Structuralevidenceforaconstant c
11
ring stoichiometry
in thesodiumF-ATP synthase
Thomas Meier
1
, Jinshu Yu
2
, Thomas Raschle
1
, Fabienne Henzen
1
, Peter Dimroth
1
and Daniel J. Muller
2
1 Institut fu
¨
r Mikrobiologie, Eidgeno
¨
ssische Technische Hochschule, Zu
¨
rich, Switzerland
2 Center of Biotechnology, University of Technology, Dresden, Germany
F-type ATP synthases are multisubunit protein com-
plexes lo cated i n the membrane of mitochondria, chloro-
plasts and bacteria. These enzymes use an electrochemical
H
+
or Na
+
gradient across the host membrane for
the synthesis of ATP, the universal energy currency of
living cells [1]. F-ATP synthases are built from two
subcomplexes: the water soluble F
1
part, with the
composition a
3
b
3
cde, and the membrane-embedded F
o
part, which inthe simplest case has the subunit com-
position ab
2
c
10)14
. Inthe crystal structure of F
1
, a
3
b
3
forms a cylinder around the extended a-helical c sub-
unit [2]. Part of the c subunit protrudes from the
bottom of the cylinder and forms the central stalk
together with the e subunit. A striking feature of the
structure is an inherent asymmetry among the catalytic
b subunits. In combination with the asymmetric loca-
tion of the central c subunit, and in accordance with
the binding change mechanism [3], this arrangement
suggests a catalytic mechanism in which the c subunit
rotates within the a
3
b
3
cylinder. Elegant experiments
have subsequently visually verified this rotation [4].
The F
o
part, a rotary motor by itself, is fueled by the
electrochemical H
+
or Na
+
gradient and, upon rota-
tion, translocates these ions across the membrane.
Recent studies of the F
o
motor have focused on the
ion path through the membrane and the coupling
Keywords
atomic force microscopy; c ring
stoichiometry; F-ATP synthase; Ilyobacter
tartaricus; Propionigenium modestum
Correspondence
T. Meier, Institut fu
¨
r Mikrobiologie,
Eidgeno
¨
ssische Technische Hochschule
Zu
¨
rich (ETH-Ho
¨
nggerberg), Wolfgang-Pauli-
Str. 10, CH 8093 Zu
¨
rich, Switzerland
Fax: +41 44 6321378
Tel: +41 44 6325523
E-mail: meier@micro.biol.ethz.ch
(Received 27 June 2005, accepted 26
August 2005)
doi:10.1111/j.1742-4658.2005.04940.x
The Na
+
-dependent F-ATP synthases of Ilyobacter tartaricus and Propioni-
genium modestum contain membrane-embedded ring-shaped c subunit
assemblies with astoichiometry of 11. Subunit c from either organism was
overexpressed in Escherichia coli using a plasmid containing the corres-
ponding gene, extracted from the membrane using detergent and then puri-
fied. Subsequent analyses by SDS ⁄ PAGE revealed that only a minor
portion of the c subunits had assembled into stable rings, while the major-
ity migrated as monomers. The population of rings consisted mainly of c
11
,
but more slowly migrating assemblies were also found, which might reflect
other c ring stoichiometries. We show that they consisted of higher aggre-
gates of homogeneous c
11
rings and ⁄ or assemblies of c
11
rings and single
c monomers. Atomic force microscopy topographs of c rings reconstituted
into lipid bilayers showed that the c ring assemblies had identical diameters
and that stoichiometries throughout all rings resolved at high resolution.
This finding did not depend on whether the rings were assembled into crys-
talline or densely packed assemblies. Most of these rings represented com-
pletely assembled undecameric complexes. Occasionally, rings lacking a few
subunits or hosting additional subunits in their cavity were observed. The
latter rings may represent the aggregates between c
11
and c
1
, as observed
by SDS ⁄ PAGE. Our results are congruent with a stable c
11
ring stoichiom-
etry that seems to not be influenced by the expression level of subunit c in
the bacteria.
Abbreviations
AFM, atomic force microscopy.
5474 FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS
between ion flux and torque generation [5–7]. Addi-
tionally, using various experimental approaches,
increasing evidence has accumulated on the overall
shape of the F
o
domain, particularly the ring-shaped
c subunit assemblies. The 2.4 A
˚
resolution structure of
the Ilyobacter tartaricus c ring, reported recently,
imposes important restrictions on proposed models for
ion translocation and torque generation [8]. Further-
more, a k ring structure from the V-ATPase of Entero-
coccus hirae shares features of ion binding with the
I. tartaricus c ring, supporting a common ion transloca-
tion mechanism inthe two types of ATPases [9]. For
detailed insights into this mechanism, high-resolution
structural data of the F
o
subunits a and b
2
are required.
