Isotope effects suggest a stepwise mechanism for Berberine Bridge Enzyme

Abstract

The flavoprotein Berberine Bridge Enzyme (BBE) catalyzes the regioselective oxidative cyclization of (S)-reticuline to (S)-scoulerine in an alkaloid biosynthetic pathway. A series of solvent and substrate deuterium kinetic isotope effect studies were conducted in order to discriminate between a concerted mechanism in which deprotonation of the substrate phenol occurs before or during hydride transfer from the substrate to the flavin cofactor and substrate cyclization, and a stepwise mechanism, in which hydride transfer results in the formation of a methylene iminium ion intermediate that is subsequently cyclized. The substrate deuterium isotope effect of 3.5 on k_red, the rate constant for flavin reduction, is pH-independent, indicating that C-H bond cleavage is rate-limiting during flavin reduction. Solvent isotope effects on k_red are one for both wild-type BBE and the E417Q mutant, indicating that solvent exchangeable protons are not in flight during or before flavin reduction, thus eliminating a fully concerted mechanism as a possibility for catalysis by BBE. An intermediate was not detected by rapid chemical quench or continuous-flow mass spectrometry experiments, indicating that it must be short-lived.

Berberine Bridge Enzyme (BBE) is a plant enzyme that participates in an alkaloid biosynthetic pathway that produces a variety of pharmacologically active alkaloids, including berberine, sanguinarine, and palmatine (1). The enzyme, an FAD-containing amine oxidase, catalyzes a unique oxidative cyclization reaction, converting (S)-reticuline to (S)-scoulerine (Scheme 1) (2). This regioselective reaction has not yet been reproduced by synthetic organic methods. Furthermore, only a couple of other enzymes have been identified that catalyze similar reactions: cannabidiolic acid synthase and Δ¹-tetrahydrocannabinolic acid synthase from Cannabis sativa (3, 4). Neither of these enzymes has been characterized enzymatically.

Scheme 1.

Reaction catalyzed by BBE.

Stopped-flow analyses of the BBE reductive half-reaction have failed to identify a flavin semiquinone intermediate, suggesting that substrate oxidation does not proceed via single electron transfer from the substrate to the cofactor (5) and instead likely occurs via hydride transfer, as observed with other flavoprotein oxidases (6, 7). Both concerted and stepwise mechanisms have been proposed for BBE (Scheme 2) (8, 9). In the stepwise mechanism, a hydride equivalent is transferred from the substrate to the flavin, forming a methylene iminium ion intermediate; subsequently, an active site base, Glu417, deprotonates a substrate phenol, making the adjacent carbon more nucleophilic and enabling its attack on the N-methylene group, yielding the cyclized product. In the concerted mechanism, deprotonation of the phenol occurs during concomitant attack on the N-methyl group by the C2′ and transfer of a hydride to the flavin.

Scheme 2.

Proposed mechanisms for the BBE-catalyzed reaction (8, 9)

Several studies support each mechanistic proposal. Oxidation by BBE of (S)-N-methylcoclaurine yields a demethylated product, suggesting that the reaction proceeds via formation of a methylene iminium ion intermediate (Scheme 3) (8). BBE oxidizes the tetrahydroprotoberberine alkaloids (S)-coreximine, (S)-scoulerine and (S)-norsteponine, forming a double bond between the nitrogen and C-8 of each (Scheme 3) (8, 10); these observations also suggest a two-step mechanism for the oxidative cyclization of (S)-reticuline (8). However, analyses of the products resulting from a different set of substrate analogs have been taken as support for a concerted mechanism (9). Oxidation of laudanosine is associated with an absorbance increase around 360 nm, suggesting formation of a double bond within the isoquinoline ring system (Scheme 3). BBE oxidizes the C1-N bond of 6,7-dimethoxy-2-methyl-1,2,3,4-tetrahydroisoquinoline, as opposed to the N-CH₃ bond expected in a two-step mechanism (Scheme 3) (9). Mutation of the proposed active site base Glu417 to glutamine leads to a 1500-fold decrease in the rate constant for flavin reduction, resulting in substrate oxidation becoming the rate-determining step during turnover (9). This has been interpreted as additional evidence for a concerted mechanism in which deprotonation of the substrate phenol is necessary for the concerted attack on the incipient methyl aminium and the transfer of a hydride to the flavin cofactor. In the present study, we performed a series of substrate deuterium and solvent kinetic isotope effect studies to discriminate between these two mechanistic proposals.

