Thiol‐Disulfide Exchange Kinetics and Redox Potential of the Coenzyme M and Coenzyme B Heterodisulfide, an Electron Acceptor Coupled to Energy Conservation in Methanogenic Archaea

Methanogenic and methanotrophic archaea play important roles in the global carbon cycle by interconverting CO2 and methane. To conserve energy from these metabolic pathways that happen close to the thermodynamic equilibrium, specific electron carriers have evolved to balance the redox potentials between key steps. Reduced ferredoxins required to activate CO2 are provided by energetical coupling to the reduction of the high‐potential heterodisulfide (HDS) of coenzyme M (2‐mercaptoethanesulfonate) and coenzyme B (7‐mercaptoheptanoylthreonine phosphate). While the standard redox potential of this important HDS has been determined previously to be −143 mV (Tietze et al. 2003 DOI: 10.1002/cbic.200390053 ), we have measured thiol disulfide exchange kinetics and reassessed this value by equilibrating thiol‐disulfide mixtures of coenzyme M, coenzyme B, and mercaptoethanol. We determined the redox potential of the HDS of coenzyme M and coenzyme B to be −16.4±1.7 mV relative to the reference thiol mercaptoethanol (E0’=−264 mV). The resulting E0’ values are −281 mV for the HDS, −271 mV for the homodisulfide of coenzyme M, and −270 mV for the homodisulfide of coenzyme B. We discuss the importance of these updated values for the physiology of methanogenic and methanotrophic archaea and their implications in terms of energy conservation.


