The Role of an Iron-Sulfur Cluster in an Enzymatic Methylation Reaction

This paper focuses on how a methyl group is transferred from a methyl-cobalt(III) species on one protein (the corrinoid iron-sulfur protein (CFeSP)) to a nickel iron-sulfur cluster on another protein (carbon monoxide dehydrogenase/acetyl-CoA synthase). This is an essential step in the Wood-Ljungdahl pathway of anaerobic CO and CO2 fixation. The results described here strongly indicate that transfer of methyl group to carbon monoxide dehydrogenase/acetyl-CoA synthase occurs by an SN2 pathway. They also provide convincing evidence that oxidative inactivation of Co(I) competes with methylation. Under the conditions of our anaerobic assay, Co(I) escapes from the catalytic cycle one in every 100 turnover cycles. Reductive activation of the CFeSP is required to regenerate Co(I) and recruit the protein back into the catalytic cycle. Our results strongly indicate that the [4Fe-4S] cluster of the CFeSP is required for reductive activation. They support the hypothesis that the [4Fe-4S] cluster of the CFeSP does not participate directly in the methyl transfer step but provides a conduit for electron flow from physiological reductants to the cobalt center.

reaction requires MeTr, the CFeSP, and CODH/ACS. Ferredoxin stimulates the reaction 4-fold. Under these conditions, the Co(II)-CFeSP must be reduced to Co(I) before it can enter the catalytic cycle. In the studies described here, we monitored the synthesis of acetyl-CoA from CO, CH 3 -H 4 folate, CoA, and the methylated form of either the wild type or C20A variant CFeSP. If the two hypotheses described above are correct, then, when the C20A variant of the methylated CFeSP is used, the rate of acetyl-CoA synthesis will slowly decrease at the rate of oxidative inactivation of the cobalt center. Furthermore, we can measure the UV-visible spectra of the C20A variant during and after the reaction to directly determine if Co(I) forms during the linear phase of the reaction and if Co(II) forms when the reaction has undergone inactivation. These results can be compared with those for the wild type protein. According to our hypotheses, when the wild type protein is used, this inactivation will not be observed because CODH and CO 2 will rapidly reduce the Co(II) protein, recruiting the inactive protein back into the catalytic cycle.

EXPERIMENTAL PROCEDURES
Materials-N 2 (99.98%) and CO (99.99%) were obtained from Linweld (Lincoln, NE). N 2 was deoxygenated by passing through a heated column containing BASF catalyst. Reagents were of the highest purity available. 14 CH 3 -H 4 folate was purchased from Amersham Pharmacia Biotech.
Organism and Enzyme Purification-Construction of the C20A CFeSP variant, cloning the gene into Escherichia coli, and reconstitution of the protein with B 12 and the FeS cluster was described earlier (20). The wild type (10) and variant (20) CFeSPs were purified under strictly anaerobic conditions as described. CODH/ACS (22), ferredoxin II (23), and MeTr (6) were purified as described under strictly anaerobic conditions at 17°C in a Vacuum Atmospheres chamber maintained below 1 ppm oxygen. Protein concentrations were determined by the Rose Bengal method (24).
Enzyme Assays-The wild type and variant CFeSPs were methylated with methyl iodide essentially as described earlier (8). 3 The as-isolated protein was first reduced by reacting with 10 mM titanium(III) citrate.
The solution was then incubated for 15-30 min at 13°C with 20-fold excess 14 CH 3 I and centrifuged through a Sephadex G-50 column (26) to remove the unreacted methyl iodide and the titanium citrate. Therefore, methyl iodide, which inhibits CODH/ACS, was absent from reactions involving the methylated CFeSP.
