Toluene Dioxygenase-Catalyzed cis-Dihydroxylation of Quinolines: A Molecular Docking Study and Chemoenzymatic Synthesis of Quinoline Arene Oxides

Molecular docking studies of quinoline and 2-chloroquinoline substrates at the active site of toluene dioxygenase (TDO), were conducted using Autodock Vina, to identify novel edge-to-face interactions and to rationalize the observed stereoselective cis-dihydroxylation of carbocyclic rings and formation of isolable cis-dihydrodiol metabolites. These in silico docking results of quinoline and pyridine substrates, with TDO, also provided support for the postulated cis-dihydroxylation of electron-deficient pyridyl rings, to give transient cis-dihydrodiol intermediates and the derived hydroxyquinolines. 2-Chloroquinoline cis-dihydrodiol metabolites were used as precursors in the chemoenzymatic synthesis of enantiopure arene oxide and arene dioxide derivatives of quinoline, in the context of its possible mammalian metabolism and carcinogenicity.

No direct evidence for cis-dihydroxylation of the electron-poor pyridyl ring was found during analysis of the biotransformation products (Boyd et al., 1987(Boyd et al., , 1993. The combined relative yields of the isolated achiral metabolites 3-hydroxyquinoline 5 (13%) and anthranilic acid 6 (27%), were however consistent with the initial formation of the transient cis-dihydrodiol 7 as a further major initial metabolite (Figure 1). Formation of 3-hydroxyquinoline 5 could result from spontaneous dehydration of intermediate 7 and of anthranilic acid 6 from enzymatic ring-cleavage (ERC) of 3-hydroxyquinoline 5. If anthranilic acid 6 and phenol 5 were derived from intermediate cis-dihydrodiol 7, the relative ratio of initially formed cis-dihydrodiols diols 2, 3, and 7 would be estimated as ca. 33:27:40. Further in silico support for the formation of transient cis-dihydrodiol 7 -the major metabolite of quinoline 1 (TDO as biocatalyst), was obtained from a joint study with a collaborating laboratory using the GOLD molecular docking program (unpublished data).This prompted our interest in employing the Autodock Vina program for the current study.
Indirect evidence for the undetected heterocyclic cisdihydrodiol 7 was acquired from the results of an earlier biotransformation (P. putida UV4) of 3-deuterioquinoline 1 (X = 87% D); it yielded 4-deuterioquinolin-3-ol 5 (23% D) along with other metabolites (Figure 1) (Barr et al., 1998). This could be accounted for by the migration and partial retention of deuterium, from the aromatization of the intermediate cis-dihydrodiol 7 (X = D, Figure 1) via an NIH shift mechanism, as observed during aromatization of the isolated carbocyclic cis-dihydrodiols of naphthalene and quinoline 2 and 3.
Major objectives of this study were to: (i) review proposed metabolic pathways of quinoline 1 and 2-chloroquinoline 8 by P. putida UV 4 cells, and to compare with TDO docking results (ii) use the isolated cis-dihydrodiol metabolites 9 and 10 and their derivatives, in the quest for improved chemoenzymatic synthetic routes to enantiopure arene oxide and dioxide derivatives of quinoline 1.

Laboratory Studies
1 H and 13 C NMR spectra were recorded on Bruker Avance DPX-300 and DPX-500 instruments. Chemical shifts (δ) are reported in ppm relative to SiMe 4 and coupling constants (J) are given in Hertz (Hz). Mass spectra were run at 70 eV, on an AE1-MS90 mass spectrometer updated by VG Autospec, using a heated inlet system. Accurate molecular weights were determined by the peak matching method, with perfluorokerosene as the standard. Elemental microanalyses were carried out on a Perkin-Elmer 2400 CHN microanalyser and IR spectra were recorded in KBr disc or in thin film, using a Perkin-Elmer Spectrum RX1 FT-IR spectrometer. ECD spectra were obtained using a Jasco J-720 instrument and MeCN as solvent. Optical rotations ([α] D ) measurements (10 −1 deg cm 2 g −1 ) were carried out at ambient temperature on a Perkin-Elmer 214 polarimeter and specified solvent concentration (g/100 ml) at sodium D-line (589 nm). Melting points were recorded in degrees Celsius using a Stuart SMP10 melting point apparatus. Column chromatography and preparative layer chromatography (PLC) were performed on Merck Kieselgel type 60 (250-400 mesh) and PF 254/366 , respectively. Merck Kieselgel type 60F 254 analytical plates were used for TLC.

