FAD dependent monooxygenase TropB and investigation of its biotransformation potential †

School, of Chemistry, University of Bristol russell.cox@oci.uni-hannover.de Chemistry of Natural and Microbial Produc Cairo, 12622, Egypt Chemistry Department, Faculty of Science, Arabia Microbiology Department, Faculty of Pharm School of Biochemistry, University of Bristol Institut für Organische Chemie, Leibniz U 30167, Hannover, Germany † Electronic supplementary informa 10.1039/c5ra06693j ‡ These authors contributed equally. Cite this: RSC Adv., 2015, 5, 49987


Introduction
The direct hydroxylation of aromatic compounds in a chemo-, regio-and stereo-selective manner remains a major challenge in synthetic organic chemistry and is "one of the most challenging elds in modern synthesis". 1 In contrast, such reactions are rapid and efficient in nature.Metabolic pathways oen involve aromatic hydroxylases and such enzymes could be potentially engineered and exploited for chemo-synthetic processes.In particular avin dependant mono-oxygenases (FMOs) are attractive candidates for industrial aromatic hydroxylation. 2 For example, an FMO from Pseudomonas azelaica (2-hydroxybiphenyl 3-mono-oxygenase) was articially re-engineered for a large-scale production of 3-tert-butylcatechol, 3 a precursor for many important chemicals and synthetically intractable targets.
The recent elucidation of the biosynthetic pathway for stipitatic acid led to the identication of a new FMO (TropB). 4This enzyme catalyses the oxidative dearomatisation of 3-methylorcinaldehyde 1 to produce a dienone 2 (Scheme 1A).This hydroxylation step is remarkably similar to one of the steps reported for the chemical synthesis of azaphilones and their analogues (Scheme 1B). 5,6he synthetic step requires extreme cold temperatures (À70 to À10 C), hazardous reagents and relatively long reaction time (30 h).It would be advantageous if related transformations could be performed bio-catalytically using TropB or a related enzyme.This would obviate the utilisation of dangerous chemicals and challenging reaction conditions to facilitate the synthesis of key synthetic intermediates.
In our previous work, TropB was overproduced recombinantly in E. coli and puried by nickel affinity chromatography, eluted as a single SDS-PAGE band (see ESI †).The puried protein was shown by UV-Vis spectrophotometry and LC-MS to possess a non-covalently bound FAD cofactor.Exogenous NADH or NADPH is required as an electron donor for in vitro activity.number of analogous compounds 4-14 (Fig. 1) were tested with TropB.Only the methyl ketone 4 was hydroxylated, whereas the other substrates were not.Here we extend these initial investigations by performing a detailed kinetic characterisation of TropB, to explore the enzyme's tolerance to a broad range of alternative aromatic substrates for example, compounds that differ in alkyl chain length and branching, and different substituents on the aromatic ring.Additionally, the effects of temperature and pH on the enzyme activity were also investigated in order to identify the optimum reaction conditions for TropB.

Results and discussion
Overproduction of TropB and assay optimisation N-terminally hexahistidine tagged TropB was produced by recombinant overexpression in E. coli BL21-CodonPlus (DE3)-RP and puried to homogeneity using a combination of Niaffinity and size exclusion chromatography (ESI Fig. S1 †).The puried protein had a distinctive yellow colour and exhibited typical avoprotein absorbance spectra with maxima at 368 and 455 nm.These observations were consistent with those reported previously. 4No exogenous avin was required to detect enzyme activity.The identity of the puried protein was conrmed by MALDI-TOF MS of a tryptic digest (see ESI †).
The effect of temperature and pH on TropB activity was investigated using the enzyme's natural substrate, 3-methylorcinaldehyde 1.This was to determine the optimum reaction conditions for later use during more detailed kinetic analyses.The enzyme was incubated with a xed concentration of substrate (200 mM) and NADPH (100 mM) at different temperatures and appropriately buffered pHs and the initial rate of reaction measured.
TropB activity was monitored spectrophotometrically by following the formation of the reaction product at 400 nm.The pH at which TropB was optimally active was pH 8.0.There was a considerable decrease in enzyme activity below pH 7 or above pH 9 (Fig. 2A).The optimum temperature for Trop B activity was found to be 30 C although surprisingly the enzyme did retain 40% of maximum activity at 60 C (Fig. 2B).

