Improved selectivity of an engineered multi-product terpene synthase

Article abstract of DOI:10.1039/c4ob00479e

Mutation of the sesquiterpene synthase Cop2 was conducted with a high-throughput screen for the cyclization activity using a non-natural substrate. A mutant of Cop2 was identified that contained three amino acid substitutions. This mutant, 17H2, converted the natural substrate FPP into germacrene D-4-ol with 77% selectivity. This selectivity is in contrast to that of the parent enzyme in which germacrene D-4-ol is produced as 29% and α-cadinol is produced as 46% of the product mixture. The mutations were shown to each contribute to this selectivity, and a homology model suggested that the mutations lie near to the active site though would be unlikely to be targeted for mutation by rational methods. Kinetic comparisons show that 17H2 maintains a k_cat/K_M of 0.62 mM^-1 s ^-1, which is nearly identical to that of the parent Cop2, which had a k_cat/K_M of 0.58 mM^-1 s^-1. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c4ob00479e

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
PAPER
Improved selectivity of an engineered
multi-product terpene synthase†
Cite this: Org. Biomol. Chem., 2014,
12, 4013
Ryan Lauchli,*‡§ Julia Pitzer,‡ Rebekah Z. Kitto, Karolina Z. Kalbarczyk and
Kersten S. Rabe*¶
Mutation of the sesquiterpene synthase Cop2 was conducted with a high-throughput screen for the
cyclization activity using a non-natural substrate. A mutant of Cop2 was identified that contained three
amino acid substitutions. This mutant, 17H2, converted the natural substrate FPP into germacrene D-4-ol
with 77% selectivity. This selectivity is in contrast to that of the parent enzyme in which germacrene
D-4-ol is produced as 29% and α-cadinol is produced as 46% of the product mixture. The mutations
were shown to each contribute to this selectivity, and a homology model suggested that the mutations lie
near to the active site though would be unlikely to be targeted for mutation by rational methods. Kinetic
comparisons show that 17H2 maintains a kcat/KM of 0.62 mM−1 s−1, which is nearly identical to that of the
Received 3rd March 2014,
Accepted 9th April 2014
DOI: 10.1039/c4ob00479e
www.rsc.org/obc
parent Cop2, which had a kcat/KM of 0.58 mM−1 s−1
.
screen the large number of mutants generated via random
mutagenesis, many of which may be misfolded. Furthermore,
Introduction
Terpene synthase enzymes exhibit exceptional control over the even when focusing on mutations in the active site or those
formation of a wide variety of products via highly complex cat- identified using homology or structural information, the
ionic cyclizations.1–3 For example, from a single 15-carbon sub- resulting mutants frequently suffer from decreased activity.
strate, farnesyl diphosphate (FPP), terpene synthases can For example, mutation to the H-α1 loop of sesquiterpene
generate over 300 distinct products. In many cases, a terpene synthases Cop4 and Cop6 severely impeded the kinetic behav-
synthase can guide the reaction of the substrate to just one of ior of these enzymes.9 In another example, thermostabilization
these hundreds of possible products with high product selecti- of the sesquiterpene tobacco epi-aristolochene synthase
vity and form multiple chiral centers in the process. Thus, caused lowered activity in the thermostabilized mutant.8 It is a
terpene synthases control some of the most complex reactions general problem in terpene synthase engineering that
in nature. Understanding and controlling these reactions to mutations in these enzymes often lead to decreased activity.
predict their products, to alter their product distributions, and
to increase the activity has proven to be difficult.
We are aware of only one previous study using a high-
throughput assay to screen terpene synthase mutants that
Several groups have recently taken on the challenge of alter- were engineered for altered product formation.4 The assay
ing and controlling the course of terpene synthase reactions identified terpene synthase mutants with an increased likeli-
using mutagenesis.4–12 These studies have largely focused on hood of correct folding, yet it provided no information on the
mutations either surrounding the active site or at locations cyclization activity of these mutants. Therefore, significant
identified in close homologues. This limitation exists because effort was expended in a secondary screen to determine which
there has been a lack of an efficient system with which to mutants maintained the activity.
We recently disclosed a high-throughput assay for the cycli-
zation activity in sesquiterpene synthases and subjected a
highly selective fungal terpene synthase to two rounds of
California Institute of Technology, Division of Chemistry & Chemical Engineering,
directed evolution.13 In the course of these two rounds, in
1200 E. California Blvd., Pasadena CA 91125, USA
which the enzyme was evolved for thermostability, the high
†Electronic supplementary information (ESI) available: The product distribution
plot, kinetic plots, nucleic acid and amino acid sequences. See DOI: selectivity of the enzyme for its native product was maintained
10.1039/c4ob00479e
and there was no apparent loss of activity. In the current study,
we applied random mutagenesis to a multi-product terpene
synthase using this high-throughput assay to identify a mutant
‡RL and JP contributed equally to this work.
§Current address: Sustainable Chemistry Solutions, Inc., 1052 W. Alameda
Avenue, #110, Burbank, California USA 91506. E-mail: ryan@bio-catalyst.com
that maintained the wild-type activity while concurrently chan-
ging and improving the product selectivity of the enzyme.
¶Current address: KIT-Campus North, Hermann-von-Helmholtz-Platz 1,
76344 Eggenstein-Leopoldshafen, Germany. E-mail: kersten.rabe@kit.edu
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 4013–4020 | 4013

View Article Online
Paper
Organic & Biomolecular Chemistry
Results
Cop2 is a multi-product sesquiterpene synthase from Coprinus
cinereus, and no crystal structure is known for this enzyme.14
The product composition formed from Cop2 includes products
believed to result from a number of distinct cyclization
chemistries.
