Role of acidic residues in the interaction of NADPH-cytochrome P450 oxidoreductase with cytochrome P450 and cytochrome c.

Site-directed mutagenesis of the acidic clusters 207Asp-Asp-Asp209 and 213Glu-Glu-Asp215 of NADPH-cytochrome P450 oxidoreductase demonstrates that both cytochrome c and cytochrome P450 interact with this region; however, the sites and mechanisms of interaction of the two substrates are clearly distinct. Substitutions in the first acidic cluster did not affect cytochrome c or ferricyanide reductase activity, but substitution of asparagine for aspartate at position 208 reduced cytochrome P450-dependent benzphetamine N-demethylase activity by 63% with no effect on KP450m or KNADPHm. Substitutions in the second acidic cluster affected cytochrome c reduction but not benzphetamine N-demethylase or ferricyanide reductase activity. The E213Q enzyme exhibited a 59% reduction in cytochrome c reductase activity and a 47% reduction in KCyt cm under standard conditions (x0.27 M potassium phosphate, pH 7.7), as well as a decreased KCyt cm at every ionic strength and a shift of the salt dependence of cytochrome c reductase activity toward lower ionic strengths. The E214Q substitution did not affect cytochrome c reductase activity under standard conditions, but shifted the salt dependence of cytochrome c reductase activity toward higher ionic strengths. Measurements of the effect of ionic strength on steady-state kinetic properties indicated that increasing ionic strength destabilized the reductase-cytochrome c3+ ground state and reductase-cytochrome c transition state complexes for the wild-type, E213Q, and E214Q enzymes, suggesting the presence of electrostatic interactions involving Glu213 and Glu214 as well as additional residues outside this region. The ionic strength dependence of kcat/KCyt cm for the wild-type and E214Q enzymes is consistent with the presence of charge-pairing interactions in the transition state and removal of a weak ionic interaction in the reductase-cytochrome c transition-state complex by the E214Q substitution. The ionic strength dependence of the E213Q enzyme, however, is not consistent with a simple electrostatic model. Effects of ionic strength on kinetic properties of E213Q suggest that substitution of glutamine stabilizes the reductase-cytochrome c3+ ground-state complex, leading to a net increase in activation energy and decrease in kcat. Glu213 is also involved in a repulsive interaction with cytochrome c3+. Cytochrome c2+ Ki for the wild-type enzyme was 82.4 microM at 118 mM ionic strength and 10.8 microM at 749 mM ionic strength; similar values were observed for the E214Q enzyme. Cytochrome c Ki for the E213Q enzyme was 17.6 microM at 118 mM and 15.7 microM at 749 mM ionic strength, consistent with removal of an electrostatic repulsion between the reductase and cytochrome c2+.

The microsomal and nuclear envelope flavoprotein NADPHcytochrome P450 oxidoreductase (P450R) 1 (NADPH:ferrihemoprotein oxidoreductase, EC 1.6.2.4) catalyzes electron transfer from NADPH to the cytochromes P450 (1) and other microsomal proteins (2)(3)(4), as well as to nonphysiologic electron acceptors such as cytochrome c, ferricyanide, menadione, and dichlorophenolindophenol (5,6). There is a substantial body of information on the structure and mechanism of this multidomain enzyme (7,8), as well as on its gene structure and regulation (8 -10). Although crystals of P450R have been obtained (11), the crystal structure has not been solved. FMN, FAD, and NADPH-binding domains of P450R have been identified by sequence comparisons with flavoproteins of known three-dimensional structure (8,12,13) and site-directed mutagenesis has identified amino acids necessary for binding of FMN and NADPH (14 -16). The orientation of the nicotinamide and FAD isoalloxazine rings has been shown to be exo, with transfer of the pro-R hydrogen of NADPH to FAD (17). The kinetic mechanism with the substrate cytochrome c is nonclassical (twosite) Ping Pong, with the reaction of cytochrome c at the electron acceptor site being Ping Pong at high ionic strength and random sequential at low ionic strength (18,19).
