Critical Role of the Heme Axial Ligand, Met95, in Locking Catalysis of the Phosphodiesterase from Escherichia coli (Ec DOS) toward Cyclic diGMP*

Heme-regulated phosphodiesterase from Escherichia coli (Ec DOS) is a gas-sensor enzyme that hydrolyzes cyclic dinucleotide-GMP, and it is activated by O2 or CO binding to the Fe(II) heme. In contrast to other well known heme-regulated gas-sensor enzymes or proteins, Ec DOS is not specific for a single gas ligand. Because Arg97 in the heme distal side in Ec DOS interacts with the O2 molecule and Met95 serves as the axial ligand on the distal side of the Fe(II) heme-bound PAS domain of Ec DOS, we explored the effect of mutating these residues on the activity and gas specificity of Ec DOS. We found that R97A, R97I, and R97E mutations do not significantly affect regulation of the phosphodiesterase activities of the Fe(II)-CO and Fe(II)-NO complexes. The phosphodiesterase activities of the Fe(II)-O2 complexes of the mutants could not be detected due to rapid autoxidation and/or low affinity for O2. In contrast, the activities even of the gas-free M95A and M95L mutants were similar to that of the gas-activated wild-type enzyme. Interestingly, the activity of the M95H mutant was partially activated by O2, CO, and NO. Spectroscopic analysis indicated that the Fe(II) heme is in the 5-coordinated high-spin state in the M95A and M95L mutants but that in the M95H mutant, like wild-type Ec DOS, it is in the 6-coordinated low-spin state. These results suggest that Met95 coordination to the Fe(II) heme is critical for locking the system and that global structural changes around Met95 caused by the binding of the external ligands or mutations at Met95 releases the catalytic lock and activates catalysis.

Heme-regulated phosphodiesterase from E. coli (Ec DOS) contains a heme-bound PAS domain in the N-terminal region and a phosphodiesterase domain with GGDEF and EAL subdomains in the C-terminal region (11,12). Although Ec DOS also contains a GGDEF subdomain, it does not appear to have guanylate cyclase activity. Thus, the precise roles of the GGDEF domain in the catalysis and structure of Ec DOS remain unclear. In contrast, the EAL domain in Ec DOS appears to contain the catalytic domain, so that this domain retains high phosphodiesterase activity that rapidly converts c-diGMP into linear dinucleotide GMP (l-diGMP) (3,4). Furthermore, E. coli with a knock-out of the Ec DOS gene display a morphology and growth rate different from those of the wild-type strain, suggesting that the enzyme is important for cell development and/or proliferation (13).
In well known heme-bound gas-sensor proteins, binding of specific gases to the Fe(II) heme (for example, CO for CooA, NO for soluble guanylate cyclase, and O 2 for FixL) causes a change in the heme environment that is transduced intramolecularly to the other domain to regulate the function (14). Each of these heme-bound gas-sensor enzymes can recognize only a single physiologically relevant gas. We recently reported, however, that the phosphodiesterase activity of full-length, dithionite-reduced Ec DOS is up-regulated upon binding of either O 2 or CO binding to the Fe(II) heme (15). Although the amino acid sequence and overall structure of the heme-containing PAS domains of Ec DOS and FixL are similar, Ec DOS is activated by both O 2 and CO, whereas FixL is down-regulated only by O 2 . Thus, Ec DOS appears to be a novel heme-bound gas-sensor enzyme with the unprecedented ability to recognize more than one type of gas molecule.
To understand why, in contrast to other gas-sensor heme proteins, Ec DOS lacks strict gas selectivity, the roles of the residues around the heme in catalytic activation must be determined. O 2 bound to the Fe(II) heme interacts directly with Arg 97 on the heme distal side (Fig. 1A) (16). Thus, Arg 97 appears to be important for recognizing different gas molecules. Met 95 binds directly to the Fe(II) heme as the endogenous axial ligand (Fig. 1B) (17), and the coordination of Met 95 to the Fe(II) heme is cleaved by binding of the external axial ligands O 2 , CO, and NO. Thus, both Arg 97 and Met 95 may play important roles in the regulation of catalysis by gas binding. To explore the role of these two residues, we generated full-length Ec DOS with mutations at Arg 97 and Met 95 , and examined their catalytic activities and activation by O 2 , CO, and NO. As in studies of SmFixL (18), in the current experiments, we measured the catalytic activity in the presence of DTT rather than sodium dithionite to more accurately reflect the physiological conditions in vivo. It was found that mutation of Arg 97 did not significantly affect the catalytic activation by gases, whereas the mutation of Met 95 led to high enzyme activity even in the absence of gases as well an inability to be activated by gas binding. Our results imply that Met 95 plays a critical role in the activation of c-diGMP hydrolysis upon gas binding by Ec DOS.
