Mechanism of Signal Peptide Cleavage in the Biosynthesis of the Major Lipoprotein of the Escherichia coli Outer Membrane*

On treatment of Escherichia coli cells with globomy- cin, a glyceride-containing precursor of the major outer membrane lipoprotein accumulates in the cytoplasmic membrane (Hussain, M., Ichihara, S., and Mizushima, S. (1980) J. Biol. Chem. 255, 3707-3712). When the en- velope fraction from such cells was incubated in a suitable buffer, this precursor could be processed to the mature lipoprotein. The processing involved removal of the signal peptide and subsequent acylation of the NH2 terminus thus bared. Two types of peptidase and an acylation enzyme(s) were found to be involved in these processes. The enzyme that cleaves the signal peptide, called signal peptidase in this paper, had many unique properties: being highly resistant to high tem- perature, having a wide optimum pH range, and being highly sensitive to detergents. The other peptidase(s), called signal peptide peptidase in this paper, was assumed to be responsible for the digestion of the signal peptide that had been cleaved from the precursor li- poprotein. This enzyme was rather heat-sensitive. Thus the processing from the precursor to the mature lipo- protein at a high temperature resulted in accumulation of a peptide that was most probably the intact signal peptide. The third enzyme(s) involved in the processing was the one that is responsible for acylation of the newly bared NH2 terminus of the lipoprotein. The en- zyme activity was also lost at 80 “C. In the light of

to diglyceride through the thiol group and to fatty acid through the terminal amino-group (3) . We have reported that on treatment of E. coli cells with a cyclic peptide antibiotic called globomycin (7, €9, a precursor form of BLP, termed pro-BLP in this paper, accumulated in the cytoplasmic membrane (9). The cysteine residue of this pro-BLP had already been modified by a glyceride similar to as in the case of BLP. The pro-BLP thus accumulated could be chased in vitro to BLP. This in vitro processing would involve 3 changes, namely, (i) cleavage of the signal peptide from the pro-BLP, (ii) digestion of the cleaved signal peptide, and (iii) acylation of the amino-group of the newly exposed NH2-terminal cysteine residue. Using this in vitro processing system we tried to characterize the enzymes involved. We have found that at least two types of peptidase are involved in the processing of pro-BLP: an enzyme which carries out cleavage of the signal peptide (signal peptidase) and an enzyme which carries out digestion of the cleaved signal peptide (signal peptide peptidase). We have also been able to show the presence of a third enzyme which is responsible for acylation of the newly exposed NH2terminal cysteine residue. Furthermore, we have identified the signal peptide released from pro-BLP. These findings help in understanding the mode of export, processing, and assembly of lipoproteins in E. coli cells.
Bacterial Strain and Media-E. coli B (the wild type strain) was used throughout the study. Cells were grown in M-S/glucose (0.2%' w/v) medium at 37 "C with shaking.
Preparation of Envelope Fractions-To exponentially growing cells (3 X IO8 cells/ml) was added a 0.5% (w/v) solution of globomycin in methanol to a final concentration of 5 pg/ml. SDS-Polyacrylamide Gel Electrophoresis-Gel electrophoresis in 0.5% SDS with 10% (w/v) acrylamide gels was carried out as described (9). Sometimes 17.5% (w/v) acrylamide gel was used as described by Anderson et al. (12). Fluorography was performed as described by Bonner and Laskey (13). Kodak X-Omat R films were exposed to an electronic flashlight immediately before use (14). After developing the x-ray films, densitometric tracing was done with a Joyce h e b l densitometer at 610 nm (filter 5-042). The relative amount of protein (radioactivity) was estimated from the weight of the peak on the paper. For direct determination of radioactivity, gel regions located by fluorography were cut out and digested in NCS tissue solubilizer (Amersham International Ltd.), and radioactivity was counted in a liquid scintillation counter as described (15).
Other Methods-Protein was estimated by the method of Lowry et al. (16). Edman degradation of anti-BLP immunoprecipitates was carried out as described (9).

