Mechanism of the Novel Prenylated Flavin-Containing Enzyme Ferulic Acid Decarboxylase Probed by Isotope Effects and Linear Free-Energy Relationships

Article abstract of DOI:10.1021/acs.biochem.6b00170

Ferulic acid decarboxylase from Saccharomyces cerevisiae catalyzes the decarboxylation of phenylacrylic acid to form styrene using a newly described prenylated flavin mononucleotide cofactor. A mechanism has been proposed, involving an unprecedented 1,3-dipolar cyclo-addition of the prenylated flavin with the α=β bond of the substrate that serves to activate the substrate toward decarboxylation. We measured a combination of secondary deuterium kinetic isotope effects (KIEs) at the α- and β-positions of phenylacrylic acid together with solvent deuterium KIEs. The solvent KIE is 3.3 on V_max/K_M but is close to unity on V_max, indicating that proton transfer to the product occurs before the rate-determining step. The secondary KIEs are normal at both the α- and β-positions but vary in magnitude depending on whether the reaction is performed in H₂O or D₂O. In D₂O, the enzyme catalyzed the exchange of deuterium into styrene; this reaction was dependent on the presence of bicarbonate. This observation implies that CO₂ release must occur after protonation of the product. Further information was obtained from a linear free-energy analysis of the reaction through the use of a range of para- and meta-substituted phenylacrylic acids. Log(k_cat/K_M) for the reaction correlated well with the Hammett σ^- parameter with p= -0.39 ± 0.03; r² = 0.93. The negative p value and secondary isotope effects are consistent with the rate-determining step being the formation of styrene from the prenylated flavin-product adduct through a cyclo-elimination reaction.

Full text of DOI:10.1021/acs.biochem.6b00170

Article
pubs.acs.org/biochemistry
Mechanism of the Novel Prenylated Flavin-Containing Enzyme
Ferulic Acid Decarboxylase Probed by Isotope Effects and
Linear Free-Energy Relationships
Kyle L. Ferguson,† Nattapol Arunrattanamook,‡,† and E. Neil G. Marsh*,†,§
‡
§
†Department of Chemistry, Department of Chemical Engineering, and Department of Biological Chemistry,
University of Michigan, Ann Arbor, Michigan 48109, United States
ABSTRACT: Ferulic acid decarboxylase from Saccharomyces
cerevisiae catalyzes the decarboxylation of phenylacrylic acid to
form styrene using a newly described prenylated flavin mono-
nucleotide cofactor. A mechanism has been proposed, involving
an unprecedented 1,3-dipolar cyclo-addition of the prenylated
flavin with the αβ bond of the substrate that serves to activate
the substrate toward decarboxylation. We measured a combi-
nation of secondary deuterium kinetic isotope effects (KIEs) at
the α- and β-positions of phenylacrylic acid together with
solvent deuterium KIEs. The solvent KIE is 3.3 on Vmax/KM but
is close to unity on Vmax, indicating that proton transfer to the product occurs before the rate-determining step. The secondary KIEs
are normal at both the α- and β-positions but vary in magnitude depending on whether the reaction is performed in H2O or D2O.
In D2O, the enzyme catalyzed the exchange of deuterium into styrene; this reaction was dependent on the presence of bicarbonate.
This observation implies that CO2 release must occur after protonation of the product. Further information was obtained from a
linear free-energy analysis of the reaction through the use of a range of para- and meta-substituted phenylacrylic acids. Log(kcat/KM)
for the reaction correlated well with the Hammett σ− parameter with ρ = −0.39 0.03; r2 = 0.93. The negative ρ value and
secondary isotope effects are consistent with the rate-determining step being the formation of styrene from the prenylated flavin−
product adduct through a cyclo-elimination reaction.
