The proton motive force, acting on acidic residues, promotes translocation of amino-terminal domains of membrane proteins when the hydrophobicity of the translocation signal is low.

We have shown previously that the first transmembrane segment of leader peptidase can function to translocate the polar amino-terminal Pf3 domain across the membrane into the periplasm independently of the proton motive force (pmf) (Lee, J. I., Kuhn, A., and Dalbey, R. E. (1992) J. Biol. Chem. 267, 938-943). We now show that when the first transmembrane segment lacks a strong hydrophobic character, the pmf is required for translocation. In addition, we find that the amino-terminal acidic residue proximal to the transmembrane domain plays a critical role in pmf-dependent amino-terminal translocation. Moreover, the pmf is required to hold the amino-terminal domain in the periplasm to prevent it from slipping such that the amino terminus is no longer exposed to the periplasm. In all cases, translocation occurs under conditions in which the function of the Sec machinery is impaired. These studies show that the low hydrophobicity of the first apolar domain (the translocation signal) can be compensated for by a negative charge in the amino-terminal region, upon which the pmf acts.

The mechanisms by which a protein integrates into the membrane and assumes its correct topology has been studied with great interest (1,2). It has been shown that membrane proteins in bacteria can utilize the Sec machinery to assemble across the plasma membrane of Escherichia coli. However, there are a growing number of membrane proteins that apparently do not use the Sec machinery (3). This sec-independent class includes membrane proteins that are made without a signal sequence and that are oriented with the amino terminus on the periplasmic side of the plasma membrane (N out C in orientation).
Although the exact mechanism by which these proteins assemble across the membrane is not known, the factors that influence the membrane topology have been elucidated. For example, it has been previously found that the asymmetric distribution of charged residues, particularly basic residues flanking a membrane-spanning domain on the cytoplasmic side, acts as a topogenic determinant for translocation (4 -7). Positive charges are retained in the cytoplasm and inhibit translocation of polar regions (8 -11). Negatively charged amino acids, in low abundance, do not inhibit translocation (11,12). Several groups have postulated that the membrane topol-ogy may be determined by alignment of dipoles and charged groups of the protein within the electrical field of the proton motive force or pmf 1 (10 -14). This electrophoretic force is a result of the transmembrane electrical gradient, ⌬⌿ (positive outside, negative inside), which may actively promote translocation of acidic residues (11,12,15) and impede translocation of basic residues (14). In support of this hypothesis, Kiefer and colleagues (16) found that acidic residues in the amino-terminal region of the Pf3 coat protein (N out C in ) are translocated across the membrane by the pmf and can act as topogenic determinants.
Another factor that affects membrane insertion of a protein is the overall hydrophobicity (or length) of a transmembrane segment (17). It is thought that apolar residues within hydrophobic domains allow their partitioning into the membrane in an energetically favorable manner, driving insertion (18 -20). Although there have been several studies conducted on the role of hydrophobicity in cleavable signal peptides (21)(22)(23)(24)(25)(26) and uncleaved signals (17,27) which undergo sec-dependent translocation in bacteria, the role of the hydrophobicity of transmembrane domains which support amino-terminal translocation has not been investigated.
What are the structural requirements for amino-terminal translocation for proteins that lack signal (or leader) peptides? In bacteria, translocation of amino-terminal sequences requires a downstream hydrophobic segment (28,29) and is most efficient when the amino-terminal region is short (14) and contains few positively charged residues (14,29). Typically, the energy source driving translocation is the pmf which is most likely correlated with the acidic charge content of the periplasmic amino-terminal domains. Translocation of amino-terminal segments is believed to be Sec-independent because translocation is unaffected by treatments that block the normal function of the Sec machinery (3). In eukaryotes, amino-terminal translocation is dependent not only on the charge difference (30) between the transmembrane flanking segments with the more positive side typically retained in the cytoplasm but also on the folding state of the aminoterminal domain and the hydrophobicity of the translocation signal (31). Long hydrophobic sequences favor amino-terminal translocation, whereas short hydrophobic sequences favor carboxyl-terminal translocation (32).
