Converting aspartase into a β-amino acid lyase by cluster screening

Article abstract of DOI:10.1002/cctc.201300986

Aspartase, despite being one of the most specific enzymes known, shows potential for application in β-amino acid synthesis. The substrate binding pocket of AspB from Bacillus sp. YM55-1 was reshaped by enzyme engineering to accommodate crotonic acid. The mutant enzyme BSASP-C6 yielded enantiopure (R)-3-aminobutyrate from crotonic acid and ammonia. To obtain this mutant, a high-throughput screening in combination with a focused permutational library was decisive for the success in this project. We achieved this by simultaneous randomisation of four residues within the substrate binding pocket and cluster screening in mixed population. Screening of 300 000 clones was necessary to find the BSASP-C6 mutant which has β-amino acid lyase activity. This exemplifies the need for efficient search and screening strategies in creating novel enzyme activities. The BSAPS-C6 enzyme developed here surpasses the natural substrate functionality of aspartase and thus represents the proof-of-concept of application of this novel synthetic route to a broader set of β-amino acids. β-Amino acids with aspartase mutants: Aspartase, despite being one of the most specific enzymes known, shows potential for application in β-amino acid synthesis. By applying a combination of focused library design and cluster screening, we created an aspartate mutant with β-amino acid lyase activity and thus open up a new platform to synthesize this interesting and important compound class. BSASP-C6= Mutant of AspB from Bacillus sp. YM55-1.

Full text of DOI:10.1002/cctc.201300986

CHEMCATCHEM
COMMUNICATIONS
DOI: 10.1002/cctc.201300986
Converting Aspartase into a b-Amino Acid Lyase by
Cluster Screening
Andreas Vogel,*[a] Ramona Schmiedel,[a] Ute Hofmann,[a] Karl Gruber,[b] and Klaus Zangger[c]
Aspartase, despite being one of the most specific enzymes
known, shows potential for application in b-amino acid synthe-
sis. The substrate binding pocket of AspB from Bacillus sp.
YM55-1 was reshaped by enzyme engineering to accommo-
date crotonic acid. The mutant enzyme BSASP-C6 yielded
enantiopure (R)-3-aminobutyrate from crotonic acid and am-
monia. To obtain this mutant, a high-throughput screening in
combination with a focused permutational library was decisive
for the success in this project. We achieved this by simultane-
ous randomisation of four residues within the substrate bind-
ing pocket and cluster screening in mixed population. Screen-
ing of 300000 clones was necessary to find the BSASP-C6
mutant which has b-amino acid lyase activity. This exemplifies
the need for efficient search and screening strategies in creat-
ing novel enzyme activities. The BSAPS-C6 enzyme developed
here surpasses the natural substrate functionality of aspartase
and thus represents the proof-of-concept of application of this
novel synthetic route to a broader set of b-amino acids.
regarded concomitantly as a- and b-amino acid due to the
presence of a second carboxylate function. However, one sub-
stantial drawback is given by the enzyme’s narrow substrate
spectrum, which impedes its use for the production of other
valuable compounds. However, extension of the substrate
range is feasible, as implied by the reaction mechanism, in
which the b-carboxylate group plays a key role in stabilising
the intermediate negative charge, whereas the a-carboxylate is
not actively involved. Based on this analysis we started an
enzyme engineering project by simultaneous randomisation of
four residues within the a-carboxylate binding site. Cluster
screening in mixed population allowed for the efficient activity
screening of the library by HPLC. The screening of
300000 clones yielded an enzyme mutant, in which all four
amino acids selected for mutation were altered, that could
convert crotonic acid into (R)-3-aminobutyrate. Detailed analy-
sis of this mutant revealed that at least three exchanges were
decisive for the new activity. This study provided an excellent
example for the synergistic effects encountered by multiple si-
multaneous mutations, the identification of which is one of
the major hurdles in enzyme engineering. In this respect, the
effectiveness of the cluster screening has proven beneficial for
the rapid screening of large library sizes. By this means the
generally very specific aspartase was turned into a unique
enzyme that represented the first step of a synthetic platform
to the important class of b-amino acids.
