2. oxygen to the aromatic ring (16, 19, 20) are shown in Fig. 1.
The reactive species, likely a di-iron(III) intermediate, attacks
the pi-electron system of the aromatic ring, forming epoxide 1
or delocalized carbocation 2. The opening of the epoxide ring
eventually provides the delocalized carbocation. The migration
of a hydride from the sp3
-hybridized carbon to the adjacent
atom then converts the carbocation to the more stable ketone
3. Finally, the dissociation of the ketone and its tautomeriza-
tion yield phenolic product 4.
The different regioselectivities shown by BMMs have been
attributed previously to differences in the shape of the active
site pocket, an idea supported by the fact that several point
mutations in the active sites of T4MO, toluene p-monooxygen-
ase, ToMO, MMO, and B. cepacia G4 T2MO cause large
variations in the regioselectivities of these BMMs (2, 7, 26, 30,
34, 35). In a previous paper (4), we have reported that muta-
tions at position 103 of the A subunit change the regioselec-
tivity of ToMO from Pseudomonas sp. strain OX1. However, a
complete study of the active site pocket of ToMO A is still
lacking.
In this paper, we report the effects of the substitution of six
residues in the ToMO A active site on substrate specificity and
regioselectivity. Furthermore, we present a detailed analysis of
the molecular determinants of regioselectivity based on the
docking of substrates and hypothetical intermediates of the
aromatic hydroxylation reaction into the active site of the crys-
tallographic structure of ToMO (Protein Data Bank [PDB]
code 1T0Q [32]), followed by a Monte Carlo optimization. The
results show that (i) the fine-tuning of TMO regioselectivity
can be achieved through a careful alteration of the shape of the
active site pocket and that (ii) the effects of mutations on
regioselectivity can be quantitatively predicted using the pro-
cedure described herein.
MATERIALS AND METHODS
Materials. Bacterial cultures, plasmid purifications, and transformations were
performed according to the procedures of Sambrook et al. (31). Escherichia coli
strains JM109 and CJ236 and vector pET22b(ϩ) were from Novagen. Plasmid
pBZ1260 (1) used for the expression of the ToMO cluster was kindly supplied by
P. Barbieri (Dipartimento di Biologia Strutturale e Funzionale, Universita`
dell’Insubria, Varese, Italy). E. coli strain JM101 was purchased from Boehr-
inger. The pGEM-3Z expression vector, Wizard SV gel, and the PCR clean-up
system for the elution of DNA fragments from agarose gel were obtained from
Promega. Enzymes and other reagents for DNA manipulation were from New
England Biolabs. The oligonucleotides were synthesized at MWG-Biotech
(Ebersberg, Germany). All other chemicals were from Sigma. The expression
and purification of recombinant catechol 2,3-dioxygenase from Pseudomonas sp.
strain OX1 are described elsewhere (36).
ToMO A mutagenesis. Plasmids for the expression of ToMO complexes with
mutated ToMO A subunits were prepared by site-directed mutagenesis of plas-
mid pTOU as described previously (4). Sequences of the mutagenic oligonucle-
otides are reported in Table S1 in the supplemental material.
Determination of apparent kinetic parameters and identification of products.
Assays were performed as described previously (4, 5) using E. coli JM109 cells
transformed with plasmid pBZ1260 or plasmid pTOU, which expresses wild-type
ToMO or ToMO mutant enzymes, respectively.
All kinetic parameters were determined using whole cells (4, 5). Enzymatic
activity on phenol was measured by monitoring the production of catechol in
continuous coupled assays with recombinant catechol 2,3-dioxygenase from
Pseudomonas sp. strain OX1 (5). The determination of apparent kinetic param-
eters for benzene, toluene, o-xylene, and naphthalene was carried out by a
discontinuous assay (5). For the calculation of the kcat values, amounts of pro-
teins were calculated as described previously (5). All the ToMO mutant enzymes
showed expression levels similar to that of the wild-type enzyme.
Reaction products were identified as described previously (5). All the regio-
specificity studies were performed using substrate concentrations higher than the
Km values. Under these conditions, absolute yields of products were proportional
to kcat values.
Modeling of substrates and intermediates into the active site of ToMO A.
Substrates and reaction intermediates were docked into the active site of ToMO
A by using the Monte Carlo energy minimization strategy. The ZMM-MVM
molecular modeling package (ZMM Software Inc. [http://www.zmmsoft.com])
was used for all calculations. This software allows conformational searches using
generalized coordinates such as torsion and bond angles instead of conventional
Cartesian coordinates (40).
Atom-atom interactions were evaluated using assisted model building with
energy refinement force fields (37) with a cutoff distance of 8 Å. Conformational
energy calculations included van der Waals, electrostatic, H bond, and torsion
components. A hydration component was not included. Electrostatic interactions
were assessed with a relative dielectric constant of 4.
Substrate and intermediate structures were prepared using the PyMOL soft-
ware (DeLano Scientific LLC). Geometry was optimized using the Zl module of
ZMM. Partial charges were attributed using the complete neglect of differential
overlap method in the HyperChem software (HyperCube Inc. [http://www.hyper
.com]).
The X-ray structure of the ToMO A-thioglycolate complex (PDB code 1T0Q)
was used to build the models of the ToMO A-substrate and ToMO A-interme-
diate complexes. To reduce computational time, a double-shell model of the
enzyme was built. The inner shell included 28 ToMO A residues surrounding the
active site cavity. During energy calculation procedures, the side chain torsion
angles—but not the backbone torsion angles—of these residues were allowed to
vary. Due to the asymmetric shape of the cavity, which is flat with the di-iron
cluster on one end and evolutionarily nonconserved hydrophobic residues on the
opposite side, the residues of the inner shell were selected manually. They
included the ligands of the iron ions, all residues with at least one side chain atom
contributing to the hydrophobic part of the active site cavity, and all residues with
at least one side chain atom less than 5 Å from the previous residues.
The outer shell included 139 residues, which did not belong to the inner shell
FIG. 1. Possible intermediates in the aromatic hydroxylation reaction catalyzed by ToMO. Intermediates 1, 2, and 3 are an epoxide, a
carbocation, and a ketone, respectively. R1 and R2 are hydrogen atoms or methyl groups. The geometrical features of the di-iron(III)-(hydro)per-
oxide intermediate and the details of the O-O bond cleavage reaction are not known.
824 NOTOMISTA ET AL. APPL. ENVIRON. MICROBIOL.
3. and were located less than about 16 Å from the active site pocket. During energy
calculation procedures, both the backbone and the side chain torsion angles of
the outer shell residues were not allowed to vary.
To further restrict the conformational freedom of the iron cluster and of the
protein-ligand complexes, two flat-bottom parabolic penalty functions—the so-
called constraints—available in ZMM were used. These functions increase the
conformational energy of the system if it deviates from specified parameters. The
atom-atom distance constraint applies a force to the system when the distance
between two specified atoms deviates from a specified value or interval. This type
of constraint was used to fix the distances between the two iron ions and between
each iron ion and the surrounding atoms, including the bridging water molecule
and the terminal water molecule. Atom-atom distance constraints were also used
to fix the conformation of the ligands of the di-iron cluster. A force constant of
1,000 kcal/mol/Å was used. The atom-atomic coordinate constraint applies a
force to the system when an atom moves farther than a specified distance from
particular Cartesian coordinates. This constrain was used to prevent the move-
ment of the Oε1 atom of the Glu103 side chain more than 0.5 Å from the original
position observed in the crystallographic structure.
A ϩ2 charge was arbitrarily assigned to each iron atom, both to account for
electron density transfer from the ligands to iron ions and to avoid strong
electrostatic attractive and repulsive interactions with negative and positive
atoms, respectively, of substrates and intermediates. Similarly, both the bridg-
ing and terminal solvent molecules observed in the 1T0Q structure were mod-
eled as neutral water molecules rather than OHϪ
ions. van der Waals radii of
iron ions and oxygen atoms of water molecules were arbitrarily set to 0.8 and 1.5
Å, respectively, in order to reduce steric hindrance inside the di-iron cluster.
Complexes with total energies of up to 8 kcal/mol higher than that of the
lowest-energy complex were stored for the analysis of energy contributions. Total
energy was partitioned into intrareceptor, intraligand, and receptor-ligand ener-
gies and energies of the constraints. Receptor-ligand energy was further parti-
tioned into van der Waals, electrostatic, and H bond components. Moreover,
receptor-ligand energy was also partitioned (i) per active site residue in order to
evaluate the contribution of each residue to the binding of ligands and (ii) per
ligand atom in order to evaluate the contributions of the ring and methyl sub-
stituents. Intrareceptor energies gave an estimation of the energy costs for
receptor (ToMO A) conformational changes upon ligand binding.
The docking procedure is described in detail in the supplemental material. The
PDB files for the initial manually generated complexes and the ZMM instruction
files containing the lists of mobile residues, constraints, and parameters used
during calculations are available upon request.
RESULTS AND DISCUSSION
Kinetic model for ToMO regioselectivity. The results of sev-
eral studies of BMMs suggest that the regioselectivities of
these enzymes depend on the shape of the active site cavity,
which is believed to influence the orientations of substrates
and reaction intermediates during catalysis (4, 7, 9, 19, 26, 34).
The ToMO active site pocket is buried deeply inside the
ToMO A subunit and is in contact with the surface through a
long tunnel whose diameter is large enough to allow the en-
trance of substrates and the exit of products (32). The active
site cavity (Fig. 2A and B) shows a lens-like shape. The dis-
tance between two atoms on opposite sides of the edge is about
10.5 to 12.5 Å. Assuming a van der Waals radius of about 2 Å,
these distances correspond to an inner diameter of 6.5 to 8.5 Å.
FIG. 2. Active site pocket of ToMO A. Panels A and B show the active site of the crystal structure of ToMO A (PDB code 1T0Q). Only residues
contributing to define the edge of the cavity are shown. In panel B, the grid cuts the cavity in such a way as to provide the largest section. Carbon
atoms are shown in red (Glu134 and Glu231), green (Ala107), yellow (Met180), blue (Glu103), magenta (Phe176), cyan (Ile100), and orange
(Phe205). (C) Superimposition of the PDB code 1T0Q crystal structure (colored as in panels A and B) onto the structure of the complex
ToMO-CCI 2 for the reaction leading to phenol production from benzene (carbon atoms are shown in white). (D) Superimposition of the complex
ToMO-CCI 2 (carbon atoms are shown in white) onto the complex ToMO-ketone 3 (carbon atoms are shown in green) for the reaction leading
to phenol production from benzene. THG, thioglycolate. CCI indicates CCI 2 and KTI indicates ketone 3.