We have used leader peptidase (lep) of Escherichia coli as a model system for studying amino-terminal translocation (14,28). Lep spans the membrane twice with its amino terminus on the periplasmic surface of the membrane and its large carboxylterminal domain protruding into the periplasm (28,33,34). We have shown previously that the first hydrophobic domain (H1) of lep can function to translocate a short, polar amino-terminal 18 amino acid antigenic peptide from the phage Pf3 coat protein across the plasma membrane of Escherichia coli (28). We have now examined the energetic requirements necessary for the insertion of amino-terminal periplasmic domains and have determined that the hydrophobicity of the transmembrane domain and the pmf are both required to varying degrees.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-E. coli strain MC1061 (⌬lacX74, araD139, ⌬(ara, leu)7697, galU, galk, hsr, hsm, strA) was from our laboratory. XL1-blue (supE44, hsdR17, recA1, endA1, gyrA46, thi relA1, lac Ϫ ) was acquired from Stratagene. Pf3-lep and its derivatives were expressed using the pING plasmids (35), which contains the arabinose promoter and the arabinose regulatory elements (Ingene, Inc. DNA Manipulations-Site-directed mutagenesis was accomplished using the Stratagene QuikChange procedure with a few alterations. The mutagenesis reaction was physically separated into two portions with Ampliwax beads. The lower portion contained 2.5 l of the 10-fold reaction buffer, the mutagenic oligonucleotides (125 ng each), deoxynucleotide triphosphates (50 M each, final concentration), and double distilled water to a final volume of 25 l. An Ampliwax bead was placed on top, and a wax layer was created by heating and chilling the tube. The upper reagent, administered above the wax layer, contained 2.5 l of the 10-fold reaction buffer, 50 ng of double-stranded DNA, 1 l of Pfu DNA polymerase and double distilled water to a final volume of 25 l. Mutagenesis was then carried out in a Progene thermocycler following the QuikChange protocol. Following DpnI treatment and transformation, mutations were screened by sequencing doublestranded DNA (36) using U. S. Biochemical Corp. Sequenase version 2.0. Mutant constructs identified by sequencing were then transformed into MC1061 using the calcium chloride method (37).
Protease Mapping Studies-MC1061 cells (1 ml) bearing the pING plasmid encoding Pf3-lep proteins were grown to the midlog phase in M9 minimal media (38) containing ampicillin (100 g/ml) at pH 7.0 with 0.5% fructose and 50 g/ml of each amino acid except methionine. The cells were induced with arabinose (0.2%, final concentration) for 1 h to express the Pf3-lep mutant proteins. Cells were labeled with 100 Ci of trans-[ 35 S]methionine for 1 min, chilled on ice, then collected by centrifugation (16,000 ϫ g, 4°C, 40 s). After resuspending in 0.25 ml of Tris acetate, pH 8.2, 0.5 M sucrose, and 5 mM EDTA, the cells were treated with lysozyme (80 g/ml, final concentration) and 0.25 ml of ice-cold water. After incubating for 5 min, 30 l of 1 M MgSO 4 were added to stabilize the spheroplasts, and the cells were collected by centrifugation (16,000 ϫ g, 4°C, 40 s). The spheroplasts were gently resuspended in 50 mM Tris acetate, 0.25 M sucrose, 10 mM MgSO 4 and incubated with or without proteinase K (1.5 mg/ml) for 1 h on ice. Another aliquot of spheroplasts was treated with 2% Triton X-100 and incubated with proteinase K (1.5 mg/ml). After quenching the protease with PMSF (5 mM, final concentration) for 5 min, the samples were acid-precipitated and immunoprecipitated with antibody against lep, ribulokinase, a cytoplasmic marker, and outer membrane protein A (OmpA), as described (11). Samples were then analyzed by SDS-PAGE with a 15% polyacrylamide gel and subjected to fluorography (39).
