All Neisseria live in association with host cells, however, little is known about the genetic potential of nonpathogenic Neisseria species to express attachment factors such as pili. In this study, we demonstrate that type IV pilin-encoding genes are present in a wide range of Neisseria species. N. sicca, N. subflava, and N. elongata each contain two putative pilE genes arranged in tandem, while single genes were identified in N. polysaccharea, N. mucosa, and N. denitrificans. Neisserial pilE genes are highly diverse and display features consistent with a history of horizontal gene transfer.
The bacterial genus Neisseria includes human commensals, human pathogens, and species isolated from other animal hosts. Commensal Neisseria species are primarily found in the upper respiratory tract, which also can be colonized by the pathogen N. meningitidis, as well as other Neisseriaceae, including Eikenella and Kingella species . Commensal Neisseria also participate with a diverse array of organisms in the formation of oral biofilms . Horizontal gene transfer and recombination have been documented between different Neisseria species [3–7], and plasmid exchange has taken place among Neisseria, Eikenella, and Kingella species . In addition, the presence of exogenous DNA from commensal Neisseria stimulates the frequency of phase variation throughout the genome of N. meningitidis, and this may contribute to the transmissibility and invasiveness of the meningococcus .
Although all Neisseria species live in relationship with a host, studies investigating bacterial interactions with host cells have primarily focused on disease development and the pathogenicity of N. meningitidis and N. gonorrhoeae. Several surface components present on pathogenic Neisseria species have been identified as virulence factors including pili, LOS, porin, Opa, and Opc. Genes encoding these factors have been identified in some commensal Neisseria species [4,7,9–13], however, in many cases little is known about the genus-wide characteristics of these gene families.
Pili are filamentous surface appendages that facilitate the initial attachment of pathogenic Neisseria to host cells. Pili found on N. meningitidis and N. gonorrhoeae display subunit features typical of the type IV family of bacterial pilins, which are produced by numerous genera belonging to the β-, δ-, and γ-proteobacteria and also share similarities with archeal flagella [14–16]. Meningococcal and gonococcal pili also are known for their ability to exhibit extensive variability. Isolates of N. meningitidis express one of two structurally distinct classes of type IV pili, termed class I and class II . In addition to this distinction, meningococcal and gonococcal pili display further antigenic differences between strains, as well as phase and antigenic variation within a strain.
Comparisons of pilin-encoding expression loci, pilE, from multiple gonococcal and class I pilin-producing meningococcal strains show a highly conserved 5′ region, which is shared by all type IV pilin genes; a central semivariable region that encodes conserved elements designated SV1–SV5; and a 3′ region consisting of a hypervariable portion flanked by conserved elements, cys1 and cys2, which contain cysteine codons conserved among all type IV pilin genes (Fig. 2). These strains also possess several truncated silent loci, pilS, which lack promoters and 5′ sequences, but play a role in pilin antigenic variation by donating variant information to the pilE gene [18–22]. In contrast, the pilE genes from class II pilin-producing meningococcal strains diverge more extensively throughout their semivariable and 3′ regions, but retain conservation of the 5′ region, SV2, and cys2 elements. Class II pilin-producing meningococci also possess a reduced number of pilS loci [23–25].
We have previously characterized pilE genes from the human commensals N. lactamica and N. cinerea. These two species share a close taxonomic relationship with N. gonorrhoeae and N. meningitidis, and their pilE genes are similar to those of class II pilin-producing meningococci. In earlier studies DNA from more distantly related nonpathogenic Neisseria species failed to hybridize to probes based on gonococcal and meningococcal pilE sequences. However, a subsequent and more sensitive screen of the genus detected the presence of conserved 5′ pilin-like sequences in all commensal species, as well as a single species isolated from a nonhuman host [9,12,13]. In this study, we have extended these observations by characterizing pilE loci from a collection of more distantly related nonpathogenic Neisseria species and analyzing the extensive interspecies diversity displayed in this gene family.
2Materials and methods
2.1Bacterial strains and growth conditions
The bacterial strains used in this study are listed in Table 1. All isolates are type strains obtained from the American Type Culture Collection except N. sicca NRL 9305, which was from the collection of P.F. Sparling. Neisseria were grown at 37°C under 5% CO2 on GC medium base agar (Difco Laboratories, MD, USA) containing the supplements described by Kellogg et al. .
Table 1. Isolation of pilE genes from nonpathogenic Neisseria species
Source of isolation
pilE specific PCR primers
GenBank Accession No.
