Dc Stimulates Membrane Translocation of the C-terminal Part of a Signal Sequence*

For several proteins in Escherichia coli it has been shown that the protonmotive force (pmf) dependence of translocation can be varied with the signal sequence composition, suggesting an effect of the pmf on the signal sequence. To test this possibility, we analyzed the effect of the membrane potential on translocation of the signal sequence. For this purpose, a precursor peptide was used (SP 1 7), corresponding to the signal sequence of PhoE with the first seven amino acids of the mature part that can be processed by purified leader peptidase. Translocation was studied in pure lipid vesicles containing leader peptidase, with its active site inside the vesicles. In the presence of a positive inside Dc , the amount of processing of SP 1 7 was significantly higher than without a Dc , indicating that the translocation of the cleavage region is stimulated by Dc . Replacement of the helix-breaking glycine residue at position 2 10 in the signal sequence for a leucine abolished the effect of Dc on the translocation of the cleavage region. It is con-cluded that Dc directly acts on the wild type signal sequence by stimulating the translocation of its C terminus. We propose that Dc acts on the signal sequence by stretching it into a transmembrane orientation. Proteins destined for export out of the cell are synthesized as precursor proteins with an N-terminal extension, the signal sequence. Signal sequences are composed of a positively charged N-terminal region, followed by a hydrophobic core region of 7–13 residues and a more polar C terminus containing the signal peptidase cleavage site (1). In Escherichia coli , the signal sequence is necessary and in most cases sufficient for translocation across the peptides were injected from a 1 mg/ml Me 2 SO solution in the subphase through a hole in the edge of the dish. All experiments were performed with saturating peptide concentrations in the subphase (3 m g/ml). To follow the kinetics of monolayer insertion, the pressure changes after addition of SP 1 7 or SP 1 7 G( 2 10)L were followed for 90 min. The area per molecule was determined as described (32). General Methods— Peptide concentrations were determined by the microbichinchonic acid protein assay reagent with bovine serum albumin as a standard (Pierce). Phospholipid concentration was determined by phosphorus assay (33). Analysis of the processing of the precursor peptides was performed by Tricine SDS-polyacrylamide gel electro- phoresis (34) and exposure of the gel in a PhosphorImager (Molecular Dynamics). The percentage processing was defined as the amount of processed form/(precursor 1 processed form) *100%, determined by the density of the bands after quantification with the program Image Quant.

For several proteins in Escherichia coli it has been shown that the protonmotive force (pmf) dependence of translocation can be varied with the signal sequence composition, suggesting an effect of the pmf on the signal sequence. To test this possibility, we analyzed the effect of the membrane potential on translocation of the signal sequence. For this purpose, a precursor peptide was used (SP؉7), corresponding to the signal sequence of PhoE with the first seven amino acids of the mature part that can be processed by purified leader peptidase. Translocation was studied in pure lipid vesicles containing leader peptidase, with its active site inside the vesicles. In the presence of a positive inside ⌬, the amount of processing of SP؉7 was significantly higher than without a ⌬, indicating that the translocation of the cleavage region is stimulated by ⌬. Replacement of the helix-breaking glycine residue at position ؊10 in the signal sequence for a leucine abolished the effect of ⌬ on the translocation of the cleavage region. It is concluded that ⌬ directly acts on the wild type signal sequence by stimulating the translocation of its C terminus. We propose that ⌬ acts on the signal sequence by stretching it into a transmembrane orientation.
Proteins destined for export out of the cell are synthesized as precursor proteins with an N-terminal extension, the signal sequence. Signal sequences are composed of a positively charged N-terminal region, followed by a hydrophobic core region of 7-13 residues and a more polar C terminus containing the signal peptidase cleavage site (1). In Escherichia coli, the signal sequence is necessary and in most cases sufficient for translocation across the cytoplasmic membrane.
Most periplasmic and outer membrane proteins in E. coli are translocated through the general secretion machinery (for a recent review, see Ref. 2). The central part of this translocase comprises the peripheral membrane subunit SecA and the integral membrane protein complex SecYEG. Besides proteinaceous components, negatively charged lipids (3,4) and nonbilayer lipids (5) are essential for efficient translocation. During or shortly after translocation, leader peptidase (Lep) 1 removes the signal sequence from the precursor protein. The catalytic site of Lep is located in a large periplasmic domain (the P2 loop) that can penetrate between lipids with a specificity for phosphatidylethanolamine (6). The energy for translocation is derived from ATP hydrolysis by SecA and the protonmotive force (pmf), that is composed of an electrical component, ⌬, and a proton gradient, ⌬pH (positive and acidic in the periplasm). The pmf plays a role in both early and late steps of translocation (7,8).