Introduction
Methanogenic archaea play an important role in the global carbon cycle by producing around 1 Gt of methane per year. [1]mong these microbes, hydrogenotrophic methanogens generate methane by complete reduction of CO 2 utilising electrons from hydrogen (4 H 2 per CO 2 ).Methanogens without cytochromes are able to grow at H 2 partial pressures below 10 Pa, [2] which allow only ca.0.5 ATP to be conserved per methane produced.Under these conditions, all reaction steps proceed close to the thermodynamic equilibrium and allow the pathway to be reversible. [3]This reversibility is exemplified by the process of anaerobic oxidation of methane, [4] where the homologous enzymes are utilised to carry out the reverse oxidative catabolism, and methane-derived electrons are delivered to sulfate-reducing partner bacteria [5,6] or coupled to nitrate reduction. [7]To enable this reversible metabolism, nature has evolved the tetrahydromethanopterin (H 4 MPT) cofactor and dedicated redox-coenzymes, such as F 420 and the heterodisulfide (HDS) of Coenzyme M (CoMÀ SH or MSH) and Coenzyme B (CoBÀ SH or BSH) to balance the steady state concentrations of metabolites.
The steps of CO 2 reduction to a methyl group in hydrogenotrophic methanogenesis are part of the reductive acetyl-CoA pathway (also known as Wood-Ljungdahl pathway).This pathway, considered to be the most energy-efficient CO 2 fixation pathway in nature, [8] is found both in methanogens and acetogens.While the acetogen-specific CO 2 -reduction pathway requires 1 ATP to be invested to load a formyl group from formate onto the carbon carrier tetrahydrofolate, [9] no direct investment of ATP is required in the first step of the methanogen-specific pathway to activate the CO 2 molecule and reduce it to a formyl-group.To enable this CO 2 activation step without ATP, the electron supply must be provided by a low potential electron donor, in the form of ferredoxins (Fd À /Fd, E 0 ' estimated to be À 500 mV [3] ).
The redox potential of molecular H 2 /H + at environmental partial pressure of H 2 and pH is not low enough to reduce a sufficient fraction of the ferredoxin pool to allow CO 2 activation, particularly at low hydrogen concentrations where methanogens are still able to thrive.Two different strategies have evolved in archaea to enable ferredoxin reduction with hydrogen-derived electrons.Both strategies are either indirectly or directly dependent on the reduction of the heterodisulfide of coenzyme M and coenzyme B (CoMÀ SÀ SÀ CoB or BSSM) to free thiols, which relates to the last step in hydrogenotrophic methanogenesis [10] (Figure 1).
In cytochrome-containing methanogens (e. g.Methanosarcinales), the HDS CoBÀ SÀ SÀ CoM is reduced to free thiols by the subunits D and E of the heterodisulfide reductase complex (Hdr) (Figure 1-A) using the membrane-residing methanophenazine as electron donor.The reduced methanophenazine pool is itself regenerated by consuming intracellular redox cofactors (F 420 , ferredoxins) in electron-translocating reactions, or external H 2 .
On the other hand, in methanogens lacking cytochromes (e. g. in Methanothermobacter), the MvhAGD-HdrABC complex catalyses the thermodynamically unfavorable reduction of ferredoxin by H 2 with the favorable reduction of CoMÀ SÀ SÀ CoB with H 2 [11] (Figure 1-B).From the reported standard redox potentials (E 0 ') of the involved species, the free energy change of this reaction in standard conditions ΔG 0 ' could be estimated to be À 52 kJ•mol À 1 . [12]Cases have also been reported where HdrABC can be complexed with a formate dehydrogenase [13] or has F 420 -accepting variants, [14] that expand the range of electron donors contributing to HDS reduction.Furthermore, the HDS CoBÀ SÀ SÀ CoM seems to be involved in other anabolic processes: The existence of heterodisulfide reductases using succinate as an electron donor, [15] implies a connection between the HDS reduction and the citric acid cycle.Additionally, the apparent essentiality of methyl coenzyme M reductase (Mcr) for a mutant methanogen not relying on methanogenesis for its energy metabolism [16] supports the requirement of HDS reduction for anabolic purposes.
Understanding energy conservation processes in methanogens requires knowledge of the redox potentials of the involved redox cofactors.The HDS CoMÀ SÀ SÀ CoB and its reduction play a central metabolic role: When catalysed by HdrDE, HDS reduction is directly linked to energy conservation in the establishment of an ion gradient.When involved in an electron bifurcating reaction, the HDS reduction is coupled with 2 other half redox reactions.The redox potential of the electron donors (hydrogen, F 420 , or formate) is known, but the redox potential of ferredoxins (second electron acceptor) is more difficult to measure, since it depends on the individual ferredoxin (which ranges from À 320 mV to À 700 mV [17] ).Furthermore, there is no consensus on the number of electrons (1 or 2) delivered per ferredoxins in redox reactions.Thus, characterising the redox properties of the HDS CoMÀ SÀ SÀ CoB, is necessary to detangle the different associated redox processes.
The standard redox potential of the HDS CoBÀ SÀ SÀ CoM has been measured via an electrochemical method, and was established to be À 143 � 10 mV relative to a standard hydrogen electrode (SHE), [18] a value that is substantially higher than e. g. glutathione (E 0 ' = À 262 mV) or most of other disulfides (see Table 1).
To re-assess the redox potential of this important HDS and to probe its reactivity, we measured the thiol-disulfide exchange kinetics between coenzyme M and coenzyme B under physiological conditions.We then equilibrated coenzyme B, coenzyme M, and mercaptoethanol thiols/disulfides and accurately quantified the concentrations of each 3 thiol and 6 disulfide species in the mixtures via 1 H-NMR spectroscopy.26][27]

Chemicals
Coenzyme M and mercaptoethanol (ESH) were purchased from Alfa Aesar (cat.no.J63989.14 and J66742).Mercaptoethanol disulfide (2-hydroxyethyl disulfide, ESSE) was purchased from Thermofischer Scientific (cat.no.380474).Coenzyme B and coenzyme B homodisulfide (BSSB) were synthetised and available from experiments described in Scheller et al., 2010. [28]oenzyme M disulfide (MSSM) was prepared by oxidation with a stoichiometric amount of KI 3 solution of commercial coenzyme M, followed by twofold precipitation from acetone and re-crystallization as described for the synthesis of methyl-Scoenzyme M. [28] KH 2 PO 4 1.00 M solution, K 2 HPO 4 1.00 M solution, thiols solutions, disulfides solutions, and 10 mM imidazole were prepared in an anaerobic chamber (3-5 % H 2 in N 2 , Coy Laboratory Products) using anoxic water.1.00 % dioxane in D 2 O was sparged under N 2 before being used in the anaerobic chamber.1.00 M potassium phosphate buffer pH 7.0 was prepared by mixing 619 μL of KH 2 PO 4 (1.00 M) and 381 μL of K 2 HPO 4 (1.00 M).

