Aromatic substitution of the FAD-shielding tryptophan reveals its differential role in regulating electron flux in methionine synthase reductase and cytochrome P450 reductase



K. R. Wolthers, Department of Chemistry, University of British Columbia Okanagan, 3333 University Way, Kelowna, BC V1V 1V7, Canada

Fax: +1 250 807 8009

Tel: +1 250 807 8663



Methionine synthase reductase (MSR) and cytochrome P450 reductase (CPR) transfer reducing equivalents from NADPH via an FAD and FMN cofactor to a redox partner protein. In both enzymes, hydride transfer from NADPH to FAD requires displacement of a conserved tryptophan that lies coplanar to the FAD isoalloxazine ring. Swapping the tryptophan for a smaller aromatic side chain revealed a distinct role for the residue in regulating MSR and CPR catalysis. MSR W697F and W697Y showed enhanced catalysis, noted by increases in kcat and kcat/Km(NADPH) for steady-state cytochrome c3+ reduction and a 10-fold increase in the rate constant (kobs1) associated with hydride transfer. Elevated primary kinetic isotope effects on kobs1 for W697F and W697Y suggest that preceding isotopically insensitive steps like displacement of W697 are less rate determining. MSR W697Y, but not MSR W697F, showed detectable formation of the disemiquinone intermediate, indicating that the polarity of the aromatic side chain influences the rate of interflavin electron transfer. By contrast, the CPR variants (W676F and W676Y) displayed modest decreases in cytochrome c3+ reduction, a 30- and 3.5-fold decrease in the rate of FAD reduction, accumulation of a FADH2–NADP+ charge-transfer complex and dramatically suppressed rates of interflavin electron transfer. We conclude for MSR that hydride transfer is ‘gated’ by the free energy required to disrupt dispersion forces between the FAD isoalloxazine ring and W697. By contrast, the bulky indole ring of W676 accelerates catalysis in CPR by lowering the energy barrier for displacement of the oxidized nicotinamide ring coplanar with the FAD.


cytochrome P450 reductase


electron transfer


ferredoxin NADP+-reductase


kinetic isotope effect


methionine synthase reductase


Cytochrome P450 reductase (CPR) and methionine synthase reductase (MSR) belong to a family of diflavin oxidoreductases that contain one equivalent of FAD and FMN per polypeptide chain [1-3]. These enzymes catalyze the transfer of reducing equivalents from the obligate two-electron donor, NADPH, to a single electron accepting metal cofactor of a transiently bound protein [4]. Catalysis begins with the donation of a hydride ion (H) from the C4 position of the nicotinamide ring of NADPH to the N5 of the FAD isoalloxazine ring. The FMN cofactor then receives a single electron from the reduced FAD hydroquinone (FADH2), which it donates to a redox partner protein [5]. Physiological electron acceptors of CPR include microsomal cytochrome P450 monooxygenases, heme oxygenases, cytochrome b5 and the fatty acid elongation system [6-9]. An N-terminal α-helical transmembrane anchor tethers CPR to the cytosolic side of the endoplasmic reticulum, where it communicates with its multiple redox partners. MSR is a cytosolic enzyme that maintains the active form of cobalamin-dependent methionine synthase [3, 4]. Methionine synthase converts methyltetrahydrofolate to tetrahydrofolate and homocysteine to methionine [10]. MSR is involved in the reductive methylation of methionine synthase-bound cob(II)alamin, which converts methionine synthase to the active methionine synthase–methylcob(III)alamin state.

Diflavin oxidoreductases are multidomain enzymes that evolutionarily arose from the genetic fusion of two prokaryotic flavoproteins [11]. The FMN domain, located at the N-terminal half of the protein, is homologous to FMN-containing flavodoxins and the C-terminal NADPH/FAD-binding domain is structurally related to ferredoxin NADP+-reductase (FNR) [12]. A centrally located connecting domain bridges the two flavin-binding motifs. In solution and in the crystalline state, CPR adopts a compact state in which the FMN and FNR domains form a binding interface that is stabilized through a network of polar and ionic interactions (Fig. 1) [12, 13]. The compact state enables rapid interflavin electron transfer (ET) as it positions the dimethylbenzyl moieties of FAD and FMN within ~ 4 Å of each other [12]. For ET to external electron acceptors, the two flavin-binding motifs are envisioned to disengage from one another, such that the FMN is more solvent exposed and amenable for interaction with partner proteins [12, 14]. NMR, small-angle X-ray scattering and mass spectrometry have provided strong evidence that CPR adopts an open conformation, and mutagenesis has established that domain motion is important for ET to redox partner proteins [15-17].

Figure 1.

Domain organization of CPR and the orientation of the FAD-stacking tryptophan. (A) The compact state of human CPR (PDB code 3QE2) showing the FMN-binding domain in slate, the connecting domain in magenta and the FNR-binding domain in gray. The cofactors are shown as green stick structures. (B) The extended conformation of CPR (PDB code: 3ES9) using the same coloring scheme as (A). (C) The orientation of residues W676, D632 and R298 in the human CPR–NADP+ complex (PDB code: 3QE2). (D) Location of the corresponding W697, D652 and K291 with NADP+ bound to the FNR module of MSR (PDB code: 2QTZ).

Diflavin oxidoreductases and the structurally related ferredoxin NADP+-reductases possess a conserved aromatic residue that lies parallel to re-face of the FAD isoalloxazine ring [17]. Sequence analysis of diflavin oxidoreductases from various organisms reveals that of the three aromatic side chains, tryptophan most commonly shields the FAD (Fig. 1C,D). By contrast, a tyrosine shields FAD in the structurally related ferredoxin NADP+-reductases [18]. Mutagenesis and crystallographic studies of CPR and ferredoxin NADP+-reductases show that the aromatic residue – located at the penultimate position of the polypeptide or at the C-terminus – sterically hinders catalytic placement of the nicotinamide moiety [18, 19]. Donation of a hydride ion from the coenzyme requires disruption of van der Waals' interactions between the FAD and the aromatic residue and rotation of the side chain away from the cofactor to accommodate the incoming nicotinamide ring. Swapping the tryptophan for a histidine in MSR significantly slowed the rate of interflavin ET, while accelerating the rate of hydride transfer [20]. By contrast, a similar substitution in CPR attenuated flavin reduction [21]. These studies indicate that structural and thermodynamic differences between MSR and CPR influence the catalytic role of the tryptophan residue. To evaluate the clear preference of the tryptophan over smaller aromatic side chains in diflavin oxidoreductases and the differential role of this residue in regulating catalysis in MSR and CPR, we generated conservative substitutions of the residue, substituting the side chain for a phenylalanine and tyrosine. Our results reveal a contrasting role for the FAD-stacking residue in regulating electron flux in MSR and CPR.


