Selection of viral variants during persistent infection of insectivorous bat cells with Middle East respiratory syndrome coronavirus

Coronaviruses that cause severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) are speculated to have originated in bats. The mechanisms by which these viruses are maintained in individuals or populations of reservoir bats remain an enigma. Mathematical models have predicted long-term persistent infection with low levels of periodic shedding as a likely route for virus maintenance and spillover from bats. In this study, we tested the hypothesis that bat cells and MERS coronavirus (CoV) can co-exist in vitro. To test our hypothesis, we established a long-term coronavirus infection model of bat cells that are persistently infected with MERS-CoV. We infected cells from Eptesicus fuscus with MERS-CoV and maintained them in culture for at least 126 days. We characterized the persistently infected cells by detecting virus particles, protein and transcripts. Basal levels of type I interferon in the long-term infected bat cells were higher, relative to uninfected cells, and disrupting the interferon response in persistently infected bat cells increased virus replication. By sequencing the whole genome of MERS-CoV from persistently infected bat cells, we identified that bat cells repeatedly selected for viral variants that contained mutations in the viral open reading frame 5 (ORF5) protein. Furthermore, bat cells that were persistently infected with ΔORF5 MERS-CoV were resistant to superinfection by wildtype virus, likely due to reduced levels of the virus receptor, dipeptidyl peptidase 4 (DPP4) and higher basal levels of interferon in these cells. In summary, our study provides evidence for a model of coronavirus persistence in bats, along with the establishment of a unique persistently infected cell culture model to study MERS-CoV-bat interactions.


MeRS-coV persistently infects insectivorous bat cells. To investigate if MERS-CoV can persistently infect cells from an insectivorous bat, we infected big brown bat (Eptesicus fuscus) kidney cells (Efk)
at a low multiplicity of infection (MOI; 0.01 TCID 50 /cell) (Fig. 1a). Twelve days after infection, most cells were dead (Fig. 1c). At 35 days post infection, surviving cells that were passaged formed a monolayer of continuously growing cells. We passaged the surviving cells and quantified the amount of virus in the supernatant by titration every week for 126 days (Fig. 1b). During the first 35 days of infection, the amount of virus in the supernatant of persistently infected bat cells was variable (Mean = 2.98 × 10 4 ; coefficient of variation = 105.4%) but after the 42 nd day, we detected relatively low, but stable amounts of infectious virus in the cell culture supernatant (Mean = 1.5 × 10 4 ; coefficient of variation = 58.8%) (Fig. 1b). Virus-like electron dense particles were also detected in these cells by electron microscopy (see Supplementary Fig. S1). In comparison, human lung cells (MRC5) that have been shown to support MERS-CoV infection 31 could not be persistently infected with MERS-CoV. All cultured cells were dead several days after infection (data not shown).

MeRS-coV n protein and RnA are detectable in persistently infected bat cells.
To determine if all bat cells in culture were persistently infected, or whether persistence was maintained by small numbers of lytically-infected cells producing virus to subsequently infect susceptible cells, we analyzed these cells for the expression of MERS-CoV nucleo-(N) protein and transcripts. All persistently-infected bat cells expressed varying levels of N protein (Fig. 1d). Consistent with our observations from immunofluorescent microscopy, we detected varying levels of transcripts for MERS-CoV N in all cells in the persistently infected culture by in-situ hybridization (Fig. 1e).

MERS-CoV ORF5 gene segments are undetectable in persistently infected bat cells. Persistent
infection is often accompanied by altered levels of virus and host gene expression, where the virus and host establish a delicate balance between cytolytic viral factors and host defensive responses 32 . To determine if there were differences in MERS-CoV replication and gene expression between acute and persistently infected bat cells, we infected bat cells with a high MOI of MERS-CoV (acute infection) and compared virus replication and gene expression in acute and persistently infected bat cells at 0, 12, 24 and 48 hours post infection (hpi) (Fig. 2a-i). For persistently infected bat cells, the time stamps in the figure indicate when RNA from these cells were harvested after seeding them in six-well plates. Expression of MERS-CoV gene segments S, E, M, N, ORFs 3, 4a, 4b and 5 increased over time in acutely infected cells (Fig. 2b-i). This was concomitant with an increase in MERS-CoV genome quantities (Fig. 2a) in acutely infected cells, as represented by upE levels 33 . In contrast, persistently infected bat cells maintained steady levels of genome (upE) quantities and with the exception of ORF5, other viral genes maintained detectable but stable levels of expression ( Fig. 2a-i). We were unable to detect transcripts for ORF5 in persistently infected bat cells (Fig. 2f).