nzyme-catalyzed decarboxylation reactions constitute an
isoalloxazine flavin nucleus (Figure 1). The cofactor is used by a
Eimportant class of biological reactions yet are inherently family of enzymes represented by UbiD, which catalyzes the
difficult to achieve under physiological conditions.1,2 This is due
to the accumulation of negative charge on the α-carbon in the
decarboxylation of 4-hydroxy-3-octaprenylbenzoic acid as part of
ubiquinone biosynthesis in many prokaryotes.10−14 PrFMN was
recently shown to be synthesized from dimethylallyl phosphate
transition state of decarboxylation, which results in a high-energy
transition state. To overcome the unfavorable energetic barrier for
decarboxylation enzymes, a wide variety of cofactors can be em-
ployed, including organic prosthetic groups such as thiamine pyro-
phosphate and pyridoxal phosphate and Lewis acids such as Mg2+,
Fe2+, and Mn2+, which serve to stabilize the negative charge.1,2
Decarboxylases have also attracted interest as industrial-scale
catalysts for the production of high value, optically pure chemicals
under mild conditions in environmentally friendly bioprocesses.3,4
Aromatic acid decarboxylases are of particular interest, as these
enzymes can be used to produce a wide variety of commercially
valuable substituted aromatic compounds such as styrene, vinyl
guiacol, and vanillin. One such enzyme, ferulic acid decarboxylase
(FDC), catalyzes the decarboxylation of a wide variety of ring-
substituted phenylacrylic acid (cinnamic acid) derivatives to
produce the corresponding substituted styrene derivatives.5−7
Recently, studies on FDC have identified a new flavin-derived
cofactor involved in the decarboxylation of α,β-unsaturated
carboxylic acids.7−9 This cofactor, which has been termed a
prenylated (pr) flavin, is a modified form of FMN (prFMN)
containing an extra 6-membered ring formed by the addition of
an isopentyl group between the N5 and C6 positions of the
and FMN by UbiX.8,9
On the basis of crystallographic evidence, a mechanism has
been proposed for FDC that involves an initial 1,3-dipolar cyclo-
addition between prFMN and the double bond of phenylacrylic
acid, which is followed by a Grob-type elimination of carbon
dioxide (Figure 1A).8,15−17 Protonation of the substrate by an
active site glutamate and finally cyclo-elimination to yield styrene
and prFMN complete the catalytic cycle. The mechanism of
FDC is of considerable interest because true thermal pericyclic
reactions are extremely rare in enzymes, and 1,3-dipolar cyclo-
additions are unprecedented.18,19 We note that a simpler mech-
anism involving a Michael addition of prFMN to the double bond
of phenylacrylic acid followed by decarboxylation and elimination
of the cofactor was also considered (Figure 1B).8 However, this
mechanism was disfavored on the basis of an X-ray structure of
a covalent inhibitor−prFMN adduct that was inconsistent with
prFMN reacting in this way.8 Here, we used a combination of
Received: February 23, 2016
Revised: April 26, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.biochem.6b00170
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Figure 1. Proposed mechanisms for the FDC-catalyzed decarboxylation of phenylacrylic acid proceeding through (A) 1,3-dipolar cyclo-addition to
the prFMN cofactor and (B) Michael addition/elimination of prFMN to the substrate. (C) Details of the active site: active site residues are shown
in magenta with the prFMN cofactor in orange. The substrate analogue α-methylphenylacrylic acid (blue) is shown bound in the active site
(PDB ID: 4ZA7).
isotope effects and linear free-energy analysis to probe the
mechanism by which this newly discovered cofactor catalyzes
decarboxylation reactions.
4-methylcinnamic acid were purchased from Acros Organic. trans-
Cinnamic acid, 3-nitrocinnamic acid, and 3-methoxycinnamic acid
were purchased from Sigma-Aldrich. 4-Cyanocinnamic acid was
purchased from Matrix Scientific. 4-Aminocinnamic acid and
4-nitrocinnamic acid were purchased from Tokyo Chemical
Industries. 4-(2-Carboxy-vinyl)-benzoic acid methyl ester was
purchased from Santa Cruz Biotechnology. All other chemicals
were purchased from Sigma-Aldrich.
MATERIALS AND METHODS
■
Materials. trans-Cinnamic acid, styrene, p-coumaric acid,
4-methoxycinnamic acid, 4-fluorocinnamic acid, 4-formylcin-
namic acid, 4-bromocinnamic acid, 4-chlorocinnamic acid, and
B
DOI: 10.1021/acs.biochem.6b00170
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
d7-trans-Cinnamic acid and 3-d1-trans-cinnamic acid were pur-
chased from Sigma-Aldrich; 2-d1-trans-cinnamic acid was synthe-
sized by reaction of benzaldehyde with d6-acetic anhydride using
standard literature procedures.20
that lacked KHCO3 were also prepared and were degassed prior
to reaction to remove dissolved CO2.
RESULTS
■
Holo-FDC was recombinantly expressed in Escherichia coli and
Hammett Analysis of FDC. Linear free-energy relationships
provide a powerful tool with which to interrogate reaction
mechanisms.22 The narrow substrate range of most enzymes
limits the utility of this approach for the investigation of enzyme
reactions, although for more promiscuous enzymes, linear free-
energy analyses have proved to be highly informative.23−25
The broad substrate range of FDC made it an excellent candidate
to conduct a Hammett analysis of the decarboxylation reaction.
The kinetics of decarboxylation were studied for a series of
14 para- and meta-substituted phenylacrylic acids with σ−
values ranging from 1.25 to −0.66, which are listed in Table 2.
purified as described previously.7
Enzyme Assay. Assays of FDC activity were routinely per-
formed in 100 mM phosphate buffer (pH 7.4) at 25 °C. Stock
solutions of substrates were prepared in DMSO. Assays contained
substrates at various initial concentrations between 10 and 50 μM.