Docking Process
All small molecule structures were created in.pdb format using UCSF Chimera (University of California). The crystal structures of TDO (PDB ID: 3en1) and NDO (1o7M) were obtained from the RCSB Protein Data Bank. The toluene contained in the TDO crystal structure was removed using UCSF Chimera. The dioxygen molecule was added to the iron prosthetic group of TDO (Fe, His222, His228, Asp376) from the iron prosthetic group (Fe, dioxygen, His-208, His-213, and Asp-362) of the NDO crystal structure using the "super" function of PyMol 2.4.0 to overlay the two partial structures at the position of 3en1 and copy the dioxygen atoms to the 3en1 model. All structures in.pdb format were then stripped of water molecules and converted to .pdbqt format using AutoDock Tools 1.5.6 (Scripps Research Institute).
The docking was performed using AutoDock Vina 1.1. This configuration includes the amino acids within 6 Å of the toluene present in the 3en1 crystal structure: Gln215, Phe216, Met220, His222, Ala223, His228, Leu 272, Ile276, Ile232, Val309, His311, Leu321, Ile324, Phe366, Phe372, and Asp376. Only Met220 is not shown in the docking structures as it blocked the view on the docking orientation and was considered irrelevant for catalytic purposes.
Based on earlier publications and current understanding of the binding site and catalytic mechanism, His 222 is involved in edge-to-face binding of the substrate to direct a planar orientation, presenting a face to the active site. The probability for the methyl group of the toluene substrate to be oriented toward Phe216 is reduced by this residue, thus resulting in the observed enantiomeric excess. His311 and Asn215 are considered to be essential for the catalytic mechanism, as His311 is a H-acceptor and Asn215 is considered to be part of a water channel.
The docking model was assessed by docking toluene and comparing obtained orientations to the toluene position in the crystal structure. A total of nine orientations were obtained ranging from −5.3 to −4.2 kcal/mol −1 . Three orientations emerged that present the 2,3-double bond to the dioxygen for dihydroxylation (see Supplementary Figures S-13A-C). These orientations showed minor differences in position compared to the orientation observed in the crystal structure (1-1.2 Å) and compared to each other. The difference in position between the docked and crystal structure was probably affected by the docking with dioxygen, as the crystal structure does not contain dioxygen, allowing the toluene to be closer to the iron prosthetic group. Two identical orientations emerged (see Supplementary Figure S-13D) which would result in the opposite enantiomer being formed. Given the enantiomeric excess of >98% of the dihydroxylated product of toluene, it appears unlikely that this orientation is readily dihydroxylated.
Based on the AutoDock Vina results, the orientations within the TDO active site, favored by heterocyclic arene substrates, e.g., 1 (X = H), 8, and 14, as well as substituted benzene substrates e.g., toluene and chlorobenzene 17, all revealed evidence of edgeto-face (T-bonding) interactions with Phe-216 and His-222. The simultaneous edge-to-face interactions, of amino acid residues Phe-216 and His-222, with both the phenyl and pyridyl rings of substrates 1, 8, and 14 in the present context, were noteworthy. Docking distances between the proximate T-bonding H atoms and arene faces of quinoline 1 (Supplementary Figures 4A 1 -A 3 ) were estimated to be within the range 2.6-2.8 Å, assuming a C-H bond length of 1.08 Å. These inter-ring distances are in accord with the calculated interacting H-to-ring center perpendicular distances, of ca. 2.6-2.8 Å and observed distances ca. 2.70-2.86 Å, from X-ray crystallographic analysis, using a range of model systems (Hoering et al., 2016;Boyd et al., 2019). Keyhole pictures, looking through the rear face of the component aromatic ring, showed both edge to face-T (Supplementary Figures 4A 3 ,B 3 ,C 3 ), and face tilted-T dockings (Supplementary Figures 4A 2 ,B 2 ,C 2 ) of quinoline 1 with TDO. This was consistent with the attractive interactions between phenyl rings (Supplementary Figures 4A 1 ,B 1 ), phenyl/pyridyl rings (Supplementary Figure 4C 1 ), phenyl/imidazoyl rings (Supplementary Figures 4A 1 ,B 1 ), and pyridyl/imidazoyl rings (Supplementary Figure 4C 1 ).

