Engineering methylaspartate ammonia lyase for the asymmetric synthesis of unnatural amino acids

Article abstract of DOI:10.1038/nchem.1338

The redesign of enzymes to produce catalysts for a predefined transformation remains a major challenge in protein engineering. Here, we describe the structure-based engineering of methylaspartate ammonia lyase (which in nature catalyses the conversion of 3-methylaspartate to ammonia and 2-methylfumarate) to accept a variety of substituted amines and fumarates and catalyse the asymmetric synthesis of aspartic acid derivatives. We obtained two single-active-site mutants, one exhibiting a wide nucleophile scope including structurally diverse linear and cyclic alkylamines and one with broad electrophile scope including fumarate derivatives with alkyl, aryl, alkoxy, aryloxy, alkylthio and arylthio substituents at the C2 position. Both mutants have an enlarged active site that accommodates the new substrates while retaining the high stereo- and regioselectivity of the wild-type enzyme. As an example, we demonstrate a highly enantio- and diastereoselective synthesis of threo-3-benzyloxyaspartate (an important inhibitor of neuronal excitatory glutamate transporters in the brain).

Full text of DOI:10.1038/nchem.1338

ARTICLES
PUBLISHED ONLINE: 29 APRIL 2012 | DOI: 10.1038/NCHEM.1338
Engineering methylaspartate ammonia lyase for
the asymmetric synthesis of unnatural amino acids
1
2,3
1
4
1
Hans Raj , Wiktor Szyma n´ ski , Jandr e´ de Villiers , Henri e¨ tte J. Rozeboom , Vinod Puthan Veetil ,
1
1
5
6
1
Carlos R. Reis , Marianne de Villiers , Frank J. Dekker , Stefaan de Wildeman , Wim J. Quax ,
4
3
2
1
Andy-Mark W. H. Thunnissen , Ben L. Feringa , Dick B. Janssen and Gerrit J. Poelarends
*
The redesign of enzymes to produce catalysts for a predefined transformation remains a major challenge in protein
engineering. Here, we describe the structure-based engineering of methylaspartate ammonia lyase (which in nature
catalyses the conversion of 3-methylaspartate to ammonia and 2-methylfumarate) to accept a variety of substituted
amines and fumarates and catalyse the asymmetric synthesis of aspartic acid derivatives. We obtained two single-active-
site mutants, one exhibiting a wide nucleophile scope including structurally diverse linear and cyclic alkylamines and one
with broad electrophile scope including fumarate derivatives with alkyl, aryl, alkoxy, aryloxy, alkylthio and arylthio
substituents at the C2 position. Both mutants have an enlarged active site that accommodates the new substrates while
retaining the high stereo- and regioselectivity of the wild-type enzyme. As an example, we demonstrate a highly enantio-
and diastereoselective synthesis of threo-3-benzyloxyaspartate (an important inhibitor of neuronal excitatory glutamate
transporters in the brain).
ptically pure a-amino acids are highly valuable as tools for a mechanism has emerged for the MAL-catalysed reaction. In this
biological research and as chiral building blocks for pharma- proposed mechanism, an (S)- or (R)-specific catalytic base abstracts
1
O
ceuticals and (agro)chemicals . Industrial and academic the C3 proton of the respective stereoisomer of 3a to generate an
interest in these compounds, combined with the potential advan- enolate intermediate that is stabilized by coordination to the essen-
tages of replacing conventional chemical processes with biocatalysis, tial active site Mg ion. Collapse of this intermediate results in the
has fueled the development of enzymatic synthesis routes for enan- elimination of ammonia and yields mesaconate.
2
þ
tiomerically pure a-amino acids. Various enzymatic synthesis
Unfortunately, the substrate scope of MAL is very narrow. Only
routes have been developed, for example using hydantoinases, dehy- with a few small substituted amines and fumarates is a reasonable
drogenases, acylases, aminotransferases, ammonia lyases and ami- activity observed, yielding a limited number of substituted aspartic
1
–5
17–19
dases , the most attractive of which are those based on acids
. Accordingly, it would be very attractive to extend the
asymmetric synthesis in which a prostereogenic substrate is con- accessible range of aspartic acid derivatives by redesigning MAL
verted to an optically pure a-amino acid with 100% theoretical to convert various unnatural substrates, thereby enlarging its bioca-
yield. For example, the asymmetric addition of ammonia to the talytic applicability. In this Article, we describe the use of an efficient
double bonds of unsaturated acids catalysed by ammonia lyases is engineering strategy based on saturation mutagenesis at carefully
5
20–24
a very attractive strategy for the synthesis of chiral a-amino acids . chosen sites lining the substrate binding pocket
, with a view
This strategy makes use of readily available starting substrates to expanding both the nucleophile and electrophile spectrum of
without a requirement for cofactor recycling, implementation of MAL. We have obtained two single-active-site mutants, one that
dynamic kinetic resolution strategies or additional catalysts. accepts a large set of non-native amines and one that accepts
However, it is limited by the narrow substrate range of various non-native fumarate derivatives. These engineered MALs
ammonia lyases.
-Methylaspartate ammonia lyase (MAL) catalyses the reversible variety of substituted aspartic acids, which are valuable as tools
addition of ammonia (2a) to mesaconate (1b) to yield threo-(2S,3S)- for biological research and as building blocks for chemical and
have potential as catalysts for the asymmetric synthesis of a large
3
2
5–31
3
-methylaspartate (threo-3a) and erythro-(2S,3R)-3-methylaspar- pharmaceutical synthesis
. This potential is illustrated by the
6–10
tate (erythro-3a) as products (Table 1) . MAL is used by the bac- mutant MAL-catalysed synthesis of threo-3-benzyloxyaspartate, a
terium Clostridium tetanomorphum as part of a degradative widely used non-transportable blocker for all subtypes of neuronal
pathway that converts (S)-glutamic acid via threo-3a to yield excitatory amino-acid transporters
2
5,26
.
