STRAP regulates alternative splicing fidelity during lineage commitment of mouse embryonic stem cells

Alternative splicing (AS) is involved in cell fate decisions and embryonic development. However, regulation of these processes is poorly understood. Here, we have identified the serine threonine kinase receptor-associated protein (STRAP) as a putative spliceosome-associated factor. Upon Strap deletion, there are numerous AS events observed in mouse embryoid bodies (EBs) undergoing a neuroectoderm-like state. Global mapping of STRAP-RNA binding in mouse embryos by enhanced-CLIP sequencing (eCLIP-seq) reveals that STRAP preferably targets transcripts for nervous system development and regulates AS through preferred binding positions, as demonstrated for two neuronal-specific genes, Nnat and Mark3. We have found that STRAP involves in the assembly of 17S U2 snRNP proteins. Moreover, in Xenopus, loss of Strap leads to impeded lineage differentiation in embryos, delayed neural tube closure, and altered exon skipping. Collectively, our findings reveal a previously unknown function of STRAP in mediating the splicing networks of lineage commitment, alteration of which may be involved in early embryonic lethality in mice.

4. Fig3h-i: Variability between the two biological replicates appears very high. how robust are these targets controlled by STRAP? For example, plekhg4 in fig3i does not appear to be regulated in strap KO. 5. Supp fig3d: looks like a large variance between biological replicates? Reviewer #2 (Remarks to the Author): This manuscript reports a study of STRAP as a novel regulator of alternative splicing in early mouse embryos. The authors carried out global studies of STRAP using RNA-seq, eCLIP-seq, complemented by experimental validations and individual assays, in knockout models of mouse embryoid body. In addition, co-IP analyses showed that STRAP is a putative spliceosome-associated factor. The study reported a relatively small number of AS events that may be regulated by STRAP. The authors also extended the study to Xenopus and showed that loss of Strap led to delayed neural tube closure. Overall, this is an interesting study that revealed a novel splicing regulator. However, there are a number of major concerns that need to be addressed.
A major concern is the relatively small number of AS events detected by RNA-seq in response to the loss of Strap. Furthermore, an even smaller number of these AS events had eCLIP peaks in their immediate neighborhood (in the exon or flanking introns). If STRAP is a putative spliceosome-associated factor, there expects to be a much larger pool of its splicing regulatory targets. The small number of AS events is likely due to the fact that only 2 biological replicates were included for each group in the RNA-seq study. Current practice standard is to include 3-5 replicates at least to better estimate sample-to-sample variation. Another reason that could underlie the small number of significant events is the limited analysis method for differential splicing discovery in RNA-seq. It is well known that existing methodologies are limited and different methods could yield very different results. It is strongly recommended that the authors use alternative methods for this analysis, in addition to rMATS, such as MAJIQ, LeafCutter, etc. Outputs of different methods should be compared, and combined in some way to enhance the quality of their results.
To make the point that N2B27-induced EBs molecularly resemble the mouse embryonic germ layer specification, the authors listed some example genes that are differentially expressed in EBs, which are enriched in expected functional categories. Instead of only listing example genes ( supplementary fig 3d), the authors should carry out a global comparison between differential gene expression profiles of EBs and those of mouse early germ layer development. To support their conclusion, there needs to be global concordance based on such comparisons.
In EBs, the loss of STRAP induced changes in gene expression. Among genes listed in Supplementary  Table 4, why weren't the genes mentioned in the first section (such as Fgf8, Gsc, Otx2 and Shh) among those that were significantly differentially expressed?
The eCLIP analysis generated a large number of binding peaks of Strap. However, only 4.4% were associated with annotated AS genes. This result indicates that Strap's main role may not lie with regulation of alternative splicing. Does it more often regulate constitutive exons? That is, is it possible Strap is a necessary factor for maintaining constitutive splicing? In Fig. 4b, for peaks binding to exons or introns, how many peaks are close to constitutive exons? In Fig. 4c and d, were all exons used in the meta-analysis? Did the differential splicing analysis using RNA-seq data allow detection of splicing changes in exons that were not alternatively spliced in the WT cells? Overall, the eCLIP results do not explain much of the AS results from RNA-seq, the reason of which needs to be further investigated.
