Requirement of Pro-sequence for the Production of Active Subtilisin E in Escherichia coli*

Subtilisin E, an alkaline serine protease of Bacillus subtilis 168, is first produced as a precursor, pre-pro-subtilisin, which consists of a signal peptide for protein secretion (pre-sequence) and a peptide extension of 77 amino acid residues (pro-sequence) between the signal peptide and mature subtilisin. When the entire coding region for pre-pro-subtilisin E was cloned into an Escherichia coli expression vector, active mature subtilisin E was secreted into the periplasmic space. When the pre-sequence was replaced with the E. coli OmpA signal peptide, active subtilisin E was also produced. When the OmpA signal peptide was directly fused to the mature subtilisin sequence, no protease activity was detected, although this product had the identical primary structure as subtilisin E as a result of cleavage of the OmpA signal peptide and was produced at a level OP approximately 10% of total cellular protein. When the OmpA signal peptide was fused to the 15th or 44th amino acid residue from the amino terminus of the pro- sequence, active subtilisin was also not produced. These results indicate that the pro-sequence of pre- pro-subtilisin plays an important role in the formation of enzymatically active subtilisin. It is proposed that the pro-sequence is essential for guiding appropriate folding of the enzymatically active conformation of subtilisin E.

I Present address: Central Research Lahoratories, Ajinomoto Co., Inc., 1-1 Suzuki-cho, Kawasaki 210,Japan. " subtilisin is thought to be first secreted across the cytoplasmic membrane with the aid of the signal peptide. Subsequently, the pro-sequence is removed either autocatalytically or by pre-existing active subtilisin to produce active mature subtilisin, which is then released into the medium (3).
Such pro-sequences have been found in many secretory proteins in various Gram-positive bacteria, although the lengths of the pro-sequences are widely different from protein to protein. For example, 8 amino acid residues are found in the penicillinase from B. licheniformis (14), 19 residues in nuclease A from Staphylococcus aureus (15), and nearIy 190 residues in the neutral protease from B. subtilis (16). The function of the pro-sequence has not been well established. The pro-sequence may be required for the association of the pro-enzyme with the cell before the release of the mature active enzyme into the medium and/or for guiding the protein to the appropriate folding for the active conformation. In our earlier study on staphylococcal nuclease, we have shown that, when the signal peptide for OmpA, a major Escherichia coli outer membrane protein, is directly fused to the mature nuclease, active nuclease is efficiently secreted into the E. coli periplasmic space (17). This result indicates that, at least in the case of staphylococcal nuclease, the prosequence is not required for the folding of the active enzyme.
In the present study, we have performed a similar construction with subtilisin E and found that, in contrast to staphylococcal nuclease, active subtilisin is not produced when the OmpA signal peptide is directly fused to mature subtilisin in spite of the fact that a large amount of a polypeptide having the identical primary structure to subtilisin E is secreted into the periplasmic space. From these results, we propose that the subtilisin pro-sequence plays an essential role in guiding the proper folding of the protein to give active subtilisin.
All the restriction enzymes used were from New England Biolabs, Bethesda Research Laboratories, International Biotechnologies (IBI), or Boehringer Mannheim and used as recommended by the suppliers.
Streptomyces subtilisin inhibitor (21-23) was a gift from Dr. S. Sato at the Biochemical Research Laboratory of Toyo Jozo Ltd., Shizuoka, Japan.
