Cell, Vol. 83, 1171-1181,
December
29, 1995, Copyright
0 1995 by Cell Press
SecA Membrane Cycling at SecYEG Is Driven
by Distinct ATP Binding and Hydrolysis Events
and Is Regulated by SecD and SecF
Anastassios
Economou,*t
Joseph A. Pogliano,*
Jonathan Beckwith,t
Donald B. Oliver,5
and William Wickner”
*Dartmouth
Medical
School
Department
of Biochemistry
Hanover,
New Hampshire
03755-3844
tuniversity
of Crete
Department
of Biology
Iraklio,
Crete
Greece
*Harvard
Medical
School
Department
of Microbiology
and Molecular
Genetics
Boston,
Massachusetts
02115
§Wesleyan
University
Department
of Molecular
Biology
and Biochemistry
Middletown,
Connecticut
06459
Summary
The SecA subunit of E. coli preprotein translocase promotes protein secretion during cycles of membrane
insertion and deinsertion
at SecYEG. This process is
regulated
both by nucleotide
binding and hydrolysis
and by the SecD and SecF proteins. In the presence
of associated preprotein,
the energy of ATP binding at
nucleotide-binding
domain 1 (NBDl) drives membrane
insertion of a 30 kDa domain of SecA, while deinsertion
of SecA requires the hydrolysis of this ATP. SecD and
SecF stabilize the inserted state of SecA. ATP binding
at NBDS, though needed for preprotein translocation,
is not needed for SecA insertion or deinsertion.
Introduction
Secretory
proteins
cross
the cytoplasmic
membrane
of
Escherichia
coli through
preprotein
translocase
(Schatz
and Beckwith,
1990; Wickner
et al., 1991). Translocation
is initiated
by the binding
energy
of ATP and is driven
by
cycles
of ATP binding
and hydrolysis
and by the proton
motive force (pmf; Schiebel
et al., 1991). The minimal
stoichiometric
subunits
of translocase
are the integral
membrane trimer
SecYEG
(Brundage
et al., 1990; Nishiyama
et al., 1994; Douville
et al., 1994, 1995) and the peripheral
subunit
SecA (Lill et al., 1989; Hart1 et al., 1990; Oliver,
1993; Douville
et al., 1995). SecYEG
and acidic
phospholipids comprise
the high affinity
membrane
receptor
of
SecA
(Lill et al., 1990;
Hart1 et al., 1990;
Hendrick
and
Wickner,
1991). SecA has two nucleotide-binding
domains
(NBDs),
the high affinity
NBDl
and the low affinity
NBD2
(Mitchell
and Oliver,
1993; see Figure
1A). SecA and SecY
are the nearest
neighbors
of a preprotein
arrested
during
its transmembrane
transit
(Joly and Wickner,
1993).
In vitro reconstitution
with purified
components
(Brundage et al., 1990) and the identification
of discrete
translocation
intermediates
(Tani et al., 1989; Schiebel
et al.,
1991) havedefineddistinctsubreactions(Hartletal.,
1990;
Schiebel
et al., 1991; Joly and Wickner,
1993; Economou
and Wickner,
1994). First, secretory
chains
associate
with
cytosolic
factors
such as the export-specific
SecB chaperone (Randall
and Hardy,
1995) and the signal recognition
particle
(Luirink
et al., 1992).
Second,
SecB-preprotein
complexes
bind to SecA by the affinities
of SecA for SecB
(Hart1 et al., 1990) and preprotein
(Lill et al., 1990; Hart1
et al., 1990). Third, upon binding
to SecA, preproteins
such
as proOmpA
(the major outer membrane
protein
A precursor) promote
the release
of ADP (Shinkai
et al., 1991) and
activate
the ATPase
activity
of SecA
(Lill et al., 1989).
Fourth,
ATP binding
causes
conformational
changes
of
SecA.
A 30 kDa domain
of SecA
inserts
into and partly
across
the membrane
(Economou
and Wickner,
1994).
Since ATP binding
energy
drives
translocation
of 20-30
aminoacyl
residuesof
the preprotein
(Schiebel
et al., 1991;
Arkowitz
et al., 1993) and preproteins
are next to SecA
throughout
their
membrane
transit
(Joly
and Wickner,
1993), SecA membrane
insertion
may be directly
coupled
to translocation
of preprotein
segments
(Economou
and
Wickner,
1994).
Fifth, upon ATP hydrolysis,
preproteins
are released
from SecA (Schiebel
et al., 1991), and SecA
deinserts
(Economou
and Wickner,
1994).
Sixth,
deinserted
SecA exchanges
with cytoplasmic
SecA (Economou and Wickner,
1994).
Seventh,
the pmf translocates
preprotein
segments
that are not engaged
with SecA
(Schiebel
et al., 1991).
Proteoliposomes
with purified
translocase
(SecA
and
SecYEG)
catalyze
multiple
rounds
of translocation
(Bassilana and Wickner,
1993) with 25% the rate and 50% the
extent
of translocation
(per SecYEG)
of E. coli inner membranes.
Additional
factors
such as the membrane
proteins
SecD and SecF (Garde1 et al., 1987,199O)
may be required
for optimal
catalysis.
SecD and SecF are essential
for cell
viability
at 23OC (Pogliano
and Beckwith,
1994a).
Overexpression
of se&F
suppresses
leader
peptide
mutations
(Pogliano
and Beckwith,
1994a),
and antibodies
that bind
to periplasmic
domains
of SecD
cause
accumulation
of
translocation
intermediates
(Matsuyama
et al., 1993).
SecD and SecF are not essential
for in vitro ATP-driven
translocation
(Arkowitz
and Wickner,
1994), and theirfunction is unknown.
Overexpression
of a 10 kb chromosomal
region containing
se&IF
and downstream
genes
leads to
membrane
insertion
of all of the cytoplasmic
SecA (Kim
et al., 1994).
We now report
that the energies
of ATP binding
and
hydrolysis
as well as translocase
interaction
with SecD
and SecF govern
the cycling
of SecA between
the inserted
and deinserted
states.
ATP
binding
and hydrolysis
at
NBDl
drive
SecA
membrane
insertion
and deinsertion,
whereas
ATP association
at NBDP couples
SecA cycling
to preprotein
translocation.
Results
SecA Mutations
in ATP-Binding
Sites
SecA
proteins
with mutations
in their ATP-binding
sites
(Mitchell
and Oliver,
1993) have been used to determine
the requirements
of SecA
insertion
and deinsertion
for
Cell
1172
ATP-derived energy. D209NSecA
has a substitution of
an aspartyl for an asparaginyl residue at position 209 (Figure 1A) in NBDl. The D209N mutation has a drastic effect
on SecA membrane deinsertion but leaves insertion unaffected. [1z51]D209N-SecA binds and inserts into inner membrane vesicles (IMVs) (Figure 1 C, closed circles) with kinetics similar to those of wild-type SecA (open squares) and
has a 30 kDa domain inaccessible to protease (Figure 1 B,
lane 4). This domain becomes accessible after membrane
solubilization with the detergent Triton X-l 00 (lane 6; also
see Economou and Wickner, 1994), and its accumulation
requires ATP, preprotein, and physiological temperature
(Figure lB), as does the wild-type protein (Economou and
Wickner, 1994). However, once inserted, D209NSecA
cannot be chased from the inserted state by an excess of
either unlabeled SecA or unlabeled D209N-SecA (Figure
lC), suggesting that D209NSecA
is defective in membrane deinsertion (Economou and Wickner, 1994). As a
result of defective deinsertion, D209NSecA
is defective
in proOmpA translocation (Figure 1D; also see Mitchell
and Oliver, 1993; van der Wolk et al., 1993), in agreement
with the observation that either the nonhydrolyzable
ATP
analog adenylyl-imidodiphosphate
(AMP-PNP) or apyrase
prevents SecA deinsertion and further translocation (Economou and Wickner, 1994). Since SecA membrane insertion and deinsertion can be mutationally uncoupled,
D209NSecA offers a unique tool with which the energetic
and ligand requirements of the SecA membrane insertion
subreaction can be studied independently of deinsertion.
proOmpA
TX-l 00
C
lime (min)
D
protease
ATP
Figure
I. SecA Mutation
+
Uncouples
Insertion
+
from
Deinsertion
(A) SecA mutants and NBDs.