Signal sequences are involved in various stages of the secretion pathway. They are necessary to target the precursor protein into the secretion machinery, and they retard folding of the nascent polypeptide, thereby increasing the time for the preprotein to stay in a translocation competent conformation (9). In early steps of translocation, the signal sequence is shown to interact with SecA (10), SecE, and SecY (11). The signal sequence was also shown to interact with Ffh protein, a component of the bacterial analogue of the signal recognition particle (12). Furthermore, signal sequences can interact with the phospholipids of the inner membrane, both in Sec-dependent (13) and Sec-independent translocation, as was observed for M13 procoat protein (14,15). From circular dichroism (CD) studies with different synthetic signal peptides, it was derived that upon interaction with anionic lipids the signal peptide undergoes a conformational change and adopts an ␣-helical structure that may be essential for its functionality (16 -19). NMR studies of the signal peptides of OmpA (20), LamB (21) and PhoE (22) showed that the ␣-helix conformation is disrupted toward the C-terminal end of the hydrophobic core near a helix breaking residue. Such residues are frequently found in signal sequences, like for example glycine at position Ϫ10 (Gly Ϫ10 ) in the signal sequence of the outer membrane protein PhoE.
Translocation of PhoE was found to be largely stimulated by the pmf (8). Remarkably, a single mutation of the Gly Ϫ10 for an ␣-helix promoting residue such as leucine abolished the pmf dependence for translocation. In this study, we investigate the possibility that the pmf can directly act on a signal sequence. For this purpose, we use the signal peptide of PhoE with the first part of the mature protein as a precursor peptide in a simplified reconstitution system with purified Lep, active only inside the vesicles. We demonstrate that the translocation of the cleavage region of the signal sequence across the bilayer is largely stimulated in the presence of ⌬. This stimulating effect was not observed with an analogous precursor peptide in which the Gly Ϫ10 was replaced by a Leu residue. This demonstrates that ⌬ can directly and specifically act on the signal sequence and suggests that ⌬ affects the stretching of the signal sequence into a transmembrane conformation and thereby facilitates initiation of translocation.

EXPERIMENTAL PROCEDURES
Materials-Lep was purified as Lep(His) 6 by immobilized metal affinity chromatography as described (23). The C-terminal His-tag has no effect on the activity of Lep (23). 1,2-Dioleoyl-sn-glycero-3-phosphoglycerol (DOPG) and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) were purchased from Avanti Polar Lipids Inc. Proteinase K was * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. obtained from Boehringer (Germany), n-octyl-␤-D-glucopyranoside (octyl glucoside), phenylmethylsulfonyl fluoride, and bovine serum albumin from Sigma (United Kingdom). 14 C-Formaldehyde was purchased from Amersham Pharmacia Biotech (specific activity 54 mCi/mmol). All other chemicals were analytical grade or better.
Precursor Peptides-The signal peptide of PhoE, elongated with the first seven amino acids of the mature part of PhoE (MKKSTLALVVM-GIVASASVQA-AEIYNKD, referred to as SPϩ7) was synthesized as described (24). The SPϩ7 in which the glycine residue at Ϫ10 position was replaced by a leucine residue (MKKSTLALVVMLIVASASVQA-AEIYNKD, SPϩ7 G(Ϫ10)L) was synthesized by Genosys Biotechnologies Inc (United Kingdom). SPϩ7 (G-10)L was further purified by semipreparative reversed phase-high pressure liquid chromatography, following the methods described in (25), and characterized by mass spectrometry. The purity of both peptides was verified by analytical reversed phase-high pressure liquid chromatography and determined to be Ͼ95%. Before use, the peptides were first dissolved in trifluoroacetic acid, dried under a stream of nitrogen, and finally dissolved in trifluoroethanol or dimethyl sulfoxide (Me 2 SO). 14 C-Radiolabeling of the peptides was performed by reductive methylation of the lysine residues, based on the procedure of Dottavio-Martin and Ravel (26). This labeling method leaves the charges intact. The labeling reaction was done as described (24) except that excess of label was removed by acetone precipitation of the peptide. Both peptides usually had a specific activity of around 30,000 dpm/nmol.
