Biofilm formation by nontypeable Haemophilus influenzae: strain variability, outer membrane antigen expression and role of pili

Background Nontypeable Haemophilus influenzae is an important cause of otitis media in children and lower respiratory tract infection in adults with chronic obstructive pulmonary disease (COPD). Several lines of evidence suggest that the bacterium grows as a biofilm in the human respiratory tract. Results Fifteen clinical isolates from middle ear fluid of children with otitis media and 15 isolates from sputum of adults with COPD were studied in an in vitro assay of biofilm formation. Striking variability among isolates was observed in their ability to form biofilms. Analysis of cell envelopes revealed minimal differences in banding patterns in polyacrylamide gels, alteration of expression of an epitope on lipooligosaccharide, and preservation of expression of selected epitopes on outer membrane proteins P2, P5 and P6 in biofilms compared to planktonically grown cells. A pilus-deficient variant showed a marked impairment in biofilm formation compared to its isogenic parent. Conclusions Nontypeable H. influenzae forms biofilms in vitro. Clinical isolates show substantial variability in their ability to grow as biofilms. Three major outer membrane proteins (P2, P5 and P6) are expressed during growth as a biofilm. Expression of lipooligosaccharide is altered during growth as a biofilm compared to planktonic growth. Pili are important in biofilm formation. As the role of biofilms in human infection becomes better defined, characterization of biofilms may be important in understanding the pathogenesis of infection and immune response to nontypeable H. influenzae in children with otitis media and adults with COPD.


Background
Nontypeable Haemophilus influenzae is an important human respiratory tract pathogen [1][2][3][4]. The bacterium causes acute otitis media as established by pure culture of the organism from middle ear fluid during disease [5][6][7][8][9]. In addition, nontypeable H. influenzae has been implicated as a cause of otitis media with effusion, which refers to the presence of fluid in the middle ear in the absence of acute symptoms. Acute otitis media and otitis media with effusion represent an enormous health problem for children worldwide [10][11][12]. A large amount of effort has been devoted to understanding the pathogenesis of otitis media with a goal toward developing ways to prevent the disease [1,13,14].
Many middle ear fluids from children with otitis media with effusion are sterile by culture. However, assays using the polymerase chain reaction (PCR) have demonstrated that DNA of nontypeable H. influenzae is present in a substantial proportion of these middle ear fluids [15][16][17][18][19][20]. Furthermore, reverse transcriptase PCR-based assays have shown the presence of bacterial mRNA, indicating that bacteria are present in a viable and metabolically active but non-culturable state [21]. These observations suggest that in otitis media with effusion, bacteria are in a physiological state which differs from that of bacteria growing planktonically in a free floating phase. Some authors have proposed that in otitis media with effusion, nontypeable H. influenzae grow in the form of a biofilm in the middle ear [15,22,23].
A biofilm is a structured community of bacterial cells enveloped in a self-produced polymeric matrix and adherent to an inert or living surface [22,24]. It is estimated that 99.9% of bacteria in nature are attached to a surface in the form of a biofilm. Over the past decade bacteria in the form of biofilms have been recognized as important causes of a variety of human infections, including infections of prosthetic devices, endocarditis, dental caries, pneumonia in cystic fibrosis, prostatitis and others [22,[25][26][27][28]. Biofilms are more resistant to immune clearance mechanisms and to antibiotics compared to planktonic bacteria [29,30]. In order to develop strategies to treat and prevent infections caused by bacteria in biofilms, including otitis media with effusion, it will be important to elucidate the characteristics of bacterial pathogens as biofilms.
In addition to being a common cause of otitis media, nontypeable H. influenzae colonizes the lower respiratory tract of adults with chronic obstructive pulmonary disease (COPD) and is an important cause of exacerbations of the disease [4,31]. An ongoing prospective study of bacterial infection in COPD at the Buffalo Veterans Affairs Medical Center has revealed interesting features regarding the dynamics of colonization in COPD. While active turnover of new strains of nontypeable H. influenzae occurs, some strains have the ability to persist for many months. Monthly sputum cultures reveal intermittent negative cultures in spite of continuous colonization by the same isolate proven by molecular typing, indicating that the organism is present in spite of negative cultures of sputum [32] (and authors unpublished observations). These observations suggest that nontypeable H. influenzae grows as a biofilm in the respiratory tract of adults with COPD, reminiscent of Pseudomonas aeruginosa and its well established propensity to grow as a biofilm while causing infection in patients with cystic fibrosis [27].
In view of these lines of evidence suggesting that nontypeable H influenzae grows as a biofilm during human respi-ratory tract infection, the goals of the present study are to 1) Develop an assay to study biofilms of nontypeable H. influenzae in vitro; 2) Characterize the extent to which isolates from the middle ear fluid of children with otitis media and from the sputum of adults with COPD form biofilms; 3) Begin to elucidate the characteristics of the outer membrane antigens of nontypeable H. influenzae during growth as a biofilm; and 4) Preliminarily assess the role of pili in biofilm formation.

