Internalization and Trafficking of Nontypeable Haemophilus influenzae in Human Respiratory Epithelial Cells and Roles of IgA1 Proteases for Optimal Invasion and Persistence

Abstract

Nontypeable Haemophilus influenzae (NTHI) is a leading cause of opportunistic infections of the respiratory tract in children and adults. Although considered an extracellular pathogen, NTHI has been observed repeatedly within and between cells of the human respiratory tract, and these observations have been correlated to symptomatic infection. These findings are intriguing in light of the knowledge that NTHI persists in the respiratory tract despite antibiotic therapy and the development of bactericidal antibodies. We hypothesized that intracellular NTHI avoids, escapes, or neutralizes the endolysosomal pathway and persists within human respiratory epithelial cells and that human IgA1 proteases are required for optimal internalization and persistence of NTHI. Virtually all strains encode a human IgA1 protease gene, igaA, and we previously characterized a novel human IgA1 protease gene, igaB, that is associated with disease-causing strains and is homologous to the IgA1 protease that is unique to pathogenic Neisseria spp. Here, we show that NTHI invades human bronchial epithelial cells in vitro in a lipid raft-independent manner, is subsequently trafficked via the endolysosomal pathway, and is killed in lysosomes after variable durations of persistence. IgaA is required for optimal invasion. IgaB appears to play little or no role in adherence or invasion but is required for optimal intracellular persistence of NTHI. IgaB cleaves lysosome-associated membrane protein 1 (LAMP1) at pHs characteristic of the plasma membrane, early endosome, late endosome, and lysosome. However, neither IgA1 protease inhibits acidification of intracellular vesicles containing NTHI. NTHI IgA1 proteases play important but different roles in NTHI invasion and trafficking in respiratory epithelial cells.

INTRODUCTION

Nontypeable Haemophilus influenzae (NTHI) is a Gram-negative, human-exclusive commensal bacterium of the nasopharynx and is a leading cause of opportunistic infections in the upper and lower respiratory tracts, including otitis media, sinusitis, conjunctivitis, community-acquired pneumonia, and exacerbations of chronic obstructive pulmonary disease (COPD) and of cystic fibrosis (1, 2).

When a new strain of NTHI is acquired, the outcome depends on a variety of dynamic host and bacterial factors. Colonization and the transition from commensal to opportunistic pathogen require NTHI to resist host innate immune defenses, including nutrient sequestration, mucociliary clearance, antimicrobial peptides, secretory IgA1, and phagocytosis by immune cells or by epithelial cells. Bacterial counterresistance mechanisms under investigation include adhesins, antimicrobial peptide degradation, IgA1 protease, biofilm production, modification of surface-exposed lipooligosaccharide moieties, production of outer membrane vesicles (OMVs), and others (3–12). Understanding these factors will help identify targets for therapeutic intervention, as NTHI infections induce acquired immunity consisting largely of strain-specific bactericidal antibodies that afford little or no cross-protection against newly acquired strains, given the high degree of antigenic heterogeneity among strains.

NTHI is considered to be an extracellular pathogen. However, for several decades, studies have reported significant numbers of NTHI within and between a variety of human respiratory epithelial and subepithelial cells and macrophages in vitro and ex vivo, including the bronchi of adults with chronic lung disease and the adenoids of children with adenoidal hypertrophy or with a history of chronic otitis media (13–27). Few studies have examined the internalization, trafficking, and fate of NTHI in host cells. It is important to understand these processes because of a possible correlation between intracellular NTHI and symptomatic infection and because NTHI infections frequently persist and recur despite antibiotic therapy and the development of bactericidal antibodies (14, 28–31). Studies in our laboratory have demonstrated that single strains of NTHI can persist in patients with COPD for months to years at a time, despite periods of culture-negative clinical evaluations between exacerbations caused by those strains (31). Collectively, these findings have fueled speculation that host intracellular locales provide temporary niches or even long-term reservoirs for chronic NTHI infection by providing nutrients and protection from immune pressures (3).

The NTHI pangenome does not encode specialized secretion systems for active invasion of host cells, suggesting that this bacterium must rely on existing host pathways to invade or survive within host cells or tissues. Host cell internalization of NTHI involves actin, tubulin, and the formation of lamellipodia and microvilli, which appear to engulf the bacteria into vesicles (13, 19, 23, 32). Internalization of NTHI is also promoted by macropinocytosis, β-glucan receptor (βGR)-mediated endocytosis in macrophages, and platelet-activating factor receptor (PAFR)-mediated endocytosis in epithelial cells (13, 19, 24, 25). The bacterial ligand for PAFR is phosphorylcholine (ChoP), and while strains with either ChoP-positive or ChoP-negative lipooligosaccharide (LOS) induce differential host cell signaling, it seems that all strains induce phosphatidylinositol-3-kinase (PI3K) signaling, phospholipase activity, and calcium release from intracellular stores (20, 22, 24, 25, 33). NTHI has consistently been found within vesicles positive for markers of endolysosomal trafficking (20, 22). Since many pathways terminate in lysosomes, NTHI must find ways to avoid, escape, or neutralize this deadly fate.

Secretory IgA1 is the predominant immunoglobulin of human mucosal surfaces. In addition to having virulostatic properties, secretory IgA1 confines pathogens to the mucosal lumen by trapping them in the mucus layer and by promoting phagocytosis (34, 35). IgA1 proteases are unique to human mucosal bacteria. Virtually all strains of NTHI encode a type I IgA1 protease gene, igaA (previously annotated as iga), and we previously identified a novel type II IgA1 protease gene, igaB, that is immunogenic and is more prevalent among pathogenic clinical isolates of NTHI than among isolates from asymptomatic colonization (36–40). Similarly, the IgA1 protease activities of symptomatic isolates are greater than those of commensal isolates of NTHI (40).

