EPR Characterization of Axial Bond in Metal Center of Native and Cobalt-substituted Guanylate Cyclase*

The nature of the metal-proximal base bond of soluble guanylate cyclase from bovine lung was examined by EPR spectroscopy. When the ferrous enzyme was mixed with NO, a new species was transiently produced and rapidly converted to a five-coordinate ferrous NO complex. The new species exhibited the EPR signal of six-coordinate ferrous NO complex with a feature of histidine-ligated heme. The histidine ligation was further examined by using the cobalt protoporphyrin IX-substituted enzyme. The Co2+-substituted enzyme exhibited EPR signals of a broad g⊥;1 component and a g∥;1 component with a poorly resolved triplet of 14N superhyperfine splittings, which was indicative of the histidine ligation. These EPR features were analogous to those of α-subunits of Co2+-hemoglobin in tense state, showing a tension on the iron-histidine bond of the enzyme. The binding of NO to the Co2+-enzyme markedly stimulated the cGMP production by forming the five-coordinate NO complex. We found that N3 − elicited the activation of the ferric enzyme by yielding five-coordinate high spin N3 − heme. These results indicated that the activation of the enzymes was initiated by NO binding to the metals and proceeded via breaking of the metal-histidine bonds, and suggested that the iron-histidine bond in the ferric enzyme heme was broken by N3 −binding.

Soluble guanylate cyclase (sGC), 1 a hemoprotein, catalyzes the conversion of GTP to cGMP. The enzyme is a major receptor for NO in cell-cell signal transduction pathways such as neuronal communication and vasodilation (1)(2)(3)(4)(5)(6)(7). NO binds to the heme iron of the enzyme and markedly activates the cyclase reaction (8 -12). The purified enzymes from rat and bovine lung were a heterodimer consisting of two similar but not identical subunits (12)(13)(14). Although the enzyme has been believed to contain 1 protoheme IX/heterodimer (12,(15)(16)(17), Stone and Marletta found recently the heme stoichiometry of 1.5 hemes/ heterodimer (18). Based on the result, they argued that the stoichiometry was actually 2 hemes/heterodimer, with each subunit binding one equivalent of heme at a homologous site between two subunits (18). In this context, it is noted that a site-directed mutagenesis study raises a new issue for the heme coordination (19). This mutagenesis study aimed at the identification of a heme axial ligand strongly suggested that the binding domain of protoheme IX located only in the smaller subunit, the ␤-subunit, and that the histidine residue at the 105-position of the ␤-subunit was an axial ligand of the enzyme heme. The ligation of the histidine residue at the 105-position to the heme iron was confirmed by a site-directed mutagenesis study, in which the ␤-subunit fragment consisting of residues 1-385 contained a stoichiometric amount of heme (20), while the ␤-subunit fragment with H105A mutation was heme-deficient (21). Although these results demonstrated that the histidine residue conserved only among the ␤-subunits was the heme binding site, there was no evidence for the heme binding site in the ␣-subunit.
Stone and Marletta (12,22) reported that the ferric and the ferrous enzyme hemes were both in a five-coordinate high spin state. Furthermore, the ferrous NO complex was demonstrated to be the active form of the enzyme with five-coordinate NO heme (12,23,24). Evidence for the formation of five-coordinate NO complex was also obtained by using partially purified, reconstituted enzyme preparations (25,26). Hence, the breaking of the heme-proximal ligand bond upon NO binding to the enzyme heme was proposed to be a trigger for the activation of the enzyme, as has been hypothesized by Traylor and Sharma (27). The spectroscopic finding for the ligation of a histidine residue at the proximal position has been obtained by a resonance Raman study (24). Deinum et al. (24) assigned the 204cm Ϫ1 Raman band to the iron-histidine stretching vibration based on analogy with that of other hemoproteins. This result was a first demonstration for the weak iron-histidine bond in sGC, but the assignment of the Fe-histidine vibration was incomplete for lack of an isotope shift experiment such as 57 Fe substitution.
Although a resonance Raman spectroscopy was an important spectroscopic probe for the analysis of properties of the ironproximal base bond, the use was usually limited to ferrous high spin enzyme heme. In contrast, an EPR technique provided structural information for the metal-axial base bond of six-coordinate ferrous NO hemoproteins (28 -32) and of Co 2ϩ porphyrin-substituted derivatives of hemoproteins (33)(34)(35)(36)(37). The Co 2ϩ derivatives that were proved to retain close structural homology to the native iron hemoproteins by x-ray crystallography (38,39) were frequently used to provide EPR response instead of EPR-silent ferrous hemoproteins. In the present work, we aimed to identify a histidine residue as the proximal ligand of the heme and to examine the nature of the metalhistidine bond of sGC by an EPR method. To take advantage of a EPR spectroscopy as a structural probe, we have prepared the six-coordinate NO complex and the Co 2ϩ -substituted derivative of sGC. The EPR findings provided firm spectroscopic evidence for the histidine ligation at the proximal position of metal. Moreover, we found that the EPR features of the NO complex and the Co 2ϩ derivative closely resembled those of the corresponding form of ␣-subunits of hemoglobin, showing the presence of tension on the iron-histidine bond. We also obtained EPR evidence that the N 3 Ϫ binding to the ferric enzyme heme formed presumably a five-coordinate high spin N 3 Ϫ heme.

