Biocatalytic carboxylation of phenol derivatives: Kinetics and thermodynamics of the biological Kolbe-Schmitt synthesis

Article abstract of DOI:10.1111/febs.13225

Microbial decarboxylases, which catalyse the reversible regioselective ortho-carboxylation of phenolic derivatives in anaerobic detoxification pathways, have been studied for their reverse carboxylation activities on electron-rich aromatic substrates. Ortho-hydroxybenzoic acids are important building blocks in the chemical and pharmaceutical industries and are currently produced via the Kolbe-Schmitt process, which requires elevated pressures and temperatures (≥ 5 bar, ≥ 100 °C) and often shows incomplete regioselectivities. In order to resolve bottlenecks in view of preparative-scale applications, we studied the kinetic parameters for 2,6-dihydroxybenzoic acid decarboxylase from Rhizobium sp. in the carboxylation- and decarboxylation-direction using 1,2-dihydroxybenzene (catechol) as starting material. The catalytic properties (K_m, V_max) are correlated with the overall thermodynamic equilibrium via the Haldane equation, according to a reversible random bi-uni mechanism. The model was subsequently verified by comparing experimental results with simulations. This study provides insights into the catalytic behaviour of a nonoxidative aromatic decarboxylase and reveals key limitations (e.g. substrate oxidation, CO₂ pressure, enzyme deactivation, low turnover frequency) in view of the employment of this system as a 'green' alternative to the Kolbe-Schmitt processes. Microbial decarboxylases are known to catalyze the reversible regioselective ortho-carboxylation of phenolic derivatives in anaerobic detoxification pathways. In order to get new insights into the catalytic action and to resolve bottlenecks in view applications, we studied the kinetics of 2,6-dihydroxybenzoic acid decarboxylase from Rhizobium sp. in the carboxylation- and decarboxylation-direction, correlating the data according to a reversible random bi-uni mechanism.

Full text of DOI:10.1111/febs.13225

Biocatalytic Carboxylation of Phenol Derivatives:
Kinetics and Thermodynamics of the Biological Kolbe-Schmitt
Synthesis
1
2,3
4
4
3
1
L. Pesci, S. M. Glueck, P. Gurikov, I. Smirnova, K. Faber, A. Liese
1
Institute of Technical Biocatalysis, Hamburg University of Technology, Germany
2
ACIB GmbH c/o
3
Department of Chemistry, Organic & Bioorganic Chemistry, University of Graz, Austria
Institute of Thermal Separation Processes, Hamburg University of Technology, Germany
4
Corresponding author: Andreas Liese, Institute of Technical Biocatalysis, Hamburg
University of Technology, Germany, Tel: +49-40-42878-3018, Fax: +49-40-42878-212,
email: liese@tuhh.de
Article type : Original Article
ABSTRACT
Microbial decarboxylases, which catalyze the reversible regio-selective ortho-carboxylation
of phenolic derivatives in anaerobic detoxification pathways, have been studied for their
reverse carboxylation activities on electron-rich aromatic substrates. Ortho-hydroxybenzoic
acids are important building blocks in the chemical and pharmaceutical industries and are
currently produced via the Kolbe-Schmitt process, which requires elevated pressures and
temperatures (≥5 bar, ≥100 °C) and often shows incomplete regio-selectivities. In order to
resolve bottlenecks in view of preparative scale applications, we studied the kinetic
parameters for 2,6-dihydroxybenzoic acid decarboxylase from Rhizobium sp. in the
carboxylation- and decarboxylation-direction using 1,2-dihydroxybenzene (catechol) as
starting material. The catalytic properties (K , Vmax) are correlated with the overall
m
thermodynamic equilibrium via the Haldane equation, according to a reversible random bi-
uni mechanism. The model was subsequently verified by comparing experimental results
with simulations. This works provides insights into the catalytic behavior of a non-oxidative
2
aromatic decarboxylase and reveals key limitations (e.g., substrate oxidation, CO pressure,
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/febs.13225
This article is protected by copyright. All rights reserved.

enzyme deactivation, low turnover frequency) in view of the employment of this system as a
green' alternative to the Kolbe-Schmitt processes.
'
Keywords: Kolbe-Schmitt reaction, biocatalytic carboxylation, non-oxidative carboxylation,
enzyme deactivation, kinetic modelling, thermodynamics.
INTRODUCTION
The Kolbe-Schmitt process for the carboxylation of hydroxy-aromatics is known since about
a century and allows access to carboxylic acids on industrial scale by employing carbon
dioxide as a C1 building block for organic synthesis [1]. The original process has been
customized to the type of substrate and the reaction conditions [2, 3]. In general, the reaction
suffers from equilibrium-limited yields of about 60% and from incomplete regio-selectivities
regarding o,p-isomeric products and requires harsh conditions in terms of pressure (50-100
bar) and temperature (100-200 °C). Less reactive mono-phenols need anhydrous reaction
conditions [3]. Recently, new investigations have been carried out to optimize these processes
by use of different reaction media, such as “reactive” ionic liquids based on bicarbonate as
anion, microwave-based heating and flow operation modes [4, 5]. On the other hand, a
biological equivalent to the Kolbe-Schmitt reaction has been discovered in the microbial
anaerobic detoxification of phenolic compounds, where non-oxidative decarboxylases act in
the C-C bond forming direction [6–8] to furnish a more polar and water soluble carboxylate
anion [9, 10]. As usual in biodegradation, the substrate promiscuity is high [11, 12]. In line
with the mechanism of electrophilic aromatic substitution (S Ar), two types of selectivities
E
were found in the biological carboxylation of phenolic substrates: ortho and para
carboxylation.
Ortho-selectivity is displayed by hydroxybenzoic and salicylic acid decarboxylases [13, 14].
These well investigated enzymes share common features, such as exclusive regio-selectivity
and absence of cofactors. Even with electron-rich substrates, conversions are limited to a
range of 10 to 40%, which is partly ascribable to the reaction equilibria. Higher carboxylation
yields are achieved in the presence of elevated bicarbonate concentrations (usually ~3M) [11,
1
5, 16]. 2,6-Dihydroxybenzoic acid decarboxylases (or γ-resorcylic acid decarboxylases)
were described from Rhizobium sp. and Pandoraea sp. [13, 17]. Ample information is
available on the substrate tolerance and the crystal structure revealing a catalytically essential
2
+
Zn , which allowed the proposal for a reaction mechanism [12, 18]. Preliminary studies on
the influence of reaction conditions on the catalytic performance revealed that the presence
This article is protected by copyright. All rights reserved.

