Both respiratory syncytial virus (RSV) and influenza A virus induce nucleotide/P2Y purinergic receptor-mediated impairment of alveolar fluid clearance (AFC), which contributes to formation of lung edema. Although genetically dissimilar, both viruses generate double-stranded RNA replication intermediates, which act as Toll-like receptor (TLR)-3 ligands. We hypothesized that double-stranded RNA/TLR-3 signaling underlies nucleotide-mediated inhibition of amiloride-sensitive AFC in both infections. We found that addition of the synthetic double-stranded RNA analog poly-inosinic-cytidylic acid [poly(I:C)] (500 ng/ml) to the AFC instillate resulted in nucleotide/P2Y purinergic receptor-mediated inhibition of amiloride-sensitive AFC in BALB/c mice but had no effect on cystic fibrosis transmembrane regulator (CFTR)-mediated Cl− transport. Poly(I:C) also induced acute keratinocyte cytokine-mediated AFC insensitivity to stimulation by the β-adrenergic agonist terbutaline. Inhibitory effects of poly(I:C) on AFC were absent in TLR-3−/− mice and were not replicated by addition to the AFC instillate of ligands for other TLRs except TLR-2. Intranasal poly(I:C) administration (250 μg/mouse) similarly induced nucleotide-dependent AFC inhibition 2–3 days later, together with increased lung water content and neutrophilic inflammation. Intranasal treatment of BALB/c mice with poly(I:C) did not induce airway hyperresponsiveness at day 2 but did result in insensitivity to airway bronchodilation by β-adrenergic agonists. These findings suggest that viral double-stranded RNA replication intermediates induce nucleotide-mediated impairment of amiloride-sensitive AFC in both infections, together with β-adrenergic agonist insensitivity. Both of these effects also occur in RSV infection. However, double-stranded RNA replication intermediates do not appear to be sufficient to induce either adenosine-mediated, CFTR-dependent Cl− secretion in the lung or severe, lethal hypoxemia, both of which are features of influenza infection.
- alveolar fluid clearance
- airway hyperresponsiveness
the interface between the respiratory epithelium and the air in the lungs is normally bathed by a thin layer of airspace lining fluid. To permit efficient alveolar gas exchange and effective mucociliary clearance in the airways, the depth of this layer must be tightly regulated (37). The bronchoalveolar epithelium clears excess fluid by active, amiloride-sensitive, vectorial Na+ transport. Inhibition of this Na+-dependent alveolar fluid clearance (AFC) mechanism can result in formation of an excessive volume of intrapulmonary fluid, impairment of gas exchange (56), narrowing of airway lumens (59), and dilution of the surface-active materials that stabilize small airways (8), all of which can contribute to small airway obstruction and hypoxemia.
We have found that both the paramyxovirus respiratory syncytial virus (RSV) (12) and the orthomyxovirus influenza A virus (57) induce nucleotide/P2Y purinergic receptor-mediated impairment of amiloride-sensitive AFC in BALB/c mice. Interestingly, influenza A virus (but not RSV) also stimulates adenosine-mediated Cl− secretion via the cystic fibrosis transmembrane regulator (CFTR) anion channel, whereas RSV (but not influenza virus) induces both bronchoalveolar epithelial and airway insensitivity to β-adrenergic agonists (13, 55). However, the viral determinants and initial signaling pathways underlying these pathophysiological effects remain undefined. Although genetically dissimilar, these viruses may induce partially homologous functional responses because double-stranded RNA intermediates generated during replication of both pathogens can act as pathogen-associated molecular patterns, which are recognized by a variety of pattern-recognition receptor systems in respiratory epithelial cells (23). Interestingly, the synthetic double-stranded RNA analog poly-inosinic-cytidylic acid [poly(I:C)], which can act as a ligand for the pattern-recognition receptor Toll-like receptor-3 (TLR-3), also stimulates ATP release from respiratory epithelial cells in vitro (31) and induces pulmonary edema in vivo (53) although effects of double-stranded RNA on AFC and pulmonary β-adrenergic agonist responsiveness have not been reported. On the basis of these findings and our observation that nucleotide-mediated AFC impairment is only triggered by replication-competent viruses (12, 57), we hypothesized that viral double-stranded RNA replication intermediates detected by TLR-3 might act as a common inducer of nucleotide release (and its downstream effects on AFC) for both viruses. The purpose of the present study was therefore to determine whether the double-stranded RNA analog poly(I:C) can induce TLR-3-dependent AFC inhibition in mice and to characterize the degree of concordance between the effects of treatment with poly(I:C), infection with RSV, and infection with influenza virus on other aspects of murine lung function, to identify potential common pathways of action of these important respiratory pathogens.
MATERIALS AND METHODS
Except where noted, all reagents were from Sigma-Aldrich (St. Louis, MO). Amiloride, CFTRinh-172 (EMD Biosciences, La Jolla, CA), forskolin, A77–1726 (EMD Biosciences), fluoxetine, and BAPTA-AM were reconstituted in DMSO, aliquoted, and stored frozen. Rho kinase inhibitor (ROCKi) (EMD Biosciences), XAMR-0721 (EMD Biosciences), poly(I:C) (Invivogen, San Diego, CA), other TLR ligands (all Invivogen), anti-keratinocyte cytokine (KC) monoclonal antibody (MAB453; R & D Systems, Minneapolis, MN), rat IgG2A (MAB006, R & D Systems), and all other reagents were reconstituted in normal saline and used fresh.
