Volume 270,
Number 35,
Issue of September 01, pp. 20568-20574, 1995
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
Asymmetry of Escherichia coli F
-ATPase as a Function of the
Interaction of 
-
Subunit Pairs with the 
and 
Subunits (*)
(Received for publication, May 12, 1995; and in revised form, June 16, 1995)
Margaret A.
Haughton
,
Roderick
A.
Capaldi (§)
From the Institute of Molecular Biology, University of Oregon,
Eugene, Oregon 97403-1229
ABSTRACT
- INTRODUCTION
- EXPERIMENTAL PROCEDURES
- RESULTS
- DISCUSSION
- FOOTNOTES
- ACKNOWLEDGEMENTS
- REFERENCES
ABSTRACT 
The asymmetry of Escherichia coli F
-ATPase
(ECF) has been explored in chemical modification
experiments involving two mutant enzyme preparations. One mutant
contains a cysteine (Cys) at position 149 of the 
subunit, along
with conversion of a Val to Ala at residue 198 to suppress the
deleterious effect of the Cys for Gly at 149 mutation (mutant
G149C:V198A). The second mutant has these mutations and also Cys
residues at positions 381 of and 108 of the 
subunit (mutant
G149C:V198A:E381C/S108C). On CuCl
treatment of
this second mutant, there is cross-linking of one copy of the
subunit to 
via the Cys at 381, a second to the subunit
(between Cys
and Cys
), while the
third subunit in the ECF complex is mostly free (some
cross-linking to 
); thereby distinguishing the three
subunits as 
, 
, and
, respectively. Both mutants have ATPase activities
similar to wild-type enzyme.
Under all nucleotide conditions,
including with essentially nucleotide-free enzyme, the three different
subunits were found to react differently with N-ethylmaleimide (NEM) which reacts with Cys
,
dicyclohexyl carbodiimide (DCCD) which reacts with Glu
,
and 7-chloro-4-nitrobenzofurazan (NbfCl) which reacts with
Tyr
. Thus, reacted with DCCD but
not NEM or NbfCl; was reactive with all three
reagents; reacted with NEM, but was poorly
reactive to DCCD or NbfCl. There was a strong nucleotide dependence of
the reaction of Cys in (but not in
) with NEM, indicative of the important role that
the subunit plays in functioning of the enzyme.
INTRODUCTION
An FF
type ATPase is found in the plasma
membrane of bacteria, as well as chloroplast thylakoid and
mitochondrial inner membranes, and this enzyme functions to synthesize
ATP in response to a light or respiratory chain-driven proton gradient.
It is a reversible enzyme, also working as an ATPase, using the
hydrolysis of ATP to establish a pH gradient for subsequent use in ion
transport processes. The simplest FF type
ATPases structurally are those from bacteria. The enzyme from Escherichia coli (ECFF) ()is made up of a total of eight different subunits, five in
the ECF part (
, , , , and ) and
three in the F part (a, b,
c
) (Futai and Kanazawa 1983; Walker et al. 1984). The F part of the enzyme from bacteria,
mitochondria, and chloroplasts is remarkably similar, particularly with
respect to , , and subunits (Cross, 1988; Senior, 1990;
Boyer, 1993). Earlier chemical studies had pointed to an intrinsic
asymmetry of F, which is expected given a stoichiometry of
three copies each of and subunits but only one copy of
, , and subunits. For example, it was shown that one of
the subunits, which contain the catalytic sites, could be
cross-linked to the subunit by the water-soluble carbodiimide EDC
(Lötscher et al., 1984a). DCCD was found
to react with only two of the three subunits, excluding the one
that is linked to the subunit (Lötscher et al., 1984a, 1984b; Tommasino and Capaldi, 1985; Stan-Lotter
and Bragg, 1986a). Also, it has been shown that NbfCl and DCCD can be
reacted with different subunits (Stan-Lotter and Bragg, 1986b).
In addition, clear asymmetry of the F was observed on
examination of F by cryoelectron microscopy (Gogol et
al., 1989a; Wilkens and Capaldi, 1994), as well as in studies of
negatively stained specimens of the enzyme (Boekema and
Böttcher, 1992).
The recent high resolution
x-ray structure of MF (Abrahams et al., 1994)
confirms and provides more details of the asymmetry of the enzyme
molecule in relation to the interaction of small subunits and
nucleotide occupancy of catalytic sites. In the crystal form examined,
one of the subunits (
) contains ADP +
Mg
, and this is close to and may make contact
with the N-terminal -helical part of the subunit. A second
subunit, containing bound AMP-PNP + Mg (
), is linked via the DELSEED region to a short
-helix (residues 82-99 in the numbering system of
ECF) in the central part of the subunit. The third
subunit, which is empty, (
), has several
interactions with the C-terminal -helical part of the
subunit.
