SF3B1-targeted Splicing Inhibition Triggers Global Alterations in Transcriptional Dynamics and R-Loop Metabolism

Daisy Castillo-Guzman, Stella R. Hartono, Lionel A. Sanz, Frédéric Chédin

Preprint posted on June 08, 2020

Debunking the interplay between RNA splicing and R-loops: Old game, new rules?

Selected by Ram


We all heard of single- or double-stranded DNA and RNA. But cells in all living organisms also produce hybrids of RNA and DNA (also called RNA-DNA hybrids). Short stretches of RNA-DNA hybrids are synthesized transiently during transcription and DNA replication, leading to the notion that RNA-DNA hybrids are byproducts of these important genomic processes.

During transcription, the nascent RNA synthesized within the transcription bubble comes out of the RNA Polymerase II (RNAPII) exit channel separated from the duplex DNA. Under favorable conditions, this nascent RNA can re-hybridize with its complementary DNA strand outside the transcription bubble forming a triple-helical nucleic acid structure with an RNA-DNA hybrid and a displaced single-strand of DNA called an R-loop. R-loop structures are ubiquitous and regulated by various cis– and trans-acting factors like sequence composition, DNA superhelicity, nucleosome dynamics, and co-transcriptional RNA processing. Interestingly, the past two decades of research demonstrate that R-loops are biologically relevant and regulate many genomic processes like transcription, DNA recombination and repair.

Besides orchestrating efficient RNA processing, several RNA processing factors were proposed to prevent R-loop formation. For example, RNA splicing factors were thought to prevent R-loops by sequestering nascent RNA from binding to complementary DNA strand and by physically limiting the homology between RNA and DNA through the splicing reaction. Dysregulation of one or more RNA splicing factors induced by inhibition or loss-of-function mutations was proposed to induce aberrant R-loop levels, leading to genome instability in cells. However, mechanistic insights into the aforementioned processes are not clear  (mostly cause of the technical challenges involved in detecting cellular R-loops5). Therefore, the authors of the current study set out to investigate how dysregulation of RNA splicing impacts co-transcriptional R-loops.

Key findings

  1. To understand the impact of RNA splicing on R-loop distribution, the authors chemically inhibited the U2 spliceosome complex using Pladienolide B (PladB) in human immortalized myelogenous leukemia cells (K-562). They gauged genome-wide R-loops, nascent RNA, and steady-state RNA levels using established sequence-based approaches DRIP-seq, EU-seq, and RNA-seq, respectively. They found that PladB mediated splicing inhibition impinges on global transcription and R-loop distribution in K-562 cells.
  2. They demonstrate that, as expected, PladB treatment caused splicing anomalies, predominantly intron-retention, and to a lesser extent skipped-exons. Surprisingly, they found that PladB-mediated R-loop gains only correlated with 2% of genomic locales showing intron-retention. Thus, they report that loci with PladB-mediated intron-retention do not associate with increased R-loops.
  3. Contrary to their initial hypothesis, they found that PladB treatment significantly reduced R-loop levels genome-wide with R-loop losses outnumbering gains (i.e., 15 to 1). PladB reduced R-loops through transcribed gene bodies, with a gradual and directional loss leading to prominent effects at gene termini. Importantly, nascent RNA levels corroborated this global R-loop loss. About 77% of genes exhibiting lower nascent transcription manifested lower R-loops, reinforcing the co-transcriptional nature of R-loops. Their data reiterates earlier findings that splicing dynamics feedbacks onto transcription rates6. Moreover, genes showing R-loop loss harbored a lower pool of RNAPII paused at promoter-proximal regions (indicating that these genes lack a robust store of promoter-proximal paused RNAPII). Thus, they report that PladB-mediated R-loop losses through gene bodies result from an inhibitory effect of splicing inhibition on transcription elongation.
  4. They further investigated the characteristics of PladB-mediated R-loop gains. About two-thirds of PladB-mediated R-loop gains mapped to intergenic regions, extending downstream of RNAPII transcribing genes present in gene-sparse neighborhoods. These new R-loop zones were ~50kb long and again correlated with nascent RNA profiles that manifested a readthrough transcription after the poly-A signals (PAS). Furthermore, PladB treatment exaggerated the pausing signature of RNAPII at promoter-proximal regions followed by deficiencies in RNAPII elongation and distal splicing events at 3’ end of these genes. Thus, PladB treatment seems to alter RNAPII regulation and propel it to readthrough towards downstream intergenic regions post-PAS (in a subset of ~300 genes).
  5. Previous findings suggest that defects in splicing proteins induce aberrant R-loop levels that cause DNA damage. However, the investigators found that PladB treatment only induced γH2AX (DNA damage marker) levels marginally, even at longer time points (24 hrs). Furthermore, they did not find any significant impact of PladB-mediated R-loops on γH2AX levels at gene loci (assessed by ChIP, chromatin immunoprecipitation). Additionally, they also found that PladB treatment reduced the expression of genes involved in DNA damage response pathways. Thus, they suggest that splicing inhibition via PladB might not directly cause R-loop mediated DNA damage.

