Autotransporters constitute a relatively simple secretion system in Gram-negative bacteria, depending for their translocation across the outer membrane only on a C-terminal translocator domain. We have studied a novel autotransporter serine protease, designated NalP, from Neisseria meningitidis strain H44/76, featuring a lipoprotein motif at the signal sequence cleavage site. Indeed, lipidation of NalP could be demonstrated, but the secreted 70 kDa domain of NalP lacked the lipid-moiety as a result of additional N-terminal processing. A nalP mutant showed a drastically altered profile of secreted proteins. Mass-spectrometric analysis of tryptic fragments identified the autotransporters IgA protease and App, a homologue of the adhesin Hap of Haemophilus influenzae, as the major secreted proteins. Two forms of both of these proteins were found in the culture supernatant of the wild-type strain, whereas only the lower molecular-weight forms predominated in the culture supernatant of the nalP mutant. The serine-protease active site of NalP was required for the modulation of the processing of these autotransporters. We propose that, apart from the autoproteolytic processing, NalP can process App and IgA protease and hypothesize that this function of NalP could contribute to the virulence of the organism.
Proteins secreted by Gram-negative bacteria are transported across the cell envelope, which is composed of the inner membrane, the peptidoglycan-containing periplasm and the outer membrane. Various elaborate protein secretion systems have evolved, which span the cell envelope and deliver proteins into the extracellular medium (Koster et al., 2000; Thanassi and Hultgren, 2000). Compared to these multicomponent systems, the autotransporter secretion mechanism is relatively simple (Henderson et al., 1998; Henderson and Nataro, 2001). Autotransporters are modular proteins consisting of an N-terminal signal sequence, a C-terminal translocator domain and, in between, a secreted passenger domain. The signal sequence directs the transport of the proteins over the inner membrane, presumably via the Sec machinery. The C-terminal domain is thought to form a β-barrel in the outer membrane, constituting a pore through which the passenger domain is transported to the bacterial cell surface (Maurer et al., 1999). However, a purified translocator domain was recently shown to form a multimeric ring-like structure with a central pore through which the passenger domain could possibly be translocated (Veiga et al., 2002). Irrespective of the secretion mechanism, there seems to be considerable flexibility in the type of passenger domains that can be translocated, as several very different types of foreign polypeptides could be transported to the cell surface, when fused to a C-terminal autotransporter translocator domain (Klauser et al., 1990; Maurer et al., 1997; Valls et al., 2000). Native passenger domains of autotransporters form a large and functionally diverse group of secreted proteins, many of them involved in pathogenesis (Henderson and Nataro, 2001). The first autotransporter studied in detail was the IgA1 protease of Neisseria gonorrhoeae (Pohlner et al., 1987). Other examples are the adhesins Aida-I of Escherichia coli (Benz and Schmidt, 1992) and Hap and Hia of Haemophilus influenzae (St. Geme et al., 1994; Hendrixson et al., 1997), the cytotoxin VacA of Helicobacter pylori (Schmitt and Haas, 1994), the IcsA/VirG protein of Shigella flexneri, which polymerizes actin (Suzuki et al., 1995), the serum-resistance protein BrkA of Bordetella pertussis (Fernandez and Weiss, 1994) and the esterase EstA of Pseudomonas aeruginosa (Wilhelm et al., 1999).
Neisseria meningitidis is a major cause of meningitis and septicaemia worldwide (Peltola, 1983; Tzeng and Stephens, 2000). A systematic search of the sequenced genomes of N. meningitidis strains MC58 and Z2491 recently led to the identification of eight open reading frames (ORFs) putatively encoding autotransporters (van Ulsen et al., 2001). Among them was the iga gene, encoding the IgA1 protease, whereas the others encode proteins with homology to various adhesins, a haemagglutinin and a serum-resistance protein. Therefore, these proteins might be involved in the virulence of N. meningitidis. Consistently, the presence of antibodies directed against these proteins in patient sera indicated the in vivo expression of at least four of these ORFs (Brooks et al., 1992; Ait-Tahar et al., 2000; Abdel-Hadi et al., 2001; van Ulsen et al., 2001). Here, the function of one of the meningococcal autotransporters, which we designated NalP, was evaluated. The cloning of this autotransporter gene, then designated aspA, was recently described (Turner et al., 2002). The authors showed that NalP/AspA is a serine protease and that the serine residue of the active site is involved in autocatalytic processing, resulting in secretion of the passenger domain. In the current paper, we show that NalP is lipidated during its biogenesis. To identify its function, a nalP mutant derivative of N. meningitidis strain H44/76 was constructed. This mutant appeared to be affected in the processing of the autotransporters App and IgA protease, resulting in altered secretion profiles of the respective passenger domains.
Sequence analysis of the NalP autotransporter
A systematic search of the published genome sequences of two N. meningitidis strains, the serogroup B strain MC58 (Tettelin et al., 2000) and the serogroup A strain Z2491 (Parkhill et al., 2000), yielded eight candidate ORFs that could encode autotransporters (van Ulsen et al., 2001). Analysis of the amino acid sequences encoded by the homologous ORFs NMB1969 and NMA0478 of MC58 and Z2491, respectively, revealed a protein that contained the prokaryotic lipoprotein motif LSA↓C (Sankaran and Wu, 1994) at the end of the signal sequence, suggesting that the mature protein is lipidated at its N-terminal cysteine. The protein was therefore tentatively designated NalP (Neisseria autotransporter lipoprotein) and the nalP gene was cloned from N. meningitidis serogroup B strain H44/76 and sequenced (GenBank acc. no. AY150284; see Supplementary material for a more detailed sequence analysis). We identified NalP originally in the genome sequences (van Ulsen et al., 2001) by its homology to the secreted serine protease (PrtS) of Serratia marcescens (Yanagida et al., 1986), but that protein does not contain a lipoprotein motif.
