Maturation of Escherichia coli Maltose-binding Protein by Signal Peptidase I in Viuo SEQUENCE REQUIREMENTS FOR EFFICIENT PROCESSING AND DEMONSTRATION OF AN ALTERNATE CLEAVAGE SITE*

Comparative analyses of a number of secretory proteins processed by eukaryotic and prokaryotic signal peptidases have identified a strongly conserved feature regarding the residues positioned -3 and -1 relative to the cleavage site. These 2 residues of the signal peptide are thought to constitute a recognition site for the processing enzyme and are usually amino acids with small, neutral side chains. It was shown previously that the substitution of aspartic acid for alanine at -3 of the Escherichia coli maltose-binding protein (MBP) peptide blocked maturation by signal peptidase I but had no noticeable or MBP across the of its biological This identified an excellent in which a detailed the In


Comparative
analyses of a number of secretory proteins processed by eukaryotic and prokaryotic signal peptidases have identified a strongly conserved feature regarding the residues positioned -3 and -1 relative to the cleavage site. These 2 residues of the signal peptide are thought to constitute a recognition site for the processing enzyme and are usually amino acids with small, neutral side chains. It was shown previously that the substitution of aspartic acid for alanine at -3 of the Escherichia coli maltose-binding protein (MBP) signal peptide blocked maturation by signal peptidase I but had no noticeable effect or MBP translocation across the cytoplasmic membrane of its biological activity.
This identified an excellent system in which to undertake a detailed investigation of the structural requirements and limitations for the cleavage site. In vitro mutagenesis was used to generate 14 different amino acid substitutions at -3 and 13 different amino acid substitutions at -1 of the MBP signal peptide.
The maturation of the mutant precursor species expressed in viuo was examined.
Overall, the results obtained agreed fairly well with statistically derived models of signal peptidase I specificity, except that cysteine was found to permit efficient processing when present at either -3 and -1, and threonine at -1 resulted in inefficient processing. Interestingly, it was found that substitutions at -1 which blocked processing at the normal cleavage site redirected processing, with varying efficiencies, to an alternate site in the signal peptide represented by the Ala-X-Ala sequence at positions -5 to -3. The substitution of aspartic acid for alanine at -5 blocked processing at this alternate site but not the normal site. The amino acids occupying the -5 and -3 positions in many other prokaryotic signal peptides also have the potential for constituting alternate processing sites. This appears to represent another example of redundant information contained within the signal peptide.
Most secretory proteins are initially synthesized with an amino-terminal extension designated the signal peptide which serves to initiate export of the protein and which is subsequently removed during or immediately following translocation (1,2). Most signal peptides share common structural features: a hydrophilic amino-terminal segment with l-3 basic residues followed by a 9-15-residue core of hydrophobic and neutral amino acids and finally a more polar carboxyl terminus that immediately precedes the cleavage site (for review, see 3). According to the loop model (4), the positively charged amino terminus interacts with the membrane surface, and the hydrophobic core inserts into and spans the lipid bilayer as a loop or reverse hairpin structure, exposing the cleavage site on the external face. Many details of this model remain a matter of conjecture; still, the experimental evidence supports this orientation of the signal peptide during translocation (1,5,6). The enzymes responsible for the endoproteolytic removal of the signal peptide are termed signal peptidases. Eukaryotic signal peptidase has been purified from several tissue sources and is an integral membrane complex of two to six polypeptides (7,8). Two distinct signal peptidases have been purified from Escherichia coli cells (9). Both are integral membrane proteins composed of a single polypeptide. Signal peptidase II cleaves the signal peptides of lipoproteins exclusively, and signal peptidase I processes all other secretory proteins.
Although signal peptides exhibit little primary sequence homology, the comparative analysis of numerous prokaryotic and eukaryotic proteins has identified a strongly conserved feature regarding the residues at positions -3 and -1 relative to the cleavage site. It was noted that amino acids with small, neutral side chains predominated at these two positions (10,11). von Heijne (12) compared 36 prokaryotic signal peptides processed by signal peptidase I, and only alanine, glycine, leucine, serine, threonine, and valine were encountered at the -3 position. Somewhat more restrictive, position -1 was found to harbor only alanine, glycine, serine, and threonine. Ala-X-Ala is the most frequently observed sequence preceding the cleavage site (10,11). Based on these frequency analyses, positions -3 and -1 have been proposed to constitute a recognition site for the processing enzyme. These observations also were utilized to formulate the A-X-B model (10) and the (-3,-l) rule (11) for predicting signal peptide cleavage sites. The (-3,-l) rule was later further modified by incorporating a larger statistical basis and a weighted matrix approach (12). Various studies with E. coli have provided support for the basic tenets of these statistically based models (6,(13)(14)(15). However, such studies did not include a systematic analysis of multiple substitutions at -3 or -1 to determine experimentally the requirements and limitations for efficient signal peptide processing.

