P. King, Monash University Department of Medicine, Monash Medical Centre, 246 Clayton Rd, Clayton, Melbourne, 3168, Australia. E-mail: email@example.com
Non-typeable Haemophilus influenzae (NTHi) is a major cause of respiratory but rarely systemic infection. The host defence to this bacterium has not been well defined in patients with chronic airway infection. The aim of this study was to assess the effect of humoral immunity in host defence to NTHi. Responses were measured in control and bronchiectasis subjects who had recurrent bronchial infection. Antibody and complement-mediated killing was assessed by incubating NTHi with serum and the role of the membrane–attack complex and classical/alternate pathways of complement activation measured. The effect of one strain to induce protective immunity against other strains was assessed. The effect of antibody on granulocyte intracellular killing of NTHi was also measured. The results showed that both healthy control subjects and bronchiectasis patients all had detectable antibody to NTHi of similar titre. Both groups demonstrated effective antibody/complement-mediated killing of different strains of NTHi. This killing was mediated through the membrane–attack complex and the classical pathway of complement activation. Immunization of rabbits with one strain of NTHi resulted in protection from other strains in vitro. Antibody activated granulocytes to kill intracellular bacteria. These findings may explain why NTHi rarely causes systemic disease in patients with chronic respiratory mucosal infection and emphasize the potential importance of cellular immunity against this bacterium.
Non-typeable Haemophilus influenzae (NTHi) is a bacterium that colonizes the throat of most healthy adults . NTHi rarely causes systemic infection, but it is a major cause of mucosal respiratory disease, particularly in chronic bronchitis. The most common cause of chronic bacterial airway colonization in patients with chronic obstructive pulmonary disease (COPD) is NTHi, accounting for up to half of all isolates [2–5]. NTHi is also the most frequently isolated pathogen in bronchiectasis, present in up to 70% of isolates in subjects with bronchiectasis [6,7]. The most common cause of exacerbations of COPD is NTHi, with studies reporting that 25% to more than 80% of exacerbations are associated with H. influenzae[2,5,8,9].
NTHi increases production of inflammatory cytokines [interleukin (IL)-6, IL-8 and tumour necrosis factor (TNF)-α] in an in-vitro model of human respiratory epithelial cells . The bacterial load of NTHi in lung airways has been shown to contribute to airway inflammation in stable chronic bronchitis . Colonization with NTHi is associated with more severe exacerbations of COPD . NTHi is also capable of extensive invasion of lung parenchyma .
NTHi has evolved a large number of mechanisms that facilitate its survival in the human host [12–14]. These include epithelial adhesion molecules, secretion of proteases, microcolony formation, antigenic drift and bio-films. As NTHi appears to cause disease in only a small proportion of the people it colonizes, the host immune response may be important in preventing disease. Previous work has established that NTHi may be killed by a combination of antibody and complement [15,16]. The mechanism of this killing in subjects with chronic airways disease is not well described.
This study compared humoral immune responses in healthy control subjects to those with bronchiectasis and chronic NTHi infection. All subjects in both groups had detectable antibody to NTHi, including immunoglobulin M (IgM), suggesting ongoing active stimulation of the immune response. Serum from both groups was highly effective in killing NTHi and this killing was mediated through the membrane–attack complex with activation of the classical complement pathway. Data from human and animal experiments suggested that infection with one strain of NTHi induces protective immunity against other strains. Antibody was also necessary for effective intracellular granulocyte killing. Humoral immune responses in both control and bronchiectasis subjects were very effective in killing NTHi, and these findings may explain why non-typeable H. influenzae is predominantly a respiratory mucosal pathogen.
A cohort of 22 subjects, aged 55 ± 15 years [mean ± standard deviation (s.d.)] who had bronchiectasis diagnosed by high resolution computed tomography scanning using standard criteria , were studied at Monash Medical Centre. Subjects had been screened for underlying causes of bronchiectasis (including cystic fibrosis mutation analysis, full blood examination, immunoglobulins, lymphocyte subsets and function, neutrophil function and aspergillus precipitins) were classified as having idiopathic disease. Subjects had moderate obstructive lung disease with a mean forced expiratory volume in 1 s (FEV1) of 66·4 ± 24·4% predicted (mean ± s.d.). The subjects had all had multiple isolates of H. influenzae from their sputum in the past 5 years (with an average of three significant isolates; defined as plentiful Gram-negative cocco-bacilli, polymorphs and a moderate to profuse growth of H. influenzae). Bronchiectasis subjects were non-smokers and did not have any other major medical conditions. These patients were all living independently and had not had an exacerbation for at least 1 month.
