MERS-CoV Accessory ORFs Play Key Role for Infection and Pathogenesis

Using our established reverse-genetics system (20), we had previously reported the generation and attenuation of a MERS-CoV mutant virus lacking accessory ORF proteins 3, 4a, 4b, and 5 (dORF3-5). To extend these finding, Calu-3 cells, an immunocompetent human airway respiratory cell line, were infected with wild-type (WT) and dORF3-5 mutant MERS-CoV at a low multiplicity of infection (MOI) (Fig. 1A). While WT virus produced robust viral replication, the dORF3-5 mutant was significantly attenuated. In addition, infection of primary human airway epithelial (HAE) cultures indicated robust attenuation of the dORF3-5 mutant relative to control virus (Fig. 1B). To determine if attenuation is governed by a replication/fitness defect, the host response, or a combination of both, Vero cells were infected at a high MOI to examine single-step growth (see Fig. S1 in the supplemental material). Following infection, both WT and dORF3-5 mutant virus grew to equivalent titers 24 h postinfection. In contrast, Calu-3 cells infected at a high MOI demonstrated robust attenuation of the dORF3-5 mutant beginning at 12 h and continuing over the course of the infection (Fig. 1C). Combined with the Vero cell results, the data suggest that host response limits the dORF3-5 mutant and confirm a critical role for the MERS-CoV accessory ORFs in successful infection of human respiratory cells.

To examine changes in host responses, RNA expression was evaluated by microarray, with analysis revealing 5,046 differentially expressed genes between the dORF3-5 mutant and the WT. Using hierarchical clustering, differentially expressed genes were grouped into clusters based on their behaviors. This resulted in clusters of augmented and/or diminished genes, with the greatest changes observed 24 h postinfection (Fig. 1D). For clusters with increased expression in the mutant virus (clusters 1 to 3), functional categories were associated with disruption of normal cell host processes, including transcription regulation, protein transport, protein catabolism, chromatin modification, zinc binding, and more. Similarly, clusters generally associated with downregulation (Fig. 1D, clusters 4 to 6) include disruption of important cell function, including RNA splicing, protein complex assembly, and chromatin organization. We next performed a detailed enrichment analysis of gene changes between WT and mutant virus at 24 h postinfection. Differentially expressed genes were divided into several behavioral categories based on comparisons between mock-infected cells and cells infected with the wild type or mutant (e.g., DDU [“down up”] refers to a gene that is downregulated in the wild type compared to mock infection, downregulated in the mutant compared to mock infection, and upregulated in the mutant compared to the wild type). Using the DAVID online resource (https://david.ncifcrf.gov ), genes from each behavioral category were used for functional enrichment analysis, thus showing what pathways likely displayed each behavior (see Tables S1 and S2 in the supplemental material). To simplify the results, DAVID groups similar enrichment hits (pathways/processes) into clusters. Accordingly, in Tables S1 and S2 each cluster of enrichment hits is represented only by the highest-scoring member of that cluster. Based on the behavioral categories and functional enrichment results, it can be seen that the dORF3-5 mutant upregulates response to virus (9.71) and defense responses (3.55) beyond the upregulation of the wild-type and upregulates genes related to ubiquitin conjugation (11.67) and zinc finger pathways (30.62) that were not upregulated in the wild type. Together, the modeling results indicate that infection with the dORF3-5 mutant stimulates a strong host response that disrupts normal cell function and host immunity and likely influences the overall outcome of infection.

Based on strong attenuation and robust induction of host defense networks in Calu-3 cells, one possibility is that the dORF3-5 mutant induces a more robust interferon (IFN) response than WT virus. To examine IFN responses, RNA expression was evaluated following infection by microarray (Fig. 2A to C). The results reveal robust augmentation of IFN-β and IFN-λ in the mutant virus compared to the control. While responses were initially equal at early time points, the dORF3-5 mutant produces more IFN-β and IFN-λ transcript beginning at 12 h postinfection and increasing at 24 h; in contrast, WT MERS-CoV produces considerably less IFN-β and IFN-λ message. Notably, neither WT nor mutant virus induced increased expression of any IFN-α tested (data not shown). Extending our analysis, downstream interferon-stimulated genes (ISGs) were also examined. Using a previously established list of consensus Calu-3 ISGs (16), the dORF3-5 mutant was found to produce augmented expression of the vast majority, especially 24 h postinfection (Fig. 2D). While wild-type MERS-CoV expression produced subsets of augmented and downregulated genes relative to mock infection, the dORF3-5 mutant produced robust induction in the majority of ISGs. Extending these results, we examined the interferon sensitivity of the dORF3-5 mutant (Fig. 2E). While the dORF3-5 mutant has reduced replication relative to the control, type I IFN treatment does not ablate viral replication, suggesting the increased magnitude of ISGs rather than a specific ISG activity mediates dORF3-5 attenuation. Together, these results suggest that the absence of MERS accessory ORFs results in augmented stimulation of the IFN-β and IFN-λ response.