Kinetic assays for thiol-disulfide exchange reactions
Liquid components (mercaptoethanol and its disulfide, 0.25 % dioxane in D 2 O, 1.00 M K 2 HPO 4 , 1.00 M KH 2 PO 4 , D 2 O, and H 2 O) were sparged with N 2 and brought into an anaerobic chamber.Solid components were weighed into Eppendorf tubes and brought into an anaerobic chamber.Solutions were prepared in an anaerobic chamber using anoxic water and neutralised using 1.00 M K 2 HPO 4 to pH = 7. Thiol and disulfide stock solutions were 100 mM and 200 mM, respectively.Individual assay mixtures (1.0 mL in 1.5 mL Eppendorf tubes) were prepared in an anaerobic chamber (T = 25 °C) to a final concentration of 10 % D 2 O, 0.1 M potassium phosphate buffer pH = 7.0 and 0.025 % dioxane.4 assays were prepared in the direction of MSSM reduction by BSH (experiments B1-4) and 1 complementary assay in the reverse direction (BSSB reduction by MSH, experiment M).The initial concentrations of thiols and disulfide for each assay were set as follows: B1/B2 -Coenzyme B 27.7 mM, Coenzyme M disulfide 9.9 mM; B3/B4 -Coenzyme B 18.0 mM, Coenzyme M disulfide 11.0 mM; M -Coenzyme M 20.0 mM, Coenzyme B disulfide 9.4 mM.
The reactions were started by the addition of the thiol stock solution.The solutions were immediately mixed, transferred to NMR tubes, and closed with a gas-tight NMR septum.The tubes were brought out of the chamber and placed into a styrofoam box containing a thermometer and ca. 1 L of water at exactly 25 °C.The box was carried to the NMR facility and the samples inserted into the spectrometer that had a default setting of 25 °C.Measurement time points correspond to the time between adding the thiol stock solution to the assay and the start time of each NMR spectrum acquisition (8 scans, measurement time = 80 s).The shortest time point obtainable was Table 1.Non-exhaustive list of reported standard redox potentials of various monothiols/disulfide systems.
b] See methods.
between 9 and 11 min.For the first time points, the samples were kept in the NMR spectrometer and re-measured several times.For later time points, the samples were stored back in the same water-filled styrofoam box, which was placed into an incubator at 25 °C.

Quantification of thiol-disulfide exchanges equilibrium concentrations
Thiol and disulfide stock solutions concentrations were measured by quantitative 1 H-NMR spectroscopy to determine their exact concentration.1.0 M potassium phosphate buffer was used to adjust the pH of each solution to 7.0.Assays (0.7 mL) were mixed in the anaerobic chamber.Each assay contained 10 % D 2 O, 0.10 M potassium phosphate buffer, 0.10 % dioxane as an internal standard for quantification, and a combination of thiols and disulfides.1.0 mM imidazole was also added as an internal pH standard.The imidazole signal from the À NHÀ CHÀ NÀ group stayed within a range of � 0.03 ppm for all experiments carried out, which corresponds to a maximal pH difference of � 0.03 units at pH 7. [29] The concentrations of thiols and disulfides for each assay were set as described in Supplementary Table I Signals for each thiol/disulfide were controlled so that their integrated value would match the ones for other signals of the same molecule, to confirm their suitability for population quantification.Surprisingly, a decrease of the À CH 2 SH signal from coenzyme B was observed (see "Influence of different 1 H-NMR measurement conditions on quantification of relevant signals" and Supplementary Figure A).Alternative NMR methods were tested but only partially solved this issue and were generally less reliable for quantification (see Supplementary Figure B).Only the data acquired with our default measurement method was used for this study.Populations at each time point were normalized by being fit to the dioxane signal.Experimental time course populations were converted to absolute concentrations using dioxane population as reference.