VOL. 75, 2009 REGIOSELECTIVITY OF PSEUDOMONAS SP. STRAIN OX1 ToMO 825
4. The distance between two atoms placed above and below the
plane of the cavity is about 8.5 Å; hence, the thickness of the
cavity is about 4.5 Å. Therefore, the cavity is slightly larger
than a benzene molecule, which has a thickness of about 4 Å
and a diameter of about 6.4 Å. The grid in Fig. 2B cuts the
cavity into two halves in such a way as to highlight the section
with the larger surface.
The edge of the pocket (Fig. 2A and B) is formed by residues
Ala107, Met180, Glu103, Phe176, Ile100, and Phe205 and two
iron ligands, Glu134, adjacent to Ala107, and Glu231, adjacent
to Phe205. It should be noted that in the case of Glu103, only
the methylene groups contribute to the pocket (4, 32). Thr201
and Phe196 form one of the faces of the cavity, whereas
Glu104 forms the opposite one.
In a previous paper (4), we explained the regiospecificity of
ToMO A for toluene by hypothesizing that there are at least
three different positions in the active site pocket which can
accommodate the methyl group of toluene. These three sub-
sites can orient the methyl group of the substrate such that its
ortho, meta, or para carbon is presented to the di-iron center.
Thus, it is the difference in the affinities of the subsites for the
binding of the methyl group which determines the relative
abundances of three different enzyme-substrate complexes,
which can account for the observed distribution of cresols
produced by ToMO. Using a manual docking procedure (4),
we mapped an ortho subsite located among Ala107, Met180,
and Glu103, a meta subsite among Glu103, Phe176, and Ile100,
and a para subsite located between Ile100 and Phe205.
Figure 3 shows a new, more complex kinetic model of
ToMO regioselectivity. According to model i, toluene would
bind to the active site in three different catalytically productive
orientations, thus giving rise to three different enzyme-toluene
(ET) complexes that lead to the production of o-, m-, and
p-cresol isomers (complexes ETo, ETm, and ETp, respectively)
through at least one enzyme-intermediate (EI) complex (EIo,
EIm, or EIp, corresponding to o-, m-, or p-cresol, respectively).
According to this model, the ET-EI conversion is the rate-
limiting step. As the interactions between toluene and the
active site cavity are limited to van der Waals interactions and
as the active site cavity is larger than the toluene molecule, it
is likely that the interconversion of the ET complexes is fast
with respect to their transformation to cresols. The new model
can be described by six equilibrium constants: Ko-m, Ko-p, and
Km-p for the conversions ETo ^ ETm, ETo ^ ETp, and ETm ^
ETp, respectively, and K‡
o, K‡
m, and K‡
p for the conversion of
each productive ET complex to the corresponding transition
state (ET‡
) complex. Each ET‡
complex can turn into the other
two activated complexes through the ET complexes. There-
fore, we can define three equilibrium constants for the conver-
sions ET‡
o ^ ET‡
m, ET‡
o ^ ET‡
p, and ET‡
m ^ ET‡
p, which
will be the products of the constants defined above:
K‡
o-m ϭ ͓ET‡
m͔/͓ET‡
o͔ ϭ Ko-mK‡
m/K‡
o ϭ exp͑ Ϫ ⌬G‡
o-m/RT͒
(1)
K‡
m-p ϭ ͓ET‡
p͔/͓ET‡
m͔ ϭ Km-pK‡
p/K‡
m ϭ exp͑ Ϫ ⌬G‡
m-p/RT͒
(2)
K‡
o-p ϭ ͓ET‡
p͔/͓ET‡
o͔ ϭ Ko-mKm-pK‡
p/K‡
o ϭ exp͑ Ϫ ⌬G‡
o-p/RT͒
(3)
where ⌬G‡
o-m, ⌬G‡
m-p, and ⌬G‡
o-p are the free-energy differ-
ences for the three conversions, R is the gas constant, and T is
the absolute temperature.
This new model also includes an undefined number of un-
productive ET (ETu) complexes. It should be noted that the
number and the stability of these unproductive complexes do
not influence regioselectivity but only the apparent kcat value,
as they decrease the relative abundances of the ETo, ETp, and
ETm complexes and of the corresponding ET‡
complexes at
equilibrium.
Each ET‡
provides the corresponding EI, which in turn
releases a cresol isomer. Each ET‡
-EI transformation should
proceed with the same rate, , which is given by the following
well-known relation: ϭ kT/h, where k and h are the Boltz-
mann and Planck constants, respectively, and T is the absolute
temperature. According to model i (Fig. 3), the relative abun-
dances of o-, m-, and p-cresol isomers produced by the enzyme
are determined by the relative abundances of the three tran-
sition state ET‡
complexes at equilibrium. Therefore, calculat-
ing the energy differences, ⌬G‡
o-m, ⌬G‡
m-p, and ⌬G‡
o-p, should
allow the prediction of the percentages of cresols formed.
Two components should contribute to these energy differ-
ences: (i) the covalent bond energy (bE‡
), which includes the
energy of the bonds among ligand atoms and between ligand
and protein atoms, for example, those between the oxygen
atom transferred to the substrate and each iron of the cluster
FIG. 3. Kinetic models for the hydroxylation reaction of toluene
and o-xylene. CCIs 5 to 12 are the CCIs deriving from toluene and
o-xylene shown in Fig. 4. oC, mC, and pC, o-, m-, and p-cresols.
826 NOTOMISTA ET AL. APPL. ENVIRON. MICROBIOL.
5. (Fig. 1), and (ii) the ligand-protein noncovalent bond energy
(nbE‡
). Assuming that bE‡
is scarcely influenced by the position
of the substituent, the ⌬G‡
values should depend essentially on
the nbE‡
contributions:
⌬G‡
o-m ϭ ͑nbE‡
m Ϫ nbE‡
o͒ ϩ ͑bE‡
m Ϫ bE‡
o͒ ϭ ͑nbE‡
m Ϫ nbE‡
o͒
ϭ ⌬nbE‡
o-m (4)
⌬G‡
m-p ϭ ͑nbE‡
p Ϫ nbE‡
m͒ ϩ ͑bE‡
p Ϫ bE‡
m͒ ϭ ͑nbE‡
p Ϫ nbE‡
m͒
ϭ ⌬nbE‡
m-p (5)
⌬G‡
o-p ϭ ͑nbE‡
p Ϫ nbE‡
o͒ ϩ ͑bE‡
p Ϫ bE‡
o͒ ϭ ͑nbE‡
p Ϫ nbE‡
o͒
ϭ ⌬nbE‡
o-p (6)
The kinetic model ii shown in Fig. 3 illustrates the case of
o-xylene. Two productive enzyme-substrate complexes (re-
ferred to hereinafter as EX complexes, where X represents the
substrate)—EX2,3 and EX3,4—and the corresponding EX‡
complexes yield 2,3-dimethylphenol (2,3-DMP) and 3,4-DMP,
respectively.
In the following sections, we discuss several pieces of evi-
dence which support these kinetic models.
Modeling toluene and o-xylene into the active site of ToMO
A. We have tried to dock substrates into the active site cavity of
ToMO A by the Monte Carlo method, as it allows effective
exploration of the conformational space with less central pro-
cessing unit time than other, more time-consuming methods
such as molecular dynamics. In all docking experiments, the
backbone of ToMO A and the structure of the di-iron cluster
were held rigid whereas at least two layers of side chains
around the active site cavity were allowed to move to improve
the fit of ligands inside the cavity. The Monte Carlo energy
minimization of ToMO A without ligands in the active site
cavity showed that only three residues contributing to the sur-
face of the cavity, i.e., Ile100, Thr201, and Phe205, were par-
ticularly mobile, being able to assume several conformations.
However, the mobility of Phe205 was limited to the 2
torsion
angle. In the second layer of residues, Leu208, Leu272, Gln204
and, to a lesser extent, His96 were able to adopt different
conformations.
When toluene and o-xylene were docked into the active site,
more than 10 low-energy orientations for each substrate were
found, giving rise to several different binding complexes (data
not shown). These results suggest that aromatic substrates can
assume several binding orientations inside the active site, in
agreement with the models in Fig. 3.
Modeling the intermediates of benzene into the active site of
ToMO A. As it is well-known that active sites are complemen-
tary to activated transition states and to unstable intermediates
rather than to substrates and products (some examples can be
found in references 11, 12, 24, 27, and 33 and references
therein), we tried to identify catalytically productive binding
modes through the docking of the intermediates of the hy-
droxylation reaction. As shown in Fig. 1, two or three inter-
mediates are supposed to be involved in the conversions of
aromatic hydrocarbons to phenols (16): (i) an aromatic carbo-
cation, (ii) an unsaturated ketone, and possibly (iii) an epoxide.
The carbocationic intermediate (CCI) has a critical role in the
regioselectivity of the reaction because, after its formation, the
nature of the product is irreversibly defined. In contrast,
the epoxide intermediate formed from toluene or from o-
xylene can yield two different isomers, depending on which
C-O bond of the epoxide ring undergoes cleavage.
We initially tested this procedure by docking the three pos-
sible intermediates of benzene hydroxylation into the active
site of wild-type ToMO. Benzene was chosen instead of tolu-
ene or o-xylene because in this case a single molecular species
exists for each intermediate. Docking was carried out by fixing
the oxygen atom transferred to the aromatic ring at the same
coordinates found for the bridging oxygen of the thioglycolate
anion in the crystal structure of ToMO A (PDB code 1T0Q).
The assumption that the oxygen atom is bound to the di-iron
cluster even after its transfer to the substrate (Fig. 1) limits the
degrees of freedom of the intermediates and provides a fixed
point which can be used as a rotation center for the ligand.
Binding energy values for the ToMO-benzene intermediate
complexes reported in Table 1 indicate that the CCI fits the
active site better than the other intermediates. The main con-
tribution to the tight binding of the CCI depends on electro-
static interactions (Table 1), but van der Waals contacts also
play an important role. In the CCI, the oxygen atom is bound
to a single carbon atom of the ring and the C-O bond forms an
angle of ϳ130° with the ring, which can thus be placed almost
exactly in the plane of the cavity as shown in Fig. 2C, thus
maximizing the steric interaction with the cavity. On the other
hand, in the molecule of the epoxide intermediate, the six-
atom ring and the epoxide ring form an angle of 105° and the
oxygen atom lies above the central point of the C-C bond of
the epoxide ring. As a consequence, when the oxygen atom is
located at the bridging position of the di-iron cluster, too-close
contacts between the six-atom ring and the active site cavity
and between the three-atom ring and the di-iron cluster are
generated (data not shown). As for the ketonic intermediate,
the oxygen atom is in the same plane as the carbon atom ring.