(8) D209NSecA
membrane
insertion.
Reactions (100 11) contain TL
buffer (50 mM Tris-HCI
[pH 8.01, 50 mM KCI, 5 mM MgCb, 2 mM
DTT), KM-9 IMVs (100 uglml), 200 pglml BSA, 5 mM phosphocreatine,
IO ug/ml creatine
kinase, SecB (48 uglml), and [‘251]D209N-SecA
(- 100,000 cpm; 5.2 nM protomer).
Reactions
in lanes 2-6 had proOmpA (28 @g/ml), ATP (1 mM), or both, as indicated. Reactions were
at 37OC (except lane 5, OOC) for 15 min, then digested with trypsin (1
mg/ml; 15 min. OOC). The sample in lane 6 received 5 ~1 of Triton
X-100 (20% v/v) 1 min after the start of proteolysis.
Samples were
precipitated
by 15% ice-cold trichloroacetic
acid (TCA) and analyzed
by SDS-PAGE
and autoradiography.
The arrowhead
indicates the
trypsin-inaccessible
30-31 kDa domain.
(C) D209NSecA
cannot deinsert. Complete translocation
reactions
minus ATP (four portions,
1.2 ml each) containing
[‘251]SecA (open
squares) or [1251]D209N-SecA (closed circles) were prepared as in (A)
and prewarmed
at 37OC for 2 min. After removal of 100 pl (0 time
control), insertion was initiated by addition of 1 mM ATP, and 100 pl
aliquots were removed as indicated. After 20 min (arrow), unlabeled
prewarmed
SecA or D209NSecA
(both at 1 mglml) were added as
indicated to 40 fig/ml (final). Aliquots were digested with proteinase
K (1 mglml; 15 min, OOC), TCA-precipitated.
and analyzed by SDSPAGE, autoradiography.
and scanning
densitometry.
Membraneinserted SecA and D209NSecA
at 20 min were set to 100%.
(D) D209NSecA
is defective
in preprotein
translocation.
Reactions
(100 al) contained
[35S]proOmpA
(80,000 cpm), TL buffer, 100 pg of
KM-9 IMVs, SecA or D209NSecA
(40 pglml), SecB (48 pglml), BSA
(0.5 mglml), IO pglml creatine kinase. 5 mM creatine phosphate,
and
ATP (1 mM) as indicated. After 20 min at 37’C, samples were digested
on ice with proteinase K (1 pg/ml, 15 min), mixed with 15% TCA, and
analyzed by SDS-PAGE
and fluorography
(Economou
and Wickner.
1994).
SecA Membrane Insertion Is Driven by ATP
Binding Energy
Though D209NSecA
has no translocation ATPase activity, it binds nucleotide (Mitchell and Oliver, 1993). We used
nonhydrolyzable ATP analogs to test whether ATP binding
energy drives SecA insertion. AMP-PNP drives insertion,
giving rise to 20%-300/o more steady-state inserted SecA
than does ATP at 37OC (Figure 2A). This interaction has
the characteristics of a binding reaction, in that AMP-PNPdriven insertion is complete by 1 min and can occur at O°C,
albeit more slowly(Figure 2A). In contrast, accumulation of
membrane-inserted
SecA during an ATP-driven reaction
reaches steady state after 20 min (Figure 2A; Economou
and Wickner, 1994).
Strikingly, when SecA membrane insertion is driven by
AMP-PNP or AMP-PCP (adenylyl-[6,y-methylenel-diphosphonate; data not shown), the preprotein is not required
(Figures 2A and 28). Another analog, ATPyS (adenosine5’-0-[3-thiotriphosphate])
can also drive SecA insertion
(Figure 2D). However, ATPyS can be slowly hydrolyzed,
and ATPyS-SecA
insertion is enhanced by preprotein.
Furthermore, membrane insertion of D209NSecA
or accumulation of the steady state of membrane-inserted
SecA requires preprotein at all ATP concentrations (Figure
2C). In ATP-driven translocation, the preprotein may promote the release of ADP from SecA(Shinkai
et al., 1991)
a rate-limiting step for translocation (Shiozuka et al., 1990).
Since D209N-SecA(which
cannot hydrolyze ATP) can still
insert, and since wild-type SecA insertion can be driven
by nonhydrolyzable
analogs, we conclude that SecA mem-
~~~;lation
of SecA
Cycling
by ATP, SecD,
and SecF
A
ATp
#ml4 + + +
AMP-PNP
AMP-PNP
ATP+pOA
proOmpA
TX-l 00
-I-
+
+
+
+
0%
B
AMP-PNP
(PM)
unlabeled
’
0
0
10
20
30
%&4ed
40 RSOQK-SecA
Time (min)
Figure
3. NBD2
Is Not Essential
for SecA Cyclrng
(A) R509KSecA
membrane insertion. All reactrons (as in Figure 1B)
contain [‘251]R509K-SecA
(- 110,000 cpm; 4 nM protomer).
(B) R509KSecA
can deinsert from the membrane. Translocation
reactions containing [?]SecA
(open squares) or [‘251]R509K-SecA (closed
circles) were treated as in Figure 1C. After 20 min (arrow), unlabeled
prewarmed
SecA (1 mglml) or R509KSecA
(1 mglml) was added as
indicated to 40 pg/ml.
Figure
2. SecA
Membrane
Insertion
(A) [‘251]SecA (- 100,000 cpm; 4 nM protomer) was added to a SecA
insertion reaction (see Figure 1) with 1 mM AMP-PNP
(squares)
or
ATP (1 mM). Reactions without nucleotides
were preincubated
for 2
min at 37OC or 0°C and 100 fd was transferred
to ice (0 time controls).
SecA insertion was initiated by nucleotide addition, and aliquots (100
ul) were removed at indicated times. Samples were digested with proteinase K (1 mglml; 15 min; 0°C) TCA-precipitated,
and analyzed by
SDS-PAGE
and scanning densitometry.
(B) AMP-PMP-driven
SecA insertion does not requrre preprotein.
Aliquots of SecA insertion reactions (minus ATP) (98 ~1) were added to
tubes containing
AMP-PNP (1 ~1 in TL buffer; 0.0001-10
mM final
concentration)
and either proOmpA (1 fd in 6 M urea in TL buffer: 28
wglml final concentration)
or an equal volume of 6 M urea in TL buffer.
After 20 min at 37%, polypeptides
were analyzed as in (A).
(C) SecA and D209NSecA
insertion requires preprotein.
[‘*Sl]SecA
and [‘251]D209N-SecA
(each at - 100,000 cpm; 4 nM protomer) were
added to 1.5 ml SecA insertion reactions
(see Figure 1) containing
0.0001-10
mM ATP. One set of samples also contained
28 pglml
proOmpA,
while the other contained the equivalent
amount of 6 M
ureaITL buffer (-pOA). After 20 min at 37OC, samples were treated as
in (A).
(D) ATPyS-driven
SecA membrane
insertion.
[1251]SecA (- 100,000
cpm; 4 nM protomer)
was added to TL buffer, KM-9 IMVs (100 pgl
ml), 200 vglml BSA, and SecB (48 uglml). After 2 min at 37%, aliquots
(100 PI) were transferred
to ice (0 time). ATPyS (1 mM) was added,
and aliquots (100 ~1) were removed as indicated. Polypeptides
were
analyzed as in (A).
brane insertion is driven by ATP binding energy and regulated by preprotein association. ATP binding energysimultaneously drives SecA membrane insertion (Figures 26
and 2C) and translocation of short preprotein segments
(Schiebel et al., 1990). This suggests that the two events
are coupled.