The igaB gene is homologous to a type II IgA1 protease of pathogenic Neisseria and was likely acquired from Neisseria by a genomic inversion event (41). The neisserial protease promotes the survival of intracellular Neisseria, seemingly via cleavage of the IgA1-like hinge region of lysosome-associated membrane protein 1 (LAMP1), a large glycoprotein important for lysosomes to acquire microbicidal activity (42, 43). LAMP1 is a major component of the membrane of late endosomes and lysosomes and is highly glycosylated to protect against degradation by lysosomal contents. The hinge region faces the lumen of late endosomes and lysosomes, where it would be accessible to bacteria and IgA1 proteases. LAMP1 spans the lysosomal membrane once, such that cleavage of the hinge region should release roughly half of the luminal portion of the molecule and might prevent lysosome acidification, perhaps by altering membrane trafficking and vesicle maturation or by exposing the membrane to luminal hydrolases, thereby compromising the integrity of the late endosome or lysosome (44, 45).

We hypothesized that NTHI is internalized by and persists in human respiratory epithelial cells and that IgA1 proteases are required for optimal internalization and persistence. To characterize the kinetics of NTHI internalization and persistence in human bronchial epithelial cells and to determine the relative roles of IgA1 proteases, we used gentamicin protection assays to measure the adherence, invasion, and survival of NTHI in an in vitro infection model system. To elucidate the trafficking of NTHI within human bronchial epithelial cells and to specify the roles of IgA1 proteases, we used confocal microscopy, gentamicin protection assays in the presence or absence of pharmacologic inhibitors, LAMP1 cleavage assays, and pH measurements of vesicles containing NTHI.

MATERIALS AND METHODS

Bacterial strains, cell lines, and media.

Nontypeable Haemophilus influenzae (NTHI) strain 11P6H, a COPD exacerbation isolate, was previously used to create isogenic IgA1 protease mutants (ΔigaA, ΔigaB, and ΔigaA ΔigaB) (37). NTHI was grown at 35 to 37°C with 5% CO₂ on chocolate agar or in supplemented brain heart infusion (sBHI) broth, which consisted of BHI broth supplemented with hemin and β-NAD.

NCI-H292, a human bronchial mucoepidermoid carcinoma cell line, was acquired from the American Type Culture Collection (ATCC). Cells were propagated at 37°C with 5% CO₂ in supplemented RPMI (sRPMI), which consisted of RPMI 1640 plus l-glutamine base medium (Life Technologies) supplemented with fetal bovine serum (FBS) (10% final concentration), HEPES (10 mM final concentration), and sodium pyruvate (1 mM final concentration). H292 cells used for experiments were passaged no more than 10 times. H292 cells were tested for mycoplasma using the Mycosensor PCR assay kit (Agilent).

RNA isolation and purification.

Total RNA was isolated from wild-type strain 11P6H grown in sBHI broth to optical densities spanning lag, log, and stationary phases. At each indicated culture density, bacteria were pelleted and resuspended in Tris-EDTA buffer and RNAprotect bacteria reagent (Qiagen). The bacteria were repelleted and stored at −80°C. RNA was extracted from thawed bacterial pellets using RNAWiz (Ambion) and chloroform and precipitated using ethanol. RNA was then washed using RiboPure bacteria columns and wash solutions (Ambion), eluted, and quantified on the NanoDrop 2000c (Thermo Scientific). DNA was removed from RNA preparations by using RNase-free DNase (Qiagen), followed immediately by RNA cleanup using an RNeasy kit (Qiagen). The absence of DNA was confirmed by PCR. RNA purity and integrity, respectively, were assessed by measuring sample absorbance at 260 and 280 nm using NanoDrop 2000c (Thermo Scientific) and Agilent 2100 bioanalyzer (Agilent) spectrophotometers and by the appearance of the sample on a MOPS (morpholinepropanesulfonic acid)-formaldehyde denaturing agarose gel. Purified RNA aliquots were stored at −80°C.

Quantitative RT-PCR.

We performed absolute quantification of igaA, igaB, and the housekeeping gene gyrA by quantitative reverse transcription-PCR (qRT-PCR) of purified RNA using the iScript cDNA synthesis kit (Bio-Rad), iQ SYBR green supermix (Bio-Rad), and the CFX384 Touch real-time PCR detection system (Bio-Rad). Primer3 Input version 0.4.0 (frodo.wi.mit.edu) was used to design primers to amplify specific regions of igaA, igaB, and gyrA. The specificity of each primer set was confirmed against genomic DNA templates from 11P6H wild-type, ΔigaA, ΔigaB, and ΔigaA ΔigaB strains in PCRs. The RNA purity and primer specificity were further confirmed using purified RNA as the template for both PCR and RT-PCR using HotMasterMix (5 Prime) and the OneStep RT-PCR kit (Qiagen), respectively, resulting in the appropriate absence and presence of single products of the expected size from samples and controls.

IgA1 cleavage assay.

H292 cells were infected according to the adherence-invasion assay protocol. Supernatants from medium collected at 1, 4, and 24 h postinoculation were incubated with purified human IgA1 (Calbiochem) at 37°C overnight. These samples were then stored at −20°C prior to being separated on 12% SDS-PAGE, transferred to nitrocellulose, probed with horseradish peroxidase (HRP)-conjugated goat anti-human IgA (1:1,000; KPL), and visualized by color development (Bio-Rad).

Adherence-invasion assays.

Twenty-four-well plates were seeded with H292 cell suspensions (2 × 10⁵ cells per well). Per time point and per strain of bacteria, duplicate wells were seeded for adherence and for invasion. NTHI strains were grown on chocolate agar overnight, and these cultures were used to inoculate sBHI broth to an optical density at 600 nm (OD₆₀₀) of 0.08. These cultures were shaken at 37°C until they reached mid-log phase (OD₆₀₀ of 0.400). Confluent monolayers were washed twice with fresh medium, inoculated with NTHI at a multiplicity of infection (MOI) of 1 bacterium per H292 cell, and incubated at 37°C with 5% CO₂. At selected time points postinoculation, medium from infected samples was harvested, diluted in phosphate-buffered saline (PBS), and cultured on chocolate agar. To determine the number of cell-associated bacteria and the number of viable intracellular bacteria, infected monolayers were washed three times with PBS, followed by a 1-h incubation with either sRPMI or sRPMI containing 50 μg · ml⁻¹ gentamicin, respectively. Monolayers were then washed three times with PBS, trypsinized, permeabilized with 0.8% saponin, diluted in PBS, and cultured in duplicate on chocolate agar. The CFU from medium and from cells not treated with gentamicin were added together to calculate the total number of CFU in a sample well. “Percent adherence” was calculated by dividing the CFU of cells not treated with gentamicin by the total CFU and multiplying the resulting value by 100. “Percent invasion” was calculated by dividing the CFU of cells treated with gentamicin by the total CFU and multiplying the resulting value by 100. Trypan blue exclusion assays were conducted to control for cytotoxicity. Data represent the means of at least four independent, replicate experiments. Statistical significance was calculated using a paired Student t test.