They were compared with 33 healthy controls aged 52 ± 17 years (mean ± s.d.) who were recruited from laboratory staff and the orthopaedic pre-admission clinic. None of these subjects were current smokers or had any current respiratory symptoms or disease. Ethical approval was obtained from the Southern Health Ethics Committee, Monash Medical Centre and informed consent was obtained from all subjects. For rabbit immunization the project was approved by an internal Millipore (Melbourne, Australia) ethics committee.
Measurement of antibody to NTHi by enzyme-linked immunosorbent assay
NTHi is a heterogeneous species with multiple outer membrane protein (OMP) subtypes. Two different forms of NTHi antigen were used: a previously described pooled heat-inactivated sonicated lysate antigen preparation  and outer membrane P6 antigen, present in all known subtypes of NTHi (gift from Professor T. F. Murphy).
A standard indirect enzyme-linked immunosorbent assay (ELISA) was used as described previously to measure total Ig, IgM and IgA using Binding Site (Birmingham, UK) antibodies.
Ten different live isolates of NTHi were used. Specimens were obtained from significant sputum isolates from the microbiology laboratory at Monash Medical Centre from 10 different subjects with bronchiectasis and COPD/chronic bronchitis. Each isolate was typed (by the Diagnostic Unit of the Department of Microbiology and Immunology, University of Melbourne) and shown to be non-typeable. The 10 isolates were also analysed for their OMP expression by a previously published method  and were shown to have distinct subtypes. The bacteria were designated as strains MMC/MU1–10. Strains were frozen in glycerol broth at −70°C. Prior to use they were cultured onto chocolate agar at 37°C in 5% CO2.
Serum bactericidal assay
Previously described methods were used to assess the effect of serum on killing of NTHi . Bacteria in logarithmic phase were washed and diluted to 5 × 104 colony-forming units (CFUs) per ml in media [RPMI-1640, 10% fetal calf serum (FCS) and l-glutamine] and 10% serum. The reaction mixture was incubated at 37°C in 5% CO2 for 2 h. Duplicate aliquots of 10 µl and 50 µl were taken at 0- and 2-h time-points and plated onto chocolate agar. The plates were incubated overnight and CFUs counted the next day.
To assess the role of the membrane–attack complex in the killing of NTHi, C5, C6 and C8 depleted human serum were used (Sigma-Aldrich, Melbourne, Australia). Bacteria were incubated in media or in media with 10% depleted serum for 2 h as described above and aliquots plated onto chocolate agar and CFUs counted the next day. To assess further the role of the terminal attack complex, human C5 (Sigma-Aldrich) was added to C5 deplete serum, human C6 (Sigma-Aldrich) to C6 deplete serum and human C8 (Sigma-Aldrich) to C8 deplete serum at a concentration of 0·1 mg per ml. To inhibit the classical pathway of complement activation 10 µmol of ethyleneglycol tetraacetic acid (EGTA) (Sigma-Aldrich) and 10 µmol of MgCl2 was incubated with serum obtained from three healthy control subjects, and the inhibition of killing was measured by number of CFUs.
Immunization of animals
Serum was obtained from three non-immunized rabbits. Rabbits were each incubated with a killed heat-inactivated strain of NTHi (rabbit 1 was incubated with a preparation of MMC/MU1, rabbit 2 was incubated with a preparation of MMC/MU2 and rabbit 3 was incubated with a preparation of MMC/MU3). An initial dose of 150 µg of protein with adjuvant was given on day 1 and two booster doses of 1 mg were given and blood obtained on day 49. The immunization of animals and taking of serum samples was performed by Millipore. All 10 strains of NTHi were then incubated individually with serum from the three different rabbits and the effect measured by the number of CFUs the next day.
Granulocyte bactericidal assay
To assess the effect of antibody on granulocyte killing, a killing assay was performed as described previously [21,22]. A pooled bacteria mixture of four live isolates (MMC/MU1–4) was washed in phosphate-buffered saline (PBS) and suspended at a concentration of 107 bacteria per ml in culture medium with peripheral blood granulocytes (at a concentration of 107 per ml) isolated from five control subjects. The pooled bacteria were prepared in three different forms: (i) control; (ii) 10% heat inactivated human serum (to block the effect of complement); and (iii) 10% human serum. For the serum samples of bacteria, specimens were pre-incubated with 20% serum for 30 min to pre-opsonize the bacteria.