Biofilm formation by clinical isolates
Fifteen isolates recovered from the sputum of adults with COPD and 15 isolates recovered from the middle ear fluid of children with otitis media were studied for biofilm formation. Figure 1a and 1b show that striking differences were observed among strains with regard to the ability to form biofilms. Values depicted in the bar graphs represent the mean of three independent experiments performed in quadriplicate. The error bars represent the standard deviation. Although some variability in values was seen, the level of biofilm formation was highly consistent from experiment to experiment. Strains which formed good biofilms (e.g., strains 5P11H1, 6P8H1, 13P26H1, 31P13H1, 56P40H1 and 74P10H1 from Figure 1b) and strains which formed poor biofims (e.g., 5P10H1, 6P5H, 7P49H1, 6P32H1 and 13P33H1) do so consistently. Inspection of Figure 1 reveals that one cannot readily divide strains into those which form biofilms and those which do not. Rather, a broad range of ability to form biofilms was seen among these randomly chosen 30 clinical isolates.

Conditions for biofilm formation
In an effort to identify growth conditions which might result in increased biofilm formation, eight strains which showed a range of ability to form biofilms were chosen for further study. The strains included 4 isolates from middle ear fluid (955, 1826, 1749, 9289) and 4 isolates from the sputum of adults with COPD (74P10H1, 5P11H1, 5P10H1, 14P13H1). The following reagents were added individually to aliquots of media: 2% glucose, 5% glucose, 2% sucrose, 5% sucrose, 0.2 M NaCl, 2% ethanol. A biofilm assay was performed and the results were compared with biofilms in growth media for each of the strains. None of the added reagents caused an increase in biofilm formation for any of the strains.

Biofilm growth curve
To characterize the kinetics of biofilm formation, the level of biofilm formation was measured at selected time points over a 24 hour period for 4 strains. Figure 2 shows that a gradual increase in biofilm is seen over approximately 16 hours. After approximately 16 hours, biofilm formation shows a leveling off and the optical density re-mains relatively stable to 24 hours at an optical density which is characteristic of each of the 4 strains.

