Rhinovirus Replication in Human Macrophages Induces NF-κB-Dependent Tumor Necrosis Factor Alpha Production

Abstract

Rhinoviruses (RV) are the major cause of acute exacerbations of asthma and chronic obstructive pulmonary disease (COPD). Rhinoviruses have been shown to activate macrophages, but rhinovirus replication in macrophages has not been reported. Tumor necrosis factor alpha (TNF-α) is implicated in the pathogenesis of acute exacerbations, but its cellular source and mechanisms of induction by virus infection are unclear. We hypothesized that rhinovirus replication in human macrophages causes activation and nuclear translocation of NF-κB, leading to TNF-α production. Using macrophages derived from the human monocytic cell line THP-1 and from primary human monocytes, we demonstrated that rhinovirus replication was productive in THP-1 macrophages, leading to release of infectious virus into supernatants, but was limited in monocyte-derived macrophages, likely due to type I interferon production, which was robust in monocyte-derived but deficient in THP-1-derived macrophages. Similar to bronchial epithelial cells, only small numbers of cells supported complete virus replication. We demonstrated RV-induced activation of NF-κB and colocalization of p65/NF-κB nuclear translocation with virus replication in both macrophage types. The infection induced TNF-α release in a time- and dose-dependent, RV serotype- and receptor-independent manner and was largely (THP-1 derived) or completely (monocyte derived) dependent upon virus replication. Finally, we established the requirement for NF-κB but not p38 mitogen-activated protein kinase in induction of TNF-α. These data suggest RV infection of macrophages may be an important source of proinflammatory cytokines implicated in the pathogenesis of exacerbations of asthma and COPD. They also confirm inhibition of NF-κB as a promising target for development of new therapeutic intervention strategies.

Acute exacerbations of asthma and chronic obstructive pulmonary disease (COPD) are the major causes of morbidity and mortality in both diseases. Rhinoviruses (RVs) are the most common trigger of acute exacerbations (19, 30, 42); however, the mechanisms by which RVs provoke exacerbations are not well understood. The airway epithelium is thought to be the site of RV replication, and many studies have observed RV induction of proinflammatory cytokines, chemokines, and adhesion molecules in epithelial cells (28, 47, 52, 59). During RV infection, the number of epithelial cells infected with virus is low both in vitro and in vivo (8, 39); nonetheless, it is currently believed that inflammatory cytokine production from RV-infected epithelial cells is an important mechanism contributing to the pathogenesis of exacerbations of asthma and COPD (37).

Tumor necrosis factor alpha (TNF-α) is a potent inflammatory cytokine implicated in the pathogenesis of asthma and COPD (25, 31, 60). TNF-α has multiple biologic effects relevant to the pathogenesis of exacerbations of airway disease, including the enhanced release of other proinflammatory/chemotactic mediators, up-regulation of adhesion molecules, enhanced migration of eosinophils and neutrophils, and induction of hypercontractile airway smooth muscle (4, 36, 60). TNF-α is detected in increased amounts in bronchoalveolar lavage during experimental RV infection in asthma and is also increased in acute exacerbations of COPD (1, 9). These data implicate TNF-α in the pathogenesis of acute exacerbations of both diseases; however, TNF-α cannot be detected in significant quantities in the supernatants of RV-infected epithelial cells, suggesting that an alternative cellular source may exist (59).

Lung macrophages (Mφ) are an important source for TNF-α, are the most numerous cells in the airway lumen, and are quickly recruited during inflammatory processes of the lung (18). Following interaction with various bacterial and viral pathogens, they become activated and secrete a wide range of antiviral, proinflammatory, and/or immunomodulatory cytokines (62). Their position in the airway and their high levels of expression of RV receptors ICAM-1 and low-density lipoprotein (LDL) receptor suggest Mφ may be a target for RV infection and TNF-α production (17, 34). Relatively few studies have investigated interactions between RVs and cells of Mφ or monocytic origin. Induction of interleukin-8 (IL-8), IL-10, IL-12, TNF-α, and monocyte chemoattractant protein-1 (MCP-1) and alteration of surface expression of CD14, CD80, and CD69 in peripheral blood mononuclear cells, monocytes, or Mφ have all been reported after exposure to RV (14, 16, 21, 29, 44, 45, 56). However, the mechanisms responsible for this monocyte/Mφ activation are unclear. Production of IL-10, TNF-α, and MCP-1 was reported to be replication independent, while IL-8 secretion and CD69 up-regulation appeared in part replication dependent, as UV-inactivated RV had a significantly reduced effect compared to live virus (14, 16, 21, 29, 44). When evidence for RV replication in Mφ was specifically investigated, attachment and entry of virus was detected, but no further evidence of replication was detected, as RV titers in supernatants decreased over time and RV RNA synthesis was not observed (14).

In addition, the molecular mechanisms regulating RV induction of inflammatory cytokines in Mφ are poorly understood, though RV induction of MCP-1 in monocytes and Mφ is reported dependent on the p38 mitogen-activated protein kinase (MAPK) MAPK/AP-1 pathway (21). Studies of RV infection of respiratory epithelial cells indicate a central role for NF-κB in up-regulation of several proinflammatory molecules (32, 46, 47, 67). We therefore hypothesized that rhinoviruses can replicate in human Mφ and that replication causes activation and nuclear translocation of NF-κB, leading to TNF-α production. To address this hypothesis, we studied RV replication in THP-1-derived Mφ and primary monocyte-derived Mφ (MDM), both well-established models of human Mφ (6, 7, 63). We also investigated the roles of virus replication and NF-κB in TNF-α production.

