Construction of a recombinant rhinovirus accommodating fluorescent marker expression

Background Rhinovirus (RV) causes the common cold and asthma exacerbations. The RV genome is a 7.3 kb single‐strand positive‐sense RNA. Objective Using minor group RV1A as a backbone, we sought to design and generate a recombinant RV1A accommodating fluorescent marker expression, thereby allowing tracking of viral infection. Method Recombinant RV1A infectious cDNA clones harboring the coding sequence of green fluorescent protein (GFP), Renilla luciferase, or iLOV (for light, oxygen, or voltage sensing) were engineered and constructed. RV‐infected cells were determined by flow cytometry, immunohistochemistry, and immunofluorescence microscopy. Results RV1A‐GFP showed a cytopathic effect in HeLa cells but failed to express GFP or Renilla luciferase due to deletion. The smaller fluorescent protein construct, RV1A‐iLOV, was stably expressed in infected cells. RV1A‐iLOV expression was used to examine the antiviral effect of bafilomycin in HeLa cells. Compared to parental virus, RV1A‐iLOV infection of BALB/c mice yielded a similar viral load and level of cytokine mRNA expression. However, imaging of fixed lung tissue failed to reveal a fluorescent signal, likely due to the oxidation and bleaching of iLOV‐bound flavin mononucleotide. We therefore employed an anti‐iLOV antibody for immunohistochemical and immunofluorescence imaging. The iLOV signal was identified in airway epithelial cells and CD45+ CD11b+ lung macrophages. Conclusions These results suggest that RV1A‐iLOV is a useful molecular tool for studying RV pathogenesis. The construction strategy for RV1A‐iLOV could be applied to other RV serotypes. However, the detection of iLOV‐expressing RV in fixed tissue required the use of an anti‐iLOV antibody, limiting the value of this construct.

approximately 100 different species A and B strains which were subsequently serotyped. 7,8 More recently, a diverse group of previously unrecognized human viruses from species C was found to be common causes of respiratory illness. 4 To understand RV pathogenesis, human and animal models have been developed. Human studies have employed experimental infection with RV-A16. 9,10 Mouse studies have used RV-A1B wild-type mice 11 or RV-A16 in mice that are transgenic for human intercellular adhesion molecule-1. 12 These models have been particularly useful in studying RV-induced exacerbations of allergic airways disease. To detect RV in the tissues, investigators have employed the monoclonal antibody R16-7. 9,13,14 This antibody, originally developed by Wai-Ming Lee at the University of Wisconsin, binds to the VP2 capsid protein of the closely related RV-A16 and RV-A1 strains, 6 but not to RV-A2, RV-B14, or RV-A49. 15 Because the presence of more than 100 different RV serotypes makes it infeasible to develop a cross-reactive antibody for RV, we sought to develop a recombinant virus with a fluorescent marker that could be used for tracking of RV infection in vivo.
Similar to other picornaviruses, RV is icosahedral, nonenveloped particle which is composed of 60 copies each of four capsid proteins, VP1, VP2, VP3, and the small myristoylated VP4. 16,17 The capsid encases a positive-sense single-stranded RNA (ssRNA) genome of approximately 7200 nucleotides. 18 Following virus entry and genome release into the host cell cytoplasm, the RV ssRNA is translated into a single polyprotein that undergoes proteolytic cleavage by viral proteases 2A pro and 3C pro, 8,19,20 with the exception of the autocatalytic cleavage of precursor VP0 into VP2 and VP4 in the presence of viral RNA during the assembly process. 21 RV infectious cDNA clones have been constructed and used as a molecular tool to study RV viral protein function and mutationphenotype association, as well as a vaccine vector for foreign gene expression. 17,[22][23][24][25] In the current study, we engineered a recombinant RV1A (RV1A-iLOV) with insertion of the coding sequence for iLOV (for light, oxygen, or voltage sensing), a small-size fluorescent marker. 26 RV1A-iLOV is viable, and its expressed iLOV protein is trackable both in vitro and in vivo, suggesting that RV1A-iLOV may be a useful tool in the study of RV pathogenesis. However, the detection of iLOV-expressing RV in fixed tissue required the use of an anti-iLOV antibody, limiting the value of this construct.

