Enzymatic total synthesis of gibberellin A4 from acetate

Article abstract of DOI:10.1271/bbb.100733

Natural terpenoids have elaborate structures and various bioactivities, making difficult their synthesis and labeling with isotopes. We report here the enzymatic total synthesis of plant hormone gibberellins (GAs) with recombinant biosynthetic enzymes from stable isotopelabeled acetate. Mevalonate (MVA) is a key intermediate for the terpenoid biosynthetic pathway. ¹³C-MVA was synthesized from ¹³C-acetate via acetyl-CoA, using four enzymes or fermentation with a MVA-secreted yeast. The diterpene hydrocarbon, ent-kaurene, was synthesized from ¹³C-acetate and ¹³C-MVA with ten and six recombinant enzymes in one test tube, respectively. Four recombinant enzymes, P450 monooxygenases and soluble dioxygenases involved in the GA ₄ biosynthesis from ent-kaurene via GA₁₂were prepared in yeast and Escherichia coli. All intermediates and the final product GA ₄ were uniformly labeled with ¹³C without dilution by natural abundance when [U-¹³C₂] acetate was used. The ¹³C-NMR and MS data for [U-¹³C₂₀] ent-kaurene confirmed ¹³C-¹³C coupling, and no dilution with the ¹²C atom was observed.

Full text of DOI:10.1271/bbb.100733

Biosci. Biotechnol. Biochem., 75 (1), 128–135, 2011
Enzymatic Total Synthesis of Gibberellin A4 from Acetate
1
2
2
Yoshinori SUGAI,1; Sho MIYAZAKI, Shinichiro MUKAI, Isamu YUMOTO,
*
y
Masahiro NATSUME,1;2 and Hiroshi KAWAIDE
1;2;
1United Graduate School of Agricultural Science, Tokyo University of Agriculture and Technology,
3-5-8 Saiwaicho, Fuchu, Tokyo 183-8509, Japan
2Graduate School of Agriculture, Tokyo University of Agriculture and Technology,
3-5-8 Saiwaicho, Fuchu, Tokyo 183-8509, Japan
Received October 12, 2010; Accepted October 22, 2010; Online Publication, January 7, 2011
[doi:10.1271/bbb.100733]
Natural terpenoids have elaborate structures and
various bioactivities, making difficult their synthesis and
labeling with isotopes. We report here the enzymatic
total synthesis of plant hormone gibberellins (GAs) with
recombinant biosynthetic enzymes from stable isotope-
labeled acetate. Mevalonate (MVA) is a key intermedi-
ate for the terpenoid biosynthetic pathway. 13C-MVA
was synthesized from 13C-acetate via acetyl-CoA, using
four enzymes or fermentation with a MVA-secreted
yeast. The diterpene hydrocarbon, ent-kaurene, was
synthesized from 13C-acetate and 13C-MVA with ten
and six recombinant enzymes in one test tube, respec-
tively. Four recombinant enzymes, P450 monooxyge-
nases and soluble dioxygenases involved in the GA4
biosynthesis from ent-kaurene via GA12 were prepared
in yeast and Escherichia coli. All intermediates and the
final product GA4 were uniformly labeled with 13C
without dilution by natural abundance when [U-13C2]
acetate was used. The 13C-NMR and MS data for
[U-13C20] ent-kaurene confirmed 13C-13C coupling, and
no dilution with the 12C atom was observed.
pathway.6) The MVA pathway involves phosphorylation
to MVA, and subsequent decarboxylation leads to IPP.
Isomerization and a carbon chain elongation reaction
afford the biosynthesis of several prenyl diphosphates.
MVA itself is biosynthesized from acetyl-CoA via
acetoacetyl-CoA
and
hydroxymethylglutaryl-CoA
(HMG-CoA). At least seven enzymes are involved in
the biosynthesis of IPP from acetate (Fig. 1A). The
structural variation of terpenoids depends on the chain
length of prenyl diphosphates, the cyclic structure
generated by cyclases, and oxidative modification to
the carbon skeleton. If terpenoid biosynthesis could be
manipulated at the enzyme level, the synthesis of many
kinds of compound with complicated and precise
structures should be achievable.
We attempted the total enzymatic synthesis of
gibberellin A4 (GA4) as a model experiment to
accomplish this objective, because all the genes in-
volved in GA4 biosynthesis have been identified and
characterized in both plants and fungi.7) GA4 biosyn-
thesis in higher plants mainly comprises three phases
depending on characterization of the biosynthetic
enzymes and cellular compartment in higher plants:
step 1, geranylgeranyl diphosphate (GGDP) synthesis
and a cyclization reaction for ent-kaurene biosynthesis;
step 2, oxidation mediated by two cytochrome P450
monooxygenases; and step 3, synthesis of an active
GA by two 2-oxoglutarate-dependent dioxygenases
(Fig. 1B). Step 1 determines the carbon chain length of
prenyl diphosphate as a diterpene (C20) and involves
several enzymes. A lot of ent-kaurene synthases have
been reported from some organisms, including higher
plants, GA-producing fungi8) and bryophytes.9,10) Two
synthetic enzymes, ent-coparyl diphosphate synthase
and ent-kaurene synthase, are involved in the biosyn-
thesis of ent-kaurene in higher plants, whereas the
bifunctional enzyme, CPS/KS, synthesizes ent-kaurene
from GGDP in GA-producing fungi and bryophytes.
Steps 2 and 3 involve some differences among higher
plants and GA-producing fungi. The ent-kaurene oxidase
(KO) genes in higher plants and GA-producing fungi
Key words: biosynthesis; 13C-NMR; isotope; recombi-
nant enzyme; terpenoid
Terpenoids are widely distributed in nature, and over
25,000 compounds have been reported from mammals,
plants and microorganisms.1) Natural terpenoids func-
tion as hormones in animals, plants and insects, drug
materials such as paclitaxel, artemisinine and cotylenin/
fusicoccin, and cell components such as cholesterol.
