Determination of pre-steady-state rate constants on the escherichia coli pyruvate dehydrogenase complex reveals that loop movement controls the rate-limiting step

Article abstract of DOI:10.1021/ja3062375

Spectroscopic identification and characterization of covalent and noncovalent intermediates on large enzyme complexes is an exciting and challenging area of modern enzymology. The Escherichia coli pyruvate dehydrogenase multienzyme complex (PDHc), consisting of multiple copies of enzymic components and coenzymes, performs the oxidative decarboxylation of pyruvate to acetyl-CoA and is central to carbon metabolism linking glycolysis to the Krebs cycle. On the basis of earlier studies, we hypothesized that the dynamic regions of the E1p component, which undergo a disorder-order transition upon substrate binding to thiamin diphosphate (ThDP), play a critical role in modulation of the catalytic cycle of PDHc. To test our hypothesis, we kinetically characterized ThDP-bound covalent intermediates on the E1p component, and the lipoamide-bound covalent intermediate on the E2p component in PDHc and in its variants with disrupted active-site loops. Our results suggest that formation of the first covalent predecarboxylation intermediate, C2α-lactylthiamin diphosphate (LThDP), is rate limiting for the series of steps culminating in acetyl-CoA formation. Substitutions in the active center loops produced variants with up to 900-fold lower rates of formation of the LThDP, demonstrating that these perturbations directly affected covalent catalysis. This rate was rescued by up to 5-fold upon assembly to PDHc of the E401K variant. The E1p loop dynamics control covalent catalysis with ThDP and are modulated by PDHc assembly, presumably by selection of catalytically competent loop conformations. This mechanism could be a general feature of 2-oxoacid dehydrogenase complexes because such interfacial dynamic regions are highly conserved.

Full text of DOI:10.1021/ja3062375

Article
pubs.acs.org/JACS
Determination of Pre-Steady-State Rate Constants on the Escherichia
coli Pyruvate Dehydrogenase Complex Reveals That Loop Movement
Controls the Rate-Limiting Step
Anand Balakrishnan, Natalia S. Nemeria, Sumit Chakraborty, Lazaros Kakalis, and Frank Jordan*
Department of Chemistry, Rutgers the State University, Newark, New Jersey 07102, United States
S
* Supporting Information
ABSTRACT: Spectroscopic identification and characteriza-
tion of covalent and noncovalent intermediates on large
enzyme complexes is an exciting and challenging area of
modern enzymology. The Escherichia coli pyruvate dehydro-
genase multienzyme complex (PDHc), consisting of multiple
copies of enzymic components and coenzymes, performs the
oxidative decarboxylation of pyruvate to acetyl-CoA and is
central to carbon metabolism linking glycolysis to the Krebs
cycle. On the basis of earlier studies, we hypothesized that the
dynamic regions of the E1p component, which undergo a disorder−order transition upon substrate binding to thiamin
diphosphate (ThDP), play a critical role in modulation of the catalytic cycle of PDHc. To test our hypothesis, we kinetically
characterized ThDP-bound covalent intermediates on the E1p component, and the lipoamide-bound covalent intermediate on
the E2p component in PDHc and in its variants with disrupted active-site loops. Our results suggest that formation of the first
covalent predecarboxylation intermediate, C2α-lactylthiamin diphosphate (LThDP), is rate limiting for the series of steps
culminating in acetyl-CoA formation. Substitutions in the active center loops produced variants with up to 900-fold lower rates of
formation of the LThDP, demonstrating that these perturbations directly affected covalent catalysis. This rate was rescued by up
to 5-fold upon assembly to PDHc of the E401K variant. The E1p loop dynamics control covalent catalysis with ThDP and are
modulated by PDHc assembly, presumably by selection of catalytically competent loop conformations. This mechanism could be
a general feature of 2-oxoacid dehydrogenase complexes because such interfacial dynamic regions are highly conserved.
FADH2 is reoxidized to FAD by NAD+, finally producing
NADH (Scheme 1A).
X-ray crystallographic studies on the E1p component, an α2
INTRODUCTION
The pyruvate dehydrogenase multienzyme complex from
Escherichia coli (PDHc) is a 4500 kDa molecular machine, a
■
homodimer, revealed a disorder-to-order transition of two
member of the superfamily of enzymes that catalyze oxidative
dynamic loops of E1p spanning residues 401−413 (inner active
decarboxylation of 2-oxoacids, generating the corresponding
acyl-Coenzyme A and NADH. The complex catalyzes the
center loop) and 541−557 (outer active center loop) seen only
in the presence of a stable predecarboxylation intermediate
oxidative decarboxylation of pyruvate according to the overall
reaction in eq 1:1
analogue C2-α-phosphonolactylThDP (PLThDP) in the active
site.5 The intermediate analogue interacts with a conserved
pyruvate + CoA + NAD+
→ acetyl−CoA + NADH + CO2 + H+
histidine residue His407, thus ordering the inner loop, and
interaction between residues Asn404 and Asp549 further orders
the outer loop over the active site, thus forming a new surface
during the life of the closed state (Figure 1). In the H407A E1p
(1)
The PDHc is comprised of three enzyme components: (1)
variant, binding and reductive acetylation of E2p lipoyl domains
pyruvate dehydrogenase (E1p, EC 1.2.4.1) present in 24 copies
were highly impaired, suggesting an important role in steps
with a mass of 99 474 Da; (2) dihydrolipoyl acetyltransferase
leading to transfer of acetyl moiety from the E1p active site to
(E2p, EC 2.3.1.12) also present in 24 copies with a mass of 65
the lipoyl group on E2p.6 In addition, extensive biochemical
959 Da; and (3) dihydrolipoyl dehydrogenase (E3p, EC
1.8.1.4) present in 12 copies with a mass of 50 554 Da.2−4 The
and biophysical studies by Kale et al. using inner loop variants
by substitutions at charged residues showed that the disorder-
E1p component catalyzes thiamin diphosphate (ThDP)-
to-order transition is essential for substrate entry and to
dependent decarboxylation of pyruvate, whose product
sequester the E1p active sites to prevent undesirable side
reductively acetylates the covalently bound lipoamide group
reactions.7 EPR studies using a nitroxide label introduced into
on E2p. The acetyl group is then transferred from
acetyldihydrolipoamide to Coenzyme A (CoA) bound at the
catalytic domain of E2p, then dihydrolipoamide is reoxidized by
a FAD group bound to the E3 component, while the resulting
Received: June 26, 2012
Published: October 22, 2012
© 2012 American Chemical Society
18644
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
Scheme 1. (A) Mechanism of E. coli Pyruvate Dehydrogenase Complex with Role of ThDP on E1p; and (B) Oxidation of
Enamine by DCPIP
the inner loop revealed that the loops adopt two conformations
in a dynamic equilibrium during the unliganded state, and the
binding of a substrate analogue, methyl acetylphosphonate
(MAP), shifts this equilibrium to the closed state.8,9 19F NMR
experiments revealed that the transitions occurred in the
millisecond−second time scales. Stopped-flow CD studies
suggested that the loop dynamics influence the covalent
addition of substrate to ThDP.9,10 On the basis of these
results, we formulated “modulation of the catalytic cycle in
PDHc by loop motion through coupling of dynamics and
catalysis” as our working hypothesis.
To further test our hypothesis, we wanted to characterize the
individual intermediates and determine their rates of trans-
formation during the catalytic cycle of PDHc. We could then
(1) assess the overall rate-limiting step(s) in the PDHc cycle,
and (2) understand the effect of coupling of E1p loop dynamics
to the catalytic events at the individual active sites. However,
Figure 1. Dynamic loops positioned near the active site channel in
E1p containing PLThDP. In the closed conformation, the ThDP C2
atom and MAP only are exposed to the solvent accessible channel.
