Paternal RNA contributions in the C. elegans zygote

Transcription

1 Manuscript EMBO Paternal RNA contributions in the C. elegans zygote Marlon Stoeckius, Dominic Grün and Nikolaus Rajewsky Corresponding author: Nikolaus Rajewsky, Max Delbrueck Centrum fuer Molekulare Medizin Review timeline: Submission date: 04 February 2014 Editorial Decision: 21 February 2014 Revision received: 18 April 2014 Editorial Decision: 29 May 2014 Revision received: 01 May 2014 Accepted: 06 May 2014 Editor: Thomas Schwarz-Romond Transaction Report: (Note: With the exception of the correction of typographical or spelling errors that could be a source of ambiguity, letters and reports are not edited. The original formatting of letters and referee reports may not be reflected in this compilation.) 1st Editorial Decision 21 February 2014 Thank you very much for submitting your study that reveals paternal RNA contributions in the C.elegans zygote for consideration to The EMBO Journal editorial office. I received two rather consistent sets of comments as well as verbal encouragement to proceed with the paper from a third expert scientist. As you will see from the written statements, both recommend changes to data presentation as to ease accessibility/comprehension. Further, ref#2 suggests two further reaching experiments (points 4/5) to exclude selected SNP-related effects and assessing the impact of the feminizing mutation on the oocyte transcriptome. With these rather straightforward and constructive suggestions, I am delighted to offer relatively rapid proceedings/invite formal revisions for this study. Please do not hesitate to get in touch regarding feasibility and anticipated timeline for the requested revisions (due to time constrains preferably via ). Please be also reminded that The EMBO Journal only considers one major round of revisions. I do very much look forward to receive/reading a suitably revised version of your study. European Molecular Biology Organization 1

2 REFEREE REPORTS Referee #1: I really liked this paper. I was interested in finding out the results of each section, which is not my usual response to most papers. The use of SNPs to mark the small RNAs of males was clever and allowed genome scale identification of the RNAs transferred. My only critique are very minor ones: 1. in the UV crosslinking experiment, the lethality is either due to the dysfunction of the crosslinked RNA or the dysfunction of the proteins crosslinked to the RNAs. It is the system of RNAs and proteins (and possibly other molecules) that is disrupted by the UV crosslinking. I liked the fact that the crosslinking is only toxic if done early. 2. The identity of the paternal mrnas and paternal sirnas needs to be shared on an Excel spreadsheet rather than a text file that is impossible for this reviewer to decode. On that excel should be some measure of the relative abundance of each mrna and sirna, the gene identity from its source, how well each of these genes is conserved at the nucleotide and at the protein level across Nematoda, or beyond nemotodes for protein coding anyway. Referee #2: The manuscript by Stoeckius et al deal with paternal inheritance of RNA in the nematode C. elegans. The authors demonstrate in a number of independent ways that indeed significant amounts of RNA in the zygote is derived from sperm. By itself this is a very useful set of data that people can mine for their genes of interest. Also biologically it makes an important point, in that paternal RNA contributions can be much more significant than often thought. In general, the work is of high quality and is suitable to be published. I do have some comments though that should be addressed before publication. 1) Presentation-wise the manuscript is not very accessible. For instance, the authors refer to Table S1 in their text, and I had to search online what table S1 was. This appeared to be a pretty rough excel sheet with little to no explanations about the figures shown. I could figure it out, but it would be wise to make this more clear. In addition, the list of genes in this table contains a lot of duplicates. For instance, the authors state they find 6 genes with less than 10% maternal contribution. When I looked into the table, I could find 8. Which made me confused, until I noticed two genes were present twice in that short list. Please clarify, perhaps by providing a small sub-table with those 8 (or perhaps 34) genes. 2) The experiment by which embryonic lethality is induced by mild UV treatment, crosslinking the paternal RNA to protein, is intriguing. However, it is of course also very indirect. I could imagine that also non-incorporated nucleotides may crosslink to proteins, hampering development. The conclusions bound to this experiment should reflect this and need to be softened. 3) I wonder to what extend the paternal loading correlates (or anti-correlates) with rpkm values in the embryo. In other words: the authors should make sure that apparent paternal loads are not caused by stochastic fluctuations in coverage due to low read counts. I ask this, because I find the reproducibility a bit low (around 50%) and wonder how much would be left if the experiment would (hypothetically) be done 10 times. Is there anything specific to the genes that were not reproducible? 4) An important point is to rule out effects of the analyzed SNPs per-sé. This can obviously be easily done by doing the Bristol-Hawaiian cross in the other direction. Ideally this would be done by a seq experiment, but perhaps an assay as in figure 3C (but on a larger set of genes) may also suffice. Feminized CB4856 could most likely be made through RNAi. 5) Finally, the authors should address to what extend the feminizing mutation (fem-1) used may influence the oocyte transcriptome and hence the results obtained. European Molecular Biology Organization 2

