Role of Xanthine Oxidase Activation and Reduced Glutathione Depletion in Rhinovirus Induction of Inflammation in Respiratory Epithelial Cells*

Rhinoviruses are the major cause of the common cold and acute exacerbations of asthma and chronic obstructive pulmonary disease. We previously reported rapid rhinovirus induction of intracellular superoxide anion, resulting in NF-κB activation and pro-inflammatory molecule production. The mechanisms of rhinovirus superoxide induction are poorly understood. Here we found that the proteolytic activation of the xanthine dehydrogenase/xanthine oxidase (XD/XO) system was required because pretreatment with serine protease inhibitors abolished rhinovirus-induced superoxide generation in primary bronchial and A549 respiratory epithelial cells. These findings were confirmed by Western blotting analysis and by silencing experiments. Rhinovirus infection induced intracellular depletion of reduced glutathione (GSH) that was abolished by pretreatment with either XO inhibitor oxypurinol or serine protease inhibitors. Increasing intracellular GSH with exogenous H2S or GSH prevented both rhinovirus-mediated intracellular GSH depletion and rhinovirus-induced superoxide production. We propose that rhinovirus infection proteolytically activates XO initiating a pro-inflammatory vicious circle driven by virus-induced depletion of intracellular reducing power. Inhibition of these pathways has therapeutic potential.

Rhinoviruses (RV) 3 are the major cause of the commonest human acute infectious disease, the common cold (1). They are also associated with the majority of acute exacerbations of asthma (2,3) and chronic obstructive pulmonary disease (COPD) (4,5). No licensed effective antiviral is currently available for the treatment of the common cold (6, 7) and treatment of virus-induced asthma and COPD exacerbations is a major unmeet therapeutic need (8). Understanding the mechanisms of virus-induced exacerbation of airway diseases is required to identify molecular targets for therapeutic intervention.
The mechanisms underlying virus-induced exacerbations of airway diseases are poorly understood. However, rhinoviruses are believed to directly infect airway epithelium inducing proinflammatory cytokine production (9 -11). This leads to recruitment and activation of inflammatory cells, resulting in airway inflammation (12,13). We have recently demonstrated that bronchial epithelial cells from asthmatic subjects have a deficient innate immune response to rhinovirus infection, responsible for: (i) increased virus replication (14,15) that could account for increased and more persistent inflammatory responses (12); (ii) increased severity and duration of lower respiratory tract symptoms and reductions in lung function (16) in rhinovirus-induced asthma exacerbations.
Increased oxidative stress is implicated in induction of the acute airway inflammation during exacerbations of asthma and COPD (17). Oxidants are directly involved in inflammatory responses via signaling mechanisms, including the redox-sensitive activation of transcription factors such as NF-B (18,19).
Recent data indicate that rhinovirus and other respiratory viruses can alter cellular redox homeostatic balance toward a pro-oxidative condition (20 -22). The molecular pathways responsible for such disequilibrium are virtually unknown. A recent study suggested NADPH oxidase involvement in rhinovirus-induced production of reactive oxygen species over a 6-h infection (23). In a previous study we documented that rhinovirus infection induces a rapid increase of intracellular superoxide anion (O 2 . ), which occurs within 15 min after infection.
This early pro-oxidative response was found to induce NF-B activation and downstream pro-inflammatory molecule production (24). O 2 . is a product of cellular metabolism and mainly originates from the activity of two enzyme systems: NADPH oxidase and xanthine dehydrogenase/xanthine oxidase (XD/XO) (25). Here we studied the molecular mechanisms by which rhinovirus induces rapid O 2 . production in respiratory epithelial cells. We also analyzed the mechanisms by which reducing agents can abolish rhinovirus-induced O 2 . production and thus can stabilize the intracellular redox state in respiratory epithelial cells following infection. Finally, we demonstrated that blocking the activity of the system responsible for rhinovirus-triggered O 2 .
generation inhibited rhinovirus-induced inflammatory mediator production in respiratory epithelial cells.

Cell Culture
Ohio HeLa cells were obtained from the MRC Common Cold Unit, Salisbury, UK, and A549 cells, a type II respiratory cell line, were obtained from the American Type Culture Collection (ATCC, Rockville, MD). Primary human bronchial epithelial cells (HBEC) were obtained by bronchial brushing from healthy volunteers, and cultured as previously described (14,24,26).

Virus Stocks
Rhinovirus type 16 (RV16, a major group rhinovirus) was obtained from the MRC Common Cold Unit. Viral stocks were prepared by infection of sensitive cell monolayers (Ohio HeLa, HeLa) as described elsewhere (24,26). TCID 50 /ml values were determined and the rhinovirus serotype was confirmed by neutralization with serotype-specific antibodies (ATCC) (27). For selected experiments rhinovirus type 1B (RV1B, minor group), obtained from the MRC Common Cold Unit, was used to evaluate whether the results were group/receptor restricted. For selected experiments filtration of the virus from inoculum, to remove viral particles, was performed as previously described (24,26). Filtered virus stocks were used as negative control. Virus at a multiplicity of infection of 1 was used for all the experiments.

