IOP PUBLISHING JOURNAL OF PHYSICS: CONDENSED MATTER
J. Phys.: Condens. Matter 23 (2011) 234102 (9pp) doi:10.1088/0953-8984/23/23/234102
Molecular basis of proton uptake in single
and double mutants of cytochrome c
oxidase
Rowan M Henry1,2, David Caplan1,2, Elisa Fadda3 and
Re´gis Pome`s1,2,4
1 Molecular Structure and Function, Hospital for Sick Children, Toronto, ON, M5G 1X8,
Canada
2 Department of Biochemistry, University of Toronto, Toronto, ON, M5S 1A8, Canada
3 Department of Chemistry, University of Galway, Ireland
E-mail: pomes@sickkids.ca
Received 10 December 2010, in final form 15 March 2011
Published 25 May 2011
Online at stacks.iop.org/JPhysCM/23/234102
Abstract
Cytochrome c oxidase, the terminal enzyme of the respiratory chain, utilizes the reduction of
dioxygen into water to pump protons across the mitochondrial inner membrane. The principal
pathway of proton uptake into the enzyme, the D channel, is a 2.5 nm long channel-like cavity
named after a conserved, negatively charged aspartic acid (D) residue thought to help recruiting
protons to its entrance (D132 in the first subunit of the S. sphaeroides enzyme). The
single-point mutation of D132 to asparagine (N), a neutral residue, abolishes enzyme activity.
Conversely, replacing conserved N139, one-third into the D channel, by D, induces a decoupled
phenotype, whereby oxygen reduction proceeds but not proton pumping. Intriguingly, the
double mutant D132N/N139D, which conserves the charge of the D channel, restores the
wild-type phenotype. We use molecular dynamics simulations and electrostatic calculations to
examine the structural and physical basis for the coupling of proton pumping and oxygen
chemistry in single and double N139D mutants. The potential of mean force for the
conformational isomerization of N139 and N139D side chains reveals the presence of three
rotamers, one of which faces the channel entrance. This out-facing conformer is metastable in
the wild-type and in the N139D single mutant, but predominant in the double mutant thanks to
the loss of electrostatic repulsion with the carboxylate group of D132. The effects of mutations
and conformational isomerization on the pKa of E286, an essential proton-shuttling residue
located at the top of the D channel, are shown to be consistent with the electrostatic control of
proton pumping proposed recently (Fadda et al 2008 Biochim. Biophys. Acta 1777 277–84).
Taken together, these results suggest that preserving the spatial distribution of charges at the
entrance of the D channel is necessary to guarantee both the uptake and the relay of protons to
the active site of the enzyme. These findings highlight the interplay of long-range electrostatic
forces and local structural fluctuations in the control of proton movement and provide a physical
explanation for the restoration of proton pumping activity in the double mutant.
(Some figures in this article are in colour only in the electronic version)
1. Introduction
Cytochrome c oxidase, an intrinsic membrane protein, is the
fourth enzyme complex in the respiratory chain. Its role is
4 Address for correspondence: Molecular Structure and Function, Hospital
for Sick Children, 555 University Avenue, Toronto, ON, M5G 1X8, Canada.
to convert dioxygen to water and harness the energy liberated
from this redox reaction to pump protons across the membrane
against an electrochemical gradient (for a review see [1–3]).
The resulting proton-motive force is utilized by ATP synthase
to drive the synthesis of ATP. The redox reaction occurs in
a stepwise manner and takes place in the binuclear centre
0953-8984/11/234102+09$33.00 © 2011 IOP Publishing Ltd Printed in the UK & the USA1

J. Phys.: Condens. Matter 23 (2011) 234102 R M Henry et al
Figure 1. Cytochrome c oxidase of R. sphaeroides. Subunit I, which contains the D channel (inset) and the binuclear centre, is highlighted in
blue. Highlighted are heme a, heme a3, CuA, CuB and key residues of the D channel, which extends from D132 to E286. This conformation is
taken from one of our simulations with N139 in the open state and a chain of 12 water molecules in the D channel. Also pictured is the cap of
57 water molecules at the entrance of the D channel used to represent bulk solution.
(BNC), which includes a copper centre (CuB) and a high-spin
Fe-heme (heme a3) (see figure 1). Throughout the catalytic
cycle, four electrons are obtained from the reduction of a
bimetallic Cu centre, CuA, by cytochrome c. The electrons
are then transferred from CuA to the low-spin heme complex
heme a. The reaction cycle utilizes eight protons, four of
which are consumed in the redox reaction and four of which
are translocated to the exit pathway towards the cytoplasmic
side of the membrane via the proton loading site (PLS).
All eight protons are taken up from the matrix side of the
mitochondrial membrane and transported to the active site
through two cavities in the enzyme, the D channel and the K
channel, which are named after highly conserved Asp132 and
Lys362 residues, respectively (numbering is from subunit I of
R. sphaeroides CcO unless otherwise noted) [1–3].
The majority of the protons used in the reaction cycle are
taken up from the D channel [4, 5]. The D channel is located
in subunit I and extends approximately 25 A˚ from residue
D132 at the entrance to residue E286, itself approximately
12 A˚ from the binuclear centre (figure 1). The negatively
charged group of residue D132 has been proposed to act as
a ‘proton antenna’, recruiting protons from the matrix side
to be transferred through the D channel [6]. Once inside
the D channel, an excess proton is thought to be relayed
throughout the cavity by a Grotthuss-like mechanism [7, 8]
involving successive exchanges of hydrogen nuclei in a chain
of water molecules forming an extended hydrogen-bonded
network [9]. Such a mechanism is supported by the presence
of water providing a pathway for the relay of protons in the D
channel [10, 11] and by simulations of proton movement in the
hydrated cavity forming the upper half of the D channel [12].
