Biochemical Journal

Research article

Role of the interface between the FMN and FAD domains in the control of redox potential and electronic transfer of NADPH–cytochrome P450 reductase

Louise Aigrain , Denis Pompon , Gilles Truan


CPR (NADPH–cytochrome P450 reductase) is a multidomain protein containing two flavin-containing domains joined by a connecting domain thought to control the necessary movements of the catalytic domains during electronic cycles. We present a detailed biochemical analysis of two chimaeric CPRs composed of the association of human or yeast FMN with the alternative connecting/FAD domains. Despite the assembly of domains having a relatively large evolutionary distance between them, our data support the idea that the integrity of the catalytic cycle is conserved in our chimaeric enzymes, whereas the recognition, interactions and positioning of both catalytic domains are probably modified. The main consequences of the chimaerogenesis are a decrease in the internal electron-transfer rate between both flavins correlated with changes in the geometry of chimaeric CPRs in solution. Results of the present study highlight the role of the linker and connecting domain in the recognition at the interfaces between the catalytic domains and the impact of interdomain interactions on the redox potentials of the flavins, the internal electron-transfer efficiency and the global conformation and dynamic equilibrium of the CPRs.

  • electron transfer
  • flavin–adenine dinucleotide (FAD)
  • flavin mononucleotide (FMN)
  • NADPH–cytochrome P450 reductase (CPR)
  • redox potential


Multidomain proteins are usually defined as continuous polypeptide chains folded into spatially distinct structural units that are still able to perform a certain function when isolated [1]. Bioinformatics tools have facilitated domain identification in protein databases [2,3], leading to the observation that 70% of eukaryotic proteins belong to multidomain protein families. The fusion of independent domains into larger units is advantageous in terms of co-localization of the components, protection of unstable species or optimization of catalytic pathways [47].

CPR (NADPH–cytochrome P450 reductase) is the major redox partner for eukaryotic cytochromes P450 and belongs to the diflavin reductase family that includes the flavin domains of NOS (nitric oxide synthase) [8], methionine synthase reductase [9], NADP+–pyruvate oxidoreductase [10], cytoplasmic NR1 protein [11] or NADPH–sulfite reductase [12]. They catalyse the transfer of electrons from reduced pyridine nucleotides to a wide variety of electron acceptors with the following pathway: NAD(P)H→FAD→FMN→acceptor [13]. CPR originates from an ancestral fusion between a flavodoxin and a ferredoxin reductase [14]. This hypothesis was confirmed further by the existence of isolated flavin domain components in prokaryotic P450 systems [15]. The union of the FMN and FAD domains is realized via a connecting domain mainly constituted of α-helices. This connecting domain is probably meant to optimize the ET (electron transfer) to acceptors while allowing the control of the global conformation of the enzyme [16,17] and reducing the production of reactive oxygen species by protecting unstable redox species [7].

Up to 2009, the X-ray structures of rat and yeast CPRs [18,19] presented a similar spatial arrangement showing the FMN and FAD cofactors in close vicinity (closed conformation), a geometry adequate for interflavin ET. In 2009, two new conformations were characterized from the crystallographic analysis of a rat CPR mutant [20] and an active yeast–human chimaera [21]. This latter protein crystallized in a widely open conformation where the FMN cofactor is completely exposed to the solvent and becomes thereby accessible to external acceptors (open conformation). The existence of alternative conformations in solution induced by redox changes and nucleotide binding was also detected by SAXS (small-angle X-ray scattering) and NMR [22]. Recent ELDOR (electron–electron double resonance) spectroscopy analysis revealed a continuum of conformational states across the energy landscape and described the nucleotide binding as a key element that controls the closing and opening of the CPR [23]. Even though the dynamic nature of CPR is now beyond doubt, the different factors (flavin redox states, presence of redox partners, cofactor binding or random oscillations) dictating such conformational switches, their modes of operation and the nature in terms of amplitude and frequencies of movements remain unclear [13,20,24].

Furthermore, the three-dimensional structures of yeast and rat wild-type CPRs cannot explain the different kinetic properties and flavin redox potentials [24,25]. Such properties may be governed by the various hypothetical geometric states of each CPR in solution. Oscillating movements of the FMN domain between the FAD domain and a haemic acceptor have been postulated in other diflavin reductases such as NOS and they constitute the rate-limiting step in the overall cycle of the enzyme [2628]. The crucial impact of domain recognition and the role of specific residues interactions at the interface between both flavin domains of NOS have also been described [29].

To analyse how individual domains influence the overall activity of multidomain proteins, domain exchange is more appropriate than multiple site-directed mutagenesis because the simultaneous modification of several residues often has a deleterious impact on the expression, folding or activity of mutant proteins. Chimaerogenesis was already used to study the influence of the electron-carrier part on the NOS activity by exchanging the native reductase domains of the NOS with the rat CPR [30]. Domain swapping produces a drastic interface change while leaving the core structure of each domain intact. Our approach consisted of building chimaeric CPRs assembled from evolutionarily distant parental enzymes (human and yeast). We therefore analysed the influence of domain interfaces on the various internal and external ET rates, the redox potentials of the cofactors and the global geometry of the enzymes.


