Loss of Asxl2 leads to myeloid malignancies in mice

ASXL2 is frequently mutated in acute myeloid leukaemia patients with t(8;21). However, the roles of ASXL2 in normal haematopoiesis and the pathogenesis of myeloid malignancies remain unknown. Here we show that deletion of Asxl2 in mice leads to the development of myelodysplastic syndrome (MDS)-like disease. Asxl2−/− mice have an increased bone marrow (BM) long-term haematopoietic stem cells (HSCs) and granulocyte–macrophage progenitors compared with wild-type controls. Recipients transplanted with Asxl2−/− and Asxl2+/− BM cells have shortened lifespan due to the development of MDS-like disease or myeloid leukaemia. Paired daughter cell assays demonstrate that Asxl2 loss enhances the self-renewal of HSCs. Deletion of Asxl2 alters the expression of genes critical for HSC self-renewal, differentiation and apoptosis in Lin−cKit+ cells. The altered gene expression is associated with dysregulated H3K27ac and H3K4me1/2. Our study demonstrates that ASXL2 functions as a tumour suppressor to maintain normal HSC function.

and provides a better understand of hematopoiesis in both normal and leukemic contexts. Overall, the experiments were well performed and the results are interesting. The manuscript could be improved by addressing the comments below.
Major comments 1-In the Figure 3: The authors should further define the proliferative capacity (for example by BrdU incorporation) and cell cycle of ASXL2 deficient HSPCs 2-Supplementary figure 3d: Because ASXL2 mutations are found in AML-ETO AML, the authors tested the possible cooperation between ASXL2 loss and AML-ETO transformation. The data presented are quite sparse and this part could be removed from the manuscript, if no more detailed experiments are added, as the work featured is insufficient to be convincing and does not add a lot to the story. Figure 4: After transplantation of the ASXl2-/-BM in recipients, the mice die from what is described as a leukemia, with increased chimerism, high proportion of MAC1+ cells from the donor, 20% of ckit+ cells in the BM, splenomegaly, anemia and invasion of blast cells in the BM and spleen. However, in case of Acute Myeloblastic leukemia (as stated at least in the abstract), one would expect to see the chimerism increase beyond 50%; indeed the leukemic cells would overtake the competitive population. It would also be interesting to have the WBC count in the blood and a blood smear to assess the invasion of the blast cells in the blood. Overall, more detailed FACS analysis of the populations in the different organs (BM, Spleen and blood) should be included to have a better idea of which type of leukemia the recipient mice develop. Figure 5: It is not clear how the authors came to look at their 3 targets genes (5d). It does not look very convincing to choose, without providing a rationale for it, 3 genes from a list of 150. If the authors want to claim that these genes are important for the observed phenotype, experiments should be conducted to this end. For example, does Knockdown of PRMD16 rescue the phenotype (in a CFU assay)?

4-
5-There is no mention of the cut-off used to determine the differentially expressed genes in term of fold change (FC)? Was it 2? 6- Figure 6: In figure 5, the authors mention 68 up and 92 downregulated genes, whereas there are around 1000 sites with changes in histone modification. Can the authors comment on what happens to the genes with changes in histone modification not associated with a difference in gene expression as they represent the majority of the genes observed? 7- Figure 6e: The authors write "Furthermore, ChIP-seq analysis confirmed the higher H3K27ac peak at the distal region of Prdm16": in the data presented in the figure, the difference is not obvious. The FC between the 2 signals should be indicated. 8-The authors wrote "the 1559 genes with differential H3k27ac signals have higher mRNA expression difference with a median FC of 1.332"; however, this way of representing data is confusing. One would expect to have genes with more H3k27ac to be upregulated, and with less H3k27ac to be downregulated. Is it the case? Terms such as "differential", "change" or "dysregulation" should be clarified. 9-The authors should provide the list of genes with the data for the RNA-seq and Chip-seq.
10-It would be important to perform ASXL2 ChIP-seq to correlate change in gene expression and histone modifications to ASXL2 binding. Figure 6a: The R values don't seem very high; could the authors comment on this point? Is there really a substantial correlation, especially for H3K27Ac? 12-It would be interesting to know if the effects observed in adult mice are also observed during development, in particular in the fetal liver of the ASXL2 ko mice.

