Electron Transfer between Cytochrome P450cin and Its FMN-containing Redox Partner, Cindoxin*

Cytochrome P450 reductase, which delivers electrons from NADPH to microsomal P450s, consists of a single polypeptide that contains both FAD and FMN. The bacterial P450cin utilizes a similar electron transport system except the FAD/FMN reductase consists of two separate polypeptides where the FMN protein, cindoxin, shuttles electrons between the FAD-containing cindoxin reductase and P450cin. Here we characterize the kinetics and specificity of electron transfer between cindoxin and P450cin as well as discuss the influence of possible binding surface interactions using homology models.

Two electrons are required for O 2 activation by the P450 heme iron that are delivered from NAD(P)H by P450 redox partners. Until recently, P450s were divided into two classes based on the types of redox partners used. Class I is a threecomponent system that uses a NAD(P)H-dependent FAD-containing protein to reduce a Fe 2 S 2 ferredoxin that then transfers two electrons, one at a time, to P450. The two-component class II consists of a FAD/FMN P450 reductase that mediates electron transfer from NADPH to P450. Not too long ago mitochondria and prokaryotes were neatly grouped into class I while microsomal P450s were placed in class II. With an increasing genomic data base and a corresponding increase in the characterization of new P450s, it now is clear that a simple two-class categorization is too limited. Indeed, a recent review defines 10 different P450 electron transfer classes (1). For example, P450BM3 is a prokaryotic fatty acid hydroxylase that uses a microsomal-like FAD/FMN reductase with the important exception that the reductase is fused to the C-terminal end of the P450 to give a catalytically self-sufficient P450 (2,3). P450RhF contains class I P450 cofactors (FMN, Fe 4 S 4 , and heme) fused into a single polypeptide (4), and CYP119, the first thermophilic P450 to be discovered, is thought to use a ferredoxin and 2-oxoacid-ferredoxin oxidoreductase (5).
P450cin from Citrobacter braakii is a three-component class I prokaryotic system except its flavin-containing components are class II-like. The P450cin monooxygenase system consists of a FAD reductase that transfers electrons to a FMN protein that then delivers electrons to P450cin. Sequence homologies clearly show that the FMN and FAD proteins are closely related to the FMN and FAD modules of P450 reductase (6). Indeed, the FMN protein cindoxin (Cdx) 3 exhibits 37% sequence identity to the FMN domain of human microsomal P450 reductase (CPR FMN ), whereas the FMN module of P450BM3 shares only 28% sequence identity with the human FMN domain.
We recently solved the crystal structure of P450cin (7) and had anticipated that the presumed electron transfer site on P450cin would resemble that of P450BM3 and other P450s that utilize an FMN module as the redox partner. Instead, we found that the architecture of the proximal docking site is more similar to P450cam, which utilizes a Fe 2 S 2 protein, putidaredoxin (Pdx). To further probe the interaction between Cdx and P450cin we have initiated a series of studies to characterize the electron transfer reaction. Here we report the kinetics and specificity of electron transfer between Cdx and P450cin.

MATERIALS AND METHODS
Protein Purification-Plasmids containing P450cin and Cdx were provided by Dr. James DeVoss. P450cin was expressed and purified as described previously (7). Adrenodoxin (Adx) and CPR expression plasmids were provided by Drs. Y. Sagara and C. Kasper, respectively. Expression and purification of Adx and Pdx were carried out as described elsewhere (8,9). The FMN domain of human CPR containing residues 1-181 was cloned into EcoRI and KpnI sites of the pProEX expression vector (Invitrogen) with a cleavable N-terminal sixhistidine tag and purified using Ni 2ϩ affinity and gel filtration chromatography. ⑀ 454 of 9.8 mM Ϫ1 cm Ϫ1 was employed to calculate concentration of oxidized CPR FMN .
