The interaction of bovine adrenodoxin with CYP11A1 (cytochrome P450scc) and CYP11B1 (cytochrome P45011beta ). Acceleration of reduction and substrate conversion by site-directed mutagenesis of adrenodoxin.

The kinetics of protein-protein interaction and heme reduction between adrenodoxin wild type as well as eight mutants and the cytochromes P450 CYP11A1 and CYP11B1 was studied in detail. Rate constants for the formation of the reduced CYP11A1.CO and CYP11B1.CO complexes by wild type adrenodoxin, the adrenodoxin mutants Adx-(4-108), Adx-(4-114), T54S, T54A, and S112W, and the double mutants Y82F/S112W, Y82L/S112W, and Y82S/S112W (the last four mutants are Delta113-128) are presented. The rate constants observed differ by a factor of up to 10 among the respective adrenodoxin mutants for CYP11A1 but not for CYP11B1. According to their apparent rate constants for CYP11A1, the adrenodoxin mutants can be grouped into a slow (wild type, T54A, and T54S) and a fast group (all the other mutants). The adrenodoxin mutants forming the most stable complexes with CYP11A1 show the fastest rates of reduction and the highest rate constants for cholesterol to pregnenolone conversion. This strong correlation suggests that C-terminal truncation of adrenodoxin in combination with the introduction of a C-terminal tryptophan residue enables a modified protein-protein interaction rendering the system almost as effective as the bacterial putidaredoxin/CYP101 system. Such a variation of the adrenodoxin structure resulted in a mutant protein (S112W) showing a 100-fold increased efficiency in conversion of cholesterol to pregnenolone.

pendent reductase and an iron-sulfur protein of the [2Fe-2S] ferredoxin type.
The reduction kinetics of the soluble bacterial cytochrome P450 cam is well understood (17)(18)(19)(20), whereas little is known about the kinetics of the membrane-bound mitochondrial cytochromes P450 (21)(22)(23). Nevertheless previous investigations point to considerable differences in the reaction mechanisms and in the regulation in these two types of cytochrome P450 systems.
To get deeper insight into the mechanism causing these differences, we intended to create adrenodoxin mutants with an increased efficiency in terms of substrate conversion and reduction rate of CYP11A1 and CYP11B1. According to sequence alignments (24,25), serine in position 112 of bovine adrenodoxin corresponds to the last amino acid in putidaredoxin, tryptophan in position 106. Tryptophan in position 106 of putidaredoxin has been shown previously to be of crucial importance for binding and electron transfer to CYP101 (20,26). Consequently, a set of four truncated adrenodoxin mutants with a C-terminal tryptophan residue in position 112 was prepared to find out how this residue affects the reactivity of adrenodoxin.
For comparison we used two truncated adrenodoxin mutants (Adx- (4 -114) and Adx-(4 -108)), which like the wild type do not contain a tryptophan, to elucidate the effect of the deleted C terminus. Both truncated mutants are known to display an increased substrate conversion activity (27), but a detailed characterization of their reduction behavior has not been performed as yet.
Furthermore, adrenodoxin mutants T54A and T54S (11) were included in this study with the intention to investigate whether a substitution in the [2Fe-2S] cluster surrounding (which is completely unrelated to the above-mentioned adrenodoxin deletions and mutations) could be able to affect affinity and/or reactivity in the adrenodoxin/mitochondrial P450 (i.e. CYP11A1 or CYP11B1) system in a similar manner.

MATERIALS AND METHODS
Biochemicals and Reagents-Pfu DNA Polymerase was from Stratagene. NADPH was purchased from SERVA, and cholesterol was from Sigma. All other chemicals were of the highest purity commercially available.
Bacterial Strains and Plasmids-Escherichia coli strains HB101 and BL21 were used as host strains for the heterologous expression of adrenodoxin. The cDNA-containing plasmid was pKKHC (28).
Protein Expression and Purification-A set of four adrenodoxin mutants was designed using wild type adrenodoxin and three Tyr-82 mutants, Y82S, Y82L, and Y82F (10). A tryptophan residue was introduced in position 112 by polymerase chain reaction-based site-directed mutagenesis, and simultaneously the 16 C-terminal amino acids of the wild type sequence were deleted.