Determination of kinetic rate constants for exchange reactions between CoB and CoM thiols/disulfides
The concentrations of individual thiols and disulfides were processed in CoPaSi v4.40 (Build 278), using a 2-thiol-moieties reaction system defined as one fixed compartment with BSH, MSH, BSSB, MSSM, and BSSM as species with an equal weight of 1. Two reactions, R1 and R2, their respective rate constants k 1 , k 2 (forward), k -1 , k -2 (backward) and their equilibrium constants K 1 and K 2 (set as ratios of k 1 over k -1 and k 2 over k -2 respectively) were set according to the two-step reaction mechanism of a thiol-disulfide exchange reaction where a heterodisulfide intermediate is formed (Eq.1), with RC = CoM and R' = CoB.
The parameter estimation was run on each assay's dataset independently over the data points obtained over a timeframe of 5 hours.The estimation was run using a genetic algorithm on a population of 1,000 elements over 10,000 generations.Start values for kinetic constants were randomized between 10 À 6 M À 1 •s À 1 and 10 6 M À 1 •s À 1 .Start values for species were set between 10 6 and the first measured value for introduced reagents, and between the first measured value and 0 for the formed products.Equilibrium constants values were added as constraints to be between the confidence interval of the values defined in the thermodynamic studies ([2.04;2.16]for K 1 and [0.41;0.45]for K 2 ) and with an overtime rate set to zero.The estimated parameters were then used as inputs for time course simulations of 30,000 s with 1,000 intervals.Combined averages X À Combined and standard deviations SD Combined were calculated as shown in Eq. 2 and Eq. 3, for a number i of average values X À i , standard deviations SD i and number of elements n i (here n i = 10,000, corresponding to the number of generations).
SD Combined ¼ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi

Calculation of standard redox potentials from equilibrium concentrations
The populations at each reaction's equilibrium were deduced from 600 MHz NMR data (400 MHz NMR data available in supporting information) and used to calculate the equilibrium constant for the different thiol-disulfide exchange reactions.The calculation of an equilibrium constant of a homo-disulfide relative to a different thiol (K 12 , see Eq. 4) does not involve the formed heterodisulfide concentration.Thus, the equilibrium constant for the reduction reaction of a disulfide ASSB by a thiol CSH can be generalized as shown in Eq. 5.The differences of redox potentials ΔE 0 ' were obtained using the Nernst equation (Eq.6) where R is the ideal gas constant (8.314J•mol À 1 •K À 1 ), T the temperature (here 298.15 K), n the number of electrons transferred in the reaction (here n = 2) and F the Faraday constant (96,485 C•mol À 1 ).

Calculation of heterodisulfides distribution over homodisulfides
The equilibrium constant for the formation of a heterodisulfide from the two corresponding homodisulfides (K HDS ) was calculated using Eq.7 where K 1 and K 2 are the equilibrium constants from the reactions shown in Eq. 1. K HDS was used to calculate the disulfide distribution constants K pref , by dividing this constant to a factor of 4 corresponding to the theoretical equilibrium constant value for a statistical distribution of homoand heterodisulfide at the equilibrium (Eq.8, see supporting information "Statistical aspect of thiol-disulfide exchange reaction").
The 95 % confidence interval over each averaged value was calculated using a Student's t distribution considering 6 replicates.

Accurate quantification of chemical species from thioldisulfide exchange experiments by 1 H-NMR spectrum simulation
Fitting the acquired 1 H-NMR spectra to simulated spectra (Figure 2) allowed accurate quantification of all species from equilibrated reaction mixtures containing CoM, CoB and mercaptoethanol thiols and disulfides.Manual spectrum simulation was preferred over integration values for quantification purposes for its efficiency to process overlapping spectra.Furthermore, spectrum simulation takes the coupling constants between different signals of the same molecule into account, while multiple analysis algorithms using Global Spectrum Deconvolution (GSD) do not and are also sensitive to background noise.
The only obstacle encountered was a decreased signal of the methylene group bonded to the thiol of coenzyme B when water presaturation was used.The physical cause for the observed water presaturation effect on the relevant BSH signal could not be elucidated.An empirical correction factor was the most suitable way to match the actual CoBÀ SH concentration of the mixture (see Supplementary Figure B).If this correction factor would not be applied, the resulting redox potential values would differ by 5 mV from the values reported in this study.

Determination of rate constants of coenzyme B and coenzyme M thiol-disulfide exchange by S N 2 reaction kinetic simulations
Exchange reactions between CoM and CoB respective disulfides and thiols were monitored over time by quantitative 1 H-NMR spectroscopy.The values for the kinetic constants described in Eq.1 were determined from the time courses resulting from the addition of MSSM to BSH (Figure 3-B1, B2, B3, B4) and inversely, after the addition of MSH to BSSB (Figure 3-M).The kinetic constants, in accordance with the reaction model shown in Eq. 1, were estimated to the following values: k 1 = (18 �

4)×10
Exchange reactions between thiols and disulfides proceed via S N 2-type nucleophilic substitution mechanism and thus, show a first order dependency on the thiol as well as the disulfide concentration.Despite the quite large error associated with the fitted kinetic constants, the determined values for the CoB/CoM thiol/disulfide system are in range with previously described values for a glutathione/coenzyme A system. [30] As in that glutathione/coenzyme A system, our measurements of CoM and CoB species also showed higher rates towards the formation of the heterodisulfide.
In our kinetic experiments, in which we used ca.20 mM of thiol and ca. 10 mM of disulfide, the resulting heterodisulfide is turned over about 2.3 • 10 À 3 M•h À 1 at 25 °C (using B1 data).The thiol-disulfide exchange rate is fast enough to fully equilibrate CoB, CoM, and mercaptoethanol thiol-disulfide mixtures after 7 days under these conditions.
Under physiological conditions, where the concentrations of thiols and disulfides are probably at least 10 times lower, this non-enzymatic thiol-disulfide exchange time will be at least 100 times lower, corresponding to a rate of 2.3 • 10 À 5 M•h À 1 .At 65 °C, this non-enzymatic thiol-disulfide exchange would be about 5.8 • 10 À 4 M•h À 1 (increased by a factor of 25, as calculated via Arrhenius equation where the pre-exponential factor is assumed to be proportional to the temperature, using an activation energy of 65 kJ•mol À 1 [31] ).The enzyme-catalysed formation rate of HDS corresponds to the methane production rate since, for every HDS, one molecule of methane is produced by the enzyme Mcr (Figure 1).Considering that the maximal methanogenesis rate of M. marburgensis at 65 °C is 176 mmol per grams of cells (dry weight) per hour, [32] and that 90 % of the cell mass corresponds to the water fraction of the cytosol, [33] this in vivo turnover number can be estimated to be ca.19.5 M•h À 1 .This enzyme-catalysed rate is at least 3 • 10 4 times faster than the non-enzyme-catalysed exchange.Even for methanogens producing methane considerably more slowly, the non-enzymatic thiol-disulfide exchange remains slow compared to the enzyme-catalysed process.

Relative redox potentials of disulfide combinations of coenzyme M, coenzyme B, and mercaptoethanol to each other
To determine standard redox potentials of CoB/CoM thiols (BSH, MSH) and their disulfides (MSSM, BSSB, and BSSM), we have carried out a comprehensive set of experiments including combinations of CoB, CoM, and mercaptoethanol thiol/disulfide exchange reactions.We consider our data to be reliable, since the measured kinetic constants confirmed that the equilibra must have been reached after our incubation time, and that these equilibria were reached from both directions.
Despite the glutathione/glutathione disulfide couple being the common reference for relative thiol redox potential measurements, we used mercaptoethanol and its homodisulfide (ESH/ESSE) as the reference thiol-disulfide couple for our experiments because "glutathione-moiety" signals overlap with the signals of interest of the quantified species.Furthermore, the signals of glutathione are more difficult to evaluate since the two hydrogen atoms of the À CH 2 SH group are diastereotopic, which complicates spectra simulation.
The measured equilibrium concentrations for each thiol and disulfide in all experiments (Supplementary Table II) were used to calculate the relative redox potentials of each homo-or heterodisulfide to specific thiols (ΔE 0 ', Table 2).
Table 2. Relative standard redox potential of all disulfide combinations between coenzyme M, coenzyme B, and mercaptoethanol.

Formation preference of heterodisulfides over homodisulfides
We calculated the preference for heterodisulfide formation relative to the statistically expected fraction of disulfides using Eq.7 and 8 from the data in Supplementary Table II.The results for all 3 heterodisulfides are provided in Table 3, whereby a K pref > 1 indicates that heterodisulfides of CoM, CoB, or mercaptoethanol are slightly preferentially formed compared to their respective homodisulfide relative to the statistically expected distribution (see supporting information "Statistical aspect of thiol-disulfide exchange reaction").