Quantitation of the Translocation Data-Fluorographs were scanned using an AppleOne Scanner. The bands were then quantitated by using the public domain program NIH Image, developed at the National Institutes of Health. 2 The percent of amino-terminal translocation was determined by taking into account the number of methionines lost during proteolysis by proteinase K. For most of the constructs there was one methionine lost out of eight (see Equations 1, 3, and 4), whereas for ⌬4 -9, ⌬4 -22, and the constructs that contain both the amino-terminal domain and the membrane-spanning domain of Pf3 coat there was one methionine lost out of seven (see . Processing (%) ϭ (8/7 processed Pf3-lep ϫ (Eq. 1) 100)/͑8/7 processed Pf3-lep ϩ unprocessed Pf3-lep).
Translocated (%) ϭ (processing %/spheroplast %) ϫ 100. (Eq. 4) Determination of Summed Hydrophobicity-The summed hydrophobicity (H) of the transmembrane region flanking the amino-terminal domain was determined as described (17) using the GES scale (40). We calculated H for the native and mutant Pf3-lep proteins by adding the hydrophobicity of the uninterrupted stretch of uncharged amino acids within the transmembrane region.

RESULTS
The Role of Amino-terminal Aspartic Acids in Amino-terminal Translocation-The amino terminus of Pf3-lep has previously been shown to translocate in a pmf-and sec-independent manner (28). This is in contrast to amino-terminal translocation of the sec-independent Pf3 coat protein (see Fig. 1A for topology), which requires the pmf (41) and the aspartyl residue at position 18 of the amino-terminal domain (16) also present in Pf3-lep. Therefore, we have investigated the role of acidic residues in amino-terminal translocation using Pf3-leader peptidase (Pf3-lep) with an F79R mutation as our model protein (Fig. 1A). This construct contains the 18 amino acid Pf3 region, from the amino terminus of Pf3 coat, fused to the fourth amino acid of leader peptidase (lep) with a threonine linking the two domains (Fig. 1B). Pf3-lep is oriented in the plasma membrane with the amino-terminal Pf3 domain in the periplasm, a single hydrophobic membrane-spanning domain (H1), and the carboxyl-terminal lep domain in the cytoplasm (Fig. 1A). The introduction of an arginine at position 79 of lep inhibits insertion of the second transmembrane domain and translocation of the carboxyl terminus (14,27). The resulting protein contains only one translocated region which allows us to monitor translocation of the amino terminus of Pf3-lep across the plasma membrane via protease mapping techniques.
We conducted site-directed mutagenesis on the Pf3 domain of Pf3-lep to determine whether a similar requirement for the aspartyl residues existed for efficient translocation of Pf3-lep. We replaced the aspartyl residues at positions 7 and 18 of the Pf3 domain with alanines, either as a single mutation or as a double mutation (Fig. 1B). Amino-terminal translocation was monitored by protease mapping in the absence or presence of the protonophore CCCP, which destroys the pmf.
Expression of Pf3-lep in exponentially growing cells was induced with 0.2% arabinose for 1 h at 37°C. Cells producing Pf3-lep or aspartyl residue mutants were then pulse-labeled with 100 Ci of 35 S-trans-labeled methionine for 1 min and then converted to spheroplasts (42). To examine translocation in the absence of a pmf, CCCP was added for 45 s prior to labeling. Protease mapping was accomplished by dividing spheroplasts into three equal fractions. The first fraction was incubated on ice for 1 h and then trichloroacetic acid-precipitated. The second fraction was incubated on ice with 1.5 mg/ml proteinase K for 1 h, quenched with PMSF (final concentration, 5 mM), followed by trichloroacetic acid precipitation. The third fraction was treated for 1 h with both 1.5 mg/ml proteinase K and 2% Triton X-100 followed by treatment with PMSF and trichloroacetic acid precipitation. After each sample was divided into two, one-half was immunoprecipitated with antileader peptidase antibody and the other half was immunopre-cipitated with anti-OmpA and anti-ribulokinase antibodies. The samples were then applied to a 15% SDS-PAGE and the results analyzed by fluorography.
