Molecular engineering study on electron transfer from NADPH-P450 reductase to rat mitochondrial P450c27 in yeast microsomes.

We have reported the localization on yeast microsomes for a modified P450c27 (mic-P450c27) that contains the microsomal targeting signal of bovine P450c17 in front of the mature form of rat mitochondrial P450c27 (Sakaki, T., Akiyoshi-Shibata, M., Yabusaki, Y., and Ohkawa, H. (1992) J. Biol. Chem. 267, 16497-16502). In this study, we found that mic-P450c27 could be reduced by NADPH in the yeast microsomes without supplement of its physiological redox partners, adrenodoxin and NADPH-adrenodoxin reductase. In order to elucidate the direct electron transfer from NADPH-P450 reductase to mic-P450c27, we carried out simultaneous expression of mic-P450c27 and yeast P450 reductase. The reduction rate of mic-P450c27 was increased by overproduction of yeast P450 reductase, roughly in proportion to the reductase content in the microsomes. In addition, we constructed a fused enzyme between mic-P450c27 and yeast P450 reductase. The reduction rate of heme iron in the fused enzyme was too rapid to be measured. These recombinant yeast microsomes showed a notable 27-hydroxylation activity toward 5beta-cholestane-3alpha,7alpha, 12alpha-triol in the absence of adrenodoxin and adrenodoxin reductase. Finally, we purified mic-P450c27 from the recombinant yeast microsomes and reconstituted the hydroxylation system in liposomal membranes using the purified mic-P450c27 and yeast NADPH-P450 reductase. Mic-P450c27 was reduced by NADPH and showed its monooxygenase activity on the reconstituted system. Therefore, yeast NADPH-P450 reductase alone was found to transfer two electrons from NADPH to mic-P450c27. These results clearly show that mic-P450c27 not only localizes on the microsomes but also functions as a microsomal cytochrome P450 that accepts electrons from NADPH-P450 reductase.

Cytochrome P450 represents a large group of structurally related hemoproteins that catalyze the monooxygenase reactions of a wide variety of both endogenous and exogenous compounds. P450 monooxygenases can be classified into two types based on the electron supplying proteins to P450. In class I P450s (1), which include most of bacterial soluble P450s and mitochondrial P450s, the electron transport chain involves an FAD-containing enzyme, ferredoxin reductase, and an ironsulfur enzyme, ferredoxin, to transfer electrons from NADH or NADPH to the terminal P450. In class II P450s, which include microsomal P450s and bacterial P450 BM-3 (2), electrons are transferred from NADPH to P450 by the catalysis of NADPH-P450 reductase that contains one molecule each of FAD and FMN. Although detailed studies have been carried out for the interaction between the P450s and their partner proteins in the both electron transfer systems (1,3,4), the interaction between mitochondrial P450s and the NADPH-P450 reductase or between microsomal P450s and the ferredoxin has not been well demonstrated yet. Recently, Jenkins and Waterman (5) reported a very interesting finding for the electron transfer system to heterologously expressed P450c17 in Escherichia coli. They purified both an FMN-containing flavodoxin and an FADcontaining NADPH-flavodoxin reductase from E. coli and found that these enzymes could support 17␣-hydroxylation activity of bovine adrenal microsomal P450c17, which explains well why P450c17 expressed in E. coli is functionally active without P450 reductase (6).
From an evolutionary standpoint, microsomal P450 is thought to have diverged from a bacterial soluble P450, followed by the appearance of mitochondrial P450 from microsomal P450 (7). Alteration of soluble P450 to microsomal one could be explained by the addition of an amino-terminal hydrophobic sequence to the soluble P450 and that of microsomal P450 to mitochondrial one could be attributed to the substitution of the signal sequence for the localization into the mitochondrial inner membranes. Indeed, we have succeeded in conversion of subcellular localization of P450c27 from yeast mitochondria to microsomes only by altering its amino-terminal targeting signal (8). This microsomally modified P450c27 (referred to as mic-P450c27) was active in both 25-hydroxylation of 1␣-hydroxyvitamin D 3 and 27-hydroxylation of cholestane-3␣,7␣,12␣-triol (THC) 1 when reconstituted in vivo and in vitro with adrenodoxin (ADX) and NADPH-adrenodoxin reductase (ADR) (8). However, no possible schemes have been suggested for the evolution or alteration of the electron transfer partners to P450.
