Resolution of α/β-amino acids by enantioselective penicillin G acylase from Achromobacter sp.

Article abstract of DOI:10.1016/j.molcatb.2015.09.008

Penicillin G acylases (PGAs) are enantioselective enzymes catalyzing a hydrolysis of stable amide bond in a broad spectrum of substrates. Among them, derivatives of α- and β-amino acids represent a class of compounds with high application potential. PGA^Ec from Escherichia coli and PGA^A from Achromobacter sp. CCM 4824 were used to catalyze enantioselective hydrolyses of seven selected N-phenylacetylated α/β-amino acid racemates. The PGA^A showed higher stereoselectivity for enantiomers of N-PhAc-β-homoleucine, N-PhAc-α-tert-leucine and N-PhAc-β-leucine. To study the mechanism of enantiodiscrimination on molecular level, we have constructed a homology model of PGA^A that was used in molecular docking experiments with the same substrates. In-silico experiments successfully reproduced the data from experimental enzymatic resolutions confirming validity of employed modeling protocol. We employed this protocol to evaluate enantiopreference of PGA^A towards seven new substrates with application potential. For five of them, high enantioselectivity of PGA^A was predicted.

Full text of DOI:10.1016/j.molcatb.2015.09.008

Journal of Molecular Catalysis B: Enzymatic 122 (2015) 240–247
Contents lists available at ScienceDirect
Journal of Molecular Catalysis B: Enzymatic
journal homepage: www.elsevier.com/locate/molcatb
Resolution of ␣/␤-amino acids by enantioselective penicillin G acylase
from Achromobacter sp.
c
a
a
a
Michal Grulicha,b,∗, Jan Brezovsky´ , Václav Steˇpánek , Andrea Palyzová , Eva Kyslíková ,
ˇ
Pavel Kyslíka
a Laboratory of Enzyme Technology, Institute of Microbiology,v.v.i., Academy of Sciences of the Czech Republic, Vídenˇská 1083, 142 20 Prague 4, Czech
Republic
b Department of Biochemistry, Faculty of Sciences, Charles University, Hlavova 8, 128 40 Prague 2, Czech Republic
c Loschmidt Laboratories, Department of Experimental Biology and Research Centre for Toxic Compounds in the Environment RECETOX, Masaryk
University, Kamenice 5/A13, 625 00 Brno, Czech Republic
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 26 June 2015
Received in revised form 19 August 2015
Accepted 18 September 2015
Available online 25 September 2015
Penicillin G acylases (PGAs) are enantioselective enzymes catalyzing a hydrolysis of stable amide bond in a
broad spectrum of substrates. Among them, derivatives of ␣- and ␤-amino acids represent a class of com-
pounds with high application potential. PGAEc from Escherichia coli and PGAA from Achromobacter sp. CCM
4824 were used to catalyze enantioselective hydrolyses of seven selected N-phenylacetylated ␣/␤-amino
acid racemates. The PGAA showed higher stereoselectivity for enantiomers of N-PhAc-␤-homoleucine,
N-PhAc-␣-tert-leucine and N-PhAc-␤-leucine. To study the mechanism of enantiodiscrimination on
molecular level, we have constructed a homology model of PGAA that was used in molecular docking
experiments with the same substrates. In-silico experiments successfully reproduced the data from exper-
imental enzymatic resolutions confirming validity of employed modeling protocol. We employed this
protocol to evaluate enantiopreference of PGAA towards seven new substrates with application potential.
For five of them, high enantioselectivity of PGAA was predicted.
Keywords:
Penicillin G acylase
Enantioselectivity
Homologous model
Docking experiments
␣/␤-amino acid
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
Recent resolutions of three-dimensional PGA crystal structures
provided detailed insight into the active site of this enzyme and
Penicillin G acylases (EC 3.5.1.11; PGA, penicillin amidohydro-
lase) are robust industrial catalysts routinely used for decades for
biotransformation of penicillins and cephalosporins [1–4]. Their
biocatalytic potential expanded tremendeously once PGAs were
found to be enantioselective and promiscuous [5]. Syntheses of
semisynthetic ␤-lactam antibiotics, peptide syntheses and res-
olutions of racemic mixtures via enantioselective acylation or
hydrolysis of broad range of substrates (e.g., amino acids, ketones,
amino acid esters or amides, amino nitriles or amino alcohols)
catalyzed by PGAs represent biocatalyses with high potential espe-
cially for pharmaceutical industry. The PGAs are heterodimers
composed of ␣- and ␤-subunit and belong to the structural super-
family of N-terminal nucleophilic hydrolases (Ntn hydrolases).
allowed better understanding of a catalytic mechanism. An active
site of PGAEc consists of two regions [6]: 1. aminic subsite preferring
hydrophilic groups including catalytic aa residues Ser1ß, and other
essential amino acid residues Gln23ß, Ala69ß and Asn241ß and 2.
acyl binding subsite which accepts hydrophobic groups [7] con-
sisting of Met142␣, Arg145␣, Phe146␣, Phe24ß, Thr32ß, Pro49ß,
Val56ß, Trp154ß and Ile177ß. Catalytic mechanism of hydrolysis
of penicillin G, synthesis of ß-lactam antibiotics or enantioselective
resolution involves the nucleophilic attack of the O␥ hydroxyl group
of Ser1ß on the carbonyl carbon of the amide bond [8,9]. As the con-
sequence of the attack, a covalent intermediate of an acyl-enzyme
complex via a tetrahedral transition state is formed. Asn241ß and
Ala69ß form the “oxyanion hole” that balances the negative charge
and thus lowers the energy of the reactive tetrahedral intermedi-
ate. Gln23ß is also shown to interact with the nucleophilic part of
a substrate and contribute to the stabilization of the tetrahedral
intermediate.