MATERIALS AND METHODS

Cell lines and viruses.

HeLa and THP-1 cell lines (European Collection of Cell Cultures) were cultured in E-MEM (Invitrogen) and RPMI 1640 (Invitrogen), respectively, with 10% fetal calf serum (FCS; Invitrogen). RV serotypes 16, 9 (major group; receptor, ICAM-1), and 1B (minor group; receptor, LDL receptor) were grown in HeLa cells and prepared as previously described (47). Virus stocks or supernatants of RV-infected Mφ were titrated on HeLa cells to ascertain their 50% tissue culture infective dose (TCID₅₀)/ml (47). The identities of all RVs were confirmed by neutralization using serotype-specific antibodies (ATCC). UV inactivation and generation of filtered virus were performed as previously described (47). Human respiratory syncytial virus (RSV) strain A2 was a gift from P. J. Openshaw.

Generation of THP-1-derived Mφ and MDM.

For all experiments, THP-1 cells were used after differentiation to Mφ. To induce differentiation THP-1, 0.75 × 10⁶/ml in RPMI 1640 with 5% FCS were treated with phorbol myristate acetate (PMA; Sigma) at 200 nM for 24 h and placed for a further 24 h in 5% FCS without PMA, to allow the cells to rest.

Mφ were generated from peripheral blood monocytes using a previously described protocol that was slightly modified (49). Briefly, mononuclear cells were separated from total blood of healthy donors by density gradient centrifugation using Ficoll-Paque (Sigma). Monocytes, isolated by positive selection using anti-CD14 magnetic beads (magnetic-activated cell sorter), were differentiated to Mφ by culture for 7 days in macrophage serum-free medium (Invitrogen) supplemented with 10 ng/ml granulocyte-macrophage colony-stimulating factor (Biosource) (49). The study was approved by the St. Mary's NHS Trust ethics committee, and informed consent was obtained from all volunteers.

Infection of Mφ.

THP-1-derived Mφ or MDM generated as described above in 12-well plates were exposed to RV or RSV at a multiplicity of infection (MOI) of 1 with continuous shaking. After 1 h, unattached virus was removed, cells were washed extensively, and 1 ml of RPMI 1640 containing 5% FCS for THP-1 or macrophage serum-free medium for MDM was added to each well. This was considered time point zero (0 h). Supernatants and RNA or protein lysates (for Western blot analysis) were harvested at different time points and stored at −80°C for further use.

RNA extraction, reverse transcription, and TaqMan real-time PCR for viral RNA quantification.

Whole-cell RNA was extracted using TRIzol according to the manufacturer's instructions (Invitrogen). Two μg of total RNA was reverse transcribed into cDNA using Omniscript reverse transcriptase and components as directed by the manufacturer (QIAGEN). RV cDNA was measured by Taqman PCR (ABI) and normalized using 18S rRNA. Primers and probe sequences along with the protocol used for Taqman real-time PCR have been published elsewhere (23, 24). Data were analyzed using version 1.0 ABI Prism 7000 SDS software (ABI) and converted to copy numbers using a standard curve for a plasmid of known concentration containing the amplified region of the RV genome.

Western blotting for RV 3C protease expression.

THP-1-derived Mφ were lysed directly into sodium dodecyl sulfate (SDS) sample buffer (Invitrogen). After electrophoresis in a 12% SDS-polyacrylamide gel electrophoresis (PAGE) gel (Invitrogen), proteins were transferred to a polyvinylidene difluoride membrane (Amersham). All these steps were performed under reducing conditions. After blocking, membranes were incubated with a 1/5,000 dilution of rabbit anti-3C RV protease antibody (provided by Svetlana Amineva, University of Wisconsin—Madison) (3) followed by horseradish peroxidase-conjugated swine anti-rabbit antibody (Serotec). Bands were visualized by chemiluminescence with the ECL Western blotting detection reagent (Amersham). After scanning, densitometry analysis of the band corresponding to the 3C protease was performed.

Infectious center assay to determine numbers of infected cells.

To asses the number of virus-releasing cells after RV infection, we employed an infectious center assay using modifications of a published protocol (39). THP-1-derived Mφ or MDM were infected with RV16 as described above and incubated at 37°C. After 1 h, cells were trypsinized and plated over monolayers of HeLa cells in different dilutions. After a further 1-h incubation, to allow the cells to settle, medium was carefully removed and RMPI 1640 containing 5% FCS and 0.3% Indubiose (Invitrogen) was added. Plates were incubated at 37°C for 4 days, and at the end of incubation cells were fixed by addition of 10% formaldehyde and stained with crystal violet, 0.1%. Virus released from infected Mφ infects only the HeLa cells on which infected Mφ have settled, thus creating a plaque. The number of virus-releasing Mφ was assessed by enumeration of plaques.

Transient transfection and NF-κB luciferase reporter assay to assess NF-κB activation.

Before RV16 infection, THP-1-derived Mφ were transiently transfected with a construct containing the luciferase gene under the control of a minimal promoter containing four NF-κB binding sites (BD Clontech) and a constitutive β-galactosidase-expressing construct (Invitrogen) using JetPEI (Polyplus) according to the manufacturer's protocol. After RV infection, cells were incubated for 48 h to allow expression of the reporter gene. Preparation of cell lysates and luciferase assays were performed as recommended by the manufacturer (Promega) using an Autolumat LB953 (Berthold Systems Inc.). All luciferase measurements were normalized to β-galactosidase activity measured using the β-galactosidase enzyme assay system (Promega).

