Transgene expression in the genome of Middle East respiratory syndrome coronavirus based on a novel reverse genetics system utilizing Red-mediated recombination cloning

Middle East respiratory syndrome coronavirus (MERS-CoV) is a high-priority pathogen in pandemic preparedness research. Reverse genetics systems are a valuable tool to study viral replication and pathogenesis, design attenuated vaccines and create defined viral assay systems for applications such as antiviral screening. Here we present a novel reverse genetics system for MERS-CoV that involves maintenance of the full-length viral genome as a cDNA copy inserted in a bacterial artificial chromosome amenable to manipulation by homologue recombination, based on the bacteriophage l Red recombination system. Based on a full-length infectious MERS-CoV cDNA clone, optimal genomic insertion sites and expression strategies for GFP were identified and used to generate a reporter MERS-CoV expressing GFP in addition to the complete set of viral proteins. GFP was genetically fused to the N-terminal part of protein 4a, from which it is released during translation via porcine teschovirus 2A peptide activity. The resulting reporter virus achieved titres nearly identical to the wild-type virus 48 h after infection of Vero cells at m.o.i. 0.001 (1 10 p.f.u. ml 1 and 3 10 p.f.u. ml , respectively), and allowed determination of the 50% inhibitory concentration for the known MERS-CoV inhibitor cyclosporine A based on fluorescence readout. The resulting value was 2.41 μM, which corresponds to values based on wild-type virus. The reverse genetics system described herein can be efficiently mutated by Red-mediated recombination. The GFP-expressing reporter virus contains the full set of MERS-CoV proteins and achieves wild-type titres in cell culture. INTRODUCTION Middle East respiratory syndrome coronavirus (MERSCoV) has been identified by the World Health Organization (WHO) as one of the priority pathogens to be studied in the context of outbreak preparedness and intervention [1]. As of 24 April 2017, 1936 cases of MERSCoV infection have been reported, including 690 (35.6%) deaths [2]. MERS-CoV poses a serious and continuing threat to international public health because of its source in a livestock species and its transmission route via the respiratory tract [3, 4]. Dromedary camels show high seropositivity, in the range 55–95%, depending on age [5]. Between 25 and 59% of camels may actively shed the virus, providing an opportunity for frequent animal-tohuman transmission in vulnerable persons [5, 6]. Serological evidence suggests that MERS-CoV in camels must have been enzootic in East Africa and the Middle East for several decades [5, 7, 8]. Although human-to-human transmissibility of MERS-CoV is generally low, introduction into health care settings can lead to virus amplification and limited onward transmission [9–11]. Hospitalassociated mortality can range between 14.5 and 65% [12–14]. Currently, there is no approved vaccine and only limited clinical data on disease outcome under empirical treatments such as interferon (IFN), ribavirin or corticosteroids (both reviewed in [15, 16]). The reasons for the high pathogenicity associated with MERS-CoV in comparison to other CoVs remain to be determined. Received 16 May 2017; Accepted 13 August 2017 Author affiliations: Institute of Virology, Helmut-Ruska-Haus, Charit e Universit€ atsmedizin Berlin, Charit eplatz 1, 10117 Berlin, Germany; Institute of Virology, University of Bonn Medical Centre, Sigmund-Freud-Str. 25, 53127 Bonn, Germany; German Centre for Infection Research (DZIF), Inhoffenstraße 7, 38124 Braunschweig, Germany; Institut für Virologie, Robert von Ostertag-Haus, Zentrum für Infektionsmedizin, Freie Universit€ at Berlin, Robert-von-Ostertag-Str. 7-13, 14163 Berlin, Germany. *Correspondence: Christian Drosten, christian.drosten@charite.de


INTRODUCTION
Middle East respiratory syndrome coronavirus (MERS-CoV) has been identified by the World Health Organization (WHO) as one of the priority pathogens to be studied in the context of outbreak preparedness and intervention [1]. As of 24 April 2017, 1936 cases of MERS-CoV infection have been reported, including 690 (35.6 %) deaths [2]. MERS-CoV poses a serious and continuing threat to international public health because of its source in a livestock species and its transmission route via the respiratory tract [3,4]. Dromedary camels show high seropositivity, in the range 55-95 %, depending on age [5]. Between 25 and 59 % of camels may actively shed the virus, providing an opportunity for frequent animal-to-human transmission in vulnerable persons [5,6]. Serological evidence suggests that MERS-CoV in camels must have been enzootic in East Africa and the Middle East for several decades [5,7,8]. Although human-to-human transmissibility of MERS-CoV is generally low, introduction into health care settings can lead to virus amplification and limited onward transmission [9][10][11]. Hospitalassociated mortality can range between 14.5 and 65 % [12][13][14]. Currently, there is no approved vaccine and only limited clinical data on disease outcome under empirical treatments such as interferon (IFN), ribavirin or corticosteroids (both reviewed in [15,16]). The reasons for the high pathogenicity associated with MERS-CoV in comparison to other CoVs remain to be determined.
