Regnase-1-mediated post-transcriptional regulation is essential for hematopoietic stem and progenitor cell homeostasis

The balance between self-renewal and differentiation of hematopoietic stem and progenitor cells (HSPCs) maintains hematopoietic homeostasis, failure of which can lead to hematopoietic disorder. HSPC fate is controlled by signals from the bone marrow niche resulting in alteration of the stem cell transcription network. Regnase-1, a member of the CCCH zinc finger protein family possessing RNAse activity, mediates post-transcriptional regulatory activity through degradation of target mRNAs. The precise function of Regnase-1 has been explored in inflammation-related cytokine expression but its function in hematopoiesis has not been elucidated. Here, we show that Regnase-1 regulates self-renewal of HSPCs through modulating the stability of Gata2 and Tal1 mRNA. In addition, we found that dysfunction of Regnase-1 leads to the rapid onset of abnormal hematopoiesis. Thus, our data reveal that Regnase-1-mediated post-transcriptional regulation is required for HSPC maintenance and suggest that it represents a leukemia tumor suppressor.

RNASe, in the regulation of hematopoiesis. Their major finding is that "Regnase-1 regulates selfrenewal of HSC through modulating the stability of Gata2 and Tal1 mRNA." They also found that "dysfunction of Regnase-1 leads to rapid onset of AML", and concluded that "Regnase-1-mediated post-transcriptional regulation is required for HSC maintenance and suggest that it represents a novel type of leukemia tumor suppressor." I will leave it to other reviewers to focus on the validity of the findings relating to the behavior of hematopoietic stem cells in the mutant mice, as well as the importance of these findings to the overall field of hematopoiesis regulation. My comments will focus on the RNA-Seq data and the conclusions regarding direct effects on Gata2 and Tal1 mRNA stability.
Despite going back and forth between the text, the methods section, and the figure legends, I was not able to figure out several aspects of the RNA-Seq data. It is possible that these pieces of information were hidden in other places, but I was not able to find them. I will list here some of my major problems with the information presented, but an overall comment is that it would be very helpful for the reviewer to have all the relevant information in one place, for example, in the methods or results section. Specific issues include, but are not limited to: 1. How were the HSC treated or cultured after the FACS separation? To my knowledge, we have no information on that. 2. How many biological replicates were used to generate the data in Fig. 5f and Fig. 6, showing the overall differential expression, i.e., how many mice were used to prepare how many independent cultures of HSC used in this experiment? 3. Regardless of the answers to 2 above, was the RNA-Seq experiment performed more than once? 4. What was the statistical method used to determine statistical significance of the differences? 5. Were the transcript data limited by an abundance cutoff? 6. Although the researchers took their gene sets through many screens before settling on the handful that they analyzed further as potential direct targets of Regnase-1, I think the reader should be given easy access to the actual gene lists, showing fold changes, an indication of abundance, and an indication of the significance of the fold changes in the original normalized RNA-Seq data. Perhaps a supplemental Excel file could be used to address most or all of these questions, as well as certain fundamental questions like the relative abundance of housekeeping transcripts, the abundance of the knocked out gene (reflecting cell purity), etc. In my view, all of this information is necessary to help the reader determine the validity of the authors' conclusions on this topic. 7. The authors then moved to their analysis of potential target transcripts. Rather than starting from the total collection of up-regulated transcripts, which most likely included many transcripts regulated by Regnase-1 either directly or secondarily, the authors moved to their exon gradient analysis, and limited the data still further by screening for up-regulated transcripts that are known to be involved in hematopoiesis. They thereby eliminated from the analysis many potential target transcripts that might well be involved in cellular function because they did not meet those screening criteria. 8. For Fig. 6c, I assume that these data were from real-time QPCR, but it is not clear that this is the case from the Results section. 9. It would be very helpful in the section on target transcript analysis to review the binding and/or cleavage sites for this protein, and the evidence to support these. 10. I don't think the co-transfection assays with luciferase readouts are adequate demonstrations that the mRNAs studied are direct targets in vivo. It would not be surprising to find that overexpression studies of this type would result in false positives, based on the high expression of the transfectedexpressed enzyme and the probable relative promiscuity of the recognition sequences in the mRNA. It should be relatively straightforward to determine mRNA stability in the original cultured HSC by labeling experiments or by using actinomycin D, perhaps supplemented by assays of direct binding (if this can be done without cleaving the mRNA).

Reviewer #1 (Remarks to the Author):
In this study, the authors found that Regnase-1 is highly expressed in adult hematopoietic stem cells. Regnase-1 knockout in hematopoietic cells results in expansion of hematopoietic progenitors and reduction of quiescent stem cells specifically in bone marrow. Transplantation analyses show that Regnase-1 deficiency impairs repopulation potential of hematopoietic stem cells. These results indicate that Regnase-1 is required for stem cell functions. Cell cycle is activated in Regnase-1-deficient stem cells. Regnase-1 homozygous and heterozygous knockout mice exhibit splenomegaly and lymphadenopathy. Lymphoid cells are decreased and myeloid cells are increased in the Regnase-1 homozygous and heterozygous knockout mice, suggesting that Regnase-1 is indispensable for normal hematopoiesis. Blasts and cells with abnormal morphology were observed in the peripheral blood and bone marrows of Regnase-1-deficient mice. Gene expression profiles of Regnase-1 are similar to those of human hematopoietic stem cells in leukemia patients but not healthy people. Based on these results, the authors insist that Regnase-1 deficiency causes leukemia development. Finally, the authors performed analysis of RNA-seq database and identified Gata2 and Tal1 mRNAs as direct targets of Regnase-1, which account for hematopoietic phenotypes of Regnase-1-deficient mice.
This study provides an interesting insight according to Regnase-1 function in hematopoiesis, especially hematopoietic stem cells. Most of the experiments are carefully done. However, this reviewer feels that the present data is not sufficient to support the authors' conclusion. To support the authors' conclusions, there are several issues that should be addressed:

Response to Reviewer 1
The authors thank the reviewer 1 for his/her review comments and valuable suggestions for improving our manuscript. Please see our responses to your comments one by one below.
1. In Figure 1, the authors analyzed Regnase-1expression levels of stem cells and progenitors. The authors should examine Regnase-1 expression levels in lineage-committed cells (e.g. B cells, T cells, myeloid, erythroid and megakaryocyte lineages), because the authors showed lineage-skewing in Regnese-1-deficient mice in Figure 4.
Our response to  According to the suggestion, we measured the expression of Regnase-1 mRNA in lineage-committed cells including B220 + B-cells, CD3 + T-cells, CD11b + myeloid-cells, Lin-Sca1 -cKit + CD34 -CD16/32erythroid and megakaryocyte lineage cells and other progenitor cells. Results showed that Regnase-1 is more strongly expressed in CD34-negative immature hematopoietic stem cells than in the other cell types. However, low Regnase-1 expression was also observed in lineage-committed cells (B-cells, T-cells and myeloid-cells). Therefore, we cannot exclude the possibility that lineage-skewing of Regnase-1-KO mice resulted from Regnase-1 deficiency in lineage-committed cells. We added these data in Supplementary Figure 1a and added comments in the Results section (lines 15-18, page 6).
Are they consequences of Regnase-1 deficiency in HSC (not lineage-committed cells)? As the authors mentioned in Discussion, Regnase-1 has cell type-specific function. The authors should discuss this issue.
Our response to Comment 1-2: As suggested, Cre-mediated Regnase-1 deletion affects not only HSCs but also lineage-committed cells in Vav1-Cre; Regnase-1 flox/flox mice. As mentioned in response to comment 1, Regnase-1 expression is also observed in lineage-committed cells, so it is possible that the changes in myelopoiesis and lymphopoiesis observed in Regnase-1-KO mice are caused by deletion of Regnase-1 in lineage-committed cells. We have considered these possibilities in the Discussion section (lines 14-17, page 23).
Splenomegaly were often observed when abnormal erythropoiesis occurred. The authors should examine erythroid cells and megakaryocytes in the Regnase-1-deficient mice.
Our response to  Thank you for your constructive criticisms. We examined erythropoiesis in Regnase-1-KO mice by counting the number of red blood cells and platelets, evaluating the fraction of megakaryocyte-erythroid progenitor cells (MEP) and erythroid subsets by flow cytometry. As shown in Supplementary Figures 2a and b, red blood cell and platelet counts in peripheral blood and hemoglobin concentrations are significantly decreased in Regnase-1 KO mice. On the other hand, MEPs and erythroid subset populations are slightly but not significantly changed (Supplementary Figure 2c,d). We also analyzed the megakaryocytes in the bone marrow of Regnase-1 KO mice by immunohistochemical staining using anti-CD41 antibody (Supplementary Figure 2e). There were no marked differences in the number of CD41 + megakaryocytes in Regnase-1 KO mice. From these results, we consider that Regnase-1 deficiency has a slight effect on terminal differentiation of erythrocytes, and we have added these results to the Results (lines 15-18, page 13).
4. In Figure 5, the authors insist that Regnase-1-deficient mice develop leukemia. How was absolute number (concentration) of blasts in peripheral blood? Is this leukemia transplantable?
Our response to  In accordance with this suggestion, we examined the frequency of blasts in peripheral blood. As shown in Figure 5c, numerous blasts (10.8% ± 6.3%) and abnormally differentiated cells were observed in Regnase-1 KO mice.
To confirm that Regnase-1 deficiency-induced leukemia cells are transplantable, we transferred bone marrow mononuclear cells from Regnase-1-KO mice to C57BL/6-Ly-5.1 congenic mice. 3 months after transplantation, leukemia-like symptoms such as abnormal proliferation of CD34 -LSK cells and splenomegaly were confirmed (Supplementary Figure 5). Based on this study, we concluded that leukemia caused by Regnase1-deficency is transplantable and have added this to the text (lines 16-19, page 15).
5. The authors identified Gata2 and Tal1 as responsible genes. Are the cell cycle-related genes p21 and p57 shown in Figure 3 direct targets of GATA2 or TAL1?
Our response to  We investigated the direct association of Gata2/Tal1 on p21/p57 gene expression by chromatin immunoprecipitation assays. Although Regnase-1 expression induces the upregulation of p21 and p53 mRNA in THP-1 leukemic cells, direct binding of Gata2 or Tal1 to p21 or p57 promoter regions was not observed (Supplementary Figure 6). Thus, we conclude that Gata2/Tal1 indirectly regulate p21 and p53 mRNA expression.
We now describe these issues in the Discussion (lines 5-8, page 25).

