The Ala95-to-Gly substitution in Aerococcus viridans l -lactate oxidase revisited - Structural consequences at the catalytic site and effect on reactivity with O2 and other electron acceptors

Article abstract of DOI:10.1111/febs.13162

Aerococcus viridansl-lactate oxidase (avLOX) is a biotechnologically important flavoenzyme that catalyzes the conversion of l-lactate and O₂ into pyruvate and H₂O_2. The enzymatic reaction underlies different biosensor applications of avLOX for blood l-lactate determination. The ability of avLOX to replace O₂ with other electron acceptors such as 2,6-dichlorophenol-indophenol (DCIP) allows the possiblity of analytical and practical applications. The A95G variant of avLOX was previously shown to exhibit lowered reactivity with O₂ compared to wild-type enzyme and therefore was employed in a detailed investigation with respect to the specificity for different electron acceptor substrates. From stopped-flow experiments performed at 20 °C (pH 6.5), we determined that the A95G variant (fully reduced by l-lactate) was approximately three-fold more reactive towards DCIP (1.0 ± 0.1 × 10⁶ M^-1·s^-1) than O₂, whereas avLOX wild-type under the same conditions was 14-fold more reactive towards O₂ (1.8 ± 0.1 × 10⁶ m^-1·s^-1) than DCIP. Substituted 1,4-benzoquinones were up to five-fold better electron acceptors for reaction with l-lactate-reduced A95G variant than wild-type. A 1.65-? crystal structure of oxidized A95G variant bound with pyruvate was determined and revealed that the steric volume created by removal of the methyl side chain of Ala95 and a slight additional shift in the main chain at position Gly95 together enable the accomodation of a new active-site water molecule within hydrogen-bond distance to the N5 of the FMN cofactor. The increased steric volume available in the active site allows the A95G variant to exhibit a similar trend with the related glycolate oxidase in electron acceptor substrate specificities, despite the latter containing an alanine at the analogous position.

Full text of DOI:10.1111/febs.13162

The Ala95-to-Gly substitution in Aerococcus viridans
L-lactate oxidase revisited – structural consequences at
the catalytic site and effect on reactivity with O2 and
other electron acceptors
Thomas Stoisser1,2, Daniela Rainer1,2,3, Stefan Leitgeb1,2, David K. Wilson3 and Bernd Nidetzky1,2,4
1 Research Center Pharmaceutical Engineering, Graz, Austria
2 Institute of Biotechnology and Biochemical Engineering, Graz University of Technology, NAWI Graz, Graz, Austria
3 Department of Molecular and Cellular Biology, University of California, Davis, CA, USA
4 Austrian Centre of Industrial Biotechnology, Graz, Austria
Keywords
Aerococcus viridans L-lactate oxidase (avLOX) is a biotechnologically
important flavoenzyme that catalyzes the conversion of L-lactate and O2
electron acceptor; flavin oxygen reactivity;
FMN; a-hydroxy acid oxidizing
flavoenzymes; L-lactate oxidase
into pyruvate and H2O2. The enzymatic reaction underlies different biosen-
sor applications of avLOX for blood L-lactate determination. The ability
of avLOX to replace O2 with other electron acceptors such as 2,6-dichlor-
Correspondence
ophenol-indophenol (DCIP) allows the possiblity of analytical and practi-
cal applications. The A95G variant of avLOX was previously shown to
exhibit lowered reactivity with O2 compared to wild-type enzyme and
therefore was employed in a detailed investigation with respect to the speci-
ficity for different electron acceptor substrates. From stopped-flow experi-
ments performed at 20 °C (pH 6.5), we determined that the A95G variant
(fully reduced by L-lactate) was approximately three-fold more reactive
towards DCIP (1.0 ꢀ 0.1 9 106 Mꢁ1ꢂsꢁ1) than O2, whereas avLOX wild-
B. Nidetzky, Institute of Biotechnology and
Biochemical Engineering, Graz University of
Technology, NAWI Graz, Petersgasse 12,
A-8010 Graz, Austria
Fax: +43 316 873 8434
Tel: +43 316 873 8400
E-mail: bernd.nidetzky@tugraz.at
(Received 9 October 2014, revised 19
November 2014, accepted 24 November
2014)
type under the same conditions was 14-fold more reactive towards O2
(1.8 ꢀ 0.1 9 106
M
ꢂsꢁ1) than DCIP. Substituted 1,4-benzoquinones were
ꢁ1
up to five-fold better electron acceptors for reaction with L-lactate-reduced
ꢀ
A95G variant than wild-type. A 1.65-A crystal structure of oxidized A95G
doi:10.1111/febs.13162
variant bound with pyruvate was determined and revealed that the steric
volume created by removal of the methyl side chain of Ala95 and a slight
additional shift in the main chain at position Gly95 together enable the
accomodation of a new active-site water molecule within hydrogen-bond
distance to the N5 of the FMN cofactor. The increased steric volume avail-
able in the active site allows the A95G variant to exhibit a similar trend
with the related glycolate oxidase in electron acceptor substrate specificities,
despite the latter containing an alanine at the analogous position.
Database
Atomic coordinates and related experimental data concerning the structure of the A95G vari-
ant of A. viridans lacate oxidase have been deposited in the Protein Data Bank under the iden-
tifier 4RJE.
Abbreviations
avLOX, Aerococcus viridans L-lactate oxidase; DCIP, 2,6-dichlorophenol-indophenol; GOX, glycolate oxidase; LOX, L-lacate oxidase; PDB,
Protein Data Bank.
562
FEBS Journal 282 (2015) 562–578 © 2014 FEBS

T. Stoisser et al.
Ala95-to-Gly substitution in L-lactate oxidase
this relationship, it is essential that the comparative
kinetic analysis of wild-type and variant enzymes is
complemented by a determination of the correspond-
ing atomic-resolution protein structures.
Introduction
The a-hydroxy acid-oxidizing enzymes constitute a dis-
tinct family of FMN-dependent dehydrogenases and
oxidases [1,2]. They share a common (b/a)8 TIM bar-
rel fold, with the FMN binding site located at the
C-terminal end of the b barrel [3–7]. Enzymes are
functional oligomers, typically homotetramers. They
catalyze the oxidation of their a-hydroxy acid sub-
strates (e.g. L-lactate, glycolate, L-mandelate) via C-H
hydrogen abstraction [8–14], probably via hydride
transfer to the oxidized form of FMN [15,16]. The res-
idues involved in FMN binding and forming the active
site are highly conserved [3–7], suggesting a common
catalytic mechanism of substrate oxidation in the first
(reductive) half-reaction of the overall conversion
(Fig. 1, Scheme 1). Strikingly, however, individual
enzymes differ strongly in how reactive their reduced-
state FMN cofactor is towards molecular oxygen.
Based on the second-order rate constant for FMN oxi-
dation by O2 in the second (oxidative) half-reaction,
the family is subdivided into a group of highly O2
reactive oxidases/monooxygenases such as L-lactate
oxidase (LOX; 1.8 9 106 Mꢁ1ꢂsꢁ1) [9] and glycolate
oxidase (GOX; 8.5 9 104 Mꢁ1ꢂsꢁ1) [8] and another
group of dehydrogenases barely reactive with O2 and
including mandelate dehydrogenase [14] and flavocyto-
chrome b2 [11,12]. Small differences in microenviron-
ment of the reactive C4a-N5 locus of FMN might be
responsible [17,18], in part at least, for these starkly
varied O2 reactivities. Enzyme structures [3–7] reveal
that strand b1 of the b barrel provides close contact
contact with the re-face, the side of the flavin ring
opposite to where the substrate binds (Fig. 1). A natu-
rally occurring Ala/Gly residue variation within strand
b1 is a prominent factor of structural variability
around the FMN N5 atom [3–7]. Evidence from site-
directed mutagenesis reveals that Ala/Gly residue
exchange in different enzymes impacts significantly
upon the O2 reactivity, in addition to having a major
effect on the a-hydroxy acid substrate specificity [19–
23]. However, the results also show that the Ala/Gly
pattern does not constitute a simply addressable oxi-
dase-dehydrogenase switch in a-hydroxy acid-oxidizing
flavoenzymes, unlike vanillyl-alcohol oxidases where a
similar pattern of Ala and Gly (or Pro) was proposed
to determine flavoenzyme activity as a dehydrogenase
and oxidase, respectively [24]. Variability in effects
caused by the Ala/Gly residue substitution in different
a-hydroxy acid-oxidizing enzymes suggests a compara-
bly complex relationship between the catalytic site
structure and the control of O2 reactivity in this fla-
voenzyme family. To gain a deeper understanding of
In the present study, we have focused on the LOX
from Aerococcus viridans (avLOX), a prototypical oxi-
dase-type member of the a-hydroxy acid-oxidizing fla-
voenzyme family that has been well characterized
biochemically [9,21,25,26] and structurally [6,27–29].
The enzyme is biotechnologically important for its
commercialized use in analytical devices for blood
L-lactate determination [30,31]. The avLOX catalyzed
conversion of L-lactate and O2 into pyruvate and
H2O2 provides convenient indirect detection of the
analyte via the H2O2 that is formed. However, avLOX
is also able to replace O2 by another oxidant such as
2,6-dichlorophenol-indophenol (DCIP), which is
exploited in the fabrication and application of L-lactate
test strips [32]. An A95G variant of avLOX was previ-
ously constructed and shown to exhibit approximately
10-fold lower O2 reactivity than the wild-type at 4 °C
[21]. The mutation was also shown to enhance the
enzyme’s specificity for long chain and generally larger
sized a-hydroxy-acid substrates [21,25].
In competition assays containing DCIP in a concen-
tration matching that of O2 at air saturation, we noted
that the avLOX preference for electron acceptor use in
L-lactate oxidation was strongly altered as result of the
substitution of Ala95 by Gly: the A95G variant uti-
lized DCIP prior to using O2, whereas the wild-type
enzyme showed reverse-order sequential utilization of
Fig. 1. Close up of the active site of A95G avLOX with bound
pyruvate at pH 8.0. The reactive C4a-N5 moiety of the FMN that is
involved in hydrogen abstraction from the a-hydroxy acid substrate
is indicated. Hydrogen bonds are indicated by dashed lines in
yellow. The crystallographic results from the present study are
shown. A comparison with the published structure of the wild-type
avLOX [27] with Ala95 is considered in the Discussion.
FEBS Journal 282 (2015) 562–578 © 2014 FEBS
563