Expression of nalP
The H44/76 nalP gene was cloned in expression vector pET11a, yielding plasmid pPU300. The plasmid contains the gene under the control of an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible T7 promoter, and nalP was, subsequently, expressed in E. coli BL21(DE3). A major overproduced gene product with an apparent molecular weight (Mr) of 114 000 was found in E. coli. This protein was isolated and used to elicit a polyclonal antiserum. These antibodies recognized the recombinant 114 kDa band as well as several degradation products in whole cell lysates of pPU300-containing E. coli BL21(DE3) (Fig. 1A). Furthermore, a specific band with an Mr of 70 000 was detected in the culture supernatant, probably representing the secreted passenger domain of NalP. The antiserum recognized a similar 70 kDa protein in the culture supernatants of N. meningitidis strain H44/76 and its unencapsulated variant HB-1 (Fig. 1A), whereas the full-length 114 kDa form was hardly (or not at all) detected in whole cell lysates. Apparently, the autotransporter secretion pathway of NalP operated inefficiently in E. coli, possibly because of the high production level.
Analysis of natural N. meningitidis isolates by PCR revealed the presence of nalP in all seven isolates tested (results not shown). In four out of the seven isolates, NalP was detected in the medium (Fig. 1B). In three of these isolates, a band with slightly higher Mr was detected in addition to the 70 kDa band, suggestive of differential processing at the cell surface. However, three out of the seven N. meningitidis isolates did not produce detectable levels of NalP (Fig. 1B). Consistently, sequencing of PCR products revealed a phase-off number of cytosines in the cytosine repeat of nalP in these isolates (results not shown), consistent with the notion that nalP expression is prone to phase-variation (Turner et al., 2002).
Lipidation of the NalP protein
Lipidation is an exceptional feature among autotransporters, and we were therefore interested to verify lipidation of NalP. First, we established the N-terminal sequence of the 70 kDa secreted NalP domain. The recombinant 70 kDa protein in the supernatant of E. coli was concentrated, purified by SDS-PAGE and submitted to N-terminal sequencing. The resulting N-terminal five-residue sequence was AGIKN, demonstrating that the secreted form of NalP starts at A65 (see Supplementary material, Fig. S1). Apparently, apart from the cleavage of the 27-residue signal sequence and the autoproteolytic step required to release the passenger domain from the C-terminal translocator domain (Turner et al., 2002), an additional proteolytic cleavage takes place in the N-terminal part of the mature protein.
To investigate whether the lipoprotein motif is recognized and processed by the modifying enzymes (Sankaran and Wu, 1994), we performed 3H-palmitate labelling experiments. Attempts to label wild-type NalP in N. meningitidis or in E. coli containing pPU300 failed, probably because the secreted 70 kDa form lacks the N-terminal peptide that includes the putative lipid moiety, as was revealed by N-terminal sequencing. Apparently, the low abundance of NalP in N. meningitidis complicated detection, whereas in E. coli most of the protein was produced in cytoplasmic inclusion bodies and not entering the autotransporter secretion pathway. Therefore, we first constructed plasmid pPU311, containing a chimeric ORF encoding the signal sequence and the following 17 residues of NalP (M1-G44) fused to the mature part of β-lactamase under the control of the T7 promoter. The fusion protein was produced in E. coli BL21(DE3) as revealed by Western-blot analysis with antibodies recognizing β-lactamase (Fig. 2A). Moreover, when the cells were incubated with 3H-palmitate, the fusion protein was clearly labelled (Fig. 2B), demonstrating that the lipoprotein motif is recognized and processed by the modifying enzymes.
To verify that also full-length NalP is lipidated, we then constructed the pET11a-derived expression plasmid pPU305, encoding a mutant form of NalP, in which the serine residue at position 427 (Turner et al., 2002) was substituted by alanine (NalP-S427A). As expected, the processed 70 kDa NalP product was not observed in culture supernatants of BL21(DE3) containing pPU305 (results not shown). Moreover, upon incubation with 3H-palmitate, a labelled band with a Mr of 110 kDa corresponding to the full-length NalP-S427A was detected (Fig. 2C).
Finally, we addressed lipidation of NalP in N. meningitidis, by testing whether globomycin, an antibiotic that specifically inhibits the lipoprotein specific signal peptidase Lsp (Inukai et al., 1978), influenced nalP expression in N. meningitidis. Indeed, in the presence of sublethal amounts of globomycin, the levels of NalP detected in the medium decreased (Fig. 2D), whereas levels of non-lipidated secreted proteins were not affected (results not shown, but see also Fig. 4C). This gave a clear indication that full-length NalP is lipidated during NalP biogenesis in N. meningitidis. However, this lipidated form is an intermediate in the secretion process, as the secreted 70 kDa form of NalP is not lipidated, as a result of further N-terminal processing.
Altered secretion profile in a nalP mutant
To gain insight in the function of NalP, nalP mutant-derivatives of strain H44/76 and of its unencapsulated variant HB-1 were constructed. A kanamycin-resistance cassette was cloned into the nalP ORF, and the resulting construct was used to transform the Neisseria strains. Inactivation of nalP, resulting from homologous recombination between the nalP::kan construct and the chromosomal nalP, was confirmed by PCR-analysis of kanamycin-resistant transformants. As expected, the mutant strains, H44/76 nalP::kan and HB-1nalP::kan, no longer produced NalP (Fig. 1A). The protein profiles of whole cell lysates of these mutants were not significantly altered (results not shown), but the profile of secreted proteins showed striking differences (Fig. 3), suggesting a role for NalP in protein secretion.