acetyl-coenzyme A6 . The crystal structure of MAL and that of
,11
the isozyme from Citrobacter amalonaticus have been solved by Results
12,13
X-ray crystallography . Based on kinetic isotope measure- Expansion of the nucleophile spectrum of MAL. The crystal
1
4,15
12,13,16
8
13
ments , structural studies
and mutagenesis experiments , structure of MAL complexed with the natural substrate threo-3a
1
Department of Pharmaceutical Biology, Groningen Research Institute of Pharmacy, University of Groningen, Antonius Deusinglaan 1, 9713 AV Groningen,
2
The Netherlands, Department of Biochemistry, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Nijenborgh 4, 9747
AG Groningen, The Netherlands, Center for Systems Chemistry, Stratingh Institute for Chemistry, University of Groningen, Nijenborgh 4, 9747 AG
Groningen, The Netherlands, Laboratory of Biophysical Chemistry, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen,
Nijenborgh 7, 9747 AG Groningen, The Netherlands, Department of Pharmaceutical Gene Modulation, Groningen Research Institute of Pharmacy,
3
4
5
6
University of Groningen, Antonius Deusinglaan 1, 9713 AV Groningen, The Netherlands, DSM Pharmaceutical Products, PO Box 18, 6160 MD, Geleen,
The Netherlands. *e-mail: g.j.poelarends@rug.nl
478
NATURE CHEMISTRY | VOL 4 | JUNE 2012 | www.nature.com/naturechemistry
2012 Macmillan Publishers Limited. All rights reserved.
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1338
ARTICLES
Table1 | Q73A-catalysed amine (2a–2u) addition to mesaconate (1b).
R
NH
*
Q73A
CO2
CO2
O2C
+
RNH2
O2C
*
1b
2
3
Entry
Nucleophile
R-group
Product
Conv. (%)*
d.e. (%)
5 (threo)
e.e. (%)
†
‡
‡
‡
‡
‡
‡
‡
‡
‡
‡
‡
‡
‡
§
§
§
§
§
§
§
§
ꢁ
1
2a
2b
2c
2d
2e
2f
2g
2h
2i
2j
2k
2l
2m
2n
2o
2p
2q
2r
H
3a
3b
3c
3d
3e
3f
3g
3h
3i
3j
3k
3l
3m
3n
3o
3p
3q
3r
79
ND
ND
.99
.99
.99
ND
2
3
4
5
6
7
8
9
Methyl
Ethyl
Propyl
Butyl
Pentyl
Hexyl
Isopropyl
Cyclobutyl
Cyclopentyl
Cyclohexyl
(Cyclopropyl)methyl
Benzyl
Cyclopropyl
Ethoxy
74
51
53
60
29
17
8
65
36
21
61
6
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
80 (threo)
}
}
}
ND
.99
.99
ND
}
}
1
1
1
1
1
1
1
1
1
1
0
1
ND
}
2
3
4
5
6
7
8
9
.99
ND
ND
ND
ND
ND
ND
ND
ND
ND
75
99
†
2-Hydroxyethyl
3-Hydroxypropyl
2-Methoxyethyl
N-Methyl-2-aminoethyl
2-Aminoethyl
37
56
60
80
90
78
.95 (threo)
.95 (threo)
.95 (threo)
.95 (threo)
2s
2t
2u
3s
3t
3u
2
2
0
1
3-Aminopropyl
†
‡
*
Reactions were allowed to proceed for 7 days to detect the formation of low amounts of erythro product isomers, if present. Reaction was allowed to proceed for 14 days. The d.e. (defined as excess of threo
1
isomer over erythro isomer) of the amino-acid product was determined by comparison of its H NMR signals in the crude reaction mixture to those of synthesized authentic standards with known threo or erythro
configuration. The purified amino-acid product could be tentatively assigned the threo configuration on the basis of analogy. ND, not determined. The e.e. of the isolated product was determined by chiral HPLC
using a synthesized authentic standard with threo-(DL) configuration.
§
ꢁ
}
suggests that the side chains of Q73 and Q172 are involved in the addition of methylamine (2b) to mesaconate (1b) was chosen
formation of the binding pocket for the amino group of the (Table 1). This small alkylamine is a poor substrate for wild-type
substrate (Fig. 1a). To engineer this putative amine binding MAL. We reasoned that if mutants displaying substantial activity
pocket, three focused libraries were generated by saturation were to be found in any of the libraries, then they could also be
mutagenesis at positions 73, 172 and 73/172, using NNS tested for their ability to process larger and more challenging
3
2
degeneracy . As the model reaction for activity screening, the amines (for example, amines 2c–2u).
The three libraries were used to transform Escherichia coli cells.
The double-site library was screened by evaluating ꢀ1,200 transfor-
mants, and ꢀ400 transformants were evaluated for each single-site
library. Although the library at position 172 failed to contain any
active mutants, screening of the libraries corresponding to positions
a
b
Tyr356
Gln172
Leu384
73 and 73/172 resulted in the identification of six mutants with pro-
nounced activity towards methylamine addition. DNA sequencing
revealed that all positive hits carry mutations at position 73,
namely Q73A, Q73G, Q73N, Q73P, Q73S and Q73T. These
mutants and the wild-type enzyme were overexpressed, purified to
homogeneity, and assayed for their ability to catalyse ammonia or
methylamine addition to mesaconate. When compared to wild-
type MAL, all mutants display a significant decrease in ammonia
addition activity and a large increase in methylamine addition
activity (Supplementary Fig. S1). The Q73A mutant showed the
highest catalytic rate for methylamine addition to mesaconate and
was therefore selected for further study.
2
.8 2.7
Phe170
4.4
2.7
6
.2
3.9
Gln73
Mg2+
Mg2+
Figure 1 | Crystal structure of wild-type MAL in complex with the natural
substrate threo-(2S,3S)-3-methylaspartate. a, Close-up of the active site
showing the hydrogen-bond interactions between the substrate’s amino
group and the side chains of Gln73 (via a water molecule) and Gln172. The
carbon atoms of the active-site residues are shown in green, and those of
the substrate in cyan. Hydrogen bonds are represented as dashed lines. The
magnesium ion and water molecule are shown as magenta and yellow
spheres, respectively. b, Close-up of the active site showing the observed
distances (in a˚ ngstroms, atoms connected by solid lines) between the
substrate’s methyl group and the side chains of the three residues (Phe170,
Tyr356 and Leu384) that are involved in the formation of the alkyl binding
pocket. Colour scheme as in a. The figures were prepared with PyMOL
A comparison of the k /K values showsthat the Q73A mutant is
cat
m
at least 140-fold more efficient in the addition of methylamine than
wild-type MAL (Supplementary Table S1). Wild-type MAL,
however, is ꢀ500-fold more efficient in the ammonia addition
reaction. Hence, the Q73A mutation moved the specificity of MAL
away from ammonia and towards methylamine (.70,000-fold shift
1
in nucleophile specificity as defined by k /Km values). H NMR
cat
spectroscopic analysis showed that the Q73A-catalysed ammonia
and methylamine additions to mesaconate lead to the same amino-
acidproducts asthe corresponding wild-type MAL-catalysed reactions
(
http://www.pymol.org).
