Biocatalytic Parallel Interconnected Dynamic Asymmetric Disproportionation of α-Substituted Aldehydes: Atom-Efficient Access to Enantiopure (S)-Profens and Profenols

Article abstract of DOI:10.1002/adsc.201800541

The biocatalytic asymmetric disproportionation of aldehydes catalyzed by horse liver alcohol dehydrogenase (HLADH) was assessed in detail on a series of racemic 2-arylpropanals. Statistical optimization by means of design of experiments (DoE) allowed the identification of critical interdependencies between several reaction parameters and revealed a specific experimental window for reaching an ′optimal compromise′ in the reaction outcome. The biocatalytic system could be applied to a variety of 2-arylpropanals and granted access in a redox-neutral manner to enantioenriched (S)-profens and profenols following a parallel interconnected dynamic asymmetric transformation (PIDAT). The reaction can be performed in aqueous buffer at ambient conditions, does not rely on a sacrificial co-substrate, and requires only catalytic amounts of cofactor and a single enzyme. The high atom-efficiency was exemplified by the conversion of 75 mM of rac-2-phenylpropanal with 0.03 mol% of HLADH in the presence of ～0.013 eq. of oxidized nicotinamide adenine dinucleotide (NAD⁺), yielding 28.1 mM of (S)-2-phenylpropanol in 96% ee and 26.5 mM of (S)-2-phenylpropionic acid in 89% ee, in 73% overall conversion. Isolated yield of 62% was obtained on 100 mg-scale, with intact enantiopurities. (Figure presented.).

Full text of DOI:10.1002/adsc.201800541

Title: Biocatalytic Parallel Interconnected Dynamic Asymmetric
Disproportionation of α-Substituted Aldehydes: Atom-Efficient
Access to Enantiopure (<i>S</i>)-Profens and Profenols
Authors: Erika Tassano, Kurt Faber, and Mélanie Hall
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Adv. Synth. Catal. 10.1002/adsc.201800541
Link to VoR: http://dx.doi.org/10.1002/adsc.201800541

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
UPDATE
DOI: 10.1002/adsc.201((will be filled in by the editorial staff))
Biocatalytic Parallel Interconnected Dynamic Asymmetric
Disproportionation of -Substituted Aldehydes: Atom-Efficient
Access to Enantiopure (S)-Profens and Profenols
Erika Tassanoa, Kurt Fabera and Mélanie Halla,*
aDepartment of Chemistry, University of Graz, Heinrichstrasse 28, 8010 Graz, Austria
Email: melanie.hall@uni-graz.at; Tel: +43-316-380-5479; Fax.: +43-316-380-9840
Received: ((will be filled in by the editorial staff))
Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/adsc.201######.
Abstract. The biocatalytic asymmetric disproportionation of conditions, does not rely on a sacrificial co-substrate, and
aldehydes catalyzed by horse liver alcohol dehydrogenase requires only catalytic amounts of cofactor and a single
(HLADH) was assessed in detail on a series of racemic 2- enzyme. The high atom-efficiency was exemplified by the
arylpropanals. Statistical optimization by means of design of conversion of 75 mM of rac-2-phenylpropanal with 0.03
experiments (DoE) allowed the identification of critical mol% of HLADH in the presence of ~ 0.013 eq. of oxidized
interdependencies between several reaction parameters and nicotinamide adenine dinucleotide (NAD+), yielding 28.1
revealed a specific experimental window for reaching an mM of (S)-2-phenylpropanol in 96% ee and 26.5 mM of
'optimal compromise' in the reaction outcome. The (S)-2-phenylpropionic acid in 89% ee, in 73% overall
biocatalytic system could be applied to a variety of 2- conversion. Isolated yield of 62% was obtained on 100 mg-
arylpropanals and granted access in a redox-neutral manner scale, with intact enantiopurities.
to enantioenriched (S)-profens and profenols following a
parallel interconnected dynamic asymmetric transformation
(PIDAT).
The reaction can be performed in aqueous buffer at ambient
Keywords:
Alcohol
dehydrogenase;
asymmetric
disproportionation; biocatalysis; Cannizzaro; design of
experiments; profens.
the hydride transfer operates through the cofactor as a
hydride shuttle (Scheme 1). Importantly, the
oxidative half-reaction proceeds on the hydrate form
of the aldehyde,[8] which – if the hydration is rate
Introduction
The Cannizzaro reaction is
a
redox-neutral
disproportionation reaction, in which two molecules
of an aldehyde undergo concurrent oxidation and
reduction, yielding an equimolar mixture of
corresponding alcohol and carboxylic acid.[1] In the
classical procedure, rather harsh conditions are
required, such as use of a strong base (e.g. alkali
metal hydroxide) and high temperature, which overall
limit the protocol to non-enolizable aldehydes. The
generally accepted mechanism starts with the
hydration of the aldehyde followed by deprotonation
by a strong base; the generated unstable dianionic
species then transfers a hydride directly onto a second
aldehyde molecule, leading to a carboxylate and an
alkoxide that gets protonated by the solvent to yield
the alcohol (Scheme 1A).[2] Various modifications
have been implemented in the past decades, as for
example, solvent-free conditions,[3] microwave
irradiation,[4] organo-base mediation[5] or Lewis acid
catalysis.[3a,6]
We established a biocatalytic formal Cannizzaro
reaction[7] that relies on the concurrent oxidative and
reductive activity of nicotinamide-dependent alcohol
dehydrogenases (ADHs) to run in an overall redox
neutral fashion. In contrast to the chemical variant,
Scheme 1. Disproportionation of A) non-enolizable
aldehydes in the base-catalyzed Cannizzaro reaction[1,2]
and B) racemic 2-arylpropanals (rac-1a-g) in the
biocatalytic formal Cannizzaro reaction catalyzed by
alcohol dehydrogenase (ADH).
1
This article is protected by copyright. All rights reserved.

