A dynamic analysis of the rotation mechanism for conformational change in F(1)-ATPase.

Abstract

Molecular dynamics trajectories for the bovine mitochondrial F(1)-ATPase are used to demonstrate the motions and interactions that take place during the elementary (120 degrees rotation) step of the gamma subunit. The results show how rotation of the gamma subunit induces the observed structural changes in the catalytic beta subunits. Both steric and electrostatic interactions contribute. An "ionic track" of Arg and Lys residues on the protruding portion of the gamma subunit plays a role in guiding the motions of the beta subunits. Experimental data for mutants of the DELSEED motif and the hinge region are interpreted on the basis of the molecular dynamics results. The trajectory provides a unified dynamic description of the coupled subunit motions involved in the 120 degrees rotation cycles of F(1)-ATPase.

Full text

Page 1

Structure, Vol. 10, 921–931, July, 2002, 2002 Elsevier Science Ltd. All rights reserved.PII S0969-2126(02)00789-X
A Dynamic Analysis of the Rotation Mechanism
for Conformational Change in F1-ATPase
of the ATP synthesis in living systems [1–3]. It is a large
multisubunit complex consisting of a proton-translocat-
ing membrane domain, Fo, attached via central and pe-
ripheral stalks to the catalytic domain F1, a spherical
globular structure outside the membrane. The F1do-
main,F1-ATPase,ismadeofthree?andthree?subunits
arrangedinalternationaroundacentral?-helicalcoiled-
coil structure that is a continuation of the central stalk
[4]. The foot of the central stalk makes extensive con-
tacts with a ring [5] of c subunits in the membrane do-
main [5]. In yeast mitochondria ten c subunits form the
ring, and there is evidence for stoichiometries in the
range from 10 to 14 in other species [6–8]. The central
stalk and the associated c ring are believed to rotate
as an ensemble relative to the rest of the enzyme, the
rotation being generatedby the transmembrane proton-
motive force via photosynthesis or respiration. The cen-
tral stalk consists of the ? subunit with the associated
? and ? subunits. The ?-helical coiled-coil domain of
the ? subunit is asymmetric, and the rotation of this
asymmetricalstructureappearstomodulatethebinding
affinities of the catalytic ? subunits for substrate and
products. Each of them, in turn, goes through three
states known as “open” (or empty), “loose,” and “tight”,
in accord with the “binding change” mechanism of ATP
synthesis [9–10]. The open subunit (which corresponds
to ?Ein the X-ray structure of Abrahams et al. [4]) has
a low affinity for either substrate or product. According
to this mechanism, during the synthesis cycle of F1Fo-
ATP synthase, a rotation of 120? converts an open site,
with low affinity for substrate or product, into a loose
substrate binding state. A further 120? rotation converts
the loose state to the tight state, and the product ATP
is formed. Another 120? rotation converts the tight state
back to the open state, releasing the ATP product. In
the X-ray structure of Abrahams et al. [4], the ?Esubunit
(no bound nucleotide) was identified as representing
the open state, the ?TPsubunit (AMP-PNP bound) was
identified as representing the loose state, and the ?DP
subunit (ADP bound) was identified as representing the
tight state. A more detailed discussion of a suggested
assignment ofthe observedconformations ofthe ?sub-
units to the different states of the binding change mech-
anism is given in Menz et al. [11].
The F1domain can be separated from the membrane
domain,anditretainstheabilitytohydrolyzeATP.Hydroly-
sis of ATP leads to the rotation of the central stalk, though
the detailed mechanism is not understood. By attaching
an actin filament or a bead to the exposed foot of the
central stalk, the rotation has been visualized in a micro-
scope [12]. It turns counterclockwise (as viewed from the
membrane) in 120? steps. During the ATP synthesis cycle,
which we study here, the rotation of the central stalk is
presumed to be in the opposite sense.
Comparisons of the subunits within a given structure
and in different structures have provided important clues
Jianpeng Ma,1,2,3Terence C. Flynn,1,2Qiang Cui,3,4
Andrew G.W. Leslie,5John E. Walker,6
and Martin Karplus3,7,8
1Department of Biochemistry and Molecular Biology
Baylor College of Medicine
One Baylor Plaza, BCM-125
Houston, Texas 77030
2Department of Bioengineering
Rice University
6100 Main, MS-142
Houston, Texas 77005
3Department of Chemistry and Chemical Biology
Harvard University
12 Oxford Street
Cambridge, Massachusetts 02138
4Department of Chemistry
University of Wisconsin, Madison
1101 University Avenue
Madison, Wisconsin 53706
5Medical Research Council Laboratory of Molecular
Biology
Hills Road
Cambridge CB2 2QH
United Kingdom
6Medical Research Council Dunn Human Nutrition
Unit
Hills Road
Cambridge CB2 2XY
United Kingdom
7Laboratoire de Chimie Biophysique, ISIS
Universite ´ Louis Pasteur
67000 Strasbourg
France
Summary
Molecular dynamics trajectories for the bovine mito-
chondrial F1-ATPase are used to demonstrate the mo-
tions and interactions that take place during the ele-
mentary (120? rotation) step of the ? subunit. The
results show how rotation of the ? subunit induces
theobservedstructuralchangesinthecatalytic?sub-
units. Both steric and electrostatic interactions con-
tribute. An “ionic track” of Arg and Lys residues on
the protruding portion of the ? subunit plays a role in
guiding the motions of the ? subunits. Experimental
data for mutants of the DELSEED motif and the hinge
region are interpreted on the basis of the molecular
dynamics results. The trajectory provides a unified
dynamic description of the coupled subunit motions
involved in the 120? rotation cycles of F1-ATPase.
Introduction
The enzyme F1Fo-ATP synthase, found in eubacteria,
chloroplasts, and mitochondria, is responsible for most
Key words: F1-ATPase; molecular dynamics simulation; ATP synthe-
sis; molecular motors; rotatary catalysis; targeted molecular dynamics
8Correspondence: marci@tammy.harvard.edu