The reaction of CH 3 -H 4 folate, CO, and CoA to form acetyl-CoA was performed as described previously (21). Under these conditions, the concentration of CODH/ACS is rate-limiting. The reaction was performed in the dark in a glass V-shaped reaction vial capped with a red rubber serum stopper. Details of the reaction mixture are given in the Fig. 2 legend. The reaction was quenched at various times by removing 5-l aliquots into 5 l of 2.2 N perchloric acid. The amount of acetyl-CoA formed was measured by Dowex 50W-H ϩ chromatography as described (27). UV-visible spectra were obtained on an OLIS-modified Cary 14 spectrometer. The data were then analyzed using the program KINSIM (28,29) and modified by Gary Xin Hua and Dr. Bryce Plapp of the University of Iowa. The input rate constants are given under "Results." The simulated data were plotted using SigmaPlot (Jandel Scientific, San Rafael, CA). 3 -H 4 folate, CO, and CoA-We measured the synthesis of acetyl-CoA from CH 3 -H 4 folate, CO, and CoA using the methylated CFeSP as the methyl carrier protein. If the variant CFeSP is used, this reaction is very slow, since the protein is in the Co(II) state and cannot be activated by physiological reductants (20). In the reaction studied here, the first turnover generates cob(I)amide, which undergoes remethylation by CH 3 -H 4 folate or oxidation to cob(II)amide. When the wild type methylated CFeSP is used, acetyl-CoA synthesis continues linearly with time until the limiting reagent is depleted. The concentration of CO (1 mM) is limiting in these reactions. With the C20A variant, the rate of acetyl-CoA synthesis initially is the same as with the wild type protein and then slowly decreases and becomes negligible after ϳ100 turnovers (Figs. 2 and 3A, inset). This reaction was performed using three different concentrations of methylated CFeSP. The data with the variant proteins were fit to single exponential equations to estimate the inactivation rate constants (Fig. 2, solid lines; Table I). These rate constants 2 CO and CODH are required for reductive activation, and CO and ACS are required for acetyl-CoA synthesis. 3 In these studies, we methylated the CFeSP with methyl iodide. This procedure was criticized by others who erroneously stated that our samples contained methyl iodide at concentrations sufficient to inhibit CO/acetyl-CoA exchange and CO oxidation activities (see Footnote 27 of Ref. 7). However, in our studies (25), we followed a protocol established after extensive studies to assure that the methyl-CODH is "viable" (8). After reacting the CFeSP with methyl iodide, the reaction mixture is chromatographed to remove methyl iodide and isolate the methylated CFeSP. Therefore, methyl iodide is absent from all enzymatic reactions in which the methylated CFeSP is used as a methyl donor. The catalytic competence of the methyl-CODH intermediate was demonstrated in three different reactions. First, in an exchange reaction between methylated CFeSP and methyl-CODH, 40% of the methyl groups underwent exchange. Second, in an exchange reaction between methyl-CODH and the methyl group of acetyl-CoA, 90% underwent exchange. Third, by measuring the conversion of methyl-CODH, CO, and CoA to acetyl-CoA, 80% of the methyl groups were converted to acetate or acetyl-CoA. Therefore, the methylation protocol is sound, and the published criticisms relating to this methodology are unjustified. were inversely proportional to the CFeSP concentration. The rate constant for reactivation of the mutant protein (k act ϭ 0.0015 s Ϫ1 ) has been measured earlier (20). It is 30 -90-fold lower than these rates of inactivation. The UV-visible spectrum of the variant CFeSP after 40 min, when the reaction is fully inhibited and acetyl-CoA is no longer being formed, is characteristic of the cob(II)amide state (Fig. 3A).

Synthesis of Acetyl-CoA from the Methylated CFeSP, CH
These results demonstrate that reductive activation is required to regenerate the active form of the CFeSP. After it converts to the Co(II) state, the C20A variant cannot be reactivated at significant rates by its physiological electron donors (CODH and CO), because its high potential 3Fe-4S cluster cannot provide the driving force to reduce Co(II) to Co(I). The wild type protein undergoes reactivation over 4000-fold faster (k act ϭ 0.88 s Ϫ1 ) than the mutant, so the reaction continues unabated until the substrates are depleted. Therefore, a low potential Fe-S cluster in the CFeSP is required for reductive activation of the CFeSP by physiological electron donors.
