© 2006 Nature Publishing Group
Production of the antimalarial drug precursor
artemisinic acid in engineered yeast
Dae-Kyun Ro1*, Eric M. Paradise2*, Mario Ouellet1, Karl J. Fisher6, Karyn L. Newman1, John M. Ndungu3,
Kimberly A. Ho1, Rachel A. Eachus1, Timothy S. Ham4, James Kirby2, Michelle C. Y. Chang1, Sydnor T. Withers2,
Yoichiro Shiba2, Richmond Sarpong3& Jay D. Keasling1,2,4,5
Malaria is a global health problem that threatens 300–500 million
people and kills more than one million people annually1. Disease
control is hampered by the occurrence of multi-drug-resistant
strains of the malaria parasite Plasmodium falciparum2,3. Syn-
thetic antimalarial drugs and malarial vaccines are currently being
testing4,5. Artemisinin, a sesquiterpene lactone endoperoxide
extracted from Artemisia annua L (family Asteraceae; commonly
known as sweet wormwood), is highly effective against multi-
drug-resistant Plasmodium spp., but is in short supply and
unaffordable to most malaria sufferers6. Although total synthesis
of artemisinin is difficult and costly7, the semi-synthesis of
artemisinin or any derivative from microbially sourced artemisi-
nic acid, its immediate precursor, could be a cost-effective,
environmentally friendly, high-quality and reliable source of
artemisinin8,9. Here we report the engineering of Saccharomyces
cerevisiae to produce high titres (up to 100mgl21) of artemisinic
acid using an engineered mevalonate pathway, amorphadiene
synthase, and a novel cytochrome P450 monooxygenase
(CYP71AV1) from A. annua that performs a three-step oxidation
of amorpha-4,11-diene to artemisinic acid. The synthesized arte-
misinic acid is transported out and retained on the outside of the
engineered yeast, meaning that a simple and inexpensive purifi-
the engineered yeast is already capable of producing artemisinic
acid at a significantly higher specific productivity than A. annua,
artemisinic acid production to a level high enough to reduce
artemisinin combination therapies to significantly below their
We engineered artemisinic-acid-producing yeast in three steps, by
(1) engineering the farnesyl pyrophosphate (FPP) biosynthetic path-
way to increase FPP production and decrease its use for sterols, (2)
introducing the amorphadiene synthase gene (ADS) from A. annua
into the high FPP producer to convert FPP to amorphadiene, and (3)
of amorphadiene to artemisinic acid from A. annua and expressing it
in the amorphadiene producer (Fig. 1). The first committed reaction
in artemisinin biosynthesis is catalysed by ADS10, which has been
characterized and used for de novo production of amorphadiene in
Escherichia coli11. To test for improvements in FPP production, we
expressed ADS under the control of the GAL1 promoter on the
pRS425 plasmid (see Supplementary Information for details). Yeast
engineered with ADS alone produced a low quantity of amorpha-
diene (Fig. 2, strain EPY201, 4.4mgl21amorphadiene).
To increase FPP production in S. cerevisiae, the expression of
several genes responsible for FPPsynthesis was upregulated, and one
gene responsible for FPP conversion to sterols was downregulated.
All of these modifications to the host strain were made by chromo-
somal integration to ensure the genetic stability of the host strain.
Overexpression of a truncated, soluble form of 3-hydroxy-3-methyl-
glutaryl-coenzyme A reductase (tHMGR)12improved amorphadiene
production approximately fivefold (Fig. 2, strain EYP208). Down-
regulation of ERG9, which encodes squalene synthase (the first step
after FPP in the sterol biosynthetic pathway), using a methionine-
repressible promoter (PMET3)13increased amorphadiene production
an additional twofold (Fig. 2, strain EPY225). Although upc2-1, a
semi-dominant mutant allele that enhances the activity of UPC2
(a global transcription factor regulating the biosynthesis of sterols in
S. cerevisiae)14, had only a modest effect on amorphadiene pro-
duction when overexpressed in the EPY208 background (Fig. 2,
strain EPY210), the combination of downregulating ERG9 and
overexpressing upc2-1 increased amorphadiene production to
105mgl21(Fig. 2, strain EPY213). Integration of an additional
copy of tHMGR into the chromosome further increased amorpha-
diene production by 50% to 149mgl21(Fig. 2, strain EPY219).
