An intermediate step in the evolution of ATPases – a
hybrid F
0
–V
0
rotor in a bacterial Na
+
F
1
F
0
ATP synthase
Michael Fritz
1,
*, Adriana L. Klyszejko
2,
*, Nina Morgner
3,
*, Janet Vonck
4
, Bernd Brutschy
3
,
Daniel J. Muller
2
, Thomas Meier
4
and Volker Mu
¨
ller
1

covalently linked subunits, and rotation of the c ring is
obligatorily coupled to ion flow across the membrane
[4–6].
Subunit c of the F
1
F
0
ATP synthases has a molecular
mass of approximately 8 kDa, and folds in the mem-
brane like a hairpin, with two transmembrane helices
connected by a cytoplasmic loop [7]. Each monomer
contains an ion-binding site (H
+
or Na
+
) [8,9]. Recent
studies have demonstrated that the c ring stoichiometry
in different organisms ranges between 10 and 15 mono-
mers (see Discussion). Assuming that each subunit
takes up one ion, each c ring revolution induces
the synthesis of three molecules of ATP. This gives a
Keywords
Acetobacterium; acetogen; ATP-synthase;
c ring; F
0
-V
0
hybrid rotor
Correspondence
V. Mu

0
-like (c
1
)
c subunit, with four transmembrane helices but only one binding site. To
determine whether both types of rotor subunits are present in the same
c ring, we have isolated and studied the composition of the c ring. High-
resolution atomic force microscopy of 2D crystals revealed 11 domains,
each corresponding to two transmembrane helices. A projection map
derived from electron micrographs, calculated to 5 A
˚
resolution, revealed
that each c ring contains two concentric, slightly staggered, packed rings,
each composed of 11 densities, representing 22 transmembrane helices.
The inner and outer diameters of the rings, measured at the density bor-
ders, are approximately 17 and 50 A
˚
. Mass determination by laser-
induced liquid beam ion desorption provided evidence that the c rings
contain both types of c subunits. The stoichiometry for c
2
⁄ c
3
: c
1
was
9 : 1. Furthermore, this stoichiometry was independent of the carbon
source of the growth medium. These analyses clearly demonstrate, for the
first time, an F
0

the Faraday constant.
The c subunit of the eukaryal V
1
V
0
ATPases present
in organelles arose by duplication and fusion of the
bacterial c subunit, giving rise to a protein of approxi-
mately 16 kDa with four transmembrane helices that
form two covalently linked hairpins in the membrane
[10]. Importantly, the ion-binding site is not conserved
in hairpin one. If one assumes the same number of
transmembrane helices in V
0
and F
0
, the rotor of
eukaryal V
1
V
0
ATPases has only half the number
of ion-binding sites compared to F
1
F
0
ATP syntheses.
This low H
+
(Na

0
subunit c gene (atpE
1
) and two
genes (atpE
2
⁄ atpE
3
) encoding identical F
0
c subunits.
The gene atpE
1
encodes an 18 kDa protein with two
predicted hairpins, and, like its eukaryotic counter-
part, is missing one ion binding site (in hairpin two).
The genes atpE
2
and atpE
3
encode two identical
8 kDa subunits with one ion-binding site each. The
three genes are encoded in the same operon and their
products are present in the same enzyme preparation
[14–16]. Here, we have addressed the question
whether both types of c subunit assemble into one
ring, and whether the c ring composition changes
with the growth conditions. We present data that
unequivocally demonstrate a V
0

Na
+
F
1
F
0
ATP synthases [18,19] as well as isolated
c rings [20], the A. woodii c ring was highly stable and
did not dissociate by boiling in 20 mm Tris ⁄ HCl, 5%
SDS (pH 8.0) for up to 30 min, but did dissociate by
autoclaving (120 °C) in the presence of 5% SDS for
5 min or in the presence of 40 mm trichloroacetic acid.
The c rings from different preparations always showed
the same migration behaviour in SDS–PAGE, and the
isolated c ring migrated to a position identical to that
of the c ring present in native ATP synthase solubi-
lized in the same detergent (Fig. 1).
Fig. 1. Isolation and subunit composition of
the c rings from the Na
+
F
1
F
0
ATP synthase
from A. woodii. Samples of isolated enzyme
(lane 1) and isolated c rings (lanes 3, 4 and
7) were boiled at 80 °C for 20 min and
applied to a 10.0% (lanes 1–3) or 13.5%
(lanes 4–9) polyacrylamide gel. The c ring

