Allyl/propenyl phenol synthases from the creosote bush and engineering production of specialty/commodity chemicals, eugenol/isoeugenol, in Escherichia coli

Article abstract of DOI:10.1016/j.abb.2013.10.019

The creosote bush (Larrea tridentata) harbors members of the monolignol acyltransferase, allylphenol synthase, and propenylphenol synthase gene families, whose products together are able to catalyze distinct regiospecific conversions of various monolignols into their corresponding allyl- and propenyl-phenols, respectively. In this study, co-expression of a monolignol acyltransferase with either substrate versatile allylphenol or propenylphenol synthases in Escherichia coli established that various monolignol substrates were efficiently converted into their corresponding allyl/propenyl phenols, as well as providing proof of concept for efficacious conversion in a bacterial platform. This capability thus potentially provides an alternate source to these important plant phytochemicals, whether for flavor/fragrance and fine chemicals, or ultimately as commodities, e.g.; for renewable energy or other intermediate chemical purposes. Previous reports had indicated that specific and highly conserved amino acid residues 84 (Phe or Val) and 87 (Ile or Tyr) of two highly homologous allyl/propenyl phenol synthases (circa 96% identity) from a Clarkia species mainly dictate their distinct regiospecific catalyzed conversions to afford either allyl- or propenyl-phenols, respectively. However, several other allyl/propenyl phenol synthase homologs isolated by us have established that the two corresponding amino acid 84 and 87 residues are not, in fact, conserved.

Full text of DOI:10.1016/j.abb.2013.10.019

Archives of Biochemistry and Biophysics 541 (2014) 37–46
Contents lists available at ScienceDirect
Archives of Biochemistry and Biophysics
journal homepage: www.elsevier.com/locate/yabbi
Allyl/propenyl phenol synthases from the creosote bush and engineering
production of specialty/commodity chemicals, eugenol/isoeugenol,
in Escherichia coli
Sung-Jin Kim, Daniel G. Vassão 1, Syed G.A. Moinuddin, Diana L. Bedgar, Laurence B. Davin,
⇑
Norman G. Lewis
Institute of Biological Chemistry, Washington State University, Pullman, WA 99164-6340, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
The creosote bush (Larrea tridentata) harbors members of the monolignol acyltransferase, allylphenol
synthase, and propenylphenol synthase gene families, whose products together are able to catalyze distinct
regiospecific conversions of various monolignols into their corresponding allyl- and propenyl-phenols,
respectively. In this study, co-expression of a monolignol acyltransferase with either substrate versatile
allylphenol or propenylphenol synthases in Escherichia coli established that various monolignol substrates
were efficiently converted into their corresponding allyl/propenyl phenols, as well as providing proof of
concept for efficacious conversion in a bacterial platform. This capability thus potentially provides an
alternate source to these important plant phytochemicals, whether for flavor/fragrance and fine chemi-
cals, or ultimately as commodities, e.g., for renewable energy or other intermediate chemical purposes.
Previous reports had indicated that specific and highly conserved amino acid residues 84 (Phe or Val)
and 87 (Ile or Tyr) of two highly homologous allyl/propenyl phenol synthases (circa 96% identity) from
a Clarkia species mainly dictate their distinct regiospecific catalyzed conversions to afford either allyl-
or propenyl-phenols, respectively. However, several other allyl/propenyl phenol synthase homologs iso-
lated by us have established that the two corresponding amino acid 84 and 87 residues are not, in fact,
conserved.
Received 5 September 2013
and in revised form 24 October 2013
Available online 1 November 2013
Keywords:
Allylphenol synthase
Propenylphenol synthase
Acyltransferase
Allylphenol
Propenylphenol
Eugenol
Isoeugenol
Ó 2013 Elsevier Inc. All rights reserved.
Introduction
necting European consumers to spice producers in Asia (mainly
India and the Indonesian archipelago), as well as in the Eastern
Monomeric allylphenols and propenylphenols are important
phenylpropanoid constituents of various essential oils/flavors of
several herbs, spices and flowers, where they are presumed to
serve mainly in defensive purposes against herbivores and para-
sites and, in specific cases, as pollinator attractants, e.g., methylc-
havicol (1, Fig. 1) [1, and references therein]. Many plant species/
tissues accumulating such compounds in significant amount have
also found uses as spices for millennia. For example, eugenol (3)
is the major component in clove oil (up to 89%) [2] which also
contains smaller amounts of eugenyl acetate (4, ꢀ5.5%) and the
terpenoid b-caryophyllene (6, ꢀ1.5%), while anethole (5) is a main
contributor to the aroma of anise [3]. Indeed, cloves were one of
the most highly sought after spices in the Age of Exploration,
when European navigators created maritime trade routes con-
African coast (e.g., Tanzania and Madagascar), where they were
later introduced. Clove oils are produced from the dried flower
buds, the leaves and stems of the clove tree (Syzygium
aromaticum, or Eugenia caryophyllata), generating a worldwide
production of ca 2000 ton year–1 [4], with clove bud oil being
the most expensive. The oil commands a gross market value of
circa US$ 30–70 million year–1 [5].
In general, however, monomeric allyl/propenyl phenols are
quite widely distributed within the plant kingdom, particularly in
the angiosperms with very few reports in other groups. As dis-
cussed earlier [1], their occurrence chemotaxonomically in other
lineages is often overlooked due to their amounts frequently being
very minor (<0.05% of tissue dry weight) and generally only being
reported as part of very complex mixtures of dozens of compo-
nents (i.e., that are listed in tables of complex mixtures). Thus, they
are often not ‘‘captured’’ in various search algorithms for chemo-
taxonomical classification purposes. Nevertheless, there are a few
reports of allyl/propenyl phenols occurring in algae [6–8], liver-
worts (bryophytes) [9–12], horsetail (pteridophytes) [13] and gym-
nosperms (e.g., Picea abies [14], Pinus sylvestris [14] and Juniper
⇑
Corresponding author. Fax: +1 509 335 8206.
E-mail address: lewisn@wsu.edu (N.G. Lewis).
Current address: Department of Biochemistry, Max Planck Institute for Chemical
1
Ecology, Hans-Knoell-Str. 8, Jena 07745, Germany.
0003-9861/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.abb.2013.10.019

38
S.-J. Kim et al. / Archives of Biochemistry and Biophysics 541 (2014) 37–46
Fig. 1. Allyl/propenyl phenol derived monomers and dimers (1–5, 7–14) and b-caryophyllene (6).
Fig. 2. Abbreviated biosynthetic pathway from phenylalanine (15) to allyl (2, 3, 31–33) and propenyl (26–30) phenols. APS, allylphenol synthase; CAAT, cinnamyl alcohol
acyltransferase; PPS, propenylphenol synthase.
brevifolia [15]), suggesting a rather broad level of occurrence in
planta.
Allyl/propenyl phenols are products of the phenylpropanoid
(C6C3) pathway from Phe (15, Fig. 2), and the biochemistry of their
formation in planta has become better understood in recent years
[1,19,21–24]. In particular, these compounds share, as common
precursors with lignin and various lignans, hydroxycinnamyl alco-
hols (monolignols), such as p-coumaryl (16) and coniferyl (18)
alcohols. Indeed, enzymes catalyzing the two last steps in allyl/
propenyl phenol formation have been characterized recently from
plants, including the creosote bush (L. tridentata) [19,20], basil
(Ocimum basilicum) [21,22], petunia (Petunia ꢁ hybrida) [21], Clar-
kia breweri [23], and anise (Pimpinella anisum) [24]. Of these, the
substrate-versatile allylphenol synthase (APS) in L. tridentata
(LtAPS1, formerly named Chavicol/Eugenol Synthase, LtCES1 [19])
is the most catalytically efficient of all APSs described in vitro thus
far. The discovery and mode of its catalytic action were deduced by
comparison to other members of the homologous pinoresinol-lar-
iciresinol [25–27], phenylcoumaran benzylic ether [28], isoflavone
[29–31], pterocarpan and leucoanthocyanidin reductases [32,33],
many of which we have extensively studied, including protein iso-
lation, gene cloning, determination and modeling of 3D structures,
as well as mechanistic studies [25–28].
