Redox Properties and Coordination Structure of the Heme in the CO-sensing Transcriptional Activator CooA*

The CO-sensing transcriptional activator CooA contains a six-coordinate protoheme as a CO sensor. Cys75 and His77 are assigned to the fifth ligand of the ferric and ferrous hemes, respectively. In this study, we carried out alanine-scanning mutagenesis and EXAFS analyses to determine the coordination structure of the heme in CooA. Pro2 is thought to be the sixth ligand of the ferric and ferrous hemes in CooA, which is consistent with the crystal structure of ferrous CooA (Lanzilotta, W. N., Schuller, D. J., Thorsteinsson, M. V., Kerby, R. L., Roberts, G. P., and Poulos, T. L. (2000) Nat. Struct. Biol. 7, 876–880). CooA exhibited anomalous redox chemistry, i.e. hysteresis was observed in electrochemical redox titrations in which the observed reduction and oxidation midpoint potentials were −320 mV and −260 mV, respectively. The redox-controlled ligand exchange of the heme between Cys75 and His77 is thought to cause the difference between the reduction and oxidation midpoint potentials.

Hemeproteins are the most popular metalloproteins exhibiting a wide variety of functions such as oxygen storage/transport, electron transfer, and redox reactions of various substrates (1,2). In addition to these traditional heme proteins, a new class of heme proteins termed the heme-based sensor proteins have been reported recently in which the heme acts as a sensor of an effector molecule (3). There are five such proteins known, soluble guanylate cyclase (sGC) 1 (4 -13), FixL (14 -23), DOS (24), HemAT (25), and CooA (26 -37). These proteins contain a heme that acts as a sensing site of NO (sGC), O 2 (FixL, DOS, and HemAT), and CO (CooA).
CooA from Rhodospirillum rubrum is a heme-based COsensing transcriptional activator. Although the function of the heme in CooA is as a sensor of a gaseous effector molecule, similar to other heme-based sensor proteins, the coordination structure of the heme in CooA is quite different from that in FixL and sGC. In the case of FixL and sGC, the heme is five-coordinate with a histidine as a proximal ligand in the resting state and binds the effector, O 2 or NO, at the distal side trans to the proximal histidine (4 -23). On the other hand, the heme in CooA is six-coordinate in the ferric, ferrous, and CObound forms (26, 28, 29, 34 -37). Therefore, CO must replace one of the axial ligands in the ferrous heme when the heme in CooA binds its effector, CO (28, 29, 34 -36). The exchange of the axial ligand upon CO binding is functionally relevant for CooA, as described below.
CO is a physiological effector regulating the activity of CooA; i.e. only the CO-bound form of CooA can be bound to the target DNA and is active as a transcriptional activator (27,29,(31)(32)(33)(34)(35). Replacement of one of the axial ligands of the ferrous heme with CO triggers the activation of CooA (28, 29, 34 -36). The release of the axial ligand from the heme upon binding CO causes conformational changes around the heme and subsequently in the whole molecule (28,29,35). These conformational changes induced by the interchange of the axial ligand are the principal part of the activation of CooA by CO.
Because the heme in CooA plays a central role in sensing CO and in regulating the transcriptional activator activity of CooA, elucidation of the coordination structure of the heme in CooA is required to understand the mechanisms of CO sensing and of the activation of CooA by CO. Mutagenesis studies, EPR, resonance Raman, and uv/vis spectroscopies have revealed that the heme is in the six-coordinate and low-spin state in ferric, ferrous, and CO-bound CooA, and that Cys 75 and His 77 are the fifth ligand of the ferric and ferrous hemes in CooA, respectively (28,29,(35)(36)(37)(38). These results suggest the possibility that exchange of the axial ligand takes place between Cys 75 and His 77 during the change in the oxidation state of the heme in CooA (29,35,36).
