The N-Terminal Region of Middle East Respiratory Syndrome Coronavirus Accessory Protein 8b Is Essential for Enhanced Virulence of an Attenuated Murine Coronavirus

ABSTRACT Middle East respiratory syndrome coronavirus (MERS-CoV) is a beta coronavirus that emerged in 2012, causing severe pneumonia and renal failure. MERS-CoV encodes five accessory proteins. Some of them have been shown to interfere with host antiviral immune response. However, the roles of protein 8b in innate immunity and viral virulence was rarely studied. Here, we introduced individual MERS-CoV accessory protein genes into the genome of an attenuated murine coronavirus (Mouse hepatitis virus, MHV), respectively, and found accessory protein 8b could enhance viral replication in vivo and in vitro and increase the lethality of infected mice. RNA-seq analysis revealed that protein 8b could significantly inhibit type I interferon production (IFN-I) and innate immune response in mice infected with MHV expressing protein 8b. We also found that MERS-CoV protein 8b could initiate from multiple internal methionine sites and at least three protein variants were identified. Residues 1-23 of protein 8b was demonstrated to be responsible for increased virulence in vivo. In addition, the inhibitory effect on IFN-I of protein 8b might not contribute to its virulence enhancement as aa1-23 deletion did not affect IFN-I production in vitro and in vivo. Next, we also found that protein 8b was localized to the endoplasmic reticulum (ER)/Golgi membrane in infected cells, which was disrupted by C-terminal region aa 88-112 deletion. This study will provide new insight into the pathogenesis of MERS-CoV infection. IMPORTANCE Multiple coronaviruses (CoV) cause severe respiratory infections and become global public health threats such as SARS-CoV, MERS-CoV, and SARS-CoV-2. Each coronavirus contains different numbers of accessory proteins which show high variability among different CoVs. Accessory proteins are demonstrated to play essential roles in pathogenesis of CoVs. MERS-CoV contains 5 accessory proteins (protein 3, 4a, 4b, 5, 8b), and deletion of all four accessory proteins (protein 3, 4a, 4b, 5), significantly affects MERS-CoV replication and pathogenesis. However, whether ORF8b also regulates MERS-CoV infection is unknown. Here, we constructed mouse hepatitis virus (MHV) recombinant virus expressing MERS-CoV protein 8b and demonstrated protein 8b could significantly enhance the virulence of MHV, which is mediated by N-terminal domain of protein 8b. This study will shed light on the understanding of pathogenesis of MERS-CoV infection.

C oronaviruses are positive-stranded RNA viruses that have the largest genomes of the known RNA viruses, typically 27-32 kb. Until now, seven human coronaviruses have been identified, including human coronavirus OC43 (HCoV-OC43), HCoV-NL63, HCoV-229E, HCoV-HKU1, severe acute respiratory syndrome coronavirus (SARS-CoV), Middle East respiratory syndrome coronavirus (MERS-CoV), and the novel coronavirus (SARS-CoV-2). The former four CoVs generally cause mild upper-respiratory-tract infection (1)(2)(3)(4) and latter three could cause severe lower respiratory tract infections and even death (5). MERS-CoV was first identified in Saudi Arabia in 2012 (6), with the highest mortality rate among SARS-CoV, MERS-CoV, and SARS-CoV-2. By the end of June 2021, 2,574 confirmed cases and 886 deaths had been reported and the mortality rate was about 35% (7). Apart from nonstructure and structure proteins, each coronavirus encodes a set of accessory proteins that are interspersed between or within the structural proteins. The accessory proteins are various in numbers and functions and share lower sequence similarity in different coronaviruses. Generally, accessory proteins considered to be dispensable for viral replication; however, growing research showed these proteins play important roles in CoV pathogenesis (8)(9)(10).