Pathogenicity and interference with the host defences in CoV may be influenced by accessory proteins enconded in the viral genome. Among the MERS-CoV accessory proteins, 4a (p4a), p4b and p5 were studied for their IFN antagonistic properties. P4a was found to inhibit the IFN induction pathway by suppressing dsRNA-mediated infection sensing. Another important component of p4a activity is interference with PKR-mediated stress response [17,18]. MERS-CoV p4b has phosphodiesterase activity and counteracts the IFN response by antagonizing activation of the 2¢, 5¢-oligoadenylate synthetase/RNase L pathway [19]. Other studies have identified activities against IFN beta expression [20,21]. P5 inhibits the induction of IFN-b production by preventing nuclear translocation of IFN regulatory factor 3 [22]. Reverse genetics is a valuable tool to modify and study the function of viral proteins in the context of a replicating virus.
Coronavirus reverse genetics is challenging because of the difficulty in stably cloning large and continuous viral genes in bacteria. Based on prior experience with SARS-CoV [23], we present a novel reverse genetics system for MERS-CoV that combines advantageous features of several classical approaches in coronavirus reverse genetics. It relies on T7based in vitro RNA transcription to avoid the risk of unwanted splicing of long virus RNA genomes in the cell. The full-length genome is propagated in an E. coli bacterial artificial chromosome (BAC) plasmid to circumvent in vitro ligation steps and provide a high yield of cDNA template. To enable more efficient mutagenesis on full-length BACcontained viral cDNA, we introduced two-step Red-mediated recombination mutagenesis [24]. Application of the technique is demonstrated by the generation of different variants of MERS-CoV reporter-expressing viruses.

RESULTS
Based on the existing reverse genetics system for SARS-CoV, a new system for MERS-CoV was established. Six subgenomic fragments were reverse transcribed from viral RNA into cDNA, amplified by PCR and cloned into BAC vector pBeloBAC via primer-added restriction sites.
Stepwise assembly of the full-length genome was achieved using naturally occurring restriction sites as shown in (Fig. 1a). An SwaI restriction site at genome position 20 764 was deleted while maintaining the amino acid sequence to enable subsequent cloning steps. This silent mutation also served as a marker mutation to distinguish wild-type from recombinant MERS-CoV (rMERS-CoV) (Fig. 1b). The T7 promoter sequence was added upstream of the MERS-CoV cDNA genome copy. A poly A tail of 20 nt followed by an MluI restricton site was added downstream of the genome. After linearization of the genome-containing BAC plasmid at the unique MluI site, capped infectious viral RNA was transcribed in vitro. In order to enhance virus rescue, genomic RNA was mixed with N gene transcript [25] and electroporated into baby hamster kidney (BHK) cells. Twenty-four hours post-electroporation, BHK cell culture supernatant was transferred to VeroB4 cells. Viral genomic RNA synthesis was monitored by real-time RT-PCR [26] and recombinant virus harvested 3 days post-infection. To determine whether rMERS-CoV had growth characteristics similar to the clinical isolate, VeroB4 cells were infected at an m.o.i. of 0.001. Wild-type and recombinant virus titres were nearly identical at 24 and 48 h p.i. (Fig. 1c).