3417
This is an Open Access article under the CC BY license.
The MBP' signal peptide is composed of 26 amino acid residues and exhibits all the features typical of prokaryotic signal peptides, including alanine residues at both -3 and -1 (16). The substitution of aspartic acid for alanine at -3 was found to block MBP processing by signal peptidase I without having a noticeable effect on translocation across the cytoplasmic membrane (6). The preMBP remained anchored to the cytoplasmic membrane by its unprocessed signal peptide but was still capable of facilitating maltose uptake efficiently (6,17). The characterization of this mutant MBP species served to underscore the nonoverlapping nature of the hydrophobic core and processing region and identified an excellent system to conduct mutational analysis of cleavage site structural requirements with affecting protein translocation. In this study, oligonucleotide-directed mutagenesis has been utilized to construct plasmids encoding an assortment of mutant preMBP species altered at specific positions near the signal peptide processing site. The effect of 14 unique single amino acid substitutions at -3 and 13 unique substitutions at -1 on processing were examined and the results compared with predictions based on statistically derived models of signal peptidase I specificity. Interestingly, certain changes at -1 uncovered a nearby alternate cleavage site in the preMBP.  (19). Plasmid pJF8 is a derivative of pJF2 which harbors an amber mutation at the codon corresponding to position 23 of the MBP signal peptide (6). Plasmid pJF13 also is a derivative of pJF2 encoding MBP with aspartic acid substituted for alanine at residue -3 (6).

Bacterial
Reagents-Minimal medium M63 supplemented with carbon source (0.2%) and thiamine (2 pg/ml), maltose tetrazolium indicator agar, and TYE agar were prepared as described previously (20). When required, ampicillin was added to minimal and complex media at concentrations of 25 and 50 @g/ml, respectively. To induce malE genes under lacUV5 promoter-operator control, IPTG was used on agar plates and in liquid media at 1 and 5 mM, respectively. [

Effect of Single Amino Acid Substitutions at -3 of the MRP
Signal Pep&-Oligonucleotide-directed mutagenesis was used to construct a set of derivatives of plasmid pJF2 which differed only in the codon corresponding to position -3 of the MBP signal peptide (see "Materials and Methods"). As a result, a set of plasmids was obtained encoding mutant MBP species in which 1 of 14 different amino acids had been substituted for the alanine normally found at -3 (Table I). Cells harboring these plasmids, as well as all the other derivatives of pJF2 described in this study, exhibited a strongly Mal' phenotype, indicating that MBP translocation across the cytoplasmic membrane was not strongly affected (6). The effect of each alteration at position -3 on signal peptide processing is shown in Table I. Cells were grown to mid-log phase, induced for synthesis of plasmid-encoded MBP by addition of IPTG, and radiolabeled for 15 min with [""S] methionine.
MBP was immune precipitated from solubilized cell extracts and analyzed by SDS-PAGE and autoradiography. Signal peptide processing, as quantitated by measuring directly the radioactivity present in the precursor and mature MBP species, is presented as the percent of the total radiolabeled MBP precipitated which is mMBP. Essentially all the wild-type MBP (98%) was processed to mMBP under these conditions.
Substitution of asparagine for alanine at -3 reduced processing to 25%. The presence of aspartic acid, arginine, histidine, tyrosine, or phenylalanine at -3 resulted in extremely inefficient processing of the signal peptide over the time course of these experiments.
