mRNA initiation and termination are spatially coordinated.

Transcriptional initiation and termination decisions drive messenger RNA (mRNA) isoform diversity but the relationship between them remains poorly understood. By systematically profiling joint usage of transcription start and end sites, we observed that mRNA using upstream starts preferentially use upstream end sites and that the usage of downstream sites is similarly coupled. Our results suggest a positional initiation termination axis (PITA), in which usage of alternative terminal sites are coupled based on their genomic order. PITA is enriched in longer genes with distinct chromatin features. We find that mRNA 5′ start choice directly influences 3′ ends depending on RNA polymerase II trafficking speed. Our results indicate that spatial organization and transcriptional dynamics couple transcription initiation and mRNA 3′ end decisions to define mRNA isoform expression. Editor's summary: The sites where RNA transcription begins and ends determine the full sequence of messenger RNA (mRNA) molecules and, consequently, the function of their encoded proteins. By analyzing both short- and long-read RNA-sequencing data across human tissues and other mammals, Calvo-Roitberg et al. revealed a coordinated relationship between transcription start and end sites that is dictated by the genomic order of alternative sites. This positional coupling, called the positional initiation-termination axis, is most prominent in long genes and is governed by the rate of RNA polymerase II elongation. These findings uncover a molecular mechanism that shapes mRNA diversity across tissues. —Di Jiang INTRODUCTION: Initiation and termination are key steps in the synthesis of mature mRNA molecules. Recent high-throughput analyses have suggested that most mRNA isoform diversity comes from the use of alternative initiation and termination sites. These choices can lead to variable protein conformations or different 5′ and 3′ untranslated regions that influence mRNA localization, translation efficiency, and stability. Thus far, these processing events have primarily been studied in isolation, with little insight into how decisions are coordinated across the transcript to govern the ultimate fate and function of mRNA molecules. RATIONALE: Analyzing short-read RNA-sequencing (RNA-seq) data across human tissues, we observed that the numbers of alternative transcription start sites (TSSs) and alternative 3′ polyadenylation sites (PASs) are correlated across genes. This led us to hypothesize that genes may be structured to use these sites in a coordinated manner. We studied how alternative RNA processing decisions at the terminal ends of genes are coordinated, aiming to better understand the regulation of full-length RNA isoforms. RESULTS: Through a systematic analysis of hundreds of long-read RNA-seq datasets across mammalian tissues, we find that mRNA initiation and termination site choice are directly coupled. Notably, this coupling is based on the order in which sites appear in the genome, with transcripts that start at an upstream TSS preferentially ending at an upstream PAS and those starting at a downstream TSS using a downstream PAS. This revealed a positional initiation termination axis (PITA), which governs coupled mRNA terminal end choices, independent of tissue- or context-specific regulation. Consistently, dCas9-CRISPR perturbations show that mRNA 5′ end choice directly influences mRNA 3′ end choice. PITA is strongly associated with the length of genomic features; PITA genes are longer and transcribed faster, have greater distances between alternative TSSs or PASs, and exhibit distinct chromatin features at TSSs. We see that the rate of RNA Polymerase II (RNAPII) elongation is dependent on where transcription begins, with downstream TSSs associated with faster RNAPII elongation. Overall, we found that PITA coupling depends on a combination of sustained RNAPII trafficking and progressively stronger alternative PAS sequences across longer genes. Specifically, slower RNAPII molecules originating at upstream TSSs are more likely to use weaker upstream PASs, whereas faster RNAPII molecules from downstream TSSs can reach stronger downstream PASs. Together, our data support a model integrating sequence and kinetic features to propose that full-length isoform expression is governed by RNAPII elongation rates within and across human genes. CONCLUSION: This study reveals widespread ordinal coupling between alternative mRNA starts and ends across tissues and species. Our results suggest an interplay between mRNA processing and the evolution of gene architecture. By showing that PITA coupling is governed by transcription elongation dynamics, we extend the classical "window of opportunity" paradigm from splicing to full-length mRNA isoforms. This work lays the foundations for new areas of investigation into the spatial control of mRNA expression and co-transcriptional RNA processing. Transcription elongation rates drive the positional coordination of mRNA initiation and termination.: RNAPII molecules initiating at downstream TSSs elongate faster throughout the gene, coupling the usage of these sites with downstream, stronger PASs in longer genes. [ABSTRACT FROM AUTHOR]