softberry Inc., Mount Kisco, NY), an HY5-binding G-box motif (CACGTGGCA) was found at −421 bp (Figure 2). Other motifs, such as SORLIP1 (GCCAC, −320), LEAFY response element (CCAATG, −545), W-box (TTGAC core, −425), and Morning element (ME, CCACAC, −537) were also found (Priest et al., 2009) (Figure 2). The entire A. annua QH6 sequence, including the upstream promoter region, was deposited in GenBank under the accession number GU929215.

When QH6 was originally isolated and characterized, there was no information suggesting that it was a single copy gene (Lu et al., 2002). This raised the possibility that other genes with a high sequence identity to QH6 might be analyzed, either instead of or together with QH6 itself. Here, by Southern blotting, a single band was detected from ClaI-, HindIII-, and ApaLI-digested genome DNA samples (2471, 2269, 1267 bp), confirming that the A. annua genome has only one copy of QH6 and providing a rationale for our subsequent studies (data not shown).

To figure out whether the G-box functions in regulating QH6 expression, we expressed β-glucuronidase gene (GUS) driven by the QH6 upstream fragment immediately before (GB+) or after G-box (GB−), respectively, in A. thaliana. For each construct, 10 independent transgenic lines were screened and genome walking was adopted to study the insertion site of each line to ensure that T-DNA insertion did not interrupt any gene known or presumed to affect either GUS expression or terpenoid metabolism (data not shown).

For both GB+ and GB− plants, rosette leaves were sampled every 2 h for 1 day in 12h/12h light/dark (LD) condition (Figure 3, from ZT0 to ZT24) and then for 24 h in constant dark (DD) (Figure 3, from ZT24 to ZT48). In GB+ plants in LD, two peaks of the steady state GUS transcripts were detected at ZT08 and ZT16, corresponding to the afternoon and midnight in nature. In DD, the fluctuation of GUS transcript abundance also showed two peaks at ZT28 and ZT36, both were 4 h ahead their LD counterparts. The ratios of transcript abundances at the highest level to the lowest were 5.39 in LD and 4.42 in DD, showing a significant fluctuation. For GB− plants, we could not identify a clear rhythmic change as the fluctuation of transcript abundance looked more frequently and the highest-to-lowest ratios were only 2.77 and 2.35 in LD and DD, respectively (Figure 3).

For better understanding the function of this G-box, we transformed Arabidopsis with luciferase gene (LUC) driven by an 1400 bp promoter fragment of QH6 with its G-box either intact or mutated from CACGTG to ACGATC to abolish its possible function. Transformed lines were screened and confirmed by genome walking as mentioned above. We monitored the luciferase activity by measuring radiance emitted with an IVIS Lumina XR Imaging System (Caliper Life Sciences) every 2 h under LD condition for 2 days. For control lines transformed with intact G-box (1400::LUC), we observed two peaks at ZT02 and ZT12 in Day 1, and another two peaks at ZT26 and ZT38 in Day 2. However, for the mutant lines (1400M::LUC), the ZT02 and ZT26 peaks no longer existed whereas the ZT12 and ZT38 peaks stayed as the control (Figure 4).

We then transformed Ler-0 and hy5-1 with 1400::LUC to study the expression of LUC when HY5 was silenced. The transcript abundance of LUC showed a synchronized accumulation in both lines under normal light/dark photoperiod (Figure 5). However, when entrained plants were placed into constant dark, only in 1400::LUC-transformed Ler-0 plants could a rhythmic fluctuation in LUC expression level be identified (Figure 5). The abundance of LUC transcripts in the transformed hy5-1 mutant decreased to a much lower level, and its fluctuation was not as significant as what we observed in Ler-0 (Figure 5).

To test whether the QH6 promoter binds HY5, a fragment from −258 to −425 bp to the QH6 translation initiation site was amplified by polymerase chain reaction (PCR) as a bait (+GB) for binding assays by yeast one-hybrid. The interaction between HY5 and this promoter fragment was screened by growth on -Trp/-Leu/-His triple dropout plates. Different concentrations of 3-aminotriazole (3AT), a histidine synthase inhibitor, were supplemented to the media to suppress background activation. These assays showed that the −258 to −425 bp region of the QH6 promoter was able to bind HY5 in vivo (Figure 6). The assay with a bait containing four consecutive repeats of the G-box motif showed strong interaction with HY5, whereas that with G-box either removed or mutated had no interaction.

EMSA was performed to further determine the binding of HY5 to a specific DNA motif in this promoter region. A biotin-labeled 52-bp double-strand DNA probe containing the G-box motif and its flanking region was synthesized, while HY5 was prokaryotically expressed and purified as described in the Materials and Methods Section. The labeled probe was incubated with or without HY5 protein preparation and separated on native polyacrylamide gels. Shifted bands were routinely observed with HY5 and QH6 G-box oligonucleotides that were abolished by using a competitor against the entire 52-bp region. Thus, we confirmed that HY5 binds to the QH6 promoter at the G-box motif (Figure 7).

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.