· Web viewEnzymatic C-H activation of aromatic compounds through CO 2 fixationGodwin A. Aleku 1,...
Transcript of · Web viewEnzymatic C-H activation of aromatic compounds through CO 2 fixationGodwin A. Aleku 1,...
Enzymatic C-H activation of aromatic compounds
through CO2 fixationGodwin A. Aleku1, Annica Saaret1, Ruth T. Bradshaw-Allen1, Sasha R. Derrington1, Gabriel R.
Titchiner1, Irina Gostimskaya1, Deepankar Gahloth1, David A. Parker2, Sam Hay1, David
Leys1*
1Manchester Institute of Biotechnology, School of Chemistry, University of Manchester,
Manchester, UK.2 Innovation/Biodomain, Shell International Exploration and Production Inc., Westhollow
Technology Center, Houston, USA.
*corresponding author: [email protected]
The direct C-H carboxylation of aromatic compounds is an attractive route
to the corresponding carboxylic acids, but remains challenging under mild
conditions. It has been proposed that the first step in the anaerobic
microbial degradation of recalcitrant aromatic compounds is a UbiD-
mediated carboxylation. In this study, we use the UbiD-enzyme, ferulic acid
decarboxylase in combination with a carboxylic acid reductase to create
aromatic degradation inspired cascade reactions. These lead to efficient
functionalization of styrene through CO2 fixation. We reveal that rational
structure-guided laboratory evolution can expand the substrate scope,
resulting in activity on a range of mono- and bi-cyclic aromatic compounds
through a single mutation. Selected variants demonstrate a 150-fold
improvement in the coumarillic acid to benzofuran + CO2 conversion and
unlock reactivity towards naphthoic acid. Our data demonstrates that
UbiD-mediated C-H activation is a versatile tool for the transformation of
aryl/alkene compounds and CO2 into commodity chemicals.
Introduction
Carboxylic acids can be prepared through direct C-H carboxylation of aromatic
compounds, a synthetically challenging reaction under mild conditions [1-4].
Enzymatic carboxylation has recently emerged as an attractive alternative and
green route to produce carboxylic acids, frequently making use of
(de)carboxylases involved in secondary metabolism [5]. Unfortunately, many
(de)carboxylases are restricted in substrate scope, and efficient carboxylation
requires high concentrations of [CO2] (Fig 1a). In contrast, the widespread UbiD
family of enzymes offers a particularly diverse range of natural substrates,
ranging from cinnamic acid and related compounds to furan or indole-containing
heteroaromatic carboxylic acids [6]. The general reaction mediated by UbiD
enzymes, the interconversion of alkene/aryl + CO2 with the corresponding
unsaturated/aromatic carboxylic acid, is mediated by the recently described
prenylated FMN cofactor (prFMN, [7]; Fig 1b).
The UbiD cofactor is synthesised by the associated UbiX, a flavin
prenyltransferase that extends the isoalloxazine FMN ring system through
prenylation with a 4th non-aromatic ring [8,9]. Oxidative maturation of the UbiX
product leads to formation of the prFMNiminium form of the cofactor. This species
has azomethine ylide characteristics, and has been proposed to form
cycloadducts with dipolarophile species such as , -α β unsaturated carboxylic acids
[10]. Indeed, a reversible 1,3 dipolar cycloaddition process has been
demonstrated to underpin the decarboxylation of cinnamic acid [11] (Fig 1b). It
is unclear to what extent the unusual 1,3 dipolar cycloaddition-based mechanism
is used in case of aromatic substrates, given the inherent barrier presented for
cycloaddition in such compounds [12]. Indeed, alternative proposals for prFMN-
based covalent catalysis have been put forward in case of activated aromatic
substrates [13,14].
While the majority of UbiDs act as decarboxylases, a subset is proposed to
function in the carboxylative direction at ambient [CO2], such as the Thauera
aromatica phenolphosphate carboxylase [15]. This enzyme converts the
phosphorylated phenol substrate to 4-hydroxybenzoic acid, thus coupling
carboxylation to dephosphorylation through an unknown mechanism. Arguably,
the most challenging reaction proposed to be catalyzed by UbiD enzymes is the
activation of highly stable aromatic compounds such as naphthalene and
benzene, a reversible process that is coupled to a second, irreversible step by
conversion of the aromatic carboxylic acid product into the corresponding acyl-
CoA thioester ([16,17]; Fig 1c). This process is believed to be essential to
anaerobic degradation of these compounds by some microbial cultures, but
corresponding enzyme activity has yet to be conclusively demonstrated.
The chemistry that underpins microbial benzene/naphthalene degradation has
inspired us to design and develop cascade reactions enabling efficient
functionalization of terminal alkenes to the corresponding aldehyde, alcohol,
amide or amine derivatives through ambient CO2 fixation. We sought to mimic
and validate the proposed pathway for microbial anaerobic degradation of
benzene/naphtalene by coupling the model UbiD enzyme Aspergillus niger
ferulic acid decarboxylase (AnFdc) (which catalyses cinnamic acid to styrene +
CO2 interconversion [10,11]) with carboxylic acid reductase (CAR) [18]. The
latter is capable of catalysing the irreversible conversion of aromatic carboxylic
acids to the corresponding aldehydes in an ATP- and NADPH-dependent manner
(Fig 1c). We demonstrate that efficient functionalization of styrene can be
achieved through CO2 fixation at ambient conditions through the use of Fdc/CAR
cascade reactions. We also establish that structure-guided laboratory evolution
can expand the substrate scope of ferulic acid decarboxylases, resulting in
activity on a range of mono- and bi-cyclic aromatic compounds. This extends the
product profile of the Fdc/CAR system enabling the C-H activation of
(hetero)aromatic compounds.
