NADPH-cytochrome P450 oxidoreductase. Structural basis for hydride and electron transfer.

NADPH-cytochrome P450 oxidoreductase catalyzes transfer of electrons from NADPH, via two flavin cofactors, to various cytochrome P450s. The crystal structure of the rat reductase complexed with NADP(+) has revealed that nicotinamide access to FAD is blocked by an aromatic residue (Trp-677), which stacks against the re-face of the isoalloxazine ring of the flavin. To investigate the nature of interactions between the nicotinamide, FAD, and Trp-677 during the catalytic cycle, three mutant proteins were studied by crystallography. The first mutant, W677X, has the last two C-terminal residues, Trp-677 and Ser-678, removed; the second mutant, W677G, retains the C-terminal serine residue. The third mutant has the following three catalytic residues substituted: S457A, C630A, and D675N. In the W677X and W677G structures, the nicotinamide moiety of NADP(+) lies against the FAD isoalloxazine ring with a tilt of approximately 30 degrees between the planes of the two rings. These results, together with the S457A/C630A/D675N structure, allow us to propose a mechanism for hydride transfer regulated by changes in hydrogen bonding and pi-pi interactions between the isoalloxazine ring and either the nicotinamide ring or Trp-677 indole ring. Superimposition of the mutant and wild-type structures shows significant mobility between the two flavin domains of the enzyme. This, together with the high degree of disorder observed in the FMN domain of all three mutant structures, suggests that conformational changes occur during catalysis.

Other physiological electron acceptors of CYPOR include heme oxygenase (3), cytochrome b 5 (4), and fatty acid elongase (5). Although non-physiological, CYPOR is also capable of reducing cytochrome c in vitro. Kinetic, spectroscopic, and potentiometric studies using a reconstituted liver microsomal monooxygenase system indicate that the hydride ion is transferred from 4 proximity is provided by the connecting domain which bifurcates the FAD domain, and brings the two flavin cofactors within ~4 Å of each other, thus facilitating direct electron transfer.
NMR spectroscopic studies (14) indicate that the pro-R hydrogen attached to the C4 atom of NADPH is transferred directly to FAD as a hydride ion. However, the wild type structure of rat CYPOR complexed with NADP + (13) shows the re-face of the isoalloxazine ring of FAD, the face closer to the surface of the molecule, to be stacked against the indole ring of the penultimate C-terminal residue, W677. The nicotinamide ring of NADP + is projected away from the isoalloxazine ring, in a disordered state towards the protein surface, thus giving no structural information regarding the interaction between the nicotinamide ring and FAD. Such topology is reminiscent of a number of related flavoproteins including FNR (15), phthalate dioxygenase reductase (16), thioredoxin reductase (17), and sulfite reductase (18); all of which are believed to undergo structural rearrangement to allow nicotinamide access to the isoalloxazine ring. Support for this hypothesis also comes from crystal structure analyses of glutathione reductase (19), NADH peroxidase (20), NAD(P)H quinone reductase (21), a mutant form of FNR (22), and most recently, adrenodoxin reductase (23); all of which show direct stacking of the isoalloxazine and nicotinamide rings, arrangements which may represent intermediates in their respective catalytic cycles. Structures of an FNR mutant, where the aromatic residue (Y308) found above the re-face of the isoalloxazine ring in the wild type structure has been mutated to serine, show the nicotinamide ring of NADP + or NADPH to be orientated over the central pyrazine ring of the isoalloxazine moiety.
