By analyzing the sequence information in transcriptome database and full genome database of Artemisia annua established by our lab (most of sequences in the databases have been annotated), and by sequence alignment with reported EIN3/EIL1s from other plant species, the gene sequence highly homologous to other EIN3/EIL1s was selected out from the transcriptome database of Artemisia annua, named as AaEIN3. Homologous alignments of nucleotide and protein sequences were performed with Protein-Blast Tool at NCBI website and Vector NTI 9.0 software. Phylogenetic analysis of AaEIN3 was done through neighbor-joining (NJ) method with MEGA 5.0 software. Full-length coding region of AaEIN3 was obtained and amplified with primers AaEIN3-ORF1 (5′-GGATCCATGGGGATGGGGATCTTTGAAG-3′) and AaEIN3-ORF2 (5′-CTGCAGTCAAAGGTACCACATTG ACATATC-3′).

Artemisia annua plants were grown in soil matrix in a chamber (16 h light/8 h dark) at 25°C for the day/22°C for the night with 65% relative humidity. For analysis of gene expression mode under ethephon (Et) and aminoxyacetic acid (AOA) treatments, 500 μM Et solution, 200 μM AOA solution and sterile water (as the mock) were sprayed evenly over 14-day-old seedlings of Artemisia annua separately. Young leaves at the same position from 8 to 10 seedlings were excised and gathered in the microfuge tube as one biological sample after 0, 1, 3, 6, 9, 12, and 24 h of the treatment, respectively. These samples were stored at -80°C for subsequent RNA extraction and qPCR analysis. For analysis of gene expression mode in transgenic plants under Et treatment, 500 μM Et solution was sprayed evenly over AaEIN3-ox, WT, and RNAi plant lines. Then the 2nd leaves counted downward from top meristems were sampled after 0 and 6 h of the treatment, and stored at -80°C for subsequent RNA extraction and qPCR analysis.

The expression levels of all genes of interest in the study were detected by quantitative PCR (qPCR). Total RNA was extracted from plant samples with the Plant RNA Extraction Kit (Tiangen Biotech, Peking, China) according to the kit’s instructions. Aliquots of 1 μg total RNA was used for cDNA synthesis in a reverse transcription system (Takara, Tokyo, Japan). The amplification reaction of qPCR was performed in a Roche LightCycler 96 Real-Time PCR Device, using SYBR Green qPCR Master Mix reagents (Tiangen Biotech, Peking, China) according to the manufacturer’s instructions. Relative expression levels of genes were normalized to the expression of β-Actin from Artemisia annua. The specific primers for each gene used in qPCR are listed in Supplementary Table S1. mRNA expression levels of the target genes were measured with 2^-ΔC_t.

For AaEIN3-overexpression vector construction, the coding region of AaEIN3 with a BamHI and a PstI restriction site on either end, respectively, was ligated into the pHB+ vector under the control of the CaMV35S promoter to generate pHB-35S:sGFP-AaEIN3:Noster construct, with the sGFP fused to the N-terminal of AaEIN3. For RNA interference (RNAi) vector construction, the primers AaEIN3-RiF (5′-CACCTGAATCGTGGCGGAACGCTAAA-3′) and AaEIN3-RiR (5′-ACTGAAACCCTGCTGGCATAAA-3′) were designed to amplify a 350 bp-long RNAi fragment with cDNA of AaEIN3 as the template. By using the gateway cloning system, the RNAi fragment was first ligated into TOPO vector and then cloned into the pHellsGate12 vector via LR reaction (Invitrogen, United States) to generate the final pHellsGate-RNAi construct. The resulting AaEIN3-overexpression and RNAi constructs were transduced into Agrobacterium tumefaciens strain EHA105, respectively, and then introduced into Artemisia annua to generate transgenic Artemisia annua plants as previously described (Shen et al., 2012). Independent transgenic lines were grown and selected in hygromycin-containing MS medium for pHB-AaEIN3 overexpression transformants, and in kanamycin-containing MS medium for pHellsGate-RNAi transformants. AaEIN3-overexpression plant lines were confirmed by PCR detection with primers AaEIN3-ORF1 (5′-GGATCCATGGGGATGGGGATCTTTGAAG-3′) and rbc48-A (5′-GCATTGAACTTGACGAACGTTGTCGA-3′). And AaEIN3-RNAi lines were confirmed by PCR detection with primers p35S-FP (5′-TTCGTCAACATGGTGGAGCA-3′) and AaEIN3-RiR (5′-ACTGAAACCCTGCTGGCATAAA-3′). These transgenic lines were transferred to soil and kept for further analyses.