Monomeric allylphenols and propenylphenols can sometimes
also coexist within the same plant species, e.g., methylchavicol
(1) and anethole (5) in fennel (Foeniculum vulgare) [16]. Addition-
ally, varying degrees of aromatic ring oxygenation and other mod-
ifications can give rise to an interesting level of structural diversity
observed within this metabolic class, e.g., to afford more elabo-
rately substituted compounds, such as the allylphenol-derived api-
ol (7), dillapiol (8) and myristicin (9), in Piper regnellii [17]. Various
plant species can also accumulate both allylphenols and presumed
propenylphenol-derived dimeric lignans, including P. regnellii,
where lignans, such as (+)-conocarpan (10), are present in roughly
equivalent amounts to the monomers 7–9 [17].
On the other hand, some plant species can apparently accumu-
late only either monomers [e.g., eugenol (3) in cloves], or their pre-
sumed coupled products, the side-chain deoxygenated lignans. An
example of the latter is the 8–80-linked lignans in the creosote bush
(Larrea tridentata), a North American desert shrub [18]. It produces
large amounts of 9,90-deoxygenated lignans [e.g., nordihydroguaia-
retic acid (11), a strong antioxidant whose derivatives (e.g., 12) are
showing promise as anticancer medicinals, as well as nor-isoguai-
acin (13) and larreatricin (14)]. A gene encoding an allylphenol
synthase (APS)2 was reported [19] following its discovery in 1999
[20].
In this study, the isolation and characterization of a L. tridentata
propenylphenol synthase (LtPPS1) catalyzing regiospecific forma-
tion of propenylphenols is described, as well as isolation of genes
encoding two acyltransferases (LtCAAT1/2) that generate acylated
monolignols; these enzymes and gene sequences were previously
preliminarily reported elsewhere [1,20,34]. Additionally, both L.
tridentata APS and PPS were separately expressed in E. coli,
2
Abbreviations used: APS, allylphenol synthase; CAAT, cinnamyl alcohol acyltrans-
ferase; PPS, propenylphenol synthase.

S.-J. Kim et al. / Archives of Biochemistry and Biophysics 541 (2014) 37–46
39
together with LtCAAT1, in order to assess their relative efficacies
using a variety of substrates. This was in order to begin to establish
proof-of-concept for large scale transformation of monolignols (or
related feedstock) into their allyl/propenylphenol counterparts in
bacteria. It was established that all three enzymes are substrate
versatile, and together can process different monolignols to afford
the corresponding allyl/propenyl phenols in vitro in high yields.
This substrate versatility is in accordance with that known for
most other phenylpropanoid pathway enzymes (e.g., cinnamyl
alcohol dehydrogenases, CAD [35] and 4-coumarate:CoA ligases,
4CL [36]).
It was also instructive to investigate factors putatively signifi-
cantly affecting enzyme regiospecificity, as it had been reported
that amino residues in Clarkia breweri CbEGS1 and CbIGS1 (specif-
ically F84/I87 for CbEGS1 and V84/Y87 for CbIGS1) were the main
determinants for regiospecific catalysis as either an APS or a PPS
[23]. We could not, however, substantiate this with the APS/PPS
homologs examined herein.
DNA sequencing. Quantitation of DNA, RNA, and protein analyses
were done using a Lambda 6 UV–visible spectrophotometer (Per-
kin–Elmer).
Chemical syntheses
The synthesis of p-coumaryl (21), caffeyl (22), coniferyl (23)
5-hydroxyconiferyl (24), and sinapyl (25) 9-acetate, as well as of
p-anol (26), 3-hydroxy-p-anol (27), 5-hydroxyisoeugenol (29),
5-methoxyisoeugenol (30), chavicol (2), 3-hydroxychavicol (31),
5-hydroxyeugenol (32) and 5-methoxyeugenol (33) is described
in the Supplementary Material Section.
cDNA isolation and cloning of L. tridentata cinnamyl alcohol
acyltransferases (CAAT)
Degenerate PCR primers were designed based on conserved re-
gions of the peptide sequences of a petunia (Petunia hybrida) acetyl
CoA:coniferyl alcohol acetyl transferase (PhCFAT; GenBank acces-
sion No. DQ767969) and two uncharacterized Arabidopsis thaliana
HXXXD-type acyl-transferase-like proteins (At1g78990 and
At5g16410) having highest sequence homology to PhCFAT (ꢀ68/
ꢀ62% similarity and ꢀ54/ꢀ46% identity, respectively) [37]. L.
tridentata total RNA was prepared from leaf tissues from 2 to
3 month old plants using the RNeasy Plant Mini Kit (Qiagen), with
this then utilized for first-strand cDNA synthesis with a Super-
ScriptÒ III First-Strand Synthesis System for RT-PCR (Invitrogen).
Degenerate PCR amplification was performed (degenerate primer
set: AT_Degen_For, 50-AARYTNGGHATHGTBGTBGAYGG-30; AT_De-
gen_Rev, 50-GGCATNGGCATNACRTANCC-30) with 12 different tem-
peratures of the gradient PCR program: 96 °C (3 min), followed by
45 cycles of denaturation at 96 °C (1 min), annealing 40 °C – 60 °C
(1 min), elongation 72 °C for (1 min), and a final incubation at 72 °C
(7 min) by GoTaqÒ DNA polymerase (Promega). An ꢀ0.4 kb PCR
product (formed with annealing temperature between 48.4 and
58.4 °C) was cloned into the pCRÒ2.1-TOPOÒ vector (Invitrogen),
then sequenced and Blast-searched (http://blast.ncbi.nlm.nih.gov/
Blast.cgi). Using the partial cDNA sequences obtained as putative
acyl transferases, an internal gene-specific primer set (AT_Int_For,
50-ATGCATGAACTGACTGAGAAACCATTGAG-30; AT_Int_Rev, 50-
CAACTCAATGGTTTCTCAGTCAGTTCATGC-30) was designed and
used for 50- and 30-RACE with a BD SMART™ RACE cDNA Amplifi-
cation Kit (Clontech) to afford a full-length cDNA which was then
sequenced. Based on the full-length cDNA sequence thus obtained,
a gene-specific primer set (AT_For, 50-ATGGGAGCTGGAGGTGA-
ATTC-30; AT_Rev (30-UTR), 50-CACAAGCAAACCCTATTATGACC-30)
was next designed and used for touch-down PCR amplification as
follows: initial denaturation at 96 °C (3 min), 10 cycles of denatur-
ation (96 °C, 1 min), annealing (1 min) with the temperature
decreasing from 65 to 55 °C with each cycle and then extension
at 72 °C (1 min) followed by 25 cycles (1 min each) of denaturation
(96 °C), annealing (55 °C), extension (72 °C), and finally extension
(72 °C, 7 min). The PCR reaction was accomplished by Expand high-
fidelity enzyme mixture (Roche).
Materials and methods
Plant material
Creosote bush (L. tridentata) seedlings (2–3 months old) were
grown from seeds (Plants of the Southwest, Santa Fe, NM), and
maintained in Washington State University growth chamber facil-
ities at 21 and 16 °C with a 15 h light/9 h dark cycle, a light inten-
sity of ꢀ150 mmol m–2 s–1, and a humidity range from 20 to 35%.
Materials
All solvents used, either HPLC or reagent grade, were purchased
from Mallinckrodt Baker (Phillipsburg, NJ), with GC–MS derivatiza-
tion reagent from Supelco (Bellefonte, PA). All other chemicals
were procured from Sigma–Aldrich (St. Louis, MO), and utilized
without further purification unless otherwise noted. Thin-layer
chromatography utilized PartisilÒ PK5F (Whatman, 20 ꢁ 20 cm,
1 mm, 150 Å) and AL SIL G/UV254 (Whatman, 20 ꢁ 20 cm,
0.25 mm) with UV being employed for visualization; column chro-
matography (CC) was performed on silica gel 60 (EM Science).
5-Methoxyeugenol (33,P90% pure) was obtained in technical
grade from Sigma–Aldrich.
