Euphorbia tirucalli β-Amyrin Synthase: Critical Roles of Steric Sizes at Val483 and Met729 and the CH–π Interaction between Val483 and Trp534 for Catalytic Action

Article abstract of DOI:10.1002/cbic.201700368

The functions of Val483, Trp534, and Met729 in Euphorbia tirucalli β-amyrin synthase were revealed by comparing the enzyme activities of site-directed mutants against that of the wild type. The Gly and Ala variants with a smaller bulk size at position 483 predominantly afforded monocyclic camelliol C, which suggested that the orientation of the (3S)-2,3-oxidosqualene substrate was not appropriately arranged in the reaction cavity as a result of the decreased bulk size, leading to failure of its normal folding into the chair–chair–chair–boat–boat conformation. The Ile variant, with a somewhat larger bulk, afforded β-amyrin as the dominant product. Intriguingly, various variants of Trp534 exhibited significantly decreased enzymatic activities and provided no aberrantly cyclized products, although the aromatic Phe and Tyr residues were incorporated and the steric sizes of the aliphatic residues were altered. Therefore, the Trp534 residue does not stabilize the transient cation through a cation–π interaction. Furthermore, the Trp residue, with the largest steric bulk among all natural amino acids, is essential for high enzymatic activity. Robust CH–π complexation between the Val483 and Trp534 residues is proposed herein. Altering the steric bulk at the Met729 position afforded the pentacyclic skeletons. Thus, Met729 is positioned at the E-ring formation site. More detailed insights into the functions of the Val483, Trp534, and Met729 residues are provided by homology modeling.

Full text of DOI:10.1002/cbic.201700368

Accepted Article
Title: Euphorbia tirucalli β-amyrin synthase: critical roles of the steric
sizes at Val483 and Met729 and the CH-πꢀ interaction between
Val483 and Trp534 for the catalytic action
Authors: Tsutomu Hoshino, Kazuya Nakagawa, Yukari Aiba, Daichi
Itoh, Chika Nakada, and Yukari Masukawa
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemBioChem 10.1002/cbic.201700368
Link to VoR: http://dx.doi.org/10.1002/cbic.201700368
A Journal of
www.chembiochem.org

10.1002/cbic.201700368
ChemBioChem
Euphorbia tirucalli -amyrin synthase: critical roles of the steric sizes at Val483 and
Met729 and the CH- interaction between Val483 and Trp534 for the catalytic action
Tsutomu Hoshino,* Kazuya Nakagawa, Yukari Aiba, Daichi Itoh, Chika Nakada, and
Yukari Masukawa
Graduate School of Science and Technology and Department of Applied Biological
Chemistry, Faculty of Agriculture, Niigata University, Ikarashi 2-8050, Nishi-ku,
Niigata 950-2181, Japan
*Corresponding author.
E-mail address: hoshitsu@agr.niigata-u.ac.jp
Electronic Supporting Information available: amino acid alignment of various triterpene
cyclases, EIMS and NMR spectral data of products 8 and 9, GC profiles of the W534X
mutants, relative enzymatic activities of the site-directed mutants, homology models of
the M729X mutants, and primers for preparation of the site-specific enzyme variants.
1
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
Abstract
The functions of Val483, Trp534, and Met729 in Euphorbia tirucalli -amyrin
synthase were revealed by comparing the enzyme activities of the site-directed mutants
against that of the wild type. The Gly and Ala variants with a smaller bulk size at
position 483 dominantly afforded monocyclic camelliol C, suggesting that the
orientation of the (3S)-2,3-oxidosqualene substrate was not appropriately arranged in
the reaction cavity as a result of the decreased bulk size, leading to failure of its normal
folding into the chair-chair-chair-boat-boat conformation. The Ile variant with a
somewhat larger bulk afforded -amyrin as the dominant product. Intriguingly, the
various variants of Trp534 exhibited significantly decreased enzymatic activities and
provided no aberrantly cyclized products, even though the aromatic Phe and Tyr
residues were incorporated and the steric sizes of the aliphatic residues were altered.
Therefore, the Trp534 residue does not stabilize the transient cation via cation-
interaction. Furthermore, the Trp residue—bearing the largest steric bulk among all the
natural amino acids—is essential for high enzymatic activity. A robust CH-
complexation between the Val483 and Trp534 residues is proposed here. Altering the
steric bulk at the Met729 position afforded the pentacyclic skeletons. Thus, Met729 is
positioned at the E-ring formation site. More detailed insights into the functions of the
Val483, Trp534, and Met729 residues are provided by homology modeling.
Key Words: biosynthesis; triterpene; β-amyrin; terpene cyclization; CH-π interaction
Introduction
Polycyclic triterpenes are biosynthesized by (3S)-2,3-oxidosqualene cyclases
(OSCs).[1] Their structural diversity is remarkable and about 150 different carbocyclic
scaffolds are currently known.[2] The multi-reaction steps are attained by a single
enzyme that generates new C-C bonds and chiral centers in a completely regio- and
stereospecific fashion. [1] -Amyrin, which is widely distributed in nature, is a
pentacyclic triterpene consisting of a 6,6,6,6,6-fused skeleton that is produced by
folding the oxidosqualene (1) substrate into a chair-chair-chair-boat-boat
conformation.[1i, 3] The biosynthetic pathway of -amyrin is shown in Scheme 1. In
contrast, tetracyclic lanosterol and cycloartenol frameworks are constructed by
arranging 1 into a chair-boat-chair-chair conformation.[4] Squalene-hopene and
lanosterol synthases have been investigated in depth,[1] but progress in the study of -
amyrin synthase has lagged in comparison. We have reported functional analyses of the
aromatic active sites in Euphorbia tirucalli -amyrin synthase(EtAS),[5] which are
2
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
strictly conserved in all -amyrin synthases (Fig. S1, Supporting Information).
Investigations of triterpene cyclases have indicated that aromatic amino acids mainly
stabilize the transient cation produced during the polycyclization reaction.[6] Our -
amyrin biosynthetic studies, however, have revealed that the conserved aromatic site
residues have the following three distinctive functions: (1) stabilization of the cations
via cation- interaction, as revealed by functional analyses of Trp257,[5c] Tyr259,[5c] and
Phe728[5a]; (2) maintenance of efficient catalysis by keeping the substrate in the correct
position within the reaction cavity, assisted by an appropriate steric bulk, as
demonstrated by the analyses of Phe413[5c] and Phe474;[5b] and (3) creation of CH-
interaction between the substrate and the aromatic amino side chain, as exemplified by
the analysis of Trp257.[5c,7d,7e] Some important structural units involved in substrate 1
have also been revealed for the normal pathway leading to the production of the -
amyrin scaffold.[7]
<Scheme 1>
In this paper, we describe the functional analyses of Val483, Trp534, and Met729 in
EtAS, which are highly conserved in pentacyclic -amyrin and lupeol synthases (Fig.