MeRS-coV ΔORF5 mutants persistently infect bat cells.
The MERS-CoV isolate used in our experiments (EMC/2012) was derived from an infected human and was subsequently propagated in primate cells (see materials and methods). To determine if persistently infected bat cells had selected for mutant variants of MERS-CoV, we sequenced the genome of the virus from persistently infected bat cells that had been passaged 15 times, over 4 months. We detected point mutations in the virus polymerase, spike and matrix genes ( Fig. 3a) but none of these mutations disrupted the coding sequences for these proteins (Table 1). However, we observed a 341 base pair deletion in ORF5. In addition, there was a frameshift mutation at the N-terminal end of the protein that introduced a stop codon at the 20 th amino acid position ( Fig. 3a and Table 1). In theory, the intact ORF5 coding sequence in the bat-adapted MERS-CoV strain would encode a putative 19 amino acid long protein. To identify when the ORF5 deletion mutant (ΔORF5 MERS-CoV) was selected for during the establishment of persistently infected bat cells, we sequenced ORF5 from early and late passage cells. As early as passage two, the detectable virus population in culture comprised of the ORF5 mutant. In the following portion of this article, we refer to the original human-derived virus as the MERS-CoV wildtype virus (W+) and the ORF5 deleted variant as ORF5 mutant (ΔORF5) MERS-CoV.
To determine if the establishment of a persistent MERS-CoV infection and the subsequent selection of viral variants was reproducible, we repeated the infection of naïve Efk (bat) cells with W + MERS-CoV at a low MOI. We recovered and sequenced ORF5 coding sequences from these cells at passages 2, 3 and 4. We detected a point mutation in ORF5 as early as passage 2, which resulted in a stop codon in the coding sequence for ORF5. In theory, the stop codon would terminate the protein at the 107 th amino acid. We confirmed this point mutation in www.nature.com/scientificreports www.nature.com/scientificreports/ ORF5 by sequencing the whole genome of MERS-CoV from persistently infected bat cells at passage 15 ( Table 1). As with the previous attempt to persistently infect Efk cells with MERS-CoV, we observed point mutations in genes other than ORF5. However, unlike in ORF5, the other point mutations did not disrupt the coding sequences for other proteins ( Table 1). The additional point mutations were at different sites compared to changes that were observed in the first attempt ( Fig. 3a and Table 1). The possible effects of these point mutations have been summarized in Table 1. To determine if the replication kinetics of ΔORF5 were different from W + MERS-CoV, we

Bat cells infected with ΔORF5 MERS-CoV express higher levels of interferon stimulated gene transcripts.
