The Plant Journal (2012) doi: 10.1111/j.1365-313X.2012.04949.

The hexanoyl-CoA precursor for cannabinoid biosynthesis is

formed by an acyl-activating enzyme in Cannabis sativa
trichomes
Jake M. Stout1,2, Zakia Boubakir1,2, Stephen J. Ambrose1, Randy W. Purves1 and Jonathan E. Page1,2,*
1
Plant Biotechnology Institute, National Research Council of Canada, 110 Gymnasium Place, Saskatoon,
SK, S7N 0W9 Canada, and
2
Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, SK, S7N 5E2 Canada

Received 26 October 2011; revised 10 February 2012; accepted 15 February 2012.

*
For correspondence (e-mail [email protected]).

SUMMARY
The psychoactive and analgesic cannabinoids (e.g. D9-tetrahydrocannabinol (THC)) in Cannabis sativa are
formed from the short-chain fatty acyl-coenzyme A (CoA) precursor hexanoyl-CoA. Cannabinoids are
synthesized in glandular trichomes present mainly on female flowers. We quantified hexanoyl-CoA using
LC-MS/MS and found levels of 15.5 pmol g)1 fresh weight in female hemp flowers with lower amounts in
leaves, stems and roots. This pattern parallels the accumulation of the end-product cannabinoid, cannabidiolic
acid (CBDA). To search for the acyl-activating enzyme (AAE) that synthesizes hexanoyl-CoA from hexanoate,
we analyzed the transcriptome of isolated glandular trichomes. We identified 11 unigenes that encoded
putative AAEs including CsAAE1, which shows high transcript abundance in glandular trichomes. In vitro
assays showed that recombinant CsAAE1 activates hexanoate and other short- and medium-chained fatty
acids. This activity and the trichome-specific expression of CsAAE1 suggest that it is the hexanoyl-CoA
synthetase that supplies the cannabinoid pathway. CsAAE3 encodes a peroxisomal enzyme that activates a
variety of fatty acid substrates including hexanoate. Although phylogenetic analysis showed that CsAAE1
groups with peroxisomal AAEs, it lacked a peroxisome targeting sequence 1 (PTS1) and localized to the
cytoplasm. We suggest that CsAAE1 may have been recruited to the cannabinoid pathway through the loss of
its PTS1, thereby redirecting it to the cytoplasm. To probe the origin of hexanoate, we analyzed the trichome
expressed sequence tag (EST) dataset for enzymes of fatty acid metabolism. The high abundance of transcripts
that encode desaturases and a lipoxygenase suggests that hexanoate may be formed through a pathway that
involves the oxygenation and breakdown of unsaturated fatty acids.

Keywords: Cannabis sativa, marijuana, cannabinoid, acyl-CoA, acyl-activating enzyme, hexanoate.

INTRODUCTION
Cannabis sativa L. (hemp, marijuana; Cannabaceae) is well that forms olivetolic acid (OA) (Dewick, 2009). Although OA
known for its content of cannabinoids, a group of more than formation has been investigated (Raharjo et al., 2004; Marks
70 natural products with unique pharmacological properties et al., 2009) and a candidate type III PKS (olivetol synthase,
(Elsohly and Slade, 2005). Major cannabinoids include OLS) identified (Taura et al., 2009), this step is unclear. OA is
D9-tetrahydrocannabinol (THC), the compound responsible geranylated to yield cannabigerolic acid (CBGA) (Fellermeier
for the psychoactive and therapeutic effects of marijuana and Zenk, 1998). The oxidocyclase tetrahydrocannabinolic
consumption, and cannabidiol (CBD), which has neuro- acid synthase (THCAS) converts CBGA to THCA while can-
protective properties (Gaoni and Mechoulam, 1964; Mec- nabidiolic acid synthase (CBDAS) forms CBDA (Sirikantara-
houlam et al., 2002). Cannabinoids are prenylated mas et al., 2004; Taura et al., 2007). THCA is the major
polyketides of mixed biosynthetic origin that are synthesized cannabinoid in drug strains of cannabis while CBDA pre-
from fatty acid and isoprenoid precursors (Figure 1). Hexa- dominates in hemp forms grown for fibre or seed. The
noyl-CoA, derived from the short-chain fatty acid hexanoate, cannabinoid acids occur in planta and undergo decarbox-
is used as a primer for a polyketide synthase (PKS) enzyme ylation to their neutral forms upon heating or pyrolysis.

ª 2012 The Authors 1

The Plant Journal ª 2012 Blackwell Publishing Ltd
2 Jake M. Stout et al.

Fatty acid Figure 1. The cannabinoid biosynthetic pathway

biosynthesis that leads to the major cannabinoids, D9-tetrahy-
drocannabinolic acid (THCA) and cannabidiolic
acid (CBDA). The carbon chain of the fatty acid
precursor hexanoate is labelled in bold to show
its incorporation into cannabinoids.
O Hexanoic acid
HO

Hexanoyl-CoA CoA
Synthetase

O OH
3 x malonyl CoA COOH Olivetolic acid
CoA-S
Hexanoyl-CoA Polyketide synthase HO

Aromatic Geranyldiphosphate
prenyltransferase

OH
COOH
Cannabigerolic acid
HO

THCA CBDA
synthase synthase

OH OH
COOH COOH

O HO

Δ9-Tetrahydrocannabinolic acid Cannabidiolic acid

Acyl-CoA thioesters such as hexanoyl-CoA are formed by We sought to clarify the formation of hexanoyl-CoA as a
members of the acyl-activating enzyme (AAE) superfamily precursor for cannabinoid biosynthesis. LC-MS/MS analysis
that activate carboxylic acids through an adenylate interme- established that hexanoyl-CoA is present in cannabis flow-
diate (Schmelz and Naismith, 2009). In plants, the substrates ers. To identify the enzyme(s) responsible for hexanoyl-CoA
of AAEs include phenylpropanoids, fatty acids and jasmo- synthesis, we used transcriptome analysis of a trichome-
nate precursors. 4-Coumarate:CoA ligase (4CL), an enzyme specific cDNA library from female flowers to identify 11
involved in phenylpropanoid metabolism, and the long- putative AAE genes. Here we report that two encode
chain acyl-CoA synthetases, are perhaps the best character- hexanoyl-CoA synthetases and provide evidence that
ized of the plant AAEs (Shockey et al., 2002; Hu et al., 2010). CsAAE1 is a cytoplasmic enzyme that functions in cannab-
The genomic organization and biochemical diversity of inoid biosynthesis. CsAAE3 encodes a peroxisomal acyl-
AAEs have been recently reviewed (de Souza et al., 2008; CoA synthetase that activates a variety of substrates that
Shockey and Browse, 2011). include hexanoate.
The primary site of cannabinoid biosynthesis is glandular
trichomes that form on female flowers (Lanyon et al., 1981). RESULTS
Glandular trichomes are also found at lower densities on the
Hexanoyl-CoA and CBDA are present in female cannabis
male flowers, where a row of trichomes is located on the
flowers
inner surfaces of anthers (Dayanandan and Kaufman, 1976).
Analysis of the transcriptome of glandular trichomes from Hexanoyl-CoA has not been identified previously in canna-
mint, hop and other plants through EST sequencing has bis. In this study we sought to measure this metabolite in
been invaluable for discovering enzymes involved in spe- tissues that were actively synthesizing cannabinoids. We
cialized metabolic pathways (Lange et al., 2000; Nagel et al., used plants of the hemp cultivar ‘Finola’, which has a can-
2008; Wang et al., 2008; Dai et al., 2010). A recent EST nabinoid profile typical of hemp varieties (high CBDA/low
analysis of cannabis trichomes has provided data on their THCA) (Figure S1). The analysis was performed initially
metabolic functions (Marks et al., 2009). using trichome cells, but measurement of acyl-CoAs was not

ª 2012 The Authors

The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
 A cytoplasmic acyl-activating enzyme involved in cannabinoid biosynthesis 3

(a) Hexanoyl-CoA standard (multiple m/z channels) Table 1 Quantity of hexanoyl-CoA and CBDA in various hemp
Benzoyl
‘Finola’ tissues
2.5x106 -CoA
Hexanoyl-CoA CBDA
Tissue (pmoles g)1 FW)a (lmoles g)1 FW)a
2.0x106
Female flowers 15.5  5.7 2.4  0.8
Ion count