Error-prone PCR (epPCR) was used to create random point
mutations in the Cop2 gene. The Cop2 mutants thus generated
were expressed as a library in E. coli. The library consisted of
about 3000 clones with an average error rate of 1.4 base
pair (bp) mutations per gene. This library was screened for
the cyclization activity using the high-throughput screen we
previously developed. The screen relies upon the usage of a
surrogate substrate (S1), which contains a vinyl methyl ether
moiety at the head of the molecule in place of the dimethyl–
alkene moiety found in the natural substrate, FPP. Terpene
synthase-mediated cyclization of S1 is measured by methanol
release using an enzyme-coupled colorimetric readout. An
active mutant denoted by 17H2 was identified in this library
with comparable activity to the parent as measured by this
screen.
Scheme 1 Proposed formation of the two major products from FPP
catalyzed by Cop2.
The mutant of Cop2 denoted by 17H2 was shown to be a
triple mutant containing mutations L59H, T65A and S310Y.
The reactions of both Cop2 and 17H2 with the natural sub-
strate FPP were investigated by GC-MS measurements. Reac-
tions of purified enzymes were run in buffer in vials with an
overlayer of hexanes. Though previous headspace analysis of
Cop2 reactions in cell culture led to the proposal that Cop2 is
a germacrene A synthase,14 analysis of the terpenes extracted
into overlayers showed that this enzyme generates a mixture of
products with α-cadinol and germacrene D-4-ol present in the
largest amounts. Because α-cadinol and germacrene D-4-ol are
alcohols rather than hydrocarbons (like germacrene A), they
were not proportionally represented by headspace analysis
because they have both lower vapor pressure and increased
solubility in the water layer. Scheme 1 shows a portion of the
proposed reaction mechanism for the formation of these two
products. No single product comprises more than 50% of the
product mixture, and Cop2 is therefore a multi-product sesqui-
terpene synthase. In contrast, the triple mutant
17H2 generates a much more selective product mixture as
shown in Fig. 1. This mixture is comprised of 77% of germa-
crene D-4-ol, and no α-cadinol is produced. Thus, the product
selectivity was significantly improved.
Fig. 1 Product distribution in % of Cop2, the triple mutant 17H2 and
the single mutants L59H, T65A and S310Y analyzed by GC-MS.
Table 1 Summary of the kinetic parameters of Cop2 and 17H2
with FPP
Enzyme
concentration kcat
+/− [µM]
Km
vmax
kcat/Km
Enzyme [µM] +/− [µM h−1
]
[s−1
]
[mM−1 s−1
]
We measured the activity of 17H2 on FPP by GC measure-
ments. Kinetic studies on both 17H2 and Cop2 were estimated
using FPP as the substrate, and Km values were determined
assuming Michaelis–Menten kinetics to be 25.5 µM for Cop2
and 22.6 µM for 17H2. The kcat values of 0.015 s−1 and 0.014 s−1
Cop2
17H2
25.5 6.7 10.7
22.6 7.1 25.3
1.0 0.2
2.0 0.5
0.015 0.58
0.014 0.62
Because 17H2 contained three point mutations, we sought
for Cop2 and 17H2, respectively, were also similar. Conse- to elucidate whether one of the single point mutations caused
quently, the kcat/Km values of 0.58 mM−1 s−1 and 0.62 mM−1 s−1 the high selectivity or whether the combination of the three
were nearly identical and show that the mutant retained the mutations was responsible for this effect. The three single
activity of the parent (see Table 1). Therefore, screening with a mutants L59H, T65A and S310Y were therefore constructed,
surrogate substrate for active enzymes gave a useful approxi- and the reactions of FPP catalyzed by these mutants were
mation of the activity on the native substrate, FPP.
analyzed.
4014 | Org. Biomol. Chem., 2014, 12, 4013–4020
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Fig. 2 Partial alignment of the amino acid sequences of Cop2 with other sesquiterpene synthases. The residues mutated in 17H2 are highlighted in
green. The residues involved in substrate binding are marked in yellow. The blue boxes highlight the conserved aspartate rich motif and the N(S/T)(E/D)
triad, necessary for the binding of Mg2+ ions. Cop2, 1DGP_A: aristolochene synthase from Penicillium roqueforti, 1PS1_A: pentalenene synthase.
Fig. 3 Homology model of Cop2 based on the pentalenene synthase structure 1HM7.15–17 The FFP analog from the structure 4KUX is shown in
order to indicate the active site. The three residues mutated in 17H2 are shown in pink as follows: S310 is at the end of the J-helix (in red), L59 is in
the upper portion of the C-helix (in orange) and T65 is in the lower portion of the C-helix. Also shown are the G-helix in light blue and the H-α1 loop
in green. The image on the right is a view from the top.
Analysis of the three single mutants showed that, in com-
The locations of the mutations in 17H2 were further eluci-
parison with Cop2, L59H and S310Y produce elevated germa- dated with a homology model of Cop2 built using SwissModel
crene D-4-ol amounts and reduced α-cadinol amounts from upon the crystal structure of the bacterial sesquiterpene
FPP. The mutant T65A also showed a small improvement in synthase, pentalene synthase (Fig. 3).18 The two enzymes
germacrene D-4-ol selectivity. None of the single mutants com- shared 21% sequence homology, and pentalene synthase had
pletely abolished α-cadinol production. These results are the highest homology to Cop2 of any terpene synthase crystal
shown in Fig. 1 and suggest that a synergistic effect of the structures. The model shows that all three mutations are
three mutations is responsible for the improvement in the located near the active site yet do not line the active site cavity.
product selectivity of 17H2 towards germacrene D-4-ol.