Results
Coupling AnFdc to CAR results in efficient CO2 fixation - To test whether
combination of a UbiD-enzyme with CAR could yield effective carboxylation
under ambient conditions, we used wild-type AnFdc [10,11] with styrene as the
model substrate. In absence of CAR, incubation of purified AnFdc or lyophilised
E. coli whole cells containing AnFdc with 5 mM styrene 1 in the presence 0.5-1 M
KHCO3 yielded cinnamic acid 2, albeit in low conversion (<2%). In contrast, a
one-pot reaction containing purified Tsukamurella paurometabola carboxylic
acid reductase (TpCAR; [19]) and AnFdc, converted 5 mM styrene 1 to
cinnamaldehyde 3 with up to 25% conversion via the cinnamic acid intermediate
(in presence of 0.05-0.5 M KHC03 or (NH4)HCO3 and with supply of
stoichiometric amounts of NADPH and ATP). Given that the biotransformation of
the intermediate cinnamic acid 2 using E. coli whole cells expressing TpCAR
afforded 97% conversion to cinnamyl alcohol 4, (due to endogenous alcohol
dehydrogenase (EcADH) activity, Extended Data Figs 1, 2), we sought to
implement the three step AnFdc-TpCAR-EcADH cascade (Fig 2). In order to
obtain high conversion of styrene 1 to cinnamyl alcohol 4, these catalysts were
assembled in one-pot, and conversion in the presence of KHCO3 was monitored.
In this case, AnFdc was utilised as purified enzyme or lyophilised cell free extract
(CFE), whereas TpCAR/EcADH were employed using E. coli whole-cell
preparation. This revealed the formation of cinnamyl alcohol as the major
product, with cinnamaldehyde and cinnamic acid only detected in trace amounts
(Fig 2, Extended Data Fig 2a). Furthermore, optimum conversion levels were
obtained at 40-60 mM KHCO3, revealing full conversion occurs near
stoichiometric molar ratios of styrene and KHCO3 (as the CO2 source; Extended
Data Fig 1b). Biotransformation conversions were similar when using AnFdc as
lyophilised CFE preparation or when using related homologues such as
Saccharomyces cerevisiae (ScFdc; Extended Data Fig 1e).
Expanding the AnFdc - CAR product scope - To further demonstrate the
potential for formation of a wide range of compounds via AnFdc-catalysed CO2
fixation, we took advantage of a recently established protocol for CAR-mediated
NADPH-free amidation [20]. We optimized the NADPH-independent TpCAR-
catalysed amidation process for cinnamic acid 2 in 500 mM (NH4)HCO3, leading
to the formation of cinnamamide 5 from 2 in up to 75 % conversion (Extended
Data Fig 3a). By coupling purified preparations AnFdc with TpCAR in an in vitro
one-pot biotransformation reaction with stoichiometric amount of ATP in 500
mM (NH4)HCO3 buffer, styrene 1 was converted to cinnamamide 5 in up to 14 %
conversion (Fig 2). The inherent reactivity of the CAR aldehyde product offers
further opportunities to extend the product profile. In order to demonstrate
access to N-alkylated secondary amines, we linked the Fdc-CAR system to a
biocatalytic reductive amination step, employing Cystobacter ferrugineus imine
reductase (CfIRED; [21]) as the reductive aminase, while Segniliparus rugosus
CAR (SrCAR) ([22]) was utilised for the carboxylate reduction. Using 2.5
equivalents of cyclopropylamine as a primary amine source, we found the three
step AnFdc-SrCAR-CfIRED cascade afforded the secondary amine 6 from styrene
1 in up to 93% conversion (Fig 2, Extended Data Figs 2c, 3).
AnFdc is a weak heteroaromatic acid decarboxylase - Encouraged by the fact
the AnFdc-CAR system enables conversion of terminal alkenes such as styrene to
respectively aldehydes, amides, secondary amines or allylic alcohols via
stoichiometric carboxylation, we next aimed to extend our approach towards the
functionalisation of (hetero)-aromatic compounds which are prevalent medicinal
and synthetic scaffolds. Unfortunately many UbiD-enzymes that act on such
compounds are oxygen-sensitive and/or require cofactor reconstitution when
produced through heterologous expression [6,7,13,14]. We thus took advantage
of the robust nature of the AnFdc enzyme and set out to evaluate and expand the
scope for activity with (hetero)aromatic acids. We identified putative alternative
substrates of AnFdc by screening a compound library comprising structurally
diverse (hetero)aromatic acids (Extended Data Fig 4) for inhibitory effect on
the decarboxylation of cinnamic acid 2. Compounds that acted as competitive
inhibitors were rescreened for AnFdc decarboxylation activity under
biotransformation conditions, enabling the detection of very weak
decarboxylation activity of the wild-type AnFdc towards the heteroaromatic
acids 7 and 8 with conversion of up to 5% achieved (Fig. 3a). In contrast, no
such activity could be observed with compounds 9 and 10, even though these
compounds exhibited competitive inhibition towards decarboxylation of 2. This
data suggests that AnFdc can potentially mimic other UbiD-enzymes such as
those involved in benzene/napthalene degradation, if the activity with
(hetero)aromatic compounds could be improved. We employed semi-rational
laboratory evolution of AnFdc to explore whether the enzyme could be evolved
to readily accept (hetero)aromatic compounds, and thus demonstrate the
potential for C-H activation of recalcitrant molecules by the UbiD-enzyme family.
Structure of AnFdc heteroaromatic acid complexes – To provide detailed
understanding of what governs AnFdc substrate specificity with respect to
(hetero)aromatic compounds, and thus guide downstream semi-rational
laboraoty evolution experiments, we determined high-resolution structures of
the wild-type enzyme in complex with respectively the weak substrate 8 and the
inhibitors 9 and 10 (Fig 3b, Extended data Fig 5). Complexes were obtained by
soaking AnFdc crystals with 100 mM - 500 mM ligand stock. Comparison of the
various ligand complex structures with previously determined cinnamic acid
substrate complexes [11] reveals aromatic ligands bind at the active site in two
distinct binding modes, one of which resembles a catalytically relevant
conformation. In this case the ligand carboxylic acid moiety is located in the
active site CO2 / Glu282 binding pocket. This places the substrate Calpha in close
proximity of the prFMNiminium C1’ atom (at ~3.1 Å). Given that wild-type AnFdc
depends on reversible cycloaddition to catalyse interconversion of cinnamic acid
2 with styrene 1 and CO2 (Fig 1b), the relative position of both the substrate
Calpha and Cbeta carbon atoms with respect to the prFMN iminium cofactor C1’ and
C4a is crucial. In the case of compounds 8 and 9, the Cbeta is located further
away from the prFMN C4a compared to cinnamic acid substrates (3.8 Å
compared to 3.3 Å; no catalytic binding mode was observed for 10). Hence, the
weak activity observed with 7 and 8 could be accounted for through an
alternative mechanism only requiring formation of a single substrate-
prFMNiminium bond (i.e. Calpha-C1’), or because these compounds are able to
explore conformations that minimize the Cbeta-C4a distance. Such
conformations are likely to require concomitant motion of residues M283 and
I327 that restrict the access and mobility of the more bulky aromatic compounds
in the active site.