Immunofluorescent staining for confocal microscopy colocalization of virus infection and NF-κB translocation.

THP-1-derived Mφ or MDM seeded in eight-well microscope slides were infected as described above. At different time points, cells were fixed in 4% paraformaldehyde and permeabilized with Triton X-100, 0.2%. After overnight blocking with phosphate-buffered saline containing 10% FCS and 1% bovine serum albumin, cells were incubated with rabbit anti-3C RV protease serum, 1/500, with or without monoclonal mouse antibody anti-p65 (1/200; Santa Cruz) for 1 h in blocking buffer followed by goat anti-mouse AlexaFluor 546, 1/200, and goat anti-rabbit AlexaFluor 488, 1/200 (Molecular Probes) as secondary antibodies in blocking buffer for 45 min. The slides were coverslipped in 4′,6′-diamidino-2-phenylindole (DAPI)-containing mounting medium and examined using an LSM 510 confocal microscope (Zeiss). To evaluate the percentage of virus-infected cells, the number of RV 3C-positive cells in 200 nucleated cells was counted and expressed as a percentage of total cells.

TNF-α, IFN-α, and IFN-β enzyme-linked immunosorbent assay (ELISA).

Levels of TNF-α and type I IFNs in supernatants of THP-1-derived Mφ or MDM were measured using paired antibodies and standards commercially available for TNF-α (Biosource) or commercially available kits for IFN-α and IFN-β (Biosource) following the manufacturer's recommendations. The sensitivity of the assay was 10 pg/ml for TNF-α, 5 IU/ml for IFN-β, and 15 pg/ml for IFN-α.

p38 MAPK and NF-κB inhibition.

To evaluate the role of NF-κB or p38 MAPK in TNF-α secretion, we carried out inhibition experiments with chemical inhibitors of these pathways. AS602868 (a gift from Ian Adcock), an inhibitor of IKKβ (12), or CAPE (Calbiochem), an inhibitor of p65 translocation (41), was used to pretreat the THP-1-derived Mφ or MDM for 1 h before infection, at concentrations ranging from 0.01 to 5 μM and 25 to 1.25 μg/ml, respectively (33, 41). The same concentration of drug was added to the medium after infection. Similar experiments were carried out with SB203580 (Calbiochem), an inhibitor of p38 MAPK, at a concentration of 10 μM (20).

Statistical analysis.

The results were analyzed using GraphPad Prism version 4.00 for Windows (GraphPad Software, California). Results of at least three separate experiments were expressed as means ± standard errors of the means (SEM) and analyzed using analysis of variance (ANOVA) for multiple comparisons, followed where appropriate by paired Student's t tests for paired comparisons.

RESULTS

Rhinovirus replicates efficiently in THP-1-derived Mφ.

To investigate whether RV16 can replicate in human Mφ, we first used THP-1-derived Mφ. After differentiation cells were exposed to RV16, and supernatants, total cell RNA, or intracellular proteins were harvested at different time points. Intracellular levels of RV RNA were determined using real-time PCR. Relatively high levels of RV16 RNA were detected at time zero (4.53 ± 0.486 log₁₀ copies/μg total RNA), followed by a small reduction at 1 h and an eclipse phase at 2 h, when a decrease of almost 1 log₁₀ was observed compared with 0 h (Fig. 1A). Following the eclipse, viral RNA production increased rapidly, reaching a peak of greater than 6 log₁₀ copies at 8 h (a 2.45-log₁₀ increase compared with 2 h [P < 0.01] and a 1.53-log₁₀ increase compared with 0 h [P < 0.05]). Thereafter, viral RNA levels decreased gradually but were still significantly elevated at 24 h and were greater than 5 log₁₀ copies at both 48 and 72 h (Fig. 1A).

FIG. 1.

RV replication in THP-1-derived Mφ. (A) THP-1-derived Mφ were infected for 1 h with RV16 (input MOI of 1 TCID₅₀/cell), and RNA was extracted from cell lysates at 0, 1, 2, 4, 8, 24, 48, and 72 h postinfection. RV RNA expression was quantified by using TaqMan, and data are presented as the number of copies per μg of total RNA. The results are expressed as means ± SEM (n = 4). Statistical significance between the eclipse (2 h) and other time points is indicated: *, P < 0.05; **, P < 0.01; statistical significance between 0 h and other time points is indicated by # for P < 0.05. (B) Cells were infected as for panel A, supernatants were harvested, and the amount of infectious virus released into the supernatants was assessed by virus titration. The results are expressed as means ± SEM (n = 4). **, P < 0.01; ***, P < 0.001. (C) Cells infected as for panel A were lysed and analyzed for the presence of RV 3C protease by Western blotting. A representative image of four different experiments is shown. (D) THP-1-derived Mφ were infected with RV16 (MOI of 1) for 1 h, and the number of virus-releasing cells was assessed in an infectious center assay. The numbers of infected THP-1 cells overlaid in duplicate on HeLa cells are indicated at the bottom (see Materials and Methods for more details). Numbers of plaques counted (each representing an infectious center derived from a rhinovirus-releasing THP-1 cell) at each concentration of overlaid THP-1 cells are indicated in the boxes with arrows. The experiment is representative of four, in which 5.66 ± 0.63% of cells were associated with plaque formation.