Using a pulse-chase analysis, the kinetics of signal peptide processing was determined for four of the substituted MBP species (Fig. 2). At the earliest chase point (2 min), maturation was complete for both wild-type and Val-" preMBP. The majority of Pro-" preMBP was processed at the 2-min chase point, with the remaining preMBP converted to mMBP by 60 min of chase. Substitution of asparagine at -3 resulted in slow processing of preMBP which was not completed by the last chase point. In contrast to the other MBP species examined, maturation of Arg-" preMBP could not be detected even after a full 60-min chase period. Substitutions at -1 Have Varied Effects on Signal Peptide Processing-Thirteen different amino acid substitutions at the -1 position of the MBP signal peptide were obtained and similarly analyzed (Table II). Precursor MBP species harboring glycine, serine, or cysteine at -1 were virtually completely processed following a 15min radiolabeling period. In contrast, the other -1 substitutions resulted in efficiencies ranging from 14 to 91%. Kinetic analyses of five representatives from this latter group revealed that processing of each MBP species examined occurred relatively slowly when compared with wild-type preMBP (Fig. 3). The finding that preMBP species with aspartic acid or certain other charged or bulky residues at -1 were processed to significant extents was unexpected. These results suggested that either the statistically based predictions for acceptable residues at the -1 position (10-12) were largely incorrect or that substitutions at -1 were redirecting processing to an alternate site. Within the MBP signal peptide, there is a second potential signal peptidase cleavage site represented by the Ala-X-Ala sequence at residues -5 to -3 (see Fig. 1). If this alternate cleavage site were being utilized due to certain substitutions at -1, the resulting Processing of E. coli Maltose-binding Protein mMBP would be expected to retain an additional 2 amino acids at its amino terminus. The possibility of cleavage at this alternate site was supported by the finding that processing of Arg-', His-', and Asn-' preMBP resulted in mMBP species that clearly migrated more slowly than wild-type mMBP when analyzed on a lower percentage polyacrylamide gel (Fig. 4). Mature MBP from the Asp-' and Tyr-' mutants also migrated slower than wild-type mMBP (data not shown), whereas the small amount of mMBP obtained from the Phe-' mutant migrated the same as wild-type mMBP (Fig. 4).
Amino-terminal Analysis of Various mMBP Species-In order to identify conclusively the processing site of four representative MBP species harboring amino acid substitutions at -1, the processed forms of these proteins were purified, and Edman protein degradation (see "Materials and Methods") was utilized to determine the actual amino acid sequence at the amino terminus (data not shown). For cells synthesizing Asp-', Val-', and Arg-' preMBP, this analysis clearly indicated that the amino-terminal residue of mMBP was the Leu-*, confirming that cleavage occurred immediately after the Ala-X-Ala sequence at residues -5 to -3. When threonine was substituted at -1, two mMBP species were detected. In this instance, approximately 75% of the preMBP processed was cleaved after residue -3, with the remainder being cleaved after the threonine at -1. Recoveries of phenylthiohydantoin derivatives were quantitated at three positions (Leui-Lys'; Lys3-Glu3; Gly'-Leu') and were consistent with approximately 3:l ratios. Note that this was the only preMBP species obtained in this study in which processing at both the normal and alternate cleavage sites could clearly be demonstrated (Table II). For cells synthesizing wild-type Vale3 or ProF3 preMBP, only authentically processed mMBP was detected.

Processing of MutantpreMBP Species Lacking the Alternate
Cleauage Site-To determine whether -1 substitutions specifically blocked processing at the normal cleavage site without the complication of processing being redirected to an alternate upstream site, the Ala+ of each of these preMBP species was converted to aspartic acid (see "Materials and Methods"). This additional substitution was expected to block cleavage at the alternate site. This substitution did have a minor effect on export and processing of preMBP with alanine (wild-type residue) at -1 ( Fig. 5) Aspartic acid at -5 was found to have a similar minor effect on processing of preMBP species having glycine, cysteine, or serine at -1. In contrast, disruption of the alternate site resulted in a total block in processing of preMBP species having aspartic acid, arginine, tyrosine, leucine, or isoleucine at -1, further confirming that the processing observed previously for the corresponding preMBP species having alanine at -5 was effected at the alternate site. Consistent with the amino-terminal analysis of mMBP obtained from the Thr-' mutant, a small amount of processing of the corresponding preMBP species harboring site. The experimental conditions are the same as those described in the legend to Fig. 4. The amino acid substitutions at -1 of preMBP are indicated above each lane by single letter code. Alanine is the residue found at -1 in wild-type preMBP. + denotes synthesis of wild-type preMBP. Note that all of the preMBP species except wild-type have an aspartic acid substituted for alanine at -5, which blocks processing at the alternate site. aspartic acid at -5 was discerned. Although the amount of processing at the normal site of this mutant was somewhat less than expected, this is most likely due to the alteration at the -5 residue which, as noted above, did have some small effect on processing of otherwise wild-type preMBP. Finally, with regard to Phem5 preMBP, repeated attempts to construct a double-mutant MBP species with an aspartic acid at -5 were unsuccessful.