Eight- to ten-week-old specific pathogen-free BALB/cAnNCr mice (National Cancer Institute, Frederick, MD) of either sex were used in these studies as indicated. Animals were maintained in autoclaved microisolators, given sterile food and water ad libitum, monitored daily for signs of respiratory distress or other illness, and were euthanized if this was detected. For all studies, data for each experimental group were derived from a minimum of two independent experiments. All studies were approved by The Ohio State University Institutional Animal Care and Use Committee.
Homozygous TLR-3−/− founders on a C57BL/6 background (B6;129S1-Tlr3tm1Flv/J) were obtained from JAX (Strain 005217; Jackson Laboratories, Bar Harbor, ME). Progeny genotype was confirmed in accordance with the vendor's instructions.
Alveolar fluid clearance measurements.
AFC was measured as previously described (10, 12, 57). Briefly, mice were anesthetized with valium (5 mg/kg body wt) followed by ketamine (200 mg/kg) 6 min later, and tracheotomized, and a trimmed sterile 18-gauge intravenous catheter was inserted caudally into the tracheal lumen. Following administration of pancuronium (0.08 μg/kg ip), each mouse was placed on a Deltaphase isothermal heating pad (Braintree Scientific, Braintree, MA) and ventilated on a Model 687 volume-controlled mouse ventilator (Harvard Apparatus, Holliston, MA), with a 200-μl tidal volume (8 ml/kg body wt), on 100% O2, at 160 breaths/min, with 18 cmH2O positive end-expiratory pressure. Once the mouse was stable on the ventilator, it was briefly disconnected to permit instillation of 300 μl of 5% BSA/saline (322 mOsm/l, isosmotic to mouse plasma) into the dependent (left) lung via the tracheal cannula, which was flushed with 100 μl air before reconnection to the ventilator. After ventilation for 30 min, instilled fluid was aspirated. Protein content was measured by the bicinchoninic (BCA) protein assay (Pierce Biotechnology, Rockford, IL). AFC rates were then calculated, as in previous studies.
In all studies, four ventilators were run concurrently (with the procedure staggered by a 3-min interval). AFC was measured with correction for excess murine alveolar protein, as previously described (57). No sex difference in AFC rate has been found in BALB/c mice. All reagents were added to the AFC instillate from stock solutions directly before instillation into mice, in a minimal volume of solvent (1–5 μl/ml). Final working concentrations of each reagent (as reported in the text) were based on previously published studies by other groups, when available; this was the case for amiloride (14), terbutaline (14), and CFTRinh-172 (57). All TLR ligands other than poly(I:C) were administered at the dose recommended by the manufacturer (Invivogen).
Intratracheal administration of poly(I:C) for bronchoalveolar lavage at 30 min.
Lightly anesthetized mice were intubated per os, and 100 μl of saline containing 500 ng poly(I:C) was administered intratracheally. Control animals received 100 μl of saline only. Mice were placed in lateral recumbency and allowed to recover for 30 min, at which time they were euthanized and subjected to bronchoalveolar lavage (see below).
Intranasal administration of poly(I:C) to determine effects at 1–3 days.
Mice were lightly anesthetized as previously described (12). One hundred microliters of saline, containing 250 or 500 μg poly(I:C) (equivalent to 10 or 20 mg/kg body wt, respectively), were administered dropwise, via the nares. Control animals received 100 μl of saline. Mice were allowed to recover and returned to their cage.
Mice were euthanized by intraperitoneal injection of ketamine (87 mg/kg)/xylazine (13 mg/kg), the tracheas were exposed surgically, and a trimmed sterile 18-gauge intravenous catheter was inserted caudally into the lumen. The lungs were then lavaged in situ with 1 ml of sterile saline. Cell viability was determined by Trypan blue exclusion, and cell types were differentiated on cytospin preparations using Wright-Giemsa stain. Cell differentials were determined from at least 200 leukocytes using standard hematological criteria.
Measurement of bronchoalveolar lavage mediators.
Bronchoalveolar lavage KC levels were measured by ELISA (R & D Systems). Lactate dehydrogenase was measured by colorimetric assay (Cayman Chemical, Ann Arbor, MI). Protein was measured by the BCA protein assay, as above. All assays were performed in accordance with manufacturers' instructions.
Lung wet:dry weight ratio.
Lung wet:dry weight ratio was measured as previously described (12). Briefly, mice were euthanized and exsanguinated, and their lungs were removed, weighed, and dried in an oven at 55°C for 7 days. After drying, the lungs were weighed again. Wet:dry weight ratio was then calculated as an index of intrapulmonary fluid accumulation, without correction for blood content.
Measurement of lung mechanics.
Mechanical properties of the mouse lung were assessed using the forced-oscillation technique (10, 20). Only female mice were used in these studies because male mice exhibit exaggerated airway responses to methacholine (4). Animals were anesthetized and tracheotomized as for AFC studies and then were mechanically ventilated on a flexiVent computer-controlled piston ventilator (SCIREQ, Montreal, Canada), with a tidal volume of 8 ml/kg, at a frequency of 150 breaths/min, against 2–3 cmH2O positive end-expiratory pressure, as in previous studies (55). Following two total lung-capacity maneuvers to standardize volume history, pressure and flow data were collected during a series of standardized volume-perturbation maneuvers. These data were used to calculate total lung resistance using the single-compartment model.
Measurement of methacholine responsiveness.