At the ‘top’ of the D channel, residue E286 plays an essential
role in the catalytic activity by shuttling protons from the D
channel to both the BNC and the PLS [13–16]. Point mutations
of some of the residues lining the D channel have been
observed to significantly affect the activity of the enzyme and
highlight the importance of the D channel to the mechanism of
redox-coupled proton pumping (see table 1) [4, 6, 17–26]. In
particular, single-point mutations of residues N139 and N207
to an aspartic acid residue produce a decoupled phenotype,
whereby the protein maintains wild-type turnover (i.e. redox
activity) while completely abolishing proton pumping [20, 23].
Moreover, the single-point mutation G204D eliminates all
protein activity [22]. Similarly, neutralization of the presumed
proton antenna in the single-point mutation D132N also results
in an inactive enzyme [6]. Most intriguing, combining the
D132N mutation with the N139D mutant (D132N/N139D)
restores wild-type activity to the formerly decoupled or
inactive enzyme [17]. These findings indicate that, despite its
distant proximity from the BNC and its presumably passive
role as a proton conduit in the uptake and relay of protons, the
D channel plays a vital role in the catalytic mechanism of the
enzyme. For this reason, uncovering the molecular basis of
proton uptake and relay via the D channel will lead to a better
understanding of proton pumping.
Numerous models of the catalytic mechanism of CcO
have been proposed, all of which identify five distinct catalytic
states [27–31]. However, the mechanism which couples redox
chemistry and proton pumping remains unclear. Recently,
we proposed a model for the catalytic cycle of CcO which
rests on the control of vectorial proton transfer by long-range
electrostatic interactions in a recurring sequence of electron
2

J. Phys.: Condens. Matter 23 (2011) 234102 R M Henry et al
Table 1. Summary of oxygen redox activity and proton pumping in the wild-type and in D-channel site-directed mutants of cytochrome c
oxidase.
Organism Mutation Activity (% of WT) Pumping Reference
R. sphaeroides WT 100 Yes —
R. sphaeroides Y33H 40 No [44]
R. sphaeroides N121A 110 Reduced [45]
R. sphaeroides N121T 100 No [45]
R. sphaeroides N121D 70 No [45]
R. sphaeroides D132A 5 No [6, 46]
R. sphaeroides D132N 5 No [6]
E. coli D135N (D132N)a 45 Reduced [26]
R. sphaeroides N139D 150–300 No [20, 47]
P. denitrificans N131D (N139D) 100–115 No [25]
E. coli N142D (N139D) 50 Reduced [26]
R. sphaeroides N139A 40 No [45]
P. denitrificans N131A (N139A) 11 No [48]
R. sphaeroides N139C 110 No [45]
P. denitrificans N131C (N139C) 86 No [48]
R. sphaeroides N139Q 60 Reduced [45]
P. denitrificans N131Q (N139Q) 52 Reduced [48]
E. coli N142V (N139V) 109 Yes [26]
R. sphaeroides N139T 40 No [49]
R. sphaeroides N139E 110 No [45]
P. denitrificans N131E (N139E) 78 No [48]
R. sphaeroides N139L 7 No [45]
R. sphaeroides N139S 90 No [45]
P. denitrificans N131V (N139V) <10 No [25]
E. coli N142V (N139V) 22 Reduced [26]
R. sphaeroides S142A 80 Reduced [45]
P. denitrificans S134A (S142A) 96 Yes [25]
R. sphaeroides S142D 2 No [45]
R. sphaeroides S200I 83 Yes [50]
R. sphaeroides G204D 2 No [22]
R. sphaeroides N207D 100 No [23]
P. denitrificans N199D (N207D) 50 No [25]
R. sphaeroides N207A 90 Reduced [45]
R. sphaeroides N207T 90 Reduced [45]
P. denitrificans N113V/N131D (N121V/N139D) 75 No [25]
R. sphaeroides D132N/N139D 20 Yes [17]
E. coli D135N/N142D (D132N/N139D) 33 Yes [26]
R. sphaeroides D132N/N139T 90 No [45]
R. sphaeroides D132N/S200I 3 No [50]
R. sphaeroides D132N/S200V/S201V 7 No [50]
R. sphaeroides D132N/S200V/S201Y 7 No [50]
R. sphaeroides S142A/N207A 80 Reduced [45]
R. sphaeroides S200V/S201V 37 Reduced [50]
R. sphaeroides S200V/S201Y 11 No [50]
aR. sphaeroides numbering is provided in parentheses.
and proton transfer steps occurring in the active site [32].