1.5x106 Leaves 4.0  0.6 0.5  0.2

Stems 2.2  1.2 0.05  0.04
1.0x106 Roots 1.5  0.7 0.004  0.001

a
0.5x106 Hexanoyl Values represent mean  standard deviation (n = 4)
-CoA

0 ers. A trichome-specific cDNA library from a marijuana

5 6 7 8 9 10 strain was sequenced to produce 9157 expressed sequence
tags (ESTs) that assembled into 4113 unique sequences
(b) Hexanoyl-CoA in flowers (m/z 866.2 → 359.2)
(1227 contigs, 2886 singletons). Unigenes were annotated
50 000 by comparison with genes on the UniProt database and
using the blastx program. Transcripts that encoded can-
40 000 nabinoid biosynthetic enzymes were abundant in the EST
dataset: OLS (Taura et al., 2009) ranked first with 340 contig
Ion count

30 000
members and THCAS (Sirikantaramas et al., 2004) ranked
11th with 48 contig members (Table 2). Genes for enzymes
20 000
of the 2-C-methyl-D-erythritol 4-phosphate (MEP) pathway,
10 000
which produces the geranyl side chain of the cannabinoids,
were highly expressed (Fellermeier et al., 2001; Phillips et
0 al., 2008). We also observed that transcripts that represented
5 6 7 8 9 10 several desaturases and enzymes of fatty acid degradation,
Time (min) specifically a lipoxygenase and an acetoacetyl-CoA thiolase,
were represented prominently in the trichome dataset.
Figure 2. Liquid chromatography (LC) mass spectrometry MS/MS analysis of
Cannabis AAE genes (CsAAEs) were identified by utilizing
hexanoyl-CoA in female flowers of the hemp cultivar ‘Finola’. (a) Hexanoyl-
CoA detected using m/z 866.2 > 359.2 and benzoyl-CoA internal standard Arabidopsis AAE sequences to query the assembled canna-
detected with m/z 872.2 > 365.1. (b) Hexanoyl-CoA in cannabis flower extract bis ESTs using TBLASTN software. Eleven unique CsAAE
detected using m/z 866.2 > 359.2.
contigs and singletons were identified and named according
to their abundance in the cDNA library (Table 3). We
considered two contigs to be strong candidates to encode
reproducible, possibly because of losses during trichome a hexanoyl-CoA synthetase. CsAAE1 was the most highly
isolation. Instead we found that hexanoyl-CoA was readily expressed AAE based on transcript abundance (42 ESTs),
detectable when whole flowers with trichomes were ana- and was the 16th most abundant transcript in the EST
lyzed (Figure 2). After adjustment for recoveries using the dataset. CsAAE1 also had the highest EST count of the four
internal standards, we quantified the hexanoyl-CoA pools in AAE genes present in the cannabis trichome EST dataset
female flowers, leaves, stems and roots. To compare these reported by Marks et al. (2009). The second candidate,
pool sizes with those of the final pathway product, we also CsAAE3, was identified due to its homology to an Arabid-
quantified the content of the cannabinoid pathway end- opsis 4CL-like gene (At4g05160) that activates several fatty
product CBDA in the same tissues. As shown in Table 1, the acids, including hexanoate (Schneider et al., 2005). Of the
levels of hexanoyl-CoA and CBDA are highest in female nine other AAEs, only CsAAE2, which was not full length,
flowers and substantially lower in the other tissues that we showed high EST counts. This protein was most similar to
analyzed. the Arabidopsis LACS4 AAE (At4g23850), which activates
long-chain fatty acids (Shockey et al., 2003).
Cannabinoid pathway transcripts are abundant in cannabis
trichomes CsAAE1 and CsAAE3 encode hexanoyl-CoA synthetases

We analyzed the transcriptome of cannabis trichome cells Of the eight full-length AAE cDNAs, CsAAE1, CsAAE3,
for genes that encoded enzymes of cannabinoid and terpe- CsAAE6, CsAAE8 and CsAAE10 yielded soluble protein
noid biosynthesis, with a focus on AAE transcripts. The when expressed in Escherichia coli (Figure S2). CsAAE5,
trichomes were obtained exclusively from female flowers, CsAAE7 and CsAAE9 formed inclusion bodies and were not
which are larger and have more trichomes than male flow- analyzed further. The hexanoyl-CoA synthetase activity of

ª 2012 The Authors

The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
4 Jake M. Stout et al.

Table 2 Transcripts in a cannabis trichome-specific cDNA library as Table 3 Acyl-activating enzymes identified in the cDNA library from
determined by EST sequencing cannabis trichome secretory cells

Description UniProt # # ESTs % ESTs Contig Closest Arabidopsis

Name size homolog, accession no. % Identity
Cannabinoid pathway
Olivetol synthase/polyketide B1Q2B6 339 3.70 CsAAE1 42 AtAAE17, At5g23050 65
synthase CsAAE2a 9 AtLACS4, At4g23850 74
Tetrahydrocannabinolic acid Q8GTB6 48 0.52 CsAAE3 5 At4CL-like, At4g05160 73
synthase CsAAE4a 2 AtAAE18, At1g55320 66
CsAAE1 B9RE68 42 0.46 CsAAE5 2 AtAAE6, At5g16340 63
Fatty acid pathway CsAAE6 1 AtACN1, At3g16910 74
Delta12-oleic acid desaturase Q6RXX0 62 0.68 CsAAE7 1 AtAAE2, At2g17650 64
Delta12-oleic acid desaturase Q41305 53 0.58 CsAAE8 1 AtAAE3, At3g48990 76
Acetoacetyl-CoA thiolase Q2L8A7 28 0.31 CsAAE9 1 AtAAE1, At1g20560 72
Lipoxygenase C4NZX3 25 0.27 CsAAE10 1 AtAAE13, At3g16170 72
Delta12-oleic acid desaturase Q6RXX0 22 0.24 CsAAE11a 1 4CL1, At1g51680 60
Delta12-oleic acid desaturase Q6RXX0 22 0.24 CsAAE12b n/a AtAAE17, At5g23050 67
Hydroperoxide lyase B9IAH7 3 0.03 CsAAE13b n/a AtAAE18, At1g55320 68
MEP pathway CsAAE14b n/a AtAAE18, At1g55320 69
4-Hydroxy-3-methylbut-2-enyl B9RZD3 185 2.02 CsAAE15b n/a ACS, At5g36680 79
diphosphate reductase
a
Isopentenyl pyrophosphate Q6EJD1 48 0.52 cDNA was not full length. Identity scores were calculated using the
isomerase incomplete amino acid sequence.
b
4-Hydroxy-3-methylbut-2-en- A9ZN14 33 0.36 Sequence identified in the Cannabis sativa genome (van Bakel et al.,
1-yl diphosphate synthase 2011).
1-Deoxy-D-xylulose-5- B9RSX4 15 0.16
phosphate synthase
1-Deoxy-D-xylulose-5-phos B9RB24 15 0.16 erties of CsAAE1 and CsAAE3 were measured for CoA and
phate reductoisomerase representative short (butanoate), medium (hexanoate and
2-C-methyl-D-erythritol Q94A03 1 0.01
decanoate), and long-chain (palmitate) fatty acids (Table 4,
4-phosphate
cytidylyltransferase Figure S4). High CoA concentrations inhibited CsAAE1.
4-(Cytidine 5¢-diphospho)-2- A9ZN11 6 0.07 Using a non-linear regression substrate inhibition model,
C-methyl-D-erythritol kinase we estimate that the Ki of CsAEE1 5.101  1.8 mM. CoA did
2-C-methyl-D-erythritol Q0Q5B3 3 0.03 not inhibit CsAAE3 at the concentrations tested. Decanoic
2,4-cyclodiphosphate
acid inhibited CsAAE3 (Ki = 120.8  47.9 lM) and slightly
synthase
Terpenoid biosynthesis inhibited CsAAE1 at concentrations above 4 mM (Ki not
Limonene synthase A7IZZ1 65 0.71 measured).
a-Pinene synthase A7IZZ2 38 0.41
Monoterpene synthase B6SCF4 35 0.38 CsAAE1 transcript is abundant in glandular trichomes
Sesquiterpene synthase B6SCF5 31 0.34
CsAAE1 and CsAAE3 transcripts were measured in different
Sesquiterpene synthase B6SCF5 27 0.29
organs of ‘Finola’ plants. Total RNA was isolated from roots,
The function of the contigs was predicted using BLASTX comparison stems, leaves, female flowers (before and after trichomes
with sequences in the UniProt database (29 November 2009 version). removal), male flowers and trichome cells. CsAAE1 tran-
script levels were high in trichome cells and low in the other
the purified recombinant proteins was tested with hexano- plant parts (Figure 5). The trichome-specific expression was
ate, CoA, ATP and MgCl2 (Figure 3). CsAAE1 and CsAAE3 further supported by the finding that removal of trichomes
formed hexanoyl-CoA, whereas the other three enzymes from the female flowers reduced CsAAE1 transcript levels.
were inactive. Both CsAAE1 and CsAAE3 exhibited temper- The expression pattern of CsAAE1, including the reduced
ature and pH optima of 40C and pH 9, respectively (Fig- expression in female flowers with trichomes removed (FF))
ure S3). CsAAE1 activity was highest with Mg2+. Although compared with intact female flowers (FF+), is similar to that
CsAAE3 preferred Co2+, Mg2+ was used for further assays. of the known cannabinoid pathway gene CBDAS. CsAAE3,
To test the substrates that CsAAE1 and CsAAE3 can which was present in all tissues (Figure 5), did not show the
activate, we assayed a range of carboxylic acids using same expression pattern as CsAAE1 or CBDAS. CsAAE1
luciferase-based assay system (Schneider et al., 2005). transcripts were more abundant in trichome cells than
CsAAE1 utilized hexanoate, heptanoate, octanoate and CsAAE3, a finding that is also reflected in their EST counts
nonanoate, with a slight preference for hexanoate (Figure 4). (Table 3). The residual CsAAE1 and CBDAS expression in
In contrast, CsAAE3 accepted short-, medium-, and long- FF) samples may reflect in the incomplete removal of tri-
chain fatty acids. To further this analysis, the kinetic prop- chome secretory cells from these tissues, or the possibility
ª 2012 The Authors
The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
 A cytoplasmic acyl-activating enzyme involved in cannabinoid biosynthesis 5