Instrumentation
1H and 13C NMR spectra were acquired on a Varian Mercury 300
and an Inova 500 spectrometer, with chemical shifts given in ppm
relative to Me4Si (1H) or solvent (13C); coupling constant (J) values
are given in Hz. Reversed-phase UPLC separations were carried out
on an ACQUITY Ultra Performance Liquid ChromatographÒ (UPLC,
Waters) BEH C18 column (2.1 ꢁ 50 mm, 1.7
lm, Waters) with
the following elution conditions at a flow rate of 0.2 ml min–1: lin-
ear gradients of CH3CN:3% AcOH in H2O (v/v) from 10:90 to 25:75
in 0.5 min, to 35:65 in 1.5 min, to 40:60 in 3 min with this compo-
sition held for an additional 11 s, then to 100:0 for 0.34 min and fi-
nally to 10:90 in 0.12 min this being held for 1 min. GC–MS
analyses of trimethylsilyloxy derivatives [bis(trimethylsilyl)trifluo-
Two different genes (each 1,392 bp encoding 463 amino acids)
were isolated and named L. tridentata cinnamyl alcohol acyltransfer-
ase 1 (LtCAAT1, GenBank accession No. KF543260) and 2 (LtCAAT2,
GenBank accession No. KF543261).
roacetamidechlorotrimethylsilane (99:1 Supelco, 50
dine (20 l)] were carried out on a HP 6890 Series GC System
equipped with a RESTEK Rtx-5Sil-MS (30 m ꢁ 0.25 mm ꢁ 0.25 m)
ll) and pyri-
l
l
cDNA isolation and cloning of L. tridentata propenylphenol synthase
(PPS)
column and a HP 5973 MS detector (electron impact mode, 70 eV).
PCR amplifications were performed on either a PTC-0220 DNA
Engine Dyad Peltier Thermal Cycler (MJ Research) or a C1000 Ther-
mal Cycler (Bio-Rad). To isolate and purify plasmid DNA, WizardÒ
Plus SV Miniprep Kit (Promega) was used, and an automated DNA
sequencer (Applied Biosystems; model 377) was employed for
Degenerate PCR primers (PPS_Degen_For, 50-GGHGSHACHGGHT
AYATHGG-30; PPS_Degen_Rev, 50-GCNAYRTCYTCYTCRTARTT-30)
were designed, including using the L. tridentata Codon Usage Data-
base (http://www.kazusa.or.jp/codon), and gradient PCR amplifica-

40
S.-J. Kim et al. / Archives of Biochemistry and Biophysics 541 (2014) 37–46
tion as described above for LtCAAT1 was performed. The resulting
PCR product (ꢀ0.6 kb formed with annealing temperature between
45.5 and 51.7 °C) was cloned, sequenced and Blast-searched
(http://blast.ncbi.nlm.nih.gov/Blast.cgi). The partial cDNA se-
quence allowed for the design of an internal gene-specific primer
set (PPS_Int_For, 50-GGTTAGAAGGGCAACTGAAGCAGCTGGAATCC-
30; PPS_Int_Rev, 50-GGATTCCAGCTGCTTCAGTTGCCCTTCTAACC-30)
for 50- and 30-RACE as above. Finally, touch-down PCR amplifica-
tion as above was performed with the following gene-specific pri-
mer set: PPS_For, 50-ATGGCTTGTGAGAAGAGCAAGATACTC-30;
PPS_Rev, 50-TCAAGCAAAAGTTGCCAATTTAGGCTTGGGAGG-30. The
final PCR product, a 957 bp cDNA encoding a 318 amino acid pep-
tide, was named L. tridentata propenylphenol synthase 1 (LtPPS1,
GenBank accession No. KF543264).
described above for LtCAAT1, but with monolignol acetates
(21–25, 0.5 mM) as substrates and NADPH (0.5 mM) as cofactor.
Optimum pH for enzymatic reactions (LtCAAT1/LtAPS1/LtPPS1)
was determined using the following buffer systems (100 mM): cit-
rate, pH 3.35–6.85; MES–KOH, pH 5.5–7.0; potassium phosphate,
pH 6.0–8.0; Tris–HCl, pH 7.0–9.0; Bis-Tris propane, pH 6.5–9.0;
all at 0.5 pH unit intervals, with assays carried out at 30 °C.
Initial velocity kinetics were determined by assaying: LtCAAT1
under standard conditions in citrate buffer (100 mM, pH 4.8) with
18 different monolignol (16–20) concentrations (3.33–500
LtAPS1 in MES–KOH (100 mM, pH 7.0) with 18 different monolig-
nol acetate (21–25) concentrations (3.33 M–3.3 mM); and LtPPS1
in MES–KOH (100 mM, pH 6.5) with 14 different monolignol
acetate (21–25) concentrations (3.33 M–1 mM). Assays were
lM);
l
l
performed in quadruplicate, with controls in the absence of
acetyl-CoA/NADPH.
Expression and purification of recombinant L. tridentata CAAT, APS
and PPS
Co-transformation/expression, monolignol incubation and product
analysis of E. coli expressing LtCAAT1/LtAPS1 or LtCAAT1/LtPPS1
LtCAAT1-, LtAPS1- (formerly named chavicol/eugenol synthase 1
(LtCES1), GenBank accession No. KF543262 [19]) and LtPPS1-coding
genes were individually cloned into a pEXP5-NT/TOPOÒ bacterial
expression vector and transformed into E. coli BL21 (DE3) cells
(Invitrogen). Heterologous gene expressions were then individu-
ally carried out in 2 ꢁ YT broth medium (30 ml) containing carben-
To generate different replication origins and antibiotic selection
markers for each individual gene within E. coli cells, LtCAAT1 was
re-cloned into the pET200/D-TOPOÒ expression vector (Invitrogen)
having a pBR322 origin and a kanamycin resistance gene. L. triden-
tata allylphenol synthase 1 (LtAPS1), formerly named chavicol/
eugenol synthase 1 (LtCES1) [19] was cloned into a pEXP5-NT/
TOPOÒ expression vector containing a pUC origin and an ampicillin
resistance gene. After sequence verification, pET200/D-LtCAAT1
and pEXP5-NT-LtAPS1 constructs were applied at once for co-trans-
formation of E. coli BL21(DE3); analogously, pET200/D-LtCAAT1 and
pEXP5-NT-LtPPS1 constructs were also co-transformed.
icillin (100
l
g ml–1
)
at 37 °C, with IPTG (isopropyl-b-D-
M final concentration when
thiogalactopyranoside) added to 10
l
a density of OD600 = ꢀ0.6 was reached. After 6 h incubation, the
cells were harvested by centrifugation at 3,000 ꢁ g for 20 min at
4 °C, with the pellets frozen and stored at –80 °C. Frozen cell pellets
were thawed at 37 °C for 5 min and resuspended in BugBusterÒ
Protein Extraction Reagent [3 ml with 75 U BenzonaseÒ Nuclease
and 20 KU rLysozymeÒ Solution (Novagen)]. After 5 min incubation
on a platform shaker, cell debris and insoluble materials were re-
moved by centrifugation at 20,000 ꢁ g for 20 min at 4 °C. Individ-
ual recombinant LtCAAT1-His6 / LtAPS1-His6 / LtPPS1-His6 were
purified through POROSÒ 20 metal (Ni2+) chelating affinity chro-
matography as follows: crude extracts of each were individually
directly applied onto 1 ml of POROSÒ 20 Metal Chelating resin (Ap-
plied Biosystems) charged with 100 mM NiSO4, and then washed
with 20 mM Tris–HCl (pH 7.9) containing 500 mM NaCl and
20 mM imidazole. Stepwise elution was carried out using each
3 ml of 40 mM, 60 mM, 80 mM, 100 mM, 120 mM, and 140 mM
imidazole solution in 20 mM Tris–HCl (pH 7.9) containing
500 mM NaCl. Fractions containing the LtCAAT1, LtAPS1 and
LtPPS1 were individually desalted using a PD-10 column (GE
Healthcare) and concentrated with AmiconÒ Ultra-4 30 K (Milli-
pore) membranes. Purified proteins were individually analyzed
on a 4–15% linear gradient SDS–PAGE (Bio-Rad) with silver-stain-
ing visualization. Protein quantification was carried out using
Bio-Rad Protein Assay (Bio-Rad) at OD595 with bovine serum albu-
min (BSA) as standard.
