Kinetic mechanism of acetyl-CoA synthase: Steady-state synthesis at variable CO/CO2 pressures

Article abstract of DOI:10.1021/ja004017t

Steady-state initial rates of acetyl-CoA synthesis (v/[E_tot]) catalyzed by acetyl-CoA synthase from Clostridium thermoaceticum (ACS) were determined at various partial pressures of CO and CO₂. When [CO] was varied from 0 to 100 μM in a balance of Ar, rates increased sharply from 0.3 to 100 min^-1. At [CO] > 100 μM, rates declined sharply and eventually stabilized at 10 min^-1 at 980 μM CO. Equivalent experiments carried out in CO₂ revealed similar inhibitory behavior and residual activity under saturating [CO] Plots of v/[E_tot] vs [CO₂] at different fixed inhibitory [CO] revealed that V_max/[E_tot] (k_cat) decreased with increasing [CO]. Plots of v/[E_tot] vs [CO₂] at different fixed noninhibitory [CO] showed that V_max/[E_tot] was insensitive to changes in [CO]. Of eleven candidate mechanisms, the simplest one that fit the data best had the following key features: (a) either CO or CO₂ (at a designated reductant level and pH) activate the enzyme (E′ + CO ? E, E′ + CO₂/2e^-/2H⁺ ? E); (b) CO and CO₂ are both substrates that compete for the same enzyme form (E + CO ? ECO, E + CO₂/2e^-/2H⁺ ? ECO, and ECO → E + P); (c) between 3 and 5 molecules of CO bind cooperatively to an enzyme form different from that to which CO₂ and substrate CO bind (nCO + ECO ? (CO)_nECO), and this inhibits catalysis; and (d) the residual activity arises from either the (CO)_nECO state or a heterogeneous form of the enzyme. Implications of these results, focusing on the roles of CO and CO₂ in catalysis, are discussed.

Full text of DOI:10.1021/ja004017t

J. Am. Chem. Soc. 2001, 123, 4697-4703
4697
Kinetic Mechanism of Acetyl-CoA Synthase: Steady-State Synthesis
at Variable CO/CO2 Pressures
Ernest L. Maynard,† Christopher Sewell,† and Paul A. Lindahl*,†,‡
Contribution from the Departments of Chemistry and of Biochemistry and Biophysics,
Texas A&M UniVersity, College Station, Texas 77843
ReceiVed NoVember 20, 2000
Abstract: Steady-state initial rates of acetyl-CoA synthesis (υ/[Etot]) catalyzed by acetyl-CoA synthase from
Clostridium thermoaceticum (ACS) were determined at various partial pressures of CO and CO2. When [CO]
was varied from 0 to 100 µM in a balance of Ar, rates increased sharply from 0.3 to 100 min-1. At [CO] >
100 µM, rates declined sharply and eventually stabilized at 10 min-1 at 980 µM CO. Equivalent experiments
carried out in CO2 revealed similar inhibitory behavior and residual activity under saturating [CO]. Plots of
υ/[Etot] vs [CO2] at different fixed inhibitory [CO] revealed that Vmax/[Etot] (kcat) decreased with increasing
[CO]. Plots of υ/[Etot] vs [CO2] at different fixed noninhibitory [CO] showed that Vmax/[Etot] was insensitive
to changes in [CO]. Of eleven candidate mechanisms, the simplest one that fit the data best had the following
key features: (a) either CO or CO2 (at a designated reductant level and pH) activate the enzyme (E′ + CO a
E, E′ + CO2/2e-/2H+ a E); (b) CO and CO2 are both substrates that compete for the same enzyme form (E
+ CO a ECO, E + CO2/2e-/2H+ a ECO, and ECO f E + P); (c) between 3 and 5 molecules of CO bind
cooperatively to an enzyme form different from that to which CO2 and substrate CO bind (nCO + ECO a
(CO)nECO), and this inhibits catalysis; and (d) the residual activity arises from either the (CO)nECO state or
a heterogeneous form of the enzyme. Implications of these results, focusing on the roles of CO and CO2 in
catalysis, are discussed.
Introduction
is the active site for acetyl-CoA synthesis6,7 while the C-cluster
is the site of CO/CO2 redox catalysis.8,9 The enzyme is
heterogeneous, in that only 30-40% of the A- and C-clusters
appear to be catalytically active.10-12
The catalytic mechanism of acetyl-CoA synthesis is thought
to involve the binding of CO to the A-cluster. An electron
generated from the oxidation of CO at the C-cluster reduces
Certain archaea and bacteria employ the Wood/Ljungdahl
pathway to grow chemoautotrophically on CO2 and H2.1
Enzymes of the pathway reduce CO2 to the methyl group of
methyltetrahydrofolate (CH3-THF).2 This methyl group is first
transferred to a corrinoid-iron-sulfur protein (CoFeSP) in a
reaction catalyzed by methyltransferase (MeTr). Acetyl-
Coenzyme A synthase (ACS; aka carbon monoxide dehydro-
genase or CODH) is a bifunctional enzyme3 that catalyzes both
the reversible reduction of CO2 (not bicarbonate)4,5 to CO and
the synthesis of acetyl-CoA from CO, CoA, and the CoFeSP-
bound methyl group.
the oxidized diamagnetic A-cluster (Aox) in association with the
1
binding of CO, yielding the S )
/
Ared-CO state.13 This
2
controversial state has been implicated as both an intermedi-
ate14,15 of acetyl-CoA synthesis and a state unable to proceed
to form products.16,17
Methyl group transfer to ACS is possible only when an
unidentified n ) 2 redox site on ACS, known as the D-site, is
reduced.16 Barondeau and Lindahl proposed that the D-site is a
disulfide/dithiol coordinated to or located near the Ni of the
A-cluster, and that it facilitates nucleophilic attack of the Ni on
the methyl group bound to CoFeSP. Migratory insertion of CO
into the resulting CH3-Ni2+ bond yields a Ni-acetyl group,
Acetogenic ACSs are R2â2 tetramers containing two types
of novel Ni-X-Fe4S4 clusters (the A- and C-clusters) and an
Fe4S4 electron-transfer cluster (the B-cluster).6-8 The A-cluster
* To whom correspondence should be addressed at the Department of
Chemistry.
† Department of Chemistry.
‡ Department of Biochemistry and Biophysics.
(1) Wood, H. G.; Ljungdahl, L. G. Variations in Autotrophic Life;
Academic Press: London, 1991.
