A zinc-dependent alcohol dehydrogenase (ADH) from Thauera aromatica, reducing cyclic α- And β-diketones

Article abstract of DOI:10.1002/adsc.201500171

Zinc-dependent alcohol dehydrogenases (ADHs) are valuable biocatalysts for the synthesis of chiral hydroxy compounds such as α-hydroxy ketones and diols, both valuable precursors for the synthesis of various pharmaceuticals. However, while highly active on aliphatic or phenyl-substituted diketones, most well characterized ADHs show no significant activity on cyclic α- and β-diketones. Therefore, this study aimed at the detection of a novel ADH capable to reduce these special targets. It involved a rational screening of biochemical pathways for enzymes with structurally related natural substrates. The so detected 6-hydroxycyclohex-1-ene-1-carbonyl-CoA dehydrogenase (ThaADH) from Thauera aromatica was cloned, expressed in Escherichia coli and purified by affinity chromatography. The characterization revealed a substrate specificity with highest activities on cyclic α- and β-diketones including 1,2-cyclohexanedione and 1,3-cyclopentanedione. Structural reasons for this extraordinary substrate spectrum were investigated with a homology model created via Swiss Model server. Although the quality of the model may be improved, it suggests that a bulky aromatic residue, that plays a crucial role in the definition of the substrate binding pockets of most ADHs, is replaced by a glycine residue in ThaADH. We propose that this structural difference leads to the formation of one large binding pocket instead of two smaller ones and consequently to a preference for cyclic diketones over linear bulky substrates. Thus, we have achieved both provision of a novel biocatalyst with high potential in chiral synthesis, and a possible explanation for the measured differences to known ADHs. The described structural motif might be used for identification of further enzymes with a related substrate scope.

Full text of DOI:10.1002/adsc.201500171

FULL PAPERS
DOI: 10.1002/adsc.201500171
A Zinc-Dependent Alcohol Dehydrogenase (ADH) from
Thauera aromatica, Reducing Cyclic a- and b-Diketones
Christoph Loderer,a Frances Morgenstern,a and Marion Ansorge-Schumachera,*
a
Chair of Molecular Biotechnology, Institute of Microbiology, Technische Universität Dresden, Zellescher Weg 20b, 01062
Dresden, Germany
Fax: (+49)-351-463-39520; phone: (+49)-351-463-39519; e-mail: marion.ansorge@tu-dresden.de
Received: February 23, 2015; Revised: April 15, 2015; Published online: May 12, 2015
Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/adsc.201500171.
Abstract: Zinc-dependent alcohol dehydrogenases spectrum were investigated with a homology model
(ADHs) are valuable biocatalysts for the synthesis of created via Swiss Model server. Although the quality
chiral hydroxy compounds such as a-hydroxy ke- of the model may be improved, it suggests that
tones and diols, both valuable precursors for the syn- a bulky aromatic residue, that plays a crucial role in
thesis of various pharmaceuticals. However, while the definition of the substrate binding pockets of
highly active on aliphatic or phenyl-substituted dike- most ADHs, is replaced by a glycine residue in
tones, most well characterized ADHs show no signif- ThaADH. We propose that this structural difference
icant activity on cyclic a- and b-diketones. Therefore, leads to the formation of one large binding pocket
this study aimed at the detection of a novel ADH ca- instead of two smaller ones and consequently to
pable to reduce these special targets. It involved a ra- a preference for cyclic diketones over linear bulky
tional screening of biochemical pathways for en- substrates. Thus, we have achieved both provision of
zymes with structurally related natural substrates. a novel biocatalyst with high potential in chiral syn-
The so detected 6-hydroxycyclohex-1-ene-1-carbon- thesis, and a possible explanation for the measured
yl-CoA dehydrogenase (ThaADH) from Thauera ar- differences to known ADHs. The described structur-
omatica was cloned, expressed in Escherichia coli al motif might be used for identification of further
and purified by affinity chromatography. The charac- enzymes with a related substrate scope.