(AFM). AFM topographs revealed crystalline and
paracrystalline membrane patches surrounded by the
lipid bilayer (Fig. 2). c rings with an outer diameter
of 5.8 ± 0.4 nm (n = 125) were surrounded by smal-
ler c rings of diameter 5.4 ± 0.4 nm (n = 150). In
agreement with previous measurements on c rings
from other F
1
F
0
ATP synthases [21–23], the occur-
rence of two diameters indicated that the reconsti-
tuted c rings had an ‘upside-down’ orientation in the
membrane and that we imaged both ring surfaces. In
further agreement with previous measurements, the
smaller c rings exhibited central protrusions that were
shown to represent lipid headgroups [24] and to
reflect the extracellular side of the ring [8,24].
Whereas the lipid bilayer exhibited a height of
4.5 ± 0.5 nm (n = 10), the proteins protruded
7.7 ± 0.5 nm (n = 10) from the supporting mica sur-
face. At a lateral resolution of approximately 1 nm,
the subunits of individual ring-shaped c oligomers
became visible (Fig. 2A,B). Cross-correlation averages
applied to further enhance common structural fea-
tures showed different assemblies of the c rings, each
being composed of 11 equally sized domains
(Fig. 2C,D). Similarly, the reference-free averages
generated by translational and rotational alignment
of single c rings showed the same stoichiometry

of the rings face in the opposite direction in the
membrane to the other two, forming the same pattern
as in the AFM surface representation of Fig. 2A. By
comparison with the 3D structure of the I. tartaricus
c ring [24], the black rings represent the view from
the cytoplasm (open rings in AFM) and the red ones
the view from the extracellular side (smaller, closed
rings in Fig. 3). The inner and outer diameters of the
rings, measured at the density borders, are approxi-
mately 17 and 50 A
˚
. However, the resolution
obtained did not enable us to distinguish c
1
from
c
2 ⁄ 3
.
Subunit composition of the c ring from A. woodii
The above structural analyses clearly assigned 22
transmembrane helices to the c ring of A. woodii.To
unravel the c
1
and c
2 ⁄ 3
subunit composition of the
potential hetero-oligomeric ring, we used laser-induced
Fig. 2. High-resolution AFM topographs of reconstituted c rings.
(A, B) Crystalline assemblies of c rings. Although the number of
subunits forming the rings can be seen, the signal-to-noise ratio

attached to the ring under the ultra-soft desorption
process [27]. The overall mass of the c ring was deter-
mined to be 93.5 ± 0.1 kDa. Harsher desorption con-
ditions, achieved by increasing the laser intensity, led
to the detachment of detergent and water molecules.
Moreover, additional energy was transferred into the
system, and the c rings (partly) dissociated into single
subunits and subcomplexes. The mass spectrum in
Fig. 4B was used to determine the c
1
to c
2 ⁄ 3
stoichio-
metry. The peak distribution contains two series of
subcomplexes. One series corresponds to subcomplex-
es containing only c
2 ⁄ 3
units of the form (c
2 ⁄ 3
)
n
,
where n = 1–5, the other is built up from c
1
and c
2 ⁄ 3
units in the form c
1
(c
2 ⁄ 3

and contains four c rings.
Hybrid V
0
–F
0
rotor in a F
1
F
0
ATP synthase M. Fritz et al.
2002 FEBS Journal 275 (2008) 1999–2007 ª 2008 The Authors Journal compilation ª 2008 FEBS
Does the subunit composition of the c ring from
A. woodii vary with the carbon source?
As outlined above, the unique presence of a eukaryal
V
0
-like c subunit in a bacterial ATP synthase raised
the question whether the stoichiometry of V
0
:F
0
-like
subunits may be flexible and thus a mechanism to
change the action of the enzyme from ATP synthase
to ATPase. To address potential variation in c ring
subunit composition depending on the growth condi-
tions, cells were grown under autotrophic conditions
(ATP synthase required) or heterotrophic fermenting
conditions (ion-pumping ATPase function required),
and c rings were purified and subjected to LILBID

A
0
,V
1
V
0
) stoichiometry [28]. The uncertainty lay
in the number of ion-translocating subunits in the
membrane-embedded rotor. Recently, the atomic struc-
ture of a c ring from the F
1
F
0
ATP synthase from
I tartaricus was solved and revealed 11 monomers [8].
Interestingly, on the basis of structural, biochemical
and genetic studies, the c ring stoichiometry in F
1
F
0
ATP synthases is apparently variable among species.
Ten monomers are found in c rings from the F
1
F
0
ATP synthases from yeast, Escherichia coli or Bacillus
PS3 [29–31], undecameric rings were found in I. tar-
taricus [22], Propionigenium modestum [17] and Clos-
tridium paradoxum [32] Na
+

