Báo cáo khoa học: Essential roles of lipoyl domains in the activated function and control of pyruvate dehydrogenase kinases and phosphatase isoform 1 pot

MINIREVIEW
Essential roles of lipoyl domains in the activated function
and control of pyruvate dehydrogenase kinases
and phosphatase isoform 1
Thomas E. Roche, Yasuaki Hiromasa, Ali Turkan, Xiaoming Gong, Tao Peng, Xiaohua Yan,
Shane A. Kasten, Haiying Bao and Jianchun Dong
Department of Biochemistry, Kansas State University, Manhattan, Kansas, USA
Four pyruvate dehydrogenase kinase and two pyruvate
dehydrogenase phosphatase isoforms function in adjusting
the activation state of the pyruvate dehydrogenase complex
(PDC) through determining the fraction of active (non-
phosphorylated) pyruvate dehydrogenase component.
Necessary adaptations of PDC activity with varying meta-
bolic requirements in different tissues and cell types are met
by the selective expression and pronounced variation in the
inherent functional properties and effector sensitivities of
these regulatory enzymes. This review emphasizes how the
foremost changes in the kinase and phosphatase activities
issue from the dynamic, effector–modified interactions
of these regulatory enzymes with the flexibly held outer
domains of the core-forming dihydrolipoyl acetyl transferase
component.
Keywords: pyruvate dehydrogenase complex; PD kinase;
PD phosphatase; dihydrolipoyl acetyltransferase; lipoyl
domain.
Introduction
The mitochondrial pyruvate dehydrogenase complex (PDC)
plays a critical fuel selection role in determining whether
glucose-linked substrates are converted to acetyl-CoA [1–4].
When carbohydrate stores are reduced, mammalian PDC
activity is down-regulated and limits the oxidative utilization

of glucose in most non-neural tissues. Extended starvation
results in PDC activity being profoundly suppressed in most
tissues; operation of the same regulatory control severely
restricts PDC activity in diabetic animals to thereby impede
consumption of abundant glucose. Following the intake of
excess dietary carbohydrate, activation of PDC in fat
synthesizing tissues accelerates fatty acid biosynthesis from
glucose. Adaptable control of PDC activity is required to
satisfy these diverse tasks in the management of fuel
consumption and storage. This is achieved by the tissue-
specific and metabolic state-specific expression and the
discrete regulatory properties of the dedicated kinase and
phosphatase isozymes [1–4]. Four pyruvate dehydrogenase
kinase (PDK) isozymes and two pyruvate dehydrogenase
phosphatase (PDP) isoforms function in governing the
activity state of PDC [5–7]. In combination these carry out a
continuous interconversion cycle that determines the pro-
portion of the pyruvate dehydrogenase (E1) component that
is in the active, nonphosphorylated state.
Among the regulatory enzymes only PDP isoform 1
(PDP1) was purified and its distinct regulatory properties
characterized [8] prior to the relatively recent development
of the capacity to recombinantly express the kinase and
phosphatase isoforms. Although prior studies established a
set of prototypical regulatory responses for kinase function,
these were ascertained from studies on purified complexes
and resolved kinase fractions containing an undefined
mixture of kinase isoforms. PDK isozymes together with
the related branched-chain dehydrogenase kinase constitute
a novel family of serine kinases, unrelated to cytoplasmic

Ser/Thr/Tyr kinases [1,2,5,6,9–12]. Based on the order in
which they were initially cloned, the four PDK isoforms
identified in mammals are designated PDK1, PDK2, PDK3
and PDK4. Though not related to the cytoplasmic Ser/Thr/
Tyr kinases, these PDK isoforms have a 2-domain structure
with a C-terminal domain that is related to another class of
ATP consuming enzymes [9–12] that broadly includes
bacterial histidine kinases. As detailed elsewhere [1],
sequence comparisons of the same PDK isozyme from
different mammals are highly conserved for each of the four
isoforms (> 94% sequence identity for human vs. rat).
Comparison of any combination of the different 45.5–
46 kDa human isoforms reveals that they share
65% ± 4% sequence identity, with only short segments
at the N-terminus that cannot be aligned.
Correspondence to T. E. Roche, Department of Biochemistry,
Willard Hall, Kansas State University, Manhattan, KS 66506,
USA. Fax: +1 785 532 7278, Tel.: + 1 785 532 6116,
E-mail:
Abbreviations: PDC, pyruvate dehydrogenase complex; E1, pyruvate
dehydrogenase component; E2, dihydrolipoyl acetyltransferase
component; L1 domain, NH
2
-lipoyl domain of E2; L2 domain,
interior lipoyl domain of E2; PDK, pyruvate dehydrogenase kinase;
PDP, pyruvate dehydrogenase phosphatase; PDP1c, catalytic subunit
of PDP1; PDP1r, regulatory subunit of PDP1; E3, dihydrolipoyl
dehydrogenase; E3 BP, E3-binding protein; L3, N-terminal lipoyl
domain of E3 BP; GST, glutathione S-transferase.
(Received 16 July 2002, revised 27 December 2002,

accepted 20 January 2003)
Eur. J. Biochem. 270, 1050–1056 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03468.x
The two PDP isoforms have catalytic subunits that are
members of the 2C class of protein phosphatases [7,13].
Both PDP activities require Mg
2+
and are regulated with
regard to their responsiveness to this essential metal [7,8].
Micromolar Ca
2+
greatly stimulates the activity of PDP1.
Polyamines, most effectively spermine, markedly reduce the
K
m
values for Mg
2+
of both PDP isoforms. The K
m
of PDP2
in the absence of polyamines is very high (16 m
M
)and
reduced to 3 m
M
[7]; whereas the K
m
of PDP1 for Mg
2+
is
lowered from 2 m