b-Amino acids are interesting target molecules for a vast range
of applications in the pharmaceutical industry. Accessibility of
such molecules by using enantioselective enzymes would be
highly desired. The industrial production of a-amino acids has
been very well established with enzymes that catalyse the ad-
dition of ammonia to unsaturated carboxylic acids (e.g., aspar-
tase and phenylalanine ammonia lyase). In contrast, the utiliza-
tion of those enzymes for the production of b-amino acids has,
so far, not been described. As a starting point for an engineer-
ing project aiming at such a reaction type, we have chosen as-
partase. This enzyme has proven robust and highly efficient in
industrial processes and its natural substrate (aspartate) can be
The enzymatic addition of ammonia to a,b-unsaturated car-
boxylic acids to yield a-amino acids is catalysed by amino acid
lyases to form phenylalanine, histidine, tyrosine, methylaspar-
tate or aspartate.[1–4] These enzymes are mechanistically very
diverse and only aspartate and methylaspartate can be regard-
ed as both a- and b-amino acids. Upon addition of ammonia
to fumarate (the natural substrate for aspartase), a negative
charge has to be stabilised in the reaction intermediate
(Scheme 1). In aspartate this is achieved through the b-carbox-
ylate group, the position of which enables the formation of an
enediolate species.[5]
[a] Dr. A. Vogel, Dr. R. Schmiedel, Dr. U. Hofmann+
c-LEcta GmbH
Perlickstr. 5, 04105 Leipzig (Germany)
Fax: (+49)341-35521433
E-mail: andreas.vogel@c-lecta.de
Following this mechanistic model, the a-carboxylate may ac-
tivate the substrate by its electron-withdrawing properties but
would not be actively involved in the charge stabilisation
during ammonia addition. We thus rationalised that we could
use aspartase for b-amino acid synthesis if the corresponding
binding site in the enzyme is adapted to accommodate other
functional groups. Several enzyme engineering studies have
shown that targeting the substrate binding site residues is the
most effective approach to gain activity with a new sub-
strate.[6–8] However, aspartase is regarded as one of the most
specific enzymes known, with extensive studies failing to iden-
tify any alternative amino acid substrates that can replace l-as-
[b] Prof. K. Gruber
Institute of Molecular Biosciences
University Graz
Humboldtstr. 50, 8010 Graz (Austria)
[c] Prof. K. Zangger
Institute of Chemistry, Organic and Bioorganic Chemistry
Heinrichstrasse 28, 8010 Graz (Austria)
[+] Current address:
Department of Medical Engineering and Biotechnology
University of Applied Sciences Jena
Carl-Zeiss-Promenade 2, 007745 Jena (Germany)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201300986.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 965 – 968 965

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
ed towards the a-carboxylate
and is therefore included in the
mutant library. The mutant li-
brary was prepared with simulta-
neous full randomisation at each
of the four sites, yielding a library
that contains every combination
of amino acids in the a-carboxyl-
ate binding site. Hence, the the-
oretical number of mutants in
Scheme 1. Rationale for using aspartase for b-amino acid synthesis. As implied by the reaction mechanism, only
the b-carboxylate is required for the reaction.
partic acid.[9,10] The only study in which the substrate scope of
aspartase could be changed is reported by Asano et al.[10] By
screening a random mutagenesis library of the Escherichia coli
aspartase gene, the authors found a K327N mutant with activi-
ty towards l-aspartic acid-a-amide, which can be regarded as
a minimal derivative of the natural substrate.