The identity of the major proteins in the culture supernatants was determined by N-terminal sequencing and mass-spectrometric identification of tryptic fragments (Fig. 3). Two forms of IgA protease and two forms of the homologue of the H. influenzae adhesin Hap, which has been designated App in N. meningitidis (Abdel-Hadi et al., 2001; van Ulsen et al., 2001), were identified. N-terminal sequencing of both secreted forms of IgA protease, which have Mrs of 160 000 and 110 000, respectively, yielded the sequence ALVRDDV, which corresponds to the sequence immediately next to the signal-peptidase cleavage site. The two secreted forms of App, with Mrs of 140 000 and 100 000, respectively, could only be identified by mass-spectrometric analysis of tryptic fragments of these bands, which revealed peptide sequences exactly matching the amino acid sequence of App of H44/76 (Fig. 3). Moreover, the results of the mass-spectrometric analysis suggested that the smaller forms of both proteins derived from proteolytic cleavage at the C terminus. The passenger domain of IgA protease of N. gonorrhoeae is a multidomain protein, of which the protease domain and the α peptide appear separately in the medium (Pohlner et al., 1987). In contrast to the 110 kDa secreted form, the 160 kDa form of IgA protease contained peptides corresponding to the α peptide (Fig. 3). Apparently, the IgA protease appears in the culture supernatant of the wild-type strain both with and without the α peptide attached. Similarly, the 140 kDa and the 100 kDa forms of App corresponded to the presence and absence, respectively, of a C-terminally located polypeptide (Fig. 3).
Influence of NalP on the processing of App
The identity of the secreted forms of App was confirmed in Western blots of HB-1 and its app::kan and nalP::kan mutant derivatives. These showed the absence of bands with Mrs of 140 and 100 kDa in the culture supernatants of the app::kan mutant (Fig. 4A). The ratio between the 140 kDa and 100 kDa forms of App was changed in the nalP::kan mutant strain when compared to HB-1, with the 140 kDa form markedly decreased and the 100 kDaA form increased in intensity. Furthermore, a new form of App with an Mr of 160 000, which could correspond to full-length App was detected in the whole cell lysates of the nalP mutant. These results demonstrated that NalP modulates the processing of App at the cell surface.
To test whether the influence of NalP on the processing of App was also detectable in natural isolates, either producing NalP or not because of phase variation (Fig. 1B), additional Western-blotting experiments were performed. The secretion profile of App in these isolates correlated perfectly with the presence or absence of NalP (Fig. 4B), as the 100 kDa form of App was most abundant in the culture supernatants of isolates not expressing NalP, whereas the 140 kDa form predominated in those isolates that expressed NalP. Apparently, NalP influences in N. meningitidis what form of the App autotransporter is secreted. In line with these results, Western-blot analysis showed that incubation of HB-1 with globomycin, which resulted in decreased amounts of NalP (Fig. 2D), also resulted in an increase of the 100 kDa form and a decrease of the 140 kDa form of secreted App (Fig. 4C). As expected, only the 100 kDa form of App was detected in the nalP disruption mutant, and incubation with globomycin did affect the amounts of this form of App detected (Fig. 4C).
Influence of NalP on the processing of IgA protease
The identity of the secreted forms of IgA protease was confirmed in Western blots of HB-1 and its iga::kan and nalP::kan mutant derivatives (Fig. 5). The blot showed the absence of the 160 kDa and 110 kDa forms of IgA protease in the culture supernatant of the iga::kan mutant (Fig. 5A) and a decreased ratio between the 160 kDa and 110 kDa forms of IgA protease in the nalP::kan mutant, when compared to HB-1. However, when the Western-blot analysis was extended to investigate the influence of NalP expression on IgA protease processing in the natural isolates (Fig. 5B), it appeared that NalP did not generally influence the profile of secreted IgA protease. In the culture supernatants of most isolates the 110 kDa form of IgA protease, which lacks the α peptide, was detected, regardless whether they expressed NalP or not. Only in the case of strain M986, which does express NalP, a 160 kDa form of IgA protease was detected in the medium. Apparently, the effect of NalP on IgA protease secretion is less general as observed for App.
The NalP active-site serine is required for the modulation of autotransporter processing
A possible explanation for the effect of NalP on the processing of App and IgA protease would be that NalP proteolytically cleaves these proteins at sites that differ from the autoproteolytic processing sites. This hypothesis would imply that the proteolytic activity of NalP is required for the alternative processing of App and IgA protease. To test this possibility, we used the neisserial replicative vector pFP10 (Pagotto et al., 2000) to construct plasmids pEN300 and pEN305 encoding wild-type NalP and the active-site mutant NalP-S427A, respectively, under the control of an IPTG-inducible lac promoter. The processing of App and IgA protease was evaluated in strain HB-1nalP::kan carrying PEN300 and pEN305. Production of wild-type NalP from plasmid pEN300 in this strain resulted in the complementation of the nalP-mutant phenotype, as appeared from the increased amounts of the 140 kDa form of App (Fig. 6A) and the 160 kDa form of IgA protease (Fig. 6B) in the culture supernatant. Therefore, the altered processing of App and IgA protease in the nalP::kan mutant is a direct effect of the nalP disruption and not a result of a polar effect of insertion of the kanamycin-resistance cassette on downstream genes, or to any other secondary mutation. In contrast, production of NalP-S427A from plasmid pEN305 did not restore the wild-type secretion profile in the nalP::kan mutant (Fig. 6A and B). Therefore, the active site of NalP is required for its role in modulating the processing of App and IgA protease.