Both L59H and T65A mutations are located within the C-helix
Though triple mutant was identified from the (orange in Fig. 3), and the S301Y mutation is located at the
a
Cop2 mutant library, any of the three single mutants would be end of the J-helix (red in Fig. 3). The location of these
found using our screening system, and the mutants all give mutations is non-obvious—they would likely not be targeted
comparable activities when measured with the assay (data not for site-directed or site-saturation mutagenesis.
shown).
An alignment of Cop2 with aristolochene synthase and pen-
talenene synthase was constructed (Fig. 2).15–17 The structures
Discussion
and the corresponding active sites of aristolochene synthase
and pentalenene synthase are known,16,17 and a comparison
Enzymatic terpene cyclization is initiated by the formation of a
reactive carbocation.1–3 Class I enzymes, which include sesqui-
with Cop2 reveals that residues which are mutated in 17H2 (as
highlighted in green) are directly adjacent to active site resi-
dues for T65 and S310, and five residues are removed from an
active site residue in the case of L59. Despite L59 being the
mutated residue furthest from the predicted active site cavity,
it causes the largest change in product selectivity.
terpene synthases like Cop2, generate the carbocation by the
release of the pyrophosphate (PPi) group of the substrate.
Enzymes of this class contain characteristic DDxxD and NSE/
DTE motifs which facilitate the binding and subsequent clea-
vage of the PPi induced by three Mg2+ ions as cofactors.19–21
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 4013–4020 | 4015

View Article Online
Paper
Organic & Biomolecular Chemistry
Additionally, a basic motif (RY) is also involved in the coordi-
Keasling et al. used a humulene synthase homology model
nation of the Mg2+ and PPi complex. Mutation of the amino to identify active site residues.10 For each of 19 amino acids
acids involved in Mg2+ and PPi binding generally results in lining the active site, site-saturation mutagenesis was con-
severely attenuated enzyme activity. Mutations in this region in ducted to generate libraries of mutants. Six mutants contain-
aristolochene synthase, for example, were reported to result in ing up to
5 mutations each demonstrated impressive
altered product distributions along with a 100–6500 fold selectivities for products up to 85.7%, and several of these
decrease in the observed kcat/Km.6 Furthermore, mutations in mutants demonstrated the activity comparable to that of the
the Mg2+ coordinating motifs of a fungal trichodiene synthase parent. The method unfortunately required the extensive and
reduced kinetic parameters as well as product specificity.7 tedious use of low-throughput GC-MS measurements. Keasling
Consistent with these previous results, the present work identi- and co-workers conducted a separate study on the selective
fied no mutations in this area of the protein that maintained Gossypium arboreum (+)-δ-cadinene synthase.4 Random muta-
the activity.
genesis by epPCR and a high-throughput solubility screen
Within the terpene synthase active site, amino acids (pri- based on fusion to chloramphenicol acetyltransferase (CAT)
marily aliphatic and aromatic) direct the propagation of the were used to identify soluble mutants. From 1000 clones of the
carbocation through complex reaction cascades.3,19,20 epPCR library, 100 mutants were re-screened for the activity by
Additionally, reaction steps involving acid–base catalysis are GC-MS, for each of two rounds of directed evolution. A total of
thought to occur during sesquiterpene cyclization. The termi- 20 clones were identified that, in addition to the native
nation of the reaction cascade can occur either via a nucleo- product (+)-δ-cadinene, also produced the new product germa-
philic attack by water, leading to the formation of terpene crene D-4-ol. The best enzyme in this study was a double
alcohols or via proton abstraction to form double bonds. The mutant that demonstrated 53% selectivity for germacrene
active site of terpene synthases acts not only as a chaperone D-4-ol with only slight decreases in kinetic properties.22
for the correct folding of the substrate but it also rearranges
The double mutant from Keasling’s study and 17H2 from
the carbocation and protects it from premature quenching by the current study both produce germacrene D-4-ol.4 Keasling’s
solvent molecules. Most previous mutational studies of sesqui- hypothesis for the increased production of germacrene D-4-ol
terpene synthases have targeted residues within the active site. was that the mutations they found allowed water into the
Schmidt-Dannert and coworkers, for example, investigated active site. When water can reach the active site it leads to the
the effects of mutation in the H-α1 loop (which caps the active termination of the reaction cascade by hydroxylation. In the
site) of several sesquiterpene synthases from Coprinus ciner- absence of water, however, another cyclization occurs before
eus.9 These synthases include the α-cuprenene synthase Cop6, the reaction is quenched.
and the multi-product terpene synthases Cop3 and Cop4.
In the case of Cop2, water quenching is responsible for for-
Though mutation of residues in the Cop6 H-α1 loop caused mation of both of the major products — α-cadinol and germa-
little change in product selectivity, both Cop4 and Cop3 crene D-4-ol. This fact suggests that simple ingress of water
showed significant changes in product compositions upon into the active site is not responsible for the increase in selecti-
mutation. The mutants were much less active than the parent vity for germacrene D-4-ol mediated by 17H2. Instead, the
enzymes with at least 3-fold slower kcat values for Cop6 and three mutations may remove non-bonding interactions that
two orders of magnitude slower kcat values for Cop4.
would favor the second cyclization to form α-cadinol.