In vivo bioconversion experiments were performed by individ-
ual addition of each monolignol (16, 18 and 20, 0.5 mM final con-
centration) into the various bacterial cell cultures (2 ꢁ YT broth
medium containing both 100 l l
g ml–1 carbenicillin and 50 g ml–1
kanamycin), at the time of IPTG induction, with or without both
cofactors (acetyl-CoA and NADPH, 0.5 mM each). Co-expression/
bioconversion experiments were carried out at different pHs of cul-
ture medium (5.0, 5.5, 6.0, and 7.0) and induced with different IPTG
concentrations (0, 0.1 and 1.0 mM). Aliquots of each culture med-
ium (150
when cell growth and enzyme reactions were stopped by addition
of glacial AcOH (15 l) followed by rapid freezing in liquid N2 until
ll) were taken over an 8 h period at 30 min intervals,
l
analysis. After thawing followed by centrifugation at 20,000 ꢁ g for
5 min at 4 °C, each cell pellet and the culture medium supernatant
were separated. The intermediate monolignol acetates (21, 23 and
25) and propenyl (26, 28 and 30)/allyl (2, 3 and 33) phenol prod-
ucts were quantified using calibration curves generated for each.
Final product confirmation was performed as follows: Et2O
(200
ubles recovered after centrifugation (13,800 ꢁ g, 5 min). An aliquot
of the organic fraction (20 l) was next derivatized with pyridine
(20 l) and BSTFA (50 l), and analyzed by GC–MS with compari-
ll) was added to culture medium (750 ll), with the Et2O sol-
l
Enzyme characterization
l
l
son of retention time and mass spectra to those of derivatized
LtCAAT1: In vitro standard enzymatic assays consisted of
authentic standards as previously described [19].
100 mM citrate buffer (pH 4.8), purified LtCAAT1 in Tris–HCl
(20 mM, pH 7.9) buffer (2
0.5 mM, from a 15 mM MeOH stock solution) and acetyl-CoA
(0.5 mM) in a total volume of 150 l. Enzyme reactions were initi-
ated by addition of substrates, and after 30 min incubation at 30 °C
were terminated by addition of glacial AcOH (15 l). An aliquot
(5 l) of each assay mixture was next subjected to reversed-phase
l
g), individual monolignols (16 – 20,
In vitro coupled reaction of LtCAAT1/LtAPS1 and LtCAAT1/LtPPS1
l
Recombinant proteins co-expressed in E. coli cells were purified
via POROSÒ 20 metal (Ni2+) chelating affinity chromatography and
analyzed on a 4–15% linear gradient SDS–PAGE as described above.
Using the same buffer systems decribed in the ‘‘enzyme character-
ization’’ section, the optimum pH of the coupled reaction was
determined for each monolignol (16–20, 0.5 mM), in presence of
l
l
UPLC analysis with UV detection, with products quantified using
calibration curves generated for each.
LtAPS1 and LtPPS1: In vitro standard enzymatic assays were car-
ried out in 100 mM MES–KOH (pH 7.0 and 6.5, respectively) as
acetyl-CoA (0.5 mM), NADPH (0.5 mM), and 4
LtAPS1 or LtCAAT1/LtPPS1 protein mixtures. After assay termination
lg of LtCAAT1/

S.-J. Kim et al. / Archives of Biochemistry and Biophysics 541 (2014) 37–46
41
with glacial AcOH (15
directly analyzed by UPLC.
l
l), aliquots of the reaction mixtures were
Arabidopsis thaliana HXXXD-type acyl-transferase-like proteins
(At1g78990 and At5g16410) having highest sequence homology
to PhCFAT] [37]. Each full-length cDNA was obtained by 50- and
30-rapid amplification of cDNA ends (RACE), followed by touch-
down PCR amplification using LtCAAT1/2 gene specific primers.
LtCAAT1 and LtCAAT2 encode ꢀ51 kDa peptides with ꢀ70/ꢀ55%
sequence similarity/identity to PhCFAT and ꢀ76 to ꢀ53% similarity
and ꢀ60 to ꢀ33% identity to four members of an uncharacterized
Arabidopsis thaliana HXXXD-type acyl-transferase-like protein
family (At1g32910, At1g78990, At5g16410, and At1g31490).
LtCAAT1/2 also had ꢀ62/ꢀ45% peptide similarity/identity to a
transferase from Zea mays (ZmAT, GeneBank accession No.
NP_001150771), while both had highest sequence similarity/iden-
tity (ꢀ80/ꢀ63%) to a putative anthranilate N-benzoyltransferase
from Ricinus communis (RcANBT, GenBank accession No.
EEF30334). Peptide sequence alignments with these BAHD super-
family members indicated that both LtCAAT1 and LtCAAT2 have
the two conserved motifs HxxxD and DFGWG [37], considered
characteristic of this superfamily (Figure S2). Based on the higher
sequence identity/similarity to PhCFAT, however, only LtCAAT1
was characterized in this study.
LtCAAT1 was heterologously expressed in E. coli, with the result-
ing recombinant His-tagged protein purified by metal chelating
affinity chromatography to apparent homogeneity. The purified
LtCAAT1 was then assayed in vitro using various monolignol sub-
strates, i.e., p-coumaryl (16), caffeyl (17), coniferyl (18), 5-hydroxy-
coniferyl (19), and sinapyl (20) alcohols, respectively. Its pH
optimum was determined to be ꢀ4.8 in citrate buffer system with
all monolignols (16 – 20) tested. While LtCAAT1 is fairly substrate
versatile, its kinetic data (Km, Vmax, kcat & kcat/Km) established that
Creation of chimeric proteins (LtAPS1/LtPPS1) to probe regions
affecting regiospecificity in APS/PPS catalytic conversions
To create chimeric LtAPS1 proteins containing a putative LtPPS1
sub-domain, gene fragments of sub-domain size were swapped be-
tween LtAPS1 and LtPPS1 by three sequential PCR steps [38]. First,
the DNA fragment of LtPPS1 encoding amino acids K71 through
E118 was amplified with forward and reverse primers containing
a 15 bp flanking region of LtAPS1 adjacent to the LtPPS1 sub-
domain using touch-down PCR as described above: primers
(LtAPS-PPS-For, 50-CTCAATGATCACGAGAAGCTTGTTTCAGTGC-30;
LtAPS-PPS-Rev, 50-CATGGGTTCGGTCAACTTCATTCCCAAATTCTG-30)
Second, PCR amplification was carried out with the LtAPS1
forward primer starting with the ATG start codon (LtAPS1-For, 50-
ATGGCACAGAAGAGCAAG-30) and the LtPPS1 reverse primer
starting with the end of the sub-domain for swapping (LtPPS1-
FGNE-Rev, 50-TTCATTCCCAAATTCTGAAGGGAC-30) by touch-down
PCR. Another PCR amplification was performed with the LtPPS1
forward primer starting with the beginning of the sub-domain
for swapping (LtPPS1-KLVS-For, 50-AAGCTTGTTTCAGTGCTTAAA-
CAAG-30) and the LtAPS1 reverse primer beginning with TAA stop
codon (LtAPS1-Rev, 50-TTAAACAAACTGACTAAGGTACTCC-30). These
two distinct rounds of PCR amplifications were done with both
LtAPS1 full length cDNA and the LtPPS1 PCR products having LtAPS1
flanking regions amplified by the first PCR reaction as PCR tem-
plates. Lastly, PCR amplification was carried out with LtAPS1 for-
ward/reverse primers (LtAPS1-For/LtAPS1-Rev) and two different
chimeric DNA fragments that were obtained from the second PCR
amplification as PCR templates. Through this three-step PCR ampli-
fication, a full length chimeric LtAPS1 containing LtPPS1 sub-domain
DNA fragment (now K69 through E116) was obtained (Figure S1).
Using the same PCR approach, the full length chimeric LtPPS1
containing a LtAPS1 sub-domain DNA fragment (now S71 through
D118) was constructed with following primers: LtPPS-APS-For,
50-CTTAGACGAACATGAGAGCTTAGTGAAGGC-30 and LtPPS-APS-
Rev, 50-CACTTACTCTGTCCACGTCGTTTCCAAATTCTG-3’ for first
PCR amplification; LtPPS1-For, 50-ATGGCTTGTGAGAAGAGCAAGA-
TACTC-30, LtAPS1-FGND-Rev, 50-GTCGTTTCCAAATTCTGAAGG-30,
LtAPS1-SLVK-For, 50-AGCTTAGTGAAGGCAATTAAGAAGG-30, and
coniferyl alcohol (18) was the preferred substrate overall, i.e., kcat
/
Km ꢀ241,000 M–1 s–1, with p-coumaryl alcohol (16) being the next
most effective (kcat/Km ꢀ26,300 M–1 s–1). The other potential sub-
strates were more poorly utilized (Table 1). When assayed in pres-
ence of coniferyl alcohol (18), kcat/Km for acetyl CoA was
determined to be 203,000 M–1 s–1
.
