Redox-mediated transcriptional activation in a CooA variant.

CooA, the carbon monoxide-sensing transcription factor from Rhodospirillum rubrum, binds CO at a reduced (Fe(II)) heme moiety with resulting conformational changes that promote DNA binding. In this study, we report a variant of CooA, M124R, that is active in transcriptional activation in a redox-dependent manner. Where wild-type CooA is active only in the Fe(II) + CO form, M124R CooA is active in both Fe(II) + CO and Fe(III) forms. Analysis of the pH dependence of the activity of Fe(III) M124R CooA demonstrated that the activity was also coordination state-dependent with a five-coordinate, high-spin species identified as the active form and Cys(75) as the retained ligand. In contrast, the active Fe(II) + CO forms of both wild-type and M124R CooA are six-coordinate and low-spin with a protein ligand other than Cys(75), so that WT and Fe(III) M124R CooA are apparently achieving an active conformation despite two different heme coordination and ligation states. A hypothesis to explain these results is proposed. This study demonstrates the utility of CooA as a model system for the isolation of functionally interesting heme proteins.

The role of small gas molecules (nitric oxide, oxygen, and carbon monoxide) in biological cell-signaling systems has attracted considerable interest in the elucidation and characterization of the respective protein receptors involved in the transduction of these signals (1,2). For the sensing of NO, soluble guanylyl cyclase serves as the receptor and has been investigated extensively (3,4). Soluble guanylyl cyclase is mildly activated upon binding CO (ϳ5% compared with NO; Ref. 4), although this activity can be increased by the presence of the indazole derivative YC-1 (5,6). Combined with recent in vivo studies (7,8), a common view is that soluble guanylyl cyclase does indeed sense and become activated by CO but only under conditions where a physiological co-effector is present. However, the identity of a specific CO-sensing protein in eukaryotic systems has yet to be demonstrated conclusively. Oxygen sensors include FixL, which is involved in anaerobic control of nitrogen fixation in Rhizobium meliloti (9,10), and the Escherichia coli protein named "direct oxygen sensor" (DOS) the function of which remains unclear (11). Finally, CooA, a transcriptional activator in the bacterium Rhodospirillum rubrum, is the first protein described that has a specific physiological role in CO-sensing and CO-mediated signal transduction (12).
All of the above gas-sensing proteins have in common a heme b prosthetic group that is directly involved in binding the effector, which leads to a conformational change in the protein that affects activity.
Upon binding CO, the homodimeric CooA undergoes a conformational change that allows it to bind to its cognate promoters in the coo regulon (13). The coo regulon contains two transcriptional operons whose polypeptide products are required for R. rubrum to anaerobically oxidize CO to CO 2 with concomitant H ϩ reduction to H 2 (14). CooA belongs to a family of transcriptional activators that include the cAMP receptor protein (CRP 1 ; Ref. 15) and the fumarate and nitrate reductase activator protein (FNR; Ref. 16). Recently the three-dimensional structure of the effector (CO)-free form of CooA has been solved (17), and its comparison to the three-dimensional structure of the effector (cAMP)-bound form of CRP (18) has indicated the conformational changes that take place upon effectordriven activation in this protein family. In addition, the structure of CooA indicated an unprecedented ligation arrangement for a heme protein wherein the N-terminal proline (Pro 2 , from the opposite subunit) and His 77 serve as the heme-axial ligands in the Fe II form. Interestingly a redox-mediated ligand exchange occurs in CooA (from Cys 75 in the Fe III form to His 77 in the Fe II form; Fig. 1) upon reduction of the heme iron (19,20). Thus, CooA functions as both a CO sensor and a redox sensor because only the Fe II form is competent to bind CO. The heme of WT CooA is hexacoordinate and low-spin in all oxidation and ligation states (21), indicating that incoming CO must displace one of the internal protein ligands. Picosecond timeresolved resonance Raman spectroscopy (22) and NMR studies (23) have indicated that His 77 is the retained ligand in the Fe II ϩ CO form of CooA. A study of CooA variants showed that alteration of Pro 2 did not significantly affect CooA activity; this is also consistent with the hypothesis of the retention of His 77 in the Fe II ϩ CO form (24).
