Reanalysis of leader-body junction regions
Using public Nanopore RNA-seq data (Materials and Methods), 575,106 out of 879,679 reads from a SARS-CoV-2-infected Vero cell sample were aligned to the SARS-CoV-2 reference genome. Among all aligned reads, 575,106 sense reads represented gRNAs(+) or sgRNAs(+), while 30 antisense reads represented gRNAs(-) or sgRNAs(-). This exceedingly high ratio (575,106 vs. 30) between sense and antisense reads may be the result of significant differences between the gRNAs(+)/sgRNAs(+) and gRNAs(-)/sgRNAs(-) degradation efficiencies. This phenomenon, however, was not reported in the previous study [2]. Another explanation for the high ratio is that gRNAs(+)/sgRNAs(+) are protected by binding to the N proteins. Another super high ratio (198,198,542 vs. 11,820,438) between contiguous and junction-spanning reads was reported in that previous study [2]. This suggested that there were significant differences between gRNAs(+)/gRNAs(-) and sgRNAs(+)/sgRNAs(-).
By reanalysis of junction-spanning reads, we found that TRS-Bs share ~ 12 nt junction regions with TRS-Ls in SARS-CoV-2 (Fig. 1A). Most junction regions in eight protein-coding genes (S, E, M, N, ORF3a, 6, 7a and 8) exhibit high similarity to those of TRS-Ls and are defined as canonical junction regions [2]. Accordingly, the sgRNAs(+) and sgRNAs(-) containing these canonical junction regions are defined as canonical sgRNAs(+) and sgRNAs(-), respectively. The junction regions of ORF7b exhibit high diversity in their sequences. Whether junction regions exist in ORF10 remains unclear [2]. Our analysis of canonical junction regions determined their core sequence (CS) of SARS-CoV-2 to be ACGAAC (Fig. 1A).
Further investigation of“leader-to-body fusion” model
Further analysis of betacoronavirus (Materials and Methods) showed that ACGAAC is a highly conserved sequence in the putative TRS-Bs of genomes from subgroup B, C and D, while this motif was changed into ACTTTC in the subgroup A. This suggests that canonical junction regions contain a specific motif for enzyme reorganization. Since all CoVs and even viruses in the entire nidovirus order adhere to the “leader-to-body fusion” model [2], this or these enzymes should be encoded by the ORF1a or 1b gene, given their likelihood to be translated. After analysing 16 non-structural proteins (nsp1–16) encoded by the ORF1a or 1b genes, we determined that nsp15 (a nidoviral RNA uridylate-specific endoribonuclease, NendoU [5]) is most likely to function in these junction regions, given that a homolog of nsp15 has cleavage sites containing “GU” [6]. Thus, the cleavage site of nsp15 was identified to follow the motif “GTTCGT|N” (the vertical line indicates the breakpoint and N indicates any nucleotide base), read in the antisense strands of CoV genomes. Almost all the genomic sites containing the motif “GTTCGT” have polyT (not less than three T) at the tail, which ensures the presence of at least one uridine for nsp5 cleavage. Furthermore, we found that two ACGAAC motifs closely flanking ORF8 upstream and downstream. This suggested that ORF8 was acquired by betacoronavirus subgroup B through “leader-to-body fusion”.
Upon searching for “GTTCGT” in the genomes of betacoronavirus subgroup B, the occurrence of “GTTCGT” on the antisense strand was found to be more than 1.6 times that on the sense strand. In particular, “GTTCGT” occurred 3 and 9 times (Table 1) on the sense and antisense strands of the SARS-CoV-2 genome, respectively. These findings suggest that the basic function of nsp15 involves the degradation of gRNAs and gsRNAs and that the high ratio between sense and antisense reads (see above) results from substantially more cleavage of gRNAs(-)/gsRNAs(-) than that of gRNAs(+)/gsRNAs(+). Among the three sites containing “GTTCGT” on the sense strand of the SARS-CoV-2 genome (referred to as internal cleavage sites—ICSs), one is located in the coding sequence (CDS) of RdRP (nsp12), while the other two are located in the ORF8 gene (see below).