Cloning, expression and purification

Chimaeric genes were constructed as described in [21]. Figure 1 illustrates the various chimaeras produced and their boundaries. For soluble chimaeric (HYs and YHs) and parental (Hs and Ys) CPRs, genes were cloned into pET15b in-frame with the His6 tag. The constructed plasmids were transformed into competent BL21(DE3) Escherichia coli cells to allow the expression of soluble Ys, Hs, YHs and HYs CPRs. For membranous CPRs (Hm, Ym, HYm and YHm), PCR products were digested, cloned into pYeDP60 vector and transformed into the CPR-deficient strain WRΔ Saccharomyces cerevisiae cells [31].

Figure 1 Construction of the chimaeric genes

MA, membranous anchor; FMN, FMN domain; connecting + FAD, connecting and FAD domains. Numbers refer to the amino acid preceding the junction points in the wild-type full-length proteins. Domains from yeast or human CPR are indicated in white and grey respectively. Hs, soluble human CPR; Hm, membranous human CPR; Ys, soluble yeast CPR; Ym, membranous yeast CPR; HYs, soluble human–yeast chimaeric CPR; HYm, membranous human–yeast CPR; YHs, soluble yeast–human CPR; YHm, yeast–human yeast CPR.

Expression and purification of soluble proteins was performed as described in [21]. Purity of the sample was determined by SDS/PAGE (4–12 % gradient gels) and spectrophotometry (>80%, see Supplementary Figure S1 at Membranous protein expression was performed at 29 °C in 2 litres of YPGE (1% yeast extract, 1% Bacto Peptone, 2% glycerol and 2% ethanol) medium complemented with 2.5 mg/l riboflavin for 36 h before induction by 20 g/l galactose overnight. Cultures were spun down at 10 000 g for 10 min, washed with 200 ml of water and suspended in 200 ml Tes (50 mM Tris/HCl, pH 7.4, 10 mM EDTA and 0.6 M sorbitol) buffer. Cell lysis was achieved in Tes buffer containing an antiprotease cocktail with a cell-disruption system (Constant Systems). Cell debris were removed by centrifugation at 26500 g for 30 min at 4 °C. Microsomal proteins contained in the supernatant were precipitated by adding 0.1 M NaCl and 5% (w/v) PEG [poly(ethylene glycol)] 4000. After centrifugation at 12000 g for 10 min, the precipitate was resuspended in TEG buffer [50 mM Tris/HCl (pH 7.4), 1 mM EDTA and 20% (v/v) glycerol] and conserved at −80 °C. Solubilization of membranous proteins was performed in 50 mM Tris/HCl (pH 7.5), 20% (v/v) glycerol, 1 mM EDTA, 1 mM PMSF, 0.5 mM DTT (dithiothreitol), 5 μM FMN and FAD, 1.25% (v/v) Triton N-101 and 0.625% sodium cholate (buffer A) at a protein concentration of 5 mg/ml for 1 h at 4 °C. After centrifugation at 60000 g for 1 h, the supernatant was applied on to a DEAE-Sepharose column equilibrated with buffer A. Fractions that were eluted at 0.5 M NaCl were kept, concentrated and desalted using a Vivacell-70 centrifugal concentrator (Vivasciences). Protein solutions were purified further via another affinity chromatography using a Q-Sepharose resin equilibrated with buffer A, and eluted with a 0.3 M NaCl step gradient. Purity of the sample was determined by SDS/PAGE (4–12% gradient gels) (see Supplementary Figure S1) and the 280/450 nm ratio was measured by optical spectroscopy (typically 5). Membranous CPRs were then concentrated to 30 μM and stored at −20 °C.

Analysis of the domain interfaces of parental CPR

Interactions at the interfaces between the FMN domain and the rest of the CPR domains were determined via the Internet server PIC (Protein Interaction Calculator, Rat CPR structure was used as a model for human CPR, since no crystallographic structure of the human enzyme is available (93% sequence identity between both proteins). Interactions were detected through the distance between lateral chains of residues and were classified according to the type of interaction involved. HIs (hydrophobic interactions) were searched between alanine, valine, leucine, isoleucine, methionine, phenylalanine, tryptophan, proline and tyrosine residues of less than 5.0 Å (1 Å=0.1 nm) distance. HBs (hydrogen bonds) were detected when the donor–acceptor distance was less than 3.5 Å and 4.0 Å for oxygen/nitrogen–hydrogen and sulfur–hydrogen interactions respectively. AIs (aromatic interactions) concerned lateral chains of lysine and arginine and aromatic groups of phenylalanine, tyrosine or tryptophan of less than 6 Å distance. Finally, SBs (salt bridges) were detected between arginine, lysine, histidine, aspartate and glutamate of less than 6 Ådistance. The inventory of those interactions and their energies is presented in Supplementary Table S1 at