Reviewer #2 (Remarks to the Author):
Kidoya et al # NCOMMS-17-23012-T -Regnase-1-mediated post-transcriptional regulation is essential for hematopoietic stem cell homeostasis These authors address a potential role for Regnase-1, a CCCH zinc finger RNA binding protein and RNASe, in the regulation of hematopoiesis. Their major finding is that "Regnase-1 regulates self-renewal of HSC through modulating the stability of Gata2 and Tal1 mRNA." They also found that "dysfunction of Regnase-1 leads to rapid onset of AML", and concluded that "Regnase-1-mediated post-transcriptional regulation is required for HSC maintenance and suggest that it represents a novel type of leukemia tumor suppressor." I will leave it to other reviewers to focus on the validity of the findings relating to the behavior of hematopoietic stem cells in the mutant mice, as well as the importance of these findings to the overall field of hematopoiesis regulation. My comments will focus on the RNA-Seq data and the conclusions regarding direct effects on Gata2 and Tal1 mRNA stability.
Despite going back and forth between the text, the methods section, and the figure legends, I was not able to figure out several aspects of the RNA-Seq data. It is possible that these pieces of information were hidden in other places, but I was not able to find them. I will list here some of my major problems with the information presented, but an overall comment is that it would be very helpful for the reviewer to have all the relevant information in one place, for example, in the methods or results section. Specific issues include, but are not limited to:

Response to Reviewer 2
The authors thank the reviewer 2 for his/her review comments and valuable suggestions for improving our manuscript. As the reviewers pointed out, there were several ambiguous descriptions in terms of the RNA-seq data analysis. According to the suggestion, we described all the relevant information to RNA-seq analysis in methods section. We have described our responses to your comments one by one as below. Moreover, in order to show in detail the "identification and variation of target genes of Regnase-1", we divided original Figure 6 into two ( Figure   6 and Figure 7) and added data. 2. How many biological replicates were used to generate the data in Fig. 5f and Fig. 6, showing the overall differential expression, i.e., how many mice were used to prepare how many independent cultures of HSC used in this experiment?
Our response to Comment 2-2: We used HSCs collected from one individual mouse of each genotype in Figures. 5f-i and Figure. 6. HSCs separated and sorted by flow cytometry were directly analyzed without culture. We inserted an explanation of the sample preparation method for RNA-seq analysis to the Figure legends.
3. Regardless of the answers to 2 above, was the RNA-Seq experiment performed more than once?
Our response to Comment 2-3: Thank you for your question. We performed two independent RNA-Seq experiments and similar results were obtained. We added this information to the Figure legends.
4. What was the statistical method used to determine statistical significance of the differences?
Our response to Comment 2-4: We detected 801 up-regurated genes and 1179 down-regulated genes in the Reg1 Δ/Δ according to the following criteria: log2 fold change of the average expression value in Reg1 Δ/Δ mice (case) compared with that in Reg1 flox/flox mice (control) was larger than 2 or smaller than 1/2, and the average expression value in the case was larger than the 90th percentile of the case. We added the data in Supplementary table 4  Because there is no biological replicate in IGV-Sashimi analysis (Figure 6a), we did not perform statistical analysis in that experiment.
Were the transcript data limited by an abundance cutoff?
Our response to Comment 2-5: We had described criteria and cutoff for selection of candidate genes ( Figure 6b) and for identification of Regnase-1 ( Figure 1a) in the Methods section (lines 1-13, page 38) and (lines 6-17, page 37). There is no cutoff for GSEA (Figures 5f, g).
6. Although the researchers took their gene sets through many screens before settling on the handful that they analyzed further as potential direct targets of Regnase-1, I think the reader should be given easy access to the actual gene lists, showing fold changes, an indication of abundance, and an indication of the significance of the fold changes in the original normalized RNA-Seq data. Perhaps a supplemental Excel file could be used to address most or all of these questions, as well as certain fundamental questions like the relative abundance of housekeeping transcripts, the abundance of the knocked out gene (reflecting cell purity), etc. In my view, all of this information is necessary to help the reader determine the validity of the authors' conclusions on this topic.
Our response to Comment 2-6: We thank the reviewer for this constructive criticism. In accordance with the reviewer's suggestion, we now add actual gene lists of original normalized RNA-Seq data including fold-changes and an indication of abundance (Supplementary Table 3). Significance of the fold-changes and abundance cutoff value described in the Methods section is also mentioned in this list.
7. The authors then moved to their analysis of potential target transcripts. Rather than starting from the total collection of up-regulated transcripts, which most likely included many transcripts regulated by Regnase-1 either directly or secondarily, the authors moved to their exon gradient analysis, and limited the data still further by screening for up-regulated transcripts that are known to be involved in hematopoiesis. They thereby eliminated from the analysis many potential target transcripts that might well be involved in cellular function because they did not meet those screening criteria.
Our response to Comment 2-7: Regarding this comment, we are not sure exactly what your question is and are therefore wondering if the response below is adequate or not; however, this is as follows. Regnase-1 binds to stem-loop element in the 3 UTR of target mRNAs and promotes degradation via its intrinsic RNase activity. While there are several studies on the binding sequences of Regnase-1, information about its interaction with target RNAs is still limited (Behrens G, et al., Nucleic Acids Res. 2018., Mino T, et al., Cell. 2015). It appears that Regnase-1 degrades various target mRNAs (Jeltsch KM and Heissmeyer V. Curr Opin Immunol. 2016); thus it seems difficult to identify all of the target genes.
Because the results of RNA-Seq analysis show that expression of numerous genes is greatly affected by Regnase-1 deficiency, we focused on transcription factors that influence the expression of a wide range of genes. We have added this information and explanation to the Results section (lines 8-11 page 17). Fig. 6c, I assume that these data were from real-time QPCR, but it is not clear that this is the case from the Results section.
Our response to Comment 2-8: As pointed out, data of Figure 6c were obtained by real-time QPCR analysis. We have changed"quantitative PCR" to "real-time QPCR" (lines 18 page 17).
9. It would be very helpful in the section on target transcript analysis to review the binding and/or cleavage sites for this protein, and the evidence to support these.
Our response to Comment 2-9: We appreciate the reviewer's advice. As mentioned in "Our response to Comment 2-7", Regnase-1 binds to the 3´ UTR of target mRNAs and promotes degradation. However, reports about the Regnase-1 binding sequences and mechanism of cleavage are still limited (Behrens G, et al., Nucleic Acids Res. 2018). Because the molecular function of Regnase-1 has not been clearly elucidated, we have added a complementary explanation. Cleavage sites of target mRNA are not clear, but our data from sashimi IGV analysis show a gradient in 3' and 5' mRNA amounts of Gata2 and Tal1 in HSC from control mice relative to Regnase-1-deficient mice. These results indicate that degradation of Gata2 and Tal1 was induced in the presence of Regnase-1. To make this clearer to understand, Bmi1 which is a transcriptional regulator important for HSC proliferation and differentiation, was added as an example of an undegraded molecule in Figure 6a. Our response to Comment 2-10: As indicated by the reviewer, overexpression studies often result in abnormal conditions for the cell and thereby cause artefactual effects. To exclude this, we performed actinomycin D blockade experiments using primary cultured Regnase-1-deficient HSC, because this design was used in a previous study of Regnase-1 (Uehata T, et al., Cell. 2013). Marked Gata2 and Tal1 mRNA degradation was observed in HSCs from control mice, but this effect was abrogated in HSCs from Regnase-1-deficient mice. We have added these results to Figure 6f (lines 12-14, page 18).
In summary, my view is that the authors have not done a convincing job of identifying what they propose as the direct targets of Regnase-1, whose accumulation results in the hematopoietic phenotype they describe. Depending on some of the answers to the questions above, the authors may already have the data for the overall changes in abundance of various transcripts in in the control and KO cells, but I think they need to do a much more detailed job of presenting those data as well as confirming the proposed changes in mRNA stability by other means.