As shown in Fig. 2A, amino-terminal translocation was observed in Pf3-lep (lanes 1-3), 7A Pf3-lep (lanes 4 -6), 18A Pf3lep (lanes 7-9), and 7,18A Pf3-lep (lanes 10 -12), in the presence (ϪCCCP) of a pmf. This is detected by the appearance of a smaller protease-resistant fragment seen in lanes treated with proteinase K (see arrow). Since the proteinase-treated fragments differ by only 18 amino acids, there is a slight shift in molecular weight. These results show that the acidic residues are not required for translocation of the amino terminus of Pf3-lep, in contrast to Pf3-coat protein (16). When the translocation study is carried out in the absence (ϩCCCP) of a pmf, amino-terminal translocation is still very efficient as seen in Fig. 2B (lanes 1-3, 4 -6, 7-9, and 10 -12). The total amount of translocated domain in the presence or absence of a pmf was quantitated by scanning the fluorograms and using the public domain program NIH Image to determine the percent translocated. As shown in Fig. 1B, 98 -99% of the proteins translocated their amino-terminal domains efficiently both in the presence and in the absence of a pmf. Successful conversion to spheroplasts was monitored in all experiments by digestion of OmpA, which is exported across the inner membrane and is completely accessible to proteinase K. The integrity of the spheroplasts was determined by monitoring the stability of cytoplasmic ribulokinase in the presence of proteinase K (Fig.  2). Furthermore, the depletion of the proton gradient was demonstrated by the inhibition of pro-OmpA translocation (ϩCCCP), as the accumulated pro-OmpA is protease-resistant in these studies (Fig. 2B).
Although both Pf3-lep and the Pf3 coat protein contain the same amino-terminal 18 amino acid residues, they do not have the same requirement for aspartic acid residues for efficient amino-terminal translocation. This indicates that there is some other important factor in the lep domain of Pf3-lep that promotes translocation.
The PMF Can Promote Translocation When Apolar Domain 1 Has Low Hydrophobicity-To determine the role of the hydrophobicity of an N out C in transmembrane segment in aminoterminal translocation, we constructed several deletion mutations within the transmembrane domain of Pf3-lep. The constructs are shown in Fig. 3 in the order of most hydrophobic to least hydrophobic transmembrane domains. The total hydrophobicity was determined for the uninterrupted stretches of hydrophobic residues (40). The hydrophobicity scale used in these studies is the transfer free energy (kcal/mol) of the amino acid side chains from water to a non-aqueous environment. Since the transmembrane domain is more hydrophobic at the amino end than the carboxyl end, deletions at this location are more detrimental to amino-terminal translocation. Protease mapping was then performed as before, in the presence of the pmf (Fig. 4A). Amino-terminal translocation was quantitated and OmpA and ribulokinase controls were also carried out as before (data not shown). Although Pf3-lep with an intact transmembrane domain translocates efficiently (99%), an effect can already be seen on the translocation of the smallest deletions, ⌬17-22 (lanes 1-3) and ⌬4-9 (lanes 4 -6), with 93 and 91% translocating, respectively. For ⌬13-22 (lanes 7-9) and ⌬4-12 (lanes 10 -12), translocation is roughly half as efficient as it is for the full-length protein (66 and 54%, respectively). Finally, translocation is completely abolished for the larger deletions ⌬9-22 (lanes 13-15), ⌬4-17 (lanes 16 -18), and ⌬4-22 (lanes 19 -21), showing that the hydrophobicity of a downstream transmembrane domain is essential for amino-terminal translocation.