H407 from the inner loop (401−413, yellow ribbon, surface) makes
H-bond contact with the intermediate analogue. D549 from the outer
loop (541−557; orange ribbon, surface) makes H-bond contact with
N404. Figure illustrated using PyMol v 1.4.1 with coordinates from
PDB ID 2G25.
obtaining net forward rate constants for the various reactions is
challenging in view of the fact that there are multiple covalent
intermediates (Scheme 1A). An ingenious method to identify
and quantify ThDP-bound covalent intermediates by their
1
C6′−H H NMR resonances after acid quench of enzyme
reaction mixtures was invented by Tittmann and Hubner
̈
(TH).11 ThDP-bound covalent intermediates are typically (i)
stable under acid conditions, (ii) released into the supernatant,
and (iii) can be simultaneously detected by their well-resolved
18645
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
C6′−H 1H NMR resonances. As early examples of this
approach, detection and kinetic characterization of covalent
intermediates on the human E1p component was achieved.12,13
To extend the capabilities of the TH method to detect and
characterize ThDP-bound covalent intermediates in the
complex reaction mixture of the entire multienzyme complex
(4 500 000 Da), we synthesized [C2,C6′-13C2]ThDP. We here
report transient state kinetics on PDHc and on variants created
by substitutions in the two mechanistically important E1p
active center loops. Using PDHc comprising E2p−E3
subcomplex together with E1p or E1p variants that was first
reconstituted with specifically labeled [C2,C6′-13C2]ThDP, we
standard assay buffer [20 mM KH2PO4 (pH 7.0), 0.5 mM ThDP, 2
mM MgCl2, 0.2 mg/mL NADH, 0.08 mg/mL yeast alcohol
dehydrogenase] at 30 °C. The reaction was started with addition of
0.05 mg/mL (0.5 μM) E1p, and the disappearance of NADH was
monitored at 340 nm using a Varian DMS 300 spectrophotometer.
Reconstitution of Apoenzymes with [C2,C6′-13C2]ThDP. E1p
and 1-lip PDHc apo-enzymes (free of ThDP) were prepared by
dialysis of the holoenzyme twice (4 h at 4 °C) against 2 L of dialysis
buffer [20 mM KH2PO4 (pH 7.0) 0.1 mM DTT, 0.1 mM
benzamidine·HCl]. All stock solutions and dilutions were made in
the reaction buffer [20 mM KH2PO4 (pH 7.0)]. [C2,C6′-13C2]ThDP
stock solution (10 mM) was prepared in 20 mM KH2PO4 (pH 5.5).
2.5 mL of a solution of E1p or variant (0.20 mM active sites)
purified in the apoenzyme form was incubated with 1 equiv of
[C2,C6′-13C2]ThDP (0.20 mM) and Mg2+ (2.5 mM) for 30 min at 4
°C. Reconstitution was tested by DCPIP assay of small aliquots (10
μL). 2.5 mL of 1-lip PDHc purified in the apoenzyme form (0.20 mM
active sites of E1p) was incubated with 1 equiv of [C2,C6′-13C2]ThDP
(0.20 mM) and Mg2+ (2.5 mM) for 30 min at 4 °C. Reconstitution
was tested by overall PDHc assay of small aliquots (10 μL).
1-lip PDHc containing E1p variants was assembled from individual
components as follows: 2.5 mL of E1p variant (0.20 mM) purified in
the apoenzyme form was incubated with an equal part by protein
content (1:1:1 mg of E1:E2:E3) of 1-lip E2p, and E3 was added from
stock. This assembled complex was incubated for 30 min at 4 °C.
Next, this mixture was loaded into an Amicon Centriprep 5 mL unit
and diluted with the reaction buffer. This diluted mixture was
centrifuged at 2900g at 4 °C for 20 min. This washing step was
repeated three more times. The final concentration of the
reconstituted apo 1-lip PDHc was adjusted to 0.20 mM with reaction
buffer [20 mM KH2PO4 (pH 7.0)] and was incubated with 1 equiv of
[C2,C6′-13C2]ThDP (0.20 mM) and Mg2+ (2.5 mM) for 30 min at 4
°C. Reconstitution was tested by overall PDHc assay of 10 μL aliquots.
Rapid Chemical Quench and NMR Methods. [C2,C6′-13C2]-
ThDP labeled enzyme was loaded into syringe A and pyruvate was
loaded into syringe B in a rapid chemical quench apparatus (Kintek
RQF-3 model, Kintek Corp.). A quench solution (12.5% TCA in 1 M
DCl/D2O) was loaded into syringe C. Enzyme and pyruvate solutions
were mixed rapidly in a 1:1 volume ratio by a computer-controlled
drive-plate and incubated for predetermined times in a reaction loop.
After this incubation period, the reaction mixture was mixed rapidly
with the quench solution from syringe C, and the resulting mixture
was collected. For longer time scales (>10 s), the samples were
assembled similarly into three different solutions. Next, equal volumes
of the enzyme (200 μL) and substrate (200 μL) solutions were mixed
in an Eppendorf tube, and after predetermined times the reaction was
stopped by addition of quench solution (200 μL). The contents of the
syringes were varied to monitor the following different reactions:
1
performed 1D H−13C gradient carbon heteronuclear single
quantum coherence (gCHSQC) NMR experiments. In the
relevant 7.1−8.7 ppm aromatic region of the spectra, protons
attached to 13C only were prominent and thus enabled
detection of ThDP-related resonances even in the presence
of large concentrations of aromatic substrates, products, and
cofactors such as NAD+, NADH, CoA, acetyl-CoA, FAD, and
FADH2. Direct mass spectral detection of the rate of reductive
acetylation of an E2p lipoyl domain construct in the presence of
E1p and pyruvate under steady-state and single-turnover
conditions provided lower limits for the rate of acetyl transfer
between the ThDP of E1p and lipoamide of the E2p
component.
The results suggest that formation of the first covalent
predecarboxylation intermediate C2α-lactylthiamin diphos-
phate (LThDP in Scheme 1A) is rate limiting for the initial
series of steps leading to the enamine intermediate, and in steps
culminating in acetyl-CoA formation. For the first time, we
could also assess the effects of complex assembly on some
individual rate constants and found that assembly modestly
ameliorated the rates of impaired active center dynamic loop
variants, suggesting that in the complex the loop dynamics that
control covalent catalysis are modulated by complex assembly.
MATERIALS AND METHODS
■
Materials. ThDP, NAD+, CoA, dithiothreitol (DTT), and
isopropyl β-D-1-thiogalactopyranoside (IPTG) were from USB
(Cleveland, OH); 2,6-dichlorophenolindophenol (DCPIP), sodium
pyruvate, and trichloroacetic acid (TCA) were from Sigma (St. Louis,
MO). [C3-13C]pyruvate, DCl, and D2O were from Cambridge Isotope
Laboratories (Andover, MA). All HPLC grade solvents were
purchased from Sigma (St. Louis, MO). Wizard Plus Minipreps
DNA Purification System was from Promega (Madison, WI).
Synthesis of [C2,C6′-13C2]ThDP is described in detail in the
Supporting Information.
(1) For E1p with pyruvate single turnover, syringe A contained E1p
(0.20 mM) reconstituted with [C2,C6′-13C2]ThDP (0.20 mM)
in reaction buffer; syringe B contained pyruvate (20 mM) in
reaction buffer.
(2) For E1p and DCPIP reductase reaction, syringe A contained
E1p (0.20 mM) reconstituted with [C2,C6′-13C2]ThDP (0.20
mM) in reaction buffer; syringe B contained pyruvate (20 mM)
and DCPIP (4 mM) in reaction buffer.
(3) For the overall PDHc oxidoreductase reaction, syringe A
contained 1-lip PDHc (0.20 mM) reconstituted with
[C2,C6′-13C2]ThDP (0.20 mM), NAD+ (5 mM), and CoA
(2 mM) in reaction buffer; syringe B contained pyruvate (20
mM).
(4) To monitor acetyl-CoA formation, syringe A contained 1-lip
PDHc (0.02 mM), ThDP (2 mM), Mg2+ (2.5 mM), NAD+ (5
mM), and CoA (2 mM) in reaction buffer; syringe B contained
[C3-13C]pyruvate (5 mM).