water-miscible organic solvents and ionic liquids as alternative reaction media increased the
yields in resorcinol carboxylation from 30% up to 50% [8]. Despite this broad knowledge,
m
only poor kinetic information is available, i.e. K values of 25.9 mM for resorcinol and 0.08
mM for its carboxylated product, and a K of 0.12 mM for 2,3-dihydroxybenzoate [17].
m
In this work, we present our investigations on enzyme kinetics together with the reaction
thermodynamics in line with the verification of the catalytic mechanism of 2,6-
dihydroxybenzoic acid decarboxylase (DHBA decarboxylase) from Rhizobium sp. with
catechol as the phenolic substrate (Scheme 1). This study elucidates intrinsic bottlenecks for
the design of biochemical carboxylation processes of phenolic compounds.
RESULTS
Reaction medium and substrate oxidation. Usually, the carboxylation activity for non-
oxidative decarboxylases is tested using KHCO close to saturating concentrations (~3M) in
3
1
00 mM potassium phosphate (KP
i
) buffer. High bicarbonate concentrations are necessary to
buffer is reported to
drive substrate conversion due to a Le Chatelier effect, while the KP
i
compensate for the pH increase due to the addition of bicarbonate [19, 16, 11].
°
Considering the equilibrium constants at 25 C for the dissociation of carbonic acid to
bicarbonate anion, it can be calculated, that regardless of the salt concentration, the pH of
such a solution would be 8.3. Consequently, due to the high concentration, it is not likely to
expect any pH shift as a consequence of carboxylate formation, which can reach
concentrations of ≤20 mM. Unless phosphate exhibits a positive effect on the enzyme
activity, one would not expect any differences in activity and long term conversion in
KHCO
.5 (Figure S1). The long term-conversions ranged in all cases between 26 and 28%,
indicating that there is no appreciable effect of the KP buffer on the enzyme stability or the
3 3 2 3 3 i
solution (2 M), KHCO /K CO buffer (2M), KHCO (2M) in KP buffer 0.1 M pH
5
i
reaction thermodynamics. Consequently, the following experiments were performed using an
unbuffered bicarbonate solution.
In order to evaluate possible chemical oxidation of the substrate, progress curves were
obtained by performing the reaction in an open and a closed reaction vessel. After around 3
hours, the product concentration reached a constant value of approximately 2.6 mM, while
catechol continued to be consumed. The progress curves clearly indicate that the substrate is
oxidized by air yielding the reactive o-quinone, which in turn polymerizes going in hand with
a color change of the solution from light brown to black (Figure S2). The same phenomenon
was observed with purified DHBA decarboxylase, which excludes the undesired action of
This article is protected by copyright. All rights reserved.

oxidases (such as Cu-dependent monooxygenases) present in crude CFE [20]. In order to
circumvent any competing chemical substrate oxidation, we tested several parameters (such
as addition of a reductant, solution degassing, protection from light) and we found the
addition of stoichiometric amounts of ascorbic acid as an effective tool to suppress the
chemical formation of quinone by autooxidation (Scheme S1 and Figure S3).
Kinetics. The carboxylation requires two substrates (catechol and bicarbonate) with the
formation of one product, 2,3-dihydroxybenzoic acid (DHBA), according to a bi-uni reaction.
As a consequence, kinetic analyses need to be conducted with purified enzyme on both
substrates (maintaining the other in saturating amounts), and on the product (due to the
reversibility of the reaction) in order to elucidate the specific activity and to calculate the
turnover frequency. Figure 1 shows initial rate measurements for catechol. The presence of
constant amounts of co-substrate bicarbonate (2 M, corresponding to 2.3 times its K ) was
m
ensured by keeping the conversion below 10 %. With limited enzyme amounts (0.20-0.30
mg/mL) the reaction was slow enough to measure initial rates directly, taking samples during
the first 20 minutes of the reaction, where zero order kinetics still occurs.
The profile shows a typical hyperbolic behavior up to 120 mM, whereas the reaction rate
steeply decreases beyond this value. This trend is not ascribable to substrate surplus
inhibition, where a second substrate molecule behaves as uncompetitive inhibitor, as it was
impossible to fit the experimental points with the corresponding equation. Experimental
points in the concentration range of 0-120 mM were fitted using the double substrate
Michaelis-Menten equation. From this plot, a K for catechol (Km,c) of 30 ± 3 mM and a
m
f
-1
-1
V
max in the carboxylation direction of 0.35 ± 0.01 µmol min mg of decarboxylase can be
calculated. In the literature a K of 25.9 mM is known for a 2,6-dihydroxybenzoic acid
m
decarboxylase from Agrobacterium tumefaciens using 1,3-dihydroxybenzene (resorcinol) as
substrate [17]. Previous studies on non-oxidative aromatic bio-carboxylation reveal that the
maximum conversion is higher at relatively low substrate concentrations, depending on the
enzyme type and the substrate. For example, in the ortho-carboxylation of 5-methyl-1,3-
dihydroxybenzene (orcinol) and resorcinol catalyzed by DHBA decarboxylase, conversions
drop from 67 and 35% at 10 to 50 mM substrate concentrations to 4 and 15% at
approximately 200 mM substrate concentrations, respectively [18, 8]. The same behavior was
found in the o-carboxylation of p-aminophenol yielding p-aminosalicylic acid, where the
maximal conversion dropped from 70% at 10 mM to 20% at 200 mM [21]. The same trend
This article is protected by copyright. All rights reserved.

