The bacterial ammonia lyase EncP: A tunable biocatalyst for the synthesis of unnatural amino acids

Article abstract of DOI:10.1021/jacs.5b07326

Enzymes of the class I lyase-like family catalyze the asymmetric addition of ammonia to arylacrylates, yielding high value amino acids as products. Recent examples include the use of phenylalanine ammonia lyases (PALs), either alone or as a gateway to deracemization cascades (giving (S)- or (R)-α-phenylalanine derivatives, respectively), and also eukaryotic phenylalanine aminomutases (PAMs) for the synthesis of the (R)-β-products. Herein, we present the investigation of another family member, EncP from Streptomyces maritimus, thereby expanding the biocatalytic toolbox and enabling the production of the missing (S)-β-isomer. EncP was found to convert a range of arylacrylates to a mixture of (S)-α- and (S)-β-arylalanines, with regioselectivity correlating to the strength of electron-withdrawing/-donating groups on the ring of each substrate. The low regioselectivity of the wild-type enzyme was addressed via structure-based rational design to generate three variants with altered preference for either α- or β-products. By examining various biocatalyst/substrate combinations, it was demonstrated that the amination pattern of the reaction could be tuned to achieve selectivities between 99:1 and 1:99 for β:α-product ratios as desired.

Full text of DOI:10.1021/jacs.5b07326

Subscriber access provided by RUTGERS UNIVERSITY
Article
The Bacterial Ammonia Lyase EncP: A Tunable
Biocatalyst for the Synthesis of Unnatural Amino Acids
Nicholas J. Weise, Fabio Parmeggiani, Syed T. Ahmed, and Nicholas J Turner
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.5b07326 • Publication Date (Web): 21 Sep 2015
Downloaded from http://pubs.acs.org on September 22, 2015
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of the American Chemical Society is published by the American Chemical
Society. 1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 8
Journal of the American Chemical Society
1
2
3
4
5
6
The Bacterial Ammonia Lyase EncP: A Tunable Biocatalyst for the
Synthesis of Unnatural Amino Acids
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Nicholas J. Weise, Fabio Parmeggiani, Syed T. Ahmed and Nicholas J. Turner*
Manchester Institute of Biotechnology & School of Chemistry, University of Manchester, 131 Princess Street, M1
7DN, Manchester, UK
KEYWORDS Unnatural Amino Acids, Biocatalysis, Enzyme Engineering, Asymmetric Synthesis
ABSTRACT: Enzymes of the class I lyase-like family catalyze the asymmetric addition of ammonia to arylacrylates, yield-
ing high value amino acids as products. Recent examples include the use of phenylalanine ammonia lyases (PALs), either
alone or as a gateway to deracemization cascades (giving (S)- or (R)-α-phenylalanine derivatives respectively), and also
eukaryotic phenylalanine aminomutases (PAMs) for the synthesis of the (R)-β-products. Herein we present the investiga-
tion of another family member, EncP from Streptomyces maritimus, thereby expanding the biocatalytic toolbox and ena-
bling the production of the missing (S)-β-isomer. EncP was found to convert a range of arylacrylates to a mixture of (S)-α-
and (S)-β-arylalanines, with regioselectivity correlating to the strength of electron-withdrawing/-donating groups on the
ring of each substrate. The low regioselectivity of the wild-type enzyme was addressed via structure-based rational design
to generate three variants with altered preference for either α- or β-products. By examining various biocatalyst/substrate
combinations it was demonstrated that the amination pattern of the reaction could be tuned to achieve selectivities be-
tween 99:1 and 1:99 for β:α-product ratios as desired.
Unnatural or non-proteinogenic amino acids are an im-
portant class of compound in medicinal chemistry as evi-
denced by their presence in numerous natural products
family members for development as biocatalysts. Use of
PALs include both small and industrial scale (S)-selective
α-amination of halocinnamates using enzymes from Pe-
1,2
12
13
with potent bioactivities. As such, amino acids are used
troselinum crispum and Rhodotorula glutinis respective-
ly (Figure 1a). Recently we reported an enantiocomple-
mentary route to the (R)-α-arylalanines which exploits
the imperfect enantiopreference of AvPAL (from Anabae-
na variabilis) in combination with an L-amino acid deam-
inase and chemical reduction as a deracemization step
as chiral building blocks for the synthesis of both phar-
3–5
maceuticals and lead compounds in drug discovery.
