Regioselective para‐Carboxylation of Catechols with a Prenylated Flavin Dependent Decarboxylase

Abstract The utilization of CO2 as a carbon source for organic synthesis meets the urgent demand for more sustainability in the production of chemicals. Herein, we report on the enzyme‐catalyzed para‐carboxylation of catechols, employing 3,4‐dihydroxybenzoic acid decarboxylases (AroY) that belong to the UbiD enzyme family. Crystal structures and accompanying solution data confirmed that AroY utilizes the recently discovered prenylated FMN (prFMN) cofactor, and requires oxidative maturation to form the catalytically competent prFMNiminium species. This study reports on the in vitro reconstitution and activation of a prFMN‐dependent enzyme that is capable of directly carboxylating aromatic catechol substrates under ambient conditions. A reaction mechanism for the reversible decarboxylation involving an intermediate with a single covalent bond between a quinoid adduct and cofactor is proposed, which is distinct from the mechanism of prFMN‐associated 1,3‐dipolar cycloadditions in related enzymes.

Abstract: The utilization of CO 2 as ac arbon source for organic synthesis meets the urgent demand for more sustainability in the production of chemicals.Herein, we report on the enzyme-catalyzed para-carboxylation of catechols, employing 3,4-dihydroxybenzoic acid decarboxylases (AroY) that belong to the UbiD enzyme family.C rystal structures and accompanying solution data confirmed that AroY utilizes the recently discovered prenylated FMN (prFMN) cofactor,a nd requires oxidative maturation to form the catalytically competent prFMN iminium species.T his study reports on the in vitro reconstitution and activation of ap rFMN-dependent enzyme that is capable of directly carboxylating aromatic catechol substrates under ambient conditions.Areaction mechanism for the reversible decarboxylation involving an intermediate with asingle covalent bond between aquinoid adduct and cofactor is proposed, which is distinct from the mechanismofprFMNassociated 1,3-dipolar cycloadditions in related enzymes.
Carboxylation reactions have received considerable attention in view of the use of CO 2 as an abundant C 1 building block for sustainable chemical production. [1] However,t o date,only afew examples of CO 2 fixation reactions have been realized on industrial scale,m ainly owing to the high energy input required for substrate activation. In recent years, biocatalysts [2] have been exploited as attractive alternatives to chemical methods [1,3] to catalyze carboxylation reactions under mild, aqueous conditions.W hereas the biocatalytic carboxylation of aldehydes (TPP-dependent pyruvate decarboxylases), [4] epoxides (epoxide carboxylases from Xanthobacter sp.), [5] and heteroaromatic compounds,such as pyrroles (pyrrole-2-carboxylate decarboxylase from Bacillus megaterium) [6] and indoles (indole-3-carboxylate decarboxylase from Arthrobacter nicotianae), [2e] exhibited narrow substrate specificity,p romising results were obtained in the biocatalytic carboxylation of phenols and styrenes. ortho-Benzoic acid decarboxylases and phenolic acid decarboxylases show ar elaxed substrate specificity for the ortho-carboxylation of phenols [7] and the b-carboxylation [8] of styrenes,r espectively, whilst maintaining their exquisite regioselectivity.