M
(+Ca
2+
)to0.4 m
M
by spermine [8,14].
A regulatory role for these polyamines effects has not been
definitively established as variation in intramitochondrial
polyamine levels has not been demonstrated under specific
metabolic conditions that meaningfully alter PDC activity.
Elevating spermidine mimicks insulin activation of PDC in
permeabilized adipocytes [4]. PDP2 is expressed in fat
synthesizing tissues [7] and is probably the primary target
of insulin regulation, which enhances PDP activity via a
mechanism that lowers the K
m
for Mg
2+
. Putative mecha-
nisms mediating insulin regulation include allosteric media-
tors [15] and phoshorylation by PKC* [16].
Particularly important is the overexpression of PDK4 [2]
under conditions of starvation, which leads to a need to
conserve carbohydrate. PDK4 expression is increased both
by glucocorticoids and by free fatty acids via the peroxisome
proliferator-activated receptor, and is blocked by an insulin-
activated pathway [2]. Impaired functioning in insulin-
induced down-regulation of PDK4 (due to lack of insulin or
insensitivity to insulin) deleteriously leads to overexpression
of PDK4 and shutting down glucose oxidation in diabetic

animals. In a complementary fashion to regulation of
PDK4 expression, starvation and diabetes reduce the
expression of PDP2 in rat heart and kidney (B. Huang,
P. Wu, K. M. Popov & R. A. Harris, Indiana University
School of Medicine, Indianapolis, IN, USA, personal
communication). The expression of PDP1, the most
abundant PDP in these tissues, is not affected. Re-feeding
restores PDP2 expression.
To conserve carbohydrate reserves, feedback suppression
of the PDC reaction when fatty acids and ketone bodies are
being used as preferred energy sources results from
enhanced kinase activity [1,3]. The resulting elevation of
the intramitochondrial NADH/NAD
+
and acetyl-CoA/
CoA ratios suppresses PDC activity by effectively stimula-
ting kinase activity, particularly PDK2 [17,18], which is
expressed in most tissues [17]. Kinase activity is reduced by
the direct inhibitory effects of ADP and pyruvate and
synergistically inhibited by combination of these effectors
[19]. Again, PDK2 is especially responsive to these inhi-
bitors [18]. Phosphate anion also enhances ADP and
pyruvate inhibition of PDK2.
The focus of this review is on the functional roles of the
dihydrolipoyl acetyltransferase (E2) component in eliciting
the predominant changes in the operation and the effector-
modulation of the PDKs (with emphasis on PDK2 and
PDK3) and PDP1. In all organisms, E2 is recognized for
providing the framework for assembly of the complex and
integrating the sequential reactions of the complex [1,20]. In

mammalian PDC, E2 also transforms kinase and phospha-
tase function and regulation through serving as an anchor-
ing scaffold, an adaptor protein directly abetting efficient
phosphorylation and dephosphorylation, a processing unit
in translating and transmitting effector signals, and in
modifying the sensitivity to alloseteric effectors that directly
bind the regulatory enzymes [1,20]. Pivotal to all these roles
are the dynamic interactions of the regulatory enzymes with
the lipoyl domains of E2 [1,18–23]. Studies using recom-
binantly expressed components have been conducted over
several years in our laboratory, and are supported by a large
number of constructs of the E2 component, particularly
involving modification of the E2 lipoyl domains.
Mammalian PDC-E2 has four globular domains (Fig. 1)
with a sequential (linker region connected) set of 2 lipoyl
domains at its N-terminal end. When expressed by itself, E2
assembles as a 60mer with an inner core formed through the
association of 20 catalytic trimers of the C-terminal domain
at the corners of a pentagonal dodecahedron. Between these
C- and N-terminal domains is a small globular domain,
flanked by linker regions, which binds the E1 component. In
all a-keto acid dehydrogenase complexes, lipoyl domains
populate the surface of the complex and consolidate the
sequential five step reaction sequence by serving as
substrates in the three central reactions and as mobile
carriers of the intermediate forms (oxidized disulfide, 6,8-
dithiol and 8-acetyl) of the lipoyl prosthetic group. The
capacity of the lipoyl domains for traversing efficiently
between the E1, E2 and E3 active sites is advanced by the
high mobility of the extended and rather stiff Ala-Pro rich