    Schematic representation of the effects of pre-mRNA splicing on the genomic distribution of R-loops. Taken and modified directly from Castillo-Guzman D et. al., 2020 under a CC-BY 4.0 international license.


The past two decades of research expound on the design principles of R-loop formation and dynamics. The consensus is that dysregulation of RNA processing factors (like splicing factors) leads to aberrant R-loops that cause genome instability, a hallmark of cancer. Moreover, targeting splicing machinery surfaced as an important strategy to alleviate cancer7. Therefore, understanding the mechanistic insights of R-loop regulation could support combinatorial therapeutic strategies to mitigate cancer progression.


I am thankful to all the authors for their support, especially Frédéric Chédin for taking the time in an often busy schedule to comment on the preLight.


Tags: genome instability, r loops, rnapii, splicing, transcription

Posted on: 6th November 2020


Read preprint (No Ratings Yet)

Author's response

Frédéric Chédin (FC) shared

1. The authors data suggest that pharmacological inhibition of splicing regulates transcription and R-loop dynamics. Notably, your data that splicing inhibition profoundly impacts transcription elongation makes total sense, considering the interplay between splicing, transcription, and chromatin architecture. Have you investigated the other factors that could contribute to differential R-loop distribution in cells with splicing defects? For example, local chromatin dynamics and topological constraints, etc.

FC: Good question and short answer: No, we haven’t so far.

2. To reinforce the authors findings, would the authors consider using other splicing inhibitors (like spliceostatin A), or genetically manipulate splicing proteins of U2 spliceosome complex? Or use cancer cells with mutations in U2 spliceosome complex proteins? This may be important to address the effects of PladB on non-splicing events.

FC: We agree that it would be important to expand these findings to other systems. The findings that U2 spliceosome components are mutated in a variety of myelodysplastic syndromes and cancers provide a particularly interesting opportunity.

3. Would the authors agree that inefficient γH2AX induction in PladB treated cells can be explained by the chosen time-points, cell cycle effects, or decreased expression of DNA damage response genes? However, would the authors consider evaluating DNA damage using other means (like qDSB-seq)?

FC: PladB treatment is known to induce only a mild DNA damage response. This could be because the number of additional R-loops caused by U2 spliceosome inhibition is limited. A model system that triggers a larger amount of excessive R-loops could be better suited to test a possible R-loop / DNA break connection. Whether yH2AX or other DSB mapping methods are most appropriate to flesh out the possible co-localization remains to be seen.

4. Recent single-molecule footprinting data8 revises the concept of paused RNAPII and reveals a high RNAPII turnover at promoter-proximal regions? How do the authors think this would impact R-loop dynamics in general?

FC: This is an interesting concept that might particularly matter for short R-loops that were proposed to form dynamically at promoter-pause sites9. How this relates to longer R-loops forming during transcription elongation is less clear.

5. The authors demonstrate readthrough transcription and longer R-loops downstream of a subset of genes. Considering the mutual correlation between R-loops and paused RNAPII, how do the authors reconcile their data? Do the authors predict multiple rounds of non-canonical transcription initiation or pooling of multiple RNAPII downstream of these genes? I can comprehend that this question does not fall under the focus of the current study. However, it would be interesting to hear the authors’ comments on how such large R-loops form?

FC: The R-loops forming downstream of genes result from readthrough transcription at a subset of a few hundred genes. For clarification, we do not believe that these R-loops are individually any longer than any other R-loop observed, most likely around a few hundred base-pairs each10. Collectively, however, they seem to pile up through large R-loop zones matching to regions that become transcribed upon PladB treatment due to readthrough. Overall, one of the strong lessons of this work is that R-loops are sensitive reporters of nascent transcription. Readthrough transcription leads to new R-loops, while reduced transcription in the gene bodies leads to decreased R-loop loads overall.

Have your say

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Register here