Processing Assay-The processing assay in a mixed micellar system was based on the method as described by van Klompenburg et al. (23). Purified Lep (0.5 g) was diluted in 50 l of assay buffer (10 mM Tris, pH 8.0, 1% octyl glucoside) and mixed with a dry lipid film of pure DOPG (lipid:detergent, 1:10 molar ratio). 14 C-Labeled peptide (1 g) was added from a 1 mg/ml trifluroethanol stock solution to the mixed micellar solution and incubated at room temperature. To stop the reaction and remove the lipids from the samples, the peptides were precipitated by addition of 1 ml of acetone (room temperature), followed by addition of 10 g of bovine serum albumin (from 2 mg/ml solution) for more efficient precipitation.
Reconstitution of Lep into Liposomes-Purified Lep was reconstituted into liposomes by the detergent dilution method based on the method described by Ohno-Iwashita et al. (27) with a few modifications. A lipid film of DOPG and DOPE (1:4 molar ratio) was prepared of appropriate amounts of lipids from chloroform stock solutions. After mixing, chloroform was removed by a stream of nitrogen, and the lipids were thoroughly dried under vacuum. Lep (dissolved in 1% (w/v) octyl glucoside containing 10 mM Tris buffer, pH 8.0) was added to the lipid film in a 1:1,000 molar ratio with respect to the lipids. Extra octyl glucoside in 10 mM Tris, pH 8.0, was added to adjust the lipid:detergent molar ratio to 1:10. Na ϩ -containing liposomes were formed by diluting the mixed micellar solution at least 30 times in a Na ϩ buffer without octyl glucoside (10 mM Hepes, pH 8.0, 100 mM Na 2 SO 4 ). To obtain liposomes with K ϩ inside, the mixed micellar solution was diluted in K ϩ buffer (10 mM Hepes, pH 8.0, 100 mM K 2 SO 4 ). After incubation overnight on ice, the proteoliposomes were collected by ultracentrifugation (60 min, TLA100.3 rotor, 60,000 rpm, 4°C). With the used detergent: lipid ratio virtually 100% single walled vesicles are produced (28). In every experiment freshly prepared vesicles were used.
Characterization of the Reconstitution Vesicles-The topology of Lep in the reconstitution vesicles was determined as described (23). The diameter of the proteoliposomes and of inner membrane vesicles of E. coli strain MC4100 were measured with the Zetasizer 3000 (Malvern Instruments).
Generation of ⌬-A transmembrane potential (⌬, positive inside) was generated by diluting Na ϩ -containing proteoliposomes 50 times in K ϩ buffer followed by addition of valinomycin in a 1:10 4 molar ratio with respect to the lipids. A reversed ⌬ (negative inside) was generated after addition of valinomycin to K ϩ -containing proteoliposomes diluted in Na ϩ buffer. As controls (no ⌬), valinomycin was added to Na ϩcontaining proteoliposomes diluted in Na ϩ buffer or K ϩ -containing proteoliposomes diluted in K ϩ buffer, respectively. The formation of a positive inside ⌬ in Na ϩ -containing vesicles in K ϩ buffer (100 M final lipid concentration) was monitored with the membrane potential sensitive fluorescent dye oxonol VI (29). The negative inside ⌬ in K ϩcontaining vesicles in Na ϩ buffer (25 M final lipid concentration) was monitored using 3,3Ј-diethylthiodicarbocyanine iodide, diSC 2 -(5) (30). Oxonol VI and diSC 2 -(5) were added from 0.5 mM stock solutions in Me 2 SO to a final concentration of 1 M. To dissipate the ⌬, gramicidin D (0.1 mM in Me 2 SO) was added in a 1:400 molar ratio with respect to the lipids. The fluorescence emission of oxonol VI was measured at 634 nm with excitation at 599 nm and of diSC 2 -(5) at 670 nm with excitation at 650 nm using a SLM AMINCO SPF-500C spectrofluorimeter.