For assessment of airway responses to methacholine, and following initial baseline recordings of lung-function parameters with nebulization of saline only, mice on the flexiVent were exposed to increasing doses of methacholine (0.1, 1.0, 10.0, 20.0, and 50.0 mg/ml, in sterile normal saline, prepared fresh daily), as previously described (55). Each methacholine dose was delivered over a 10-s period via an AeroNeb vibrating plate ultrasonic nebulizer, in series with the inspiratory limb of the flexiVent Y-tube. Ten recordings of total lung resistance were generated following administration of each methacholine dose, and average values of all 10 measurements for each parameter at each methacholine dose were thereby obtained. Values for percent change from baseline were then calculated for each individual mouse at each methacholine dose.
Measurement of airway responsiveness to β-adrenergic agonists.
For assessment of β-agonist responsiveness, a second, age-matched, group of female mice was evaluated in parallel for each treatment condition. As in previous studies (55), initial baseline measurements of lung function with nebulization of saline were followed by exposure to the nebulized β-agonist terbutaline (100 μM), then increasing doses of methacholine (1, 10, 20, and 50 mg/ml), with a further dose of β-agonist between each dose of methacholine. Average values of 10 total lung-resistance measurements at each methacholine or terbutaline dose were thereby obtained, and percent changes from baseline were calculated, as above. To avoid loss of β-agonist bronchodilator effects, total lung-capacity maneuvers were not performed between doses. However, the nebulizer was carefully cleaned and dried with a swab between each methacholine or terbutaline dose to eliminate carry over. Both the nebulizer and tubing were also thoroughly cleaned between mice to eliminate drug transfer from one mouse to the next.
Descriptive statistics were calculated using Instat 3.05 (GraphPad Software, San Diego, CA). Gaussian data distribution was verified by the method of Kolmogorov and Smirnov. Differences between group means were analyzed by ANOVA, with Tukey-Kramer multiple-comparison posttests. P < 0.05 was considered statistically significant. All data are presented as means ± SE.
The double-stranded RNA analog poly(I:C) rapidly inhibits alveolar fluid clearance in normal BALB/c mice.
AFC was evaluated in live mice with normal oxygenation and acid-base balance, as previously described (12). Initial dose-response studies indicated that addition of 50 ng/ml of the double-stranded RNA analog poly(I:C) to the AFC instillate fluid had no effect on AFC rate in BALB/c mice (data not shown). In contrast, addition of 500 ng/ml poly(I:C) to the AFC instillate resulted in significant AFC inhibition at 30 min (Fig. 1A). The degree of AFC inhibition elicited by 500 ng/ml poly(I:C) (34%) was identical to the maximal AFC inhibition induced by RSV infection at day 2 (35%), but significantly less than that induced by influenza A virus infection at day 2 (51%), in both present and prior studies (12, 57). Importantly, when 500 ng/mouse of poly(I:C) (in 100 μl saline) was instilled intratracheally, no increase in lung water content (wet:dry weight ratio) was found 30 min later (Table 1). Bronchoalveolar lavage fluid levels of protein, lactate dehydrogenase, and KC (the murine homolog of IL-8) were reduced after 30 min of exposure to poly(I:C), compared with saline-treated controls. Although this decrease was only statistically significantly for lactate dehydrogenase, it probably occurred in all three cases as a result of greater dilution of lavage fluid following poly(I:C) exposure; because poly(I:C) acutely inhibits AFC, the original instillate will be cleared less rapidly from the lungs, so that a greater volume will still be present at 30 min when lavage is performed. A small, but nonsignificant, increase in bronchoalveolar lavage total cell counts was also noted in poly(I:C)-treated mice, primarily as a result of a significant rise in alveolar macrophage numbers. Very few neutrophils were present in lavage fluid in either treatment group. Finally, no significant histopathological changes in lung tissue were observed at 30 min after intratracheal administration of either saline or poly(I:C) (not shown). Taken together, these data indicate that 30 min of exposure to poly(I:C) does not induce respiratory epithelial cell death, impaired epithelial barrier function, or significant pulmonary inflammation.
We and others have shown previously that addition of either the epithelial Na+ channel inhibitor amiloride (1.5 mM) or the CFTR Cl− channel inhibitor CFTRinh-172 (100 μM) to the AFC instillate results in significantly reduced AFC in normal BALB/c and C57BL/6 mice (12, 14). In the present study, we found that amiloride reduced the mean AFC rate to 12% in untreated mice and 16% in poly(I:C)-treated animals (Fig. 1B). These AFC rates did not differ significantly from each other. However, in poly(I:C)-treated mice, the amiloride-sensitive component of AFC was reduced to 30%, compared with 69% in untreated mice. This indicates that poly(I:C) has no significant effect on amiloride-insensitive AFC and primarily inhibits its amiloride-sensitive fraction.
In contrast to amiloride, which inhibited AFC in both untreated and poly(I:C)-treated mice, CFTRinh-172 only inhibited AFC in untreated animals and had no effect on AFC in poly(I:C)-treated animals (Fig. 1B). Likewise, the AFC rate in poly(I:C)-treated mice was comparable to that in untreated mice under Cl−-free conditions (when NaCl in the instillate is replaced with Na+ gluconate), which maximize the electrochemical gradient for Cl− secretion into the lung (19) (Fig. 1C). These data indicate that, unlike influenza A virus (57), poly(I:C) does not stimulate CFTR-mediated Cl− secretion.