Evolution of AnFdc towards aromatic substrates – Using the insights gained
from the wt AnFdc structural studies, we evaluated the effect of M283 and I327
on ligand position/conformational freedom on the desired reactivity. Using a
semi-rational laboratory evolution approach, we constructed NDT libraries of
M283 and I327 and assessed in vivo activity for the decarboxylation of 7 or 8 by
reversed-phase HPLC. Five variants (bearing bulky hydrophobic residues at
M283 or small non-hydrophobic residues at I327) displayed improved activity
towards compound 7 with I327S and I327N variants affording 95% and 61%
conversion respectively when used as whole-cell preparation, and both variants
afforded >99% conversion when employed as purified enzyme preparation
(Extended data Fig 6). In order to validate and understand the effect of the
I327S mutation, we determined high resolution structures of I327S variants in
complex with 8 or 9. These structures confirm subtle changes have occurred in
the active site and relative position of ligand/prFMNiminium cofactor as a result of
the mutation (Fig 3c, Extended data Fig 5 ). The reduction in side chain size at
position 327 is accompanied by a minor rotation of the ligand in the active site,
reducing the Cbeta-C4a distance by ~0.1 Å to 3.7 Å. While this distance remains
larger than the 3.3 Å observed for the cinnamic acid (WT enzyme complex),
conformational heterogeneity in the position of the M283 side chain induced by
the I327S mutation is likely to offer opportunity for further transient rotational
movement of the substrate along the Calpha-C1’ axis, and hence additional
reduction of the Cbeta-C4a distance. Crucially, the I327S structure with 2-
naphthoic acid 10 reveals this compound is now able to occupy a catalytically
relevant conformation, with both Calpha-C1’ and Cbeta-C4a distances closely
resembling the cinnamic acid distances observed previously (respectively 3.0 Å
and 3.4 Å).
AnFdC I327S/N-catalysed aromatic decarboxylation - An assessment of the
substrate scope of AnFdc I327S and I327N variants against a wide range of
(hetero)aromatic carboxylic acids was performed in the decarboxylation
direction, revealing an extended substrate repertoire for both variants.
Conversion of >99% was achieved with 7, using the purified AnFdc I327S, and
this variant displayed a total turnover number of >5000 constituting a 150-fold
improvement in specific activity when compared to the wild-type (Fig 4a,
Extended data Fig 6). Furthermore, related compounds such as 8 and 9 were
decarboxylated by the I327S variant, affording conversion of 42% and 93%
respectively. In addition, a range of bicyclic and monocyclic heteroaromatic
(di)carboxylic acids were also decarboxylated (Fig 4a). The AnFdc variants
displayed excellent regioselectivity for oxygen and sulphur-containing
heterocycles (7, 9, vs 12, 13, Fig 4a, Extended data Fig 6) displaying clear
preference towards heterocyclic 2-carboxylic acids (e.g. 7, 9) and rejecting the
heterocyclic 3-carboxylate 12 and 13. Remarkably, 2-naphthoic acid 10 was
decarboxylated by AnFdc I32S/N albeit in low conversion (7 %), representing
the first example of an isolated enzyme capable of decarboxylating non-activated
aromatic carboxylic acids (Fig 4a, Extended Data Figs 6, Fig 7). This is in
agreement with the observed catalytically active binding mode as revealed from
structural studies, Fig. 3c.
DFT computational studies of aromatic decarboxylation - In order to gain
further mechanistic insight into decarboxylation of aromatic acids by AnFdc
I327S, we used density functional theory (DFT) calculations to study the
formation of the covalent adduct and subsequent decarboxylation reaction.
Similar to previous work [11], an active site ‘cluster’ model was generated based
on the crystal structure of AnFdc I327S with non-covalently bound 2-naphthoic
acid 10 (Extended Data Fig 9). Stable covalent substrate-prFMN adducts were
found for both coumarillic 7 and naphthoic acid 10 substrates, corresponding to
two distinct species prior to decarboxylation (Int1cyclo and Int1open) and the
decarboxylation product Int2CO2. In agreement with solution studies, the
potential energy profile for coumarillic acid decarboxylation is substantially
lower than for naphthoic acid (Extended Data Fig 9e). In both cases, the
cycloadduct Int1cyclo is substantially higher in energy than the Int1open species,
and decarboxylation proceeds from Int1open. This suggests that, instead of a
concerted cycloaddition between the prFMN azomethine ylide dipole and the
aromatic substrate leading to the formation of the pyrrolidine adduct Int1cyclo, a
stepwise process is followed, whereby only the Ca-C1’ bond is fully formed in the
Int1open adduct. While Int1open resembles a putative arenium ion or Wheland
intermediate following aromatic electrophilic addition [23], analysis of the
charge distribution suggests through-space interactions between the closely
stacked prFMN and substrate aromatic planes (Cβ-C4a distance for the Int1open is
2.73 Å for 7 and 2.77 Å for 10) appear to avoid significant charge separation
between the aromatic substrate and prFMN (Extended Data Fig 9e). Similar
through-space electronic interactions have previously been used to control
abiotic electrophilic aromatic substitutions [24].