Unanswered, as yet, is whether differences in the structure
of - subunit pairs such as observed by Abrahams et
al.(1994) are due to the association of small subunits, nucleotide
binding, or both interactions. This is particularly relevant as a
crystal form of rat liver F has been reported which is
claimed to be symmetrical with respect to features of the and
subunits and appears to have the small subunits, including
and , scrambled (Bianchet et al., 1991; Pedersen et
al., 1995).
We are using a combination of molecular biological
and biophysical approaches to examine structure-function relationships
in ECFF (e.g. Aggeler and Capaldi,
1992; Turina and Capaldi, 1994a, 1994b) Here, we describe studies in a
mutant G149C:V198A, a functioning enzyme with a Cys residue in the
catalytic site region (Iwamoto et al., 1993), thereby allowing
probing of the conformation of this region under different nucleotide
conditions. We have used this mutant, and a second mutant which
includes the above mutations along with a Cys introduced into the
DELSEED region of the subunit (through which different
subunits can be cross-linked to and subunits, respectively)
to study the asymmetry of the enzyme in relation to different
nucleotide occupancies of catalytic sites.
EXPERIMENTAL PROCEDURES
Materials
[
C]NEM and
[C]DCCD were purchased from NEN DuPont, and
Amersham Corp. respectively; Sephadex (G-50 DNA grade, fine, and G-25,
medium) was from Pharmacia Biotech Inc.; the BCA protein assay was from
Pierce; ATP assay kit was from BioOrbit (Finland); restriction enzymes
were from Boehringer Mannheim and New England Biolabs; and NEM, Nbf-Cl
(NBD-Cl), AMP-PNP, and other chemicals were from Sigma.
Strains and Construction of Plasmids Containing Mutations
in the uncD, uncC, and uncG Genes
A 1.44-kb RsrII/EagI fragment containing the uncD mutation G149C:V198A was isolated from pBMUD14 (a generous
gift from Dr. M. Futai, Osaka University, Japan) (Iwamoto et
al., 1993) and inserted into the 11.23-kb RsrII/EagI fragment of the plasmid pRA100 coding for
wild-type (derivitized pAN45 vector described in Aggeler et
al., 1992), or pRA112 containing the uncG mutation
S8C (Aggeler and Capaldi, 1992). The resulting plasmids pRA123 and
pRA124 contain the mutations G149C:V198A and
G149C:V198A/S8C, respectively, and all of the genes encoding
the ECFF ATPase. Experiments showed that the
presence of a Cys at position 8 of the subunit (for subsequent
studies in which this site is reacted with a fluorophore) had no effect
on modification by any of the reagents used here. Therefore, in some
experiments, enzyme including this mutation was used. The plasmid
pRA137, containing the mutation G149C:V198A:E381C/S108C, was
constructed by ligation of the 6.7-kb EagI/EagI
fragment of pRA123 (containing G149C:V198A) to the 5.9-kb fragment
of pRA134 (containing E381C/S108C, Aggeler et al.,
1995). The unc
E. coli strain AN888 (uncB
Mu::416 argH pyrE entA nalA
recA) (Aggeler et al., 1992) was transformed with the
plasmids pRA134 (Aggeler et al., 1995), pRA123, pRA124, and
pRA137 (this work) for expression of mutant ECF. The
wild-type E. coli strain AN1460 (pAN45/unc413::Mu argH pyrE entA nalA recA) is described in Aggeler et
al.(1992). The plasmid pAN45, and strains AN1460 and AN888, were
the generous gift of Graeme B. Cox, The Australian National University,
Canberra). The E. coli strain XL1 Blue (Stratagene) was used
for routine subcloning procedures (Davis et al., 1986;
Sambrook et al., 1989).
Preparation of ECF and
ECFF
ECF and
ECFF were isolated by a modification of the
methods of Wise et al.,(1981), Foster and Fillingame, (1979),
described in Aggeler et al.(1987) and Gogol et al. (1989b). ECF, either wild-type or mutant, was
precipitated in 70% (NH
)SO for 1 h
at 4 °C, then pelleted by centrifugation at 10,000 
g for 20 min. The protein was dissolved in 50 mM MOPS, pH
7.0, 0.5 mM EDTA, and 10% glycerol (v/v) (MOPS pH 7.0 buffer)
then passed through two consecutive centrifuge columns (Sephadex G-50,
fine, 0.5 5.5 cm) (Penefsky, 1977), equilibrated in the same
buffer at a concentration of 8-26 µM (3-10
mg/ml) in order to remove loosely bound nucleotides, as described in
Aggeler et al.(1992).
Removal of Endogenous Nucleotides and Analysis of Bound
ATP and ADP
ECF (pRA124) was freed of endogenous
nucleotides as described by Senior et al.(1992). Briefly,
protein (10 mg) was precipitated twice in 70% saturated
(NH)SO then passed through a column
of Sephadex G-25, medium, (1 cm 100 cm, flow rate 0.8 ml/h, 23
°C) equilibrated in 100 mM
Tris-HSO, pH 8.0, 4 mM EDTA, and 50%
(v/v) glycerol (Tris pH 8.0 buffer). Peak fractions were precipitated
in 70% saturated (NH)SO,
centrifuged, dissolved in 100 mM MOPS, pH 7.0, 1 mM EDTA, and 50% (v/v) glycerol, and stored in liquid N.
