Activation mechanism of methanol:5-hydroxybenzimidazolylcobamide methyltransferase from Methanosarcina barkeri.

Methanol:5-hydroxybenzimidazolylcobamide methyltransferase (MT1) is the first of two enzymes involved in the transmethylation reaction from methanol to 2-mercaptoethanesulfonic acid in Methanosarcina barkeri. MT1 only binds the methyl group of methanol when the cobalt atom of its corrinoid prosthetic groups is present in the highly reduced Co(I) state. Formation of this redox state requires H2, hydrogenase, methyltransferase activation protein, and ATP. Optical and electron paramagnetic resonance spectroscopy studies were employed to determine the oxidation states and coordinating ligands of the corrinoids of MT1 during the activation process. Purified MT1 contained 1.7 corrinoids per enzyme with cobalt in the fully oxidized Co(III) state. Water and N-3 of the 5-hydroxybenzimidazolyl base served as the upper and lower ligands, respectively. Reduction to the Co(II) level was accomplished by H2 and hydrogenase. The cob(II)amide of MT1 had the base coordinated at this stage. Subsequent addition of methyltransferase activation protein and ATP resulted in the formation of base-uncoordinated Co(II) MT1. The activation mechanism is discussed within the context of a proposed model and compared to those described for other corrinoid-containing methyl group transferring proteins.

produced, which is the substrate for the final step in methanogenesis in all methanogens studied so far (3).
The corrinoid prosthetic group of MT 1 can only be methylated by methanol when the central cobalt atom of the cobamide is present in the highly reduced Co(I) state (4,5). Since this state is extremely sensitive toward oxidation, MT 1 readily inactivates upon manipulation and even during catalysis. Reactivation is possible and requires participation of a reducing system, methyltransferase activation protein (MAP), and ATP (4, 6 -8). The reducing system consists of hydrogen, hydrogenase, and ferredoxin. Ferredoxin is not absolutely required, though it stimulates the apparent reaction rate of methyl group transfer (6).
Here, we report the UV-visible absorbance and electron paramagnetic resonance properties of the corrinoid prosthetic groups of MT 1 under various additions of the reducing system, MAP, and ATP. From these results, the sequence of events leading to the formation of the cob(I)amide of MT 1 is deduced. The activation of MT 1 proceeds by a novel mechanism, which is presented in a model and compared to those described for other corrinoid-containing methyl group-transferring proteins.

EXPERIMENTAL PROCEDURES
Cell Material-Cells of M. barkeri strain MS (DSM 800) were grown and harvested, and cell extract was prepared anaerobically as described previously (6,9).
Enzyme Assays-Incubation mixtures were prepared in an anaerobic glove box, and reactions were performed in crimp-sealed 10-ml serum vials. MT 1 activity was determined by measuring the methanol-dependent HS-CoM conversion to CH 3 -S-CoM when added to a reaction mixture containing MT 2 /hydrogenase, MAP, and ferredoxin fractions obtained after DEAE-Sepharose fractionation of cell extract of M. barkeri (6). A typical reaction mixture (final volume, 100 l) contained 50 mM TES/K ϩ buffer (pH 7.0), 24 mM MgCl 2 , 10 mM methanol, 10 mM HS-CoM, 2 mM ATP, 1 mM 2-bromoethanesulfonic acid, 20 l of MT 2 / hydrogenase fraction, 5 l of ferredoxin fraction, 25 l of MAP fraction, and an amount (usually 25 l) of MT 1 to be tested (6). After gassing with 50% H 2 , 50% N 2 (100 kPa), the vials were kept on ice. Reactions were started by placing the vials at 37°C. After appropriate incubation periods, routinely 0, 15,30,45, and 60 min, reactions were stopped by placing the vials on ice. Activity of methyl group transfer of methanol to HS-CoM was routinely assayed by measuring the decrease in the amount of HS-CoM (see below). The methyltransferase activity obtained was linearly dependent on the amount of MT 1 added.