In the motif analysis using HOMER, what were used as the background exons? The choice of background can dramatically change the results, which needs to be carefully justified.
The authors attempted to related STRAP function with that of other splicing factors. Different analyses eluded to different factors without a consensus (e.g., PRPF3, SF3, SRSF1, SRSF2, SNRPA, U2AF2). A unifying analysis should be conducted to draw clear conclusions on which factors may be directly related to Strap function.
Minor issues: The terminology used for alternatively skipped exons (SE) should be updated. Sometimes "CA" was used, in other cases, "ES" was used. Neither is standard in the splicing literature. Instead, "SE" should be used to refer to skipped exons (i.e., cassette exons).
The manuscript is very long. Some contents could be condensed/removed from the main text, especially those not directly focusing on the function of STRAP, for example, the first section "Substantial AS events occur during mouse early organogenesis".
Reviewer #3 (Remarks to the Author): This work highlights a neurodevelopmental role for the serine threonine kinase receptor-associated protein (STRAP) in mouse embryos, embryonic stem cell (mESC) derived embryoid bodies (EBs) with validation in Xenopus. The authors use a range of orthogonal models and experimental models to convincingly demonstrate the role of STRAP in neurodevelopment. Overall, this is interesting work and the conclusions are broadly supported by the experimental data provided. There do remain some unanswered questions (e.g. how STRAP binding to the same positions can exert opposing effects in different circumstances), but these are generally acknowledged with plausible hypotheses in the manuscript. Overall I do believe that this work is of the calibre to be published in Nature Communications and the authors have done a good job in validation of core findings, orthogonal methods to confirm particular results and a balanced conclusion. Although I'm not suggesting that these should preclude publication, it is noteworthy that there is no actual mechanism presented; i.e. how is STRAP actually regulating these splicing events. On balance, with the appropriate revisions, I'm supportive of the work being published. Specifically, it would be further strengthened by the authors considering / addressing the following: 1) The first paragraph in the introduction should state explicitly that it is referring to mouse development 2) The rationale for choosing the timepoints is morphological change. However, it is likely that the transcriptional program driving this precedes physical changes. Therefore, could the story be richer / more comprehensive by having an earlier timepoint as a comparator? Also I believe that these are data generated from another previously published study (Chen et al 2004) that have been re-analysed here. Of course, this does not detract from the current findings but it would be more transparent to state this explicitly i.e. a previous study demonstrated X (Chen et al 2004). We reanalysed the data to confirm X but also extend this observation to show Y.
3) 'To delineate the alteration in molecular features during this transitional course, we applied global transcript profiling of the above two stages.' -please describe what the sample was (i.e. homogenized tissue, specific micro-dissected tissue, purified cell types?) 4) 'Strap -/-mutants failed to express several early brain developmental markers (such as Fgf8, Gsc, Otx2 and Shh18) as early as E8.5, (Fig.2c); however, other brain markers (i.e. En1, Hoxb1 and Six3) were not affected by the ablation (Supplementary Fig. 2b)' -Is this because specific brain regions are not formed or is it because the cells within the region do not express the correct markers? Again, it was previously reported that no differences exist at early embryonic stages, but by 9.5 (after organogenesis) there are striking differences. Therefore organogenesis may begin unaffected and the impact of STRAP comes into play later in development. Embryonic lethality occurs between E10.5 and E11.5. This has been previously published and Chen 2004 and should be cited clearly at this point. It will be important to determine at which point the markers displayed in Fig2C,D are lost, as they study the expression at the time point in which embryonic lethality is occurring. 5) Association of STRAP with splicesomal factors does not necessarily imply that it is involved in the process of spliceosome assembly. STRAP could be sequestering splicing factors away for example. How can these correlative data (co-IP, IF) be taken further to prove that STRAP itself is actually crucial for spliceosome assembly as the authors propose? 6) 'As expected, little is altered in transcriptome profiling of Strap-KO ESCs relative to that of their counterpart WT cells (data not shown)' -Can these data please be shown together with transcript level differences between STRAP-KO and WT. Similarly, in existing tissue from these experiments, it would strengthen the study to see the results validated in the early embryo to see if they are consistent between the EB model and in vivo.