this library is 160000 on the amino acid level but approximate-
ly 1 million on the DNA level (NNK degeneration contains 32
codon combinations), which is the relevant number for the
screening coverage.[13,14] As the main screening substrate, we
used crotonic acid (trans-2-butenoic acid), which carries
a methyl instead of a carboxylate group, in contrast to fumaric
acid. The wild-type enzyme did not show any activity to cro-
tonic acid. Given the large library size and the non-existent
starting activity, we applied cluster screening[15] as it is excep-
tionally well-suited to find rare events. A similar strategy was
published by the group of Bommarius.[16] The cluster complexi-
ty (e.g., number of clones mixed per well, see the Experimental
Section) was chosen based on the assumption that only a frac-
tion of the wild-type activity (<0.1%) could be achieved, re-
sulting in cluster sizes of approximately 20000 clones per
As a starting enzyme for our study we used AspB from Bacil-
lus sp. YM55-1 (BSASP).[11,12] The enzyme is highly active, very
stable, well expressed in E. coli and a crystal structure has been
solved,[5,12] making it an excellent starting point for enzyme en-
gineering. At the project start, only the apo structure of the
enzyme was available (PDB entry 1J3U) and therefore used for
modelling the complex with aspartate. This model was later
confirmed by the published crystal structure for the substrate-
bound BSASP (PDB entry 3R6V). In the complex structure, as-
partate is rigidly bound by a number of salt bridges and H-
bonds from amino acid side chains (Figure 1).
plate. After screening
a considerable amount of clones
(ꢀ300000) without any new signal, we obtained a cluster
(from a 96-well plate with a total of ꢀ10000 clones) with
a signal at the expected retention time for the product and,
consequently, singled out this hit clone (Figure 2). The signal
from the single clone was tremendously high and delivered
the first confirmation of the activity in the new mutant.
Direct polar interactions with the a-carboxylate group are
thus formed by the side chains of Thr187, Lys324 and Asn326.
His188 and Asn142 show secondary interactions but interact
mainly with the amino group of aspartate and thus remain un-
altered in the mutagenesis experiments. Met321 is also orient-
The newly acquired b-amino acid lyase activity was con-
firmed further on the isolated clone BSASP-C6. The specific ac-
tivity with crotonic acid was 0.1 UmgÀ1 at 378C and 0.45 at
608C, approximately 10000-fold lower than the activity of the
wild-type BSASP on its native substrate fumarate.[17] The Mi-
Figure 2. Identification of 3-aminobutyrate-forming enzyme by cluster
screening. Chromatogram from HPLC with fluorescent detection after deriva-
tisation with ortho-phthaldialdehyde from screening of aspartase mutant li-
braries for 3-aminobutyrate-forming enzymes. The hit cluster shows a signal
corresponding to the retention time of the expected product. The cluster
complexity was 10000 clones per 96-well plate. After singling out the posi-
tive clone from the hit cluster, the signal of the single clone is significantly
improved. RFI=Relative fluorescent intensity.
Figure 1. Identification of residues for a combinatorial mutant library. Close-
up view of the substrate binding site of BSASP (PDB entry 3R6V). The bound
l-aspartate is shown in yellow, neighbouring residues (within 6 ꢁ) in grey.
The four residues that were modified in the active variant are shown in ma-
genta. Polar interactions between the substrate and the enzyme are depict-
ed as dashed green lines. The figure was prepared by using the program
PyMOL (Schrodinger Inc.).
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 965 – 968 966

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
chaelis constant Km of BSASP-C6 for crotonic acid was deter-
mined as 120 mm and the kcat to 0.09 sÀ1 at 378C. Compared
to the published constants of the wild-type enzyme with fuma-
rate (Km =30 mm and kcat =640 sÀ1),[17] the reduced catalytic ef-
ficiency was mainly due to the limited maximum catalytic rate,
whereas the Km was just moderately higher. The missing acti-
vating effect of the electron-withdrawing a-carboxylate may
have contributed further to the reduced activity. The BSASP-C6
enzyme is absolutely regio- and enantioselective: only (R)-3-
aminobutyrate was detected by NMR and HPLC analyses of the
reaction product (see the Supporting Information). The mutant
was applied to a preparative reaction at a substrate concentra-
tion of 300 mm. A final conversion of approximately 60% was
obtained after 100 h, which was most likely a result of the un-
favourable kinetic constants of this first-round mutant (Fig-
ure S4). Several aliphatic and aromatic a,b-unsaturated com-
pounds were investigated for turnover with BSASP-C6. Turn-
over could be detected with trans-2-pentanoic acid only. No
activity with the wild-type substrate fumarate was detected.