Standard redox potentials of disulfide combinations of coenzyme M, coenzyme B, and mercaptoethanol
We report E 0 ' = À 281 mV value for the standard redox potential of the HDS BSSM.This value is 10.5 mV lower than the measured standard redox potential of BSSB and MSSM, respectively À 270 mV and À 271 mV (Table 4-E 0 '), even though there is only a slight preference for heterodisulfide formation (K pref = 1.22).The main difference is explained by the formal definition of the redox potential of the heterodisulfide, where the homodisulfides are not considered in the calculation of the redox potential.If K pref = 1, the statistical distribution on disulfide reduction causes a 0.5 difference between the resulting equilibrium constants for homodisulfide and heterodisulfide reduction which, in redox potential calculations, causes a difference of 8.9 mV (see supporting information "Statistical aspect of thiol-disulfide exchange reaction").The observed difference in redox potential between homo and heterodisulfides is thus mostly related to a probabilistic aspect of the exchange reaction in a thiol-disulfide system (i.e. using a thiol as a reducing agent).Under physiological conditions in methanogens and anaerobic methanotrophic archaea, putatively most of the oxidized forms of CoM and CoB are expected to be present as heterodisulfide.Homodisulfides that are slowly formed via the exchange reactions measured here are probably efficiently removed via the NAD(P)H-dependent CoM-homodisulfide reductase. [34]If the amount of homodisulfide BSSB or MSSM formed is negligible due to the high turnover rate of enzymes oxidizing the BSH and MSH thiols selectively to the heterodisulfide, the use of the redox potential value excluding this statistical factor E 0 '* = À 272 mV, is recommended to be used.
Furthermore, the actual in vivo redox potential might be higher considering the low thiol/disulfides concentrations.Indeed, the standard redox potential E 0 ' is defined at 1 molar concentration of thiols and disulfides each.As one molecule is converted to 2, the redox potential (E') increases at lower concentrations.At 1 mM concentrations of thiols and disulfides, the equilibrium constant deviates by a standard factor of 10 3 , resulting in values that are 89 mV less negative.These standard conditions using 1 mM concentrations, are referred to as E' m [35] (Table 4 -E' m À E' m *).
Considering E 0 ' = À 262 mV for the reference thiol Glutathione [19] and ΔE 0 ' = À 2.3 mV for the relative redox potential of mercaptoethanol disulfide to glutathione, [24] we used the value E 0 ' = À 264 mV for mercaptoethanol standard redox potential.This calculation method was preferred from directly using published values for mercaptoethanol standard redox potential, as the standard redox potential of glutathione relative to NADPH was reported to be accurately measured in standard conditions.The equilibrium constant of the exchange reaction between mercaptoethanol and glutathione disulfide in standard conditions (K eq = 1.2) is a consensual value that does not rely on a potentially erroneous reference. [24]b] K pref : Disulfide distribution constant (see supporting information "Statistical aspect of thiol-disulfide exchange reaction").

Table 4.
Standard redox potentials of coenzyme M, coenzyme B, and mercaptoethanol homo-and heterodisulfides relative to the SHE.À 183 [a] Values deduced using the published standard redox potential of Glutathione and the ΔE 0 ' of mercaptoethanol disulfide to glutathione.The uncertainties represent the precision of the measurement.c] *: Values calculated without the À 8.9 mV factor relative to heterodisulfide statistical distribution.
potentials.Overall, the largest source of error in our calculated redox potentials comes from the ultimate reference NADPH (E 0 ' = À 323 mV), for which an error of � 10 mV is associated. [19]hese values might need to be slightly adjusted, if a new "best estimation" for the redox potential of glutathione would be released.
Our determined standard redox potentials of À 281 mV for the HDS CoBÀ SÀ SÀ CoM, À 271 mV for CoMÀ SÀ SÀ CoM, and À 270 mV for CoBÀ SÀ SÀ CoB relative to SHE contrast with the previous report, where the standard redox potentials of the HDS CoBÀ SÀ SÀ CoM was determined to be À 143 � 10 mV, and those of the homo-disulfides CoBÀ SÀ SÀ CoB and CoMÀ SÀ SÀ CoM to be respectively À 139 � 5 mV and À 177 � 7 mV relative to SHE. [18] The reason for this discrepancy might be due to the use of a hanging mercury drop working electrode to directly measure redox potentials via cyclic voltammetry in the previous study, because the formation of stable metal-thiolate complexes with the electrode can bias the measured value. [19,36]In addition, the E' of a thiol-disulfide couple is strongly dependent on the concentrations, even if the concentrations of thiols and disulfides are equal, and exact measurement conditions have not been provided, or been considered for the calculation of E 0 ' in those studies.Overall, CoM and CoB thiols have comparable redox properties to those of thiols involved in biological systems (see Table 1).