N. elongata ATCC 25295
N. mucosa ATCC 19696
N. polysaccharea ATCC 43768
N. sicca NRL 9304
N. subflava ATCC 49275
N. denitrificans ATCC 14686
Guinea pig oropharynx
2.2Pilin gene amplification
Neisserial chromosomal DNA was purified using the Qiagen genomic tip 20 (Qiagen Inc., CA, USA) and pilin genes were amplified by PCR using primers indicated in Table 1. The pilE locus from N. polysaccharea was amplified with primers homologous to upstream and 3′ internal regions of the published N. cinerea pilE locus using previously described standard PCR conditions . A vectorette PCR strategy was used to isolate pilin genes from the remaining species as previously described using the Vectorette II starter pack T and JumpStart REDAccutaq LA DNA polymerase (Sigma–Genosys, TX, USA). Briefly, a 102 bp region representing the conserved 5′ end of pilE was first amplified from each neisserial species using degenerate PCR primers . This region was sequenced and the data were used to design species-specific pilE primers (Table 1). A species-specific pilE primer was then used together with a vectorette-specific primer to amplify pilin-encoding loci from vectorette libraries constructed for each species according to the manufacturer's instructions. In some cases additional products were amplified by primer walking in order to extend the isolated region.
2.3DNA sequence analyses
PCR products were purified using the QIAquick PCR purification kit (Qiagen Inc.) and either sequenced directly or cloned into plasmid pCRII as directed in the TOPO TA cloning kit (Invitrogen, CA, USA). Plasmids were purified using the QIAprep spin miniprep kit (Qiagen Inc.). PCR products and recombinant plasmid inserts were sequenced at the Iowa State University DNA Sequencing and Synthesis Facility on a Prism 377 DNA sequencer (Applied Biosystems, CA, USA) using BigDye cycle sequencing reactions (Applied Biosystems). The complete DNA sequences of both strands were determined for each product. Sequence database searches utilized Pfam , the Conserved Domain Database , and BLAST . Sequence alignments were initially constructed with ClustalW  and then manually modified based on known structural features of pilin. Fractional G + C content was calculated with the EMBOSS program GEECEE (http://bioweb.pasteur.fr/seqanal/interfaces/geecee.html) and Jalview was used to calculate pairwise comparisons . A phylogenetic tree was estimated from the amino acid sequence alignment using the maximum likelihood method available in the TreePuzzle software package . The WAG model of amino acid replacement was utilized with a gamma distribution of among-site rate variation (parameters available on request). Southern blotting experiments were performed using the DIG high prime DNA labeling and detection starter kit II (Roche Diagnostics Corporation, IN, USA). DNA sequence data were deposited in GenBank under Accession Nos. DQ007932–DQ007937.
3Results and discussion
3.1Isolation of pilE genes from nonpathogenic Neisseria species
We used PCR-based approaches to isolate pilE genes from several Neisseria species that are human commensals, and from the guinea pig isolate N. denitrificans. Primers complementary to sequences conserved among previously studied neisserial pilE genes were used to isolate the closely related locus from N. polysaccharea. An asymmetric vectorette PCR strategy  was required to isolate the more diverse pilin-like sequences found in N. denitrificans, N. elongata, N. mucosa, N. sicca, and N. subflava. The region encoding amino acid +2 through the C-terminal cys1 element was isolated from all species and used for comparative analyses. In some cases, the gene isolation strategy also yielded sequences flanking the target region. Available upstream and downstream regions display possible promoter sequences, ribosome binding sites, and transcription termination sites. Several species also possess neisserial DNA uptake sequences  flanking their pilE genes. The chromosomal regions we have characterized are diagrammatically represented in Fig. 1 and the predicted amino acid sequences encoded by pilE loci located in these regions are shown in Fig. 2.
Putative pilE genes were identified in N. denitrificans, N. mucosa, N. polysaccharea, N. sicca, and N. subflava based on Conserved Domain Database searches and significant Pfam matches to entry PF00114 pilin (E values ranged from 1.2e-22 to 1.3e-5). Local alignment searches detected significant similarity (E values as low as 9e–32) between N. elongata sequences and known type IV pilin expression loci, however, this similarity did not extend over the entire length of the gene. Thus, it is presently unclear whether the pilin-like N. elongata sequence is a legitimate member of the neisserial pilE gene family. N. elongata appears to be a divergent member of the genus in analyses of other genes , and the N. elongata sequence we have identified is more closely related to pilE than to other neisserial pilin-like genes including pilV, comP, pilX, and pil H–L (data not shown). Confirmation of the functions of the putative pilE loci identified in this study will require additional analyses. Electron microscopy studies revealed the presence of pili on N. denitrificans, N. sicca, and N. polysaccharea (data not shown). The absence of pili from the remaining strains may be due to phase variation, deficiencies in pilus assembly functions, or other unknown factors.