Synthesis of substrate analogues
Compounds 15-27 (Fig. 3), were designed to probe the substrate selectivity of TropB in addition to the previously tested substrates.Metabolites 25-27 are natural products and were extracted from their natural sources according to established protocols. 5,7The analogues 4 and 8 were purchased from Sigma.The remaining compounds were produced synthetically.
Aryl alkyl ketones were obtained directly from Friedel-Cras acylation using 2,5-dimethylresorcinol 9 with the appropriate acid chlorides in the presence of zinc powder as a catalyst (Scheme 2). 8The yield of desired products was very low (<5%).This was unsurprising given the presence of unprotected   hydroxyl groups within 9 leading to the formation of many side products which were detected by LC-DAD-MS.However, the desired products were easily identied by their mass and UV absorption which was higher than those of the oxygen-acylated side products.As the aim was to rapidly produce only a few milligrams of material for enzyme assays the low yields were not problematic.The only exception was for 15 where it was not possible to generate enough material for full kinetic analysis.
The syntheses of 4-nitro-2,5-dimethylresorcinol 22 and 4,6dinitro-2,5-dimethylresorcinol 24 were achieved by the mono and di-nitration of 9 using a mixture of diluted nitric acid and acetic anhydride. 9,10The 4-nitroso-2,5-dimethylresorcinol compound 23 was synthesised using diluted sulphuric acid and sodium nitrite following a previously reported synthesis. 11ll the compounds were puried using preparative LC-MS and the structure of each was conrmed by high resolution mass spectrometry and NMR.To the best of our knowledge none of these compounds have previously been characterised and only 23 and 15 had previously been synthesised. 10,12Characterisation of compounds is given in the ESI.†