To test whether ATP binding and hydrolysis at NED2
are required for SecA cycling, we examined R509KSecA,
which has a substitution of an arginyl for a lysyl residue
at position 509 within NBD2 (see Figure 1A; Mitchell and
Oliver, 1993). The R509K mutation abolishes ATP binding
to NBD2 but does not affect ATP binding at NBDl (Mitchell
and Oliver, 1993). lodinated R509KSecA
undergoes
membrane insertion with the same requirements (Figure
3A) and kinetics (Figure 3B) as wild-type SecA. Furthermore, R509K-SecA
can efficiently chase membraneinserted SecA and, unlike D209NSecA,
completes deinsertion as efficiently as SecA itself (Figure 38). Since ATP
hydrolysis is essential for deinsertion (Economou and
Wickner, 1994; see Figure 1 C), R509K-SecA membrane
cycling must be driven by its low translocation ATPase
activity (approximately 4% of wild-type; Mitchell and Oliver, 1993) at NBDl (Mitchell and Oliver, 1993). Though
R509KSecA
interacts with preprotein and has normal
membrane cycling, it does not support preprotein translocation (Mitchell and Oliver, 1993). This suggests that
R509K-SecA undergoes futile cycles of SecA insertion and
deinsertion. Thus, SecA cycling is driven by events at
NBDl, while ATP interactions at NBD2 are needed to couple SecA movement to preprotein translocation.
Cell
1174
123456
78
A
123
.
ATP+proOmpA
AMP-PNP
proieolysis
Figure
4. SecA
(min)
Membrane
Insertion
into SecYEG
Proteoliposomes
[?]SecA
(71,000 cpm; 4 nM protomer) was added to SecYEG proteoliposomes (5.7 ~1; 58 pg of reconstituted
protein/ml; lanes 1-6) or acidic
phospholipid
liposomes (lanes 7 and 8) plus TL buffer, SecB (46 wg/
ml), and BSA (200 pglml). Samples in lanes 3,4, and 6 also contained
ATP (1 mM) and proOmpA (26 pglml), while reactions in lanes 5 and
6 contained
AMP-PNP (1 mM). After 5 min at 37OC, samples were
transferred
to ice and incubated for 15 min with proteinase
K (0.5 mgl
ml). Polypeptides
were analyzed as in Figure 2. Markers of M, 43, 30,
and 21 kDa are indicated.
0.4
. DF0 DF+++
03
SecA Insertion into SecYEG Proteoliposomes
Only SecA bound at SecYEG undergoes membrane insertion (Economou and Wickner, 1994; Douville et al., 1995).
To test whether SecYEG is sufficient for SecA cycling, we
examined [1251]SecA insertion into proteoliposomes
with
purified SecYEG (Brundage et al., 1990; Bassilana and
Wickner, 1993). Limited (1 min) proteolysis of reactions
withATPandproOmpA(Figure4,lane3)orwithAMP-PNP
(lane 5) revealed an - 30 kDa proteolytic intermediate as
observed in native membranes (Economou and Wickner,
1994). In the absence of translocation ligands, less than
15% of the 30 kDa proteolytic intermediate is observed
(lane 1). A similar low amount of intermediate can be derived from [‘251]SecA bound to acidic phospholipid
liposomes (lane 7). This background level is unaffected by
translocation ligands (lane 6). Therefore, ATP- and AMPPNP-driven SecA insertion into reconstituted proteoliposomes occurs at SecYEG.
Though there is substantial SecA insertion into SecYEG
proteoliposomes,
the proteoliposome-inserted
30 kDa
SecA is unstable during longer (15 min) proteolytic digestion on ice (Figure 4, lanes 4 and 6). In contrast with integral membrane proteins, whose apolar membrane anchor
domains block transbilayer movement, the polar SecA is
capable of moving across the membrane at 0% (see
Figure 2A). Since SecA inserted into IMVs is stable to
proteolysis for up to 2 hr (Economou and Wickner, 1994),
additional stabilizing factors may be present in inner membranes but absent from the purified system. We find (below) that SecD and SecF stabilize membrane-inserted
SecA.
Membranes Depleted of SecD and SecF
To determine whether SecD and SecF stabilize membrane-inserted SecA, we constructed JP352, a derivative
of E. coli MC41 00 in which chromosomal se&If operons
are expressed under the araf3 operon promoter (Pogliano
and Beckwith, 1994b). JP352 thus synthesizes SecD and
SecF only when grown in the presence of arabinose.
Membranes from JP352 grown in arabinose (Figure 5A,
lane 3) have - 3-fold increased levels of SecD and SecF
Figure
5. Membrane
Depletion
of SecD
and SecF
(A) lmmunostaining
of membrane
preparations
with various levels of
SecD and SecF. I MVs were from MC41 00 grown in M63,0.2%
glucose
(lane 1) and from JP352 (P,,::yajCsecDfl
grown in 0.2% arabinose,
0.2% glucose (lane 3) or, after overnight
growth in arabinose,
grown
for 6 hr in arabinose-free
medium (lanes 2). IMV polypeptides
were
analyzed by SDS-PAGE
(15% acrylamide
gels [lto et al.. 19801 for
SecA, leader peptidase
[Lep], SecD, and SecF, and high Tris gels
(Douville et al., 19951 for SecY and SecG). Proteins were transferred
electrophoretically
to PVDF membranes
and immunostained
(Douville
et al., 1995). Positions of molecular weight markers
(in kilodaltons)
are indicated.
(B) Scatchard
analysis of SecA binding to urea-treated
SecDF+++ or
SecDF- IMVs was as described
(Hart1 et al., 1990; Economou
and
Wickner,
1994).
compared with MC4100 grown in glucose (lane 1). JP352
membranes prepared after growth without arabinose are
essentially devoid of SecD and SecF (lane 2). The three
membrane preparations have similar polypeptide profiles
(data not shown) and similar levels of SecY, SecG, SecA,
and leader peptidase (Figure 5A). Furthermore, depletion
or overexpression
of SecD and SecF does not affect the
activity of SecYEG as a high-affinity SecA receptor as determined by Scatchard analysis (Figure 5B), showing that
SecD and SecF are not components of the SecA receptor.
In agreement with previous observations (Arkowitz and
Wickner, 1994), depletion of SecD and SecF does not detectably perturb translocase or the membrane composition. IMVs with varying SecD and SecF content were
treated with 6 M urea to inactivate endogenous SecA (Cunningham et al., 1969) and were used in assays of SecA
membrane cycling and preprotein translocation.
SecD and SecF Stabilize Membrane-Inserted
SecA
To test whether SecD and SecF depletion affects SecA
membrane cycling, IMVs with varying levels of SecD and
Regulation
1175
of SecA Cycling
by ATP,
SecD,
and SecF
(A) Membranes
depleted of SecD and SecF lack stably inserted SecA.
SecA insertion reactions
(see Figure 1) contained
IMVs (100 pglml)
of varying SecD and SecF content plus 20 PM CCCP. After 10 min
at 37%, samples were transferred
to ice, proteolyzed,
and analyzed
as in Figure 2. The amount of membrane-inserted
SecA domain was
calculated after correction
for the size difference.
(6) Steady-state
yield of membrane-inserted
[‘?]SecA
depends on the
level of the SecD and SecF proteins. IMVs were prepared from strain
JP352 grown to an ODmo of 0.6 in M63 with 0.2% arabinose
(lane
1) or from successive
generations
of the same cells grown without
arabinose as described
in Experimental
Procedures.
IMVs (3 Kg/lane)
In lanes 2-6 represent five successive
generations
of cells. IMVs (100
pglml) were added to translocation/SecA
cycling reactions,
and the
30-31 kDa domain of SecA (bottom; arrowhead)
was analyzed
as
In (A).
(C) Quantitation
of data from (8).
(D) Effect of pmf on SecA membrane
insertion. Urea-treated
KM-9
IMVs (200 uglml) were added to translocation/SecA
cycling reactions
(20 min, 37OC) with 0 to 10 mM ATP. Reactions
marked plus pmf
contain 5 mM succlnate to establish a transmembrane
proton gradient
(Schiebel 81 al., 1991).
SecA after only a limited reduction of SecD and SecF (Figure 8B), suggesting that loss of inserted SecA is not an
indirect effect. The amounts of inserted SecA are roughly
proportional to the concentration of SecF (Figure 6C).
SecD and SecF are required for the maintenance of pmf
(Arkowitz and Wickner, 1994), but this is unlikely to be
related to their role in SecA cycling. The IMVs employed
here are proton permeable as a result of F, removal (Geller
et al., 1986; Yamadaet al., 1989), and all reactionscontain
the uncoupler CCCP (carbonyl cyanide m-chlorophenylhydrazone). Moreover, pmf does not affect ATP-driven
[1251]SecA insertion into IMVs that show wild-type levels of
SecD and SecF but that lack the F,FbATPase (Klionsky
et al., 1984). Imposition of a pmf by addition of succinate
stimulates SecA insertion at up to 10 PM ATP (Figure 6D),
in agreement with pmf lowering the apparent K, for ATP
of preprotein translocation (Shiozuka et al., 1990). However, no such stimulation was observed at physiological
(1 mM) ATP concentrations (Figure 6D). It is therefore unlikely that the pmf plays a major role in normal SecA membrane cycling.
