Cohesin-independent STAG proteins interact with RNA and R-loops and promote complex loading

Essential revisions:

1) The R-loop IFs shown detect large spots of fluorescence that seem to mark the nucleolus. This needs to be clarified, using a nucleolus marker, since it would be possible that most of the effect related to R loops are occurring mainly at the rDNA regions

Based on existing literature, it is expected that a strong R-loop signal would be detected within the nucleolus. R-loops have been identified within the nucleolus by IF ^1,2 and specifically at the rDNA locus using ChIP-qPCR ³. Indeed, our results recapitulate these observations, showing S9.6 signal within a Nucleolin mask by IF and S9.6 peaks within the intergenic spacer (IGS) of the rDNA consensus (Author response image 1a, b).

Author response image 1

Download asset Open asset

We agree that some of the effect related to SA proteins and R-loops described here may indeed be occurring at rDNA. In fact, the literature supports this possibility. For example, it is known that cohesin is necessary for nucleolar integrity in yeast. Cohesin has been shown to bind to the nontranscribed region of the rDNA locus ⁴ and the 35S and 5S genes form loops that are dependent on Eco1, the cohesin subunit known to acetylate Smc3 and stabilize cohesin on chromatin ⁵. Consequently, yeast with eco1 mutations exhibit disorganised nucleolar structure and defective ribosome biogenesis. Similarly, we have discovered that Stag1 regulates nascent rDNA transcription and nucleolar integrity for mouse embryonic stem cell (mESC) pluripotency ⁶. This is likely to be direct since Stag1 is bound to both the mouse and human rDNA coding region and IGS, as well as at LINE and SINE repetitive elements important for nucleolar biology. Importantly, SA1/2 binding at rDNA and SINEs is maintained in IAA-treated RAD21^mAC cells and SA1^ΔCoh interacts with nucleolar proteins involved in rDNA transcription and processing, including Polr1a, UBF1 in the FC; Fibrillarin (FBL) in the DFC; and Nucleophosmin (NPM) and Nucleolin (NCL) in the GC (Review Figure 1c). Taken together, we argue that SA1 has an important and conserved role in maintaining nucleolar R-loops and structure.

2) Authors should show by DRIP-qPCR and ChIP of SA at a number of regions that SA is enriched at the sites where R loops are enriched and that recruitment is diminished by RNH treatment. Similar experiments could be performed by using inhibitors of transcription such as DRB or some other.

To further support our observations that SA proteins are localised to R-loops, we have prepared DRIP-seq in RAD21^mAC cells treated with RNase H (RNH), to confirm specificity of the DRIP signal, and with EtOH or IAA to assess the impact of cohesin loss on R-loops. We combined these datasets with our ChIP-seq for SA proteins and RAD21 in EtOH or IAA conditions. These results are discussed below.

Despite our best efforts, we were unable to address the question of whether SA1^ΔCoh is diminished by RNH treatment. First, RAD21^mAC cells experienced massive cell death after both IAA treatment and RNH overexpression. Next, we did manage to prepare SA ChIP-seq and DRIP-seq libraries upon triptolide treatment to inhibit transcription, however the ChIP signal/noise was extremely poor and we were not confident in these results. In the interest of time and resources, we proceeded with the analysis of R-loop and SA using DRIP-seq.

This revealed several interesting observations which have now been incorporated into the revised manuscript in a new layout (Lines 296-350, 1160-1186, Figure 3h, i, Figure 3 —figure supplement 1), and to our existing accession GSE167887. We confirmed the specificity of the experiment by identifying RNH-sensitive R-loops, among which we observed two regimes of SA association with R-loops. A small proportion of R-loop sites directly overlapped with SA1/2 in control conditions. These sites were enriched at genes and the SA signal was sensitive to RAD21 loss. On the other hand, a larger proportion of R-loops had SA signals adjacent (bound within 2kb of the R-loop peak). Interestingly, these SA sites were enriched in repressed chromatin and were not sensitive to RAD21 loss, in fact they were enriched.

Thus, our results support those already provided from IF, STORM, mass spec and coIP to show that cohesin and SA are localised to R-loops in control conditions, and that acute depletion of RAD21 leads to the continued associated of residual SA proteins at R-loops.

3) The IF and global IP data shown in gels need to be quantified and measured properly with right controls. In this sense, authors seem to be concluding that the stability of SA proteins depends on its binding to RNA. This could be tested more specifically bby showing that after RNH treatment the protein disappear at higher speed whereas other proteins remain.

Thank you for raising this point. While our results show that SA1^ΔCoh binds RNA and interacts with a number of RBP, we acknowledge that we do not definitively know that the stability of SA proteins (with or without cohesin) depends on its binding to RNA. We acknowledge that the treatment of RNases can impact global levels of many proteins and that the observation that SA1^ΔCoh is reduced upon RNH treatment is not sufficient in this context. Thus, we have removed the data from the manuscript and figures.