Substrate selectivity
TropB and its potential substrates (2.5 mM) were incubated for 1 hour with an excess of NADPH (3.0 mM) at pH 8 and 30 C. LCMS analysis was used to detect the presence of reaction products.Control experiments with heat-denatured protein were also performed.Compounds which were demonstrated as substrates by TropB in this assay were then taken forward for full kinetic characterisation.All kinetic assays were performed using a continuous spectrophotometric assay monitoring either product formation at 400 nm or conversion of NADPH to NADP + at 340 nm (Table 1) to determine the initial rate.
TropB was found to catalyse the NADPH dependent conversion of the natural substrate 3-methylorcinaldehyde 1 to the dienone 2 with a k cat of 0.298 AE 0.006 s À1 .This compares unfavourably to other FMO enzymes such as 2-methyl-3hydroxypyridine-5-carboxylic-acid dioxygenase 13,14 (MHPCO) which has a k cat of 11.8 s À1 and 3-hydroxybenzoate 6-hydroxylase 14 (3HB4H) which has a k cat of 36 s À1 .However, the K M determined for TropB (20.0 AE 1.42 mM) is similar to the values published for MHPCO (25.4 mM) and 3HB4H (27 mM).
The acetophenone derivative 4 is turned over by the enzyme at a similar rate to the natural substrate but binds more weakly as indicated by a 9-fold increase in K M .This is probably due to the missing 6-methyl group because compounds 16 and 17 which retain the 6-methyl group have lower K M values than 4. The fact that these K M values are still higher than the natural substrate is probably due to steric hindrance from the longer carbon chains.This less than optimal binding was probably the reason for the lower k cat values.Compound 8 which lacks both methyl groups was not accepted by the enzyme.
Compounds 18-21 which have long or branched alkyl ketone chains and aryls on the C-1 position are not well tolerated by the enzyme and yielded un-or barely-detectable levels of reaction product.The nitro and nitroso derivatives 22 and 23 were accepted as substrates but the turnover was slower than for the natural substrate and was similar to compounds 16 and 17.The K M value for these compounds was higher than the natural substrate and nitroso had a higher K M than the nitro analogue.26 and 27 are substrates for SorbC, a homologue of TropB, and these were chosen for analysis along with 24 and 25. 7 These compound were not substrates for TropB.
Attempts to further purify and characterise the oxidation products failed, since they were not stable in organic solvents. 15owever oxidation of the nitroso compound 23 resulted in the formation of the stable oxime 28 (Fig. 4).The reaction using 23 was scaled up to 16 ml and the product 28 puried by LC-MS.Seven milligrams of the product (99%) was obtained as a colourless oil.The molecular formula C 8 H 9 NO 4 was deduced from the HRMS data (found: 206.0425 [M]Na + ; calcd: 206.0429 for C 8 H 9 NNaO 4 ).Analysis of the 1 H and 13 C-NMR spectra of the product demonstrated that oxidation occurred via hydroxylation at C-2 of the benzenoid ring (see ESI † for data).
In an attempt to rationalise, the substrate selectivity of TropB a homology model of the enzyme was generated.TropB shows 25% identity to the FAD-dependent 3-hydroxybenzoate 6hydroxylase (HB6H) from Rhodococcus jostii RHA1 for which a crystal structure has been obtained (pdb accession 4bjy). 16This was used as a template for the development of a threaded model structure using the SWISS-MODEL protein structure homologymodelling server.The FAD cofactor was placed into the active site of the homology model by transferring the coordinates from the template (4bjy) into the homology model and subjecting the protein-ligand complex to energy minimisation.
To determine the substrate binding mode and possible catalytic mechanism, a sequence alignment was made between TropB and HB6H (Fig. 5).Two residues involved in substrate interactions in 4bjy are conserved in TropB and could play similar roles.These are Y217 (¼ Y239) which H-bonds to the C1 substituent and H213 (¼ H235) which was proposed to act as a base to deprotonate the phenolic hydroxyl.The latter was shown to be essential for catalysis in HB6H by mutagenesis studies.
The TropB substrate 1 was manually docked into the homology model near the FAD in an orientation consistent with the known stereo-chemical outcome of the reaction. 4The resulting protein-ligand complex was energy minimised using the YASARA algorithm (Fig. 6).P329 and I237 make hydrophobic interactions with the substrate and F252 directly interacts with substrate 6-methyl group.N248 is also within Hbonding distance of the carbonyl oxygen (2.4 Å).The substrate and FAD are orientated approximately perpendicular to each other which is similar to the arrangement of these moieties in HB6H. 17239 is located near the 2-hydroxy of the substrate (2.4 Å) and the carbonyl oxygen (1.9 Å), while H235 is near the substrate 4-hydroxy (within 4.5 Å).These two residues could therefore play a role in catalysis (Scheme 3).Deprotonation of the 4-hydroxy by H235 would increase the nucleophilicity of the 3-carbon thus aiding in its attack of the FAD hydroperoxide.The tyrosine could then act as a general acid protonating the resulting alkoxide during or aer reaction.Such an acid-base mechanism would be consistent with the pH prole (Fig. 2) which shows an optimum at pH 8.0.
This mechanism would also be consistent with TropB accepting nitro and nitroso derivatives and would provide an explanation for why TropB rejects the carboxylate derivative 3 as   the carboxylate ion would make deprotonation of the phenolic hydroxyl groups by H235 more difficult.Consistent with our kinetic analyses, the active site in our model is large enough to accommodate alkyl chains of up to 3 carbons beyond the carbonyl oxygen (Fig. 7), but not the bulkier aromatic substrates.Interestingly, docking of bromoalkyl substrate 19 shows that the active site pocket could reorganise to increase the available volume for some substrates, although this is clearly not sufficient to allow 19 or the aromatic substrates 20 and 21 to be efficient substrates.The active site surface also appears to be highly charged and thus is not favourable for binding hydrophobic groups.The kinetic data showed a role for the 6-methyl group in binding and the model shows this group interacts with F252, possibly xing the substrate in the correct orientation for effective deprotonation by H235.

