Mutations in salt-bridging residues at the interface of the core and lid domains of epoxide hydrolase StEH1 affect regioselectivity, protein stability and hysteresis

Article abstract of DOI:10.1016/j.abb.2010.01.007

Epoxide hydrolase, StEH1, shows hysteretic behavior in the catalyzed hydrolysis of trans-2-methylstyrene oxide (2-MeSO)1Abbreviations used: SO, styrene oxide; 2-MeSO, trans-2-methylstyrene oxide.¹. Linkage between protein structure dynamics and catalytic function was probed in mutant enzymes in which surface-located salt-bridging residues were substituted. Salt-bridges at the interface of the α/β-hydrolase fold core and lid domains, as well as between residues in the lid domain, between Lys¹⁷⁹-Asp²⁰², Glu²¹⁵-Arg⁴¹ and Arg²³⁶-Glu¹⁶⁵ were disrupted by mutations, K179Q, E215Q, R236K and R236Q. All mutants displayed enzyme activity with styrene oxide (SO) and 2-MeSO when assayed at 30 °C. Disruption of salt-bridges altered the rates for isomerization between distinct Michaelis complexes, with (1R,2R)-2-MeSO as substrate, presumably as a result of increased dynamics of involved protein segments. Another indication of increased flexibility was a lowered thermostability in all mutants. We propose that the alterations to regioselectivity in these mutants derive from an increased mobility in protein segments otherwise stabilized by salt bridging interactions.

Full text of DOI:10.1016/j.abb.2010.01.007

Archives of Biochemistry and Biophysics 495 (2010) 165–173
Contents lists available at ScienceDirect
Archives of Biochemistry and Biophysics
journal homepage: www.elsevier.com/locate/yabbi
Original paper
Mutations in salt-bridging residues at the interface of the core and lid domains
of epoxide hydrolase StEH1 affect regioselectivity, protein stability and hysteresis
*
Diana Lindberg, Shabbir Ahmad, Mikael Widersten
Department of Biochemistry and Organic Chemistry, Uppsala University, Box 576, SE-751 23 Uppsala, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
Epoxide hydrolase, StEH1, shows hysteretic behavior in the catalyzed hydrolysis of trans-2-methylsty-
rene oxide (2-MeSO)1. Linkage between protein structure dynamics and catalytic function was probed
in mutant enzymes in which surface-located salt-bridging residues were substituted. Salt-bridges at
Received 24 November 2009
and in revised form 18 December 2009
Available online 15 January 2010
the interface of the
a/b-hydrolase fold core and lid domains, as well as between residues in the lid
domain, between Lys179-Asp202, Glu215-Arg41 and Arg236-Glu165 were disrupted by mutations, K179Q,
E215Q, R236K and R236Q. All mutants displayed enzyme activity with styrene oxide (SO) and 2-MeSO
when assayed at 30 °C. Disruption of salt-bridges altered the rates for isomerization between distinct
Michaelis complexes, with (1R,2R)-2-MeSO as substrate, presumably as a result of increased dynamics
of involved protein segments. Another indication of increased flexibility was a lowered thermostability
in all mutants. We propose that the alterations to regioselectivity in these mutants derive from an
increased mobility in protein segments otherwise stabilized by salt bridging interactions.
Ó 2010 Elsevier Inc. All rights reserved.
Keywords:
Epoxide hydrolase
Salt-bridges
Hysteresis
Mutagenesis
Substrate specificity
Introduction
The mechanism invokes that the substrate-free enzyme exists
in two different states, E and E0, with distinct substrate-binding
Structure dynamics is an inherent component of enzyme catal-
ysis and is a hallmark of allosteric enzymes [1–5]. We have previ-
ously reported that a plant epoxide hydrolase, StEH1 from Solanum
tuberosum, exhibits hysteretic kinetics [6,7] during the pre-steady
state phase of the catalyzed hydrolysis of trans-2-methylstyrene
oxide (2-MeSO) (Fig. 1) implying that this enzyme is present in dif-
ferent conformations in the substrate-free state [8].
properties. In addition, conformational transitions between the dif-
ferent Michaelis complexes (ES and E0S) are included. The model
mechanism in Scheme 1 explains the slow transitions detectable
in the pre-steady state kinetics and also describes how two differ-
ent diol products, as observed with (1R,2R)-2-MeSO as substrate,
may be formed (Fig. 1). Similar kinetic mechanisms have been pro-
posed for other a/b-hydrolase fold enzymes: butylcholine esterase
The tertiary structure of StEH1 consists of a canonical
a
/b-
[18] and acetylcholine esterase [19].
hydrolase fold core domain roofed by a lid domain [9,10] with
the active-site situated in a cleft between these two domains and
catalytic groups being contributed from both domains (Fig. 2A)
[11,12]. The catalytic mechanism involves a covalent enzyme-
bound intermediate formed after nucleophilic attack by an en-
zyme-contributed carboxylate on one of the oxirane carbons so
forming a covalent ester intermediate. The reaction is concluded
by base-catalyzed hydrolysis of the alkylenzyme to the vicinal diol
product [13–17].
The structural basis for the difference between E and E0 is un-
known and crystal structures of wild-type and mutant enzyme
forms have not provided information of possible structural flexibil-
ity; the available structures are virtually identical within the
experimental error [11,12]. It is possible, however, that the struc-
tural differences may escape detection in the X-ray crystallo-
graphic experiments either from being too small for detection at
the achieved resolution or that one conformer predominates at
equilibrium.
The hysteretic behavior of StEH1 is manifested in the transient
state reaction kinetics as a decrease in the observed rate of alkylen-
zyme formation at increasing substrate concentrations [8], and can
be modeled by the kinetic mechanism in Scheme 1.
Interactions at the interface between the core and lid do-
mains are expected to influence the fine structure and dynamics
of the active-site. From the available 3-D structures, a number of
salt-bridges can be identified, either connecting the different do-
mains, or present as intra-domain interactions (Fig. 2). The ami-
no acid residues participating in the different interaction
networks are relatively well conserved in related plant proteins
which may further suggest potential roles for these interactions
(Fig. 3).
* Corresponding author. Fax: +46 (0) 18 55 8431.
E-mail address: mikael.widersten@biorg.uu.se (M. Widersten).
Abbreviations used: SO, styrene oxide; 2-MeSO, trans-2-methylstyrene oxide.
1
0003-9861/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.abb.2010.01.007

166
D. Lindberg et al. / Archives of Biochemistry and Biophysics 495 (2010) 165–173
asy.ch/tools/blast/). Hits representing bona fide or inferred epoxide
hydrolases from different species were picked and aligned to the
StEH1 sequence with ClustalW [20].