Intracellular survival assay.

The intracellular survival assay followed the adherence-invasion assay protocol through 24 h postinoculation, at which time samples were cultured for CFU to calculate percent adherence and percent invasion at 24 h, while remaining samples were washed with PBS and treated with sRPMI containing gentamicin (50 μg · ml⁻¹). This gentamicin treatment was kept constant and was refreshed every 24 h. At daily time points, monolayers were washed with PBS, trypsinized, permeabilized with 0.8% saponin, diluted in PBS, and cultured on chocolate agar to determine the number of viable intracellular bacteria. Percent survival was calculated by dividing the intracellular CFU at a given time point by the total CFU at 24 h and multiplying the resulting value by 100. Trypan blue exclusion assays were conducted to control for cytotoxicity. The data represent the means of three independent, replicate experiments. Statistical significance was calculated using a paired Student t test.

Adherence-invasion assays with pharmacological inhibitors.

Noncytotoxic concentrations of each inhibitor were established through lactate dehydrogenase (LDH) release assays of uninfected cells using a cytotoxicity detection kit (Roche) and through trypan blue exclusion assays of infected cells. A 100 mM concentrated stock of chlorpromazine hydrochloride (Sigma) was used to make a 1 mM working stock in deionized water. Stocks were stored in foil at 4°C. Aliquots of a 25 μg · ml⁻¹ working stock of concanamycin A (Sigma) in dimethyl sulfoxide (DMSO) were stored at −20°C. A 500 μg · ml⁻¹ working stock of filipin III (Cayman Chemical) in DMSO was made fresh for same-day use. A 50 mM concentrated stock of nystatin (Sigma) was used to make a 5 mM working stock in DMSO. Working-stock aliquots were stored at −20°C. A 30 mg · ml⁻¹ working stock of 3-methyladenine (Acros Organics) dissolved in double-deionized water heated to 56°C was made fresh for same-day use.

For adherence-invasion assays using pharmacological inhibitors, the adherence-invasion assay protocol was modified to include a 2-h preincubation with sRPMI containing either inhibitor or inhibitor diluent, followed by inoculation with bacteria and a 4-h incubation at 37°C. For experiments with concanamycin A, H292 cells were infected for 16 h, washed, treated with gentamicin for 1 h, washed, and treated with sRPMI containing concanamycin A or diluent for 3 h.

Adherence and invasion ratios were calculated by dividing the percent adherence or percent invasion in the presence or absence of inhibitor by the respective percentage in the absence of inhibitor. Similarly, for experiments with concanamycin A, rescue data are presented as ratios that were calculated by dividing the intracellular CFU following the 3-h treatment with concanamycin A or with diluent by the intracellular CFU following the 3-h treatment with diluent. Data represent the averages of at least three independent replicate experiments. Statistical significance was calculated using a paired Student t test.

Inhibitor activity was evaluated using the following fluorescent control cargo conjugates: cholera toxin subunit B Alexa Fluor 488 conjugate (Invitrogen) (2 μg/ml final concentration), transferrin Alexa Fluor 488 conjugate (Invitrogen) (50 μg/ml final concentration), and dextran 70000 Texas red conjugate (Invitrogen) (50 μg/ml final concentration). Chambered slides of H292 cells were washed and incubated with sRPMI containing inhibitor or inhibitor diluent for 2 h, followed by the addition of cargo conjugates, incubation for 20 min on ice, and incubation for 30 min to 2 h at 37°C with 5% CO₂. Samples were washed, fixed, and examined using a Zeiss LSM-Meta NLO (laser-scanning microscope with nonlinear optics) confocal microscope.

Confocal microscopy, antibodies.

Lab-Tek 4-well chambered coverglasses (Nunc) were seeded with H292 cells and infected according to the adherence-invasion assay protocol. Samples were fixed with 2% paraformaldehyde for 1 h at room temperature and stored in PBS at 4°C. Bacteria were visualized using a directly conjugated rabbit polyclonal antibody that was raised against fixed whole bacteria of strain 11P6H, which recognizes a large number of bacterial epitopes. This antibody was purified from rabbit serum using HiTrap protein G HP affinity columns (GE Healthcare) and directly conjugated to Alexa Fluor 488 (Life Technologies). Host cell components were visualized using filipin III fluorescent dye (final concentration of 333 μg/ml; Cayman Chemical), primary monoclonal antibodies including mouse anti-caveolin-1 antibody (1:50; BD), mouse anti-early endosomal antigen 1 (EEA-1) antibody (1:200; BD), mouse anti-flotillin-1 antibody (1:20; BD), mouse anti-human CD107a antibody (LAMP1) (1:200; BD), and goat anti-human secretory component antibody (1:200; Sigma), and secondary antibodies including highly cross-adsorbed goat anti-mouse IgG conjugated to Alexa Fluor 568 (Life Technologies, 1:200) and donkey anti-goat IgG conjugated to Alexa Fluor 568 (Life Technologies, 1:200). All antibodies were tested against bacteria alone and H292 cells alone to control for nonspecific staining. Samples were examined and images were captured using a Zeiss LSM-Meta NLO confocal microscope.

rhLAMP1 cleavage assay.