Cdx was purified as follows. Overnight cultures of Escherichia coli DH5␣ harboring the Cdx expression plasmid were used to inoculate 1 liter of terrific broth (TB) medium in a 2.8-liter flask. Cells were grown at 37°C and 220 rpm to a 600-nm absorbance of 1-1.5 followed by addition of 1 mM isopropyl ␤-D-1-thiogalactopyranoside. Cell growth was continued at 27°C and 70 rpm; 48 h after induction cells were harvested by centrifuging at 6000 rpm and kept frozen at Ϫ20°C. All lysis and purification steps were carried out at 4°C in 50 mM potassium phosphate, pH 7.4, and 0.5 mM dithiothreitol (buffer A).
Cell pellets were resuspended in buffer A containing 0.1 mM phenylmethanesulfonyl fluoride, 1 g/ml leupeptin and pepstatin A and stirred for 1 h with 0.1 mg/ml lysozyme. Lysis was completed by sonication with a Branson Sonifier 450 using four to six alternations of 20-s pulses and 40-s incubations on ice. Cell debris was removed by centrifugation at 17,000 rpm for 1 h. The supernatant was diluted 3-to 5-fold with buffer A and loaded on a DEAE-Sepharose column equilibrated with buffer A. The column was washed with three column volumes of buffer A, two volumes of buffer A containing 100 mM KCl, and finally with four volumes of buffer A containing 180 mM KCl. Cdx was eluted with a gradient of 180 -300 mM KCl in buffer A. Yellow-colored fractions were checked for purity by SDS-PAGE, and those containing a 16-kDa band were combined and concentrated to 2 ml using an Amicon YM10 concentrator. The resulting sample was loaded onto a S-200-Sepharose gel filtration column and eluted with buffer A. Fractions were checked for purity using SDS-PAGE and by absorption spectra typical of a flavoprotein, with a well separated peak at 280 and a ratio of absorbance at 450:375 nm of 1.2. The extinction coefficient of oxidized Cdx was determined by boiling a small amount of Cdx to release free FMN and estimating the amount of FMN released (⑀ 450 ϭ 12.2 mM Ϫ1 cm Ϫ 1). Assuming a one-to-one FMN:protein ratio, the value of ⑀ 450 nm for Cdx was calculated to be 10.5 mM Ϫ1 cm Ϫ1 .
UV-visible Spectroscopy-All spectroscopic measurements were carried out on a Varian Cary 3E UV-visible spectrophotometer in a 1.0-ml gas-tight quartz cuvette with a rubber septum. Additions were made using a 50-l Hamilton gas-tight syringe.
Stopped-flow-All solutions were made anaerobic by alternating evacuating and flushing with pure argon. Concentration of the sodium dithionite solutions was determined by titrating cytochrome c solutions and measuring the amount of cytochrome c reduced upon addition of small aliquots of the reductant. Solutions were loaded into gas-tight syringes in an anaer-obic glove box. Because the stopped flow instrument was in an aerobic environment, an oxygen-scrubbing system of 1 mM glucose, 1 unit/ml glucose oxidase, and 1 unit/ml catalase was included in the wash buffers and experimental solutions. Absorbance changes were monitored at 450 nm to follow formation of the ferrous-CO complex formation. Kinetic data were analyzed using IgorPro software (WaveMetrics, Inc.). Data were analyzed according to Reaction 1 where Cdx(hq) and Cdx(sq) are the hydroquinone (two-electron-reduced) and semiquinone (one-electron-reduced) forms of Cdx, respectively. The dissociation constant, K D , and maximum rate of electron transfer, k, were estimated from double reciprocal plots of 1/k obs rate versus 1/[Cdx] concentration. Homology Model-The amino acid sequence of Cdx was submitted to the online program 3D Jigsaw, which returned a PDB coordinate file with the sequence of Cdx threaded on to the x-ray crystal structure of the FMN binding domain of human cytochrome P450 reductase (PDB code 1B1C). FMN was manually added to the model, and the model was energy-minimized. Flavodoxins and P450 reductase FMN modules have small, compact, and highly homologous structures. As a result, there is a high level of confidence that the Cdx homology model is a good approximation of the correct structure.
