ABSTRACT
Middle East respiratory syndrome coronavirus (MERS-CoV) causes a highly lethal pneumonia that emerged in 2012. There is limited information on MERS-CoV pathogenesis, as data from patients are scarce and the generation of animal models reproducing MERS clinical manifestations has been challenging. Human dipeptidyl peptidase 4 knock-in (hDPP4-KI) mice and a mouse-adapted MERS-CoV strain (MERSMA-6-1-2) were recently described. hDPP4-KI mice infected with MERSMA-6-1-2 show pathological signs of respiratory disease, high viral titers in the lung, and death. In this work, a mouse-adapted MERS-CoV infectious cDNA was engineered by introducing nonsynonymous mutations contained in the MERSMA-6-1-2 genome into a MERS-CoV infectious cDNA, leading to a recombinant mouse-adapted virus (rMERS-MA) that was virulent in hDDP4-KI mice. MERS-CoV adaptation to cell culture or mouse lungs led to mutations and deletions in genus-specific gene 5 that prevented full-length protein expression. In contrast, analysis of 476 MERS-CoV field isolates showed that gene 5 is highly stable in vivo in both humans and camels. To study the role of protein 5, two additional viruses were engineered expressing a full-length gene 5 (rMERS-MA-5FL) or containing a complete gene 5 deletion (rMERS-MA-Δ5). rMERS-MA-5FL virus was unstable, as deletions appeared during passage in different tissue culture cells, highlighting MERS-CoV instability. The virulence of rMERS-MA-Δ5 was analyzed in a sublethal hDPP4-KI mouse model. Unexpectedly, all mice died after infection with rMERS-MA-Δ5, in contrast to those infected with the parental virus, which contains a 17-nucleotide (nt) deletion and a stop codon in protein 5 at position 108. Expression of interferon and proinflammatory cytokines was delayed and dysregulated in the lungs of rMERS-MA-Δ5-infected mice. Overall, these data indicated that the rMERS-MA-Δ5 virus was more virulent than the parental one and suggest that the residual gene 5 sequence present in the mouse-adapted parental virus had a function in ameliorating severe MERS-CoV pathogenesis.
IMPORTANCE Middle East respiratory syndrome coronavirus (MERS-CoV) is a zoonotic virus causing human infections with high mortality rate (∼35%). Animal models together with reverse-genetics systems are essential to understand MERS-CoV pathogenesis. We developed a reverse-genetics system for a mouse-adapted MERS-CoV that reproduces the virus behavior observed in humans. This system is highly useful to investigate the role of specific viral genes in pathogenesis. In addition, we described a virus lacking gene 5 expression that is more virulent than the parental one. The data provide novel functions in IFN modulation for gene 5 in the context of viral infection and will help to develop novel antiviral strategies.
INTRODUCTION
Middle East respiratory syndrome coronavirus (MERS-CoV) is a life-threatening human virus that emerged during the summer of 2012 in Saudi Arabia (1–3). As of 8 June 2020, a total of 2,494 cases have been confirmed, distributed over 27 countries, causing 858 deaths (fatality rate of ∼35%) (http://www.who.int/emergencies/mers-cov/en/). The high case fatality rate, continuous MERS-CoV outbreaks, and risk of virus adaptation potentially resulting in pandemic spread make research on MERS-CoV an international priority (see the WHO February 2018 document Annual Review of Diseases Prioritized under the Research and Development Blueprint [https://www.who.int/emergencies/diseases/2018prioritization-report.pdf]) (4). MERS-CoV has a zoonotic origin, with camels as the intermediate host transmitting the virus to humans (5, 6). In fact, MERS-CoV is endemic in camels, as shown by the high seroprevalence observed in camels from the Middle East and Africa (7–9). In general, MERS-CoV infection of healthy individuals results in subclinical or mild disease. In contrast, in patients with comorbidities, MERS-CoV infection often causes acute respiratory distress syndrome and multiorgan failure leading to death (6). To date, there is limited information on MERS-CoV pathogenesis, as data from patient autopsies are scarce and most animal models do not reproduce MERS clinical manifestations (10). Specifically, knowledge of viral genes involved in virulence and of the host cell pathways modified by the infection is required to understand MERS-CoV pathogenesis. To that end, reverse-genetics systems facilitate modification of the viral genome for analysis of both virulence factors and virus-host interactions. In addition, this understanding will accelerate the development of novel vaccine candidates based on attenuated viruses and the identification of novel antiviral drug targets.
MERS-CoV is classified within the subgenus Merbecovirus of genus Betacoronavirus (11, 12). The positive-sense single-stranded RNA genome of MERS-CoV is approximately 30 kb and contains 11 open reading frames (ORFs) in the order 5′-ORF1a, ORF1b, S, 3, 4a, 4b, 5, E, M, N, 8b-3′ which are expressed from a nested set of eight mRNAs (13, 14). Genus-specific genes 3, 4a, 4b, and 5 are nonessential for virus replication (15, 16). In overexpression analyses, these genes have been involved in the modulation of virus-host interaction (17–21), although there is limited information on the role of genus-specific genes in the context of viral infection (16, 22–24). We previously engineered a MERS-CoV infectious cDNA clone maintained as a bacterial artificial chromosome (BAC) (15), enabling the study of the role of viral proteins in modulating the innate immune response (22, 23).
Human dipeptidyl peptidase 4 (hDPP4) is the receptor for MERS-CoV (25), but mouse DPP4 (mDPP4) does not sensitize cells for infection (26). To overcome this problem, several transgenic or knock-in (KI) mouse models have been generated (27–34). Mice transgenically expressing hDPP4 are susceptible to MERS-CoV, but in general, they develop a lethal encephalitis (27, 29, 30, 33). Recently, we developed hDPP4-KI mice by replacing three mDPP4 exons with their human counterparts. hDPP4-KI mice were susceptible to MERS-CoV, with virus replication restricted to the lung, although no clinical signs were observed (30). To further improve the model, a mouse-adapted MERS-CoV strain (MERSMA-6-1-2) was obtained by passaging the virus 30 times in hDPP4-KI mice. Infection with MERSMA-6-1-2 virus resulted in a lethal pneumonia (30).