In the present study, we examined by the use of molecular engineering methods whether mic-P450c27 could accept electrons efficiently from yeast NADPH-P450 reductase. In addition, we have constructed a highly active fused enzyme between mic-P450c27 and yeast P450 reductase. These findings clearly indicated that yeast P450 reductase can transfer electrons from NADPH directly to mic-P450c27. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Present address: Sumitomo Pharmaceuticals Research Center, Kasugadenaka, Konohana-ku, Osaka 554, Japan.
Recombinant DNA Procedures-Recombinant DNA procedures and transformation of E. coli and S. cerevisiae cells were performed as reported previously (9). Sequencing of the synthesized linker DNAs and the junction region of the fused enzyme cDNA was performed as described previously (11). Recombinant E. coli and S. cerevisiae strains were cultivated in L broth and concentrated SD medium, respectively, as described (12). Polymerase chain reaction was carried out with Pfu DNA polymerase (Stratagene Cloning Systems, CA) by using Perkin Elmer PC2000 apparatus.
Measurement of P450 Hemoprotein-P450 hemoprotein content in whole cells or subcellular fractions was determined by a reduced CO difference spectrum (9) using an extinction coefficient of 91 mM Ϫ1 cm Ϫ1 (13).
Western Blot Analysis-Total cellular proteins from the recombinant yeast strain were subjected to an electrophoresis on SDS-polyacrylamide gel and then transferred electrophoretically from the gel to a nitrocellulose filter (9). The filter was probed with anti-mic-P450c27 antiserum (8) or anti-yeast NADPH-P450 reductase Ig (14), followed by 125 I-labeled Protein A as described previously (9). Purification of Mic-P450c27 and Reconstitution of the Hydroxylation System in Liposomal Membranes-Microsomal fraction was prepared from the recombinant yeast strains as described previously (9). Mic-P450c27 was purified from AH22/pAMS25 microsomes to an electrophoretically homogeneous form by the same procedures as reported (8). The final preparation (12 nmol of P450/mg of protein) was incorporated into liposomal membranes, and the hydroxylation system was reconstituted in vitro with yeast NADPH-P450 reductase (specific activity of 150 units/mg protein, provided by Dr. Y. Aoyama, Soka University) (15). The molar ratio of mic-P450, P450 reductase, and phospholipid was 1:1:3000.
THC 27-Hydroxylation Activity-The monooxygenase activity of the microsomal fraction was measured as follows. The reaction mixture contained each of the microsomal fractions prepared from the recombinant yeast strains (0.025-0.1 M P450), 100 mM Tris-HCl (pH 7.8), 0.5 mM EDTA, and [ 3 H]THC (20, 50, 100, or 200 M). The reaction was initiated by addition of NADPH to a final concentration of 0.5 mM. After incubation at 37°C for 3 or 5 min, the reaction was stopped by vigorous mixing with CH 2 Cl 2 . The organic phase was recovered and dried up. The resulting residue was solubilized with acetonitrile for high performance liquid chromatography analysis with a linear gradient of 50 -100% acetonitrile aqueous solution at a flow rate of 1.0 ml/min at 50°C (8). [ 3 H]THC and its hydroxylated metabolite were analyzed by high performance liquid chromatography with monitoring the radioactivity on a Flow-one ␤ system (Packard). The THC 27-hydroxylation activity on the reconstituted liposomal membrane was similarly measured. The cellular THC 27-hydroxylation activity of the recombinant yeast was measured by incubation of the cell culture with 10 M [ 3 H]THC at 30°C as described previously (8).
