Rapid Biocatalytic Synthesis of Aromatic Acid CoA Thioesters utilizing Microbial Aromatic Acid CoA Ligases

Abstract

Chemically labile ester linkages can be introduced into lignin by incorporation of monolignol conjugates, which are synthesized in planta by acyltransferases that use a coenzyme A (CoA) thioester donor and a nucleophilic monolignol alcohol acceptor. The presence of these esters facilitates processing and aids in the valorization of renewable biomass feedstocks. However, the effectiveness of this strategy is potentially limited by the low steady-state levels of aromatic-acid thioester donors in plants. As part of an effort to overcome this, aromatic-acid CoA ligases involved in microbial aromatic degradation were identified and screened against a broad panel of substituted-cinnamic and benzoic acids involved in plant lignification. Functional fingerprinting of this ligase library identified four robust highly active enzymes capable of facile, rapid, and high-yield synthesis of aromatic-acid CoA thioesters under mild aqueous reaction conditions mimicking in planta activity.

Entry for the Table of Contents

Adenosine triphosphate (ATP)-dependent aromatic acid-Coenzyme A (CoA) ligases in microbial lignin degradation follow a different mechanism of CoA ligation than plant phenylpropanoid ligases contributing to lignification. Microbial ligases form an acyl-phosphate intermediate instead of acyl-adenylate and enable metabolism of the highly substituted sinapic acid. Such ligases have promise as tools for biocatalysis and lignin engineering.

Introduction

Lignin is a complex plant polymer, prevalent in the secondary cell walls of specific plant tissues, such as wood tissue. It has potential as a renewable source of phenolics. However, the prevalence of strong carbon-carbon (C–C) and ether (C–O) inter-unit linkages between the monomeric units represents a significant bottleneck to the industrial processing of plant biomass.^[1] Enzymes in the BAHD acyltransferase family are known to produce esters that can be incorporated at low levels into the lignin of naturally occurring plants.^[2] Introduction of ester bonds into lignin is advantageous as they can be easily cleaved by mild alkaline hydrolysis (saponification). This permits facile recovery of aromatic chemicals from the lignin and improved access to other plant cell wall polymers such as cellulose and hemicelluloses for easier carbohydrate release.^[3]

BAHD acyltransferases are a superfamily of acyltransferases in plants named after the first four enzymes identified in the family (benzoyl alcohol O-acetyltransferase, anthocyanin O-hydroxycinnamoyltransferase, N-hydroxycinnamoyltransferase, and deacetylvindoline 4-O-acetyltransferase).^[4] This family of enzymes uses a histidine-aspartate catalytic dyad (conserved HXXXD catalytic motif) to generate esters/amides from electrophilic coenzyme A (CoA) thioesters of hydroxycinnamic and hydroxybenzoic acids as donors and nucleophilic alcohols/amines as acceptors. Products from these reactions are both primary and secondary metabolites in plants that can be incorporated into lignin via radical polymerization reactions as depicted in Scheme 1.^[1,5] Recent studies have shown that overexpressing BAHD transferases in bioenergy plants can lead to increased ester incorporation into lignin allowing for easier downstream processing.^[3,6] Expression of substrate-specific acyltransferases in planta supported attempts at precision modification of plant cell wall composition.^[7] However, low steady-state levels of aromatic acid-CoA thioester precursors may limit the capability to increase ester synthesis. To overcome this bottleneck, we have created a library of ligases that may increase steady-state levels of the aromatic acid-CoA thioesters as per Scheme 1.

Scheme 1.

Aromatic acid-CoA ligases produce aromatic acid-CoA thioesters, that are the donor substrates for acyltransferases. Certain acyltransferases can use aromatic alcohols such as coniferyl alcohol as acceptors to produce ester-linked monomer conjugates in planta leading to increased incorporation of chemically labile ester linkages (highlighted by a dashed box) into lignin. These ester linkages can be easily cleaved by base treatment and allow for increased lignin valorization.

