Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Artemisinin, in the form of artemisinin-based combination therapies (ACTs), is currently the most important compound in the treatment of malaria. The current commercial source of artemisinin is Artemisia annua, but this represents a relatively expensive source for supplying the developing world. In this study, the possibility of producing artemisinin in genetically modified plants is investigated, using tobacco as a model. Heterologous expression of A. annua amorphadiene synthase and CYP71AV1 in tobacco led to the accumulation of amorphadiene and artemisinic alcohol, but not artemisinic acid. Additional expression of artemisinic aldehyde Δ11(13) double-bond reductase (DBR2) with or without aldehyde dehydrogenase 1 (ALDH1) led to the additional accumulation dihydroartemisinic alcohol. The above-mentioned results and in vivo metabolic experiments suggest that amorphane sesquiterpenoid aldehydes are formed, but conditions in the transgenic tobacco cells favour reduction to alcohols rather than oxidation to acids. The biochemical and biotechnological significance of these results are discussed.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Artemisinin, in the form of artemisinin-based combination therapies (ACTs), is currently the most important compound in the treatment of malaria (Weina, 2008). Two Artemisia species that contain artemisinin (Liersch et al., 1986), including Artemisia annua, have a long history of use in traditional Chinese medicine (Li et al., 2006). Artemisinin, per se, was isolated and characterized about 40 years ago (Liu et al., 1979) and shown to be active against the malaria parasite, Plasmodium falciparum. Since then, a number of derivatives of artemisinin have been developed and marketed as highly effective anti-malarial agents that have overcome issues of the parasite’s resistance to some of the established anti-malarials (Weina, 2008).
The successful development of ACTs has created problems for the consistent, low-cost supply of artemisinin for malaria treatment in the developing world (Kindermans et al., 2007). Artemisinin is currently extracted from cultivated A. annua. This is relatively expensive, and there has been considerable interest in recent years in understanding how artemisinin is produced in A. annua and how its production costs could be reduced (Covello, 2008).
Some genetic engineering approaches have been directed towards improvement of artemisinin in A. annua itself, with modest success. Examples include the suppression of squalene synthase (Zhang et al., 2009) and the overexpression of farnesyl diphosphate synthase (FPS) (Chen et al., 2000). In tobacco, heterologous expression of amorpha-4,11-diene synthase (ADS) resulted in the accumulation of amorphadiene in transgenic plants (Wallaart et al., 2001). Chappell et al. (Wu et al., 2006) found that targeting FPS and ADS to the plastids had a large positive effect on the degree of accumulation of amorphadiene.
Given the successful production of amorphadiene in a heterologous host and the availability of additional genes encoding enzymes of artemisinin biosynthesis, we have undertaken to genetically engineer tobacco (Nicotiana tabacum) with a view to producing the most advanced precursors possible. Our efforts have been important in highlighting the “bottlenecks” that challenge the efficient production of artemisinin in heterologous plants and tissues.
Production of amorphadiene in transformed tobacco
The initial step in engineering artemisinin precursor production was the replication of amorphadiene production (Wallaart et al., 2001; Wu et al., 2006). The plasmid pYZ044 (see Figure 1) was designed for plastid-localized expression of farnesyl diphosphate synthase and amorphadiene synthase. The plasmid was introduced into tobacco using a standard Agrobacterium tumefaciens-dependant protocol. Some transgenic lines transformed with pYZ044 showed reduced growth rates and a bleached appearance (data not shown). The expression of ADS was confirmed by RT-PCR (see Figure 2). The mass spectra of various N,O-bis-(trimethylsilyl)acetamide-treated amorphane sesquiterpenoid standards are shown in Figure 3. This information was used to determine the production of amorphadiene in the leaves of 21 independent transformed tobacco lines, as shown in Figures 4 and 5. Amorphadiene levels ranged from undetectable, as in wild-type leaves, to approximately 4 μg/g fresh weight in leaves of the transformed tobacco. This is comparable to levels reported previously for very similar experiments (Wu et al., 2006).