Less is known about the substrate binding sites of P450R. A number of electron-transfer complexes are stabilized by electrostatic interactions which play a role both in complex formation and in orienting the two redox centers (20 -22); however, evidence also exists for involvement of multiple hydrophobic and van der Waals interactions (23)(24)(25). Chemical modification and cross-linking studies (26 -29) have provided evidence for electrostatic interactions between P450R and its substrates. For example, neutralization of carboxyl groups on P450R by 1-ethyl-3(3-dimethyl-aminodipropyl)carbodiimide (EDC) has been shown to inhibit both cytochrome c reductase activity and cytochrome P450-dependent monooxygenase activity, with no effect on electron transfer to ferricyanide or cytochrome b 5 (27)(28)(29). Modification of Lys 382 of CYP2B4 with fluorscein isothiocyanate has been shown to inhibit reductase-dependent but not cumene-hydroperoxide-dependent monooxygenase activity (30) and Shimizu et al. (31) have identified by site-directed mutagenesis seven lysyl and/or arginyl residues in CYP1A2 which affect cytochrome P450-dependent monooxygenase activity. In contrast, studies on the ionic strength dependence of P450 reduction argue against charge-pairing between P450R and cytochrome P450 (32)(33)(34).
In attempts to identify specific side chain interactions between P450R and substrate, Nisimoto (26) characterized an EDC cross-linked complex between P450R and cytochrome c, where a lysyl residue in cytochrome c was covalently linked to an acidic residue in the region between residues 200 -220 of the reductase. This region contains two clusters of acidic amino acids, Asp 207 -Asp 208 -Asp 209 and Glu 213 -Glu 214 -Asp 215 ; one or more of which could charge-pair with cytochrome c. This study investigates the role of these residues in substrate binding and electron transfer by site-directed mutagenesis.

MATERIALS AND METHODS
Construction of the P450R expression plasmid, pOR263, and methods for the expression and purification of the Escherichia coli-expressed reductase have been described previously (14). For site-directed mutagenesis, a 750-base pair BamHI/SacI fragment from pOR263 was cloned into M13mp19. Mutagenesis was carried out by the method of Zoller and Smith (35) or by using the oligonucleotide-directed mutagenesis kit obtained from Amersham Corp. The following oligonucleotides were synthesized at the University of Wisconsin Biotechnology Center:  D207N, 5Ј-CGTCATCATTACCAAG-3Ј; D208N, 5Ј-CCCGTCATTATCA-CCA-3Ј; D209N, 5Ј-TTCCCGTTATCATCAC-3Ј; E213Q, 5Ј-ATCCTCTT-GCAAGTTC-3Ј; E214Q, 5Ј-GAAATCCTGTTCCAAG-3Ј; E215Q, 5Ј-GA-TGAAATTCTCTTCC-3Ј; D207N/D208N, 5Ј-TTCCCGTCATTATTACC-AAG-3Ј; E213Q/E214Q/D215N, 5Ј-ATGAAATTCTGTTGCAAGTTC-3Ј;  D207N/D208N/D209N, 5Ј-TTCCCGTTATTATTACCAAG-3Ј. Mutant phage were identified by plaque hybridization and confirmed by sequencing. The mutant BamHI/SacI fragments were cloned into pOR263, and the mutant expression plasmids were verified by restriction mapping and sequencing. Mutants with multiple substitutions were prepared using combinations of the above oligonucleotides. Cytochrome c and ferricyanide reductase activities and flavin content were determined as described previously (14). Assays in potassium phosphate were carried out in 271 mM potassium phosphate, pH 7.7, giving an ionic strength of 765 mM. Assays in KCl contained 10 mM potassium phosphate, pH 7.7, and KCl to achieve the desired ionic strength. Cytochrome c 2ϩ was prepared by dissolving 15 mg of cytochrome c 3ϩ in 0.5 ml of 10 mM potassium phosphate, pH 7.7, adding 1 mg of sodium dithionite, and passing the reduced cytochrome c over a 5-ml Sephadex G10 column to remove excess dithionite. Concentrations of reduced and oxidized cytochrome c were determined by measuring the absorbance at 550 nm (⑀ ϭ 21 mM) with and without added dithionite. Cytochrome P450 was purified from phenobarbital-induced rat liver microsomes as described by Guengerich and Martin (36). Assays for benzphetamine N-demethylation contained 0.19 M P450R, 0.04 -0.72 M cytochrome P450, 20 g/ml sonicated dilauroylphosphatidylcholine, 1 mM benzphetamine, 1 mM NADPH, and 54 mM HEPES, pH 7.4. P450R, cytochrome P450, and lipid were