Ala95-to-Gly substitution in L-lactate oxidase
T. Stoisser et al.
the two electron acceptors present, as expected for a
true oxidase. This result therefore suggested ‘dehydro-
genase-like’ behavior of the A95G variant contrary to
the wild-type enzyme and prompted further research
aiming to clarify the changes in avLOX electron
acceptor substrate specificity brought about by the
Change in electron acceptor reactivity as a result
of substitution of Ala95 by Gly
When comparing specific enzyme activities determined
from initial-rate assays of L-lactate oxidation by DCIP
or O2, we noted the A95G variant to react approxi-
mately four-fold faster with DCIP than O2, in contrast
to the wild-type, which showed four-fold higher reac-
tivity with O2 than DCIP. A faster reaction of A95G
variant with DCIP compared to O2 was also seen
when L-lactate was replaced by alternate enzyme sub-
strates such as glycolate (4.2-fold), (S)-2-hydroxy-buty-
rate (5.5-fold) and (S)-mandelate (8.2-fold). By
comparing O2 consumption with H2O2 formation rates
and finding that both rates were identical within limits
of experimental error (ꢀ 10%), we confirmed that L-
lactate oxidation by O2 followed the uncoupled oxi-
dase path in both wild-type and mutant enzyme
(Scheme 1). The alternative coupled path utilized by L-
lactate monooxygenase [10,19,20,33,34] where either
none or only a small amount of H2O2 is released
(Scheme 1) was thus ruled out with respect to having
any significance under the reaction conditions used. To
determine whether the change in relative specific activ-
ity caused by the substitution of Ala95 also resulted in
altered preference for utilization of DCIP compared to
O2, we performed competition assays in which both
electron acceptors were present at a comparable con-
centration ( ꢃ 200 lM) and the time-dependent con-
sumption of each acceptor substrate during L-lactate
oxidation was measured. The results shown in Fig. 2
show that DCIP was the preferred electron acceptor
substrate for A95G variant, whereas the wild-type
clearly preferred O2.
ꢀ
site-directed replacement of Ala95 by Gly. A 1.65-A
crystal structure of A95G variant (FMN oxidized) in
complex with pyruvate was determined, and detailed
stopped-flow kinetic analysis of the reoxidation of
L-lactate-reduced enzymes by different electron accep-
tors was performed. The A95G variant is shown to
place the reactive C4a-N5 moiety of its FMN cofactor
into a more hydrated structural microenvironment,
somewhat similar to that seen in GOX from the same
flavoenzyme family [3], and also, similar to GOX, to
exhibit electron acceptor specificities somewhere inbe-
tween an oxidase and dehydrogenase.
Results
A95G avLOX was recombinantly produced in E. coli
and purified to apparent homogeneity (data not
shown). An identical wild-type avLOX is commercially
available and, unless stated otherwise, this highly pure
recombinant enzyme preparation from E. coli was
used. However, wild-type avLOX was additionally pre-
pared as described for the A95G variant. To ensure
that the preparations of wild-type enzyme obtained
commercially and isolated by us were comparable, we
performed a relative kinetic characterization of the
two. Apparent turnover numbers for L-lactate oxida-
tion at air-saturated oxygen conceꢁn1tration and 37 °C
were identical ( ꢃ 140 sꢁ1ꢂmol H2O2 ꢂmol FMNꢁ1ꢂsꢁ1
)
and within limits of the experimental error (ꢀ 10%).
The transient rate constant for oxidation of L-lactate
in anaerobic stopped flow experiments at 20 °C
(described below) was also identical (220–250 sꢁ1) for
the two preparations. Absorbance spectra in the wave-
length range 320–600 nm were highly similar for all
enzyme preparations, wild-type or A95G variant.
However, we noted the relative FMN content of
enzyme preparations as isolated was somewhat lower
than that of the commercial preparation. Active-site
concentrations are therefore always based on the
amount of FMN present.
The wild-type avLOX used in the present study dif-
fers from the published enzyme structure [6,27] at
four sequence positions as a result of substitutions:
T102A, S163G, G232A and R255A. None of the
substituted residues is close to the active site, nor do
they have any reported role in enzyme function or
stability.
Scheme 1. Proposed reaction pathways of L-lactate oxidase
(uncoupled half-reactions: H2O2 releasing) and
L-lactate
monooxygenase (coupled half-reactions: acetate releasing; rate
constants marked with asterisk). E-Flox and E-Flred are enzymes
with FMN oxidized and reduced, respectively. The scheme is
adapted from Maeda-Yorita et al. [9].
564
FEBS Journal 282 (2015) 562–578 © 2014 FEBS

T. Stoisser et al.
Ala95-to-Gly substitution in L-lactate oxidase
clearly resolved in time (Fig. 4), stopped flow absor-
bance traces from reactions at 20 °C did not exhibit
similar dynamics of an absorbance band at 530 nm.
The proposed pathway of the reductive half-reaction
therefore involves Michaelis complex formation, con-
version to reduced enzyme–product complex and dis-
sociation of the a-keto acid (Scheme 1). However, the
product release step might be too fast at 20 °C to
allow for build-up of enzyme–product complex in sig-
nificant amounts or the elevated temperature prevents
detection of such complex. Therefore, under all condi-
tions used at 20 °C, a decrease in A455 occurred in a
single (exponential) phase (Fig. 3), whereas, in con-
trast, it was clearly biphasic (bi-exponential) for reac-
tions with L-lactate at 4 °C (Fig. 4). Pseudo-first-order
rate constants (kobs) determined from single- or dou-
ble-exponential fits of stopped flow curves exhibited
hyperbolic dependence on the a-hydroxy acid substrate
concentration (Fig. 5). Table 1 summarizes the kinetic
parameters (kred, Kd) calculated from the data. The
results obtained at 4 °C are consistent with Yorita
et al. [9,21], whereas differences in parameter values
are attributed to varied buffer and pH conditions as
used in the present study and in the earlier study. Buf-
fer effects have been noted in earlier studies of differ-
ent a-hydroxy acid oxidases [10,35]. The kred for
oxidation of L-lactate was similar for wild-type and
A95G avLOX. Binding of L-lactate (Kd) was weakened
7.4-fold as result of the mutation. In kred terms, the
A95G variant was much more reactive than wild-type
enzyme when used with the substrates (S)-mandelic
acid (66-fold) and (S)-2-hydroxy butyric acid (41-fold).
The improved reactivity of these bulky a-hydroxy acid
A
B
Fig. 2. Utilization of O2 (dotted line) and DCIP (dashed line) as
competing electron acceptors for enzymatic L-lactate oxidation by
wild-type (A) and A95G avLOX (B). Reactions were performed at
37 °C using 50 mM L-lactate as substrate. Potassium phosphate
buffer (50 mM; pH 6.5) was used. The concentration of wild-type
avLOX was 25.6 nM and that of the A95G variant was 7.7 nM. The
retardation of O2 consumption by A95G avLOX at the time when
DCIP was largely depleted (≥ 2 min) was noted but not analyzed
further.
Kinetics of the reductive half-reaction with
different a-hydroxy acid substrates
substrates was mainly the result of an increase in kred
.
Anaerobic stopped flow experiments were performed
to determine the kinetics of FMN reduction in wild-
type and A95G avLOX by L-lactate and also other
(bulkier) a-hydroxy acid substrates. Working at 20 °C
(Fig. 3), the present study extends earlier mechanistic
accounts of the same two enzymes [9,21,25] where the
transient kinetic analysis was conducted at 4 °C. With
L-mandelate, however, kinetic studies of wild-type and
A95G avLOX were subsequently also performed at
25 °C [25].
Irrespective of the substrate and its concentration
used, stopped flow absorbance traces at different wave-
lengths of A95G avLOX were found to be similar to
those of the wild-type enzyme, revealing a time-depen-
dent decrease in A455 as a result of a progressing
reduction of the FMN cofactor (Fig. 3). Unlike at
4 °C, where the formation and decay of reduced
enzyme-pyruvate complex (measured at A530) was
Kinetics of the oxidative half-reaction using
different electron acceptors
Reoxidation of the reduced lactate oxidases was mea-
sured in stopped flow experiments at 20 °C using dif-
ferent electron acceptors applied under varied
concentrations. Stopped flow absorbance traces
recorded at different wavelengths indicated the com-
plete reoxidation of both enzymes under all conditions
used. Figure 3 shows the oxidation of wild-type
avLOX by DCIP as an example. An increase in A455
occurred in a single (exponential) phase from which
kobs was determined. The kobs was also determined
from the corresponding decrease in DCIP A600 (Fig. 3)
and the kobs values obtained from measurements of
A455 and A600 agreed within the limits of error (typi-
cally ꢀ10%). Unless explicitly noted, however, the kobs
data are from stopped flow absorbance traces at A455
.
FEBS Journal 282 (2015) 562–578 © 2014 FEBS
565