11-
Minor comments -The manuscript may benefit from proofreading by a native English reader - Figure 2d: in the figure legend it is indicated "quantification for both LSK and LKS-population" however only LSK is found in the figure - Figure 5: The number of differentially expressed genes does not appear in the main text, only in the figure legend; for the sake of clarity, these numbers should be cited in the main text. - Figure 5: Please correct the numbering of the figure (c is missing) - Figure 5b: The IPA representation is too small (text too small) -Line 106: The authors mention pancytopenia but increased neutrophil counts. Lymphocyte counts would be useful.
-Line 156: The authors should discuss the significance of the expanded GMP population in the MDS/AML context? -Methods -line 407-410: More details or reference would be appreciated -Line 415: Could the authors detail what FACS antibodies were used? -Line 440: "The self-renewal ability and differentiation capability of these single CD34-LSK cells were determined by the potential of retaining multipotent lineage differentiation potential or lineage commitment": please clarify how (single cell colony assay)?
1. Asxl2 KO HSPCs shows growth advantage in culture and are less apoptotic in vivo. These phenotypes of Asxl2 KO HSPCs in gene trap mice are not compatible with MDS. In addition, ASXL2 mutations are rare in MDS. What is known about ASXL2 in MDS? Is it downregulated in MDS?

Response:
We agree with the reviewer that apoptosis in BM cells of MDS patients is elevated. We found a lower apoptotic rate in the LK cells of Asxl2 -/mice. However, there was a higher apoptotic rate in mature lineage positive cells in Asxl2 -/mice compared to WT controls. In response to the reviewer's question, we examined the levels of ASXL2 mRNA expression in the BM CD34 + cells of a cohort of MDS patients (n=18) and compared with that in CD34 + cord blood cells and G-CSF mobilized CD34 + cells. A lower expression of ASXL2 was observed in four out of 18 MDS patients examined. These data suggest that ASXL2 is down regulated in at least some of the MDS patients. Future study with a larger cohort of patients is necessary to true frequency of downregulated ASXL2 expression in MDS patients.
2. In contrast to the growth advantage of Asxl2 KO BM HSPCs in culture, it shows lower repopulating capacity in competitive repopulation assays. How do the authors explain this discrepancy? Why Asxl2 +/did not show impaired repopulation?

Response:
We agree with the reviewer that there is a discrepancy in the phenotypes on HSPCs between the in vivo and in vitro assays. Our in vivo phenotypic analysis revealed that loss of Asxl2 mainly affected LT-HSCs with less apoptosis. In contrast, Asxl2 loss increased the apoptosis in lineage committed and matured cell populations, which might cause the transient reduction of repopulating capacity of Asxl2 deficient BM cells at early time points after transplantation. However, after five months, Asxl2 null LT-HSCs expanded in the recipient mice, resulting in a higher reconstitution rate. Our data showed that Asxl2 haploinsufficient mice exhibited mild hematopoietic phenotypes, except lower survival rate in the Asxl2 +/-BM transplanted recipients. These data may suggest a gene dosage dependent effect of Asxl2 in the hematopoiesis.
3. Do the straight Asxl2 KO also develop leukemia?