was observed with hetero-aromatics (pyrrole to pyrrole-2-carboxylate) [19]. Although these
findings are generally explained by substrate inhibition, our kinetic plot for catechol does not
fit to this hypothesis. The drastic drop in reaction rate could better be explained by enzyme
deactivation at elevated substrate concentration. To verify this hypothesis, we performed a
deactivation test developed by M. J. Selwyn in 1954 [22]. The test is based on the fact that in
absence of enzyme deactivation, product formation over time is only a function of time and
enzyme concentration, if all the other parameters are kept constant. It follows that performing
enzymatic reactions with different enzyme amounts and plotting product concentration vs.
time multiplied by enzyme concentration, the points must coincide with the same curve. If
denaturation occurs, the enzyme amount itself becomes a function of time, and different
curves will be obtained. Figure 2 shows the results of the deactivation test at substrate
concentrations of 30 and 200 mM.
Reaction progress clearly shows that at high substrate concentrations the enzyme
concentration is not constant anymore, since higher conversions were obtained with higher
enzyme amounts. Recently, Ienaga et al. have shown that replacement of Tyr 64 and Phe 195
(
corresponding to Asn60 and Phe 189 in 2,6-dihydroxybenzoic acid decarboxylase from
Rhizobium sp.) by Thr and Tyr, respectively, around the active site of salicylic acid
decarboxylase from T. moniliiforme yields a mutant which reached the same maximal
conversion of p-aminosalicylic acid (70%) within a range of 10 to 200 mM of substrate [21].
This shows how key amino acid residues may be substituted resulting in enhanced protein
stabilization, possibly as a consequence of the introduction of additional OH groups (Thr64
and Tyr195). To assess the kinetic behavior with regard to bicarbonate, 100 mM of catechol
was used at saturating concentration (3.3 times Km,c) and the co-substrate concentration was
varied from 0 to 3 M (approx. saturation). Figure 3 shows the double substrate Michaelis-
Menten fit.
m
The non-linear fit revealed a K for bicarbonate (Km,b) equal to 839 ± 4. From the typical
rectangular hyperbola it can be concluded that enzyme kinetics are not significantly affected
by different ionic strength values. Complete evaluation of the enzyme kinetics requires the
catalytic properties in the reverse decarboxylation direction. Since in this case, bicarbonate
cannot be introduced, it was decided to buffer the system at pH 8 (the same value as in the
forward carboxylation direction) to counteract the pH shift due to the acidic substrate. Since
the reaction rate in the (downhill) decarboxylation direction was significantly higher as
This article is protected by copyright. All rights reserved.

compared to the (uphill) carboxylation direction, in a 20 minutes assay it was possible to
generate progress curves of the reaction, from which initial rates were derived. The one-
substrate kinetic fit is shown in Figure 4.
r
The catalytic constants Km,HA and V max for the decarboxylation reaction were 1.2 ± 0.3 mM
-1
-1
and 1 ± 0.1 µmol min mg . The catalytic parameters K
m max
, V , kcat and the specificity
constant K are summarized in Table 1.
a
In general, the enzyme appears more efficient in catalyzing the decarboxylation reaction, as
HA
shown by the K
a
, the combination of Km,HA and kcat. As previously assumed, the presence
f
of the lower V max is the result of the thermodynamically more difficult 'uphill' C-C bond
formation compared to the cleavage [19].
Activation Energy. The completion of the kinetic study requires the determination of the
activation energy of the rate limiting step in the overall mechanism. The maximum reaction
o
rate was measured at 20, 30, 40 and 50 C, which was accompanied by an increase in
carboxylation activities. The Arrhenius equation was used to calculate the activation energy
-1
-1
(
Figure S4), which resulted in 29 ± 0.2 kJ mol (7.00 ± 0.06 kcal mol ). It is interesting to
note that from theoretical studies on the Kolbe-Schmitt reaction for phenol carboxylation,
-1
similar activation energies of 5.4 and 10.3 kcal mol were calculated for the nucleophilic
attack of the phenolate anion to the electrophilic carbon atom of CO [24]. The close
2
similarity of the E of the chemical and the biological carboxylation further supports the
a
strong mechanistic relationship of benzoic acid decarboxylases with the Kolbe-Schmitt
reaction [12].
Thermodynamics. Determination of the equilibrium point, controlled by thermodynamics,
can be hampered by product inhibition and enzyme deactivation. Thus, long-term
experiments were performed using non-limiting enzyme amounts and varying the
concentration of the two substrates. In the absence of product inhibition or enzyme
deactivation, a linear correlation of maximal conversion is expected by increasing the
substrate (catechol) concentration at fixed bicarbonate concentration. Conversely, higher
conversion is expected at increasing bicarbonate concentration, while maintaining a fixed
substrate concentration. The enzyme concentrations employed in these set of experiments
ensured the occurrence of steady state concentrations after few hours of reaction. The results
This article is protected by copyright. All rights reserved.

are in accordance with previously reported trends from similar systems [16, 8, 19] and are
depicted in Figs. 5 and 8.
Figure 5 shows that a steady maximal conversion of approximately 23 % is obtained at 10 to
8
0 mM substrate concentration, where deactivation effects are negligible. In this range, it can
be assumed that the maximal conversion corresponds to the equilibrium conversion. Beyond
this threshold, the conversion drastically drops due to enzyme deactivation (Figure 2) and the
value is therefore depending on the enzyme concentration. For the converse experiment, a
fixed 10 mM concentration of catechol was used to measure the equilibrium shift at increased
amounts of bicarbonate. From the law of mass balance, the equilibrium constant for the
-4
-6
-1
carboxylation was calculated to be 1.6 10 ± 8.0 10 M . From this, it is possible to
calculate the difference in standard free energy at 303 K using the Van´t Hoff isotherm
0
-1
equation: ∆G = 5.2 ± 0.03 kcal mol .
Carboxylation under CO
hetero-aromatics under CO
2
pressure. The biocatalytic carboxylation of phenolics and
2
pressure (even supercritical conditions) was reported [13, 25].
To study the influence of gaseous CO on the enzyme, long-term experiments were
2
undertaken at pressures of 50 and 80 bar. To ensure the supercritical state of the gas at 80 bar,
tests were performed at 40°C. Additional tests were performed to exclude temperature effects
on enzyme stability and thermodynamics, as well as pH. Although water pressurized with
CO
units for both KP
.0) was used to balance the pH decrease caused by CO dissolution in the reaction media.
2
at >30 bar has a pH of 3 at room temperature [26], we found a drop of only 0.5 – 1 pH
i
and bicarbonate buffers pressurized by CO . Hence, KP buffer (0.2 M, pH
2
i
8
2
The results are shown in Table 2.
Close to the limits of accuracy, conversions of 5-7 % were found at 50 and 80 bar, indicating
no significant influence of dissolved CO concentration on the conversion. In contrast, a
conversion of 25% was achievable only in the presence of 2 M KHCO under a gas pressure
2
3
of 80 bar. According to literature data, the amount of dissolved carbon dioxide levels off at a
pressure of ~100 bar, which corresponds to ~1.2 M solution and at 50 and 80 bar, the
2
solubility of CO is 0.87 M and 1.1 M, respectively [27, 28], giving approximately half of the
bicarbonate concentration used in unpressurized experiments. Taking the pH of the
pressurized solution into account, the bicarbonate concentration can be estimated as 4 mM,
which is considerably lower than the corresponding KHCO
3
concentration used at ambient
This article is protected by copyright. All rights reserved.