Several mutants of both rat (24) and human (25;26) CYPOR have been constructed which have mutations involving the stacking aromatic residue, and their kinetic properties determined. To fully assess the role of W677 in the hydride transfer mechanism, we selected two by guest on July 9, 2020 http://www.jbc.org/ Downloaded from 5 mutants of rat CYPOR for crystallographic studies, W677X and W677G. The W677X mutant has both the aromatic residue that stacks on top of the isoalloxazine ring of FAD (W677) and the adjacent C-terminal residue (S678) deleted. This was designed to prevent the C-terminus from precluding nicotinamide binding, and allow NADP + to bind in a productive manner. The W677G protein, which has the stacking tryptophan residue replaced with glycine, was made to determine that any changes observed in the W677X structure compared to the wild type structure are solely due to the absence of W677, and not the presence of the carboxylate group of V676. Finally, the structure of a third mutant form (S457A/C630A/D675N, or "triple mutant") was determined in which the structure of the carboxy terminus is preserved, but three catalytically important residues (S457, C630, and D675) have been substituted. The structures of the W677X and W677G mutants reveal the detailed binding mode of the nicotinamide moiety of NADP(H) and provide insights into the energetics of its binding. The structure of the triple mutant, together with those of the two W677 mutants, correlates with previously observed kinetic data and defines the role of these residues in the hydride transfer mechanism. 7 formation of air-stable semiquinone was observed after addition of NADPH (200 µM) under aerobic conditions. For measurements of NADP + difference spectra, NADP + (130 µM) was added to samples of the oxidized enzyme and spectra re-measured; the oxidized enzyme spectra were then subtracted from this to determine difference spectra. Cytochrome c reductase activities and primary deuterium isotope effects were determined as described previously (27;29). Synchrotron data were collected at beam-line BM-14C, Advanced Photon Source. Complete datasets were reduced and scaled using the HKL suite (31). Initial phases were calculated using the difference Fourier method with the refined wild type structure [(13); PDB accession code, 1AMO]. Multiple rounds of reciprocal space refinement were carried out using CNS (32), and subsequent manual fitting with the graphic visualization program O (33) was performed using 2Fo -Fc and Fo -Fc maps. Automatic addition of water molecules was done using the program DDQ (34), followed by manual addition or deletion of water molecules until R-free convergence was achieved.

General Properties of the Mutants-
The visible absorption spectra of the oxidized forms of the W677X and W677G mutants are similar to that of the wild-type enzyme ( Figure 1A). Furthermore, addition of NADPH under aerobic conditions reduces the enzyme and produces the air-stable semiquinone spectrum, illustrated by the broad peak around 585 nm ( Figure 1A), suggesting that these mutants are competent in hydride and interflavin electron transfer. In addition, as seen by the peak at ~510 nm in the difference spectrum ( Figure 1B), both mutants are capable of forming charge transfer complexes between the FAD isoalloxazine moiety and oxidized nicotinamide ring of NADP + . Comparison of the 510 nm absorbance peak in the difference spectra shows that the amounts of the charge transfer complex between FAD and the nicotinamide group of NADP + are substantial in the two mutants, whereas that in the wild type is negligible. This suggests that, unlike the wild type, both mutants are capable of forming equally stable complexes.
Cytochrome c reductase activities of the W677X and W677G mutants are 4% and 2%, respectively, of that of wild-type (Table I), indicating the importance of the tryptophan residue in catalysis. Notably, K m NADPH is decreased substantially by both of these mutations, suggesting that NADPH may bind more tightly to the protein in the absence of W677. This is further confirmed by a 5-to 13-fold decrease in K i NADP+ , and a decrease in K i NAD+ by 2 orders of magnitude (Table I). Furthermore, the deuterium isotope effect is decreased in W677X, suggesting a change in the rate-limiting step from hydride transfer to cofactor release.
General Crystal Properties-Crystals of all three mutants took about two to three weeks to reach their maximum size of 0.2 mm x 0.2 mm x 0.02 mm. These crystals appeared as thin 9 yellow plates and were prone to forming multiple layers. Upon perfecting the cryo-conditions, crystals could be cooled to -180 °C and diffracted reasonably well to between 1.8 Å and 2.7 Å resolution, depending on the mutant crystal type (Table II), for the entire duration of data collection (~16 hours in-house, ~0.5 hours at the synchrotron). For the W677X and W677G mutants, multiple data sets were collected in an attempt to reduce the relatively high mosaicity, however these all gave similar results suggesting that this was an inherent property of these crystals.
W677X Structure (deletion of W677 and S678)-Crystals used for data collection were of a similar quality to those of the wild type form, diffracting to 2.6 Å resolution and having a relatively high mosaicity of ~1° (Table II). As with all the mutant structures studied, initial difference Fourier analysis of the diffraction data made use of the refined wild type rat CYPOR structure (13). Overall, superimposition of the complete molecule reveals very little difference between the wild type and mutant structure folds (Cα r.m.s. deviation ~0.6 Å). After one round of simulated annealing, however, the Fo-Fc map clearly showed an extension of density (> 2σ) from the pyrophosphate linker of NADP + to the region above the re-face of the isoalloxazine ring of FAD, which is occupied by the indole ring of W677 in the wild type structure. In contrast, the 2Fo-Fc map was distorted with respect to the positioning of W677 in the wild type structure, and there was poor density extending back into the preceding amino acids of the peptide. However, well-defined electron density for the phospho-AMP region of NADP + showed this part of the cofactor was bound to the enzyme in the same orientation as observed in the wild type structure. This immediately suggested that the nicotinamide ring had substituted for the indole ring of W677, and that significant torsion about the pyrophosphate linker of NADP + occurred to achieve this stacked conformation. Further refinement resulted in the nicotinamide moiety being tilted ~30° with respect to the isoalloxazine ring (Figure 2), in a similar orientation to that seen in the Y308S mutant of FNR (22). The A-face of the nicotinamide ring is stacked against the re-face of the isoalloxazine ring, with the pro-R hydrogen of the C4 atom positioned for transfer to the N5 atom of FAD, consistent with the stereospecificity of hydride transfer (14).