terization revealed a substrate specificity with high-
est activities on cyclic a- and b-diketones including
1,2-cyclohexanedione and 1,3-cyclopentanedione. Keywords: biocatalysis; carbonyl reduction; enzyme
Structural reasons for this extraordinary substrate catalysis; oxidoreductases
Introduction
as aliphatic diols, hydroxy acids, hydroxylated iso-
prene derivatives and benzyl alcohol derivatives.[4,5,6,7]
Alcohol dehydrogenases (ADHs) are a diverse group A very interesting class of products are a-hydroxy ke-
of enzymes that catalyze the reversible reduction of tones that can be obtained either by selective oxida-
carbonyl functions to the corresponding alcohols. tion of a chiral diol or reduction of a prochiral dike-
Their often high chemo-, regio- and stereoselectively tone.[8] A number of ADHs were already demonstrat-
make them potential tools for biocatalytic application. ed to efficiently catalyze the reduction of linear ali-
For asymmetric synthesis, zinc-dependent medium phatic diketones such as 2,3-pentanedione, to the cor-
chain ADHs have been described as particularly responding
hydroxy
ketone.[9,10,11,12]
Substrate
useful. Enzymes in this category are constituted of specificity and stereoselectivity in these reactions are
about 340 amino acids and share the Rossman fold described by a widely accepted molecular model in
motif containing two zinc atoms per subunit.[1] The which the two substituents of the carbonyl function
first zinc atom is directly involved in catalysis while are fitted into a small and a large alkyl site within the
the second is important for the overall structure of enzyme, respectively, defining the orientation of the
the enzyme.[2,3]
substrate in the active site.[13,14,15] In contrast, little in-
Zinc-dependent ADHs have been applied in the formation exists on ADH-catalyzed conversion of
synthesis of various chiral hydroxy compounds such cyclic diketones. Most common ADHs were either
1872
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 1872 – 1880

FULL PAPERS
A Zinc-Dependent Alcohol Dehydrogenase (ADH) from Thauera aromatica
Results and Discussion
Screening of Metabolic Pathways
In order to identify ADHs with a probable capability
for reduction of cyclic diketone compounds, metabolic
pathways were scanned on the KEGG metabolic
pathway database. Metabolic steps with carbonyl re-
duction or alcohol oxidation reactions on natural sub-
strates with a structural similarity to 1,2-cyclohexane-
dione and 1,3-cyclohexanedione were screened. Six
promising reactions involved in different metabolic
pathways were detected. Where available, enzymes
proposed or described to catalyze these reactions
were assigned to their protein families by applying
the Pfam database. This assignment yielded one
promising alcohol oxidation being catalyzed by
mainly-zinc-dependent ADHs. It is part of the bacte-
rial degradation of benzoate and involved oxidation
of 6-hydroxycyclohex-1-ene-1-carboxyl-CoA to 6-oxo-
Figure 1. A) Physiological reaction of 6-hydroxycyclohex-1-
ene-1-carbonyl-CoA dehydrogenase (6HCoADH) in the
benzoate degradation pathway.[22] B) Target substrates for
enzymatic reduction.
not tested or determined to be not active towards cyclohex-1-ene-1-carboxyl-CoA (Figure 1A) by a 6-
these substrates.[10,11,12,16] A few reports describe reduc- hydroxycyclohex-1-ene-1-carbonyl-CoA dehydrogen-
tion of 1,2-cyclohexanedione, 1,3-cyclohexanedione ase (6HCoADH, EC 1.1.1.368).[22] In the product 6-
and 1,3-cyclopentanedione (Figure 1B), which were oxocyclohex-1-ene-1-carbonyl-CoA, the carbon atom
used as model compounds in this study. A non-identi- in the a-position to the reactive hydroxy function is
fied NADPH-dependent enzyme from yeast was sp2-hybridized as is the case in 1,2-cyclohexanedione
shown to reduce 1,2-cyclohexanedione with decent ac- (Figure 1B). Instead of a second carbonyl group an
tivities.[17] As well, ADHT from Thermoanaerobacter extremely bulky CoA-thio ester group is present
ethanolicus 39E converted 1,2-cyclohexanedione, but which suggests that, from a sterical point of view, 1,2-
only with 5% of the reaction rate, revealed on its best cyclohexanedione should also fit in the active site.
substrate.[18] A not further specified carbonyl reduc- 6HCoADH was therefore a promising candidate for
tase KRED101 from Merck was shown to be able to the reduction of the model cyclic diketones.
catalyze the reduction of 4,4-dimethoxytetrahydro-
The benzoate degradation pathway is mainly found
2H-pyran-3-one, which is structurally related to the in proteobacteria such as Thauera, Geobacter and
described model substrates.[19,20] The NADPH-depen- Rhodopseudomonas species with 6HCoADH being
dent 3-hydroxycyclohexanone dehydrogenase from described as a characteristic enzyme.[23] Its physiologi-
a denitrifying Pseudomonas species exhibits good ac- cal role has been demonstrated for the enzyme from
tivities for the oxidation of cyclic diols but only very Thauera aromatica which was cloned and recombi-
low relative activities for the reduction of 1,2-cyclo- nantly expressed in E. coli for this purpose. The
hexanedione and 1,3-cyclohexanedione (3% each).[21]
enzyme was described as a homo dimer with a molecu-
In conclusion, no well characterized ADH is avail- lar weight of 38 kDa per subunit[24] and shown to
able to date that is able to reduce cyclic diketones accept NAD+, but not NADP+ as cofactor. It formed
such as 1,2-cyclohexanedione and 1,3-cyclohexane- the basis of all the following experiments and for sim-
dione with acceptable reaction rates. Consequently, plicity was renamed ThaADH.
this study aimed for the detection, provision and char-
acterization of a novel ADH to fill this gap. Homolo-
gy modeling was used to identify structural differen- Provision of ThaADH
ces between the novel enzyme and well-established
reference ADHs such as carbonyl reductase from The ThaADH gene obtained from Thauera aromatica
Candida parapsilosis (CPCR2) and thus provide an genomic DNA was fused to a DNA-sequence coding
explanation for the differences in substrate specifici- for a C-terminally attached StrepTagII to facilitate
ties.
protein purification and identification, and was ex-
pressed in E. coli BL21(DE3). Expression resulted in
a protein with a molecular weight of about 40 kDa
(Figure 2A), which was predominantly in the soluble
protein fraction of the cell lysate (Figure 2B). The
identity of the ThaADH-StrepTagII fusion protein
Adv. Synth. Catal. 2015, 357, 1872 – 1880
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1873

FULL PAPERS
Christoph Loderer et al.