conditions (from 0.1 to 3 M). These data underline that for 2,6-DHBA decarboxylase from
Rhizobium the co-substrate is bicarbonate rather than (dissolved) CO . From literature data, it
2
seems that the ability to use CO
2
is strongly dependent on the enzyme type, even though they
accept similar substrates and share similar biological role.
2
+
For instance, Mn -dependent phenol carboxylase from Thauera aromatica catalyzes the
para-carboxylation of phenyl phosphate under CO atmosphere (1 bar) in the presence of 100
mM of NaHCO [7]. Likewise, metal-independent pyrrole-2-carboxylate decarboxylase from
2
3
B. megaterium exhibited a higher reaction rate and conversion in the presence of 3 M KHCO
3
-
with a bell-shaped dependency on CO
2
pressure [25], although the enzyme uses HCO
3
as the
reactive species [19]. In contrast, the behavior of 2,6-DHBA decarboxylase from Rhizobium
is similar to that of 2,6-dihydroxybenzoic acid decarboxylase from Pandorea sp. acting on
3 2
resorcinol, where the conversion decreased in presence of KHCO under supercritical CO
pressure [13]. Influence of moderate pressure on enzymatic reactions is versatile and may
lead either to inhibitory effects or extra activation [29, 30]. In addition, drastic effects can be
2
observed in presence of organic (co)solvents, which strongly manipulate CO solubility [8].
The direct use of gaseous CO in biocatalytic carboxylation is currently being studied.
2
Modelling. The development and validation of a kinetic model for an enzymatic reaction
allows to verify the quality of the determined catalytic constants. Since the carboxylation
using bicarbonate as co-substrate consists of a formal addition of two substrates, selectively
binding to two different sites, forming a single product followed by water elimination, a bi-
uni mechanism is proposed. The proposed mechanism for DHBA decarboxylase assumes
2
+
binding of bicarbonate by Zn and the phenolic substrate by Asp 287 [12, 18].
Macroscopically the sequence of binding is not fixed and therefore we rather assume a
random mechanism; ordered binding being more typical for external cofactors-dependent
enzymes (e.g., NADH/NADPH-dependent reductions), where usually the coenzyme binds
first [31, 32]. Assuming that the C-C bond formation during the conversion of the ternary
enzyme-two-substrates complex to the enzyme-product complex is the rate-limiting step,
while other binding events are in rapid equilibrium, the following catalytic mechanism
(
Figure 6) and kinetic equation (Equation 1) can be drawn:
This article is protected by copyright. All rights reserved.

(
1)
The only unknown parameter in the equation is the dissociation constant of the ternary
b
complex (K ). At reaction equilibrium, the net reaction velocity equals zero, where the law
c
of mass action is related to the catalytic constants by the Haldane equation (Equation 2).
(
2)
b
From equation 2, the dissociation constant can be calculated as K
c
= 87 mM. The solutions
of the differential equation are compared with the experimental results obtained at different
reaction conditions (Figure 7).
The good agreement between experimental and theoretical values supports the proposed
kinetic mechanism as a viable model for the biocatalytic carboxylation. Moreover, as shown
in Figure 8, also the equilibrium positions can be modeled and predicted with very good
agreement.
DISCUSSION
The reversible regioselective carboxylation of catechol catalyzed by 2,6-dihydroxybenzoic
acid decarboxylase from Rhizobium sp. has been kinetically characterized and the kinetic
properties of the enzyme have been linked to the overall thermodynamic equilibrium, which
further strengthened the proposed mechanism (Scheme 2).
In comparison, related (de)carboxylases, such as pyrrole-2-carboxylate decarboxylase from
m m
Bacillus megaterium shows a K of 61 mM for its substrate pyrrole, a K of 560 mM for the
co-substrate bicarbonate and a Vmax in the forward (carboxylation) direction of 47.2 µmol
-1
-1
min mg [19]. In a related fashion, salicylic acid decarboxylase from T. moniliiforme shows
-2
-1
-1
K
m
and Vmax of 123 mM and 1.23 10 µmol min mg for phenol as substrate and a K
m
and
-1
-1
V
max of 1.08 mM and 0.47 µmol min mg in the reverse direction for the o-carboxylation
product [16]. For 2,3-dihydroxybenzoic acid decarboxylase from Aspergillus oryzae, K and
m
This article is protected by copyright. All rights reserved.