The 3Ј polymerase chain reaction primers consisted of the following sequences: 5Ј-GGGGAAGCTTACCACACGGCATCAGGTACTCG-3Ј (reverse primer) and 5Ј-GGTGACTCTCTGCTAGATGTTGTGG-3Ј (forward primer). The stop codon and the tryptophan codon are bold-faced; in addition, the tryptophan codon is italicized, and the HindIII site is underlined. Oligonucleotides for polymerase chain reaction were purchased from Biotez (Berlin, Germany). Ligation, transformation, plasmid preparation, and dideoxy sequencing were carried out according to standard protocols (29).
Bacteria were grown as previously reported (30) with slight modifications. Recombinant adrenodoxin was purified after enzymatic cell lysis as described previously, and the final concentration of adrenodoxin was determined using ⑀ 414 ϭ 9.8 mM Ϫ1 cm Ϫ1 (31). The purity of the adrenodoxin preparation was estimated by determining the Q-value (A 414 /A 273 ).
SDS-polyacrylamide gel electrophoresis was done using standard procedures (36). Western blot analysis of adrenodoxin was performed according to Sambrook et al. (29) using rabbit polyclonal antibodies.
Kinetics by Rapid Mixing-Stopped flow measurements were carried out with a single channel stopped flow SX-17MV spectrophotometer equipped with polyether-ether-ketone tubing (Applied Photophysics) at 15°C. The system was made anaerobic by incubation with argon-bubbled buffer containing 5 mM dithionite for 1 h followed by repeated flushing with excessively argon-bubbled reaction buffer to completely remove the dithionite.
Conditions were optimized for the reduction of CYP11A1 and CYP11B1 avoiding an oxygen-scavenging system that had been used in former studies (23,27). Anaerobic measurements were carried out under an argon atmosphere. The reaction buffer for CYP11A1 consisted of 50 mM potassium phosphate (pH 7.3), 0.1 M KCl, 0.1 mM EDTA, 0.1 mM dithioerythritol, and 0.05% sodium cholate (35). Syringe A contained 2 M P450 scc and 40 M cholesterol. Syringe B contained 2 mM NADPH, 2 M adrenodoxin reductase, 40 M cholesterol, and adrenodoxin in varying concentrations (ranging from 2 up to 128 M). The mixture in syringe B was allowed to age for at least 10 min to assure a complete reduction of the adrenodoxin. The solutions in the two syringes were CO-saturated prior to loading into the driving syringes. The reaction buffer for CYP11B1 contained 50 mM potassium phosphate (pH 7.3), 0.1 M KCl, 0.1 mM EDTA, 0.1 mM dithioerythritol, 0.3% sodium cholate, and 0.3% Tween 20.
The reaction was monitored at 450 nm recording 4000 data points with a split time base of 10/200 s. The dead time for the measurements was 1.3 ms with the SX-17MV. For each adrenodoxin concentration, three to seven reaction traces were averaged. The reduction of CYP11A1 by all adrenodoxin species, except S112W and Y82F/S112W, is best described by monoexponential relaxations. For adrenodoxin concentrations exceeding 5-fold cytochrome concentration pseudo-first order reaction conditions can be presumed. Monoexponential functions were fitted to the data resulting in apparent rate constants (k app ). The dependence of k app from the adrenodoxin concentration was best described by a hyperbolic saturation. Extrapolation to infinite adrenodoxin concentration resulted in limiting rate constants (k app, max ) for the reduction process. Due to considerable differences in binding strength that lead to experimental limitations (i.e. the absorbance of the solution limits a proper measurement), saturation could not be reached in all cases (especially for wild type adrenodoxin). The reduction of CYP11A1 by S112W and Y82F/S112W and all the corresponding experiments with CYP11B1 resulted in biphasic relaxations. Biexponential functions were fitted to the data. The first, rapid phase showed a hyperbolic dependence on the adrenodoxin concentration and could be extrapolated as described above. Formation of the reduced species was moni-tored by the respective CO complex, showing an absorbance maximum at 450 nm.
Optical Difference Spectroscopy-Titrations were carried out in tandem cuvettes using the same buffer system as described for the stopped flow measurements with 0.03% Tween 20 added to convert CYP11A1 to the low spin state. The sample cuvette typically contained 1.5-2 M CYP11A1 and 30 M cholesterol and was titrated with adrenodoxin. The interaction of adrenodoxin with CYP11A1 promotes the binding of cholesterol, which causes a shift of the low spin (417 nm) to the high spin (393 nm) form of the cytochrome heme iron. The absorbance changes were plotted versus the adrenodoxin concentration and fitted by numerical integration using the program CHEMSIM (37).