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
limiting – induces a lag in the coupling of the two
redox reactions and thus furnishes a sub-optimal ratio
of products.[7] The biocatalytic system exists
therefore because of the oxidative catalytic activity of
ADHs on geminal diols, which, although well-
studied,[9] remains largely in the shadow of reductive
processes.
loading along with high conversion). First, a range of
experiments on the model substrate 2-phenylpropanal
was performed to identify the key elements impacting
the four reaction outputs. Thereafter, a design of
experiments (DoE) statistical analysis was conducted
to determine potential (cross-) interactions between
the different reaction parameters that impact the
overall outcome of the bioconversion. Finally, the
scope of the reaction was evaluated with a particular
focus on compounds yielding enantioenriched 2-
arylpropanols (profenols) – useful intermediates in
medicinal chemistry and fragrances[18] – and 2-
arylpropanoic acids (profens) – compounds belonging
to an important subclass of NSAIDs (non-steroidal
anti-inflammatory drugs) with pharmacological
activity attributed to the (S)-form.[19]
Owing to the limited scope of the chemical
reaction, cases of stereoselective Cannizzaro reaction
are rare. Only a few intramolecular versions have
been reported in literature to date, almost all applying
arylglyoxals as starting material and exploiting Lewis
acids complexed by sterically hindered chiral
ligands.[10] On the other hand, several ADHs showed
significant enantioselectivity in the oxidation of -
substituted aldehydes[11] and we could exploit this
feature to develop an intermolecular asymmetric
biocatalytic formal Cannizzaro reaction employing a
single ADH[7] (Scheme 1B). From a synthetic
standpoint, the system possesses attractive features: i)
Overall redox-neutral, the reaction requires only
catalytic amounts of cofactor; ii) no sacrificial
auxiliary substrate (coupled-substrate approach) nor a
second enzyme (coupled-enzyme approach)[12] are
necessary to regenerate the nicotinamide cofactor; iii)
starting from a single racemic compound, which
spontaneously racemizes in the reaction buffer, two
Results and Discussion
Analysis of the reaction on model substrate 2-
phenylpropanal (1a)
The disproportionation of 1a was previously
evaluated with a panel of alcohol dehydrogenases
(ADHs), and it was noted that horse liver ADH
(HLADH) provided both (S)-configurated products
with high stereoselectivity.[7] Under the tested
conditions (0.5 mg/mL HLADH – ~ 13 µM – and 0.1
eq. each of NADH and NAD+), HLADH showed a
predominant reductive half-reaction (ratio 3a/2a
0.38) and a poor overall conversion (12%, Figure 1,
sample 1). A 10-fold increase in the biocatalyst
amount (5 mg/mL, i.e. ~1.3 mol%) led to higher
conversion with modest erosion of the enantiopurity
of the alcohol (Figure 1, sample 2). However, the
high cofactor loading applied appeared in
contradiction to the catalytic amount theoretically
needed (i.e. only 1 equivalent of cofactor per enzyme
due to internal recycling). Furthermore, we assumed
that the oxidation state of the nicotinamide cofactor
would play a crucial role to achieve a balanced ratio
of products. Reducing the amount of NAD+ to 0.05
eq. at low enzyme concentration (13 µM) led to an
unexpected 5-fold increase in conversion level
(Figure 1, sample 4) and a more balanced product
ratio (3a/2a ~ 0.76). Hence, providing only NAD+
seems key to achieving the latter and is likely
associated with sufficient aldehyde hydration
required for the first oxidative half-reaction to take
place (vide infra). However, a dramatic drop in the
stereoselectivity of both half-reactions was observed.
Next, the substrate loading was increased (Figure 1,
samples 5-7). Remarkably, HLADH sustained
aldehyde concentrations up to 75 mM, typically not
well tolerated by enzymes due to the formation of
Schiff bases with free amino groups of surface amino
acids,[20] which leads to protein deactivation. This
translated into markedly improved product
concentrations (19.6 mM 2a and 18.6 mM 3a),
excellent enantioselectivity towards both alcohol and
carboxylic acid [(S)-2a in 97% ee and (S)-3a in 86%
ee] and well balanced product ratio (3a/2a 0.95).
enantioenriched
products
(not
necessarily
homochiral) can be obtained via dynamic kinetic
resolution. While this in theory should translate into
highly atom-efficient biotransformations, our initial
set-up applied to 2-phenylpropanal (10 mM) still
required ~ 1 mol% catalyst (5 mg/mL) and 0.2 eq. of
cofactor in form of a 1:1 NADH/NAD+ mixture to
reach close to full conversion (Table 1, entry 6).[7]
Optimization of the reaction is associated with the
necessary consideration of the four outputs: i)
conversion level (highest possible), ii) ratio of
products (close to 1:1), iii) enantiopurity of the
alcohol and iv) enantiopurity of the carboxylic acid
(both ideally >95%). Because several parameters
affect the process simultaneously (e.g. amount of
catalyst and cofactor, substrate concentration), the
outcome of the biocatalytic disproportionation may
be efficiently improved through statistical analysis.
The use of chemometrics in biotechnology has been
increasing over the last decade,[13] and the
employment of statistical experimental design[14] in
particular can contribute to process optimization
while minimizing the number of experiments.[15] For
instance, the production of an oligosaccharide by
yeast whole cells could be improved (in terms of
yield and product concentration) by using a fractional
factorial statistical design,[16] while Kara et al.
similarly applied this method to enhance the
productivity of
a
bi-enzymatic cascade for
caprolactone synthesis.[17] Herein, we present this
approach to extend the asymmetric ADH-catalyzed
disproportionation to
a range of racemic 2-
arylpropanals under synthetically relevant conditions,
aiming at elevated substrate concentration and high
atom-efficiency (i.e. low catalyst- and cofactor
2
This article is protected by copyright. All rights reserved.

