Effects of Active‐Site Modification and Quaternary Structure on the Regioselectivity of Catechol‐O‐Methyltransferase

Abstract Catechol‐O‐methyltransferase (COMT), an important therapeutic target in the treatment of Parkinson's disease, is also being developed for biocatalytic processes, including vanillin production, although lack of regioselectivity has precluded its more widespread application. By using structural and mechanistic information, regiocomplementary COMT variants were engineered that deliver either meta‐ or para‐methylated catechols. X‐ray crystallography further revealed how the active‐site residues and quaternary structure govern regioselectivity. Finally, analogues of AdoMet are accepted by the regiocomplementary COMT mutants and can be used to prepare alkylated catechols, including ethyl vanillin.

Experimental methods pp. S6-S9. Synthesis of reaction standards pp. S10-S30. Supplementary figures and tables Figure S1. Biosynthesis of vanillin from glucose Figure S2. Methylated catechols in APIs Figure S3. Irreversibility of COMT-catalysed methylation Table S1. Activity and regioselectivity of COMT mutants Figure S4. Activity and regioselectivity of COMT mutants Table S2. Kinetic parameters for selected COMT enzymes Figure S5. pH profile of WT and K144A COMT Table S3. Crystallography data for WT dimeric and Y200L dimeric COMT Figure S6. Preliminary X-ray crystal structure of Y200L COMT with DHBAL and SAH Figure S7. Gel filtration chromatogram for COMT monomer and dimer Figure S8. Calibration for gel filtration chromatography Table S4. WT monomer and dimer activity and regioselectivity Figure S9. Oligomeric forms of COMT in solution Figure S10. Stability of WT COMT under dilution and typical reaction conditions Figure S11. Stability of WT and Y200L monomeric and dimeric forms Figure S12. Regioselectivity of WT and Y200L monomeric and dimeric forms Figure S13. X-ray crystal structure of dimeric COMT with DNC and AdoMet bound Figure S14. AdoMet analogue assays Table S5. Primers used for COMT mutagenesis Figure S15. SDS PAGE of Ni-NTA purification of COMT Figure S16. SDS PAGE of COMT anion exchange and gel filtration purification Figure S17. HPLC analyses of COMT assays with authentic standards Figure S18. Calibration curves for substrates 1a-d and products 2a-d and 3a-d Figure S19. 1 H NMR of enzymatic 4a Figure S20. HPLC chromatogram of 4a Figure S21. 1 H NMR of enzymatic 4b Figure S22. HPLC chromatogram of 4b Figure S23. 1 H NMR of enzymatic 4c Figure S24. HPLC chromatogram of 4c p. S31.

Michaelis-Menten enzyme kinetics.
To determine the kinetic constants of wild-type and mutant COMT, enzyme (0.2-0.5 µM) was assayed against a range of substrate concentrations (2-5000 µM) with the initial rate of reaction measured by monitoring substrate conversion over several time points. The rates were plotted against substrate concentrations using Sigmaplot 12.0 and the Km and kcat constants generated from the resulting Michaelis-Menten plot.
Crystallogenesis. The sitting drop vapour diffusion technique was used to grow crystals of both wild type COMT dimer and the Y200L mutant. The purified protein was collected, diluted (1 µM) and mixed with SAM (10 µM) and DNC (10 µM) overnight at 4 ˚C through gentle rocking. The COMT proteins were then concentrated to a final concentration of 10 mg/mL with bound DNC and SAM ligands in a 20 mM Tris buffer (with 300 mM NaCl) at pH 7. Crystallography data collection and structure determination. Data were collected from single cryo frozen crystals at Diamond Light Source, full details of data and refinement statistics are presented in Table S2. The data was scaled and integrated using Xia2 [S1] and the structures subsequently solved by molecular replacement in Phaser. [S2] All models were subsequently completed and refined using iterative cycles of rebuilding and refinement in COOT [S3] and Phenix.refine. [S4] Validation with Molprobity was integrated into the iterative rebuilding and refinement cycle. [S5] Final models with R and Rf of 15.5 & 18.8 for the WT dimer structure and 15.9 and 17.8 for the Y200L dimer structure with DNC have been deposited with the protein data bank, accession codes 5FHQ, 5FHR.