(8) Hu, Z. G.; Spangler, N. J.; Anderson, M. E.; Xia, J. Q.; Ludden, P.
W.; Lindahl, P. A.; Mu¨nck, E. J. Am. Chem. Soc. 1996, 118, 830-845.
(9) Anderson, M. E.; Lindahl, P. A. Biochemistry 1994, 33, 8702-8711.
(10) Shin, W.; Anderson, M. E.; Lindahl, P. A. J. Am. Chem. Soc. 1993,
115, 5522-5526.
(11) Shin, W.; Lindahl, P. A. J. Am. Chem. Soc. 1992, 114, 9718-9719.
(12) Fraser, D. M.; Lindahl, P. A. Biochemistry 1999, 38, 15697-15705.
(13) Ragsdale, S. W.; Ljungdahl, L. G.; DerVartanian, D. V. Biochem.
Biophys. Res. Commun. 1982, 108, 658-663.
(14) Gorst, C. M.; Ragsdale, S. W. J. Biol. Chem. 1991, 266, 20687-
20693.
(15) Menon, S.; Ragsdale, S. W. Biochemistry 1996, 35, 12119-12125.
(16) Barondeau, D. P.; Lindahl, P. A. J. Am. Chem. Soc. 1997, 119,
3959-3970.
(17) Grahame, D. A.; Khangulov, S.; Demoll, E. Biochemistry 1996, 35,
593-600.
(2) Abbreviations: CH3-THF, methyl tetrahydrofolate; CoFeSP, corrinoid
iron-sulfur protein; MeTr, methyltransferase; ACS, acetyl-Coenzyme A
synthase from Clostridium thermoaceticum, also known as CODH; CoA,
coenzyme A; MV, methyl viologen; AS, ammonium sulfate; Dred, reduced
state of the D-site; Dox, oxidized state of the D-site; Ared-CO, one-electron
reduced CO-bound form of the A-cluster; Aox, the oxidized form of the
A-cluster; SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel elec-
trophoresis; DTT, dithiothreitol; RDS, rate-determining step.
(3) Ragsdale, S. W.; Kumar, M. Chem. ReV. 1996, 96, 2515-2539.
(4) Bott, M.; Thauer, R. K. Z. Naturforsch. 1989, 44, 392-396.
(5) Ensign, S. A. Biochemistry 1995, 34, 5372-5381.
(6) Xia, J.; Sinclair, J. F.; Baldwin, T. O.; Lindahl, P. A. Biochemistry
1996, 35, 1965-1971.
(7) Xia, J.; Hu, Z.; Popescu, C. V.; Lindahl, P. A.; Mu¨nck, E. J. Am.
Chem. Soc. 1997, 119, 8301-8312.
10.1021/ja004017t CCC: $20.00 © 2001 American Chemical Society
Published on Web 04/27/2001

4698 J. Am. Chem. Soc., Vol. 123, No. 20, 2001
Maynard et al.
monitored continuously (Teledyne model 310 analyzer). ACS was
purified from cell paste, using a modified procedure (10 mM DTT was
included in all buffers).33 ACS was 90-94% pure, as quantified by
imaging Coomassie Blue (Bio-Rad) stained SDS-PAGE gels with an
AlphaImager 2000 (Alpha Innotech Corp.) densitometer. ACS catalyzed
CO oxidation33 and acetyl-CoA synthesis (assayed as described below
with 1 atm CO2) with specific activities of 350 and 1.3 µmol min-1
mg-1, respectively. Protein concentrations were determined by the
Biuret method.34 ACS, CoFeSP, and MeTr have molecular masses of
154 700 Da /Râ,35 89 000 Da /Râ,36 and 57 280 Da/R2,37 respectively.
Buffer A (50 mM MES pH 6.3) was rendered CO2-free as follows.
MES (Sigma) was dissolved in distilled-deionized water to a final
concentration of 50 mM, pH 3.3, filtered, degassed using an anaerobic
Schlenk line, and brought into the box. The pH was adjusted to 6.30
by using anearobically prepared 50% (w/v) KOH. Buffer A was sparged
with O2-scrubbed (Oxisorb, MG Industries) Ar for 30 min prior to use.
ACS activity assays were used to detect residual CO2 dissolved in Buffer
A. Without added CO or CO2, ACS catalyzed the synthesis of acetyl-
CoA at an initial rate of 0.1 µM min-1, corresponding to 0.6 µM CO2
in solution.
subsequent nucleophilic attack of which by CoA yields acetyl-
CoA and the rereduced D-site.16,18
We recently examined the effect of CO2 on the catalytic
synthesis of acetyl-CoA, and discovered that it is a substrate
for this activity.19 This implies that CO2 is reduced to CO at
the C-cluster and that CO migrates to the A-cluster for use in
acetyl-CoA synthesis. We also found that CO is not released
into solution for this migration, but travels through a protein-
encapsulated tunnel from the C- to the A-cluster. Since these
clusters are located in separate subunits and show no magnetic
1
interactions when both are in S ) /2 paramagnetic states, the
tunnel may be >10 Å long. A subsequent study by Ragsdale
and co-workers provided further evidence for this tunnel.20
Molecular tunnels connect active sites in other multifunctional
enzymes.21-23
Since CO is known to react with the C-cluster and the
A-cluster, there must be at least 2 CO-binding sites on the
enzyme. However, additional sites have been proposed.9,24-28
Anderson and Lindahl suggested that CO or CO2 (in the
presence of a reductant) reactivate cyanide-inhibited enzyme
by binding to a “modulator” site.9 Seravalli et al. suggested that
two CO molecules bind to the C-cluster.25 Russell and Lindahl
interpreted their CO/CO2 potentiometric titrations in terms of a
redox-cooperativity in the presence of CO2.26 Ludden and co-
workers have recently reported that CODH from Rhodospirillum
rubrum is activated upon binding a CO molecule at the
C-cluster.24 CO has also been found to partially inhibit the ACS-
catalyzed exchange of free CoA with acetyl-CoA.29,30
These studies suggest that the effects of CO and CO2 with
ACS are complicated, possibly involving multiple roles. More-
over, the discovery that both CO2 and CO are substrates for
the synthesis of acetyl-CoA raises issues regarding the kinetic
mechanism. CO2 behaves as a classical Michaelis-Menten
substrate (Km ) 320 ( 50 µM; kcat/Km ) 0.53 ( 0.07 µM-1
min-1),19 while analogous kinetic parameters for CO have not
been reported. Also unknown are whether CO and CO2 are
competitive substrates and whether their catalytic properties are
additive. To address these issues, we measured steady-state
acetyl-CoA synthase activity at various concentrations of CO
and CO2, and fit kinetic models to the data. In this paper, these
experiments are described, a kinetic mechanism emphasizing
the roles of CO and CO2 in catalysis is proposed, and
implications are discussed.
