INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Both Neisseria meningitidis and Neisseria gonorrhoeae express Type IV pili that are composed of multimers of the PilE protein subunit (17 to 20 kDa) and a possible tip-associated adhesin, PilC (27). Type IV pili are necessary for initial attachment of these pathogens to epithelial and endothelial cell monolayers (27, 37), twitching motility (47), and natural competence (7).

Gonococci and some meningococci express a type IV class I pilus that is identified by reactivity with monoclonal antibody (MAb) SM1 (45). Class I type IV pili are notable for high-frequency antigenic variation, which is the result of primary nucleotide sequence changes that occur through a recombination mechanism involving several silent pilin loci (pilS) and the expressed pilin locus (pilE) (8). Meningococci express either class I type IV pili or class II pili, which are distinguished by nonreactivity with MAb SM1. The class II pilE locus, which is not found in the same location of the genome as class I pilE, was recently identified in strain FAM18 (1). A comparison of class I and II PilE subunits revealed structural similarities between the two proteins including a conserved N-terminal domain and hypervariable central and C-terminal domains. Class II pilE loci have also been identified in the commensal Neisseria spp. N. lactamica and N. cinerea, which may act as reservoirs for allelic exchange (2).

The complete crystallographic structure of class I pilin from N. gonorrhoeae strain MS11 has been determined, and three posttranslational modifications have been identified (6). These attachments include an -glycerophosphate moiety on Ser94, a phosphate group on Ser68, and a disaccharide consisting of Gal(1-3)GlcNAc attached to Ser63. Class I pilin from N. meningitidis isolates is also substituted in this way (9, 21), but in some instances the disaccharide attachment is replaced with a trisaccharide group consisting of Gal(1-4)Gal(1-3)-2,4-diacetamido-2,4,6-trideoxyhexose (Gal-Gal-DATDH) (38, 39).

Various proposals have been made regarding the role of glycosylation in pilus function. Marceau et al. (21) found that changing the primary amino acid sequence to prevent glycosylation of pili resulted in the loss of soluble pilin production in meningococci but not in gonococci (22). Glycosylation of pili may also play a significant role in the resistance of meningococci to complement-mediated lysis by normal human serum (NHS). One model suggests that nonbactericidal anti-Gal immunoglobulin A (IgA) antibodies bind to glycosylated pili, thereby effectively blocking the access of bactericidal antibodies to the bacterial-cell surface (9).

Examination of the amino acid sequence of the class II type IV pilin reveals that the sites for posttranslational processing are conserved (1). We have examined the role of the pgl locus in the glycosylation of class II pilin expressed by N. meningitidis and determined whether removal of the glycan moieties on pili effect resistance to complement-mediated lysis by human sera.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. Meningococcal and gonococcal strains were cultured under aerobic conditions with 3.5% CO₂ at 37°C on GC agar (Difco) supplemented with 0.4% glucose, 0.01% glutamine, 0.2 mg of cocarboxylase/liter, and 5 mg of Fe(NO₃)₃/liter. The wild-type strains and constructed mutants used in this study are shown in Table 1. Antibiotic selection for meningococcal mutants was performed in brain heart infusion medium (BHI) supplemented with 2.5% fetal calf serum and 80 µg of kanamycin (sulfate salt)/ml or in GC medium containing 60 µg of spectinomycin/ml. Escherichia coli JM109 was used as a host for all DNA manipulations. It was routinely grown on Luria-Bertani medium which, where appropriate, was supplemented with antibiotics at the following concentrations: ampicillin and spectinomycin at 50 µg/ml, and kanamycin and chloramphenicol at 25 µg/ml.

Detection of polymorphisms in the pgl locus. For colony PCR, a single colony was resuspended in 20 µl of sterile water and 1 µl of this preparation was used as a template in the PCRs. The following primer pairs were used to detect polymorphisms in the pgl locus: 11610 and 11487, 11486 and 12420, 10861 and 12420, 11593 and 12420, 10370 and 10372, 10370 and 11445, 10370 and 11216, 10370 and 11601, 11610 and 12420 (Table 2). PCR amplification of this locus from N. gonorrhoeae strain FA1090 using the same primers was used as a control for the PCR conditions. The locations of primers in the pgl locus of N. meningitidis strain NMB are shown in Fig. 1.