In addition to robust IFN response, differential expression (DE) between mutant and wild-type virus showed statistical enrichment of inflammatory genes in the dORF3-5 mutant. IL6, PTGS2, and TNF and several other inflammation regulatory genes were found central to the network of genes activated in the mutant virus (Fig. 3A). Examination of regulator expression (Fig. 3B, top panel) shows clear stimulation in the mutant that is absent in the wild-type MERS-CoV infection. Extending this analysis to NF-κB-dependent cytokines and chemokines, the RNA expression data provides a clear signature of augmented inflammation in the dORF3-5 mutant. To verify this augmented inflammation response, protein analysis was conducted on cell supernatants 12 and 24 h postinfection (Fig. 3C). The results show robust induction of inflammatory cytokines and chemokines at 12 h postinfection, including interleukin-1 receptor antagonist (IL-1Ra), IL-1β, RANTES, and macrophage inflammatory protein 1α (MIP-1α). Overall, the results indicate that the dORF3-5 mutant prompts a robust inflammatory response in addition to a strong interferon response.

Having demonstrated replication attenuation as well as stimulation of both IFN and inflammation responses in vitro, the dORF3-5 mutant was evaluated in vivo utilizing a newly developed mouse model for MERS-CoV (22). Using a clustered regularly interspaced short palindromic repeat (CRISPR)-generated mouse targeting positions 288 and 330 of Dpp4 (288-330^+/+), the dORF3-5 mutant replicated less efficiently in the lung, but only at day 4 postinfection (Fig. 4A). While no weight loss was observed in any of these models because of the wild-type genome backbone (22), the results confirm in vivo attenuation of the dORF3-5 mutant. Importantly, despite replication attenuation, the dORF3-5 mutant was also capable of protection from a lethal MERS-CoV challenge. Following vaccination with the wild type or dORF3-5 mutant or mock infection, 288-330^+/+ mice were challenged with a lethal dose of a mouse-adapted strain of MERS-CoV 15 (Fig. 4B and C) (22). While mock-infected mice experienced rapid weight loss through day 4, both wild-type- and dORF3-5-vaccinated mice were protected from weight loss (Fig. 4B) and viral replication (Fig. 4C). Notably, 50% plaque reduction neutralization test (PRNT₅₀) titers showed robust antibody neutralization with wild-type and dORF3-5 mutant vaccination (Fig. 4D), suggesting that replication attenuation has no significant impact on the generation of a protective immune response. Together, the data indicate that the MERS dORF3-5 mutant has some potential as a live attenuated vaccine candidate.

While replication attenuation was observed in the wild-type MERS-CoV backbone, it remained possible that the dORF3-5 mutant may not be sufficiently attenuated in a virulent backbone. To test this question, the dORF3-5 mutant was generated in the context of a mouse-adapted infectious clone. Derived from a plaque-purified isolate of MERS-CoV 15, the MERS-C2 clone had several mutations in its backbone, including point mutations in nonstructural protein 2 (NSP2), NSP6, and NSP8 as well as a large deletion of a portion of ORF4b (22). The removal of an accessory ORF in this and other mouse-adapted clones raised concerns that accessory ORF mutation may augment pathogenesis. However, following infection with the dORF3-5 mutant in the mouse-adapted backbone, 288-330^+/+ mice showed no pathogenesis in terms of weight loss (Fig. 5A). Similarly, viral replication of the mutant virus is attenuated at both days 2 and 4 postinfection (Fig. 5B). Notably, examination of the cytokine response in the lung indicated attenuated inflammatory responses following infection with the dORF3-5 mutant relative to the mouse-adapted control virus (Fig. 5C); the largest differences were noted at day 2, suggesting that dORF3-5 attenuation occurs early during in vivo infection. Together, the results illustrate clear attenuation of the dORF3-5 mutant in the context of a virulent backbone and signal its possible utility as a live attenuated vaccine platform.

Having fully established dORF3-5 attenuation in vitro and in vivo, we next examined the role of the individual ORFs in contribution to attenuation. As expected, the accessory ORFs have the greatest diversity even within the group (Fig. 6A). For 2C CoVs which include MERS-CoV, the greatest amino acid diversity is observed in ORF3, -4a, -4b, and -5, suggesting that evolutionary pressures have shaped proteins, likely to adapt to specific aspects of their host species. Several reports have highlighted both ORF4a and -4b as IFN antagonists as well as disrupting elements of the interferon-stimulated gene response (18, 23, 24). In contrast, little characterization of dORF5 has occurred in the context of MERS-CoV infection. With this in mind, we utilized a previously described MERS mutant that replaces ORF5 with red fluorescent protein (RFP) in order to characterize the host response in its absence. Following infection of primary HAE cells, the dORF5 mutant had a modest, but significant attenuation in replication relative to WT virus 24 and 48 h postinfection; by 72 h, the dORF5 mutant had nearly equivalent titers (Fig. 6B). Similar to Vero cells (20), the dORF5 mutant had no significant replication attenuation in Calu-3 cells following high-MOI infection (Fig. 6C). However, broad changes in the host response were observed in RNA expression (Fig. 6D); upon examination of expression relative to wild-type virus, the dORF5 mutant had more modest changes than the dORF3-5 mutant (Fig. 1). However, several pathways associated with apoptosis, NF-κB signaling, and serine/threonine kinase activity were augmented in the dORF5 mutant (clusters 5, 6, 7, and 10). Notably, the dORF5 mutant failed to induce robust type I and type III IFN responses, suggesting no specific role in IFN antagonism (Fig. 6E and F). In contrast, inflammatory cytokines demonstrated augmented production following dORF5 infection relative to the WT (Fig. 6G). While this stimulation pales compared to that of the dORF3-5 mutant, the data argue for a role in dORF5 in disrupting elements of NF-κB activation and the host inflammation response following MERS-CoV infection.