Infections, Harvesting of Cells, Preparation of Cell Homogenates, and Preparation of Membrane and Cytosolic Fractions
Confluent A549 or HBEC cells were exposed to rhinovirus, medium alone, or filtered virus (f-RV) inoculum for different time intervals (20 min to 8 h). Cell layers were thereafter washed three times in cold phosphate-buffered saline (PBS) before harvesting by scraping. Harvested cells were centrifuged and the cell pellet was resuspended in phosphate buffer (10 mM, pH 7.2). Cell lysis was obtained by repeated (three times) freezing and thawing. For preparation of cytosolic fractions, the cell homogenate was then ultracentrifuged at 20,000 ϫ g for 30 min, the cell fragments pelleted, and the supernatant (cytosol) collected. Where indicated, to obtain the membrane fraction, the cell homogenates were centrifuged at 800 ϫ g for 10 min to separate nuclei from cell membranes. Supernatants were harvested and again centrifuged at 2,000 ϫ g, supernatants discarded, and membrane pellets diluted in 0.1 M sucrose solution. A final centrifugation at 11,000 ϫ g for 20 min was performed at 4°C. Pellets containing membranes were diluted in 100 l of PBS buffer. Protein content was determined photometrically using the Bio-Rad protein assay (Bio-Rad).

Cytochrome c Reduction Kinetics
The intracellular production of O 2 . was spectrophotometrically evaluated by superoxide dismutase (SOD)-inhibitable cytochrome c reduction kinetics, as previously described (24,35). Kinetics were carried out in 2-ml quartz cuvettes at 37°C for 20 min in a Uvikon 860 (Kontron Instruments) spectrophotometer in the presence or absence of SOD (500 IU/ml, Sigma). Concentration of cytochrome c from beef heart (Sigma) was 10 Ϫ5 M. Absorbance readings were taken at 550 nm (peak of reduced cytochrome). Newly generated O 2 . was measured in each sample and expressed as micromolar, according to standardized procedures (25). Measurements were based on absorbance differences in the presence or absence of SOD, after 5 min of kinetics, when the kinetic slope of cytochrome c reduction was steepest. Data were normalized per mg of protein.

Uric Acid Kinetics
Uric acid kinetics were performed at 1 h infection to evaluate the involvement of XD/XO in rhinovirus-induced O 2 .
generation, as uric acid represents the other end product of xanthine degradation by XO. Uric acid kinetics were spectrophotometrically monitored at 293 nm in a UviKon spectrophotometer (Kontron), according to standard procedures (36,37), on 500 l of supernatant collected in PBS, pH 7.4, after addition of xanthine (0.1 mM, Sigma). After 15 min of kinetics, uricase (1.0 units/ml) was added to evaluate the amount of uric acid produced.

NADPH Oxidase Assay
NADPH oxidase assay was performed at different time intervals (20 min to 3 h) to evaluate the involvement of this system in rhinovirus-induced O 2 . generation. Cell homogenates were centrifuged as described above to separate nuclei from cell membranes. To reconstitute NADPH oxidase, supernatants containing membranes were centrifuged again at 40,000 ϫ g. The reaction mixture contained 200 l of supernatant and 50 l of diluted membrane pellet. After 2 min, 200 M NADPH and 5 mM MgCl 2 were added in the presence or absence of the specific inhibitor of NADPH oxidase diphenylene iodonium chloride (0.92 g/ml, Sigma) (38). Cytochrome c was added to a concentration of 0.1 mM and PBS, pH 7.2, to a final volume of 0.5 ml, and the reduction kinetics were monitored for 15 min at 37°C as previously described.

Western Blot Analysis for Xanthine Dehydrogenase/Oxidase
Whole cell proteins were extracted from A549 cells as previously described (39). At least 50 mg/lane of whole cell proteins were subjected to a 4 -12% Tris glycine gel electrophoresis, and transferred to nitrocellulose filters by blotting. Filters were blocked for 45 min at room temperature in Tris-buffered saline (TBS), 0.05% Tween 20, 5% nonfat dry milk. The filters were then incubated with rabbit anti-human XD/XO (LS-C26419; from LifeSpan Biosciences) for 1 h at room temperature in TBS, 0.05% Tween 20, 5% nonfat dry milk at dilution of 1:500. Filters were washed three times in TBS, 0.5% Tween 20 and after being incubated for 45 min at room temperature with goat anti-rabbit antibody conjugated to horseradish peroxidase (Dako) in TBS, 0.05% Tween 20, 5% nonfat dry milk, at a dilution of 1:4000. After three further washes in TBS, 0.05% Tween 20 visualization of the immunocomplexes was performed using ECL as recommended by the manufacturer (Amersham Biosciences). As an internal control we reprobed each filter with an antihuman actin antibody (Santa Cruz Biotechnology). The 145and 85-kDa bands of the XD/XO system ((full-length XD and the post-cleavage fragment containing the active site, respectively (40,41)) and the 43-kDa (actin) band were quantified using densitometry with VisionWorks LS software (UVP) and expressed as the ratio with the corresponding actin optical density value of the same lane.