The three different preparations of bacteria were then added to granulocytes from controls; with 1 ml of granulocytes suspension added to 1 ml of bacterial suspension. This mixture was then incubated at 37°C with rotation for 5 min. The tubes were washed and resuspended in 2 ml of culture medium (and 10% serum added to serum samples). This was taken as the 0 h time-point. From each tube 200 µl was taken and granulocytes permeabilized with 2% saponin (Sigma-Aldrich) and left at room temperature for 5 min. Samples were then diluted and 10 and 50 µl aliquots taken and applied to chocolate agar plates which were then incubated overnight at 37°C. The same technique was repeated at 1 h and 2 h time-points. The degree of neutrophil killing was assessed by comparing the colony count at the 1 h and 2 h time-points with the 0 h time-point.
Results were tabulated as either the percentage of killing at 2 h compared to the 0 h time-point or as the number of CFUs at 2 h between modified serum samples (for addition of mediators of complement) and expressed as means ± standard error of the mean (s.e.m.) (unless specified otherwise) using Prism 2·0/5·0 (Graphpad Software, San Diego, CA, USA). Differences were analysed using the paired or unpaired t-test as appropriate.
Antibodies to NTHi in control and bronchiectasis subjects
We have demonstrated previously that control subjects and bronchiectasis subjects with recurrent infection with NTHi all had detectable antibody to NTHi (pooled heat-inactivated lysate antigen) with similar titres of total Ig, but higher levels of IgG1 and IgG3. These results were extended by measuring total Ig to both the lysate antigen and P6 in a larger cohort of subjects (Fig. 1a). All control (n = 33) and bronchiectasis (n = 22) subjects had detectable antibody to NTHi. All subjects also had detectable IgA and IgM (Fig. 1b). There was no significant difference between the groups in any of the antibodies measured. The presence of IgM in subjects implies ongoing stimulation of the immune response by NTHi.
Effect of serum on killing of NTHi
The addition of 10% serum obtained from control subjects (n = 10) and bronchiectasis patients (n = 10) was very effective in killing NTHi. In the control subjects the addition of 10% serum resulted in a mean reduction in NTHi of 95·7% ± 0·8 after 2 h; the range was 90·3% ± 5 for stain MMC/MU8 to 99·9% ± 0·1 for strain MMC/MU2. The range for single tests was 51·4–100% killing (Fig. 2a).
In the bronchiectasis subjects the addition of 10% serum resulted in mean reduction in NTHi of 96·1% ± 0·7 after 2 h; the range was 92·9% ± 1·8 for stain MMC/MU5 to 99·9% ± 0·1 for strain MMC/MU10. The range for single tests was 73–100% killing (Fig. 2b).
Killing was equally effective in control and bronchiectasis subjects who had recurrent isolation of NTHi from their sputum. In the 200 individual tests conducted the minimum killing was greater than 50%, with an average of 96% reduction in NTHi after 2 h.
Role of terminal attack components in serum killing of NTHi
Serum killing of NTHi is likely to be mediated by the membrane–attack complex of complement. To test whether this was the case, human serum preparations that had been depleted of C5, C6 and C8 were each incubated with the 10 strains of NTHi (i.e. three different serum preparations). The addition of 10% deplete serum had no significant effect on killing of NTHi after 2 h of incubation. For the C5 deplete serum, NTHi CFUs were 142 000 ± 26 000 CFUs per ml compared with control experiment values of 155 000 ± 25 000. For the C6 deplete serum, NTHi CFUs were 143 000 ± 10 0 CFUs per ml compared with control experiment values of 124 000 ± 34 000. For the C8 deplete serum, NTHi CFUs were 137 000 ± 21 000 CFUs per ml compared with control experiment values of 157 000 ± 36 000. To confirm further that killing of NTHi is mediated by the membrane–attack complex, C5 was added back to the C5 deplete serum, C6 to the C6 deplete serum and C8 to the C8 deplete serum. This addition resulted in restoration of membrane–attack complex-mediated killing. The addition of C5 back to C5 deplete serum resulted in a reduction of NTHi CFUs (P = 0·003). The addition of C6 back to C6 deplete serum resulted in a reduction of NTHi CFUs (P < 0·001). The addition of C8 back to C8 deplete serum resulted in a reduction of NTHi CFUs (P = 0·03). The results are demonstrated in Fig. 3.
Effect of EGTA/MgCl2 on complement killing of NTHi
The membrane–attack complex may be activated by the classical pathway through antibody complement interactions or through the alternate pathway through stabilization on bacterial membranes. As it has been demonstrated previously that killing is dependent upon a combination of antibody and complement it is likely that the pathway for complement-dependent killing is predominantly through the classical pathway. To test whether this was the case, EGTA with MgCl2 was used to inhibit the classical pathway of complement activation. The 10 strains of NTHi were incubated with 10% serum from three healthy controls. EGTA caused complete inhibition of killing, suggesting that the pathway is arising predominantly from classical pathway activation. At 2 h the mean number of bacteria in media without serum was 137 000 ± 17 600 CFUs per ml and in the media with 10% serum and EGTA was 135 000 ± 17 200 (mean results of 30 experiments for each group). The results are demonstrated in Fig. 4.