Response:
We thank the reviewer for this great question. To address this question, we analyzed a cohort of aged Asxl2 -/mice and found that Asxl2 -/mice around 12 months of age did develop leukemia as evidenced by high WBC counts, increased proportion of Gr1 + /Mac1 + , GMP in the BM, splenomegaly and hepatomegaly, as well as white femurs. LT-HSCs in the BM were also much higher in these aged Asxl2 -/mice than that in WT controls. These data are now added as Supplementary Fig. 4a-e in the revised manuscript on page 9. 4. In competitive repopulation assays, some of the mice show MDS-like phenotype and others develop AML. Does this indicate that AML develop from the MDS clone? Do all the recipient mice show MDS phenotype at the early time points? Please present the frequency of the Asxl2 +/and Asxl2 -/mice developing MDS. Do the MDS mice show increased apoptosis not only in differentiated cells but also in HSPCs?
Response: MDS-like phenotypes can be seen in most of the Asxl2 +/or Asxl2 -/-BM recipient mice before leukemic transformation. It is likely that transformation of leukemia in these recipients were evolved from the MDS clone. We presented these data in Supplementary  Table 3  5. In Figure 6, without ASXL2 ChIP-seq data or biochemical data of ASXL2 complex, we cannot evaluate the association of ASXL2 function with enhancer histone modifications. The notion that ASXL2 regulates enhancer H3K27ac has not been supported by any direct experimental data in this study. Is there any molecular mechanism known about this?
Response: We thank the reviewer pointed out an important question. We performed ASXL2 ChIP-seq assay to determine ASXL2 binding sites on the genome using WT LK cells. ChIPseq analyses revealed that a 12% of ASXL2 peaks was located at promoter regions, and the rest of ASXL2 peaks were at gene body (41%) and intergenic regions (47%). GO/KEGG analysis revealed that ASXL2 binding sites were enriched at genes that are associated with cell differentiation, positive regulation of transcription from RNA polymerase II promoters, etc. These data are now shown as Supplementary Fig. 10b,c and on page 14 in the revised manuscript. Of note, similar to Dr. Abdel-Wahab's group, we only detected a low number of ASXL2 peaks, which might due to the low affinity of the ASXL2 antibody or relative low expression of ASXL2. Generation of high quality ASXL2 antibody is required for confirming these observations. In addition, two histone deacetylation-related proteins, HDAC1 and HDAC2 were identified to associate with ASXL2 by immunoprecipitation (IP) and western blot, implying the potential of involvement by ASXL2 in the regulation of H3K27 deacetylation. We recognize that further studies to elucidate the mechanism of ASXL2 in regulating HDAC activity and H3K27 acetylation is warranted. These data are now shown as Supplementary Fig.  13a-c and on page 16 in the revised manuscript.
Minor: 1. Although it has been already reported, please briefly describe the targeting strategy of Asxl2 in mice. Which portion of Asxl2 is deleted by gene trap? Are they null mice?

Response:
We apologize for missing this information. The targeting strategy of the Asxl2 -/mice is now briefly added to the revised manuscript in the sections of Results on page 6 and Methods on page 21. Briefly, the gene trap cassette was integrated within the first intron of Asxl2, the exact site of integration is at 5016 bp downstream of exon 1. The cassette contains a polyA signal at the 3′ end, interrupting endogenous splicing and causing translation to stop. Thus, the Asxl2allele encodes a fusion protein containing the first 19 amino acids of Asxl2 joined to the β-gal reporter. This fusion protein is likely functionally null because it is missing all the conserved domains of wild-type ASXL2. Success targeting/deletion of Asxl2 gene transcription and therefore protein in hematopoietic cells are shown by PCR, qPCR and western blot, respectively ( Supplementary Fig. 1).
2. Is there any specific differentiation block in lymphocytes? Please show L-MPP, CLPs, B-cell progenitor fractions in BM and thymocytes in the thymus.

Response:
We thank the reviewer pointed out this important question. Flow cytometric analysis of BM revealed that the LMPP and CLP populations in Asxl2 -/mice were lower than those of WT controls. Furthermore, the early stage of B cells (pre-B1 and pre-B2) was decreased in the bone marrow of Asxl2 -/mice. These data are now added as Supplementary  Fig. 3a,b. To investigate the T cell populations in the thymus in vivo, we also performed flow cytometric analysis using antibodies against CD25, CD44, CD4 and CD8 to distinguish the T cell populations as DN1 (CD25 -CD44 + CD4 -CD8 -), DN2 (CD44 + CD25 + CD4 -CD8 -), DN3 (CD44 -D25 + CD4 -CD8 -), or DN4 (CD44 -CD25 -CD4 -CD8 -). The DN1, DN4, CD4 + and CD8 + populations were higher in Asxl2 -/thymus compared to WT controls. In contrast, DN2, DN3 and CD4 + /CD8 + population were lower in Asxl2 -/thymus compared to WT controls. These data suggest that deletion of Asxl2 also interrupted lymphocyte development. These data are now shown as Supplementary Fig. 3c and on page 8 in the revised manuscript.
3. In Figure 2, please show the total BM cell numbers (age).