linker regions [24]. These lipoyl domain roles and movement
in support of the three complete reactions catalyzed by E1,
E2 and E3 are included in Fig. 3. Apart from E2, the three
domain E3-binding protein (E3BP, Fig. 1) also has an
anchoring C-terminal domain, an E3-binding domain and a
mobile lipoyl domain. Again, the lipoyl domain contributes
as a substrate and intermediate carrier in the PDC reaction
sequence and, to a lesser extent than E2 lipoyl domains, to
regulatory enzyme function (below).
Fig. 1. E2 and E3BP domains and their binding interactions. E2 subunit
domains: L1, N-terminal lipoyl domain; L2, inner lipoyl domain; B, E1
binding domain; I, oligomer-forming, acetyl-transferase-catalyzing
inner domain. E3BP domains: L3, N-terminal lipoyl domain; B, E3
binding domain; I, inner domain which associates with the inner
domain of E2. Dotted connections indicate binding interactions to the
catalytic components (E1 and E3), kinase isoforms (PDK1, PDK2,
PDK3, PDK4) and Ca
2+
-binding of PDP1.
Ó FEBS 2003 Lipoyl domain-aided kinases and phosphatase action (Eur. J. Biochem. 270) 1051
Activated PDK2 and PDK3 function
PDK2 exhibits the full set of the prototypical regulatory
responses of mammalian PDKs described above. The E2
component transforms the efficiency of PDK2 catalysis and
intervenes to create or alter all of these regulatory responses
[18] (X. Yan, H. Bao, S. A. Kasten and T. E. Roche,
unpublished results). PDK2 can phosphorylate free E1 but
E2 enhances the rate of inactivation of PDC by several fold
at low micromolar levels of complex and up to 5 000-fold
with dilute (n

M
) complexes (Y. Hiromasa & T. E. Roche,
unpublished results). This clearly involves PDK2 gaining
efficient access to many E2-bound E1 via agile intercession
of the outer domains of the E2 60mer. A combination of
characterization of complexes with the analytical ultracen-
trifuge and kinase assays using very dilute complexes have
provided important insights (Y. Hiromasa & T. E. Roche,
unpublished work). PDK2 preferentially interacts with the
inner lipoyl domain (L2 domain, Fig. 1) of E2 via an
interaction that requires the lipoyl prosthetic group. Binding
to a free L2 domain is not readily detected but reduction of
the lipoyl group leads to detectable binding. Binding to two
L2 in glutathione S-transferase–L2 (GST–L2) dimer struc-
ture is readily observed (K
d
¼ 4
M
) and this affinity is
increased more than 10-fold upon reduction of the pros-
thetic group (i.e. by GST–L2
red
). Thus, the PDK2 dimer
binds two lipoyl domains and is tightly bound by E2-E1,
particularly when the lipoyl groups are reduced. Interest-
ingly, lipoate reduction stimulates PDK2 activity (regulat-
ory mechanism below). Even with oxidized lipoyl groups,
E2 supports maximal PDK2 activity when complexes
containing < 0.5 PDK2 dimers per complex are diluted
to 30 n

M
indicating that the above affinities do not fully
explain PDK2 function.
Direct binding of the free L2 domain with an oxidized
lipoyl group to PDK3 is much tighter than its binding to
PDK2 and has a potent effect in directly enhancing PDK3
activity [18]. Fifty-fold lower levels of the dimeric GST–L2
domain activate PDK3 (P. Tao, Y. Hiromasa & T. E.
Roche, unpublished results). A portion of the 13-fold
activation of PDK3 by L2 [18] is, in fact, due to
preventing or reversing a decrease in activity of PDK3
undergoing self association in the absence of L2. Binding
of two L2 or one GST–L2 dimer stabilizes PDK3 as a
dimer. Short-term assays, following dilution of the
L2-stabilized PDK3 dimer into assay mixtures established
that excess L2 or GST–L2 still promulgates a several-fold
increased kinase activity, by inducing a more active PDK3
conformation. Beyond this direct allosteric activation, the
E2 60mer further enhances PDK3 activity and, in contrast
to activation by L2, high activity is sustained with very
dilute (< 3 n
M
) complexes.
The bifunctional binding to L2 is likely to underpin the
ability of PDK2 and PDK3 to maintain rapid initial rates
when only a few E1 bound to the E2 60mer or with dilute
complexes (Y. Hiromasa & T. E. Roche, unpublished
observations). To gain efficient access to E2-bound E1,
successive Ôhand-over-handÕ transfer is proposed to proceed
via a kinase dimer being successively bound to two lipoyl

domains, randomly dissociating from one lipoyl domain
and then rebinding to a second mobile lipoyl domain faster
than releasing from the singly held state [21]. Each of the
lipoyl domains of E2 is concentrated within the exterior of
the complex at > 1 m
M
[25,26], so that the rate at which a
singly held dimer associates with a second lipoyl domain
readily exceeds the rate for complete dissociation. This
delivery mechanism may be particularly important for
efficient kinase function within the mitochondrion where
the high protein concentration (> 400 mgÆmL
)1
) limits
diffusion of macromolecules.
We have taken advantage of the direct activation of
PDK3 by L2 to discern the surface structure of the L2
domain that engages in leveraging the conformational
change in PDK3 that elicits this substantial increase in
activity (X. Gong, T. Peng, and T. E. Roche, unpublished
results). Modified L2 was prepared by substituting surface
amino acid residues and by enzymatically inserting cofactor
analogs for the lipoyl prosthetic group. As shown in Fig. 2,
critical residues for PDK3 activation span > 25 A
˚
range on
L2 surface (indicated by * or **, Fig. 2). Most are located
near the lipoylated end of L2 (Leu140, Asp172 and Ala174,
Fig. 2A; Asp197 and Arg196, Fig. 2B). Also critically
important were acidic residues (Glu162 and Glu179)