Reduction of Heme Iron-Stopped flow analysis was performed at 10°C using a dual-wavelength stopped-flow device (Unisoku Co., Ltd., Hirakata, Japan) as described previously (16,17). The reaction mixture contained the microsomal fraction (0.2 M P450 or its fused enzyme), 20 mM potassium phosphate (pH 7.4), 20% glycerol, 50 mM glucose, 20 units/ml glucose oxidase, 200 units/ml catalase, and 100 M THC. The reaction was initiated by rapid mixing of the microsomes with an equal volume of 80 M NADPH in the presence of CO, and the change in the absorbance difference between 450 and 490 nm was recorded. A computer program was employed to fit the kinetic data as described (17).
Other Methods-Cytochrome c reductase activity derived from NADPH-P450 reductase or the fused enzyme in the microsomal fractions was measured as described previously (14,18). Protein concentration was determined by the method of Lowry et al. (19), using bovine serum albumin as a standard.

Construction of Yeast Expression Plasmids-
The expression plasmid pAMS25R for simultaneous expression of mic-P450c27 and yeast NADPH-P450 reductase was constructed by inserting the NotI fragment of pAMS25 containing the expression unit for mic-P450c27 into the unique NotI site of the plasmid pARR3N (10). The plasmid pF250 for the fused enzyme between mic-P450c27 and yeast P450 reductase was constructed by adapting the procedure for the construction of pAFCR1 (17). A BamHI-XhoI fragment (250 base pairs) encoding the carboxyl-terminal portion of mic-P450c27 was amplified by polymerase chain reaction. The resulting BamHI-XhoI fragment was doubly inserted with the HindIII-BamHI fragment prepared from pAMS25 between the HindIII and XhoI sites of pBluescript II. The plasmid pGYR encoding yeast NADPH-P450 reductase (20) was digested with PvuII to insert a linker containing an XhoI site. Both HindIII-XhoI fragments from the resulting plasmids that encode mic-P450c27 and P450 reductase, respectively, were doubly inserted into the unique HindIII site of the expression vector pAAH5N in the correct orientation. Fig. 1 schematically represents the three expression plasmids used in the present study, pAMS25 for mic-P450c27, pAMS25R for both mic-P450c27 and yeast P450 reductase, and pF250 for the fused enzyme between mic-P450c27 and yeast P450 reductase. The individual expression plasmids were introduced into S. cerevisiae AH22 cells to obtain the corresponding recombinant yeast strains. Fig. 2 shows Western blot analysis of whole cellular proteins prepared from the recombinant yeast strains. A clear band reactive with anti-mic-P450c27 antiserum was detected in both AH22/pAMS25 and AH22/pAMS25R at the position of apparent molecular mass of about 55 kDa (Fig. 2, left, lanes 2 and 3). AH22/pF250 shows a protein band reactive with both anti-mic-P450c27 antiserum and anti-P450 reductase Ig (Fig. 2, left,  lane 4 and right, lane 2). The apparent molecular mass of the protein was about 130 kDa, which agrees with the structure of the fused enzyme as deduced from its cDNA sequence. The expression levels of mic-P450c27 in AH22/pAMS25 and AH22/ pAMS25R and the fused enzyme in AH22/pF250 were estimated to be about 2.0, 1.6, and 1.1 ϫ 10 5 molecules/cell, respectively, on the basis of reduced CO difference spectra of the recombinant yeast cells.

