Main

Although Haemophilus influenzae normally inhabit the respiratory tracts of humans with no evidence of disease, they are also responsible for a spectrum of human infections. Organisms possessing the type b capsule may cause serious, invasive infections, such as meningitis, bacteremia, or septic arthritis, in nonimmune children. Nontypeable organisms that lack a polysaccharide capsule, on the other hand, are associated with localized respiratory infections, such as otitis media, sinusitis, and bronchitis, in both children and adults.

The first event in the pathogenesis of H. influenzae infections is colonization of the nasopharynx, which depends upon adherence of the bacteria to cells of the respiratory tract. Among the several H. influenzae adherence factors that have been postulated, pili, which are large, polymeric surface proteins, have been characterized most completely. These pili, which mediate hemagglutination, are found on both type b and nontypable organisms and the pilins (major structural subunits of pili) from type b and nontypable strains are immunologically cross-reactive(1). The role of pili in H. influenzae adherence to human cells has been commonly defined using shed buccal epithelial cells(1–3), although HEp-2 tissue culture cells(4), cells from organ culture preparations(5–7), and intact tissue sections have also been used(8).

In addition to adherence to epithelial cells, invasion of respiratory tissues and bacterial internalization by respiratory cells may also play a role in pathogenesis. Using electron microscopy to document invasion, Farleyet al.(5) showed that nonpiliated type bH. influenzae invaded human adenoidal organ cultures, as evidenced by the appearance of bacteria between damaged mucosal cells, better than piliated organisms. These authors also showed, in another study(9), that Brazilian purpuric fever strains of nontypeableH. influenzae (which possess pili) and non-Brazilian purpuric fever strains (whose piliation status was not defined) were internalized by human adenoidal cells in organ culture but not Chang conjunctival cells. Using gentamicin to kill extracellular, noninternalized organisms, St. Geme and Falkow(10) showed that nonpiliated, nontypeableH. influenzae were internalized by Chang conjunctival cells; piliated strains were not tested.

The goal of this study was to investigate the role of pili in the adherence of H. influenzae type b and nontypeable H. influenzae to cells from a variety of different respiratory tissues. Pilus-mediated adherence was assessed by testing piliated organisms, nonpiliated phase variants, and nonpiliated mutants possessing insertionally inactivated pilin genes and confirmed by the ability of the ganglioside GM1, which as been shown by Van Alphen et al.(3) to be a H. influenzae pilus receptor analog, to inhibit adherence. In addition, we sought to investigate the role of pili in invasion of human respiratory cells.

METHODS

Bacteria. The piliated and nonpiliated variants of H. influenzae used in this study have been described previously(1, 2). Strains E1a and M43 are type b strains (as determined by smooth phenotype and reactivity with type b antisera) and Mr13 and AAr176 are nontypeable strains that possess a rough phenotype and do not react with antisera directed against H. influenzae capsular types a through f. E1ap/mut is an isogenic mutant of H. influenzae type b strain E1a with the pilin structural gene insertionally inactivated(11). All bacteria were stored and grown on Levinthal agar or broth [brain heart infusion agar or broth (Difco) supplemented with 100 μg/mL hemin and 20 μg/mL NAD) as previously described(12). For use in the adherence assays, the bacteria were confirmed to be piliated or nonpiliated by hemagglutination and then biotinylated using the methods of Forney et al.(13).

Epithelial cells. Table 1 lists the epithelial cells used in these studies. 16HBE14o- cells, kindly provided by D. Gruenert, University of California, San Francisco, are simian virus 40-transformed ciliated bronchial epithelial cells obtained from a normal child(14). The nasal cells (which were obtained from ATCC and have not been well characterized) showed a positive reaction with preliminary studies using AE1/AE3 anti-keratin antibody (Boehringer Mannheim Biochemicals, Indianapolis, IN), with negative reaction of tracheal fibroblasts (7285 and 7291 cells) as negative controls and positive reaction of HEp-2 cells as positive controls. These results indicate that the nasal cells contain keratin in their cytoskeleton and are epithelial in origin. The cell lines were maintained in Dulbecco's modified Eagle's medium, Sigma Chemical Co., St. Louis, MO (CCL30, mink cells, 16 HBE-140-, HeLa, and HEp-2) or Ham's-F-12, Sigma (A549, 7285, and 7291) with 10% fetal bovine serum and 10% antibiotic/antimycotic mixture (100 U of penicillin, 100 μg of streptomycin, and 0.5 μg of amphotericin B/mL) at 37° in 5% CO2. The cells were fed three times weekly with either Dulbecco's modified Eagle's medium or Ham's F-12 and passaged every 7 d. For use in the adherence assays, cells were plated into 96-well, flat bottom polystyrene plates and grown to confluency.