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
2a (mM)
2a (mM)
3a (mM)
3a (mM)
2a ee%
2a ee%
3a ee%
3a ee%
80
70
60
50
40
30
20
10
100
80
60
40
20
0
Sample HLADH (µM)
Cofactor (mM)
1:1 NADH/NAD+
1:1 NADH/NAD+
1:1 NADH/NAD+
0.5 NAD+
79%
0.73
1a
2a
3a
4
13
73%
0.94
128
256
13
56%
51%
0.95
5
13
0.5 NAD+
0.73
40%
0.78
6
13
0.5 NAD+
46%
7
13
0.5 NAD+
97% >99%
0.80
51%
0.5:0.5 NADH/NAD+
0.5 NAD+
0.87
8
13
1.41
12%
0.38
0.76
9b
10c
13 + 13
13 + 13
0
1
2
3
4
5
6
7
8
9
10
Samples
0.5 + 0.5 NAD+
[1a]
10
30
50
75
mM)
Figure 1. HLADH-catalyzed asymmetric disproportionation of rac-1a. Conditions: phosphate buffer (50 mM, pH 7.5),
30 °C, 120 rpm, 24 h (see experimental section); entries 9-10: 4 vol% MTBE; 13 µM HLADH corresponds to 0.5 mg/mL;
conversions (%) to (alcohol+acid) (bold black numbers on top of the bars) and ee values were determined by chiral GC
analysis after product extraction (see Supporting Information); carboxylic acid to alcohol product ratio (bold orange
c)
numbers in the bars). a) Data from ref. 7. b) After 24 h, a second aliquot of enzyme was added; total reaction time 48 h.
After 24 h, a second aliquot of enzyme and cofactor was added; total reaction time 48 h.
Only modest increase in conversion along with out-
of-balance product ratio were observed with the pair
of cofactors (Figure 1, sample 8). Collectively, the
data suggest that combining higher substrate
concentration (75 mM) with catalytic amounts of
NAD+ (0.007 eq.) and low enzyme concentration (~
0.02 mol%) offers an excellent compromise to
maximize all reaction outputs, i.e. high conversion,
ideal product ratio and excellent stereocontrol in both
redox reactions.
Finally, a further improvement of the overall
conversion was achieved through the addition of
fresh enzyme after 24 h (Figure 1, samples 9-10),
with excellent stereocontrol retained in both half-
reactions. Moreover, a concomitant addition of NAD+
after 24 h (Figure 1, sample 10) proved to be crucial
for a well-balanced product ratio (73% overall
conversion, with 28.1 mM of (S)-2a in 96% ee and
26.5 mM of (S)- 3a in 89% ee). Both
supplementations likely compensate for slow enzyme
deactivation and (spontaneous) cofactor degradation
over time, respectively.
limited to low substrate concentration (0.5 mM, Table
1, entry 1), ADH-10 from Sulfolobus solfataricus was
applied on a 5 mM scale and yielded (S)-2a in high
enantiopurity with only 0.01 eq. NADH (Table 1,
entry 2). In conjunction with 1,4-butanediol as 'smart'
co-substrate,[22] the efficiency of HLADH could be
increased on a 5 mM scale using 0.02 eq. NADH
(98% conversion, Table 1, entry 3). Hollmann et al.
then developed an oxidative variant based on a
coupled-enzyme approach, which relied on NAD(P)H
oxidase (NOX) for cofactor recycling.[11] Only a few
cofactor turnovers could be reached at high NAD+
concentration and 5 mM 1a, which was attributed to
non-overlapping process windows of both enzymes
(Table 1, entries 4-5). In this regard, the parallel
interconnected dynamic asymmetric transformation
(PIDAT)[23] developed here is highly competitive
among all known biocatalytic DKRs of 1a, with
lowest cofactor concentration needed, highest product
concentration and highest TTN of the biocatalyst
reached.
The disproportionation of 1a was performed on a
semi-preparative scale (100 mg) according to the
improved procedure (Figure 1, entry 10). After
isolation and purification of the products, 31 mg of
(S)-2a in 97% ee and 35 mg of (S)-3a in 86% ee (no
racemisation, data not shown) were obtained in an
overall isolated yield of 62% and perfect product
ratio of 1:1.03 (see experimental section for details).
The overall performance of the process favorably
compares to related studies on the combination of
ADH-mediated dynamic kinetic resolution (DKR) of
rac-2-arylpropanals, such as rac-1a, with cofactor
recycling (Table 1). Reductive DKR[21a] was first
developed in a coupled-substrate approach using
ethanol as co-substrate.[18a,21b] While HLADH was
3
This article is protected by copyright. All rights reserved.

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
Table 1. ADH-mediated DKR approaches to enantiopure profen derivatives with cofactor recycling
Entrya)
[1a]
(mM)
0.5
[Product]
(mM)
[NAD+]
(mM / eq.) Recyclingb)
NAD
[ADH]
(mg/mL)
TONNAD TTNADH eeproduct
(S) %
1[18a]
2[21b]
3[22]
4[11]
5[11]
6[7]
0.27 (2a)
0.01 / 0.02 Coupled-substrate 0.01
44
1038
n.d.
1884
5
90 (2a)
98 (2a)
95 (2a)
50 (3a)
88 (3a)
EtOHc)
0.05 / 0.01 Coupled-substrate n.r.d)
EtOH
5
3.7 (2a)
4.9 (2a)
2.4 (3a)
3.2 (3a)
74
5
0.1 / 0.02
Coupled-substrate 0.1
1,4-butanediol
49
5
2 / 0.4
Coupled-enzyme
NOX
20
37.5
5
1.2
3.2
4.9f)
26.5i)
5
1 / 0.2
Coupled-enzyme
NOX
PIDAT
3
10
75
5.2 (2a)
1e) / 0.1
1h) / 0.01
76
93 (2a)
88 (3a)
96 (2a)
89 (3a)
+4.5 (3a)
28.1 (2a)
+26.5 (3a)
7g)
PIDAT
1h)
2100
a)
Data from literature, see references. b) NOX: NAD(P)H oxidase; PIDAT: Parallel Interconnected Dynamic Asymmetric
c)
d)
Transformation (one enzyme, no co-substrate). 10% THF as co-solvent. n.r. not reported (0.5 mU/mL on benzyl
alcohol; enzyme obtained with 1.4 U per liter culture). e) NADH/NAD+, 1:1, 1 mM each. f) TON of each redox state of
nicotinamide molecule (NADH/NAD+ pair). g) 4 vol% MTBE. h) Total amount (added in two equal portions over 24 h, total
reaction time 48 h). i) Each turn-over converts two molecules of substrate. n.d. not determined.