S1─S3, Supporting Information). We constructed many site-specific mutants by
altering the steric sizes of aliphatic residues and/or the type of aromatic residues, and
then the relative enzyme activities of the variants against that of the wild-type were
evaluated to identify their functions. As shown in Fig. S1, the Val483 in EtAS
corresponds to Ile or Val in the tetracyclic cycloartenol or lanosterol synthases. The
functions of Ile481 in Arabidopsis thaliana cycloartenol synthase (AtCAS1) and Val454
in Saccharomyces cerevisiae lanosterol synthase (Erg7) have been reported. [8a,b]
However, the function of the conserved Val residue in pentacyclic -amyrin synthases
has not been reported. The EtAS variants V483G and V483A, constructed by us,
produced monocyclic camelliol C in a significantly high yield. Trp534 and Met729 are
not conserved in the tetracyclic terpene synthases. The homology-modeled structure of
EtAS suggests that Trp534 and Val483 form a robust CH-complexation between them
to build an organized protein local structure. The Trp534X mutants, irrespective of the
aliphatic or the aromatic variants, exhibited significantly decreased enzymatic activity,
possibly due to a collapse of the organized protein local structure. The role of Met729
residue in the OSC-mediated polycyclization reaction is unknown. A decrease in the
steric bulk at position 729 gave aberrant cyclization products. Nevertheless, the variant
having almost the same steric size as Met, did not yield the aberrant cyclization
products. Therefore, the appropriate steric volume at the three positions is crucial in
3
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
correctly positioning the substrate in the reaction cavity. Herein, we report the detailed
functional analyses of the three residues. Homologically modeled variant structures at
the three sites provide more detailed insights into their mechanistic functions.
Results and discussion
Identification of triterpene products 8–15
Four site-directed variants were constructed by substituting Val483 with Gly, Ala, Ile,
and Phe. Seven variants were constructed for the Trp534 position by replacing it with
the substituents Ala, Val, Ile, Met, His, Phe, and Tyr. Seven mutants were produced
using Gly, Ala, Val, Leu, Phe, Trp, and Asn in lieu of Met729. Mutagenesis of these
positions in the wild-type pYES2-EtAS/CT, constructed earlier by our group,[9] was
performed using the QuikChange site-directed mutagenesis method. These mutants
were expressed in 100 mL cultures of Saccharomyces cerevisiae GIL77, and the
triterpene products accumulated in the yeast cells were extracted with hexane after
saponification with 15% KOH/MeOH. The hexane extracts were dried under the
reduced pressure and subjected to gas chromatography (GC) analyses (Figs. 1A and
1B). Compounds 10 and 11 were produced by the wild-type EtAS and previously
identified as butyrospermol and tirucalla-7,24-dien-3-ol, respectively.[9]
As shown in Fig. 1A (V483X variants), compounds 8 and 9, which are not
produced by the wild type, were detected from the V483G and the V483A mutants. A
large-scale culture (24 L) was conducted to identify the structures of 8 and 9. The
cultured cells were collected by centrifugation, and the triterpenes were extracted with
hexane after saponification. The extracts were loaded onto a SiO2 column for
chromatography using a mixed solvent of hexane:EtOAc (= 100:15), yielding pure
compound 8. Compound 9 was produced in a very small amount, rendering its isolation
difficult. Repeated 5% AgNO3-SiO2 column chromatography (hexane:EtOAc = 100:15
and 100:7.5) finally gave the 9-enriched fraction. This fraction was acetylated with
Ac2O/Py, and the acetate fraction was subjected to normal-phase high-performance
liquid chromatography (hexane:tetrahydrofuran = 100:0.5), affording 9-acetate in an
almost pure state. The two products were subjected to nuclear magnetic resonance
(NMR) analyses, including two-dimensional NMR (Figs. S2 and S3). In the 1H-NMR
spectrum of 8 (600 MHz, CDCl3; Fig. S2.2), six olefinic methyl (Me) protons and five
olefinic protons were observed, indicating that product 8 possesses the same number of
double bonds as that of substrate 1. However, an alcoholic carbon was found at C 75.0
ppm (d, C-3) in the 13C-NMR spectrum (150 MHz). Thus, product 8 is suggested to be a
4
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
monocyclic compound. In the heteronuclear multiple-bond correlation (HMBC)
spectrum, Me-23 (H 1.04, 3H, s) and Me-24 (H 0.901, 3H, s) had cross-peaks with C-3
and C-5 (C 48.9, d). Furthermore, olefinic Me-25 (H 1.79, 3H, s) had strong HMBC
correlations with C-5 (C 48.9, d) and C-6 (C 137.1, s), and with C-1 (C 118.3, d).
These data strongly support a cyclohexene ring in the molecular structure of product 8.
Detailed analyses of the COSY, HOHAHA, NOESY, HSQC, and HMBC spectra (Figs.
S2.4–S2.8) together with the distortionless enhancement by polarization transfer
(DEPT) spectra definitively revealed compound 8 to be camelliol C. The assignments of
the NMR data are shown in Fig. S2.9. In the 1H-NMR spectrum (400 MHz, CDCl3) of
product 9-acetate (Fig. S3.2), five olefinic Me protons and four olefinic protons were
found, in addition to methylidene protons (H 4.99, s; 4.82, s, H-25, each 1H). The
presence of the methylidene group was further confirmed by observing the signal at C
109.4 ppm (t) in the 13C-NMR (Fig. S3.3) and DEPT 135 spectra (100 MHz, C6D6). In
the HMBC spectrum (Fig. S3.8), H-25 had definitive cross-peaks with C-5 (C 51.7, d),
C-6 (C 147.1, s), and C-1 (C 31.9, t). Me-23 (H 1.05, 3H, s) and Me-24 (H 0.985. 3H,
s) had clear HMBC cross-peaks with C-5, C-3 (C 78.3, d), and C-4 (C 39.4, s). These
analyses indicate that product 9 is monocyclic achilleol A (Fig. S3.9).
Gas chromatograms of the W534X variants are shown in Fig. S4. No aberrant
cyclization products were found. The product patterns were the same as that of the wild
type; namely, -amyrin 2, butyrospermol 10, and tirucalla-7,24-diene-3-ol 11. The
product distribution ratios of the W534X mutants are shown in Table S1. As shown in
Fig. S4, production of 2 by all the enzyme variants, including the aliphatic and aromatic
mutants, was significantly decreased compared with that produced by the wild type;
thus, minor products 10 and 11, which are generated by the wild type, were scarcely
detected in the GC profiles of certain mutants (Fig. S4 and Table S1).
Fig. 1B depicts the GC analyses of triterpene fractions from various M729X
variants. Products 12, 13, and 15 were identified as lupeol, germanicol, and -
taraxasterol, respectively,[5] by comparing their GC/mass spectrometry (MS) spectra
(MS fragment pattern and retention times) with those of our previously identified
authentic samples.[5] The MS fragment patterns for the acetylated products of the
M729G variants are shown in Fig. S5.
Cyclization pathway from substrate 1 into products 8–15
The product structures and the cyclization pathway from 1 are illustrated in Scheme 2.
Substrate 1 was cyclized by the V483G and V483A mutants to yield the monocyclic
cation 3. Deprotonation of H-1 and Me-25 afforded camelliol C 8 and achilleol A 9,
5
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
respectively. Further cyclization reaction via a chair-chair-chair-boat folding
conformation gave the tetracyclic dammarenyl cation 4. By contrast, the cyclization
reactions through a chair-chair-chair-chair conformation provided the 17-epi-
dammarenyl cation 4’. These tetracyclic cations 4 and 4’ underwent backbone
rearrangements (H-17C-20, H-13C-17, Me-30C-13, and Me-18C-14),
followed by proton elimination of H-7 to yield 10 and 11. A ring expansion of 4
afforded the baccharenyl cation 5, followed by formation of the lupanyl cation 6.