been found to encode F
0
and V
0
c subunit genes [14].
The genes have been found to be expressed [15] and
the subunits have been found in the purified enzyme
[16]. However, a critical question that was solved here
was whether both subunits are part of one rotor or
whether there are two populations of enzymes, one
having only the F
0
-like c subunit and the other only
Fig. 4. Mass spectra of the c ring taken
under various laser desorption conditions.
Under ultrasoft desorption conditions (A),
the c ring is detected unfragmented with a
charge distribution of one to four as indi-
cated by red vertical bars. The broadening
of the peaks towards higher masses is due
to the attachment of detergent and water
molecules. Under harsh desorption condi-
tions (B), the c ring is fragmented, which
leads to two series of subcomplexes con-
taining only c
2 ⁄ 3
subunits (indicated by blue
vertical bars) or one c
1
subunit and 1–9 c

showed no peaks that contained two or more c
1
sub-
units, excluding the existence of more than one
c
1
unit per ring and of course rings formed from c
1
only. No mass was detectable corresponding to a ring
made by c
2 ⁄ 3
subunits only or more than nine c
2 ⁄ 3
subunits.
The stoichiometry of the subunits in the c ring of
A. woodii was determined to be 1 : 9 (c
1
: c
2 ⁄ 3
), with a
total of 22 transmembrane helices. This value is identi-
cal to the value obtained for the other Na
+
F
1
F
0
ATP
synthases. Additionally, the size of the A. woodii
c rings (approximately 58 A

across the membrane) and occasionally bind to and
thus compensate the negative charge on the rotor. The
freed positive charge on the stator attracts the next
negative charge on the rotor, thus leading to rotation
of the ring.
As most of the c rings investigated so far have a
number of monomers that cannot be divided by three,
this implies that the translocated ion to ATP ratio is
not an integer. It has been suggested that an elastic
power transmission between F
1
and F
0
is important
for operation of the enzyme under symmetry mismatch
conditions [37]. In the enzyme from A. woodii, rotation
of the c ring over each phase of 120° is coupled to at
least two different numbers of ions. Obviously, the
force that has to be applied to overcome the spatial
difference between three hairpins (c
2 ⁄ 3
)
–c
1,N-term
)
c
1,C-term
neutral
– c
2 ⁄ 3

tionable but remains to be addressed experimentally.
The electrochemical ion potential across the cyto-
plasmic membrane of A. woodii has not yet been
determined due to high nonspecific binding of the
radioactive probes, but it is reasonable to assume that
is similar to that in other bacteria, i.e. in the range of
)180 to )200 mV. Therefore, the enzyme will work as
an ATP synthase under physiological conditions. Its
capability to synthesize ATP despite the presence of
the V
0
-like c subunit has been demonstrated very
recently in a proteoliposome system [16].
Experimental procedures
Growth of cells and isolation of membranes
A. woodii (DSM 1030) was grown in 20-1iter fermentors to
mid-exponential growth phase as described previously [43].
Fructose (20 mm), betaine (40 mm), methanol (60 mm)or
formate (80 mm) were used as carbon and energy sources.
The NaCl concentration was 20 m m, unless otherwise
stated. The ATP synthase was purified to apparent homo-
geneity by solubilization with 1% dodecyl- b -d-maltoside
followed by chromatography as described previously [16].
All preparations were routinely analyzed by SDS–PAGE
using the buffer system described by Scha
¨
gger and von
Jagow [44]. Polypeptides were visualized by staining with
Coomassie brilliant blue [45] or silver [46]. The protein
concentration of samples was determined according to the

in a first step by filtration (filter paper, 2.6 lm pore size,
Schleicher & Schuell, Dassel, Germany) followed by filtra-
tion through a 0.2 lm filter (4 mm syringe filters, Nalgene,
Rochester, NY, USA). The filtrate was dialyzed overnight
at 4 °C against 10 mm Tris ⁄ HCl, 200 mm NaCl, pH 8.0,
followed by addition of b-octylglycoside (Biomol, Mu
¨
n-
chen, Germany) to a final concentration of 1.5%. To fur-
ther concentrate the c rings and to remove excess salt, the
sample was loaded onto an Amicon Ultra-4 tube
(30 000 Da molecular mass cut-off; Amicon, Hanover,
Germany) and concentrated to about 2–4 mgÆmL
)1
.
Western blot analysis
After separation by SDS–PAGE, the ATP synthase subun-
its were blotted onto a nitrocellulose membrane as
described previously [48]. Western blot enhanced chemilu-
minescence (ECL) detection reagents were either purchased
from Perkin Elmer Life Sciences (Boston, MA, USA) or
produced in our laboratory. Blot membranes were incu-
bated in a mixture of 4 mL of solution A (0.1 m Tris ⁄ HCl,
pH 6.8, 50 mg luminol in a total volume of 200 mL),
400 lL of solution B (11 mg p-hydroxycoumaric acid in
10 mL dimethylsulfoxide) and 1.2 lLofH
2
O
2
for 2 min