3 1st Revision - authors' response 18 April 2014 Point-by-point response (Comments in blue, reply in black, specific edits to the manuscript in red) Referee #1: I really liked this paper. I was interested in finding out the results of each section, which is not my usual response to most papers. The use of SNPs to mark the small RNAs of males was clever and allowed genome scale identification of the RNAs transfered. My only critique are very minor ones: 1. in the UV crosslinking experiment, the lethality is either due to the dysfunction of the crosslinked RNA or the dysfunction of the proteins crosslinked to the RNAs. It is the system of RNAs and proteins (and possibly other molecules) that is disrupted by the UV crosslinking. I liked the fact that the crosslinking is only toxic if done early. We thank the reviewer for the critical comment. We softened the introduction of the experiment and included further discussion about the results in the revised version of the discussion. Results section: This reflects the functional importance of RNA molecules in the worm and therefore this method can provide evidence for phenotypes induced by loss of RNA function. Discussion section: However, these experiments can only reveal potential functionality of the complement of paternally provided RNAs and we cannot exclude that the observed lethality is in part caused by non-incorporated paternal nucleotides crosslinked to proteins or by dysfunction of RNA-binding proteins crosslinked to paternal RNA. 2. The identity of the paternal mrnas and paternal sirnas needs to be shared on an Excel spreadsheet rather than a text file that is impossible for this reviewer to decode. On that excel should be some measure of the relative abundance of each mrna and sirna, the gene identity from its source, how well each of these genes is conserved at the nucleotide and at the protein level across Nematoda, or beyond nemotodes for protein coding anyway. We agree with the reviewer that the Excel spreadsheet provided by us was not very intuitive. We therefore made better Excel spreadsheets available for the revised version of the manuscript as supplementary tables with detailed legends. As requested by the reviewer we provide expression quantification for mrnas and small RNAs along with information on evolutionary conservation. We extracted PhastCons conservation scores from genome-wide alignments across six nematode species since gene models are not well annotated in other nematode species and therefore orthologs cannot be aligned directly. For this reason we also do not have annotated open reading frames in other nematode species and prefer to only quantify conservation on the nucleotide level. For mrnas, we also performed protein alignments to the human proteome and supply predicted homologs based on best hits form blast alignments (E-value < 10-10). These predicted homologs were also included in the excel sheet. In summary, we added the following supplementary tables to the revised version of the manuscript: Table S1: Paternal contribution of protein-coding mrnas. Table S2: Expression of protein-coding mrnas in our measured samples. European Molecular Biology Organization 3

4 Table S3: Paternal contribution of 21U RNAs and endo-sirnas. Table S4: Expression of 21U RNAs and endo-sirnas in oocytes, 1-cell embryos, and sperm/spermatocytes. Referee #2: The manuscript by Stoeckius et al deal with paternal inheritance of RNA in the nematode C. elegans. The authors demonstrate in a number of independent ways that indeed significant amounts of RNA in the zygote is derived from sperm. By itself this is a very useful set of data that people can mine for their genes of interest. Also biologically it makes an important point, in that paternal RNA contributions can be much more significant than often thought. In general, the work is of high quality and is suitable to be published. I do have some comments though that should be addressed before publication. 