Western-blot analysis was performed to verify the production of NalP-S427A (Fig. 6C). As expected, expression from pEN305 led to accumulation of full-length NalP-S427A in the whole cell lysates. Surprisingly, however, also a smaller NalP fragment was detected in the whole cell lysate as well as in the culture supernatant (Fig. 6C). This NalP fragment, with a Mr of about 76 kDa, was somewhat larger than the 70 kDa secreted domain found in the culture supernatant of wild-type HB-1. Apparently, when expressed in sufficient amounts, the mutant NalP protein can be processed by other proteases at the cell surface. To evaluate a possible effect of App or IgA protease on the processing of wild-type NalP in HB-1, the presence of the 70 kDa secreted form of NalP in the culture supernatants of HB-1 and of its iga-, and app-mutant derivatives was compared (Fig. 6D). However, no major effects on the expression levels or appearance of NalP were detected, suggesting that in HB-1, the autoproteolytic cleavage of NalP predominates over possible processing by other proteases.
Genome analysis of N. meningitidis led to the identification of eight putative autotransporters (van Ulsen et al., 2001). The product of one of these contains a lipoprotein motif and was therefore designated NalP. Recently, Turner et al. (2002) reported the cloning of nalP, for which they coined the name AspA (for Autotransported serine protease A). However, despite the earlier report, we favour the acronym NalP for two reasons: (i) it is descriptive for a unique feature of the protein, and (ii) the acronym AspA is already in use for the enzyme aspartate ammonia-lyase in N. meningitidis as well as in other bacteria and can therefore not be used for this totally unrelated protein. Turner et al. (2002) showed that NalP (AspA) is a serine-protease autotransporter, identified the active-site serine and showed that antibodies in the serum of patients infected with N. meningitidis recognized the protein. In the present study, we investigated the possible lipidation of NalP and provide evidence for its function, being a modulator of autotransporter processing. Both findings are novel features of autotransporter proteins.
Analysis of the NalP protein sequence revealed the presence of a lipoprotein motif at the signal-sequence processing site. Because the secreted form of NalP lacked, in addition to the signal sequence, a peptide including the putative lipid moiety from the N terminus of the passenger domain, lipidation could not be demonstrated for the wild-type NalP. However, an active-site mutant of NalP (NalP-S427A), which was not autoproteolytically processed at the cell surface, could be labelled with 3H-palmitate in E. coli. Moreover, the antibiotic globomycin, which specifically interferes with lipoprotein biogenesis (Inukai et al., 1978), inhibited NalP secretion in N. meningitidis. Taken together, these results strongly suggest that NalP is lipidated.
Lipidation is an uncommon feature among autotransporters. Previously, the signal sequence of the AlpA autotransporter of H. pylori, which contains the rather uncommon (because of the large residue at position − 1) lipoprotein motif LAL↓C, was fused to mature β-lactamase and shown to be lipidated (Odenbreit et al., 1999), but lipidation of AlpA itself has not been shown. The SphB1 autotransporter of B. pertussis, which shows 28% identity and 40% similarity to NalP, also contains a lipoprotein motif LAA↓C (Coutte et al., 2001). Strong evidence for the lipidation of this motif was obtained by 3H-palmitate labelling of a fusion to β-lactamase and by the inhibition of the production of the protein in the presence of globomycin (Coutte et al., 2003a). In the case of NalP, we could also demonstrate lipidation of the full-length protein directly in E. coli. Apparently, lipidation, although it results in increased hydrophobicity of the protein and, thus, in increased affinity for membranes, does not hamper the transport of the passenger domain across the outer membrane. Interestingly, pullulanase of Klebsiella oxytoca is also N-terminally lipidated, but in this case the protein is secreted to the extracellular medium via a type II secretion system (Pugsley and Kornacker, 1991). In the case of NalP, in contrast to pullulanase, an N-terminal oligopeptide, including the lipid moiety, appeared to be removed upon secretion. Thus, the lipidated form of NalP is an intermediate in the biogenesis of the protein. A similar mode of secretion has been proposed for AlpA (Odenbreit et al., 1999). Furthermore, removal of the lipid moiety is reminiscent of the transport of the extracellular endoglucanase of the plant-pathogen Ralstonia solanacearum (Huang and Schell, 1990), presumably via a type II secretion system. This protein is also transiently lipidated during passage across the cell envelope, after which an N-terminal peptide with the lipid moiety attached is proteolytically removed.
Analysis of the extracellular proteins showed that IgA protease and App were the major secreted proteins of N. meningitidis grown in vitro. Moreover, two forms of both proteins were detected. The passenger domain of IgA protease of N. gonorrhoeae is known to consist of subdomains, i.e. the protease domain, a small γ peptide and a larger α peptide (Pohlner et al., 1987). The function of the α peptide is unknown, but it has been reported to contain nuclear localization signals, which could target fused proteins to the nucleus of eukaryotic cells (Pohlner et al., 1995). Mass-spectrometric analysis indicated that the larger IgA protease polypeptide detected in the culture supernatant of H44/76 corresponded to the protease domain, fused, necessarily via the γ peptide, to the α peptide, whereas the smaller polypeptide consisted of the protease domain alone. Remarkably, the App autotransporter appears to share this organization, as two forms of App were detected in the extracellular medium, with the smaller form lacking a C-terminal domain that was found in the larger one. Moreover, two putative nuclear localization signals could be assigned to this C-terminal domain (Fig. 2), which was therefore designated α peptide, in analogy to IgA protease. The presence of an α peptide in App appears to be unique to the N. meningitidis protein, as the H. influenzae Hap protein, which is very homologous to App (56% identity over the complete proteins), does not contain a similar domain. The secreted form of H. influenzae Hap and its translocator domain are separated by a peptide of only 35 amino acids (Fink et al., 2001), in which no nuclear localization sequence could be detected. Others have also noted the appearance of App in two forms (Abdel-Hadi et al., 2001; Serruto et al., 2003), and both groups reported differences in the relative amounts of the 100 kDa or 140 kDa forms, depending on the N. meningitidis isolates tested, whereas in E. coli only the 100 kDa form of App was observed. Our results now provide and explanation for these observations. NalP, which is phase-variably expressed, affects the processing of App, and only when NalP is produced, App is secreted with the α peptide attached.