The amino acids at all four sites were exchanged in BSASP-
C6 to hydrophobic residues (T187C, M321I, K324M and N326C),
which would be expected for the transformation of a binding
site to accommodate a methyl group instead of a carboxylate.
However, this unusual acquisition of two new cysteine residues
in a substrate binding site raised the question of whether a di-
sulphide bridge was formed. The distance between the Ca
atoms of the corresponding amino acids in the wild-type
enzyme structure was 8.4 ꢁ, longer than the maximum dis-
tance for disulphide-forming sites (6.6 ꢁ). The region of the
amino acids T187 and N326 was also very rigid (as opposed to
the M321/K324 region),[5] which also contradicted disulphide
bridge formation in the BSASP-C6 mutant.
back-mutations retained this activity, however, with significant-
ly lower activity (maximum 37%) compared to BSASP-C6. Fur-
ther variants with fewer mutations that retained K324M and
created all double combinations of the remaining three sites
resulted in a near complete loss of the activity towards 3-ami-
nobutyrate (Table 1).
Single-site saturation libraries on the wild-type enzymes did
not yield any detectable activity either and single-site satura-
tion libraries on the BSASP-C6 mutant did not yield improved
variants (data not shown). It thus became apparent that, at
least with the strategy applied here, a minimum of three muta-
tions had to occur simultaneously within the substrate binding
pocket to produce significant b-amino acid ammonia lyase ac-
tivity. To achieve this, a high-throughput screening similar to
the cluster-screening in combination with a focused permuta-
tional library was decisive for the success in this project.
The newly acquired substrate of BSASP-C6 possesses mainly
electronic changes and no further steric burden. Such
a change in substrate scope is achieved much more easily with
other enzymes classes such as alcohol dehydrogenases, transa-
minases or nitrilases.[18–20] The present aspartase engineering
study thus represents an example that evolvability of highly
specialised enzymes from primary metabolism is much less
pronounced than in enzymes from secondary metabo-
lisms.[21,22]
The study shown here demonstrates that focused library
design in combination with cluster screening is paving the
way to novel enzyme activities. Previous attempts to use am-
monia lyases for b-amino acid synthesis are based on minimal
substrate changes[10] or mimicking natural activity.[23] The
BSASP-C6 developed here surpasses the natural substrate func-
tionality and thus represents the proof-of-concept for applica-
tion of this synthetic route to a broader set of compounds.
As the crotonic acid-converting mutant was very unique in
this library and special with regard to the resulting amino acid
sequence, we analysed it further. To check the individual con-
tribution of each mutation, we performed back-mutation to
the wild-type amino acid on each site, resulting in three triple
mutants (Table 1). The K324M mutation turned out to be cru-
cial for the activity towards crotonic acid, whereas all other
Experimental Section
A synthetic gene with optimised codon usage for the amino acid
sequence of Bacillus sp. YM55-1 aspartase AspB (Uniprot Q9LCC6)
was obtained from Genscript (USA) and cloned into the c-LEctas
expression vector pLE1A10, which placed the gene under control
of a T7 promoter. Expression took place in BL21(DE3) in ZYM505
medium.[24] After induction with 0.1 mm isopropyl b-d-1-thiogalac-
topyranoside, cells were grown at 308C overnight and harvested.
Cells containing recombinant enzyme were lysed by repeated
freeze–thaw cycles in a mixture of 50 mm 4-(2-hydroxyethyl)-1-pi-
perazineethanesulfonic acid (HEPES, pH 7.0), 2 mm MgCl2,
0.1 mgmLÀ1 deoxyribonuclease I and 0.5 mgmLÀ1 lysozyme. Culti-
vation and cell extract preparation was adapted to 96-well plate
format to allow cluster screening. The mutant library was prepared
by creating polymerase chain reaction fragments with NNK-rando-
mised oligonucleotides and assembled by an overlap extension
polymerase chain reaction and restriction and ligation.[25]
Table 1. Dissection of the individual contributions of the mutations in
BSASP-C6 to the b-amino acid lyase activity.
