INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Haemophilus influenzae is a fastidious, gram-negative bacterium that is commonly found as a commensal organism in the human nasopharynx (28). H. influenzae is characterized as encapsulated (possessing one of six chemically and immunologically distinct polysaccharide capsules, i.e., types a to f) or nonencapsulated (i.e., nontypeable H. influenzae). Invasive infections, such as bacteremia, cellulitis, septic arthritis, and meningitis, occur in nonimmune hosts and are usually caused by organisms possessing the type b capsule. In children, immunocompromised individuals, and individuals with underlying pulmonary disease (e.g., cystic fibrosis, chronic bronchitis, and chronic obstructive pulmonary disease), H. influenzae can cause localized respiratory infections, such as otitis media, sinusitis, conjunctivitis, and pneumonia, and acute exacerbations of chronic lung diseases (16, 28, 30, 34).

Colonization of the upper respiratory tract is an essential step in the pathogenesis of H. influenzae disease and is a likely target for therapeutic intervention. Both typeable and nontypeable H. influenzae organisms have been shown to adhere to cultured epithelial cells and human nasopharyngeal tissues (33). One of the cell surface molecules shown to mediate attachment to epithelial cells is the polymeric hemagglutinating pilus found on both typeable and nontypeable H. influenzae (15).

Five genes (hifA, hifB, hifC, hifD, and hifE) are required for the synthesis of mature H. influenzae pili, and they are located on an approximately 6-kb chromosomal locus (15, 26, 40). hifA encodes the major pilin subunit and lies on one end of the pilus gene cluster (26, 40). The HifA pilin is approximately 24 kDa and comprises the primary structural component of the shaft of the mature pilus (9, 27, 35). The hifA pilin genes of 11 H. influenzae strains, including 5 type b strains and 6 nontypeable strains (including 2 H. influenzae biotype aegyptius strains), have been cloned and their nucleotide sequences have been determined in earlier studies by several investigators (3, 10, 12, 20, 22, 37, 39, 43).

Immunologic characterizations of intact H. influenzae pili and the HifA pilins have been complicated by the fact that intact pili are highly conformational and are immunologically diverse while denatured pilins are immunologically homogeneous (11, 13). Further, polyclonal antisera raised against native pili from type b strains Eagan and M43 bind to homologous piliated type b H. influenzae but do not bind to homologous denatured HifA pilins, suggesting that epitopes defined by these sera may be assembled by protein folding or by protein-protein interactions and are not available on denatured pilins (13). Similarily, polyclonal antisera raised against pilins of strains M43 and Eagan do not bind to intact pili of the homologous strains (11, 13).

The two goals of this work were (i) to identify differences in the HifA sequences from several different typeable and nontypeable H. influenzae isolates that might explain the pilus immunologic heterogeniety and (ii) to identify sequence similarities that might relate to functional importance in bacterial pilus assembly. To do this analysis, the hifA genes from 26 typeable and 16 nontypeable strains were amplified by PCR and subjected to restriction fragment length polymorphism (RFLP) analysis with AluI and RsaI. Six different RFLP patterns were displayed with this analysis, and four more patterns were revealed from the nucleotide sequences of cloned hifA genes. Cloning and sequencing of representative hifA genes from each of the six RFLP patterns were performed and used for further analysis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and growth conditions. The H. influenzae strains used in this study are presented in Table 1. Except for strains AAr108 and AA61, the strains listed were isolated from individuals in a variety of geographical areas over a number of years and thus probably represent different bacterial clones. Strains AAr108 and AA61 were isolated from a mother and her son and may be the same strain. Bacterial strains designated AA and AAr were obtained from the clinical laboratories at the University of Michigan from 1983 to 1988, while strains designated M and Mr were obtained from the clinical laboratories at the University of Minnesota from 1979 to 1982. Bacterial strains were grown on Levinthal agar (37 g of brain heart infusion broth [Difco Laboratories, Detroit, Mich.], 18 g of Bacto agar [Difco], 2,000 µg of NAD [Sigma Chemical Co., St. Louis, Mo.], and 2,000 µg of hemin [Sigma] in 1,000 ml of deionized water) at 37°C with 5% CO₂ for 18 to 24 h (13). The H. influenzae strains were classified by using type-specific anticapsular antisera (for types a to f [Difco]) in a slide agglutination test.