Ala95-to-Gly substitution in L-lactate oxidase
T. Stoisser et al.
A
B
C
D
Fig. 3. Kinetic analysis of reduction and re-
oxidation of FMN in avLOX analyzed in
anaerobic stopped flow experiments at
20 °C. (A, B) Single wavelength traces of
FMN reduction (455 nm) in wild-type (A,
6 lM; 2 mM L-lactate) and A95G avLOX (B,
3 lM; 12.5 mM L-lactate). The grey traces
show the measurements and the black
lines are single exponential fits to the
data. (C) Single wavelength traces of FMN
re-oxidation (455 nm; black) by 25 lM
DCIP (600 nm; blue) using L-lactate
reduced enzyme (5 lM). Single exponential
fits of averaged traces for the increase in
A455 and decrease in A600 gave values of
kobs that agreed within 10%. (D) Enzyme
(5 lM) absorbance spectra recorded in
50 mM potassium phosphate buffer (pH
6.5) in the presence of 25 lM DCIP, for
oxidized enzyme (black line), enzyme
reduced by the molar equivalent of L-
lactate (black dashed line) and dithionite-
reduced enzyme (grey line). (E)
E
Superimposition of time-resolved
absorbance spectra recorded during the
reoxidation of L-lactate-reduced enzyme
(6 lM) by 30 lM DCIP.
The kobs values exhibited a linear dependence on the
acceptor substrate concentration used (Fig. 6). The
second-order rate constant kox was obtained as the
slope of straight-line fit to the data (Fig. 6). A sum-
mary of the kinetic parameters is provided in Table 2.
The O2 reactivity was decreased 5.5-fold in A95G
compared to wild-type avLOX, confirming evidence
obtained in the study reported by Yorita et al. [21] as
performed at 4 °C. The reactivities of alternative elec-
tron acceptors, however, were higher with A95G than
wild-type avLOX: DCIP (7.7-fold); 2,5-dichloro-1,4-
benzoquinone (3.3-fold); and 2-methyl-1,4-benzoqui-
none (4.3-fold). The electron acceptors used spanned a
representative range of standard redox potentials from
+0.176 V to +0.281 V (Table 2). Unlike other fla-
voenzymes in which the kox exhibited clear dependence
on acceptor substrate redox potential [38], the kox val-
ues of the reduced lactate oxidases did not reveal a
recognizable trend correlation that would have sup-
ported quantitative structure–activity relationship
analysis.
Crystal structure of A95G avLOX
The A95G avLOX crystal structure was determined
and the crystallographic data collection and refinement
statistics are summarized in Table 3. As with the
wild-type structure [6,27], the A95G avLOX structure
is tetrameric with an entire C4 tetramer within the
asymmetric unit. Disorder prevented the fitting of resi-
due ranges in each subunit in the final refined model:
A (1–5, 203–204), B (1–6, 199–204), C (1–7, 203–204,
214–217), D (1–7, 203–204). FMN cofactors and pyru-
vates were found in locations consistent with the wild-
type structure [Protein Data Bank (PDB) accession
number: 2E77] [27]. Unless stated otherwise, structural
analysis was carried out using subunit A (Fig. 7)
because the overall B-factors for each subunit (A:
2
2
2
2
ꢀ
ꢀ
ꢀ
ꢀ
13.6 A ; B: 14.9 A ; C: 14.5 A ; D: 14.0 A ) indicated
that it had the lowest disorder. These subunits are sim-
ilar overall with the r.m.s.d. of Cas from subunit A
ꢀ
calculated to be 0.26, 0.21 and 0.16 A, respectively, for
subunits B, C and D. As previously determined, the
566
FEBS Journal 282 (2015) 562–578 © 2014 FEBS

T. Stoisser et al.
Ala95-to-Gly substitution in L-lactate oxidase
A
C
B
D
Fig. 4. Reductive half-reaction of A95G avLOX analyzed in enzyme-monitored turnover experiments at 4 °C in 50 mM potassium phosphate
buffer (pH 6.5). (A) Enzyme (5 lM) absorbance spectra at reaction start (black line) and after mixing with 8 mM L-lactate and reaction for
18 ms (red line). The grey line shows the spectrum after reaction for 200 ms. Inset shows the time-resolved absorption trace at 530 nm,
normalized on the maximum absorption observed at this wavelength in the reaction. According to previous studies of avLOX [9,21], the
change in the 530-nm band is a result of the formation and decay of a complex between reduced enzyme and pyruvate, and the inset
shows the dynamics of the complex. (B) Stopped flow trace at 455 nm monitoring FMN reduction, with the inset showing the
corresponding trace at 530 nm. Solid lines are fits of Scheme 1 for anaerobic L-lactate oxidation (steps k1 to k6) to the data using the
differential equation solver BERKELEY MADONNA, version 8.3.18 (www.berkeleymadonna.com). (C, D) The same analysis carried out for wild-type
avLOX with the exception that 10 mM L-lactate and 19.4 lM enzyme were used. In (C), the red line shows the reaction after 14 ms. Useful
ꢁ1
fits were obtained with the rate constants: A95G variant: k2/k1 = 6 mM, k3 = 77 sꢁ1, k4 = 0 sꢁ1, k5 = 24 sꢁ1 and k6 = 10 mM ꢂsꢁ1. Wild-
ꢁ1
type: k2/k1 = 1 mM, k3 = 88 sꢁ1, k4 = 0 sꢁ1, k5 = 87 sꢁ1 and k6 = 10 mM ꢂsꢁ1. Rate constants k3 and k5 agree reasonably with the kobs
data obtained from double-exponential fits of the absorbance trace at 455 nm (Table 1).
overall fold is a (b/a)8 barrel and the active site is
found adjacent to the FMN cofactor at the C-terminal
end of the inner b barrel (Fig. 7). In addition to the
A95G substitution, this protein contains four other
residue replacements (T102A, S163G, G232A and
R255A), which distinguish it from the reported avLOX
structure [27]. In most cases, these were well defined in
the electron density. From a structural perspective,
they are likely to be functionally inert because they are
these residues is important with respect to access to
the active site of the enzyme.
A comparison of the structure of the pyruvate-
bound A95G and wild-type active sites reveals almost
identical structures (Figs 1 and 8A). In particular,
there are no changes in any of the residues within van
der Waals contact of the pyruvate (i.e. Tyr40, Tyr124,
Tyr146, Arg181, Tyr215, His265 or Arg268). There is
a significant change in the main-chain conformational
torsion angles of Gly95 relative to the wild-type
ꢀ
all on the periphery of the structure and > 18 A from
ꢀ
the oxidized substrate carbon on the pyruvate in the
active site (Fig. 7). The pyruvate occupying the active
site is almost completely sequestered from solvent by
the loops connecting strands and helices, as well as by
two longer irregular excursions composed of amino
acids 34–44 and 179–218. One or more of these loops
must therefore move to allow substrate binding and
product exit. We note B-factors trend higher for amino
acids in this second loop, particularly in subunits B
and C, suggesting that the mobility of some or all of
enzyme, leading to a 0.8-A displacement of the a-car-
bon, creating somewhat more volume adjacent to the
pyruvate methyl binding site. The missing methyl
group in the wild-type alanine is replaced with a water
molecule that hydrogen bonds with N5 of the isoallox-
azine ring (Figs 1 and 8A).
The FMN cofactor is bound at the C-terminal end
of the internal b-barrel and interactions between it and
the enzyme are almost identical to those found in the
wild-type structure. One notable difference is the
FEBS Journal 282 (2015) 562–578 © 2014 FEBS
567