β-Adrenergic agonists can improve AFC in animal models of lung injury (40) and have shown promise as treatments for impaired AFC in humans (45, 48). However, the β-adrenergic agonist terbutaline (100 μM) had no stimulatory effect on AFC in poly(I:C)-treated mice (Fig. 1D). This phenotype is also found in RSV-infected mice (13) but is absent following influenza infection (57). Bronchoalveolar epithelial β-agonist insensitivity in RSV-infected mice results from CXCR2 chemokine receptor activation by KC (13, 55). As in RSV infection, a neutralizing rat monoclonal antibody to KC (50 μg/ml) had no effect on poly(I:C)-induced AFC inhibition at 30 min but restored terbutaline sensitivity. Isotype-matched rat-irrelevant IgG at the same concentration had no effect. Importantly, the ability of terbutaline to increase AFC in the presence of KC blockade confirms that the β-agonist itself is functional but that the response to it is impaired in poly(I:C)-exposed mice.
TLR-3 is required for inhibition of alveolar fluid clearance by the double-stranded RNA analog poly(I:C) at 30 min.
To confirm that the inhibitory effect of poly(I:C) on AFC was specific, we investigated the effects of exposure to other TLR ligands for 30 min on AFC in normal mice (Table 2). TLR-7 and TLR-8 are also associated with antiviral responses, but ligands for these TLRs had no significant inhibitory effect on AFC at 30 min. Likewise, neither the TLR-4 ligand endotoxin, which has been reported as being activated by RSV (18, 27) and shown to inhibit AFC chronically (41), nor ligands specific for TLR-5 and TLR-9 had any effect on AFC. Interestingly, however, ligands for both TLR-2 (heat-killed Listeria monocytogenes) and TLR-2/6 (FSL-1 synthetic lipoprotein) did inhibit AFC at 30 min.
To support these observations, we compared the effects of poly(I:C) exposure for 30 min, RSV infection for 2 days, and influenza infection for 2 days on AFC in C57BL/6-congenic TLR-3-knockout mice. Although TLR-3-knockouts had a lower basal AFC rate than BALB/c mice, this was completely unaffected by either addition of poly(I:C) to the AFC instillate or infection with RSV for 2 days (Fig. 2A). Interestingly, however, as in BALB/c and C57BL/6 mice (57, 61), AFC was impaired by 49% in TLR-3-knockouts at day 2 following influenza infection. This indicates that the inhibitory effect of influenza infection is both TLR-3 independent and unrelated to basal AFC rate.
In addition to its TLR-3-mediated effects, double-stranded RNA can activate cytoplasmic protein kinase R in a TLR- and retinoic acid-inducible gene I (RIG-I)-like helicase-independent fashion (36). However, we found that poly(I:C)-induced AFC inhibition in BALB/c mice was not reversed by the protein kinase R inhibitors 2-aminopurine (1 mg/ml) and PKRi (20 μM) (Fig. 2B).
Blockade of the pyrimidine-P2Y purinergic receptor axis prevents poly(I:C)-induced inhibition of alveolar fluid clearance.
Our previous studies demonstrated that inhibitory effects of both RSV and influenza virus on amiloride-sensitive AFC are mediated by purinergic 5′-nucleotides, acting on P2Y purinergic receptors (12, 57). We therefore wished to determine whether a similar mechanism might underlie the effect of poly(I:C) on amiloride-sensitive AFC. We have previously shown that agents used in this part of the study have no effect on AFC rate in normal mice (11, 12).
The de novo pyrimidine synthesis inhibitor A77–1726 (20 μM) reversed poly(I:C)-induced suppression of AFC at 30 min (Fig. 3A). This effect was blocked by concomitant addition of 20 μM uridine, which allows pyrimidine synthesis via a salvage pathway (11). This indicates that the A77–1726 block is specific to the de novo pyrimidine synthesis pathway.
Volume-regulated anion channels have been shown to mediate ATP release from respiratory epithelial cells in vitro (43) and to be necessary for AFC inhibition by both RSV and influenza (11, 57). Opening of these channels can be facilitated by ROCK activation (42). Poly(I:C)-mediated inhibition of AFC was blocked by both the volume-regulated anion channel inhibitor fluoxetine [10 μM (33)] and ROCKi (50 μM) but was not reversed by the Ca2+ chelator BAPTA-AM (100 μM), which prevents vesicle exocytosis (Fig. 3B). This indicates that poly(I:C)-induced AFC inhibition is a result of nucleotide release through volume-regulated anion channels.
AFC in poly(I:C)-treated mice was significantly increased by apyrase (5 U/ml), which degrades ATP and UTP to their monophosphate forms, but was unaffected by hexokinase (5 U/ml, with 10 mM glucose), which hydrolyzes ATP and (more slowly) UTP to their diphosphate forms (29) (Fig. 3C). UDP-glucose-pyrophosphorylase (5 U/ml), which metabolizes UTP to UDP-glucose in the presence of inorganic pyrophosphatase (5 U/ml) and glucose-1-phosphate (10 mM), also significantly increased AFC at 30 min in poly(I:C)-treated mice (Fig. 3C). This effect was lost in the absence of inorganic pyrophosphatase, which provides the energy to drive the otherwise fully reversible degradation of UTP (data not shown). Finally, the P2Y purinergic receptor antagonists suramin and XAMR-0721 (both 100 μM) also reversed the inhibitory effect of poly(I:C) on AFC at 30 min (Fig. 3D). Together, these data indicate that, as in RSV- and influenza-infected mice (12, 57), activation of P2Y receptors by the 5′-nucleotide UTP (3) mediates impairment of epithelial Na+ transport in poly(I:C)-treated animals.