Protein Purification-Because several of the enzymes involved in the methanol:HS-CoM methyltransferase reaction are oxygen-labile (1,4,6,8), all handlings were performed in an anaerobic glove box (97.5% N 2 , 2.5% H 2 ) at room temperature. The purification procedure started by applying 10 ml of cell extract to a DEAE-Sepharose-CL-6B column and separating the proteins involved in the methyltransferase reaction as described previously (6). MT 2 and hydrogenase activity eluted between 0.20 and 0.22 M NH 4 Cl. Fractions between 0.25 M and 0.34 M NH 4 Cl * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
MT 1 was purified to homogeneity with an anaerobic Perkin-Elmer fast protein liquid chromatography system equipped with a TSK DEAE-5-PW column (7.5 cm by 0.75 cm). After application of a 550-l DEAE-Sepharose MT 1 fraction, the column was washed with 15 ml of 50 mM TES/K ϩ buffer (pH 7.0) containing 15 mM MgCl 2 , 1 mM dithiothreitol, and 10% ethylene glycol followed by a 113-ml linear gradient of 0 -0.25 M NH 4 Cl in the buffer. Subsequently, the column was washed with 15 ml of 0.5 M NH 4 Cl in the buffer to remove all bound proteins. The eluate was monitored at 280 nm, and fractions of 0.9 ml were collected at a flow rate of 0.45 ml min Ϫ1 . MT 1 activity was present in a large peak that was eluted at about 0.20 M NH 4 Cl. Fractions eluting between 0.18 and 0.22 M NH 4 Cl were pooled. This purification step was repeated for the remainder of the DEAE-Sepharose MT 1 fraction. The purified MT 1 pools were combined, washed, and concentrated by Amicon YM-10 ultrafiltration to the desired volume, usually 0.4 ml.
MAP was partially purified on a hydroxylapatite column packed with TSK-Gel HA-1000 (7.5 by 0.75 cm) and equilibrated in 50 mM TES/K ϩ buffer (pH 7.0) containing 1 mM dithiothreitol, 0.5 mM CHAPS, and 10% ethylene glycol. After application of the sample (1 ml), the column was washed with 15 ml of equilibration buffer. Bound protein was eluted by a 50-ml simultaneous linear gradient of 50 -0 mM TES/K ϩ and 0 -300 mM ammonium phosphate buffer (pH 7.0); both buffers contained 1 mM dithiothreitol, 0.5 mM CHAPS, and 10% ethylene glycol. Fractions (0.9 ml) were collected at a flow rate of 0.45 ml min Ϫ1 . MAP activity was recovered in a peak that was eluted at 145 mM ammonium phosphate and 26 mM TES. The purification step was repeated for the remainder of the DEAE-Sepharose MAP fraction. Purified MAP pools were combined, desalted, and concentrated by Amicon YM-10 ultrafiltration by washing with 50 mM TES/K ϩ buffer (pH 7.0) containing 1 mM dithiothreitol and 10% ethylene glycol. At this stage, MAP was purified 23-fold with a recovery of 6.5%; apart from MAP (60 kDa) 8 -25% native polyacrylamide gel electrophoresis (PAGE) showed the presence in amounts approximately equal to MAP of two contaminating proteins (96 and 32 kDa) that apparently are not related with the methanol:HS-CoM methyltransferase reaction (8). Since the procedure to isolate homogeneous MAP gave very low yields (8), the two-step purification method was employed to obtain the activation protein in the large amounts required for the experiments described.
In order to remove fortuitous Mn 2ϩ that interfered with the EPR experiments, MT 2 /hydrogenase and ferredoxin fractions were applied separately to a Sep-Pak CM cation exchange cartridge and eluted with 50 mM TES/K ϩ buffer (pH 7.0) containing 1 mM dithiothreitol and 10% ethylene glycol.
