|Home | About | Journals | Submit | Contact Us | Français|
Rhinoviruses (RVs) are picornaviruses that are causative agents of the majority of upper respiratory tract infections, or “common colds,” in humans. RVs infect both the upper and lower respiratory tract, and in addition to the common cold may also cause pneumonia, complications in patients with chronic lung diseases such as cystic fibrosis, and asthma exacerbations. Convenient animal models are not available to study the pathogenesis of rhinovirus-induced illness. Rhinovirus RV1A replicates poorly in mouse cells; variants with improved replication were selected by serial passage through mouse embryonic fibroblasts and mouse lung epithelial cells. Adaptation for improved growth in mouse cells was mediated by amino acid changes in the RV1a non-structural protein 3A. Mouse cell-adapted RV1A was capable of productively infecting mice in which the airway was subjected to chemical permeabilization. A mouse model for RV infection will permit studies of RV pathogenesis and may identify targets for therapeutic intervention.
Rhinovirus infections cause the majority of common colds, as well as serious acute exacerbations of asthma, chronic lung diseases such as chronic obstructive pulmonary disease (COPD) (Varkey and Varkey, 2008), and pneumonia in infant, elderly, and immunosuppressed patients. Common colds are major economic burdens, resulting in massive healthcare costs and decreased working productivity. Asthma, COPD, and pneumonia are all serious conditions that can lead to severe morbidity resulting in hospitalization, and which are often fatal. Because of the diversity of rhinovirus serotypes, there are no vaccines to prevent rhinovirus infection, and no effective treatment.
The development of novel therapies that could reduce the incidence or severity of rhinovirus infections is hindered by the lack of a practical animal model. Studies performed in human volunteers and non-human primates are limited in scope and often prohibitively expensive (Xatzipsalti and Papadopoulos, 2007). Small-animal models developed for rhinovirus infection are also limited, as there is no current model that supports robust replication of infectious virus comparable to a clinically relevant human infection. These models require that mice are either immunosuppressed (Yin and Lomax, 1986) or are accompanied by modest production of infectious virus in the respiratory tract (Bartlett et al., 2008).
A major obstacle to development of a small animal model that meaningfully reproduces many of the clinical features of human rhinovirus infection is that mouse cells are not permissive for replication of most rhinovirus types. The block to replication appears to be at the level of RNA replication (Yin and Lomax, 1983). Furthermore, the ~90 rhinovirus types that bind human intercellular adhesion molecule 1 (ICAM-1) cannot enter mouse cells because murine ICAM-1 does not bind the viral capsid (Bella and Rossmann, 1999; Bella and Rossmann, 2000).
Several rhinovirus types have been selected for growth in mouse cells (Harris and Racaniello, 2003; Harris and Racaniello, 2005; Yin and Lomax, 1983; Yin and Lomax, 1986). Repeated passages of RV2, RV16, and RV39 between human cells and mouse fibroblasts lead to the isolation of host range variants capable of robust replication in mouse cells. The genomes of these viruses have mutations leading to amino acid changes in the 2BC and 3A region of the viral genome. By passing RV1a in mouse lung epithelial cells, we isolated viruses with amino acid changes in the 3A protein that can multiply in mouse cells and in the respiratory epithelium of mice.
RV1A was subjected to alternate passages in cultured in MDA5−/− murine embryonic fibroblasts and HeLa cells, or by serial passage through MEFs with amplification in HeLa cells as needed. MEFs deficient in MDA5, the cytoplasmic sensor of picornavirus infection (Kato et al., 2006; Wang et al., 2010) were used for viral passage to diminish the interferon response and possibly enhance HRV replication. MDA5 binds double-stranded viral RNA and induces innate antiviral immune responses, and has also been shown to mediate airway epithelial responses in response to RV infection (Wang et al., 2009). Following 35 cell passages in these cells, virus was passaged 27 times in LA4 mouse lung epithelial cells or for 17 alternating passages in LA4 and HeLa cells, yielding isolates designated RV1A/M2M and RV1A/M2H, respectively.