Construction of Expression Plasmid and Site-directed Mutagenesis-The cloning of Ec DOS and construction of the expression plasmid pET28a(ϩ) were previously described (19). Mutants were generated as described in the supplementary materials. All constructs were confirmed by DNA sequencing.
Protein Overexpression and Purification-The wild-type and mutant Ec DOS proteins were overexpressed in E. coli BL21(DE3) using the pET28a(ϩ) expression plasmid, which contains an N-terminal His 6 tag and a thrombin cleavage site. A single colony was put in 50 ml of Luria-Bertani (LB) medium containing 20 g/ml kanamycin and 0.5% glucose and then shaken overnight at 200 rpm and 37°C. The cultured medium was added to 1 liter of terrific broth (TB) medium containing 20 g/ml kanamycin. Then, the medium was shaken for 4 h at 120 rpm and 37°C. After the medium was cooled down to 20°C, ␦-aminolevulinic acid was added to 0.45 mM and incubated with shaking for another 20 h. The E. coli cells were harvested by centrifugation for 10 min at 5000 ϫ g and 4°C, and finally stored at Ϫ80°C.
The following protein purification procedures were performed on ice or at 4°C. E. coli cells frozen at Ϫ80°C were suspended in 50 mM potassium phosphate buffer (pH 7.5) containing 20 mM imidazole, 150 mM NaCl, 5% glycerol, 1 mM phenylmethylsulfonyl fluoride (buffer A), and 0.1 mg/ml lysozyme. The solution was sonicated and then centrifuged at 100,000 ϫ g for 30 min. The supernatants were applied to a nickel-nitrilotriacetic acid-agarose column. The Ec DOS fractions were eluted with a linear gradient of 20 -300 mM imidazole in buffer A. The solution containing Ec DOS was dialyzed against 20 mM Tris-HCl (pH 8.0) with 5% glycerol (buffer B), and then thrombin protease (Wako Chemicals, Osaka, Japan) (1 unit/mg of protein) was added to cleave the His tag. Subsequently, the sample solution was applied to a nickel-nitrilotriacetic acid-agarose column, and the His tag-free Ec DOS was eluted in the flow-through fractions using buffer B. The Ec DOS fractions were then applied to DEAE column, and Ec DOS was eluted with a linear gradient of 0 -1 M NaCl in buffer B. After the proteins were dialyzed against buffer B, the purified Ec DOS was concentrated with Centriprep (Millipore, Billerica, MA) and Centrisalt I (Sartorius AG, Goettigen, Germany). The purified proteins were Ͼ95% homogenous as confirmed by SDS-PAGE. The sample was stored at Ϫ20°C in the presence of 40% glycerol.
Optical Absorption Spectroscopy-All spectral data were obtained using a UV-1600PC (Shimadzu, Kyoto, Japan) spectrophotometer in a glove box that can maintain the oxygen concentration under 100 ppm (Hirasawa, Tokyo, Japan). The heme of Ec DOS proteins was reduced using either 10 mM DTT or small quantities of sodium dithionite in PDE buffer (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, and 5 mM MgCl 2 ) under anaerobic conditions. The Fe(II) heme, Fe(II) heme-O 2 , and Fe(II) heme-CO complexes were prepared in anoxic (10% H 2 and 90% N 2 gas), O 2 -saturated, or CO-saturated buffer, respectively. The gas-saturated buffer solutions were obtained by bubbling the buffer with the appropriate gas for at least 1 h. The Fe(II) heme-NO complexes were prepared by adding of 50 or 100 M NOC9 to the solution of Fe(II) heme complex.