Dithionite-reduced ACS was thawed, concentrated using a Centricon-
100 (Amicon), and chromatographed using a Sephadex G-25 column
(1 cm × 20 cm) equilibrated with Buffer A containing 1.0 mM DTT.
Dithionite-free enzyme was eluted at 0.5 mL/min. Samples were divided
into aliquots and simultaneously frozen in liquid N2.
CoFeSP Purification. Rust-brown colored fractions eluting prior
to ACS on DEAE Sephacel (Pharmacia) were concentrated by
ultrafiltration through a YM30 membrane (Amicon), made 10% in
ammonium sulfate (AS), and applied to a phenyl Sepharose (Pharmacia)
column (5 cm × 17 cm) equilibrated in Buffer B (50 mM Tris-Cl, pH
8.0, 2.0 mM dithionite, 10 mM DTT) plus 10% AS.16 The column was
washed with 500 mL of Buffer B plus 10% AS, and proteins were
eluted with a linear gradient containing 10-2.5% AS (750 mL of each).
Rust-brown fractions were combined, concentrated, diluted with 5
volumes of Buffer B, and loaded onto a DEAE Sephacel column (5
cm × 16 cm) equilibrated with Buffer B. The column was washed
with 800 mL of Buffer B containing 0.1 M NaCl. CoFeSP eluted with
a linear gradient containing 0.1-0.4 M NaCl (500 mL of each).
Combined CoFeSP fractions were removed from the box and frozen
in liquid N2. Active fractions were 92-95% pure according to SDS-
PAGE analysis. CoFeSP was assayed (as described below with 1 atm
of CO2) for its ability to assist in catalyzing the synthesis of acetyl-
CoA, by varying its concentration at fixed [ACS] (2 µM) and [MeTr]
(10 µM). Within the range tested (0 - 8 µM CoFeSP) the rate of acetyl-
CoA synthesis was linear, with a specific activity of 0.10 µmol min-1
mg-1
.
Dithionite-reduced CoFeSP was thawed, concentrated using a
Centricon-30, and chromatographed using Sephadex G-25 (1 cm × 20
cm) equilibrated in Buffer A. CoFeSP was collected into a 1-cm quartz
cuvette and scanned from 350-620 nm (Spectral Instruments model
SI 440). Residual dithionite was reacted with thionin (Aldrich) (2.5
mM, in Buffer A) until the absorbance at 604 nm, due to unreacted
thionin, increased.16 The sample was concentrated and excess thionin
was removed by G-25 chromatography, as above. Dithionite-free
thionin-oxidized CoFeSP was concentrated, divided into aliquots, and
frozen as above.
Experimental Procedures
ACS Purification. Clostridium thermoaceticum cells were grown
and harvested as described.31,32 Protein purification was performed in
a Vacuum/Atmospheres HE-453 glovebox containing <1 ppm O2, as
(18) Ragsdale, S. W.; Wood, H. G. J. Biol. Chem. 1985, 260, 3970-
3977.
(19) Maynard, E. L.; Lindahl, P. A. J. Am. Chem. Soc. 1999, 121, 9221-
9222.
(20) Seravalli, J.; Ragsdale, S. W. Biochemistry 2000, 39, 1274-1277.
(21) Thoden, J. B.; Holden, H. M.; Wesenberg, G.; Raushel, F. M.;
Rayment, I. Biochemistry 1997, 36, 6305-6316.
(22) Hyde, C. C.; Ahmed, S. A.; Padlan, E. A.; Miles, E. W.; Davies,
D. R. J. Biol. Chem. 1988, 263, 17857-17871.
MeTr Purification. MeTr assay solution contained the following
(final concentrations): Buffer C (50 mM Na-phosphates, pH 6.3, 1.0
(31) Lundie, L. L., Jr.; Drake, H. L. J. Bacteriol. 1984, 159, 700-703.
(32) Loke, H.-K.; Bennett, G. N.; Lindahl, P. A. Proc. Natl. Acad. Sci.
U.S.A. 2000, 97, 12530-12535.
(33) Shin, W.; Lindahl, P. A. Biochim. Biophys. Acta 1993, 1161, 317-
322.
(23) Krahn, J. M.; Kim, J. H.; Burns, M. R.; Parry, R. J.; Zalkin, H.;
Smith, J. L. Biochemistry 1997, 36, 11061-11068.
(24) Heo, J.; Staples, C. R.; Halbleib, C. M.; Ludden, P. W. Biochemistry
2000, 39, 7956-7963.
(25) Seravalli, J.; Kumar, M.; Lu, W.-P.; Ragsdale, S. W. Biochemistry
1997, 36, 11241-11251.
(34) Pelley, J. W.; Garner, C. W.; Little, G. H. Anal. Biochem. 1978,
86, 341-343.
(26) Russell, W. K.; Lindahl, P. A. Biochemistry 1998, 37, 10016-10026.
(27) Ensign, S. A.; Hyman, M. R.; Ludden, P. W. Biochemistry 1989,
28, 4973-4979.
(35) Morton, T. A.; Rundquist, A. R.; Ragsdale, S. W.; Shanmu-
gasundaram, T.; Wood, H. G.; Ljundahl, L. G. J. Biol. Chem. 1991, 266,
23824-23828.
(28) Hyman, M. R.; Ensign, S. A.; Arp, D. J.; Ludden, P. W. Biochemistry
1989, 28, 6821-6826.
(36) Lu, W. P.; Schiau, I.; Cunningham, J. R.; Ragsdale, S. W. J. Biol.
Chem. 1993, 268, 5605-5614.
(29) Lu, W.; Ragsdale, S. W. J. Biol. Chem. 1991, 266, 3554-3564.
(30) Pezacka, E.; Wood, H. G. J. Biol. Chem. 1986, 261, 1609-1615.
(37) Roberts, D. L.; Zhao, S. Y.; Doukov, T.; Ragsdale, S. W. J.
Bacteriol. 1994, 176, 6127-6130.