Figure 3. Liquid chromatography (LC) mass CsAAE1 CsAAE3

spectrometry MS/MS analysis of CsAAE1 and
CsAAE3 enzyme assays to produce hexanoyl- Enzyme assay
CoA. Purified recombinant enzymes were as- 8.0 x 106 8.0 x 106 Hexanoyl-CoA
sayed with hexanoate, CoA, Mg2+ and ATP.
Hexanoyl-CoA
Products were identified by multiple reaction 6.0 x 10 6
6.0 x 106
monitoring (MRM) transitions and retention time
compared with an authentic hexanoyl-coenzyme 4.0 x 106 4.0 x 106
A (CoA) standard. Negative controls with boiled

Ion count (m/z 866.2>359.2)

enzymes show the lack of hexanoyl-CoA synthe-
2.0 x 106 2.0 x 106
sis.

0 0
0 2 4 6 8 10 0 2 4 6 8 10

Enzyme assay with boiled enzyme

8.0 x 106 8.0 x 106

6.0 x 106 6.0 x 106

4.0 x 106 4.0 x 106

2.0 x 106 2.0 x 106

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Time (min)

that some cannabinoid biosynthesis occurs in internal floral contain targeting sequences and is therefore likely to be in
tissues. the cytosol. The co-infiltration of YFP:CsAAE1 and OLS:CFP
resulted in overlapping fluorescence signals for YFP and
CsAAE1 and CsAAE3 are localized to different subcellular
CFP, indicating that these proteins are cytosolic (Figure 6).
compartments
Phylogenic relationships of cannabis AAEs
Many plant AAEs are located in the peroxisome where they
function to activate substrates for b-oxidation (Shockey et We examined the evolutionary relationships between the
al., 2003; Souza et al., 2008). Peroxisomal proteins either CsAAEs and those from other plants. We first generated a
contain a C-terminal peroxisome targeting sequence 1 maximum likelihood phylogenetic tree using cannabis and
(PTS1) of 12 amino acids ending with SKL (or variants Arabidopsis AAE protein sequences (Shockey et al., 2003;
thereof), or an N-terminal peroxisome targeting sequence 2 Shockey and Browse, 2011) (Figure S5). CsAAE1 is a mem-
(PTS2) sequence with a RLx5HL motif (Reumann, 2004). ber of clade II, which includes AtAAE17 (At5g23050), AtA-
CsAAE1 and CsAAE3 were analyzed for PTSs using a AE18 (At1g55320) and a plastidic acetyl-CoA synthetase
BLOCKS-based algorithm (http://www.peroxisomedb.org). (ACS, At5g36880). CsAAE3 grouped with the clade V pro-
The C-terminal sequence of CsAAE1 is not predicted to have teins, which includes enzymes that accept a range of fatty
a functional PTS1 signal (e-value 0.32), whereas CsAAE3 acids, hormone precursors and phenylpropanoids as sub-
contains a non-canonical PTS1 sequence (e-value 0.04). strates (Koo et al., 2006; Kienow et al., 2008; Souza et al.,
Neither protein contains a PTS2 sequence. 2008). All clade V AAEs possess canonical PTS1 signals and
We sought to verify these predictions by co-localizing are predicted to be located in the peroxisome.
fluorescent protein fusions of the CsAAE1 and CsAAE3 with To gain further insight into the evolution of CsAAE1, we
organelle-specific marker proteins using transient expres- expanded the analysis of AAE clade II to include homologs
sion in Nicotiana benthamiana leaves (Sparkes et al., 2006). from Chlamydomonas, basal plants and higher eudicots.
To avoid disrupting the C-terminal PTS1 signal, CsAAE1 and Using data from the recent sequencing of the cannabis
CsAAE3 were expressed as N-terminal yellow fluorescent genome (van Bakel et al., 2011), we identified three addi-
protein (YFP) fusions. When co-infiltrated leaves were tional clade II AAEs (CsAAE12, 13 and 14). The proteins in the
examined using confocal microscopy, YFP:CsAAE3 was expanded clade II group into three distinct subclades
found in discrete compartments that co-localized with the (Figure 7a). We designated the group that included CsAAE1
PX-CK peroxisome cyan fluorescent protein (CFP) marker and AtAAE17 as subclade IIa, the group that included
protein (Figure 6). In contrast, YFP:CsAAE1 localized to the AtAAE18 as subclade IIb, and the group that contained
cytosol. We compared the localization of CsAAE1 with the ACS and homologs as subclade IIc. To investigate if perox-
putative cannabinoid pathway enzyme OLS, which does not isomal localization is a hallmark of subclades IIa/b, the

ª 2012 The Authors

The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
6 Jake M. Stout et al.

CsAAE1 mentally validated plant PTS1 sequences were added (Reu-

100 mann, 2004; Lingner et al., 2011). In total, 29 out of 30 were
predicted to have a functional PTS1 (e-values £0.25), which
80 indicated that all but one of the clade IIa/b AAEs were
(% ATP depletion)

targeted to the peroxisome (Figure 7b). CsAAE1 is the only

Enzyme activity

60 clade II AAE that lacks a PTS1.

40 DISCUSSION
Role of hexanoyl-CoA in cannabinoid biosynthesis
20
Cannabinoids are a rare example of a plant polyketide bio-
0 synthesized from hexanoate or hexanoyl-CoA precursors.
C ara te
2
3
is C4
4
is C5
5
6
7
8
C9
10

C2
C4
C6
p- Cin C18
umam 0

Fe ffea e
Si ru te
pa e
te
C
C

C
C
C
C
C

1
1
1

co n 2

a t

na lat
a
Many plants and microbes use fatty acyl-CoAs as polyketide
C
o

primers (e.g. alkylresorcinols in Sorghum (Cook et al.,

2010)), yet there are only a few examples of hexanoate/
CsAAE3 hexanoyl-CoA-derived polyketides in plants and fungi. Pri-
100 min, the allergenic principle of Primula obconica, is postu-
lated to be formed from hexanoyl-CoA (Horper, 1996).
80 Outside of the plant kingdom, Dictyostelium discoideum
(% ATP depletion)

synthesizes the DIF-1 polyketide from hexanoyl-acyl carrier

Enzyme activity

60 protein (ACP) thioester (Austin et al., 2006; Ghosh et al.,

2008). The well known fungal toxins of the aflatoxin/sterig-
40 matocystin class are formed from hexanoate (Townsend
et al., 1984; Yabe and Nakajima, 2004) as are the antitumor
20
benastatins in Streptomyces (Xu et al., 2009).
The detection of hexanoyl-CoA in cannabis flowers dem-
0
onstrates that this compound is present in the tissues where
C ara te
2
3
is C4
4
is C5
5
6
7
8
C9
10

