(Received for publication, December 3, 1996, and in revised form, February 3, 1997)
From the Department of Biology, Georgia State University, Atlanta, Georgia 30303
Membrane vesicles were prepared from Escherichia coli cells in which SecE was depleted to 2% of wild-type membranes. SecE depletion had pleiotropic effects; SecD, SecF, SecG, and SecY were decreased 4-6-fold, whereas SecA was increased about 16-fold over that of wild-type membranes. These membranes were substantially active in the in vitro translocation of proOmpA, which was mediated by the SecA pathway since it was inhibited by azide. Similar substantial translocation activities were observed for proLamB and proLpp in the SecE-depleted membranes. However, the translocation of proPhoA was more severely impaired. These data indicate that SecE may enhance but is not obligatorily required for the translocation of at least certain precursors, and suggest that the effects of the SecE depletion on protein translocation may be precursor-dependent.
The general secretion pathway of SecA-dependent protein translocation in Escherichia coli has been studied extensively (1). The Sec machinery is generally believed to be composed of several inner membrane proteins: SecA, SecD, SecE, SecF, SecG, and SecY (PrlA). SecA plays a central role in this machinery, and its requirement for secretion has been well established (2-4). SecA initiates an early step of translocation by hydrolyzing ATP (5), after forming a complex with secretory precursors and the cytoplasmic chaperone, SecB (6). The energy supplied by ATP hydrolysis is believed to drive the insertion and deinsertion of SecA onto and off the plasma membrane (7, 8). However, recent data indicate that some SecAs are permanently integrated into membrane and are active in protein translocation (9, 10). SecD and SecF are also integral membrane proteins and are necessary for efficient protein translocation but not for cell viability (11). Both are presumed to be involved in a late step of the translocation reaction (12, 13). However, they do not exhibit significant effects on the translocation activity of reconstituted membranes (14).
Together, SecE, SecG, and SecY appear to form an integral part of the translocation machinery in the inner membrane since these proteins are copurified chromatographically and coimmunoprecipitated with anti-SecY antibody (15-17). Additionally, when overexpressed, these three proteins can be recovered in the plasma membrane fraction (18). However, SecG is not essential for translocation since reconstituted membranes without SecG are still active in protein export, although the translocation activity can be enhanced by the addition of SecG (19). In addition, strains deleted of secG grow normally but are cold-sensitive (20). On the other hand, SecE and SecY are thought to be indispensable components of protein export in reconstituted membranes (15, 21, 22). Nevertheless, it has been reported that reconstituted membranes without SecY are active in protein translocation (4, 23).
The requirement for SecE is addressed in this study. SecE is a 13.6-kDa integral membrane protein, spanning E. coli inner membrane three times with its amino terminus in the cytoplasm (24). Complementation studies revealed that the third transmembrane segment and its adjacent cytoplasmic domain is sufficient for the function of SecE (25, 26). Genetic data showed that the secE gene product is essential for cell growth (25), and its mutation causes precursor accumulation (27). In this study, we depleted cells of SecE in vivo to elucidate the role of SecE in protein translocation in vitro. The results show that membranes prepared from SecE-depleted cells still have substantial translocation activity for several precursors, indicating that SecE is not essential for in vitro translocation of certain secretory proteins.
Wild-type E. coli
MC1000 (28) membranes were prepared as described (29). SecE-depleted
membranes were prepared from E. coli strain PS289 (MC1000,
leu+, ara+,
phoAPvuII, pcnB80,
zadL::Tn10 (Tcs Strr),
secE
19-111,
recA::cat/pBAD22
secE+), a gift from C. Murphy and J. Beckwith,
Harvard Medical School, Boston, MA (30), carrying a chromosomal
deletion in secE and a plasmid with secE behind
an arabinose-activable promoter. PS289 cells were grown to mid-log
phase at 37 °C in LinA medium (29) containing 0.2%
L-arabinose, harvested, washed twice, and resuspended to
the same medium without arabinose. Cells were inoculated to 0.1 A600 unit in LinA supplemented with either 0.5%
glucose or 0.2% arabinose, and harvested after 4 doublings (for
arabinose medium) or after growth ceased (for glucose medium). Inverted inner membrane vesicles from PS289 cells were prepared as described (29).
35S-proLamB and 35S-proPhoA were synthesized from plasmid pEB (lamB under control of pT7-6, constructed in this laboratory) and pAE2.2 (phoA under control of T7 RNA polymerase promoter, a gift from D. Oliver, Wesleyan University, Middletown, CT), respectively, using an in vitro transcription-translation coupled system (31) with addition of 2% polyethylene glycol 6000. 35S-proOmpA and 35S-proLpp were synthesized with mRNA prepared from OmpA- and Lpp-overproducing cells, respectively, as described (29). After translation, ribosomes and endogenous membranes were removed by centrifugation, and this posttranslational supernatant was used as labeled precursors.
