Plants, as green photosynthetic organisms, synthesize and accumulate a diverse array of specialized compounds through secondary metabolic pathways. These compounds (estimated at 200,000) are not normally involved in basic plant growth and development but are essential for plants to survive in response to their environment. Secondary metabolites, on the other hand, are not only important for plant adaptation strategies but also have benefits for humans. For instance, over the last two decades, more than 70% of FDA-approved medicine has been based on secondary metabolites found in plants (Schramek et al., 2010; Yang et al., 2016). Artemisia annua L. is a valuable medicinal plant used in Chinese traditional medicine that produces artemisinin, a unique secondary metabolite (Tu, 2011). The impact of artemisinin against Plasmodium falciparum, the primary malaria agent, has received considerable attention (Mutabingwa, 2005). Malaria is a deadly disease that kills thousands of people each year and is spread by female Anopheles mosquitos (World Health Organization, 2016). Although artemisinin is widely considered the most effective anti-malarial compound, the low rate of artemisinin production in plants (< 1%) necessitates numerous efforts to improve the artemisinin yield in plants (Ro et al., 2006; Paddon et al., 2013; Shen et al., 2016). Even though semi-synthesis of artemisinin via artemisinic acid in engineered yeast was achieved (Turconi et al., 2014), it was unable to meet industrial demands due to significantly higher costs than artemisinin produced in plants (Peplow, 2016). As a result, identifying enzyme-coding genes as well as the regulatory network of transcription factors is critical for transformation strategies and the production of higher yield lines. So, in recent years, there has been a significant amount of research to understand the molecular mechanism underlying the regulation of artemisinin biosynthesis (Hassani et al., 2020; Kayani et al., 2021; Shu et al., 2022).

Artemisinin, like other secondary metabolites, is synthesized by a multi-enzymatic biosynthetic pathway. Terpenoid biosynthesis is initiated by the cytosolic mevalonate pathway (MVA) and plastid methylerythritol phosphate (MEP) derived isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP) (Vranová et al., 2013). IPP and DMAPP produced by the MVA pathway are used primarily to power the production of polyisoprenoids, triterpenes, and sesquiterpenes. MEP products, on the other hand, are typically used to drive monoterpene, diterpene, and tetraterpene biosynthesis (Vranová et al., 2013; Yang et al., 2016). However, the biosynthesis of artemisinin is initiated by farnesyl diphosphate synthase (FPS) activity on plastidial and cytosolic derived IPP and DMAPP to convert them to farnesyl diphosphate (FPP) ( Figure 1 ) (Schramek et al., 2010). Furthermore, the amorpha-4,11-diene synthase (ADS) forms a double bond or a new ring structure on phosphates, breaking the chemical bond of FPP, generating amorpha-4,11-diene, which is then hydroxylated, and oxidized by cytochrome P450 monooxygenase (CYP71AV1) and cytochrome P450 oxidoreductase (CPR) to artemisinic alcohol, artemisinic aldehyde, and artemisinic acid (Ro et al., 2006; Teoh et al., 2006; Alam et al., 2010). Furthermore, artemisinic aldehyde Δ11(13) reductase (DBR2) reduces artemisinic aldehyde to dihydroartemisinic aldehyde, which is then converted to dihydroartemisinic acid (DHAA) by aldehyde dehydrogenase 1 (ALDH1) as the final compound generated by enzymatic reactions in A. annua (Teoh et al., 2006).

Among the various techniques used to boost artemisinin production, including breeding, chemosynthesis, and microbial metabolic engineering (Brisibe and Chukwurah, 2014; Corsello and Garg, 2015), overexpression strategies of the enzyme-coding genes of the artemisinin biosynthetic pathway have proven to be a successful approach. Among the key enzymes well-studied, 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGR) is an important regulatory enzyme responsible for diverting HMG-CoA into the mevalonate arm of the isoprenoid pathway where it is converted to (IPP) and further to FPP, respectively (Ram et al., 2010). Since FPP is the precursor for essential isoprenoids, including artemisinin, its production by overexpression of farnesyl diphosphate synthase (FPS) has been investigated in A. annua (Chen et al., 2000; Han et al., 2006a). DBR2, another essential enzyme in artemisinin biosynthesis, is not only crucial for its catalytic activity but can also be considered as a checkpoint to direct the conversion of artemisinic aldehyde to dihydroartemisinin aldehyde toward artemisinin production (Yuan et al., 2015). Minimizing the activity of competitive branch pathways is a practical strategy for a successful transformation toward a higher yield of the desired product. HMGR, FPS, and DBR2 with checkpoint features could be suitable candidates in this regard.