Knockdown of Xanthine Dehydrogenase Expression
RNA interference was used to specifically suppress expression of XD in A549 cells. Cells were transfected in 6-well plates with small interfering RNA (siRNA) using siPORT TM NeoFX TM Transfection Agent (Applied Biosystem), as described by the manufacturer. The following siRNA (all from Ambion) were used: siRNAs for XD (s14918; target sequence: sense, GCAUCGUCAUGAGUAUGUAtt; antisense, UUUAU-AGCAUCCUCAAUUGtg), siRNA for GAPDH (4390849) and nonsilencing siRNA (4390843). Total mRNA was extracted by using the RiboPure TM kit (Ambion) as per the manufacturer's instructions. 1 g of mRNA was used to perform the reverse transcription assay with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystem). XD mRNA expression was monitored by Real Time RT-PCR using the TaqMan Gene Expression Assay (Applied Biosystem) specific for XD (catalog number Hs00166010_m1) following the manufacturer's recommendations. The reaction was carried out in a Rotor-Gene TM 6000 instrument (Corbett Life Science). Results were normalized to 18S rRNA (sense, 5Ј-CGC CGC TAG AGG TGA AAT TCT-3Ј; antisense, 5Ј-CAT TCT TGG CAA ATG CTT TCG-3Ј 300 nM each, probe, 5Ј-FAM ACC GGC GCA AGA CGG ACC AGA TAMRA-3Ј, 175 nM) and expressed as XD mRNA relative levels as compared with nonsilencing siRNAtransfected cells by using the RotorGene software (Corbett Research) and the two standard curve methods for relative quantitation (42).

High Performance Liquid Chromatography (HPLC) Analysis of Intracellular GSH
A549 cells were cultured at 85% confluence, incubated with RV16, medium alone, or f-RV for different periods (20 min to 1 h), then trypsinized, collected, and harvested in cryovials with 1.2 ml of 3% metaphosphoric acid in sterile conditions to avoid GSH oxidation and finally frozen in liquid nitrogen until used. Cell homogenate was obtained and protein content was determined as previously described. Intracellular GSH concentration was evaluated by HPLC in a Kontron Instruments apparatus (Milan, Italy) equipped with a C18 hydrophobic column (5 m particle size, 4.6 ϫ 250 mm), a 420 pump (range 0.005-10 ml/min), a 425 gradient former, and an injection valve with a 20-l sampling loop. Elution was carried out at room temperature in isocratic gradient (75% methanol and KH 2 PO 4 buffer, pH 3, 1 ml/min speed). GSH was analyzed at 200 nm by a 432 UV detector (Kontron) with an IBM integrated software PC Pack. Homogenates of cells were centrifuged at 40,000 ϫ g for 20 min at 4°C and the supernatant collected and concentrated on Amicon Ultra 10,000 centrifugal filter devices (Millipore, Bedford, MA) to a final volume of about 300 l. Samples were analyzed without derivatization against standards of pure lyophilized GSH (Biomedica Foscama), diluted in 1 ml of normal saline solution. The final concentration was obtained by serial dilutions of the lyophilized product. In selected experiments cells were pretreated, before infection, as previously

Rhinovirus Replication
Titration Assay in a Sensitive Cell Line-Rhinovirus replication was evaluated by titration assay in a sensitive cell line (HeLa) (9). Cells were seeded in a 96-well plate. Where indicated, subconfluent cells were treated with the highest concentration used in the study for each of the tested compounds for the time intervals previously specified (4 h for Leu, Pep, PMSF, Apr, E-64, oxypurinol; 12 h for H 2 S and GSH) or diluent alone before the infection. Cells were exposed for 1 h to 10-fold serial dilution of RV16, from not diluted down to 10 Ϫ8 (4 wells per condition). After a 1-h infection, virus unbound to cultured cells was removed and fresh medium added. The cells were incubated in 4% minimal essential medium (Invitrogen) at 37°C for 5 days, fixed in methanol, and stained with 0.1% crystal violet. The cytopathic effect was evaluated by visual assessment and assessment of the continuity of the monolayer. For each experiment TCID 50 /ml values were calculated (27).
TaqMan Real-time PCR-A549 cells were seeded in 6-well plates at 1.7 ϫ 10 5 cells/ml. Where indicated, subconfluent cells were treated with the highest concentration used in the study for each of the tested compounds for the time intervals previously specified (4 h for Leu, Pep, PMSF, Apr, E-64, and oxypurinol; 12 h for H 2 S and GSH) or diluent alone before the infection. Cell lysates were harvested at 4 and 8 h following the infection. Total RNA was extracted from cell lysates by using a commercially available kit (RNeasy Kit, Qiagen) following the manufacturer's recommendations. Viral RNA in cell lysates was measured by TaqMan RT-PCR. For this purpose 2 g of total RNA were used for cDNA synthesis (Omniscript RT kit, Qiagen).
TaqMan quantitative PCR was carried out using primers and probe for rhinovirus (sense, 5Ј-GTG AAG AGC CSC RTG TGC T-3Ј 50 nM; antisense, 5Ј-GCT SCA GGG TTA AGG TTA GCC-3Ј 300 nM; probe, 5Ј-FAM-TGA GTC CTC CGG CCC CTG AAT G-TAMRA-3Ј, 175 nM) and 18S rRNA (see above) (15). Reactions consisted of 2 l of cDNA (cDNA for 18S was diluted 1:100), 12.5 l of 2ϫ QuantiTect Probe PCR Master Mix (Qiagen), primers, and probes at the final concentrations listed above and RNase-free water to a total volume of 25 l. Reactions were performed on a Rotor-Gene TM 6000 instrument (Corbett Life Science). Viral RNA expressions were normalized to 18S rRNA and compared with standard curves and expressed as copies per g of RNA.