There are also promising efforts to incorporate unnatural
amino acids into proteins and peptide mimics as tools in
6,7
chemical biology or to improve their therapeutic prop-
8,9
14
erties. Due to the prevalence of amino acid metaboliz-
(Figure 1b). Work by Janssen et al. has focused on the
ing enzymes in nature, biocatalysis represents an attrac-
tive strategy to develop greener and more generally appli-
cable routes to these compounds.
characterization and engineering of an (R)-selective ami-
The class I lyase-like family is a group of structurally
and mechanistically related enzymes, which, in nature,
catalyze the deamination of aromatic amino acids to yield
the corresponding arylacrylic acids. Enzymes which pro-
ceed no further than this are known as ammonia lyases
(
ALs) and can be histidine (HAL - EC 4.3.1.3), phenylala-
nine (PAL - EC 4.3.1.24/25) or tyrosine (TAL - EC
.3.1.23/25) specific. Other family members also allow
4
subsequent reamination at the β-position (aminomutases
or AMs), thus catalysing overall α- to β-isomerization of
phenylalanine (PAM - EC 5.4.3.10/11) or tyrosine (TAM -
10,11
Figure 1. The complete PAL/PAM toolbox for production
of either enantiomer of α-phenylalanine derivatives (a,
b) and β-phenylalanine derivatives (c, d) from the cor-
responding acrylic acids.
EC 5.4.3.6).
PALs and PAMs have been shown to accept many ring-
substituted cinnamate derivatives as substrates for amina-
tion reactions, making them the most broadly applicable
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 8
1
4
nomutase from Taxus wallichiana var. chinensis to access
autoinduction on an 800 mL scale , yielding wet whole
cells. Protein over-production was confirmed via poly-
acrylamide gel electrophoresis of denatured protein frac-
tions following nickel affinity chromatography. Initial
testing of the whole cell catalyst under previously report-
1
5–
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
enantiopure (R)-β-phenylalanine derivatives (Figure 1c).
1
7
This leaves only the (S)-β-aromatic amino acids which
cannot be produced via direct amination with current
methods (Figure 1d).
1
4
15
The enzyme EncP has been demonstrated to be a PAL in
the thermotolerant bacterium Streptomyces maritimus,
siphoning the primary metabolite phenylalanine into nat-
ural product biosynthetic pathways for enterocin and
ed conditions for AvPAL and Taxus PAM gave no de-
tectable activity with cinnamate 1a as a substrate - ascer-
tained via HPLC on a non-chiral phase in comparison
with authentic standards. After sampling different am-
monium salts and changing buffer pH and incubation
temperature (Tables S1-S4 in the Supporting Infor-
mation), a method was developed by which ~80% conver-
sion of 1 mM cinnamate could be observed after a 22 hour
reaction period. As such, the following method was ap-
1
8,19
wailupemycin antibiotics.
More recent studies of this
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
enzyme, along with the closely related AdmH from Pan-
toea agglomerans (63% sequence identity), revealed both
enzymes to be thermobifunctional (displaying a de-
20
creased lyase:mutase ratio with increasing temperature).
-
1
Despite high sequence similarity, AdmH has been re-
vealed to be an aminomutase physiologically, catalyzing
the interconversion of (S)-α- to (S)-β-phenylalanine as a
plied to all subsequent analyses: 40 mg mL whole cells in
o
4 M (NH ) SO (pH 8.3), 55 C, 550 rpm, 22 hours.
4
2
4
Initially a panel of cinnamate derivatives 1a-v was sub-
jected to biotransformations with EncP-overproducing E.
coli BL21(DE3) cells. Formation of the corresponding ami-
no acids 2-3a-v was monitored by HPLC (Figure 3) reveal-
ing that the enzyme was able to convert all 22 substrates
tested. Interestingly conversions tended to follow the pat-
tern ortho- > meta- > para-, with most ortho-substituted
cinnamates showing conversion greater than or equal to
the unsubstituted ‘natural’ substrate 1a. There was also
little disparity between the conversions achieved with all
the ortho-substituents regardless of size, with para- and
to a lesser extent meta-substituted substrates displaying
decreased conversions with increasing substituent bulki-
ness. This substrate profile is in contrast to that reported
for the Taxus PAM reaction, where para-substituents are
converted with higher efficiency and larger ortho- and
21
precursor for the peptide antibiotic andrimid. As with all
class I lyase-like enzymes this unique amine chemistry is
mediated by the family-specific 4-methylideneimidazole-
5-one (MIO) prosthesis. The electrophilic MIO moiety is
formed post-translationally in the active site by cycliza-
tion and dehydration of the tripeptide [AT]SG, and is
suggested to bind and release the primary amine / am-
monia covalently, as required for (re)amination or deami-
1
0,11
nation (Figure 2).
The active site geometry of AdmH
has been demonstrated to limit the phenylalanine isomer-
ization reaction to a single rotamer of the cinnamate in-
termediate, with the MIO catalysing amination / deami-
nation on the same face, thus giving the particular stereo-
22
chemistry of β-amination. As neither these enzymes
(with proven or predicted (S)-selective aminomutase ac-
tivity) has been investigated in the amination direction,
EncP was selected as a template for development of the
missing biocatalyst in the PAL/PAM toolbox (Figure 1).