Figure 2. A) Coomassie stained SDS-PAGE of the ThaADH after recombinant expression in E. coli and after purification,
respectively. Cl: cleared lysate, Fl: column flowthrough, W: washing fraction, E1: elution fraction 1, E2: elution fraction 2.
B) Coomassie stained SDS-PAGE and immunodetection of StrepTagII in the insoluble (ins) and soluble (sol) fraction of E.
coli lysate after ThaADH-expression. Strep-tagged CPCR2 was used as a positive control.
was confirmed via Anti-StrepTagII Western blot (Fig- highest catalytic activities on compounds similar to its
ure 2B). After purification, no major impurities (Fig- natural substrate. Hence the prediction that an
ure 2A) were observed. The total protein yield was enzyme with structurally related natural substrates
22 mg from 400 mL culture volume.
should be able to reduce the model compounds was
confirmed.
Since 1,2-cyclohexanedione yielded the highest re-
action velocities it was applied for the measurements
of pH and temperature dependences of ThaADH.
Reactivity of ThaADH
The substrate spectrum of ThaADH with regard to The pH profile shows a sharp pH optimum for the re-
carbonyl reduction was measured for various alde- duction reaction at 6.0 with a residual activity of 40%
hydes, ketones and especially diketones (Figure 3). In- at pH 7.0 (Figure 4A). For acidic pH the activity de-
terestingly, ThaADH converts neither acetophenone crease is even stronger. Concerning temperature the
nor benzaldehyde which are both preferred substrates highest activity was measured at 508C (Figure 4B).
of many known zinc-dependent ADHs, such as Higher temperatures lead to a very fast precipitation
CPCR2 from Candida parapsilosis, SsADH from Sul- of the enzyme disturbing the photometrical measure-
folobus solfataricus, TADH from Thermus sp., ADH- ment.
A from Rhodococcus ruber and HlADH from horse
Dependency of enzyme activity on substrate con-
liver.[10,12,16,25,26] Moreover, the enzyme actually prefers centration was determined for different substrates.
substrates with the reactive carbonyl function being For 2,3-pentanedione, a nearly linear increase in
located inside a cyclohexyl or cyclopentyl structure. enzyme activity was measured up to 200 mmol·LÀ1 in-
With 0.14 U·mgÀ1 the highest specific activity was dicating a KM value >100 mmol·LÀ1. In the case of cy-
measured for 1,2-cyclohexanedione, which is in ac- clohexanone and 1,2-cyclohexanedione KM could be
cordance with the highest structural similarity of this more exactly determined yielding 12.1 and
compound to the natural substrate, 6-hydroxycyclo- 9.3 mmol·LÀ1, respectively (Figure 4C). Thus, despite
hex-1-ene-1-carbonyl-CoA. 1,3-Cyclohexanedione is the lack of exact information for 2,3-pentanedione, it
converted four-fold slower and 1,4-cyclohexanedione is obvious that ThaADH shows a significantly higher
is not accepted at all. The specific activity for unsub- affinity to both cyclic carbonyl substrates. With
stituted cyclohexanone is 0.10 U·mgÀ1 and therefore 5.1 U·mgÀ1 the maximal specific activity for reduction
lower than for the ortho-substituted 1,2-cyclohexane- of 1,2-cyclohexanedione is only 50% lower than for
dione. Hence, ortho-substitution of the reactive car- the natural substrate (11.8 U·mgÀ1).[24] In contrast, KM
bonyl group is obviously beneficial, meta-substitution for oxidation of 1,2-cyclohexanedione is 150-fold
is accepted and para-substitution is detrimental for higher than for reaction on the natural substrate
the reductive activity of ThaADH. Summarizing these (60 mmol·LÀ1). However, although the affinity to the
observations we could show that ThaADH exhibits its cyclic diketone is lower compared to the natural sub-
1874
asc.wiley-vch.de
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 1872 – 1880

FULL PAPERS
A Zinc-Dependent Alcohol Dehydrogenase (ADH) from Thauera aromatica
Figure 3. Substrate specificity of ThaADH. Assay conditions: buffer: triethanolamine (pH 6.4, 100 mmol·LÀ1), T=508C, sub-
strate (3 mmol·LÀ1), NADH (250 mmol·LÀ1), ThaADH (100 mg·mLÀ1). All measurements were conducted in triplicate.