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Locations of primers in the pgl locus of N. meningitidis strain NMB. Small half arrows and primer designations show the relative positions and directions of the primers. Below the ORF names are the location identifiers from the complete genome of N. meningitidis strain Z2491 (28).

Construction of mutants in the pgl locus. The methods used for the preparation and manipulation of DNA have been described previously (14). Internal fragments of each open reading frame (ORF) in the pgl cluster were individually amplified by PCR from strain NMB genomic DNA (Table 3). Where no convenient restriction sites were present in the gene, the splice overlap extension (SOE) technique (46) was used to introduce restriction sites into the appropriate positions within the PCR product (Table 3). The PCR products were purified using QIAquick spin columns, treated with T4 DNA polymerase, and cloned into the pHSG576 vector (43). The aphA-3 cassette from pUC18K (26) was released by restriction enzyme cleavage or amplified by PCR, and the resultant fragments inserted into unique restriction sites of each ORF cloned into pHSG576. The orientation of the aphA-3 cassette in each construct was determined by directional PCR using the KANC primer (14) and the reverse primer for each ORF. Every construct was sequenced to confirm an in-frame fusion of the aphA-3 cassette with the ORF into which it had been inserted. In some cases, the ORF::aphA-3 cassette was subcloned into pHSG298 (43) in an attempt to increase the frequency of transformation into strain NMB (Table 3). Transformation of strain NMB was performed as described previously (14), and mutants were checked by colony PCR using the oligonucleotide primer pairs listed in Table 3. A negative stain was performed on each mutant, and results were examined by electron microscopy (JEOL 100) to confirm piliation.

Construction of an insertionally inactivated mutant in the class II pilE locus in N. meningitidis strain NMB. Primers for the class II pilE locus (pilEII) were designed from an alignment of the pilEII loci from N. meningitidis strain FAM18 (1) and commensal neisseria isolates (2). The pilEII gene from strain NMB was amplified using the primer pair 13412 and 13411 and was completely sequenced (GenBank accession no. AF320321). To construct a pilEII:: cassette, a unique HincII site was incorporated into pilEII using the primer pair 13513 and 13537 and the primer pair 13511 and 13512 to generate two products which were subsequently joined together using the primer pair 13412 and 13411. This product was cloned into the HincII site of pHSG298. The  spectinomycin cassette was released from pHP45 on a SmaI fragment and was ligated into the introduced HincII site to form pJKD2426. pJKD2426 was used to transform NMB, and spectinomycin-resistant mutants were screened for acquisition of the cassette using the primer pair 13412 and 13411.

Conditions for SDS-PAGE and Western immunoblotting. MAb SM1 (45) was used for the detection of class I pili, and class II pili were detected using MAb AD211 (34). For the detection of pili, 10 µg of total protein from each strain was loaded per lane on a 12.5% (wt/vol) Laemmli sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel (20). The proteins were transferred to a carboxymethylcellulose membrane in standard transfer buffer (25 mM Tris [pH 8.3], 192 mM glycine, and 20% [vol/vol] methanol) at 30 V overnight using a Bio-Rad Protean transfer apparatus. After the transfer was completed, the membranes were blocked for 1 h with bovine serum albumin (2% BSA) in TBS (50 mM Tris-150 mM NaCl [pH 7.4]). This was followed by two 10-min washes in TBS before incubation overnight with a 1:200 dilution of MAb SM1 or a 1:3,000 dilution of MAb AD211 in TBS. The membrane was washed five times (10 min each time) before incubation for 2 h with a 1:2,000 dilution of anti-mouse IgG-horseradish perioxidase (HRP) conjugate. Again the membrane was thoroughly washed as before, and the assay was developed with the addition of the color substrate solution for HRP (20).

Detection of glycosylated pilin. Glycosylation of pilin was detected using the digoxigenin (DIG) glycan detection kit (Boehringer Mannheim) according to the manufacturer's instructions (9). In brief, the immunoblots of whole-cell lysate proteins were incubated with periodate, which oxidizes adjacent hydroxyl groups of sugars to aldehyde groups. After removal of the periodate solution and extensive washing, the derivatized sugars were labeled with a spacer-linked steroid hapten DIG covalently linked to a hydrazide group. In the final steps, DIG was detected in a standard immunoblot format using an anti-DIG-alkaline phosphatase conjugate.

