Coronaviruses have been known for quite some time as viruses that cause a variety of diseases in humans and animals (32
). The discovery of a coronavirus as the causative agent of SARS revived the interest in coronaviruses and resulted in a rapid increase of the number of identified coronaviruses, as well as of the number of full coronavirus genome sequences. Until this study, lineage C of the genus Betacoronavirus
(formerly known as subgroup 2c) included virus isolates from bats. Here, we determined and analyzed the complete genome sequence of a previously unknown lineage C betacoronavirus that was isolated from the sputum of a 60-year-old male suffering from acute pneumonia and renal failure in the Kingdom of Saudi Arabia whose death was probably a consequence of this infection (24
The sequencing of the full HCoV-EMC/2012 genome was greatly facilitated by the advent of high-throughput techniques. Using an optimized random amplification deep-sequencing approach, approximately 90% of the virus genome was covered with high accuracy in a single run. Using the data from this first run, primers could be designed to perform conventional Sanger sequencing for confirmation. This combination of techniques allowed the determination of the complete virus genome within a few days, without a requirement for prior knowledge of the virus genome under investigation. The error rate in 454 deep sequencing was generally higher than in Sanger sequencing, but the high coverage across the HCoV-EMC/2012 virus genome (up to 5,697 reads per nucleotide position) corrected for most of the incorrect base callings. The sequence obtained using the 454 platform aligned almost perfectly with that obtained by Sanger sequencing, with the exception of two nucleotide positions. The deep-sequencing data revealed variation at position 11623 (U or G), with G occurring in 44% of the reads, suggesting that ORF1a-encoded residue 3782 can be either valine (codon GUC) or glycine (codon GGC). The valine codon was the more abundant codon at this position in HCoV-EMC/2012, and valine is also present in most other betacoronaviruses. At position 27162, both G and A were detected in different runs, with an A in 45% of the reads. This G-to-A substitution introduces a premature stop codon (UGG to UAG) in ORF5. The virus stock that we sequenced was derived from passage of the virus from a sputum specimen six times in Vero cell culture. Hence, the observed sequence variants may reflect either natural heterogeneity or emerging genetic changes that occurred during virus passage in cell culture. Additional HCoV-EMC/2012 virus isolates or patient materials are currently not available to verify these genome sequence ambiguities at positions 11623 and 27162.
Adaptation to cell culture leading to a loss of functionality of genes, and in particular in relation to the so-called “accessory protein genes,” has previously been described for a variety of coronaviruses, including SARS-CoV (2
). These genes, like ORF3 through ORF5 of HCoV-EMC/2012, are dispersed between the structural protein genes (35
) and in some cases may even overlap such a gene, as in the case of the ORF overlapping the N protein gene in betacoronaviruses () (23, 36). The origin of most accessory protein genes remains unclear, although for some, acquisition by recombination with cellular or heterologous viral sequences seems plausible (37
). Accessory gene functions have been probed by reverse genetics (knockout mutants) for a variety of coronaviruses, including SARS coronavirus (39
), which established that they are not essential for replication in cell culture systems. In animal models, on the other hand, profound effects on pathogenesis after the inactivation (or transfer to a heterologous coronavirus) of accessory protein genes have been previously described (40
). In some cases, accessory gene products have been implicated in immune evasion, e.g., by interfering with cellular innate immune signaling (43
The apparent absence of selection pressure on coronavirus accessory protein genes during cell culture passage may explain the relatively high frequency with which loss of functionality appears to occur. The detection of an internal termination codon in part of the HCoV-EMC/2012 ORF5 sequences (45% of the reads) may constitute another example of such an event, which would lead to the truncation of the ORF5 protein after 107 amino acids. This would resemble a 29-nt deletion that occurred in the SARS-CoV genome, which resulted in the truncation of ORF8 (34
), and a 45-nt in-frame deletion in ORF7b of the same virus that emerged upon cell culture passage (23