Mouse-adaptation of selected viruses was assessed by one-step growth analysis. In HeLa cells, RV1A/M2M and RV1A/M2H replicated at a slightly reduced rate compared to wild-type RV1A at both MOI=1 (Figure 1a) and MOI=0.1 (Figure 1c). In LA4 cells, RV1A/M2M and RV1A/M2H replicated at a faster rate and to higher titers than wild-type RV1A at both MOI=1 (Figure 1b) and MOI=0.1 (Figure 1d), indicating that both variants are adapted for more efficient growth in mouse airway epithelial cells.
Cytopathic effect over the course of virus replication was determined in HeLa and LA4 cells by measuring cell viability by trypan blue exclusion. Cells infected with all viruses displayed cytopathic effect compared to mock-infected HeLa cells (Figure 2a). The viability of HeLa cells infected with wild-type RV1A steadily decreased; a slightly slower decrease in cell viability was observed in cells infected with RV1A/M2M and RV1A/M2H. Very little cytopathic effect was observed in LA4 cells infected with wild-type RV1A compared to RV1A/M2M and RV1A/M2H, both of which produced steady decreases in cell viability (Figure 2b). The increased replication of mouse-adapted RV1A produce higher levels of cell killing in mouse cells compared to wild-type virus, perhaps because the viruses replicate to higher levels.
To identify the changes that accompany enhanced replication in mouse cells, complete nucleotide sequences of the genomes of wild-type RV1A, RV1A/M2M, and RV1A/M2H were determined. The genomes of both RV1A/M2M and RV1A/M2H contain multiple changes from the wild-type sequence at both the nucleotide and amino acid level. These changes are found in both the 5′ untranslated region (UTR) and the open reading frame (ORF) encoding the viral polyprotein (Figure 3). The changes within the coding region are located in the P1 structural and P3 non-structural regions of the genome, specifically in VP1, VP3, and 3A. Both RV1A/M2M and RV1A/M2H shared several common changes but contain unique silent mutations, indicating that each virus was isolated independently.
Several individual molecular clones of wild-type RV1A contained the N85D change in VP1 observed in the RV1A/M2M and the H5Y change in VP3 observed in both RV1A/M2M and RV1A/M2H. Because these amino acids are also present in database sequences of RV1A (Palmenberg et al., 2009), they may represent normal fluctuations within the viral population. To evaluate the effect that these amino acid changes have on the mouse cell-adapted phenotype, they were introduced into wild type HRV1a. No differences were observed between viruses containing VP1 N85D/VP3 H5Y and wild-type virus in LA4 cells (Figure 4), indicating that changes in the structural proteins have no effect on RV1A adaptation to efficient growth in mouse cells.
Analysis of previously selected mouse cell-adapted RV2, RV16, and RV39 indicates that the P2/P3 nonstructural proteins mediate changes in host range (Harris and Racaniello, 2003; Harris and Racaniello, 2005; Lomax and Yin, 1989; Yin and Lomax, 1983; Yin and Lomax, 1986). Specifically, changes in protein 3A enhance replication of RV39 in mouse fibroblasts (Harris and Racaniello, 2005). These changes may promote interactions with host proteins that are essential for viral replication. If this hypothesis is correct, then it should be possible to render mouse cells permissive for poliovirus replication by synthesis of P2/P3 proteins from picornaviruses that are competent for growth in mouse cells. Mouse cells are permissive for the replication of wild type poliovirus type 1/Mahoney (P1/M), indicating that the poliovirus P2/P3 proteins interact with the host machinery in a manner that promotes viral replication. To determine whether poliovirus 2BC or 3A proteins can enhance growth of RV1A, stable LA4 cells were produced by DNA-mediated transformation with plasmids encoding the poliovirus proteins. We determined protein expression by immunoblotting (Figure S1). RV1a replication was enhanced in cells that synthesize P1/M proteins 2BC and 3A, either individually or together (Figure 5).