1
5
meta-products do not appear to be tolerated. This differ-
ence is likely due to the cinnamate being bound in alter-
nate orientations within the active sites of (S)- and (R)-
selective PAMs, as is evident from their overlaid crystal
A
pET expression vector containing the codon-
20
23
optimized gene encoding the EncP protein was used to
transform E. coli BL21(DE3) cells according to standard
procedures. The biocatalyst was then prepared via
structures . The methoxycinnamates 1p-r do not seem to
follow this pattern with the lower than expected conver-
sion of the ortho-isomer, possibly due to steric clash of
this bulky group with the activated β-carbon.
The regioselectivity was also found to be affected by
the ring substituents with respect to their electron-
withdrawing / donating nature, position and number.
Compared to cinnamate 1a, which gave only a slight ex-
cess of β-product, it was found that nitro groups any-
where on the aromatic ring gave predominantly α-amino
acids. Of these, 2- and 4-nitro- 1m and 1o gave much less
of the β-isomer than 3-nitro- 1n, which clearly suggests
resonance effects. The halogenated compounds 1b-l also
exhibited increased α-selectivity with ortho- (and to a
lesser extent meta-) substituents. The number of elec-
tron-withdrawing groups was found to have an effect, as
seen with the increase in α-product between 3-fluoro-,
3,5-difluoro- and 2,3,4,5,6-pentafluorocinnamate (1c,e,f).
An increased excess of β-amino acid was displayed with
all electron-donating substituents, with the exception of
2-methylcinnamate 1s. The methoxycinnamates 1p-r
Figure 2. The catalytic mechanism of the proposed (S)-
selective MIO-dependent amination of cinnamate de-
rivatives.
2
ACS Paragon Plus Environment

Page 3 of 8
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
(
a) 100
WT
80
60
40
20
0
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1a 1b 1c 1d 1e
ee 2 (%) >95 75 >95 >95 n.d.
ee 3 (%) >95 83 >95 >95 72 n.d. 81 >95 >95 >95 >95 >95
1f
ꢀ
1g 1h
n.d. >95 >90 >90 >95 n.d.
1i
1j
1k
1l 1m 1n 1o 1p 1q 1r
1s
1t
1u 1v
85 >95 84 n.d.
>95 >95 86
ꢀ
>95 n.d. n.d. >95
92 64 >95 >95
0
ꢀ
6
ꢀ
(
b) 100
R299K
80
60
40
20
0
1
a
1b 1c 1d 1e
ee 2 (%) >95 65 n.d. >95 n.d.
ee 3 (%) >95 >95 >95
1f
ꢀ
1g 1h
n.d. >95 >90 >90 >95 n.d.
1i
1j
1k
1l 1m 1n 1o 1p 1q 1r
1s
1t
1u 1v
ꢀ
n.d.
ꢀ
n.d. >95
0
ꢀ
84 >95 86
ꢀ
ꢀ
ꢀ
>95 n.d. >95 >95
ꢀ
>95 n.d.
ꢀ
68 >95 >95
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
(
c) 100
E293M
80
60
40
20
0
1
a
1b 1c 1d 1e
n.d. n.d.
1f
ꢀ
1g 1h
1i
>80
1j
ꢀ
1k
ꢀ
1l 1m 1n 1o 1p 1q 1r
n.d. >95
1s
1t
1u 1v
ee 2 (%) >95
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
0
>95 >95 82
ꢀ
ee 3 (%) >95 56 87 >95 48 n.d. 60 74 >95 72 62 >95 20 78 70 >95 >95 >95 >95 >95 >95 >95
Figure 3. Amination of arylacrylic acids 1a-v to the corresponding amino acids by EncP and two rationally designed,
regioselective variants (n.d. = not determined, - = not required).
showed mirrored regioselectivity to the nitrocinnamates
m-o, also implying resonance effects. Oddly the 4-
1
100
80
WT
4ꢀCH3
halocinnamates (1d,i,l) also showed an increase in β-
production relative to α-.
To probe the contribution of electronic effects to the
regioselectivity of the amination reaction, substituent-
and position-specific core electron binding energy shift
4
ꢀOCH3
2ꢀOCH3
3
ꢀCH3
4ꢀBr
60
4ꢀCl
3ꢀBr
3ꢀOCH
3
H
2
4
4ꢀF
(
ΔCEBE) values were plotted against the percentage β-
3
ꢀF
R² = 0.9191
40
3ꢀCl
product of each reaction (Figure 4). These calculated val-
ues are described to be a measure of the difference in en-
ergy required to remove an electron from the ipso-carbon
of a substituted aryl ring as compared to the unsubstitut-
ed π-system. As such ΔCEBE was used as a quantitative
measure of the electron richness/deficiency of the aro-
matic ring of each substrate and therefore its relative abil-
ity to activate the molecule for ammonia addition as
compared to the carboxyl moiety. There was a strong
negative correlation, consistent with either the conjugate
addition or the concerted hydroamination mechanism,
previously proposed for related aminomutases.