The concentration of endogenous nucleotide (ATP + ADP) in
denatured F plus standard ATP was determined by the
luciferin/luciferase technique as described in Fromme and
Gräber(1990), using the BioOrbit ATP assay kit and
a FlowTech Engineering Luminometer model 1030. Total ATP content was
estimated following incubation of denatured protein with phosphoenol
pyruvate and pyruvate kinase (Fromme and Gräber,
1990; Senior et al., 1992). Mutant ECF was found
to contain 1.1-1.4 mol of nucleotide (ATP + ADP) cf. 1.3-1.5 mol of nucleotide/mol in wild-type enzyme
after two gel centrifugation columns (and initial precipitation with
(NH)SO). Enzyme treated to remove
all nucleotides was found to contain 0.2-0.3 mol of (ATP +
ADP)/mol.
Maleimide Reaction of ECF and
ECFF
For modification by maleimides,
mutant ECF, either G149C:V198A or
G149C:V198A:S8C (1-2 µM) from which
loosely bound nucleotides had been removed (as above), was equilibrated
for 30 min-1 h at room temperature in a buffer of 50 mM MOPS,
pH 7.0, 0.5 mM EDTA, and 10% glycerol, with or without the
addition of nucleotide. As specified in individual experiments,
nucleotide was added to give aliquots containing 0.5 mM EDTA
(EDTA), 5 mM ATP + 0.5 mM EDTA (ATP +
EDTA), 5 mM ADP + 5.5 mM MgCl +
5 mM NaHPO (ADP + Mg + P
), or 5 mM ATP + 5.5 mM MgCl (ATP + Mg) (equivalent to
ADP + Mg + P after catalysis).
[C]NEM was incorporated by the addition of 25
µM [C]NEM (2.6 mM stock in
MeSO, specific activity 40.0 mCi/mmol) or 200 µM [C]NEM (10 mM stock in
MeSO, specific activity 8.3 mCi/mmol) to
100-200-µl aliquots of F, where the final
concentration of MeSO did not exceed 2-4% (v/v). At
specific time intervals (1-120 min), aliquots were withdrawn, and
the reaction quenched by the addition of 10 mML-cysteine or 10 mM NEM. 20-50 µg of
samples were dissociated in a DTT-containing buffer, pH 6.2, and
subjected to 10-18% linear gradient SDS-PAGE (see below). For the
analysis of cross-linked products, DTT was omitted from the
dissociation buffer. The radioactivity in subunits was determined
from gel slices and expressed as moles of [C]NEM
incorporated per mole of enzyme (as described in Aggeler et
al., 1987). The stoichiometry of incorporation of
[C]NEM into the subunits was determined
from an aliquot which was denatured in 2% SDS prior to reaction with
200 µM [C]NEM.
CuCl-induced Cross-link
Formation
ECF (7.5-12.5 µM) from
which loosely bound nucleotides had been removed (as above) was passed
through two consecutive centrifuge columns equilibrated in 50 mM MOPS, pH 7.0, 30 µM CuCl, and 10%
glycerol to induce disulfide bond formation between -,
-, -, and - (Aggeler et al.,
1995). After incubation at room temperature for a further 30 min, the
cross-linking reaction was stopped with 1 mM EDTA. For the
analysis of cross-linked products, samples were dissolved in DTT-free
dissociation buffer prior to 10-18% linear gradient SDS-PAGE.
Modification by DCCD and NbfCl
ECF (2
µM) was reacted with 200 µM DCCD (10 mM ethanolic stock), or 200 µM [C]DCCD (10 mM ethanolic stock,
26.6 mCi/mmol) in MOPS pH 7.0 buffer (50 mM MOPS, 0.5
mM EDTA, and 10% (v/v) glycerol) for up to 3 h. For
determination of ATPase activity, aliquots were withdrawn at time
intervals of 30 s, 2, 5, 10, 15, 30, 60, 120, and 180 min and assayed
immediately (5 µg/ml). In a parallel experiment, the rate of
incorporation of [C]DCCD into Glu was determined from incorporation of C into gel
slices (as described above), at the same time intervals. Where
indicated in figure legends, ECF was equilibrated with
nucleotides ATP, or ADP + P + Mg, prior to
reaction with DCCD, or [C]DCCD. Unreacted
[C]DCCD was removed by passage through a
centrifuge column of Sephadex G-50, fine, equilibrated in MOPS pH 7.0
buffer when subsequent chemical modification was required, or by
addition of 10 mM NaOAc, pH 5.2 if loaded directly on
SDS-polyacrylamide gels. Formation of cross-links between -,
-, -, and - (as described above) was
induced both prior to, and subsequent to, incorporation of
[C]DCCD.Prior to modification by NbfCl,
ECF was reacted with CuCl to induce formation
of cross-links and the reaction stopped by addition of 1 mM EDTA. NEM (200 µM) was then added (for 1 h) to modify
accessible Cys residues prior to reaction with NbfCl. Excess NEM was
removed by transfer to a buffer containing 50 mM
Tris-HSO, 1 mM EDTA, and 10% (v/v)
glycerol (Tris pH 7.5 buffer) by column centrifugation. Cross-linked
ECF (2-5 µM) was reacted with 500
µM NbfCl (10 mM ethanolic stock) in Tris pH 7.5
buffer for 1 h at 30 °C then unreacted NbfCl was removed by passage
through a centrifuge column equilibrated in MOPS pH 7.0 buffer. The
Nbf-modified F was split into two aliquots, one of which
was further modified by reaction with [C]DCCD
(200 µM) for 3 h at room temperature, after which time
unreacted [C]DCCD was quenched by addition of 10
mM NaOAc, pH 5.2. Aliquots were subjected to modified SDS-PAGE
where the pH of the separating gel was lowered from 8.6 to 8.0 in order
to increase the stability of the Tyr-Nbf adduct.