RV1A/M2M was passaged seven times by intranasal inoculation in BALB/c mice with intermittent amplification in HeLa cells, producing virus RV1A/M2M7. When this virus was inoculated intranasally in mice, virus titers in respiratory tissues declined 1000-fold within 24-48 hours (data not shown). We considered that the respiratory mucosa might act as a barrier to infection of mice with RV1A/M2M. In support of this hypothesis, it has been shown that disruption of the intestinal mucosal barrier is required to render poliovirus receptor transgenic mice susceptible to poliovirus infection by oral inoculation (Kuss, Etheredge, and Pfeiffer, 2008). Hypochlorous acid and hydrogen peroxide have been shown to increase respiratory permeability in a variety of rodent models in vivo (Greiff et al., 1999; Guo et al., 1996; Guo, Schneider, and Wangensteen, 1995). Therefore the effects of these treatments on susceptibility of mice were determined. Mice were instilled intranasally with 30 μM hypochlorous acid or Ringer’s solution several hours prior to inoculation with RV1A/M2M7. Virus titers in the respiratory tissues declined nearly 1,000-fold with 24 hours in mice treated with Ringer’s solution (Figure 6). In contrast, virus titers declined only slightly in respiratory tissues of mice treated with 30 μM hypochlorous acid. These observations suggest that permeabilization of the respiratory mucosa allows RV1A/M2M7 to penetrate to target cells and replicate to levels sufficient to maintain virus titer. Pre-treatment of mice with 20 μmol hydrogen peroxide lead to a 100-fold drop in virus titer 24 hours following inoculation. Hydrogen peroxide might be less efficient at inducing respiratory permeability, or could reduce infectivity of the inoculum.
We determined the replication kinetics of RV1a in mice pre-treated with hypochlorous acid. After intranasal inoculation with 8 × 106 PFU of RV1A/M2M7, viral titers in the nose, trachea, and lungs were determined by plaque assay. In mice infected with wild-type RV1A, viral titers steadily decreased, and were at the limit of detection of the plaque assay after two days. In contrast, in mice infected with RV1A/M2M7, there was no decrease in virus titer by 24 h post-infection, and a ten-fold decrease 48 hours post-infection. These observations indicate that RV1A/M2M7 replicates in the murine respiratory tract, while the wild-type RV1A does not.
The replication cycle of RVs is poorly understood on a molecular level, in that specific host cell proteins required for permissive infection are largely unknown. Most studies that focus on how picornavirus proteins interact with the host cell to facilitate replication of viral RNA and production of progeny virus have been carried out on poliovirus-infected cells. As poliovirus is capable of robust replication in mouse cells, such studies have not revealed species-specific host cell requirements for viral growth.
Passage of RV1a in murine embryonic fibroblasts yielded viruses with amino acid changes in VP1, VP3, and 3A. Introduction of amino acid changes into the capsid region of wild-type RV1A did not lead to enhanced replication in mouse cells. It seems likely that the amino acid change in 3A protein is responsible for the enhanced replication of RV1a in mouse embryonic fibroblasts. It is curious that 2BC protein from P1/M alone was capable of enhancing wild-type RV1A replication in mouse cells, despite the fact that changes in 2BC were not identified in any of the mouse-adapted RV1A variants isolated in this study. This observation suggests that multiple viral proteins are involved in the function that is deficient in mouse cells.