C2
C4
16
p- Cin C18
umam 0
af te
Si ru te
pa e
te
C
C

C
C
C
C
C

1
1

co n 2

na lat
a

Fe ea
C

cannabinoid biosynthesis occurs (Table 1). The concentra-

o

tion of hexanoyl-CoA was the highest in female flowers,

Substrate which also contained the highest concentration of CBDA.
CBDA was also detected in leaves, stems and roots, albeit at
Figure 4. Substrate specificities of CsAAE1 and CsAAE3. Diverse carboxylic
acid substrates were tested with purified recombinant CsAAE1 or CsAAE3, concentrations that were orders of magnitude less com-
CoA, Mg2+ and ATP. The amount of unconsumed ATP that remained at pared with flowers.
completion was quantified using a luciferase-based system (Schneider et al., The small size of the pool, measured to be 15.5 pmol g)1
2005). Values represent the percent difference of ATP depletion compared
with control reactions that contained no substrate. C2 to C20 indicate the fresh weight, is perhaps not surprising considering that this
number of carbons in saturated fatty acids; iso denotes branched-chain compound is a pathway intermediate. This analysis may
substrates. Error bars represent the percent error of the ratio, n = 3. also be complicated by the possible association of hexanoyl-
CoA with acyl-CoA binding protein (Xiao and Chye, 2011). To
our knowledge, a few reports describe the quantification of
C-terminal 12 amino acids of each protein in the clade IIa/b short- and medium-chain acyl-CoA thioesters in plants.
phylogeny were aligned using Muscle, and the presence of a Perera et al. (2009) used LC-MS/MS to quantify the butyryl-
functional PTS1 signal was assessed using a BLOCKS-based CoA/isobutyryl-CoA content of Arabidopsis siliques and
algorithm (http://www.peroxisomedb.org) to which experi- leaves to be 380 and 160 pmol g)1 fresh weight, respectively.

Table 4 Kinetic properties of CsAAE1 and CsAAE3

CsAAE1 CsAAE3

Substrate Km Vmax (pKat) kcat (s)1) Km Vmax (pKat) kcat (s)1)

CoA 0.26  0.05 lM – – 0.16  0.01 lM – –

Butanoate >10 mM – – >10 mM – –
Hexanoate 3.7  0.7 mM 6.8  0.7 2.0 261  37 lM 1.8  0.05 57.6
Decanoate 1.7  0.2 mM 1.8  0.7 0.5 16.1  5.8 lM 1.6  0.1 10.0
Palmitate n.d.a – – 1.3  0.5 lM 0.4  0.01 2.4

a
Not determined due to lack of catalytic activity.

ª 2012 The Authors

The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
 A cytoplasmic acyl-activating enzyme involved in cannabinoid biosynthesis 7

CsAAE1 data provide evidence that CsAAE1 is the enzyme involved

150 5
in cannabinoid biosynthesis. CsAAE1 was the most abun-
4 dant transcript in the EST catalogue, and quantitative
3
100 2 reverse transcriptase polymerase chain reaction (qRT-PCR)
1
0 data showed that its expression was over 100-fold higher in

–
FF
FF
50 trichome cells compared with other tissues (Figure 5). Fur-
thermore, we have localized CsAAE1 to the cytosol, which is
0 the same compartment where the putative cannabinoid en-
zyme OLS is localized. The substrate preference of CsAAE1
Transcript abundance (fold difference compared to leaves)

–50 provides additional evidence for its role in cannabinoid

R

F
FF

M
biosynthesis as it shows more specificity for hexanoate and
other short-chain fatty acyl-CoAs than CsAAE3 (Figure 4).
CsAAE3
0 Unfortunately, there have been no reports of successful
–2 genetic transformation of cannabis, and it was not possible
–4
10 to probe the function of CsAAE1 or CsAAE3 by using trans-
–6
–8 genic approaches such as RNAi. Our attempts at virus-
+

5 induced gene silencing (VIGS) using tobacco rattle virus

FF
FF

were also not successful (data not shown).

0 An argument against the involvement of CsAAE1 in
cannabinoid biosynthesis is its high Km value (3.7 mM) for
–5 hexanoate. Km values from plant AAEs tend to be in the low
lM range (Table S1) but most of these values are from clade
–10 IV and V AAEs, and it is not clear if these values are
R

F
FF

consistent for members of other AAE clades. The Arabidop-

sis malonyl-CoA synthetase (AtAAE13; clade VII) has a Km for
CBDAS malonate of 529 lM, which is approximately sevenfold lower
2500
than that of CsAAE1 for hexanoate (Chen et al., 2011).
150
Analysis of the kinetic properties of CsAAE1 showed that this
100
2000 enzyme could only utilize hexanoate and, to a lesser extent,
50
decanoate as substrate, with hexanoate acid yielding the
1500 0
highest kcat value (Table 4). These data are corroborated
+

–
FF
FF

1000 with those findings from the luciferase-based ATP depletion

assay (Figure 4). In contrast to CsAAE1, CsAAE3 accepted a
500
broader range of substrates (Figure 4) and showed reason-
0 able kinetic properties for hexanoate, decanoate and palm-
itate (Table 4). Interestingly, CsAAE3 was inhibited by
R

FF

F
M

decanoate but not hexanoate or palmitate. A similar inhib-

Figure 5. Quantitative reverse transcription polymerase chain reaction (qRT- itory effect of C8–C12 fatty acids has been observed for
PCR) analysis of CsAAE1, CsAAE3 and CBDAS expression in different tissues
CsAAE3’s close Arabidopsis homolog (At4g05160) (Schnei-
of the hemp cultivar ‘Finola’. Gene expression values relative to actin were
plotted as fold differences compared with leaves, with leaf expression der et al., 2005).
assigned a value of 1. Insets depict gene expression in female flowers with One route to overcoming catalytic inefficiency and low
and without trichomes, with values also indicated as fold differences
turnover of secondary metabolic enzymes is to increase
compared with leaves. R, roots; S, stems; L, leaves; FF+, female flowers with
trichomes; FF), female flowers with trichomes removed by the Beadbeater expression levels (reviewed in Pichersky and Gang (2000)).
method; T, trichomes; MF, male flower. Values are mean  SD, n = 3. Although we did not quantify the CsAAE1 protein, the high
transcript levels indicated by ESTs counts (Table 2) and qRT-
PCR (Figure 5) showed that it is highly expressed.
These pools are approximately an order of magnitude greater CsAAE1 lacks a PTS1 (Figure 7b) and its recruitment to the
than what we observed in cannabis flowers. cannabinoid pathway may have resulted from altered sub-
cellular localization. Given that all other clade II AAEs are
CsAAE1 is a cannabinoid biosynthetic gene that
localized in the peroxisome or plastid (i.e. clade IIc), it is
neofunctionalized by different subcellular localization
possible that this neofunctionalization is a recent evolution-
Through a sequence-based analysis of the trichome EST ary event specific to cannabis. CsAAE1 may have arisen
dataset and biochemical assay of five AAEs, we identified through a tandem duplication event and has undergone
two that possessed hexanoyl-CoA synthetase activity. Our subsequent divergence from its ancestral function through a
ª 2012 The Authors
The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
8 Jake M. Stout et al.

YFP:CsAAE1 CFP:CsOLS Merged Figure 6. Subcellular localization of CsAAE1 and

(a) (b) (c) CsAAE3. Cannabis acyl-activating enzymes
(AAEs) were expressed transiently as N-terminal
YFP fusions in Nicotania benthamiana leaves. (a)
YFP:CsAAE1. (b) OLS:CFP co-expressed in the
same cells as YFP:CsAAE1. (c) Overlay of A and
B. (d) YFP:CsAAE3. (e) PX-CK peroxisome marker
co-expressed in the same cells as YFP:CsAAE3.
(f) Overlay of (d) and (e). CFP, cyan fluorescent
protein; YFP, yellow fluorescent protein.
20 µm