Similar to that of effector-bound CRP, the structure of effector-free CooA indicated that a single ␣-helix (designated the C-helix) in each monomer serves to create the intersubunit dimerization domain (18). In CRP and FNR, alterations of particular amino acids in the center of this helix have a variety of effects on activity (25)(26)(27). In this report, we have identified CooA variants altered in the C-helix that have a perturbed ligation structure in the oxidized (Fe III ) form. The Fe III form of one of these variants can also bind specific target DNA and is competent in transcriptional activation. This novel activity can be modulated by the redox state of the heme iron, demonstrating the ability of CooA to serve as a model system for engineering unique sensing capabilities in heme proteins.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-The construction of strains bearing WT CooA and CooA variants in an E. coli overexpression system and in a ␤-galactosidase reporter system (19) have been described previously. Variants randomized in their amino acid sequences at selected locations in CooA were generated as described previously (24) or by modified oligonucleotide-directed polymerase chain reaction (28).
Purification of WT CooA and CooA Variants-The purification of WT CooA (to Ͼ95% homogeneity) was performed as described previously (19). Because of poor accumulation in the E. coli expression system, M124R CooA was purified to ϳ80% homogeneity. We also isolated M124R CooA under anaerobic conditions in the presence of sodium dithionite as described by Shelver et al. (19) but observed little improvement in the yield of heme-containing CooA. When necessary, aliquots of the reduced anaerobically isolated M124R CooA were chemically oxidized with a slight excess of potassium ferricyanide, which was then removed by G-25 gel filtration. The heme content of CooA preparations was quantified using the reduced pyridine-hemochromogen assay (29), and the protein content was measured using the BCA assay (Pierce).
In Vivo Activity Assays-In this study, we examined the in vivo activity of a CooA-dependent ␤-galactosidase reporter under different culture regimens that were designed to promote a particular oxidation state of the CooA present. Strains containing a promoter of cooF gene (P cooF )-lacZ reporter fusion (19) were grown in 2ϫ Luria broth with 0.9% (w/v) NaCl containing 100 g/ml ampicillin. For assays of aerobic activity, inocula (250 l) were diluted into 10 ml of Luria broth with 0.45% (w/v) NaCl containing 100 g/ml ampicillin and 25 M isopropyl-␤-D-thiogalactopyranoside in 250-ml flat-bottomed flasks. Aerobic cultures were grown with vigorous shaking at 30°C for ϳ12 h to mid to late log phase. For assays with and without CO, 20-ml cultures were grown in MOPS-buffered medium (24) in 120-ml stoppered serum vials made anaerobic on a nitrogen manifold. For CO induction of cultures, CO gas was added to the headspace to a final concentration of 2%, and 500 l of inocula were injected. Anaerobic cultures were then incubated at 30°C with gentle lateral shaking for ϳ12 h to mid to late log phase. Whole-cell ␤-galactosidase assays were then performed as described previously (30).
In Vitro Activity Assays-In vitro DNA binding assays were performed using the fluorescence polarization technique described previously (24,31). In this study, however, we also measured the DNA binding activity of the Fe III (as isolated aerobically or chemically oxidized), Fe II (as isolated anaerobically or chemically reduced with dithionite), and Fe II -CO CooA forms. The binding data were fitted by nonlinear regression using the equation (corrected for sample quenching) described by Lundblad (31). The influence of pH dependence on activity of our CooA preparations was measured in a solution consisting of CaCl 2 (6 mM), KCl (50 mM), glycerol (5% (v/v)), dithiothreitol (1 mM), and different buffering components (pH 8.0, 8.5, and 9.0; 20 mM Tris-HCl and pH 9.5, 10.0, and 10.5; 20 mM glycine/NaOH). The pK a of the activity transition (see "Results") was calculated by fitting the pH titration data to the Henderson-Hasselbach equation assuming a single ionization. Different preparations of the labeled P cooF showed slightly different baseline values (no protein added) in the polarization assay, although the total change in polarization at saturation (ϳ0.030 -0.035 anisotropy units) was roughly constant.
UV-Visible Absorption Spectroscopy-UV-visible absorption spectroscopy of CooA samples was performed at room temperature in quartz cuvettes sealed with rubber septa using a Shimadzu UV-PC2100 spectrophotometer (slit width ϭ 0.5 nm). pH titrations of isolated M124R CooA were performed in the following buffers (100 mM and containing 5% glycerol) at these pH values: pH 8.0, 8.5, and 9.0, Tris-HCl; pH 9.5, 10.0, and 10.5, glycine/NaOH. The pK a of the spectral transition was calculated by fitting the titration data to the Henderson-Hasselbach equation assuming a single ionization.