Previous studies have demonstrated that selection of RVs with improved replication in mouse cells is accompanied by changes in the viral 2BC and 3A proteins (Lomax and Yin, 1989; Yin and Lomax, 1983; Yin and Lomax, 1986). The mechanism by which changes in these proteins enhance the permissivity of mouse cells is presently unknown. The results of yeast two-hybrid studies have shown that the RV 3A protein interacts with proteins involved in vesicular transport and lipid biosynthesis (Harris and Racaniello, 2003; Harris and Racaniello, 2005). These findings are intriguing given the essential role of membrane structures in picornavirus RNA replication (Quiner and Jackson, 2010). Furthermore, poliovirus has been shown to subvert autophagic processes to generate double-membraned vesicular structures needed for assembly of the viral replication complex, with notable co-localization of poliovirus 3A and the autophagosomal marker LC3 (Taylor and Kirkegaard, 2007; Taylor and Kirkegaard, 2008). This observation is also consistent with reports that GTPases regulating vesicular transport and membrane lipid composition are required for assembly of specialized organelles supporting enterovirus replication (Belov, Fogg, and Ehrenfeld, 2005; Belov et al., 2007b; Hsu et al.; Lanke et al., 2009).
The mechanism by which protein 3A facilitates RNA replication-associated membrane remodeling in rhinovirus-infected cells is unclear. Although induction of autophagy does not enhance the growth of HRV2 and does not appear to be required for replication (Brabec-Zaruba et al., 2007), the importance of 3A in membrane remodeling has been conserved among many other members of genus Enterovirus. Furthermore, amino acid changes in RV1A structural proteins have no effect on virus replication, suggesting that mouse adaptation is mediated post-entry. When synthesized alone, poliovirus 3A causes substantial swelling of the ER membranes, but does not induce formation of membranous replication structures (Doedens and Kirkegaard, 1995). During infection, 3A directly binds guanine nucleotide exchange factors (GEFs) mediating activation and recruitment of ADP-ribosylation factor (Arf) family GTPases that regulate anterograde ER-to-Golgi secretory trafficking pathways into the membranes associated with RNA replication complexes (Belov et al., 2007a; Belov and Ehrenfeld, 2007; Belov et al., 2008; Belov, Fogg, and Ehrenfeld, 2005; Doedens and Kirkegaard, 1995). Amino acid changes in RV1A/M2M 3A may enhance the ability of RV1A to interact with murine host proteins required for these processes in the course of the viral life cycle.
It is likely that additional passaging of RV1A/M2M in vivo will select for variants capable of more robust growth in the mouse respiratory tract. However, development of more virulent mouse-adapted RV1A variants may not eliminate the requirement for permeabilization of the respiratory tract prior to experimental infection in this model. The necessity for permeabilization may be a consequence of the remarkable ability of the respiratory mucosa to prevent pathogens from accessing the susceptible epithelial cells lining the airway. This requirement may mimic the normal physiological factors predisposing human patients to RV infection. It is well-documented that patients who smoke or are exposed to cigarette smoke (Gryczynska, Kobos, and Zakrzewska, 1999; Jedrychowski and Flak, 1997), are afflicted with cystic fibrosis or similar pre-existing condition (McManus et al., 2008), or who have suffered another type of pulmonary injury are more susceptible to upper respiratory tract infections. This enhanced susceptibility may be due to either direct mechanical injury creating entry routes for virus, or by increased mucosal permeability and enhanced RV receptor expression induced by inflammatory mediators up-regulated in response to pulmonary damage (Altman et al., 1993; Bianco et al., 2000; Subauste and Proud, 2001; Whiteman et al., 2003). Furthermore, infants who have not developed a fully differentiated respiratory mucosa are also prone to contracting upper respiratory infections. These data suggest that a loss of host mucosal barrier integrity is a critical early step in RV pathogenesis, and may determine host susceptibility. The fact that basal cells, located deeper within the respiratory epithelium are more susceptible to rhinovirus infection (Jakiela et al., 2008) is consistent with these observations
Further studies should be conducted to characterize and improve this mouse model. Studies assessing cellular responses and cytokine expression to determine the breadth of the murine immune response are underway, as are histopathology studies to determine the level of injury and inflammation present in the mouse respiratory tract. This model represents an advance in the ability to study both the pathogenesis of RV infection and the development and pathogenesis of allergic sensitization and asthma in vivo.