Abbreviation: N-PhAc-amino acid, N-phenylacetyl amino acid.
∗
The PGAA from the mutant strain Achromobacter sp. CCM 4824
Corresponding author at: Laboratory of Enzyme Technology, Institute of Micro-
biology,v.v.i., Academy of Sciences of the Czech Republic, Vídenˇská 1083, 142 20
Prague 4, Czech Republic. Fax: +420 296442347.
ˇ
[10] has been firstly characterized by Skrob et al. [11]. Molecular
mass of ␣-subunit and ␤-subunit is 27.0 and 62.5 kDa, respectively.
E-mail address: grulich@biomed.cas.cz (M. Grulich).
http://dx.doi.org/10.1016/j.molcatb.2015.09.008
1381-1177/© 2015 Elsevier B.V. All rights reserved.

M. Grulich et al. / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 240–247
241
The results showed that the enzyme is more efficient biocatalyst in
comparison with PGAEc as regards syntheses of semi-synthetic ␤-
lactam antibiotics. Its potential for kinetically controlled syntheses
of semisynthetic ␤-lactam antibiotics has been shown recently [12]
and has already been used at industrial scale by Fermenta Biotech
Ltd.
Here we present results of the research into enantioselectivity of
PGAA. First, the enantioselective resolutions of racemic mixtures of
␣/␤-amino acids having considerable potential for pharmaceutical
applications were determined experimentally. Then we describe a
construction of a homology model of PGAA and consequent molec-
ular docking experiments employed to understand molecular basis
of PGAA enantioselectivity. Experimentally validated model of the
enzyme was used to predict PGAA enantioselectivity towards new,
hitherto non-investigated substrates.
of 0.4 mL/min)—2 mM CuSO4 containing 5% or 2% isopropanol; for
Daicel Crownpak CR(+) column (flow rate of 1 mL/min)—aqueous
solution of HClO4 (pH 1) containing methanol (10% v/v). Reten-
tion times of enantiomers of reaction products are summarized in
Supplementary data (Table S1).
E values describing enantioselectivities of PGAEc or PGAA were
determined using Eq. (1):
ꢀ
ꢁ
ln A/A0
ꢀ
ꢁ
E =
(1)
ln B/B0
where A0 and A represent the concentrations of the faster react-
ing enantiomer at the reaction times 0 and t, resp. B0 and B denote
the concentrations of the slower reacting enantiomer at the reac-
tion times 0 and t, resp. [16]. The enantiomeric ratio E was also
determined by non-linear regression using Eq. (2) which is derived
from Sihı´’s equation describing a relationship between E, degree of
conversion (c) and the enantiomeric excess eep [16].
2. Experimental
ꢂ
ꢃ
1/ E−1
(
)
ꢀ
ꢁ
ꢀ
ꢁ−E
2.1. Microorganisms and culture conditions
c = 1 −
1 + eep
×
1 − eep
(2)
Recombinant strain E. coli BL21(pKX1P1) and E. coli RE3(pKA18)
was used to prepare biomass for purification of PGAA and PGAEc
,
2.5. Molecular modeling
respectively. Fed batch cultures of the strains in a stirred bioreactor
were described earlier [13,14].
A homology model of PGAA was prepared using the sequence
of amino acids from GenBank (AAY25991). The most suitable tem-
plate for modeling was identified by PSI-BLAST search using the
ExPaSy server against sequences in the Protein Data Bank. The PGAA
query sequence was then aligned with the best template—PGAEc
enzyme (PDB-ID 1GM7, sequence identity 50%) using ClustalOmega
[17]. Identity of amino acid residues forming the active site between
the template and the query was 88%. Only two substitutions
occurred within the seventeen amino acid residues forming the
active site (Val56ß → Leu56ß and Ser149␣ → Ala149␣), and no
insertions or deletions were observed for these positions. Align-
ments were prepared with a constraint between the Ser1␤ residue
of the query sequence and the corresponding serine residue of the
template. The homology modeling based on alignment of PGAA and
PGAEc was performed using the Swiss-model webserver [18]. The ␣
and ␤ chains of penicillin acylases were modelled separately by cal-
culating a number of minimized intermediate models which were
ranked by the structure quality Z-score. One model of the ␣ chain
and one of the ␤ chain had to be chosen from this ensemble of struc-
tures, not only on the basis on the Z-score, but also by taking into
account the reciprocal positions of the two chains. The final model
was evaluated using a program Verify 3D [19] with the template
validation data used as the baseline to assess the respective models.
To facilitate reproducibility of the work, the model was deposited
to Protein Model DataBase [20] under the following access code
PM0080082. The analysis of Ramachandran plot generated by RAM-
PAGE server [21] suggested a high quality of the homology model
with more than 95% of residues located in the favored region, for
detailed report from the analysis see Supplementary data (Fig. S8).
This finding was further confirmed by low Z-score of −0.29 reported
by QMEAN server [22,23] indicating that the quality of the model is
comparable to the high-resolution crystal structures of proteins of
similar size. Next, the PGAA model was protonated via H++ server at
pH 7.5 using default settings. Protonation of catalytically important
amino acid residues was modified manually so that they conform
to specific reaction mechanism, i.e., enantioselective hydrolysis of
N-PhAc-␣/␤-amino acids [24].
2.2. Enzyme purification and hydrolytic activity assay
Ec
PGAA was purified as described by Skrob et al. [11] and PGA
ˇ
according to Kutzbach and Rauenbusch [15]. Purified PGAEc and
PGAA has the specific activity of 60 and 50 U/mg protein, respec-
tively. The activity of 1 unit (U) was defined as the amount of an
enzyme producing 1 ␮mol of phenyl acetic acid per minute from
PenG (2% w/v) in 0.05 M sodium phosphate buffer (pH 8.0) at 37 ◦C.