Effect of immunization of rabbits on serum killing of NTHi
NTHi is an exclusively human pathogen that does not infect other species normally, although immunization in animals including rodents and rabbits has been shown to induce bactericidal antibodies. All controls and bronchiectasis patients in this study had killing of all 10 strains of NTHi. This result suggests that infection with one strain of NTHi may induce bactericidal antibodies against other strains. Serum was obtained from three non-immunized rabbits and no bactericidal killing was demonstrated against all 10 strains of NTHi. Rabbits were then each immunized with a different strain and immune serum collected and then incubated with all 10 strains of NTHi. All three rabbits demonstrated killing of all 10 strains. In the three immunized rabbits the addition of 10% serum resulted in a mean reduction in NTHi of 96·0% ± 1·3 after 2 h. The range for single tests was 68–100% killing. Results are demonstrated in Fig. 5a.
To establish whether the effect of killing using serum from immunized rabbits was due to the classical pathway, serum was incubated with EGTA and MgCl2. Similar to the findings in human subjects, this caused almost complete inhibition of killing. At 2 h the mean number of bacteria in media without serum was 135 000 ± 24 800 CFUs per ml and in the media with 10% serum and EGTA was 127 000 ± 21 700 (mean results of 30 experiments for each group). The results are demonstrated in Fig. 5b.
Granulocyte killing of NTHi
Antibody binds antigen and the antigen–antibody complex is then taken up into phagocytes by binding between the Fc fragment of antibody and the Fc receptor on the phagocyte. This Fc binding not only facilitates phagocytosis but has a major role in enhancing the effectiveness of intracellular killing [21,22]. Antibody is likely to have a role in the activation of granulocytes to kill phagocytosed intracellular NTHi. To test whether this was the case, bacteria were incubated with serum and granulocytes from control subjects and cells were permeabilized to release intracellular NTHi.
The addition of serum enhanced granulocyte killing of NTHi. After 2 h of incubation the number of intracellular bacteria actually increased compared to baseline for the media without serum. The addition of heat-inactivated serum (which removes the effect of complement) reduced bacteria at 2 h significantly compared to control (from 187 ± 44% of baseline numbers to 4·2 ± 1·3 of baseline numbers, P = 0·01). This intracellular killing was enhanced further by the addition of 10% serum (with complement intact) with a reduction to 0·8 ± 0·3 (P = 0·01). These results demonstrate that serum is important in activating intracellular granulocyte killing of NTHi. This effect appears to be predominantly through the activation of Fc receptors (heat-inactivated serum) but is enhanced further by the addition of complement (normal 10% serum). Results are shown in Fig. 6.
This study demonstrated that levels of antibody to NTHi were similar in controls and in bronchiectasis subjects, and killing by serum of a variety of different strains was extremely effective in both groups. Killing was through activation of the classical pathway of complement activation and by the terminal attack complex. Human and animal experiments suggest that development of a humoral immune response to one strain is protective against other strains. Antibody was necessary for granulocyte intracellular killing. The findings may explain why NTHi is predominantly a mucosal pathogen.
Previous studies have demonstrated that NTHi strains cause infection in patients despite the presence of specific antibodies in serum and sputum [23,24]. Groeneveld et al. measured antibodies to five different NTHi isolates in 27 COPD subjects and 13 control subjects. All subjects in both groups had clearly detectable antibody to NTHi from serum. COPD subjects had detectable antibody in sputum that had a similar specificity to the serum. As was the case with the current study, all subjects had detectable IgM which was of similar titre. The IgM findings suggest that NTHi causes ongoing stimulation of the immune response in both patient and control subjects.
A number of previous studies have established clearly that serum is bactericidal for NTHi [15,16]. The effectiveness of serum killing has not been well characterized in subjects with chronic bronchial infection with NTHi. In this cohort of bronchiectasis patients all subjects demonstrated marked killing of a number of different strains of NTHi and this was similar to controls. The results suggest that these subjects with chronic NTHi infection are able to mount a highly effective humoral immune response that prevents systemic infection.