Isolation of genomic DNA from H. influenzae. Genomic DNA was isolated from H. influenzae either by a modification of the Marmur procedure (25, 42) or by the Wizard genomic DNA purification kit (Promega, Madison, Wis.).

Amplification of hifA from H. influenzae by PCR. PCR was used for the amplification of hifA from H. influenzae genomic DNA. Primers used were based on the conserved 5' and 3' regions of six hifA genes from strains Eagan, M43, AM30, 86-1249, 86-0295, and 81-0384 (3, 10, 12, 20, 22, 37, 39, 43). The primer sequences were derived from the 5' and 3' regions in the hifA gene that show significant nucleotide sequence identity among the six H. influenzae pilin sequences. The nucleotide sequences of these primers were 5'-ATGAAAAAAACACT(AT)CTTGGTAGC-3' and 5'-TTAT(CT)CGTAAGCAATT(GT)GGAAACC-3'. Fifty nanograms of H. influenzae genomic DNA was mixed with 20 pmol of each primer and 45 µl of PCR SuperMix (Gibco BRL) to a final volume of 50 µl and overlayed with mineral oil (Sigma). The final PCR mixture contained 20 mM Tris-HCl (pH 8.4), 50 mM KCl, 1.5 mM MgCl₂, 200 µM (each) deoxynucleoside triphosphates (dATP, dCTP, dGTP, and dTTP), and 1 U of recombinant Taq DNA polymerase along with the H. influenzae genomic DNA and primers. The published error frequency for Taq DNA polymerase ranges from 1.1 × 10⁴ errors/bp to 8.9 × 10⁵ errors/bp (2, 38). The mixture was first incubated for 1 min at 95°C and then for 35 cycles of 95°C for 1 min, 50°C for 1 min, and 72°C for 2 min, followed by a final elongation step for 3 min at 72°C in a model PTC-100 programmable thermal controller (MJ Research, Inc., Watertown, Mass.). After amplification, samples were separated on 1% agarose gels and bands were visualized after staining with ethidium bromide (Sigma) and illumination by UV light. Molecular weight markers were run to estimate PCR fragment sizes (50- and 100-bp ladders [Gibco BRL]).

RFLP analysis of hifA PCR fragments. Amplified hifA PCR fragments were digested with restriction endonucleases AluI and RsaI according to the manufacturer's directions (Gibco BRL). Digestion products were resolved on 1% agarose gels and were visualized on a UV transilluminator after ethidium bromide staining. Molecular weight markers were run to estimate AluI and RsaI fragment sizes (50- and 100-bp ladders [Gibco BRL]). H. influenzae strains were grouped according to the hifA AluI and RsaI digestion patterns displayed after agarose gel electrophoresis.

Cloning of representative hifA genes. Two representative strains from each RFLP group were selected for cloning and sequencing of their hifA genes (Table 1). Amplified hifA fragments were electrophoresed on a preparative 1% agarose gel and purified with the GeneClean II kit (Bio 101, Inc., La Jolla, Calif.). The purified hifA PCR fragments were ligated into the SrfI site of pCR-Script Amp SK(+) after generation of blunt ends with a PCR Polishing Kit (Stratagene Cloning Systems, La Jolla, Calif.) and transformed into competent E. coli DH5 (Gibco BRL). Transformants were selected on LB agar containing 100 µg of ampicillin per ml and 40 µg of X-Gal per ml. Plasmid DNA from putative transformants was isolated with Qiagen Minipreps (Qiagen, Chatsworth, Calif.). BamHI/NotI (Gibco BRL) double digests were used to confirm DNA inserts in the isolated recombinant plasmids.