Ala95-to-Gly substitution in L-lactate oxidase
T. Stoisser et al.
A
C
Fig. 5. Dependencies of pseudo-first-order
rate constants of anaerobic enzyme-
monitored turnover at 4 °C (A, B) and
20 °C (C, D) on the L-lactate concentration.
(A, C) Results of reductive half-reactions
of wild-type avLOX. (B, D) Results of
reductive half-reactions of A95G variant.
The shown kobs values are from fits of
double- (A, B) and single- (C, D)
B
D
exponential decay functions to stopped-
flow absorbance traces (455 nm) of FMN
reduction. Black and red symbols show
kobs values associated with the fast and
slow kinetic transient, respectively. Solid
lines show hyperbolic fits of the data,
giving rise to a kred value at saturating L-
lactate and a dissociation constant of L-
lactate (Kd). Values of kred and Kd for the
fast and slow reduction phase are
summarized in Table 1.
planarity of the flavin ring. The prior structure appears
to be refined as a planar ring system with C7 and C8,
two peripheral methyl groups allowed to diverge
slightly from the plane. Attempts to refine the A95G
structure with strict planarity restraints being imposed
resulted in clear residual electron density, which was
consistent in all four monomers, showing that the
rings were not planar in this case. The final structure
resulted in a flavin ring with an approximately 18°
‘butterfly angle’, which is within the range of 0–34°
found in other examples of reduced flavin ring systems
[39]. This FMN structure is clearly more consistent
with the unbiased high-resolution electron density
(Fig. 9). Crystals were grown and maintained in an
ambient air environment, which should strongly pro-
mote the oxidized form of the cofactor; however, pho-
toreduction of flavins during X-ray exposure appears
to be a widespread phenomenom and provides a likely
explanation for the lack of FMN planarity seen in all
four subunits of the A95G structure [40]. The possibil-
ity therefore exists that the FMNs are present in a
one-electron-reduced form. However, the collection of
further evidence supporting the tentative semiquinone
state of the FMN in the structure was beyond the
scope of the present study.
Table 1. Kinetic constants of FMN reduction by a-hydroxy acid
substrate in wild-type and A95G avLOX determined in enzyme-
monitored anaerobic stopped flow experiments using 50 mM
potassium phosphate buffer (pH 6.5) at 20 °C. ND, not determined.
Enzyme
Substrate
Wild-type
A95G
L-Lactate
At 20 °C
kred (sꢁ1 a
Kd (mM)b
At 4 °C
)
274 ꢀ 11
1.0 ꢀ 0.2
235 ꢀ 13
7.4 ꢀ 1.5
kred1 (sꢁ1 c,d
kred2 (sꢁ1 c,d
)
138 ꢀ 6
72 ꢀ 3
142 ꢀ 16
27 ꢀ 1
)
Kd1 (mM)
0.35 ꢀ 0.10
1.0 ꢀ 0.1
5.0 ꢀ 1.0
0.8 ꢀ 0.2
Kd2 (mM)
(S)-Mandelic acid
kred (sꢁ1
Kd (mM)
)
0.08 ꢀ 0.01
5.3 ꢀ 0.5
0.6 ꢀ 0.2
ND
(S)-2-Hydroxybutyric acid
kred (sꢁ1
Kd (mM)
)
9.0 ꢀ 0.5
0.9 ꢀ 0.2
398 ꢀ 28
2.3 ꢀ 0.4
a
From Scheme 1, the measured kred is an expression of k3 and k5.
Equality of k3 and k5 is inferred from the fact that FMN reduction
occurred in a single exponential phase. Therefore, the value for k3
and k5 is 274 sꢁ1 (= kred).
b
From Scheme 1, Kd is the substrate dissociation constant defined
as the ratio of k2/k1.
c
From Scheme 1, kred1 is k3 and kred2 is k5.
Discussion
d
Fits of increase and decrease of A530 for the reaction at saturat-
ing L-lactate concentration (20 mM) gave similar (ꢀ15%) values of
kred1 and kred2, respectively.
Determination of the mechanistic implications of
Ala/Gly residue variation at the active site of
568
FEBS Journal 282 (2015) 562–578 © 2014 FEBS

T. Stoisser et al.
Ala95-to-Gly substitution in L-lactate oxidase
A
D
B
E
Fig. 6. Kinetic analysis of re-oxidation of
reduced FMN in avLOX wild-type and
A95G by different electron acceptors (A-E)
in 50 mM potassium phosphate buffer (pH
6.5) at 20 °C. Enzyme (10 lM) was
C
reduced anaerobically by L-lactate (20 lM)
and then mixed with electron acceptor at
the final concentration indicated.
Absorbance traces at 455 nm were fitted
with single exponentials to obtain kobs. Red
circles show A95G; black circles show
wild-type avLOX. Error bars indicate the
SD from three independent experiments.
Straight-line fits of the data are shown and
used to obtain kox (Table 2).
a-hydroxy acid-oxidizing flavoenzymes promises
a
obtained in the present study and in previous studies
[21,25], in terms of changes in active-site conformation
induced. Detailed kinetic analysis of the reoxidation
of wild-type and A95G avLOX revealed previously
unknown effects of the mutation on modulating
enzyme specificity for the reaction with different elec-
tron acceptor substrates. These findings are relevant in
connection to the oxidase/dehydrogenase dichotomy in
a-hydroxy acid-oxidizing flavoenzymes. Structures of
relevant Ala-to-Gly (flavocytochrome b2) [43] and Gly-
to-Ala (mandelate dehydrogenase) [44] mutants of
dehydrogenase-type members of the enzyme family are
available, although the structural consequences of the
same mutation in an oxidase from this family have not
been reported.
deeper insight into the role of the microenvironment
of the FMN cofactor’s reactive C4a-N5 locus on the
efficiency of each catalytic step of the enzymatic reac-
tion. Reactivity in the oxidative half-reaction with O2
compared to reactivity with alternative electron ac-
ceptors distinguishes oxidase from dehydrogenase-
type members of the family and is therefore
considered particularly relevant. Identification of the
structural principles of O2 reactivity is of high cur-
rent interest in mechanistic flavoenzyme research in
general [17,18,41,42].
The replacement of Ala95 by Gly in avLOX
has already been described in the literature and its
functional consequences have been analyzed in consid-
erable detail [21,25]. These prior studies notwithstand-
ing, the present study represents a significant advance
via A95G avLOX structure determination, which, for
the first time, enables an interpretation of the kinetic
consequences of the site-directed substitution, as
Reaction pathway of L-lactate oxidation by O2
Kinetic data for L-lactate oxidation by wild-type and
A95G avLOX at low (4 °C) and ambient (20 °C)
FEBS Journal 282 (2015) 562–578 © 2014 FEBS
569

Ala95-to-Gly substitution in L-lactate oxidase
T. Stoisser et al.
Table 2. Kinetic constants of FMN reoxidation by different electron acceptors in wild-type and A95G avLOX determined at 20 °C in enzyme-
monitored stopped flow experiments using 50 mM potassium phosphate buffer (pH 6.5). Enzyme (10 lM) was reduced by L-lactate (20 lM)
in the first reaction step performed anaerobically. ND, not determined.
Wild-type
kox (Mꢁ1ꢂsꢁ1 d,e
A95G
a
Em (V)
Electron acceptor
)
2,5-Dichloro-1,4-benzoquinone
2-Methyl-1,4-benzoquinone
2,6-Dichlorophenolindophenol
2,5-Dimethyl-1,4-benzoquinone
O2/H2O2
+0.310
+0.251
+0.217b
+0.176
+0.281c
2.7 ꢀ 0.2 9 104
3.7 ꢀ 0.2 9 104
1.3 ꢀ 0.1 9 105
5.9 ꢀ 0.4 9 104
1.8 ꢀ 0.1 9 106
8.9 ꢀ 0.5 9 104
1.6 ꢀ 0.1 9 105
1.0 ꢀ 0.1 9 106
ND
3.3 ꢀ 0.2 9 105
a
Em is the midpoint reduction potential (30 °C, pH 7.0) from Rao and Hayon [36].
Em is the midpoint reduction potential (30 °C, pH 7.0) from Dawson et al. [37].
Em is the midpoint reduction potential (30 °C, pH 7.0) from Mattevi [17].
b
c
d
kox is the second-order rate constant of FMN re-oxidation determined from linear dependencies of pseudo-first-order stopped flow rate
constants, as shown in Fig. 3. Standard errors are from straight-line fits of the data.
e
From Scheme 1, kox equals k7.
Table 3. Crystallographic data collection and refinement statistics.
pyruvate was not significant under the conditions used,
Values given in parentheses are for data in the high-resolution bin.
which is consistent with the literature [21,25]. (Note
that microscopic reversibility of a-hydroxy acid/a-keto
acid interconversion is not called into question and
was clearly shown for reactions catalyzed by mande-
late dehydrogenase [45] and flavocytochrome b5 [46].
However, oxidation of flavocytochrome b5 by pyruvate
occurred with a rate constant of only 0.08 sꢁ1). There-
fore, the apparent turnover number of the lactate oxid-
ases at air-saturated reaction conditions (kcat_app) is
described with rate constants from Scheme 1 [19,20]:
Data collection
Unit cell dimensions
ꢀ
a, b, c (A)
b (°)
122.59, 124.36, 106.89
124.29
ꢀ
Resolution range (A)
Completeness (%)
Rmerge
50.00–1.65 (1.68–1.65)
98.2 (95.1)
5.9 % (37.1 %)
581 586/155 981
Observations/unique
reflections
Mean I/r (I)
26.4 (4.1)
Refinement statistics
Rwork (%)
15.2
18.0
kcat
¼ 1=½ðk3 þ k5Þ=ðk3k5Þ þ 1=ðk7½O2ꢄÞꢄ
ð1Þ
app
Rfree (%)
Rfree test set
7791
Using a value of 250 lM for [O2] and applying the
relevant rate constants from Tables 1 and 2, the calcu-
lated kcat_app of wild-type avLOX is 105 sꢁ1, which is
in useful agreement with the kcat_app value of 88 sꢁ1
obtained from steady-state kinetic analysis conducted
at the same [O2]. For the A95G enzyme, the calculated
kcat_app of 48 sꢁ1 was somewhat higher than the exper-
Deviations from ideal geometry
ꢀ
Bond lengths (A)
0.013
1.54
Bond angles (°)
2
ꢀ
Average B values (A )
Main-chain atoms
Side-chain atoms
FMN
13.1
15.3
10.3
19.3
14.6
imental value of 30 sꢁ1
The Km for oxygen and L-lactate can also be calcu-
lated:
.
Water molecules (920)
All atoms
Ramachandran plot (%)
Most favoured
Additionally allowed
Outliers
96.6
3.0
0.4
KmðO2Þ ¼ k3k5=½k7=ðk3 þ k5Þꢄ
ð2Þ
temperature suggest a common reaction pathway for
both enzymes, as shown in Scheme 1. Evidence sug-
gesting that reduced lactate oxidases were not detecta-
bly oxidized by excess of pyruvate under anaerobic
conditions at 20 °C indicated that, practically, k4 ꢃ 0.
Model fits to anaerobic stopped flow curves at 4 °C
(Fig. 4) also suggested that the reverse reaction from
Kmðl-lactateÞ ¼ ½ð1=k1Þ þ k2k5=ðk1k3k5Þꢄ=ð1=kcat appÞ
ð3Þ
Values of 69 and 356 lM are thus obtained for Km
(O2) of wild-type and A95G avLOX, respectively. The
calculated Km (L-lactate) values are 0.71 and 1.5 mM
570
FEBS Journal 282 (2015) 562–578 © 2014 FEBS