Expression of Mic-P450c27 and Its Fused Enzyme in Yeast-
P450 and NADPH-P450 Reductase Contents in the Microso- mal Fraction Prepared from the Recombinant Yeast Cells-Mic-P450c27 expressed in yeast was localized in the microsomal fraction as reported previously (8). The fused enzyme exhibited a subcellular distribution quite similar to mic-P450c27. Thus, we determined the P450 and P450 reductase contents in the yeast microsomal fractions based on reduced CO difference spectra and NADPH-cytochrome c reductase activity, respectively (Table I). While the P450 content in AH22/pAMS25R microsomes was slightly less than in AH22/pAMS25 microsomes, the P450 reductase content in AH22/pAMS25R microsomes was about 24-fold higher than in AH22/pAMS25 microsomes. The ratio of mic-P450c27 to P450 reductase contents was roughly estimated to be 1:0.06 and 1:1.7 in AH22/pAMS25 and AH22/pAMS25R microsomes, respectively. The AH22/ pF250 microsomes exhibited the highest cytochrome c reductase activity, suggesting that the reductase portion of the fused enzyme can reduce cytochrome c more efficiently than P450 reductase itself. These results are consistent with those obtained for the fused enzyme between rat P4501A1 and yeast P450 reductase (17).
NADPH-dependent Heme Reduction in Mic-P450c27 and the Fused Enzyme- Fig. 3A shows the reduced CO difference spectra of AH22/pAMS25 microsomes in the presence of 100 M THC when reduced by addition of NADPH or sodium hydrosulfite. Most of mic-P450c27 in AH22/pAMS25 microsomes was reduced by NADPH within 2 min. Similarly, almost all of mic-P450c27 on the reconstituted membrane was reduced within 1 min (Fig. 3B). Thus, it seems most likely that mic-P450c27 was reduced by endogenous yeast NADPH-P450 reductase present in the microsomes.
The reduction rate of the heme iron by NADPH in the presence of substrate was also examined by stopped flow analysis. The rate constant k (s Ϫ1 ) in AH22/pAMS25 and AH22/ pAMS25R microsomes was estimated to be 0.040 and 0.67, respectively (Table II). Therefore, the reduction rate of the heme iron in mic-P450c27 was found to be roughly proportional to the yeast P450 reductase content in the microsomes. The reduction rate constant of mic-P450c27 was slightly larger than the reduction rate constant for rat microsomal P4501A1 similarly expressed in the yeast microsomes (17). In contrast, the reduction rate of the heme iron of the fused enzyme was too rapid to be measured, as was the case for the fused enzyme between rat P4501A1 and yeast P450 reductase (17). These results also strongly support the finding that yeast P450 reductase is an electron donor enzyme that can transfer electrons directly to mic-P450c27.
THC 27-Hydroxylation Activity-We reported previously that AH22/pRXMS25 strain containing mic-P450c27 and mature forms of bovine ADX and ADR converts THC into its 27-hydroxylated product 5␤-cholestane-3␣,7␣,12␣,27-tetrol (8). The substrate THC was also added to the cell culture of AH22/ pAMS25R and AH22/pF250 strains. After 14 h incubation, 5␤-cholestane-3␣,7␣,12␣,27-tetrol was detected in the culture supernatants of both strains (data not shown). Both mic-P450c27 and its fused enzyme were localized on the yeast endoplasmic reticulum membranes. Thus, the observed hydroxylation activity in the recombinant cells could depend on the electron transfer system on endoplasmic reticulum membranes and/or cytoplasm. The mitochondrial electron donor protein(s) could not be involved in the mic-P450c27-dependent activity in the cell culture of AH22/pAMS25R and AH22/pF250 strains.   AH22/pAMS25 0.040 Ϯ 0.008 AH22/pAMS25R 0.67 Ϯ 0.22 AH22/pF250 ND THC 27-hydroxylation activity in AH22/pAMS25, AH22/ pAMS25R, and AH22/pF250 microsomes was further examined without addition of ADR and ADX. AH22/pAMS25R and AH22/ pF250 microsomes showed a notable THC 27-hydroxylation activity, whereas AH22/pAMS25 microsomes showed little activity. The kinetic parameters, apparent K m (M) and V max (mol product/min/mol P450) values, were calculated by Lineweaver-Burk plots (Table III). The kinetic parameters for AH22/ pAMS25 microsomes could not be estimated because of its low activity. Mic-P450c27 and its fused enzyme had nearly the same K m values for THC, whereas the fused enzyme showed five times larger V max values than mic-P450c27. These results strongly suggest that the two electrons essential for the P450c27dependent THC hydroxylation are transferred from NADPH through NADPH-P450 reductase to mic-P450c27. The V max value of the fused enzyme (110 min Ϫ1 ) was about 2-fold higher than for mic-P450c27 in the AH22/pAMS25 microsomes in the presence of ADX and ADR (8). This high monooxygenase activity of the fused enzyme is likely due to a very efficient electron transfer from the reductase part to the P450 part of the fused enzyme. Efficient electron transfer has been observed in several fusion constructs between microsomal P450 and NADPH-P450 reductase (11,12,17,(21)(22)(23).