Analytical Procedures-UV-visible light absorption spectra were recorded in 1-ml quartz cuvettes on a Hitachi U-3200 spectrophotometer. Spectra of MT 1 were recorded against a reference that contained the same components except MT 1 . EPR spectroscopy was carried out on a Bruker 200 D spectrometer equipped with cryogenics, peripheral equipment, and data acquisition/manipulation facilities as described previously (10). Incubation mixtures were anaerobically transferred to EPR tubes under a slight overpressure of hydrogen, and directly frozen in liquid nitrogen. EPR spectra of controls containing the same components except MT 1 , were also recorded and subtracted from the spectra of the MT 1 -containing samples. In the controls MT 1 was replaced by TES/K ϩ buffer. EPR spectra were simulated by using the program KOPER (11)(12)(13).
Protein was determined with the Bio-Rad protein reagent with bovine serum albumin as a standard. Molar concentrations of MT 1 were calculated from the molecular mass (122,000 Da) of the protein (1). Molar amounts of hydroxylapatite-purified MAP were estimated from the reported M r ϭ 60,000 of the protein, taking into account that approximately one-third of the total protein was MAP (8). Corrinoids were quantified after conversion into the dicyanocobamide derivatives. Samples of MT 1 were diluted in 50 mM CAPS buffer (pH 10) containing 5 mM potassium cyanide and incubated for 5 min at 90°C (14). Concentrations were calculated from the absorption at 580 nm (⑀ 580 ϭ 10.2 ϫ 10 3 M Ϫ1 cm Ϫ1 ) (15). High performance liquid chromatography analysis of corrinoids was performed as described before (6). HS-CoM was measured by the method of Ellman (16). Samples of 25 l were mixed with 3 ml 0.48 mM 2,2Ј-dinitro-5,5Ј-dithiobenzoic acid in 150 mM Tris/Cl Ϫ buffer (pH 8.0) and immediately measured at 412 nm. Total iron was determined as described by Fish (17). Manganese was measured as described by Bartley and co-workers (18). The effect of bathophenanthroline disulfonate was tested anaerobically essentially as described by Rouvière and Wolfe (19).

RESULTS
Purification of MT 1 -Methanol:5-hydroxybenzimidazolylcobamide methyltransferase was purified to homogeneity from cell extract of M. barkeri strain MS by the simple two-step procedure summarized in Table I. After this stage only one band was detected upon nondenaturing 8 -25% gradient PAGE showing an apparent M r ϭ 121,000 (Fig. 1). Superose-6 chromatography indicated a molecular mass of about 120 kDa. These values favorably agree with the M r ϭ 122,000 described before (1). Prior to staining of the gel, the protein could be observed as a red-colored band indicative of the presence of corrinoids. Denaturating SDS-PAGE demonstrated the ␣ 2 ␤ subunit composition of polypeptides of 33,000 and 54,000, respectively, as described before (1). With the protein determination employed (Bio-Rad), an average 1.7 Ϯ 0.4 mol of 5-hydroxybenzimidazolylcobamide (B 12 -HBI)/mol of MT 1 was  (Fig. 2) was typical of Co(III) corrinoids (20 -22). Exposure to air did not further alter the spectrum. The prosthetic group showed absorption maxima at 352 (2.89), 493 (1.07), and 522 (1.04) nm with shoulders at 388 (1.16), 403 (1.04), and 465 (1.0) nm (the numbers in parentheses express the absorbance relative to that of the shoulder at 465 nm). The major visible absorption band at 352 nm is characteristic of the presence of water as the upper ligand (21,22). Compared to free aquo-B 12 -HBI, the absorbance of the corrinoid of MT 1 was relatively increased in the 375-475 nm region. Incubation of purified cob(III)amide-MT 1 with up to equimolar amounts of MAP and/or ATP (up to 5 mM) had no effect on the UV-visible spectrum of the corrinoid prosthetic groups.