1) Presentation-wise the manuscript is not very accessible. For instance, the authors refer to Table S1 in their text, and I had to search online what table S1 was. This appeared to be a pretty rough excel sheet with little to no explanations about the figures shown. I could figure it out, but it would be wise to make this more clear. In addition, the list of genes in this table contains a lot of duplicates. For instance, the authors state they find 6 genes with less than 10% maternal contribution. When I looked into the table, I could find 8. Which made me confused, until I noticed two genes were present twice in that short list. Please clarify, perhaps by providing a small sub-table with those 8 (or perhaps 34) genes. We thank the reviewer for the comment. Please see our response to reviewer 1, comment 2. 2) The experiment by which embryonic lethality is induced by mild UV treatment, crosslinking the paternal RNA to protein, is intriguing. However, it is of course also very indirect. I could imagine that also non-incorporated nucleotides may crosslink to proteins, hampering development. The conclusions bound to this experiment should reflect this and need to be softened. Please see response item 1 to Ref 1. 3) I wonder to what extend the paternal loading correlates (or anti-correlates) with rpkm values in the embryo. In other words: the authors should make sure that apparent paternal loads are not caused by stochastic fluctuations in coverage due to low read counts. I ask this, because I find the reproducibility a bit low (around 50%) and wonder how much would be left if the experiment would (hypothetically) be done 10 times. Is there anything specific to the genes that were not reproducible? We thank the referee for these helpful comments. We agree that stochastic fluctuations due to low read coverage could be a confounding factor. However, we observed that the fraction of SNPs that is reproducible across replicates only weakly depends on the expression cutoff (although the total number of SNPs does decrease at higher cutoffs). We therefore think that most of the observed variability is due to slightly different biological conditions in the two replicates and hypothesized that the paternal contribution might be sensitive to external conditions. We therefore had a closer look at the set of paternal genes which are reproducible and the ones which were not reproducible across the two replicates. GO functional enrichment analysis revealed that genes associated with regulation of growth rate (P < ) and cellular component size (P < ) are over-represented among genes with reproducible paternal component. In contrast, stress response was the most European Molecular Biology Organization 4

5 highly enriched functional category for genes with non-reproducible paternal component (P < ) while it was not over-represented among the reproducible genes. We added this result to the revised version of the manuscript: About half of the paternal contribution is reproduced in our two independent experiments (Figure 3B). Could the non-reproducibility of the other half be due to low read coverage or expression? This seems not to be the case since the fraction of paternal SNPs in transcripts that were reproduced only weakly depends on expression levels. We therefore think that most of the observed variability is due to slightly different biological conditions in the two replicates and hypothesized that part of the paternal contribution might be sensitive to external conditions. Consistently, GO term functional enrichment analysis revealed that genes associated with regulation of growth rate (P < 1.6e-7) and cellular component size (P < 2.3e-5) were highly statistically significantly over-represented among genes with reproducible paternal component, while stress response was the most highly significantly enriched functional category for genes with non-reproducible paternal component (P < 4e-10) and not statistically significantly over-represented among the reproducible genes. 4) An important point is to rule out effects of the analyzed SNPs per-sé. This can obviously be easily done by doing the Bristol-Hawaiian cross in the other direction. Ideally this would be done by a seq experiment, but perhaps an assay as in figure 3C (but on a larger set of genes) may also suffice. Feminized CB4856 could most likely be made through RNAi. We agree with the reviewer that it is important to address potential biases of our analysis because of the SNPs or genetic background of the strains used. However, we believe the insights from the proposed reverse experiment are limited. In case the reverse experiment would yield the same set of paternally derived mrnas in the 1-cell embryo, one could conclude that this set is independent of the nematode strain of origin. A meaningful insight, but it would not alter our conclusion that there is reproducible paternal mrna in the embryo. However, if there would be differences, we would conclude that the paternal contribution may at least in part depend on the species used. Also this result would not alter our current conclusion. In any case, the proposed experiment is difficult and data will be hard to interpret because: a) It requires RNAi of fem-1 (or spe-9) in the Hawaiian instead of using a temperature sensitive allele. To obtain enough 1-cell embryos from the cross we performed the cross with over feminized worms. Performing RNAi on a large population of worms can result in variable occurrence of the