IMVs lacking SecD and SecF could, like SecYEGproteoliposomes,
allow SecA insertion but fail to stabilize
the inserted state during proteolysis at O°C. Reactions
containing [?]SecA
and either SecDF- or DF+++ membranes were incubated for 20 min at 37OC and proteolyzed
at 0%. After 1 min at O’X, a proteolytic intermediate of
30-31 kDa was observed in both SecDF+++ (Figure 7A,
lane 2) and SecDF- (lane 1) IMVs. Membranes lacking
SecD and SecF have 400/o-60% less 30 kDa material,
which, after 15 min of proteolysis, becomes completely
digested (lane 3). In contrast, in reactions containing
SecDF+++ membranes, approximately 60% of the 30 kDa
domain seen after 1 min of digestion (lane 2) remains protease-inaccessible
(lane 4). We next performed a systematic analysis of the kinetics of proteolysis of [1*51]-labeled
30 kDa SecA either in aqueous solution or in reactions
that contain membranes with different levels of SecD and
SecF, either with or without ATP and proOmpA (Figure
78). Soluble SecA yields significant amounts of a 30 kDa
fragment at short times of proteolysis ([A] in Figure 7B),
but these are substantially reduced in the presence of
ligands ([B] in Figure 78) or IMVs ([Cl and [E]) and could
be unrelated to the domain of SecA that undergoes membrane insertion at SecYEG. Significant levels of the 30
kDa intermediate were seen after short digestion times
with SecDF- or SecDF+++ membranes but only after preprotein translocation had occurred ([D] and [F] in Figure
78; contrast with [C] and [El, no preprotein or ATP). However, only with SecD and SecF is the 30 kDa domain stabilized to remain inaccessible after prolonged digestion ([F]
in Figure 78).
SecF, i.e., wild-type levels(SecDF+), depleted(SecDF-),
or
3-fold overexpressed (SecDF+++), were added to complete
translocation reactions containing [l*SI]SecA (Economou
and Wickner, 1994). Strikingly, IMVs depleted of SecD and
SecF maintain very low amounts of steady-state inserted
SecA (Figure 6A). IMVs show reduced inserted state of
SecD and SecF Are Not Needed to Stabilize
Membrane-Inserted
SecA When ATP
Hydrolysis Is Prevented
SecA deinsertion is driven by ATP hydrolysis (Economou
and Wickner, 1994). We therefore examined whether the
proteolytic stability of the membrane-inserted
[1251]-labeled
30 kDa domain of SecA requires ATP hydrolysis. Com-
0
0
50
100
150
Total SecA (,q /ml)
200
0Lzl
0123456
SecF (artxbary umts)
Figure 6. Membranes
Depleted
High Levels of Membrane-Inserted
of SecD
SecA
and SecF
Do Not Sustain
Cell
1176
A
123456
7
0
SWDF
Proteolysis(min)
un
123456
.
SecDF
ATP
Figure
7. SecD
and SecF Stabilize
Membrane-Inserted
SecA
(A) Effect of SecD and SecF on proteolysis
of the 30 kDa domain
of SecA. TranslocationlSecA
cycling reactions containing
SecDF- or
SecDF+++ IMVs (100 uglml) plus [‘251]Se~A (- 100,000 cpm; 4 nM protomer) and 20 nM CCCP were incubated at 37’C for 20 min, transferred to ice, digested with trypsin (1 min or 15 min as indicated), and
analyzed by SDS-PAGE
and autoradiography.
Full-length SecA (open
arrow) and the 30 kDa domain (closed arrow) are indicated.
(B) Kinetics of proteolysis
of the 30 kDa domain of SecA. Complete
translocation
reactions (graphs D and F) and reactions without membranes (A and B) or that contained
SecDF- or SecDF”’
membranes
but no translocation
ligands (C and E, respectively)
were analyzed as
in Figure 7A. The plus and minus symbols in the right margin indicate
the presence or absence of ATP and proOmpA.
The 30 kDa domain
of SecAwasquantitated
bypcountingof
gel-excised
bands. Reactions
in (C)-(F) contained
100 nglml IMVs.
(C) Effect of ATP hydrolysis
on stability of inserted SecA. Complete
SecA insertion reactions
(50 nl; lanes 5-8) and reactions with AMPPNP (1 mM; lanes 5-6) containing [‘251]SecA and IMVs with different
SecD and SecF contents
(as indicated) were prepared as described
in the legend to Figure 1. Samples in lanes 1-4 were incubated for
16 min, then mixed with apyrase (2.5 U/50 nl reaction) and incubated
for 2 min, chilled on ice, and analyzed as in Figure 1. Full-length SecA
(open arrow) and the 30 kDa domain (closed arrow) are indicated.
(D) Membrane insertion of [1251]D209N-SecA (143,000 cpm; 4 nM protomer) in the absence of SecD and SecF. TranslocationlSecA
cycling
reactions contained
IMVs (100 uglml) of varying SecD and SecF content (as indicated) and 20 nM CCCP. Lanes 4-6 also contain ATP (1
mM), while lanes l-3 have an equal volume (1 nl) of TL buffer. The
30 kDa domain is indicated (closed arrow).
plete translocation reactions or reactions containing AMPPNP were incubated for 20 min at 37OC. ATP was removed
by a 2 min incubation with apyrase. All samples were then
proteolyzed on ice for 1 or 15 min. Apyrase addition increases the amounts of 30 kDa SecA by 30% (Economou
and Wickner, 1994). Strikingly, even in the absence of
the SecD and SecF, removal of free ATP stabilizes the
membrane-inserted
30 kDa domain that remains after proteolytic digestion of the rest of SecA (Figure 7C, lane 1
versus lane 3; compare with Figure 7A, lanes 1 and 3,
without apyrase). AMP-PNP also stabilized this domain
against proteolysis in the absence of SecD and SecF (lane
7). The 30 kDa proteolytic fragment may thus be capable
of ATP hydrolysis.
To test the role of ATP hydrolysis in the stability of membrane-inserted
SecA further, we examined D209NSecA
insertion into membranes that have, or lack, SecD and
SecF. [1251]D209N-SecA was added to translocation reactions with IMVsof varying SecD and SecF content. D209NSecA insertion requires ATP (Figure 7D, lanes 4-6) but
not SecD and SecF (lane 5). D209NSecA that is inserted
into SecDF- IMVs gives rise to a 30 kDa proteaseinaccessible domain that is stable after 15 min of digestion
(Figure 7D). Therefore, when ATP hydrolysis is prevented
by either a mutation in the NBDl of SecA, by ATP removal
by apyrase, or by a nonhydrolyzable
ATP analog, the stability of membrane-inserted
SecA becomes independent
of the presence of SecD and SecF. However, under physiological conditions of ATP hydrolysis, the SecD and SecF
proteins regulate the stability of inserted SecA. If SecD
and SecF stabilize the inserted form of SecA after ATP
hydrolysis, their depletion may affect preprotein translocation. Under conditions allowing multiple turnovers, there
is a 30%~50% reduction in the apparent V,,, for proOmpA
translocation (Figure 8A) and a 20%-30%
reduction of
translocation ATPase activity (Figure 86, closed circles)
in the absence of SecD and SecF. This correlates with
the reduction in the steady-state levels of membraneinserted SecA seen in the absence of SecD and SecF (see
Figure 7).
Cells depleted of SecD and SecF are cold-sensitive
(Pogliano and Beckwith, 1994a), but it is unlikely that altered SecA cycling is directly responsible. Comparisons
of assays containing SecDF+++ and SecDF- IMVs at 37OC
and at 23OC reveals that translocation ATPase activity is
reduced by 3- to 5-fold at 23OC (Figure 86) and preprotein
translocation and SecA membrane insertion by less than
2-fold (Figures 8C and 8D, respectively) in either the presence or absence of SecD and SecF. Perhaps, in the absence of SecD and SecF, the failure to stabilize the inserted SecA may exacerbate the inherent cold sensitivity
of the in vivo translocation process (Pogliano and Beckwith, 1993).