Phosphodiesterase Assay-Ec DOS hydrolyzes c-diGMP to l-diGMP as demonstrated by reversed-phase liquid chromatography (supplementary materials Fig. 1S). On the other hand, in our phosphodiesterase assay, we used a colorimetric method, namely, l-diGMP that is converted to a guanylyl (3Ј,5Ј)guanosine by Ec DOS. CIAP reacts with only l-diGMP because CIAP lacks phosphodiesterase activity and thus reacts only with terminal phosphate groups. CIAP therefore converts l-diGMP to GpG (guanylyl (3Ј,5Ј)guanosine) and phosphate. One product, phosphate, was quantified by colorimetrically using BIOMOL GREEN reagent (Biomol). The change in absorbance at 630 nm (supplementary materials Fig. 2S, A) was measured with a Benchmark Microplate Reader TM (Bio-Rad) and used to determine phosphate concentration by comparison with a standard curve (supplementary materials Fig. 2S, B).
All steps were performed in the globe box. Wild-type and mutant Ec DOS proteins were diluted in PDE buffer. The wildtype and Arg 97 mutant enzymes were fully reduced to the Fe(II) form by adding 10 mM DTT. For the Met 95 mutants, a trace amount of sodium dithionite was used to completely reduce the heme iron to the Fe(II) complex because DTT cannot reduce the heme iron in the Met 95 mutant proteins. Subsequently, excess dithionite was removed by gel filtration on Superdex G-25. After complete reduction the Fe(II) form was confirmed by optical absorption spectroscopy, Fe(II)-CO and Fe(II)-O 2 complexes were prepared by diluting the sample in gas-saturated PDE buffer containing 10 mM DTT. The Fe(II)-NO sample was prepared by adding 50 M or 100 M NOC9 to the sample of Fe(II) enzyme. The

RESULTS
Optical Absorption Spectra of the Mutants-We first collected optical absorption spectra for the gas-free and gas-bound Fe(II) heme complexes of the R97A, R97I, and R97E mutants of full-length Ec DOS (Fig. 2). The absorption maxima of the mutant proteins are summarized in Table 1. The optical absorption spectra of these Arg 97 mutants in the absence or presence of CO and NO were essentially the same as for the wild-type protein, suggesting that the coordination structure (6-coordinated low-spin form) and the heme environment were not substantially changed by the Arg 97 mutations. The spectra of the Fe(II)-O 2 complexes for the Arg 97 mutants as measured with a conventional absorption spectrometer, however, could not be detected due to rapid autoxidation (Ͼ10 s Ϫ1 ) and/or a low affinity for O 2 (data not shown). It is likely, therefore, that Arg 97 is important for stabilizing the Fe(II)-O 2 complex.
Optical absorption spectra of the Met 95 mutants of the fulllength enzyme for the gas-free and gas-bound Fe(II) heme complexes were essentially the same as those previously reported for the mutants of the isolated heme-bound PAS domain (20 -22). Note that both the M95A and M95L mutants were 5-coordinate high-spin complexes, whereas the M95H mutant was a 6-coordinate low-spin complex in which His must be an axial ligand.
In the remaining experiments, we reduced the wild-type and Arg 97 mutant enzymes to the Fe(II) heme state using DTT, whereas we reduced the Met 95 mutant proteins with sodium Gas-responsive Heme-regulated Phosphatase JULY 20, 2007 • VOLUME 282 • NUMBER 29 dithionite. It was, therefore, examined whether there were differences in the optical absorption spectra of the Fe(II) heme complexes when the enzymes were reduced with DTT and sodium dithionite. We found that the spectra of the Fe(II) heme complexes were essentially the same, even if the heme iron was reduced by DTT or sodium dithionite, suggesting that the heme environment of the full-length enzyme is not influenced by the reducing agent.