Importantly, only the lower molecular weight forms of App and IgA protease predominated in the culture supernatant of a nalP disruption mutant. Similarly, the processing of another autotransporter of the serine-protease family, identified in the genome sequence of MC58 (NMB1998; van Ulsen et al., 2001) and tentatively called AusI, appeared to be affected by the nalP mutation (P. van Ulsen, B. Adler, P. van der Ley, L. van Alphen and J. Tommassen, in prep.). The results demonstrated that NalP affects the processing of other autotransporters in N. meningitidis H44/76, presumably at the cell surface. The analysis of the extracellular protein profiles of seven N. meningitidis isolates, either expressing NalP or not, corroborated this conclusion. The major secreted form of App detected correlated perfectly with the NalP-expression status of the isolate. Thus, the larger 140 kDa form predominated in NalP+ isolates, whereas the smaller 100 kDa form predominated in NalP– isolates. However, for IgA protease, such a correlation was not found. A higher molecular weight form was only observed in H44/76 and one of the four other NalP+ isolates tested. Apparently, the NalP-mediated modulation of IgA-protease processing is only effective in a subset of N. meningitidis isolates.
The results obtained with the NalP active-site mutant indicate that the serine protease domain of NalP is involved in the processing of App and IgA protease of H44/76. Although an indirect role of NalP cannot be excluded completely, we favour a model whereby NalP, being a serine protease of the subtilase family, cleaves the App and IgA protease precursors at a site between the passenger domain and the translocator domain, before autoproteolytic cleavage could have occurred, thereby disfavouring further processing of the passenger domain into the separate subdomains. This cleavage would then occur in competition with the autoproteolytic cleavage between the subdomains, which explains the presence of the lower Mr forms of App and IgA protease, besides their higher Mr forms, also in HB-1 expressing NalP. The observation that autoproteolytic cleavage of H. influenzae Hap occurs intermolecularly rather than intramolecularly (Fink et al., 2001; Fink and St Geme, 2003) supports a model of intermolecular cleavage between different autotransporters. Moreover, the multimeric ring-like structures observed for translocator domains of IgA protease (Veiga et al., 2002) might be instrumental to intermolecular processing. Possibly, hetero-oligomeric complexes could be formed between the translocator domains of different autotransporters, thus facilitating the processing of App and IgA protease by NalP. It remains to be investigated why NalP is lipidated. We hypothesize that lipidation of NalP might influence the topology of the protein, by bringing the N terminus of NalP containing the protease domain close to the cell surface. This topology might facilitate the processing of the other autotransporters at sites close to the cell surface, resulting in the release of the higher Mr forms of these proteins.
The fact that IgA protease is not in all strains a target for NalP-mediated processing could result from the sequence variation and mosaic gene structure of the iga gene (Halter et al., 1989; Lomholt et al., 1995), and the variability in the α peptide of IgA protease among neisserial isolates (Jose et al., 2000), including the reported absence of the processing site between the α peptide and the translocator domain in one N. meningitidis isolate (Lomholt et al., 1995). Our preliminary results indicate that iga of H44/76 differs from those in the sequenced genomes, specifically in the region where the α peptide and the translocator domain are adjoining (our unpublished observation). In contrast, the available sequences of App suggest that the α peptide of this protein is much more conserved. Therefore, our results identify App, a protein expressed by N. meningitidis during colonization and infection of the human body (van Ulsen et al., 2001), as the primary target of NalP. Additional targets, such as the IgA protease, might be strain dependent.
The observation that a processed form of the NalP active-site mutant was secreted in N. meningitidis, indicates that also the reverse, i.e. cleavage of NalP by another protease can occur. This other protease is, likely, also an autotransporter, as homologues of known outer membrane proteases like OmpT or OmpP are missing in N. meningitidis. Presumably, NalP contains a cleavage sites for this autotransporter, which appears to be of low-affinity in H44/76, as it becomes apparent only when autoproteolytic cleavage cannot occur. However, in other N. meningitidis isolates, NalP may contain a site of higher affinity, which could explain the observed two forms of NalP observed in the culture supernatants of those isolates.
Interestingly, the NalP-related SphB1 autotransporter of B. pertussis (see above) was found to be involved in the secretion of filamentous haemagglutinin (FHA; Coutte et al., 2001) and to be essential for B. pertussis colonization in a mouse model (Coutte et al., 2003b). FHA derives from a very large preprotein of 367 kDa called FhaB, which is transported over the outer membrane by the dedicated transporter protein FhaC (Willems et al., 1994). Proteolytic cleavage by SphB1 was shown to result in the release of FHA from the cell surface into the medium (Coutte et al., 2001), and this processing did not occur when a mutant non-lipidated SphB1 was expressed (Coutte et al., 2003a). Open reading frames encoding FHA-like proteins have also been identified in the N. meningitidis genomes and therefore NalP might be involved in the maturation of these FHA-like proteins as well. However, analysis of the secretion profiles on Coomassie-stained SDS-PAGE gels did not reveal major bands being affected by NalP, other than the ones described in this study. We therefore propose that NalP might function in the release of a specific form of N. meningitidis autotransporters, i.e. fused to their α peptides. The relevance of NalP-mediated modulation of autotransporter secretion for neisserial virulence remains to be investigated. However, the ability of the IgA protease α-peptide to mediate nuclear targeting (Pohlner et al., 1995) suggests a role for this domain in localization and targeting of the att-ached protease domain. Similarly, the nuclear-targeting sequences in the App α-peptide might influence localization of secreted App. Strikingly, Serruto et al. (2003) recently showed that App could mediate adherence to cultured human epithelial cells. They localized the domain involved in binding in the region, which we now identify as the α-peptide. This suggests that secreted App, when the α peptide is attached, might be targeted to epithelial cells. Like many other N. meningitidis genes encoding virulence factors (Saunders et al., 2000), expression of nalP appears to be phase variable (Turner et al., 2002; this paper). Moreover, NalP seems to contribute to the final localization of the App protease domain, by preventing the cleavage of the α-peptide from the rest of the passenger domain, which might influence virulence. If so, NalP also should be considered a virulence factor.
