Probing the Role of Lysines and Arginines in the Catalytic Function of Cytochrome P450d by Site-directed Mutagenesis INTERACTION WITH NADPH-CYTOCHROME P450 REDUCTASE*

To identify amino acids of cytochrome P450d (P450d) which participate in the interaction with NADPH-cy-tochrome P450 reductase, we changed conserved ionic amino acids of P450d to others by site-directed mutagenesis. Turnover numbers (0.032-0.008 min") of purified mutants Lyss4-Glu, Lysss-Glu, Lysl''-G1u, Ly~~~'-Glu, Ly~~'~-Glu, Arg4"-G1u, and Ly~~'~-Glu toward 7-ethoxycoumarin were much lower than that (0.380 min") of the wild type at 25 "C. Reduction rates (less than 0.054 s-') of the heme of all mutants (0.1 WM) in the presence of NADPH and the reductase (0.3 WM) were much lower than that (5.9 s-') of the wild type. Furthermore, a turnover number (0.042 min-') of a microsomal triple mutant (Arg13'-Leu + Arg13'-Leu + Arg'37-Le~) of a conserved under these conditions. Buffer solutions containing sonicated DLPC were the same as those used for obtaining activities. For sample solutions a 1-cm cell was used, while two 5-mm cells placed in tandem were used for reference solutions to each cell P450d and the reductase being added separatedly. Experiments were repeated at least three times and their averaged values are described. Experimental errors were less than 20% except for the mutant Ly~'~~-Glu in which the correct value was not estimated due to very small spectral changes. Rate constants were expressed by s-' and were obtained by monitoring the Soret spectral peak at 447 nm of the CO-reduced forms of P450d at 25 "C. Rate constants were estimated from the fast phase. Since the observed kinetic tracing was the sum of the absorbance changes produced by reduction of both P450d and the reductase, the absorption contribution from P450d was determined by subtracting the spectrum produced by the reductase from that produced by the complete system. Reduction was initiated by adding 40 pl of the NADPH (1 mM) -DLPC (32 pM)-CO (approximately 0.8 mM) solution to 40 pl of the P450d (0.2 pM)-reductase (0.6 ~M)-DLPC (32 pM)-Co (approximately 0.8 mM) solution. The stopped flow spectrometer with micro mixing cells has a dead time, 4-4.5 ms, a cell path, 10 mm and minimal volume for one shot, 30-40 pl. Both buffer solutions were the same as used for measuring activities. Opaque solution containing DLPC was sonicated until the solution became transparent, and then the enzymes were added to the solution. Before the reduction was started, 30 mM glucose, 50 units/ml glucose oxidase, and 1,000 unitslmi catalase were added to both the P450d and NADPH solutions to eliminate oxygen gas and HzOz solved in the solutions. The DLPC solutions were always kept more than 2 h at 25 "C to reach equilibrium after the enzymes were added (3). Experiments were repeated at least three times, and their averaged values are described. Experimental errors were less than 10%.

NADPH-cytochrome P450 reductase is a flavoprotein and works as only one electron mediator to microsomal P450s from NADPH. Taking in consideration the above mentioned findings, it seemed very likely that ionic amino acids such as Lys and Arg on the protein surface of microsomal P450s directly interact with the reductase. Thus we made thirteen mutants of conserved Lys and Arg of P450d in the present study. Mutated positions of Lys and Arg of P450d and corresponding sequences of other P450s are shown in Fig. 1 (22,23). Arg4*'-Leu and L y~~~~-G l u mutants were obtained in our previous study (24). We obtained turnover numbers of mutants, dissociation constants of the reductase from mutants, and rates of the reduction of the heme of mutants by NADPH in the presence of the excess reductase. From those findings, we strongly suggest that seven amino acids out of those mutated Lys and Arg residues of P450d are very important in the ionic interaction with the reductase and/or in orientating the best geometry of the two proteins for the electron transfer.