This geometry prevents the positioning of the ring in the plane
of the cavity but, interestingly, pushes it toward the tunnel (Fig.
2D) which connects the active site to the exterior of the mol-
ecule.
An interesting observation which stems from these docking
experiments is that the orientation of the carbocation (Fig. 2C)
is very similar to that of the thioglycolate found in the crystal
structure of ToMO (this feature is even more evident in the
case of the carbocations deriving from o-xylene, as discussed
below).
Modeling the intermediates of toluene and o-xylene into the
active site of ToMO A. Given the increased stability of the
TABLE 1. Interaction energy values for ToMO A-benzene
reaction intermediates
Ligand
Interaction energy (kcal/mol)
van der Waals Electrostatic Total
Epoxide Ϫ12.27 Ϫ6.42 Ϫ18.69
Carbocation Ϫ15.06 Ϫ19.44 Ϫ34.50
Ketonea
Ϫ13.41 Ϫ9.08 Ϫ22.49
Ketoneb
Ϫ13.35 Ϫ9.43 Ϫ22.78
a
sp3
carbon near Thr201.
b
sp3
carbon near Ala107.
VOL. 75, 2009 REGIOSELECTIVITY OF PSEUDOMONAS SP. STRAIN OX1 ToMO 827
6. ToMO-CCI complex relative to the ToMO-epoxide and
ToMO-ketone complexes, in the case of benzene we decided
to pay particular attention to the modeling of the ToMO-CCI
complexes corresponding to toluene and o-xylene. Docking the
CCIs of toluene and o-xylene is a complex procedure because
several isomers exist. As shown in Fig. 4, toluene may generate
up to five CCIs and four possible intermediates may be pro-
duced from reactions starting with o-xylene. Intermediate cou-
ples 5 and 6, 7 and 8, 10 and 11, and 12 and 13 are enantiomers.
CCIs 5 and 6 yield o-cresol, CCIs 7 and 8 yield m-cresol, CCIs
10 and 11 yield 2,3-DMP, and CCIs 12 and 13 yield 3,4-DMP.
CCI 9 yields p-cresol.
Our docking experiments indicate that only CCIs 5, 7, and 9
from toluene and CCIs 10 and 12 from o-xylene can bind to the
active site of wild-type ToMO with the same orientation as the
CCI derived from benzene (data not shown), i.e., with the ring
in the plane of the grids in Fig. 2B and C. As described in the
following sections, using only the binding energy values relative
to these five intermediates, we obtained very good agreement
between predicted and experimentally determined percentages;
therefore, we will discuss only the docking analyses of these
intermediates.
Figure 5A shows a model of the positioning of CCIs 10 and
12, which lead to 2,3-DMP and 3,4-DMP, respectively, into the
active site. The models of CCIs 5 and 7, leading to o- and
m-cresol, respectively, are completely superimposable onto the
model of CCI 10 (data not shown), whereas CCI 9, which yields
p-cresol, has an orientation similar to that of CCI 12 (data not
shown). Figure 5B shows that the ring of CCI 12 is placed
exactly in the plane of the grid in Fig. 2 and that it mimics the
orientation of thioglycolate even better than the CCI of ben-
zene shown in Fig. 2C. Thus, our data indicate that CCIs 5, 7,
9, 10, and 12 dock into the active site of ToMO A and place
their methyl groups into subsites located on the border of the
pocket, as hypothesized previously (4). However, the ToMO-
CCI complexes suggest rather different positioning of the sub-
sites for methyl groups from that in the original model. The
model in Fig. 5A shows that only the ortho and para subsites
are unambiguously defined. In this new model, the ortho sub-
site, located among residues Glu134, Leu192, and Ala107, is
closer to the di-iron cluster than that in our previous model
whereas the new para subsite is defined by residues Glu103,
Phe176, and Ile100 (the meta subsite in the original model).
Moreover, two alternative meta subsites (designated m1a and
m1b) can be mapped. The existence of two alternative meta
subsites may depend on the close proximity of the ortho and
para subsites. This geometry is incompatible with the simulta-
neous docking of three adjacent methyl groups. Therefore,
when CCI 10 is docked into the cavity, the ortho methyl groups
block the ortho subsite whereas the meta methyl group partially
fills the para subsite, thus defining the m1b subsite (residues
Glu103, Phe176, and Met180). On the other hand, when CCI
12 is docked into the cavity, the para methyl group fills the para
subsite, whereas the meta group partly occupies the ortho sub-
site, thus defining the m1a subsite (residues Ala107, Glu103,
and Met180). The van der Waals contributions of the methyl
groups of CCI 10 (subsites o and m1b) are Ϫ1.12 and Ϫ2.02
kcal/mol, respectively, whereas their contributions in the case
of CCI 12 (subsites m1a and p) are Ϫ1.64 and Ϫ1.65 kcal/mol,
respectively. Hence, it seems that a methyl group positioned
into the m1b subsite gives a greater contribution to the binding
energy than one positioned into the m1a subsite. An indirect
confirmation of this observation comes from the docking of
CCI 7, which leads to m-cresol. In this case, the single methyl
group is predicted to occupy subsite m1b.
In order to further analyze the subsites of the active site, we
docked the intermediate of the hydroxylation of m-xylene,
which yields 2,4-DMP, into the cavity (Fig. 5C). In this case,
both the ortho and para methyl groups were forced slightly out
from the ortho and para subsites. Because of the steric hin-
drance between the ortho methyl and residue Glu134 and be-
tween the para methyl and residue Ile100, the contributions of
the ortho and para methyl groups to the binding energy de-
creased to Ϫ0.05 and Ϫ0.63 kcal/mol, respectively. These two
values are significantly lower than those found for the inter-
mediates deriving from o-xylene (see above). Interestingly, the
total van der Waals contributions of CCIs 10 and 12 and of the
intermediate produced from m-xylene to the binding energy
FIG. 4. Chemical structures of the possible CCIs deriving from
toluene and o-xylene. CCIs 5 and 6, CCIs 7 and 8, and CCI 9 are the
possible intermediates for the transformation of toluene into o-, m-,
and p-cresols, respectively. CCIs 10 and 11 and CCIs 12 and 13 are the
possible intermediates for the transformation of o-xylene into 2,3- and
3,4-DMPs, respectively.
828 NOTOMISTA ET AL. APPL. ENVIRON. MICROBIOL.
7. were Ϫ15.54, Ϫ16.01, and Ϫ13.37 kcal/mol, respectively. This
finding would indicate that the cavity is better tailored to ac-
commodate the intermediates from the physiological substrate
o-xylene than those deriving from a nonphysiological substrate
such as m-xylene.
In conclusion, our docking data provide a very detailed
map of the active site residues potentially involved in regio-
selectivity.
From binding energies to percentages of products. In the
hypothesis that the CCI is the first intermediate formed during
the hydroxylation reaction, as the toluene3CCI reaction is
endergonic, the ET‡
transition states should be similar to en-
zyme-CCI complexes. Therefore, the ⌬nbE‡
o-m, ⌬nbE‡
m-p, and
⌬nbE‡
o-p values defined by equations 4 to 6 can be estimated
through the Monte Carlo docking of the CCIs. Using equa-
tions 1 to 6 and the binding energy values provided by the
ZMM software for toluene CCIs 5, 7, and 9, we predicted the
percentages of cresols. Predicted percentages of o-, m-, and
p-cresols (39.7, 19.9, and 40.4%, respectively) were very similar
to experimentally determined percentages (36, 19, and 45%,
respectively). Similarly, using the binding energy values for
o-xylene CCIs 10 and 12, we found that predicted percentages
of 2,3-, and 3,4-DMPs (17.9 and 82.1%, respectively) were very
similar to experimentally determined percentages (19 and
81%, respectively). Thus, it may be concluded that the hypoth-
eses of completely steric control of regioselectivity and the use
of the ToMO A-CCI noncovalent bond energies for calculating
the relative stabilities of the ET‡
or EX‡
complexes are essen-
tially correct.
As a control, the docking procedure was repeated using the
two possible epoxides deriving from toluene, i.e., toluene 2,3-
epoxide and toluene 3,4-epoxide, which provide o-cresol/m-
cresol and m-cresol/p-cresol, respectively. The binding energy
of toluene 3,4-epoxide was found to be about 2 kcal/mol higher
than that of toluene 2,3-epoxide. In a system at equilibrium,
this energy difference would imply the formation of less than
5% 2,3-epoxide. Even assuming that the 2,3-epoxide interme-
diate converts entirely to o-cresol, this finding is not in agree-
ment with the experimentally determined percentage of
o-cresol (36%). Similarly, in the case of the ketonic interme-
diates, the isomer leading to m-cresol showed the higher bind-
ing energy (data not shown), in disagreement with the exper-
imental data.
Modeling the CCIs of toluene and o-xylene in the active sites
of ToMO A mutant forms. To further test our hypothesis, all
the residues located on the edge of the cavity (Ala107, Met180,
Glu103, Phe176, Ile100, and Phe205) were selected for muta-
tional studies to experimentally verify whether changes at these
sites would affect the regioselectivity of the enzyme in a pre-
dictable way. These residues were all changed to hydrophobic
residues to preserve the hydrophobic nature of the pocket.
Ala107 was changed to larger residues, such as Val and Ile,
in order to hinder the ortho site. Met180 was changed to Ile in
an enzyme already carrying the mutation E103G in order to
obtain a double-mutant enzyme designated (E103G, M180I)-
ToMO A, in which all the residues facing the active site pocket
are identical to the corresponding residues present in homol-
ogous T4MO.
Glu103, previously mutated to Gly, Leu, and Met (4), was
changed to the -branched residues Val and Ile in order to
FIG. 5. Structures of the ToMO-CCI 10 and ToMO-CCI 12 com-
plexes. (A) Superimposition of the ToMO-CCI 10 complex (carbon
atoms are in white) onto the ToMO-CCI 12 complex (carbon atoms
are in green). (B) Superimposition of the PDB structure 1T0Q (carbon
atoms are in magenta) onto the structure of the ToMO-CCI 12 com-
plex (carbon atoms are in green). (C) Superimposition of the com-
plexes ToMO-CCI 10 and ToMO-CCI 12 (colored as in panel A) onto
the complex of ToMO with the CCI of the reaction m-xylene 3
2,4-DMP (carbon atoms are in magenta). The surface of the cavity of
the complex ToMO-CCI 10 is shown as a mesh. The mesh is colored
to show the contributions of residues Glu134, Ala107, Met180, Glu103,
Phe176, and Ile100.