All cells were propagated in cell culture medium and supplements from Invitrogen (Carlsbad, CA) unless otherwise specified. HeLa S3 cells were propagated in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% bovine calf serum (Hyclone, Logan, UT), 40 mM MgCl2, 20 mM HEPES buffer, and 1% penicillin/streptomycin. LA4 mouse lung epithelial cells were propagated in F-12/Ham’s Nutrient Solution supplemented with 15% fetal bovine serum, 0.01% sodium bicarbonate, 40 mM MgCl2, and 1% penicillin/streptomycin. MDA5−/− murine embryo fibroblasts (MEFs) were generously provided by Marco Colonna (Washington University, St. Louis, MO) and maintained in DMEM supplemented with 10% fetal bovine serum, 1% nonessential amino acids, and 1% penicillin/streptomycin.
LA4 cells stably expressing 2BC and/or 3A proteins from poliovirus type 1/Mahoney strain were generated by introducing plasmid DNA encoding the viral protein (pcDNA4-TO/FLAG-2BC and/or pcDNA4-TO/FLAG-3A, kindly provided by Juliet Morrison) or with empty pcDNA4-TO vector using Lipofectamine 2000 (Invitrogen). Prior to DNA-mediated transformation, LA4 cells were cultured with varying concentrations of zeocin (Invitrogen) ranging from 100 μg/mL to 1 mg/mL for 10 days to determine the optimal concentration for selection. Fresh LA4 medium containing zeocin was added every 48 hours. The optimal concentration was defined as the lowest concentration needed to kill 100% of the cells in the shortest amount of time. Ultimately, 1 mg/mL zeocin was capable of killing over 90% of the cells in 8 days, with total cell death occurring at this concentration at 9 or 10 days following application of zeocin. LA4 medium containing 1 mg/mL zeocin was applied to cells 24 hours post-DNA transformation and maintained for 14-17 days until colonies of thriving cells were observed. Production of FLAG-2BC and/or FLAG-3A by these cell lines was determined by SDS-PAGE and subsequent immunoblotting with M2 anti-FLAG antibody (Sigma-Aldrich, St. Louis, MO).
Human rhinovirus 1A (RV1A) was obtained from the American Type Culture collection and stocks were prepared in HeLa S3 cells. Briefly, virus diluted in PBS containing 0.5% bovine serum albumin (VBS) was adsorbed to confluent HeLa cell monolayers for 1 hour at 33° C. HeLa cell culture medium was added and cells were cultured for 24-72 h or until 90% cytopathic effect was observed by light microscopy. Virus titer was determined on confluent HeLa cell monolayers by standard plaque assay in RV1A plaquing medium composed of DMEM supplemented with 2% bovine calf serum, 40 mM MgCl2, 0.01% sodium bicarbonate, 1% penicillin/streptomycin, and 1% low-gelling type VII agarose (Sigma).
Virus stocks of RV1A were centrifuged in a SW41 rotor (Beckman) for two hours at 35,000 rpm with a 2 mL cushion of 25% sucrose in PBS. The pellet was resuspended directly in 1 mL TRIzol reagent (Invitrogen), and RNA was prepared using the manufacturer’s protocol. The RNA pellet was briefly air-dried and resuspended in 50 μL DEPC-treated dH2O. First-strand cDNA was produced from viral RNA by reverse transcription with Superscript III (Invitrogen) at 55° C for 1 hour using oligo dT (Fermentas, Glen Burnie, MD), followed by a 30 minute incubation with RNase H (Invitrogen).