To expand the toolbox for biocatalytic carboxylation, we searched for enzymes enabling the regiocomplementary paracarboxylation of phenols.M ost of the already characterized enzymes either require an ATP-consuming activation (phosphorylation) step prior to carboxylation (phenylphosphate carboxylases), [9] suffer from ar apid loss of activity under aerobic conditions,e specially after purification (4-hydroxybenzoate [10] and 3,4-dihydroxybenzoate decarboxylases [2d, 11] ) or have not been extensively examined in vitro yet, [12] which limits their usability for biotransformations.Based on aliterature survey and ap reliminary activity screen of hetero-logously expressed potential para-carboxylases (see the Supporting Information), 3,4-dihydroxybenzoic acid decarboxylases from Enterobacter cloacae (EcAroY) and Klebsiella pneumoniae (KpAroY,89%identical) were selected for further studies. [2d, 13] Both enzymes belong to the UbiD family and are related to aferulic acid decarboxylase (Fdc1), which has been shown to facilitate the (de)carboxylation of cinnamic acids in the presence of ar ecently discovered prenylated flavin (prFMN) cofactor. [14] Owing to the natural occurrence of the UbiD-associated prenyltransferase UbiX in the E. coli expression host, lyophilized whole cells displayed high decarboxylation activity in initial screenings (see the Supporting Information). However, upon purification of the decarboxylase from the E. coli host, only little enzyme activity could be detected for either of the two enzymes,despite additional co-expression with the UbiDassociated prenyltransferase UbiX to provide sufficient prFMN in vivo.U pon in vitro reconstitution with reduced prFMN (Figure 1a), [14a,15] decarboxylation activity could be detected with 3,4-dihydroxybenzoic acid (3,4-DHBA, 1) following brief exposure to oxygen to generate the active prFMN iminium form ( Figure 1b). Thel ack of activity for anaerobically reconstituted protein clearly demonstrates the requirement for oxidative maturation of the prFMN cofactor. EPR and UV/Vis spectroscopy revealed the presence of ar adical semiquinone intermediate following in vitro reconstitution and oxidation, which is reminiscent of an intermediate observed with the related E. coli UbiD [15] (see the Supporting Information). After reconstitution and maturation, the enzyme activity has ahalf-life of only 5-8 min under aerobic conditions but remained unchanged for at least 14 h under anaerobic conditions ( Figure 1c).
Ac omparison of the KpAroY and EcAroY crystal structures revealed only small differences,i na ccordance with the similar activities observed in solution experiments. TheA roY monomer consists of an N-terminal prFMN-binding domain (residues 1-339), an oligomerization domain (residues 340-475), and aC -terminal a-helix (residues 476-495;F igure 2a). Ac omparison with the previously reported structures of the fungal Fdc1 and E. coli UbiD [14] shows that AroY structures adopt an "open" conformation, where the position of the prFMN domain is more akin to that observed for UbiD,c ompared to the more closed conformation observed for Fdc1 (see the Supporting Information). This open conformation is observed for all crystallographically independent AroY monomers,a nd in the 4.6 cryo-EM solution structure of apo-EcAroY (Figure 2a).  Thea ctive site of AroY is clearly defined by the prenylated isoalloxazine moiety of the cofactor and the presence of key conserved residues (Figure 2b). It is situated near the hinge point of the prFMN-binding domain motion, and at the cleft between the oligomerization and the prFMNbinding domains.T wo water molecules are clearly defined in the active site;one of them forms hydrogen bonds to His327 and Lys363 while the other one interacts with Lys363, His436, and Arg188. We hypothesize that the two water molecules mimic the two hydroxy groups at the meta-and para-positions of the substrate protocatechuic acid (1). Given the relative rigidity of the substrate,w es uperimposed ac atechol moiety onto the two water molecules,and the carboxylate moiety was positioned in close proximity to the prFMN iminium group. While this has some similarity to the structure of the Fdc1:substrate complex (PDB-ID:4 ZA7), [14a] the exact relative position of the substrate carboxylate moiety and the prFMN N5ÀC1' iminium linkage is different. In Fdc1, the substrate a-carbon atom adjacent to the carboxyl group is located directly above the prFMN C1' atom whereas in AroY, the a-carbon atom is located above the isoalloxazine N5 atom (see the Supporting Information).
Exchanging several amino acids within the putative catechol-binding motif (Arg188, His327, Lys363 to Ala and His436 to Lyso rT hr) led to ac omplete loss of activity.T he same effect was observed upon exchange of Arg181 or Glu289, which are located near the carboxylate group of the substrate,toAla. Theanalogous Glu282 in Fdc1 is proposed to be required for the donation of aproton to the covalently bound intermediate, [14a,16] which hints at as imilar role for Glu289 as acatalytic acid in AroY.