The effect of co-expression of CYP71AV1 with FPS and ADS
CYP71AV1 is the A. annua cytochrome P450 responsible for oxidation of amorphadiene (Ro et al., 2006; Teoh et al., 2006). In yeast, CYP71AV1 is capable of oxidizing amorphadiene to artemisinic acid via artemisinic alcohol and artemisinic aldehyde. Thus, one might expect tobacco transformed with pYZ046 to accumulate artemisinic acid. The plasmid pYZ046 represents the addition of a module for CYP7AV1 to the pYZ044 construct. As indicated in Figure 5, leaves of tobacco transformed with pYZ046 accumulated artemisinic alcohol and to a lesser extent the precursor amorphadiene. No artemisinic aldehyde or acid was detectable in the same leaves. Thus, CYP71AV1 was apparently functional in the transgenic tobacco plants and was capable of accessing amorphadiene (presumably synthesized in the plastids), but it was not able to effect the accumulation of the more oxidized amorphadiene derivatives, artemisinic aldehyde and artemisinic acid.
The reason for the lack of accumulation of artemisinic aldehyde and acid in plants expressing CYP71AV1 was investigated through in vivo metabolic experiments (see Figure 6). Wild-type tobacco leaves were supplied with artemisinic and dihydroartemisinic aldehydes, incubated and analyzed by GC/MS. The supplied aldehydes were recovered from the leaves. Interestingly, when artemisinic aldehyde was supplied, wild-type leaves also accumulated artemisinic alcohol, but no artemisinic acid. Similarly, when dihydroartemisinic aldehyde was supplied, both the aldehyde and dihydroartemisinic alcohol accumulated in the absence of the formation of dihydroartemisinic acid. The earlier observations suggest at least two possibilities: (i) CYP71AV1 is not active in the oxidation of artemisinic aldehyde in tobacco leaves or (ii) the conditions in tobacco leaves favour the reduction of sesquiterpene aldehydes to their cognate alcohols and any oxidation of artemisinic alcohol may be countered by endogenous alcohol dehydrogenase activity catalyzing the reverse reaction.
The effect of co-expression of DBR2 and ALDH1 with FPS, ADS and CYP71AV1
Given the propensity for tobacco leaves to reduce sesquiterpene aldehydes, the possibility of trapping the sesquiterpenoid production in transformed plants as artemisinic or dihydroartemisinic acid was investigated. The rationale is that if CYP71AV1 catalyzes the oxidation of artemisinic alcohol to the aldehyde in tobacco, but it is then being reduced by endogenous enzymes, it might be possible to convert the aldehyde to the acid, a form less likely to undergo reduction.
To this end, transgenic tobacco lines were transformed with pYZ047 and pYZ048 to provide for expression of artemisinic aldehyde Δ11(13) double-bond reductase and aldehyde dehydrogenase 1 (ALDH1) in addition to FPS, ADS and CYP71AV1. Plants of 6 positive pYZ047 transgenic lines and 22 pYZ048 lines were analyzed by GC–MS. The accumulation of amorphadiene, artemisinic alcohol and dihydroartemisinic alcohol was found in many of the transgenic leaves (Figure 5). Neither artemisinic acid nor dihydroartemisinic acid was found in the extracts from the pYZ047 and pYZ048 transgenic lines. The presence of dihydroartemisinic alcohol, in both sets of transgenic plants, indicates that DBR2 was functional. Furthermore, given the fact that DBR2 is not thought to act on artemisinic alcohol (Zhang et al., 2008), it suggests that artemisinic aldehyde is being formed in the transformed plants, presumably by CYP71AV1, providing DBR2 with substrate. In other words, in plants expressing ADS, CYP71AV1 and DBR2, it appears that artemisinic aldehyde was probably formed by CYP71AV1, reduced to dihydroartemisinic aldehyde by DBR2 and further reduced to dihydroartemisinic alcohol, probably by an endogenous enzyme.