Reviewer #3 (Remarks to the Author):
This manuscript by Kidoya et al reports the function of Renase-1 in regulation of hematopoietic stem cells and hematopoietic pathology. There are a number of novel observations but many conclusions are not convincing and require further development. The stated mechanism that Gata2 and Tal1 hyperactivation mediates the Regnase-1KO HSC self-renewal and leukemia is not well supported.

Response to Reviewer 3
Thank you very much for your review comments and valuable suggestions to improve our manuscript. In response, we have added new data and further discussion. The definition of HSC was ambiguous in the manuscript, thus we described it more exactly by rewriting the terms (e.g., "LT-HSC" into "CD34 -HSC"). It was not clear whether AML-like phenotype of Vav1-Cre; Regnase-1 flox/flox mice was caused by Regnase-1 deficiency or Gata2 and Tal1 hyperactivation in HSC. In accordance with reviewer's suggestions, we examined the influence of blood cell differentiation or hematocyte proliferation for leukemia symptom. Please see our responses to your comments one by one below.
Major concerns: 1. The authors states that Regnase-1 selectively affects self-renewal of LT-HSC but the data to support the selectivity is not strong. First, Fig. 1d c-Kit + cells are not all HSC. Second, Fig.3a  Our response to Comment 3-1: 1. We thank the reviewer for this constructive criticism. The definition of HSC subsets was imprecise in many parts of the manuscript, and for this reason we have now rewritten all descriptions of HSCs more precisely ("LT-HSC" is now "CD34 -HSC", "ST-HSC" is now "CD34 + HSC"). As indicated by the reviewer, c-kit + cells in Figure 1d are not always HSCs, and we rewrote this to read "c-kit + cells including hematopoietic progenitor cells". In the same way, In accordance with the reviewer's suggestion, we investigated whether Regnase-1 degrades CD34 mRNA in CD34 -HSCs (LT-HSCs). CD34 -HSCs collected from control and Regnase1-KO mice were cultured under conditions where RNA synthesis was inhibited by actinomycin D, and the degradation of CD34 mRNA was examined. Gata2 and Tal1 mRNA degradation was suppressed in CD34 -HSCs from Regnase-1 KO mice, but CD34 mRNA degradation was not significantly different between control and Regnase-1 KO CD34 -HSC as shown in Figure 6f (lines 12-14, page18). From this result, we considered that CD34 could be used as a marker not affected by Regnase-1.
2. The conclusion that Regnase-1 does not affect differentiation was also not well supported. Fig. 4e show changes in frequency of lymphoid and myeloid cells. The cell number and the age of mice needs to be shown. If these were 8 weeks old mice, how early could the defects be seen? If early changes in lineage composition were seen, then differentiation could be affected in Reg-1 KO mice.
Our response to Comment 3-2: Thank you for the important suggestions. We used 8 week-old mice for lymphoid and myeloid cell analysis in 3. The authors stated that apoptosis of LT-HSC and ST-HSC was not affected in Reg-1 KO, however, Fig. 3i indicate significant (p<0.005) difference.