EXPERIMENTAL PROCEDURES
Enzyme Purification-Fresh bovine lung (4 kg) was minced and homogenized with a Waring blender in 12 liters of 50 mM TEA buffer, pH 7.6, containing 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 1 mM EDTA, and 55 mM mercaptoethanol (buffer A). Throughout the purification, these protease inhibitors and mercaptoethanol were included in the buffer unless otherwise stated. After the homogenate was centrifuged at 13,500 ϫ g for 20 min, 1.2 kg of DEAE-cellulose A-500 (Seikagaku Kogyo) equilibrated with buffer A was added to the supernatant. The slurry was stirred for 1 h at 4°C, and the resin was collected by sedimentation. Subsequently, the resin was washed three times with buffer A and was poured into a column. The enzyme was eluted with a 3.5-liter linear NaCl gradient of 0 -0.35 M in buffer A. After active fractions were pooled, solid ammonium sulfate (0.29 g/ml) was added. The precipitate collected by centrifugation at 10,000 ϫ g for 15 min was dissolved in 200 ml of 40 mM potassium phosphate buffer, pH 7.6, containing 55 mM mercaptoethanol. The sample was washed with the phosphate buffer by using a Minitan system (Millipore Corp.) to remove ammonium sulfate. Then the enzyme was applied to the column of Matrex Blue A (Amicon) equilibrated with 40 mM phosphate buffer described above and eluted with a linear gradient of 0 -1 M KCl gradient. The pooled enzyme was applied to a ceramic hydroxylapatite column (Bio-Rad). The enzyme was eluted by increasing phosphate concentration from 0 to 0.45 M at pH 7.6 containing 55 mM mercaptoethanol and the protease inhibitors except for EDTA. The concentrated sample was further purified with a Superdex 200-pg HPLC column (Amersham Pharmacia Biotech). Then fractions with the cyclase activity were applied to an 80-ml column of GTP-Sepharose 4B with a 12-atom spacer attached through ribose hydroxyl (40). The column was exhaustively washed to remove contaminated proteins with 25 mM Tricine-NaOH buffer, pH 7.6. The enzyme was eluted with a 1-liter gradient running from 0 to 0. 15 M NaCl. The fractions with a specific activity over 8000 nmol/min/mg of protein in the presence of NO were pooled. Then the sample was finally purified to an apparently homogenous state with a Protein Pak G-DEAE HPLC column (Waters). The overall yield was about 10%. The resultant homogenous enzyme was stored in 50 mM TEA buffer, pH 7.6, containing 10% glycerol and 5 mM DTT at Ϫ80°C until use.
Co 2ϩ Protoporphyrin IX Substitution-Ignarro et al. (17) have reported a method to prepare the apoenzyme by lowering pH to 5.7. We attempted to prepare the apoenzyme by this method, but the recovery of the Co 2ϩ -substituted enzyme was very low at the final purification step. We tested the heme depletion as a function of pH and found that the heme in sGC was depleted by the DEAE cellulose chromatography under alkaline conditions. In brief, the supernatant fraction of homogenized tissue described above was adjusted to pH 8.5, and DEAE cellulose A-500 equilibrated with 50 mM TEA buffer at pH 8.5 containing protease inhibitors was poured to the supernatant. The enzyme was eluted by a linear gradient of 0 -0.35 M NaCl. The fractions with cyclase activity that was assayed in the presence of protoporphyrin IX were further purified by GTP-agarose and Superdex 200-pg columns under the conditions described above. The apoenzyme was pooled and reconstituted with Co 2ϩ protoporphyrin IX under anaerobic conditions. The remaining purification steps were the same as those used for the native enzyme purification. The enzyme Co 2ϩ porphyrin-substituted by our method exhibited essentially the same optical and EPR spectral properties as the Co 2ϩ -substituted enzyme, which was obtained by the method of Ignarro et al. (17).
Spectral Measurements-Absorption spectra were recorded with a Shimadzu MPS-2000 or a Perkin-Elmer Lamda 18 spectrophotometer at room and subzero temperatures. The temperature of the cuvette holder was controlled with thermomodule elements. The buffer systems used were 50 mM TEA buffer (pH 7.6) containing 5% glycerol and the same buffer containing 40% ethylene glycol for room and subzero temperature measurements, respectively. Other details were described in the figure legends.
EPR spectra were measured on a Varian E-12 X-band EPR spectrometer with 100-kHz field modulation. An Oxford flow cryostat (ESR-900) was used for liquid helium temperature measurements. The microwave frequency was calibrated with a microwave frequency counter (Takeda Riken, model TR 5212), and the magnetic field strength was determined by the nuclear magnetic resonance of water protons. Accuracy of g values was Ϯ0.01 in the low magnetic field and Ϯ0.005 in the high field. Other details were as described elsewhere (41).
NO complexes for EPR measurements were prepared in buffer containing 5% glycerol at Ϫ5°C or in buffer containing 40% ethylene glycol at Ϫ24°C as follows. The enzyme solution was transferred to a septumcapped EPR tube and flushed with oxygen-free argon gas for 10 min. Then NO gas previously washed with 1 N NaOH or an aliquot of SNAP solution was introduced to the tube with a gas-tight syringe. The formations of NO complexes in five-and six-coordinate states were ensured by directly measuring the optical spectrum of the sample in the EPR tube at Ϫ5 or Ϫ24°C.
The ferric enzyme was prepared by adding a 2-fold excess of ferricyanide to the DTT-free ferrous enzyme, where DTT in the enzyme solution was removed by a Superdex 200HR (Amersham Pharmacia Biotech) HPLC column. For EPR measurements, the residual ferricyanide was freed of the solution by passing through a Superdex 200HR HPLC column. The ferric enzyme was converted to N 3 Ϫ complexes by adding a desired amount of NaN 3 . The EPR spectra of ferric N 3 Ϫ complex were measured at 5 or 15 K.