located toward the other end of the domain (Fig. 2A).
Even at very high levels, the well-folded and fully lipoylated
Glu179fiAla–L2 did not activate PDK3, suggesting that
this residue is particularly important for promoting a
conformational change required for activation.
The full length of the lipoyl-lysine prosthetic group was
absolutely required to in order to promulgate PDK3
activation. L2 with any of several amino acid substitutions
for Lys173 failed to activate PDK3. 8-Thiol-octanoyl-L2
enhanced PDK3 activity beyond the native lipoyl-L2.
Heptanoyl-Lys173-L2 inhibited PDK3 activity and effect-
ively hindered activation by native L2. These results support
the importance of interactions throughout the prosthetic
group and fit the concept that extended reach of the 8-thiol
group upon lipoate reduction contributes to interactions
fostering kinase activation by NADH (see regulation
below). The need for specific structure spread out on the
surface of the L2 domain acting in concert with fully
extended lipoyl lysine prosthetic group clearly implies that
these extended regions work together to convert PDK3 to a
more active conformation.
E2-mediated regulation of kinase activity
As indicated above, feedback suppression of PDC activity,
when fatty acids and ketone bodies are primarily being
consumed, results from kinase activity being greatly
enhanced due to the resulting elevation of NADH/NAD
+
and acetyl-CoA/CoA ratios. The rise in these ratios is sensed
by the rapid and reversible E3 and E2 reactions which act to
increase the proportion of the lipoyl groups of E2 and E3BP

that are reduced and acetylated (Fig. 3) [22,27–29]. Short-
term reduction of the lipoyl group gives rise to an 80%
increase in kinase activity (PDK*), and acetylation stimu-
lates kinase activity up to threefold (PDK**). In vitro,full
stimulation can also be achieved by low levels of pyruvate
reacting through the rate-limiting E1 reaction. Indeed, this
approach provided important insights into the mechanism
of stimulation as blocking E1 catalysis prevents reductive
acetylation and consequently kinase activation. Stimulation
can be observed with peptide substrates and free lipoyl
1052 T. E. Roche et al.(Eur. J. Biochem. 270) Ó FEBS 2003
domains pointing to the importance of direct allosteric
interactions of the reacted lipoyl group with the kinase [22].
8-Thiol-octanoyl-L2 undergoes E2-catalyzed acetylation
(using the inner E2-core stripped of lipoyl domains) and
this acetylation stimulates kinase activity. Thus, the thiol at
the 6-position of the dihydrolipoyl group is not required.
The kinases associated with purified bovine kidney
complex (PDK2 and PDK3 [18]) were shown to respond
to this control mechanism in a remarkably sensitive manner
that was enhanced by ADP inhibition [28]. Half-maximal
stimulation of the activity of bovine kidney kinase is reached
when < 10% of the lipoyl groups in the assembled complex
are acetylated. Near-maximal stimulation of the kinase
activity is attained with an NADH/NAD
+
ratio of 0.1 and
acetyl-CoA/CoA ratio of 0.2. The most responsive human
kinase isoform is PDK2 [17,18]. The greater extended reach
of the reduced and acetylated lipoyl group over that of the

oxidized prosthetic group (Fig. 3) probably bestows the
ability for interactions to induce activating conformations
within the kinase active site. The much stronger binding of
Fig. 2. Surface residues of the L2 domain of E2 that are required for
activating PDK3 and PDP1. Panels A and B show opposing sides of
space-filled models of the human L2 domain. The position of Lys173
(lipoylated lysine) is at the top of the structures; unstructured regions
that are not part of the connecting linker regions are shown at the
N- and C-termini at the bottom of the structures. Lys173 must be
lipoylated for an L2 construct to bind and activate PDK3 or bind and
competitively prevent E2 activation of PDP1. ** or * designate resi-
dues whose substitution removes 80% or 50%, respectively, of L2
activation of PDK3 (X. Gong, T. Peng, and T. E. Roche, unpublished
results). In these panels, ++ or + indicates a mutation that reduces
PDP1 binding to L2 by 75% or 45%, respectively [32].
Fig. 3. Signal mechanism for stimulation of PDK activity by elevation of
the NADH/NAD
+
and acetyl-CoA/CoA ratios. In the PDC reaction
(forward direction proceeds in the counterclockwise direction), the
rapid and reversible E3 and E2 reactions respond to changes in these
products-to-substrate ratios and react to thereby determine the pro-
portion of lipoyl groups that are in the oxidized, reduced or acetylated
forms. When a kinase dimer binds to a lipoyl domain with its lipoyl
group in the oxidized form (bound PDK, top left), E2 facilitates higher
rates of phosphorylation of the E2-bound E1. Binding of a kinase
dimer to a lipoyl domain containing a reduced lipoyl group fosters a
further increase in kinase activity (PDK* state). Binding of a PDK
dimer to a lipoyl domain with its lipoyl group reductively acetylated yet
additionally enhances kinase activity (PDK** state). See the text for