The mutations at Met729 gave rise to an aberrant cyclization pathway. A decrease
in the steric bulk (the Gly, Ala, and Val variants) halted the polycyclization cascade at
the lupanyl cation 6 stage; the deprotonation of Me-29 or Me-30 furnished lupeol 12.
The lupanyl cation 6 underwent a ring expansion to yield the final oleanyl cation 7, but
germanicol 13 and -taraxasterol 15 were produced in addition to the normal
cyclization product 2. H-18 of 7 was subjected to proton elimination to give 13. 1,2-
Shift of Me-29 to C-19 gave the taraxasteryl cation 14, followed by abstraction of H-
21 to yield 15.
<Scheme 2>
Functional analyses of the Val483 residue
Table 1 shows the product distribution ratio obtained by mutating the Val483
residue. Decrease in the steric bulk (Gly and Ala variants) dramatically altered product
distribution and the aberrant cyclization products. A significantly large quantity of
monocyclic 8 and a very small amount of monocyclic 9 were produced along with the
correctly cyclized pentacyclic 2 (2:8:9 = 1.7:18.1:1 for the V483G mutant). However,
the Ile variant did not yield the two aberrant products; instead, the amount of 2
produced by this mutant was nearly the same as that produced by the native enzyme
(Table 1). It is noteworthy that 10 and 11 are produced by the wild-type EtAS, as we
reported before,[9] and the production yields by the V483X mutants remained almost
unchanged (Table 1), indicating that the alteration in steric bulk at the Val483 position
had little influence on formation of the tetracyclic cation 4. Fig. 2 compares the relative
enzyme activities of the V483X variants with that of the wild-type EtAS. The total
amounts of the triterpene products, which were estimated by GC analyses, were divided
by the enzyme amounts expressed in vivo, which were determined by western blot
analyses (Figs. S6.1–S6.3). This procedure has been successfully applied to the
estimation of the relative enzymatic activities of the mutants against that of the native
EtAS.[1i,5,9] The relative activities for the total amounts of triterpenes produced increased
6
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
in proportion to the increase in steric bulk (van der Waals volume): 65% for the Gly
mutant, 82% for the Ala mutant, and 89% for the Ile mutant. The relative activities for
the production of -amyrin also increased with larger steric bulk (see the right-pointing
arrow in Fig. 2). In contrast, the production of monocyclic 8 and 9 was inversely
proportional to the increase in the van der Waals volume;[10] no production of 8 and 9 by
the Ile mutant was found, although the steric size of Ile is somewhat larger than that of
Val (the wild type). The decreased steric bulk at position 483 could possibly result in an
inaccurate arrangement of the substrate near the B-ring formation site within the
reaction cavity, leading to polycyclization abortion at the monocyclic stage and the
decreased enzyme activity for the total amounts of triterpenes produced. The aromatic
Phe mutant had no enzymatic activity (Fig. 2), which would have occurred as a result of
the failure to achieve CH- complexation, leading to a collapse of the organized protein
local structure, as discussed below in the section on the homology-modeled structures.
Intriguingly, production of 8 and 9 is also found by mutating the corresponding amino
acids of Erg7 and AtCAS1. In the case of Erg7, site-specific mutagenesis of Val454 into
Gly and Ala gave 9 only, besides lanosterol.[8b] On the other hand, the mutation of
Ile481in AtCAS1 into Gly and Ala afforded both 8 and 9, but the production yield of 9
was significantly higher than that of 8.[8a] Joubert et al. suggested that the steric size of
Val (or Ile) residue may participate in prefolding the C-10─C-11 double bond of 1 to
facilitate the formation of chair-shaped B-ring.[8b] In contrast, an overwhelming
production of 8, compared with that of 9, was observed for the EtAS V483X variants.
At the present time, we have no explanation for this difference of product distribution
among the different triterpene cyclases. The Matsuda group discovered camelliol C
synthase (AtCAMS1) from A. thaliana, which produces 8 in an overwhelming yield
(98%), besides 9 (2%) and 2 (0.2%).[11] Amino acid alignment study indicated that the
corresponding position is substituted with Ala, [11]despite Val and Ile being conserved in
tetracyclic and pentacyclic triterpene cyclases (see Figs. S1.1 and S1.3). The Matsuda
group also reported that AtCAMS1 is a descendant of the enzymes that form pentacyclic
triterpenoids. [11] Mutations of this site in EtAS corroborate the proposal of the Matsuda
group that camelliol C synthase evolved from -amyrin synthases. Furthermore, the
triad consisting of the Val483–Trp534–Tyr259 motif (EtAS), of which Trp and Tyr
residues are conserved in AtCAMS1, as shown in Fig. S1.3. This fact may provide
additional evidence that camelliol C synthase evolved from 6,6,6,6,6- or 6,6,6,6,5-fused
pentacyclic triterpene synthase.
<Table 1 and Fig. 2>
7
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
Functional analyses of the Trp534 residue
Figs. 3 and S7.2 show the relative enzymatic activities of the aliphatic and aromatic
variants targeted for residue Trp534. All the aliphatic variants showed significantly
decreased activities (0.8–5.0%). Enhancement of the activities in the Phe and Tyr
variants was slight or negligible, and no significant difference between the aromatic and
aliphatic variants (e.g., the Met and Ile mutants) was observed (see also Fig. S7.2). By
contrast, the EtAS variants of Trp612 into Phe or Tyr exhibited the relative activities of
45─65% against that of the wild-type, but little activities for the aliphatic mutants, in
addition to the generation of aberrant cyclization products (unpublished results by us),
which are significantly different from the experimental results of Trp534 variants (Figs.
3 and S7.2). The homology modeling shows that the Trp612 residue is horizontal to the
lanosterol ligand.[1i] Thus, the -electrons of Trp612 can stabilize the cationic
intermediate via cation- interaction. This distinct behavior between Trp534 and Trp612
mutants supports that Trp534 residue is perpendicular to the lanosterol ligand as shown
in the homology modeling (depicted below). As described in the section of the
homology modeling, the Val483 residue is supposed to participate in the CH-
interaction with the Trp534 that is arranged perpendicularly to the ligand. It is likely that
this complexation build the rigidly organized protein architecture around the A/B/C-ring
formation sites in the enzyme cavity. The CH- interaction between the Val and the Trp
is most robust compared with that between Val and Phe or Val and Tyr, due to the higher
-electron density of Trp relative to that of Phe and Tyr. Thus, a robustly organized
protein architecture would be created. No aberrant cyclization products, such as 8, 9, 12,
13, or 15, were generated by all the mutants examined (Fig. S4 and Table S1). The
significant disruption of the robust protein architecture around the B/C-ring formation
sites, which results from the mutations at this position, would have led to the
significantly decreased enzyme activity and absence of aberrant cyclization product
generation. In addition to the robust CH- complexation of Trp534 with Val483, the
Trp534 residue possessing the largest bulk size among natural amino acids may form a
tightly packed cavity to enable close contact with substrate 1 around A/B/C-ring
formation sites, by which the desired folding conformation is acquired to facilitate the
polycyclization reaction. The steric size at this position of the Phe and the Tyr variants
are smaller than that of the wild-type (Trp residue), which could prevent the folding of
the desired conformation and yield the significantly decreased activities of the Phe and
Tyr variants, that is, the steric size at position 534 also could be crucial for the normal
catalytic action. Thus, the integral structural unit, formed through the CH-
8
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
complexation between the Val and the Trp, builds a tight packing cavity to have close
contact with substrate 1 around A/B/C-ring formation sites, as discussed in the section
of the homology model.