Bioactive natural terpenoids that are only available in
small amounts in plants or microorganisms are targets
for total and combinatorial chemical syntheses.2–5)
However, the chemical synthesis of terpenoids requires
stereochemical control and involves structures with
extensive functional groups.
All terpenoids are biosynthesized from dimethylallyl
diphosphate (DMAPP) and isopentenyl diphosphate
(IPP), both of which are derived from mevalonate
(MVA) or 2C-methyl-D-erythritol 4-phosphate (MEP)
y
*
To whom correspondence should be addressed. Fax: +81-42-367-5698; E-mail: hkawaide@cc.tuat.ac.jp
Research Fellow of the Japan Society for the Promotion of Science.
Abbreviations: AACT, acetoacetyl-CoA thiolase; ACS, acetyl-CoA synthase; DMAPP, dimethylallyl diphosphate; DMDC, diphosphomevalonate
decarboxylase; GA, gibberellin; GGDP, geranylgeranyl diphosphate; GGPS, geranylgeranyl diphosphate synthase; HMG-CoA, hydroxymethylglu-
taryl-CoA; HMGR, hydroxymethylglutaryl-CoA reductase; HMGS, hydroxymethylglutaryl-CoA synthase; IPI, isopentenyl diphosphate isomerase;
IPP, isopentenyl diphosphate; KAO, ent-kaurenoic acid oxidase; KO, ent-kaurene oxidase; MEP, 2C-methyl-D-erythritol 4-phosphate; MVA,
mevalonate; MVK, mevalonate kinase; PMVK, phosphomevalonate kinase; 3ox, gibberellin 3-oxidase; 20ox, gibberellin 20-oxidase

Enzymatic Total Synthesis of Gibberellins
129
A
B
Fig. 1. In Vitro Synthetic Route to GGDP and GA Dependent on the MVA Pathway and Biosynthesis of GAs in Plants.
Panel A shows the biosynthetic route from an acetate to GGDP via the MVA pathway. Nine enzymes were responsible for the biosynthesis of
GGDP. The synthesis of one molecule of IPP consumed six molecules of ATP and two molecules of NADPH. Panel B shows the GA4
biosynthetic pathways in plants and in the enzyme cocktail. ent-Kaurene, the common precursor of GA, was synthesized via the MEP pathway
and MVA pathway in plants and the enzyme cocktail, respectively. Both KO and KAO in microsomes in the enzyme cocktail produced
synchronously in Pichia yeast were used for preparing the GA12 synthetic cocktail. Both 20ox 3 and 3ox 1 were added to the GA12 synthetic
cocktail after determining the GA12 synthesis.
belong to the CYP701A family and CYP503 family,
respectively, even though these enzymes catalyze the
same reaction.8,11) In addition, the synthesis of GA4
from GA12 is mediated by 2-oxoglutarete-dependent
dioxygenases in higher plants, but the same reaction is
mediated by cyctochrome P450 monooxygenases in GA-
producing fungi.
Materials and Methods
Chemicals. ent-Kaurene was extracted from the leaves of Crypto-
meria japonica and purified by SiO2 column chromatography as
previously described.18) The following cofactors were used: ATP
(Wako, Osaka, Japan), NADPH (Nacalai Tesque, Kyoto, Japan), and
Coenzyme A (Sigma Aldrich, MO, USA). [U-13C6] glucose and
[U-13C2] acetate (99.8% labeled for each) were purchased from Watari
Co. (Yokohama, Japan).
To undertake the synthesis of GA4 by using biosyn-
thetic enzymes, fourteen recombinant enzymes have to
be prepared by heterologous expression in Escherichia
coli and yeast. We have already conducted the in vitro
functional analysis of all GA biosynthetic recombinant
enzymes in higher plants, as well as of bifunctional ent-
kaurene synthases in mosses and fungi.9–15) We recently
succeeded in the recombinant protein production and
functional analysis of both KO (CYP701A) and ent-
kaurenoic acid oxidase (KAO, CYP88A) from lettuce
and rice by heterologously expressing their genes in
the yeast, Pichia pastoris.16,17) We describe here the
enzymatic total synthesis of bioactive GA4 from acetate.
The enzymes responsible for the MVA pathway were
selected for the enzymatic synthesis, because the starting
materials were easily obtained. All the genes for the
biosynthesis of GA4 from acetate via acetyl-CoA, MVA,
GGDP, and ent-kaurene were expressed in E. coli and
Pichia to produce recombinant enzymes. Then, we
established an in vitro synthesis of uniformly 13C-
labeled GA4 from starting materials of [U-13C2] acetate
and [U-13C6] MVA by our system of using the
recombinant enzyme mix.
Cloning of open reading frames for the recombinant biosynthetic
enzymes. The source of the biosynthetic enzyme genes responsible for
the MVA pathway was Neurospora crassa. This fungus was cultured
in a liquid medium (0.5% yeast extract, 0.5% KH2PO4, 0.2% MgSO4
and 0.1% CH3COONH4) for 4 d at 28 ꢀC on a reciprocal shaker
(140 rpm). The mycelia were filtered and frozen in liquid nitrogen. The
frozen mycelia were ground to a fine powder in liquid nitrogen, and
then total RNA was extracted by using the EASY Prep RNA reagent
(Takara Bio, Shiga, Japan) according to the manufacturer’s protocol.
