An international research team led by investigators at Huazhong Agricultural University assembled the first cross-genus integrated analysis of coronavirus transcriptomes and RNA-RNA interactomes, spanning all four coronavirus genera: alpha, beta, gamma, and delta. The team performed new RNA sequencing and RIC-seq experiments on the porcine epidemic diarrhea virus, an alpha-coronavirus, and porcine deltacoronavirus, a delta-coronavirus, filling a critical gap left by earlier studies that focused almost exclusively on beta-coronaviruses such as SARS-CoV-2 and MERS-CoV. By combining their new data with publicly available datasets, the researchers constructed a resource of 234 transcriptomes and 12 RNA-RNA interactome maps, hosted in a publicly accessible database called the Coronavirus Omics Data Explorer.

The first major finding to emerge from this integrated analysis was a conserved U-shaped distribution of RNA-RNA interaction frequencies across all four coronavirus genera. Interactions follow a power-law decay at short genomic distances, reach a minimum between 5 and 20 kilobases, and then rise again at distances exceeding 20 kilobases. This hierarchical folding pattern, preferentially enriching both short-range and long-range contacts, suggests that coronaviruses employ a shared genome organization strategy that may help coordinate transcription and replication during infection, regardless of which genus the virus belongs to.

The researchers then systematically classified canonical, TRS-dependent, and non-canonical, TRS-independent, transcription junctions across 103 high-quality RNA sequencing samples covering 14 different coronaviruses. Unsupervised clustering revealed three distinct groups of samples: a canonical-dominant group accounting for 63.1 percent of samples, a mixed group at 19.4 percent, and a non-canonical-dominant group at 17.5 percent. Notably, the balance between canonical and non-canonical transcription shifted with infection time, serial passage, viral genetic background, and specific experimental perturbations. Non-canonical junctions accumulated with serial passage in five of six independent passage datasets, and treatments such as luteolin altered the junction ratio, whereas others, including the nucleoside analog NHC, had minimal impact.

Turning to the structural basis of these events, the team used aggregate peak analysis of RNA-RNA interaction data to examine the genomic regions flanking canonical junction sites. Across PEDV, SARS-CoV-2, PDCoV, and MERS-CoV, interaction hotspots preferentially occurred between the flanking regions of the leader transcription regulatory sequence and body transcription regulatory sequences oriented in the same genomic direction, producing pronounced diagonal enrichment in the heatmaps. This same-directionality pattern, validated independently by COMRADES, SPLASH, and simplified-SPLASH in SARS-CoV-2, suggests that parallel spatial proximity between these regulatory elements facilitates the template-switching step carried out by the viral polymerase during canonical sgRNA synthesis.

Non-canonical junctions told a different structural story. Short-range non-canonical junctions, those spanning less than roughly 1.7 kilobases, showed anti-diagonal enrichment in the interaction heatmaps, indicating convergent, or antiparallel, orientations between the participating genomic regions. These convergent interactions are consistent with reverse-complementary base pairing that stabilizes RNA stem-loop structures. In SARS-CoV-2-infected cells, the researchers identified 5,014 non-canonical junctions between three TRS-like sequence pairs co-localized within a single stem-loop spanning nucleotides 29,262 to 29,870, demonstrating that such structures can physically bring distant sequence motifs into contact to drive discontinuous transcription without any involvement of the canonical leader sequence.

Perhaps the most striking connection emerged when the team mapped genomic deletion hotspots. Across multiple genome alignments, they identified 105 deletions exceeding 20 nucleotides in SARS-CoV-2, 31 in PEDV, 17 in MERS-CoV, and 8 in PDCoV. The interaction patterns at deletion boundaries mirrored those at short-range non-canonical junctions, with significant anti-diagonal enrichment. Moreover, 29.5 percent of SARS-CoV-2 deletions, 45.2 percent of PEDV deletions, and 62.5 percent of PDCoV deletions overlapped with predicted stem-loop regions. In SARS-CoV-2, a 2,250-nucleotide window in the 3-prime proximal genome, encompassing the 3’SL1, 3’SL2, and 3’SL3 stem-loops, contained 16 deletions concentrated in the ORF7a, ORF7b, and ORF8 accessory genes, a density far exceeding the genome-wide expectation. These findings suggest that stem-loop-mediated short-range non-canonical junctions may serve as initiating events for genomic deletions that shape viral evolution.

The conserved 3’SL1/2/3 region also revealed a mechanism for expressing ORF10, a SARS-CoV-2 gene that lacks its own canonical body TRS. The 5-prime portions of these stem-loops sit downstream of the TRS-B elements of upstream genes such as N, ORF7a, and ORF7b, while their 3-prime portions lie upstream of ORF10, an arrangement that allows TRS-B transfer from upstream genes to ORF10. Nanopore sequencing confirmed the existence of double-junction sgRNAs containing ORF10, upstream TRS-B sequences, and the leader, with an average of 11.37 percent of non-canonical ORF10 sgRNAs arising through this route. The remaining ORF10 transcripts were formed by a single long-range junction between the N gene and ORF1a, generating what the researchers designated sgRNA-N-ORF1a.

Remarkably, this N-ORF1a junction proved to be a recurring feature across coronavirus genera. Full-length sgRNA-N-ORF1a molecules were confirmed by nanopore sequencing in PEDV, MERS-CoV, MHV, and HCoV-229E, and reproducible RNA-RNA interactions between the two participating genomic regions were detected across different viruses. The transcription level of sgRNA-N-ORF1a tracked changes in interaction strength over the infection time course, declining in parallel with a 12.76 percent reduction in N-ORF1a genomic interaction between 24 and 48 hours post-infection. These sgRNAs carry the 3-prime untranslated region at their termini, and within these regions the team identified potential open reading frames, designated ORF-UTR3, in MERS-CoV, PEDV, and PDCoV, predicted using non-canonical GUG or UUG initiation codons.

Functional experiments lent biological weight to these predictions. When the researchers engineered a PDCoV ORF-UTR3 variant with its non-canonical GUG start codon converted to AUG and overexpressed it in LLC-PK1 cells, the construct significantly reduced the transcription of IFN-alpha1, IFN-beta, ISG15, and OAS1 upon PDCoV infection, mirroring the immune-suppressive function previously reported for SARS-CoV-2 ORF10. Viral N sgRNA abundance also increased in ORF-UTR3-overexpressing cells, indicating enhanced replication. Taken together, the study proposes that architecturally programmed discontinuous RNA synthesis links non-canonical sgRNAs to genomic plasticity and immune modulation, offering a framework for understanding coronavirus adaptation and potentially informing the design of more stable attenuated vaccines, since deletion-prone stem-loop regions appear to be central to both viral evolvability and passage-based attenuation strategies.

Subject of Research: Architecturally programmed discontinuous transcription and subgenomic RNA synthesis patterns across coronavirus genera

Article Title: Cross-genus analysis reveals architecturally programmed sgRNA synthesis patterns in coronaviruses

Article References: Cross-genus analysis reveals architecturally programmed sgRNA synthesis patterns in coronaviruses. (n.d.). https://doi.org/10.1038/s44320-026-00239-0

Image Credits: AI Generated

DOI: 10.1038/s44320-026-00239-0

Keywords: coronavirus, subgenomic RNA, discontinuous transcription, RNA-RNA interactome, RNA architecture, stem-loop, genomic deletion, ORF10, immune modulation, SARS-CoV-2, PEDV, PDCoV