High-throughput screening of antiviral inhibitors would be greatly facilitated by using a MERS-CoV reporter virus. Scobey et al. described a recombinant MERS-CoV expressing RFP rather than ORF5 [27]. However, removal of ORF5 eliminates a viral protein and the generated reporter virus replicated less efficiently than the recombinant wild-type virus. In order to maintain the natural viral protein repertoire, GFP was released from a fusion protein by interruption of translation conferred by a porcine teschovirus 2A peptide sequence [28]. This sequence prevents the ribosome from forming a peptide bond. Translation is continued, but two peptide chains are generated. In order to account for the potential influence of an additional sequence on genome structure and regulatory elements, two candidate sites for GFP insertion were chosen (Fig. 2a). The first site was the 5¢-end of ORF4 (rMERS-CoV-O4-GFP), whose transcription-regulatory sequence (TRS) presumably ends in front of the start codon and should thus allow insertion of foreign sequence elements without affecting regulation of transcription. The second insertion site was the 3¢-end of ORF5 (rMERS-CoV-O5-GFP). The ORF5 stop codon is located well upstream of the body TRS of the ensuing envelope (E) gene, leaving a non-coding sequence of 58 nt. Insertion of an additional gene immediately upstream of the ORF5 stop codon might not cause important changes in the sequence context upstream of the E gene that is considered relevant for transcription [29]. Two-step Red-mediated recombination [24] was used to insert GFP into the full-length MERS-CoV clone at the described genome positions. This system combines homologous recombination using the Red system of bacteriophage l with cleavage by homing endonuclease I-SceI to produce seamless insertions in target genes by recombination. The reporter viruses rMERS-CoV-O4-GFP and rMERS-CoV-O5-GFP were rescued from cDNA. Generated stocks were quantified by plaque titration revealing that rMERS-CoV-O4-GFP produced significantly smaller plaques than rMERS-CoV (average diameters 1.0 and 2.2 mm, respectively; t-test, P<0.0001), while rMERS-CoV-O5-GFP produced even smaller and more heterogeneous plaques than rMERS-CoV (average diameter 0.7 mm; t-test, P<0.0001, Fig. 2b). Growth of the reporter viruses was compared to that of rMERS-CoV. VeroB4 cells were infected with the three recombinant viruses at an m.o.i. of 0.001 and supernatants titred at 24 and 48 h p.i. Both reporter viruses reached slightly lower titres compared to rMERS-CoV, with growth pattern rMERS-CoV>rMERS-CoV-O4-GFP>rMERS-CoV-O5-GFP (Fig. 2c). Immunofluorescence microscopy showed GFP expression in cells infected with both types of reporter virus, but not in cells infected with rMERS-CoV (Fig. 2d).
Because of its more efficient growth, only rMERS-CoV-O4-GFP was included in further experiments. A Western blot analysis confirmed that rMERS-CoV-and rMERS-CoV-O4-GFP-infected cells expressed similar amounts of ORF4a-encoded p4a, but only cells infected with the reporter virus expressed GFP (Fig. 3a). Interruption of translation mediated by the 2A peptide was verified by the absence of detectable GFP-p4a fusion protein. To correlate the production of infectious particles against the expression of the fluorescence marker, VeroB4 cells were infected with rMERS-CoV-O4-GFP. Cell culture supernatants were titred and the fluorescence intensity measured quantitatively at various time points. At 48 h p.i., infectious virus titres and fluorescence reached maximum values (Fig. 3b). Fluorescence signal declined subsequently, with infectious viral titres remaining at plateau level. Cells infected with rMERS-CoV showed virus production but no fluorescence signal in the quantitative readout.