Recently, Lanzilotta et al. (39) have reported the crystal structure of ferrous CooA, which shows that the N-terminal Pro 2 and His 77 are the axial ligands of the ferrous heme in CooA and that the Pro 2 of one subunit provides one ligand to the heme of the other subunit in the CooA homodimer (39). Although the crystal structure of ferrous CooA has been reported, the sixth ligand of the ferric heme in CooA remains to be elucidated. The proximal ligand of CO-bound CooA also remains controversial. We have proposed that His 77 is the proximal ligand of the CO-bound heme in CooA (28,29), whereas Vogel et al. (34) and Dhawan et al. (38) have pointed out the possibility that CO displaces His 77 upon CO binding to the ferrous heme in CooA.
We report herein the coordination structure of the heme in CooA revealed by alanine-scanning mutagenesis and EXAFS analyses. In this study, we also examined the redox properties of the heme in CooA by means of electrochemical redox titrations.

EXPERIMENTAL PROCEDURES
Site-directed mutagenesis was carried out by using the Quick-change site-directed mutagenesis kit (Stratagene). The expression and purification of the recombinant CooA were carried out as reported previously (26,29). A Sephacryl S-100 (Amersham Pharmacia Biotech) gel filtration column equilibrated with 50 mM Tris-HCl buffer, pH 8.0 containing 100 mM NaCl was used for the final step of the purification. The purified CooA was concentrated to about 30 M and about 2 mM by ultrafiltration with a YM-10 membrane (Amicon, Inc.) for electrochemical redox titration and EXAFS measurements, respectively.
The PCR product containing the cooA⌬N5 gene was cloned into the pCR2.1 vector using the TA cloning kit (Invitrogen) to construct pCR-CO⌬N5. An expression vector of CooA⌬N5, pKK-CO⌬N5, was constructed by inserting the EcoRI fragment containing the cooA⌬N5 gene, which was cut from pCR-CO⌬N5, into the EcoRI site of pKK223-3 (Amersham Pharmacia Biotech). The activity of CooA⌬N5 was measured as reported previously (27,29).
The electrochemical experiments were made using an electrochemical cell equipped with a quartz optical cell (40 (height) ϫ 10 (width) ϫ 1 mm (thickness)). The optical path length of the optical cell was 1 mm. A working electrode of gold mesh (40 ϫ 9 ϫ 0.7 mm) was immersed in the optical cell. A platinum wire and Ag 2ϩ /AgCl (3 M KCl) electrodes (RE-1B, BSA) were used as auxiliary and reference electrodes, respectively. The potential was controlled by a potentiostat (HA-301, Hokuto Denko Co.). The electronic absorption spectra were measured on a U-3300 Hitachi uv/vis spectrophotometer.
For electrochemical redox titrations, the following redox mediator dyes CooA solution containing the mediator dyes was repeatedly degassed and flushed with argon prior to the measurement and then ϳ2.5 ml of the sample solution was transferred into the electrochemical cell, which was sealed by a rubber septum, by using a gas-tight syringe under argon atmosphere. The electrochemical cell was kept in a thermoelectric cell holder of a U-3300 Hitachi uv/vis spectrophotometer at 15°C during the electrochemical titration. The redox reaction of CooA was followed by recording the absorbance change in the regions of the Soret and Q bands.
EXAFS spectra at the iron K-edge were measured at liquid nitrogen temperature using monochromatized synchrotron radiation at the BL 12C of the Photon Factory in the National Laboratory for High Energy Physics (Tsukuba, Japan). The EXAFS spectra were measured as fluorescence excitation spectra using a bent cylinder type focusing mirror and a silicon (111) double-crystal monochromator. The sample was placed at an angle of 45 degrees against the incident x-ray beam, and the fluorescent x-ray intensity perpendicular to the beam was measured using a solid state 19-element detector (Canberra Industries, Inc.) The analysis of the EXAFS data was performed using the program XFIT (43) as previously reported (44 -46). The k-windows used for the EXAFS analyses are shown in figures. The multiple scattering from the outershell atoms of the porphyrin ring and axial ligand molecules were taken into account. In constrained refinement, the number of parameters was reduced by treating a set of scattering atoms as a unit (47).