Table 1
The genomic sites of motif GTTCGT
| Motif | Position | Region(Start-end) | Type |
| GTTCGT | 16014 | nsp12(13483–16236)* | ICS |
| | 28198 | ORF8(27894–28259) | ICS |
| | 28233 | ORF8(27894–28259) | ICS |
| ACGAAC | 70 | 5' UTR(1-265) | TRS-L |
| | 21556 | S(21563–25384) | TRS-B |
| | 25385 | ORF3a(25393–26220) | TRS-B |
| | 26237 | E(26245–26472) | TRS-B |
| | 26473 | M(26523–27191) | TRS-B |
| | 27041 | ORF6(27202–27387) | TRS-B |
| | 27388 | ORF7a(27394–27759) | TRS-B |
| | 27888 | ORF8(27894–28259) | TRS-B |
| | 28260 | N(28274–29533) | TRS-B |
| The motifs (in the first column) were mapped to the SARS-CoV-2 genome (GenBank: MN908947.3) using positions (in the second column) of the first nucleotides. ACGAAC indicated the motif GTTCGT read on the antisense strand of the SARS-CoV-2 genome. The third column shows the regions influenced by the motifs (in the first column). * this site is located in the CDS of RdRP (nsp12). TRS-L: transcription regulatory sequence in the body; TRS-B: transcription regulatory sequence in the leader; ICS: internal cleavage site; RdRP: RNA-dependent RNA polymerase; CDS: coding sequence. |
These two ICSs are also located in ORF8 of most SARS-CoV-2, SARS2-like CoV and SARS-like CoV genomes; however, they are absent in the genomes of SARS-CoVs obtained from humans (GenBank: AY274119 and AY278489) and SARS-like CoV genomes from civets (GenBank: AY304486, AY515512 and AY572034). One of the two ICSs is present in the genome of the SARS-like CoV strain WIV1 from bats (GenBank: KF367457), which is the ancestor of SARS-CoV. Deletions of ORF8 were reported to be associated with attenuation of SARS-CoV (GenBank: AY274119) [8] and SARS-CoV-2 [10]. The ORF8 gene of SARS-CoV is considered to have played a significant role in adaptation to human hosts following interspecies transmission [7] via the modification of viral replication [8]. The loss of two nsp15 ICSs in ORF8 of SARS-CoV is the key clue revealing the functions of ORF8 and the pandemic of SARS-CoV.
Based on the above analyses, we modified the “leader-body fusion” model and proposed the molecular basis of it. In our model, nsp15 cleaves synthesized gRNAs(-) and sgRNAs(-). The cleavage occurs at TRS-Bs(-) synthesized using TRS-Bs as templates. Next, the free 3' ends (~ 6 nts) of TRS-Bs(-) hybridize the junction regions of TRS-Ls for template switching. Alternatively, the cleaved TRS-Ls(-) are used as templates to synthesize TRS-Ls that then hybridize the junction regions of the cleaved sgRNAs(-) without TRS-Ls(-) for template switching and gRNAs(+)/sgRNAs(+) synthesis. These findings suggests that RdRP (nsp12) is so active that it must have a very high enzyme activity.
Some non-canonical sgRNAs(+) and sgRNAs(-) are synthesized as a result of occasional hybridization between the free 3' ends of TRS-Bs(-) and highly similar sequences of TRS-L junction regions. The sgRNAs(-) without TRS-Ls are also synthesized due to missing hybrids. In the previous study [2], sgRNAs(-) without TRS-Ls, non-canonical sgRNAs(+) and sgRNAs(-) were reported, supporting our model. Furthermore, cleavage also occurs at ICSs on the sense strands, after which the free 3' ends of the cleaved 5' ends hybridize the antisense strands to synthesize recombinant sgRNAs(+) or even gRNAs(+). This may also contributes to the multiple recombination events in betacoronavirus genomes.