I327S/N plus CAR affords aromatic C-H activation - The AnFdc I327S/N
variants thus offer the possibility of using Fdc-CAR cascade reactions to
functionalise aromatic compounds through CO2 fixation. We employed AnFdc
I327S for the functionalisation of benzofuran 18 via transient formation of 7 as a
model reaction (Fig 4b). We demonstrated reduction of benzofuran-2-carboxylic
(coumarillic) acid 7 using E. coli whole cells expressing TpCAR afforded 1-
benzofuran-2-ylmethanol 20. To form the three step AnFdc I327S-TpCAR-EcADH
cascade reaction, purified AnFdc I327S or CFE was combined with fresh resting
E. coli whole-cells expressing TpCAR. This system converted benzofuran 18 to 1-
benzofuran-2-ylmethanol 20 with supply of 60-100 mM KHCO3 (Fig 4b,
Extended Data Fig 6c). Similarly, the use of purified AnFdc I327S and TpCAR or
SrCAR in (NH4)HCO3 reaction buffer leads to conversion of 18 to the
corresponding amide 21 in up to 18 % conversion (~1% of the carboxylic acid
intermediate 7 was detected). Finally, the use of a three step AnFdc, I327S-SrCAR
and reductive aminase from Aspergillus oryzae (AspRedAm; [25]) leads to
approximately 22% of the substrate converted, affording the corresponding
secondary amine 22 (13 %) (Fig 4b, Extended Data Fig 8).
Discussion - Taken together, our data confirms UbiD-mediated
(de)carboxylation can readily occur for a wide range of substrates, including
heteroaromatic and non-activated aromatic compounds. While the reaction
equilibrium is poised towards decarboxylation at ambient conditions, carboxylic
acids formed can be irreversibly converted to a range of C+1 compounds using
cascade reaction systems, thus achieving efficient CO2 fixation and substrate
activation. Hence, this confirms a similar system can operate in those microbial
cultures degrading benzene/naphthalene under anaerobic conditions that
contain suitable UbiD carboxylases [16,17]. While the aromatic stability of these
substrates renders them recalcitrant to transformation, prFMNiminium mediated
carboxylation offers a route to activation of these compounds. The energetic
barriers imposed by such transformation might offer insights into the
remarkably slow growth rates of such cultures. The one-pot biocatalytic
cascades developed enable the functionalisation of unactivated terminal alkenes
such as styrene derivatives or (hetero)aromatic compounds via carboxylation at
near stoichiometric supply of CO2. When using CAR mediated conversion as the
irreversible second step, the inherent reactivity of the aldehyde generated offers
opportunities to access an extensive product range. Given the broad substrate
tolerance of CAR enzymes towards , -α β unsaturated carboxylic acids as well as
(hetero)aromatic acids [26] is complementary to the substrate scope of
(evolved) Fdc, the Fdc-CAR system can be exploited to provide access to a broad
range of industrially important products. Additionally, it is likely that coupling
Fdc-mediated CO2-fixation to other irreversible and synthetically useful
enzymatic reactions such as enoate reduction [27] or CoA-ligation [28] can
further extend the scope for UbiD-mediated CO2 transformation into commodity
chemicals. This approach expands the repertoire of existing biocatalytic
functionalisation of alkenes [29], and contributes a new route to using CO2 as a
chemical building block [30,31].
Acknowledgements
This work was supported by the grants BBSRC BB/K017802 (DL) and ERC pre-
FAB 695013 (DL). We acknowledge assistance via use of the Manchester Protein
Structure Facility and Diamond Light Source for access (proposal numbers
MX12788 and MX17773) which contributed to the results presented here. We
also acknowledge the assistance given by IT Services and the use of the
Computational Shared Facility at The University of Manchester. We thank James
Marshall and Nicholas J. Turner (Manchester) for kindly providing the CfIRED
plasmid. D.L. is a Royal Society Wolfson Merit Award holder.
Author contributions
G. A. A & R. T. B. performed enzyme inhibition studies, mutagenesis and protein
engineering experiments. G. A. A. & I. G. performed crystallisation and D.L.
determined associated crystal structures. G. A. A., S. R. D. & D. G. identified
suitable carboxylic acid reductases. G. A. A., S. R. D. & G. R. T. identified suitable
reductive aminases. G. A. A. & R. T. B. performed decarboxylation
biotransformations. G. A. A. constructed CO2-fixation cascades, and performed
and analysed cascade biotransformations. A.S. & S.H. performed computational
studies. G. A. A. & D. L. designed experiments and wrote initial draft of the
manuscript. D.P. advised on all aspects. All authors discussed the results and
commented on the manuscript. D. L. initiated and coordinated the project.
Competing interests
The authors declare no competing interests.
References
1. Banerjee, A., Dick, G.R., Yoshino, T. & Kanan, M.W. Carbon dioxide utilization via carbonate
promoted CH-carboxylation. Nature 531, 215-219 (2016).
2. Luo, J., & Larrosa, I. C-H carboxylation of aromatic compounds through CO2 fixation.
ChemSusChem 10, 3317-3332 (2017).
3. Hong, J., Li, M., Zhang, J., Sun, B., & Mo, F. C-H bond carboxylation with carbon dioxide.
ChemSusChem 12, 6-39 (2019).
4. Xiao, D.J. et al. A closed cycle for esterifying aromatic hydrocarbons with CO2 and alcohol. Nat.
Chem. 11, 940-947 (2019).
5. Payer, S.E., Faber, K. & Glueck, S.M. Non-oxidative enzymatic (de)carboxylation of
(hetero)aromatics and acrylic acid derivatives. Adv. Synth. Cat. 361, 2402-2420 (2019).
6. Marshall, S.A., Payne, K.A.P & Leys, D. The UbiX-UbiD system: the biosynthesis and use of
prenylated flavin (prFMN). Arch. Biochem. Biophys. 632, 201-221 (2017).
7. Leys, D. Flavin metamorphosis: cofactor transformation through prenylation. Curr. Opin. Chem.
Biol. 47, 117-125 (2018).
8. White M.D, et al. UbiX is a flavin prenyltransferase required for bacterial ubiquinone
biosynthesis. Nature 522, 502-506 (2015)
9. Marshall S.A., et al. The UbiX flavin prenyltransferase reaction mechanism resembles class I
terpene cyclase chemistry. Nat Comms 10, 2357 (2019)
10. Payne K.A.P. et al. New cofactor supports -unsaturated acid decarboxylation via 1,3
dipolar cycloaddition. Nature 522, 497-501 (2015).
11. Bailey S.S. et al. Enzymatic control of cycloadduct conformation ensures reversible 1,3
dipolar cycloaddition in a prFMN dependent decarboxylase. Nat. Chem. 11, 1049-1057 (2019).
