Endogenous RNAi pathway evolutionarily shapes the destiny of the antisense lncRNAs transcriptome

A genome-wide comparative analysis of “cryptic” aslncRNAs decay in RNAi-capable and RNAi-deficient budding yeasts suggests an evolutionary contribution of RNAi in shaping the aslncRNAs transcriptome.

--An editable version of the final text (.DOC or .DOCX) is needed for copyediting (no PDFs).
--High-resolution figure, supplementary figure and video files uploaded as individual files: See our detailed guidelines for preparing your production-ready images, http://www.life-sciencealliance.org/authors --Summary blurb (enter in submission system): A short text summarizing in a single sentence the study (max. 200 characters including spaces). This text is used in conjunction with the titles of papers, hence should be informative and complementary to the title and running title. It should describe the context and significance of the findings for a general readership; it should be written in the present tense and refer to the work in the third person. Author names should not be mentioned.

B. MANUSCRIPT ORGANIZATION AND FORMATTING:
Full guidelines are available on our Instructions for Authors page, http://www.life-sciencealliance.org/authors We encourage our authors to provide original source data, particularly uncropped/-processed electrophoretic blots and spreadsheets for the main figures of the manuscript. If you would like to add source data, we would welcome one PDF/Excel-file per figure for this information. These files will be linked online as supplementary "Source Data" files. ***IMPORTANT: It is Life Science Alliance policy that if requested, original data images must be made available. Failure to provide original images upon request will result in unavoidable delays in publication. Please ensure that you have access to all original microscopy and blot data images before submitting your revision.*** Reviewer #1 (Comments to the Authors (Required)): The authors report the identification of many ncRNAs in the RNAi-competent yeast N. castellii, analyses of their numbers and distribution relative to S.cerevisiae and characterization of the effects of loss of RNA degradation factors, particularly Dcr1. The experimental and bioinformatic analyses appear to have been well performed and the major conclusions are supported by the data presented.
It was perhaps surprising, and no doubt disappointing for the authors, that the loss of Dcr1 did not have a more pronounced phenotype. However, this is a valuable contribution to the field and I am happy to recommend publication in LSA.
2. Figure 4 and P21: The analysis is described as "immune-FISH", but does not involve hybridization or detection of nucleic acids.
Reviewer #2 (Comments to the Authors (Required)): This study examines the role of the nuclear exosome, the exoribonuclease Xrn1p, and the endoribonuclease Dicer in regulating the antisense (as)lncRNA transcriptome in the budding yeast Naumovozyma castellii. It extends the findings on budding yeast lncRNA transcriptomes of Alcid and Tsukiyama (2016). Alcid and Tsukiyama examine the role of RNAi in restricting aslncRNA expression in N. castellii and demonstrate that a strain lacking the main catalytic subunit of the nuclear exosome (Rrp6p) grow more slowly than wild-type cells, perhaps due to the expression of aslncRNAs in the cytoplasm, which might pair with mRNAs to create a Dicer substrate. Supporting this idea, deleting the gene encoding Dicer in this nuclear exosome mutant strain partially rescues the growth phenotype. In the current study, the authors performed RNA sequencing and reanalyzed the data of Alcid and Tsukiyama. Whereas disrupting Dicer by itself had very little effect on the lncRNA transcriptome, the loss of Dicer in an xrn1 deletion strain increased the expression of antisense Xrn1-sensitive unstable transcripts (asXUTs) and reduced growth (which contrasts to the effect of losing Dicer in the rrp6 deletion background, previously reported by Alcid and Tsukiyama). The current study confirmed that these asXUTs are substrates of Dicer and extended the comparative analysis of aslncRNAs in different yeast species, showing that asXUTs overlap coding regions to a greater degree in S. cerevisiae, while antisense cryptic unstable transcripts (asCUTs) overlap coding regions to a greater degree in N. castellii, suggesting that the nuclear exosome-sensitive aslncRNA transcriptome has expanded in N. castellii. Overall, this article provides increased insight into budding yeast lncRNAs and the potential role that RNAi, the nuclear exosome, and Xrn1p have in restricting the expression of these lncRNAs in the cytoplasm.