Conclusions
In summary, we have identied the optimum reaction conditions for the FAD dependent monooxygenase TropB and probed the substrate selectivity of this enzyme using a range of newly synthesised substrate analogues.We have identied the key role of a 6-methyl group for substrate binding within the TropB active site.Nitro and nitroso containing substrates are tolerated by the enzyme as are aryl alkyl ketones with chains of 4 carbons or less.Substrates containing branched chains and bicyclic systems are not tolerated.These insights will be of value in guiding the design of new TropB substrates for enzymatic conversion to high-value chemicals and future efforts to engineer the substrate selectivity of these enzymes, thus illustrating the potential utility of TropB as a potential biocatalyst.

Experimental
Chemicals FAD, NADPH, 2,4-dihydroxy-3,5-dimethylacetophenone and 2,4dihydroxy benzaldehyde were purchased from Sigma.NMR was performed using a Varian VNMRS 400 or Varian VNMRS 500 spectrometer.High resolution mass spectrometry was achieved with a Bruker Daltonics Apex IV FT-ICR.Å depending on the orientation of the H.After energy minimisation the H is 6.2 Å away from the His but if the H is rotated to point towards the His then this is reduced to 4.5 Å. Colour scheme: yellow, cofactor; green potential catalytic residues; grey; hydrophobic and H-bonding interactions.

Bioinformatic analysis
Similar amino acid sequences were identied by a BLASTP search (http://blast.ncbi.nlm.nih.gov).Multiple sequence alignments of TropB were performed using Multalin (http:// multalin.toulouse.inra.fr/multalin/) 17and formatted by ESPript (http://espript.ibcp.fr/ESPript/ESPript/). 18olation of natural products 3-Methylorcinaldehyde 1 and 5-hydroxy-3-methylorcin-aldehyde 25 were isolated from T. stipitatus knockout strains DtsL1 and the dienone 2 was isolated from DtsR5 by LC-MS as described before. 8Dihydrosorbicillin 26 and 2,3-sorbicillin 27 were isolated from Penicillium chrysogenum as described before. 7emical synthesis Aryl alkyl ketones.Aryl alkyl ketones were synthesised by Friedel-Cras acylation using acyl chlorides and aromatic compounds according to established protocol. 9The reaction was performed on a small scale with 13.8 mg (0.1 mmol) of 2,5dimethyl-resorcinol and 6.5 mg (0.1 mmol) of very ne zinc powder in a 5 ml reaction vial.Particles were ground together using a small glass stirring rod before the addition of 0.11 mmol of an alkyl or aryl acyl chloride.This was mixed for 5 minutes at room temperature.The reaction was quenched with 2 ml of methanol.The solution was ltered and subjected directly to LC-DAD-MS for analysis and purication.1-Nitroso-2,4-dihydroxy-3,6-dimethylbenzene 23 (ref.13).The mono-nitrosation of 2,5-dimethylresorcinol was achieved using diluted sulphuric acid and sodium nitrite. 12Diluted sulphuric acid (1 ml H 2 SO 4 in 40 ml water) was added drop wise over a vigorously stirred solution of 2,5-dimethylresorcinol (138 mg, 1 mmol) and sodium nitrite (140 mg, 2 mmol) in 10 ml aqueous ethanol 50%.The formation was monitored using TLC and LC-MS during the addition of diluted acid until the reaction reached to 80% completion.The reaction was diluted with 300 ml water and extracted by ethyl acetate.The crude extract was dissolved in methanol and the compounds puried by massdirected LC-MS.

Preparative LC-MS
The purication of compounds was achieved using a Waters mass-directed autopurication system comprising of a Waters 2767 autosampler, Waters 2545 pump system, a Phenomenex LUNA column (5m, C 18 , 100 Å, 10 Â 250 mm) equipped with a Phenomenex Security Guard precolumn (Luna C 5 300 Å) running at 4 ml min À1 and a larger column reversed-phase HPLC (Phenomenex, Kinetex, 5 mm C 18 100 Å, AXIA Packed LC Column 250 Â 21.2 mm) with a ow rate of 16 ml min À1 .Solvents were HPLC grade including H 2 O, CH 3 CN and MeOH with the addition of 0.05% formic acid in all the solvent.The post-column ow was split (100 : 1) and the minority ow was made up with MeOH + 0.045% formic acid to 1 ml min À1 for simultaneous analysis by diode array (Waters 2998), evaporative light scattering (Waters 2424) and ESI mass spectrometry in positive and negative modes (Waters Quatro Micro).Detected peaks were collected into glass test tubes.Combined tubes were evaporated under a ow of dry N 2 gas, weighed, and residues dissolved directly in deuterated solvents for NMR analysis.Solvent gradients for LC/MS analytical and purication were variable but generally runs of 15, 30 and 60 min were designed to start with 95% Water (A) for 2 min to remove any salts present in a sample.Next, the polarity of the solvent system was increased accordingly for good separation.Before the run ended, 2 min wash of 95% organic solvent CH 3 CN (B) or MeOH (C) was applied to ensure the elution of all compounds.Finally, the HPLC column was prepared for the next run by adding 2 min 95% Water at the end of the programme.