3-allyloxy-4-hydroxybenzaldehyde (4b).
60% sodium hydride in mineral oil (3.8 g, 81 mmol) was washed free of oil with four portions of dry hexane (30 mL) under nitrogen. Dry DMSO (60 mL) was then added and the mixture was then cooled to 0 ˚C with stirring. A solution of 3,4-dihydroxybenzaldehyde (1a) (5.5 g, 40 mmol) in dry DMSO (20 mL) was then added dropwise. The suspension was stirred until no solid was visible, whereupon a solution of allyl bromide (4.8 g, 3.5 mL, 40 mmol) in dry DMSO (20 mL) was added dropwise. The ice bath was then removed and the solution was left to reach room temperature whilst stirring overnight. The solution was then added to ice cool water (100 mL) and acidified with aqueous HCl (1.0 M). The product was then extracted with ethyl acetate (3 x 100 mL). The organic layers were combined, washed with brine and dried over magnesium sulphate. The ethyl acetate was then removed under reduced pressure. The product was then purified through flash chromatography using hexane/ethyl acetate/acetic acid (80:20:1) as the eluant to give 3-allyloxy-4-hydroxybenzaldehyde (

3-benzoxy-4-hydroxybenzaldehyde (4c)
. 60% sodium hydride in mineral oil (3.8 g, 81 mmol) was washed free of oil with four portions of dry hexane (30 mL) under nitrogen. Dry DMSO (60 mL) was then added and the mixture was then cooled to 0 ˚C with stirring. A solution of 3,4-dihydroxybenzaldehyde (1a) (5.5 g, 40 mmol) in dry DMSO (20 mL) was then added dropwise. The suspension was stirred until no solid was visible, whereupon a solution of benzyl bromide (3.9 g, 2.7 mL, 23 mmol) in dry DMSO (20 mL) was added dropwise. The ice bath was then removed and the solution was left to reach room temperature whilst stirring overnight. The solution was then added to ice cool water (100 mL) and acidified with aqueous HCl (1.0 M). The product was then extracted with ethyl acetate (3 x 100 mL). The organic layers were combined, washed with brine and dried over magnesium sulphate. The ethyl acetate was then removed under reduced pressure. The product was then purified through flash chromatography using hexane/ethyl acetate/acetic acid (80:20:1) as the eluant to give 3-benzoxy-4hydroxybenzaldehyde ( 3-hydroxy-4-benzoxybenzaldehyde (5c). A suspension of 3,4-dihydroxybenzaldehyde (1a) (3.1 g, 23 mmol), acetone (100 mL), potassium carbonate (3.1 g, 23 mmol) and benzyl bromide (3.9 g, 2.7 mL, 23 mmol) was stirred for 5 hours at 60 ˚C. After 4 hours the suspension was filtered and the acetone removed under reduced pressure.

S9
The resulting residue was dissolved in diethyl ether (50 mL), mixed with H2O and acidified to pH 3 with aqueous sulphuric acid (6.0 M). The layers were then separated and the aqueous layer further extracted with diethyl ether (2 x 50 mL). The organic fractions were combined and dried over magnesium sulphate. The solvent was removed under reduced pressure and the crude product was purified by flash chromatography (80:20:1, hexane/ethyl acetate/acetic acid) to give 4-benzoxy-3-hydroxybenzaldehyde (0.30 g, 7%  Figure S1. Engineered biosynthetic pathway to vanillin from glucose in fission yeast. [3] 3DHSD: 3dehydroshikimate dehydratase; ACAR: aromatic carboxylic acid reductase; UGT: UDP-glycosyltransferase. Figure S2. Methylated catechols as potential building blocks in APIs. [S7-S12] Table S1. Percentage conversions and regiomeric excesses of COMT mutants with substrates 1a-d. 1a and 1b are key intermediates in the biosynthesis of vanillin from glucose, 1c possesses an ionisable R-group that can be influenced by the active site mutations introduced, whereas 1d was selected due to its potential as a compound for development of colormetric screens for methyltransferase activity. Conversions were calculated as the percentage of substrate converted to metaand para-products combined. Regioisomer excesses (r.e.s) were calculated as the percentage excess of the major regioisomer over the minor regioisomer. Positive r.e.s denote a regiomeric excess of the meta-isomer, whereas negative r.e.s denote excess of the para-isomer. Figure S4. Comparison of activity and regioselectivity between substrates 1a-d for all COMT mutants. Conversions are shown as blue bars, r.e.s are shown as red bars. Conversions were calculated as the percentage of substrate converted to metaand para-products combined. Positive r.e.s denote a regioisomer excess of the metaregioisomer, whereas negative r.e.s denote excess of the para-regioisomer. Table S2. Michaelis-Menten kinetics of WT COMT, WT COMT monomeric and dimeric forms, and mutants. Kinetic parameters were determined for those mutants that showed most significant shifts in regioselectivity. Mutations W38D and W38R perturbed the binding of substrates DHBAL 1a and NOC 1d to a greater extent, leading to higher Km values. This is consistent with the structural studies which suggest that W38 could π-stack with the catechol ring and thereby contribute towards substrate binding. [5a, S13, S14] The Km values for Y200L with 1a and 1d are similar to those of the wild-type enzyme; Y200 is more distal to the catechol ring and is thus less likely to make contact with and affect the substrate binding affinity. The Km values were significantly higher for all mutants with DHBA 1b, indicating that the binding of this substrate is more sensitive to changes in the active site.

