Rhinovirus infection liberates planktonic bacteria from biofilm and increases chemokine responses in cystic fibrosis airway epithelial cells

Abstract

Background

Intermittent viral exacerbations in cystic fibrosis (CF) patients with chronic P. aeruginosa (PA) infection are associated with increased bacterial load. A few clinical studies suggest that rhinoviruses (RV) are associated with majority of viral-related exacerbations in CF and required prolonged intravenous antibiotic treatment. These observations imply that acute RV infection may increase lower respiratory symptoms by increasing planktonic bacterial load. However, the underlying mechanisms are not known.

Methods

Primary CF airway epithelial cells differentiated into mucociliary phenotype were infected with mucoid PA (MPA) followed by RV and examined for bacterial density, biofilm mass, levels of chemokines and hydrogen peroxide (H₂O₂). Requirement of dual oxidase 2 in RV-induced generation of H₂O₂ in CF cells was assessed by using gene-specific siRNA.

Results

Super infection with RV increased chemokine responses in CF mucociliary-differentiated airway epithelial cells with pre-existing MPA infection in the form of biofilm. This was associated with the presence of planktonic bacteria at both the apical and basolateral epithelial cell surfaces. Further, RV-induced generation of H₂O₂ via dual oxidase 2, a component of NADPH oxidase in CF cells was sufficient for dispersal of planktonic bacteria from biofilm. Inhibition of NADPH oxidase reduced bacterial transmigration across mucociliary-differentiated CF cells and IL-8 response in MPA and RV-infected cells.

Conclusion

We show that acute infection with RV liberates planktonic bacteria from biofilm. Planktonic bacteria, which are more proinflammatory than their biofilm counterpart stimulates increased chemokine responses in CF airway epithelial cells, which in turn may contribute to pathogenesis of CF exacerbations.

Pulmonary manifestations due to chronic endobronchial infection are the leading cause of morbidity and mortality in individuals with cystic fibrosis (CF). P. aeruginosa (PA), a principal pathogen in CF, chronically colonizes the airways and often exists in a biofilm matrix. Despite chronic infection, CF patients experience acute exacerbations only periodically, indicating that biofilm bacteria may act as a resilient reservoir for planktonic bacteria, rather than a trigger for exacerbations. Consistent with this hypothesis, increased PA density was observed during episodes of acute exacerbations and required antibiotic treatment (reviewed in ¹). Further, PA isolated from sputa collected during exacerbations was similar to colonizing flora by genotypic analysis, suggesting a clonal expansion of bacteria from the biofilm reservoir ².

The significance of respiratory viral infections in the pathogenesis of asthma and COPD exacerbations has long been recognized. Respiratory viral infection is implicated in 44 to 80% of asthma exacerbations and 46–50% of COPD exacerbations ³. Similarly, improved method of viral detection suggested a significant role for viral infections in CF exacerbations ^4–7. Respiratory viruses associated with CF exacerbation include influenza viruses A/B, rhinoviruses, respiratory syncytical viruses, parainfluenza viruses and adenoviruses.

Rhinoviruses (RV), which cause common cold are also responsible for majority of viral-related exacerbations in asthma and COPD ³. A handful of clinical studies suggested that RV is also associated with majority of viral-related exacerbations in CF ^4,7–9. In one study, patients with RV infection required prolonged intravenous antibiotic treatment ¹⁰. In another study, RV infection was associated with increased use of antibiotics, prolonged hospitalization and decline in lung function ⁸. Similarly, respiratory syncytial virus infection in patients with intermittent or chronic PA infection were associated with increased psuedomonal antibody levels ¹⁰. These circumstantial evidences suggest that viral infection may increase planktonic bacterial load. This may in turn stimulate an intense inflammatory response and increase clinical symptoms that can be reduced by treatment with antibiotics. Consistent with this notion, in vitro studies suggested a more intense host inflammatory response to motile planktonic bacteria ^11,12. However, the mechanisms by which viral infections increase density of planktonic bacteria are not well known.

In the present study, we examined whether super infection with RV liberates planktonic bacteria and increases chemokine responses of CF airway epithelial cells pre-infected with mucoid PA. Further, we also investigated the mechanisms by which RV causes dispersal of planktonic bacteria from biofilm.