A recent study demonstrated that in contrast to its ability to suppress innate antiviral responses in human cells, MERS-CoV induces a robust type I interferon response in bat cells 31 . Since ΔORF5 MERS-CoV replicated better in bat cells than W + MERS-CoV, we examined the differences in their ability to induce innate antiviral responses. We compared the levels of transcripts for interferon beta (IFNβ) and selected downstream IFN stimulated genes (ISGs), such as guanylate binding protein 1 (GBP1), myxovirus resistance 1 (Mx1), interferon regulatory factor 7 (IRF7), melanoma differentiation-associated protein 5 (MDA5) and interferon inducible protein 6 (IFI6) in human and bat cells infected with a high MOI of W+ or ΔORF5 MERS-CoV for 24 hours. IFNβ transcript levels in ΔORF5 infected bat cells were higher than W + MERS-CoV infected human cells at 24 hpi (Fig. 4a). Bat cells infected with ΔORF5 MERS-CoV contained significantly (p ≤ 0.05) higher levels of transcripts for IFI6 and MDA5 (Fig. 4b,e) than cells infected with W + virus. IFI6 transcript levels in bat cells infected with W + MERS-CoV were also significantly higher than W + MERS-CoV infected human cells (Fig. 4b). For other genes that were examined, there was an increasing trend for gene expression in bat cells infected with ΔORF5, but the differences were not statistically significant during the duration of this experiment (p = 0.08-0.11; Fig. 4c persistently infected bat cells are resistant to superinfection with W + MeRS-coV. Viruses often exist as quasispecies, including virus stocks that are prepared in laboratories 34 . Coronaviruses have been detected in multiple bat species globally 35 and wild-caught bats can be persistently infected with coronaviruses 18,25 . The co-evolution of CoVs and bats, and the advantage of long-term CoV infection, if any, have not been explored in bats. In this study, we observed ΔORF5 MERS-CoV as the dominant strain in persistently infected bat cells in two independent experiments. To identify if infection with ΔORF5 MERS-CoV confers a survival advantage upon bat cells, we determined if persistently infected bat cells were resistant to superinfection with W + virus. To differentiate between W+ and ΔORF5 MERS-CoV replication, we examined the levels of genomic (upE) RNA (present in both strains), and ORF5 transcripts, (detectable in W + MERS-CoV only). On re-infecting  www.nature.com/scientificreports www.nature.com/scientificreports/ persistently infected bat cells with W + virus (MOI = 1 TCID 50 /cell), we did not observe an increase in upE or ORF5 RNA levels in the superinfected cells, suggesting a lack of replication of W+ and ΔORF5 virus at 24 and 48 hpi (Fig. 5a,b). In contrast, we observed a significant increase in upE and ORF5 transcript levels at 24-and 48-hours post infection of naïve bat cells that were infected with an equal amount of W + MERS-CoV (Fig. 5a,b). Immediately following infection, ORF5 transcripts in superinfected bat cells (ΔORF5 + W + MERS-CoV) were significantly higher than in persistently infected bat cells (Fig. 5b), but the expression levels did not increase over time, suggesting that W + virus did not replicate and that the initial difference was due to input virus. MERS-CoV uses DPP4 as a receptor for cellular entry and both MRC5 and Efk cells contained comparable amounts of DPP4 transcripts (Fig. 5c -right panel). To determine if the persistently-infected cells were resistant to infection by W + MERS-CoV because of reduced DPP4 expression, we measured relative amounts of DPP4 transcripts in the samples described in Fig. 5a,b. Efk cells infected with W + MERS-CoV contained high levels of DPP4 transcripts, which did not decrease significantly in the 48 hrs following infection (Fig. 5c). While we could amplify DPP4 transcripts in all cells (see gel insert in Fig. 5c), in contrast to naïve Efk cells infected with W + MERS-CoV, the persistently infected cells contained almost 100 fold (average − ΔΔCT values of 6.1) less DPP4 transcripts. These levels did not change significantly in the 48 hrs following infection with W + MERS-CoV.
To determine the protective role of type I interferon in bat cells that were persistently infected with MERS-CoV, we compared the amount of IFNβ transcripts in naïve bat cells infected with W + MERS-CoV and bat cells persistently infected with ΔORF5 MERS-CoV. We observed increased levels of IFNβ transcripts in persistently infected www.nature.com/scientificreports www.nature.com/scientificreports/ bat cells, relative to naïve cells infected with W + virus (at 0-and 12-hours post-infection of naïve cells and post-seeding of persistently infected cells; p = 0.06 and 0.002; Fig. 5d). In addition, persistently-infected Efk cells consistently displayed higher levels of IFNβ transcripts, relative to levels in naïve, uninfected Efk cells that were used to normalize the data. As shown previously 31 , naïve Efk cells infected with W + MERS-CoV displayed a robust induction of IFNβ transcripts at 48 hpi (p = 0.003; Fig. 5d). 36 and it is involved in the regulation of antiviral responses to MERS-CoV in bat cells 31 . To determine if IRF3-mediated innate antiviral responses play a role in the maintenance of persistent infection in bat cells, we reduced IRF3 protein levels (Fig. 6b) in these cells using small-interfering RNA (siRNA). siRNA-mediated reduction of IRF3 in persistently infected bat cells increased MERS-CoV titers by ten-fold (Fig. 6a). Reducing IRF3 protein levels in persistently infected bat cells also increased cell death within 48 hours of siRNA treatment compared to control siRNA-treated persistently infected cells (data not shown).