H292 cells were infected according to the adherence-invasion assay protocol. Supernatants from media collected at 24 h postinoculation were filter sterilized and cultured to ensure sterility. The pH of each supernatant was measured, and portions of each supernatant were subsequently adjusted to pHs representative of the plasma membrane and of endolysosomal compartments. Samples were cultured to ensure that the supernatants were still sterile following pH adjustments. Each supernatant was aliquoted, stored at −20°C, and never thawed more than once. Controls were included for pH changes induced by freeze-thaw, by the addition of PBS (LAMP1 diluent), and by overnight incubation. Thawed supernatants were incubated with purified recombinant human LAMP1 (rhLAMP1; R&D Systems) at 37°C overnight. Digestions were separated on NuPAGE Novex 10% bis-tris gel (Life Technologies), transferred to polyvinylidene difluoride (PVDF) membrane (Bio-Rad), probed with mouse IgG anti-human LAMP1 (1:1,000; BD) and HRP-conjugated goat anti-mouse IgG (1:1,000; KPL), and visualized using Amersham ECL Prime (GE) chemiluminescence reagent for exposure to film.

SNARF-1 pH measurements.

NTHI was grown on chocolate agar overnight, and these cultures were used to inoculate sBHI broth to an OD₆₀₀ of 0.08. These cultures were shaken at 37°C until they reached an OD₆₀₀ of 0.4. Aliquots of each culture were pelleted at 13,000 × g for 1 min, washed, resuspended in 0.1 M sodium phosphate buffer (pH 7.5), stained with SNARF-1, a pH-sensitive fluorophore with biphasic emission (final concentration of 125 μg/ml; Life Technologies), for 30 min at room temperature, pelleted, washed, and resuspended in FBS-free RPMI. Lab-Tek 4-well chambered coverglasses (Nunc) of confluent H292 cells were washed with FBS-free RPMI. These slides and four-chambered slides containing only FBS-free RPMI (no H292 cells) were inoculated with SNARF-1-stained bacteria (MOI of 100) and incubated at 37°C with 5% CO₂. At 6 h postinoculation, wells containing bacteria alone were washed and immersed in sodium phosphate buffers of pH 5, 6, 7, or 8. These calibrators and the infected H292 cells were immediately examined with a Zeiss LSM-Meta NLO confocal microscope, using an excitation wavelength of 514 nm and emission wavelength band pass filters of 555 to 600 nm and 640 to 700 nm. The gain settings for each emission wavelength were kept constant for all images captured in a given experiment. Representative images were selected from image stacks by using ImageJ64 (rsbweb.nih.gov/ij/index.html), and the pixel intensities of each emission color (yellow and red) were quantified using ImageProPlus (MediaCybernetics) for a minimum of 20 bacteria or bacteria-containing vesicles per sample. The pixel intensity ratios (yellow/red) of the calibrators were used to generate a standard curve by which the pixel intensity ratios of infected samples were converted to their corresponding pH values.

RESULTS

IgA1 proteases of strain 11P6H are constitutively transcribed and exhibit different cleavage activities in a respiratory epithelial cell model system.

To characterize IgA1 protease gene expression through all phases of bacterial growth, we performed quantitative RT-PCR using samples collected from a growth curve of the wild-type strain grown in broth. Both igaA and igaB were constitutively expressed through the lag, log, and stationary phases of growth, which is consistent with other IgA1 proteases and is expected given the lack of regulatory sequences upstream of either gene (Fig. 1). The expression of igaB was approximately 10-fold greater than the expression of igaA.

FIG 1.

Quantitative RT-PCR. NTHI IgA1 protease gene expression is constitutive through lag, log, and stationary phases. Wild-type NTHI parent strain 11P6H was grown in sBHI broth and pelleted at the culture densities indicated. Absolute quantification of igaA, igaB, and gyrA transcripts was achieved using qRT-PCR, and all reactions were performed in triplicate. Error bars represent standard errors of the means of three independent experiments.

To confirm and characterize the activity of IgaA and IgaB in our H292 bronchial epithelial cell infection model system over time, we performed an IgA1 cleavage assay using media collected from NTHI-infected H292 cells at 1, 4, and 24 h postinoculation. The IgA1 cleavage patterns generated using samples from infected H292 cells were identical to those generated using samples from parallel experiments with bacteria grown in broth in the absence of H292 cells. IgaB-generated cleavage fragments appeared sooner than IgaA-generated cleavage fragments through the 24-h time point (Fig. 2). However, IgaA-generated cleavage fragments were absent from wild-type samples (Fig. 2). Together, the quantitative RT-PCR data and IgA1 cleavage assay observations indicate that the dominance of IgaB-mediated cleavage in the wild-type strain is not due to a lack of igaA gene expression or to instability of the igaA transcript but is perhaps instead due to a greater abundance of the igaB transcript, preferred translation of the igaB transcript, or preferred substrate affinity for IgaB.

FIG 2.

IgA1 cleavage assay. NTHI IgA1 proteases are active and exhibit distinct cleavage activities in the H292 infection model system. H292 cells were infected according to the adherence-invasion assay protocol. Supernatants from medium collected at 1, 4, and 24 h postinoculation were incubated with purified human IgA1 at 37°C overnight. Digestions were separated on SDS-PAGE, transferred to nitrocellulose, probed with goat anti-human IgA, and visualized by color development. The top band is intact IgA1, and the bands at the levels of the arrows are IgA1 cleavage products.

IgaA is required for optimal invasion of bronchial epithelial cells by NTHI.

Once we confirmed and characterized IgA1 protease expression and activity in our infection model system, we performed adherence-invasion and persistence assays to ascertain the kinetics of NTHI entry into and persistence within respiratory epithelial cells and the relative roles of IgaA and IgaB. We chose an MOI of 1 because it is more likely to represent an early in vivo situation than a higher inoculum. The MOI reflected the concentration of bacteria at the time of inoculation and early infection. This concentration increases over the course of the infection. At 4 h postinfection, we observed a high number of infected cells with some degree of scatter, and by 24 h postinfection, all cells were infected. The adherence-invasion assay measured viable bacteria in the medium, in the cell layer (adherence), and in the cell layer treated with the membrane-impermeant antibiotic gentamicin (invasion), and these values are presented as percentages of the total number of bacteria in a given sample, since percentages account for the normal experimental variability of CFU. The data analyzed as percentages yielded identical results as when analyzed as CFU.