The 5′ ectonucleotidase inhibitor α,β-methyleneADP (100 μM), which blocks degradation of ATP to adenosine (28), had no effect on AFC rate at 30 min in poly(I:C)-treated mice. Likewise the A1-adenosine receptor antagonist 8-cyclopentyl-1,3-dipropylxanthine (100 μM) did not reverse poly(I:C)-induced AFC impairment at 30 min (Fig. 3E). These data demonstrate that, unlike in influenza infection (57), adenosine/A1-adenosine receptor-induced stimulation of CFTR-mediated Cl− secretion plays no role in mediating poly(I:C)-induced AFC impairment. Our findings are also consistent with our observation that poly(I:C) does not stimulate CFTR-mediated Cl− secretion (see Fig. 1, B and C).
Intranasal exposure to the double-stranded RNA analog poly(I:C) induces alveolar fluid clearance inhibition at 1–3 days in normal BALB/c mice.
To more closely mimic the effects of respiratory viral infection on AFC, we established an intranasal double-stranded RNA exposure model. We found that intranasal exposure of mice to 250 μg/mouse poly(I:C) resulted in significant AFC impairment 2–3 days later (Fig. 4A). Exposure to a higher dose of poly(I:C) (500 μg/mouse) resulted in more rapid onset of AFC impairment with some recovery by day 3, whereas intranasal exposure of control animals to saline only had no detrimental effect on AFC at day 2 (data not shown). Because the lower dose of poly(I:C) (250 μg/mouse) induced AFC impairment with kinetics similar to those seen in RSV infection, we chose to use this dose for all subsequent studies in this model.
Poly(I:C)-induced AFC impairment at day 2 could be reversed by addition of 50 μM A77-1726 to the AFC instillate but was not affected by A77-1726 in the presence of 20 μM uridine (Fig. 4B). This suggests that a similar nucleotide-mediated mechanism (unrelated to cell death) underlies AFC inhibition following acute or chronic exposure to poly(I:C). Moreover, addition of 100 μM terbutaline had no effect on AFC rate in poly(I:C)-treated mice at day 2 (Fig. 4C). This indicates that, as with acute exposure via the AFC instillate, intranasal poly(I:C) induces prolonged insensitivity to β-agonists. However, unlike in 30-min poly(I:C) exposure studies, terbutaline responsiveness was not restored by addition to the AFC instillate of a neutralizing rat monoclonal antibody to KC (50 μg/ml).
Intranasal exposure to the double-stranded RNA analog poly(I:C) induces mild lung injury at day 2 in normal BALB/c mice.
No mortality or peripheral hypoxemia was found 1–8 days after intranasal exposure to either saline or 250 μg/mouse poly(I:C) (data not shown). However, treatment with poly(I:C), but not saline, resulted in a small (3%) but statistically significant (P < 0.001) loss of starting body weight at day 1 only. Exposure to poly(I:C), but not saline, also resulted in a significant increase in lung water content, which was maximal at day 2 (Table 3) and comparable in extent to that found in RSV- or influenza-infected mice at the same time point (12, 57). Lung wet:dry weight ratios in saline-treated animals were lower at day 2 than at 30 min (see Table 1), but this probably reflects the persistence of instilled fluid in the bronchoalveolar space at 30 min. Intranasal poly(I:C) treatment also moderately increased epithelial permeability and cell death at day 2, as indicated by elevated bronchoalveolar lavage fluid protein and lactate dehydrogenase content, respectively.
Poly(I:C) treatment resulted in limited lung histopathology at day 2, consisting primarily of a relatively mild parenchymal and perivascular mixed inflammatory cell infiltrate, with sporadic hypertrophy of alveolar type II cells. Mild perivascular, but not alveolar, edema was also observed. However, we found no evidence of respiratory epithelial cell death or sloughing following poly(I:C) exposure. Poly(I:C)treatment resulted in a very significant increase in bronchoalveolar lavage fluid total cell counts at day 2, primarily reflecting increased alveolar macrophage and neutrophil numbers (Table 3). Neutrophil infiltration was temporally associated with increased lavage fluid KC levels.
Intranasal exposure to the double-stranded RNA analog poly(I:C) induces airway β-agonist insensitivity at day 2 in normal BALB/c mice.
We have previously demonstrated that RSV infection is associated with impairment of both bronchoalveolar epithelial (AFC) and airway responsiveness to β-adrenergic agonists (13, 55). Likewise, in the present study, we found that poly(I:C) induced terbutaline-insensitive AFC impairment at both 30 min and 2 days (Figs. 1D and 4C). We therefore investigated the chronic effects of poly(I:C) treatment on bronchodilation to terbutaline at day 2.
To determine whether airways remained capable of bronchodilation in response to inhaled β-adrenergic agonists 2 days after intranasal poly(I:C) treatment, we compared values for the maximal change in total lung resistance induced by nebulization of 50 mg/ml methacholine between groups of mice receiving only nebulized methacholine and mice receiving both methacholine and terbutaline by nebulizer, as in previous studies with RSV (55). This approach was necessary because large airways in the mouse are almost completely relaxed at baseline, so β-adrenergic agonists will cause virtually no additional decrease in lung resistance (20). Therefore, to demonstrate a β-adrenergic agonist bronchodilatory effect of sufficient magnitude at which any poly(I:C)-induced changes could be detected, we had to first induce bronchoconstriction with methacholine. Additionally, to measure poly(I:C)-induced changes in responsiveness to β-adrenergic agonists, it was important to ensure that poly(I:C) did not significantly alter bronchoconstrictive responses to methacholine.