L. tridentata propenylphenol synthase 1 (LtPPS1) and allylphenol
synthase 1 (LtAPS1)
A putative LtPPS1 homolog was obtained using a PCR-based
approach with several combinations of degenerate primer sets,
and first-strand cDNA synthesized from total RNA isolated from
L. tridentata leaves. The full-length putative LtPPS1 cDNA sequence
was obtained by 50-/30-RACE and touchdown PCR. It encoded a
ꢀ35.8 kDa peptide, having ꢀ83/68, ꢀ78/61, ꢀ77/56, and ꢀ85/
69% similarity/identity to isoeugenol synthases from Petu-
nia ꢁ hybrida (PhIGS1 [41]), Pimpinella anisum (PaAIS1 [24]), C.
breweri (CbIGS1 [23]) and P. regnellii (PrPPS1, GenBank accession
No. KF543267), respectively (Table 2). As indicated beforehand,
P. regnellii contains both allylphenol derived monomers 7–9 and
presumed propenylphenol-derived lignans, e.g., 10 [17]. Two APS
genes were also isolated (PrAPS1 and PrAPS2, GenBank accession
No. KF543265 and KF543266, respectively) from P. regnellii.
LtPPS1-Rev,
50-TCAAGCAAAAGTTGCCAATTTAGGCTTGGGAGG-30
for second and third rounds of PCR amplification.
The two different chimeric LtAPS1 and LtPPS1 cDNA’s were then
individually cloned into a pEXP5-NT/TOPOÒ expression vector and
transformed into E. coli BL21 (DE3) cells (Invitrogen) after se-
quence verification. Heterologous gene expression and purification
were individually performed as described above. Purified chimeric
LtAPS1 and LtPPS1 proteins were then assayed using coniferyl 9-
acetate (23) as a substrate.
Results and discussion
LtPPS1 was then heterologously expressed in E. coli, with the
resulting His-tagged protein purified to apparent homogeneity by
metal chelating affinity chromatography for in vitro assays. LtPPS1,
when incubated with monolignol acetates [p-coumaryl (21), caf-
feyl (22), coniferyl (23), 5-hydroxyconiferyl (24) and sinapyl (25)
9-acetates], in the presence of NADPH, catalyzed formation of the
corresponding propenylphenols [p-anol (26), 5-hydroxy-p-anol
(27), isoeugenol (28), 5-hydroxyisoeugenol (29) and 5-methoxy-
isoeugenol (30), respectively]. Its pH optimum was ꢀ6.5 in MES–
KOH buffer system with all monolignol acetates (21–25) tested.
Kinetic data established that coniferyl 9-acetate (23) (kcat/Km
L. tridentata cinnamyl alcohol acyltransferases (LtCAAT1 and 2)
Initially, two putative monolignol acyltransferase genes
(LtCAAT1 and 2) were isolated from creosote bush (L. tridentata)
leaves using degenerate primers designed based on peptide
sequences conserved among members of the BAHD (benzyl
alcohol-acetyl-, anthocyanin-O-hydroxy-cinnamoyl-, anthrani-
late-N-hydroxy-cinnamoyl/benzoyl-, deacetyl-vindoline [39,40])
acyltransferase superfamily. These included an acetyl CoA: conife-
ryl alcohol acyltransferase from Petunia ꢁ hybrida (PhCFAT;
GenBank accession No. DQ767969) and two uncharacterized
ꢀ130,000 M–1 s–1
) was the preferred substrate, with caffeyl

42
S.-J. Kim et al. / Archives of Biochemistry and Biophysics 541 (2014) 37–46
Table 1
Kinetic parameters of LtCAAT1 determined with (hydroxy)cinnamyl alcohols, p-coumaryl (16), caffeyl (17), coniferyl (18), 5-hydroxyconiferyl (19) and sinapyl (20) alcohols as
well as acetyl-CoA with coniferyl alcohol (18).
Km
95
82
(
l
M)
Vmax (pkat/
lg)
kcat (s–1
)
kcat/Km (M–1 s–1
)
p-Coumaryl alcohol (16)
Caffeyl alcohol (17)
6
8
48.5 1.2
10.3 0.4
2.5
0.5
26,300
6,100
Coniferyl alcohol (18)
5-Hydroxyconiferyl alcohol (19)
Sinapyl alcohol (20)
174 8.5
476 27
302 33
821.0 17.7
81.7 2.8
24.5 1.3
42.1
4.2
1.3
241,000
8,800
4,300
Acetyl-CoA [with coniferyl alcohol (18)]
133
9
527.0 14.7
27
203,000
Table 2
Taken together, these data are thus in support of a broader sub-
strate versatility for these enzymes than had previously been con-
cluded, with the profiles of products being formed by the different
plants presumably resulting from substrate availability, and not an
inherent inability of the enzymes to bind and/or react with these
substrates. Thus, in our view, these enzymes should be regarded
more generally as allylphenol and propenylphenol synthases
(APSs/PPSs), to reflect their broader substrate preferences, i.e., in
a manner similar to the aforementioned substrate-versatile CADs
and 4CLs operating in the phenylpropanoid pathway [35,36].
Percent similarity and identity of LtPPS1 peptide sequence with other characterized
APS/PPS. Sequences were obtained from the NCBI protein sequence database. The
BLOSUM62 matrix was applied for scoring.
LtPPS1
PrPPS1
PhIGS1
PaAIS1
PrPPS1 (KF543267)*
PhIGS1 (DQ372813)
PaAIS1 (EU925388)
CbIGS1 (EF467238)
85/69
83/68
78/61
77/56
81/64
74/55
73/51
78/58
72/51
74/52
*
GeneBank accession numbers are given in parenthesis.
In vitro co-incubation of LtCAAT1 and either LtAPS1 or LtPPS1
9-acetate (22) (kcat/Km ꢀ66,700 M–1 s–1) being the next (Table 3);
the others were somewhat poorer substrates, particularly 5-
hydroxyconiferyl 9-acetate (24).
In order to establish whether both LtCAAT1 and LtAPS1 or
LtPPS1 could readily function in tandem, in vitro assays were next
individually performed in buffer at different pHs containing
LtCAAT1, acetyl-CoA, NADPH, various monolignols (16–20) and
either LtAPS1 or LtPPS1. In all cases, the monolignols tested were
processed to form their acylated monolignol derivatives (21–25)
and the corresponding propenyl (26–30) or allyl (2, 3, 31–33) phe-
nols. The optimum pH values for the coupled conversion of a
monolignol into its corresponding allylphenol (by LtAPS1) or
propenylphenol (by LtPPS1) were determined to be ꢀ7.0 and 6.5
in MES–KOH buffer system, respectively, with >75% overall activity
being maintained at pH values ranging from 6.25 to 7.5 and 5.5 to
7.0 (data not shown).
LtAPS1, formerly named chavicol/eugenol synthase 1 (LtCES1)
[19], was previously heterologously expressed in E. coli, with the
resulting His-tagged LtAPS1 purified by metal chelating affinity
chromatography to apparent homogeneity. For comparative pur-
poses, LtAPS1 was also shown to catalyze conversion of the corre-
sponding monolignol acetates (21–25) into the corresponding
allylphenols [chavicol (2), 3-hydroxychavicol (31), eugenol (3), 5-
hydroxyeugenol (32) and 5-methoxyeugenol (33), respectively].
Its optimum pH herein was ꢀ7.0 in MES–KOH buffer system with
all monolignol acetates tested as substrates. Kinetic parameters
determined that coniferyl 9-acetate (23) (kcat/Km ꢀ87,000 M–1 s–
1) was the most effective substrate, with p-coumaryl 9-acetate
(21) (kcat/Km ꢀ83,000 M–1 s–1) being the next (Table 4). These more
comprehensive enzymatic kinetic parameters indicate that LtAPS1
is even more efficient than previously described [19].