T. Stoisser et al.
Ala95-to-Gly substitution in L-lactate oxidase
between 350 homologous Ca atoms in these monomers
ꢀ
is 1.23 A and so they can therefore be considered to
be similar overall. Notably, GOX has an alanine
(Ala79) at the equivalent position to the A95G muta-
tion in avLOX and the conformation is similar to the
alanine in the wild-type enzyme. Moreover, the flavin
ring found in GOX appears to be planar, which has
the potential to further restrict binding of larger elec-
tron acceptors. Although there is no substrate within
the structure of the GOX active site, the residues that
potentially interact are in a conformation creating a
generally larger active site than the pyruvate-bound
A95G active site (Fig. 8B). There are several differ-
ences between the active sites that may be able to
explain the promiscuous reactivity of GOX. These
include the displacement of the Tyr146 (GOX: Tyr129)
ꢀ
side chain by 1.8 A, the Arg181 (Arg164) side chain
ꢀ
ꢀ
by 4.6 A and the His265 (His254) side chain by 1.1 A
away from the pyruvate bound in the avLOX wild-
type active site. Another gain of steric volume in GOX
results from the lack of residues homologous to
Leu211 and Tyr215, which approach the pyruvate
Fig. 7. The Ca trace of the
A subunit of A95G avLOX (blue)
showing the position of the FMN (yellow), the pyruvate (green),
the sites of the inert mutations (T102A, S163G, G232A and R255A,
pink) and the active site A95G mutation (red). The stretch of
residues from 199–217 with a higher mobility in some subunits is
indicated in light blue.
ꢀ
within 4.0 and 2.5 A respectively. It can be concluded
from this comparison that there is no common single
structural feature or sequence signature that facilitates
the binding of larger electron acceptors such as DCIP
but rather a general increase in the steric volume of
the active site. Therefore, this provides a plausible
explanation as to why the A95G variant rather than
wild-type avLOX exhibits trend resemblance with
GOX in electron acceptor substrate specificities. We
also note that, consistent with a scenario where elec-
tron acceptor reactivity at the catalytic center is mainly
governed by access to the FMN site, the re-oxidation
rate constant k7 did not show an apparent correlation
with the intrinsic electronic reactivity (Em) (Table 2) of
the electron acceptor substrate used.
for wild-type and A95G avLOX, respectively, and can
be compared to the corresponding values of
0.5 ꢀ 0.1 mM and 1.0 ꢀ 0.2 mM obtained from
steady-state kinetic analysis.
Comparison of active-site structures and electron
acceptor reactivities of A95G avLOX and
glycolate oxidase
Similar to A95G avLOX, GOX is also able to effi-
ciently utilize electron acceptors different from and lar-
ger than molecular oxygen. For example, GOX from
pig liver was reported to exhibit a two-fold higher
turnover number with DCIP than O2 [35]. The kox
(O2) of GOX from spinach (8.5 9 104
Reactivity with a-hydroxy acid substrates in the
reductive half-reaction
ꢁ1
M
ꢂsꢁ1) [8] is
21-fold lower than the corresponding kox for wild-type
avLOX (Table 2). The Ala95-to-Gly substitution in
avLOX therefore creates an oxidase, which in terms of
kox (O2) is intermediate between wild-type avLOX and
GOX. Because reactivity with DCIP compared to O2
appears to have been strongly enhanced in GOX com-
pared to wild-type avLOX and the same trend in elec-
tron acceptor reactivities is observed for the A95G
mutant, a structural comparison between GOX and
A95G avLOX was suggested.
Specificity for substituents at position 2 in a-hydroxy
acid substrates is assumed to be determined by resi-
dues surrounding the methyl group in L-lactate, which
includes Ala95 [21,25,43]. Comparison of the A95G
avLOX structure with the structure of wild-type
enzyme shows that the mutation of this residue to gly-
cine increases the volume available at this position by
removing the steric bulk of the alanine side chain and
allowing the main chain to relax slightly away from
the L-lactate binding site. This extra volume potentially
allows the substrate to rotate slightly to permit a
re-direction of the substituent at position 2. Moreover,
The crystal structure of spinach GOX is available
(PDB code: 1GOX) [3]. A structural comparison of
these 36% identical enzymes reveals that the r.m.s.d.
FEBS Journal 282 (2015) 562–578 © 2014 FEBS
571

Ala95-to-Gly substitution in L-lactate oxidase
T. Stoisser et al.
A
B
Fig. 8. A comparison of the active sites
from wild-type avLOX, the A95G variant
and GOX. (A) An overlap of the A subunit
(lavender) of the wild-type avLOX
structure (PDB code: 2E77) and the A95G
variant (colored by atom type) shows the
additional steric volume available for DCIP
binding as a result of the smaller side
chain at position 95, the slightly altered
main-chain conformation at that position
and the lack of flavin planarity. Leu211 is
not present in the wild-type structure as a
result of disorder. (B) An overlap of the
A95G variant (colored by atom type;
pyruvate in red) and GOX (lavender; PDB
code: 1GOX) active-site residues
surrounding the A95G-bound pyruvate
(red) shows the increased steric volume
available for electron acceptor binding in
GOX versus A95G.
it must be emphasized that the sp3?sp2 hybridization
change at the reactive C2 during substrate oxidation
involves substantial reorientation of the C2 substitu-
ents when tetrahedral substrate converts via a more
planar transition state into planar product. Geometri-
cal requirements of the catalytic reaction coordinate
for oxidation of a-hydroxy acid substrates larger than
L-lactate are likely to be more easily accommodated by
the more spacious binding pocket of A95G avLOX
compared to wild-type enzyme. This could explain the
mutant enzyme’s markedly increased reactivities (kred
)
with (S)-2-hydroxybutyric acid and (S)-mandelic acid
(Table 1) and generally its higher specificity for larger
a-hydroxy acid substrates. It also may account for the
decreased efficiency (Table 1) with L-lactate because
the substrate would have fewer structural constraints
572
FEBS Journal 282 (2015) 562–578 © 2014 FEBS

T. Stoisser et al.
Ala95-to-Gly substitution in L-lactate oxidase
implies that there is a conformational change required
for the binding and release of pyruvate, as well as for
the binding and release of similarly sized electron
acceptors. According to the proposed reaction path-
way of avLOX (Scheme 1), pyruvate must also dissoci-
ate so that O2 can react with reduced flavin. The
region of residues near the range 199–204 is a good
candidate for participation in such a conformational
rearrangement as a result of its high apparent mobil-
ity. Various residues in this segment are often missing
in crystal structures and those that are present usually
exhibit high temperature factors [3–7,27–29]. In the
A95G mutant subunits, certain residues in this region
cannot be observed in the electron density: A, 203–
204; B, 199–204; C, 203–204; D, 203–204. Although
ꢀ
this general region is > 10 A distant from the pyru-
vate binding site, it is tempting to speculate that it
may function as a hinge region, facilitating the move-
ment of a short helix composed of residues 210–217,
which buries the active site from solvent (Fig. 7). This
is supported by the high disorder observed in residues
214–217 in subunit C, which prevented the fitting of
this segment. Other published structures show similar
gaps in this region as well, although some encompass
a larger range of 196–213, which includes a portion of
the sequestering helix. We also note that, at 4 °C,
A95G variant released pyruvate 2.7-fold more slowly
than wild-type avLOX did (Table 1).
In flavoenzyme oxidases [18], O2 is assumed to
approach the reactive C4a-N5(H) locus from the edge
of the plane in which the FMN molecule lies. In
monooxygenases, by contrast, an attack of O2 is con-
sidered to occur from one of the sides (re or si face) of
the cofactor plane and typically involves a well-defined
O2 binding site [18]. Crystal structures of wild-type
and A95G lactate oxidases appear to be compatible
with both ‘edge-on’ and ‘face-on’ approaches of the O2
substrate to the FMN. However, a re-face interaction
appears to be unlikely because of steric hindrance
from the methyl side chain of Ala95 in wild-type
enzyme (Fig. 8A) and bound water in the A95G vari-
ant (Figs 1 and 8A), as well as additional protection
provided by Pro93 and Ile94. The structures also
reveal the absence of a discrete binding pocket for O2
in the active site, consistent with oxidase-only reaction
pathways in both enzymes. To address the decrease in
O2 reactivity in A95G variant compared to wild-type
avLOX, it is relevant to consider changes in the local
environment of the C4a-N5 locus resulting from the
site-directed substitution. The water molecule accomo-
dated in the A95G variant reduces the hydrophobicity
around the putative O2 reacting site compared to the
wild-type lactate oxidase that has the methyl group of
Fig. 9. Unbiased Fo – Fc difference density contoured at 2r
superimposed on the final refined model showing the nonplanarity
of the flavin ring. Also shown for reference are the site of the
A95G mutation and the bound pyruvate.
in binding and has the potential to bind in a more dis-
ordered fashion. Modeling experiments indicate a pos-
sible steric conflict between the substituent at position
2 and the side- and/or main-chain atoms of Tyr124
and, to a lesser extent, the side chain of Tyr191. The
A95G avLOX crystal structure generally supports the
interpretation of kinetic evidence on the A95G variant
previously reported by Yorita et al. [21,25].
Reactivity with O2 and alternative electron
acceptors in the oxidative half-reaction
The bound FMN and pyruvate molecules are almost
completely sequestered from the solvent with calcu-
2
lated accessibilities ranging from 2.2 to 5.2 A for the
ꢀ
2
FMN and from 0.4 to 3.0 A for the pyruvate. This
ꢀ
FEBS Journal 282 (2015) 562–578 © 2014 FEBS
573