Interferon regulatory factor 3 (IRF3) and mitogen-associated protein kinase (MAPK) signaling regulates persistent infection in bat cells. Interferon regulatory factor 3 (IRF3) is a central regulator of innate antiviral responses in mammalian cells
In little brown bats (Myotis lucifugus) that are co-infected with a bat CoV and White Nose Syndrome-causing fungus (P. destructans), suppression of the MAPK pathway has been speculated to increase CoV replication 26 .
To test if this pathway played a similar role in regulating viral load in persistently infected bat cells in culture, we inhibited the MAPK signaling pathway using a commercial MAP kinase inhibitor, URMC-99 37 . We observed a significant increase in MERS-CoV replication (upE levels) upon inhibiting the MAPK pathway (Fig. 6c). We confirmed the suppression of the MAP kinase pathway in bat cells by demonstrating a reduction in the transcript levels of a MAPK pathway-related Ras superfamily protein, RAS-like estrogen regulated growth inhibitor (RERG; Fig. 6d) 38,39 .
Since ΔORF5 MERS-CoV was repeatedly selected for in persistently infected Efk cells, we predicted the cellular localization of ORF5 to identify if there was a potential detrimental effect on cells expressing ORF5. By using transmembrane domain prediction softwares and immunofluorescent microscopy, we observed that MERS-CoV ORF5 has 3 transmembrane domains (see Supplementary Fig. S4a) and ectopically expressed MERS-CoV ORF5 localized to the endoplasmic reticulum (ER) in bat cells (see Supplementary Fig. S4b). As bat coronaviruses from the beta-CoV sub-family have ORF5 homologues 18 , we compared MERS-CoV ORF5 with that of ORF5 found in bat beta-CoVs. We found differences in the predicted cytoplasmic domain of MERS-CoV ORF5 (see Supplementary Figs. S4c,d), which may be associated with high toxicity and cell death that we observed with W + MERS-CoV infections in human and bat cells 40 .

Discussion
Successful spillover of a virus to a new species requires viruses to adapt to the use of viral receptors and circumvent innate antiviral defense mechanisms that are unique to each host species. Due to the lack of well characterized in vivo experimental models, it is difficult to examine the processes by which coronaviruses, such as those that cause SARS, MERS, porcine epidemic diarrhea (PED) and swine acute diarrhea syndrome (SADS) have adapted from maintenance in bats to successful infection in their spillover species, such as humans, camels, civet cats and pigs. While closely related coronaviruses have been detected in insectivorous bat species 41,42 , the "culprit" viruses that have caused serious disease in spillover mammalian species have not been isolated from, or characterized in bats.
In this study, we established a long-term persistent infection in bat cells to examine how MERS-CoV (EMC/2012), isolated from a human infection, adapts to replication in cells from an insectivorous bat. Here we demonstrate that unlike in human cells, where infection leads to death of all cells, MERS-CoV establishes a persistent infection in a bat cell line. We observed that: a) Infection of bat cells led to a long-term persistent infection. b) The viability of the cells in culture did not depend on the presence of uninfected resistant cells, as seen in some models of coronavirus persistence 43 . All bat cells in culture contained viral RNA and protein. c) Viral persistence www.nature.com/scientificreports www.nature.com/scientificreports/ was accompanied by the rapid selection of mutations in the accessory gene -ORF5, and this mutation was "fixed" for at least 15 passages of the culture over 4 months. While viral persistence was accompanied by mutations in other genes, none of the mutations appeared to be as drastic as the ORF5 mutations. The genomic locations of other mutations in MERS-CoV also varied between the two independent attempts at establishing persistence. d) The ΔORF5 mutant virus replicated more efficiently in bat cells than W + MERS-CoV, and cells infected at high MOIs by the ΔORF5 mutant contained higher levels of some interferon-stimulated and anti-apoptotic transcripts. e) Cells persistently infected with the ΔORF5 mutant were resistant to infection by W + MERS-CoV. Compared to uninfected or W + MERS-CoV-infected Efk cells, cells persistently-infected with ΔORF5 MERS-CoV expressed greatly reduced levels of DPP4 transcripts, the known receptor for W + MERS-CoV. f) Persistence was maintained by the retention of higher basal levels of cellular interferon response, as suppression of the response by down-regulating IRF3 increased virus replication.