When MT 1 was incubated under hydrogen with a small amount of MT 2 /hydrogenase and ferredoxin, the protein turned yellow. The spectrum (Fig. 3) displayed the typical characteristics of cob(II)amide (5,20,21). Absorption maxima were observed at 417 (1.31) and 465 (1.29) nm with shoulders at 315 (3.64), 356 (2.11), and 535 (1.0) nm (the numbers in parentheses indicate the absorbance relative to that of the shoulder at 535 nm). Compared to the spectrum of free Co(II) B 12 -HBI, the Co(II) prosthetic group of MT 1 exhibited a relative increase in the absorbances around 400 nm and in the 500 -600 nm regions. Prolonged incubation or the addition of higher amounts of hydrogenase and ferredoxin did not result in any alterations of the spectrum. Ferredoxin was not absolutely required for the reduction.
The spectrum of the hydrogen-reduced Co(II) form of MT 1 was not altered by the subsequent addition of MAP, ATP, or methanol alone. When both MAP and ATP were added the shoulder at 315 nm of the cob(II)amide of MT 1 disappeared while no significant changes occurred at higher wavelengths (Fig. 3). Interestingly, a distortion of the absorbance band at 315 nm is the major feature, when the HBI nucleotide in free Co(II) B 12 -HBI becomes decoordinated ("base-off") upon acidi- fication (21). When MT 1 was incubated in the presence of the reducing system (H 2 , hydrogenase, ferredoxin), MAP and ATP, we never detected the characteristic absorption peak of cob(I) amide at 390 nm. However, the additional presence of methanol in even low concentrations (Ͻ100 M) resulted in the instantaneous formation of methyl-B 12 -HBI as found by high performance liquid chromatography analysis of the extracted corrinoids (data not shown; see also Ref. 6).
EPR Spectroscopy-EPR observations of corrinoids are limited to the paramagnetic Co(II) state of cobalt. Thus, the purified, cob(III)amide-containing MT 1 was EPR-silent. After incubation of MT 1 with hydrogenase and ferredoxin under hydrogen the typical signal of the nucleotide base-coordinated ("base-on") form of Co(II) corrinoids (Fig. 4A) was observed (23). The spectrum could be simulated on the basis of a near-axial, low spin Co(II) complex with hyperfine splitting from the I ϭ 7/2 cobalt nucleus and superhyperfine splitting from an axially coordinating nitrogen with I ϭ 1 (Table II and Fig. 4B). For a reasonable fit, it was required that the line width in the zdirection was a function of the cobalt nuclear quantum number m I . In the perpendicular (or xy-) direction, there was no reso-lution. Therefore, the combined xy values for g, A, and W were not uniquely determined. However, it was found by extensive fitting that the low field shoulder to the experimental spectrum could not be simulated by any combination of these parameters. This shoulder might well be a manifestation of non-collinearity of the g and A(Co) tensors, i.e. the actual symmetry of the coordination site was probably triclinic.
