Comprehensive resolution and classification of the Epstein Barr virus transcriptome

Virus genomes harbor compacted repertoires of genes and regulatory elements. Through long-read sequencing, we provide a comprehensive Epstein Barr virus (EBV) transcriptome analysis, identifying 1453 transcript isoforms and resolving the major isoform of all but one lytic reading frame. Further, we categorize each transcript according to their dependence on viral DNA replication. We show that the late gene viral preinitiation complex, vPIC also activates early promoters/genes, we identify active alternate promoters with distinct dependencies on viral DNA replication, we discover biphasic promoters with embedded features of both early and late promoters. Genetic and chromatin interaction studies identify an enhancer function for the viral lytic origin of replication (OriLyt). We also observe substantial viral read-through transcription that likely causes transcriptional interference and fine tuning of viral promoter activity. In some loci with same direction overlapping gene configurations, polyA read-through is necessary to facilitate transcription through entire ORFs while also giving rise to highly abundant viral lncRNAs due to the partial nature of read-through. Altogether, this study identifies extensive viral transcriptome diversity, it resolves the major isoforms for nearly all lytic ORFs, and it identifies the alternative regulatory modes driving the temporal regulation of EBV lytic gene expression. This study systematically mapped the Epstein-Barr virus (EBV) transcriptome during lytic reactivation using two EBV-positive cell models (Akata and Mutu) and fluorescence-activated cell sorting to purify lytically reactivated cells. High-depth Oxford Nanopore Technologies (ONT) long-read cDNA sequencing was performed to capture full-length viral transcripts. Transcript features were validated and annotated using CAGE-seq (for TSS mapping), short-read RNA-seq (for splice junctions and expression), and ONT 3′ end analysis. Chromatin interactions were assessed with Micro-C/HiC, and polysomal profiling was used to evaluate translation of novel isoforms. Protein structures from new transcripts were predicted with AlphaFold3. All data were processed and integrated using a custom computational pipeline developed in Python, Perl, R, and Bash, enabling comprehensive classification of over 1,400 EBV transcript isoforms and revealing new regulatory mechanisms of viral gene expression.