Our response to Comment 3-3:
We thank the reviewer for pointing out our mistake. The notation of Figure 3i was an error and we have corrected this in the revised manuscript.

Comment 3-4:
4. It is not clear what was the criteria for classifying the pathology seen in Reg-1 KO mice as AML-M4. More detailed characterizations need to be provided. Also, is the disease transplantable?
Our response to  In accordance with the reviewer's suggestion we described the details of the pathological examination in Results (lines 2-8, page 15). In the peripheral blood from Regnase-1 KO mice, we observed 1 to 14% of lymphoblasts and 16 to 22% of atypical lymphocytes exhibiting abnormal nuclear morphology, 2 to 4% of mature myelocyte and metamyelocyte and 1 to 16% of atypical cells including stab cells in neutrophils. In addition, 3 to 8% of monoblasts and 6 to 20% of atypical mononuclear cells were present, and proerythroblast-polychromatophilic erythroblasts were observed. Furthermore, 1% of myeloblasts were found. Therefore, we diagnosed AML.
To investigate whether the AML was transplantable, we transferred mononuclear cells (MNCs)  5. It is not clear why the authors switched to Tie2Cre for the RNA-seq experiment. As Tie2Cre is also expressed in endothelial cells, they are not hematopoietic specific as the authors stated. It would be more logical to examine LT-HSC by RNA-seq using the VavCre mice where the hematological phenotypes were observed. The definition of "similarities" between Reg-1KO HSC and AML needs to be described. Also, it is not clear why the authors compared the gene expression of Reg-1KO HSC with human BCR-ABL+ AML. If the authors think that the disease seen in Reg-1 KO mice resembles AML-M4, then it would make sense to compare to AML-M4 patients or patients with reduced REG-1 expression. Can the authors show that Regnase-1 expression itself associates with prognosis?
Our response to  As indicated by the reviewer, Tie2Cre causes recombination in vascular endothelial cells in addition to blood cells.
We performed RNA-seq analysis by using CD34 + HSCs from BM of Vav-Cre Reg1 flox/flox mice and obtained similar results (please see the additional figure for the reviewer). However, these RNA-seq data seem technically noisy, and we were unable to acquire precisely the same results especially in IGV-Sashimi plot analyses (Figure 7a).
Technical noise of low-input RNA-seq analysis is reported to affect downstream analyses for detecting subtle biological differences of gene expression in small amounts of cells (Sheng K, et al. Nat Methods. 2017., Jia C, et al. Nucleic Acids Res. 2017). Therefore, we have omitted the data from Vav-Cre Reg1 flox/flox mice from the manuscript, and retained the previous results from Tie2Cre Reg1 flox/flox mice. We anticipate that the reviewer will appreciate this technical limitation. Regarding the subtype of AML, to properly respond to comment 3-4, we repeated the experiments using more mice. As a result, a wide range of AML-like cells representing not only specifically AML-M4 was confirmed. Therefore, we have corrected the description of the AML-M4 phenotype in the text and have performed GSEA analysis with human AML-M1 or -M3 cells (Figure 5g). The "similarities of GSEA" means that sample genes sets show statistically significant concordant differences with a priori-defined sets of genes which are known to have an association with disease phenotypes. We have added this information to the Results 6. The authors show that Regnase-1 protein levels in human AML cell lines were much lowered than healthy PBMC. Does the author know that Regnase-1 is also expressed higher in human HSC compared to more mature populations? The conclusion that Regnase-1 deletion in HSC leads to gene expression signature that is similar to AML HSC is not well supported.
Our response to  We thank you for your constructive criticisms. We agree with the comment that analysis of Regnase-1 expression in human HSC was necessary in the context of relevance for human AML. We evaluated protein expression of Regnase-1 in human BM-derived CD34 -HSCs, which are commercially available. As shown in Supplementary showing Gata2 and Tal1 mRNA degradation need to be shown.
Our response to  As suggested by the reviewer, it is important to show the degradation of Gata2 and Tal1 mRNA in the presence of Regnase-1. In accordance with this suggestion, we performed Northern blotting and added these data to Figure 6e and the Results (lines 8-14, page 18). Because it is a challenge to prepare sufficient mouse HSCs for these assays, we analyzed a THP 1 leukemia cell line overexpressing Regnase-1. As shown in Supplementary Figure 6a, we confirmed that Regnase-1 overexpression decreases Gata2 and Tal1 mRNA expression. Northern blotting revealed that overexpression of Regnase-1 causes degradation of Gata2 and Tal1 mRNA (Figure 6e). Moreover, we analyzed mRNA degradation using actinomycin D, in order to respond to Reviewers' comments 2-3 and 3-1, and confirmed that half-lives of Gata2 and Tal1mRNA were significantly extended in Regnase-1-deficient CD34 -HSCs compared with CD34 -HSCs from control mice (Figure 6f).
The effects of Gata2 and Tal1 knock-down on Reg-1 KO LT-HSC cell cycle need to be shown to support this claim.
It is, however, difficult to ensure that LT-HSC is maintained following the cytokine stimulation and retroviral transduction. Therefore, the effects in proliferation seen is not necessarily effects in LT-HSC. In this context, understanding whether Regnase-1 selectively affects self-renewal of LT-HSC or also affects MPP and later progenitor (comment 1) is important to interpret these results.
Our response to  To determine whether Gata2 and Tal1 hyperactivation are involved in the cell cycle abnormalities seen in Regnase-1-deficient HSCs, Gata2 and Tal1 expression by CD34 -HSCs (LT-HSCs) from Regnase-1 KO BM was downregulated by shRNA transfection and then those cells were transplanted into C57BL/6-Ly-5.1 congenic mice.