Determination of the kinetic parameters of soluble CPRs towards artificial acceptors

Reduction of external acceptors was measured at 25 °C using a Cary 300 spectrophotometer (Varian). Horse heart cyt c (cytochrome c) reduction was monitored at 550 nm (ϵ=21000 M−1·cm−1) and ferricyanide reduction was monitored at 420 nm (ϵ=1020 M−1·cm−1). The CPR concentration ranged from 0.5 to 50 nM. CPR was diluted directly in a spectrophotometer cuvette containing 20 mM sodium phosphate buffer (pH 7.4) supplemented with 0.5 unit of catalase and 0.5 unit of superoxide dismutase. The NADPH concentration was 200 μM when cyt c or ferricyanide concentrations varied from 0.15 to 40 μM and 3 to 400 μM respectively. The cyt c concentration was held at 50 μM when the NADPH concentration varied from 0.5 to 200 μM. Reactions were started by the addition of NADPH and were monitored over 2 min. Experiments were repeated in triplicate. Kinetic parameters were determined by non-linear regression using a hyperbolic function to retrieve K1/2 and Vmax values. Analyses were performed using SigmaPlot® 11 (Systat Software).

Preparation of FMN-depleted CPRs

Soluble CPRs (0.2 μM) were diluted in 20 mM sodium phosphate buffer (pH 8.5) with 2 M KBr. BSA-saturated coal was added and the solution was gently shaken overnight at 4 °C. After centrifugation at 5000 g for 10 min, supernatants were concentrated and desalted using Vivaspin-15 centrifugal concentrators (Vivasciences) to a final concentration of 6 μM. Saturation with either FMN of FAD was carried out with 6 μM FMN or FAD solutions (respectively). Achievement of the saturation was assessed by cyt c activity measurements.

Determination of FMN and FAD contents and CPR concentration

FAD and FMN contents were determined by HPLC analysis. Flavins were released from CPRs by denaturation into 50 μl of 6 M guanidium chloride for 10 min at room temperature (20 °C) before centrifugation at 16100 g for 10 min. FMN and FAD quantifications were performed by reverse-phase HPLC on an XTerra® MS C18 column (Waters) using pure FAD and FMN as standards.

Enzymatic reconstitution of cytochrome P450 activity

The P450 reductase activity tests were performed in reconstituted systems made of sonicated lipidic micelles of didodecylphosphatidylcholine (1 mg/ml) in 10 mM Mops buffer (pH 7.4). Each sample contained 100 nM human P450 3A4 (Invitrogen), 30 mM MgCl2, 200 μM testosterone, 3 mM GSH and 200 nM cytochrome b5. The molar ratio of CPR to P450 varied from 0 to 10. The reaction was started with the addition of 2 mM NADPH and, after 30 min of incubation at 37 °C, the reaction was stopped by the addition of 1 vol. of acetonitrile. Experiments were performed in duplicate. Testosterone and 6β-hydroxytestosterone quantifications were performed by reverse-phase HPLC on an XTerra® MS C18 column using pure testosterone and 6β-hydroxytestosterone as standards.

Stopped-flow experiments and data processing

Measurements of the ET rate constants were performed on a sequential SX.18MV stopped-flow coupled to a diode-array absorption spectrometer (Applied Photophysics). Experiments were carried out at room temperature in an oxygen-free 20 mM sodium phosphate buffer (pH 7.4). The first syringe contained the CPR solution at 20 μM and the second contained the NADPH solution at 20 or 200 μM. Before each experiment, CPR was fully oxidized by ferricyanide and desalted on a Sephadex G-20 column (GE Healthcare). Experiments were repeated in triplicate. Data fitting and analysis were performed using SigmaPlot® 11 with appropriate multi-exponential functions (see Supplementary Table S2 at evolutions of the absorption at 455 nm corresponding to the biphasic flavin reduction were fitted by decreasing biexponentials, evolutions of the absorption at 585 nm during the first 0.25 s corresponding to sq (semiquinone) formation were fitted by ‘rise to a maximum’ mono-exponentials and evolutions of the absorption at 585 nm over 4 s corresponding to the first sq formation and the further CPR reduction were analysed as described by Guttierez et al. [32] and fitted to eqn (1): Embedded Image