Response:
As suggested by the reviewer, we now included the data showing the total BM cellularity in Supplementary Figure 1f in the revised version.
4. In the abstract, the authors claim that AML1-ETO9a cooperated with Asxl2 deficiency to promote HSC self-renewal. But, from this experiment (Supplemental Figure 3d), we cannot judge the HSC self-renewal. This is just a progenitor assay.

Response:
We agree with the reviewer and performed additional experiments to further evaluate the cooperative effect between Asxl2-loss and AML1-ETO. We firstly determined the effect of Asxl2 deletion and AML1-ETO expression on HSC self-renewal by transducing AML1/ETO 9a (AE9a) into fetal liver cells of WT and Asxl2 +/embryos. Asxl2 +/− cells transduced with AE9a had a significantly higher replating potential than Asxl2 +/cells transduced with WT vector. We also genetically intercrossed Asxl2 +/mice with AML1/ETO knockin mice (AML1-ETO-stop/+ Mx1-Cre + , noted as AE) and generated Asxl2 +/-;AE mice. These mice were undergone three intraperitoneal injections with pI:pC. Three months after the pI:pC injection, we performed hematopoietic phenotypic analysis on PB and BM cells. The results showed that Asxl2 +/-;AE;MxCre + mice had a higher proportion of LT-HSC in the BM, and higher frequencies of Mac1 + cells in the BM and the PB than those in Asxl2 +/or AE mice. The frequency of CFU-Cs was also significantly higher in the BM of Asxl2 +/-;AE mice (see data below). These data suggest that AML1-ETO cooperated with Asxl2 deficiency to promote myeloid differentiation. We still have limited data to support the notion of Asxl2 deletion in cooperating with AML1-ETO in leukemia transformation. As suggested by reviewer 2, "The data presented are quite sparse and this part could be removed from the manuscript", we now removed these data from the manuscript.

Reviewer #2
We are pleased that the reviewer feels "the experiments were well performed and the results are interesting".
Major comments 1-In the Figure 3: The authors should further define the proliferative capacity (for example by BrdU incorporation) and cell cycle of Asxl2 deficient HSPCs.

Response:
As suggested by the reviewer, we performed BrdU incorporation assays to determine cell cycle of Asxl2 deficient HSPCs. The result showed that Asxl2 +/and Asxl2 -/-LK cells had an increased frequency of cells at S-phase compared to WT LK cells, suggesting a higher proliferation of Asxl2 +/and Asxl2 -/-LK cells. These data are now provided in the revised manuscript as in Supplementary Fig. 6d.
2-Supplementary figure 3d: Because ASXL2 mutations are found in AML-ETO AML, the authors tested the possible cooperation between ASXL2 loss and AML-ETO transformation. The data presented are quite sparse and this part could be removed from the manuscript, if no more detailed experiments are added, as the work featured is insufficient to be convincing and does not add a lot to the story.