Although overexpression of the gene encoding FPP synthase
(ERG20) had little effect on total amorphadiene production (Fig. 2,
strain EPY224), the specific production increased by about 10%
owing to a decrease in cell density. Combining all of these modifi-
cations resulted in a strain (EPY224) able to produce 153mgl21
amorphadiene, a sesquiterpene production level nearly 500-fold
higher than previously reported15.
To create a strain that produced artemisinic acid from amorpha-
diene, we isolated genes encoding enzymes responsible for oxidizing
amorphadiene to artemisinic acid in A. annua. Artemisinin is a
sesquiterpene lactone derivative, which is the most widespread and
characteristic class of secondary metabolites found in Asteraceae
(also known asCompositae)16. We hypothesized that plants belonging
to the Asteraceae family would share common ancestor enzymes for
the early steps in the biosynthesis of sesquiterpene lactones, and
therefore undertook a comparative genomic analysis of plants in the
Asteraceae family. Previous cell-free assays have indicated that a
cytochrome P450 monooxygenase (P450) catalyses the first regio-
retrieved P450-expressed-sequence tags (ESTs) from the Asteraceae
EST-database generated from two Asteraceae crops, sunflower and
lettuce (http://www.cgpdb.ucdavis.edu). Use of degenerate primers
highly specific to the Asteraceae CYP71 and CYP82 subfamilies (the
most abundant P450 subfamilies in Asteraceae) enabled the isolation
1California Institute of Quantitative Biomedical Research,2Department of Chemical Engineering,3Department of Chemistry,4Department of Bioengineering, and5Berkeley Center
for Synthetic Biology, Lawrence Berkeley National Laboratory, University of California, Berkeley, California 94720, USA.6Amyris Biotechnologies Inc., Emeryville, California
*These authors contributed equally to this work.
Vol 440|13 April 2006|doi:10.1038/nature04640
© 2006 Nature Publishing Group
fragments against sunflower and lettuce ESTs, weidentified a single A.
annua P450 gene fragment that had surprisingly high sequence
identity (85–88% at the amino-acid level) to ESTs of unknown
function from both sunflower and lettuce. Sequence identity of this
A. annua P450 fragment to other P450 fragments outside the Aster-
aceae family was much lower (,50% at the amino-acid level),
indicating that this P450 is highly conserved in three distantly related
genera in the Asteraceae family, but not in plants outside the
Asteraceae family. This P450 gene was therefore a good candidate
an open reading frame of 495 amino acids, was recovered from
A. annua. Phylogenetic analysis showed that CYP71AV1 shares a
close lineage with other P450s that catalyse the hydroxylation of
monoterpenoids (CYP71D13/18; ref. 19), sesquiterpenoids
(CYP71D20; ref. 20) or diterpenoids (CYP71D16; ref. 21), further
suggesting the potential involvement of this P450 in terpenoid
metabolism (Supplementary Information). For functional, hetero-
logous expression of CYP71AV1, its native redox partner, NADPH:
cytochrome P450 oxidoreductase (CPR), was also isolated from
A. annua, and its biochemical function was confirmed in vitro.
(Michaelis–Menten constants (Km) for cytochrome c and NADPH
were determined to be 4.3 ^ 0.7mM and 23.0 ^ 4.4mM (mean ^
s.d., n ¼ 3), respectively.)
Using A. annua CPR as a redox partner for CYP71AV1, we then
investigated whether CYP71AV1 could catalyse the conversion of
amorphadiene to more oxygenated products in vivo. The transgenic
yeast strain EPY224 was transformed with a vector harbouring CPR
and CYP71AV1 under the control of galactose-inducible promoters.
After galactose induction, the ether-extractable fraction of the yeast
culturemedium and cellpellet wereanalysed by gas chromatography
mass spectrometry (GC–MS). A single chromatographic peak
unique to EPY224 co-expressing CYP71AV1 and CPR was detected
in both the yeast culture medium and cell pellet, but was not present
in control yeast (EPY224 expressing CPR only). However, more than
95% of this novel compound was associated with the cell pellet. In
GC–MS analysis, the electron-impact mass spectrum and retention
time of this compound were identical to those of artemisinic
acid isolated from A. annua (Fig. 3). In a shake-flask culture,
32 ^ 13mgl21(mean ^ s.d., n ¼ 7) artemisinic acid was produced
from EPY224 expressing CYP71AV1 and CPR. Notably, the pathway
intermediates, artemisinic alcohol and artemisinic aldehyde, were
present at negligible levels in the culture medium and cell pellets of
EPY224 engineered with CYP71AV1 and CPR. (Artemisinic alcohol
was present at less than 5% of the artemisinic acid in the cell pellet,
and no artemisinic aldehyde was detected.)