The reporter virus rMERS-CoV-O4-GFP was assessed for its suitability in viral inhibitor testing using the CoV inhibitor cyclosporin A (CsA) [23,30]. The spectrum of activity of CsA is known to include MERS-CoV [31]. VeroB4 cells were infected with rMERS-CoV-O4-GFP at an m.o.i. of 0.1, and cultivated in the presence of CsA at concentrations between 1 and 16 µM for 48 h. Virus growth (in percentage) was plotted against the logarithmic CsA concentrations (Fig. 4a). IC 50 was calculated to be 2.41 µM. At this concentration cell viability was still 100 %. Potential differences in CsA activity against rMERS-CoV and rMERS-CoV-O4-GFP were determined by virus titration. IC 50 values for rMERS-CoV and rMERS-CoV-O4-GFP were 5.92 and 4.62 µM, respectively, differing by a factor of 1.3 (Fig. 4b).

DISCUSSION
Here we present a novel MERS-CoV reverse genetics system that combines the advantages of existing systems for CoV reverse genetics. It uses a BAC-maintained cDNA fulllength genome [32] that does not require in vitro ligation of fragments before RNA transcription [27]. This approach may improve the reliability of viral genome cDNA recovery. Second, it uses in vitro transcribed RNA to initiate replication in transfected cells, obviating the need for the viral cDNA genome to be transcribed in the nucleus, where it may undergo unwanted splicing [32]. Recombination-based mutagenesis was more rapid than classical mutagenesis because it circumvents the introduction of mutations at the level of subclones, necessary in classical mutagenesis [23]. Recombination-based mutagenesis moreover yielded a higher frequency of mutated clones due to the use of an additional positive selection marker. The genomic insertion site of reporter genes is critical. The reporter can be fused to a viral gene or replace a viral gene. The advantage of the first option is that no viral gene is lost that would be critically needed for replication or interaction with the host cell. Nevertheless, the genetic modification may also affect the function and/or expression of the protein. Insertion of the reporter gene rather than a nonessential, accessory gene might minimize the impact on reporter virus replication in cell culture [33,34]. The previously published reporter-expressing rMERS-CoV has its accessory gene ORF5 replaced by RFP [27]. However, p5 inhibits the induction of IFN-b production by preventing nuclear translocation of IFN regulatory factor 3 [22]. In a growth analysis this virus yielded 10-fold lower virus yield compared to the parental virus in an immortalized cell line.
In primary alveolar type II pneumocytes, primary lung microvascular endothelial cells and primary lung fibroblasts, the reporter virus yielded between 10-and 100-fold less progeny [22]. Although some loss of replication level can be tolerated in many experimental settings, we favoured a less invasive reporter expression strategy based on insertion of a picornavirus 2A peptide. This sequence of 22 amino acids from porcine teschovirus facilitates autonomous and highly efficient intra-ribosomal self-processing of polyproteins [28,35]. The viral gene to which the reporter gene is fused will be left unaffected by this strategy, as previously demonstrated for other plus-strand RNA viruses such as dengue-, Sindbis-and Seneca Valley virus [36][37][38].
The coronaviral genome is sensitive to alterations. Mutations can influence the transcription level if located in core TRS elements, in the upstream (5¢-proximal) TRS context, as well as in distant genome regions [39,40]. In this study, two alternative sites were chosen for the insertion of GFP to account for proximal sequence interactions. The first insertion site chosen was the 5¢-end of ORF4a. Excepting the ORF4a start codon, all other parts of the TRS are located upstream of ORF4a. At the alternative insertion site in the 3¢-part of ORF5, the ensuing E gene starts 58 nt downstream of the termination codon, preserving much of the upstream context of the E-gene TRS. The observed loss of growth efficiency associated with this insertion site may be due to long-distance RNA-RNA interactions [39]. For instance, it has been described for transmissible gastroenteritis virus that secondary structures within 600 nt upstream of the nucleocapsid gene's core TRS are necessary for optimal transcription. CoVs lacking the E gene are replication-incompetent or severely attenuated [41,42]. However, as we have not focused on the ORF5-GFP construct for establishing the reporter virus, we have not tested p5 and E expression.