MATERIALS AND ETHODS

For additional detail, please see the online data supplement.

Rhinovirus

Rhinovirus serotype 39 was purchased from ATCC (Manassas, VA). Viral stocks were generated and tissue culture infectious dose (TCID₅₀) was determined as described ¹³.

Bacteria and growth conditions

A clinical isolate of mucoid P. aeruginosa (MPA) was grown in tryptic soy broth (BD Diagnostics, Sparks, MD), and suspended in PBS.

CF cell cultures and infection

Primary CF airway epithelial cells were obtained from 3 CF patients and grown either at air-liquid interface to promote mucociliary differentiation or as monolayers ¹⁴. Use of CF bronchial segments was reviewed by the University of Michigan Institutional Review Board. Mucociliary-differentiated cells were sequentially infected with MPA (at 0.01 MOI) followed by RV (1 × 10⁶ TCID₅₀), UV-irradiated RV or sham. Chemokine levels in the basolateral medium were determined by ELISA (R&D systems, Minneapolis, MN). In some experiments, cells were preincubated with diphenylene iodonium (DPI) or 1140W (both from Sigma-Aldrich, St. Louis, MO) as indicated in the Results section. Cells treated with 0.1% DMSO or media served as controls. Levels of lactose dehydrogenase, (an index of cytotixcity) was measured in the basolateral medium or cell culture supernatants from CF cell monolayers by using Cytotox⁹⁶ non-radioactive cytotoxicity kit (Promega, Madison, WI).

Dispersal of biofilm

MPA biofilms were formed on the pegs of MBEC assay system (Innovatech, Edmonton, CA) as described ¹⁵. Briefly, the device consists of 96 conical pegs attached to a plastic lid and rests on a 96 well plate containing bacterial suspension (OD₆₀₀ of 0.01) diluted in Lauria-Bertani broth and incubated for 48h. The pegs with biofilm were rinsed with sterile PBS and exposed to conditioned media from CF airway epithelial cells for 30 minutes. The pegs were washed and stained with 0.2% crystal violet. Crystal violet bound to pegs was eluted with methanol and the optical density was measured at 590 nm to quantify the biofilm mass.

Measurement of H₂O₂

H₂O₂ produced from cells was measured using the Amplex red hydrogen peroxide/peroxidase kit (Invitrogen, Carlsbad, CA).

Transfection of cells

CF cells were transfected with Duox2 siRNA (Dharmacon, Inc., Lafayette, CO) or non-targeting (NT) siRNA using Lipofectamine^™ RNAiMAX (Invitrogen) by reverse transfection method following manufacturer’s instructions. Reduction in expression of Duox2 was determined by real time PCR in RV-infected cells, using the following primers. Forward primer 5′-aac cta agc agc tca caa ct-3′ and reverse primer 5′-cag aga gca atg atg gtg at-3′. Specificity of the product was determined by melting curve analysis.

Confocal Microscopy

Cultures were fixed in methanol and incubated with a mixture of antibodies to P. aeruginosa (Abcam, Cambridge, MA)and zona occludins (ZO)-1 (BD-biosciences, San Jose, CA).

Scanning electron microscopy

Cell cultures were processed for scanning electron microscopy as described previously ¹⁴.

Statistical analysis

Results are expressed as mean± SEM or median with range. To compare groups, one-way analysis of variance (ANOVA) with Tukey-Kramer post-hoc analysis or ANOVA based on ranks with Dunn’s post-hoc analysis was performed, as appropriate.

RESULTS

Super infection with RV increases chemokine responses in MPA-infected CF airway epithelial cells

CF airway epithelial cells differentiated into mucociliary phenotype were infected with MPA or treated with PBS. Twenty four hours later, cells were infected with sham, RV or replication deficient UV-irradiated RV (UV-RV). Compared to cell cultures treated with PBS, cultures infected with MPA, RV or UV-RV showed significant increases in IL-8, GRO-α and ENA-78 (Figure 1A, 1B and 1C). Cell cultures co-infected with MPA/RV, showed synergistic increases in the levels of all three chemokines measured compared to MPA or RV-alone infected cells. Such synergistic increase in chemokines was not observed in cells infected with MPA/UV-RV, indicating that infectious virus is required for the observed increases.

Figure 1.