In conclusion, we demonstrate here the lipidation of NalP in N. meningitidis and the function of NalP as a modulator of the processing of App and IgA protease, although the latter seems not a general target for NalP in N. meningitidis isolates. Both findings are novel features of autotransporter proteins.
Bacterial strains, plasmids and growth conditions
Neisseria meningitidis strain H44/76 (B:15:P1.7,16) was originally isolated from the cerebrospinal fluid of a meningitis patient in Norway (Caugant et al., 1988). HB-1 is a H44/76-derivative in which a streptomycin-resistance cassette disrupts the capsule gene cluster. The other N. meningitidis serogoup B isolates used were MC58, M986, and 881607; the serogroup A isolates used were Z2491 and 13077; the serogroup C isolates used were FAM18 and 126E. All were present in the lab collection of isolates at the NVI (Bilthoven). Isolates were plated on GC-agar plates (Oxoid) supplemented with Vitox (Oxoid), and liquid cultures were grown in Tryptic soy broth (TSB; Oxoid). Mutant N. meningitidis strains containing disrupted genes were plated on GC-agar plates containing 100 µg ml−1 kanamycin. Neisseria meningitidis transformants carrying plasmids pEN300 or pEN305 were plated on GC-agar plates and grown in TSB containing 10 µg ml−1 chloramphenicol. Escherichia coli strains used, DH5α, Top10F′ (Invitrogen) and BL21(DE3) (Invitrogen), were grown on Luria–Bertani broth (LB) supplemented with 100 µg ml−1 ampicillin, 100 µg ml−1 kanamycin, 25 µg ml−1 chloramphenicol, or 12.5 µg ml−1 tetracycline for plasmid maintenance, where applicable, and with 0.5% glucose for full repression of the lac promoter. The cloning and expression vectors used were pCRII-topo (Invitrogen), pET11a (Invitrogen), pUC4K (Vieira and Messing, 1982), pFP10 (Pagotto et al., 2000) and pBR322.
Primer pairs NalPstart (5′-GGAATTCCATATGCGAACGAC CCCAACCTTCCCTA-3′) and NalPend (5′-CAAGATCTCA GAACCGGTAGCCTACGCCGA-3′) were designed to amplify the ORF of nalP from the start codon up to and including the stop codon, without any up- or downstream sequences. They were based upon the MC58 genome sequence (Tettelin et al., 2000). NdeI and BglII restriction sites (underlined) were included in the primers to facilitate cloning. Polymerase chain reaction was performed using chromosomal DNA of strain H44/76 as the template. Polymerase chain reactions and subsequent cloning steps were performed as described earlier (van Ulsen et al., 2001). Briefly, the PCR products were cloned into the pCRII-topo vector, yielding pCRII-nalP. The additional NdeI site present in the nalP ORF was removed by changing the recognition site 5′-CATATG-3′ into 5′-CGTATG-3′, a mutation that did not change the amino acid sequence of NalP. The mutagenesis was done by a three-primer method. First, the primers NalPstart and NalPmutNdel (5′-GGGCATAGGTGTTGGGCTGAGCT-3′) were used in a PCR reaction, with pCRII-nalP as the template. The resulting PCR product was purified from agarose gel and used as a primer in a second PCR reaction, together with primer NalPend and again pCRII-nalP as the template. The resulting 3260 bp product was isolated and TA-cloned in the pCRII-topo vector, yielding pCRII-nalP N. Finally, the altered nalP ORF was obtained by cutting pCRII-nalP N with NdeI and BglII and cloned into pET11a cut with NdeI and BamHI, resulting in plasmid pPU300. Plasmid pPU300 contained the ORF under control of the T7 promoter, which is inducible in strain BL21(DE3) by adding IPTG to the growth medium. A three primer strategy was also used to construct the mutant nalP-S427A, encoding a NalP in which the active site serine S427 was substituted by alanine. Primers NalPstart and NalPmutSA (5′-GGGTGCGGAAAAGGCTGTTCCGGCAA-3′; sequence resulting in the mutated triplet underlined) were used in the first PCR reaction with pCRII-nalP N as the template. The resulting fragment and primer NalPend were used in the second PCR, again with pCRII-nalP N as the template, after which the PCR fragment was isolated and subcloned in pCRII-topo, yielding pCRII-nalPS427A. From there the fragment was cloned after restriction with NdeI and BglII into the NdeI and BamHI sites of pET11a, yielding pPU305. The pPU300 and pPU305 cloning steps were checked by restriction enzyme digestion and sequencing.
The nalP ORF was completely sequenced using pCRII-nalP as template, as it contained the primary PCR product. Primers used were the universal M13 primers and a collection of primers that were based upon the sequence of nalP of MC58. All sequencing reactions were done with the BigDye sequencing kit (Perkin Elmer) and analysed on ABI310 or ABI377 sequencers (Perkin Elmer). The resulting sequences covered the ORF completely on both strands of the DNA. The sequence data was submitted to GenBank under accession number AY150284.