Full productive replication was then investigated by assessment of virus release into supernatants of cells. Very low titers of RV16 were detected in supernatants at early time points, 0 to 4 h, but they remained unchanged during this period. RV16 titers then increased progressively and were significantly elevated compared to 0 h at 8, 24, 48, and 72 h, peaking at 24 h (2.56 log₁₀ TCID₅₀/ml increase compared with 0 h; P < 0.001), followed by a gradual decline thereafter (Fig. 1B).

Synthesis of new RV proteins was also investigated by Western blot analysis of cell lysates using a polyclonal rabbit serum against RV 3C protease, which is a nonstructural protein expressed only during viral replication (3, 54). RV 3C protease was undetectable at time zero and thereafter could be detected in increasing amounts (Fig. 1C). When analyzed by densitometry, the 3C protease expression was significantly elevated between 8 and 48 h (P < 0.01), peaking at 24 h (P < 0.001) (data not shown). In addition 3ABC protease, a precursor of 3C protease, was also clearly detected at both 8 and 24 h (Fig. 1C).

Despite being acknowledged as the primary site of RV replication in the lower airway, only a small minority of bronchial epithelial cells are infected in vitro and similar frequencies are observed in vivo (5, 8, 39). Therefore, having demonstrated productive replication in THP-1-derived Mφ, we next wished to determine the frequencies of virus-infected cells, to permit comparison with epithelial cells. For this, an infectious center assay was used to determine the number of RV16-releasing cells, using methods adapted from those previously used to determine frequencies of infected epithelial cells (39). When THP-1 Mφ were exposed to RV16, 5.66 ± 0.63% of cells (n = 4) released sufficient infective virus to lead to plaque formation on HeLa cells (Fig. 1D).

We next investigated frequencies of cells in which replication was demonstrable by immunofluorescent staining for RV16 3C protease, with the polyclonal rabbit serum, using confocal microscopy to enumerate infected and noninfected cells. RV 3C protease was detected as early as 2 h after RV16 infection of THP-1-derived Mφ and persisted until 48 h (data not shown). The highest frequency of positive cells for RV16 3C protease (9 ± 1.4%; n = 4) was detected at 6 h after infection (see Fig. 3B, below). No staining was detected when cells were treated with medium or UV RV16 (data not shown).

FIG. 3.

NF-κB activation after RV infection of Mφ. (A) THP-1-derived Mφ transiently transfected with NF-κB-Luc minimal promoter were infected with RV16 (MOI of 1) or medium treated and analyzed for luciferase expression at 48 h. The results were normalized to β-galactosidase and are expressed as the relative fold induction over the medium control. The results are expressed as means ± SEM (n = 4). *, P < 0.05 compared with medium. (B) THP-1-derived Mφ or MDM in eight-well chambered slides were infected with RV16 (MOI of 1). At 6 h postinfection (for THP-1-derived Mφ) or 4 h postinfection (for MDM), cells were fixed, permeabilized, and stained with RV 3C protease rabbit antiserum (green) and mouse anti-p65 monoclonal antibody (red), followed by an appropriate secondary antibody. A representative image of four independent experiments for THP-1-derived Mφ and three for MDM is presented. Magnification, ×800. Green, RV 3C staining; red, p65 staining; blue, overlay with DAPI staining for nuclei. Cells with rhinovirus infection are indicated by green arrows, and cells with NF-κB nuclear translocation are shown by the red arrows. In the overlay, most cells have dual staining, indicating that p65 nuclear translocation occurred principally in virus-infected cells.

Consistent with these infection frequencies and with data recently reported in primary bronchial epithelial cells from normal volunteers (65), no cytopathic effects were observed in RV16-infected THP-1 Mφ when inspected by inverted microscopy.

Limited rhinovirus replication can be detected in MDM.

It has been reported that human RVs are unable to replicate in Mφ (14). However, UV inactivation experiments have suggested replication is required for up-regulation of several genes following RV infection in monocytes or Mφ (16, 21, 29, 44, 45), and we have shown efficient replication in THP-1-derived Mφ. Therefore, we wished to investigate to what extent rhinovirus replication can take place in primary Mφ. For this we used MDM, a model with close resemblance to alveolar Mφ (2).

MDM were exposed to RV16, and supernatants or total cell RNA was harvested at different time points. RV16 was detected at low titers at 0 h and, after a small increase between 2 and 8 h, virus titers continuously decreased thereafter; however, virus was still detectable at 72 h (Fig. 2B) (P > 0.05). We next investigated intracellular levels of RV RNA using real-time PCR. High levels of viral RNA were detected at 0 h (6.767 ± 0.098 log₁₀ copies/μg total RNA), and these levels had decreased by 1 log by 24 h but then remained constant at around 6 logs at 48 and 72 h (Fig. 2A) (P > 0.05).

FIG. 2.