In this model, E286 plays a vital role by relaying most of
the chemical and pumped protons to the BNC and the PLS,
respectively. This role requires that its pKa or proton affinity
be at once high enough to guarantee proton uptake from the
D channel and low enough to relay chemical protons on to
the BNC and vectorial protons on to the PLS. To evaluate our
model, the pKa of E286 was calculated to gauge its ability to
deliver chemical and vectorial protons throughout the catalytic
cycle, successively in the wild-type and in the N139D, N207D
and G204 mutants. This study ascribed the mutant phenotypes
to long-range electrostatic forces. Results showed that the
introduction of an anionic group into the D channel induces an
increase in the pKa of E286 and suggested that the decoupled
(N139D, N207D) and inactive (G204D) phenotypes stem from
compromised delivery of vectorial and chemical protons due
to this elevated pKa. In this process, the magnitude of the pKa
shift is modulated by the proximity of the anionic charge to
the carboxylic group of E286. In the case of the decoupled
mutants, the pKa of E286 was raised significantly, which may
still enable the delivery of chemical protons, but is too high to
deliver pumped protons. In the inactive G204D mutant, whose
side chain lies closer to E286 by 3 A˚, the pKa of E286 further
increased by an additional unit relative to the wild-type, which
is presumably too high to allow any proton delivery.
Absent from our previous study was an analysis of the
effect of the D132N/N139D double mutant on the pKa of
E286. This double mutant is particularly intriguing since
it is able at once to restore activity to a decoupled mutant,
N139D, and to an inactive mutant, D132N (see table 1).
3

J. Phys.: Condens. Matter 23 (2011) 234102 R M Henry et al
Since we have proposed that the phenotype of the single-
point mutant N139D is due to the altered pKa of E286,
the phenotype of the double mutant puts our electrostatic
model to the test. If the decoupling of proton pumping in
N139D is due to the introduction of a negative charge in the
D channel, why is the double mutant, which conserves the
charge of residue N139D, functional? In another development,
our recent comparative study of functional hydration in the
wild-type enzyme and in two uncharged single-point mutants
(respectively alanine and valine, N139A and N139V) revealed
the existence of multiple rotameric states in the side chain of
residue N139 [11]. Specifically, this side chain was found
to predominantly adopt the ‘closed’ conformation seen in
the crystallographic structures [10], whereby it interrupts the
hydrogen-bonded chain of water molecules in the D channel,
and to adopt a metastable ‘open’ conformation in which
hydrogen-bond connectivity is established and water-mediated
proton relay may proceed according to a Grotthuss mechanism.
Interestingly, the occlusion of this conformational gate in the
N139V mutant corresponds to an inactive enzyme, whereas
the removal of the gate in N139A corresponds to a decoupled
phenotype, suggesting that conformational gating of proton
uptake by residue N139 is indeed required for the proper
function of the enzyme [11].
The existence of a conformational gate modulating
proton uptake in the wild-type enzyme suggests that multiple
rotameric states may also exist for the N139D side chain.
If so, these states must be accounted for in our electrostatic
model of proton pumping. To this end, we now examine
the physical basis for the phenotype of single and double
mutants. We first present systematic free energy calculations
to characterize the conformational isomerization of the side
chain of residue 139 successively in the wild-type protein
and in the N139D and D132N/N139D mutants. We then
assess the effect of this conformational isomerization on the
pKa of E286 using continuum electrostatic calculations. The
results reveal the existence of an additional conformation
in which the side chain of residue 139 faces out, towards
the entrance of the D channel. This ‘out-facing’ conformer
predominates in the double mutant, eliminating the shift in
the pKa of E286 induced by long-range electrostatic repulsion
with N139D while putatively restoring the proton antenna
capability compromised in the D132N mutant. Together,
these findings offer a simple explanation for the phenotype of
the double mutant consistent with the electrostatic control of
proton uptake and delivery. By restoring the charge distribution
of the enzyme at the entrance of the D channel, the double
mutant restores the proton pumping activity of the wild-type
enzyme compromised in the two single-point mutants.
2. Methods
2.1. Molecular system
The initial conformation of the protein was obtained from
the structure of R. sphaeroides cytochrome c oxidase solved
at 2.3 A˚ resolution by x-ray crystallography (PDB ID code
1M56) [10]. Mutant structures were obtained by performing
mutations to N139 and D132 in silico, followed by 2 ns of
equilibration at 298 K. All simulations were performed with
12 water molecules present in the D channel following our
systematic study of hydration [11]. As in that previous study,
a hemispherical ‘cap’ of water molecules was placed at the
entrance of the D channel to model bulk solution on the matrix
side of the protein. The 57 water molecules were restrained by
a spherical boundary potential with a radius of 9 A˚ centred at
the channel opening and a force constant of 5 kcal mol−1 A˚−2.
An axis connecting the C
α
atoms of D132 and E286 was
used to define the D channel and was aligned with the z axis.
Residues with at least one heavy atom within 5 A˚ from the
D-channel axis as well as the D-channel water molecules and
cap water molecules were allowed to move during the MD
simulations. All remaining atoms in the system were held
fixed.
The CHARMM force field, version 22 [33], was used to
model the protein, and the TIP3P force field [34] was used
to model water molecules. The charge distribution of the
binuclear centre was calculated as described in Fadda et al
[32]. The enzyme was simulated in the fully reduced R state.
Titratable residues were simulated at standard protonation
states. In particular, E286 was modelled in its deprotonated
state to mimic conditions favourable to proton uptake in the D
channel [32].
