SARS-CoV-2 and ORF3a: Nonsynonymous Mutations, Functional Domains, and Viral Pathogenesis

1The rapid spread of coronavirus (CoV) disease 2019 (COVID-19), caused by severe acute respiratory syndrome CoV 2 (SARS-CoV-2), caused a major global concern. Coronaviruses are enveloped positive-sense RNA viruses and are broadly distributed in humans and mammals. The genome of SARS-CoV-2 showed 96.2% sequence similarity to a bat SARS-related coronavirus (SARS-CoV RaTG13) collected in Yunnan Province, China, and 79% and 50% similarities to SARS-CoV and Middle East respiratory syndrome CoV (MERS-CoV), respectively. A 91% similarity to pangolin CoV suggested that pangolins can be considered possible hosts in the emergence of the novel coronavirus. The 3a protein (NCBI accession number YP_009724391.1) showed 72% sequence similarity to that detected in SARS-CoV.

2We investigated the presence in SARS-CoV-2 of functional domains in the 3a protein linked to virulence, infectivity, ion channel formation, and virus release. We then studied the diverse nonsynonymous mutations in ORF3a and investigated the effect of newly introduced mutations in the localization and tree topology of the 3a protein in SARS-CoV-2.

3Signature mutations within SARS-CoV-2 ORF3a cause the isolates to cluster into defined phylogenetic clades (Fig. 1). We observed microclonality within the ORF3a gene tree defined by the nonsynonymous mutations separating the isolates into distinct subpopulations, highlighted in Fig. 1. Moreover, three isolates of the Q57H clade with the Q57H mutation were identified to contain second mutations: D173Y (EPI_ISL_419177), W131C (EPI_ISL_418188), and L129F (EPI_ISL_418241). One isolate, namely, EPI_ISL_411929 from the G251V clade, also had a W128L mutation.

4The 3a protein showed a 97.82% sequence similarity to a nonstructural protein, NS3, of bat coronavirus RaTG13 (NCBI accession number QHR63301.1). The alignment of the ORF3a protein sequences extracted from the 2,782 available genomes revealed in total 51 different nonsynonymous amino acid (aa) substitutions (Table 1). Q57H and G251V were the most common and identified in 17.43% (n = 485) and 9.71% (n = 270) of the genomes, respectively.

5We divided the 3a protein into six functional domains (I to VI) based on previously reported data and color-coded each domain for its role within the host cell (see Table S1 in the supplemental material and Fig. 2). Then, we aligned and compared the amino acid sequences in SARS-CoV (NCBI accession number P59632), SARS-CoV-2 (UniProtKB accession number P0DTC3/NCBI accession number YP_009724391), RaTG13 (EPI_ISL_402131), pangolin CoV (EPI_ISL_410721), and civet SARS (NCBI accession number AAU04650.1) to determine whether or not SARS-CoV-2 has similar functional domains and to accordingly follow and determine whether any of the introduced nonsynonymous mutations has a potential impact on the virus’ virulence and pathogenesis.

6One of the immediately observed differences was the absence of the previously detected N terminus putative signal peptide (Fig. 2, aa 1 to 15, domain I) in SARS-CoV (Table S1), confirmed using Protter v1.0, from all the other studied strains, including SARS-CoV-2 (domain I). The 3a protein in SARS-CoV-2 and all other studied strains, as with SARS-CoV, had three transmembrane regions (Fig. 2).

7Domain II contained the TRAF3-binding motif in SARS-CoV, which was also detected in SARS-CoV-2. We observed in domain II of SARS-CoV-2 two amino acid substitutions (positions 36 and 40; amino acid substitutions are shown in bold in the motifs below). A PLQAS motif was conserved in SARS-CoV and civet SARS in domain II, while the L37I substitution (PIQAS motif) was detected in SARS-CoV-2 and RaTG13, with the pangolin CoV additionally having an S40T substitution (PIQAT motif) (Fig. 2; Table S1).

8Domain III consisted of a K+ ion channel (positions 91 to 133) and a cysteine-rich domain (positions 81 to 160) in SARS-CoV. We noticed that in this domain, Y91 and Y109 were conserved (Fig. 2). Several mutations were identified within this domain in SARS-CoV-2 and included H93Y, L127I, W128L, L129F, and W131C. The sequence alignment of 2,782 SARS-CoV-2 3a proteins revealed a 0.5% (n = 14) prevalence of H93Y (source, Wales, UK; date, 12 March 2020 to 20 March 2020) and a 0.036% (n = 1) prevalence of L127I (EPI_ISL_418264; source, Greece; date, 18 March 2020), W128L (EPI_ISL_411929; source, South Korea; date, January 2020), L129F (EPI_ISL_418241; source, Algeria; date, 02 March 2020), and W131C substitutions. The W131C mutation detected in strain EPI_ISL_418188 (source, USA; date, 23 March 2020) added a third cysteine residue to this domain in SARS-CoV-2.

9A cysteine-rich region was also observed between positions 81 and 160. Cysteine residues were previously reported as being involved in the homodimerization of the 3a protein in SARS-CoV. The most important residue for homodimerization was C133 and was conserved in all studied strains (Fig. 2, domain III).

10Domain IV consisted of a caveolin-binding motif (Fig. 2, positions 141 to 149; Table S1). A single amino acid substitution was observed in SARS-CoV-2, RATG13, pangolin CoV (YDANYFLCW motif), and civet SARS (YEANYFVCW motif).