Electron Paramagnetic Resonance Spectroscopy-Isolated M124R Fe III CooA was first concentrated to ϳ200 M (in heme) using ultrafil-tration under nitrogen gas pressure in an Amicon cell fitted with a Diaflo YM-10 membrane. The protein (200 l) was then mixed with 50 l of 500 mM buffer (pH 7.5, MOPS and pH 10.5, glycine/NaOH) to provide a final buffer concentration equal to 100 mM and a heme concentration of ϳ160 M. Samples were then quickly degassed on an argon manifold to remove dissolved oxygen, frozen, and stored at 77 K. The EPR spectra were then recorded as described previously (24). Analysis of theoretical g values was performed using a rhombogram computer program (Rhombo Version 1.0; see Appendix in Ref. 32).

RESULTS
In Vivo ␤-Galactosidase Assay-Following random mutagenesis of that portion of cooA encoding the Ser 122 -Cys 123 -Met 124 residues, mutagenized clones were screened in a strain of E. coli in which CooA regulates lacZ expression (19). Some clones were noted that expressed ␤-galactosidase under aerobic growth conditions; this was not a property of clones expressing WT CooA. Similar to WT, however, these clones also showed activity under anaerobic conditions only when CO was present. Upon sequencing, these clones contained sequences that created M124R or M124K substitutions. The results with colonies were supported by quantitative ␤-galactosidase activity assays. Cells with WT CooA have ϳ1-3% activity under aerobic or anaerobic conditions compared with anaerobic conditions in the presence of CO. Cells with M124R CooA have activity similar to that of WT CooA when grown anaerobically in the presence of CO but substantially higher activity aerobically (ϳ30%) and anaerobically (ϳ13%). Cells with M124K CooA displayed aerobic effector-independent activity severalfold lower than did cells with M124R CooA, and cells with M124L, M124A, and M124I CooA were indistinguishable from those with WT CooA aerobically with slight differences anaerobically in the presence of CO (data not shown). To test whether M124R CooA was unique in its substantial activity under aerobic conditions, we mutagenically randomized the 124 position as we have done previously (24) and screened for aerobic expression of ␤-galactosidase. Variants M124R, M124K, and M124I were repeatedly isolated, although quantitative analysis of strains with M124K and M124I showed only low activity aerobically indicating that only the Arg substitution at position 124 has a substantial effect. Interestingly, P2Y and P2H CooA, which have similar spectral features in the Fe III form to that of M124R CooA (see below), showed no aerobic activity and only slightly perturbed activity in the presence of CO (24). It is important to note that CooA is in large excess in these in vivo assays, so that levels of specific activity below that of WT can allow maximal reporter activity. The in vitro characterization (see below) allows the manipulation of CooA levels.
Isolation of M124R CooA and Analysis of DNA Binding in Vitro-To understand the biochemical basis for the "aerobically active" phenotype, we isolated M124R and M124K CooA using the procedures described above. Because M124R CooA consistently showed a higher amount of aerobic activity when compared with that of M124K CooA, we chose to extensively characterize M124R CooA. Neither M124R nor M124K CooA accumulated well in the E. coli expression system (ϳ5-10% of WT CooA). Although the isolated material was ϳ80% pure, the BCA protein assay combined with the heme assay revealed  (19) and, based on indirect evidence, by Pro 2 (24). Upon reduction of the heme iron to the Fe II form, Cys 75 is replaced by His 77 . Based on a variety of evidence, CO binding to the heme displaces Pro 2 (22)(23)(24). that our M124R CooA preparations contained only 0.2 hemes/ dimer compared with 1.6 hemes/dimer for the isolated WT protein (21). M124R CooA was also unstable in dilute solutions but was stabilized by addition of glycerol to 5-10% (v/v). Diluting stock Fe III M124R CooA into a solution at pH 7.0 caused instantaneous precipitation, which is not seen in WT CooA. Therefore, all manipulations were carried out at pH 8.0 in the presence of 5% glycerol. Although we were still unable to isolate M124R CooA containing greater than 0.2 hemes/dimer, we will demonstrate in this study that the ability of M124R CooA to bind DNA is both redox-and CO-specific, which is consistent with heme dependence.