<Fig. 3>
Functional analyses of the Met729 residue
As illustrated in Scheme 2, alteration of the steric bulk at this position renders
germanicol 13 and -taraxasterol 15, which are generated from the oleanyl cation 7. In
addition, lupeol 12 was produced from the lupanyl cation 6. In other words, this Met
residue is located at the E-ring formation site, and its appropriate steric bulk enables the
accurate placement of the substrate in the reaction cavity, leading to its correct folding
(a boat conformation) around the site. Indeed, the production of -amyrin was roughly
increased in proportion to the increase in van der Waals volume (see the right-pointing
arrow in Fig. 4). Furthermore, the increments of the total amounts of triterpenes
produced were in proportion to the steric size, as shown in Fig. 4; namely, 69.9% for the
Gly mutant and 101.1% for the Leu variant possessing nearly the same steric bulk as the
native EtAS. However, the activity of the Phe mutant decreased to 80.2% due to its size
being larger than that of the wild type. The Trp mutant, bearing too large a side chain,
exhibited no enzyme activity. In our experiments, Phe728, positioned prior to Met729,
stabilized the oleanyl cation 7.[5a] Decreasing of the steric bulk at this position afforded
lupane 12 (5-membered E-ring), germane 13 and ursane 15 scaffolds (different
deprotonation products), suggesting that the steric size of Met may have participated in
impelling a correct boat conformation around the E-ring formation site to yield the 6-
membered E-ring (oleanyl cation 7) from the 5-membered E-ring (premature lupanyl
cation 6). The distorted boat structure given by the decreased steric bulk would have
generated the different deprotonation products according to Scheme 2. The expressed
quantities of the mutated EtASs and of each triterpene produced are shown in Figs.
S8.1–S8.7. In the tetracyclic cycloartenol and lanosterol synthases, the Asn residue is
highly conserved at this position (see Figs. S1.1 and S1.2). The Asn mutant produced
13 in a significantly high distribution ratio (54%), but the production of tetracyclic
products 10 and 11 did not increase. Thus, the substitution of Met with Asn does not
contribute at all to the production of the tetracyclic triterpenes.
<Table 2 and Fig. 4>
9
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
Mechanistic insights from the homology-modeled structures
No X-ray structure of -amyrin synthase has been reported hitherto. To gain a
deeper insight into the functions of the Val483, Trp534, and Met729 residues, a
homology models, based on the X-ray structure of human lanosterol synthase (HsLAS),
[13] were constructed, in which the three aforementioned residues were highlighted in
Fig. 5A. The Me groups of Val483 are perpendicular to the indole ring, so a strong CH-
 interaction may occur between V483 and Trp534, and organized local protein
structures are assembled around the A/B/C ring formation sites. Moreover, Val483 is
situated near the lanosterol ligand (ca. 3─4 Å), suggesting that this residue plays a vital
role in the substrate 1 polycyclization reaction. A red arrow indicates the position of
Trp534 in the cavity. The indole ring of Trp is perpendicular to the ligands. Brandl et al.
reported that CH- interaction is observed in various protein structures when the
distance between the carbon atom and the mass center of the aromatic system (dC-X) is
ca 3.4─4.4 Å. [11a] As shown in Fig. 5A, the dC-X are 3.9─4.3 Å for the wild-type EtAS.
Furthermore, the shortest contact distance (3.5─3.9 Å) is also reported as examples of
CH- interaction in proteins.[11b, c] In the case of EtAS, 3.6 Å is estimated as the shortest
distance between the Me group and the indole ring (a green solid line in Fig. 5A).
Therefore, CH- complex between V483 and Trp534 residues can be credible. Met729
is also located near the ligand around the E-ring formation site (ca. 3─4 Å). Fig. 5B
illustrates the arrangements of the three active site residues on substrate 1, which is
folded in a normal chair-chair-chair-boat-boat conformation. Figs. 6 and 7 show the
structures of V483X and W534X, respectively. Models of the M729X variants are
shown in Fig. S9. The orientations of the mutated residues are optimized in the models
by minimizing the steric hindrance between the ambient- and mutated residues.
As depicted in Fig. 6, the CH- interaction in the Gly mutant (A) is minimal,
because the Me group is missing, and a larger space (marked with a blue oval) is
generated between the substrate and the Gly residue. This vacant space and the
distortion of the organized local structure, which could result from the mobility of the
Trp534, could lead to less contact with substrate at B-ring formation site and halt the
polycyclization reaction at premature monocyclic stage (production of monocycles 8
and 9). For the Ala mutant (C), the Ala residue has the Me group and thus CH-
complexation is possible, but this interaction is weaker than that created by the Val
residue (wild-type) because only one Me group is involved. Furthermore, the smaller
steric size could afford the free space around the A/B-ring sites, and thus the premature
cyclization products 8 and 9 were also produced in a similar way as by the Gly mutant.
For the Ile mutant (D), the free space is minimal between the Ile residue and the
10
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
substrate, and the local protein architecture is maintained because of the strong CH-
interaction between the two Me groups of Ile and the indole ring. In the case of the Phe
variants (E and F), the embedded Phe residue with its large steric bulk pushes the
Trp534 residue (labeled as W534 in Fig. 6) aside (due to the collision between Phe and
Trp534) and CH- complexation therefore fails, leading to disorganization of the local
structure. The models give additional evidence for the crucial role of the steric size in
the normal polycyclization cascade, which is consistent with the outcomes illustrated in
Fig. 2.
As depicted in Fig. 7, the Trp534 residue is arranged in proximity to the B/C-ring
formation sites and perpendicularly to the ligand (Fig. 5). Thus, the function of the
Trp534 residue is not to offer the -electrons to the intermediary cation produced during
the polycyclization reaction (cation- interaction), but to place the substrate at the
correct position in the reaction cavity (by virtue of the residue’s appropriate steric size)
in order to furnish a chair-chair-chair conformation for the fused A/B/C-ring system.
The aliphatic mutants replaced by Ala, Ile, and Met are not capable of forming the CH-
complexation, because -electrons are lacking at position 534, and thus the organized
local structure at the A/B/C-ring sites collapses. Furthermore, large empty spaces
(depicted as blue-filled ovals in Fig. 7) are generated, as shown in the models of the
aliphatic mutants (B, C, and D). This indicates that the substrate is moving freely within
the reaction cavity, resulting in failure to organize its normal folding conformation. The
models for the Phe and Tyr mutants (aromatic variants) are depicted as (E) and (F),
respectively, in Fig. 7. Because these aromatic substituents are not arranged
perpendicularly to the ligand (as shown by the green arrows), they also fail to bind the
substrate tightly in the cavity, leading to a significantly decreased enzyme activity
(<10% of the wild-type activity; see Fig. 3).
The models of the mutated EtASs targeted for the Met729 position are shown in
Fig. S9. The Gly and Ala mutants have a significantly large free space (blue-filled
ovals) between the residues and the E-ring site. The Val mutant has a somewhat
compact space. The Leu variant has a similar space as the wild type (tightly packed),
because the side chains of Leu and Met possess a similar van der Waals volume (see
Fig. 4). [10] The Phe and Asn residues are oriented relatively far from the E-ring
formation site (see the green arrows in Fig. S9; the long axes of the side chains differ
from that of Met). As a result, these mutations have decreased enzyme activities and
produce aberrant cyclization products (12, 13, and 15). The Asn mutant has a larger
space than the Phe variant and may generate greater amounts of the aberrantly
deprotonated product 13. The homology-modeled structures further support the theory
11
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
that the steric bulk at site 729 is essential in the normal polycyclization pathway, as
discussed above (see Fig. 4 ).