Response:
We agree with the reviewer. As also suggested by the Reviewer 1, we further evaluated the cooperative effect between Asxl2-loss and AML1-ETO in vivo, by genetically intercrossing Asxl2 +/mice with AML1/ETO knockin mice (AML1-ETO-stop/+ Mx1-Cre + , noted as AE) to generate Asxl2 +/-;AE mice. These mice were undergone three intraperitoneal injections with pI:pC over a six-day period of time. Three months after the pI:pC injection, we performed hematopoietic phenotypic analysis on PB and BM cells. The results showed that Asxl2 +/-;AE;MxCre + mice had an increased proportion of LT-HSC in the BM, and higher frequencies of Mac1 + cells in the BM and PB compared to those in Asxl2 +/or AE mice. The frequency of CFU-Cs was also significantly higher in the BM of Asxl2 +/-;AE mice. These data now shown as below. We still feel there are limited data to support the notion of Asxl2 deletion in cooperating with AML1-ETO in leukemogenesis. As suggested by the reviewer, we removed the Asxl2 +/-;AML1-ETO data from the manuscript. Figure 4: After transplantation of the Asxl2 -/-BM in recipients, the mice die from what is described as a leukemia, with increased chimerism, high proportion of Mac1 + cells from the donor, 20% of ckit + cells in the BM, splenomegaly, anemia and invasion of blast cells in the BM and spleen. However, in case of Acute Myeloblastic leukemia (as stated at least in the abstract), one would expect to see the chimerism increase beyond 50%; indeed the leukemic cells would overtake the competitive population. It would also be interesting to have the WBC count in the blood and a blood smear to assess the invasion of the blast cells in the blood. Overall, more detailed FACS analysis of the populations in the different organs (BM, Spleen and blood) should be included to have a better idea of which type of leukemia the recipient mice develop.

3-
Response: According to the reviewer's comment, we now provided more detailed information with flow cytometric analyses. The data showed that leukemic mouse had higher percentage of Gr1 + /Mac1 + cells in the BM, spleen and PB, as well as much higher percentage of cKit + /Mac1 + cells in BM and spleen compared to WT controls. These data are now added as Supplementary Fig. 7f in the revised manuscript. We agree with the reviewer and changed "AML" to myeloid leukemia in the "Abstract" to more precisely reflect the disease characteristics. Besides, a detailed FACS analysis of the populations in the different organs (BM, Spleen and blood) shown as Supplementary Fig. 7f-h in the revised manuscript. In addition, the WBC count in PB and a blood smear were shown as Supplementary Fig. 7d,e. Figure 5: It is not clear how the authors came to look at their 3 targets genes (5d). It does not look very convincing to choose, without providing a rationale for it, 3 genes from a list of 150. If the authors want to claim that these genes are important for the observed phenotype, experiments should be conducted to this end. For example, does Knockdown of Prdm16 rescue the phenotype (in a CFU assay)?

4-
Response: We choose these three target genes based on their known roles in apoptosis (for Fas and Casp3) or HSC self-renewal regulation (for Prdm16). Given the fact that Prdm16 is required for hematopoiesis and maintenance of HSPCs self-renewal and Prdm16 is higher expressed in AML, as suggested by the reviewer, we knocked down Prdm16 using lentiviralsystem in the fetal liver LK cells from WT and Asxl2 -/embryos. Prdm16-KD in Asxl2 -/fetal liver cells significantly decreased the percentage of LSK cells after seven days of culture as well as the frequency of CFU-C. These data are now shown as Supplementary Fig. 9b,c in the revised manuscript.
5-There is no mention of the cut-off used to determine the differentially expressed genes in term of fold change (FC)? Was it 2? Response: We thank the reviewer for this comment. We identified a set of 565 differentially expressed genes (DEGs) in Asxl2 -/-LK cells (P<0.05), in which 160 DEGs (P<0.05, FDR<0.25 and fold change≥1.5) are shown in Figure 5a. According to the reviewer's comment, we provided this information in the sections of Results and Figure legend in the revised manuscript on page 13. 6- Figure 6: In figure 5, the authors mention 68 up and 92 downregulated genes, whereas there are around 1000 sites with changes in histone modification. Can the authors comment on what happens to the genes with changes in histone modification not associated with a difference in gene expression as they represent the majority of the genes observed?