VOL. 75, 2009 REGIOSELECTIVITY OF PSEUDOMONAS SP. STRAIN OX1 ToMO 829
8. increase hindrance in the region between the meta and para
sites. Phe176 was changed to Leu and Ile in order to enlarge
this region.
Ile100 contributes to defining the hypothetical para site, but
it is also at the boundary between the active site pocket and the
tunnel which connects the pocket to the surface of the protein.
Moreover, it is less closely packed than the other residues of
the active site and bulges from the surface of the pocket. The
corresponding residue in MMOs, Leu110, has been defined
previously as the gate that controls the access to the active site
(2, 29). Therefore, to explore the entire range of side chain
dimensions, Ile100 was changed to Ala, Val, Leu, Met, Phe,
and Trp.
Phe205, which is located at the boundary between the active
site pocket and the tunnel, like Ile100, was changed to Leu.
All mutant enzymes were assayed with phenol, benzene,
toluene, o-xylene, and naphthalene. The majority of the mu-
tations did not change the catalytic efficiency with respect to
that of wild-type ToMO or caused only minor changes. Only
mutations I100A, I100W, F205L, A107V, and A107I were
found to reduce significantly the kcat values for all the sub-
strates (Table 2). Possible explanations are discussed briefly in
the supplemental material.
The docking procedure for the prediction of the regioselec-
tivity was repeated using the Monte Carlo-optimized models of
ToMO A mutant enzymes. As in wild-type ToMO A, the
backbone and the structure of the di-iron cluster were held
rigid; i.e., we assumed that the mutations did not significantly
change the structure of the protein. Recently, Murray and
coworkers have published the crystallographic structure of
ToMO carrying the mutation I100W [the (I100W)-ToMO en-
zyme] (18). In spite of the large increase in the hindrance of
the side chain at position 100, the structure of the mutated
ToMO A is essentially unchanged. This effect happens because
the bulky tryptophan side chain is accommodated inside the
active site cavity partially hindering it. As all the residues we
have mutated are at the border of the active site cavity, it is
likely that variations in the steric hindrance of these side chains
can be easily accommodated without relevant changes to the
backbone structure, as observed in the case of the I100W
mutation. We want also to underline that the Monte Carlo-
optimized model of (I100W)-ToMO correctly predicted the
orientation of the mutated side chain (data not shown).
Tables 3 and 4 report the results obtained after the docking
of toluene- and o-xylene-derived CCIs, respectively, to ToMO
A mutant enzymes at positions 107, 103, 180, and 176. The
regioselectivities of all mutant enzymes, with the exception of
those of (E103L)-ToMO for toluene and (F176L)-ToMO for
o-xylene, were predicted with fairly good accuracy. Minor dif-
ferences between experimental and predicted percentages may
depend on the small differences in binding energy values (⌬G‡
)
TABLE 2. Apparent kcat values of ToMO and ToMO mutant
proteins on benzene, toluene, o-xylene, and naphthalene
ToMO variant
or mutation(s)
kcat (sϪ1
)a
for substrate:
Phenol Benzene Toluene o-Xylene Naphthalene
Wild type 1.0 0.37 0.43 0.26 0.033
I100A 0.019 0.05 0.03 0.025 0.0015
I100V 0.4 0.22 0.27 0.19 0.011
I100M 0.35 0.14 0.18 0.078
I100L 0.76 0.3 0.56 0.3
I100F 0.15 0.13 0.3 0.16
I100W 0 0.018 Very low Very low
F176I 0.25 0.43 0.28 0.21 0.016
F176L 0.32 0.36 0.32 0.23 0.016
A107V 0.07 0.064 0.024 0.032
A107I 0.027 0.0017 0.0009 0.0005
F205L 0.01 0.0015 0.00094 0.0006
E103G 1.0 0.43 0.42 0.42
E103L 0.89 0.29 0.32 0.36
E103 M 0.29 0.26 0.3 0.2
E103V 0.24 0.22 0.22 0.11
E103I 0.21 0.03 0.03 0.01
E103G, I100L 0.53 0.58 0.67 0.75
E103V, I100V 0.30 0.11 0.097 0.029
E103G, M180I 0.50 0.22 0.34 0.2
a
Standard errors for kcat values were Ͻ20% for phenol and Ͻ25% for ben-
zene, toluene, o-xylene, and naphthalene.
TABLE 3. Comparison between experimentally determined and calculated percentages of cresol isomers produced by wild-type ToMO and
ToMO mutant proteins
ToMO variant or
mutation(s)
Binding energya
(kcal/mol) for: Calculated % ofb
:
Experimentally
determined %c
of:
ET‡
o ET‡
m1b ET‡
m1a ET‡
p o-C m-C p-C o-C m-C p-C
Wild type Ϫ34.69 Ϫ34.28 Ϫ34.70 39.7 19.9 40.4 36 19 45
E103G Ϫ34.00 Ϫ33.91 Ϫ35.44 7.5 6.5 86.0 9 6 85
E103G, M180I Ϫ32.74 Ϫ33.85 Ϫ35.28 1.2 8.1 90.7 6 10 84
A107V Ϫ27.50 NAd
NA Ϫ34.04 0.0 2.9 97.1 0 6 94
A107I Ϫ27.30 NA NA Ϫ33.90 0.0 3.3 96.7 0 4 96
F176I Ϫ33.28 Ϫ33.31 Ϫ34.91 5.6 5.9 88.4 7 6.5 87.5
F176L Ϫ33.27 Ϫ33.40 Ϫ34.82 6.3 7.8 85.9 6 7 87
E103V Ϫ33.70 Ϫ33.35 Ϫ33.45 45.2 25.1 29.7 49 34 17
E103I Ϫ33.73 Ϫ33.03 Ϫ32.89 64.6 19.8 15.6 59 24 17
E103M Ϫ33.86 Ϫ33.29 Ϫ33.21 58.5 22.3 19.2 47 19 34
E103L Ϫ33.46 Ϫ31.61 Ϫ32.26 85.1 3.7 11.2 20 11 69
a
Total binding energy for the complex ToMO variant-CCI 5 (ET‡
o), ToMO variant-CCI 7 (ET‡
m1b and ET‡
m1a), or ToMO variant-CCI 9 (ET‡
p).
b
o-, m-, and p-C, o-, m-, and p-cresols.
c
Error, Ͻ1%.
d
NA, not applicable. The hindrance of the branched side chains of Val and Ile forces the meta methyl group to occupy a position intermediate between the m1b
and p subsites. The binding energies for the (A107V)-ToMO–CCI 7 and (A107I)-ToMO–CCI 7 complexes are Ϫ31.95 and Ϫ31.90 kcal/mol, respectively.
830 NOTOMISTA ET AL. APPL. ENVIRON. MICROBIOL.
9. among ET‡
o, ET‡
m, ET‡
p, EX‡
3,4, and EX‡
2,3 complexes. In
fact, our approach predicts ⌬G‡
values of 0.1 to 0.3 kcal/mol,
which are 1% or less of the calculated binding energy values
(30 to 36 kcal/mol).
Moreover, as the docking procedure provides the binding
energy values of each EI, it is now possible to understand also
how each mutation influences the stability of the interaction of
toluene and o-xylene methyl groups at each subsite. For exam-
ple, mutation E103G significantly reduces the production of
o-cresol, increasing the difference in stability between the ortho
and para orientations, which changes from 0.01 kcal/mol in the
case of the wild-type protein to 1.44 kcal/mol in the case of the
E103G mutant enzyme. As shown in Table 3, this increase is
due both to the destabilization of the ortho orientation (com-
plex ET‡
o) and to the stabilization of the para orientation
(complex ET‡
p). Moreover, the ZMM software yields individ-
ual group contributions to the total binding energy (see Ma-
terials and Methods). Our data (not shown) indicate that the
contribution of the methyl group in the ortho site to the binding
energy, due essentially to van der Waals interactions, is barely
influenced by the E3G mutation, changing from Ϫ1.34 kcal/
mol for the wild-type protein to Ϫ1.47 kcal/mol for the E103G
mutant form. Rather, the predicted destabilization of the ET‡
o
complex (0.69 kcal/mol) depends on the loss of van der Waals
contacts between the C-H groups at positions 3 and 4 of the
ring (Fig. 4 numbering) and the larger active site pocket. On
the other hand, the enlargement of the para subsite removes
clashes between the para methyl group and the pocket and
increases the contribution of the para methyl group to the
binding energy (from Ϫ0.76 to Ϫ2.1 kcal/mol). In the case of
o-xylene, the mutation E103G increases the relative stability of
the EX‡
3,4 complex, which provides 3,4-DMP, both increasing
the stability of the EX‡
3,4 complex and decreasing the stability
of the EX‡
2,3 complex (Table 4). The values for the individual
components of the calculated binding energy indicate that the
increase in the stability of the EX‡
3,4 complex is due mainly to
improved binding of the methyl groups and that the decrease
in the stability of the EX‡
2,3 complex depends on the worse
accommodation of the substrate ring.
Mutations F176I and F176L have effects similar to that of
mutation E103G on the regioselectivity of ToMO A, but these
effects depend on different factors. Indeed, these two muta-
tions leave the stability of the ET‡
p complex (Table 3) almost
unchanged with respect to that of the wild-type enzyme,
whereas they decrease the stability of ET‡
o and ET‡
m com-
plexes by 1.42 and 0.95 kcal/mol, respectively. Moreover, the
values for the individual components of the binding energy
indicate that the decrease in the stability of the ET‡
o complex
is due mainly to a loss of van der Waals contacts between the
active site pocket and the ring of the substrate (about 0.9
kcal/mol) and that the contribution of the methyl group to the
binding energy is not affected by the mutation (the change is
less than 0.1 kcal/mol). In contrast, in the case of the ET‡
p
complex, the loss of van der Waals contacts between the active
site pocket and the ring of the substrate (about 0.8 kcal/mol) is
counterbalanced by an increased contribution by the methyl
group of the substrate (about 1 kcal/mol).