Phosphodiesterase Activity of the Wild-type Enzyme-Activities of other heme-bound oxygen-sensor enzymes, such as SmFixL, in the presence of DTT are much more sensitive to oxygen and appear to reflect the activity of the native enzyme under hypoxic conditions (18). In the present study, therefore, we used 10 mM DTT to reduce the wild-type enzyme, and we measured the catalytic activity in the glove box to provide anoxic conditions. High pressure liquid chromatography (as shown in supplementary materials Fig. 1S) is used in general to measure phosphodiesterase cleavage of c-diGMP. However, this method is time consuming. In this report, we use colorimetric detection of products in the development of a rapid c-diGMP cleavage assay ("Experimental Procedures" and supplementary materials Fig. 2S).  Table 2. The binding of O 2 , CO, or NO to the heme enhanced the activity 6 -7fold. The finding that even NO binding markedly enhanced the catalytic activity was surprising because other heme gas-sensor proteins such as FixL, CooA, and soluble guanylate cyclase strictly distinguish O 2 , CO, and NO, respectively (14). This emphasizes that Ec DOS is a novel gas-sensor enzyme that has the unprecedented ability to be activated by all three of these gases.
Effects of Arg 97 Mutations on the Phosphodiesterase Activity-The optical absorption spectra of the Arg 97 mutants in the absence or presence of CO and NO were similar to those of the wild-type enzyme ( Table 1), suggesting that the heme environments of the mutants and wild-type enzyme are essentially the same. Therefore, we examined the activities of the mutant enzymes in the presence of CO or NO. It was found that CO and NO binding markedly enhanced (6.5-13fold) the catalytic activity of the three Arg 97 mutants, although in the absence of these gases, the activities of the R97A and R97E mutants were lower than that of the wildtype enzyme ( Table 2 and supplementary materials Fig. 1S). Because the autoxidation of the O 2 -bound Arg 97 mutant proteins was very rapid (Ͼ10 s Ϫ1 ) and/or the affinities of the Fe(II) heme complexes for O 2 were low, we could not measure the catalytic activity of the O 2 -bound Arg 97 mutant enzymes.
Effects of Met 95 Mutations on the Phosphodiesterase Activity-Addition of up to 100 mM DTT was not sufficient to fully reduce the Met 95 mutants due to the relatively low redox potential of the heme iron in the mutant proteins. This was also observed for the same mutants of the isolated PAS domain (21). Therefore, we used a trace amount of sodium dithionite to completely reduce the heme to the Fe(II) complex. After removing excess

TABLE 1 Optical absorption spectral maxima (nm) of the Fe(II) complexes for the wild-type, Arg 97 mutant, and Met 95 mutant proteins in the absence and presence of O 2 , CO, and NO
Coordination structures and spin states of the heme iron are shown in parentheses. sodium dithionite by gel filtration on Sephadex G-25, we measured the catalytic activities of the Met 95 mutant proteins in the same buffer containing 10 mM DTT. Interestingly, the gas-free M95A and M95L mutants showed high activities without gas binding; specifically, the activities of these two mutants were 7-8-fold higher than that of the gasfree wild-type enzyme (Table 1 and supplementary materials Fig. 2S). The activities of the gas-bound bound forms of the mutants were almost as high as the gas-free forms ( Table 2). In other words, the catalytic activities of the mutant enzymes were not enhanced by gas binding. In contrast, the activity of the gasfree M95H mutant was 2-fold higher than that of the wild-type enzyme, and the gas molecules enhanced its activity 2-3-fold.

DISCUSSION
The PAS fold is characterized by several ␣-helices flanking a five-stranded antiparallel ␤-sheet scaffold (17, 29 -33). PAS domains are found in many signaling proteins, where they are used either as signal sensor domains or for protein dimerization (29 -32). Several PAS domain proteins are known to detect signals through the use of prosthetic groups such as heme or flavin (30,31). Heme-based PAS sensor proteins detect oxygen, CO, and possibly the cellular redox state (14,31).
