Genetic and functional analysis of the phosphorylcholine moiety of commensal Neisseria lipopolysaccharide

Authors


Summary

Phosphorylcholine (ChoP) is a common surface feature of many mucosal organisms, including Neisseria spp., in which it is present exclusively on pili of pathogenic Neisseria and on the lipopolysaccharide (LPS) of commensal Neisseria (Cn). Its presence in Cn has been confirmed by nuclear magnetic resonance. It appears that choline is the main source for the production of ChoP by Cn. We have sequenced a locus, containing four genes (licA–D) with 47–73% identity to the lic1 locus of Haemophilus influenzae (Hi) and 21–40% identity to lic genes in Streptococcus pneumoniae, involved in the production and incorporation of ChoP. The arrangement of the Cn genes and the presence of CAAT repeats, responsible for phase variation of ChoP expression, resemble Hi and differ from S. pneumoniae. Cn DNA flanking the lic locus contains genes ilvE and NMA2149 with >85% identity to the pathogenic Neisseria genes. However, there are no lic genes in the corresponding location or elsewhere in pathogenic Neisseria. This suggests either the loss of the locus from pathogenic Neisseria or a horizontal transfer of genes to Cn, perhaps from H. influenzae spp. As in Hi, ChoP enhances adherence to and invasion of human epithelial cells via the receptor for platelet-activating factor. However, ChoP expression also increases susceptibility to serum killing mediated by complement and C-reactive protein. Taken together, these observations support the hypothesis that the ability of many organisms to switch off ChoP expression rapidly represents an important adaptation to dif-ferent environments encountered during the colonization/infection process and that the ChoP moiety apparently synthesized by distinct means in pathogenic and commensal Neisseria represents an advantage in the colonization properties of these bacteria.

Introduction

Phosphorylcholine (ChoP) has been detected on a number of prokaryotes, including Gram-positive and Gram-negative bacteria. Many of these are pathogens, e.g. Streptococcus pneumoniae, Bacillus spp., Clostridium spp., Haemophilus influenzae, Neisseria meningitidis, Neisseria gonorrhoea and Pseudomonas aeruginosa (Mosser and Tomasz, 1970; Sanchez-Beato et al., 1995; Gillespie et al., 1996; Kolberg et al., 1997; Weiser et al., 1998a). It has also been demonstrated on a number of nematodes (Harnett and Harnett, 1999). Recent studies of the oral flora have identified additional species that have structural molecules bearing ChoP (Gmur et al., 1999), and one of these is the Gram-negative Actinobacillus actinomycetemcomitans, which is associated with periodontal diseases (Schenkein et al., 2000). Most of these organisms contain phosphorylcholine within structural glycans or glycolipids, such as teichoic acids, lipoteichoic acids or lipopolysaccharides (LPS), whereas some express ChoP on surface proteins, for example in P. aeruginosa on a 43 kDa protein and in pathogenic Neisseria on pili (fimbriae).

Recent studies performed in our laboratories have shown that, despite considerable similarities between commensal and pathogenic neisserial pili as well as LPS, ChoP epitope occurs exclusively on commensal neisserial LPS and on pili of pathogenic Neisseria, but not on any other moieties in these bacteria (Serino and Virji, 2000), thus differentiating pathogenic Neisseria from the commensal species.

The phosphorylcholine moiety of distinct organisms appears to have distinct properties, in some cases beneficial to the infected host and in others detrimental. Considering the exclusive presence of ChoP on pathogenic Neisseria pili and on commensal Neisseria LPS, we asked questions regarding the genetic basis of ChoP incorporation in Neisseria. In addition, as commensal Neisseria (rarely pathogenic) and H. influenzae (with the potential to cause serious infections) both express ChoP on their LPSs as well as sharing an environmental niche, we examined how the biological function of ChoP compared in the two species.

The function of the phosphorylcholine moiety in pathogenesis has been particularly investigated in S. pneumoniae and H. influenzae. Experiments using animal models of S. pneumoniae infection have suggested that ChoP may mediate adhesion and invasion of S. pneumoniae in the lung (Cundell et al., 1995) and brain (Ring et al., 1998) by interacting with the receptor for platelet-activating factor (PAF), present on epithelial and endothelial cells. Some oral bacterial species behave like the pneumococcus, gaining access to the circulatory system by binding to the PAF receptor on endothelial cells and inducing elevated levels of anti-ChoP antibodies (Schenkein et al., 2000). In the case of H. influenzae, it has been suggested that ChoP contributes to its persistence in the human respiratory tract and to the sensitivity to serum killing mediated by C-reactive protein (CRP) (Weiser et al., 1998b). However, ChoP may be incorporated into the LPS of H. influenzae at different locations (on a chain extension from heptose III or heptose I), and these differences may determine the accessibility of ChoP to the binding of antibody or CRP and the sensitivity of the organisms to CRP (Lysenko et al., 2000a). Moreover, ChoP renders the organisms more resistant to the antimicrobial activity of peptides, such as the cathelicidin LL-37/hCAP18, which contribute to the innate defence of the human respiratory tract (Lysenko et al., 2000b).

In the case of H. influenzae, choline is acquired from the growth medium, phosphorylated and attached as ChoP to a terminal hexose residue on the LPS (Weiser et al., 1997; Lysenko et al., 2000a). The incorporation of choline into LPS requires the expression of four genes, licA–D (Weiser et al., 1989). Genes homologous to licA–D have also been identified recently in S. pneumoniae (Zhang et al., 1999), although the genes have a different organization, and two similar copies of licD (licD1 and licD2) are involved in the attachment of the two residues of ChoP per repeating unit of the teichoic acid.

