Metabolic engineering of biosynthesis and sequestration of artemisinin

The sesquiterpenoid artemisinin (AN) is the most important medicine for the treatment of malaria in humans. The industrial production of AN still mainly depends on extraction from the plant <em>Artemisia</em> <em>annua</em>. However, the concentration of AN in <em>A. annua </em>is low. Although different engineering strategies have been used in both <em>A. annua</em> and heterologous plant and yeast production platforms, the worldwide capacity and production costs for AN are not in balance with its demand (<strong>Chapter 1</strong>). Although the genes encoding for the entire AN biosynthesis pathway (AN-PW) of the AN precursor dihydroartemisinic acid (DHAA) have been identified, the application of these genes in pathway engineering seem to be limited by lack of control over product transport and sequestration. At the onset of this thesis project there was no information on transport in the AN-PW. However, it was known that DHAA is converted into AN outside the glandular trichome cells of <em>A. annua</em>. Therefore, in this thesis I tried to gain more knowledge on transport within the AN-PW and the use of different metabolic engineering strategies to improve the production of AN. At the onset of my PhD project, the AN-PW genes from two different <em>A. annua</em> chemotypes were compared to understand the basis of different relative activities in the two branches of the AN-PW (<strong>Chapter 2</strong>). For these assays we used transient expression in <em>N. benthamiana</em>. In the AN-PW, artemisinic aldehyde (AAA) is at a branch point as it can be converted to artemisinic acid (AA) by amorphadiene oxidase (<em>AMO</em>), or to dehydroartemisinic aldehyde (DHAAA) by artemisinic aldehyde Δ11 (13) reductase (<em>DBR2</em>). AA is the precursor for arteannuin B (AB) while DHAAA may be converted by a <em>CYP71AV1 or an ALDH1 </em>to dehydroartemisinic acid (DHAA), the precursor for AN. In this chapter we demonstrate that the CYP71AV1 from a high AN production (HAP) chemotype has reduced activity in the AB branch of the pathway compared to the CYP71AV1 from a low AN production (LAP) chemotype. In addition, we show that the relative expression levels of <em>DBR2</em> and <em>ALDH1</em> also affect the AN/AB chemotype. The low catalytic efficiency of AMO from the HAP chemotype may be caused by a deletion of seven amino acids at the N-terminus of the protein compared to CYP71AV1 from LAP. Ectopic expression of the AN-PW genes in <em>N. benthamiana</em> showed that the bulk of the PW products are modified by glycosylation and glutathione conjugations. These side reactions therefore compete with the biosynthesis flux towards the AN precursor DHAA. At this point in my thesis the ectopic expression of AN-PW genes in <em>N. benthamiana</em> had not yielded any AN. At a later stage it became clear that this was due to harvest of leaves at 5-7 days post agro-infiltration (dpi), while AN in <em>N. benthamiana</em> leaves expressing AN-PW genes only becomes detectable after 7 dpi. Glycosylation of the bulk of the AN-PW products in <em>N. benthamiana</em> stresses the need for an efficient transport of (DH)AA to the outside of cells in order to escape from the glycosylation reactions. In Chapter 3, transport and sequestration of AN precursors was investigated by studying the effect of membrane transporters (PDRs) and lipid transfer proteins (LTPs). Hereto, two membrane transporters with activity towards AN-PW products were made available by the group of Prof. Marc Boutry and we isolated three <em>LTP</em> genes from <em>Artemisia annua</em> which showed expression in the glandular trichomes. In this chapter we show that AaLTP3 displays specific activity, together with AaPDR2 towards transport of (DH)AA to the apoplast in <em>N. benthamiana</em>. Moreover, infiltration experiments with (DH)AA in <em>N. benthamiana</em> leaves revealed that these compounds are rapidly taken up by the cells and that inside the cells there is a strong reverse flux in the AN-PW by conversion of (DH)AA towards (DH)AAA and (DH)AAOH. Subsequently we demonstrated that AaLTP3 has a stronger activity in keeping products in the apoplast than the AaPDR2 membrane transporter. Therefore, I suggest that by removal of (DH)AA from the cytosol through transport over the plasma membrane