Substitutions at Position +2 of the Mature Moiety-The efficiency of processing at the alternate site was clearly influenced by the residue occupying the -1 position relative to the normal cleavage site (see Table II and Fig. 3). The substitutions at -1 are actually positioned +2 to the alternate cleavage site. The possibility that substitutions at +2 relative to the normal cleavage site could affect the efficiency of processing at this site was investigated. Once again, in uitro mutagenesis was employed to obtain cells synthesizing preMBP species with the following amino acids substituted for the isoleucine normally present at the second position of the mature moiety: arginine, asparagine, aspartic acid, cysteine, histidine, leucine, proline, tyrosine, and valine. In each instance, the signal peptide also had aspartic acid substituted for alanine at -5, thus blocking the alternate cleavage site. The substitutions generated at +2 included residues that had widely varying effects on alternate site processing when they were present at -1 (e.g. aspartic acid and tyrosine). Processing of these mutant preMBP species was analyzed by pulse-chase experiments similar to those shown in Figs. 3 and 4. Even at the earliest chase point (in this case 30 s), there were no detectable differences in the processing kinetics observed for any of these mutant preMBP species compared with that discerned for the parental preMBP species with a wild-type mature moiety (data not shown).

Additional Alterations
Near the MBP Cleavage Site-It was of interest to determine if the small amount of processing of Aspm3 preMBP which could be discerned (see Fig. 4, third lane, and Ref. 6) was due to specific recognition of the altered cleavage site by signal peptidase I or by some less specific protease activity in the periplasm. It was found that mMBP could not be detected if preMBP was synthesized with aspartic acid residues at both -3 and -1, indicating that the previously observed maturation of Aspm3 preMBP was most likely achieved by specific processing at the normal site. Likewise, maturation of a preMBP species deleted for residues -3 to -1 could not be detected (data not shown).

DISCUSSION
Among prokaryotic signal peptides, only alanine, glycine, serine, leucine, threonine, and valine have been found at position -3 (12). Consistent with the predictions of Perlman and Halvorson (10) and von Heijne (11,12) this study demonstrated that the presence of any of these residues at -3 was compatible with very efficient preMBP processing. In addition, cysteine at -3 had no noticeable effect on the efficiency of preMBP processing (see below). Cleavage of Prod3 preMBP also was accomplished with high efficiency, although the kinetics indicated that this precursor species was somewhat less than an optimum substrate for the processing enzyme. Proline is conspicuously absent from the -3 to +l regions of prokaryotic signal peptides (12). It is frequently encountered at positions -4 to -6, where it is believed to induce a P-turn that may be crucial for efficient presentation of the cleavage site to the processing enzyme (see below). Amino-terminal analysis of the processed forms of Prom3, Valm3, and wild-type MBP indicated that each species was cleaved exclusively at the normal site.
Kuhn and Wickner (14) reported previously that phage Ml3 procoat with a phenylalanine at -3 was a very poor substrate for signal peptidase I, and Fikes and Bassford (6) found that preMBP with an aspartic acid substituted for alanine at -3 was translocated efficiently but not processed. In agreement with these results and the von Heijne model (29), the substitution of bulky or charged residues (asparagine, aspartic acid, arginine, histidine, tyrosine, and phenylalanine) at -3 of preMBP resulted in extremely inefficient processing. The small amount of processing observed for AsnT3 and ASP-~ preMBP was probably due to specific recognition of the altered cleavage site by signal peptidase I since the introduction of an aspartic acid residue at -1 of the ASP-~ preMBP or deletion of residues -3 to -1 from otherwise wild-type preMBP resulted in a total block in processing. The signal peptide of unprocessed preMBP is normally highly susceptible to proteolytic degradation, resulting in a polypeptide that migrates identically with mMBP on SDS-PAGE (30, 31). However, it was shown previously that ASP-~ preMBP is efficiently translocated into the periplasm but remains firmly anchored to the cytoplasmic membrane by the signal peptide (6). The signal peptides of ASP-~ preMBP and the other mutant preMBP species described in this study presumably are shielded from proteolytic attack by their membrane association.