As in the analysis of Figure 7 f-h, 2 months after transplantation, we confirmed that abnormal proliferation was suppressed in transplanted HSCs; we then analyzed the cell cycle of CD34 -HSCs and confirmed that cell cycle abnormalities in CD34 -HSCs due to Regnade-1-deficiency was partially recovered by knockdown of Gata2 and Tal1 expression. We have added these results to Figure 7h and the Results section (lines 14-18, page 19). As described in response to Comment 3-1, cell cycle abnormalities due to Regnase-1 deletion is mainly observed in CD34 -HSCs and CD34 + HSCs; thus we considered that hyperactivation of Gata2 and Tal1 in immature HSCs may be involved in the development of AML-like symptoms in Regnase-1 KO mice.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): The authors well answered my concerns. In the rebuttal, the authors found that Regnase-1 is expressed in differentiating cells, implying that Regnase-1 may play a role in the differentiation of linage-committed cells. This reviewer expects that the authors examine this issue in their next study.
Reviewer #2 (Remarks to the Author): In my view, this revised paper did not correct many of the fundamental problems that were present in the original. For example, focusing on the differential gene expression aspects of the paper, they clarified that they were using a single mouse per genotype, and a single biological replicate, for their analysis pictured in Fig. 5e. They mentioned in the figure legend for Fig. 5e that there were two independent experiments, but they did not clarify whether Fig. 5E represented a single experiment or was an average of the two. In any case, in my opinion, this number of biological replicates is considerably fewer than most journals and reviewers would accept for adequately controlled, statistically validated differential expression analysis. As best I can tell, the authors provided a bit more explanation for this but did not address the fundamental problem, i.e., too few biological samples.
Another fundamental problem that was addressed in the original review was that most of the experiments were performed with mice expressing one particular Cre, whereas the differential expression analysis used a completely different Cre. In the authors' response, they said that they obtained "similar results" using the Vav-cre mice, but that the data seemed "technically noisy, and we were unable to acquire precisely the same results". I don't think this is an adequate explanation.
My view is that the differential gene expression aspects of the paper are fundamental to their conclusions, but, based on some of the factors enumerated above, we are unable to rely on these data as presented.
Reviewer #3 (Remarks to the Author): In general, this manuscript by Kidoya et al has improved but a number of significant concerns remain. Particularly, it is not convincing that the pathology seen in the Reg1 deficient mice are spontaneous AML. It is generally thought that AML pathogenesis require multiple "hits". It seems quite unlikely that there is full malignant transformation in 2-3 months. In addition, CD34-HSC transplant do not seem to develop leukemia-like pathology, this argues against Regnase1 lost is sufficient to cause leukemia. Given the role of Regnase1 in regulating inflammatory genes in immune cells, it is important to consider possible consequences on the observed phenotype not necessarily the consequences of its role in CD34-HSC.
Major issues: 1. If Reg1 loss accelerates CD34-HSC cell cycling as the author show, it is possible that would lead to HSC exhaustion and explain the results shown for transplantation experiment (Fig. 2f,h,I,g). Is there expansion of progenitors at earlier time points? Do the author have engraftment/repopulation data for earlier time points? It would help clarify the effects on progenitor proliferation/differentiation and HSC exhaustion.
2. It is rather confusing to state that Reg1 deficiency leads to bone marrow failure like disease (nonmalignant) and promote leukemogenesis (malignant) in what appears to be the same animals and same time point? Was there increase total WBC counts in the diseased mice? The cell numbers in PB shown in Fig 4e, m seem very low and the abnormal cells seen in the PB includes both lymphpoid and myeloid cells (Fig. 5c). These observations are not typical for a diagnosis of AML as the authors claimed. The authors now show transplantation of cells from diseased mice but it is not mentioned whether these mice also developed "leukemia". A transplantable disease means that the disease phenotypes, including major pathological features defining the types of leukemia in this case is reproduced in the transplants. Given the data presented here, it is difficult to conclude that the abnormality seen in Reg1KO at 8 weeks is AML. It remains possible that what is seen is some form of MPD or MPN-like condition. The authors further made a strong statement that "results in the spontaneous development of leukemia even in heterozygous mice" (Page 26, line 3). There is very limited characterization of the heterozygous mice to support this claim.
3. Down regulation of Regnase1 in AML cells is not sufficiently supported by just comparing AML cell line to PBMC. Particularly since the author showed that Regnase1 is also expressed in lymphoid cells (and appears to be higher than myeloid cells) (Supplemental Figure 1a), the lower levels seen could be just due to the different cellular composition.
Minor issues: 1. In Fig. 1c, which one is CD34-HSC? 2. In Fig. 2d, are all cells CD34-? Is the expansion of CD48+LSK shown in Fig 2d related to changes cell cycle?
3. Fig. 2g show increased % of CD34-cells in donor derived LSK for Regnase-1KO but there is no mention about absolute cell number. Given that the % of donor derived LSK is greatly reduced, this information alone is not sufficient to conclude that self -renewal capacity of CD34-HSC is increased (page 10). Also, can the authors explain the difference here (almost no CD34+HSC cells vs. increased CD34+HSC cells in Fig 1i) and also no difference in proliferation in Fig. 2e? 4. Fig.3 e,f show significant increase of all cell cycle phases (G0, G1, S/G2M) in Reg1KO CD34-HSC but the text mention reduction of G0 phase (page 12, line 3)? Which is correct? If all phases of cell cycle are increased, how to interpret this? 5. Page 21, line 13 in discussion, should be "promoting" cell cycle progression, not suppressing.
6. Page 21, Line 16-17, please clarify where "Regnase-1 is down-regulated in human AML cells and that its expression is inversely associated with AML prognosis" is shown. 7. Page 25 line 13, not sufficient evidence to suggest that Regnase-1 is the "causative gene for human AML. Line 16, "Regnase-1 might be useful for predicting AML prognosis" contradicts the author the main text that Ragnase 1 expression is not associated with prognosis.