Recombinant protein overproduction and purication
Recombinant hexa-histidine tagged TropB produced by overexpression in BL21-CodonPlus (DE3)-RP (Stratagene) using the plasmid pET-28a-TropB originally described by Davison and coworkers. 4Overproduction and purication of the protein was according to established protocols with the exception of the inclusion of 10% glycerol in all purication buffers and the use of a HiLoad 16/60 Superdex 200 column (GE Healthcare) for size exclusion chromatography (see ESI Fig. S1 †).

Protein Identication from SDS-PAGE gels by MALDI
The appropriate coomassie stained band was excised from the SDS-PAGE gel, cut into 1 mm cubes and destained overnight with 50% ethanol.The sample was washed with 100 mM ammonium bicarbonate and dehydrated several times with acetonitrile.Reduction and alkylation was achieved by incubation with 50 mM DTT at 56 C for 1 hour followed by cooling to room temperature and incubation with 50 mM iodoacetamide for 45 minutes in the dark.Gel pieces were washed with 100 mM ammonium bicarbonate and dehydrated with acetonitrile several times before the addition of 12.5 ng ml À1 trypsin solution (Promega) in 50 mM ammonium bicarbonate and incubated at 37 C overnight.Peptide fragments were extracted from the gel by several rounds of sonication in acetonitrile using a water bath sonicator.The solution was dried overnight under vacuum and resuspended in 10 ml of 50% acetonitrile.
Samples were spotted onto a stainless steel plate with an equal volume of matrix (a saturated solution of a-cyano-4hydroxycinnaminic acid in 50% acetonitrile) and le to dry.A mass spectrum was recorded with a MALDI-TOF spectrometer (4700 Proteomics Analyzer, Applied Biosystems) in linear mode.
The peptides masses resulting from this analysis were submitted to the Mascot Peptide Mass Fingerprint search engine (http://www.matrixscience.com;Matrix Science, London, UK) to search for matching proteins in NCBInr database (see ESI Fig. S2 †).

LC-MS activity assay
Enzyme activity was measured in a 1 ml nal volume with 50 mM phosphate buffer pH 8, 2.5 mM for each individual compound and 20 mg ml À1 (0.38 mM) TropB.The reaction was initiated by adding 2.5 mg of NADPH (nal concentration 3 mM) and was incubated for 60 minutes under aerobic conditions at 30 C. The reaction product was extracted by ethylacetate and detected by LC-MS.A Phenomenex LUNA column (5 m, C 18 , 100 Å, 4.6 Â 250 mm) was used with a gradient elution of acetonitrile: 0 min, 10% B; 15 min, 90% B; 16 min, 95% B; 17 min, 95% B; 18 min, 10% B; 20 min, 10% B. Mass spectrometry was performed with a Waters ZQ micromass spectrometer.
Control reactions were carried out under identical conditions but using boiled enzyme.