7)
Have the EBs been characterized at all? Germ layer markers, percentage of neural cells, regional identity, capacity for terminal differentiation and acquisition of electrophysiological properties would be helpful to include 8) '…indicating these variants have potential biological functions instead of undergoing nonsensemediated mRNA decay' -This could be tested formally by genetically and/or pharmacologically inhibiting NMD and looking for a change in the transcript level 9) Stage-specific description of alternative splicing programs has recently been investigated using human stem cell approaches and found that intron retention was a predominant event early in lineage restriction, whilst exon skipping dominated at later stages. It would be useful for readers to contextualise the present study with this work perhaps in the discussion. 10) 'In contrast, there was a low number of AS outcomes (n=130) affected by STRAP in ESCs (data not shown), raising the possibility that STRAP might mediate splicing in a cell type-specific manner.' -Please show the data.
11) The authors can begin to address the cell type specificity of STRAP-mediated splicing effects by comparing their data from different stages of lineage restriction (e.g. iPSC, EB, terminal differentiation) 12) Supp Figure  15) The claim that STRAP doesn't interact with DDX15 in the nucleus could be substantiated by absence of DDX15 on STRAP Co-IP and vice versa. The colocalisation by ICC is only partially convincing 16) The team could have use a conditional knock out in the nervous system to overcome embryonic lethality. However, I do not expect the authors to do this as it is a major undertaking and I feel not a reasonable request for this paper given i) the conceptual advance that has been presented and ii) time frame of a rebuttal. 17) Some in vitro binding assay validation of the motifs would be useful, either generating a mutant and showing that STRAP binding is lost, or using a blocking agent (e.g. MIXmer ASO) to mask it.
18) Suggesting STRAP interacts with other splicing regulators seems unsubstantiated.
19) Fat1 is one of the least affected genes following STRAP loss, why did the authors choose this one and not another? 20) Can the authors perform WB to evaluate an increase in specific protein isoforms resulting from their finding of exon inclusion for example. 21) Why is the length of the exon displayed as a log10 value? The findings in the majority of figure 6 appear purely descriptive and could be moved to supplementary. The motif analysis could have been supplementary when showing the interaction of STRAP to DNA.
22) There is limited information on whether STRAP is in a complex with these splicing regulators or is simply nearby?
Minor changes: 1) Bottom of page 3: Gene oncology (GO) should be Gene ontology (GO) 2) What were the number of biological and technical repeats?
3) 'We further compared splice site scores in responsive and unresponsive CAs. We observed that both STRAP-enhanced and -repressed exons has weaker 3' splice site (ss) than those found in unresponsive CAs (unpaired Wilcoxon test, enhanced P<0.0001, repressed P <0.067), and that enhanced CAs had even weaker 3'ss than repressed ones (Fig. 3d,e).' -Have instead of has 4) Figure 3D can be moved to supplementary, it does not seem important enough to the overall narrative to justify being in the main text.
5) The schema (last panel of last figure) would benefit from a key within the actual figure.
6) 'the total transcripts of these genes did not change much' seems quite casual and should be rewritten.
Reviewer #4 (Remarks to the Author): In this manuscript, Jin et al. identified STRAP as a novel regulator of alternative splicing (AS) in early mouse development. In developing mouse embryos, the authors uncovered the previously uncharacterized AS signatures in between E8 and E9 mouse embryos when embryos undergo organogenesis including massive neurogenesis and brain patterning. Using Strap knockout embryoid body (EB) differentiation model, the authors found stage specific AS patterns regulated by direct Strap binding and identified the Strap binding motifs. Moreover, using amphibian Xenopus model, authors demonstrated that the AS regulation by Strap is evolutionarily conserved and important for neural development.
Major Points 1. The AS of early developmental genes show highly stage specific patterns. Thus the developmental stages should be carefully considered (and need to be well-presented in the figure panels to be easily recognized). In EB differentiation model, the authors found 454 AS events in the Strap knockout EBs (Fig  3) at D9 (9 days of differentiation). However, at gene level analysis (Fig 5), authors compared splicing patterns of Nnat and Mark3 along the differentiation until D14 and found that Strap affects their AS events somewhere in between D11 and D14. Are there any examples of genes whose ASs are mainly regulated at around D9 EBs? Testing and validating individual AS patterns at this stage (D9) would strengthen the authors' hypothesis that the altered AS pattern in the Strap KO embryos might have caused leathality beyond E10.5 (and impaired neurogenesis in D11-D14 EB model shown in Supplementary Fig 5j). More specifically, according to the data in Supplementary Fig 5j, KO EBs at D14 might have been already lost its potential to further differentiate (exemplified with the impaired gene expressions of Ncam1, Ntf3 and Map2). Therefore, it is reasonable to consider that D14 KO EBs are not in the same differentiation states with WT D14 EBs but rather arrested earlier along the differentiation course. Therefore, the altered splicing patterns of D14 KO EBs do not fully support the idea that Strap directly regulates AS events at later stages (i.e., D14). Authors should validate that the altered AS patterns in Strap KO are not due to the indirect effects of developmental arrest.
2. Again, I think the direct correlation between Strap binding and the AS regulation is largely missing. Since Strap KO might lead to developmental arrest in EB differentiation models (see Supplementary Fig  5j), I would suggest the authors to do the transient knockdown experiments using indicible shRNA expression at D14 EB (or inducible KO model if possible) and see whether Strap knockdown at this specific stage (or at earlier stages) could still affects AS of some of the neuronal target genes (such as Nnat and Mark3).
3. The Xenopus Strap morphant phenotype of delayed neural tube closure is too vestigial and insufficient to justify the authors' claim of evolutionary conservation of the mechanisms and functions of Strap. I think the Strap morphant Xenopus embryos should be re-analyzed for expressions of early neural genes (pan-neural, neuronal or region specific neural marker genes) or for Strap target candidate genes identified in mouse EB model. The authors previous paper (also cited as #17) already showed that Strap knockdown causes impaired neural/body patterning. Perhaps neutralized animal cap (e.g., by injecting Noggin) model can be utilized for gene expression analysis and AS changes instead of the whole embryo samples. This might particularly helpful since the splicing defects shown in Fig 6g are  2. The authors argue that "Loss of STRAP selectively affected expression of brain-regional markers" (p5, line 176-177). However, it is unclear how the reduction of Strap sensitive early brain marker genes (such as Fgf8, Gsc, Otx2 and Shh) did not affect other brain markers (En1, Hoxb1 and six3) at E8.5 while it led to the failure of further developmental progression and later germ layer marker gene expressions (such as Irx3, Nkx2.5, Gata6 and Tcf15) in E9.5 KO embryos. It seems that Strap deficiency leads to overall developmental arrest by affecting widespread genes throughout all three germ layers rather than specifically affecting a subset of brain specific target genes. Authors should discuss about the results in more detail.
3. Immunocytochemistry analysis of Strap localization indicated its co-localization with several splicing factors but not with spliceosome disassemble protein DDX15 (Fig 2i and Supplementary Fig 2f). However, the immunostaining signals of STRAP and other proteins seem often over-saturated especially in the case which the authors argue the co-localization. Can the authors provide under saturated images to justify co-localizations better?
We would like to thank the reviewers for the helpful comments on our manuscript. We have addressed all the points raised and feel that our manuscript is now much improved with the changes in the text and additional new experiments. We have discussed each point in detail as follows:

Response to Comments of Reviewer 1
General Comment: "This manuscript by Jin et al. shows STRAP knockout affects many alternative splicing events and early organogenesis during embryonic development. The authors first found Strap KO exhibits dysmorphogenesis as early as E8.5-9 and lethality at E10.5. Transcriptome analysis showed a major defect in alternative splicing regulation but modest transcriptional abnormalities. Proteomics on Strap co-IP samples identified spliceosome components as Strap's major interacting partners in the nucleus, suggesting STRAP be a splicing regulator. The authors used eCLIP to locate STRAP's RNA binding sites and found many in introns. Although the informatics data are largely descriptive, many are derived from in vivo samples, which is a significant strength. The manuscript is easy to read. As presented, the novelty of the paper is proposing STRAP as an alternative splicing regulator. Some clarifications are needed (see below). The novelty can be enhanced with mechanistic studies of how STRAP regulates splicing and/or functional analysis of STRAP's targets." Response: We appreciate the encouraging comments of the reviewer regarding novelty and strength of this interesting study. The concerns raised by this reviewer are being addressed as follows.
Major Comments 1. Comment: "Fig 2: is the STRAP-spliceosome interaction RNA dependent? Because TFIIIC complex is pulled down, the interaction could be even DNA dependent? The authors can answer these questions with DNase and RNase treatment of cell lysates. If the authors decide to pursue the mechanistic aspect, mapping of interaction domains and functional consequence of interaction mutations should be tested." Response: We agree with the reviewer that it is important to explore whether the interactions between STRAP and spliceosome components are in an RNA-dependent manner, which is being addressed here. Overall, the results show that the association of STRAP with U2 proteins remained unchanged or enhanced (new Fig. 3c) after treatment with RNase. In contrast, other U proteins bind to STRAP in the presence of RNA (new Fig. 3c). We then have focused on whether and how STRAP participates in steps of U2 snRNP biogenesis and found that STRAP is involved in the assembly of 17s U2 snRNP complex (new Fig. 7). Additionally, we indeed have observed the contact between STRAP-TFIIIC110 and DNA (new Supplementary Fig. 3e), suggesting that STRAP might play a role in the co-transcriptional process. We will explore this possibility in the future work. Regarding the binding domain specificity of STRAP and functional activity in terms of splicing events, we have performed the gain-of-function assay using a wellcharacterized SR140-dependent UPF3A minigene reporter. We observed that STRAP, specifically through its third WD40 repeat domain, plays a critical role in UPF3A exon4 splicing (new Fig. 7h). We also have obtained similar conclusion using other two minigenes (new Supplementary Fig. 7c,d). Collectively, these results support the notion that STRAP is required for the recruitment of certain splicing factors to spliceosome complex.  Fig. 5f). Therefore, combing the in vivo and vitro results, we demonstrate that STRAP physically interacts with RNA.
3. Comment: "RNA-seq targets and eCLIP-Seq targets do not substantially overlap (fig 5a). This does not necessarily invalidate the authors' conclusions, but the authors should provide a thorough explanation." Response: This is a good point. The overlap percentages we presented in the original manuscript were not ideal (previous Fig. 5a Fig. 7c and d).
5. Comment: "The phenotypic characterization of the KO mice is rather limited and need to be elaborated". 'However, E9.5 Strap -/embryos are small in size with delayed development as compared to WT littermates, namely no body turning with truncated frontonasal region, dilated heart cavity (Fig. 2a), and open neural tube (data not shown), consistent with a previous report of gene trap mutagenesis study 19 '. 7. Comment: "An important question is whether STRAP is a general splicing regulator or a lineage/stage specific regulator. STRAP's expression patterns and its KO/KD in a context other than germline deletion can answer this question. The latter also helps to mitigate the concern of indirect effects from germline deletion for 8 days." 'Additionally, the assigned ranked scores by COMPLEAT tool 29 showed 30% of hits (45 out of 152) participate in multiple predicted networks between splicing protein interactions ( Fig.3b and Supplementary Fig.3c) Fig.4 g-j) show highly repeated close values between each duplicate. We also conducted RT-PCR assay to assess the accuracy for analyzing AS events (new Supplementary  Fig. 4k). Together, these support the robustness of our findings.  Supplementary Fig. 4e, f). To avoid confusion of the reviewers and readers, we have removed previous Supplementary Fig. 3d from the revised manuscript.