FIG. 3. Reduced CO difference spectra of the microsomal fraction prepared from AH22/pAMS25 cells (A) and mic-
The direct electron transfer from NADPH-P450 reductase to mic-P450c27 was definitely demonstrated in the membrane reconstituted system using the purified mic-P450c27 and yeast P450 reductase. The apparent K m value for THC of mic-P450c27 in the liposomes containing mic-P450c27 and the reductase (1:1 in molar ratio) was similar to that of AH22/ pAMS25R microsomes, although the V max value was about a third of that obtained in the microsomes that contain the reductase about 1.7 times more than mic-P450c27.

DISCUSSION
P450c27 was first purified from rat liver mitochondria by Okuda et al. (24), who demonstrated the requirement of ADX and ADR for its activity in a reconstituted system. For the interaction of P450s with their electron donor proteins, Bernhardt and Gunsalus (25) examined heterologous electron transfer from the bacterial electron donor putidaredoxin to rabbit microsomal P4502B4 and found very low affinity of the bacterial redoxin for the microsomal P450. Jenkins and Waterman (5) reported the electron transfer from NADPH to heterologously expressed microsomal P450c17 through E. coli flavodoxin and flavodoxin reductase system, although the 17␣hydroxylation activity supported by the E. coli redox system was 10-fold less efficient as compared with purified rat P450 reductase. This flavodoxin and flavodoxin reductase system can be replaced by another ferredoxin and ferredoxin reductase system from spinach for human P4501A2 and 3A4 expressed in E. coli (26,27). Recently, Black et al. (28) reported the monooxygenase activity of the fusion construct of P450scc with NADPH-P450 reductase expressed in COS-1 cells, although the mitochondrial localization of the fused enzyme was absolutely required for its activity. However, because the mitochondrial fraction of COS-1 cells contains an electron transfer system consisting of ferredoxin and ferredoxin reductase, any other protein(s) than P450 reductase may participate in the electron transfer from NADPH to P450scc part. Thus, no direct evidence has been reported for heterologous electron transfer from the microsomal electron donor P450 reductase to mitochondrial P450.
On the expression of mic-P450c27 in yeast, we have previously concluded that endogenous yeast NADPH-P450 reductase was insufficient for mic-P450c27 to show monooxygenase activity, because 1␣-hydroxyvitamin D 3 25-hydroxylation activity could not be detected without the addition of both ADX and ADR (8). However, to our surprise, stopped flow analysis revealed that the reduction rate of mic-P450c27 heme iron in AH22/pAMS25 microsomes is somewhat more rapid than that observed for rat P4501A1 in AH22/pAMC1 microsomes (17). These findings suggest the existence of the electron donor protein(s) to mic-P450c27 on the yeast microsomal membrane.
In order to confirm the electron transfer from NADPH-P450 reductase to mic-P450c27, we first constructed the simultaneous expression plasmid pAMS25R for mic-P450c27 and yeast P450 reductase. The resulting AH22/pAMS25R strain contained a 24-fold increased amount of the reductase as compared with AH22/pAMS25 cells. The reduction rate of mic-P450c27 heme iron in AH22/pAMS25R microsomes was 17-fold more rapid than mic-P450c27 in AH22/pAMS25 microsomes, roughly in proportion to the reductase content. In addition, AH22/ pAMS25R microsomes showed THC 27-hydroxylation activity without addition of ADX and ADR. These results strongly support our hypothesis that mic-P450c27 can accept two electrons from P450 reductase in the P450 reaction cycle.