MERSMA-6-1-2 virus contained mutations at several locations in the genome. Specifically, it contained a stop codon and a deletion in genus-specific gene 5, abrogating expression of full-length protein. This MERS-CoV protein is homologous to SARS-CoV 3a, HCoV-NL63 protein 3, or HCoV-229E protein 4 (35, 36). Although they differ in sequence, their predicted structures are quite similar and contain three transmembrane domains. Protein 5 accumulates at the ERGIC (endoplasmic reticulum-Golgi intermediate compartment) when overexpressed (19, 20), similar to proteins 3 and 4 from HCoV-NL63 and -229E, respectively (37, 38). Most CoV proteins similar to MERS-CoV protein 5 have ion channel activity (38–40), are involved in morphogenesis (38, 41), and, in some cases, are incorporated into the viral envelope (37, 42). These proteins, although not essential for virus growth in cell culture, modulate virus-host interactions and pathogenesis (39, 41, 43).
Here, we engineered a mouse-adapted MERS-CoV infectious cDNA (rMERS-MA), based on the MERSMA-6-1-2 virus sequence and showed that the resulting virus was virulent in hDPP4-KI mice. A recombinant virus lacking full-length gene 5 sequence (rMERS-MA-Δ5) was also engineered and shown to be more virulent than the parental rMERS-MA virus, suggesting that it expressed a gene 5 fragment. To further characterize protein 5 functions, a normal lung cell system, in which increased MERS-CoV stability was observed, was used. An rMERS-CoV-Δ5 virus lacking protein 5 expression induced a decreased innate immune response compared with the parental rMERS-CoV virus expressing full-length protein 5. Moreover, in a transgenic mouse model system, rMERS-CoV-Δ5 virus was more virulent than the parental rMERS-CoV. Altogether, our data indicated that protein 5 had an innate immune modulatory function required to prevent enhanced lethality in two different mouse model systems.
RESULTS
Engineering of a mouse-adapted infectious cDNA clone for MERS-CoV.A MERS-CoV mouse-adapted virus was selected after 30 passages in hDPP4-KI mice and cloning in vitro (MERSMA-6-1-2) (30). The genome of this virus was sequenced, and 17 mutations were found compared with the reference MERS-CoV sequence (strain EMC/2012 [GenBank accession no JX869059]) (Fig. 1A). Three of them were silent mutations and were not further analyzed. The rest of the mutations introduced amino acid changes in several viral proteins, and mutations in the 5′ and 3′ untranslated regions (UTRs) (Fig. 1A). It is worth noting that the mutation in nsp3 and the premature stop codon in gene 5 were already present in the inoculum virus grown in Vero cells before passages were initiated in mice (30). In addition, with passage in mice, a small 17-nucleotide (nt) deletion appeared in genus-specific gene 5, shifting the translation frame of protein 5 (Fig. 1A). Both gene 5 mutations together abrogated the expression of full-length protein 5.
Engineering of a mouse-adapted MERS-CoV infectious cDNA. (A) Diagram of mutations present in the mouse-adapted MERSMA-6-1-2 genome (30). The letters above each box indicate viral genes. UTR, untranslated region. Vertical bars in the boxes indicate mutations: synonymous mutations in black, mutations in the UTR in gray, or missense mutations in gray. The white box in gene 5 indicates a small 17-nt deletion, and the asterisk indicates the G27162A mutation that created a premature stop codon. The boxes below the genome indicate the amino acid changes caused by the missense mutations. (B) The upper schematic represents the cDNA of the mouse-adapted MERS-CoV genome, as shown in panel A, flanked by cytomegalovirus promoter (CMV) and the hepatitis delta virus ribozyme (Rz) and bovine growth hormone termination sequence (BGH). pA, poly(A) tail. The bar below represents the six fragments (F1 to F6, in dark gray) originally designed to assemble MERS-CoV infectious cDNA (15), flanked by the indicated restriction sites (positions in the viral genome indicated by numbers in parentheses). Boxes in light gray indicate the chemically synthesized mouse-adapted fragments. In the case of the PacI-SanDI fragment, three different DNA sequences were used, depending on the final virus to be obtained: MERS-MA, for the sequence similar to that of MERSMA-6-1-2 virus; MERS-MA-5FL, containing full-length gene 5 sequence; and MERS-MA-Δ5, lacking gene 5 sequence.
The mutations identified in MERSMA-6-1-2 virus were introduced into the corresponding intermediate cDNA fragments and plasmids, which were assembled to engineer a mouse-adapted infectious cDNA (pBAC-MERSFL-MA) (Fig. 1B) (15). Recombinant mouse-adapted MERS-CoV (rMERS-MA) was recovered after transfection of Huh-7 cells with the infectious cDNA. The rMERS-MA virus titer was 3.8 × 106 PFU/ml, similar to that of MERSMA-6-1-2 virus.
Virulence of rMERS-MA in the hDPP4-KI mice model.Sixteen-week-old KI mice were mock infected or infected with the recombinant and nonrecombinant viruses, and clinical signs were monitored daily. Mock-infected mice lost no weight, and all survived (Fig. 2). In contrast, animals infected with rMERS-MA virus rapidly lost weight, similar to the positive-control MERSMA-6-1-2 virus (Fig. 2A). All mice infected with rMERS-MA virus died by day 8 postinfection (Fig. 2B). These results indicated that rMERS-MA was as virulent as parental MERSMA-6-1-2 virus.
Virulence of rMERS-MA in hDPP4-KI mice. Sixteen-week-old hDPP4-KI mice were intranasally mock infected (Mock [empty squares]) or infected with 104 PFU/animal of MERSMA-6-1-2 control virus (gray circles) or rMERS-MA recovered from infectious cDNA (black squares). Weight loss (A) and survival (B) are indicated. The values represent the mean from five animals per group. Error bars indicate the SEM for each value.
Stability of MERS-CoV gene 5 in vivo.It was previously observed that MERS-CoV genus-specific genes 3, 4a, 4b, and especially 5 tended to accumulate mutations and deletions when virus was grown in cultured cells (Table 1) (14, 15, 44). Consistent with these results, mutations were also observed after MERS-CoV mouse adaptation, with selection of deletions in gene 5 (Table 1) (28, 30, 45).