Table 1 Epithelial cells used in the adherence assays
Full size table

Buccal epithelial cells were prepared as described previously(15).

Adherence assay. In our earlier adherence assays(1), epithelial cells were applied to microtiter plate wells that had been activated with lysine and gluteraldehyde, and the adherent bacteria were detected by immunoassay using rabbit antisera that contains antibodies cross-reactive with H. influenzae. In this study, the mink cell assay and its control buccal cell assay were performed using this technique.

Subsequent studies in our laboratory revealed that the variability of cell adherence to the microtiter plate wells was significantly reduced by fixing the cells to the wells with 1% glutaraldehyde(16). The remainder of our assays have used this technique as follows: 100 μL of 1% gluteraldehyde in PBS were added to each well containing the epithelial cells prepared as above and incubated at room temperature for 1 h. The wells were then rinsed twice with 0.2% Triton X-100 (Sigma) to rehydrate the cell membranes.

Before adding bacteria, the wells were blocked with 200 μL of PBS containing 2% BSA for 1 h at 37°C. Fifty microliters of PBS with 0.1% gelatin, containing 1 × 107 cfu of biotinylated bacteria(A620 = 0.16) were added and incubated at room temperature for 1 h. The plates were then washed three times with 200 μL of PBS. Fifty microliters of ExtrAvidin-peroxidase (Sigma), diluted 1:2500 in PBS, were added to each well and incubated at room temperature for 1 h. After rinsing three times with PBS, the quantity of bacteria adherent to the cells was determined by adding 50 μL of enzyme substrate [34 mg ofo-phenylenediamine (Sigma)] in 100 μL of 0.1 M citrate phosphate buffer (pH 5.01) with 10 μL of 30% H2O2 to each well. The reaction was terminated after 5 min by the addition of 50 μL of 4 N H2SO4. The intensities of the reactions were determined atA490 nm on an automated ELISA spectrophotometer (Dynatech). Control wells contained epithelial cells with no bacteria. Preliminary studies showed very low level or no binding of bacteria to wells containing no epithelial cells. Incubation of organisms with epithelial cells for at least 6 h did not result in a change in their piliation phenotype. Each variable was tested using two to six wells in two to four different assays.

GM1 interference assay. To test the ability of GM1 to interfere with the adherence of piliated and nonpiliated bacteria to the cells, the ganglioside GM1 (Sigma) as added to biotinylated bacteria to a final concentration of 25 μg/mL and incubated at 37° for 1 h. Using the GM1-treated bacteria, the adherence assay was then performed as above.

Internalization of bacteria by A549 cells. To quantitate internalization of piliated and nonpiliated H. influenzae type b by A549 cells, we used a modification of the assay described by Isberg and Falkow(17) in which gentamicin was used to kill extracellular, noninternalized bacteria. In our system 100 μL of veronal-buffered saline containing 600 cfu of bacteria were added to A549 cells in 96-well plates for 0.5, 1, 2, 3, and 4 h. The wells were washed three times with Hanks' balanced salt solution, and 100 μL of gentamicin (10 mg/mL), which kills extracellular H. influenzae but not intracellular organisms(9), were added to each well for 30 min. After washing four times, 100 μL of 0.5% Triton X-100 (Sigma), diluted in distilled water, were added to each well, and the cells were lysed by gently pipetting up and down for 2 min. Serial 10-fold dilutions of the contents of the wells were plated onto Levinthal agar plates, and the surviving colonies were counted. The percent internalization was calculated using the followingformula.

Statistical analysis. Comparisons of mean adherence, as measured by OD of the immunoassays, was made using nonpaired t tests, calculated using the Stat View statistical program.

RESULTS

Adherence to epithelial cells. Adherence of piliated type b(E1a and M43) and nontypeable (Mr13 and AAr176) H. influenzae to 7291 and 7285 cells, human tracheal fibroblast cell lines, was no greater than that of nonpiliated variants, indicating that bacterial attachment to these respiratory fibroblast cells was not pilus-dependent (Fig. 1). Although some nonpiliated organisms appeared to adhere better than their piliated variants, this was not a consistent finding, and the magnitude of the differences, although statistically significant, were modest. Similarly, adherence of these organisms to human nasal epithelial cells and to A549 human alveolar epithelial cells, was not pilus-dependent(Fig. 1). In contrast, piliated organisms adhered better to bronchial epithelial cells (Fig. 1) than did nonpiliated organisms. Similar to results reported previously(1, 2, 13), piliated organisms adhered better than nonpiliated variants to shed buccal epithelial cells. Although no differences were seen in the binding of piliated and nonpiliated organisms to HeLa cells (Fig. 2), nonpiliated organisms consistently, and significantly, adhered better to HEp-2 cells than did piliated cells.