2.3. Chemical synthesis of N-phenylacetyl-amino acid racemic
mixtures
Phenylacetyl chloride (0.044 mol) was added dropwise to
100 mL of NaOH solution (10%, w/v) of racemic mixture of an
amino acid (0.04 mol) kept on ice bath. The reaction mixture was
acidified to pH 2 with 6 M HCl, and N-phenylacetyl amino acid (N-
PhAc amino acid) was extracted three times with ethyl acetate and
recrystallized from the ethyl acetate solution. Purity of the product
was determined by HPLC using C-8 reverse phase column. Eluent
consisted of H2O (containing 0.1% TFA): acetonitrile in ratio 7:3. For
more details on products yields, chemical shifts in the 1H NMR and
13C NMR spectra and data from MS analyses see Supplementary
data (Figs. S1–S7).
2.4. Enantioselective hydrolysis of N-PhAc-amino acid racemic
mixtures
The reactions were carried out in a pH-stat at 30 ◦C under
continuous stirring. A water solution (25 mL) containing 0.025 M
N-PhAc-amino acid racemic mixture was incubated at pH 7.5 for
30 min. 50 U of PGAEc or PGAA were added to the reaction mixture
and the pH was maintained at 7.5 by titration (2 M NH4OH).
Concentrations of reactants were analyzed by HPLC using
Dionex P580 Pump, C-8 column and a Dionex PDA-100 detector set
at 215 nm. The mobile phase consisted of a mixture of acetonitrile:
water (containing 0.1% TFA) = 3: 7.
The enantiomeric excess of the products (eep) was determined
by HPLC using Dionex P580 Pump, Sumichiral OA 5000 column and
UV detector set at 215 nm or Daicel Crownpak CR(+) column using
the Dionex PDA-100 detector set at 200 nm. Composition of eluent
solution was as follows: for Sumichiral OA 5000 column (flow rate
Molecular docking calculations were performed through
AutoDock Vina plug-in for PyMol [25]. Structures of substrates
were prepared using Avogadro molecular editor [26] and ener-
getically minimized in four steps of the steepest descent using
MMFF94 force field [27]. Substrate and protein structures were
converted to AutoDock Vina compliant format by AutoDock Vina

242
M. Grulich et al. / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 240–247
plug-in for PyMol. For docking, the grid box (22.5 × 22.5 × 22.5 Å)
covering the PGAA active site was defined and centered on the cen-
ter of mass of Ser1␤, Ala69␤ and Asn241␤ residues. The docking
calculations were performed using Autodock Vina software [28]
with an exhaustiveness of 200, maximum of 9 generated binding
modes and maximum energy difference between the best and the
worst binding modes of 3 kcal/mol. Visualization of docked bind-
ing modes and analysis of its geometries were performed in PyMol
[29] software. Mechanism-based geometric criteria for prediction
of substrate reactivity [30] were derived accordingly to the catalytic
mechanism of PGAs. To evaluate reactivity of docked substrate con-
formations, the binding modes were assessed by using following
criteria: (i) distance between nucleophilic oxygen atom Oꢀ of Ser1␤
and attacked carbon from carbonyl group of the substrate reflect-
ing the probability of nucleophilic attack, (ii) lengths of hydrogen
bonds between nitrogen atoms from amino groups of Ala69␤ and
Asn241␤ and the oxygen atom from the carbonyl group of sub-
strate that stabilize the negative charge of the reactive tetrahedral
intermediate, and (iii) length of hydrogen bond between oxygen
atom from carbonyl group of Gln23␤ and leaving nitrogen group of
substrate that also contributes to the stabilization of the tetrahe-
dral intermediate (Fig. 1). To derive particular cutoff distances for
these criteria, high-quality structures of PGA complexed with sub-
strates and their analogs were analyzed (Supplementary data, Table
S2). Since such evaluation provides only qualitative two-state pre-
diction of substrates reactivity—reactive and non-reactive, we can
accurately predict only potential for highly enantioselective dis-
crimination. In such a case, one enantiomer is predicted as poorly
active, while the other one as well active. It is important to note
that minor differences in enantioselectivity cannot be predicted
using employed approach. Additionally, the interactions present in
predicted binding modes were analyzed by PoseView [31].
Fig. 1. Mechanism-based geometric criteria developed for evaluation of reactivity
of docked substrates in active site of PGAA. The essential amino acid residues of
PGAA’s active site are shown as green sticks. Binding mode of (S)-N-PhAc-tert-leucine
is shown as cyan sticks. The analyzed distances D1 (OGln23␤ → H), D2 (OSer1␤ → C),
D3 (NAsn241␤ → O) and D4 (NAla69␤ → O) are displayed as yellow dashed lines. (For
interpretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)
3. Results and discussion
3.1. Experimental determination of enantioselectivity of PGAA
on PGAA dealt with syntheses of several important semi-synthetic
␤-lactam antibiotics. In order to asses a broader potential of PGAA,
we have focused on seven ␣/␤-amino acids (Table 1) that have
application potential as building blocks of active pharmaceutical
ingredients. These substrates were used to compare enantioselec-
Research into enantioselectivity of PGAs started in 60s of the
last century and dealt mainly with resolutions of racemic mixtures
of ␣-amino acids and their derivatives [32,33]. So far, main concern
Table 1
Substrates subjected to biocatalysis by PGAA and PGAEc
.
