A versatile biosynthetic approach to amide bond formation

Article abstract of DOI:10.1039/c8gc01697f

The development of versatile and sustainable catalytic strategies for amide bond formation is a major objective for the pharmaceutical sector and the wider chemical industry. Herein, we report a biocatalytic approach to amide synthesis which exploits the diversity of Nature's amide bond forming enzymes, N-acyltransferases (NATs) and CoA ligases (CLs). By selecting combinations of NATs and CLs with desired substrate profiles, non-natural biocatalytic pathways can be built in a predictable fashion to allow access to structurally diverse secondary and tertiary amides in high yield using stoichiometric ratios of carboxylic acid and amine coupling partners. Transformations can be performed in vitro using isolated enzymes, or in vivo where reactions rely solely on cofactors generated by the cell. The utility of these whole cell systems is showcased through the preparative scale synthesis of a key intermediate of Losmapimod (GW856553X), a selective p38-mitogen activated protein kinase inhibitor.

Full text of DOI:10.1039/c8gc01697f

Green Chemistry
View Article Online
View Journal
COMMUNICATION
A versatile biosynthetic approach to amide bond
formation†
Cite this: DOI: 10.1039/c8gc01697f
Helena K. Philpott,a Pamela J. Thomas,b David Tew,a Doug E. Fuerstc and
Received 29th May 2018,
Accepted 21st June 2018
Sarah L. Lovelock *‡a
DOI: 10.1039/c8gc01697f
rsc.li/greenchem
The development of versatile and sustainable catalytic strategies cessing.2 In addition, reactions involving complex molecules
for amide bond formation is a major objective for the pharma- are often challenging in terms of chemo-, regio- and stereo-
ceutical sector and the wider chemical industry. Herein, we report selectivity. It is therefore perhaps unsurprising that more atom
a biocatalytic approach to amide synthesis which exploits the economical and preferably catalytic methods for amide bond
diversity of Nature’s amide bond forming enzymes, formation were voted number one on the list of desirable
N-acyltransferases (NATs) and CoA ligases (CLs). By selecting com- transformations by the American Chemical Society Green
binations of NATs and CLs with desired substrate profiles, non- Chemistry Institute.3 This challenge has certainly contributed
natural biocatalytic pathways can be built in a predictable fashion to the growing interest, in both academic and industrial lab-
to allow access to structurally diverse secondary and tertiary oratories, in developing a new generation of sustainable strat-
amides in high yield using stoichiometric ratios of carboxylic acid egies for amide synthesis.4
and amine coupling partners. Transformations can be performed
Despite the successful development of a number of elegant
in vitro using isolated enzymes, or in vivo where reactions rely catalytic approaches including the use of boronic acid derived
solely on cofactors generated by the cell. The utility of these catalysts,5 oxidative strategies using transition metal- and
whole cell systems is showcased through the preparative scale syn- organo-catalysts,6 and those involving biocatalysts,7 the uptake
thesis of a key intermediate of Losmapimod (GW856553X), a selec- of these technologies in the pharmaceutical and agrochemical
tive p38-mitogen activated protein kinase inhibitor.
industries has not been realized. The reasons for this slow
translation from basic research to application are almost cer-
tainly multi-faceted, but it is clear that these emerging strat-
egies must compete with very established and robust method-
ologies, in an environment with considerable time pressures
along the product discovery, lead optimization and production
pipeline.
In Nature, N-acyltransferases (NATs) catalyze amide for-
mation between activated thioesters and amines, and can be
found in a number of pathways including xenobiotic detoxifi-
cation8 and the biosynthesis of secondary metabolites (e.g.
antibiotics9 and plant phenylpropanoid amides10). The thio-
ester precursors are generated from the corresponding acid
and coenzyme A (CoA) catalyzed by ATP-dependent CoA ligases
(CL) (Scheme 1). CLs and NATs are believed to readily adapt to
changing environmental pressures via evolutionary processes
Amide bond formations are amongst the most commonly uti-
lized transformations in the pharmaceutical industry, as evi-
denced by the widespread occurrence of amide linkages in
pharmaceuticals and other bioactive molecules.1 As a result,
amide synthesis is often overlooked as a challenge in modern
organic chemistry due to the availability of established
methods of performing this fundamental transformation.