Kinetic Mechanism of Acetyl-CoA Synthas
J. Am. Chem. Soc., Vol. 123, No. 20, 2001 4699
mM DTT), 50 µM Râ CoFeSP, 1.0 mM CH3-THF (Sigma, disodium
salt), and 500 µM Ti(III) citrate prepared38 from Ti(III) chloride
(Aldrich) at 30 ( 2 °C. To 0.50 mL of assay solution in a 0.5 cm
quartz cuvette was added 50- 100 µL of solution containing MeTr.
[Co1+FeSP] (calculated using ∆ꢀ390 nm ) 7.9 mM-1 cm-1, experimen-
tally determined) was monitored vs time.
and [CO2]). This assumption should be valid, as the initial [MV1+]/
[MV2+] ratio of the stock solution (2.1 ( 0.2) was nearly the same for
each experiment, and all steps performed after introducing this reductant
were strictly anaerobic. Because [MV] and [H+] were constant, the
“CO2” included in our mechanisms (and conclusions) should be viewed
as an undifferentiated species CO2/2e-/2H+ rather than a CO2 molecule.
The initial CH3-THF concentration was 200 times its Km of 10 µM,44
and the amount consumed during reaction (∼0.3 mM) was insignificant.
The high concentrations of CH3-THF and MeTr ensured that methyl
group transfer to CoFeSP did not affect the rate of acetyl-CoA synthesis
within the conditions employed. These high concentrations also ensured
that the concentration of substrate CH3-Co3+FeSP would remain
constant and high, and that the concentration of product Co1+FeSP
would remain constant and low, relative to [ACS]. The large volume
of the reaction vessels (∼270 mL) ensured that [CO] and [CO2] did
not change significantly during reaction. The initial [CoA] was
substantially greater than its Kd of 10 µM45 and its Km of 50 µM for
the CO/Acetyl-CoA exchange reaction.46 Thus, under the conditions
employed, only changes in [CO] and [CO2] affected the initial rate of
reaction.
Active fractions, which eluted after ACS on the DEAE Sephacel
column (from the ACS prep), were concentrated by ultrafiltration using
a YM30 membrane, diluted with 2 volumes of Buffer C, and applied
to a DEAE Biogel-A (Bio-Rad) column (5 cm × 12.7 cm) equilibrated
in Buffer C. The column was washed with 700 mL of Buffer C
containing 0.15 M NaCl. Proteins were eluted with a linear gradient
containing 0.15-0.4 M NaCl (600 mL of each). Active fractions were
concentrated, diluted with 5 volumes of Buffer D (50 mM Tris-Cl, pH
7.6, 1.0 mM DTT), made 6.0% in AS, and applied to a phenyl
Sepharose column (5 cm × 14 cm) equilibrated in Buffer D containing
6.0% AS. The column was washed with 1 L of Buffer D containing
6.0% AS. Proteins were eluted with a linear gradient containing 6.0-
0% AS in Buffer D (600 mL of each). Active fractions were combined,
concentrated, and immediately frozen in liquid N2. MeTr was >95%
pure according to SDS-PAGE analysis and had a specific activity of
5.0 µmol min-1 mg-1. Purified MeTr was thawed, concentrated using
a Centricon-10 (Amicon), diluted with 25 volumes of Buffer A
containing 1.0 mM DTT, and reconcentrated. After repeating this
procedure 5×, aliquots were frozen as above.
Modeling and Simulations. Eleven kinetic mechanisms were
analyzed, designated U1AT, U2AT, U3AT, U4AT, U5AT, U6AT,
C4AT, M4AT, R4AT, U4AO, and U4NT.47 These included a “residual”
activity at high inhibitor concentrations that arose either from a partially
active inhibited enzyme form (U4AO) or from a distinct enzyme form
not inhibited by CO (all other models). Each mechanism involved N
enzyme forms. Differential equations describing the time-dependent
concentrations of N - 1 of these forms were set equal to 0 in accordance
with the steady-state approximation. Resulting equations and the
Acetyl-CoA Synthase Assay. CH3-THF and Coenzyme A (Sigma,
sodium salt) were dissolved in Buffer A to give 63.5 ( 0.2 and 34.5
( 0.3 mM stock solutions, respectively. Concentrations were determined
using ꢀ290 ) 30.8 mM-1 cm-1 for CH3-THF39,40 and ꢀ260 ) 16.8 mM-1
cm-1 for CoA.41 DTT was added (1.0 mM final) to the CoA solution
to prevent oxidation. Each solution was divided into aliquots and frozen
as above.
Various amounts of CO (MG Industries, research grade), CO2 (MG
Industries, anaerobic grade), and Ar were mixed with a flowmeter (MG
Industries, series 7941-AS2 4-tube) and passed into a reaction vessel.19
The flowmeter was calibrated by measuring in triplicate the rate of
water displaced from a volumetric flask. Henry’s law constants for CO
and CO2 at 30 °C are 0.98 mM/atm41 and 31 mM/atm,42 respectively.
Methyl viologen (Sigma; MV2+/1+; Eo′ ) -0.44 V vs NHE43) was
enzymatically reduced,19 standardized by titration against K3Fe(CN)6,
and used immediately (average of 12 measurements yielded 68 ( 2%
reduction). To a 5.0 mL conical vial were added, in the following order,
Buffer A, CH3-THF (2.0 mM), CoFeSP (30 µM), MeTr (10 µM), and
MV1+ (1.0 mM), yielding 0.50 mL total volume (final concentrations).
The assay solution was mixed and then transferred into a reaction vessel.
The vessel was sealed, removed from the glovebox, flushed with 15-
20 volumes of the desired mixture of gases, and returned to the box.