The data points were best fitted by two trend lines:
one shallower in gradient through the majority of the
ortho-substituents and the other through the remaining
2ꢀCH3
3ꢀNO
2
2
ꢀF
20
R² = 0.9093
2ꢀBr
4ꢀNO2
3ꢀCN
0.6
2ꢀCl
2ꢀNO
0.8 1
2
0
ꢀ
0.6
ꢀ0.4
ꢀ0.2
0
0.2
∆CEBE / eV
0.4
Figure 4. The dependence of β- vs. α-amination by EncP
on the core electron binding energy shift (ΔCEBE) due
to substrate ring substituents.
1
7
25
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 8
substrates, possibly indicating different binding modes in
the active site. It is noteworthy that 2-methoxycinnamate
p is the only ortho-substituted compound to give an
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
amination pattern that fits with the steeper trend line.
These observations (along with additional deviations,
such as increased β-selectivity for the electron deficient 4-
halocinnamate rings) imply that the specific interactions
of each substrate with the active site contribute to reac-
tion parameters, as recently proposed for amino acid
25
isomerization reactions with AdmH. In fact it could be
reasoned that interactions of ring substituents with the
active site may affect the positioning of each substrate.
For example the orientation of ortho-groups relative to
the rest of the substrate might enable different accom-
modation of these by the aryl binding pocket of the en-
zyme, compared to the elongated meta- and para-
analogues. This may impact upon positioning of the β-
and α-carbons relative to the catalytic residues and con-
tribute to the differing trends in amination preference
observed with these subsets of compounds.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
As expected, the amination reaction was found to
yield predominantly the (S)-amino acids ((S)-2 and (S)-3)
for the majority of substrates (Figure 3). In many cases
good to excellent ee values were obtained, with 9 exam-
ples of completely stereoselective addition at both β- and
α-positions. The ee of the α-products 3m-o for the nitro-
compounds was imperfect, with 2-nitrocinnamate 1m (the
highest converted of the three) giving a particularly low
value of just 6% ee for 3m. This low ee may be due to a
competing MIO independent amination process previous-
Figure 5. EncP active site catalytic and substrate posi-
tioning residues as inferred by homology modelling
based on AdmH.
the positioning of the substrate in the active site and thus
affecting the β- and/or α-addition preference.
Previous efforts aimed at engineering the carboxyl-
binding pocket in a plant PAL have revealed a delicate
hydrogen bonding network which is easily disturbed by
1
4
ly observed with AvPAL, allowing amination from either
face of the substrate. Interestingly, the 4-
26
mutagenesis. For this reason the original residues in
EncP were substituted for amino acids with the highest
similarity. Residue R299 was mutated to a K to maintain
the charge interaction with the substrate and this variant
showed increased β-selectivity (from 56:44 to 88:12) with
cinnamate 1a (Figure 3). This change in selectivity could
possibly be rationalized by the different shape and higher
charge density of the lysine terminal amine (versus the
arginine guanidinyl moiety) sliding the β-carbon of the
substrate in a more central position for amination by the
catalytic MIO and tyrosine residues (Figure 6b). Next the
neighbouring E293 was substituted for Q to generate a
variant that displayed opposite regioselectivity, producing
an excess of α-phenylalanine from 1a (43:57). With little
shape difference between glutamate and glutamine, the
change in selectivity may be due to reduced interaction
strength between positions 293 and 299. By removing the
salt bridge here it is likely that R299 is repositioned, shut-
tling the substrate in the active site so as to allow easier α-
amination versus β- (Figure 6c). This hypothesis was test-
ed further with an E293M substitution, to remove any
residual polar interactions between the two amino acid
side chains. This variant displayed further improved α-
selectivity (18:82) compared with EncP-E293Q. To our
knowledge the E293 variants are the first examples of
PAM enzymes with significantly increased α-selectivity
methoxycinnamate 1r was found to produce racemic 2r
under the reaction conditions. This outcome may be
linked to the steric and electronic similarities between the
4-methoxyphenylalanines 2-3r and tyrosine β- and α-
regioisomers (4-OCH vs. 4-OH substituent). Tyrosine is
3
disfavoured by PAL and PAM enzymes and may therefore
not be correctly bound in the active site for stereoselec-
tive amination/deamination reactions.