In contrast, when translocation was carried out in the absence of a pmf (CCCP treated cells), only the most hydrophobic  lanes 1-3), was partially translocated (54%, Fig. 4B), whereas the other deletion mutants were completely blocked. Therefore, amino-terminal translocation only occurs with the full-length Pf3-lep and ⌬17-22 in the absence of a pmf. This indicates that the pmf can promote membrane insertion of proteins with apolar domains with low hydrophobicity.
Acidic Residues Are Required for Amino-terminal Translocation When the Hydrophobicity of the Transmembrane Domain Is Decreased-Although we have shown that negatively charged amino acids are not required for translocation of the amino terminus of Pf3-lep with a full-length transmembrane domain (Fig. 2), we wanted to test whether translocation of the amino terminus of Pf3-lep with a truncated H1 requires the amino-terminal aspartyl residues. This is of particular interest since ⌬13-22 requires the presence of a pmf for translocation, as shown in Fig. 4. We used site-directed mutagenesis to create mutations within the Pf3 domain of ⌬13-22, by substituting the aspartic acid residues at positions 7 and 18 of the aminoterminal domain with alanines as before (Fig. 5A). We then conducted protease mapping experiments using cells that were pulse-labeled with trans-[ 35 S]methionine in the presence and in the absence of the pmf to determine the competency for amino-terminal translocation. Whereas the amino-terminal domains of ⌬13-22 and 7A ⌬13-22 translocate with equal efficiency in the presence of a pmf, at 66 and 58%, respectively, the amino-terminal domain does not translocate in the 18A ⌬13-22 mutant and the double mutant 7,18A ⌬13-22 (Fig. 5B). We should note that the low molecular weight bands observed in lanes 5, 8, and 11 are uncharacterized proteolytic fragments. In the absence of the pmf, translocation of all four of these proteins is abolished (data not shown). Therefore, we found that the acidic residue proximal to the transmembrane domain is absolutely required for pmf-dependent translocation of the amino-terminal domain of ⌬13-22 (Pf3-lep with a truncated transmembrane domain), in agreement with the results found for Pf3 coat (16). This suggests that the pmf is acting on the aspartyl residue at position 18 of the amino-terminal Pf3 domain of Pf3-lep to drive translocation when the hydrophobicity of the membrane-spanning domain is low.
Role of PMF in Sustaining Amino-terminal Translocation-We wanted to determine if the pmf was also required to sustain amino-terminal translocation of Pf3-lep in which the hydrophobicity of apolar domain 1 had been decreased. In this study, we characterized Pf3-lep with a full-length transmembrane segment, ⌬4-9, which is dependent on the pmf for translocation, and ⌬13-22, which had an even stronger defect in translocation as shown in Fig. 4. We pulse labeled a 2-ml culture for 1 min using 200 Ci of trans-[ 35 S]methionine followed by a 5-min chase of cold methionine (500 g/ml). Onethird of the sample was then chilled, and the remaining cells were treated with 5 l of 10 mM CCCP for 1 min. At this time, one-half of the remaining sample was then removed and chilled as above, whereas the remaining cell culture was treated with 2-mercaptoethanol to inactivate the CCCP and allow the pmf to regenerate (43). These samples were then washed to remove the 2-mercaptoethanol and incubated at 37°C for 10 min to re-establish the pmf (see figure legend). All the samples were then converted into spheroplasts and protease-mapped as described previously. We found that Pf3-lep with a full-length apolar domain 1 was unaffected by this treatment (Fig. 6A,  lanes 1-9). To our surprise, we found that both ⌬4-9 and ⌬13-22, which had undergone amino-terminal translocation (91 and 66%, respectively) after a 5-min chase (Fig. 6, B and C, lanes  1-3; denoted as Pulse ϪCCCP), were not as accessible to protease when the pmf was abolished following pulse labeling (Fig.  6, B and C, lanes 4 -6; denoted as Pulse ϩCCCP). We found that the extent to which the amino terminus had "slipped" out of the periplasmic space was correlated with the total amount of hydrophobicity in the apolar domain 1. Although full-length Pf3-lep does not "slip" out from the periplasmic space (hydrophobicity of Ϫ46 kcal/mol) when the pmf was abolished, only 62% of ⌬4-9 with a total hydrophobicity of Ϫ28 kcal/mol (Fig. 3) and none of ⌬13-22 with a total hydrophobicity of Ϫ25 kcal/mol remain translocated. We are convinced that this effect was due to the abolishment of the pmf because when we allow the pmf to regenerate in the presence of 2-mercaptoethanol for 10 min, translocation of the amino-terminal domains is recovered to the same extent as before and are digested by protease (81 and 50%, respectively) Fig. 6, B and C, lanes 7-9 (denoted as P/ϩCCCP 10Ј Recover). The results indicate that when the transmembrane domain is not adequately hydrophobic, as in the cases of ⌬4-9 and ⌬13-22, the pmf is required not only for initiation of amino-terminal translocation but also to sustain translocation of the Pf3 region. We hypothesize that the pmf holds the translocated amino-terminal domain with the proximal acidic residue in the periplasm and therefore prevents this domain from "slipping" out from the periplasmic space.