3.2Features of predicted pilin proteins encoded by nonpathogenic Neisseria species
The putative pilE genes from nonpathogenic Neisseria species encode key features characteristic of type IV pilin proteins (Fig. 2). These include: a short leader sequence, conserved residues recognized by prepilin peptidases (−1 glycine, +1 phenylalanine, and +5 glutamic acid), a highly conserved hydrophobic N-terminal domain, and two C-terminal cysteine residues. The prepilin peptidase recognition sites are of particular interest. Dupey and Pugsley  demonstrated the presence and expression of pilD, which encodes prepilin peptidase, in all commensal Neisseria species they examined; including N. cinerea, N. lactamica, N. polysaccharea, N. mucosa, N. sicca, and N. subflava. Therefore, it appears these species have the genetic potential both to express pilin proteins and to carry out the subsequent cleavage and N-methylation steps required to produce mature pilin molecules.
The predicted pilin sequences encoded by the nonpathogenic Neisseria species examined in this study do not consistently display many features characteristic of previously identified neisserial pilin proteins (Fig. 2). Sites of post-translational modification found in some strains of N. meningitidis and N. gonorrhoeae are variably displayed by nonpathogenic Neisseria species. These include residues associated with glycosylation (+63 serine, all numbering is for N. gonorrhoeae) [20,35], phosphoethanolamine and phosphocholine modification (+68 serine) , and α-glycerophosphate modification (+94 serine) . The regions SV2 and cys2, previously noted for their conservation among N. gonorrhoeae, N. meningitidis, N. cinerea, and N. lactamica isolates, also are not absolutely conserved in the species examined in this study. All pilin sequences from nonpathogenic Neisseria species are missing the hypervariable portion found in the C-terminal region of gonococcal and class I meningococcal pilins. This hypervariable region is subject to positive selection and is likely exposed on the outer surface of the pilus fiber, where it would be available for interactions with host cells [20,38,39]. Interestingly, meningococci possessing class II pilin proteins that also lack this hypervariable region retain the ability to cause invasive disease in humans.
3.3Chromosomal arrangement of pilin genes in nonpathogenic Neisseria species
The N. sicca and N. subflava isolates examined in this study each possess two complete pilE genes arranged in tandem and separated by a short intergenic region. The pilin-like region characterized in N. elongata also contains two tandem loci. The presence of tandem pilE loci is atypical of previously studied pathogenic Neisseria species, however, it has been observed in the type IV pilin-producing genera Eikenella, Kingella, and Xanthomonas[40–42]. In addition, N. sicca and N. subflava share strong DNA sequence similarity in flanking regions, with 93.3% identity in the 200 bp immediately downstream of their pilE loci and 71.3% identity in the intergenic region. Both species also possess copies of the neisserial DNA uptake sequence at the end of the homologous 3′ flanking region. The N. sicca and N. subflava DNA sequences diverge substantially beyond this region and the two species possess different downstream genes (Fig. 1). The conservation of the intergenic and flanking regions suggests a history of horizontal gene transfer involving the pilE genes in N. sicca, N. subflava, and possibly an additional DNA donor. We searched for flanking region similarities with type IV pilin-encoding sequences in other genera, but found no evidence of recent intergenus transfer events with organisms currently represented in sequence databases.
We have isolated single pilE genes from N. mucosa, N. polysaccharea, and N. denitrificans in this study, and from N. lactamica and N. cinerea in previous work . Our gene isolation strategies did not yield extensive flanking sequences in these species, therefore we cannot rule out the possible presence of additional adjacent pilin genes in these organisms. However, we carried out Southern blotting experiments that indicated pilin-encoding sequences are localized to a single, relatively small region in the genomes of nonpathogenic Neisseria. Chromosomal DNA from each species displayed a single five to 16 kb chromosomal Cla I fragment that hybridized to a species-specific pilE probe (data not shown). Although more complete genomic characterization is required, the studies we have carried out have not presented evidence suggesting the presence of pilS loci in nonpathogenic Neisseria species.