NATURE CHEMISTRY | VOL 4 | JUNE 2012 | www.nature.com/naturechemistry
2012 Macmillan Publishers Limited. All rights reserved.
479
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1338
ARTICLES
a
demonstrates its surprisingly broad nucleophile scope. Control
experiments showed that the amines do not react with mesaconate
in the absence of enzyme.
100
For amines 2c–2m (Table 1, entries 3–13), the single products of
the Q73A-catalysed additions to mesaconate were identified as the
threo isomers of the corresponding N-substituted 3-methylaspartic
8
6
4
2
0
0
0
0
0
1
acids (3c–3m) by comparison of their H NMR signals in the crude
reaction mixtures to those of chemically synthesized authentic stan-
dards with known relative configuration (Supplementary Fig. S3).
No other regio- or diastereoisomers were observed. For a few
selected amines (2c–2e, 2h, 2i and 2l), preparative-scale reactions
were performed (1.2 mmol 1b). Purification of the amino-acid pro-
ducts and analysis by high-performance liquid chromatography
Methylamine (2b)
Cyclobutylamine (2i)
Ethoxyamine (2o)
(
HPLC) on a chiral stationary phase (Supplementary Fig. S4)
revealed that the Q73A mutant is also highly enantioselective, pro-
ducing 3c–3e, 3h, 3i and 3l with .99% enantiomeric excess (e.e.)
0
1
2
3
23
24
25
Time (h)
(
Table 1). For the other amines (Table 1, entries 14–21), the
Q73A-catalysed addition reactions were performed on a 3.1 mmol
scale (1b). The products were isolated and identified as the corre-
sponding N-substituted 3-methylaspartic acids (3n–3u). Although
the relative configuration of products 3n–3u has not been deter-
mined by comparison to authentic standards, we assume the relative
configuration to be threo for all products 3b–3u.
To further demonstrate the preparative usefulness of the Q73A
mutant, we optimized the reaction conditions for a few structurally
distinct amines (2b, 2i and 2o). With 0.05 mol% biocatalyst and a
b
1
00
8
6
4
2
0
0
0
0
0
1
15-fold molar excess of 2b or 2o (3 M each) over mesaconate
26 mM), the reactions were complete within 20–60 min at 25 8C,
achieving final conversions of almost 100% (Fig. 2a). For amine
i, a lower concentration (2 M) had to be used to avoid protein pre-
(
Fumarate (1a)
-hexylfumarate (1g)
-(benzyloxy)fumarate (1j)
2
2
2
cipitation. Nevertheless, the reaction reached equilibrium (ꢀ65%
conversion) in just a few hours. These results clearly demonstrate
the potential of the Q73A mutant for application in the selective
synthesis of various N-substituted 3-methylaspartic acids.
0
1
2
3
23
24
25
Time (h)
Expansion of the electrophile spectrum of MAL. The structure of
Figure 2 | Enzyme-catalysed transformations under optimized reaction
conditions. a, Progress curves for the Q73A-catalysed (0.05 mol%) addition
1
3
MAL in complex with threo-3a suggests that the side chains of F170,
Y356 and L384 are involved in the formation of the binding pocket
for the substrate’s methyl group (Fig. 1b). To engineer this putative
alkyl binding pocket, three focused libraries were generated by
saturation mutagenesis at positions 170, 356 and 384 using NNS
degeneracy. As the model reaction for activity screening, the
addition of ammonia to 2-hexylfumarate (Table 2, 1g) was chosen.
Compound 1g is not converted at all by wild-type MAL, even after
of methylamine (2b, 3 M), cyclobutylamine (2i, 2 M) or ethoxyamine
1
(
2o, 3 M) to mesaconate (26 mM), as monitored by H NMR spectroscopy.
b, Progress curves for the L384A-catalysed (0.05 mol%) addition
of ammonia (5 M) to fumarate (1a), 2-hexylfumarate (1g) or
2
-(benzyloxy)fumarate (1j) (each at 30 mM), as monitored by
1
H NMR spectroscopy.
(Supplementary Fig. S2). This demonstrates that the Q73A mutation an incubation period of several weeks. The hope was that active
does not affect the regio- or diastereoselectivity of the enzyme. mutants with the ability to aminate electrophiles with large
Whereas the enzyme-catalysed ammonia addition reaction yields an substituents at the C2 position (for example, 1e–1g, 1i, 1j, 1l–1n)
ꢀ
1:1 mixture of threo and erythro isomers of 3-methylaspartate could be identified in this way.
8
(
Table 1, 3a) , the enzyme-catalysed methylamine addition reaction
The three libraries were introduced into E. coli cells, after which
is highly diastereoselective and gives exclusively the threo isomer of the library corresponding to position 384 was screened by evaluating
1
9
N,3-dimethylaspartate (3b) (Table 1, Supplementary Fig. S2) .