(d) YFP:CsAAE3 (e) CFP:PX-CK (f) Merged

20 µm

change in subcellular localization. The identification of a a C6 fatty acid. Although C8- and C10-specific thioesterases
second clade IIa cannabis AAE (CsAAE12) through our are found in plants that accumulate medium-chain fatty
cannabis genome sequencing effort (van Bakel et al., 2011) acids (e.g. Cuphea; Dehesh et al., 1996), there are no C6-
supported this hypothesis. CsAAE12 is adjacent to CsAAE1 specific thioesterases reported. We searched the cannabis
in the genome, encodes a protein that has 77% amino acid trichome transcriptome and found three contigs that were
identity to CsAAE1 and possesses a functional PTS1 signal similar to acyl-ACP thioesterases. Represented by only
(Figure 7b). Alternatively, CsAAE1 may have evolved eight, three and two ESTs, they were not as prominent as
through allelic drift away from the ancestral clade IIa the cannabinoid and terpenoid pathway transcripts. We
function. were unable to detect any thioesterase-like sequences in
the EST dataset reported by Marks et al. (2009). A second
CsAAE3 is a peroxisomally targeted enzyme that is likely
possible route to hexanoate is the degradation of branched-
involved in b-oxidation
chain amino acids and elongation of the resulting a-keto
We found that CsAAE3 is more efficient than CsAAE1 at acids (a-KA) (Kroumova et al., 1994; Kroumova and Wag-
synthesizing hexanoyl-CoA. However, CsAAE3 is localized to ner, 2003). We considered this route to be unlikely given
the peroxisome (Figure 6) and it is not clear how hexanoyl- that 13C feeding provided no evidence to support this route
CoA formed in this compartment could be exported to the (Fellermeier et al., 2001) and the lack of high expression for
cytoplasm where the polyketide synthesis phase of the transcripts for branched-chain amino acid metabolism. A
cannabinoid pathway is located. CsAAE3 accepts a very third option is the formation of C18 unsaturated fatty acids
broad range of substrates, which indicated that it is a more by the action of desaturases, and their cleavage by lipoxy-
generalized acyl-CoA synthetase that may function in per- genase and hydroperoxide lyase to yield C6 and C12
oxisomal b-oxidation. products. Volatile C6-aldehydes in soybean, apple and to-
mato are formed by a lipoxygenase pathway (Matoba et al.,
Origin of hexanoate
1985; Gray and Prestage, 1999; Rowan et al., 1999). The
Although the incorporation of 13C-label into cannabinoids high expression of three desaturases (62, 53 and 22 ESTs)
supports the hypothesis that hexanoate is formed from and a lipoxygenase (25 ESTs) provides evidence that the
fatty acid precursors (Fellermeier et al., 2001), the origin of fatty acid degradation pathway maybe active in trichomes,
hexanoate for activation by CsAAE1 has not been clarified. although hydroperoxide lyase was represented by three
There are several possible metabolic routes to short- and ESTs.
medium-chain fatty acids in plants such as de novo bio-
Metabolic engineering of cannabinoid biosynthesis
synthesis, elongation of a-keto acids, and fatty acid degra-
dation. Marks et al. (2009) concluded that hexanoate was Our discovery of AAEs that catalyze the biosynthesis of
likely formed from the de novo pathway, based on the high hexanoyl-CoA and other acyl-CoA esters may further the
expression of ACP and 3-keto-ACP reductase in trichomes breeding of low-cannabinoid cannabis varieties. Hemp cul-
compared with leaves. The de novo route would require an tivars must contain <0.3% of THCA to be permitted to be
acyl-ACP thioesterase to terminate fatty acid biosynthesis at grown legally in Canada and in the European Union. While

ª 2012 The Authors

The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
 A cytoplasmic acyl-activating enzyme involved in cannabinoid biosynthesis 9

(a)

(b)

Figure 7. Analysis of acyl-activating enzyme (AAE) clade II homologs from the plant kingdom. (a) Maximum likelihood inference of the phylogenetic relationship of
CsAAE1 and representative plant clade II homologs. Branch point bootstrap values were calculated with 1000 replicates, and values <60 are not shown. At,
Arabidopsis thaliana; Cr, Chlamydomonas reinhardtii; Cs, Cannabis sativa; Gm, Glycine max; Mt, Medicago truncatula; Os, Oryza sativa; Pp, Physcomitrella patens;
Pt, Populus trichocarpa; Rc, Ricinus communis; Sm, Selaginella moellendorffii; Vc, Volvox carteri; Vv, Vitis vinifera; Zm, Zea mays. (b) Alignment of the C-terminal
amino acid residues of clade II acyl-activating enzyme (AAE) proteins. The e-values calculated using a peroxisome targeting sequence 1 (PTS1) signal prediction
algorithm (http://www.peroxisomedb.org) are listed adjacent to each sequence.

ª 2012 The Authors

The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
10 Jake M. Stout et al.

breeding has led to low-THCA hemp, reduction in CsAAE1

Analysis of hexanoyl-CoA and CBDA in cannabis flowers
activity through targeted mutagenesis may lead to further
decreases and also reduce other non-psychoactive canna- A solution that contained benzoyl-CoA and hexanoyl-CoA
binoids such as CBDA. standards (1 lM in 95:5 water:methanol with 5 mM TEA and
Cannabinoids have received increased attention as drugs 3 mM acetic acid) was used to optimize chromatographic
to treat a number of diseases and conditions (Mechoulam, conditions. Benzoyl-CoA was used as the internal standard.
2005; Izzo et al., 2009). The discovery of hexanoyl-CoA Female ‘Finola’ plants were divided into flower, leaf, stem
synthetases, together with the identification of other genes and root samples, which were then ground to homogeneity
encoding the cannabinoid biosynthetic pathway (Figure 1), in liquid nitrogen. Approximately 400 mg of homogenized
allows for the reconstruction of the pathway in heterologous tissue was extracted with 5% trichloroacetic acid (TCAA) for
systems using a synthetic biology approach. Such a plat- 5 min at 4C. The samples were centrifuged at 1800 g for
form would enable the discovery of genes that encode 5 min and the supernatant purified using cation exchange
enzymes that catalyze the formation of the less-abundant (Waters MCX1CC Oasis cartridge, www.waters.com): after
cannabinoid derivatives, and may allow for scalable fer- activation with 1 ml methanol and 1 ml water, the TCAA
mentation production of these compounds so that their supernatant was loaded, washed with 1 ml 0.1 N HCl and
efficacy can be tested. 1 ml 100% MeOH, and eluted with 1 ml 25 mM NH4OH in
82% MeOH. The eluate was dried and the residue resus-
EXPERIMENTAL PROCEDURES
pended in 100 ll of the starting liquid chromatography (LC)
Plant cultivation and isolation of glandular trichomes buffer. UPLC-MS/MS analysis was performed using a Wa-
ters Acquity UPLC system with a diode array detector and a
Female plants of a marijuana strain (12% THCA by dry
Quattro Ultima mass spectrometer in electrospray ioniza-
weight, low CBDA) were propagated from cuttings and
tion (ESI) multiple reaction monitoring (MRM) positive ion
grown in a secure controlled environment chamber under
mode. Chromatography was performed on a reverse phase
long day (18 h) vegetative conditions. Flowering was
UPLC column (2.1 · 50 mm, 1.7 lm, BEH C18; Waters)
induced by transfer to 12-h day length. Female plants of the
using 7.5 ll injections. Separations were achieved with a
hemp cultivar ‘Finola’ (<0.3% THCA, high CBDA) were grown
binary solvent system that consisted of solvent A (water
from seed under 16-h day conditions. Trichome cells were
with 5 mM triethylamine and 3 mM acetic acid, pH 10) and
isolated from female flowers using a Beadbeater method
solvent B (methanol and water (95:5) with 5 mM triethyl-
adapted from (Gershenzon et al., 1992). Yield was ca. 0.9 g
amine and 0.3 mM acetic acid, pH 11). The elution gradient
from 60 g of flowers.
(200 ll min)1) was: 5.3% B for 1 min, 15% B at 6 min and
cDNA library and ESTs 50% B at 10 min. The column was flushed and returned to
starting conditions at 30 min. Consistent with previous
Total RNA from trichomes from a high-THCA marijuana strain work (Perera et al., 2009), optimum MRM transitions in
was isolated using a CTAB-based method (Meisel et al., 2005) positive ionization mode resulted in a loss of 507 amu.
followed by clean-up using an RNeasy Midi (Qiagen, Quantification was carried out by using a peak area ratio of
www.qiagen.com) kit. Genomic DNA was removed by on- the MRM transitions for hexanoyl-CoA (m/z 866 > 359,
column digest with DNase I (Qiagen). mRNA was isolated from 9.25 min) relative to benzoyl-CoA (m/z 872 > 365,
total RNA using Dynabeads Oligo(dT) (Invitrogen, www.invi- 8.45 min).
trogen.com). mRNA (3.7 lg) was used as a template for cDNA CBDA was measured in the same tissues used to quantify
library synthesis (Stratagene, www.stratagene.com) to pro- hexanoyl-CoA. Approximately 200 mg of homogenized tis-
duce cDNAs with EcoRI (5¢) and XhoI (3¢) sites. cDNA was sue was extracted with 50% methanol at 65C for 10 min.
separated on a 1% agarose gel into >2.5, 1.5–2.5 and <1.5 kb The extracts were filtered using a Spin-X column (Corning,
fractions and each ligated separately into pBluescript SK+ www.corning.com). HPLC analysis was performed using a
vector digested with EcoRI and XhoI before electroporation Waters 2695 system coupled to a Waters 3100 single
into E. coli DH10B T1 Phage Resistant Cells (Invitrogen). Bac- quadrupole mass detector and a 996 photodiode array
terial cultures were used for TempliPhi amplifications (GE detector. Separations were achieved on a reverse phase
Healthcare, www.gehealthcare.com). TempliPhi products column (2.1 mm · 5 cm, 2.7 lm, Ascentis Express C18;
were sequenced with the T3 primer. BigDye reactions were Sigma-Aldrich, www.sigmaaldrich.com) using a binary sol-
resolved on an ABI 3730xl sequencer (Applied Biosystems, vent system consisting of solvent A (water containing 10%
www.appliedbiosystems.com). Sequences were processed acetonitrile, 0.01% formic acid) and solvent B (99.09%
using a high-performance cluster computer using the proce- acetonitrile, 0.01% formic acid) with gradient elution
dures described in Nagel et al. (2008). Sequences were anno- (250 ll min)1) of: 45% B from 0 to 8 min; 95% B from 8 to
tated by comparison to the UniProt database using the BLASTX 10.5 min, 45% B at 13 min with a hold to 20 min. Peaks were
program. detected by absorbance at 270 nm. CBDA was quantified

