The Potential Active Site of the Lipoprotein-specific (Type II) Signal Peptidase of Bacillus subtilis *

Type II signal peptidases (SPase II) remove signal peptides from lipid-modified preproteins of eubacteria. As the catalytic mechanism employed by type II SPases was not known, the present studies were aimed at the identification of their potential active site residues. Comparison of the deduced amino acid sequences of 19 known type II SPases revealed the presence of five conserved domains. The importance of the 15 best conserved residues in these domains was investigated using the type II SPase of Bacillus subtilis, which, unlike SPase II of Escherichia coli, is not essential for viability. The results showed that only six residues are important for SPase II activity. These are Asp-14, Asn-99, Asp-102, Asn-126, Ala-128, and Asp-129. Only Asp-14 was required for stability of SPase II, indicating that the other five residues are required for catalysis, the active site geometry, or the specific recognition of lipid-modified preproteins. As Asp-102 and Asp-129 are the only residues invoked in the known catalytic mechanisms of proteases, we hypothesize that these two residues are directly involved in SPase II-mediated catalysis. This implies that type II SPases belong to a novel family of aspartic proteases.

Signal peptidases (SPases) 1 remove the targeting signals (i.e. signal peptides) from proteins that are translocated across the bacterial cytoplasmic membrane. This is a prerequisite for the release of the protein at the trans side of the membrane and, in some cases, the posttranslational modification of its amino terminus (for reviews, see Refs. [1][2][3][4]. Although the primary structure of signal peptides is poorly conserved, three functional domains have to be present: first, a positively charged amino terminus (N-region); second, a central hydrophobic domain (H-region); and third, a polar carboxyl-terminal domain (C-region), specifying the SPase cleavage site (1). We have previously shown that five paralogous type I SPases are involved in the processing of secretory precursor proteins in Bacillus subtilis (5)(6)(7). Two of these, denoted SipS and SipT, are of major importance for protein secretion. In this respect, B. subtilis is representative for Gram-positive eubacteria and archaea, many of which contain paralogous sip gene families (8). Considerable similarities can be observed between the known type I SPases when individual amino acid sequences are compared, including strictly conserved serine and lysine residues, which form a catalytic dyad (8 -11).
In contrast to the sip genes, B. subtilis and other eubacteria of which the genome has been sequenced completely contain only one gene for lipoprotein-specific (type II) SPases (12)(13)(14). As estimated from published genome sequences, lipoprotein precursors, which are the substrates of these enzymes, represent about 1-3.5% of the known eubacterial proteomes (14). The major difference between signal peptides of lipoproteins and those of secretory proteins is the presence of a well conserved "lipobox" of four residues in the C-region of lipoprotein signal peptides (3,15). Invariably, the carboxyl-terminal residue of the lipobox is cysteine, which, upon lipid modification, forms the signal for the retention of the mature lipoprotein at the membrane-cell wall interface of Gram-positive eubacteria, or the inner and outer membranes of Gram-negative eubacteria (16,17). Modification of this cysteine residue by the diacylglyceryl transferase (Lgt) is a prerequisite for processing of the lipoprotein precursor by SPase II. In Escherichia coli, mature (apo-)lipoproteins are further modified by amino-fatty acylation of the diacylglyceryl-cysteine amino group (17,18). The latter modification is probably not conserved in all eubacteria, as B. subtilis and Mycoplasma genitalium lack an lnt gene for the lipoprotein aminoacyltransferase (14).
Lipoprotein processing by SPase II is essential for cell viability of E. coli and other Gram-negative eubacteria (19,20). In contrast, the SPase II of B. subtilis is not essential for viability, although the activity of several lipoproteins seems to be strongly impaired in the absence of SPase II (14,21). The latter applies, for example, to the PrsA protein (14), which is required for the folding of translocated secretory proteins (22). Consequently, the secretion of ␣-amylase, a nonlipoprotein, was strongly impaired in cells lacking SPase II (14).