Steady-state kinetic analysis

Substitution of the FAD-stacking tryptophan for a smaller aromatic side chain has differential affects on the steady-state kinetic parameters of cytochrome c3+ reduction for CPR and MSR (Table 1). In MSR, W697F and W697Y showed a 2.4- and 3.4-fold increase in kcat and a 4.8- and 6.7-fold increase in catalytic efficiency for NADPH. By contrast, swapping the corresponding residue for Tyr in CPR resulted in a 1.5-fold decrease in kcat and 1.2-fold decrease in catalytic efficiency, whereas swapping the residue for a Phe led to a 6.4-fold and 2.4-fold reduction in kcat and kcat/Km for NADPH, respectively.

Table 1. Kinetic and inhibition constants for wild-type and variant MSR- and CPR-catalyzed reduction of cytochrome c3+
Enzyme specieskcat (s−1)Km(NADPH) (× 10−6 m)Ki (× 10−6 m)kcat/Km (× 106 m−1·s−1)
  1. a

    Values are from reference [22].

Wild-typeMSRa7.2 ± 0.12.4 ± 0.237 ± 31.4 ± 0.13.1 ± 0.3
W697FMSR17.0 ± 0.31.2 ± 0.13.0 ± 0.30.1 ± 0.014.8 ± 1.5
W697YMSR24.3 ± 0.41.2 ± 0.14.2 ± 0.50.7 ± 0.120.8 ± 2.3
Wild-typeCPR20.0 ± 0.20.7 ± 0.11.0 ± 0.10.6 ± 0.128.2 ± 1.8
W676FCPR3.8 ± 0.10.3 ± 0.10.4 ± 0.11.9 ± 0.311.9 ± 1.2
W676YCPR13.1 ± 0.20.6 ± 0.10.7 ± 0.11.2 ± 0.123.8 ± 2.5

Previous isothermal titration calorimetry experiments (performed in phosphate-free buffer) gave dissociation constants of 50 nm and 37 μm for the NADP+–CPR and NADP+–MSR complexes, respectively, revealing that the oxidized coenzyme has a ~ 740-fold higher binding affinity for wild-type CPR [22, 23]. Product inhibition results shown in Table 1 reveal only a ~ 40-fold difference in the Ki for NADP+ between the two enzymes. The difference is likely due to the fact that steady-state experiments for CPR were performed in 50 mm potassium phosphate buffer, whereas 50 mm Tris/HCl was used for MSR. The orthophosphate anion is known to competitively inhibit the binding of the pyridine nucleotide in house-fly CPR, leading to an inflation of the apparent Ki for 2′,5′-ADP and NADP+ and Km for NADPH [24]. Indeed, isothermal titration calorimetry experiments demonstrated that the binding affinity of the oxidized coenzyme is 15-fold higher in a phosphate-free buffer [23]. We were unable to establish an accurate estimate of the Km in phosphate-free buffer for CPR using cytochrome c3+ assays, because the true Km is too low for the detection limits of a 1 cm path length cuvette.

In MSR, the W697F (Ki = 3.0 μm) and W697Y (Ki = 4.2 μm) variants showed improved NADP+ binding affinity compared with wild-type (Ki = 37 μm). The effects of the corresponding substitutions were not as pronounced, because the W676Y and W676F variants elicited a Ki of 0.7 and 0.4 μm for NADP+ with the wild-type enzyme having a Ki of 0.95 μm. Comparison of these three Ki values assumes that the phosphate dianion, which likely competes with the coenzyme for the 2ʹ-AMP subsite, inhibits the enzyme to the same extent in native CPR as it does for the W676 variants. Lower apparent Km values of NADPH observed for the CPR and MSR variants suggest that the reduced coenzyme also binds with higher affinity. The MSR variants also had slightly increased binding affinity for 2′,5′-ADP with MSR W697F and W697Y eliciting a Ki of 0.1 and 0.7 μm, respectively, for the coenzyme analog. CPR W676F and W676Y, however, showed slightly weaker binding affinity for 2′,5′-ADP compared to the native enzyme.

Stopped-flow analysis of flavin reduction in CPR

Figure 2 illustrates a mechanistic scheme for the NADPH-catalyzed reductive half-reaction of CPR. Stopped-flow spectroscopic analysis of mammalian forms of CPR reveals that NADPH-catalyzed reduction of CPR occurs in three kinetic phases over 400 s [25]. These kinetic phases are illustrated in Fig. 3A,D, which shows the spectral changes of wild-type CPR upon NADPH reduction and the deconvoluted spectra of the intermediates that form along the reaction pathway (generated from global analysis of the time-resolved photodiode array data). The ‘fast’ phase, with a rate constant of 20 s−1 (human CPR) or 30–60 s−1 (rat CPR), involves concomitant bleaching of the flavin absorption maxima at 454 nm and an absorbance increase at 600 nm [17, 25, 26]. Spectral changes occurring in this first kinetic phase indicate that H transfer from NADPH to FAD is kinetically coupled to interflavin ET (Steps 4–6 of Fig. 2) [25, 27]. The second kinetic phase reflects a population of the two-electron reduced form of CPR oxidizing a second equivalent of NADPH, noted by a further bleaching of absorbance at 454 and 600 nm (Step 7 of Fig. 2). The observed rate constant for this reaction drops to ~ 5 s−1, owing to the reduced thermodynamic pull of the higher potential FMN [25, 28]. A final slow kinetic phase involves a further absorbance decrease at 454 nm and an increase at 600 nm and represents the coenzyme/enzyme solution reaching a thermodynamic equilibrium [25].

Figure 2.