As for mutations E103I and E103V, it should be noted that
they insert -branched residues at position 103. This insertion
generates clashes between the side chain and the ring of the
substrates, thus lowering the stability of all the ToMO-CCI
complexes with respect to that of the wild-type enzyme (Tables
3 and 4). The greater destabilization of the ET‡
p complex than
of ET‡
o and ET‡
m (Table 3) leads to increased production of
o- and m-cresols, whereas slightly different conformations of
the valine and isoleucine side chains (data not shown) make
the (E103V)-ToMO–ET‡
m1a complex more stable than the
(E103I)-ToMO–ET‡
m1b complex. These results give a molec-
TABLE 4. Comparison between experimentally determined and calculated percentages of DMP isomers produced by wild-type ToMO and
ToMO mutant proteins
ToMO variant
or mutation(s)
Binding energya
(kcal/mol) for: Calculated % of:
Experimentally
determined %b
of:
EX‡
2,3 EX‡
3,4 2,3-DMP 3,4-DMP 2,3-DMP 3,4-DMP
Wild type Ϫ35.27 Ϫ36.17 17.9 82.1 19 81
E103G Ϫ35.18 Ϫ36.39 11.5 88.5 1 99
E103G, M180I Ϫ34.1 Ϫ36.23 2.7 97.3 2 98
A107V Ϫ26.20 Ϫ32.65 0.0 100 0 100
A107I Ϫ25.60 Ϫ31.80 0.0 100 0 100
F176I Ϫ33.60 Ϫ35.25 5.8 94.2 2.5 97.5
F176L Ϫ34.8 Ϫ35.47 24.4 75.6 2.5 97.5
E103V Ϫ33.81 Ϫ34.15 36.0 64.0 39 61
E103I Ϫ31.67 Ϫ33.20 7.0 93.0 13 87
E103M Ϫ34.36 Ϫ35.02 24.7 75.3 18 82
E103L Ϫ32.75 Ϫ33.76 15.4 84.6 6 94
I100A Ϫ34.61 Ϫ35.85 11.0 89.0 18 82
I100V Ϫ35.33 Ϫ36.01 24.1 75.9 12 88
I100L Ϫ35.43 Ϫ36.16 22.6 77.4 12 82
I100M Ϫ35.48 Ϫ36.51 15.0 85.0 28 72
I100F Ϫ35.79 Ϫ36.84 14.5 85.5 19 81
I100L, E103G Ϫ35.09 Ϫ36.29 11.6 88.4 4 96
I100V, E103V Ϫ33.62 Ϫ34.17 28.3 71.7 34.5 65.5
a
Total binding energy for the complex ToMO variant-CCI 10 (EX‡
2,3) or ToMO variant-CCI 12 (EX‡
3,4).
b
Error, Ͻ1%.
VOL. 75, 2009 REGIOSELECTIVITY OF PSEUDOMONAS SP. STRAIN OX1 ToMO 831
10. ular reason for the enhanced production of m-cresol by mutant
(E103V)-ToMO with respect to that by mutant (E103I)-
ToMO. Moreover, due to the slightly different orientations of
the two side chains, mutation E103V decreases the stability of
the EX‡
3,4 complex much more than that of the EX‡
2,3 com-
plex and mutation E103I has the opposite effect (Table 4). As
a consequence, mutation E103V increases the percentage of
2,3-DMP whereas mutation E103I decreases it with respect to
that produced by wild-type ToMO. In both cases, the decrease
in the stability of the EX‡
2,3 and EX‡
3,4 complexes is due to
worse accommodation of both the substrate ring and the
methyl groups.
The Ala107 side chain, according to our model, contributes
to the surface of the ortho site. In agreement with the model,
the mutation of Ala107 to Val or Ile, whose side chains fill the
cavity of the ortho site, completely abolishes the production of
o-cresol and 2,3-DMP.
Modeling intermediates in the case of Ile100 mutant en-
zymes. Several mutations at position 100, which defines the
para subsite, surprisingly increase the production of m-cresol
(Table 5). Replacement with alanine, leucine, phenylalanine,
and valine increases the production of m-cresol, whereas re-
placement with methionine and tryptophan decreases it. It is
also interesting that the effects of mutations at positions 100
and 103 are additive, as shown by the behavior of the double-
mutant enzymes (I100L, E103G)-ToMO A and (I100V,
E103V)-ToMO A. The individual mutations I100V and E103V
increase the percentage of m-cresol from 19 to 47 and 34%,
respectively, whereas the corresponding double mutant pro-
duces 62% m-cresol. Furthermore, mutation I100L increases
the percentage of m-cresol from 19 to 38%, whereas mutation
E103G decreases it to 6%. In this case, the double-mutant
enzyme (I100L, E103G)-ToMO A produces 17% m-cresol. As
shown in Fig. 5A, a second hypothetical meta site (the m2 site)
may exist between Ile100 and Phe205 (the para subsite in our
previous model [4]) which is occupied by the methyl group
when CCI 8 (Fig. 4) is docked into the active site. However, the
binding of CCI 8 is different from the binding of the other
intermediates because the hindrance of the side chains of
Ile100 and Phe205 forces the ring of the intermediate into a
plane which forms an angle of about 25° with the plane that
contains intermediates 5, 7, 9, 10, and 12 (Fig. 6A). Thus, it
may be that this different geometry of the wild-type ToMO
A-CCI 8 complex—with the aromatic ring out of the plane of
the grid in Fig. 2B—makes the complex catalytically unproduc-
tive. In fact, this geometrical constraint is the reason that, in
the case of the wild-type protein, we have limited our docking
analysis to intermediates 5, 7, 9, 10, and 12 only (see above).
Several mutations of residue 100 remove part or all of the
hindrance and allow intermediate 8 to dock to the active site in
a conformation more similar to those of intermediates 5, 7, 9,
10, and 12 (Fig. 6A). This finding makes kinetic model i inad-
equate to predict the effects of mutations at position 100. To
take into account the extra methyl subsite m2, we modified
model i by introducing a fourth ET complex (ETm2) and the
corresponding transition state ET‡
m2 (model iii in Fig. 3).
Model iii includes three extra ⌬G‡
values, ⌬G‡
o-m2, ⌬G‡
m1-m2,
and ⌬G‡
p-m2, which determine the relative abundance of the
ET‡
m2 complex with respect to total ET‡
complexes and,
hence, the amount of m-cresol produced by the pathway ETm2
^ ET‡
m2 3 enzyme-CCI 8.
As the orientation of CCI 8 inside the active site is different
for each mutant protein and different from those of interme-
diates 5, 7, 9, 10, and 12, the hypothesis that the ⌬G‡
values
depend essentially on the noncovalent bond energy contribu-
tions is no longer valid. On the contrary, because of nonopti-
mal binding of the activated substrate to the cluster, the cova-
lent bond energy component of the ET‡
m2 complex, bE‡
m2,
should be generally higher than that of the other three ET‡
complexes, bE‡
:
⌬G‡
o-m2 ϭ ͓͑nbE‡
m2 Ϫ nbE‡
o͒ ϩ ͑bE‡
m2 Ϫ bE‡
͔͒ ϭ ͑⌬nbE‡
m2-o
ϩ ⌬bE‡
͒ (7)
⌬G‡
m1-m2 ϭ ͓͑nbE‡
m2 Ϫ nbE‡
m1͒ ϩ ͑bE‡
m2 Ϫ bE‡
͔͒ ϭ ͑⌬nbE‡
m2-m1
ϩ ⌬bE‡
͒ (8)
⌬G‡
p-m2 ϭ ͓͑nbE‡
m2 Ϫ nbE‡
p͒ ϩ ͑bE‡
m2 Ϫ bE‡
͔͒ ϭ ͑⌬nbE‡
m2-p
ϩ ⌬b
E‡
͒ (9)
where ⌬bE‡
ϭ (bE‡
m2 Ϫ bE‡
) Ն 0.
TABLE 5. Comparison between experimentally determined and calculated percentages of cresol isomers produced by wild-type ToMO and
ToMO variants mutated at position 100
ToMO variant
or mutation(s)
Binding energy (kcal/mol)a
for:
eϪ⌬bE‡/RT
Calculated % ofb
:
Experimentally
determined %c
of:
ET‡
o ET‡
m1b ET‡
m1a ET‡
p ET‡
m2 o-C m-C p-C o-C m-C p-C
Wild type Ϫ34.69 Ϫ34.28 Ϫ34.70 Ϫ34.50 0.00 40.0 20.0 40.0 36 19 45
I100A Ϫ33.90 Ϫ33.88 Ϫ35.07 Ϫ35.20 0.58 7.0 43.0 50.0 9 42 49
I100V Ϫ33.80 Ϫ33.86 Ϫ34.30 Ϫ34.75 0.38 16.0 47.0 37.0 13 47 40
I100L Ϫ34.66 Ϫ34.15 Ϫ34.35 Ϫ35.23 0.32 35.0 44.0 21.0 31 38 31
I100 M Ϫ34.48 Ϫ33.98 Ϫ35.04 Ϫ37.76 0.00 22.0 10.0 68.0 29 5 66
I100F Ϫ34.71 Ϫ34.52 Ϫ35.85 Ϫ36.02 0.90 5.0 52.0 42.0 37 35 28
I100W Ϫ34.55 Ϫ34.40 Ϫ36.52 Ϫ36.29 0.55 3.0 28.0 70.0 25 5 70
I100L, E103G Ϫ33.98 Ϫ33.81 Ϫ35.30 Ϫ34.08 1.00 8.0 16.0 76.0 6 17 77
I100V, E103V Ϫ35.50 Ϫ32.85 Ϫ33.46 Ϫ35.51 2.70 20.0 61.0 19.0 19.5 62 18.5
a
Total binding energy for the complex ToMO variant-CCI 5 (ET‡
o), ToMO variant-CCI 7 (ET‡
m1b and ET‡
m1a), ToMO variant-CCI 9 (ET‡
p), or ToMO variant-CCI
8 (ET‡
m2).
b
o-, m-, and p-C, o-, m-, and p-cresols.
c
Error, Ͻ1%.