-1
-1
V
max values for 2,3-dihydroxybenzoic acid were 0.42 mM and 4.8 µmol min mg [33]. Our
results from 2,6-dihydroxybenzoic acid decarboxylase from Rhizobium sp. are in accordance
with these values, as higher Vmax and lower K values were generally found in the more
m
favorable 'downhill' direction. The fact, that irrespective of the type of phenolic or hetero-
aromatic substrates, the catalytic constants are approximately in the same order of magnitude,
even among different enzymes sources, is an example how catalytic constants evolved up to
the thermodynamic limit. This connection between catalytic parameters and thermodynamic
equilibrium has been pointed out by I. Segel in 1983 [34] and for this enzyme class, it can be
assumed that evolutionary events have tended to maximize the carboxylation efficiency by
m
increasing Vmax and decreasing K for the carboxylates, in order to leave the equilibrium
constants unchanged. The relationship between the catalytic parameters of an enzyme and the
thermodynamic equilibrium (equation 2) is meaningful only when the chemical equilibrium
is reached and in the case of the biocatalytic carboxylation this seems the case at relatively
low substrate concentrations, where no deactivation is present. The fact that conversions do
not exceed 7% under CO
the binding of (neutral) carbon dioxide to the enzyme and the low conversion under CO2
pressure is the result of the slow equilibrium between CO and bicarbonate. In contrast, the
latter clearly appears to be the co-substrate, which is favorably bound to the catalytically
2 m
pressure of 50 or 80 bar can be translated into an infinite K for
2
2
+
essential Zn ion in the active site. The preference of the enzyme for bicarbonate as co-
substrate (rather than CO ) can be explained by its higher availability in water.
2
The mechanism of enzymatic carboxylation shows a strong resemblance to that of the Kolbe-
Schmitt reaction (Scheme 2, A and B). Although the mechanism of the latter is still not fully
understood, it is generally accepted that the first step involves formation of a complex
between the reactants (phenolate anion and CO
24, 35]. This arrangement makes CO more electrophilic and puts it in position for
nucleophilic attack, and nicely explains the regioselectivity through a 'chelate-effect': ortho in
2
), linked by the positively charged metal (A)
[
2
+
+
case of small Na , para with large K [36, 37]. The formation of a phenyl hemicarbonate salt
as intermediate (not shown) fails to explain para-selectivity and is therefore less likely. C-C
Bond formation occurs via classic electrophilic aromatic substitution (Ar-S
resonance-stabilized non-aromatic intermediate, which finally regains aromaticity via
tautomerization. In contrast, in the biocatalytic equivalent (B), bicarbonate serves as CO
E
) through a
2
2
+
source, which is electronically activated by a divalent Zn atom serving as Lewis acid to
make it amenable to nucleophilic attack by catechol, which is deprotonated by Asp 287
(
assisted by His 218 and Glu 221) to enhance its electron density. The (postulated) non-
This article is protected by copyright. All rights reserved.

2
aromatic oxyanion intermediate regains aromaticity via elimination of H O and concomitant
tautomerization in a similar fashion as the chemical variant [18].
CONCLUSIONS
The kinetics and thermodynamics of the reversible non-oxidative carboxylation of catechol to
,3-dihydroxybenzoic acid catalyzed by 2,6-dihydroxybenzoic acid decarboxylase from
2
Rhizobium sp. were studied and a model based on a reversible random bi-uni mechanism was
proposed, which shows good correlation between experimental and theoretical values. In
view of the application of this regio-specific biocatalytic variant of the Kolbe-Schmitt
process, the following key limitations were identified: (i) limited conversions obtained at
high substrate concentrations were due to enzyme deactivation (rather than thermodynamic
limitations), (ii) instability of the phenolic substrate due to spontaneous air-oxidation and (iii)
unfavorable catalytic parameters.
In order to shift the unfavorable equilibrium into the 'uphill' (carboxylation) direction,
reaction engineering techniques are currently being developed towards the design of a
2
processes which makes use of CO as waste gas as C1 building block for the carboxylation of
phenol derivatives.
MATERIALS AND METHODS
General. All chemicals were purchased from Sigma Aldrich (Buchs, Switzerland).
Concentrations of substrate and product were determined by HPLC with UV detection using
calibration curves obtained with authentic reference material. Activity measurements were
performed by measuring five to six product/substrate concentration values in the appropriate
time range. All kinetic and thermodynamic measurements were conducted in duplicate. Non-
linear regression curve fits were performed minimizing the sum of the squares for the double
substrate Michaelis Menten model using Microsoft Excel (Excel 2010, Microsoft).
Mathematical solutions for the kinetic model were obtained using the ODE45 solver in
Matlab (Matlab 2010, Mathworks).
Overexpression and preparation of 2,6-dihydroxybenzoate decarboxylase. Chemically
competent Escherichia coli BL21(DE3) cells were transformed with the plasmidic vector
pET21a (+) containing the His-tagged DHBA decarboxylase gene from Rhizobium sp. and
heterologous expression was performed as previously reported [8]. For the preparation of
This article is protected by copyright. All rights reserved.

cell-free extracts (CFEs), wet cells (0.6 g/mL) were suspended in potassium phosphate (KP
i
)
buffer (50 mM, pH 7.0), the suspension was sonicated (5 mm sonication tip, 70 % power,
5
0% duty cycle, 3 min) on ice for three times and centrifuged at 25000 rpm (75,465 g) for 30
o
min at 4 C. Enzyme purification was carried out using a Nickel-NTA-agarose packed column
(
9 cm high, conical 0.8 x 4 cm). After washing the column with equilibration buffer (Tris 20
mM, NaCl 300 mM, imidazole 10 mM, pH 7.4) with 5 column volumes, the cell free extract
was loaded by gravity using a ratio of 0.5 mL of CFE per mL of chromatographic matrix.
Unspecific adsorbed proteins were removed by washing with 10 column volumes with
washing buffer (Tris 20 mM, NaCl 300 mM, imidazole 20 mM, pH 7.4), DHBA
decarboxylase was eluted using 5 column volumes of elution buffer (Tris 20 mM, NaCl 300
o
mM, imidazole 250 mM, pH 7.4). All purification steps were carried out at 4 C. Imidazole
i
was removed by ultrafiltration using exchange buffer (KP buffer, 50 mM, pH 7.0) for
storage. The biocatalyst was used either as purified enzyme or CFE.
General procedures for the enzymatic transformations. If not specified differently,
carboxylation reactions were carried out in HPLC screw capped glass vials using an
o
unbuffered potassium bicarbonate solution (pH 8, total volume 1 mL) with shaking at 30 C
and 500 rpm. After addition of potassium bicarbonate, ascorbic acid and catechol in the
desired concentration, the reactions were started by the addition of DHBA decarboxylase in
the appropriate amount. Data points were taken by diluting samples (50 μL) with a water-
acetonitrile solution (20:80, 450 μL) supplemented with trifluoroacetic acid (3% v/v) to stop
the reaction. The samples were centrifuged (5 min, 13000 rpm 15,870 g) and analyzed by RP-
HPLC. Decarboxylation reactions were performed under the same conditions apart from
using KP
Enzymatic transformation under pressurized CO
acid and decarboxylase CFE were mixed in KP (0.2 M buffer, pH 8.0) in a teflon test tube,
i
buffer 0.1 M, pH 8.0 as the reaction medium.
2
. Desired amounts of catechol, ascorbic
i
which was mounted to a Berghof RHS 295 high pressure autoclave equipped with heating
jacket, digital manometer and magnetic stirring, while the temperature was directly controlled
by a thermocouple. The autoclave was sealed and heated to 40 ±1 °C and carbon dioxide
preheated to 40 °C was slowly introduced to reach the target pressure. Once the desired
pressure and temperature were reached, stirring was turned on (500 rpm) and these conditions
were kept constant for 24 h. Then, the pressure was slowly released (5 bar/min) and the
temperature was lowered to room conditions.
This article is protected by copyright. All rights reserved.