Figure 4. A) pH-dependent activity profile of ThaADH. Assay conditions: buffers: pH 4.0, 5.0: acetate (100 mmol·LÀ1), pH 6.0:
phosphate (100 mmol·LÀ1), pH 7.0, 8.0, 9.0 triethanolamine (TEA) (100 mmol·LÀ1), T=508C, 1,2-cyclohexanedione
(3 mmol·LÀ1), NADH (250 mmol·LÀ1), ThaADH (100 mg·mLÀ1). B) Temperature-dependent activity profile of ThaADH. Assay
conditions: buffer: TEA (pH 6.4, 100 mmol·LÀ1), T=508C, 1,2-cyclohexanedione (3 mmol·LÀ1), NADH (250 mmol·LÀ1),
ThaADH (100 mg·mLÀ1). C) Kinetic parameters of ThaADH and CPCR2 for the reduction of carbonyl compounds.
strate, it is higher than for the best linear diketone.
The reaction products of 1,2-cyclohexanedione re-
This again confirms the preference of ThaADH for duction were identified via GC-MS. Thereby 2-hy-
cyclic diketones.
droxy-cyclohexanone was identified as sole reaction
Adv. Synth. Catal. 2015, 357, 1872 – 1880
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1875

FULL PAPERS
Christoph Loderer et al.
product (Supporting Information, Figures S2–S6), enzyme should possess a significantly different struc-
whereas the corresponding diol was not detected. ture in comparison to the ADHs described before.
This indicates a specific mono-reduction of the dike- Since a crystal structure of ThaADH is not available,
tone substrate by ThaADH.
homology models of the enzyme were created to elu-
cidate this matter. The best models obtained via the
Swiss model homology modeling server were based
on cod liver class I alcohol dehydrogenase (1cdo.1.B)
and an ADH from Bacillus stearothermophilus
Structural Determinants of ThaADH Activity
The amino acid sequence of ThaADH shows a signifi- (1rjw.1.A), respectively, as templates.[27,28] Unfortu-
cant similarity to the well described ADHs such as nately, the overall quality of the models is quite poor
CPCR2 (47%) and ADH ꢁAꢂ (44%). Despite these due to the only low sequence identities between
sequence similarities the substrate spectrum of ThaADH and the template enzymes (33% and 34%,
ThaADH clearly shows that the active site of the respectively). This is indicated by the low GMQE
value (Table 1), which describes the expected accura-
cy of the model with a maximum value of 1. On the
models based on the templates cod liver class I alcohol de- other hand, the two models derived from completely
Table 1. Parameters and scores of the ThaADH homology
hydrogenase (1cdo.1.B) and ADH from Bacillus stearother-
mophilus (1rjw.1.A).
different templates are in good agreement (Fig-
ure 5A). To check whether the homology model is ap-
propriate to describe the active site, the local quality
(IDDT) of the model was reviewed for each residue.
The regions with lowest local quality were identified
as loops, located on the protein surface, and as parts
of the subunit interface. In contrast, the active site
residues, which are located in a cleft, exhibit a signifi-
cantly better quality than the overall model (Table 1).
The derived models were compared to the available
Model
Template
PDB file Sequence similarity GMQE[a] IDDT
Average[b]
Model quality
1
2
1cdo.1.B 0.33
1rjw.1.A 0.34
0.6
0.6
0.74
0.69
[a]
GMQE: expected accuracy of the homology model;[33,34]
IDDT: expected local similarity to the target;[33,34]
[b]
IDDT average: average IDDT values from six active site crystal structure of CPCR2 from Candida parapsilo-
residues (MM).
sis, ADH-A from Rhodococcus ruber, SsADH from
Figure 5. A) Superposition of the two homology models of ThaADH based on the known crystal structures of cod liver class
I alcohol dehydrogenase (1cdo.1.B, dark blue) and an ADH from Bacillus stearothermophilus (1rjw.1.A, bright blue), respec-
tively, as template. B) Superposition of the homology model of ThaADH based on cod liver class I alcohol dehydrogenase
and known crystal structure of CPCR2. Highlighted are the catalytic zinc atom (magenta), the cofactor NADH (grey) as
well as G316 in ThaADH (blue) and W286 in CPCR2 (red). The surface of ThaADH residues in the relevant part of the
active site is shown in blue. C) Superposition of the crystal structures of the zinc-dependent ADHs CPCR2 (red), SsADH
from Sulfolobus solfataricus (bright orange), ADH-A from Rhodococcus ruber (dark orange) and HlADH from horse liver
(yellow). Highlighted are the catalytic zinc atom (magenta), the cofactor NADH (grey) and the aromatic amino acid corre-
sponding to W286 in CPCR2 (red).
1876
asc.wiley-vch.de
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 1872 – 1880

FULL PAPERS
A Zinc-Dependent Alcohol Dehydrogenase (ADH) from Thauera aromatica
Figure 6. Schematic model of the proposed function of residue G316 in ThaADH and the corresponding residue W286 in
CPCR2. A) Catalytically active binding of acetophenone in CPCR2. B) Possible clashes in a potential non-productive bind-
ing of 1,2-cyclohexanedione in CPCR2. C) Catalytically active binding of 1,2-cyclohexanedione in ThaADH.