In plants, CoA thioesters of hydroxycinnamic acids, typically p-coumaric, caffeic, and ferulic acids are produced by enzymes dependent on adenosine triphosphate (ATP) that are collectively known as 4-coumarate:CoA ligases (4CL). These enzymes function upstream in the plant phenylpropanoid pathway by ligating CoA to aromatic acids derived from cinnamic acid, which is produced by elimination of ammonia, via phenylalanine ammonia lyase, from phenylalanine.^[8] The hydroxycinnamic acid-CoA thioester products of these enzymes are substrates for a large number of downstream enzymes and are incorporated into a wide range of biologically produced aromatic compounds.^[9] Some representative compound classes derived from p-coumaroyl-CoA with example members are highlighted in Scheme 2.

Scheme 2.

Aromatic CoA thioesters are building blocks in a wide range of biological scaffolds. The molecule of p-coumaroyl-CoA is shown in the center with the transferable aromatic acyl-moiety highlighted in blue; biological scaffolds derived from this molecule are shown around it with the incorporated aromatic acyl-moiety highlighted in blue. The compound classes are highlighted in blue, and the names of the individual compounds are in black.

Evolutionarily, 4CL is part of the ANL (Acyl-CoA ligase, Non-ribosomal peptide synthase (NRPS), Luciferase) family that contains a characteristic C-terminal flexible domain annotated in protein databases as “AMP-binding”. Catalysis in ANL enzymes proceeds via a “domain alternation” strategy, typically in two half-reactions that are associated with distinct conformational states of the protein.^[10] The highly conserved, invariant and reversible first half-reaction involves the attack of the acid carboxylate on the α-phosphate of ATP to generate an acyl-adenylate intermediate with the concomitant release of pyrophosphate. The ANL family shares this reaction with other adenylate-forming enzymes, which includes aminoacyl-tRNA synthetases, NRPS-independent siderophore synthetases, carboxylic acid reductases, amide bond synthetases,^[11] and ubiquitin activating enzymes (E1 ligases) among others.^[12] In the second half-reaction, a nucleophile attacks the activated carbonyl to release AMP and the ligated product. The ANL family appears to have specialized towards thiol nucleophiles derived from pantothenic acid (vitamin B5), and so utilizes CoA and acyl/peptidyl-carrier protein (ACP/PCP) linked 4’-phosphopantetheine as nucleophiles in acyl-CoA ligase and NRPS enzymes respectively.^[10] A notable exception to this are luciferases that catalyze the oxidative decarboxylation of d-luciferyl-adenylate under aerobic conditions as the major reaction but remain capable of CoA ligation activity under anaerobic conditions with the natural substrate d-luciferin.^[13]. The CoA ligase activity of luciferases dominates with the enantiomeric l-luciferin as well as some fatty acids under aerobic conditions. These acids can act as competitive inhibitors to the light producing reaction, which suggests that the bioluminescence reaction evolved more recently in the CoA ligase family by leveraging the molecular characteristics of d-luciferin.^[13] An instance of the natural separation of the CoA ligase reaction into adenylation and thioesterification reactions on two separate ligases has also been recently reported. The PtmA2 ligase is a rare example of a non-adenylating ANL enzyme capable of the thioesterification reaction, to which adenylation activity could be partially restored by installation of a strategic lysine residue known to be critical for the adenylation reaction.^[14] We also note that ANL ligases are distinct from pimelate-CoA ligases involved in biotin biosynthesis, structures for which have been reported recently.^[15] Catalysis in these ligases also proceeds via an acyl-adenylate intermediate, but they lack the C-terminal domain and consequently the domain alternation conformational change that is a hallmark of ANL enzymes.

Although numerous examples of plant 4CL enzymes are known, they typically display low activity, and broad substrate specificity, albeit with a known inability to ligate the highly substituted sinapic acid.^[8] There are few reports of benzoate-CoA ligases (BCLs) in plants despite the presence of many metabolites containing substituted benzoate moieties in planta.^[16] We therefore sought to identify additional ligases that might overcome these deficiencies and provide a toolset for altering the metabolic flux of the aromatic acid-CoA thioesters of interest.