RESULTS
In Vitro Conversion of Pro-BLP to BLP-It has previously been shown that pro-BLP that accumulates in the envelope fraction of E. coli cells labeled with [3H]leucine in the presence of globomycin can be converted into BLP when the envelope fraction is incubated at 37 "C (9). For Fig. 1 arginine. Envelope fractions were solubilized in 10 nm sodium phosphate buffer (pH 7.1), 1% SDS, 1% 2-mercaptoethanol, 10% glycerol at 100 "C for 5 min, and analyzed on SDS-polyacrylamide gel (9). Slot 1 was a control sample without globomycin treatment. The envelope fraction from globomycin treated cells (8 X 10' cells) (slot 2) was washed with and suspended in 2 ml of 10 n w sodium phosphate buffer (pH 7.1) at 4 "C and incubated at 37 "C for 2 h in the absence (slot 3) or presence (slot 4) of globomycin (5 pg/rnl). After incubation, the envelope fraction was pelleted at 100,000 X g for 30 min and analyzed by gel electrophoresis as mentioned above. The gel was processed for fluorography. Each slot contained 30,000 cpm. The x-ray film was exposed for 3 days. The positions of BLP and pro-BLP are indicated.
prepared from cells labeled with [3H]arginine. As the numbers of arginine in pro-BLP and BLP are the same, quantitative conversion of pro-BLP to BLP was seen more clearly. For the conversion of pro-BLP to BLP, removal of globomycin from the envelope fraction was essential. When the envelope fraction was incubated in a small amount of buffer (e.g. envelope fraction from 8 X lo8 cells in 10 pl) at 37 "C, the conversion did not take place even after repeated washing of the envelope fraction at 4 "C. On the other hand, the peptidase reaction took place when the envelope fraction was incubated in a large volume of buffer (e.g. envelope fraction from 8 X 1 0 ' cells in 2 m l ) even at 37 "C. The results can be interpreted as follows. The interaction of globomycin with the envelope is so strong at lower temperatures that the drug cannot be removed even after repeated washing, while the interaction is weakened at an elevated temperature. Therefore, incubation of the envelope fraction in a large volume of buffer at a higher temperature stimulates the dissociation of the drug from the envelope, which in turn, initiates the peptidase reaction. As described later, we found that the peptidase was active at a temperature as high as 80 "C. At this temperature the reaction took place even when the envelope fraction was suspended in a small volume of buffer. This may be explained by efficient removal of the drug from the envelope at the extremely high temperature.
In order to isolate the peptidase(s) responsible for the removal of the signal peptide, solubilization of the activity with detergents was attempted. Hereafter this enzyme will be tentatively called signal peptidase. The envelope fraction was mixed with a detergent in 100 mM Tris-HC1 buffer (pH 7.1), incubated at 30 "C for 30 min, and examined on polyacrylamide gel for the processing. However, the treatment with the following detergents resulted in complete inactivation of the signal peptidase activity; 0.1% SDS, 2% sodium deoxycholate, 0.5% sodium sarcosinate, 2% Triton X-100, and 0.5% Nikkol. The activity could not be restored even when the detergenttreated envelope fraction was dialyzed against 10 mM Tris-HC1,5 m MgC12,0.02% NaN3 (pH 8.0) at 25 "C for 2 days to remove the detergent according to the method of Yamada and Mizushima (17). The latter experiment was not carried out with Nikkol or Triton X-100, since these detergents are hardly removed by dialysis. Therefore, the following studies on the signal peptidase activity were carried out using the intact envelope. The envelope fraction could be stored in an ice box for at least a week or at -80 "C for more than a month without loss of the peptidase activity.
Effect ofpH on the Signal Peptidase Reaction-The effect of pH on removal of the signal peptide upon incubation of the envelope fraction is shown in Fig. 2. It is notable that the processing took place to almost the same extent over the pH range of 4.0 to 11.0. Although the processing did not take place at pH 1.0 or 2.3, the activity was restored when the envelope fraction was resuspended in 10 mM sodium phosphate buffer at pH 7.0. Contrary to this, treatment at an extreme alkaline pH resulted in irreversible inactivation of the peptidase reaction.