Design of experiments
0.01 eq.) allows a tight control of the PIDAT as all
steps are connected to the oxidative half-reaction
taking place first (Figure 1).
Besides the two redox half-reactions, the system
includes two equilibria involving the aldehyde
substrate, i.e. racemization via keto-enol tautomerism
and hydration (Scheme 2). To reach full conversion
of rac-1a to two enantiopure products, e.g. (S)-2a and
(S)-3a, two main requirements must be fulfilled:
Given the intricacy of all steps involved (Scheme
2), a concomitant optimization of four responses
(conversion level, ratio of products, enantiopurity of
alcohol and carboxylic acid) appears challenging. A
multivariate statistical analysis was thus performed
by means of design of experiments (DoE), using
Design-Expert® from Stat-Ease, Inc. As described
above, three parameters (variables) were considered
for the statistical analysis (Table 2): concentration of
the enzyme (A), of the oxidized nicotinamide
cofactor (B), and of the substrate (C), while
temperature (30 °C), buffer (KPi, pH 7.5) and
reaction time (24 h) were kept constant. The selected
variables were employed for the development of a
Central Composite Design (CCD), which is a
Response Surface Method (RSM) that analyzes the
variables at five different levels and which strength
red
red
red
red
(1) kR << kS ≤ krac (kR < krac < kS has been
shown possible[24]
)
According to (1), a dynamic kinetic resolution
(DKR) would establish a progressive and desired
accumulation of (S)-2a.
(2) kRox << kSox < khyd < krac (ideal: kSox + khyd ≈ kS
)
red
Since the oxidation is non-reversible, the
consumption of the hydrate form of (S)-1a controls
the equilibria by pulling the overall system towards
hydrate formation of the faster reacting enantiomer,
which is the underlying reason for choosing the
oxidized cofactor to initiate the reaction (red event,
Scheme 2).[23] Maximal efficiency is in theory
attained when racemization is the most rapid event
[equation (2)]. Increasing the enzyme concentration
usually correlated with decreased stereoselectivity
and it was speculated that the too slow racemization,
combined with imperfect enzyme stereoselectivity,
could not provide a steady pool of the faster reacting
enantiomer (e.g. krac < kRox < kSox).[11,21]
In order to sustain a constant ratio of products, the
ideal system should possess a reductive half-reaction
rate for the desired enantiomer comparable to the
overall rate of the hydration step and oxidative half-
Scheme 2. Equilibria and reactions involved in the
asymmetric biocatalytic disproportionation of rac-1a to
(S)-2a and (S)-3a (key controlling enzymatic event in red).
red
ox
reaction combined (kS
= kS + khyd < krac).
Importantly, the low concentration of NAD+ (0.007-
4
This article is protected by copyright. All rights reserved.

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
Table 2. Selected variables and relative levels in the
Table 3. Validation of the model by prediction of the four
responses and corresponding experimental results at pre-
defined conditions.a)
Central Composite Design.
Levels (coded)
Response
Conversion (%) 39
Predicted Experimentalb) 95% CIc)
Factora)
Low Central High Axial Axial
(-1) (0)
(+1) (-)b (-)b
16
0.40 0.63
25 48
40
38-41
Ratio 3a/2a
ee (S)-2a (%)
ee (S)-3a (%)
0.79
91
84
0.77
92
83
0.77-0.81
91-92
A [HLADH] 10
22
0.85 0.25
70 10
7
26
1.00
85
B [NAD+]
83-85
C [1a]
a)
b)
A = 13 M; B = 0.5 mM; C = 50 mM. Mean of
duplicate experiments. c) Interval of confidence. Data from
the prediction were rounded for clarity.
a) A in µM, B and C in mM. b) 'Star points' were set at  =
1.68179 to maintain rotability
consists in the efficient evaluation of interactions and
quadratic effects.[25] CCD consists of a factorial
design (2k, where k is the number of the parameters)
augmented with a set of 'star points' (2k, placed at the
distance of ± from the center point) to estimate the
curvature. Considering the number of variables (k =
3), duplicates and center points (n = 4), the overall
final number of experiments is 36 (2k + 2k + n). The
standard reaction conditions are described in the
experimental section and for the entire experimental
matrix and corresponding results, see the Supporting
Information.
Four different responses were evaluated:
conversion (%), product ratio (3a/2a), ee of (S)-2a
(%) and ee of (S)-3a (%). The collected data were
analyzed by ANOVA, which assesses the
significance of the effect of each parameter and their
interactions (for the calculations, see Supporting
Information). Multiple regression analysis provided
the final reduced model equations, one for each
specific response (3) - (6) [coefficients are given as
coded values and significance levels were assessed by
stars (* p ≤ 0.1; ** p ≤ 0.05; *** p ≤ 0.01)]:
predictive power of the statistical models applied to a
biological system is remarkable. The equations and
their relative perturbation plots, in which the
sensitivity of a response to each significant factor
(cross-interactions not taken into account) is
described graphically (Figure S2), were analyzed,
bearing in mind that a steep slope or curvature
correlates with a strong influence of the variable on
the outcome (response). Remarkably, at least one
significant quadratic term was included in all models
[italic terms in (3)-(6)]: for instance, increasing the
substrate concentration (C) suggested a substantial
positive effect on the overall stereoselectivity of the
reaction, as well as on the product ratio, but appears
detrimental to the conversion. Moreover, C acted as
the only significant variable able to control the
product ratio, whereas the cofactor amount (B) had a
strong influence on both conversion and
enantiopurity of (S)-2a, however with opposite trend.