11The YXXΦ motif was detected in all studied strains, including SARS-CoV-2 in domain V (motif, YNSV; positions 160 to 163). Finally, domain VI in SARS-CoV consisted of a diacidic motif, ExD, at positions 171 to 173 (Table S1; Fig. 2). The diacidic EGD motif was conserved in SARS-CoV and civet SARS, while E171S changed the motif to SGD in SARS-CoV-2, RaTG13, and pangolin CoV. A D173Y substitution was detected in one SARS-CoV-2 strain (EPI_ISL_419177; source, France; date, 22 March 2020), completely disrupting the diacidic motif.

12ORF3a encodes a minor structural protein of 274 aa residues in SARS-CoV. In this study, we divided the 3a protein into six functional domains (I to VI) based on previously reported data from SARS-CoV and color-coded each domain for its role within the host cell (Fig. 2).

13We linked the TRAF3-binding motif in SARS-CoV to domain II and found that we have a similar one in SARS-CoV-2. The 3a protein in SARS-CoV, associated with TRAF3 through the TRAF3-binding motif, was found to activate NF-κB and the NLRP3 inflammasome.

14Domain III had the K+ ion channel and cysteine-rich domain in SARS-CoV. We observed several mutations within this domain in SARS-CoV-2. H93Y was particularly important, previously being linked in SARS-CoV to the loss of the K+ channel and reduced proapoptotic activity. A cysteine-rich region between positions 81 and 160 was also detected in SARS-CoV-2. 3a in SARS-CoV forms interchain disulfide bonds on the interior side of the viral envelope with the spike (S) protein though cysteine-rich regions, and the biological function of the 3a protein was correlated with that of the S protein in SARS-CoV.

15Additionally, cysteine residues were associated with the homodimerization of the 3a protein in SARS-Co. C133 was particularly important in maintaining the homodimer, which was conserved in all viruses, including SARS-CoV-2.

16Domain IV consisted of a caveolin-binding motif in SARS-CoV. Potential interactions with caveolin-1 may regulate the uptake and trafficking of the 3a protein to the plasma or endomembranes.

17In all the studied strains, the conserved YXXΦ motif in domain V, which had a significant role in the transport of the 3a protein from the Golgi apparatus to the plasma membrane in SARS-CoV, was another important finding. Mutations in this motif were linked to the aggregation of the 3a protein in the Golgi apparatus. Maintaining the YXXΦ motif in all strains confirms its role in 3a intracellular trafficking and surface transport, which otherwise would be targeted to lysosomal degradation via lipid droplets. A diacidic motif on the C terminus of SARS-CoV, which was also detected in SARS-CoV-2, was also linked to intracellular protein sorting and trafficking signals.

18Our study showed the functional importance of conserved domains across the species barrier and revealed the possible roles of the 3a protein in the viral life cycle. The observations reported in this study merit experimental confirmation.

19A total of 2,825 genomes, as of 5 April 2020, were downloaded from GISAID. Genomes were selected based on both completeness (>29,000 bp) and the high coverage option. Sequences were piped into Prokka v1.14.6 with the “- -kingdom Viruses” flag enabled. ORF3a protein sequences were extracted, and amino acid sequences were further parsed for sequencing related artifacts, such as N characters and N strings. Based on these criteria, 2,782 genomes were selected for downstream analysis.

20The selected genomes were aligned using MAFFT v7.450, and the multiple-sequence alignment (MSA) was viewed in Jalview v2.10.5. Nonsynonymous amino acid variants were manually extracted from the amino acid MSA. The variant site locations were put into the Protein Variation Effect Analyzer, known as PROVEAN v1.1.3. The selected −2.50 cutoff value represents a mean balanced accuracy (specificity versus sensitivity) of 78.17%.

21The protein sequences for protein 3a in SARS-CoV and SARS-CoV-2 were downloaded from the Swiss model repository with the UniProtKB accession numbers P59632 and P0DTC3/YP_009724391, respectively. Both sequences were aligned with MAFFT for direct comparison. Domain and motif scanning was performed through option 3 in the Web-based ScanProsite tool available at https://prosite.expasy.org/scanprosite/. The consensus patterns for various domains were manually entered, and scans were run with high sensitivity to eliminate unwanted matches. Identified domains and motifs were manually inspected and identified through the sequence alignment and correlated with the various nonsynonymous amino acid variants.

22Membrane topology of the ORF3a protein was detected using Protter. Default parameters were adopted for sequence-based topology visualization of SARS-CoV (NCBI accession number P59632), SARS-CoV-2 (UniProtKB accession number P0DTC3), RaTG13 (NCBI accession number QHR63301.1), pangolin CoV (EpiCoV accession number EPI_ISL_410721), and civet SARS (GenBank accession number AY572035) 3a proteins. The Protter server collects protein topology data from UniProt or Phobius.

23Amino acid sequences of all 3a protein loci were aligned using MAFFT v7.450. The alignments was passed through BMGE to infer entropy values relevant to the phylogeny, with minimal reconstruction artifacts. A phylogenetic tree of aligned Orf3a amino acid sequences was constructed using FastME 2.0, which builds an initial neighbor-joining (NJ) tree and improves topology by implementing the nearest-neighbor interchanges (NNIs) algorithm along with Felsenstein’s bootstrap iterations for branch support.