ꢀ400 transformants; ꢀ200 transformants were evaluated for each of
1
To investigate its nucleophile scope, we used H NMR spec- the other two libraries. Although the libraries corresponding to pos-
troscopy to determine the ability of the Q73A mutant to add a itions 170 and 356 failed to give any active mutants, screening of the
variety of structurally different amines (2c–2u, Table 1) to mesaco- library corresponding to position 384 resulted in several positive
nate (tenfold molar excess of amine over mesaconate, 0.01 mol% of hits. Analysis by DNA sequencing revealed that all positive hits
biocatalyst, room temperature). These amines are not accepted as carry either an alanine or glycine mutation at position 384. These
substrates by wild-type MAL, with the exception of 2c, 2n, 2p two mutants and the wild-type enzyme were overexpressed, purified
and 2t, which show minimal conversion (,3%) following pro- to homogeneity, and tested for their ability to catalyse the addition
longed (two weeks) incubation. Given that MAL catalyses the fast of ammonia to either mesaconate or 2-hexylfumarate. Gratifyingly,
anti-addition and much slower syn-addition of ammonia to mesa- the L384G and L384A mutants exhibit pronounced amination
8
conate , leading to threo and erythro isomers of the corresponding activity towards 2-hexylfumarate (Supplementary Fig. S5). Both
product, respectively, we allowed the unnatural amine addition reac- mutants display a significant decrease in amination activity
tions to run for seven days to detect the formation of low amounts of towards mesaconate when compared to wild-type MAL. In other
erythro product isomers, if present. Remarkably, the Q73A mutant words, the mutations moved the electrophile specificity of MAL
processed all amines tested (Table 1), including those having bulky away from the natural substrate mesaconate and towards the unna-
substituents such as hexyl, cyclohexyl and benzyl, which tural substrate 2-hexylfumarate. Because the L384A mutant showed
480
NATURE CHEMISTRY | VOL 4 | JUNE 2012 | www.nature.com/naturechemistry
2012 Macmillan Publishers Limited. All rights reserved.
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1338
ARTICLES
Table2 | MAL- and L384A-catalysed ammonia addition to fumarates 1a–1n.
NH2
*
L384A
CO2
CO2
O2C
+
NH3
O2C
*
R
R
1
2a
4
,
†
§
Entry
Electrophile
R-group
Product
Conv. (%)* wild-type
Conv. (%)* L384A
Conv. (%)* L384A
d.e. (%)
e.e. (%)
‡
§§
1
1a
1b
1c
1d
1e
1f
1g
1h
1i
1j
1k
1l
1m
1n
H
4a
4b
4c
4d
4e
4f
4g
4h
4i
4j
4k
4l
4m
4n
100
80
74
57
0
0
0
46
0
0
99
75
76
67
59
52
53
43
44
60
50
34
42
90
99 (80)
–
.99 (S)
‡
ꢁ
ꢁꢁ
2
3
4
5
6
7
8
9
Methyl
Ethyl
Propyl
Butyl
Pentyl
73 (60)
73 (45)
56 (threo)
.95 (threo)
.95 (threo)
ND
‡
ꢁ
ꢁ
ND
ND
ND
ND
ND
ND
ND
‡
66 (46)
‡
}
57 (36)
52 (23)
53 (48)
2
‡
}
.95
‡
}
Hexyl
.95
‡
Ethoxy
Phenoxy
Benzyloxy
Ethylthio
Phenylthio
Benzylthio
Benzyl
45 (24)
70 (threo)**
‡
}
46 (20)
65 (50)
.95
‡
††
}}
1
1
1
1
1
0
1
2
3
4
.95 (threo)
36 (threo)
30 (threo)
20 (threo)
.99
‡
‡‡
36
0
0
81 (22)
45 (13)
55 (30)
ND
ND
ND
ND
‡
‡‡
‡
‡‡
‡
0
90 (55)
.95 (threo)**
†
‡
§
ꢁ
*
Reactions were allowed to proceed for 7 days. Preparative scale. Yield (%) of isolated product after ion exchange chromatography. The d.e. is defined as excess of threo-isomer over erytho-isomer. The relative
configuration of the isolated product(s) was assigned by H NMR using the product(s) of the corresponding wild-type MAL-catalysed reaction as reference sample and by referring to the literature . }The relative
1
17
configurations have not been determined. **The relative configuration of the isolated product(s) was assigned by referring to the literature29,31. The relative configuration of isolated product was assigned by
††
1
H NMR spectroscopy using an authentic standard with known threo configuration and by referring to the literature30
‡‡
. The isolated amino-acid products were first esterified and acetylated, after which the
§§
1
relative configuration of the products could be tentatively assigned by H NMR spectroscopy on the basis of comparison to chemically synthesized reference molecules (Supplementary Methods). The
e.e. value of isolated product was determined by chiral HPLC using authentic standards with known R or S configuration. ND, not determined. The e.e. value of isolated product was determined by
chiral HPLC using authentic threo-(DL)-4j as standard.
ꢁ
ꢁ
}}
higher activity towards a range of fumarate derivatives, it was configuration (Table 2). Analysis of products 4a and 4j by chiral
selected for further study.
HPLC revealed that the L384A-catalysed amination reactions are
Kinetic parameters were determined for the L384A mutant using highly enantioselective, producing these amino acids with .99%
substrates 1a–1g, which have varying 2-alkyl chain length (Table 2), e.e. (Table 2).
and compared to those measured for the wild-type enzyme
(
The preparative usefulness of the L384A mutant was further
Supplementary Table S2). As previously reported1 , wild-type demonstrated by transformations using 0.05 mol% biocatalyst,
7,18
MAL only displays activity towards fumarates with a short alkyl sub- 5 M ammonia and 30 mM of the structurally distinct unsaturated
stituent (1a–1d). Mutant L384A, however, shows activity for all acids 1a, 1g or 1j (167-fold molar excess of ammonia over unsatu-
tested substrates. It displays both higher affinity (lower Km values) rated acid). Under these optimized conditions, the reactions were
and higher activity (larger kcat values) for substrates with a long complete within 10–60 min at 25 8C, achieving final conversions
alkyl chain (1d–1g) than for those with a short alkyl chain of ꢀ85–100% (Fig. 2b). These results clearly demonstrate the poten-
(1a–1c). Although wild-type MAL is at least 100-fold more efficient tial of the L384A mutant for applications in the selective synthesis of
in the amination of 1a–1c, the L384A mutant is ꢀ13-fold more effi- valuable a-amino acids.
cient in the amination of 1d. In the case of the large substrates
1
e–1g, which are not converted at all by wild-type MAL, it was Structural basis for the expanded substrate spectrum. To obtain
not possible to determine the rate enhancement, but it must be insight into how the active sites of the Q73A and L384A mutants
much higher than that observed for 1d. are remodelled to accommodate the new substrates, we
To further explore the electrophile scope of the L384A mutant, determined their X-ray crystal structures at resolutions of 2.0 Å
its ability to catalyse the amination of substrates with aryl, alkoxy, and 1.9 Å, respectively (Supplementary Table S3). A comparison
aryloxy, alkylthio and arylthio substituents at the C2 position of the Q73A and L384A structures with that of wild-type MAL
1
(
1h–1n) was tested using H NMR spectroscopy (Table 2). The revealed no significant changes in overall structure nor in main-
wild-type enzyme only displays amination activity for compounds chain and side-chain conformations of the active-site residues
h and 1k, which, like substrates 1a–1d, have small R-groups at (Fig. 3). These results demonstrate that the expanded substrate
1
the C2 position (Table 2, entries 1–4, 8, 11). Strikingly, the L384A spectrum of the Q73A and L384A mutants is not due to major
mutant exhibits activity towards all substrates tested (Table 2), structural changes but is related to the replacement of a large
demonstrating its broad electrophile scope. Control experiments active site residue by a smaller one, resulting in an enlarged active
showed that the fumarate derivatives do not undergo amination in site that can accommodate the new substrates (Fig. 3).
the absence of enzyme.