Ala95-to-Gly substitution in L-lactate oxidase
T. Stoisser et al.
Ala95 in an approximately analogous position. For
different flavoenzymes, an optimum combination of
hydrophobicity and steric constraint creates a physi-
cally confined microenvironment that can strongly
facilitate the ability of O2 to rapidly react with the
reduced flavin [17,18]. A lowering of polarity of the
flavin environment in (S)-mandelate dehydrogenase
upon substitution of Gly81 by Ala was suggested to
account for the slight increase in O2 reactivity in the
mutant compared to wild-type enzyme [44].
with the lactate oxidases in an apparently strictly
bimolecular fashion (Fig. 6) supports the notion that
specific binding of the acceptor to a discrete binding
site in the protein may not have been involved in the
electron transfer process from the reduced flavin. The
electron transfer between the flavin and the heme
group in flavocytochrome b2 presents a relevant exam-
ꢀ
ple as it proceeds over a long distance (9.7 A). It does
not involve direct contact between the two chromo-
phores but appears to be aided by protein residues
[1,12].
The water molecule in A95G avLOX also forms a
hydrogen bond with the flavin N5. Studies of other fla-
vin-dependent oxidases/monooxygenases strongly sup-
port the idea that hydrogen bonding with the
cofactor’s N5 atom represents an important strategy
for the control of O2 reactivity [18]. In pyranose oxi-
dase [47], for example, the substitution of a threonine
that interacts with the flavin N5 resulted in removal of
a characteristic C4a-hydroperoxyflavin stabilization,
such that flavin oxidation and H2O2 formation
occurred without any detectable intermediate, as is
normally the case for oxidases. Kinetic isotope effect
studies of the pyranose oxidase also suggested that
N5-H bond cleavage controls the process of H2O2
elimination during flavin oxidation [48]. Hydrogen
bonding to N5(H) would consequently disfavor rapid
and direct formation of H2O2. Thus, installing a
hydrogen bond to flavin N5 from water is relatively
less effective in altering the reaction with O2 in avLOX
than removing the native hydrogen bond to flavin N5
from threonine in pyranose oxidase, and there is no
evidence for C4a-hydroperoxyflavin intermediate for-
mation during O2 reduction by A95G avLOX. How-
ever, hydrogen bonding from water might protect the
N5-H proton from proton transfer reactions (e.g. pro-
tonation of the departing hydrogen peroxide anion)
and this could slow down the overall process of H2O2
production from O2 in A95G variant compared to
wild-type avLOX.
Conclusions
Characteristic Ala/Gly residue variation in the catalytic
centers of a-hydroxy acid oxidases has attracted signifi-
cant attention in structure–function studies of these
enzymes, including avLOX. The Ala or Gly adopts a
structural position close to the C4a-N5 locus at the
re face of the FMN cofactor and therefore contributes
to the microenvironment of the reaction site. In av-
LOX, the A95G variant has already been subject of
ꢀ
mechanistic investigations. The 1.65-A crystal structure
of A95G avLOX bound with pyruvate provides a
detailed interpretation of different kinetic consequences
of the site-directed substitution: decreased reactivity
with molecular oxygen; increased specificity for reduc-
tion of alternative electron acceptors; and increased
specificity for oxidation of a-hydroxy acid substrates
larger than L-lactate. Generally, from the structural
results, it may be concluded that the increased ability
of the mutant to utilize larger donor or acceptor sub-
strates is a consequence of the increased steric volume
arising from the loss of the alanine methyl group. A
comparison of the LOX-A95G structure with the
homologous wild-type GOX structure shows that a
glycine at this position is not absolutely necessary for
the acceptance of alternative electron acceptors such as
DCIP as a substrate for this family of enzymes. Low-
ered O2 reactivity in A95G variant compared to wild-
type avLOX might be a consequence of a hydrogen
bond from bound water to flavin N5 in the mutant lac-
tate oxidase, which could protect the N5-H proton in
reduced flavin from undergoing proton transfer reac-
tions directly conducive to H2O2 elimination in the
overall process of O2 reduction.
With respect to the question of how larger electron
acceptors might bind to the active site of the A95G
mutant, attempts to manually model DCIP into the
binding site were made with particular attention
focused on the site of the mutation. Ultimately, it was
considered that there were too many possible modes of
binding, particularly when possible conformational
changes in active site residues were considered. How-
ever, there is also the possibility that electron transfer
from reduced flavin to DCIP takes place assisted by
the protein matrix. In that case, a relatively unspecific
interaction between DCIP and the protein might be
sufficient for flavin reoxidation to occur. Evidence that
DCIP and other alternative electron acceptors reacted
Experimental procedures
Materials
The a-hydroxy acids (free acid or sodium salt) and electron
acceptors used were of the highest quality available at
574
FEBS Journal 282 (2015) 562–578 © 2014 FEBS

T. Stoisser et al.
Ala95-to-Gly substitution in L-lactate oxidase
Sigma-Aldrich (Deisenhofen, Germany). DCIP was
obtained as sodium salt. Solutions of electron acceptors
were prepared fresh and kept in the dark prior to use. av-
LOX wild-type was a kind gift from Roche Diagnostics
(Penzberg, Germany). The lyophilized protein was dis-
solved to a concentration of 10 mgꢂmLꢁ1 in 50 mM potas-
sium phosphate buffer (pH 6.5). The avLOX migrates as a
single protein band in SDS/PAGE and was therefore used
without further purification. All other materials have been
described previously [49].
(PreSens GmbH, Regensburg, Germany) connected to a
miniaturized lock-in amplifier (MICROX TX3), as also
obtained from PreSens GmbH. Reactions were performed
in magnetically stirred glass vials (total volume of 4.5 mL;
300 r.p.m. agitation rate) sealed air-tight using a septum
through which the electrode could be fitted into the system.
The total reaction volume was 1.5 mL and air-saturated
buffer was used at 37 °C. The L-lactate concentration was
50 mM and fully saturating at steady-state. A 50 mM potas-
sium phosphate buffer (pH 6.5) was used. Reactions were
started by injecting a suitably concentrated enzyme solution
(10 lL) through the septum and were followed for approxi-
mately 3 min. Experimental reaction rates varied between
approximately 5.0 and 50 lMꢂminꢁ1. The standard error of
r_H2O2 and r_O2 determined in identical replicates (n = 5)
was approximately 5% and 7%, respectively. We found
that r_H2O2 was equal to r_O2 within limits of the experi-
mental error.
Construction of A95G avLOX, enzyme production
and purification
Plasmid vector pLO-1, which harbors the avLOX gene
under control of the tac promoter, was used as template
for mutagenesis by a two-stage PCR protocol as reported
previously [49]. The two mutagenic oligonucleotide primers
used were:
Because of its comparably lower standard error, r_H2O2
was used for determination of Michaelis–Menten parame-
ters. Reaction rates were measured at 10 or more a-
hydroxy acid substrate concentrations in the concentration
range indicated. Unless noted otherwise, a temperature of
37 °C was used and air-saturated buffer conditions were
applied.
50-CCAATTGGTGCCCATGGTTTAGCTCACGCTAC
TAAAGAAGCTGG-30 and 50-AGCGTGAGCTAAACCA
TGGGCACCAATTGGGGCCATGATGAATGG-30.
The mismatched bases are underlined. The desired muta-
tion was verified by sequencing.
Enzyme production was carried out in E. coli BL21
(DE3) containing pLO-1 vector encoding wild-type or
A95G avLOX. Shaken flask and bioreactor cultivations of
the strains, processing of E. coli cells and protein purifica-
tion were carried out by protocols exactly as described
previously [49]. Briefly, soluble extract obtained by high-
pressure cell disruption was purified by ammonium sulfate
precipitation followed by hydrophobic interaction and
anion exchange chromatography. Pooled fractions were de-
salted and concentrated to approximately 20 mgꢂmLꢁ1 in
50 mM potassium phosphate buffer (pH 7.0), using 10-kDa
cut-off centrifugal concentrator tubes (Vivaspin 20; Sarto-
Absorbance spectra were recorded with a Varian Cary
50 Bio UV-Vis spectrophotometer (Agilent, Santa Clara,
CA, USA) using Hellma quartz cuvettes (Hellma Analytics,
€
Mullheim, Germany) of 10-mm light path. Aꢁn1 extinction
coefficient of protein bound FMN of 12 500 M ꢂcmꢁ1 was
determined using the protocol reported previously [50].
Enzyme turnover numbers (kcat) are based on the molar
concentrations of FMN.
Reaction with DCIP and O2 as competing
electron acceptors
€
rius Vivascience, Gottingen, Germany). Enzymes were
A quartz cuvette closed with a Teflon stopper and fitted
with the above-described microoptode was used. The cuv-
ette was placed in the Varian spectrophotometer so that
the consumption of DCIP (change in A600) and O2 (mic-
rooptode) could be measured simultaneously over time
once the reaction had started via injection of enzyme solu-
tion (10 lL). The L-lactate concentration was 50 mM. Air-
saturated potassium phosphate buffer ([O2] ꢅ 200 lM; pH
6.5) was used, and the DCIP concentration was also
200 lM. Reactions were carried out at 37 °C.
stored at ꢁ70 °C.
Assays
Protein was measured with Roti-Quant reagent (Carl-Roth,
Karlsruhe, Germany) using BSA as reference. The standard
LOX activity assay was based on in situ peroxidase-coupled
detection of the H2O2 released during L-lactate (50 mM)
oxidation at pH 6.5. It was carried out at 37 °C as
described previously [49]. To examine the degree of uncou-
pling (Scheme 1) in catalytic reactions of wild-type and
A95G avLOX, we compared the H2O2 formation rate
(r_H2O2) with the corresponding O2 consumption rate
(r_O2), considering that, in a fully uncoupled reaction, the
absolute value of the rate ratio (|r_H2O2/r_O2|) is unity.
r_H2O2 was measured using the coupled peroxidase assay.
r_O2 was measured using a fiber-optic oxygen microoptode
Stopped flow experiments
An Applied Photophysics (Leatherhead, Surrey, UK)
model SX17MV stopped-flow analyzer equipped with a
diode array absorbance detector was used to perform
enzyme-monitored turnover experiments. Alternatively,
FEBS Journal 282 (2015) 562–578 © 2014 FEBS
575