Sequencing of cloned representative hifA PCR fragments. Cloned representative hifA genes were sequenced at the University of Michigan Medical School DNA Core Facility with an Applied Biosystems model 373A automated sequencer (Applied Biosystems, Inc., Foster City, Calif.). Sequencing primers were purchased from Stratagene (M13 20 and reverse primers) and synthesized at the University of Michigan Medical School DNA Core Facility with the hifA DNA sequences from the representative cloned genes. DNA and protein sequences were analyzed with Lasergene Biocomputing software for the Macintosh from DNASTAR, Inc. (Madison, Wis.) and the Wisconsin Package, version 9.0, from the Genetics Computer Group (GCG) (Madison, Wis.) (5).

Nucleotide sequence accession numbers. The GenBank accession numbers for the hifA DNA sequences and the derived protein sequences determined in this study are as follows: AAr176, AF020908; AAr49, AF020909; 1712MEE (LB5), AF020910; ATCC 9007, AF020911; AAr73, AF020912; AAr160, AF020913; AA18, AF020914; AAr32, AF020915; ATCC 9006, AF020916; AAr157, AF020917.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

PCR amplification and RFLP analysis of hifA from typeable and nontypeable H. influenzae. The hifA genes from representative typeable (n = 26) and nontypeable (n = 16) H. influenzae (Table 1) were amplified directly from genomic DNA. The resultant hifA PCR products were approximately 650 bp in length (data not shown). H. influenzae Rd, which lacks the hif gene cluster (8), was used as a negative control and did not yield PCR products with the hifA primers.

View larger version (89K): [in this window] [in a new window]   FIG. 1.   AluI and RsaI RFLP analysis of PCR amplified hifA from genomic DNAs of representative H. influenzae strains. (A) hifA samples digested with AluI. (B) hifA samples digested with RsaI. Lanes: 1 and 14, 100-bp ladder; 2, Eagan (E1a; LKP3); 3, AAr176; 4, 1712MEE (LB5); 5, AAr49; 6, M43; 7, ATCC 9007; 8, AAr73; 9, AAr160; 10, AAr32; 11, AA18; 12, ATCC 9006; 13, AAr157.

Correlation between the hifA RFLP grouping of strains and immunoreactivity with antipilus polyclonal antisera. Gilsdorf et al. (13) have shown that polyclonal antisera raised against intact pili from H. influenzae type b strains Eagan and M43 each reacted with a different subset of 22 piliated type b isolates. Table 2 presents representative type b strains reactive with each of the antipilus sera and representative nontypeable strains that do not react with either antipilus serum (13). Those type b strains that demonstrate greater reactivity with antiserum directed against strain M43 pili (i.e., 4⁺) tend to fall into RFLP pattern 3, while those type b strains that react equally (i.e., 2⁺) or greater (2⁺ to 4⁺) with antiserum directed against strain Eagan pili tend to fall into RFLP pattern 1 (² = 6.875; 0.02 < P  0.05). Strain AAr103p⁺ is the exception, in that it shows greater reactivity with anti-Eagan pilus serum and yet falls into RFLP pattern 3 (Table 2).

Sequence analysis of representative hifA genes. In order to confirm the validity of the RFLP analysis and to explore sequence differences between each RFLP group, we chose 10 different representative hifA genes to clone and sequence from the H. influenzae strains in Table 1 to complement the existing hifA sequence database. The strains chosen were from several different sources and represent the six RFLP patterns defined in this study.

View larger version (51K): [in this window] [in a new window]   FIG. 2.   Comparison of predicted amino acid sequences of HifA from representative H. influenzae strains. Identical residues throughout all HifA sequences are in boldface type and are shown in the bottom line as "Consensus." Arrowheads denote the conserved cysteine residues; asterisks denote the positions of the conserved tyrosine and glycine residues 2 and 14 amino acids, respectively, from the C terminus. Hydrophilic regions I, II, and III are underlined (10), and pilin motifs "Segment S3" and "Segment S6" (17) are in shaded boxes. The comparison was performed with the Pileup program of the Wisconsin Package, version 9.0, from GCG (5).