In contrast to the eubacterial type I SPases, very little is known about the mechanism that type II SPases employ for catalysis (17). In the present studies, which were aimed at the identification of potential active site residues of type II SPases, we made use of the fact that SPase II is not essential for viability of B. subtilis. As a first approach, all residues of the SPase II of B. subtilis that are conserved in the 19 known eubacterial type II SPases were mutated. The results showed that two strictly conserved aspartic acid residues are essential for the activity, but not the stability, of this enzyme, indicating that type II SPases employ an aspartic acid catalytic dyad for signal peptide cleavage of lipid-modified proteins, similar to aspartic proteases of the pepsin family. A third conserved as-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
partic acid residue appears to be required for the stability of the SPase II of B. subtilis. Table I lists the plasmids and bacterial strains used. Tryptone/yeast extract medium contained Bacto tryptone (1%), Bacto yeast extract (0.5%), and NaCl (1%). S7 media 1 and 3, used for labeling of B. subtilis proteins with [ 35 S]methionine (Amersham Pharmacia Biotech), were prepared as described in Refs. 23 and 24. When required, medium for E. coli was supplemented with kanamycin (20 g/ml) or ampicillin (40 g/ml); media for B. subtilis were supplemented with kanamycin (10 g/ml) or Em (1 g/ml).

Plasmids, Bacterial Strains, and Media-
DNA Techniques-Procedures for DNA purification, restriction, ligation, agarose gel electrophoresis, and transformation of E. coli were carried out as described in Ref. 29. Enzymes were from Roche Molecular Biochemicals. B. subtilis was transformed as described in Ref. 8. PCR was carried out with pWO DNA polymerase (Roche Molecular Biochemicals) as described in Ref. 10. The BLAST algorithm (30) was used for protein comparisons in GenBank TM . To construct B. subtilis 8G5 lsp, a genomic DNA fragment lacking the complete lsp gene (Fig. 1, deletion 1111-base pair BclI fragment) was cloned in the chromosomal integration plasmid pORI280 (25) resulting in pINT11d. This plasmid carries, in addition to the mutant copy of the ileS-pyrR locus, an Em r marker and the E. coli lacZ gene. Upon transformation of B. subtilis 8G5 with pINT11d, Em r and blue transformants were selected on plates with X-gal. These were obtained as a result of a Campbell-type integration of pINT11d (single cross-over recombination) into the homologous ileS-pyrR sequences on the chromosome. After growth for about 200 generations in the absence of Em, cells were selected (Em s and white on plates with X-gal) that had excised the integrated plasmid from the chromosome. As verified by PCR and Southern blotting, some of these, denoted B. subtilis 8G5 lsp, lacked the 1111-base pair BclI fragment from the ileS-pyrR locus. Note that the latter fragment also contains the ylyA gene and part of the ylyB gene, which are not involved in lipoprotein processing by SPase II, as shown by complementation studies with pGDL150 (see below).
To construct the plasmids pGDL150 (specifying wild-type SPase II), pGDL151 (specifying SPase II ⌬C), and pGDL152 (specifying SPase II-Myc), a PCR with primers Lsp-3 and primers Lsp-4, L-dC, or Lsp-6 (Table II) was performed, using chromosomal B. subtilis 168 DNA as a template. Amplified fragments were subsequently cleaved with SalI and EcoRI and ligated into the corresponding sites of pGDL48 (26). Consequently, the wild-type or mutant lsp genes on pGDL150 -152 were transcribed from the constitutive promoter of the Em r gene present on pGDL48. Site-directed mutations were introduced into plasmidborne copies of lsp-myc by a two-step PCR approach (10), using primers Lsp-3 and Lsp-6 in combination with mutagenic oligonucleotides (Table  II). Amplified fragments were cleaved with SalI and EcoRI and ligated into the corresponding sites of pGDL48. The resulting plasmids were named pL-x, where x indicates the position and type of amino acid substitution in the corresponding SPase II-Myc mutant proteins.
Western Blot Analysis-Western blotting was performed as described in Ref. 30. After separation by SDS-PAGE, proteins were transferred to Immobilon polyvinylidene difluoride membranes (Millipore Corporation). To detect PrsA or carboxyl-terminally Myc-tagged SPase II, B. subtilis cells were separated from the growth medium, and samples for SDS-PAGE were prepared as described previously (23). For the separation of samples with Myc-tagged SPase II, 1 mM dithiothreitol, and 1% Triton X-100 were added to the loading buffer for SDS-PAGE as described in Ref. 31. The PrsA protein was visualized with specific antibodies and horseradish peroxidase-anti-rabbit-IgG conjugates (Amersham Pharmacia Biotech); carboxyl-terminally Myc-tagged SPase II was visualized with monoclonal anti c-Myc antibodies (Roche Molecular Biochemicals) and horseradish peroxidase-anti-mouse-IgG conjugates.