Redox Potential Measurements-The dye photoreduction method with Safranine T as the indicator was used to determine the redox potential of adrenodoxin (38) and its mutants. Experiments were analyzed by using the Nernst equation.
Substrate Conversion-CYP11A1-dependent conversion of cholesterol to pregnenolone as well as the CYP11B1-dependent conversion of 11-deoxycorticosterone to corticosterone in dependence of the adrenodoxin concentration was performed as described previously (39) with slight modifications (40). The cholesterol side chain cleavage activity was assayed at 37°C in a reconstituted system catalyzing the conversion of the respective steroid. The buffer we used for the substrate conversion was the same as described for the stopped flow assays but, in addition, contained an NADPH-regenerating system. The samples were analyzed by reversed phase HPLC using acetonitrile/isopropanol (30:1) (CYP11A1) and acetonitrile/phosphoric acid (10 mM) (1:1) (CYP11B1), respectively.

Production of Adrenodoxin S112W and Double Mutants with
Additional Changes in Position 82-The mutants were produced by site-directed mutagenesis using the primers described under "Materials and Methods." Dideoxy sequencing revealed that the adrenodoxin cDNA contained only the desired mutations except for mutant S112W where an additional silent mutation had occurred in codon 87 (AGA 3 AGG), which, however, does not change the coded amino acid.
Expression and Purification-The mutant proteins were expressed into the cytoplasm of E. coli BL21. The specific content of cells per liter of culture was 4.3-5.3 g. Production of the heterologous proteins was verified by SDS-polyacrylamide gel electrophoresis and Western blot analysis (data not shown).
The efficiency of each purification step is reflected by the increase in the ratio A 414 /A 276 (Q-value, see Table I). Mutant proteins in which a tryptophan residue was introduced display lower Q-values as compared with the wild type protein due to the additional tryptophan, which increases the protein absorbance at 276 nm. The final yield of each adrenodoxin mutant after purification is given in Table I. All four tryptophan mutants gave lower yields in purification than the wild type protein due to a slightly different behavior of the mutants during the hydrophobic interaction chromatography step, i.e. the salt gradient used for wild type purification (starting from 2.2 M ammonium sulfate down to 1.0 M) was not sufficient to completely elute the respective mutant protein from the column. Spectral Characterization-UV/visible spectra of the investigated adrenodoxin mutants clearly display differences in the UV region at 280 nm as compared with the wild type protein due to the newly introduced tryptophan residue (Fig. 1). Also the contribution of the amino acid in position 82 (Tyr, Phe, and nonaromatic, respectively) to the absorbance around 280 nm is detectable, whereas a difference between Y82S/S112W and Y82L/S112W could not be observed.
Redox Potentials-The redox potentials of wild type and mutant adrenodoxins are displayed in Table II. Interestingly the redox potential of all four tryptophan mutants is more than 60 mV lower as compared with the wild type protein. It is readily noted that these values are within the same range as those of the adrenodoxin mutants that do not contain a tryptophan but either are truncated or contain an amino acid substitution in the immediate cluster environment.
K S Values-The binding affinity of adrenodoxin to its natural redox partner CYP11A1 has been determined for the adrenodoxin mutants (Table III). Differences in K S of the wild type as compared with other publications (27) could be attributed to the different buffer conditions used for the titrations, i.e. the buffer we used contained 0.1 M potassium chloride, whereas in former studies a potassium phosphate buffer without additional salt was used to determine the K S values. The ionic strength is known to largely affect the interaction between adrenodoxin and CYP11A1 (22). Moreover, in contrast to most former studies where different buffers were used for the determination of binding constants, steady-state kinetic constants, and pre-steady-state kinetic constants, respectively, we used the same buffer system for the binding studies, the CYP11A1-dependent conversion of cholesterol, and the stopped flow assays to assure similar assay conditions for all experiments and thus to allow the determination of affinities relevant in the reduction reaction. An example for these titrations is given in Fig. 2.
C-terminal truncation of adrenodoxin leads to an increased affinity that is even more pronounced together with a C-terminal tryptophan residue and is dramatically enhanced for the combination of truncation, aromatic amino acid in position 82 (i.e. adrenodoxin mutants S112W and Y82F/S112W), and C-terminal tryptophan (Table III).