Proposed mechanism for the NADPH-catalyzed reduction of diflavin oxidoreductases. Step I: NADPH binding to the diflavin enzyme with the flavin cofactors in the fully oxidized state (FAD and FMN). Step II: A 180° flip of the conserved tryptophan (see 'Discussion'). Step III: Displacement of the tryptophan by the nicotinamide ring of NADPH. Step IV: Transfer of a hydride ion from C4 of the nicotinamide ring to the N5 of the FAD isoalloxazine ring, producing the hydroquinone form of the cofactor (FADH2) and NADP+. Step V: Internal ET from FADH2 to FMN producing the semiquinone forms of both cofactors (FADḢ and FMNḢ); denoted by an increase in absorbance at 600 nm. Step VI: second single ET to the FMNḢ and step VII encompasses multiple steps that involve full four-electron reduction of the enzyme by a second molecule of NADPH.

Figure 3.

Resolved spectral intermediates following rapid mixing of CPR with NADPH. Wild-type CPR (22 μm), W676F (16 μm) and W676Y (36 μm) were mixed in the stopped-flow with 10-fold excess of NADPH under anaerobic reactions in 50 mm Tris/HCl, pH 7.5 at 25 °C. The time-dependent spectral changes of the flavin absorbance were monitored by photodiode array spectroscopy over 400 s and are shown for wild-type (A), W676F (B) and W676Y (C). For clarity, only select spectra are shown. The multiwavelength data, comprising 1000 individual spectra from 380 to 700 nm, were subject to singular value decomposition analysis and global fitting. For wild-type, the data were fit to a three-step model (a → b → c → d), generating kobs1 of 19.0 ± 0.01 s−1, kobs2 of 2.72 ± 0.01 s−1 and kobs3 of 0.04 ± 0.01 s−1 and deconvoluted spectral intermediates shown in (D). The spectral data for W676F and W676Y were fit to a two-step model of (a → b → c) giving a kobs1 of 0.64 ± 0.01 s−1 and 0.005 ± 0.001 s−1 for W676F and 6.12 ± 0.03 s−1 and 0.033 ± 0.03 s−1 for W676Y and the deconvoluted spectral intermediates shown in (E) (W676F) and (F) (W676Y).

The stopped-flow photodiode array spectra generated upon mixing CPR W676F and W676Y with an excess of NADPH reveals partial bleaching of absorbance at 454 nm accompanied by the appearance of a flat absorbance band > 532 nm (Fig. 3B,C). The conversion of fully oxidized enzyme (a) to partially reduced (b), occurring at 6.1 s−1 for W676Y and 0.64 s−1 for W676F (Fig. 3E,F), is accompanied by a unique shift in the flavin spectral profile. For wild-type CPR (Fig. 3D), the absorption maximum is located at 454 nm, with a shoulder at 474 nm. The oxidized W676Y spectra (Fig. 3F) is similar to the native enzyme with a peak at 457 nm and a shoulder at 474 nm; however, the flavin spectra of the W676F variant is shifted, such that the absorbance peak is at 474 nm, with a shoulder at 456 nm (Fig. 3E). For the CPR variants, NADPH reduction causes the flavin absorbance maxima to decrease and shift (more so for W676F) to reveal a prominent shoulder at 474 nm. A comparison of the deconvoluted spectral intermediates of ‘a’ and ‘b’ shown in Fig. 3E,F clearly shows these spectral shifts. The shoulder at 474 nm in ‘b’ (a diagnostic feature of the oxidized FMN cofactor), suggests that interflavin ET does not significantly reduce the FMN cofactor to the semiquinone state in the first kinetic phase for W676F and W676Y. Instead, the first kinetic phase represents a single NADPH hydride transfer event leading to the formation of the NADP+–FADH2–FMN complex. Thus, the broad flat absorbance band ≥ 532 nm observed in ‘b’ for W676F and W676Y likely represents the formation of a charge-transfer complex arising from stacking of the oxidized nicotinamide and the FAD isoalloxazine rings. Moreover, the lack of a clear absorbance maxima centered at 600 nm, which would otherwise signify the occurrence of the semiquinone, suggests that this band is primarily attributed a charge-transfer species.

The larger change in the flavin absorbance maxima for W676Y versus W676F, suggests that the Tyr is able to further shift the enzyme to the two-electron reduced state (i.e. E–FADH2–FMN). The conversion of ‘b’ to ‘c’ occurs on a much slower time scale (rate constant of 0.03 s−1 for W676Y and 0.005 s−1 for W676F) and like wild-type CPR, this kinetic phase is associated with accumulation of the flavin semiquinone (noted by the increase in the absorbance band centered at 600 nm). This final kinetic phase indicates that W676F and W676Y function in interflavin ET as the enzyme reaches thermodynamic equilibrium. From these data, we conclude that substitution of Trp676 to a smaller aromatic residue leads to a decrease in the observed rate of FAD reduction, 3.5-fold for W676Y and 30-fold for W676F. Absence of a detectable disemiquinone intermediate during the first kinetic phase indicates suppressed interflavin ET, and the appearance of a charge-transfer band suggests that the oxidized nicotinamide ring remains coplanar to the FAD. Similar results were observed for the W676H variant, where it was concluded that the indole ring serves to trigger release of oxidized cofactor following hydride transfer [21].

Primary kinetic isotope effects with (R)-[4-2H]-NADPH on CPR reduction

Single-wavelength stopped-flow traces at 454 nm for CPR W676Y and W676F with (R)-[4-2H]-NADPH and NADPH are shown in Fig. 4A,B. Absorbance traces at 454 nm were monophasic over 10 s and a fit of the data to a single exponential equation generated observed rate constants of 5.6 ± 0.1 s−1 (NADPH) and 6.0 ± 0.1 s−1 ((R)-[4-2H]-NADPH) for W676Y and 0.66 ± 0.01 s−1 (NADPH) and 0.73 ± 0.01 s−1 ((R)-[4-2H]-NADPH) for W676F. For both W676F and W676Y, reduction by (R)-[4-2H]-NADPH leads to small increase in the rate of flavin reduction in CPR W676Y and W676F, generating inverse kinetic isotope effects (KIE) of 0.90 (W676F) and 0.94 (W676Y). The inverse KIE suggests that reverse hydride/deuteride transfer to NADP+ is a component of the observed rate constant. A single exponential fit of the absorbance traces at 600 nm gave rate constants (0.71 ± 0.02 s−1 for W676F and 7.5 ± 0.1 s−1 for W676Y) that are comparable with that obtained at 454 nm, indicating that formation of the NADP+–FADH2 charge-transfer complex is concomitant with FAD reduction (Fig. 4C,D).

Figure 4.