In Vitro Protein TranslocationA typical 100-µl translocation reaction contained 70 µl of 35S-labeled precursor solution, 2 µl of an ATP-regenerating stock (29), and 25 µg of inverted inner membrane vesicles. Purified SecA was added to 20 µg/ml where indicated. The reaction was carried out at 37 °C for 10 min unless otherwise indicated, and then followed the procedures as described (29).
Carbonate, Urea, and Heparin TreatmentMembranes (40 µg) were suspended in 100 µl of 0.1 M Na2CO3, or of 6 M urea, incubated at 0 °C for 30 min, then recovered by centrifugation with Beckman TLA 100.2 rotor at 73,000 rpm for 2 h over a 200-µl cushion (0.5 M sucrose in DTK buffer (1 mM dithiothreitol, 10 mM Tris-HCl, pH 7.6, and 50 mM KCl)). Heparin treatment was the same as described (4).
Purification of SecA, 35S-SecA, ProOmpA, and SecBPurified SecA and 35S-SecA are from laboratory stock purified from pT7-secA (2). ProOmpA was purified from BA13/pAM103 (2, 10). Purified SecB was from laboratory stock obtained as described (32).
Other MethodsSDS-PAGE1 was
carried out according to Laemmli (33). For Lpp, a Tricine-15% SDS-PAGE
system was performed by the method of Schagger and von Jagow (34).
Protein amounts were determined by Bio-Rad assay kit according to
Bradford (35) with -globulin as a standard. All autoradiogram
quantitations were carried out by PDI Image Analyzing System (Protein
Databases Inc., Huntington Station, NY).
Proteinase K was from Boehringer Mannheim. [35S]Methionine was from DuPont NEN. All other chemicals are of reagent grade and were obtained from Sigma or other commercial sources.
Without arabinose, growth of PS289 cells ceased after 3 doublings, which is consistent with the findings that secE
is essential for cell growth (25). Inverted inner membrane vesicles
were prepared from these cells, and all Sec proteins in the membranes were examined by immunoblots. The depletion of SecE in these membranes was confirmed (Fig. 1). Surprisingly, not only SecY and
SecG decreased as expected; SecD and SecF were also reduced (Fig. 1).
The amount of membrane used for immunoblotting was adjusted to
visualize the relative levels of Sec proteins, and were quantitated
(Table I). The results revealed that membranes isolated
from PS289 cells grown with arabinose (referred to as
SecE++ membranes hereafter), contained around 12-fold more
SecE (expressed from multicopy plasmid), 2-fold more SecA and SecG, and
normal amounts of SecD, SecF, and SecY as compared with isogenic
wild-type MC1000 membranes. SecE in the membranes from PS289 cells
shifted from arabinose to glucose media was depleted 600-fold and
reduced to about 2% of wild-type MC1000 membranes. The depletion of
SecE had pleiotropic effects on other Sec proteins: SecD, SecY, SecF, and SecG also decreased to about 25, 15, 22, and 17%, respectively. In
contrast, SecA increased about 16-fold in these membranes. This
increase was due to increased expression and not to the redistribution of SecA protein between the cytosol and the membranes (data not shown).
|
SecE-depleted membranes were examined for in
vitro proOmpA translocation activity. Surprisingly, SecE-depleted
membranes had the same translocation activity as wild-type MC1000
membranes with either SecA-depleted S30 or normal SecA-containing S30
that were used to synthesize nascent proOmpA (Fig. 2).
More importantly, translocation in SecE-depleted membranes included the
processing of precursor form to mature OmpA, and the translocated
proteins were still resistant to 1 mg/ml proteinase K, indicating that SecE-depleted membranes were not porous and behaved similarly to normal
membranes in translocation. (The SecE-depleted membranes, however, were
more sensitive to freeze-thawing.) Since SecE-depleted membranes had 16 times more SecA than wild-type membranes, this might account for the
high activity of these membranes. To test for this, additional SecA was
added in translocation assays. The translocation activity of wild-type
MC1000 or SecE++ membranes increased up to 250%, reaching
saturation at 20 µg/ml additional SecA (Fig.
3A). For SecE-depleted membranes, 20 µg/ml SecA only enhanced translocation activity to a maximum of 1.4-fold. Even under this condition, the SecE-depleted membranes still had about
50% translocation activity of MC1000 membranes. The lower enhancement
of translocation activity for SecE-depleted membranes by SecA was not
due to reduced binding of SecA, since SecE-depleted membranes had the
same SecA binding capacity as MC1000 and SecE++ membranes
(data not shown).