Expression of outer membrane antigens during growth as a biofilm
To preliminarily assess the expression of outer membrane antigens by nontypeable H. influenzae during growth as biofilms, cell envelopes of strain 74P10H1 grown as a biofilm in a 24 well plate were purified. Simultaneously, cell envelopes of strain 74P10H1 grown on agar plates were purified by the same method. Figure 3 is a Coomassie blue stained gel of cell envelopes of strain 74P10H1 grown on agar medium compared to cell envelopes of the same strain grown as a biofilm. Differences are noted in minor bands and differences in the intensity of bands are noted. The major bands appear similar in the two preparations.
To further assess the expression of epitopes on selected individual outer membrane antigens, purified cell envelopes were assayed in immunoblot assays with well characterized polyclonal and monoclonal antibodies. Polyclonal antibodies to a peptide corresponding to loop 6 of the outer membrane protein P2 [33], which is the only known porin of nontypeable H. influenzae, revealed expression of P2 in the bacteria grown as a biofilm and in planktonically grown bacteria (Figure 4). Similarly, the epitopes recognized by monoclonal antibodies to outer membrane protein P5, an OMP A-like protein, and P6, a highly conserved protein in all strains of nontypeable H. influenzae, were expressed in the outer membranes of cells grown as biofilms and cells grown planktonically ( Figure  4).     Optical density 10 15 20 25 Time (hours) When bacteria were grown on agar plates using the same media as for growth in biofilms (brain heart infusion + hemin + nicotinamide adenine dinucleotide), the expression of a lipooligosaccharide epitope was reduced compared to the level of expression in biofilms (Figure 4). This epitope appears to be expressed at similar levels as in biofilms when the bacteria are grown on chocolate agar and in brain heart infusion broth (data not shown). Bacteria grow more slowly on the brain heart infusion agar compared to chocolate agar. The observation that the expression of epitopes on outer membrane proteins P2, P5 and P6 was similar in biofilms and agar ( Figure 4) indicates that the alteration in expression of the epitope on lipooligosaccharide is not explained by slow growth, but represents altered expression of lipooligosacharide under different growth conditions.

Role of pili in biofilm formation
To assess the potential role of pili in biofilm formation, a variant of a strain of nontypeable H. influenzae which lacks expression of pili was tested for biofilm formation along with its isogenic parent strain. Strain M37-1 lacks expression of pili as indicated by absence of hemagglutination and absence of a 22 kDa band on SDS-PAGE corresponding to pilin [34]. In an experiment comparing biofilm formation in 4 duplicate wells, strain M37-1, the pilus deficient variant, yielded an average optical density of 0.204 whereas its isogenic parent strain, M37, yielded an average optical density of 0.884. This experiment was repeated multiple times and the pilus deficient variant showed a 3 to 4 fold decrease in biofilm formation compared to its isogenic parent strain in every experiment.

Discussion
Assays of 30 randomly chosen clinical isolates from the middle ear of children with otitis media and from the sputum of adults with COPD show that isolates of nontypeable H. influenzae demonstrate striking differences in their propensity to grow as biofilms in vitro. These differences are highly reproducible in that individual isolates show similar levels of biofilm formation on repeated experiments The assay for biofilm formation used in this study is based on the ability of bacteria to adhere to plastic wells in microtiter plates. This method has been used to study biofilm formation by a variety of other gram-positive and gram-negative bacteria [35][36][37][38]. The method has both advantages and disadvantages compared to studying biofilm formation in flow cells, the other widely used method. The microtiter plate assay yields reproducible results and is convenient, allowing one to study large numbers of strains and conditions. Furthermore, the method yields quantitative results, based on measuring the optical density of wells. An important potential limitation of the mi-  crotiter plate method is that one might question whether the method is capable of distinguishing true biofilm formation from bacteria simply adhering to plastic wells. Two lines of evidence suggest that the method is measuring biofilm formation. First, Watnick and Kolter [36] showed by confocal laser microscopy that the morphology of biofilms of Vibrio cholerae grown on cover slips in static culture is similar to the morphology of biofilms grown in a flow cell. And second, with regard to this study, the protein patterns in SDS polyacrylamide gel electrophoresis and expression of lipooligosaccharide epitopes are different in bacteria recovered from biofilms grown in micotiter wells compared to planktonic cells grown in media. These observations suggest that the assay is detecting bacteria as biofilms rather than simply measuring planktonic bacteria adhering to plastic wells. However, more definitive answers regarding biofilm formation in human infection caused by nontypeable H. influenzae awaits de-tailed studies involving confocal laser microscopy of clinical samples containing the organism.
The importance of nontypeable H. influenzae as a human respiratory tract pathogen, has stimulated a large body of research to elucidate the antigenic characteristics of the outer membrane of the organism. This work is important in understanding mechanisms of pathogenesis, characterizing the human immune response to infection and guiding vaccine development. This large body of work has been performed exclusively on bacteria grown planktonically. In view of the mounting evidence that nontypeable H. influenzae grows as a biofilm in otitis media with effusion and in the respiratory tract of adults with COPD, it will be important to characterize the bacterial surface antigens of the organism during growth as a biofilm. The present study used well characterized polyclonal and monoclonal antibodies to begin to address this important question.
P2, the major outer membrane protein in the outer membrane is the only known porin protein of H. influenzae. Immunoblot assays with antiserum raised to a conserved surface exposed loop of P2 showed that expression of P2 was similar in amount, molecular mass, and immunoreactivity in bacteria grown planktonically and as a biofilm. This observation is in contrast to expression of porin proteins of Escherichia coli and Pseudomonas aeruginosa whose expression are altered during growth as a biofilm [38,39].
Similarly, expression of outer membrane protein P5, an OMP A-like protein, and P6, a highly conserved 16 kDa protein, is preserved in cells grown as biofilms based on immunoblot assays with monoclonal antibodies ( Figure  4). The expression of P5 and P6 in biofilms is especially important since these proteins are under consideration as vaccine antigens. Finally, alteration of expression of an epitope on lipooligosaccharide suggests that the organism expresses a different lipooligosaccharide molecule during growth as a biofilm. Similarly, expression of lipopolysaccharide of Pseudomonas aeruginosa is altered during growth as biofilms [40]. These studies of outer membrane antigen expression are limited by the sensitivity of immunoblot assays. While the assays establish that P2, P5 and P6 and the epitopes recognized by the antibodies are expressed in biofilms and that expression of lipooligosaccharide is altered, one must be cautious in drawing firm conclusions regarding the level of expression and the surface accessibility of the epitopes recognized by the antibodies based on results of immunoblot assays alone because they are not quantitative assays.