Next, we constructed the fused enzyme between mic-P450c27 and yeast NADPH-P450 reductase to further investigate the electron transfer from NADPH through P450 reductase to mic-P450c27. The rate of the first electron transfer from NADPH to the heme iron was too rapid to be measured, as was the case for the fused enzyme between rat P4501A1 and yeast P450 reductase (17). In addition, the fused enzyme showed a 5-fold higher THC 27-hydroxylation activity than mic-P450c27 in AH22/pAMS25R microsomes. These observations also strongly suggested the direct electron transfer from the reductase part to the heme iron in the mic-P450c27 part of the fused enzyme. To exclude the possibility of contamination of other electron transfer proteins in the yeast microsomes, mic-P450c27 was purified from the recombinant yeast microsomes. The purified preparation of mic-P450c27 in the liposomal membranes reconstituted with yeast P450 reductase showed a significant THC 27-hydroxylation activity. This is clear evidence that electrons are transferred directly from NADPH to mic-P450c27 through P450 reductase. The discrepancy between the observation of Okuda et al. (24) and that of the present study may be attributable to the amount of P450 reductase used for reconstitution. As compared with the reconstituted system used by Okuda et al. (24), we employed equal or more amounts of yeast P450 reductase for the reconstitution of the liposomal membranes and for coexpression in the present study.
Mic-P450c27 has the microsomal targeting signal derived from bovine microsomal P450c17 that is absent in the mature form of native mitochondrial P450c27. Therefore, it seems likely that the amino-terminal hydrophobic sequence of mic-P450c27 plays an important role not only in its localization to the microsomal membrane but also in its membrane topology resulting in efficient interaction with NADPH-P450 reductase. The redox potential of mitochondrial P450s is quite similar to that of microsomal P450s. Therefore, in the view of electrochemistry there is no difficulty for the electron transfer from  (29,30). The direct electron transfer from microsomal P450 reductase to mic-P450c27 is also of interest from the evolutionary aspect of P450 superfamily. Comparison of primary structures of more than 200 P450 species revealed that microsomal P450s have diverged from an ancestral form of a soluble bacterial P450 and mitochondrial P450s then evolved from a microsomal P450 (7). We substantiated the conversion of P450c27 localization from mitochondria to microsomes by altering its amino-terminal targeting signal (8), showing that only the amino-terminal sequences of 20 -40 residues decide its subcellular localization. On the other hand, the conversion of the electron transport chain from class I (bacteria) to class II (microsomes), and then back again to class I (mitochondria), accompanied with the change in subcellular localization could not have been clearly explained. As shown in the previous (8) and present reports, mic-P450c27 is an example that accepts electrons from both class I and class II electron donors. Accordingly, it is very plausible that P450c27 is located at the divergence point between microsomal and mitochondrial P450s. It is noteworthy that P450c27 appears most closely related to microsomal P450 in evolution (31). Jenkins and Waterman (5) discussed the evolution of eukaryotic P450 reductase based on the result of electron transfer capability of E. coli flavodoxin and flavodoxin system to microsomal P450c17. Flavodoxin and NADPH-flavodoxin reductase in E. coli show some similarity in their sequences to NADPH-P450 reductase within regions of functional importance (5). Thus, it seems likely that P450 reductase evolved from a bacterial flavodoxin and flavodoxin reductase system by a fusion event (5,32) We have established a method for construction of fused enzymes between microsomal P450 and NADPH-P450 reductase to exhibit higher monooxygenase activities with rapid intramolecular electron transfer (10,12,17,(21)(22)(23). In this report, we applied the similar technique to construct a fused enzyme between mitochondrial P450 with an amino-terminal hydrophobic sequence and microsomal NADPH-P450 reductase. Thus, this novel technique for construction of a highly active single chain enzyme consisting of mitochondrial P450 and microsomal P450 reductase would be very useful to study structure-function analysis of P450 monooxygenases.