Protein expression and purification
Mutated cDNAs were subcloned into pGTacStEH1-5H and puri-
fied by Ni(II)-IMAC and size exclusion chromatography as de-
scribed previously [17]. Enzyme concentrations were determined
from UV absorbance spectra of homogeneous proteins in buffer.
Steady state kinetics
Kinetic parameters were determined from the initial rates of en-
zyme-catalyzed hydrolysis of both enantiomers of styrene oxide
(SO) and trans-2-methylstyrene oxide (2-MeSO) in the presence of
wild-type or mutant StEH1. Initial rates were measured in 0.1 M so-
dium phosphate, pH 7.5, at 30 °C. Substrates were dissolved in ace-
tonitrile and diluted in buffer so that the final concentration of
acetonitrile in the reaction mixtures became 1% (v/v) and substrate
concentrations varied in a range of 0.02–0.50 mM. Enzyme concen-
trations were 20 nM in the reactions with (1S)-SO and (1S,2S)-MeSO
and 0.20–0.35
lM in the (1R)-SO and (1R,2R)-MeSO reactions.
Epoxide hydrolysis was followed by recording the decrease in
= ꢀ2.75 and ꢀ4.3 mMꢀ1 cmꢀ1 for SO
Fig. 1. (A) Epoxide hydrolase substrates studied. (B) Outcome of hydrolysis of
(1R,2R)-2-MeSO resulting from nucleophilic attack at either the benzylic (C-1) or
the aliphatic carbon (C-2).
absorbance at 225 nm,
D
e
and 2-MeSO, respectively. Kinetic parameters kcat, KM and kcat/KM
were determined after fitting the Michaelis–Menten equation using
MMFIT and RFFIT in SIMFIT (http://www.simfit.man.ac.uk) to the
experimental data. Due to the practically feasible upper substrate
concentration limit, the values of kcat and KM were in some sub-
strate/enzyme combinations determined after extrapolation to sat-
urating conditions, and hence, less well determined. The kcat/KM
parameters were in all cases well determined.
The object of the current work was to analyze the contribution
of salt-bridges, located at the enzyme domain interface and within
the lid domain, to protein stabilization and, as an extension, their
influence on structure dynamics reflected in the enzyme kinetics.
Site-directed mutagenesis targeted to destabilize salt-bridge and
hydrogen bonding interactions was performed and the mutant en-
zymes were analyzed for their catalytic and kinetic properties and
for thermal stability.
Pre-steady state kinetics
In order to determine microscopic rate constants the catalyzed
hydrolysis of the 2-MeSO enantiomers was followed during the
pre-steady state phase. Formation of the alkylenzyme intermediate
was detected as a concomitant decrease in intrinsic tryptophan
fluorescence of the enzyme [21]. Measurements were performed
in a sequential stopped-flow spectrophotometer (Applied Photo-
physics SX.20MV) with an excitation wavelength of 290 nm and
detecting fluorescent light collected through a 320 nm cut-off filter.
Reactions were performed at 30 °C in 0.1 M sodium phosphate, pH
7.5. Substrate concentrations ranged between 25–1200, or 85–
Materials and methods
Chemicals, reagents and bacterial strains
Epoxide substrates were purchased from Sigma–Aldrich at
highest purity available. Enantiomeric purity was analyzed by chi-
ral HPLC. Oligonucleotides were custom-made by Thermo Scien-
tific, restriction and DNA modifying enzymes supplied by
Fermentas and Promega Corp. Escherichia coli XL1-Blue was sup-
plied by Stratagene. Chromatographic resins were purchased from
GE Healthcare.
1200
lM, with (1S,2S)- or (1R,2R)-2-MeSO, respectively. Enzyme
concentrations were 2 or 10
lM with the (1S,2S)- or (1R,2R)-2-
MeSO respectively. Depending on the level of complexity of the cat-
alyzed reactions, apparent rate constants (kobs) were determined by
fitting single (Eq. (1)) or double (Eq. (2)) exponential functions with
floating endpoints to averaged (8–12 traces) progression curves. F is
the fluorescence signal, A, the initial fluorescence value at t = 0, kobs
is the apparent rate constants and C, the end point. F-tests were per-
formed to validate the use of the higher order equation.
Enzyme mutagenesis
Site-directed mutagenesis was performed by polymerase chain
reactions using mutagenic primers and plasmid pGTacStEH1-5H,
encoding wild-type StEH1 [17], as template. Sequences of muta-
genic oligonucleotides and deduced codon replacements are given
in Table 1. Mutated cDNA fragments were subcloned between the
MunI and SacI (K179Q, R236K and R236Q) or the XhoI and SacI
(E215Q) restriction sites of pGTacStEH1-5H. All mutated cDNA in-
serts were sequenced in full to confirm expected mutations and ex-
clude other alterations.
F ðtÞ ¼ A exp ðꢀkobstÞ þ C
ð1Þ
F ðtÞ ¼ A1 exp ðꢀkobs1tÞ þ A2 exp ðꢀkobs2tÞ þ C
ð2Þ
When the apparent rates displayed a hyperbolic dependence on
substrate concentration, indicative of a mechanism as depicted in
Scheme 2, Eq. (3) was fitted to the data. Applying this equation as-
sumes rapid pre-equilibrium of association and dissociation of ES;
[ES] = [S]/(KS + [S]). Curve-fitting allows for determination of rate
Sequence alignment
The primary structure of StEH1 was used as query in a BLASTP
search of the plant sequence database at Expasy (http://www.exp-
constants for alkylenzyme formation, k2, and decomposition, k
and k3, and the ES dissociation constant, KS.
ꢀ2

D. Lindberg et al. / Archives of Biochemistry and Biophysics 495 (2010) 165–173
167
Fig. 2. Salt-bridges at the interface of the core and lid domains of StEH1. (A) Overview of the StEH1 structure with the canonical
a/b-hydrolase folded core domain shown in
green. The mainly alpha-helical lid domain is colored in yellow. Residues in purple stick representation, two Tyr residues and an Asp, contribute catalytic groups and are
included to mark the active-site location. Residues in yellow or green stick representation are involved in charged polar interactions. (B) Arg236 is located at the junction
between the core and lid domains of StEH1 and adjacent to one of the catalytic Tyr residues (Tyr235). This residue forms a salt-bridge with Glu165 and hydrogen bonds with
main- and side-chain groups of nearby residues. (C) Glu215 and Arg41 interact through an inter-domain salt-bridge and via a number of hydrogen bonds with side- and main-
chain groups of nearby residues from both domains of the protein. (D) Salt-bridge involving Asp202 and Lys179. The Lys residue also interacts with the backbone carbonyl of
Ile200. Image created using the atomic coordinates in 2cjp [11] with PyMOL (DeLano Scientific; www.pymol.org).