Aromatic acid CoA ligases are also found in microbes^[17] but have not been characterized as extensively as plant ligases. These enzymes typically act upstream in aerobic/anaerobic aromatic acid degradation pathways that feed carbon from aromatic acids into central metabolic intermediates. We hypothesized that such ligases may have distinct kinetic and specificity properties as compared to plant ligases due to their different roles in metabolism, which could make them useful as biocatalysts and as genetic components in engineered biosynthetic pathways. In this work, we utilize a simple thiol detection assay to systematically characterize the substrate scopes of a microbial ligase library against an aromatic acid panel and identify highly active ligases for biocatalysis and future plant transformation experiments.

Results and Discussion

Ligase Library Generation via Database Mining

Ten genes annotated in MetaCyc^[18] to catalyze the ATP-dependent thioesterification of hydroxycinnamic acids to their CoA-thioesters were added to the ligase library. The genes were chosen based on the strength of the literary evidence in support of their annotated function that in some cases involved knock-out/knock-in studies in the wild-type strain and studies of recombinantly produced protein in others. Analysis of these genes revealed that the 8 encoded proteins belonged to the classical ANL family (L2, L4-L8, L10 and L11) and contained the signature C-terminal AMP binding domain. The two remaining proteins (L9 and L14) were from microbial prokaryotic ferulic acid degradation clusters. These two had a domain architecture like acetyl-CoA synthetase (NDP-forming)^[19] and represent the only two instances of ADP-forming 4CLs reported to date. We note that the ADP-forming reaction proceeds via an acyl-phosphate intermediate instead of an acyl-adenylate and is exclusively restricted to the metabolic context of aerobic microbial degradation of lignin-derived hydroxycinnamic acids.^[20] Three microbial benzoate CoA ligases (L1, L3 and L12) reported to react with benzoates were also included in the ligase library. After review of the properties of plant ligases, we also included 4-coumarate ligase isoform-2 from Nicotiana tabacum (Nt4CL2 – abbreviated here as L0) in our library as a well-characterized control to validate our methods and assay.^[21] 4-Coumarate ligase isoform 4 from Arabidopsis thaliana (At4CL4 – abbreviated here as L13) was also included in the library as it has been reported to be a rare example of plant ANL ligase capable of sinapic acid thioesterification.^[22] Ligase library details including source organism, Uniprot accession codes for protein sequences, and source publication for the gene (if available) are listed in Table S1. Metabolic contexts of the aromatic acid-CoA ligases sourced for this study are illustrated in Scheme 3. Collected DNA sequences were codon-optimized for protein expression in Escherichia coli (E. coli), synthesized as gene blocks (Integrated DNA Technologies – IDT) and ligated via Gibson assembly into a pVP67K vector with a kanamycin resistance marker and an N-terminal 8x polyhistidine tag to enable affinity purification (Figure S1).^[23] Protein production was induced in a small-scale growth of E. coli cells containing ligase genes inserted into this plasmid via chemical induction using isopropyl β-d-1-thiogalactopyranoside (IPTG). Post growth, cells were lysed using a commercial chemical lysis reagent, centrifuged to clear insoluble debris and the soluble fractions were screened by SDS-PAGE to identify genes producing soluble protein products by comparison against an uninduced control (Figures S2 and S3).

Scheme 3.

Metabolic contexts of the aromatic acid CoA ligases sourced for this study. (A) 4-Coumarate ligases involved in microbial aerobic degradation of lignin derived hydroxycinnamic acids utilize an ADP-forming mechanism to ligate CoA to aromatic acids prior to degradation of the acid into central metabolism intermediates. (B) 4-Coumarate ligases in plant phenylpropanoid metabolism convert endogenously produced hydroxycinnamic acid into CoA thioesters that are funneled into several downstream plant pathways including lignification. (C) Benzoate CoA ligases are involved in the microbial degradation of benzoic acid into central metabolism intermediates and act in aerobic/anaerobic pathways depending on the specific organism.