Serum bactericidal assays. The serum bactericidal assays were performed as previously described (15).

Quantitative ELISA for the detection of capsular polysaccharide on whole cells. A whole-cell enzyme-linked immunosorbent assay (ELISA) was performed according to the published procedure (41) with minor modifications using the serogroup B capsule-specific MAb 2-2-B. One hundred microliters of a 1:3 dilution of the cell suspension at an optical density at 650 nm (OD₆₅₀) of 0.1 was added to the microtiter plates, and the incubation was carried out at 37°C instead of 33°C.

Nucleotide sequence accession number. The sequence of the pilEII gene from strain NMB has been assigned GenBank accession no. AF320321. The orf3-pglB2-orf8 region from this strain has been assigned accession no. AF320320.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Glycosylation of class II pili but not of class I pili is detected by periodate oxidation. We examined the glycosylation phenotypes of pili from a limited selection of meningococcal strains and included N. gonorrhoeae strain MS11 as a control for the expression of glycosylated class I pilin (Fig. 2). Colony PCR of the eight meningococcal isolates in this panel using the pilEII primer pair 13412 and 13411 indicated that three strains, NMB, FAM18, and F8229, contained the class II pilin locus (Materials and Methods). Western immunoblotting using MAb AD211 confirmed that strains NMB and FAM18, but not F8229, expressed class II pilin (Fig. 2C). Conversely, all of the remaining strains, except M981, expressed MAb SM1-reactive class I pili. The glycosylation immunoblot showed that the pilin of strain NMB reacted strongly with the glycan detection reagents, whereas a much fainter band corresponding in size to the pilin of FAM18 was detected by this assay. In all instances, the glycosylation assay did not detect pilin-linked glycans in class I pilin-expressing isolates. Even though reactive bands were present in some strains, they did not correspond in size to the pilins and were variable in appearance when the assay was repeated (data not shown). Based upon this survey, we decided to investigate the genetics for the glycosylation of class II pili in strain NMB.

View larger version (74K): [in this window] [in a new window]   FIG. 2.   Glycosylation status of class I and class II pilin from eight meningococcal isolates. Whole-cell lysates were electrophoresed on an SDS-PAGE gel and stained with Coomassie blue (A). Duplicate gels were transferred to nitrocellulose membranes and either incubated with MAb SM1 (B) to detect class I pilin or with MAb AD211 (C) to detect class II pilin or assayed for detection of glycoproteins using the DIG-glycan kit (D). White asterisks indicate the positions of glycosylated pilin in panel D. Lanes: 1, N. gonorrhoeae strain MS11; 2, N. meningitidis strain 6083; 3, N. meningitidis strain GA0929; 4, N. meningitidis strain F8229; 5, N. meningitidis strain FAM18; 6, N. meningitidis strain M981; 7, N. meninigitidis strain H44/76; 8, N. meningitidis strain NMB; 9, N. meningitidis strain MC58.