The redox titration of parental and chimaeric CPRs was performed under anaerobic conditions in a 100 mM sodium phosphate buffer (pH 7.4) that was made oxygen-free beforehand by repeated vacuum evacuation and flushing with argon purified through an Oxy-Trap column (Alltech). The CPR concentration was 50 μM and four mediators were added at 0.5 μM to facilitate the electronic exchange between the enzyme and electrodes: safranin (E °=−280 mV), phenosafranin (E °=−240 mV), 2-hydroxy-1,4-naphthoquinone (E °=−150 mV) and Benzyl Viologen (E °=−350 mV). A few units of catalase and superoxide dismutase were also added to avoid the formation of reactive oxygen species. CPRs were sequentially reduced by addition of sodium dithionite aliquots and a time lapse was kept to allow the sample to reach equilibrium (assessed by the optical spectrum and solution potential stability). Absorption spectra were recorded using a Cary 300 spectrophotometer and the solution potential was measured through an silver/AgCl/KCl [210 mV/SHE (standard hydrogen electrode)] reference electrode and a platinum electrode. Experiments were performed in duplicate. Data were analysed with the Electrofilter laboratory-made software based on a method described previously [9,24,33,34] using eqn (2) which describes the sum of two two-electron redox processes derived by extension to the Nernst equation and the Beer–Lambert Law: Embedded Image where E °1, E °2, E °3 and E °4 correspond respectively to the standard potentials of FMN and FAD ox (oxidized)–sq and sq–hq (hydroquinone) couples; ac are the values of the FMN absorption coefficients respectively at the ox, sq and hq states, and de are the same values for the FAD cofactor. The complexity of the system (four-electron titration with probable overlap of midpoint potentials) necessitated the use of a two-stage fitting process. In order to first determine the redox potentials of ox–sq couples, data recorded at 430 nm (isosbestic point for sq and hq forms) were fitted with the constraints b=c and e=f since absorption coefficients for the sqs and reduced flavins are equal at their isosbestic point. The same argument was used to determine the redox potentials of sq–red (reduced) couples with the data at 500 nm (isosbestic point for oxidized and sq forms, a=b and d=e). Midpoint potential values obtained from the fit at isosbestic points were then used as starting points to enable accurate fitting of absorption against reduction potential data at the maxima for oxidized flavins (455 nm) and neutral blue sq forms (585 nm). The oxidized flavins were assumed to have equal absorbance coefficients, especially a=d [33]. Reduction potentials for the system were determined through a iterative process in which midpoint potentials associated with one isosbestic point were fixed while the other pair were varied. The process was repeated iteratively until no further change was observed [24]. Values obtained this way are similar to those originally estimated using the fit at the isosbestic points (see Supplementary Table S3 at


Analysis of the interactions at the interfaces between CPR domains

Domain swapping between yeast and human CPRs induces a massive change in the interface between the FMN and the connecting/FAD domains. To analyse precisely the impact of this chimaerogenesis on the intramolecular interactions between domains, we identified the residues involved in interdomain contacts in parental enzymes. Since no crystallographic structure of human CPR is available, we used the rat CPR as a model for the human protein (93% sequence identity). We mapped three different regions of interaction in parental CPR (Figures 2A and 2B): a first between the FMN and the FAD domains, a second between the FMN and the connecting domains, and the last one between the FMN domain and the loop connecting the FMN domains (linker).

Figure 2 Interactions between CPR domains

(A and C) Rat (PDB code 1J9Z), (B and D) yeast (PDB code 2BF4) and (E) YH (PDB code 3FJO) CPR. (A and B) Residues in dark grey, black and grey refer to the interface between FMN/FAD domains, FMN/connecting domains and FMN domain/linker respectively. (C and D) Residues in black represent specific interactions, whereas residues in white represent conserved or equivalent interactions. (E) Residues in white spheres represent the network of interactions between the FMN domain and the linker, and dotted grey and black spheres represent the conserved or equivalent interactions at the FMN domain–linker and FMN–FAD domain interfaces respectively.

Conserved interactions (present in both parental enzymes) and equivalent interactions (neighbour positions in a CPR sequence alignment) should remain in the chimaeric CPRs (Figures 2C and 2D and see Supplementary Table S1). They represent globally 31 and 46% of the interactions present in yeast and rat CPR respectively. In contrast, specific interactions present in parental enzymes are probably lost in the chimaeric CPRs. Since these specific interactions involve strong contact energies (100% of the total AIs and more than 50% of the SBs), they probably lead to the destabilization of the closed form in HY and YH, coherent with the widely open structure of YHs (Figure 2E) [21].

Approx. 75% of interactions present at the FMN domain–linker interface are lost in the chimaeras. However, a new network of side-chain interactions involving the same residues is still detectable in the open structure of YHs (Figure 2E). The reorganization of the side-chain contacts in this region may be related to a necessary control of the distance between catalytic domains during CPR domain reorganization, preventing them from wandering too far from each other and preserving a good probability of closing back the structure. In this case, the linker, which appears to be quite flexible (high B factor values in all crystal structures), should not only act as a string but also as a spring, guiding the opening and closing movement of domains.