Kinetic Measurements-The NO binding to the ferrous enzyme was analyzed by a Photal stopped flow spectrophotometer, model RA-401, equipped with a photodiode array detector. The buffer solution in reservoirs was bubbled with oxygen-free argon for 10 min, and then the catalytic amount of glucose oxidase and catalase and 2 mM glucose was added to assure anaerobic conditions. The enzyme and an aliquot of NO-saturated solution were then added to the buffer solution under a constant stream of argon.
Resonance Raman Measurements-The resonance Raman spectra were measured with a JASCO NR-1800 spectrometer equipped with a liquid nitrogen-cooled CCD detector (Princeton Instruments). Excitation wavelengths were 413.1-and 406.7-nm lines from a Krypton ion laser (Coherent, Innova 90). Calibration of the Raman spectrometer was performed by using indene.
Activity Measurements-The enzyme activity during the purification was measured in a reaction mixture containing 2 mM GTP, 5 mM DTT, 3 mM MgCl 2 , and an appropriate amount of the enzyme solution in a total volume of 0.5 ml of 40 mM TEA buffer, pH 7.4. When desired, 1 mM isobutylmethylxantine was added to inhibit phosphodiesterase activity. The reaction was started by the addition of 0.2 mM SNAP and conducted at 37°C for 10 min. The reaction was terminated by the addition of 20 l of 30% acetic acid. The mixture was centrifuged for 10 min at 15,000 rpm, and cGMP was quantitated with a C18 HPLC column at a constant flow rate of 1 ml/min of 40 mM potassium phosphate buffer, pH 6.0, containing 10% methanol.
The activity of the homogenous enzyme was assayed under the same condition as that described above, except that the concentration of GTP was increased to 4 mM. The activation by NO was performed in a septum-capped sample tube. The assay mixture in the tube was flushed with a purified argon, and the reaction was started by the addition of 30 l of saturated NO solution with a gas-tight syringe. For the activation by CO, the reaction mixture was saturated with CO gas prior to the addition of the enzyme.
Electrophoresis-Reducing SDS-polyacrylamide gel electrophoresis was carried out by using 9% acrylamide running gel. Protein was visualized with a silver stain method (Daiichi Chemical Co.).
Reagents-GTP and cGMP were purchased from Wako Pure Chemical Industries. Research grade NO was obtained from Takachiho Chemical Co. S-Nitroso-N-acetyl-DL-penicillamine was purchased from DOJINDO or ALEXIS. Other chemicals, purchased from Nacalai Tesque Co., were of highest commercial grade and were used without further purification.

RESULTS
Properties of Native Enzyme-The homogenous enzyme exhibited a basal activity of 98 nmol/min/mg of protein at 37°C in the presence of Mg 2ϩ . The activity increased to 26,811 nmol/ min/mg of protein upon the addition of NO, while it increased to 650 nmol/min/mg of protein upon the addition of CO. An addition of protoporphyrin IX (2.4 M) slightly increased the activity (580 nmol/min/mg of protein). The enzyme preparation stimulated by a combined addition of protoporphyrin and NO exhibited an activity (25,560 nmol/min/mg of protein) rather lower than the NO-stimulated activity. The SDS-polyacrylamide gel electrophoresis analyses indicated that the enzyme was a heterodimeric protein consisting of an ␣-subunit of 75 kDa and a ␤-subunit of 71 kDa. The enzyme contained 0.97 Ϯ 0.04 protoheme IX/heterodimer, in which protein was determined by the modified biuret method of Yonetani (42) using bovine serum albumin as the standard, and the heme content was determined by the pyridine hemochromogen method (43). When protein was determined by the Bradford protein assay, essentially the same heme stoichiometry was obtained (0.95 Ϯ 0.03). The cyclase activity per heme, defined as turnover number (mol of cGMP/min/mol of heme) was 3800 min Ϫ1 at 37°C in the presence of NO.
Detection and Characterization of Six-coordinate Ferrous NO Complex-The ferric enzyme exhibited a Soret band at 390 nm (⑀ mM 103), which was indicative of the five-coordinate high spin state (data not shown). The ferrous enzyme, prepared by adding a slight excess of Na 2 S 2 O 4 to the ferric enzyme under anaerobic conditions, exhibited the Soret maximum at 431 nm (⑀ mM 105). The optical spectrum of the enzyme reduced by Na 2 S 2 O 4 was identical with that of the ferrous enzyme as isolated in the presence of mercaptoethanol. An anaerobic addition of SNAP or NO to the ferrous enzyme yielded the fivecoordinate NO complex with the Soret maximum at 400 nm (⑀ mM 75). These optical and EPR spectral properties of the ferric enzyme (see Figs. 6 and 8) agreed with previous results (18,20) but significantly differed from the result obtained by using a partially purified reconstituted enzyme (25). Deinum et al. (24) pointed out that the partially purified reconstituted enzyme preparation had a different heme environment from the native enzyme.
When the ferrous enzyme was mixed with SNAP at Ϫ24°C in the presence of 40% ethylene glycol used as an antifreeze, a new spectral species with a sharp Soret band at 419 nm and 544-and 579-nm bands in the visible region was produced (Fig.  1A). The peak positions closely agreed with those of the sixcoordinate NO complex of hemoglobin (44), suggesting that the new species was six-coordinate ferrous NO complex. By raising the temperature to Ϫ15°C, the species fully converted to the five-coordinate ferrous NO complex with 400-nm Soret maximum, giving clear isosbestic points (Fig. 1A, inset).