sterile/feminizing phenotype, which is a prerequisite for the SNP analysis. Embryos from non-sterile worms can be mistaken for hybrid embryos from the cross and thus paternal contribution calculated from these data will potentially be an underestimate. b) It requires hand-collection of 1-cell stage hybrid embryos instead of automated FACS purification. The temperature sensitive sterile strain that we have been using in our crosses also expressed OMA-1-GFP in the oocytes and early embryos. This allowed us to specifically sort the 1-cell stage embryos resulting from the cross in large numbers for sequencing, and the GFP also allowed us to distinguish true hybrid (GFP positive) embryos from 1-cell (GFP negative) embryos derived from Hawaiian hermaphrodites that are still present in very low numbers after Hawaiian male purification by filtration. Thus, hand collection of embryos after the cross can potentially enrich for 1-cell stage embryos from Hawaiian hermaphrodites and therefore these data will overestimate the paternal contribution. Taken together, we believe that the data obtained from the reverse experiment European Molecular Biology Organization 5

6 can be hard to interpret. We therefore included another entirely independent dataset from a parallel study (now submitted also to EMBO J) where we profiled absolute mrna expression between oocytes and 1-cell embryos from N2 worms as another results and discussion paragraph in the manuscript. Because the 1-cell embryo is thought to be transcriptionally silent, we would expect transcripts delivered by sperm into the oocyte upon fertilization to be up-regulated in the 1-cell embryo compared to the unfertilized oocyte. Indeed, we observed numerous significantly up-regulated mrnas in 1-cell embryos in this study. 18 out of the 31 (60%) more than threefold significantly up-regulated mrnas have indeed a paternal contribution from our SNP analysis in the (completely independent) crossing experiment. The number of genes with paternal contribution is therefore highly significantly over-represented among the up-regulated genes (Fisher s exact test P ~ 0). Conversely, out of all 164 transcripts with a paternal contribution, 24 (15%) were significantly up-regulated while 87 (53%) did not change significantly. Up-regulated genes are thus also highly significantly enriched among genes with paternal contribution (Fisher s exact test P < 10-10). Moreover in another approach, we compared mrna expression in oocytes, 1- cell stage embryos and sperm/spermatocytes. mrnas that are up-regulated in the 1-cell embryo are also more highly expressed in sperm/spermatocytes compared to oocytes, which further suggests that the majority of up-regulated transcripts are of paternal origin. We believe that these results from a different experimental setup and different worm strains (for oocytes we used N2; for 1-cell stage embryos an N2 based GFP reporter strain; both hermaphrodites) further substantiate the conclusions drawn in our manuscript. We refer to these results in the revised version of the manuscript: Transcripts with paternal SNPs in the zygote are also upregulated at the oocyte-to-embryo transition. In C. elegans, RNA polymerase II is transcriptionally silent in the embryo until the 3-4-cell embryo stage (Guven-Ozkan et al, 2008; Seydoux & Dunn, 1997; Baugh, 2003). Thus, we reasoned that if sperm transports mrnas into the oocyte we would expect these transcripts to be upregulated in the 1-cell embryo compared to the unfertilized oocyte. We thus closely examined our mrna-sequencing data of a parallel study (Stoeckius et al., submitted), where we profiled mrna expression by sequencing of polyadenylated transcripts (mrna-seq) in oocytes and 1-cell stage embryos in N2 worms, i.e. in different genetic backgrounds compared to the crossing experiment. The data suggest that 193 mrnas are significantly upregulated more than 2-fold upon fertilization (P < 0.05). Strikingly, 14 out of 16 mrnas that are upregulated more than 4-fold after fertilization contain paternal SNPs in our Hawaiian crossing experiment arguing for a paternal contribution (Figure 3D). We next tested if we could quantitatively validate the paternal mrna contribution by subtracting the amount of paternal transcripts as computed from the SNP frequency from the transcript level in the 1-cell embryo to achieve an improved regression to the oocyte mrna or small RNA expression levels. Interestingly, in comparison to the unmodified embryonic transcript levels, we observed a significant increase in correlation to oocyte mrna levels after in silico subtraction of the paternal contribution from 1-cell mrna expression (R2=0.78 versus R2=0.66, Supplementary Figure 2B). The significance of this observation was reflected by a clear shift of the distribution of residuals (P < 1e-9, Wilcoxon s rank sum test, Supplementary Figure 2B). Moreover, in another independent approach we directly compared mrna expression in oocytes and 1-cell stage embryos to sperm (and spermatocytes). For the majority (>80%) of mrnas that are highly upregulated in the 1-cell embryo we observe higher expression in sperm (and European Molecular Biology Organization 6