Two amplified products were produced by PCR using Herculase II fusion polymerase (Stratagene), encompassing the entire genome of RV1A. The first PCR product comprising nucleotides 1-2655 was amplified using the following primers designed to add a XmaI site to the 5′ terminus and a ClaI site to the 3′ terminus:
Forward (Xma-RV1A-T7-5′UTR): 5′-CGCCCCGGGGGATCCCGCGGAAATTAATACGACTCACTATAGGTTAAAACTGGGTGTGGGTTGTTCCCACTCACACCACCCAATGGGTGTTGTACTCTGTTATTCCGGTAACTTTG-3′
Reverse (RV1A-ClaIrev): 5′-GCGATCGATGGCATATAATTAGTCACATGAC-3′
The second PCR product comprising nucleotides 2632-7100 was amplified using the following primers designed to add a ClaI site to the 5′ terminus and a XhoI site to the 3′ terminus:
Forward (RV1A-ClaIfwd): 5′-GCGATCGATGGAAAATCACACTACAGGAAATGG-3′
Reverse (3′-RV1AXhoI): 5′-CGCCCTCGAGGTTTTTTTTTTTTTTTTATAGAATTAAAGAATCATTCATTC-3′
Each full-length PCR product was cleaved with XmaI and ClaI or ClaI and XhoI and cloned into pACYC177 between the XmaI and ClaI sites (pACYC177-5′RV1A) and the ClaI and XhoI sites (pACYC177-3′RV1A), respectively. Both subclones were cleaved with XmaI and SacI, and the 2500 bp fragment from pACYC177-5′RV1A and the 7000 bp fragment from pACYC177-3′RV1A were gel purified using the QIAGEN Gel Purification Kit (QIAGEN) per manufacturer’s instructions. The 7000 bp fragment from pACYC177-3′RV1A was treated for one hour with Antarctic phosphatase (New England Biolabs, Ipswich, MA) at 37° C. Both fragments were ligated using T4 DNA ligase and propagated in XL-10 Gold Ultracompetent E. coli (Stratagene/Agilent, Santa Clara, CA). The complete nucleotide sequence was determined by sequencing using primers designed to fully sequence both strands of infectious clones and PCR products. The nucleotide sequences were aligned by Geneious (Auckland, NZ) sequence analysis software, and a consensus was determined and used as the reference for wild-type RV1A.
Four-to-eight week old BALB/cJ mice (Jackson Laboratories) were bred and maintained in a barrier facility according to institutional animal care protocols. Mice were housed in microisolator cages and fed sterilized food and water ad libitum. Infection experiments were independently performed at least 3 times with 3-5 mice/condition/timepoint/experiment. Mice were anesthetized with isoflurane gas and intranasally instilled with 50 μL 30 mM hypochlorous acid to permeabilize the respiratory tract. Mice were allowed to recover 6-18 hours prior to isofluorane anesthesia and intranasal instillation of 8 × 106 plaque forming units of either RV1A or RV1A/M2M in a 0.1 mL inoculum. At 0 and 24 (Figure 6), or 0, 24, and 48 hours post-infection (Figure 7), mice were humanely sacrificed via asphyxiation with carbon dioxide gas. Nose, trachea, and lungs were removed on ice, weighed, homogenized in 1 mL phosphate-buffered saline using a Brinkmann tissue homogenizer, and subjected to 3 freeze/thaw cycles. Homogenates were centrifuged to pellet solid tissue debris, and supernatants were filtered through 0.2 μm syringe filters (Nalgene). Virus titers from clarified organ homogenates were determined by plaque assay as described above.
Supplementary figure S1. Immunoblot analysis with anti-FLAG antisera of extracts from LA4 cells expressing FLAG-2BC, FLAG-3A, and FLAG-2BC/FLAG-3A.
This work was supported in part by Public Health Service Grants AI50754 and T32AI07161 from the National Institute of Allergy and Infectious Diseases.
We thank Marco Colonna for mda-5−/− murine embryonic fibroblasts, and Juliet Morrison for providing FLAG-2BC and FLAG-3A expression constructs.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.