The coordination state of the new species was examined by an EPR method at 35 K. When the new species, prepared in EPR tube at pH 7.6 and Ϫ24°C, was immediately frozen by immersing into liquid nitrogen, the EPR signal of the new species was negligibly small in the spectrum, where the threeline signal of five-coordinate NO complex was predominant. We thought that the new species was rapidly degraded to the five-coordinate NO complex upon freezing. The examination of stability of the new species as a function of pH at Ϫ24°C revealed that the species was more stable at pH 8.6 than at pH 7.6. As expected, the EPR spectrum of the species prepared at pH 8.6 displayed a new EPR signal at g ϭ 1.979 besides the three-line signal of the five-coordinate NO complex (b in Fig.  1B). The new EPR species was not a modified form of the five-coordinate NO complex produced by the effect of pH or the binding of antifreeze, since the changes in pH from 7.6 to 8.6, the changes in the antifreeze from ethylene glycol to glycerol, or changes in the concentration of antifreeze did not alter the EPR spectrum of the five-coordinate NO complex. The EPR signal of the new species was obtained by subtracting the three-line signal of the five-coordinate species (a in Fig. 1B) from the spectrum of trace b. The resultant spectrum (c in Fig.  1B) was typical of a six-coordinate ferrous NO complex as indicated by a triplet superhyperfine splitting of 14 NO in the central resonance signal around g ϭ 2. This was the first clear identification of a six-coordinate ferrous NO complex of sGC. The six-coordinate NO complex was prepared by adding 100 M SNAP to the ferrous enzyme (2.1 M as heme) at Ϫ24°C in 50 mM TEA, pH 7.6, containing 40% ethylene glycol. Inset, the degradation of the six-coordinate ferrous NO complex to the five-coordinate ferrous NO complex after raising the temperature to Ϫ15°C. B, EPR spectrum of the ferrous NO complex (16 M as heme). The EPR spectrum illustrated in trace a was that of the fully five-coordinate NO complex, which was prepared by adding NO under anaerobic conditions at Ϫ5°C in 50 mM TEA buffer, pH 8.6, containing 40% ethylene glycol, and that in trace b was that of the six-coordinate NO complex prepared by adding SNAP under anaerobic conditions at Ϫ24°C in 50 mM TEA buffer, pH 8.6, containing 40% ethylene glycol. In the spectrum of trace b, the presence of the six-coordinate NO species at g ϭ 1.979 besides that of the five-coordinate one was noted. The EPR signal of the six-coordinate NO complex, which was obtained by subtracting the three-line signal of the fivecoordinate NO complex from spectrum b, was shown in trace c. The EPR measurements were done at 35 K and at microwave power of 10 milliwatts. NO A is the coupling constant for hyperfine splitting by 14 NO.
Superhyperfine structure of a triplet of triplets in the g z region, which was indication of the ligation of axial ligand with 14 N nucleus, a histidine residue (29), was unclear in the spectrum due to the low signal quality.
The formation of the six-coordinate NO complex was examined at room temperature by monitoring the absorbance at 400, 419, and 430 nm under stopped flow conditions ( Fig. 2A). It was particularly noted that the magnitude of the absorbance changes at 419, 400, and 430 nm was different from that expected. For instance, if the five-coordinate NO complex is assumed to be directly produced in the reaction between the ferrous enzyme and NO, the absorbance decrease at 430 nm must be much larger than the increase at 400 nm, indicative of the formation of the five-coordinate NO complex. However, the absorbance change at 430 nm was smaller than that at 400 nm ( Fig. 2A). Furthermore, the absorbance decrease at 419 nm was unexpectedly large. The most reasonable interpretation was that the six-coordinate NO complex was produced within a dead time of the apparatus (about 2.5 ms), and then converted to the five-coordinate NO complex. To confirm the formation of the six-coordinate NO complex within a dead time of the apparatus, the reaction was analyzed by a rapid scan spectrophotometer (Fig. 2B). The spectrum taken at 4 ms after mixing agreed with that of the six-coordinate NO complex shown in Fig. 1A and was converted to that of the five-coordinate NO complex with the Soret band at 400 nm through one set of isosbestic points.
The time course at 419 nm ( Fig. 2A) obeyed that of a first order reaction with a rate constant of 38 s Ϫ1 . The conversion rates did not show significant pH-dependent changes between pH 7.0 and 8.6 at 15°C (Fig. 2A, inset), being different from the above described results at Ϫ24°C in the presence of ethylene glycol. The finding that the six-coordinate NO complex formation was completed within the dead time indicated that the binding rate of NO to the ferrous enzyme, i.e. the formation rate of the six-coordinate NO complex, was much faster than 1 ϫ 10 7 M Ϫ1 s Ϫ1 . These results together were the first clear evidence of the formation of the six-coordinate NO complex.