When the apparent rates displayed a negative dependence on
substrate concentration, a situation described in Scheme 1 involv-
ing rate-limiting hysteretic steps, Eq. (4) was fitted to the data
which is a rewritten form of Eq. (5), the rate law for the kinetic
mechanism in Scheme 1 [18]. This fit does not allow for determina-
tion of individual rate constants but provides the sum of the rates
of interconversion of the ES¢E0S (k5 + kꢀ5) as determined by the
end-point rate at saturating [S].
Scheme 1.
k2½Sꢁ
kobs
kobs
kobs
¼
þ ðkꢀ2 þ k3Þ
ð3Þ
ð4Þ
ð5Þ
KS þ ½Sꢁ
Thermostability of enzyme activity
Â
Ã
2
KSK0 ðk0 þkꢀ0Þþ½Sꢁ KSðK0 þkꢀ5ÞþK0 ðKꢀ0 þk5Þ þ½Sꢁ ðk5 þkꢀ5Þ
S
S
To investigate thermostability of mutants and wild-type StEH1,
enzyme activity was measured after incubation for different time
periods at three different temperatures: 30, 45 and 55 °C. The
activity was measured spectrophotometrically at 225 nm, by
observing the decrease in absorbance due to hydrolysis of epoxides
¼
2
KSK0 þ½SꢁðKS þK0 Þþ½Sꢁ
S
S
k0 þ ðk5=KSÞ½Sꢁ
1 þ ð½Sꢁ=KSÞ
k
ꢀ0 þ ðkꢀ5=K0 Þ½Sꢁ
1 þ ð½Sꢁ=K0SSÞ
¼
þ

168
D. Lindberg et al. / Archives of Biochemistry and Biophysics 495 (2010) 165–173
Fig. 3. Sequence alignment of plant epoxide hydrolases. Residues in black are involved in salt-bridge or hydrogen bond interactions at the lid and core domain interface of
StEH1. s, salt-bridge interaction; b, hydrogen bond through backbone peptide group; c, hydrogen bond involving side-chain functional group. The catalytic residues Tyr235 and
His300 are shown in white on black. Black letters on gray indicates positions of possible compensatory mutations preserving the ability to form ionic interactions. Q76E11,
Citrus jambhiri [41], Q39856, Glycine max [42], B3VMR3, Nicotiana benthamiana [accession EU779658], A2Q320, Medicago truncatula [accession AC154391], Q0DJJ0, Oryza
sativa subsp. japonica, [43], Q8H289, Ananas comosus [44], Q84ZZ3, Euphorbia lagascae [45], Q9M9W5, Arabidopsis thaliana [accession AC011620], Q42566, Arabidopsis thaliana
[46], B4FF27, Zea mays [accession BT035715], Q8SAY7, Oryza sativa [accession AC096687], Q8LPE6, Cicer arietinum [accession AJ487037], A5C6E7, Vitis vinifera [47].
Regioselectivity
Table 1
Oligonucleotide sequences of mutagenic primersa,b
.
To investigate the regioselectivity of the StEH1 variants, hydro-
lysis reactions with both enantiomers of 2-MeSO were performed
(in triplicates) at 30 °C with continuous agitation for 24 h in the
presence of acetonitrile (3% (v/v)), substrate (7.32 mM) and en-
Name
Sequence (50–30)
Codon change
AAA to CAA
K179Q
TTG CTC CAA TTG GTG CTA AGT CTG TTC TTA
AGC AAA TAT TGA CAT ACC GCG ATC
CTT GTT GGC ATA GTA ATC CAA CTG TTC CTC
AGA AAG CCA CGA TG
E215Q-1
GAG to CAG
zyme (2 lM). The reaction volume was adjusted to 500 ll by addi-
E215Q-2
R236KQ
GTT GGA TTA CTA TGC CAA CAA G
–
tion of 0.1 M sodium phosphate, pH 7.5 buffer. After completed
hydrolysis, the reaction mixture was evaporated and the diol prod-
ucts were dissolved in the solvent used as mobile phase in the sub-
sequent HPLC separation. Samples were loaded onto a straight
phase system and separated over a chiral column (Chiralpak, AS-
H, 0.46 cm Ø ꢂ 25 cm) with a mobile phase consisting of hexane/
isopropanol (93:7). The column was coupled to a Shimadzu Prom-
inence LC-20AD pump and peaks were detected at 220 nm using a
Prominence SPD-M20A diode array detector. Molar ratios of diols
were determined from the peak areas of the separated reaction
products.
TAA CTT GAG CTC CTG TCC ATG GTG CTG TGA
GTT CCC AGT TTA TGG GTA AAG CCT KGTAAT
AGT TAA CTG CAC CAG T, K = T/G
CGT to
AAG/CAG
a
Italicized letters: restriction endonuclase recognition sites.
Bold letters: nucleotide replacements.
b
Scheme 2.
Results
to diol, in 0.1 M sodium phosphate, pH 7.5, at 30 °C as described
above. (1S,2S)-2-MeSO at 0.30 mM was used as substrate for all
StEH1 variants. Enzyme was added to a final concentration of
20 nM. Apparent rate constants of inactivation (kinactive) were
determined from fitting an equation for single exponential decay
to the experimental data (Eq. 6). The half-lives were calculated
Residue selection, mutagenesis, protein expression and purification
Salt-bridging residues were identified from the X-ray crystal
structure of StEH1 (Fig. 2). Residues Lys179, Glu215 and Arg236, situ-
ated at the protein surface were selected for mutation and were re-
placed by glutamine (K179Q, E215Q and R236Q) or lysine (R236K)
residues. Residue replacements were chosen to preserve the polar
from t1/2 = ln0.5/ꢀkinactive
.
ActivityðtÞ ¼ A exp ðꢀkinactivetÞ
ð6Þ

D. Lindberg et al. / Archives of Biochemistry and Biophysics 495 (2010) 165–173
169
character of the protein surface and thereby lowering the risk of
Enzyme kinetics
adverse effects on protein solubility. The StEH1 mutants were ex-
pressed in E. coli XL1-Blue and purified to homogeneity by Ni(II)-
IMAC chromatography and size exclusion chromatography. Final
yields of purified proteins ranged from 8 to 10 mg per liter of batch
culture, comparable to purifications of the wild-type enzyme. The
purification yield of mutant R236Q, however, was lower (2–5 mg
per liter culture) indicating lower stability and/or solubility of this
variant. Purified proteins were stored at 4 °C without loss in en-
zyme activity over the time period of the measurements.