View larger version (40K): [in this window] [in a new window]   FIG. 3.   Dendrogram showing the relationships of the predicted amino acid sequences of HifA from the 19 representative H. influenzae strains. The dendrogram was generated with the Pileup program of the Wisconsin Package, version 9.0, from GCG (5). NT, nontypeable H. influenzae.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The goal of this work was to compare the hifA genes from several independent typeable and nontypeable H. influenzae isolates and identify any sequence differences that might explain the pilus immunologic heterogeneity. Further, these hifA genes were compared to pilins from other bacteria to identify conserved regions potentially important for pilus assembly.

The relationships of H. influenzae strains based upon hifA RFLP analysis (Table 1; Fig. 1) were confirmed by analysis of the derived amino acid sequences (Fig. 2 and 3; Table 3). All of the strains contain highly conserved 18- to 20-residue signal sequences (Fig. 2) and several regions of high sequence identity. Of note are the equally spaced pairs of cysteine residues at positions 45 and 85 in all of the HifA pilin sequences (Fig. 2). This Cys-Cys loop is conserved among pilins from several different bacteria, is postulated to play a role in the maintenance of protein structure, and is thought to be a dominant immunogenic epitope in the PapA pilins of uropathogenic E. coli (4, 17). The amino acids contained within the region of the Cys-Cys loop (residues 58 to 121 [Fig. 2]) of H. influenzae M37 were demonstrated by Palmer and Munson, Jr. (31), to possess a significant part of the epitope defined by the pilin-specific monoclonal antibody 3H12, emphasizing the immunogenic potential of this region in H. influenzae pilins.

The region in HifA from amino acids 156 to 205, which is the most variable region within the pilin sequences, is analogous to the variable region found in the PapA pilins of uropathogenic E. coli and the pilins from other bacteria and may account for the immunologic diversity in H. influenzae pili (4, 17). For PapA, strain-to-strain differences in the variable region and the Cys-Cys loop are thought to constitute the basis for the serological diversity of these pili (4).

In an effort to identify common regions of type b pili that are surface exposed and represent antigenic epitopes, Forney et al. (10) analyzed the hydrophilicity of the pilin proteins expressed by the type b strains M43 and Eagan. They identified three hydrophilic regions within the HifA sequence (regions I, II, and III [Fig. 2]) and proposed that these regions might constitute conserved antigenic epitopes. The present study shows that these three regions are highly conserved within all HifA sequences (Fig. 2), with 10 of 19 residues being absolutely conserved in region I. Further, 10 of 18 and 4 of 12 residues are absolutely conserved in regions II and III, respectively, in the HifA comparison (Fig. 2). Along with the absolutely conserved amino acids in each region, several conserved amino acid substitutions result in high degrees of sequence similarity in these regions. To determine if these regions contain surface-exposed, immunogenic epitopes, Gilsdorf et al. (14) constructed 14- to 15-amino-acid peptides corresponding to regions I, II, and III and raised polyclonal rabbit antisera to these peptides. Sera to these peptides demonstrated poor to no reactivity to native pili and moderate to strong reactivity to denatured pili, suggesting that the epitopes determined by these peptides are not present on assembled pili in a conformation that can be recognized by the antipeptide antibodies (14). These results were supported by the previous observation that antibodies raised against denatured pilin and an internal peptide of strain M43 HifA recognized epitopes on denatured pilins of both type b and nontypeable H. influenzae better than native pili on the same strains (11, 13). Therefore, although these regions are highly conserved on denatured pilins, they are not available for antipeptide or antipilin antibody binding on native pili.

Several conserved features characteristic of pilus proteins assembled by E. coli PapD-like molecular chaperones are seen in the C-terminal sequences of the HifA pilins (Fig. 2). For example, the derived HifA sequences have strong amino acid homology to one another in the C terminus (Fig. 2) and contain absolutely conserved tyrosines and glycines at 2 and 14 residues, respectively, from the C terminus (3, 10, 19, 21, 22, 37, 39, 43). These conserved HifA sequence features are shared with the minor pilin HifD and the C terminus of the putative adhesin, HifE (26, 40). Recently, HifB-HifA and HifB-HifD chaperone-pilin complexes have been isolated, demonstrating that the biogenesis of H. influenzae pili is dependent upon the periplasmic chaperone HifB (36).