Scheme 3.

BBE-catalyzed oxidation of various substrate analogs (8–10).

MATERIALS AND METHODS

Materials

All reagents were purchased from Sigma-Aldrich Corp. (St. Louis, MO) or Fisher Scientific (Waltham, MA) unless otherwise stated. Superdex 200 prep grade and Sephacryl S-200 FPLC columns were purchased from GE Life Sciences (Pittsburgh, PA). (R,S)-Norreticuline was purchased from Hang Zhou Sage Chemical Co. (Hang Zhou, China).

Syntheses

(R,S)-Reticuline was synthesized from (R,S)-norreticuline in two steps, using the method of Rice and Brossi (11). Deuterated (R,S)-reticuline (-N-CD₃) was synthesized with the same protocol using ethyl formate-d₁ (CDN Isotopes, Quebec, Canada) and BD₃ in THF. ¹H NMR (Bruker Avance 500 MHz) analysis of the deuterated substrate yielded essentially the same NMR spectrum as that reported for protiated (R,S)-reticuline (12), but with the singlet of the -N-CH₃ protons at 2.45 ppm missing.

Protein Expression and Purification

Wild-type BBE and the E417Q mutant were expressed, purified, and quantified as described previously (13, 14); protein was expressed with Pichia pastoris strains provided by Professor Peter Macheroux (University of Graz, Austria). Volumes were adjusted for a 2.5-l fermentation in a BioFlo/Celligen 115 Fermentor (Edison, NJ). A Sephacryl S-200 column was sometimes used instead of a Superdex 200 prep grade column for purification of BBE. An additional chromatographic purification step was added during the purification of the E417Q mutant: fractions eluted from the Superdex 200 column that contained BBE were concentrated, dialyzed against buffer containing no salt (100 mM Tris, pH 8.0), applied onto a Mono Q column equilibrated with the same buffer, and eluted with a gradient of increasing salt concentration (0–250 mM NaCl in 100 mM Tris buffer, pH 8.0).

Enzyme Assays

HPLC-based steady-state assays with BBE were based on those described previously (14), using 0.5 nM enzyme and varying concentrations of (R,S)-reticuline (0.19–6.0 μM) in 50 mM of the appropriate buffer at 25 °C. The assays were initiated by the addition of enzyme. The following buffers were used: MOPS, pH 6.75–7.75; Bicine, pH 7.75–8.75; and CHES, pH 8.75–9.75. Instead of the continuous assay described previously, reactions were quenched with an equal volume of 1 N NaOH at varying time intervals (40 s – 20 min). Subsequently, the pH was neutralized with the same volume of either 1 N HCl or 1 N acetic acid before quantifying product formation by HPLC analysis. The HPLC analysis was performed as described previously (14), but using a Phenomenex (Torrance, CA) C18 column (250 × 4.6 mm) and a Waters (Milford, MA) HPLC system. Substrate and product were separated using an isocratic elution of 65% methanol and 35% 50 mM potassium phosphate, pH 7.0.

Wild-type enzyme k_cat values were determined in 50 mM CHES, pD 9.5, in 98–99% D₂O, and wild-type and mutant flavin reduction rate constants were measured in 100 mM of the appropriate buffer listed above, pD 7–10, in 94–95% D₂O. All experiments were performed at saturating substrate concentrations (100–400 μM (R,S)-reticuline or 100–400 μM deuterated-(R,S)-reticuline). Solvent isotope effects were calculated by directly comparing the kinetic parameters obtained in H₂O and D₂O. The solvent isotope effect on k_cat was determined by comparing rates measured at pH 9 and pD 9.5. The effects on k_red with the wild-type enzyme were calculated using the pL-independent values obtained by fitting the entire k_red-pL curves to eq 1, and the effects on k_red with the E417Q mutant were determined by comparing the values in the pH-independent region, pH 8.5–9.5 and pD 9–10.