The biosynthesis of artemisinin in A. annua occurs in glandular secretory trichomes (GSTs) of the leaves. GST structures are made up of 10 symmetric cells and are thought to be the main biofactories for artemisinin production, storage, and secretion, however, there is also a report on the production of artemisinin in the non-glandular trichomes of self-pollinated inbred A. annua (Judd et al., 2019). The exclusive expression of artemisinin pathway key enzymes in GSTs, nominate them to be the true cell biofactories and the target for metabolic engineering and breeding strategies to improve their efficiency, such as by increasing their density (Mutabingwa, 2005; Schilmiller et al., 2008; Chen et al., 2021). However, the lack of a model plant for GST study, as well as poor knowledge of the mechanism underlying their development, are two limiting constraints. The initiation of GSTs and T-shaped non-secretory trichomes (TSTs) via distinct transcription factor (TF) networks, as well as information from key enzyme expression profiles, have led to a possible solution (Ro et al., 2006). Among the most common TF families in A. annua, the AaORA an APETALA2/ETHYLENE RESPONSIVE FACTOR (AP2/ERF) TF was cloned and found to be trichome specific. Meanwhile, the qPCR analysis showed that its expression pattern decreases throughout leaf development, with the highest expression in young leaves. The HPLC results confirmed the elevation of artemisinin and dihydroartemisinic acid in AaORA overexpressed A.annua (Lu et al., 2013).

Plant hormones, in addition to transcription factors that regulate gene expression, play an important role in increasing secondary metabolite production (Wasternack and Hause, 2013). Recent studies in A. annua have suggested a possible relationship between Jasmonic acid (JA) and the homeodomain-leucine zipper (HD-ZIP) TF, leading to epidermal cell differentiation into glandular trichome (Katsir et al., 2008; Yan et al., 2017). It has been reported that AaTCP15 (teosinte branched1/cycloidea/proliferating) TF responds to both JA and ABA and is required for artemisinin biosynthesis by binding to and activating the promoters of the DBR2 and ALDH1 genes; additionally, AaORA (octadecanoid-derivative responsive AP2-domain protein), a positive regulator of artemisinin biosynthesis, responds to JA and ABA and activates AaTCP15 transcripts. As a result, they collaborate synergistically to activate DBR2, a critical gene for the production of artemisinin, by forming an AaORA-AaTCP15 module (Ma et al., 2021). We have previously demonstrated that AaHD1 interacts with Jasmonate ZIM-domain 8 (AaJAZ8) and suppresses its biological activity, therefore mediating GST initiation and increasing artemisinin content (Yan et al., 2017). There is also research on the role of light in JA-induced artemisinin promotion; binding of AaHY5 to the promoter of AaWRKY9 results in its activation, and AaWRKY9 is a positive regulator of AaDBR2 and AaGSW1 in artemisinin biosynthesis. There is also an interaction between AaWRKY9 and AaJAZ9, which is a repressor in the JA signaling pathway; the absence of JA results in transcriptional repression of AaWRKY9 by AaJAZ9 (Fu et al., 2021).

Based on previous research, we hypothesized that simultaneous cotransformation of GST-specific TFs and artemisinin biosynthetic pathway enzyme-coding genes in A. annua would increase the density of the trichome and ultimately the content of artemisinin. So, three independent vectors ProCYP71AV1::AaORA, 35S::HD1 (GST specific TF), and 35S::EPSPS+HMGR+FPS+DBR2 (artemisinin biosynthetic pathway enzyme-coding genes) were simultaneously co-transferred to A. annua to generate the T0 lines. Furthermore, the transformation frequency, regeneration capacity, and artemisinin yield of the transgenic plants of the T1 line were investigated. The results indicated that this simultaneous cotransformation could increase trichome density and the content of artemisinin not only in the T0 line but also in the T1 line.