However, current strategies rely on the use of stoichiometric
activating or coupling reagents which result in poor atom
economy and lead to the generation of often toxic or hazar-
dous by-products, which severely complicate downstream pro-
aAdvanced Manufacturing Technologies, GlaxoSmithKline, Medicines Research
Centre, Gunnels Wood Road, Stevenage, SG1 2NY, UK.
E-mail: sarah.lovelock@manchester.ac.uk
bMolecular Design, Computational and Modelling Sciences, GlaxoSmithKline,
Medicines Research Centre, Gunnels Wood Road, Stevenage, SG1 2NY, UK
cAdvanced Manufacturing Technologies, GlaxoSmithKline, 709 Swedeland Road,
King of Prussia, PA 19406, USA
†Electronic supplementary information (ESI) available: Experimental details,
supplementary tables and figures. See DOI: 10.1039/c8gc01697f
‡Present address: Manchester Institute of Biotechnology, University of
Manchester, 131 Princess Street, Manchester, M1 7DN, UK.
Scheme 1 CoA ligase and N-acyltransferase catalyzed amide bond
formation.
This journal is © The Royal Society of Chemistry 2018
Green Chem.

View Article Online
Communication
Green Chemistry
Table 1 Activity of CoA ligases towards carboxylic acids 1a–1o
and a large number of wild-type enzymes have been reported
which are involved in the synthesis of structurally diverse
amides.11 However, at present the utility of these enzymes as
versatile biocatalysts for sustainable amide bond formations
has not been realized.
Herein, we describe a versatile and sustainable biosynthetic
cascade strategy for amide bond formation. By selecting com-
binations of NATs and CLs with desired substrate profiles,
non-natural biocatalytic pathways can be built in a predicable
fashion to allow access to structurally diverse secondary and
tertiary amides. Furthermore, we demonstrate that cascades
performed in vivo rely solely on cofactors generated by the cell,
avoiding the need to supply expensive co-factors.
We selected a panel of nine CoA ligases, including ipfF12
and CBL13 which have emerged in recent years in response to
environmental contaminants (ibruprofen and chlorinated aro-
matics
respectively)
from
anthropogenic
sources.
Phenylacetate CoA ligase from Penicillium chrysogenum (phl) is
involved in the biosynthesis of penicillins and has previously
been reported to have broad substrate promiscuity and accepts
a range of aromatic and aliphatic acids.9 Following expression
in E. coli, the activity of purified CLs towards a panel of fifteen
acids (1a–1o) was determined using a spectrophotometric pyr-
ophosphate detection assay (EnzChek, Invitrogen). The
selected CLs have complimentary substrate profiles (Table 1),
and combined are able to operate across broad chemical
space. CLs were identified with activity towards structurally
diverse acids containing aromatic (1a–i), heteroaromatic (1j–k),
allylic (1m) and aliphatic (1l & 1n–o) functionality. RpCL,14
LcCL, ArCL and MsCL are annotated as cyclohex-1-ene carboxy-
late CLs in the UniProt database (and share 77%, 60% and
38% sequence identity to RpCL respectively). These proteins
demonstrated specificity towards cyclohexanoic acid deriva-
tives, although other substrates were accepted to a lesser
extent. Nonetheless, the activities of RpCL and its homologues
are complimentary to the other CLs in the panel, and allow
access to a wide range of CoA ester intermediates. Selected
Each reaction contained 2.5 µg CL, 0.4 mM substrate, 0.4 mM ATP,
0.4 mM CoA and Enzchek assay components: 0.2 mM 2-amino-6-
mercapto-7-methylpurine
ribonucleoside,
purine
nucleoside
phosphorylase (0.025 U), inorganic pyrophosphatase (0.0075 U), 1 mM
MgCl2, 0.1 mM sodium azide in 50 mM Tris-HCl, pH 7.5. The increase
in absorbance was monitored at 360 nm.
reactions were repeated on a preparative scale (5 ml, 2.5 mM) representing NATs from across both superfamilies were selected
using stoichiometric ATP and CoA, with successful conversions and synthesized (Genscript). Expression trials in E. coli per-
to the CoA esters confirmed by UPLC (see ESI†).