It has been shown that these subunits flank the c ring
on the outside, but details of their structures have not
yet been explored [10]. The number of c subunits in the
rotor ring is not fixed but varies among species. For
example, rings comprising 10, 14 and 11 subunits have
been found in yeast [11], chloroplasts [12] and the bac-
terium I. tartaricus [13], respectively. Recently, a ring
of 15 subunits was found inthe alkaliphilic cyanobacte-
rium Spirulina platensis, demonstrating that a sym-
metry mismatch between the F
1
and F
o
motor is not an
essential feature for function [14]. The number of
binding sites on the c ring determines the ATP to
proton ⁄ Na
+
ratio, and therefore this stoichiometry is
an important bioenergetic parameter forthe cell. As
the number of subunits can vary among species, the
question was raised whether this stoichiometry could
also vary within one species in order to adapt to speci-
fic energetic requirements of the cells [15]. This proposi-
tion seemed to be supported by an effect of the carbon
source on the expression level of subunit c in Escheri-
chia coli. However, structural analyses of rotors from
I. tartaricus and from chloroplasts showed that their
stoichiometry seems to be constrained by the nearest
neighbor interaction between the subunits [16]. This
question has also been addressed with subunit c from
Escherichia coli [17], where annular shaped particles
were detected by electron microscopy after reconstitu-
tion from single c subunits. In agreement with the
above conclusions, it has been suggested that the
primary protein structure determines the ability of
subunit c to form rings. Furthermore, it was shown, by
gradient gel analysis, that the number of subunits in
the oligomer III isolated from the Chlamydomonas
reinhardtii chloroplast ATP synthase is not affected
by the metabolic state of the cells [18]. However, to
date, no structural methods have been applied to
clarify whether thestoichiometry of the c rings is
influenced by variation of the expression level of
subunit c.
In the present study we investigated whether the
c subunits from thesodiumF-ATPsynthase of
I. tartaricus and of Propionigenium modestum assemble
into uniform rings after heterologous overexpression in
E. coli. Atomic force microscopy (AFM) topographs
showed that complete rings were composed exclusively
of 11 subunits and that defective rings exhibited the
same diameter as intact ones. Hence, intrinsic features
of the I. tartaricus c subunits are responsible for the
formation of c
11
oligomers inthe fully assembled rings.
Results and Discussion
Synthesis and assembly of subunit c from
I. tartaricus in E. coli
To investigate whether heterologously synthesized c
subunits from I. tartaricus assemble properly into
rings, we used the E. coli strain BL21(DE3) as a host.
From wild-type P. modestum or I. tartaricus cells, the
c
11
ring is easily purified by extraction from mem-
branes using lauroylsarcosine and subsequent ammo-
nium sulfate precipitation. As a result of their extreme
stability, these rings are easily recognized by
SDS ⁄ PAGE. The E. coli BL21(DE3) cells transformed
with recombinant plasmids harboring the c subunit
genes of I. tartaricus or P. modestum under control of
the strong T7 promoter produced large amounts of the
appropriate c subunit inthe monomeric state, but also
sizeable amounts of oligomeric assemblies with a ratio
of 9:1 (c
1
:c
oligo
). For further analyses, these
assemblies were purified by sucrose density gradient
centrifugation and subjected to SDS ⁄ PAGE. The
results shown in Fig. 1 indicate that the assemblies
consisted not only of c
11
, but also of higher aggregates.
However, these aggregates are made up exclusively of
c subunits because they are converted completely into
the monomeric form by treatment with trichloroacetic
acid. A similar pattern of bands was observed with the
recombinant T67C mutant, with the exception of an
additional band corresponding to (c
11
)
2
and aggregates
at the top of the gel. For comparison, the c
11
ring pre-
paration of wild-type I. tartaricus cells is shown. Here,
the c
11
ring formed the most prominent species, and
higher aggregates or the c monomer were less abun-
dant than inthe preparations from recombinant E. coli
cells.
Investigation of the aggregation status of c ring
preparations by Blue Native PAGE
As shown above (Fig. 1) our c ring preparations con-
tained various amounts of the c monomer unit and a
T. Meier et al. Structuralevidenceforaconstant c
11
ring stoichiometry
FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS 5475
number of aggregates that resisted disassembly into c
11
by SDS. To further investigate the aggregation status
of these preparations, we performed Blue Native
PAGE inthe first dimension followed by SDS ⁄ PAGE
in the second dimension. The results (Fig. 2) indicate
that the c ring prepared from I. tartaricus wild-type
cells in octylglucoside contained not only c
11
and the
stable aggregates observed in Fig. 1, but also a number
of supercomplexes corresponding to (c
11
)
n
, with n ran-
ging from two to approximately five. The most abun-
dant species was c
11
, and each supercomplex of higher
order was present in two- to threefold lower quantities
than the previous one. All of these supercomplexes dis-
assembled by SDS into the c
11
oligomer (and the SDS-
stable aggregates of c
11
, see below) as shown by the
equal mobility during SDS ⁄ PAGE (the second dimen-
sion in Fig. 2). The aggregation into supercomplexes
was prevented if the detergent octylglucoside was
replaced by Triton X-100 (Fig. 2A,B). The formation
of supercomplexes was also investigated inthe recomb-
inantly synthesized T67C mutant. Here, in addition to
higher aggregates, a (c
11
)
2
form was observed which
Fig. 1. SDS gel electrophoresis of purified c ring preparations. The
c rings from Ilyobacter tartaricus and Propionigenium modestum
were heterologously expressed in Escherichia coli and purified as
described inthe Experimental procedures. Two to three micro-
grams of each sample was subjected to SDS ⁄ PAGE and the gels
were stained with silver. The positions of the monomeric c subunit
(c
1
), the c ring (c
11
) and the c ring dimer (c
11
)
2
are marked on the
left side. Lanes 1 and 4, purified c rings from P. modestum and
I. tartaricus, respectively, isolated from the heterologous expression
cultures. Lanes 2 and 5, disintegration of c rings to the c monomer
by treatment with trichloroacetic acid. Lane 3, mutant T67C
harbouring an SDS-stable c
11
dimer. Lane 6, c ring purified from
wild-type I. tartaricus cells. A molecular mass standard is shown.