Spectroelectrochemical Titration-Redox titrations were carried out in a cuvette that was assembled in the anaerobic Coy glove box. The cuvette was sealed with air-tight septa through which the Ag/AgCl reference electrode, the gold working electrode, and gas-tight Hamilton syringe were inserted. A small magnetic stir bar was placed at the bottom of the cuvette to mix the reagents. The temperature was maintained at ϳ25°C. The titration buffer (50 mM potassium phosphate, 50 mM potassium chloride, pH 7.4) was made anaerobic by flushing it with ultra-  SEPTEMBER 14, 2007 • VOLUME 282 • NUMBER 37 pure argon while stirring. The final experimental volume was 1.3 ml containing 10 -15 M protein. The following typical redox mediators were used: methyl viologen (Ϫ430 mV), benzyl viologen (Ϫ374 mV), 2-hydroxy-1,4-naphthoquinone (Ϫ145 mV), and anthraquinone 2-sulfonate (Ϫ230 mV) to a final concentration of 2 M each. All the potentials here are reported against standard hydrogen electrode (SHE).

P450cin-Cindoxin Electron Transfer
The protein and mediator mixtures were deoxygenated under the flow of argon gas for several minutes. The protein was reduced by addition of a small excess of anaerobically prepared sodium dithionite solution (concentration was determined using a molar absorption coefficient ⑀ 315 ϭ 8.05 mM Ϫ1 cm Ϫ1 ). The reduced protein spectrum was recorded to confirm complete reduction. After each addition of a small aliquot of oxidant/reductant, the solution was stirred until equilibration (stabilization of the potential) was reached (ϳ15-20 min), and then the spectrum (350 -800 nm) was recorded. The titration continued until the sample solution was maximally oxidized by ferricyanide (10).
The electrochemical potential was monitored using an Orion pH/mV meter (Model SA 720) coupled to a gold electrode and an Ag/AgCl reference electrode from Bioanalytical Systems, Inc. The gold electrode was modified using 4,4Ј-dithiodipyridine. The electrode system was calibrated using the ferrousferric ammonium sulfate couple (ϩ675 mV) (11). The observed potential was obtained relative to the Ag/AgCl reference electrode. Hence, they were corrected (using the calibration data for the ferrous-ferric ammonium sulfate solution) to values relative to the standard hydrogen electrode by addition of 197 mV to the data obtained using the Ag/AgCl electrode.

RESULTS
Redox Potential of Cytochrome P450cin-The UV-visible spectrum of the oxidized substrate-free low spin protein has maxima at 416, 535, and 569 nm. The UV-visible spectrum of the oxidized substrate-bound protein has maxima at 391, 510, and 540 nm, which is typical of high spin P450s. The reduced protein has maxima at 408 and 540 nm. Any of these spectral differences could be used to monitor the change in population from the oxidized (Fe 3ϩ ) to the reduced (Fe 2ϩ ) state. Because the Soret absorbance in the 390 -416-nm range overlapped with the absorbance of the mediators, the Q-bands from 500 -570 nm were used to monitor the titrations. The percentage of oxidized/reduced protein was calculated from the spectral data and was plotted against the observed potential. The data were analyzed and titration curves were fitted to the Nernst equation where ⌬E is the observed potential and is plotted against the logarithm of the ratio of the concentration of oxidized (C OX ) and the reduced species (C RED ). The intercept of the plot gives ⌬E°, the redox potential of the protein, and the slope divided by 59 mV gives n, the number of electrons involved in the reaction   1A). This finding is in contrast to previously reported results (12) where P450cin was bound to an electrode and showed no redox shift upon substrate binding. An increase in redox potential upon substrate binding is observed in many other P450s and is consistent with known structural changes that result from substrate binding. The shift in spin state from low to high is due in part to the displacement by the substrate of the axial water ligand from the heme iron, thus shifting the heme iron from low spin hexacoordinate to high spin pentacoordinate. In addition, substrate binding dehydrates the active site, which further destabilizes the lower redox potential. The change in redox potential also provides control of electron transfer. In some P450s electron transfer to the low spin substrate-free form is thermodynamically unfavorable whereas the increase in redox potential upon substrate binding removes the thermodynamic barrier to electron transfer. This prevents the futile consumption of electrons by the substrate-free enzyme (13)(14)(15). Redox Potentials of Cdx-During the purification of Cdx, blue and green bands were observed during the loading of the cell-free extract on to the DEAE column, indicating the presence of partially reduced Cdx. Moreover, upon reduction of purified Cdx with small amounts of dithionite, the blue, neutral semiquinone form was readily observed with an absorption maximum of 590 nm. When the reduced Cdx is exposed to an aerobic environment, the blue semiquinone is reoxidized to the yellow quinone. The sensitivity of the Cdx semiquinone to oxygen is not shared by full-length CPR or the FMN and FAD domains (16). Despite the difference in the stability of the semiquinone, these initial spectral observations suggested that, like P450 reductase, the redox potential of the Sq/Hq couple should be lower than the Ox/Sq couple.