RV replication in MDM. (A) MDM were infected for 1 h with RV16 (MOI of 1), and RNA was extracted from cell lysates at 0, 2, 4, 8, 24, 48, and 72 h postinfection. RV RNA expression was quantified by using TaqMan, and data are presented as the number of copies per μg of total RNA. The results are expressed as means ± SEM (n = 3). Significance is at a P value of >0.05 by ANOVA. (B) Cells were infected as for panel A, supernatants were harvested, and the amount of infectious virus released into the supernatants was assessed by virus titration. The results are expressed as means ± SEM (n = 3). Significance is at a P value of >0.05 by ANOVA. (C) MDM, seeded in chambered slides, were infected with RV16 (MOI of 1) for 1 h. At 4, 8, and 24 h postinfection cells were fixed, permeabilized, and stained using rabbit anti-RV 3C protease serum and an appropriate secondary antibody (green). The slides were coverslipped using DAPI-containing mounting medium and analyzed using confocal microscopy. A representative image of three independent experiments is presented. Magnification, ×800. Green represents, RV 3C staining and overlay with DAPI staining for nuclei (blue). (D and E) THP-1-derived Mφ or MDM were infected for 1 h with RV16 (MOI of 1), supernatants were harvested at 24 h, and the amount of IFN-β (D) or IFN-α (E) released into supernatants was assessed by ELISA. The results are expressed as means ± SEM (n = 3). *, P < 0.05; **, P < 0.01.

Having observed limited virus release and high levels of viral RNA persisting to at least 72 h, we next investigated if synthesis of new viral proteins could be demonstrated by immunofluorescent staining for RV16 3C protease, using confocal microscopy. RV 3C protease was clearly detectable at 4 h (frequency of positive cells for RV16 3C protease, 19 ± 1.7%; n = 3) after RV16 infection of MDM and persisted until 24 h. No staining was detected when rabbit polyclonal immunoglobulin G (data not shown) was used and when cells were treated with medium (Fig. 2C).

To assess frequencies of virus-releasing cells in MDM, we next used the infectious center assay in a manner similar to that with THP-1 Mφ. Only 0.1 ± 0.01% of cells (n = 4) released sufficient infective virus to lead to plaque formation on HeLa cells (data not shown).

These data indicate that RV can replicate well in THP-1-derived Mφ, but replication is clearly limited in MDM. To attempt to explain these differences, we next investigated release of type I IFNs by THP-1-derived Mφ and MDM in response to RV infection. Both IFN-β (P < 0.05) and IFN-α (P < 0.01) were significantly increased at 24 h in the supernatants of RV16-infected MDM, while no IFN-β and only trace amounts of IFN-α were detected in RV16-infected THP-1-derived Mφ supernatants (Fig. 2D and E).

NF-κB is activated after RV infection of both THP-1-derived and monocyte-derived Mφ.

As NF-κB activation is implicated in activation of bronchial epithelial cells by RVs (32, 47, 67) and having demonstrated full RV replication in THP-1-derived Mφ and limited replication in MDM, we next examined if NF-κB was activated following infection of Mφ. We first assessed whether an NF-κB-dependent reporter gene was activated by RV infection. When THP-1-derived Mφ were transiently transfected with an NF-κB-Luc minimal promoter/reporter and analyzed for luciferase expression at 48 h after infection, there was a 2.12-fold increase in RV16-infected cells compared with medium-treated cells (Fig. 3A), (P < 0.05), confirming RV induction of NF-κB activation.

We next sought evidence of NF-κB activation by assessing NF-κB nuclear translocation by immunostaining for p65/NF-κB and analysis by confocal microscopy in infected and noninfected cells. Clear evidence of NF-κB nuclear translocation was observed in RV16-infected THP-1-derived Mφ at 3, 6 (Fig. 3B), and 24 h, while no nuclear staining was observed with medium-treated cells or UV-inactivated RV16-treated cells (data not shown). To determine whether NF-κB translocation occurred only in virus-infected cells or in noninfected cells (as a consequence of paracrine stimulation by cytokines released from infected cells), THP-1-derived Mφ were costained for RV 3C protease and NF-κB and analyzed using confocal microscopy. p65/NF-κB translocation was clearly colocalized with RV 3C protease staining (Fig. 3B), indicating that NF-κB activation occurred only in virus-infected cells.

Finally, to determine whether RV infection resulted in NF-κB activation in primary human MDM, similar experiments were carried out with MDM. Both NF-κB activation and RV 3C protease expression were observed in MDM at 2, 4, 8, and 24 h, with peak expression for NF-κB at 4 h (Fig. 3B). As with THP-1-derived Mφ, RV 3C protease and p65 translocation were almost exclusively colocalized, indicating that NF-κB activation occurred only in RV-infected MDM (Fig. 3B).

TNF-α is released from RV-infected Mφ in a time- and dose-dependent manner.

We next assessed TNF-α production by Mφ after RV16 infection. THP-1-derived Mφ and MDM released TNF-α after RV16 infection in a time-dependent manner. In the case of THP-1-derived Mφ, there was a steady increase of TNF-α in the supernatant starting at 4 h, reaching statistical significance at 24 h (P < 0.05) and continuing thereafter until 72 h (P < 0.01) (Fig. 4A). For MDM, significantly elevated concentrations of TNF-α in supernatants were observed from 8 to 48 h; however, the peak was reached earlier, at 24 h (P < 0.01) (Fig. 4B). TNF-α was also secreted in a dose-responsive manner after RV16 infection. The time point with the greatest levels of cytokine in the supernatant was chosen to carry out these experiments. Infection of either THP-1-derived Mφ or MDM with increasing concentrations of RV16 resulted in release of increasing amounts of TNF-α into the supernatants (Fig. 4C and D).

FIG. 4.