Reviewer #1 (Remarks to the Author):
The authors well answered my concerns. In the rebuttal, the authors found that Regnase-1 is expressed in differentiating cells, implying that Regnase-1 may play a role in the differentiation of linage-committed cells. This reviewer expects that the authors examine this issue in their next study.

Reviewer #2 (Remarks to the Author):
Thank you for your valuable comments on our paper which we have amended according to your suggestions.
Comment 2-1: In my view, this revised paper did not correct many of the fundamental problems that were present in the original.
For example, focusing on the differential gene expression aspects of the paper, they clarified that they were using a single mouse per genotype, and a single biological replicate, for their analysis pictured in Fig. 5e. They mentioned in the figure legend for Fig. 5e that there were two independent experiments, but they did not clarify whether Fig. 5E represented a single experiment or was an average of the two. In any case, in my opinion, this number of biological replicates is considerably fewer than most journals and reviewers would accept for adequately controlled, statistically validated differential expression analysis. As best I can tell, the authors provided a bit more explanation for this but did not address the fundamental problem, i.e., too few biological samples.
Our response to Comment 2-1: Regarding the results of gene expression analysis by RNA-seq, we had performed two independent experiments using a single mouse per genotype, and shown the average of the two. As you pointed out, there are too few biological samples and the results are unreliable. In order to deal with this problem, it is necessary to perform multiple experiments with increased numbers of biological samples. Hence, we performed two independent experiments with n=3 per genotype and n=2 per genotype. As a result of this approach, we have now obtained data on gene expression with greater reproducibility, and have again averaged them. These new results are very similar to the previous data in the first version of the paper, but we do consider them to be more reliable now. They now replace the previous figure (Figure 2j and 5f-i). Another fundamental problem that was addressed in the original review was that most of the experiments were performed with mice expressing one particular Cre, whereas the differential expression analysis used a completely different Cre. In the authors' response, they said that they obtained "similar results" using the Vav-cre mice, but that the data seemed "technically noisy, and we were unable to acquire precisely the same results". I don't think this is an adequate explanation.
Our response to Comment 2-2: As you pointed out, despite the fact that we had performed all biochemical analyses using Vav1-iCre : Reg1 fl/fl mice, bioinformatics analysis of RNA-seq was performed using Tie2-Cre : Reg1 fl/fl mice. To improve the reliability of this research by ensuring consistency, we have now performed RNA-seq analysis using Vav1-iCre : Reg1 fl/fl mice. The results are similar to the previous data from Tie2-Cre: Reg1 fl/fl mice, but not identical. In particular, we could not reproduce the data shown in the previous Figure 6 a,b, and we have therefore deleted this. We searched for candidate genes affected by Regnase expression, not only by informatics analysis but also by biochemical analysis using luciferase-based screening assays. We have now added this information and the data to the Results section (line 17, page 17 -line 9, page 18).
Regarding the reason why the results of gene expression analysis were different between Tie2-Cre : Reg1 fl/fl and Vav1-iCre : Reg1 fl/fl , it seems that Tie2Cre causes recombination in vascular endothelial cells in addition to hematopoietic cells, whereas Vav1Cre acts only in hematopoietic cells specifically.
My view is that the differential gene expression aspects of the paper are fundamental to their conclusions, but, based on some of the factors enumerated above, we are unable to rely on these data as presented.
Our response to Comment 2-3: We believe that the quality and reliability of this paper have been improved greatly by using the appropriate number of mice (Vav1-iCre : Reg1 fl/fl ) so that there are sufficient biological samples and new data included. I'd like to thank you for pointing out the essential issues to improve the paper

Reviewer #3 (Remarks to the Author):
In general, this manuscript by Kidoya et al has improved but a number of significant concerns remain. Particularly, it is not convincing that the pathology seen in the Reg1 deficient mice are spontaneous AML. It is generally thought that AML pathogenesis require multiple "hits". It seems quite unlikely that there is full malignant transformation in 2-3 months. In addition, CD34-HSC transplant do not seem to develop leukemia-like pathology, this argues against Regnase1 lost is sufficient to cause leukemia. Given the role of Regnase1 in regulating inflammatory genes in immune cells, it is important to consider possible consequences on the observed phenotype not necessarily the consequences of its role in CD34-HSC.