HPLC analysis was used to detect the contents of DHAA and artemisinin in leaves of Artemisia annua. Samples were prepared as described previously (Lu et al., 2013). Artemisia annua leaves were dried in a drying oven at 45–50°C for 48–72 h, and then ground to powder in a mortar. 0.1 g dried leaf powder was added into 1.5 ml methanol and ultrasonically oscillated for 30 min at 25°C/50W. After centrifugation at 10000g for 10 min, the clear supernatant was collected and the extraction was repeated once more. The resulting supernatant was filtered through a 0.22-μm membrane. The filtrates were analyzed by a Waters Alliance 2695 HPLC system coupled with a Waters 2420 ELSD detector (Milford, MA, United States). The mobile phase was methanol/H₂O (v:v = 6:4) for artemisinin measurement, and acetonitrile/0.1% acetate (v:v = 6:4) for DHAA measurement. The HPLC conditions were set as described previously (Lu et al., 2013). The standard of artemisinin was purchased from Sigma-Aldrich (Shanghai, China), and the standard of DHAA was purchased from Honsea Sunshine Bio Science & Technology Co., Ltd. (Guangzhou, China).

An earlier report showed that applying 15 mg/L ethephon (Et) in the growing medium of Artemisia Annua seedlings would lead to a decrease of artemisinin content in the roots (Weathers et al., 2005). To detect whether ethylene has an impact on the expression of artemisinin biosynthesis-related genes, we sprayed 500 μM Et solution and 200 μM AOA (an inhibitor of endogenous ethylene biosynthesis) respectively, to 14-day old wild type seedlings of Artemisia annua. The plants of mock group were treated with sterile water. qPCR was done to detect the expression mode of ADS, DBR2, CYP71AV1 (CYP), and AaORA, four important artemisinin biosynthesis-associated genes at different time points after the treatment. The result showed that Et treatment significantly downregulated the expression level of ADS, DBR2, CYP71AV1, and AaORA in leaves, compared to the mock group (Figures 1A,C). The expression level of the four genes dropped to the lowest level after 6 h of Et treatment, and gradually rose back to the normal level afterward (Figure 1A). Meanwhile, treatment with AOA significantly upregulated the expression level of ADS, DBR2, CYP71AV1, and AaORA in leaves, compared to the mock group (Figures 1B,C). The expression level of the four genes rose to the peak after 3 h of AOA treatment, and gradually dropped back afterward (Figure 1B). This indicated that suppressing endogenous ethylene biosynthesis could enhance the expression of ADS, DBR2, CYP71AV1, and AaORA. The above result demonstrated that ethylene negatively regulates the expression of three key enzyme genes ADS, DBR2, CYP71AV1, and a positive regulator gene AaORA that are involved in artemisinin biosynthesis.

ETHYLENE-INSENSITIVE3 (EIN3) is a key transcription factor in ethylene signaling pathway. Ethylene regulates the expression of ethylene-responsive genes downstream via EIN3/EIL1’s functioning, thus further affecting diverse physiological courses of plants as germination and development, secondary metabolism and stress response. We speculated, the repression of the expression of these artemisinin biosynthesis-associated genes by ethylene may involve the mediation of EIN3/EIL1 protein. Therefore, dissecting the role of EIN3 in ethylene’s regulating artemisinin biosynthesis became our major goal in the research.

Through homologous alignments among the sequences from transcriptome database of Artemisia Annua and EIN3/EIL1 gene sequences reported in other species, we isolated the gene sequence of EIN3 from Artemisia Annua transcriptome database, which is highly homologous to EIN3/EIL1 sequences from other species, named AaEIN3. AaEIN3 encodes 604 amino acids, with a coding region of 1815 bp. Multiple alignment showed an identity of 56–67% in protein sequence between AaEIN3 and EIN3/EIL1s from other plant species (Figure 2A). Phylogenetic analysis showed AaEIN3 is closest in evolutionary relationship to EIL1 from Citrus sinensis (CsEIL1) (Figure 2B).