For the disruption of nalP by allelic exchange, the PstII-fragment of pUC4K, containing the kanamycin-resistance gene (kan), was cloned into the NsiI sites of nalP in pPU300. In this construct, kan replaces a 2112 bp internal fragment of nalP, which includes the DNA encoding the serine-protease active site residues and the putative autoproteolytic cleavage sites. A comparable strategy was used to insert the kanamycin-resistance cassette into the unique PstI site of app. The plasmid containing an iga::kan construct was a kind gift from G. Vidarsson (NVI). The plasmids for allelic exchange were used to transform N. meningitidis H44/76 and HB-1, according to standard procedures (Van der Ley and Poolman, 1992). Polymerase chain reaction analysis and Western blotting confirmed double crossover events resulting in target-gene disruptions.
The construct encoding the fusion of the signal sequence and 17 amino acids of mature NalP to mature β-lactamase was made by PCR using a four-primer method. Primers Bla1 (5′-CAATGCAGGCCCAGAAACGCTGGTGAAAGTAAAAGA-3′) and Bla2 (5′-GTTACCAATGCTTAATCAGTGAGGCA-3′) were used to generate a PCR product from pBR322 en-coding mature β-lactamase starting at position + 2 after the signal sequence cleavage site. Primers ssNalP1 (5′-GATCGAGATCTCGATCCCGCGAAA-3′) and ssNalP2 (5′-CGTTTCTGGGCCTGCATTGAAGTCGGGCGCA-3′) were used to generate a PCR product from pPU300 that contained the T7 promoter and the part of nalP encoding the signal sequence of NalP up to position + 17 after the signal sequence cleavage site. The primers Bla1 and ssNalP2 had extensions of 10 nucleotides complementary to each other, and a subsequent PCR reaction, containing the two PCR products and primers ssNalP1 and Bla2, resulted in a fragment that contained the promoter and the ORF encoding the fusion protein. This PCR product was cloned into pCRII-topo and the resulting plasmid was cut with EcoRI and PstI. The fragment with the hybrid ORF was ligated to pBR322 cut with the same enzymes, yielding plasmid pPU311 encoding the fusion protein under control of the T7 promoter.
We made use of the neisserial replicative vector pFP10 (Pagotto et al., 2000) to construct plasmids for the expression of nalP or nalPS427A in N. meningitidis. The pFP10-derived plasmid pRV2000 (Voulhoux et al., 2003) contains the neisserial omp85 gene under control of a lacIOP cassette. Primers PFP1 (5′-GCTTTCGCGAGCTCGAGTTCTAGATATT-3′) and PFP2 (5′-GACGTCAGATCTGTTTCAGTTTCATATG CAGTTCCTTGT-3′) were used to generate a fragment with the lacIOP cassette and the DNA corresponding to the untranslated 5′ leader sequence of omp85, including the ribosome-binding site, but without the omp85 coding region. Instead, NdeI and BglI I restriction sites were introduced (underlined in primer PFP2), with the ATG of the NdeI site at the position of the initiation codon of omp85. Furthermore, primer PFP1 contained a substitution resulting in the disruption of the BglI I site present upstream of lacI. The fragment was subcloned in pCRII-topo and subsequently cut out with XhoI and AatI I, and used to replace the corresponding fragment of pRV2000, resulting in plasmid pEN10. Subsequently, nalP and nalP S427A were cloned into the NdeI and BglII sited of pEN10, using pCRII-nalP N and pCRII-nalP S427A, respectively, yielding plasmids pEN300 (nalP) and pEN305 (nalPS427A), placing the ORF under control of the lac promoter.
Collection of cells and culture supernatants
Fresh overnight cultures of E. coli strain BL21(DE3) containing plasmids pET11a or pPU300 were diluted 1:100 in LB and grown to an optical density at 600 nm (OD600) of 0.6. IPTG was added to a final concentration of 0.1 mM, after which incubation was prolonged for another two hours. All N. meningitidis isolates were grown in TSB for approximately 6 h to a final OD600 of ∼3.0. Neisseria meningitidis HB-1 isolates, carrying either plasmid pEN300 or pEN305, were inoculated in TSB supplemented with 0.25 mM IPTG and chloramphenicol, when the expression of nalP was required, and were also grown to a final OD600 of ∼3.0. In the experiments where globomycin was added, HB-1 cells were grown to an OD600 of ∼0.8, after which globomycin, dissolved in ethanol, was added to a final concentration of 3 µg ml−1. Incubation was prolonged for 3 h.
In all cases, cells were harvested by centrifugation (4500 g, 5 min) and resuspended in phosphate-buffered saline (PBS), pH 7.4, to an OD600 of 10.0. The culture supernatants were centrifuged again (16 000 g, 5 min) to remove residual cells, and then, the protein content was precipitated by adding ice-cold trichloroacetic acid to a final concentration of 5% and incubation for at least 1 h at 20°C. Samples were then centrifuged (16 000 g, 15 min), and the pellets were washed with acetone and dissolved in 40 µl PBS, pH 7.6, per ml of collected supernatants, corrected for the final OD600 of the culture.
Fresh overnight cultures of E. coli strain BL21(DE3) containing plasmids pPU311, or pPU305 were diluted 1:100 in LB and grown to an OD600 of 0.6. Then, 3H-palmitate in ethanol (Amersham) was added to a final concentration of 20 µCi ml−1, and incubation was prolonged for 20 min. Next, IPTG was added to a final concentration of 1 mM, after which incubation was prolonged for another two hours. Cells were harvested and whole cell lysates, as well as medium samples, were subjected to SDS-PAGE. After fixing, gels were incubated for 30 min in 3H-enhancer (DuPont) and dried before autoradiography. Typically, films were developed after 2 weeks.