ª 2012 The Authors

The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
 A cytoplasmic acyl-activating enzyme involved in cannabinoid biosynthesis 11

using a standard curve prepared from an authentic CBDA gation. The wash step was repeated twice and the samples
standard (40–9500 pmoles injected, r2 1.0). were loaded onto a chromatography column. After washing
with 10 ml of HWB, His-tagged proteins were eluted with
qRT-PCR of CsAAE1 and CsAAE3 expression
2.5 ml of His-tag elution buffer (50 mM Tris–HCl pH 7,
‘Finola’ plants were grown from seed until mid-flowering 500 mM NaCl, 250 mM imidazole, 10% v/v glycerol, 10 mM
stage. Roots, stems, leaves, female flowers (with trichomes b-mercaptoethanol). The eluates were exchanged into
and after trichome isolation using the Beadbeater), trichome 50 mM HEPES pH 9, 10% v/v glycerol, 2 mM MgCl2, 2 mM
cells and male flowers were sampled from three plants. dithiothreitol using PD10 columns (Amersham Biosciences,
Total RNA was isolated as described above. RNA had an www.gelifesciences.com). The purity of the isolated proteins
Abs260:Abs280 of >1.9 and showed distinct ribosomal bands was verified by SDS-PAGE (Figure S1), and the protein
on a denaturing gel. First-strand cDNA was synthesized concentration was determined by Bradford assay.
using 0.5 lg RNA with a QuantiTect cDNA Synthesis kit Hexanoyl-CoA synthetase assays were performed in a
(Qiagen). Each 20 ll cDNA sample was diluted 1:4 with 20 ll reaction mix that contained 0.1 lg recombinant AAE,
water, and 1 ll used as a PCR template. Gene-specific 50 mM HEPES pH 9, 8 mM ATP, 10 mM MgCl2, 0.5 mM CoA
primers were designed to produce amplicons of 90–200 bp and 4 mM sodium hexanoate for 10 min at 40C. Reactions
(Table S2). PCR reactions (20 ll) were performed in 96-well were terminated with 2 ll of 1 N HCl and stored on ice until
plates using a SYBR Green-based assay (QuantiFast SYBR analysis. After centrifugation at 15 000 g for 5 min at 4C,
Green kit; Qiagen) with a StepOne Plus instrument (Applied the samples were diluted 1:100 in water and analyzed by
Biosystems). The cycling parameters used were 95C for UPLC-MS/MS as described above.
5 min followed by 40 cycles of 95C for 10 sec, 60C for The kinetic properties of CsAAE1 and CsAAE3 with CoA
30 sec, and a standard dissociation protocol (95C 15 sec, were tested in assays containing 50 mM HEPES pH 9.0,
60C for 1 min, 60–95C in 0.3C increments for 15 sec). 10 mM sodium hexanoate, 8 mM ATP, 10 mM MgCl2,
Experiments were performed using cDNAs from three plants 0.02–2 mM CoA, and 290 ng of CsAAE1 or 10 ng of CsAAE3.
with two technical replicates. Actin, which we found had Detection of hexanoyl-CoA was performed by analyzing 5 ll
stable expression in all tissues tested, was used as a refer- of the reaction mixtures by HPLC-MS using a Waters 2695
ence gene. The efficiencies for all primer pairs were 90–110% system coupled to a Waters 3100 single quadrupole mass
as calculated using the standard curve method. Ct values detector. The isocratic LC conditions consisted of 87% 2 mM
were calculated using STEPONE Software (Applied Biosys- ammonium formate pH 4.6 and 13% acetonitrile, with a flow
tems). The 2)DDCt method was used for relative gene rate of 250 ll min)1. Compounds were separated on a
expression analysis. reverse phase column (2.1 mm · 5 cm, 2.7 lm, Ascentis
Express C18; Sigma-Aldrich). Hexanoyl-CoA was detected in
Recombinant protein expression, purification and assay
positive mode by scanning for m/z of 866 Da (CV = 45 V,
CsAAE ORFs were amplified using gene-specific primers span = 0.2 Da, dwell time = 0.01 sec). Curve fitting of the
(Table S2) using full-length cDNAs or trichome cDNA as data was performed with PRISM 5.0 (GraphPad, www.graph-
template. The PCR products were cloned into pCR8/GW/ pad.com) using the non-linear regression with substrate
TOPO (Invitrogen) and recombined into the pHIS8/GW using inhibition model for CsAAE1 and the Michaelis–Menten
LR recombinase (Invitrogen). Plasmids were transformed model for CsAAE3. Kinetics were measured in assays that
into Rosetta 2 (DE3) cells (Merck, www.merck.com) and contained 50 mM HEPES pH 9.0, 8 mM ATP, 10 mM MgCl2,
protein expressed in 500 ml LB medium with 0.2 lM iso- and various concentrations of fatty acid. [CoA] was 0.5 mM
propyl-beta-D-thiogalactopyranoside (IPTG). CsAAE1 was for CsAAE1 and for 2 mM CsAAE3. Compounds were
expressed at 12C for 24 h, whereas CsAAE3 was expressed separated as above, except that isocratic conditions were
at 37C for 16 h. Cells were harvested by centrifugation and 93% ammonium formate for butyryl-CoA, 72% for decanoyl-
resuspended in 10 ml His-tag lysis buffer (50 mM Tris–HCl CoA and 45% for palmitoyl-CoA analysis. Masses scanned
pH 7, 500 mM NaCl, 2.5 mM imidazole, 10% v/v glycerol, were 838 Da (butyryl-CoA), 922 Da (decanoyl-CoA), and
10 mM b-mercaptoethanol, 1% v/v Tween 20, 750 lg ml)1 1006.5 (palmitoyl-CoA). Reaction products were quantified
lysozyme), incubated on ice for 1 h, and lysed by sonication. using standard curves of authentic CoA-thioester standards
After centrifugation for 20 min at 12 000 g at 4C, the (r2 of 1.0, butyryl-CoA; 0.99, hexanoyl-CoA; 1.0, decanoyl-
supernatant was added to 160 ll of Talon resin (Clontech, CoA; 0.99, palmitoyl-CoA).
www.clontech.com) that had been washed with His-tag
Substrate preference assay
wash buffer (HWB; 50 mM Tris–HCl pH 7, 500 mM NaCl,
2.5 mM imidazole, 10% v/v glycerol, 10 mM b-mercaptoeth- Assays conditions were identical to those used in (Schneider
anol). The samples were incubated with rocking at 4C and et al., 2005). To each well of a 96-well plate, 90 ll of 100 mM
the resin pelleted by centrifugation. The resin was resus- Tris pH 7.8, 1 mM MgCl2, 2.3 lg luciferin and 0.5 lg of lucif-
pended in HWB, washed at 4C, and recovered by centrifu- erase were added. After shaking for 2 sec the luminescence
ª 2012 The Authors
The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
12 Jake M. Stout et al.