Our analysis identified a potential ORF underlying the N protein gene (ORF8a), which is a common feature in betacoronaviruses. This ORF was not previously described for BtCoV-HKU4 and BtCoV-HKU5 (8
) but is conserved in the genome sequences of both viruses (see Fig. S1J
in the supplemental material). Remarkably, in HCoV-EMC/2012, both the 5′ and 3′ parts of the ORF appear to have been truncated. In BtCoV-HKU4 and BtCoV-HKU5, the ORF8b AUG codon would be the second AUG on sg mRNA8, making leaky ribosomal scanning a likely mechanism for translation initiation. In HCoV-EMC/2012, however, this AUG codon (positions 28606 to 28608) seems to have been mutated to AUA. Conservation of the sequence immediately downstream of this position, which is now formally upstream of ORF8b in HCoV-EMC/2012, was observed with BtCoV-HKU4 and BtCoV-HKU5, suggesting that the putative loss of this AUG codon may also have been a relatively recent event. In the 3′ part of ORF8b, sequence alignment of HCoV-EMC/2012 with BtCoV-HKU4 and BtCoV-HKU5 suggests that the former acquired a premature termination codon at positions 29099 to 29101 (UAA). Although we cannot at present assess the timing of these events in HCoV-EMC/2012 evolution, due to the lack of alternative samples for this species, the presumed loss of ORF8b functionality may also be a consequence of virus passage in cell culture.
To classify newly identified coronaviruses as the prototype of a novel virus species, it is required that the amino acid sequence identity in the conserved replicase domains in all intervirus pairwise comparisons is below the 90% threshold (1
). Here, we propose HCoV-EMC/2012 to represent a novel species of the betacoronavirus genus, since the amino acid sequence identities between HCoV-EMC/2012 and its closest relatives BtCoV-HKU4 and BtCoV-HKU5 in the seven conserved domains of ORF1ab were 75% and 77%, respectively. These viruses were originally detected in Asia in lesser bamboo bats (Tylonycteris pachypus
) and Japanese house bats (Pipistrellus abramus
), respectively (8
). This proposed classification will remain provisional until approved by ICTV.
The ICTV guidelines for coronavirus species demarcation require the availability of a (nearly) complete genome sequence prior to virus classification. However, there is considerable correlation between the results based on full-genome sequence analysis and those determined using the most conserved part of the ORF1b-encoded RdRp domain, which is commonly used in screening for new coronaviruses. In 2010, this partial sequence was reported for a betacoronavirus (VM314/2008) that was isolated 2 years earlier from a Pipistrellus pipistrellus
bat in The Netherlands. This virus was provisionally classified a betacoronavirus based on a 332-nt fragment from the RdRp-encoding domain of ORF1b (31
), which shares 88% nucleotide sequence identity with HCoV-EMC/2012, the highest identity with any coronavirus sequence available in the public domain. Although this high similarity is not sufficient to resolve the taxonomic relation between HCoV-EMC/2012 and isolate VM314/2008, it suggests that they may both belong to the same coronavirus species. Establishing the genome sequence of VM314/2008, or closely related viruses, is urgently required to verify this hypothesis. Based on the genetic relation between HCoV-EMC/2012 and bat coronaviruses, it is tempting to speculate that HCoV-EMC/2012 emerged from bats—either directly or via an intermediate animal host, possibly Pipistrellus
bats. This bat species is known to be present in the Kingdom of Saudi Arabia and neighboring countries.
Although most infections of human coronaviruses are relatively mild, the infection by HCoV-EMC/2012 with fatal outcome, and a similar severe case of an infection with a closely related coronavirus in London (25
), is a reminder that certain coronaviruses may cause severe and sometimes fatal infections in humans. It is important to develop an animal model that can be used to fulfill Koch’s postulates for the novel virus, by demonstrating that the isolated virus can indeed cause the observed disease. The availability of the HCoV-EMC/2012 genome sequence will facilitate the development of a variety of diagnostic assays that can be used to study the prevalence and clinical impact of HCoV-EMC/2012 infections in humans. The first generation of assays for this purpose has recently been described (45
). We anticipate that the availability of this full-length virus genome sequence will be valuable for the development of additional applied and fundamental research.