Reactions were initiated by addition of FDC to a final concentration
between 50 and 500 nM depending on the activity of the enzyme
with a particular substrate. We followed the activity spectrophoto-
metrically by monitoring depletion of the substrates. The wave-
lengths used to monitor the enzyme activity and extinction coeffi-
cients of the various substrates are given in Table 1. Care was taken
Table 2. kcat/KM Values Measured for FDC-Catalyzed
Decarboxylation of Various Phenylacrylic Acid Derivatives
Table 1. Assay Wavelengths and Extinction Coefficients for
Phenylacylic Acid Derivatives Used in This Study
substituent
σ−
kcat/KM (M−1 s−1
)
log(kcat/KM)
εsubstrate
εproduct
p-NH2−
p-HO−
p-CH3O−
p-CH3−
p-H−
p-F−
m-CH3O−
p-Cl−
−0.66
−0.37
−0.27
−0.17
0
12400 670
4550 250
4.09 0.023
3.66 0.023
4.45 0.027
4.43 0.025
4.39 0.023
4.44 0.027
4.44 0.031
4.29 0.026
4.33 0.026
4.06 0.012
4.16 0.030
4.11 0.029
3.88 0.015
3.94 0.008
substituent
assay wavelength (nm)
(M−1 cm−1
)
(M−1 cm−1
)
p-NH2−
p-HO−
p-CH3O−
p-CH3−
p-H−
p-F−
m-CH3O−
p-Cl−
292
294
290
280
276
276
276
280
278
300
280
284
304
314
17900
16900
18900
19000
17900
14300
14300
21200
22000
19000
14800
26600
24800
18000
1840
7930
5410
28100 1800
27000 1600
24500 1400
27500 1800
27000 2000
19700 1200
21400 1300
11500 320
14300 1000
12800 880
8690 300
0.02
0.12
0.23
0.26
0.66
0.71
0.88
1.13
1.25
948
p-Br−
1810
2630
p-CH3OOC−
m-NO2−
p-CN−
p-CHO−
p-NO2−
p-Br−
p-CH3OOC−
m-NO2−
p-CN−
p-CHO−
p-NO2−
7240
7230
8690 160
9760
The relatively high KM values for some substrates of >0.35 mM
precluded reliable measurements of full Michaelis−Menten curves.
Therefore, initial velocities were measured at low substrate concen-
trations relative to KM, where velocity is proportional to substrate
concentration, allowing the determination of kcat/KM for each
substrate (Table 2).
The resulting Hammett plot for a series of 12 of the substituted
phenylacrylic acids listed in Table 2 is shown in Figure 2.
The data were poorly correlated with the standard Hammett
parameter σ (ρ = −0.5; r2 = 0.65) but showed a strong correlation
with the Hammett parameter σ− (ρ = −0.39 0.03; r2 = 0.93).
Two substrates, 4-amino- and 4-hydroxyphenylacrylic acid,
reacted much more slowly than expected based on their highly
negative σ− values. These substrates are the only compounds that
contain hydrogen bond-donating functional groups. Although it
is unclear why they react so slowly, the large deviation of these
compounds from the trend line suggests they may react by a dif-
ferent mechanism. For this reason, they were excluded from the
calculation of ρ in our analysis.
The apparent linearity of the Hammett analysis provides
strong evidence that a chemical step is rate-determining in the
reaction. The negative slope of the plot, which is caused by
electron-releasing groups that increase the rate of reaction, is very
unusual, as decarboxylation reactions (both enzymatic and
nonenzymatic) involve a buildup of negative charges in the
transition state and are therefore associated with large positive
ρ values. The strong correlation with σ− indicates that, in addition to
inductive effects, delocalization of electrons through the extended
to ensure that the assay remained linear at the concentrations of
substrate and enzyme used.
pH Dependence and Solvent and Secondary Deute-
rium Isotope Effect Measurements. The pH dependence of
FDC activity was studied with the following buffers: pH 4.0−6.5,
100 mM sodium citrate; pH 6.5−8.0, 100 mM sodium phos-
phate; pH 8.0−9.0, 100 mM Tris chloride. For measurements in
deuterated solvents, the buffer components were dissolved in
D2O (99.9%) and repeatedly lyophilized to exchange protium.
pD was corrected using eq 1, where χ is the mole fraction of D2O.
pD = pH + 0.076χ2 + 0.3314χ + 0.00009
(1)
To measure the solvent deuterium kinetic isotope effect on
styrene formation, the reactions were performed in buffers con-
taining various mole fractions of D2O, and the deuterium content
of the styrene produced was analyzed by GC-MS as described
previously.17,21
Secondary KIEs were measured at substrate concentrations of
50 μM and enzyme concentrations of 50 nM so that the measure-
ments represent the KIE on Vmax/KM. Experiments were performed
at a pD and pH of 6.5, where the activity of the enzyme is indepen-
dent of pH.