When catalytic efficiencies at saturating substrate concentra-
tion were determined, it was clear that the mutants in general dis-
played kcat values lower than the wild-type enzyme (Table 2). The
turnover numbers showed
a general trend of wild-type >
K179Q > E215Q > R236K > R236Q. Pre-steady state kinetics have
identified hydrolysis of the alkylenzyme intermediate to be rate-
limiting in the wild-type-catalyzed reaction with (1S,2S)-2-MeSO
[8]. This appears to be true also for the salt-bridge mutants as
SS-2-MeSO
SS-2-MeSO
the trend for k3
is similar to kcat
(Tables 2 and 3).
The catalytic efficiencies under non-saturating conditions, as
expressed in kcat/KM, did not follow the same pattern. With (1R)-
SO, K179Q and E215Q promoted hydrolysis 2 to 2.5-fold more effi-
ciently as compared to the wild-type enzyme (Table 2), while other
mutants performed on par with the wild-type. In the (1S)-SO reac-
tion, relatively small effects were observed from the mutations
with the exception of R236Q which displayed approximately
threefold lower activity as compared to the wild-type enzyme.
kcat/KM for hydrolysis of (1R,2R)-2-MeSO follows the pattern of kcat
with this substrate, indicating that chemical turnover is primarily
rate-limiting for the enzyme efficiency in this reaction. The largest
effect on enzyme efficiencies was observed with (1S,2S)-2-MeSO.
All mutants showed lower activities, with R236Q again as the most
affected mutant, displaying 7.5-fold lower catalytic efficiency as
compared to the wild-type enzyme.
Hysteresis
The hysteretic behavior in the kinetics of StEH1 have been pre-
viously established with both 2-MeSO enantiomers as substrates
[8]. In all mutants, except for R236Q, the same slow transitions
were detected in the pre-steady state phase of the reactions with
either 2-MeSO enantiomer (Fig. 4A and C). In the case of the
R236Q mutant the transient state kinetics with (1S,2S)-2-MeSO
as substrate could only be modeled according to Scheme 2 and
did not display an observed rate that decreased with increasing
substrate concentration (Fig. 4). The stopped-flow measurements
with (1R,2R)-2-MeSO as substrate could not be determined with
this mutant due to the low protein purification yields.
A 0.8
B 0.2
C 0.4
0.15
0
0.15
0
0.15
single
single
single
0
-0.15R236K/(R,R)-2-MeSO
-0.15
0.15
-0.15R236Q/(S,S)-2-MeSO
double
0
0.1
0.4
0.2
-0.15 R236K/(S,S)-2-MeSO
0
0
0
0
10
20
t (ms)
30
40
50
0
10
20
t (ms)
30
40
50
0
10
20
t (ms)
30
40
50
D 800
600
80
140
100
60
E
F
wild type
K179Q
E215Q
R236K
R236Q
wild type
K179Q
E215Q
R236K
wild type
K179Q
E215Q
R236K
60
40
20
400
200
20
0
200 400 600 800 1000 1200 1400
[(1S,2S)-2-MeSO]
0
200 400 600 800 1000 1200 1400
[(1S,2S)-2-MeSO]
0
200 400 600 800 1000 1200 1400
[(1R,2R)-2-MeSO]
Fig. 4. (A) Typical example of an averaged experimental trace (grey dots) of fluorescence quenching indicative of alkylenzyme formation during the pre-steady state of the
reaction with 500 M (1S,2S)-2-MeSO and 2 M R236K mutant. Lines represent fits of single (dashed line) or double (solid line) exponential functions. The double exponential
fit provides two observable apparent rates, kobs1 and kobs2. Similar traces were recorded for the wild-type, K179Q and the E215Q enzymes with this substrate. Residual plots of
the respective fits are shown as inset. (B) As in (A), but with 350 M (1R,2R)-2-MeSO as substrate. Solid line represents the fit of a single exponential function. (C) As in (A), but
with 2 M R236Q as catalyst. Solid line represents the fit of a single exponential function. (D–F) Substrate concentration-dependence of kobs. Hyperbolic concentration-
l
l
l
l
dependence was observed for the more rapid apparent rates obtained from the double exponential fit with (1S,2S)-2-MeSO as substrate and the wild-type, K179Q, E215Q and
R236K. The observed rates in the reactions catalyzed by R236Q were obtained from single exponential fits as exemplified in (C). Negative substrate concentration-dependence
was observed for the slower observed rates with (1S,2S)-2-MeSO (E) as well as for the single rate obtained with the (1R,2R)-enantiomer as substrate (F). See the Materials and
methods section for detailed descriptions of fitted equations. Determined parameter values are presented in Table 3.

170
D. Lindberg et al. / Archives of Biochemistry and Biophysics 495 (2010) 165–173
In reactions following Scheme 2 (both enantiomers of SO, and in
practice (1S,2S)-2-MeSO [8]) KM is described by Eq. (7). The value of
KM will decrease as formation and stabilization of ES and alkylen-
zyme is promoted, i.e. low KS and high k2 relative to k and k3.
ꢀ2
KS ðkꢀ2 þ k3Þ
KM
¼
ð7Þ
ðk2 þ kꢀ2 þ k3Þ
In all mutants, with (1R)-SO as substrate, KM was two to four-
fold lower than the wild-type enzyme value, suggesting that en-
zyme-intermediate complexes are stabilized to a higher degree in
these enzyme variants. This may be linked to the slightly lower val-
ues of kcat in the mutants indicating slower hydrolysis rates. With
(1S)-SO and (1R,2R)-2-MeSO no clear effects on KM were observed;
variations were less than twofold when comparing the different
enzyme forms. The case of (1R,2R)-2-MeSO should be considered
special since this reaction is best modeled by the mechanism in
Scheme 1. In this case the stabilization and decay of ES, E0S, E-alkyl1
and E-alkyl2 will all contribute to KM. Hence, compensatory effects
may be present precluding differences in stabilization of individual
enzyme-intermediate forms. The KM value for (1S,2S)-2-MeSO, was
relatively unchanged with the exception of the R236Q mutant,
which displayed a value more than fourfold higher than the
wild-type enzyme, mainly due to a lower alkylation rate (k2)
(Table 3).