Subsequently, these targets were scaled up for protein production at a 500 mL scale in auto-induction medium^[23] (Figure S4) and recombinant proteins were purified by Ni-NTA affinity chromatography in high yield (>100 mg protein from 500 mL culture) with good purity (Figure S5). Post elution, these proteins were desalted into low salt buffer using a spin column via centrifugation and assayed against benzoic and 4-coumaric acid as an initial test for aromatic acid CoA ligation activity. A red shift in the optical spectrum corresponding to the formation of the CoA thioester from the parent aromatic acid showed that L14 was active on 4-coumaric acid (Figure S6). It was determined using the same method that L0 and L9 were also active on 4-coumaric acid, whereas L3 was active on benzoic acid.

Development of an Optical Assay for Ligase Screening

The CoA thioester from aromatic acids can be detected by wavelength-scanning the reaction mixtures over the 280–400 nm region, but different extinction coefficients for the various aromatic acid products complicate the systematic comparison of enzymatic reaction conversions. To overcome this, we implemented a microplate-based optical assay based on thiol quantification using 5,5’-dithiobis(2-nitrobenzoic acid) (DTNB, Ellman’s reagent) to quantitate free CoA at a fixed time end point for the ligase reaction.^[24] This assay provided a standardized readout at 412 nm at physiological pH and a method to test ligase activity on non-chromogenic acids such as acetate as shown in Figures 1A and S7. Standard curves prepared using β-mercaptoethanol and authentic CoA standard were found to be identical showing that the assay robustly and specifically detects thiol content in the 10–100 μM range with a sensitivity of μ1 μM (Figures S8, S9, S10, and S11). The molar absorptivity of stoichiometrically released 2-nitro-5-thiobenzoate (TNB^2-) anion was calculated to be μ14000 M⁻¹ cm⁻¹ which is in agreement with values in the published literature.^[24] However, in attempts to measure the activity of the ADP-forming ligase L9 using this assay, it was observed that the baseline increased over time post addition of DTNB to the ligase reaction (Figure S12). The absence of this problem for the AMP-forming ligase L0 suggested it was an enzyme-mechanism specific issue. The ADP-forming ligase contains catalytic domains homologous to those found in the central metabolism enzyme succinyl-CoA synthetase, which is known to catalyze reversible CoA ligation. The ability of L9 to catalyze the reverse reaction was therefore tested by incubating the enzyme with ADP, inorganic phosphate (P_i) and feruloyl-CoA. It was shown that both L9 and L14 can catalyze feruloyl-CoA degradation (Figure S13) with concomitant formation of ATP (Figure S14) proving that the ADP-forming mechanism retains its reversibility despite the switch in the preference for the acid substrate. This explains the baseline increase upon DTNB treatment as the L9 enzyme begins to regenerate CoA from the acyl-CoA thioester upon conjugation of free CoASH to DTNB to restore equilibrium. To overcome this, the DTNB detection solution was supplemented with 6 M GuHCl to quench the ligase reaction by protein denaturation without disrupting the pH. As an equal volume of the ligase reaction was mixed with the detection solution prior to readout by optical spectroscopy, the final concentration of GuHCl was 3 M and this was found to be sufficient to stabilize the baseline (Figure S12). A DTNB standard curve using CoA standard was prepared in 3 M GuHCl as shown in Figure S10 and was found to be indistinguishable from the standard curve under non-denaturing conditions, consistent with literature reports.^[24] Subsequent assays were performed under denaturing conditions.

Figure 1.

DTNB assay used to map ligase substrate scope. (A) Principle of DTNB assay. Leftover CoA at the end of ligase reaction is reacted with a ~3-fold molar excess of Ellman’s reagent DTNB. The amount of stoichiometrically released TNB^2- anion is quantified using absorbance at 412 nm and corresponds to amount of unreacted CoA. (B) Mapping L0 ligase substrate scope against a panel of structurally diverse aromatic acids. The horizontal line on the graph denotes the limit of detection for the assay.