The activity of ALDH1 in the transgenic plants is unclear. In the presence of NAD(P), ALDH1 is capable of oxidizing both artemisinic aldehyde and dihydroartemisinic aldehyde to the corresponding acids (Teoh et al., 2009). Given the evidence for aldehyde formation and the lack of sesquiterpenoid acid formation, it seems that, in tobacco leaves, ALDH1 is either not capable of competing with other enzymes for aldehydes, the redox conditions are unfavourable for ALDH1 activity, or it is not active. Unfortunately, it was not possible to detect ALDH1 activity in extracts of pYZ048 transgenic leaves (data not shown), and it is therefore difficult to distinguish between these possibilities.
The metabolism of artemisinic and dihydroartemisinic aldehydes was investigated by supplying them to pYZ048 transgenic tobacco leaves (data not shown). As found with wild-type leaves, aldehydes were converted to alcohols. In a slight twist, artemisinic aldehyde was converted to dihydroartemisinic alcohol, as well as artemisinic alcohol. The former alcohol was presumably produced by the action of DBR2 (on artemisinic aldehyde) and an endogenous alcohol dehydrogenase activity. It is notable that a similar conversion of artemisinic aldehyde to dihydroartemisinic alcohol was catalyzed by the extracts of A. annua reported previously (Bertea et al., 2005).
The production of amorphane sesquiterpenoid acids in yeast has been successful, at least qualitatively, for both artemisinic (Ro et al., 2006) and dihydroartemisinic (Zhang et al., 2008) acids. The artemisinin pathway-specific enzymes involved were ADS, CYP71AV1 and DBR2. Both products can be converted chemically to artemisinin and its pharmaceutical congeners (Covello, 2008). It therefore seemed reasonable to attempt similar engineering experiments to produce sesquiterpenoid acids in transgenic plants. The results show accumulation of precursors to the acids, but not of the acids themselves (see Figure 5).
Amorphadiene was clearly produced in at least some lines for each construct used in transformation. Furthermore, in at least some lines transformed with each plasmid designed for CYP71AV1 expression, artemisinic alcohol accumulated. Clearly, the cytochrome P450 was active in those lines and had access to the amorphadiene substrate (presumably biosynthesized in the plastids), to reduced cytochrome P450 reductase and to molecular oxygen. It is worth noting that, because of the volatility of amorpha-4,11-diene, the overall production of the compound in transgenic tobacco is likely to be significantly underestimated (Wu et al., 2006).
If CYP71AV1 is active in transgenic tobacco, one would expect it to produce artemisinic aldehyde from the alcohol that accumulated. There is in fact, indirect evidence that artemisinic aldehyde is formed from CYP71AV1. When DBR2 is co-expressed, dihydroartemisinic alcohol accumulates (see Figure 5 results for pYZ047 and pYZ048). Given the in vivo metabolism experiments and the transgenic results, the best explanation for the earlier observation is that CYP71AV1 produces artemisinic aldehyde, and this is converted to dihydroartemisinic aldehyde by the action of DBR2 and to dihydroartemisinic alcohol by an endogenous alcohol dehydrogenase under conditions that favour reduction. Thus, both artemisinic aldehyde and, in the case of DBR2 expression, dihydroartemisinic aldehyde are probably produced, but do not accumulate, nor are they converted to acids in transgenic tobacco.
The possible reasons for lack of sesquiterpene acid accumulation in transgenic tobacco include the following:
1 CYP71AV1 alcohol oxidase and aldehyde oxidase activities are weak relative to amorphadiene hydroxylase activity. This is consistent with the results herein but is not in agreement with the yeast results (Ro et al., 2006).
2 ALDH1 activity is weak or the cellular environment favours reduction. Unfortunately, it was not possible to confirm ALDH1 activity in the extract of transgenic tobacco leaves.