Calculation of Kinetic Constants
Kinetic constants
describing the incorporation of [C]NEM and
[C]DCCD were determined assuming
pseudo-first-order reactions of one, two, or three independently
reacting cysteinyl residues on ECF using
KaleidaGraph
data analysis and graphics program for
personal computer. For example, the following equation describes the
reaction of two cysteinyl residues with NEM,
where P = Cys-1-[C]NEM
adduct, P = Cys-2-[C]NEM
adduct, C = concentration of protein (ECF), B
= initial concentration of ligand
([C]NEM), k, k = second-order reaction constants for
Cys-1, Cys-2, respectively, and t = time.
Other Methods
ATPase activity was measured with a
regenerating system described by Lötscher et
al. (1984b). Protein concentrations were determined using the BCA
protein assay (Pierce). Trypsin cleavage (1:50 w/w, protease/protein)
of wild-type and mutant (G149C:V198A) ECF (3
µM) was carried out in 50 mM Tris-HCl, pH 8.0,
0.5 mM EDTA, and 5 mM ATP, for 1 h at room
temperature. SDS-polyacrylamide gel electrophoresis was routinely
performed with a 4% stacking gel and 10-18% linear gradient
separating gel (Laemmli, 1970) stained with Coomassie Brilliant Blue R
(Downer et al., 1976). Samples were dissociated in 0.12 M Tris-HCl buffer, pH 6.2, 2% SDS, 5% glycerol, and 50 mM DTT prior to electrophoresis. For quantitation of yield of
cross-linked products, Coomassie Brilliant Blue-stained gels were
scanned with a Microtek flat-bed scanner and the intensity
of protein bands digitized by the NIH Image 1.53 processing
and analysis program for Macintosh. For identification of
subunit composition of cross-linked products, gels were electroblotted
to Immobilon polyvinylidene difluoride membranes
(Millipore), incubated with monoclonal antibodies, and visualized as
described previously (Gogol et al., 1989a; Aggeler et
al., 1990; Mendel-Hartvig and Capaldi, 1991).
RESULTS
The double mutant G149C:V198A was obtained by Futai and
colleagues as an enzymatically active revertant of the deleterious
mutation G149C (Iwamoto et al., 1993). For our purposes,
these two mutations were inserted into the plasmid pRA100 as described
under ``Experimental Procedures.'' The ATPase activity of
ECF isolated from the mutant was 8-14 µmol of ATP
hydrolyzed/min/mg as assayed at pH 7.5 in the presence of 2 mM ATP and 5 mM MgCl, an activity in the same
range as that of the wild-type enzyme when measured under the same
conditions. After removal of the subunit by trypsin cleavage, the
activity of the mutant G149C:V198A increased to 40-55
µmol of ATP hydrolyzed/min/mg, again the same value obtained with
wild-type ECF.
The Different Reactivities of the Three Cys149 Residues
Indicates Asymmetry of ECF under All Nucleotide
Conditions
The presence of Cys allows probing of
the catalytic site region of each of the three subunits under
various nucleotide conditions. For most experiments, enzyme was
subjected to two column centrifugation steps to remove loosely bound
nucleotide, and then reacted with [C]NEM, either
25 or 200 µM, for varying lengths of time. Such
preparations were found to retain 1.1-1.4 mol of tightly bound
nucleotide which, based on other studies, were assumed to be in
non-catalytic sites. The time courses of labeling of the enzyme
prepared under these conditions showed incorporation of around 2 mol of
[C]NEM. This involves modification of only two
of three subunits (see later) which react with rate constants
differing by a factor of around 2 (Fig. 1, trace A). In
the experiment in Fig. 1, the rate of incorporation of the first
mole of NEM was calculated to be 88 M
s and that of the second mole 38 M s. As the endogenous
Cys in the subunits is buried, based on the lack of reactivity of
this site in the wild-type enzyme, this labeling is in the introduced
Cys. Some modification of the third copy of Cys could be achieved by prolonged incubation with high concentration
of [C]NEM (500 µM), when
incorporation of as much as 2.6 mol of maleimide/mol ECF was obtained. However, the rate constant for reaction of this
third subunit (0.2 M s) is negligible under the conditions used in
the experiments reported here.
