Journal of the American Chemical Society
Communication
As well as Amycolatopsis species, NAAAR activity has also
been detected in Streptomyces atratus Y-53 and in several
thermophiles.9−14 In these bacteria, NAAAR is believed to act
as a bifunctional enzyme, as an o-succinyl benzoate synthase
(OSBS) that catalyzes the formation of o-succinylbenzoate (a
precursor of menaquinone) from 2-hydroxy-6-succinyl-2,4-
cyclohexadiene carboxylate (Scheme 1c)15 as well as an N-
succinyl-amino acid racemase (NSAAR) involved in D-amino
acid detoxification (Scheme 1d).16 NAAAR has been well
studied as a member of the enolase superfamily, and
mechanistic analysis has shown that the NAAAR reaction
proceeds via a divalent cation-dependent, enolate-based proton
transfer catalyzed by two lysine residues.17 The racemization of
N-succinyl amino acids (and the OSBS reaction) is known to
be ∼20-fold faster than with the equivalent N-acetylated
substrates. As the present acylase process for the production of
enantiopure α-amino acids requires N-acetylated substrates,
high activity toward these derivatives is essential. Our aim was
to isolate a NAAAR variant with sufficiently high activity, such
that the current acylase KR process could be run as an efficient,
coupled NAAAR/acylase DKR. We achieved this by NAAAR
gene mutagenesis followed by genetic selection in an
engineered Escherichia coli host.
amino acids to act as a source of useful L-amino acids. As a
result of these deletions, when grown in minimal media
containing N-acetyl-D-methionine as the sole methionine
source, the selection host is now dependent on recombinant
NAAAR activity to racemise this to N-acetyl-L-methionine. An
endogenous hydrolase then provides the source of L-
methionine from the N-acetylated substrate. Using the λ red-
mediated gene replacement/deletion method, dadA was
deleted, and metB replaced with a chloramphenicol resistance
gene.18 This double-knock out host, named SET21, is now
completely unable to synthesize L-methionine or racemise D-
methionine. Importantly, the SET21 host is viable on N-acetyl-
L-methionine allowing for a NAAAR-dependent screening
method shown in Scheme 2. The ability of NAAAR to
complement the SET21 strain and produce viable colonies was
confirmed by transformation of SET21 with a plasmid
expressing NAAAR wild-type (WT) and selection on minimal
media supplemented with N-acetyl-D-methionine as the sole
methionine source (see Supporting Information, SI).
Using the SET21 selection host, mutagenic libraries of the
NAAAR gene were screened, and the best variant used as the
template for the next round of mutagenesis/selection. Variants
were generated using XL1-Red (>107 variants), error prone
PCR (>105 variants), and saturation mutagenesis (>102
variants). Active variants were selected by SET21 colony size
when grown on minimal media supplemented with N-acetyl-D-
methionine or by high-throughput HPLC screening (see SI).
Selection of libraries generated by XL1-Red mutagenesis led to
the isolation of the G291E mutation, which was subsequently
improved upon by saturation mutagenesis to generate G291D.
Error prone PCR on this clone generated NAAAR G291D
F323Y, which could not be improved upon after saturation
mutagenesis at residue 323 and a further round of error prone
PCR. Saturation mutagenesis of other residues (D316 and
S135) within the active site of NAAAR G291D F323Y failed to
generate improved variants. The NAAAR activity of each
variant was determined by expression and purification of
recombinant protein followed by kinetic analysis with a range of
amino acids. The rate of racemisation was monitored with
chiral HPLC, and the activities of the best variant from each
round of selection are shown in Table 1. As well as testing both
enantiomers of methionine (used in the plate selection assay),
we also chose two synthetically useful amino acid precursors to
probe the versatility of the new variant: D-(4-fluoro)phenyl
glycine and D-allylglycine since they are used in the preparation
of various drugs. Gratifyingly, the NAAAR G291D F323Y
variant displayed the most improved activity with each of the
substrates tested, showing up to 6-fold improved kcat values
over WT. The kinetic data would suggest that the mutations are
directly affecting turnover and not substrate binding, as no large
changes in Km values were observed with each amino acid.
To measure the improvement in the DKR of N-acetyl-DL-amino
acids with these variant NAAARs, small scale resolutions were
performed using whole cell biocatalysts expressing an L-acylase
and one of NAAAR WT, G291E, G291D, or G291D F323Y.
Combinations of the acylase and NAAAR hosts were mixed,
and the resolution of N-acetyl-DL-methionine to L-methionine
monitored over 5 h by chiral HPLC (Figure 1). As expected,
with no NAAAR present, N-acetyl-D-methionine showed no
conversion to free L-methionine (data not shown). When
BL21(DE3) cells expressing NAAAR WT were used, N-acetyl-
D-methionine was now a substrate, via racemisation, for the L-
specific acylase, however the rate of racemization (krac) was
To efficiently screen mutagenic Amycolatopsis sp. Ts-1−60
NAAAR libraries, a selection method was required that linked
the rate of racemisation of an N-acetyl amino acid to the
viability of an E. coli host. This has been achieved by disabling
the natural E. coli L-methionine biosynthetic pathway, and at the
same time removing a D-amino acid racemization pathway
(Scheme 2). Conversion of L-homoserine to L-methionine was
Scheme 2. Disabled L-Methionine Biosynthesis in E. coli
SET21 Host and NAAAR-Dependent Growth on N-Acetyl-D-
methionine
abolished by deletion of the second enzyme in the L-
methionine biosynthetic pathway, cystathionine γ-synthase
(metB), while D-amino acid racemization was prevented by
deletion of a nonspecific D-amino acid dehydrogenase (dadA).
DadA combines with an aminotransferase to racemise a variety
of amino acids via their prochiral keto acid allowing toxic D-
19311
dx.doi.org/10.1021/ja305438y | J. Am. Chem. Soc. 2012, 134, 19310−19313