To detect whether AaEIN3 affects artemisinin biosynthesis, we constructed AaEIN3-overexpressing (ox) vector and RNAi vector, respectively, and transformed them into Artemisia Annua plants. PCR was done to select out AaEIN3-overexpressing (ox) and RNAi transgenic plant lines. These transgenic plants were kept for further analysis. qPCR result showed that the expression level of AaEIN3 in AaEIN3-ox plants was significantly elevated, compared to wild type plants (WT), while the expression level of ADS, DBR2, and CYP71AV1, three key enzyme genes in artemisinin biosynthesis, was significantly lower than that of WT (Figure 3A). Meanwhile, the expression level of AaORA, a transcription factor positively regulating these key enzyme genes expression, also significantly declined relative to that in WT (Figure 3A). This indicated that AaEIN3 overexpression reduced the expression of ADS, DBR2, CYP71AV1, and AaORA, which are involved in artemisinin biosynthesis. On the other hand, in RNAi plants, the expression of AaEIN3 got significantly repressed, compared to that in WT, while the expression level of ADS, DBR2, CYP71AV1, and AaORA was significantly higher than that in WT (Figure 3A). This indicated that repression of AaEIN3 expression led to an increase of ADS, DBR2, CYP71AV1, and AaORA expression. These results demonstrated AaEIN3 negatively regulates the expression of ADS, DBR2, CYP71AV1, and AaORA that are involved in artemisinin biosynthesis.

Then we detected the contents of DHAA and artemisinin in AaEIN3-ox, RNAi and WT plants by HPLC. The result showed that in AaEIN3-ox plants, both of DHAA and artemisinin contents were significantly lower than that in WT plants; while in RNAi plants, both of DHAA and artemisinin contents were significantly higher than that in WT plants (Figure 3B). These above results demonstrated AaEIN3 is a negative regulator in artemisinin biosynthesis.

To detect whether the downregulation of artemisinin biosynthesis by ethylene involves the function of AaEIN3, we evenly sprayed AaEIN3-ox, WT, and RNAi plants with 500 μM Et solution, and detected the expression mode of ADS, DBR2, CYP71AV1, and AaORA in leaves of these plant lines at 0 and 6 h after the treatment. The result is shown in Figure 4. Et treatment led to a decline of expression of ADS, DBR2, CYP71AV1, and AaORA in all of AaEIN3-ox, WT, and RNAi plant lines. But in WT, the downregulation effect of Et on the four genes’ expression was the most significant of all. When AaEIN3 expression got repressed by RNAi, the downregulation effect of Et on the expression of ADS, DBR2, CYP71AV1, and AaORA got significantly attenuated relative to that in WT, or rather the responsiveness of the four genes to ethylene signals got attenuated when AaEIN3 expression was repressed by RNAi. These results indicated that ethylene signals downregulate artemisinin biosynthesis via the mediation of AaEIN3.

What is more, overexpression of AaEIN3 significantly reduced the expression level of ADS, DBR2, CYP71AV1, and AaORA, and Et treatment further attenuated the four genes’ expression in AaEIN3-ox plants. But in AaEIN3-ox plants, the extent of the reduction in the four genes’ expression level was not so significant as that in WT after Et treatment (Figure 4). We speculate the expression amount of AaEIN3 in AaEIN3-ox plants has surpassed the normal need for regulating the genes downstream, and is sufficient to downregulate the expression level of those genes to the full extent. Ethylene could repress the degradation of EIN3 protein to make it accumulate gradually. Under the context of AaEIN3 overexpression driven by 35S promoter, the further accumulation of EIN3 impelled by exogenous Et strengthened the negative regulation effect of EIN3 on genes expression not so significantly as that in WT. This may explain why the decline in the four genes’ expression level in AaEIN3-ox plants after Et treatment is much lesser than that in WT plants. This result also suggested a crucial role of AaEIN3 in the regulation of artemisinin biosynthesis by ethylene signaling pathway from the other side.

Previous study of our lab found that the expression level of ADS, DBR2, CYP71AV1, and AaORA, and the content of DHAA, the end-product of enzymatic reactions in artemisinin biosynthesis, are relatively higher in younger leaves. As leaves get matured and senescent, the four genes expression level and DHAA content get lower rapidly, that demonstrated that the process of leaf maturation and senescence repressed the expression of these genes participating in artemisinin biosynthesis (Zhang et al., 2012; Lu et al., 2013). Besides, an earlier study reported that EIN3 transcription in leaves of Arabidopsis would increase with leaves aging, and that EIN3 protein could directly repress miR164 transcription to increase the expression of senescence-associated genes as NAC2 (also named ORE1) and SAG12, thus accelerating leaf senescence (Li et al., 2013). Therefore, EIN3 is a senescence-associated gene, which can promote the process of ethylene- or age-dependent leaf senescence. Based on the previous findings, we infer that ethylene’s negative regulation of artemisinin biosynthesis may correlate with EIN3-induced leaf senescence.