Enzyme-linked Immunosorbent Assays for Chemokines
Subconfluent A549 cells were pretreated for 4 h with oxypurinol (20 M) before RV16 or f-RV inoculum or medium alone treatment. After a 1-h infection unbounded virus was removed and fresh medium added. Supernatants were harvested at 4 h and levels of IL-8 and GRO-␣ were assessed using commercially available enzymelinked immunosorbent assay kits (R&D System) following the manufacturer's instructions. Detection limits for IL-8 and GRO-␣ enzymelinked immunosorbent assay were ϳ10 and 15 pg/ml, respectively.

NF-B Transcription Factor Activation
Nuclear extracts were prepared from A549 cells using the Nuclear Extract Kit (Active Motif). NF-B activation was assessed in A549 cell nuclear extracts using the TransAM TM p65 Transcription Factor Assay Kit (Active Motif) following the manufacturer's recommendations. Nuclear extract of Jurkat cells provided by the manufacturer (Active Motif) were used as positive controls.

Statistical Analysis
Group data were expressed as mean Ϯ S.E. Analysis of variance was used to determine differences between groups. Paired or unpaired Student's t tests were performed after the analysis of variance when appropriate. All experiments were carried out at least 5 times. Bonferroni adjustment was applied . ) in A549 respiratory epithelial cells and HBEC. The production is independent of the NADPH oxidase system. In panels A, B, and E, confluent cells were exposed to live RV16 (closed circles), medium alone (open circles), or RV16 physically removed by filtration (f-RV, dash line, diamonds) for different time intervals (20 min to 8 h). O 2 . production was evaluated in cytosolic fractions of A549 cells (A, n ϭ 7) and HBEC (E, n ϭ 5) and in total A549 cell homogenates (B, n ϭ 7) by SOD-inhibitable cytochrome c reduction kinetics. C, O 2 . production by a mixture of cytosolic and membrane fraction from A549 cells was evaluated at different intervals following exposure to live RV16 (black circles), RV16 physically removed by filtration (open circles), medium alone (diamonds), without (continuous line) or with (dashed line) NADPH added to the reaction mixture. The addition of NADPH did not affect RV16-induced cytosol O 2 . production (n ϭ 5). D, O 2 . production was not changed when a specific inhibitor of NADPH oxidase (diphenylene iodonium chloride) was added to the reaction mixture (RV16 physically removed by filtration (open circles), medium alone (diamonds) or live RV16 infection (black circles), without (continuous line) or with (dashed line) NADPH) confirming lack of involvement of NADPH oxidase (n ϭ 5). F and G, O 2 . production evaluated in the cytosolic fraction of HBEC (F, n ϭ 5)) and A549 cells (G, n ϭ 7) exposed to live RV1B (closed circles), medium alone (open circles), or RV1B physically removed by filtration (dash line, diamonds). Newly generated O 2 . was measured in each sample and expressed as micromolar and normalized per mg of protein (***, p Ͻ 0.001; *, p Ͻ 0.05 compared with medium alone treated cells and f-RV inoculated cells).
where indicated. A probability value of Ͻ0.05 was considered significant.