Kinetics of CYP11A1 Reduction-The stopped flow measurements allow the CO complex formation at 450 nm that occurs after electron transfer to the heme iron to be followed. Binding of CO to the reduced heme iron can be considered as an irreversible and not rate-limiting step under the experimental conditions used. Fig. 3 shows the apparent rate constants of the observed reactions plotted versus the concentration of adrenodoxin. These plots were best described by hyperbolic satura-FIG. 1. Absorbance spectra of adrenodoxin wild type and the different mutants. Spectra were collected using cuvettes with a path length of 0.1 cm. Proteins were in storage buffer (10 mM potassium phosphate, pH 7.4). Due to the fact that there were no spectral differences to detect between mutant Y82S/S112W and Y82L/S112W, only the absorbance spectrum of Y82S/S112W is shown. Starting from the bottom, the spectra correspond in order of appearance to wild type adrenodoxin, mutant S112W, Y82S/S112W, and Y82F/S112W.

TABLE II Redox potentials of adrenodoxin and its mutants
The measurements were carried out as described under "Materials and Methods." Each determination was done at least three times, and the deviation was less than Ϯ5 mV for each protein. References are given in parentheses.

Adrenodoxin species
Redox potential mV Wild type Ϫ270 S112W Ϫ334 Y82F/S112W Ϫ332 Y82L/S112W Ϫ333 Y82S/S112W Ϫ332 Adx-(4-114) Ϫ342 (7) Adx-(4-108) Ϫ344 (12) T54A Ϫ329 (11) T54S Ϫ340 (11)  2. Adx-induced low spin (418 nm) to high spin (393 nm) conversion of CYP11A1 followed by differential spectroscopy. Shown is the titration of 1.5 M CYP11A1 with varying concentrations of adrenodoxin mutant Y82F/S112W. tion, which allowed the extrapolation to apparent maximal velocity constants for the respective adrenodoxin species. Thus, the different adrenodoxin mutants can be compared with respect to their efficiency to reduce the cytochrome. Table IV displays the maximal velocity constants (k app, max ) derived from the fits shown in Fig. 3. For six mutants (Y82S/ S112W, Y82L/S112W, Adx-(4 -108), Adx-(4 -114), T54A, and T54S) and the wild type, a monoexponential function could be fitted to the data, whereas the time courses for the other mutants (Y82F/S112W and S112W) were best described by biexponential functions. The first, rapid phase of the biexponential covered about 33% of the total reaction amplitude and was used to calculate k app, max (see below). Examples for the different time courses are given in Fig. 4. The second, slow phase (see Table IV) appeared similar for S112W and Y82F/S112W.
Kinetics of CYP11B1 Reduction-Rate constants determined from the stopped flow measurements are shown in Table IV. All time courses were best described by biexponential functions. Interestingly in this study no significant difference between the adrenodoxin species used is observed.
Substrate Conversion-Conversion of cholesterol to pregnenolone was performed according to Hannemann et al. (40). k cat and K m values for the different adrenodoxin species are listed in Table V. Data clearly show an increased k cat and, correspondingly, a lowered K m for the truncated mutants, particularly for the two mutants S112W and Y82F/S112W. The k cat value is increased 6.6-fold (S112W) and 9.4-fold (Y82F/S112W), whereas the K m is decreased 9-fold (S112W) and 9.7-fold (Y82F/ S112W). All truncated mutants display similar K m values, being lowered by a factor of about 10 as compared with the wild type. In contrast, the K m values for the "full-length mutants" are not changed in this dramatic way, being 2-fold higher (T54A) and 1.6-fold lower (T54S) than the wild type. Besides mutants S112W and Y82F/S112W (see above), a considerable effect on k cat is observable only for the truncated mutant Adx-(4 -108) where the k cat is 2.8-fold increased.
The data obtained from CYP11B1-dependent conversion of 11-deoxycorticosterone to corticosterone are summarized in Table VI. The large differences between the adrenodoxin species observed for CYP11A1-dependent conversion of cholesterol to pregnenolone are not detected in the CYP11B1 reaction. The differences in k cat are marginal (largest difference factor is 1.8 for T54S and wild type) when comparing wild type adrenodoxin and the mutant species. The K m values differ by a factor of 2.4 for adrenodoxin Y82F/S112W as compared with the wild type, which is the most pronounced difference observed. DISCUSSION An interesting and so far not understood feature in cytochrome P450-dependent reactions is the observation that most of the bacterial systems display 100 -1000-fold higher turnover rates as compared with microsomal and mitochondrial ones. Attempts to improve the efficiency of mitochondrial P450 systems should thus lead to deeper insight into the mechanism of protein-protein interactions and reduction kinetics in this important class of enzymes.