Bacteria, Plasmids, Protein Overexpression, and Purifica-
tion. The methods for plasmid purification, protein overexpression,
and purification were described earlier.7,9,10,14 On the basis of our early
observations that the kinetic behavior of 1-lip PDHc was very similar
to that of 3-lip PDHc,15 the 1-lip E2p and its truncated versions of LD-
E2p and DD-E2p were created.14 The lipoyl domain in 1-lip E2p
component is a hybrid lipoyl domain. The 85-amino acid residues
starting from the N-terminus comprise residues 1−33 from the N-
terminal end of the first lipoyl domain and residues 238−289 from the
C-terminal end of the third lipoyl domain of the wild-type 3-lip E2
component.16
Enzyme Activity Measurements. Overall PDHc activity was
measured by reconstitution of E1p variants with independently
expressed 1-lip E2p and E3 components as described earlier.17 The
E1p component-specific activity was determined by monitoring the
reduction of DCPIP at 600 nm.18 The E1p component-specific
acetaldehyde production was coupled to conversion of acetaldehyde to
ethanol by yeast alcohol dehydrogenase in the presence of NADH.19
The activity was measured in the presence of 5 mM pyruvate in
The quenched reaction mixture was collected and centrifuged at 15
700g for 30 min, the supernatant was separated and filtered through
Gelman 0.45 μM discs, and the filtrate was analyzed by gCHSQC
NMR. All NMR experiments were carried out on a Varian 600 MHz
18646
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
Figure 2. Distribution of ThDP-bound covalent intermediates in reactions of E1p and PDH complex. gCHSQC NMR spectra of the supernatant
after acid quench of PDHc and removal of protein from the reaction. From top: (first) 20 ms quench of the reaction of PDHc; (second) 50 ms
quench of the reaction of E1p with pyruvate and DCPIP; (third) 20 ms quench of the reaction of E1p with pyruvate; and (fourth) control, PDHc
reaction mixture quenched before addition of pyruvate. ThDP-derived peaks are marked at 9.71 ppm (C2−H), 8.62 ppm (AcThDP), 8.01 ppm
(C6′−H), 7.37 ppm (AcThDP), and 7.34 ppm (HEThDP). Other peaks are due to NAD+, CoA, DCPIP, NADH, acetyl-CoA, or DCPIPH2.
NMR instrument. Data acquisition was at 25 °C, and sample pH was
product over total acetyl groups was plotted and fit to a single-
exponential eq 6.
1
∼0.75. 16 384 transients in the H dimension were collected with a
recycle delay time of 2 s.
f = f0 + f1 × (1 − e−k t
)
1
The relative integrals of C6′−H signals of respective ThDP-bound
covalent intermediates represent their relative ratios after correction
for the amount of unbound ThDP estimated using eq 2.
(6)
Lipoylation of the E2p Lipoyl Domain and E2p Didomain in
Vitro. To ensure full lipoylation of the lipoyl domain (LD-E2p,
residues 1−96 in E2p) and didomain, comprising lipoyl domain,
subunit binding domain, and linkers (DD-E2p, residues 1−190 in
E2p) required for reductive acetylation assay, these domains were
lipoylated in vitro using E. coli lipoyl protein ligase.20 The reaction
mixture contained in 50 mM ammonium bicarbonate (pH 7.0): ATP
(1.2 mM), MgCl2 (1.2 mM), lipoic acid (0.6 mM), lipoyl protein
ligase (10 μM,) and lipoyl domain (60 μM) or didomain (60 μM) for
1 h at 25 °C followed by incubation overnight at 4 °C. Lipoylation was
confirmed by Fourier transform ion cyclotron resonance mass
spectrometry (FT-ICR MS) using an electrospray ionization (ESI)
sampling method.
[ThDP]0 + [E]0 + KD
[E − ThDP] =
2
([ThDP]0 + [E]0 + KD)2
−
− [ThDP]0 [E]0
4
[ThDP] = [ThDP]0 − [E − ThDP]
(2)
Here, [E−ThDP] represents the concentration of E−ThDP
complex; [ThDP] and [ThDP]0 represent concentrations of free
and total ThDP; KD is the dissociation constant.
Time course of the fraction abundance of an intermediate over the
total active sites present (sum total of integrals for C6′−H signals of
ThDP and all ThDP-bound covalent intermediates) was fit to single
exponential eq 3 or double exponential equations of the form in eq 4
or 5.
Reductive Acetylation of the E2p Lipoyl Domain by E1p
under Steady-State Conditions. Experimental details are presented
in the legend of Supporting Information Figure S3. A FT-ICR MS-
based discontinuous assay method was employed to simultaneously
detect LD-E2p and acetyl-LD-E2p at various times.
f = f1 × (1 − e−k t
)
Reductive Acetylation of the Lipoyl Domain in a Single-
Turnover Experiment. For this experiment, the lipoyl domain (40
μM) and E1p (40 μM) or H407A variant (30 μM) in 50 mM
(NH4)2CO3 (pH 7.0) containing MgCl2 (8 mM) and ThDP (0.80
mM) in syringe A was mixed with pyruvate (4 mM) in syringe B, and
the reaction was stopped at times of 0.005−0.15 s by addition of 83 μL
of quench solution containing 50% methanol and 2% formic acid.
Samples were diluted to result in a concentration of LD-E2p of 1−2
μM in 50% methanol and 0.2% formic acid and were analyzed by ESI-
FT ICR-MS. The fraction of acetylated LD-E2p at different times was
determined by taking a ratio of the relative intensity of the acetylated
form to the total relative intensity (sum of unacetylated and acetylated
1
(3)
(4)
(5)
f = f1 × (1 − e−k t) + f2 × (1 − e
)
−k2t
1
f = f0 + f1 × (e−k t) − f2 × (e
)
−k2t
1
In experiments monitoring acetyl-CoA formation, [C3-13C]pyruvate
was used, and the product [C2-13C]acetyl-CoA was detected using the
gHSQC NMR method. The C3−H signals from pyruvate in both
ketone and hydrate forms and the C2−H signal from acetyl-CoA were
integrated, and the time course of fraction relative abundance of
18647
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
LD-E2p). Time dependence of this fraction was plotted, and the data
were fitted to eq 3 using Sigma Plot 10.0.
RESULTS
■
Detection of ThDP-Bound Covalent Intermediates by
NMR. Isotopically labeled [C2,C6′-13C2]ThDP enabled
selective detection of ThDP-bound covalent intermediates
using gCHSQC NMR experiments. In a typical H-decoupled
experiment, data were collected in the H dimension, and only
1
1
protons directly attached to 13C atoms were detected. Given
the 1% natural abundance of 13C, [C2,C6′-13C2]ThDP-bound
covalent intermediates could be detected and quantified even in
the presence of large molar excess of aromatic cofactors over
enzymes, except when 1H resonances from other cofactors and
intermediates overlap (Figure 2). This enabled us to extend the
capabilities of the TH method to include large complexes such
as the PDHc, which require additional aromatic substrates.
NMR spectra from typical experiments show well-resolved
resonances of [C2,C6′-13C2]ThDP-bound covalent intermedi-
1
ates in the H dimension at the same positions as do their
1
unlabeled counterparts in H NMR spectra (see Supporting
Information Table S1 for chemical shifts). Integration of areas
under these peaks provides a quantitative estimate of their
fractional abundance. The LThDP intermediate was not
detected in any of our studies; presumably it did not
accumulate under the various reaction conditions tested. Just
as the TH method, the current method could not distinguish
between the enamine and HEThDP intermediates; the former
is detected as HEThDP under acid quench conditions.
Formation of LThDP Is the Rate-Limiting Step in E1p
Reaction with Pyruvate. In the absence of E2p−E3
subcomplex, the E1p component converts pyruvate to
acetaldehyde, and acetaldehyde release (kcat = 0.12 s−1) is the
rate-determining step.21 E1p reconstituted with [C2,C6′-13C2]-
ThDP was mixed with pyruvate, and the reaction was stopped
at times in the range of 0.005−30 s (see gCHSQC NMR
spectra in Figure 3). Only a resonance at 7.34 ppm pertaining
to the 13C6′−H of HEThDP could be detected. The resonance
pertaining to LThDP (7.26 ppm) could not be detected
(Figure 3) and leads to the conclusion that the decarboxylation
of LThDP is much faster than its formation. The transient
nature of LThDP both from chemical models and during the
catalysis of many pyruvate decarboxylating enzymes is discussed
by Kluger and Tittmann.22 The 13C6′−H resonances for
HEThDP and ThDP (at 8.01 ppm) show a time-dependent
increase and decrease, respectively, indicating conversion of
ThDP to HEThDP and accumulation of HEThDP in E1p
active sites while approaching steady state. Integration of the
signals and a correction for fraction unbound ThDP (eq 2) at
each time point yields fraction relative abundance of HEThDP
(Figure 3, bottom; Table 1), which fitted to a single exponential
(eq 3) yields an apparent first-order rate constant of 117 14
s−1 for formation of HEThDP. This composite rate constant
includes contributions from net forward rate constants for
formation of the Michaelis complex (MM, which is formed
within the dead-time of the instrument), LThDP formation,
and decarboxylation to enamine (Scheme 1). However, the
major contribution to the rate is from the LThDP formation
step (the other steps are faster), and so it can alternatively be
viewed as the rate of formation of LThDP. We also emphasize
that under acid quench conditions the enamine intermediate is
protonated to HEThDP, its conjugate acid.