Rhinovirus-induced O 2 . Production in Respiratory Epithelial
Cells Is Cytosolic Not Membrane Associated-O 2 . production was evaluated by SOD-inhibitable cytochrome c reduction kinetics. In our previous study we found that RV16 infection rapidly induced intracellular O 2 . production, which was maximal at 1 h in A549 cells, a type II respiratory epithelial cell line (24,35). That study evaluated intracellular O 2 . generation, i.e.
cell membranes were precipitated before SOD-inhibitable cytochrome c reduction assay was performed. Because other workers had implicated NADPH oxidase in RV induction of reactive oxygen species (23) and because NADPH oxidase is a membrane bound system, we first sought to identify the cellular site of O 2 . production. In the search of cellular sources of RV16induced O 2 . production, we first confirmed our previous findings of rapid induction of O 2 . by RV16 in membrane-free cyto-solic fractions (Fig. 1A). We next investigated total cell homogenates, which included cell membranes (Fig. 1B)  . production could derive from the activity of NADPH oxidase in experimental conditions where both the cytosolic and membrane fractions of the homogenates were pooled together. The membrane fractions were added, together with the substrate of NADPH oxidase, NADPH, in the presence and absence of an inhibitor of the NADPH enzyme system, diphenylene iodonium chloride (38). NADPH addition would increase O 2 . production if production is NADPH oxidasedependent. We found that both the addition of NADPH (Fig. 1C) and diphenylene iodonium chloride (Fig. 1D)   ) production. A (n ϭ 5) and B (n ϭ 5), uric acid production kinetics in confluent A549 cells and HBEC, respectively. Cells were exposed to live RV16 (squares), RV16 physically removed by filtration (triangles), or medium alone (open circles). The arrow indicates when uricase was added to the system. Newly generated uric acid was measured spectophotometrically in each sample and expressed as micromolar and normalized per mg of protein. C, uric acid production in A549 exposed to minor group rhinovirus RV1B (live RV1B (squares), RV1B physically removed by filtration (triangles), or medium alone (open circles)) (n ϭ 5). D (n ϭ 5) and E (n ϭ 5), A549 cells (D) and HBEC (E) were exposed for 1 h to live RV16 (RV16), RV16 physically removed by filtration (f-RV16), medium alone (Medium), or diluent alone (Diluent). Where indicated cells were pretreated 4 h before the infection with oxypurinol and than exposed for 1 h to medium alone (Oxy) or RV16 (OxyϩRV16) (***, p Ͻ 0.001 versus medium, diluent, f-RV16, Oxy-treated cells and versus RV16-infected cells pretreated with Oxy).
virus-specific (24) and confirmed this in the present study, as an inoculum from which virus had been removed by molecular weight filtration (f-RV16) did not induce any O 2 . production ( Fig. 1, A and B). We next wished to investigate RV induction of O 2 . in primary bronchial epithelial cells, as we had only previously demonstrated this at a single time point of 20 min. Similar results were found in primary HBEC obtained by bronchial brushing from healthy volunteers, with induction being significant at 20 and 120 min and peaking at 60 min (Fig. 1E). To investigate whether these results were confined to the major RV group, to which RV16 belongs, which binds ICAM-1 as surface receptor (43,44), experiments were performed using RV1B of the minor RV group, which uses members of the low density lipoprotein receptor family. The fact that following RV1B infection the kinetics of O 2 . production both in the A549 cell line (Fig. 1F) and HBEC (Fig. 1G) were not different to those observed with RV16 indicates that the effect is receptor independent.

Rhinovirus Induces O 2 . Production via the XD/XO Enzyme
System-To confirm that O 2 . is generated via the XD/XO enzyme system, we next investigated RV induction of uric acid, the other product, besides O 2 . , of XO degradation of the purine base xanthine. These experiments were conducted at 1 h after infection, i.e. at the peak of O 2 . generation. Fig. 2A shows the kinetics of uric acid production, after addition of substrate xanthine, in cytosolic fractions of A549 cell homogenates after exposure for 1 h to RV16, f-RV16, or medium alone. Uric acid was detected only in RV16-infected samples. Its identification was confirmed by addition, after 15 min of the kinetic assay, of uricase, which degrades uric acid to allantoin (36, 37). As expected, uric acid was rapidly degraded by uricase confirming that the XD/XO system was activated by rhinovirus infection (Fig. 2A). Similar findings were observed in homogenate samples obtained from HBEC, in the same experimental conditions (Fig. 2B), confirming rhinovirus activation of XD/XO in primary cells. The fact that following RV1B infection the kinetics of uric acid production were identical (Fig.  2C) to those observed with RV16 ( Fig. 2A) (Fig. 2D). Similar findings were observed in homogenate samples obtained from HBEC in the same experimental conditions (Fig. 2E), as in both cases oxypurinol reduced O 2 . production to levels observed with medium or diluent alone or inactivated virus. Rhinovirus Induces Proteolytic Activation of XD/XO Enzymatic System-In other experimental systems, the XD/XO enzymatic system is able to produce O 2 . when it is converted to the oxidase form by proteolytic activity, exerted by a serine protease, which partially hydrolyzes the enzyme to its active form (40,45). To investigate the mechanisms of rhinovirus induction of XD/XO, cells were next infected in the presence or absence of serine protease inhibitors, or with cysteine or metalloprotease inhibitors as controls. RV16 infection (20 min to 1 h) failed to generate O 2 . in the cytosol when epithelial cells were pretreated for 4 h with protease inhibitors PMSF, leupeptin (Leu), pepstatin (Pep), or aprotinin (Apr), which all act as serine protease inhibitors (Fig. 3A). Serine protease involvement was confirmed as neither the metalloprotease inhibitor phenanthroline (Phe, Fig. 3B) nor the cysteine protease inhibitor E-64 (Fig. 3C)   . production in A549 cells. In A-C, where indicated (ϩ) cells were pretreated for 4 h with protease inhibitors, then exposed, for 20 min (empty bars) or 1 h (filled bars) to RV16 or medium alone (in A, *** p Ͻ 0.001 versus all other conditions; B and C, *** p Ͻ 0.001 versus medium alone treated cells with or without inhibitor pre-treatment).