PMF-dependent Translocation of the Amino Terminus of Pf3lep When H1 of Lep Is Replaced with the Transmembrane Segment of the Pf3 Coat Protein-Although the Pf3 coat protein requires a pmf, Pf3-lep does not. One possible reason for this difference is that the transmembrane segment of Pf3 coat is less hydrophobic than that of lep. In view of our results with respect to hydrophobicity, we wanted to determine the outcome of replacing the H1 of lep, with a summed hydrophobicity of Ϫ46 kcal/mol, with that of the Pf3 coat protein, which has a hydrophobicity of Ϫ39 kcal/mol. Oligonucleotide-directed mu- tagenesis was used to create this mutant. The ⌬4-22 Pf3-lep mutant, in which the first transmembrane domain of lep was deleted, was used as a template for mutagenesis. The insertion of the Pf3 coat transmembrane domain was completed in two rounds of mutagenesis in which nine amino acids were first inserted, followed by the insertion of the remaining nine amino acids. The completed mutant was sequenced and the transmembrane segment was found to be genetically identical to that of the Pf3 coat protein. This mutant protein, Pf3H1-lep, contains the amino-terminal domain and the transmembrane domain of the Pf3 coat protein fused to the 23rd amino acid of lep. Pf3H1-lep was protease-mapped in cells that were analyzed for translocation in the presence and absence of a pmf, as described previously (Fig. 7). Pf3H1-lep translocates efficiently in the presence of a pmf (lanes 1-3) (82%) and is impeded without a pmf (lanes 4 -6) (61%). However, it has been shown that translocation of Pf3 coat is abolished in the absence of a pmf (16). Since lep also has two additional hydrophobic regions (H2 and H3) unlike Pf3 coat, we deleted them to determine if amino-terminal translocation occurs in a manner more similar to that of Pf3 coat. In the presence of a pmf, Pf3H1-lep ⌬62-98 translocated with a 99% efficiency (lanes 7-9), and 33% was translocated in the absence of a pmf (lanes 10 -12). These results show that pmf-dependent translocation of the amino terminus of Pf3-lep with the transmembrane segment of the Pf3 coat protein is enhanced when H2 and H3 are omitted.