Finally, increasing the enzyme concentration (A)
dramatically affected the stereoselectivity, while
having a positive influence on the conversion.
Successful optimization of the system requires
above all identification of possible cross-interactions
between variables; these and identified trends can be
visualized in the corresponding contour plots [NAD+
concentration set to 0.5 mM, Figure 2(a-d)]. Isolines
parallel to the x-axis in plot (b) illustrate the sole
(3) (conversion) = 43.80 + 4.08A*** + 1.57B*** –
4.53C*** + 2.21AC*** + 1.01BC* – 1.56 A2*** –
2.08C2***
(4) (3a/2a)3 = 0.4950 + 0.0197B* + 0.0413C*** –
dependence of the ratio on
C
(substrate
0.0408C2***
concentration). In the other three plots, the concurrent
effect of both A and C (enzyme and aldehyde
concentration, respectively) is depicted as diagonal
isopleths: their curvature indicates existing cross-
interactions. Importantly, the impracticality of a
simultaneous optimization of all four responses is
thereby highlighted, and an optimum compromise
must be sought instead. Contour plots obtained at
fixed enzyme concentration (C = 50 mM) revealed
weaker interactions between A and B (Figure S3) as
well as opposite behaviors on conversion and
stereoselectivity.
In order to determine suitable 'sweet spots' in
which pre-defined responses (desired reaction
outcome) would be fulfilled, the optimization module
of the software was used to screen for suitable
combinations of parameters. Following constraints
were applied: conversion >38%, ratio 3a/2a >0.77, ee
of (S)-2a >90% and ee of (S)-3a >83%. A section of
the experimental domain matching these criteria at
(5) (ee (S)-2a)2 = 7424.87 – 964.96A*** –
407.09B*** + 1480.24C*** + 340.29AC***
+ 92.90BC** + 138.56B2*** – 440.29C2***
(6) (ee (S)-3a)3 = 571200 – 16442.8A*** –
11175.59B** + 57493.75C*** + 13803.96AC** –
24716.31C2***
To validate the models and assess their predictive
power, a defined set of experimental conditions was
chosen (A = 13 M; B = 0.5 mM; C = 50 mM) and
the models were applied to anticipate the outcome of
the corresponding reaction. The experimental data
perfectly matched with the prediction (Table 3),
thereby validating the four equations. Given the
strong variations in product enantiopurity reported
under varying conditions (Figure 1 and Table 1), the
5
This article is protected by copyright. All rights reserved.

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
Figure 2. Contour plots (DesignExpert®) of all responses; warmer (colder) colors indicate higher (lower) values; A =
[HLADH] (M; x-axis); B = [NAD+] or C = [1a] (mM; y-axis). Conversion of 1a (a), ratio 3a/2a (b), ee of (S)-2a (c) and
ee of (S)-3a (d); conversion of 1e (e), ratio 3e/2e (f), ee of (S)-2e (g) and ee of (S)-3e (h). [NAD+] set to 0.5 mM in plots
(a-d); [1a] set to 30 mM in plots (e-h).
0.5 mM NAD+ could be identified, thereby
highlighting the rather narrow operational
experimental window (yellow area, Figure 3); we
could observe that the experimental conditions
previously selected for the model validation were
situated within the optimum range.
(S)-2e, already hinting at a difficult optimization of
the disproportionation of 1e; ii) the cofactor amount
(B) had a similar influence, except on the ratio, where
a marked curvature effect was observed; iii) the
substrate concentration (C) was identified as a key
parameter involved in interactions with an opposite
effect to that of A and B.
Faced Centered Design using 2-(p-tolyl)propanal
(1e).
While initial experiments indicated that it is
possible to reach high conversion levels at low
substrate concentration (Table S4, entries 7-8), the
conserved excellent enantioselectivity in the
Encouraged by the results and the information
obtained from the DoE analysis applied to the
conversion of 1a, we adopted the same approach for
the reaction with 2-(p-tolyl)propanal (1e). The
resulting model could again be experimentally
validated at 95% confidence (Table S6 and
Supporting Information). Interestingly, the resulting
model equations for both conversion and ee of (S)-2e
were composed only of linear terms and interactions,
whereas those for ratio 3e/2e and ee of (S)-3e
required also 'pure quadratic' terms. The effect of
each single parameter was evaluated by analyzing the
perturbation plots (Figure S4), which yielded the
following observations: i) Increasing the enzyme
concentration (A) had a positive effect on overall
oxidative
half-reaction
was
unfortunately
accompanied by a decrease in the enantiopurity of
(S)-2e and a product ratio in favor of the alcohol. The
described general trends were far more evident from
the contour plots, obtained this time at fixed substrate
concentration ([1e] = 30 mM, Figure 2(e-h)). An
exquisite stereoselectivity in the reductive half-
reaction was not only coupled with poor to moderate
conversion, but the unlikelihood of a concurrent
perfect enantioselectivity in both half-reactions was
also manifest, since the two optimum areas pointed in
opposite directions [Figure 2(g-h)]. The quadratic
effect of the cofactor amount on 3e/2e was also
evident, with slight deviations from the optimum
value of B leading to out-of-balance product ratio
[Figure 2(f)].
conversion
–
as generally expected
–
and
enantiopurity of (S)-3e, but an adverse effect on
product ratio and – unexpectedly – enantiopurity of
6
This article is protected by copyright. All rights reserved.

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
Table 4. Substrate scope of HLADH-catalyzed asymmetric
disproportionation.a)
Entry
R
[1]
(mM) (%)b
Conv. 3/2c) ee (S)-2
ee (S)-3
(%)
(%)
1
2
3
4
5
6
7
8
F
(1b)
50
30
10
50
30
10
50
30
10
50
30
10
27
32
34
6
1.01 >99
0.89 99
0.82 >99
0.96 >99
0.93 >99
0.97 >99
1.24 99
1.29 98
1.40 96
0.86 99
0.88 99
0.91 99
1.01 99
0.99 99
0.98 98
1.00 >99
1.09 99
1.09 91
94
94
93
77
84
90
87
89
84
87
89
90
93
93
93
84
90
90
CF3
(1c)
12
36
14
25
51
11
18
29
19
31
71
8
Br
(1d)
Figure 3. Overlay plot of the four responses
(DesignExpert®); A = [HLADH] (M; x-axis); C = [1a]
(mM; y-axis); the area where all required criteria are
fulfilled at 0.5 mM NAD+ is highlighted in yellow
(conversion >38%; ratio 3a/2a >0.77, ee (S)-2a >90%; ee
(S)-3a >83%).