Comparison of the crystal structures of the oxygen-free and oxygen-bound PAS domains of Ec DOS (16,17) shows that Arg 97 undergoes a large change in position when oxygen binding occurs. It might thus be expected that Arg 97 should play a prominent role in DOS function. Notably, however, our study clearly shows that this is not the case. The R97A and R97I mutants have activity profiles similar to that of the wild-type enzyme. On the other hand, the M95A and M95L mutants are clearly constitutively "on." In the absence of Met 95 , the PAS domain presumably assumes a conformation similar to that of the gas-bound protein. Consistent with this idea, the M95H mutant, with a histidine bound to the heme, exhibits properties similar to those of the wild-type enzyme. These data strongly suggest that changes in Met 95 coordination structure caused by gas binding (mimicked by mutations at Met 95 ) result in larger conformational changes at the distal side, which subsequently communicate ligand binding information to the catalytic domain.
Effects of Sodium Dithionite and DTT on the Phosphodiesterase Activity of Ec DOS-Although the optical absorption spectra of the Fe(II) complexes for the wild-type and Arg 97 mutant proteins were essentially the same when reduced by DTT or sodium dithionite (Fig. 2 and Table 1), the phosphodiesterase activities of the DTT-reduced enzymes were at least 5-fold higher than those of the dithionite-reduced enzymes (Fig. 3 and supplementary materials Fig. 6S). Similar results have been reported for other heme-bound oxygen-sensor enzymes, including the histidine kinase SmFixL (18). The low activity of the dithionite-reduced forms of SmFixL were thought to be due to the formation of an aberrant disulfide bond at Cys 301 between Fe(III) homodimers during the preparation of samples under aerobic conditions in the absence of reducing agents. In SmFixL treated with DTT, the disulfide bond was cleaved, leading to more enhancing the enzyme activity.
The wild-type and some mutants of Ec DOS are sometimes purified as the Fe(II)-O 2 form (data not shown), suggesting that Ec DOS is present as the Fe(II) form in cells. In general, however, the Fe(III) form along with minor components of Fe(II) and Fe(II)-O 2 forms are obtained when Ec DOS is purified under aerobic conditions. Ec DOS contains eight cysteine residues in each monomer. It is possible, therefore, that some cysteine residues in Ec DOS are already oxidized and form disulfide bond(s), leading to lower activity. The following results support the idea that dispensable oxidation caused partial inactivation of Ec DOS. When DTT was removed by G-25 column chromatography, DTT-reduced Ec DOS showed low phosphodiesterase activities for Fe(II), Fe(II)-O 2 , and Fe(II)-NO complexes, but showed high activity only for the Fe(II)-CO complex (supplementary materials Fig. 5S). O 2 and NO molecules are oxidants, and thus, may oxidize cysteine residue(s) or cleave disulfide bonds in Ec DOS, partially inactivating the phosphodiesterase activity.
Role of Arg 97 in the Catalysis-The Arg 97 mutants autoxidized rapidly (Ͼ10 s Ϫ1 ) and/or had low affinities for oxygen. Therefore, Arg 97 must be important for stabilizing the O 2 molecule bound to the Fe(II) heme by directly binding to the O 2 molecule on the heme distal side (16). In contrast, optical absorption spectra suggest that Arg 97 is not important for stabilizing the CO and NO molecules bound to the Fe(II) heme. The enhancement of catalytic activity by CO and NO was similar for the Arg 97 mutants and the wild-type enzyme. Therefore, Arg 97 must not be critical for the enhancement of catalysis by CO and NO.
The structure of the heme-bound sensor PAS domains of SmFixL and Ec DOS are similar. Arg 214 in SmFixL, which cor-  (24,25). The optical absorption spectra suggest that Arg 97 in Ec DOS stabilizes the Fe(II)-O 2 form and that the interaction of Arg 97 with the gas ligand may not be important for the activation of catalysis by ligand binding (Table 2 and Fig. 4). It has been suggested that specific steric and/or electrostatic effects destabilize the Fe(II)-O 2 complex (20,28). In general, an increase in accessibility of the distal pocket to either solvent water molecules or to other polar residues, or protonation of the O 2 molecule bound to the Fe(II) heme, accelerate autoxidation of the Fe(II)-O 2 complex. The increase in autoxidation rate seen in the Arg 97 mutants suggests that Arg 97 mutations lead to increases in interaction of polar molecules with the Fe(II)-O 2 complex. These findings contrast with data from tests with M95A and M95L mutants, where the mutations significantly decreased the autoxidation rate (20). Isolation of the PAS domain also decreased the autoxidation rate (20).
