كتابة تعليق

Overall assessment

This preprint reports that a subset of chromatin-enriched RNA-binding proteins (RBPs) activate transcription through intrinsically disordered regions (IDRs) whose composition partially resembles that of transcription factor activation domains. Using a classic transcriptional reporter assay, the authors show that selected IDRs activate transcription, and that aromatic residues are required for this activity. A particularly strong result is that the characterized composition of activator IDRs was predictive of new activators, including the CTD of Rpb1. Together with the observation that inducible Rbp1-CTD condensates recruit transcriptional machinery, this work supports a model in which the IDRs of RBPs help recruit coactivators. The identified molecular grammar is compelling, and their model can help explain the coupling of transcription and RNA processing. Overall, the novel transcriptional activator role of RBPs is well supported by the data presented, and the more detailed descriptions of methods and data requested below are mainly meant to allow a better evaluation and understanding of the work.

Major points

1. The conclusion that splicing changes were not associated with up-regulated genes after RBP overexpression is central to attributing upregulation to transcription. Fig S3 shows only alternative 3' splice site (A3SS) events, described as an example and it is unclear whether other splicing defects were tested (skipped exon, alternative 5' splice site, mutually exclusive exon, retained introns, etc). Moreover, other post-transcriptional effects are not examined. A summary across all event classes and all upregulated genes would test more rigorously whether upregulation is secondary to splicing changes. Similarly, a discussion on why other post-transcriptional changes are not considered could strengthen their conclusion.

2. RBP ChIP-seq profiles at upregulated versus neutral super-enhancer-associated genes on Fig. 2D could strongly support a direct link. As presented, however, the analysis is difficult to evaluate. The Results do not state the source of the ChIP-seq data or whether the signal reflects baseline or overexpression conditions, and the number of genes in each profile is not provided. The profiles also show enrichment at the transcription end site (TES) and across the gene body, but the implications of this binding pattern for a TF-like mechanism are not discussed. Specifying the datasets and discussing what the observed occupancy implies about how these RBPs engage these genes would help readers judge whether occupancy is consistent with direct transcriptional regulation.

Minor points

1. "Chromatin-enriched" is not defined in the Methods regarding the selected RBPs on Fig. 1. For example, how far above background is the binding and across how many genes? It is also unclear how the 30 RBPs were chosen and whether others meet the criteria. Describing the criteria would clarify the degree of chromatin binding that was considered.

2. Protein levels are shown for IDRs from three RBPs on Fig. S1B, but FUS and SNRNP70 are also presented as potential activators on Fig. 2D. Showing expression for all fusions of interest would help rule out that differences in luminescence reflect differences in stability rather than intrinsic activation potential.

3. It is unclear whether the TT-seq data in Fig 1A are re-analyzed public data or were generated here. Stating the source of every dataset in the legends or Methods would improve transparency.

4. The co-IP on Fig. S4C of HNRNPA1 and HNRNPH1 with RNAPII is interpreted as suggesting direct recruitment of RNAPII. However, the methods do not describe nuclease treatment, so RNA could bridge the interaction. Repeating the co-IP with benzonase would help distinguish direct protein-protein interaction from RNA-mediated association and would clarify the proposed recruitment model.

5. Only RBM22 is labeled among the activators and non-activators plotted on Fig. S4B, but the "tested RBPs" differ between figures, so it is unclear which RBPs are being analyzed. Labeling all points and/or stating in each legend which RBPs were analyzed and why (for example, in a summary table) would make results easier to compare across figures.

6. The Methods describe the filters applied to identify novel activator RBPs on Fig. 4D-E, but not how the nine predicted activators and three non-activators were chosen among candidates passing them, or what their scores were. Reporting the number of candidates at each step and the score of each tested IDR would allow readers to judge how predictive the grammar is.

7. The Discussion describes the activators as regulators of RNA splicing, but no figure specifically shows this. It is unclear whether this reflects the RBPs chosen for testing or whether splicing regulators are generally more likely to be activators. Supplementary annotation of the selected RBPs (e.g., by function) and a comparison of activator frequency between splicing regulators and other RBPs would support this statement.

I reviewed this preprint as part of the Transparent and Accountable Peer Review training program. This report was written with the support of my research mentor

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.