Discussion
Preproteins initiate translocation by binding to SecA (Hart1
et al., 1990) triggering a remarkable cascade of events
(Schiebel et al., 1991; Economou and Wickner, 1994). This
Regulation
1177
of SecA Cycling
Figure 8. Effect of SecD
Translocation
ATPase
by ATP,
and SecF
Se&,
Depletion
and SecF
on Translocation
and
(A) ProOmpA translocation.
[YS]proOmpA
(1,500,OOO cpmlml. O-40
uglml) was added to 50 ul translocation
reactions
containing
SecDF+++or SecDF- IMVs and 20 HIM CCCP. Samples were removed after
10 min at 37OC and digested on ice with TPCK-treated
trypsin (1 mgl
ml; 15 min). Polypeptides
were analyzed as in Figure 1. Translocated
[%]OmpA
was visualized
by fluorography
and quantitated
by scanning densitometry.
(B)Translocation
of ATPase. Urea-treated
IMVs(100 pg/ml)from either
MC4100 (DF’) or JP352 grown with arabinose (DF”‘)
or afler removal
of arabinose (DF-) were incubated at 37% or at 23OC as indicated in
TL buffer containing
SecA (40 rig/ml) and proOmpA (28 nglml) in 50
nl reactions.
Minus preprotein
background
incubations
contained 6
M urea, 50 mM Tris-HCI
(pH 8.0) instead of proOmpA (0.5 and - 1.3
nmol P, per mg of IMV protein at 0 and 45 min, respectively),
which
were subtracted
from the values shown. At indicated times, the P,
content of 10 nl aliquots was determined
(Lill et al., 1989; Douville et
al., 1995).
(C and D) In vitro translocation
of proOmpA and membrane
insertion
of SecA in the absence of SecD and SecF is not sensitive to cold.
[%]proOmpA
translocation
(C)and SecA membrane insertion (D) into
DF~ or DF+++ IMVs were assayed
as described
in Figures 2 and 3,
except that reactions were incubated at 37°C and at 23”C, the temperature at which in vivo cold sensitivity
can be observed
(Pogliano and
Beckwith,
1994a). The protease-inaccessible
30 kDa domain of SecA
in (D) is indicated by arrowheads.
association activates cycles of SecA membrane insertion
and deinsertion (Economou and Wickner, 1994) that are
driven by the ATPase activity of SecA (Lill et al., 1989) and
constitute the central engine of the ATP-driven segment of
preprotein translocation.
Using mutations in the NBDs of SecA and ATP ana-
logs, as well as in vitro biochemical reconstitution experiments, we have dissected SecA membrane cycling and
translocation into subreactions. These results and others
lead to a working hypothesis. Deinserted SecA-ADP is
bound to SecYEG. Preprotein binding to SecA (Hart1 et
al., 1990) causes conformational
changes that promote
nucleotide exchange (Shinkai et al., 1993) of ATP for ADP
at NBDl. The 30 kDa domain of SecA then inserts into
the membrane at SecYEG, and 20-30 aminoacyl residues
of the preprotein coinsert. This may be promoted by ATP
binding at NBD2. SecD and SecF interact with inserted
SecA, SecYEG, or both. ATP hydrolysis at NBDl influences ATP hydrolysis at NBD2. Translocated domains of
the preprotein are released from SecA, and SecA deinserts, allowing pmf-driven translocation of preprotein domains not bound to SecA. The succeeding preprotein domain can then bind to SecA. The cycle of SecA insertion
and deinsertion is governed by preprotein binding and dissociation, ATP binding and hydrolysis, and SecD and
SecF.
The data strongly indicate that the 30 kDa domain of
SecA becomes protease-inaccessible
by insertion into the
membrane rather than by assuming a protease-resistant
conformation or by being protected from protease by other
polypeptides. First, proteolysis of translocation reactions
only yields 30 kDa SecA when each of the components
needed fortranslocation,
namely preproteins, ATP, physiological temperature, and SecYEG, is present (Economou
and Wickner, 1994; Douville et al., 1995). Second, disruption of membrane integrity in the presence of protease by
sonication, nonionic detergents, or a single freeze-thaw
cycle allows digestion of the 30 kDa domain, whereas sonication or freeze-thaw treatments, followed by addition of
protease to the resealed vesicles, preserves the integrity
of the 30 kDa domain (Economou and Wickner, 1994).
Third, membrane-inserted
SecA can be proteolyzed or biotinylated from the periplasmic surface of the membrane
(Kim et al., 1994). Fourth, a cross-linker, attached to a
translocation-arrested
preprotein, reacts with SecA whenever it is at a position to react with SecY (Joly and Wickner,
1993). Fifth, after proteolysis has removed all but the 30
kDa domain of SecA, this domain is stabilized by SecD
and SecF (Figures 6 and 7) inner membrane proteins that
are largely exposed to the periplasm. It is unlikely that the
30 kDa domain of SecA is protected from protease by
the small cytoplasmic loops of SecD and SecF. Finally, in
addition to causing a single, irreversible SecA membrane
insertion (Figures 2A and 2B), AMP-PNP causes a single
“loopful” of 20-30 aminoacyl residues of the preprotein to
undergo a single, irreversible membrane insertion (Schiebel et al., 1991). This strongly suggests that SecA membrane cycling drives preprotein movement.
The two nucleotide-binding
sites of SecA have very distinct roles. The observation that D209N inserts normally
but is blocked in deinsertion allowed independent study
of the insertion step. ATP binding at NBDl drives a 30 kDa
domain of SecA to insert into the membrane at SecYEG
(Figure 4). Concomitant with SecA movement, limited preprotein translocation is observed (Schiebel et al., 1991;
Cell
1178
Arkowitz et al., 1993). NBD2 is essential for translocation
but apparently is not required for SecA membrane insertion, since a mutant that cannot bind ATP at NBD2 cycles
normally. Translocated preproteins are released from SecA
upon ATP hydrolysis (Schiebel et al., 1991) which could
take place at both NBDl and NBD2. Mutations in either
NBD reduce (R509K) or abolish (D209N) the translocation
ATPase activity of SecA, suggesting that ATP interactions
at the two sites are interdependent.
The finding that removal of free ATP by apyrase stabilizes inserted SecA
(Figure 7D) and prevents its deinsertion (Economou and
Wickner, 1994) indicates that additional ATP binding at
NBDl may be required for deinsertion. Other membrane
transporters, such as the cystic fibrosis transmembrane
conductance regulator (Anderson and Welsh, 1992; Gunderson and Kopito, 1995) and the multidrug resistance
P-glycoprotein (Gill et al., 1992), also contain two ATPbinding domains with distinct functions.
Is the ATP-driven SecA membrane insertion mechanistically coupled to preprotein translocation?
After 1 min of
proteolysis at 0°C has removed most of SecA, the stability
of the remaining membrane-inserted
30 kDa domain is
remarkably different in the presence of ATP or of the nonhydrolyzable analog AMP-PNP (Figure 7A versus lanes
5-8 of Figure 7C). This suggests that ATP can be hydrolyzed by the 30 kDa proteolytic fragment. The D209N
point mutation in NBDl (Figure 7D) and AMP-PNP have
similar effects on 30 kDa stability, suggesting that this
proteolytic fragment of SecA contains NBDl. ATP bound
to the membrane-inserted
30 kDa domain at NBDl may
also be apyrase-inaccessible
(Figure 7C). Strikingly, a region of SecA between NBDl and NBD2 can be crosslinked to preproteins in solution (Kimura et al., 1991) and
point mutations at positions 373 and 488 have a Prl phenotype (Fikes and Bassford, 1989; Bieker-Brady and Silhavy,
1992). This region may bind leader peptides (Lill et al.,
1990; Akita et al., 1990) and mature preprotein domains
(Andersson and von Heijne, 1993; Schiebel et al., 1991;
Arkowitz et al., 1993). Mature preprotein segments recognized by SecA have a consistent size of 25-40 aminoacyl
residues (Andersson and von Heijne, 1993; Schiebel et
al., 1991). A preprotein-binding
site on the 30 kDa/NBDl
domain of SecA would offer a simple link between SecA
membrane insertion and insertion of 20-30 aminoacyl residues of the preprotein.