Figure 3. Distribution of ThDP-bound covalent intermediates during
reaction of E1p with pyruvate. (Top) gCHSQC NMR spectra of the
C6′-H resonances of ThDP-bound intermediates (6.5−9.0 ppm).
Reaction of E1p (0.1 mM) with pyruvate (10 mM) was quenched in
acid at the indicated times. (1) Control sample: no pyruvate, (2) 0.02
s, (3) 0.05 s, (4) 0.1 s, (5) 0.5 s, (6) 5 s, (7) 30 s. ThDP (C6′−H), δ
8.01 ppm; HEThDP (C6′−H), δ 7.34 ppm. (Bottom) Fraction
○
HEThDP ( ) calculated using ratio of C6′−H signal of HEThDP to
total C6′−H signals from HEThDP and ThDP determined at times 0,
0.005, 0.02, 0.05, 0.1, 0.5, 5, and 30 s during E1p reaction with
pyruvate. The data points were fit using eq 8, and the line is the
regression fit trace.
Chemical Oxidation of the Enzyme-Bound Enamine
Affords Transient Detection of AcThDP on E1p. The
central enamine intermediate reductively acetylates the lip-
oamide group on E2p, a reaction that can be viewed as (1) a 2-
electron oxidation to 2-acetyl-ThDP and concomitant reduc-
tion to dihyrolipoyl-E2p, followed by (2) transfer of the
resultant acetyl group to the dihydrolipoyl-E2p with the
dithiolane ring of lipoamide participating in both steps. In
related oxidative decarboxylases such as pyruvate oxidase and
pyruvate ferredoxin oxidoreductase, the enamine is oxidized by
enzyme-bound cofactors such as FAD or FeS clusters,
respectively, and the resultant acetyl moieties are transferred
to phosphate or CoA acceptor groups, respectively. This allows
accumulation of 2-acetylthiamin diphosphate (AcThDP) in the
latter enzymes when the acceptors are absent.23−25 However,
during the overall PDHc pathway, because the lipoamide serves
both functions and AcThDP does not appear to accumulate at
significant levels as shown by Frey et al.,26 this hinders pre-
steady-state studies aimed at determining the rates of oxidation.
Alternatively, an artificial electron acceptor can be used to study
18648
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
Table 1. Comparison of Pre-Steady-State Rate Constants for E1p and Variants under Various Conditions
kHEThDP (s−1
)
a
b
c
variant
kcat
,
s−1
pyruvate only
pyruvate + DCPIP
73.2 8.8
PDHc reaction
kr (s−1) reductive acetylation
E1p
95 12
0.7 0.1
117 14
12.9 1.0
1.16 0.05
0.13 0.02
25.9 1.2
116 25
12.1 0.5
2.5 0.19
0.59 0.05
51.7 5.4
D549A
H407A
E401K
Y177A
0.08 0.01
0.83 0.05
3.31 0.3
0.06 0.01
0.02 0.001
0.016 0.003
a
b
The kcat values were determined from the overall PDHc activity assay on reconstitution with E2−E3 subcomplex. E1p-specific reaction with
c
DCPIP assay on reconstitution with E2−E3 subcomplex. For reaction of reductive acetylation of LD-E2p with E1p or variant and pyruvate using
ESI-FT-MS.
the oxidation step of E1p with pyruvate, and DCPIP as a two-
electron acceptor is commonly used to assay ThDP-dependent
oxidative decarboxylases. DCPIP (monitored at λmax = 600 nm,
Scheme 1B) oxidizes the central enamine intermediate
producing the reduced form DCPIPH2. While the kcat of 0.65
0.16 s−1 determined by this assay is 100-fold lower than the
overall kcat of 95 12 s−1 for PDHc, DCPIP serves as a quick
and efficient assay to study oxidation of the enzyme-bound
enamine for E1p variants, confirming that decarboxylation has
taken place.
In pre-steady-state experiments, acid quench of reaction
mixtures of E1p reconstituted with [C2,C6′-13C2]ThDP with
pyruvate and DCPIP during 0.005−30 s, the NMR spectra
revealed resonances at 7.34 ppm for 13C6′−H of HEThDP, and
at 7.36, 7.37, and 8.62 ppm for the 13C6′−H of AcThDP
(acetyl, hydrate, and internal carbinolamine forms), but none
for LThDP (Figure 4). The resonance for HEThDP displays a
time-dependent increase reaching a steady state during early
times (0−0.5 s). The resonances for AcThDP show a time-
dependent rapid increase during short time scales (0.0−0.5 s)
and a decrease at longer time scales (0.5−30 s) corresponding
to the total consumption of DCPIP, and subsequent slow
hydrolysis of AcThDP to acetate and ThDP. Data treatment, as
described above, yielded an apparent first-order rate constant of
57
6 s−1 for AcThDP formation and 73.2
8.8 s−1 for
enamine formation (Figure 4, bottom inset; Table 1). It is
assumed that the total amount of enamine during any time
point is the sum total of AcThDP and HEThDP). The
fractional abundance of AcThDP was also fit to a biexponential
process (eq 5) yielding an apparent rate constant of 52 7.5
s−1 for AcThDP formation, and a first-order rate constant of
0.23 0.05 s−1 for its hydrolysis (Figure 4, bottom; see also
Table 2 in the Discussion).
Figure 4. Distribution of ThDP-bound covalent intermediates during
the reaction of E1p with pyruvate and DCPIP. (Top) gCHSQC NMR
spectra of the C6′−H resonances of ThDP-bound intermediates (6.5−
9.0 ppm). Reaction of E1p (0.1 mM) with pyruvate (10 mM) and
DCPIP (2 mM) was quenched in acid at the indicated times. (1) 0.02
s, (2) 0.05 s, (3) 0.1 s, (4) 0.5 s, (5) 5 s, (6) 30 s. ThDP (C6′−H), δ
8.01 ppm; HEThDP (C6′−H), δ 7.34 ppm; 2-acetylThDP (C6′−H),
δ 8.62 ppm (carbinolamine), 7.38 and 7.37 ppm (hydrate and keto
forms). (Bottom) Time course of AcThDP: Steady state is reached
within 0.5 s; next, consumption of DCPIP leads to depletion of
AcThDP and formation of HEThDP. The red trace is the regression fit
line to eq 5. (Inset) Fraction intermediate (AcThDP (red), HEThDP
(black)) calculated using ratio of C6′−H signals of intermediate to
total C6′−H signals from AcThDP, HEThDP, and ThDP determined
during the first 0.5 s.
The kinetic data show that the enamine is rapidly oxidized on
E1p in the presence of DCPIP.27
The hydrolysis of AcThDP appears to be the slow step (k′5 =
0.23 0.05 s−1; see Scheme 2A in the Discussion) explaining
the discrepancy between the steady-state kcat values derived
from the DCPIP-based assay (0.65
activity-based assay (95
0.16 s−1) and PDHc
12 s−1). While the enamine is
oxidized by DCPIP at a rate comparable to its formation (see
above), the AcThDP intermediate thus generated is hydrolyzed
at a very slow rate, similar to the nonenzymatic hydrolysis rate
of synthetic 2-acetylThDP at pH values of 7−7.5.28 Even when
bound to enzyme, the rate of AcThDP hydrolysis is unaffected,
hence the low kcat. Nevertheless, the DCPIP-based artificial
oxidation reaction is a relevant model for studying the oxidation
of the enamine on ThDP dependent oxidative decarboxylases at
the individual active site level using pre-steady-state kinetics
(Scheme 1B). While the mechanism of this artificial oxidation
step is unclear (location of DCPIP on the enzyme, electron
transfer pathway, etc.), it is clear that AcThDP is the
intermediate in this reaction.