incubated at 37°C for 5 min in covered microtiter plates in a volume of 20 l. Reactions were then initiated by addition of 180 l of HEPES buffer containing NADPH and benzphetamine. After 4 min, reactions were terminated by addition of 50 l of 30% trichloroacetic acid. After centrifugation to remove protein, formaldehyde in the supernatants was assayed by the Nash method, as described by Werringloer (37). Fig. 1 shows the amino acid sequence of rat P450R between residues 200 and 221. This region contains two clusters of acidic residues which have been identified as being cross-linked by EDC to cytochrome c (26). Comparison of the mammalian, yeast, and bacterial reductases identifies invariant (Asp 207 ) and conserved (Glu 213 ) acidic residues, as well as invariant or conserved residues at positions 204, 216, 219, 220, and 221. Comparison of the rat reductase sequence with that of Desulfovibrio vulgaris flavodoxin reveals identity at positions 208, 215, and 219 as well as the conservation of an acidic residue at position 214. The results presented here focus on the effects of introduction of the corresponding amides for the acidic residues in these two clusters, changes which produce minimal steric effects but remove charge-pairing interactions and introduce the potential for hydrogen-bonding. Table I shows that substitution of asparagine for Asp 208 affects primarily the interaction of P450R with cytochrome P450, as evidenced by a 63% reduction in the P450-dependent benzphetamine N-demethylase activity and no effect on cytochrome c and ferricyanide reduction. Similar changes at either positions 207 or 209 had no effect on the catalytic activity with cytochrome P450, cytochrome c, or ferricyanide. The activity of the double mutant D207N/D208N was similar to that of D208N. The drop in N-demethylase activity was accompanied by a decrease in k cat of 55% for D208N and 66% for D207N/ D208N, with no change in K m P450 (Table II). No significant changes in K m NADPH or K m Cyt c were seen in any of the cluster I single mutants.

RESULTS
In contrast to the results seen with cluster I substitutions, mutations in the second acidic cluster affected interaction of the reductase with cytochrome c but not with cytochrome P450 (Table III). None of the cluster II substitutions affected P450dependent N-demethylase activity; however, under standard assay conditions (0.27 M potassium phosphate, pH 7.7), cytochrome c reductase activity of the E213Q mutant was decreased by 59%. E214Q, D215N, D207N/E214Q, and D207N/ D215N exhibited normal catalytic activities, while the cytochrome c reductase activities of the D207N/E213Q and E213Q/E214Q/D215N mutants were decreased by 70%. As with cluster I mutations, ferricyanide activity was not affected by any of the cluster II mutations.
Each of the cluster I and cluster II single mutants and mutants carrying two or three substitutions had the expected flavin stoichiometry, i.e. 1 mol of FMN and 1 mol of FAD per mol of protein. Replacement of all six acidic residues with their corresponding amides, however, produced a protein with the normal complement of FAD but only 0.4 mol of FMN/mol of protein. The specific activity of this six-place mutant was 10% of wild type and was significantly less than that of E213Q and E213Q/E214Q/D215N. This reduced activity is probably the result of structural perturbations affecting FMN binding.
Examination of the kinetic properties of the cluster II mutants in 0.27 M potassium phosphate, pH 7.7, shows that the decreased specific activity of the E213Q mutant was accompanied by a 60% decrease in k cat and a 47% decrease in K m Cyt c (Table IV). Similar decreases were observed in each of the  Cyt c was increased 1.6-fold for the E214Q and D207N/E214Q mutants, with no change in k cat . Kinetic properties of the D215N mutant were the same as wild type. K m P450 and benzphetamine N-demethylase activity were not affected by any of the cluster II mutations. Substitution at Glu 213 produced a slight, but statistically insignificant, decrease in K m NADPH . However, K m NADPH for the three-(E213Q/ E214Q/D215N) and six-place (D207N/D208N/D209N/E213Q/ E214Q/D215N) mutants was significantly decreased to 58 and 44%, respectively, of wild type.