Our observations lead us to suggest a scenario for the molecular mechanisms that govern persistence of MERS-CoV in bat cells (Fig. 7): MERS-CoV stocks grown in permissive primate (Vero) cells largely comprise W + virus that is cytolytic in both primate and bat cells. Due to the error-prone nature of the viral RNA polymerase, the laboratory stock likely contains variants that are less successful for replication in primate cells than the W + virus. These mutants include some with alterations in the ORF5 gene that render the virus less cytolytic for bat cells. Infection of bat cells at a low MOI leads to the infection of cells by either the W+ or ΔORF5 mutant virus. Due to the cytolytic nature of the W + virus, cells infected by it die with the subsequent release of more W + virus. While cells infected by the less cytolytic ΔORF5 mutants also produce virus, they survive. ΔORF5 mutant virus infected cells survive because the virus induces a more robust innate antiviral response in these cells and the apoptotic processes that accompany W + virus infection are dampened (see IFI6 expression; Fig. 4b).
Since infection with ΔORF5 MERS-CoV renders bat cells resistant to infection by W + virus, W + virus is eventually replaced in the culture by ΔORF5 MERS-CoV. Persistence is maintained by elevated levels of basal innate antiviral responses in these cells and suppression of this response leads to an increase in viral replication and cell death.
An interesting observation in our study is the rapid selection of mutations in the gene for ORF5, suggesting that deletion of the coding sequences or other mutations in the gene alter its function in the W + virus and may be an important adaptation for maintenance in bat cells. The role of ORF5 and other accessory genes of beta-coronaviruses has not been clearly established. CoV accessory genes do not appear to be required for virus replication in cultured cells but are thought to be important for infection in vivo or for virulence in specific hosts. In human cells, MERS-CoV structural proteins, such as M, and accessory proteins, such as ORFs 4a, 4b and 5 are known to effectively inhibit innate antiviral responses, however, the extensive role of ORF5 in modulating cellular processes has not been investigated 24,44 . Experiments with adapting MERS-CoV to cause disease in humanized mice suggest a role for ORF5 and other accessory viral genes in disease outcome 45,46 . However, the role of ORF5 in MERS-CoV infection of bat cells or bats has not been explored.
An in-silico analysis of the W + MERS-CoV ORF5 amino acid sequence revealed three potential trans-membrane regions (see Supplementary Fig. S4a), suggesting that it is a membrane-associated protein. Three trans-membrane accessory proteins in alphacoronaviruses, such as feline infectious peritonitis virus 47 , porcine www.nature.com/scientificreports www.nature.com/scientificreports/ epidemic diarrhea virus 48 and human CoV 229E 49 , as well as in betacoronaviruses, such SARS-CoV 50 also appear to be associated with virulence.
We co-expressed ORF5-linked to red fluorescent protein with Luman-GFP, a green fluorescent protein-linked protein that localizes to the endoplasmic reticulum (ER). The two proteins co-localized in the ER indicating that the protein coded by ORF5 is associated with the ER (see Supplementary Fig. S4b). Given these data, we have suggested an ER-associated structure of the ORF5 protein with its carboxyl terminal portion in the cytoplasm (see supplementary Fig. S4c). Interestingly, several bat beta coronaviruses also possess an ORF5 homologue. However, the putative cytoplasmic domains of these proteins differ from those of human and camel MERS-CoV isolates at specific residues (see Supplementary Fig. S4d). While the role of MERS-CoV ORF5 in adaptation for replication in different hosts will require further investigation, based upon the reduced cytotoxicity of ORF5 deletion mutant, we hypothesize that the protein promotes ER-stress induced apoptosis. Viruses with a deleted ORF5 gene or gene specifying a deleted (or mutated) carboxyl terminus would be less likely to trigger apoptotic cell death in ER-stressed persistently infected cells.