Thermostability of enzyme activity
To probe the contribution of salt-bridges to protein structure
stability, the wild-type and mutated enzymes were subjected to
elevated temperatures followed by activity measurements. All
salt-bridge mutations affected thermal stability negatively to dif-
ferent degrees. The most stable mutants K179Q and E215Q dis-
played half-lives approximately 50-fold shorter than the wild-
type enzyme at 55 °C (Table 4). The Arg236 mutants were more se-
verely affected with R236Q being the most temperature sensitive.
Regioselectivity
The regioselectivity of (1R,2R)-2-MeSO hydrolysis is not abso-
lute and the StEH1-catalyzed hydrolysis produces both possible
diol products at a certain molar ratio [8]. When diol product ratios
were analyzed after complete hydrolysis of (1S,2S)- or (1R,2R)-2-
MeSO differences were observed between the wild-type and mu-
tant enzymes (Table 5). With (1S,2S)-2-MeSO the nucleophilic at-
tack occurs almost exclusively at carbon-1. This property appears
unaltered by the salt-bridge mutations. The regioselectivity in
hydrolysis of (1R,2R)-2-MeSO, however, was altered to different
degrees in the mutants. The relatively low discrimination between
the two epoxide carbons shown by the wild-type enzyme was re-
tained in K179Q. E215Q was totally promiscuous while both Arg236
mutants displayed substantially higher discrimination favoring at-
tack at carbon-1 and thereby increasing the ee value from 15 to
>40% of the (1S,2R)-diol (Table 5).
Discussion
Due to the chiral nature of epoxides stereospecificity in en-
zymes acting upon them as substrates will influence catalysis. This
includes enantiomer discrimination, and with asymmetric epox-
ides, regiospecificity of epoxide ring opening. Most epoxide hydro-
lases that have been examined do exhibit stereoselectivity to
different degrees. Although detailed descriptions of the active-site
structures of several epoxide hydrolases are available from X-ray
crystallographic studies [11,12,22–27] it is yet non-trivial to draw
up general descriptions of structural determinants for substrate

D. Lindberg et al. / Archives of Biochemistry and Biophysics 495 (2010) 165–173
171
Table 3
Pre-steady state kinetic parameters in 0.1 M sodium phosphate, pH 7.5, 30 °C.
Enzyme
Substrate
(1S,2S)-2-MeSO
KS (mM)
(1R,2R)-2-MeSO
(k5 + kꢀ5) (sꢀ1
k2 (sꢀ1
)
k2/KS (sꢀ1 mMꢀ1
)
k
(sꢀ1
)
k3 (sꢀ1
110 10
69
44 10
83 30
78 20
)
(k5 + kꢀ5) (sꢀ1
)
)
ꢀ2
Wild-type
K179Q
E215Q
R236K
R236Q
0.47 0.1
0.75 0.2
0.84 0.4
0.16 0.1
0.23 0.2
370 20
650 50
370 70
160 50
130 20
770 200
870 200
440 200
1000 700
590 400
170 20
230 20
290 20
200 70
85 30
32
23
22
36
2
4
2
1
8.1
56
35
3
8
2
4
7
35
–
n.d.
Table 4
Time-dependent heat-inactivation of enzyme activity.
and enzyme–substrate complexes, resulting in hysteretic kinetics
[8]. The model adequately explains the enzyme behavior behind
the kinetic data obtained both during the pre-steady state and
the steady state phases of catalyzed reactions.
Enzyme
t1/2 (min)a
T (°C)
StEH1 displays relatively high enantiospecificity for the sub-
strate epoxides SO and 2-MeSO with E-values ranging from 40 to
150 depending on epoxide and pH (Table 2) [8,32]. The distribution
of diol products, however, is strongly dependent on the enantio-
meric form of epoxide substrate. With SO, StEH1-catalyzed hydro-
lysis results in enantioconvergence with the (1R)-phenylethanediol
being formed at high enantiomeric excess [32]. This is due to that
regiospecificity is almost totally inverted from attack at the achiral
carbon-2 with (1R)-SO to reaction at the chiral carbon-1 with the S-
enantiomer. Hydrolysis of 2-MeSO is more complex with very high
regiospecificity for attack at carbon-1 of the (1S,2S)-enantiomer
(ee > 98%), but with lower selectivity with the (1R,2R)-enantiomer
[8]. Furthermore, the steady state kinetics with the latter enantio-
mer displays cooperativity at lowered assay temperatures [D. Lind-
berg, M. De La Fuente, M. Widersten, unpublished], indicative of
multiple enzyme–substrate complexes which is another hallmark
of hysteresis in kinetics [7].
Although there is no evidence from X-ray crystal structures that
StEH1 exists in different metastable conformations, productive
binding of an asymmetric epoxide would demand a particular ac-
tive-site structure possessing adequate dynamic properties for
catalysis to proceed efficiently. The example of the 2-MeSO enan-
tiomers demonstrates this: hydrolysis of (1R,2R)-2-MeSO is par-
tially rate-limited by an observed rate that decreases with
increasing concentration of substrate, suggesting that structural
interconversion between different structural forms of the ES com-
plex exists.
30
45
55
Wild-type
K179Q
E215Q
R236K
R236Q
>500
>500
>500
>500
>500
>500
>500
38
9.0 0.5
53
1.4 0.3
5.0
1.1 0.4
0.31 0.04
5
1
2
460 100
a
Half-life of enzyme activity after incubation at the given temperatures, mea-
sured at 30 °C in the presence of 0.3 mM (1S,2S)-2-MeSQ in 0.1 M sodium phos-
phate, pH 7.5.
specificity, and to an even lesser extent, stereospecificity. Available
descriptions explaining the structure–function relationships be-
hind enantiospecificity in a given enzyme are essentially empirical
and has primarily been based on results from experiments in
which stereospecificity of epoxide hydrolases has been manipu-
lated by directed evolution [27–31]. If the mutagenesis regime
has involved random-site substitutions, amino acid residue altera-
tions augmenting enantiospecificity may in the evolved enzyme be
located both near or away from first-sphere substrate-interacting
residues [28,29,31]. On the other hand, restricting mutagenesis to
first-sphere active-site residues may also generate higher enantio-
specificity [30]. A recent report explains the basis for enhanced
enantiopreference for the S-enantiomer of phenylglycidyl ether
in an in vitro-evolved variant of the A. niger isoenzyme. In this case,
introduction of active-site residues sterically interfering with pro-
ductive binding of the R-enantiomer was responsible for substrate
discrimination [27].