To further demonstrate that the active sites of the Q73A and
The L384A-catalysed aminations of 1a–1n were also performed L384A mutants can accommodate larger substrates than the active
on a preparative (3.1 mmol) scale (Table 2). Reactions (tenfold site of wild-type MAL, we performed comparative docking simu-
molar excess of ammonia over unsaturated acid, 0.01 mol% of bio- lations. Figure 4a shows an overlay of the docking model of wild-
catalyst, room temperature) were allowed to run for seven days to type MAL in complex with the natural substrate threo-3a and the
detect the formation of low amounts of erythro product isomers, docking model of mutant Q73A complexed with the unnatural sub-
if present. The products were isolated and identified as aspartate strate threo-(2S,3S)-2-(cyclohexylamino)-3-methylbutanedioic acid
(4a) and its 3-substituted derivatives 4b–4n, respectively. While (3k), respectively. This comparison clearly shows that compound
these L384A-catalysed amination reactions provided high regio- 3k does not fit into the active site of wild-type MAL because
selectivities, varying degrees of diastereoselectivity were obtained, residue Q73, instead of the shorter A73, partially fills the volume
with the single or predominant diastereoisomer having the threo that is occupied by the cyclohexylamino moiety of substrate 3k in
NATURE CHEMISTRY | VOL 4 | JUNE 2012 | www.nature.com/naturechemistry
2012 Macmillan Publishers Limited. All rights reserved.
481
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1338
ARTICLES
a
c
b
Ala384
Leu384
Leu384
Ala73
Gln73
Gln73
Figure 3 | Structural comparison of wild-type MAL and the Gln73Ala and Leu384Ala mutants. a, Overlay of the structures of wild-type MAL (cyan
ribbons) and the Gln73Ala (green ribbons) and Leu384Ala (orange ribbons) mutants. The structures of the Gln73Ala (PDB code 3ZVH) and Leu384Ala
(
PDB code 3ZVI) mutants can be superimposed on the wild-type MAL structure12 (PDB code 1KCZ) with a root-mean-square deviation (RMSD) of
0.53–0.74 Å, considering all atoms in the two subunits that occupy the different crystallographic asymmetric units. When considering only the atoms of
residues that are within 5 Å of the residues at positions 73 and 384, the RMSD is 0.19–0.50 Å. b, Overlay of the active sites of wild-type MAL and the
Gln73Ala mutant. Colour scheme as in a. c, Overlay of the active sites of wild-type MAL and the Leu384Ala mutant. Colour scheme as in a. In all three
figures, the residues at positions 73 and 384 are shown as sticks. For clarity, these sticks are shown in grey for wild-type MAL. The figures were prepared
with Discovery Studio 2.5.
a
b
Lys331
Gln329
Lys331
Ala73
Gln329
Mg2+
Mg2+
Ala384
3.2 Å
3.2 Å
Leu384
Leu384
Gln73
Gln73
His194
His194
Figure 4 | Molecular docking of substrates in the active sites of wild-type MAL and the two MAL mutants. a, Overlay of the most favourable docked
conformations of threo-3a and 3k in the active sites of wild-type MAL (grey) and the Gln73Ala mutant (green), respectively. Residue Gln73 in wild-type MAL
is represented by a solvent-accessible surface calculated with a probe radius of 1.4 Å. b, Overlay of the most favourable docked conformations of threo-3a
and 4f in the active sites of wild-type MAL (grey) and the Leu384Ala mutant (orange), respectively. Residue Leu384 in wild-type MAL is represented by a
2þ
solvent-accessible surface calculated with a probe radius of 1.4 Å. In both figures, interactions between active-site residues or the Mg ion and substrate
threo-3a in wild-type MAL are displayed as orange dashed lines. The distance between the substrate’s C3 atom and the side chain of Lys331 is shown in
a˚ ngstroms (atoms connected by a solid line). The figures were prepared with Discovery Studio 2.5.
the Q73A–substrate complex. The observation that residue Q73 benefit of this structure-guided engineering strategy lies in the
occludes the amine binding pocket thus explains why cyclohexyla- reduction of the sequence space that has to be covered to identify
2
0–24
mine 2k and other large amines are not accepted as substrates by valuable mutants
.
wild-type MAL. An overlay of the docking model of wild-type Engineering of the amine binding pocket (Fig. 1a) yielded a
MAL complexed with threo-3a and that of mutant L384A in mutant (Q73A) that has a very broad nucleophile scope and excel-
complex with the unnatural substrate threo-(2S,3S)-3-pentylaspartic lent regio- and stereoselectivity, producing the threo isomers of the
acid (4f) is shown in Fig. 4b. This comparison shows that the N-substituted 3-methylaspartic acids 3b–3u with very high diastereo-
mutation of residue L384 to the shorter A384 prevents a clash meric and enantiomeric excess (Table 1). These results demonstrate
between the substrate’s pentyl group and the large leucine residue that the high regio- and stereoselectivity of the Q73A-catalysed
found at position 384 in wild-type MAL, thus rationalizing the addition reactions is not influenced by the different substituents
ability of the L384A mutant to convert large electrophiles.