XD Knockdown Reduces Rhinovirus-induced O 2 . Production-
To further confirm the role of the XD/XO system in rhinovirusinduced O 2 . production, we performed experiments in which XD expression in A549 cells was knocked down by siRNA. Transfection of XD siRNAs resulted in marked suppression of XD mRNA expression as compared with scrambled siRNAtransfected cells (Fig. 4D), with a peak of inhibition at 24 h. At this time point, XD knockdown suppressed RV16-induced O 2 .

Rhinovirus Depletes Intracellular Reduced GSH-To investigate the consequences of RV-induced O 2 .
production on intracellular redox equilibrium, we evaluated whether the concentration of the intracellular reducing agent GSH is modified by RV16 infection. In A549 cells, as the duration of RV16 infection increased, endogenous stores of GSH were progressively reduced and complete depletion was observed at 1 h after infection (Fig.  5A). Similar results were found with RV1B (Fig. 5B). The HPLC absorbance peak for GSH is representatively shown for medium-treated cells in Fig. 5C and depletion of the HLPC peak in RV16-infected cells is representatively shown in Fig. 5D. As observed with O 2 . induction, RV16-induced depletion of intracellular GSH at 1 h after infection was completely inhibited when A549 cells were pretreated with either oxypurinol or serine protease inhibitors, but not with metalloprotease inhibitor phenanthroline (Phe) or cysteine protease inhibitor (E-64) (Fig. 5E). Thus, RV16-induced depletion of intracellular GSH occurs via XO activation and subsequent cytosol O 2 . production.
The same findings were also observed when A549 cells were infected with minor group RV1B (Fig. 5F), to indicate that the effect was receptor independent. Increasing Intracellular GSH Inhibits Rhinovirus-induced Intracellular GSH Depletion and O 2 . Production-Because reduction of intracellular reducing power is "per se" a known mechanism of activation of XO (46,47), we next investigated whether by increasing intracellular GSH with the reducing agents H 2 S or exogenous GSH, we could block the activation of XO induced by RV infection. We first showed that pretreatment of A549 cells with H 2 S and exogenous GSH increased intracellular GSH levels in a dose-dependent manner (Fig. 6A). A549 cells were then pretreated for 12 h with 2 mM H 2 S to enhance intracellular GSH (Fig. 6B) and then infected for 1 h with RV16. No reduction of endogenous GSH was observed after a 1-h RV16 infection (Fig. 6C), confirming that enhancing intracellular GSH protected cells against virus-induced GSH depletion. Similar protection was observed with 10 mM exogenous GSH (data not shown) treatment before the infection. Pretreatment with either 2 mM H 2 S or 10 mM exogenous GSH, not only increased intracellular GSH levels but also completely inhibited RV16 induced O 2 . production at 1 h after infection, as assessed by SOD-inhibitable cytochrome c reduction assay ( Fig. 6D). In these experiments uric acid production was similarly suppressed (Fig. 6E). Thus, the results indicated that increasing GSH intracellular storage with exogenous GSH or H 2 S completely inhibited rhinovirus-induced XO activation and O 2 . production.

XO Inhibition Reduces Rhinovirus-induced Chemokine
Production-RV infection of bronchial epithelial cells induces the expression of several proinflammatory cytokines, including many that are involved in neutrophil chemoattraction and activation (i.e. IL-8 and Gro-␣). Neutrophil inducing cytokines are not effectively suppressed by the currently available asthma therapies, i.e. steroids or long acting ␤ agonists (10). Because the induction of these mediators occurs through oxidative sensitive pathways (e.g. NF-B signaling activation (17)), we evaluated whether oxypurinol inhibition of XO-mediated O 2 . generation affects rhinovirus-induced IL-8 and Gro-␣ production in A549 respiratory epithelial cells. Significant induction of IL-8 and Gro-␣ was apparent at 4 h post-RV16 infection (Fig. 6, A  and B). Oxypurinol pre-treatment significantly reduced both IL-8 (Fig.  7A) and Gro-␣ RV16-induced production (Fig. 7B), whereas diluent had no effect. These data confirmed that inhibition of XO was effective in suppressing rhinovirus-induced neutrophil chemokine production.