4) Provide experimental evidence of the existence of SA1 and/or SA2 bound to chromatin in isolation (ie without other components of the cohesin complex) in unperturbed cells. If data exists in the literature already, this can also be sufficient to infer their model. No evidence is provided of the existence of SA1 (or SA2) complexes bound to R-loops (or anywhere else) in normal circumstances (ie without induced degradation of RAD21). A Chip experiment (SA1 and/or SA2 vs RAD21) or co-localization imaging data would be at least supportive for the existence of SA1 and SA2 independent of the cohesin complex. Otherwise, we might be well just looking at remnants of what were cohesin holo-complexes after induced degradation of RAD21 subunits. Thus,

There are a number of pieces of evidence in the literature to support the existence of SA proteins on chromatin without core cohesin. These include a role for SA1 at telomeres ⁷; the effect of NIPBL loss on SA levels in hepatocytes (Figure 1) ⁸; and in vitro experiments of SAs ability to bind nucleic acids without cohesin ⁹.

To formally address this in our cell system, we have performed a GFP-TRAP in untreated (EtOH) RAD21^mAC cells. The results show efficient TRAP of RAD21, interactions with SA proteins as expected and ~20% ‘residual’ SA protein in the flow-through material after GFP-TRAP. We note that this amount of chromatin-bound SA which is not interacting with RAD21 in unperturbed cells is reminiscent of the amount of ‘residual’ chromatin-bound SA we observe upon acute degradation of RAD21. This result has been added to Figure 1 —figure supplement 1 and commented on at Lines 116-120 in the manuscript where we have also included Reference 7 above.

5) The analysis only considers exon 31/32 inclusion in HCT cells. It would be interesting to know whether the pattern seen here is consistent in other cell types, including non-cancer cell lines – this would indicate how relevant the specific SA1/2 differences seen here are to other cell types.

We have now included an additional analysis of exon 31/32 inclusion data for SA proteins from mouse embryonic stem (mESC) and mouse Neural progenitor cell (mNPC) RNA-seq datasets, including ones we have produced ourselves as well as published work. The data show that the splicing in of exon 31 of SA1 and the splicing out of exon 32 of SA2 is conserved across cell types and species, arguing for an important role for this splicing event in SA function in general. This data has been added to Figure 5 —figure supplement 1 and commented on at Line 456-457 in the manuscript.

References:

1. García-Rubio, M. L. et al. The Fanconi Anemia Pathway Protects Genome Integrity from R-loops. PLoS Genet. 11, e1005674 (2015).

2. Barroso, S. et al. The DNA damage response acts as a safeguard against harmful DNA-RNA hybrids of different origins. EMBO Rep. 20, e47250 (2019).

3. Abraham, K. J. et al. Nucleolar RNA polymerase II drives ribosome biogenesis. Nature 1–27 (2020). doi:10.1038/s41586-020-2497-0

4. Laloraya, S., Guacci, V. & Koshland, D. Chromosomal addresses of the cohesin component Mcd1p. Journal of Cell Biology 151, 1047–1056 (2000).

5. Harris, B. et al. Cohesion promotes nucleolar structure and function. Mol Biol Cell 25, 337–346 (2014).

6. Pezic, D. et al. bioRxiv 1–60 (2021). doi:10.1101/2021.02.14.429938

7. Bisht, K. K., Daniloski, Z. & Smith, S. SA1 binds directly to DNA through its unique AT-hook to promote sister chromatid cohesion at telomeres. J. Cell. Sci. 126, 3493–3503 (2013).

8. Schwarzer, W. et al. Two independent modes of chromatin organization revealed by cohesin removal. Nature 551, 51–56 (2017).

9. Pan, H. et al. Cohesin SA1 and SA2 are RNA binding proteins that localize to RNA containing regions on DNA. Nucleic Acids Research 24, 105–17 (2020).

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

We consider that the manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below, before its final acceptance.

The authors have improved considerably the manuscript by clarifying a number of issues raised in the first revision. As a consequence of this, conclusions are supported in its large part. However, the main conclusion that SA has a direct functional role in R-loop regulation is not fully supported. Given the association of many proteins with transcribed regions, it is not unexpected that part of those regions may form R-loops, so an overlap of binding is observed. However, the fact that authors are not able to find a reduction by treating with RNH is worrying and provided that no biochemistry is added either. The study is certainly interesting, but conclusions in this regard should be lessened, particularly in the title and Abstract, and limitations regarding this point of the study raised in the Discussion.

As the SA ChIP in RNH conditions was not technically possible, we agree that we cannot definitively comment on whether SA has a role in R-loop regulation per se. We are however confident in the observation that cohesin-independent SA proteins interact with RBP and localise to R-loops in HCT cells. Thus, we have made adjustments to the Title, Abstract and Discussion to reflect this.

– The title has been changed from “Cohesin-independent STAG proteins interact with RNA and localise to R-loops to promote complex loading” to “Cohesin-independent STAG proteins interact with RNA and R-loops and promote complex loading”.

– We have also removed the comment that STAG proteins regulate R-loops from the abstract and replaced with “Accordingly, SA proteins interact with RNA, RNA binding proteins and R-loops, even in the absence of cohesin”.

Two comments in the Discussion have been changed:

1. Line 542 “…or perhaps sequentially, and facilitate the loading of cohesin. Future work investigating the binding profiles of SA proteins and re-loaded cohesin following R-loop depletion in vivo, would help to confirm these findings. Furthermore, biochemical assessment of SA interaction with R-loops will help identify the mechanisms of SA’s roles at R-loops both in the presence or absence of the cohesin ring”.

2. Line 569 “This study also reveals SA1 as a novel regulator of R-loop homeostasis” has been removed and replaced with “Our study suggests that SA1 may act as a novel regulator of R-loop homeostasis”.