2.2. Molecular dynamics simulations
The MD trajectories were generated using the program
CHARMM, version 28 [35]. The Langevin equations of
motion were propagated at 298 K with an integration step
of 2 fs and a friction coefficient of 5 ps−1 applied to all
heavy atoms. The SHAKE algorithm [36] was employed to
fix all covalent bonds involving hydrogen atoms with a bond
deviation tolerance of 1.0 × 10−6. Non-bonded interactions
were calculated with a force-based switching function acting
between 14 and 16 A˚. Trajectories and structures were viewed
using visual molecular dynamics (VMD) [37].
2.3. Free energy calculations
The reversible free energy change or potential of mean
force (PMF) for the rotation about the C
α
–C
β
torsion angle
χ1 of residue 139 was calculated using umbrella sampling
(US) [38] together with distributed replica sampling (DR) [39]
to reduce systematic sampling errors [40, 41]. In the
US scheme, the χ1 torsion of residue 139 was subjected
to confinement by quadratic biasing potentials of the form
Vi(χ1) = 12 ki(χ1 − χ1,i)
2
, where ki is the restoring force
constant and χ1,i ranged from 0◦ to 320◦ in 10◦ increments. DR
is a generalized-ensemble algorithm designed to improve the
efficiency of Boltzmann sampling by achieving a random walk
in temperature or (as in the present case) in conformational
space [39, 42]. Multiple replicas of a protein system
differing in reaction coordinate (here, umbrella restraint Vi
centred at χ1,i ) are simulated independently. Periodically,
individual replicas are halted and a stochastic move replacing
Vi by adjacent umbrella Vi±1 is attempted. Construction
4

J. Phys.: Condens. Matter 23 (2011) 234102 R M Henry et al
Figure 2. Representative snapshots of the D channel for three
different conformations of residue 139: (a) closed (crystallographic)
conformation of N139 in the wild-type enzyme; (b) open
conformation of the N139D single-point mutant; and (c) out-facing
conformation of N139D in the D132N/N139D double mutant. Water
molecules present in or near the D channel are shown in spherical
representation together with licorice representations of the side
chains of residues 132, 139 and 286. The C
α
trace of the protein is
depicted as ribbons.
of all starting structures began from equilibrated structures
of residues N139 and N139D in their closed positions of
χ1 ≈ 195◦ and 210◦, respectively [11]. From these initial
conformations, the remaining structures were generated by
running successive 400 ps simulations. Here, the final structure
of a replica with χ1,i = X was used as the starting structure
for the replica with χ1,i = X + 10◦ or χ1,i = X − 10◦,
as required. All replicas were then equilibrated for another
400 ps with a force constant ki = 0.02 kcal/mol/deg2. Next, a
series of 2 ns simulations of all replicas were performed while
adjusting both the umbrella centres χ1,i and force constants
ki in an attempt to obtain a near 20% overlap between all
adjacent umbrellas in the sampled χ1 space. This procedure
resulted in small differences in the final umbrella positions and
force constants for the wild-type, N139D, and D132N/N139D
systems. Weight values (also called ‘A values’) required to
achieve a random walk in the DR scheme were calculated
adaptively as described before [40, 42] using 1 ns of simulation
time per replica with no exchanges. All replicas were then run
while allowing umbrella replacements every 2 ps. The total
sampling time was 631 ns, 772 ns and 834 ns for the wild-type,
N139D and D132N/N139D systems, respectively. Throughout
the simulation, weight values were adjusted based on the
current calculated free energy surface. PMFs were generated
using Alan Grossfield’s implementation of WHAM [43]. Error
was calculated using block averaging. A separate PMF was
generated for each block of data and a global fit was used
to align all the PMFs thus generated. The standard deviation
between the PMFs was then calculated.
Figure 3. PMF for the χ1 rotation of N139 (blue), N139D (red) and
N139D in the D132N/N139D double mutant (green).
2.4. Continuum electrostatic calculations
The pKa shift calculations were performed with the PBEQ
module of CHARMM [35]. Details pertaining to atomic
radii used and dielectric constant values assigned to different
regions of the system can be found in our previous study [32].
No explicit water molecules were included in the PBEQ
calculation. The Poisson–Boltzmann (PB) equation was solved
numerically by finite difference on a coarse grid (0.60 A˚ mesh
size) focused on regions of interest up to a 0.30 A˚ mesh size.
The pKa values are averaged over 150 snapshots selected
randomly from five independent 2.5 ns molecular dynamics
trajectories of the wild-type protein and both mutants. These
calculations were performed separately for the closed and out-
facing conformational states of residue N139D in the single or
double mutants.
3. Results and discussion
3.1. Conformational isomerization of residue 139
The rotameric states of residue 139 in the wild-type, single-
point mutant N139D, and double mutant D132N/N139D
were characterized using free energy simulations combin-
ing umbrella sampling [38] and distributed replica sam-
pling [39, 40, 42]. Figures 2 and 3 depict representative
conformations of residue 139 in the wider context of the D
channel and the PMF for the conformational isomerization of
the χ1 torsion of residue 139, respectively. Our previous study
of functional hydration in the D channel characterized the
conformational equilibrium between the ‘closed’ and ‘open’
conformations of N139, in which the side chain prevents and
allows the formation of a hydrogen-bonded chain of water
molecules, respectively (figures 2(a) and (b)) [11]. The
conformational isomerization between the preferred closed
state and the metastable open state was inferred to provide a
conformational gate for proton uptake, since the presence of
hydrogen bonds is a prerequisite for water-mediated proton
relay [9] across the narrow bottleneck of the D channel.