In our previous studies (Serino and Virji, 2000), we demonstrated the presence of a licA homologue in commensal Neisseria, but there was no evidence for a similar gene in pathogenic Neisseria, indicating that a different pathway is responsible for the expression of ChoP on pili of N. meningitidis and N. gonorrhoeae. In addition, we observed incorporation of radiolabelled choline from the growth medium into LPS of commensal Neisseria, suggesting that a genetic locus similar to that in H. influenzae could be present in commensal Neisseria. In this report, we describe the sequencing of the entire lic locus of commensal Neisseria and show that it contains the counterparts of H. influenzae licA–D genes. The locus has the same gene arrangement in H. influenzae. In addition, a similar feature of multiple tandem repeats of 5′-CAAT-3′ within the coding region of the first gene, licA, is present in various commensal Neisseria strains, and the number of repeats correlates with the spontaneous phase variation of the ChoP epitope (Serino and Virji, 2000).

This study also suggests that the expression of ChoP might aid in the adherence of commensal Neisseria to human respiratory tract epithelial cells and may play a role in their colonization of the respiratory niche. However, by binding to the acute-phase reactant CRP, it may also contribute to the sensitivity of the ChoP-expressing bacteria to killing mediated by complement and CRP.

Results

Nuclear magnetic resonance (NMR) analysis of LPS from commensal Neisseria strains confirms the presence of phosphorylcholine

The presence of an epitope reacting with anti-ChoP monoclonal antibodies (mAbs) indicated strongly that this moiety is present on the LPS of commensal Neisseria. The final proof of its presence was obtained by NMR. For this purpose, LPS was extracted from commensal Neisseria strains that were either reacting (Neisseria subflava C450) or non-reacting (Neisseria lactamica NL2) with anti-ChoP mAb and subjected to 1H-NMR analysis. The presence of ChoP on the LPS was indicated by a strong signal at ≈ 3.25 p.p.m., resulting from the ChoP methyl protons (Fig. 1) in C450 but not in NL2 LPS. Confirmation of the presence of ChoP in C450 LPS was obtained by reanalysis of the LPS samples after the addition of exogenous phosphorylcholine. This resulted in an increased peak at the position (≈ 3.25 p.p.m.) indicative of the presence of ChoP.

Figure 1.

1H-NMR spectra of LPSs from N. lactamica ChoP strain NL2 (A and B) and N subflava ChoP+ strain C450 (C and D), with (B and D) and without (A and C) the addition of exogenous ChoP. The signal at ≈ 3.25 p.p.m. in (C) (arrow) is indicative of the presence of ChoP. The signal appeared in (B) and increased in (D) after the addition of exogenous phosphorylcholine. No signal is observed in (A). Spectra were recorded at 500 MHz in D2O with HOD as an internal chemical shift standard.

Commensal neisserial genome contains a locus highly homologous to the Haemophilus influenzae lic1 locus

In our previous studies, we identified a gene in commensal Neisseria strains homologous to the H. influenzae licA gene (Serino and Virji, 2000). A polymerase chain reaction (PCR)-based strategy, which allows direct amplification of unknown DNA flanking a known DNA, was used to amplify and sequence products downstream of the licA gene. This technique uses a specific biotinylated primer and a degenerate flanking primer in an initial PCR step. The amplified fragments are isolated using streptavidin-coated magnetic beads. The purified fragments are separated, and the non-biotinylated strand is used in a second PCR reaction with a specific nested primer and a primer corresponding to the non-degenerate 5′ part of the flanking primer. These amplified PCR products are then purified and sequenced directly (see Experimental procedures).

Using this method, we obtained the complete licA gene sequence from N. lactamica strain C501 and N. subflava strain C450 and discovered the presence of three additional genes downstream of licA, with homology to the licB, licC and licD genes of H. influenzae. The complete LicA homologue of commensal Neisseria is 60% identical to the Haemophilus gene product encoding a choline kinase and 30% identical to the S. pneumoniae LicA gene product (Fig. 2A). The second gene in the H. influenzae lic1 locus, licB, is reported to encode a protein involved in choline transport (Weiser et al., 1997). The commensal Neisseria LicB homologue shows 47% identity with the H. influenzae protein and 21% with the S. pneumoniae protein (as determined from the deduced amino acid sequence; Fig. 2B). The third gene in the locus encodes a protein with 68% identity to the licC gene product of H. influenzae, a protein with similarity to several pyrophosphorylases (Weiser et al., 1997), and 40% identity to the S. pneumoniae gene product (Fig. 2C). Finally, a fourth gene in the locus encodes a protein that is 73% identical to the licD gene product of H. influenzae (Fig. 2D). This protein appears to be involved in the transfer of phosphorylcholine from its donor choline diphosphonucleoside to a specific oligosaccharide structure on the LPS (Weiser et al., 1997). Different licD alleles have been found in two Haemophilus strains (Rd or Eagan), and specific alleles apparently determine the chain extension to which ChoP is attached. Three amino acids (DYD), at position 201, are present in the Eagan strain but not in Rd (Lysenko et al., 2000a). These amino acids are also absent from the commensal LicD protein.

Figure 2.