MERS-CoV has five putative accessory proteins, including protein 3 (encoded by ORF3), protein 4a (encoded by ORF4a), protein 4b (encoded by ORF4b), protein 5 (encoded by ORF5), and protein 8b (encoded by ORF8b) (11). All four protein (protein 3, -4a, -4b, -5) deletions could attenuate MERS-CoV viral replication and augment host responses (8,9). Plasmid transfection-mediated accessory protein overexpression could activate the inflammatory response or antagonize type I IFN response by various pathways. Several studies have shown that protein 4a, protein 4b and protein 8b can interfere with innate antiviral and proinflammatory responses in vitro (7,10,(12)(13)(14)(15)(16). The protein 8b encoded by an internal ORF within the N gene via a ribosomal leaky scanning mechanism (11), which is similar to protein internal (I) in MHV, protein 9b in SARS-CoV, protein N2 in HKU1, and protein 9b in SARS-CoV-2 (17) SARS-CoV 9b can induce caspase-dependent apoptosis (18), or suppresses innate immunity by targeting mitochondria and the MAVS/TRAF3/TRAF6 signalosome (19). SARS-CoV-2 9b can inhibit IFN-I responses by targeting TOM70 (20,21), or suppress IFN-I/III production induced by RIG-I/MDA5-MAVS signaling (22). MERS-CoV protein 8b can antagonize nuclear factor kappa B (NF-k B) activation or IFN-b promoter activation by luciferase reporter assay (23,24). However, these results were generated from luciferase reporter assay and functional validation are required (25). Lok-Yin Roy Wong et al. found MERS-CoV protein 8b suppresses IFN-I expression by impeding the interaction of IKK« and HSP70 (7). However, since ORF8b is fully overlapping with nucleocapsid (N) protein gene, it would be difficult to generate ORF8b mutants in the context of MERS-CoV genome. Therefore, an alternative experimental approach is required to study the role of protein 8b in MERS-CoV infection. Mouse hepatitis virus (MHV) strain rJ2.2 is an attenuated neurotropic betacoronavirus that causes mild encephalitis (26). It contains 5 accessory genes, including NS2A, HE, ORF4, ORF5A an ORFI. Generally, accessory genes are unique in sequence and functions among different betacoronavirus (27). ORF4 in MHV was demonstrated to have no specific functions in MHV pathogenicity (28), which makes it a perfect alternative model for studying the roles of viral accessory proteins from the same genus, including SARS-CoV and MERS-CoV, by replacing this gene using a reverse genetic approach. No sequence homology between MERS-CoV accessory proteins and MHV viral proteins were found, suggesting it is unlikely that there will be functional redundancy between MHV and MERS-CoV accessory genes. Previously, Pewe and colleagues revealed that SARS-CoV protein 6 enhances viral virulence using the same MHV reverse genetic system (29).
Here, we constructed a set of recombinant MHV expressing individual MERS-CoV accessory proteins and demonstrated that accessory protein 8b could enhance viral replication in vivo and in vitro, as well as increase the mortality rate of infected mice. Protein 8b is 112 amino acids in length, with less than 30% similarity to any of structural or accessory proteins of MHV virus (by EMBOSS Needle tool). Protein 8b expression has also been detected in MERS-CoV infected cells using Western blot (WB). Further, we demonstrated that protein 8b could significantly inhibit type I interferon production (IFN-I) and innate immune response in mice infected with MHV expressing protein 8b. Protein 8b could initiate from multiple internal methionine sites and at least three protein variants were identified. Residues 1-23 of protein 8b was demonstrated to be responsible for increased virulence in vivo. We also found that protein 8b was localized to the endoplasmic reticulum (ER)/Golgi membrane in infected cells, which could be disrupted by C-terminal region deletion. This study will provide new insight into the pathogenesis of MERS-CoV infection.