Substrates
Structure
Product application
References
[35]
N-PhAc-␣-phenylalanine
␣-(S)-phenylalanine—direct precursor of neuromodulator phenylethylamine
N-PhAc-␣-homophenylalanine
(S)-homophenylalanine—building block of drugs for threatment of
[36]
hypertension and cardiovascular diseases
N-PhAc-␤-leucine
N-PhAc-␣-leucine
Building block for terpenoids or proteins
[37]
[38]
(S)-enantiomer—building block of proteins
N-PhAc-␣-isoleucine
N-PhAc-␤-homoleucine
N-PhAc-␣-tert-leucine
(S)-isoleucine—food complement
Enhancer of activity of myeloperoxidase, building block of biologically active
tripeptides
[39,40]
[24,41]
(S)-tert-leucine - building block for anti-AIDS and anti-cancer pharmaceuticals

M. Grulich et al. / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 240–247
243
Table 2
studied experimentally (Table 1). The binding energies predicted
by molecular docking strongly correlated (R = −0.76) with the
molecular weight of investigated substrates. However, the largest
difference in the predicted binding energies between enantiomers
of a given substrate was 0.7 kcal/mol only (Table S3). Since such
difference is within the error margin of the scoring function,
the observed difference between binding affinities of individual
enantiomers should be considered as negligible without any sig-
nificant influence on the enantioselectivity of PGAA. The predicted
structures of complexes were further analyzed in respect to the
interactions of substrates with amino acid residues involved in
the catalysis (i.e., Ala69␤, Asn241␤, Gln23␤ and Ser1␤). For all
N-PhAc-␣-amino acids and N-PhAc-␤-homoleucine, the distances
from functional groups of important amino acid were markedly
shorter for (S)-enantiomer which predicted this enantiomer as the
preferred form (Table 3). On the contrary, (R)-N-PhAc-␤-leucine
was preferred enantiomer. Acyl moieties of all enantiomer pairs
adopted binding modes that allowed the moiety to fit in the acyl-
binding site of PGAA, in which their phenyl rings formed ␲–␲
interactions with Phe24␤ and most frequently also hydrophobic
interactions with Phe24␤, Thr68␤ and Ala69␤, see Supplementary
data (Figs. S9 and S10)· Based on our docking experiments we rec-
ognized, in addition to acyl binding subsite, also the “secondary
hydrophobic” subsite formed mainly by the following amino acid
residues: Ala69␤, Phe71␤, Ala149␣ and Phe146␣.
Experimentally determined enantioselectivity of PGAA and PGAEc towards N-PhAc-
␣/␤-amino acids.
Substrate
E-values
PGAA
PGAEc
N-PhAc-␣-phenylalanine
N-PhAc-␣-homophenylalanine
N-PhAc-␤-leucine
90
80
68
80
80
5
5
3
5
5
4
5
90
80
60
80
80
55
75
5
5
5
5
5
5
5
N-PhAc-␣-leucine
N-PhAc-␣-isoleucine
N-PhAc-␤-homoleucine
N-PhAc-␣-tert-leucine
85
105
tivities of PGAA with PGAEc, an enzyme exhibiting 50 % sequence
identity with PGAA [34].
Racemates of substrates listed in Table 1 were used in enantios-
elective hydrolytic reactions catalyzed by purified PGAEc and PGAA
and their enantioselectivity values (E) were determined using HPLC
analysis (Table 2). A high degree of enantioselectivity (E = 90 5) for
both PGAA and PGAEc was observed in enzyme-catalyzed hydrol-
yses of N-PhAc-␣-phenylalanine. Both PGAA and PGAEc exhibited
similar degree of enantioselectivity towards substrate N-PhAc-
␣-leucine and N-PhAc-␣-isoleucine. On the other hand, PGAA
exhibited markedly higher enantioselectivity (in descending order)
with: N-PhAc-␤-homoleucine, N-PhAc-␣-tert-leucine and N-PhAc-
␤-leucine (Table 2).
Two mechanisms of PGAA’s enantioselectivity may be recog-
nized accordingly to the bulkiness of substrate’s substituent.
3.2. Structural basis of enantioselectivity of PGAA
3.3. Enantioselectivity mechanism adopted by substrates with
bulky substituents
A lot of effort has been invested into the structural analysis
of PGAs and understanding of its substrate-enzyme interactions.
Amino acid residues Ser1␤, Gln23␤, Ala69␤ and Asn241␤ [42] are
recognized as essential amino acid residues of aminic subsite in
the eleven PGAs so far characterized [5]: Ala69␤ and Asn241␤
participate in formation of “oxyanion - hole”, Gln23␤ stabilizes
E-S tetrahedral intermediate and the Ser1␤ residue serves as a
catalytic residue. It is proposed that the general topology and
the quarternary structure of the active site cavity might be con-
served throughout the PGA family [43]. However, slight variations
in amino acid residues are evident in the acyl binding subsite that is
formed in E. coli by Met142␣, Arg145␣, Phe146␣, Phe24␤, Thr32␤,
Pro49␤, Val56␤, Trp154␤, and Ile177␤ [42,44]. Due to so far unsuc-
cessful effort to crystallize the PGAA (data not shown), we decided
to prepare homology model of PGAA to study molecular mechanism
beyond enantioselectivities of this enzyme.
N-PhAc-␣-phenylalanine, N-PhAc-␣-homophenylalanine, N-
PhAc-␣-isoleucine, and N-PhAc-␣-tert-leucine all adopted similar
binding mode in the aminic subsite without exception (Fig. 2A).