The probable interplay of substrate binding versus
electronic effects in EncP prompted rational design efforts
to tailor regioselectivity for compounds with undesirable
β:α ratios. A homology model of EncP was constructed
23
using the structure of AdmH (PDB ID: 3UNV) as a guide
(Figure 5 and Figures S1-S2 in the Supporting Infor-
mation). Residues considered to be important for sub-
strate positioning were selected and point mutations de-
signed. The most striking enzyme-substrate interactions
in this model seemed to be a hydrogen bond (from N340)
and a salt bridge (from R299) to the carboxyl group of the
substrate(Figure 5a). Residues N340 and R299 seemed, in
turn, to be coordinated by ionic and hydrogen bonding
interactions, courtesy of the neighbouring E293. It was
postulated that by substituting R299 and E293 in EncP
the nature of these interactions could be altered, shifting
4
ACS Paragon Plus Environment

Page 5 of 8
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
2
ꢀOCH 3ꢀCH
4ꢀBr 4ꢀCl
4ꢀF
4ꢀOCH3
3
3
100
2
ꢀCH
3
4ꢀCH3ꢀOCH
3
R299K
3
3ꢀBr
H
80
60
40
3ꢀCl
2ꢀBr
3ꢀF
2ꢀCl
2F
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
20
0
3
ꢀNO2
4
ꢀNO2
2ꢀNO
2
ꢀ
0.6
ꢀ0.4
ꢀ0.2
0
0.2
0.4
0.6
0.8
1
∆
CEBE / eV
100
E293Q
4ꢀCH3
8
0
0
4ꢀOCH3
6
4ꢀBr
3ꢀCH3
4ꢀCl
4ꢀF
3
ꢀOCH3
2
ꢀCH3
40
20
0
H
3ꢀBr
3ꢀF
2ꢀOCH3
Figure 6. The hypothesized effect of varying active site
residues on substrate positioning and ammonia addi-
tion preference.
3ꢀCl
2ꢀF
2ꢀCl
3ꢀNO2
2ꢀBr
2ꢀNO2
3
ꢀCN
4ꢀNO2
achieved through enzyme engineering. Further details on
the rational design are provided in the Supporting Infor-
mation.
ꢀ0.6
ꢀ0.4
ꢀ0.2
0
0.2
CEBE / eV
0.4
0.6
0.8
1
∆
1
00
E293M
The regio- and stereoselectivity of each of the three
variants was then tested with the entire panel of com-
pounds 1a-v under the same conditions as for the wild-
type enzyme (Figure 3, Figure 7). EncP-R299K was found
to convert all substrates with the exception of 3-
cyanocinnamate 1v, where no product was detected. For
the majority of substrates a large increase in β-selectivity
was observed, with 9 substrates producing exclusively the
β-amino acid (all 3 para-halocinnamates 1d, i and l, and
all 6 methyl- and methoxy-compounds 1p-u). Interesting-
ly the 3-fluoro- compound 1c showed only a minor β-:α-
product shift compared the other meta-halogenated sub-
strates (45:55 to 52:48) causing it to remain on the upper
wild-type trend line in the same region as the ortho-
halocompounds (Figure 7). Apart from this cluster and
the largely unaltered pattern for the nitro-products (m-o)
the previously influential effect of substrate electronic
effects seemed to have been greatly disrupted by the
R299K amino acid substitution. Both EncP-E293Q and
EncP-E293M showed conversion with the entire substrate
panel, with increased α-selectivity in most cases. EncP-
E293M in fact gave a β:α-product ratio of <30:70 with 20
of the 22 substrates (all except 1r and 1u). These 20 points
were found to lie around or below the lower trend line for
the wild-type reactions (Figure 7), indicating a similar but
opposite disruption of the regioselectivity-ΔCEBE
8
0
0
4
ꢀOCH3
6
4
ꢀCH3
40
4ꢀCl
3ꢀCH3
ꢀOCH3
ꢀ0.2
4ꢀBr
2ꢀCH3
20
4ꢀF
3ꢀCl
3ꢀF
3
H
2
ꢀBr 3ꢀBr
3ꢀNO
2
2ꢀNO2
2
ꢀOCH3
2ꢀCl
0.2
∆CEBE / eV
2
ꢀF
3ꢀCN
0.6
4ꢀNO2
0
ꢀ
0.6
ꢀ0.4
0
0.4
0.8
1
Figure 7. The dependence of β- vs. α-amination by EncP
rationally designed variants on the core electron bind-
ing energy shift (ΔCEBE) due to substrate ring substitu-
ents. Trends for the wild-type enzyme are shown as
dashed lines for comparison.
correlation, as seen with the first variant. The changes
observed for E293Q were more subtle (the results for con-
versions and enantioselectivity are given in Figure S7 in
the Supporting Information), with almost all para-
substitutions remaining on or around the wild-type upper
trend line (Figure 7). The majority of meta- and the un-
substituted compounds, however, had moved to an in-
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
termediary position on the graph; between the two origi-
Page 6 of 8
Table 1. Time course of the amination of methylcin-
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
namate 1s to methyl-β-phenylalanine 2s by the EncP-
R299K variant.
nal trend lines, with the remaining ortho- and 3-cyano-
substrates lying close to the lower trend line. The previ-
ously ‘outlying’ 1p displayed a marked jump from β- to α-
selectivity (β:α ratio from 87:13 to 37:63), having fully mi-
grated between the two trend lines. Again compounds
with strongly electron-withdrawing groups (ΔCEBE > 0.6
eV) showed little change in regioselectivity with either
E293 variant, as with the β-selective enzyme. This implies
that a highly electron deficient ring activates the α-carbon
strongly enough to counteract any changes in substrate
positioning we had hypothesized. By contrast the greatest
O
NH2 O
EncPꢀR299K
OH
OH
(NH4)2SO4 (pH 8.3)
55°C
1
s
(S)ꢀ2s
Time
Conv.