1
We monitored the proton exchange experiments using H
NMR at 400 MHz at 22 °C by monitoring the disappearance of
the styrene resonance at 5.2 ppm. The reaction buffer was
comprised of 100 mM potassium citrate in D2O (pD 6.5), 10 μM
FDC, 20 mM KHCO3, and 2.5 mM styrene. Reaction mixtures
C
DOI: 10.1021/acs.biochem.6b00170
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Figure 2. Hammett plot for the FDC-catalyzed decarboxylation of
14 different para- and meta-substituted phenylacrylic acids. Data points
in red were used to determine ρ; the two outliers in blue were excluded
from the analysis.
Figure 3. pL-rate profile for decarboxylation of phenylacylic acid by
FDC in H2O (blue) and D2O (red) buffers.
π-system associated with the aromatic ring is important in
stabilization of the transition state in a step that contributes
significantly to the overall rate of the reaction.19,25
Kinetic Isotope Effects. The proposed mechanism of the
FDC reaction involves the creation and breaking of bonds at three
carbon atoms in the substrate. The kinetic significance of these
bonding changes in the overall reaction was examined through
the use of kinetic isotope effects. In particular, proton incorporation
into styrene should be subject to a solvent kinetic isotope effect,
and the transient changes in hybridization at Cα and Cβ of the
substrate should be subject to secondary kinetic isotope effects.
pH Dependence and Solvent Isotope Effects. Initial
characterization by 1H NMR of the styrene produced in the FDC
reaction established that the solvent proton is incorporated trans
to the phenyl ring of styrene, as evidenced by the disappearance of
the signal at 5.2 ppm when the reaction was performed in D2O
(see Deuterium Exchange into Styrene). Therefore, decarboxyla-
tion occurs with the retention of the configuration at Cα, which is
in accord with the geometry of the active site and the proposed
role of Glu282 acting as the proton donor.
Because solvent isotope effects may be influenced by pH, we
first investigated the activity of FDC as a function of pH and pD.
Under Vmax conditions, the enzyme exhibits a bell-shaped pH
dependence, typical of many enzymes, with an activity maximum
at pH 6.5. The acidic limb is characterized by pKa = 5.3 0.1, and
the basic limb is characterized by pKa = 8.0 0.1. The enzyme
activity in deuterated buffers was very similar, although the pKa of
the acidic limb was shifted to a slightly lower value (pKa = 5.0
0.1) (Figure 3). On the basis of the pL curves, the solvent KIE on
Figure 4. Proton inventory for FDC. The mole fraction of D2O in the
solvent is plotted against the ratio of protiated to deuterated styrene.
The solid line represents the best fit to eq 2. A linearized plot of the data
is shown in the inset.
we can determine it more accurately by fitting the full proton inven-
tory data, shown in Figure 4, to eq 2.
⎛
⎜
⎞
⎠
V
max (DVsolvent) was measured at pL = 6.5. The value is very close
1
2
⎟
⎟
χproduct
=
D OSIEobs⎜
− 1
to unity (DVsolvent = 0.95 0.05; n = 5), indicating that proton
transfer to the product is not a kinetically significant step in the
overall reaction.
χ
⎝
(2)
solvent
From this analysis, a substantial solvent KIE on Vmax/KM is
evident, with DV/Ksolvent = 3.33
0.09. We note that eq 2
To investigate the protonation step in more detail, we conducted
a proton inventory analysis, which allowed us to measure the
solvent KIE on Vmax/KM (DV/Ksolvent). Reaction mixtures were set
up at pL 6.5 in buffers containing 0.1 mole fraction increments of
assumes that only a single proton is in motion in the transition
state; the excellent fit of the data to the equation suggests that this
assumption is valid. KIEs on Vmax are observed only if the
isotopically sensitive step is rate-determining under saturating
substrate concentrations, whereas the KIE on Vmax/KM
represents the KIE determined at low substrate concentrations
and reflects all steps up to and including the first irreversible step.
deuterium (χD O). The styrene produced by the reaction was
2
isolated, and the mole fraction of deuterium (χstyrene) was
D
determined by GC-MS. While V/Ksolvent may be estimated by
the determination of χstyrene at χD O =0.5, forwhichDV/Ksolvent =∼3.3,
2
D
DOI: 10.1021/acs.biochem.6b00170
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Our data therefore indicate that the steps that control Vmax/KM
and Vmax are different. We interpret these results as an indication
that protonation of the product is associated with a significant
energetic barrier and occurs before the first irreversible rate-
determining step of the reaction. As discussed below, these steps
are likely associated with release of the products from the enzyme.