Adherence by the ΔigaB mutant through 24 h was largely unchanged compared to the adherence of the wild-type strain (Fig. 3). Adherence by the ΔigaA and ΔigaA ΔigaB mutants was enhanced at 1 and 4 h and diminished at 24 h, such that adherence by strains possessing igaA (wild-type and ΔigaB strains) did not surpass adherence by strains lacking igaA (ΔigaA and ΔigaA ΔigaB strains) until 24 h (Fig. 3). These data indicate that IgaA plays an unclear role in adherence by NTHI and that IgaB plays no role in adherence by NTHI under the conditions of this assay. Invasion by the ΔigaB mutant through 24 h was largely unchanged compared to invasion by the wild-type strain (Fig. 3). Invasion by the ΔigaA and ΔigaA ΔigaB mutants was diminished through 24 h (Fig. 3). These data indicate that IgaA is required for optimal invasion and suggest that IgaB plays little or no role in invasion by NTHI under the conditions of this assay.

FIG 3.

Adherence-invasion assay. IgaA is required for optimal invasion by NTHI. Percent adherence was calculated by dividing the CFU of infected H292 cells in the absence of gentamicin by the total CFU per well and multiplying the resulting value by 100. Percent invasion was calculated by dividing the CFU of infected H292 cells treated with gentamicin by the total CFU per well and multiplying the resulting value by 100. Bacterial counts at each of the time points for adherence were as follows: 1 h, ∼10⁵ CFU/ml; 4 h, ∼10⁶ CFU/ml; 24 h, 10⁷ CFU/ml. Bacterial counts at each of the time points for invasion were as follows: 1 h, ∼10⁴ CFU/ml; 4 h, ∼10⁶ CFU/ml; 24 h, ∼10⁷ CFU/ml. Error bars represent standard errors of the means for four independent experiments. A paired Student t test was used to calculate the statistical significance of the differences observed between the wild type and each mutant. *, P ≤ 0.05; **, P ≤ 0.005; ***, P ≤ 0.0005.

IgaB is required for optimal intracellular persistence of NTHI.

To ascertain the kinetics of NTHI persistence within human respiratory epithelial cells and the relative roles of IgaA and IgaB beyond 24 h postinoculation, we performed intracellular survival assays with constant gentamicin exposure beginning at 24 h postinoculation in order to track the viability of bacteria within infected H292 cells. Controls for H292 cell viability were performed in parallel and confirmed that sustained gentamicin exposure does not affect H292 cell viability.

Each strain exhibited declining intracellular viability through 120 h, with the data indicating that intracellular NTHI died or was killed in respiratory epithelial cells in vitro and that intracellular NTHI persisted for as long as 96 h postinoculation in this infection model system (Fig. 4). Relative to the wild-type strain, the ΔigaA mutant died more slowly and the ΔigaB mutant died more rapidly (Fig. 4). These data indicate that IgaB, like its neisserial homologue, is required for optimal intracellular persistence of NTHI. The data also suggest that IgaA promotes a more rapid bacterial death, possibly by mediating a different, more lethal or more rapid trafficking pathway or by competing with IgaB.

FIG 4.

Intracellular survival assay. IgaB is required for optimal intracellular persistence of NTHI, while IgaA might promote a more rapid death of NTHI. Percent invasion (1 to 24 h) was calculated by dividing the CFU of cells treated with gentamicin by the total CFU per well. Percent survival (48 to 120 h) was calculated by dividing the CFU of cells at a given time point by the total CFU per well at 24 h. Gentamicin treatment durations beyond 24 h were constant and refreshed daily. Error bars represent standard errors of the means of four independent invasion assay experiments and three independent survival assay experiments. A paired Student t test was used to calculate the statistical significance of the differences between the wild type and each mutant. *, P ≤ 0.05; **, P ≤ 0.005; ***, P ≤ 0.0005.

Pathways of bronchial epithelial cell invasion by NTHI.

To elucidate the internalization and trafficking of NTHI and to specify the role(s) of IgaA and IgaB, we examined the relationship between NTHI and commonly usurped host pathways by using confocal microscopy and adherence-invasion assays in the presence and absence of pharmacologic inhibitors. Literature searches and reviews informed our selection of inhibitors, and noncytotoxic concentrations of each inhibitor were established through LDH release assays of uninfected cells and through trypan blue exclusion assays of infected cells.

Clathrin-mediated endocytosis is also known as classical or receptor-mediated endocytosis. In addition to mediating the endocytosis of ligands bound to protein receptors, this pathway is usurped for the internalization of certain viruses and bacteria. Adherence-invasion assays were performed by using pharmacologic inhibition of clathrin-mediated endocytosis by chlorpromazine, which inhibits clathrin-mediated internalization by sequestering clathrin. Chlorpromazine had no effect on NTHI adherence, but it equally and significantly inhibited invasion by the wild type, the ΔigaB mutant, and the ΔigaA ΔigaB mutant (Fig. 5). These data suggest that clathrin plays a role in invasion by NTHI, which is consistent with findings that ChoP-PAFR binding mediates NTHI adherence and invasion, since PAFR internalization appears to be clathrin dependent. We do not have a good explanation for why inhibition of invasion by the ΔigaA mutant with chlorpromazine did not reach statistical significance.

FIG 5.

Adherence-invasion assay in the presence and absence of chlorpromazine. To assess the role of clathrin in the internalization of NTHI, H292 cells were pretreated with chlorpromazine (CPZ) or with diluent for 2 h, followed by a 4-h infection conducted according to the adherence-invasion assay protocol. Data are normalized to samples in the absence of chlorpromazine. Error bars represent standard errors of the means of three independent experiments. A paired Student t test was used to calculate statistical significance. *, P ≤ 0.05; **, P ≤ 0.005.

Lipid raft-mediated endocytosis is another well-characterized form of endocytosis that is usurped for the internalization of certain viruses, bacterial toxins, and bacteria. To determine whether NTHI colocalizes with lipid rafts, infected H292 cells were stained with antibodies for caveolin-1 and flotillin-1, which are markers of caveolar lipid rafts. No colocalization of NTHI was observed with either marker at 4 and 24 h postinoculation (Fig. 6A). Unlike caveolin-1 and flotillin-1, cholesterol is a conserved component of all lipid rafts and can be stained with filipin since filipin binds to cholesterol and is UV autofluorescent. No colocalization of NTHI was observed with filipin at 4 and 24 h postinoculation (Fig. 6A).

FIG 6.