Effect of Temperature on the Signal Peptidase Reaction-It is evident from Fig. 3 that the processing activity of the envelope fraction was stable over a broad temperature range. Even at 80 "C the envelope fraction showed considerable activity. The enzyme activity was totally lost on incubation at 96 "C for 2 min. There was no processing below 20 "C either. It is uncertain whether the lack of processing below 20 "C is due to inactivity of the enzyme or to incomplete dissociation of globomycin from the envelope. The processing activity was inhibited by globomycin at all temperatures. This indicates that the processing of pro-BLP at higher temperatures was PH FIG. 2. The processing activity of the signal peptidase at different pH values. The envelope fraction from 8 X 10' cells labeled with r3H]arginine in the presence of globomycin ( Fig. 1, slot 2) was suspended in 2 ml of solutions of different pHs and incubated at 37 "C for 1 h. Solutions used were diluted HC1 for pH 1.0 and 2.3, 0.1 N NaOH for pH 13.0, and Universal buffer (23) for pH 3.0 to 12.0. After incubation, the envelope fractions were pelleted, analyzed on SDSpolyacrylamide gel as mentioned in Fig. 1, and a fluorogram was taken. Each slot contained about 60, OOO cpm. By scanning the x-ray film, the relative amounts of BLP and pro-BLP were estimated and the percentage of conversion from pro-BLP to BLP at a definite pH was calculated. The processing activity of the signal peptidase at different temperatures. Tubes containing 2 ml of 10 mM sodium phosphate buffer (pH 7.1) were equilibrated at different temperatures. To each of these tubes, the envelope fraction from 8 X 1 0 ' cells labeled with ['Hlarginine in the presence of globomycin suspended in 10 pl of 10 n" sodium phosphate buffer (pH 7.1) was added and incubation was carried out for 20 min in the absence (0) or presence (A) of globomycine (5 pg/ml). Then the tubes were chilled and the envelope fraction was pelleted and analyzed on SDS-polyacrylamide gel as described in Fig. 1. Each slot contained 6 0 , O O O cpm. From the fluorogram, the percentage of conversion from pro-BLP to BLP was calculated as described in Fig. 2. not merely a chemical reaction but was catalyzed by an enzyme.
It should be stressed here that the peptidase reaction was assayed using an envelope fraction that had an organized membranous structure, and the substrate as well as the enzyme are proteins. Therefore, inactivation of the peptidase activity does not necessarily mean inactivation of the enzyme itself. It may also be due to denaturation of the substrate (pro-BLP) or the membrane structure.

Acylation of the Terminal Amino Group of BLP after
Signal Peptide Cleavage-The mature form of BLP contains three fatty acid residues at the NH2-terminal glycerylcysteine, two being ester-linked to glycerol and the other being amidelinked to the terminal amino-group (3). It is known that the pro-BLP that accumulates in the globomycin-treated cells contains the ester-linked fatty acids at the glycerylcysteine residue, the residue that becomes the NHz terminus of BLP (9). To determine whether the BLP formed by in vitro processing can also be acylated at the newly exposed terminal amino group, the increase of acyl residues during the conversion of pro-BLP to BLP was measured. Cells were treated with globomycin and labeled with both [3H]palmitic acid and ["C]arginine. As the numbers of arginine in pro-BLP and BLP are the same, ["Clarginine was used as an internal standard for BLP. In E. coli B, the palmitic acid contents in the amide-linked and ester-linked fatty acids are about 65% and 45%, respectively (18). Provided that [3H]palmitic acid is not converted to other species of fatty acid, it is assumed that if acylation of the NH2 terminus takes place, the 3H/'4C ratio should increase by 70% upon conversion of pro-BLP to BLP. The envelope fraction was isolated and incubated at 37 or 80 "C as described in the legend to Fig. 4. The envelope fraction was then delipidated with chloroform/methanol(2:1), solubilized in 1% SDS and immunoprecipitated with antiserum against BLP. The immunoprecipitates were subjected to SDS-polyacrylamide gel electrophoresis and a fluorogram was taken (Fig. 4A). Samples which were not incubated (slot 2) or incubated in the presence of globomycin (slots 6 and 8) contained pro-BLP only. Incubation at 37 "C for 10, 20, and 60 min gave rise to increasing amounts of BLP as expected (slots 3,4, and 5). Incubation at 80 "C for 7 min also gave rise to a substantial amount of BLP (slot 7). Although lipopolysaccharide, another fatty acid-containing molecule, should give a characteristic band, if it exists, at a position near that of pro-BLP in the gel system employed here (19), such a band can not be seen in Fig. 4, indicating that these immunoprecipitated samples were not contaminated by lipopolysaccharide. The bands of pro-BLP and BLP were cut out from the gel, digested, and the 3H and 14C counted. No signflcant changes were observed in the 3H/14C ratio of pro-BLP during the incubation, while a nearly 50% increase in the 3H/'4C ratio of the corresponding BLP bands was observed upon conversion from pro-BLP at 37 "C, suggesting that the terminal NH2 group of BLP thus formed had been modified by fatty acid. In contrast to the increase in 3H/14C ratio upon the processing at 37 "C, the ratio did not increase at 80 "C ( Fig. 4 A , slot 7). This indicates that at 80 "C, cleavage of the signal peptide was not accompanied by acylation of the NH2 terminus. This further suggests that the cleavage of the signal peptide and the acylation of the new NHP terminus is a 2-step process, the signal peptidase being active at 80 "C, and the acylation enzyme(s) being inactive at this temperature.