We observed that cells persistently infected with MERS-CoV with deleted or truncated ORF5 were resistant to superinfection with W + MERS-CoV. This phenomenon may explain the rapid selection of the mutant (ΔORF5 MERS-CoV) and dilution of W + virus. While we have not clearly established the molecular mechanisms that render the persistently infected cells resistant to superinfection, our results indicate that down-regulation of DPP4, the cellular receptor for MERS-CoV (Fig. 5c) in the persistently infected cells provide an explanation. While we detected DPP4 transcripts in the cells (Fig. 5c), the levels in persistently infected cells (relative to a universally expressed house-keeping gene) were about a hundred times lower than in W + MERS-CoV infected naïve Efk cells.
Superinfection exclusion, a phenomenon by which viruses, following infection, exclude infection by similar viruses is well known and occurs in cells persistently or lytically infected with herpesvirus 51,52 , alphaviruses 53 , flaviviruses 54 , papillomaviruses 55 and influenza viruses 56 . The mechanisms by which infected cells are rendered resistant to further infection vary, but often rely on the occlusion of receptors for the superinfecting virus. We found that the levels of DPP4 transcripts in infected cells did not decrease following infection of Efk cells with either W + or ΔORF5 MERS-CoV (Figs. 4f and 5c). This suggests that during a normal infection at a high MOI, MERS-CoV does not down-regulate the expression of the receptor. However, a lack of bat specific immunological reagents prevented us from detecting bat DPP4 on the surface of MERS-CoV infected Efk cells, so we cannot rule out a decrease in the expression of the protein on the cell surface or occlusion by the viral spike protein.
While persistence in bat cultured cells, devoid of the in vivo environment, may not exactly recapitulate MERS-CoV -bat interactions, our model of MERS-CoV persistence mirrors some features observed during enhanced replication of persistently-infecting CoV in bats suffering from White Nose Syndrome 25,26 . Our model also allows us to develop hypotheses that can then be tested by observations in a wild free-living species. Virus titration. Supernatant from persistently infected bat cells were titrated in triplicates on Vero 76 cells using tissue culture infectious dose 50 (TCID 50 ) assay. Briefly, 10 4 cells were seeded in each well of a 96-well plate. The plates were incubated overnight to obtain a confluent layer of Vero 76 cells. The virus sample (supernatant) was diluted ten-fold and 50μl of the diluted virus sample was added to each well of the 96-well plate. The plates were incubated at 37 °C for 1 hr. After incubation, the virus containing supernatant was discarded and 100μl of complete media with 5% FBS was added to the plates. The plates were incubated for 5 days and cytopathic effect (2020) 10:7257 | https://doi.org/10.1038/s41598-020-64264-1 www.nature.com/scientificreports www.nature.com/scientificreports/ was observed under a light microscope. Tissue culture infectious dose 50/ml (TCID 50 /ml) was calculated using the Spearman and Karber algorithm 58,59 . immunoblot. Persistently infected Efk cells were seeded at a concentration of 3 × 10 5 cells/well in six well plates and incubated at 37 °C overnight. Next day, cells were harvested in sample buffer for western blots. Western blots were carried out as previously mentioned 60 . Briefly, samples were denatured in a reducing sample buffer and electrophoresed on a reducing gel. Proteins were blotted from the gel onto polyvinylidene difluoride (PVDF, GE Healthcare, Cat #10600023) membranes and detected using primary and secondary antibodies. Primary antibodies used were: 1: 1,000 rabbit anti-IRF3 (Abcam; Cat# ab68481), 1: 1,000 mouse anti-GAPDH (EMD Milipore, Cat #AB2302) and 1: 1,000 rabbit anti-MERS-CoV N protein (Sino Biological, Cat #40068-RP01). Secondary antibodies used were: 1:10,000 goat anti-mouse Alexa 488 (Molecular Probes, Cat #A-11001) and 1: 10,000 goat anti-rabbit Cy5 (GE Healthcare, Cat #PA45012). Blots were observed and imaged using a Typhoon Scanner (Amersham Biosciences). electron microscopy. Persistently infected Efk cells were scraped in phosphate buffered saline and pooled in a 15 ml screw-cap centrifuge tube. The cells were recovered by centrifugation at 300xg for 5 mins and the supernatant was discarded. The cell pellet was resuspended in 10% neutral buffered formalin (10% NBF; Sigma, Cat #HT501128) for virus inactivation. The cells were later processed for electron microscopy as mentioned previously 25 . Briefly, cells were treated with osmium tetroxide (1% OsO 4 , 0.1 M sodium cacodylate buffer) for one hour at room temperature. Samples were quickly rinsed with water, gradually dehydrated in ethanol and en-bloc stained with uranyl acetate. After rinsing three times (5 min each) in propylene oxide, samples were infiltrated with Epon/Araldite (electron Microscopy Sciences, Cat #50-980-381). Samples were placed in molds and freshly prepared Epon/Araldite was added. The samples were then polymerized at 60 °C for 24-48 h. Sections of 90 nm were cut and observed by transmission electron microscopy (TEM -Hitachi HT 7700, Tokyo, Japan).