Construction of Plasmids-Cloning of the subtilisin E gene was performed as shown in Fig. 1 by ligating the 2.5-kilobase pair KpnI-EcoRI fragment from the total chromosomal DNA of B. subtilis 168 directly to pUC18 cloning vector (IBI) on the basis of the DNA sequence of the subtilisin gene (9). A clone of the subtilisin E gene was isolated by the colony hybridization technique using a synthetic oligonucleotide, 5'-AAAGGGTTAATCAACG-3' a5 a probe. Various   Ghrayeb et al. (20). DNA fragments cloned with PIN-111-ompA were obtained by various restriction enzyme digestions of the cloned gene for subtilisin E: AccI-XmnI digestion for pHI126, DraI-XmnI for pHIlO2, HpuI-XmnI for pHI103, FspI-XmnI for pHI100, and NaeI-XmnI for pHI114. These fragments are shown by lines with arrows, at the bottom. The number at the kft-hand side for each line is the residue number from the amino terminus of mature subtilisin (+l). The plasmids pHI212, 214, and 215 were derived from pH1126 by oligonucleotide-directed site-specific mutagenesis. In the linear restriction map, the solid box represents the coding region for mature subtilisin, the shaded box for the pro-sequence, and the empty box for the pre-sequence. DNA fragments obtained by restriction enzyme digestions from this clone were inserted into the unique Hind111 site of the PIN-111-ompA expression vector (20).
The AccI-XmnI fragment of the subtilisin E gene was inserted into a PIN-111-ompA3 vector after the vector plasmid was digested with HindIII and treated with DNA-polymerase Klenow fragment. As shown in Fig. 1 and Fig. 2, the resultingplasmid (pH11261 contains the entire coding sequence of subtilisin plus 48 extra bases from the 5' uncoding region including its own Shine-Dalgarno sequence after the coding region for the OmpA signal peptide. As a result, synthesis of a peptide starting from the OmpA signal peptide terminates at residue -108 (see Fig. 2). However, in this plasmid (pHI126), protein synthesis can be independently initiated from the subtilisin signal peptide at residue -106. Similarly, the DraI-XmnI fragment and the HpaI-XmnI fragment were inserted into a PIN-111-ompA, vector, yielding pH1102 and pHI103, respectively (Fig. 1). The plasmid pH1100 was prepared by ligating the FspI-Xmnl fragment to the large linear fragment of the pIN-III-ompA2 vector resulting from the digestion of the vector with EcoRI and HindIII followed by treatment with DNA-polymerase Klenow fragment. The NaeI-XmnI fragment was also inserted into the PIN-111-ompA3 vector like pHI126. The resulting plasmid was digested with EcoRI, treated with DNA-polym-erase Klenow fragment, and then religated to adjust its reading frame, yielding pHIl14.
In order to construct pHI212, pHI214, and pHI215, site-specific mutagenesis was carried out on pH1126 as described later.
Site-specific Mutagenesis-Oligonucleotides were synthesized on a Systec Microsyn 1450 DNA synthesizer using phosphoamidite chemistry (24). Monomers for the synthesis were purchased from American BioNuclear. Oligomers were purified by polyacrylamide gel electrophoresis. Site-specific mutagenesis was carried out according to the method of Inouye and Inouye (25) directly on the plasmids and the mutation was confirmed by DNA sequencing.
In order to delete a DNA sequence between the -366th and -250th base from pH1126 (see Fig. 2), a synthetic oligonucleotide, oligomera (Fig. 3), was used for site-specific mutagenesis. As a result of this deletion, the OmpA signal peptide was fused to the -83rd residue from the pro-sequence cleavage site (or the -6th residue from the pre-sequence cleavage site; see Fig. 2) to generate plasmid pH1212 (see also Fig. 1). Similarly, oligomer-b and -c (Fig. 3) were used in order to delete DNA sequences from the -366th to -238th base, and from the -366th to -232nd base, respectively (see Fig. 2). As a result of the former deletion, the OmpA signal peptide was fused to the -79th amino acid residue from the pro-sequence cleavage site (or the FIG. 2. Partial DNA sequence of the plasmid pH1126 around the translation initiation site and the following aminoterminal region of pre-pro-subtilisin, with the amino acid sequences derived from the DNA sequence. Numbers indicate the base positions (above the DNA sequence) and amino acid positions (under the amino acid sequence, in parentheses) from the prosequence cleavage site. The OmpA signal peptide sequence and the subtilisin pre-sequence are underlined with arrows indicating cleavage sites, a filled arrow for the OmpA signal peptide and an open arrow for the subtilisin pre-sequence, respectively. The cleavage site of the subtilisin pro-sequence is also shown by a filled arrow. The sites where the OmpA signal peptide were fused are shown by arrows with plasmid names, or by residue numbers with plasmid names. The amino acid replacement in pH1216 at residue 32 is also shown. there is an extra sequence, Asn-Met-Ser-Ala-Gln-Ala, between the OmpA signal peptide cleavage site and the pro-sequence; and for pHI214, the extra sequence Gln-Ala. Another oligonucleotide, oligomer-d (Fig. 3), was used to alter Asp-32 to Asn-32 on pHI212, which is known to eliminate protease activity completely (3). This plasmid was designated pH1216.