(Willcox, 2009; Peplow, 2016)

In this study, we overexpressed the artemisinin mainstream enzyme-coding genes to reduce substrate flux to side branches and increase artemisinin production. To gain this, we isolated, cloned, and over-expressed HMGR, which correlates the carbon flux from the cytosolic MVA pathway to the artemisinin pathway, FPS, which catalyzes the initial step of essential isoprenoid synthesis, and DBR2, which is a checkpoint for directing the conversion of artemisinic aldehyde to dihydroartemisinic aldehyde toward artemisinin production (Han et al., 2006b; Ram et al., 2010; Ikram et al., 2019). The high expression of HMGR, FPS, and DBR2 in T0 transgenic plants was followed by higher artemisinin content (2.8- and 3.2-fold higher artemisinin) in Line E and F, respectively; indicating that these genes are suitable candidates for overexpression strategies and useful for increasing artemisinin content in A. annua. Several studies have shown that individually transforming enzyme-coding genes or TFs can boost the production of target metabolites in plants (Aquil et al., 2009; Yuan et al., 2015; Shi et al., 2017; He et al., 2022). Over-expression of DBR2 (Yuan et al., 2015), HMGR (Aquil et al., 2009), and FPS (Banyai et al., 2010), for example, increased artemisinin content by up to 2, 2.2, and 2.5 folds, respectively. Another study found that overexpressing HMGR from Catharanthus roseus in A. annua via Agrobacterium resulted in a nearly 50% increase in artemisinin content when compared to non-transgenic plants (Nafis et al., 2011). Also, AaWRKY 1 overexpression in A. annua resulted in CYP71AV1 upregulation, which increased artemisinin content by up to 1.8 times higher than wild-type A. annua (Han et al., 2014).

To increase GST density in transgenic plants, we also overexpressed two important trichome-specific TFs (Lu et al., 2013; Yan et al., 2017). The over-expressing pCAMBIA2300-proCYP71AV1::AaORA and pHB::AaHD1 vectors were constructed and transformed into the plants. The higher artemisinin content in the T0 and T1 transgenic lines was associated with a higher expression level of these genes as well as an increased number of GSTs. Previous studies indicated that overexpression of AaORA and AaHD1 (trichome-specific TFs) can increase trichome density in transgenic lines, thus increasing the artemisinin content by up to 1.5 folds (Lu et al., 2013; Yan et al., 2017). Overexpression of AaGSW2, a specific WRKY transcription factor for trichomes, has also been reported in trichomes, resulting in increased GST density in A. annua; however, knockdown lines showed impaired GST initiation. They also discovered that the ectopic expression of the Mentha spicata and Mentha haplocalyx AaGSW2 homologs in A. annua resulted in GST formation (Xie et al., 2021). AaMYB17 and AaMIXTA1 are also reported to positively regulate glandular trichome formation in A. annua and overexpressing lines demonstrated increased glandular initiation, whereas repression of these TFs resulted in decreased glandular trichome density (Shi et al., 2018; Qin et al., 2021). Also, AaMIXTA1 and AaHD8 form a regulatory complex that positively regulates trichome initiation and cuticle development in A. annua, ultimately increasing the expression of AaHD1 and cuticle development genes (Yan et al., 2018).

Our findings from the T0 and T1 transgenic lines suggested that the co-transformation of these genes could be an efficient strategy to increase artemisinin content while also forming a stable transformation in progeny plants.

Since the WHO recommended artemisinin combination therapy as the first line of treatment in the battle against malaria, there has been an increase in demand for its supply, but there is still a significant shortfall to meet global demand. Although semisynthesis artemisinin production from genetically modified yeast is conceivable but more expensive than production from plants, A. annua remains the primary commercial outlet for artemisinin. Therefore, transformation strategies will continue to be the primary approach to meet market expectations. We demonstrated in this study that the simultaneous co-transformation of three mainstream enzyme-coding genes and two trichome-specific TFs can promote artemisinin content in Artemisia annua to generate a high artemisinin-yielding progeny plant, which may secure a steady supply of artemisinin globally at affordable prices.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

DH designed and performed the experiments and drafted the manuscript. AT evaluated and analyzed the data, prepared figures, and revised the manuscript. XF provided the overexpression vectors and revised the manuscript. WQ, LH, and YM revised the manuscript. KT served as the principal investigator and corresponding author of this research. All authors contributed to the article and approved the submitted version.