The DTNB assay was benchmarked by testing an initial panel of six acids against the plant ligase L0 as shown in Figure-1B. Reactions were initiated by the addition of ATP, incubated for an hour and then the thiol content was determined by the DTNB assay. Conversion of free CoA to CoA thioester was calculated by comparison of thiol content to a control reaction lacking ATP to determine percentage of CoA converted to thioester. L0 showed high conversion to CoA thioester with cinnamic (1), p-coumaric (3), and ferulic (5) acids, but was lacking activity with sinapic acid (6) as established previously. No activity was detected with benzoic acid (9), suggesting that the propenyl extension on the aromatic ring was important for substrate recognition. Furthermore, although the bulky triple-substituted sinapic acid (6) was not accepted as a substrate, the enzyme showed a high conversion on p-methoxycinnamic acid (4) suggesting that the multiple substituents on the aromatic acid ring in sinapic acid (6) could introduce steric clashes in the active site or interfere with protein conformational dynamics essential for catalysis. The inability of plant ligases to convert sinapic acid (6) to sinapoyl-CoA may reflect its more recent emergence in land plants as compared to p-coumaric (3) or ferulic acid (5) as evidenced by studies of the ferulate 5-hydroxylate (F5H) gene, the protein from which catalyzes the first committed step in sinapate biosynthesis.^[25]

Mapping Substrate Scope of Ligase Library by DTNB Assay

The DTNB assay was used to map the substrate scope of the other three active ligases (L3, L9 and L14) against a broader panel of aromatic acids relevant to plant lignification. Fifteen acids were analysed as a part of this panel as shown in Figure 2A. This number allowed full utilization of the 96-well plate to read ligase substrate scope with the inclusion of a no-acid negative control in a single experiment. Calculated conversions of the four active ligases against this panel are shown as a heatmap in Figure 2B and as individual enzymatic reactions in Figures S15–S18. Consistent with prior studies,^[16,21] there was no significant cross-reactivity between enzymes catalyzing reaction on benzoic acid and those that prefer p-coumaric acid. None of the enzymes tested showed any activity on 4-hydroxyphenylacetic acid (8) despite this acid having identical aromatic ring substitution as p-coumaric (3) and p-hydroxybenzoic acid (12). This emphasizes the fact that the length of the carbon chain connecting the carboxylate reactive centre to the aromatic moiety is a key molecular determinant of ligase specificity. No conversion was seen for any of the tested ligases with acetate (15), confirming that the aromatic moiety is crucial in positioning the carboxylate for catalysis.

Figure 2.

Systematic mapping of substrate scope for active aromatic acid CoA ligases. (A) 15 acid panel comprising of acids relevant to plant lignification examined for ligation. (B) Heatmap of the % conversion of 15 aromatic acids to the respective CoA thioester by different ligases as measured by DTNB assay. Reactions contained 1 μM enzyme for L0 and L3, and 0.2 μM enzyme for L9 and L14. (C) Ligase reaction (10 mL scale) using sinapic acid (6). The characteristic yellow color of sinapoyl-CoA is observed in the tube utilizing L9 enzyme for reaction.

In addition to benzoic acid (9), L3 was found to catalyze CoA ligation to benzoic acids with single substitutions on the aromatic acid ring including salicylic (10), anthranilic (11) and p-hydroxybenzoic (12) acids. However, no conversion was detected with the di-substituted vanillic acid (13) or the tri-substituted syringic acid (14). In this regard, the ANL ligase L3 is like L0 in that multiple substituents on the benzene ring are not tolerated and pose a hindrance to catalytic conversion.

L0, L9, and L14 ligases all catalyzed CoA ligation to p-coumaric acid (3) as well as to the saturated dihydro-p-coumaric acid (7), suggesting that the double bond in the propanoic acid extension on the aromatic ring is not essential for substrate recognition.