DHBAL
The kcat values for W38D, W38R and Y200L were similar to that of the wild-type, whilst K144 mutants generally exhibited lower catalytic rates. Surprisingly, given the postulated role of K144 as the catalytic general base [5a, S14, S15] significant enzyme activity was still retained with K144 mutants (Fig. 2). Indeed with 1d as a substrate, K144A/V173Y exhibited similar turnover number to the WT COMT, albeit with lower efficiency. This suggests that whilst K144 may participate in deprotonating the catechol hydroxyl group, it is not essential. Presumably the coordination of the hydroxyl groups with the Mg 2+ cation, and the presence of the more electron-withdrawing nitro group of 1d can serve to lower the pKa of the substrate hydroxyl group sufficiently to allow a water molecule to abstract the proton. To explore this phenomenon further, the effect of pH on enzyme activity was explored with the wild-type and K144A mutant and the substrate 1a (Fig. S5).  Figure S5. pH profile of WT and K144A COMT activity with substrate DHBAL 1a. The WT and K144A enzymes were assayed with substrate 1a over a pH range of 5.29 to 8.55. The conversion of 1a to metaand para-methylated products combined for each enzyme at each pH was plotted to give the pH dependent activity profile. Over a pH range of 5.3-8.6, little change in activity occurred with the wild-type enzyme, but a marked increase in activity was observed with increasing pH for the K144A mutant. Previous literature has suggested that K144 acts as a catalytic base to deprotonate the hydroxyl group closest to the AdoMet sulphonium centre, [S14-S16] with the aid of Mg 2+ coordination lowering the hydroxyl pKas. [5a] We suggest that in the absence of K144, increasing solvent pH results in greater ionization of the catechol hydroxyl and faster methylation. With the wild-type enzyme, the presence of K144 as a general base results in substrate deprotonation that is independent of the solvent pH.   Figure S6. Model of the active site of Y200L with SAH and active substrate DHBAL 1a based on a preliminary crystal structure. The binding modes of 1a and SAH are consistent with those observed in the structures of Y200L and WT dimer with DNC and AdoMet. The loss of E199 interaction with the aldehyde of 1a leads to preferential binding in the orientation for meta-methylation i.e. the aldehyde is positioned at the solvent interface.  Figure S8. Gel filtration calibration with protein standards. The calibration curve was plotted with a series of proteins with known molecular weights, and used to verify the molecular weights of COMT monomeric and dimeric forms as predicted by gel filtration FPLC. The monomeric and dimeric forms of WT COMT were separated by gel filtration FPLC and assayed with substrates 1a, 1b and 1d at 37 °C, 800 rpm shaking, 20 min. Conversions and r.e.s were calculated as described previously. Kinetic parameters were also determined for monomeric and dimeric COMT with 1a (Table S2). The monomeric form showed a higher catalytic rate per active site (2.3 ± 0.050 min −1 ) compared with the dimer (1.6 ± 0.030 min −1 ), and a lower Km (1.3 ± 0.18 μM monomer, 3.7 ± 0.28 μM dimer), as reflected in the activity assays shown above. Thus, monomeric COMT appears to bind more tightly and with greater reaction turnovers, yet exhibits considerably lower regioselectivity than the dimer.  Figure S9. Oligomeric forms of WT and Y200L in solution. The ratio of dimer to monomer by peak area for WT COMT was determined to be 0.8:1, whereas for Y200L the monomer:dimer ratio was 1.1:1. Figure S10. Stability of COMT WT dimer and monomer under dilution and incubation under assay conditions. The dimeric and monomeric forms were isolated and diluted from the gel filtration FPLC purification concentration (~6 mg/mL 232 µM) to an approximate assay concentration of 0.5 mg/mL (19 µM). Dilution does not appear to cause any change in oligomeric state, with both dimer and monomer remaining at 100% (blue lines). Incubation of the same samples for 20 min under assay conditions (37 °C, 800 rpm agitation) results in a decrease in the amount of the original oligomeric state (red lines), as some of the other oligomeric state begins to be formed.  