We used wild type adrenodoxin and a set of eight mutant proteins in stopped flow experiments as well as in differential spectroscopy to investigate the kinetics and thermodynamics of the protein-protein interaction and the reduction rate of CYP11A1 and CYP11B1 by adrenodoxin. The electron transfer was indirectly monitored at 450 nm where CO binding to P450, which occurs upon reduction of the protein-bound heme, can be detected.
To correlate the initial reduction process determined in stopped flow experiments with the subsequent redox cycles (one more cycle in the case of the CYP11B1-dependent conver- sion of 11-deoxycorticosterone to corticosterone and five subsequent redox cycles for the conversion of cholesterol to pregnenolone by CYP11A1, respectively (16,41)) and as a monitor for the overall reactivity of the respective adrenodoxin species, we studied the CYP11A1-and CYP11B1-dependent substrate conversions mediated by wild type adrenodoxin and the various mutants.
The components of the bacterial putidaredoxin/CYP101 sys-tem are homologous to the mitochondrial adrenodoxin/ CYP11A1 and adrenodoxin/CYP11B1 system and have been extensively studied due to the ready availability of the different proteins (19,26,(42)(43)(44). Therefore, it became the best understood and commonly used model system for electron transfer investigations in ferredoxin-mediated P450 reactions, although experiments revealed considerable differences between the respective systems. (i) They differ by about 1 order of magnitude in the velocity of formation of the reduced CO complex displaying rate constants of 2.5-3 s Ϫ1 for CYP11A1 reduction (22) and 33 s Ϫ1 for the CYP101 system (45). (ii) The aromatic character IV Determined maximal rate constants for the reduction of CYP11A1 and CYP11B1 with Adx and its mutants Rate constants for the respective P450s were determined at different adrenodoxin concentrations from raw data fitted by a monoexponential (Mono) or a biexponential (Bi) equation. These rate constants were plotted against the corresponding adrenodoxin concentration and fitted according to a hyperbolic equation (f(x) ϭ y 0 ϩ ax/(b ϩ x)).   of the C-terminal tryptophan in position 106 of putidaredoxin (26,44) is indispensable for the strong binding of putidaredoxin and CYP101, whereas wild type adrenodoxin has no tryptophan at all and also comprises a more extended C-terminal end (adrenodoxin consists of 128 amino acids). (iii) Putidaredoxin and adrenodoxin are unable to substitute for one another in the respective reductase reaction (5) or to serve as an effector for substrate turnover in the corresponding system of the other one (18), whereas the respective reduced P450⅐CO complexes could be generated (46). To design more efficient adrenodoxin mutants, i.e. proteins with putidaredoxin-like properties, we constructed a set of four truncated adrenodoxin mutants with a C-terminal deletion of 16 amino acids and introduced a tryptophan in position 112, which according to sequence alignments corresponds to tryptophan 106 in putidaredoxin. For comparison, two truncated adrenodoxin mutants, Adx-(4 -114) and Adx-(4 -108), which do not contain a tryptophan, and the two full-length adrenodoxin mutants, T54A and T54S, showing a substitution in the surrounding of the [2Fe-2S] cluster, were used.
The redox potentials of all the truncated forms of adrenodoxin as well as of the full-length mutants containing a replacement in position Thr-54 are similar (Table II) and differ from that of the wild type by about Ϫ60 mV. Therefore, a contribution of the amino acid in position 82 to the redox potential can be ruled out in concordance with former studies where fulllength adrenodoxin Tyr-82 mutants showed wild type-like midpoint potentials (10). Consequently the drop in midpoint potential is due to the deletion of the adrenodoxin C terminus in the truncated mutants and/or the presence of a tryptophan residue. The latter explanation, however, seems unlikely since the redox potential of putidaredoxin is significantly higher (47) as compared with wild type adrenodoxin. Additionally, sitedirected mutagenesis studies on putidaredoxin proved that the amino acid at position 106 (Trp in the wild type protein was mutated to Phe, Leu, Lys, Val, Tyr, respectively) or even its absence (mutant ⌬106) does not significantly modulate the redox potential (20). Moreover, the truncated adrenodoxin mutants Adx-(4 -114) and Adx-(4 -108) also exhibit lowered redox potentials.