Having developed versatile methods for the preparation of NATs which were produced solubly under these conditions.
CoA esters, a panel of phylogenetically diverse NATs were identi-
To begin the bimolecular substrate profiling of the NAT
formed in 96-well plate format led to the identification of 45
fied from the UniProt database (EC 2.3.1) and selected for evalu- panel, each enzyme was initially evaluated with four amines
ation as amide bond forming catalysts. N-Acyltransferases are (2a–d) in combination with a single CoA ester, which was
divided into several unrelated classes, which proceed via selected from a small panel of naturally occurring substrates.
different catalytic mechanisms. Arylamine N-acyltransferases are For each NAT, the CoA ester selected was either a known sub-
a class of xenobiotic metabolizing enzymes which are structu- strate or a close structural analogue (see Table S4†). Activity
rally related to transglutaminases and cysteine proteases and assays were performed using clarified cell lysates and for-
utilize a conserved Cys-His-Asp/Glu catalytic triad to promote mation of the amide products were detected by UPLC-MS. Out
acyl transfer via a double displacement Ping Pong Bi Bi mecha- of the 45 soluble NATs, a total of 31 enzymes accepted either
nism.15
A
second superfamily are the GCN5-related 2,4-dimethoxyaniline (2a) and/or 4-(aminomethyl)benzonitrile
N-acyltransferases, which are implicated in a large number of (2b). To explore thioester promiscuity of these 31 NATs, one
secondary metabolic pathways. These enzymes proceed via an pot reactions were performed using amines 2a or 2b, and a
ordered Bi Bi mechanism in which acyl groups are directly trans- range of CoA esters produced in situ from the corresponding
ferred from CoA esters to amine coupling partners, without the acids and a suitable CL (determined in Table 1) in the pres-
generation of acyl-enzymes intermediates.16 A total of 67 genes ence of 1 mol% CoA. Amide product formation was detected
Green Chem.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Green Chemistry
Communication
This journal is © The Royal Society of Chemistry 2018
Green Chem.

View Article Online
Communication
Green Chemistry
by UPLC-MS by comparison to chemically synthesized stan-
dards. NATs were identified with activity towards structurally
diverse thioesters derived from acids 1a, 1c–e, 1g, 1km & 1o.
Furthermore, a large number of the NATs screened demon-
strated considerable substrate promiscuity and catalyzed the
formation of diverse amide products with high conversions
(Table 2). In contrast, several NATs (01StAT, 02MsAT, 19CpAT,
35MaAT, 38ShAT, 45SpAT, and 64ScAT) showed high specificity
for acetyl CoA, and were not active towards thioesters derived
from the acids screened (1a, 1c–e, 1g, 1k–m & 1o) (see ESI†).
Other NATs (23DmAT, 28BaAT, 36PaAT and 65PsAT) showed
only low conversion with a single thioester substrate as
detailed in Table S5.† It is noteworthy, that in the presence of
alternative thioesters, NATs from the transglutaminase super-
family (05PaAT, 07NfAT, 41GmAT, 42GmAT, 43GmAT and
66BcAT) yielded a by-product corresponding to the acetylated
amine, presumably as a result of a competing reaction with
acetyl CoA present in the cell lysate.