, 10% glycerol, pH 7.8) was placed onto
freshly cleaved mica for about 30 min. Then the sample
was rinsed with dialysis buffer to remove weakly attached
material. Contact-mode AFM topographs were recorded in
dialysis buffer at 25 °C, with a loading force of approxi-
mately 100 pN and a line frequency of 4–6 Hz. No differ-
ences between topographs recorded in the trace and retrace
directions were observed, indicating that the scanning pro-
cess did not influence the appearance of the sample. For
image processing, individual particles of the AFM topo-
graphs were subjected to reference-free alignment and
averaging using the SPIDER image processing system
(Wadsworth Labs, New York, NY, USA). Correlation
averages were calculated using the SEMPER image process-
ing system (Synoptics Ltd, Cambridge, UK). To assess the
rotor symmetry, the rotational power spectra of reference-
free averages and of single rotors were calculated.
Electron microscopy and image processing
Two-dimensional crystal samples were prepared in 4.5%
w ⁄ v trehalose on molybdenum grids (Pacific Grid-Tech,
San Diego, CA, USA) by the back-injection method. Grids
were examined in a JEOL 3000 SFF helium-cooled electron
microscope (JEOL Ltd., Tokyo, Japan) at 4 K at an accel-
erating voltage of 300 kV. Images were recorded by a spot-
scanning procedure, using 24 spots by 30 spots per image
on Kodak SO-163 film (Kodak, Stuttgart, Germany) at a
magnification of 53 000 · and with an electron dose of 20–
30 electrons ⁄ A
˚
2

M. Fritz et al. Hybrid V
0
–F
0
rotor in a F
1
F
0
ATP synthase
FEBS Journal 275 (2008) 1999–2007 ª 2008 The Authors Journal compilation ª 2008 FEBS 2005
SFB 579 to BB, Cluster of Excellence ‘Macromolecular
Complexes’ Project EXC 115 to TM), the Fonds der
chemischen Industrie (to BB), and the EU (grant
NEST2004 PathSYS29084 to DM). We thank Deryck
Mills for assistance with electron microscopy and
Werner Ku
¨
hlbrandt (MPi of Biophysics, Frankfurt,
Germany) for comments on the manuscript.
References
1 Boyer PD (1997) The ATP synthase – a splendid molec-
ular machine. Annu Rev Biochem 66, 717–749.
2 Walker JE (1994) The regulation of catalysis in ATP
synthase. Curr Opin Struct Biol 4, 912–918.
3 Yoshida M, Muneyuki E & Hisabori T (2001) ATP
synthase – a marvellous rotary engine of the cell. Nat
Rev Mol Cell Biol 2, 669–677.
4 Sambongi Y, Iko Y, Tanabe M, Omote H, Iwamoto-
Kihara A, Ueda I, Yanagida T, Wada Y & Futai M
(1999) Mechanical rotation of the c subunit oligomer in

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
+
-AT-
Pase from Enterococcus hirae. Science 308, 654–659.
10 Mandel M, Moriyama Y, Hulmes JD, Pan Y-CE, Nel-
son H & Nelson N (1988) cDNA sequence encoding the
16-kDa proteolipid of chromaffin granules implies gene
duplication in the evolution of H
+
-ATPases. Proc Natl
Acad Sci USA 85, 5521–5524.
11 Nelson N & Taiz L (1989) The evolution of H
+
-ATPas-
es. Trends Biochem Sci 14, 113–116.
12 Cross RL & Mu
¨
ller V (2004) The evolution of A-, F-,
and V-type ATP synthases and ATPases: reversals in
function and changes in the H
+
⁄ ATP stoichiometry.
FEBS Lett 576, 1–4.
13 Forgac M (1999) Structure and properties of the vacuo-
lar H
+
-ATPases. J Biol Chem 274, 12951–12954.
14 Rahlfs S, Aufurth S & Mu

ller V (2007) An intermediate step in the
evolution of ATPases – the F
1
F
0
-ATPase from Aceto-
bacterium woodii contains F-type and V-type rotor
subunits and is capable of ATP synthesis. FEBS J 274,
3421–3428.
17 Meier T, Matthey U, von Ballmoos C, Vonck J, Krug
von Nidda T, Ku
¨
hlbrandt W & Dimroth P (2003) Evi-
dence for structural integrity in the undecameric c-rings
isolated from sodium ATP synthases. J Mol Biol 325,
389–397.
18 Reidlinger J & Mu
¨
ller V (1994) Purification of ATP
synthase from Acetobacterium woodii and identification
as a Na
+
-translocating F
1
F
0
-type enzyme. Eur J Bio-
chem 223, 275–283.
19 Laubinger W & Dimroth P (1989) The sodium translo-
cating adenosintriphosphatase of Propionigenium mode-