Confocal microscopy and adherence-invasion assays with pharmacologic inhibitors. Lipid raft-independent invasion of bronchial epithelial cells by NTHI. (A) Confocal microscopy. At 4 and 24 h postinoculation of H292 cells, NTHI does not colocalize with vesicles positive for the following markers of lipid rafts: caveolin-1 (shown at 24 h), flotillin-1 (shown at 24 h), and cholesterol (shown at 4 h). NTHI was labeled with anti-11P6H antibody conjugated to Alexa Fluor 488 (green). Caveolin-1 was labeled with anti-caveolin-1 antibody and goat anti-mouse antibody conjugated to Alexa Fluor 568 (red). Flotillin-1 was labeled with anti-flotillin-1 antibody and goat anti-mouse antibody conjugated to Alexa Fluor 568 (red). Cholesterol was labeled with filipin (blue), and those samples were also labeled with anti-human secretory component antibody and donkey anti-goat antibody conjugated to Alexa Fluor 568 (red) to provide additional visualization of the plasma membrane. (B) Confocal microscopy. Filipin (FLP) and nystatin (NST) inhibit the internalization of fluorescently conjugated cholera toxin B subunit (white arrows), a known cargo of lipid raft-mediated endocytosis. H292 cells were pretreated with sRPMI containing inhibitor or inhibitor diluent, followed by addition of the cargo conjugate, incubation for 20 min on ice, and incubation for 2 h at 37°C. (C) Adherence-invasion assays. Filipin and nystatin do not inhibit invasion by NTHI. H292 cells were pretreated with inhibitor or inhibitor diluent for 2 h, followed by a 4-h infection conducted according to the adherence-invasion assay protocol. Data are normalized to samples in the absence of inhibitor. Error bars represent standard errors of the means of three independent experiments. A paired Student t test was used to calculate statistical significance. *, P ≤ 0.05; **, P ≤ 0.005.

For adherence-invasion assays with pharmacologic inhibitors, we selected nystatin and filipin, which inhibit lipid raft-mediated internalization by binding to cholesterol. To confirm that nystatin and filipin effectively inhibit lipid raft-mediated internalization under the conditions of this assay, positive controls were prepared using a fluorescent conjugate of cholera toxin B subunit, a known cargo of lipid raft-mediated internalization. Confocal microscopy confirmed that filipin, and perhaps nystatin, inhibited lipid raft-mediated internalization under the conditions of this assay (Fig. 6B).

Nystatin and filipin each had no effect on invasion by NTHI (Fig. 6C). Interestingly, filipin inhibited adherence by the ΔigaA ΔigaB mutant (Fig. 6C). Although the significance of this is unclear, it is possible that bound filipin affects the distribution or availability of host plasma membrane molecules that interact with NTHI IgA1 proteases or adhesins. The absence of such an effect in cells treated with nystatin could be due to the distinct properties of these molecules, including their sizes and mechanisms of engaging cholesterol (46). Together, the absence of NTHI colocalization with lipid raft markers in confocal microscopy images and the inability of filipin to inhibit NTHI invasion indicate that lipid rafts play no significant role in the internalization of NTHI.

NTHI is trafficked via endolysosomes and is killed in lysosomes.

Endosomes are commonly trafficked via the endolysosomal system. To determine whether NTHI colocalizes with endolysosomal markers, infected H292 cells were examined for bacterial colocalization with early endosomal antigen 1 (EEA1) and LAMP1. Distinct EEA1-positive vesicles and LAMP1-positive vesicles containing NTHI were observed at 4 and 24 h postinoculation (Fig. 7A). These observations indicate that NTHI is trafficked via the endolysosomal pathway, and the survival assay data indicate that NTHI is killed over time, presumably in lysosomes.

FIG 7.

Confocal microscopy and survival assay with pharmacologic inhibitor of lysosome acidification. NTHI traffics to early endosomes and lysosomes and is killed in lysosomes. IgA1 proteases are required for optimal survival in lysosomes. (A) Confocal microscopy. At 4 and 24 h postinoculation of H292 cells, NTHI is found within vesicles positive for EEA1 (shown at 4 h) and within vesicles positive for LAMP1 (shown at 24 h). NTHI was labeled using anti-11P6H antibody conjugated to Alexa Fluor 488 (green). EEA1 was labeled using anti-EEA1 antibody and goat anti-mouse antibody conjugated to Alexa Fluor 568 (red). LAMP1 was labeled using anti-human LAMP1 antibody and goat anti-mouse antibody conjugated to Alexa Fluor 568 (red). (B) Survival assay in the presence and absence of concanamycin A (CMA). H292 cells were infected for 16 h, treated with gentamicin for 1 h, and treated with concanamycin or diluent for 3 h. Survival data are normalized to samples treated with diluent. Error bars represent standard errors of the means of three independent experiments. A paired Student t test was used to calculate statistical significance. *, P ≤ 0.05; **, P ≤ 0.005.

To determine whether NTHI is killed in lysosomes, H292 cells were infected for 16 h and treated with concanamycin A, a specific inhibitor of vacuolar-type H⁺-ATPases responsible for lysosome acidification. Concanamycin A improved the survival of each strain of NTHI, with the greatest improvements being in the viabilities of the ΔigaB and ΔigaA ΔigaB mutants (Fig. 7B). Together, these data indicate that NTHI is trafficked to and killed in lysosomes and that both IgA1 proteases, but especially IgaB, are required for optimal intracellular survival, potentially by cleavage of LAMP1 and subsequent inhibition of lysosome acidification.

IgaB cleaves LAMP1 at pHs characteristic of the plasma membrane and endolysosomes.

To determine whether IgaA or IgaB cleaves LAMP1, LAMP1 cleavage assays were performed using purified rhLAMP1 incubated with supernatant that was collected from infected H292 cells and subsequently pH adjusted to a pH within the reported pH ranges of the lysosome (4.65), late endosome (5.15), early endosome (5.65), and plasma membrane (7.50) and to an intermediate pH (6.50). Western blots of incubation products probed for LAMP1 demonstrate that IgaB cleaves rhLAMP1 at pHs characteristic of endolysosomal trafficking and of the plasma membrane and that cleavage is most efficient at higher pHs (Fig. 8).