To determine whether the Fe III M124R CooA activity was detectable in vitro, we tested isolated M124R and WT CooA in different redox states using the fluorescence polarization assay for CooA-DNA interactions. At saturating (40 g/ml protein; 800 nM heme) concentrations of protein, Fe III WT CooA showed a basal value of anisotropy in this particular experiment (Fig.  2). Dithionite reduction to the Fe II form had no effect in anisotropy, whereas addition of CO to the Fe II WT CooA generated a large increase in the value of anisotropy as a result of CooA-DNA interaction. When an equivalent protein concentration (40 g/ml protein; 80 nM heme) of Fe III M124R CooA in the Fe III form was added to the system, a large increase in the value of anisotropy was evident compared with Fe III WT CooA, indicating Fe III M124R CooA-DNA interaction. However, upon dithionite reduction to the Fe II form, M124R CooA less effectively interacted with the DNA target. Finally, when CO was added to the sample, M124R CooA had an in vitro activity similar to that of WT CooA, indicating that with both Fe II ϩ CO samples the level of functional CooA had saturated the target DNA. These results indicate both a redox-dependent activity and a CO-dependent activity in M124R CooA.
To address the heme concentration dependence of M124R CooA-DNA interactions, we measured the fluorescence polarization of samples that contained increasing amounts of M124R CooA on a heme basis (Fig. 3). When based on heme-containing M124R CooA, Fe III M124R CooA (at pH 8.0) showed an increase in anisotropy reflecting a K D of ϳ90 nM. This value is approximately 5 times greater than the K D for Fe II ϩ CO M124R CooA, which was ϳ20 nM. In comparison, Fe II ϩ CO WT CooA has a K D of ϳ10 nM (Fig. 3), which is a value similar to that found in a previous study (24). Therefore, the CO-specific DNA binding of WT and M124R CooA to target DNA is roughly similar. The anisotropy value for Fe II proteins was near baseline in this particular experiment but tended to increase at the higher concentrations used in this assay (data not shown) indicating the presence of nonspecific interactions at moderately high protein levels (for WT CooA, Ͼ500 nM heme ϭ 25 g/ml protein).
Fe III M124R CooA Has Unusual Coordination Properties-To determine whether the active phenotype of M124R CooA is correlated with structural perturbations near the heme center, we examined isolated M124R CooA using UV-visible absorption spectroscopy. Fig. 4A shows the UV-visible absorption spectra of Fe III M124R CooA compared with that of Fe III WT CooA. Significant spectral changes were observed in Fe III M124R CooA (Soret ϳ387 nm with a strong ligand-to-metal charge transfer band ϳ640 nm) when compared with that of Fe III WT CooA (Soret ϳ424 nm with a very weak ligand-tometal charge transfer band ϳ650 nm) and are consistent with a high-spin, five-coordinate heme in Fe III M124R CooA. In contrast, the spectra of the Fe II (Fig. 4B) and Fe II ϩ CO (Fig.  4C) forms of M124R CooA were indistinguishable from those of the Fe II and Fe II ϩ CO forms of WT CooA, indicating that only the Fe III form of M124R CooA is significantly perturbed by this substitution. The five-coordinate, high-spin spectral signature of Fe III M124R CooA was similar to that observed for Fe III P2Y CooA (24). However, at pH 7.4, there was a higher percentage of the five-coordinate form in Fe III M124R CooA than that observed in Fe III P2Y CooA (24), indicating that their heme environments are somewhat different at identical pH.
In Fe III M124R CooA, the five-coordinate heme indicates that an open coordination should exist for exogenous ligands to bind. However, addition of cyanide, azide, or imidazole to ϳ1000-fold molar excess produced no significant changes in the UV-visible absorption spectrum of Fe III M124R CooA (data not shown), suggesting either electrostatic repulsion of these ligands or a relatively inaccessible heme iron. This result is in FIG. 2. M124R CooA binds DNA in the Fe III form and is modulated by redox. An in vitro DNA binding assay using fluorescence polarization at pH 8.0 is shown. In this experiment, target DNA ϭ 6.4 nM. Samples were equal in protein concentration (40 g/ml) but contained different amounts of heme (see text). For WT, the heme concentration was ϳ800 nM, and for M124R, the heme concentration was ϳ80 nM. Error bars represent data from triplicate measurements. contrast to that of Fe III P2Y CooA, which is also five-coordinate and high-spin yet binds imidazole albeit with a relatively low affinity (24). The differential ability of Fe III M124R CooA and Fe III P2Y CooA to bind exogenous ligands implies significant differences in their respective heme environments (see below).