Figure 1:
Kinetics of incorporation of
[C]NEM into Cys of mutant
ECF (G149C:V198A) under different nucleotide
conditions. ECF (2 µM) in MOPS pH 7.0 buffer
containing EDTA (A, crosses), ATP + EDTA (B, circles), AMP-PNP + Mg (C, triangles), or ADP + Mg + P (D, diamonds) was
incubated with 25 µM [C]NEM for
varying times up to 2 h. Aliquots were withdrawn at the times
indicated, quenched with 10 mML-Cys, and
electrophoresed on a 10-18% linear gradient SDS-polyacrylamide
gel. The incorporation of radioactivity into Cys (mol [C]NEM/mol F) was
determined in gel slices of subunits (described under
``Experimental Procedures''). The curve fits represent
modification of 2 Cys residues according to .
One possible explanation of the
observed asymmetry described above is that 1 mol of nucleotide is
tightly bound in catalytic sites rather than in non-catalytic sites
after two passages through centrifuge columns. To avoid this ambiguity,
labeling studies were also conducted with ECF freed of
nucleotide according to Senior et al.(1992). In this
procedure, the enzyme is twice precipitated from solution in
(NH)SO and then subjected to gel
filtration in a buffer containing 50% glycerol, 4 mM EDTA.
After such treatment, the G149C:V198A mutant was found to retain
less than 0.3 mol of nucleotide/mol of enzyme. Reaction of this
essentially nucleotide-free enzyme with [C]NEM
still gave the same labeling profiles as that obtained with enzyme
retaining from 1 to 2 mol of tightly bound nucleotide (result not
shown).
The profile of [C]NEM reaction with
Cys of ECF in the presence of 5 mM ATP (trace B) or 5 mM AMP-PNP + 5 mM MgCl (trace C), conditions in which all three
catalytic sites would be occupied by nucleotide, are also shown in Fig. 1. The rates of incorporation of NEM under these conditions
were 84 M s and 16 M s in ATP, and 81 M s and 17 M s in AMP-PNP,
respectively in Fig. 1. The data are similar to those obtained
for ECF without added nucleotides (trace A),
except for an approximately 2-fold lower rate of incorporation of the
second mole of NEM.
[C]NEM modification of
ECF in the presence of 5 mM ADP + 5 mg
Mg + 5 mM P, both added
directly or generated by catalytic turnover of added ATP (Fig. 1, trace D), also resulted in incorporation of
only 2 mol of reagent. However, with all three catalytic sites occupied
by ADP, the rates of NEM incorporation were considerably slower, i.e. 10 and 1 M s leading to incorporation of less that 1 mol of reagent under the
conditions described in Fig. 1(i.e. 25
µM, 2 h of incubation). The stoichiometry of incorporation
and calculation of kinetic constants (notably k),
under this nucleotide condition, was therefore determined using a
higher concentration of [C]NEM (200
µM) when 2 mol of reagent could be reacted with the
enzyme.
In different experiments, the absolute rates of NEM
incorporation varied by a factor of around 20% with the relative
differences observed under different nucleotide conditions always
maintained. From Fig. 1, it is clear that the conformation of
one or both of the NEM reactive, catalytic site regions is
significantly different when ADP is bound compared with when ATP is
bound or when the catalytic sites are empty.
[C]NEM Reaction with
ECFF from the Mutant
G149C:V198A
Profiles of [C]NEM
incorporation into Cys in ECFF isolated from the mutant G149C:V198A are shown in Fig. 2. Again, only 2 mol of NEM were incorporated when
experiments were conducted as described. For the two nucleotide
conditions, AMP-PNP + Mg and ADP +
Mg + P (generated by turnover of the
enzyme), the rates of NEM incorporation were similar to that in
ECF. Importantly, the difference in rates of modification
of Cys in ATP compared with ADP is retained in the intact
ATP synthase complex (ECFF).
Figure 2:
Kinetics of incorporation of
[C] NEM into Cys of mutant
ECFF (G149C:V198A) under different
nucleotide conditions. FF (5 µM)
in MOPS pH 7.0 buffer containing AMP-PNP + Mg (triangles), or ATP + Mg (ADP
+ P + Mg after catalysis) (squares), was
incubated with 200 µM [C]NEM for up
to 2 h. At the times indicated, the incorporation of radioactivity into
Cys of subunits (mol
[C]NEM/mol F) was determined as
described in Fig. 1.