Ala95-to-Gly substitution in L-lactate oxidase
T. Stoisser et al.
turnover was determined from absorption measurements at
a single wavelength (450 nm, 530 nm) recorded with a pho-
tomultiplier detector. Diode array data were recorded every
0.1 ms and the instrument deadtime was below 0.5 ms
under the conditions used. Data from diode array measure-
ments (280–700 nm) were deconvoluted using PRO-K soft-
structure. PROCHECK analysis was performed as implemented
in CCP4I (http://www.ccp4.ac.uk/ccp4i_main.php). WINCOOT
07-PRE 1 [53] was used for fitting and modeling.
Acknowledgements
ware (Applied
Photophysics).
Anaerobic reaction
This work was funded within the Austrian COMET
Program under the auspices of the Austrian Federal
Ministry of Transport, Innovation and Technology
(bmvit); the Austrian Federal Ministry of Economy,
Family and Youth (bmwfj); and the State of Styria
(Styrian Funding Agency SFG). The COMET Pro-
gram is managed by the Austrian Research Promotion
Agency (FFG). Professor Marco Fraaije (Groningen,
The Netherlands) is thanked for introducing T.S. to
anaerobic stopped flow methodology. EU COST
Action CM0701 ‘CASCAT’ is thanked for providing
financial support.
conditions were established by flushing all solutions exhaus-
tively with nitrogen. Traces of oxygen were removed by
adding 10 mM glucose and 200 nM glucose oxidase (Roche
Diagnostics). All parts of the stopped-flow apparatus were
rinsed with glucose/glucose oxidase solution. A 50 mM
potassium phosphate buffer (pH 6.5) was used. Each mea-
surement was repeated five times or more and the stopped-
flow traces were averaged.
The reductive half-reaction was followed by collecting
spectra with the diode array detector or by measuring the
decrease A450 after mixing anaerobically the enzyme (final
concentration of 5–20 lM) with varied concentrations of
a-hydroxy acid substrate [L-lactate, (S)-mandelate, (S)-2-hy-
droxybutyrate]. The standard temperature was 20 °C but
oxidation of L-lactate was also monitored at 4 °C. The oxida-
tive half-reaction was followed at 20 °C by mixing enzyme,
which had been anaerobically reduced by the double molar
equivalent of L-lactate, with varied concentrations of electron
acceptor substrate (O2, DCIP, 2,5-dichloro-1,4-benzoqui-
none, 2,5-dimethyl-1,4-benzoquinone or methyl-1,4-benzo-
quinone). A450 was recorded with the photomultiplier
detector, except for reaction with DCIP where diode array
data were obtained. Single wavelength traces were fitted to
the appropriate single- or double-exponential decay/rise
function to obtain pseudo-first-order reaction rate constanst
(kobs).
Author contributions
B.N., S.L. and D.K.W. designed the research. T.S.
performed the biochemical experiments and analyzed
data with S.L. and B.N. Crystallization was performed
by D.R. D.R. and D.K.W. determined the structure.
B.N. and D.K.W. wrote the paper.
References
1 Palfey BA & Massey V (1998) Flavin-dependent
enzymes. In Comprehensive Biological Catalysis: A
Mechanistic Reference, Volume III Radical Reactions
and Oxidation/Reduction (Sinnott M, ed.), pp. 83–154.
Academic Press Limited, London, UK.
Crystallization, structure determination and
refinement
^
2 Le KHD & Lederer F (1991) Amino acid sequence of
long-chain a-hydroxy-acid oxidase from rat kidney, a
member of FMN-dependent a-hydroxy-acid-oxidizing
enzymes. J Biol Chem 266, 20877–20881.
A95G avLOX was crystallized using hanging drop vapor
diffusion in air atmosphere at 20 °C. One microliter of pro-
tein solution at 10 mgꢂmLꢁ1 in 50 mM potassium phos-
phate (pH 7.0) was mixed with 1 lL of reservoir solution
of 30% polyethylenglycol in 50 mM Tris-HCl buffer (pH
8.0) with the addition of 50 mM pyruvate. Crystals were
harvested in 85% (v/v) well solution with 15% (v/v) ethyl-
ene glycol. X-ray measurements were carried out at the
Stanford Synchrotron Radiation Laboratory (Menlo Park,
CA, USA).
3 Lindquist Y (1989) Refined structure of spinach
ꢀ
glycolate oxidase at 2 A resolution. J Mol Biol 209,
151–166.
4 Xia Z & Mathews FS (1990) Molecular structure of
ꢁ
ꢀ
flavocytochrome b2 at 2.4 A resolution. J Mol Biol 212,
837–863.
5 Sukumar N, Dewanti AR, Mitra B & Mathews FS
(2004) High resolution structures of an oxidized and
reduced flavoprotein. The water switch in a soluble
form of (S)-mandelate dehydrogenase. J Biol Chem 279,
3749–3757.
Molecular replacement using PHASER [51] was conducted
to determine rotation and translation using the avLOX
wild-type structure (PDB code: 2DU2) as a search model.
Appropriate changes in sequence were made manually to
the structure, which was refined using REFMAC5 [52] to
reach a final R of 15.4% and Rfree of 18.0%. Iterations of
refinement and manual refitting were used to identify and
add waters, ethylene glycol and pyruvate molecules to the
6 Umena Y, Yorita K, Matsuoka T, Kita A, Fukui K &
Morimoto Y (2006) The crystal structure of L-lactate
ꢁ
ꢀ
oxidase from Aerococcus viridans at 2.1 A resolution
reveals the mechanism of strict substrate recognition.
Biochem Biophys Res Commun 350, 249–256.
576
FEBS Journal 282 (2015) 562–578 © 2014 FEBS