pH Dependence of Fe III M124R CooA Spectra and in Vitro Activity-Because the UV-visible absorption spectrum of Fe III P2Y CooA, which contains a fraction of five-coordinate, highspin heme, exhibited a pH dependence (24), we examined whether Fe III M124R CooA had similar behavior. The UVvisible absorption spectra of Fe III M124R CooA as a function of buffer pH are shown in Fig. 5A. At pH 8.0, Fe III M124R CooA exhibited spectral features that indicated a spin mix of the five-coordinate, high-spin form (Soret peak at 387 nm) and a shoulder at 423 nm, indicating the presence of the six-coordinate, low-spin form (contrast with Fig. 4A at pH 7.4). Increasing the pH of the solution increased the fraction of the sixcoordinate, low-spin form with a clear isosbestic point at ϳ410 nm (Fig. 5A). At pH 10.5, the UV-visible absorption spectrum of Fe III M124R CooA was essentially identical to that of Fe III WT CooA at pH 7.4 (Fig. 4A). Fitting the ABS 387 nm (the Soret maximum of the five-coordinate, high-spin form) to the Henderson-Hasselbach equation and assuming a single ionization gave a pK a of ϳ9.0 (Fig. 5B). This result suggests that hydroxide may be the sixth ligand in six-coordinate, low-spin Fe III M124R CooA at strongly basic pH.
Because Fe III M124R CooA undergoes a spectral transition with pH, we explored the possibility that the redox-mediated activity might also be influenced by pH. Fig. 5B shows the change in anisotropy as a function of pH and indicates that at a more basic pH Fe III M124R CooA becomes less active and is essentially incapable of binding target DNA at pH 10.5. This result indicates that not only is the DNA-binding ability of Fe III M124R CooA redox-dependent, but it also exhibits a coordination-state correlation as well. The pK a of the transition for DNA binding activity was ϳ8.9, which is essentially identical to that found in the spectral transition (Fig. 5B). This result indicates that a five-coordinate, high-spin fraction of Fe III M124R CooA represents the species that is active in binding DNA. No pHdependent effects (pH 8.0 -10.5) were observed in the Fe II M124R CooA (inactive) and Fe II ϩ CO M124R CooA (active) samples tested (data not shown). Under the same varied pH conditions, there was a slight decrease (ϳ10%) in the in vitro activity of Fe II ϩ CO WT CooA at the higher pH range (data not shown).
EPR Spectroscopy of Fe III M124R CooA-EPR spectroscopy of Fe III M124R CooA corroborated the results of the UV-visible absorption spectroscopy. An increase in the pH of the solution resulted in a dramatic decrease in the intensity of high-spin forms of Fe III M124 R CooA and a concomitant increase in the intensity of the low-spin form (Fig. 6). Inspection of the lowfield region of the spectrum of Fe III M124R CooA revealed the presence of two distinct S ϭ 5/2 systems. For the first spin system, analysis of the g values using a rhombogram computer program (32) for an S ϭ 5/2 system with g ϭ 2 indicates that the spectrum is composed of a system with theoretical g values of g z ϭ 1.84 (assumed; see below), g y ϭ 4.30, and g x ϭ 7.53 (Table I) and are similar to those that are found in Fe III fivecoordinate, high-spin thiolate-ligated hemes such as P-450cam (33) and the hypothesized arrangement of Fe III P2Y CooA (24). For the second spin system, theoretical g values of g z ϭ 1.94 (assumed; see below), g y ϭ 4.99, and g x ϭ 6.96 (Table I) are similar to those found in Fe III five-coordinate, high-spin histidine-ligated hemes such as FixL (34) and soluble guanylyl cyclase (35). Concerning the former spin system, the signal at g y ϭ 4.3 does not change significantly with pH, suggesting that a majority of this signal is arising from non-heme iron. Because Cys 75 is the normal ligand in Fe III WT CooA, it is very likely that this residue is the source of the thiolate signal. The origin of the neutral nitrogen signal remains unclear, although Pro 2 is a possible candidate.