3 An endogenous alcohol dehydrogenase or related enzyme converts sesquiterpenoid aldehydes to alcohols in an environment that favours reduction. This is consistent with the in vivo metabolism of artemisinic and dihydroartemisinic aldehydes in wild-type tobacco (Figure 6) and with the conversion of artemisinic aldehyde to dihydroartemisinic alcohol in A. annua extracts (Bertea et al., 2005).
4 The toxicity of sesquiterpene aldehydes or acids may prevent survival or propagation of functional transgenic plants that would otherwise be capable of aldehyde and acid accumulation. This could best be tested through the use of inducible promoters.
At this point, the best explanation would appear to be that the cellular compartment in transgenic tobacco, in which the metabolism of sesquiterpenoid aldehydes occurs, favours reduction. This highlights the importance of designing a transgenic plant system that takes into account the tissue and cellular compartment in which transgenes are expressed. In nature, artemisinin is produced in glandular secretory trichomes. Indeed, there is evidence that biosynthesis occurs in the nonphotosynthetic apical cells of the trichome (Olsson et al., 2009). These cells may provide a biochemical environment that is particularly conducive to the production of sesquiterpenoid acids. In other plants, sesquiterpene lactones are produced in different tissues including roots, in many cases, accumulating in laticifers (Hagel et al., 2008). Thus, the compartmentalization of sesquiterpene lactone biosynthesis appears to be a general phenomenon.
In this study, the 35S promoter of the cauliflower mosaic virus was used to provide constitutive expression of transgenes in tobacco. It remains possible that specific tobacco cell types, such as those found in tobacco trichomes, could provide an environment suitable for amorphane sesquiterpenoid acid production. The lack of even traces of such acid production in our experiments suggests that this may not be the case. However, it would be interesting to investigate the possibility further in tobacco or other species, especially in the Asteraceae.
From a commercial point of view, the amount of artemisinin and its precursors produced in plants is important. Some A. annua cultivars produce up to 2% dry weight artemisinin. In this study, sesquiterpenoid alcohols were produced at levels of about 0.0001% fresh weight or roughly 0.001% dry weight. Consequently, while this study provides important qualitative indications of the possible bottlenecks in the engineering of plant-derived artemisinin, considerable effort will be required to match native A. annua using heterologous plant hosts. As has been suggested, sesquiterpene lactone-producing members of the Asteraceae may be suitable hosts (Covello, 2008).
The transformation of N. tabacum with combinations of A. annua genes, which include those encoding amorpha-4,11-diene synthase (ADS), amorpha-4,11-diene oxidase (CYP71AV1) and artemisinic aldehyde Δ11(13) double-bond reductase (DBR2), results in the accumulation of amorphadiene and amorphane sesquiterpenoid alcohols. This indicates that the cellular environment does not favour the formation of commercially desirable sesquiterpenoid acids.
Artemisinic acid and dihydroartemisinic acid were isolated in this laboratory from dichloromethane extracts of A. annua flower buds and leaves as described for artemisinic acid (Teoh et al., 2006). Artemisinic aldehyde, artemisinic alcohol, dihydroartemisinic aldehyde and dihydroartemisinic alcohol were prepared from the corresponding isolated acid as described previously (Zhang et al., 2008). Amorpha-4,11-diene was synthesized from artemisinic acid as described previously (Chang et al., 2000). Octadecane was obtained from Sigma-Aldrich (St. Louis, MO, US). Unless otherwise specified, all commercial enzymes were obtained from New England Biolabs (Cambridge, MA, USA).
Preparation of plasmids for tobacco transformation
In this study, four plasmids (pYZ044, pYZ046, pYZ047 and pYZ048) were prepared for use in A. tumefaciens-mediated transformation of tobacco. The ORFs introduced into these plasmids included subsets of tpFPS, tpADS, CYP71AV1, DBR2 and ALDH1. The ORFs for A. annua FPS, ADS, CYP71AV1, DBR2 and ALDH1 were isolated by RT-PCR from the total RNA of A. annua prepared using RNeasy Plant Mini Kit (Qiagen, Mississauga, ON, Canada). The N. tabacum rbcS transit peptide was isolated from the Xanthi cultivar by PCR using genomic DNA as the template. The oligonucleotide primers used for all the above-mentioned PCRs and the nucleotide range that they span are listed in the Table 1. The resulting PCR products were initially cloned into pCR2.1-TOPO (Invitrogen, Carlsbad, CA, US) and confirmed by DNA sequencing.