The bulky, hydrophobic nucleophilic substituents were all oriented
to the secondary hydrophobic subsite. In the case of N-PhAc-
␣-phenylalanine and N-PhAc-␣-homophenylalanine containing
the second aromatic ring in the structure of their nucleophilic
substituents, additional ␲–␲ and hydrophobic interactions were
formed between this ring and either with Phe71␤ or Phe146␣
residues within this site see Supplementary data (Fig. S9). The car-
boxyl moieties of all four substrates were oriented towards part of
the active site with positively charged and hydrophilic amino acid
residues (e.g., Arg263␤, Ser390␤, Asn214␤ and N-terminal part of
Ser1␤). Oxygens from the carbonyl group of (S)-enantiomers were
always properly stabilized by amino acid residues of the “oxyanion-
hole” in the orientation enabling nucleophilic attack upon the
The model of PGAA was used as a receptor in molecular dock-
ing experiments with both enantiomeric forms of seven substrates
Table 3
Reactivity of binding modes of ␣/␤-amino acids in PGAA. Disfavoring interactions beyond the cutoff values are highlighted in bold.
Substrate
Enantiomer
Distance (Å)
Predicted as reactive
Predicted
preference
OSer1␤ → C
NAla69␤ → O
NAsn241␤ → O
OGln23␤ → H
(S)
(R)
(S)
(R)
(R)
(S)
(S)
(R)
(S)
(R)
(S)
(R)
(S)
(R)
3.4
5.2
3.3
4.6
3.2
3.4
3.4
3.4
3.6
4.5
3.2
3.4
3.3
4.7
4.0
6.3
3.7
6.2
3.7
3.9
3.9
4.0
4.1
6.2
3.8
3.8
4.0
6.3
4.0
7.9
3.9
7.4
4.0
4.0
3.8
4.1
3.9
7.4
3.7
4.0
4.1
7.5
3.9
5.4
3.3
5.2
3.5
4.4
3.0
4.6
2.8
5.1
2.9
4.4
3.3
4.8
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
N-PhAc-␣-phenylalanine
N-PhAc-␣-homophenylalanine
N-PhAc-␤-leucine
(S)
(S)
(R)
(S)
(S)
(S)
(S)
N-PhAc-␣-leucine
N-PhAc-␣-isoleucine
N-PhAc-␤-homoleucine
N-PhAc-␣-tert-leucine

244
M. Grulich et al. / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 240–247
Fig. 2. Predicted binding modes of seven experimentally tested substrates within the active site cavity of PGAA. (A) Binding modes of enantiomers with bulky substituents
(moieties: ␣-phenylalanine, ␣-homophenylalanine, ␣-isoleucine, and ␣-tert-leucine). (B) Binding modes of enantiomers with less bulky substituents. Grey surface—the
active site cavity; red surface—the acyl-binding subsite; orange surface—the secondary hydrophobic subsite. Crucial catalytic amino acid residues of the active site are green;
preferred and non-preferred enantiomers are cyan and yellow, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)
carbon atom of the carbonyl by Ser1␤. Also hydrogens from amino
groups of substrates were properly stabilized by Gln23␤. On the
contrary, binding modes of (R)-enantiomers had their peptide
bonds oriented in the opposite direction (Fig. 2A). This orienta-
tion resulted into larger spatial separation of substrates’ carbonyl
oxygen and hydrogen of stabilizing functional amino group of the
enzyme. Such binding modes also effectively shielded the attacked
carbon from the nucleophilic attack by Ser1␤. All this facts com-
bined suggest that (R)-enantiomers of this group of substrates were
poorly reactive at best, while (S)-enantiomers of these compounds
were better stabilized for reaction as is illustrated by all four mea-
sured distances being much shorter contributing to its predicted
preference (Table 3).
N-PhAc-␣-leucine to adopt similar orientation as (S)-enantiomers
of N-PhAc-␤-homoleucine and N-PhAc-␣-leucine and therefore
to partially restore necessary stabilization of their carbonyl oxy-
gens by the residues of the “oxyanion-hole” and orientation of
attacked carbon towards catalytic nucleophile (Table 3). However,
the amino groups of substrates were not stabilized by Gln23␤,
and the obviously non-complementary interactions required by
these binding modes should further disfavor their formation. In
the case of N-PhAc-␤-leucine racemic mixture, binding modes of
both enantiomers followed the same pattern as described above
but (R)-enantiomer was the preferred form (Table 3).
3.5. Prediction of enantioselectivity of PGAA towards new
substrates
3.4. Enantioselectivity mechanism adopted by substrates with
less bulky substituents
Since the molecular docking of seven substrates into the homol-
ogy model of PGAA confirmed experimental results obtained with
purified PGAA (Table 2), we employed this approach for in-silico
prediction of enantioselective hydrolyses of non-conventional ␣/␤-
amino acid enantiomers of which have an application potential
(Table 4). We should stress here that enantioselective hydrol-
yses of majority of racemates of these substrates catalyzed by
so far characterized PGAs have not yet been reported. A refer-
ence to stereoselectivity of PGA with N-PhAc-3-aminopent-4-ynoic
acid derivative and N-PhAc-p-Cl-␤-phenylalanine can be found in
research studies [45,46].
Results of docking experiments with pure enantiomers of race-
mates of seven N-PhAc-␣/␤-amino acids into active site cavity of
homology model of PGAA are shown in Table 5.
Similarly to previous analysis, both enantiomers exhibited com-
parable binding energies ruling out a differential binding as a source
This group of substrates includes N-PhAc-␤-homoleucine, N-
PhAc-␣-leucine and N-PhAc-␤-leucine. The binding modes of
(S)-enantiomers of N-PhAc-␤-homoleucine and N-PhAc-␣-leucine
followed the scheme for substrates with the bulky substituents
(Fig. 2B), with the exception that these substituents were not ori-
ented towards the secondary hydrophobic subsite of the binding
cavity. The predicted binding modes of these two (S)-enantiomers
fulfilled all criteria for their reactivity (Table 3).