Almost all (.96%) of the synthesized artemisinic acid was
removed from the cell pellet by washing with alkaline buffer (pH9
Tris-HCl buffer supplemented with 1.2M sorbitol), with less than
2% remaining in the washed cell pellet or culture medium. Thus, it
seems that artemisinic acid is efficiently transported out of yeast cells
but remains bound to the cell surface when it is protonated under
acidic culture conditions. We used this feature to develop a one-step
purification method: a single silica gel column chromatographic
separation of ether-extracted artemisinic acid from the wash buffer
routinely yielded .95% pure artemisinic acid. In a one-litre aerated
bioreactor, 115mg of artemisinic acid was produced, of which 76mg
was purified using this method. The1H and13C nuclear magnetic
resonance spectra of this yeast-derived artemisinic acid were identical
to those of artemisinic acid isolated from the leaves of A. annua, and
are consistent with previously reported values22,23. We can therefore
confirm that structurally authentic artemisinic acid is synthesized by
Figure 1 | Schematic representation of the engineered artemisinic acid
biosynthetic pathway in S. cerevisiae strain EPY224 expressing CYP71AV1
and CPR. Genes from the mevalonate pathway in S. cerevisiae that are
directly upregulated are shownin blue;those that are indirectlyupregulated
by upc2-1 expression are in purple; and the red line denotes repression of
ERG9 in strain EPY224. The pathway intermediates IPP, DMAPP and GPP
geranyl pyrophosphate, respectively. Green arrows indicate the biochemical
pathway leading from farnesyl pyrophosphate (FPP) to artemisinic acid,
which was introduced into S. cerevisiae from A. annua. The three oxidation
steps converting amorphadiene to artemisinic acid by CYP71AV1 and CPR
S. cerevisiae strains are described in the text. Cultures were sampled after
144h of growth, and amorphadiene levels were quantified. Data, shown as
total production, are mean ^ s.d. (n ¼ 3).
NATURE|Vol 440|13 April 2006
© 2006 Nature Publishing Group
transgenic yeast de novo. The transgenic yeast produced artemisinic
acid at a biomass fraction comparable to that produced by A. annua
(4.5% dry weight in yeast compared to 1.9% artemisinic acid and
0.16% artemisinin in A. annua) but over a much shorter time (4–5
days for yeast versus several months for A. annua). As such, the
specific productivity of the engineered yeast strain is nearly two
orders of magnitude greater than A. annua.
Three-step oxidations by P450 enzymes have been previously
reported in plant hormone gibberellin biosynthetic pathways24,25.
We conducted in vitro enzyme assays to identify whether CYP71AV1
catalyses all three oxidation reactions from amorphadiene to arte-
misinic acid. Microsomes were isolated from S. cerevisiae strain
YPH499 expressing either CPR alone or CPR and CYP71AV1, and
incubated with pathway intermediates (amorphadiene, artemisinic
alcohol or artemisinic aldehyde) (Fig. 4). Microsomes from the CPR
control did not catalyse the conversion of any pathway intermediate
to more oxidized products, whereas efficient conversion of amor-
phadiene, artemisinic alcohol and artemisinic aldehyde to the final
product artemisinic acid was detected in microsomes containing
CYP71AV1 and CPR. These in vitro assays demonstrate unambigu-
ously that recombinant CYP71AV1 is able to catalyse three oxidation
reactions at the C12 position of amorphadiene. Previous in vitro
enzyme assays using A. annua protein extract have suggested that
soluble alcohol and aldehyde dehydrogenases and a C11,13 double-
bond reductase (which acts on the aldehyde) are involved in
artemisinin biosynthesis17. Although we cannot exclude a catalytic
role for additional alcohol and aldehyde dehydrogenases in artemi-
sinin synthesis in A. annua, the efficient in vivo conversion of
CYP71AV1 indicates that the membrane-bound, multifunctional
CYP71AV1 is a key contributor to artemisinin biosynthesis.