For experiments involving varied [CO2], CO2 was injected by syringe
(Hamilton Gastight). The reaction mixture was incubated 15 min at 30
( 2 °C in dim light. Acetyl-CoA synthesis was initiated by syringe
injection of a CoA/ACS solution affording 1.0 mM CoA, 0.3 µM Râ
ACS, and 50 µM DTT (final concentrations). Aliquots (80 µL) were
removed by syringe at various times and analyzed for acetyl-CoA by
reversed phase HPLC.19 Initial rates were determined by linear least-
squares regression analysis of [acetyl-CoA] vs time plots. Ninety-seven
data points (rates at various [CO] and [CO2]) were analyzed.
conservation relationship [Etot] ) ∑N [Ei] were solved using Maple
i)1
version 6.0 (Waterloo Maple, Inc.), for [E] and [Eres] (the enzyme form
that gives rise to the residual activity) as a function of [CO], [CO2],
and [Etot]. These expressions were substituted into the rate expression
d[P]/dt ) υ ) kp[E] + kres[Eres], where kp and kres are turnover numbers
for the major and residual enzyme forms. The expression for maximal
rate Vmax ) kcat[Etot] was determined by setting each substrate
concentration to ∞ and all inhibitor concentrations to 0. The expression
for the Km of a given substrate was obtained by setting the other
substrate and inhibitor concentrations to 0 and solving for the substrate
1
concentration when υ ) /2Vmax. Expressions describing kcat/Km and
Km in terms of microscopic rate constants were substituted into the
equation υ ) kp[E] + kres[Eres] as a means of minimizing the number
of unknowns.
Resulting steady-state velocity equations were fit to 14 data sets
(plots of (υ/[Etot])dat vs either increasing [CO]dat at fixed [CO2]dat or
increasing [CO2]dat at fixed [CO]dat) using a computer program written
in C. The Adaptive Simulated Annealing (ASA) algorithm48 was used
to search for best-fit values of the parameters for each equation. A
cost function returned the sum of the squares of the residuals between
the data points and the graph of the equation evaluated at the current
parameter values. ASA was used to find parameter values that
minimized this difference. Relative errors for each model are reported
in Table 1. ASA was also used to determine the uncertainty in each
parameter by separately finding the value that would make the cost
1.5 times its minimal value, fixing all other parameters at best-fit values.
Simulations of each data set (i.e. υsim/[Etot] vs either [CO]sim at fixed
[CO2]sim or [CO2]sim at fixed [CO]sim) consisted of 1000 simulation points
spanning the same range as the data.
Conditions Employed in Assays. A reductant was required when
CO2 was used as the substrate. Although the resulting solution potential
was sufficient to yield only ∼20% of the theoretical Vmax/[Etot],19 MV1+
was chosen because of its stability at the experimental pH. Our analysis
assumes that the steady-state ratios of [MV1+]/[MV2+] and [Dred]/Dox]
were the same under all conditions employed (i.e. independent of [CO]
(44) Zhao, S. Y.; Roberts, D. L.; Ragsdale, S. W. Biochemistry 1995,
34, 15075-15083.
(45) Wilson, B. E.; Lindahl, P. A. J. Biol. Inorg. Chem. 1999, 4, 742-
748.
(46) Ramer, S. E.; Raybuck, S. A.; Orme-Johnson, W. H.; Walsh, C. T.
Biochemistry 1989, 28, 4675-4680.
(38) Seefeldt, L. C.; Ensign, S. A. Anal. Biochem. 1994, 221, 379-386.
(39) Blair, J. A.; Saunders: K. J. Anal. Biochem. 1970, 34, 376-381.
(40) Gupta, V. S.; Huennekens, F. M. Arch. Biochem. Biophys. 1967,
120, 712-718.
(47) Mechanisms were distinguished by a four-alphanumeric code. The
first alphanumeric (U, C, M, or R) indicates Uncompetitive, Competitive,
Mixed, or Reductive inhibition. The second (1-6) indicates the number of
CO’s presumed to inhibit activity. The third alphanumeric (A or N) indicates
whether CO/CO2 Activation was assumed or Not assumed. The fourth
alphanumeric (O or T) indicates whether one or two terms are present in
the rate equation.
(41) Budavari, S. The Merck Index, 11 ed.; Merck & CO., Inc.: Rahway,
NJ, 1989.
(42) Butler, J. N. Carbon Dioxide Equilibria and Their Applications;
Addison-Wesley: Reading, MA, 1982.
(43) Michaelis, L.; Hill, E. S. J. Am. Chem. Soc. 1933, 55, 1481.
(48) Ingber, L. Math. Comput. Modeling 1989, 12, 967-973.

4700 J. Am. Chem. Soc., Vol. 123, No. 20, 2001
Maynard et al.
Table 1
mechanism
description
unknowns
rel error
U1AT
U2AT
U3AT
U4AT
U5AT
U6AT
C4AT
M4AT
R4AT
U4NT
U4AO
Uncompetitive; 1 CO’s bind ECO; Activation; Two terms
Uncompetitive; 2 CO’s bind ECO; Activation; Two terms
Uncompetitive; 3 CO’s bind ECO; Activation; Two terms
Uncompetitive; 4 CO’s bind ECO; Activation; Two terms
Uncompetitive; 5 CO’s bind ECO; Activation; Two terms
Uncompetitive; 6 CO’s bind ECO; Activation; Two terms
Competitive; 4 CO’s bind ECO; Activation; Two terms
Mixed; 4 CO’s bind ECO; Activation; Two terms
10
11
12
13
14
15
13
17
13
10
12
5.4
1.8
1.1
1.0
0.98
0.96
4.6
1.0
5.9
2.2
1.1
1 CO Reduces and 3 CO’s bind ECO; Activation; Two terms
Uncompetitive; 4 CO’s bind ECO; No activation; Two terms
Uncompetitive; 4 CO’s bind ECO; Activation; One term
Figure 2. Initial rate of acetyl-CoA synthesis at increasing [CO] under
3.8 (b) and 1.0 mM (9) [CO2]. Other conditions were as in Figure 1.
Inset: Double reciprocal plot and comparison with data of Figure 1
(1).
Figure 1. Initial rate of acetyl-CoA synthesis at increasing [CO] under
an Ar atmosphere. Data (b) were obtained as described in the
Experimental Procedures. The solid line is the best-fit simulation gen-
erated by using eq 1 (model U4AT, Table 1) and the best-fit parameters
(see text). Inset: Double reciprocal plot of data (b), best fit (s), and
fit without activation (mechanism U4NT) by CO/CO2 (- - -).
inactive or less active than the CO-bound or CO-reduced form.
Thus, CO is both an activator and a substrate of ACS-catalyzed
acetyl-CoA synthesis.
