henhaobio
Sign in

bioRxivBiochemistryPreprint

A sequence-encoded promoter proximal super pause stabilizes an offline RNA polymerase II state

Roberto J. Vazquez NuneziD, Serena Kesha, Seychelle M. VosiD

PDFDOI

In-depth summary

Claude Sonnet 4.6 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (24 Jul 2026) · CC BY 4.0

A reconstituted, direct-RNA-sequencing transcription assay on 1,000 human promoter-proximal sequences identified a TFIIS-insensitive consensus super pause, whose cryo-EM structure shows a one-nucleotide sidetracked polymerase state.

Cell-based assays such as NET-seq and PRO-seq are mostly steady-state with limited temporal resolution, and earlier biochemical work covered only a handful of model sequences at low nucleotide concentrations. GATO-seq gives time-resolved, factor-controlled pausing measurements across a library of human sequences.

Key results

  1. ShownA consensus built from TFIIS-insensitive pauses produces a very stable pause lasting beyond 10 minutes. It is only mildly affected by 1 mM NTPs, and the polymerase is neither terminated nor irreversibly arrested.

    RNA extension at 10 μM and 1 mM NTPs kept the pause band prominent over 10 min. G+1 to U substitution gave full-length product at two minutes. Pyrophosphate caused strong pyrophosphorolysis. EC* conversion gave full-length RNA at 30 seconds and 1 minute. · GATO-seq uncovers consensus “super pause”…; RNA polymerase II is not arrested at the super…

  2. ShownCryo-EM of actively transcribing Pol II on the consensus sequence resolved a reversible one-nucleotide backtracked 'sidetracked' state held by a threonine pocket. Overlay with a yeast structure suggests it clashes with TFIIS domain III.

    From 2,546,600 particles, 862,521 showed high-resolution Pol II features. Classes: pretranslocated 23,701, posttranslocated 24,149, sidetracked 36,583. Active site ~2.9 Å. RNA 3′ end bent ~50° vs ~120° in canonical backtracking. Pocket: T850, T854, T1100, T1103. · Cryo-EM structure of RNA polymerase II at the…

  3. ShownMost pause sites are sensitive to TFIIS, but a subset is insensitive. TFIIS also increased the fraction of full-length transcripts.

    873 unique pause sites without TFIIS vs 510 with. 759 sites (59.8%) were seen only without TFIIS, 396 (31.2%) only with, and 113 (8.9%) were shared. Full-length transcripts at 5 min were 29 ± 6% with TFIIS vs 14.8 ± 1% without. · GATO-seq detects TFIIS specific effects on pause…

  4. ShownDSIF-NELF roughly halved the average elongation rate and shortened median pause half-life. Near-physiological NTP concentration sharply reduced the number of detected pause sites.

    Porcine Pol II on human templates: 25.6 ± 3.8 bp/min alone vs 13.3 ± 0.43 bp/min with DSIF-NELF; 101 ± 13 bp/min at 1 mM NTP, likely an underestimate. Pause sites: 1702 (low NTP), 516 (high NTP), 2718 (DSIF-NELF). Median t1/2 4.24 vs 5.71 min with vs without DSIF-NELF. · GATO-seq reports on elongation velocity; GATO-seq reports RNA polymerase II pausing…

  5. SuggestedScanning human promoter-proximal regions found many super pause-like motifs. Pol II profiles at genes carrying them were mixed: ChIP-exo showed a more TSS-focused signal, while PRO-seq showed no difference from background.

    Scan of 10,000 HEK293T protein-coding genes for the 43 sequences found 4,455 sites across 2,227 genes vs 1,819 sites in dinucleotide-matched background. RPB1 ChIP-exo was more focused near the TSS; PRO-seq was similar to background. · RNA polymerase II occupancy is enriched at super…

Methods, data and reproducibility

GATO-seq assembles human promoter-proximal templates with purified porcine Pol II and initiation factors, quenches over time, and maps nascent RNA 3′ ends by direct RNA nanopore sequencing. Candidate pauses were validated by RNA extension assays and cryo-EM.