LtCAAT1/LtAPS1 and LtCAAT1/LtPPS1 dual constructs, heterologous co-
expression in E. coli and in vivo metabolism of monolignols into allyl/
propenylphenols
Thus, both L. tridentata PPS and APS are also demonstrably more
substrate-versatile than the previously reported other homologs
in, for example, basil (ObEGS1), petunia (PhIGS1) and C. breweri
(CbEGS1, CbEGS2), which were only characterized with coniferyl
9-acetate (23) as substrate [21,23].
It was next instructive to establish whether E. coli cultures har-
boring LtCAAT1/LtAPS1 and LtCAAT1/LtPPS1 genes were able to di-
rectly convert monolignols into the corresponding allyl/propenyl
phenols. If so, this potentially could be a means for producing
Table 3
Kinetic parameters of LtPPS1 determined with (hydroxy)cinnamyl acetates, p-coumaryl (21), caffeyl (22), coniferyl (23), 5-hydroxyconiferyl (24) and sinapyl (25) acetates.
Km
(
lM)
Vmax (pkat/
lg)
kcat (s–1
)
kcat/Km (M–1 s–1
)
p-Coumaryl 9-acetate (21)
Caffeyl 9-acetate (22)
Coniferyl 9-acetate (23)
5-Hydroxyconiferyl 9-acetate (24)
Sinapyl 9-acetate (25)
99 10
61.2 2.3
60.1 3.3
103.0 3.3
13.1 0.3
34.7 1.0
2.2
2.2
3.7
0.5
1.2
22,200
66,700
130,000
6,600
33
36
76
44
6
4
5
4
27,300
Table 4
Kinetic parameters of LtAPS1 determined with (hydroxy)cinnamyl acetates, p-coumaryl (21), caffeyl (22), coniferyl (23), 5-hydroxyconiferyl (24) and sinapyl (25) acetates.
Km
(
lM)
Vmax (pkat/
lg)
kcat (s–1
)
kcat/Km (M–1 s–1
)
p-Coumaryl 9-acetate (21)
Caffeyl 9-acetate (22)
Coniferyl 9-acetate (23)
5-Hydroxyconiferyl 9-acetate (24)
Sinapyl 9-acetate (25)
212 33
107 18
100 19
236 50
697 37
517.0 24.3
73.9 3.5
253.9 12.5
144.0 8.4
557.7 10.6
17.6
2.5
8.7
4.9
19.0
83,000
23,400
87,000
20,800
27,300

S.-J. Kim et al. / Archives of Biochemistry and Biophysics 541 (2014) 37–46
43
the latter compounds in a bacterial system, including with addi-
tional modifications as needed using synthetic biology approaches
in the future. Thus, the plasmid containing LtCAAT1, as well as that
harboring either LtAPS1 or LtPPS1, were individually assembled
with different antibiotic resistance selectable markers. These were
then used to transform E. coli cells, with levels of gene expression
promoted by IPTG. Co-expression was confirmed after purification
using metal chelation affinity chromatography followed by SDS–
PAGE analysis (see Fig. 3 with LtCAAT1 and LtPPS1 as an example).
The engineered E. coli lines were thus next examined over their
logarithmic growth period (OD600 = ꢀ0.6) for their abilities to
metabolize coniferyl alcohol (18) in the presence of exogenously
provided cofactors (acetyl-CoA and NADPH) and with/without
IPTG (0, 0.1 and 1.0 M) induction. Bacterial suspensions were ana-
lyzed every 30 min up to 8 h after IPTG induction/substrate addi-
tion and were demonstrated to generate either eugenol (3) when
co-transformed with LtCAAT1 and LtAPS1 (Fig. 4A) or isoeugenol
(28) when co-transformed with LtCAAT1 and LtPPS1 (Fig. 4B). In
both cases, in the absence of IPTG, the overall yields of either euge-
nol (3) – or isoeugenol (28) – were higher than when 0.1 or 1.0 M
IPTG were employed. This can be rationalized by the fact that
monolignol acetates are not very stable at pH >6.0. [For instance,
the recovery of coniferyl 9-acetate (23) after its incubation at 7.0
for 5 min was only ꢀ40%]. Thus, while higher expression levels
were achieved for all three genes upon IPTG induction (data not
shown), in the absence of IPTG, the lower expression level of
LtCAAT resulted in slower formation of coniferyl 9-acetate (23),
which could then be rapidly processed into eugenol (3) or isoeuge-
nol (28).
As an example, Fig. 5 shows UPLC chromatograms of culture
medium at 0 (Fig. 5A, D and G), 1.5 and 3 h during incubation of
coniferyl alcohol (18). Accumulation of coniferyl 9-acetate (23)
and eugenol (3) occurs when E. coli cells were co-transformed with
LtCAAT1/LtAPS1 (Fig. 5E and F), as well as 23 and isoeugenol (28)
with the LtCAAT1/LtPPS1 (Fig. 5H and I). In the latter case, the cor-
responding allylphenol, eugenol (3), was also detected in trace
amounts. As expected, untransformed E. coli cells did not produce
either allyl/propenyl phenols (Fig. 5B and C).
Fig. 4. Accumulation levels of coniferyl acetate (23), eugenol (3) and isoeugenol
(28) in E. coli culture medium at different time points. At t = 0 min, coniferyl alcohol
(18) was added to E. coli cells co-transformed with LtCAAT1/LtAPS1 (A) or LtCAAT1/
LtPPS1 (B) with/without IPTG (0, 0.1 and 1.0 M) induction. Aliquots were analyzed
every 30 min up to 8 h after IPTG induction/substrate addition.
In an analogous manner, p-coumaryl (16) and sinapyl (20) alco-
hols were converted into their corresponding allyl (2 and 33) or
propenyl (26 and 30) phenols when incubated with the co-trans-
formed E. coli cells harboring LtCAAT1/LtAPS1 and Lt CAAT1/LtPPS1,
respectively. In all cases, products were individually identified/
confirmed by their retention times and GC–MS mass spectral data
of the corresponding silylated derivatives, relative to those of
authentic standards.
Interestingly, efficient (comparable) conversions took place
even in the absence of added cofactors (data not shown). This pre-
sumably indicated that these modified E. coli lines can efficiently
take up monolignol substrates in culture medium, then utilize
them with the co-expressed LtCAAT1/LtAPS1 or Lt CAAT1/LtPPS1
and with their own cofactors [e.g., acetyl-CoA and NAD(P)H] inside
the cells. In all assays, however, the final (allyl/propenylphenols) or
intermediate (monolignol acetates) products formed were de-
tected in the culture media only, and not in sonicated cell pellet
samples, i.e., presumably indicative of the final and/or intermedi-
ate products formed inside the cell being secreted.
Phylogenetic comparisons and differing regiospecificities of APS/PPS
homologs
The most recent phylogenetic classification of the APS/PPS gene
family, whose enzymatic properties have been established with
the corresponding recombinant proteins, is shown in Fig. 6. To gen-
erate this classification, a multiple sequence alignment was per-
formed using ClustalW, with the neighbor-joining method used
to construct the unrooted phylogenetic tree. From this treatment,
a clear segregation into three distinct clades occurs for the APS
and PPS families with a ‘‘PPS clade’’ including LtPPS1, PrPPS1,
PaAIS1 and PhIGS1, an ‘‘APS clade’’ formed by LtAPS1, LtAPS2,
Fig. 3. Co-expression of LtCAAT1 and LtPPS1 in E. coli. Lane 1 shows molecular
weight markers, lane 2 shows the crude enzyme extract, lanes 3 to 8 show the His-
tagged fusion proteins (LtCAAT1/LtPPS1) eluted with imidazole (40–140 mM at
20 mM concentration intervals, respectively) in Tris–HCl buffer (pH 7.9).

44
S.-J. Kim et al. / Archives of Biochemistry and Biophysics 541 (2014) 37–46
Fig. 5. Reversed-phase UPLC analyses of culture medium of E. coli cells co-transformed with LtCAAT1/LtAPS1 or LtCAAT1/LtPPS1. Coniferyl alcohol (18) was added to culture
medium with aliquots taken after 0, 1.5 and 3 h incubation. (A–C) Untransformed (control) E. coli cell culture. (D–F) E. coli cell culture co-transformed with LtCAAT1/LtAPS1.
(G–I) E. coli cell culture co-transformed with LtCAAT1/LtPPS1.