In the Thr-54 mutants, on the other hand, a change in the hydrophobicity of the cluster surrounding may account for the lowered redox potential; in fact, this explanation might hold true for the truncated mutants as well considering a structural rearrangement within the molecule due to deletion of the C terminus that is transmitted to the environment of the cluster. Indeed, Burova et al. (48) described a more compact overall structure of adrenodoxin mutant Adx-(4 -108) as determined by calorimetric experiments and limited proteolysis indicating that differences between the wild type Adx and the truncated Adx exist in solution (48). The lowered redox potentials of the truncated mutants should improve their ability to reduce CYP11A1, which, in fact, has been observed (Tables II and IV). In this context, it is interesting to note that the two Thr-54 mutants, despite their lowered redox potentials, behave wild type-like in the kinetic assays (discussed below).
The ability of the adrenodoxin mutants to reduce CYP11A1 and CYP11B1 and to promote the cytochrome P450-dependent substrate conversion as well as the affinity of the adrenodoxin mutants to CYP11A1 was studied by performing different spectroscopic measurements using steady-state and stopped flow kinetics.
The data we provide clearly show dramatic differences in the interaction of the designed adrenodoxin species with CYP11A1 (Table III), whereas it was not possible to determine the respective binding constants for CYP11B1 due to the inherent instability of this cytochrome in the low spin state. Wild type adrenodoxin as well as the adrenodoxin mutants T54A and T54S (which are identical in length with the wild type protein) bind weaker by a factor of 8.5-12.3 to CYP11A1 than the C-terminally truncated mutants S112W and Y82F/S112W. The other mutants (Adx-(4 -114), Adx-(4 -108), Y82S/S112W, and Y82L/S112W), which are also lacking 14 -20 amino acids at the C terminus, display increased affinities to CYP11A1 as well. Obviously truncation of the C-terminal end of adrenodoxin enhances the affinity of the iron-sulfur protein to its redox partner CYP11A1. Thus, one explanation for the function of the C-terminal end of adrenodoxin in vivo might be the modulation of binding strength to a moderate level to permit an optimal interaction of adrenodoxin with its various reaction partners.
The mutant adrenodoxins S112W as well as Y82F/S112W, besides improved affinity to CYP11A1, also show enhanced activity in the cholesterol conversion reaction (see Table V) with increased k cat values (6.6-and 9.4-fold) and decreased K m values (9-and 9.7-fold) resulting in a 50 -100-fold increase of the reaction efficiency (k cat /K m ). Looking at the other mutants of this study, only adrenodoxin-(4 -108) displays a higher k cat , being increased by about 2.8-fold as compared with the wild type. Interestingly the four tryptophan mutants differ largely in their binding affinity to CYP11A1, which could be attributed to the mutation in position 82 on the surface of adrenodoxin where an aromatic residue seems to be necessary for an amplification in binding strength (Table III). On the other hand, the K m values of all truncated mutants are within the same range and are decreased by a factor of ϳ10 as compared with wild type adrenodoxin. This lowered K m in CYP11A1-dependent cholesterol conversion is paralleled by decreased K S values of the respective CYP11A1⅐adrenodoxin complexes. Surprisingly the differences found for the CYP11A1-dependent substrate conversion mediated by adrenodoxin mutants S112W and Y82F/S112W in comparison to the adrenodoxin wild type could not be observed with CYP11B1 as the final electron acceptor (see Table VI). Only minor differences between the adrenodoxin species were detected in the CYP11B1-dependent conversion of 11-deoxycorticosterone to corticosterone with a maximum difference in k cat /K m of 3.8-fold for adrenodoxin mutant Y82F/S112W. These data can be interpreted in two ways: either none of the structural alterations in the different adrenodoxin mutants affect the interaction with CYP11B1, or the rate-limiting step of the CYP11B1-dependent substrate conversion reaction occurs after the interaction of adrenodoxin with the cytochrome. In the latter case, substrate hydroxylation or product release from the enzyme could be considered ratelimiting. In fact, Imai et al. (49) demonstrated that the ratelimiting step in the CYP11B1-dependent aldosterone production is the release of the product from the enzyme. If this was also true for corticosterone formation, possible alterations in the interaction of the different adrenodoxin species with CYP11B1 would not be detected in this experimental system. It has to be kept in mind, however, that the interactions between the different components of the P450 systems described here are at least partially mediated by electrostatic forces. Therefore, the salt concentration used in the assays most probably affects binding affinities and efficiency of the respective system so that former results (11,27) obtained under conditions different from ours show comparable tendencies indeed but cannot be related directly to the values presented in this work.