Stopped-flow single-wavelength absorbance traces of CPR W676Y and W676F. (A,B) Average of five absorbance traces at 454 nm generated upon mixing of 200 μm NADPH (black line) and 200 μm [4(R)-2H] NADPH (gray line) with 20 μm CPR W676F (A) and 20 μm CPR W676F (B). Conditions: 50 mm Tris/HCl, pH 7.5, 25 °C. At 454 nm, five separate absorbance traces were fit to a single exponential equation generating observed rate constants for CPR W676Y of 5.60 ± 0.04 s−1 (NADPH) and 5.96 ± 0.11 ([4(R)-2H]NADPH) and a primary KIE of 0.94 ± 0.02. The observed rate constants for W676F are 0.66 ± 0.01 s−1 (NADPH) and 0.73 ± 0.05 s−1 [4(R)-2H]NADPH, giving a primary KIE of 0.90 ± 0.03. (C,D) Absorbance traces at 600 nm for W676F (C) and W676Y (D) generated by mixing 200 μm NADPH with 10 μm of each of the variants. Monoexponential fits of these traces gave rate constants of 0.70 ± 0.01 (W676F) and 7.5 ± 0.1 (W676Y).

Thermodynamic analysis of CPR W676Y and W676F

Anaerobic redox potentiometry was performed to investigate if the altered kinetic behavior of W676Y and W676F is a result of the substitutions altering the thermodynamic driving force for electron flux through the flavoprotein. To determine the midpoint potentials of the FAD and FMN cofactors, the absorbance spectra of W676Y and W676F were collected as sodium dithionite was titrated into the protein solution (Fig. 5). At each redox state of the flavoprotein, the potential value and absorbance spectrum was recorded. The midpoint potentials were extracted by plotting the summed absorbance values from 590 to 605 nm (over the semiquinone absorption maximum) against the potential value, normalized to the standard hydrogen electrode. For W676F, a fit of the data to Eqn (1), generated midpoint potential values for the FMNox/sq (−99 ± 1 mV) and FMNsq/hq (−215 ± 5 mV) couples that were similar to that of the native enzyme, whereas the FADox/sq (−252 ± 10 mV) and FADsq/hq couples (−275 ± 8) were ~ 30 and ~ 110 mV more electropositive [27]. Redox titrations of W676Y also generated midpoint potentials of the FADox/sq (−263 ± 56 mV) and FADsq/hq (−260 ± 25 mV) couples that were more electropositive than that of wild-type CPR. Unexpectedly, the FMNox/sq (−166 ± 3 mV) and FMNsq/hq (−198 ± 4 mV) were also slightly perturbed by the tryptophan to tyrosine substitution. It is not clear why the FMN cofactor potential is altered by a substitution neighboring the FAD, but it may reflect a slightly altered electronic environment surrounding the FMN cofactor in W676Y. Although the midpoint potentials are more compressed in W676F and W676Y, the thermodynamically favored flow of electrons remains NADPH to FAD to FMN, as it is in the native enzyme. Thus, slower rates of NADPH to FAD hydride transfer and suppressed interflavin ET observed in W676F and W676Y cannot be attributed to changes in the flavin midpoint potentials.

Figure 5.

Potentiometric analysis of CPR W676F and W676Y. Absorbance spectra of CPR W676F (A) and W676Y (C) were recorded during anaerobic titration with the reductant dithionite. A plot of summed absorbance values between 590 and 605 nm (across the semiquinone maximum) was plotted against the reduction potential, normalized to the standard hydrogen electrode for W676F (B) and W676Y (D). The data (B) and (D) were fit to the four-electron Nernst equation (Eqn (1)), as described in 'Experimental procedures'. A fit of Eqn (1) to the data in (B) gave midpoint potential values of FMNox/sq (−99 ± 1 mV); FMNsq/hq (−215 ± 5 mV); FADox/sq (−252 ± 10 mV); FADsq/hq (−275 ± 8 mV). A fit of Eqn (1) to the data in (D) gave midpoint potential values of FMNox/sq (−166 ± 3 mV); FMNsq/hq (−198 ± 14 mV); FADox/sq (−263 ± 56 mV); FADsq/hq (−260 ± 25 mV).

Multiwavelength stopped-flow analysis of MSR flavin reduction

Stopped-flow photodiode array spectra with the W697F and W697Y show bleaching of the flavin absorbance maxima (454 nm), without appreciable formation of an absorbance band centered at 600 nm (Fig. 6A,B). Preliminary inspection of the single-wavelength absorbance trace at 454 nm for both variants reveals three resolvable kinetic phases. The value of the initial rate constant could not be accurately determined as the photodiode array detector scans every 1.5 ms and the first kinetic phase occurred in < 6 ms. Fortunately, single-wavelength absorbance traces at 454 nm, shown in Fig. 7 (see below), enabled us to extract an observed rate constant (kobs1) of 213 and 215 s−1 for W697Y and W697F, respectively. Accordingly, the time-resolved spectral data were globally fit to a three-step model, with the first rate constant fixed at the value determined from single-wavelength analysis.

Figure 6.

Resolved spectral intermediates following rapid mixing of MSR W697 variants with NADPH. MSR W697F (12 μm) and W697Y (10 μm) were mixed in the stopped-flow with 10-fold excess of NADPH under anaerobic conditions in 50 mm Tris/HCl, pH 7.5 at 25 °C, and the optical changes in the flavin spectra over 200 s are shown in (A) (W697F) and (B) (W697Y). The spectral data in (A) and (B) were globally fit to a three-step model of a → b → c → d, and the deconvoluted spectral intermediates resulting from the analysis are displayed; (C) W697F, (D) W697Y. For W697F and W697Y, the first rate constant, reporting on conversion of a to b, was fixed at 215 s−1 and 213 s−1, respectively. The following are the observed rate constants generated from the global analysis: for W697F b → c, 1.15 ± 0.01 s−1; c → d, 0.016 ± 0.001 s−1 and for W697Y; b → c, 1.32 ± 0.01 s−1; c → d, 0.015 ± 0.001 s−1.

Figure 7.