How does the preprotein regulate SecA cycling? Preprotein binding causes a conformational change in SecA and
ADP release (Shinkai et al., 1991) indicative of an altered
structure at the nucleotide-binding
sites. AMP-PNP may
differ from ATP in that it may bind to SecA of either conformation, the forms with or without bound proOmpA, to promote the insertion (Figure 28). AMP-PNP may not only
promote forward movement but also prevent deinsertion,
thus alleviating the requirement for preprotein. Without
preprotein, membrane-bound
SecA has a basal ATPase
activity (Lill et al., 1990) and modest SecA membrane insertion (less than 15% of that seen with preprotein; Economou and Wickner, 1994). These results and nucleotide
binding studies (Lill et al., 1989; Matsuyama et al., 1990;
Shinkai et al., 1991; Mitchell and Oliver, 1993) demonstrate that the preprotein is not essential for ATP binding
to SecA. This conclusion is strengthened
by the finding
that the binding energy of AMP-PNP also drives high levels
of SecA membrane insertion in the absence of preprotein.
However, normal SecA cycling and membrane insertion of
the hydrolysis-defective
D209NSecA
require preprotein.
Furthermore, SecA membrane insertion, driven by the
poorly hydrolyzable ATPyS, is increased 3-fold in the presence of preprotein. Preprotein binding might increase the
affinity of SecA for ATP. In the ATP- and ATPyS-driven
reactions, preprotein may overcome a rate-limiting step
of translocation such as ADP release (Shiozuka et al.,
1990; Shinkai et al., 1991).
While SecYEG is the only integral membrane protein
that is essential for SecA cycling (Figures 4 and 7; Economou and Wickner, 1994; Douville et al., 1995) optimal
cycling and translocation also require SecD and SecF.
These two proteins, which may form a dimer (Pogliano
and Beckwith, 1994a; Sagara et al., 1994), are of clear
importance in vivo (Garde1 et al., 1990; Bieker-Brady and
Silhavy, 1992; Pogliano and Beckwith, 1994a). We propose that SecD and SecF regulate SecA deinsertion, presumably by interacting with either SecYEG or SecA. This
supports previous proposals for a direct interaction of
SecD and SecF with translocase (Bieker-Brady and Silhavy, 1992) late in the catalytic cycle (Garde1 et al., 1990;
Bieker-Brady and Silhavy, 1992; Matsuyama et al., 1993;
Pogliano and Beckwith, 1994a). Such interactions could
be disrupted by a-SecD antibodies (Matsuyama et al.,
1993). The aggreement between in vivo and in vitro experiments that use overexpression, depletion, or antibody inactivation suggest that it is unlikely that SecD and SecF
affect SecA cycling indirectly. This is further supported by
the similar properties of SecD and SecF-depleted
IMVs
(Figures 6 and 7) and SecYEG proteoliposomes
(Figure
4) and by the rapid effects of SecD inactivation in vivo
(Matsuyama et al., 1993; Pogliano and Beckwith, 1994a).
Associations between the polytopic SecD and SecF proteins and translocase could involve membrane domains
of SecD and SecF, or their large periplasmic regions (- 45
kDa and - 11 kDa, respectively; Garde1 et al., 1990; Pogliano and Beckwith, 1994b), which are required for function
(Matsuyama et al., 1993). Domains of membrane-inserted
SecA are exposed to (or near) the periplasm (Kim et al.,
1994; Economou and Wickner, 1994) and SecY (Joly and
Wickner, 1993) and could also be near SecD and SecF.
SecD and SecF may also prevent the leakage of ions via
translocase, required for the maintenance of a stable pmf
(Arkowitz and Wickner, 1994).
Several questions derive from our working model. First,
how is ATP binding energy transformed into conformational changes of the two SecA protomers? Second, what
interactions within SecA ensure communication
between
NBDl and NBD2? Third, how does preprotein interact with
SecA, and what is the role of NBD2 in coupling SecA membrane penetration to preprotein transfer? Fourth, what is
the catalytic role of SecD and SecF interactions with translocase? To address these questions, we will need to use
Regulation
1179
of SecA Cycling
by ATP,
SecD,
and SecF
pure proteins in a reconstituted in vitro system and to determine the 3-dimensional structures of translocase and
its domains. The requirement for SecD and SecF for stability of inserted SecA may offer a biochemical assay for their
functional purification.
Experimental
Procedures
Bacterial
Strains and Membrane
Purification
D209N-SecA.
R509KSecA,
and Tl OSN-SecA were purified from strain
BL21.19 (~DE3, supF(Ts), trp(Am), Zch::TnlO,
secA73(Am),
clpA379::
kan, recA::cat;
Mitchell and Oliver, 1993).
To manipulate
se&F
expression,
we constructed
strain JP352
(araA.774,
A[argF-lac]Ul69,
rpsL7.50,
relA7, tbi, flb5307,
deoC7,
ptsf25.
recA::cat.
tgt::kan, araCk, P,,::yajCsecDF),
a derivative
of
MC41 00 (araD739,
A(argNacJU769,
rpsL750, relA 7, fbi, flbB5307,
deoC7, ptsF25, fbsR; Silhavy et al., 1984) in which the chromosomal
yajCSecDFoperon
was fused 3’to the arabinose promoter Psao. First,
in plasmid pKJ1, which carries the yajCSecDF operon and upstream
sequences
(Pogliano and Beckwith, 1994b). tgt, a nonessential
gene
lying upstream of yajCSecDF,
was replaced with a kanamycin
resistance gene @an). The kan gene was isolated on a Pstl DNA fragment of
plasmid pUC4K (Pharmacia),
was blunt-ended
with Klenow fragment
polymerase,
and was cloned into the Smal site of tgr. Second, tgt::
kan isolated as an EcoRV fragment
was ligated into the Ndel site
upstream of P,o::yajSecDFof
pGAP1 (Pogliano and Beckwith, 1994a)
to form pGAP12. In pGAP12, the kan gene is transcribed
divergently
from yajCSecDF. Third, the chromosomal
tgt-yajCsecDFwas
replaced
with that from linearized
pGAP12 by homologous
recombination
in
strain JCB43.3 (recD7903::miniTnlO).
Fourth, the tgr::kan-Peao::yajCsecDFregion
was transduced
into JP313(MC4100
araA774). Finally,
the recA::cat
allele was transduced
into the resulting strain, yielding
JP352.
IMVswere
from E. coli MC4100, JP352, and KM9 (uric ::TnlO, re/Al,
spoT7, metB7; Klionsky et al., 1984). Media and antibiotics
were according totheprotocolsof
Sambrooketal.
(1989). Cultures weregrown
at 37% with aeration in M63 medium, 10 uglml thiamine, and glucose
or arabinose (0.2% w/v). For complete depletion of SecD and SecF,
JP352 (P,,o::yajCsecDF)
cells were grown to an OD,
of 0.6 in M63,
glucose (0.2% w/v), and arabinose
(0.2% w/v). harvested
(5 min,
10,000 rpm, Sorval JA20 rotor, 4°C) washed in medium without arabinose, and resuspended
In M63 containmg
glucose (0.2% w/v) to an
ODsoo of 0.3. Cells were further grown to an OD, of 0.6 and diluted
1:2 with M63 with glucose (0.2%). This process was repeated for a
total growth of 6.5 hr, when SecD and SecF are depleted beyond
rmmunodetection
(Frgure 4). IMVs were prepared from 8-12 liters of
culture (Douville et al., 1995) and treated with 6 M urea (35 mm, O°C;
Cunningham
et al., 1989).
a-SecY antisera, by measuring translocation
ATPase, and by translocation of [%]proOmpA.
SecYEG proteoliposomes
and acidic phospholipid vesicles (Economou
and Wickner,
1994) were frozen in liquid
nitrogen and stored at -80%
lodinated
SecA
SecA, D209NSecA,
and R509KSecA
were iodinated and assayed
for binding to urea-treated
IMVs (100 pglml) as described
(Economou
and Wickner,
1994). Fluorograms
were scanned with an Apple Color
OneScanner
with Ofoto software (version 2.02) and were analyzed
with NIH Image software (version 1.52) on a Macintosh
Quadra 650
computer.