Translocation Assay-The proteoliposomes were treated with 200 g/ml proteinase K to remove the Lep molecules with their active site outside the vesicles. After 20 min at room temperature, 2 mM phenylmethylsulfonyl fluoride was added, and the mixture was incubated for 5 min on ice. The "shaved" proteoliposomes were collected by centrifugation, washed, and resuspended in buffer to a final phospholipid concentration of 10 mM. 1 l of vesicle suspension was diluted 50 times in buffer followed by addition of valinomycin to generate either a positive inside ⌬ (Na ϩ inside/K ϩ outside), a negative inside ⌬ (K ϩ inside/Na ϩ outside), or to act as a control (Na ϩ inside/Na ϩ outside or K ϩ inside/K ϩ outside, respectively). Immediately after the addition of valinomycin, 1 g of 14 C-labeled SPϩ7 or SPϩ7 G(Ϫ10)L was added from a 1 mg/ml trifluoroethanol stock solution and incubated at room temperature. To stop the incubation, the peptides were precipitated as in the processing assay described above.
Monolayer Experiments-Monolayer surface pressures were measured by the (platinum) Wilhelmy plate method (31) in a thermostatically controlled box at 37°C, using a Cahn 2000 microbalance. The monomolecular lipid layers were spread from a pure DOPG solution in CHCl 3 /CH 3 OH (3:1, v/v) at the air/water interface in a Teflon dish to give an initial surface pressure of 20 mN/m. The Teflon dish had a volume of 5 ml and a constant area of 8.81 cm 2 . As a subphase, a buffer consisting of 10 mM Tris, pH 8.0, and 100 mM NaCl was used and continuously stirred with a magnetic bar. The buffer was filtered through a 0.22-m pore filter and degassed before use. The precursor peptides were injected from a 1 mg/ml Me 2 SO solution in the subphase through a hole in the edge of the dish. All experiments were performed with saturating peptide concentrations in the subphase (3 g/ml). To follow the kinetics of monolayer insertion, the pressure changes after addition of SPϩ7 or SPϩ7 G(Ϫ10)L were followed for 90 min. The area per molecule was determined as described (32).
General Methods-Peptide concentrations were determined by the microbichinchonic acid protein assay reagent with bovine serum albumin as a standard (Pierce). Phospholipid concentration was determined by phosphorus assay (33). Analysis of the processing of the precursor peptides was performed by Tricine SDS-polyacrylamide gel electrophoresis (34) and exposure of the gel in a PhosphorImager (Molecular Dynamics). The percentage processing was defined as the amount of processed form/(precursor ϩ processed form) *100%, determined by the density of the bands after quantification with the program Image Quant.

RESULTS
Strategy-To investigate the effect of the pmf on the translocation of the signal sequence, a precursor peptide was used, corresponding to the signal sequence of PhoE with the first seven amino acids of the mature part (SPϩ7). In this peptide the cleavage region for Lep is intact. Processing of this precursor peptide in the lumen of reconstitution vesicles by Lep can then in principle be used as a tool for translocation of the C terminus of the signal sequence. In such systems a ⌬ can be generated by applying ion diffusion potentials. No other proteins were incorporated in the vesicles to exclude possible effects of ⌬ via these proteins. To study the effect of conformational flexibility of the signal sequence and to be able to compare the results with those obtained for intact precursors under translocation conditions also a precursor peptide with a Leu at the position of the helix-breaker was used (SPϩ7 G(Ϫ10)L).
Cleavage of the Precursor Peptides by Leader Peptidase in Mixed Micelles-A prerequisite for the translocation assay is that Lep can process the precursor peptides and that this can be analyzed. Processing was tested in mixed micelles of octyl glucoside and DOPG. The inset of Fig. 1 shows the processing reaction in time, analyzed on Tricine gels. Both precursor peptides are cleaved, resulting in a decrease of the intensity of the precursor band and the appearance of a band at a lower position, corresponding to the signal peptide. The amount of processing was determined as the ratio of the intensity of processed form over the total intensity of the two bands. Processing of both peptides occurs rapidly and levels off around 55 and 65% for SPϩ7 and SPϩ7 G(Ϫ10)L, respectively (Fig. 1). Processing of the mutant precursor peptide occurs at a slightly faster rate. In the quantification it is assumed that both peptides are labeled at the N-terminal lysines, which is obvious from the gels. In the case that the lysine at the ϩ6 position is also labeled, the actual amount of processing would be even higher. The processing efficiency of the precursor peptides is similar to that of the whole precursor prePhoE in this mixed micellar system (23) indicating that these precursor peptides are valid model compounds.