on the amine substrate. This is of high synthetic significance,
especially when compared to the Q73A-catalysed ammonia
addition to mesaconate (the natural reaction), which shows low dia-
Discussion
Wild-type MAL accepts only a few amines as alternative stereoselectivity (Table 1, entry 1). Recently, a mechanism was pro-
8
,12,13
nucleophiles for ammonia and a few fumarate derivatives as posed for MAL
in which the proton abstraction/addition steps
alternative electrophiles for mesaconate17–19. Therefore, the goal of are effected by K331 and H194 as the (S)- and (R)-specific general
this study was to expand both the nucleophile and electrophile base/acid catalysts, respectively, which are juxtaposed on either side
scope of MAL. This goal was achieved by using an engineering of the chiral C3 carbon of the amino-acid substrate. In view of this
strategy that involves randomization at carefully chosen sites mechanism, the high diastereoselectivity of the Q73A-catalysed
lining the substrate binding pocket of the enzyme. The major amine addition reactions could be explained by assuming that the
482
NATURE CHEMISTRY | VOL 4 | JUNE 2012 | www.nature.com/naturechemistry
2012 Macmillan Publishers Limited. All rights reserved.
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1338
ARTICLES
substituents on the amine substrate sterically hinder protonation of and 50 ml of D
O. In the case of amines 2f, 2g and 2m, 500 ml of a 250 (2f) or
25 mM (2g and 2m) stock solution was added to the reaction mixtures to avoid
2
1
the C3 carbon (numbering according to the amino-acid product) of
mesaconate by H194, thereby preventing the formation of erythro
isomers. Structural analysis of the Q73A mutant showed that this
protein precipitation. Reactions were started by the addition of 200 mg of freshly
purified enzyme, and reaction mixtures were incubated at 22 8C. H NMR spectra
were recorded 3 h, 3 days and 7 days after the addition of enzyme. Product amounts
1
mutant enzyme has an enlarged amine binding pocket, without were estimated by integration of the respective product signals, if present.
changes in the orientation of active-site residues, thus rationalizing
its ability to convert the new amine substrates.
Engineering of the alkyl binding pocket (Fig. 1b) yielded a electrophiles (1a–1n) was examined using H NMR spectroscopy. Reaction mixtures
mutant (L384A) that has a very broad electrophile scope and excel- consisted of 500 ml of a 1 M stock solution of NH
lent regio- and enantioselectivity in the amination reaction, produ-
cing (S)-aspartic acid (4a) and its 3-substituted derivatives 4b–4n
General procedure for monitoring the amination of various electrophiles. The
ability of wild-type MAL and the L384A mutant to add ammonia to different
1
Cl in water (pH 9.0, containing
4
2
5
0 mM MgCl ), 50 ml D O, and 200 ml of a 250 mM stock solution (or 100 ml of a
00 mM stock solution) of electrophile in water (pH 9.0, containing 20 mM MgCl2).
2
2
Reactions were started by the addition of 200 mg of freshly purified enzyme, and
(Table 2). The structure of L384A showed that this mutant
1
reaction mixtures were incubated at 22 8C. H NMR spectra were recorded 3 h,
enzyme has an enlarged alkyl binding pocket, with an otherwise 3 days and 7 days after addition of enzyme. Product amounts were estimated by
unchanged active site geometry, thus rationalizing its ability to integration of the respective product signals, if present.
convert fumarate derivatives with large substituents. In contrast to
the Q73A-catalysed amine additions, the L384A-catalysed amin-
Starting substrates. Amines 2a–2u and unsaturated acids 1a and 1b were purchased
from Sigma-Aldrich or Merck. Unsaturated acids 1c–1n were prepared according to
ation of the fumarate derivatives gives varying degrees of diastereo- experimental procedures described in the Supplementary Methods.
selectivity. This irregular pattern of diastereoselectivities is an
interesting phenomenon and suggests that some of the fumarate
Identification of the amino-acid products. Procedures for the enzymatic synthesis
and purification of amino acids 3n–3u and 4a–4n are described in the
derivatives, like the natural substrate mesaconate, are optimally Supplementary Methods. The relative configuration of products 4b–4d, 4h, 4j and
1
positioned to undergo protonation by both K331 and H194 from 4n was determined by comparison of their H NMR signals to those of commercially
1
7
available or wild-type MAL-synthesized authentic standards with known
either face of the substrate, producing the undesired mixture of
threo and erythro isomers, whereas others are bound in a position
that does not favour protonation by H194, yielding only the
2
9–31
configuration, and/or by referring to the literature
purchased from Tocris Bioscience. The procedures to establish the relative
configuration of products 4k–4m are provided in the Supplementary Methods. The
. Compound threo-DL-4j was
desired threo isomers. Nonetheless, the synthetic potential of the e.e. of products 4a and 4j was determined by chiral HPLC using a Chirex 3126-(D)-
penicillamine column (250 mm × 4.6 mm, Phenomenex) or a Nucleosil chiral-1
column (250 mm × 4 mm, Macherey-Nagel), respectively. Authentic standards
L-4a and D-4a were purchased from Sigma-Aldrich. HPLC conditions and observed
retention times are given in the Supplementary Methods.
L384A mutant is convincingly demonstrated by its use as catalyst
in the stereoselective synthesis (diastereomeric excess (d.e.) values
are .95%) of threo-3-benzyloxyaspartate (4j) and threo-3-benzyl-
aspartate (4n), which are important compounds in studying the
role of glutamate transporters in the brain26–28
Identity and configuration of amino-acid products 3a–3m. The identity and
.
1
relative configuration of products 3a–3m were determined by comparative H NMR
In summary, we provide evidence that a single amino acid sub-
stitution is sufficient to dramatically expand the nucleophile or elec-
analysis of the crude reaction mixtures and authentic compounds with known
relative configuration. The experimental procedures for the synthesis of these
trophile spectrum of MAL. The results provide support for the authentic standards are provided in the Supplementary Methods. Products 3c–3e,
notion that structure-guided saturation mutagenesis of sites lining 3h, 3i and 3l were also purified from preparative scale (0.15 g 1b) reactions and their
e.e. was determined by chiral HPLC. HPLC conditions, procedures for the synthesis
of the threo-DL standards, and the observed retention times are provided in the
Supplementary Methods.