Difference between BLP Formed at 37°C
and That Formed at 80 "C-The results mentioned above suggest that BLP formed at 80 "C should be smaller in molecular weight than that formed at 37 "C. The band of BLP formed at 80 "C ( Fig. 4, slot 7) seemed to be in a position slightly below the position of native BLP or BLP formed at 37 "C. To c o n f i i this we analyzed the same samples in a 17.5% polyacrylamide slab gel with a discontinuous buffer system (12); the pattern is shown in Fig. 5 arginine in the presence of globomycin, suspended in 10 mM sodium phosphate buffer (pH 7.1) (slot 2). and incubated a t 37 "C for 10,20, and 60 min, or at 80 "C for 7 min (slots 3, 4, 5, and 7, respectively). Each tube contained the envelope fraction from 8 X 10' cells suspended in 2 ml of the buffer. To slots 6 and 8 globomycin (5 pg/ml) was added before incubation at 37 "C for 60 rnin and at 80 "C for 7 min, respectively. The pelleted envelope fractions were then extracted 3 times with chloroform/ methanol (2:1), solubilized in 1% SDS, and immunoprecipitated with antiserum against BLP. The immunoprecipitates were solubilized and subjected to SDS-polyacrylamide gel electrophoresis, as described in Fig. 1, and a fluorogram was taken. Slot I was the immunoprecipitate of the envelope fraction from control cells (no globomycin treatment). Radioactivity in individual slots was 9,600 cpm (slots I , 5, and 6). 8,000 cpm (slots 2 4 , and 10,000 cpm (slots 7 and 8). B, the bands of pro-BLP and BLP were cut out from the gel and digested with NCS tissue solubilizer as described (151, and the radioactivity was counted. The :'H/I4C ratios of pro-BLP (W) and BLP (0) are shown. A background value was obtained by determining the radioactivity in an approximately equal sized slice from an indicated area (bkg); this was subtracted from the radioactivity in the pro-BLP and BLP bands. In all cases the background radioactivity was very low. same position as the native BLP, while the BLP formed at 80 "C moved faster. This was evident when both the samples were run together. Fig. 5 also shows that the immunoprecipitated samples contained only BLP and pro-BLP, and were essentially free from any radioactive contaminants. The modification by fatty acid of the new NHz terminus was further investigated by using the same samples as follows. If the amino group of the terminal glyceride-bearing cysteine residue of BLP formed in vitro has been modified by fatty acid, this terminal residue should be resistant to Edman degradation. These anti-BLP immunoprecipitated samples were, therefore, subjected to one-cycle Edman degradation. Liberation of the 'H count from the 37 "C-chased sample was only 8%, while that from the 80 "C-chased sample was about 30%. Liberation of the I4C count during the one-cycle Edman degradation was almost negligible. These results further substantiated the interpretation of the results in Fig. 4. The relatively low efficiency of the ' H liberation even with the 80 "C-chased sample may be due to the unusual structure of the NH2 terminus.