The Amino-terminal Domain of Pf3-lep Translocates under Conditions Where the Sec Machinery Is Impaired-We tested whether Pf3-lep becomes dependent on the Sec machinery when the hydrophobicity of the first hydrophobic domain is decreased. To test this, we conducted protease mapping experiments on cells expressing ⌬13-22 in which the function of the Sec machinery was compromised. As a control, we confirmed that Pf3-lep with a full-length apolar domain 1 inserts independently of a functioning Sec machinery (14). Cells were grown to mid-log phase and induced with arabinose to express the plasmid-encoded proteins for 1 h. The cells were treated with azide for 5 min, followed by pulse labeling for 1 min, and then analyzed by protease mapping; azide has been shown to inhibit the SecA ATPase activity (44). We show that azide treatment does not affect amino-terminal translocation of either of these proteins (Fig. 8, A and B, lanes 1-3). Furthermore, translocation is not affected in SecA ts or SecY ts cells when grown at the non-permissive temperature, 42°C. As shown in Fig. 8A, Pf3-lep translocates its amino terminus to the same extent in cells with impaired SecA (lanes 4 -6) or SecY (lanes 7-9) as in cells with functional SecA and SecY ( Fig. 2A, lanes  1-3). Similar results are found with ⌬13-22 (Fig. 8B, lanes 4 -9). Under these experimental conditions where the function of SecA or SecY is impaired, the translocation of OmpA is completely blocked as indicated by the fact that pro-OmpA is protease-resistant. These results suggest that amino-terminal translocation occurs in a Sec-independent manner when the transmembrane segment has either high or low hydrophobicity.

DISCUSSION
In this report we have examined why amino-terminal translocation of some proteins, such as Pf3 coat (41) and ProW (45), requires the pmf and leader peptidase (28) does not. Our results show that when the hydrophobic character of a transmembrane segment is high, amino-terminal translocation occurs independently of the pmf. However, translocation is dependent on the pmf which acts on a negative charge when the hydrophobic character of the transmembrane segment is low.
As a general rule, we found that amino-terminal transloca-FIG. 6. The PMF is required to sustain the amino-terminal region in the periplasmic space when the first apolar domain lacks strong hydrophobic character. MC1061 cells harboring plasmids that encode Pf3-lep (A), ⌬4-9 (B), or ⌬13-22 (C) were grown, induced, and labeled as described previously in Fig. 2. The procedures are outlined under "Results" and the methods for the pmf recovery followed Wolfe and Wickner (43) with the following alterations. Cells pulse-labeled, chased, and CCCP-treated were washed in M9 media twice and incubated in M9 media containing 10 mM 2-mercaptoethanol, 500 g/ml cold methionine, and 1 mg/ml chloramphenicol for 2 min on ice. The cells were then washed to remove the 2-mercaptoethanol and incubated in M9 media for 10 min at 37°C with shaking. The cells were then harvested and converted to spheroplasts and protease-mapped as before. tion was decreased as the hydrophobicity of the transmembrane domain 1 was decreased. Even the deletion of four amino acids had a noticeable affect on translocation, where translocation decreased from 99 to 91%. Almost equally surprising was the finding that amino-terminal translocation could be supported, although less efficiently, by a mutant that contained a stretch of only nine uncharged amino acids, ⌬13-22. However, translocation was blocked by decreasing the hydrophobic character further.
Strikingly, we observed an increased pmf requirement for translocation when the hydrophobicity of apolar domain 1 was decreased (Figs. 3 and 4). The only mutant that was able to translocate in a pmf-independent manner, ⌬17-22, had a very small deletion in the first transmembrane domain. This suggests that it is the overall high hydrophobic content of transmembrane segment 1 of the native leader peptidase that enables translocation of its amino terminus across the membrane independently of the pmf. This may explain the pmf requirement of Pf3 coat (16) since the Pf3 coat has a transmembrane segment that is less hydrophobic than the first transmembrane segment of lep. Indeed, we observed pmf-dependent translocation of the amino terminus of Pf3-lep when H1 of lep was replaced with the transmembrane segment of the Pf3 coat protein. The pmf dependence was even more dramatic when the H2 and H3 domains of lep were deleted preventing them from aiding translocation of the amino-terminal domain (Fig. 7). Fig. 9 illustrates the results in which the translocation data of the Pf3-lep mutants are plotted against the hydrophobicity of the respective H1 regions. The hydrophobicity scale of Engelman et al. (40) was used to determine the total hydrophobicity. Translocation of the amino-terminal region can occur efficiently in the presence of a pmf (ϪCCCP) when the summed hydrophobicity of the transmembrane segment exceeds a threshold of roughly Ϫ21 kcal/mol, whereas translocation is inefficient even in the presence of a pmf below a threshold of roughly Ϫ18 kcal/mol. These results indicating that hydrophobicity of a signal domain is required for translocation of aminoterminal domains are in strong agreement with Lee and Manoil (17), where they studied an N in C out uncleaved signal of the E. coli serine chemoreceptor (Tsr). In their study, they used a hybrid protein where alkaline phosphatase (phoA) was fused immediately after the first transmembrane segment of Tsr. They found that arginine mutants showing high export of phoA had summed hydrophobicities greater than Ϫ23 kcal/mol (for the largest, uninterrupted stretch of hydrophobic amino acids), whereas low export (Ͻ15%) of phoA mutants had summed hydrophobicities of less than Ϫ23 kcal/mol. We obtained similar results with our single and double arginine mutants (data not shown).