rMERS-CoV-O4-GFP replicated to viral titres comparable to rMERS-CoV and showed wild type-like expression of p4a. The selected insertion site at the 5¢-end of ORF4a thus allowed seamless integration of GFP and is most likely to be useful for integration of other foreign genes into the MERS-CoV genome. No GFP-p4a fusion protein was detected by Western blot analysis, proving the highly efficient activity of the 2A peptide. An observation of practical importance was that the fluorescence signal of rMERS-CoV-O4-GFP was highest at 48 h p. i. and declined afterwards, while the level of viral particles in the cell culture supernatant remained constant. GFP may be degraded during apoptotic or normal cell metabolism, whereas viral particles accumulate outside cells and remain infectious for a considerable time in the supernatant. GFP readout for the antiviral inhibitor test was therefore done at 48 h p.i.
When testing the known MERS-CoV inhibitor CsA, the IC 50 based on fluorescence signal was in the same range as that using infectious particles and corresponded to the effective CsA concentration identified in a previous study [31]. The IC 50 values determined using rMERS-CoV-O4-GFP versus rMERS-CoV differed by a factor of 1.3 only. This is most likely an intra-assay variation, but we cannot exclude causation by a slightly impaired growth of the reporter virus, requiring reduced CsA concentrations for efficient growth inhibition.
In summary, we describe a new reverse genetics system for MERS-CoV that can be mutated easily and time-efficiently by homologue recombination. Based on this system we identified a suitable genome position for the insertion of a reporter gene without losing expression of viral proteins.

Virus growth kinetics and plaque titration
Twenty-four hours prior to infection, Vero B4 cells were seeded in 6-well plates at a concentration of 2Â10 5 cells ml À1 . The supernatant was removed and cells washed once with PBS (Thermo Fisher Scientific) before virus (diluted in OptiPro serum-free medium, Thermo Fisher Scientific) adsorption. After incubation for 1 h, virus-containing supernatant was removed, and cells washed twice with PBS and supplied with DMEM as described above. Plaque titration was done as described elsewhere [23,43]. Briefly, 1Â10 5 VeroB4 cells were seeded per well in a 24-well plate 24 h prior to infection with a serial dilution of virus-containing cell culture supernatant diluted in OptiPro. One hour after adsorption, supernatants were discarded and cells overlaid with 2.4 % Avicel (FMC BioPolymers, Brussels, Belgium) diluted 1 : 2 in 2Â DMEM supplemented with 20 % fetal bovine serum, 2 % penicillin/streptomycin, 2 % non-essential amino acids, 2 % L-glutamine and 2% sodium pyruvate. Four days post-infection the overlay was removed, and cells were fixed in 6 % formaldehyde and stained with a 0.2 % crystal violet, 2 % ethanol and 10 % formaldehyde solution.
All infection experiments were done under biosafety level 3 conditions with enhanced respiratory personal protection equipment.
All plasmids were digested at the naturally occurring restriction sites indicated, digested plasmids separated on a 0.8 % agarose gel and desired fragments purified using Nucleo-Spin Gel and PCR Clean-up (Macherey-Nagel) or QIAEXX II Gel Extraction Kits (QIAGEN). Ligation was performed for 16 h at 14 C using T4 DNA Ligase (Roche). The fulllength genome-containing plasmid was Sanger sequenced. The integrity of the genomic and sub-genomic inserts was checked using PCRs that amplify the whole insert in overlapping fragments of 3-4 kb prior to downstream applications.