Identification of the Signal Peptide Cleaved from Pro-BLP-During processing, the signal peptide region of pro-BLP could either be digested from one end or clipped between the glycine'" and cysteine" residues. In the latter case, the signal peptide released could further be digested by another peptidase(s). With the hope that the second peptidase is inactivated at 80 "C, identification of the signal peptide of BLP released during in vitro processing was attempted. Cells were labeled with [I4C]arginine, [14C]leucine, or [''SJmethionine in the presence of globomycin. Envelope fractions were isolated and an in vitro chase was performed at 37 "C for 60 min or at 80 "C for 7 min, and then analysis by SDS-polyacrylamide gel electrophoresis was carried out. The gel pattern of samples processed at 37 "C contained no new bands (data not shown), while the same samples processed at 80 "C showed a new band below the position of BLP (Fig. 6). This new band was observed concurrently with the processing of pro-BLP to BLP. When the processing was inhibited by the addition of globomycin, the new band did not appear. Moreover, 'this band was labeled with [14C]leucine or [:"S]methionine but not with [I4C]arginine. It is known that the signal peptide region of pro-BLP has no arginine residue. These results strongly suggest that the new band was the signal peptide of pro-BLP. The signal peptide of pro-BLP has only one methionine residue that is present at the NH:! terminus (6). As this new band was significantly labeled with [' %I methionine, the NH2 terminus of pro-BLP was suggested to be conserved in the signal peptide released during in vitro processing. It should be mentioned, however, that the signal peptide thus identified is assumed to contain a small amount of signal peptides derived from minor lipoproteins discovered recently (19). This will be discussed later.
The results in Fig. 6 also show that the signal peptide of BLP is metabolized by at least two types of peptidase; a heatresistant endopeptidase that removes the signal peptide from pro-BLP (signal peptidase) and a heat-sensitive peptidase that digests the cleaved signal peptide (signal peptide peptidase).  (slots 2,5, and 8), [lJC]leucine (slots 3, 6, and 9) or ["'SI methionine (slots 4, 7, and IO) were isolated. A part of them was solubilized and analyzed on SDS-polyacrylamide gel (slots 2-4) as described in Fig. 1. The rest was incubated at 80 O C for 7 min in the absence (slots 5-7) or present (slots [8][9][10] of globomycin (5 pg/ml). After incubation, the envelope fractions were solubilized and analyzed by SDS-polyacrylamide gel electrophoresis. Slot I was the envelope fraction from control cells (no globomycin treatment)) labeled with ["S]methionine. The gel was then fixed and dried, and an autoradiogram was taken. Each slot contained 1.8-2.0 X 10" cpm and the xray film was exposed for 9 days. The positions of pro-BLP and BLP are indicated. SP, signal peptide. The band indicated by an arrow is a cytoplasmic membrane protein (24).

DISCUSSION
Biosynthesis and assembly of BLP in E. coli is a complicated process. The BLP is fist synthesized as a precursor form that possesses a signal peptide of 20 amino acid residues extending from the NH2 terminus (6). Then it is converted to the mature form through cleavage of the signal peptide and modification of the newly bared NH2-terminal cysteine residue. During these processes, BLP is exported through the cytoplasmic membrane and assembled into the outer membrane.
With the aid of an antibiotic called globomycin, we found pro-BLP that had already been modified by glyceride in the cytoplasmic membrane (9). This indicates that the modification by glyceride takes place before the cleavage of the signal peptide. We have also presented evidence showing that pro-BLP can be exported through the cytoplasmic membrane while it still retains the signal peptide that is, most probably, held in the cytoplasmic membrane (20). This indicates that the cleavage of the signal peptide is not required for the export of the major part of the protein.
In the present work, we further found that at least two types of peptidase are involved in the processing of pro-BLP; an enzyme that removes the signal peptide (signal peptidase) and an enzyme that digests the cleaved signal peptide (signal peptide peptidase). We have also been able to show a third enzyme activity which is responsible for acylation of the newly exposed NH:! terminus of the cysteine residue. Fig. 7 is a summary of these processes which are discussed in detail below.