QmnAA -271
A r g G l y --- B, the sequences of pHI212,214, and 215; C, the sequences of pH1100 and pHT700; D, the sequences of pHI102, 103, and 114. The amino acid residues deleted from pH1100 are boxed.

I~e P h a T h r~e t A l a P h e S e r A~n K e t S a r
In order to delete an insertion (Glu-Leu) at the fusion site between the OmpA signal peptide and the subtilisin E mature sequence from pHI100, oligomer-e (Fig. 3) was used, yielding pHT700. The amino acid sequences at the fusion site for pH1100 and pHT700 are shown in Fig. 4C. In the case of pHT700, the OmpA signal peptide was directly fused to the mature subtilisin sequence.
Expression of the Subtilisin E Gene-Expression of the subtilisin E gene was performed at 37 "C in M9 medium (26) supplemented with casamino acids (2%). At a Klett reading of 50 with a blue filter, IPTG was added to the culture medium to a final concentration of 2 mM in order to induce gene expression. After 2-h induction, cells were harvested and the gene products were analyzed. In some experiments, induction was carried out with 0.005 mM IPTG at a Klett reading of 100 at 23 "C for 3 h. Cell fractionation was carried out as described previously (17). For protease activity assay, an aliquot of a sample solution was incubated at 37 "C in 750 pl of 50 mM Tris-HC1 (pH 8.5) containing 0.13 mM succinyl-Ala-Ala-Pro-Phe-p-nitroaniline (Vega). After adding 250 pl of 2 M sodium citrate (pH 5.0) to the solution, the amount of released p-nitroaniline was measured by the absorbance at 410 nm (27). The activity was calculated as unit/ml of starting cell culture (1 unit was defined as 1 pmol of p-nitroaniline per h).
Purification of Subtilisin-Subtilisin E from B. subtilis 168 was purified according to the method described by Estell et al. (7) using a CM52 column. In the case of active subtilisin E produced in E. coli, the periplasmic fraction of induced cells was applied to a CM52 column, since most of the activity was found in the periplasmic space as described later.

Production of Active Subtilisin in
E. coli-When the plasmid pH1126 was constructed as an intermediate for the following mutageneses (see Fig. l), it was found that the cells carrying this plasmid produced a small amount of active subtilisin in the presence of IPTG, and the cell fractionation experiment showed that active subtilisin was secreted into the periplasmic space (data not shown). Since the subtilisin gene can be expressed only from its own initiation codon at amino acid residue -106 (see Fig. 2) by this plasmid, we infer that the subtilisin signal sequence is functional in E. coli and that prosubtilisin can be processed to active subtilisin in the E. coli periplasmic space.