7 spermatocytes) compared to oocytes, which further suggests that the upregulated transcripts are of paternal origin (Supplementary Figure 2B). Notably, transcripts coding for structural sperm proteins that are among the most highly expressed genes in sperm (and spermatocytes) (e.g. major sperm proteins; Supplementary Figure 1A) do not appear to be transferred into the embryo, which argues for a specific selection of mrnas that are transferred into the oocyte upon fertilization. Transcripts that are up-regulated (>2-fold) in the zygote are apparently very lowly expressed in oocytes (on average ~0.3 RPKM) and exhibit still very low, but significantly higher expression in 1-cell stage embryos (on average ~3.2 RPKM) compared to oocytes, thus leading to a strong positive fold-change. For comparison, the 2,000 most highly expressed transcripts in the 1-cell stage have RPKM values ranging from 100 to 24,000 with a mean of around 300 RPKM. The low expression of these genes in the 1-cell would be consistent with this hypothesis, given the small volume of sperm, limiting its capacity to deliver mrnas into the zygote. Together, these data provide additional evidence for sperm derived mrnas in the zygote. 5) Finally, the authors should address to what extend the feminizing mutation (fem-1) used may influence the oocyte transcriptome and hence the results obtained. We agree with the reviewer that feminizing mutations can potentially influence the oocyte transcriptome, although we have not yet observed this using different feminization mutations (e.g. spe-9 and fem-1) interrogating -1 oocytes for other studies. We believe that the large overlap of the paternal mrna candidates from the crossing experiment with the mrnas that are up-regulated in the oocyte to 1- cell (where wildtype oocytes were used from N2 hermaphrodites) argues against an effect of the feminizing mutation on our conclusions. 2nd Editorial Decision 29 May 2014 I received final comments from the critical referee on your revised paper. As discussed, this scientist is currently also assessing the related submission from your lab. Accordingly, s/he is identifying two items that should be presented more consistently between the two studies. I have to note though, that this does not yet indicate potential further proceedings of the newly submitted paper! Conditioned on appropriate consolidation in the text file, I am really pleased to proceed with formal acceptance of the paternal RNA-contribution paper. For efficient editorial proceedings, I kindly ask you to attend to the specific requests below: - Your last article was split in two parts and uploaded as PDF-files. Also, the section "Conflict of interest" was missing. Please include this section and ensure that your final article is uploaded as one text file in Word-format. - Your last figures and SI-figures were uploaded within one PDF. => Please upload all main figures as individual figure files (1 file per figure) and upload the supplementary figures as individual "Expanded view content"-files. - You had uploaded several datasets, which is fine, but please ensure that they are all referred to appropriately in your article text. It would be most helpful if you were to provide a short, (two up to four bullet-points) summary that outlines major advance provided by your study! European Molecular Biology Organization 7

8 We have also the opportunity to graphically feature the paper and kindly ask you to suggest a summarizing image 550 x 150(max 400) pixel in size. I do take the liberty to congratulate you at this stage and would be intrigued to receive an ultimate version for rapid publication in The EMBO Journal. REFEREE REPORT Referee #3: The revised version of this manuscript represents a very nice piece of work that can be published in EMBO J as it is. The concerns have been addressed adequately. I have one remark though: the authors should be more consistent in their remarks/explanations between the two manurscripts that are currently under review at EMBO J. For instance, in their rebuttal to this manuscript the authors state that they have not observed influence of fem-1 mutation in the transcripttome of oocytes, while in the parallel paper they write that commonly used methods used to isolate oocytes (that make use of fem-1 mutants; suppl figure 5B in the parallel paper) do result in changes in the transcriptome. This is not a major point of these manuscripts, but such inconsistencies may raise some doubts. A second case is related to the observation that a significant number of genes whose mrna levels go up upon fertilization are in fact paternally providing some mrna. In the accompanying paper that actually describes this data the authors seem to favor another explanation, however, namely that these mrnas become re-adenylated, and hence that they are therefore more effectively detected. Both explanations may be true, but the authors cannot use one explanation in one manuscript, and one in another. To be concrete, the re-adenylation explanation should also be voiced in the manuscript at hand (Paternal RNA contributions in the C. elegans zygote). 