Effect of Rhinovirus Infection on NF-B Activation, Modulation by
Oxypurinol-To assess whether rhinovirus infection activates NF-B and whether this is mediated by O 2 . production, we measured activated NF-B in nuclear extracts in A549 cells following RV16 infection with or without a 4-h oxypurinol pretreatment. p65 nuclear concentration was 40.7 Ϯ 6.3 pg/l in unstimulated conditions. In accordance with previous data (24,26) we found that 30 min RV16 infection significantly induced p65 nuclear translocation (2-fold versus unstimulated; p Ͻ 0.05). Four h pre-treatment with 20 M oxypurinol significantly inhibited RV16-induced NF-B activation (p Ͻ 0.05 versus RV16-infected cells not pretreated) (Fig. 7C). These data indicate that rhinovirus-induced O 2 . is involved in the activation of the NF-B-dependent pro-inflammatory pathways.
Effect of Tested Compounds on Rhinovirus Replication and Infectivity-Control experiments were performed to assess whether any of the antioxidant approaches used above had any antiviral activity that could provide an alternative explanation for our findings. HeLa cells were infected with RV16 for 1 h, i.e. until the peak of O 2 . generation was reached and intracellular GSH fully depleted, in the presence or absence of antiproteases, oxypurinol or GSH, or H 2 S pretreatment for the time intervals previously specified for each compound. The chosen concentrations for each compound were the highest utilized in the present study. No difference was found in the progression of RV16- In panels C and D an arrow indicates the position of the GSH peak. E (n ϭ 5), HPLC evaluation of intracellular GSH in A549 cells exposed for 1 h to medium alone (white bar), live RV16 (RV16), or RV16 physically removed by filtration (f-RV16, light gray bar). Where indicated A549 cells were pretreated for 4 h with oxypurinol (Oxy) or protease inhibitors PMSF, Leu, Pep, Apr, Phe, and E-64, then exposed for 1 h to medium alone (gray bars) or live RV16 (black bars). In panel F (n ϭ 5) A549 cells were exposed to RV1B (***, p Ͻ 0.001).
induced cytopathic effects measured on a daily basis by visual assessment (data not shown). At 5 days there was no difference in virus yields expressed as mean TCID 50 /ml values between treated and control samples for all tested compounds (Fig. 8A). These results were confirmed by TaqMan PCR assay showing progressive rhinovirus replication (i.e. increased rhinovirus RNA), which was not affected by antiprotease pretreatment (data not shown at 4 h infection; data at 8 h represented in Fig. 8B). . production and to affect intracellular GSH levels.