Major concern: 1. RNA sequencing experiments were normalized using ERCC RNA spike-ins, which was appropriate, but the small RNA sequencing libraries were "normalized on tRNAs signals." Because the abundance of tRNA fragments can vary in different libraries for reasons that have nothing to do with sequencing depth, normalizing to these fragments is not appropriate. The authors should repeat the small RNA sequencing after adding spike-ins (synthetic RNAs that are ~23 nt in length) that don't map to the N. castellii genome. Alternatively, they could perform small RNA Northerns for several siRNAs using the existing samples and normalize the small RNA sequencing data based on these blots.
Minor points: 1. The number of DUTs listed on the left in figure 1B disagrees with the sum of the sense and antisense transcripts on the right and the number of DUTs listed in the main text.
2. No legends are provided to interpret the heatmaps presented in figures S2A and S2B.
3. There appears to be a discrepancy between the small RNA sequencing data presented in figures S3A and S4B, particularly in the dcr1 deletion strains. Were the data presented in these two figures each of the two biological replicates mentioned in the methods? Do the 23 nt species from the dcr1 deletion strain in figure S4B have other features of siRNAs? 4. The preferred first nucleotide of the 22 nt and 23 nt reads of this study (A) differs from the preferred first nucleotide of the 22 nt and 23 nt genome mapping reads in Drinnenberg et al. 2009 (U). Were reads mapping to rRNA and tRNA removed from the analyses in figures S3A and S4B? Do the small RNAs of these distributions map to the genome or are these the pre-mapped library reads?
5. In the next-to-last paragraph in the results, what do the percentages refer to ("8.1% vs 12.9%")? Also, are the asCUTs and the asXUTs plotted in figure 5E exclusively asCUTs and asXUTs (i.e., the transcripts that are not in the union of the Venn diagram in figure S1E)?
6. In the discussion, a sentence reads, "To which extent the generated small RNAs are properly loaded into Argonaute to mediate post-transcriptional gene silencing, for example at the level of translation regulation, remains unknown." Why is translational regulation proposed as the mechanism for these siRNAs that are known to be capable of directing slicing?
Reviewer #3 (Comments to the Authors (Required)): RNA surveillance pathways play key roles in regulation of the coding and noncoding transcriptome. In particular, several classes of long noncoding RNAs are targeted for degradation by these pathways. Previous studies have identified roles for the 3'-5' exosome ribonuclease, the 5'-3' Xrn1 ribonuclease, and the RNAi pathway in processing of antisense long noncoding RNAs (aslncRNAs). This study examines the interplay these surveillance pathways in the yeast N. castellii, which unlike the more studied S. cerevisiae has a functional RNAi pathway, in addition to the exosome and Xrn1. The authors use genome-wide RNA profiling and other methods to show that (1) the exosome and Xrn1 are primarily responsible for degradation of aslncRNAs in N. castellii, (2) loss of the RNase III family dsRNA ribonuclease of the RNAi pathway, Dcr1, results mainly in the accumulation of Xrn1sensitive lncRNAs, termed XUTs, (3) dcr1 and xrn1 mutants display synergistic growth defects, suggesting that Dcr1 becomes important in the absence of Xrn1, and (4) the exosome-sensitive antisense transcriptome in N. castellii is expanded relative to S. cerevisiae, suggesting that this yeast has adapted to the presence of cytoplasmic RNAi by increasing nuclear RNAi surveillance to prevent aslncRNA-mRNA pairs from becoming RNAi targets. The results are interesting and provide insight into adaptation strategies that allow coordination between RNAi and other RNA surveillance pathways. The comparison of two budding yeasts, one with RNAi and one without, is very powerful for this purpose. The main conclusions of the paper are supported by the results. The paper is suitable for publication in LSA.
I only have minor comments.
1. In addition to aslnRNAs (<200 nt), have the authors considered read through transcription? This would also produce a dsRNA substrate for Dcr1 at convergently transcribed gene pairs. In this regard, overexpression of Dcr1 in S. pombe results in production of siRNAs from nearly all convergent transcription units (Yu et al., Mol Cell. 2014 Jan 23;53(2):262-76), suggesting that Dcr1 can target nearly all sense-antisence RNA pairs.