The bactericidal effect of serum of NTHi appears to be mediated through a combination of antibody (IgM and IgG) and complement [15,16]. This bactericidal action suggests killing by the terminal attack complex of complement with insertion of complement components C5–9 to form a pore in the bacterial membrane. The effect of the membrane–attack complex has well characterized in Neisseria meningitidis (meningococcal), Streptococcus pneumoniae and Escherichia coli infection [25,26]. This study demonstrates that NTHi is also killed by this mechanism. The killing appeared to be mediated predominantly through the classical pathway of complement activation and this is expected, as previous studies have demonstrated that both complement and antibody are required for killing of NTHi. The alternative pathway may still have a role in killing of NTHi, however . The killing of encapsulated type b H. influenzae has been shown to be dependent upon the classical and particularly the alternative pathway of complement activation [27,28]. It has also been demonstrated that C-reactive protein (CRP) is capable of activating complement to kill NTHi .
Data from the killing of human subjects and animal data suggest that developing an immune response to one strain of NTHi is protective against other strains of this bacterium. Therefore, there may be common antigens between strains which induce bactericidal antibodies. Groeneveld et al. described strong cross-reactivity between five different strains in serum from COPD and control subjects . Neary et al. immunized a rabbit with P2 outer membrane protein, the most abundant outer membrane protein of NTHi . The antibodies to P2 were then demonstrated to cause complement-mediated killing of eight of 15 strains of NTHi. Antibodies to P6 have also been shown to be bactericidal to multiple strains.
Antibody and complement are also important for activating intracellular killing of phagocytosed bacteria [21,22]. In this study intracellular numbers actually increased over a 2 h time-point in human granulocytes and similar results have been demonstrated for E. coli. The use of heat-inactivated serum enhanced intracellular killing markedly, emphasizing the importance of Fc binding and activation of the granulocyte function. This effect was enhanced further by the addition of complement (in normal serum). These findings imply that how NTHi enters into cells may be important in whether the host defence is activated effectively. One study has described how NTHi is able to enter monocytes by the beta-glucan receptor  and this may effect intracellular survival.
The data from this and previous studies demonstrate that killing of NTHi by antibody and complement is very effective, and this may account for the rarity of systemic NTHi infection. Reports of systemic infection with NTHi have been described in hypogammaglobulinaemia [32–36]. Infection with H. influenzae is recognized to be a feature of C3 deficiency but not deficiency of the terminal complement components in human subjects, suggesting that opsonization with C3 and not the membrane–attack complex killing is the important protective mechanism . The rarity of deficiency of the terminal complement components makes definitive studies difficult, however, and there is one report of invasive Hib infection in a subject with C7 deficiency . It is also possible that both mechanisms of complement-mediated killing and Fc uptake and activation of intracellular killing are important and can compensate for deficiency of the other mechanism. Considering the frequency of NTHi colonization, the incidence of systemic NTHi infection is extremely low, suggesting the effectiveness of humoral immunity in this regard.
Some strains may be more resistant to serum killing, and Williams et al. describe an invasive isolate of NTHi that was able to inhibit C3 deposition . The role of antibody in respiratory mucosal defence against NTHi is not well defined, although it is likely to be significant. IgA would prevent the binding of NTHi to respiratory mucosa and a major pathogenic mechanism of this bacterium is the production of IgA proteases [12–14]. Inflamed bronchi also produce complement and IgG/IgM [39,40], which may enhance bacterial killing.
NTHi is the most common pathogenic bacterium isolated from subjects with bronchiectasis and COPD both in terms of chronic colonization and acute exacerbations. How much of this represents infections with new strains [41,42] or persistent infection with one strain [43,44] is not known. In the bronchiectasis patients in this study apparently effective humoral immune responses were associated with significant bronchial infection with NTHi. In this circumstance NTHi may attempt to evade the immune response by living intracellularly and thus be sheltered from the bactericidal effect of antibody/complement. Recently the ability of NTHi to live intracellularly in monocytes/macrophages and epithelial cells has been described [2,45–50]. For intracellular infections host defence is reliant upon the cell-mediated immune response of the cytotoxic and helper T lymphocytes. The authors have demonstrated previously that subjects with bronchiectasis and chronic NTHi infection have defective cell-mediated responses when compared with healthy controls [18,51].
There is no widely used effective vaccine for NTHi. There have been problems of antigenic variability and finding a suitable animal model. Most work has focused generally upon enhancing humoral immunity. Developing a vaccine to enhance T lymphocyte responses to NTHi may be important.
This study has demonstrated that both control and patient subjects with recurrent NTHi make an effective humoral immune response to this bacterium which enhances microbial killing. The intracellular survival and the corresponding cellular immune response may determine if clinical disease occurs.
The authors would like to thank Professor T. F. Murphy of the State University of New York, Buffalo for the supply of the P6 antigen and Dr Philip Peake of the University of New South Wales, Sydney for advice about complement. We would also like to thank the Department of Clinical Immunology.