Identification of the pgl cluster in the class II pilin-expressing N. meningitidis strain NMB. Previous studies of glycoproteins in archaebacteria (10, 17, 18, 40, 48) and Bacillus alvei (11) have shown that glycans destined for attachment to proteins are synthesized as lipid-linked intermediates. These studies identified the lipid carrier as dolichol phosphate and undecaprenol phosphate (UdP) (10), thereby suggesting that synthesis and transport of these glycans are similar to those of O antigens. Based on this premise, we hypothesized that the synthesis of lipid-linked glycans for pilin glycosylation in Neisseria spp. would require a UdP transferase that would have significant amino acid sequence similarity to either the UdP-galactose-phosphotransferase, RfbP, or the UdP-N-acetylglucosamine transferase, RfbU. A BLAST search of all six translated frames of the meningococcal Z2491 genome using the amino acid sequence of Salmonella enterica serovar Typhimurium RfbP (accession no. S15314) identified one strong candidate that was 37% identical to this protein and was encoded by a single ORF, termed pglB (NMA0639). Analysis of the surrounding nucleotide sequence identified a further six ORFs in this cluster (Fig. 3). The subsequent genome annotation of Z2491 identified all of the ORFs in this cluster as potentially encoding enzymes necessary for lipopolysaccharide or capsule biosynthesis (28). The cluster of seven ORFs was delineated upstream by an ORF encoding an enzyme necessary for ribose synthesis, ribD (NMA0644), and downstream by an ORF (NMA0636) encoding a protein with sequence similarity to a hypothetical protein from Haemophilus spp. (Fig. 1 and 3). The potential involvement of this locus in pilin glycosylation was further supported by the discovery of a flagellin glycosylation locus in Campylobacter jejuni (42) in which four ORFs were homologous to those found in the meningococcal locus (data not shown). Further, Power et al. (29) described the involvement of three of these ORFs, pglB, pglC, and pglD, in the glycosylation of class I pilin from N. meningitidis strain C311.

View larger version (21K): [in this window] [in a new window]   FIG. 3.   Polymorphisms in the pgl locus of N. gonorrhoeae and N. meningitidis. The sequences of the pgl clusters in strain FA1090 (available at the Gonococcal Genome Sequencing Project website, http://www.genome.ou.edu), N. meningitidis strain Z2491 (28), and N. meningitidis strain MC58 (44) can be found in the genome databases. The organization of the pgl locus in strain NMB and strain MS11 was determined by PCR, and the region corresponding to orf3-pglC in strain NMB has been sequenced (GenBank accession no. AF320320). The C-terminal domains of pglB (striped) and pglB2 (crosshatched) are indicated. The unique orf8 (stippled arrow) and the Correia element (solid box) are indicated in the pgl locus of strain NMB.

Glycosylation phenotype of mutants in the pgl cluster. To determine the roles of the eight ORFs found in the pgl cluster of strain NMB, each ORF was disrupted by the insertion of a nonpolar kanamycin resistance cassette and whole-cell lysates of each mutant were assayed for the presence of glycosylated pili (See Materials and Methods; also Table 1). Electron microscopy confirmed that all the mutants in the pgl cluster had normal piliation phenotypes (data not shown). As expected, strain NMB expressed glycosylated pilin, whereas the negative control strain CMK28, in which the class II pilin locus was disrupted, did not react with the DIG-glycan reagents (Fig. 4). This assay revealed that pglF, pglB2, pglC, and pglD mutants produced pili with a modified glycosylation pattern (partly or completely removed), whereas orf2, orf3, orf8, and avtA were normal for this phenotype.

View larger version (72K): [in this window] [in a new window]   FIG. 4.   Pilin glycosylation phenotypes of pgl mutants in strain NMB. Whole-cell lysates of the pgl mutants and control strains were separated on an SDS-PAGE gel and transferred to a nitrocellulose membrane. Glycoproteins were detected using the DIG-glycan kit. Lanes: 1, N. gonorrhoeae strain MS11; 2, N. meningitidis strain MC58; 3, N. meningitidis strain NMB; 4, CMK28 (pilEII); 5, CMK20 (pglF); 6, CMK21 (orf2); 7, CMK22 (orf3); 8, CMK23 (pglB2); 9, CMK24 (orf8); 10, CMK25 (pglC); 11, CMK26 (pglD); 12, CMK27 (avtA).

Glycosylated pili are not required for resistance to NHS. Hamadeh et al. (9) have proposed that elevated levels of anti-Gal IgA antibody in human serum may protect N. meningitidis from killing by binding to glycosylated pili. We therefore examined whether removal of the glycans attached to pili would result in increased sensitivity to NHS. Mutant M7, in which capsule biosynthesis has been abolished (36), was used as a positive control for sensitivity to NHS (15). After 30 min in 50% NHS, wild-type NMB and strains carrying mutations in pglF, orf2, orf8, pglD, and avtA remained resistant to killing (Fig. 5). In contrast, mutations in pglB2 and pglC significantly decreased the resistance of strain NMB to NHS (>2-log kill; P < 0.05). Interestingly, inactivation of orf3, which does not affect pilin glycosylation, also resulted in a modest decrease in serum resistance (~1-log kill; P = 0.018) (Fig. 5). Therefore, the loss of pilin glycosylation did not necessarily result in serum sensitivity, suggesting that other factors were responsible for this phenotype in orf3, pglB2, and pglC mutants.