18649
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
ThDP-Bound Covalent Intermediates on E1p during
the Full Catalytic Cycle of PDHc. Having confirmed that the
major ThDP-bound covalent intermediates could be detected
in a time-resolved manner using our method, PDHc was
reconstituted in the presence of all substrates to determine: (a)
the fate of the covalent intermediates during the catalytic cycle
and (b) if complex assembly has any effect on catalysis in the
E1p active sites. PDHc reconstituted with [C2,C6′-13C2]ThDP
was mixed with pyruvate, CoA, and NAD+, and the reaction
was stopped at 0.005−30 s (Figure 5). As with E1p alone, only
method. While AcThDP could be detected during the artificial
oxidation reaction, it presumably does not accumulate to
detectable levels according to NMR during the overall PDHc
reaction. This is consistent with the following scenarios: (i) the
rate of acetyl transfer from E1p to dihydrolipoylE2p is faster
than the rate of oxidation of the enamine to AcThDP in a
stepwise mechanism, or (ii) a concerted reductive acetylation
mechanism proceeding by a tetrahedral intermediate transiently
cross-linking the E1p and lipoylE2p, without a distinct AcThDP
accumulating, as modeled by Pan and Jordan.30
The resonances for HEThDP and ThDP display a time-
dependent increase and decrease, respectively, indicating a
conversion of ThDP to HEThDP. Analysis (eq 3) yields a
fraction of relative abundance of HEThDP leading to an
apparent first-order rate constant of 117
formation of HEThDP (Figure 5, bottom; see also Tables 1
and 2), as compared to the overall kcat of 95
12 s−1
determined from Vmax (57 7 activity units per E1 active
25 s−1 for the
center) for NADH production for 1-lip PDHc.
Conversion of Pyruvate to Acetyl-CoA. We also tested
the kinetic competence of the intermediates by directly
monitoring the conversion of [C3-13C]pyruvate to [C2-13C]-
acetylCoA by PDHc using gCHSQC NMR (Supporting
Information Figure S1). Quantification using integration of
1
13C bound H signals yielded an initial velocity of 1219 66
μM/s for acetyl-CoA production under the reaction conditions
(10 μM E1p active centers in the reaction loop). Assuming full
saturation of all components, this initial velocity corresponds to
a kcat of 122
6.6 s−1 per E1p active centers (Supporting
Information Figure S1, bottom), a number remarkably similar
in magnitude to the other rate constants on the pathway
determined in this work (Table 1).
Reductive Acetylation of LD-E2p Is Not the Rate-
Limiting Step in the Overall PDHc Reaction. Reductive
acetylation of the lipoyl moiety covalently bound to the lipoyl
domain of E2p is the final step involving ThDP-bound covalent
intermediates. The reaction of enamine derived from C2α-
methoxybenzyl-3,4,5-trimethylthiazolium salt with S-methylli-
poic acid methyl ester, a mimic of the S-protonated form, was
shown to be a chemically competent model for this step.30 The
free lipoic acid itself is a poor substrate both in chemical
models31 and for E1p-bound enamine.32 Also, E1p has been
shown to successfully reductively acetylate the independently
expressed LD-E2p,6,7 providing the most appropriate model
reaction to study the rate of the final step in E1p catalysis while
conserving both the chemistry and the intercomponent
communication due to its specific recognition of E1p.33−35
Therefore, a method was developed to detect LD-E2p and
acetyl-LD-E2p simultaneously in the quenched reaction
mixtures using ESI-FT-ICR MS (Supporting Information
Figure S2). Fortuitously, the acetyl-LD-E2p and indeed the
acetyl-DD-E2p are stable under the reaction and MS
conditions.
Reductive Acetylation of LD-E2p under Steady-State
Conditions. First, the reductive acetylation by E1p (5 nM)
was monitored at varying concentrations of LD-E2p (1.5−20
μM) in the presence of 2 mM pyruvate. Under these
conditions, the LD-E2p is a substrate for E1p, and during
multiple turnovers the acetylated LD-E2p accumulates as the
product of the reaction (Supporting Information Figure S3 and
Figure 6). A plot of the initial slope/min calculated from the
progress curve of a time-dependent accumulation of acetylated
LD-E2p versus concentrations of LD-E2p resulted in a
Figure 5. Distribution of ThDP-bound covalent intermediates during
PDHc reaction. (Top) gCHSQC NMR spectra of the C6′−H
resonances of ThDP-bound intermediates (6.5−9.0 ppm). Reaction of
E1p (0.1 mM) with pyruvate (10 mM), NAD+ (2.5 mM), and CoA (1
mM) was quenched in acid at the indicated times. (1) Control sample:
PDHc and cofactors without pyruvate, (2) 0.02 s, (3) 0.05 s, (4) 0.1 s,
(5) 0.5 s, (6) 5 s, (7) 30 s. ThDP (C6′−H), δ 8.01 ppm; HEThDP
(C6′−H), δ 7.34 ppm. Other peaks due to NAD+, CoA substrates, and
○
NADH and acetyl-CoA products. (Bottom) Fraction HEThDP ( )
calculated using ratio of C6′−H signal of HEThDP to total C6′−H
signals from HEThDP and ThDP determined at various times (0,
0.005, 0.02, 0.05, 0.1, 0.5, 5, 30 s) during the PDHc overall reaction.
The trace is the regression fit line to eq 3.
HEThDP could be detected at all times; that is, LThDP and
AcThDP do not accumulate to detectable levels. Two studies
on the transient nature of the AcThDP during PDHc catalysis
showed that (i) it was chemically competent and could be
transferred to the dihydrolipoate when generated on the
enzyme using 3-fluoropyruvate, and (ii) using [C3-14C]-
pyruvate, 0.5% of PDHc active sites were found to contain
AcThDP during the overall PDHc reaction, while up to 12% of
active-sites were found to contain AcThDP in an artificial
enamine oxidation reaction using ferricyanide.28,29 These
results are similar to the current findings using the NMR
18650
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
Figure 6. Michaelis−Menten plot for steady-state reductive acetylation
of LD-E2p by E1p and pyruvate. For experimental details, see legend
of Figure S3. Data were fitted to a Michaelis−Menten equation (vo =
vomax[LD-E2p]/(Km + [LD-E2p]), and the trace is the regression fit
line.
hyperbolic curve yielding a KM of 0.73 0.15 μM for LD-E2p
(Figure 6), a much lower value than the 26 μM reported
earlier.32
Pre-Steady-State Kinetics of Reductive Acetylation of LD-
E2p by E1p. The KM for LD-E2p from above suggested
conditions for a single turnover kinetic experiment, a better
model of the reductive acetylation as it occurs during PDHc
catalysis. At high concentrations of E1p (40 μM active sites)
and LD-E2p (40 μM), formation of the E1p−LD-E2p complex
is ensured to be spontaneous, and chemical steps become rate
limiting in a pseudo first-order process. Also, in contrast to
steady-state conditions described above, single turnover
conditions ensure that product release steps do not impede
rates. Samples collected in the time range of 0.005−0.15 s were
analyzed using ESI−MS, and a rate constant of 52 5 s−1 for
reductive acetylation of LD-E2p by E1p was calculated (Figure
7, top). In a similar experiment, a value of 50 5 s−1 confirmed
reproducibility of the data (not shown). This pseudo first-order
rate constant is composed of the rates of conversion of pyruvate
to the enamine and the reductive acetylation of the lipoyl
domain by the enamine (Scheme 1A). Because the rate
constant for reductive acetylation of LD-E2p (52 s−1) was
lower than that of acetyl-CoA formation (kcat = 122 s−1), a
similar experiment was also conducted with an E2p didomain
comprising the lipoyl domain, the subunit-binding domain, and
linker regions (DD-E2p), presenting a better mimic of E2p by
providing a binding domain for E1p. Rate constants of 134 and
96 s−1 (Figure 7, bottom) were obtained in two independent
experiments, also clearly indicating that reductive acetylation of
the lipoyl domain is not a rate-limiting step in the overall PDHc
reaction.
In contrast, the H407A substitution in E1p drastically
reduces the rate of reductive acetylation of LD-E2p and DD-
E2p, lowering the rate constants to 0.02 0.001 s−1 (LD-E2p)
and 0.04 0.01 s−1 (DD-E2p) (Table 1), indicating that the
residue His 407 participates in communication between E1p
and E2p, as was suggested earlier.6,7 Moreover, for this variant,
reductive acetylation becomes rate determining, as the
preceding steps up to formation of HEThDP (2.5 s−1) are
50-fold faster in the reaction reconstituting H407A E1p to
PDHc (Table 1 and Supporting Information Figure S5).