Analytical methods. Catechol and 2,3-dihydroxybenzoic acid concentrations were measured
by an Agilent LC-1100 HPLC equipped with a diode array detector using an RP Phenomenex
°
Luna C18 column (150 x 4.60 mm, 5 µm) at 25 C. Water/trifluoroacetic acid (0.1 %) and
-1
acetonitrile/trifluoroacetic acid (0.1 %) with a flow rate of 1 mL min with a gradient (0-2
min 15%; 2-10 min 15-100%; 10-12 min 100%; 12-16 min 100-15%) were used. Catechol
and 2,3-dihydroxybenzoic acid were detected at 254 nm. Retention times were 6.2 min for
catechol and 6.8 min for 2,3-dihydroxybenzoic acid.
ACKNOWLEDGEMENTS
This work has been supported in part by the Austrian BMWFJ, BMVIT, SFG,
Standortagentur Tirol and ZIT through the FFG-COMET funding program.
ACKNOWLEDGEMENTS AUTHOR CONTRIBUTIONS
LP participated in planning the experiments, performed the experiments, analyzed the data
and wrote the paper. PG participated in planning and realization of the experiments under
carbon dioxide pressure and reviewed the paper. IS participated in planning the experiments
under carbon dioxide pressure and reviewed the paper. SG and KF provided the plasmidic
vector, participated in planning the experiments and wrote the paper. AL participated in
planning the experiments, analyzed the data and wrote the paper.
REFERENCES
1
. Schmitt R (1885). Beitrag zur Kenntnis der Kolbe´schen Salicylsäure Synthese. J. Prakt.
Chem. 31, 397–411.
2
. Baine O, Adamson G F, Barton J W, Fitch J L, Swayampati D R & Jeskey H (1954). A
study of the Kolbe-Schmitt reaction. II. The carbonation of phenols. J. Org. Chem. 19, 510–
5
3
4
14.
. Lindsey A & Jeskey H (1957). The Kolbe-Schmitt reaction. Chem. Rev. 57, 583–620.
. Stark A, Huebschmann S, Sellin M, Kralisch D, Trotzki R & Ondruschka B (2009).
Microwave-Assisted Kolbe-Schmitt Synthesis Using Ionic Liquids or Dimcarb as Reactive
Solvents. Chem. Eng. Technol. 32, 1730–1738.
This article is protected by copyright. All rights reserved.

5
. Hessel V, Krtschil U, Hessel V, Reinhard D & Stark A (2009). Flow Chemistry of the
Kolbe-Schmitt Synthesis from Resorcinol: Process Intensification by Alternative Solvents,
New Reagents and Advanced Reactor Engineering. Chem. Eng. Technol. 32, 1774–1789.
6
. Lack A & Fuchs G (1992). Carboxylation of phenylphosphate by phenol carboxylase, an
enzyme system of anaerobic phenol metabolism. J. Bacteriol. 174, 3629–3636.
7
. Aresta M, Quaranta E, Liberio R, Dileo C & Tommasi I (1998) Enzymatic synthesis of 4-
OH-benzoic acid from phenol and CO2: the first example of a biotechnological application of
a Carboxylase enzyme. Tetrahedron 54, 8841–8846.
8
. Wuensch C, Schmidt N, Gross J, Grischek B, Glueck SM & Faber K (2013). Pushing the
equilibrium of regio-complementary carboxylation of phenols and hydroxystyrene
derivatives. J. Biotechnol. 168, 264–270.
9
. Brackmann R & Fuchs G (1993). Enzymes of anaerobic metabolism of phenolic
compounds. Eur. J. Biochem. 213, 563–571.
1
0. Zhang X & Young LY (1997). Carboxylation as an initial reaction in the anaerobic
metabolism of naphthalene and phenanthrene by sulfidogenic consortia. Appl. Environ.
Microbiol. 63, 4759–4764.
1
1. Wuensch C, Glueck SM, Gross J, Koszelewski D, Schober M & Faber K (2012).
Regioselective Enzymatic Carboxylation of Phenols and Hydroxystyrene Derivatives. Org.
Lett. 14, 1974–1977.
1
2. Goto M, Hayashi H, Miyahara I, Hirotsu K, Yoshida M & Oikawa T (2006). Crystal
Structures of Nonoxidative Zinc-dependent 2,6-Dihydroxybenzoate ( γ-Resorcylate)
Decarboxylase from Rhizobium sp. Strain MTP-10005. J. Biol. Chem. 281, 34365–34373.
1
3. Matsui T, Yoshida T, Yoshimura T & Nagasawa T (2006). Regioselective carboxylation
of 1,3-dihydroxybenzene by 2,6-dihydroxybenzoate decarboxylase of Pandoraea sp. 12B-2.
Appl. Microbiol. Biotechnol. 73, 95–102.
14. Yoshida M, Fukuhara N & Oikawa T (2004). Thermophilic, Reversible γ-Resorcylate
Decarboxylase from Rhizobium sp. Strain MTP-10005: Purification, Molecular
Characterization, and Expression. J. Bacteriol. 186, 6855–6863.
1
5. Santha R, Savithri HS, Rao A & Vaidyanathan CS (1995). 2,3–Dihydroxybenzoic Acid
Decarboxylase from Aspergillus niger. A novel decarboxylase. Eur. J. Biochem. 230, 104–
1
1
10.
6. Kirimura K, Gunji H, Wakayama R, Hattori T & Ishii Y (2010). Enzymatic Kolbe–
Schmitt reaction to form salicylic acid from phenol: Enzymatic characterization and gene
identification of a novel enzyme, Trichosporon moniliiforme salicylic acid decarboxylase.
Biochem. Biophys. Res. Comm. 394, 279–284.
1
7. Yoshida T, Hayakawa Y, Matsui T & Nagasawa T (2004). Purification and
characterization of 2,6-dihydroxybenzoate decarboxylase reversibly catalyzing nonoxidative
decarboxylation. Arch. Microbiol. 181, 391–397.
This article is protected by copyright. All rights reserved.