Secondary Kinetic Isotope Effects. Secondary kinetic
isotope effects report on changes in the stiffness of bonds adjacent
to the site of the reaction and may be either normal or inverse.26−29
They are particularly informative for an examination of changes in
the geometry of carbon atoms: the transition from tetrahedral to
planar geometry is associated with a normal secondary KIE, where-
as the transition from planar to tetrahedral geometry is associated
with an inverse secondary KIE. To investigate the mechanism of
the FDC reaction, we measured secondary deuterium KIEs at both
the α- and β-positions of phenylacrylic acid in both D2O and H2O
buffers (the commercially available β-deuterated phenylacrylic acid
was also deuterated on the phenyl ring; the remote isotope was
assumed to not influence the KIE at the β-position). KIEs were
determinedat25 °C in100 mMsodium citratebuffer (pL 6.5)bya
direct comparison of reaction rates at low substrate concentrations
relative to KM so that the measurements represent KIEs on
Vmax/KM. The data are presented in Table 3.
occur. This implies that dissociation of CO2 does not occur until
after the product is protonated when the reaction proceeds in the
decarboxylation direction.
DISCUSSION
■
The novel isopentenyl modification of FMN converts a cofactor
generally associated with redox reactions into one that supports
decarboxylation. Its discovery raises a number of interesting
questions regarding the mechanism by which FDC catalyzes
the decarboxylation of aromatic substrates and the nature of the
rate-determining step. The proposed mechanism involving a
1,3-dipolar cyclo-addition of prFMN with phenylacrylic acid
(Figure 1A) seems plausible, as the alkylation of N6 of the flavin
introduces functionality into the cofactor which is expected to
have an ylide character similar to that of an azomethine ylide.8,31
The kinetic experiments described here provide insight into the
rate-determining step in the reaction mechanism and provide
further lines of support for the mechanistic proposal shown in
Figure 1, which is based primarily on X-ray crystallography of
enzyme−inhibitor complexes.
The stereospecificity with which deuterium is incorporated
into styrene is consistent with the α-carbon remaining bonded to
prFMN after decarboxylation. If the α-carbon of the substrate
was free to rotate about the Cα−Cβ bond, then a scrambling of
the isotope would be observed. The stereochemistry of deuterium
incorporation, which occurs with a retention of the configuration,
is also consistent with the proposed role of Glu282 acting as the
proton donor in the reaction. The basic limb of the pH-rate profile
(Figure 3), which is characterized by an apparent pKa of 8.0, may
be attributed to Glu282, as this residue must be protonated for the
reaction to proceed. Although a pKa of 8.0 is quite high even for a
buried glutamate side chain, the proximity of the negatively
charged substrate is expected to further raise its pKa.
Table 3. Summary of Secondary Kinetic Isotope Effects
Measured for the FDC-Catalyzed Decarboxylation of
Deuterated Phenylacrylic Acids in H2O and D2O
deuterium position
2° DV/K (H2O)
2° DV/K (D2O)
Cα
0.99 0.02
1.10 0.016
1.15 0.017
1.12 0.030
1.01 0.027
1.32 0.035
Cβ
Cα and Cβ
In H2O, no apparent 2° KIE was observed at the α-position,
whereas at the β-position a large, normal 2° KIE was measured
Furthermore, the observation that proton exchange into
−
styrene requires HCO3 indicates that release of CO2 from
(2°DV/Kβ(H O) = 1.10
0.03; n = 9). In contrast, in D2O,
the apparent 2° KIE at the α-position became significant
the enzyme must occur after proton transfer to the substrate
(we assume that HCO3− serves as a CO2 source rather than the
enzyme utilizing HCO3− directly). It would also be consistent if
the proton transfer preceded decarboxylation in the reaction
mechanism, although this seems unlikely. Lastly, the large solvent
2
(2°DV/Kα(D O) = 1.12 0.03; n = 9), whereas the 2° KIE at
2
the β-position was suppressed and was at unity within error.
When dideuterated phenylacrylic acid was the substrate, large 2°
KIEs were measured in both H2O and D2O. The fact that these
KIEs are normal indicates that they both arise from the
rehybridization of the α- and β-carbons from tetrahedral to
planar geometry during the reaction.
deuterium KIE on Vmax/KM (DV/Ksolvent = 3.33
0.09),
measured by internal competition, contrasts with the KIE on
max, measured by a direct comparison of reaction rates in H2O
V
and D2O, which are close to unity. This is consistent with
protonation of the styrene−prFMN adduct being the rate-
determining step in the decarboxylation/protonation portion of
the reaction. However, it is unlikely that protonation is the rate-
determining step that limits kcat. This would be inconsistent with
the fact that the solventKIE onVmax is unity and thatthe secondary
KIEs are normal, rather than inverse, as discussed below.