Figure S11. Stability of WT and Y200L COMT oligomeric forms over time. The monomeric and dimeric forms were separated, diluted to approximately 15 µM and incubated at 37 °C with 800 rpm agitation for one and three minutes before analysing by gel filtration FPLC. Both WT and Y200L monomer and dimer remained mostly stable over a shorter time period (one minute) and in their initial oligomeric state. Figure S12. Regioselectivity of monomeric and dimeric COMT over time. The monomeric and dimeric forms of WT and Y200L COMT were assayed with DHBAL 1a for one, two and three minutes (15 µM enzyme, 0.5 mM 1a). The WT monomer showed low regioselectivity (+39% r.e.) whereas the dimer showed a high r.e. of +85%. In contrast, with Y200L both monomeric and dimeric forms possessed high r.e.s of +86 and +92% respectively. , and the E199-Y200 region pulled out of the active site (long arrows). Given that the wild-type dimeric form shows the E199 residue flipping out of the active site in a similar manner to that observed with the Y200L mutant, it is unsurprising that the higher Km for Y200L with DHBAL 1a is also reflected in wild-type dimeric COMT (Table  S2), and that the r.e.s of dimeric COMT with 1a and 1d are similar to that of Y200L (Table S1, Fig. S11). The regioselectivity assay data for the WT and mutant COMT enzymes described in Fig. S4 and Table S1 were obtained using protein solutions prepared from Ni-NTA purifications. Further purification of WT COMT by gel filtration FPLC revealed that in solution, COMT exists as a mixture of monomeric and dimeric forms (Fig. S7). Separation of the WT monomer and dimer and determining the regioselectivity of each with substrates 1a, 1b and 1d revealed low r.e.s for the monomeric form but high r.e.s for the dimer (Table S4). In order to investigate whether the Y200L mutation was in itself responsible for loss of the E199-substrate interaction, or whether the mutation was causing the enzyme to preferentially form a meta-regioselective dimeric state, the Y200L COMT mutant was also analysed by gel filtration FPLC (Fig. S9). The small difference in the dimer:monomer ratio for Y200L (1.1:1) compared with WT (0.8:1) did not appear to correlate well with the large difference in meta:para ratios between Y200L (19.5:1) and WT (3.3:1). Separation of the Y200L monomer and dimer by FPLC and incubation under assay conditions (37 °C with 800 rpm agitation) revealed that under a short time duration (one minute), the monomeric and dimeric forms remained mostly stable (Fig. S11). Reactions of Y200L monomer and dimer under the same conditions (one to three minutes duration to ensure stability of the separated monomeric and dimeric forms) with substrate 1a revealed high meta-r.e.s for both forms (Fig. S12) unlike with the WT COMT. Thus, with the Ni-NTA purified COMT assays (Fig. S4, Table S1), the 54% r.e. observed with WT COMT is derived from a mixture of highly regioselective dimer and low regioselectivity monomer. In contrast, Y200L monomer and dimer are both meta-regioselective.

Substrate
The crystal structure of dimeric WT COMT (Fig. S13) reveals movement of E199 out of the active site in a manner similar with that of the Y200L mutant, which was also crystallised as a dimer. However, whereas the WT monomer, with E199 capable of forming an H-bond with the substrate R-group, shows low meta-selectivity, the Y200L monomer has a meta-selectivity on par with that of both WT and Y200L dimeric forms. Thus, we suggest that with WT COMT the dimerization and domain swapping appear to cause loss of the E199-substrate H-bonding, whereas with Y200L the mutation alone is sufficient, allowing the monomer to be equally regioselective as the dimer.

B.
A.    Figure S17. HPLC chromatograms of wild-type COMT assays compared with authentic standards for compounds 1-3(a-d). 5′-S-methyl-thioadenosine (MTA) is a degradation product of AdoMet. [S21] Figure S18. Calibration curves for substrates 1a-d and products 2-3(a-d). Calibrations were used to adjust HPLC peak areas to compensate for differences in extinction coefficients. All calibrations were run in triplicate.        Figure S23. 1 H NMR of enzymatically generated benzyl vanillin 4c in CD3CN. Figure S24. HPLC chromatogram of enzymatic and synthetic benzyl vanillin 4c, tR = 7.37 min.