The effects of ionic strength on the catalytic activities of the wild-type and mutant enzymes are shown in Fig. 2. When ionic strength was increased by increasing the concentration of potassium phosphate ( Fig. 2A), maximum cytochrome c reductase activity was observed at an ionic strength of approximately 452 mM for the wild-type enzyme, 339 mM for the E213Q enzyme, and 678 mM for the E214Q enzyme. Maximum activity of the E213Q/E214Q/D215N mutant was observed at 502 mM (not shown). When ionic strength was varied by increasing the concentration of KCl (Fig. 2B), wild-type k cat increased to a maximum in the range between 208 and 479 mM, while E213Q k cat was maximal at 208 mM and declined thereafter. k cat for the E214Q enzyme was maximal at ionic strengths greater than 479 mM. While k cat of the E213Q enzyme was only 28% of wild type at 749 mM, it was the same as that of wild type at low ionic strength.    The ionic strength dependence of K m Cyt c for the wild-type, E213Q, and E214Q proteins is shown in Table V. For all three proteins, K m Cyt c increased with increasing ionic strength. At each ionic strength, K m Cyt c for E213Q was lower than that found for either the wild-type or E214Q enzymes; this difference was more pronounced at higher ionic strengths. In addition, K m Cyt c for the wild-type enzyme was also dependent upon the anion, being 34.6 M at 749 mM ionic strength with KCl (Table V) (Table IV). This anion dependence was not observed with E213Q and E214Q (Tables IV and V).
Cytochrome c 2ϩ was a competitive inhibitor versus cytochrome c 3ϩ . Cytochrome c 2ϩ K i for the wild-type enzyme was dependent upon ionic strength, with K i at 118 mM being 7 times greater than that at high (749 mM) ionic strength (Table V). A similar trend was noted for the E214Q mutant; however, cytochrome c 2ϩ K i for the E213Q mutant was similar to wild type at high ionic strength but did not increase with decreasing ionic strength.
The ionic strength dependence of many electron transfer reactions has been proposed to be a consequence of complementary electrostatic interactions necessary for formation and/or stabilization of productive electron transfer complexes (19 -21, 43). This ionic strength dependence is also observed for the reaction of P450R with cytochrome c. Fig. 3 shows that k cat / K m Cyt c for the wild-type enzyme decreases as ionic strength increases. The E214Q k cat /K m Cyt c shows a similar ionic strength dependence, but is shifted to lower ionic strengths, consistent with removal of a weak electrostatic interaction. The E213Q mutant displays a different ionic strength dependence, probably because this substitution alters interactions with both oxidized and reduced cytochrome c.
A number of theoretical models have been developed to describe the effects of ionic strength on biomolecular rate constants when electrostatic interactions are important in formation of the electron transfer complex. Models such as that of Wherland and Gray (44) treat the interacting species as small and uniformly reactive, while Watkins et al. (45) have developed a model incorporating asymmetric protein surface charge distributions and localization of charges at the site of electron transfer.
Nonlinear least-squares fits of the ionic strength dependence of k cat /K m Cyt c to the Watkins model are shown in Fig. 3 for the wild-type, E213Q, and E214Q proteins. The wild-type and E214Q proteins fit the model quite well, yielding 5.8 and 20.2 Å, respectively, for the radius of the interaction domain. These values compare favorably with radii of 7-20 Å obtained for other electrostatically stabilized electron transfer complexes (44). Similar fits and radii for these two proteins were obtained with the Wherland-Gray model (not shown). These results are consistent with the presence of electrostatic interactions in the transition state which are weakened by removal of Glu 214 . The ionic strength dependence for E213Q, however, did not fit these simple electrostatic models, giving poor fits to both the Watkins and Wherland-Gray equations. Addition of dipole terms to the Watkins model improved the fit (Fig. 3); however, the interaction radius remained unreasonably small (1.4 Å).