Deuterium Exchange into Styrene. The decarboxylation
catalyzed by FDC can be driven in the reverse direction in the
presence of styrene and high concentrations of bicarbonate.30
We used the reversibility of the reaction to examine the require-
ments for proton transfer between Glu282 and substrate. Reaction
mixtures were set up in citrate-buffered D2O solutions containing
2 mM styrene that were either purged of dissolved CO2 or
supplemented with 20 mM KHCO3. We monitored the exchange
of deuterium into styrene using 1H NMR by following the disap-
pearance of the resonance at 5.20 ppm. Deuterium incorporation
at the trans-position of C1 also resulted in a simplification of
the signal at 5.75 ppm due to the cis-C1 hydrogen from a doublet
of doublets to a doublet together with a slight upfield isotope-
induced chemical shift. In CO2-purged buffers, little to no exchange
of deuterium was observed (Figure 5), whereas the addition of
KHCO3 resulted in deuterium exchange into styrene occurring
quite rapidly with an apparent rate constant kapp of ∼1.5 min−1.
This result demonstrates that CO2 must be present in the active site
for protonation/deprotonation of the styrene−prFMN adduct to
The strong correlation of log(kcat/KM) with the Hammett σ−
parameter (Figure 2) points to a chemical step in the reaction as
rate-determining, rather than a substrate-binding or product
release step, in which case ρ would be expected to be close to zero.
Interestingly and unexpectedly, for a decarboxylation reaction, ρ is
negative (ρ = −0.39). Therefore, it seems unlikely, regardless of
the precise details of the mechanism, that the decarboxylation step
is rate-determining. Otherwise, the buildup of negative charge in
the transition state should lead to a positive ρ value, which is
generally observed in decarboxylation reactions.22,25
The negative ρ value provides evidence that the 1,3-cyclo-
elimination reaction that leads to the release of styrene from
prFMN is likely to be rate-determining in the FDC reaction; this
E
DOI: 10.1021/acs.biochem.6b00170
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
−
Figure 5. FDC catalyzes deuterium exchange into styrene in the presence of 20 mM HCO3 . 1H NMR spectra were recorded at 500 MHz between 0 and
−
30 min after addition of FDC. The bottom spectrum was recorded under the same conditions but in the absence of HCO3 . For details, see the text.
would be consistent with the 2° KIE measurements discussed
below. We note that, although the correlation of reactivity
with σ− is generally discussed in terms of resonance effects,
this formalism derives from the overlap of p-orbitals to form
extended π-systems and is thus correlated with the HOMO-
LUMO analyses used to rationalize thermal pericyclic reactions.
In 1,3-dipolar cyclo-addition reactions, azomethine ylides have a
nucleophilic character and react increasingly rapidly with increas-
ingly electron-withdrawing dipolarophiles (i.e., the rate of reaction
is dominated by the difference in energy between the HOMO of
the azomethine ylide and the LUMO of the dipolarophile).31
Indeed, these types of cyclo-addition reactions have been found to
exhibit linear free-energy relationships that correlate well with the
electronic properties of the dipolar ylide and the dipolarophile.31
This implies that the rate of the reverse cyclo-elimination reaction
will, conversely, be slowed by electron-withdrawing substituents.
In the present case, the structure of prFMN suggests that its reac-
tivity toward dipolarophiles will resemble that of an azomethine
ylide; hence, electron-withdrawing substituents on the phenyl ring
will slow the final cyclo-elimination reaction leading to product
formation, consistent with our experimental observations.
geometry at the carbon atoms from tetrahedral to planar,27 which
again points to the rate-determining step involving the formation
of the styrene double bond in the final cyclo-elimination reaction.
The change in the apparent 2° KIEs observed when the reaction
is performed in D2O is unusual but may be explained by the fact
that, when the α-carbon undergoes rehybridization, it contains an
additional deuterium from the solvent. This will introduce a
cryptic 2° KIE at the α-carbon that could mask the 2° KIE at the
β-carbon. When the enzyme is reacted with α-deuterated pheny-
lacrylic acid in D2O, the α-carbon will contain two deuterium
atoms; thus, the observed 2° KIE will be further elevated and
will appear as a normal KIE. A similar argument can be made to
explain the further increase in 2° KIEs when α,β-dideuterated
substrates are used. The pattern of 2° KIEs further suggests that
the 1,3-cyclo-elimination reaction that leads to the formation of
the styrene double bond occurs in a concerted but asynchronous
reaction in which bond cleavage is advanced at the α-carbon with
respect to the β-carbon.
Our results do not definitively rule out a previously considered
mechanism in which a Michael addition of prFMN to pheny-
lacrylic acid facilitates decarboxylation, as shown in Figure 1B.8
However, we consider this possibility as less likely because in
this mechanism the 2° KIEs indicate that decarboxylation was rate-
limiting, concomitant with a change in geometry from tetrahedral
to planar at the α- and β-carbons. As discussed above, postulating
decarboxylation as the rate-limiting step appears inconsistent with
the negative value of ρ associated with the overall reaction.