Figure 1
figure 1

Adherence of piliated (light bars) and nonpiliated (dark bars) H. influenzae, type b strains = E1a and M43, and nontypeable strains = Mr13 and AAr176 to human tracheal fibroblasts (7291 and 7285 cells), nasal epithelial cells, A549 alveolar pneumocytes, bronchial epithelial cells, and buccal epithelial cells.Solid light and dark bars represent cells without GM1 treatment and slashed light and dark bars represent cells that were pretreated with GM1 before incubation with the bacteria. In these experiments, adherence was measured using biotinylated H. influenzae. Error bar = SEM.

Full size image
Figure 2
figure 2

Adherence of piliated (light bars) and nonpiliated (dark bars) H. influenzae to HEp-2 and HeLa cells. In these experiments, adherence was measured using biotinylatedH. influenzae. Error bar = SEM.

Full size image

Using the older adherence assay that detects adherent organisms by their reactivity with H. influenzae cross-reactive antisera, we showed that neither piliated nor nonpiliated H. influenzae adhered to nonhuman (mink) respiratory epithelial cells (Fig. 3). The lower OD values seen with the buccal cells in this experiment, compared with the results seen in Figure 1, reflect the difference in technique between the two types of assays.

Figure 3
figure 3

Adherence of piliated (light bars) and nonpiliated (dark bars) H. influenzae to nonhuman mink epithelial cells and to human buccal epithelial cells. In these experiments, adherent H. influenzae were detected with cross-reacting antisera.Error bar = SEM.

Full size image

To confirm that pili mediated the increased binding of piliated H. influenzae to certain cells, we compared the binding of piliated organisms and their nonpiliated phase variants with that of E1ap/mut, whose pilin gene(encoding the major pilus subunit) had been insertionally inactivated. In these experiments we used epithelial cells that had shown pilus dependent binding (buccal epithelial cells), cells showing pilus independent binding(HeLa cells), and cells that demonstrated increased binding of nonpiliated organisms (HEp-2 cells). The results (Fig. 4) showed that the binding to these cells by E1ap/mut was similar to that of the nonpiliated variant. The observation that the binding of E1ap- to buccal cells is intermediate between the binding of E1ap+ and E1ap/mut may reflect the heterogeneous population of E1ap-(2), in which most bacterial cells are nonpiliated but a few are spontaneous piliated variants. Because E1ap/mut possesses an insertional mutation in the single copy of the pilin gene, no spontaneous piliated variants arise from this strain.

Figure 4
figure 4

Adherence of piliated (light bars), nonpiliated (dark bars) and pilin gene mutant (hatched bars) H. influenzae to buccal epithelial cells (BEC), HeLa, and HEp-2 cells. Error bar = SEM.

Full size image

Inhibition of pilus mediated binding by GM1.Figure 1 also shows the effect of the pilus receptor analog GM1 on adherence of H. influenzae to various respiratory epithelial cells. The nonpilus mediated H. influenzae adherence to the human tracheal fibroblast cells (7291 and 7285) was not inhibited by GM1. However, adherence of nonpiliated M43 to both cell lines was reproducibly, and significantly, increased by preincubation of the organisms with GM1. The meaning of this unexpected finding, which is confined to this strain, is unclear.

The nonpilus mediated H. influenzae adherence to nasal epithelial cells and to A549 cells was also not inhibited by GM1 (Fig. 1). In contrast, the pilus mediated adherence of H. influenzae to bronchial epithelial-derived airway cells and to buccal cells was significantly inhibited by GM1; the OD values of the piliated organisms preincubated with GM1 were similar to those of nonpiliated organisms. These findings confirm those of Van Alphen(19) that GM1 is a pilus receptor analog, and inhibits pilus mediated H. influenzae adherence to respiratory cells.

Internalization of H. influenzae by A549 cells. Because other pulmonary pathogens [group B streptococci(18) andPseudomonas aeruginosa(19)] have been shown to be internalized by A549 cells, which possess many characteristics of human type II pneumocytes(20), we tested the ability of A549 cells to internalize piliated and nonpiliated H. influenzae. The internalization of piliated and nonpiliated organisms did not differ after each of the four incubation time points tested (Fig. 5), and the majority of internalization had occurred by 1 h (percent internalization at 1 h, ± SEM, E1ap+ = 3.0% ± 0.5 and E1ap- = 2.3% ± 0.3, p > 0.1).