The kinetic constants shown in Table V and Figs. 2 and 3 can be used to obtain the changes in the free energies of binding for wild-type, E213Q, and E214Q reaction intermediates when ionic strength is varied from 118 to 749 mM (46,47). Fig. 4 shows the effects of ionic strength on the relative free energies of binding for the reductase-cytochrome c 3ϩ ground state, reductase-cytochrome c transition state, and reductase-cytochrome c 2ϩ complexes of the three enzymes, as well as on the activation energy. For all three enzymes, increasing the ionic strength from 118 to 749 mM increased the free energies of binding of both the ground state and transition state complexes. For the wild-type and E214Q proteins, the ground state complexes were destabilized more than the transition state, producing, respectively, 209 and 354 cal/mol decreases in the activation energies and leading to the observed increases in k cat at high ionic strength. The opposite case holds for the E213Q enzyme, where destabilization of the ground state complex was less than that of the transition state by 509 cal/mol, accounting for the observed decrease in k cat at high ionic strength.
Increasing the ionic strength from 118 to 749 mM stabilized the wild-type and E214Q reductase-cytochrome c 2ϩ complexes by nearly 1 kcal/mol but had no effect on the E213Q reductasecytochrome c 2ϩ complex (Fig. 4). At low ionic strength, the E213Q reductase-cytochrome c 2ϩ complex was stabilized by 1.2 kcal/mol relative to wild type. At high ionic strength, however, this substitution had no effect on binding of cytochrome c 2ϩ . DISCUSSION The current study examines the role of two clusters of acidic amino acids, located in the amino-terminal half of P450R, in substrate recognition. The results indicate that the sites and mechanisms of interaction are distinct for the two substrates, as well as for the oxidized and reduced forms of cytochrome c. Asp 208 of cluster I affects P-450 catalyzed N-demethylation while Glu 213 of cluster II influences cytochrome c reduction. Although this region is adjacent to the FMN binding domain (8,(12)(13)(14), none of these substitutions, with the exception of the D207N/D208N/D209N/E213Q/E214Q/D215N substitution, affected FMN content. Ferricyanide reductase activity was also unchanged, suggesting that the observed effects are not due to conformational changes in the vicinity of the FAD/NADPH domain.
Only Asp 208 located in cluster I interacts measurably with cytochrome P450. Importantly, mutagenesis of adjacent residues (Glu 207 and Glu 209 ) did not alter any enzymatic properties. Furthermore, activity of the double mutant, D207N/ D208N, was similar to that of D208N, suggesting a specific interaction with Asp 208 . Since K m P450 was unchanged and k cat and k cat /K m P450 were reduced 54 and 56%, respectively, Asp 208 appears to be involved in interactions affecting electron transfer rather than cytochrome P450 binding, either directly or indirectly by orienting the P450R cytochrome P450 complex for optimum electron transfer (48,49).
In contrast, interaction of P450R with cytochrome c involves primarily the second acidic cluster (residues 213-215). The largest changes in cytochrome c reductase activity were seen with the E213Q mutant, where catalytic activity was reduced 59% and K m Cyt c was reduced 47%. Conversion of Asp 214 to its corresponding amide affected only the ionic strength depend-ence of k cat and K m Cyt c . The D215N substitution had no effect on catalytic activity. Properties of the double and triple mutants were similar to those of the corresponding single mutants, suggesting that each residue interacts independently with cytochrome c.
The ionic strength dependence of wild-type and E214Q k cat / K m Cyt c values fits well for models incorporating electrostatic interactions and is consistent with removal of a weak electrostatic interaction in the E214Q mutant (Fig. 3). In contrast, the ionic strength dependence of E213Q does not fit these models without the introduction of additional terms, either as a result of an unfavorable conformation of the electron transfer complex or as a consequence of altered interactions with reduced cytochrome c.