Conclusions
Nontypeable Haemophilus influenzae forms biofilms in vitro. Clinical isolates show substantial variability in their ability to grow as biofilms. Three major outer membrane proteins, P2, P5 and P6, are expressed during growth as a biofilm. Expression of an epitope on lipooligosaccharide is altered during growth as a biofilm. Pili are important in biofilm formation. As the role of biofilms in human infection becomes better defined, characterization of biofilms may be important in understanding the pathogenesis of infection and immune response to nontypeable H. influenzae in children with otitis media and adults with COPD. Bacteria were grown on chocolate agar or in brain heart infusion broth supplemented with hemin and nicotinamide adenine dinucleotide both at 10 µg/ml. In the experiments to compare cell envelopes of bacteria grown planktonically with those grown as biofilms, bacteria were grown on brain heart infusion agar supplemented as noted above. Media was supplemented with glucose, sucrose, NaCl and ethanol in selected experiments as described in the Results. A volume of 230 µl of 95% ethanol was added to each well and the OD 570 was measured. All strains were tested in quadruplicate. Each plate included 4 wells which contained sterile broth instead of bacteria but were treated identically otherwise. The OD 570 was standardized against these wells.

Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS PAGE) and immunoblot assays
Cell envelopes were subjected to SDS PAGE and Coomassie blue staining using previously described methods [44]. Immunoblot assays were performed as described previously [45]. After incubating with the antibodies, blots were developed with the appropriate murine or rabbit peroxidase conjugated secondary antibody.

Purification of cell envelopes
To study the outer membrane of nontypeable H. influenzae, bacteria were grown as a biofilm overnight in a 24 well Linbro tissue culture plate. Broth was aspirated and the wells were washed 3 times with 1 ml of 0.01 M HEPES, pH 7.4. Bacteria in the biofilm were harvested by scraping the wells with the tip of a 1 ml micropipette tip. To prepare cell envelopes, cells were suspended in 0.01 M HEPES, pH 7.4 and sonicated on ice by using a Branson Sonifier (small tip, setting 7) [44]. The suspension was centrifuged at 12,000 x g for 2 minutes at 4°C to remove unbroken cells and debris. The supernatant was recovered and centrifuged at 12,000 x g for 45 minutes. The resulting pellet was suspended in sample buffer and subjected to SDS-PAGE and immunoblot assay.