Alignments of LicA (A), LicB (B), LicC (C) and LicD (D) homologues of commensal Neisseria strain C501 (Cn Lic), H. influenzae (Hi Lic) and S. pneumoniae (Stpn Lic). The predicted amino acid sequences of the lic genes were aligned by CLUSTALW, and identical amino acids (black) are highlighted by BOXSHADE. Spaces (–) were introduced to maximize sequence homology. The striped box indicates the presence of multiple CAAT repeats at the 5′ end of the licA gene in commensal Neisseria and H. influenzae. The dotted box shows the presence of a consensus sequence for protein kinases (A). Three amino acids, DYD, indicated at position 201, are present in strain Eagan of H. influenzae but are absent in Rd strain and in commensal Neisseria LicD (D).

In S. pneumoniae, two tandem but distinct copies of licD, licD1 and licD2, are present and are transcribed in a direction divergent from licA–C, separated by three additional open reading frames (ORFs; Zhang et al., 1999). The two licD gene products are homologous to each other (44% identity) but share 33% and 31% identity, respectively, with the commensal Neisseria licD protein. Thus, the general organization of the lic locus is similar to that of the lic1 locus of H. influenzae, and differs from the organization of the lic genes in S. pneumoniae (Fig. 3A).

Figure 3.

A. Organization of the lic1 locus in different organisms. Comparison of the S. pneumoniae, H. influenzae and commensal Neisseria lic gene arrangement. The directions of ORFs are indicated by arrows.

B. Correlation between ChoP expression and the genotype of the licA gene. The nucleotide sequence at the 5′ end of the licA gene from different commensal Neisseria strains is shown, containing a variable number of CAAT tandem repeats and two potential initiation codons (α and β). Variation in the number of repeats creates a translational switch with possible translation products in phase with the licA ORF. Accordingly, only when one of the two initiation codons was in frame with the rest of the licA ORF was the expression of ChoP (indicated by + or –)observed.

Further sequencing of PCR products amplified from the downstream DNA region of licD of commensal Neisseria did not reveal the presence of any tandem copy of the same gene. However, sequencing of the downstream DNA region revealed a high degree of similarity (≈ 97%) to a gene retrieved from the total genomic sequence of pathogenic N. meningitidis strain Z2491 (serogroup A; Sanger Centre), strain MC58 (serogroup B; TIGR) and N. gonorrhoeae strain FA1090 (University of Oklahoma). This gene, named ilvE in the sequenced N. meningitidis genomes, encodes for a putative branched-chain amino acid aminotransferase. Moreover, partial DNA sequencing upstream of commensal Neisseria licA revealed the presence of a gene with great similarity (≈ 84–87%) to a putative inner membrane hypothetical protein (NMA2149, Sanger Centre; NMB0338, TIGR Genome Project; and RNGO1565, University of Oklahoma), whose function has not yet been assigned. On the genome of pathogenic Neisseria, these two genes are separated from each other by a 174-nucleotide DNA region (NMA2150, Sanger Centre) of unknown function, whereas in commensal Neisseria, the lic locus is found between the two genes in place of the 174 bp nucleotide region. Therefore, it is unclear whether the entire locus was present in pathogenic Neisseria at one time and has been lost by deletion and replacement. In order to investigate whether other features associated with the expression and incorporation of ChoP in H. influenzae LPS were also present in commensal Neisseria, we carried out further studies, described below.

The genetic basis for phase variation of ChoP on LPS of commensal Neisseria

Chromosomal DNA extracted from various distinct and isogenic commensal Neisseria strains was used in a PCR reaction to isolate the 5′ end of the licA gene to determine the number of 5′-CAAT-3′ repeats within the gene. As predicted, by comparison with the H. influenzae licA gene, the ChoP-expressing strains and phase variants within a strain had a number of repeats (six or thirteen) permissive for the correct translation of the full-length licA gene product from one of the two potential ATG initiation codons α and β (Fig. 3B). In the case of strains C501 and C450, only the start codon α is in frame with the rest of the gene, whereas for both NL4 ChoP+ variants (H and L), only the start codon β can be used for the full translation of the licA gene. The ChoP strain NL2 and the ChoP phase variant of strain NL4 (variant O) had either seven or five repeats, in which there was no full-length translation product, because neither of the two upstream initiation codons were in frame with the rest of the licA gene (Fig. 3B). Thus, changes in the number of CAAT repeats within the licA gene are responsible for the phase variation of ChoP expression observed on commensal LPS (Serino and Virji, 2000).

Commensal Neisseria preferentially use exogenous choline for ChoP synthesis

We have shown previously that commensal Neisseria are able to incorporate radiolabelled choline from the growth medium into LPS-associated ChoP (Serino and Virji, 2000). However, H. influenzae strains also appear to use other choline-containing compounds for the same purpose (Fan et al., 2001). To evaluate whether such choline sources could also be used by commensal Neisseria, four strains (three ChoP-expressing strains, C619, C501 and C450, and one ChoP, NL2) were grown in a chemically defined choline-free medium (CDM) (Copley and Egglestone, 1983) supplemented with choline, phosphorylcholine or glycerolphosphorylcholine. The three ChoP-expressing strains were able to incorporate choline from the medium into LPS, as determined by Western analysis using an anti-ChoP mAb. Up to fivefold increase in ChoP expression was observed in C619 in the presence of free choline (Fig. 4). Similar results were also obtained with ChoP-expressing C501 and C450 (data not shown). A relatively small increase (twofold) was observed in the presence of the other two choline sources. In contrast, the ChoP NL2, with an out of frame licA gene, did not show incorporation of any of the choline sources into LPS, consistent with the role of LicA in the phosphorylation of choline.