RESULTS
Generation and screening of recombinant MHVs expressing MERS-CoV accessory proteins that had increased virulence. MHV is a well-characterized neurotropic mouse coronavirus and offers several advantages for CoV studies as describe in introduction (29). So, we inserted each of the MERS-CoV accessory proteins gene (ORF3, ORF4a, ORF4b, ORF5, and ORF8b) into ORF4 of rJ2.2 (an attenuated neurotropic variant of MHV), respectively ( Fig. 1A and B), as disruption of ORF4 of MHV does not affect its replication in cells nor cause acute encephalitis disease in mice (29). For subsequent detection, a Flag tag and an HA tag were inserted to the N-and C-end of each accessory protein, respectively. All insertions of the five proteins were confirmed by PCR and Western blot (WB) (Fig. 1C and D). Of note, multiple bands were observed in the lane of ORF8b in WB assay (Fig. 1D), which could result from multiple methionine sites in ORF8b sequence. We further characterized this phenomenon in Materials and Methods. To characterize recombinant virus replication, growth kinetics of these viruses were analyzed. As shown in Fig. 1E, all of the recombinant virus showed similar replication kinetics with slightly lower titers compared with wild type (wt) rJ2.2 virus. Next, C57BL/6 mice were challenged with 1000 PFU of each recombinant and WT rJ2.2 virus and clinical manifestations were monitored. As shown in Fig. 1F and G, rJ2.2.4a and rJ2.2.8b (rj2.2 expressing protein 4a or protein 8b) infections resulted in enhanced virulence in mice, but not recombinant viruses inserted with ORF3, ORF4b, and ORF5, as determined by more weight loss and greater mortality. Here, we chose rJ2.2.8b for further analysis since protein 4a has been extensively studied and several studies had reported that protein 4a significantly inhibits type I interferon (IFN-I) signaling by its dsRNA binding domain (14,30). We also confirmed that rJ2.2.8b did not change cell tropism in the infected brain, as both rJ2.2, rJ2.2.8b mainly infected oligodendrocytes (Fig. 2). Of note, the virus titers showed no significant difference between rJ2.2 and rJ2.2.8b in infected mouse brains (Fig. 2E).
Protein 8b enhanced virus replication and increased viral protein accumulation in vitro. In order to control the influence of exogenous sequence insertion into MHV genome, we developed another recombinant virus in which a mutated start codon of ORF8b together with an additional stop codon introduced at residue 25. This virus was designated rJ2.2.8b-KO. The KO mutation was confirmed by WB (Fig. 3A). Of note, multiple bands were frequently observed in the lane of rJ2.2.8b in WB assay when using anti-HA antibody (C-terminal) (Fig. 3A), but never found using Flag antibody (N-terminal) (Fig. 1D), which could result from multiple methionine sites in ORF8b sequence. We further characterized this phenomenon in Fig. 6. Previous studies had shown that MERS-CoV accessory proteins have different subcellular localization (10). Immunofluorescence analysis (IFA) revealed that the protein 8b was colocalized with endoplasmic reticulum (ER) and Golgi compartments in infected 17CL-1 cells (Fig. 3B).
Protein 8b increased morbidity and mortality of rJ2.2.8b infected mice. Consistent with Fig. 1F and G, mice infected with rJ2.2.8b lost more body weight and exhibited greater morbidity as determined by clinical scores than rJ2.2.8b-KO infected mice ( Fig. 4A and B). More than 60% of mice died by day 17 post rJ2.2.8b infection. In contrast, all mice infected with rJ2.2.8b-KO survived until 21 days. p.i. (Fig. 4C). Consistent with these results, viral titers were significantly higher in the brains of rJ2.2.8b infected mice than those of rJ2.2.8b-KO at 3 days p.i. (Fig. 4D). These results demonstrated that protein 8b from MERS-CoV could significantly enhance the virulence in MHV infected mice.
Protein 8b suppressed innate immune response after rJ2.2.8b infection in vivo and in vitro. As significant differences in viral titers were observed at 3 days. p.i., but not at following time points (Fig. 4D), we proposed that the protein 8b might affect innate immune pathways at early stage of infection. RNA-seq was performed using brain samples harvested at 3 days p.i. Differentially expressed genes (DEGs) in the brain tissues of mice infected with rJ2.2.8b or rJ2.2.8b-KO were compared. We found 645 DEGs containing 202 upregulated and 443 downregulated genes in rJ2.2.8b groups compared with rJ2.2.8b-KO groups (Fig. 5A). Most downregulated genes were associated with innate immune pathways and inflammation pathways, including interferon stimulated genes (ISGs) and inflammatory cytokines in rJ2.2.8b infected mice ( Fig. 5B and C). Several key genes were validated by real-time RT-PCR analysis. As shown in Fig. 5D, IFN-I, ISG15, TNF, and IL-6 were significantly downregulated in rJ2.2.8b infected mice, indicating protein 8b efficiently inhibited