Response to Comments of Reviewer# 2
General comment: "This manuscript reports a study of STRAP as a novel regulator of alternative splicing in early mouse embryos. The authors carried out global studies of STRAP using RNA-seq, eCLIP-seq, complemented by experimental validations and individual assays, in knockout models of mouse embryoid body. In addition, co-IP analyses showed that STRAP is a putative spliceosome-associated factor. The study reported a relatively small number of AS events that may be regulated by STRAP. The authors also extended the study to Xenopus and showed that loss of Strap led to delayed neural tube closure. Overall, this is an interesting study that revealed a novel splicing regulator. However, there are a number of major concerns that need to be addressed."

Response: We appreciate the reviewer's affirmative comments and insightful suggestions that are helpful in improving the manuscript.
Major Comments 1. Comment: "A major concern is the relatively small number of AS events detected by RNA-seq in response to the loss of Strap. Furthermore, an even smaller number of these AS events had eCLIP peaks in their immediate neighborhood (in the exon or flanking introns). If STRAP is a putative spliceosome-associated factor, there expects to be a much larger pool of its splicing regulatory targets. The small number of AS events is likely due to the fact that only 2 biological replicates were included for each group in the RNA-seq study. Current practice standard is to include 3-5 replicates at least to better estimate sample-to-sample variation. Another reason that could underlie the small number of significant events is the limited analysis method for differential splicing discovery in RNA-seq. It is well known that existing methodologies are limited and different methods could yield very different results. It is strongly recommended that the authors use alternative methods for this analysis, in addition to rMATS, such as MAJIQ, LeafCutter, etc. Outputs of different methods should be compared, and combined in some way to enhance the quality of their results." Supplementary  Fig. 4e,f). Second, we agree with the reviewer that rMATS tool can sometimes be underpowered. We, therefore, re-analyzed the RNA-seq data using MAJIQ tool, as suggested by this reviewer. We obtained 1,462 altered local splicing variants (out of 972 genes) upon deletion of STRAP in Day 9 EBs (new Supplementary Table 7), which has a significant concordance with the data from rMATS (new Fig. 4h). By combining these two datasets, we obtained 1687 AS events (new Fig. 4h, including classic-and non-classic splicing patterns) regulated by STRAP. This number is greatly increased from 4.4% to 15.6%. In detail, we found that 15.6% (319 out of 2047) of STRAP binding events is associated with its regulated AS genes (both classic-and non-classic types) and 22.4% (319 out of 1423) of unique genes with AS events in 9-day-old EBs are linked to STRAP binding, supporting our hypothesis that STRAP developmentally regulates certain AS events through direct binding. We hope the additional data make our results more informative.

Comment: "
To make the point that N2B27-induced EBs molecularly resemble the mouse embryonic germ layer specification, the authors listed some example genes that are differentially expressed in EBs, which are enriched in expected functional categories. Instead of only listing example genes (supplementary fig 3d), the authors should carry out a global comparison between differential gene expression profiles of EBs and those of mouse early germ layer development. To support their conclusion, there needs to be global concordance based on such comparisons." Response: To address the reviewer's concern, we have added a systematic global comparison of our findings with three datasets previously published (new Supplementary Fig. 4e,f). In these figures, the previous findings from induced neuroectoderm-like EBs by Li et al. (Cell Reports, 2020) are shown to be very close to the context of our model; however, the data from either early mouse brain tissues (Ayoub et al, PNAS, 2011) or three germ layer-derived multiple organs in mouse embryo early development (Werb et al, 2014) is far away with our data in cluster analysis. We further characterized the EBs during the course of time using molecular markers (New Supplementary Fig. 4c) and electrophysiological assay (new Fig. 6d,