FIG 8.

LAMP1 cleavage assay. IgaB cleaves rhLAMP1 at pHs characteristic of the plasma membrane, early endosomes, late endosomes, and lysosomes. At 24 h postinoculation, medium from H292 cells infected with wild-type and mutant strains as indicated was filter sterilized and incubated with rhLAMP1 at 37°C overnight. Digestions were separated on protein gels, transferred to PVDF membrane, probed with anti-human LAMP1 antibody and goat anti-mouse antibody conjugated to HRP, and visualized by chemiluminescence for exposure to film. The large top band is intact rhLAMP1, and the smaller bands at the arrows are rhLAMP1 cleavage fragments.

pH of NTHI-containing vesicles.

We then hypothesized that NTHI is trafficked to acidified lysosomes and that IgaB-mediated cleavage of LAMP1 inhibits endolysosomal acidification. To test this hypothesis, each strain was stained with SNARF-1, a pH-sensitive fluorophore with biphasic emission, prior to infection of H292 cells. At 6 h postinoculation, samples were visualized using confocal microscopy, and pixel intensities were used to calculate the pH of bacteria adhering to the plasma membrane (Fig. 9A) and of bacteria in perinuclear vesicles (lysosomes) (Fig. 9B).

FIG 9.

SNARF-1 pH measurements. IgA1 proteases do not inhibit lysosome acidification. NTHI was stained with SNARF-1, a pH-sensitive fluorophore with dual emission wavelengths. Stained bacteria were used to infect H292 cells for 6 h. Immediately following infection, samples were examined using confocal microscopy. Image stacks were captured, and representative optical sections were selected for image analysis. Pixel intensities of each emission wavelength were quantified from images of calibrators (bacteria alone incubated in buffers of known pHs) and of SNARF-1-stained bacteria at the plane of the plasma membrane (A, C) and at the plane of lysosomes, which present as large perinuclear vesicles (B, C). Pixel intensities (Yellow/Red) were converted to pH measurements using a standard curve generated by the calibrators for each experiment. Infection pretreatments with concanamycin A (CMA) or diluent were used to control for lysosome acidification, and measurements of pH at the plasma membrane (orange shading) were included to serve as controls by providing data from which to draw comparisons. Lysosome pH values in the absence of CMA treatment are shown with yellow shading.

To test the usefulness of SNARF-1 in our infection model system, H292 cells were pretreated with or without concanamycin A prior to infection. As expected, in H292 cells treated with concanamycin A, the average pH of NTHI-containing vesicles was roughly the same as the pH of bacteria at the plasma membrane and higher than the pH of NTHI-containing vesicles in untreated cells (Fig. 9C). However, contrary to our hypothesis, the pH of lysosomes containing the wild type was no less acidic than the pH of lysosomes containing the ΔigaA ΔigaB mutant (Fig. 9C). We then performed the experiment using H292 cells infected with the wild type and each IgA1 protease mutant strain. The pH measurements of NTHI-containing vesicles were all within the reported ranges of lysosomal pH (Fig. 9C). These data indicate that IgA1 proteases do not inhibit the acidification of endolysosomes and that IgaB-mediated cleavage of LAMP1 does not mediate an increase in lysosomal pH under the conditions of this assay and model system.

DISCUSSION

Although NTHI is classically considered to be an extracellular pathogen, increasing evidence of intracellular NTHI and its possible correlation with symptomatic and chronic infection has fueled speculation that host intracellular environments provide temporary niches or even long-term reservoirs for chronic NTHI infection. Few studies have examined the NTHI invasion process and the responsible host and bacterial factors. In previous work, we identified a novel NTHI human IgA1 protease, igaB, which is associated with pathogenic clinical isolates of NTHI (36, 37). igaB is homologous to an IgA1 protease gene that is unique to pathogenic Neisseria and promotes its intracellular survival, which was correlated to cleavage of LAMP1 (43). We hypothesized that intracellular NTHI escapes or neutralizes the endolysosomal trafficking pathway in order to persist within human respiratory epithelial cells. We also hypothesized that both NTHI IgA1 proteases, but especially IgaB, are required for optimal invasion and intracellular persistence of NTHI. Here, we show that NTHI invades human bronchial epithelial cells in vitro in a lipid raft-independent manner, is subsequently trafficked via the endolysosomal pathway, and is killed in lysosomes after various durations of persistence. While IgaA is required for optimal invasion by NTHI, IgaB is required for optimal intracellular persistence by NTHI. IgaB cleaves lysosome-associated membrane protein 1 (LAMP1) at pHs characteristic of the plasma membrane, early endosome, late endosome, and lysosome. However, neither IgA1 protease inhibits acidification of intracellular vesicles containing NTHI. Thus, NTHI IgA1 proteases play important but different roles in NTHI invasion and trafficking in respiratory epithelial cells.

Intracellular survival assay data demonstrate that NTHI persists in bronchial epithelial cells in vitro for up to 96 h postinoculation (Fig. 4), representing a level of persistence that could possibly translate as enhanced colonization or as delayed clearance in an in vivo situation, particularly if the intracellular state of NTHI is temporary. For example, NTHI has been observed within and between subepithelial cells and macrophages, so in order to gain access to these locations, NTHI might traverse the epithelial barrier via transcytosis, requiring temporary residence in epithelial cells. IgA1 protease mutants of pathogenic Neisseria exhibited transcytosis deficiencies in vitro, although it is unknown whether this was indirectly due to undefined trafficking defects (47). Our survival assay data also indicate that IgaB is required for optimal persistence, suggesting that IgaB plays a role in the persistence of NTHI in the airways of adults with COPD, for example.

The adherence and invasion assay data demonstrate that IgaA plays an undefined role in adherence and is required for optimal invasion and that IgaB plays little or no role in adherence and invasion (Fig. 3). As invasion is a function of both bacterial invasion and intracellular survival, it is logical to consider that the reduced invasion of the ΔigaA mutant could be due to the possibility that this strain invades at the same rate as the wild-type strain but is killed more rapidly. However, this seems less likely given that the survival assay data show that the ΔigaA mutant dies more slowly than the wild-type strain (Fig. 4).