  • GATO-seq template library: first 262 bp of human genes with focused TSS, fused to an AdML promoter · human (genes); Sus scrofa Pol II · n = 1,000 genes · GEO GSE309328 (reference files)
  • Cryo-EM dataset of Pol II transcribing the consensus super pause sequence for 5 min · Sus scrofa · n = 2,546,600 particles collected
  • Scan of promoter-proximal regions (+1 to +300) of expressed protein-coding genes in HEK293T · human · n = 10,000 genes
  • Published HEK293T PRO-seq and RPB1 ChIP-exo datasets used for the occupancy comparison · human · GSM4730174, GSM4730175 (PRO-seq); GSM6703891 (ChIP-exo)
  • RNA extension assays on synthetic consensus and mutant scaffolds · Sus scrofa · n = three replicates
Methods & tools
GATO-seq, Oxford Nanopore SQK-RNA004 direct RNA sequencing, Dorado 9, Bowtie2 v2.4.5, deepTools v3.5.5, kpLogo, Bedtools v2.31.1, MEME fasta-shuffle-letters and FIMO v5.4.1, GOrilla, Cryo-EM single-particle analysis, ImageJ v1.53K, GraphPad Prism 10
Reagents & environment
Dorado 9 (rna004_130bps_sup@v5.1.0), Bowtie2 v2.4.5, deepTools v3.5.5, Bedtools v2.31.1, MEME v5.4.1, MinKNOW v23.07 and above, ImageJ v1.53K, GraphPad Prism 10.
Experimental conditions
MinION runs of 3-7 h to acquire ~1,000,000 raw reads per library; no computing hardware specified.
Reproducibility gaps
PDB, EMDB, Zenodo code, SRA and GEO processed-data entries are placeholders in the preprint. Some code was initially ChatGPT-generated. Materials are available upon request.
Limitations (5)
  • Experiments used linear DNA without chromatin or topological constraints and a viral AdML promoter, so native promoter sequence effects were not examined. TFIID and Mediator were also absent and may influence pause duration or site selection. · stated by the authors
  • Only three time points were collected, so finer time sampling is needed to model pause behavior. The assay also does not account for premature termination (for example by Integrator or ARMC5). · stated by the authors
  • Poly-A tailing for library preparation could deplete pause sites whose terminal nucleotide is adenosine. · stated by the authors
  • The sidetracked state was resolved only on the consensus sequence. TFIIS exclusion rests on overlay with a yeast TFIIS domain III structure rather than a direct TFIIS-bound structure. · noted in this summary
  • In-cell relevance of super pauses is indirect: ChIP-exo shows a more focused signal, PRO-seq shows no enrichment, and occupancy at individual predicted sites was not validated. · noted in this summary

Read this if you study transcriptional pausing or TFIIS rescue, or want a time-resolved reconstituted assay paired with cryo-EM evidence for a sequence-encoded offline Pol II state.

Citations
2

Abstract

Promoter proximal pausing by RNA polymerase II is critical for regulating gene expression in multicellular eukaryotes. How nucleic acid sequence and protein factors contribute to pausing remains incompletely understood. We developed Gene-specific Analysis of Transcriptional Output (GATO)-seq, which for the first time enables massively parallel, temporally resolved, reconstituted transcription in an assay that uses direct RNA sequencing to map 3'ends of nascent transcripts from a library of human genes. GATO-seq identified a "super pause" sequence that potently induces RNA polymerase II pausing and is not relieved by rescue factor Transcription Factor (TF) IIS. Cryogenic-electron microscopy (cryo-EM) structures of RNA polymerase II on the super pause sequence reveal a previously unobserved, reversible single-nucleotide backtracked state ("sidetracked"), stabilized by a threonine-lined pocket that limits further backtracking. We introduce a powerful in vitro technique that can be employed to study transcription regulation and through its use show that nucleic acid sequence encodes pausing propensity and traps sequence specific offline states, linking sequence to pausing control.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISVishwakarma '23Mann '24Fong '26Saba '18Yang '24Herrero-Ruí '20
70% semantic matchMolecular Biology
Rishi Kishore Vishwakarma, M. Zuhaib Qayyum, Paul Babitzke+1 more·Proceedings of the National Academy of Sciences·Feb 6, 2023· 30 citations