PolyAþ RNA isolation and double-stranded cDNA synthesis were
performed as previously reported.12) The Arabidopsis cDNA library
previously reported was used.19)
Eight ORFs (acetyl-CoA synthase, ACS; acetoacetyl-CoA thiolase,
AACT; HMG-CoA synthase, HMGS; HMG-CoA reductase, HMGR;
mevalonate kinase, MVK; phosphomevalonate kinase, PMVK; di-
phosphomevalonate decarboxylase, DMDC; and IPP isomerase, IPI)
which were necessary for generating the biosynthetic enzymes from
acetate to DMAPP, were amplified from an N. crassa cDNA library
by PCR. A bacterial GGDP synthase (GGPS) gene, which encoded
an enzyme converting DMAPP/IPP to GGDP, was cloned from
Sulfolobus acidocaldarius.20) The plasmid containing SaGGPS was
used as the PCR template. The CYP701A gene encoding KO cloned
from Pisum sativum was supplied by Dr. S. M. Swain.21) The CYP88A
gene encoding KAO from Lactuca sativa was the same as that

130
Y. SUGAI et al.
previously used.16) These P450 genes were amplified from the
plasmid containing each ORF. Two ORFs of the GA biosynthetic
genes, GA20-oxidase 3 (20ox 3; Q39112) and GA3-oxidase 1 (3ox 1;
Q39103), were cloned from the A. thaliana cDNA library. The
polymerase chain reaction (PCR) was performed by using an
Advantage-HF 2 PCR kit (Clontech, CA, USA), based on the
manufacturer’s protocol. The fourteen sets of primers used for
amplification are listed in Supplemental Table S1; see Biosci.
Biotechnol. Biochem. Web site. All PCR products were ligated into
the pCR2.1 vector (Invitrogen) to confirm their DNA sequences by
using the ABI PRISMꢀ 3130xl Genetic Analyzer (Applied Biosystems,
CA, USA).
Enzymatic synthesis of [U-13C6] MVA from acetate. The in vitro
synthesis of MVA combined four recombinant enzymes (90 mg of
ACS, 35 mg of AACT, 35 mg of HMGS, and 210 mg of HMGR) with a
50 mM Tris–HCl buffer at pH 7.5, 20 mM MgCl2, 16 mM ATP, 10 mM
NADPH, 4 mM CoA, and 4 mM [U-13C2] acetate as the substrate (a
500-mL reaction scale). After an overnight incubation at 28 ꢀC, the
reaction mixture was acidified by adding HCl and left for 30 min at
room temperature to convert MVA to the lactone form. The acidified
sample was extracted three times with EtOAc and concentrated in a
gentle stream of dry N2 gas. The concentrated sample was loaded into
an SiO2 cartridge column (Varian, CA, USA) that had previously been
washed with 3 mL of EtOAc and equilibrated with n-hexane/EtOAc
(1:1, v/v). The mevalonolactone was eluted with 10 mL of n-hexane/
EtOAc (1:9, v/v) which was evaporated to dryness with a gentle
stream of dry N2 gas at room temperature.
Construction of the protein expression vectors. Nine ORFs for the
enzymatic synthesis from acetate to GGDP were introduced into the
multi-cloning site of the pQE-30 vector (Qiagen, Hilden, Germany).
The ORF of bifunctional ent-kaurene synthase (CPS/KS) cloned from
the liverwort, Jungermannia subulata, was introduced into the pCold
vector (Takara Bio).9) Two GA oxidase ORFs, 20ox 3 and 3ox 1,
were introduced into the pMALc2 vector (NEB, MA, USA) to
produce fusion forms of maltose-binding protein (MBP).14) Two P450
ORFs, KO and KAO, were introduced into the multicloning site of
pPICZ-A (Invitrogen, CA, USA) in tandem to synchronously produce
two P450 proteins in one host cell. The whole KAO ORF was first
introduced into the pPICZ-ꢀPmeI vector in which the PmeI site in the
promoter region had been deleted. This KAO expression cassette
containing the AOX promoter and terminal region in the pPICZ-
ꢀPmeI vector was then amplified by PCR and introduced downstream
of the pPICZ-KO terminal region. The pPICZ-KO/KAO tandem
vector was finally propagated in E. coli and linearized with PmeI for
transformation.
Synthesis of ent-kaurene from acetate and MVA. The MVA
substrate was added to an ent-kaurene synthetic cocktail which
contained six recombinant enzymes (15 mg of MVK, 310 mg of PMVK,
20 mg of DMDC, 150 mg of IPI, 20 mg of GGPS, and 60 mg of CPS/KS),
5 mM MgCl2, and 10 mM ATP. An additional four enzymes for
synthesizing MVA were mixed into the ent-kaurene synthetic cocktail
as already described when acetate was used as the starting material.
The prepared cocktail was incubated at 28 ꢀC for 8 h, and then the
product was extracted three times with cyclohexane and concentrated
with dry N2 gas.
Synthesis of GA4 from ent-kaurene. A GA12 synthetic cocktail was
prepared by suspending two P450 enzymes in the reaction buffer. ent-
Kaurene was incubated with this GA12 synthetic cocktail in the
presence of 5 mM NADPH. After 2 h of incubation at 28 ꢀC, the GA4
synthesis was performed for 2 h at 28 ꢀC after adding two enzymes
(20ox 3 and 3ox 1) and cofactors (5 mM ascorbate, 5 mM ꢁ-ketoglu-
tarate and 0.5 mM FeSO4). The product was extracted with EtOAc and
analyzed by GC-MS after derivatization.
Expression and production of soluble recombinant biosynthetic
enzymes. Twelve enzymes (ACS, AACT, HMGS, HMGR, MVK,
PMVK, DMDC, IPI, GGPS, CPS/KS, 20ox 3 and 3ox 1) that had been
obtained from the soluble fraction were produced in E. coli grown in a
2 ꢁ YT medium (800 mL) containing ampicillin (100 mg mLꢂ1) at
37 ꢀC for 3 h. Induction of the recombinant protein was initiated by
adding IPTG (final concentration of 1 mM) when the OD600 value had
reached 0.6. After an additional 20 h of incubation at 18 ꢀC, the cells
were collected, washed with a 50 mM Tris–HCl buffer (pH 7.5) and
resuspended in 5 mL/g fresh weight of a lysis buffer (100 mM Tris–
HCl at pH 7.5, 10% glycerol, and 0.5 mM EDTA). The cells were then
treated with lysozyme and sonicated to obtain a soluble protein after
centrifuging at 9,000 rpm for 30 min. The His-tagged recombinant
enzymes were purified by Ni-NTA affinity column chromatography
according to the manufacturer’s protocol. Both the recombinant 20ox 3
and 3ox 1 proteins fused with MBP were used for enzymatic synthesis
without purification.