antiviral innate response. Further, rJ2.2.8b infected bone marrow-derived macrophages (BMM) also produced less IFN-I in vitro as determined by real-time RT-PCR, which were consistent with RNA-seq data in vivo (Fig. 5E). A previous study showed that fibroblasts cells did not produce IFN-I after MHV infection (31). We next examined whether poly(IÁC) treatment could reverse this effect in rJ2.2.8b infected cells. The inhibitory effect of protein 8b on poly(IÁC) induced IFN-I production was analyzed in L929 cells. Cells were infected with rJ2.2.8b or rJ2.2.8b-KO for 12 h, and then transfected with poly(IÁC) for 4 h. Consistent with previous report, rJ2.2 infection alone did not induce significant IFN-I production (Fig. 5F). Although poly(IÁC) treatment resulted in higher IFN-I production in both  (Fig. 5F). Together, these results suggested that protein 8b inhibited innate immune response after MHV infection. The N-terminal region of protein 8b was essential for enhanced viral virulence. As shown in Fig. 1D and Fig. 2A, protein 8b potentially encodes multiple variants since it contains three methionines in its sequence (Fig. 6A). To determine which variant is responsible for enhanced virulence, several recombinant rJ2.2 (rJ2.2.8b D1-8 and rJ2.2.8b D1-23 with a Flag-tag at the N-end and an HA-tag at C-end of protein 8b), as well as eukaryotic expressing plasmids (without any tags which reflected the real sizes of protein 8b variants) inserted with truncated ORF8b were generated, where methionnines were replaced with alanines (Fig. 6A). Protein sizes of truncated protein 8b were  analyzed by WB using rabbit polyclonal serum against protein 8b (Fig. 6B). First, multiple variants of protein 8b were confirmed using authentic MERS-CoV infected cell lysate. The protein size from cells transfected with full-length protein 8b was identical with the protein 8b band with the highest molecular weight in MERS-CoV infected cells. While protein 8b bands from 8b D1-8 and 8b D1-23 transfected cells were consistent with the two lower molecular weight bands, indicating protein 8b could initiate from internal methionines and encode three truncated protein 8b (Fig. 6B).
To further determine the role of these protein 8b variants and deletions on virulence,17Cl-1 cells were infected with truncated rJ2.2.8b viruses, respectively. Virus titers were determined and compared. All recombinant viruses exhibited similar growth kinetics. The titer of rJ2.2.8b D1-23 was slightly higher than the other viruses (Fig. 6G). To determine the role of protein 8b truncations in vivo, mice were infected with these mutant recombinant viruses and weight loss and survival rates were monitored. As shown in Fig. 6H and I, mice infected with rJ2.2.8b D1-8, rJ2.2.8b D88-112, and rJ2.2.8b displayed similar weight loss with 60-80% mortality in mice. However, the mice infected with rJ2.2.8b D1-23 were all survived (Fig. 6I). Viral titers in the brains of infected mice were measured at 3 days. p.i. As shown in Fig. 6J, no significant difference was observed between rJ2.2.8b and rJ2.2.8b D1-23, although rJ2.2.8b D1-23 infection showed slightly lower viral replication in the brain. These results indicated that the aa 1-23 region of protein 8b was essential for enhanced virulence in the attenuated murine coronavirus. To determine whether aa 1-23 deletion impairs the inhibitory function of protein 8b on IFN-I production, IFN-I mRNA levels in the brains of mice infected with rJ2.2.8b variants were analyzed. No significant difference of IFN-I mRNA levels was detected, suggesting that aa 1-23 domain was not responsible for the inhibition of IFN-I production of protein 8b (Fig. 6K).

DISCUSSION
MERS-CoV could led to acute pneumonia with renal failure (6) and also has the highest fatality rate among all human respiratory coronaviruses according to WHO. MERS-CoV genome encodes a variety of accessory proteins which has no significance sequence homology even among closely related CoVs. The role of accessory proteins on the innate immune response has been investigated in some coronaviruses. Most of them play important roles in inhibition of IFN-I response or facilitate the inflammatory response. Previous studies showed that the MERS-CoV accessory protein 4a can suppress IFN-I production (13) and inhibit PKR-mediated antiviral response (14). Protein 4b inhibits NF-k B-dependent innate immune response (12), protein 8b inhibits NF-k B signaling (24) or type I IFN expression (7,24). However, most of these studies were performed by eukaryotic plasmid overexpression. Some recent studies utilizing recombinant MERS-CoV have elucidated the functions of some of these proteins, such as protein 4a, protein 4b, and protein 5 (32)(33)(34). Although MERS-CoV protein 8b suppresses IFN-I expression in MERS-CoV-D8b (protein 8b knockout) infected cells (18), little is known about the role of protein 8b on viral virulence in vivo.