Role of Met 95 in the Catalysis-Met 95 is the sixth axial ligand in the gas-free Fe(II) complex. Met 95 must move away from the heme iron upon binding of exogenous ligand to Fe(II) heme (Fig. 1). Previous studies indicate that mutations at Met 95 in the isolated heme-bound PAS domain of Ec DOS substantially increase the affinities of the heme iron for O 2 , CO, and CN Ϫ and that the heme redox potential is also markedly changed by Met 95 mutations (20,21,26). The role of Met 95 in catalytic regulation, however, has been unclear.
In the present study, we found that the activities of the gasfree M95A and M95L mutant enzymes were already similar to those of the gas-bound wild-type enzymes (Table 2). Therefore, it appears that Met 95 is critical for regulating the catalytic activity of Ec DOS. In the gas-free form, the M95A and M95L mutants are 5-coordinate high-spin complexes, in contrast to the wild-type and Arg 97 mutant enzymes, which are 6-coordinate low-spin complexes. In the wild-type and Arg 97 mutant enzymes, cleavage of the Met 95 coordination to the Fe(II) heme occurs upon gas binding. On the other hand, in the M95A and M95L mutants, as in the gas-bound wild-type and Arg 97 mutant proteins, there is no bond between the residue at position 95 and the Fe(II) heme iron. Certain structural change(s) near the main or side chain of Met 95 on the heme distal side must occur upon binding of the gas ligand. There may be a similar structure(s) in the Fe(II) complex of the M95A and M95L mutants even in the absence of the gas molecules, which lead to the high activities of the gas-free Fe(II) heme enzymes ( Table 2 and Fig. 4).
It is interesting that the M95H mutant is the 6-coordinate low-spin complex, probably as a result of His 95 coordination to the Fe(II) heme iron. The catalytic activity of the gas-free form of this mutant was lower than those of other Met 95 mutants that are 5-coordiante high-spin complexes. On the other hand, gas binding by this mutant enhanced its activity 2-3-fold, which is higher than that observed for other Met 95 mutant proteins. Perhaps a certain conformational change(s) caused by the M95H mutation influences the activity of the gas-free M95H mutant enzyme, leading to activities intermediate between the wild-type/Arg 97 and M95A/M95L mutants. Collectively, the results suggest that Met 95 coordination to the Fe(II) heme is a key factor in the down-regulation of catalysis so that its removal by gas binding or the M95A or M95L mutations enhances the catalytic activity.
Comparison with Catalysis of Ec DOS toward cAMP-Recent studies of the cAMP phosphodiesterase activity of Ec DOS indicate that cAMP is hydrolyzed by the Fe(II) heme form but not the Fe(III) form of Ec DOS (19). Furthermore, binding of CO or NO to the Fe(II) heme complex eliminates the ability to hydrolyze cAMP. X-ray crystal structural studies of the isolated heme-bound PAS domain of Ec DOS indicate that profound structural changes of the heme-bound PAS domain occur upon changes in the redox state (17). Such redox-dependent structural changes appear to reflect the redox-dependent regulation in the cAMP phosphodiesterase activity of Ec DOS. Furthermore, knocking down the Ec DOS gene in E. coli causes changes in cell growth rate, cell morphology, and intracellular cAMP concentration, suggesting that Ec DOS plays an important role in modulating the cellular effects of cAMP (13).
In the present study, we showed that, in addition to cAMP, Ec DOS hydrolyzes c-diGMP. The mechanism by which c-diGMP hydrolysis is regulated differs from that regulating cAMP hydrolysis. Namely, the c-diGMP phosphodiesterase activity of Ec DOS is substantially up-regulated by the binding of gas mol- ecules. Furthermore, it is important to note that, unlike other heme-bound gas-sensor enzymes and proteins (14), Ec DOS does not discriminate between O 2 , CO, and NO. Further studies are needed to understand the mechanisms of gas sensing and intramolecular signal transduction by this novel enzyme. It will also be interesting to investigate the intracellular cross-talk between the two substrates of Ec DOS, cAMP and c-diGMP (27).