T. Stoisser et al.
Ala95-to-Gly substitution in L-lactate oxidase
^
7 Cunane LM, Barton JD, Chen Z, Le KH, Amar D,
Lederer F & Mathews FS (2005) Crystal structure
analysis of recombinant rat kidney long chain hydroxy
acid oxidase. Biochemistry 44, 1521–1531.
site-directed mutagenesis of alanine 95 to glycine. J Biol
Chem 271, 28300–28305.
22 Dewanti AR, Xu Y & Mitra B (2004) Role of glycine
81 in (S)-mandelate dehydrogenase from Pseudomonas
putida in substrate specificity and oxidase activity.
Biochemistry 43, 10692–10700.
8 Macheroux P, Massey V & Thiele DJ (1991) Expression
of spinach glycolate oxidase in Saccharomyces
cerevisiae: purification and characterization.
Biochemistry 30, 4612–4619.
23 Daff S, Manson FD, Reid GA & Chapman SK (1994)
Strategic manipulation of the substrate specificity of
Saccharomyces cerevisiae flavocytochrome b2. Biochem J
301, 829–834.
9 Maeda-Yorita K, Aki K, Sagai H, Misaki H & Massey
V (1995) L-lactate oxidase and L-lactate
monooxygenase: mechanistic variations on a common
structural theme. Biochimie 77, 631–642.
24 Leferink NG, Fraaije MW, Joosten HJ, Schaap PJ,
Mattevi A & van Berkel WJ (2009) Identification of a
gatekeeper residue that prevents dehydrogenases from
acting as oxidases. J Biol Chem 284, 4392–4397.
25 Yorita K, Janko K, Aki K, Ghisla S, Palfey B &
Massey V (1997) On the reaction mechanism of
L-lactate oxidase: quantitative structure-activity analysis
of the reaction with para-substituted L-mandelates. Proc
Natl Acad Sci U S A 94, 9590–9595.
10 Lockridge O, Massey V & Sullivan PA (1972)
Mechanism of action of the flavoenzyme lactate
oxidase. J Biol Chem 247, 8097–8106.
11 Lederer F, Belmouden A & Goudry M (1996) The
chemical mechanism of flavoprotein-catalysed a-
hydroxy acid dehydrogenation: a mutational analysis.
Biochem Soc Trans 24, 77–83.
12 Boubacar AKO, Pethe S, Mahy J-P & Lederer F (2007)
Flavocytochrome b2: reactivity of its flavin with
molecular oxygen. Biochemistry 46, 13080–13088.
13 Sobrado P & Fitzpatrick PF (2003) Solvent and
primary deuterium isotope effects show that lactate CH
and OH bond cleavages are concerted in Y254F
flavocytochrome b2, consistent with a hydride transfer
mechanism. Biochemistry 42, 15208–15214.
26 Yorita K, Misaki H, Palfey B & Massey V (2000) On
the interpretation of quantitative structure-function
relationship data for lactate oxidase. Proc Natl Acad
Sci U S A 97, 2480–2485.
27 Li SJ, Umena Y, Yorita K, Matsuoka T, Kita A,
Fukui K & Morimoto Y (2007) Crystallographic study
on the interaction of L-lactate oxidase with pyruvate at
ꢀ
1.9 Angstrom resolution. Biochem Biophys Res Commun
14 Lehoux IE & Mitra B (1999) (S)-mandelate
dehydrogenase from Pseudomonas putida: mechanistic
studies with alternate substrates and pH and kinetic
isotope effects. Biochemistry 38, 5836–5848.
15 Fitzpatrick PF (2004) Carbanion versus hydride
transfer mechanisms in flavoprotein-catalyzed
dehydrogenations. Bioorg Chem 32, 125–139.
16 Cao Y, Han S, Yu L, Qian H & Chen J-Z (2014) MD
and QM/MM studies on long-chain L-a-hydroxy acid
oxidase: substrate binding features and oxidation
mechanism. J Phys Chem B 118, 5406–5417.
17 Mattevi A (2006) To be or not to be an oxidase:
challenging the oxygen reactivity of flavoenzymes.
Trends Biochem Sci 31, 276–283.
358, 1002–1007.
28 Leiros I, Wang E, Rasmussen T, Oksanen E, Repo H,
Petersen SB, Heikinheimo P & Hough E (2006) The 2.1
ꢀ
A structure of Aerococcus viridans L-lactate oxidase
(LOX). Acta Crystallogr Sect F Struct Biol Cryst
Commun 62, 1185–1190.
29 Furuichi M, Suzuki N, Dhakshnamoorhty B,
Minagawa H, Yamagishi R, Watanabe Y, Goto Y,
Kaneko H, Yoshida Y, Yagi H et al. (2008) X-ray
structures of Aerococcus viridans lactate oxidase and its
complex with D-lactate at pH 4.5 show an a-
hydroxyacid oxidation mechanism. J Mol Biol 378,
436–446.
30 Minagawa H, Nakayama N, Matsumoto T & Ito N
(1998) Development of long life lactate sensor using
thermostable mutant lactate oxidase. Biosens
Bioelectron 13, 313–318.
18 Chayen P, Fraaije M & Mattevi A (2012) The
enigmatic reaction of flavins with oxygen. Trends
Biochem Sci 37, 372–380.
19 Sun W, Williams CH Jr & Massey V (1996) Site-
directed mutagenesis of glycine 99 to alanine in L-
lactate monooxygenase from Mycobacterium smegmatis.
J Biol Chem 271, 17226–17233.
31 Romero MR, Ahumada F, Garay F & Baruzzi AM
(2010) Amperometric biosensor for direct blood lactate
detection. Anal Chem 82, 5568–5572.
32 Hirano K, Yamato H, Kunimoto K & Ohwa M (2001)
Design of novel electron transfer mediators based on
indophenol derivatives for lactate sensors. Biosens
Bioelectron 17, 315–322.
20 Sun W, Williams CH Jr & Massey V (1997) The role of
glycine99 in L-lactate monooxygenase from
Mycobacterium smegmatis. J Biol Chem 272, 27065–
27076.
€
33 Muh U, Massey V & Williams CH Jr (1994) Lactate
21 Yorita K, Aki K, Ohkuma-Soyejima T, Kokubo T,
Misaki H & Massey V (1996) Conversion of L-lactate
oxidase to a long chain alpha-hydroxyacid oxidase by
monooxygenase. I. Expression of the mycobacterial
gene in Escherichia coli and site-directed mutagenesis of
lysine 266. J Biol Chem 269, 7982–7988.
FEBS Journal 282 (2015) 562–578 © 2014 FEBS
577

Ala95-to-Gly substitution in L-lactate oxidase
T. Stoisser et al.
€
34 Muh U, Williams CH Jr & Massey V (1994) Lactate
and its complex with two of its substrates. Acta
monooxygenase. III. Additive contributions of active
site residues to catalytic efficiency and stabilization of
an anionic transition state. J Biol Chem 269, 7994–
8000.
Crystallogr D Biol Cryst 65, 543–552.
45 Dewanti AR & Mitra B (2003) A transient intermediate
in the reaction catalyzed by (S)-mandelate
dehydrogenase from Pseudomonas putida. Biochemistry
42, 12893–12901.
35 Schuman M & Massey V (1971) Effect of anions on the
catalytic activity of pig liver glycolic acid oxidase.
Biochim Biophys Acta 227, 521–537.
46 Urban P, Alliel PM & Lederer F (1983) On the
transhydrogenase activity of baker’s yeast
36 Rao PS & Hayon E (1973) Experimental determination
of the redox potential of the superoxide radical.
Biochem Biophys Res Commun 51, 468–473.
37 Dawson RMC, Elliott DC, Elliott WH & Jones KM
(1986) Data for Biochemical Research. Oxford
University Press, NY.
flavocytochrome b2. Eur J Biochem 134, 275–281.
47 Pitsawong W, Sucharitakul J, Prongjit M, Tan TC,
Spadiut O, Haltrich D, Divne C & Chaiyen P (2010) A
conserved active-site threonine is important for both
sugar and flavin oxidations of pyranose 2-oxidase. J
Biol Chem 285, 9697–9705.
38 Sollner S, Deller S, Macheroux P & Palfey BA (2009)
Mechanism of flavin reduction and oxidation in redox-
sensing quinone reductase Lot6p from Saccharomyces
cerevisiae. Biochemistry 48, 8636–8643.
48 Sucharitakul J, Wongnate T & Chaiyen P (2011)
Hydrogen peroxide elimination from C4a-
hydroperoxyflavin in a flavoprotein oxidase occurs
through a single proton transfer from flavin N5 to a
peroxide leaving group. J Biol Chem 286, 16900–16909.
39 Senda T, Senda M, Kimura S & Ishida T (2009) Redox
control of protein conformation in flavoproteins.
Antioxid Redox Signal 11, 1741–1766.
€
49 Unterweger B, Stoisser T, Leitgeb S, Birner-Grunberger
R & Nidetzky B (2012) Engineering of Aerococcus
viridans L-lactate oxidase for site-specific PEGylation:
characterization and selective bioorthogonal
modification of a S218C mutant. Bioconj Chem 23,
1406–1414.
40 Røhr AK, Hersleth HP & Andresson KK (2010)
Tracking flavin conformations in protein crystal
structures with Raman spectroscopy an QM/MM
calculations. Angew Chem Int Ed 49,
2324–2327.
50 Macheroux P (1999) UV-visible spectroscopy as a tool
to study flavoproteins. In Methods in Molecular
Biology, vol 131, Flavoprotein Protocols (Chapman SK
& Reid GA, eds), pp. 1–7. Humana Press Inc., Totowa,
NJ, USA.
41 Gadda G (2012) Oxygen activation in flavoprotein
oxidases: the importance of being positive. Biochemistry
51, 2662–2669.
42 Roth JP & Klinman JP (2003) Catalysis of electron
transfer during activation of O2 by the flavoprotein
glucose oxidase. Proc Natl Acad Sci U S A 100, 62–67.
43 Mowat CG, Wehenkel A, Green AJ, Walkinshaw MD,
Reid GR & Chapman SK (2004) Altered substrate
specificity in flavocytochrome b2: structural insights into
the mechanism of L-lactate dehydrogenation.
Biochemistry 43, 9519–9526.
51 Collaborative computational project 4 (1994) The CCP4
suite: programs for protein crystallography. Acta
Crystallogr D Biol Cryst 50, 760–763.
52 Murshudov GN, Vagin AA & Dodson EJ (1997)
Refinement of macromolecular structures by the
maximum-likelihood method. Acta Crystallogr D Biol
Cryst 53, 240–255.
44 Sukumar N, Dewanti A, Merli A, Rossi GL, Mitra B
& Mathews FS (2009) Structures of G81A mutant form
of the active chimera of (S)-mandelate dehydrogenase
53 Emsley P, Lohkamp B, Scott WG & Cowtan K (2010)
Features and development of COOT. Acta Crystallogr
D Biol Cryst 66, 486–501.
578
FEBS Journal 282 (2015) 562–578 © 2014 FEBS

Copyright of FEBS Journal is the property of Wiley-Blackwell and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.