Sulfolobus solfataricus and HlADH from horse ble due to a conformational adaption of the sub-
liver.[29,30,31,32] In Figure 5B, the superposition of the strate,[35] illustrated by the high, but detectable KM
monomeric structures of CPCR2 and ThaADH is value (Figure 4C). For inflexible cyclic substrates such
shown. The superposition yielded a RMSD of 1.407 Š as 1,2-cyclohexanedione no catalytically active bind-
over 301 residues, confirming the expected structural ing is possible due to potential clashes with W286 or
similarity. ThaADH residues C55, H78 and D169 other active site residues such as S46. In ThaADH
align with the CPCR2 residues C44, H65 and D154 the presence of a glycine residue in the corresponding
coordinating the catalytic zinc residue. ThaADH resi- position leads to the formation of one large binding
due T57 aligns with S46 in CPCR2, which acts as pocket. Thus, even the sterically inflexible 1,2-cyclo-
proton donor during the reduction reaction (Fig- hexanedione can be bound in a catalytically active po-
ure 5B).
sition (Figure 6C) with a KM value of 9.3 mmol·LÀ1. In
Comparison of the neighboring residues constitut- turn, the absence of two distinct binding pockets for
ing the active site of the different enzymes reveals the substituents of linear substrates leads to a reduced
one major structural difference. W286 which is re- affinity for this type of compounds, as measured for
stricting the active site of CPCR2 in a distance of 2,3-pentanedione. Unfortunately, replacement of
5.3 Š to the catalytic zinc atom is exchanged for a gly- W286 in CPCR2 by another amino acid does not
cine (G316) in ThaADH. A bulky aromatic residue at yield active variants.
this position is structurally conserved amongst all ref-
This was observed during site saturation of this resi-
erence ADHs and was described to be a constituent due in a previous study (detailed data will be pub-
of the small substrate binding pocket on the boundary lished elsewhere). The physical replacement of trypto-
to the large one (Figure 5C).[29] Thus, it could act as phan in the active site most likely resulted in a severe
a wedge separating the two substrate binding pockets, destabilization, preventing an observable phenotype.
but could also limit the available space for cyclic sub- Thus, a direct proof for the hypothesized effect of
strates. Assuming the homology model is accurate the a W286 analogue missing in ThaADH has still to be
removal of this wedge results in the formation of one delivered.
large hydrophobic binding pocket improving access of
cyclic substrates to the active site.
Based on this observation and the measured kinetic Conclusions
parameters we propose the following model for the
impact of the structural difference between ThaADH It was demonstrated in this study that data mining in
and CPCR2 in positions G316 and W286: When the metabolic pathways based on structural similarities of
high affinity substrate acetophenone binds to CPCR2, substrates and defined reaction types is an appropri-
the small methyl and the large phenyl substituent of ate tool for identification of biocatalysts with novel
the substrate are located in the small and the large specificities. Accordingly, a novel zinc-dependent
alkyl sites, respectively (Figure 6A).
ADH, with a preference for mono-reduction of steri-
The alkyl sites are separated by W286. In a compa- cally demanding cyclic diketones was identified. Sup-
rable binding position, the ring structure of cyclic sub- ported by homology modeling and kinetic parameters,
strates may clash with the residue separating the alkyl an exchange of a bulky aromatic residue to glycine in
sites (Figure 6B). In case of comparably flexible ring the active site of this enzyme can be proposed to play
structures such as cyclohexanone, binding is still possi- a crucial role for its extraordinary substrate specifici-
Adv. Synth. Catal. 2015, 357, 1872 – 1880
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1877

FULL PAPERS
Christoph Loderer et al.
the temperature was reduced to 158C. The cell were har-
vested after 22 h by centrifugation (48C, 8000 rpm, 10 min)
and stored at À208C.
ty. However, due to the potential inaccuracy of the
homology model, a crystal structure will be necessary
to proof this postulation.
Due to its exceptional substrate specificity
ThaADH has a high potential for a biocatalytic appli-
cation in the reduction of diketones. In addition, this
enzyme may be used as a starting point in the search
for further biocatalysts accepting sterically demanding
diketone substrates. A further structural investigation
of ThaADH will also be of great value for the eluci-
dation of structural determinants in the substrate spe-
cificity of ADHs in general.
Purification of ThaADH-Streptag Fusion Protein
Cell lysis was conducted by a combination of lysozyme di-
gestion and ultrasonification. The cells were first resuspend-
ed in cell lysis buffer containing 100 mmol·LÀ1 triethanola-
mine (TEA), 150 mmol·LÀ1 NaCl, 1 mmol·LÀ1 PMSF,
50 mg·mLÀ1 lysozyme and 5 mg·mLÀ1 DNAseI (AppliChem
GmbH, Darmstadt, Germany). After incubation for 90 min
at 68C and 2 min of sonification (Sartorius Labsonic M, Set-
tings: Cycle=0.6, Amplitude=100%), cell debris were re-
moved by centrifugation (48C, 16000 rpm, 30 min).