DISCUSSION
Several studies have described oxidant generation following respiratory virus infection both in vitro and in vivo (20,48). In some studies a role for virus-induced oxidants in the production of inflammatory responses/mediators has been identified (49). However, the molecular mechanisms regulating generation of oxidant species by viruses in biological systems have never been fully investigated.
The findings of the present study indicate a complex mechanism of oxidant induction following activation of XO induced by rhinovirus infection (Fig. 9). A "vicious circle" would represent the final scenario where activation of XO, initiated immediately after infection via proteolysis of XD to XO, is thereafter implemented via a non-proteolytic mechanism mediated by oxidative consumption of intracellular reducing capacity via depletion of GSH stores. Depletion of intracellular reducing agents is a known mechanism of activation of XO and O 2 .
production (46,47). The involvement of rhinovirus-induced oxidants in GSH consumption was confirmed by the finding that, when XO activation was inhibited, rhinovirus infection did not result in GSH depletion. The sequence of events represented in Fig. 9 is supported by the timing of the different steps involved, with O 2 . production being rapidly induced 20 min after infection ( Fig. 1), whereas GSH depletion is undetectable 20 min after infection and thereafter progressively increases for 40 min, being complete at 60 min after infection (Fig. 5, A and B).
The fact that in a reducing environment no uric acid was produced and that the intracellular concentration of GSH was unchanged after rhinovirus infection confirms the inverse relationship between GSH intracellular concentra-  5). B, representative HPLC profiles of intracellular GSH detection in control uninfected A549 cells pretreated with 2 mM exogenous H 2 S. C, representative HPLC profile of intracellular GSH detection in A549 cells pretreated with 2 mM exogenous H 2 S and infected with RV16. In panels B and C an arrow indicates the GSH peak. D, A549 cells were infected for 1 h with RV16 or with or without 12 h pretreatment with 2 mM H 2 S (H 2 S) or 10 mM exogenous GSH (GSH). Superoxide anion production was assessed with SOD-inhibitable cytochrome c reduction assay (n ϭ 5, ***, p Ͻ 0.001) . E, uric acid production kinetics in confluent A549 cells exposed to RV16 infection with diluent (squares), 2 mM H 2 S (open circles), or 10 mM GSH (triangles) with a 12-h pre-treatment. The arrow indicates when uricase was added to the system. Newly generated uric acid was measured spectrophotometrically in each sample and expressed as micromolar and normalized per mg of protein (n ϭ 5).
tion and XO activation. Previous studies have documented GSH depletion following viral infections (50), however, the underlying mechanisms were not described. The involvement of a NADPH oxidase-like enzyme in rhinovirus-induced oxidative stress has been previously described (23). In contrast to the study by Kaul and colleagues (23), our study focuses strictly on the early oxidative events occurring within cells immediately after rhinovirus infection (with a peak at 1 h after infection). Also, by using a method able to specifically detect newly generated O 2 . , i.e. the SOD-inhibitable cytochrome c reduction assay (51), we directly evaluated O 2 . intracellular production, and not oxidative stress in general.
Moreover, at variance with Kaul and colleagues (23), we employed oxypurinol, which completely inactivates XO by direct binding to the enzyme active site (28), and not allopurinol, which only partially inhibits the enzyme. Treatment with oxypurinol completely abolished rhinovirus-induced O 2 . generation, thus confirming the specificity of our findings on XO activation. These and other differences between the two studies, in particular different samples for analyses (intracellular versus extracellular compartments) and different timing are likely explanations for the different results reported. In the experimental setting evaluating the functional effect of oxypurinol on cytokine production, to exclude nonspecific interference, we tested the effect of oxypurinol on IL-1␤induced IL-6 production. This is a pathway for which mechanisms other than oxidant generation are considered relevant (52,53). We found that IL-1␤ significantly induced IL-6 in a dose-dependent manner and that oxypurinol does not affect IL-6 production induced by 4-h IL-1␤ stimulation (data not shown). Pretreatment with all four serine antiproteases investigated, but not the cysteine and metalloprotease inhibitors phenanthroline and E-64, also completely prevented O 2 . production in response to rhinovirus infection. Previous studies have documented that limited proteolysis of XD with serine proteases converts the enzyme to the active XO form (40). With the exception of phenanthroline and E-64, all antiproteases here utilized can act as serine protease inhibitors, or serine-cysteine protease inhibitors (29,30,32). Pepstatin, an inhibitor of aspartate proteases, may also act on serine proteases because of active site target similarities (31). The fact that phenanthroline and E-64, two protease inhibitors for which no serine protease inhibitory activity is documented, do not prevent rhinovirus-induced O 2 .
production and intracellular GSH depletion supports the serine specificity of the pathway of activation of XD to XO following rhinovirus infection. Not only activity but also the amount of XD protein was found to be decreased after rhinovirus infection with a parallel increase of the proteolytic fragment containing the active site of  5) and B (n ϭ 5), A549 cells were exposed for 1 h to medium alone (Medium), RV16 physically removed by filtration (f-RV16) or live RV16 (RV16). Where indicated cells were pretreated for 4 h with oxypurinol (OxyϩRV16) or diluent (DiluentϩRV16) before the infection. Supernatants were harvested at 4 h and levels of IL-8 (panel A) and GRO-␣ (panel B) were assessed (***, p Ͻ 0.001). C, A549 cells were exposed for 20 min to RV16 with (dark gray bar) or without (black bar) 20 M oxypurinol (Oxy) after a 4-h pretreatment and NF-B activation was assessed in nuclear extracts (n ϭ 3, *, p Ͻ 0.05 versus medium alone). Control experiments were performed to evaluate the effects of the compounds used in the study on rhinovirus replication, to assess whether they have antiviral activities that could provide an alternative explanation for our findings. Virus binding to host cells, virus cell entry, and infectivity would be the most relevant events in our experimental conditions, where the mechanisms analyzed begin a few minutes after the infection has started (24). By using HeLa cells we were able to examine the effects of the tested compounds independently from the up-regulatory effect of rhinovirus on its own receptor (ICAM-1) observed in respiratory epithelial cells (26). Indeed, ICAM-1 surface expression decreases on rhinovirus-infected HeLa cells in parallel with the appearance and severity of cytopathic effects (data not shown). Moreover, we further exclude any direct anti-RV16 effect of the proteases used in the present study by assessing viral replication by TaqMan Real Time PCR (Fig. 8B). Taken together these experiments demonstrated that the tested compounds at the highest concentrations used in our experimental setting had no effect on rhinovirus replication and infectivity.
Although we cannot exclude the involvement of rhinovirus 3C proteolytic enzyme in XD/XO activation, we believe this possibility is unlikely to be relevant in our experimental conditions where oxidant activation was already detectable at 20 min and peaked at 1 h after infection, whereas 2-4 h infection is required for rhinovirus 3C protease to be produced (54).
Despite the effort and resources expended, no antiviral drugs are currently marketed for the prevention or treatment of rhinovirus infection (6). In the absence of effective anti-viral ther-apies, development of therapies that blocked the inflammatory responses to infection would be a major advance. Our demonstration of the molecular mechanisms by which rhinovirus activates oxidant generation, a crucial step in the complex inflammatory response to infection (23,24) could open new possibilities in the search for therapeutic targets for future intervention. In particular, the documentation that: (a) exogenous interventions able to increase intracellular reducing agent storage can block rhinovirus-mediated activation of the vicious circle that leads to sustained production of O 2 . , and (b) inhibition of XO, with protease inhibitors or specific inhibitors (oxypurinol) inhibits O 2 . and pro-inflammatory mediator production, indicate promising options for the development of treatment for rhinovirus-induced diseases including the common cold and exacerbations of asthma and COPD.