EXPERIMENTAL PROCEDURES
Expression of P450d mutants in yeast was done as described previously (24)(25)(26)(27)(28). Site-directed mutageneses were done by using an in vitro mutagenesis kit of Amersham (United Kingdom) as previously described (24,26). Mutation was confirmed by determination of nucleotide sequences by a SequenaseTM DNA sequencing kit of U. S. Biochemical Corp. We strictly checked every mutant for whether a mutation(s) at another position(s) had occurred.
Preparations of yeast microsomes and purification of P450d mutants were done as previously described (24)(25)(26)(27)(28).* All mutants were purified as a high spin form, and most of them were stable both for oxidized and reduced forms. It should be noted, however, that the Ly~"'~-Glu and LyP3-G1u mutants were unstable at 37 "C in terms of the Soret absorption spectrum of the CO-reduced form. It was confirmed spectrometrically that these mutants are stable at 25 "C. Catalytic activities of microsomal and purified mutants were obtained as previously described (24,28). Concentration of P450d mutants was obtained from molar absorptivity, 1.09 X lo5 M" cm" at 393 nm of the high spin oxidized form or 1.20 X lo5 M" cm" at 447 nm of the CO-reduced form? Concentration of the reductase was determined spectrometrically as previously described (6). Dissociation constants ( K d ) of the reductase from the mutants were obtained from difference absorption spectra of the Soret region on a Jasco Uvidec-510 recording spectrometer.* Rate constants of the reduction of the heme of the mutants were obtained on a stopped flow spectrometer (Union Rapid Reaction Analyzer RA-601) equipped with micromixing cells and directly connected with a NEC PC-9801VM personal computer. Table I shows turnover numbers of microsomal ionic P450d mutants toward 7-ethoxycoumarin obtained at 37 "C. Turnover numbers (0.045-0.110 min") of Lys mutants at 94, 99, and 105 in the amino-terminal region were much lower than that of the wild type. Turnover numbers (0.223-0.620 rnin") of single mutants of the "Arg cluster," Arg'35-Arg'36-Arg'37, were not much lower than that (0.674 min-') of the wild type, while that (0.042 min") of the triple mutant of the Arg cluster was much lower than that of the wild type. Both A r P -L e u and Ar$75-Trp mutants did not bind the heme in the active site of this enzyme in terms of optical absorption spectros-  Boldface letters of P450d are amino acids mutated in the present study. The center column of the lower part is a sequence alignment modified by us.

RESULTS
copy. Mutant Lys4"-G1u had sufficient catalytic activities. M u t a n t s L y~~~" -G l u , L y~~~~-G l u , Arg455-Glu, and L y~~~~-G l u had much lower activities (0.013-0.040 rnin") than that of the wild type.
To test the role of important ionic amino acids in the catalytic function, we further purified Lysg4-Glu, LyP-Glu, L y~' '~-G l u , L y~~~' -G l u , L y~~~~-G l u , Arg4@-Glu, and L y~~~~-G l u mutants. All mutants were purified as the high spin form (27).* The triple mutant Arg'35-Le~ + Arg'36-Le~ + Arg'37-Leu could not be purified because the heme dissociated from the protein during the purification procedure. Activities of all purified mutants toward 7-ethoxycoumarin, which were obtained a t 25 "C to avoid denaturing some mutants (cf. "Experimental Procedures"), were much lower than that of the wild type (Table I).
Dissociation constants ( Kd) of the reductase from the mutants were obtained by the Soret absorption spectral change.
To test whether electrons transfer to the heme of P450d through the reductase, we observed the Soret peak at 447 nm of the reduced P45od-co form in the presence of NADPH and triple amounts of the reductase. The peak at 447 nm of the wild type-reductase-CO solution quickly appeared after adding NADPH, while those of all mutants slowly appeared after more than 1 min. The rate of the appearance of the peak at 447 nm, which corresponds to the rate of the reduction of the heme, was estimated with a stopped flow spectrometer. Rate constants (<0.001-0.054 s-') of the reduction of the heme of the mutants by NADPH in the presence of triple amounts of the reductase were much lower than that (5.9 s-') of the wild type.