When absence of all four accessory proteins (protein 3, -4a, -4b, and -5), MERS-CoV attenuated in mice (8), indicating accessory proteins might play important roles on virulence enhancement. However, only deletion of protein 5 increase the virulence of a mouse-adapted MERS-CoV strain (33). In order to reduce the functional redundancy or compensatory between different proteins on the virulence of the virus, we constructed series of recombinant rJ.2.2 viruses inserted with only one MERS-CoV accessory protein.
In order to ensure the transcription of ORFs, an MHV traditional TRS (transcription regulating sequence) sequence was added before the ORFs. Our previous study also found the insertion of exogenous genes could slightly reduce rJ2.2 replication, so we con-  (15). However, in this study, no weight change and survival rate of mice infected with MHV-ORF 4b recombinant virus were provided. Based on our results in Fig. 1, protein 4b increased rMHV virulence as determined by increased mortality rate after infection. Another studies described that MERS-CoV protein 4a increases recombinant encephalomyocarditis virus (EMCV) replication efficiency by suppressing the PKR-dependent stress response pathway (14), which is consistent with our results. All of the mice infected with rJ2.2.4a succumbed to the infection (Fig.  1). In this study, we demonstrated that two MERS-CoV accessory proteins, protein 8b could enhance viral virulence in MHV infected mice as determined by increased weight loss and mortality rate. In vitro growth curve showed that rJ.2.2.8b infection could produce more progeny virus at the early phase of infection and significant differences in titers in mouse brains were also observed in the early time of infection. We proposed that the protein 8b might enhance virulence in mice by affecting innate immune pathways. RNA-seq analysis showed that the innate immune signaling pathway was suppressed in the brains of mice infected with rJ2.2.8b. In vitro experiment was also confirmed the role of protein 8b on the innate immune response. However, subsequent experiments showed that rJ2.2 virus with aa 1-23 deletion had reduced mortality rate while did not affect IFN-I production in mouse brains. indicating although N-terminal domain is responsible for increased viral virulence, protein 8b potentially inhibits IFN-I response through its internal domains, which requires further investigation.
Our result showed that MRES-CoV protein 8b colocalized with ER and Golgi compartment. SARS-CoV accessory protein 6 which has the same cellular localization had been reported to be an amphipathic protein and its hydrophobic N-terminal region is sufficient to enhance virus replication (35). In this study, we found the ahelix C-region (aa 88-112) of protein 8b is critical for its subcellular location. However, the intracellular membrane location of protein 8b was not associated with virulence enhancement since deletion of C-terminus domain did not affect mouse survival after rJ2.2 infection. Interestingly, the distribution of Golgi and ER seem diffused in rJ2.2.8b D 88-112 infected cells which need further study.
In SARS-CoV, accessory protein 6 suppresses IFN-I production by interacting with karyopherin a2 via its cytoplasmic C-terminal region and inhibits STAT1 nuclear translocation (36). However, MERS-CoV accessory protein 8b did not change STAT1 translocation in 17CL-1 cells and BMMs (data not shown), suggesting different pathways used for IFN-I suppression between MERS-CoV protein 8b and SARS-CoV protein 6.
Unexpectedly, three truncated protein 8b proteins were detected in rJ2.2.8b infected cells, which could potentially reflect that the initiation of protein 8b protein translation could occur at different ATG codons on the 59 end of the ORF8b mRNA. protein 8b might mainly initiated at the first two ATGs since more abundant protein bands were detected in rJ2.28b and MERS-CoV infected cells. The full-length protein 8b and truncated protein 8b with aa 1-8 deletion were responsible for the enhanced virulence in vivo, while protein 8b with 1-23 AA deletion expressed at minimal level did not cause lethal disease in MHV infected mice.
In summary, protein 8b was demonstrated to increase viral replication in vivo and in vitro and enhance virulence in vivo. protein 8b also inhibited IFN-I production, but the IFN-I antagonism effect did not contribute to its virulence enhancement in vivo. Similar to SARS-CoV protein 6 (37), MERS-CoV protein 8b had a modest effect on virus replication, while its expression in virus-infected cells resulted in greatly enhanced mortality in mice, suggesting a potentially important role when expressed in authentic MERS-CoV infected hosts.