| Design and construction of recombinant RV1A-iLOV cDNA clone
The RV infectious cDNA clone encoding replication-competent RV-1A, pMJ3-RV1A, was kindly provided by W. T. Jackson, University of Maryland 27 and served as a backbone for either GFP-, RL-, or iLOVexpressing viruses. GFP, RL, and iLOV open reading frames (ORFs) were designed to be flanked by the edited nucleotide sequences encoding the viral 2A pro cleavage site with silent mutations introduced as described previously 24 ( Figure 1A, Table S2). Respective GFP, RL, and iLOV inserts were PCR amplified from existing clones using the primers listed in Table S1. The PCR products, which contained Apa I restriction enzyme cleavage sites on the 5′ and 3′ ends, were digested with Apa I, ligated to pMJ3-RV1A, and transformed in E. coli (DH5α, Thermo Fisher Scientific). The resultant clones were sequenced to confirm the correct orientation of the inserts.

| Generation of recombinant RV
Infectious cDNA clones encoding RV1A, RV1A-GFP, RV1A-RL, and RV1A-iLOV were linearized by Mlu I restriction enzyme digestion. To produce replication-competent virus, full-length viral RNA transcripts were generated using the MEGAscript T7 Transcription Kit (Thermo Fisher Scientific) and transfected into H1-HeLa cells using Lipofectamine MessengerMAX (Thermo Fisher Scientific).

| Quantitative real-time PCR of lung cytokines
Lung RNA was extracted with TRIzol Reagent (Thermo Fisher Scientific) combined with on-column digestion of genomic DNA (QIAGEN, Valencia, CA). cDNA was synthesized from 1 μg of RNA and subjected to quantitative real-time PCR using specific mRNA primers encoding for IL-1β, IFNβ, IFNγ, CXCL1, CXCL2, CXCL10, CCL2, CCL5, and IL-10 (Table S2). For each sample, the level of gene expression was normalized to its own GAPDH mRNA.

| Data analysis
Data are represented as mean ± SE. Statistical significance was assessed using an unpaired t test or one-way ANOVA, as appropriate. Group differences were pinpointed by a Tukey's multiplecomparison test.

| Incompatibility of RV1A genome with GFP insert
GFP with flanking 2A pro cleavage sites was designed to be inserted between the RV genomic sequences encoding the VP1 and 2A proteins ( Figure 1A). To stabilize the genome, silent mutations were introduced into the coding sequences of both the flanking 2A pro cleavage sites (Table S1), as described previously. 24 To generate viral stocks, genomic RNA transcripts made from the RV1A-GFP infectious clone were transfected into H1-HeLa cells followed by three consecutive passages (P1-P3). A cytopathic effect (CPE) was observed in HeLa cell-infected RV1A-GFP ( Figure 1B). However, we were unable to detect GFP expression by Western blot ( Figure 1C To examine the presence of intact GFP and RL ORFs from RV1A-GFP and RV1A-RL, RT-PCR was performed for plaque-purified RV1A-GFP and RV1A-RL using RV-specific flanking primers. Results showed that the GFP and RL sequences were deleted ( Figure 1E). Sequence analysis of inserts confirmed these results (data not shown).

| Generation and characterization of RV1A-iLOV
We reasoned that the approximately 753 nt GFP and 1005 nt RL ORFs (including engineered flanking sequences) exceeded the limited packaging capacity of RV. 24 We therefore chose an alternative smaller fluorescent protein, iLOV (~366 nt). RV1A-iLOV was generated following the design and generation procedure of RV1A-GFP.
CPE and growth kinetics of RV1A-iLOV were determined in HeLa cells. In comparison with the parental virus, the RV1A-iLOV displayed a slightly reduced cytopathic effect and growth rate (Figure 2A,B).
By immunofluorescence, iLOV signal (green) appeared in RV1A-iLOV-infected cells only, further confirming the expression of iLOV ( Figure 2C). Western blot analysis using rabbit sera recognizing the iLOV protein showed expression of a product of the predicted molecular weight in RV1A-iLOV-infected cells ( Figure 2D), while viral capsid proteins VP0 and VP2 were detected in both RV1A-and RV1A-iLOV-infected cells.