832 NOTOMISTA ET AL. APPL. ENVIRON. MICROBIOL.
11. Therefore, the equilibrium constants which determine the
relative abundance of the ET‡
m2 complex can be expressed as
function of the ⌬bE‡
values:
K‡
o-m2 ϭ ͓ET‡
m2͔/͓ET‡
o͔ ϭ exp͑ Ϫ ⌬nbE‡
m2-o/RT͒
ϫ exp͑ Ϫ ⌬bE‡
/RT͒ (10)
K‡
m1-m2 ϭ ͓ET‡
m2͔/͓ET‡
m1͔ ϭ exp͑ Ϫ ⌬nbE‡
m2-m1/RT͒
ϫ exp͑ Ϫ ⌬bE‡
/RT͒ (11)
K‡
p-m2 ϭ ͓ET‡
m2͔/͓ET‡
p͔ ϭ exp͑ Ϫ ⌬nbE‡
m2-p/RT͒
ϫ exp͑ Ϫ ⌬bE‡
/RT͒ (12)
Using the Monte Carlo In equations 10 through 12, correct to
add symbol “ϫ” between “exp . . .” expressions [e.g.,
exp(Ϫ⌬nbE‡
m2-p/RT) ϫ exp(Ϫ⌬bE‡
/RT)]? If not, please clar-
ify relationship between these expressions; should space be-
tween them be deleted, or should a chem point or some other
symbol be added in place of the multiplication sign? strategy,
we predicted nbE‡
o, nbE‡
m1, nbE‡
p, and nbE‡
m2 for each muta-
tion at position 100. Then, using equations 1 to 12, we deter-
mined the exp(Ϫ⌬bE‡
/RT) values, which provided predicted
cresol percentages similar to the experimentally determined
ones. Table 5 shows that, except for mutations I100F and
I100W, for each mutation at position 100, a single exp(Ϫ⌬bE‡
/
RT) value exists which yields good agreement between exper-
imental and predicted percentages of products. Moreover, Fig.
6B shows that there is good correlation between these
exp(Ϫ⌬bE‡
/RT) values and the angle ␦ formed between the
plane in which intermediates 5, 7, 9, 10, and 12 lie (the plane
of the grid in Fig. 2) and the plane where intermediate 8 is
located (Fig. 6A). This interesting finding suggests that the
energy of the transition state increases with the angle ␦ and
further supports the hypothesis that the predicted Monte
Carlo-minimized complexes ToMO A-benzene-CCI and
ToMO A-CCI 5, 7, 9, 10 or 12 illustrate the geometry of the
catalytically productive orientation of aromatic substrates in-
side the ToMO A active site pocket. Moreover, we want also to
underline that our model provides a simple explanation for the
additive effects of mutations at positions 100 and 103. In fact,
according to our model, the observed percentage of m-cresol is
the sum of the percentages of m-cresol produced from toluene
docked with its methyl group in subsite m1 and from toluene
docked with its methyl group in subsite m2. Moreover, the
residue at position 103 prevalently contributes to the m1 site,
whereas the residue at position 100 contributes to the m2 site.
Thus, the combined effect of a mutation at position 103, which
improves interactions at subsite m1, and a mutation at position
100, which partially opens subsite m2, is the production of a
percentage of m-cresol close to the sum of the percentages
produced by the single mutations. This is the case for the
double mutant (I100V, E103V)-ToMO. In the case of the
double mutant (I100L, E103G)-ToMO, the effect of mutation
E103G, which decreases the ability of the m1 site to anchor a
methyl group, is counterbalanced by the effect of mutation
I100L on the catalytic efficiency of the m2 site. Consequently,
no apparent change in m-cresol production is observed.
As for mutations I100F and I100W, the lower level of agree-
ment between predicted and experimental data may be due to
the fact that mutations I100F and I100W increase significantly
the volume of the side chain at position 100. As described
above, the crystallographic structure of the (I100W)-ToMO
mutant enzyme (18) shows that the tryptophan side chain
points toward the active site cavity, partially hindering it. Most
likely, the simultaneous accommodation inside the active site
cavity of the bulky side chain of tryptophan—or phenylala-
nine—and of the substrate may require changes in the confor-
mation of the ToMO A backbone which cannot be predicted
by our strategy, as all the docking experiments were carried out
by holding the backbone of ToMO A in a rigid position and
allowing the movement of only the side chains closer to the
cavity.
As for o-xylene, our docking data indicate that there is no
mutation at position 100 which allows the positioning of inter-
mediates 11 and 13 in the active site with a conformation
similar to those of intermediates 10 and 12 (data not shown).
Thus, in the case of o-xylene, there is no need for hypothesizing
FIG. 6. Comparison between the orientation of CCI 10 docked into
wild-type ToMO (WT) and the orientations of CCI 8 docked into
wild-type and mutant ToMOs. (A) The surface of the cavity of the
ToMO-CCI 10 complex is shown as a mesh. ␦ is the angle between the
ring of CCI 10 docked into wild-type ToMO and CCI 8 docked into
wild-type and mutant ToMOs. (B) Correlation between angle ␦ shown
in panel A and exp(Ϫ⌬bE‡
/RT) values. Data for the mutation I100V
were not used in the linear fit.
VOL. 75, 2009 REGIOSELECTIVITY OF PSEUDOMONAS SP. STRAIN OX1 ToMO 833
12. a kinetic model which takes into account the involvement of an
m2 subsite. Consequently, we used kinetic model ii and calcu-
lated the data reported in the lower part of Table 4. Also in this
case, the predicted regioselectivities of the mutant enzymes
were in good agreement with the experimental values.
Regioselectivity on naphthalene. From a steric point of view,
naphthalene can be described as an ortho disubstituted ben-
zene derivative bearing groups larger than methyl groups. Four
CCIs can be produced from naphthalene, similar to those
shown in Fig. 4 for o-xylene. Our docking data indicate that
only intermediates equivalent to CCIs 10, 12, and 13 in Fig. 4
can fit the active site (in Table 6, these complexes are named
EN‡
␣, EN‡
1, and EN‡
2, respectively, and their models are
shown in Fig. S5 in the supplemental material). However, none
of these three intermediates can assume exactly the same ori-
entation as the intermediates derived from benzene, toluene,
and o-xylene. This finding is in agreement with the observation
that the kcat value of wild-type ToMO for naphthalene is con-
siderably lower than those measured for the more physiologi-
cal substrates. Data in Table 6 show that in the case of naph-
thalene, our predictions are only qualitatively correct. In fact,
for all the mutant enzymes we have studied, the calculated
␣-naphthol/-naphthol ratio was higher than that observed
experimentally. However, the model correctly predicts that
mutations which reduce the volume of the residues at positions
100 and 176 (Ile and Phe, respectively) increase the percentage
of -naphthol. Ile and Phe side chains likely impair the correct
positioning of the naphthalene reaction intermediates corre-
sponding to CCIs 12 and 13 (i.e., the intermediates which form
the complexes EN‡
1 and EN‡
2, respectively).
The poor quantitative agreement of our predictions with
experimental data may depend on the rigidity of the backbone
and of the major part of the side chains in the docking proce-
dure. It may be possible that greater flexibility of the backbone
is needed to accommodate this large substrate. Moreover, it
should also be remembered that the models shown in Fig. 3 are
based on the hypothesis that EX complexes are generated
through fast equilibrium events. From a molecular point of
view, this means that substrates can easily change their orien-
tation inside the active site. Most likely, the hypothesis holds
true for monocyclic, small substrates, but it could not hold in
the case of a bulky molecule like naphthalene (see Fig. S5 in
the supplemental material). If naphthalene, entering the active
site, generates an EN complex (as the docking of naphthalene
CCIs suggests [see Fig. S5C in the supplemental material]) and
the rate of conversion of this initial complex to EN␣ is not
higher than the rate of the hydroxylation reaction, more
-naphthol than the amount predicted by a model based on
fast equilibrium events will be formed.
Regioselectivity on polar substrates: the case of phenol. An
intriguing feature of several multicomponent monooxygenases
is their specificity in the second hydroxylation step, which pro-
duces exclusively catechol from phenol and (di)methyl-
catechols from cresols and DMPs. It should also be remem-
bered that Tao et al. and Vardar and Wood (34, 35), using
random mutagenesis, were able to obtain several ToMO and
T4MO mutant forms which produce hydroquinone in different
amounts. Mutant (I100Q)-ToMO is particularly interesting, as
it produces 80% hydroquinone and only 20% catechol, the
physiological product of ToMO. Even if a detailed analysis is
beyond the scope of this paper, we have tested our approach
with this mutant enzyme and with wild-type ToMO.
We have docked the two possible enantiomeric CCIs of the
phenol-to-catechol hydroxylation reaction (CCIs 14 and 15 in
Fig. 7) into the ToMO active site on the hypothesis that, like
those for methyl groups, subsites for the hydroxyl group should
exist.
Our results indicate that the hydroxyl groups can be posi-
FIG. 7. Chemical structures of the CCIs deriving from phenol.
CCIs 14 and 15 are the possible intermediates of the phenol-catechol
reaction. CCI 16 is the intermediate of the phenol-hydroquinone re-
action. The positive charge is delocalized on both the ring and the OH
group.
TABLE 6. Comparison between experimentally determined and calculated percentages of naphthol isomers produced by wild-type ToMO
and ToMO mutant proteins
ToMO
variant or
mutation
Binding energya
(kcal/mol) for: Calculated %b
of:
Experimentally
determined %c
of:
EN‡
␣ EN‡
1 EN‡
2 ␣-N
-N produced
by EN‡
1
-N produced
by EN‡
2
␣-N -N
Wild type Ϫ38.66 Ϫ31.08 Ϫ35.49 99.5 0.0 0.5 87 13
F176I Ϫ36.32 Ϫ32.88 Ϫ34.10 97.4 0.3 2.3 57 43
F176L Ϫ37.29 Ϫ28.65 Ϫ34.17 99.5 0.0 0.5 57 43
I100A Ϫ38.58 Ϫ33.37 Ϫ36.10 98.5 0.02 1.5 53 47
I100V Ϫ38.02 Ϫ30.30 Ϫ35.91 97.3 0.0 2.7 81 19
a
Total binding energy for the complex ToMO variant-naphthalene analogue of CCI 10 (EN‡
␣), ToMO variant-naphthalene analogue of CCI 12 (EN‡
1), or ToMO
variant-naphthalene analogue of CCI 13 (EN‡
2).
b
␣- and -N, ␣- and -naphthol.
c
Error, Ͻ1%.