Interestingly, whereas L0 showed conversion on the unsubstituted cinnamic acid (1) as well as o-coumaric acid (2), L9 and L14 activity seemed to have an absolute dependence on the presence of an oxygen substitution at the para-position on the aromatic ring as evidenced by high conversion of p-coumaric (3) and p-methoxy cinnamic acids (4). This is consistent with their natural metabolic context in microbial lignin degradation, as cinnamic acid is completely absent in lignin because radical formation reactions require the phenolic (4-OH) substituent. Both the microbial ligases showed high conversion of sinapic acid (6), which is a challenging substrate for plant ANL ligases. This suggests that pivoting to a different central metabolism type mechanism for the CoA-ligase reaction may have arisen as an adaptation at the organismal level that has allowed for utilization of sinapic acid as a microbial carbon source.

Milligram-scale CoA Thioester Synthesis using Ligases

During the assays, it was apparent that the microbial 4CL ligases were highly efficient and more rapidly achieved higher conversion on substrates with a 5-fold lower loading of catalyst when compared to L0. The previously established, widely used protocol for aromatic acid CoA thioester synthesis^[26] utilizes L0 as a catalyst and requires a long reaction time (μ20 h) which is disadvantageous as the CoASH thiol degrades in solution over time, thereby reducing yield and leading to waste of expensive CoA reagent. Pilot 10 mL scale reactions were set up utilizing L9 and it was found that the enzyme achieved visualizable conversion in 15 min as shown in Figure 2C. Quantification of reaction progress by optical spectroscopy, detecting the aromatic acid-CoA thioester product and by the DTNB assay, indicated >90% conversion within an hour.

Encouraged by these results, we adapted the literature procedure for thioester purification,^[26] using a C-18 column to bind the hydrophobic aromatic acid-thioesters in aqueous solution supplemented with 4% ammonium acetate and eluting residual aromatic free acid and excess CoA. Once the absorbance at the λ_max for each compound returned to the baseline, then five column volumes of milliQ water removed the excess ammonium acetate, and the pure thioesters were eluted in methanol. The isolated yields for hydroxycinnamoyl- CoA thioesters synthesized using L9 enzyme were quantified by optical spectroscopy using published literature extinction coefficients for 4-coumroyl-CoA (21 mM⁻¹ cm⁻¹ at 333 nm), feruloyl-CoA (19 mM⁻¹ cm⁻¹ at 346 nm), and sinapoyl-CoA (20 mM⁻¹ cm⁻¹ at 352 nm) ^[27] and are shown in Table 1.^[21] The thioesters of p-coumarate (3) and ferulate (5) were successfully synthesized and isolated in a high yield by this method. Sinapic acid (6) required longer washing with 4% ammonium acetate to completely remove it from the column and the yield of sinapoyl-CoA was lower than for the other hydroxycinnamates as seen in Table 1. The product thioesters of hydroxycinnamic acids could be easily distinguished from the parent acids by screening for a red-shift by optical spectroscopy (Figure S6). The identity and purity of the hydroxycinnamoyl-CoA thioesters were further confirmed by high-resolution mass spectrometry in negative-ion mode (Figure S19).

Table 1.

Isolated yields of synthesized thioesters.^[a]

Entry	Catalyst	Substrate	% Yield (± SD) [b]
1	L9	p-coumarate	89 (3) [c]
2	L9	ferulate	89 (4) [c]
3	L9	sinapate [e]	75 (2) [c]
4	L3	benzoate	92 (2) [d]
5	L3	p-hydroxy benzoate	94 (1) [d]
6	L3	anthranilate	91 (3) [d]
7	L3	salicylate [f]	70 (2) [d]

^[a]Unless otherwise noted, reactions were carried out in 100 mM HEPES pH 7.5 for 1 h and included 2 mM MgATP, 1 mM of corresponding acid, 0.5 mM CoA and 0.4–1 μM enzyme.