Weiss investigated the effects of structure guided mutations
Residues that cause an alteration of product selectivity upon
in tobacco epi-aristolochene synthase (TEAS).8 The crystal mutation in sesquiterpene synthases have been called plasticity
structure for this enzyme was previously solved, and the struc- residues.10,11 The search for plasticity residues has primarily
ture was used to introduce thermostabilizing mutations that been limited to active site residues regardless of whether active
were distant from the active site. Despite this distance, the site residues are ideal for this purpose. In Cop2, residues L59,
thermostabilized mutant identified in this work suffered from T65, and S310 are not predicted to be active site residues, yet
losses in the product selectivity and activity.
they are plasticity residues that alter product distributions and
Though some studies have indeed identified mutations that maintain the activity. The ideal plasticity residues may in fact
maintain or even improve the wild-type activity, these studies be located remote from the active site, and this appears to be
generally rely on structures and enzymes with close homologs. the case for Cop2. Further application of random mutagenesis
For example, Chen and Zhang solved the crystal structure of a and the high-throughput cyclization activity screen used here
bisabolene synthase, and used it to guide the mutation of the may reveal other instances of plasticity residues located in posi-
homolog amorphadiene synthase (82% amino acid identity).11 tions not currently amenable to structural prediction.
Mutation of T399 to serine increased the kcat of the enzyme
two-fold without a decline in product selectivity. In another
study, Noel and Chappell utilized the structure of TEAS to
guide the inter-conversion of the product distribution of this
enzyme and the close homolog henbane premnaspirodiene
Experimental
Instrumentation
synthase (HPS) (72% amino acid identity) without negatively Gas chromatography (GC) was performed on a Shimadzu
affecting the activity.5
GC-17A instrument equipped with an FID detector. The
4016 | Org. Biomol. Chem., 2014, 12, 4013–4020
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
column was an HP-5 column (Agilent, 30 m × 0.32 mm × suggested by the manufacturer. 3 μL of the gene and 2 μL of
0.25 μm film). For kinetic measurements, the temperature the vector backbone were combined with 15 μL of the mix
program was an initial temp of 90 °C for 1.0 min, followed by described for the one-step isothermal DNA assembly by
a temp ramp of 9.0 °C min−1 to 174 °C, followed by a ramp of Gibson et al.24 and the assembly was performed for 60 min at
45 °C min−1 to 260 °C, and the temp was held at 260 °C for 50 °C. An aliquot of 5 μL of the resulting solution was trans-
2.0 min. The split ratio was 1.0, the column flow was 1.68 mL formed into chemically competent E. coli DH5α. After plasmid
min−1, the injector temp was 250 °C, and the detector temp purification with the Zyppy™ Plasmid Miniprep Kit (Zymo
was 300 °C. Absorbance data from 96-well microtitre plates Research), correct assembly was verified by sequencing
were collected on a Tecan Infinite® M200. Proteins were puri- (Laragen). However a point mutation resulting in a frame shift
fied on an ÄKTAxpress with 1 mL HisTrap columns (GE and premature abortion of the translation was identified.
Healthcare Life Sciences). GC-MS measurements were con- Therefore the gene sequence was fixed in a site-directed muta-
ducted on a Shimadzu GCMS-QP2010 SE instrument in elec- genesis using the primers Cop2mt-for GACGAAATCA
tron ionization (EI) mode. The column used was a Shimadzu TATTTTTCTCGATTCCTGAAGACCGCTG and Cop2mt-rev
SHRXI-5MS (30 m × 0.25 mm × 0.25 μm film). The temperature GAGAAAAATATGATTTCGTCATTAGTCCTGCGCGTTTATC and
̲-
̲
program was an initial temp of 75 °C for 6.0 min, followed by Phusion polymerase (NEB) with HF buffer and 60 °C as the
a temp ramp of 9.0 °C min−1 to 180 °C, followed by a temp annealing temperature. The base pair to be inserted is under-
ramp of 35 °C min−1 to 280 °C, and the temp was held at lined. After digestion for 120 minutes at 37 °C with DpnI (Neb)
280 °C for 3.0 min. The split ratio was 20, the column flow was 5 μL were transformed into chemically competent E. coli
1.5 mL min−1 and the injector temp was 250 °C.
DH5α. After plasmid purification with the Zyppy™ Plasmid
Miniprep Kit (Zymo Research) the success of the base inser-
tion was verified by sequencing (Laragen). The plasmid was
General procedures
Alcohol oxidase (AOX) was obtained from MP Biomedicals. then transformed into BL21(DE3) for protein overexpression.
Surrogate 1 (S1) was synthesized as described.13 Farnesyl
diphosphate (FPP) was synthesized from farnesol according to
Cop2 library formation
the method of Poulter.23 Primers were purchased from Inte- A library of Cop2 mutants was produced through error-prone
grated DNA Technologies (IDT). E. cloni Express electrocompe- PCR using the primers pETCop2Fwd2 (AAAAAACCATATGC-
tent E. coli BL21 (DE3) was purchased from Lucigen. Alkaline CAAGTCCCGCGGGAG) and pETCop2Rev2 (TTTTTTCT-
phosphatase, Taq polymerase, T4 ligase, and restriction CGAGTTTAGGCCAAAACCCGCAC). The template was the
enzymes DpnI, XhoI, and NdeI were purchased from New Cop2 gene in pET-22b, using Taq (NEB) along with 200 μM
England Biolabs (NEB). DNA sequencing was performed by MnCl2 and 0.2 mM of each dNTP (Roche). These conditions
Laragene (Culver City, CA, USA). TB refers to terrific broth gave an approximate mutation rate of 1.4 base pairs per gene.