In the case of (R)-enantiomers of N-PhAc-␤-homoleucine
and N-PhAc-␣-leucine, the carboxyl groups pointed towards the
secondary hydrophobic subsite and simultaneously their non-
polar substituents pointed towards the positively charged and
hydrophilic residues (Fig. 2B). These binding modes allowed the
peptide bonds of (R)-enantiomers of N-PhAc-␤-homoleucine and

M. Grulich et al. / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 240–247
245
Table 4
Novel substrates used for prediction of PGAA enantioselectivity.
Substrates
Structure
Product application
References
N-PhAc-3-aminopent-4-ynoic acid
(S)-enantiomer—building block of Elarofiban
(S)-enantiomer—building block for Xemilofiban
[47]
[48]
N-PhAc-3-amino-3-pyridin-propanoic acid
N-PhAc-3-amino-3-benzyl-2-hydroxybutanoic
acid
(S)-enantiomer—building block for Paclitaxel
(S)-enantiomer—building block of Lidamycin
[49]
[50]
N-PhAc-3-amino-3-(3-chloro-4,5-
dihydroxyphenyl) propanoic
acid
N-PhAc-p-Cl-␣-phenylalanine
N-PhAc-p-F-␣-phenylalanine
(S)-enantiomer—precursor for synthesis of analgesic Zolmitriptan
(R)-enantiomer—building block of Abarelix
[51]
[51]
N-PhAc-p-Cl-␤-phenylalanine
(S)-enantiomer market product
Sigma–Aldrich
of potential enantiopreference of PGAA (Table S3). Therefore we fur-
ther focused on evaluation of differences in potential reactivity of
these novel substrates using validated geometric criteria. As shown
for ␣/␤-amino acids, such analysis was able to successfully predict
high enantioselectivity in cases, when reactivity of one enantiomer
was seriously disfavored. Based on this analysis, we predicted
high enantiopreference of PGAA for three (R)-enantiomers and
two (S)-enantiomers of novel substrates (Table 5). However,
we could not entirely exclude possibility of moderate-to-low
enantioselectivity for remaining two substrates. In the case of N-
PhAc-3-aminopent-4-ynoic acid, binding of its rigid and extended
ethynyl group into the secondary hydrophobic subsite of the PGAA
prevented stabilization of carbonyl oxygen of both enantiomers
by the residues of “oxyanion hole” as well as proper stabiliza-
tion by interactions with Gln23␤, predicting both enantiomers as
inactive or poorly active. The binding modes of (R)-enantiomers
of both N-PhAc-3-amino-3-pyridin-propanoic acid and N-PhAc-
3-amino-3-(3-chloro-4,5-dihydroxyphenyl) propanoic acid had
their aromatic nucleophilic substituents located in the secondary
hydrophobic subsite as was the case of the binding modes for the
group of substrates containing bulkier substituents. Interestingly,
the (S)-enantiomers of these compounds were bound with their
carboxyl groups in the secondary hydrophobic subsite, while their
aromatic nucleophilic substituents pointed towards hydrophilic
part of the active site in the same fashion as the group of substrates
containing less bulky substituents. Since the aromatic substituents
of these two substrates carry also some polar groups, their bind-
ing modes seemed to be relatively easy to accommodate into the
Table 5
In-silico prediction of enantioselective hydrolyses of novel substrates catalyzed by PGAA. Disfavoring interactions beyond the cutoff values are highlighted in bold.
Substrate
Enantiomer
Distance (Å)
Predicted
Predicted
as reactive
preference
OSer1␤ → C
NAla69␤ → O
NAsn241␤ → O
OGln23␤ → H
–
(R)
(S)
(R)
(S)
(S)
(R)
(R)
(S)
(S)
(R)
(R)
(S)
(R)
(S)
3.9
3.5
3.2
4.2
3.3
4.2
3.1
3.6
3.2
4.8
3.6
3.4
3.2
4.4
4.5
4.4
3.7
4.2
3.8
6.1
3.6
4.1
3.8
6.6
4.1
4.0
3.6
6.7
4.7
4.3
3.9
5.0
4.0
7.8
4.0
4.3
3.8
8.0
4.0
4.1
4.0
7.9
4.7
5.3
3.1
5.0
3.4
3.8
2.9
4.2
2.0
4.2
2.6
3.7
3.2
3.4
No
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
Yes
Yes
No
N-PhAc-3-aminopent-4-ynoic acid
N-PhAc-3-amino-3-pyridin-propanoic acid
N-PhAc-3-amino-3-benzyl-2-hydroxybutanoic acid
(R)
(S)
(R)
N-PhAc-3-amino-3-(3-chloro-4,5-dihydroxyphenyl)
propanoic acid
N-PhAc-p-Cl-␣-phenylalanine
(S)
–
N-PhAc-p-F-␣-phenylalanine
N-PhAc-p-Cl-␤-phenylalanine
(R)

246
M. Grulich et al. / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 240–247
PGAA’s binding site. In any case observed binding modes of (S)-
enantiomers provided poor interactions with all four functionally
important residues (Table 5). This analysis suggested high enantio-
preference towards (R)-enantiomers of these two substrates. In the
case of N-PhAc-3-amino-3-benzyl-2-hydroxybutanoic acid, both
enantiomers complied to the binding modes observed for group
of substrates having the bulky substituent, with the exception
that hydroxyl group of (S)-enantiomer took over the interaction
of carboxyl group with hydrophilic part of the active site cavity.