Page 2

Structure
922
Figure 1. Views of the F1-ATPase Structure
(A) Ribbonstructure ofF1-ATP synthaseshowing ?3?3???based on thestructures ofAbrahams etal. [4]and Gibbons etal. [15](see Experimental
Procedures). ? subunits, red; ? subunits, yellow; ? subunit, purple; ? subunit, green; ? subunit, light yellow.
(B) View showing ?Eand ?TP, plus ?, ?, and ?, from same structures as in (A) with labels, including ? bulge, ? groove, and ? protrusion.
(C) Detail from Figure 1B showing central portion of ?TPsubunit with coiled coil of ? subunit (purple) and illustrating some important structural
elements mentioned in the text. The central domain is shown in yellow, with the ? strands 3 and 7 and helix B highlighted in green and the
hinge region plus the flexible loop made up of residues 119–134 in red. A portion of the C-terminal domain is shown in gold, with the Phe418
and 424 loop region in red.
(D) Stereo view showing ?TPin red and ?TPin green, with the N-terminal domain and the top of the central domain superposed. See text for
subunit labels.
concerning the rotational mechanism of F1-ATPase. How-
ever, a direct study of the transition at an atomic level
of detail is essential for determining the dynamics of the
structural changes and the relative order in which they
occur. The analysis reported here, which is based pri-
marily on simulations with the targeted molecular dy-
namics (TMD) method [13], is an important step toward
this goal. We supplement the TMD results with those
based onthe biasedmolecular dynamics(BMD) method
[14]. The structural model we use for the simulation was
obtained (see Experimental Procedures) by combining
the original X-ray structure of F1-ATPase [4] with a more
recent structure that has resolved most of the ?, ?, and
? subunits [15]. The simulations correspond to a single
120? rotation step in the direction of ATP synthesis, with
the?subunitrotatinginaclockwisedirection,asviewed
from the bottom of Figure 1A. The ? and ? subunits do
not rotate but undergo conformational changes based
on the endstates inferred from the X-ray structure of
Abrahamsetal.[4].AsdescribedinExperimentalProce-
dures, the TMD simulation applies constraints to the
entire system to find a low energy pathway connecting
the two endstates. By constrast, in the BMD simulation
(see Experimental Procedures), a biasing force is ap-