from RPI, and LB refers to Luria broth from the same vendor. The inserts produced by these epPCR reactions were double
The Cop2 gene was a gift from Professor Claudia Schmidt- digested with XhoI and NdeI (NEB). A pET-28a(+) plasmid with
Dannert and was supplied as pUC-Cop2. The terpene synthase a low-activity terpene synthase gene (sscg_03688, from Pro-
gene 03688 in pET-28a(+) was a gift from Professor David fessor David E. Cane) was double digested with XhoI and NdeI
E. Cane. Standard buffer consisted of 50 mM PIPES, 10 mM (NEB) to remove the gene and then dephosphorylated using
MgCl2, 100 mM NaCl, and 2 mM DTT at pH 7.6. Transfers of calf intestinal alkaline phosphatase (NEB). Both digestions
cultures and other liquids in 96-well format were done using a were followed by gel purification and extraction using a kit
Hamilton Microlabs Nimbus Liquid Handling Robot.
(Promega) to produce purified DNA. The insert and backbone
were ligated using T4 ligase (NEB) at 16 °C overnight and the
Cloning of Cop2 into pET-22b
DNA was cleaned and concentrated using
a kit (Zymo
The gene for Cop2 was amplified from the vector pUC-Cop2 Research). Electro-competent BL21 (DE3) cells were trans-
with the primers Cop2-for TTAACTTTAAGAAGGAGATATACA- formed with 1 µL of the cleaned ligation product and the
T̲A̲T̲G̲
CCCGCACAGC using Phusion polymerase (NEB) with HF picks. The 96-well plates contained 300 µL of LB media per
buffer and 60 °C as the annealing temperature. The part of the well and 0.05 g L−1 of kanamycin. Cultures were grown over-
sequence that is homologous to the vector backbone is in night at 37 °C while shaking at 225 rpm and 80% rh. Next,
italics; the start codon is underlined. The vector backbone was 50 µL aliquots of the cultures were transferred into 96-well deep
amplified from pET-22b using the primers pET22-for well plates (2 mL volume per well) containing 800 μL of TB media
GCTTTGTTAGCAGCCGGATCTCA and pET22b-rev CCCCTCTA- as well as 0.05 g L−1 of kanamycin. Cryostocks of the libraries
GAAATAATTTTGTTTAACTTTAAGAAGGAGATATACAT and the were made during transfer by transferring 120 µL per well of the
LongRange PCR Kit from Qiagen at the annealing temperature cultures to 96-well microtiter plates in 20% glycerol. These cryo-
of 50 °C. After separation on a 1% agarose gel the PCR pro- stocks were flash-frozen on dry ice and stored in the −80 °C
ducts were excised and purified using the Zymoclean Gel DNA freezer. The deep-well expression plate cultures were grown at
Recovery Kit (Zymo Research) following the procedures 37 °C for 4 h shaking at 225 rpm and 80% rh. Protein production
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 4013–4020 | 4017

View Article Online
Paper
Organic & Biomolecular Chemistry
was then induced by adding 50 µL per well of IPTG in LB media determined with a spectrophotometer (Nanodrop). The corres-
for a total IPTG concentration of 0.5 mM. The plates were then ponding fragments were combined via fragment assembly PCR
returned to the shaker to sit at 25 °C with 225 rpm and 0% rh for and the amount of the large and the small fragments was set
24 h. The cells were then pelleted by centrifugation at 3000g for according to their relative sizes. After 15 cycles with an exten-
10 min at 4 °C. The plates containing the cell pellets were frozen sion time of 20 s, the primers Cop2fwd2 and Cop2rev2 were
at −20 °C overnight until the screening protocol began.
added and another 20 cycles were started to amplify the full
construct.
The following primers were used for the preparation of the
single mutants of Cop2:
High-throughput screening
Screening for all libraries and substrates was initiated by
thawing deep well plates with cell pellets at room temperature
for 30 min. Then, to lyse the cells, 300 µL of lysis buffer were
added to each well. The lysis buffer consisted of Standard
Buffer (pH 7.6, 50 mM PIPES, 10 mM MgCl2, 100 mM NaCl,
2 mM DTT) containing 10% betaine, 0.5 mg mL−1 lysozyme
and 0.02 mg mL−1 DNAse I. The plates were vortexed to resus-
pend the pellets, and then were incubated at 37 °C for 1 h. Cell
debris was pelleted by centrifugation at 5000g at 4 °C for
15 min. Using a liquid-handling robot, 100 µL of the lysate
from each well was transferred to 96-well microtiter plates, and
then 100 µL of 0.50 mM S1 in Standard Buffer was added.
After the lysate and S1 were incubated for 1 h at 25 °C,
10 µL of alcohol oxidase (50 µL of AOX stock solution dissolved
in 950 µL of 0.1 M KPi, pH 8.0) were added to each well, and
the plates were shaken on a table top orbital shaker at 600 rpm
for 10 minutes at room temperature. Next 16 µL of 0.5 M ethy-
lenediaminetetraacetic acid (EDTA, pH 8.0) were added to each
well, and the plates were shaken at 600 rpm for 1 min at room
temperature. Finally, 50 µL of 0.16 M Purpald® (Sigma-
Aldrich) in 2 M NaOH was added to each well. The plates were
shaken at 600 rpm for 25 min at room temperature, after
which a microtiter plate reader was used to record the absor-
bance of the wells at 550 nm.