TNF-α is released from RV-infected Mφ in a time- and dose-dependent manner. (A and B) THP-1-derived Mφ (A) or MDM (B) were infected for 1 h with RV16 (MOI of 1), supernatants were harvested at 4, 8, 24, 48, and 72 h, and the amount of TNF-α released into the supernatants was assessed by ELISA. (C) THP-1-derived Mφ were infected for 1 h with RV16 at MOIs of 0.01, 0.1, and 1. Supernatants were harvested 72 h postinfection, and concentrations of TNF-α were assessed by ELISA. (D) MDM were infected for 1 h with RV16 at MOIs of 0.2, 0.5, 1, and 5. Supernatants were harvested 24 h postinfection, and concentrations of TNF-α were assessed by ELISA. The results are expressed as means ± SEM (n = 4). Significance (compared to medium control): *, P < 0.05; **, P < 0.01.

TNF-α secretion is serotype and receptor independent and is largely dependent on viral replication.

To investigate whether RV-induced TNF-α secretion is serotype or receptor restricted, we investigated two other serotypes in addition to RV16: RV9, another major group serotype, and RV1B, a minor group serotype. Because all three virus stocks were used as unpurified preparations, we also wished to confirm that the observed induction was virus specific and not the result of other soluble factors present in the inoculum used for infection. For this, the inoculum was molecular weight filtered using a 30-kDa filter to remove all virus particles but not small molecules, such as cytokines (47). We also wished to determine whether the production of TNF-α is virus replication dependent and therefore we used UV-inactivated RV. As for the dose-response studies, these experiments were carried out at time points of maximum production for each cell type. RV9 and RV1B were also capable of inducing the release of similar amounts of TNF-α from THP-1-derived Mφ and MDM (Fig. 5A and B), indicating that induction was not serotype or receptor restricted. When filtered RV16 inoculum was used, the production of TNF-α was abolished, confirming that induction was virus specific (Fig. 5A and B). When comparing UV-inactivated RV16 with live RV16, the levels of TNF-α measured in supernatants were reduced by 83% in the case of THP-1-derived Mφ and completely suppressed in the case of MDM (Fig. 5A and B), confirming that induction was largely or completely replication dependent, respectively. Similar results were obtained with filtered and UV-inactivated RV9 and RV1B (data not shown).

FIG. 5.

RV-induced TNF-α secretion from Mφ is serotype and receptor independent and is largely dependent on viral replication. THP-1-derived Mφ (A) or MDM (B) were infected for 1 hour with major group (receptor, ICAM-1) viruses RV16 and RV9 and the minor group (receptor, LDL receptor) virus RV1B (all at an MOI of 1), medium alone (m), UV-inactivated RV16 (MOI of 1; UV), or the same volume of RV16 inoculum from which virus had been removed by molecular weight filtration (filtered). Supernatants were harvested at 72 h for THP-1-derived Mφ or at 24 h for MDM, and the amount of TNF-α released was quantified by ELISA. The results are expressed as means ± SEM (n = 3). Significance (compared to medium control): *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Inhibition of NF-κB, but not p38 MAPK, inhibits TNF-α production from Mφ in response to RV16 infection.

The transcription factor NF-κB is required for inflammatory gene up-regulation after RV infection of epithelial cells (32, 47, 67) and is also important for induction of TNF-α in other systems (11, 53). Having demonstrated activation of NF-κB after RV infection of Mφ, we wished to asses its involvement in TNF-α production. For this we used two inhibitors, AS602868 (a specific inhibitor of the upstream kinase IKKβ) and CAPE (a specific inhibitor of the p65 subunit of the NF-κB/Rel complex). AS602868 inhibited RV16-induced TNF-α production in THP-1-derived Mφ in a dose-responsive manner, achieving statistically significant inhibition at concentrations of 0.5 and 1 μM (P < 0.05) and complete inhibition at 5 μM (P < 0.001) (Fig. 6A). CAPE also induced a dose-dependent inhibition of RV-induced TNF-α secretion in THP-1-derived Mφ but was less effective than AS602868. Significant inhibition was observed at 5 and 10 μg/ml (P < 0.05), and 60.26% inhibition was reached with a 25-μg/ml concentration (P < 0.001) (Fig. 6B), with higher doses having a toxic effect on cells. To confirm the above findings in primary MDM, we tested the effect of AS602868 on TNF-α release from RV-infected MDM. We investigated 5 μM, because this concentration proved to have the maximal effect without any cell toxicity in THP-1-derived Mφ. Again, an effective inhibition was observed, with TNF-α production after RV infection of MDM being reduced by 95% (P < 0.01) (Fig. 6C).

FIG. 6.

Requirement for NF-κB in TNF-α production from RV-infected Mφ. THP-1-derived Mφ (A, B, and D) or MDM (C) were pretreated for 1 h with inhibitors at the doses indicated, before infection with RV16 or with RSV (for p38 MAPK inhibition experiments only) at an MOI of 1. The same concentration of drug was added to the medium after infection. Supernatants were harvested at 72 h for THP-1-derived Mφ or at 24 h for MDM, and the amount of TNF-α released was quantified by ELISA. (A) AS602868, an inhibitor of IKKβ, in concentrations between 0.01 and 5 μM; (B) CAPE, an inhibitor of p65 translocation, in concentrations between 1.25 and 25 μg/ml; (C) AS602868 at a concentration of 5 μM; (D) SB203580, an inhibitor of p38 MAPK, at a concentration of 10 μM. The results are expressed as means ± SEM (n = 6 [A, B, and C] or n = 4 [D]). Significance (compared to medium control): *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Another molecule shown to be important for cytokine production by RV-infected epithelial cells and Mφ is p38 MAPK (20, 21). To assess its role in TNF-α secretion, the p38 MAPK inhibitor SB203580 was added to the medium at 10 μM before and after RV or RSV infection. Inhibition of p38 MAPK with SB203580 had no effect on TNF-α secretion from RV16-infected THP-1-derived Mφ (RV16, 212.4 ± 9.88; RV16 plus SB203580, 268.5 ± 40.10 ng/ml; P > 0.05), while RSV-induced TNF-α secretion was reduced by 63% (P < 0.05) (Fig. 6D).