the magnitude of these stimulatory effects on PDK activity.
Ó FEBS 2003 Lipoyl domain-aided kinases and phosphatase action (Eur. J. Biochem. 270) 1053
PDK2 to the L2 domain upon reduction of the lipoyl group
(above) is consistent with additional interactions supporting
preferential binding and eliciting a conformational change
in the PDK structure.
Elevation ADP and phosphate constitute the primary
metabolic indicators for a low energy state. Abundant
pyruvate manifests the availability of glucose-linked sub-
strates for solving this energy demand. Reasonably, these
metabolites act together to impose a marked reduction in
kinase-catalyzed inactivation of PDC activity. Elevation of
K
+
to physiological levels slows kinase catalysis in
conjunction with decreasing the K
i
for ADP. E2 activation
of PDK2 activity transforms PDK2 from being poorly
inhibited by pyruvate or dichloroacetate to being markedly
inhibited [18]. Inhibition results from pyruvate (or dichlo-
roacetate) binding PDK2ÆADP (and PDK2ÆATP) but not
the free PDK2. This results in a synergistic inhibition by
pyruvate and ADP by reducing the rate of ADP dissoci-
ation (X. Yan, H. Bao, S. A. Kasten & T. E. Roche,
unpublished observation). Phosphate inhibition of PDK2 is
also favored and acts together with ADP and pyruvate in
reducing PDK2 activity. Under conditions of physiological
salts, the full set of regulatory properties of E2-activated
PDK2 are consistent with an encompassing mechanism in

which effectors either enhance or reduce a rate-limiting step
of ADP dissociation. Accordingly, dissociation of ADP is
slowed by pyruvate and speeded up upon reductive
acetylation of lipoyl groups [1,18,22] (X. Yan, H. Bao,
S. A. Kasten & T. E. Roche, unpublished results).
As an example of isozyme differences, pyruvate is a very
weak inhibitor of PDK3. However, the combination of
phosphate and ADP synergistically inhibit PDK3 activity.
Human PDK3 activity is enhanced only when inhibited (e.g.
by ADP) or when other conditions reduce its nonstimulated
activity [18]. For instance, PDK3 activity is increased several
fold when the poorly activating L1 domain of E2 provides
the only reactive lipoyl group undergoing reductive acety-
lation. The gain in the capacity for PDK3 activity to be
enhanced by reductive acetylation in the presence of ADP
and phosphate appears to be the primary basis for the
higher fractional stimulation of bovine kidney kinases
observed in the presence of these inhibitors [18].
To express human PDK4 (J. Dong, L. Hu & T. E. Roche,
unpublished results), we first added a Gly-Glu-Glu amino
acid sequence after the C-terminal Val-Ala-Met, as this
hydrophobic sequence triggers degradation when recombi-
nantly expressed in Escherichia coli. Subsequently we
generated unmodified PDK4 in an E. coli strain lacking
the ClpP proteosome. Interestingly the unmodified PDK4
differed from the GEE-modified PDK4 in being stimula-
ted by NADH and acetyl-CoA. This suggests that the
C-terminal region of PDKs is important for the lipoyl-
mediated product stimulation. PDK4 is preferentially
bound by the L1 domain of E2 and L3 domain of E3BP

(Fig. 1) and reduction and acetylation of the lipoyl groups
of these domains enhance PDK4 activity. PDK1 binds to
the L1 and L2 domains of E2 (Fig. 1). PDK1 is particularly
effective in phosphorylating the third phosphorylation site
on E1 [30].
The PDK2 [11] and branched chain dehydrogenase
kinase [12] 3D structures appear to be ideally suited for
bifunctional binding of two lipoyl domains as was previ-
ously proposed based on the studies reviewed above. The
PDK2 subunits have a wedge shape formed from C- and
N-terminal domains. In the dimmer, the combined wedges
form is located in extended apposed positions, with the
extended wedge openings running in opposite direc-
tions. The monomer association involves the ATP-using
C-terminal domains interacting near the base of that side of
the wedge outside; the N-terminal domains form the outside
of the wedges. The wedge pockets have an extended seam
that extends the length of the interface between these
domains with the c-phosphate of bound ATP exposed at
one end. On the reverse side from these pocket openings, the
combined wedges produce a large horseshoe-shaped cavity
with the outer N-terminal ends twisting in opposite
directions. Significant segments of the PDK2 subunit
structure were not resolved in the 3D structure, including
a loop over each active site, a large C-terminal segment, and
a shorter section at the N-terminus [11]. It seems likely that
lipoyl domain binding occurs either in the extended wedge
pockets or within the horseshoe-shaped cavity, aided by
stabilizing interactions involving one or more of the flexible
termini. As noted above, the C-terminal segment may inter-

act with the lipoyl domain or prosthetic group following
reduction and acetylation of the lipoyl group because
modification of the human PDK4 (by adding Gly-Glu-Glu
to the C-terminus) prevented stimulation of PDK4 by
NADH and acetyl-CoA but did not prevent lipoyl domain
binding.
E2 mediated Ca
2+
-activation of PDP1
As intramitochondrial ATP decreases, free Mg
2+
rises as a
secondary signal of the diminished cellular energy state. In
conjunction with exercise, growth, and many other neural
and hormone-initiated cellular transitions that demand
energy, cellular Ca
2+
is elevated in the cytoplasm, leading to
a concomitant rise in intramitochondrial Ca
2+
[4]. PDP1
activity is considerably up-regulated in response to both of
these energy-demanding up shifts in free Mg
2+
and Ca
2+
and facilitates an increase in the proportion of active PDC.
PDP1 is a heterodimer composed of a 52-kDa catalytic
subunit [13], PDP1c, and 96-kDa regulatory subunit [31].
PDP1c alone has a low K