ORF5 variation in different MERS-CoVsa
Deletions in genes 3, 4a, and 4b have also been reported in different field isolates (46, 47). To assess whether gene 5 mutations also occurred in human and camel isolates, the 476 MERS-CoV GenBank sequences containing gene 5 (data collected on 20 June 2018) were aligned. Interestingly, 99.5% of the sequences (474 out of 476), both from camels (54%) and from humans (44%), contained full-length gene 5 sequence. In these sequences, 60 mutations were identified: 65% were silent, and 35% were nonsynonymous. Two MERS-CoV isolates contained small deletions in gene 5 (Table 1): (i) strain D383/15, isolated from a camel, (GenBank accession no. KX108941) contains a 2-nt deletion located 19 nt downstream of the start codon, resulting in nearly complete absence of full-length protein 5 expression, and (ii) human isolate Hu/Jeddah-KSA-3RS2702/2015 (GenBank accession no. KU851859) contains a 19-nt deletion 368 nt downstream of the ATG codon, resulting in a truncated protein 5.
Thus, genus-specific gene 5 was more stable during the evolution of field isolates than genes 3, 4a, and 4b, which is at variance with the sequence modifications observed after MERS-CoV adaptation to mice and cell culture. These observations prompted us to analyze whether protein 5 modulates MERS-CoV pathogenesis.
Recovery of mouse-adapted rMERS viruses with differences in gene 5.As indicated above, MERSMA-6-1-2 virus was obtained after 30 mouse lung passages of a virus that originally contained a stop codon in gene 5 (Fig. 1A). Moreover, MERSMA-6-1-2 and rMERS-MA viruses also contained a small deletion in gene 5, acquired in mice (Fig. 1A) (30). Both modifications prevented the translation of full-length protein 5, although transcription of subgenomic mRNA of protein 5 (sgmRNA-5) was unaffected (Fig. 3A). To further investigate the role of protein 5 in MERS-CoV replication and pathogenesis, two additional viruses were engineered: rMERS-MA-5FL, containing full-length gene 5, and rMERS-MA-Δ5, lacking gene 5 sequence, including the transcription start site (Fig. 3B). The rMERS-MA-Δ5 virus titer (2.9 × 106 PFU/ml) was similar to that of the parental virus in cell cultures (Fig. 3B). In contrast, the rMERS-MA-5FL virus titer was 5-fold lower than that of the parental virus (Fig. 3B). The sequence of the genome containing all accessory genes and E protein was determined after PCR amplification of the viral genome. Both parental rMERS-MA and mutant rMERS-MA-Δ5 viruses led to expected bands (Fig. 3C) and included just the engineered modifications (data not shown). In contrast, when rMERS-MA-5FL sequence was analyzed, no changes were detected in genes 3 or 4a (data not shown), whereas genetic instability was detected in the region including genes 4b, 5, E, and M, as bands with a lower size than expected were detected (Fig. 3C). Sequencing of the reverse transcription-PCR (RT-PCR) products showed that gene 5 was completely deleted at early passages (data not shown), in agreement with previous observation that genus-specific ORFs 3, 4a, 4b, and especially gene 5 tend to accumulate point mutations and deletions when the virus was grown in cell culture (Table 1) (14, 15, 44).
Engineering of mouse-adapted rMERS-CoVs expressing different gene 5 sequences. (A) Analysis of sgmRNA-5 accumulation. Huh-7 cells were mock infected or infected with either rMERS-CoV (human) or mouse-adapted rMERS-MA (MA) virus. Total RNA was extracted at 18 h postinfection (hpi), and the presence of sgmRNA-5 was analyzed by RT-PCR. (B) Design of rMERS-MA viruses containing different gene 5 sequences. The upper bar represents the MERS-CoV genome. Three different viruses were engineered, expressing the same gene 5 sequence as MERSMA-6-1-2 virus (rMERS-MA), full-length gene 5 sequence (rMERS-MA-5FL), or lacking the gene 5 sequence (rMERS-MA-Δ5). The dark gray vertical bar represents the mouse-adapted mutation in the 4b gene. The asterisk indicates the G27162A mutation creating a premature stop codon, and the white rectangle represents the 17-nt small sequence deletion in gene 5. Letters above the bar indicate the viral genes. UTR, untranslated region. Viral titers obtained in cell cultures are indicated to the right. (C) Specific oligonucleotide pairs were used for overlapping PCRs (PCR1 and PCR2) covering the MERS-CoV genome region from the end of the S gene to the M gene, as indicated in the upper schematic. Analysis of passage 0 (p0) or passage 1 (p1) of rMERS-CoVs in Huh-7 cells is shown in the panel below. Arrows indicate the expected PCR product sizes: PCR1 in black and PCR2 in gray. Molecular size in base pairs is indicated to the left. Mock, noninfected cells; MA KI P30, mouse-adapted MERSMA-6-1-2 virus; WT, rMERS-MA from infectious cDNA; 5FL, rMERS-MA-5FL; Δ5, rMERS-MA-Δ5. Similar results were obtained by independent transfection of two infectious cDNAs per virus.
rMERS-MA-5FL genetic instability precluded further analysis. Consequently, for in vivo analysis, we chose to study parental rMERS-MA and mutant rMERS-MA-Δ5 viruses. None of them expressed the full-length protein 5, although the parental virus contains most of the gene 5 sequence, which was absent in rMERS-MA-Δ5 virus.
Virulence of rMERS-MA-Δ5 in hDPP4-KI mice.To analyze the role of protein 5 in pathogenesis, we infected 38-week-old hDPP4-KI mice with a sublethal dose of virus. The 38-week-old mice were used because our preliminary results showed that these mice were more resistant to infection than 16-week-old mice (data not shown), which are used in most studies (30).