Results
CO Inhibition of ACS and the Residual Activity. As [CO]
increased above 100 µM, rates declined sharply. By 300 µM
CO, the rate was 20 min-1, 20% of maximal. This reveals that,
in addition to being a substrate and an activator, CO inhibits
ACS from catalyzing acetyl-CoA synthesis. The rate of catalysis
continued to decline as [CO] increased, eventually stabilizing
at a residual rate of 10 min-1 under an atmosphere of CO (980
µM). This rate is similar to those reported previously under
Assay Conditions. The catalytic synthesis of acetyl-CoA by
ACS is exceedingly complex, involving three proteins (ACS,
CoFeSP, and MeTr), seven components (CO, CO2, CH3-THF,
CoA, MV1+, MV2+, and H+), and the strict exclusion of O2. In
the past, this complexity has made detailed kinetic studies
difficult. We have succeeded in performing such experiments
by controlling and fixing numerous variables and focusing on
the effects of CO and CO2. As a result, the kinetic parameters
reported here are apparent rather than true values, applicable
only under the conditions employed. Refer to Experimental
Procedures for details.
CO Activation of ACS. To determine the steady-state kinetic
parameters for CO, initial steady-state rates of acetyl-CoA
synthesis, normalized to the total enzyme concentration (υ/[Etot]),
were measured as the concentration of CO was varied in a
balance of Ar (Figure 1, solid circles). As [CO] increased from
0 to 100 µM, υ/[Etot] increased from 0.3 to 100 min-1. When
plotted in double-reciprocal form (Figure 1, inset), the data in
this range yielded a straight line, suggesting a hyperbolic
dependence of rate on [CO], as expected for a Michaelis-
Menten substrate. However, the slope of this line is far steeper
than would be observed for a standard substrate. As we show
below, the only models able to simulate this slope (and the
observed maximal activity) assume that the binding of CO to
ACS, or the reduction of some site on ACS caused by the
binding and subsequent oxidation of CO to CO2, activates ACS
for catalysis. The unbound or oxidized form of ACS is either
saturating CO conditions,49 including 7,50 15,51 and 20 min-1 52
.
Equivalent data obtained with 3.8 mM CO2 (Figure 2, solid
circles) showed a maximal rate of nearly 200 min-1 at 10 µM
[CO]. At higher [CO], rates again declined and eventually
stabilized at 10 min-1. These data demonstrate that CO also
inhibits catalysis in the presence of CO2. Similar data were
obtained with 1.0 mM CO2 (Figure 2, solid squares), except
that the maximal rate was somewhat less (150 min-1). Given
the previously determined Km value for CO2 (320 µM),19 this
difference in maximal rates probably reflects incomplete satura-
tion of ACS. For both CO2 concentrations used, residual rates
were similar to that obtained under Ar. A double-reciprocal plot
(49) The latter two rates were obtained by dividing the published rates
(obtained at 55 °C) by 8 to account for the temperature difference. Rates
were not corrected for the lower solubility of CO in solution at 55 °C. This
solubility difference may account for the ∼2-fold faster rates of catalysis
obtained.
(50) Lu, W.; Harder, S. R.; Ragsdale, S. W. J. Biol. Chem. 1990, 265,
3124-3133.
(51) Roberts, J. R.; Lu, W.-P.; Ragsdale, S. W. J. Bacteriol. 1992, 174,
4667-4676.
(52) Menon, S.; Ragsdale, S. W. Biochemistry 1998, 37, 5689-5698.

Kinetic Mechanism of Acetyl-CoA Synthas
J. Am. Chem. Soc., Vol. 123, No. 20, 2001 4701
Figure 3. Initial rate of acetyl-CoA synthesis at increasing [CO2] and
66 (b), 114 (2), 163 (9), and 282 µM ([) [CO].
Figure 5. (A) Kinetic mechanism of the “major” activity from U4AT.
(B) Mechanism U4AO. See text for details.
increased, and the corresponding double-reciprocal plots (Figure
4, inset) coalesced toward a single [Etot]/υ value. This type of
pattern is consistent with a competitive mode of interaction and
indicates that CO and CO2 compete for the same enzyme form
during the catalytic synthesis of acetyl-CoA. Plots of [Etot]/υ
vs 1/[CO2] at nonzero [CO] revealed nonlinear behavior (Figure
4, inset). This can be attributed to CO’s ability to serve as a
substrate in catalysis.
Data Analysis. Eleven kinetic mechanisms were evaluated
for their ability to simulate the data (as determined by the
relative errors in Table 1), and to do so with the fewest unknown
parameters. Using these criteria, the best mechanism, called
U4AT47 and illustrated in Figure 5A, is described by the steady-
state rate equation
Figure 4. Initial rate of acetyl-CoA synthesis at increasing [CO2] and
0 (b), 10 (2), 12 (3), 22 (9), 25 (O), 44 ([), and 46 µM (4) [CO].
Inset: Double reciprocal plot.
of the inhibitory region (Figure 2, inset) indicates a similar [CO]-
dependence for titrations performed in CO2 and Ar.
CO Functional Regions. CO is an activator, as well as a
substrate for catalysis. At high concentrations, CO partially
inhibits catalysis, as evidenced by the residual activity. These
functions operate throughout the [CO] range employed in these
experiments, but certain functions dominate within a given
concentration “region”. Despite an overlap of functions, 0-20
µM CO will be called the ActiVation region, 20-100 µM will
be called the Substrate region, 100-300 µM will be called the
Inhibitory region, and 300-980 µM will be called the Residual
region.
Relationship of CO and CO2 in the Inhibitory Region.
We wanted to determine whether CO inhibited catalysis by a
competitive or uncompetitive mechanism, or by a mixture of
the two (mixed inhibition). These mechanisms were distin-
guished graphically, by plotting rates as a function of [CO2] at
different fixed [CO] in the Inhibitory region. The resulting plots
(Figure 3) did not coalesce toward a single υ/[Etot] value (i.e.
kcat) as [CO2] increased (which would have indicated competitive
inhibition). Rather, υ/[Etot] decreased with increasing [CO],
consistent with uncompetitive inhibition.
kcat
kcat
[CO] +
[CO2]
( )
( )
Km
Km
kres[CO]
CO
CO2
υ
[Etot]
)
+
(1)
[CO2]
Km,res + [CO]
[CO]
T
+ T
+ A
Km,CO
K
m,CO2
where
and
[CO]j
n
T ) 1 +
(2)
(3)
∑
j
j)1
K
(
)
∏
ik
k)1
Probing the Substrate/Activation Region. The rate of
catalysis was measured as a function of [CO2] at different [CO]
in the Substrate/ActiVation region to determine whether substrate
CO and CO2 competed for the same enzyme form. The resulting
direct plots (Figure 4) approached the same υ/[Etot] as [CO2]
Kact(ka1 + ka3)
A ) 1 +
ka1[CO] + ka3[CO2]

4702 J. Am. Chem. Soc., Vol. 123, No. 20, 2001
Maynard et al.