Response to Reviewer 3
Thank you for your valuable comments on our paper which we have amended according to your suggestions.
As you say, AML develops through multiple genetic hits and we agree that it is unlikely that leukemia was caused solely by Regnase-1 deficiency. Therefore, we have added the following sentence to the paper: "Excessive proliferation of HSPC induced by deficiency of Regnase-1 may provide the possibility to progress to AML by promoting mutagenesis." in Discussion (lines 18-19, page 21). In addition, we changed the designation from "AML and leukemia" into "AML-like phenotype", because the phenotype of Regnase-1-deficient mice does not completely match AML, as you point out. If you recommend that we should describe the phenotype as AML from the above results nonetheless, we would be glad to re-write this again.
Moreover, responding to major comment 2, we analyzed peripheral blood from CD34 -HSC-transplanted mice and confirmed that they exhibited an AML-like phenotype. Furthermore, in the Discussion, we introduce the possibility that the phenotype of Regnase-1-deficient mice may be caused by an effect on differentiated blood cells.  (Fig. 5c). These observations are not typical for a diagnosis of AML as the authors claimed. The authors now show transplantation of cells from diseased mice but it is not mentioned whether these mice also developed "leukemia". A transplantable disease means that the disease phenotypes, including major pathological features defining the types of leukemia in this case is reproduced in the transplants. Given the data presented here, it is difficult to conclude that the abnormality seen in Reg1KO at 8 weeks is AML. It remains possible that what is seen is some form of MPD or MPN-like condition. The authors further made a strong statement that "results in the spontaneous development of leukemia even in heterozygous mice" (Page 26, line 3). There is very limited characterization of the heterozygous mice to support this claim.
Our response to Comment 3-2: Thank you for these important suggestions. As you pointed out, the exact phenotype of Regnase-1-deficient mice needs to be described. Because we could not clarify whether "bone marrow failure" is really appropriate to describe the symptoms of Regnase-1-deficient mice, we have deleted this phrase "Bone marrow failure".
In terms of WBC counts, we have analyzed the number of WBC in Regnase-1-KO mice and have observed increases in these cells (line 15, page 15, Figure 5d). Moreover, to test whether the Regnase-1 deficiency-induced leukemic cells are transplantable, we transferred CD34 -LSK cells from Regnase-1-KO mouse bone marrow to C57BL/6-Ly-5.1 congenic mice. Three months after transplantation, we analyzed recipient peripheral blood and found that abnormal blood cells were present, but no blasts (lines 7-8, page 16, Supplementary Figure 5d). Furthermore, in mice heterozygous for Regnase-1, pathological examination of the peripheral blood of such 1-year old heterozygotes also revealed an AML-like phenotype, as in Regnase-1-KO mice (lines 16-17, page 15, Supplementary Figure 4c). Nonetheless, we agree that these results of peripheral blood analyses may not to be sufficient to diagnose AML, as you have indicated. Therefore, we now refer throughout the paper to Regnase-1 deficiency causing "aberrant differentiation" instead of "AML and leukemia" (lines 10-11, page 16). 3. Down regulation of Regnase1 in AML cells is not sufficiently supported by just comparing AML cell line to PBMC.
Particularly since the author showed that Regnase1 is also expressed in lymphoid cells (and appears to be higher than myeloid cells) (Supplemental Figure 1a), the lower levels seen could be just due to the different cellular composition.
Our response to  In addition to the comparison of Regnase-1 protein in AML cells and PMBCs, we confirmed that human HSCs express more Regnase-1 protein than PMBCs (lines 8-11, page 17, Figure 5j). However, we cannot exclude the possibility that the decrease in Regnase-1 protein expression in AML cell lines is due to the differentiation stage of the myeloid cells, and we were unable to estimate the proportion of lymphoid cells in PBMCs. Therefore, we deleted sentences from the Discussion which referred to a causal relationship between Regnase-1 and AML that was based on the analysis of Regnase-1 protein expression levels in AML cell lines. 1. In Fig. 1c, which one is CD34-HSC?
Our response Thank you for spotting our mistake. The notation of Figure 1c was an error and we have corrected this in the revised manuscript. The graph in Figure 2d did not show data from CD34cells. Therefore, we performed an analysis of SLAM using CD34 -KSL cells. While CD48 -CD150 + cells were the main population in CD34 -KSL cells from control mice, we observed an increase in the number of CD48 + cells in both CD150 + and CD150populations in CD34 -KSL cells of Rregnase-1-deficient mice (Supplementary Figure 1b).
Analysis of the cell cycle status of CD48 + LSK showed that there are many proliferating cells at G1 and G2/M phases consistent with increased numbers of cells in Regnase-1-deficient mice relative to wild-type mice (Supplementary Figure 1c). We have added this information and explanation to the Results section (lines 3-5, page 9).  Fig. 2g show increased % of CD34-cells in donor derived LSK for Regnase-1KO but there is no mention about absolute cell number. Given that the % of donor derived LSK is greatly reduced, this information alone is not sufficient to conclude that self -renewal capacity of CD34-HSC is increased (page 10). Also, can the authors explain the difference here (almost no CD34+HSC cells vs. increased CD34+HSC cells in Fig 1i) and also no difference in proliferation in Fig. 2e?

Our response
In accordance with your suggestion, we confirmed increased CD34-HSC numbers in Regnase-1-deficient BM and replaced the graph showing the absolute cell number (Figure 2g).
It is difficult to explain the difference in growth of HSC in Figure 1i and Figure 2e, but we are considering the possibility that reduction of colony-forming ability is caused by unbalanced HSC differentiation and proliferation, for example, by decreased stemness due to excessive cell cycling. However, as this is speculative, we have not mentioned this possibility in the paper. If you recommend that we include this in the Discussion, we would be glad to do so.