In order to examine whether the subtilisin signal peptide can be replaced with the OmpA signal peptide, the OmpA signal peptide was fused to pro-subtilisin a t three different sites within the pre-pro-subtilisin signal sequence. This gave rise to three plasmids, pHI212,214, and 215, as shown in Fig.   1. The amino acid sequences of these proteins at the fusion regions are shown in Fig. 4B. In all cases, when induced with IPTG, protease activity was detected in both the culture medium and the soluble fraction of the sonicated cells. No significant differences in the total activity were found among the three constructions described. When cells carrying pH1212 were induced for 2 h with 2 mM IPTG at 37 "C, the total activity per milliliter of culture was 69 units. These results indicate that active enzyme was produced from these fusion proteins.
When total cellular protein from the cells described were analyzed by SDS-polyacrylamide gel electrophoresis as shown in Fig. 5, lanes 2-4,  authentic subtilisin E ( l a n e I). In the case of cells carrying pH1212 ( l a n e 2 ) , in addition to the mature subtilisin band, two new major bands were seen, a band migrating a little slower and the one a little faster than the band for elongation factor Tu. The apparent molecular masses of the upper and the lower bands were estimated to be 44 and 42 kDa, respectively. From these molecular masses, the upper band is deduced to be the full-length OmpA-pro-subtilisin from pHI212, and the lower band to be pro-subtilisin resulting from the cleavage of the OmpA signal peptide. Judging from the densities of these bands, approximately half of OmpA-pro-subtilisin appears to be processed to pro-subtilisin. In the case of pH1214 and pH1215 (lanes 3 and 4, respectively), the upper bands migrated almost at the same position as elongation factor Tu and pro-subtilisin migrated a little faster than prosubtilisin from pH1212 ( l a n e Z ) , as expected from the structure shown in Fig. 1. In all cases, the plasmid-derived proteins (the upper band plus the lower band) became the major products of the cells. It was not possible to determine the cellular location of the OmpA-subtilisin hybrid proteins owing to their insolubility. It should be noted that, in all cases, cells carrying these plasmids were unable to form colonies on plates containing 0.5 mM IPTG. In liquid cultures, cells lysed when grown in the presence of IPTG. This lysis is probably a major reason for the protease activity found in culture medium.
When the cells were induced by lower concentrations of IPTG or at lower temperatures, this lysis occurred much later, and less activity was found in culture medium. When the cells carrying pH1212 were induced with 0.005 mM IPTG at 23 "C for 3 h, the total activity per milliliter of culture increased to 1100 units from 69 units, which was obtained at 37 "C with 2 mM IPTG. Under these conditions, 89% of the total activity was recovered in the periplasmic fraction of the induced cells. Furthermore, lesser amounts of the precursor forms were detected when these cells were analyzed by SDS-polyacrylamide gel electrophoresis (data not shown).
The active enzyme produced in E. coli harboring pHI212, 214, and 215 was identified as subtilisin E since the protease activity was inhibited by the serine protease inhibitor phenylmethanesulfonyl fluoride (1) and a specific subtilisin inhibitor, streptomyces subtilisin inhibitor (21-23). The SDS-polyacrylamide gel electrophoresis of purified enzyme from the cells harboring pH1212 is shown in Fig. 6.
Since subtilisin activity is thought to be required for maturation of subtilisin in B. subtilis (3), a mutation abolishing that activity was constructed on pHI212. In this mutant plasmid, pHI216, Asp-32, which is known to be essential for protease activity, was replaced by Asn-32. Both precursor bands produced by pH1212 were observed in pHI216-bearing strains (Fig. 5, lane 5); however, no mature band was seen.