m
of 0.4 m
M
for Mg
2+
[13].
However, in holo-PDP1 the regulatory subunit, PDP1r,
raises the K
m
of the heterodimer for Mg
2+
to 2 m
M
with
Ca
2+
present and 3.5 m
M
in the absence of Ca
2+
[14].
Binding of spermine to the PDP1r subunit returns the K
m
of
holo-PDP1 for Mg
2+
to the lower level (0.5 m
M
). This leads
to a marked increase in PDP1 activity at physiological levels

of intramitochondrial free Mg
2+
(0.5–1.2 m
M
).
E2 activation is facilitated by the Ca
2+
-dependent
binding of PDP1, or PDP1c alone, to the L2 domain of
E2 [23]. Even at saturating Mg
2+
,E2plusCa
2+
accelerates
the rates of dephosphorylation by PDP1 by 10-fold and by
PDP1c by sixfold. Greater-fold increases are facilitated by
E2 when phosphatase activity is assessed with subsaturating
levels of Mg
2+
or when activities are compared with levels
of free vs. E2-bound phosphorylated-E1 set at < 1 l
M
.We
have used a set of over 30 mutant L2 domains and
substitutions within the lipoyl-lysine prosthetic group to
elucidate two major regions of L2 that contribute to binding
PDP1 [32].
1054 T. E. Roche et al.(Eur. J. Biochem. 270) Ó FEBS 2003
The first region includes the lipoyl prosthetic group and
neighboring residues (Fig. 2A, ++ residues). Marked

reductions in binding result from substitution of the
adjacent residues Ala172 and Asp173 as well as Leu140.
Full binding was retained after replacing the lipoyl group
with an octanoyl group, but no activation remained with
nonlipoylated L2 or following any of a series of amino acid
substitutions for lipoylated Lys173. These results indicate
the lipoyl cofactor probably interacts at an extended
hydrophobic pocket in the surface structure of PDP1c. In
contrast to the effectiveness of the octanoyl group, the
dithiolane ring character of the lipoyl prosthetic group
contributes to L2 binding to E1 [33] and, as indicated above,
the 8-thiol is crucial for L2 activating PDK3 [23]. Given that
the L1 domain has identical amino acids in aligned sequence
positions, this region of L2 is not expected to contribute to
the high specificity of PDP1 for binding to L2.
In a second region at the other end of the L2 domain,
mutation of glutamates 162, 179 and 182, and glutamine
181 greatly reduces binding. Indeed, substitution of alanine
or glutamine for Glu182 blocks binding of PDP1 and
PDP1c to the L2 domain. As can be seen in Fig. 2A, a
distinct pocket exists in the center of this set of residues.
Binding of PDP1c-Ca
2+
to this region may be reinforced by
a resulting removal of the mutual repulsion by the three
acidic residues. Only a complete 3D structure can establish
how these proteins interact and elucidate whether residues
such as Glu182 directly participate in forming a tight Ca
2+
-

binding site. The conversion of the Val–Gln residues
connecting Glu179 and Glu182 in L2 to the Ser–Leu
sequence between equivalent acidic residues in L1 markedly
reduced binding of PDP1 to L2 [32]. This bisubstituted L2
had a substantial but lesser effect on the binding of PDP1c.
The dual mutation did not reduce use of L2 in the E1
reaction despite the fact Glu179 was a key specificity residue
for E1 [33]. Five other single site mutants (Ala174fiSer and
Arg196fiGln at the lipoylated end of L2, Asp213fiAsn
and Tyr220fiAla in the C-terminal lobe and the Tyr129fi
Ala in the N-terminal segment) had greater effects on L2
binding to PDP1 than PDP1c, indicating that the PDP1r
subunit of PDP1 promotes a more precise interaction with
L2.
Contrary to expectations for 1 l
M
binding of Ca
2+
,
a characteristic Ca
2+
-binding EF-hand sequence is not
apparent in either PDP1c or L2. Isothermal titration
calorimetry measurements revealed that Ca
2+
does not
bind to either L2 or PDP1c alone (A. Turkan and
T. E. Roche, unpublished results). Ca
2+
mayplayadirect

bridging role in the PDP1c–L2 interaction or enhance
binding by a capture mechanism. A conformational change
in the PDP1c could create such a high-affinity Ca
2+
-binding
site to stabilize the nascent interaction between the protein
components. Because the highly activating binding of PDP1
to the L2 domain of E2 requires, as described above, both
domain-aided hydrophobic interaction by the exterior lipoyl
group at one end of L2 and electrostatic interactions at the
opposite end of the L2 domain [32], it seems likely that these
essential regions of L2 act in concert. An appealing prospect
is that this extensive interaction surface of L2 supports a
conformation transition that fosters and stabilizes a tight
Ca
2+
-binding site in PDP1c. Further insight into the
constitution of the Ca
2+
-dependent complexes formed
between L2 and PDP1 or PDP1c will require detailed
structures that will probably depend on crystallizing these
complexes.
Conclusion
Among four PDK isoforms and two PDP isoforms, we have
focused on PDK2, PDK3, PDP1, and its catalytic subunit,
PDP1c, to illustrate how the consequential variation in the
functional capacity and the responsiveness of these regula-
tory enzymes is dependent upon differences in their effector–
modified interactions with the flexibly held outer domains of