Response:
We appreciate the reviewer's comment. There are a total of ~600 and ~900 genes with increased and decreased signals in H3K27ac, respectively. Overall, these 600 genes with increased histone signals were concordantly upregulated in Asxl2 -/-LK cells compared to WT cells (P<0.01 and FDR<0.01), as shown by Gene Set Enrichment Analysis (GSEA) analysis. 40% (~260) were identified to be the leading-edge subset that contributes most to statistical significance of this analysis. The leading-edge subset identified with GSEA analysis is the core that accounts for the gene set's enrichment signal (shown in the following plot). In contrast, 68 upregulated genes mentioned were identified with Cuffdiff2 analysis at single gene level with P<0.05, FDR<0.25 and fold change≥1.5. These results suggest that a significant number (~260) of genes with increased levels of H3K27ac modification were upregulated, and a smaller set (68) of genes meet a more stringent criteria as upregulated. Similarly, these 900 genes with decreased histone signals were concordantly downregulated (P<0.01 and FDR<0.01), in which approximately 30% were identified to be the leading-edge subset that contributes most to statistical significance. 92 downregulated genes mentioned were identified with Cuffdiff2 analysis at single gene level with P<0.05, FDR<0.25 and fold change≥1.5. Similar analyses showed that genes with increased and decreased signals in H3K4me1/2 were concordantly upregulated and downregulated, respectively. To address the reviewer's question, we now provided the new data in Supplementary Fig. 11a. Figure 6e: The authors write "Furthermore, ChIP-seq analysis confirmed the higher H3K27ac peak at the distal region of Prdm16": in the data presented in the figure, the difference is not obvious. The FC between the 2 signals should be indicated.

Response:
We thank the reviewer for this suggestion. We examined the differences at single peak level, and identified several peaks with increased histone signals in this combined region. Among them, the peak in the promoter region of Prdm16 had a greatest increase of three histone modifications (2.11, 1.81 and 1.68 for fold changes of H3K27ac, H3K4me1 and H3K4me2, respectively). Accordingly, we indicated this most prominent peak with fold changes between the histone signals in WT and Asxl2 -/cells in the revised Fig. 6e. We also re-worded the description in the Figure legends. 8-The authors wrote "the 1559 genes with differential H3k27ac signals have higher mRNA expression difference with a median FC of 1.332"; however, this way of representing data is confusing. One would expect to have genes with more H3k27ac to be upregulated, and with less H3k27ac to be downregulated. Is it the case? Terms such as "differential", "change" or "dysregulation" should be clarified.

Response:
We agree with the reviewer's comment. Accordingly, we divided the 1559 genes with differential H3K27ac signals into the genes with increased H3K27ac signal and the genes with decreased H3K27ac signal. Overall, these genes with increased signals in H3K27ac, H3K4me1 or H3K4me2 were concordantly upregulated, and the genes with decreased signals in H3K27ac, H3K4me1 or H3K4me2 were concordantly downregulated, in Asxl2 -/-LK cells compared to WT cells, as shown by Gene Set Enrichment Analysis (GSEA) analysis (P<0.01 and FDR<0.01). These new data are now shown as Fig. 6b,c and Supplementary Fig. 11a in the revised manuscript. 9-The authors should provide the list of genes with the data for the RNA-seq and Chip-seq.
Response: According to the reviewer's suggestion, we now provided this data as Supplementary Table 4 and Supplementary Table 6 to show the list of genes with both RNAseq and ChIP-seq data.
10-It would be important to perform ASXL2 ChIP-seq to correlate change in gene expression and histone modifications to ASXL2 binding.

Response:
We thank the reviewer pointed out an important question. We performed ASXL2 ChIP-seq assay to determine ASXL2 binding sites on the genome using WT LK cells. ChIPseq analyses revealed that a 12% of ASXL2 peaks was located at promoter regions, and the rest of ASXL2 peaks were at gene body (41%) and intergenic regions (47%). GO/KEGG analysis revealed that ASXL2 binding sites were enriched at genes that are associated with cell differentiation, positive regulation of transcription from RNA polymerase II promoters, etc. These data are now shown as Supplementary Fig. 10b,c and on page 14 in the revised manuscript. Of note, similar to Dr. Abdel-Wahab's group, we only detected a low number of ASXL2 peaks, which might due to the low affinity of the ASXL2 antibody or relative low expression of ASXL2. Generation of high quality ASXL2 antibody is required for confirming these observations. 11- Figure 6a: The R values don't seem very high; could the authors comment on this point? Is there really a substantial correlation, especially for H3K27Ac?