Table 1. Oligonucleotide primers used in this study. Restriction endonuclease sites are underlined, and start and stop codons are indicated in bold
5′-end of ORF
3′-end of ORF
5′-end of ORF
3′-end of ORF
5′-end of ORF
3′-end of ORF
5′-end of ORF
3′-end of ORF
5′-end of ORF
3′-end of ORF
rbcS transit peptide
rbcS transit peptide
Overlap extension PCR
Overlap extension PCR
Overlap extension PCR
Overlap extension PCR
The DNA sequence encoding the tobacco rbcS transit peptide was fused separately to the ORFs of FPS and ADS by overlap extension PCR. For tpFPS, the initial transit peptide and FPS PCR products were generated using the corresponding cDNA clones as templates and the primer pairs 11/14 and 2/13, respectively. The resulting PCR products were mixed, and a tpFPS product was amplified with the primers 2 and 11. The same strategy was used to amplify a transit peptide/ADS fusion product using primers 4, 11, 15 and 16 (Table 1), and the final tpADS product was obtained by PCR using primers 4 and 11.
The starting point for the construction of plasmids for tobacco transformation was pRD400 (Datla et al., 1992). The vector was modified with the introduction of new multiple cloning sites to give the binary vector pYZ040 (Figure 1).
A range of DNA cassettes for plant expression were prepared in the vector pUC19 giving pYZ030, pYZ032 and pYZ039. These plasmids include the 35S promoter and the Nos terminator [which were PCR amplified using the binary vector pBI121 (Genbank accession number AF485783) as the template]. They differ specifically in the multiple cloning sites used for insertion of ORFs and the fact that pYZ039 has two promoter/terminator combinations. A two-gene construct, pYZ039-tpFPS-tpADS was prepared by inserting the tpFPS and tpADS ORFs into the multiple cloning sites of pYZ039 using NheI and PacI and subsequently, ApaI and SbfI. CYP71AV1 was inserted into pYZ032 at the ApaI and SbfI restriction sites resulting in a one-gene construct, pYZ032-CYP71AV1. DBR2 and ALDH1 were ligated separately into pYZ030 at the NheI and PacI restriction sites to give the one-gene constructs pYZ030-DBR2 and pYZ030-ALDH1, respectively. A three-gene construct, pYZ039-tpFPS-tpADS-CYP71AV1, was made by excising a XhoI/AsiSI fragment from pYZ032-CYP71AV1 (with promoter, ORF and terminator) and introducing it into pYZ039-tpFPS-tpADS at the SalI and AsiSI restriction sites. A four-gene construct, pYZ039-tpFPS-tpADS-CYP71AV1-DBR2, the DBR2 expression cassette from pYZ030-DBR2 was made by excising a XhoI/AsiSI fragment and introducing into pYZ039-tpFPS-tpADS-CYP71AV1 using SalI and AsiSI restriction enzymes. The same approach was applied to make the five-gene construct pYZ039-tpFPS-tpADS-CYP71AV1-DBR2-ALDH1 using pYZ030-ALDH1, and XhoI and AsiSI, and the above-mentioned four-gene construct at the SalI and AsiSI restriction sites.
Plasmids for plant transformation were prepared from these two-, three-, four-, and five-gene constructs by digestion with AscI and AsiSI and ligation into the binary vector pYZ040 at AscI and AsiSI sites, to give pYZ044, pYZ046, pYZ047 and pYZ048, respectively (Figure 1). All the binary constructs were confirmed by DNA sequencing before being transferred into A. tumefaciens, strain GV3101(pMP90), by electroporation.