The rate constants determined for the formation of the reduced CYP11A1⅐CO complex by the various adrenodoxin species differ by up to a factor of 10 (Table IV) that again is in contrast to the values observed for the CYP11B1 reduction where all the different adrenodoxin species behave quite sim-ilarly (Table IV). The time courses of the CYP11B1 reduction were best described by biexponential functions whereby only the first maximum rate constant could be determined within a useful range of deviation. On the other hand, the impossibility to correlate the time courses with a monoexponential function could be cautiously interpreted as a hint to a second process taking place during the monitored reaction.
For CYP11A1, on the other hand, a comparison of kinetic and thermodynamic parameters for the different adrenodoxin mutants studied unravels pronounced dependences. Stronger interaction between CYP11A1 and adrenodoxin is paralleled by faster reduction of CYP11A1 (as detected by CYP11A1⅐CO complex formation) and faster substrate conversion. Mutants S112W and Y82F/S112W display greatly increased rate constants as compared with wild type adrenodoxin, being 10-fold faster, and also the rate constants of the other mutants can be discriminated according to their respective C-terminal extension. An aromatic residue in position 82 of adrenodoxin appears to be important for an additional increase of the rate constant that again is in concordance with the amplified binding affinity of those adrenodoxin mutants to CYP11A1 and with their accelerated kinetics of the CYP11A1-dependent substrate conversion.
Moreover, biexponential time courses of CYP11A1 reduction could only be observed for adrenodoxin S112W and Y82F/ S112W, i.e. those tryptophan mutants containing an aromatic residue in position 82 (which is on the surface of adrenodoxin). This result might be explained by the formation of at least two different complexes of adrenodoxin and CYP11A1 with only one being productive while the other one is unable to transfer electrons (Fig. 5). From the smaller amplitude (see Fig. 4) of the first, rapid phase, it can be concluded that the productive complex is thermodynamically less favored. This mechanistic model is supported by the fact that the second rate constant is independent from adrenodoxin concentration as it would be assumed for a rearrangement of an unproductive complex to form the productive configuration. The formation of additional hydrophobic interactions between the respective adrenodoxin mutants and CYP11A1 seems to be possible and may result in a conformational change of the cytochrome P450. In a computer model (50) two aromatic residues (Trp-401 and Phe-411) are located on the CYP11A1 surface close to the site of ionic interaction with adrenodoxin (51). These two residues might form stacking interactions with Trp-112 of the adrenodoxin mutants (S112W and Y82F/S112W). Obviously an aromatic amino acid in position 82 of adrenodoxin is also required for this potential interaction as no biphasic behavior was observed with the other adrenodoxin species.
The CYP11A1 reduction courses of all other adrenodoxin species tested were apparently monoexponential, i.e. no second phase was observed. This means that either no unproductive adrenodoxin⅐CYP11A1 complex exists with these proteins, or the rearrangement of an unproductive complex into the productive complex is as fast or even faster than the reduction of CYP11A1 mediated by the respective adrenodoxins. It would, therefore, be of great interest to obtain three-dimensional information on the adrenodoxin⅐CYP11A1 complex(es) and to elucidate the structural basis of the formation of different (i.e. productive and unproductive) forms.
A comparison of the rate constants for the formation of the reduced CYP11A1⅐CO complex, e.g. 23 s Ϫ1 for the fastest (S112W) mutant, with the k cat value of pregnenolone formation (0.07 s Ϫ1 for S112W, see Table V) shows that these rate constants cannot be directly correlated. The same holds true for the relation of these values of all other adrenodoxin species. This leads to the conclusion that the rate-limiting step in the CYP11A1-dependent substrate conversion, although being correlated to adrenodoxin binding and CO reduction rate, remains to be elucidated.
Taken together the truncated adrenodoxin mutants containing a tryptophan at the C-terminal end show the tendency to a more "putidaredoxin-like" behavior, i.e. a stronger binding to the redox partner CYP11A1, an enhanced substrate conversion activity, and a faster formation of the reduced CYP11A1⅐CO complex. The introduced mutations lead to a significant increase in substrate specificity (i.e. k cat /K m ) by 2 orders of magnitude.