Signal Peptidase-In the present work, we were able to identify the signal peptide that was released from pro-BLP upon incubation of the globomycin-treated envelope. The signal peptide thus identified has methionine at the NHn terminus and its molecular weight was roughly estimated to be 2,000 daltons or higher.2 The results indicate that the signal peptidase responsible for pro-BLP is an endopeptidase that M. Hussain, S. Ichihara, and S. Mizushima, unpublished experiments.
clips the glycyl'"-glycerylcysteine2' bond. It should be noted that the signal peptide thus identified might have contained that from the minor lipoproteins (19), since precursors of these lipoproteins accumulate in globomycin-treated cells, and at least some of them could be chased to the mature form after the removal of globomycin. However, the total amount of these lipoproteins, numerically, is very small compared to that of BLP (19). Therefore, a greater part of the signal peptide is supposed to be that from pro-BLP. The signal peptide has been isolated in quantity and detailed characterization is now in progress in this laboratory.
The signal peptidase in the envelope fraction has many unique properties as a protease; being highly resistant to heat, having a wide optimum pH range, and being highly sensitive to detergents. These results suggest that a hydrophobic environment is important for the peptidase reaction; in other words, the expression of the peptidase activity may be closely related to the membrane structure. The signal peptidase is also unique in that its activity is not inhibited by the many protease inhibitors listed in the text, while it is inhibited by globomycin that does not inhibit cleavage of the signal peptide of other major outer membrane proteins (9). But it may be possible that the signal peptidase was not accessible to the protease inhibitors, especially to the hydrophilic ones, because of the localization of the enzyme in the membrane. As discussed in a previous paper (19), this signal peptidase is supposed to be specific for lipoproteins that include BLP and other minor lipoproteins. Wickner and his associates purified and characterized a signal peptidase for phage M13 procoat protein (21). This signal peptidase is apparently different from the signal peptidase for pro-BLP in that it is resistant to Triton X-100.
Signal Peptide Peptidase-Under the usual conditions, the

Signal Peptide
Cleavage in E . coli signal peptidase reaction did not result in accumulation of the signal peptide, indicating that the signal peptide must have been digested into smaller fragments immediately after release from pro-BLP. Incubation at 80 "C resulted in the appearance of a signal peptide on a gel, indicating that the E. coli cells possess a rather heat-sensitive protease that is responsible for digestion of the cleaved signal peptide. We tentatively named it signal peptide peptidase, although it is unclear whether the enzyme is specific for the digestion of signal peptides or not. It should be noted that pro-BLP was quite stable in the globomycin-treated envelope. This may suggest that the signal peptide region is stable against the action of the signal peptide peptidase as long as it is linked to BLP. Of course, the possibility that globomycin inhibits the signal peptide peptidase as well as the signal peptidase cannot be ruled out. However, this is unlikely, since the action of globomycin is so far limited to the inhibition of cleavage of the peptide bond next to the cysteine residue that is linked with glyceride. Acylation of the Terminal Amino Group That Was Newly Exposed as a Result of Signal Peptide Cleavage--In E. coli cells, removal of the signal peptide is accompanied by acylation of the newly exposed NH2 group of the cysteine residue. The results shown in Fig. 4 strongly suggest that cleavage of the signal peptide in vitro is accompanied by the acylation. The acylation was prevented when the cleavage reaction was carried out at 80 "C, suggesting that the cleavage of the signal peptide and the modification by fatty acid are sequential but independent reactions, only the enzyme(s) for the latter reaction being inactivated at 80 "C. The BLP formed by in vitro processing at 37 "C is most probably NHz-acylated as mature BLP is, since it migrated to the same position as that of the mature BLP, and the NHz-terminal cysteine residue was resistant to Edman degradation. These results also indicate that neither an external supply of energy nor additional cofactors are required for the modification by fatty acid as well as for the cleavage of the signal peptide. It has been reported that the acyl moiety of phospholipids is the precursor for the NHe-terminal fatty acid residue in BLP (22).
It should be noted that the in vitro system employed in the present work is a very simple one, and yet it most likely allows the entire biochemical reactions involved in the conversion of pro-BLP to BLP to occur. The relationship between these biochemical events and the translocation and assembly of BLP into the outer membrane is now being studied in this laboratory.