5

J. Phys.: Condens. Matter 23 (2011) 234102 R M Henry et al
Here we have extended the range of the PMF to include
rotation of the terminal group towards the entrance of the D
channel, in the vestibule lying between residues 132 and 139
(figure 2(c)). This conformation will be referred to as the
out-facing conformation from this point on. The free energy
profiles reveal that, although the side chain of residue 139 in
all three enzymes can adopt all three rotameric states (closed,
open, out-facing), the stability of the out-facing conformer
relative to the other two conformations varies dramatically
(figure 3).
As we have shown previously [11], the closed state of
residue N139 is favoured over the open state by approximately
4 kcal mol−1. Furthermore, the out-facing conformation is
highly disfavoured in the wild-type enzyme, with a free energy
of approximately 12 kcal mol−1 relative to the closed state.
In the single-point mutant N139D, the equilibrium between
open and closed states does not differ significantly from that
of the wild-type, with a 3 kcal mol−1 preference for the closed
state. However, the free energy of the out-facing conformer
is now comparable to that of the open state. In contrast, in
the N139D/D132N double mutant the out-facing conformer
undergoes a dramatic stabilization relative to the single-point
mutant and becomes preferred over closed and open states by
5 and 6 kcal mol−1, respectively. Based on these results, the
most populated conformations of residue 139 (over 98%) are
predicted to be the closed, closed and out-facing states in the
wild-type enzyme and in N139D and D132N/N139D mutants,
respectively. These results justify, a posteriori, the neglect of
the out-facing conformation in our previous studies of the wild-
type and single-point mutant [32] while forcing us to consider
its role in the restored phenotype of the double mutant.
The high energetic penalty incurred when residue N139
adopts the out-facing state is likely due to poor hydrogen-
bonding interactions between the amide terminus of N139
and chemical groups lining the vestibule of the D channel.
In the single-point mutant N139D, the relative stabilization
of the out-facing residue is likely to be due at least in part
to the better hydration of the charged carboxylate group in
the vestibule of the D channel, compared to the closed and
open conformers of N139D (compare the hydrated state of the
N139D side chain in figures 2(b) and (c)). However, the out-
facing conformation also brings the two carboxylate groups of
D132 and N139D in closer proximity to each other, increasing
the Coulombic repulsion between these two groups. Indeed,
we hypothesize that the dramatic stabilization of the out-facing
conformation in the double mutant is due to the removal of
the charge–charge repulsion resulting from the neutralization
of residue 132. Residue 132, at the entrance of the channel, lies
approximately 6 A˚ from residue 139. As shown in our previous
study of the single N139D mutant [32] and in the following
section, electrostatic interactions between the charged groups
of N139D and E286, which lie at least 18 A˚ apart in the D
channel, are strong enough to induce significant pKa shifts.
If the charge state of residue 132 modulates the
conformational equilibrium of N139D, it may be expected that,
reciprocally, the conformational equilibrium of D132 depends
on the charge and the conformational state of residue 139.
Specifically, both the introduction of a charge in the single
Figure 4. pKa of E286 in successive catalytic steps in the wild-type
enzyme (light blue), in the closed state of the single-point mutant
N139D (dark blue) and of the double mutant D132N/N139D (green),
and in the out-facing conformation of the double mutant (red). The
letter codes refer to the catalytic states of the enzyme and the square
brackets [0|1], [1|0] and [1|1] represent the distribution of charge in
the two moieties of the active site, as defined in [32]. According to
our electrostatic model, these three states correspond to proton
delivery from E286 to the PLS, proton delivery from E286 to the
BNC, and proton pumping steps, respectively. The average pKa of
E286 in proton-delivery states [0|1] and [1|0] are 10.8 and 11.7 for
the wild-type enzyme and the N139D single mutant, respectively.
The lines connecting the pKa values are shown to guide the eye.
Standard deviations are within 0.3 units for all the calculated pKa
values.
N139D mutant and the conformational isomerization of that
residue to its out-facing conformer may be expected to shift the
conformational equilibrium of D132. However, the analysis
of the rotameric states of the D132 side chain sampled in
our simulation did not show any such dependence (results
not shown). Resolving the issue of conformational coupling
between residues 132 and 139 would necessitate a systematic
characterization, using PMF calculations, of the rotameric
equilibrium of residue D132 in different rotameric states of
N139 and N139D. Such a study is beyond the scope of the
present paper.
3.2. Effect of mutations on the pKa of the proton shuttle E286
In light of the above results, continuum electrostatic
calculations were performed on the double mutant in the out-
facing state and, for comparison, the closed state. Since the
PMF profiles described above support the assumption that the
closed states of N139 and N139D (single-point mutant) are the
preferred conformers, calculations of the pKa of E286 in the
closed state of the wild-type and of the N139D single-point
mutant are also included for comparison.