The 2° KIEs measured for FDC provide further evidence that a
chemical step, rather than a substrate binding or product release
step, is rate-determining. Normal 2° KIEs are observed at both the
α- and β-positions of phenyl acrylate, although they are solvent-
dependent, which complicates their interpretation as discussed
below. Normal 2° isotope effects are indicative of a change in
F
DOI: 10.1021/acs.biochem.6b00170
Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
(9) White, M. D., Payne, K. A. P., Fisher, K., Marshall, S. A., Parker, D.,
Rattray, N. J. W., Trivedi, D. K., Goodacre, R., Rigby, S. E. J., Scrutton, N.
S., Hay, S., and Leys, D. (2015) UbiX is a flavin prenyltransferase
required for bacterial ubiquinone biosynthesis. Nature 522, 502−506.
(10) Bentinger, M., Tekle, M., and Dallner, G. (2010) Coenzyme
Q - Biosynthesis and functions. Biochem. Biophys. Res. Commun. 396,74−79.
(11) Gulmezian, M., Hyman, K. R., Marbois, B. N., Clarke, C. F., and
Javor, G. T. (2007) The role of UbiX in Escherichia coli coenzyme Q
biosynthesis. Arch. Biochem. Biophys. 467, 144−153.
(12) Jacewicz, A., Izumi, A., Brunner, K., Schnell, R., and Schneider, G.
(2013) Structural Insights into the UbiD Protein Family from the
Crystal Structure of PA0254 from Pseudomonas aeruginosa. PLoS One
8, e63161.
(13) Meganathan, R. (2001) Ubiquinone biosynthesis in micro-
organisms. FEMS Microbiol. Lett. 203, 131−139.
(14) Rangarajan, E. S., Li, Y. G., Iannuzzi, P., Tocilj, A., Hung, L. W.,
Matte, A., and Cygler, M. (2004) Crystal structure of a dodecameric
FMN-dependent UbiX-like decarboxylase (Pad1) from Escherichia coli
O157: H7. Protein Sci. 13, 3006−3016.
(15) Gothelf, K. V., and Jorgensen, K. A. (1998) Asymmetric 1,3-
dipolar cycloaddition reactions. Chem. Rev. 98, 863−909.
(16) Pellissier, H. (2007) Asymmetric 1,3-dipolar cycloadditions.
Tetrahedron 63, 3235−3285.
(17) Waugh, M. W., and Marsh, E. N. G. (2014) Solvent Isotope
Effects on Alkane Formation by Cyanobacterial Aldehyde Deformylat-
ing Oxygenase and Their Mechanistic Implications. Biochemistry 53,
5537−5543.
(18) Pindur, U., and Schneider, H. G. (1994) Pericyclic Key Reactions
in Biological Systems. Chem. Soc. Rev. 23, 409−415.
(19) Miyamoto, K., and Ohta, H. (1992) Purification and Properties of
a Novel Arylmalonate Decarboxylase from Alcaligenes-bronchisepticus
KU-1201. Eur. J. Biochem. 210, 475−481.
(20) Kurti, L., and Czako, B. (2005) Strategic Applications of Named
Reactions in Organic Synthesis, Elsevier Academic Press.
In conclusion, the results presented here are consistent with
the proposed mechanism for FDC, which involves a novel cyclo-
addition reaction of the substrate with the prenylated flavin
cofactor. The results suggest that the rate-determining step in the
catalytic cycle is resolution of the product−prFMN adduct
through a cyclo-elimination reaction. Further mechanistic studies
will be necessary to better establish the timing of the decar-
boxylation and protonation events, for example, by the measure-
ment of heavy atom KIEs associated with decarboxylation. So far,
evidence for the cyclic adduct between the substrate and prFMN
remains indirect, and further experiments are needed to sub-
stantiate the formation of this key intermediate. Our results
suggest it may be possible to stabilize this adduct sufficiently
to permit its characterization by reaction of the enzyme with
highly electron-withdrawing substrates that slow the rate of cyclo-
elimination. The substrate reactivity patterns uncovered here
may also be useful in guiding efforts to engineer FDC toward
alternative substrates or greater catalytic efficiency.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: nmarsh@umich.edu.
Funding
This research was supported in part by grants from the National
Science Foundation (CHE 1152055 and CBET 1336636) to
E.N.G.M.; N.A. acknowledges the support of a Royal Thai
Government Scholarship.
Notes
The authors declare no competing financial interest.
(21) Gadda, G., and Fitzpatrick, P. F. (2013) Solvent isotope and
viscosity efects on the steady state kinetics of the flavoprotein
nitroalkane oxidase. FEBS Lett. 587, 2785−2789.
(22) Jaffe, H. H. (1953) A Reexamination of the Hammett Equation.
Chem. Rev. 53, 191−261.
ABBREVIATIONS
■
FDC, ferulic acid decarboxylase; prFMN, prenylated flavin
mononucleotide; SIE, solvent isotope effect; KIE, kinetic isotope
effect
(23) Neims, A. H., Deluca, C. D., and Hellerman, L. (1966) Studies on
Crystaline D-Amino Acid Oxidase. III. Substrate Specificity and sigma-pi
Relationship. Biochemistry 5, 203−207.