Nevertheless, a difference in measured distances indicated strong
(S)-preference (Table 5). Substrates N-PhAc-p-Cl-␣-phenylalanine
and N-PhAc-p-Cl-␤-phenylalanine fully followed behavior of the
group of substrates containing bulkier substituents including pre-
dicted enantiopreference towards their (S)- and (R)-enantiomers,
respectively. Enantiomers of N-PhAc-p-F-␣-phenylalanine adopted
extended conformations because their nucleophilic substituents
could fit better into aminic subsite being possibly stabilized by
␲–␲ interactions with Phe71␤. Although the carboxyl group of (R)-
enantiomer was being oriented into the secondary hydrophobic
subsite, the distances to interacting groups were favorable for both
enantiomers predicting their good reactivity (Table 5). Therefore,
no conclusion on possible enantioselectivity of PGAA towards this
substrate could be drawn.
Young Scientists for International Cooperation Empowerment”
(CZ.1.07/2.3.00/30.0037) project co-financed by the European
Social Fund and the state budget of the Czech Republic.
Special thanks to Peter Babiak for discussion on organic synthe-
ˇ
ses and to Eva Sebestová on homology model construction.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.molcatb.2015.09.
008.
References
[1] M. Arroyo, I. de laMatta, C. Acebal, M. Pillar Castillon, Appl. Microbiol.
Biotechnol. 60 (2003) 507–514.
[2] A.K. Chandel, L.V. Rao, M.L. Narasu, O.V. Singh, Enzyme Microb. Tech. 42
(2007) 199–207.
[3] R.P. Elander, Appl. Microbiol. Technol. 61 (2003) 385–392.
[4] C.F. Sio, W.J. Quax, Curr. Opin. Biotechnol. 15 (2004) 349–355.
ˇ
[5] M. Grulich, V. Steˇpánek, P. Kyslík, Biotechnol. Adv. 31 (2013) 1458–1472.
[6] A. Basso, P. Braiuca, S. Clementi, C. Ebert, L. Gardossi, P. Linda, J. Mol. Catal. B
Enzyme 19–20 (2002) 423–430.
[7] H.J. Duggleby, S.P. Tolley, C.P. Hill, E.J. Dodson, G. Dodson, P.C.E. Moody,
Nature 373 (1995) 264–268.
[8] R.C. Giordano, M.P. Ribeiro, R.L. Giordano, Biotechnol. Adv. 24 (2006) 27–41.
[9] K. Srirangan, V. Orr, L. Akawi, A. Westbrook, M. Moo-Young, C. Perry Chou,
Biotechnol. Adv. 31 (2013) 1319–1332.
4. Conclusions
ˇ
[10] K. Plhácˇková, S. Becˇka, F. Skrob, P. Kyslík, Appl. Microbiol. Biotechnol. 62
(2003) 507–516.
ˇ
[11] F. Skrob, S. Becˇka, K. Plhácˇková, V. Fotopulosová, P. Kyslík, Enzyme Microb.
In this study, we present results of the research into enan-
tioselectivity potential of a novel industrial catalyst PGAA towards
racemic mixtures of ␣/␤-amino acids, pure enantiomers of which
have an application in pharmaceutical industry. We have demon-
strated that PGAA exhibited markedly higher enantioselectivity
than PGAEc (in descending order) with: N-PhAc-␤-homoleucine, N-
PhAc-␣-tert-leucine, and N-PhAc-␤-leucine. We have constructed a
homology model of PGAA using availability of a crystal structure of
PGAEc. We were able to evaluate plausibility of the model by using
molecular docking calculations: the results of molecular model-
ing were identical with the enantiopreference of PGAA found for all
experimentally studied substrates. Follow-up docking experiments
with seven non-conventional N-PhAc-␣/␤-amino acids enabled us
to predict the enzyme enantiopreference for most of these sub-
strates. Predictions indicated high PGAA’s enantiopreference for
(R)-enantiomers of three compounds and (S)-enantiomers of two
substrates. In-silico prediction of enantiopreference revealed PGAA
convenience towards production of (S)-enantiomer of N-PhAc-
3-amino-3-benzyl-2-hydroxybutanoic acid, a potential building
block for paclitaxel. In-silico prediction of PGAA’s enantiomer pref-
erence could be powerful tool for selection of compounds that are
not available on the market but can serve as a source of chiral
compounds for pharmaceutical industry.
Tech. 32 (2003) 738–744.
[12] S. Becˇka, V. Steˇpánek, R.W. Vyasarayani, M. Grulich, J. Marsˇálek, K. Plhácˇková,
ˇ
M. Dobisˇová, H. Maresˇová, M. Placˇková, R. Valesˇová, A. Palyzová, A. Datla, T.K.
Ashar, P. Kyslík, Appl. Microbiol. Biotechnol. 98 (2014) 1195–1203.
ˇ
[13] L. Sobotková, V. Steˇpánek, K. Plhácˇková, P. Kyslík, Enzyme Microb. Tech. 19
(1996) 389–397.
ˇ
[14] P. Kyslík, V. Steˇpánek, L. Hollerová, S. Becˇka, W.R., Vyasarayani, A., Datla, K.
Plhácˇková, J. Marsˇálek US patent 8039604B2, 2011 Oct 18.
[15] C. Kutzbach, E. Rauenbusch, Hoppe-Seyler’s Z. Physiol. Chem. 354 (1974)
45–53.
[16] C.S. Chen, Y. Fujimoto, G. Girdaukas, C.J. Sih, J. Am. Chem. Soc. 104 (1982)
7294–7299.
[17] J.D. Thompson, D.G. Higgins, T.J.W. Gibson, Nucleic Acids Res. 22 (1994)
4673–4680.
[18] K. Arnold, L. Bordoli, J. Kopp, T. Schwede, Bioinformatics 22 (2006) 195–201.
[19] J.U. Bowie, R. Luthy, D. Eisenberg, Science 253 (1991) 164–170.