DISCUSSION
Nelson and Strobe1 (22) suggested that there are several ionic domains on the surface of P450s. Stayton et al. (17) suggested from the molecular model of P450,,, and cytochrome b5 that basic amino acids such as Arg7', Arg"', Lys344, and Ar$'j4 of P45OC,, will be involved in the interaction with putidaredoxin. As observed in Fig. 1 (22), Arg7' of P450,,, corresponds to Lysg4 of P450d. P450d has two other Lys residues, Lysg9 and LYS"~, in this region (Fig. 1). It seems that this region may be one of the ionic domains of membranebound P450s (22). Thus, we reversed the ionic character by changing these amino acids to Glu.
Since it seemed that the Arg'35-Glu type mutant may form self-ionic bond(s) on the protein surface of P450d, we purposely changed Arg to a nonpolar amino acid Leu for the mutants of this region.
It was reported that A r p of P45oc21 (number for P450d) is essential for steroid 21 hydroxylation, since Trp and Leu mutants of this Arg have no activities (29). However, this Arg is conserved only for P45Ol7, and P450cz1 ( Fig. 1) (22). One notices that a highly conserved Arg is located at the upper site toward the amino terminus by two amino acid residues, which corresponds to Arg375 for P450d or to A r e 1 for P450,,, (Fig. 1) (22). We thus replaced this conserved Arg with Trp and Leu. Turnover numbers were expressed by nmol min-' (nmol P450)". Turnover numbers of microsomal mutants were obtained at 37 "C (24), while those of purified mutants in the presence of P450 reductase were obtained at 25 "C in the presence of DLPC because some purified mutants such as Lysl"'-Glu and L y~~~~-G l u were not stable at 37 "C (cf. "Experimental Procedures"). Concentrations used for activities were 0.2 p~ P-450d, 500 p~ 7ethoxycoumarin, 0.6 p~ reductase, and 1 mM NADPH. By titrating the reductase, this concentration of the reductase was sufficient to have full activity of the wild type P450d, which was comparable to that previously reported (24,311. For catalytic activities of purified mutants, 5 mg of DLPC/I ml of buffer solution was sonicated until the opaque solution was changed to transparency. We always kept the DLPC solution for more than 2 h at 25 "C after we added the enzymes to the DLPC solution to ensure equilibrium (7). Buffer solutions consisted of 20 p M EDTA, 20 p M dithiothreitol, 4% glycerol, and 0.1 M potassium phosphate (pH 7.2). Experiments were repeated at least three times, and their averaged values are described. Experimental errors were less than 20% and less than 10% for the microsomal solutions and for the reconstituted solutions, respectively.
'Apparent dissociation constants (&) of the reductase were expressed by nM and were obtained from the Soret spectral change caused by adding the reductase to the P450d-acetanilide solution at 25 "C.2 The spin state of the high spin type P450d was converted partially to the low spin state by adding 20-30 mM acetanilide.' The low spin portion of the P450d-aCetanilide solution was changed back to the high spin state with two isosbestic points around 400 nm and around 425-460 nm by adding the reductase. Double reciprocal plots of the spectral change versus the concentration of the free reductase formed a straight line, indicating that a 1:l redUCtaSe-P450d complex is formed? Thus, we used these Soret spectral changes accompanied with the spin change for obtaining K d values. We always kept the P450d-reductase solution for more than 40 min at 25 "C in each titrating step to reach equilibrium (3).
Estimation of Kd less than 28 nM was not feasible because the concentration of free reductase could not be obtained under these conditions. Buffer solutions containing sonicated DLPC were the same as those used for obtaining activities. For sample solutions a 1-cm cell was used, while two 5-mm cells placed in tandem were used for reference solutions to each cell P450d and the reductase being added separatedly. Experiments were repeated at least three times and their averaged values are described. Experimental errors were less than 20% except for the mutant L y~'~~-G l u in which the correct value was not estimated due to very small spectral changes. Rate constants were expressed by s-' and were obtained by monitoring the Soret spectral peak at 447 nm of the CO-reduced forms of P450d at 25 "C. Rate constants were estimated from the fast phase. Since the observed kinetic tracing was the sum of the absorbance changes produced by reduction of both P450d and the reductase, the absorption contribution from P450d was determined by subtracting the spectrum produced by the reductase from that produced by the complete system. Reduction was initiated by adding 40 pl of the NADPH (1 mM) -DLPC (32 pM)-CO (approximately 0.8 mM) solution to 40 pl of the P450d (0.2 pM)-reductase (0.6 ~M ) -D L P C (32 pM)-Co (approximately 0.8 mM) solution. The stopped flow spectrometer with micro mixing cells has a dead time, 4-4.5 ms, a cell path, 10 mm and minimal volume for one shot, 30-40 pl. Both buffer solutions were the same as used for measuring activities. Opaque solution containing DLPC was sonicated until the solution became transparent, and then the enzymes were added to the solution. Before the reduction was started, 30 mM glucose, 50 units/ml glucose oxidase, and 1,000 unitslmi catalase were added to both the P450d and NADPH solutions to eliminate oxygen gas and HzOz solved in the solutions. The DLPC solutions were always kept more than 2 h at 25 "C to reach equilibrium after the enzymes were added (3). Experiments were repeated at least three times, and their averaged values are described. Experimental errors were less than 10%.