The dependence of BBE activity on oxygen concentration was determined using an oxygen electrode (Model 5300A, Yellow Springs Instruments, Yellow Springs, OH) to measure oxygen consumption at 25 °C. A 1-ml reaction contained 50 mM CHES, pH 9, and 200 μM (R,S)-reticuline, and the reaction was initiated by the addition of BBE (0.25 or 0.125 μM final). The oxygen concentration was varied from 20–1250 μM by bubbling the reaction solution with the appropriate concentration of oxygen for 15 minutes before addition of enzyme.

Flavin reduction during substrate oxidation was monitored at 445 nm using an Applied Photophysics (United Kingdom) SX-20 MV stopped-flow spectrophotometer under anaerobic conditions. The system was made anaerobic by overnight incubation of the flow cell and syringes with 100 mM appropriate buffer containing 5 mM glucose. The buffer was deoxygenated in a tonometer by exposing it to 16 cycles of vacuum and oxygen-scrubbed argon; subsequently, glucose oxidase (30 nM final) was added to maintain anaerobic conditions and remove any remaining oxygen. The following buffers were used: MES, pH 6.0; MOPS, pH 6.75–7.75; Bicine, pH 8.0–8.55; and CHES, pH 9–9.75. Substrate solutions were made anaerobic by bubbling them with oxygen-scrubbed argon for 5 minutes. Oxygen was removed from enzyme solutions by exposing them to 16 cycles of vacuum and oxygen-scrubbed argon. Both substrate and enzyme solutions included 5 mM glucose, and 30 nM glucose oxidase was added after either bubbling with argon or the vacuum/argon cycles. Experiments were performed using final concentrations of 100 mM buffer, 17 μM BBE, and 75–400 μM (R,S)-reticuline. Substrate deuterium kinetic isotope effects on the rate constant for reduction were measured in 100 mM buffer under anaerobic conditions with saturating deuterated substrate concentrations (100–400 μM (R,S)-reticuline) at 25 °C.

Rapid-Quench Analyses

Rapid-quench experiments were performed in air-saturated buffers using an SFM-400 Bio-Logic (France) instrument. Reactions contained final concentrations of 10 μM BBE, 200 μM (R,S)-reticuline, and 100 mM CHES, pH 9.0, at 25 °C. Reactions were quenched with 0.67 M NaOH. The solution was neutralized with 1 N acetic acid before HPLC analysis to determine the amount of product formed.

Continuous Mass Spectrometry (MS) Experiments

BBE reaction progress was analyzed using an LTQ Orbitrap Discovery Mass Spectrometer (Thermo Scientific) equipped with a custom nano-electrospray ionization source containing a small volume continuous-flow mixer driven by a NanoFlow Metering System (Eksigent, Dublin, CA). Wild-type and E417Q BBE were exchanged into 50 mM ammonium acetate, pH 9.0, using a series of concentration and dilution steps. Reactions were initiated by mixing substrate and enzyme solutions in a 1:1 ratio at varying flow rates of 1–10 μl/min at 10 and 25 °C; the delay volume was approximately 0.04 μl, yielding reactions times of 0.25 – 2.5 s. The reactions consisted of final concentrations of 10 μM wild-type or E417Q BBE and 40 or 400 μM (R,S)-reticuline in 50 mM ammonium acetate, pH 9.0. The following were monitored by high-resolution MS: reticuline (m/z 330.1705), products scoulerine and coreximine, (m/z 328.1549), the imine intermediate (m/z 328.1549) and the demethylated intermediate norreticuline (m/z 316.1549).