RV synergistically increases chemokine expression in MPA infected cells. Mucociliary-differentiated CF airway epithelial cells were infected apically with MPA or treated with PBS for 24h. Cells were then infected with RV and incubated for another 24 h. Sham or UV-RV- infected cells were used as controls. Chemokine levels were measured in the basolateral medium. A, IL-8; B, Gro-α; and C, ENA-78. Data represents median and range calculated from 2 independent experiments, performed in duplicates on cells obtained from 3 CF donors. Numbers within each panel represents medians calculated from 6 replicates for MPA, RV or MPA/RV- infected cells to show the synergistic increases. (* different from PBS-treated group, p ≤0.05; †different from all other treatment groups, p ≤0.05, ANOVA on ranks with Dunn’s post-hoc analysis).

Small increases in both bacterial and viral loads was observed in cells co-infected with MPA and RV compared to cells infected with MPA alone or MPA/UV-RV (Supplemental Tables 1 and 2). Together, these results suggest that the observed synergistic or additive chemokine responses depend on other factors besides small increases in bacterial and/or viral load.

RV-infection facilitates transmigration of P. aeruginosa across mucociliary-differentiated CF airway epithelial cells

Cells infected with MPA/UV-RV or MPA/RV were immunostained with antibody to P. aeruginosa and zona occludin (ZO)-1, a component of tight junction, and subjected to confocal indirect immunofluorescence microscopy. Basolateral media from these cultures were plated to determine the number of translocated bacteria. Confocal microscopy showed normal distribution of ZO-1 in the periphery of the cells and the presence of bacterial microcolonies (appearing as a green haze) with very few individual bacteria on the apical surface of MPA/UV-RV-infected cells (Figure 2A). In contrast, MPA/RV-infected cell cultures showed dissociation of ZO-1 from the periphery of cells in some areas, and the presence of bacterial microcolonies along with numerous individual bacteria on the apical surface (Figure 2B). Z-sections of these cultures revealed that while bacteria were mainly on the apical surface in MPA/UV-RV-infected cultures (Figure 2C), they were observed between the cells and at the basolateral surface in MPA/RV-infected cultures (Figure 2D). Further, bacteria were recovered consistently (4.2 × 10⁴ ± 1.6 × 10⁴ CFU/ml) in the basolateral media obtained from MPA/RV-infected cells, but not from MPA/UV-RV- infected cells. These results indicate that secondary infection with RV facilitates dispersal of bacteria from the microcolonies and transmigration of dispersed bacteria across the differentiated CF cultures and this may potentially increase interaction of bacteria with basolateral receptors.

Figure 2.

RV facilitates transmigration of MPA across CF airway epithelial cells. Well-differentiated CF airway epithelial cells were sequentially infected with MPA followed by UV-RV (A and C) or RV (B and D). Cells were immunostained with antibodies to mixture of P.aeruginosa (green) and ZO-1 (red) and counterstained with DAPI (blue). Images are representative of three independent experiments. A and B, Apical view of cell cultures; C and D, Z-section generated from the area marked with white line in A and B respectively. Arrows in panel A and B represent bacterial microcolonies, Arrowhead and * in panel B represent 2 or 3 bacteria in a group and dissociation of ZO-1 from the periphery of cells respectively.

RV infection causes dispersal of bacteria from MPA biofilm

Scanning electron microscopy of MPA/UV-RV-infected mucociliary differentiated cell cultures revealed a film covering majority of the bacteria, suggestive of presence of bacteria in a biofilm (Figure 3A). In contrast, MPA/RV-infected cultures showed hollows in the biofilm which were surrounded by bacteria that were not covered by a film, an indication of dispersal of bacteria from the biofilm (Figure 3B). These results imply that RV infection can disperse planktonic bacteria form biofilm.

Figure 3.