Single Turnover Kinetics of the Variant Methylated CFeSP-When the wild type methylated CFeSP is reacted with CODH/ ACS, a low potential metal center (cluster A) on ACS undergoes methylation as the CFeSP is converted to the Co(I) state (25). It remained a possibility that there is some degree of radical chemistry in this reaction (see below for discussion). We considered that disabling the electron transfer pathway to the cobalt center might uncover or shift the predominantly S N 2 mechanism to a radical pathway. If so, the initial cobamide product of the methyl transfer reaction would be cob(II)amide, instead of cob(I)amide. We mixed the methylated C20A variant CFeSP and CODH/ACS in the presence of CoA and CO and followed the spectrum of the reaction mixture. The broad absorption band from methyl-Co(III) at 450 nm decreases as the intensity of the 390-nm peak from cob(I)amide increases (Fig.  3B). The reaction yields clean isosbestic points, clearly demonstrating that demethylation of the variant methyl-CFeSP (like the wild type protein) by CODH/ACS is accompanied by conversion of CH 3 -Co(III) to Co(I). The final absorption changes at 390 nm yield a difference extinction coefficient (⌬⑀) of 17 mM Ϫ1 cm Ϫ1 , which indicates complete conversion of methyl-Co(III) to Co(I). The rate of Co(I) formation is 0.21 M min Ϫ1 , which yields a k cat for CODH/ACS under these conditions of 1 min Ϫ1 . This is similar to values measured earlier under these condi-  Table I. [4Fe-4S] Cluster Requirement for B 12 Reductive Activation tions of pH and ionic strength (21). As described above (Fig.  2B), Co(I) undergoes oxidation to Co(II) over a longer time.

DISCUSSION
The Wood-Ljungdahl pathway of acetyl-CoA synthesis is unusual in that the so-called "Western half" (1) of the pathway occurs through enzyme-bound intermediates. The first of these is methylcobamide, formed by transfer of the methyl group from CH 3 -H 4 folate to the CFeSP. Studies of the C. thermoaceticum MeTr and an analogous reaction catalyzed by the E. coli methionine synthase strongly indicate that this reaction occurs by an S N 2 mechanism (6, 20, 30). The results described here clearly show that methylation of Co(I) competes with oxidative inactivation. The rate constant for Co(I) oxidation, calculated by fitting the data shown in Fig. 2 to an exponential, is 0.05 min Ϫ1 . The competition between oxidative inactivation and methyl transfer can also be measured from these data. As shown in Fig. 3A, there is one inactivation event per 100 turnovers, where one turnover consists of 1 mol of acetyl-CoA formed. These two values are internally consistent, since the inactivation rate is 100-fold lower than the turnover number for the acetyl-CoA synthase reaction (60 min Ϫ1 at pH 5.8). A variety of conditions can affect this rate constant. For example, if the protein is exposed to oxygen, the inactivation rate constant increases dramatically. Oxidation of Co(I)-CFeSP with an air-saturated buffer occurs with a rate constant of Ͼ300 s Ϫ1 . 4 However, acetogenic bacteria growing in their natural habitats would rarely experience such high concentrations of oxygen. If the conditions of our assay approximate physiological conditions, the need to extract one electron equivalent from reduced CODH or ferredoxin per 100 mol of acetyl-CoA formed would not be overly demanding.
We observe a striking decay in the rate of acetyl-CoA synthesis when the C20A variant CFeSP is used as the methyl carrier protein (Fig. 2). We can estimate the inactivation rate constant by fitting the data to an exponential equation. To analyze the data more quantitatively, we simulated the progress curve. This procedure requires inputting the relevant rate constants and concentrations of substrates and enzymes for each reaction in this multistep process. We have determined all of the elementary reaction rates and the steady-state kinetic parameters for the MeTr 5 and the CODH (32) reactions. We also have determined the rate of formation of Co(I) from Co(II) for the wild type and C20A CFeSP (20). However, the full set of rate constants for acetyl-CoA synthesis from the methylated CFeSP, CO, and CoA have not yet been measured. A full simulation that included all of these steps would contain about 40 rate constants. We made several simplifying assumptions to wrap all of these reactions into three steps (Scheme 1). The first step is the MeTr-catalyzed methylation of the CFeSP (B ϩ E 3 C ϩ E). The CFeSP and methyl-CFeSP concentrations are below their K m values, and the CH 3 -H 4 folate concentration is much higher than its K m value. Therefore, the rate constants for the methylation the CFeSP and demethylation of the methyl-CFeSP by MeTr can be substituted by the k cat /K m values for the CFeSP and the methylated CFeSP, respectively. Since the CH 3 -H 4 folate concentration (2.5 mM) is much higher than its K m (10 M) value in these reactions, we also omitted this term from the simulation. In addition, we simplified the acetyl-CoA synthesis steps by summarizing them into a single reaction (C ϩ F ϩ G 3 B ϩ F ϩ H). Since the CoA and CO concentrations are significantly above their K m values, the rate of regeneration of Co(I) by CODH/ACS is governed by the k cat /K m for the methylated CFeSP. The k cat for acetyl-CoA synthesis is approximately 1 s Ϫ1 . This was an adjustable parameter. Our first estimates used a K m value of 10 M for the methylated CFeSP, which is similar to that for MeTr. Furthermore, acetyl-CoA synthesis is irreversible at these high concentrations of CO and CoA, so the reverse rate constant is set to 0. For the third step (B % BЈ), the rate of reactivation of Co(II) to Co(I) is 0.88 s Ϫ1 for the wild type and 0.0015 s Ϫ1 for the C20A variant (20) CFeSP. These simplifying assumptions allowed us to focus attention on the oxidative inactivation and reductive activation reactions and quantitatively test the reductive activation hypothesis.