The analysis of changes at the -1 position was somewhat more complicated. Alanine is clearly the residue most frequently encountered at -1, although serine, glycine, and, in one instance, threonine also have been reported (12). Indeed, altered preMBP species having glycine, serine, and also cysteine (see below) substituted for alanine at -1 seemed to be processed as well as wild-type preMBP. Processing of Thr-' preMBP appeared to be nearly as efficient. However, processing at higher than expected efficiencies also was observed for signal peptides having other amino acids substituted at -1. For example, maturation of both Asp-' and Val-' preMBP approached nearly 100% in a pulse-chase experiment. For these and other preMBP species with bulky or charged residues at -1, it was found that processing was not occurring at the normal site. Instead, these preMBP species were primarily cleaved between residues -3 and -2, immediately downstream from the sequence Ala-Ser-Ala that served as an alternate processing site when maturation at the normal site was blocked. Examples of alternate cleavage sites for signal peptidase I in mutant lipoproteins have been described previously (32,33).
Processing at the alternate site of preMBP was initially indicated by the finding that in some cases, the mature species migrated slightly slower on SDS-PAGE than did the wildtype mMBP. This altered electrophoretic mobility was detected when aspartic acid, arginine, asparagine, histidine, or tyrosine was substituted at position -1 and presumably resulted from the retention of the charged or polar residue at the amino terminus of the mature protein. Amino-terminal analysis subsequently confirmed that the Asp-', Arg-', as well as Val-' preMBP species were processed after residue -3. This analysis also revealed that Thr-' preMBP exhibited processing at both the normal and alternate sites, primarily the latter (approximately 75%). Although threonine has been reported at the -1 position of one naturally occurring prokaryotic signal peptide (12), the results obtained with preMBP were more consistent with those of Kuhn and Wickner (14), who found that Thr-' of MI3 procoat was a poor substrate for signal peptidase I, both in viva and in uitro.
In lieu of determining the cleavage site of each mutant preMBP species altered at -1 by amino-terminal analysis of the purified mature protein, the alternate processing site of these proteins was disrupted by substituting aspartic acid for alanine at -5. In these instances, only preMBP species with alanine, serine, glycine, or cysteine were processed efficiently. A small amount of processing of Thr-' preMBP was discerned; otherwise, disruption of the alternate site appeared to block maturation when aspartic acid, arginine, valine, tyrosine, leucine, isoleucine, histidine, or asparagine was present at -1. These results are very consistent with the conclusions of von Heijne (11) that aromatic, charged and large polar residues are excluded from this position of the cleavage site. As noted previously, attempts to substitute aspartic acid for alanine at -5 of Phe-' preMBP were unsuccessful for an unknown reason. However, it seems highly likely that the small amount of processing observed for this MBP species was also accomplished at the alternate site.
The finding that cysteine residues at either -3 or -1 permitted efficient processing of preMBP was not necessarily surprising.
Although not yet encountered in the -4 to +2 region of prokaryotic precursor proteins processed by signal peptidase I, cysteine is a small, neutral residue that is present in this position in many eukaryotic signals (12). Its exclusion from the processing site of prokaryotic signals may be to help distinguish precursor nonlipoproteins from precursor lipoproteins that require a cysteine residue in the +l position (34). The possibility that mutant preMBP species with cysteine at either -3 or -1 were modified and processed as lipoproteins by signal peptidase II was not specifically investigated. However, the consensus sequence for lipoproteins at residues -3 to +l (Leu-Ala-Gly-Cys) (34) is considerably different from the corresponding sequences of these proteins (Ser-Ala-Ser-Cys and Ser-Ala-Leu-Cys, respectively; see Fig. 1). Since the mature MBP produced by processing of Cysm3 and Cys-i preMBP migrates on SDS-PAGE identically to wild-type mMBP (see Fig. 4), this possibility seems highly unlikely.
Overall, the results obtained experimentally with altered preMBP species agree fairly well with the statistical analyses of prokaryotic signal peptides by von Heijne (11,12), with the exceptions that cysteine was found to permit efficient processing when present at either -3 and -1, and threonine at -1 resulted in inefficient processing. The data presented here do not address directly the accuracy of the weighted matrix approach for predicting the cleavage sites of prokaryotic signal peptides processed by signal peptidase I (12). Nonetheless, the statistically derived assumptions concerning positions -3 and -1 which constitute a key element to this approach now have been tested directly and found to be correct with only several minor exceptions.