Both rMERS-MA and rMERS-MA-Δ5 viruses replicated to a similar extent in the lungs of infected hDPP4-KI mice, as shown by the accumulation of viral gRNA (Fig. 4A) or virus titration (Fig. 4B). Compared with mock-infected mice, all animals infected with rMERS-MA or rMERS-MA-Δ5 viruses lost weight (Fig. 4C). In agreement with our previous observations (data not shown), 80% of the animals infected with parental rMERS-MA survived. In contrast, all animals infected with rMERS-MA-Δ5 mutant virus died by 10 days postinfection (dpi) (Fig. 4D). In line with this result, when hDPP4-KI mice were infected with different doses of rMERS-MA and rMERS-MA-Δ5 viruses, increases in weight loss (Fig. 5A) and animal death (Fig. 5B) were observed for the animals infected with rMERS-MA-Δ5 virus, compared with those infected with the parental rMERS-MA virus. These data strongly suggested that the complete absence of gene 5 sequence in rMERS-MA-Δ5 mutant, present in parental rMERS-MA virus, had a role in modulating virulence.
Growth and virulence of mouse-adapted rMERS-CoVs in hDPP4-KI mouse lungs. Thirty-eight-week-old KI mice were intranasally mock infected (Mock [gray]) or infected with 105 PFU/animal of recombinant viruses recovered from infectious cDNA. Mice were inoculated with parental rMERS-MA (black) or mutant rMERS-MA-Δ5 (white) virus. Lung samples were obtained at the indicated time postinfection, and viral gRNA accumulation (A) and virus titers (B) were determined (n = 3 mice per time point). Weight (C) and survival (D) were monitored. The values represent means from 10 animals per group. Error bars indicate the SEM for each value.
Mouse-adapted rMERS-CoV’s dose-dependent effect on virulence in hDPP4-KI mice. Thirty-eight-week-old KI mice were intranasally mock infected (Mock [black]) or infected with 104, 103, or 102 PFU/animal of recombinant viruses recovered from infectious cDNA. Mice were inoculated with parental rMERS-MA (black symbols) or mutant rMERS-MA-Δ5 (open symbols) viruses. Weight (A) and survival (B) were monitored. The values represent means from five animals per group. Error bars indicate the SEM for each value.
MERSMA-6-1-2 virus virulence in hDPP4-KI mice is associated with lung injury, including inflammatory cell infiltration, diffuse alveolar damage, edema, and hyaline membrane formation (30). Histopathological examination showed no damage in noninfected animals (Fig. 6A, upper panels). Mice infected with rMERS-MA virus showed extensive lung injury with cell infiltration and edema at 4 dpi, in agreement with previous data (30) (Fig. 6A, middle panels). By 6 dpi, cell infiltration produced by parental rMERS-MA virus was reduced (Fig. 6A, middle panels), suggesting recovery from the infection. In contrast, lung injury was significantly delayed in mice infected with rMERS-MA-Δ5 mutant virus, as at 4 dpi edema was only observed in some areas covering around 40% of the lung tissue, with limited inflammatory infiltration (Fig. 6A, bottom panels). Extensive lung damage, including massive inflammatory cell infiltration, hemorrhage, and edema, was observed in rMERS-MA-Δ5-infected mice at 6 dpi (Fig. 6A, bottom panels). Histopathological quantification showed, in rMERS-MA-Δ5-infected lungs, significantly reduced damage at 4 dpi and increased damage at 6 dpi, compared with parental virus-infected mice (Fig. 6B). Altogether, these data suggested that lung pathology caused by rMERS-MA-Δ5 mutant virus was delayed and ultimately was more severe than that caused by rMERS-MA virus.
Lung damage in rMERS-MA-Δ5-infected hDPP4-KI mice. Lung samples, collected at 4 and 6 days postinfection, were stained with hematoxylin-eosin. (A) Representative sections are shown. Pictures were obtained with a 10× objective. (B) Lung damage score, calculated from the observation of 50 microscopy fields per animal. Values represent the mean of the score, and error bars indicate the SEM for each value. ***, P < 0.001.
The pathology produced by respiratory CoVs, and specifically by MERS-CoV, has been associated with an exaggerated proinflammatory innate immune response (6, 48–51). The observed results led us to hypothesize that the complete absence of gene 5 sequences altered the innate immune responses triggered by the infection. To analyze this issue, we analyzed lung expression of several genes involved in the innate immune response, including those coding for interferon beta (IFN-β), IFN-γ, cytokines such as interleukin-6 (IL-6), IL-10, tumor necrosis factor (TNF), CCL2, CCL4, and CXCL10, and IFN-stimulated genes (ISGs), such as those coding for ISG15, MX1, and DDX58. Interestingly, a moderate but significant decrease in the levels of IFN-β, IL-10, CCL4, CCL2, CXCL10, TNF, and ISGs was observed at 4 dpi in the lungs of mice infected with rMERS-MA-Δ5 compared to rMERS-MA virus (Fig. 7A [data not shown]), in agreement with the delayed lung damage found in the rMERS-MA-Δ5-infected mice at that time point.
Cytokine accumulation in the lungs of rMERS-MA-Δ5-infected hDPP4-KI mice. (A) Selected cytokine mRNA accumulation. Total RNA was extracted from lung samples, collected at 4 and 6 dpi, of mock-infected animals (gray) or animals infected with parental rMERS-MA (black) or mutant rMERS-MA-Δ5 (white) virus. Quantification of mRNAs encoding IFN-β, ISG15, IL-10, CCL4, CCL2, and CXCL10 was performed by RT-qPCR using specific TaqMan assays; relative mRNA levels were based on comparison with mock-infected animals. (B) The levels of IFN-β in extracts from lung from mock-infected animals (gray) or animals infected with parental rMERS-MA (black) or mutant rMERS-MA-Δ5 (white) viruses were quantified by ELISA. Means from three animals per point are shown; error bars represent SEMs. r.u., relative units. *, P < 0.05; **, P < 0.01.
Only IFN-β was significantly decreased at 4 and also at 6 dpi in mice infected with rMERS-MA-Δ5 virus compared with those infected with the parental rMERS-MA virus (Fig. 7A). Indeed, a reduction in IFN-β protein levels in the lungs of mice infected with rMERS-MA-Δ5 was also observed at both 4 and 6 dpi (Fig. 7B). Overall, the results strongly suggested that loss of gene 5 sequence contributed to an increased virulence of mouse-adapted MERS-CoV-Δ5 virus, through immune modulation of the induction of IFN-β and other cytokines.