1
(
8. Wuensch C, Gross J, Steinkellner G, Lyskowski A, Gruber K, Glueck SM & Faber K
2014). Regioselective ortho-carboxylation of phenols catalyzed by benzoic acid
decarboxylases: a biocatalytic equivalent to the Kolbe–Schmitt reaction. RSC Adv. 4, 9673-
9
679.
1
9. Wieser M, Fujii N, Yoshida T & Nagasawa T (1998). Carbon dioxide fixation by
reversible pyrrole-2-carboxylate decarboxylase from Bacillus megaterium PYR2910. Eur. J.
Biochem. 257, 495–499.
2
0. Ramsden CA, Riley PA (2014). Tyrosinase: the four oxidation states of the active site and
their relevance to enzymatic activation, oxidation and inactivation. Bioorg. Med. Chem. 22,
2
2
388–2395.
1. Ienaga S, Kosaka S, Honda Y, Ishii Y & Kirimura K (2013). p-Aminosalicylic Acid
Production by Enzymatic Kolbe-Schmitt Reaction Using Salicylic Acid Decarboxylases
Improved through Site-Directed Mutagenesis. Bull. Chem. Soc. Jpn. 86, 628–634.
2
2. Selwyn MJ (1954). A simple test for inactivation of an enzyme during assay. Biochim.
Biophys. Acta 105, 193–195.
2
3. Pierpont CG & Lange CW (1994). The Chemistry of Transition Metal Complexes
Containing Catechol and Semiquinone Ligands. Progr. Inorg. Chem. 41, 331–442.
2
4. Markovic Z, Engelbrecht JP & Markovic S (2002). Theoretical study of the Kolbe-
Schmitt reaction mechanism. Z. Naturforsch A Phys Sci 57, 812–818.
2
5. Matsuda T, Ohashi Y, Harada T, Yanagihara R, Nagasawa T & Nakamura K (2001).
Conversion of pyrrole to pyrrole-2-carboxylate by cells of Bacillus megaterium in
supercritical CO2. Chem. Commun., 2194–2195.
2
6. Hofland GW, Rijke A de, Thiering R, van der Wielen,Luuk A.M & Witkamp G (2000).
Isoelectric precipitation of soybean protein using carbon dioxide as a volatile acid. J.
Chromatogr. B 743, 357–368.
2
7. Bamberger A, Sieder G & Maurer G (2000). High-pressure (vapor+liquid) equilibrium in
binary mixtures of (carbon dioxide+water or acetic acid) at temperatures from 313 to 353 K.
J. Supercrit. Fluids 17, 97–110.
2
8. Dodds WS, Stutzman LF & Sollami BJ (1956). Carbon Dioxide Solubility in Water. Ind.
Eng. Chem. Chem. Eng. Data Series 1, 92–95.
2
9. Jaenicke R (1981). Enzymes under extremes of physical conditions. Ann. Rev. Biophys.
Bioeng. 10, 1–67.
3
0. Brummund J, Meyer F, Liese A, Eggers R & Hilterhaus L (2011). Dissolving carbon
dioxide in high viscous substrates to accelerate biocatalytic reactions. Biotechnol. Bioeng.
1
08, 2765–2769.
3
1. Gerhard M & Schomburg D (2013) Biochemical pathways: an atlas of biochemistry and
molecular biology 2nd edition, John Wiley and sons.
3
2. Everse J (1982) The Pyridine nucleotide coenzymes. Elsevier.
This article is protected by copyright. All rights reserved.

3
3. Santha R, Appaji Rao N & Vaidyanathan C (1996). Identification of the active-site
peptide of 2,3-dihydroxybenzoic acid decarboxylase from Aspergillus oryzae. Biochim.
Biophys. Acta 1293, 191–200.
3
4. Schmidt ND, Peschon JJ & Segel IH (1983). Kinetics of enzymes subject to very strong
product inhibition: Analysis using simplified integrated rate equations and average velocities.
J. Theor. Biol. 100, 597–611.
3
5. Hales JL, Jones JI & Lindsey AS (1954). Mechanism of the Kolbe-Schmitt reaction. Part
I. Infra-red studies. J. Chem. Soc., 3145–3151.
3
6. Hirao I & Kito T (1973). Carboxylation of Phenol Derivatives. XXI. The Formation
Reaction of the Complex from Alkali Phenoxide and Carbon Dioxide. Bull. Chem. Soc. Jpn.
4
6, 3470–3474.
3
7. There is evidence that a part of potassium p-hydroxybenzoate is derived by rearrangement
+
+
of initially formed K salicylate, while Na salicylate does not rearrange, see: Ota K (1974).
The Conversion Reaction of Alkali 4-Hydroxyisophthalates into Hydroxybenzoic Acids.
Bull. Chem. Soc. Jpn. 47, 2343–2344.
Tables
Table 1. Catalytic parameters for enzymatic (de)carboxylation.
Carboxylation
f
K
m
V
max
k
cat
K
a
-
1
-1
-1
-1 -1
(
mM)
(µmol min mg )
(s )
(mM s )
-
2
-3
catechol bicarbonate
0.35 ± 1.1 10
0.12 ± 3.0 10
catechol
bicarbonate
c
b
a
(
K )
m,c
(Km,b
)
(K
a
)
(K
)
-
3
-4
3
0 ± 3
839 ± 4
4.2 10
±
1.5 10
±
-
4
-6
6.5 10
5.7 10
Decarboxylation
r
HA
K
m
V
max
k
cat
K
a
-1
-1
-1
-1 -1
(
mM)
(µmol min mg )
(s )
(mM s )
DHBA (Km,HA
)
-2
1
.2 ± 0.3
1 ± 0.1
0.52 ± 0.025
0.44 ± 9 10
This article is protected by copyright. All rights reserved.