View larger version (20K): [in this window] [in a new window]   FIG. 5.   (A) Serum resistance profiles of N. meningitidis strain NMB pgl mutants CMK22 (orf3), CMK23 (pglB2), and CMK25 (pglC). The graph represents a time course assay over 30 min using 50% pooled NHS. Each point on the graph represents the mean percent survival in duplicate experiments, and the error bars are standard errors of the mean for each data set. The acapsulate mutant derivative of strain NMB, M7, was used as a serum-sensitive control. (B) Capsule expression profiles of CMK22 (orf3), CMK23 (pglB2), and CMK25 (pglC) nonsibling mutants. Each column represents the mean optical density at 405 nm (OD405) from five capsule ELISA readings, while the error bars represent the standard deviation of each data set. Percentage values were calculated on the assumption that the mean value for the parental strain NMB was equivalent to 100%. Similarly, two-tailed Student t tests were performed using these data.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The glycosylation of proteins in procaryotes has been most extensively studied in relation to archaebacterial S-layer proteins and pilin (10, 17, 18, 40, 48). Together these studies indicate that the sugar groups destined to be attached to glycoproteins are first synthesized on a lipid anchor, either C₆₀-dolichol or C₅₅-dolichol, and then transferred to the protein. Many aspects of this process resemble the steps in the assembly of lipid-linked oligosaccharide intermediates for capsular polysaccharide (33), O-antigen (31), and peptidoglycan (12) biosynthesis. Until recently, the genes encoding the enzymes necessary for the synthesis of these lipid-linked sugars and the transfer of these groups to a glycoprotein were unknown. Investigations into the genetic basis for glycosylation of flagella from C. jejuni (42) and Caulobacter crescentus (19) revealed that many of the genes necessary for this process encode proteins that are highly related to transferases involved in O-antigen and capsule biosynthesis. In particular, the genetic cluster required for glycosylation of flagella in C. jejuni encodes a protein with significant sequence identity to undecaprenol galactose phosphotransferase (RfbP), which is necessary for transfer of galactose 1-phosphate from UDP-galactose to UdP (31). These studies suggested to us that an RfbP homologue might be necessary for pilin glycosylation in Neisseria spp., and this has resulted in the identification of a cluster of eight ORFs that were subsequently investigated for involvement in pilin glycosylation.

We detected glycosylation of the class II pili from strain NMB after periodate oxidation of the glycoproteins. Interestingly, this assay was unable to detect glycosylated class I pili from N. gonorrhoeae strain MS11, which has the disaccharide substitution (6). Power et al. (29) also noted a similar phenomenon with the trisaccharide glycosylated class I pilin from N. meningitidis strain C311 but were able to overcome this lack of sensitivity by pretreating samples with galactose oxidase. The basis for this difference in sensitivity to periodate oxidation is unclear, although it may indicate that the class II pilin from strains NMB and FAM18 have different or modified glycan attachments.

The pgl cluster in strain NMB contained a modified version of the pglB found in the database sequences and was designated pglB2. In strain NMB, a novel ORF, orf8, was also inserted between pglB2 and pglC. Although PglB and PglB2 have different C-terminal domains, insertional inactivation of pglB2 in strain NMB confirmed that the protein encoded by this gene is required for pilin glycosylation. Power et al. (29) have previously shown that pglB is necessary for the addition of the trisaccharide glycan to class I pili in N. meningitidis strain C311. They proposed that PglB may operate as a bifunctional enzyme, with the N-terminal domain acting as a potential undecaprenol transferase and the C-terminal domain functioning as an acetylase required for the biosynthesis of the 2,4-diacetamido sugar residue of the glycan (DATDH). PglB2 has likely retained the undecaprenol transferase activity, since the N-terminal domain is intact. The C-terminal domain of PglB2, however, may be functionally different and may result in a different modification pattern of the acetamido sugar. This hypothesis could also potentially explain why the glycosylated class II pilin of strain NMB is more sensitive to periodate oxidation.