Induction of subtilisin from pH1216 was also performed a t 23 "C. Although this temperature favors production of mature subtilisin from strains harboring pHI212, 214, and 215, no mature subtilisin was observed (data not shown). Therefore, these results indicate that the production of mature subtilisin in E. coli is absolutely dependent on subtilisin activity. Direct Fusion of OmpA Signal Peptide to Mature Subtilisin-In the plasmid pH1100 the OmpA signal peptide was directly fused to mature subtilisin except that there are 2 extra amino acid residues (Glu-Leu) between Ala-1 and Gln-2 as shown in Fig. 4C. Assays of lysates from induced cells harboring pH1100 did not show any detectable subtilisin activity, despite the fact that SDS-polyacrylamide gel electrophoresis of the induced cultures showed a large amount of a protein with a mobility close to that of subtilisin (Fig. 5,  8). This product accounted for as much as 10% of total cellular protein after a 2-h induction. Like the major products of pHI212,214,215, and 216, the pH1100 product was insoluble and was easily purified from a low-speed centrifuge pellet. Amino acid sequencing from the amino terminus to the 10th residue of this product by Edman degradation confirmed the processing of the OmpA signal peptide.
The lack of detectable activity in the pH1100 product could be due to the 2 extra amino acid residues inserted in the amino-terminal region of mature subtilisin. Therefore, these 2 residues were deleted by site-specific mutagenesis (see Fig.  4C), giving a fusion in which the OmpA signal sequence is fused directly to a sequence encoding authentic mature subtilisin. The cells harboring the new plasmid pHT700 again showed no detectable subtilisin activity, as was the case with pHI100. This is despite the fact that these cells produced an amount of authentic mature subtilisin equivalent to 10% of total cellular protein. The product was again insoluble, as was the pH1100 product, and migrated slightly faster than the pH1100 product, as expected from the fact that the pHT700 product is shorter by 2 amino acid residues (see Fig. 7). Thus, this product is considered to have the identical sequence as mature subtilisin.
These results indicate that the pHT700 product does not have the same three-dimensional structure as a native subtilisin E. Since pHI212,214, and 215 are able to produce active subtilisin, it appears that the pro-sequence is essential for folding of subtilisin into its native structure..
A similar result was obtained with plasmid pH1114 (see

FIG. 7. SDS-polyacrylamide gel electrophoresis of total cellular protein carrying pHT700 (lane I ) and pH1100 (lane 2).
Figs. 2 and 40). This plasmid is able to produce a polypeptide of which the amino-terminal 14 amino acid residues of the mature subtilisin are replaced with a hexapeptide, Gly-Ile-Asn-Ser-Lys-Leu, due to the plasmid construction procedure. No subtilisin activity was detected. The product was again produced in a large amount in an insoluble form and migrated a t 31 kDa as shown in Fig. 5, lane 9.
OmpA Subtilisin Fusion Proteins with Shortened Pro-sequences-In order to examine further the role of the prosequence for the production of active subtilisin, two plasmids, pH1102 and pHI103, were constructed, in which the OmpA signal peptide was fused to the -63rd (pHI102) and -33rd (pHI103) residue in the mature subtilisin sequence (see Fig.  2). The resulting product of pH1102 has the carboxyl-terminal 63 amino acid residues of the pro-sequence with 5 extra amino acids (Ala-Glu-Phe-Gln-Ala) at its amino terminus, as shown in Fig. 4 0. The product of pH1103 has the carboxyl-terminal 33 amino acids of the pro-sequence and 5 extra amino acids (Ala-Glu-Phe-Gln-Ala) at its amino terminus, also shown in Fig. 40. No subtilisin activity was detected from cells harboring either of these plasmids. SDS-polyacrylamide gel electrophoresis showed that pH1102 produced a major product of apparent molecular mass 41 kDa (Fig. 5, lane 6) when induced with IPTG. This product was found to be insoluble and could be pelleted by low-speed centrifugation. Besides this major product, a few new bands are observed as shown by dots.
However, none of these bands corresponds to mature subtilisin. All the other minor bands are probably derived by degradation of the major product. Similarly, pH1103 produced a major product of apparent mass 37 kDa (Fig. 5, lane 7). This product was also insoluble, and no band corresponding to mature subtilisin was seen. The differences of these apparent molecular masses from that of mature subtilisin on SDSpolyacrylamide gel electrophoresis were in good agreement with the actual molecular mass differences between mature subtilisin and the products after OmpA signal peptide cleav-age. Inductions of 23 "C with 0.005 mM IPTG gave the same results.