Page 3

Dynamic Analysis of Rotation Mechanism of F1-ATPase
923
plied only to the ? subunit to determine directly the
response of the ? and ? subunits to the rotation of the
? subunit; it provides some important supplementary
results, as described below, but it does not yield the
full transition path obtained with the TMD method over
the nanosecond timescale of the trajectory.
The trajectory has the spatial and temporal resolution
necessary to reveal the structural and energetic origins
of the conformational changes that occur during the
120? rotation step. Specifically, the results demonstrate
how the rotation of the ? subunit induces the major
changes observed in the structure of the ? subunits.
The lowerpart of the? subunit, including the?? domain,
which protrudes from the ?3?3complex (it is referred to
as the “? protrusion” below), is shown to play an essen-
tialroleinthedynamicsofthetransition.Some ?subunit
motions occur during the transition, although all three
are very similar in the X-ray structure. Certain side chain
interactions, such as those involving the ionic track of
the ? subunit, are shown to be important in guiding the
transition. The closing motion of the ?Esubunit, with
ADPbound,involvesprimarilyelectrostaticinteractions,
while the outward movement of the ?TPsubunit, with no
ligand, is due primarily to steric interactions with the ?
subunit. Main chain polar interactions of the DELSEED
motif supplement those from the negatively charged
side chains and explain the fact that mutation of the
latter does not inactivate the rotational mechanism. A
more detailed description of the motion in the hinge
region than that inferred from the X-ray structure is ob-
tained.Themoleculardynamicstrajectoryreportedhere
provides a unified description of the subunit motions of
F1-ATPase in the 120? rotation cycle that forms the basis
for its function.
N-terminal domains of the six subunits form a ring with
roughly 6-fold symmetry that is believed to play a major
role in the stability of the ?3?3complex and undergoes
small changes during the transition. The conformations
of the central domains and the C-terminal domains of
?TPand ?DP, which have AMP-PNP and ADP bound, re-
spectively, are very similar to each other (for notation,
see Figure 1B and its legend). The empty ?Esubunit has
a very different conformation, although the overall fold
is similar to that of the other subunits. The nucleotide
binding site in the central domain of ?Eis open, when
compared with ?TP, due to a conformational change. It
involves the separation of the C-terminal portions of ?
strands 7 and 3, movement (and increased disorder) of
the loop containing Phe418 and Phe424, and a global
increase in the distance between the ? and ? subunits
in the vicinity of the binding site, as described and illus-
trated in Abrahams et al. [4]; we focus here on the sepa-
ration of ?strands 7 and 3 as amarker for the nucleotide
binding site opening in what follows (see Figure 1C). In
?E, there is also a displacement of helix B of the central
domain and a large downward rotation of the C-terminal
domain when compared with the corresponding do-
mains of ?TPand ?DP. The three ? subunits, which bind
AMP-PNP, have a closed nucleotide binding site and
N-terminal and central domains that are very similar to
those of ?TPand ?DP. However, the C-terminal domains
of the ? subunits are significantly different from those
of ?TPand ?DP(Figure 1D); this is found to be important
intherotationalmotion.The?subunitispositionedsuch
that its “bulge,” formed by the N-terminal helix above
theprotrusion,facestheC-terminaldomainoftheempty
?Esubunit and its “groove,” formed by the other side of
the N-terminal and C-terminal coiled-coil helices, faces
the C-terminal domains of the ?DPand ?TPsubunits (Fig-
ure 1B). The helix-turn-helix (hth) motifs (Figure 1B) of
the ? subunit C-terminal domains play an essential role
in the interaction with the ? subunit. In the crystal struc-
ture, those of ?DPand ?TPare in van der Waals contact
with each other within the ? groove, while that of the ?E
subunit is pushed outward by the ? bulge.
? Subunit Motion
Since the conformational changes of the ? subunits in
the ATP synthesis cycle are “driven” by the ? subunit,
we first describe its motions. We use the fractional time
(in percent) along the trajectory to order the events that
occur during the transition. The calculation time (be-
tween 250 ps and 1 ns) is much shorter than the actual
timescale (on the order of 250 ?s for a 120? rotation in
hydrolysis [16]), but the correspondence of the results
obtained with simulations of different lengths (see Ex-
perimental Procedures) provides support for identifying
the dynamics that take place during the trajectory with
the physical motions. The rotation of the ? protrusion is
rather uniform up to about 90? (60% of the trajectory);
the remainder of the rotation occurs more slowly and is
essentially complete at 90% of the trajectory, with only
small adjustments after that. There is little “trembling”
of the axis during the transition, indicating that the de-
scription of the ? subunit motion as a rotation is mean-
ingful. Although the motions of the portions of the N- and
C-terminal helices within the ? protrusion are essentially
synchronized with it, the rotation of the coiled-coil por-
Results
Coupled Subunit Motions in the 120?
Synthesis Step
Overall Structure
The overall structure of the bovine mitochondrial F1-
ATPase is available from Abrahams et al. ([4]; see Figure
1A). It consists of nine subunits referred to as ?3?3???,
has an approximately spherical domain about 100 A˚
across and 80 A˚high, with a stalk that protrudes by
about 45 A˚, and interacts with the membrane-bound Fo
portion of the F1Fo-ATP synthase complex. The three ?
and three ? subunits that form the spherical domain
are associated with each other like the segments of an
orange.The?subunitconsists ofaleft-handed?-helical
coiled coil formed by a shorter N-terminal helix and a
longer C-terminal helix, which occupy a central hole in
the?3?3complex,andamoreglobularportionthatforms
the ? protrusion, which includes the lower part of the
coiled coil. The ? and ? subunits are associated with the
? protrusion but are not in contact with the ?3?3portion
of the complex and do not play an important role in
the rotation studied here, though they are likely to be
involved in coupling to the Focomplex. The ? and ?
subunitshavesimilarfolds,andeachonecanbedivided
into an N-terminal ? barrel domain, a central nucleotide
binding domain, and a C-terminal helical domain. The