PCR-analysis for the presence of nalP
Bacteria were swabbed from freshly grown plates and resuspended in PBS to an OD600 of approximately 2. Five µl of the suspension were added to the PCR mixture, together with primers NalPstart and NalPmatEnd (5′-AGATCTCAGAAGAT GCGTACACCGTCGGCGGCA-3′). The latter primer hybridizes to region 2322–2346 of nalP. The PCR was performed with Taq DNA polymerase (MBI Fermentas) in a buffer according to the manufacturer's recommendations. Mixtures were first incubated for 5 min at 95°C, after which 32 cycles were performed of 1 min 95°C, 1 min 55°C and 3 min 72°C, using a Biometra PCR machine (Boehringer Mannheim). Samples were analysed on ethidium bromide-stained agarose gels.
Polyclonal antisera were raised against NalP and App. Recombinant proteins were purified from inclusion bodies produced in E. coli BL21(DE3) carrying plasmids pPU200, a pET-derivative containing the app gene of strain H44/76 (van Ulsen et al., 2001), or pPU300. Fresh overnight cultures were diluted 1:100 in 100 ml LB and grown to an OD600 of 0.6, after which IPTG was added to a final concentration of 0.1 mM and incubation was prolonged for two hours. Cells were harvested by centrifugation (7000 g, 10 min) and resuspended in 20 ml of 10 mM Tris-HCl, pH 8.0, 3 mM EDTA, and the protease-inhibitor cocktail Complete (Roche) at the recommended concentration. They were disrupted in batches by sonication for 2 × 5 min in a Branson sonifier, while on ice. Lysates were centrifuged (2000 g, 4°C, 10 min), and the pellet was dissolved in 20 mM Tris-HCl, pH 8.0, 0.1 M glycine, 7 M urea. Residual membrane fractions were removed by centrifugation (200 000 g, 4°C, 1 h). A further purification involved preparative SDS-PAGE, after which the protein band corresponding to the full-length recombinant protein (Mr 160 000 for App and Mr 114 000 for NalP) was excised from gel and extracted from the gel-slice using electroelution (Electroeluter, Bio-Rad), according to the manufacturer's recommendations. The final sample was in 2.5 mM Tris-HCl, pH 8.0, 0.1 M glycine and 0.1% SDS. These samples were sent to Eurogentec (Leuven) and used to immunize rabbits.
Polyclonal antiserum against IgA protease was kindly donated by G. Vidarsson (RIVM).
SDS-PAGE and Western blotting
An equal volume of 2× sample buffer (0.125 M Tris-HCl, pH 6.8, 20% v/v glycerol, 4% w/v SDS, 0.02% w/v bromophenol blue and 5% v/v β-mercaptoethanol; 2× SB) was added to either whole cell lysates or proteins precipitated from culture supernatants, and samples were then boiled for 10 min. Protein profiles were analysed by SDS-PAGE, using a Protean III minigel system (Bio-Rad) and an 8% (w/v) polyacrylamide running gel. The gels were either stained with Coomassie brilliant blue G250, or the proteins were blotted upon 0.45 µm Protran filters (Schleicher and Schuell), using the Protean II minigel blotting system (Bio-Rad) at 100 V for 1 h. Unspecific binding of antibodies to filters was prevented by overnight incubation in PBS with 0.5% Protifar (Nutricia) and 0.1% (v/v) Tween-20 (Merck). The anti-IgA protease and anti-App sera were diluted 1:20 000, the anti-NalP serum was diluted 1:30 000 in the same buffer and applied to the blots for 1 h. After extensive washing, the blots were incubated with goat anti-rabbit IgG serum conjugated to horseradish peroxidase (Biosource International) at a dilution of 1:15 000 in the same buffer. Binding of antibodies was visualized by chemiluminescence using the ECL kit (Pierce).
Proteins precipitated from culture supernatants were separated by SDS-PAGE with 0.4 mM thioglycolic acid in the 11% separating gel and then blotted upon polyvinilydene difluoride membranes (Millipore). The membranes were stained with Coomassie brilliant blue G250, and the bands to be analysed were excised. Samples were subjected to 5–7 steps of Edman degradation at the Protein Sequencing Facility, Utrecht University, using a protein sequencer model 476 A (Perkin Elmer).
Protein samples of the culture supernatants of N. meningitidis strain HB-1 and its nalP mutant derivative were separated by SDS-PAGE. Gel slices with the bands of interest were cut from gel and sliced in 1 mm3 pieces. Slices were then washed three times with water, shrunk with acetonitril, swollen again with 100 mM ammoniumbicarbonate and dried under vacuum. The dried gel pieces were reduced by adding 10 mM dithiothreitol in 100 mM ammoniumbicarbonate and heated for 45 min at a temperature of 56°C. Alkylation was done with 55 mM iodoacetamide during 30 min. The proteins in the slices were digested with modified trypsin (Promega) at a temperature of 37°C for 15 h. The resulting peptides were extracted with 5% formic acid and subsequently analysed by nano-HPLC-MS/MS (Meiring et al., 2002). The resulting spectra were interpreted and compared to computer-derived peptide patterns of the MC58 genome and the complete GenBank database.
We gratefully acknowledge Dr G. Vidarsson for his gift of the IgA protease knock-out construct and the anti IgA protease serum, Dr Miyakoshi (Sankyo, Japan) for his gift of globomycin, Dr A. de Jong for technical help and valuable suggestions, and M. Feijen and W. ten Hove for technical assistance.