The kinetics of translocation in SecE-depleted, SecE++, and MC1000 membranes were compared (Fig. 3B). The translocation kinetic curves were virtually linear up to 5 min, indicating that SecE-depleted membranes are indeed active, albeit less efficient, in protein translocation. The ratio of translocation activity of SecE-depleted membranes reached 68% relative to MC1000 membranes after 10 min.
Translocation Capacity in SecE-depleted MembranesTo
determine whether the apparent high activity of SecE-depleted membranes
was due to the limited amount of nascent precursors, and since the
kinetic curves in Fig. 3B suggested that in the later time
points this may be significant, saturation experiments were carried
out. Membrane amounts used for the translocation assays were decreased
from 25 µg to 5 µg, and titrated with post-translational 35S-proOmpA supernatant (Fig. 4) to
determine the saturation point. Translocation activities of MC1000 and
SecE++ membranes showed that saturation was achieved with
140 µl of supernatant, and that of SecE-depleted membranes was 70 µl of supernatant. The translocation activity of SecE-depleted
membranes with 140 µl of supernatant was 45% of wild-type membranes,
this activity is comparable to that observed in typical translocation assays in which 5-fold more (25 µg) membrane was used (Fig.
3B).
To further compare the translocation capacity of these membranes,
competition experiments were carried out. As shown in Fig. 5A, the amounts of nonlabeled purified
proOmpA needed for 50% competition in MC1000 and SecE-depleted
membranes were 127 and 54 µg/ml, respectively. Thus, SecE-depleted
membranes had about 40% capacity of wild-type membranes for proOmpA
translocation, which is consistent with the translocation activity
ratio mentioned above. The concentration of purified proOmpA used in
this experiment was extremely high relative to nascent
35S-proOmpA, which was estimated as 0.1 µg/ml in the
posttranslational supernatant (data not shown). Since proOmpA was
purified in 6 M urea, it may behave differently or may lose
its competence for translocation when it is diluted 50-fold in
translocation reaction. Therefore, similar experiments were carried out
with nonlabeled nascent proOmpA. The competition results are similar to
those with purified proOmpA (Fig. 5B).
Without additional SecA, SecE-depleted membranes had the same
translocation activity as MC1000 membranes, and just slightly lower
than SecE++ membranes with normal S30 (Fig. 2). Thus the
competition experiment was also performed with nonlabeled purified
proOmpA in the absence of SecA. As expected, there was no significant
difference between competition curves of SecE++ and
SecE membranes (Fig. 5C), supporting the
previous observation that SecE-depleted membranes are active in
translocation and the notion that the translocation characteristics of
these membranes are similar.
SecE-depleted
membranes had reduced levels of other Sec proteins, except for SecA,
which was increased. These membranes were still active in translocation
of proOmpA without exogenous SecA (Fig. 2, lane 3),
indicating that the SecA present in the membranes is sufficient for
activities. To further examine the dependence of SecA in SecE-depleted
membranes, the SecA inhibitor azide was added during translocation. The
translocation activity of SecE-depleted membranes was significantly
inhibited by azide (Fig. 6), indicating that
translocation in SecE-depleted membranes remained
SecA-dependent. In fact, the translocation of proOmpA in
SecE-depleted membranes were more sensitive to azide than those in
SecE++ and MC1000 membranes. In the presence of 2 µg of
SecA, there was a 2-fold difference in I50 (50%
inhibition) between these two membranes: the I50 was 1.7 mM in SecE-depleted membranes, and 3.7 mM in
SecE++ membranes. The difference was even more pronounced
in the presence of 6 µg of SecA; the I50 values of
SecE-depleted and SecE++ membranes were 2.3 and 8 mM, respectively (data not shown).
Translocation of proOmpA and proLamB are known to be
SecB-dependent (36-38). In our in vitro system,
the addition of purified SecB still enhanced the translocation of
proOmpA and proLamB across both SecE++ and
SecE membranes (data not shown). This evidence also
indicates that translocation across SecE
membranes is
dependent on Sec system.