Characteristics of the Reconstitution Vesicles-Lep was reconstituted in vesicles composed of DOPE/DOPG (4:1 molar ratio) imitating the phospholipid composition of the E. coli inner membrane. The size of the reconstituted vesicles was compared with that of E. coli inner membrane vesicles by dynamic light scattering and determined to be very similar with average diameters of 175 and 170 nm, respectively. In the reconstitution vesicles, Lep molecules are expected to be oriented with their periplasmic domain both outside and inside the vesicles, as schematically illustrated in Fig. 2A. To determine this topology and to obtain vesicles with Lep molecules with their active site only on the lumenal site of the bilayer, the vesicles were treated with proteinase K and analyzed by Western blotting with an antibody against the P2 loop of Lep. Fig.  2B shows the Western blot of vesicles without protease treatment (lane 1) and after incubation with proteinase K (lane 2). Externally oriented Lep molecules are fully degraded by proteinase K, resulting in loss of immunoreactive material. Of the internally located Lep molecules, the P2 loop is protected from protease digestion. The P1 loop is exposed on the exterior of the vesicles and can be cleaved by proteinase K, resulting in a slight shift to a lower molecular weight (cleaved). Digestion of the P1 loop does not destroy the Lep activity, because only the second membrane-spanning domain and the P2 loop are critical for catalytic activity (35). A small amount of full-length Lep remains after protease treatment, as was also observed in inner membrane vesicles (23). Solubilization of the vesicles with Triton X-100 prior to addition of proteinase K, resulted in the digestion of all Lep molecules (lane 3). In intact vesicles some Lep molecules are protected (lane 2), demonstrating that proteinase K cannot reach the vesicle lumen. The topology of Lep molecules in the reconstitution vesicles was calculated from the intensity of the band corresponding to the partially degraded product compared with the amount of Lep present before protease treatment. For different batches of vesicles it was found that 8 -12% of the Lep molecules is oriented with its periplasmic domain in the inside of the vesicle, similar to what was found in an earlier study (36).
Next, the possibility to generate a ⌬ in these protease pretreated proteoliposomes was investigated. This is not trivial because it is known that reconstitution vesicles containing integral membrane proteins are often rather leaky. Fig. 3A shows that a positive inside ⌬ (resembling the situation in E. coli) could be generated in the Na ϩ -containing proteoliposomes diluted in K ϩ buffer by addition of valinomycin. This ⌬ was highest immediately after addition of valinomycin (30 mV) but was not stable in time. After about 10 min, only a small ⌬ remained, which could be dissipated with gramicidin. Fig. 3B shows that a negative inside ⌬ could be generated by addition of valinomycin to the reversed system. The negative inside ⌬ was more stable in time, although it is also decreased after 10 min. Again, the remaining ⌬ could be dissipated by gramicidin. In both cases, even without addition of valinomycin, a moderate spontaneous positive or negative inside potential was generated (not shown), most likely as the result of the differences in rate of spontaneous diffusion of K ϩ and Na ϩ . No ⌬ was observed in a system of Na ϩ inside and Na ϩ outside, or K ϩ inside and K ϩ outside after addition of valinomycin. The possibility of generating a ⌬ in both directions in reconstitution vesicles in which all Lep molecules have their active site inside the vesicles, results in a valid system in which the effect of ⌬ on the translocation of the cleavage region of a signal sequence can be studied.
A Positive Inside ⌬ Enhances Processing of Wild Type SPϩ7, but not of SPϩ7 G(Ϫ10)L-With the proteinase K pretreated proteoliposomes, the effect of ⌬ on translocation of the precursor peptides could be analyzed. In the absence of ⌬ (Na ϩ inside and Na ϩ outside, Fig. 4A, upper gel left) a small amount of SPϩ7 is processed. The Tricine gel shows that in time the band corresponding to the signal peptide appears. Apparently, the precursor peptide can spontaneously insert into the pure lipid membrane and the C terminus of at least part of the molecules can cross the bilayer to the lumen of the vesicle where Lep is active. Strikingly, when a ⌬ is applied in this system, directed positive inside (Na ϩ inside and K ϩ outside, upper gel right), processing of SPϩ7 is largely stimulated. Already after only 5 s of incubation a band of the signal peptide has appeared, whereas this band is barely visible in the situation without a ⌬ after short incubation times. In contrast, SPϩ7 G(Ϫ10)L is processed efficiently both in the absence and in the presence of a positive inside ⌬, which is evident from the lower gels in panel A. Fig. 4B quantifies the results and shows that the amount of processing of wild type SPϩ7 (solid line) in the presence of a positive inside ⌬ at any time point is significanly increased compared with the situation without a ⌬, but that the amount of processing of SPϩ7 G(Ϫ10)L is not enhanced in the presence of ⌬. The same results were found in several independent experiments with different batches of labeled peptides. When no valinomycin was added, stimulation of SPϩ7 processing was observed also in the Na ϩ inside and K ϩ outside situation compared with Na ϩ inside and Na ϩ outside due to the generation of a spontaneous potential, but it was a less pronounced effect. This demonstrates that valinomycin itself is not responsible for the observed effects. The results indicate that the signal sequence can insert into the membrane and translocate its C terminus across the bilayer without the presence of the complete translocase and that this process is largely accelerated and enhanced by ⌬, but the single substitution of the helix-breaking Gly Ϫ10 for a Leu residue relieves the ⌬ dependence.