The substrate tolerance of the NAT panel was further evalu-
ated towards structurally diverse amines and anilines (2a–2i)
in combination with a selected thioester produced in situ from
the corresponding acid (Table 2). NATs classified as tyramine
hydroxycinnamoyl transferases (32NtAT, 48CaAT, 49StAT,
54SlAT, 55SlAT, 56SlAT, 67CaAT) were highly promiscuous and
accepted a number of amine substrates. These NATs also dis-
played considerable thioester substrate promiscuity, making
them highly attractive biocatalysts for versatile amide bond for-
mations. The spermidine/agmatine hydroxycinnamoyl trans-
ferases gave variable results; 10HvAT and 12AtAT were specific
for substituted benzylamine 2b, whereas 11AtAT and 14AtAT
accepted a number of amine substrates. The NATs from the
transglutaminase superfamily were more specific and had a
preference for aniline analogues and to a lesser extent trypta-
mine. Remarkably 05PaAT and 07NfAT are able to utilize the
highly deactivated p-nitroaniline (2h) as a substrate. In con-
trast 4-aminopyridine was not tolerated by any of the NATs
evaluated. Combined, the NATs were able to operate across
broad chemical space and accepted amines and anilines with
a range of pKa values. NATs were also evaluated for activity
towards the single enantiomers of 2d and 2e. The NATs which
accepted these amines showed complete selectivity towards
the (S)-enantiomers, highlighting the potential utility of these
enzymes for performing kinetic resolutions.
Fig. 1 Activity of 11AtAT towards secondary amines 2j–2o.
secondary amines (2k–2o) in combination with cinnamoyl CoA
(produced in situ) as the acyl donor (Fig. 1). A total of five sec-
ondary amines were accepted, with the highest conversion
observed with N-methylbenzylamine (2k). Of the substrates
tested, L-proline methyl ester (2o) was the only secondary
amine which was not tolerated.
The in vitro reactions described above provide a detailed
understanding of the substrate tolerance of individual NATs
and CLs, thus guiding the predictable design of biosynthetic
cascades to target amide products. However, these transform-
ations rely on the use of stoichiometric ATP as an activating
agent. To minimize production costs and improve reaction
sustainability, it is advantageous to perform transformations
in vivo to alleviate the requirement to supply costly CoA and
stoichiometric ATP cofactors. This approach also creates the
opportunity to integrate amide bond formation into the ‘syn-
thetic biology toolbox’ for the creation of designer cell factories
for the production of high-value molecules, which are not
compatible with alternative biocatalytic approaches which
utilize lipases in organic solvents.
To highlight the utility of our in vivo biosynthetic approach
to amide bond formation, we selected a key intermediate in
the production of Losmapimod (GW856553X), a selective p38-
mitogen activated protein kinase inhibitor, as a target.18 Based
on our earlier substrate profiling, we selected 4-chlorobenzoate
CL from Alcaligenes sp. (CBL) and serotonin hydroxycinnamoyl
transferase from Capsicum annuum (66CaAT) as attractive can-
didates to perform the desired transformation (Fig. 2). The con-
version of 6-chloronicotinic acid (1p, 10 mM) and neopentyl-
amine (2p, 10 mM) to 3 proceeded in >99% conversion using
E. coli whole cells expressing CBL and 66CaAT in LB medium.
Transformations performed in Tris buffer (50 mM, pH 8.0)
with and without 10% v/v glycerol resulted in reduced conver-
sions of 71% and 28% respectively, suggesting that co-factor
regeneration by the E. coli cells is important to achieve high
conversions. A subsequent preparative scale transformation
(10 mM, 0.6 mmol) (with reduced catalyst loading) proceeded
in 83% conversion with an isolated yield of 74% following
In general, NATs are reported to be highly specific for
primary amines. To our knowledge the only enzymes known to
catalyze tertiary amide formation are those which acetylate
close structural analogues of proline including L-azetidine-2-
carboxylic acid (AZC) and 4-hydroxy-L-proline.17 Two AZC acyl-
transferases were expressed and assayed; however, they were
highly specific for AZC and acetyl CoA (see ESI†). Evaluation of
our NAT panel towards the secondary amine 4-piperidinone
(2j) led to the identification of a single enzyme, an agmatine
courmaroyltransferase from Arabidopsis thaliana (11AtAT) that
successfully catalyzed the formation of a tertiary amide basic extraction and evaporation of residual volatile neopentyl-
product from amine 2j and cinnamoyl CoA (14% conversion). amine (b.p. 83 °C), with no requirement for chromatographic
11AtAT was subsequently evaluated towards a small panel of separations.