9
10
11
12
13
14
15
16
17
18
Me
(1e)
MeO 50
(1f)
30
10
50
30
10
i-Bu
(1g)
Taken together, the results of the DoE analysis
highlight the practical difficulty of attaining
optimization of all outputs of the disproportionation
reaction and its strong dependence on the substrate
type.
16
31
a) Conditions: phosphate buffer (50 mM, pH 7.5), HLADH
(0.5 mg/mL, ~ 13 µM), 0.5 mM NAD+, 5 vol% MTBE as
co-solvent, 30 °C, 120 rpm, 24 h (see experimental
section). b) Conversions (to [alcohol+acid]) and ee values
were determined by chiral GC analysis after product
extraction (see Supporting Information). c) Carboxylic acid
to alcohol ratio.
Substrate scope of the PIDAT
Finally, the substrate scope of the HLADH-driven
disproportionation reaction was investigated with
emphasis on 2-arylpropanals bearing different groups
in the p-position, as this substitution pattern is a
common feature in the profen NSAID sub-class.
Varying substrate concentrations (10-50 mM) were
employed to evaluate the enzyme tolerance and
performance at higher aldehyde loading (Table 4).
Based on the findings derived from the DoE analysis,
NAD+ was added at 0.5 mM concentration.
selectivity.
An
atom-efficient[26]
convergent
enzymatic route to (S)-ibuprofen based on HLADH-
mediated PIDAT of 1g is currently being developed
and will be reported elsewhere. Altogether, no clear
pattern in line with the electronic effects of the
substitution could be identified, despite anticipation
that electron-withdrawing groups would be beneficial
by enhancing racemization and, possibly, hydration[8]
rates. Indeed, TTN of HLADH was the highest with
non-substituted 1a (~ 3 x 103), followed by p-fluoro-
derivative 1b, while bulky compound 1c led to the
lowest value. p-Bromo-compound 1c and p-methoxy-
compound 1f gave intermediate TTNs (Table S7).
Overall, the p-substitution was well tolerated and
HLADH catalyzed the asymmetric formal Cannizzaro
reaction with exquisite stereocontrol in both half-
reactions on 1b-g [(S)-2 in 99- >99% ee and (S)-3 in
87-94% ee]. Compound 1b was particularly well
accepted and yielded highest product amount (total
13.5 mM from 50 mM substrate, Table 4, entry 1)
with perfect product ratio and highest enantiopurity
for both products. Some general trends could be
observed: product ratios were generally well-
balanced (except for 1d) and conversions increased
Conclusion
with
decreasing
substrate
concentration.
The biocatalytic asymmetric formal Cannizzaro
reaction of -substituted aldehydes catalyzed by
HLADH was investigated in detail, with particular
attention to enantioselective formation of both
alcohol and carboxylic acid products. In order to
develop this intricate reaction to synthetically
relevant conditions, aiming at high product
Enantioselectivity toward (S)-2b-g was consistently
excellent, indicating that the conditions applied were
in this regard ideal. For most substrates, stereocontrol
in the oxidation reaction tends to decrease upon
increasing substrate concentration.
Interestingly, 2-(4-isobutylphenyl)propanal (1g,
precursor to ibuprofen) was also accepted (Table 4,
entries 16-18): despite rather low conversion, ratio
3g/2g was perfectly balanced, with ideal enantio-
concentration and perfect stereoselectivity,
a
statistical investigation by means of design of
7
This article is protected by copyright. All rights reserved.

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
(OD600 = 0.6). IPTG solution (165 μL of a 1 M stock
solution, 0.5 mM final concentration) was added. The
cultures were shaken overnight at 25 °C and 120 rpm.
Cells from 2 L of culture were harvested by centrifugation
(4000 rpm, 20 min, 4 °C), washed with phosphate buffer A
(50 mM, pH 7.5), centrifuged again (4000 rpm, 20 min,
4 °C) and resuspended in buffer A containing 20 mM
imidazole. After sonication (Branson Sonifier 250; 3 min,
30% amplitude, 1 sec pulse, 4 sec pause), the disrupted
cells were centrifuged (17000 rpm, 20 min, 4 °C). The
supernatant was transferred through a fluted filter into a 50
mL Sarstedt tube, loaded on a HisTrapTM FF column (5
mL), and washed with 20 column volumes of buffer A
containing 20 mM imidazole. The column was washed
with buffer A containing 150 mM imidazole (10 column
volumes). The protein was eluted from the column with
buffer A containing 300 mM imidazole (seven column
volumes). Possible residual proteins were washed from the
column with buffer A containing 500 mM imidazole (five
column volumes). After SDS-PAGE (Figure S1), fractions
containing highly pure HLADH were combined, desalted
overnight at 4 °C in a dialysis tube submerged into buffer
A (4 L), freeze-dried and stored at -21 °C.
experiments (DoE) was performed. This provided a
global understanding of the system and highlighted
interactions between the parameters involved in the
reaction. While a common optimum cannot be
attained, an excellent compromise between
conversion, product ratio and stereocontrol of both
concurrent reactions was achieved at 75 mM
concentration of aldehyde, yielding products in up to
99% ee. The transformation proceeds in an atom-
efficient manner, does not need a sacrificial co-
substrate and relies on a single enzyme and catalytic
amounts of oxidized cofactor (max. 0.05 eq.). In 100
mg-scale transformation, this translated into an
effective requirement of 2 mol% NAD+ related to the
amount of isolated products. A range of 2-
arylpropanals bearing substituents in the p-position
were converted to corresponding enantioenriched (S)-
'profenols' and 'profens', thereby highlighting the
synthetic
potential
of the
redox-neutral[27]
disproportionation. While some substrates are still
posing a challenge in terms of reactivity, the use of
other ADHs could improve the reaction output.