Primer
Nucleotide sequence
pETCop2Fwd2
pETCop2Rev2
Cop2L59HrevCor
Cop2L59Hfwd
Cop2T65AfwdAg
Cop2T65ArevAg
Cop2S310Yfwd
Cop2S310Yrev
AAAAAACCATATGCCAAGTCCCGCGGGAG
TTTTTTCTCGAGTTTAGGCCAAAACCCGCAC
CTCGCCCGCTTTGTGTGCCATAAAC
GTTTATGGCACACAAAGCGGGCGAG
CGAGCTCGCTGCTGCATGC
GCATGCAGCAGCGAGCTCG
GGGTCTCTAAATTGGTACATCGACGGGACA
TGTCCCGTCGATGTACCAATTTAGAGACCC
The DNA inserts were gel extracted and purified and 1 μg
DNA was digested with 1 µL XhoI and 1 µL NdeI for 1 h at
37 °C. The same applies for the vector DNA. The pET28a(+)
vector backbone was cut with XhoI and NdeI for 1 h and
dephosphorylated with 1 µL alkaline phosphatase (Calf Intes-
tine, NEB) for 30 min.
Ligation and transformation
After another purification step by gel extraction, ligation of the
cut insert (10 μL) and the digested and dephosphorylated
vector backbone (2.5 μL) was performed with T4 DNA ligase
(1 μL) overnight at 16 °C. The ligation products were worked
up with the DNA Clean & Concentrator kit (Zymo Research)
and eluted in 6 μL dH2O. For the transformation 1 μL of the
concentrated ligation product was added to 50 μL of electro-
competent E. coli BL21 (DE3) cells. The cells were transferred
into a cooled electroporation cuvette and pulsed with a voltage
of 2.5 kV. Immediately afterwards 950 μL of SOC media was
added and the cells were incubated for 1 h at 37 °C and 225
rpm. After recovery 30 μL and 200 μL of the cell suspension
were plated on LBKan agar plates with a kanamycin sulfate con-
centration of 50 mg L−1. The plates were incubated overnight
at 37 °C.
In order to identify positive clones, overnight cultures were
prepared and Minipreps were prepared with the QIAprep Spin
Miniprep Kit and sent for sequencing to Laragen with the
primers T7 Promoter and T7 Terminator. Glycerol stocks with
a final glycerol concentration of 20% were prepared from
clones with the correct sequences, frozen on dry ice and stored
at −80 °C.
Construction of the single mutants of 17H2 via fragment
assembly PCR
In order to analyze the three mutations of 17H2 individually,
the corresponding single mutants L59H, T65A and S310Y were
prepared. The forward and reverse primers with the respective
mutations were designed and the PCR reactions were con-
ducted with 65 ng Cop2 DNA template, 0.5 μL Phusion Poly-
merase, 1 μL of forward and reverse primer [10 μM], HF buffer
5×, and 1 μL dNTPs [10 mM].
The following temperature program was used for the PCR
for 30 cycles.
Temperature
Time
98 °C
98 °C
58 °C
72 °C
72 °C
10 °C
30 s
10 s
20 s
15 s/1 kb
10 min
Until end
Cultivation and protein expression
The proteins were overexpressed in E. coli BL21. Two overnight
The samples were loaded on a 1% agarose gel and purified cultures were prepared from the strain of interest by inoculat-
using the Promega Kit for DNA extraction and purification. ing 5 mL of LBKan medium with cells from a cryostock or from
The DNA was eluted in 50 μL dH2O and its concentration was a plate. The cultures were grown at 37 °C and 225 rpm
4018 | Org. Biomol. Chem., 2014, 12, 4013–4020
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
overnight. For the main cultures, 1 L TB media was prepared ettes that were stuffed with small pieces of KimwipesTM tissue
in 2 L Erlenmeyer flasks and autoclaved before the antibiotic and 250 μL of the organic layer were then filtered through the
kanamycin sulfate was added to a final concentration of 50 mg sodium sulfate. The filtrates were collected in a GC vial with a
L−1. The TBKan media were inoculated with the overnight cultures low-volume insert and used for GC-MS analysis. Relative
and incubated at 37 °C and 200 rpm. The ODs at 600 nm were quantifications were performed using ion abundances.4 Reac-
measured with a spectrophotometer. After ∼3–4 h OD600s of tions were run in triplicate.
0.5–1.0 were obtained and protein expression was induced by the
addition of IPTG to an end concentration of 0.5 mM (0.12 g IPTG
Kinetic measurements
were dissolved in 5 mL dH2O and added to 1 L culture media). In order to estimate the kinetic parameters of the parent Cop2
After the induction, the temperature was decreased to 25 °C and and the triple mutant 17H2, GC-scale reactions were con-
the cultures were incubated for another 24 h at 200 rpm. Then ducted as above with different substrate concentrations. All
the cells were harvested by centrifugation at 4000g and 4 °C for reactions were performed in 2 mL GC vials in 500 μL Standard
15 min. The pellets were resuspended in ∼20 mL Buffer A Buffer with 10% betaine. In the case of the wild type Cop2, an
(pH 7.0, 50 mM Tris/HCl, 5 mM MgCl2, 300 mM NaCl, 2 mM enzyme concentration of 0.2 μM was used, whereas for the
DTT, 10% (v/v) glycerol, 10 mM imidazole for Buffer A and mutant 17H2 a concentration of 0.5 μM was used. FPP concen-
500 mM for Buffer B) and frozen at −20 °C.