Viral mRNA accumulation in cells and hDDP4-KI mice infected with rMERS-MA or rMERS-MA-Δ5 virus.The described effects of gene 5 absence were unexpected and could reflect changes in genomic sequence outside gene 5. However, we detected no additional changes when the complete genome of both inoculum viruses and the virus extracted from hDPP4-KI mouse lungs was sequenced.
CoV transcription involves the production of subgenomic mRNAs (sgmRNAs) (52), some of which encode immunoevasive proteins (22, 23), and it was possible that gene 5 deletion had affected their levels. The accumulation of viral sgmRNAs was evaluated in infected Huh-7 cells and hDPP4-KI mice by semiquantitative RT-PCR analysis. All sgmRNAs except that encoding protein 5 were detected in all cases (Fig. 8A), and no significant differences in the accumulation of sgmRNAs for E, 3, and 4a/4b genes was noted (Fig. 8B), strongly suggesting that deletion of gene 5 did not change the levels of other viral proteins.
Viral sgmRNA accumulation by recombinant mouse-adapted viruses. Total RNA was extracted at 18 hpi from mock-infected cells or cells infected with rMERS-CoV (WT), rMERS-MA (MA), or rMERS-MA-Δ5 (Δ5). RNA was also extracted from lungs of infected hDPP4-KI mice at 4 dpi. Viral RNA accumulation was evaluated by semiquantitative RT-PCR. (A) Representative images of PCR results for each viral sgmRNA, indicated in the left. (B) Densitometric estimation of sgmRNA accumulation in Huh-7 cells (left panel) and hDPP4-KI mouse lungs (right panel) infected with rMERS-CoV (dark gray), rMERS-MA (black), or rMERS-MA-Δ5 (white) virus. Viral sgmRNA levels were relative to gRNA levels in all cases. All values were relative to those obtained in the infection of Huh-7 cells with rMERS-CoV virus, considered as 100%. Error bars represent SEMs.
Increased stability of MERS-CoV in normal human lung fibroblasts.As indicated above, in contrast with the viruses in humans and camels, MERS-CoV gene 5 was highly unstable in cell cultures, preventing the recovery of a mouse-adapted virus encoding the full-length gene 5 to compare MERS-CoV pathogenesis in the presence or the absence of full-length protein 5. Therefore, additional cell systems were explored. Both human rMERS-CoV and mouse-adapted rMERS-MA viruses efficiently grew in human MRC-5 cells (Fig. 9A). After 10 passages in MRC-5 cells, the sequence of the genome containing all accessory genes and E protein was determined after RT-PCR amplification. Both rMERS-CoV and rMERS-MA viruses led to the expected bands (Fig. 9B), and sequencing analysis showed no additional mutations or deletions in the expected sequences (data not shown). Moreover, similar virus titers were obtained for passage 10 viruses (Fig. 9C). Altogether, these data indicated that MERS-CoV was more stable in MRC-5 cells than in other cell types, such as Vero or Huh-7 cells.
Increased MERS-CoV stability in MRC-5 cells. (A) Growth kinetics of recombinant mouse-adapted (rMERS-MA [black]) and human (rMERS-CoV [gray]) viruses in MRC-5 cells. Cells were infected at a multiplicity of infection (MOI) of 0.1, and cell culture supernatant was recovered at the indicated times postinfection. (B) Stability of accessory genes in MRC-5 cells. Cells were infected with recombinant mouse-adapted (rMERS-MA) and human (rMERS-CoV) viruses. Blind passages were performed each 24 hpi. Total cellular RNA was recovered from the infected cells, and the genomic region containing the accessory genes was amplified with specific oligonucleotides, as in Fig. 3C. (C) Growth kinetics of passage 10 mouse-adapted (rMERS-MA [black]) and human (rMERS-CoV [gray]) viruses in MRC-5 cells. Cells were infected at a MOI of 0.1, and cell culture supernatant was recovered at the indicated times postinfection. The values indicate the mean from three independent infections; error bars represent SEMs.
Virus growth and innate immune response induced by rMERS-CoV-Δ5 virus in normal lung fibroblasts.Based on the increased MERS-CoV stability, human MRC-5 cells were then used to rescue mouse-adapted viruses after transfection with the engineered cDNAs (Fig. 3B). Both rMERS-MA and rMERS-MA-Δ5 viruses were efficiently recovered, with titers of 1.7 × 106 and 9.0 × 105 PFU/ml, respectively. The rMERS-MA-5FL virus was also rescued, with a significantly lower titer of 2.0 × 103 PFU/ml. Passage of these viruses in MRC-5 cells did not increase the rMERS-MA-5FL virus titer, and, unfortunately, small deletions were detected in gene 5 after passage 2 (data not shown). These results precluded the analysis of the effect of protein 5 absence or presence on MERS-CoV pathogenesis using the mouse-adapted virus and hDPP4-KI mouse model.
Therefore, to analyze protein 5 functions in the context of viral infection, human rMERS-CoV and rMERS-CoV-Δ5 viruses were rescued from MRC-5 cells transfected with the corresponding infectious cDNAs (15). Both rMERS-CoV and rMERS-CoV-Δ5 viruses were efficiently recovered, with titers of 1.67 × 106 and 1.7 × 106 PFU/ml, respectively. Moreover, both viruses presented similar growth kinetics in MRC-5 cells (Fig. 10), in line with the results previously obtained in Huh-7 cells (15). Interestingly, the mRNA accumulation of IFN-β, ISG15, and proinflammatory cytokines CCL2, CCL4, CXCL10, and TNF was significantly reduced in MRC-5 cells infected with rMERS-CoV-Δ5 virus compared with the levels observed in those infected with the parental rMERS-CoV virus (Fig. 11). In agreement with the previous observations from hDPP4-KI mice, these data indicated that the absence of protein 5 expression led to a decreased innate immune response.
Growth kinetics of rMERS-CoV-Δ5 virus in human lung fibroblasts. MRC-5 cells were infected at an MOI of 0.1 with the recombinant mutant rMERS-CoV-Δ5 (open circles) and the parental rMERS-CoV (black circles) viruses. Viral titers were determined at the indicated times postinfection. The values indicate the mean from three independent infections; error bars represent SEMs.