No major modifications of the backbone structure in the connecting domain between the closed and open crystallographic structures of CPR had been detected [21], leaving the function of this connecting domain in the various suspected CPR movements unexplained. Nevertheless, by simply increasing the number of contacts, the connecting domain may favour the recognition and the positioning of the FMN and FAD domains for interflavin ET. Interactions between the FMN and the connecting domains represent indeed one-third of all interactions and account for one-third of their bond energy (see Supplementary Table S1).

The drastic modification of the interfaces between domains in our chimaeric CPRs leads to a considerable loss of the interactions between domains. We then question how such dramatic changes could affect CPR function.

Kinetics of reduction of artificial or natural acceptors supported by chimaeric and parental CPRs

To evaluate and compare the ET capabilities of the chimaeric and parental CPRs, kinetic parameters for the reduction of artificial acceptors (ferricyanide and cyt c) were measured (Table 1). Since the formation of a stable complex between CPR and artificial acceptors is not required for ET, the half-saturation constant, K1/2, was used to describe this mechanism rather than the Michaelis–Menten constant Km.

View this table:
Table 1 Comparison of the specific activities and turnover numbers of chimaeric and parental CPRs

Values are means±S.D. Cyt c and ferricyanide reductions were measured with soluble proteins (Hs, Ys, HYs and YHs), whereas P450 assays were performed with membranous (Hm, Ym, HYm, YHm) and soluble (values in parantheses) CPRs.

Ferricyanide has the particularity to be reduced predominantly by the FAD domain [14,35], whereas cyt c behaves like a normal physiological acceptor, accepting the electrons flowing out of the FMN after an internal (FAD→FMN) ET [36]. Both YHs and HYs are still able to transfer electrons to external acceptors even though their activities are lower than those measured with parental CPR (Table 1). ET rates to ferricyanide are equivalent for parental and chimaeric CPRs sharing the same FAD domains (same kcatferricyanide for YHs and Hs or for HYs and Ys). Therefore the origin of the FMN domain has no effect on the ET rate from FAD to ferricyanide. However, the K1/2ferricyanide value obtained with YHs is one order of magnitude higher than the one obtained with Hs CPR, reflecting either a modification in the FAD accessibility to ferricyanide or a change in a rate-limiting step occurring before FAD→ferricyanide ET.

The kcatcytc values of HYs and YHs are respectively 10- and 15-fold lower than the human CPR kcatcytc value. K1/2cytc values are in the same order of magnitude for all CPRs (between 2.4 and 3.1 μM), with the exception of YHs which exhibits a 10-fold lower K1/2cytc (0.32 μM). Again, this change in the K1/2cytc value could reflect either an overall change in the geometrical conformation leading to a better accessibility to the FMN domain and/or a modification in any rate-limiting step occurring before cyt c reduction. Globally, the decrease in cyt c reductase activity in the chimaeras is rather due to low kcat values, i.e. low ET velocities.

Activities of the chimaeric CPRs were also measured with a natural acceptor: the human cytochrome P450 3A4 (P450–3A4). As described previously, human CPR is more efficient than yeast CPR towards P450–3A4 [37] (Table 1). Interestingly, despite their low efficiencies, all soluble CPRs were able to support P450 activities (approx. 15% of the activities measured with membranous CPRs). This result proves that the membranous anchor of CPR is not mandatory for ET to P450, although it plays an important role in the recognition and coupling between both proteins [38]. Again, chimaeric CPRs are less active than parental enzymes towards P450–3A4, but the observed decrease in activities is less important when measured with P450 as an acceptor than with cyt c. P450 specific activities in the presence of YHm and HYm (~15 nmol·min−1·nmol−1) are almost equivalent to the ones measured with Ym and only 3.6-fold lower than the activities measured in the presence of Hm (Table 1). The effect of domain substitution is therefore more important for cyt c reduction than on supporting P450–3A4 activity.

Conservation of the electronic pathway in chimaeric CPRs

To understand why ET from the FMN domain to acceptors for both chimaeras was lowered compared with parental enzymes, we investigated the conservation of the electronic pathway in YHs and HYs. We first verified that no bimolecular ET between the FAD and FMN domains originating from two different chimaeric CPRs occurred. Initial cyt c reduction velocity was measured as a function of CPR concentration (ranging from 0.5 to 50 nm, see Supplementary Figure S2 at This velocity is linearly dependent on the YHs or HYs concentration and the slopes are coherent with kcatcytc measured in steady-state kinetics experiments. This result confirms that the ET between the FMN and FAD is intramolecular and not between two different chimaeric CPRs.