[20] T. Castrignanò, P.D. De Meo, D. Cozzetto, I.G. Talamo, A. Tramontano, Nucleic
Acids Res. 34 (2006) 306–309.
[21] S.C. Lovell, I.W. Davis, W.B. Arendall III, P.I.W. de Bakker, J.M. Word, M.G.
Prisant, J.S. Richardson, D.C. Richardson, Proteins Struc. Funct. Genet. 50
(2002) 437–450.
[22] P. Benkert, M. Künzli, T. Schwede, Nucleic Acids Res. 37 (2009) 510–514.
[23] P. Benkert, M. Biasini, T. Schwede, Bioinformatics 27 (2011) 343–350.
[24] S.-L. Liu, Q.-X. Song, D.-Z. Wei, Y.-W. Zhang, X.-D. Wang, Prep. Biochem.
Biotechnol. 36 (2006) 235–241.
[25] D. Seeliger, B.L. deGroot, J. Comput. Aided Mol. Des. 24 (2010) 417–422.
[26] M.D. Hanwell, D.E. Curtis, D.C. Lonie, T.M. Vandermeersch, E. Zurek, G.R.
Hutchinson, J. Cheminform. 1 (2012) 4–17.
[27] T.A. Halgren, J. Comput. Chem. 17 (1996) 490–519.
[28] O. Trott, A.J. Olson, J. Comput. Chem. 31 (2010) 455–461.
[29] The PyMOL Molecular Graphics System [computer program]. Version 1.4,
DeLano Scientific, San Carlos, CA, USA, 2004.
Ethical
[30] L. Daniel, T. Buryska, Z. Prokop, J. Damborsky, J. Brezovsky, J. Chem. Inf. Model.
55 (2015) 54–62.
The authors are aware of ethical standards and declare that they
have no conflicts of interests. The research does not involve human
being or animals.
[31] K. Stierand, P. Maaß, M. Rarey, Bioinformatics 22 (2006) 1710–1716.
[32] M. Cole, Biochem. J. 115 (1969) 741–745.
[33] G. Lucente, A. Romeo, D. Rossi, Experientia 21 (1965) 317–318.
[34] S.E. McVey, M.A. Walsh, G.G. Dodson, K.S. Wilson, J.A. Brannigan, J. Mol. Biol.
313 (2001) 139–150.
[35] M.D. Berry, J. Neurochem. 90 (2004) 257–271.
Aknowledgements
[36] A.L. Ahmad, P.C. Oh, S.R.A. Ahukor, Biotechnol. Adv. 27 (2009) 286–296.
[37] G.C. Barrett, Chemistry and Biochemistry of the Amino Acids, Chapmann and
Hall, 1985.
[38] L. Breen, T.A. Churchward-Venne, J. Physiol. 590 (9) (2012) 2065–2066.
[39] W. Haag, S., Becquerelle, S., Valvani, O. Abril-Horpel, EP patent 2296693A1,
2012 Mar 7.
[40] D. Seebach, J.L. Matthews, Chem. Commun. (1997) 2015–2022.
[41] A.S. Bommarius, M. Schwam, K. Stingl, M. Kottenhahn, K. Huthmacher, K.
Drauz, Tetrahedron Asymmetr. 6 (1995) 2851–2888.
[42] B. Oh, K. Kim, J. Park, J. Hoon, D. Han, Y. Kim, Biochem. Biophys. Res. Commun.
319 (2004) 486–492.
This publication is supported by the project “BIOCEV - Biotech-
nology and Biomedicine Centre of the Academy of Sciences and
Charles University”(CZ.1.05/1.1.00/02.0109), from the European
Regional Development Fund, by long-term research development
project RVO 61388971 of the Institute of Microbiology, the project
SVV 260079/2014 and partially sponsored by Fermenta Biotech
Ltd., India. Work of JB was supported by the “Employment of Best

M. Grulich et al. / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 240–247
247
[43] J.L. Barbero, J.M. Buesa, G.G. deBuitrago, E. Mendez, A. Pez-Aranda, J.L. Garcia,
C.A. Maryanoff, J. Mills, S. Muller, A. Roessler, L. Scott, K.L. Sorgi, F.J. Villani,
R.R.H. Webster, C.h. Weh, Org. Proc. Res. Dev. 7 (2003) 866–872.
Gene 49 (1986) 69–80.
[44] J.M. Van der Laan, A.M., Riemens, W.J. Quax, US patent 5891703A, 1999 Apr 6.
[45] R.S. Topgi, J.S. Ng, B. Landis, P. Wang, J.R. Behling, Bioorg. Med. Chem. 7 (1999)
2221–2229.
[48] L.G. Frederick, O.D. Suleymanov, J.A. Szalony, B.B. Taite, A.K. Salyers, L.W. King,
L.P. Feigen, N.S. Nicholson, Circulation 25 (1998) 813–820.
[49] J. Rudat, B.R. Brucher, C. Syldatk, AMB Express 2 (2012) 11–21.
[50] J. Hu, Y.C. Xue, M.Y. Xie, R. Zhang, T. Otani, Y. Minami, Y. Yamada, T. Tarunaka,
J. Antibiot. 41 (1988) 1575–1579.
[46] V.A. Soloshonok, N.A. Fokina, A.V. Rybakova, I.P. Shishkina, S.V. Galushko, A.E.
ˇ
Sorochinsky, V.P. Kukhar, M.V. Savchenko, V.K. Svedas, Tetrahedron
Asymmetr. 6 (1995) 1601–1610.
[51] X. Gong, E. Su, P. Wang, D. Wei, Tetrahedron Lett. 52 (2011) 5398–5402.
[47] J.H. Cohen, M.E. Bos, S. Cesco-Cancian, B.D. Harris, J.T. Fortenstine, M. Justus,