Cytokine mRNA accumulation in rMERS-CoV-Δ5-infected cells. MRC-5 cells were mock infected (gray) or infected at an MOI of 1 with mutant rMERS-CoV-Δ5 (white) or parental rMERS-CoV-WT (black) virus. Total RNA was extracted at the indicated times postinfection. Quantification of mRNAs encoding IFN-β, ISG15, CXCL10, CCL2, CCL4, and TNF was performed by RT-qPCR using specific TaqMan assays; relative mRNA levels were based on comparison with mock-infected cells. Means from three independent infections per point are shown; error bars represent SEMs. r.u., relative units. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Virulence of rMERS-CoV-Δ5 in mice.To analyze the virulence in the presence or in the absence of MERS-CoV protein 5, transgenic K18 TghDpp4 mice (53) were infected with rMERS-CoV or rMERS-CoV-Δ5 virus. These mice were susceptible to the human MERS-CoV, which causes a lethal disease associated with encephalitis, lung mononuclear cell infiltration, and lung edema (53). This model has been proven adequate to test protection strategies against MERS-CoV-induced lung pathology (54, 55). Significantly higher weight losses were observed in the mice infected with rMERS-CoV-Δ5 virus, compared with those infected with the parental virus (Fig. 12A). Moreover, the mortality of mice infected with the mutant virus lacking protein 5 was higher than that caused by the parental virus expressing protein 5 (Fig. 12B). Histopathological examination showed no damage in noninfected animals and limited localized damage in infected mice (data not shown). Mice infected with rMERS-Δ5 virus showed increased cell infiltration at 3 dpi compared with rMERS-CoV-infected mice (data not shown). These results indicated that rMERS-CoV-Δ5 virus was more virulent than rMERS-CoV virus, in agreement with the previous observations.
Virulence and growth of rMERS-CoVs in TghDpp4 mouse lungs. Thirty-one-week-old TghDpp4 mice were intranasally mock infected (Mock [black]) or infected with 105 PFU/animal of recombinant viruses recovered from infectious cDNA. Mice were inoculated with parental rMERS-CoV-WT (black) or mutant rMERS-CoV-Δ5 (white) viruses. Weight (A) and survival (B) were monitored. The values represent means from five animals per group. Lung samples were obtained at the indicated time postinfection, and viral gRNA accumulation (C) and virus titers (D) were determined (n = 3 mice per time point). Error bars indicate the SEM for each value. *, P < 0.05.
Interestingly, rMERS-CoV-Δ5 virus replicated more efficiently in the lungs of infected mice, as shown by the increased genomic RNA (gRNA) accumulation (Fig. 12C) and viral titer (Fig. 12D) compared with the parental rMERS-CoV virus. Altogether with the data obtained in MRC-5 cells and hDPP4-KI mice, these results may suggest a reduced innate immune response triggered by rMERS-CoV-Δ5 virus in the lungs of infected TghDpp4 mice compared with the one induced by the parental rMERS-CoV virus.
DISCUSSION
Our results, using two different animal models, showed that viruses completely lacking gene 5 (rMERS-MA-Δ5 and rMERS-CoV-Δ5) were more virulent than the parental rMERS-MA and rMERS-CoV viruses, indicating a role for protein 5 in the modulation of the innate immune response. rMERS-MA-Δ5 virus caused delayed induction of IFN-β, ISGs, and proinflammatory cytokines in the lungs of infected hDPP4-KI mice and rapid accumulation of edema in the lungs without associated inflammatory infiltration. Previous studies have also indicated extensive edema in mice with poor outcomes after infection with SARS-CoV or MERS-CoV (6, 30, 50, 56, 57).
It was previously proposed that protein 5 modulated NF-κB-mediated inflammation, as infection of cultured cells with a recombinant MERS-CoV in which protein 5 was replaced by red fluorescent protein increased proinflammatory cytokines, such as IL-1α, IL-1β, CCL5, and CCL4 (16). We observed no increase in IL-1α, IL-1β, or CCL5 in the lungs of mice infected with both recombinant viruses, in agreement with previous observations (30). Moreover, rMERS-MA-Δ5 virus caused a decrease in the accumulation of CCL4 in the lungs of infected hDPP4-KI mice, highlighting the different behaviors of protein 5 in the cultured cell systems used to date and in mice. Interestingly, rMERS-CoV infection of MRC-5 cells significantly increased the accumulation of IFN-β and proinflammatory cytokines. Furthermore, the observed fold changes between infected and noninfected cells were comparable to those observed in vivo in the lungs of infected hDPP4-KI mice. These data were in contrast with previous observations in Huh-7 cells, in which infection with rMERS-CoV does not induce IFN-β and other proinflammatory cytokines (22). Indeed, in Huh-7 cells no difference was found in the absence or the presence of protein 5 (22). The data presented in this work indicated that normal lung fibroblasts were a more appropriate cell culture system to maintain MERS-CoV genetic stability and to analyze the innate immune response during MERS-CoV infection.
KI mouse models developed for studies of MERS are based on the functional replacement of mouse DPP4 (mDPP4) by hDPP4, with substitution of two amino acids (28), three exons (30), or the complete gene (31). These mouse models reproduce the respiratory pathology caused by MERS-CoV without infection of tissues not infected in humans. To achieve lethality, adaptation of MERS-CoV to hDPP4-KI mice was required (28, 30). In the mouse-adapted virus used in this work, most of the viral genome is intact, including accessory genes 3, 4a, 4b, and 8b (30), which are involved in the modulation of virus-host interactions, specifically the innate immune response (16, 22–24). Thus, rMERS-MA has advantages over other mouse-adapted viruses that have extensive deletions and modifications in accessory genes, preventing the analysis of the function of these genes in pathogenesis (28, 45).
Accessory gene 5 is the only gene that is mutated in rMERS-MA. This gene is homologous to other CoV genus-specific genes (35, 36). These genes are nonessential for viral growth in cell culture, although there is increasing evidence that they may modulate in vivo pathogenicity (39, 41, 43). Our engineered rMERS-MA-Δ5 and rMERS-CoV-Δ5 viruses were more virulent than the parental ones, in contrast to what has been reported for the homologous genes in other CoVs. For instance, we showed that recombinant severe acute respiratory syndrome (SARS)-CoV-Δ3a is attenuated in vivo (39).