A
B
C
Fig. 2. Supercomplex formation of c
11
rings visualized by Blue
Native gel electrophoresis. Five micrograms of c ringin buffer con-
taining 10 m
M Tris ⁄ HCl, pH 8.0, and 1.5% (w ⁄ v) octylglucoside
was loaded on a Blue Native gel (5–17% acrylamide gradient), as
described inthe Experimental procedures. The samples contained
purified c ring from wild-type Ilyobacter tartaricus cells without (A)
and with (B) addition of 0.2% (v ⁄ v) Triton X-100. The c ring mutant
T67C of Propionigenium modestum was used in (C). After the run
in the first dimension, the gel lane was loaded onto a SDS gel for
the run inthe second dimension. The gels were subsequently
stained with silver. Intact c
11
ring and its supercomplexes were
marked with c
11
with the indexed numbers (n ¼ 1–4) correspond-
ing to the amount of complexed rings (c
11
)
n
. The monomeric c sub-
unit is marked with c
1
. A molecular mass standard is shown.
Structural evidenceforaconstant c
11
ring stoichiometry T. Meier et al.
5476 FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS
did not disintegrate into the c
11
oligomer by SDS, indi-
cating that a covalent disulfide bond had been formed
by the newly introduced cysteine residues.
Composition of the SDS-resistant c
11
aggregates
As described above, our c
11
ring preparations also con-
tained a distinct number of SDS-resistant complexes of
higher molecular weight, which might signify rings
with different amounts of tightly bound phospholipids.
To investigate this possibility, the c ring preparation
was incubated with phospholipase C, phospholipase
A2 or lipase, and the products were analyzed by
SDS ⁄ PAGE. The results (Fig. 3) indicate that none of
these enzymes significantly decreased the amount of
the stable c
11
aggregates. A similar observation was
made after incubating the sample for 5 min at 95 °Cin
SDS-containing loading buffer; this result confirms the
extreme stability of these aggregates. We conclude,
from these results, that the higher molecular weight of
the stable aggregates cannot be attributed to strongly
bound phospholipids. This conclusion is in agreement
with those of previous experiments, which showed that
the detergent-purified c ring contained no bound
phospholipids and the ones observed on one side of
the rings originated from the reconstitution procedure
[19].
On SDS ⁄ PAGE, the SDS-resistant aggregates migra-
ted between c
11
and (c
11
)
2
. We therefore reasoned that
these complexes might consist of c
11
with one or more
c monomers attached. To investigate this possibility,
the homogeneous c
11
ring was isolated by electro-
elution of the c
11
band excised from the SDS gel
(Fig. 4). During storage for at least 1 month, no aggre-
gates or monomeric c units were formed from pure
c
11
ring preparations. However, after addition of isola-
ted c monomers and incubation overnight, the stable
aggregates were formed again. This suggests that the c
monomer assembled with other c subunits and rings to
form a ladder of higher aggregates. To test this hypo-
thesis, higher aggregates were specifically electroeluted
from the gel and subjected to SDS ⁄ PAGE without
heat treatment. The results showed that some aggre-
gates converted to c
11
and c
1
. It may therefore be con-
cluded that c
11
and c
1
form stable aggregates and that
these aggregates are in dynamic equilibrium with c
11
and c
1
. The addition of palmitoyl-oleyl-phosphatidyl-
choline to pure c
11
did not result inthe formation of
any stable aggregates, confirming our conclusion that
these aggregates do not represent c
11
rings with bound
phospholipid molecules.
Further experiments showed that the aggregation of
c
11
and c
1
was faster at 25 °Cor37°C than at 4 °C
Fig. 4. In vitro aggregation of c
11
with c
1
to complexes resistant to
SDS. Forthe preparation of homogeneous c
11
, 1 mg of wild-type c
ring from Ilyobacter tartaricus (lane 1) was applied onto a prepara-
tive SDS gel. After the run, the c
11
band was cut out with a scalpel
and the protein was electroeluted from the gel pieces to obtain
pure c ring, as described inthe Experimental procedures (lanes 2
and 5). As a control, c-ring bands migrating more slowly were cut
out and electroeluted (lane 3). Upon incubation of 2 lg of pure c
11
with 2 lgofc
1
purified in detergent, the slower migrating band
reappeared (lanes 4 and 8). Upon incubation of 2 lg of pure c
11
with 2 and 10 lgofc
1
purified in chloroform ⁄ methanol, the slower
migrating c ring aggregates did not reappear (lanes 6 and 7). Lane
9, c
1
purified by extraction with chloroform ⁄ methanol. Lane 10, c
1
purified by sucrose density gradient centrifugation with octylgluco-
side as the detergent. Lane 11, 2 lg of c ring after incubation with
5 lg of palmitoyl-oleyl-phosphatidylcholine. A molecular mass
standard is shown.
Fig. 3. Incubation of c ring with phospholipases and lipase. The
c ring samples isolated from Ilyobacter tartaricus wild-type cells
were incubated with phospholipase C (PLC), phospholipase A2
(PLA2) and lipase (Lip), as described inthe Experimental proce-
dures, and 4 lg aliquots were loaded onto an SDS gel. The
enzymes alone were applied to separate lanes, as indicated. Also
shown is the nontreated c ring (–) and the c ring incubated at 95 °C
for 5 min. The gel was stained with silver. A molecular mass stand-
ard is shown.