Mutagenesis of CooA
Alanine-scanning Mutagenesis-Mutagenesis and spectroscopic studies have revealed that the heme in CooA is sixcoordinate in the ferric, ferrous, and CO-bound forms (28 -30, 34 -36). To date, the sixth ligand of the ferric and ferrous hemes in CooA is not known, whereas Cys 75 and His 77 are assigned to be the fifth ligand of the ferric and ferrous hemes, respectively (29,35). We have reported that His, Met, Cys, and Lys are not candidates for the axial ligands of the heme in CooA, except for Cys 75 and His 77 (29). To identify the unknown ligand of the heme in CooA, we carried out alanine-scanning mutagenesis on all Arg, Asp, Glu, Asn, Gln, Ser, Thr, and Tyr residues in the heme-binding domain of CooA in the present work. These side chains could potentially coordinate to the heme iron, located in the N-terminal region from Met 1 to Met 131 (27). The mutant CooA proteins constructed in this work are summarized in Table I. The individual mutant proteins were partially purified by means of a Q-Sepharose column and their uv/vis spectra were measured in the ferric, ferrous, and CO-bound forms. All the mutants thus prepared exhibited almost the same electronic absorption spectra as did wild-type CooA (data not shown), indicating that no candidate for the sixth ligand of the heme in CooA was found by alanine-scanning mutagenesis. 2 These results suggest that side chains of any amino acid residues, except for Cys 75 and His 77 , do not coordinate to the heme iron as an axial ligand in CooA.
Truncated Mutant Lacking Four Residues in the Aminoterminal of CooA-To elucidate whether or not the amino group of the amino-terminal residue is coordinated to the heme in CooA, a truncated mutant (CooA⌬N5), in which four residues were deleted from the amino-terminal of CooA, was constructed and its electronic absorption spectra and transcriptional activator activity were measured. The amino-terminal amino acid sequences of wild-type and CooA⌬N5, which were deduced from the DNA sequences, are 1 MPPRFNIANV and 1 MNIANV, respectively. The four underlined residues in this sequence were deleted in CooA⌬N5. CooA⌬N5 showed the Soret peak at 422, 424, and 422 nm in the ferric, ferrous, and CO-bound forms, respectively, as shown in Fig. 1. The spectral features of CooA⌬N5 were almost the same as the corresponding ones in the wild-type though the ferric CooA⌬N5 showed a slightly blue-shifted and broad Soret peak with a shoulder at 399 nm compared with wild-type. The shoulder at 399 nm and a CTband at 638 nm in the spectrum of ferric CooA⌬N5 suggest that  Table stand for the positions of the residues at which the mutation was introduced. The individual residues were replaced by Ala in these mutants except for Tyr. For the Y55F and Y67F mutants, Tyr was replaced by Phe. Target residues  Mutant proteins   Arg  R4A, R21A, R24A, R51A, R53A, R61A, R87A, R91A, R96A, R118A, R125A  Asp  D15A, D40A, D49A, D72A, D94A, D129A  Glu  E17A, E38A, E41A, E59A, E60A, E62A, E83A, E86A, E89A, E99A, E128A  Asn N6A, N9A, N42A Gln Q100A, Q103A Ser S13A, S25A, S32A, S64A, S70A, S78A, S107A, S122A Thr  T18A, T36A, T69A, T85A, T88A, T97A, T104A, T121A, T126A  Tyr  Y55F, Y67F a high-spin form of the ferric heme exists to some extent in ferric CooA⌬N5. The transcriptional activator activity of CooA⌬N5 was 13.2 and 0.29 units/mg of protein in the presence or absence of CO, respectively. These values were comparable with those for wild-type CooA; the activity of wild-type CooA is 15.7 and 0.23 units/mg of protein in the presence or absence of CO, respectively (27,29). CooA⌬N5 was also a CO-dependent transcriptional activator as is wild-type CooA. The deletion of four amino acid residues from the amino-terminal of CooA did not perturb the electronic absorption spectra and CO-dependent transcriptional activator activity of CooA⌬N5. A similar result has been reported recently for the different N-terminal truncated mutants of CooA (48).