Mutation[a]
Activity[b] [UmgÀ1
]
Relative activity [%]
T187C, M321I, K324M, N326C[c]
T187C, M321I, K324M
M321I, K324M, N326C
T187C, K324M, N326C
T187C, M321I, N326C
K324M, N326C
0.45
0.17
0.12
100
39
26
0.13
28
<0.01[d]
<0.01[d]
<0.01[d]
<0.01[d]
<2
<2
<2
<2
M321I, K324M
T187C, K324M
[a] Mutation to AspBwild-type amino acid sequence. [b] Activity for 3-ami-
nobutyrate formation at 608C of the bacterial crude extract. Specific ac-
tivities were calculated based on the BSASP content from sodium dodecyl
sulfate polyacrylamide gel electrophoresis analysis. [c] Originally isolated
BSASP-C6 mutant. [d] Activities below the detection limit of the applied
analytics (<0.01 UmgÀ1).
Biotransformations were performed with the bacterial crude ex-
tracts or whole cells in a mixture of 100 mm HEPES (pH 8.0) and
2 mm MgCl2.
In the initial screening, a substrate concentration of 20 mm croton-
ic acid, 50 mm ammonia and 378C was used. Crude extracts were
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 965 – 968 967

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
[1] M. M. Heberling, B. Wu, S. Bartsch, D. B. Janssen, Curr. Opin. Chem. Biol.
2013, 17, 250–260.
precipitated with 60% ammonium sulfate and pellets containing
aspartase variants were resolubilised in HEPES buffer.
[2] N. J. Turner, Curr. Opin. Chem. Biol. 2011, 15, 234–240.
[3] Y. Asano, Y. Kato, C. Levy, P. Baker, D. Rice, 2004, 22, DOI: 10.1080/
10242420410001703496.
[4] M. de Villiers, V. Puthan Veetil, H. Raj, J. de Villiers, G. J. Poelarends, ACS
Chem. Biol. 2012, 7, 1618–1628.
Characterisation of BSASP-C6 and the other mutants was done at
pH 8.0 with 300 mm crotonic acid and 300 mm ammonium chlo-
ride and incubation at 37 and 608C. Reactions were stopped at ap-
propriate time intervals by heating to 808C for 20 min, followed by
centrifugation and derivatisation of the supernatant for HPLC de-
tection. Two different derivatisation reagents were used depending
on the sensitivity required. For the cluster screening, derivatisation
with ortho-phthaldialdehyde and ethanethiol[26] and subsequent
HPLC with fluorescent detection was applied. For single clones and
determination of specific activity, derivatisation with 1-fluoro-2,4-di-
nitrobenzene and HPLC with UV detection at 360 nm was applied.
[5] G. Fibriansah, V. P. Veetil, G. J. Poelarends, A.-M. W. H. Thunnissen, Bio-
chemistry 2011, 50, 6053–6062.
[6] K. L. Morley, R. J. Kazlauskas, Trends Biotechnol. 2005, 23, 231–237.
[7] M. T. Reetz, M. Bocola, J. D. Carballeira, D. Zha, A. Vogel, Angew. Chem.
Int. Ed. 2005, 44, 4192–4196; Angew. Chem. 2005, 117, 4264–4268.
[8] M. D. Toscano, K. J. Woycechowsky, D. Hilvert, Angew. Chem. Int. Ed.
2007, 46, 3212–3236; Angew. Chem. 2007, 119, 3274–3300.
[9] R. E. Viola, Adv. Enzymol. RAMB 2000, 74, 295–341.
[10] Y. Asano, I. Kira, K. Yokozeki, Biomol. Eng. 2005, 22, 95–101.
[11] Y. Kawata, K. Tamura, M. Kawamura, K. Ikei, T. Mizobata, J. Nagai, M.
Fujita, S. Yano, M. Tokushige, N. Yumoto, Eur. J. Biochem. 2000, 267,
1847–1857.