the E2 assemblage. Novel mechanisms have been uncovered
whereby E2 greatly enhances kinase and phosphatase
catalysis by enormously increasing access to their E1
substrates and by direct allosteric activation (particularly
PDK3), mediates kinase stimulation in feedback effector
control by NADH and acetyl-CoA, facilitates Ca
2+
-activa-
tion of PDP1, and modifies the allosteric control by effectors
that bind directly to the regulatory enzymes (e.g. enhances
pyruvate inhibition PDK2 and alters the Mg
2+
requirement
of PDP1). Requisite interactions of PDK3 and PDP1 with
L2, the inner lipoyl domain of E2, are shown to engage
extensive regions of the surface of L2 and to require the lipoyl
prosthetic group and, in thecase of the PDKs, to be markedly
modified by reaction of this prosthetic group.
Acknowledgements
This work was supported by the National Institutes of Health Grant
DK18320 and by the Kansas Agriculture Experiment Station –
contribution 03-27-J.
References
1. Roche, T.E., Baker, J., Yan, X., Hiromasa, Y., Gong, X., Peng,
T., Dong, J., Turkan, A. & Kasten, S.A. (2001) Distinct regulatory
properties of pyruvate dehydrogenase kinase and phosphatase
isoforms. Prog. Nucleic Acid Res. Mol. Biol. 70, 33–75.
2. Harris, R.A., Huang, B. & Wu, P. (2001) Control of pyruvate
dehydrogenase kinase gene expression. Adv. Enzyme Regul. 41,
269–288.

3. Randle, P.J. & Priestman, D.A. (1996) Shorter term and longer
term regulation of pyruvate dehydrogenase kinases. In Alpha-Keto
Acid Dehydrogenase Complexes (M. S. Patel, T. E. Roche &
R. A. Harris, eds), pp. 151–161. Birkhauser Verlag, Basel,
Switzerland.
4. Denton, R.M., McCormack, J.G., Rutter, G.A., Burnett, P.,
Edgell, N.J., Moule, S.K. & Diggle, T.A. (1996) The hormonal
regulation of pyruvate dehydrogenase complex. Adv. Enzyme
Regul. 36, 183–198.
5. Gudi, R., Bowker-Kinley, M.M., Kedishvili, N.Y., Zhao, Y. &
Popov, K.M. (1995) Diversity of the pyruvate dehydrogenase
kinase gene family in humans. J. Biol. Chem. 270, 28989–28994.
6. Rowles, J., Scherer, S.W., Xi, T., Majer, M., Nickle, D.C.,
Rommens, J.M., Popov, K.M., Harris, R.A., Riebow, N.L., Xia,
J., Tsui, L C., Bogardus, C. & Prochazka, M. (1996) Cloning and
characterization of PDK4 on 7q21.3 encoding a fourth pyruvate
dehydrogenase kinase isozyme in human. J. Biol. Chem. 271,
22376–22382.
7. Huang, B., Gudi, R., Wu, P., Harris, R.A., Hamilton, J. & Popov,
K.M. (1998) Isozymes of pyruvate dehydrogenase phosphatase.
DNA-derived amino acid sequences, expression, and regulation.
J. Biol. Chem. 273, 17680–17688.
Ó FEBS 2003 Lipoyl domain-aided kinases and phosphatase action (Eur. J. Biochem. 270) 1055
8. Reed, L.J. & Damuni, Z. (1987) Mitochondrial protein phos-
phatases. Adv. Prot. Phosphatases 4, 59–76.
9. Bower-Kinley, M. & Popov, K.M. (1999) Evidence that pyruvate
dehydrogenase kinase belongs to the ATPase/kinase superfamily.
Biochem. J. 344, 47–53.
10. Wynn, R.M., Chaung, J.L., Cote, C.D. & Chuang, D.T. (2000)
Tetrameric assembly and conservation in the ATP-binding

domain of rat branched-chain a-ketoacid dehydrogenase kinase.
J. Biol. Chem. 275, 30512–30519.
11. Steussy, C.N., Popov, K.M., Bowker-Kinley, M.M., Sloan, R.B.,
Harris, R.A. & Hamilton, J.A. (2001) Structure of pyruvate
dehydrogenase kinase. Novel folding pattern for a serine protein
kinase. J. Biol. Chem. 276, 37443–37450.
12. Machius, M., Chaung, J.L., Wynn, R.M., Tomchick, D.R. &
Chuang, D.T. (2001) Structure of rat BCKD kinase: Nucleotide-
induced domain communication in a mitochondrial protein
kinase. Proc. Natl Acad. Sci. USA 98, 11218–11223.
13. Lawson, J.E., Niu, X D., Browning, K.S., Trong, H.L., Yan, J. &
Reed, L.J. (1993) Molecular cloning and expression of the catalytic
subunit of bovine pyruvate dehydrogenase phosphatase and
sequence similarity with protein phosphatase 2C. Biochemistry 32,
8987–8993.
14. Yan, J., Lawson, J.E. & Reed, L.J. (1996) Role of the regulatory
subunit of bovine pyruvate dehydrogenase phosphatase. Proc.
NatlAcad.Sci.USA93, 4953–4956.
15. Larner, J., Huang, L.C., Suzuki, S., Tang, G., Zhang, C.,
Schwartz, C.F.W., Romero, G., Luttrell, L. & Kennington, A.S.
(1989) Insulin mediators and control of the pyruvate dehydro-
genase complex. Ann. N.Y. Acad. Sci. 573, 297–305.
16. Caruso, M., Maitan, M.A., Bifulco, G., Miele, C., Vigliotta, G.,
Oriente,F.,Formisano,P.&Beguinot,F.(2001)Activationand
mitochondrial translocation of protein kinase C delta are neces-
sary for insulin stimulation of pyruvate dehydrogenase complex
activity in muscle and liver cells. J. Biol. Chem. 276, 45088–45097.
17. Bowker-Kinley, M.M., Davis, W.I., Wu, P., Harris, R.A. &
Popov, K.M. (1998) Evidence for existence of tissue-specific reg-
ulation of the mammalian pyruvate dehydrogenase complex.