was measured with a 1420 Multilabel counter (PerkinElmer, Figure S1. HPLC chromatogram of hemp ‘Finola’ flower extract.
www.perkinelmer.com) for 15 sec without a filter in place. Figure S2. SDS-PAGE analysis of purified recombinant AAEs.
Figure S3. Biochemical properties of CsAAE1 and CsAA3.
Subcellular localization of CsAAE1 and CsAAE3 Figure S4. Kinetic plots of CsAAE1 and CsAAE3.
Figure S5. Maximum likelihood inference of the phylogenetic
YFP:CsAAE1 and YFP:CsAAE3 fusions were constructed by relationship of cannabis AAEs and the Arabidopsis AAE superfamily.
recombination into pEARLEYGATE104 (Earley et al., 2006) Table S1 Km values of acyl-activating enzymes reported in the
using LR recombinase (Invitrogen). To generate an OLS:CFP literature
Table S2 Primers sequences used in this study
construct, OLS lacking a stop codon was cloned into pCR8/
Table S3 Protein sequences used for phylogenetic analysis
GW/TOPO before recombination into pEARLYGATE102 Please note: As a service to our authors and readers, this journal
using LR recombinase. The peroxisome marker PX-CK provides supporting information supplied by the authors. Such
(Nelson et al., 2007) was from ABRC (http://www.arabidop materials are peer-reviewed and may be re-organized for online
sis.org). Plasmids were transformed into Agrobacterium delivery, but are not copy-edited or typeset. Technical support
issues arising from supporting information (other than missing
tumefaciens GV3101 by electroporation and selected on LB
files) should be addressed to the authors.
plates that contained 10 lg ml)1 rifampicin and 50 lg ml)1
kanamycin. Leaves of 2-week-old Nicotiana benthamiana REFERENCES
plants were infiltrated with the Agrobacterium solution at an Austin, M.B., Saito, T., Bowman, M.E., Haydock, S., Kato, A., Moore, B.S., Kay,
OD600 of 0.02 (Sparkes et al., 2006). Leaf epidermal cells were R.R. and Noel, J.P. (2006) Biosynthesis of Dictyostelium discoideum dif-
ferentiation-inducing factor by a hybrid type I fatty acid–type III polyketide
visualized using a Zeiss, www.zeiss.com LSM510 confocal
synthase. Nature Chem. Biol. 2, 494–502.
microscope at 2 days post-infiltration. CFP was visualized van Bakel, H., Stout, J.M., Cote, A.G., Tallon, C., Sharpe, A.G., Hughes, T.R.
with excitation at 458 nm and image collection with a 475– and Page, J.E. (2011) The draft genome and transcriptome of Cannabis
sativa. Genome Biol. 12, R102. oi:10.1186/gb-2011-12-10-r102.
525 nm bandpass filter; YFP was measured at 514 nm with a
Chen, H., Kim, H.U., Weng, H. and Browse, J. (2011) Malonyl-CoA synthetase,
530–600 nm bandpass filter. Images were collected and encoded by ACYL ACTIVATING ENZYME13, is essential for growth and
analyzed using the Zeiss LSM software package. development of Arabidopsis. Plant Cell, 23, 2247–2262.
Cook, D., Rimando, A.M., Clemente, T.E. et al. (2010) Alkylresorcinol synth-
Phylogenetic analyses ases expressed in Sorghum bicolor root hairs play an essential role in the
biosynthesis of the allelopathic benzoquinone sorgoleone. Plant Cell, 22,
Arabidopsis, Physcomitrella and Populus AAE sequences 867–887.
Dai, X., Wang, G., Yang, D.S., Tang, Y., Broun, P., Marks, M.D., Sumner, L.W.,
corresponded to those identified previously (Shockey and Dixon, R.A. and Zhao, P.X. (2010) TrichOME: a comparative omics database
Browse, 2011). Additional clade II homologs were retrieved for plant trichomes. Plant Physiol. 152, 44–54.
by using the blastp program to compare the Arabidopsis Dayanandan, P. and Kaufman, P.B. (1976) Trichomes of Cannabis sativa L.
(Cannabaceae). Am. J. Bot. 63, 578–591.
clade II sequences against the phytozome database (http:// Dehesh, K., Edwards, P., Hayes, T., Cranmer, A.M. and Fillatti, J. (1996) Two
www.phytozome.net). Phylogenetic analyses were per- novel thioesterases are key determinants of the bimodal distribution of acyl
formed with MEGA v5.05 software (Tamura et al., 2011) using chain length of Cuphea palustris seed oil. Plant Physiol. 110, 203–210.
Dewick, P.M. (2009) Medicinal Natural Products: A Biosynthetic Approach, 3rd
the maximum likelihood algorithm with distances computed edn. Chichester: John Wiley & Sons.
using 1000 bootstrap replicates. Earley, K.W., Haag, J.R., Pontes, O., Opper, K., Juehne, T., Song, K. and Pik-
aard, C.S. (2006) Gateway-compatible vectors for plant functional genom-
GenBank accession numbers ics and proteomics. Plant J. 45, 616–629.
Elsohly, M.A. and Slade, D. (2005) Chemical constituents of marijuana: the
GenBank accession numbers for the nucleotide sequences complex mixture of natural cannabinoids. Life Sci. 78, 539–548.
Fellermeier, M. and Zenk, M.H. (1998) Prenylation of olivetolate by a hemp
reported in this study are: CsAAE1-11 (JN717233–JN717243) transferase yields cannabigerolic acid, the precursor of tetrahydrocan-
and CsAAE12-15 (JN859185–JN859188). nabinol. FEBS Lett. 427, 283–285.
Fellermeier, M., Eisenreich, W., Bacher, A. and Zenk, M.H. (2001) Biosynthesis
ACKNOWLEDGEMENTS of cannabinoids. Incorporation experiments with 13C-labeled glucoses.
Eur. J. Biochem. 268, 1596–1604.
Funding was provided by the Natural Sciences and Engineering Gaoni, Y. and Mechoulam, R. (1964) Isolation, structure, and partial synthesis
Research Council of Canada, the National Research Council of of an active constituent of hashish. J. Am. Chem. Soc. 86, 1646–1647.
Canada, Genome Canada, Genome Prairie and the Government of Gershenzon, J., McCaskill, D., Rajaonarivony, J.I., Mihaliak, C., Karp, F. and
Saskatchewan. We are grateful to L. Holbrook (Prairie Plant Systems Croteau, R. (1992) Isolation of secretory cells from plant glandular tric-
homes and their use in biosynthetic studies of monoterpenes and other
Inc, http://www.prairieplant.com/) for providing access to cannabis
gland products. Anal. Biochem. 200, 130–138.
trichome samples. We thank J. Noel (Salk Institute) for pHIS8/GW, P. Ghosh, R., Chhabra, A., Phatale, P.A., Samrat, S.K., Sharma, J., Gosain, A.,
Covello for substrates, S. Qiu for cDNA library construction, R. Mohanty, D., Saran, S. and Gokhale, R.S. (2008) Dissecting the functional
Taschuk for cloning OLS, S. Polvi for growing hemp plants, D. Cram role of polyketide synthases in Dictyostelium discoideum: biosynthesis of
for bioinformatics, and M. Smith and J. Zou for comments on the the differentiation regulating factor 4-methyl-5-pentylbenzene-1,3-diol. J.
manuscript. Biol. Chem. 283, 11348–11354.
Gray, D. and Prestage, S. (1999) Fresh tomato specific fluctuations in the
composition of lipoxygenase-generated C6 aldehydes. Food Chem. 64,
SUPPORTING INFORMATION
149–155.
Additional Supporting Information may be found in the online Horper, W. (1996) Biosynthesis of primin and miconidin and its derivatives.
version of this article: Phytochemistry 41, 451–456.

ª 2012 The Authors

The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x
 A cytoplasmic acyl-activating enzyme involved in cannabinoid biosynthesis 13