Green Chem.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Green Chemistry
Communication
Notes and references
1 S. D. Roughley and A. M. Jordan, J. Med. Chem., 2011, 54,
3451.
2 E. Valeur and M. Bradley, Chem. Soc. Rev., 2009, 38, 606.
3 D. J. C. Constable, P. J. Dunn, J. D. Hayler, G. R. Humphrey,
J. L. Leazer Jr., R. J. Linderman, K. Lorenz, J. Manley,
B. A. Pearlman, A. Wells, A. Zaks and T. Y. Zhang, Green
Chem., 2007, 9, 411.
4 V. R. Pattabiraman and J. W. Bode, Nature, 2011, 480, 471;
H. Lundberg, F. Tinnis, N. Selander and H. Adolfsson,
Chem. Soc. Rev., 2014, 43, 2714; R. M. de Figueiredo,
S. Suppo and J. M. Campagne, Chem. Rev., 2016, 116,
12029.
Fig. 2 In vivo production of a key amide intermediate in the synthesis
of Losmapimod.
5 K. Arnold, B. Davies, D. Herault and A. Whiting, Angew.
Chem., Int. Ed., 2008, 47, 2673; N. Gernigon, R. M. Al-Zoubi
and G. Hall, J. Org. Chem., 2012, 77, 3886; K. Ishihara and
Y. Lu, Chem. Sci., 2016, 7, 1276; H. Noda, M. Furutachi,
Y. Asada, M. Shibasaki and N. Kumagai, Nat. Chem., 2017,
9, 571.
6 C. Gunanathan, Y. B. David and D. Milstein, Science, 2007,
317, 790; S. de Sarkar and A. Studer, Org. Lett., 2010, 12,
1992; B. Shen, D. M. Makley and J. N. Johnston, Nature,
2010, 465, 1027; H. Lundberg and H. Adolfsson, ACS Catal.,
2015, 5, 3271; H. Lundberg, F. Tinnis, J. Zhang, A. Algarra,
F. Himo and H. Adolfsson, J. Am. Chem. Soc., 2017, 139,
2286.
7 V. Gotor-Fernandez, E. Busto and V. Gotor, Adv. Synth.
Catal., 2006, 348, 797; A. Goswami and S. G. Van Lanen,
Mol. Biosyst., 2015, 11, 338; J. Pitzer and K. Steiner,
J. Biotechnol., 2016, 235, 32; A. J. L. Wood, N. J. Weise,
J. D. Frampton, M. S. Dunstan, M. A. Hollas,
S. R. Derrington, R. C. Lloyd, D. Quaglia, F. Parmeggiani,
D. Leys, N. J. Turner and S. L. Flitsch, Angew. Chem., Int.
Ed., 2017, 56, 14498; R. Hara, K. Hirai, S. Suzuki and
K. Kino, Sci. Rep., 2018, 8, 2950.
8 E. P. Karagianni, E. Kontomina, B. Davis, B. Kotseli,
T. Tsirka, V. Garefalaki, E. Sim, A. E. Glenn and
S. Boukouvala, Sci. Rep., 2015, 12900, 1; X. Kubiak, I. L. de
la Sierra-Gallay, A. F. Chaffotte, B. Pluvinage, P. Weber,
A. Haouz, J.-M. Dupret and F. Rodrigues-Lima, J. Biol.
Chem., 2013, 288, 22493; I. M. Westwood and E. Sim,
Biochemistry, 2007, 8, 3.