(%)
a
ee of 2s
a
β:α ratio
b
(h)
(%)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
4
8
10
19
22
24
26
28
99:1
99:1
99:1
99:1
99:1
99:1
>96
95
92
90
84
76
shift
in
regioselectivity
was
seen
with
2-
methoxycinnamate 1p which, despite lowered conversion,
could be produced as either the β- or α-amino acid with
no detectable trace of the other regioisomer with EncP-
R299K or EncP-E293M respectively.
The enantioselectivity of the β-amination reactions
across all variants was found to be similar for each sub-
strate regardless of the differences in conversion and re-
gioselectivities observed. By contrast the ee values for α-
amination were found to decrease for many of the sub-
strates with increasing α-regioselectivity (R299K > WT >
E293Q/M). With the β-selective EncP-R299K only one
substrate (2-nitrocinnamate 1m) gave the α-product (S)-
22
4
8
a
Conversion and product ratios determined by non-chiral
b
HPLC. Enantiomeric excesses determined by chiral HPLC.
EXPERIMENTAL SECTION
General methods and materials. Analytical grade rea-
gents and solvents were obtained from Sigma-Aldrich,
AlfaAesar or Fisher Scientific and used without further puri-
fication, unless stated otherwise. Reference standards of op-
tically pure amino acids were purchased from Sigma-Aldrich,
AlfaAesar or PepTech Corporation. Plasmid DNA purification
was performed with the QIAprep Spin miniprep kit (Qiagen),
according to the manufacturer’s instructions. Automated
DNA sequencing was performed by MWG Eurofins. E. coli
DH5α was used as a cloning host for plasmid propagation, E.
coli BL21(DE3) as an expression host for protein production.
Chemically competent cells of both strains were purchased
from New England Biolabs. Solid and liquid media were sup-
3
m with <95% ee, with all others showing no detectable
trace of the opposite enantiomer. The more α-selective
variants gave moderate to poor ees with 10 of the 22 con-
verted substrates of which 5 were <60% (S) for E293Q and
4 for E293M.
As a further test of the enantioselectivity, a time
course experiment was set up to follow the enantiomeric
excess for conversion of 1s to 2s using the EncP-R299K
variant (Figure 3) to see if a similar erosion of ee was ob-
served for β-amination by EncP to that reported for α-
–1
plemented with kanamycin (50 μg mL final concentration).
1
4
amination in AvPAL. After 48 h it was found that, alt-
hough conversion tailed off after the first 8 h, the enanti-
omeric excess of 2s continued to fall (Table 1). Interest-
ingly, this was found not to be the case with the meta-
and para-isomers 2t-u (Table S11, Supporting Infor-
mation). This may demonstrate a difference in the prefer-
ence of the enzyme for the two possible binding modes of
ortho-substituted arylacrylates, as previously alluded to
Solid media were prepared by addition of agar (1.5% w/v) to
liquid media. The pET-28a-EncP plasmid (pET-28a vector
containing the codon-optimised gene encoding the EncP
protein) was obtained from Prof. Jason Micklefield and used
as previously reported.
Molecular visualisations and modelling. The protein
structure for the aminomutase AdmH was downloaded
from the Research Collaboratory for Structural Bioinformat-
ics’ Protein Data Bank – RCSB PDB (PDB ID: 3UNV). Homol-
ogy models were created using the experiment option in the
YASARA Structure version 14.7.17, with the ‘slow’ modelling
method. For this the amino acid sequence of the query en-
zyme was input in FASTA format along with the unmodified
PDB file for the appropriate structure. The number of PSI-
BLAST iterations to build a position-specific scoring matrix
from related sequences was set to 3 with an E-value cut off of
20
23
25,27
by Walker et al.
In summary, we have demonstrated that EncP can be
engineered to enhance the regioselectivity of the addition
of ammonia to cinnamic acid derivatives. The varying and
often undesirable regioselectivity of the enzyme was
found to be heavily influenced by the electronic proper-
ties of the substrates. Rational design of three EncP vari-
ants revealed that both the β- and α-selectivity of the en-
zyme could be enhanced, allowing conversion to either
(S)-β or (S)-α-arylalanines with 99:1 selectivity and high
ee. Further engineering is currently underway to improve
conversions for the β-products in order to fully develop a
biocatalyst for the preparative synthesis of (S)-β-amino
acids.