Using this approach, we found that baseline total lung resistance values 2 days after intranasal treatment with saline (100 μl) or poly(I:C) (250 μg/mouse in 100 μl saline) did not differ from those of normal BALB/c mice (Fig. 5A) and were similar to those reported previously (1). Likewise, exposure to nebulized terbutaline had no effect on baseline total lung resistance values in either saline- or poly(I:C)-treated mice (Fig. 5A). This demonstrates that the large airways of both treatment groups were fully relaxed at baseline. Poly(I:C) treatment also had no effect on methacholine responsiveness at day 2 (Fig. 5B), which confirms that poly(I:C) does not cause airway hyperresponsiveness at this time point. Finally, we found that poly(I:C)-exposed animals were unresponsive to terbutaline at day 2; nebulized terbutaline resulted in a significant (46%, P < 0.05) decrease in the maximal change in total lung resistance in saline-exposed mice but had no effect on maximal total lung-resistance values in poly(I:C)-treated animals (Fig. 5C). These data indicate that intranasal poly(I:C) treatment did not result in airway hyperresponsiveness to the bronchoconstrictor methacholine at day 2 but did induce airway insensitivity to the bronchodilatory effects of the β-adrenergic agonist terbutaline.
Respiratory epithelial cells, alveolar macrophages, and intraepithelial plasmacytoid dendritic cells play a primary role in the initiation and orchestration of innate and adaptive immune responses to invading pathogens in the lung. All three cell types can express pattern-recognition receptor systems capable of recognizing a variety of pathogen-associated molecular patterns (15). Recognition results in transcription of multiple genes involved in innate and adaptive immunity, including type I interferons (α and β) and proinflammatory cytokines such as IL-8 (23), which have downstream consequences for both lung function and host survival. Three pattern-recognition receptor families have been identified to date: TLRs, the RIG-I-like helicases, and inflammasome proteins (15). RSV and influenza A virus have been shown to interact in a partially-overlapping fashion with the different pattern-recognition receptors (reviewed in Ref. 22); influenza can stimulate TLR-3, TLR-7, RIG-I, and the inflammasome, whereas RSV can activate TLR-2, TLR-3, TLR-4, and RIG-I. RSV does not appear to activate either TLR-7 (because endosomal fusion is not a feature of RSV cellular entry) or the inflammasome although a related paramyxovirus (Sendai virus), which has similar nucleotide-mediated effects on AFC to those of RSV (26), does stimulate this latter pathway. Finally, double-stranded RNA, which is a cytoplasmic byproduct of viral genome replication and transcription for both pathogens (34), can activate TLR-3 and the inflammasome and can also be recognized by RIG-I.
Previously, we demonstrated that replication-competent RSV (12) and influenza A virus (57) can both induce nucleotide/P2Y receptor-mediated impairment of amiloride-sensitive AFC in BALB/c mice. The artificial TLR-3 ligand and double-stranded RNA analog poly(I:C), which has been used for several years to mimic the inflammatory and pathophysiological consequences of RNA viral infections of the lung, had also been shown to stimulate ATP release from respiratory epithelial cells in vitro (31). We therefore hypothesized that viral double-stranded RNA replication intermediates recognized by TLR-3 might act as a common initiator of nucleotide release for both viruses. Furthermore, we proposed that by characterizing the degree of concordance between the effects of double-stranded RNA, RSV, and influenza virus on murine lung function, we could determine the relative contribution of TLR-3 to the pathogenesis of lung dysfunction in both viral diseases, which differ in both their degree of AFC impairment and overall severity/lethality in mice.
When we compared poly(I:C) treatment to respiratory viral infection, we found that exposure to poly(I:C) elicited prolonged lung dysfunction and other outcomes almost identical to those caused by the A2 strain of RSV, but quite different in most cases to those induced by influenza A (Table 4). Specifically, poly(I:C) treatment induced mild hypoxemia and weight loss, a moderate increase in lung water content, and β-adrenergic agonist insensitivity in both the bronchoalveolar epithelial and airway compartments, all of which are evident in RSV-infected but not influenza-infected mice. Exposure of BALB/c, but not TLR-3−/−, mice to poly(I:C) also induced nucleotide-mediated AFC impairment, which we have also observed previously in both RSV and influenza infections (12, 57). However, poly(I:C) did not trigger the severe hypoxemia, cachexia, pulmonary edema, and mortality seen in influenza-infected mice and did not induce either airway hyperresponsiveness [reported by others in influenza-infected mice (2, 60)] or an adenosine-mediated Cl− secretory response, which contributes to more severe AFC impairment in influenza-infected animals (57). Moreover, influenza inhibited AFC even in TLR-3-knockout mice. These data suggest that the pathophysiological effects of RSV infection in BALB/c mice can be fully recapitulated by double-stranded RNA interacting with endosomal TLR-3, but those of influenza cannot. Furthermore, our findings suggest that the severe pathophysiological effects of influenza A virus are not TLR-3 mediated and may involve interaction of other influenza viral genetic determinants with additional pattern-recognition receptors, such as TLR-7. Finally, we should note that, since we found that TLR-2 agonists could also induce the same degree of AFC impairment as poly(I:C), we cannot exclude a role for TLR-2 activation in mediating this effect, particularly given the recent finding that activation of TLR-2 by RSV is involved in the innate immune response to this virus (39). However, because influenza also inhibits amiloride-sensitive AFC but has not been shown to activate TLR-2-mediated signaling, this possibility seems unlikely.