In summary, we have created a strain of S. cerevisiae capable of
producing high levels of artemisinic acid by engineering the FPP
biosynthetic pathway to increase FPP production and by expressing
amorphadiene synthase, a novel cytochrome P450 and its redox
yielding chemistry for the conversion of artemisinic acid to artemi-
sinin or any other derivative that might be desired8,9, microbially
produced artemisinic acid is a viable source of this potent family of
antimalarial drugs. Upon optimization of product titres, a conserva-
tive analysis suggests that artemisinin combination therapies could
Information). In addition to cost savings, this bioprocess should not
be subject to factors like weather or political climates that may affect
plant cultivation. Furthermore, artemisinic acid from a microbial
source can be extracted using an environmentally friendly process
without worrying about contamination by other terpenes that are
produced by plants, thereby increasing the ease with which it can be
produced while reducing purification costs.
acid by recombinant
Figure 3 | GC–MS analysis of artemisinic acid produced from A. annua and
transgenicyeast. a, Cell pellets fromS. cerevisiaestrain EPY224expressing
CPR or CPR and CYP71AV1 were washed using an alkaline buffer followed
by acidification and ether extraction. Artemisinic acid was extracted from
A. annua leaves using hexane. Methyl esters of both samples were prepared
with trimethylsilyl-diazomethane before GC–MS analysis. The internal
standard (IS) is the methyl ester of 4-octylbenzoic acid. b, c, Mass spectra
and retention times of artemisinic acid from yeast (b) and A. annua (c).
RT, retention time (in min).
Figure 4 | In vitro enzyme assays for CYP71AV1 activities. Microsomes
were isolated from S. cerevisiae strain YPH499 expressing CPR (control) or
CPR and CYP71AV1 (CYP71AV1). a–c, For each enzyme assay, 10mM
amorphadiene (a), 25mM artemisinic alcohol (b) or 25mM artemisinic
aldehyde (c) was used. Chromatographic peaks for the substrates used are
indicated with asterisks. Ether-extractable fractions were derivatized and
analysed by GC–MS in selective ion mode (m/z: 121, 189, 204, 218, 220 and
248). Enzymatic products are as indicated: 1, artemisinic alcohol (retention
time 13.20min); 2, artemisinic aldehyde (retention time 11.79min);
3, artemisinic acid (retention time 13.58min, detected as methyl ester).
NATURE|Vol 440|13 April 2006
© 2006 Nature Publishing Group Download full-text
of S. cerevisiae EPY strains, the cloning of CYP71AV1 and CPR, and the semi-
synthesis of artemisinic alcohol and artemisinic aldehyde are provided in
Chemicals and plant material. Authentic artemisinic acid was purchased from
Apin Chemicals or extracted from A. annua leaves with hexane, as described in
Supplementary Information. A. annua plants were started from seeds (Sandeman
Seeds) and grown in a greenhouse at the University of California, Berkeley.
GC–MS analysis of amorphadiene. Amorphadiene production by the various
strains was measured by GC–MS using a dodecane layer to trap volatile
amorphadiene (see Supplementary Information for details). Amorphadiene
(90% pure) was prepared by fermentation using an E. coli strain11, and was used
to construct a standard curve to determine amorphadiene production levels.
In vivo production, purification and chemical analysis of artemisinic acid.
Pre-cultured EPY224 strains transformed with pESCURA::CPR or pESCUR-
A::CPR/CYP71AV1wereinoculatedat anabsorbanceof0.05at 600nm(A600) in
25ml synthetic defined medium lacking histidine, leucine, methionine and
uracil, and supplemented with 0.2% dextrose, 1.8% galactose and 1mM
methionine. After 120h of culture at 308C, the cells were centrifuged and the
cell pellet was washed using 50mM Tris-HCl buffer (pH9). The buffer was
acidified to pH2 using 2M HCl, and extracted with ethyl acetate spiked with
4-octyl benzoic acid (10mgml21). The extracts were derivatized by 50ml of 2M
TMS-diazomethane (Aldrich) with 10% methanol. For qualitative analysis by
GC–MS, the product was purified by silica gel column chromatography eluted
with ether and pentane (1:1).
Products were analysed using a gas chromatography mass spectrometer
(70eV, Agilent Technologies) equipped with a DB5 capillary column (0.25mm
internal diameter £ 0.25mm £ 30m, J&W Scientific). The gas chromatography
oven programme used was 808C (held for 2min), 208Cmin21ramp to 1408C,
product separation by a 58Cmin21increment upto2208C.Forquantificationby
gas chromatography-flame ionization detection, samples were analysed without
column purification using the same gas chromatography oven programme.