0.5. Sequences profiles were obtained from YASARA’s posi-
tion specific scoring matrix (PSSM). The maximum number
of top scoring templates was set to 5, with a maximum of 5
alignments per template and a maximum oligomerization
state for templates of 4. The number of sampling experi-
ments to optimise loop structures was set to 50 with a termi-
nal extension value of 10. Side-chain rotamers were selected
6
ACS Paragon Plus Environment

Page 7 of 8
Journal of the American Chemical Society
using electrostatic and knowledge-based packing interac-
tions as well as solvation effects with implicit solvent. Hy-
Funding Sources
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
th
This work was funded by the European Union’s 7 Frame-
work program FP7/2007-2013 under grant agreement no.
drogen bonding network were optimized to including pH-
dependence and ligands bound via hydrogen prediction and
visualisation. A high-resolution energy minimization with a
shell of explicit solvent molecules was run using YASARA
version-specific knowledge-based force field parameters.
Site-directed mutagenesis of EncP. Specific mutations
were introduced in the pET-28a-EncP template using the
Phusion site-directed mutagenesis kit (Thermo Scientific)
according to the protocol provided. The PCR reactions were
performed on an Eppendorf Mastercycler Gradient using the
following primers (mutated codon underlined): R299K_Fw =
2
89646 (KYROBIO). F.P. and S.T.A. were supported by the
Biotechnology and Biological Sciences Research Council
BBSRC) and Glaxo-SmithKline (GSK) under the Strategic
(
Longer and Larger (sLoLa) grant initiative ref. BB/K00199X/1.
We thank the Royal Society for a Wolfson Research Merit
Award (N. J. T.)
ACKNOWLEDGMENT
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
We thank Professor Jason Micklefield and collaborators for
providing the construct for production of EncP. Thanks also
go to Scott P. France and Matthew M. Heberling for help
with analytics.
5
’-AGCATTAAGTGTACACCG-3’, R299K_Fw = 3’-CCGTTAA-
CTTCTACGTATA-5’, E293Q_Fw = 5'-GGCAATTCAGGAT-
GCATATAGC-3', E293M_Fw = 5'-GGCAATTATGGATGCA-
TATAGC-3', E293Q_Rv (= E293M_Rv) = 3'-CAACGTTTTCG-
TCCATCGGA-5'. PCR conditions: 98°C (30 s), [98°C (15 s),
REFERENCES
1.
Walker, K. D.; Klettke, K.; Akiyama, T.; Croteau, R. J. Biol.
Chem. 2004, 279, 53947–53954.
5
8°C (30 s), 98°C (150 s)] × 30, 72°C (300 s). Ligated PCR
product or template plasmid was transformed into E. coli
DH5α as per the manufacturer’s protocol and grown on LB-
agar plates for 16 h at 37°C. Single colonies from these were
used to inoculate LB cultures (5 mL) grown for 16 hours at
2.
Jin, M.; Fischbach, M. A.; Clardy, J. J. Am. Chem. Soc. 2006,
1
28, 10660–10661.
3.
Huang, X.; Brien, E. O.; Thai, F.; Cooper, G. Org. Process
Res. Dev. 2010, 14, 592–599.
37°C (250 rpm shaking). Plasmid DNA was isolated and the
4
.
Ruf, S.; Buning, C.; Schreuder, H.; Horstick, G.; Linz, W.;
Olpp, T.; Pernerstorfer, J.; Hiss, K.; Kroll, K.; Kannt, A.;
Kohlmann, M.; Linz, D.; Hübschle, T.; Rütten, H.; Wirth,
K.; Schmidt, T.; Sadowski, T. J. Med. Chem. 2012, 55, 7636–
presence of the mutation verified by automated DNA se-
quencing. The sample was then transformed into E. coli
BL21(DE3) for protein production.
Whole cell biocatalyst production. LB medium (8 mL,
supplemented with kanamycin) was inoculated with a single
colony of E. coli BL21(DE3) containing the suitable plasmid
and grown for 16 h at 37°C and 250 rpm. The starter culture
7649.
5
6.
.
Dunn, C. J.; Faulds, D. Drugs 2000, 60, 607–617.
Kálai, T.; Altenbach, C.; Cascio, D.; Peters, F. B.; Hideg, K.;
Schultz, P. G.; Bonneville, J.; Tichtinsky, G.; Von, D.;
Springer, A.; Reinbothe, S. Proc. Natl. Acad. Sci. 2010, 107,
21637-21642.
28
was used to inoculate LB-based autoinduction medium
(
800 mL, supplemented with kanamycin), which was incu-
bated at 18°C and 250 rpm for 4 days. The cells were harvest-
ed by centrifugation (4000 rpm, 12 min) and stored at –20°C
until further use. Protein over-production was confirmed via
SDS-PAGE analysis and by purification by immobilized metal
affinity chromatography (Supporting Information, Figure S3).
Optimized biotransformation procedure for the
amination of substrates 1a-v. E. coli BL21(DE3) whole cells
containing the PAL variant (40 mg wet weight) were resus-
pended in a solution of substrate 1a-v (1 mM) and (NH ) SO
7.
8.
Peters, F. B.; Brock, A.; Wang, J.; Schultz, P. G. Chem. Biol.
2009, 16, 148–152.