Figure 4 shows the calculated pKa of E286 in successive
catalytic steps of the enzyme. In this plot, [0|1], [1|0], and
[1|1] denote the sequence of three alternating charge states of
the active site of the enzyme, where the numbers on the left and
on the right denote the total charge of heme a and the BNC,
6

J. Phys.: Condens. Matter 23 (2011) 234102 R M Henry et al
respectively [32]. Ten catalytic states denoted F through MV
are shown, which are thought to involve the uptake of protons
through the D channel, as described in detail in Fadda et al
[32]. Together, these ten distinct catalytic states make up three
out of four recurring subcycles in which electron and proton
transfer are coupled electrostatically: the uptake of a proton
by the PLS following reduction of heme a in state [0|1], the
uptake of a chemical proton by the BNC following electron
transfer from heme a to heme a3 in state [1|0], and the pumping
of a proton upon the ensuing maximization of the positive
charge in the active site of the enzyme (state [1|1]) [32]. As
seen in figure 4, the periodic recurrence of the three charge
states induces a pseudo-periodic cycle in the pKa of E286, the
residue at the top of the D channel which shuttles protons from
the D channel alternatively to the PLS and to the BNC in states
[0|1] and [1|0], respectively. In each of these cycles, the pKa
of E286 drops by up to two units in state [1|1].
As previously discussed [32], the introduction of a
negatively charged residue at position 139 destabilizes the
ionic state of residue E286 relative to its neutral, protonated
state, resulting in a further increase in its already elevated
proton affinity. This perturbation is significant in the closed
conformer of N139D, which is the most populated state of the
single mutant. However, this additional pKa shift is smaller in
the closed state of the double mutant and becomes negligible
in the out-facing state, which is the predominant conformation
of N139D in the double mutant (figure 4). The restoring effect
of the double mutant, which brings the average pKa of E286
within statistical uncertainty to that of the wild-type enzyme
in the two proton-delivery states, is due to the increasing
separation and better hydration of the carboxylate group of
N139D in the out-facing state (see figure 2).
3.3. Mechanism for the restoration of proton pumping activity
The phenotype of the D132N/N139D double mutant is
remarkable in its ability to restore wild-type activity to both the
N139D decoupled mutant and the inactive D132N mutant [6].
As such, any proposed mechanism regarding the restored
activity of the double mutant should include an explanation
for both of these phenomena. Our results provide plausible
explanations for both these experimental results.
When D132 is mutated to any non-carboxylate residue,
the activity of the enzyme is reduced to 5% of wild-type [6]
(see table 1). This is likely to be due to the removal of the
ability for residue 132 to recruit protons to the entrance of
the D channel, since the activity of the D132A mutant can
be increased by increasing the concentration of protons on the
matrix side of the protein [6]. Consequently, it is likely that
the return to wild-type activity in the double mutant is due to
the N139D carboxylate group acting as a replacement proton
antenna. Our PMF results further support this hypothesis, since
the low-energy out-facing rotamer of N139D in the double
mutant places the carboxylate group of N139D a mere 6 A˚
from the location of the wild-type D132 side chain in the
crystallographic structure. In this orientation, it is inferred that
residue N139D is sufficiently close to the matrix side of the
protein as to act as a proton antenna. In addition, this out-
facing rotamer is critical in explaining the phenotype of the
double mutant in regards to the restoration of proton pumping
activity to the N139D decoupled mutant.
In our previous study of decoupled single-point mutants
N139D and N207D, we attributed the decoupling of proton
pumping from redox chemistry to an increase in the pKa of
E286 small enough to allow proton delivery to the BNC but
high enough to compromise proton delivery to the PLS [32].
Importantly, we also found that the magnitude of this pKa shift
was modulated by the proximity of the new charge to E286.
Consistently with these findings, the above results show that
swinging the N139D side chain to its out-facing conformation
(figure 2), which is only marginally populated in the single
mutant due to electrostatic repulsion with D132 but becomes
the preferred conformer in the double mutant (figure 3) thanks
to the neutralization of the charge of residue 132, brings
the pKa of E286 back to near-wild-type values (figure 4),
consistent with the restoration of proton pumping [17].
4. Conclusions
We have examined the structural and electrostatic basis for
decoupling and re-coupling of proton pumping and redox
chemistry in the N139D and D132N/N139D mutants of CcO.
The above results shed light onto the role of residues lining the
D channel in the uptake and relay of protons to the active site
of the enzyme. These findings are consistent with our previous
analysis of structural fluctuations in the D channel [11]
and support our previously proposed electrostatic model of
proton pumping [32]. Detailed free energy simulations of the
conformational equilibrium of residue 139 confirm that the
closed conformer found in crystal structures of the enzyme is
the lowest energy state in both the wild-type and the N139D
mutant. In addition, this study has uncovered an out-facing
conformer which is metastable in the wild-type protein and
in the single N139D mutant but is predominant in the double
mutant. Finally, electrostatic calculations suggest that the
preference of the N139D carboxylate group for the out-facing
conformation, where it may replace the ‘proton antenna’ lost
in the neutralization of D132 while at the same time restoring
the proton affinity of E286 and therefore its ability to deliver
vectorial protons, restores proton pumping activity to the
double mutant.
Thus, this study provides a consistent model explaining
how the D132N/N139D double mutant re-establishes the
electrostatic balance governing the uptake of protons in the
enzyme, which was compromised both in D132N and N139D.