^[b]Purified yield of thioester reported with respect to CoA. Average and standard deviation for three independent runs.

^[c]Determined by optical spectroscopy using published extinction coefficients.

^[d]Determined by weighing of purified, lyophilized product and confirmed by optical spectroscopy.

^[e]Purified using an extended step gradient to separate the thioester from residual acid.

^[f]Reaction included 0.05 U/mL of pyrophosphatase and carried out for 2h to increase conversion. Thioester was purified using an extended step gradient to remove residual CoA.

L3 ligase was used to synthesize the CoA thioesters in high yield for benzoate (9), salicylate (10), anthranilate (11) and p-hydroxybenzoate (12) as shown in Table 1. As reliable extinction coefficients for all the compounds synthesized were not available, yield was calculated by weighing the purified thioesters after lyophilization. Salicyl-CoA co-eluted with unreacted CoA, the presence of which was confirmed by DTNB reaction. To ensure the reaction was permitted to achieve completion, 0.05 U/mL of inorganic pyrophosphatase (New England Biolabs) was included in the reaction and the reaction time was extended to 2 h. This reduced the amount of remaining CoA and salicyl-CoA was separated from CoA utilizing the same step gradient elution method employed for sinapoyl-CoA. The absence of unreacted CoA in the collected fractions and the resuspended lyophilized thioester was confirmed by negative reaction with the DTNB reagent. The optical spectra of the benzoyl-CoA thioesters showed a π→π^∗ transition around 260 nm corresponding to the adenyl-moiety of CoA and the thioester linkage, alongside the n→π^∗ transition in the 280–400 nm region corresponding to the substituted phenolic ring (Figure S20 and S21) consistent with literature.^[28] The extinction coefficient for the π→π^∗ transition at 260 nm for all the synthesized benzoyl-CoAs was calculated to be μ20000 M⁻¹ cm⁻¹, agreeing with the published value of 21000 M⁻¹ cm⁻¹ for benzoyl-CoA.^[28] The red shift observed for the n→π^∗ transition ranged from 280–400 nm and was dependent on aromatic ring substituents, and comparison of optical spectra showed that anthraniloyl-CoA displayed a more pronounced red-shift with an absorption maximum at 356 nm as compared to 320 nm for salicyl-CoA (Figure S21). The benzoyl-CoA thioesters were further characterized by high-resolution mass spectrometry in negative-ion mode (Figure S22). In the high m/z region, they were observed as singly charged pseudo-molecular ions [M-H]^- and their monosodium adducts [M+Na-2H]^- (Figure S22, panel A). Comparison of the ¹H NMR spectra of benzoyl-CoA and p-hydroxybenzoyl-CoA to CoA revealed a significant shift of the proton attached to the carbon adjacent to the reactive thiol group and smaller shifts in the aromatic protons of the CoA adenine group for the benzoyl-CoA thioesters (Figure S23). The prepared thioesters can be used for the assay of enzymes such as acyltransferases in which they act as donor substrates.^[5,7]

Conclusion

In summary, we screened a library of CoA ligases against a panel of hydroxycinnamic and benzoic acids relevant to plant metabolism and identified four robust, highly active enzymes suitable for further plant manipulation experiments. For hydroxycinnamic acids, we report enzymes unique to microbial aerobic lignin degradation that use an ADP-forming mechanism to generate aromatic acid-CoA thioesters, thus expanding the options beyond previously known plant ANL enzymes. We leverage the high reactivity of the ADP-forming ligase L9 to establish a facile and rapid protocol for the lab-scale synthesis of aromatic acid CoA thioesters that are of importance as substrates for a wide range of enzymes such as BAHD acyltransferases involved in lignification and natural product synthase enzymes. The molecules reported herein also represent starting points for the study of many other catalytic processes. Our work highlights the notion that surveying diverse natural metabolic contexts is an effective strategy for the identification of useful enzymes of practical interest and establishes a toolbox of aromatic acid-CoA ligases.