Co 2ϩ Protoporphyrin IX-substituted Enzyme-Apoenzyme used for the reconstitution with Co 2ϩ protoporphyrin IX exhibited the basal and NO-stimulated activities of 38 and 165 nmol/min/mg of protein, respectively. The following titration experiments indicated that the apoenzyme preparation retained a correct binding site for protoheme IX. When the apoenzyme was titrated with protoheme under anaerobic conditions maintained by the addition of a slight excess of Na 2 S 2 O 4 , the absorbance at 431 nm as well as the cyclase activity increased with an increased amount of protoheme, giving a clear inflection point. At the point the cyclase activity reached a plateau (data not shown), with the NO-stimulated activity of 2550 nmol/min/mg of protein. The basal activity of the reconstituted enzyme was 16 nmol/min/mg of protein. The resultant reconstituted enzyme exhibited an optical spectrum essentially identical to that of the native enzyme. Similar results were also obtained when titrated with Co 2ϩ protoporphyrin IX. The Co 2ϩ -substituted enzyme further purified by the method of the previous section exhibited retention times identical to those of the native enzyme when analyzed by a Superdex 200HR column and a Protein Pak G-DEAE HPLC column (data not shown). The results indicated that the Co 2ϩ -enzyme had the same metal binding site as that of native enzyme, and had essentially identical molecular mass and protein surface charges to those of the native enzyme. The Co 2ϩ -enzyme exhibited a specific activity of 8600 nmol/min/mg of protein in the presence of NO. Since an attempt to purify it to a homogenous state was unsuccessful, the comparisons of activity between cobalt-and iron-enzymes were done using a turnover number defined as mol of cGMP/min/mol of heme or cobalt porphyrin. The turnover number of the partially purified Co 2ϩ -substituted enzyme was 5840 min Ϫ1 in the presence of NO, which was about 1.5-fold higher than that of native enzyme (3800 min Ϫ1 ). The activation of the Co 2ϩ -substituted enzyme by NO was about 50-fold, which was significantly low when compared with the native enzyme (270-fold). The lower activation of the Co 2ϩ -substituted enzyme by NO was attributable to the high basal activity of 115 min Ϫ1 , which was about 8-fold higher than that of native enzyme (14 min Ϫ1 ).
The partially purified Co 2ϩ -substituted enzyme showed the Soret band at 405 nm and the visible band at 559 nm (Fig. 3A), which were nearly identical to those of Co 2ϩ -substituted myoglobin (45). The shoulder absorption around 430 nm marked by an asterisk was attributed to the contamination of the native iron-enzyme by the pyridine hemochromogen assay. The content was estimated to be less than 15%. The addition of NO to the Co 2ϩ -enzyme slightly blue-shifted the Soret band to 399 nm and red-shifted the visible band to 569 nm (Fig. 3B). The   FIG. 2. NO binding analyses by a stopped flow method. A, time course of NO binding to ferrous sGC measured by the stopped flow method. In this experiment, 1.8 M DTT-free ferrous enzyme was mixed with 60 M NO at 15°C. Absorbance changes were monitored at 400, 419, and 430 nm. Inset, pH dependence of the conversion rate from the six-coordinate NO to the five-coordinate NO complex was illustrated. B, spectral changes in the ferrous sGC followed by a rapid scan spectrophotometer after mixing with NO. The DTT-free ferrous enzyme (2.4 M) was mixed with 60 M NO, and spectra were recorded with 4-ms gate time at times indicated in the figure at 15°C. For comparison, the spectrum of the ferrous enzyme (Fe 2ϩ ) is illustrated. In these experiments, 50 mM TEA buffer, pH 7.6, containing 5% glycerol was used, and DTT was omitted to avoid the reaction between NO and DTT. spectral pattern of the NO complex was entirely different from that of Co 2ϩ -myoglobin NO complex in a six-coordinate state, which exhibited the Soret band at 421 nm and 539-and 577-nm bands in the visible region. These results suggested that the NO complex of Co 2ϩ -substituted sGC was in a five-coordinate state. The coordination state was confirmed by a resonance Raman spectroscopy as described below.
The resonance Raman spectra of the Co 2ϩ -substituted enzyme and the NO derivatives were summarized in Fig. 4, A and B. The Co 2ϩ -enzyme exhibited the 4 and 3 Raman bands at 1371 cm Ϫ1 and at 1504 cm Ϫ1 , respectively, which closely agreed with those of Co 2ϩ -myoglobin and -hemoglobin in a six-coordinate state (46). The addition of 14 NO to the Co 2ϩ -substituted enzyme shifted the 4 band to 1376 cm Ϫ1 from 1371 cm Ϫ1 with an appearance of the Raman band at 1682 cm Ϫ1 . The resonance Raman spectrum was markedly different from those of 14 NO complexes in a six-coordinate state (46). The replacement of 14 NO by 15 NO eliminated the Raman band at 1682 cm Ϫ1 with a concomitant appearance of the 1648-cm Ϫ1 band and without detectable shift of other bands (Fig. 4A, c and d). In the low frequency region, we detected the shift of 523 cm Ϫ1 band upon the replacement of 14 NO with 15 NO (Fig. 4B, b and inset). These results indicated that the 1682-and 523-cm Ϫ1 bands were assigned to the NO stretching vibration ( N-O ) and Co-NO vibration ( Co-NO ), respectively. Both N-O and Co-NO values agreed with those of the corresponding vibration of five-coordinate NO complexes of Co 2ϩ model porphyrins (48) but not of six-coordinate NO complexes (47). These findings indicated that the NO complex of Co 2ϩ -substituted sGC was in a fivecoordinate state.
The EPR spectrum of Co 2ϩ -substituted enzyme was shown in Fig. 5A, with that of Co 2ϩ -myoglobin for comparison. The Co 2ϩsubstituted enzyme exhibited five-coordinate low spin signals at g Ќ ϭ 2.37 and g ϭ 2.04 with poorly resolved eight-line hyperfine splitting due to 59 Co nucleus ( Co A ϭ 7.4 mT). The g Ќ ϭ 2.37 component was significantly broad compared with that of Co 2ϩ -myoglobin. The hyperfine splitting constant ( Co A ϭ 7.4 mT) agreed with that of other Co 2ϩ -substituted hemoproteins with proximal histidine (34,37), suggesting the presence of a histidine residue as the proximal ligand in sGC. However, the triplet superhyperfine splitting due to the 14 N nucleus of the axial ligand was not well resolved in Fig. 5A. To gain firm evidence for the histidine ligation, EPR signals with 20-mT sweep width centered at 300 mT were accumulated to obtain a high quality spectrum. As shown in Fig. 5B, we could detect the three-line superhyperfine splitting due to the 14 N nucleus ( N a ϭ 1.7 mT). Thus, the EPR signal of Co 2ϩ -substituted sGC was characterized by a poorly resolved triplet splitting of the 14 N nucleus and the relatively broad g Ќ component. These features resembled those of the ␣-subunit in Co 2ϩ -hemoglobin FIG. 3. Optical absorption spectra of Co 2؉ protoporphyrin IXsubstituted sGC. A, optical spectrum of Co 2ϩ -substituted enzyme under anaerobic conditions at 5°C. B, optical spectrum of the NO complex of Co 2ϩ -substituted enzyme at 5°C. The NO complex was prepared by adding NO under anaerobic conditions. In these experiments, 50 mM TEA buffer, pH 7.6, containing 5% glycerol and 5 mM DTT was used.