T. Meier et al. Structuralevidenceforaconstant c
11
ring stoichiometry
FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS 5477
and reached approximately 90% completion after
1 day. Interestingly, the stable aggregates of c
11
and c
1
,
or of several c
1
moieties, were formed with c
1
isolated
in detergent (by sucrose density centrifugation) but not
after the extraction of c
1
with chloroform ⁄ methanol
and reconstitution into a water ⁄ detergent mixture.
These results suggest different structures forthe two
different preparations of the c monomer.
AFM of c subunit preparations
To investigate whether the heterologously expressed c
rings exhibited stoichiometries other than the previ-
ously observed undecameric composition, all purified
samples were reconstituted into lipid bilayers, as des-
cribed previously [20], and imaged by AFM. High-
resolution AFM topographs of c ring preparations
from I. tartaricus (Fig. 5) and P. modestum (Fig. 6)
showed surveys of crystalline (A) and densely packed
(B) regions of the reconstituted c subunits. The
undecameric subunit stoichiometry of the c rings was
more clearly visible inthe densely packed regions of
the unprocessed topographs. Those c rings that were
assembled into a 2D crystal exhibited an upside-down
orientation, with one oligomer neighbored by three
oligomers showing an opposite orientation. In agree-
ment with previous results, the more elevated oligo-
mers (bright white areas) protruded from the lower
and wider c rings by about 1.1 ± 0.2 nm (n ¼ 50) and
thus partly prevented the AFM stylus from contouring
the wider rings [13]. However, for statistical analyses
we performed reference-free single particle analysis of
the densely packed c rings. All classes of complete c
rings exhibited 11 subunits forming the donut-like o ligo-
mer (first image of Figs 5D and 6D). However, some
rings were incomplete, missing one or more subunits.
Compared with AFM topographs of c rings isolated
from wild-type I. tartaricus ATP synthase [13,16], the
reconstituted samples investigated inthe present study
showed more of these structural inconsistencies.
The presence of incompletely assembled c rings from
I. tartaricus and spinach chloroplast F-ATP synthases
was previously observed by AFM [16]. As the dia-
meter of the incomplete c rings did not change in any
Fig. 5. Atomic force microscopy (AFM) topographs of c subunit oligomers from Ilyobacter tartaricus F-ATPsynthase overexpressed in
Escherichia coli. The undecameric oligomers were reconstituted into the lipid bilayer and imaged in buffer solution. (A) A survey of oligomers
assembled into a 2D crystal. The donut-shaped oligomers were inserted into the membrane exposing either one of their surfaces to the
AFM stylus. (B) A survey of densely assembled oligomers. Arrows point out oligomers either missing one subunit or showing additional
subunits inside their central cavities. (C) A gallery of c rings observed from the densely packed arrangement. The first topograph represents
a reference-free single particle average obtained from more than 300 c rings. Most of the examples selected exhibit additional central pro-
trusions. (D) A gallery of c rings observed from the crystalline arrangement. Examples selected exhibit additional central protrusions. The
dashed circles with a diameter of 5.7 nm demonstrate that the outer diameters of the c rings are very consistent with each other. Topo-
graphs exhibit a gray scale corresponding to a vertical height of 3 nm.
Structural evidenceforaconstant c
11
ring stoichiometry T. Meier et al.
5478 FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS
preparation investigated, it was concluded that the
diameter of the c rings may be determined by the struc-
ture of the c monomer and not by the number of assem-
bled subunits. This finding is in agreement with our
present analysis of the defective rings overexpressed in
E. coli, which exhibited the same outer diameter
5.7 ± 0.3 nm (n ¼ 300) as the complete rings
5.6 ± 0.3 nm (n ¼ 280) within an experimental error of
0.1 nm. This structural agreement did not depend on
the number of subunits missing to complete the c ring.
The incompletely assembled c rings prepared from
chloroplasts and bacteria represented less than 5% of
all rings imaged [16]. In contrast, c rings from I. tar-
taricus (Fig. 5) or P. modestum (Fig. 6) synthesized
recombinantly in E. coli showed an increased amount
of incomplete c rings, exhibiting a total content of
8% (n ¼ 2000). Among these, c
10
,c
9
and c
8
assem-
blies represented the most abundant species. These
defective rings could probably not be observed on the
SDS gel because the detergent dissociates the less
stable c
2
to c
10
assemblies into monomeric units.
Therefore, we assume that upon insertion of the last,
11th, c subunit, the assembly becomes resistant to SDS
or heat treatment. The observed accumulation of the
incomplete c
10
complex inthe recombinant c ring
preparations suggests that the insertion of the last c
subunit forms the limiting step inthe assembly process
of a functional oligomer.
Upon closer inspection, the occurrence of additional
protrusions inthe cavity, and sometimes at the side of
some oligomers, became apparent (galleys of Figs 5
and 6). It may be assumed that these protrusions rep-
resent one or more c subunits attached to the ring-
shaped oligomer. Such a finding is in agreement with
the observation presented in Fig. 4, in which the com-
plete c subunit oligomers, hosting additional c sub-
units, migrate at higher molecular weights inthe SDS
gel electrophoresis. Furthermore, it also corresponds
to the recent observation that the analogous rotor
from chloroplast F-ATPsynthase may accommodate
small transmembrane proteins within its central cavity
[21].