Substitutions at the Mobile Active-Center Loops
Impair the First C−C Bond Formation Step. Next, to test
the role of active center dynamic loops on individual catalytic
Figure 7. Time-dependence of fraction acetylated LD-E2p and DD-
E2p during single turnover reaction. Top: The LD-E2p (40 μM) and
E1p (concentration of active centers/subunit = 40 μM) in syringe A
was mixed at 25 °C with pyruvate (4 mM) in syringe B (see details
under Materials and Methods). Reaction was stopped at times 0.005−
0.15 s by addition of quench solution, and samples were analyzed by
FT ICR- MS. The intensity of the MS signal for acetylated and
unacetylated LD-E2p was detected. The relative intensity of acetylated
LD-E2p versus total intensity (sum of acetylated and unacetylated LD-
E2p) was plotted versus time. Data were fitted to a single exponential
(eq 3). Bottom: The E1p (concentration of active centers/subunit =
30 μM) and DD-E2p (30 μM) in 50 mM ammonium bicarbonate (pH
7.0) containing 0.4 mM ThDP and 2.0 mM MgCl2 in syringe A were
mixed at 25 °C with pyruvate (4 mM) in 50 mM ammonium
bicarbonate (pH 7.0) in syringe B. Reaction was stopped at times
0.005−1.0 s by addition of quench solution, and samples were
analyzed by FT ICR-MS. Data were fitted to a double exponential (eq
4).
steps, we selected E401K (inner loop hinge residue
substitution), H407A (inner loop His interacting with
PLThDP), and D549A (outer loop Asp interacting with inner
loop Arg404) for pre-steady-state kinetic studies (see Figure 1).
In an experiment similar to that carried out with E1p, loop
variants (E401K, H407A, D549A, or Y177A E1p) were
reconstituted with [C2,C6′-13C2]ThDP and mixed with
pyruvate, and the reactions were stopped at 0−300 s. Only
the resonance at 7.34 ppm pertaining to the 13C6′−H of
HEThDP could be detected for all variants at all time points,
again indicating no LThDP accumulation in any of the loop
variants tested. Data treatment yielded apparent first-order rate
constants for formation of HEThDP (Supporting Information
Figures S4−S6, and Table 1). The rates decrease in the
following order: E1p > Y177A > D549A > H407A > E401K,
with increasing disruption of loop dynamics expected from
these substitutions. The Y177A E1p, a substitution in a
nonmobile active-site loop, was used as a control for
nonspecific effects of disruptions to the active site and showed
only a 4-fold reduction as compared to E1p (Supporting
Information Figure S7, and Table 1), whereas substitutions in
the dynamic loops elicited reductions of 10−900-fold. These
18651
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
apparent rate constants, which represent the rates of formation
of LThDP, indicate that the substitutions dramatically impair
the first C−C bond formation step. We also note that we
cannot determine the effect of these substitutions on the rate of
LThDP decarboxylation relative to E1p.
The 100-fold rate reduction observed in the H407A E1p
variant, where the H-bond connecting the inner loop to the
predecarboxylation intermediate is abolished, suggests a
significant role for this interaction in the predecarboxylation
steps. The drastic reduction of a 900-fold in the case of E401K
E1p (charge reversal substitution at a hinge residue), in
conjunction with the disordered loops seen in the structure of
E401K E1p in presence of PLThDP,7 suggests that active-site
sequestration, to avoid carboligase-type side reactions, is also
essential for formation of the first C−C bond.
Enamine Oxidation by DCPIP in Inner Loop E1p Variants.
When using DCPIP to convert the enamine to 2-acetylThDP,
unlike with E1p, only HEThDP could be detected in both
E401K E1p and H407A E1p. The rates of HEThDP formation
(0.13 s−1 for E401K E1p, and 1.16 s−1 for H407A E1p) with
pyruvate as the only reactant are similar to the rate of hydrolysis
of AcThDP (0.23 s−1) accounting for our inability to observe
significant buildup of this intermediate (Supporting Informa-
tion Figures S4−S6; see also Tables 1 and 2). Also, the
apparent rates of HEThDP accumulation in these experiments
(0.016 s−1 for E401K E1p) and (0.06 s−1 for H407A E1p) are
much lower than the corresponding rates in the absence of
DCPIP as presented above.
the-art.22 During the course of this study, synthesis of
[C2,C6′-13C2]ThDP provided the first observation by NMR
of the fate of ThDP-bound covalent intermediates during
catalysis by the entire PDHc from E. coli. In studies with model
C2α-lactylthiazolium salts, Zhang et al. showed the rate of
decarboxylation increased with decreasing solvent polarity from
water to tetrahydrofuran, and in tetrahydrofuran the first-order
rate constant exceeds 50 s−1.36 Presumably, the E1p active site
provides such a hydrophobic environment. In studies using
HEThDP as a substrate for E1p, Zhang et al. estimated up to
107-fold rate acceleration for ionization of the C2α−H bond of
HEThDP bound to E1p. This rate acceleration corresponds to
10 kcal/mol stabilization of the enamine intermediate on the
enzyme,27 and thus channels this electron-rich intermediate
(the enamine) toward oxidation by the lipoyl moiety covalently
attached to E2p. The free lipoic acid is a poor acceptor as
compared to an analogue mimicking its protonated form in
chemical models (108-fold rate enhancement)30 and LD-E2p in
enzymic models (kcat/KM enhancement of >104-fold).32 These
studies provided fundamental groundwork for our under-
standing of the results here presented: (a) LThDP decarbox-
ylation could be fast with low barriers in general, unless some
elements of the protein environment slow it down; (b) in E1p,
the enamine is highly stabilized, and its concentration almost
certainly corresponds to that of the HEThDP detected as a
result of the acid quench; and (c) the catalytic contributions for
reductive acetylation originate from E1p (activation of lipoic
acid) and LD-E2p.
Assembly to PDHc improves reaction rates of E1p mobile
active-center loop variants suffering from drastically impaired
rates.
Determination of Net Forward Rate Constants. A
kinetic model for catalysis in PDHc was constructed according
to Cleland’s net rate constants method,37 by substituting net
rate constants denoted with prime for individual forward and
reverse rate constants. The net rate constants denote the flux
through each step, and in case of irreversible steps, the net rate
constants would be identical to the true forward rate constants
(Scheme 2). For a series of reactions, the reciprocal of the net
rate constant for any individual step is its transit time. This
transit time model can be applied to determine each individual
net rate constant as a summation of transit times of the
previous steps.11,38
Next, the effect of reconstitution of the mobile loop variants
(E401K E1p, H407A E1p, and D549A E1p) with E2p−E3
subcomplex on catalysis was determined in the presence of
pyruvate, NAD+, and CoA. Similarly to all earlier experiments,
only HEThDP could be detected during the various time scales,
without apparent accumulation of LThDP. The apparent rates
of formation of HEThDP for reconstituted PDHc reactions are
summarized in Table 1 (also see Supporting Information
Figures S4−S6). Again, these apparent rate constants, which
represent the rates of formation of LThDP, decrease in the
order E1p > D549A > H407A > E401K, in direct relation to
increasing disruption of loop dynamics expected from these
substitutions.
Comparison of the rates of E1-specific reactivity (reaction
with pyruvate only) of parental and variant E1p’s provided
additional insights. There is an approximately 5-fold improve-
ment in the rate of HEThDP formation for E401K E1p, 2-fold
improvement for H407A E1p, and no change for D549A E1p
and for the E1p on assembly to complex. These results suggest
that assembly of complex has a modest yet clearly discernible
effect on catalysis by the E1p component. Presumably, the
interaction with lipoyl domains of E2p orders the inner loop
over the E1p active site leading to a modest recovery in active-
site sequestration. This PDHc assembly induced recovery is
maximal for the highly disrupted inner loop variants, and there
is no change for the outer loop variant or the PDHc.
1
1
k′1
1
k′2
1
k′3
=
+
+
and
kHEThDP
k1 × [pyruvate] × k2′
[pyruvate]
k1′ =
=
× k2′
× k4′
k−1 + k2′
KM
(7)
(8)
1
1
1
k′D
1
k′4
=
+
+
where
kAcThDP
kHEThDP
kon × [DCPIP] × k4′
[DCPIP]
KMDCPIP
k′D =
=
koff + k4′
1
1
1
k′D
1
+
=
+
where
kreductive acetylation
kHEThDP
k′4
kon × [LD] × k4′
koff + k4′
[LD]
KMLD
k′D =
=
× k4′
(9)
DISCUSSION
From eq 7, k2′, the net rate constant for formation of
■
Pioneering efforts in the syntheses and characterization of
central ThDP-bound covalent intermediates opened up
avenues for the application of NMR methods to thiamin
enzymology leading to the development of the current state-of-
LThDP, approximates to the experimentally observed rate
k
HEThDP for the following reasons: (i) under pyruvate saturating
conditions, the Michaelis complex accumulates within 2.5
ms;9,10 and (ii) decarboxylation is not rate limiting in any of the
18652
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
particular catalytic step(s) and their contribution to rate
enhancements during these steps. Our results suggest that,
similar to PDHc, formation of the first covalent predecarbox-
ylation intermediate LThDP is the rate-determining step for the
active-site loop variants. The rate of LThDP formation tended
to decrease drastically with increasingly disruptive substitutions
(Tables 1 and 2). Taken together with earlier structural and
thermodynamic studies, this implies that the loop dynamics and
the ensuing conformational changes upon substrate binding to
E1p contribute toward catalysis and LThDP formation. These
results also point to a sequential mechanism for loop closure,
whereby the inner loop has to close first providing a new
surface for interactions and closing of the outer loop. In such a
mechanism, affecting the inner loop would have strong
detrimental effects on catalysis, while the effect due to
disruptions of the outer loop will be of lesser importance to
LThDP formation.