Recently, Girardeau and Bertin (17), using two-dimensional sequence analysis, described other conserved markers of the bacterial pilin family; these features are conserved within the representative HifA sequences (Fig. 2). Among the motifs identified, segments S3 (FxlxLxxC [where x is any residue]) and S6 (Ax[G/N]VGVQi [where i is a hydrophobic residue]) were the most conserved. Girardeau and Bertin (17) suggest that the S3 and S6 motifs, along with the conserved Cys-Cys loop and C-terminal homology, play a role in the function or maintenance of the structural integrity of the protein.

The intact pili of H. influenzae demonstrate a high degree of immunologic heterogeneity with both polyclonal and monoclonal antipilus sera (1, 11, 13, 23, 31). Brinton et al. (1) used polyclonal anti-LKP pilus sera to differentiate clinical isolates of H. influenzae into seven different LKP pilus types (LKP1 to LKP7). Four LKP serotypes are represented in this study [LKP3, Eagan (E1a); LKP4, 86-1249; LKP1, 86-0295; LKP5, 81-0384] and each displays a different RFLP pattern (Table 1).

Polyclonal antipilus sera raised against the pili of type b strains Eagan and M43 each reacted with a different subset of 22 type b H. influenzae strains (13, 23). Those type b strains showing equal or greater reactivity with Eagan antipilus serum displayed hifA RFLP pattern 1, while those strains showing greater reactivity with M43 antipilus serum displayed hifA RFLP pattern 3 (Table 2). Several strains, though, show reactivity with both antipilus sera, suggesting that epitopes defined by these antisera are shared by some strains and not others. These findings support those of Denich et al. (4), who found common immunogenic domains among PapA pilins from different strains of uropathogenic E. coli.

The amino acid sequences of representative HifA pilins from RFLP patterns 1 and 3 range between 78 and 81% identity (Table 3). Further, strain AAr103p⁺ reacts more strongly with Eagan antipilus serum yet displays hifA RFLP pattern 3, to which strain M43 belongs (Table 2). This result demonstrates that although the HifA pilin sequence of strain AAr103p⁺ has an overall amino acid sequence identity closer to that of M43, it shares certain critical antigenic residues with that of Eagan. Few amino acid differences between HifA pilins could explain the varying reactivities with antipilus antisera. For example, H. influenzae AAr176 bacteria do not react with Eagan antipilus serum (Table 2) (11). This serum appears to be specific for the Eagan HifA pilin and decorates the entire pilus shaft on whole bacteria subjected to immunoelectron microscopy (9). The derived HifA sequences from strains Eagan and AAr176 are 99% identical and differ by only three amino acids (residues 43, 217, and 221) (Table 3; Fig. 2).

The first amino acid difference (residue 43 [Fig. 2]) lies within the predicted hydrophilic region I originally identified by Forney et al. (10) and near a pair of conserved cysteines that define the Cys-Cys loop that may represent a major surface-exposed antigenic region in HifA. The second pair of amino acid differences between the Eagan and AAr176 HifA sequences (residues 217 and 221) are at the C terminus, a region that tends to be conserved among pilins and is thought to play a role in pilus subunit interactions (17, 21, 26). However, Palmer and Munson, Jr. (31), observed that monoclonal antibody 3H12 reactivity with M37 HifA was enhanced by the addition of M37 C-terminal sequences to the M37-MinnA HifA chimeras, suggesting that this region may itself be antigenic or that amino acids in this region contribute to nonlinearly assembled epitopes.

The rationale for the immunological heterogeneity of H. influenzae pili is not well understood. Limited studies have shown that humans can produce serum antibodies directed against H. influenzae pili (7, 32). Further, H. influenzae has been shown to undergo pilus phase variation and antigenic variation in such cell surface molecules as major outer membrane protein P2, immunoglobulin A1 proteases, and lipopolysaccharide (6, 18, 24, 29, 41). Taken together, the antigenic diversity of H. influenzae pili may be due to small changes in immunodominant surface-exposed epitopes in HifA and may play a role, along with phase variation and antigenic drift of other surface molecules, in the organism's ability to evade the host immune system.