2. It looks like the authors detect fewer Dcr1 foci using the anti-GFP nanobody fewer fixation compared to live Dcr1-GFP imaging. Can they exclude the possibility of nuclear Dcr1 foci using their live imaging data (imaging of Dcr1-GFP with another nuclear fluorescent protein).
In summary, the results presented in this paper are valuable. The demonstration of a role for Dicer in processing of aslncRNA, and the relationship to other surveillance pathways, is important and raises the possibility that Dicer may perform this function broadly across evolution.

July 22, 2019
Please find hereafter (in italic & blue) our point-by-point answer to the comments of the three reviewers.

Reviewer #1 (Comments to the Authors (Required)):
The authors report the identification of many ncRNAs in the RNAi-competent yeast N. castellii, analyses of their numbers and distribution relative to S.cerevisiae and characterization of the effects of loss of RNA degradation factors, particularly Dcr1. The experimental and bioinformatic analyses appear to have been well performed and the major conclusions are supported by the data presented. It was perhaps surprising, and no doubt disappointing for the authors, that the loss of Dcr1 did not have a more pronounced phenotype. However, this is a valuable contribution to the field and I am happy to recommend publication in LSA. We are grateful to reviewer #1 for the positive feedback, acknowledging the quality of our work and supporting the publication of our manuscript in LSA.
2. Figure 4 and P21: The analysis is described as "immune-FISH", but does not involve hybridization or detection of nucleic acids. We agree with the reviewer's comment. We now use the term "immunofluorescence".
This study examines the role of the nuclear exosome, the exoribonuclease Xrn1p, and the endoribonuclease Dicer in regulating the antisense (as)lncRNA transcriptome in the budding yeast Naumovozyma castellii. It extends the findings on budding yeast lncRNA transcriptomes of Alcid and Tsukiyama (2016). Alcid and Tsukiyama examine the role of RNAi in restricting aslncRNA expression in N. castellii and demonstrate that a strain lacking the main catalytic subunit of the nuclear exosome (Rrp6p) grow more slowly than wild-type cells, perhaps due to the expression of aslncRNAs in the cytoplasm, which might pair with mRNAs to create a Dicer substrate. Supporting this idea, deleting the gene encoding Dicer in this nuclear exosome mutant strain partially rescues the growth phenotype. In the current study, the authors performed RNA sequencing and re-analyzed the data of Alcid and Tsukiyama. Whereas disrupting Dicer by itself had very little effect on the lncRNA transcriptome, the loss of Dicer in an xrn1 deletion strain increased the expression of antisense Xrn1sensitive unstable transcripts (asXUTs) and reduced growth (which contrasts to the effect of losing Dicer in the rrp6 deletion background, previously reported by Alcid and Tsukiyama). The current study confirmed that these asXUTs are substrates of Dicer and extended the comparative analysis of aslncRNAs in different yeast species, showing that asXUTs overlap coding regions to a greater degree in S. cerevisiae, while antisense cryptic unstable transcripts (asCUTs) overlap coding regions to a greater degree in N. castellii, suggesting that the nuclear exosome-sensitive aslncRNA transcriptome has expanded in N. castellii. Overall, this article provides increased insight into budding yeast lncRNAs and the potential role that RNAi, the nuclear exosome, and Xrn1p have in restricting the expression of these lncRNAs in the cytoplasm. We thank reviewer #2 for the critical reading of our manuscript and for her/his positive and constructive feedback.
Major concern: 1. RNA sequencing experiments were normalized using ERCC RNA spike-ins, which was appropriate, but the small RNA sequencing libraries were "normalized on tRNAs signals." Because the abundance of tRNA fragments can vary in different libraries for reasons that have nothing to do with sequencing depth, normalizing to these fragments is not appropriate. The authors should repeat the small RNA sequencing after adding spike-ins (synthetic RNAs that are ~23 nt in length) that don't map to the N. castellii genome. Alternatively, they could perform small RNA Northerns for several siRNAs using the existing samples and normalize the small RNA sequencing data based on these blots. We agree that including RNA spike-in is important. As requested, we have repeated the small RNA sequencing analysis following the addition of an aliquot of total RNA from S. pombe in the total RNA samples from N. castellii. The 22-23 nt small RNAs derived from the centromeric repeats of S. pombe constitute the RNA spike-in used as the reference for the normalization of the small RNA-Seq signals. The results of the new experiment upon normalization on the spike-in signals are very similar to the previous data, based on the normalization on the tRNAs signals. In fact, upon spike-in normalization, we observed that the tag densities for the tRNAs are globally unaffected in the different samples (see Figure R1 above).