The Val483 and Trp534 residues are exclusively conserved in the -amyrin and
lupeol synthases, which produce the pentacyclic skeletons (Figs. S1.1 and S1.3). The
Val or Ile residue is also highly conserved in the cycloartenol and lanosterol synthases at
the position equivalent to Val483 in EtAS (Figs. S1.1 and S1.3). Trp534 in EtAS
corresponds to Tyr in the cycloartenol and lanosterol synthases. The Wu group has
reported the importance of His234 and Tyr510 in Erg7, and further pointed out that a
hydrogen-bonded complex is formed by these two residues.[1f, 13] We propose that this
Tyr residue forms the CH- complex with the Val or Ile residues, in addition to the
hydrogen-bond complex with His (Fig. S10, ESI); these robust interactions build an
organized local structure through the A–D-ring formation sites. It was proposed that the
function of the hydroxyl group of Tyr is to abstract H-9 of the final protosteryl cation
for lanosterol formation.[14] However, because the Tyr Phe mutation still afforded
lanosterol, despite the loss of the deprotonating OH group,[15] the function of the Tyr
remains unknown. Fig. S10 shows the structure of human lanosterol synthase (HsLAS)
superimposed onto that of EtAS. The three residues in HsLAS are in close proximity to
each other (ca. 3-4 Å) but the side chains of Tyr503 and Trp534 are oriented very
differently (green and orange arrows in Fig. S10). The disposition of Tyr503 is
essentially the same as that indicated in Fig. 7F. The CH- interaction between Val and
Tyr is weaker than that between Val and Trp because the former has a lower -electron
density than the latter. Nevertheless, the local structure of the tetracyclic terpene
cyclases is reinforced by hydrogen bonding with His232 (shown in green in Fig. S10).
Thus, a more robust CH- interaction occurs in EtAS (Fig. S10, ESI) than in HsLAS.
Functional analyses of Val483 and Trp534 in EtAS indicate that the CH- interaction
between these residues accounts for the robustly organized reaction cavity around the
A/B/C-ring formation sites. This configuration is necessary for EtAS-mediated
cyclization. By contrast, the Val (Ile)–Tyr–His triad complex forms a robust protein
architecture around the A–D-ring formation sites in lanosterol and cycloartenol
biosynthesis. Fig. S1.3 show the multiple alignment of amino acid sequences for various
OSCs. Protostadienol synthase also folds substrate 1 into a chair-boat-chair-chair
conformation. In contrast, the dammarenediol-, baccharis oxide-, baruol-, and thalianol
synthases force 1 into a chair-chair-chair folding conformation. As discussed in Fig.
S11, we propose that the Val453–Tyr503–His-232 triad complex (HsLAS numbering)
may direct 1 into a chair-boat-chair-chair conformation, notably to shape a boat
12
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
structure around the B-ring formation site. By contrast, the Val483–Trp534–Tyr259 triad
(EtAS numbering) would probably impel 1 into the chair-chair-chair folding
conformation, notably to shape a chair structure around the B-ring formation site.
Mutations of Val and Ile into Gly and Ala in lanosterol and cycloartenol synthases also
afford monocycles 8 and 9.[8a, b] This would have occurred by the disruption of the CH-
interaction between the Val (Ile) and the Tyr residues (in a similar way as the EtAS
homology modelling shown in Figs. 6B and 6C and Fig. S10). As described above, the
different orientations and the steric volumes of the side chains of Tyr503 and Trp534
(see Figs. 5 and S10) may confer the different folding conformation (either pre-boat or
pre-chair structure). However, no experimental evidence for this idea is provided at the
present time. To validate this theory and to clarify the molecular mechanism for the
boat- or chair- conformational selectivity, further studies are required.
Conclusions
The present study highlights the importance of the steric bulk of the aliphatic amino
acids at positions 483 and 729 for normal folding of the substrate to complete the
polycyclization cascade, and for the deprotonation position in order to lead to the -
amyrin scaffold as a normal end-product. We propose that Val483 and Trp534 form a
robust CH- complex between the active site residues. This is the first study to report
that the CH- interaction has a crucial role in reinforcing the protein architecture of
OSCs in order to ensure the correct placement of substrate 1 within the active site in the
reaction cavity. Furthermore, it was noted that all the site-specific variants of position
Trp534 had little or no enzymatic activity and afforded no aberrant cyclization products,
which differed significantly from the behaviors of the V483X and M729X variants. This
finding reinforces the fact that the Trp534 residue has a crucial role in building the
rigidly organized protein architecture around the A/B/C-ring formation sites, which
occurs through its CH- complexation with the Val483 residue. The insights derived
from the homology-modeled structures further are consistent with the functions inferred
from the various mutants of Val483, Trp534, and Met729.
Acknowledgments
This work was supported by JSPS KAKENHI (Grant Nos. 25450150 and 18380001).
13
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
Experimental Section
Instruments
NMR spectra were recorded in C6D6 with a Bruker DMX 600 or DPX400
spectrometer. The chemical shifts () are given in ppm relative to the residual solvent
peak as the internal reference (H=7.28 and C=128.0 ppm for C6D6, and H=7.26 and
C=77.0 ppm for CDCl3). GC analyses were done on a Shimadzu GC-2014
chromatograph fitted with a flame ionization detector (J&W DB-1 capillary column,
0.25 mm x 30 m). GC/MS spectra were obtained with a JMS-Q1000 GC K9 (JEOL)
instrument under electronic impact at 70 eV with a Zebron ZB-5ms capillary column
(0.25 mm x 30 m), the oven temperature being elevated from 220 to 270 ºC (3 ºC/min).
Mutagenesis experiments
Mutagenesis of V483, W534 and Met729 in wild-type pYES2–EtAS/CT was
performed with the QuikChange site-directed mutagenesis method, according to the
previous report.9 The oligonucleotide primers used were shown in Table S2. The
integrity of the mutated gene was verified by sequencing.
Expression of the mutated EtAS in S. cerevisiae GIL77, and analyses of triterpene
amounts produced.
The expression plasmid pYES2-EtAS was transformed into S. cerevisiae GIL77
strain lacking lanosterol synthase (erg7, ura3-167, hem3-6, gal2), by using Frozen-EZ
Transformation II Kit (Zymo Research). The transformants were plated onto synthetic
complete medium without uracil (SC-U), supplemented with ergosterol (20 g/mL),
hemin chloride (13 g/mL), and Tween 80 (5 mg/mL), and then incubated at 30ºC for 3
days. Expression of EtAS in the yeast was conducted as reported by Kushiro et al.[16]
100 mL culture of S. cerevisiae GIL77 harboring pYES2-EtAS was grown in SC-U
medium that was supplemented with ergosterol (20 g/mL), hemin chloride (13
g/mL), and Tween 80 (5 mg/mL) at 30ºC with shaking (150 rpm). After grown for 2
A
days, cells were collected and resuspended in SC-U medium without glucose that was
supplemented with ergosterol, hemin chloride, and Tween 80, and then 2% galactose
was added to induce EtAS expression at 30 ºC for 24 h. The yeast cells were collected
and resuspended in 0.1 M potassium phosphate (pH 7.0) supplemented with 3% glucose
and hemin chloride, and further incubated at 30 ºC for 24 h. To the grown cells
collected, was added 15% KOH/MeOH and the suspension was refluxed by heating to
disrupt the cells. The lipophilic materials including the cyclization products were
extracted with hexane, the extract having being subjected to GC/MS and GC analysis
for identifying the triterpenes produced and for quantifying the triterpene produced,
14
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
respectively. As the internal reference, geranylgeraniol was used for the quantification
of the triterpene products.