12. Baunach M, & Hertweck, C. Natural 1,3-dipolar cycloadditions. Angew. Chem. Int Ed 54,
12550-12552 (2015)
13. Payer, S.E. et al. Regioselective para-carboxylation of catechols with a prenylated flavin
dependent decarboxylase. Angew. Chem. Int. Ed. 56, 13893-12897 (2017).
14. Payne, K.A.P. et al. Enzymatic carboxylation of 2-furoic acid yields 2,5-furandicarboxylic acid
(FDCA). ACS Catalysis 9, 2854-2865 (2019).
15. Schuhle, K., & Fuchs, G. Phenylphosphate carboxylase: a new C-C lyase involved in anaerobic
phenol metabolism in Thauera aromatica. J. Bacteriol. 186, 4556-4567 (2004).
16. Meckenstock, R.U. et al. Anaerobic degradation of benzene and polycyclic aromatic
hydrocarbons. J. Mol. Microbiol. Biotechnol. 26, 92-118 (2016).
17. Luo, F. et al. Metatranscriptome of an anaerobic benzene-degrading, nitrate reducing
enrichment culture reveals involvement of carboxylation in benzene ring activation. Appl.
Environ. Microbiol. 80, 4095-4107 (2014).
18. Winkler M. Carboxylic acid reductase enzymes (CARs). Curr. Opin. Chem. Biol. 43, 23-29
(2018).
19. Finnegan W. et al. Characterization of carboxylic acid reductases as enzymes in the toolbox
for synthetic chemistry. ChemCatChem 9, 1005-1017 (2017).
20. Wood, A.J.L. et al. Adenylation activity of carboxylic acid reductases enables the synthesis of
amides. Angew. Chem. Int. Ed. 56, 14498-14501 (2017).
21. Zawodny, W. et al. Chemoenzymatic synthesis of substituted azepanes by sequential
biocatalytic reduction and organolithium-mediated rearrangement. J. Am. Chem. Soc. 140, 17872-
17877 (2018).
22. Gahloth, D. et al. Structures of carboxylic acid reductase reveal domain dynamics underlying
catalysis. Nat. Chem. Biol. 13, 975 (2017).
23. Olah, G.A. Aromatic substitution. XXVIII. Mechanism of electrophilic aromatic substitutions.
Acc. Chem. Res. 4, 240-248 (1971).
24. Murphy, K.E. et al. Precise through-space control of an abiotic electrophilic aromatic
substitution reaction. Nat. Comm. 8, 14840 (2017).
25. Aleku, G.A. et al. A reductive aminase from Aspergillus oryzae. Nat. Chem. 9, 961-969 (2017)
26. Aleku, G.A., Roberts, G.W., & Leys, D. Biocatalytic reduction of alpha,beta-unsaturated
carboxylic acids to allylic alcohols. Green Chemistry DOI:10.1039/D0GC00867B (2020)
27. Winkler C.K, Faber K, & Hall. M. Biocatalytic reduction of activated C-C bonds and beyond:
emerging trends. Curr. Opin. Chem. Biol. 43, 97-105 (2018).
28. Villemur R. Coenzyme A ligases involved in anaerobic biodegradation of aromatic
compounds. Can J. Microbiol. 41, 855-861 (1995).
29. Wu, S., Zhou, Y. & Li Z. Biocatalytic selective functionalization of alkenes via single-step and
one-pot multi-step reactions. Chem. Commun. 55, 883-896 (2019).
30. Dabral, S. & Schaub, T. The use of carbon dioxide (CO2) as a building block in organic synthesis
from an industrial perspective. Adv. Synt Catal. 361, 223-246 (2019).
31. Hepburn, C. et al. The technological and economic prospects for CO2 utilization and removal.
Nature 575, 87-97 (2019).
Figure legends
Fig 1. UbiD (de)carboxylases offer new routes to unsaturated carboxylic
acid derivates. a) Previous biocatalytic carboxylation work using large molar
excess of KHCO3 and applying enzymes with narrow substrate scope ([5]; i.e.
phenolic acid decarboxylase, PAD and 2,3-dihydroxybenzoic acid decarboxylase,
DHBD). b) A chemical scheme of a model UbiD reaction, interconversion of
styrene + CO2 with cinnamic acid, using covalent catalysis by the prFMNiminium
cofactor mediated by a reversible 1,3 dipolar cycloaddition process [10,11]. c)
The proposed pathway for microbial anaerobic benzene/naphthalene
degradation mediated through initial UbiD mediated carboxylation coupled to
irreversible CoA-ester production by benzoyl-CoA ligase (BCL) [16,17].
Fig 2. CO2-fixation synthetic cascades linking AnFdc and CAR-catalysed
reactions for the functionalisation of terminal alkenes. One-pot reaction
containing AnFdc and resting E. coli whole-cells expressing TpCAR converts
styrene 1 to cinnamyl alcohol 4 (95% conversion) via AnFDc catalysed
carboxylation, CAR-catalysed carboxylate reduction and a carbonyl reduction
step catalysed by E. coli endogenous ADH (EcADH). A one-pot two-enzyme in
vitro cascade containing AnFDC and CAR, and excluding NADPH afforded the
conversion of styrene 1 to cinnamide 5 (14% conversion) via cinnamic acid 2.
An alternative one-pot three-step cascade applying the CfIRED imine reductase
to catalyse the reductive coupling of the resulting aldehyde to a primary amine
provides access to the secondary amine 6 (93% conversion). Conversion values
were determined from HPLC/GC-MS analyses of biotransformation reactions.
Biotransformation products (2, 3, 4, 5) were confirmed by HPLC analyses
against commercial authentic reference standards; product 6 was confirmed by
analysis of the MS spectra.
Fig 3. Structure guided laboratory evolution of AnFdc. a) Wild-type AnFdc
demonstrates weak decarboxylation activity with hetero-aromatic bicyclic
compounds 7 and 8, but not 9 and 10. b) Overlay of the crystal structures of
AnFdc WT with compounds 8 and 10 and cinnamic acid 2 (individual structures
and omit electron densities are shown in Extended Data Fig 6). Key amino acids
are shown in atom coloured sticks (green carbons), while compound 8 (cyan
carbons) is shown in the catalytically active conformation (grey carbons),
compound 10 (magenta carbons) binds in a non-catalytic mode. Grey dotted
lines indicate hydrogen bonds established with the carboxylic acid moiety of
compound 8, while red dotted lines indicate the distance between the prFMN C1’
and C4a atoms and the compound 8 Calpha and Cbeta. Residues M283 and I327
restrict compound 8 from binding in a conformation whereby the Calpha-beta
double bond overlays with the corresponding cinnamic acid Calpha-Cbeta atoms.
c) Overlay of the crystal structures of AnFdc I327S with compounds 9 and 10
(individual structures and electron densities are shown in Extended Data Fig 6).