SecA has been found to have cytoplasmic, peripheral, and integral membrane forms (39). The integral form has been suggested to be the active form for translocation (4, 9, 10). To determine what form of SecA is prevalent in SecE-depleted membranes, different membrane extractions were performed. Table II shows that about two-thirds of the membrane-bound SecA in SecE-depleted membranes were resistant to extraction by 0.1 M sodium carbonate (pH 11), 6 M urea, or 1 M heparin, suggesting that the two-thirds of SecA may be in the integral membrane form that is active for protein translocation.
|
To determine whether the substantial
translocation activity in SecE-depleted membranes is limited to
proOmpA, the effects of SecE depletion on the translocation of proLamB,
proLpp, and proPhoA were examined in our in vitro system. It
is notable that SecE++ membranes had higher activity than
wild-type membranes for those four preproteins tested (Figs.
3B and 7). The activity ratios of SecE-depleted membranes to
wild-type membranes for LamB and Lpp precursors were 50 and 38%,
respectively, after 2 min of translocation reaction (Fig.
7). After 10 min, the ratios changed to 80 and 63%,
respectively. Thus, with a 50-fold decrease of SecE, the SecE-depleted
membranes still contained substantial translocation activity for LamB
and Lpp precursors, similar to proOmpA.
Since OmpA, LamB, and Lpp all are outer membrane proteins, the kinetics of translocation of the soluble periplasmic protein, PhoA, was also examined. Interestingly, the translocation of proPhoA with SecE-depleted membranes was not as active as those of the other three precursor proteins. The activity of SecE-depleted membranes for proPhoA was only 23% of wild-type membrane after translocation for 2 min, and 36% after 10 min (Fig. 7). Thus the requirement of SecE or other Sec proteins in translocation may be precursor-dependent. It is unclear whether the destination of secretory protein is critical for SecE dependence in translocation.
In SecE-depleted membranes, SecA increases 16-fold. The increased expression of SecA by depletion of SecE was predictable since the secE gene was identified by SecA derepression (27), a common phenotype observed in most mutants with protein secretion defects (28, 40, 41). Genetic evidence has shown that SecE is functionally in excess and is the first membrane protein to interact with SecA (42, 43). The SecYEG complex has been proposed to be a receptor for SecA, existing as a high affinity binding site (6, 7, 18). However, our SecE-depleted membranes only contain 2% SecE, 15% SecY, and 17% SecG, but have 16 times more SecA. They also have the same binding capacity for exogenous SecA as wild-type membranes. Additionally, these SecA proteins are active in protein translocation (Fig. 2, lane 3), and most of them are found in integral form in membrane (Table II). Although the nature of the binding of SecA to SecE-depleted membranes is not certain, it is clear that the binding of SecA to the SecYEG complex is not obligatorily required for protein translocation. Since there is little SecE and much reduced SecYG in SecE-depleted membranes, it is also evident that the YEG membrane complex is not the essential membrane-embedded domain of preprotein translocase, in contrast to what has been proposed previously (16, 18).
The interdependence of the Sec system is well documented. SecY overproduction is SecE-dependent (44). However, SecE overproduction does not require the overproduction of SecY (45). This is in agreement with our data that SecE++ membranes have 12-fold more SecE but a normal amount of SecY. Additionally, SecF can stabilize SecD and SecY in an overproducing strain (46). (On the other hand, depletion of SecD and SecF does not significantly influence the amounts of SecY and SecE (13).) Furthermore, the amounts of SecY and SecG are also decreased in the SecE reduced background, secE501 (17). Combining with our data that the depletion of SecE is accompanied by the depletion of other Sec proteins except for SecA, these results suggest an interdependence of all the Sec proteins. Traxler and Murphy (30) have shown that integral membrane proteins are not inserted and degraded even under mild SecE-depleted condition, providing a possible explanation for the reduction of other Sec proteins.
Is There a Substitute for SecE?SecE has been reported to be
essential for protein translocation across reconstituted
proteoliposomes providing the basis for current dogma that SecYEG forms
the translocation channel (15, 21, 45). It is not yet clear why the
requirements of SecE in vitro are different in various
protein translocation systems. The translocation of SecE-depleted
membranes is at least as active as these reconstituted systems.
However, the efficient processing of secretory precursors with these
reconstituted proteoliposomes has never been reported. Furthermore, the
possibility of functional substitution of any Sec protein by another
protein has not been ruled out. It is possible that there is another
protein(s) that can replace the function of SecE; some other non-Sec
protein(s) may be involved in active translocation across SecE-depleted
membranes. First, YajC, a product encoded from the gene located within
secD operon, has been shown to interact with SecY (47).
Second, phage shock protein A (PspA), which is induced in response to
heat, ethanol, osmotic shock, and infection by filamentous
bacteriophages (48), has been reported to have a stimulatory effect on
protein export in E. coli (49). Third, another new gene
product, Syd (suppressor of secY dominance), has been found
to suppress secYd1 mutant, and to stabilize
overproduced SecY (50). The possible involvement of these proteins in
translocation across SecE-depleted membranes needs to be
investigated.