For protein purification Strep-Tactinꢃ Sepharoseꢃ 5 mL
gravity flow columns (IBA GmbH, Goettingen, Germany)
were utilized. The columns were equilibrated with 25 mL
washing buffer (100 mmol·LÀ1 TEA, 150 mmol·LÀ1 NaCl,
1 mmol·LÀ1 EDTA at pH 7.0). The cleared lysate containing
ThaADH-streptag fusion protein was applied on the
column. After washing with 10 mL of washing buffer, bound
protein was released with 15 mL elution buffer
(100 mmol·LÀ1 TEA, 150 mmol·LÀ1 NaCl, 2.5 mmol·LÀ1 Des-
thiobiotin at pH 7.0) in 2 mL fractions. The fractions with
the highest activity were pooled. After addition of glycerol
to a final concentration of 40% the protein was stored at
À208C.
Experimental Section
General Information
If not stated differently, all chemicals and oligonucleotides
were purchased from Sigma–Aldrich (Sigma–Aldrich, St.
Louis, MO, USA), media components were purchased from
Carl Roth (Carl Roth GmbH & Ko. KG, Karlsruhe, Germa-
ny) and enzymes for molecular cloning were purchased
from Thermo Scientific (Thermo Fisher Scientific inc, Wal-
tham, MA, USA).
Cloning of the thaADH Gene
CPCR2 was expressed and purified according to the pro-
tocol described by Jakoblinnert.[7]
The thaADH gene (Genbank GI: 19571180) was obtained
via PCR from genomic DNA from Thauera aromatica pro-
vided by Leibniz-Institut DSMZ (Deutsche Sammlung von
Mikroorganismen und Zellkulturen GmbH, Braunschweig,
Germany). The PCR was conducted with Phusionꢃ High-Fi-
delity DNA Polymerase (Thermo Fisher Scientific, Wal-
tham, MA, USA) according to the manufacturerꢂs instruc-
tions. The primers Thauera NdeI for: cattcatATGGCGGC
GAAGAGTTC and Thauera_HindIII_rev: gtataagcttGG-
GCGTGAGGATGG were applied introducing a NdeI and
a HindIII restriction site in the PCR product. With these re-
striction sites the thaADH gene was ligated into the multi-
ple cloning site of a modified pET22b(+) expression vector
under the control of a T7 promoter. Thereby a linker coding
for ten alanine residues followed by a StrepTagII was fused
to the C-terminus of the gene. E.coli DH5a was transformed
with this vector and screened for positive clones via colony
PCR. For selected positive clones a plasmid preparation was
performed followed by sequencing by GATC biotech
(GATC Biotech AG, Konstanz, Germany).
Polyacrylamide Gel Electrophoresis and Western Blot
SDS-PAGE was performed according to the procedure of
Lämmli[37] applying the Invitrogen XCell SurelockTM
system (Life Technologies, Carlsbad, CA, USA). For the ex-
pected 40 kDa target protein a tris-glycine gel containing
12% acrylamide was utilized. Samples were incubated for
10 min at 958C in loading buffer and 15 mL were loaded on
the gel. In addition 5 mL of PageRuler Prestained Protein
Ladder were loaded on the gel. The gel was run for 80 min
at 180 V. In case of colorimetric visualization, staining was
performed with Coomassie brilliant blue R-250.
For immunodetection of the ThaADH-streptag fusion
protein, the SuperSignalꢃ West Femto kit (IBA GmbH,
Goettingen, Germany) was applied, which is based on
a streptactin-horse radish peroxidase conjugate. Protein
transfer from the SDS-gel to a methanol activated PVDF
membrane (Biotrace, Auckland, New Zealand) was per-
formed in the XCell SurelockTM system (Life Technologies,
Carlsbad, CA, USA) for 2 h at 30 V. The PVDF membrane
was then washed three times with TBS-T buffer
[20 mmol·LÀ1 Tris, 500 mmol·LÀ1 NaCl, 0.05% (v/v) Tween
20, pH 7.5] followed by the colorimetric assay, applying Su-
perSignalꢃ Femto Maximum Sensitivity Substrate (Thermo
Fisher Scientific, Waltham, MA, USA).
Expression of ThaADH-Streptag Fusion Protein
In order to utilize the T7 expression system, the sequenced
vector was transformed into E.coli BL21(DE3).[36] Induction
time, inductor concentration, expression temperature and
expression time were optimized. The highest expression was
achieved with the following protocol: 400 mL LB medium
containing 100 mg·mLÀ1 ampicillin were inoculated from an
overnight preculture to a final OD600 of 0.05 and incubated
at 378C and 150 rpm in an Infors Minitron shaker (Infors
GmbH, Einsbach, Germany) to an OD600 of 0.8. Then, ex-
pression was induced by addition of 1 mmol·LÀ1 IPTG and
Enzyme Activity Assays
Assays for ADH activity were performed photometrically
by measuring the consumption of the cofactor NADH. The
standard assay was performed in 100 mmol·LÀ1 TEA buffer
at a pH value of 6.4 with 3 mmol·LÀ1 substrate (1,2-cyclo-
1878
asc.wiley-vch.de
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 1872 – 1880

FULL PAPERS
A Zinc-Dependent Alcohol Dehydrogenase (ADH) from Thauera aromatica
hexanedione) and 250 mmol·LÀ1 NADH in 1 mL scale at
508C. The reaction was started by addition of ThaADH
stock solution to a final concentration of 100 mg·mLÀ1. Initial
reaction rates were measured at a wavelength of 340 nm in
10 s intervals for 2 min. For determination of the pH opti-
mum, acetate (pH 4.0, 5.0), phosphate (KPi) (pH 6.0) and
TEA buffer (pH 7.0, 8.0, 9.0) were utilized. For determina-
tion of the temperature optimum pH values were adjusted
to 6.4 for each tested temperature.