Data Analysis

Kinetic data were analyzed using the KaleidaGraph program (Synergy Software, Reading, PA). Steady-state data were fit to the Michaelis-Menten equation to determine k_cat and K_m values. pH-Dependence data from steady-state and stopped-flow experiments were analyzed using equation 1, which accounts for a single pK_a value with rates increasing to a constant value at high pH. Here Y is the parameter being measured, C is the pH-independent value of the parameter, H is the hydrogen ion concentration, and K is the dissociation constant of the ionizable group. Stopped-flow experiments measured the rate of flavin absorbance decrease during reduction of the flavin cofactor by substrate. For wild-type BBE, the flavin absorbance decrease over time was fit to equation 2; here A is the absorbance at time t, A_∞ is the final absorbance, and A₁ and A₂ are the absorbance changes associated with each of the first-order rate constants, k₁ and k₂, respectively. One of the first-order rate constants, attributed to k_red, corresponds to the majority of the flavin absorbance change, while the other constant, likely due to product release, is approximately 10-fold slower and corresponds to only 10–20% of the absorbance change. Rate constants were measured at three saturating substrate concentrations in triplicate and averaged. The E417Q mutant stopped-flow data were fit to Equation 3, which describes a single exponential decay.

$$logY=\frac{\text{C}}{1+\frac{\text{H}}{\text{K}}}$$ (1)

$$A=A_{\infty }+A_1{\text{e}}^{-k1t}+A_2{\text{e}}^{-k2t}$$ (2)

$$A=A_{\infty }+{A\text{e}}^{-kt}$$ (3)

RESULTS AND DISCUSSION

Steady-state kinetics

BBE activity under steady-conditions was determined using an HPLC-based assay to quantify product formation and an oxygen electrode to measure oxygen consumption. The resulting steady-state kinetic parameters are listed in Table 1. The HPLC-based assay performed in air-saturated buffers at 25 °C yielded a k_cat(app) value consistent with the value previously reported at 37 °C (14). The K_m(app) value of (S)-reticuline for BBE has been most recently reported to be 3 μM (15), but has been previously reported to be as low as 0.14 μM (16), and this latter lower value is more consistent with our observations. The k_cat and oxygen K_m values were measured at varying oxygen concentrations (0.02 – 1.25 mM) at 25 °C. The oxygen K_m value of 280 ± 70 μM (Table 1) indicates that assays performed in air-saturated buffers ([O₂] ≈ 250 μM) only yield apparent k_cat values. The k_cat/K_O2 value is consistent with the k_ox rate constant for flavin reoxidation that has been reported previously (5).

Table 1.

Kinetic Parameters of Wild-Type BBE^a

parameter	value
kcat(app)	5.4 ± 0.3 s−1
Km	0.27 ± 0.03 μM
kcat/Km(app)	20 ± 2 μM−1 s−1
kcat	10.5 ± 0.7 s−1
Km(O2)	280 ± 70 μM
kcat/Km(O2)	37 ± 9 mM−1 s−1
kred	98 ± 4 s−1

^aAll kinetic constants were measured at 25 °C in 50 or 100 mM CHES, pH 9.0 with wild-type BBE. The k_cat(app) and K_m values were determined in air-saturated buffer and k_red was determined in anaerobic solutions.

Measurement of substrate deuterium and solvent kinetic isotope effects necessitates identification of a pH-independent region of activity for accurate analyses of results. The k_cat(app)-pH profile with wild-type BBE at 25 °C fit to a curve with a pK_a of pH 7.8 ± 0.1 for a single ionizable group that must be deprotonated for maximal activity (Figure 1). It was previously reported that BBE activity is maximal at pH 8.9, but decreases below and above this pH value (2). The decrease in enzymatic activity observed above pH 9 could be due to enzyme instability at high pH over the prolonged, hour-long incubation times utilized in that study, compared to the initial rate assay utilized here. The solvent kinetic isotope effect on k_cat was measured in the pH-independent region, at pH 9 and pD 9.5, in buffers containing 1.25 mM O₂, yielding a value of 1.4 ± 0.1 (Table 2). A solvent isotope effect on k_cat/K_m could not be determined accurately due to the low (S)-reticuline K_m value and the sensitivity limits of the assay.

Figure 1.

The effects of pH on k_cat(app) at 25 °C. The curve is a fit of the data to eq 1.