LtPPS1, whereas PrAPS1, PrAPS2, LtAPS1, LtAPS2, PhEGS1 and
CbEGS2 (APS clade) have only ꢀ62–66% similarity and ꢀ41–46%
identity to LtPPS1. In addition, although the remaining Clarkia
and Ocimum homologs (CbIGS1, CbEGS2 and ObEGS1) segregate
away from the PPS clade, the levels of similarity and identity are
ꢀ75–77% and ꢀ53–56% to LtPPS1, respectively.
Interestingly, in our hands, the in vitro enzymatic activities of
LtAPS/LtPPS are higher, by more than an order of magnitude, than
those from reports for the corresponding C. breweri CbEGS1 and
CbIGS1 [23]. Yet, the Clarkia enzymes, in this distinct clade, are also
highly homologous to each other (ꢀ96% identity): they differ by
only 13 amino acid residues, nine of which are within a small re-
gion between positions 73 and 95 (Fig. 7). This level of similar-
ity/identity prompted Koeduka et al. [23] to begin initial studies
to probe the molecular basis of differing regiospecificities between
CbEGS1 and CbIGS1. Thus, when they initially fused the first nucle-
otides corresponding to amino acids 1–95 of CbEGS1 with the
remaining sequence of the CbIGS1, this resulted in a partially chan-
ged regiospecificity. That is, in terms of eugenol (3):isoeugenol (28)
ratios obtained, this was 0:100 for the CbIGS1 wild type and 76:24
for the hybrid protein. By contrast, the reciprocal hybrid protein
(amino acids 1–95 of CbIGS1 with amino acids 96–318 of CbEGS1)
lost activity completely.
Next, following generation of various single and multiple site-
directed mutants, it was concluded that there were two amino acid
residues mainly responsible for the differing regiospecificities, i.e.,
F84 and I87 were reported as being mainly required for APS activ-
ity, versus V84 and Y87 for PPS, respectively. For the CbIGS, the
V84F/Y87I mutant mainly generated eugenol (3), the opposite reg-
iospecific product, in a eugenol (3):isoeugenol (28) ratio of 75:25,
relative to the WT enzyme which only formed isoeugenol (28).
The CbEGS1 F84V/I87Y mutant, however only partially changed
its catalytic activity from a eugenol (3):isoeugenol (28) ratio of
100:0 (WT) to 31:69. In contrast, the F86V/I89Y mutant of CbEGS2
Fig. 6. Phylogenetic analysis of peptide sequences in APS/PPS gene family previ-
ously studied from various plants. Peptide sequence alignment was carried out
using ClustalW program, and the relationship was shown as an unrooted neighbor-
joining phylogenetic tree. The multiple alignments and the phylogenetic tree were
constructed by using MEGA5.2 software. LtAPS1/2 are from Larrea tridentata
(KF543262/KF543263, respectively); PrAPS1/2 are from Piper regnellii (KF543265/
KF543266, respectively); CbEGS1/2 are from Clarkia breweri (EF467239/EF467240,
respectively); ObEGS1 is from Ocimum basilicum (DQ372812); PhEGS1 is from
Petunia hybrida (EF467241); LtPPS1 is from L. tridentata (KF543264); PrPPS1 is from
Piper regnellii (KF543267); PaAIS1 is from Pimpinella anisum (EU925388); PhIGS1 is
from Petunia hybrida (DQ372813) and CbIGS1 is from C. breweri (EF467238). Note:
GeneBank accession numbers are given in parenthesis.
PrAPS1, PrAPS2, PhEGS1 and CbEGS2, and a ‘‘PPS/APS clade’’ with
CbEGS1, CbIGS1 and ObEGS.
Table 5 also depicts the levels of similarity/identity between
each homolog, relative to LtPPS1. PhIGS1, PrPPS1, and PaAIS1
(PPS clade) show ꢀ78–83% similarity and ꢀ61–68% identity to

S.-J. Kim et al. / Archives of Biochemistry and Biophysics 541 (2014) 37–46
45
Table 5
Percent similarity and identity of APSs (EGSs) and PPSs (IGSs) peptide sequence. Sequences were obtained from the NCBI protein sequence database. The BLOSUM62 matrix was
applied for scoring.
LtPPS1
PhIGS1
PrPPS1
PaAIS1
CbIGS1
CbEGS1
ObEGS1
PrAPS1
PrAPS2
LtAPS1
LtAPS2
PhEGS1
LtPPS1
PhIGS1
PrPPS1
PaAIS1
CbIGS1
CbEGS1
ObEGS1
PrAPS1
PrAPS2
LtAPS1
LtAPS2
PhEGS1
CbEGS2
83/68
85/69
78/61
77/56
76/56
75/53
65/44
65/46
63/43
62/41
65/46
66/45
81/64
78/58
72/51
71/50
70/51
60/42
63/43
62/42
60/41
62/45
63/44
74/55
73/51
72/51
73/52
65/47
64/44
63/43
60/42
63/47
63/45
74/51
73/51
75/54
60/40
63/44
61/45
60/43
62/45
63/44
98/96
77/57
61/44
62/41
61/42
59/39
61/43
63/44
77/58
6243
66/44
65/42
64/44
62/41
64/45
67/46
62/41
61/42
59/39
61/42
64/44
86/72
82/73
80/68
84/73
84/74
82/70
82/67
82/67
83/67
88/79
86/79
87/79
83/71
85/74
88/82
Fig. 7. Peptide sequence alignments of core region of APS (EGS) and PPS (IGS) previously claimed as core domain which affects their regiospecificity. The red dotted line
represents separation of proteins with distinct APS (EGS) and PPS (IGS) catalytic properties. The red } indicate the 9 amino acids that are different between CbIGS and CbEGS
between amino acids 73 and 95. The red triangles indicate amino acid residues at 84 and 87 (in case of CbEGS1 and CbIGS1). (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)
changed the product ratio of eugenol (3):isoeugenol (28) ratio
100:0 to 88:12.
dictated by the overall steric changes in the binding pocket result-
ing from variations in the amino acid residues over the ‘‘core sub-
strate binding/catalysis domains’’.
Thus, to further explore the regiospecificity of the core domain,
we generated two different chimeric proteins using a three
sequential PCR amplification process [38]. Unlike the Clarkia genes,
the peptide sequences of LtAPS1 and LtPPS1 had only 63% similar-
ity and 43% identity to each other. Therefore, we made chimeric
proteins instead of fusion proteins. Each core domain in LtAPS1/
LtPPS1 that was reported as a key region for regiospecificity was
thus completely swapped (see Materials and Methods). The chime-
ric LtAPS1/LtPPS1 cDNAs were then individually cloned and sepa-
rately expressed in E. coli, with the corresponding chimeric
proteins purified by metal chelating affinity chromatography, and
subjected to enzyme assays with coniferyl 9-acetate (23). As for
the reverse construct from the Clarkia series, the chimeric LtAPS1
containing the LtPPS1 core domain (from K69 to E116) was also
catalytically inactive. By contrast, the chimeric LtPPS1 containing
the LtAPS1 core domain (from S71 to D118) produced eugenol
(3):isoeugenol (28) in a 16:84 ratio compared to isoeugenol (28)
being essentially only produced by the WT enzyme. Thus, in our
hands, manipulation of this core domain did not significantly affect
APS/PPS regiospecificity.
Conclusion
In this study, genes encoding catalytically active CAAT/APS and
CAAT/PPS were isolated from L. tridentata. Kinetic properties estab-
lished that LtAPS1 and LtPPS1 have the highest catalytic activities
yet reported. Using E. coli as a bacterial platform, LtCAAT/LtAPS1
and LtCAAT/LtPPS1 were able to convert monolignols efficaciously
into their corresponding allyl/propenyl phenols. Interestingly,
these genes segregate away from the clade in C. breweri (which
produces minute levels of monomeric allyl/propenylphenol metab-
olites). Phylogenetic analyses indicated that amino acid at posi-
tions 84 and 87 in CbEGSI/CbIGS1 were not conserved, and did
not control regiospecificities in other clades.
Nevertheless, these developments now offer a new means to
produce these important plant natural products via bacterial pro-
duction systems.
In addition, the amino acid residues at equivalent positions to
F84 and I87 in CbEGS1 are not conserved in the catalytically more
active L. tridentata and P. regnellii APS and PPS enzymes (see Fig. 7).