Stopped-flow single-wavelength absorbance traces of MSR W676F and W676Y. An average of five absorbance traces at 454 nm upon mixing of 250 μm NADPH (black line) and 250 μm [4(R)-2H] NADPH (gray line) with 12.5 μm MSR W697F (A) and 12.5 μm MSR W697Y (B). To determine observed rate constants, five individual traces at 454 nm were fitted to double-exponential equation from 0.001 to 5 s. For MSR W697F, the average values of kobs1 and kobs2 are 215 ± 13 s−1 and 0.91 ± 0.12 s−1 (NADPH) and 106 ± 10 and 1.4 ± 0.1 s−1 ([4(R)-2H]NADPH), respectively, giving inline image of 2.0 ± 0.1 and inline image of 0.67 ± 0.10. For MSR W697Y, the average values of kobs1 and kobs2 are 213 ± 12 and 1.1 ± 0.1 s−1 (NADPH) and 104 ± 1 and 1.3 ± 0.2 s−1 ([4(R)-2H]NADPH), respectively, giving inline image of 2.0 ± 0.1 and inline image of 0.87 ± 0.14. (C,D) Absorbance traces at 600 nm following mixing of 12.5 μm MSR W697F (C) and 12.5 μm W697F (D) with 250 μm NADPH.

Global analysis of the photodiode spectra generated similar rate constants for both (kobs2 = 1.2–1.3 s−1 and kobs3 = 0.015 s−1) and spectral profiles of the intermediates formed during the course of the reaction (Fig. 5C,D). The value of kobs2 is fivefold higher than that observed for native MSR, but similar to kobs2 for S698A and S698Δ [20]. The final observed rate constant (kobs3), reporting on the full-four-electron reduction of the enzyme by a second equivalent of NADPH is similar for W697Y, W697F, and wild-type MSR. The deconvoluted spectral intermediates show a progressive loss in the flavin absorbance maxima at 454 nm, without significant build up of an absorbance band at 600 nm (observed in spectral species C for the wild-type enzyme) [29]. These data indicate that reducing the size of the aromatic side chain significantly increases the rate of the first hydride transfer step, but suppresses the rate of interflavin ET.

Primary kinetic isotope effects with (R)-[4-2H]-NADPH on MSR

Single-wavelength stopped-flow absorbance traces at 454 nm following the rapid mixing of the reduced coenzyme with MSR W697F and W697Y were biphasic over 5 s (Fig. 7A,B), with a third slower kinetic phase appearing over 200 s (not shown). The stopped-flow traces were fit to a double-exponential equation from 0.001 to 5 s, generating a kobs1 of 215 and 213 s−1, and kobs2 of 0.91 and 1.1 s−1 for W697F and W697Y, respectively. The 10-fold increase in kobs1 indicates that the indole ring of W697 significantly attenuates step(s) in the mechanism from the binding of substrate up to the first hydride transfer event. Stopped-flow experiments repeated with (R)-[4-2H]-NADPH resulted in a reduction of kobs1 (108 s−1 for W697F and 106 s−1 for W697Y) generating a primary kinetic isotope effect of 2.0 for W697F and W697Y. These primary KIE values are slightly higher than that observed for native MSR (KIE = 1.7) indicating that hydride transfer is more rate determining in the variants [29]. The reversible nature of the reductive half-reaction is noted in the second kinetic phase, where inverse isotope effects of 0.68 for W679F and 0.80 for W697Y were observed. The percent amplitude changes associated with kobs1, kobs2, and kobs3 are 19%, 29% and 52% for W697Y and 14%, 21% and 65% for W697F, similar to that of the wild-type enzyme [29].

Previous studies of MSR suggested that the FAD-stacking tryptophan of MSR accelerated FADH2 to FMN ET, as the disemiquinone intermediate was not detected upon substitution of the residue for a histidine or serine [20]. To examine if a smaller aromatic side chain could partially substitute for the tryptophan in mediating interflavin ET, we monitored single-wavelength traces at 600 nm during the course of the reductive half-reaction. As shown in Fig. 7D, there was a small accumulation of the disemiquinone intermediate in W697Y noted by the transient absorbance change at 600 nm. This intermediate was essentially absent in W697F (Fig. 7C). For W697Y, the small amplitude change associated with disemiquinone formation prevented this species from being resolved as a discrete spectral intermediate through global analysis of the photodiode array spectra. Nevertheless, the single-wavelength data indicate that the tyrosyl side chain does partially substitute for the indole moiety in accelerating interflavin ET.


A number of studies with diflavin reductases have shown that the aromatic side chain has a multifaceted role in catalysis: controlling coenzyme binding affinity and selectivity and the rates of flavin-mediated ET [20, 30-32]. We were interested in determining whether the tryptophan has additional functionality not present in the smaller aromatic side chains, given that there is a high preference for the indole moiety in diflavin oxidoreductases (e.g. novel oxidoreductase 1, P450 BM3) and the fact that the evolutionary progenitor, FNR, contains an FAD-stacking tyrosine. In other words, we wanted to determine whether the tryptophan was a beneficial amino acid substitution that evolved with the fusion of FNR and flavodoxin. We also wanted to further explore the disparate role of the FAD-stacking residue in controlling catalysis in MSR and CPR.

Structural rationale for differences in coenzyme binding affinity between CPR and MSR

It is apparent from steady-state inhibition studies and previous isothermal titration calorimetry experiments that NADP+ binds more tightly to CPR than MSR [22, 23]. A comparison of the CPR–NADP+ and MSR(FNR)–NADP+ crystal structures provide a plausible structural rationale for these differences as they show different orientations of the FAD-stacking tryptophan and a neighboring loop containing a conserved Asp (Aps632 in human CPR and Asp652 in MSR; Fig. 1) [12, 22]. In the MSR(FNR)–NADP+ structure, the entire indole moiety of Trp697 is stacked against the FAD isoalloxazine ring (maximizing π–π overlap). In addition, the side-chain carboxylate of Asp652 points towards the 5′-pyrophosphate of the cofactor, creating electrostatic repulsion that potentially weakens coenzyme binding [22]. The MSR(FNR) substrate-free structure also shows the same orientation of the Trp697 and Asp652, indicating that coenzyme binding does not induce conformational changes in these residues [22]. By contrast, the indole ring of Trp676 in the CPR–NADP+ structure is flipped such that the phenyl portion of the side chain is parallel with the FAD. Asp632 of CPR is also rotated away from the NADP+ 5′-pyrophosphate, reducing electrostatic repulsion with the coenzyme. CPR adopts this active site topology upon binding of the oxidized coenzyme, as the active site of ligand-free CPR – albeit one with a genetically engineered disulfide linkage – resembles that of MSR, with the entire indole moiety of Trp697 stacked against the FAD and Asp632 pointing downwards [17]. Active site differences between the two enzymes that manifest into coenzyme-induced structural changes in one enzyme and not the other are not known, but they may account for the different binding affinities for NADP+.