GC-MS and NMR analyses. The sample was analyzed by the JEOL
(Tokyo, Japan) JMS-Bu25 GC-MS system (ionization energy of 70 eV,
and filament current of 300 mA) coupled with a DB-5 capillary column
(0.25 mm in diameter, 15 m long, and 0.25 mm in film thickness; J&W
Scientific, Folson, CA, USA). The oven temperature program for ent-
kaurene and GAs has been described in previous reports.10,14) The
samples were derivatized to the methyl ester and methyl ester
trimethylsilyl ether by using diazomethane and diazomethane/N-
methyl-N-trimethylsilyl-trifluoroacetamide (MSTFA, Thermo Fisher
Scientific, MA, USA) for the identification of GA12 and GA4,
respectively. The 13C-NMR (150 MHz) spectra were recorded at room
temperature with a JEOL JNM-A600 FT-NMR system in CDCl3, the
solvent signals being used as the references.
Preparation of the membrane-binding protein by using the Pichia
expression system. KO and KAO proteins were produced in Pichia
yeast basically according to the details in our previous report.19) After
MeOH induction, the cells were retrieved by centrifugation, disrupted
by glass beads, and removed from the crude enzyme solution. A
microsomal fraction was prepared by centrifugation using PEG4000
and resuspended in a reaction buffer (100 mM Tris–HCl at pH 7.5, 10%
glycerol, and 0.5 mM EDTA) for use in the enzymatic synthesis
reaction.
Results
Preparation of the basic enzyme cocktail for synthe-
sizing GGDP from acetate
We initially prepared a basic enzyme cocktail for the
in vitro synthesis of GGDP, a common precursor of
diterpenoids. Nine recombinant enzymes responsible for
the conversion steps from an acetate to GGDP (Fig. 1A)
were prepared as components of the enzyme cocktail.
The eight ORFs (Fig. 1A) for biosynthetic enzymes from
the acetate to DMAPP were cloned and amplified by
PCR from a N. crassa cDNA library. Among these,
HMGR is an integral membrane binding proteins to the
endoplasmic reticulum, although the others are soluble
proteins. It has been reported that the membrane-
spanning domain near the N-terminal region of HMGR
was not necessary for the enzyme activity.23) Hence,
N. crassa HMGR in soluble protein form (the residual
510 amino acids near the C-terminal, ꢀHMGR) was
produced in E. coli by truncating the N-terminal region
[U-13C6]MVA production by Saccharomycopsis fibuligera fermen-
tation. The yeast, S. fibuligera, was incubated at 26 ꢀC for 6 d in 50 mL
of a medium containing 0.5% polypeptone, 0.25% yeast extract, 0.1%
KH2PO4, 0.05% MgSO4, 0.1% CaCO3 and 10% [U-13C6] D-glucose.
The resulting culture filtrate was extracted three times with 2-butanone
at pH 2.0. This crude extract was purified in an SiO2 column (silica gel
60, Merck, NJ, USA) and eluted with n-hexane/EtOAc (1:1). Purified
[U-13C6] mevalonolactone (364 mg), the dehydrated form of MVA,
was obtained and used for the subsequent experiments. NMR data, ꢀc
(CDCl3): 29.7 (d, 40 Hz, 3-CH3), 35.8 (dd, 36 Hz, C-4), 44.6 (dd,
36 Hz, C-2), 66.0 (d, 35 Hz, C-5), 68.2 (ddd, 36 Hz, C-3), 170.5 (d,
51 Hz, C-1). MS data, m=z: 136 (Mþ, 2%), 75 (83), 61 (30), 45
(100).22)

Enzymatic Total Synthesis of Gibberellins
131
100
(kDa) 1
2
2
2
3
4
5
6
A
A
100
70
100
50
70
173
218
50
40
50
292
310
40
100
200
m/z
300
100
B
(kDa) 1
3
4
5
6
7
8
B
100
70
100
70
50
161
50
40
50
40
204
272
290
30
25
30
25
100
200
m/z
300
Fig. 3. Mass Spectra of Enzymatically Synthesized [U-13C20
Geranylgeraniol and Authentic Geranylgeraniol.
]
1
3
4
(kDa)
C
Panels A and B show the spectra of enzymatically synthesized [U-
] geranylgeraniol and authentic geranylgeraniol, respectively.
13
C
20
A, GC tR ¼ 7:43 min, m=z 310 ([M]þ, 4.7%). B, GC tR ¼ 7:43 min,
m=z 290 ([M]þ, 4.2%).
to determine the enzyme products by GC-MS and 13C-
NMR. The substrate [U-13C6] MVA (97% 13C-labeled),
which was easily obtained from [U-13C6] glucose by
fermentation with S. fibuligera,24) was added to the basic
cocktail, and the mixture was incubated at 28 ꢀC for
8 h. The GGDP synthesized was dephosphorylated with
alkaline phosphatase and analyzed by GC-MS. The
spectrum showed that the molecular and fragment ion
peaks were shifted relative to the weight of the 13C
isotope, even though the retention time by GC was the
same as that of the natural form (Fig. 3). The yield
of GGDP was estimated as 27% from the results of the
GC-MS analysis.
Although we could obtain [U-13C6] MVA by the
fermentation method just described, this procedure
carried the risk of dilution of the isotope-labeled
molecules. Therefore, MVA was enzymatically synthe-
sized from a labeled acetate to avoid this problem. Four
recombinant enzymes (ACS, AACT, HMGS and
ꢀHMGR), CoA, ATP and Mg2þ were utilized for an
in vitro synthesis of MVA. [U-13C2] sodium acetate
(99.8% 13C-labeled) was incubated in the MVA syn-
thetic cocktail at 28 ꢀC for 8 h, and MVA was obtained.
The GC-MS and 13C-NMR data for mevalonolactone
were the same as those for fermentated [U-13C6]
mevalonolactone. The yield of MVA from the acetate
was estimated to be about 54% (0.7 mg).
100
70
100
70
Fig. 2. SDS–PAGE of the Recombinant Soluble Proteins Expressed
in E. coli.