Table 2.

Kinetic Isotope Effects for Wild-Type and E417Q BBE^a

	Wild-Type	E417Q
D2O(kcat)H	1.4 ± 0.1	N.D.
D2O(kred)H	1.00 ± 0.02	0.96 ± 0.05
D2O(kred)D	0.88 ± 0.10	0.91 ± 0.03
D(kred)H2O	3.5 ± 0.3	6.7 ± 0.3
D(kred)D2O	3.2 ± 0.3	6.3 ± 0.2

^aConditions: 25 °C in either 50 or 100 mM CHES at pH 9.0 or pD 9.5. N.D. - not determined. The ^D2O(k_red)_H values were determined from the rate constant for flavin reduction in the pL-independent region measured in H₂O and D₂O.

Rapid-reaction kinetics

The modest solvent isotope effect suggested that amine oxidation may not be rate-limiting for turnover by BBE (5). Consequently, we measured isotope effects on k_red, the rate constant for flavin reduction at saturating substrate concentrations, using stopped-flow spectroscopy (Figure 2). The overall shape of the k_red-pH profile was similar to that for k_cat; the data fit to the same equation yielded a pK_a value of 7.3 ± 0.1 (Figure 3). This pK_a value is likely due to the necessity for the substrate nitrogen to be deprotonated for oxidation; this has been observed with other flavin amine oxidases (17–22). The pK_a value of the tertiary amine in reticuline does not appear to have been reported, but it is likely close to that of trimethylamine of approximately 9.8. While we cannot rule out the possibility that the pK_a is due to an amino acid residue in the protein, the most likely candidates are Glu417, His459 and Tyr106; mutagenesis of His459 or Tyr106 does not have a drastic effect on k_red (9), while the data below rule out Glu417.

Figure 2.

Reduction of wild-type BBE by protiated (circles) and methyl-deuterated (squares) reticuline at 25 °C. Curves are fits to eq 2. Only 10% of the data points are shown for clarity.

Figure 3.

The effects of pH on k_red at 25 °C with protiated (circles) and methyl-deuterated (squares) reticuline in H₂O and with protiated reticuline in D₂O (open circles). The curves are fits to eq 1.

Substrate deuterium kinetic isotope effects on k_red were measured over pH 6–10; Figure 2 compares flavin reduction at pH 9 with protiated and deuterated reticuline. The ^Dk_red values are pH-independent with the wild-type enzyme (Figure 3, Table 2), demonstrating that reduction is not limited by substrate binding (23). The k_red-pH profile with the deuterated substrate yielded a similar pK_a value of 7.4 ± 0.1 (Figure 3).

The k_red was measured at varying pD values; the pK_a value increased to 7.9 ± 0.1 in D₂O (Figure 3). In contrast to the solvent isotope effect on k_cat, the pH-independent value for the solvent isotope effect on the rate constant for flavin reduction is 1.00 ± 0.02, within error of one. The value is similar with the deuterated substrate (Table 2). Solvent isotope effects near one on the rate of flavin reduction indicate that the substrate phenolic proton is not in flight during C-H bond cleavage. The ^Dk_red did not change significantly in D₂O, consistent with CH bond cleavage remaining rate-limiting for flavin reduction.

Kinetics of E417Q BBE

The possibility of rapid and equilibrium deprotonation before slower C-H bond cleavage remained; hence, isotope effects were measured with a variant lacking the proposed glutamate active-site base. In the E417Q mutant, the k_red value decreases approximately 1500-fold and becomes equal to k_cat, which itself decreases, indicating that reduction is rate-limiting for turnover with this enzyme (9). If the glutamate residue functions to deprotonate the phenolic proton, a significant solvent isotope effect should be observed in the mutant protein as the deprotonation becomes more rate-limiting (24).