Fig. 6. Atomic force microscopy (AFM) topographs of c subunit oligomers from Propionigenium modestum F-ATPsynthase overexpressed
in Escherichia coli. The oligomers were reconstituted into the lipid bilayer and imaged in buffer solution. (A) Survey of undecameric oligo-
mers assembled into a 2D crystal. The donut-shaped oligomers were inserted into the membrane exposing either one of their surfaces to
the AFM stylus. (B) A survey of densely assembled oligomers. Arrows point out oligomers either missing one subunit or showing additional
subunits inside their central cavities. (C) A gallery of c rings observed from the densely packed arrangement. The first topograph represents
a reference-free single particle average obtained from more than 200 c rings. Most of the examples selected exhibit additional central protru-
sions. (D) A gallery of c rings observed from the crystalline arrangement. Examples selected exhibit additional central protrusions. The
dashed circles with a diameter of 5.7 nm demonstrate that the outer diameter of the c rings is very consistent with each other. Topographs
exhibit a gray scale corresponding to a vertical height of 3 nm.
T. Meier et al. Structuralevidenceforaconstant c
11
ring stoichiometry
FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS 5479
Conclusion
Recently, the crystal structure of the rotor ring from
the I. tartaricus F-ATPsynthase has been solved and
provides striking details concerning the mechanism of
the F
o
motor of ATP synthase [8]. Of particular
interest in this structure is the architecture of the
Na
+
-binding site, which closely resembles that of the
k ring from the E. hirae V-ATPase [9]. As a result of
its extreme stability [22], the c
11
rotor ring from the
Na
+
-translocating F-ATPsynthase from I. tartaricus
seems to be particularly suitable forstructural investi-
gations. Fora more detailed characterization of this
system, and to increase experimental options, we have
now investigated the aggregation behavior of the c
11
ring isolated from wild-type I. tartaricus cells, and we
have explored the subunit c assembly of the protein
expressed heterologously in E. coli. Under all investi-
gated conditions, these assemblies were found to con-
sist exclusively of rings of uniform size, allowing tight
packaging of 11 monomeric units. In accordance with
previous observations, some of these rings had gaps
indicative of the absence of, in most cases, one c sub-
unit [16]. As these rings had the same diameter as
the c
11
rings, they were regarded as incompletely
assembled. In preparations derived from recombinant
E. coli cells, the incomplete assemblies were more
abundant than in preparations derived from wild-type
I. tartaricus cells. In both wild-type and recombinant
preparations, the majority of the incomplete rings
lacked only one monomer. It can therefore be conclu-
ded that the insertion of the last monomer is the lim-
iting step inthe assembly of the ring. Whether this
step, which is particularly demanding, requires a spe-
cific assembly factor, is completely unknown. A can-
didate for such a factor is the membrane insertion
protein, YidC, which was recently shown to be
required forin vitro assembly of the c ring from
E. coli [23]. Irrespective of the assembly mechanism,
our results clearly show that the size of thering is
not changed by massive overexpression of subunit c
in the E. coli host cells, indicating that intrinsic fea-
tures of the monomeric unit determine the number of
subunits that can be packed into the ring. These data
are fully compatible with the recent finding that the
stoichiometry of the subunit III cylinder within the
ATP synthase of the green algae, C. reinhardtii,is
not affected by the metabolic state of the cells [18].
However, such findings are difficult to reconcile with
the proposed variation of c ring stoichiometries in
E. coli, which are dependent on the expression level
or the nutritional status of the cells [15].
Knowledge on the aggregation behavior of the c ring
has been of considerable value in exploring suitable
crystallization conditions for structure determination.
Two types of aggregates had to be taken into account.
The first were supercomplexes of the (c
11
)
n
type. The
formation of these supercomplexes is dependent on the
detergent because they are formed in octylglucoside,
but not in Triton X-100. These supercomplexes dis-
aggregate completely into the c
11
rings inthe presence
of SDS. The second type of aggregate appears as a
ladder above the original c
11
band on SDS ⁄ PAGE and
consists of c
11
rings hosting varying amounts of the c
monomer. Aggregates are particularly abundant in
c ring preparations from E. coli expression clones
where the c monomer is present in high amounts. Once
these aggregates were formed they remained stable and
were minimally influenced by additives such as deter-
gents, organic solvents, salts or lipids (like 1-palmitoyl-
2-oleyl-sn-glycero-3-phosphocholine). AFM topographs
of these samples showed an exclusively undecameric
stoichiometry inthe completely assembled rings. This
is also observed inthe noncrystalline areas of the
reconstituted vesicles, demonstrating that it is not an
artifact from the 2D crystallization.
Slower migrating bands of c rings, as observed on
SDS gels, suggest that a certain fraction of c rings may
host additional subunits. The AFM topographs indi-
cate that these additional subunits may be hosted at the
outer sides and within the central cavities of the rings.
That these bands are composed exclusively of c sub-
units has been proven by biochemical analyses and, in
addition, the formation of these aggregates from pure
c
11
and c
1
has been demonstrated inthe present study.
Experimental procedures
Materials
Chemicals were purchased from Fluka (Buchs, Switzerland)
including lipase from Aspergillus oryzae. N-Lauroylsarcosine
sodium salt and n-octyl-beta-d-glucopyranoside were pur-
chased from Sigma (Buchs, Switzerland) and Glycon
Biochemicals (Luckenwalde, Germany), respectively. Primers
were custom synthesized by Microsynth (Balgach, Switzer-
land). Phospholipase A2 from hog pancreas, and phospho-
lipase C from Bacillus cereus, were purchased from Sigma
(St Louis, MO, USA).
Construction of plasmid pt7cIT
The atpE gene from I. tartaricus [24] was amplified with
Pfu polymerase and the following two primers:
Structural evidenceforaconstant c
11
ring stoichiometry T. Meier et al.