Previous studies with CPR and MSR have shown that swapping Trp for a smaller residue (e.g. Ser, Ala) increases the binding affinity for NADP+ and reduces the coenzyme preference for NADPH over NADH [20, 30]. For example, a MSR W697S variant resulted in 60-fold higher binding affinity for NADP+, presumably because the substitutions open up a binding cavity within the active site that enables the nicotinamide ring to stack against the FAD [20]. As expected, the effects of the Tyr and Phe variants of MSR were not as dramatic, because they led to only relatively modest decreases in the inhibition constants. The corresponding substitutions in CPR had even less of an affect on the coenzyme binding affinity. These data indicate that the aromatic side chains in the CPR and MSR variants are coplanar to the FAD isoalloxazine ring, and act, albeit to a lesser degree than Trp, to preclude catalytic placement of the nicotinamide ring.

Rotation of MSR W697 gates hydride transfer

Steady-state and pre-steady-state experiments revealed that relatively conserved substitutions in CPR and MSR have opposite effects on catalysis. In MSR, substitution of W697 for a smaller aromatic side chain resulted in increased steady-state cytochrome c3+ reduction and catalytic efficiency for NADPH. Improved catalytic performance was likely the result of the 10-fold rate enhancement in the initial ‘fast’ phase of flavin reduction, kobs1. The observed rate constant, kobs1, is not only a function of hydride transfer, but also the steps leading up to this chemical step, including binding of NADPH and displacement of W697 (Steps 1–3 of Fig. 2). Thus, an increase in kobs1 can result from the substitutions affecting one or more of these steps of the mechanism. Given that the extensive π–π stacking between W697 and the FAD likely constitutes an energetic barrier for productive NADPH binding and hydride transfer, it is reasonable to propose that a reduction in this noncovalent contact in W697Y and W697F lowers this energy barrier, resulting in an increase in kobs1. A small amplification in the primary KIE observed with (R)-[4-2H]-NADPH on kobs1 over that of the native enzyme, suggests that isotopically insensitive steps, like displacement of the aromatic residue, become less rate determining. In essence, the substitutions are unmasking the intrinsic isotope effects associated with bond breaking/forming steps of H transfer. The polarity of the aromatic ring (i.e. Phe versus Tyr) does not appear to influence the rate of displacement from the FAD, as both variants elicited, within error, the same elevated KIEs and kobs1.

From the discussion above, it is plausible that in MSR the indole ring undergoes two separate conformational switches prior to hydride transfer, as has been suggested for CPR [17]. In the first step, the indole ring potentially flips 180° from its position depicted in Fig. 1D to that shown in the CPR–NADP+ complex (Fig. 1C). Evidence that MSR W697 undergoes this initial flip with respect to the FAD is provided by single-wavelength stopped-flow absorbance traces at 600 nm, which show transient formation of a charge-transfer complex, with rates of formation and decay equal to 120 and 20 s−1, respectively [29]. Previously, the charge-transfer band was thought to arise from the van der Waals' interaction between the FAD and nicotinamide ring, but given that these broad absorbance bands are nondescript (similar absorbance bands arise from electronic coupling between the aromatic side chains and FAD and between the nicotinamide ring and FAD) it is reasonable to assume that it arises from flipping of the indole ring with respect to the FAD [33]. Similar Trp–FAD charge-transfer bands have been described for FNR and flavodoxins [34, 35]. Moreover, we do not observe this absorbance increase at 600 nm at this approximate time domain in stopped-flow experiments with W697F and W697Y, suggesting that this transient signal originates from the conserved tryptophan. In wild-type MSR, the disappearance of the charge-transfer band at 600 nm is concomitant with the first phase of flavin reduction observed at 454 nm, indicating that rotation of the indole ring away from the FAD isoalloxazine ring is coupled to catalytic placement of the nicotinamide ring and hydride transfer [29]. Thus, we propose for MSR that a reduction in the size of the aromatic ring may lower the energy barrier for two separate realignments of the residue, which in turn may account for the dramatic increase in kobs1 for MSR W697F and W697Y.

In CPR, swapping the FAD-shielding aromatic side chain does not seem to dramatically reduce the energy barrier for catalytic placement of the reduced nicotinamide ring as the observed rate of flavin reduction was slower in the variants. The reason for this difference may be attributed to the ability of the incoming reduced pyridine nucleotide to induce a stable 180° flip of the indole ring, leaving only the phenyl portion of the side chain stacked against the FAD. Thus, the energy barrier for displacement of aromatic side chain by the nicotinamide ring may be the same in W676F, W676Y as in the native enzyme.

Polarity of aromatic FAD-stacking residue influences the rate of interflavin ET in MSR

In MSR, the rate constant that encompasses hydride transfer is two orders of magnitude faster than that of subsequent FAD to FMN ET. The origin of the attenuated interflavin ET rate constant is unknown, but it may be linked to the thermodynamically preferred conformational state of the full-length enzyme or its dynamic behavior. For example, a more extended conformation of MSR (akin to Fig. 1B) increases the distance between the FAD and FMN and slows electronic communication between the cofactors. Although interflavin ET is severely attenuated in MSR, we have shown previously that W697 accelerates this catalytic step [20]. Substitution of the residue to a His or Ser completely abolished the transient absorbance signal at 600 nm that forms during the course of the reductive half-reaction. Evidence that the MSR variants are able to mediate interflavin ET (albeit at suppressed rates) is provided by their ability to be fully reduced by two equivalents of NADPH under double-turnover conditions and to catalyze cytochrome c3+ reduction (a reaction that is dependent on ET from the FMN cofactor). A transient absorption signal at 600 nm was observed in the W697Y variant, but the amplitude of the absorbance trace was smaller compared with wild-type, indicating that it only partially supplants the role of the indole ring in accelerating FAD to FMN ET. The faster rates of interflavin ET exhibited by W697Y, may explain why this variant elicits a slightly higher turnover number for cytochrome c3+ reduction compared to MSR W697F.