Apart from structural restrictions on productive enzyme–sub-
strate interactions, the kinetics of enzyme structure dynamics
and of individual catalytic steps will be expressed in observed
enantio- and regiospecificities. We recently proposed a model for
the kinetic mechanism of epoxide hydrolase StEH1 (Scheme 1)
which takes into account promiscuous regioselectivity as well as
partially rate-limiting conformational changes in the free enzyme
Salt-bridges in folded proteins have been demonstrated to af-
fect protein stability and dynamics [33–36]. Observed effects from
replacing salt-bridging residues have been explained as due to loss
in direct binding energies in the ionic interactions, changes in free
energies of desolvation and changes in overall polar interactions in
the folded proteins [37]. In StEH1, a number of such interactions
are found (Fig. 2), in some cases located distantly from the catalytic
site. The overall effects observed by disturbing these interactions
Table 5
Regioselectivity in hydrolysis of trans-2-methylstyrene oxide in 50 mM sodium phosphate, pH 7.5, 30 °C.
Enzyme
Enantiomeric excess of diol products
Substrate
(1S,2S)-2-MeSO
(1R,2R)-2-MeSO
*
b
*
b
DDG (kJ/mol)
ee (%)a
DD
G
(kJ/mol)
Major:minor product
ee (%)a
Major:minor product
Wild-type
K179Q
E215Q
R236K
R236Q
98.2 0.006
99.8 0.05
99.5 0.2
99.7 0.007
99.7 0.02
ꢀ11.8
ꢀ17.4
ꢀ14.6
ꢀ16.4
ꢀ16.4
(1R,2S):(1S,2R)
(1R,2S):(1S,2R)
(1R,2S):(1S,2R)
(1R,2S):(1S,2R)
(1R,2S):(1S,2R)
15
20
1.8
42
44
4
2
2
3
6
ꢀ0.7
ꢀ1.0
(1S,2R):(1R,2S)
(1S,2R):(1R,2S)
(1S,2R):(1R,2S)
(1S,2R):(1R,2S)
(1S,2R):(1R,2S)
ꢀ0.09
ꢀ2.3
ꢀ2.4
a
(Cmajor diol ꢀ Cminor diol)/(Cmajor diol + Cminor diol) ꢂ 100.
DDGà = ꢀRTln [major diol]/[minor diol] [40].
b

172
D. Lindberg et al. / Archives of Biochemistry and Biophysics 495 (2010) 165–173
through conservative replacements can be summarized as affect-
ing thermostability, hysteretic behavior and regioselectivity.
(1) Arg236-Glu165. This salt-bridge, also involving other residues
through hydrogen bonding (Fig. 2B) associates a loop (the ‘‘cap-
loop” [38]) connecting the core domain, through Arg236, with the
lid domain, the ‘‘NC-loop” [38]. Arg236 is in the linear amino acid
sequence adjacent to one of the catalytic tyrosine residues
(Tyr235) which suggests a possible stabilizing role of the catalytic
site through this salt-bridge. The interaction is highly conserved
in other plant enzymes (Fig. 3) further indicating an important
structural role.
for protons generated during catalytic turnover [39]. This salt-
bridge interaction appears to be conserved in plant enzymes
(Fig. 3). Replacing the carboxylate with an amide in E215Q was ex-
pected to break the ionic interaction with Arg41 and alter the
hydrogen bonding network formed with surrounding groups.
The catalytic efficiency of E215Q was almost threefold higher
than the wild-type due to a lower KM value for (1R)-SO. The lower
KM is not likely caused by stronger interactions between enzyme
and substrate since the mutated residue is located 18 Å away from
the active-site, at the protein surface. Since kcat is virtually un-
changed, as compared to the wild-type (Table 2), the KM-effect is
expected to primarily be caused by an increase in the ratio of k2/
The anticipated effect of replacing Arg236 with Lys was to de-
crease the valency in hydrogen bonding with Val160, Phe158 and
Asn233, and to weaken the ionic interaction with Glu165. The result-
ing mutant is catalytically active with both SO and 2-MeSO, albeit
with lower efficiencies as compared to the wild-type enzyme (Ta-
ble 2). The largest effect by the replacement is observed in the sta-
bility at elevated temperatures (Table 4). If the destabilization of
the replacement is due to losses in enthalpic bonding energy or
unfavorable solvation effects is unknown. Although the efficiencies
of catalysis of the different reactions were quite similar to the
wild-type enzyme, clear effects could be observed in the pre-stea-
dy state phase. With (1R,2R)-2-MeSO, the sum of the rates for the
interconversion between ES and E0S (Scheme 1) shows a higher va-
lue demonstrating a more rapid interconversion between these en-
zyme–substrate complexes (Table 3, Fig. 4F). If the change is in
mainly one direction, or in a compensatory manner, is unresolved.
The mutation has also affected regiospecificity in this reaction con-
siderably. The preference for nucleophilic attack at carbon-1 is
higher in this mutant as compared to the wild-type enzyme. The
estimated change in free-energy differences is subtle, however,
and corresponds to ꢀ1.6 kJ/mol (Table 5). This demonstrates that
small alterations to protein dynamics affecting the flux through
the different pathways of Scheme 1, directly influence the outcome
of product ratios. In the case of (1S,2S)-2-MeSO, the attack of the
enzyme nucleophile is in all cases almost exclusively at carbon-1
and proceeds through a route E + S ? ES ? E-alkyl1 ? E + diol1
(Schemes 1 and 2), detected as an observed rate of alkylenzyme
formation increasing with substrate concentration (Eq. 3). In addi-
tion, a route E ? E0 ? E0S ? ES ? E-alkyl1 ? E + diol1 has to be in-
voked to explain the slower observed rate which decreases with
increasing substrate concentration (Fig. 4E). The enantiomeric ex-
cess of the major (1R,2S)-diol is unaffected (Table 5).
k
leading to accumulation of alkylenzyme. With the exception
ꢀ2
of a slightly lower KM(SS)-2-MeSO, caused by a lower alkylenzyme
hydrolysis rate, the E215Q mutant retains as a whole the kinetic
properties of the wild-type enzyme. A clear effect from the muta-
tion, however, is the loss in regioselectivity in the hydrolysis of
(1R,2R)-2-MeSO. The E215Q-catalyzed hydrolysis produced an
essentially racemic mixture of diol products (Table 5). The sum
of the rates of interconversions between ES and E0S was similar
to that of R236K which, in contrast to E215Q, displays an increase
in preference for attack at carbon-1 of the epoxide (Table 5). Hence,
if hysteresis is coupled to regioselectivity it is implied that the
changes in the individual values of k5 and k are not equal in
ꢀ5
the different mutants. The 10-fold shorter half-life at 55 °C reveals
that the affected interactions are indeed important to maintain a
functional form of the protein (Table 3).