Key amino acids are shown in atom coloured sticks (green carbons), while
compound 9 (blue carbons) is shown in the catalytically active I327S
conformation as well as the conformation observed in the WT structure (in grey
carbons), compound 10 (magenta carbons) now binds in a catalytic mode as
compared to the corresponding WT complex (see panel b). Grey dotted lines
indicate hydrogen bonds established with the carboxylic acid moiety of
compound 10, while red dotted lines indicate the distance between the prFMN
C1’ and C4a atoms and the ligand 10 Calpha and Cbeta. The latter are near
identical to the previously determined relative position of cinnamic acid-prFMN
[11].
Fig 4. Enzymatic C-H activation of (hetero)aromatic compounds by the I327S
AnFdc – CAR system a) Expanded substrate panel for AnFdc variant I327S. b)
The use of I327S AnFdc in cascade reactions similar to those shown in Fig 2
allows enzymatic activation of benzofuran 18 through CO2 fixation at ambient
conditions. Conversion values were determined from HPLC/GC-MS analyses of
biotransformation reactions. Biotransformation products were confirmed by
HPLC/GC-MS analyses against commercial authentic reference standards, except
for product 22 which was confirmed by analysis of the MS spectra.
Fig 1
Fig 2.
Fig 3.
Fig 4.
Online Methods
No statistical methods were used to predetermine sample size. The experiments were
not randomized, and investigations were not blinded to allocation during experiments
and outcome assessment.
Chemicals and reagents
Chemicals and reagents including all substrates and reference product standards
(except compounds 6 and 22) of highest purity were purchased from Sigma-Aldrich
(Poole, Dorset, UK), or Fluorochem (Hadfield, Derbyshire, UK). HPLC solvents were
obtained from Sigma-Aldrich (Poole, Dorset, UK) or ROMIL (Waterbeach, Cambridge,
UK) and GC gases from BOC gases (Guildford, UK). Enzyme nicotinamide cofactors
NADP+ and NADPH were purchased from Bio Basic (Markham, Ontario, Canada), and
Adenosine triphosphate (ATP) was sourced from Sigma-Aldrich (Poole, Dorset, UK).
Production and preparation of AnFdcUbiX biocatalysts
Cloning, expression and purification of AnFdcUbiX were performed according to
previously described methods [10,32] and AnFdcUbiX whole-cell biocatalyst was prepared
as described previously [32].
Preparation of CARSfp whole-cell Biocatalyst
Cloning of carboxylic acid reductases (TpCAR and SrCAR) and phosphopantetheine
transferase (Sfp) was performed as previously reported [22]. A single colony of
recombinant E. coli BL21 (DE3) containing pET28-b-FL-TpCAR and Sfp from Bacillus
subtilis (pCDF1b-Sfp) was inoculated into 10 mL lysogeny broth (LB) (1% tryptone,
0.5% yeast extract, 1% NaCl) and incubated overnight at 37 °C in an orbital shaker at
200 rpm. This starter culture was used as the inoculum. A 2 L flask with 500 mL LB was
supplemented with kanamycin (30 g mLμ -1) and streptomycin (50 g mLμ -1) and
inoculated with 5 mL of starter culture. Cultivation was performed at 37 °C in an orbital
shaker with 200 rpm shaking. At an optical density (OD600 nm) of between 0.6 and 0.8,
isopropyl -D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 0.2β
mM to induce protein expression. Incubation was continued at 20 °C and 200 rpm for 18
h. Cells from a 500 mL culture were harvested by centrifugation at 6000 rpm for 15 min,
and washed in sodium phosphate buffer (NaPi, 50 mM, pH 7.5). Harvested cells
containing CARSfp biocatalyst system were used as fresh resting whole-cell biocatalyst or
purified according to previously reported method [22].
Biotransformation procedures for Fdc-catalysed decarboxylation
(i) For Fdc-catalysed decarboxylation of acrylic acid derivatives, e.g. cinnamic acid 2
using purified holo-Fdc, a 500 µL reaction mixture contained cinnamic acid 5 mM (or up
to 20 mM), 2% (v/v) dimethylsulfoxide (DMSO) and 0.2-1mg mL -1 of purified Fdcubix in
NaPi (50 mM or 100mM, pH 6).
(ii) For Fdc-catalysed decarboxylation of (hetero)aromatic carboxylic acid, e.g.
benzuran-2-carboxylic acid 7, a 500 µL reaction contained 0.5-1mg AnFdc I327S/N,
(hetero)aromatic carboxylic acid 2 mM (or up to 20 mM for benzofuran-2-carboxylic
acid), 2% (v/v) dimethylsulfoxide (DMSO) in NaPi (50 mM, pH 6).
(iii) For Fdc-catalysed decarboxylation reaction using the whole-cell biocatalyst, a
typical 500 µL reaction mixture contained 5 mM carboxylic acid substrate, 2% (v/v)
DMSO and fresh resting E. coli whole cells containing the overexpressed FdcUbiX (to a
final optical density, OD 600 nm of 30), or 30 mg of the lyophilized whole-cell biocatalyst
in NaPi (50 mM, pH 6).
Reaction mixture in a 2 mL tightly closed glass vials were incubated at 30 °C with
shaking at 250 rpm for 18 h. The reaction was stopped by addition of 3 volumes of
acetonitrile and vigorous mixing. The reaction mixtures were centrifuged (15°C, 13 000
rpm, 20 min); the clear supernatant was further centrifuged and analysed by RP-HPLC
(Agilent system,Santa Clara, CA, USA, equipped with a G1379A degasser, G1312A binary
pump, a G1367A well plate autosampler unit, a G1316A temperature-controlled column
compartment and a G1315C diode array detector). ARP-HPLC column, syncronis; C18;
250 mm length, 4.6 mm diameter, 5 m particle size (Thermo Scientific; Waltham, MAμ
USA) was used.