Page 4

Structure
924
Figure 2. Structures from the Molecular Dy-
namics Trajectory Showing ?E, ?TP, ?, ?, and
? in Same Colors as in Figure 1A
Intitial structure (A); 150 ps (B); 300 ps (C);
350 ps (D); 425 ps (E); final structure (F).
tion that extends above the ? protrusion lags behind
somewhat. It has rotated by about 80? at 80% of the
trajectory, and the top part of the C-terminal helix com-
pletes its rotation at the end of the trajectory. The ?
subunit rotation, as well as the conformational change
of?TPand?Eduringthetrajectory,arevisualizedinFigure
2. During ATP synthesis, the rotation of the ? subunit
is driven by the translocation of protons through the
transmembrane segment, Fo. The observed lag and the
distortionofthecoiledcoilreflectstheenergeticbarriers
that must be overcome in inducing the conformation
change of the ? subunits during the synthesis cycle.
That the coiled coil is flexible has been suggested from
a comparison of X-ray structures [11]. A normal mode
analysisindicatesthatthelowestmodesofthe?subunit
involve a twisting/untwisting of the coiled coil, with the
? protrusion undergoing a small rocking motion (data
not shown).
?3?3Subunit Motions
At the start of the trajectory (see Figure 2), the ? bulge
of the N-terminal helix is oriented directly toward ?E;
after the 120? rotation, the ? bulge is oriented toward