RESULTS
Conserved Domains in Type II SPases-Thus far, the nucleotide sequences of 19 lsp genes for SPase II are known, allowing the detailed comparison of the deduced amino acid sequences of the corresponding proteins ( Fig. 2). As demonstrated for the SPase II of E. coli (50), all these type II SPases have four predicted transmembrane (TM) domains (denoted TM-A to -D) (Fig. 3). Two periplasmic (Gram-negative eubacteria) or cell wall-exposed (Gram-positive eubacteria) regions are localized between the TM-A and TM-B regions and between the TM-C and TM-D domains, respectively (Fig. 3). Furthermore, five highly conserved domains (I-V) were detected in these SPases (Figs. 2 and 3): Domain I, containing no strictly conserved residues, is located in TM-A; domain II, containing the strictly conserved residues Asn-45 and Gly-47, is located in the extracytoplasmic region between TM-A and -B; domain III, containing the strictly conserved residues Asn-99 and Asp-102, is located at the junction of TM-C and the extracytoplasmic region between TM-C and -D; domain IV, containing the strictly conserved residues Val-109 and Asp-111, is located in the extracytoplasmic region between TM-C and -D; and domain V, containing the strictly conserved residues Phe-125, Asn-126, Ala-128 and Asp-129, is located at the junction of TM-D and the extracytoplasmic region between TM-C and -D (Fig. 3).
A Carboxyl-terminally Myc-tagged SPase II Has Wild-type Activity-Positively charged residues in the cytosolic aminoand carboxyl-terminal regions of type II SPases have been invoked in catalysis (3). Notably, the SPase II of B. subtilis lacks an amino-terminal cytosolic region with positively charged residues, refuting that such a region could be required for catalysis (12). Like other known type II SPases, the B. subtilis SPase II does, however, contain positively charged carboxyl-terminal residues (four Lys residues). To determine whether the positively charged residues in the carboxyl terminus of the B. subtilis SPase II are important for catalysis, a mutant lsp gene encoding SPase II ⌬C (lacking the six carboxyl-terminal residues KKKKEQ), was constructed by PCR and cloned into plasmid pGDL48 (Table III). To obtain a positive control for SPase II activity, the wild-type lsp gene was also amplified by PCR and cloned into pGDL48. The resulting plasmids, denoted pGDL150 (SPase II) and pGDL151 (SPase II ⌬C), were used to transform B. subtilis 8G5 lsp, lacking the lsp gene. Next, pre-PrsA processing to the mature form, which required SPase II activity, was analyzed by pulse-labeling with [ 35 S]methionine for 90 s. As shown in Fig. 4A, almost all labeled PrsA was processed to the mature form in cells producing SPase II ⌬C, similar to cells producing the wild-type SPase II (Fig. 4A). These findings demonstrated that the positively charged carboxyl-terminal residues of the B. subtilis SPase II are not required for catalysis. This suggested that the carboxyl terminus of SPase II could be tagged with a Myc epitope for the immunological detection of this enzyme, without inhibition of catalytic activity. The latter idea was tested by the construction of plasmid pGDL152, which specifies the SPase II-Myc protein.
SPase II-Myc contains an extension of nine residues with the sequence KLISEEDLN, which, together with the two authentic carboxyl-terminal residues (EQ) of SPase II, forms the Myc epitope. As shown by pulse labeling, Myc-tagged SPase II could replace wild-type SPase II, as pre-PrsA was efficiently pro-cessed to the mature form (Fig. 4A). Furthermore, the presence of SPase II-Myc in these cells could be demonstrated by Western blotting using Myc-specific antibodies (Fig. 4B).
Mapping of Functionally Important Conserved Residues of SPase II-Functionally important residues of SPase II were mapped by site-specific mutagenesis. To this purpose, SPase II-Myc was used because the integrity of mutant proteins can be verified with Myc-specific antibodies. Residues present in at least 17 of the 19 known type II SPases were replaced by alanine. The latter residue was chosen because it is small and it has a chemically inert side chain, minimizing conformational strain and indirect effects on catalysis. The strictly conserved residue Ala-128 was replaced by valine, which has a more bulky side chain. Furthermore, Trp-50 was replaced by alanine, because all known type II SPases contain a residue with an aromatic side-chain at this position (tryptophan or phenylalanine) (Fig. 2).