| Genetic stability of RV1A-iLOV in cell culture
Next, we examined the genetic stability of RV1A-iLOV in HeLa cells. RV1A-iLOV stocks by RT-PCR revealed that exogenous iLOV DNA was stably retained within the RV genome over five passages (P1 to P5) ( Figure 3E). Sequence analysis of the iLOV insert confirmed these results and revealed no mutations (data not shown).

| Assessment of antiviral effects using RV1A-iLOV in vitro
We explored the application of RV1A-ilOV to antiviral drug screening. Bafilomycin has previously been shown to inhibit RV F I G U R E 1 GFP ORF insertion into the rhinovirus genome is deleted. A, Schematic presentation of the insertion of GFP into RV genome. RV proteins are presented in boxes. 2A pro cleavages at points indicated by yellow solid triangles separate RV structural from nonstructural proteins and releases GFP proteins. Solid blue arrows indicate 3C pro cleavage sites. B, Plaque morphology of HeLa cells infected with parental wild-type RV1A or RV1A-GFP. C, Western blot analysis of whole-cell lysates from HeLa cells infected with RV1A or RV1A-GFP. Samples were probed for the presence of GFP and the RV structural proteins VP0 and VP2. GFP input was made from whole-cell lysate of pEGFP-N1-transfected HeLa cells. D, Detection of RV1A-GFP-infected cells by live-cell imaging and immunofluorescence staining. HeLa cells were infected at an MOI of 0.1 with either sham, parental-RV1A or RV1A-GFP for 16 h. HeLa cells were transfected with pEGFP-N1 for 16 h (bar, 50 μm). RV VP2/0 protein was detected using AF555-conjugated anti-VP2/0 Ab (red); iLOV (green) was directly detected by blue laser; nuclei were stained by DAPI (shown in black; bar, 50 μm). E, RT-PCR analysis of RV-GFP and RV-RL genomes. RV-GFP and RV-RL genomic RNA were isolated from plaque-purified virus stocks. RV1A-GFP and RV1A-RL infectious cDNA clones were used as a template for amplification of complete GFP or RL sequence infection. 30

| Induction of cytokines in RV1A-iLOVinfected mice
We previously showed RV1B, a minor group virus, triggers inflammation and cytokine expression in mice. 11 We therefore tested whether the iLOV insert influences RV replication and RV-induced inflammatory responses in vivo. Eight-week-old mice were infected with RV1A and RV1A-iLOV for up to 4 days, and lungs were harvested at dif-  Table   S1). Retention of the iLOV insert was approximately 90% at the indicated time points ( Figure 5B,C). We next examined cytokine mRNA expression in RV1A-iLOV-infected mice. Lungs were harvested one day after infection. Similar to the parental RV1A virus, RV1A-iLOV increased lung mRNA levels of Il1b, Ifnb1, Ifng, Cxcl1, Cxcl10, Cxcl2, and Ccl2 ( Figure 5D). However, Ifng and Cxcl10 mRNA expressions were decreased for RV1A-iLOV compared to RV1A, perhaps because of the slightly reduced growth rate. No induction of Ccl5 or Il10 was observed for either RV1A or RV1A-ilOV.