Figure 5
figure 5

Internalization of piliated (open square) and nonpiliated (closed diamond) H. influenzae strain E1a by A549 cells.

Full size image

DISCUSSION

Colonization of the human respiratory tract is an important first event in the pathogenesis of infection by H. influenzae. Using a primate model, Weber et al.(21) showed that piliatedH. influenzae colonize the nasopharynx better than nonpiliated organisms. However, whether piliated H. influenzae adheres uniformly to all cells of the respiratory tract has not been well defined. Sterket al.(8) investigated H. influenzae adherence to a variety of cell types using light microscopic examination of tissue sections, and showed that piliated organisms adhered to ciliated cells. These results conflict with those of Farley et al.(5) who showed, using organ cultures, that piliatedH. influenzae adhered best to nonciliated cells.

Results of our study show that H. influenzae adhered to all of the human respiratory tract cell types that we examined, and did not adhere to nonhuman mink lung cells. Although pili mediated the adherence to some cell types, piliated organisms did not adhere better than nonpiliated organisms to nasal epithelial cells, pulmonary alveolar cells (A549), or tracheal fibroblast cells. These results corroborate those of other studies(1, 9, 10) suggesting that H. influenzae possess nonpilus adhesins that facilitate its adherence to human cells. Two such adhesins, two high molecular weight proteins of nontypeableH. influenzae that mediate adherence to Chang conjunctivae cells(22), have been identified. Piliated organisms did, however, adhere better than nonpiliated organisms to buccal epithelial cells, confirming many previous studies(1, 2, 9, 23). In addition, piliated organisms also adhered better than nonpiliated cells to bronchial epithelial cell(14).

In this study, nonpiliated organisms adhered better than piliated organisms to HEp-2 cells, which are derived from human laryngeal carcinoma cells. These results confirm those of Sable et al.(4) and suggest the existence of a nonpilus adhesin specific for a receptor on HEp-2 cells. In addition, the factor that mediates HEp-2 binding of the nonpiliatedH. influenzae phase variant (Fig. 4) appears to function similarly in E1a/mut, in which the pilin gene is insertionally inactivated. This suggests that the nonpilus adhesin is either unmasked by the absence of pili, or is negatively coregulated with pili.

No difference in adherence of piliated or nonpiliated organisms to HeLa cells, which were derived from human cervical carcinoma, was observed in our study, using E1ap+, originally isolated from the spinal fluid of a child with meningitis, or M43p+, isolated from the throat of a child with Hib meningitis. Rosenau et al.(24) described peritrichous pili on 14 of 17 H. influenzae biotype IV strains that correlated with adherence to HeLa cells; however, only 3 of the 14 piliated strains agglutinated red blood cells, suggesting that these pili are different from the buccal cell adherent pili described in our studies. Biotype IV H. influenzae are frequently isolated from infected neonates or from the genital tracts of adults, are genetically homogenous, and are distinct from other H. influenzae strains. Thus, pili of biotype IV H. influenzae may possess adhesins that mediate the specific tropism of these strains to urogenital cells, distinct from the respiratory cell tropism seen with organisms expressing hemagglutinating pili.

Van Alphen et al.(3) observed that gangliosides GM1, GM2, GM3, and GD1a inhibited the adherence of piliatedH. influenzae to oropharyngeal cells and erythrocytes, suggesting that sialyl-lactosylceramide (GM3) may be the minimal binding component of the host cell pili receptors. Our studies confirmed their findings, as GM1 inhibited the adherence of piliated H. influenzae to buccal cells and to bronchial epithelial cells. GM1 did not, however, inhibit the nonpilus-mediated adherence of either piliated and nonpiliated cells to nasal cells, alveolar cells, or tracheal fibroblast cells.

Our results show that pulmonary alveolar cells (A549) internalized both piliated and nonpiliated H. influenzae, and the magnitude of the internalization was similar to that described by St. Geme et al.(10) for the internalization of nonpiliated H. influenzae by Chang conjunctival cells. Thus, pili do not play a role in the adherence to, nor the internalization of, H. influenzae by A549 cells.

These data indicate that the colonization of the human respiratory tract with H. influenzae may be facilitated by both pili and nonpilus factors. Although not understood completely, H. influenzae colonization of the respiratory tract may parallel that of Escherichia coli K-1 in the gastrointestinal tract, in which pili do not appear to play a direct role in colonization of the lower gastrointestinal tract, but rather promote oral colonization which, in turn, assures that a high concentration of organisms flood the lower tract to allow intestinal colonization by other mechanisms(25).