To monitor the activity of SPase II mutant proteins specified by plasmids (pL-x) ( Table I), processing of pre-PrsA to the mature form was studied in pulse labeling experiments with B. subtilis 8G5 lsp. As shown in Fig. 4A, processing of pre-PrsA was not affected in cells producing the SPase II mutant proteins G47A, F50A, G95A, and F125A, showing that these residues are not required for activity. Processing of pre-PrsA was mildly affected in cells producing the SPase II mutant proteins K18A, R103A, V109A, indicating that these residues are of minor importance for catalysis. Notably, pre-PrsA was not processed, or was processed very inefficiently, in cells producing FIG. 1. Construction of B. subtilis 8G5 lsp. Schematic presentation of the ileS-pyrR region of B. subtilis 8G5 (parental strain) and B. subtilis 8G5 lsp. B. subtilis 8G5 lsp lacks a 1111-base pair BclI fragment containing the complete ylyA and lsp genes and the 5Ј sequences of the ylyB gene. The latter strain was obtained as a result of a Campbell-type integration of pINT11d, containing a mutant copy of the ileS-pyrR region, into the homologous ileS-pyrR sequences on the chromosome of B. subtilis 8G5. After growth in the absence of antibiotics, cells were selected that had excised the integrated plasmid from the chromosome and lacked the 1111-base pair fragment (see under "Experimental Procedures"). The restriction sites relevant for the construction are shown (Bc, BclI; Bg, BglII; Nd, NdhI); ЈylyB, 5Ј truncated ylyB gene. To investigate which of the residues that are important for SPase II activity are determinants for the stability of the enzyme, Western blotting experiments were performed. As shown in Fig. 4B, only Asp-14 was essential for the stability of SPase II-Myc, whereas all other mutant proteins were detectable.
The Conserved Domains III and V Contain Critical Residues for SPase II Activity-To verify the (in-)activity of the SPase II mutants with strongly reduced activities, pulse-chase labeling experiments were performed with B. subtilis 8G5 lsp. In these experiments, cells were chased with an excess of nonradioactive methionine for 10 min, as it was previously shown that within this period of time no pre-PrsA is converted to the mature form in cells lacking SPase II (14). As shown in Fig. 5, no labeled pre-PrsA was converted to the mature form in cells producing the SPase II mutant proteins D14A, D102A, A128V, and D129A, suggesting that these proteins are inactive. In contrast, low amounts of labeled mature PrsA were detectable in cells producing the N99A and N126A mutant proteins, and significant levels of pre-PrsA processing were detected in cells producing the N45A and D111A mutant proteins.
To determine the effects of the mutations in conserved residues of SPase II at steady state, the accumulation of pre-PrsA in B. subtilis 8G5 lsp was analyzed by Western blotting. It has to be noted that cells lacking SPase II display alternative processing of pre-PrsA to mature-like forms that, on SDS-PAGE, migrate at a slightly reduced rate compared with mature PrsA (14). As shown in Fig. 6, only the cells producing the SPase II mutant proteins D14A, N99A, D102A, N126A, A128V, and D129A accumulated precursor and mature-like forms of PrsA.   Taken together, our findings show that residues Asn-99, Asp-102, Asn-126, Ala-128, and Asp-129 are critical for SPase II activity and that Asp-14 is critical for SPase II stability. DISCUSSION In the present paper, we document the mapping of six functionally important residues of SPase II of B. subtilis. These are Asp-14, Asn-99, Asp-102, Asn-126, Ala-128, and Asp-129. All of these residues are predicted to be localized close to the external surface of the cytoplasmic membrane. Only one residue, Asp-14, was required for the stability of the enzyme, showing that it is an important structural determinant. This view is supported by the fact that the replacement of the equivalent Asp residue in the SPase II of E. coli (Asp-23) by glycine merely resulted in temperature sensitivity of the enzyme (3). In addition, Asp-14 of the B. subtilis SPase II is not conserved in the type II SPases of M. genitalium and M. pneumoniae, which have been shown to contain active type II SPases (52,53). In contrast, mutation of the other five residues required for activity of the B. subtilis SPase II did not significantly affect the stability of this enzyme, showing that these residues are directly or indirectly required for catalysis. Interestingly, unlike the SPase II of E. coli (38), the B. subtilis SPase II did not require positively charged residues at the carboxyl terminus for activity, which implies that these residues are structural, rather than catalytic, determinants for the E. coli SPase II.
The observation that SPase II lacks conserved serine residues and that the only conserved lysine residue is not required for activity rules out the possibility that type I and type II SPases make use of similar catalytic mechanisms. Furthermore, the lack of conserved cysteine and histidine residues and the previous finding that purified SPase II of E. coli was active in the absence of metal ions (32) demonstrate that SPase II does not employ the well defined catalytic mechanisms of thiolor metalloproteases. Consequently, the present observation that two strictly conserved aspartic acid residues are essential for SPase II activity indicates that this enzyme belongs to the aspartic proteases. This hypothesis is supported by the obser- vation that the SPase II of E. coli could be inhibited by pepstatin, a known inhibitor of aspartic proteases (32).