Consistently with our proposed model of proton pumping,
these findings underline the importance of electrostatic driving
forces in the control of proton movement in the enzyme
interior. In particular, our results suggest that preserving
the spatial distribution of charges at the entrance of the D
channel is necessary to guarantee both the uptake of protons
and the subsequent relay of these protons to the catalytic
site and the pumping element. More generally, our results
also illustrate how energetic factors driving proton movement
are modulated by conformational fluctuations of the enzyme.
Although understanding the physical basis of proton uptake
is an important step, much remains to be done to elucidate
7

J. Phys.: Condens. Matter 23 (2011) 234102 R M Henry et al
the molecular basis for the kinetic control necessary for
preventing leakage and ensuring the directionality of proton
movement in the pump cycle. In that perspective, further
characterization of the interplay of conformational fluctuations
and proton translocation is a challenge for future theoretical
and computational work.
Acknowledgments
We thank the Centre for Computational Biology of the Hospital
for Sick Children for a generous allocation of CPU resources
and we gratefully acknowledge the Canadian Institutes of
Health Research (grant MOP43949) for support. RP is a CRCP
chairholder.
References
[1] Wikstro¨m M and Verkhovsky M I 2006 Towards the
mechanism of proton pumping by the haem-copper oxidases
Biochim. Biophys. Acta 1757 1047–51
[2] Bra¨nde´n G, Gennis R B and Brzezinski P 2006 Transmembrane
proton translocation by cytochrome c oxidase Biochim.
Biophys. Acta 1757 1052–63
[3] Brzezinski P and Adelroth P 2006 Design principles of
proton-pumping haem-copper oxidases Curr. Opin. Struct.
Biol. 16 465–72
[4] Konstantinov A A et al 1997 The roles of the two proton input
channels in cytochrome c oxidase from Rhodobacter
sphaeroides probed by the effects of site-directed mutations
on time-resolved electrogenic intraprotein proton transfer
Proc. Natl Acad. Sci. USA 94 9085–90
[5] Brzezinski P and Adelroth P 1998 Pathways of proton transfer
in cytochrome c oxidase J. Bioenerg. Biomembr. 30 99–107
[6] Mills D A et al 2000 Where is ‘outside’ in cytochrome c
oxidase and how and when do protons get there? Biochim.
Biophys. Acta 1458 180–7
[7] de Grotthuss C J T 1806 Me´moire sur la de´compostion de l’eau
et des corps qu’elle tient en dissolution a` l’aide de
l’e´lectricite´ galvanique Ann. Chim. 58 54–74
[8] de Grotthuss C J T 2006 Memoir on the decomposition of water
and of the bodies that it holds in solution by means of
galvanic electricity Biochim. Biophys. Acta 1757 871–5
[9] Pome`s R and Roux B 1996 Structure and dynamics of a proton
wire: a theoretical study of H+ translocation along the
single-file water chain in the gramicidin channel Biophys. J.
71 19–39
[10] Svensson-Ek M et al 2002 The x-ray crystal structures of
wild-type and EQ(I-286) mutant cytochrome C oxidases
from rhodobacter sphaeroides J. Mol. Biol. 321 329–39
[11] Henry R et al 2009 Functional hydration and conformational
gating of proton uptake in cytochrome c oxidase J. Mol.
Biol. 387 1165–85
[12] Xu J and Voth G A 2005 Computer simulation of explicit
proton translocation in cytochrome c oxidase: the
D-pathway Proc. Natl Acad. Sci. USA 102 6795–800
[13] Riistama S et al 1997 Bound water in the proton translocation
mechanism of heam-copper oxidases FEBS Lett. 414 275–80
[14] Hofacker I and Schulten K 1998 Oxygen and proton pathways
in cytochrome c oxidase Proteins Struct. Funct. Gen.
30 100–7
[15] Pome`s R, Hummer G and Wikstro¨m M 1998 Structure and
dynamics of a proton shuttle in cytochrome c oxidase
Biochim. Biophys. Acta 1365 255–60
[16] Adelroth P et al 2000 Proton transfer from glutamate 286
determines the transition rates between oxygen intermediates
in cytochrome c oxidase Biochim. Biophys. Acta 1459 533–9
[17] Bra¨nde´n G et al 2006 Controlled uncoupling and recoupling of
proton pumping in cytochrome c oxidase Proc. Natl Acad.