Figure 4.

Western blot analysis to investigate the ability of commensal Neisseria to use distinct sources of choline. The presence of ChoP in proteinase K-treated bacterial lysates was detected using anti-ChoP mAb TEPC-15. Results of experiments with strains C619 (ChoP+) (lanes 1–4) and NL2 (ChoP) (lanes 5–8) are shown. Bacteria were grown in a chemically defined medium (CDM) supplemented with the same amount (36 μM) of choline (lanes 2 and 6), phosphorylcholine (3 and 7) and glycerophosphorylcholine (4 and 8). Lanes 1 and 5 contained no additions. Molecular size markers are shown in kDa. The numbers at the bottom refer to the densitometric analysis of the relative colour intensity (area in arbitrary units) determined by the NIH SCIONIMAGE program.

ChoP increases the susceptibility of commensal Neisseria to serum bactericidal activity

In previous investigations, we isolated phase variants of N. lactamica NL4 with none (O), low (L) or high (H) levels of ChoP expression. To determine the effect of ChoP on sensitivity to killing mediated by CRP, these phase variants of strain NL4 were compared for their ability to survive the bactericidal effect of human serum. The specific contribution of CRP to serum killing was determined by preincubating the bacteria for 20 min in the presence of purified CRP (5 μg ml–1) before the incubation with human serum. The survival of bacteria over 15 min in the presence of 10% immunoglobulin (Ig)-depleted human serum was compared with controls incubated with decomplemented serum. The addition of CRP had no significant effect on the serum killing of the ChoP (O) variant, but decreased the survival of the ChoP+ variants in a ChoP-dependent manner. The variant with high ChoP expression was the most sensitive (Fig. 5). The effect of CRP was shown to require the presence of calcium, consistent with the requirement of Ca2+ ions for CRP binding to the phosphorylcholine ligand (Volanakis and Kaplan, 1971; Shrive et al., 1996).

Figure 5.

Serum bactericidal assay showing the effect of ChoP on sensitivity to killing mediated by CRP. N. lactamica NL4 ChoP variants (O, L and H) were compared for their ability to survive 15 min incubation in the presence of purified Ig-depleted human serum after being preincubated with purified CRP (5 μg ml–1). The percentage survival represents the number of CFU in the presence of untreated serum compared with controls containing heat-inactivated serum. Standard deviations of triplicate experiments are shown. O, L and H represent strains expressing none, low or high levels of ChoP.

ChoP increases adherence and invasion of commensal Neisseria to human epithelial cells

As the ChoP moiety is central to recognition by PAF receptor on human epithelial and endothelial cells, the role of phosphorylcholine of commensal Neisseria in adherence to human epithelial cells was examined. Two respiratory epithelial cell lines, a human bronchial cell line (16HBE14) and human type II alveolar cells (A549) were used. These studies demonstrated that NL4 ChoP O variant was less adherent than the ChoP+ H variant (Fig. 6A). The effect of ChoP on adhesion to epithelial cells was confirmed by pretreating the bacteria with the anti-ChoP IgA mAb TEPC-15. Adhesion of the variant H decreased in the presence of anti-ChoP mAb to the levels observed for variant O.

Figure 6.

Figure 6.

Effect of ChoP on adhesion and invasion of lung epithelial cells.

A. Monolayers of human bronchial cell line 16HBE14 were infected with N. lactamica NL4 variants O (white columns) and H (black columns). and adhesion was determined after 3 h incubation. Hatched bars represent bacteria preincubated with anti-ChoP mAb for 20 min before infection of the cell monolayers. Percentage inoculum represents the portion of the inoculum that adhered to target cells. Standard deviations of triplicate experiments are shown within an experiment. The data represent one of several typical experiments.

B. Inhibition of adherence (a) and invasion (i) by PAF-Ra correlates with the presence of ChoP. Monolayers of 16HBE14 cells were pretreated with PAF-Ra L659,989 (5 nM) and infected with N. lactamica NL4 variant O (ChoP) and variant H (ChoP+). The effects of PAF-Ra on adherence (open bars) and invasion (black bars) are expressed as a percentage of untreated controls. (± standard deviations of triplicate experiments).

Figure 6.

Figure 6.

Effect of ChoP on adhesion and invasion of lung epithelial cells.

A. Monolayers of human bronchial cell line 16HBE14 were infected with N. lactamica NL4 variants O (white columns) and H (black columns). and adhesion was determined after 3 h incubation. Hatched bars represent bacteria preincubated with anti-ChoP mAb for 20 min before infection of the cell monolayers. Percentage inoculum represents the portion of the inoculum that adhered to target cells. Standard deviations of triplicate experiments are shown within an experiment. The data represent one of several typical experiments.

B. Inhibition of adherence (a) and invasion (i) by PAF-Ra correlates with the presence of ChoP. Monolayers of 16HBE14 cells were pretreated with PAF-Ra L659,989 (5 nM) and infected with N. lactamica NL4 variant O (ChoP) and variant H (ChoP+). The effects of PAF-Ra on adherence (open bars) and invasion (black bars) are expressed as a percentage of untreated controls. (± standard deviations of triplicate experiments).