RV disperses bacteria from MPA biofilm. Well-differentiated CF airway epithelial cells were sequentially infected with MPA followed by UV-RV (A) or RV (B) and subjected to scanning electron microscopy. * and arrowhead represents MPA biofilm and individual bacteria; arrows in panel B points to hollow in the biofilm surrounded by bacteria not embedded in the biofilm, an indication of dispersed bacteria from biofilm. Images are representative of three independent experiments. C. MPA biofilm grown on plastic pegs of MBEC device were incubated with PBS, Sham, RV inoculum, or UV-RV and biofilm mass determined. D. MPA biofilm grown on plastic pegs of MBEC device were incubated with conditioned media from CF airway epithelial cells exposed to Media, sham, RV or UV-RV and biofilm mass determined. Data represent average ± SEM calculated from 4 independent experiments, performed in quadriplicates (*different from CM from cells exposed to media, sham and UV-RV, p ≤0.05; ANOVA with Tukey-Kramer post-hoc analysis).

To determine whether RV is sufficient to disperse bacteria from biofilm, MPA biofilm grown on plastic pegs of an MBEC device (Innovotech, Edmonton, CA) was exposed to PBS, cell culture media, RV inoculum equivalent to 1 × 10⁶ TCID₅₀ or UV-RV for 3 h, and the biofilm mass left on the peg was quantified. None of these treatments affected the biofilm mass (Figure 3C), suggesting that a factor(s) secreted by epithelial cells upon RV infection is responsible for dispersal of bacteria from biofilm. This led us to examine the effect of conditioned media from RV- or UV-RV- infected cells on biofilm density. In the subsequent experiments, monolayers of CF primary cells were used, because mucus present on the apical surface of differentiated cells interfered with the assay. Exposure of MPA biofilm grown on plastic pegs to conditioned media from RV-infected, but not from UV-RV-infected cells decreased biofilm density (Figure 3D). These results indicate that a factor(s) released from CF cells in response to infectious RV is responsible for biofilm disruption.

RV-induced H₂O₂ generation is sufficient for disruption of biofilm

RV has been shown induce oxidative stress ^16,17 and such environmental change can cause disruption of bacterial biofilm ^18–20. Therefore, we examined the capacity of conditioned media from CF cells infected with RV or UV-RV in the presence of antioxidants to disrupt MPA biofilm. Treatment with diphenylene iodonium (DPI), an inhibitor of NADPH oxidase completely reversed the capacity of conditioned media from RV-infected cells to disrupt biofilm (Figure 4A; Supplemental Figure 1A). In contrast, conditioned media from cells infected with RV in the presence of 1140W, an inhibitor of inducible nitric oxide synthase, had no affect on bioflm mass (Supplemental Figure 1C). Neither DPI nor 1140W alone had effect on biofilm mass. Also, these inhibitors did not affect cell viability or viral load in the cells (Supplemental Figure 1E and 1F). These results indicate that DPI-inhibitable oxidative stress caused by RV is responsible for dispersion of biofilm bacteria.

Figure 4.

Generation of H₂O₂ in CF airway epithelial cells in response to RV-infection is responsible for bacterial dispersal from biofilm. A. Conditioned media from monolayers of CF airway epithelial cells infected with UV-RV or RV in the presence of 0 (−) or 5 μM (+) DPI was incubated with MPA biofilm and biofilm mass determined. B. CF airway epithelial cells infected with UV-RV or RV or treated with media in the presence of DPI (0 or 5 μM) and H₂O₂ present in the media was measured. C. CF cells were transfected with non-targetting (NT) or Duox2 siRNA, infected with RV or UV-RV and expression of Duox2 was determined by immunoblot analysis, representative three independent experiments. D. Levels of H₂O₂ in media from monolayers of CF cells transfected with siRNA specific to Duox2 or NT siRNA and infected with UV-RV or RV. E. MPA biofilm grown on plastic pegs was exposed to conditioned media from cells transfected with Duox2 siRNA or NT siRNA followed by UV-RV or RV infection and biofilm mass was quantified. Cells treated with media served as control. F. MPA biofilm was exposed to conditioned media from RV-infected cells or H₂O₂ (10 μM) in the presence or absence of catalase and biofilm mass was measured. Data represent average ± SEM calculated from 4 independent experiments, performed in duplicates or quadriplicates (* different from respective control p ≤0.05; † different from RV-infected group in the absence of DPI or NT siRNA-transfected cells infected with RV, ANOVA with Tukey-Kramer post-hoc analysis).