For determination of specific activity, enzyme was purified by heat
precipitation from crude extract by incubation at 658C for 30 min.
The purity obtained by this method was >90% (Figure S5). Protein
concentration was determined with the Bradford assay and bovine
serum albumin as standard. Specific activity was measured at
pH 8.0 with 300 mm crotonic acid and 300 mm ammonium chlo-
ride by incubation at 37 and 608C.
[12] T. Fujii, H. Sakai, Y. Kawata, Y. Hata, J. Mol. Biol. 2003, 328, 635–654.
[13] W. M. Patrick, A. E. Firth, Biomol. Eng. 2005, 22, 105–112.
[14] M. T. Reetz, D. Kahakeaw, R. Lohmer, ChemBioChem 2008, 9, 1797–
1804.
[15] T. Greiner-Stçffele, M. Struhalla, Method from the Selection of Biomole-
cules from Biomolecule Variant Libraries, WO2005040376, 2003.
[16] K. M. Polizzi, M. Parikh, C. U. Spencer, I. Matsumura, J. H. Lee, M. J.
Realff, A. S. Bommarius, Biotechnol. Prog. 2006, 22, 961–967.
[17] Y. Kawata, K. Tamura, S. Yano, T. Mizobata, J. Nagai, N. Esaki, K. Soda, M.
Tokushige, N. Yumoto, Arch. Biochem. Biophys. 1999, 366, 40–46.
[18] U. T. Bornscheuer, G. W. Huisman, R. J. Kazlauskas, S. Lutz, J. C. Moore, K.
Robins, Nature 2012, 485, 185–194.
[19] A. Vogel, D. Schwarze, T. Greiner-Stçffele, Nitrilases, WO2012048865,
2011.
[20] A. Vogel, R. Schmiedel, S. Kçpke, Method For The Alcohol Dehydrogen-
ase Reduction Of A Secosteroid Derivative, WO2013102619, 2012.
[21] J.-K. Weng, J. P. Noel, Cold Spring Harbor Symp. Quant. Biol. 2012, 77,
309–320.
For cluster screening,[15] the number of colony-forming units in the
primary transformation medium was determined by plating on se-
lective agar. The medium was inoculated with the desired number
of cells and evenly spread onto a 96-deep-well plate to ensure a de-
fined cluster complexity (e.g., a number of different cells in each
well). Cells were grown, lysed and assayed as described above.
After initial identification of an active hit cluster, cells from this
cluster were spread onto a new 96- -deep-well plate while lower-
ing the complexity, thereby increasing the content of active cells in
certain wells. The procedure was repeated until the complexity
was reduced to allow plating and single clone measurement (sin-
gling out of positive clones).
[22] J.-K. Weng, R. N. Philippe, J. P. Noel, Science 2012, 336, 1667–1670.
´
[23] B. Wu, W. Szymanski, G. G. Wybenga, M. M. Heberling, S. Bartsch, S.
de Wildeman, G. J. Poelarends, B. L. Feringa, B. W. Dijkstra, D. B. Janssen,
Angew. Chem. Int. Ed. 2012, 51, 482–486; Angew. Chem. 2012, 124,
497–501.
Acknowledgements
Financial support by the Europꢀischer Fonds fꢁr regionale En-
twicklung (EFRE) and the state of Saxonia under the ERA-IB proj-
ect “Eng-Biocat” is greatly acknowledged. We thank Dr. Andreas
Buthe for carefully proof-reading the manuscript.
[24] F. W. Studier, Protein Expression Purif. 2005, 41, 207–234.
[25] T. Greiner-Stçffele, C. Feller, M. Struhalla, Process for Generating a Variant
Library of DNA Sequences, WO2009146892, 2009.
[26] R. Hanczkꢂ, A. Jꢃmbor, A. Perl, I. Molnꢃr-Perl, J. Chromatogr. A 2007,
1163, 25–42.
Received: November 17, 2013
Revised: January 8, 2014
Published online on April 1, 2014
Keywords: aspartase · amino acids · cluster screening ·
enzyme engineering · mutagenesis
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 965 – 968 968