Aspartic proteases are a group of proteolytic enzymes of the pepsin family that share the same catalytic mechanism and usually function in acidic environments (54 -56). The known aspartic proteases of eukaryotes are monomeric enzymes, which consist of two subdomains, both containing the conserved sequence Asp-Thr-Gly. All three residues contribute to the active site (57,58). Furthermore, conserved hydrogen bonds between the catalytic Asp residues and conserved Ser or Thr residues that are located at the ϩ3 position relative to the catalytic Asp are present in most pepsin-like aspartic proteases. These hydrogen bonds are, most likely, responsible for the low pK a values of the active-site aspartic acid residues (59,60). Aspartic proteases from retroviruses and some plant viruses are related to the eukaryotic aspartic proteases, but the active protease is a homodimer, the active site(s) of which contain the conserved Asp-Thr/Ser-Gly motif (61,62). In contrast to the eukaryotic aspartic proteases, conserved Ser or Thr residues are absent from the ϩ3 position relative to the active site Asp residue of retroviral aspartic proteases. Consequently, the active site Asp residues of retroviral aspartic proteases lack the conserved hydrogen bonding, which explains why these enzymes have a much higher optimum pH than pepsin-like proteases (63). Notably, type II SPases lack the conserved Asp-Thr/Ser-Gly motif of previously described (eukaryotic and viral) aspartic proteases. Moreover, like in the viral aspartic proteases, conserved Ser or Thr residues are absent from the ϩ3 position relative to the putative active site Asp residues. Instead, type II SPases contain strictly conserved Asn residues at the Ϫ3 position (Fig. 2), which are very important for activity. These observations imply that the type II SPases belong to a novel class of aspartic proteases. As no type II SPases or otherwise related proteins have been identified in archaea or eukaryotes, it seems that this novel class of aspartic proteases has evolved exclusively in eubacteria.
The present observations suggest that SPase II of B. subtilis employs Asp-102 and Asp-129 for catalysis. By analogy to the known catalytic mechanism of aspartic proteases, this implies that Asp-102 and Asp-129, and their equivalents in other type II SPases, form a catalytic dyad. It remains unclear whether the pK a of these residues is reduced by hydrogen bonding, as described for eukaryotic aspartic proteases. If such hydrogen bonds do not exist in type II SPases, this would explain the high optimum pH (7.9) of the SPase II of E. coli (39). The absence of conserved Ser/Thr residues does not, however, exclude the possibility that the pK a of active site aspartic acid residues is modulated by other residues. In fact, this could be one possible role of other residues (i.e. Asn-45, Asn-99, Asp-111, Asn-126, and Ala-128) required for the activity of SPase II of B. subtilis. Alternatively, the latter residues could be required for the geometry of the active site of type II SPases, or the specific recognition of the diacyl-glyceryl-modified cysteine residues in the lipobox of preproteins. Based on the catalytic mechanism of the aspartic protease of the human immunodeficiency virus 1 (64,65), which is required for viral replication (66), we propose the following mechanism for type II SPases. At the start of catalysis, the active site contains a so-called "lytic" water molecule, and only one of the active site aspartic acid residues is protonated (Fig. 7A). Upon binding of a lipid-modified precursor, the carbonyl carbon of the scissile peptide bond is hydrated (Fig. 7B), resulting in a tetrahedral intermediate (Fig. 7C). During this event, a proton is transferred (via the lytic water molecule) from one active site aspartic acid residue to the other (Fig. 7, C and D). Next, one hydroxyl group of the tetrahedral intermediate donates a proton to the charged aspartic acid residue, and simultaneously, the nitrogen atom at the scissile peptide bond accepts a proton from the other catalytic aspartic acid residue. The latter event results in peptide bond cleavage (processing), regeneration of the catalytic site of SPase II, and release of the mature lipoprotein and the cleaved signal peptide from the enzyme (Fig. 7D). This model is particularly attractive, because both essential aspartic acid residues are predicted to be located in close proximity to the extracytoplasmic surface of the membrane, similar to the active site serine residue of type I SPases (9,10). This is the place where C-regions of exported precursors are likely to emerge from the translocation apparatus.