REFERENCES
■
(1) Jordan, F., and Patel, H. (2013) Catalysis in Enzymatic
Decarboxylations: Comparison of Selected Cofactor-Dependent and
Cofactor-Independent Examples. ACS Catal. 3, 1601−1617.
(2) Kourist, R., Guterl, J.-K., Miyamoto, K., and Sieber, V. (2014)
Enzymatic Decarboxylation-An Emerging Reaction for Chemicals
Production from Renewable Resources. ChemCatChem 6, 689−701.
(3) Claypool, J. T., Raman, D. R., Jarboe, L. R., and Nielsen, D. R.
(2014) Technoeconomic evaluation of bio-based styrene production by
engineered Escherichia coli. J. Ind. Microbiol. Biotechnol. 41, 1211−1216.
(4) McKenna, R., Thompson, B., Pugh, S., and Nielsen, D. R. (2014)
Rational and combinatorial approaches to engineering styrene
production by Saccharomyces cerevisiae. Microb. Cell Fact. 13, 123.
(5) Huang, Z. X., Dostal, L., and Rosazza, J. P. N. (1994) Purification
and Characterization of a Ferulic Acid Decarboxylase from Pseudomo-
nas-Fluorescens. J. Bacteriol. 176, 5912−5918.
(6) Mukai, N., Masaki, K., Fujii, T., Kawamukai, M., and Iefuji, H.
(2010) PAD1 and FDC1 are essential for the decarboxylation of
phenylacrylic acids in Saccharomyces cerevisiae. J. Biosci. Bioengin. 109,
564−569.
(7) Lin, F., Ferguson, K. L., Boyer, D. R., Lin, X. N., and Marsh, E. N.G.
(2015) Isofunctional enzymes PAD1 and UbiX catalyze formation of a
novel cofactor required by ferulic acid decarboxylase and 4-hydroxy-3-
polyprenylbenzoic acid decarboxylase. ACS Chem. Biol. 10, 1137−1144.
(8) Payne, K. A. P., White, M. D., Fisher, K., Khara, B., Bailey, S. S.,
Parker, D., Rattray, N. J. W., Trivedi, D. K., Goodacre, R., Beveridge, R.,
Barran, P., Rigby, S. E. J., Scrutton, N. S., Hay, S., and Leys, D. (2015)
New cofactor supports alpha,beta-unsaturated acid decarboxylation via
1,3-dipolar cyclo-addition. Nature 522, 497−501.
(24) Yorita, K., Misaki, H., Palfey, B. A., and Massey, V. (2000) On the
interpretation of quantitative structure-function activity relationship
data for lactate oxidase. Proc. Natl. Acad. Sci. U. S. A. 97, 2480−2485.
(25) Trahanovsky, W. S., Cramer, J., and Brixius, D. W. (1974)
Oxidation of Organic Compounds with Cerium(IV) 0.18. Oxidative
Decarboxylation of Substittued Phenylacetic Acids. J. Am. Chem. Soc. 96,
1077−1081.
(26) Cleland, W. W. (1987) Secondary isotope effects on enzymatic
reactions. In Isotopes in Organic Chemistry (Buncel, E., and Lee, C. C.,
Eds.) Elsevier, Amsterdam.
(27) Cleland, W. W. (2005) The use of isotope effects to determine
enzyme mechanisms. Arch. Biochem. Biophys. 433, 2−12.
(28) Kurtz, A. K., and Fitzpatrick, P. F. (1997) pH and Secondary
Kinetic Isotope Effects on the Reaction of D-Amino Acid Oxidase with
Nitroalkane Anions: Evidence for Direct Attack on the Flavin by
Carbanions. J. Am. Chem. Soc. 119, 1155−1156.
(29) Northrop, D. B. (1975) Steady-State Analysis of Kinetic Isotope
Effects in Enzyme Reactions. Biochemistry 14, 2644−2651.
(30) Wuensch, C., Pavkov-Keller, T., Steinkellner, G., Gross, J., Fuchs,
M., Hromic, A., Lyskowski, A., Fauland, K., Gruber, K., Glueck, S. M.,
and Faber, K. (2015) Regioselective Enzymatic beta-Carboxylation of
Hydroxy-styrene Derivatives Catalyzed by Phenolic Acid Decarbox-
ylase. Adv. Synth. Catal. 357, 1909−1918.
(31) Perez, P., Domingo, L. R., Aurell, M. J., and Contreras, R. (2003)
Quantitative Characterization of the global electrophilicty pattern of
some reagents involved in 1,3-dipolar cyclo-addition reaction.
Tetrahedron 59, 3117−3125.
G
DOI: 10.1021/acs.biochem.6b00170
Biochemistry XXXX, XXX, XXX−XXX