DISCUSSION
When the subtilisin signal peptide (pre-sequence) was replaced with the OmpA signal peptide, as in the OmpA-prosubtilisin fusions pHI212, 214, and 215, active mature subtilisin was found in the periplasmic space, as well as insoluble and inactive pro-subtilisin. The processing to the mature, enzymatically active subtilisin from OmpA pro-subtilisin can be either before or after the cleavage of OmpA signal peptide. However, since longer incubation did not increase the amount of active subtilisin, the pro-subtilisin which accumulated in large amount appeared to be defective for further processing to active subtilisin. It is interesting to note that lower temperatures (23 "C) and/or reduced concentrations (0.005 mM) of IPTG were found to be effective in increasing the production of active subtilisin with less accumulation of pro-subtilisin. This result suggests that a lower rate of synthesis and lower temperature may improve the processing of pro-subtilisin.
The activation of pro-subtilisin in E. coli occurred in the case of pHI212. By contrast, no band corresponding to the mature subtilisin was observed in the case of pH1216 having a mutation at the active site. This result suggests that at least initial activation of pro-subtilisin (cleavage of the pro-sequence) is likely to be an autocatalytic process.
The importance of the pro-sequence for the production of active subtilisin was clearly demonstrated by the product of pHT700 in which the OmpA signal peptide was directly fused to mature subtilisin by removing the pro-sequence of 77 residues. In this case, the product could not exhibit any detectable subtilisin activity, even though it has the identical sequence as mature subtilisin as a result of the cleavage of the OmpA signal sequence, indicating that the polypeptide failed to fold properly to an active enzyme, resulting in an insoluble protein. This result suggests that the pro-sequence functions in guiding the folding of pro-subtilisin molecule into the proper conformation necessary for activity.
The analysis of peptide sequences derived from DNA coding sequences (13) of subtilisins from different species of Bacilli has shown the existence of a highly homologous region in the carboxyl-terminal part of the pro-sequence, and a common sequence, Tyr-Ile-Val-Gly-Phe-Lys, in the amino-terminal region of the pro-sequence. In the case of subtilisin E, this sequence is found from residue -68 to residue -62 in the prosequence (see Fig. 4). These regions may be important for the production of active subtilisin. Furthermore, the predicted secondary structure and the hydrophobic residue distribution of the pro-sequence (13) show a high degree of similarity among all subtilisins, even in regions where their primary structures are quite different. This suggests the notion that the secondary structure of the pro-sequence plays an important role for the production of active subtilisin. In addition, it is interesting to note that there are exceptionally large numbers of charged residues in the pro-sequence (23 out of 77 in the case of subtilisin E). This unique feature may also be important for the function of the pro-sequence of subtilisin.
Restoring the carboxyl-terminal 33 amino acid residues of the pro-sequence (pHI103) did not allow the production of active subtilisin. Furthermore, restoring the carboxyl-terminal 63 amino acid residues (pHI102) was still not enough to produce active subtilisin. The latter construction lacks only 15 residues from the amino terminus of the pro-sequence including 5 residues from the common sequence discussed above. The results indicate that almost the entire pro-sequence is important for its function.
From the tertiary structure obtained from x-ray crystallography of subtilisin, one can find that charged residues are unevenly distributed on the surface of the molecule. Especially along the cleft in which the active site is located, only a few charged residues are found. On the other hand, the prosequence is structurally well conserved among the various subtilisins and is highly enriched with charged residues. Therefore, it is feasible to speculate that the pro-sequence interacts in a very specific manner with the mature portion of subtilisin, and this interaction is essential for folding of subtilisin into the enzymatically active conformation. In this regard, it is of great interest to purify pro-subtilisin and to study its crystallographic structure. Currently, purification of pro-subtilisin from pH1215 is in progress.