DISCUSSION

We have investigated the ability of RV to infect human Mφ and the molecular mechanisms involved in TNF-α production from RV-infected Mφ. We have demonstrated that RV can replicate efficiently in THP-1-derived Mφ by showing increasing intracellular levels of viral RNA and viral replicative protein over time and release of infectious virus into supernatant. As has been observed with epithelial cells, 5% of cells released sufficient virus to cause plaques in HeLa cells (Fig. 1). Also, we showed evidence of replication in MDM by showing persistence of virus release and high levels of intracellular viral RNA up to 72 h and by demonstrating synthesis of new viral proteins. However, only 0.1% of cells released sufficient virus to cause plaques in HeLa cells, and virus titers released into supernatants did not increase significantly over time, indicating that replication was limited (Fig. 2). For both THP-1-derived Mφ and MDM, we also demonstrated activation of NF-κB following RV infection and colocalization of p65/NF-κB nuclear translocation with viral replication. Virus infection was accompanied by release of TNF-α in a time- and dose-dependent, serotype- and receptor-independent manner, largely or completely dependent upon viral replication. Finally, we established the requirement for NF-κB, but not p38 MAPK, in RV induction of TNF-α.

Mφ are a heterogeneous population. The differentiation process, which is under the control of local environments, will render Mφ with significant differences from one body site to another (26, 34, 64). Therefore, significant differences will be present not only between monocytes and Mφ but also between Mφ from different anatomical locations (2, 22, 34, 49). Although using THP-1-derived Mφ and MDM instead of primary airway Mφ could be considered a limitation of this study, primary alveolar Mφ are difficult to obtain in sufficient number and purity for extensive experiments, and it is not known how representative they would be of Mφ found in the trachea and bronchial tree, where rhinovirus infections are likely to occur. For these studies, we therefore chose two models extensively used as replacements for alveolar Mφ. The first used THP-1 cells, a monocytic cell line which can be differentiated to a macrophage phenotype by PMA treatment (6, 7, 58, 63). The second used Mφ derived from peripheral blood monocytes by treatment with granulocyte-macrophage colony-stimulating factor. The Mφ obtained by this protocol have a close resemblance to alveolar Mφ (2).

The first important finding of this study is that RV can replicate in Mφ. This confirms the preliminary results of a previous study which showed a low-grade release of rhinovirus from nondifferentiated THP-1 cells (29). The pattern of infection in THP-1-derived Mφ is similar with that found in respiratory epithelial cells, where only a small percentage of the cells release sufficient virus to produce plaques in an infectious center assay (39). Persistence of viral RNA over 72 h, presence of RV 3C staining in 19% of cells, virus release by 0.1% of infected cells, and replication-dependent TNF-α release in MDM suggest RV can replicate, to a limited degree, in this system as well. The discrepancy between the number of RV 3C protease-positive cells and the number of virus-releasing cells suggests that virus replication is abortive in the majority of MDM, with only a minority of cells supporting a complete cycle of virus replication. Also, the release of type I IFNs in the supernatants of MDM is indirect evidence of viral replication, as these cytokines are only induced during viral replication (27). Type I IFN production could also explain the differences seen in viral replication between THP-1-derived Mφ and MDM, as type I IFN production was almost absent in THP-1-derived Mφ. We have recently described profoundly deficient IFN-β production in response to rhinovirus infection of primary bronchial epithelial cells from asthmatic subjects (65). It is interesting that if a similar deficiency were observed in Mφ from asthmatic subjects, then THP-1-derived Mφ could be a good model for Mφ from asthmatic subjects. Even if this is not the case, we have provided evidence of limited rhinovirus replication in MDM sufficient to induce NF-κB translocation and TNF-α production. These data suggest rhinovirus infection of Mφ in vivo may be an important source of proinflammatory mediators in the context of exacerbations of respiratory disease. The outcome of RV replication in primary macrophages is similar to the results obtained for RSV and influenza virus. Both viruses can infect airway Mφ, but the release of virions is absent or very limited (10, 51).

The differences seen in permissiveness to RV replication and in type I IFN production between THP-1 Mφ and MDM were not found when NF-κB translocation or TNF-α production was studied; indeed, TNF-α production occurred at levels approximately 100-fold higher in THP-1-derived Mφ compared with MDM. These differences imply that the signaling pathways leading to NF-κB activation and TNF-α production are not affected in THP-1-derived Mφ, while specific pathways involved in type I IFN production and antiviral responses are deficient in THP-1-derived Mφ compared with MDM.