FIG. 4. Resonance Raman spectra of Co 2؉ protoporphyrin IXsubstituted sGC. A, high frequency resonance Raman spectra of Co 2ϩ porphyrin-substituted enzyme and its NO complexes. The spectra of the Co 2ϩ -enzyme, Co 2ϩ -14 NO complex, and Co 2ϩ -15 NO complex are illustrated in a, b, and c, respectively. The difference spectrum between 14 NO and 15 NO complexes ( 14 NO Ϫ 15 NO) is presented in d. B, low frequency resonance Raman spectra of Co 2ϩ porphyrin-substituted enzyme (trace a) and the 14 NO complex (trace b). The difference spectrum between 14 NO and 15 NO is shown in the inset. These spectra were taken at 406.7 nm excitation wavelength. The buffer used in these experiments was 50 mM TEA buffer, pH 7.6, containing 5% glycerol and 5 mM DTT. tetramer in T-state (34) rather than that of the ␤-subunit of Co 2ϩ -hemoglobin in T-state (34), Co 2ϩ -myoglobin, or Co 2ϩhorseradish peroxidase (35).
Azide Complex-As shown in Fig. 6A, the addition of N 3 Ϫ to the ferric enzyme caused a decrease in the intensity of the Soret band and a remarkable increase in the intensity of the 635-nm band at room temperature, confirming the previous result (22). These spectral changes were unusual, because the N 3 Ϫ addition to other ferric hemoproteins such as metmyoglobin produced 420-and 540-nm low spin bands and reduced the 640-nm band intensity (49). The low spin bands in the metmyoglobin N 3 Ϫ complex were intensified by lowering the temperature, showing that the spin state was in a thermal spin equilibrium between low and high spin states (50,51). In contrast, the N 3 Ϫ complex of sGC did not display the spectral change by lowering the temperature to 77 K (data not shown). This indicated that the ferric heme of the N 3 Ϫ -bound sGC was in a high spin state, not in a thermal spin equilibrium.
The ferric enzyme exhibited Raman bands assignable to the oxidation state marker at 1371 cm Ϫ1 ( 4 ), the coordination marker at 1492 cm Ϫ1 ( 3 ), and 2 at 1568 cm Ϫ1 (Fig. 6B). The result showed that the coordination and spin states of the enzyme heme were categorized to ferric five-coordinate high spin heme (52). The addition of N 3 Ϫ only intensified and sharpened the coordination marker band ( 3 ) at 1492 cm Ϫ1 without detectable shifts of the other bands, indicating that the coordination state of the ferric enzyme heme remained unchanged upon the addition of N 3 Ϫ . Although we tried to detect the ligation of N 3 Ϫ to the enzyme heme through the detection of Fe-N 3 Ϫ stretching vibration, it was unsuccessful. The N 3 Ϫ addition markedly enhanced the cyclase activity of the ferric enzyme, but not the ferrous enzyme (Fig. 7). The activation reached a maximum at 50 mM N 3 Ϫ and gradually decreased with an increase in the N 3 Ϫ concentration (Fig. 7). The reason for the activity decrease over 50 mM N 3 Ϫ was unknown. The maximum activity expressed as turnover number FIG. 5. EPR spectra of Co 2؉ protoporphyrin IX-substituted sGC. A, EPR spectrum of Co 2ϩ protoporphyrin IX-substituted sGC (Co sGC) at 35 K (upper trace). The EPR spectrum of Co 2ϩ protoporphyrin IX substituted myoglobin (Co Mb) is shown for comparison (lower trace). B, accumulated EPR spectrum of Co 2ϩ protoporphyrin IX-substituted sGC between 0.29 and 0.31 mT. In this experiment, 40 scans were averaged. The buffer used in these experiments was 50 mM TEA buffer, pH 7.6, containing 10% glycerol and 5 mM DTT. N a and Co A denote coupling constants for hyperfine splitting by 14 N nucleus of proximal base and 59 Co nucleus, respectively.
(mol of cGMP/min/mol of heme) was 970 min Ϫ1 , which corresponded to the specific activity of 9100 nmol of cGMP formed/ min/mg of protein. This was about one-third of the specific activity of the ferrous NO complex. The result was a first observation for the activation of the enzyme in the ferric state. The addition of 10 mM KCN markedly inhibited the activity in the presence of 50 mM N 3 Ϫ , but the inhibitory effect was significantly diminished by increasing the N 3 Ϫ concentration to 150 mM (Fig. 7). This strongly suggested that N 3 Ϫ and cyanide combined to the same site, i.e. the ferric enzyme heme.