Based on the available structural, mutational, and kinetic data as well as DFT calculations (see the Supporting information), we propose ar eaction mechanism involving aq uinoid intermediate (Figure 2c). This intermediate has chemical similarity to that proposed for the phenolic acid decarboxylases (PAD) [8a] and is different from the 1,3-dipolar cycloaddition mechanism proposed for Fdc1. [14a, [16][17][18] The calculations,e mploying al arge model of the active site with 283 atoms (Figure 2b), suggested that the generation of ac ycloadduct is unlikely in the case of AroY as it would require the formation of av ery strained intermediate (Figure S29). Thelatter is not required in the case of the cinnamic acid like substrates of Fdc1, for which previous calculations have validated the proposed 1,3-dipolar cycloaddition mechanism to the exocyclic alkene. [16b,17b] Instead, the mechanism involving aq uinoid intermediate was calculated to have feasible energy barriers,w hich are significantly lower than those of the 1,3-cycloaddition mechanism. Thec alculated energy profile and optimized structures of all intermediates and transition states along the reaction pathway are given in Figures S30-S32.
To force the reaction equilibrium towards the carboxylation of suitable substrates,w esubjected phenols 4-8 to a3m potassium bicarbonate buffer as aC O 2 source (Scheme 1b). Whereas the carboxylation of simple phenols was not successful (nonsubstrates are shown in Table S7), EcAroY catalyzed the regioselective para-carboxylation of catechol (4)i nt he presence of either bicarbonate or pressurized CO 2 (30 bar). Apart from 4, EcAroY also accepted pyrogallol (5) as as ubstrate,a nd carboxylation with bicarbonate predominantly occurred in the para-position to the central hydroxy group to give 2.C arboxylation in vicinity to the peripheral hydroxy groups occurred only to am inor extent to give 3.
As the catechol scaffold appears to be crucial for substrate acceptance,t he carboxylation of substituted catechols 6-8 with bicarbonate was tested (Scheme 1b). Small electronwithdrawing (3-F, 6)t om edium-sized electron-donating (3-OMe, 7;3 -Me, 8)s ubstituents were tolerated in the 3-position, and carboxylation occurred exclusively at the 5-position, that is,i npara-position to the central hydroxy group,w hich was confirmed by NMR spectroscopy (see the Supporting Information).
In conclusion, our data contribute to ab etter understanding of the ATP-independent para-carboxylation of phenolic substrates.C rystal structures and in vitro reconstitution data unambiguously demonstrate that prFMN is employed as ac ofactor, and that oxidative maturation is required for activity.T he exact mechanism of oxidative maturation and the cause of the observed oxygen sensitivity remain unclear at this stage.Owing to the preference of AroY for catechols over simple phenols,asecond hydroxy group seems to be mandatory for ideal substrate positioning (with Scheme 1. EcAroY substrate screening with lyophilized E. coli whole cells containing the heterologously expressed decarboxylase. Potassium bicarbonate (3 m)w as used as the CO 2 source in the carboxylation assays.
ahydrogen bond between OH and His327;F igure 2b)inthe active site of the enzyme.T he second hydroxy group further enhances the nucleophilicity of the (catechol) substrate to facilitate the nucleophilic addition step onto prFMN.E lectron-withdrawing and -donating groups are tolerated in the meta-position relative to the carboxylation site whereas substitution in the ortho-position was not tolerated owing to steric hindrance in the active site.
In the context of the wider UbiD family,t he metalassisted binding and the associated oxidative maturation of prFMN are common to all biochemically and structurally characterized enzymes (the fungal Fdc1, E. coli UbiD,a nd AroY). Thesubstrate binding specificity is distinct for each of these enzymes,but in all cases appears to be largely governed by residues derived from the oligomerization domain that are involved in binding to the non-carboxylate substrate moiety. Thecarboxylate group,ont he other hand, is bound near the conserved Glu-Arg-Glu/Asp triad of ionic residues,w hich is located near the prFMN N5=C1' iminium moiety.Akey difference is the relative position of the oligomerization and prFMN domains,a nd thus the relative position of the substrate-binding and carboxylate-binding motifs.T hese are considerably closer in the fungal Fdc1 structure than in bacterial UbiD and AroY.Aputative domain motion might allow the UbiD/AroY enzymes to adopt am ore Fdc1-like conformation, but this has not been directly observed. While the quinoid-based mechanism proposed for AroY suggests an alternative to 1,3-dipolar cycloadditions in the case of aromatic substrates,i td oes not explain how the (de)carboxylation of non-phenolic substrates [19] is achieved. Ab etter understanding of the UbiD enzyme family will require further studies of these and additional family members. [20]