Biochem. J. 329, 191–196.
18. Baker,J.C.,Yan,X.,Peng,T.,Kasten,S.A.&Roche,T.E.(2000)
Marked differences between two isoforms of human pyruvate
dehydrogenase kinase. J. Biol. Chem. 275, 15773–15781.
19. Pratt, M.L. & Roche, T.E. (1979) Mechanism of pyruvate
inhibition of kidney pyruvate dehydrogenase
a
kinase and
synergistic inhibition by pyruvate and ADP. J. Biol. Chem. 254,
7191–7196.
20. Roche, T.E. & Cox, D.J. (1996) Multifunctional 2-oxo acid
dehydrogenase complexes. In Channeling in Intermediary Meta-
bolism (L. Agius & H. S. A. Sherratt, eds), pp. 115–132. Portland
Press Ltd., London, UK.
21. Liu, S., Baker, J.C. & Roche, T.E. (1995) Binding of the pyruvate
dehydrogenase kinase to recombinant constructs containing the
inner lipoyl domain of the dihydrolipoyl acetyltransferase com-
ponent. J. Biol. Chem. 270, 793–800.
22. Ravindran, S., Radke, G.A. & Roche, T.E. (1996) Lipoyl domain-
based mechanism for integrated feedback control of pyruvate
dehydrogenase complex by enhancement of pyruvate dehydro-
genase kinase activity. J. Biol. Chem. 271, 653–662.
23.Chen,G.,Wang,L.,Liu,S.,Chang,C.&Roche,T.E.(1996)
Activated function of the pyruvate dehydrogenase phosphatase
through Ca
2+
-facilitated binding to the inner lipoyl domain of
the dihydrolipoyl acetyltransferase. J. Biol. Chem. 271, 28064–
28070.
24. Perham, R.N. (2000) Swinging arms and swinging domains in

multifunctional enzymes: catalytic machines for multistep reac-
tions. Ann. Rev. Biochem. 69, 961–1004.
25.Wagenknecht,T.,Grassucci,R.,Radke,G.A.&Roche,T.E.
(1991) Cryoelectron microscopy of mammalian pyruvate
dehydrogenase complex. J. Biol. Chem. 266, 24650–24656.
26. Roche, T.E., Powers-Greenwood, S.L., Shi, W., Zhang, W., Ren,
S.Z., Roche, E.D., Cox, D.J. & Sorensen, C.M. (1993) Sizing of
the bovine pyruvate dehydrogenase complex and the dihydro-
lipoyl transacetylase core by quasielastic light scattering.
Biochemistry 32, 5629–5637.
27. Cate, R.L. & Roche, T.E. (1978) A unifying mechanism for sti-
mulation of mammalian pyruvate dehydrogenase kinase activity
by NADH, dihydrolipoamide, acetyl-CoA, or pyruvate. J. Biol.
Chem. 253, 496–503.
28. Cate, R.L. & Roche, T.E. (1979) Function and regulation of
mammalian pyruvate dehydrogenase complex: acetylation, inter-
lipoyl acetyl transfer, and migration of the pyruvate dehydro-
genase component. J. Biol. Chem. 254, 1659–1665.
29. Yang, D., Gong, X., Yakhnin, A. & Roche, T.E. (1998)
Requirements for the adaptor protein role of dihydrolipoyl
acetyltransferase in the upregulated function of the pyruvate
dehydrogenase kinase and pyruvate dehydrogenase phosphatase.
J. Biol. Chem. 273, 14130–14137.
30. Korotchkina, L.G. & Patel, M.S. (2001) Site specificity of four
pyruvate dehydrogenase kinase isozymes toward the three phos-
phorylation sites of human pyruvate dehydrogenase. J. Biol.
Chem. 279, 37223–37229.
31. Lawson,J.E.,Park,S.H.,Mattison,A.R.,Yan,J.&Reed,L.J.
(1997) Cloning, expression and properties of the regulatory sub-
unit of bovine pyruvate dehydrogenase phosphatase. J. Biol.

Chem. 272, 31625–31629.
32. Turkan, A., Gong, X., Peng, T. & Roche, T.E. (2002) Struc-
tural requirements within the lipoyl domain for the Ca
2+
-dependent binding and activation of pyruvate dehydrogenase
phosphatase isoform 1 or its catalytic subunit. J. Biol. Chem. 277,
14976–14985.
33. Gong, X., Peng, T., Yakhnin, A., Zolkiewski, M., Quinn, J.,
Yeaman, S.J. & Roche, T.E. (2000) Specificity determinants for
the pyruvate dehydrogenase component reaction mapped with
mutated and prosthetic group modified lipoyl domains. J. Biol.
Chem. 275, 13645–13653.
1056 T. E. Roche et al.(Eur. J. Biochem. 270) Ó FEBS 2003