DPI inhibits flavoenzymes including NADPH oxidases. RV infection increased mRNA expression of dual oxidase 2 (Duox2), a component of NADPH oxidase by 5.6 ± 1.2 fold compared to sham or UV-RV infected cells. Duox2 generates and secretes H₂O₂ directly into extracellular milieu. In addition, RV infection also stimulated the generation of H₂O₂ in CF airway epithelial cells (12.6±5.81 μM), and this was inhibited by DPI (Figure 4B; Supplemental Figure 1B). To determine whether Duox2 stimulated by RV infection is required for generation of H₂O₂, monolyaers of CF cells were transfected with non-targetting (NT) siRNA or siRNA specific to Duox2. CF cells transfected with siRNA specific to Duox2, but not NT siRNA, blocked the expression of Duox2 (Figure 4C and supplemental Figure 1) and inhibited H₂O₂ generation from RV-infected cells (Figure 4D). In addition, conditioned media from Duox2 siRNA infected with RV was attenuated in its capacity to disrupt biofilm (Figure 4E). This was not due to change in RV load in Duox2 siRNA transfected cells as there was no difference in RV load between NT-siRNA and Duox2 siRNA transfected cells (Supplemental Figure 1C).

To further confirm that H₂O₂ induced by RV contributes to biofilm disruption, initially we determined the effect of H₂O₂ on the biofilm mass grown on plastic pegs. H₂O₂ decreased the biofilm mass in a concentration dependent manner (Supplemental Figure 2). Catalase, which neutralizes H₂O₂, inhibited the reduction of biofilm mass caused by both H₂O₂ and conditioned media from RV-infected cells (Figure 4F). These results confirmed that H₂O₂ generation induced by RV is required for disruption of biofilm.

DPI inhibits bacterial transmigration and IL-8 increase in differentiated CF cells super infected with RV

To assess whether H₂O₂ induced by RV is required for the bacterial transmigration and observed synergistic increases in IL-8 production, well-differentiated CF airway epithelial cells were infected with PBS or MPA and incubated for 24h. Cells were then infected with UVRV or RV and incubated in the presence 0 or 5 μM DPI and the basolateral media was examined for IL-8 levels and the presence of bacteria. As observed earlier, in the absence of DPI, MPA/RV-infected cultures showed synergistic increases in IL-8 (Figure 5A) and bacteria in the basolateral media (Figure 5B). In contrast, MPA/RV-infected cells incubated in the presence of DPI showed decreased IL-8 response which was similar to IL-8 response stimulated by MPA alone and very few bacteria in the basolateral chamber. Confocal microscopy of cultures immuno-stained with antibodies to P. aruginosa and ZO-1 showed presence of biofilm along with numerous individual bacteria on the apical surface of MPA/RV-infected cell cultures in the absence of DPI (Figure 5C). Z-sections of these cultures revealed presence of bacteria both in the apical and basolateral surface (Figure 5E). On the other hand, DPI treated MPA/RV-infected cultures showed predominantly biofilm bacteria with very few individual bacteria on the apical surface (Figure 5D) and Z-sections (Figure 5F) showed bacteria in the basolateral surface rarely. Together these results indicate that RV-induced DPI-inhibitable NADPH oxidase activity is responsible for dispersal of planktonic bacteria from biofilm, transmigration of bacteria and the related synergistic increase in IL-8 in CF airway epithelial cells.

Figure 5.

DPI inhibits bacterial transmigration and reduces IL-8 response in mucociliary differentiated cultures sequentially infected with MPA followed by RV. Mucociliary-differentiated CF cultures were infected with MPA. Twenty four hours later cultures were infected with sham, UV-RV or RV in the presence of DPI (0 or 5 μM) for another 24 h. IL-8 concentration in the basolateral media was determined by ELISA. B. Number of bacteria in the basolateral chamber was determined by plating. Data represent average ± SEM or median with range from 3 independent experiments, performed in duplicates (* different from Media or UV-RV-treated control groups p ≤0.05; **significantly different from respective RV-infected group p ≤0.05, ANOVA with Tukey-Kramer post-hoc analysis or ANOVA on ranks with Dunn’s post-hoc analysis). C to F Representative confocal images of cell cultures infected with MPA/RV in the presence of 0 or 5μM DPI respectively showing bacteria (green), ZO-1 (red) and nuclei (blue). C and D. Apical view of cell cultures; E and F. Z sections generated from an area marked with white line in E and F respectively.