Fermentation and product analyses. A one-litre bioreactor (New-Brunswick
an A600of 5.0. The dissolved oxygen level was maintained at 40% by altering
acid was removed from the cell pellet by an alkaline wash as before, and purified
through a silica column eluted with 78% hexane, 20% ethylacetate and 2% acetic
1H and13C NMR using a 500MHz NMR spectrometer (Bruker DRX-500) in the
College of Chemistry NMR Facility at the University of California.
Invitro enzyme assays. A one-litre culture of S. cerevisiae YPH499 transformed
with pESCURA::CPR or pESCURA::CPR/CYP71AV1 was induced with 2%
followed by an additional ultracentrifugation step to remove cytosolic protein
contamination, as previously described26. Approximately 500mg of total micro-
somal protein was used in a 1-ml reaction containing 100mM potassium
phosphate buffer pH7.5, 10 or 25mM substrate, 100mM NADPH and an
NADPH regeneration system (5mM glucose-6-phosphate and two units of
glucose-6-phosphate dehydrogenase). Reactions were incubated for 2h at 248C
with gentle agitation, acidified to pH2, and extracted with ethyl ether. Products
were separated using the same gas chromatography oven programme as above.
Selective ion mode (SIM), including six ions characteristic to the products (121,
189, 204, 218, 220 and 248), was used for detection.
Received 22 December 2005; accepted 9 February 2006.
World Health Organization. World Malaria Report 2005. (WHO, Geneva, 2005).
Korenromp, E. L., Williams, B. G., Gouws, E., Dye, C. & Snow, R. W.
Measurement of trends in childhood malaria mortality in Africa: An
assessment of progress toward targets based on verbal autopsy. Lancet Infect.
Dis. 3, 349– -358 (2003).
Marsh, K. Malaria disaster in Africa. Lancet 352, 924 (1998).
Hoffman, S. Save the children. Nature 430, 940– -941 (2004).
Vennerstrom, J. L. et al. Identification of an antimalarial synthetic trioxolane
drug development candidate. Nature 430, 900– -904 (2004).
Enserink, M. Infectious diseases: Source of new hope against malaria is in short
supply. Science 307, 33 (2005).
Schmid, G. & Hofheinz, W. Total synthesis of Qinghaosu. J. Am. Chem. Soc.
105, 624– -625 (1983).
Acton, N. & Roth, R. J. On the conversion of dihydroartemisinic acid into
artemisinin. J. Org. Chem. 57, 3610– -3614 (1992).
Haynes, R. K. & Vonwiller, S. C. Cyclic peroxyacetal lactone, lactol and
ether compounds. US patent 5,310,946 (1994).
10. Mercke, P., Bengtsson, M., Bouwmeester, H. J., Posthumus, M. A. & Brodelius,
P. E. Molecular cloning, expression, and characterization of amorpha-4,11-diene
synthase, a key enzyme of artemisinin biosynthesis in Artemisia annua L. Arch.
Biochem. Biophys. 381, 173– -180 (2000).
11.Martin, V. J., Pitera, D. J., Withers, S. T., Newman, J. D. & Keasling, J. D.
Engineering a mevalonate pathway in Escherichia coli for production of
terpenoids. Nature Biotechnol. 21, 796– -802 (2003).
12. Donald, K., Hampton, R. & Fritz, I. Effects of overproduction of the catalytic
domain of 3-hydroxy-3-methylglutaryl coenzyme A reductase on squalene
synthesis in Saccharomyces cerevisiae. Appl. Environ. Microbiol. 63, 3341– -3344
13. Gardner, R. G. & Hampton, R. Y. A highly conserved signal controls
degradation of 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA)
reductase in eukaryotes. J. Biol. Chem. 274, 31671– -31678 (1999).
14. Davies, B. S. J., Wang, H. S. & Rine, J. Dual activators of the sterol biosynthetic
pathway of Saccharomyces cerevisiae: Similar activation/regulatory domains but
different response mechanisms. Mol. Cell. Biol. 25, 7375– -7385 (2005).