5480 FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS
5¢-GGAGGAAATAAGCATATGGATATG-3¢ (forward),
containing an NdeI site, and 5¢-CCTTTCAGGAAGCT
TCCTCC-3¢ (reverse), containing a HindIII site. The PCR
product and plasmid pt7-7 [25] were digested with these
two restriction enzymes and ligated before transformation
into E. coli DH5a. Plasmid pt7c [26] was mutagenized with
the Quick Change Site Directed Mutagenesis Kit (Strata-
gene, La Jolla, CA, USA) to yield the single mutation,
T67C, inthe P. modestum subunit c (plasmid pt7cT67C).
Synthesis and purification of c oligomers
from strain BL21(DE3) transformed with various
plasmids
E. coli BL21(DE3) (Novagen, Madison, WI, USA) was
transformed with plasmids pt7c, pt7cIT and pt7cT67C, as
described above. The transformed E. coli cells were grown
in 2 L of Luria–Bertani (LB) medium to reach an attenua-
nce (D) of 0.6 at 37 °C inthe presence of 200 lgÆmL
)1
ampicillin. After cooling on ice for 5 min, the expression
was induced with 0.7 mm isopropyl thio-b-d-galactoside
and allowed to continue for 6 h at 30 °C, to yield typically
2.5 g of cells per L of medium. The cells (1 g wet weight)
were suspended in 8 mL of 50 mm potassium phosphate
buffer, pH 8.0, containing 1 mm 1,4-dithio-dl-threitol,
0.1 mm diisopropylfluorophosphate and a spatula tip of
DNaseI. Preparation of membranes was performed at 4 °C.
The cell suspension was passed twice through a French
pressure cell at 12 000 psi (8.3 · 10
4
kPa). After the
removal of cell debris by centrifugation at 15 000 g for
20 min, ultracentrifugation was performed at 200 000 g for
60 min. The membrane pellet was washed once with 4 mL
of 20 mm Tris, 5 mm EDTA, and then adjusted to pH 8.0
with HCl. Solubilization of the membranes was accom-
plished with 2 mL of 20 mm Tris ⁄ HCl, pH 8.0, containing
5mm EDTA and 1% (w ⁄ v) N-lauroylsarcosine for 10 min
at 65 °C. After ultracentrifugation at room temperature,
the pellet was discarded and contaminating membrane pro-
teins were precipitated with (NH
4
)
2
SO
4
at 65% (w ⁄ v) sat-
uration. After 20 min of incubation at 20 °C, the sample
was centrifuged for 20 min at 39 000 g. The filtrated super-
natant containing the c oligomer was dialysed against 5 L
of 10 mm Tris buffer, which was adjusted to pH 8.0 with
HCl using a dialysis membrane with a molecular cut-off of
6000 Da.
The protein sample was concentrated by ultrafiltration
with Centricon tubes YM-10 (Millipore, Billerica, MA,
USA) to a concentration of 1 mgÆmL
)1
and applied to the
top of a density gradient (5 mL) of 5–30% sucrose con-
taining 20 mm Tris ⁄ HCl, pH 8.0, 10 lm 1,4-dithio-dl-thre-
itol and 1% (w ⁄ v) octylglucoside. After ultracentrifugation
(4 °C, 16 h, 150 000 g) ina Beckman SW55-Ti rotor
(Beckman, Coulter, Inc., Fullerton, CA, USA), fractions
of 0.5 mL were collected from the top and analysed by
SDS ⁄ PAGE [27]. The c ring-containing samples were
pooled and concentrated by ultracentrifugation (18 h,
200 000 g,4°C). The final protein concentration was typ-
ically between 1.5 and 3 mgÆmL
)1
. Fractions containing
the monomeric c subunit were also collected and used to
study the association with c
11
rings to stable c
11
(c
1
)
n
aggregates.
Reconstitution of densely packed and 2D
crystalline c ring samples
The c rings purified from these expression cultures were
crystallized in two dimensions by mixing octylglucoside-sol-
ubilized protein with 1 mgÆmL
)1
1-palmitoyl-2-oleyl-sn-gly-
cero-3-phosphocholine at a lipid : protein ratio of 0.8
(w ⁄ w) ina total volume of 50 lL, followed by dialysis for
24 h at 25 ° C against 200 mL of buffer (10 mm Tris ⁄ HCl,
pH 7.5, containing 200 mm NaCl and 0.02% NaN
3
), then
for another 24 h at 37 °C. The crystals were stored at 4 °C
for further analysis. Subunit c monomers solubilized in
chloroform ⁄ methanol were purified according to the proce-
dure described previously [26].
Purification of pure c
11
without supercomplexes
and attached monomers
The c ring was isolated from wild-type I. tartaricus cells as
previously described [20]. One milligram of the protein was
loaded onto a preparative SDS-containing gel, according to
Scha
¨
gger et al. [27], together with a prestained marker.
After the run, the c
11
ring was visible, without staining, as
a result of the high local protein concentration. The band
was excised from the gel with a scalpel and subjected to
electroelution (at 25 mA) for 6 h at 4 °C.
Phospholipase A
2
, phospholipase C and lipase
digestions
Eighty micrograms of purified subunit c
11
ring from I. tar-
taricus [20] was incubated for 14 h at 37 °C with 10 U of
phospholipase A2, 2 U of phospholipase C or 5 U of
lipase, inthe presence of 50 mm Tris ⁄ HCl buffer with a pH
adjusted to 8.0, 7.2 or 8.0, respectively, and 1.5% (w ⁄ v)
octylglucoside.