The role of PAF receptor in adhesion and invasion of epithelial cells by commensal Neisseria strains was evaluated by testing the effects of a PAF receptor antagonist (PAF-Ra), L659,989 (Merck Research Laboratories). Pretreatment of the cell monolayers with 5 nM PAF-Ra decreased the invasion of the ChoP+ H variant by ≈ 60%. No significant inhibition of adherence or invasion was observed for the O variant in PAF-Ra-pretreated cells (Fig. 6B).

Susceptibility of commensal Neisseria to the antimicrobial peptide LL-37/hCAP18

A number of antimicrobial peptides are expressed by epithelial cells of the upper and lower respiratory tract. One of these peptides, LL-37/hCAP18, belonging to the cathelicidin family, has been reported previously to have a bactericidal effect on a number of mucosal pathogens (Turner et al., 1998). The presence of ChoP on the LPS of H. influenzae strains or on the teichoic acid of S. pneumoniae has been reported to augment the resistance of the bacteria to killing by LL-37/hCAP18 (Lysenko et al., 2000b).

In this study, the hypothesis that ChoP on commensal Neisseria LPS could also act to decrease killing by the antimicrobial peptide was tested on ChoP variants of N. lactamica. In addition, H. influenzae strains expressing or lacking ChoP on LPS were also used for comparison. The linear LL-37 peptide was synthesized chemically and used in bactericidal assays in a low-ionic-strength buffer (medium E), shown previously to optimize the antibacterial activity of LL-37 (Johansson et al., 1998). In contrast to previous reports on H. influenzae (Lysenko et al., 2000b) and as observed in our experiments with distinct H. influenzae strains, the expression of ChoP on commensal LPS did not confer increased resistance to the bactericidal effects of LL-37. Compared with the H. influenzae strains, both ChoP+ and ChoP variants of commensal Neisseria were considerable more sensitive to the peptide (Fig. 7).

Figure 7.

Effect of ChoP on susceptibility to the antimicrobial activity of LL-37/hCAP18. H. influenzae 7004 (dotted lines), ChoP+ (upper line) and ChoP derivatives (lower line) and commensal Neisseria NL4 (solid lines) ChoP+ (solid symbols) and ChoP (open symbols) variants are shown. Bacteria were incubated for 60 min at 37°C with LL-37/hCAP18 at the concentrations indicated, and the numbers of survivors were determined by plating on agar. Means of at least three independent determinations and standard deviations are shown.

Discussion

Phosphorylcholine appears to be a common feature on the cell surface of many mucosal pathogens, particularly those residing in the human respiratory tract, such as H. influenzae, S. pneumoniae and Neisseria spp.

The ChoP moiety has been associated with the virulence of mucosal pathogens and has gained much attention as a potential vaccine candidate. In this paper, we present the basis for ChoP incorporation in commensal Neisseria and its functional implications. The studies should be of importance in our understanding of mucosal ecology and in considering the moiety as a potential vaccine candidate.

In previous reports, the ChoP moiety was demonstrated on pili from N. meningitidis and N. gonorrhoeae as well as on LPS from commensal Neisseria as an epitope reacting with anti-ChoP antibodies (Weiser et al., 1998a), whose specificities were addressed and demonstrated later (Serino and Virji, 2000).

In this study, we first undertook to demonstrate the physical presence of the moiety on commensal Neisseria LPS to facilitate our major lines of investigation, i.e. the genetic basis and functional importance of ChoP in commensal Neisseria. Similar investigations in pathogenic Neisseria are currently in progress.

Using a two-step PCR technique, we sequenced an entire locus containing four genes (licA–D) with high homology to the lic1 locus of H. influenzae. The locus also shows homology to the S. pneumoniae lic locus but to a lesser extent. The commensal licA gene contains, at the 5′ end, CAAT repeats similar to the Haemophilus gene, responsible for the phase variation in ChoP expression, which are absent from the Streptococcus licA gene. The number of CAAT repeats determines whether one of the two potential initiation codons is in frame with the rest of the gene and, therefore, full translation results in ChoP expression. However, the different levels of ChoP expressed by isogenic variants of a commensal strain do not appear to depend on which start codon is used, as both NL4 variants H and L use the same initiation codon. Therefore, it is possible that, as with H. influenzae (Lysenko et al., 2000a), it is the location on distinct sites on LPS that leads to different levels of ChoP detection in distinct strains and variants. In addition, in commensal Neisseria, the arrangement of the lic genes is similar to that in H. influenzae but different from that in S. pneumoniae, in which the licABC genes are separated by two ORFs from two copies of the licD gene. The overall greater similarity of the commensal lic genes to those of Haemophilus perhaps suggests a common ancestry for these genes. However, sequencing of flanking regions at the 5′ end and 3′ end on the lic locus revealed no further homology to any H. influenzae genes. Instead, two genes identical to pathogenic Neisseria genes, closely positioned on the chromosome of N. meningitidis serogroup A strain Z2491, serogroup B strain MC58 and N. gonorrhoeae strain FA1090, were found. The lic genes are absent from the two N. meningitidis and one N. gonorrhoeae sequenced genomes, and no hybridization with a licA probe was observed with N. meningitidis strains of serogroups B and C and N. gonorrhoeae strains, all of which reacted strongly with anti-ChoP mAb (Serino and Virji, 2000).

In addition to the absence of the lic1 locus in pathogenic Neisseria, no incorporation of radiolabelled choline was observed into pili or any other structure in the pathogens (Serino and Virji, 2000). Therefore, a distinct genetic machinery is involved in the production and incorporation of ChoP on pathogenic pili, and current studies in our laboratories are investigating this pathway.