Translocation channels in SecE-depleted membranes seem to have different efficiencies for translocating different proteins. The translocation activities of proOmpA, proLamB, and proLpp are only moderately inhibited in SecE-depleted membranes even in the presence of excess SecA. However, for precursor of PhoA (alkaline phosphatase), the translocation activity is reduced to 20-36% relative to wild-type membranes, implying that SecE, or other Sec proteins, might be required for the translocation of proPhoA. Alternatively, some other factor(s) needed for translocation of proPhoA might be reduced in SecE-depleted membranes. Indeed, protein export was found to be severely defective in vivo with the same strain used here, PS289 (30).
The cold-sensitive SecE mutants raise an interesting question. The secE mutation is lethal to the cell at nonpermissive temperatures. Some mutants express less than 50% SecE even at permissive temperatures (25). However, cold sensitivity has been found to be an inherent property of the secretion pathway itself (51). The reduced synthesis of SecE in these mutants, therefore, may not be the direct reason for lethality at nonpermissive temperatures, since we also found that the translocation activity of SecE-depleted membranes is not cold-sensitive (data not shown). (It has been reported more recently that the secE gene is absent in the genome of Mycoplasma genitalium (52) and Methanococcus jannaschii (53), both of which are among the smallest free-living organisms. However, this needs to be further confirmed.) If SecE is not essential for translocation, why does cell growth halt when SecE is depleted? One possibility is that SecE may have another physiological function(s) relative to cell growth. In other words, SecE depletion might cause secondary effects, which lead to cessation of growth. Thus, it may be possible that some other proteins, which are directly involved in cell growth, are affected by depletion of SecE. Alternatively, we favor the notion that, like PhoA shown here, translocation of some other essential proteins are more strictly dependent on SecE. In this case, the specific dependence can explain the cessation of cell growth in vivo after SecE depletion, and the genetic evidences that SecE is essential.
SecA Is the Only Indispensable Component of the Sec MachineryEven with a reduction of SecE down to 2%, SecE-depleted membranes are still active in translocation of proOmpA, proLamB, and proLpp, at about 50% of the activity of wild-type membranes in the presence of excess SecA, and at an even higher ratio in its absence. Therefore we conclude that SecE is not absolutely essential for in vitro translocation of, at least, certain secretory preproteins (but probably not all precursors). Then, which Sec protein is responsible for the translocation activity in SecE-depleted membranes for these precursors? SecD and SecF have been shown to only have an enhancing effect on translocation (14), and so does SecG (20). Even the requirement for SecY, which has 10 transmembrane segments, is still controversial, since it has been shown that translocation can occur with reconstituted membranes depleted of SecY (4, 23), and we have confirmed this as well.2 Thus, the best candidate for translocation activities in SecE-depleted membranes is SecA. We have recently found that some SecA is permanently embedded in membranes in an active integral form and may form part of the translocation channel (10). Therefore, we propose that SecA is the only indispensable component of the Sec machinery, although it may not be sufficient alone. It is possible that, together with all other Sec proteins in proper ratios as in normal membranes, SecA can translocate proteins more efficiently. However, when other Sec proteins are diminished, SecA is still active in transporting proteins, albeit with less efficiency. This interpretation is analogous to earlier observations on the energy requirement for protein translocation; ATP is essential, but less efficient, as energy source for translocation without proton motive force (54-56). Interestingly, ATP is required for SecA function. Thus most of the translocation activity of SecE-depleted membranes may come from the 16-fold increase in SecA, in conjunction with other reduced Sec proteins, as active translocation sites with reduced efficiency. That may explain why SecE-depleted membranes are more sensitive to azide than SecE++ and MC1000 membranes, and why their translocation activity cannot be enhanced as much as wild-type membranes by exogenous SecA, even though both membranes have the same binding capacity for it.
Thus, while SecE depletion may have caused severe defects leading to cessation of cell growth, it is also clear from the data presented here that SecE is not essential for translocation of some proteins but may be required for other precursors. It follows that since the YEG complex is not obligatory for protein translocation in general, the active integral form of SecA may play a more important role than previously realized in forming the translocation channel in the inner membrane.
We thank J. Beckwith, P. Schatz, D. Oliver, and C. Murphy for strains, plasmids, and SecE peptide antibodies; P. Li for establishing the in vitro cotranscription-translation system; and A. Boyer for reading of the manuscript. We also acknowledge the numerous discussions with the laboratory members who also contributed purified proteins, plasmids, and antibodies.