The above results suggested that the first five residues from the amino terminus are not responsible for the coordination of the heme in CooA. However, this conclusion is now known to be incorrect. During the course of the reviewing process of this paper, Lanzilotta et al. (39) have reported the crystal structure of the reduced form of CooA in which His 77 and the nitrogen atom from Pro 2 are the axial ligands of the heme (39). Though Pro 2 is deleted in CooA⌬N5, CooA⌬N5 showed almost the same uv/vis spectra and activity as those of wild-type CooA. These results indicate that some compensation for the mutation takes place to maintain CooA⌬N5 in order to have a six-coordinate heme and to be active as the CO-dependent transcriptional activator. However, it is not known at present what is coordinated to the heme in place of Pro 2 in the case of CooA⌬N5.

EXAFS Analyses of CooA
To analyze the coordination atoms of the heme iron more directly, we measured the iron K-edge EXAFS of CooA in the ferric, ferrous, and CO-bound states. The k 3 -weighted EXAFS raw data and the corresponding Fourier transforms are shown in Figs. 2 and 3, respectively. The following models for the axial ligands of the heme were used to fit the experimental data: i.e. Cys and Pro, His and Pro, and His and CO were assumed to be the axial ligands of the ferric, ferrous, and CO-bound hemes in CooA, respectively. The atomic coordinates of ferrous CooA (39) were used as a starting model for refinement of the fitting in the case of ferric and ferrous CooA. The structural parameters obtained by the fitting are shown in Table II.
The model of the coordination structure of the heme in CooA elucidated by EXAFS analyses is shown in Fig. 4. Cys 75 and the nitrogen atom from Pro 2 are thought to be coordinated to the ferric heme in CooA. The distances of the Fe-S Cys and Fe-N Pro bonds in the ferric heme were estimated to be 2.25 and 2.19 Å, respectively. In the ferrous heme, His 77 and the nitrogen atom from Pro 2 are thought to be the axial ligands. The distances of the Fe-N His and Fe-N Pro bonds in the ferrous heme were estimated to be 2.02 and 2.16 Å, respectively. Good fitting results were obtained for CO-bound CooA when His 77 was thought to be retained to be the proximal ligand of the CObound heme, as shown in Figs. 2 and 3. The distances of the Fe-N His and Fe-C CO bonds in the CO-bound CooA were estimated to be 1.98 and 1.79 Å, respectively. Details of the coordination structure of CooA in each state are discussed in the following sections.
Coordination Structure of Ferrous CooA-Lanzilotta et al. (39) have reported the crystal structure of ferrous CooA, which reveals that His 77 and Pro 2 are the axial ligands of the ferrous heme in CooA (39). As shown in Figs. 2B and 3B, the EXAFS curves can be fitted by using the atomic coordinates of ferrous CooA as a starting model for refinement of the fitting. Compared with the crystallographic result, the EXAFS analysis in this study reveals a slightly short Fe-N His and a slightly long Fe-N Pro bond distances by 0.17 and 0.04 Å, respectively. Although the bond distance of Fe-N Pro is slightly longer than that of Fe-N His , a similar bond distance for Fe-N His and Fe-N Pro does not show that one bond is more labile to be replaced with CO than the other (39). However, EXAFS and time-resolved resonance Raman spectroscopies reveal that His 77 remains coordinated to the heme when CO reacts with ferrous CooA, as described below.
Coordination Structure of the Ferrous CO Complex of CooA-EXAFS analysis of the CO-bound CooA indicated that His 77 would be the proximal ligand of the CO-bound CooA, which is consistent with our model reported previously (28,29), i.e. the mixed coordination of CO and the His 77 imidazole, as shown in Fig. 4.