trations from 6.25 to 100 μM and from 10 to 200 μM were used
for the wild type and mutant enzymes, respectively. All
samples were closed with a septum cap and incubated at room
Protein purification
After thawing the cell pellets, the lysis of the resuspended cells temperature. The reactions were stopped after 5, 10, 15, 20, 30,
was performed by ultrasonication for 2 min (pulse on 5 s, 40, 50 and 60 min by the addition of 40 μL of EDTA (0.5 M, pH
pulse off 5 s). The cell lysates were constantly kept on ice and 8) and vortexing. The organic layer was filtered through
after sonication they were centrifuged at 20 000g for 30 min at sodium sulfate as described above and used for GC measure-
4 °C. The supernatants were filtered through 0.2 μm cellulose ments. Double determination was made for all samples. The
filters and if necessary 0.02 mg mL−1 DNAse was added. The curve fitting was done with Prism. The same results were
purification of the proteins was performed with an ÄKTAxpress obtained with the program Mathematica. The product for-
system from GE Healthcare. Since the proteins were hexahisti- mation increased linearly for the first 30–40 min. The peak
dine-tagged, 1 mL HisTrap Ni-NTA columns were used for the areas were converted into the estimated product concen-
purification and an imidazole gradient was applied for elution trations with the help of the internal standard dodecane at a
of the protein using Buffers A and B above. The fractions con- known concentration of 200 µM. Therefore, the products are
taining proteins were identified by monitoring the absorbance quantified as dodecane equivalent μM units. Though this
at 280 nm, and then pooled. The buffer was exchanged with method is only for estimation of product concentrations, the
Standard Buffer containing 10% betaine (w/v) using Amicon® use of the same comparison for both Cop2 and 17H2 gives a
Ultra-15 Ultracel® cellulose 10 000 MW-cutoff centrifugal useful approximation of relative kinetic behavior.
filters. The purified proteins were aliquoted and stored at
Identification of germacrene D-4-ol and α-cadinol
−80 °C until they were further used. The protein concentration
was determined by measuring the absorbance at 280 nm. Compounds were identified by comparison of their retention
Therefore the protein was denatured in 500 μL 6 M guanidi- time, mass fragmentation patterns, and Kovat’s indices with
nium chloride and the average of three different measure- the known values. Fragmentation patterns were searched for
ments was calculated. To check the purity and molecular similarity in the NIST library in the Shimadzu 2010 GC-MS
weights of the proteins, SDS-PAGE analysis was performed.
software. Kovat’s indices were determined by running a stan-
dard sample containing linear alkanes from C10 to C16. Ger-
macrene D-4-ol was readily determined using these methods.
GC-MS analysis
To test the conversion of the natural substrate FPP, GC-scale The retention times and fragmentation patterns of α-cadinol,
reactions were performed at an enzyme concentration of 2 μM however, are quite similar to other diastereomers. To aid in
and a substrate concentration of 300 μM. The reactions were product identification, T-muurolol synthase sscg_03688 was
performed in 2 mL GC vials in a total reaction volume of expressed in E. coli containing the gene in pET-28a(+) as pro-
500 μL. First, Standard Buffer containing 10% betaine was vided by David E. Cane.25 The enzyme was purified as
added to the vials. The enzyme and substrate were added and described in the literature.25 GC-scale reactions were con-
the reactions were overlaid with 500 μL hexanes containing ducted using 2 μM of enzyme for Cop2, 17H2, and sscg_03688
200 μM dodecane as the internal standard (4.6 μL dodecane and 300 μL of FPP. Cop2 and 17H2 reactions were run in stan-
per 100 mL hexanes). The samples were closed with a septum dard buffer with 10% w/v betaine. The sscg_03688 reaction
cap and in the case of Cop2 they were incubated at 30 °C for was run in standard buffer with 20% glycerol. The reactions
2 h. The incubation time and temperature were chosen were incubated overnight at room temperature and worked up
depending on the current enzyme and experiment. Afterwards and analyzed by GC-MS as described above. Analysis of the
the reactions were stopped by adding 40 μL of EDTA (0.5 M, product by GC-MS showed that the product from Cop2 had a
pH 8) and vortexing. Sodium sulfate was added to glass pip- slightly longer retention time than T-muurolol, and a similar
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 4013–4020 | 4019

View Article Online
Paper
Organic & Biomolecular Chemistry
mass spectral fragmentation pattern, and was therefore identi- 12 D. B. Little and R. B. Croteau, Arch. Biochem. Biophys., 2002,
fied as α-cadinol.26–33
402, 120.
13 R. Lauchli, K. S. Rabe, K. Z. Kalbarczyk, A. Tata, T. Heel,
R. Z. Kitto and F. H. Arnold, Angew. Chem., Int. Ed., 2013,
52, 5571.
14 S. Agger, F. Lopez-Gallego and C. Schmidt-Dannert, Mol.
Microbiol., 2009, 72, 1181.
15 A. Marchler-Bauer, C. Zheng, F. Chitsaz, M. K. Derbyshire,
L. Y. Geer, R. C. Geer, N. R. Gonzales, M. Gwadz,
D. I. Hurwitz, C. J. Lanczycki, F. Lu, S. Lu, G. H. Marchler,
J. S. Song, N. Thanki, R. A. Yamashita, D. Zhang and
S. H. Bryant, Nucleic Acids Res., 2013, 41, D348.
16 J. M. Caruthers, I. Kang, M. J. Rynkiewicz, D. E. Cane and
D. W. Christianson, J. Biol. Chem., 2000, 275, 25533.
17 C. A. Lesburg, G. Zhai, D. E. Cane and D. W. Christianson,
Science, 1997, 277, 1820.