N. tabacum cv ‘Xanthii’ was transformed with the A. tumefaciens containing the binary constructs described earlier using a modified A. tumefaciens-mediated transformation protocol (Lloyd et al., 1986), followed by selection with 100 mg/L kanamycin. The genomic DNA of the regenerated seedlings was isolated using a DNeasy Plant Mini Kit (Qiagen) and used as the template for PCR screens using gene-specific oligonucleotide primers (Table 1). Six to twenty-two independent PCR-positive transformants were generated per construct and analyzed by GC–MS.
Analysis of the transgene expression
Representative tobacco plants transformed with the constructs pYZ044, pYZ046, pYZ047 and pYZ048 were used for testing transgene expression. Total RNA was extracted from leaves of transgenic plants using RNeasy Plant Mini Kit (Qiagen) and treated with RNase-Free DNase (Qiagen) following the manufacturer’s instructions. First strand cDNA was synthesized using Superscript II reverse transcriptase (Invitrogen) utilizing 1 μg of total RNA as the template. The primers for each gene in RT-PCR were the same as the ones used for PCR screening of transgenic tobacco plants (Table 1). The thermal cycling conditions were as follows: 95 °C for 2 min, followed by 30 cycles of 95 °C for 30 s, 60 °C for 30 s and 72 °C for 30 s, and 1 cycle of 72 °C for 10 min.
In vivo metabolism of amorphane sesquiterpenoids
Young leaves from wild-type or transgenic tobacco plants (approximately 1–2 g) were excised from the plant before the bolting stage, and the petioles were trimmed under water and placed in 100 μL 25% (v/v) ethanol containing 0.5 mg of different sesquiterpenoids and allowed to take up the solution (approximately 1 h). The leaves were then placed in 0.5 mL water in a sealed desiccator overnight prior to work-up for GC–MS analysis.
Phytochemical analysis by GC–MS
Leaf material was collected and processed immediately or frozen at −80 °C. The leaf material (0.5–3 g) was placed in a 50 mL Teflon centrifuge tube with 20 mL of pentane and 10 μg octadecane, as internal standard, per g fresh weight of leaves. The material was homogenized with a Polytron and centrifuged at 2100 g for 10 min. The pentane layer was removed and concentrated under nitrogen to approximately 50 μL, then placed in a conical-bottom 250-μL GC vial insert (Agilent). Twenty microlitres of a N,O-bis-(trimethylsilyl)acetamide/pyridine solution (1:1 v/v; Sigma-Aldrich, St. Louis, MO, US) was added, and the mixture was concentrated to 20 μL for GC–MS analysis.
GC–MS analysis was accomplished using a 2 μL sample with a 40:1 split injection on an Agilent 6890 GC with oven temperature set to ramp from 125 °C to 300 °C at 5 °C/min and equipped with a DB-5MS column (30m, 0.25 mm ID, 0.25 μm film thickness, Agilent Technologies), coupled to an Agilent 5973 mass selective detector set to scan between 50 and 700 atomic mass units, under standard EI conditions (70 eV). Compounds were identified by comparison of retention time and mass spectra to standard compounds and quantified by their relative single-ion responses (amorphadiene, m/z 204, relative response factor (rrf) = 5.2; artemisinyl trimethylsilyl ether, m/z 202 rrf = 0.6 and dihydroartemisinyl trimethylsilyl ether, m/z 204, rrf = 7.1; artemisinic aldehyde, m/z 218, rrf = 5.8; dihydroartemisinic aldehyde, m/z 162, rrf = 19.4; trimethylsilyl artemisinate, m/z 216 rrf = 7.0; trimethylsilyl dihydroartemisinate, m/z 162, rrf = 19.4) in comparison with the internal standard (octadecane, m/z 254, rrf = 1.0).
We are grateful to the PBI DNA Technology and Plant Growth Facility for technical support, to Joan Krochko and David Taylor for reviewing the manuscript.