the substrate binding pocket is a powerful strategy to create
enzymes that can accommodate non-native substrates and
22
perform reactions not observed in nature . The new designer
enzymes catalyse attractive and straightforward asymmetric trans-
formations in water, making use of readily available starting sub-
strates, without the need for organic (co)solvents or the
Optimization of reaction conditions for amino acid synthesis. By increasing both
the amount of biocatalyst and the molar excess of amine over mesaconate, excellent
conversions were achieved for several Q73A-catalysed amine additions within short
reaction times. The optimized reaction conditions were as follows. Reaction mixtures
implementation of dynamic kinetic resolution strategies. These are (30 ml) consisted of 3 M amine (either 2b or 2o) and 26 mM 1b in 20 mM MgCl2
very desirable features in view of the quest for environmentally (final pH 9). In the case of amine 2i, a lower concentration (2 M) was used to avoid
protein precipitation. The reactions were started by the addition of freshly purified
benign and atom efficient processes. The potential of the two engin-
enzyme (0.05 mol%), and reaction mixtures were incubated at 25 8C. Aliquots of
eered MALs for application in the selective synthesis of new and
each reaction mixture (5 ml) were withdrawn 5 min, 15 min, 45 min, 90 min,
valuable aspartic acid derivatives has been demonstrated, given
180 min and 24 h after the addition of enzyme. For each sample, the reaction was
that the product yields may be further improved by optimizing stopped by incubating the reaction mixture at 100 8C for 5 min.
the reaction conditions and purification protocols. The Q73 and
L384 residues represent good targets for future combinatorial muta-
genesis experiments (Supplementary Results and Discussion),
potentially yielding new biocatalysts with an even larger
substrate scope.
Optimized conditions for several L348A-catalysed amination reactions were as
follows. Reaction mixtures (6–12 ml) consisted of 5 M NH Cl and 30 mM 1a, 1g or
4
1
j in 20 mM MgCl (final pH 9). The reactions were started by the addition of
2
freshly purified enzyme (0.05 mol%), and reaction mixtures were incubated at 25 8C.
Aliquots (1–2 ml) of the reaction mixtures were withdrawn 5 min, 15 min, 45 min,
90 min, 180 min and 24 h after the addition of enzyme. For each sample, the
reaction was stopped by incubating the reaction mixture at 100 8C for 5 min.
Each sample was dried in vacuo, followed by resuspending the resulting residue
in ꢀ10 ml of water for lyophilization. This process was repeated 5 to 10 times to
Methods
Mutant library preparation and screening. Site-saturation mutagenesis libraries
1
33
remove the excess of amine/ammonia to enable analysis by H NMR spectroscopy.
were generated by overlap extension polymerase chain reaction using plasmid
8
The final residue was dissolved in 800 ml D O. The conversion of substrate to
pBAD(MAL-His) as the template and the primers listed in Supplementary
2
product was estimated by integration of the respective substrate and product signals.
Table S4. Detailed procedures for library preparation and screening are provided in
the Supplementary Methods.
Crystallization, structure determination and molecular docking. The procedures
used for crystallization, structure determination and comparative docking
simulations are described in the Supplementary Methods.
Enzyme expression, purification and kinetic characterization. Wild-type MAL
and MAL mutants were overproduced in E. coli TOP10 cells and purified to
8
homogeneity as described before . Procedures for the kinetic characterization of
Received 16 January 2012; accepted 21 March 2012;
published online 29 April 2012
MAL (wild-type or mutants) are provided in the Supplementary Methods.
General procedure for monitoring the addition of various amines to mesaconate.
The ability of wild-type MAL and the Q73A mutant to add different amines (2a–2u) References
1
to mesaconate (1b) was examined using H NMR spectroscopy. Reaction mixtures
consisted of 500 ml of a 1 M stock solution of amine in water (pH 9.0, containing
1. Sonke, T., Kaptein, B. & Schoemaker, H. E. Use of enzymes in the synthesis of
amino acids, in Amino Acids, Peptides and Proteins in Organic Chemistry Vol. 1
2
0 mM MgCl ), 100 ml of 500 mM 1b in water (pH 9.0, containing 20 mM MgCl )
(ed. Hughes, A. B.) 77–117 (Wiley-VCH, 2009).
2
2
NATURE CHEMISTRY | VOL 4 | JUNE 2012 | www.nature.com/naturechemistry
2012 Macmillan Publishers Limited. All rights reserved.
483
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1338
ARTICLES
2.
3.
4.
Wohlgemuth, R. Biocatalysis—key to sustainable industrial chemistry. Curr.
Opin. Biotechnol. 21, 713–724 (2010).
Panke, S. & Wubbolts, M. Advances in biocatalytic synthesis of pharmaceutical
intermediates. Curr. Opin. Chem. Biol. 9, 188–194 (2005).
Panke, S., Held, M. & Wubbolts, M. Trends and innovations in industrial
biocatalysis for the production of fine chemicals. Curr. Opin. Biotechnol. 15,
23. Arnold, F. H. & Georgiou, G. (eds) Methods in Molecular Biology (Directed
Enzyme Evolution) Vol. 230 (Humana Press, 2003).
24. Turner, N. J. Directed evolution drives the next generation of biocatalysts. Nature
Chem. Biol. 5, 567–573 (2009).
25. Shimamoto, K. et al. Characterization of novel L-threo-b-benzyloxyaspartate
derivatives, potent blockers of the glutamate transporters. Mol. Pharmacol. 65,
1008–1015 (2004).
272–279 (2004).
5.
6.
7.
Turner, N. J. Ammonia lyases and aminomutases as biocatalysts for the synthesis
of a-amino and b-amino acids. Curr. Opin. Chem. Biol. 15, 234–240 (2011).
Barker, H. A., Smyth, R. D., Wilson, R. M. & Weissbach, H. The purification and
properties of b-methylaspartase. J. Biol. Chem. 234, 320–328 (1959).
Goda, S. K., Minton, N. P., Botting, N. P. & Gani, D. Cloning, sequencing, and
expression in Escherichia coli of the Clostridium tetanomorphum gene encoding
b-methylaspartase and characterization of the recombinant protein.
Biochemistry 31, 10747–10756 (1992).
26. Shimamoto, K. Glutamate transporter blockers for elucidation of the function of
excitatory neurotransmission systems. Chem. Rec. 8, 182–199 (2008).
27. Esslinger, C. S. et al. The substituted aspartate analogue L-b-threo-benzyl-
aspartate preferentially inhibits the neuronal excitatory amino acid transporter
EAAT3. Neuropharmacology 49, 850–861 (2005).