(3) Lys179-Asp202. Lys179, situated in one of the lid helices, is in-
volved in a salt-bridge with Asp202 and hydrogen bonded to the
backbone carbonyl of Ile200 (Fig. 2D). Lys179 was replaced by a
Gln to remove ionic interactions, yet maintaining the polar nature
of the original residue. Since Lys179 and Asp202 are situated in the
same protein domain at the very surface of the protein, it was
not anticipated that disturbing this salt-bridge would influence
protein structure/function to a large degree. This salt-bridge is also
only moderately conserved as judged from the sequence alignment
(Fig. 3). The thermostability of the K179Q mutant, however, was
considerably shorter than that of the wild-type, with a half-life of
less than two minutes, as compared to 53 min, at 55 °C (Table 4).
The interactions between Lys179 and Asp202 and Ile200 connects
two different helical segments of the lid domain and disruptions
caused by the mutation to Gln are clearly important for maintain-
ing a functional structure at elevated temperatures.
The anticipated effect by the R236Q mutation was to fully re-
move any ionic interactions and thereby abolish this contribution
to the enzyme’s structural properties. The R236Q mutant resembles
functionally its R236K relative. The enzyme activity is decreased
with all substrates, although to a moderate degree (Table 2). The
main change is a drastically shortened half-life at elevated temper-
atures (Table 4), demonstrating a further destabilization of the en-
zyme structure. In the pre-steady state analysis of the (1S,2S)-2-
MeSO reaction the slower rate observed previously with the wild-
type enzyme [8] was absent, with only a single rate detectable
(Fig. 4C). This can be interpreted such that structural interconver-
sions in this mutant are more rapid and therefore preventing accu-
mulation of reaction intermediates to detectable levels.
The effects of the K179Q mutation on enzyme function during
steady state were minor. A slight increase in activity could be seen
with both SO enantiomers resulting from lower KM values and sim-
ilar activities with the 2-MeSO substrates, as compared to the wild-
type enzyme were also detected (Table 2). Effects on the micro-
scopic rate constants for the reaction with (1S,2S)-2-MeSO re-
vealed an elevated rate of alkylenzyme formation (k2) by almost
twofold to a value of 650 sꢀ1. Although the relative increase is
moderate, the result suggests that an increased mobility in the
lid domain favors epoxide ring opening and alkylenzyme forma-
tion. Since both catalytic Tyr residues which stabilizes the anionic
alcoholate leaving group are situated in the lid domain changes in
the dynamics of the protein structure are expected to influence this
reaction step. In addition, the sum of the interconversion rates be-
tween ES and E0S (k5 + kꢀ5) was also elevated in this mutant further
indicating an increased mobility in structural elements influencing
the dynamics in the enzyme–substrate complexes.
(2) Glu215-Arg41. Glu215 is situated at the N-terminal end of one
of the lid domain helices forming the cap-loop [38] at the interface
with the core domain. One of the carboxylate oxygens (OE2) forms
hydrogen bonds with the hydroxyl group and backbone nitrogen of
the lid Ser212 (Fig. 2C). The other oxygen (OE1) participates in a
salt-bridge with the guanidinium group of Arg41, which in turn is
hydrogen bonded to the side-chains of Tyr219 and Gln304. Glu215
is also the terminal residue in a hydrogen-bond chain leading from
the hydrolytic water molecule in the active-site to solvent. This
chain of interaction has been proposed to function as an exit route
Conclusions
Disruption of intra- and inter-domain salt-bridges located away
from the active-site in StEH1 cause relatively modest effects on the

D. Lindberg et al. / Archives of Biochemistry and Biophysics 495 (2010) 165–173
173
[18] P. Masson, L.M. Schopfer, M.-T. Froment, J.-C. Debouzy, F. Nachon, E. Gillon, O.
Lockridge, A. Hrabovska, B.N. Goldstein, Chem. Biol. Interact. 157–158 (2005)
143–152.
[19] A. Badiou, M.T. Froment, D. Fournier, P. Masson, L.P. Belzunces, Chem. Biol.
Interact. 175 (2008) 410–412.
[20] J.D. Thompson, D.G. Higgins, T.J. Gibson, Nucleic Acids Res. 22 (1994) 4673–
4680.
[21] L.T. Elfström, M. Widersten, Biochemistry 45 (2006) 205–212.
[22] M. Nardini, I.S. Ridder, H.J. Rozeboom, K.H. Kalk, R. Rink, D.B. Janssen, B.W.
Dijkstra, J. Biol. Chem. 274 (1999) 14579–14586.
[23] M.A. Argiriadi, C. Morisseau, B.D. Hammock, Proc. Natl. Acad. Sci. USA 96
(1999) 10637–10642.
[24] J. Zou, B.M. Hallberg, T. Bergfors, F. Oesch, M. Arand, S.L. Mowbray, T.A. Jones,
Structure 8 (2000) 111–122.
[25] G.A. Gomez, C. Morisseau, B.D. Hammock, D.W. Christianson, Biochemistry 43
(2004) 4716–4723.
[26] G.A. Gomez, C. Morisseau, B.D. Hammock, D.W. Christianson, Protein Sci. 15
(2006) 58–64.
overall enzyme function at moderate temperature. However, the
mutations result in clear alterations to enzyme function as re-
flected in the kinetic parameters determined under the pre-steady
state reaction phase and in the resulting regioselectivity of epoxide
ring opening. The observed effects shall probably be attributed to
altered dynamic properties of the enzyme structure. One indication
of effects on dynamics is the severely lowered thermostability of
these enzyme variants. In addition, the sum of microscopic rates
of isomerizations between Michaelis complexes are in all cases in-
creased in the mutants in the catalyzed reaction with (1R,2R)-2-
MeSO, a substrate that undergoes epoxide ring opening at either
carbon, further suggesting less rigid enzyme–substrate complexes.
Higher rates in this isomerization step in the reaction scheme
should have a direct consequence on the flux through the different
pathways leading to different product diol enantiomers. This is in-
deed the observable results. All mutants, although K179Q to a les-
ser extent, display altered regioselectivity in the hydrolysis of this
epoxide.
[27] M.T. Reetz, M. Bocola, L.W. Wang, J. Sanchis, A. Cronin, M. Arand, J. Zou, A.
Archelas, A.L. Bottalla, A. Naworyta, S.L. Mowbray, J. Am. Chem. Soc. 131
(2009) 7334–7343.
[28] B. van Loo, Chem. Biol. 11 (2004) 981–990.