To dismiss the possibility of a direct ET from FAD to natural or artificial acceptors, both parental and chimaeric CPRs were depleted from their FMN cofactor and their activities were measured and compared with the initial ones (see Supplementary Figure S3 at HPLC–MS analysis confirmed that the FAD cofactor was not removed during the FMN-depleted CPR preparation. FMN-depleted chimaeric CPRs no longer reduce cyt c (kcatcytc are between 3 and 15% of the initial activity). YHs and HYs regain 70–100% of their ET capacities towards cyt c when free FMN was added (two equivalent FMN to CPR). In contrast, addition of free FAD has absolutely no effect on the activities of either FMN-depleted or FMN-saturated chimaeric CPRs. Hence the FAD domains of HYs and YHs do not promote ET to external acceptors such as cyt c. We can therefore conclude that the electronic pathway is definitely conserved in both chimaeras, with a first internal ET between the two flavins and a second external ET from the FMN to the acceptor.

Rapid kinetics analysis

To understand further the origin of the loss of cyt c and P450 reduction activities, we analysed, using rapid kinetic techniques, CPR reduction by NADPH. FAD reduction and internal ET from FAD to FMN were followed using a stopped-flow apparatus coupled to an absorption spectrophotometer. Flavin reduction is characterized by the diminution of the absorption at 455 nm (Figures 3A, 3C and 3E) while the interflavin ET leading to the formation of blue sq is associated with an increase in the absorption around 585 nm (Figures 3B, 3D and 3F and see Supplementary Figure S4 at

Figure 3 Kinetics of Hs, Ys, YHs and HYs reduction

(A and B) Evolution of the absorption at 455 and 585 nm (respectively) from 0 to 4 s. (C) and (D) correspond to a shorter time range (0–0.25 s) at the same wavelengths for equimolar ratio between CPR and NADPH. (E) and (F) are equivalent to (A) and (B), but with a 10-fold excess of NADPH. Hs (■), Ys (◆), HYs (●) and YHs (▲).

Fitting of the data recorded within short durations (0–25 ms) (Figure 3C) allows the evaluation of the rate constants for the biphasic FAD reduction kinetic (Table 2). The first (rapid) rate constant is 1.5-fold higher in chimaeras than in parental CPR sharing the same FAD domain. In contrast, the second phase is slower for both chimaeras (approx. 1 s−1) compared with the yeast and human enzymes (3.4 and 7.3 s−1 respectively). An inversion of the amplitude values of the exponentials used for fitting is also observed: rapid and slow phases correspond respectively to 80 and 20% of the FAD reduction in parental CPR, whereas they represent respectively approx. 25 and 75% of the FAD reduction in chimaeras. Like yeast and human CPRs, both chimaeras can be distinguished by their capacity to be rapidly reduced by NADPH since the first rapid phase in HYs is 3.5-fold higher than for YHs (Table 2).

View this table:
Table 2 Apparent rate constants and amplitudes of flavin reduction and sq formation in parental and chimaeric CPR

Experiments were repeated in triplicate with an S.D. under 5%. S.D. of the fit and R values are given in Supplementary Table S2 at

Concerning sq formation (absorption at 585 nm), Ys accumulates less sq than Hs as described previously [25]. The rate of sq formation is lower in chimaeras compared with Hs (approx. 7- and 12-fold lower for HYs and YHs respectively), but one order of magnitude greater than the rate measured for Ys. Surprisingly, after 1 s, both chimaeric enzymes have accumulated almost 2-fold more sq than Hs (Figure 3B and Table 2).

The reduction of chimaeric and human CPRs continues after 1 s, leading to a further decrease in both absorptions at 455 nm and 585 nm. This reduction is even greater when NADPH is added in excess (Figures 3E and 3F, and see Supplementary Table S2) and corresponds to a second internal ET (disappearance of sq) followed by a novel FAD reduction [32].

In equimolar conditions (NADPH compared with CPR), less than 10% of CPR molecules are concerned by this slow phase (Figure 3B). This proportion reaches 18, 26 and 33% of YHs, HYs and the human CPR respectively when NADPH is used in 10-fold excess (Figure 3F). Chimaeric enzymes present the same global behaviour as human CPR except that the second internal ET rate is approximately one order of magnitude lower in YHs and HYs (0.13 and 0.10 s−1 respectively) compared with Hs (1.45 s−1).

Redox potentiometry

Flavin redox potentials are characteristic of the nature of the CPR [25] and are affected by the chemical and electrostatic environment [8,34,39]. We have measured the effect of domain substitution on the various redox potentials of FAD and FMN species (Table 3). We used the method of Das and Sligar [34] to fit the data obtained. This two-stage process enables an accurate evaluation of the flavin redox potentials (see Supplementary Table S3 and Figure S5 at The theoretical fitting curves are in good agreement with the experimental data (Figure 4) except for the absorption at 500 nm (isosbestic point) of HYs. However, this deviation has a very low impact on the final redox potential determination since values that are first calculated with the data recorded at these isosbestic points are recalculated and refined by fitting data corresponding to the maximum absorption wavelengths (450 and 585 nm) [34] for which the theoretical curve is very close to the experimental data points. Redox potentials of FAD and FMN in Hs and Ys parental CPR are in very good agreement with published values [25,33,40]. Redox titration experiments were repeated twice, and the estimation of the error on standard redox potential values is approx.±10 mV for E °1, E °2 and E °3 and between±10 and±15 mV for E °4 since the quality of spectra was slightly affected at low potentials.