Radioactivity
of [‘251]-labeled polypeptides
was determined
by y-counting
excised gel bands. Sodium dodecyl sulfate polyacrylamide gel electrophoresis
(SDS-PAGE)
using 15% polyacrylamide
or
high Tris gels (Douville et al., 1995), fluorography.
and protein transfer
to PVDF membranes
were as described
(Economou
and Wickner,
1994). lmmunostaining
was with rabbit antisera to synthetic peptides,
a-SecF and a-SecD (Arkowitz and Wickner,
1994) a-SecY (Brundage
et al., 1990) and a-SecG (Douville et al., 1994) or against purified
proteins, a-SecA (Lill et al., 1989) and a-leader peptidase (Wolf et al.,
1982).
Acknowledgments
A. E. dedicates this paper to the memory of Dr. Apostolos Economou.
Correspondence
should be addressed
to A. E. at Dartmouth.
We are
grateful to Marilyn Rice Leonard, Bernadette Hils, and Barbara Atherton for expert assistance,
and to Franck Duong and Lily Karamanou
for useful discussions.
This work was supported
by grants from the
National Institutes of General Medical Sciences. A. E. was supported
by postdoctoral
fellowships
from the European Molecular Biology Organization and the Human Frontier Science Project Organization.
Received
September
5, 1995; revised
October
26, 1995.
References
Akita, M., Sasaki, S., Matsuyama,
S., and Mizushima.
S. (1990). SecA
interacts with secretory
proteins by recognizing
the positive charge
at the amino terminus of the signal peptide in Escherichia
co/i. J. Biol.
Chem. 265, 8164-8169.
Anderson,
M.P., and Welsh, M.J. (1992). Regulation by ATP and ADP
of CFTR chloride channels that contain mutant nucleotide-binding
domarns. Science 257, 1701-l 704.
Andersson.
H., and von Heifne, G. (1993). Set dependent
and set
independent
assembly of E. colimner membrane proteins: the topological rules depend on chain length. EMBO J. 72, 683-691.
Arkowitz, R.A., and Wickner, W. (1994). SecD and SecF are required
for the proton electrochemical
gradient stimulation of preprotem translocation. EMBO J. 73, 954-983.
Arkowitz.
R.A., Joly, J.C., and Wickner,
W.T. (1993). Translocation
can drove unfolding of a preprotein
domain. EMBO J. 72. 243-253.
Proteins
and Chemicals
Proteins were purified as described:
SecA, D209N-SecA,
R509KSecA, and T108N-SecA,
Cunningham
et al. (1989); SecB, Weiss et
al. (1988) and Lecker et al. (1989); proOmpA,
Crooke et al. (1988);
[%]proOmpA,
Crooke and Wickner (1987). Protein concentration
was
determined
with Bradford reagent (Bio-Rad) wtth bovine albumin as
standard. AMP-PNP, AMP-PCP, ATP$S, creatme kinase, and creatine
phosphate were from Boehringer. Arabinose, ATP. dithiothreitol
(DTT).
bovine serum albumin (BSA), phenylmethylsulfonyl
fluoride (PMSF),
CCCP, potato apyrase (grade VIII; 470 U ATPase activitylmg),
and
N-tosyCL-phenylalanine
chloromethylketone-treated
trypsin were from
Sigma. Protern molecular werght markers were from Amersham (Rainbow markers) and Bethesda Research Laboratories
LifeTechnologies
(‘C-labeled
markers).
Bieker-Brady,
K., and Silhavy, T.J. (1992). Suppressor
analysis suggests a multistep, cyclic mechanism
for protein secretion.
EMBO J.
77, 3165-3174.
SecYEG Reconstituted
Proteoliposomes
KM-9 membranes
were solubilized
rn 1.5% 8-octyl glucoside,
20%
glycerol, 3.75 mglml E. coli acidic phospholipids.
1 mM DTT. 20 mM
Tris-HCI
(pH 8.0) and chromatographed
on DE-52, then reconstituted
Into proteoliposomes
(Brundage et al., 1990; Bassrlana and Wickner,
1993, Douville et al., 1995). Proteoliposomes
bearing SecYEG were
rdentrfied by silver staining of SDS gels, by rmmunodetection
using
Crooke. E., and Wrckner,
W.T. (1987). The trigger factor: a soluble
protein whrch folds proOmpA into a membrane
assembly competent
form. Proc Natl. Acad. Sci. USA 84, 5216-5220.
Bassilana, M., and Wickner, W. (1993). PurifiedE. colipreprotein
translocase catalyzes
multiple cycles of precursor
protein translocation.
Biochemistry
32, 2626-2630.
Brundage, L., Hendrick, J.P., Schiebel, E., Driessen,A.J.M..and
Wickner, W. (1990). The purified E. coli integral membrane
protern SecY/
E is sufficient for reconstitution
of SecA-dependent
precursor
protein
translocation.
Cell 62, 649-657.
Chang, C.N., Blobel, G., and Model, P. (1978). Detection of prokaryotic
signal peptrdase in an Escherichia colimembranefraction:
endoproteolyhc cleavage of nascent fl pre-coat
protein. Proc. Natl. Acad. Sci.
USA 75, 361-365.
Crooke, E., Brundage. L.. Rice, M., and Wickner, W. (1988). ProOmpA
spontaneously
folds In a membrane
assembly competent
state which
Cdl
1180
trigger
factor
stabilizes.
EMBO
J. 7, 1831-1835.
Cunningham,
K., Lill, R., Crooke, E., Rice, M., Moore, K., and Wickner,
W. (1989). Isolation of SecA protein, a peripheral
protein of the E.
co/i plasma membrane that is essential for the functional binding and
translocation
of proOmpA.
EMBO J. 8, 955-959.
Douville, K., Leonard, M., Brundage,
L., Nishiyama,
K., Tokuda, H.,
Mizushima.
S., and Wickner, W. (1994). Band 1 subunit of Eschericbia
co/i preprotein translocase
and integral membrane
export factor PI2
are the same protein. J. Biol. Chem. 269, 18705-18707.
Douville, K., Price, A., Eichler, J., Economou,
A., and Wickner,
W.
(1995). SecYEG and SecA are the stoichiometric
components
of preprotein translocase.
J. Biol. Chem. 270, 20106-20111.
Economou,
A., and Wickner,
W. (1994). SecA promotes
preprotein
translocation
by undergoing
ATP-driven cycles of membrane insertion
and deinsertion.
Cell 78, 835-843.
Fikes, J.D., and Bassford, P.J.. Jr. (1989). Novel secA alleles improve
export of maltose-binding
protein synthesized
with a defective signal
peptide. J. Bacterial. 171, 402-409.
Gardel, C.. Johnson,
K., Hunt, J., Michaelis,
(1987). secD, a new gene involved in protein
co/i. J. Bacterial. 769, 1286-1290.
S., and Beckwith,
J.
export in fscherichia
Gardel, C.. Johnson, K., Jacq, A., and Beckwith. J. (1990). The se&
locus of E. co/i codes for two membrane proteins required for protein
export. EMBO J. 72, 4205-4216.
Geller. B.L., Movva, N.R., and Wickner,
W. (1986). Both ATP and the
electrochemical
potential are required for optimal assembly
of proOmpAintoEscherichiacoliinner
membranevesicles.
Proc. Natl.Acad.
Sci. USA 83, 4219-4222.
Gill, D.R., Hyde, SC., Higgins, C.F., Valverde, M.A., Mintenig, G.M.,
and Sepulveda, F.V. (1992). Separation of drug transport and chloride
channel functions of the human multidrug resistance
P-glycoprotein.
Cell 77, 23-32.
Gunderson,
K.L., and Kopito, RR. (1995). Conformational
states of
CFTR associated
with channel gating: the role of ATP binding and
hydrolysis.
Cell 82, 231-239.
Hart& F.-U., Lecker, S., Schiebel, E., Hendrick, J.P., and Wickner, W.
(1990). The binding of SecB to SecA to SecYlE mediates preprotein
targeting to the E. coli membrane.
Cell 63, 269-279.
Hendrick,
J.P., and Wickner.