¨
ller DJ (2005) The c
15
ring of the Spirulina platensis
F-ATP synthase: F
1
⁄ F
0
symmetry mismatch is not
obligatory. EMBO Rep 6, 1040–1044.
24 Vonck J, von Nidda TK, Meier T, Matthey U, Mills
DJ, Ku
¨
hlbrandt W & Dimroth P (2002) Molecular
architecture of the undecameric rotor of a bacterial
Na
+
-ATP synthase. J Mol Biol 321, 307–316.
25 Morgner N, Kleinschroth T, Barth HD, Ludwig B &
Brutschy B (2007) A novel approach to analyze mem-
brane proteins by laser mass spectrometry: from protein
subunits to the integral complex. J Am Soc Mass Spec-
trom 18, 1429–1438.
26 Meier T, Morgner N, Matthies D, Pogoryelov D,
Keis S, Cook GM, Dimroth P & Brutschy B (2007)
A tridecameric c ring of the adenosine triphosphate
(ATP) synthase from the thermoalkaliphilic Bacillus
sp. strain TA2.A1 facilitates ATP synthesis at low
Hybrid V
0

cation of noninteger 10:3 H
+
⁄ ATP ratio and permissive
elastic coupling. Proc Natl Acad Sci USA 101, 12159–
12164.
32 Meier T, Ferguson SA, Cook GM, Dimroth P & Vonck
J (2006) Structural investigations of the membrane-
embedded rotor ring of the F-ATPase from Clostridium
paradoxum. J Bacteriol 188, 7759–7764.
33 Pogoryelov D, Reichen C, Klyszejko AL, Brunisholz R,
Muller DJ, Dimroth P & Meier T (2007) The oligo-
meric state of c rings from cyanobacterial F-ATP synth-
ases varies from 13 to 15. J Bacteriol 189, 5895–5902.
34 Oster G & Wang H (1999) ATP synthase: two motors,
two fuels. Structure 7, R67–R72.
35 Dimroth P, von Ballmoos C, Meier T & Kaim G (2003)
Electrical power fuels rotary ATP synthase. Structure
11, 1469–1473.
36 Junge W, Lill H & Engelbrecht S (1997) ATP synthase:
an electrochemical transducer with rotatory mechanics.
Trends Biochem Sci 22, 420–423.
37 Junge W, Pa
¨
nke O, Cherepanov DA, Gumbiowski K,
Mu
¨
ller M & Engelbrecht S (2001) Inter-subunit rotation
and elastic power transmission in F
0
F

Muller DJ (2005) Structural evidence for a constant c
11
ring stoichiometry in the sodium F-ATP synthase.
FEBS J 272, 5474–5483.
43 Heise R, Mu
¨
ller V & Gottschalk G (1993) Acetogenesis
and ATP synthesis in Acetobacterium woodii are cou-
pled via a transmembrane primary sodium ion gradient.
FEMS Microbiol Lett 112, 261–268.
44 Scha
¨
gger H & von Jagow G (1987) Tricine–sodium
dodecylsulfate-polyacrylamide gel electrophoresis for
the separation of proteins in the range from 1 to
100 kDa. Anal Biochem 166, 369–379.
45 Weber K & Osborne M (1969) The reliability of the
molecular weight determination by dodecyl sulfate poly-
acrylamide gel electrophoresis. J Biol Chem 244, 4406–
4412.
46 Blum H, Beier H & Gross HJ (1987) Improved silver
staining of plant proteins, RNA, and DNA in poly-
acrylamide gels. Electrophoresis 8, 93–98.
47 Lowry OH, Rosebrough NJ, Farr AL & Randall RJ
(1951) Protein measurement with the folin–phenol
reagent. J Biol Chem 193, 265–275.
48 Kyhse-Andersen J (1984) Electroblotting of multiple
gels: a simple apparatus without buffer tank for rapid
transfer of proteins from polyacrylamide to nitrocellu-
lose. J Biochem Biophys Methods 10, 203–209.