15. Jackson, B. E., Hart-Wells, E. A. & Matsuda, S. P. T. Metabolic engineering to
produce sesquiterpenes in yeast. Org. Lett. 5, 1629– -1632 (2003).
16. Seaman, F. C. Sesquiterpene lactones as taxonomic characters in the
Asteraceae. Bot. Rev. 48, 121– -592 (1982).
17. Bertea, C. M. et al. Identification of intermediates and enzymes involved in the
early steps of artemisinin biosynthesis in Artemisia annua. Planta Med. 71,
40– -47 (2005).
18. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local
alignment search tool. J. Mol. Biol. 215, 403– -410 (1990).
19. Lupien, S., Karp, F., Wildung, M. & Croteau, R. Regiospecific cytochrome P450
limonene hydroxylases from mint (Mentha) species: cDNA isolation,
characterization, and functional expression of (2)-4S-limonene-3-hydroxylase
and (2)-4S-limonene-6-hydroxylase. Arch. Biochem. Biophys. 368, 181– -192
20. Ralston, L. et al. Cloning, heterologous expression, and functional
characterization of 5-epi-aristolochene-1,3-dihydroxylase from tobacco
(Nicotiana tabacum). Arch. Biochem. Biophys. 393, 222– -235 (2001).
21. Wang, E. et al. Suppression of a P450 hydroxylase gene in plant trichome
glands enhances natural-product-based aphid resistance. Nature Biotechnol. 19,
371– -374 (2001).
22. Elmarakby, S. A., El-Feraly, F. S., Elsohly, H. N., Croom, E. M. & Hufford, C. D.
Microbiological transformations of artemisinic acid. Phytochemistry 27,
3089– -3091 (1988).
23. Kim, S., Han, J. & Lim, Y. Revised assignment of1H-NMR signals of artemisinic
acid. Planta Med. 62, 480– -481 (1996).
24. Helliwell, C. A., Poole, A., Peacock, W. J. & Dennis, E. S. Arabidopsis
ent-kaurene oxidase catalyzes three steps of gibberellin biosynthesis. Plant
Physiol. 119, 507– -510 (1999).
25. Helliwell, C. A., Chandler, P. M., Poole, A., Dennis, E. S. & Peacock, W. J. The
CYP88A cytochrome P450, ent-kaurenoic acid oxidase, catalyzes three steps
of the gibberellin biosynthesis pathway. Proc. Natl Acad. Sci. USA 98,
2065– -2070 (2001).
26. Pompon, D., Louerat, B., Bronine, A. & Urban, P. Yeast expression of animal
and plant P450s in optimized redox environments. Methods Enzymol. 272,
51– -64 (1996).
Supplementary Information is linked to the online version of the paper at
Acknowledgements We thank D. Nelson for assigning an official CYP number,
and D. Ockey and D. McPhee for artemisinic acid isolation from the A. annua
plant. We are also grateful to N. A. Da Silva for pd-UB, R. Y. Hampton for
pRH127-3 and pRH973, and J. Rine for pBD33 and pBD36. We thank
R. Michelmore and other members in the Compositae Genomics Project for the
support of this project. This research was conducted under the sponsorship of
the Institute for OneWorld Health, through the generous support of The Bill and
Melinda Gates Foundation, and through funding from the Akibene Foundation,
the United States Department of Agriculture, a University of California
Discovery Grant, the Diversa Corporation and the National Science Foundation.
Author Contributions D.-K.R., E.M.P. and J.D.K. designed the project and
experiments. D.-K.R., E.M.P., Y.S., M.C.Y.C., S.T.W. and J.K. performed
experiments. K.J.F. conducted NMR analysis of artemisinic acid. J.M.N. and R.S.
semi-synthesized artemisinic alcohol and artemisinic aldehyde. T.S.H. performed
bioinformatics analysis of the Compositae EST-database. M.O., R.A.E. and
K.A.H. provided technical assistance. D.-K.R., E.M.P., K.L.N. and J.D.K. wrote the
paper. All authors discussed the results and commented on the manuscript.
Author Information Artemisia annua CYP71AV1 and CPR gene sequence
information has been deposited in GenBank under accession numbers
DQ268763 and DQ318192, respectively. Reprints and permissions information is
available at npg.nature.com/reprintsandpermissions. The authors declare
competing financial interests: details accompany the paper at www.nature.com/
nature. Correspondence and requests for materials should be addressed to
NATURE|Vol 440|13 April 2006