Blue Native PAGE
Blue Native PAGE was carried out as described previously
[28]. Separation gels with a linear gradient of 5–17% acryl-
amide were prepared and overlayed with 4% sample gels.
Samples of 2–5 lg protein each were mixed with sample
buffer [50 mm Tris ⁄ HCl, pH 6.8, containing 12% (v ⁄ v) gly-
cerol and 0.01% (w ⁄ v) Serva blue G]. After running for 1 h
at 100 V with cathode buffer (50 mm Tricine, 15 mm Bis-
Tris ⁄ HCl, pH 7.0) containing 0.02% (w ⁄ v) Serva blue G,
T. Meier et al. Structuralevidenceforaconstant c
11
ring stoichiometry
FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS 5481
the cathode buffer was replaced with buffer containing only
0.002% (w ⁄ v) Serva blue G and the run continued at
400 V. Native protein complexes were then analyzed by
SDS ⁄ PAGE, as described previously [27], with the lanes
from the Blue Native PAGE embedded into a 4% stacking
gel.
Atomic force microscopy
The samples were diluted to a concentration of 10
lgÆmL
)1
in 200 mm NaCl, 10 mm Tris ⁄ HCl, pH 7.5.
To allow adsorption of the membranes, a drop of 30 lL
was placed onto freshly cleaved mica. After an adsorption
time of 15 min, the sample was gently washed using the
above buffer solution containing no membrane proteins to
remove weakly attached material from the mica surface.
Contact mode AFM topographs were then recorded in the
same buffer, at room temperature, at forces of < 100 pN
applied to the AFM stylus, and at scanning line frequencies
of typically 4–6 Hz. The AFM used was a Nanoscope E
(Digital Instruments, Santa Barbara, CA, USA) equipped
with a 120 lm piezo scanner and a fluid cell. Cantilevers
(Olympus, Tokyo, Japan) had oxide-sharpened Si
3
N
4
tips
and a spring constant of 0.09 NÆm
)1
. No differences
between topographs recorded simultaneously in trace and
in the retrace direction were observed, indicating that the
scanning process did not influence the appearance of the
biological sample.
AFM data analysis and image processing
Individual particles of the AFM topographs were selected
manually and subjected to reference-free alignment and
averaging using the SPIDER image processing system
(Wadsworth Laboratories, New York, NY, USA). Refer-
ence-free averages generated by translational and rotational
alignment of single particles enhanced common structural
features among the c oligomers (Figs 5 and 6). To assess
the rotor symmetry, the rotational power spectrum of the
averaged image was calculated using the semper image
processing system (Synoptics Ltd, Cambridge, UK). Alter-
natively, the rotational power spectrum of each individual
particle was calculated and then averaged (data not
shown). It appeared that all averaged classes showed a stoi-
chiometry of 11 subunits except for those of defect parti-
cles. The diameter of intact and defective c rings was
determined as described previously [16].
Other methods
Gels were stained with silver [29]. The protein concentra-
tion of samples was determined according to the bicinchon-
inic acid method [30] with bovine serum albumin as the
standard.
Acknowledgements
The authors thank Marijke Koppenol for critically
reading the manuscript. This work was supported by
the free state of Saxiona, the European community,
and the Deutsche Forschungsgemeinschaft (DFG).
References
1 Capaldi RA & Aggeler R (2002) Mechanism of the
F(1)F(0)-type ATP synthase, a biological rotary motor.
Trends Biochem Sci 27, 154–160.
2 Abrahams JP, Leslie AG, Lutter R & Walker JE (1994)
Structure at 2.8 A
˚
resolution of F
1
-ATPase from bovine
heart mitochondria. Nature 370, 621–628.
3 Boyer PD (1993) The binding change mechanism for
ATP synthase – some probabilities and possibilities.
Biochim Biophys Acta 1140, 215–250.
4 Noji H, Yasuda R, Yoshida M & Kinosita K (1997)
Direct observation of the rotation of F
1
-ATPase. Nature
386, 299–302.
5 Dimroth P, von Ballmoos C, Meier T & Kaim G (2003)
Electrical power fuels rotary ATP synthase. Structure
(Camb) 11, 1469–1473.
6 Feniouk BA, Kozlova MA, Knorre DA, Cherepanov
DA, Mulkidjanian AY & Junge W (2004) The of ATP
synthase: ohmic conductance (10 fS), and absence of
voltage gating. Biophys J 86, 4094–4109.
7 Aksimentiev A, Balabin IA, Fillingame RH & Schulten
K (2004) Insights into the molecular mechanism of
rotation inthe F
o
sector of ATP synthase. Biophys J 86,
1332–1344.
8 Meier T, Polzer P, Diederichs K, Welte W & Dimroth P
(2005) Structure of the rotor ring of F-Type Na
+
-ATPase
from Ilyobacter tartaricus . Science 308, 659–662.
9 Murata T, Yamato I, Kakinuma Y, Leslie AG &
Walker JE (2005) Structure of the rotor of the V-Type
Na
+
-ATPase from Enterococcus hirae. Science 308,
654–659.
10 Mellwig C & Bo
¨
ttcher B (2003) A unique resting posi-
tion of the ATP-synthase from chloroplasts. J Biol
Chem 278, 18544–18549.
11 Stock D, Leslie AG & Walker JE (1999) Molecular
architecture of the rotary motor in ATP synthase.