The k_red with the E417Q mutant was measured over pH 7–9.5; the pK_a value decreased to 6.5 ± 0.1 with this variant (Figure 4). The deuterium kinetic isotope effect is 6.7 ± 0.3 at pH 9 with the E417Q mutant (Table 2). This effect indicates that C-H bond cleavage is also rate-limiting during flavin reduction in the mutant enzyme. The elevated substrate deuterium kinetic isotope effects in the mutein may be due to longer donor-acceptor distances in the mutein, allowing tunneling to occur only with the protiated substrate, compared to wild-type enzyme where tunneling may occur with both isotopes due to the short distances between the donor and acceptor, resulting in lower substrate deuterium kinetic isotope effects (25). The solvent kinetic isotope effect on k_red with the E417Q mutant with both the protiated and deuterated substrates is the same, with an average value of 0.94 ± 0.04 (Table 2), indicating that solvent-exchangeable protons are not in flight during flavin reduction even in this slow mutant enzyme.

Figure 4.

The effects of pH on k_red at 25 °C with the BBE E417Q mutant. The curve is a fit to eq 1.

Product analyses

The dramatic decrease in the flavin reduction rate with the E417Q mutant enzyme has been used as evidence supporting a concerted mechanism for BBE catalysis (9). The decrease in the rate of flavin reduction could, however, also result from improper positioning of the substrate in the active site of the mutant enzyme. Hydride transfer to the flavin cofactor requires precise positioning of the substrate by the flavin cofactor (26). Indeed, as demonstrated previously, the mutant enzyme yields two products, coreximine in addition to scoulerine (see Scheme 3 for structure of coreximine) (9). Coreximine could only be produced if the substituted benzene ring of the substrate were more flexible than in the wild-type enzyme; coreximine’s production requires a 180° rotation of the aromatic ring in the active site (9), demonstrating that substrate binding in the mutant and wild-type enzyme are not the same.

A stepwise mechanism involves formation of a discrete iminium intermediate. Rapid-quench analyses of the reaction with the WT and E417Q enzymes (10 μM) were performed to obtain direct evidence for this intermediate. The imine would be expected to hydrolyze to its demethylated form, norreticuline, after the quench in 0.7 M NaOH. The lower limit for detection of norreticuline with HPLC analysis after rapid-quench was approximately 350 nM. We detected only the substrate and the product (S)-scoulerine with WT BBE and the products (S)-scoulerine and (S)-coreximine with the E417Q mutant.

Continuous-flow mass spectrometry (MS) experiments were also used in an effort to detect the intermediate directly; these also failed to identify any intermediate (Figure 1S). The only major peaks resulting from the reaction of wild-type or E417Q BBE with reticuline corresponded to the masses of the substrate reticuline and the products scoulerine and coreximine. The iminium intermediate would have the same m/z value as the product scoulerine (m/z 328.1549), which complicated its detection by MS. However, MS/MS analyses of the m/z 328.1537 product peak from the WT and E417Q mutant reactions after various time periods yielded the same m/z product ions, indicating that the parent peak likely belonged to the products scoulerine and coreximine, and not the iminium intermediate. Norreticuline was not observed in MS spectra of the reactions above the levels observed in the MS spectra of the substrate reticuline. These results suggest that any iminium intermediate is short-lived and quickly cyclizes to the final product.

Conclusions

The combination of substrate and deuterium solvent isotope effects described here for wild-type and mutant BBE demonstrates that deprotonation is not occurring before or during C-H bond cleavage, suggesting that the concerted mechanism in Scheme 2 is unlikely. It is possible that deprotonation of the substrate phenol is unnecessary for catalysis. Even a protonated aromatic hydroxyl group functions as an electron-donating group, making the adjacent ortho carbon atoms more nucleophilic. A hydroxyl group must be present at C3′ for cyclization of the substrate to occur by BBE in that its substitution with a methoxy group prevents cyclization (8). This could be due to the methoxy group being less electron-donating than a hydroxyl substituent or to disruption of the interaction between the glutamate and the substrate, causing altered binding. Still, a concerted attack of a methyl amine by the C2′ atom and hydride transfer is less likely than attack on a methylene iminium ion intermediate by the C2′ atom, indicating that a stepwise mechanism is the likely mechanism for catalysis.

ABBREVIATIONS

BBE

Berberine Bridge Enzyme