3.4Comparisons among pilin sequences
The pilE loci of Neisseria species are highly diverse outside of the N-terminal encoding region. In order to construct an accurate alignment of the sequences for further comparative analyses, we excluded the N. elongata pilin-like sequences, which shared similarity with other Neisseria sequences at the 5′ end of the genes but could not be aligned with confidence in the remaining regions. Inclusion of amino acid sequences translated from published pilE sequences from N. gonorrhoeae, N. lactamica and N. cinerea; and the pilE and pilS loci from three strains of N. meningitidis[21–23,25] helped to refine the alignment, which was then used to infer a phylogenetic tree (Fig. 3). Generally, the relationships among pilE loci follow patterns seen for other neisserial gene families [5–7]. The pilE sequences from N. gonorrhoeae and class I pilin-producing N. meningitidis strains MC58 and Z2491 represent the most similar sequence group and share 83.0–87.7% predicted amino acid identity with one another. All meningococcal pilS genes also cluster within this homology group. The pilE locus of class II pilin-producing N. meningitidis strain FAM18 and those of the commensals N. lactamica, N. cinerea, and N. polysaccharea comprise a second related sequence group with 70.2–85.2% amino acid identity among species. Pilin genes from N. mucosa, N. denitrificans, N. subflava, and N. sicca are all more distantly related to the first two groups and exhibit 30.9–58.3% amino acid identity with one another. It is interesting to note N. denitrificans does not represent an outlying group in this analysis.
Comparisons of the N. sicca and N. subflava pilE loci reveal additional characteristics of the pilin gene family. Intraspecies comparisons show the two tandem genes in a single isolate are highly diverse and do not cluster together in the phylogeny (Fig. 3). Other type IV pilin-producing genera display similarly high levels of pilin sequence variability between tandem pilE loci [15,40–42]. Interspecies comparisons show 51.9–53.6% DNA sequence identity among the pilE coding sequences in N. sicca and N. subflava, while the intergenic and downstream sequences surrounding these loci are more closely related between the two species (see Section 3.3). It is also noteworthy that pilE genes in N. sicca, N. subflava, and other nonpathogenic Neisseria species possess lower levels of identity among predicted amino acid sequences than among corresponding DNA sequences (data not shown). These findings are consistent with a high rate of protein evolution, which has been previously demonstrated for meningococcal pilin genes .
Fractional G + C content values also vary among neisserial pilin genes. The pilE loci from N. gonorrhoeae and class I pilin-producing N. meningitidis strains display a G + C content of 0.53, which is similar to that of sequenced neisserial genomes [21,22]. The pilE genes of class II pilin-producing N. meningitidis strain FAM18 and the commensal Neisseria species isolated from the human pharynx display lower G + C contents ranging from 0.44 to 0.49. The G + C content of the presumably geographically isolated guinea pig strain, N. denitrificans, is 0.56.
Our data provide further evidence that class II pilin-producing meningococci may have acquired a pilE locus by horizontal gene transfer from a commensal species. The sequences upstream of pilE in the class II pilin-producing N. meningitidis strain FAM18 are 100% identical to the short region we have characterized upstream of N. polysaccharea pilE (Fig. 1). The pilE sequence from strain FAM18 also clusters with commensal pilin genes in both the phylogentic and G + C content analyses, while the FAM18 pilS genes cluster with gonococcal and class I meningococcal pilin genes in the phylogeny. Genome sequencing has shown the pilE and pilS loci are located in separate regions of the FAM18 chromosome, while these loci are adjacent in the class I pilin-producing meningococcal strains MC58 and Z2491 [21,22]. Strain FAM18 appears to have undergone a deletion of the chromosomal region that contains pilE in class I pilin-producing meningococci, and the class II pilE locus of strain FAM18 is located on an insertion this strain possesses relative to other sequenced meningococcal genomes (Stephen Bentley, personal communication). Thus, it appears fully virulent strains of meningococci may possess pilin sequences derived from commensal Neisseria species.
In conclusion, we have isolated one or two pilE loci from representative strains of several species of nonpathogenic Neisseria. Characterization of these genes and comparison with meningococcal and gonococcal pilE and pilS gene sequences showed that pilin loci comprise a diverse gene family, with high rates of protein evolution and a history of horizontal exchange among species. Further studies of a larger dataset will determine the extent of pilE sequence diversity within as well as between Neisseria species and help further elucidate the evolutionary relationships among the pilin genes of these species.
We thank Heidi Jordheim and Rebecca Rameden for assistance with this study, and Claudia Hagen for assistance with manuscript preparation. This work was supported by the Howard Hughes Medical Institute through the Undergraduate Biological Sciences Education Program. The sequence database for the serogroup C, ST11 complex N. meningitidis strain FAM18 was produced at the Sanger Institute and can be obtained from http://www.sanger.ac.uk/Projects/N_meningitidis/seroC.shtml.