Science 286, 1700–1705.
12 Seelert H, Poetsch A, Dencher NA, Engel A, Stahlberg
H&Mu
¨
ller DJ (2000) Proton-powered turbine of a
plant motor. Nature 405, 418–419.
13 Stahlberg H, Mu
¨
ller DJ, Suda K, Fotiadis D, Engel A,
Meier T, Matthey U & Dimroth P (2001) Bacterial
Na
+
-ATP synthase has an undecameric rotor. EMBO
Rep 2, 229–233.
14 Pogoryelov D, Jinshu J, Meier T, Dimroth P & Muller
DJ (2005) The c
15
ring of the Spirulina platensis F-ATP
Structural evidenceforaconstant c
11
ring stoichiometry T. Meier et al.
5482 FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS
synthase: F
1
⁄ F
0
symmetry mismatch is not obligatory.
EMBO Rep.
15 Schemidt RA, Qu J, Williams JR & Brusilow WS
(1998) Effects of carbon source on expression of F
0
genes and on thestoichiometry of the c subunit in the
F
1
F
0
ATPase of Escherichia coli. J Bacteriol 180, 3205–
3208.
16 Mu
¨
ller DJ, Dencher NA, Meier T, Dimroth P, Suda K,
Stahlberg H, Engel A, Seelert H & Matthey U (2001)
ATP synthase: constrained stoichiometry of the trans-
membrane rotor. FEBS Lett 504, 219–222.
17 Arechaga I, Butler PJ & Walker JE (2002) Self-assembly
of ATP synthase subunit c rings. FEBS Lett 515, 189–
193.
18 Meyer Zu Tittingdorf JM, Rexroth S, Scha
¨
fer E, Schl-
ichting R, Giersch C, Dencher NA & Seelert H (2004)
The stoichiometry of the chloroplast ATP synthase
oligomer III in Chlamydomonas reinhardtii is not
affected by the metabolic state. Biochim Biophys Acta
1659, 92–99.
19 Meier T, Matthey U, Henzen F, Dimroth P & Mu
¨
ller
DJ (2001) The central plug inthe reconstituted undeca-
meric c cylinder of a bacterial ATP synthase consists of
phospholipids. FEBS Lett 505, 353–356.
20 Meier T, Matthey U, von Ballmoos C, Vonck J, Krug
von Nidda T, Ku
¨
hlbrandt W & Dimroth P (2003) Evi-
dence forstructural integrity inthe undecameric c-rings
isolated from sodium ATP synthases. J Mol Biol 325,
389–397.
21 Seelert H, Dencher NA & Mu
¨
ller DJ (2003) Fourteen
protomers compose the oligomer III of the proton-rotor
in spinach chloroplast ATP synthase. J Mol Biol 333,
337–344.
22 Meier T & Dimroth P (2002) Intersubunit bridging by
sodium ions as rationale forthe unusual stability of the
turbines of Na
+
-F
1
F
o
ATP synthases. EMBO Rep 3,
1094–1098.
23 Van der Laan M, Bechtluft P, Kol S, Nouwen N &
Driessen AJ (2004) F
1
F
0
ATP synthase subunit c is a
substrate of the novel YidC pathway for membrane
protein biogenesis. J Cell Biol 165, 213–222.
24 Meier T, Neumann S, Dimroth P & Kaim G (2002)
DNA sequence of the entire atp-operon encoding the
sodium dependent F
1
F
o
ATP synthase from Ilyobacter
tartaricus. Biochim Biophys Acta 1625, 221–226.
25 Tabor S & Richardson CC (1985) A bacteriophage T7
RNA polymerase ⁄ promoter system for controlled exclu-
sive expression of specific genes. Proc Natl Acad Sci
USA 82, 1074–1078.
26 Matthey U, Kaim G, Braun D, Wu
¨
thrich K & Dimroth
P (1999) NMR studies of subunit c of the ATP synthase
from Propionigenium modestum in dodecylsulfate
micelles. Eur J Biochem 261, 459–467.
27 Scha
¨
gger H & Jagow G (1987) Tricine-sodium dodecyl
sulfate-polyacrylamide gel electrophoresis for the
separation of proteins inthe range from 1 to 100 kDa.
Anal Biochem 166, 368–379.
28 Scha
¨
gger H & von Jagow G (1991) Blue native electro-
phoresis for isolation of membrane protein complexes
in enzymatically active form. Anal Biochem 199,
223–231.
29 Nesterenko MV, Tilley M & Upton SJ (1994) A simple
modification of Blum’s silver stain method allows for 30
minute detection of proteins in polyacrylamide gels.
J Biochem Biophys Methods 28, 239–242.
30 Smith PK, Krohn RI, Hermanson GT, Mallia AK,
Gartner FH, Provenzano MD, Fujimoto EK, Goeke
NM, Olson BJ & Klenk DC (1985) Measurement of
protein using bicinchoninic acid [Erratum in: Anal
Biochem (1987) 163, 279]. Anal Biochem 150, 76–85.
T. Meier et al. Structuralevidenceforaconstant c
11
ring stoichiometry
FEBS Journal 272 (2005) 5474–5483 ª 2005 FEBS 5483
. the amount of
the stable c
11
aggregates. A similar observation was
made after incubating the sample for 5 min at 95 °Cin
SDS-containing loading buffer;. the retrace direction were observed, indicating that the
scanning process did not in uence the appearance of the
biological sample.
AFM data analysis and