Conclusions
In this study, a multi-product terpene synthase, Cop2, was con-
verted to a more selective mutant by random mutagenesis. The
results suggest that random mutagenesis and screening com-
bined with an assay for the cyclization activity may be an
effective strategy to identify residues that influence the product
selectivity without imposing activity penalties. No structural
information was used to guide the introduction of mutations,
and the mutations were found at residues that would be non-
obvious targets for mutagenesis in structure-based approaches.
18 SWISS-MODEL, Protein Structure Bioinformatics group at
the Swiss Institute of Bioinformatics and the Biozentrum
University of Basel http://swissmodel.expasy.org/.
19 Y. Gao, R. B. Honzatko and R. J. Peters, Nat. Prod. Rep.,
2012, 29, 1153.
20 D. W. Christianson, Chem. Rev., 2006, 106, 3412.
21 D. J. Miller and R. K. Allemann, Nat. Prod. Rep., 2012,
29, 60.
22 Keasling reports higher selectivity and productivity when
no organic overlayer is used to sequester products. We do
not use these data for comparison because the more vola-
tile (+)-δ-cadinene is removed into the vapor phase and
thereby biases both the selectivity and productivity in favor
of germacrene D-4-ol.
23 V. J. Davisson, A. B. Woodside, T. R. Neal, K. E. Stremler,
M. Muehlbacher and C. D. Poulter, J. Org. Chem., 2006, 5,
4768.
Acknowledgements
We are grateful to Frances H. Arnold, Claudia Schmidt-
Dannert, David Cane, Scott Virgil, Robert M. Coates, and David
Christianson for their pioneering studies, invaluable advice,
encouragement, and materials. We thank Sabine Brinkmann-
Chen, John McIntosh, Thomas Heel, and Chris Farwell for
advice and assistance. RL acknowledges the support of the
NIH fellowship 1F32GM095061. The content is solely the
responsibility of the authors and does not necessarily rep-
resent the official views of the NIH. KSR thanks the Deutscher
Akademischer Austauschdienst (DAAD) for a postdoctoral fel-
lowship. KZK and RZK acknowledge the support of Summer
Undergraduate Research Fellowships from the California Insti-
tute of Technology.
24 D. G. Gibson, L. Young, R. Y. Chuang, J. C. Venter,
C. A. Hutchison III and H. O. Smith, Nat. Methods, 2009, 6,
343.
25 Y. Hu, W. K. W. Chou, R. Hopson and D. E. Cane, Chem.
Biol., 2011, 18, 32.
26 B. Pickel, D. P. Drew, T. Manczak, C. Weitzel,
H. T. Simonsen and D.-K. Ro, Biochem. J., 2012, 448, 261.
27 M. Hazzit, A. Baaliouamer, M. L. Faleiro and M. G. Miguel,
J. Agric. Food Chem., 2006, 54, 6314.
28 M. Jalali-Heravi, B. Zekavat and H. Sereshti, J. Chromatogr.
A, 2006, 1114, 154.
References
1 E. Breitmaier, Terpenes; Flavors, Fragrances, Pharmaca,
Pheromones, Wiley-VCH, Weinheim, 2006.
2 J. S. Dickschat, Nat. Prod. Rep., 2011, 28, 1917.
3 R. K. Allemann, Pure Appl. Chem., 2008, 80, 1791.
4 Y. Yoshikuni, V. J. Martin, T. E. Ferrin and J. D. Keasling,
Chem. Biol., 2006, 13, 91.
5 B. T. Greenhagen, P. E. O’Maille, J. P. Noel and J. Chappell,
Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 9826.
6 B. Felicetti and D. E. Cane, J. Am. Chem. Soc., 2004, 126, 7212.
7 L. S. Vedula, J. Jiang, T. Zakharian, D. E. Cane and 29 W. A. Asuming, P. S. Beauchamp, J. T. Descalzo, B. C. Dev,
D. W. Christianson, Arch. Biochem. Biophys., 2008, 469, 184.
8 J. E. Diaz, C.-S. Lin, K. Kunishiro, B. K. Feld, S. K. Avrantinis,
V. Dev, S. Frost and C. W. Ma, Biochem. Syst. Ecol., 2005,
33, 17.
J. Bronson, J. Greaves, J. G. Saven and G. A. Weiss, Protein 30 R. Baranauskiene, P. R. Venskutonis, P. Viskelis and
Sci., 2011, 20, 1597. E. Dambrauskiene, J. Agric. Food Chem., 2003, 51, 7751.
9 F. Lopez-Gallego, G. T. Wawrzyn and C. Schmidt-Dannert, 31 G. Gkinis, O. Tzakou, D. Iliopoulou and V. Roussis,
Appl. Environ. Microbiol., 2010, 76, 7723. Z. Naturforsch., C: Biosci., 2003, 58, 681.
10 Y. Yoshikuni, T. E. Ferrin and J. D. Keasling, Nature, 2006, 32 V. Roussis, M. Tsoukatou, P. V. Petrakis, I. Chinou,
440, 1078.
M. Skoula and J. B. Harborne, Biochem. Syst. Ecol., 2000,
28, 163.
11 J.-X. Li, X. Fang, Q. Zhao, J.-X. Ruan, C.-Q. Yang,
L.-J. Wang, D. J. Miller, J. A. Faraldos, R. K. Allemann, 33 J. P. Mevy, A. Bousquet-Melou, S. Greff, J. Millogo and
X.-Y. Chen and P. Zhang, Biochem. J., 2013, 451, 417.
C. Fernandez, Biochem. Syst. Ecol., 2006, 34, 815.
4020 | Org. Biomol. Chem., 2014, 12, 4013–4020
This journal is © The Royal Society of Chemistry 2014