Figure 4 Redox titration of parental and chimaeric CPRs

(A) Hs; (B) Ys; (C) HYs and (D) YHs. The absorbance is plotted against the redox potential of the solution (mV/SHE) at several wavelengths. Curves 1, 2, 3 and 4 correspond to the absorbance at 455 nm (▲, oxidized flavins), 585 nm (●, protonated sq), 500 nm (○, isosbestic point between oxidized flavins and sq) and 430 nm (△, isosbestic point between sq and hq). Symbols correspond to the experimental data points and continuous lines represent the fitting to the four-electrons Nernst equation [34].

View this table:
Table 3 Standard potentials of the flavinic redox pairs for parental and chimaeric CPR

Potentials are given in mV/SHE. Experiments were repeated twice and errors on values are estimated to ±10 mV for E °1, E °2 and E °3 and ±10–15 mV for E °4.

The differences between FAD potentials Δ(E °3E °4) are in the same order of magnitude in all CPRs (50–100 mV) and the FAD domain of HYs exhibit redox potentials that are equivalent to those of Ys and Hs. This behaviour is similar to other native diflavin reductases where the FAD domain does not seem to be affected by the nature and presence of other catalytic domains [8,9,33]. However, FAD potentials for YHs are lower than the parental ones, rendering FAD reduction by NADPH thermodynamically less favourable.

FMN standard potentials of the two chimaeric CPRs diverge from the values measured in parental proteins. The high E °1 value (approx. 0 mV) and the wide gap between E °1 and E °2 (>300 mV) clearly reflects a significant modification of the FMN environment in chimaeras. The E °1 values for YHs and HYs are coherent with a good stabilization of the FMN blue sq giving a strong green colour to YHs and HYs during their purification (results not shown) and a relatively strong accumulation of sq in the stopped-flow experiments. Again, these important modifications of the FMN redox potentials probably reflect a change in the chemical environment of the flavin in the two chimaeras.


Progress in the description of the various factors governing the domain movements of CPR has highlighted the role of redox potential and cofactor binding [13,20,24]. How the various domains control the geometry of the enzyme is still under heavy investigation. The exact role of the connecting domain is still unclear because it remains structurally unchanged between the open and closed forms of CPR [20,21]. However, a careful analysis of the interfaces between CPR domains highlights at least a static role of the connecting domain as it expands the interaction surface by a factor of 4 (~2000 Å2 compared with 500 Å2). The connecting domain permits an extension of the interaction surface between the FMN and FAD domains, in contrast with class III bacterial P450 systems that have smaller interaction surfaces between their redox partners. The linker itself does not present any secondary structure, but the interaction network between this loop and the FMN domain avoids its complete elongation. The role of this flexible loop was also highlighted by Hamdane et al. [20]: the deletion of several residues within the linker of the rat CPR led to an enzyme that crystallized in intermediate conformations between the closed and fully open one. This deletion had also a dramatic impact on the activity of the mutants which were no longer capable of internal FAD→FMN ET. It is therefore conceivable that this flexible loop serves as a spring or rail guiding the catalytic domains during their recognition before internal ET.

To highlight how the domain interfaces control CPR geometry, ET rates and cofactor redox potentials, we designed chimaeric enzymes substituting only the FMN domain while leaving the linker, connecting and FAD domains altogether. The biochemical characterization of HYs and YHs chimaeric CPRs strengthens the importance of interdomain recognition and positioning in the overall activity of CPR. The assigned role for each domain is preserved in our chimaeras and the three-dimensional structure of YHs is coherent with an absence of significant backbone conformational changes in each domain in comparison with the ones from the closed conformations [21]. Kinetic analysis demonstrated that both chimaeras have similar ET velocity with external acceptors. A simple model for CPR ET mechanism is presented in Figure 5.

Figure 5 Model mechanism of CPR ET during the reduction of external acceptors

Ared and Aox represent the reduced and oxidized acceptor respectively. Initial CPR species could contain either no electron (FMNox) or one electron (FMNsq), corresponding respectively to the yeast and human stable species.

This model describes the ET to external acceptors without the formation of any complex between CPR and the acceptor (no Km). However, rapid equilibria (k1 and k4) followed by rate-limiting elementary reaction steps (internal ET rates k3 and k6) would define cyt c reduction against cyt c concentration as a hyperbolic function. In such a mechanism, the decrease in the internal (FMN→FAD) ET in chimaeras leads to the observed drastic decrease in kcatcytc values (YHs and HYs), but also in the diminution of K1/2cyt c (YHs). We can therefore conclude that the recognition and the ET to acceptors may not be affected in our chimaeras, and that the defect in the acceptor reduction rates may only come from a decrease in the rate of internal ET.