Picosecond time-resolved resonance Raman spectroscopy reveals that a new intense line because of (Fe-His) at 211 cm Ϫ1 is observed immediately after photolysis of CO-bound CooA (49). The transient 211 cm Ϫ1 band is completely absent in the case of H77Y CooA in which His 77 is replaced by Tyr (49). These results are consistent with the notion that His 77 is the proximal ligand of the CO-bound CooA.
Resonance Raman spectroscopy has revealed the absence of any significant interactions between the bound CO and the distal heme pocket (28). These results indicate that the sixth ligand (Pro 2 , Ref. 39) replaced by CO moves far away beyond the coordination sphere of the heme.
CO reacts with the six-coordinate ferrous heme under physiological conditions to form CO-bound CooA, by which CooA is activated as the transcriptional activator. In other words, CO replaces one of the axial ligands of the ferrous heme to induce a conformational change required for the activation of CooA (29). The replacement of the axial ligand is functionally relevant for the activation of CooA by CO. It seems reasonable for us to anticipate that a large conformational change, including a concomitant movement of the main chain, is induced by the replacement of Pro 2 with CO. Replacement of Pro 2 will cause rotation of the C-helices about the dimer interface, as proposed by Lanzilotta et al. (39).
Coordination Structure of Ferric CooA-The mutagenesis analyses suggested that side chains of any amino acid residues, except for Cys 75 and His 77 , do not coordinate to the heme iron as an axial ligand in CooA. It has been reported that His 77 is not coordinated to the ferric heme in CooA (29,35,38). Therefore, a possible candidate for the sixth ligand would be either an exogenous molecule such as H 2 O or OH Ϫ , or the nitrogen atom from Pro 2 that is the sixth ligand of the ferrous heme (39).
Given that an exogenous molecule is coordinated to the heme as the sixth ligand, it might be H 2 O or OH Ϫ . This is mainly because the EPR spectral feature of the ferric CooA bears good resemblance to that of cytochrome P450 in the low spin state with the Cys-Fe 3ϩ -H 2 O/OH Ϫ coordination (29,37). However, the coordination of H 2 O or OH Ϫ to the ferric heme is unlikely because all the two axial ligands must be changed upon reduction of the heme, if this is the case. Furthermore, experiments on ligand binding with imidazole have shown that the heme center in CooA is remarkably resistant to exogenous ligand binding (35). The electronic absorption spectra of CooA have been reported to be unchanged between pH 6.5 and 11 (35). These results apparently indicate that the sixth ligand of the heme in CooA will be an endogenous ligand, not an exogenous one.
Dawson et al. (50) have reported spectroscopic investigations  of ferric cytochrome P450cam ligand complexes. The spectroscopic properties of ferric CooA are similar to those of the nitrogen donor complexes of ferric cytochrome P450cam rather than those of the oxygen donor complexes. Dhawan et al. (38) have reported that magnetic circular dichroism (MCD) spectroscopy reveals the coordination of a neutral nitrogen atom as the sixth ligand to the ferric heme in CooA. As shown in Figs. 2A and 3A, good fitting results were obtained for the EXAFS spectra of ferric CooA when Cys 75 and Pro 2 were assumed to be the axial ligands of the ferric heme. These results suggest that the nitrogen atom from Pro 2 is the most plausible candidate for the sixth ligand of the ferric heme in CooA.

Electrochemical Properties of CooA
Electrochemical Redox Titrations-The present mutagenesis and EXAFS experiments showed ligand switching of the heme iron in CooA coupled with the change in its oxidation state. Then, we examined the effect of the ligand switching on the redox properties of CooA by electrochemical techniques. In the presence of redox mediator dyes, electron transfer proceeded between a gold mesh electrode and CooA. In electrochemical redox titrations, the electronic absorption spectra of CooA were measured at each potential after an electric current reached Յ0.5 A. Fig. 5A shows a typical spectral change of CooA in the visible region during reductive titration of ferric CooA, when the potential was varied from Ϫ100 to Ϫ600 mV with a step width of 25 or 50 mV. The electronic absorption spectrum at a potential of Ϫ100 mV was identical to that of ferric CooA as isolated, giving the Soret peak at 423.5 nm. As the applied potential was shifted negatively, the Soret peak at 423.5 nm decreased in intensity, whereas a new Soret peak at 424.5 nm increased in intensity, with a concomitant appearance of the ␣ (558.5 nm) and ␤ (528.0 nm) bands. No spectral change was observed below Ϫ600 mV. The final spectrum was identical to the spectrum of ferrous CooA obtained by the chemical reduction of the ferric protein with sodium dithionite.