The Catalytic Site Asymmetry of ECF is
Related to Interaction of Subunits with the and
Subunits
We have recently described a mutant,
E381C:S108C, in which CuCl treatment induces
essentially quantitative cross-linking between one subunit and
the subunit (via the Cys at 381 of and probably the
intrinsic Cys
of the subunit) (Aggeler et
al., 1995). A second subunit is cross-linked, again in near
100% yield, to the subunit (via Cys at 381 of and
Cys of the subunit). A portion of the third
subunit becomes cross-linked to the subunit, but the majority
migrates on gels as free subunit (Aggeler et al., 1995).
With this mutant, it is possible to differentiate the three
subunits by their interactions with small subunits. By combining the
mutants G149C:V198A and E381C/S108C, a novel mutant was
constructed in which the different reactivities of Cys residues could be related to the association of the different
subunits with the single copy and subunits.The
mutant G149C:V198A:E381C/S108C had an ATPase activity of
12-16 µmol of ATP hydrolyzed/min/mg, within the range
obtained for wild-type enzyme. ECF for this mutant was
reacted first with CuCl to generate disulfide bonds between
Cys residues and the and subunits,
respectively (Fig. 3). The cross-linked enzyme was then reacted
with [C]NEM which becomes incorporated into
Cys along with any Cys not
involved in disulfide bond formation with or . The rate of
reaction of NEM with the different subunits was followed by
slicing and counting C incorporated into the cross-linked
products as well as into the non-cross-linked subunit, as a
function of time. As shown in Fig. 4A, there was no
incorporation of [C]NEM into the -
cross-linked product under any of the nucleotide conditions tested.
Therefore, the subunit, to which the short central -helix of
is bound, is the one most shielded from maleimide reaction. There
was incorporation of up to 1 mol of C into the -
cross-linked product, the rate of which was nucleotide dependent (Fig. 4B). NEM modification of Cys in
this subunit occurred at the fast rate observed for the reaction
of Cys in the mutant G149C:V198A when ATP +
EDTA was present, but occurred at the slower of the two rates in this
mutant when ADP + Mg + P was
present. The nucleotide dependent switching in rates of modification of
Cys in the subunit linked to is, therefore,
more than 50-fold. In the time course of the labeling experiment, two
NEM molecules were bound into the free subunit, 1 mol into
Cys and the second into the (free) Cys.
There was no significant nucleotide dependence of the rates of
modification of these sites in the free subunit (Fig. 4C). Experiments with the mutant E381C have
shown no alteration in the rate of [C]NEM
modification of the Cys at 381 under different nucleotide conditions
(results not shown).
Figure 3:
Disulfide bond formation between
Cys and , , and subunits of mutant
ECF (G149C:V198A:E381C/S108C). Coomassie
Brilliant Blue-stained 10-18% linear gradient SDS-PAGE of
untreated (lane 1) and cross-linked mutant (lane 2)
ECF. Cross-links were induced by passage of ECF (7.5-12.5 µM) through two consecutive
CuCl-containing centrifuge columns then the reaction
stopped by the addition of 1 mM EDTA. 40 µg aliquots were
incubated in dissociation buffer in the absence of DTT prior to
SDS-PAGE.
Figure 4:
Kinetics of incorporation of
[C]NEM into cross-linked subunits of
mutant ECF (G149C:V198A:E381C/S108C).
Cross-linked mutant ECF in MOPS pH 7.0 buffer containing
ATP + EDTA (circles), or ADP + Mg + P (diamonds), was incubated with 200
µM [C]NEM for varying times up to
2.5 h. At the times indicated, the incorporation of radioactivity into
Cys (mol [C]NEM/mol
F) was determined as described in Fig. 1for
- (A), - (B), and free
subunit (C).
Activity Effects of the Modification of Cys by NEM
Fig. 5plots the residual ATPase activity of
ECF in the mutant G149C:V198A as a function of
incorporation of NEM into the Cys at 149. Reaction of 1 mol of NEM
occurring predominantly in the subunit linked to when the
reaction was done in ATP + EDTA (from Fig. 4) caused only a
60% inhibition of ATPase activity. At 2 mol of NEM incorporated, and
therefore with modification of both the linked to and free
subunit, ATPase activity was reduced to less than 2% of the
untreated enzyme.
Figure 5:
Correlation of ATPase activity with
incorporation of [C]NEM into subunits of
mutant ECF (G149C:V198A). ECF, 1
µM or 2 µM, was equilibrated in MOPS pH 7.0
buffer containing ATP + EDTA (circles) or ATP +
Mg (ADP + P + Mg after
catalysis) (squares), and incubated with 25 µM [C]NEM (ATP + EDTA) or 200 µM [C]NEM (ATP + Mg
for up to 2 h. The incorporation of radioactivity into
149 (mol [C]NEM/mol
F) was determined at various times from gel slices of the
subunit isolated by SDS-PAGE. In a parallel experiment, the
residual ATPase activity (% basal activity) was determined at the same
time intervals under the same conditions. The dashed and dotted lines plot for full inhibition at 1 and 2 mol of NEM
incorporated, respectively.