DISCUSSION

In the present study, we demonstrate that super infection with RV stimulates robust chemokine responses from CF airway epithelial cells pre-infected with MPA. We also show that increased cytokine responses is associated with RV-facilitated dispersal of bacteria from MPA biofilm and transmigration of planktonic bacteria from the apical to basolateral surface of mucociliary-differentiated CF airway epithelial cells. Further, we provide evidence that RV infection-induces H₂O₂ production in CF airway epithelial cells and this in turn causes dispersal of planktonic bacteria from biofilm. Together, these results suggest that acute super infection with RV increases chemokine responses and planktonic bacteria by causing dispersing biofilm.

RV stimulates oxidative stress and chemokine responses from airway epithelial cells ^16,21–23. Therefore, it is plausible that RV-induced subtle change in the microenvironment of CF airways is sufficient to increase pathogenesis of existing flora. Consistent with this notion, RV induced mild oxidative stress appears to be sufficient for liberation of plnaktonic bacteria from the biofilm. Planktonic bacteria are motile and express virulence factors and hence can stimulate inflammatory responses readily compared to relatively immotile biofilm bacteria ¹². In CF airways, MPA persists in biofilm and is less stimulatory to airway mucosa due to ‘Sandwich binding’ ²⁴. Therefore, dispersal of planktonic bacteria from the biofilm caused by RVinfection may lead to increased chemokine responses.

Exposure to oxidative or other environmental stress, bacteria within the biofilm undergoes coordinated dispersal events releasing free-swimming planktonic bacteria ^18–20. RV which can stimulate expression of NO and H₂O₂ ^16,17,22,25, therefore can disperse planktonic bacteria from the biofilm. In the present study, we found that RV-stimulated H₂O₂ mediated by Duox2, but not NO is responsible for RV-induced liberation of planktonic bacteria from biofilm. Duox2 is one of the six homologues of NADPH oxidase catalytic phox subunit gp91^{phox 26} and bronchial epithelial cells express Duox2 in response to treatment with Th1 and Th2 cytokines, TLR3 agonist or infection with RV ²⁷. Duox2 is located in the plasma membrane and can secrete H₂O₂ directly into the extracellular milieu^26,28, and thus can aid dispersal of bacteria from biofilm. Consistant with this notion, either inhibition of Duox2 expression, or catalase which neutralizes the H₂O₂, abrogated the capacity of conditioned media from RV-infected cells to disrupt biofilm, suggesting that Duox2-mediated H₂O₂ generation plays a role in RV caused dispersal of planktionic bacteria from the biofilm. Lack of NO generation in CF cells in response to RV infection, may be due to inducible nitric oxide synthase deficiency in these cells ^29,30.

In addition to dispersing bacteria from the biofilm, RV can also facilitate transmigration of bacteria across the airway epithelium, because RV has capacity to compromise barrier function ^31,32. This in turn may promote new interactions of dispersed planktonic bacteria with basolateral receptors which are otherwise not accessible, leading to increased production of cytokines from airway epithelial cells.

It is well established that bacteria in biofilms are more resistant to antibiotics than planktonic bacteria³³. In this regard, one can argue that disrupting biofilms to liberate planktonic bacteria that are more easily killed by antibiotics should be beneficial in CF, where bacteria are thought to persist in biofilm. In addition RV-induced H₂O₂ should also be beneficial in CF, because H₂O₂ treatment has been proposed to kill planktonic bacteria²⁰. However, planktonic bacteria dispersed from biofilm are often as resistant to antibiotics as their biofilm counterparts ³⁴ and also express antioxidant enzymes such as alkyl peroxidase and catalase, each of which can neutralize H₂O₂ in the extracellular milieu³⁵. Therefore, it is conceivable that RV-mediated dispersal of planktonic bacteria from biofilm increases inflammation in CF, requiring prolonged hospitalization and use of intravenous antibiotics.

In conclusion, our results suggest that RV-induced oxidative stress liberates planktonic bacteria from the biofilm, and facilitates transmigration of planktonic bacteria across the mucociliary differentiated CF airway epithelial cells. This in turn may increase interaction of planktonic bacteria with the basolateral receptors, leading to increased production of chemokines. These findings provide a novel mechanism by which RV may increase planktonic bacterial load and chemokine levels in CF.