Haemophilus influenzae incorporates choline from its environment but is also able to use phosphorylcholine or glycerophosphorylcholine (GPC) to express ChoP (Fan et al., 2001). Commensal Neisseria also incorporate radiolabelled choline in the ChoP moiety on LPS (Serino and Virji, 2000). Here, we addressed the question whether they had the ability, like H. influenzae, to use alternative sources of choline. Our results suggest that choline is the main source for the production of ChoP, as choline is incorporated much more efficiently compared with phosphorylcholine or GPC. In addition, the use of these choline sources requires in frame translation of the putative choline kinase, licA, as no expression of ChoP is observed when the NL2 ChoP strain, with an out of frame licA gene, is incubated in the presence of either choline or phosphorylcholine. The observation that ChoP is expressed by commensal Neisseria strains even in the absence of added choline suggests a possible endogenous synthesis of choline by the bacteria. However, active transport of choline into the cell for its incorporation into ChoP on LPS is undoubted, as radiolabelled choline is incorporated and a fivefold increase in ChoP expression is observed after the addition of exogenous unlabelled choline.

Taken together, the studies suggest a high likelihood of the acquisition of the lic locus by commensal Neisseria from H. influenzae. It is possible that pathogenic Neisseria also contained the genes at some time, but clearly, there are no lic-like genes in N. meningitidis and N. gonorrhoeae. The possibility also remains that the transfer could have occurred from commensal Neisseria to H. influenzae. However, it should also be considered that a greater diversity of ChoP sources can be used by H. influenzae compared with commensal Neisseria, which may argue for the transfer of the lic locus (but not other genes) from H. influenzae to commensal Neisseria. Alternatively, H. influenzae could have developed further mechanisms for generating ChoP since the acquisition of the lic genes.

In further studies, we sought to investigate the biological role of ChoP in the context of nasopharyngeal colonization. To investigate whether the expression of ChoP would be beneficial for the bacteria in colonizing epithelial cells, we compared the adherence and invasion of human lung and bronchial epithelial cells by variants that differ in the expression of ChoP. The results of these experiments provide evidence that the expression of ChoP has a significant role in enhancing bacterial adherence and invasion. This finding is consistent with previous studies, which suggested the involvement of ChoP on pneumococcus and Haemophilus strains in the attachment to host cells by direct interaction with the PAF receptor (Cundell et al., 1995; Swords et al., 2000). PAF-R is found on a variety of endothelial and epithelial cell types, and the interaction with ChoP results in a series of host cell signalling events. The ability of the bacteria to interact with the PAF receptor, expressed on cells, via the ChoP moiety was corroborated further by the fact that pretreatment of epithelial cells with PAF receptor antagonists (PAF-Ra) inhibits bacterial invasion in direct correlation with ChoP expression. However, and in accordance with results obtained with H. influenzae (Swords et al., 2000), no significant inhibition of bacterial adhesion was obtained in PAF-Ra-pretreated cells. Taken together, these results suggest that the expression of ChoP confers a greater potential for cell association.

Conversely, ChoP is detrimental in some host niches, as it is the target of innate immunity based on the binding to CRP and subsequent activation of complement (Weiser et al., 1998b). Our previous studies clearly demonstrated that the level of CRP associated with the bacteria correlated directly with the level of ChoP expression on their LPS. Here, we examined the role of CRP in the antibody-independent complement activation by the classical pathway at a concentration (5 μg ml–1) that is within the range observed in inflammatory specimens (0.17–42 μg ml–1) (Gould and Weiser, 2001). We observed an in-creased sensitivity of the strains in direct correlation with increased expression of ChoP. An implication of our results is that the innate immunity mediated by CRP may be particularly important in protection from mucosal inhabitants, including the pneumococcus, some H. influenzae and Neisseria strains. Thus, against the colonizing potential of ChoP via the PAF receptor, the action of CRP via ChoP may exert several detrimental effects. CRP aids in opsonization and complement-dependent clearance mechanisms but, in addition, the presence of CRP on the mucosal surfaces could also block the interaction between ChoP-expressing bacteria and PAF-R on epithelial cells (Gould and Weiser, 2001). This also suggests that the ability of many organisms to switch off ChoP expression rapidly represents an important adaptation to environmental shifts encountered during the infection process.

The ability to use ChoP to evade the host immune response has also been considered for other organisms containing ChoP covalently attached to surface proteins. In particular, some filarial nematodes secrete phosphorylcholine-containing glycoproteins during parasitism of their host, and a number of immunomodulatory properties have been attributed to ChoP (Harnett and Harnett, 2001). Data have been produced revealing that ChoP-containing molecules are able to interfere with lymphocyte activation (Harnett and Harnett, 1999). It has also been demonstrated that the ChoP moiety induces interleukin-10 (IL-10) production in B1 cells of mice (Palanivel et al., 1996), which is directly associated with the development of Th2-type immune response and downregulation of Th1 response.

ChoP expression by H. influenzae has also been associated recently with increased resistance to host antimicrobial peptide killing (Lysenko et al., 2000b). It has been suggested that the cathelicidin peptide LL-37/hCAP18, expressed by many cells of the upper and lower respiratory tract, interacts with LPS and that the LPS structure affects sensitivity to antimicrobial peptides in general (Turner et al., 1998). In our studies, in contrast to previous observations on H. influenzae, ChoP did not appear to increase the resistance of commensal Neisseria to the antimicrobial peptide LL-37/hCAP18. For this reason, one might speculate that structural features of commensal Neisseria LPS differ from H. influenzae LPS, such that they influence the ability of the amphipathic peptide (perhaps because of altered surface charge) to insert effectively into the outer membrane.