After supplying the above rate constants, we input rate constants for inactivation of Co(I) (beginning with those derived from the exponential fits as a starting point) into the KINSIM program. The dashed lines in Fig. 3    the methylated CFeSP. The simulated progress curves fit the data well, indicating that our simplifying assumptions are reasonable. To better understand the factors that control whether the CFeSP will undergo oxidative inactivation, reductive reactivation, or methylation, we used KINSIM to output the concentrations of the three states of the CFeSP, Co(I), Co(II), and methyl-Co(III) for the mutant and wild type proteins. Fig. 4 shows these species for the reaction shown in Fig. 2A with 8 M CFeSP. The pre-steady state reaction is identical for the mutant and the wild type proteins. With the wild type protein, the concentration of Co(II) quickly reaches a value that is about 2% of the total CFeSP present and slowly decays to about 0.4%. The predominant species at the end of the reaction is methyl-Co(III). However, with the C20A mutant, Co(II) increases in an exponential fashion as Co(I) decays. At the end of the reaction, there is a mixture of methyl-Co(III) (32% of the total CFeSP) and Co(II) (58% of the total). In both reactions, there is a significant amount of methyl-Co(III) at the end of the reaction, because the CH 3 -H 4 folate concentration is higher than that of CO. The essential parameters are the relative rates of inactivation relative to the rates of reactivation and Co(I) methylation. With an inactivation rate constant of about 0.1 min Ϫ1 , the reaction profile begins to deviate from linearity even when the reactivation rate is as high as the inactivation rate constant. Interestingly, increasing the reactivation rate constant above the measured value has minimal effect on the progress curve. For example, increasing the reactivation rate 10-fold (to 8.8 min Ϫ1 ) increases the slope of the reaction's progress curve by only 3%. This indicates that the environment and location that nature has selected for the FeS cluster in the CFeSP is nearly optimal for its role in reductive activation.
The redox potential for the [4Fe-4S] 2ϩ/1ϩ couple is about 20 mV more negative than that of the Co(II)/(I) couple and is nearly equal to that of the CO 2 /CO couple. The design of a low potential cluster is clearly important; however, based on a purely electrochemical argument, one would expect CO to reduce the Co(II) state about as easily as the cluster 2ϩ state.
However, the requirement for the cluster in reductive activation of Co(II)-CFeSP by physiological electron donors is nearly absolute. The cluster is located in a separate subunit from the cobamide. Solving the crystal structures of the CFeSP, MeTr, and CODH/ACS would greatly enhance our understanding of how these proteins interact in electron and methyl transfer reactions.