Where analysis of biotransformations were performed on the GC-MS, an equal volume of
EtOAc (containing a known concentration of an internal standard where necessary) was
added to biotransformation mixture, vigorously mixed, centrifuged and the organic
layer was extracted. The aqueous layer was then acidified to a pH of ~2 and further
extracted with EtOAc and centrifugation (15 °C, 13 000 rpm, 10 min) to improve the
separation of phases. The organic layers were combined and dried over anhydrous
MgSO4 and samples were analysed by GC-MS (Agilent 5977A Series GC/MSD System
with an Agilent 7890B Series GC coupled to Mass Selective Detector. Column: 30 m DB‐WAX column with 0.25 mm inner diameter and 0.25 m film thickness (Agilent, Santaμ
Clara, CA, USA) was used. Analysis method: Inlet temperature: 240 °C, detector
temperature: 250 °C, MS source 230 °C, helium flow: 1.2 mL min−1; oven temperature
40–240 °C, 15 °C min−1).
Construction and screening of AnFdc I327NDT and M283NDT libraries
Site-saturation mutant libraries were generated using the following overlapping primer
sets: AnFdc I327NDT, forward: CCCATACGATGNDTGGCAGCCTGGCC (5'>3'); reverse:
GGCCAGGCTGCCAHNCATCGTATGGG (5'>3'). AnFdc M283NDT, forward:
GGCCCGTTTGGTGAANDTCATGGCTATATTTTC (5'>3'); reverse:
GAAAATATAGCCATGAHNTTCACCAAACGGGCC (5'>3').
E. coli NEB 5-alpha competent cells (New England Biolabs, Hitchin, UK) were
transformed with Dpn1-treated PCR products. Six colonies were sequenced from each
library to ascertain even distribution of mutations. Plates containing >60 colonies were
washed with 5 mL LB into a falcon tube and the plasmids were isolated using Qiagen
miniprep kits. The extracted plasmid mixture was used to transform E. coli BL21 (DE3)
cells. Protein expression was performed in deep-well plates; each well contained a
single E. coli BL21 (DE3) colony (bearing AnFdc variant in pET-30a (+) vector) in 800 Lμ
LB supplemented with kanamycin (30 g mLμ -1). The culture was then incubated at 37 °C
in an orbital shaker at 300 rpm, induced at OD600 nm between 0.6 and 0.8 with IPTG (final
concentration of 0.2 mM). Further cultivation was continued at 20 °C, 300 rpm shaking
for 18 h. Cell cultures were harvested at 4000 rpm for 20 min, washed with NaPi (50
mM, pH 7.5) and were used for biotransformations as fresh resting cells.
Biotransformation reactions was performed in deep-well plates; 250 L NaPμ i buffer (50
mM, pH 6) containing 2 mM substrate and 2 % (v/v) DMSO was added to pellets in each
well, sealed and incubated at 30 °C with 250 rpm shaking for 18 h, after which the
reaction was stopped by addition of 3 volumes of MeCN. Product formation was
monitored using RP-HPLC. Five mutants (displaying best activity) identified from the 36
transformants initially screened were further rescreened using cell-free extract.
Variants of interests were purified and further characterised.
Identification of competitive inhibitors and alternative substrates
A compound library containing aromatic and heteroaromatic carboxylic acids was
constructed. The inhibitory effect of these compounds on the AnFdc-catalysed
decarboxylation of cinnamic acid 2 was investigated; AnFdcUbiX (0.1 mg mL-1) was
incubated with 5 mM 2 in the presence of 0-20 mM inhibitor (in 100-200 mM NaPi
buffer, pH 7.5), and the conversion of 2 was determined after 30 min incubation.
(Hetero)aromatic carboxylic acids that inhibited the decarboxylation of 2 were
rescreened under 24 h biotransformation conditions to assess for decarboxylation
product. In all cases, control reactions lacking AnFdcUbiX were performed.
Biotransformation procedures for the conversion of cinnamic acid to cinnamyl alcohol
Conversion of cinnamic acid 2 to cinnamyl alcohol 4 via the intermediate
cinnamaldehyde 3 was achieved using recombinant resting E. coli cells expressing CARsfp
and containing an endogenous alcohol dehydrogenase (EcADH); glucose was added to
enable in situ generation of cofactors. A typical 500 uL reaction mixture contained 5 mM
cinnamic acid 2, 5% v/v DMSO, E. coli resting whole-cells containing over-expressed
CARsfp to the final OD600 nm of 30 in NaPi or Tris-HCl (50 mM, pH 7.5). Reaction mixtures in
2 mL Eppendorf tubes were incubated at 30 °C with 250 rpm shaking for 18 h, after
which samples were prepared for RP-HPLC.
One-pot biotransformation for the conversion of styrene to cinnamaldehyde via CO2-
fixation
For Fdc-CAR mediated conversion of styrene 1 to cinnamaldehyde 3 in one pot, a typical
1 mL reaction mixture contained 5 mM styrene 1, 2% v/v DMSO, purified FdcUbiX (1-2mg
mL-1), purified CARsfp (0.6-1 mg mL-1), 10-200 mM KHCO3, 10 mM NADPH and 10 mM
ATP, 15 mM MgCl2 in Tris-HCl buffer (50 mM, pH 7.5). Reaction mixtures in 2 mL tightly-
closed glass vials were incubated at 30 °C with 250 rpm shaking for 18 h, after which the
enzyme was inactivated by the addition of 3 volumes of MeCN, vigorously mixed and
samples were prepared for RP-HPLC.
One-pot biotransformation for the conversion of alkene or heteroaromatic to the
corresponding alcohol via CO2-fixation
For Fdc-CAR-ADH mediated conversion of styrene 1 to cinnamyl alcohol 4 in one pot, a
typical 1 mL reaction mixture contained 5 mM 1, 2% v/v DMSO, AnFdcUbiX either as
purified form (1-2 mg mL-1) or lyophilized cell free extract (10mg), CARsfp as fresh
resting whole-cells preparation (to the final OD 600 nm of 50), 20 mM D-glucose and 10-
200 mM KHCO3 in NaPi (50 mM, pH 7.5). Reaction mixtures in 2 mL tightly-closed glass
vials were incubated at 30 °C with 250 rpm shaking for 18 h, after which the enzyme
was inactivated by the addition of two volumes of MeCN and vigorously mixed. The
reaction mixtures were centrifuged (15 °C, 13 000 rpm, 10 min); the clear supernatant
was filtered and analysed by RP-HPLC.