The oxidized resting enzyme exhibited only rhombic high spin signals with g values of g 1 ϭ 6.62, g 2 ϭ 5.36, and g 3 ϭ 1.98 at 5 K (Fig. 8, trace A). The EPR spectrum was not altered by raising the temperature to 15 K (not shown), indicating that the ferric enzyme heme did not contain a low spin component. The result essentially agreed with the previous report measured at 20 K (22). These results with resonance Raman data confirmed that the ferric enzyme heme in sGC was in a single coordination and in pure high spin state. The addition of 50 mM N 3 Ϫ produced two types of new high spin species with EPR signals of gЈ 1 ϭ 6.94 and gЉ 1 ϭ 6.10 at 5 K (Fig. 8, trace B). To estimate the amount of N 3 Ϫ -bound heme in the presence of 50 mM N 3 Ϫ , the spectrum of trace A divided by some factors was subtracted from the spectrum of trace B. As illustrated in the spectrum BЈ), the subtraction of the spectrum divided by 2 satisfactorily eliminated the residual unreacted ferric enzyme, indicating that a half of the enzyme combined N 3 Ϫ ion. The reason why the N 3 Ϫ addition yielded two types of high spin species remained unclear. These new high spin signals were intensified by increasing the N 3 Ϫ concentration to 150 mM (Fig.  8, trace C). The EPR spectrum at 15 K (Fig. 8, trace D) was essentially identical with that at 5 K, indicating that the N 3 Ϫ complex did not contain low spin components. DISCUSSION Stone and Marletta (53) have reported the mechanism for the five-coordinate NO complex formation. They proposed that NO first combined with the ferrous heme to produce a sixcoordinate NO complex, which then converted to a five-coordinate NO complex. The conversion was a complex process, which was interpreted by assuming two populations of the enzyme with different activities for NO. About 30% of the population of the heme rapidly converted to the five-coordinate NO complex from the six-coordinate one via a single exponential pathway (20 s Ϫ1 at pH 7.4 and 10°C), while the conversion of the remaining 70% of the population was very slow and depended on NO concentration. The slow NO-dependent conversion of the latter major population was explained by assuming an unidentified nonheme iron binding site of NO on the protein. However, these experimental findings were inadequate to argue the formation of a six-coordinate NO complex and inconsistent with our results in some points. In the present work, we obtained definite evidence to show the formation of a six-coordinate NO complex in the reaction (Figs. 1 and 2). In contrast to the above findings by Stone and Marletta, our data showed that the entire population of the six-coordinate NO complex rapidly converted to the five-coordinate NO complex through a single FIG. 6. Optical and resonance Raman spectra of the ferric azide complex. A, the optical absorption spectrum of ferric sGC in the absence and presence of azide (50 mM). The spectra were taken at 5°C, and the buffer used was 50 mM TEA buffer, pH 7.6, containing 5% glycerol. B, the resonance Raman spectra of ferric sGC with or without 50 mM of azide in 50 mM TEA buffer, pH 7.6, containing 5% glycerol. The spectra were taken at 406.7-nm excitation wavelength. exponential pathway (38 s Ϫ1 at pH 7.6 and 15°C). Furthermore, the conversion rate was almost unchanged at the lower concentration of NO (10 M). Our results did not contradict an interpretation that the enzyme heme was the sole binding site for NO. One might claim that the discrepancy for the kinetic results was due to the difference in the heme content. The 1.5-heme stoichiometry reported by Stone and Marletta (18) was estimated by the precise protein determination, in which the protein content obtained by the Bradford protein assay was corrected by a factor of about 1.6 based on the quantitative amino acid analyses; i.e. the Bradford protein assay overestimated the amount of protein. When the correction was not made, the heme content in their preparations essentially agreed with the 1-heme stoichiometry obtained by the present and the recent studies (54). Thus, the difference was considered apparent, although efforts to obtain the precise amount of protein were not made in this and other studies (54). Whatever the reason for the discrepancy, our data including the EPR and resonance Raman studies agreed with a view that the enzyme heme in our preparation was a single population with a single coordination structure and definitely indicated that the fivecoordinate NO complex was produced via the formation of the six-coordinate NO complex.
The optical and resonance Raman spectra of the Co 2ϩ -reconstituted enzyme presented here differed considerably from the results of Dierks et al. (48), especially for the NO complex. They reported the Soret absorption maximum at 390 nm of the NO complex, but the corresponding complex in our preparation exhibited the maximum at 399 nm. The discrepancy might be attributed to the difference in the heme environment. Indeed, Deinum et al. (24) pointed out that the apoenzyme obtained by Dierks et al. had a different environment from the native enzyme. By using the apoenzyme preparation with a correct heme binding site (see "Results"), we demonstrated that the Co 2ϩsubstituted enzyme formed a five-coordinate NO complex with a high cyclase activity. Hence, the coordination structure of the active NO complex was essentially the same irrespective of whether the central metal of porphyrin was iron or cobalt.
The iron-histidine stretching frequency of sGC essentially agreed with that of ␣-hemes of hemoglobin in T-state (55), implying a tension to pull the proximal histidine from the porphyrin plane in sGC (24). EPR characterization of native and Co 2ϩ -enzymes also revealed the similarity in metal environment between ␣-subunits of hemoglobin and sGC. The EPR spectrum of the six-coordinate NO complex of native sGC dif-fered from those of horseradish and cytochrome c peroxidases with an anionic proximal histidine residue but closely resembled those of hemoglobin and myoglobin with a neutral proximal histidine residue. Among the latter hemoproteins, the sixcoordinate NO complex of ␣-subunits of hemoglobin (31,32) exhibited an EPR spectrum similar to that of sGC with the paramagnetic center of rhombic symmetry. The Co 2ϩ -substituted enzyme also exhibited an EPR signal analogous to ␣(Co) subunits in Co 2ϩ -hemoglobin tetramer in T-state with the broad g Ќ component and the poorly resolved 14 N superhyperfine splitting (33), where the Co 2ϩ -proximal histidine bond in ␣(Co) subunits was reported to be more tensioned than that of ␤(Co) subunits (56). The present data revealed that the nature of the metal-histidine bond in sGC was strikingly analogous to that of ␣-subunits of hemoglobin and provided additional evidence for tension on the metal-proximal histidine bond of sGC proposed by a resonance Raman study (24).