The first term of U4AT describes the “major” activity that
dominates in the ActiVation, Substrate, and Inhibitor regions.
The second term represents the residual activity evident in the
Residual region. In this mechanism, the residual activity arises
from a second enzyme form which utilizes CO as a classical
Michaelis-Menten substrate. At infinite [CO], only the first
term approaches zero. According to U4AT, enzyme in the
inactivated form E′ reacts with either CO or CO2, yielding
activated form E. In the next step, CO and CO2 compete for E,
weak (Ki values >20 M CO) that this binding step could be
ignored, and removing this step did not affect the simulation.
All other mechanisms examined either yielded poorer fits
relative to U4AT or else contained substantially more unknown
parameters. U1AT, U2AT, U3AT, U5AT, and U6AT are
identical to U4AT except for involving 1, 2, 3, 5, or 6 inhibitory
CO molecules, respectively. Simulations using U1AT and U2AT
did not fit the data satisfactorily. Fits using U3AT, U5AT, and
U6AT were acceptable. U6AT was excluded because the
marginal improvement in fit relative to U5AT could not be
justified by the additional complexity. C4AT was identical to
U4AT except that CO inhibited catalysis by binding to the same
enzyme form as that to which substrates CO and CO2 bind (form
E). The relative error of C4AT excluded it. R4AT was identical
to U4AT except that the first inhibition step was assumed to be
a reduction producing CO2 rather than a binding event (the other
three were binding steps). However, simulations using R4AT
yielded a relative error high enough for it to be excluded. U4NT
lacked the activation step of U4AT, and failed to fit the data at
low [CO] (Figure 1 inset, dashed line), and was excluded.
yielding CO bound form ECO. ECO reacts with CH3-Co3+
-
FeSP and CoA (represented as a single step) yielding product
(acetyl-CoA) and E. Four CO molecules bind sequentially to
ECO, yielding unproductive (CO)nECO (n ) 1-4) forms of
the enzyme. Thus, U4AT assumes that 4 CO molecules inhibit
the synthesis of acetyl-CoA uncompetitively with respect to
“substrate” CO and CO2. The solid lines in Figures 1-4
represent the best-fit simulation obtained using U4AT. Best-fit
kinetic parameters for CO and CO2, obtained at 30 °C, were as
follows: kcat,CO ) 900 ( 300 min-1, Km,CO ) 300 ( 100 µM,
(kcat/Km)CO ) 3.2 ( 0.4 µM-1 min-1, kcat,CO ) 200 ( 30 min-1
,
,
2
Km,CO ) 380 ( 40 µM, (kcat/Km)CO ) 0.52 ( 0.04 µM-1 min-1
2
2
Kact ) 6 ( 3 µM, ka1 ) 10 ( 8 µM-1 min-1, ka3 ) 6000 (
3000 µM-1 min-1, kres ) 10 ( 5 min-1, and Km,res ) 200 (
100 µM. The best-fit inhibition constants were Ki1 ) 900 (
300 µM, Ki2 ) 50 ( 10 µM, Ki3 ) 40 ( 10 µM, Ki4 ) 50 (
30 µM.53
Discussion
In this study, steady-state rates of acetyl-CoA synthesis were
measured at various [CO] and [CO2]. Results were analyzed
by constructing various candidate kinetic models, deriving the
corresponding steady-state rate equations, and then attempting
to simulate the data using these equations. Comparing which
model could or could not simulate the data provided insight
into the actual mechanism used by ACS, especially with regard
to the roles of CO and CO2.
U4AT fits the data with high fidelity. The lower values of
the last three inhibition constants (Ki2, Ki3, and Ki4) relative to
Ki1 indicate positive-cooperative binding and inhibition of the
enzyme by CO. The (kcat/Km) values reflect the efficiency by
which CO or CO2 form ECO. The best-fit (kcat/Km)CO is 6 times
greater than that of CO2, suggesting that CO is a better substrate
than CO2. However, given that these values are apparent (see
Experimental Procedures), this difference may simply reflect
the sub-optimal reducing conditions used. Using CO2 as a
substrate, we have studied the dependence of solution redox
potential on initial velocity and found that under maximally
Our results and analysis indicate that ACS is activated for
catalysis by binding CO and possibly by an associated reduction
of a redox center in the enzyme. Candidate mechanisms that
did not include activation failed to fit the data, especially in
the region between 0 and 20 µM CO. As evident from the
double-reciprocal plot in the inset of Figure 1, the activity
increased more sharply than it would have in the absence of
activation. Since ACS is active in the presence of CO2 (and
reductant) and in the absence of CO, we presume that CO2 and
reductant also activate the enzyme. This suggests that activation
involves the binding of these molecules in conjunction with a
redox process. Relative to inactivated enzyme form E′, activated
enzyme E may have either CO or CO2 bound, and/or it may be
two-electrons more reduced. Activation probably does not
simply involve the binding of CO or CO2, as this would not
result in the same activated state. The site of activation may be
the C-cluster as it is the site of CO/CO2 redox catalysis. Another
possibility that is compatible with an earlier study9 is that CO
or CO2/2e-/2H+ activate the enzyme by binding to the modula-
tor site.
Our results and analysis also indicate that, at higher [CO],
CO and the undifferentiated substrate composed of CO2,
reductant, and protons compete for the same activated enzyme
form (E). While our data do not specify the CO/CO2 binding
site, it is undoubtedly the C-cluster, the active site where CO2
is reduced to CO, as this reduction must occur before CO2 could
be used as a substrate for acetyl-CoA synthesis. Our previous
study19 suggests that CO molecules obtained by reducing CO2
migrate to the A-cluster via a molecular tunnel. The competition
observed here between CO2 and CO molecules not derived from
CO2 suggests that such CO molecules also migrate to the
A-cluster via the C-cluster and tunnel.
reducing conditions, kcat,CO was enhanced by a factor of 4.6.19
2
Everything else being equal, correcting the best-fit kcat for this
factor would increase (kcat/Km)CO to a value (namely 2.4 µM-1
2
min-1) approaching that for CO at the same pH. Identical kcat/
Km values would indicate that at sufficiently negative redox
potentials, and at concentrations well below their Km values,
CO and CO2 are utilized at equal rates for the ACS-catalyzed
formation of acetyl-CoA. That kcat,CO obtained at these potentials
2
(920 min-1) is within error of that obtained for CO (900 min-1
)
indicates that these alternate pathways to product may share
the same RDS.