The high-spin system has unusual relaxation properties in that it is only observable at relatively high power and very low temperature (200 microwatts and 4 K). Under these conditions, the low-spin (S ϭ 1/2) features are saturated and appear as a dispersion line shape that dominated the high-field signals arising from the S ϭ 5/2 system (data not shown). At lower powers and higher temperature (20 -50 microwatts, 23 K), the low-spin system can be observed as well defined derivativeshaped features, although the high-spin features are completely unobservable under these conditions. Therefore, the g z values from the high-field region (S ϭ 5/2) systems can only be assumed from the rhombogram analysis.
The six-coordinate, low-spin system exists in equilibrium with the two five-coordinate, high-spin systems and is evident even at near neutral pH (Fig. 6). This signal increases with pH and represents a thiolate/strong field ligation based on the g values (g x ϭ 1.90, g y ϭ 2.26, and g z ϭ 2.45) that are identical to those of WT CooA (Table I). We note that there appears to be a higher proportion of low-spin signal in the EPR spectrum compared with that seen in the UV-visible absorption spectrum at similar pH values, and we propose that this reflects the very different temperatures at which these experiments were performed. Apparently the low temperature stabilizes the low-spin form for reasons that are presently unknown. The identity of the ligand trans to thiolate could be either H 2 O or hydroxide, which is observed in P-450 (36), or presumably some nitrogen donor ligand (perhaps Pro 2 ) provided by another residue in Fe III M124R CooA. DISCUSSION Residues in Regions Homologous to Position 124 of CooA Are Important for Activity in the CRP Superfamily-In the case of both CRP and FNR, there are different alterations that can cause effector-independent activity, but the region homologous to Cys 123 and Met 124 in CooA is particularly interesting. In FNR, substitutions at Asp 154 (Asp 154 is homologous to Cys 123 in CooA) render the protein insensitive to redox, which is the primary effector for this protein. Normally FNR exists as a monomer under oxidizing conditions, and it is only under reducing conditions after the formation of Fe 4 S 4 centers in each monomer that FNR can dimerize and thus bind palindromic DNA sequences with high affinity. The D154A substitution causes the protein to more readily dimerize under all conditions, although there is no evidence that it lies particularly close to the Fe 4 S 4 center (25). The mechanism by which cluster formation leads to dimerization is a topic of central importance to understanding FNR action, but there is no reason to believe that it has any mechanistic similarities to CO binding to the heme of CooA. CRP is more similar to CooA in that it is a dimer under all conditions, and the binding of its effector, cAMP, leads to a conformational change that is presumably somewhat similar to that of CooA. Ser 128 of CRP, which is homologous with Met 124 in CooA, makes contact with the cAMP molecule bound to the other protein monomer (18). An S128P substitution severely reduces the ability of CRP to bind DNA in response to cAMP (26), and an S128T substitution results in a lack of discrimination between cAMP from cGMP as effectors (37). In addition, the binding of the cAMP to S128A CRP occurs with magnified negative cooperativity compared with that of WT CRP (26).
It seems paradoxical that these three proteins, with what appear to be completely different sites and mechanisms for effector recognition and response, should all be so strikingly affected by changes in this region of the protein. As noted below, it is our working hypothesis (17) that repositioning of the C-helices upon CO binding is important for activation of WT CooA, and preliminary results on alterations of the leucine zipper motif (38) along these helices are generally consistent with that hypothesis. 2 However, the effects described for M124R CooA do not appear to be based on this specific mechanism because the M124R substitution should not have a dramatic effect on the leucine zipper motif. We therefore favor the hypothesis that the M124R substitution exerts its effects by a different mechanism of communication between the heme and the DNA binding domains; it is unclear whether this pathway is also present in WT CooA as well.