The absence of protein 5 caused a decrease in IFN production, both in human lung fibroblasts and in infected hDPP4-KI mice, which could partly explain the higher pathogenesis caused by rMERS-MA-Δ5 and rMERS-CoV-Δ5 mutant viruses. A dysregulated type I IFN response has been associated with poor outcomes in SARS-CoV-infected patients and mice (58, 59) and in MERS-CoV-infected mice (60). Increased virulence due to decreased type I IFN production has also been reported in the context of influenza A virus infections (61–63). In most cases, this occurred concomitantly with higher virus replication efficiency in target tissues. Both a 10-fold increase and a 5-fold increase in lung titers for rMERS-CoV-Δ5 and rMERS-MA-Δ5 compared to parental viruses were observed in TghDpp4 and hDPP4-KI mice, respectively (Fig. 12D and 4B). These results are concordant with previous studies that showed that overexpression of protein 5 activated the IFN-β promoter, although this effect was attributed to a general effect on cell transcription (20). Additional studies are required to clarify the molecular mechanism of gene 5 effects on the MERS-CoV-induced immune response.
Although MERS-CoV gene 5 is highly unstable in cell culture, our analysis of 476 human and camel MERS-CoV isolates showed that gene 5 is the most stable MERS-CoV accessory gene in field isolates. This observation is consistent with the described stability of HCoV-229E ORF4, which is homologous to MERS-CoV gene 5 (43). All HCoV-229E virus clinical isolates contain a full-length ORF4, while in cell-adapted strains, this gene is either mutated to carry two ORFs—4a encoding the N terminus of protein 4, including the three transmembrane domains, and 4b coding for the C terminus of full-length protein 4—or replaced by an ORF encoding a truncated protein containing just the first transmembrane domain (43). In addition, although our rMERS-MA virus did not express full-length protein 5, our data indicate that there is a difference in virus virulence when gene 5-derived sequence is present. sgmRNA-5 expression levels were similar in human and mouse-adapted MERS-CoVs, although in rMERS-MA virus, it contains a 17-nt deletion. Notably, there are downstream ATG codons in rMERS-MA that could lead to the expression of the carboxy-terminal domain of protein 5. Similar findings were described for SARS-CoV 3a protein, homologous to MERS-CoV protein 5. The sgmRNA for 3a protein contains several ATG codons leading to the expression of full-length protein 3a and a truncated protein corresponding to the 174 amino acids (aa) of the C terminus (64; C. Castaño, I. Sola, and L. Enjuanes, unpublished results). Therefore, one possibility is that a MERS-CoV protein 5 fragment is indeed expressed and retains part of protein 5 activity in the modulation of pathogenesis.
MATERIALS AND METHODS
Ethics statement.Experiments involving animals were performed in strict accordance with EU (2010/63/UE) and Spanish (RD 53/2013 and 32/2007) guidelines. All the protocols were approved by the on-site Ethical Committee (permit no. PROEX 112/14). Infected mice were housed in a self-contained ventilated rack (Allentown, NJ).
Cells and viruses.Baby hamster kidney cells (BHK) and human lung fibroblasts (MRC-5) were obtained from American Type Culture Collection (ATCC CCL-10 and CCL-171, respectively). Human hepatoma (Huh-7) cells were kindly provided by R. Bartenschlager (University of Heidelberg, Germany). All cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 25 mM HEPES and 5% (BHK) or 10% (Huh-7 and MRC-5) fetal bovine serum (FBS) (HyClone; ThermoFisher). Mouse-adapted MERS virus (MERSMA-6-1-2) was described previously (30). Recombinant rMERS-CoV and rMERS-CoV-Δ5 viruses were obtained from MERS-CoV infectious cDNA as previously described (15). Virus growth and titration were performed as previously described (15). All work with MERS-CoV infectious viruses was performed in biosafety level 3 (BSL3) facilities at CNB-CSIC according to the guidelines set forth by the institution.
Plasmid constructs.Nine DNA fragments containing the mutations present in the MERSMA-6-1-2 virus (30) were chemically synthesized by and purchased from GeneArt (ThermoFisher Scientific, Germany) (Table 2). The fragments were introduced into the corresponding intermediate pBAC plasmids using enzymes shown in Table 2. Sequential assembly of the MERS-CoV fragments was performed as previously described (15), using the unique restriction sites indicated in Fig. 1B, to obtain pBAC-MERSFL-MA. Two additional infectious cDNAs were engineered, containing either the full-length gene 5 sequence (pBAC-MERSFL-MA-5FL) or lacking the whole gene 5 sequence (pBAC-MERSFL-MA-Δ5) (Fig. 1B). To obtain pBAC-MERSFL-MA-5FL, synthetic fragment 6b, containing the gene 5 modifications (Table 2), was not introduced into intermediate plasmid pBAC-MERS-3′ (15) (Fig. 1B). Synthetic DNA fragments 6a and 6c (Table 2) were introduced in the plasmid pBAC-MERS-Δ5 (15), to engineer pBAC-MERSFL-MA-Δ5 (Fig. 1B). All cloning steps were checked by sequencing. For each mutant sequence, two independent cDNAs were constructed.
DNA fragments containing the mouse-adapted mutations
Transfection and recovery of infectious rMERS viruses from cDNA clones.BHK cells grown to 90% confluence in 12.5-cm2 flasks were transfected with 6 μg of the corresponding pBAC and 18 μl of Lipofectamine 2000 (Invitrogen) according to the manufacturer’s specifications. At 6 h posttransfection (hpt), BHK transfected cells were trypsinized and plated over confluent Huh-7 or MRC-5 monolayers grown in 12.5-cm2 flasks. After a 3-day incubation period, the cell supernatants were harvested (passage 0) and passaged once on fresh Huh-7 or MRC-5 cells, and viral titers were determined (15).