While the presence of MAP or ATP alone did not alter the EPR spectrum of Co(II) MT 1 , the combination of both induced a remarkable change (Fig. 4C). The spectrum was now a sum of two Co(II) spectra: a base-on and a base-off spectrum. The cobalt hyperfine splitting of the latter could be seen to extend to higher field values than that of the former. This pattern became somewhat obscured when we subtracted the pure base-on spectrum from the sumspectrum (resulting in trace D in Fig. 4). Because of the low signal-to-noise ratio, however, the xy pattern of the base-off spectrum was now more easily identified, thus allowing for a simulation of this spectrum (Table II and Fig. 4E). From the simulated spectra (Fig. 4, B and E) and the combined spectrum (Fig. 4C) it could be estimated that 38% of the cob(II)amide in MT 1 was base-uncoordinated. Lower concentrations of MAP gave lower amounts of the base-off species. The formation of base-off Co(II) MT 1 indicated that, as the result of the action of MAP and ATP, the HBI-base was no longer coordinated and that no ligand or a weak ligand with poor delocalization properties was present (23). DISCUSSION Optical and EPR spectroscopical studies were employed to determine the oxidation state of the central cobalt atom and the coordination of the ligands in the corrinoid protein MT 1 under various additions of MAP, ATP, and a reducing system. After isolation MT 1 contained somewhat less then 2 mol of B 12 -HBI/ mol of ␣ 2 ␤ protein. The UV-visible light spectrum (Fig. 2) indicated that the prosthetic group was present in the hexacoordinated Co(III) oxidation state with the nucleotide 5-hydroxybenzimidazole and water as the lower and upper ligands, respectively. Purified MT 1 is inactive, and reactivation apparently requires the reduction of Co(III), which is brought about by a reducing system (H 2 , hydrogenase, and ferredoxin), MAP, and ATP (4, 6, 8) (Fig. 5). Incubation of MT 1 with hydrogen, hydrogenase, and ferredoxin resulted in the reduction to the Co(II) state. EPR spectroscopy demonstrated that the HBIbase was still coordinated at this stage (base-on). Subsequent addition of both MAP and ATP induced a conversion of base-on  Table II for details on the simulations. EPR conditions: microwave frequency, 9.18 GHz; modulation frequency, 100 kHz; modulation amplitude, 0.63 (trace A) and 0.8 millitorr (trace C); microwave power, 0.08 (trace A) and 0.32 milliwatts (trace C); temperature, 24 K (trace A) and 19 K (trace C). into base-off Co(II) MT 1 . We previously showed MAP to be autophosphorylated by ATP; MAP-phosphate is able to substitute for the requirement of MAP and ATP (8). The phosphorylated protein, thus effects MT 1 in such a way that the HBI-base becomes dissociated. In the experiment shown in Fig. 4, an estimated 38% of the corrinoids became base-uncoordinated when MT 1 (15.8 M) was incubated with 7.5 M MAP and excess ATP (5 mM). From this, it follows that the concentration of base-off MT 1 amounted to 6.0 M, which is about equimolar with respect to MAP added. In agreement with this conclusion, lower amounts of base-off cob(II)amide were obtained when the MAP concentration was decreased in the EPR experiments.
The conversion of base-on into base-off cob(II) amide has an important implication. In non-protein bound B 12 -HBI, such base-off conversion causes the midpoint redox potential of the Co(II)/Co(I) couple to increase from -592 to -500 mV (21). In a similar way, the action of MAP and ATP may facilitate the reduction in MT 1 of cob(II)amide to the catalytically active species, the powerful nucleophile cob(I)amide. Although we could not detect the direct formation of cob(I)amide in our experiments, the findings that (i) the species is produced upon demethylation of methylated MT 1 (5) and (ii) methyl-B 12 -HBI bound to MT 1 is formed during the activation of the enzyme in the presence of H 2 , hydrogenase, MAP, ATP and methanol (this paper; Refs. 4 -6), demonstrate that cob(I)amide must play a role in the catalytic and reductive activation cycles. The cob(II)amide/cob(I)amide midpoint redox potentials are strongly influenced by the protein environment. In the corrinoid/ironsulfur proteins involved in acetyl-CoA synthesis and degradation that have been isolated from Clostridium thermoaceticum (24) and from Methanosarcina thermophila (25), reduction to the catalytically active Co(I) state occurred at midpoint redox potentials of Ϫ504 mV (26) and Ϫ486 mV (25), respectively. Here, reduction has to be performed by electrons derived from the carbonyl (CO)/CO 2 oxidation (E 0 Ј ϭ Ϫ520 mV). It is important to note that the enzymes do not require ATP for the activation; in the purified, inactive cob(II)amide state, the corrinoids are already contained in the (at neutral pH) thermodynamically unfavorable base-off state by the protein backbone. Here, the observed midpoint redox potentials about equaled the E 0 Ј ϭ Ϫ500 mV of the free base-off cob(II)amide/cob(I)amide couple. For a number of corrinoid-containing methyltransferases, midpoint potentials have been measured that were significantly higher than found for the corresponding B 12 derivatives in solution (25)(26)(27)(28). For example, the reduction of base-on cob(II)amide to cob(I)amide in the membrane-bound methyltetrahydromethanopterin:HS-CoM methyltransferase complex from Methanosarcina mazei showed an E 0 Ј as high as Ϫ426 to Ϫ450 mV, which is about 150 mV more positive than the analogous reduction of free B 12 -HBI (28). In comparing the UV-visible light spectra of MT 1 with aqueous solutions of B 12 -HBI, we noticed some differences in the 400 nm and 500 -600 nm regions (Figs. 2 and 3). Since MT 1 does not contain Fe-S clusters (this paper) or other chromophoric groups (results not shown), this had to be caused by a conformational distortion of the corrin ring structure by the protein (29). Such distortion is likely to change the reduction potential of the prosthetic group. Incubation of methyltetrahydromethanopterin:HS-CoM methyltransferase with ATP and the methyl donor (methyltetrahydromethanopterin) raised the apparent midpoint potential another 200 mV (E 0 Ј ϭ Ϫ245 mV) (28), i.e. to a level where reduction becomes feasible at even very low hydrogen concentrations (E 0 Ј ϭ Ϫ414 mV). Remarkably, the 200 mV shift was not observed when the methyl donor was omitted, and the authors (28) proposed that the simultaneous action of ATP and the methylating substrate in a ternary enzyme complex is required for raising the redox potential. This may also apply to MT 1 . As pointed out above, we never observed the characteristic UV-visible light spectral features of cob(I)amide upon incubation under even high (100 kPa) hydrogen partial pressure of MT 1 with MAP and ATP. However, the additional presence of methanol results in the formation of methyl-B 12 -HBI in MT 1 , suggesting a cooperative action of MAP-phosphate and methanol in the reductive activation (Fig.  5). The EPR experiments outlined above indicated that the amount of base-off cob(II) amide formed was dependent on the amount of MAP added. Yet, MT 1 may be fully activated in the presence of substoichiometric amounts of MAP provided ATP and methanol is present (6,8). This may be explained by assuming that MAP becomes dephoshorylated after completion of the activation cycle of the MT 1 molecule (8) (Fig. 5). Rephosphorylation by ATP then yields MAP-phosphate for activation of another molecule. Future investigations have to clarify questions with respect to the corrinoid midpoint redox potentials in MT 1 and the effects hereon of MAP-phosphate and methanol.
In order to be active, corrinoid-dependent methyltransferases often require an ATP-dependent reductive activation (3,30). As yet, only the activation mechanism of methionine synthase has been elucidated (31). Here, ATP is the substrate in the formation of the potent methylating agent, S-adenosyl methionine, which traps Co(I) out of the thermodynamic unfavorable Co(II) to Co(I) reduction equilibrium (31). In this paper, we have presented evidence that nature developed another approach to facilitate the generation of the active enzymes, notably by inducing in an ATP-dependent process the conformational change of the prosthetic group. Perhaps other corri- In the base-off Co(II) state, no ligand is shown but it is possible that a ligand with poor delocalization properties, e.g. water, is present. The dashed line indicates that it is not fully known how and at which stage MAP-phosphate is dephosphorylated. By the combined action of the reducing system, MAP-phosphate and methanol, catalytically active MT 1 containing methyl-B 12 -HBI is subsequently produced. Hyd, hydrogenase; Fd, ferredoxin. Ferredoxin is shown in parentheses because activation can occur in the absence of this electron carrier (6). noid-containing methyltransferases from methanogens (3) and other obligate anaerobic organisms (32) are activated in a similar fashion.