Table 2. Carboxylation under CO
2
pressure. Reactions were performed using 10 mM of
°
catechol and 10 mM of ascorbic acid in KP buffer (0.2 M pH 8.0) for 24h at 40 C and 500
i
rpm.
CO
2
pressure (bar)
Max. conversion (%)
5
8
0
0
5
7
8
0, KHCO (2M)
25
3
Schema
2
,6-dihydroxybenzoic acid decarboxylase
OH
OH
from Rhizobium sp.
OH
OH
HCO -
3
CO H
2
catechol
2
,3-dihydroxybenzoic
acid
Scheme 1. Enzymatic carboxylation (forward) and decarboxylation (reverse reaction) of
catechol catalyzed by DHBA decarboxylase using bicarbonate as CO source.
2
This article is protected by copyright. All rights reserved.

A
Na
Na
H
Na
Na
H
δ-
δ+
O
O
C
O
O
O
HO
O
O
O
O
O
O
+
2 resonance structures
B
Asp287
O
Asp287
O O
Zn2
O
Asp287
O
Zn2
O
OH
O
H
O
O
H
O
Zn2
O
O
O
HO
OH
HO
HO
O
OH
H
oxyanion intermediate
Scheme 2. Comparison of the key steps in the mechanism of the chemical (A) and
biological (B) Kolbe-Schmitt reaction.
This article is protected by copyright. All rights reserved.

Fig. 1. Initial rate measurements from carboxylation of catechol and Michaelis-Menten
non-linear fit. Reactions were performed using different catechol concentrations and
stoichiometric amounts of ascorbic acid, 2 M KHCO
0°C and 500 rpm. Dotted line: non-linear fit using the double substrate Michaelis-Menten
model.
3
, 0.2-0.3 mg/mL of decarboxylase, at
3
This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.

Fig. 2. Deactivation test. Progress curves obtained with varying amounts of DHBA
decarboxylase CFE at fixed catechol concentrations of 30 mM (A) and 200 mM (B); black
diamonds: 60 µL (0.21 mg/mL), white squares: 80 µL (0.28 mg/mL), black circles: 100 µL
(
3
0.35 mg/mL). The reactions were performed using 2 M of KHCO at 30°C and 500 rpm.
This article is protected by copyright. All rights reserved.

Fig. 3. Michaelis-Menten non-linear fit for hydrogen carbonate in the carboxylation of
catechol. Reactions were performed at varying KHCO concentrations using 100 mM of
3
catechol, 0.3-0.4 mg/mL of decarboxylase at 30°C and 500 rpm. Dotted line: non-linear fit
using double substrate Michaelis-Menten model.
This article is protected by copyright. All rights reserved.

Fig. 4. Michaelis-Menten non-linear fit for the decarboxylation of 2,3-dihydroxybenzoic
acid. Reactions were performed using various concentrations of DHBA, KP buffer 0.1 M at
i
pH 8.0, 0.2-0.3 mg/mL of decarboxylase, and equimolar amounts of ascorbic acid at 30°C
and 500 rpm. Dotted line: non-linear fit using single substrate Michaelis-Menten model.
This article is protected by copyright. All rights reserved.

Fig. 5. Long-term experiments at fixed bicarbonate and varying catechol
concentrations. Reactions were performed at 30°C, 500 rpm in the presence of 2 M of
3
KHCO with varying catechol concentration, decarboxylase concentrations ranging from 0.5
to 1 mg/mL.
This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.

Fig. 6. Kinetic equation and catalytic steps for a rapid equilibrium random and
reversible bi-uni mechanism for biocatalytic carboxylation. HA: 2,3-dihydroxybenzoic
acid; E: enzyme = DHBA decarboxylase; C: substrate catechol; B: co-substrate bicarbonate;
f
r
V
max: maximum rate in carboxylation (forward) direction; V max: maximum rate in
decarboxylation (reverse) direction; K
constant for bicarbonate; K
dissociation constant of ternary enzyme-two-substrate complex; k
= kcat in reverse direction
m
,
c
: Michaelis constant for catechol; K
m
,
b
: Michaelis
b
m
,
HA: Michaelis constant for 2,3-dihydroxybenzoic acid; K
c
:
f
= kcat in forward direction;
k
r
This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.

Fig. 7. Simulation of batch reactions and comparison with experimental data.
Continuous lines result from simulation. A: catechol 40 mM, KHCO 2 M, DHBA
decarboxylase 0.22 mg/mL; B: catechol 60 mM, KHCO 2 M, decarboxylase 0.25 mg/mL;
3
3
3
C: catechol 30 mM, KHCO 1 M, decarboxylase 0.25 mg/mL.
This article is protected by copyright. All rights reserved.

Fig. 8. Comparison of experimental (white rectangles) and simulated (black line)
equilibrium conversions. Reactions were performed at 30°C, 500 rpm in the presence of 10
mM of catechol with various KHCO
3
starting concentrations and decarboxylase
concentrations ranging from 0.5 to 1 mg/mL.
This article is protected by copyright. All rights reserved.