Page 5

Dynamic Analysis of Rotation Mechanism of F1-ATPase
925
Figure 3. Space-Filling View of Parts of ?TPin Yellow and the Coiled Coil of the ? Subunit, with the N-Terminal Helix and Its ? Bulge in Magenta
and the C-Terminal Helix and its ? Bend in Dark Purple
(A) Intermediate structure corresponding to Figure 2D showing the hth motif of ?TPbeing “pushed out” by the ? bend of the C-terminal helix.
(B) Final structure showing ?TPin contact with the ? bulge of the N-terminal helix; see text.
?TP. The downward and outward displacement of the
hth motif of ?TP, which is located in the ? groove at
the beginning of the trajectory, is initiated by the steric
repulsion between the hth motif and the “? bend” (i.e.,
thecurvedC-terminalhelixofthe?subunit)inthemiddle
ofthetrajectory(50%),whenthe?protrusionhasrotated
by about 70?, and is completed near the end of the
rotation cycle (90%), when the hth motif is in contact
with the ? bulge (Figure 3). Throughout the transition,
the hth motif is guided by its interactions with the ionic
track (see below). The motion of the hth motif leads to
the displacement of the entire C-terminal domain of ?TP
and is coupled to the conformational change in the cen-
tral domain; e.g., the breaking of the hydrogen bonds
between ? strands 7 and 3, which opens the nucleotide
binding site. The lower portion of the central domain
and the C-terminal domain move approximately as a
unit, although there are also significant internal motions
during the transition. As pointed out by Masaike et al.
[17], the largest main chain dihedral angle changes in-
volve three “hinge” residues, His177, Gly178, and
Gly179. However, there are many smaller changes in the
region around the hinge, indicating that the opening
transition is more like a relatively localized elastic body
motion. Examples are the deformation of the B helix
during the dynamics, which results in an extra turn in
?Erelative to ?TP, and the rearrangements in a loop (resi-
dues 119–134) connecting the two parts of the central
domain; the structures sampled during the dynamics
involve motions not evident from an interpolation be-
tween the end structures. The result of Masaike et al.
[17], that mutating two hinge residues, Gly178 and
Gly179, to Ala had only a small effect on the hydrolysis
reaction, indicates the complexity of the motion. Only
when His177 was also mutated to Ala was essentially
complete inhibition achieved. The latter mutation is not
likelytocontributestericallytopreventingtheconforma-
tion change. As can be seen from the crystal structure,
His177 in the closed state (?TP) forms a strong hydrogen
bond with the main chain carbonyl of Tyr180 and a
weaker one with Asp250. During the dynamics, His177
moves outward, and it points away from these residues
in the open state. Since Tyr180 is at the other end of
theturn formedbyresidues177–180, theHis177,Tyr180
interaction is expected to play a significant role in stabi-
lizing the closed form.
The inward motion of ?Einduced by the ? subunit
rotation is somewhat more complex than the outward
motion of ?TPjust described. The initial position of ?Eis
constrained by the steric (van der Waals) interactions
between its hth motif and the ? bulge. As the ? subunit
rotates and the ? bulge passes by the hth motif (near
40%), the latter is distorted by it. The hth motif then
assumes a less distorted structure and starts to be
pulled into the ? groove, mainly by electrostatic interac-
tions (see below). This causes an upward motion of the
rest of the C-terminal domain and leads to the closing
of the nucleotide binding site with restructuring of the
? sheet by hydrogen bond formation between strands
3 and 7 in the central domain; this occurs slightly later
(60%). As was found for ?TP, the relative motions of ?E
and the ? subunit are guided by the ionic track on the
protrusion of the latter (see below). The motion of ?E
involves both repulsive and attractive interactions with
the coiled coil of the ? subunit; i.e., steric repulsion with
the ? subunit first prevents entrance into the groove,
but, once the ? bulge has passed, attractive ionic inter-
actions pull the hth motif inward in the simulation. This
occurs spontaneously,as shown by theBMD simulation
(see below). By the choice of the occupation of the
? subunit nucleotide binding sites, in accord with the
endstate (i.e., that which occurs after the 120? rotation),
the contributions of the ligands to the conformation
change are included implicitly, i.e., the empty ?TP-site
opens, while the ADP-containing ?E-site is induced to
close.
Previously, it has not been possible to infer anything
concerning the motions of the ? subunits because they
are very similar in the crystal structures (e.g., Abrahams
et al. [4]). They remain in the closed form, in accord with