For determination of the kinetic parameters of ThaADH,
reaction conditions were modified. The enzyme concentra-
tion was reduced to 17 mg·mLÀ1 and the reaction time was
increased to 10 min. Reactions were performed in
100 mmol·LÀ1 KPi buffer (pH 6.0). For determination of the
kinetic parameters of CPCR2 all reactions were performed
under the same conditions in 100 mmol·LÀ1 TEA buffer
(pH 7.33) and 258C. From activity assays performed at dif-
fering substrate concentrations, kinetic parameters were de-
rived by non-linear regression. Activity data are shown in
the Supporting Information (Tables S2–S8)
ful discussions regarding molecular modeling. Prof. Gerold
Barth and Sebastian Stark from TU Dresden are acknowl-
edged for technical support in StrepTagII Immunodetection.
References
[1] B. Persson, J. Hedlund, H. Joernvall, Cell. Mol. Life
Sci. 2008, 65, 3879–3894.
[2] H. Joernvall, B. Persson, J. Jeffrey, Eur. J. Biochem.
1987, 167, 195–201.
[3] E. Magonet, P. Hayen, D. Delforge, E. Delaive, J. Re-
macle, Biochem. J. 1992, 287, 361–365.
[4] J. Haberland, W. Hummel, T. Daussmann, A. Liese,
Org. Process Res. Dev. 2002, 6, 458–462.
[5] M. Amidjojo, E. Franco-Lara, A. Nowak, H. Link, D.
Weuster-Botz, Appl Microbiol Biotechnol. 2005, 69, 9–
15.
[6] W. Kroutil, H. Mang, K. Edegger, K. Faber, Curr.
Opin. Chem. Biol. 2004, 8, 120–126.
[7] A. Jakoblinnert, R. Mladenov, A. Paul, F. Sibilla, U.
Schwaneberg, M. B. Ansorge-Schumacher, P. Domi-
nguez de Maria, Chem. Commun. 2011, 47, 12230–
12232.
[8] P. Hoyos, J. V. Sinisterra, F. Molinari, A. R. Alcantara,
P. Dominguez de Maria, Acc. Chem. Res. 2010, 43, 288–
299.
Product Identification
Identification of the reaction product of 1,2-cyclohexane-
dione reduction was performed on an Agilent 6890N Net-
work Gas Chromatograph system with an Agilent HP
5973N Mass Selective Detector equipped with an Optima
35 MS column. A temperature program from 50–3208C with
a gradient of 108C minÀ1 was applied. The substrate was de-
tected at 9.85 min (Supporting Information, Figures S2, S3,
S5). The product 2-hydroxycyclohexanone was detected at
9.97 min (Sipporeting Informatipon, Figures S2 and S3).
[9] J. Peters, T. Zelinski, T. Minuth, M. R. Kula, Tetrahe-
dron: Asymmetry 1993, 4, 1683–1692.
[10] G. Gonzalo, I. Lavandera, K. Faber, W. Kroutil, Org.
Lett. 2007, 9, 2163–2166.
[11] P. Schweiger, H. Gross, U. Deppenmeier, Appl Micro-
biol Biotechnol. 2010, 87, 1415–1426.
[12] S. Leuchs, L. Greiner, Chem. Biochem. Eng. Q. 2011,
25, 267–281.
[13] V. Prelog, Pure Appl. Chem. 1964, 9, 119–130.
[14] J. B. Jones, I. Jakovac, J. Chem. 1982, 60, 19–28.
[15] J. Peters, T. Zelinski, T. Minuth, M. R. Kula, Biocataly-
sis 1993, 8, 31–46.
Homology Modeling
Homology modeling of ThaADH structure was performed
via the Swiss Model protein structure homology-modeling
server (ProMod Version 3.70.).[38,39,40,41] The template search
yielded several hits with sequence identities ranging be-
tween 25 and 30%. 25 models were generated applying the
25 best hits as template. The quality of the models was
judged by the provided GMQE value. This value reflects
the expected accuracy of the model taking into account the
template structure as well as the model geometry and solva-
tation (QMean4).[33,34] To judge the local quality of the
model, six residues (H78, P102, A103, D169, G316, Y322) in
the active site were chosen and their local IDDT values
were utilized. The IDDT value is a measure for the local ex-
pected similarity to the target.[33,34] Structural alignments
and structure comparisons were performed with Yasara (ver-
sion 12.10.3). The structural alignment and the superposition
were conducted with build-in MUSTANG module.[42]
[16] J. Peters, T. Zelinski, T. Minuth, M. R. Kula, Enzyme
Microb. Technol. 1993, 15, 950–958.