While U4AT provides the best fit with the lowest level of
complexity, M4AT and U4AO cannot be excluded (refer to
Table 1 for relative errors and number of unknowns for each
mechanism). M4AT is identical to U4AT except that inhibitory
CO molecules are assumed to bind both forms ECO and E.
Simulations using M4AT fit the data as well as U4AT; however,
the inhibitory binding of CO to E was so weak (4CO + E a
(CO)4E; the Ki values were >400 mM) that this process did
not contribute noticeably to the inhibition. U4AO, shown in
Figure 5B, differs from U4AT in that residual activity is
assumed to arise from the partially inhibited enzyme form
(CO)4ECO. The best-fit simulation using U4AO fit the data as
well as U4AT, but the competitive binding of CO was made so
(53) To take the heterogeneity of ACS into account, the apparent kinetic
parameters reported here should be divided by 0.3-0.4. Thus, the kcat values
reported here, while higher than any others reported to date, may
underestimate the true values by a factor of 2.5-3.3.
This study demonstrates that at even higher [CO], CO inhibits
acetyl-CoA synthesis by binding to a different form of the

Kinetic Mechanism of Acetyl-CoA Synthas
J. Am. Chem. Soc., Vol. 123, No. 20, 2001 4703
enzyme than that to which substrates bind. The inhibition is
unusually sensitive to small changes in [CO], and only simula-
tions that assumed positive cooperative binding of >2 CO’s
were able to mimic this sensitivity. In a potentiometric study
by Russell and Lindahl26 ACS was found to exhibit positive
redox cooperativity, further evidence for the presence coopera-
tive interactions between CO and the enzyme. That the inhibition
is uncompetitive became apparent when we tried to fit models
assuming mixed or competitive inhibition to the data. Moreover,
if CO inhibition were competitive, the extent of inhibition would
be depressed in the presence of CO2. In fact, inhibition is not
affected by CO2 (Figure 2). The site to which these CO
molecules bind is uncertain, but the A-cluster is a likely
possibility, and the Ared-CO state of this cluster may be unable
to proceed to products.3,14,15,54 Other scenarios in which the
inhibitor binds to a distinct enzyme species that arises after
product release are also possible.55
Our results and analysis reveal that CO-inhibition is partial,
and that a residual activity remains at the highest CO concentra-
tions employed. This is the only activity that was evident from
previous studies, which were performed at concentrations (980
µM, 1 atm) higher than those at which the major activity is
evident.15,50-52 We have considered two ways in which the major
and residual activities may be related. The residual activity may
arise from a heterogeneous form of the enzyme, and result from
a catalytic mechanism distinct from the major activity. Accord-
ing to this view, CO would be a Michaelis-Menten substrate
but not an inhibitor or an activator of the residual activity.
Alternatively, the residual activity may arise from the CO-
Figure 6. Proposed mechanism by which ACS discriminates between
substrate CO/CO2 and inhibitor CO in terms of different pathways to
the A-cluster. See text for details.
inhibited form of the enzyme associated with the major activity
mechanism of catalysis (form (CO)nECO), which may cor-
respond to Ared-CO. We are unable to distinguish these
possibilities.
Finally, why is CO2 a substrate but not an inhibitor of the
major activity while CO is both a substrate and inhibitor? One
possibility is that at moderate [CO], CO and CO2 bind at the
C-cluster, and that the resulting CO molecules travel through
the tunnel to the A-cluster where they serve as substrates in
acetyl-CoA synthesis. CO likely inhibits the enzyme by binding
the A-cluster, for example, by converting the diamagnetic
oxidized state Aox state to the Ared-CO state. Do these inhibitory
CO molecules access the A-cluster directly from solution or
from the C-cluster through the tunnel? The latter possibility
appears unlikely, because once a CO molecule dissociates from
the C-cluster and migrates toward the A-cluster, ACS would
be unable to differentiate between CO destined to serve as a
substrate and CO destined to inhibit catalysis. We propose
(Figure 6) that CO molecules inhibiting catalysis bind to the
A-cluster directly from solution, and do not travel to the
A-cluster via the C-cluster and tunnel. In contrast, CO molecules
that serve as substrate travel through the tunnel. It is fascinating
to consider mechanisms that might enable ACS to distinguish
between CO molecules arriving from the tunnel and those
coming directly from solution. While molecular tunnels have
been found in other multifunctional enzymes, this study may
be the first example of an enzyme that utilizes its tunnel to
discriminate between two identical molecules with diametrically
opposed effects.
(54) Not all previously published results may appear compatible with
this proposal. Analysis of the hyperfine splitting of the NiFeC EPR signal
(from the Ared-CO state) as well as the corresponding ENDOR spectra of
13CO-reacted ACS indicate a single CO bound at the A-cluster.56 A single
CO is also suggested by the IR spectra of enzyme in this state.57 Strictly
viewed, these results are compatible with this proposal, since our results
do not require that all inhibitory CO’s bind to the same site, only that they
bind cooperatively. Another potential discrepancy involves the strength of
CO inhibition. In an EPR/redox titration study, Russell and Lindahl
estimated the Kd for CO binding to the A-cluster in the Ared-CO state to be
∼3 and 0.3 µM in the absence and presence of CO2, respectively,26 while
the values obtained here are substantially weaker. This apparent discrepancy
may arise from differences in the way the two experiments were performed.
The experiments by Russell and Lindahl were performed in the absence of
the other substrates required for acetyl-CoA activity, while they were present
in this study. If one of those substrates (e.g. CH3-Co3+FeSP) competed
with the inhibitory CO’s for binding to the A-cluster, the apparent Kd values
measured here could underestimate true values. Further studies are required
to settle this issue.
Acknowledgment. The National Institutes of Health
(GM46441) and the Robert A. Welch Foundation (A1170)
sponsored this study.
Supporting Information Available: Derivation of steady-
state rate equations for all mechanisms (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
(55) Cennamo, C. J. Theor. Biol. 1968, 21, 260-277.
(56) Fan, C.; Gorst, C. M.; Ragsdale, S. W.; Hoffman, B. M. Biochemistry
1991, 30, 431-435.
(57) Kumar, M.; Ragsdale, S. W. J. Am. Chem. Soc. 1992, 114, 8713-
8715.
JA004017T