After the reductive titration, the potential was shifted positively to reoxidize CooA in the electrochemical oxidative titration. As shown in Fig. 5B, the spectral changes of CooA during the oxidative titration were the reverse of those observed in the reductive titration. At the end point of the oxidative titration, the electronic absorption spectrum of ferric CooA was restored. In the spectral change during the redox titrations, isosbestic points were observed at 411.0, 434.5, 510.0, and 569.5 nm, as shown in Fig. 5. These results show that the redox reaction of CooA proceeded quantitatively and without any side reaction in the electrochemical redox titration system described herein.
Nernst plots to determine the reduction and oxidation midpoint potentials of wild-type CooA are shown in Fig. 6A. The reduction and oxidation midpoint potentials of wild-type CooA were calculated from the plots to be Ϫ320 mV and Ϫ260 mV, respectively. In the case of wild-type CooA, the reduction and oxidation midpoint potentials were not identical; i.e. the oxidation midpoint potential was shifted positively by 60 mV compared with the reduction midpoint potential. , Ϫ350 (f), Ϫ375 (g), Ϫ400 (h), Ϫ450 (i), Ϫ500 (j), and Ϫ550 mV (versus NHE) (k). No spectral change was observed at Ϫ100 mV. The spectra measured at potentials of Ϫ600 and Ϫ650 mV were identical to spectrum k. B, the difference spectra, the electronic absorption spectra at the appropriate potential minus the spectrum at a potential of Ϫ650 mV (the spectrum of reduced CooA), are shown. The spectra were measured at potentials of Ϫ400 (a), Ϫ150 (i), and Ϫ100 mV (versus NHE) (j). No spectral change was observed at potentials of Ϫ600, Ϫ550, Ϫ500, and Ϫ450 mV.
If a simple redox reaction of the heme in CooA takes place during the electrochemical redox titrations, the midpoint potentials should be identical to each other. However, this is not the case for CooA. On the basis of the coordination structure described in this work, the reduction and oxidation reactions of the heme iron are accompanied by the ligand exchange and are thought to proceed according to the scheme as shown in Scheme 1. A similar scheme has been reported for analyzing the redox titrations of a mutant (F82H/C102S) of yeast iso-1cytochrome c (51) and of cytochrome cd1 from Paracoccus pantotrophus (52). A similar hysteresis in the redox titrations has been observed for cytochrome cd1 from Paracoccus pantotrophus (52). The oxidized Paracoccus pantotrophus cytochrome cd1 has His/His axial ligation at the c heme iron and Tyr/His axial ligation at the d1 heme iron (53,54). Upon reduction, the ligation of the c heme iron switches to His/Met concomitant with dissociation of the Tyr from the d1 heme iron (53,54). Koppenhöfer et al. (52) have reported that the hysteresis in the redox titrations is observed in the case of Paracoccus pantotrophus cytochrome cd1, because the true equilibrium is not reached for the ligand switching reactions during the redox titrations.