In summary, our studies support the notion that ChoP decoration of LPS may contribute to the ability of bacteria to colonize and perhaps persist within the human respiratory tract by aiding epithelial invasion. In this res-pect, they add to the body of investigation that suggests the importance of the phosphorylcholine moiety on the surface of a number of phylogenetically divergent pathogens in colonizing similar host niches. The finding that similar mechanisms for ChoP biogenesis operate in commensal Neisseria and H. influenzae is consistent with the notion of a convergent evolution within these respiratory inhabitants. However, the similarity of their genomic structures also suggests that a horizontal exchange might have occurred, and that the acquisition and/or retention of the lic1 locus may confer an advantage in the colonization properties of commensal Neisseria.

Experimental procedures

Bacterial strains, media and chemicals

Commensal Neisseria strains used in these studies have been described previously (Serino and Virji, 2000) and included N. subflava C450, N. flavescens C619 and N. lactamica strains C501, NL2 and NL4. All Neisseria strains were grown for 16–18 h at 37°C with 5% CO2 in brain–heart infusion (BHI) broth supplemented with 5% heated horse blood. H. influenzae 7004 (type b) and its lic1 derivative were generously provided by Dr L. van Alphen and Dr J. Weiser respectively. H. influenzae strains were grown in BHI agar supplemented with Levinthal base (Virji et al., 1990). Chemicals were purchased from Sigma, unless otherwise specified.

Structural analysis of LPS

LPSs from strains expressing or lacking ChoP (C450 and NL2 respectively) were isolated from agar-grown bacteria by a modification of the hot phenol extraction procedure (Johnson and Perry, 1976). Briefly, cells were scraped from agar plates and suspended in 40 mM NaPO4–5 mM EDTA. The bacterial suspensions were treated first with lysozyme (6 mg g–1 wet weight of bacteria) for 16 h at 4°C, and then with 200 U of RNase One (Promega) for 3 h at 37°C. LPS was extracted from the cell lysates using one volume of prewarmed 80% phenol and stirred vigorously at 70°C for 15 min. The mixture was centrifuged at 10 000 r.p.m. for 20 min, and the aqueous phase was collected. A second extraction was performed as before, and the combined aqueous phases were precipitated overnight at –20°C using NaCl to 0.25 M and three volumes of 96% ethanol. After centrifugation at 10 000 r.p.m. for 30 min, LPS pellets were washed in 70% ethanol, vacuum dried and resuspended in the appropriate volume of sterile distilled water. A second RNase treatment was performed using 100 U of the enzyme, followed by a single hot phenol extraction and precipitation of the aqueous phase as described before. The final pellet was redissolved in a small volume of distilled water and analysed on a 16.5% tricine gel by silver staining.

The crude LPS samples were analysed by 1H-NMR spectroscopy. Spectra were recorded at 500 MHz on samples in D2O at 37°C on a Jeol Alpha 500 spectrometer with HOD (resonance: δ 4.75 p.p.m.) as an internal chemical shift standard.

DNA and PCR techniques

Neisserial chromosomal DNA was prepared by the CTAB method (Sambrook et al., 1989). All PCR amplifications were performed using a MiniCycler (MJ Research). PCR products were purified using a PCR purification kit (Amicon, Millipore).

Isolation and sequence analysis of unknown flanking DNA by a two-step PCR

Total genomic DNA from commensal Neisseria strains C450 and C501 was used as a template for PCR amplification of the 5′ region of the licA gene by a method for the isolation of unknown flanking DNA regions using magnetic strepta-vidin beads (Dynal). Standard reactions (50 μl volume) were carried out using 10–20 pmol of each primer and consisted of 35 cycles of denaturation at 94°C for 1 min, annealing at 62°C for 1 min and extension at 72°C for 1 min, followed by a 10 min extension at 72°C. Primers 1 and 2 were specific primers in the known DNA region, whereas the flanking primer (FP) was a partly degenerated primer, consisting of a combination of random nucleotides and a 3′ end of five fixed nucleotides (e.g. 5′-CAGTTCAAGCTTGTCCAGGAATTC NNNNNNNGGCCT-3′). The method included two PCR steps: during the first step, using primers FP and a biotinylated primer 1 (5′-CCGGGCAGATCCAGCAGGATA-3′), biotinylated fragments were amplified and isolated using magnetic streptavidin-coated beads. After separation of the DNA strands, the non-biotinylated strands were used as a template for a second PCR, using a nested specific primer 2 (5′-CGGCGTGGCGCGTGAGAAAAT-3′) and primer 3 (corresponding to the first 24 nucleotides of the non-degenerate 5′ end of primer FP). The amplified products for this PCR were gel extracted and sequenced using primer 2. This procedure was used to sequence the entire commensal lic locus, comprising the four genes licABCD, from the N. lactamica strain C501 and N. subflava strain C450, using primers designed from the sequences obtained. Searches for sequence homology were performed at the NCBI site (National Centre for Biotechnology Information) using the BLAST programs (Altschul et al., 1990). The genomic sequence databases of N. meningitidis serogroup A (strain Z2491), serogroup B (strain MC58) and N. gonorrhoeae strain FA1090 are available at the websites of the Sanger Centre (http://www.sanger.ac.uk), the Institute for Genomic Research (http://www.tigr.org) and the University of Oklahoma (http://www.genome.ou.edu/gono.html) respectively.