Response:
We appreciate the reviewer's comment. Our original Fig. 5a showed correlations between the changes in three histone modifications and mRNA expression difference in Asxl2 -/cells compared to WT cells. Now, we also examined the correlation between histone modifications and mRNA expression in one condition. New analysis showed a high positive correlation with spearman correlation coefficient R=0.68 between H3K27ac enrichment and gene expression in WT and Asxl2 -/-LK cells (P<0.001, Supplementary Fig. 10f), which is consistent with previous studies by others (reference 21,22 in the revised manuscript). Similarly, R values ranging from 0.5 to 0.7 were identified for associations between H3K4me1/2 and mRNA expression in WT and Asxl2 -/-LK cells, respectively (P<0.001). The relative low R values with about 0.25 may reflect the fact that change in mRNA expression is influenced by multiple factors, such as transcription factors, chromatin conformation, one or multiple longrange regulatory elements. These results are now shown as Supplementary Fig. 10f and on page 15 in the revised manuscript.
12-It would be interesting to know if the effects observed in adult mice are also observed during development, in particular in the fetal liver of the ASXL2 KO mice.

Response:
The reviewer raised an important question. In response to the reviewer's suggestion, we performed flow cytometric analysis on the fetal liver cells of each of the mouse genotypes. An increased MPP and Gr1 + /Mac1 + populations were observed in Asxl2 -/fetal liver cells, while the frequencies of LSK and GMP were comparable amongst the three genotypes. Consistently, the CFU-GEMM, CFU-GM and BFU-E were significantly higher in Asxl2 -/fetal liver cells than WT and Asxl2 +/fetal liver cells. These data are now provided as Supplementary  Fig. 5a,b in the revised manuscript.
Minor comments -The manuscript may benefit from proofreading by a native English reader.

Response:
The current manuscript has been improved significantly after being edited by a native English reader.
- Figure 2d: in the figure legend it is indicated "quantification for both LSK and LKS-population" however only LSK is found in the figure.

Response:
The data for quantification of LSKis shown in Supplementary Fig.3d in the revised version.
- Figure 5: The number of differentially expressed genes does not appear in the main text, only in the figure legend; for the sake of clarity, these numbers should be cited in the main text.

Response:
The information for the number of differentially expressed genes is now provided in the Result section on page 13 in the revised manuscript.

Response:
The text for IPA representation is enlarged as suggested in the revised manuscript.
-Line 106: The authors mention pancytopenia but increased neutrophil counts. Lymphocyte counts would be useful.
Response: According to the reviewer's suggestion, we added the lymphocyte counts of the peripheral blood in the Fig. 1a of the revised manuscript.
-Line 156: The authors should discuss the significance of the expanded GMP population in the MDS/AML context? Response: Britta Will et al. revealed that lower-risk MDS is characterized by expansion of phenotypic common myeloid progenitors (CMP). In contrast, higher-risk MDS have higher proportion of granulocyte/monocyte progenitors (GMP) and these patients have the higher risk of AML transformation. The higher proportion of GMP in Asxl2 deficient mice may associate with disease progression in vivo. We added the reference and this description to the result section on page 9 of the revised manuscript.
-Methods -line 407-410: More details or reference would be appreciated.

Response:
As the reviewer suggested, we provided detailed information regarding the histological staining method on page 21, 22 in the revised manuscript.
-Line 415: Could the authors detail what FACS antibodies were used?
Response: As the reviewer suggested, the detailed information of FACS antibodies were provided in Supplementary Table 1 in the revised manuscript.
-Line 440: "The self-renewal ability and differentiation capability of these single CD34 -LSK cells were determined by the potential of retaining multipotent lineage differentiation potential or lineage commitment": please clarify how (single cell colony assay)?