Thus, the new small RNA sequencing analysis confirms our initial conclusions, which remain unchanged. Again, we are grateful to reviewer #2 for her/his comment which allowed us to reinforce our conclusions, in a more robust and rigorous manner.
Minor points: 1. The number of DUTs listed on the left in figure 1B disagrees with the sum of the sense and antisense transcripts on the right and the number of DUTs listed in the main text. There was an error in Figure 1B and this has been corrected. There are 10 DUTs.

No legends are provided to interpret the heatmaps presented in figures S2A and S2B.
A scale has been added.
3. There appears to be a discrepancy between the small RNA sequencing data presented in figures S3A and S4B, particularly in the dcr1 deletion strains. Were the data presented in these two figures each of the two biological replicates mentioned in the methods? Do the 23 nt species from the dcr1 deletion strain in figure S4B have other features of siRNAs?
The results shown in the previous Fig S3A and S4B corresponded to distinct datasets (ie different libraries & sequencing). This might explain (at least partly) the minor variations in terms of first base distribution between the two datasets. These figures have been updated in the revised version of the manuscript, as the small RNA-Seq analysis has been repeated (response to major concern). The new libraries for the WT, xrn1, dcr1, xrn1 dcr1 and Dcr1-GFP conditions were constructed and sequenced in parallel, so that the profile obtained for the Dcr1-GFP strain (new Fig S4C) can be directly compared to the profile of the WT and dcr1strains (Fig S3A). The small 23 nt peak detected in dcr1 in the previous Fig S4B, but not in the previous Fig S3A, is absent in the new dataset. 4. The preferred first nucleotide of the 22 nt and 23 nt reads of this study (A) differs from the preferred first nucleotide of the 22 nt and 23 nt genome mapping reads in Drinnenberg et al. 2009 (U). Were reads mapping to rRNA and tRNA removed from the analyses in figures S3A and S4B? Do the small RNAs of these distributions map to the genome or are these the pre-mapped library reads? In the initial version of the manuscript, Fig S3A and S4B used reads that uniquely mapped to the N. castellii genome, with no additional filter to remove the reads matching rRNAs (unlikely as they are repeated sequences) and tRNAs (used for normalization of the signals). In the revised manuscript, the small RNA-Seq analysis has been repeated (response to major comment), and the reads matching to rRNAs and tRNAs were filtered out. In the new Fig S3A and S4C (previously S4B), the preferred first base of the 22-23 nt small RNAs now appears to be 'U', which is consistent with what has been previously reported by Drinnenberg et al (2009). 5. In the next-to-last paragraph in the results, what do the percentages refer to ("8.1% vs 12.9%")? Also, are the asCUTs and the asXUTs plotted in figure 5E exclusively asCUTs and asXUTs (i.e., the transcripts that are not in the union of the Venn diagram in figure S1E)? These percentages correspond to the global coverage of the coding transcriptome by aslncRNAs in N. castellii and S. cerevisiae. This has been clarified in the main text. The numbers in Fig 5E correspond to the full set of asCUTs or asXUTs in each species (ie 868 asCUTs and 622 asXUTs in N. castellii), including those that overlap an asXUT or an asCUT, respectively.
6. In the discussion, a sentence reads, "To which extent the generated small RNAs are properly loaded into Argonaute to mediate post-transcriptional gene silencing, for example at the level of translation regulation, remains unknown." Why is translational regulation proposed as the mechanism for these siRNAs that are known to be capable of directing slicing? This was fully speculative. In absence of experimental data supporting this hypothesis, the sentence has been modified to avoid confusion, now only stating: "To which extent the generated small RNAs are properly loaded into Argonaute to mediate post-transcriptional gene silencing remains unknown."