Ligation Arrangement of Fe III M124R CooA-The results of the UV-visible and EPR spectroscopy in this paper show that a portion of Fe III M124R CooA has a heme coordination that most closely resembles the five-coordinate, high-spin heme observed for P-450cam (33), and the correlation between pK a for the coordination state change and pK a for activity strongly implies 2 R. L. Kerby, unpublished data.  that it is one or both of the five-coordinate, high-spin forms that are active. The EPR analysis also suggests that the majority of this species has a thiolate ligand that we assume to be Cys 75 because that is the normal ligand to the Fe III form of WT CooA. It would therefore be tempting to conclude that the M124R substitution somehow causes the loss of the sixth protein ligand (Pro 2 ) to yield the five-coordinate Fe III form with Cys 75 as the protein ligand and that this form would be necessary and sufficient for the observed activity. This hypothesis is incorrect, however, as P2Y and P2H variants also have a substantial fraction of five-coordinate, high-spin heme in the Fe III form, and at least a portion of this appears to have a thiolate ligand (24), yet these proteins display almost no activity in vivo or in vitro under oxidizing conditions (data not shown). Indeed, the P2Y and P2H variants are not significantly affected in their response to CO (24), indicating that the loss of the "Pro 2 arm" of the N terminus (17) is not critical for the CO-dependent activation of WT CooA. We also note that the g values of these five-coordinate forms of M124R and P2Y CooA are not identical (Table I), and the difference in imidazole binding between these two proteins indicates some difference in their respective heme environments. It is therefore our working hypothesis that although both these variants have a five-coordinate Cys-ligated heme in the Fe III form, the heme position is different in a subtle but functionally important way as described below.
Structural Consequences of the M124R Substitution-The recently published crystal structure of Fe II WT CooA (17) provides a framework to form hypotheses about why M124R CooA might be active in the Fe III form. It must be emphasized, however, that the published crystal structure of WT CooA is in the Fe II form, which has a completely different ligation and coordination state than Fe III M124R CooA and is inactive in DNA binding. Nevertheless, there are a few salient possibilities that arise from inspection of the published structure.
Inspection of the Fe II WT CooA structure in Fig. 7 shows the orientations of the Met 124 residues in relation to the hemes. Using the SwissPro structure-viewing program to analyze M124R gives a variety of different energetically reasonable conformers of the "new" amino acid. Two favored configurations have the guanidino cation of Arg 124 hydrogen-bonded to either the 7-proprionate of the heme or to the peptide chain carbonyl of Ser 78 . The former of these would have the potential of perturbing the heme itself, perhaps destabilizing the Pro 2 ligand in the process, whereas the latter is immediately adjacent to the His 77 and Cys 75 region and possibly perturbs the heme directly or indirectly through that interaction. The current working hypothesis of CooA function (17) is that CO binding leads to a shift in heme position such that it now interacts with the nearby portion of the C-helices, and it is the repositioning of these helices with respect to each other that induces the conformational change at the opposite end of the protein. Consistent with this hypothesis, we assume that activity in the Fe III form of M124R is caused by an interaction of the heme with the C-helices that is somewhat similar to that of the CO-bound heme of WT CooA. However, because the heme must itself be in a somewhat different position (it is still ligated to Cys 75 rather than the expected His 77 of Fe II WT CooA), the heme-helix interactions for activity are certainly unlike the normal pathway in the WT protein; they simply have the same end effect. We rationalize the failure of P2Y and P2H to have similar activity because they presumably have the Fe III heme in a slightly different position albeit in a roughly similar ligation state. The inability of Fe III M124R CooA (in contrast in Fe III P2Y) to bind imidazole is consistent with this steric association of the heme with the C-helices.
In summary, we have demonstrated that M124R CooA is active in transcriptional activation in the Fe III form and that this activity is dependent upon both redox and coordination state of the heme. The M124R substitution certainly perturbs ) are shown as well as the relative positions of the Met 124 residues. The A monomer is colored in light gray, whereas the B monomer is colored in dark gray. The DNA binding domain, which in this view is above the C-helices, is not shown. The B monomer (dark gray, left side of panel) shows the view from solvent looking toward the interior of CooA. The model was generated using the freeware program Web Lab Viewer Lite. In the perspective shown, the closest approach of Met 124 of the A monomer to the heme-7-proprionate of the B monomer is ϳ6.5 Å. the heme ligation and probably the heme position, which ultimately leads to this novel phenotype. The results strongly support the notion of extreme flexibility in the heme-protein contacts of CooA as suggested by the analysis of the crystal structure (17). This study and another on cyanide binding variants of CooA (39) demonstrate that unique sensing capabilities can be engineered into CooA and should encourage efforts to search for other variants of CooA that are responsive to other heme-site ligands.