Promising remains the exquisite stereocontrol of the
asymmetric dismutation reaction, which is achieved
through concurrent oxidative and reductive reactions
regulated by the oxidized cofactor acting as connector
and applied in catalytic quantity. In several cases,
HLADH appears well-suited to install the
stereochemistry with high precision in the final two
products at almost 'no cost'.[28] Further studies on
practical applications of the parallel interconnected
dynamic asymmetric transformation (PIDAT) for
stereoselective synthesis are currently underway in
our laboratories.
Standard procedure for disproportionation
reactions (PIDAT)
Enzymatic reactions were run as follows: an aliquot of
purified HLADH was added to a phosphate buffer solution
(final volume 1 mL, 50 mM, pH 7.5) containing rac-1a-g
and the nicotinamide cofactor(s). The reaction mixture was
incubated at 30 °C and 120 rpm. After 24 h, products were
extracted with EtOAc (300 μL), the aqueous phase was
acidified with HCl (100 μL, 3 N) and extracted again with
EtOAc (300 μL). The combined organic fractions were
dried over Na2SO4 and analyzed by chiral GC to determine
both conversion (using calibration lines) and enantiomeric
excesses of both products. For disproportionation reactions
in presence of co-solvent (1b-g), MTBE was present in the
buffer in 5 vol%; the workup was performed as above.
Experimental Section
General information
Preparative scale disproportionation of 1a
Chemical reagents and solvents were used as received
from commercial sources. Rac-1b-g and reference
compounds (rac- and (S)-2b-g and rac- and (S)-3b-g) were
prepared according to modified procedures reported in
literature, and fully characterized (see Supporting
Information). Absolute configuration of enantiopure
reference compounds was determined by comparison of
experimental optical rotation values with literature data.
Absolute stereochemistry of products from the
biotransformations was assigned by chiral GC analysis by
comparison with elution profiles of available racemic and
corresponding enantiopure reference compounds (see
Supporting Information).
The reaction was performed on 100 mg 1a according to the
procedure developed during the optimization phase (Figure
1, entry 10). Briefly, 100 mg of rac-1a (0.75 mmol) added
from
concentration),
a
stock solution in MTBE (4 vol% final
mg of freshly purified HLADH
5
supplemented with ZnCl2 (100 µM final concentration)
and 3 mg of NAD+ (5 µmol) were incubated at 30 °C and
120 rpm in 10 mL phosphate buffer (50 mM, pH 7.5).
After 24 h, a second aliquot of both enzyme preparation (5
mg) and fresh cofactor (3 mg) were added, and the
incubation was continued for 24 h. The mixture was then
extracted with EtOAc and the solvent evaporated under
vacuum. The crude was purified by flash chromatography
(cyclohexane/EtOAc 8:2), obtaining 31 mg (0.23 mmol) of
1
Cloning and overexpression of HLADH
(S)-2a, in 97% ee (determined by chiral GC). H NMR
For expression of HLADH (E-isoenzyme of ADH from
Equus caballus forming dimer EE,[29] cloned in pET28a
witn N-terminal His-tag), E. coli BL21 (DE3) cells
transformed with pEG 54 plasmid were grown overnight at
30 °C and 120 rpm in a 50 mL Sarstedt tube containing
LB/Kan medium (10 mL, 50 g/mL kanamycin). LB/Kan
medium (330 mL, 50 μg/mL kanamycin) in 1 L baffled
shaking flasks was inoculated with 2 mL of the ONC.
These cultures were shaken at 25 °C and 120 rpm for 4.5 h
(300 MHz, CDCl3): δ = 7.38-7.19 (m, 5H), 3.70 (d, J = 6.8
Hz, 2H), 2.95 (h, J = 6.9 Hz, 1H), 1.46 (bs, 1H), 1.28 (d, J
= 7.0 Hz, 3H); NMR data in agreement with the
literature.[30] The aqueous phase was acidified with 3 N
HCl and extracted with EtOAc; evaporating the solvent
yielded 35 mg (0.23 mmol) of pure (S)-3a in 86% ee
1
(determined by chiral GC). H NMR (300 MHz, CDCl3): δ
= 7.61-7.25 (m, 5H), 3.74 (q, J = 7.2 Hz, 1H), 1.52 (d, J =
8
This article is protected by copyright. All rights reserved.

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
7.2 Hz, 3H); NMR data in agreement with the literature.[31]
Overall isolated yield: 62%; ratio 3a/2a: 1.03.
[15] C.-F. Mandenius, A. Brundin, Biotechnol. Progr.
2008, 24, 1191-1203.
[16] S. M. Kritzinger, S. G. Kilian, M. A. Potgieter, J. C.
du Preez, Enzyme Microb. Technol. 2003, 32, 728-737.
Acknowledgements
[17] A. Bornadel, R. Hatti-Kaul, F. Hollmann, S. Kara,
The Austrian Science Fund (FWF) is gratefully acknowledged for
financial support (P30519). E. T. thanks the Austrian Ministry of
Science, Research and Economy (BMWFW) for an 'Ernst Mach –
worldwide' fellowship. The authors thank Snježana Maliurić and
Adrian Tripp for their assistance in the preparation of reference
materials. HLADH-carrying plasmid was a kind gift from Prof.
Dr. M. Pohl (Forschungszentrum Jülich, Germany). The authors
wish to thank the reviewer who anonymously suggested the
correct terminology 'formal Cannizzaro reaction'.
Tetrahedron 2016, 72, 7222-7228.
[18] a) D. Giacomini, P. Galletti, A. Quintavalla, G.