Hu, Y., Gai, Y., Yin, L., Wang, X., Feng, C., Feng, L., Li, D., Jiang, X.-N. and cDNA encoding a polyketide synthase from Cannabis sativa. Plant Physiol.
Wang, D.-C. (2010) Crystal structures of a Populus tomentosa 4-couma- Biochem. 42, 291–297.
rate:CoA ligase shed light on its enzymatic mechanisms. Plant Cell, 22, Reumann, S. (2004) Specification of the peroxisome targeting signals type 1
3093–3104. and type 2 of plant peroxisomes by bioinformatics analyses. Plant Physiol.
Izzo, A.A., Borrelli, F., Capasso, R., Di Marzo, V. and Mechoulam, R. (2009) 135, 783–800.
Non-psychotropic plant cannabinoids: new therapeutic opportunities from Rowan, D.D., Allen, J.M., Fielder, S. and Hunt, M.B. (1999) Biosynthesis of
an ancient herb. Trends Pharmacol. Sci. 30, 515–527. straight-chain ester volatiles in red delicious and granny smith apples
Kienow, L., Schneider, K., Bartsch, M., Stuible, H.-P., Weng, H., Miersch, O., using deuterium-labeled precursors. J. Agric. Food Chem. 47, 2553–
Wasternack, C. and Kombrink, E. (2008) Jasmonates meet fatty acids: 2562.
functional analysis of a new acyl-coenzyme A synthetase family from Schmelz, S. and Naismith, J.H. (2009) Adenylate-forming enzymes. Curr.
Arabidopsis thaliana. J. Exp. Bot. 59, 403–19. Opin. Struct. Biol. 19, 666–671.
Koo, A.J.K., Chung, H.S., Kobayashi, Y. and Howe, G.A. (2006) Identification of Schneider, K., Kienow, L., Schmelzer, E., Colby, T., Bartsch, M., Miersch, O.,
a peroxisomal acyl-activating enzyme involved in the biosynthesis of jas- Wasternack, C., Kombrink, E. and Stuible, H.-P. (2005) A new type of per-
monic acid in Arabidopsis. J. Biol. Chem. 281, 33511–33520. oxisomal acyl-coenzyme A synthetase from Arabidopsis thaliana has the
Kroumova, A.B. and Wagner, G.J. (2003) Different elongation pathways in the catalytic capacity to activate biosynthetic precursors of jasmonic acid. J.
biosynthesis of acyl groups of trichome exudate sugar esters from various Biol. Chem. 280, 13962–13972.
solanaceous plants. Planta, 216, 1013–1021. Shockey, J. and Browse, J. (2011) Genome-level and biochemical diver-
Kroumova, A.B., Xie, Z. and Wagner, G.J. (1994) A pathway for the biosyn- sity of the acyl-activating enzyme superfamily in plants. Plant J. 66,
thesis of straight and branched, odd- and even-length, medium-chain fatty 143–160.
acids in plants. Proc. Natl. Acad. Sci. USA, 91, 11437–11441. Shockey, J.M., Fulda, M.S. and Browse, J.A. (2002) Arabidopsis contains nine
Lange, B.M., Wildung, M.R., Stauber, E.J., Sanchez, C., Pouchnik, D. and long-chain acyl-coenzyme a synthetase genes that participate in fatty acid
Croteau, R. (2000) Probing essential oil biosynthesis and secretion by and glycerolipid metabolism. Plant Physiol. 129, 1710–1722.
functional evaluation of expressed sequence tags from mint glandular Shockey, J.M., Fulda, M.S. and Browse, J. (2003) Arabidopsis contains a large
trichomes. Proc. Natl. Acad. Sci. USA, 97, 2934–2939. superfamily of acyl-activating enzymes. Phylogenetic and biochemical
Lanyon, V.S., Turner, J.C. and Mahlberg, P.G. (1981) Quantitative analysis of analysis reveals a new class of acyl-coenzyme A synthetases Plant Physiol.
cannabinoids in the secretory product from capitate-stalked glands of 132, 1065–1076.
Cannabis sativa L. (Cannabaceae). Bot. Gaz. 142, 316–319. Sirikantaramas, S., Morimoto, S., Shoyama, Y., Ishikawa, Y., Wada, Y.,
Lingner, T., Kataya, A.R., Antonicelli, G.E., Benichou, A., Nilssen, K., Chen, X.- Shoyama, Y. and Taura, F. (2004) The gene controlling marijuana psy-
Y., Siemsen, T., Morgenstern, B., Meinicke, P. and Reumann, S. (2011) choactivity: molecular cloning and heterologous expression of D1-tetra-
Identification of novel plant peroxisomal targeting signals by a combina- hydrocannabinolic acid synthase from Cannabis sativa L. J. Biol. Chem.
tion of machine learning methods and in vivo subcellular targeting analy- 279, 39767–39774.
ses. Plant Cell, 23, 1556–72. Souza, C. de A., Barbazuk, B., Ralph, S.G., Bohlmann, J., Hamberger, B. and
Marks, M.D., Tian, L., Wenger, J.P., Omburo, S.N., Soto-Fuentes, W., He, J., Douglas, C.J. (2008) Genome-wide analysis of a land plant-specific
Gang, D.R., Weiblen, G.D. and Dixon, R.A. (2009) Identification of candidate acyl:coenzyme A synthetase (ACS) gene family in Arabidopsis, poplar, rice
genes affecting D9-tetrahydrocannabinol biosynthesis in Cannabis sativa. and Physcomitrella. New Phytol. 179, 987–1003.
J. Exp. Bot. 60, 3715–3726. Sparkes, I.A., Runions, J., Kearns, A. and Hawes, C. (2006) Rapid, transient
Matoba, T., Hidaka, H., Narita, H., Kitamura, K., Kaizuma, N. and Kito, M. expression of fluorescent fusion proteins in tobacco plants and generation
(1985) Lipoxygenase-2 isozyme is responsible for generation of n-hexanal of stably transformed plants. Nat. Protocol. 1, 2019–2025.
in soybean homogenate. J. Agric. Food Chem. 33, 852–855. Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M. and Kumar, S.
Mechoulam, R. (2005) Plant cannabinoids: a neglected pharmacological (2011) MEGA5: Molecular evolutionary genetics analysis using maximum
treasure trove. Br. J. Pharmacol. 146, 913–915. likelihood, evolutionary distance, and maximum parsimony methods. Mol.
Mechoulam, R., Parker, L.A. and Gallily, R. (2002) Cannabidiol: an overview of Biol. Evol. 28, 2731–2739.
some pharmacological aspects. J. Clin. Pharmacol. 42, 11S–19S. Taura, F., Sirikantaramas, S., Shoyama, Y., Yoshikai, K., Shoyama, Y. and
Meisel, L., Fonseca, B., Gonzaelz, S., Baeza-Yates, R., Cambiazo, V., Campos, Morimoto, S. (2007) Cannabidiolic-acid synthase, the chemotype-deter-
R., Gonzalez, M., Orellana, A., Retamales, J. and Silva, H. (2005) A rapid and mining enzyme in the fiber-type Cannabis sativa. FEBS Lett. 581, 2929–
efficient method for purifying high quality total RNA from peaches (Prunus 2934.
persica) for functional genomics analyses. Biol. Res. 38, 83–88. Taura, F., Tanaka, S., Taguchi, C., Fukamizu, T., Tanaka, H., Shoyama, Y. and
Nagel, J., Culley, L., Lu, Y., Liu, E., Matthews, P.D., Stevens, J.F. and Page, J.E. Morimoto, S. (2009) Characterization of olivetol synthase, a polyketide
(2008) EST analysis of hop glandular trichomes identifies an O-methyl- synthase putatively involved in cannabinoid biosynthetic pathway. FEBS
transferase that catalyzes the biosynthesis of xanthohumol. Plant Cell, 20, Lett. 583, 2061–2066.
186–200. Townsend, C.A., Christensen, S.B. and Trautwein, K. (1984) Hexanoate as
Nelson, B.K., Cai, X. and Nebenführ, A. (2007) A multicolored set of in vivo a starter unit in polyketide biosynthesis. J. Am. Chem. Soc. 106, 3868–
organelle markers for co-localization studies in Arabidopsis and other 3869.
plants. Plant J. 51, 1126–1136. Wang, G., Tian, L., Aziz, N., Broun, P., Dai, X., He, J., King, A., Zhao, P.X. and
Perera, M.A., Choi, S.-Y., Wurtele, E.S. and Nikolau, B.J. (2009) Quantitative Dixon, R.A. (2008) Terpene biosynthesis in glandular trichomes of hop
analysis of short-chain acyl-coenzymeAs in plant tissues by LC-MS-MS (Humulus lupulus L.). Plant Physiol. 148, 1254–1266.
electrospray ionization method. J. Chromatogr. 877, 482–488. Xiao, S. and Chye, M.-L. (2011) New roles for acyl-CoA-binding proteins
Phillips, M.A., León, P., Boronat, A. and Rodrı́guez-Concepción, M. (2008) The (ACBPs) in plant development, stress responses and lipid metabolism.
plastidial MEP pathway: unified nomenclature and resources. Trends Plant Prog. Lipid Res. 50, 141–151.
Sci. 13, 619–623. Xu, Z., Metsä-Ketelä, M. and Hertweck, C. (2009) Ketosynthase III as a gateway
Pichersky, E. and Gang, D.R. (2000) Genetics and biochemistry of secondary to engineering the biosynthesis of antitumoral benastatin derivatives. J.
metabolites in plants: an evolutionary perspective. Trends Plant Sci. 5, 439–445. Biotechnol. 140, 107–113.
Raharjo, T.J., Chang, W.T., Verberne, M.C., Peltenburg-Looman, A.M., Lin- Yabe, K. and Nakajima, H. (2004) Enzyme reactions and genes in aflatoxin
thorst, H.J. and Verpoorte, R. (2004) Cloning and over-expression of a biosynthesis. Appl. Microbiol. Biotechnol. 64, 745–755.

ª 2012 The Authors

The Plant Journal ª 2012 Blackwell Publishing Ltd, The Plant Journal, (2012), doi: 10.1111/j.1365-313X.2012.04949.x