Panel A shows the purified recombinant enzymes involved in the
biosynthesis of MVA from the acetate. Lanes 1 and 6, molecular
mass markers; lane 2, ACS; lane 3, AACT; lane 4, HMGS; lane 5,
ꢀHMGR. Panel B shows the purified recombinant enzyme involved
in the biosynthesis of ent-kaurene from MVA. Lanes 1 and 8,
molecular mass markers; lane 2, MVK; lane 3, PMVK; lane 4,
DMDC; lane 5, IPI; lane 6, GGPS; lane 7, JsCPS/KS. Panel C
shows the E. coli cell lysate containing oxidases. Lanes 1 and 4,
molecular mass markers; lane 2, 20ox 3; lane 3, 3ox 1. Each
enzyme is indicated by an arrow.
(ꢀ2–664). The resulting eight ORFs and the bacterial
GGPS gene amplified by PCR were introduced into the
pQE-30 vector. All recombinant proteins were produced
in E. coli as His-tagged fusion proteins and purified by
Ni-NTA resin (Fig. 2A and B). The DDBJ accession
number of each cDNA is listed in Supplemental Table S1.
The basic enzyme cocktail in this context indicates a
mixed enzyme solution containing recombinant en-
zymes and cofactors in one test tube for synthesizing
prenyl diphosphates. We first carried out the in vitro
synthesis of GGDP from MVA by using an enzyme
cocktail containing five recombinant enzymes (MVK,
PMVK, DMDC, IPI, and GGPS), Mg2þ, and ATP.
Uniformly 13C-labeled MVA was used as the substrate
Synthesis of [U-13C20] ent-kaurene by using a specific
enzyme cocktail
ent-Kaurene is a crucial hydrocarbon intermediate in
GA synthesis, and is biosynthesized by a two-step

132
Y. SUGAI et al.
B
275
A
100
50
257
100
50
272
292
229
246
100
200
300
100
200
300
m/z
m/z
D
C
319
300
100
50
100
50
241
258
348
329
360
380
100
200
300
400
100
200
300
400
m/z
m/z
302
284
F
E
100
50
100
50
241
225
437
418
100
200
300
400
100
200
300
400
m/z
m/z
Fig. 4. Mass Spectra of ent-Kaurene, the GA12 Methyl Ester Derivative and GA4 Methyl-TMSi Derivative.
13
Panels A and B show the spectra of enzymatically synthesized [U-
C ] ent-kaurene and non-labeled ent-kaurene, respectively.
20
A, tR ¼ 6:38 min, m=z 292 ([M]þ, 72%); B, tR ¼ 6:38 min, m=z 272 ([M]þ, 75%). Panels C and D show the enzymatically synthesized [U-13C20
]
GA12 methyl ester and non-labeled GA12 methyl ester, respectively. C, tR ¼ 9:18 min, m=z 380 ([M]þ, 3%); D, tR ¼ 9:18 min, m=z 360 ([M]þ,
13
2%). Panels E and F show the enzymatically synthesized [U-
C ] GA4 methyl ester trimethylsilyl ether and non-labeled GA4 methyl ester
19
trimethylsilyl ether, respectively. E, tR ¼ 7:54 min, m=z 437 ([M]þ, 35%); D, tR ¼ 7:54 min, m=z 418 ([M]þ, 35%).
cyclization reaction with GGDP. The enzymatic syn-
thesis of GGDP from an acetate via MVA has been
described in the previous section. We used a bifunc-
tional ent-kaurene synthase from the liverwort, J.
subulata (JsCPS/KS), to efficiently convert GGDP into
ent-kaurene in vitro, and prepared recombinant JsCPS/
KS (Fig. 2B).9)
The enzyme cocktail specific to ent-kaurene synthesis
was prepared by adding JsCPS/KS to the GGDP basic
enzyme cocktail using [U-13C6] MVA. After incubating
the reaction mixture, ent-kaurene was trapped in cyclo-
hexane overlayed on the aqueous phase in a test tube and
then extracted again with cyclohexane. The yield of
[U-13C20] ent-kaurene from [U-13C6] MVA reached 3%
by this method. We also synthesized [U-13C20] ent-
kaurene from [U-13C2] acetate by using ten recombinant
enzymes and achieved a total synthesis with a 0.3%
yield. The GC-MS data demonstrate the successful
synthesis of [U-13C20] ent-kaurene from [U-13C6] MVA
by comparing with a natural authentic specimen. The
molecular ion peak of [U-13C20] ent-kaurene had an m=z
value of 292 corresponding to twenty 13C isotope atoms.
The other fragment ion peaks showed good agreement
with fragment ions completely labeled with 13C isotopes
(Fig. 4A and B, Table 1). A scaled-up synthesis was
performed for a 13C-NMR analysis of [U-13C20] ent-
kaurene. The ent-kaurene synthetic cocktail supplied
with 3.7 mg of [U-13C6] MVA was incubated to afford
150 mg of [U-13C20] ent-kaurene. The proton-decoupled
13C-NMR spectrum of [U-13C20] ent-kaurene is shown
in Fig. 5. The carbon atoms completely labeled with the
13C isotope gave high-intensity peaks and carbon-carbon
coupling in the 13C-NMR analysis. 13C-NMR of [U-
13
C ] ent-kaurene (150 MHz; CDCl3) ꢀ: 17.6 (d, C-20,
20
J10{20 35 Hz), 18.2 (dd, C-11, J9{11,11{12 33 Hz), 18.7 (dd,
C-2, J1{2,2{3 33 Hz), 20.3 (dd, C-6, J5{6,6{7 34 Hz), 21.6
(d, C-19, J4{19 35 Hz), 33.2 (dddd, C-4 J3{4,4{5,4{18,4{19
34 Hz), 33.3 (dd, C-12, J11{12,12{13 32 Hz), 33.7 (d, C-18,

Enzymatic Total Synthesis of Gibberellins
Table 1. The GC-MS Data for the Synthetic Products
133
genes in the present work were synchronously expressed
in the Pichia transformant, and the microsome was used
for synthesizing GA12 from ent-kaurene.