Serial inactivation of the genes in the pgl cluster in strain NMB confirmed that, in addition to pglB2, pglF, pglC, and pglD were involved in pilin glycosylation. Power et al. (29) have also recently shown that pglC and pglD are necessary for the addition of the trisaccharide glycan to class I pili in N. meningitidis strain C311. They have tentatively proposed that PglC and PglD are enzymes necessary for the biosynthesis of DATDH. PglF, which was not included in the study by Power et al. (29), contains a 70-amino-acid domain that has significant identity with a corresponding region of a potential repeat unit transporter, Cps2J, required for capsule transport in Streptococcus pneumoniae (13). Repeat unit transporters, or "flippases," are proposed to transport lipid-linked oligosaccharide intermediates from the cytoplasm to the external surface of the inner membrane prior to polymerization and transfer to the final anchor (in the case of capsule and O antigen, phospholipids and lipid A core, respectively) (31, 33). If indeed pilin glycosylation proceeds through a lipid-linked intermediate, then PglF would be necessary to transfer the trisaccharide-undecaprenol intermediate to a position where the glycan could be transferred to the pilin subunit before polymerization into pili. Future experiments will explore whether this model is possible.

Hamadeh et al. (9) have suggested that glycosylation of pili may increase resistance to complement-mediated lysis by NHS. They proposed that anti-Gal IgA antibody, which is normally found in human serum and recognizes terminally linked -galactosyl residues, could bind to glycosylated pili and block activation of the alternative complement pathway. At the time these experiments were conducted, mutants expressing nonglycosylated pili were not available. Therefore, we decided to test the NHS sensitivity of the glycosylation mutants of the serum-resistant strain NMB (15). These experiments revealed that glycosylation of pili in strain NMB had no direct effect on serum sensitivity. Although pglB2 and pglC mutants were significantly more serum sensitive than wild-type NMB, pglF and pglD mutants that were also defective in pilin glycosylation retained a normal serum resistance profile. In addition, the mutation of orf3, which did not affect pilin glycosylation, also resulted in a modest reduction in resistance to NHS. We subsequently found that orf3, pglB2, and pglC mutants, but not pglF and pglD mutants, expressed less capsular polysaccharide. Moreover, the reduction in capsule expression in the orf3, pglB2, and pglC mutants appeared to correlate with the serum sensitivity profiles of these mutants. In other words, a 10% reduction in capsule expression resulted in a 1-log kill of orf3 mutants, whereas a reduction of approximately 50% resulted in a 2-log kill of pglB2 and pglC mutants. Therefore, these experiments indicate that the expression of capsular polysaccharide is the predominant serum resistance determinant in meningococci irrespective of pilin glycosylation. Masson and Holbein have also previously observed that decreased levels of capsular polysaccharide, apart from complete abolition of capsule expression, have significant effects on the serum resistance of meningococci (23, 24). On the basis of these results, we hypothesize that lipid-linked oligosaccharide intermediates may accumulate in the pglB2 and pglC mutants and therefore interfere in capsular assembly or transport. The basis for the capsule deficiency of orf3 mutants is less clear but may also have a similar explanation if this gene is proven to encode an enzyme necessary for UdP processing.

In conclusion, we have explored the pgl cluster necessary for pilin glycosylation of a class II isolate, strain NMB, and shown that the genetic organization of this cluster differs from that in class I meningococci and N. gonorrhoeae. Our preliminary data may also indicate that a modified glycan is attached to class II pili. Since pglF, pglC, and pglD are conserved in gonococcal and meningococcal isolates expressing glycosylated (class I) pilin with a disaccharide (strain MS11) or a trisaccharide (strain C311) addition, the functions of the proteins encoded by these genes are likely to be necessary for common steps in pilin glycosylation. However, pglB is a polymorphic gene, and different forms of the encoded protein may be involved in specific modifications of the acetamide sugar. Further, blockage of pilin glycosylation at certain points in this pathway has unexpected pleiotropic effects on other cell surface structures, particularly the expression of capsular polysaccharide.