Comparison of the effects of ionic strength on the kinetic properties of the wild-type, E213Q, and E214Q enzymes permits the identification of the reaction intermediates affected by changes in salt concentration and by removal of either the Glu 213 or Glu 214 side chains (Fig. 4). For the wild-type enzyme, increasing ionic strength increased the free energy of binding for both the ground state and transition state complexes, consistent with the presence of electrostatic interactions between P450R and cytochrome c. High ionic strength increased the binding energy of the ground state complex more than that of the transition state, leading to a net decrease in activation energy and the observed increase in k cat . This is an example of the use of the free energy of binding of the enzyme-substrate  complex to increase the rate of catalysis (46). It is likely that this mechanism also underlies the inhibition of cytochrome c reduction by polyols observed by Voznesensky and Schenkman (32,33). Similarly, removal of a weak interaction between Glu 214 and cytochrome c has a larger destabilizing effect on the P450R-cytochrome c ground state complex than on the transition state complex, leading to a net increase in k cat at high ionic strength.
Replacement of Glu 213 with glutamine had no effect at low ionic strength; however, at higher ionic strengths, the ground state complex was more stable than that of the wild-type enzyme, leading to a net increase in activation energy and decrease in k cat . Glu 213 in the wild-type enzyme may interact weakly or not at all with cytochrome c; however, glutamine at position 213 must be able to form hydrogen bonds with cytochrome c which act to stabilize the P450R-cytochrome c 3ϩ complex in a conformation which is less favorable for electron transfer.
Although Glu 213 does not interact strongly with oxidized cytochrome c, cytochrome c 2ϩ K i of the wild-type enzyme decreased with increasing ionic strength, consistent with the presence of an electrostatic repulsion between P450R and cytochrome c 2ϩ (Table V). The E213Q mutant bound cytochrome c 2ϩ tightly at both high and low ionic strength as a result of removal of this repulsive interaction. Based on K i Cyt c 2ϩ values (Table V), the E213Q-cytochrome c 2ϩ complex was stabilized by 965 cal/mol relative to wild type at 118 mM ionic strength, close to the theoretical 1 kcal/mol destabilization of the flavodoxincytochrome c complex resulting from loss of a positive charge upon reduction of cytochrome c (50).
The free energy changes associated with the D208N, E213Q, and E214Q substitutions range from nil to 1.3 kcal/mol, substantially less than those commonly associated with loss of an ionic bond. This may be because these substitutions do not affect rate-limiting steps in benzphetamine demethylation or cytochrome c reduction or because an electrostatic interaction has been replaced with a hydrogen bonding interaction (47), as has been proposed for binding of human adrenal ferredoxin to P450 scc (51). Free energy changes of this magnitude are also to be expected if multiple electrostatic, hydrophobic, and van der Waals interactions contribute to binding. Mutagenesis of cationic residues of P450 cam (23) and cytochrome b 5 (24) which are involved in charge-pairing interactions with putidaredoxin and cytochrome b 5 reductase, respectively, produced binding energy changes similar to those observed here, and it was proposed that a single electrostatic interaction made a relatively small contribution to overall binding. The altered ionic strength dependence of the E213Q mutant may be an indication of an increased contribution from these other interactions.
In summary, these results demonstrate that residues located in two acidic clusters identified by chemical cross-linking experiments have specific effects on the interaction of P450R with cytochrome c and cytochrome P450. Cytochrome c and cytochrome P450 interact with separate residues in this region. Asp 208 of P450R influences electron transfer to cytochrome P450 with no effect on binding. Although additional factors are necessary for electron transfer to cytochrome P450 (31,(32)(33)(34)52), the ionic strength data indicate that electrostatic interactions do predominate in formation of the P450R-cytochrome c complex (Fig. 3) (6, 52). Glu 214 interacts weakly with cytochrome c and Glu 213 is involved in a repulsive interaction with reduced cytochrome c. Glu 213 may not form contacts with oxidized cytochrome c; however, replacement of Glu 213 with glutamine may introduce new hydrogen bonding interactions which stabilize the reductase-cytochrome c complex in an altered conformation where electron transfer is less efficient. The effects of ionic strength on the kinetic properties of these mutants point toward multiple charge-pairing, hydrophobic, and repulsive interactions involving these residues and other, as yet unidentified residues, which must also play a role in electron transfer from P450R to its substrates.