834 NOTOMISTA ET AL. APPL. ENVIRON. MICROBIOL.
13. tioned in two hydrophilic sites defined by residues Glu134 and
Glu197 on one side of the di-iron cluster (see Fig. S6A in the
supplemental material) and by Glu231 on the other side (see
Fig. S6B in the supplemental material). The contributions of
the hydrogen bonds to the binding energy are about 0.5 kcal/
mol in both cases. When the intermediate leading to hydro-
quinone (CCI 16 in Fig. 7) was docked into the active site of
ToMO, we noticed that no hydrogen bond partner was present
in the catalytic pocket, thus giving a molecular basis for the
inability of the wild-type enzyme to produce hydroquinone. On
the contrary, in the (I100Q)-ToMO mutant enzyme, the car-
bonyl group of the Gln100 side chain could form a hydrogen
bond with the hydroxyl group on the intermediate, leading to
hydroquinone (see Fig. S6C in the supplemental material),
with an estimated contribution to the binding energy of 0.7
kcal/mol. This finding led to a prediction which is in agreement
with the experimental observation that the (I100Q)-ToMO
mutant enzyme produces more hydroquinone than catechol.
Conclusions. BMMs have broad substrate specificities, cou-
pled with specific regioselectivity properties, in the hydroxyla-
tion reaction of aromatic substrates. These features are meta-
bolically relevant, because they are the basis for the capabilities
of several microorganisms to grow on selected molecules.
Moreover, given the catalytic potentials of BMMs, they may
constitute a powerful tool for the bioremediation of harmful
substances and may serve as specific biocatalysts in (regio)s-
elective syntheses.
Results from several structural and functional studies sug-
gest that the different regioselectivities of BMMs depend on
differences in the shape of the active site pocket (2, 4, 7, 17, 26,
30, 34, 35). However, a detailed description of the molecular
basis for the regioselectivities of these enzymes is still lacking.
This situation is particularly inconvenient because it impairs
the possibility to attempt rational modifications to produce
new catalysts and/or new microorganisms endowed with spe-
cific, advantageous properties.
In this study, we have developed a procedure based on the
docking of the intermediates of the hydroxylation reaction into
the active site pocket of a specific monooxygenase, ToMO A.
This approach allows for (i) a detailed analysis of the molec-
ular determinants of the enzyme’s regioselectivity, (ii) the pre-
diction of the regioselectivity properties of mutant forms of the
enzyme, in the absence of any experimental data, and (iii) the
prediction of the catalytically productive orientation of a sub-
strate inside the active site pocket. Thus, this procedure is a
valuable tool for the design of mutant monooxygenases for use
in biosynthesis and bioremediation procedures, and its appli-
cability may also be extended to other kinds of substrates and
other multicomponent monooxygenases.
Finally, the results of the docking experiments reported in
this paper have very interesting implications for the catalytic
mechanism of TMOs. The optimal fit between the ToMO
active site pocket and the delocalized carbocation and the good
agreement between experimentally determined regioselectivity
and the regioselectivity predicted using the delocalized carbo-
cations as ligands strongly suggest that the delocalized carbo-
cation is a crucial intermediate in aromatic hydroxylation re-
actions.
ACKNOWLEDGMENTS
We are indebted to Giuseppe D’Alessio, Matthew H. Sazinsky, and
Anna Tramontano for critically reading the manuscript.
This work was supported by grants from the Ministry of University
and Research (PRIN/2002 and PRIN/2004).
REFERENCES
1. Bertoni, G., F. Bolognesi, E. Galli, and P. Barbieri. 1996. Cloning of the
genes for and characterization of the early stages of toluene catabolism in
Pseudomonas stutzeri OX1. Appl. Environ. Microbiol. 62:3704–3711.
2. Borodina, E., T. Nichol, M. G. Dumont, T. J. Smith, and J. C. Murrell. 2007.
Mutagenesis of the “leucine gate” to explore the basis of catalytic versatility
in soluble methane monooxygenase. Appl. Environ. Microbiol. 73:6460–
6467.
3. Cafaro, V., V. Izzo, R. Scognamiglio, E. Notomista, P. Capasso, A. Casbarra,
P. Pucci, and A. Di Donato. 2004. Phenol hydroxylase and toluene/o-xylene
monooxygenase from Pseudomonas stutzeri OX1: interplay between two en-
zymes. Appl. Environ. Microbiol. 70:2211–2219.
4. Cafaro, V., E. Notomista, P. Capasso, and A. Di Donato. 2005. Mutation of
glutamic acid 103 of toluene o-xylene monooxygenase as a means to control
the catabolic efficiency of a recombinant upper pathway for degradation of
methylated aromatic compounds. Appl. Environ. Microbiol. 71:4744–4750.
5. Cafaro, V., E. Notomista, P. Capasso, and A. Di Donato. 2005. Regiospeci-
ficity of two multicomponent monooxygenases from Pseudomonas stutzeri
OX1: molecular basis for catabolic adaptation of this microorganism to
methylated aromatic compounds. Appl. Environ. Microbiol. 71:4736–4743.
6. Cafaro, V., R. Scognamiglio, A. Viggiani, V. Izzo, I. Passaro, E. Notomista,
F. Dal Piaz, A. Amoresano, A. Casbarra, P. Pucci, and A. Di Donato. 2002.
Expression and purification of the recombinant subunits of toluene/o-xylene
monooxygenase and reconstitution of the active complex. Eur. J. Biochem.
269:5689–5699.
7. Fishman, A., Y. Tao, L. Rui, and T. K. Wood. 21 October 2004. Controlling
the regiospecific oxidation of aromatics via active site engineering of toluene
para-monooxygenase of Ralstonia pickettii PKO1. J. Biol. Chem. 280:506–
514. [Epub ahead of print.]
8. Gallagher, S. C., R. Cammark, and H. Dalton. 1997. Alkene monooxygenase
from Nocardia corallina B-276 is a member of the class of dinuclear iron
proteins capable of stereospecific epoxygenation reactions. Eur. J. Biochem.
247:635–641.
9. Hemmi, H., J. M. Studts, Y. K. Chae, J. Song, J. L. Markley, and B. G. Fox.
2001. Solution structure of the toluene 4-monooxygenase effector protein
(T4moD). Biochemistry 40:3512–3524.
10. Horinouchi, M., K. Kasuga, H. Nojiri, H. Yamane, and T. Omori. 1997.
Cloning and characterization of genes encoding an enzyme which oxidizes
dimethyl sulfide in Acinetobacter sp. strain 20B. FEMS Microbiol. Lett.
155:99–105.
11. Kraut, D. A., P. A. Sigala, B. Pybus, C. W. Liu, D. Ringe, G. A. Petsko, and
D. Herschlag. 28 March 2006. Testing electrostatic complementarity in en-
zyme catalysis: hydrogen bonding in the ketosteroid isomerase oxyanion
hole. PLoS Biol. 4:e99. [Epub ahead of print.]
12. Lahiri, S. D., G. Zhang, D. Dunaway-Mariano, and K. N. Allen. 13 March
2003. The pentacovalent phosphorus intermediate of a phosphoryl transfer
reaction. Science 299:2067–2071. [Epub ahead of print.]
13. Leahy, J. G., P. J. Batchelor, and S. M. Morcomb. 2003. Evolution of the
soluble diiron monooxygenases. FEMS Microbiol. Rev. 27:449–479.
14. Lippard, S. J. 2005. Hydroxylation of C-H bonds at carboxylate-bridged
diiron centres. Philos. Trans. A 363:861–877, 1035–1040.
15. Lipscomb, J. D. 1994. Biochemistry of the soluble methane monooxygenase.
Annu. Rev. Microbiol. 48:371–399.
16. Mitchell, K. H., C. E. Rogge, T. Gierahn, and B. G. Fox. 14 March 2003.
Insight into the mechanism of aromatic hydroxylation by toluene 4-mono-
oxygenase by use of specifically deuterated toluene and p-xylene. Proc. Natl.
Acad. Sci. USA 100:3784–3789. [Epub ahead of print.]
17. Mitchell, K. H., J. M. Studts, and B. G. Fox. 2002. Combined participation
of hydroxylase active site residues and effector protein binding in a para to
ortho modulation of toluene 4-monooxygenase regiospecificity. Biochemistry
41:3176–3188.
18. Murray, L. J., R. Garcia-Serres, M. S. McCormick, R. Davydov, S. G. Naik,
S. H. Kim, B. M. Hoffman, B. H. Huynh, and S. J. Lippard. 29 November
2007. Dioxygen activation at non-heme diiron centers: oxidation of a prox-
imal residue in the I100W variant of toluene/o-xylene monooxygenase hy-
droxylase. Biochemistry 46:14795–14809. [Epub ahead of print.]
19. Murray, L. J., and S. J. Lippard. 23 May 2007. Substrate trafficking and
dioxygen activation in bacterial multicomponent monooxygenases. Acc.
Chem. Res. 40:466–474. [Epub ahead of print.]
20. Murray, L. J., S. G. Naik, D. O. Ortillo, R. Garcia-Serres, J. K. Lee, B. H.
Huynh, and S. J. Lippard. 30 October 2007. Characterization of the arene-
oxidizing intermediate in ToMOH as a diiron(III) species. J. Am. Chem.
Soc. 129:14500–14510. [Epub ahead of print.]
21. Murrell, J. C., B. Gilbert, and I. R. McDonald. 2000. Molecular biology and
regulation of methane monooxygenase. Arch. Microbiol. 173:325–332.
VOL. 75, 2009 REGIOSELECTIVITY OF PSEUDOMONAS SP. STRAIN OX1 ToMO 835
14. 22. Newman, L. M., and L. P. Wackett. 1995. Purification and characterization of
toluene 2-monooxygenase from Burkholderia cepacia G4. Biochemistry 34:
14066–14076.
23. Notomista, E., A. Lahm, A. Di Donato, and A. Tramontano. 2003. Evolution
of bacterial and archaeal multicomponent monooxygenases. J. Mol. Evol.
56:435–445.
24. Otyepka, M., P. Banas, A. Magistrato, P. Carloni, and J. Damborsky. 2007.
Second step of hydrolytic dehalogenation in haloalkane dehalogenase inves-
tigated by QM/MM methods. Proteins 70:707–717.
25. Pikus, J. D., J. M. Studts, C. Achim, K. E. Kauffmann, E. Munck, R. J.
Steffan, K. McClay, and B. G. Fox. 1996. Recombinant toluene-4-monoox-
ygenase: catalytic and Mossbauer studies of the purified diiron and rieske
components of a four-protein complex. Biochemistry 35:9106–9119.
26. Pikus, J. D., J. M. Studts, K. McClay, R. J. Steffan, and B. G. Fox. 1997.
Changes in the regiospecificity of aromatic hydroxylation produced by active
site engineering in the diiron enzyme toluene 4-monooxygenase. Biochem-
istry 36:9283–9289.