The other intriguing finding (see Fig. 9) is that to achieve approximately 50% amino-terminal translocation, a summed H1 hydrophobicity of Ϫ21 kcal/mol is required in the presence of the pmf, compared with a summed hydrophobicity of Ϫ32 kcal/mol in the absence of the pmf. This suggests that the pmf contributes energy that is equivalent to Ϫ11 kcal/mol of hydrophobicity, i.e. four leucine residues.
Our results also demonstrate that the amino-terminal aspartyl residue flanking the transmembrane segment is essential for pmf-dependent translocation of ⌬13-22. This confirms that aspartic residues can play an active role in amino-terminal were analyzed for protease mapping as described in Fig. 2. For SecA ts studies, E. coli CJ105 cells bearing the plasmid encoding Pf3-lep and ⌬13-22 were grown to the mid-log phase at 30°C, shifted to 42°C, and induced with 0.2% arabinose for 1 h and then labeled with 100 Ci of trans-[ 35 S]methionine for 1 min. For SecY ts studies, E. coli CJ107 cells bearing the plasmid encoding Pf3 lep or ⌬13-22 were grown and labeled as described for the SecA studies. Following pulse labeling, the cells were converted to spheroplasts and analyzed as described in Fig. 2. CJ105 and CJ107 are derivatives of HJM114 (FЈ lac, pro[⌬lac pro]) and have been described (46) .   FIG. 9. Plot of amino-terminal translocation versus transmembrane hydrophobicity. These results are derived from the experiments in Fig. 4. The quantitation of translocation and the summed hydrophobicity of the first transmembrane region for the various mutants is shown in Fig. 3. The pmf was dissipated by the addition of CCCP.
translocation as previously suggested (16). It is striking that the pmf is not only required for initiation of translocation but also to stabilize the amino-terminal domain of ⌬4-9 and ⌬13-22 in the periplasm to prevent it from slipping away from the periplasmic space so that it is not accessible to protease (Fig. 6, B and C). In contrast, slipping is not observed for Pf3-lep which has a transmembrane segment with high hydrophobicity (Fig.  6A). How the pmf holds the amino-terminal domain on the periplasmic side of the inner membrane is not known at this time. Although it is possible that the amino-terminal polar domain is merely slipping into the membrane bilayer, we think that this would be unlikely because it would be energetically unfavorable for the polar domain to reside within the apolar membrane bilayer.
One possible role of a pmf is that it is affects translocation, via direct interaction of the negatively charged residues of the inserting membrane protein, with its electrical potential component (positive outside, negative inside). Our data do not directly address this possibility. However, if such a simple electrophoretic mechanism was operating in E. coli it does not explain why the acidic residue at position 18 would be required and not the one at position 7. A second possibility is that the pmf acts via a transmembrane pH gradient. If this were the case, a protein machinery, which protonates and deprotonates amino acid residues, would most likely be involved.
In conclusion, the present data provide strong evidence that the pmf, acting on an acidic residue, can compensate for the low hydrophobicity of the translocation signal. This indicates that hydrophobic forces and proton motive forces work together in the translocation process.