| Detection of iLOV in lungs of infected mice
HeLa cells were plated on coverslips, infected with RV1A-iLOV, and fixed in 4% paraformaldehyde. A fluorescent signal was visualized, but the signal was rapidly quenched, likely due to the oxidation and bleaching of iLOV-bound flavin mononucleotide. Similarly, imaging of fixed lung tissue from RV1A-iLOV-infected mice failed to reveal a fluorescent signal. We therefore employed an anti-iLOV antibody for immunohistochemical staining and immunofluorescence F I G U R E 2 Construction of the infectious recombinant rhinovirus-iLOV. A, Plaque morphology of HeLa cells infected with parental wild-type RV1A or RV1A-iLOV. iLOV (green) expression was directly detected by blue laser for RV plaques. B, Viral copy number and titers in RV-infected HeLa cells. Cells were infected with RV1A or RV1A-iLOV at an MOI of 0.1. At specified times, cells were harvested for analysis. Viral copy number was analyzed by quantitative polymerase chain reaction. Viral titer was determined by TCID 50 (n = 3, mean ± SD). C, Fluorescence imaging for RV-iLOV. HeLa cells were infected with parental wild-type RV1A or RV1A-iLOV for 24 h at an MOI of 0.1. RV VP2/0 protein was detected using AF555-conjugated anti-VP2/0 Ab (red); iLOV (green) was directly detected by blue laser; nuclei were stained by DAPI (shown in black; bar, 50 μm). D, Western blot assay to detect the expression of iLOV F I G U R E 3 Stability of RV1A-iLOV. A, Live-cell imaging and immunofluorescence staining of RV1A-iLOV-infected HeLa cells. HeLa cells were infected with P1 and P5 of RV1A-iLOV for 24 h at an MOI of 0.1 RV VP2/0 protein, which was detected using AF555-conjugated anti-VP2/0 Ab (red); iLOV (green) was directly detected by blue laser; nuclei were stained by DAPI (shown in black; bar, 50 μm). B, The number of iLOV-positive cells out of 50 VP2/0-positive cells was counted. All VP2/0-positive cells were iLOV-positive. C and D, iLOV detection in HeLa cells by flow cytometry. Cells were transfected with pXJ41-iLOV or infected with RV1A or RV1A-iLOV, harvested 24 h later, and analyzed as a percentage of single cells (n = 3, mean ± SEM). E, RT-PCR analysis of RV1A-iLOV genomes. Parental RV1A or RV1A-iLOV genomic RNA was isolated from HeLa cells infected with P1 or P5 virus stocks. The RV1A-iLOV infectious cDNA clone was used as a template for amplification of complete iLOV sequence imaging. We infected mice with RV1A-iLOV and harvested lungs one day postinfection. We have previously shown that, besides airway epithelial cells, RV colocalizes with CD68+ macrophages. 13,14 Immunohistochemical staining with anti-iLOV showed signal in both the epithelium and macrophages ( Figure 6A). Immunofluorescence similarly indicated localization in the airway epithelium, as well as colocalization of iLOV (green) with VP2/0 (red) and a macrophage marker, CD68 (blue, Figure 6B).
We have previously shown that RV infection induces lung infiltration with CD11b-positive, M2-polarized exudative macrophages. 33 For the analysis of intracellular iLOV, aliquots of lung mince were fixed, permeabilized, and incubated with the Cy3-tagged anti-iLOV prior to flow cytometry. Flow cytometric analysis showed similar increases in the percentage for CD45 + CD11b+ cells in both RV1A-and RV1A-iLOV-infected mice ( Figure 6C), confirming the colocalization of lung macrophages and RV. However, a signal was detected in CD45+ CD11b+ cells only in RV1A-iLOV-infected mice. Taken together, these results suggest RV1A-iLOV as a potential tool to study RV-induced responses in immature mice.