[17] J. Heidlas, R. Tressl, Eur. J. Biochem. 1990, 188, 165–
174.
[18] T. Daussmann, H. G. Hennemann, European Patent
EP 1763577 B, 2005.
[19] J. C. Moore, D. J. Pollard, B. Kosjek, P. N. Devine, Acc.
Chem. Res. 2007, 40, 1412–1419.
[20] B. Kosjek, J. Nti-Gyabaah, K. Telari, L. Dunne, J. C.
Moore, Org. Process Res. Dev. 2008, 12, 584–588.
[21] W. Dangel, A. Tschech, G. Fuchs, Arch. Microbiol.
1989, 152, 273–279.
[22] M. Carmona, M. T. Zamarro, B. Blazquez, G. Durante-
Rodriguez, J. F. Juarez, J. A. Valderrama, M. J. L. Bar-
ragan, J. L. Garcia, E. Diaz, Microbiol. Mol. Biol. Rev.
2009, 73, 71–130.
Acknowledgements
Deutsche Forschungsgemeinschaft (DFG) is greatly acknowl-
edged for financial support within the research project AN
387/5-1 as well as the collaborative research network BioNo-
Co (GRK1166) for scientific support and software provision.
Marco Boccola and Gaurao Dhoke from RWTH Aachen
University are acknowledged for their kind support and fruit-
[23] D. A. Pelletier, C. S. Harwood, J. Bacteriol. 2000, 182,
2753–2760.
[24] D. Laempe, M. Jahn, G. Fuchs, Eur. J. Biochem. 1999,
263, 420–429.
[25] V. Hoellrigl, F. Hollmann, A. C. Kleeb, K. Buehler, A.
Schmid, Appl. Microbiol. Biotechnol. 2008, 81, 263–273.
Adv. Synth. Catal. 2015, 357, 1872 – 1880
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1879

FULL PAPERS
Christoph Loderer et al.
[26] A. Trincone, L. Lama, V. Lanzotti, B. Nicolaus, M.
De Rosa, M. Rossi, A. Gambacorta, Biotechnol.
Bioeng. 1990, 35, 559–564.
[27] A. S. Ramaswamy, M. El Ahmad, O. Danielsson, H.
Joernvall, H. Eklund, Protein Sci. 1996, 5, 663–671.
[28] C. Ceccarelli, Z. X. Liang, M. Strickler, G. Prehna,
B. M. Goldstein, J. P. Klinman, B. J. Bahnson, Biochem-
istry 2004, 43, 5266–5277.
[35] A. Jakoblinnert, J. Wachtmeister, L. Schukur, A. V. Shi-
vange, M. Bocola, M. B. Ansorge-Schumacher, U.
Schwaneberg, Protein Eng. Des. Sel. 2013, 26, 291–298.
[36] S. Tabor, in: Current Protocols in Molecular Biology,
(Eds.: F. M. Ausubel, R. Brent, R. E. Kingston, D. D.
Moore, J. G. Seidman, J. A. Smith, K. Struhl), John
Wiley & Sons, Inc., Hoboken, New Jersey, 1990,
pp 16.2.1–16.2.11.
[29] H. Man, C. Loderer, M. B. Ansorge-Schumacher, G.
Grogan, ChemCatChem 2014, 6, 1103–1111.
[30] M. Karabec, A. Łyskowski, K. C. Tauber, G. Steinkell-
ner, W. Kroutil, G. Grogan, K. Gruber, Chem.
Commun. 2010, 46, 6314–6316.
[31] L. Esposito, F. Sica, C. A. Raia, A. Giordano, M. Rossi,
L. Mazzarella, A. Zagari, J. Mol. Biol. 2002, 318, 463–
477.
[32] S. Ramaswamy, M. Scholze, B. V. Plapp, Biochemistry
1997, 36, 3522–3527.
[33] P. Benkert, M. Kuenzli 1, T. Schwede, Nucleic Acids
Res. 2009, 37, 510–514.
[34] P. Benkert, M. Biasini, T. Schwede, Bioinformatics
2011, 27, 343–350.
[37] U. K. Laemmli, Nature 1970, 227, 680–685.
[38] M. Biasini, S. Bienert, A. Waterhouse, K. Arnold, G.
Studer, T. Schmidt, F. Kiefer, T. G. Cassarino, M. Ber-
toni, L. Bordoli, T. Schwede, Nucleic Acids Res. 2014,
42, 252–258.
[39] K. Arnold, L. Bordoli, J. Koppl, T. Schwede, Bioinfor-
matics 2006, 22, 195–201.
[40] N. Guex, M. C. Peitsch, T. Schwede, Electrophoresis
2009, 30, 162–173.
[41] F. Kiefer, K. Arnold, M. Kuenzli, L. Bordoli, T.
Schwede, Nucleic Acids Res. 2009, 37, 387–392.
[42] A. S. Konagurthu, J. C. Whisstock, P. J. Stuckey, A. M.
Lesk, Proteins Struct. Funct. Bioinf. 2006, 64, 559–574.
1880
asc.wiley-vch.de
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 1872 – 1880