A Reversed ⌬ Inhibits Processing of Wild Type SPϩ7-To further investigate whether ⌬ acts on the signal sequence, a reversed ⌬, negative inside, was applied to the proteoliposomes. In Fig. 5, the data are presented as the percentage of processing caused by the presence of a ⌬ relative to the situation without a ⌬. Incubation of wild type SPϩ7 with reconstitution vesicles in the presence of a reversed ⌬ (K ϩ inside and Na ϩ outside) led to an inhibition of processing compared with the control without ⌬ (K ϩ inside and K ϩ outside; Fig. 5,  line, closed symbols). This is an additional indication that ⌬ acts on the wild type signal sequence, but not on the analogue with a Leu residue replacing the Gly residue in the hydrophobic core.
Different Kinetics of Insertion of the Precursor Peptides Into a Lipid Monolayer-The differences in behavior of the two peptides in the translocation assay might result from a different mode of insertion into the lipid bilayer. It has been observed earlier that the wild type signal peptide of PhoE inserts into a lipid monolayer with two-phase kinetics. The second phase was characterized by insertion of additional peptide and a two-fold reduction of the area per molecule, suggesting a transition from a destabilized looped conformation to an extended helical conformation (32). This effect could be most clearly observed for a pure DOPG monolayer. We first investigated whether SPϩ7 exhibits a similar insertion behavior. Fig. 6 shows that SPϩ7 inserts rapidly into the monolayer, followed by a more ratelimiting further penetration, similar to the two-phase kinetics previously observed for the signal peptide. The second insertion step of SPϩ7 is slower than that previously observed for the signal peptide (compare Ref. 32), probably as a result of the C-terminal extension. In contrast to the wild type SPϩ7, the mutant precursor peptide (SPϩ7 G(Ϫ10)L) inserted rapidly into the DOPG monolayer and soon reached a stable level (Fig.  6). The molecular area of this peptide inserted in the monolayer was determined to be 138 Ϯ 21 Å 2 /molecule, which is similar to the molecular area of wild type signal peptide (32) and SPϩ7 (24) in the second phase of monolayer insertion. This value is consistent with the cross-section of an ␣-helix. These results suggest that the presence of a helix-breaking residue in the signal sequence is responsible for the two-phase insertion into the lipid monolayer. DISCUSSION Translocation of proteins across the inner membrane of E. coli depends on the pmf. Previous studies have shown that mutations in the signal sequence (8) and directly after the cleavage site (37,38) affect the pmf dependence of translocation, suggesting an involvement of the pmf early in translocation. However, whether the pmf directly acts on the signal sequence remained unclear, because effects via components of the secretion machinery could not be excluded. This study provides the first evidence that the electrical component of the pmf (⌬) can directly act on the signal sequence and stimulates the translocation of the cleavage region, in a way that is strikingly dependent on the presence of the helix-breaking residue in the hydrophobic region. The precursor peptide, analogous to the signal sequence of PhoE with the first seven amino acids of the mature part, was found to be processed in the lumen of the Lep-containing reconstitution vesicles, demonstrating the translocation of the cleavage region across the lipid bilayer. This was largely stimulated by the presence of a positive inside ⌬ (corresponding to the situation in E. coli). A reversed ⌬ inhibited the translocation. These results cannot be ascribed to an effect of ⌬ on Lep, because it was shown that the replacement of the helix disrupting Gly at position Ϫ10 by a Leu abolished the effect of ⌬ on translocation. Upon translocation, the positively charged N terminus stays most likely anchored to the side of the membrane were it was added, as a result of electrostatic interactions with the negatively charged lipids (13) and because it is energetically highly unfavorable to move it across the lipid bilayer, especially against the direction of the positive inside ⌬ (39).