Implications of the measured heterodisulfide standard redox potential for energy conservation in methanogenesis pathways
The difference between our value for CoBÀ SÀ SÀ CoM (E 0 ' = À 281 mV) and the previously reported value [18] is ΔE 0 ' = À 138 mV.For HDS reduction, the difference in the standard redox potential translates to a ΔΔG 0 ' of + 26.6 kJ•mol À 1 , which corresponds to the energy value of about 1 = 2 ATP under physiological conditions.Since the overall thermodynamics of methanogenesis pathways are known, updating the free energy for the HDS reduction reaction (ΔG 0 ' = À 25.7 kJ•mol À 1 ) requires the adjustment of the free energies of other reactions by À 26.6 kJ•mol À 1 to match the overall values.
For the reduction of methanol with hydrogen to methane (ΔG 0 ' = À 112.5 kJ•mol À 1 [1] ), only 3 reactions are involved: A methyl transfer from methanol (MeOH) to coenzyme M (Mta step), the reduction of methyl coenzyme M (MeÀ SÀ CoM) to methane and generation of the HDS (Mcr step), and the reduction of the HDS to CoMÀ SH and CoBÀ SH (Hdr step).The currently reported thermodynamic values for these 3 reactions are À 27.5 kJ•mol À 1 , [37] À 30 kJ•mol À 1 , [38] and À 52.3 kJ•mol À 1 , [12] respectively.Considering our remeasurement, À 26.6 kJ•mol À 1 remains to be attributed between the Mta and the Mcr steps.The reported value for the Mta step has been deduced from values relying on bond-energies calculations, [37] which do not take entropic factors into account and cannot be considered reliable.To our knowledge, there was no experimental data available to estimate the free energy of the Mta step.The free energy for the Mcr step is also difficult to estimate.This value, which had been deduced from the known steps and the overall reaction thermodynamics, was initially reported to be ΔG 0 ' = À 45 kJ•mol À 1 [1,39] and was later adjusted to ΔG 0 ' = À 30 kJ•mol À 1 [40] after the publication of the "erroneous" HDS E°' value. [18]Regarding the available experimental data that could help estimating the ΔG 0 ' of the Mcr step, only one single experiment involving purified Mcr has been reported where MeÀ SÀ CoM was directly formed from methane and the heterodisulfide at 60 °C. [41]Assuming that the equilibrium had been reached under these conditions and that the CoBÀ SH concentration had equaled the MeÀ SÀ CoM concentration during incubation with Mcr, the equilibrium constant K = 10 À 8 is obtained in the methane oxidizing direction, which corresponds to a ΔG 0 ' of À 51 kJ•mol À 1 .Using ΔG 0 ' = À 51 kJ•mol À 1 as a minimum value for the Mcr step implies a maximum value for the Mta step of ΔG 0 ' = À 35.8 kJ•mol À 1 (at least 7.4 kJ•mol À 1 lower than the previously reported value).
Regarding hydrogenotrophic methanogenesis (ΔG 0 ' = À 131 kJ•mol À 1 ), aside from the already discussed uncertainty of the Mcr step value, the part of the pathway from CO 2 to MeÀ SÀ CoM (figure 1) needs to be adjusted by À 7.4 kJ•mol À 1 when assuming a value ΔG 0 ' = À 51 kJ•mol À 1 for the Mcr step.To obtain a more refined overview of free energies within methanogenesis pathways, additional thermodynamic data need to be acquired.A promising next step would be the direct measurement of the Mta reaction in methanogens (equilibrium concentrations of methanol, CoMÀ SH, and MeÀ SÀ CoM), and of the acetogenic version of Mta (equilibrium concentrations between methanol, tetrahydrofolate, and methyl-tetrahydrofolate).

Conclusions
The accuracy of standard redox potentials and free energy changes is of major interest to assess energy conservation along metabolic pathways that involve redox reactions, especially when these are not clearly established, hypothetical or synthetic.
The standard redox potential of the heterodisulfide CoBÀ SÀ SÀ CoM was determined to be E 0 ' = À 281 mV.For reactions happening in biological context, the use of the value E 0 '* = À 272 mV that assumes CoBÀ SÀ SÀ CoM to be present exclusively as the heterodisulfide without significant equilibration to homodisulfides, is preferable.We estimate this E 0 '* value to be justified, since our kinetic data imply that the in vivo enzyme-catalysed selective formation of the heterodisulfide is much faster than exchange reactions that would lead to a mixture of hetero-and homodisulfides.
While the redox potential of the HDS is 138 mV lower than previously stated, the electron bifurcation and potential difference in the reaction catalysed by HdrABC-MvhAGD remains substantial.Hence, no conceptual mistake has been discovered, but thermodynamic values of individual steps in all methanogenesis pathways must be re-adjusted to match the given overall free energy changes.