Proposal of a negative feedback model
Based on the above analyses, we proposed a negative feedback model (Fig. 1B) to explain the regulation of replication and transcription in—but not limited to—CoVs. In this model (Fig. 1), nsp15 regulates the synthesis of subgenomic RNAs (sgRNAs) or genomic RNAs (gRNAs) by the cleavage of TRS-Bs(-). The expression level of nsp15 determines the relative proportions of sgRNAs and gRNAs. An increase in nsp15 expression results in less gRNAs(-) and more gsRNAs(-), after which fewer gRNAs(+) and more gsRNAs(+) are synthesized, respectively. A decrease in gRNAs(+) results in a decrease of nsp15 expression, given that nsp15 is synthesized using gRNA(+) as the template. Furthermore, the nsp15 ICS in the CDS of SARS-CoV-2 nsp12 (Table 1) enhances the negative feedback. Via this negative feedback mechanism, CoVs reach equilibrium between the replication and transcription (Fig. 1B); thus, this mechanism is important for the persistence of CoVs in hosts.
Our negative feedback model is based on the determination of the molecular basis of the “leader-body fusion” model. This molecular basis is associated to the cleavage function of nsp15, which is different from other models proposed in the previous studies [4], mainly due to the integration of information from many aspects, particularly considering: (1) The direct RNA sequencing data [2]; (2) The nsp15 structure in complex with GpU (pdb code: 6 × 1B); (3) the polyT at the tail of “GTTCGT”; and (4) the nsp15 ICSs in ORF8 (see above). These discoveries confirmed the identification of the cleavage sites of nsp15. However, they were not used to construct the models in the previous studies [4].
The necessity of negative feedback regulation
In our previous study [7], we proposed that first hairpins (immediately upstream of ORF1a) have an important role in the functions (e.g. regulation of translational initiation) of ribosome binding sites (RBSs) in 5' UTRs of the SARS-CoV-2 genomes. SARS-CoV and SARS-CoV-2 have an identical first hairpin, which may enhance the translation of downstream genes. To indirectly prove that negative feedback is a basic mechanism, we designed preliminary experiments to show that first hairpins from SARS-CoV-2 enhance the translation of the downstream enhanced green fluorescent proteins (EGFPs); however, over-expression of EGFP without negative feedback regulation will cause cell death.
In total, three types of plasmids containing EGFP reporter genes—named pEGFP-C1, pSARS, and pCoV-ba (betacoronavirus subgroup A)—were used in the experiments (Fig. 2A). The plasmid pEGFP-C1 was used as a control, given that it contains 17-nt sequences, encoding the first hairpin from Cytomegalovirus (CMV). Two types of plasmids proceeded by 30- and 29-nt inserts were used to evaluate their influence on translation (Materials and Methods). These two inserts encoded the two first hairpins from SARS-CoV-2 and the subgroup A of betacoronaviruses, respectively. Comparing the fluorescent brightness of cells transfected with three types of plasmids, the expression of EGFP in pSARS was markedly higher than that in pEGFP-C1 and pCoV-ba (Supplementary 1). Moreover, pSARS caused cell death at 48 hours after plasmid transfection. We then performed 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2-H-tetrazolium bromide (MTT) and lactate dehydrogenase (LDH) experiments to further evaluate the influence of plasmid transfection (Materials and Methods). Both MTT and LDH experiments consistently suggested that pSARS caused significantly more HEK293T and Hela cell death at 48 hours after plasmid transfection, than pEGFP-C1 and pCoV-ba. However, this phenomenon appeared in HEK293 cells at 56 hours after plasmid transfection (Fig. 2B). Given that the only difference among the three types of plasmids is their 17-, 30-, and 29-nt sequences encoding different hairpins, we concluded that these hairpins determined the translation efficiency of the downstream EGFPs. The hairpin in pSARS resulted in the over-expression of EGFP, which caused more cell death. To determine whether the factor acts at the translation level and to rule out other possible factors that may exert influence at the replication or transcription level, we performed the following experiments: (1) using HEK293 cells to rule out the possible influence by the differences of plasmid copy numbers, since all three types of plasmids containing the SV40 origins can be replicated to a copy number of between 400 ~ 1000 plasmids per cell within HEK293T; and (2) using qPCR to rule out the possible influence by differential transcription (Supplementary 1).