This stereochemistry is identical to that seen in FNR (22); however, in the case of the CYPOR structure, the center of the nicotinamide ring is positioned over the N5 atom of the isoalloxazine ring, whereas with FNR the centers of both the nicotinamide and isoalloxazine rings are directly opposing each other. This suggests, as with the Y308S FNR mutant, that the structure mimics the productive complex between NADPH and FAD, allowing direct hydride transfer to occur.
The arrangement of three residues known to be critical for hydride transfer, S457, C630, and D675 (27), is similar to that of the wild type enzyme ( Figure 3). However, the position of the nicotinamide group juxtaposed to the isoalloxazine ring of FAD shows the three catalytic residues forming contacts with NADP + not seen in the wild type structure ( Figure 4). The potential van der Waal's interaction seen in the wild type structure between the hydroxyl group of S457 and the N5 atom of FAD (~3.9 Å) is maintained in the W677X structure; however a weak hydrogen bond (~3.5 Å) is present between S457 and the carboxamide group of NADP + .
D675 forms a relatively strong hydrogen bond (~2.5 Å) to the carboxamide group, presumably leading to displacement of the side chain of D675 with respect to the wild type structure, resulting in loss of hydrogen bonds to both S457 and C630. C630 is positioned in the same manner as in the wild type, but in close contact with the C4 atom of the nicotinamide ring. A similar arrangement of these three catalytic residues around the nicotinamide ring is also seen in the FNR Y308S mutant structure. However, the carboxylate group of V676 in the W677X mutant, which is present two residues before that of the wild type and the two other mutant proteins, forms a hydrogen bond to the carboxamide moiety of NADP + .
The hydrogen bonding scheme around the pyrimidine side of the FAD isoalloxazine ring is the same as that seen in the wild type structure ( Figure 4). The N1 and O2 atoms of the flavin ring form hydrogen bonds with a water molecule (WAT 1 in Figure  and untraceable density confirmed this disorder, and therefore this region was omitted prior to any further refinement.  (Table II). However, difference Fourier analysis using phases taken from the wild type structure produced an initial 2Fo-Fc map which showed, unlike the W677X structure, the FMN domains of both molecules in the asymmetric unit to be relatively well ordered. Inspection of this map above the re-face of the isoalloxazine ring of FAD showed that the indole ring of W677 was absent, confirming this residue had been substituted for glycine. The Fo-Fc map suggested the nicotinamide of NADP + to be positioned in a similar manner to that seen in the W677X structure. However, in contrast to the W677X structure, the final 2Fo-Fc and Fo-Fc OMIT maps show that the nicotinamide ring and the connecting ribose group are not as well defined as in the W677X structure. This suggests that there is likely to be a significant population (20~30 % as estimated from B-factor values) of molecules in the crystal lattice adopting the nicotinamide moiety in a wild type-like conformation, being out towards the protein surface. This is consistent with the fact that this mutant has a slightly higher reductase activity than the W677X mutant (Table I). As in the W677X structure, the carboxamide group of the bound nicotinamide ring forms a hydrogen bond with D675, pulling it away from S457 and C630, resulting in loss of these hydrogen bonds that are found in the wild type structure. The  (Table II). The scaled data showed a significant drop in mosaicity, being approximately half that seen in the other CYPOR forms studied (Table II). This large improvement in crystal quality appears to be at least in part due to mutation of the three catalytic residues; S456, C630, and D675. However, inspection of the crystal lattice does not give any clear indication as to their involvement in crystal packing, and hence diffraction resolution.
Data were analyzed using the difference Fourier method as before. The resulting Fo-Fc and 2Fo-Fc maps were used to improve the initial model, with introduction of the three mutated residues. This was followed by the refinement procedure as described in the methods section. As with the W677X and W677G mutants, the final model shows the same overall fold as seen in the wild type structure. This extends to the hydrogen bond networks, including the two conserved water molecules that are bound to the FAD isoalloxazine ring in all the structures presented.