Wild-type and mutant MERS-CoVs were previously described (20) and cultured on Vero 81 cells, grown in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, CA) and 5% fetal clone serum (HyClone, South Logan, UT) along with an antibiotic/antimycotic (Gibco, Carlsbad, CA). Growth curves in Vero, Calu-3 2B4, and human airway epithelial (HAE) cells were performed as previously described (9, 30). Briefly, cells were washed with phosphate-buffered saline (PBS) and inoculated with virus or mock diluted in PBS for 40 min at 37°C. Following inoculation, cells were washed 3 times, and fresh medium was added to signify time zero. Samples were harvested at the described time point. For IFN pretreatments, 100 U/ml of recombinant human IFN-β (PBL Laboratories) was added to cells 16 h prior to inoculation, and cells were infected as described above. All virus cultivation was performed in a biosafety level 3 (BSL3) laboratory with redundant fans in biosafety cabinets as described previously by our group (31, 32). All personnel wore a 3M Breathe Easy powered air-purifying respirator with Tyvek suits, aprons, and booties and were double gloved.

Both wild-type and mutant viruses were derived from either the MERS-CoV wild type or the corresponding mouse-adapted (MA1 [herein referred to as WT]) infectious clone as previously described. Generation of the dORF3-5 and dORF5 mutants was previously described.

RNA isolation and microarray processing from Calu-3 cells were carried out as previously described (40). Differential expression (DE) was determined by comparing virus-infected replicates to time-matched mock replicates. Criteria for DE in determining the consensus ISG list were an absolute log₂ fold change of >1.5 and a false discovery rate (FDR)-adjusted P value of <0.05 for a given time point.

Genes identified as differentially expressed were used to generate clustered expression heat maps. Hierarchical clustering (using Euclidean distance and complete linkage clustering) was used to cluster gene expression according to behavior across experimental conditions. The DAVID online resource (https://david.ncifcrf.gov ) was used to acquire functional enrichment results for the genes in each cluster. DAVID output was manually summarized for each cluster. Plots were generated with R.

For the analysis in Tables S1 and S2, three comparisons were used to group genes into distinct behavioral categories: wild-type virus compared to mock infected, mutant virus compared to mock infected, and mutant virus compared to wild type, all at the 24-h time point. Genes were assigned a behavioral category based on whether they were up- or downregulated in each of these three comparisons. The genes in each category were then submitted to DAVID analysis, which resulted in clusters of functional enrichment hits grouped according to gene set overlap (accession no. 17784955). To simplify the results, each cluster is represented only by the highest-scoring enrichment hit in each cluster and scored according to the negative log₁₀P value of the significance of the enrichment for that top hit.

This study was carried out in accordance with the recommendations for care and use of animals by the Office of Laboratory Animal Welfare (OLAW), National Institutes of Health. The Institutional Animal Care and Use Committee of the University of North Carolina at Chapel Hill (IACUC, UNC [permit no. A-3410-01]) approved the animal study protocol (IACUC protocol no. 15-009 and 13-072) followed in this article.

Ten- to 20-week old CRISPER-generated 288-330^+/+ C57BL/6 mice were anesthetized with ketamine and xylazine (as per IACUC, UNC guidelines) and intranasally inoculated with a 50-µl volume containing 10⁶ PFU of MERS-CoV WT, SARS-CoV dORF3-5 virus, mouse-adapted variants, or the PBS mock infection control as indicated in the figure legends. Infected animals were monitored for weight loss, morbidity, and clinical signs of disease, and lung titers were determined as described previously (36). For vaccination experiments, 10- to 20-week-old 288-330^+/+ mice were infected with 10⁶ PFU of dORF3-5 as described above, monitored for clinical symptoms for 7 days, and then challenged 4 weeks postvaccination with 10⁶ PFU MERS-CoV MA1-passaged virus. Animal housing, care, and experimental protocols were in accordance with IACUC, UNC guidelines.

Raw microarray data for these studies were deposited in publicly available databases in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (41) and are accessible through GEO accession no. GSE65574 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE65574 ).