The wavelengths used for determining the redox potential of Cdx were 474 and 583 nm, which are close to the absorption maximum of the oxidized flavin and the absorption maximum for the blue semiquinone, respectively. These absorbances were used to calculate the percentage of oxidized FMN that was plotted against the observed potential (Fig. 1B). Data were fit using the Nernst equation where the two one-electron steps in reduction were analyzed separately because only oxidized and   SEPTEMBER 14, 2007 • VOLUME 282 • NUMBER 37 semiquinone species were present during most of the first step and only semiquinone and hydroquinone forms were present during most of the second step. Data points in the region of maximal semiquinone accumulation were not included in the Nernst plot as in this region all three redox species could be present (14,17,18). The values of the midpoint potentials obtained are: (a) Ϫ93 Ϯ 17 mV for the Ox/Sq couple and (b) Ϫ226 Ϯ 5 mV for the Sq/Hq couple. The redox potential of the mixed couple Ox/Red is Ϫ167 Ϯ 2 mV. These values are close to those of CPR FMN ( Table 1) and suggest that the two-electron-reduced FMN transfers electrons to P450cin because the hydroquinone is a better reducing agent than the semiquinone. Despite these similarities, the Ox/Sq potential for Cdx is lower than that for P450 reductase, which may in part account for the greater air instability of the semiquinone in Cdx.

P450cin-Cindoxin Electron Transfer
Kinetics-The first electron transfer between Cdx and P450cin was monitored by mixing oxidized CO-saturated P450cin and Cdx, fully reduced by a stoichiometric amount of dithionite, in a stopped-flow spectrophotometer and recording absorbance changes at 450 nm over time (Fig. 2). P450cin shows saturation at ϳ12-fold excess Cdx, with a K D ϭ 3 M and k ϭ 15 s Ϫ1 (Fig. 2B). The rate of electron transfer value is comparable with that calculated for other class I and II redox couples (Table 2) (19 -22). Electron transfer from Cdx to P450cin was also found to depend on ionic strength. As the salt concentration increased, the rate of P450cin reduction dramatically decreased (Fig. 2C). A full kinetic analysis (Fig. 2C, inset) at 200 mM salt shows that the maximum rate of electron transfer is not salt-dependent but K D increases from 3 to 30 M in going from 5 to 200 mM salt. This suggests that complementary electrostatic interactions play an important role in the P450cin-Cdx complex formation.
To test the specificity of P450cin, we investigated how the hemoprotein reacts with class I ferredoxins Adx and Pdx (Fig. 3). Adx-to-P450cin electron transfer was very slow, with a secondary reaction (data not shown) that quenched absorbance from reduced P450cin at longer time scales and higher Adx concentrations. This made it difficult to make an accurate determination of k and K D from double reciprocal plots (Fig. 3C, inset). For Pdx k is 27-fold lower and K D 45-fold higher than for Cdx, whereas for CPR FMN k is 23-fold slower and K D 6-fold higher (Fig. 3, Table 2). Homology Model-Cdx has 37% identity and 56% similarity with human CPR FMN . The energy-minimized model of Cdx, generated using the program 3D Jigsaw, showed minor deviations from the crystal structure of CPR FMN . CPR FMN is longer and has an extra small helix at the N terminus. In addition, the fragment comprising the third helix in CPR FMN is not well structured in the Cdx model. The P450cin-Cdx redox complex was modeled based on the known structure of the P450BM3 complexed with its FMN module (PDB code 1BVY) (23). P450cin was first superimposed on P450BM3 by aligning the heme groups. Next, the energy-minimized homology model of Cdx was superimposed on the FMN domain of P450BM3. Fig.  4A shows the hypothetical P450cin-Cdx complex with the superimposed CPR FMN . Even without any energy optimization of the complex, the fit between P450cin and Cdx is good with few steric clashes and good ionic pairing and the FMN within electron transfer distance of 20 Å from the P450cin heme (Fig.  4A). The ionic strength dependence of the Cdx-to-P450cin electron transfer reaction (Fig. 2) indicates that electrostatic interactions are important, similar to other P450s. Not surpris- ingly, the docking surfaces have complementary electrostatic potentials with P450cin serving as the positively charged partner and Cdx the negatively charged partner.