by AaPDR2 and subsequent sequestration in the apoplast by AaLTP3, AaLTP3 creates sink activity which prevents reflux of (DH)AA from the apoplast back into the cells. AaLTP3 therefore contributes to a directional flux through the AN-PW towards the end product (DH)AA. Finally, in this work we could also for the first time detect AN and AB in <em>N. benthamiana</em> leaves by extraction of necrotic leaves at 13 dpi. Because in <em>A. annua</em> glandular trichome cells both the AN sesquiterpene biosynthesis pathway and the flavonoid biosynthesis pathway are active, we explored whether there is a functional interaction between these two major secondary metabolite biosynthesis pathways. In Chapter 4 we describe how we manipulate the flavonoid biosynthesis pathway in <em>N. benthamiana</em> leaves using the <em>Antirhinum majus</em> transcription factor <em>Rosea1</em> (<em>ROS</em>) and test coexpression of <em>ROS</em> with AN-PW genes. The co-expression of <em>ROS</em> stimulates AN-PW product accumulation. Subsequent analysis indicates that this is most likely from transcriptional activation of the enzyme Mevalonate Kinase (<em>MVK)</em> in the mevalonate pathway, which provides precursors for the sesquiterpene biosynthesis pathway. In addition, we demonstrate that production of flavonoids competes with AN-PW product accumulation, as co-expression of AN-PW genes with <em>ROS</em>, but simultaneous inhibition of chalcone synthase (CHS) by a <em>CHSRNAi</em> construct, results in higher AN-PW product levels. However, accumulation of the end products AN and AB was not affected significantly. Finally, the combined expression of <em>AN-PW+ROS+AaPDR2+AaLTP3+ CHSRNAi</em> results in highest sequestration of (DH)AA in the apoplast and highest accumulation of the end products AN and AB in <em>N. benthamiana</em>. During my thesis work, in a related project it was found that expression of another sesquiterpene biosynthesis gene (caryophyllene synthase; <em>CST</em>) in transgenic <em>Arabidopsis</em> resulted in higher caryophyllene emission for a transformant expressing a genomic DNA of <em>CST</em>, compared with a similar transformant expressing a <em>CST</em> cDNA described in literature. This suggested that ectopic expression of intron containing genes is more efficient than ectopic expression of cDNAs. To test whether in the context of metabolic engineering the use of genomic (intron-containing) genes is more efficient than the use of the corresponding cDNA we generated a set of stable transformed <em>Arabidopsis</em> lines with either genomic <em>CST</em> (<em>gCST</em>), cDNA <em>CST</em> (<em>cCST</em>), genomic amorphadiene synthesis (<em>gADS</em>) and cDNA <em>ADS</em> (<em>cADS</em>). In chapter 5 we show that indeed the lines with overexpression of the genomic clones yield higher levels of the anticipated products (caryophyllene or amorphadiene) than the lines with overexpression of the corresponding cDNAs. Transcript analysis showed that for <em>gCST</em> the increase in caryophyllene production was higher than can be explained solely by the increase in <em>CST</em> transcription. In the context of transient expression in <em>N. benthamiana</em> leaves the intron-mediated-enhancement effect was less pronounced. In the final discussion chapter 6 I review limitations and potential solutions to metabolic engineering of the AN-PW in plants, and I discuss the impact of our findings on AN production capacity using transient expression versus natural production in <em>A. annua</em>. Moreover, I discuss how the finding of this thesis go beyond just insights into the AN-PW as especially the identification of the role of LTPs in sesquestration of (sesqui)terpenes into the apoplast may have an impact on the metabolic engineering efforts of many other (sesqui)terpene pathways. Because some plant hormones are also terpenoid products the newly identified role of LTPs may also have impact on a deeper understanding of hormone signalling in plants. I have already started exploring this path by generating a set of <em>Arabidopsis</em> plants with overexpression of different <em>Arabidopsis</em> LTP genes to test whether any hormone related traits are altered (Chapter 6). Preliminary results do indeed confirm a role of LTPs in endogenous plant hormone balance, something worthwhile to be further explored in future research.