| D ISCUSS I ON
In the present study, we sought to design and generate a recombi- Construction of recombinant viruses expressing fluorescent markers, especially GFP, has been applied through reverse genetics to RNA viruses including influenza virus, 34 Zika virus, 35 West Nile virus, 36 respiratory syncytial virus, 37 murine coronavirus, 38 and porcine reproductive and respiratory syndrome virus. 39 However, insertion of large fluorescent protein coding sequences into the genome of the picornaviruses, a group of small RNA viruses whose genome sizes range from 7.2 to 8.5 kb, has been problematic. Insertion of the GFP ORF into poliovirus severely impaired viral replication and was deleted in the course of cell culture serial passage. 40 In addition, an attempt to construct a recombinant foot-and-mouth disease virus (FMDV) expressing GFP or Renilla luciferase protein failed, likely due to the limited packaging capability. 41 Subsequent construction of viruses containing increasingly larger inserts suggested 300-400 nt F I G U R E 6 Presence of iLOV signal in the airway epithelium and lung macrophages of RV1A-iLOV-infected mice. A,Twenty-four hours after infection, lungs were fixed in formaldehyde overnight, embedded in paraffin, sectioned at 5 μm, and incubated with a 1:1000 dilution of anti-iLOV or isotype control IgG (bar, 50 μm). B, Lung sections were costained with AF-488-conjugated anti-iLOV (green), AF-555conjugated anti-VP2/0 (red), and Cy5-conjugated CD68 (far red optical spectrum, shown in blue). C, Lung CD45 + , CD11b+, iLOV+ cells from RV-infected BALB/c mice were identified one day postinfection and analyzed as a percentage of CD45 + cells (n = 4 from one experiment). Data are presented as mean ± SEM (*different from sham, P < 0.05; one-way ANOVA) as the maximum size to be inserted into FMDV genome. Consistent with this, RV1A has been used to express a 393-nt-long fragment of HIV gag gene. 24 Shorter antigenic tags have also been inserted into the nonstructural proteins of poliovirus. 42 Compared to GFP, fluorescent proteins based on flavin-binding LOV (light, oxygen, or voltage sensing) domain offer advantages owing to their smaller size (354 nt), pH, and thermal stability. 43  Similar to viral proteins, iLOV protein is released from the RV polyprotein through viral proteinase-mediated auto-cleavage during viral protein production, an early step of viral replication. Detection of iLOV in mouse tissue is therefore highly suggestive of viral replication, particularly in the airway epithelium. However, it remains unclear whether iLOV signal in macrophages represents replication or engulfment of the virus by phagocytosis. Viral replication in cultured macrophages is limited, 45 though it has recently been shown that airway epithelial cells promote rhinovirus replication in monocytic cells. 46 Because there are more than 100 different RV serotypes (in species A and B alone), it is infeasible to develop a cross-reactive antibody for RV. Until now, only one antibody has been available for this purpose, the monoclonal antibody R16-7. This antibody binds to the VP2 capsid protein of the closely related RV-A16 and RV-A1 strains 6 but not to RV-A2, RV-B14, or RV-A49. 15 We developed a recombinant virus with a fluorescent marker that could be used for tracking of RV infection in vivo. We designed the iLOV sequence to be flanked with 2A pro cleavage sites and then inserted between the RV genomic sequences encoding the VP1 and 2A proteins. 2A pro mediates auto-cleavage between VP1 and 2A proteins. 20,47 Since self-catalytic cleavage is a characteristic of picornavirus replication, this design should allow extension of our technique to all human RVs. Given the fact that iLOV sequence was stably maintained within RV1A genome during consecutive passages, it is plausible that other RVs serotypes would accommodate and maintain the stability of iLOV sequence. Though the 2A pro cleavage sites of numerous RV serotypes are heterogeneous, 48 designing the flanked 2A pro cleavage sequence to be serotype-specific would guarantee the release of iLOV. Besides VP1-2A cleavage site, the junction site between 5′UTR and the N-terminus of VP4 has been used to insert GFP in the genome of coxsackie A16 virus 49 ; however, the insertion impaired viral replication. Taken together, these data suggest that the construction strategy for RV1A-iLOV could be applied to other RV serotypes for studying the life cycle of RV in cultured cells, screening for antiviral drugs and for the pathogenesis of RV.

ACK N OWLED G EM ENT
The authors thank Dr. William T. Jackson (University of Maryland School of Medicine) for his gift of pMJ13-RV1A infectious cDNA clone. The authors also thank Dr. Wai-Ming Lee (Biological Mimetics) for his constructive criticism of this research. This work was supported by NIH R01 HL134369. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.