NF-κB is important for cytokine production from RV-infected epithelial cells (32, 47, 67). During viral infections, NF-κB activation can be the result of several different events, including recognition of double-stranded or single-stranded RNA, viral enzymes, stress induced by viral entry and/or replication (38). In the case of RV infection of epithelial cells, oxidative stress has been shown to induce early activation (48). Recognition of double-stranded RNA by protein kinase R has been proposed as another possible mechanism leading to cytokine production, but there is no clear evidence at this time for involvement protein kinase R in cytokine induction after RV infection (15). Also, RV 3C protease was reported to be important in NF-κB activation and cytokine production from RV-infected epithelial cells (13, 66). In monocytes/Mφ, transient degradation of IκB was shown to take place and suggested to be important for MCP-1 production (21). In this study we showed, using a minimal promoter, that NF-κB activation occurs during RV infection of THP-1-derived Mφ. Because only a minority of cells proved to support RV replication, we assessed whether NF-κB activation occurred in virus-infected cells or in neighboring cells through paracrine effects of an unknown mediator. RV 3C protease is a nonstructural protein produced either transiently by primary translation of input viral RNA or, to a greater degree and at later time points, by translation from viral RNA produced as a result of replication. Its expression at the levels and time points observed therefore constitutes evidence of replication. When RV 3C protease was detected in THP-1-derived Mφ or MDM, it colocalized with nuclear translocation of p65 (Fig. 3). This is thus the first study to show a direct colocalization between RV replication and p65 translocation, indicating that NF-κB activation at the time points studied occurs principally in virus-infected and not neighboring cells through paracrine mechanisms.

One of the most important and well-studied NF-κB-dependent genes is TNF-α, which has a very-well-established role in inflammation (11, 53). In the airway there are two major potential sources: mast cells as a source of preformed TNF-α and Mφ as a source of newly formed TNF-α (60). Respiratory viruses are known to induce TNF-α release from Mφ (40, 43). Because only very low amounts, if any, are released from epithelial cells after RV infection in vitro, it is logical to investigate whether Mφ could be a source during rhinovirus infection of the lower airways (59). Here we show that infection of Mφ leads to sustained TNF-α production. This has been suggested in a previous study; however, in that study TNF-α production was reported not to be associated with replication (14). Using UV-inactivated RVs we showed that that TNF-α secretion is almost completely dependent on viral replication. These findings are in agreement with previous studies which showed replication to be important for efficient IL-8 production from RV-infected monocytes (29).

Finally, we showed a clear requirement for NF-κB activation for TNF-α to be released from RV-infected Mφ using two specific chemical inhibitors of NF-κB. NF-κB has been reported to be required for induction of several inflammatory mediators induced in response to RV infection of epithelial cells, suggesting inhibition of NF-κB might have therapeutic potential in treatment of virus-induced exacerbations of asthma and COPD (32, 46, 47, 55, 67). The present data showing that NF-κB is also required for RV induction of TNF-α in Mφ reinforce the evidence that inhibition of NF-κB could be a rewarding approach. At the highest specific doses used, IKKβ inhibition was the most efficient approach, inhibiting TNF-α release by >95% in both model systems used (Fig. 6A and C), while inhibition of p65 translocation only partially blocked release (by ∼60%) (Fig. 6B). This may be because the IKKβ inhibitor is either more specific or more potent that CAPE. An alternative explanation is that IKKβ inhibition would be expected to block activation of all forms of NF-κB, while p65 inhibition would block only those species of NF-κB that included this protein subunit. These data suggest p50 homodimers or other Rel family proteins excluding p65 may contribute to RV induction of TNF-α and that therapeutic approaches based on NF-κB inhibition may be most successful if they target upstream activation events that would inhibit all family members, rather than just p65.

Previous studies have implicated p38 MAPK in cytokine/chemokine production from RV-infected epithelial cells and monocyte/Mφ (20, 21). We therefore also investigated its role in TNF-α production in response to RV infection of Mφ. p38 MAPK inhibition had no effect of TNF-α production, suggesting that although this pathway has been reported to be activated during RV infection of Mφ and involved in induction of MCP-1 (21), it does not appear to be required for induction of TNF-α. However, the same concentration of p38 MAPK inhibitor efficiently inhibited RSV-induced TNF-α production. Further studies will be required to increase our understanding of signaling pathways involved in induction of inflammatory cytokine production in response to RV infection of Mφ.

TNF-α has strong proinflammatory activities, and its inhibition has therapeutic effects in a number of chronic inflammatory diseases (35, 50, 57, 61). Its role in allergic inflammation has been recognized, and recent studies implicate it in severe stable asthma and COPD (25, 31). There is less information available relating to acute exacerbations, though preliminary studies suggest it may be important (1, 9). Further studies are required to investigate its role in virus-induced asthma and COPD exacerbations to help determine whether pharmacological inhibition or use of blocking antibodies/soluble receptor could be considered as a possible therapeutic intervention in asthma and COPD exacerbations. The present data indicating that TNF-α release by Mφ during RV infection is strongly induced to high-nanogram levels suggest it is likely to play an important role in the exacerbation process and, therefore, inhibition is likely to be rewarding.

In conclusion, we have demonstrated that RV replication occurs in Mφ, is accompanied by NF-κB activation in virus-infected cells, and strongly induces TNF-α secretion. We also showed that TNF-α secretion is mediated by NF-κB but not p38 MAPK. These studies suggest inhibition of both TNF-α and NF-κB may be useful in treatment of exacerbations of asthma and COPD and indicate that further study on the role of RV-induced macrophage activation in the pathogenesis of these conditions is required.