1 H
-NMR-spectra simulations and fitting for quantification of thiols and disulfides1 H-NMR spectra simulations and fitting to experimental spectra were performed on iNMR v6.4.5.Only the À CH 2 SÀ signals for thiols and disulfides between 3.05 and 2.40 ppm and the dioxane reference signal were simulated (see supporting information "Detailed procedures for NMR spectra simulation").An empirical correction factor was applied to the intensity of the coenzyme B signal used for quantification (1.492 for 400 MHz measurements and 1.403 for 600 MHz measurements, see Supplementary FigureAand "Calculation the correction factor for coenzyme B quantification").

Figure 2 .
Figure 2. Example of quantification of all individual species present in a thiol-disulfide exchange reaction (species at the equilibrium from experiment 9, see Supplementary TableI).The range shown between 2.45 ppm and 3.05 ppm contains the signals of the À CH 2 À SÀ groups of the studied thiols and disulfides.(A) Individual 1 H-NMR spectra of the 3 thiols and the 3 homodisulfides (B) 1 H-NMR spectrum of the resulting exquilibrated mixture, which contains 3 thiols, 3 homodisulfides, and the corresponding 3 heterodisulfides.(C) Simulated 1 H-NMR spectrum of the expected mixture fit to the experimental spectrum B. (D) Individual signals of each species composing the simulated spectrum C, allowing the quantification of each thiol, homo-and heterodisulfide.Thiols (RSH) and disulfides (RSSR') are shown as combination of "thiol moieties" (R, R').M: coenzyme M moiety; B: Coenzyme B moiety; E: mercaptoethanol moiety.
Figure 2. Example of quantification of all individual species present in a thiol-disulfide exchange reaction (species at the equilibrium from experiment 9, see Supplementary TableI).The range shown between 2.45 ppm and 3.05 ppm contains the signals of the À CH 2 À SÀ groups of the studied thiols and disulfides.(A) Individual 1 H-NMR spectra of the 3 thiols and the 3 homodisulfides (B) 1 H-NMR spectrum of the resulting exquilibrated mixture, which contains 3 thiols, 3 homodisulfides, and the corresponding 3 heterodisulfides.(C) Simulated 1 H-NMR spectrum of the expected mixture fit to the experimental spectrum B. (D) Individual signals of each species composing the simulated spectrum C, allowing the quantification of each thiol, homo-and heterodisulfide.Thiols (RSH) and disulfides (RSSR') are shown as combination of "thiol moieties" (R, R').M: coenzyme M moiety; B: Coenzyme B moiety; E: mercaptoethanol moiety.
Figure 2. Example of quantification of all individual species present in a thiol-disulfide exchange reaction (species at the equilibrium from experiment 9, see Supplementary TableI).The range shown between 2.45 ppm and 3.05 ppm contains the signals of the À CH 2 À SÀ groups of the studied thiols and disulfides.(A) Individual 1 H-NMR spectra of the 3 thiols and the 3 homodisulfides (B) 1 H-NMR spectrum of the resulting exquilibrated mixture, which contains 3 thiols, 3 homodisulfides, and the corresponding 3 heterodisulfides.(C) Simulated 1 H-NMR spectrum of the expected mixture fit to the experimental spectrum B. (D) Individual signals of each species composing the simulated spectrum C, allowing the quantification of each thiol, homo-and heterodisulfide.Thiols (RSH) and disulfides (RSSR') are shown as combination of "thiol moieties" (R, R').M: coenzyme M moiety; B: Coenzyme B moiety; E: mercaptoethanol moiety.

Figure 3 .
Figure 3. Simulated and experimental time courses of CoB and CoM thiol-disulfide exchange reactions.B1, B2, B3, B4: Reaction started with the addition of CoBÀ SH to CoMÀ SÀ SÀ CoM M: Reaction started with the addition of CoMÀ SH to CoBÀ SÀ SÀ CoB.