In addition to signaling via endosomal TLR-3, double-stranded RNA can induce cellular changes by binding to cytoplasmic RNA-dependent protein kinase (protein kinase R). Protein kinase R is induced by type I interferons and recognizes RNA with short-stem loops in a TLR- and RIG-I-like helicase-independent fashion (36). Activation of protein kinase R can mediate enhanced responses to double-stranded RNA in RSV-infected mice (17). However, we found that the inhibitory effect of poly(I:C) on AFC at 30 min in BALB/c mice was not reversed by specific inhibitors of this kinase. This indicates that poly(I:C)-induced AFC inhibition is not dependent on protein kinase R signaling. Nevertheless, our findings do not preclude a role for protein kinase R in mediating other effects of poly(I:C) on lung function and inflammation in a TLR-3-independent manner.
Previous in vitro studies have shown that treatment of murine or human primary respiratory epithelial cells or cell lines with poly(I:C) induces secretion of multiple chemokines, particularly macrophage inflammatory protein-1α, regulated on activation normal T-expressed and presumably secreted (RANTES), and IL-8 (9, 16, 24, 54), together with increased expression of genes encoding TLRs, including trl3 (47, 51). In vivo studies using mice have demonstrated that poly(I:C) treatment results in TLR-3- and CXCR2-dependent neutrophilic pulmonary inflammation, interstitial edema, bronchiolar epithelial hypertrophy, and altered lung function (32, 53). These changes were accompanied by elevated bronchoalveolar lavage fluid proinflammatory cytokine levels (53) and increased airway epithelial cell TLR-3 protein expression (32). Our studies demonstrating that exposure to 250 μg/mouse poly(I:C) results in elevated bronchoalveolar lavage KC levels and neutrophil counts at day 2, together with mild pneumonitis, moderate pulmonary edema, and limited epithelial impairment, are consistent with these earlier findings and utilize comparable poly(I:C) doses. However, we are the first to show that poly(I:C) induces nucleotide-dependent AFC inhibition and β-agonist insensitivity in both bronchoalveolar epithelium and airways. These functional changes are similar to those caused by RSV.
Several pathways have been proposed for nucleotide release from cells (50). These include facilitated diffusion, exocytosis of nucleotide-filled cytoplasmic granules, and channel-mediated release. ATP release from respiratory epithelial cells in vitro is mediated by volume-regulated anion channels (43), the opening of which is facilitated by ROCK activation (42). In previous studies, we found that AFC inhibition by both RSV and influenza is dependent on ROCK and volume-regulated anion channel activity (11, 57). Likewise, in the present study, we found that inhibitors of both Rho kinases and volume-regulated anion channels could block poly(I:C)-mediated AFC inhibition at 30 min (33), but the Ca2+ chelator BAPTA-AM had no such effect. This indicates that poly(I:C)-induced AFC inhibition is a result of nucleotide release through volume-regulated anion channels, rather than exocytosis of ATP-containing cytoplasmic granules, which is a Ca2+-dependent process (44). Furthermore, the ability of volume-regulated anion channel inhibition to rapidly reverse AFC inhibition in poly(I:C)-treated mice indicates that, as in RSV- or influenza-infected animals, nucleotide release is not a nonspecific consequence of epithelial cell death. We should note, however, that out findings provide no information regarding nucleotide release pathways in the normal lung. Moreover, we cannot exclude the possibility that, rather than being the release pathway, volume-regulated anion channels merely facilitate nucleotide release via another mechanism or modulate extracellular nucleotide degradation.
Respiratory epithelial cells are the primary replication site for both RSV and influenza virus and are therefore the most likely candidates for TLR-3-mediated nucleotide release following either viral infection or exposure to double-stranded RNA (35, 58). However, both viruses can also infect alveolar macrophages, albeit at lower levels (35, 58). Because alveolar macrophages also express TLR-3 (46), it is therefore possible that the effects of double-stranded RNA on alveolar fluid clearance are mediated by nucleotides released by poly(I:C)-stimulated macrophages, rather than epithelial cells. However, there are several points that render this mechanism less likely. First, in vitro studies have shown that both influenza virus (6, 30) and RSV (5, 52) can inhibit amiloride-sensitive Na+ transport by respiratory epithelial cells in the absence of alveolar macrophages and without altering ouabain-sensitive Na+, K+ ATPase activity (5). Second, as noted above, this cell type is not the primary site of replication for either virus, which suggests that alveolar macrophages are unlikely to be the primary source of the nucleotides that are responsible for inhibition of amiloride-sensitive AFC following infection with influenza or RSV, or exposure to double-stranded RNA. Finally, Punturieri et al. (46) demonstrated that resident alveolar macrophages in normal mice are unresponsive to stimulation by both poly(I:C) and bacterial ligands, as a result of impaired STAT-1 signaling. Taken together, these studies argue against a significant role for alveolar macrophages in poly(I:C)-induced inhibition of amiloride-sensitive AFC although they clearly do not exclude a role for this cell type in other aspects of lung dysfunction that result from exposure to double-stranded RNA.