Evaluation of the enzyme activities for the wild- and mutated EtASs.
The quantitative assays of the expressed EtASs were performed by Western blot
analyses according to the previous paper.[9] To minimize the experimental errors,
glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used. GAPDH is widely
known to be housekeeping protein and frequently used as the internal reference.[17]
Spectroscopic data of products 8 and 9.
Product 8: Oil; 1H-NMR (600 MHz, CDCl3): =0.900 (s, 3H, Me-24), 1.04 (s, 3H, Me-
23), 1.42 (m, 1H, H-7), 1.67 (s, 9H, Me-27, Me-28 and Me-29), 1.71 (s, 3H, Me-26),
1.75 (s, 3H, Me-30), 1.79 (4H: s, 3H for Me-25; m, 1H for H-7), 2.06 (m, 6H: 1H for H-
2, 4H for H-15 and H-19, and 1H for H-8), 2.10 (m, 4H, H-11 and H-12), 2.14 (m, 4H
for H-16 and H-20), 2.24 (m, 1H, H-8), 2.31 (m, 1H, H-2), 3.53 (t-like, J=7.0 Hz, H-3),
5.24─5.15 (4H, m, H-10, H-13, H-14 and H-21), 5.31 (br s, 1H, H-1); 13C-NMR (150
MHz, CDCl3): =15.98 (q), 16.04 (q) and 16.09 (q, 2C) that are assignable for Me-24,
Me-26, Me-27 and Me-28, 17.7 (q, Me-29), 22.6 (q, Me-25), 25.3 (q, Me-23), 25.7 (q,
Me-30), 26.7 (t) and 26.8 (t) for C-16 and C-20, 27.2 (t, C-7), 28.2 (t, 2C) for C-11 and
C-12, 31.8 (t, C-2), 38.1 (s, C-4), 39.7 (t, 2C, C-15 and C-19), 42.0 (t, C-8), 48.9 (d, C-
5), 75.0 (d, C-3); 118.3 (d, C-1), 122.42 (d), 124.24 (d), 124.4 (d) and 124.7 (d) that are
assignable for one carbon among C-10, C-13, C-17 and C-21; 131.2 (s, C-22); 134.9 (s),
135.2 (s), 135.4 (s) that are assignable for one carbon among C-9, C-14 and C-18; 137.1
(s, C-6); MS (EI): m/z (%): 69 (76), 81 (100), 95 (36), 109 (35), 121 (55), 136 (34), 149
(17), 203 (13), 259 (1.3), 274 (2.4), 339 (0.8), 357 (1.2), 408 (1.2), 426 (M+, 1.4).
Product 9 acetate: Oil; 1H-NMR (400 MHz, C6D6): =0.985 (s, 3H; Me-24), 1.05 (s,
Me-23), 1.69 (s, 3H, Me-29), 1.73 (s, Me-27 or Me-28), 1.75 (s, 3H, Me-27 or Me-28),
1.69 (m, 1H, H-2), 1.81 (bs, 3H, Me-30), 1.81 (3H, 1H for H-5 and 2H for H-7, buried
in the large Me-30 signal), 1.82 (s, 3H, COMe), 1.96 (m, 1H, H-2), 2.00 (m, 1H, H-1),
2.02 (m, 1H, H-8), 2.23 (m, 4H, H-15 and H-19), 2.28 (m, 4H, two CH2 protons among
H-11, H-12, H-16 or H-20), 2.28 (m, 4H, two CH2 protons among H-11, H-12, H-16 or
H-20), 2.31 (m, 1H, H-8), 2.33 (m, 4H, two CH2 protons among H-11, H-12, H-16 and
H-20), 4.82 (s, 1H, H-25), 4.96 (m, 1H, H-3), 4.97 (s, 1H, H-25), 5.37 (1H, t, J=6.8 Hz,
H-21), 5.42 (m, 3H, H-10, H-13 and H-17); 13C-NMR (100 MHz, C6D6): =16.1 (q, one
carbon among Me-26, Me-27 and Me-28), 16.2 (2C, q, two carbons among Me-26, Me-
15
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
27 and Me-28), 17.3 (q, Me-24), 20.7 (COCH3), 24.4 (t, C-7), 25.8 (Me-30), 26.2 (q,
Me-23), 27.1(1C, t, either C-11, C-12, C-16 or C-20), 27.3 (1C, t, one carbon among C-
11, C-12, C-16 and C-20), 28.7 (1C, t, one carbon among C-11, C-12, C-16 and C-20),
28.8 (1C, t, one carbon among C-11, C-12, C-16 and C-20), 29.1 (t, C-2), 31.9 (t, C-1),
38.9 (t, C-8), 39.4 (s, C-4), 40.2 (2C, C-15 and C-19), 51.7 (d, C-5), 78.3 (d, C-3), 109.4
(t, C-25), 124.8 (d, 2C), 124.9 (2C, d), 131.1 (s, C-22), 134.9 (s), 135.3 (s), 135.5 (s),
147.2 (s, C-6), 169.6 (COCH3). MS (EI): m/z (%): 69 (100), 81 (82), 95 (28), 107 (23),
121 (21), 136 (18), 175 (20), 203 (14), 257 (1.1), 271 (2.1), 339 (2.3), 365 (1.1), 408
(2.2), 468 (M+, 1.0).
16
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
References
[1] Reviews for the polycyclization of triterpene cyclases. (a) K. U. Wendt, G. E.
Schulz, E. J. Corey and D. R. Liu, Angew. Chem., Int. Ed. 2000, 39, 2812–2833. (b)
T. Hoshino and T. Sato, Chem. Commun. 2002, 291–301. (c) R. A. Yoder and J. N.
Johnston, Chem. Rev. 2005, 105, 4730–4756. (d) M. J. R. Segura, B. E. Jackson and
S. P. T. Matsuda, Nat. Prod. Rep., 2003, 20, 304-317. (e) I. Abe, Nat. Prod. Rep.
2007, 24, 1311–1331. (f) T. K. Wu, C. H. Chang, Y. T. Liu and T. T. Wang, Chem.
Rec. 2008, 8, 302–325. (g) W. D. Nes, Chem. Rev. 2011, 111, 6423–6451. h) G.
Siedenburg, D. Jendrossek, Appl. Environ. Microbiol., 2011, 77, 3905-3915. i) T.
Hoshino, Org. Biomol. Chem., 2017, 15, 2869-2891.
[2] R. Xu, G. C. Fazio and S. P. T. Matsuda, Phytochemistry 2004, 65, 261–291.
[3] (a) A. Eschenmoser, L. Ruzicka, O. Jeger and D. Arigoni, Helv. Chim. Acta. 1955,
38, 1890-1940. (b) T. Suga and T. Shishibori, Phytochemistry, 1975, 14, 2411-2417.