Analyzed compound
tR (min)
MS m=z (rel. int., %)
ent-kaurene 12C (natural) 603800
272 ([M]þ, 75), 257 (100),
229 (70), 213 (37)
292 ([M]þ, 72), 276 (100),
246 (56), 229 (38)
272 ([M]þ, 72), 257 (100),
229 (60), 213 (29)
These two genes were inserted in tandem with the
pPICZ vector and transformed into Pichia which already
had a P450 reductase gene in the AOX promoter
region.19) The transformant by two-step selection, using
distinct concentrations of zeocin (0.1 mg/mL and 1.0
mg/mL) was incubated to induce both the CYP701A
and CYP88A genes by adding MeOH. The microsomal
fraction prepared by centrifugation was incubated with
[U-13C20] ent-kaurene in a test tube at 28 ꢀC for 2 h. The
products were extracted with EtOAc and analyzed by
GC-MS, after their derivatization with diazomethane.
The mass chromatogram monitored at m=z 348 and 319
detected the [U-13C20] GA12 methyl ester at 9.18 min
by GC. The full-scan mass spectrum indicated that
all carbon atoms of the GA12 methyl ester were
substituted by the 13C isotope without any natural
abundance (Fig. 4C and D, Table 1). The yeast micro-
somal fraction containing the CYP701A and CYP88A
enzymes converted ent-kaurene to GA12. Such synthetic
intermediates as ent-kaurenoic acid and 7ꢁ-hydroxy-ent-
kaurenoic acid were not detected by GC-MS. It was
impossible to synthesize GA12 from MVA or acetate in
one step when these P450 enzymes were added to the
ent-kaurene synthetic enzyme cocktail.
13C-labeled
authentic
603800
603800
a
GA12
12C (natural) 901800
360 ([M]þ, 2), 328 (25),
300 (100), 241 (34)
380 ([M]þ, 3), 348 (28),
319 (100), 258 (30)
360 ([M]þ, 3), 328 (29),
300 (100), 241 (30)
13C-labeled
authentic
901800
901800
b
GA4
12C (natural) 705400
418 ([M]þ, 35), 328 (34),
284 (100), 225 (66)
437 ([M]þ, 35), 347 (35),
302 (100), 241 (60)
418 ([M]þ, 34), 328 (37),
284 (100), 225 (64)
13C-labeled
authentic
705400
705400
aGA12 was analyzed as the methyl ester.
bGA4 was analyzed as the methyl ester trimethylsilyl ether.
Cyclohexane
CDCl3
12
13
11
17
20
10
1
4
14
16
9
2
3
8
15
4,12,18
7
5
6
2,11,20
19,6
Enzymatic synthesis of GA4 from acetate
18
19
The biosynthetic conversion of GA12 and GA9 into
GA9 and GA4 was catalyzed by 20ox and 3ox,
respectively (Fig. 1B). Although fungal cyctochrome
P450 monooxygenases also catalyzed the same reac-
tions, two dioxygenases were selected as components of
the enzyme cocktail. Both oxidases are 2-oxoglutarate-
dependent dioxygenases and require 2-oxoglutarate,
ascorbate, and Fe2þ. Among the numerous 20ox and
3ox enzymes which have been cloned from higher plants
and functionally characterized, we selected Arabidopsis
20ox 3 and 3ox 1 for the enzymatic synthesis of GA4
from GA12. These 20ox 3 and 3ox 1 enzymes converted
GA12 to GA9 and GA9 to GA4, respectively, and
minimally accumulated intermediates and shunt product
such as GA25.25) These recombinant enzymes were
produced in E. coli in the MBP fusion form13) and used
for the enzymatic synthesis without purification
(Fig. 2C).
As described above, we used two enzyme cocktails
which separately synthesized ent-kaurene from acetate
and GA12 from ent-kaurene. After confirming the GA12
synthesis by GC-MS, the 20ox 3 and 3ox 1 enzymes
were added to the residual reaction mixture in the same
test tube and incubated at 28 ꢀC for 2 h. The EtOAc
extract from the enzyme reaction mixture was derivat-
ized by MSTFA and diazomethane. Figure 4E and F
show the result of the GC-MS analysis of the product,
the [U-13C19] GA4 methyl ester trimethylsilyl ether. A
comparison with the authentic GA4 methyl ester
trimethylsilyl ether showed the retention time of both
derivatives by GC to be the same, while the molecular
ion peak and fragmentation ion peaks of the [U-13C19]
GA4 methyl ester trimethylsilyl ether were both shifted
according to the number of 13C atoms. The total yield of
GA4 from the acetate was estimated to be 0.15%.
3,1,7,
14,10
17
8,13
5,9
15
16
150
125
100
75
50
25 (ppm)
Fig. 5. 13C-NMR (150 MHz, CDCl3) Spectrum of [U-13C20] ent-
Kaurene Synthesized by the Enzyme Cocktail.
J4{18 36 Hz), 39.9 (dd, C-14, J8{14,13{14 32 Hz), 39.3
(dddd, C-10, J1{10,5{10,9{10,10{20 34 Hz), 40.5 (dd, C-7,
J6{7,7{8 34 Hz), 42.1 (dd, C-1, J1{2,1{10 35 Hz), 44.1 (ddd,
C-13, J12{13,13{14,13{16 34 Hz), 44.2 (dd, c-3, J2{3,3{4
33 Hz), 44.2 (dddd, C-8, J7{8,8{9,8{14,8{15 34 Hz), 49.3
(dd, C-15, J8{15,15{16 37 Hz), 56.1 (ddd, C-5, J4{5,5{6,5{10
33 Hz), 56.3 (ddd, C-9, J8{9,9{10,9{11 33 Hz), 102.8 (d,
C-17, J16{17 74 Hz), 156.1 (ddd, C-16, J13{16,15{16 37 Hz,
J16{17 74 Hz).