DISCUSSION
Both Autodock and Gold programs proved to be very successful in matching the preferred in silico orientations of arene substrates with the experimentally confirmed regiochemistry and absolute stereochemistry of the isolated cis-diol metabolites. Previous docking studies of arene substrates with TDO indicated that the preferred orientations were controlled by: (i) attractive edge-toface T shaped interactions with the orthogonal phenyl group of Phe-216 and imidazole ring of His 222 and (ii) Van der Waals interactions with the proximate hydrophobic amino acids Ile-276, Leu-272, Ile-324, Val-309, Leu-272, Phe-352 (Hoering et al., 2016;Vila et al., 2016aVila et al., ,b, 2017Boyd et al., 2017Boyd et al., , 2019. Theoretical, crystallographic and experimental support for phenyl-phenyl and phenyl-pyridyl edge-to-face bonding (Tbonding) interactions, between arene and heteroarene rings, has been reported (Jennings et al., 2001;Escudero et al., 2009;Gonzalez-Rosende et al., 2017;Aliev and Motherwell, 2019). Little evidence was available, from crystalline protein structures, for similar edge-to-face interactions between the imidazole ring of histidine with other aromatic residues, e.g., Phe, Tyr, Trp, His (Bhattacharyya et al., 2003). The possibility of further types of edge-to-face interactions between heteroarene substrates 1 (Figures 4A-C), 8 (Figures 5A-C), and 14 (Figure 6) with Phe-216 and His-222 in the active site of TDO (Figures 4-6), is also apparent.
To account for the unexpected isolation of (3S,4S)-diol 11 as a product from P. putida UV4 biotransformations of 2-chloroquinoline 8, the undetected cis-dihydrododiol 12 of unknown absolute configuration, was originally proposed as a possible intermediate in the metabolic sequence 8 → 12 → 11 (Figure 2)  . Later GC-MS identification of 2-quinolone 13 as a very minor metabolite of substrate 8, was consistent with an alternative or additional metabolic pathway (8 → 13 → 11, Figure 2) (Boyd et al., 2002). Stronger evidence for the latter biosynthetic route was provided when 2-quinolone 13 was used as substrate and (3S,4S)-diol 11 was found to be the only identifiable metabolite. Since 2hydroxyquinoline is a minor tautomer of 2-quinolone 13, the formation of cis-diol metabolite 11 could also be considered as a TDO-catalyzed cis-dihydroxylation within a substituted pyridyl ring. It is therefore possible that cis-diol metabolite 11 was formed from 2-chloroquinoline 8, via both 2-quinolone 13 and cis-dihydrodiol 12 intermediates.
The previously reported regioselective and enantioselective cis-dihydroxylation, of the carbocyclic ring of quinoline 1, 2chloroquinoline 8 and 2-chloropyridine 14, was rationalized by molecular docking (Autodock Vina) studies of these substrates. The in silico results revealed novel edge-to-face attractive interactions at the active site of TDO. These studies also provided support for the TDO-catalyzed cis-dihydroxylation of: (i) the pyridyl ring in substrates, 1, 8, and 14, to yield the corresponding unstable cis-dihydrodiol metabolites, 7, 12, and 15, and (ii) the carbocyclic ring to give stable cis-dihydrodiol metabolites 2, 3, 9, and 10 having the correct absolute configurations. Evidence for the monooxygenase-catalyzed epoxidation and dioxygenasecatalyzed cis-dihydroxylation of a pyridyl ring was discussed, in the context of both bacterial and mammalian metabolism of quinoline 1.
Our findings are consistent with the current thinking on the catalytic mechanism for these enzymes, which is not fully understood. Barry and Challis (2013) discuss an important point, i.e., whether the Fe(III)-OOH complex in the TDO catalytic cycle reacts directly with the substrate or first undergoes rearrangement to an Fe(V)-O(OH) complex. They highlight the question of which arene carbon atom first forms a bond with an oxygen atom.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
CA, DB, NS, and PS supervised the research, designed the research, obtained the funding, wrote the manuscript, and designed the experiments. PH conducted the modeling and docking. JC, PL, and NS conducted the laboratory synthesis/biotransformations. All authors contributed to the article and approved the submitted version.

FUNDING
Financial support for post-graduate studentships, RA funding, and academic staff funding was gratefully received from BBSRC (grant number 81/ABC09451 to DB and NS for RA and academic funding, grant number BB/E013848/1 to CA for academic funding) and the Leverhulme Trust (to CA/PH for Ph.D. and academic funding), the European Social Fund (JC for Ph.D. funding) and Queen's University of Belfast, Oxford Glycosciences, and the Overseas Research Studentship (to PL for Ph.D. funding).