[4] (a) A. Eschenmoser and D. Arigoni, Helv. Chim. Acta, 2005, 88, 3011-3050. (b) E. J.
Corey and S. C. Virgil, J. Am. Chem. Soc., 1991, 113, 4025-4026. (c) T. Hoshino, A.
Chiba and N. Abe, Chem. Eur. J., 2012, 18, 13108-13116.
[5] (a) R. Ito, I. Hashimoto, Y. Masukawa, Chem.-Eur. J., 2013, 19, 17150-17158. (b) R.
Ito, Y. Masukawa, C. Nakada, K. Amari, C. Nakano and T. Hoshino, Org. Biomol.
Chem., 2014, 12, 3836-3846. (c) R. Ito, C. Nakada and T. Hoshino, Org. Biomol.
Chem. 2017, 15, 177-188.
[6] M. Morikubo, Y. Fukuda, K. Ohtake, N. Shinya, D. Kiga, K. Sakamoto, M.
Asanuma, H. Hirota, S. Yokoyama and T. Hoshino, J. Am. Chem. Soc., 2006, 128,
13148-13194, and references cited therein.
[7] (a) I. Abe, Y. Sakano, H. Tanaka, W. Lou, H. Noguchi, M. Shibuya and Y. Ebizuka,
J. Am. Chem. Soc., 2004, 126, 3426-3427. (b) I. Abe, Y. Sakano, M. Sodeyama, H.
Tanaka, H. Noguchi, M. Shibuya and Y. Ebizuka, J. Am. Chem. Soc., 2004, 126,
6880-6881. (c) T. Hoshino, Y. Yamaguchi, K. Takahashi and R. Ito, Org. Lett., 2014,
16, 3548-3557. (d) T. Hoshino, Y. Miyahara, M. Hanaoka, K. Takahashi and I.
Kaneko, Chem. Eur. J., 2015, 21, 15769-15784. (e) I. Kaneko and T. Hoshino, J.
Org. Chem. 2016, 81, 6657-6671. (f) Y. Terasawa, Y. Sasaki, Y. Yamaguchi, K.
Takahashi and T. Hoshino, Eur. J. Org. Chem., 2017, 287-295.
[8] (a) S.P.T. Matsuda, L. B. Darr, E. A. Hart, J.B.R. Herrera, K. E. MacCann, M. M.
Meyer, J. Pang and H. G. Schepmann, Org. Lett., 2000, 2, 2261-2263. (b) B. M.
Joubert, L. Hua and S. P. T. Matsuda, Org. Lett., 2000, 2, 339-341.
17
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
[9] R. Ito, Y. Masukawa and T. Hoshino, FEBS J., 2013, 280, 1267-1280.
[10] Z-h Lin, H-x Long, Z. Bo, Y-q Wang, and Y-z Wu, Peptides, 2008, 29, 1798-1805.
[11] (a) M. Brandl, M. S. Weiss, A. Jabs, J. Sühnel and R. Hilgenfeld, J. Mol. Biol. 2001
307, 357-377. (b) M. Nishio, Y. Umezawa, J. Fantini, M. S. Weiss and P.
Chakrabarti, Phys. Chem. chem. Phys., 2014, 16, 124648-12683. (c) D. Pal and P.
Chakrabarti, J. Mol. Biol., 1999, 294, 271-288.
[12] M. D. Kolesnikova, W. K. Wilson, D. A. Lynch, A. C. Obermeyer and S. P. T.
Matsuda, Org. Lett., 2007, 9, 5223-5226.
[13] (a) T. K. Wu, Y. T. Liu and C. H. Chang, ChemBioChem, 2005, 6, 1177-1181. (b) T.
K. Wu, Y. T. Liu, C. H. Chang, M. T. Yu and H. J. Wang, J. Am. Chem. Soc., 2006,
128, 6414-6419.
[14] R. Tohma, T. Schulz-Gasch, B. D’Archy, J. Benz, J. Arbi, H. Dehmlow, M.
Henning, M. Stihle and A. Ruf, Nature, 2004, 432,118-122.
[15] S. Lodeiro, W. K. Wilson and S. P. T. Matsuda, Org. Lett., 2006, 8, 439-442.
[16] T. Kushiro, M. Shibuya and Y. Ebiduka, Eur. J. Biochem, 1998, 256, 238-244.
[17] (a)A. VigelsøR. Dybboe, C. N. Hansen, F. Dela, J. W. Helge and A. G. Grau, J.
Appl. Physiol, 2015, 118, 386-394. (b) M. Salmon, R. B. Thimmappa, R. E. Minto,
R. E. Melton, R. K. Highes, P. E. O’Maille, A. M. Hemmings and A. Osbourn, Proc.
Natl. Acad. Sci. USA, 2016, 113, E4407-4414.
18
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
Legends to Figures and Schemes
Scheme 1. Cyclization pathway of (3S)-2,3-oxidosqualene 1 to generate -amyrin
triterpene 2. Substrate 1 is folded into a chair-chair-chair-boat-boat conformation in the
enzyme cavity, and the proton released from the DCTA motif attacks the epoxide ring,
leading to the sequential ring-forming reactions and construction of the 6,6,6,6,6-fused
pentacyclic ring scaffold 7 via carbocationic intermediates 3–6. Oleanyl cation 7
undergoes the rearrangement reactions of 1,2-hydride shifts and deprotonation of the
axial-oriented H-12 to yield -amyrin 2.
19
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
(A)
(B)
2
2
10
13
11
Wild
Wild
2
10
11
11
10
15
15
M729G
12
2
13
8
12
M729A
M729V
11
10
9
9
V483G
V483A
2
2
13
10
11
11
2
8
2
M729L
M729F
2
2
11
V483Ｉ
M729W
M729N
13
2
V483F
12
11
10
15
25
35
Time (min)
45
30
45
35
40
Time (min)
Figure 1. Gas chromatography (GC) profiles of the wild-type Euphorbia tirucalli -
amyrin synthase and its V483X (A) and M729X (B) variants. The Gly and Ala variants
of Val483 afforded a large amount of monocyclic camelliol C (8) and a very small
amount of monocyclic achilleol A (9). By contrast, the Gly and Ala mutants of Met729
produced germanicol (13) in large quantities. The GC conditions for (A) and (B) were
as follows: column, J&W, DB-1, capillary (length 30 m, I.D. 0.32 mm, film thickness
0.25 µm); injection temp., 300C; column temp., 190–250C (10C/min) and 250–
268C (0.35C/min); injection volume, 3.0 μL. Products 10–13 and 15 were identified
as follows: 10, butyrospermol; 11, tirucalla-7,24-dien-3-ol; 12, lupeol; 13, germanicol;
15, -taraxasterol.
20
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
Scheme 2. Product structures generated by the V483X, W534X, and M729X variants,
and the cyclization pathways to forming products 8–13 and 15.
21
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
Table 1. Product distribution ratio (%) for V483X mutants.
Oleanyl cation 7
β-Amyrin 2
Monocyclic cation 3
Dammarenyl cation 4
Butyrospermol
Tirucalla-7,24-
Camelliol C 8
Achilleol A 9
10
dien-3β-ol 11
Wild
94.7
8.3
─
─
3.1
3.5
3.4
2.2
0.5
0.5
V483G
V483A
86.9
56.8
4.8
1.1
38.1
V483I
V483F
94.0
─
─
─
─
3.6
2.4
─
─
─
─: not detected.