Determination of optimum temperature and pH
Enzyme activity was measured across a range of 5 to 40 C in 50 mM Tris-HCl buffer (pH 8.0) at a xed concentration of enzyme.NADPH (100 mM) and the natural substrate 3-methylorcinaldehyde (200 mM) were included in each reaction mix.Reactions were performed in a 100 ml volume in transparent 96 well microplates (Costar).Activity was measured by monitoring the initial rate of increase of product formation at 400 nm in a Molecular Devices SpectraMAX 190 microplate reader.Known concentrations of product (dienone) were used to generate a standard curve from which the increase in absorbance could be converted into an increase of product concentration.The reaction rate was recorded from the linear portion of a graph showing product concentration against time.To ensure the reactions occurred at the required temperature, solutions were pre-incubated at that temperature for 30 minutes in either a heat block or cooling shaker.All reactions were repeated in triplicate.
To measure the effect of pH on activity, the above protocol was used except temperature was xed at 30 C and different reaction buffers were used to create a pH range of 5-12.The following buffers were used: 50 mM sodium acetate pH 5.0, sodium phosphate pH 6.0, Tris-HCl pH 7.0, Tris-HCl-8.0,Tris-HCl-9.0,CHES pH 10.0 and CHES pH 12.0.

Determination of kinetic parameters
Reactions were monitored by following either the increase of product formation at 400 nm or the depletion of NADPH at 340 nm using a Carey 50 spectrophotometer (Agilent).In cases where the product absorbance was monitored the reactions were allowed to reach completion and the nal absorbance value was used to calculate the extinction coefficient using the Beer-Lambert law.The extinction coefficient was used to convert absorbance units into product concentration for plotting graphs of concentration against time.When NADPH depletion was monitored the universally recognised extinction coefficient of 6.2 Â 10 3 M À1 cm À1 was used for the same purposes.
Kinetic parameters were estimated by plotting initial reaction rate against substrate concentration and tting the data to the Michaelis-Menten equation using GraphPad Prism version 5.00 for Windows, GraphPad Soware, San Diego California USA, http://www.graphpad.com.

TropB modeling and docking studies
1][22][23] The PDB entry 4bjy was used as the template structure. 17This enzyme shares 25% sequence identity to TropB and is a known FAD dependent hydroxylase.The resulting model was structurally aligned with the template 4bjy in Pymol (The PyMOL Molecular Graphics System, Version 1.3 Schrödinger, LLC.) and the 3D co-ordinates of the FAD ligand were merged from 4bjy into the homology model le.The resulting protein-ligand complex was subjected to energy minimisation using the YASARA energy minimization server. 24The substrate, 3-methylorcinaldehyde 1 was manually positioned within the TropB active site using Pymol and the resulting protein-ligand complex again subjected to energy minimization using the YASARA energy minimization server.To evaluate TropB substrate tolerability 3-methylorcinaldehyde 1 was substituted for a representative selection of compounds listed in Table 1 and their ability to be accommodated with the TropB active site assessed using the same methodology.The electrostatic surface was generated using the PDB2PQR server 25 and the APBS Pymol Plugin. 26

Fig. 3
Fig. 3 Structures of the compounds for evaluating the substrate selectivity of TropB.

Fig. 4
Fig. 4 Conversion of 23 to 28 by TropB over 40 minutes showing consumption of 23 and formation of 28.

Fig. 5
Fig. 5 Alignment of TropB and HB6H (4bjy) showing the conserved His and Tyr residues involved in substrate interactions.

Scheme 3
Scheme 3 Possible mechanism for the hydroxylation of 1 catalysed by TropB.

Fig. 7
Fig. 7 Active site surface of TropB homology model coloured according to electrostatic potential.(A) shows the boundary of the active site surface when the natural substrate (green) was docked.This active site could accommodate up to 3 carbons beyond the carbonyl oxygen.(B) shows the extra space opened up by conformational change during energy minimisation when 19 (grey) was docked.

Fig. 6
Fig. 6 Docking of 3-methylorcinaldehyde inside the active site of the TropB homology model.His-235 and Tyr-239 are conserved from the template (4bjy).His235 to H (hydroxyl) ¼ either 4.5 A or 6.2Å depending on the orientation of the H.After energy minimisation the H is 6.2 Å away from the His but if the H is rotated to point towards the His then this is reduced to 4.5 Å. Colour scheme: yellow, cofactor; green potential catalytic residues; grey; hydrophobic and H-bonding interactions.

Table 1
Substrate tolerance and kinetic parameters of TropB with the substrates tested.All kinetic assays were done in triplicate."Trace" means the amount of product detected was at the lower limit of detection.Errors show standard deviation.n.d.not determined