Enzymatic synthesis of GA12 with CYP701A and
CYP88A
The oxidation of ent-kaurene and ent-kaurenoic acid
on the GA biosynthetic pathway of higher plants is
catalyzed by two distinct P450 monooxygenases, KO
and KAO, that produce ent-kaurenoic acid and GA12,
respectively (Fig. 1B). It was found that recombinant
CYP701A and CYP88A catalyzed the respective steps.
Our previous report has demonstrated that both P450
genes were heterologously expressed in a Pichia yeast
system, and a functional analysis was performed in
vitro.16,17,19) Both CYP701A (pea) and CYP88A (lettuce)

134
Y. SUGAI et al.
Discussion
various terpenoids in a natural biosynthetic manner.
Since the GA biosynthetic genes in this work were
cloned from several plant and fungal materials,7) the
total synthesis of bioactive GA4 from an acetate could
be achieved by using only recombinant enzymes,
cofactors and substrates. There are two procedures for
synthesizing GA4 from GA12 by applying enzymes:
using dioxygenases from plants and using P450s from
fungi. We used here dioxygenases for the GA4 synthesis,
because preparing a synthetic cocktail with various types
of enzyme will be applicable to the enzymatic synthesis
of other terpenoids.
Our enzymatic total synthesis system could also
provide fully isotopic labeled terpenoids. We have
demonstrated the synthesis of fully 13C-labeled GA4
and its intermediates without dilution of natural abun-
dance. The 13C-NMR and EI-MS data for [U-13C20] ent-
kaurene indicated 13C-enrichiment in which ent-kaurene
was not diluted with 12C during the enzyme reaction
(Figs. 4A and 5). The enzymatic synthesis from ent-
kaurene to GA12 and GA4 also proceeded without
dilution. Experimental techniques using 13C-enrichment
and high-resolution NMR have been performed in
metabolic studies carried out on living cells and organs
of mammals.33–35) For example, studies on the last
two steps of the MEP pathway have demonstrated that
the metabolism of 13C-enriched substrates and direct
13C-NMR measurement of the cell lysate revealed the
products from enzyme reactions.36,37) This approach also
provides a new strategy to determine the function of
unknown biosynthetic genes. The methods presented
here will enable the enzymatic total synthesis of other
terpenoids and uniform labeling of stable isotopes. Our
strategy provides a new insight into the biosynthesis and
metabolomics of terpenoids.
The goal of an enzymatic total synthesis of bioactive
GA4 from acetate was achieved by successful produc-
tion of recombinant proteins and adjustment of reaction
conditions. All the recombinant enzymes involved in
the synthesis of GA4 from acetate were prepared by a
heterologous expression system using E. coli and yeast
(Fig. 2). These fourteen enzymes for GA4 synthesis
included soluble enzymes and membrane-bound P450s.
The soluble enzymes were produced in E. coli in the
His-tagged and MBP fusion forms. The appropriate
combination of the expression constructs and host cells
was critical for mass production of the recombinant
enzymes. We have previously established the heterolo-
gous production of recombinant P450 enzymes for plant
hormone biosynthesis.16,17,19) Two independent P450
enzymes were synchronously produced in a single
transformant of Pichia yeast in this present work. This
procedure involved the efficient preparation of an
enzyme cocktail for GA12 synthesis rather than having
to prepare two microsomal fractions separately contain-
ing KO and KAO enzymes. Such synchronous expres-
sion of multiple P450 products has been reported by
using human CYP1A1, CYP2B6 and CYP2C19 in
planta.26) These three independent P450s were active
in in vivo experiments, whereas the reactions catalyzed
by our P450s (CYP701A3 and CYP88A) operated in
vitro by using the microsomal fraction prepared from
Pichia cells.
Cell-free systems have been used for several decades
in numerous biosynthetic studies and for the in vivo
synthesis of such natural products as biosynthetic
intermediates. GA biosynthetic studies have used cell-
free systems prepared from immature seeds of plants for
GA biosynthesis and for the preparation of radioisotope-
labeled biosynthetic intermediates.27–29) However, natu-
ral materials may often be difficult to prepare in a cell-
free extract on a large scale. In addition, it is difficult
to standardize cell-free systems because the quality and
characteristics of enzyme extracts are not consistent
with separate preparations from intact materials or
organisms. Furthermore, the enzyme reactions and
processes in a cell-free system cannot be regulated; for
example, a natural-based cell-free system cannot selec-
tively accumulate biosynthetic intermediates.30,31) Re-
combinant enzyme-based cocktails are powerful tools to
solve these problems in the synthesis of natural
products. If farnesyl diphosphate is necessary to study
sesquiterpene biosynthesis, farnesyl diphosphate syn-
thase is used instead of GGPS in the cocktail. The
substitution of farnesyl diphosphate synthase and amor-
phadiene synthase for GGPS and JsCPS/KS in the
enzyme cocktail successfully produced amorphadiene
(Supplemental Fig. S1A). Another example is taxadiene,
a diterpene hydrocarbon, which could be synthesized
when taxadiene synthase was added to the cocktail
instead of JsCPS/KS (Supplemental Fig. S1C). Diverse
structural platforms of terpenoids are also determined by
terpene cyclases in nature; for example, rice has several
diterpene cyclase genes in the genome.32) Each cyclase
plays a role in strictly determining the structures of GA
and phytoalexins. Our enzyme cocktail thus has the
ability to generate multiple synthetic reactions for
Acknowledgments
We are grateful to Dr. S. M. Swain (CISRO-Plant
Industry, Australia) for providing the genes encoding
CYP701A. We thank Drs. H. Yamashita, W. Matsumoto
and J. Shibasaki of ADEKA Corp. (Tokyo, Japan) for
technical support regarding the fermentation production
of [U-13C6] MVA. This work was supported in part
by Grant-in-Aids for Scientific Research (C) to H. K.
(18580103) and for JSPS Fellows to Y. S. (227977) from
Japan Society for the Promotion of Science.
References
1) Gershenzon J and Dudareva N, Nat. Chem. Biol., 3, 408–414
(2007).
2) Enserink M, Science, 307, 33 (2005).