e), which show that N2B27-induced EBs molecularly resemble the neuroectodermal lineage commitment and have a terminal differentiation toward matured neuronal cells after long-term induction of ascorbic acid.
3. Comment: "In EBs, the loss of STRAP induced changes in gene expression. Among genes listed in Supplementary Table 4, why weren't the genes mentioned in the first section (such as Fgf8, Gsc, Otx2 and Shh) among those that were significantly differentially expressed?" Response: We agree that there is no mention of the above-mentioned genes in previous Supplementary Table4, in which differentially expressed genes with a P value < 0.01 as well as a log2 fold change > 1 were listed. According to the thresholds, there are no significant changes of these genes when compared KO with WT groups at day 9. We have now quantified these markers by qPCR assays at two time points (day 9 and day 11) and found several genes downregulated by loss of STRAP at day 11, consistent with the in vivo data (new Fig. 2c). These results are now included in new Supplementary Fig. 4d.

Comment: "
The eCLIP analysis generated a large number of binding peaks of Strap. However, only 4.4% were associated with annotated AS genes. This result indicates that Strap's main role may not lie with regulation of alternative splicing. Does it more often regulate constitutive exons? That is, is it possible Strap is a necessary factor for maintaining constitutive splicing? In Fig. 4b, for peaks binding to exons or introns, how many peaks are close to constitutive exons? In Fig. 4c and d, were all exons used in the meta-analysis? Did the differential splicing analysis using RNA-seq data allow detection of splicing changes in exons that were not alternatively spliced in the WT cells? Overall, the eCLIP results do not explain much of the AS results from RNA-seq, the reason of which needs to be further investigated." Response: We appreciate reviewer's suggestion of using alternative method to analyze splicing events. In previous Fig. 5a, the overlap percentages between STRAP peaks and annotated AS genes were not ideal. We have now reanalyzed the intersected percentages using combined AS genes (from rMATS and MAJIQ) with eCLIP genes, which results in the increased percentages from 4.4% to 15.6% and has been updated in new Fig. 6a. Fig. 4c). Thus, the binding of STRAP preferentially in close proximity to both 5' and 3' of splicing sites suggests its potential role in demarcating exonic regions; (2)  In previous Fig. 4c and d, Fig. 5f).  Fig. 3c) indicating that these interactions depend on protein-protein contact. We therefore focus on the relationships between U2-specific proteins and STRAP. CO-IP assays reveal that STRAP regulates the assembly of 17S complex at the stage of U2 snRNP biogenesis (New fig.7c-f). Please also see our response to reviewer# 1 comment 1 for additional details on this point.

Minor Comments
7. Comment: "The terminology used for alternatively skipped exons (SE) should be updated. Sometimes "CA" was used, in other cases, "ES" was used. Neither is standard in the splicing literature. Instead, "SE" should be used to refer to skipped exons (i.e., cassette exons)." Response: We agree with the reviewer and have maintained the consistency of using "SE" in the revised manuscript.
8. Comment: "The manuscript is very long. Some contents could be condensed/removed from the main text, especially those not directly focusing on the function of STRAP, for example, the first section "Substantial AS events occur during mouse early organogenesis."

Response to Comments of Reviewer# 3
General Comment: "This work highlights a neurodevelopmental role for the serine threonine kinase receptor-associated protein (STRAP) in mouse embryos, embryonic stem cell (mESC) derived embryoid bodies (EBs) with validation in Xenopus. The authors use a range of orthogonal models and experimental models to convincingly demonstrate the role of STRAP in neurodevelopment. Overall, this is interesting work and the conclusions are broadly supported by the experimental data provided. There do remain some unanswered questions (e.g. how STRAP binding to the same positions can exert opposing effects in different circumstances), but these are generally acknowledged with plausible hypotheses in the manuscript. Overall, I do believe that this work is of the caliber to be published in Nature Communications and the authors have done a good job in validation of core findings, orthogonal methods to confirm particular results and a balanced conclusion. Although I'm not suggesting that these should preclude publication, it is noteworthy that there is no actual mechanism presented; i.e. how is STRAP actually regulating these splicing events. On balance, with the appropriate revisions, I'm supportive of the work being published". Specifically, it would be further strengthened by the authors considering / addressing the following": Response: We thank the reviewer for positive comments and for providing insightful suggestions that provided opportunities to improve the manuscript.
Major Comments 1. Comment: "The first paragraph in the introduction should state explicitly that it is referring to mouse development".
Response: We have revised the "Introduction" part and referred to mouse development (page 1).