[29] M.T. Reetz, C. Torre, A. Eipper, R. Lohmer, M. Hermes, B. Brunner, A. Maichele,
M. Bocola, M. Arand, A. Cronin, Y. Genzel, A. Archelas, R. Furstoss, Org. Lett. 6
(2004) 177–180.
[30] M.T. Reetz, L.-W. Wang, M. Bocola, M. Angew, Chemie Int. Ed. 45 (2006) 1236–
1241.
[31] B. van Loo, J. Kingma, G. Heyman, A. Wittenaar, J.H. Lutje Spelberg, T. Sonke,
D.B. Janssen, Enzyme Microb. Technol.44 (2009) 145-153.
[32] M.I. Monterde, M. Lombard, A. Archelas, A. Cronin, M. Arand, R. Furstoss,
Tetrahedron Asym. 15 (2004) 2801–2805.
We propose that the observed alterations to regioselectivity in
these mutants derive from an increased mobility in protein seg-
ments otherwise stabilized by salt bridging interactions.
Acknowledgments
[33] J.J. Tanner, R.M. Hecht, K.L. Krause, Biochemistry 35 (1996) 2597–2609.
[34] Y.-Y. Cheung, S.Y. Lam, W.-K. Chu, M.D. Allen, M. Bycroft, K.-B. Wong,
Biochemistry 44 (2005) 4601–4611.
[35] G. Valentini, L. Chiarelli, R. Fortin, M.L. Speranza, A. Galizzi, A. Mattevi, J. Biol.
Chem. 275 (2000) 18145–18152.
This work was supported by the Swedish Research Council
[Grant number 2008-3579] and The Carl Trygger Foundation
[Grant number CTS 07:380].
[36] R.L. Baldwin, J. Mol. Biol. 371 (2007) 283–301.
References
[37] S. Kumar, R. Nussinov, ChemBioChem 3 (2002) 604–617.
[38] S. Barth, M. Fischer, R.D. Schmid, J. Pleiss, Proteins Struct. Funct. Bioinf. 55
(2004) 846–855.
[39] A. Thomaeus, J. Carlsson, J. Åqvist, M. Widersten, Biochemistry 46 (2007)
2466–2479.
[40] R.E. Gawley, J. Org. Chem. 71 (2006) 2411–2416.
[41] K. Gomi, H. Yamamoto, K. Akimitsu, Plant Mol. Biol. 53 (2003) 189–199.
[42] M. Arahira, V.H. Nong, K. Udaka, C. Fukazawa, Eur. J. Biochem. 267 (2000)
2649–2657.
[43] International Rice Genome Sequencing Project, Nature 436 (2005) 793–
800.
[1] J. Monod, J.-P. Changeux, F. Jacob, J. Mol. Biol. 6 (1963) 306–329.
[2] G.G. Hammes, Biochemistry 41 (2002) 8221–8228.
[3] V.L. Schramm, Arch. Biochem. Biophys. 433 (2005) 13–26.
[4] P.K. Agarwal, J. Am. Chem. Soc. 127 (2005) 15248–15256.
[5] I. Bahar, C. Chennubhotla, D. Tobi, Curr. Opin. Struct. Biol. 17 (2007) 633–640.
[6] C. Frieden, Ann. Rev. Biochem. 48 (1979) 471–489.
[7] K.E. Neet, R.G. Ainslie Jr., Methods Enzymol. 64 (1980) 192–226.
[8] D. Lindberg, A. Gogoll, M. Widersten, FEBS J. 275 (2008) 6309–6320.
[9] D.L. Ollis, E. Cheah, M. Cygler, B. Dijkstra, F. Frolow, S.M. Franken, M. Harel, S.J.
Remington, I. Silman, J. Schrag, J.L. Sussman, K.H.G. Verschueren, A. Goldman,
Protein Eng. 5 (1992) 197–211.
[44] L.W. Neuteboom, W.Y. Kunimitsu, D.A. Christopher, Plant Sci. 163 (2002)
1021–1035.
[45] J. Edqvist, I. Farbos, Planta 216 (2003) 403–412.
[46] T. Kiyosue, J.K. Beetham, F. Pinot, B.D. Hammock, K. Yamaguchi-Shinozaki, K.
Shinozaki, Plant J. 6 (1994) 259–269.
[10] M. Arand, A. Cronin, F. Oesch, S.L. Mowbray, T.A. Jones, Drug. Metab. Rev. 35
(2003) 365–383.
[11] S.L. Mowbray, L.T. Elfström, K.M. Ahlgren, C.E. Andersson, M. Widersten,
Protein Sci. 15 (2006) 1628–1637.
[12] A. Thomaeus, A. Naworyta, S.L. Mowbray, M. Widersten, Protein Sci. 17 (2008)
1275–1284.
[13] H.-F. Tzeng, L.T. Laughlin, S. Lin, R.N. Armstrong, J. Am. Chem. Soc. 118 (1996)
9436–9437.
[14] H.-F. Tzeng, L.T. Laughlin, R.N. Armstrong, Biochemistry 37 (1998) 2905–2911.
[15] R. Rink, J. Kingma, J.H. Lutje Spelberg, D.B. Janssen, Biochemistry 39 (2000)
5600–5613.
[16] R.N. Armstrong, C.S. Cassidy, Drug. Metab. Rev. 32 (2000) 327–338.
[17] L.T. Elfström, M. Widersten, Biochem. J. 390 (2005) 633–640.
[47] R. Velasco, A. Zharkikh, M. Troggio, D.A. Cartwright, A. Cestaro, D. Pruss, M.
Pindo, L.M. FitzGerald, S. Vezzulli, J. Reid, G. Malacarne, D. Iliev, G. Coppola, B.
Wardell, D. Micheletti, T. Macalma, M. Facci, J.T. Mitchell, M. Perazzolli, G.
Eldredge, P. Gatto, R. Oyzerski, M. Moretto, N. Gutin, M. Stefanini, Y. Chen, C.
Segala, C. Davenport, L. Dematte, A. Mraz, J. Battilana, K. Stormo, F. Costa, Q.
Tao, A. Si-Ammour, T. Harkins, A. Lackey, C. Perbost, B. Taillon, A. Stella, V.
Solovyev, J.A. Fawcett, L. Sterck, K. Vandepoele, S.M. Grando, S. Toppo, C.
Moser, J. Lanchbury, R. Bogden, M. Skolnick, V. Sgaramella, S.K. Bhatnagar, P.
Fontana, A. Gutin, Y. Van de Peer, F. Salamini, R. Viola, PLoS ONE 2 (2007)
e1326.