Rapid kinetics experiments showed an inversion of the amplitudes in the rapid and slow FAD reduction phases between chimaeric and parental CPRs (Table 2). This behaviour could be explained by different equilibrium states between several conformations adopted by the CPR in solution [22]. Some conformations could be competent for ligand binding and/or FAD reduction (80% of parental CPR and 20–30% of chimaeric CPR), while the rest of the proteins would remain in an unfavourable conformation for hydride transfer (20% of parental and 70–80% of chimaeric CPR). The interconversion between these conformations could constitute a rate-limiting step accounting for the second (slow) FAD reduction phase [23,41].

In a recent publication, Hay et al. [23] described the conformational landscape of the CPR geometry and the impact of interdomain distance on FAD→FMN ET efficiency. Our results confirm that interdomain interactions and internal ET are essential factors governing the ET efficiency, but also highlight that a dramatic change in the interface does not abolish domain–domain ET. The modification of the domain interactions should also lead to changes in the CPR geometry and domain mobility.

The FMNox/sq redox potential values for YHs and HYs reveal that the FMN protonated sq form is greatly stabilized, leading to a blue sq even more stable than its mammalian counterpart. Other flavoproteins also present unusual FMN redox potentials (positive E °1 values like HYs) [4244]. Alterations of the FAD and the FMN redox potentials are likely to reflect a change in their chemical and electrostatic environment induced by a modification of the enzyme geometry. Several groups have pointed out the influence of cofactor binding and environmental modification on the FAD potentials and hydride transfer in diflavin reductases or ferredoxin reductases [16,32,34,45,46]. Surprisingly, FMN and FAD redox potentials remain almost identical in native human CPR and in isolated FMN and connecting/FAD domains “indicating that the flavins are located in discrete environments and that these environments are not significantly disrupted by genetic dissection of the domains” [33]. We should, however, keep in mind that potentiometry experiments are carried out in the absence of any ligand or acceptors and the measured values of flavin redox potentials originate from CPR molecules that are probably in equilibrium between several conformations. The conservation of the redox potential values of the cofactors in isolated domains and native CPR demonstrates that, under these chemical conditions, CPR adopts a flexible geometry where each domain is highly independent and probably apart from each other. In this context, the significant modifications of flavin potentials measured in our chimaeric CPRs may also reflect a change in the dynamic equilibrium of HYs and YHs in solution in comparison with parental CPR.

Our work confirms the essential role of the interdomain interface in controlling CPR geometry and mechanism. The surface electrostatic potentials of the two catalytic domains are well conserved across CPRs from different species and even across diflavin reductases in general [4750]. However, the biochemical properties of our chimaeras demonstrate that preserving the global electrostatic potential is not sufficient to sustain a high internal ET efficiency. Our results also support the idea that the interface between FMN and connecting/FAD domains in CPR may affect the equilibrium between subpopulations.

Nucleotide binding seems to govern the closing movement since all X-ray structures of the CPR in the closed form were obtained in the presence of NADP+ [18,19] and several biochemical analyses also support this hypothesis [23,24,41,45]. The crystallographic structure of the open form of YHs in the absence of nucleotide [21] and the characterization of a dynamic equilibrium between an open and a more closed forms by SAXS analysis [22] tend to confirm that the release of the ligand is one of the factors allowing CPR opening. Other factors, notably redox states of the flavins, are also suspected to promote or reflect CPR domain movements [22]. The control of the opening/closing phenomenon involves various type of effectors: some directly connected to the structure of the protein (redox potentials and interfaces) and the others being external factors (ligands, acceptors). The biochemical behaviour of CPR may be the result of a combination of these various effectors. Our current hypothesis is that the binding of external effectors has a direct effect on the structure of the interface, resulting in surface modifications allowing the conformational reorganization of the CPR domains. The coupling of methods specifically designed for the study of dynamic objects such as SAXS, NMR or fluorescence spectroscopy will probably constitute a more appropriate way to further question the relationship between CPR mechanism and movements during its interactions with redox partners.


Louise Aigrain, Denis Pompon and Gilles Truan conceived the work. Louise Aigrain carried out the experiments and analysed the data. Louise Aigrain, Denis Pompon and Gilles Truan prepared the paper.


This research recieved no specific grant from any funding agency in the public, commercial or not for-profit sectors.

Abbreviations: AI, aromatic interaction; CPR, NADPH–cytochrome P450 reductase; cyt, c, cytochrome c; ET, electron transfer; HB, hydrogen bond; hq, hydroquinone; HI, hydrophobic interaction; NOS, nitric oxide synthase; ox, oxidized; red, reduced; SAXS, small-angle X-ray scattering; SB, salt bridge; SHE, standard hydrogen electrode; sq, semiquinone


View Abstract