To confirm such inference, qPCR analysis was done to detect the expression mode of genes in leaves at different ages or developmental stages in the same phyllotaxy. On a stem of the plant, the closer to top meristems, the younger the leaf is; the lower located, the more aged and senescent the leaf is. qPCR result showed that AaEIN3 expression level is lowest in the youngest leaf closest to top meristems (marked as L1 in Figure 5), and gets higher as leaf becomes aged and senescent in all of AaEIN3-ox, WT, and RNAi plant lines (Figure 5A). But the expression mode of ADS, CYP71AV1, DBR2, and AaORA, is on the contrary to that of AaEIN3: the more aged and senescent the leaf is, the lower their expression level gets (Figures 5B–E). This result is consistent with the previous report, indicating the aging and senescence of leaves attenuated the expression of the four genes involved in artemisinin biosynthesis.

Then, to detect whether AaEIN3 expression could accelerate leaves’ senescence process, we screened Artemisia Annua transcriptome database and its full genome database (the sequence information of both databases has not been published yet), and selected out the cDNA sequences of a senescence-associated gene NAC2 and a senescence marker gene SAG12 in Artemisia Annua, named AaNAC2 and AaSAG12, respectively. qPCR result showed that the expression level of AaNAC2 and AaSAG12 increased with the increase of leaves age (Figures 5F,G), as does the expression level of NAC2 (ORE1) and SAG12 in Arabidopsis thaliana reported previously (Li et al., 2013). Through comparing the expression level of AaNAC2 and AaSAG12 in the leaves at the same developmental stage (ranking at the same position in the phyllotaxy) among AaEIN3-ox, WT, and RNAi plants, we found AaEIN3 overexpression increased AaNAC2 and AaSAG12 expression, while repressing AaEIN3 expression by RNAi reduced AaNAC2 and AaSAG12 expression (Figures 5F,G). This indicated AaEIN3 could promote the expression of leaf senescence-associated genes.

Moreover, we observed the phenotype of leaves in the same phyllotaxy in AaEIN3-ox, WT, and RNAi plants. In AaEIN3-ox plants, the point at the edge of the 5th leaf counted downward from the top meristems (L5) began to show slight etiolation, and the etiolation phenomenon became more and more visible in the 7th, 9th, and 16th leaves counted downward from the top meristems (L7, L9, L16). Compared to AaEIN3-ox plants, L5 in WT plants has not shown etiolation, and the point at the edge of L7 and L9 in WT only exhibited slight trace of etiolation. The etiolation signs got more visible in L16 of WT plants, but the etiolation extent of L16 in WT plants was still lesser than that in AaEIN3-ox plants. On the other hand, none of the leaves L1–L16 in RNAi plants exhibited etiolation signs (Figure 6). These phenomena indicated AaEIN3 overexpression accelerated leaf senescence, and repression of AaEIN3 expression by RNAi delayed leaf senescence, which further demonstrated AaEIN3 is a senescence-associated gene that can induce leaf senescence in Artemisia annua.

Taken together, our study proposed a mechanism by which ethylene negatively regulates artemisinin biosynthesis, shown in Figure 7. Ethylene signal promotes the senescence process of leaves in Artemisia annua via the mediation of AaEIN3, a key component in ethylene signaling pathway. And leaf senescence attenuates the expression of ADS, DBR2, CYP71AV1, and AaORA, thus reducing DHAA biosynthesis and finally leading to the decrease of artemisinin accumulation in Artemisia annua. Our work discovered the first negative regulator (AaEIN3) for artemisinin biosynthesis and provided more novel knowledge and clues on how plant hormone signals regulate artemisinin metabolic pathway. However, it’s still to be explored and studied as for the issues how the process of maturation and senescence of leaves modulates the expression of key genes involved in artemisinin biosynthesis, and what transcription factors or other signaling pathways get involved in this modulation. All of these issues need further research.