W. (1991). SecA protein needs both
acidic phospholipids
and SecYlE protein for functional,
high-affinity
binding to the E. co/i plasma membrane.
J. Biol. Chem. 266, 2459624600.
Ito, K., Date, T., and Wickner, W. (1980). Synthesis,
assembly into the
cytoplasmic
membrane, and proteolytic processing
of the precursor
of
coliphage Ml3 coat protein. J. Biol. Chem. 255, 2123-2130.
Joly, J.C., and Wickner,
W. (1993). The SecA and SecY subunits of
translocase
are the nearest neighbors
of a translocating
preprotein,
shielding it from phospholipids.
EMBO J. 72, 255-263.
Kim, Y.J., Rajapandi, T., and Oliver, D.B. (1994). SecA protein
posed to the periplasmic
surface of the E. coli inner membrane
active state. Cell 78, 845-853.
is exin its
Kimura, E., Akita, M., Matsuyama,
S., and Mizushima.
S. (1991). Determination of a region in SecA that interacts with presecretory
proteins
in fscherichia
co/i. J. Biol. Chem. 266, 6600-6606.
Klionsky, D.J., Brusilow, W.M., and Simoni. R.D. (1984). In viva evidence for the role of the E subunit as an inhibitor of the protontranslocating
ATPaseof Escherichia
co/i. J. Bacterial. 160,1055-1066.
Lecker, S., Lill, R., Ziegelhoffer,
T.. Bassford,
P.J., Jr., Kumamoto,
C.A., and Wickner, W. (1989). Three purechaperone
proteinsof Escherichia co/i, SecB, trigger factor, and GroEL, form soluble complexes
with precursor
proteins in vitro. EMBO J. 8, 2703-2709.
Lill, R.. Cunningham,
K., Brundage, L.. Ito, K., Oliver, D., and Wickner,
W. (1989). The SecA protein hydrolyzes
ATP and is an essential component of the protein translocation
ATPase of E. co/i. EMBO J. 8.961966.
Lill, R., Dowhan, W., and Wickner, W. (1990). The ATPase activity of
SecA is regulated by acidic phospholipids,
Secy. and the leader and
mature domains of precursor
proteins, Cell 60, 259-269.
Luirink. J., High, S., Wood, H.. Giner, A., Tollervey,
D., and Dobberstein, 8. (1992). Signal-sequence
recognition
by an Escherichia
co/i ribonucleoprotein
complex. Nature 359, 741-743.
Matsuyama,
S., Kimura, E., and Mizushima,
S. (1990). Complementation of two overlapping
fragments
of SecA, a protein translocation
ATPase of Escherichia
co/i, allows ATP binding to its amino-terminal
region. J. Biol. Chem. 265, 8760-8765.
Matsuyama,
S., Fujita, Y., and Mizushima,
S. (1993). SecD is involved
in the release of translocated
secretory
proteins from the cytoplasmic
membrane of Escherichia
co/i. EMBO J. 72, 265-270.
Mitchell, C., and Oliver, D. (1993). Two distinct ATP-binding
domains
are needed to promote protein export by Escherichia
co/i SecA ATPase. Mol. Microbial.
70, 483-497.
Nishiyama,
K.-i., Hanada, M.. and Tokuda, H. (1994). Disruption
of
the gene encoding
~12 (SecG) reveals the direct involvement
and
important function of SecG in the protein translocation
of Escberichia
co/i at low temperature.
EMBO J. 73, 3272-3277.
Oliver, D.B. (1993). SecA protein: autoregulated
ATPase
preprotein insertion and translocation
across the Escherichia
membrane.
Mol. Microbial. 7, 159-165.
Oliver, D.B., and Beckwith,
ponent required for protein
319.
catalyzing
co/i inner
J. (1982). Regulation of a membrane comsecretion in Escherichia
coli. Cell 30,31 l-
Pogliano, K.J., and Beckwith, J. (1993). The Cs set mutants of Escherichia co/i reflect the cold sensitivity
of protein export itself. Genetics
733, 763-773.
Pogliano, J.A., and Beckwith,
J. (1994a). SecD and SecF
protein export in Escherichia
co/i. EMBO J. 73, 554-561.
Pogliano. K.J., and Beckwith, J. (1994b). Genetic
acterization
of the Eschericbia
co/i secD operon
Bacterial. 776, 804-814.
facilitate
and molecular charand its products,
J.
Randall, L.L., and Hardy, S.J. (1995). High selectivitywith
low specificity: how SecB has solved the paradox of chaperone
binding. Trends
Biochem. Sci. 20, 65-69.
Rensing, S.A., and Maier, U.G. (1994). The SecY protein family: comparative analysis and phylogenetic
relationships.
Mol. Phylogenet.
Evol. 3, 187-191.
Sagara, K., Matsuyama,
S.. and Mizushima,
S. (1994). SecFstabilizes
SecD and SecY, components
of the protein translocation
machinery
of the Escherichia
co/icytoplasmic
membrane. J. Bacterial. 7 76,411 l4116.
Sambrook,
J., Fritsch, E.F., and Maniatis, T. (1989). Molecular Cloning: A Laboratory
Manual, Second Edition (Cold Spring Harbor, New
York: Cold Spring Harbor Laboratory
Press).
Schatz, P.J., and Beckwith. J. (1990). Genetic analysis of protein
port in Escherichia
co/i. Annu. Rev. Genet. 24, 215-248.
ex-
Schiebel, E., Driessen,
A.J.M., Hartl, F.-U., and Wickner,
W. (1991).
ApLH+and ATP function at different steps of the catalytic cycle of preprotein translocase.
Cell 64, 927-939.
Shinkai, A., Mei, L.H., Tokuda, H., and Mizushima,
S. (1991). The
conformation
of SecA, as revealed by its protease sensitivity,
is altered
upon interaction with ATP, presecretory
proteins, everted membrane
vesicles, and phospholipids.
J. Biol. Chem. 266, 5827-5833.
Shiozuka,
K., Tani, K., Mizushima,
S., and Tokuda, H. (1990). The
proton motive force lowers the level of ATP required for the in vitro
translocation
of a secretory
protein in Escherichia
co/i. J. Biol. Chem.
265, 18843-16647.
Silhavy, T.J.. Berman, M.I., and Enquist, L.W. (1984). Experiments
with Gene Fusions (Cold Spring Harbor, New York: Cold Spring Harbor
Laboratory
Press), pp. xi-xii.
Tani, K., Shiozuka, K., Tokuda, H., and Mizushima,
S. (1989). In vitro
analysis of the process of translocation
of OmpA across the Escherichia co/i cytoplasmic
membrane.
J. Biol. Chem. 264, 18582-18588.
Ueguchi, C., and Ito, K. (1990). Escherichia
co/i set mutants accumulate a processed
immature form of maltose-binding
protein (MBP), a
late-phase intermediate
in MBP export, J. Bacterial. 772, 5643-5649.
van der Walk. J., Klose,
Freudl, R.. and Driessen,
M., Breukink,
E., Demel, R.A., deKruijff, B.,
A.J.M. (1993). Characterization
of a Bacillus
Regulation
1181
subtilis
release
of SecA
Cycling
SecA mutant protein
from the membrane.
by ATP, SecD,
and SecF
deficient in translocation
Mol. Microbial. 8, 31-42.
ATPase
and
Weiss, J.B., Ray, P.H.. and Bassford,
P.J., Jr. (1988). Purified SecB
protein of Escherichia
co/i retards folding and promotes
membrane
translocation
of the maltose-binding
protein in vitro. Proc. Natl. Acad.
Sci. USA 85, 8978-8982.
Wickner, W., Driessen, A.J.M., and Hark. F.-U. (1991). The enzymology of protein translocation
across the Escherichia
co/i plasma membrane. Annu. Rev. Biochem. 60. 101-124.
Wolfe, P.B., Zwizinski,
C.. and Wickner,
W. (1982). Purification
characterization
of leader peptidase from E. co/i. Meth. Enzymol.
40-46
and
97,
Yamada, H., Matsuyama.
S-i., Tokuda, H., and Mizushima, S. (1989).
A high concentration
of SecA allows proton motive force-independent
translocation
of a model secretory
protein in Escherichia
co/i membrane vesicles. J. Biol. Chem. 264, 18577-18581.