Ref. 24.
e Heme did not bind to the mutant protein.
Lys' ' ' is highly conserved for all P450s (Fig. 1) (22). It was ever, when one modifies amino acid alignment of this region claimed that this Lys is involved in the interaction with the as the center column of the lower part in Fig. 1, several basic  reductase (13, 15). Thus we made the Lys401-G1u mutant.
amino acids such as Arg or Lys are modestly conserved in Arg4" of P450d nearly corresponds to Arg342 of P450,,, (Fig. this region. Lys344 of P450,., perhaps corresponds to Lys440 1, left column of lower part). The heme did not bind to the (Fig. 1). We changed Lys440 and Arg455 to Glu. A mutant mutant Arg429-Le~ protein in our previous study (24). How-L y~~~~-G l u was already made in our previous study (24).
Arp364 of P450,,, corresponds to Lys463 of P450d (Fig. 1)  (22). The basic amino acid of this region is well conserved as Arg or Lys (Fig. 1) (22), leading to the L y~~~~-G l u mutant. We selected 7-ethoxycoumarin as a substrate because catalytic activity of P450 toward this substrate is less influenced by change(s) of the ternary structure of the substrate binding site caused by mutations (24, 28). Thus a marked decrease in activity toward this substrate caused by mutations (Table I) will be associated with intrinsic and/or essential function for catalytic activity of P450d. Thus, one may rule out the possibility that a structural change of the substrate binding site caused by the mutation is a reason for the decrease in activity of microsomal and purified mutants.
Since the heme did not bind to the and Arg4" mutants, these amino acids may be so important to retain appropriate tertiary protein structure of P450d as to hold the heme. Similarly the conserved Lys4" does not appear to be directly involved in the essential function of P450d.
Affinities of the reductase to the Lyslo5-G1u, Arg455-Glu, and Lys4'j3-G1u mutants were comparable to or higher than that to the wild type (Table I), even if ionic characters of these amino acids were reversed. Thus decreased catalytic activities and rate constants of the heme reduction of these mutants are not attributed to the dissociation of the two proteins. It seems likely thus that LysIo5, Arg455, and Lys463 are important to orient appropriate geometry of the whole molecule of the two proteins and/or interfacial surface amino acids of the two proteins for efficient intermolecular electron transfer.
On the other hand, by reversing the ionic character of Lysg4, Lysg9, L Y S~~" , and Lys453 of P450d, affinities of the reductase to the mutants decreased (Table I). Thus it appears that these amino acids may play a role in anchoring the reductase to P450d in addition to orienting proper geometry of the proteins for electron transfer. However, we cannot conclusively state that these amino acids are a prerequisite for anchoring the reductase since the Kd values of these mutants are not remarkably different from that of the wild type and the binding of the reductase may induce only a small Soret spectral change of these mutants.
Although it was not feasible to study in detail on the triple mutant of the Arg cluster, it appears that the Arg cluster may be involved in the essential function of P450d, perhaps in the interaction with the reductase.
Arg7', Arg"', and Ar$64 of P450,., are located in the B, C, and L helices, respectively, and reside on the same proximal surface of the protein molecule (17,30). Correspondingly L Y S~~, Lysg9, and Lyslo5 of P450d (Fig. 1) must be located at the B helix, assuming that the tertiary structure of P450d is the same as that of P450,,,. Similarly the Arg cluster and Lys-463 of P450d (Fig. 1) must be situated at the C and L helices, respectively, on the same assumption. Examination of the molecular model generated by replacing amino acid side chains in the crystal structure of P450,., suggests that Lys453, Arg455, and perhaps Lys440 are located on the proximal protein surface (24). Therefore, it is implied that these im-portant basic residues and their neighboring residues of P450d for the interaction with the reductase must be located on the same surface of the P450d molecule if we assume that the tertiary structure of P450d is the same as that of P450,,,.

CONCLUSIONS
Ionic amino acids Lysg4, Lysg9, LysIo5, LYS~~', Lys453, Arg455, and Lys463 and perhaps the Arg cluster Arg'35-Arg'36-Arg'37 appear very important for the catalytic function of P450d, probably by participating cooperatively in forming an electron transfer complex with the reductase. Ionic regions of the surface of the P450d molecule which directly interact with the reductase may be similar, if not identical, to those of P450,,, (30) which directly interact with putidaredoxin.