Comment: "
The rationale for choosing the time points is morphological change. However, it is likely that the transcriptional program driving this precedes physical changes. Therefore, could the story be richer / more comprehensive by having an earlier timepoint as a comparator? Also I believe that these are data generated from another previously published study (Chen et al 2004) that have been re-analysed here. Of course, this does not detract from the current findings but it would be more transparent to state this explicitly i.e. a previous study demonstrated X (Chen et al 2004). We reanalysed the data to confirm X but also extend this observation to show Y." Response: We agree with the reviewer that, as originally written, it is not clear why we chose 9day-old EBs to perform a serial of experiments. Our rationale is as follows: (1) this timepoint relatively matches the mouse embryonic day on which the morphology of STRAP KO embryo is severely impaired.
(2) Based on our culture protocol, EBs could undergo the stage of neuroectoderm lineage on day 9. As the reviewer's request, we quantified the expression of markers at different timepoints. The results show that the expression of neuroectoderm-related genes was comparable between two groups at day 6-9 and brain developmental markers appeared as early as day 9. We have now included these results in new Supplementary Fig. 4d and revised the text accordingly. We also have rearranged the text for the citation Chen's paper (page 4). Please note that in this paper authors only show the Strap KO mouse embryo lethality after E10.5 and describe the alteration in morphology upon Strap deletion. There is no more information shown in this paper to explore the mechanism at cellular and molecular levels.
3. Comment: " 'To delineate the alteration in molecular features during this transitional course, we applied global transcript profiling of the above two stages.' -please describe what the sample was (i.e. homogenized tissue, specific micro-dissected tissue, purified cell types?)."

Response: We have revised the text as follows (page 3):
'To delineate the molecular features of mouse embryos during post-gastrulation (E8.0) to early organogenesis (E9.0), we applied global transcript profiling using whole mouse embryos at these two stages'. 4. Comment: " 'Strap -/-mutants failed to express several early brain developmental markers (such as Fgf8, Gsc, Otx2 and Shh18) as early as E8.5, (Fig. 2c); however, other brain markers (i.e. En1, Hoxb1 and Six3) were not affected by the ablation (Supplementary Fig. 2b)' -Is this because specific brain regions are not formed or is it because the cells within the region do not express the correct markers? Again, it was previously reported that no differences exist at early embryonic stages, but by 9.5 (after organogenesis) there are striking differences. Therefore organogenesis may begin unaffected and the impact of STRAP comes into play later in development. Embryonic lethality occurs between E10.5 and E11.5. This has been previously published and Chen 2004 and should be cited clearly at this point. It will be important to determine at which point the markers displayed in Fig. 2C,D are lost, as they study the expression at the time point in which embryonic lethality is occurring." -/-embryos (E8.5) shows that they are normal in their own body proportions and orientation. So we think the down-regulated expression of the early brain developmental markers (including Fgf8,Gsc,Otx2 and Shh18) is mainly caused by loss of STRAP in region-specific cells. However, we cannot exclude the possibility that low levels of these genes in Strap-/-mutants were present due to the delayed development of curtained brain regions. Combined with our previous study in Xenopus (ref 18) and current mouse embryo model, we conclude that STRAP contributes its role in this regard between the late gastrulation and early organogenesis stages. We have quantified and compared the above markers using E7.5 embryos (new Supplementary Fig. 2b). Our data show that germ layers markers were comparable between WT, Het and KO. We have rewritten the results section as follows (page 5).

Response: The reviewer raises an interesting point. Morphologic evaluation of Strap
'Strap deficiency initially has a negative impact on mouse embryo fore-and midbrain development at E8.5 and subsequently causes a uniform and constant developmental delay from E9.5.' We have changed the write up in results section for citing observations from Chen, 2004, Nature genetics (page 4).

Comment
: "Association of STRAP with spliceseomal factors does not necessarily imply that it is involved in the process of spliceosome assembly. STRAP could be sequestering splicing factors away for example. How can these correlative data (co-IP, IF) be taken further to prove that STRAP itself is actually crucial for spliceosome assembly as the authors propose?"  Supplementary Fig. 6g). We have also assessed the terminal differentiation of EBs regarding electrophysiological properties and included the data in new Fig.  6d,e.
8. Comment: " '…indicating these variants have potential biological functions instead of undergoing nonsense-mediated mRNA decay' -This could be tested formally by genetically and/or pharmacologically inhibiting NMD and looking for a change in the transcript level." Response: We appreciate the reviewer's suggestion. We have treated EBs with NMD inhibitor cycloheximide (Durand, et al, J Cell Biol, 2007;Dang, et al, J Biol Chem, 2009) Supplementary Fig. 2a).
13. Comment: "Supp Figure 2 C The teratoma validation is morphologically consistent but markers for the 3 germ layers should be examined by IHC."