Rescue of recombinant MERS-CoV
Rescue of recombinant virus was carried out as already described for SARS-CoV [23]. Genomic viral RNA was transcribed in vitro from 1 µg MluI linearized and phenolchloroform extracted full-length plasmid using Ambion's mMESSAGE mMACHINE Kit. Additionally, capped nucleocapsid transcript was generated using 1 µg of a purified PCR product spanning the N gene (forward primer adding SP6 promoter sequence: 5¢-ggccatttaggtgacactatagatggcatcccctgctgcac-3¢, reverse primer: 5¢-ttttttttttttttttttttgcaaatcatctaattagcc-3¢). Ten µg of genomic viral RNA and 2 µg of N gene transcript were added to 4Â10 6 BHK cells resuspended in 100 µl OptiPro and electroporated in a 2 mm gap cuvette by applying one pulse of 140 V in 25 msec in a Gene Pulser Xcell (Biorad). Electroporated cells were grown for 24 h in a T75 cell culture flask. Cell culture supernatant was then transferred to susceptible VeroB4 cells and viral genomic RNA synthesis monitored for 3 days by real-time RT-PCR [26] before progeny virus was harvested, plaque titred and stocked for further use.

Generation of MERS-CoV reporter viruses by Red-mediated recombination
Red-mediated recombination was done as previously described [24]. The molecular procedure is shown in Fig. 5 using the insertion of the GFP at the 5¢-end of ORF4a as an example. The full-length cDNA plasmid was transformed into E. coli strain GS1783. This strain carries chromosomally encoded genes for Red recombination under a temperature-inducible promoter and the I-sceI gene under an arabinose-inducible promoter. The Porcine teschovirus 2A peptide was cloned to the N-and C-terminus of the GFP gene (cDNA for codon-optimized Aequorea coerulescens GFP was kindly provided by Dr Sebastian Hauka, Institute of Virology, University of Bonn Medical Centre). The universal transfer construct was created by inserting a selection marker cassette, consisting of an I-SceI restriction site, a kanamycin resistance gene and a short-sequence duplication (50 nts) of GFP into the GFP sequence using the unique restriction site AgeI. The universal transfer construct was PCR amplified using specific primers that contained 5¢extensions of approx. 50 nts, corresponding to the MERS-CoV sequence at the site of recombination. PCR primers MERS-GFP-4-F (5¢-agtgtgaatctttttgacgttggttactcagttaattaacgaactctatggtgtctaaaggagccgagctg-3¢) and MERS-GFP-4-R (5¢-ggtgagttaaggtacttctgccaaatttgattaagcagagacacgtaatccatag gtccagggttctcctccac-3¢) were used for the insertion of GFP at the 5¢-end of ORF4a. Primers MERS-GFP-5-F (5¢-cgcgcgattcagttcctcttcacataatcgccccgagctcgcttatcgttggaagcggagctactaac ttc-3¢) and MERS-GFP-5-R (5¢-actaatggattagcctctacacg ggacccatagtagcgcagagctgcttaatacagttcgtccatgccgtg-3¢) were used for the insertion of GFP at the 3¢-end of ORF5. The PCR product was DpnI digested and gel purified to reduce input plasmid contamination. The purified PCR product was electroporated into E.coli strain GS1783 containing the full-length MERS-CoV BAC clone and recombination was initiated. Selection of positive clones was done on LB-agar plates containing 30 µg ml À1 chloramphenicol and 30 µg ml À1 kanamycin at 32 C for 48 h. Positive clones were screened by a PCR spanning the recombination site. To remove the kanamycin resistance cassette, in vivo cleavage was induced at the I-SceI site using 1 % L-(+) arabinose. A second recombination was induced by incubation of E. coli GS1783 at 42 C for 30 min. Positive clones were selected on LB-agar plates containing 1 % L-(+) arabinose and 30 µg ml À1 chloramphenicol. Successful recombination resulting in a full-length MERS-CoV cDNA genome carrying the GFP gene was confirmed by sequencing of recombination sites and the insert.