Aguilar, M.-I.; Purcell, A. W.; Devi, R.; Lew, R.; Rossjohn, J.;
Smith, A. I.; Perlmutter, P. Org. Biomol. Chem. 2007, 5,
2884–2890.
9
.
Steer, D. L.; Lew, R. A.; Perlmutter, P.; Smith, A. I.; Aguilar,
M.-I. Curr. Med. Chem. 2002, 9, 811–822.
1
0. a) Turner, N. J. Curr. Opin. Chem. Biol. 2011, 15, 234–240. b)
Heberling, M. M.; Wu, B.; Bartsch, S.; Janssen, D. B. Curr.
Opin. Chem. Biol. 2013, 17, 250–260.
4
2
4
(
4 M, pH 8.3, 0.8 mL total volume). The mixture was incu-
11.
Poppe, L. Curr. Opin. Chem. Biol. 2001, 5, 512–524.
bated at 55°C and 400 rpm for 22 h unless otherwise stated.
Biotransformation samples were mixed with an equal volume
of MeOH, vortexed and centrifuged (13000 rpm, 3 min). The
supernatant was transferred to a 0.45 μm filter vial and used
directly for HPLC analysis. Methods and conditions are re-
ported in the Supporting Information.
12. Gloge, A.; Zoń, J.; Kövári, A.; Poppe, L.; Rétey, J. Chem. Eur.
J. 2000, 6, 3386–3390.
1
3. De Lange, B.; Hyett, D. J.; Maas, P. J. D.; Mink, D.; van
Assema, F. B. J.; Sereinig, N.; de Vries, A. H. M.; de Vries, J.
G. ChemCatChem 2011, 3, 289–292.
14. a) Parmeggiani, F.; Lovelock, S. L.; Weise, N. J.; Ahmed, S.
T.; Turner, N. J. Angew. Chem. Int. Ed. Engl. 2015, 54, 4608–
4611. b) Lovelock, S. L.; Lloyd, R. C.; Turner, N. J. Angew.
Chem. Int. Ed. Engl. 2014, 53, 4652–4656.
15. Szymanski, W.; Wu, B.; Weiner, B.; de Wildeman, S.;
Feringa, B. L.; Janssen, D. B.; J. Org. Chem. 2009, 74, 9152-
ASSOCIATED CONTENT
Supporting Information. Figures S1-S7, Tables S1-S11, Fur-
ther experimental details, analytical methods and additional
data. This material is available free of charge via the Internet
at http://pubs.acs.org.
9
157.
16. Bartsch, S.; Wybenga, G. G.; Jansen, M.; Heberling, M. M.;
Wu, B.; Dijkstra, B. W.; Janssen, D. B. ChemCatChem 2013,
AUTHOR INFORMATION
Corresponding Author
5
, 1797–1802.
17. Wu, B.; Szymański, W.; Wybenga, G. G.; Heberling, M. M.;
Bartsch, S.; de Wildeman, S.; Poelarends, G. J.; Feringa, B.
*
E-mail: nicholas.turner@manchester.ac.uk
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 8
L.; Dijkstra, B. W.; Janssen, D. B Angew. Chem. Int. Ed.
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Engl. 2012, 51, 482–486.
18. Xiang, L.; Moore, B. S. J. Biol. Chem. 2002, 277, 32505–
32509.
1
9. Xiang, L.; Moore, B. S. J. Bacteriol. 2005, 187, 4286–4289.
0. Chesters, C.; Wilding, M.; Goodall, M.; Micklefield, J. An-
gew. Chem. Int. Ed. Engl. 2012, 51, 4344–4348.
1. Magarvey, N. A.; Fortin, P. D.; Thomas, P. M.; Kelleher, N.
L.; Walsh, C. T. ACS Chem. Biol. 2008, 3, 542–554.
2. Ratnayake, N. D.; Wanninayake, U.; Geiger, J. H.; Walker,
K. D. J. Am. Chem. Soc. 2011, 133, 8531–8533.
23. Strom, S.; Wanninayake, U.; Ratnayake, N. D.; Walker, K.
2
2
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
D.; Geiger, J. H. Angew. Chem. Int. Ed. Engl. 2012, 51, 2898–
2902.
2
2
2
4. Segala, M.; Takahata, Y.; Chong, D. P. J. Mol. Struct.
THEOCHEM 2006, 758, 61–69.
5. Ratnayake, N. D.; Liu, N.; Kuhn, L. A.; Walker, K. D. ACS
Catal. 2014, 4, 3077–3090.
6. Bartsch, S.; Bornscheuer, U. T. Protein Eng. Des. Sel. 2010,
23, 929–933 (2010).
27. Wanninayake, U.; Walker, K. D. J. Am. Chem. Soc. 2013,
35, 11193–11204.
8. Studier, F. W. Prot. Expr. Purif. 2005, 41, 207-234.
1
2
_
___________________________________________________
TOC graphic
_
_________________________________________________
8
ACS Paragon Plus Environment