27. Pompliano, D. L., A. Peyman, and J. R. Knowles. 1990. Stabilization of a
reaction intermediate as a catalytic device: definition of the functional role of
the flexible loop in triosephosphate isomerase. Biochemistry 29:3186–3194.
28. Reisner, E., T. C. Abikoff, and S. J. Lippard. 31 October 2007. Influence of
steric hindrance on the core geometry and sulfoxidation chemistry of car-
boxylate-rich diiron(II) complexes. Inorg. Chem. 46:10229–10240. [Epub
ahead of print.]
29. Rosenzweig, A. C., H. Brandstetter, D. A. Whittington, P. Nordlund, S. J.
Lippard, and C. A. Frederick. 1997. Crystal structures of the methane mono-
oxygenase hydroxylase from Methylococcus capsulatus (Bath): implications
for substrate gating and component interactions. Proteins 29:141–152.
30. Rui, L., Y. M. Kwon, A. Fishman, K. F. Reardon, and T. K. Wood. 2004.
Saturation mutagenesis of toluene ortho-monooxygenase of Burkholderia
cepacia G4 for enhanced 1-naphthol synthesis and chloroform degradation.
Appl. Environ. Microbiol. 70:3246–3252.
31. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a
laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold
Spring Harbor, NY.
32. Sazinsky, M. H., J. Bard, A. Di Donato, and S. J. Lippard. 2004. Crystal
structure of the toluene/o-xylene monooxygenase hydroxylase from Pseudo-
monas stutzeri OX1. Insight into the substrate specificity, substrate channel-
ing, and active site tuning of multicomponent monooxygenases. J. Biol.
Chem. 279:30600–30610.
33. St. Maurice, M., and S. L. Bearne. 2000. Reaction intermediate analogues
for mandelate racemase: interaction between Asn 197 and the alpha-hy-
droxyl of the substrate promotes catalysis. Biochemistry 39:13324–13335.
34. Tao, Y., A. Fishman, W. E. Bentley, and T. K. Wood. 2004. Altering toluene
4-monooxygenase by active-site engineering for the synthesis of 3-methoxy-
catechol, methoxyhydroquinone, and methylhydroquinone. J. Bacteriol. 186:
4705–4713.
35. Vardar, G., and T. K. Wood. 2004. Protein engineering of toluene-o-xylene
monooxygenase from Pseudomonas stutzeri OX1 for synthesizing 4-methyl-
resorcinol, methylhydroquinone, and pyrogallol. Appl. Environ. Microbiol.
70:3253–3262.
36. Viggiani, A., L. Siani, E. Notomista, L. Birolo, P. Pucci, and A. Di Donato.
2004. The role of conserved residues H246, H199 and Y255 in the catalysis
of catechol 2,3-dioxygenase from Pseudomonas stutzeri OX1. J. Biol. Chem.
279:48630–48639.
37. Weiner, S. J., P. A. Kollman, D. A. Case, U. C. Singh, C. Ghio, G. Alagona,
S. Profeta, and P. Weiner. 1984. A new force field for molecular mechanical
simulation of nucleic acids and proteins. J. Am. Chem. Soc. 106:765–784.
38. Whited, G. M., and D. T. Gibson. 1991. Toluene-4-monooxygenase, a three-
component enzyme system that catalyzes the oxidation of toluene to p-cresol
in Pseudomonas mendocina KR1. J. Bacteriol. 173:3010–3016.
39. Zhao, M., B. Helms, E. Slonkina, S. Friedle, D. Lee, J. Dubois, B. Hedman,
K. O. Hodgson, J. M. Frechet, and S. J. Lippard. 2008. Iron complexes of
dendrimer-appended carboxylates for activating dioxygen and oxidizing hy-
drocarbons. J. Am. Chem. Soc. 11:11.
40. Zhorov, B. S., and P. D. Bregestovski. 2000. Chloride channels of glycine and
GABA receptors with blockers: Monte Carlo minimization and structure-
activity relationships. Biophys. J. 78:1786–1803.
836 NOTOMISTA ET AL. APPL. ENVIRON. MICROBIOL.
![oxygen to the aromatic ring (16, 19, 20) are shown in Fig. 1.
The reactive species, likely a di-iron(III) intermediate, attacks
the pi-electron system of the aromatic ring, forming epoxide 1
or delocalized carbocation 2. The opening of the epoxide ring
eventually provides the delocalized carbocation. The migration
of a hydride from the sp3
-hybridized carbon to the adjacent
atom then converts the carbocation to the more stable ketone
3. Finally, the dissociation of the ketone and its tautomeriza-
tion yield phenolic product 4.
The different regioselectivities shown by BMMs have been
attributed previously to differences in the shape of the active
site pocket, an idea supported by the fact that several point
mutations in the active sites of T4MO, toluene p-monooxygen-
ase, ToMO, MMO, and B. cepacia G4 T2MO cause large
variations in the regioselectivities of these BMMs (2, 7, 26, 30,
34, 35). In a previous paper (4), we have reported that muta-
tions at position 103 of the A subunit change the regioselec-
tivity of ToMO from Pseudomonas sp. strain OX1. However, a
complete study of the active site pocket of ToMO A is still
lacking.
In this paper, we report the effects of the substitution of six
residues in the ToMO A active site on substrate specificity and
regioselectivity. Furthermore, we present a detailed analysis of
the molecular determinants of regioselectivity based on the
docking of substrates and hypothetical intermediates of the
aromatic hydroxylation reaction into the active site of the crys-
tallographic structure of ToMO (Protein Data Bank [PDB]
code 1T0Q [32]), followed by a Monte Carlo optimization. The
results show that (i) the fine-tuning of TMO regioselectivity
can be achieved through a careful alteration of the shape of the
active site pocket and that (ii) the effects of mutations on
regioselectivity can be quantitatively predicted using the pro-
cedure described herein.
MATERIALS AND METHODS
Materials. Bacterial cultures, plasmid purifications, and transformations were
performed according to the procedures of Sambrook et al. (31). Escherichia coli
strains JM109 and CJ236 and vector pET22b(ϩ) were from Novagen. Plasmid
pBZ1260 (1) used for the expression of the ToMO cluster was kindly supplied by
P. Barbieri (Dipartimento di Biologia Strutturale e Funzionale, Universita`
dell’Insubria, Varese, Italy). E. coli strain JM101 was purchased from Boehr-
inger. The pGEM-3Z expression vector, Wizard SV gel, and the PCR clean-up
system for the elution of DNA fragments from agarose gel were obtained from
Promega. Enzymes and other reagents for DNA manipulation were from New
England Biolabs. The oligonucleotides were synthesized at MWG-Biotech
(Ebersberg, Germany). All other chemicals were from Sigma. The expression
and purification of recombinant catechol 2,3-dioxygenase from Pseudomonas sp.
strain OX1 are described elsewhere (36).
ToMO A mutagenesis. Plasmids for the expression of ToMO complexes with
mutated ToMO A subunits were prepared by site-directed mutagenesis of plas-
mid pTOU as described previously (4). Sequences of the mutagenic oligonucle-
otides are reported in Table S1 in the supplemental material.
Determination of apparent kinetic parameters and identification of products.
Assays were performed as described previously (4, 5) using E. coli JM109 cells
transformed with plasmid pBZ1260 or plasmid pTOU, which expresses wild-type
ToMO or ToMO mutant enzymes, respectively.
All kinetic parameters were determined using whole cells (4, 5). Enzymatic
activity on phenol was measured by monitoring the production of catechol in
continuous coupled assays with recombinant catechol 2,3-dioxygenase from
Pseudomonas sp. strain OX1 (5). The determination of apparent kinetic param-
eters for benzene, toluene, o-xylene, and naphthalene was carried out by a
discontinuous assay (5). For the calculation of the kcat values, amounts of pro-
teins were calculated as described previously (5). All the ToMO mutant enzymes
showed expression levels similar to that of the wild-type enzyme.
Reaction products were identified as described previously (5). All the regio-
specificity studies were performed using substrate concentrations higher than the
Km values. Under these conditions, absolute yields of products were proportional
to kcat values.
Modeling of substrates and intermediates into the active site of ToMO A.
Substrates and reaction intermediates were docked into the active site of ToMO
A by using the Monte Carlo energy minimization strategy. The ZMM-MVM
molecular modeling package (ZMM Software Inc. [http://www.zmmsoft.com])
was used for all calculations. This software allows conformational searches using
generalized coordinates such as torsion and bond angles instead of conventional
Cartesian coordinates (40).
Atom-atom interactions were evaluated using assisted model building with
energy refinement force fields (37) with a cutoff distance of 8 Å. Conformational
energy calculations included van der Waals, electrostatic, H bond, and torsion
components. A hydration component was not included. Electrostatic interactions
were assessed with a relative dielectric constant of 4.
Substrate and intermediate structures were prepared using the PyMOL soft-
ware (DeLano Scientific LLC). Geometry was optimized using the Zl module of
ZMM. Partial charges were attributed using the complete neglect of differential
overlap method in the HyperChem software (HyperCube Inc. [http://www.hyper
.com]).
The X-ray structure of the ToMO A-thioglycolate complex (PDB code 1T0Q)
was used to build the models of the ToMO A-substrate and ToMO A-interme-
diate complexes. To reduce computational time, a double-shell model of the
enzyme was built. The inner shell included 28 ToMO A residues surrounding the
active site cavity. During energy calculation procedures, the side chain torsion
angles—but not the backbone torsion angles—of these residues were allowed to
vary. Due to the asymmetric shape of the cavity, which is flat with the di-iron
cluster on one end and evolutionarily nonconserved hydrophobic residues on the
opposite side, the residues of the inner shell were selected manually. They
included the ligands of the iron ions, all residues with at least one side chain atom
contributing to the hydrophobic part of the active site cavity, and all residues with
at least one side chain atom less than 5 Å from the previous residues.
The outer shell included 139 residues, which did not belong to the inner shell
FIG. 1. Possible intermediates in the aromatic hydroxylation reaction catalyzed by ToMO. Intermediates 1, 2, and 3 are an epoxide, a
carbocation, and a ketone, respectively. R1 and R2 are hydrogen atoms or methyl groups. The geometrical features of the di-iron(III)-(hydro)per-
oxide intermediate and the details of the O-O bond cleavage reaction are not known.
824 NOTOMISTA ET AL. APPL. ENVIRON. MICROBIOL.](data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7)