Immunofluorescence
Vero B4 cells were seeded at a density of 5Â10 4 cells ml À1 in a 24-well plate containing glass slides 24 h before infection. Cells were mock-infected or infected with rMERS-CoV, rMERS-CoV-O4-GFP or rMERS-CoV-O5-GFP at an m.o.i. of 0.1. Forty-eight hours post-infection, cells were washed twice with PBS and fixed with 6 % formaldehyde for 1 h. Cells were washed twice with PBS and glass slides were mounted using ProLong Gold Antifade mounting medium containing DAPI (Thermo Fisher Scientific). Slides were left to dry in the dark and stored at 4 C until fluorescence microscopy was performed.  . This universal transfer construct was PCR amplified (c), and 50 nt corresponding to the site of recombination (d) were added. During the first Red recombination, GFP and the marker cassette were inserted at the 5¢end of ORF4a (e). The marker cassette was then excised by in vivo cleavage at the I-SceI site and a second Red recombination (f). Adapted from [24].

Western blot
Forty-eight hours post-infection, cells were washed once with ice-cold PBS, scraped into 1 ml PBS and centrifuged for 10 min at 300 g and 4 C. Cell pellets were lysed in 50 µl Pierce IP Lysis Buffer (Thermo Fisher Scientific) supplied with 100Â Protease Inhibitor Cocktail Set III (Merck Millipore) for 30 min at 4 C. Cell debris was pelleted for 10 min at 13 000Âg and 4 C and the supernatant transferred to a fresh tube and mixed with 4Â NuPAGE LDS Sample Buffer (Thermo Fisher Scientific) supplied with 10 % 2-mercaptoethanol (Roth). Samples were inactivated for 10 min at 99 C. Proteins were separated on a 12 % sodium dodecyl sulfatepolyacrylamid gel and transferred onto a 0.2 µm PVDF membrane (Merck Millipore) by semi-dry blotting (Biometra). Detection of 12 kDa MERS-CoV p4a was done using a polyclonal rabbit anti-p4a antibody (1 : 1250, Eurogentec), a horseradish peroxidase (HRP)-coupled goat anti-rabbit antibody (1 : 10,000, Dianova) and SuperSignal West Femto Chemiluminescence Substrate (Thermo Fisher Scientific). The membrane was stripped of antibodies using ReBlot Plus Strong Antibody Stripping Solution (10Â, Merck Millipore) and probed again with mouse anti-b actin (1 : 5000, Sigma Aldrich) and rabbit anti-GFP (1 : 2000, Life Technologies) antibodies. Visualization was done as described above, additionally using goat anti-mouse HRP antibody (1 : 10,000, Dianova).

Inhibitor test
Vero B4 cells were seeded at a concentration of 2Â10 5 cells ml À1 in a black 96-well plate 24 h prior to infection. Cells were infected with rMERS-CoV or rMERS-CoV-O4-GFP at an m.o.i. of 0.1 as described above. After 1 h of virus adsorption the inoculum was removed, and DMEM without phenol red containing 10 % FCS and different concentrations of cyclosporine A (1-16 µM) was applied. Fluorescence intensity was measured 48 h post-infection, for which infection medium was removed, cells washed once with PBS and 50 µl of PBS applied to each well. Fluorescence readout was performed without the plate lid using a Synergy Mx microplate reader (Biotek) at the following settings: excitation (gap): 475 nm (13.5); emission (gap): 505 nm (13.5); gain: 100 V; readings per well: 30; read height: 4.5 mm; temperature: 37 C. Titration of supernatants was performed as described above. Cytotoxicity of CsA was determined using the Cell-Titer-Glo Luminescent Cell Viability Assay (Promega). Vero B4 cells were seeded exactly as described above and treated with the same concentrations of CsA that were used for the inhibitor assay. After 48 h, 50 µl of medium was replaced by 50 µl of CellTiter-Glo substrate and incubated for 10 min. Luminescence was read according to the manufacturer's instructions. In order to calculate the percentage of virus growth or cell viability, fluorescence or luminescence signals determined at 0 µM CsA were set to 100 %.

Funding information
This work was funded by the German Research Foundation through project number DR 772/12-1.