Many studies have shown a stimulating effect of the membrane potential on translocation of parts of a protein. In E. coli, M13 procoat protein needs the pmf for translocation of the periplasmic region (40,41) and translocation of N-terminal domains requires the pmf when the hydrophobicity of the first transmembrane domain is low (42). In mitochondria, ⌬ is required for the translocation of mitochondrial presequences (43). Also in other biological processes, translocation of parts of proteins occurs under influence of a membrane potential, such as in the gating of pore-forming colicins and in the working mechanism of some toxins (44 -46). In all situations the effect of the membrane potential could be explained by electrophoretic forces on charged residues in the translocated domains. We propose a similar mechanism to explain our data: ⌬ can act on the C-terminal part of the precursor peptide, which contains an excess of negative charges. In agreement with such a mechanism an acidic residue is often found at position ϩ2 of the mature sequence in bacterial precursor proteins (47). An additional possibility could be that ⌬ acts on the helix dipole moment, which is negative on the C-terminal part of the ␣-helix of the signal sequence.
Before presenting a molecular model for the observed effects, it is important to recall what is known about the conformational behavior of the PhoE signal sequence and its relation to the pmf dependence of translocation. Translocation of prePhoE is highly dependent on the pmf. Strikingly, the pmf dependence is relieved by a single mutation in the signal sequence replacing Gly Ϫ10 for different helix stabilizing residues (8). This strongly suggests that the pmf dependence in translocation of precursor proteins is related to the conformational behavior of the signal sequence. Our study provides two important results in favor of such explanation. First, introduction of a helixpromoting Leu in the signal sequence abolished the effect of ⌬ on translocation across the lipid bilayer of the C terminus of the precursor peptide, which is in accordance with the observed pmf independence of translocation of the corresponding pre- PhoE mutant across the inner membrane (8). Second, the monolayer results demonstrate that the mode of insertion into a hydrophobic environment is largely affected by the Gly to Leu substitution. Wild type SPϩ7 displayed a two-phase kinetics of monolayer insertion, as previously observed for the wild type signal peptide. This has been explained as the initial insertion in a looped conformation, which changes to an extended ␣-helix (32). However, SPϩ7 G(Ϫ10)L is inserted into the monolayer in one step. The lack of the second phase suggests that it penetrates into the lipids directly in a stable ␣-helical conformation. In a previous study, two-dimensional NMR analysis of the wild type PhoE signal sequence in SDS micelles showed a clear destabilization of the helix around Gly Ϫ10 (22). CD measurements of both precursor peptides in the same system (not shown), performed as described in (22), with a 1:100 peptide: detergent molar ratio, showed a 10% higher ␣-helical content for SPϩ7 G(Ϫ10)L than for wild type SPϩ7 (67 and 57%, respectively) as determined according to the method of Greenfield and Fasman (48). This supports our proposal that SPϩ7 G(Ϫ10)L most likely contains a more stable, extended ␣-helix in a membrane environment. A speculative model to explain these observations is presented in Fig. 7. It proposes that the wild type signal sequence initially adopts a destabilized helix that can partially insert into the membrane, which converts into a transmembrane helical conformation, which process is highly facilitated by the transmembrane potential. When a helix-promoting residue is present at the position of Gly Ϫ10 , the extended transmembrane orientation is achieved rapidly, even without a transmembrane potential.
Studies in the analogous eukaryotic secretion system have provided strong evidence that the signal sequence early in translocation is functional in a transmembrane orientation (49). The model poses an intriguing question; why did nature bother to use a helix-breaking residue in the signal sequence if only a transmembrane state is essential? Because the signal sequence interacts with many proteins of the translocation machinery, including SecA, the SecYEG complex, the lipids, and Lep, it might be essential that the signal sequence needs a certain conformational flexibility to be able to undergo all these distinct interactions. The presence of the Gly residue might be a good compromise for efficient functioning in all stages of the process. It should be realized that the pmf also acts on the translocase and possibly sets it in a more relaxed conformation (50). It is therefore an attractive possibility that the pmf acts in concert on the signal sequence and the translocase.
In conclusion, we propose that ⌬ in the prokaryotic secretion pathway acts on the signal sequence by stretching it into a transmembrane conformation and thereby initiating the translocation of the mature part of a precursor protein across the membrane.