However, the nicotinamide ring and connecting ribose group of NADP + are disordered in a similar fashion to that seen in the wild type. As with all other structures solved so far, the loops between residues #500 and #506 of both molecules in the asymmetric unit are disordered; however, sufficient electron density was present to allow these residues to be modeled in for both molecules. Inspection of the mutated residues shows that the Cα -Cβ bonds of the alanine side chains at positions 457 and 630 are oriented in the same manner as their wild type counterparts, and that the side chain of N675 is positioned almost identically to that of the wild type D675.  Crystal structures of glutathione reductase (19), NADH peroxidase (20), and more recently a substitution mutant of FNR (22), show that the nicotinamide moiety of the pyridine nucleotide cofactor can position itself in place of the aromatic residue. These structures demonstrate that NAD(P)H binding occurs in a bipartite manner, with the 2'-phosphate of AMP bound to a positively charged region at the protein surface, and a second interaction between the nicotinamide and isoalloxazine rings towards the protein core. In the case of CYPOR, one of the surface residues that interact with the negatively charged 2'-phospho AMP region, R597, has been shown to provide approximately 5 Kcal mol -1 of binding energy (35). The aromatic interactions are probably dominated by π-π orbital overlap, where electrons from the π orbitals of both the nicotinamide and isoalloxazine rings hybridize to form a weak π-π interaction (typically 1 -3 Kcal mol -1 (36)). This range of interaction energy agrees with recent data obtained using house fly CYPOR with various nucleotide analogs (37). However, the nature of this π orbital interaction will change upon transfer of the hydride ion, since the electronic characters of both moieties are altered.

Downloaded from
Structural investigation on the interactions between the nicotinamide moiety of the pyridine nucleotide and CYPOR has been elusive. The structure of the wild type enzyme complexed with NADP + revealed that the nicotinamide ring is disordered above the protein surface. In the current study, we demonstrate the direct interactions of the nicotinamide ring with FAD and the polypeptide chain using W677 mutants. As is seen in the wild type crystal interactions. In the wild type structure, the indole ring of W677 clearly acts to block binding of the nicotinamide ring of NADP + to the isoalloxazine ring of FAD, thus explaining why, in the case of the W677X mutant, deletion of this residue (and S678) results in 10-fold decreases of both the K m for NADPH and K i for NADP + (24), and a 100-fold decrease in K i for NAD + (Table   I). The larger change in affinity observed for NAD + than for NADP + is probably due to the fact that strong binding interactions between the phospho-AMP and the enzyme are not as necessary for cofactor binding in the absence of repulsive interactions between the indole ring of W677 and the nicotinamide moiety. This change in affinity is also observed for the W677G mutant. Similar results have recently been reported for W676 mutants of the human enzyme (corresponding to W677 in the rat enzyme) (25). Tighter binding of NADP(H) likely alters the rate-limiting step for cytochrome c reduction from hydride transfer to NADP + release, accounting for both the overall decrease in catalytic activity and elimination of the deuterium isotope effect with NADP[ 2 H] (Table I). It therefore appears that W677 functions in regulating the binding and release of NADPH and NADP + , respectively. The W677G structure confirms the role of the tryptophan residue in regulating nicotinamide binding, and the cofactor binding mode determined for the W677X mutant is not an artifact of the premature carboxy terminus.
Furthermore, movement of W677 is probably limited to the indole side chain, since the backbone topology of the W677G mutant is comparable to the wild type structure.
The geometry of the pyridine nucleotide and flavin cofactors observed in the present studies are in contrast to those seen in a number of other flavoenzymes where there is a direct interaction between the nicotinamide and isoalloxazine moieties. These include glutathione reductase (19), NADH peroxidase (20), and adrenodoxin reductase (23) that must be formed upon release of the hydride ion. In all the CYPOR structures determined so far, S457 is located ~4 Å away from the N5 atom of FAD and on the same plane as the flavin ring (Figure 3), strongly suggesting this residue may be involved in stabilizing the semiquinone form of FAD. Also, as seen in the wild type structure, S457, in conjunction with D675 and C630, forms a hydrogen bond network (Figure 4) that may promote proton release from the hydroquinone/semiquinone. It is interesting to note that this hydrogen bond network is broken when the nicotinamide group binds to the active site. As is seen in the W677X and W677G structures, the carboxylate group of D675 is moved away from both S457 and C630 to form a hydrogen bond with the carboxamide group of NADP + . This precludes D675 from participating in the hydrogen bond network when the nicotinamide ring of NADP(H) binds to the enzyme. It is therefore not unreasonable to expect that the nicotinamide ring must be released from the active site before the proton dissociates from the N5 atom of FAD.
Given the kinetic and structural data available, a plausible mechanism for hydride transfer is proposed in Figure 6.