Response:
We have now provided a histological characterization of the teratoma (new Fig. 2e). Results section has been updated accordingly (page 5).
14. Comment: "Supp Figure 2 E: The IgG is not as clean as expected and why are so few peptides detected"?
Response: We acknowledge that due to longer development, the gel showed little background and some non-specific protein bands with IgG. We definitely obtained a number of protein bands due to specific binding with STRAP, which is clear from the raw data of manually enriched STRAP-specific targets as listed in new Supplementary Table 3. 15. Comment: "The claim that STRAP doesn't interact with DDX15 in the nucleus could be substantiated by absence of DDX15 on STRAP Co-IP and vice versa. The co-localization by ICC is only partially convincing." Response: We agree with the reviewer and have now added the data for Co-IP assay. We did not see the interaction between STRAP and DDX15 in the absence or presence of RNase A (new Supplementary Fig. 3d). The text has been updated accordingly (page 5).
16. Comment: The team could have use a conditional knock out in the nervous system to overcome embryonic lethality. However, I do not expect the authors to do this as it is a major undertaking and I feel not a reasonable request for this paper given i) the conceptual advance that has been presented and ii) time frame of a rebuttal.
Response: We thank the reviewer for his/her interest along this line and the understanding that it is not possible to finish those experiments in the time frame. We will definitely explore this in our future work.
17. Comment: "Some in vitro binding assay validation of the motifs would be useful, either generating a mutant and showing that STRAP binding is lost, or using a blocking agent (e.g. MIXmer ASO) to mask it." Response: We thank the reviewer for this comment. We have now performed RNA EMSA assays (new Fig. 5f). Please see our response to reviewer# 1 comment 2 for additional details on this point.
18. Comment: "Suggesting STRAP interacts with other splicing regulators seems unsubstantiated." Response: We have made major changes by including data in response to this comment. Overall, the result shows that the association of STRAP with U2 proteins remained unchanged or enhanced with RNase treatment (new Fig. 3c). In contrast, other U proteins bound to STRAP in the presence of RNA (new Fig. 3c). Additionally, STRAP involved in the assembly of 17S U2 snRNP complex (new Fig.7). Please see our response to Reviewer# 1 comment 1 for further details on this point.

Comment
: "Fat1 is one of the least affected genes following STRAP loss, why did the authors choose this one and not another"?
Response: Based on the STRAP-RNA binding map (new Fig. 6b,c), we have performed UV crosslinking and immunoprecipitation (RIP) with anti-STRAP antibody to arbitrarily examine three eCLIP-hit genes, Fat1, Tcf7l2 and Ctnnd1. We have now removed results for Fat1 to new Supplementary Fig. 6c. 20. Comment: "Can the authors perform WB to evaluate an increase in specific protein isoforms resulting from their finding of exon inclusion for example" Response: We appreciate the reviewer's suggestion. We have carefully checked the commercially available antibodies against the protein isoforms or the antibodies that distinguish two isoforms identified by our RT-PCR assays (new Supplementary Fig. 4k). Unfortunately, only the antibody for specific NAB2 isoform is available. As shown in results below, less short Nab2 isoform was observed in STRAP KO EBs, which is consistent with the change at transcript level.
21. Comment: "Why is the length of the exon displayed as a log10 value? The findings in the majority of figure 6 appear purely descriptive and could be moved to supplementary. The motif analysis could have been supplementary when showing the interaction of STRAP to DNA." Response: Because the length of STRAP target exons and introns has a wide range, we have to narrow down the value via log10 conversion. As the reviewer suggested, plots have been moved to new Supplementary Fig. 8a-e. Please see Source data file for detailed raw data. Considering the limited words in text, we have removed the plots for the motif analysis (previous Fig. 6d-e and Supplementary Fig.6 c-g) from the revised manuscript.

Comment
: "There is limited information on whether STRAP is in a complex with these splicing regulators or is simply nearby"? 25. Comment: " 'We further compared splice site scores in responsive and unresponsive CAs. We observed that both STRAP-enhanced and -repressed exons has weaker 3' splice site (ss) than those found in unresponsive CAs (unpaired Wilcoxon test, enhanced P<0.0001, repressed P <0.067), and that enhanced CAs had even weaker 3'ss than repressed ones (Fig.  3d,e).' -Have instead of has." Response: Thanks for this point. It is now fixed.
26. Comment: " Figure 3D can be moved to supplementary, it does not seem important enough to the overall narrative to justify being in the main text." Response: Previous Fig. 3d was a schematic diagram to define alternative and constitutive exons as well as 5'/3'splicing sites. Accordingly, we extracted the corresponding DNA sequences and uploaded them to MaxEntScan tool to leverage the splicing scores (as shown in previous Fig. 3e). Therefore, these two subfigures are interdependent. We have now revised them as new Fig. 3d and 3e.