This suggests instead that these two amino acid residues do not
function as a biochemical switch, with control instead being
Acknowledgments
This research was supported by the U.S. Department of Agricul-
ture National Institutes for Food and Agriculture (#683A757612),

46
S.-J. Kim et al. / Archives of Biochemistry and Biophysics 541 (2014) 37–46
[21] T. Koeduka, E. Fridman, D.R. Gang, D.G. Vassão, B.L. Jackson, C.M. Kish, I. Orlova,
S.M. Spassova, N.G. Lewis, J.P. Noel, T.J. Baiga, N. Dudareva, E. Pichersky, Proc.
Natl. Acad. Sci. USA 103 (2006) 10128–10133.
the Boeing Company (#9M-PS01-MEC08-062), and the G. Thomas
and Anita Hargrove Center for Plant Genomic Research.
[22] D.G. Vassão, D.R. Gang, T. Koeduka, B. Jackson, E. Pichersky, L.B. Davin, N.G.
Lewis, Org. Biomol. Chem. 4 (2006) 2733–2744.
[23] T. Koeduka, G.V. Louie, I. Orlova, C.M. Kish, M. Ibdah, C.G. Wilkerson, M.E.
Bowman, T.J. Baiga, J.P. Noel, N. Dudareva, E. Pichersky, Plant J. 54 (2008) 362–
374.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.abb.2013.10.019.
[24] T. Koeduka, T.J. Baiga, J.P. Noel, E. Pichersky, Plant Physiol. 149 (2009) 384–
394.
[25] A.T. Dinkova-Kostova, D.R. Gang, L.B. Davin, D.L. Bedgar, A. Chu, N.G. Lewis, J.
Biol. Chem. 271 (1996) 29473–29482.
[26] M. Fujita, D.R. Gang, L.B. Davin, N.G. Lewis, J. Biol. Chem. 274 (1999) 618–627.
[27] T. Min, H. Kasahara, D.L. Bedgar, B. Youn, P.K. Lawrence, D.R. Gang, S.C. Halls, H.
Park, J.L. Hilsenbeck, L.B. Davin, N.G. Lewis, C.H. Kang, J. Biol. Chem. 278 (2003)
50714–50723.
[28] D.R. Gang, H. Kasahara, Z.-Q. Xia, K. Vander Mijnsbrugge, G. Bauw, W. Boerjan,
M. Van Montagu, L.B. Davin, N.G. Lewis, J. Biol. Chem. 274 (1999) 7516–7527.
[29] N.L. Paiva, R. Edwards, Y.J. Sun, G. Hrazdina, R.A. Dixon, Plant Mol. Biol. 17
(1991) 653–667.
[30] N.L. Paiva, Y. Sun, R.A. Dixon, H.D. VanEtten, G. Hrazdina, Arch. Biochem.
Biophys. 312 (1994) 501–510.
[31] X. Wang, X. He, J. Lin, H. Shao, Z. Chang, R.A. Dixon, J. Mol. Biol. 358 (2006)
1341–1352.
[32] J. Bogs, M.O. Downey, J.S. Harvey, A.R. Ashton, G.J. Tanner, S.P. Robinson, Plant
Physiol. 139 (2005) 652–663.
[33] S. Gagné, S. Lacampagne, O. Claisse, L. Gény, Plant Physiol. Biochem. 47 (2009)
282–290.
[34] S.-J. Kim, D.G. Vassão, S.G.A. Moinuddin, D.L. Bedgar, L.B. Davin, N.G. Lewis,
Joint Meeting of the American Society of Pharmacognosy and the
Phytochemical Society of North America, St. Petersburg Beach FL, July 10–14,
2010.
References
[1] D.G. Vassão, K.-W. Kim, L.B. Davin, N.G. Lewis, in: L. Mander, H.-W. Lui (Eds.),
Comprehensive Natural Products II Chemistry and Biology, vol. 1, Elsevier,
Oxford, UK, 2010, pp. 815–928.
[2] K. Chaieb, H. Hajlaoui, T. Zmantar, A.B. Kahla-Nakbi, M. Rouabhia, K.
Mahdouani, A. Bakhrouf, Phytother. Res. 21 (2007) 501–506.
[3] A. Orav, A. Raal, E. Arak, Nat. Prod. Res., Part A 22 (2008) 227–232.
[4] K. Bauer, D. Garbe, H. Surburg, in: K. Bauer, D. Garbe, H. Surburg (Eds.),
Common Fragrance and Flavor Materials: Preparation and Uses, fourth ed.,
Wiley-VCH, 2001, pp. 167–226.
[5] George Uhe Company, Flavor and fragrance ingredients Market Reports
Newsletter (April 2006) http://www.uhe.com/mktreport-04_06.htm.
[6] N.E. Awad, Bull. Fac. Pharm. (Cairo Univ.) 40 (2002) 233–243.
[7] N.E. Awad, Bull. Fac. Pharm. (Cairo Univ.) 40 (2002) 169–174.
[8] M.Y. Moussa, Bull. Fac. Pharm. (Cairo Univ.) 41 (2003) 139–144.
[9] Y. Asakawa, N. Tokunaga, M. Toyota, T. Takemoto, C. Suire, J. Hattori Bot. Lab.
45 (1979) 395–407.
[10] Y. Asakawa, N. Tokunaga, T. Takemoto, S. Hattori, M. Mizutani, C. Suire, J.
Hattori Bot. Lab. 47 (1980) 153–164.
[35] S.-J. Kim, M.-R. Kim, D.L. Bedgar, S.G.A. Moinuddin, C.L. Cardenas, L.B. Davin, C.-
H. Kang, N.G. Lewis, Proc. Natl. Acad. Sci. USA 101 (2004) 1455–1460.
[36] M.A. Costa, D.L. Bedgar, S.G.A. Moinuddin, K.-W. Kim, C.L. Cardenas, F.C.
Cochrane, J.M. Shockey, G.L. Helms, Y. Amakura, H. Takahashi, J.K. Milhollan,
L.B. Davin, J. Browse, N.G. Lewis, Phytochemistry 66 (2005) 2072–2091.
[37] R. Dexter, A. Qualley, C.M. Kish, C.J. Ma, T. Koeduka, D.A. Nagegowda, N.
Dudareva, E. Pichersky, D. Clark, Plant J. 49 (2007) 265–275.
[38] R. Grandori, K. Struck, K. Giovanielli, J. Carey, Protein Eng. 10 (1997) 1099–
1100.
[39] J.C. D’Auria, Curr. Opin. Plant Biol. 9 (2006) 331–340.
[40] St-Pierre, V. De Luca, in: R. Ibrahim, L. Varin, V. De Luca, J.T. Romeo (Eds.),
Recent Advances in Phytochemistry, vol. 40, 2000, pp. 285–315.
[41] T. Koeduka, I. Orlova, T.J. Baiga, J.P. Noel, N. Dudareva, E. Pichersky, Plant J. 58
(2009) 961–969.
[11] S.R. Gradstein, R. Matsuda, Y. Asakawa, J. Hattori Bot. Lab. 50 (1981) 231–248.
[12] S.R. Gradstein, R. Matsuda, Y. Asakawa, Nova Hedwigia 80 (1985) 63–86.
[13] N. Radulovic´, G. Stojanovic´, R. Palic´, Phytother. Res. 20 (2006) 85–88.
[14] A. Wibe, A.-K. Borg-Karlson, T. Norin, H. Mustaparta, J. Comp. Physiol. A. 180
(1997) 585–595.
[15] R.P. Adams, Biochem. Syst. Ecol. 26 (1998) 637–645.
[16] M. Marotti, R. Piccaglia, J. Essent. Oil Res. 4 (1992) 569–576.
[17] P.J.C. Benevides, P. Sartorelli, M.J. Kato, Phytochemistry 52 (1999) 339–343.
[18] S.G.A. Moinuddin, S. Hishiyama, M.-H. Cho, L.B. Davin, N.G. Lewis, Org. Biomol.
Chem. 1 (2003) 2307–2313.
[19] D.G. Vassão, S.-J. Kim, J.K. Milhollan, D. Eichinger, L.B. Davin, N.G. Lewis, Arch.
Biochem. Biophys. 465 (2007) 209–218.
[20] N.G. Lewis, L.B. Davin, S.-J. Kim, D.G. Vassão, A.M. Patten, D. Eichinger, PCT
Application no. PCT/US2007/69911, filed May 29, 2007.