2nd Revision - authors' response 01 May 2014 (Comments in blue, reply in black, specific edits to the manuscript in red) Referee #3: The revised version of this manuscript represents a very nice piece of work that can be published in EMBO J as it is. The concerns have been addressed adequately. I have one remark though: the authors should be more consistent in their remarks/explanations between the two manuscripts that are currently under review at EMBO J. For instance, in their rebuttal to this manuscript the authors state that they have not observed influence of fem-1 mutation in the transcriptome of oocytes, while in the parallel paper they write that commonly used methods used to isolate oocytes (that make use of fem-1 mutants; suppl figure 5B in the parallel paper) do result in changes in the transcriptome. This is not a major point of these manuscripts, but such inconsistencies may raise some doubts. We agree with the reviewer that we did not make this important point clear enough in the rebuttal letter. Commonly used methods to obtain oocytes in biochemical quantities use sterile feminized (e.g. fem-1 or spe-9) worms. These worms accumulate unfertilized oocytes in their uterus that can be isolated by cutting the animals and specifically filtrating the oocytes. It is known that uterusaccumulated oocytes can undergo meiosis and endoreduplication (Gu et al., 2014; McNally and McNally, 2005) and are most likely not fertilization competent. European Molecular Biology Organization 8

9 We observed that oocytes that accumulate in the uterus have a different structural appearance compared to -1 oocytes (in both, wildtype as well as in feminized worms) and, as we describe in the accompanying paper, also observe changes in transcript and protein content in the uterus-accumulated, easy to purify oocyte population. At least in PCRs we do not observe these differences in hand-dissected -1 oocytes from feminized worms. The -1 oocytes reside in the distal germline next to the spermatheca. Thus, taking the literature and our observations into account we believe that the oocytes that are fertilized in feminized worms when crossing them to wildtype males are the -1 oocytes and not those that accumunated in the uterus. We will also clarify this more precisely in the accompanying manuscript that is currently under review at EMBO J. In any case, it is true that we can not fully exclude at least some influence of the feminizing mutations on the oocyte transcriptome, but we believe that the large overlap of the paternal mrna candidates from the crossing experiment (where fem-1 worms were used) with the mrnas that are up-regulated in the oocyte to 1-cell (where wildtype oocytes were used) argues against an effect of the feminizing mutation on our conclusions. A second case is related to the observation that a significant number of genes whose mrna levels go up upon fertilization are in fact paternally providing some mrna. In the accompanying paper that actually describes this data the authors seem to favor another explanation, however, namely that these mrnas become re-adenylated, and hence that they are therefore more effectively detected. Both explanations may be true, but the authors cannot use one explanation in one manuscript, and one in another. To be concrete, the re-adenylation explanation should also be voiced in the manuscript at hand (Paternal RNA contributions in the C. elegans zygote). We agree with the reviewer that it is important to also address readenylation in the paternal RNA manuscript. We will also further address this issue in the accompanying paper that we submitted to EMBO J. We added one sentence to the discussion of the results in the manuscript: Over 80% of these strongly upregulated mrnas have a paternal contribution as assayed by our crossing experiment (Figure 3D, Figure E2B). However, the paternal contribution estimated based on the SNP frequency does not explain the full magnitude of up-regulation in most cases. In addition to paternally contributed transcripts, readenylation of maternal mrnas can contribute to the observed fold change. Additional comment: While this paper was under review, an exciting study has been published by the Mansuy lab, suggesting that in stressed male mice can inherit small non-coding RNAs to the next generation through sperm (Gapp et al., 2014). Also we became aware of a study from the Hirsh lab in 1980 dissecting parental contribution of 24 temperature sensitive embryonic lethal phenotypes (Wood et al., 1980). The paper describes one paternal effect mutant. Today it is known that this gene - zyg-8 - encodes a protein kinase essential for mitosis in the one-cell embryo; hence, another paternally inherited essential protein - comparable to SPE-11 that is also sperm provided and essential for early embryogenesis. We are citing both papers in the final version of the manuscript and thank David Hirsh in the Acknowledgments for pointing out Wood et al European Molecular Biology Organization 9