Conclusions
We have developed a versatile biosynthetic approach to amide
bond formation which addresses the growing sustainability
concerns associated with traditional amine and carboxylic acid
coupling methods. The methodology developed utilizes renew-
able whole cell catalysts which do not require the addition of
external co-factors. The transformations operate in aqueous
media under ambient conditions and do not require a signifi-
cant excess of either coupling partner. Furthermore, it is
anticipated that many desirable amide products can be iso-
lated following simple acid/base extractions, as highlighted
herein during the production of a key Losmapimod inter-
mediate, or through selective crystallization of amide products.
At present, substrate loadings are modest and there is a clear
requirement for future process intensification. However, there
is considerable precedent for dramatically increasing substrate
loadings in biotransformations through biocatalyst engineer-
ing (directed evolution),19 including several products in GSK’s
portfolio. Our methodology allows access to structurally
diverse secondary and tertiary amides by exploiting the natural
diversity of CLs and NATs. It is anticipated that laboratory evol-
ution will further expand the substrate range and practical
utility of these enzyme families. Substrate profiling of NATs
and CLs enables the predictable design of self-sufficient whole
cell catalysts, thus adding amide bond formation to the
growing repertoire of transformations accessible to synthetic
biology.
9 Z. D. Dunn, W. J. Wever, N. J. Economou, A. A. Bowersand
and B. Li, Angew. Chem., Int. Ed., 2015, 54, 5137;
M. J. Koetsier, P. A. Jekel, M. A. van den Berg,
R. A. L. Bovenberg and D. B. Janssen, Biochem. J., 2009,
417, 467.
Conflicts of interest
There are no conflicts to declare.
10 K. Kang and K. Back, Metab. Eng., 2009, 11, 64; S.-M. Jang,
A. Ishihara and K. Back, Plant Physiol., 2004, 135, 346.
11 A. Sabbagh, C. Veyssière, E. Lecompte, S. Boukouvala,
E. S. Poloni, P. Darlu and B. Crouau-Roy, Evol. Biol., 2013,
13, 62.
Acknowledgements
The authors would like to thank the global GSK Synthetic 12 R. W. Murdoch and A. G. Hay, Microbiology, 2013, 159, 621.
Biochemistry team for helpful discussions, and Tomasz 13 R. Wu, A. S. Reger, J. Cao, A. M. Gulick and D. Dunaway-
Kubowiz for analytical support.
Mariano, Biochemistry, 2007, 46, 14487.
This journal is © The Royal Society of Chemistry 2018
Green Chem.

View Article Online
Communication
Green Chemistry
14 J. Kuver, Y. Xu and J. Gibson, Arch. Microbiol., 1995, 164, 337.
15 J. Sandy, A. Mushtaq, S. J. Holton, P. Schartau, M. E. Noble
and E. Sim, Biochem. J., 2005, 390, 115.
16 F. Dyda, D. C. Klein and A. B. Hickman, Annu. Rev. Biophys.
Biomol. Struct., 2000, 29, 81.
17 R. Nasuno, Y. Hirano, T. Itoh, T. Hakoshima, T. Hibi and
H. Takagi, Proc. Natl. Acad. Sci. U. S. A., 2013, 110, 11821.
18 L. K. Newby, et al., J. Lancet, 2014, 384, 1187.
19 T. Li, J. Liang, A. Ambrogelly, T. Brennan, G. Gloor,
G. Huisman, J. Lalonde, A. Lekhal, B. Mijts, S. Muley,
L. Newman, M. Tobin, G. Wong, A. Zaks and X. Zhang,
J. Am. Chem. Soc., 2012, 134, 6467; C. K. Savile, J. M. Janey,
E. C. Mndorff, J. C. Moore, S. Tam, W. R. Jarvis,
J. C. Colbeck, A. Krebber, F. J. Fleitz, J. Brands,
P. N. Devine, G. W. Huisman and G. J. Hughes, Science,
2010, 329, 305.
Green Chem.
This journal is © The Royal Society of Chemistry 2018