Scheme 2. Transit Time Kinetic Model of E1p Reaction with
(A) DCPIP and (B) LD-E2p under Single Turnover
Conditions Modeling PDHc
The effect of complex assembly in the presence of the E2p
and E3 components, which can be interpreted as recovery of
E1p active-site sequestration, also follows a general trend. The
recovery seen is in the order E401K E1p (5-fold) > H407A E1p
(2-fold) > D549A E1p = E1p (no change) from minimal to
maximal recovery. Complex assembly helps maximally dis-
rupted variants, while it has minimal effect on the outer loop
variant and E1p. The overall kcat and rate of reductive
acetylation of the lipoyl domains are at least 10-fold lower
than the rate of HEThDP formation in the loop variants
(Tables 1 and 2), suggesting that in contrast to the PDHc,
reductive acetylation is rate-determining in the case of the inner
and outer loop variants. The active-site loops are disordered in
the crystal structures of both E401K E1p and H407A E1p in
the presence of PLThDP, whereas they are ordered in the E1p
(see yellow surface in Figure 1), suggesting that in PDHc, the
preformed surface is near optimal for interactions with the
lipoyl domain; hence assembly elicits no further changes.
However, in the loop variants, such a surface has to be induced
upon binding of E2p to enable the observed reductive
acetylation of lipoamide. Presumably, the effect of loop
dynamics on catalysis has been optimized by evolution for
the E1p.
experiments with E1p or variants. The net forward rate
constants, summarized in Table 2, suggest that the enamine is
similarly susceptible to oxidation by DCPIP, LD-E2p, or DD-
E2p. The net rate constants for reductive acetylation from this
study provide a lower limit for the physiological reaction,
because the attachment of the LD or DD-E2p to the core
domain of E2p by the flexible linker could provide added rate
enhancement by providing stringent diffusion limitations.39
This is evident upon comparison with the overall kcat (122 s−1)
for acetyl-CoA production by the PDHc. While reductive
acetylation was believed to be the rate-limiting step in PDHc
catalysis,40 the current study provides similar net rate constants
for several steps, which are within range of the overall kcat,
suggesting equalized barriers41,42 for the interconversion of
these intermediates by evolutionary optimization for substrate
channeling.43,44
The Effects of Assembly on Inner and Outer Loop
Variants. Thermodynamic analysis of substrate analogue MAP
binding to E1p produced nonlinear van’t Hoff plots (ΔCp =
−246 cal mol−1 K−1), suggesting a coupling of a conformational
change with ligand binding, and the corresponding conforma-
tional states were observed in the X-ray structure and in
solution studies using NMR and EPR spectroscopy.5,8,9 In
contrast, for the H407A E1p and E401K E1p variants with rigid
body binding behavior (ΔCp = 0 cal mol−1 K−1), the closed
conformation of the loops could not be observed in the crystal
structure. Our current study with substrate reveals the roles of
loop dynamics during the catalytic cycle by identifying the
The dynamic loop regions described in this study are highly
conserved in other homodimeric α2 and heterotetrameric α2β2
E1 components of 2-ketoacid dehydrogenases, as observed in
X-ray structures.45−50 The loops that undergo dynamic
equilibrium over the active sites in ThDP-dependent enzymes
carry out similar functions with different triggers. For example,
in human α2β2 E1 branched chain ketoacid dehydrogenase and
Table 2. Net Forward Rate Constants for Individual Steps in Oxidative Decarboxylation Reactions of E1p
a
b
reaction
k1′, s−1
k2′, s−1
k3′, s−1
k4′, s−1
k5′, s−1 (DCPIP)
a
PDH complex
>600
117 25
117 25
0.13 0.02
1.16 0.05
2.5 0.2
NA
fast
fast
fast
fast
fast
50
93.6
111.2
0.05
0.13
0.02
a
E1p DCPIP
>600
0.23 0.05
a
c
E401K E1p DCPIP
H407A E1p DCPIP
H407A E1p PDHc
>600
0.23 0.05
a
c
>600
0.23 0.05
a
>600
d
chemical models
3 × 10−3
6.6 × 104 (M−1 s−1
)
0.23
a
b
The Michaelis−Menten (MM) complex accumulates within 2.5 ms in E1p and loop variants.9 Because LThDP does not accumulate, we can only
infer k3′ ≫ k2′, and consequently the effect of loop substitutions on k3′ could not be determined. k4′ is the composite net rate constant involving
binding of DCPIP or lipoyl domain to E1p and reaction with enamine. k5′ is the forward rate constant for hydrolysis of 2-acetylThDP in DCPIP
reaction. For PDHc and H407A PDHc, because AcThDP could not be detected, k5′ is not determined, and k4′ includes the rate for oxidation of the
c
d
enamine and transfer of the acetyl group to the lipoyl domain. Assuming k5′ is similar to the uncatalyzed rate. The thiamin C2−H deprotonation is
rate-determining in k1′.55The rate constants for the chemical models are from refs 32, 28, and 29. NA is not available in the literature.
18653
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
(4) Stephens, P.; Lewis, H.; Darlison, M.; Guest, J. Eur. J. Biochem.
1983, 135, 519.
human α2β2 E1p, phosphorylation of an active site loop
destabilizes the closed/ordered loop conformation,51−53 where-
as in other examples, the stabilization of closed conformation is
induced by ThDP binding.47,50,54
In the present study, the closed conformation of the active
center loops is only apparent on formation of the
predecarboxylation LThDP intermediate, while the loop
variants indicate potential effects of complex assembly on
loop closing.
(5) Arjunan, P.; Sax, M.; Brunskill, A.; Chandrasekhar, K.; Nemeria,
N.; Zhang, S.; Jordan, F.; Furey, W. J. Biol. Chem. 2006, 281, 15296.
(6) Nemeria, N.; Arjunan, P.; Brunskill, A.; Sheibani, F.; Wei, W.;
Yan, Y.; Zhang, S.; Jordan, F.; Furey, W. Biochemistry 2002, 41, 15459.
(7) Kale, S.; Arjunan, P.; Furey, W.; Jordan, F. J. Biol. Chem. 2007,
282, 28106.
(8) Song, J.; Park, Y. H.; Nemeria, N. S.; Kale, S.; Kakalis, L.; Jordan,
F. J. Biol. Chem. 2010, 285, 4680.
(9) Kale, S.; Ulas, G.; Song, J.; Brudvig, G. W.; Furey, W.; Jordan, F.
Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 1158.
(10) Kale, S.; Jordan, F. J. Biol. Chem. 2009, 284, 33122.
(11) Tittmann, K.; Golbik, R.; Uhlemann, K.; Khailova, L.;
Schneider, G.; Patel, M.; Jordan, F.; Chipman, D. M.; Duggleby, R.
G.; Hubner, G. Biochemistry 2003, 42, 7885.
(12) Seifert, F.; Ciszak, E.; Korotchkina, L.; Golbik, R.; Spinka, M.;
Dominiak, P.; Sidhu, S.; Brauer, J.; Patel, M. S.; Tittmann, K.
Biochemistry 2007, 46, 6277.
(13) Seifert, F.; Golbik, R.; Brauer, J.; Lilie, H.; Schroder-Tittmann,
K.; Hinze, E.; Korotchkina, L. G.; Patel, M. S.; Tittmann, K.
Biochemistry 2006, 45, 12775.
CONCLUSION
■
Using de novo synthesized 13C-labeled thiamin and FT-MS, we
have developed new approaches capable of determining the
pre-steady-state rate constants for conversion of the 2-oxoacid
starting material to all intermediates, also including inter-
component communication between the E1p and E2p, and
formation of acetyl-CoA on E2p in the entire multimegadalton
pyruvate dehydrogenase complex.