Gucciardo, F. Paradisi, Chem. Commun. 2007, 4038-
4040; b) D. Quaglia, M. Pori, P. Galletti, E. Emer, F.
Paradisi, D. Giacomini, Process Biochem. 2013, 48,
810-818.
[19] R. Kourist, P. Dominguez de Maria, K. Miyamoto,
Green Chem. 2011, 13, 2607-2618
References
[20] a) F. Ledl, E. Schleicher, Angew. Chem. Int. Ed.
1990, 29, 565-594; b) H. K. Weber, J. Zuegg, K. Faber,
J. Pleiss, J. Mol. Catal. B: Enzym. 1997, 3, 131-138.
[1] S. Cannizzaro, Justus Liebigs Ann. Chem. 1853, 88,
129-130.
[21] a) G. A. Applegate, D. B. Berkowitz, Adv. Synth.
Catal. 2015, 357, 1619; b) J. A. Friest, Y. Maezato, S.
Broussy, P. Blum, D. B. Berkowitz, J. Am. Chem. Soc.
2010, 132, 5930-5931.
[2] J. Clayden, N. Greeves, S. Warren, Organic Chemistry,
Oxford, 2nd edn., 2012.
[3] a) M. M. Mojtahedi, E. Akbarzadeh, R. Sharifi, M. S.
Abaee, Org. Lett. 2007, 9, 2791-2793; b) K.
Yoshizawa, S. Toyota, F. Toda, Tetrahedron Lett.
2001, 42, 7983-7985.
[22] S. Kara, D. Spickermann, J. H. Schrittwieser, C.
Leggewie, W. J. H. van Berkel, I. W. C. E. Arends, F.
Hollmann, Green Chem. 2013, 15, 330-335
[4] R. S. Varma, K. P. Naicker, P. J. Liesen, Tetrahedron
Lett. 1998, 39, 8437-8440.
[23] For Parallel Interconnected Kinetic Asymmetric
Transformation (PIKAT) using two substrates and two
enzymes, see: A. Rioz-Martinez, F. R. Bisogno, C.
Rodriguez, G. de Gonzalo, I. Lavandera, D. E. Torres
Pazmino, M. W. Fraaije, V. Gotor, Org. Biomol. Chem.
2010, 8, 1431-1437.
[5] D. Basavaiah, D. S. Sharada, A. Veerendhar,
Tetrahedron Lett. 2006, 47, 5771-5774.
[6] a) M. Curini, F. Epifano, S. Genovese, M. C.
Marcotullio, O. Rosati, Org. Lett. 2005, 7, 1331-1333;
b) X. Yang, J. Guo, G. Zou, Lett. Org. Chem. 2005, 2,
145-147
[24] M. Kitamura, M. Tokunaga, R. Noyori, J. Am. Chem.
Soc. 1993, 115, 144-152.
[7] C. Wuensch, H. Lechner, S. M. Glueck, K. Zangger,
[25] NIST/SEMATECH, e-Handbook of Statistical
M. Hall, K. Faber, ChemCatChem 2013, 5, 1744-1748
Methods
2017,
Chapter
5.3.3.6;
http://www.itl.nist.gov/div898/handbook/pri/section893
/pri336.htm (accessed November 820, 2017)
[8] K. T. Eisses, Bioorg. Chem. 1989, 17, 268-274.
[9] a ) A. J. Irwin, J. B. Jones, J. Am. Chem. Soc. 1977, 99,
1625-1630; b) I. J. Jakovac, H. B. Goodbrand, K. P.
Lok, J. B. Jones, J. Am. Chem. Soc. 1982, 104, 4659-
4665.
[26] B. M. Trost, Science 1991, 254, 1471-1477.
[27] N. Z. Burns, P. S. Baran, R. W. Hoffmann, Angew.
Chem. Int. Ed. 2009, 48, 2854-2867.
[10] a) K. Ishihara, T. Yano, M. Fushimi, J. Fluorine
Chem. 2008, 129, 994-997; b) P. Wang, W.-J. Tao, X.-
L. Sun, S. Liao, Y. Tang, J. Am. Chem. Soc. 2013, 135,
16849-16852; c) W. Wu, X. Liu, Y. Zhang, J. Ji, T.
Huang, L. Lin, X. Feng, Chem. Commun. 2015, 51,
11646-11649.
[28] T. Newhouse, P. S. Baran, R. W. Hoffmann, Chem.
Soc. Rev. 2009, 38, 3010-3021.
[29] D. H. Park, B. V. Plapp, J. Biol. Chem. 1991, 266,
13296-13302.
[30] X. Cui, Y. Li, C. Topf, K. Junge, M. Beller, Angew.
Chem. Int. Ed. 2015, 54, 10596-10599.
[11] P. Könst, H. Merkens, S. Kara, S. Kochius, A. Vogel,
R. Zuhse, D. Holtmann, I. W. C. E. Arends, F.
Hollmann, Angew. Chem. Int. Ed. 2012, 51, 9914-9917.
[31] M. Gaydou, T. Moragas, F. Juliá-Hernández, R.
Martin, J. Am. Chem. Soc. 2017, 139, 12161-12164.
[12] K. Faber, Biotransformations in Organic Chemistry,
Springer, 6th edn., 2011.
[13] P. Braiuca, C. Ebert, A. Basso, P. Linda, L. Gardossi,
Trends Biotechnol. 2006, 24, 419-425.
[14] G. E. P. Box, J. S. Hunter, W. G. Hunter, Statistics
for experimenters: design, innovation, and discovery,
Wiley-Interscience, 2005.
9
This article is protected by copyright. All rights reserved.

10.1002/adsc.201800541
Advanced Synthesis & Catalysis
UPDATE
Biocatalytic Parallel Interconnected Dynamic
Asymmetric Disproportionation of -Substituted
Aldehydes: Atom-Efficient Access to Enantiopure
(S)-Profens and Profenols
Adv. Synth. Catal. Year, Volume, Page – Page
Erika Tassano, Kurt Faber and Mélanie Hall*
10
This article is protected by copyright. All rights reserved.