One-pot biotransformation for the conversion of terminal alkene or heteroaromatic to the
corresponding amine via CO2-fixation
For the one-pot, three-step Fdc-CAR-RedAm-mediated conversion of styrene 1 to the
corresponding N-alkylated cinnamylamine 6, a typical 1 mL reaction mixture contained
5 mM 1, 5% v/v DMSO, purified AnFdcUbiX (1.5 mg mL-1), purified SrCAR (0.6 mg mL-1 ),
0.8 mg mL-1 purified imine reductase/reductive aminase, 12.5 mM cyclopropylamine,
10 mM ATP, 15 mM MgCl2, 20 mM D-glucose , 1 mM NADP+, 0.2 mg mL-1 purified GDH, in
100 mM or 500mM (NH4)HCO3 buffer, pH8. Reaction mixtures in 2 mL tightly closed
glass vials were incubated at 30 °C with 250 rpm shaking for 18 h. The reaction was then
basified with NaOH to pH >12; 500 uL of EtoAc added, vigorously mixed, centrifuged (15
°C, 13 000 rpm, 5 min); and organic layer collected. The remaining aqueous layer was
then acidified to pH ~2 and further extracted with EtOAc with centrifugation (15 °C, 13
000 rpm, 10 min). The organic layers were combined and dried over anhydrous MgSO4
and samples were analysed by GC-MS. The same reaction conditions were used for the
conversion of benzofuran 18 to the corresponding N-alkylated amine derivative 22.
Products 6 and 22 obtained from analytic biotransformation reactions were confirmed
from analysis of the MS spectra. GCMS analyses were performed on Agilent 5977A
Series GC/MSD System with an Agilent 7890B Series GC coupled to Mass Selective
Detector. Data analysis was performed using GC/MSD MassHunter Data Acquisition and
ChemStation Data Analysis. A 30 m x 0.25 mm x 0.1 m VF-5HT column (Agilent, Santaμ
Clara, CA, USA) was used. The parameters of the method include: Inlet temperature =
240C, detector temperature = 250C, MS source= 230 C, helium flow = 1.2 mL min-1;
oven temperature between 50 - 360C, 30C min-1.
One-pot biotransformation for the conversion of terminal alkene or heteroaromatic to the
corresponding amide via CO2-fixation
For the one-pot, two step carboxylation-amidation cascade mediated by Fdc-CAR system
for the conversion of styrene 1 to the corresponding cinnamide 5, a typical 1 mL
reaction mixture contained 5 mM 1, 2% v/v DMSO, purified AnFdcUbiX (1-2 mg mL-1),
purified SrCAR (0.6 mg mL-1 ), 10 mM ATP, 15 mM MgCl2, in 500 mM (NH4)HCO3 buffer,
pH 8. Reaction mixtures in 2 mL tightly closed glass vials were incubated at 30 °C with
250 rpm shaking for 18 h. The reaction was then basified with NaOH to pH >12; 500 uL
of EtoAc added, vigorously mixed, centrifuged (15 °C, 13 000 rpm, 5 min); and organic
layer collected. The remaining aqueous layer was then acidified to pH ~2 and further
extracted with EtOAc with centrifugation (15 °C, 13 000 rpm, 10 min). The organic
layers were combined and dried over anhydrous MgSO4 and samples were analysed by
GC-MS.
Protein crystallization and structure determination
Crystallisation was performed by sitting drop vapor diffusion using a seed stock based
on previously established crystallization conditions for AnFdc ([10]; 0.2 M potassium
thiocyanate, Bis-Tris propane pH 6.5, 20% w/v PEG 3350 by mixing 0.05 L seed stock,μ
0.25 L protein solution and 0.3 L reservoir solution at 4 °C). Where FMN containingμ μ
AnFdc was used, production, purification and crystallisation were performed as
previously described [11]. Crystals were cryoprotected in reservoir solution
supplemented with 10% PEG 200 and flash cooled in liquid nitrogen. To obtained co-
crystals, the protein was incubated with 1-5 mM ligand prior to the crystallisation while
ligand soaks was performed in cryoprotectant solution supplemented with 100 mM-500
mM ligand stock solution in MeOH. Diffraction data were collected at Diamond
beamlines and data were processed using XDS [33]. Iterative refinement and manual
model building were performed with REFMAC5 and COOT [34]. Final data collection and
refinement statistics are given in Supplementary Table 1.
Computational studies
Active site DFT ‘cluster’ models comprising ~350 atoms (figure B) were built from the
crystal structure of AnFdc I327S with non-covalently bound naphthoic acid bound and
modelled at the B3LYP/6-31G(d,p) level of theory with the D3 version of Grimme’s
dispersion with Becke–Johnson damping [35] and a generic polarizable continuum with
= 5.7 using the polarizable continuum model. Calculations were performed usingε
Gaussian 09 revision D.01, the models and additional details are given in Extended
Data Fig. 9 and are similar to our previous work [11], but make use of a larger model
comprising more of the active site.
Data availabilityThe atomic coordinates and experimental data have been deposited in the
Protein Data Bank with accessions codes 6TIH, 6TIJ, 6TIO, 6TIN, 6TIL, 6TIB, 6TIC
and 6TIE. All other data are available from the corresponding author on
reasonable request.
Methods only References
32. Aleku, G.A. et al. Terminal alkenes from acrylic acid derivatives via non-oxidative enzymatic
decarboxylation by ferulic acid decarboxylases. ChemCatChem 10, 3736-3745 (2018).
33. Kabsch, W. XDS. Acta Crystallogr. D 66, 125–132 (2010).
34. Winn, M. D. et al. Overview of the CCP4 suite and current developments. Acta Crystallogr. D
Biol. Crystallogr. 67, 235–242 (2011).
35. Grimme, S., Ehrlich, S. & Goerigk, L. Effect of the damping function in dispersion corrected
density functional theory. J. Comp. Chem. 32, 1456–1465 (2011).