The mechanism by which Trp, and to a lesser extent Tyr, accelerate interflavin ET is unknown, but given that they lie at the interface of the FNR and FMN domains, it is reasonable to suggest that they affect the conformational alignment of the flavin cofactors (known to impact rates of interflavin ET) [17]. Alternatively, the polar aromatic side chains may stabilize the FAD semiquinone intermediate, leading to faster splitting of electrons between the redox cofactors [36]. Finally, the side chains may lower the energy level for ET by acting as a direct electron conduit between the cofactors. This latter hypothesis is supported by the fact that Tyr and Trp are the only two naturally occurring amino acid residues known to function in a variety of biological ET processes [37].

W676 displaces the oxidized nicotinamide ring, a step that controls electron flux

By contrast to MSR, hydride and interflavin ET are kinetically coupled in NADPH-catalyzed CPR reduction. The FADox/FADhq redox couple (−332 mV) is slightly more electronegative compared with the NADP+/NADPH couple (−320 mV), whereas the FMNox/FMNsq couple is −66 mV [28, 38]. Thus, the higher potential FMN is the thermodynamic driving force for forward ET through the flavoprotein. The compact nature of the CPR structure seemingly allows for tight kinetic coupling as the FAD and FMN isoalloxazine rings are within van der Waals' contact, enabling instantaneous flavin redox communication following FADH2 formation. Strikingly, substitution of the FAD-stacking tryptophan to a smaller aromatic side chain greatly suppresses interflavin ET, because stopped-flow experiments show bleaching of absorbance at 454 nm (FADH2 formation), without the simultaneous appearance of the disemiquinone. Thus, NADPH-catalyzed reduction of W676F and W676Y leads to a distribution of NADPH–FAD–FMN and NADP+–FADH2–FMN states of the enzyme in the first kinetic phase. The 3.5- and 30-fold decreases in the rate of FAD reduction observed in W676Y and W676F potentially results from a lack of ET to the higher potential FMN, which would otherwise drive electron flux through the protein. As a result, there is a larger commitment to reverse catalysis (i.e. hydride transfer back to NADP+), as evidenced by the inverse KIE observed with (R)-[4-2H]-NADPH for both W676 variants. Moreover, the charge-transfer band observed in the photodiode array spectra suggests that the nicotinamide ring remains coplanar with the FAD, enabling reverse hydride transfer. Therefore, the stopped-flow spectroscopic data suggest that the W676 indole ring lowers the energy barrier for displacement of the oxidized nicotinamide ring from the FAD following H transfer. The lack of coupled hydride and interflavin ET in W676F and W676Y potentially result from the displaced aromatic side chain enforcing a conformational realignment of the FAD and FMN isoalloxazine rings that minimizes their electronic overlap. Thus, for CPR it is plausible that introduction of a smaller aromatic side chain approximating the size of the nicotinamide ring increases the thermodynamic barrier for displacement of the nicotinamide ring from the FAD. Moreover, the energy barrier is likely higher for NADP+ versus NADPH release, given that coenzyme oxidation converts the nicotinamide ring from a ‘boat’ to planar conformation. Thus, in CPR, W676 may promote NADPH oxidation (versus NADP+ reduction performed by ferredoxin NADP+-reductase), by lowering the energy barrier for displacement of the planar oxidized nicotinamide ring.

Experimental procedures


NADPH, NADP+, 2′,5′-ADP, cytochrome c3+, ethanol-d6 and alcohol dehydrogenase from Thermoanaerobium brockii were purchased from Sigma-Aldrich (Oakville, ON, Canada). Liquid chromatography resins were from GE Biosciences (Baie d'Urfe, QC, Canada), Pfu Turbo DNA polymerase and Xl1 Blue cells were from Agilent Technologies (Mississauga, ON, Canada). BL21(DE3)pLysS and Rosetta2(DE3)pLysS competent cells and yeast alcohol dehydrogenase were from EMD Millipore (Billerica, MA, USA). All other chemicals and reagents were purchased from Fisher Scientific (Ottawa, ON, Canada). (R)-[4-2H]-NADPH (A-side NADPD) was synthesized and isolated as previously described [39].

Expression and purification of MSR and CPR variants

The MSR W697Y, MSR W697F, CPR W676Y and CPR W676F variants were generated using the QuikChange site-directed mutagenesis kit (Agilent Technologies), using oligonucleotides obtained from Integrated DNA Technologies (Coralville, IA, USA). DNA sequencing confirmed that single amino acid substitutions had been successfully generated and that no other amino acid substitutions had been introduced into any of the four variants. The plasmids harboring the designed variants of MSR were transformed into the Rosetta2(DE3)pLysS strain of Escherichia coli and recombinant proteins were expressed and purified as previously described [40]. Wild-type and variant forms of CPR were transformed into BL21(DE3)pLysS and expressed as for MSR [40]. A typical purification of CPR involved resuspending 40 g of cells (wet weight) in 300 mL of 50 mm Tris/HCl, pH 7.5 with 1 mm phenylmethylsulfonyl fluoride and benzamidine. The cells were disrupted by sonication (8 s pulses for 30 min with a 55 s interval, power setting 22%) and then clarified by centrifugation (25 000 g for 50 min). Imidazole (20 mm) and NaCl (0.5 m) were added to the supernatant before it was applied to a 5 mL Ni2+-nitrilotriacetic acid column equilibrated with 50 mm Tris/HCl, pH 7.5, 0.5 m NaCl and 20 mm imidazole. The column was washed with 30 mL of 50 mm Tris/HCl, 0.5 m NaCl, pH 7.5 and 20 mm imidazole, and the protein was eluted with 350 mm imidazole. Fractions containing CPR were pooled and dialyzed against 50 mm Tris/HCl, pH 7.5, 1 mm EDTA, 5 mm β-mercaptoethanol for 16 h at 4 °C. The dialysate was loaded onto a 58 mL Q-sepharose HP column (2.6 × 11 cm) equilibrated with 50 mm Tris/HCl, pH 7.5. The column was washed with 120 mL of 50 mm Tris/HCl, pH 7.5, and the protein was eluted with a linear gradient of 0–0.5 m NaCl at a flow rate of 2 mL·min−1. Fractions containing CPR, as judged by the flavin absorbance spectra and SDS/PAGE analysis, were pooled, concentrated by ultrafiltration and flash-frozen in liquid nitrogen and stored at −80 °C. The final protein concentration was calculated from an absorbance reading at 454 nm (extinction coefficients of 25 600 and 21 600 m−1·cm−1 for MSR and CPR, respectively [26].