In the case of CooA, the change in the oxidation state of the heme iron accompanies the axial ligand switching between Cys 75 and His 77 , which is similar to ligand switching for Paracoccus pantotrophus cytochrome cd1. The reaction intermediates 2 and 4 shown in Scheme 1 seem to be formed during the reduction and oxidation of CooA, respectively, whereas the attempts to observe these intermediates by uv/vis spectroscopy with stopped flow method were unsuccessful so far. The reduction and oxidation of CooA should proceed with the same midpoint potential, if all the reaction steps shown in Scheme 1 are in equilibrium. However, this is not the case for wild-type CooA as shown in Fig. 6A, i.e. the apparent reduction and oxidation midpoint potentials are different from each other. Because the equilibrium is not reached for the ligand switching reactions during the redox titrations, the hysteresis in the redox titrations is thought to be observed also in the case of CooA, as is the case for Paracoccus pantotrophus cytochrome cd1.
To support the above suggestion, we also examined the redox titrations of the H77G mutant of CooA in which His 77 , which is one of the axial ligands of the ferrous heme, is replaced by Gly. The results of the electrochemical redox titrations for H77G CooA are shown in Fig. 6B. The redox properties of H77G CooA were significantly different from those of wild-type CooA in which the reduction and oxidation midpoint potentials (Ϫ420 mV) were identical to each other. Because H77G CooA lacks His 77 , only redox reactions of the heme iron seem to take place during the redox titrations. The ligand switching during the redox titrations, the vertical transitions in Scheme 1, does not take place in the case of H77G CooA, which is the reason why the hysteresis in the redox titrations was not observed for H77G CooA.
Redox Properties of CooA Relevant to its Functional and Structural Properties-The reduction midpoint potential of ferric CooA (Ϫ320 mV) is comparable with that of the low-spin form of Pseudomonas putida d-camphor hydroxylase cytochrome P450cam (E1 ⁄2 ϭ Ϫ300 mV, Ref. 55), whose axial ligands are a thiolate derived from Cys and H 2 O/OH Ϫ . This observation is consistent with the coordination of the Cys 75 thiolate to the ferric heme iron in CooA, because the low reduction potential is thought to be caused by the strong electron-donating nature of the Cys thiolate (56). The thiolate coordination is apparently supported by EPR results, in which the spectral features of FIG. 6. Reductive (OE) and oxidative (q) titrations of wild-type (A) and H77G CooA (B). The fraction of the ferric form was plotted as a function of potential. The individual data points are an average of at least three independent measurements. The solid lines are theoretical Nernst curves for one-electron reduction (OE) and oxidation (q) with midpoint potentials of Ϫ320 and Ϫ260 mV for wild-type CooA and with those of Ϫ420 mV for H77G CooA, respectively. SCHEME 1.
ferric CooA are similar to those of the low-spin form of P450cam (29,37).
It is also characteristic of CooA that the heme having the histidyl imidazole as the axial ligand gives Ϫ260mV for the oxidation midpoint potential. Compared with the redox potentials of other heme proteins having a histidyl imidazole, the value is similar to that of horseradish peroxidase (Ϫ250 mV, Ref. 57) rather than those of myoglobin (ϩ50 mV, Refs. 58, 59)) and cytochrome b562 (ϩ113 mV, Ref. 60). The difference in the redox potentials has been explained in terms of the difference in the character of the ligand imidazole group; i.e. the imidazole ring of peroxidase has an anionic character, resulting from the strong hydrogen bond of the NH with carboxylate (61), whereas those in myoglobin and cytochrome b562 have a neutral character (58,59,62). In the case of CooA, deprotonation of Pro 2 that is coordinated to the ferrous heme may be responsible for the low oxidation potential of CooA.
The low oxidation potential of CooA could be concerned with the regulation of CooA activation. CooA is required for CO-dependent expression of the coo operons encoding a CO-oxidizing system that allows R. rubrum to grow on CO as the sole energy source (31). CooA is activated only in the presence of CO under anaerobic conditions to induce the expression of the coo operons (30,32,33). Even in the presence of CO, however, CooA must not be activated under aerobic conditions because CO dehydrogenase, a key enzyme for the CO-oxidizing system encoded in the coo operons, is labile to oxygen (63). The low oxidation potential of CooA would facilitate the oxidation of the heme to prevent CooA from being activated in vivo, once oxygen is present in the cells.