DCCD Modification of ECF Examined with the
Mutant E381C:S108C
Reaction of F with DCCD
has been shown to lead to incorporation of 1 or 2 mol/mol of enzyme
(Satre et al., 1979; Esch et al., 1981; Yoshida et al., 1981; Tommasino and Capaldi, 1985), with evidence
presented that the subunit linked to (based on
cross-linking with EDC) is only poorly reactive with the reagent
(Lötscher and Capaldi, 1984). The use of the mutant
E381C/S108C allows the specificity of DCCD for the different
mutants to be examined in more detail than previously possible. Fig. 6A shows the time dependence and Fig. 6B the molar incorporation of DCCD in relation to
inhibition of ATPase activity of the mutant. The results are
essentially the same as for wild-type (not shown, but see Tommasino and
Capaldi, 1985) and show a Mg dependence but no
nucleotide dependence of the modification of the enzyme by DCCD.
Figure 6:
DCCD modification of mutant ECF (E381C/S108C) under different nucleotide conditions and
correlation of ATPase activity with incorporation of
[C]DCCD. A, mutant ECF
(2 µM) was equilibrated in MOPS pH 7.0 buffer
containing EDTA (crosses), ATP + EDTA (circles),
AMP-PNP + Mg
(triangles), or ATP + Mg (ADP + P
+ Mg after
catalysis) (squares), and incubated with 200 µM DCCD. Aliquots were withdrawn for assay of residual ATPase
activity at various time intervals up to 3 h. B, in a
parallel experiment, aliquots of ECF (2 µM)
were equilibrated in MOPS pH 7.0 buffer containing ATP + EDTA (circles), AMP-PNP + Mg (triangles), or ATP + Mg (ADP
+ P + Mg after catalysis) (squares), and
incubated with 200 µM [
C]DCCD. At
the times indicated in A, and at 2 and 3 h, aliquots were
withdrawn and the reaction was quenched by addition of 10mM NaOAc, pH 5.2. The incorporation of radioactivity into
Glu
(mol [C]DCCD/mol
F) was determined as described in Fig. 1and plotted
against the corresponding ATPase activity.
Table 1gives data on incorporation of DCCD into the different
subunits assessed by CuCl-induced disulfide bond
formation in the E381C/S108C mutant. In these experiments,
enzyme was reacted for 1 h in EDTA-containing buffer, by which time
between 1.3 and 1.6 mol of [C]DCCD had become
incorporated into the enzyme with more than 90% inhibition of activity.
The modification of ECF by the hydrophobic carbodiimide
occurred predominantly in the subunit linked to and in the
free subunit, with no major preference between the two.
Incorporation of DCCD into the linked to was low, as
expected from previous studies, and may in part represent background
labeling, given that there was a proportionally small labeling of
and subunits in our experiments (result not shown).
Importantly, the same distribution of label, mainly into -
and into the free subunit, was observed whether the DCCD reaction
occurred first followed by disulfide bond formation to link
subunits to their partner small subunits, or if cross-linking was
performed first, followed by DCCD labeling.
NbfCl Reaction and DCCD Labeling of NbfCl-reacted
ECF
NbfCl has been found to inhibit F (95%) when reacted under conditions that modify a Tyr in the
subunit, i.e. Tyr on ECF (Andrews et al., 1984). Recently, Weber et al. (1994) have shown that the NbfCl effect requires binding of 1 mol
of reagent and that the resulting modification causes selective loss of
the lowest affinity binding site for nucleotide, i.e. that
with a K
for MgATP of around 25
µM.Fig. 7shows the labeling of ECF from the mutant E381C/S108C with NbfCl. The reagent,
detected by its fluorescence, binds predominantly to the free
subunit (seen both in the subunit band and in the -
cross-linked product), with a small amount of reaction in -,
but no labeling of that linked to the subunit.
Figure 7:
NbfCl modification of cross-linked mutant
ECF (E381C/S108C) followed by incorporation of
[C]DCCD. A, cross-links were
induced by passage of mutant ECF (7.5-12.5
µM) through two consecutive CuCl
-containing
centrifuge columns and the reaction stopped by the addition of 1 mM EDTA. Cross-linked mutant ECF was transferred to a
Tris pH 7.5 buffer, and incubated with 500 µM NbfCl for 1
h at 30 °C. Excess NbfCl was removed by column centrifugation and
an 80-µg aliquot subjected to SDS-PAGE in the absence of DTT. Lane 1, Coomassie Brilliant Blue-stained gel; lane 2,
fluorogram of the same gel. B, relative intensities of
NbfCl incorporation into the different subunits as observed upon
UV illumination (302 nm) (A, lane
2).
DCCD
reaction of NbfCl-labeled ECF led to incorporation of
[C]DCCD into the subunit linked to ,
and into the free subunit, to the same extent as with enzyme that
had no prior reaction with NbfCl (Table 1).