Virus infection of mice.hDPP4-KI mice (30) and transgenic K18 TghDpp4 mice (53) were maintained at the CNB-CSIC animal core facility. All work with infected animals was performed in a BSL3+ laboratory at the Center for Animal Health Research (CISA-INIA). Mice were anesthetized with isoflurane and intranasally inoculated with 104 PFU/animal of the different viruses in DMEM. Weight loss and mortality were evaluated daily. Animals approaching 25 to 30% weight loss were euthanized (45). Three animals per group were euthanized and necropsied at days 3, 4, and 6 after inoculation, depending on the experiment. Lungs were collected and examined for macroscopic lesions. Lungs were then divided and were either frozen for subsequent virus titration, placed in RNAlater stabilization reagent (Life Technologies) for RNA extraction, or fixed in zinc formalin (Sigma-Aldrich) for histopathology. To determine MERS-CoV titers, lungs were homogenized in phosphate-buffered saline (PBS) containing 100 IU/ml penicillin, 0.1 mg/ml streptomycin, 50 μg/ml gentamicin, and 0.5 μg/ml amphotericin B (Fungizone), using a gentleMACS dissociator (Miltenyi Biotec, Inc.).
Analysis of lung damage.Lung sections were fixed with zinc formalin and stored in 70% ethanol at 4°C. Paraffin embedding, sectioning, and hematoxylin-eosin staining were performed by the histology service at the CNB-CSIC. Samples were examined with a ZEISS Axiophot fluorescence microscope. Determination of the lung damage score was obtained from unbiased observation of 50 microscopy fields per animal and assigning a score of 0 to 3 related to the amount of interstitial, peribronchiolar, and perivascular inflammation (65).
Analysis of RNA by quantitative RT-PCR.Total intracellular RNA was purified from Huh-7 cells, MRC-5 cells, or tissue samples using an RNeasy minikit (Qiagen) according to the manufacturer's specifications. Total cDNA was synthesized using 100 ng of total RNA as a template, random hexamers, and a high-capacity cDNA transcription kit (Life Technologies), following the manufacturer’s recommendations. MERS-CoV gRNA was evaluated using a custom TaqMan assay previously described (15). To evaluate cellular mRNA levels, commercially available TaqMan assays were used (Table 3). The human hydroxymethylbilane synthase (HMBS) gene (Table 3) was used as a reference housekeeping gene, since its expression remains constant in infected cells compared to that in noninfected cells (66; data not shown]. For mouse-derived RNA quantification, rRNA 18S (Mm03928990_g1) was used as a reference gene (66), as mRNAs for all commonly used housekeeping genes have been found to be altered both in noninfected mice and in mice infected with different SARS-CoV and MERS-CoV viruses. Data were acquired with a 7500 real-time PCR system (Applied Biosystems) and analyzed with 7500 software v2.0.6. Relative quantifications were performed using the threshold cycle (2−ΔΔCT) method (67). All experiments and data analysis were MIQE compliant (68).
TaqMan assay resultsa
IFN-β quantification.Extracts from frozen tissues, used also for virus titration as described above, were used to evaluate the levels of IFN-β. An enzyme-linked immunosorbent assay (ELISA) kit specific for mouse IFN-β was used (Quantikine ELISA; R&D systems), following the manufacturer’s specifications.
Semiquantitative analysis of viral sgmRNAs.RNA from infected cells and lungs of hDPP4-KI mice at 4 dpi and cDNA were prepared as described above. cDNA was then used as a template for quantitative PCR (qPCR), using primer SA-33VS (5′-CCTCGTTCTCTTGCAGAAC-3′) and specific oligonucleotides for each viral subgenomic mRNA (sgmRNA): SA-558RS (5′-GATTTTCACAAGCAATGAG-3′) for gRNA, SA-26009RS (5′-CAGCAGATTCAGTATAAC-3′) for sgmRNA-3 and sgmRNA-4, SA-27201RS (5′-CAAACAGTGGAATGTAGG-3′) for sgmRNA-5, and SA-27812RS (5′-CTCGTCAGGTGGTAGAGG-3′) for sgmRNA-E. PCR products were resolved in 1.5% agarose gels. DNA amounts were estimated by densitometric analysis using ImageLab 5.2 software (Bio-Rad).
Sequence analysis.On 20 June 2018, 521 MERS-CoV sequences, including strains from humans (44%) and camels (54%), were downloaded from GenBank in FASTA format. The FASTA file was then curated by eliminating the following: (i) 39 partial sequences that did not include the gene 5 genome region; (ii) one partial sequence (GenBank accession no. KT806050) lacking genome 3′-end sequence, including from mid-gene 5 to the 3′ UTR; and (iii) 5 MERS-like bat-CoV sequences, as they have low sequence identity to MERS-CoV. The remaining 476 MERS-CoV sequences, including the gene 5 region, were then aligned using Multiple Sequence Comparison by Log-Expectation (MUSCLE) v3.8.31 software (69), downloaded from the European Bioinformatics Institute (EBI) web page. Alignments were visualized using Jalview application (70). Further analysis of specific MERS-CoV sequences containing mutations or deletions was performed using Lasergene 14 software (DNASTAR, Inc., Madison WI, USA).
Statistical analysis.Two-tailed, unpaired Student’s t tests were used to analyze the difference in mean values between groups. All results are expressed as the mean ± the standard error of the mean (SEM). P values of <0.05 were considered significant.
Data availability.Sequences have been deposited in GenBank under accession no. MT576585.
ACKNOWLEDGMENTS
We thank Marga Gonzalez (CNB-CSIC) for technical assistance. In vivo experiments were performed at INIA-CISA BSL-3 (Madrid, Spain).
This work was supported by grants from the Government of Spain (BIO2013-42869-R and BIO2016-75549-R AEI/FEDER, UE and SEV 2017-0712), the CSIC (202020E079), the European Zoonotic Anticipation and Preparedness Initiative (ZAPI; IMI_JU_115760), and the U.S. National Institutes of Health (NIH: 0258-3413/HHSN266200700010C awarded to L.E., 2P01AI060699 awarded to L.E., P.B.M., and S.P., and R01 AI129269 awarded to S.P.). L.W. was supported by a fellowship from the China Scholarship Council (CSC). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
FOOTNOTES
- Received 10 June 2020.
- Accepted 8 July 2020.
- Accepted manuscript posted online 3 November 2020.
- Copyright © 2021 American Society for Microbiology.