By studying the processing of several prokaryotic proteins in microsomal extracts, Ghrayeb et al. (32) concluded that signal peptidase I and eukaryotic signal peptidase possess similar substrate specificities. More recently, Folz et al. (35) used site saturation mutagenesis to obtain 13 unique amino acid substitutions at the -1 position of a mutant human preproapolipoprotein(a-II). Efficient processing in an in vitro system was observed with alanine or cysteine in the -1 position. Processing, albeit less efficient, was also discerned with glycine, serine, threonine, or proline at -1. These results may indicate that there are minor differences in the substrate specificities of the prokaryotic and eukaryotic enzymes, although this analysis is somewhat difficult to interpret since kinetic studies were not employed, and an alternative processing site also could be utilized, in this case representing the normal processing site of the wild-type preproapolipoprotein.
In this study, processing at the alternate preMBP cleavage site was only detected when processing at the normal site was blocked or rendered highly inefficient by substitutions at -1. Why is this alternate site not utilized at some detectable frequency in the wild-type signal peptide? In those instances in which the alternate site was cleaved, the processing kinetics were much slower than those exhibited by MBP species processed at the normal site.* Thus, in the wild-type signal peptide, extremely rapid cleavage at the normal site may effectively preclude the slower, alternate processing event. Less efficient recognition of the alternate site by signal peptidase I probably results from its proximity to the &turn structure that represents the boundary between the hydrophobic core and the signal peptidase recognition sequence (10) and that is known, from mutational analyses, to be essential for signal peptide processing (13-l&36). In prokaryotic signal peptides, the cleavage site is usually separated from the end of the core by six residues. As proposed by Perlman and Halvorson (lo), proper juxtaposition of the cleavage site with signal peptidase probably depends strongly on its location relative to the start of the turn. This was demonstrated recently in a convincing fashion for eukaryotic signal peptidase (37). Thus, for preMBP, the alternate processing site seems to be positioned 2 residues too close to the ,&turn initiated by the serine at -6 (10) for efficient recognition and cleavage by signal peptidase I.
It was somewhat puzzling to find that there was a wide variation in the efficiency of alternate processing in different mutant preMBP species in which processing at the normal site was blocked. These alterations at -1 relative to the normal cleavage site are actually positioned +2 to the alternate site. Previous studies with Ml3 procoat indicated that changes at +2 could profoundly affect processing efficiency (38, 39), although stepwise deletions from the cleavage site into the mature region of /I-lactamase provided no indication that sequences on the carboxyl side of the cleavage site could influence the reaction kinetics (15). Nine different substitutions for the isoleucine normally present at +2 of preMBP were found to have no noticeable effect on processing at the normal site, This included several amino acids (e.g. arginine and tyrosine) that, when present at -1, significantly reduced the efficiency of processing at the alternate site. These results indicated that there probably is not a strong restriction per se on the amino acid that can occupy the +2 position relative to the cleavage site. A recent study by Duffaud and Inouye (40) demonstrated that mutational alterations in the adjacent mature region which decreased the probability of a turn structure at the cleavage site itself could strongly affect the processing efficiency of an OmpA-staphylococcal nuclease chimeric protein expressed in E. coli. The possibility was 2  considered that the alternate preMBP cleavage site, compared with the normal site, was much more sensitive to changes at +2 which affected the turn probability through that region. However, there does not appear to be a strong correlation between alternate site processing efficiencies and the propensity of the residue substituted at -1 of the normal cleavage site to participate in a turn structure.3 This is being investigated further. Finally, what is the significance of the finding of an alternate processing site in the preMBP? Its presence may represent the rule rather than the exception. The LamB, galactosebinding protein, and ribose-binding protein signal peptides also harbor the sequence Ala-X-Ala-X-Ala at positions -5 to -1 (41). Overall, the frequency of compatible residues at -5 and -3 for constituting potential alternate processing sites in prokaryotic signal peptides is quite high (41). There may be circumstances in which the presence of alternate cleavage sites ensures efficient processing. It may also ensure processing in the event of a mutational alteration elsewhere in the signal which shifts the proximity of the core to the normal cleavage site. The existence of alternate cleavage sites also appears to reflect a general redundancy in the features that have been identified as being important to promoting protein export. For example, many prokaryotic signal peptides have more than 1 basic residue in the amino-terminal hydrophilic segment, whereas several studies have shown that a single such residue is sufficient (42,43). Likewise, the hydrophobic cores of most signal peptides appear to be longer than required (44,45). Although this built-in redundancy may seem superfluous in the laboratory setting, nature may have deemed it important in maintaining protein translocation and signal peptide cleavage as rapid and highly efficient cellular processes.