DISCUSSION
The studies reported here exploit the recently described
mutant of ECF, E381C/S108C, in which CuCl induces high yield cross-linking between one subunit and
, and a second subunit and the subunit. As a
consequence, it is possible to distinguish the three subunits by
their interaction with the small subunits. Additionally, a mutant was
constructed that contained a Cys introduced into the catalytic site
region ( Cys) as well as Cys at residue 381
and at residue 108. Chemical modification studies were conducted:
(i) with NEM to modify Cys, (ii) with DCCD, which reacts
with Glu, and (iii) with NbfCl, which reacts at
Tyr.
With DCCD, the modification was introduced
both prior to, and following, induction of disulfide bond formation
between subunits and and subunits, respectively. Up
to 2 mol of reagent were incorporated under both conditions with the
same distribution of reagent between the three different
subunits. This result establishes that the asymmetrical distribution of
DCCD is not induced by the incorporation of the reagent, but reflects
an intrinsic asymmetry of the enzyme. Furthermore, the data with DCCD
are reassurance that cross-linked enzyme fairly reflects the structure
of ECF, which is not significantly altered by disulfide
bond formation between with , and with subunits.
The key finding of the work presented here is that the three
different subunits have different conformations (shown
schematically in Fig. 8), in the absence of nucleotides in
catalytic sites and even when both catalytic and non-catalytic
nucleotide-binding sites are empty. Therefore, there must be an
intrinsic asymmetry of F induced by interactions of the
- pairs with the small subunits and . One
subunit, that which interacts directly with the short central
helix of the subunit (residues 82-99 in E. coli)
(), is reactive to the hydrophobic carbodiimide
DCCD but does not react with NbfCl. A Cys introduced at residue 149 in
place of Gly in this copy of the subunit is shielded from
reaction with NEM. The subunit linked to the subunit
() does not appear to bind DCCD at Glu and has very poor reactivity to NbfCl. Cys in this
subunit is the most reactive of the three to NEM. The third
subunit, the free subunit (), reacts readily
with DCCD, Cys is modified by NEM, and this copy of the
subunit is the primary site of reaction of NbfCl.
Figure 8:
Schematic representation of the different
conformations of the three subunits of ECF,
designated , , and
, relative to interactions of the -
subunit pairs with the single-copy subunits and .
forms a cross-link between Cys and Cys, reacts with DCCD but not with NbfCl,
and residue Gly 
Cys (P-loop) of this subunit
is shielded from reaction with NEM; forms a
cross-link between Cys and Cys,
incorporates NEM into Cys, but reacts only very poorly
with DCCD and NbfCl; is the primary site of
reaction with NbfCl, reacts with DCCD, and Cys of this
subunit is modified by NEM.
Asymmetry of
the enzyme, as reflected by the different reactivities of the three
subunits, is retained after binding of nucleotides into catalytic
(and non-catalytic) sites under conditions where all sites would be
occupied (i.e. 5 mM nucleotide). With either ATP
+ EDTA, AMP-PNP + Mg, or ADP +
Mg + P bound, DCCD reacted with
and but not
, while NEM reacted with ,
but not . The recent
structure determination of Abrahams et al.(1994) establishes
an asymmetry of F under conditions different from those
used here, in this case for enzyme with ADP in one catalytic site,
AMP-PNP in a second catalytic site (that which is in the subunit
linked to the short helix of and, therefore in our terminology), and a third catalytic site empty of
nucleotide. It is interesting to note that Weber et al.(1994)
have shown that the three catalytic sites have the same affinity for
ATP or ADP in the absence of Mg (i.e. in
EDTA). Therefore, the catalytic sites can become equivalent without
loss of the asymmetry induced by binding of the small subunits.
The
chemical modification studies presented here provide interesting data
on nucleotide-dependent conformational changes occurring in
ECF. Binding of ATP to the enzyme under conditions that
prevent hydrolysis of the substrate (ATP + EDTA or AMP-PNP +
Mg) does not greatly alter the reactivity of
Cys compared with when catalytic sites are empty of
nucleotide. However, binding of ADP + Mg +
P has a profound effect. The reactivity of the enzyme to
DCCD, in contrast, is not sensitive to nucleotide conditions. It is
modulated by Mg, probably indirectly, by interaction
of the cation with carboxyl groups, resulting in some shielding from
the carbodiimide. Importantly, the nucleotide-dependent conformational
change reflected in the altered reactivity of Cys occurs
specifically in that subunit linked to the subunit.
Conformational changes have been detected in the subunit during
ATP hydrolysis (Mendel-Hartvig and Capaldi, 1991; Aggeler et
al., 1992; Turina and Capaldi, 1994a) and ATP synthesis (Richter
and McCarty, 1987), in concert with translocation of this subunit
between an and a subunit (Wilkens and Capaldi, 1994). The
results presented here, then, add evidence to the proposal (Capaldi et al., 1994) that the subunit plays an important role
in the functioning of ECF