The studies described here also provide insight into the mechanism of methyl group transfer from the methylated CFeSP to CODH/ACS. There are three possible mechanisms (33). The first is heterolytic cleavage of the Co-C bond by an S N 2-type mechanism involving attack by a nucleophilic center of ACS (Mechanism I). This nucleophilic center has not been unambiguously identified. Evidence from our laboratory, summarized in recent reviews (1,2,34,35), and from Lindahl's laboratory (7) indicates that the nucleophile is the nickel site of cluster A. We favor a mechanism involving a Ni(I) nucleophile; however, Barondeau and Lindahl (7) favor a Ni(II) nucleophile. A second possibility (Mechanism II) is a homolytic mechanism in which electron transfer from the reduced Fe-S cluster of the CFeSP would form a methyl-Co(II) species that would disproportionate to form Co(I) and methyl-nickel. 6 A third possibility (Mechanism III), that the CH 3 -Co(III) species is cleaved homolytically to form a CH 3 -Ni species and Co(II), is inconsistent with stopped flow studies of the wild type CFeSP. Co(I) forms at the same rate that methyl-Co(III) decays, apparently ruling out a Co(II) intermediate (36). Mechanisms II and III invoke transfer of a methyl radical to CODH/ACS.
Methylation of CODH/ACS has been modeled using compounds such as CH 3 -Co(dmgBF 2 )py (where dmgBF 2 represents (difluoroboryl)dimethylglyoximato and py represents pyridine) as the methyl group donor complex and Ni(tmc) ϩ (where tmc represents 1,4,8,11-tetramethyl-1,4,8,1-tetraazacyclotetradecane) as the methyl acceptor (37,38). Two equivalents of the Ni(I) complex were required, one to reduce methyl-Co (III) to methyl-Co(II) and the other to capture the methyl radical generated upon cleavage of the methyl-Co(II) species. Thus, studies of the inorganic model system support Mechanism II.
On the other hand, several criteria favor Mechanism I for the analogous enzymatic methyl transfer. Martin and Finke (39) noted the thermodynamic problem with mechanism II: reduction of CH 3 -Co(III) requires redox potentials that are too low for physiological electron donors. In the case of model compounds, Co-C bond homolysis is possible, since the Ni(tmc)(II/I) couple, which has a reduction potential of Ϫ1.18 V (versus SHE), is capable of reducing CH 3 -Co(III), with an E m of Ϫ1.2 V. However, neither CO nor any of the redox centers present in CODH/ACS from C. thermoaceticum have E m values below Ϫ550 mV (8, 40 -43). Thus, CODH is too weak as a reductant to reduce methyl-cob(III)amide. We were unable to detect any reduction or cleavage of the methyl-Co(III) state of the CFeSP after several hours of incubation in the presence of CO and CODH/ACS (9). Mechanism I is also supported by studies of acetyl-CoA synthesis using chiral CH 3 -H 4 folate, which is converted to acetyl-CoA with retention of configuration (44). The most straightforward interpretation of these studies is that transfer of the methyl group to CODH occurs with inversion of configuration, as expected for an S N 2 type displacement. Although reactions exist in which the radical is transferred before it has the chance to rotate (and thus randomize), if the methyl group is transferred as a radical species, racemization would be the most likely outcome.
Further support for Mechanism I comes from recent studies FIG. 4. Simulation of the cobalt states for the mutant and wild type enzymes during acetyl-CoA synthesis. The simulation was performed for the 8 M CFeSP using the parameters shown in Table I and following Scheme I. using the variant C20A CFeSP (20). The methylated C20A variant, which cannot accept electrons from CODH and, therefore, cannot reduce methyl-Co(III) to methyl-Co(II), generates acetyl-CoA as rapidly as the wild type protein. This result appears to be inconsistent with Mechanism II. However, this study did not evaluate the redox state of the CFeSP during the reaction. The finding that electron transfer to the cobalt center is crippled in the variant offers the possibility of uncovering reaction intermediates that would be masked by the rapid reduction by CODH of the wild type protein. Furthermore, it is possible that a radical pathway (such as Mechanism III), which may be a minor component of the reaction with the wild type protein, could become the dominant pathway used by the variant form of the CFeSP. However, just as with the wild type protein, demethylation of methyl-Co(III) is accompanied by formation of Co(I), which is in redox equilibrium with Co(II). We conclude that methylation of CODH/ACS, like the CH 3 -H 4 folate-dependent methylation of the CFeSP, proceeds through an SN 2 displacement mechanism and does not involve radical chemistry. An important question remains. What conditions determine whether a methyl transfer reaction will occur through a homolytic or heterolytic mechanism?