28. Bridges, R. J. & Esslinger, C. S. The excitatory amino acid transporters:
pharmacological insights on substrate and inhibitor specificity of the EAAT
subtypes. Pharmacol. Ther. 107, 271–285 (2005).
8
.
Raj, H. et al. Alteration of the diastereoselectivity of 3-methylaspartate
ammonia lyase by using structure-based mutagenesis. ChemBioChem 10,
29. Mavencamp, T. L., Rhoderick, J. F., Bridges, R. J. & Esslinger, C. S. Synthesis and
preliminary pharmacological evaluation of novel derivatives of L-b-threo-
benzylaspartate as inhibitors of the neuronal glutamate transporter EAAT3.
Bioorg. Med. Chem. 16, 7740–7748 (2008).
30. Shimamoto, K. et al. Syntheses of optically pure b-hydroxyaspartate derivatives
as glutamate transporter blockers. Bioorg. Med. Chem. Lett. 10,
2407–2410 (2000).
31. Spengler, J., Pelay, M., Tulla-Puche, J. & Albericio, F. Synthesis of orthogonally
protected L-threo-b-ethoxyasparagine. Amino Acids 39, 161–165 (2010).
32. LaBean, T. H. & Kauffman, S. A. Design of synthetic gene libraries encoding
random sequence proteins with desired ensemble characteristics. Protein Sci. 2,
1249–1254 (1993).
33. Ho, S. N., Hunt, H. D., Horton, R. M., Pullen, J. K. & Pease, L. R. Site-directed
mutagenesis by overlap extension using the polymerase chain reaction. Gene 77,
51–59 (1989).
2236–2245 (2009).
9
1
1
.
Bright, H. J. Divalent metal activation of b-methylaspartase. The importance of
ionic radius. Biochemistry 6, 1191–1203 (1967).
0. Bright, H. J. On the mechanism of divalent metal activation of
b-methylaspartase. J. Biol. Chem. 240, 1198–1210 (1965).
1. Kato, Y. & Asano, Y. 3-Methylaspartate ammonia-lyase as a marker enzyme of
the mesaconate pathway for (S)-glutamate fermentation in Enterobacteriaceae.
Arch. Microbiol. 168, 457–463 (1997).
1
2. Asuncion, M., Blankenfeldt, W., Barlow, J. N., Gani, D. & Naismith, J. H. The
structure of 3-methylaspartase from Clostridium tetanomorphum functions via
the common enolase chemical step. J. Biol. Chem. 277, 8306–8311 (2002).
3. Levy, C. W. et al. Insights into enzyme evolution revealed by the structure of
methylaspartate ammonia lyase. Structure 10, 105–113 (2002).
4. Bright, H. J., Ingraham, L. L. & Lundin, R. E. The mechanism of the
methylaspartate ammonia–lyase reaction: deuterium exchange. Biochim.
Biophys. Acta 81, 576–584 (1964).
1
1
Acknowledgements
The authors thank B. Wu, B. Kaptein and O. May for helpful discussions. This research was
financially supported by VENI grant 700.54.401 and ECHO grant 700.59.042 (both to
G.J.P.) from the Division of Chemical Sciences of the Netherlands Organisation of
Scientific Research (NWO-CW), and by the Netherlands Ministry of Economic Affairs and
the B-Basic partner organizations (www.b-basic.nl) through B-Basic, a public-private
NWO–ACTS programme (ACTS, Advanced Chemical Technologies for Sustainability).
Financial support from the Royal Netherlands Academy of Arts and Sciences (KNAW to
W.S. and B.L.F.) is also gratefully acknowledged.
15. Bright, H. The mechanism of the b-methylaspartase reaction. J. Biol. Chem. 239,
2307–2315 (1964).
16. Babbitt, P. C. et al. The enolase superfamily: a general strategy for enzyme-
catalyzed abstraction of the a-protons of carboxylic acids. Biochemistry 35,
1
6489–16501 (1996).
7. Akhtar, M., Botting, N. P., Cohen, M. A. & Gani, D. Enantiospecific synthesis of
-substituted aspartic acids via enzymic amination of substituted fumaric acids.
1
1
1
3
Tetrahedron 43, 5899–5908 (1987).
8. Botting, N. P., Akhtar, M., Cohen, M. A. & Gani, D. Substrate specificity of the
Author contributions
3-methylaspartate ammonia-lyase reaction: observation of differential relative
H.R. performed mutagenesis, library screening and activity measurements. W.S. and J.d.V.
synthesized starting substrates and reference compounds. H.R., W.S., M.d.V. and F.J.D.
performed preparative biocatalysis. H.J.R. performed X-ray crystallography experiments.
V.P.V. performed chiral HPLC. C.R.R. performed the molecular docking experiments.
S.d.W., W.J.Q., A.M.W.H.T., B.L.F., D.B.J. and G.J.P. supervised scientific work. All authors
contributed to writing the paper.
reaction rates for substrate-product pairs. Biochemistry 27, 2953–2955 (1988).
9. Gulzar, M. S., Akhtar, M. & Gani, D. Preparation of N-substituted aspartic acids
via enantiospecific conjugate addition of N-nucleophiles to fumaric acids using
methylaspartase: synthetic utility and mechanistic implications. J. Chem. Soc.
Perkin Trans. 1, 649–656 (1997).
2
0. Lutz, S. & Bornscheuer, U. T. Protein Engineering Handbook. Vols 1–2
(
Wiley-VCH, 2009).
2
1. Reetz, M. T., Bocola, M., Carballeira, J. D., Zha, D. & Vogel, A. Expanding the Additional information
range of substrate acceptance of enzymes: combinatorial active-site saturation
test. Angew. Chem. Int. Ed. 44, 4192–4196 (2005).
The authors declare no competing financial interests. Supplementary information and
chemical compound information accompany this paper at www.nature.com/
naturechemistry. Reprints and permission information is available online at http://www.
nature.com/reprints. Correspondence and requests for materials should be addressed to G.J.P.
22. Morley, K. L. & Kazlauskas, R. J. Improving enzyme properties: when are closer
mutations better? Trends Biotechnol. 23, 231–237 (2005).
484
NATURE CHEMISTRY | VOL 4 | JUNE 2012 | www.nature.com/naturechemistry
©
2012 Macmillan Publishers Limited. All rights reserved.