As noted above, we have shown previously that infection of mice with both RSV and influenza virus induces increased release of ATP into the bronchoalveolar lavage fluid (10–12, 57). However, only in influenza infection is elevated ATP release accompanied by increased activation of A1-adenosine receptors (12, 57). By stimulating CFTR-mediated Cl− secretion, A1-adenosine receptor activation contributes to AFC impairment in influenza-infected mice (57). However, in other nonviral models, adenosine stimulation of A1-adenosine receptors on neutrophils has also been shown to play a significant role in the initiation and promotion of acute lung injury (49). This suggests that increased adenosine generation may be a determinant of the increased severity of lung injury associated with influenza infection in mice although this does not preclude a role for other factors, such as differential type I interferon production (21) and induction of the cytokine storm in severe influenza (7). The mechanism by which influenza induces adenosine production remains unclear although our present data suggest that this effect is not mediated by double-stranded RNA replication intermediates because it cannot be replicated by poly(I:C).
We have shown previously that primary RSV infection results in both bronchoalveolar epithelial and airway insensitivity to β-agonists (13, 55). Interestingly, we have also found that influenza A virus does not induce epithelial β-agonist insensitivity at 2 days postinfection (57) although airway responses to β-agonists in influenza-infected mice have not been determined to date. Bronchoalveolar epithelial β-agonist insensitivity in RSV-infected mice does not result from receptor internalization or degradation (13). Moreover, this effect is not characterized by any dramatic shift in β2-adrenergic receptor binding kinetics (13). Rather, bronchoalveolar epithelial and airway β-agonist insensitivity in RSV infection appears to result from heterologous CXCR2-mediated desensitization of β2-adrenergic receptors. Desensitization is induced the neutrophil chemoattractant KC, which is released in response to RSV (13, 55). In the present study, we found that intranasal treatment with poly(I:C) could likewise induce KC production and β-agonist insensitivity in both the bronchoalveolar epithelium and airways at day 2. These findings suggest that β-agonist insensitivity in RSV may be triggered by double-stranded RNA intermediates. Interestingly, however, only the bronchoalveolar epithelial β-agonist sensitivity caused by exposure to poly(I:C) for 30 min could be prevented by KC antagonism. This suggests that double-stranded RNA induces epithelial β-agonist insensitivity at day 2 by a different mechanism to that triggered by either exposure to double-stranded RNA for 30 min or infection with RSV. The mechanism underlying β-agonist insensitivity at day 2 after exposure to poly(I:C) presently remains undefined. However, because it is unlikely that poly(I:C) persists in the lungs for the entire 2 days after intranasal administration, it may be that effects of prolonged exposure on epithelial β-agonist insensitivity occur secondary to some other aspect of the inflammatory response to poly(I:C). More specifically, we postulate that β-agonist insensitivity may be related to release of proteases from infiltrating inflammatory cells (particularly neutrophils) that are present (and probably activated) in very large numbers at day 2, but which are absent at 30 min (25).
In the present study, we found that intranasal poly(I:C) treatment did not induce airway hyperresponsiveness at day 2. This result is in agreement with the findings of Cooper et al. (9), who reported no change in responsiveness to the muscarinic agonist carbachol in precision-cut human lung slices following exposure to poly(I:C) for 24 h. However, our data contrast with those of Stowell et al. (53), who reported airway hyperresponsiveness following poly(I:C) treatment. The underlying reason for this discordance is unclear although it may relate to the fact that, although the latter group administered a comparable total amount of poly(I:C) to mice in their study (300 μg/mouse, compared with 250 μg/mouse in our study), they did so over a 3-day period, rather than in a single dose. Alternatively, the discrepancy may result from the fact that these investigators assessed airway hyperresponsiveness to intravenous methacholine 24 h after 3 days of poly(I:C) treatment, rather than airway hyperresponsiveness to aerosolized methacholine at day 2, as in the present study. Indeed, previous studies have shown that airway responses to methacholine differ between these two routes of administration (2). Nevertheless, given that previous studies have shown that influenza A virus (2, 60), but not the A2 strain of RSV (38, 55), causes airway hyperresponsiveness to methacholine in BALB/c mice, our data also suggest that poly(I:C) induces an airway dysfunction phenotype analogous to that of RSV-infected mice but dissimilar to that of influenza-infected animals.
In conclusion, our data show that exposure of the lung to poly(I:C) recapitulates the lung dysfunction phenotype of RSV-infected mice but not the severe lung injury characteristic of influenza. Viral double-stranded RNA replication intermediates can induce nucleotide-mediated impairment of amiloride-sensitive AFC in both infections, together with β-adrenergic agonist insensitivity in RSV. However, these replication intermediates are insufficient to induce either adenosine-mediated Cl− secretion in the lung or severe hypoxemia, both of which are features of influenza. Identification of the viral determinants and pattern-recognition receptors involved in triggering these influenza-specific processes may allow development of new therapies to ameliorate influenza-associated lung injury.
This research was supported by a Beginning Grant-in-Aid (0765209B) from the American Heart Association Great Rivers Affiliate to I. Davis and by funds from the Department of Veterinary Biosciences, The Ohio State University. F. Aeffner was supported by the Eli Lilly Foundation. E. Yu was supported by funds from the Laboratory Animal Residency Program, The Ohio State University.
No conflicts of interest, financial or otherwise are declared by the authors.
The current address of Dr. Yu is Division of Animal Care, Vanderbilt University, 1161 21st Ave. South, AA-6206 Medical Center North, Nashville, TN 37232.
- Copyright © 2011 the American Physiological Society