Because CPR FMN and Cdx are so similar in structure, a more detailed comparison between these two might provide some insight into why CPR FMN is a poor reductant of P450cin compared with Cdx. Although Cdx and CPR FMN both present electronegative surfaces to their respective redox partners, the distribution of acidic side chains is quite different, as shown in Fig.  4, C and D. The acidic residues that can directly interact with P450 are represented as green spheres. CPR FMN has a larger number of acidic residues on the docking surface. Thus, the optimal electrostatic interactions with P450cin are likely to lead to different orientations in the P450cin-Cdx and P450cin-CPR FMN complexes. One particularly noteworthy difference at the docking interface involves Glu-10 in Cdx (Gln-27 in CPR FMN ), predicted to interact with His-342 in P450cin (Fig.  4B). A second potentially important difference is Asp-87 in CPR FMN , which is Gly-66 in Cdx. Gly-66 approaches close to Gln-66 in P450cin. In addition, the sequence in the immediate vicinity of Gly-66 in Cdx is Ala-Gly 66 -Gly-Gly whereas the corresponding sequence in CPR FMN is Thr-Asp 87 -Asn-Ala. Thus, CPR FMN might experience steric clashes with P450cin in this region. It thus appears that even though the polypeptide topology is very likely conserved at the docking interface, the distribution of negative charges and specific predicted contact points are sufficiently different to give significantly different docked complexes, which translates into lower k and higher K D for CPR FMN compared with Cdx.

DISCUSSION
The present study shows that the Cdx-P450cin electron transfer reaction is specific for Cdx and proceeds at a rate similar to other known P450 systems. The reaction is highly ionic strength-dependent, which also is similar to other P450s and consistent with the electrostatic complementarity of the docking surfaces. The redox potentials of Cdx suggest that the fully reduced hydroquinone is the electron-donating species because the redox potential of the Sq/Hq Cdx couple is lower than that of substrate-bound P450cin (Ϫ226 versus Ϫ202mV, respectively). Thus, reduction of P450cin by the Cdx Hq is a thermodynamically favorable reaction. This again is similar to microsomal P450 reductase and further illustrates that P450cin is more similar to microsomal P450s with respect to electron transfer than its prokaryotic homologue, P450cam. Given such similarities with microsomal P450s, it was surprising to find that the detailed topography of the expected proximal docking site on P450cin is far more similar to P450cam than microsomal P450s or P450BM3. It thus appears that a similarly shaped docking site can exhibit large differences in specificity. Part of such selectivity is no doubt due to the obvious electrostatic difference but also due to more subtle differences in orientation and other possible allosteric effects and docking-induced structural changes. For example, the P450cam redox partner Pdx is known to have a significant effector role (24) where binding to the proximal side of the heme leads to perturbations in the distal substrate binding pocket that are thought to be coupled to conformational changes required for oxygen activation (25,26). Whether or not Cdx plays a similar effector role remains to be seen. Finally, the expected high structural homology between CPR FMN and Cdx and yet very different electron transfer parameters provide some insights into which regions of the docking surfaces are critical for electron transfer. The hypothetical model for the P450cin-Cdx complex predicts Glu-10 and Gly-66 as two regions that differ substantially from CPR FMN , which will be tested experimentally in our future studies.