For the first time, these approaches enabled us to deduce the
effect of complex assembly on microscopic rate constants. (a)
The results all point to the notion that formation of the first
covalent bond between substrate and the C2 atom of ThDP
(formation of LThDP) is rate limiting overall, and all of the
subsequent rate constants measured are very similar within
experimental error. This is consistent with rather similar
barriers having evolved for the various chemical conversions on
the enzyme. (b) Assembly to PDHc had no effect on the rate
constants on E1p, but assembly did improve the rate constants
for the least active inner loop variants.
Our earlier results pointing to a relationship between the rate
of formation of LThDP and mobility of dynamic active center
loops make these results even more intriguing. Combining the
information from this study and our previous one on loop
mobility on E1p leads to the important conclusion that loop
movement on the E1p component controls the rate-limiting
step in the entire multienzyme complex. The generality of our
findings here summarized remains to be tested on related 2-
oxoacid dehydrogenase complexes, both bacterial and human.
(14) Wei, W.; Li, H.; Nemeria, N.; Jordan, F. Protein Expression Purif.
2003, 28, 140.
(15) Yi, J.; Nemeria, N.; McNally, A.; Jordan, F.; Machado, R. S.;
Guest, J. R. J. Biol. Chem. 1996, 271, 33192.
(16) Ali, S. T.; Guest, J. R. Biochem. J. 1990, 271, 139.
(17) Nemeria, N.; Volkov, A.; Brown, A.; Yi, J.; Zipper, L.; Guest, J.
R.; Jordan, F. Biochemistry 1998, 37, 911.
(18) Nemeria, N.; Yan, Y.; Zhang, Z.; Brown, A. M.; Arjunan, P.;
Furey, W.; Guest, J. R.; Jordan, F. J. Biol. Chem. 2001, 276, 45969.
(19) Holzer, H.; Schultz, G.; Villar-Palasi, C.; Juntgen-Sell, J. Biochem.
Z. 1956, 327, 331.
(20) Jones, D. D.; Horne, H. J.; Reche, P. A.; Perham, R. N. J. Mol.
Biol. 2000, 295, 289.
(21) Zhang, S., Rutgers, the State University of New Jersey, 2000.
(22) Kluger, R.; Tittmann, K. Chem. Rev. 2008, 108, 1797.
(23) Wille, G.; Meyer, D.; Steinmetz, A.; Hinze, E.; Golbik, R.;
Tittmann, K. Nat. Chem. Biol. 2006, 2, 324.
(24) Tittmann, K.; Wille, G.; Golbik, R.; Weidner, A.; Ghisla, S.;
Hubner, G. Biochemistry 2005, 44, 13291.
(25) Juan, E. C.; Hoque, M. M.; Hossain, M. T.; Yamamoto, T.;
Imamura, S.; Suzuki, K.; Sekiguchi, T.; Takenaka, A. Acta Crystallogr.,
Sect. F: Struct. Biol. Cryst. Commun. 2007, 63, 900.
(26) Frey, P. A. BioFactors 1989, 2, 1.
(27) Zhang, S.; Zhou, L.; Nemeria, N.; Yan, Y.; Zhang, Z.; Zou, Y.;
Jordan, F. Biochemistry 2005, 44, 2237.
(28) Gruys, K. J.; Datta, A.; Frey, P. A. Biochemistry 1989, 28, 9071.
(29) Flournoy, D. S.; Frey, P. A. Biochemistry 1986, 25, 6036.
(30) Pan, K.; Jordan, F. Biochemistry 1998, 37, 1357.
(31) Chiu, C. C.; Chung, A.; Barletta, G.; Jordan, F. J. Am. Chem. Soc.
1996, 118, 11026.
ASSOCIATED CONTENT
■
S
* Supporting Information
Synthesis of [C2,C6′-13C]ThDP and seven figures. This
material is available free of charge via the Internet at http://
pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
frjordan@rutgers.edu
■
(32) Graham, L. D.; Packman, L. C.; Perham, R. N. Biochemistry
1989, 28, 1574.
(33) Jones, D. D.; Perham, R. N. Biochem. J. 2008, 409, 357.
(34) Jones, D. D.; Stott, K. M.; Reche, P. A.; Perham, R. N. J. Mol.
Biol. 2001, 305, 49.
Notes
The authors declare no competing financial interest.
(35) Fries, M.; Stott, K. M.; Reynolds, S.; Perham, R. N. J. Mol. Biol.
2007, 366, 132.
(36) Zhang, S.; Liu, M.; Yan, Y.; Zhang, Z.; Jordan, F. J. Biol. Chem.
2004, 279, 54312.
ACKNOWLEDGMENTS
This work was supported at Rutgers by NIH-GM-050380 (to
F.J.).
■
(37) Cleland, W. W. Biochemistry 1975, 14, 3220.
(38) Fersht, A. Structure and Mechanism in Protein Science: A Guide to
Enzyme Catalysis and Protein Folding; W. H. Freeman and Co.: New
York, 1998; p 123.
REFERENCES
■
(1) Koike, M.; Reed, L. J.; Carroll, W. R. J. Biol. Chem. 1960, 235,
1924.
(39) Jones, D. D.; Stott, K. M.; Howard, M. J.; Perham, R. N.
(2) Stephens, P.; Darlison, M.; Lewis, H.; Guest, J. Eur. J. Biochem.
Biochemistry 2000, 39, 8448.
1983, 133, 155.
(40) Berg, A.; Westphal, A. H.; Bosma, H. J.; de Kok, A. Eur. J.
Biochem. 1998, 252, 45.
(3) Stephens, P.; Darlison, M.; Lewis, H.; Guest, J. Eur. J. Biochem.
1983, 133, 481.
18654
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655

Journal of the American Chemical Society
Article
(41) Huhta, D. W.; Heckenthaler, T.; Alvarez, F. J.; Ermer, J.;
Hubner, G.; Schellenberger, A.; Schowen, R. L. Acta Chem. Scand.
1992, 46, 778.
(42) Burbaum, J. J.; Raines, R. T.; Albery, W. J.; Knowles, J. R.
Biochemistry 1989, 28, 9293.
(43) Perham, R. N. Annu. Rev. Biochem. 2000, 69, 961.
(44) Perham, R. N.; Jones, D. D.; Chauhan, H. J.; Howard, M. J.
Biochem. Soc. Trans. 2002, 30, 47.
(45) Frank, R. A.; Price, A. J.; Northrop, F. D.; Perham, R. N.; Luisi,
B. F. J. Mol. Biol. 2007, 368, 639.
(46) Frank, R. A. W.; Pratar, J. V.; Pei, X. Y.; Perham, R. N.; Luisi, B.
F. Structure 2005, 13, 1119.
(47) Frank, R. A.; Titman, C. M.; Pratap, J. V.; Luisi, B. F.; Perham,
R. N. Science 2004, 306, 872.
(48) Ciszak, E. M.; Korotchkina, L. G.; Dominiak, P. M.; Sidhu, S.;
Patel, M. S. J. Biol. Chem. 2003, 278, 21240.
(49) Aevarsson, A.; Chuang, J. L.; Wynn, R. M.; Turley, S.; Chuang,
D. T.; Hol, W. G. J. Structure 2000, 8, 277.
(50) Nakai, T.; Nakagawa, N.; Maoka, N.; Masui, R.; Kuramitsu, S.;
Kamiya, N. J. Mol. Biol. 2004, 337, 1011.
(51) Wynn, R. M.; Kato, M.; Machius, M.; Chuang, J. L.; Li, J.;
Tomchick, D. R.; Chuang, D. T. Structure 2004, 12, 2185.
(52) Machius, M.; Wynn, R. M.; Chuang, J. L.; Li, J.; Kluger, R.; Yu,
D.; Tomchick, D. R.; Brautigam, C. A.; Chuang, D. T. Structure 2006,
14, 287.
(53) Kato, M.; Wynn, R. M.; Chuang, J. L.; Tso, S. C.; Machius, M.;
Li, J.; Chuang, D. T. Structure 2008, 16, 1849.
(54) Pei, X. Y.; Titman, C. M.; Frank, R. A.; Leeper, F. J.; Luisi, B. F.
Structure 2008, 16, 1860.
(55) Kern, D.; Kern, G.; Neef, H.; Tittmann, K.; Killenberg-Jabs, M.;
Wikner, C.; Schneider, G.; Hubner, G. Science 1997, 275, 67.
18655
dx.doi.org/10.1021/ja3062375 | J. Am. Chem. Soc. 2012, 134, 18644−18655