Incorporation of choline analogues into LPS

Bacteria were grown overnight in chemically defined medium (CDM) (Copley and Egglestone, 1983) lacking choline, at 37°C + 5% CO2, with the addition of choline, phosphorylcholine or glycerolphosphorylcholine at 36 μM final concentration. After growth, bacteria were collected, washed in PBS and an aliquot was removed for the determination of cell density. The cells were resuspended at the appropriate concentration in tricine sample buffer and treated at 100°C for 5 min. The lysates were treated with proteinase K at 0.5 mg ml–1 for 3 h at 64°C and then boiled before loading on the gel. LPS was separated on 16.5% tricine-SDS–PAGE gels (Invitrogen), transferred on Immobilon-P (Millipore) and immunoblotted with anti-ChoP mAb TEPC-15, as described previously (Serino and Virji, 2000). Densitometric measurements were determined using the NIH SCIONIMAGE program for Windows (Scion Corporation).

Serum bactericidal assay

Complement-mediated serum bactericidal assays of commensal Neisseria strains were performed using an Ig-depleted human serum (Sigma). Assays were performed with 25 μl of a suspension of microorganisms diluted to 2 × 105 colony-forming units (cfu) ml–1 with Dulbecco’s complete phosphate-buffered saline (PBSB), 10 μl of 10% human serum, 2.5 μl of purified CRP (0.2 mg ml–1), 1 μl of Ca2+ (1 M) and adjusted to 100 μl with PBSB. After a 20 min preincubation period of the bacteria in the presence of CRP, the mixtures were incubated for 15 min in the presence of Ig-depleted human serum as a source of complement. To calculate the percentage survival, colony counts were compared with controls in which complement had been inactivated by prior heating of the serum to 56°C for 30 min. Human CRP added to the bactericidal assays was purchased from Sigma Chemical and purified further by dialysis in PBSB to remove sodium azide.

Adherence and invasion assays

To determine the ability of N. lactamica ChoP variants to adhere to human epithelial cells, human type II alveolar cells A549 and immortalized bronchial epithelial cells 16HBE14 (available from the American Type Culture Collection) were used in adherence assays. A549 cells were propagated in Ham’s F12 medium (Sigma) supplemented with 10% (v/v) fetal calf serum (FCS) according to the supplier’s instructions. The bronchial cells 16HBE14 were cultured in Dulbecco’s modified Eagle medium (DMEM; Sigma), supplemented with 10% (v/v) FCS and 1%L-glutamine. The cells were seeded into 96-well plates (≈ 105 cells per well) and incubated at 37°C in 5% CO2 in air (v/v). The bacteria (≈ 5 × 107 per well) were incubated with the confluent monolayers for 3 h at 37°C in infection medium 199 (Sigma) supplemented with 2% FCS. After four washes with infection medium to remove non-adherent bacteria, the monolayers were lysed by the addition of sterile 1% saponin for 10 min at 37°C, and the cells were serially diluted in PBSB and plated on HBHI agar plates. After overnight incubation, the numbers of colonies, representing associated bacteria, were counted and used to estimate the percentage of adherent bacteria compared with the inoculum. Values represented the averages of three wells from at least three independent experiments. Bacterial invasion was estimated using a standard gentamicin survival assay (Virji et al., 1992). Cell monolayers were infected with bacterial suspension as described above. After 3 h incubation, the monolayers were washed three times in medium 199 supplemented with 2% FCS, and 250 μl of fresh medium containing gentamicin (200 μg ml–1) was added to kill extracellular bacteria. After 1.5 h of incubation, the monolayers were washed three times with fresh medium and lysed by the addition of sterile 1% saponin. The released bacteria were serially diluted and plated as described above.

For inhibition of the PAF receptor, cell monolayers were preincubated for 30 min in the presence of the PAF-R antagonist L659,989, which binds competitively to the active site of the PAF-R. PAF-Ra concentrations (5 nM) were maintained throughout the course of the experiments. PAF-Ra L659,989 was kindly provided by Merck Research Laboratories.

Bactericidal assays with LL-37/hCAP18

LL-37/hCAP18 was chemically synthesized as described previously and diluted from a stock of 1.0 mg ml–1 in water into a buffer of 0.01% acetic acid and 0.02% bovine serum albumin (BSA; Bals et al., 1998). For bactericidal assays, bacteria were diluted in medium E (0.8 mM MgSO4, 9.6 mM citric acid, 57.4 mM K2HPO4, 16.7 mM NaH4HPO4; Johansson et al., 1998), containing 0.1% gelatin to maintain bacterial viability, at a density of 105 cfu ml–1. Bacteria (90 μl) were then mixed with 10 μl of 0–15 μg ml–1 LL-37 in 0.01% acetic acid and 0.02% BSA, incubated for 60 min at 37°C, and the number of survivors was determined by plating on agar as above.

Acknowledgements

This work was carried out in the Spencer Dayman Meningitis Laboratories and was supported by the Medical Research Council and by the National Meningitis Trust. M.V. is an MRC Senior Fellow. We would like to thank Dr Martin Murray, Chemistry Department, University of Bristol, for performing the NMR analyses of LPS samples.

Ancillary