Sci. USA 103 317–22
[18] Adelroth P et al 1997 Glutamate 286 in cytochrome aa3 from
rhodobacter sphaeroides is involved in proton uptake during
the reaction of the fully-reduced enzyme with dioxygen
Biochemistry 36 13824–9
[19] Aagaard A et al 2000 Redesign of the proton-pumping
machinery of cytochrome c oxidase: proton pumping does
not require Glu(I-286) Biochemistry 39 15847–50
[20] Pawate A S et al 2002 A mutation in subunit I of cytochrome c
oxidase from rhodobacter sphaeroides results in an increase
in steady state activity but completely eliminates proton
pumping Biochemistry 41 13417–23
[21] Mitchell D M et al 1996 Site-directed mutagenesis of residues
lining a putative proton transfer pathway in cytochrome c
oxidase from rhodobacter sphaeroides Biochemistry
35 13089–93
[22] Han D, Morgan J E and Gennis R B 2005 G204D, a mutation
that blocks the proton-conducting D-channel of the aa3-type
cytochrome c oxidase from rhodobacter sphaeroides
Biochemistry 44 12767–74
[23] Han D et al 2006 Replacing Asn207 by aspartate at the neck of
the D channel in the aa3-type cytochrome c oxidase from
rhodobacter sphaeroides results in decoupling the proton
pump Biochemistry 45 14064–74
[24] Backgren C et al 2000 Proton translocation by cytochrome c
oxidase can take place without the conserved glutamic acid
in subunit I Biochemistry 39 7863–7
[25] Pfitzner U et al 2000 Tracing the D-pathway in reconstituted
site-directed mutants of cytochrome c oxidase from
paracoccus denitrificans Biochemistry 39 6756–62
[26] Garcia-Horsman J A et al 1995 Proton transfer in cytochrome
bo3 ubiquinol oxidase of Escherichia coli: second-site
mutations in subunit I that restore proton pumping in the
mutant Asp135 Biochemistry 34 4428–33
[27] Belevich I et al 2007 Exploring the proton pump mechanism of
cytochrome c oxidase in real time Proc. Natl Acad. Sci. USA
104 2685–90
[28] Michel H 1998 The mechanism of proton pumping by
cytochrome c oxidase Proc. Natl Acad. Sci. USA
95 12819–24
[29] Faxe´n K et al 2005 A mechanistic principle for proton pumping
by cytochrome c oxidase Nature 437 286–9
[30] Belevich I, Verkhovsky M I and Wikstro¨m M 2006
Proton-coupled electron transfer drives the proton pump of
cytochrome c oxidase Nature 440 829–32
[31] Wikstro¨m M and Verkhovsky M I 2007 Mechanism and
energetics of proton translocation by the respiratory
heme-copper oxidases Biochim. Biophys. Acta
1767 1200–14
[32] Fadda E, Yu C H and Pome`s R 2008 Electrostatic control of
proton pumping in cytochrome c oxidase Biochim. Biophys.
Acta 1777 277–84
[33] MacKerell A D J et al 1998 All-atom empirical potential for
molecular modeling and dynamics studies of proteins
J. Phys. Chem. B 102 3586–616
[34] Jorgensen W L et al 1983 Comparison of simple potential
functions for simulating liquid water J. Chem. Phys.
79 926–35
[35] Brooks B R et al 1983 CHARMM: a program for
macromolecular energy minimization and dynamics
calculations J. Comput. Chem. 4 187–217
[36] Ryckaert J P, Ciccotti G and Berendsen H J C 1977
Numerical-integration of cartesian equations of motion of a
system with constraints—molecular-dynamics of N-alkanes
J. Comput. Phys. 23 327–41
[37] Humphrey W, Dalke A and Schulten K 1996 VMD: visual
molecular dynamics J. Mol. Graph. 14 33–8
8

J. Phys.: Condens. Matter 23 (2011) 234102 R M Henry et al
[38] Torrie G M and Valleau J P 1977 Nonphysical sampling
distributions in Monte Carle free-energy estimation:
umbrella sampling J. Comput. Phys. 23 187–99
[39] Rodinger T, Howell P L and Pome`s R 2006 Distributed replica
sampling J. Chem. Theory Comput. 2 725–31
[40] Neale C, Rodinger T and Pome`s R 2008 Equilibrium exchange
enhances the convergence rate of umbrella sampling Chem.
Phys. Lett. 460 375–81
[41] Rodinger T, Howell P L and Pome`s R 2008 Calculation of
absolute protein–ligand binding free energy using distributed
replica sampling J. Phys. Chem. 129 155102
[42] Rauscher S, Neale C and Pome`s R 2009 Simulated tempering
distributed replica sampling, virtual replica exchange, and
other generalized-ensemble methods for conformational
sampling J. Chem. Theory Comput. 5 2640–62
[43] Kumar S et al 1995 Multidimensional free-energy calculations
using the weighted histogram analysis method J. Comput.
Chem. 16 1339–50
[44] Namslauer I et al 2010 A pathogenic mutation in cytochrome
c oxidase results in impaired proton pumping while retaining
O2-reduction activity Biochim. Biophys. Acta 1797 550–6
[45] Zhu J et al 2010 Decoupling mutations in the D-channel of
the aa3-type cytochrome c oxidase from rhodobacter
sphaeroides suggest that a continuous hydrogen-bonded
chain of waters is essential for proton pumping Biochemistry
49 4476–82
[46] Fetter J R et al 1995 Possible proton relay pathways in
cytochrome c oxidase Proc. Natl Acad. Sci. USA 92 1604–8
[47] Namslauer A et al 2003 Redox-coupled proton translocation in
biological systems: proton shuttling in cytochrome c oxidase
Proc. Natl Acad. Sci. USA 100 15543–7
[48] Du¨rr K L et al 2008 A D-pathway mutation decouples the
Paracoccus denitrificans cytochrome c oxidase by altering
the side-chain orientation of a distant conserved glutamate
J. Mol. Biol. 384 865–77
[49] Lepp H et al 2008 Impaired proton pumping in cytochrome c
oxidase upon structural alteration of the D pathway Biochim.
Biophys. Acta 1777 897–903
[50] Lee H J et al 2010 Intricate role of water in proton transport
through cytochrome c oxidase J. Am. Chem. Soc.
132 16225–39
9