While it was previously shown that IgA1 cleavage fragments enhance adherence by Streptococcus pneumoniae, it is unprecedented and unclear why an IgA1 protease deletion mutation would enhance adherence by NTHI (48). However, the IgA1 cleavage assay data suggest that IgaA activity gradually increased through 24 h in our infection model system, such that adherence could be mediated by a competitor molecule at 1 and 4 h. Alternatively, the varied adherence data could be due to changes in the expression of other bacterial adhesins and corresponding host cell receptors. An inherent limitation of in vitro cell lines is that they can contain or lack factors that might or might not be present on human airway epithelial cells in vivo, which could also influence the adherence and invasion of NTHI in our model system.

Data from confocal microscopy and from adherence-invasion assays using filipin indicate that NTHI invasion of bronchial epithelial cells is lipid raft independent (Fig. 6). Previous studies of NTHI invasion of human alveolar epithelial cells have generated differing observations about the role of lipid rafts (20, 22). Additionally, internalization of NTHI outer membrane vesicles (OMVs) by human pharyngeal epithelial cells in vitro was shown to be lipid raft dependent, although it is conceivable that OMVs would have different properties than whole bacteria (8). Such variability illustrates the importance of controls and of varied model system components, such as bacterial strains, host cell types, and experimental conditions, in studies examining internalization and trafficking pathways. Carefully developed protocols and control experiments confirmed the activity and noncytotoxicity of pharmacologic inhibitors in our infection model system, supporting our conclusion that lipid rafts play no role in NTHI invasion of bronchial epithelial cells.

In light of current knowledge about eukaryotic internalization pathways, the lack of a role for lipid rafts in our infection model system lends weight to data suggesting that NTHI internalization is clathrin mediated, including data from studies of ChoP-PAFR-mediated internalization, from our adherence-invasion assay with chlorpromazine (Fig. 5), and from our confocal microscopy observations that NTHI is trafficked via endolysosomes (Fig. 7A) (24, 25). Regardless, but perhaps especially in the case of clathrin-mediated internalization, endosomes are commonly trafficked via the endolysosomal pathway. Interestingly, confocal microscopy showed no clear colocalization of NTHI with clathrin or within clathrin-coated vesicles (data not shown), suggesting that clathrin plays an indirect role in the internalization of NTHI, as was observed and suggested by data from a study of an invasive clone of Haemophilus influenzae biogroup aegyptius, the etiologic agent of Brazilian purpuric fever (49). We observed no visible differences in the degree of trafficking of the wild type and of each IgA1 protease mutant to endolysosomes, suggesting that IgA1 proteases do not mediate differential trafficking of NTHI.

Data from our rescue assays using concanamycin A, an inhibitor of lysosome acidification, indicate that NTHI is killed in lysosomes and suggest that IgA1 proteases, especially IgaB, improve NTHI survival in lysosomes (Fig. 7B). Indeed, our LAMP1 cleavage assays show that IgaB cleaves rhLAMP1 at pHs characteristic of the plasma membrane and endolysosomes (Fig. 8). However, the SNARF-1 experiments show that NTHI traffics to acidified lysosomes and that IgaB does not inhibit lysosome acidification in our infection model system (Fig. 9). These data could reflect the LAMP1 cleavage assay observation that IgaB-mediated cleavage of LAMP1 is less efficient at low pH (Fig. 8).

LAMP1 and LAMP2 comprise approximately 50% of all proteins in the late endosomal-lysosomal membrane. Single- and double-knockout models, respectively, indicate that while LAMP1 and LAMP2 can compensate for each other and, thus, likely have overlapping functions, LAMP deficiency inhibits lysosome maturation (42, 45). While it was previously unknown whether the host cell would be able to compensate for real-time cleavage of LAMP1 by IgA1 proteases, our data suggest that either the host cell is able to compensate or compensation is not necessary for lysosome acidification.

Studies of the IgaB homologue in pathogenic Neisseria have correlated IgA1 protease-mediated cleavage of LAMP1 with improved survival. However, because IgA1 protease cleavage of LAMP1 in vitro was less efficient at low pH than at neutral pH, it is hypothesized that cleavage takes place not in lysosomes but at the plasma membrane (43). Indeed, neisserial porin and pilin trigger endosome and lysosome exocytosis, redistributing LAMP1 to the plasma membrane, which is thought to reduce the total number of lysosomes in infected cells (50–52). Our data show that IgaB-mediated cleavage of LAMP1 is also more efficient at neutral pH, but we observed no redistribution of lysosomes in our infection model system.

Future studies will examine the dominance of IgaB activity in strains expressing both igaA and igaB. Our qRT-PCR data (Fig. 1) indicate that the dominance of IgaB-mediated cleavage observed in IgA1 cleavage assays (Fig. 2) is not due to a lack of igaA gene expression but is perhaps due instead to the greater abundance of the igaB transcript or to possible preferred translation of the igaB transcript, greater abundance of the IgaB protein, or preferred substrate affinity for IgaB. Our laboratory has identified clinical isolates with varied IgaA- and IgaB-mediated cleavage activities, such that gene- and protein-level analyses of these strains should provide additional insight into the regulation and clinical relevance of IgaA and IgaB.

Regulatory sequences are absent from gene regions upstream from igaA and igaB in NTHI strain 11P6H, and constitutive expression is characteristic of all IgA1 protease genes examined in detail (53). However, certain studies have called into question the constitutive nature of IgA1 proteases. For example, Neisseria gonorrhoeae was reported to increase its production of IgA1 protease during growth under iron-restricted conditions. Also, a tetranucleotide repeat region (5′-CAAT-3′) has been identified upstream from igaB in strain 2019, a strain that is also a clinical isolate obtained from a patient with COPD (54). This tetranucleotide repeat region promotes slipped-strand mispairing responsible for phase variation of genes in H. influenzae. In a novel experimental human nasopharyngeal colonization study, phase variation of igaB in strain 2019 shifted from phase-off to phase-on over the course of a 6-day colonization period compared to the phase of the inoculation strain (54). This observation indicates that there is a selection for igaB-on clones of NTHI in the human airway during early colonization and provides additional evidence that IgaB is important for NTHI colonization and persistence.