Heme iron was reported to lie about 0.4 Å out of the porphyrin plane toward the proximal side in a five-coordinate ferrous high spin state and to move into the porphyrin plane upon NO ligation, yielding an ϳ0.4-Å movement of the iron atom from the initial position (57)(58)(59)(60). The movement upon NO ligation might impose further tension on the iron-proximal histidine bond in sGC but did not cause the immediate cleavage of the proximal bond, as demonstrated by the formation of the sixcoordinate NO complex with the proximal histidine ( Figs. 1 and  2). There were several factors to facilitate the cleavage of the iron-histidine bond, including a repulsive trans effect of NO (27,61) and the protonation of the proximal histidine residue (62,63). The latter was proposed for the proximal histidine release of myoglobin and peroxidases at acidic pH. This possibility might be excluded, because the rate for the release was almost unchanged between pH 7.0 and 8.6 ( Fig. 2A, inset), and the pK value higher than 8.6 might be unlikely for the imidazole group. The repulsive trans effect of NO on the proximal histidine, therefore, was concluded to be an important driving force for the proximal histidine release, as has been hypothesized by others (24,27). The resultant five-coordinate NO complex possibly retained the iron atom displaced from the porphyrin plane toward NO, as shown for the iron-porphyrin complexes with five-coordinate NO structure (64). The displacement of the iron atom toward NO might stabilize the five-coordinate NO complex by preventing the access and the rebinding of the proximal histidine residue to the heme iron. Thus, the tension imposed by the in plane iron movement and FIG. 8. EPR spectra of ferric sGC and its ferric azide complex. Trace A, EPR spectrum of the ferric enzyme at 5 K; trace B, EPR spectrum of the ferric N 3 Ϫ (50 mM) complex at 5 K; trace BЈ, difference spectrum (trace B Ϫ 0.5 ϫ trace A); trace C, spectrum of the ferric N 3 Ϫ (150 mM) complex at 5 K; trace D, spectrum of the ferric N 3 Ϫ (150 mM) complex at 15 K. EPR spectra were taken at a microwave power of 5 milliwatts and by 100-kHz modulation with 0.5-mT width, and the enzyme concentrations were 75 M as heme. Throughout these EPR measurements, 50 mM TEA buffer, pH 7.6, containing 10% glycerol was used. the repulsive trans NO effect are crucial for the iron-proximal histidine bond cleavage of sGC upon NO ligation.
The overall movement of metal associated with the formation of five-coordinate NO complex significantly differed between Fe 2ϩ and Co 2ϩ porphyrins. The overall movement from the initial position was ϳ0.35 Å for cobalt atom and ϳ0.6 Å for iron atom, which were assessed from the results of model porphyrin complexes and myoglobins (39, 64 -66). If the estimation is valid for the Co 2ϩ -substituted sGC, one may assume that the 0.35-Å movement is sufficient for breaking the metal-histidine bond in Co 2ϩ -substituted enzyme upon NO binding. Thus, it was predicted that the proximal base bond of the Co 2ϩ -substituted enzyme was more readily broken than in the native iron-enzyme, and the instability of the bond correlated with the higher basal activity in the Co 2ϩ -enzyme.
It has been known that N 3 Ϫ activated sGC in the presence of exogenously added catalase and DTT under aerobic conditions (16). In this case, catalase oxidized N 3 Ϫ to NO using H 2 O 2 produced by autoxidation of DTT, resulting in the NO complex formation of sGC. In the present study, DTT and catalase were not exogenously added. Under these conditions, we found that the ferric enzyme was capable of catalyzing cGMP production in the presence of N 3 Ϫ . The optical and EPR spectral studies revealed that the N 3 Ϫ complex was in the five-coordinate high spin state (Figs. 6 and 8). The five-coordinate N 3 Ϫ heme has been proposed for the high spin N 3 Ϫ complex of carp hemoglobin in T-state with inositol hexaphosphate (67), and McCoy and Caughey (68) have found the similarity between the infrared N 3 Ϫ stretching frequency of high spin N 3 Ϫ complex of hemoglobin and that of the five-coordinate N 3 Ϫ protoheme complex. Although we could not obtain a clear indication for the formation of the six-coordinate N 3 Ϫ heme in the reaction between the ferric enzyme and N 3 Ϫ , the formation of the six-coordinate one was assumed important as a trigger for the release of the proximal histidine in sGC; the formation moved the Fe 3ϩ atom into the porphyrin plane from the position initially displaced toward the proximal side as reported for model compounds (69,70). The movement probably imposed further tension on the proximal histidine bond of sGC. Consequently, the six-coordinate N 3 Ϫ heme might experience the release of the proximal histidine, yielding a small fraction of a five-coordinate N 3 Ϫ heme with in plane configuration of the iron atom. The five-coordinate N 3 Ϫ heme once formed would be stabilized by the iron displacement from the porphyrin plane toward the N 3 Ϫ ion. These considerations led to a reasonable conclusion for why CO and cyanide complexes of sGC were practically inactive; to our knowledge it has not been known that both CO and cyanide complexes formed the stable five-coordinate complex with the iron displacement from the porphyrin plane toward the corresponding ligand. 2 It is emphasized that our results provide the first clear experimental evidence for the five-coordinate high spin N 3 Ϫ heme, no matter whether or not the mechanism for the proximal histidine release is correct.