Steady-state turnover analysis

The initial rates of MSR and CPR reduction of cytochrome c3+ were determined at 25 °C by following absorbance change over 1 min at 550 nm (Δε =21.1 mm−1·cm−1) on a Lambda 25 UV–visible spectrometer (Perkin–Elmer, Massachusetts, MA, USA) as previously described [22]. For MSR, the 1-mL reaction mixture contained 50 mm Tris/HCl, pH 7.5, 8 μm cytochrome c3+, variable NADPH (0.25–50 μm), NADP+ (0, 1, 2.5, 10 μm) or 2′,5′-ADP (0, 1, 2.5, 10 μm) concentrations. For CPR, the concentration of cytochrome c3+ was 8 μm, and concentrations of NADPH, NADP+ and 2′,5′-ADP were similar to those used for MSR, but the buffer used was 50 mm potassium phosphate, pH 7.5. The reactions were initiated with the addition of MSR or CPR. Inhibition data were fit to the equation for competitive inhibition by nonlinear least-squares analysis using origin 8.5 software (OriginLab Co., Northampton, MA, USA).

Stopped-flow kinetic measurements

A SF-61DX2 stopped-flow (TgK Scientific, Bradford-on-Avon, UK) housed in an anaerobically maintained glove box (Belle Technology, Weymouth, UK) was used to perform pre-steady-state measurements of NADPH-catalyzed reduction of the CPR and MSR variants. All stopped-flow experiments were performed at 25 °C in 50 mm Tris/HCl, pH 7.5 as previously described [20]. Anaerobic solutions of buffer (50 mm Tris/HCl pH 7.5), protein, NADPH and (R)-[4-2H]-NADPH were prepared as previously described [20]. For multiwavelength stopped-flow analysis of CPR, a 10-fold excess of NADPH was mixed with an equal volume of the enzyme and spectral changes were recorded with a photodiode array detector over 10–20 and 400 s on a log time base. For the MSR variants, the spectral data were recorded over 1.5 and 300 s. For both proteins, the spectral data sets were combined and evaluated in reactlab kinetics software (Jplus Consulting Pty Ltd., Karawara, Australia) by singular value decomposition. For wild-type CPR and MSR W697F and W697Y, the reduced singular value decomposition data were best fitted to a three-step model with four discrete spectral species. By contrast, the singular value decomposition data for CPR W676Y and W676F were best fitted to a two-step model. NADPH or (R)-[4-2H]-NADPH reduction of wild-type and variant forms of MSR and CPR were also followed at 454 and 600 nm at 25 °C under pseudo-first-order conditions with reduced coenzyme in 10-fold excess of enzyme concentration. Single-wavelength data were fit to a single or double-exponential equation. The concentrations of enzyme and substrate written in the figure legends are before the mixing event (i.e. syringe concentrations); both solutions are diluted twofold in the reaction cell.

Potentiometric titrations

As for the stopped-flow experiments, the electrochemical redox titrations of CPR W676F and W676Y were preformed in a glove box maintained under a nitrogen atmosphere at 25 °C. The titration buffer, 50 mm Hepes/KOH, pH 7.0 was made anaerobic by extensive bubbling with nitrogen, followed by > 16 h equilibration in the glove box. Hepes buffer was used in place of 100 mm potassium phosphate pH 7.0 as the protein precipitated in the latter buffer during the titration. Two milliliters of concentrated CPR W676F and W676Y were introduced into the glove box and gel filtered over a 10 mL size-exclusion column (Bio-Rad Econo-Pac 10 DG column, Mississauga, ON, CA) equilibrated with anaerobic titration buffer. The protein was diluted to 14–28 μm in a total volume of 3.5 mL. The following redox mediators were added to the protein solution: benzyl viologen (1 μm), methyl viologen (0.3 μm), 2-hydroxy-1,4-naphthoquinone (5 μm) and phenazine methosulfate (2 μm). The fiber optic probes of the stopped-flow were connected to a specially designed cuvette holder to record the flavin absorption spectra throughout the titration. The sample was illuminated with the xenon lamp of the stopped-flow and absorption spectra were recorded from 280 to 703 nm using a photodiode array detector. The electrochemical potential was monitored on a Mettler Toledo FiveEasy voltmeter coupled to a platinum–Ag/AgCl electrode. The potential and spectra were recorded after each addition of reductant, dithionite, following the protocol of Dutton [41]. The observed potentials were normalized to the standard hydrogen electrode with the addition of 197 mV to the potential values. All spectra were corrected to the same baseline absorbance at 703 nm (highest wavelength recorded) due to baseline drift during the redox titration. There was no evidence of protein precipitation at the end of the titration.

The midpoint potentials of FAD and FMN were determined by summing the absorbance between 590 and 605 nm (the flavin semiquinone absorption maxima) and plotting the summed value against the potential, normalized against the standard hydrogen electrode. The data were fit to Eqn (1), which is derived from extension of the Beer–Lambert Law and the Nernst equation; the equation is the sum of a two 2-electron redox process [28, 42].

display math(1)

In these equations, E is the observed potential. E1′ and E2′ are the oxidized/semiquinone and semiquinone/hydroquinone midpoint potentials of one flavin. E3′ and E4′ are the corresponding midpoint potentials of the second flavin. A is the total absorbance; a, b and c are the component absorbance values contributed by one flavin in the oxidized, semiquinone and reduced states, respectively, and d, e and f are the corresponding absorbance values associated with the second flavin. During the fitting routine, it was assumed that the absorbance contribution of the oxidized and reduced forms of the FAD and FMN were equal by setting a = d and c = f. Estimates for E1′, E2′, E3′ and E4′, taken from previously published work, were entered and fixed during the fitting routine. After several iterations, reasonable estimates of the individual flavin absorbance values (parameters a–f) were obtained. Combinations of these absorbance values (i.e. a and d, b and e, and c and f) were fixed, and then the potential values were then allowed to vary during the fitting routine. This process was repeated several times, allowing the potential values and absorbance values a, c, d and f to vary during the final iteration. The parameters b and e were fixed to prevent the data from becoming over parameterized.


This work is supported by a grant from the Natural Sciences and Engineering Research Council of Canada.