In-situ hybridization. As for immunofluorescence, persistently infected and mock/acute infected Efk cells were seeded at a concentration of 3 × 10 4 cells/chamber in chamber slides (Nunc, Thermo Scientific). Each chamber was harvested at the respective time points by removing media, a quick rinse with PBS and addition of 10% NBF. 10% NBF was added to inactivate and fix the cells over a period of 48 h. MERS-CoV RNA was detected in the cells using ViewRNA ISH tissue 2-plex assay kit (Affymetrix, Thermofischer Scientific, Cat #QVT0012) and probes targeting MERS-CoV nucleocapsid sequence (Thermofischer Scientific, United States) as per manufacturer's instructions, with the exception of xylene treatment of slides, which was omitted. Counterstaining of cells was done using Gill's hematoxylin I (Sigma Aldrich, Cat #GHS-132). Coverslips were applied at the end of the assay and images were taken using brightfield microscope with 20X and 40X objectives.
RNA extraction, cDNA preparation and quantitative real-time PCR (qRT-PCR). All RNA extractions were performed using the RNeasy Plus Mini kit (Qiagen, Cat #74136) as per manufacturer's instructions. cDNA was prepared using iScript gDNA clear kit (Bio-Rad) as per manufacturer's instructions. 500 ng of RNA was used for cDNA preparation. cDNA was used as a template for the quantification of target genes. Quantitative real-time polymerase chain reaction (qRT-PCR) assays amplifying targeted cellular genes and the normalizer gene (Glyceraldehyde-3-phosphate; GAPDH) were performed for both MRC5 and Efk cells. Primer sequences are listed in Table 2. MERS-CoV gene primers were limited in their capacity to detect genomic and sub-genomic RNA pertaining to the respective gene segment. For example, nucleocapsid primer was designed to detect nucleocapsid gene segment present in the genomic RNA as well as in the sub-genomic RNA. Primers for quantifying GBP1, IL8, Mx1, IRF3, MDA5, IRF7, TNFα and IFI6 were used as mentioned previously 61 . Bio-Rad's CFX96 Touch PCR thermocycler was used in conjunction with Bio-Rad's Ssofast Evagreen supermix (Bio-Rad, Cat #1725204) and samples were prepared as previously described 62 . For qRT-PCR, after the initial denaturation step of 95 °C for 5 minutes, two step cycling for 40 cycles was performed at 95 °C/10 s, 56 °C/30 s. Absorbance readings were acquired after each cycle. The final three steps were carried out at 95 °C/1 min, 55 °C/30 s and 95 °C/30 s to generate the dissociation curve. Absorbance readings for the dissociation curve were acquired at every degree from 55-95 °C. Relative fold change in gene expression was calculated after normalizing the Ct values using GAPDH and the data was normalized to mock infected samples or as mentioned in the figure legends. Difference of one Ct indicates a two-fold difference in gene expression. Graphs were plotted using GraphPad prism version 7.
ORF5 domain prediction and multiple sequence alignment. ORF5 domain prediction was performed by inputting the ORF5 amino acid sequence into the TMHMM Server v.2.0 65 . Amino acid sequence from CoVs were retrieved from NCBI's database (accession numbers in supplementary Fig S4d), and multiple sequence alignment was performed using MacVector software 66 .