3) Holton RA, Juo RR, Kim HB, Williams AD, Harusawa S,
Lowenthal RE, and Yogai S, J. Am. Chem. Soc., 110, 6558–
6560 (1988).
4) Holton RA, Somoza C, Kim HB, Liang F, Biediger RJ,
Boatman PD, Shindo M, Smith CC, Kim S, Nadizadeh H,
Suzuki Y, Tao C, Vu P, Tang S, Zhang P, Murthi KK, Gentile
LN, and Liu JH, J. Am. Chem. Soc., 116, 1597–1598 (1994a).
5) Holton RA, Kim HB, Somoza C, Liang F, Biediger RJ,
Boatman PD, Shindo M, Smith CC, Kim S, Nadizadeh H,
Suzuki Y, Tao C, Vu P, Tang S, Zhang P, Murthi KK, Gentile
LN, and Liu JH, J. Am. Chem. Soc., 116, 1599–1600 (1994b).
6) Eisenreich W, Bacher A, Arigoni D, and Rohdich F, Cell. Mol.
Life Sci., 61, 1401–1426 (2004).

Enzymatic Total Synthesis of Gibberellins
135
7) Yamaguchi S, Annu. Rev. Plant Biol., 59, 225–251 (2008).
8) Kawaide H, Biosci. Biotechnol. Biochem., 70, 583–590
(2006).
21) Swain SM, Singh DP, Helliwell CA, and Poole AT, Plant Cell
Physiol., 46, 284–291 (2005).
22) Gray RT, Org. Mass Spectrom., 9, 346–349 (1974).
23) Croxen R, Goosey MW, Koen JPR, and Hargreaves JA,
Microbiol., 140, 2363–2370 (1994).
9) Kawaide H, Hayashi K, Kawanabe R, Sakigi Y, Matsuo A,
Natsume M, and Nozaki H, FEBS J., 278, 123–133 (2011).
10) Hayashi K, Kawaide H, Notomi M, Sakigi Y, Matsuo A, and
Nozaki H, FEBS Lett., 580, 6175–6181 (2006).
24) Yamashita H, Fragrance J. (in Japanese), 28, 62–65 (2000).
25) Phillips AL, Ward DA, Uknes S, Appleford NEJ, Lange T,
Huttly AK, Gaskin P, Graebe JE, and Hedden P, Plant Physiol.,
108, 1049–1057 (1995).
11) Helliwell CA, Sheldon CC, Olive MR, Walker AR, Zeevaart
JAD, Peacock WJ, and Dennis E, Proc. Natl. Acad. Sci. USA,
95, 9019–9024 (1998).
26) Inui H, Shiota N, Motoi U, Ido Y, Inoue T, Kodama T, Ohkawa
Y, and Ohkawa H, J. Pesticide Sci., 26, 28–40 (2001).
27) Graebe JE, Dennis DT, Upper CD, and West CA, J. Biol.
Chem., 240, 1847–1854 (1965).
12) Kawaide H, Imai R, Sassa T, and Kamiya Y, J. Biol. Chem.,
272, 21706–21712 (1997).
13) Toyomasu T, Kawaide H, Mitsuhashi W, Inoue Y, and Kamiya
Y, Plant Physiol., 118, 1517–1523 (1998).
28) Oster MO and West CA, Arch. Biochem. Biophys., 127, 112–
123 (1968).
14) Rebers M, Kaneta T, Kawaide H, Yamaguchi S, Yang YY, Imai
R, Sekimoto H, and Kamiya Y, Plant J., 17, 241–250 (1999).
15) Itoh H, Tanaka-Ueguchi M, Kawaide H, Chen X, Kamiya Y,
and Matsuoka M, Plant J., 20, 15–24 (1999).
29) Kamiya Y and Graebe JE, Phytochem., 22, 681–689 (1983).
30) Hedden P and Graebe JE, Phytochem., 20, 1011–1015 (1981).
31) Graebe JE, Planta, 102, 261–271 (1972).
32) Toyomasu T, Biosci. Biotechnol. Biochem., 72, 1168–1175
(2008).
16) Sawada Y, Katsumata T, Kitamura J, Kawaide H, Nakajima M,
Asami T, Nakaminami K, Kurahashi T, Mitsuhashi W, Inoue Y,
and Toyomasu T, J. Exp. Bot., 59, 3383–3393 (2008).
17) Ko K, Lin F, Katsumata T, Sugai Y, Miyazaki S, Kawaide H,
Okada K, Nojiri H, and Yamane H, Biosci. Biotechnol.
Biochem., 72, 3285–3288 (2008).
33) Eisenreich W and Bacher A, Phytochem., 68, 2799–2815
(2007).
34) Rodrigues TB, Fonseca CP, Castro MMCA, Cerdan S, and
Geraldes CFGC, Q. J. Nucl. Med. Mol. Imaging, 53, 631–645
(2009).
18) Hayashi K, Horie K, Hiwatashi Y, Kawaide H, Yamaguchi S,
Hanada A, Nakashima T, Nakajima M, Mander LN, Yamane H,
Hasebe M, and Nozaki H, Plant Physiol., 153, 1085–1097
(2010).
35) Eakin RT, Morgan LO, Gregg CT, and Matwiyoff NA, FEBS
Lett., 28, 259–264 (1972).
36) Hecht S, Eisenreich W, Adam P, Amslinger S, Kis K, Bacher A,
Arigoni D, and Rohdich F, Proc. Natl. Acad. Sci. USA, 98,
14837–14842 (2001).
19) Katsumata T, Hasegawa A, Fujiwara T, Komatsu T, Notomi M,
Abe H, Natsume M, and Kawaide H, Biosci. Biotechnol.
Biochem., 72, 2110–2117 (2008).
37) Rohdich F, Hecht S, Gartner K, Adam P, Krieger C, Amslinger
¨
20) Ohnuma S, Suzuki M, and Nishino T, J. Biol. Chem., 269,
14792–14797 (1994).
S, Arigoni D, Bacher A, and Eisenreich W, Proc. Natl. Acad.
Sci. USA, 99, 1158–1163 (2002).