Butyrospermol
Camelliol C
β-Amyrin
Tirucalla-7,24-dien-3β-ol
van der Waals’ volume (nm3)
Achilleol A
120.0
100.0
80.0
60.0
40.0
20.0
0.0
0.8
0.6
0.4
0.2
0
100
89.1
82.0
65.2
-0.2
0.0
wild
V483G V483A
V483I
V483F
Figure 2 Enzyme activities of V483X variants relative to that of wild-type Euphorbia
tirucalli -amyrin synthase. Open circles denote the van der Waals volume (nm3) of the
side chains of the amino acids.[10] The values were reported to be as follows: Val (wild
type), 0.25674; Gly, 0.00279; Ala, 0.05702; Ile, 0.37671; and Phe, 0.55298. The
22
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
production of -amyrin roughly increased in proportion to the increase in the van der
Waals volume (see the right-pointing arrow). The relative enzymatic activities of the
variants were determined according to the data in Fig. S6.
23
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
Tirucalla-7,24-dien-3β-ol
Butyrospermol
β-Amyrin
van der Waals volume (nm3) of side chain
120
100
80
60
40
20
0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
100
8.4
1.4
6.9
5
3.6
4.1
0.8
Figure 3. Enzymatic activities of Trp534 variants relative to that of the wild-type
Euphorbia tirucalli -amyrin synthase. Significant differences between the aromatic
variants (Phe and Tyr mutants) and the aliphatic variants (e.g., Leu and Met mutants)
were not observed (see also Fig. S7.2), indicating that the -electrons at this position are
unlikely to be responsible for the polycyclization cascade. The van der Waals volumes
were plotted according to literature data.[10] The relative enzymatic activities of the
variants were determined according to the data in Fig. S7.
24
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
Table 2. Product distribution ratio (%) for the M729X mutants.
Oleanyl cation 7
Lupanyl cation 6
Dammarenyl cation 4
2
13
15
12
11
10
─
30.6
27.0
4.8
─
─
2.7
1.1
─
─
─
─
-
─
4.1
2.6
─
Wild
95.9
53.5
62.8
88.6
98.9
87.8
─
2.0
3.1
1.8
3.3
─
2.1
6.0
4.7
3.3
1.1
9.0
─
M729G
M729A
M729V
M729L
M729F
M729W
M729N
─
─
─
3.2
─
─
─
34.5
54.4
1.8
6.5
2.8
─: not detected.
Germanicol (13)
Lupeol (12)
Tirucalla-7,24-dien-3β-ol (11)
β-Amyrin (2)
ψ-Taraxasterol (15)
Butyrospermol (10)
van der Waals
120
100
80
60
40
20
0
1
100
101.1
0.8
79.7
80.2
72.8
69.9
0.6
65.2
0.4
0.2
0
-0.2
0.0
Wild M729G M729A M729V M729L M729F M729W M729N
Figure 4. Enzymatic activities of the M729X variants relative to that of wild-type
Euphorbia tirucalli -amyrin synthase. Open circles denote the van der Waals volume
(nm3) of the side chains of the amino acids.[10] The production of -amyrin was roughly
increased in proportion to the increase in the van der Waals volume (see the right-pointing
arrow). The relative enzymatic activities of the variants were determined according to the
data in Fig. S8.
25
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
(B)
(A)
3.9
4.1
4.3
2.9
3.6
2.6
2.9
3.3
Figure 5. (A) Homology modeling of -amyrin synthase displayed with PyMOL
(http://www.pymol.org). This model was constructed by ESyPred3D
(http://www.unamur.be/sciences/ biologie/urbm/ bioinfo/esypred/), based on the X-ray
crystal structure of human lanosterol cyclase (PDB ID: 1w6k).[13] Gray, lanosterol
molecule; Black, Met729 residue; Beige, Val483 residue; Pink, Trp534. The numbers
indicate distances in Å units. The distance between the carbon atom and the center of
the -system (dC-X) in the EtAS structure are estimated to be 3.9─4.3 Å. Brandl et al.
reported that CH- interaction frequently occurs when the dC-X is ca 3.3-4.4 Å.[11a] The
shortest distance between the methyl group of Val483 and the indole ring (green solid
line) are estimated to be 3.6 Å, which is further indicative of CH-complexation
between the side chains, because the shortest distance (3.5─3.9 Å) is also reported as
the index of CH- interaction in protein structures. [11b, c] The values of 2.6–3.0 Å
indicate that the three amino acids are in close proximity to lanosterol or the substrate
ligand.
(B) The arrangements of the three active site residues on the folded 2,3-oxidosqualene
substrate 1 are illustrated.
26
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
(A) Wild-type
(B) V483G
(C) V483A
3.9
4.1
4.3
2.9
2.6
2.9
3.3
(E) V483F (1)
(F) V483F (2)
(D) V483I
Figure 6. Homology models of the mutated Euphorbia tirucalli -amyrin synthases
targeted for the Val483 position. The preferred conformations (rotamers of the side
chain) were optimized by PyMOL software. The blue oval shapes display the free space
between the mutated residues and the ligand.
(A)Wild type. CH- complexation with Trp534. The space between the ligand and
Val483 is compact owing to the appropriate steric bulk of the Val residue.
(B) The V483G mutant. A large space exists between Gly and Trp534, leading to
decreased CH- interaction.
(C) The V483A variant. A somewhat larger space and decreased CH- interaction are
evident compared with those of the wild-type.
(D)The V483I variant. Strong CH- interaction and compactness between the ligand are
evident.
(E) and (F) for the V483F mutant. The orientation of the Phe side chain is different
between (E) and (F), which were constructed by minimizing the steric hindrance
between the ambient amino acids and Phe. A large space is still generated, and the
organized local structure around the A/B-ring formation sites collapses as a result of
collision between the side chains of Phe and Trp534, which would probably lead to
incorrect positioning of the Trp534 side chains.
27
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
(B) W534A
(C) W534I
(A) Wild-type
3.9
4.1
4.3
2.9
Large space
Large space
2.6
2.9
3.3
(E)W534F
(D) W534M
(F) W534Y
Large space
Figure 7. Homology models of the mutated Euphorbia tirucalli -amyrin synthases
targeted for the Trp534 position.
The Trp residue is oriented perpendicularly to the ligand around the B/C-ring formation
sites (marked with a red arrow). The side chains of the mutated residues are arranged so
as to minimize the steric hindrance between the surrounding residues in the cavity.
Blue-filled ovals represent the free space between the mutated residues and the ligand.
For the aliphatic variants substituted with Ala (B), Ile (C), or Met (D), large vacant
spaces are generated between the ligand and the substituents and therefore CH-
interaction does not occur. The green arrows in (E) and (F) display the orientation of the
side chains of the Phe and Tyr residues, respectively. These aromatic residues are not
arranged perpendicularly to the ligand, which is different from the red arrow (wild-type
(A)), although CH- interactions between the benzene rings and the methyl group of the
Val483 residue likely occur because of their close proximity to each other (3.2 or 3.4 Å).
28
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700368
ChemBioChem
TOC (Table-of-Contents)
The appropriate steric sizes of the three residues facilitate the folding the substrate into a chair-
chair-chair-boat-boat conformation, notably CH- interaction between the Val and the Trp
confers the robust protein architecture around A/B/C-ring formation site.
29
This article is protected by copyright. All rights reserved.